Evolutionary and functional potential of ploidy increase within individual plants: somatic ploidy mapping of the complex labellum of sexually deceptive bee orchids2018 Annals of Botany
doi: 10.1093/aob/mcy048pmid: 29672665
Abstract Background and Aims Recent tissue-level observations made indirectly via flow cytometry suggest that endoreplication (duplication of the nuclear genome within the nuclear envelope in the absence of subsequent cell division) is widespread within the plant kingdom. Here, we also directly observe ploidy variation among cells within individual petals, relating size of nucleus to cell micromorphology and (more speculatively) to function. Methods We compared the labella (specialized pollinator-attracting petals) of two European orchid genera: Dactylorhiza has a known predisposition to organismal polyploidy, whereas Ophrys exhibits exceptionally complex epidermal patterning that aids pseudocopulatory pollination. Confocal microscopy using multiple staining techniques allowed us to observe directly both the sizes and the internal structures of individual nuclei across each labellum, while flow cytometry was used to test for progressively partial endoreplication. Key Results In Dactylorhiza, endoreplication was comparatively infrequent, reached only low levels, and appeared randomly located across the labellum, whereas in Ophrys endoreplication was commonplace, being most frequent in large peripheral trichomes. Endoreplicated nuclei reflected both endomitosis and endocycling, the latter reaching the third round of genome doubling (16C) to generate polytene nuclei. All Ophrys individuals studied exhibited progressively partial endoreplication. Conclusions Comparison of the two genera failed to demonstrate the hypothesized pattern of frequent polyploid speciation in genera showing extensive endoreplication. Endoreplication in Ophrys appears more strongly positively correlated with cell size/complexity than with cell location or secretory role. Epigenetic control of gene overexpression by localized induction of endoreplication within individual plant organs may represent a significant component of a plant’s developmental programme, contributing substantially to organ plasticity. Confocal microscopy, Dactylorhiza, flow cytometry, genome duplication, infra-organ ploidy mapping, labellum, Ophrys, orchid, plasticity, polyteny, progressively partial endoreplication INTRODUCTION In recent years, the general perception of the significance of ploidy change in plants has shifted from presumed evolutionary dead-end to an important – possibly even the most important – driver not only of speciation (e.g. Wendel, 2015; Van de Peer et al., 2017) but also of the origin of major lineages such as flowering plants (e.g. Jiao et al., 2011). Today, polyploidy stands accused of being responsible for evolutionary transitions that range from phylogenetically profound rare or unique events, such as the much-debated origin of the angiosperms, to frequent but smaller-scale events, such as the development of morphologically cryptic species complexes (Pillon et al., 2007; Trávníček et al., 2012). Inevitably, studies seeking process-based interpretations of ploidy change in plants have focused on model organisms such as Arabidopsis (Melaragno et al., 1993; Traas et al., 1998; Kato and Lam, 2003; Cookson et al., 2006; Berr and Schubert, 2007; Roeder et al., 2010; Adachi et al., 2011; Massonet et al., 2011; Schubert et al., 2012; Del Pozo and Ramirez-Parra, 2015; Sliwinska et al., 2015), Gossypium (Guan et al., 2014; Snodgrass et al., 2017) and Nicotiana (Renny-Byfield et al., 2011; McCarthy et al., 2016). However, research emphasis has consistently been placed firmly on mechanisms of ploidy change that affect whole organisms and thereafter characterize the entire evolutionary lineage. Here, we focus on genome duplication events that occur in only some cells of the affected organisms and are not directly heritable (though the ability to generate cells with expanded nuclei evidently is heritable). Only recently have the potentially profound consequences of duplication events affecting only specific cells or tissues been recognized, and most of the insights thus far have emerged from the homogenized tissues necessary for flow cytometric analysis. We also have utilized flow cytometry, but in an unusually precise fashion that targeted localized epidermal regions of a single organ. One innovation reported here is that we have combined flow cytometry with confocal microscopy, which permitted mapping of somatic ploidy that was both direct and unusually finely targeted on specific tissues. Nature and terminology of endoreplication Also known as endoreduplication, the term endoreplication describes replication of the nuclear genome occurring within the nuclear envelope in the absence of spindle formation or subsequent cell division, thereby elevating (typically doubling) the DNA content of the affected cell (Nagl, 1978; Galbraith et al., 1991; Breuer et al., 2010, 2014; Edgar et al., 2014). Two contrasting routes to endoreplication are widely recognized (e.g. Gutierrez et al., 2014; Scholes and Paige, 2015) (Fig. 1). In endomitosis, the cell undergoes partial mitosis but fails to execute telophase or cytokinesis, consequently forming lobed or even multiple nuclei within a single cell if mitosis has progressed sufficiently. By contrast, in endocycling the cell avoids mitosis, so that physical connection often persists between duplicated chromatids. Thus, the chromatids can eventually form large, structurally complex polytene chromosomes if the sister chromatids remain tightly configured to form parallel arrays (Huskins, 1947; Pearson, 1974; Zhimulev and Koryakov, 2009; Maluszynska et al., 2013). In either case, in order to unequivocally qualify as endoreplicated, a cell must bypass the M checkpoint in the cell cycle (Fig. 1), a definition that automatically encompasses all cells exceeding 4C (mitotically dividing cells cycle routinely between 2C and 4C DNA amounts). Fig. 1. View largeDownload slide Major phases, checkpoints and potential short-cuts in the standard mitotic cell cycle, illustrating terminology relevant to endoreplication. Fig. 1. View largeDownload slide Major phases, checkpoints and potential short-cuts in the standard mitotic cell cycle, illustrating terminology relevant to endoreplication. The textbook epitome of well-understood polytene chromosomes is zoological rather than botanical – specifically, the salivary gland of the fruit fly Drosophila, wherein multiply duplicated genes greatly increase the transcriptional rate of late-larval ‘glue’, thereby facilitating pendent pupation (e.g. Rodman, 1967; Zhimulev et al., 2012). Since then, zoological interest has expanded rapidly (reviewed by Neiman et al., 2017), reaching as far as biomedicine (e.g. Biesele and Poyner, 1943; Vitrat et al., 1998; Cremer and Cremer, 2001). The mitotic cell cycle is traditionally divided into a series of phases and checkpoints (e.g. Polyn et al., 2015) (Fig. 1); in plants it is determined primarily by antagonism between auxins and cytokinins (notably cyclin-dependent kinase). On average, larger cells cycle more slowly. Genome size sets a minimum duration of cell cycle and a minimum cell size. Thus, both cell size and replication time are causally positively correlated with genome size, but not with each other. In both plants and animals, endoreplication often leads to increases in cell size/complexity and in transcriptional rates/levels, reflecting massed gene duplications (Traas et al., 1998; Sugimoto-Shirasu and Roberts, 2003; Magyar et al., 2012). Visual detection of endoreplication events is less straightforward than might be imagined. Doubling of genome size would from first principles be predicted to double the nuclear volume but to expand the radius by only 26 % (volume = 4/3πr3) – a comparatively subtle increase that can prove difficult to detect in two-dimensional images, particularly since in practice the nuclei often deviate from perfect spheres and carry differentially condensed chromatin. Similar problems exist when measuring cell sizes. Nevertheless, a broadly allometric positive correlation has been demonstrated between genome size and stomatal guard-cell size in the leaves of phylogenetically broad samples of both angiosperms (‘flowering plants’: Beaulieu et al., 2008) and gymnosperms (Lomax et al., 2014). Endoreplication in plant tissues can more easily be detected using flow cytometry, provided sufficient endoreplicated cells are present in the targeted tissue. By applying flow cytometry to a wide range of orchid species, Travnicek et al. (2015) were able to demonstrate the occurrence of the under-researched phenomenon of progressive partial endoreplication (PPE), which remains poorly understood and to date has only been observed in orchids. During most reported cases of PPE, endoreplication affects not the entire nuclear genome but rather the great majority of it, such that the genome undergoes stepwise increases of less than double (Jersakova et al., 2015; Travnicek et al., 2015; Hribova et al., 2016). Brown et al. (2017, their Table 3) subsequently summarized data on PPE in a diverse suite of 75 orchid taxa and reported replicate fractions that ranged from 19 to 105 %, predictably generating a large variance around the mean value for all taxa. All Orchidoideae, most Cypripedioideae and a significant minority of Epidendroideae were affected. Similarly major effects of PPE had earlier been observed in model animals such as Drosophila (e.g. Bosco et al., 2007). Present study Flowers of cultivated azaleas were shown by Schepper et al. (2001) to be fundamentally diploid but nonetheless to contain cell lineages that often switched via endopolyploidy to tetraploidy towards the margin of the flower, which also diverged from the central region of the perianth in both cellular structure and anthocyanin pigmentation. Kudo and Kimura (2002) observed a range of endopolyploid levels from 2C to 32C in contrasting regions of the petals of cabbage, speculating that the larger proximal epidermal cells contribute to rapid petal expansion in the opening bud. Similar investigations have since been performed on orchid flowers – an intensively researched family that therefore has the potential to provide a stronger and more rounded model system for studying the evolutionary significance of ploidy change and genome duplication. However, all of these previous orchid studies (1) treated each organ of the flower as a homogeneous entity and (2) relied upon indirect estimates of nucleus size obtained via flow cytometry of homogenized tissue samples, rather than through direct microscopic observation of individual cells (cf. Mishiba et al., 2001; Lim and Loh, 2003; Yang and Loh, 2004; Barow and Jovtchev, 2007; Chen et al., 2011; Teixeira et al., 2014; Travnicek et al., 2015; Ho et al., 2016; reviewed by Hribova et al., 2016). Such studies are unable to distinguish between the two contrasting modes of endoreplication (endomitosis and endocycling), though they do offer the newly recognized advantage of being able to detect all but the most subtle cases of PPE. Here, we combine the analytical power of flow cytometry to detect PPE and quantify levels of endoreplication with the anatomical precision and cytostructural insights offered by direct microscopical examination, using in particular confocal microscopy to explore endoreplication within the labellum, a specialized insect-attracting petal that is characteristic of orchids (e.g. Rudall and Bateman, 2002; Mondragón-Palomino and Theissen, 2011). The orchid labellum is an exceptionally complex floral structure that typically exhibits many spectacular adaptations at both the macroscopic and the cellular scale (e.g. Box et al., 2008; Bradshaw et al., 2010). It therefore allows simultaneous comparison of an unusually large number of biological features that might influence, or indeed be influenced by, endoreplication. Two model systems of European orchids in subfamily Orchidoideae subtribe Orchidinae were selected by us to contrast strongly in both phenotypic complexity and predisposition to polyploid speciation (Table 1). The genus Dactylorhiza has a diploid complement of 2n = 2x = 40, develops a labellum that is micromorphologically relatively simple, indulges in no known secretion of either nectar or scent, and frequently undergoes polyploid (especially allopolyploid) speciation (e.g. Pillon et al., 2007; Paun et al., 2010; Hedrén et al., 2011). In contrast, the primary subject of this study, bee orchids of the genus Ophrys, have a diploid complement of 2n = 2x = 36 and, on present (limited) evidence, rarely undergo polyploid speciation. However, they reliably possess a labellum that is exceptionally micromorphologically complex (Fig. 2); it exhibits several specialized epidermal cell types that contrast in both appearance and function (e.g. Bradshaw et al., 2010). Table 1. Details of staining and microscopy methods applied to each orchid species studied Species Clade 2n Staining method Microscopy method PI (all DNA, red) Feulgen (all DNA, dark grey) DAPI (A+T-rich DNA, blue) FISH (18S rDNA, green) Confocal Light Fluorescence Dactylorhiza fuchsii A 40 ✓ ✓ Dactylorhiza foliosa A 40 ✓ ✓ Dactylorhiza praetermissa AB 80 ✓ ✓ Ophrys insectifera 1 36 ✓ ✓ ✓ ✓ Ophrys speculum 2 36 ✓ ✓ ✓ ✓ Ophrys tenthredinifera s.l. 2 36 ✓ ✓ Ophrys apifera 3 36 ✓ ✓ ✓ ✓ Ophrys sphegodes 3 36 ✓ ✓ ✓ Species Clade 2n Staining method Microscopy method PI (all DNA, red) Feulgen (all DNA, dark grey) DAPI (A+T-rich DNA, blue) FISH (18S rDNA, green) Confocal Light Fluorescence Dactylorhiza fuchsii A 40 ✓ ✓ Dactylorhiza foliosa A 40 ✓ ✓ Dactylorhiza praetermissa AB 80 ✓ ✓ Ophrys insectifera 1 36 ✓ ✓ ✓ ✓ Ophrys speculum 2 36 ✓ ✓ ✓ ✓ Ophrys tenthredinifera s.l. 2 36 ✓ ✓ Ophrys apifera 3 36 ✓ ✓ ✓ ✓ Ophrys sphegodes 3 36 ✓ ✓ ✓ View Large Table 1. Details of staining and microscopy methods applied to each orchid species studied Species Clade 2n Staining method Microscopy method PI (all DNA, red) Feulgen (all DNA, dark grey) DAPI (A+T-rich DNA, blue) FISH (18S rDNA, green) Confocal Light Fluorescence Dactylorhiza fuchsii A 40 ✓ ✓ Dactylorhiza foliosa A 40 ✓ ✓ Dactylorhiza praetermissa AB 80 ✓ ✓ Ophrys insectifera 1 36 ✓ ✓ ✓ ✓ Ophrys speculum 2 36 ✓ ✓ ✓ ✓ Ophrys tenthredinifera s.l. 2 36 ✓ ✓ Ophrys apifera 3 36 ✓ ✓ ✓ ✓ Ophrys sphegodes 3 36 ✓ ✓ ✓ Species Clade 2n Staining method Microscopy method PI (all DNA, red) Feulgen (all DNA, dark grey) DAPI (A+T-rich DNA, blue) FISH (18S rDNA, green) Confocal Light Fluorescence Dactylorhiza fuchsii A 40 ✓ ✓ Dactylorhiza foliosa A 40 ✓ ✓ Dactylorhiza praetermissa AB 80 ✓ ✓ Ophrys insectifera 1 36 ✓ ✓ ✓ ✓ Ophrys speculum 2 36 ✓ ✓ ✓ ✓ Ophrys tenthredinifera s.l. 2 36 ✓ ✓ Ophrys apifera 3 36 ✓ ✓ ✓ ✓ Ophrys sphegodes 3 36 ✓ ✓ ✓ View Large Fig. 2. View largeDownload slide Scanning electron micrograph of the labellum of Ophrys speculum with enlargements of three micromorphologically variable regions of the adaxial epidermis that were examined for endoreplication during the present study (speculum, appendix, median/mid-lobe and lateral lobe hirsute margins) and one (the stigmatic surface) that was not (see also Bradshaw et al., 2010). As O. speculum lacks an apical appendix, the appendix shown represents that of O. apifera (see also Fig. 9A). Scale bar: (main figure) = 2 mm. Fig. 2. View largeDownload slide Scanning electron micrograph of the labellum of Ophrys speculum with enlargements of three micromorphologically variable regions of the adaxial epidermis that were examined for endoreplication during the present study (speculum, appendix, median/mid-lobe and lateral lobe hirsute margins) and one (the stigmatic surface) that was not (see also Bradshaw et al., 2010). As O. speculum lacks an apical appendix, the appendix shown represents that of O. apifera (see also Fig. 9A). Scale bar: (main figure) = 2 mm. When the labellum of Ophrys is viewed from a functional perspective, contrasting labellar cell types are collectively responsible for seducing naive male insects into participating in this textbook example of sexually deceptive pollination. The labellum first secretes a cocktail of pseudopheromones as an olfactory cue, then offers a complex mosaic of brightly coloured and reflective regions as visual cues, and finally stimulates the now alighted insect with strategically placed and contrasting patches of trichomes – tactile cues that differ between species in size, shape and/or orientation and therefore supposedly mimic the conspecific female insect sufficiently well to encourage pseudocopulation (e.g. Schiestl, 2005; Bateman et al., 2011; Vereecken et al., 2011; Vignolini et al., 2012). In its attempt to mate with the flower, the naive male removes the cohesive pollen masses from the gynostemium – the fused hermaphrodite structure that defines the orchid family, providing the location of meiosis for both pollen and ovule production. The primary goals initially set for the present study were to use somatic ploidy mapping: (1) To assess, via confocal microscopy applied to the labellar petals of multiple representatives of two orchid genera that contrast strongly in cellular complexity, the frequency and location of endoreplicated nuclei, and to estimate the number of endocycles that each has undergone. (2) To determine in epidermal cells whether the presence and size (average and range distribution) of endoreplicated nuclei are associated with, and thus could have helped to determine, one or more of the following three features: size and/or complexity of cells; location of cells within a specific organ, the labellum (especially central versus marginal locations); presumed overall level of transcriptional activity within cells (notably energy-intensive secretions of nectar and/or fragrances). (3) To determine whether PPE occurs in Ophrys labella and whether it can be detected via direct microscopic observation (this goal was added post hoc in the wake of the recognition through flow cytometry of the existence of PPE in orchids by Travnicek et al., 2015). (4) To begin to explore the external morphology and internal structure of endoreplicated (including polytene) nuclei in orchids, permitting the distinction between the products of endomitosis and endocycling. (5) To use somatic ploidy changes observed in the labella as a proxy for endoreplication within the gynostemium – the site of both ovule and pollen formation in the orchid flower, and thus of the unreduced gametes necessary for polyploid speciation in Dactylorhiza. [This goal was effectively undermined after our laboratory study had concluded, when it was shown that the gynostemium is the one organ in an orchid flower that is not prone to endoreplication (Travnicek et al., 2015.)] In retrospect, we are satisfied that we met objectives (1) to (3), made some progress with objective (4), but were naive to believe that we could achieve objective (5). MATERIALS AND METHODS Materials The three species of Dactylorhiza that we selected for detailed study consist of a diploid that is endemic to the Atlantic island of Madeira (D. foliosa), a much more geographically widespread diploid that has acted as the ovule-parent in a series of speciation events driven by allopolyploidy (D. fuchsii), and one of the allotetraploid species (D. praetermissa) that D. fuchsii has generated in combination with the diploid pollen-parent D. incarnata (e.g. Hedrén et al., 2011). The genus Ophrys is exceptionally taxonomically controversial; some traditional morphological and ethological studies recognize more than 350 species (cf. Delforge, 2006; Vereecken et al., 2011; Paulus, 2015; Delforge, 2016), whereas molecular phylogenies suggest ten or fewer species (cf. Devey et al., 2008; Bateman et al., 2011), a pattern maintained even when next-generation sequencing technologies are employed (Fig. 3) (Bateman et al., 2018). We selected five relatively uncontroversial species to represent each of the three subgenera of Ophrys: the monotypic subgenus insectifera plus morphologically and molecularly divergent species pairs from the two remaining subgenera (Table 1, Fig. 3). Fig. 3. View largeDownload slide Backbone phylogeny of the genus Ophrys based on RAD-Seq data, showing the relationships of the nine macrospecies recognized by Bateman et al. (2018), each of which is illustrated to the right. The five macrospecies included in the present study are highlighted in red, and the three main clades here regarded as subgenera are numbered in red. Black numbers below branches indicate all bootstrap support values. Scale: the long axis of each rectangular flower image = 25 mm. Fig. 3. View largeDownload slide Backbone phylogeny of the genus Ophrys based on RAD-Seq data, showing the relationships of the nine macrospecies recognized by Bateman et al. (2018), each of which is illustrated to the right. The five macrospecies included in the present study are highlighted in red, and the three main clades here regarded as subgenera are numbered in red. Black numbers below branches indicate all bootstrap support values. Scale: the long axis of each rectangular flower image = 25 mm. Fresh inflorescences were sampled from either the living collections of the Royal Botanic Gardens, Kew, for Dactylorhiza, and from either natural populations in the UK and the Mediterranean or the private seed-raised horticultural collection of Barry Tattersall for Ophrys. Plant material intended for microscopic examination was placed immediately in fixative (3:1 glacial acetic acid:ethanol) and later refrigerated at 4 °C. Fresh material for analysis via flow cytometry was simply placed in damp tissue. Staining methods Four different staining techniques, each providing different cytological emphases, were employed for contrasting purposes (Table 1). Non-fluorescent Feulgen (Schiff reagent) staining provided the focus for preliminary observations. Of the three remaining, uniformly fluorescent stains employed, propidium iodide (PI) stains DNA uniformly, whereas 4,6-diamidino-2-phenylindole (DAPI) stains predominantly A+T-rich regions. Fluorescent in situ hybridization (FISH) was used to label 18S ribosomal DNA (rDNA) in order to determine whether signals were dispersed or alternatively clustered as an indication of polytene chromosomes. Although fluorescent stains are less convenient than Feulgen stains, they allowed us to pursue both FISH experiments and confocal examination on the same preparation. Cold hydrolysis was employed in the Feulgen procedure to encourage the stain to pass through the cell membrane and reach the nucleus. Inflorescences were immersed in 5 m HCl in test tubes placed in a water bath at 20 °C for 30 min. Samples were then subjected to four 5-min washes in distilled water on a low-speed rotator. Flowers were subsequently stained in Feulgen tubes, enclosed in aluminium foil, and placed in the dark for 30 min. Individual labella were then excised from the stained inflorescence, placed on glass slides, mounted under coverslips and sealed using vulcanizing rubber solution. For PI staining, labella were immediately excised from the fixed inflorescence and passed through four 5-min washes in distilled water on a low-speed rotator before being placed on a glass slide. The PI was diluted 1:1000 in distilled water before application. Staining with DAPI followed a broadly similar protocol but was confined to O. speculum and O. apifera (Table 1). One drop (~25 μL) of photoreactive DAPI, dissolved in Vectashield® to preserve fluorescent reactivity, was added with care in subdued light, and the labellum coverslipped. Any substandard DAPI slides were washed in the non-ionic detergent Tween® before being re-stained with PI. The more complex FISH procedure was adapted from Lim et al. (2006). The protocol requires a three-day preparatory period in total [a ‘standard wash’ is here defined as a 5-min wash at 37 °C in 2× saline sodium citrate buffer (SSC; 3 m NaCl, 0.3 m sodium citrate, pH7)]. Each labellum was quartered, and each quarter was placed in a 1.5-mL microtube. Samples received two 5-min washes in citric buffer (4 mm citric acid plus 6 mm sodium citrate) followed by enzyme digestion in 0.3 % w/v cellulase R10, 0.3 % w/v pectolyase Y23 and 0.3 % w/v Driselase in enzyme buffer for 20 min, and transfer to enzyme buffer for 2 h at room temperature. Samples then received two further 5-min washes before being exposed to 0.12 mg mL−1 RNase in 2× SSC at 37 °C for 60 min. Three further standard washes rendered the labellar fragments fragile, so they were rinsed with care in 0.01 m HCl for 2 min followed by pepsin treatment: 0.1 μg mL−1 pepsin in 0.01 m HCl for 5 min at room temperature. Material was then fixed in 2 % formaldehyde in water at room temperature for 10 min and washed in 2× SSC in preparation for probe hybridization. The petal fragments were then immersed in the hybridization mix overnight. This mix contained the 18S rDNA probe, which had been isolated by PCR amplification of 18S rDNA from Allium cernuum using PCR primers (18S2F 59-CGGAGAATTAGGGTTCGATTC-39 and AB101R 59-ACGAATTCATGGTCCGGTGAAGTGTTCG-39) and labelled with digoxigenin-11-dUTP by nick translation, as described by Renny-Byfield et al. (2012). The hybridization mix contained 4 µg mL−1 labelled 18S rDNA probe in 50 % w/v formamide, 10 % w/v dextran sulphate and 0.1 % w/v sodium dodecyl sulphate in 2× SSC. After overnight incubation, the samples were denatured in a water bath at 70 °C for 3 min, followed by a further phase of hybridization for 24 h at 37 °C. Material was then given three standard washes plus two 5-min, 42 °C washes in a mixture of 20 % formamide in 0.1× SSC. Sites of probe hybridization were then detected by immersing the material in 20 mg mL−1 fluorescein conjugated anti-digoxigenin immunoglobulin G (Roche Biochemicals) in 2× SSC for 4 h. Three further standard washes preceded a wash in 4× SSC/0.2 % Tween at room temperature for 5 min. The surviving tissues were then mounted in subdued light on glass slides in a single drop of DAPI in Vectashield®, together with a single drop of PI as a counterstain. All prepared slides were stored in a refrigerator at 4 °C prior to microscopic examination. Microscopy methods Feulgen-stained slides were observed under standard compound light microscopy and digitally imaged at magnifications ranging from ×10 to ×100, allowing an initial phase of rapid screening. Slides stained in PI and DAPI were then subjected to more detailed examination under a Leica DMRA2 epifluorescence microscope, and images were captured using a Hamamatsu Orca ER camera. There exists a trade-off when measuring genome size through direct microscopic observation rather than automated flow cytometry; a much smaller number of observations is feasible, but in compensation nuclear size can be related to nuclear micromorphology. For each of the three Dactylorhiza species, areas of 100 nuclei were measured from multiple labella using confocal microscopy and Improvision Openlab software. A nuclear volume was calculated from each measured area by assuming that the recorded area represented a circle and the volume a sphere. Cumulative curves of the resulting measures were then constructed in search of comparatively high-angle break points in size distributions that would allow us to assign each nucleus to one of multiple bins in order to construct histograms competent to indicate contrasting ploidy levels (while bearing in mind the possible modifying effects of PPE). Interpretation focused on the proportions of 4C and especially 8C nuclei, which are the smallest nuclei incontrovertibly resulting from endoreplication. The PI-based slides that constituted the backbone of this study, together with the FISH slides of O. sphegodes (Table 1), were in addition transferred to a controlled-environment suite for high-resolution microscopy using a Leica SP5 confocal laser microscope. This approach allowed three-dimensional reconstruction of nuclei, albeit incurring image acquisition times considerably exceeding those necessary for epifluorescence microscopy. In order to detect multiple fluorescence channels simultaneously, successive parallel planes through a specimen were typically captured at intervals of 1 μm, though closer intervals were used when z-stacking successive images collectively representing single nuclei. The resulting sets of high-quality two-dimensional images and stacks were converted into rotatable three-dimensional images (arguably more accurately described as two-and-a-half dimensional images) using Improvision’s Velocity® software. Flow cytometry methods Flow cytometry was introduced into the present study only towards its close. It is in many ways complementary to direct microscopic observation of nuclei; many more nuclei can be measured, arguably with greater accuracy, but at the expense of losing precise knowledge of the spatial relationships of individual nuclei (Supplementary Data). Flow cytometric study was confined to leaves and flowers of two individuals of O. tenthredinifera villosa (a putative microspecies/subspecies confined to the eastern Mediterranean) that were grown from seed wild-collected at two localities on Crete, together with leaf apices of a further two Ophrys species: O. speculum and O.sphegodes incubacea. Analyses were performed separately on two flowers from each plant of O. tenthredinifera, run 2 weeks apart. Each labellum was dissected into six regions for separate analysis according to the dominant micromorphology of the adaxial epidermis: top-left margin, lower-left margin (both dominantly trichomes), appendix (domed), speculum margin, speculum, and stigmatic surface (all typically papillate). It is noteworthy that regrettably, the small excised regions of labellum that were subjected to flow cytometry also encompassed the underlying tissues of the mesophyll and abaxial epidermis, together with any vascular tissues penetrating the mesophyll. Also, given the small areas of tissues excised, counts of nuclei were inevitably smaller than would be ideal. We assessed the nuclear DNA content of each sample by flow cytometry, using interphase nuclei stained with PI and following the one-step procedure described in detail by Doležel et al. (1998). Individual labellum fragments were placed in Petri dishes containing 1 mL of general-purpose isolation buffer (GPB) with 3 % polyvinylpyrrolidone PVP40 (Loureiro et al., 2007) plus leaf tissue of the selected internal standard (Pisum sativum ‘Ctrirad’, 2C = 9.09 pg; Doležel et al., 1998), and each combined sample was diced using a fresh razor blade. Nuclear suspensions were then filtered through a nylon mesh (30 µm pore size) to remove unwanted debris. The filtrate was stained with 1 mg mL−1 PI to a final concentration of 60 µg mL−1. After incubation on ice for 20 min, the relative fluorescence of at least 1000 (typically 5000) particles was recorded using a Partec Cyflow SL3 (Partec, Münster, Germany) flow cytometer fitted with a 100-mW, 532-nm green solid-state laser (Samba, Cobolt, Solna, Sweden). The resulting histograms were analysed with FlowMax software (v2.4, Partec). The individual genomic DNA contents were estimated as 2C values by multiplying the known 2C value of the chosen standard (2C = 1 pg) by the ratio between the mean relative fluorescence intensities of the G1 peak of the Ophrys tissue sample and that of the G1 peak of the standard. Calculation of genome sizes (2C and 4C nuclei, respectively) assumes that 1 pg of unmodified statistical dsDNA represents 978 Mbp (Doležel et al., 1998). We here use 1C to indicate the monoploid genome size, as recommended by Greilhuber et al. (2005). RESULTS AND DISCUSSION Dactylorhiza: contrasting diploidy with allopolyploidy Speciation in European Dactylorhiza occurs dominantly through hybridization plus chromosome doubling (allopolyploidy) between the phylogenetically divergent D. fuchsii and D. incarnata groups (Pillon et al., 2007; Bateman, 2011; Hedrén et al., 2011). Several analytical approaches employed during the last half-century have all indicated multiple independent origins via allopolyploidy of species that are only subtly morphologically distinct, having originated from different yet conspecific parental races that exhibit contrasting habitat preferences (certainly the case in the D. majalis allopolyploid complex). The relevant Dactylorhiza species have a uniformly papillate labellum (Box et al., 2008) and non-secretory spur; operating by food-deceit, they are dominantly pollinated by typically wide spectra of bee species (e.g. Claessens and Kleynen, 2011). We therefore elected to use this genus to test indirectly whether a predisposition to endoreplication might also indicate a predisposition to polyploid speciation. Previous chromosome counts and genetic studies have both demonstrated D. foliosa and D. fuchsii to be diploid (the latter with 2C of 5.94 pg; Aagaard et al., 2005), whereas D. praetermissa is the allotetraploid product of a diploid ovule-parent broadly resembling D. fuchsii and a diploid pollen-parent broadly resembling D. incarnata (e.g. Hedrén et al., 2011). Nonetheless, our analysis of labellar tissues in our three Dactylorhiza study species revealed nuclear size distributions that differed little from expectations of phases G1 or G2 (Fig. 4); we found no evidence of highly reduplicated cells. The distributions of microscopically estimated nucleus size in the three Dactylorhiza species showed less pronounced peaks and troughs than we anticipated (Fig. 5). Nonetheless, the troughs did permit the identification of break points that could be used to delimit bins, and mean values calculated from these bins yielded arithmetically credible estimates of the volumes of 2C, 4C and (in the case of D. praetermissa) <5 % of putative 8C nuclei – presumably the products of small-scale endoreplication. Fig. 4. View largeDownload slide Propidium iodide confocal images of the central region of the labellum of the diploid Dactylorhiza fuchsii (A) and allotetraploid D. praetermissa (B), presented at the same magnification to show the comparatively low variation in the nuclear genome size of the epidermal cells in each species compared with those observed in Ophrys species. Insets show inflorescences of the respective species. Scale bar = 25 μm; long axis of insets = 30 mm. Fig. 4. View largeDownload slide Propidium iodide confocal images of the central region of the labellum of the diploid Dactylorhiza fuchsii (A) and allotetraploid D. praetermissa (B), presented at the same magnification to show the comparatively low variation in the nuclear genome size of the epidermal cells in each species compared with those observed in Ophrys species. Insets show inflorescences of the respective species. Scale bar = 25 μm; long axis of insets = 30 mm. Fig. 5. View largeDownload slide Histogram comparing nuclear volume distributions of the more or less planar labella of the diploid species Dactylorhiza fuchsii (red bars) and D. foliosa (blue bars) with that of the micromorphologically similar but allotetraploid species D. praetermissa (green bars). Arrows indicate the mean size that corresponds with each modal genome size. Fig. 5. View largeDownload slide Histogram comparing nuclear volume distributions of the more or less planar labella of the diploid species Dactylorhiza fuchsii (red bars) and D. foliosa (blue bars) with that of the micromorphologically similar but allotetraploid species D. praetermissa (green bars). Arrows indicate the mean size that corresponds with each modal genome size. In summary, epifluorescence microscopy images of PI-stained labella of the three species of Dactylorhiza studied by us (Fig. 4) revealed uniform nuclear sizes consistent with prior assumptions of their respective ploidy levels (Fig. 5). No clear evidence of endoreplication was found in the two diploid Dactylorhiza species, and only a hint of a low-frequency, low-level endoreplication was detected in the allotetraploid species. Given that the comparative epidermal homogeneity of the labellum is mirrored in the comparative homogeneity of nuclear size, we found no evidence that predisposition to allopolyploid speciation within this genus is positively correlated with predisposition to endoreplication within individual dactylorchid plants. Admittedly, this outcome contrasts with a previous study (Chen et al., 2011) that used flow cytometry to estimate levels of endoreplication in the tropical epidendroid orchid Phalaenopsis aphrodite and found that diploid plants assigned to this species maintained higher overall levels of endoreplication than did corresponding tetraploid plants. Given the comparative homogeneity of patterns of nuclear size variation in Dactylorhiza, we elected to focus our study on the genus Ophrys. Ophrys: presence of partial progressive endoreplication Leitch et al. (2009, their Fig. 1) reported 1C genome size values for 42 species of subfamily Orchidoideae that collectively yielded a mode of 6–7 pg and a mean of 8.4 pg. Four of their 42 data points were derived from unspecified species of Ophrys, which collectively yielded a slightly higher mean of 10.2 pg from a comparatively narrow range of 1C = 9.5–10.8 pg; this figure remains lower than the mean of 18.3 pg/1C calculated by Leitch et al. for aggregated terrestrial species of all orchid subfamilies. Although the two plants analysed by us for flow cytometry and presented in Fig. 6 ostensibly represent the same microspecies (O. villosa, placed molecularly within the macrospecies O. tenthredinifera; Devey et al., 2008), we unexpectedly discovered that the plant that was marginally larger in both vegetative organs and flowers was tetraploid (supposedly an unusual phenomenon within the genus), whereas the somewhat smaller plant yielded the expected diploid result. The labellar results are more readily interpreted when considered in the context of flow cytometry results obtained from leaves of the same two plants (Fig. 6A, D). In particular, the 2C peak that could not be detected with confidence in the labellum data derived from the diploid plant is clearly present, though admittedly modest in size, in the leaf-based histogram. Arithmetic comparison of flow cytometric count peaks over five runs of contrasting regions of the labellum also made clear that these plants routinely show partial replication throughout the labellum; recorded values for the transition from 2C to 4C nuclei demonstrated size increases of 86 ± 3 % rather than the 100 % that would be expected from complete replication of the nuclear DNA. Halving the mean value obtained from the 2C peaks in the flow histogram yields an estimated 1C value of 10.0 ± 0.3 pg – midway within the range previously reported for the genus by Leitch et al. (2009). Fig. 6. View largeDownload slide Fluorescence intensity distributions (a proxy for nuclear genome size) generated by flow cytometry from populations of cells in diploid and tetraploid individuals of Ophrys tenthredinifera villosa. The fluorescence peaks generated by nuclei isolated from leaf tissue of diploid (A) and tetraploid (D) individuals are presented to enable the different endoreplication fluorescence peaks in floral tissues to be correctly interpreted. (B, E) Fluorescence peaks from the trichome-rich lateral lobe of labellum of (B) diploid and (E) tetraploid plants. (C, F) Fluorescent peaks from the labellum appendix in (C) diploid and (F) tetraploid plants. The inverted triangle indicates the expected position of 2C peaks in the lateral lobe and appendix of the labellum; such peaks were absent from the diploid plant. Fig. 6. View largeDownload slide Fluorescence intensity distributions (a proxy for nuclear genome size) generated by flow cytometry from populations of cells in diploid and tetraploid individuals of Ophrys tenthredinifera villosa. The fluorescence peaks generated by nuclei isolated from leaf tissue of diploid (A) and tetraploid (D) individuals are presented to enable the different endoreplication fluorescence peaks in floral tissues to be correctly interpreted. (B, E) Fluorescence peaks from the trichome-rich lateral lobe of labellum of (B) diploid and (E) tetraploid plants. (C, F) Fluorescent peaks from the labellum appendix in (C) diploid and (F) tetraploid plants. The inverted triangle indicates the expected position of 2C peaks in the lateral lobe and appendix of the labellum; such peaks were absent from the diploid plant. Thus, our flow cytometric data suggest that ~14 % of the chromosomal material is lost during endoreplication, offering clear evidence of PPE. Interestingly, 86 % is precisely the fixed fraction reported for PPE in a Lebanese accession attributed to O. fusca by Brown et al. (2017). However, these modest losses of nuclear material are not evident in our small number of nuclear measurements obtained via confocal microscopy. It is therefore possible that whichever elements in the genome fail to replicate during PPE do not affect nuclear diameters as perceived via direct microscopic observation. Further research is urgently required in this area. Ophrys: location and spatial extent of endoreplication Previous detailed micromorphological studies of Ophrys labella using light microscopy, scanning electron microscopy and transmission electron microscopy highlighted a spectacular diversity of cell shapes and sizes, both within and between labella (Bradshaw et al., 2010; Francisco and Ascensão, 2013). These studies showed that the labella of all species examined – even relatively simple, early-divergent species such as O. insectifera and O. speculum – could readily be divided into at least four micromorphologically distinct and acceptably homogeneous regions of the adaxial epidermis suitable for detailed cytometric and microscopic investigation of endoreplication (Fig. 2), the number of distinct epidermal cell types generally increasing in the more evolutionarily derived species. The four regions chosen for study here were (1) the often reflective speculum in the centre of the labellum, (2) the appendix that terminates the labellar apex of most Ophrys species (although this feature was well developed only in O. apifera and O. tenthredinifera among the five species studied here; Table 1, Fig. 3), and the variably papillate–trichomatous margins of (3) the mid-lobe and (4) the lateral lobes, respectively. Future studies could usefully examine other regions, such as the typically papillate stigmatic surface of the gynostemium immediately above the labellar base (Fig. 2), as the stigma should share with the appendix (a putative osmophore) the trait of being secretory and thus presumably constituting a focus of significant gene expression. Preliminary light microscope investigation of Feulgen-stained O. speculum (Fig. 7A) demonstrated an absence of endoreplication in the well-developed speculum located at the centre of the labellum (Fig. 7C) but extensive formation of endoreplicate nuclei in the trichomes (Fig. 7B, D) that form the unusually well-developed hirsute border that extends across both the middle and lateral lobes (Figs 2 and 7A). It also revealed some details of both the diploid and the endoreplicated nuclei, indicating the simultaneous occurrence of at least two kinds of endoreplication: regular nuclei formed through endocycling (Fig. 7F; see also Fig. 9A) and irregular lobed (‘lobulated’) nuclei most likely produced through endomitosis (Fig. 7E, G; see also Fig. 9E). Fig. 7. View largeDownload slide Feulgen light microscopy images of the labellum of Ophrys speculum (A), contrasting the comparative uniformity of nucleus size in the speculum (C) with the greater variability in nucleus size and shape observed along the hirsute margin (B, D). (E–G) Higher-magnification images contrast the likely products of endomitosis (E, G) versus the more spheroidal nuclei thought to result from endocycling (F). The locations on the flowers of the remaining images are shown in (A). Scale bars: (A) = 5 mm; (B) 200 μm; (C, D) 100 μm; (E–G) = 20 μm. Fig. 7. View largeDownload slide Feulgen light microscopy images of the labellum of Ophrys speculum (A), contrasting the comparative uniformity of nucleus size in the speculum (C) with the greater variability in nucleus size and shape observed along the hirsute margin (B, D). (E–G) Higher-magnification images contrast the likely products of endomitosis (E, G) versus the more spheroidal nuclei thought to result from endocycling (F). The locations on the flowers of the remaining images are shown in (A). Scale bars: (A) = 5 mm; (B) 200 μm; (C, D) 100 μm; (E–G) = 20 μm. Ophrysapifera (Fig. 8A) was examined under the confocal microscope using both PI and DAPI staining. Like O. speculum, this species showed no evidence of endoreplication in its comparatively poorly developed speculum (Fig. 8C). Unlike O. speculum, this species possesses a well-developed appendix, but that too showed only modest evidence of endoreplication (Fig. 8B). Had endoreplication been present in the appendix, we would have found it difficult to determine whether its presence reflects the marginal location or physiological activity of the osmophoric appendix, but its comparatively weak expression suggests that neither explanation may apply (but see the flow cytometry evidence for O. sphegodes discussed below). Fig. 8. View largeDownload slide Microscopic images of the labellum of Ophrys apifera (A), contrasting the uniformity of nuclear genome size in the speculum (C) and the apical appendix (B), which is believed to be responsible for much of the pseudopheromone production in the flower, with the much greater variability in size and shape evident along the hirsute margins (D–F). Some of the endoreplicate nuclei occur towards the distal ends of large unicellular trichomes (F) wherever the trichomes are clustered on the labellum. The locations on the flowers of the remaining images are shown in (A). Propidium iodide confocal (C, F) and DAPI confocal (B, D, E). Scale bars: (A) = 2 mm; (B) = 100 μm; (C) = 50 μm; (D) = 500 μm; (E) = 1 mm; (F) = 100 μm. Fig. 8. View largeDownload slide Microscopic images of the labellum of Ophrys apifera (A), contrasting the uniformity of nuclear genome size in the speculum (C) and the apical appendix (B), which is believed to be responsible for much of the pseudopheromone production in the flower, with the much greater variability in size and shape evident along the hirsute margins (D–F). Some of the endoreplicate nuclei occur towards the distal ends of large unicellular trichomes (F) wherever the trichomes are clustered on the labellum. The locations on the flowers of the remaining images are shown in (A). Propidium iodide confocal (C, F) and DAPI confocal (B, D, E). Scale bars: (A) = 2 mm; (B) = 100 μm; (C) = 50 μm; (D) = 500 μm; (E) = 1 mm; (F) = 100 μm. In contrast, the hirsute regions of the labellum margin, rich in unicellular yet highly elongated trichomes, once again proved to be rich in a diverse spectrum of endoreplicate nuclei (Fig. 8D, E). These trichomes were arguably the most complex, and certainly the largest, epidermal cells that we investigated. As well as being larger and more structurally complex than diploid nuclei, the endoreplicate nuclei of O. apifera also exhibited unusual behaviours. Some – particularly those adorning the prominent horn-like lateral lobes – had apparently been squeezed outward until becoming lodged in the narrowly conical apices of the trichomes (Fig. 8D, F), rather than remaining in the much more spacious basal regions of the cells. Similar results were obtained through PI staining of O. insectifera and O. tenthredinifera (results not shown). We sought more quantitative confirmation of these confocal observations in our flow cytometry data. Examples of relative fluorescence histograms for the lateral-lobe trichomes and appendix cells of the diploid and tetraploid plants, respectively, of O. tenthredinifera villosa are given in Fig. 6 (when interpreting this figure, it is important to remember that 2C peaks clearly evident in leaf tissue were far less evident in the corresponding flower tissues). Comparisons between the labella of the two plants analysed, different labella of the same plant and different regions of the same labellum all suggested that broad patterns of nuclear size distributions were reliable but that the quantitative details of those distributions were not, as significant discrepancies were evident at every hierarchical level. Interestingly, contrasting results derived from replicated analyses were most obvious in those labellar regions that are characterized by a papillate epidermis – the stigma (not examined under the confocal microscope) and the speculum. The only confident generalization that can be made here is that the flow cytometry data indicate a greater inclination towards multiple endoreplication events in the appendix than was suggested by confocal examination, levels within the appendix at least matching those evident in the trichomes. This observation encourages us to suspect that the detailed discrepancies reflect the fact that our confocal observations were carefully adjusted to chosen levels in the cellular ‘stratigraphy’ of the labellum, whereas in the case of our flow cytometry samples, size data derived from the adaxial epidermal cells were subordinate in number to those derived from other tissues present in the analysed homogenate. Any such distortion would impact less upon the appendix, a structure that is comparatively small, is composed of comparatively small cells and is more dorsiventrally homogeneous when viewed at a cellular level. Ophrys: genome size and micromorphology of endoreplicated nuclei In terms of levels of endoreplication reached, our flow cytometric data (Fig. 6) indicated that, for the diploid plant of O. tenthredinifera (and irrespective of precise location on the labellum; Fig. 6B, C, E, F), 8C cells were either dominant or were co-dominant with 4C cells; 2C and 16C cells were either infrequent or, less commonly, seemingly absent. Results differed according to precise location on the labellum, obliging us to rely upon the corresponding leaf data for clear evidence of the presence of 2C nuclei (Fig. 6A, D). Similar nucleus size distributions were observed in the tetraploid plant, except that its fundamental ploidy level meant that, strictly, it was 4C cells that were most frequent, whereas clear evidence of a small minority of 16C cells was found only among the trichomes of the lateral labellar lobes (Fig. 6B, E). Similar flow cytometry profiles were obtained from the leaves of the four plants analysed, the only slight deviation being that no clear peak was identified at 16C in the leaves of either O. tenthredinifera (Fig. 6A, D) or O. speculum (not shown). Interestingly, irrespective of the ploidy of the plant, results obtained from leaves matched those obtained from the labellar appendix more closely than those obtained from trichome-rich regions of the labellum (cf. Fig. 6). Closer examination of endoreplicated nuclei through confocal means relied primarily upon PI and FISH preparations of O. sphegodes (Fig. 9). Individual nuclei were selected for more detailed examination and the ploidy level of each was estimated by measuring its nuclear diameter. Still images of nuclei carefully selected by repeatedly pausing movies obtained through z-stacked confocal microscopy (Fig. 9) contrasted nuclei that are diploid (likely to be 2C; Fig. 9A left, B), endoreplicate non-polytene (likely to be 8C; Fig. 9A right) – both clearly possessing single, proportionately sized nucleoli – and endoreplicate polytene (likely to be 16C; Fig. 9D, E). The mean presumed diploid diameter was 21.7 μm. The right-hand nucleus shown in Fig. 9A had a mean diameter of 36.6 μm and that in Fig. 9D of 38 μm, and these are therefore assumed to be 8C, whereas that shown in Fig. 9E averaged 44 μm in diameter and is therefore assumed to be 16C, implying that it had undergone three rounds of endoreplication. No significantly larger nuclei were observed in any Ophrys species, either through confocal microscopy (Figs 7–10) or flow cytometry (Fig. 6). Fig. 9. View largeDownload slide Propidium iodide images of the labellum margin of Ophrys sphegodes (C) selected from within confocal image stacks of reconstructed nuclei contrast nuclei that are diploid at 2C (A left, B), endoreplicate non-polytene at an estimated 8C (A right) (both possessing single, proportionately sized nucleoli) and endoreplicate polytene at an estimated 16C (D, E). (A right) This nucleus is assumed to be endocyclic whereas the remainder of the illustrated endoreplicate nuclei are assumed to be endomitotic. Vertical dimensions of nuclei: (A) left = 21.7 μm, right = 36.6 μm; (B) = 19 μm, (D) = 38 μm, (E) = 44 μm; (C) vertical dimension of image = 25 mm. Fig. 9. View largeDownload slide Propidium iodide images of the labellum margin of Ophrys sphegodes (C) selected from within confocal image stacks of reconstructed nuclei contrast nuclei that are diploid at 2C (A left, B), endoreplicate non-polytene at an estimated 8C (A right) (both possessing single, proportionately sized nucleoli) and endoreplicate polytene at an estimated 16C (D, E). (A right) This nucleus is assumed to be endocyclic whereas the remainder of the illustrated endoreplicate nuclei are assumed to be endomitotic. Vertical dimensions of nuclei: (A) left = 21.7 μm, right = 36.6 μm; (B) = 19 μm, (D) = 38 μm, (E) = 44 μm; (C) vertical dimension of image = 25 mm. Fig. 10. View largeDownload slide FISH confocal images of a fragmented labellum of Ophrys sphegodes, distinguishing diploid (A) from endoreplicate (B, C) nuclei (PI, red fluorescence; FISH, green fluorescence). The diploid nucleus (A) has a single nucleolus containing diffuse 18S rDNA chromatin (FITC, yellow fluorescence) at various intensities, suggesting differential condensation. The most diffuse chromatin is assumed to be the most transcriptionally active. The endoreplicate nuclei (C) are larger, and have a larger nucleolus, with more highly intense 18S rDNA signals, revealing greatly amplified copy numbers of rDNA units resulting from endoreplication. The higher-magnification image (C) shows most clearly 18S rDNA captured at three contrasting condensation states. In (B) and (C), condensed rDNA units surround the nucleolus (red arrow) and envelop diffuse, decondensed (transcriptionally active) rDNA within. Concentrations of rDNA occurring elsewhere in the nucleus (e.g. green arrow) may be inactive and extranucleolar. Scale bars: (A–B) = 20 μm; (C) = 10 μm. Fig. 10. View largeDownload slide FISH confocal images of a fragmented labellum of Ophrys sphegodes, distinguishing diploid (A) from endoreplicate (B, C) nuclei (PI, red fluorescence; FISH, green fluorescence). The diploid nucleus (A) has a single nucleolus containing diffuse 18S rDNA chromatin (FITC, yellow fluorescence) at various intensities, suggesting differential condensation. The most diffuse chromatin is assumed to be the most transcriptionally active. The endoreplicate nuclei (C) are larger, and have a larger nucleolus, with more highly intense 18S rDNA signals, revealing greatly amplified copy numbers of rDNA units resulting from endoreplication. The higher-magnification image (C) shows most clearly 18S rDNA captured at three contrasting condensation states. In (B) and (C), condensed rDNA units surround the nucleolus (red arrow) and envelop diffuse, decondensed (transcriptionally active) rDNA within. Concentrations of rDNA occurring elsewhere in the nucleus (e.g. green arrow) may be inactive and extranucleolar. Scale bars: (A–B) = 20 μm; (C) = 10 μm. Moreover, structural differences are evident between the different categories of nucleus. Diploid nuclei clearly have chromatin concentrated around the periphery (Fig. 10B), whereas the endoreplicate polytene nuclei (Fig. 9D, E) have a much more complex micromorphology that features interchromosomal domains; frayed telomeric regions form a terminal ‘fan’ of chromatids (Fig. 9E, green arrow). There exists a long history of the study of plant polytene chromosomes (Tschermak-Woess, 1956; Barlow, 1974; Gostev and Asker, 1978; Carvalhiera, 2000), but previous reports have been sporadic, have represented few plant families, and have been confined to tissues intimately associated with ovules (Maluszynska et al., 2013). When applied to O. sphegodes, FISH not only readily distinguished diploid from endoreplicate nuclei, but also clearly demonstrated the presence of a single nucleolus around which rDNA transcription was concentrated (Fig. 10). It imaged 18S rDNA at three contrasting condensation states. Condensed rDNA units surrounded the nucleolus, enveloping diffuse, decondensed (i.e. transcriptionally active) rDNA within (red arrows in Fig. 10B, C). By contrast, the rDNA located elsewhere in the nucleus was probably inactive (green arrow in Fig. 10), an overall situation commonly found in plants (Leitch et al., 1992). There was no indication that rDNA was under-replicated and might therefore be a component of the DNA regions that fail to replicate during PPE; it is more likely that PPE reflects failure to replicate some non-functional regions of the nuclear genome. These results make clear that rDNA number has substantially expanded and most likely been completely amplified during endoreplication – an essential prerequisite for mRNA translation to proteins and indicative of higher overall metabolic activity in the endocycled cells. The clear presence of a single nucleolus containing labelled rDNA in both diploid and endoreplicated nuclei suggests that the chromosomes carrying rDNA remain in close proximity, even when clear signs of polyteny are absent. Placing our observations in a broader orchidological context Previous studies of endoreplication in orchids have primarily examined tropical taxa using flow cytometry, with decidedly mixed results. All investigations found evidence of endoreplication in their case-study species, though they differed in which organs of the plant showed the most frequent endoreplication and in what maximum C-value those endoreplicated nuclei could attain (cf. Mishiba et al., 2001; Barow and Meister, 2003; Lim and Loh, 2003; Yang and Loh, 2004; Barow and Jovtchev, 2007; Chen et al., 2011; Teixeira et al., 2014; Travnicek et al., 2015). Collectively, these studies implicate most of the organs that constitute a typical orchid plant: root hair, root, stem, leaf, pedicel, ovary, gynostemium and tepals (i.e. labellum, lateral petals, sepals), though most authors reported that at least some of the organs of their study species appeared to lack endoreplication. The most commonly reported exceptions are the two organs/part-organs that provide the sites of meiosis: ovary (reliably 2C) and pollinium (reliably 1C) (Chen et al., 2011; Travnicek et al., 2015). Reports typically limit observed endoreplication in any particular orchid organ to two or three rounds of duplication, that is, showing a minority of cells attaining either 8C or 16C (though rare 32C nuclei were reported in the intergeneric hybrid ×Doritaenopsis by Mishiba et al., 2001). These results fall well short of the estimated 256C once recorded in the pericarp of Solanum lycopersicum fruits (Yang and Loh, 2004), but are consistent with previous observations on members of several other flowering plant families, such as Fabaceae (Kocova et al., 2016), Brassicaceae, Caryophyllaceae (Agullo-Antón et al., 2013), Cucurbitaceae and Aizoaceae (Travnicek et al., 2015). The most ambitious flow cytometry study on orchids published thus far was conducted by Travnicek et al. (2015), who compared roots, leaves (base and apex), tepals (not differentiated into individual organs, but usually focusing on sepals), ovaries and pollinaria in 48 orchid species: 36 from subfamily Epidendroideae and six from subfamily Orchidoideae (none of them European), plus smaller numbers from the remaining three subfamilies. Clade (i.e. subfamily) membership appeared to be a strong predictor of endoreplicatory behaviour, as it was absent from all tested members of Apostasioideae and Cypripedioideae and from half of the tested members of Epidendroideae. Where endoreplication was observed, it took two forms: ‘classically’ complete (i.e. basic nucleus size doubled) or PPE (basic nucleus size less than doubled; Bory et al., 2008; Trávníček et al., 2012) – a phenomenon thus far observed within the plant kingdom only in a small minority of orchids and one that is presently poorly understood (Travnicek et al., 2015; Hribova et al., 2016). Remarkably, no orchid was observed to exhibit both forms of endoreplication, suggesting plant-wide control of this potentially important mechanistic divergence. According to Travnicek et al. (2015), all tested members of subfamilies Vanilloideae and Orchidoideae showed only partial endoreplication, along with one-sixth of species sampled in subfamily Epidendroideae (mainly tribe Pleurothallideae). The remaining one-third of Epidendroideae reportedly showed no endoreplication of any kind, a conclusion that contrasts with those of all previously published studies of orchids. Lastly, these authors argued that in those orchids that show PPE, tepals can entirely lack G1-phase (i.e. 2C) nuclei. All six Orchidoideae genera examined by Travnicek et al. (2015) showed PPE, though these authors (1) failed to target labella when examining tepals and (2) analysed only non-European species placed phylogenetically outside the limits of subtribe Orchidinae, whereas we analysed only European species confined to subtribe Orchidinae. We can now confirm that, as expected, PPE extends to subtribe Orchidinae in the form of the genus Ophrys, and affects all regions of the labellum (Fig. 6). It appears increasingly likely that the ability to undergo the unusual phenomenon of PPE is clade-specific and hence is phylogenetically constrained. Functional and evolutionary implications of endoreplication Recent insights obtained through genomics have reduced the formerly predominant evolutionary focus on nuclear protein-coding genes, reflecting increasing recognition that potentially all eukaryotes conservatively maintain 15–30 k functional protein-coding genes irrespective of their phenotypic complexity (e.g. Liu et al., 2013). Orchids are no exception – the genome sequence of the classic model orchid Phalaenopsis equestris projected a total of ~29 k genes (Cai et al., 2015) and that of the earliest-divergent orchid Apostasia shenzhenica predicted ~22 k genes (Zhang et al., 2017). Instead, phenotypic complexity shows a far better positive correlation with the amount of non-protein DNA responsible for RNA transcripts and other regulatory elements (Liu et al., 2013). This observation implies a greater evolutionary role for other factors (both endogenous and exogenous) that influence the detailed progress of organismal development – a cycle repeated through time ad nauseam in the relevant phylogenetic lineage. Endogenously, it is likely that mediation of complex epidermal differentiation in plants such as Ophrys is achieved at least partly through influencing production of a minimum of three ‘molecular patterning modules’: (1) the MYB-bHLH-WD40 protein complex; (2) the transmembrane calpain protease DEFECTIVE KERNEL1 (DEK1); and (3) homeodomain leucine zipper (HD-ZIP) class IV transcription factors acting in concert with SIAMESE-related, cyclin-dependent kinase inhibitors. This combination of factors has been shown to be critical to epidermal patterning in several model angiosperms (reviewed by Robinson and Roeder, 2015). Numerous quantitative trait loci have proved capable of modulating successive endocycles to increase resistance to both biotic and abiotic stresses. For example, endoreplication is enhanced by exposure to increased levels of UV-B (Gegas et al., 2014), increased growth temperature (Lee et al., 2007) or more intense herbivory (Scholes and Paige, 2014), in turn enhancing growth rate and yield in crop plants (Breuer et al., 2014). However, the crucial extent to which endoreplication is pre-programmed into ontogeny rather than induced by environmental change remains controversial (Yang and Loh, 2004). The habitat preferences of the orchids in question may be of relevance, given that most tropical orchids are epiphytes and most temperate orchids, though terrestrial geophytes, occupy low-competition soils. In both cases, nutrients and often also water are limited resources, a situation that might confer economic advantage on plants that can increase nuclear size (and thus nuclear products via gene upregulation) without being obliged to expend energy duplicating the remaining components of the cell. Unsurprisingly, the bulk of research into endoreplication in plants has been conducted on that most academically ubiquitous of species, Arabidopsis thaliana (Galbraith et al., 1991, et seq.). Greater frequency and/or number of cycles of endopolyploidy have been shown to increase organ size, not only in Arabidopsis but also in the model epidendroid orchid Phalaenopsis (Ho et al., 2016). Indeed, phenotypic expression of endoreplication is not confined to mere size differences; for example, the complexity of each unicellular but multiply branched trichome that adorns an Arabidopsis leaf is positively correlated with, and potentially dictated by, the number of endoreplications undergone by its nucleus (Folkers et al., 1997; Traas et al., 1998). The range of nuclear sizes found in the hypocotyl and epidermal pavement cells of Arabidopsis leaves precisely matched that observed by us in bee orchid labella (2C–16C), though Arabidopsis trichomes can exceptionally reach 64C (Traas et al., 1998). In contrast, no correlation was observed between degree of endoreplication and length of collet (hypocotyl) root hairs in a later study of six mutant strains of Arabidopsis (Sliwinska et al., 2015), weakening any attempt on our part to draw general conclusions. Traas et al. (1998) argued that, at least in the case of Arabidopsis pavement cells and trichomes, endoreplication represents a response to both internal and external stimuli. More startlingly, they also claimed that successive endoreplication events occurring within the same cell lineage can consistently reflect different combinations of factors. If both of these statements prove upon further evidence to be valid, endoreplication has the potential to constitute a family of complex processes and influences that together are capable of exceptionally rapid and fine-tooled responses during organismal development. There may even be a requirement for a degree of environmental predictivity by the affected cell, as current evidence suggests that endocycles determine the final size and shape of the affected cell but cease early in cell development, before the large central vacuole has formed and certainly before the bulk of cell expansion has occurred. Thus, on the basis of their observations on the model epidendroid orchids Phalaenopsis and Oncidium, Ho et al. (2016; see also Lee et al., 2004) argued that once endoreplication in an orchid flower ceases, so does its ability to further expand through cell division. Traas et al. (1998, p. 500) therefore warned that the correlation between nucleus size and cell size/complexity would most likely be clearest in those plant cell types that are the least physically constrained, such as trichomes and root hairs, noting that ‘in expanding tissues, growth of neighbouring cells must be tightly coordinated in order to avoid local distortions of tissues’. Similar observations were made by Guan et al. (2014) on the trichome-homologue cotton fibres of Gossypium. We speculate that such ‘local distortions of cells’, driven by an appropriate number of endoreplication cycles, would form a ready explanation for the unusual degree of three- dimensionality that characterizes most orchid labella, including the deep invagination that forms the labellar spur in orchids such as Dactylorhiza (Box et al., 2008; Bell et al., 2009) and the overall convexity plus horn-like and/or breast-like ‘evaginations’ that characterize the labellum of most Ophrys species (Bradshaw et al., 2010). Our previous observations suggest that both sets of features emerge late in flower development. We introduced this paper by outlining the multiple adaptations present in Ophrys that are assumed to be required for efficient pseudocopulatory pollination. The labellum first secretes a cocktail of pseudopheromones as an olfactory cue, then offers a complex mosaic of brightly coloured and reflective regions as visual cues, and finally further stimulates the now alighted insect with strategically placed and contrasting patches of trichomes – tactile cues that differ in size, shape and/or orientation and therefore supposedly mimic the conspecific female insect. It seems reasonable to assume that the number of endoreplications undergone by nuclei in the labellum influences the quantities generated of the many biochemicals that they are competent to produce, even if the relationship between genome size and gene products is not precisely arithmetic. If so, it becomes feasible to hypothesize that the epidermal gradation evident in the Ophrys labellum from flat pavement cells through domed and papillate cells to short straight-sided trichomes and longer spiral trichomes could also reflect concentration-dependent biochemical interactions in different epidermal regions. In other words, differential endoreplication could, through dosage effects, dictate differentiation of the panoply of subtly variable visual and tactile cues offered by the bee orchid labellum. In particular, overexpression induced by highly localized endoreplication could substantially shift the concentrations (both relative and absolute) of the biochemicals present in the cocktail of pseudopheromones emitted by Ophrys flowers – changes that are likely to have at least some impact on the spectrum of pollinating insects attracted to the flowers (cf. Gögler et al., 2009; Breitkopf et al., 2013; Sedeek et al., 2014, 2016). Admittedly, demonstrating such an effect would not necessarily resolve current impassioned debates regarding the nature of species and speciation within the genus; opinions already expressed regarding the downstream consequences of such pollinator shifts have ranged from greatly enhanced speciation rate driven by novel pollinator specificities (e.g. Vereecken et al., 2011; Paulus, 2015) to prevention of speciation through enhanced gene flow (e.g. Bateman et al., 2011; Hennecke et al., 2015). Superimposed on continuing uncertainty regarding the nature and significance of endoreplication per se is our almost complete ignorance regarding the nature of progressively partial endoreplication (Fig. 6). Does this phenomenon occur beyond the bounds of the orchid family? Which modestly sized portions of the orchid genome fail to copy during the genome replication process, and why? Hribova et al. (2016, p. 2003) concluded that ‘the mechanism behind PPE is the incomplete replication of nuclear DNA. Together with the precise control of the extent of DNA under-replication, our results indicate that PPE is a highly controlled process accompanying cell and tissue differentiation’. They then ‘hypothesize[d] that PPE is part of a highly controlled transition mechanism from proliferation phase to differentiation phase of plant tissue development’ (p. 1996). The results obtained from Ophrys labella during the present study are consistent with, though by no means conclusively demonstrate, the views recently expressed by Hribova et al. Perhaps the most interesting of the many as yet unanswered questions is whether endoreplication does indeed offer selective advantages or is merely the happenstance product of relaxation in developmental control – relaxation occurring towards the end of the ontogenetic trajectories that add the final details to determinate, disposable organs such as petals. But in practice the converse argument may apply. The ability to determine the approximate levels of endoreplication occurring in different specified regions of a single tissue (in this case the epidermis) in a single organ (in this case the labellar petal) could be taken as indicating remarkably subtle and sophisticated (epi)genetic control of development – one that permits the considerable phenotypic plasticity that is evident in so many bona fide orchid species. CONCLUSIONS (1) Endoreplicated nuclei were observed in all five species of Ophrys examined by us, involving both endomitosis and endocycling. In contrast, endoreplicated nuclei were less evident in the corresponding three species of Dactylorhiza. Thus, no link has been demonstrated here between the stronger predisposition to allopolyploid speciation evident in Dactylorhiza and the stronger predisposition to localized endopolyploidy here demonstrated in Ophrys. (2) Endoreplication in the labella of Ophrys species appears to be more strongly positively correlated with cell size/complexity (it especially characterizes trichomes) than with secretory role or marginal location (less evident in the appendix, which is both marginal to the labellum and highly physiologically active, yet yielded nucleus size distributions more consistent with the corresponding leaves). However, this provisional conclusion should be tested further, using even smaller microdissected tissue samples that reliably encompass only a single cell type. (3) Both fluorescence microscopy and flow cytometry revealed three size categories of endoreplicated nuclei in both O. sphegodes and O. tenthredinifera villosa, the polytene nuclei reaching 16C in size (i.e. maximally having undergone three rounds of endoreplication). (4) Progressively partial endoreplication may occur throughout tribe Orchideae within the orchid family, indicating strong phylogenetic control of the underlying mechanism. It is now essential to determine which portion(s) of the orchid genome are escaping replication during PPE events. (5) The possibility therefore exists for epigenetic control of gene overexpression via local induction of endoreplication in particular tissues. If so, endoreplication should be viewed as an important element in the epigenetic palette available to a plant, and a possible explanation of the plastic responses that are being observed with increasing frequency in plants. The evolutionary-developmental significance of endoreplication may thus far have been massively underestimated by the biological community. (6) Combining flow cytometry with confocal microscopy represents a powerful approach to determining the nature and scale of endoreplication within organisms. Direct visualization through confocal microscopy allows observations to be focused precisely within target tissues and to include the nanomorphology of the nucleus, whereas indirect measurement via flow cytometry provides fully quantitative size distributions. In future studies, we would aim to microdissect and then bulk up the organs of interest so that the flow cytometry results can be more rigorously confined to the target tissue. SUPPLEMENTARY DATA Supplementary data are available online at https://academic.oup.com/aob. They consist of text relating to measurement of nuclear size, comparing flow cytometry with direct microscopic observation. The text incorporates a table (Table S1) of observed and estimated mean nuclear peaks in the confocal size distributions of nuclei located in the labella of the diploid D. fuchsii and the allotetraploid D. praetermissa analysed in parallel with the diploid D. foliosa. It also incorporates a figure (Fig. S1) showing fluorescence histograms of leaf material in diploid (A, B) and tetraploid (C, D) cytotypes of O. tenthredinifera villosa. ACKNOWLEDGEMENTS We thank Michaela Egertova for guidance on the use of the confocal microscope, Heike Brinkman for additional technical support, and Thomas Cremer and Rachel Walker for sharing insights from their own related confocal research. We are grateful to Catalina Romila for measuring the nuclei of Dactylorhiza fuchsii and D. praetermissa, and Beth Bradshaw for capturing some of the scanning electron micrographs inset into Fig. 2. Barry Tattersall kindly provided material for flow cytometry. The London-focused Centre for Ecology and Evolution provided a modest but invaluable pump-priming grant in autumn 2012 (awarded to P.J.R and A.R.L.) that funded the 2013 summer internship of J.J.G. LITERATURE CITED Aagaard SMD , Sastad SM , Greilhuber J , Moen A . 2005 . A secondary hybrid zone between diploid Dactylorhiza incarnata ssp. cruenta and allotetraploid D. lapponica (Orchidaceae) . Heredity 94 : 488 – 496 . Google Scholar CrossRef Search ADS PubMed Adachi S , Minamisawa K , Okushima Y , et al. 2011 . Programmed induction of endoreduplication by DNA double-stranded breaks in Arabidopsis . Proceedings of the National Academy of Sciences of the USA 108 : 10004 – 10009 . Google Scholar CrossRef Search ADS PubMed Agullo-Antón MA , Olmos E , Pérez-Pérez JM , Acosta M . 2013 . Evaluation of ploidy level and endoreduplication in carnation (Dianthus spp.) . Plant Science 201–202 : 1 – 11 . Google Scholar CrossRef Search ADS PubMed Barlow PW . 1974 . The polytene nucleus of the giant hair cell of Bryonia anthers . Protoplasma 83 : 339 – 349 . Google Scholar CrossRef Search ADS Barow M , Jovtchev G . 2007 . Endopolyploidy in plants and its analysis by flow cytometry . In: Dolezel J , Greilhuber J , Suda J , eds. Flow cytometry with plant cells: analysis of genes, chromosomes and genomes . Weinheim : Wiley , 349 – 372 . Google Scholar CrossRef Search ADS Barow M , Meister A . 2003 . Endopolyploidy in seed plants is differently correlated to systematics, organ, life strategy and genome size . Plant, Cell and Environment 26 : 571 – 584 . Google Scholar CrossRef Search ADS Bateman RM . 2011 . Glacial progress: do we finally understand the narrow-leaved marsh orchids ? New Journal of Botany 1 : 2 – 15 . Google Scholar CrossRef Search ADS Bateman RM , Bradshaw E , Devey DS , et al. 2011 . Species arguments: clarifying concepts of species delimitation in the pseudo-copulatory orchid genus Ophrys . Botanical Journal of the Linnean Society 165 : 336 – 347 . Google Scholar CrossRef Search ADS Bateman RM , Paun O , Sramkó G . 2018 . Integrating restriction site-associated DNA sequencing (RAD-Seq) with morphological cladistic analysis clarifies evolutionary relationships among major species groups of bee orchids . Annals of Botany 121 : 85 – 105 . Google Scholar CrossRef Search ADS PubMed Beaulieu JM , Leitch IJ , Patel S , Pendharkar A , Knight CA . 2008 . Genome size is a strong predictor of cell size and stomatal density in angiosperms . New Phytologist 179 : 975 – 986 . Google Scholar CrossRef Search ADS PubMed Bell AK , Roberts DL , Hawkins JA , Rudall PJ , Box MS , Bateman RM . 2009 . Comparative morphology of nectariferous and nectarless labellar spurs in selected clades of subtribe Orchidinae (Orchidaceae) . Botanical Journal of the Linnean Society 160 : 369 – 387 . Google Scholar CrossRef Search ADS Berr A , Schubert I . 2007 . Interphase chromosome arrangement in Arabidopsis thaliana is similar in differentiated and meristematic tissues and shows a transient mirror symmetry after nuclear division . Genetics 176 : 853 – 863 . Google Scholar CrossRef Search ADS PubMed Biesele JJ , Poyner H . 1943 . Polytene chromosomes in two mammary carcinomas of the human subject . Cancer Research 3 : 779 – 783 . Bory S , Catrice O , Brown S , et al. 2008 . Natural polyploidy in Vanilla planifolia (Orchidaceae) . Genome 51 : 816 – 826 . Google Scholar CrossRef Search ADS PubMed Bosco G , Campbell P , Leiva-Neto JT , Markow TA . 2007 . Analysis of Drosophila species genome size and satellite DNA content reveals significant differences among strains as well as between species . Genetics 177 : 1277 – 1290 . Google Scholar CrossRef Search ADS PubMed Box MS , Bateman RM , Glover BJ , Rudall PJ . 2008 . Floral ontogenetic evidence of repeated speciation via paedomorphosis in subtribe Orchidinae (Orchidaceae) . Botanical Journal of the Linnean Society 157 : 429 – 454 . Google Scholar CrossRef Search ADS Bradshaw E , Rudall PJ , Devey DS , Thomas MM , Glover BJ , Bateman RM . 2010 . Comparative labellum micromorphology in the sexually deceptive temperate orchid genus Ophrys: diverse epidermal cell types and multiple origins of structural colour . Botanical Journal of the Linnean Society 162 : 502 – 540 . Google Scholar CrossRef Search ADS Breitkopf H , Schlüter PM , Xu S , Schiestl FP , Cozzolino S , Scopece G . 2013 . Pollinator shifts between Ophrys sphegodes populations: might adaptation to different pollinators drive population divergence ? Journal of Evolutionary Biology 26 : 2197 – 2208 . Google Scholar CrossRef Search ADS PubMed Breuer C , Ishida E , Sugimoto K . 2010 . Developmental control of endocycles and cell growth in plants . Current Opinion in Plant Biology 13 : 654 – 660 . Google Scholar CrossRef Search ADS PubMed Breuer C , Braidwood L , Sugimoto K . 2014 . Endocycling in the path of plant development . Current Opinion in Plant Biology 17 : 78 – 85 . Google Scholar CrossRef Search ADS PubMed Brown SC , Bourge M , Maunoury N , et al. 2017 . DNA remodelling by strict partial endoreplication in orchids, an original process in the Plant Kingdom . Genome Biology and Evolution 9 : 1051 – 1071 . Google Scholar CrossRef Search ADS Cai J , Liu X , Vanneste K , et al. 2015 . Genome sequence of the orchid Phalaenopsis equestris . Nature Genetics 47 : 65 – 72 . Google Scholar CrossRef Search ADS PubMed Carvalhiera GMG . 2000 . Plant polytene chromosomes . Genetics and Molecular Biology 23 : 1043 – 1050 . Google Scholar CrossRef Search ADS Chen W-H , Tang C-Y , Lin T-Y , Weng Y-C , Kao Y-L . 2011 . Changes in the endopolyploidy pattern of different tissues in diploid and tetraploid Phalaenopsis aphrodite subsp. formosana (Orchidaceae) . Plant Science 181 : 31 – 38 . Google Scholar CrossRef Search ADS PubMed Claessens J , Kleynen J . 2011 . The flower of the European orchid: form and function . Voerendaal : published by the authors. Cookson SJ , Radziejwoski A , Granier C . 2006 . Cell and leaf size plasticity in Arabidopsis: what is the role of endoreduplication ? Plant, Cell and Environment 29 : 1273 – 1283 . Google Scholar CrossRef Search ADS Cremer T , Cremer C . 2001 . Chromosome territories, nuclear architecture and gene regulation in mammalian cells . Nature Reviews Genetics 2 : 292 – 301 . Google Scholar CrossRef Search ADS PubMed Delforge P . 2006 . Orchids of Europe, North Africa and the Middle East . London : A. & C. Black . Delforge P . 2016 . Orchidés d’Europe, d’Afrique du Nord et do Proche-Orient , 4th edn . Paris : Delachaux et Niestle . Devey DS , Bateman RM , Fay MF , Hawkins JA . 2008 . Friends or relatives? Phylogenetics and species delimitation in the controversial European orchid genus Ophrys . Annals of Botany 101 : 385 – 402 . Google Scholar CrossRef Search ADS PubMed Doležel J , Greilhuber J , Lucretti S , et al. 1998 . Plant genome size estimation by flow cytometry: inter-laboratory comparison . Annals of Botany 82 ( Supplement A ): 17 – 26 . Google Scholar CrossRef Search ADS Doležel J , Bartoš J , Voglmayr H , Greilhuber J . 2003 . Nuclear DNA content and genome size of trout and human . Cytometry 51A : 127 – 128 . Google Scholar CrossRef Search ADS Edgar BA , Zielke N , Gutierrez C . 2014 . Endocycles: a recurrent evolutionary innovation for post-mitotic cell growth . Nature Reviews 15 : 197 – 210 . Google Scholar CrossRef Search ADS PubMed Folkers U , Berger J , Hülskamp M . 1997 . Cell morphogenesis of trichomes in Arabidopsis: differential control of primary and secondary branching by branch initiation regulators and cell growth . Development 124 : 3779 – 3786 . Google Scholar PubMed Francisco A , Ascensão L . 2013 . Structure of the osmophore and labellum micromorphology in the sexually deceptive orchids Ophrys bombyliflora and Ophrys tenthredinifera (Orchidaceae) . International Journal of Plant Sciences 174 : 619 – 636 . Google Scholar CrossRef Search ADS Galbraith DW , Harkins KR , Knapp S . 1991 . Systematic endopolyploidy in Arabidopsis thaliana . Plant Physiology 96 : 985 – 989 . Google Scholar CrossRef Search ADS PubMed Gegas VC , Wargent JJ , Pesquet E , Grandqvist E , Paul ND , Doonan JH . 2014 . Endopolyploidy as a potential alternative adaptive strategy for Arabidopsis leaf size variation in response to UV-B . Journal of Experimental Botany 65 : 2757 – 2766 . Google Scholar CrossRef Search ADS PubMed Gögler J , Stökl J , Sramkova A , et al. 2009 . Ménage a trois – two endemic species of deceptive orchids and one pollinator species . Evolution 63 : 2222 – 2234 . Google Scholar CrossRef Search ADS PubMed Gostev A , Asker S . 1978 . Polytene chromosomes in glandular hairs of Salvia horminum . Hereditas 88 : 133 – 135 . Google Scholar CrossRef Search ADS Greilhuber J , Dolezel J , Lysak M , Bennett MD . 2005 . The origin, evolution and proposed stabilization of the terms ‘genome size’ and ‘C-value’ to describe nuclear DNA contents . Annals of Botany 95 : 255 – 260 . Google Scholar CrossRef Search ADS PubMed Guan X , Song Q , Chen ZJ . 2014 . Polyploidy and small RNA regulation of cotton fiber development . Trends in Plant Science 19 : 516 – 528 . Google Scholar CrossRef Search ADS PubMed Gutierrez C , Sequiera-Mendes J , Aragüez I . 2014 . Replication of the plant genome . In: Howell SH , ed. Molecular biology. The plant sciences , Vol. 2 . New York : Springer . Google Scholar CrossRef Search ADS Hedrén M , Nordström S , Bateman RM . 2011 . Plastid and nuclear DNA marker data support the recognition of four tetraploid marsh orchids (Dactylorhiza majalis s.l., Orchidaceae) in Britain and Ireland . Biological Journal of the Linnean Society 104 : 107 – 128 . Google Scholar CrossRef Search ADS Hennecke M , Munzinger S , Miller G , et al. 2015 . Stress-induced changes of floral scent in Ophrys taxa challenge ex situ evidence of supposed pollinator specificity . Berichte aus den Arbeitskreisen Heimische Orchideen 32 : 112 – 130 . Ho T-T , Kwon AR , Yoon Y-J , Paek K-Y , Park S-Y . 2016 . Endoreduplication level affects flower size and development by increasing cell size in Phalaenopsis and Doritaenopsis . Acta Physiologiae Plantae 38 : 190 . Google Scholar CrossRef Search ADS Hribova E , Holusova K , Travnicek P , et al. 2016 . The enigma of progressively partial endoreplication: new insights provided by flow cytometry and next-generation sequencing . Genome Biology and Evolution 8 : 1996 – 2005 . Google Scholar CrossRef Search ADS PubMed Huskins CL . 1947 . The subdivision of the chromosomes and their multiplication in non-dividing tissues: possible interpretations in terms of gene structure and gene action . American Naturalist 81 : 401 – 434 . Google Scholar CrossRef Search ADS PubMed Jersakova J , Vrana J , Hribova E , Dolezel J , Suda J . 2015 . Challenges of flow-cytometric estimation of nuclear genome size in orchids, a plant group with both whole-genome and progressively partial endoreplication . Cytometry A 87 : 958 – 966 . Google Scholar CrossRef Search ADS PubMed Jiao Y , Wickett NJ , Ayyampalayam S , et al. 2011 . Ancestral polyploidy in seed plants and angiosperms . Nature 473 : 97 – 100 . Google Scholar CrossRef Search ADS PubMed Kato M , Lam R . 2003 . Chromatin of endoreduplicated pavement cells has greater range of movement than that of diploid guard cells in Arabidopsis thaliana . Journal of Cell Science 116 : 2195 – 2201 . Google Scholar CrossRef Search ADS PubMed Kocova V , Strakova N , Kolarcik V , Rakai A , Martonfi P . 2016 . Endoreduplication as a part of flower ontogeny in Trifolium pratense cultivars . Botanical Studies 57 : 34 . Google Scholar CrossRef Search ADS PubMed Kudo N , Kimura Y . 2002 . Nuclear DNA endoreduplication during petal development in cabbage: relationship between ploidy levels and genome size . Journal of Experimental Botany 53 : 1017 – 1023 . Google Scholar CrossRef Search ADS PubMed Lee HC , Chiou DW , Chen WH , Markhart AH , Chen YH , Lin TY . 2004 . Dynamics of cell growth and endoreduplication during orchid flower development . Plant Science 166 : 659 – 667 . Google Scholar CrossRef Search ADS Lee HC , Chen YJ , Markhart AH , Lin TY . 2007 . Temperature effects on systemic endoreduplication in orchid during floral development . Plant Science 172 : 588 – 595 . Google Scholar CrossRef Search ADS Leitch AR , Mosgoller W , Shi M , Heslop-Harrison JS . 1992 . Different patterns of rDNA organization at interphase in nuclei of wheat and rye . Journal of Cell Science 101 : 751 – 757 . Google Scholar PubMed Leitch IJ , Kanhandawala J , Suda J , et al. 2009 . Genome size diversity in orchids: consequences and evolution . Annals of Botany 104 : 469 – 481 . Google Scholar CrossRef Search ADS PubMed Lim KY , Kovarik A , Matyasek R , et al. 2006 . Comparative genomics and repetitive sequence divergence in the species of diploid Nicotiana section Alatae . Plant Journal 48 : 907 – 919 . Google Scholar CrossRef Search ADS PubMed Lim WL , Loh CS . 2003 . Endopolyploidy in Vanda Miss Joaquim (Orchidaceae) . New Phytologist 159 : 279 – 287 . Google Scholar CrossRef Search ADS Liu G , Mattick JS , Taft RJ . 2013 . Meta-analysis of the genomic and transcriptomic composition of complex life . Cell Cycle 12 : 2061 – 2072 . Google Scholar CrossRef Search ADS PubMed Lomax BH , Hilton J , Bateman RM , et al. 2014 . Reconstructing relative genome size of vascular plants through geological time . New Phytologist 201 : 636 – 644 . Google Scholar CrossRef Search ADS PubMed Loureiro J , Rodriguez E , Doležel J , Santos C . 2007 . Two new nuclear isolation buffers for plant DNA flow cytometry: a test with 37 species . Annals of Botany 100 : 875 – 888 . Google Scholar CrossRef Search ADS PubMed Magyar Z , Horváth B , Khan S , et al. 2012 . Arabidopsis E2FA stimulates proliferation and endocycle separately through RBR‐bound and RBR‐free complexes . EMBO Journal 31 : 1480 – 1493 . Google Scholar CrossRef Search ADS PubMed Maluszynska J , Kolano B , Sas-Nowosielska H . 2013 . Endopolyploidy in plants . In: Leitch IJ , Greilhuber J , Doležel J , Wendel JF , eds. Plant genome diversity, Vol. 2. Physical structure, behaviour and evolution of plant genomes . Vienna : Springer , 99 – 119 . Google Scholar CrossRef Search ADS Massonet C , Tisne S , Radziejwoski A , et al. 2011 . New insights into the control of endoreduplication: endoreduplication could be driven by organ growth in Arabidopsis leaves . Plant Physiology 157 : 2044 – 2055 . Google Scholar CrossRef Search ADS PubMed McCarthy EW , Chase MW , Knapp S , Litt A , Leitch AR , Le Comber SC . 2016 . Transgressive phenotypes and generalist pollination in the floral evolution of Nicotiana polyploids . Nature Plants 2 : 161119 . Google Scholar CrossRef Search ADS Melaragno JE , Mehrotra B , Coleman AW . 1993 . Relationship between endopolyploidy and cell size in epidermal tissue of Arabidopsis . Plant Cell 11 : 1661 – 1668 . Google Scholar CrossRef Search ADS Mishiba K-I , Okamoto T , Mii M . 2001 . Increasing ploidy level in cell suspension cultures of Doritaenopsis by exogenous application of 2,4-dichlorophenoxyacetic acid . Physiologia Plantarum 112 : 142 – 148 . Google Scholar CrossRef Search ADS PubMed Mondragón-Palomino M , Theissen G . 2011 . Conserved differential expression of paralogous DEFICIENS- and GLOBOSA-like MADS-box genes in the flowers of Orchidaceae: refining the ‘orchid code’ . Plant Journal 66 : 1008 – 1019 . Google Scholar CrossRef Search ADS PubMed Nagl W . 1978 . Endopolyploidy and polyteny in differentiation and evolution . Amsterdam : Elsevier . Neiman M , Beaton MJ , Hessen DO , Jeyasingh PD , Wider LJ . 2017 . Endopolyploidy as a potential driver of animal ecology and evolution . Biological Reviews 92 : 234 – 247 . Google Scholar CrossRef Search ADS PubMed Paulus H . 2015 . Bestäuber als Isolationsmechanismen: Freilandbeobachtungen und Experimente zur Spezifität der Bestäuberanlockung in der Gattung Ophrys (Orchidaceae und Insecta, Hymenoptera, Apoidea) . Berichte aus den Arbeitskreisen Heimische Orchideen 32 : 142 – 199 . Paun O , Bateman RM , Fay MF , Hedrén M , Civeyrel L , Chase MW . 2010 . Stable epigenetic effects impact evolution and adaptation in allopolyploid orchids . Molecular Biology and Evolution 27 : 2465 – 2473 . Google Scholar CrossRef Search ADS PubMed Pearson MJ . 1974 . Polyteny and the functional significance of the polytene cell cycle . Journal of Cell Science 15 : 457 – 479 . Google Scholar PubMed Van de Peer Y , Mizrachi E , Marchal K . 2017 . The evolutionary significance of polyploidy . Nature Reviews Genetics 18 : 411 – 424 . Google Scholar CrossRef Search ADS PubMed Pillon Y , Fay MF , Hedrén M , et al. 2007 . Evolution and biogeography of European species complexes in Dactylorhiza (Orchidaceae) . Taxon 56 : 1185 – 1208 . Google Scholar CrossRef Search ADS Polyn S , Willems A , De Veylder L . 2015 . Cell cycle entry, maintenance, and exit during development . Current Opinion in Plant Biology 23 : 1 – 7 . Google Scholar CrossRef Search ADS PubMed Del Pozo JC , Ramirez-Parra E . 2015 . Whole genome duplications in plants: an overview from Arabidopsis . Journal of Experimental Botany 66 : 6991 – 7003 . Google Scholar CrossRef Search ADS PubMed Renny-Byfield S , Chester M , Kovarik A , et al. 2011 . Next generation sequencing reveals genome downsizing in allotetraploid Nicotiana tabacum, predominantly through the elimination of paternally derived repetitive DNAs . Molecular Biology and Evolution 28 : 2843 – 2854 . Google Scholar CrossRef Search ADS PubMed Renny-Byfield S , Kovařík A , Chester M , et al. 2012 . Independent, rapid and targeted loss of highly repetitive DNA in natural and synthetic allopolyploids of Nicotiana tabacum . PLoS ONE 7 : e36963 . Google Scholar CrossRef Search ADS PubMed Robinson DO , Roeder AHK . 2015 . Themes and variations in cell type patterning in the plant epidermis . Current Opinion in Genetics & Development 32 : 55 – 65 . Google Scholar CrossRef Search ADS PubMed Rodman TC . 1967 . DNA replication in salivary gland nuclei of Drosophila melanogaster at successive larval and prepupal stages . Genetics 55 : 375 – 386 . Google Scholar PubMed Roeder AHK , Chickarmane V , Cunha A , Obara B , Manjunat BS , Meyerowitz EM . 2010 . Variability in the control of cell division underlies sepal epidermal patterning in Arabidopsis thaliana . PLoS Biology 8 : e1000367 . Google Scholar CrossRef Search ADS PubMed Rudall PJ , Bateman RM . 2002 . Roles of synorganisation, zygomorphy and heterotopy in floral evolution: the gynostemium and labellum of orchids and other lilioid monocots . Biological Reviews 77 : 403 – 441 . Google Scholar CrossRef Search ADS PubMed Schepper S , Leus L , Mertens M , Debergh P , Bockstaele E , Loose M . 2001 . Somatic polyploidy and its consequences for flower coloration and morphology in azalea . Plant Cell Reports 20 : 583 – 590 . Google Scholar CrossRef Search ADS Schiestl FP . 2005 . On the success of a swindle: pollination by deception in orchids . Naturwissenschaften 92 : 255 – 262 . Google Scholar CrossRef Search ADS PubMed Scholes DR , Paige KN . 2014 . Plasticity in ploidy underlies plant fitness compensation to herbivore damage . Molecular Ecology 23 : 4862 – 4870 . Google Scholar CrossRef Search ADS PubMed Scholes DR , Paige KN . 2015 . Plasticity in ploidy: a generalized response to stress . Trends in Plant Science 20 : 165 – 175 . Google Scholar CrossRef Search ADS PubMed Schubert V , Berr A , Meister A . 2012 . Interphase chromatin organisation in Arabidopsis nuclei: constraints versus randomness . Chromosoma 121 : 369 – 387 . Google Scholar CrossRef Search ADS PubMed Sedeek KEM , Scopece G , Staedler YM , et al. 2014 . Genic rather than genomewide differences between sexually deceptive Ophrys orchids with different pollinators . Molecular Ecology 23 : 6192 – 6205 . Google Scholar CrossRef Search ADS PubMed Sedeek KEM , Whittle E , Guthörl D , Grossniklaus U , Shanklin J , Schlüter PM . 2016 . Amino acid change in an orchid desaturase enables mimicry of the pollinator’s sex pheromone . Current Biology 28 : 1505 – 1511 . Google Scholar CrossRef Search ADS Sliwinska E , Mathur J , Bewley JD . 2015 . On the relationship between endoreplication and collet hair initiation and tip growth, as determined using six Arabidopsis thaliana root-hair mutants . Journal of Experimental Botany 66 : 3285 – 3295 . Google Scholar CrossRef Search ADS PubMed Snodgrass SJ , Jareczek J , Wendel JF . 2017 . An examination of the nucleotypic effects in diploid and polyploid cotton . AoB Plants 9 : plw082 . Google Scholar CrossRef Search ADS PubMed Sugimoto-Shirasu K , Roberts K . 2003 . “Big it up”: endoreduplication and cell-size control in plants . Current Opinion in Plant Biology 6 : 544 – 553 . Google Scholar CrossRef Search ADS PubMed Teixeira da Silva JA , Giang DTT , Dobránszki J , Zeng S , Tanaka M . 2014 . Ploidy analysis of Cymbidium, Phalaenopsis, Dendrobium and Paphiopedilum (Orchidaceae), and Spathiphyllum and Syngonium (Araceae) . Biologia (Botany) 69 : 750 – 755 . Traas J , Hülskamp M , Gendreau E , Höfte H . 1998 . Endoreduplication and development: rule without dividing ? Current Opinion in Plant Biology 1 : 498 – 503 . Google Scholar CrossRef Search ADS PubMed Trávníček P , Jersáková J , Kubátová B , et al. 2012 . Minority cytotypes in European populations of the Gymnadenia conopsea complex (Orchidaceae) greatly increase intraspecific and intrapopulation diversity . Annals of Botany 110 : 977 – 986 . Google Scholar CrossRef Search ADS PubMed Travnicek P , Ponert J , Urfus T , et al. 2015 . Challenges of flow-cytometric estimation of nuclear genome size in orchids, a plant group with both whole-genome and progressively partial endoreplication . Cytometry A 87 : 958 – 966 . Google Scholar CrossRef Search ADS PubMed Tschermak-Woess E . 1956 . Karyologische Pflanzenanatomie . Protoplasma 46 : 798 – 834 . Google Scholar CrossRef Search ADS Vereecken NJ , Streinzer M , Ayasse M , et al. 2011 . Integrating past and present studies on Ophrys pollination: a comment on Bradshaw et al . Botanical Journal of the Linnean Society 165 : 329 – 335 . Google Scholar CrossRef Search ADS Vignolini S , Davey MP , Bateman RM , et al. 2012 . The mirror crack’d: both structure and pigment contribute to the metallic blue appearance of the mirror orchid, Ophrys speculum . New Phytologist 196 : 1038 – 1047 . Google Scholar CrossRef Search ADS PubMed Vitrat N , Cohen-Solai K , Pique C , et al. 1998 . Endomitosis of human megakaryocytes are due to abortive mitosis . Blood 91 : 3711 – 3723 . Google Scholar PubMed Wendel JF . 2015 . The wondrous cycles of polyploidy in plants . American Journal of Botany 102 : 1753 – 1756 . Google Scholar CrossRef Search ADS PubMed Yang M , Loh CS . 2004 . Systematic endopolyploidy in Spathoglottis plicata (Orchidaceae) development . BMC Cell Biology 5 : 33 . Google Scholar CrossRef Search ADS PubMed Zhang G-Q , Liu K-W , Li Z , et al. 2017 . The Apostasia genome and the evolution of orchids . Nature 549 : 379 – 383 . Google Scholar CrossRef Search ADS PubMed Zhimulev IF , Koryakov DE . 2009 . Polytene chromosomes . Encyclopedia of Life Sciences . www.els.net. Zhimulev IF , Belyaeva ES , Vatolina TY , Demakov SA . 2012 . Banding patterns in Drosophila melanogaster polytene chromosomes correlate with DNA-binding protein occupancy . BioEssays 34 : 498 – 508 . Google Scholar CrossRef Search ADS PubMed © The Author(s) 2018. Published by Oxford University Press on behalf of the Annals of Botany Company. All rights reserved. For permissions, please e-mail: [email protected]. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)
Immediate vs. evolutionary consequences of polyploidy on clonal reproduction in an autopolyploid plant2018 Annals of Botany
doi: 10.1093/aob/mcy071pmid: 29726889
Abstract Background and Aims Clonal reproduction in polyploids is expected to exceed that in diploids, due to either the immediate direct effects of whole-genome duplication (WGD) or selection during establishment. The timing of polyploidy effects on clonality are largely unknown despite its hypothesized influence on polyploid success. This study tests the direction and timing of divergence in clonal traits in diploid and polyploid Chamerion angustifolium. Methods Root bud production and biomass allocation patterns were compared between diploids and synthesized tetraploids (neotetraploids), and between neotetraploids and naturally occurring tetraploids grown in a common environment. Key Results Neotetraploids produced more root buds and fewer sexual structures than diploids and natural tetraploids; diploids and natural tetraploids had similar root bud numbers and sexual investment. The root bud:inflorescence biomass ratio was 71 % higher in neotetraploids than in natural tetraploids. Root bud location suggests that ramet density in neotetraploid genets could be higher than in diploid genets. Conclusions WGD immediately increases investment in asexual vs. sexual reproduction in C. angustifolium, potentially promoting within-cytotype mating and establishment for neopolyploids. However, evolutionary change after the polyploidization event negates the direct effects of WGD. Natural polyploids and diploids have similar root bud production and biomass allocation patterns, probably resulting from habitat- and ploidy-mediated selection on polyploids to become more like diploids. These results highlight the value of studying the effects of polyploidization in young vs. established polyploids. Clonal reproduction, colchicine, Chamerion angustifolium (fireweed), neopolyploid, autopolyploidy, root buds, whole-genome duplication INTRODUCTION Polyploidy is considered to be a driving force behind adaptive divergence and diversification in the angiosperms (Otto and Whitton, 2000; Wood et al., 2009, Soltis and Soltis, 2016), although the mechanisms leading to the formation and establishment of polyploid populations are still poorly understood (Soltis et al., 2010). Autopolyploids often differ ecologically and phenotypically from their lower ploid parents (Stebbins, 1950; Levin, 1983; Ramsey and Schemske, 2002; Husband et al., 2016), but because studies are commonly performed on long-established cytotypes it is unclear whether differences are due to instantaneous changes associated with the whole-genome duplication (WGD) event (Stebbins, 1971; Otto and Whitton, 2000; Comai, 2005) or divergence through selection after the fact (Bretagnolle and Lumaret, 1995; Weiss-Schneeweiss et al., 2013). Newly produced, synthetic polyploids (i.e. neopolyploids) provide an opportunity to study the direct phenotypic effects of WGD separate from changes wrought by generations of selection in naturally occurring polyploid cytotypes. However, using neopolyploids to study the immediate ecological and evolutionary consequences of a polyploidization event is uncommon (Bretagnolle and Lumaret, 1995; Husband et al., 2008; Maherali et al., 2009; Baldwin and Husband, 2013; Oswald and Nuismer, 2010; Ramsey, 2011; Martin and Husband, 2012; Husband et al., 2016), and it frequently remains unknown how phenotypically similar neopolyploids are to diploids, or what processes contribute to differences between new and established polyploids. The timing of phenotypic divergence can have a significant impact on the likelihood of polyploid establishment and persistence in sympatry with its progenitor (Ramsey, 2011; Husband et al., 2016). One of the most crucial barriers to establishment that neopolyploids must overcome is a frequency-dependent mating disadvantage, where rare polyploids experience a lack of same-cytotype mates resulting in a disproportionate number of between-cytotype fertilizations [minority cytotype exclusion (MCE); Levin, 1975]. Polyploid gametes are usurped by incompatible pollen, and subsequent odd-ploidy offspring have low viability and fertility, leading to low polyploid fitness and their eventual exclusion from the population (Husband, 2000; Baack, 2005). MCE may be weakened by increasing same-cytotype mating through shifts in mating system (Barringer, 2007; Husband et al., 2008), self-compatibility (Thompson and Lumaret, 1992; Robertson et al., 2011), flowering time (Husband and Sabara, 2004), flower morphology (Segraves and Thompson, 1999; Lim et al., 2008; Vallejo-Marin, 2012), eco-geographic differentiation (Martin and Husband, 2012; Thompson et al., 2014) and asexual reproduction via apomixis (Rodriguez, 1996; Otto and Whitton, 2000). Asexual reproduction through clonal reproduction could also facilitate polyploid establishment (Stebbins, 1950; Thompson and Lumaret, 1992; Otto and Whitton, 2000; Weiss-Schneeweis et al., 2013). Both the evolutionary relationship between polyploidy and clonal reproduction and the underlying mechanisms behind any association remain understudied (Baldwin and Husband, 2013; Freeling, 2017; Herben et al., 2017; Kolář et al., 2017). Clonal reproduction can be broadly defined as vegetative asexual reproduction through the propagation of plant parts not involving seeds (Vallejo-Marin et al., 2010), such that a genetic individual (a genet) may consist of multiple, genetically identical, daughter ramets. Clonal reproduction can occur through a wide variety of modes (e.g. rhizomes, stolons, bulbils, plantlets and corms; Klimešová and De Bello, 2009). Early researchers posited that a perennial life history and the ability to reproduce clonally may be direct consequences of a polyploidization event (Müntzing, 1936; Gustafsson, 1948), but it is more likely that WGD would quantitatively alter the clonal ability normally present in a diploid (Stebbins, 1950; Soltis et al., 2016). A WGD event can immediately alter phenotypic and reproductive patterns (Otto and Whitton, 2000; Ramsey and Schemske, 2002), and differences in the size or number of clonal propagules produced by a neopolyploid could be due to gene dosage effects (Guo et al., 1996; Ramsey and Schemske, 2002), or new gene interactions and functions (Osborn et al., 2003; Parisod et al., 2010; Roulin et al., 2013; Soltis et al., 2016). While the direction of genetic effects of WGD can be unclear, it is generally assumed that the genetic consequences of WGD will result in heightened rates of clonal reproduction in polyploids (Stebbins, 1950; Otto and Whitton, 2000; Herben et al., 2017). Neopolyploids could also experience modification to clonal traits as an artefact of the direct effects of WGD on other traits. Neopolyploids often have faulty sexual reproduction because of chromosome segregation irregularities (Comai, 2005; Cifuentes et al., 2010), resulting in sexual structures not being fully formed, or the abortion of malformed ovules (Stebbins, 1950; Levin, 2002). In this case, a re-allocation of resources towards clonal growth as an alternative form of reproduction could occur within the lifetime of a neopolyploid (Gustafsson, 1948; van Kleunen et al., 2002; Vallejo-Marin et al., 2010). Enhanced clonal reproduction in neopolyploids could facilitate survival through the bottle-neck stage of establishment via physical competition and persistence even in the absence of sexual reproduction (Stebbins, 1950; Ramsey and Schemske, 2002; Husband et al., 2008). Moreover, since mating generally involves near neighbours (Levin and Kerster, 1974; Vallejo-Marin et al., 2010), producing spatially proximate clones will increase rates of geitonogamous pollen transfer between ramets in a genet (Charpentier, 2002; Baack, 2005), and increase the number of potential same-cytotype mates within a predominantly diploid population. If the costs of selfing are less than those of between-cytotype mating, the probability of polyploid establishment will increase as more same-cytotype matings occur and the negative effects of MCE are circumvented (Husband and Schemske, 1997; Barringer, 2007; Husband et al., 2008). During the establishment process, we expect that selection will maintain or further increase clonal reproduction in polyploids to extend the associated advantageous conditions for spreading within the population until more sexually produced same-cytotype mates become available (Müntzing, 1936; Eckert, 2002; Honnay and Bossuyt, 2005). Moreover, since polyploids are expected to have lower inbreeding depression than diploids (Lande and Schemske, 1985; Barringer, 2007; Husband et al., 2008), geitonogamous selfing in polyploids should result in a smaller fitness reduction than in diploids. Thus, once fully established, naturally occurring polyploid cytotypes should be able to tolerate larger, more compact genets and have optimal fitness at higher rates of clonal reproduction than diploids (Baldwin and Husband, 2013). Previous research quantifying clonal reproduction in polyploids has presented conflicting results. There is some support for the prediction that higher ploidies will have increased levels of clonality in comparison with diploids (Bruneau and Anderson, 1988; Hroudová and Zákravský, 1993; Schlaepfer et al., 2010), while other studies have found that polyploids produce fewer clonal propagules than diploids (Schulze et al., 2013; Baldwin and Husband, 2013; Hanzl et al., 2014; Martínková et al., 2015), and some have demonstrated no differences between ploidies (Keeler, 2004). The above sudies have all focused on naturally occurring ploidy levels, and there are no studies that have used neopolyploids in determining the immediate effects of WGD on clonal reproduction. Here, we use the mixed-ploidy species Chamerion angustifolium to investigate the differences in clonal reproduction between diploids and tetraploids and whether such differences are the direct result of WGD or differential selection in natural populations. Baldwin and Husband (2013) previously tested for differences in size and spatial extent of clones between diploids and tetraploids in natural populations of this species, but did not assess the capacity for clonal reproduction in neotetraploid C. angustifolium. We investigate the following three questions. (1) Are polyploids more clonal than their diploid progenitors? (2) Do differences in clonal reproduction occur immediately following WGD? (3) Do differences arise through evolutionary modification in natural populations? To address these questions, we compare clonal reproduction via root bud production between diploid, naturally occurring tetraploid and synthetic neotetraploid C. angustifolium grown under common greenhouse conditions. Neotetraploids are contrasted with diploids to address the second question explicitly, and with naturally occurring tetraploids to address the third. MATERIALS AND METHODS Study species Chamerion angustifolium is a perennial, self-compatible, insect-pollinated herb occurring widely across the northern hemisphere in predominantly open or disturbed habitat. This species exhibits ploidy variation with diploid (2n = 2x = 36), tetraploid (2x = 4x = 72) and infrequent triploid individuals occurring naturally in North America (Sabara et al., 2013). Tetraploids are autotetraploids, derived from the doubling of the diploid C. angustifolium genome (Roy, 2008). Typically, diploids occur at higher latitude and altitude than tetraploids, but mixed populations occur in a diffuse contact zone across southern parts of the boreal forest and in the Rocky Mountains in North America (Husband and Schemske, 2000; Sabara et al., 2013). Chamerion angustifolium reproduces both sexually through perfect flowers, and asexually through the production of independent plants via vertically growing adventitious root buds (Mosquin, 1966; Stocklin, 1992). Both diploids and tetraploids have high inbreeding depression and strongly outcrossing mating systems (Husband et al., 2008; Ozimec and Husband, 2011). In contrast, previous studies have found that newly synthesized neotetraploid C. angustifolium have a significantly lower cost of inbreeding than naturally occurring tetraploids (Husband et al., 2008). Consequently, fitness gains through sexual reproduction for diploids, naturally occurring tetraploids and neotetraploids can be greatly impacted by the spatial arrangement of ramets within a genet as determined by patterns of investment in clonal reproduction (Charpentier, 2002; Vallejo-Marin et al., 2010; Van Drunen et al., 2015). Source material, neopolyploid synthesis and ploidy determination Root bud production was compared between greenhouse-grown plants from three available cytotypes of C. angustifolium (diploid, naturally occurring tetraploid and newly synthesized tetraploids) in the summer of 2016. Diploid and natural tetraploid seed of C. angustifolium were the F1 progeny of within-population within-cytotype crosses performed during 2015 on plants collected from ten locations in the Rocky Mountains (four diploid populations, four tetraploid populations and two of mixed ploidy; Table 1). A total of 18 diploids and 18 natural tetraploids were used in this study, from three crosses within each of the six populations per naturally occurring cytotype (Table 1). Table 1. Source population locations for C. angustifolium within the Rocky Mountains and the number of individuals of each ploidy used in the current study Population Ploidy Latitude Longitude Number of individuals 2x 4xNeo 4x Jasper Park Boundary 2x N 51°26.674 W 116°12.532 3 3 – Marmot Basin 2x N 52°48.100 W 118°04.956 3 2 – Wilcox Creek 2x N 52°13.082 W 117°10.621 3 – – Fortress Mountain 2x N 50°49.534 W 115°12.052 3 4 – Moose Meadows 4x N 51°15.231 W 115°52.336 – – 3 Powderface Low 4x N 51°01.318 W 114°53.861 – – 3 Sibbald Mountain 4x N 51°03.160 W 114°56.847 – – 3 Barrier Lake 4x N 51°01.785 W 115°02.097 – – 3 Coleman Clearcut Mixed N 50°18.648 W 114°36.761 3 2 3 Rampart Creek Mixed N 52°02.511 W 116°51.767 3 2 3 Mount Kitchener Mixed N 52°16.461 W 117°18.538 – 2 – Total 18 15 18 Population Ploidy Latitude Longitude Number of individuals 2x 4xNeo 4x Jasper Park Boundary 2x N 51°26.674 W 116°12.532 3 3 – Marmot Basin 2x N 52°48.100 W 118°04.956 3 2 – Wilcox Creek 2x N 52°13.082 W 117°10.621 3 – – Fortress Mountain 2x N 50°49.534 W 115°12.052 3 4 – Moose Meadows 4x N 51°15.231 W 115°52.336 – – 3 Powderface Low 4x N 51°01.318 W 114°53.861 – – 3 Sibbald Mountain 4x N 51°03.160 W 114°56.847 – – 3 Barrier Lake 4x N 51°01.785 W 115°02.097 – – 3 Coleman Clearcut Mixed N 50°18.648 W 114°36.761 3 2 3 Rampart Creek Mixed N 52°02.511 W 116°51.767 3 2 3 Mount Kitchener Mixed N 52°16.461 W 117°18.538 – 2 – Total 18 15 18 Neotetraploids (4xNeo) are listed under the source population of their maternal parent. View Large Table 1. Source population locations for C. angustifolium within the Rocky Mountains and the number of individuals of each ploidy used in the current study Population Ploidy Latitude Longitude Number of individuals 2x 4xNeo 4x Jasper Park Boundary 2x N 51°26.674 W 116°12.532 3 3 – Marmot Basin 2x N 52°48.100 W 118°04.956 3 2 – Wilcox Creek 2x N 52°13.082 W 117°10.621 3 – – Fortress Mountain 2x N 50°49.534 W 115°12.052 3 4 – Moose Meadows 4x N 51°15.231 W 115°52.336 – – 3 Powderface Low 4x N 51°01.318 W 114°53.861 – – 3 Sibbald Mountain 4x N 51°03.160 W 114°56.847 – – 3 Barrier Lake 4x N 51°01.785 W 115°02.097 – – 3 Coleman Clearcut Mixed N 50°18.648 W 114°36.761 3 2 3 Rampart Creek Mixed N 52°02.511 W 116°51.767 3 2 3 Mount Kitchener Mixed N 52°16.461 W 117°18.538 – 2 – Total 18 15 18 Population Ploidy Latitude Longitude Number of individuals 2x 4xNeo 4x Jasper Park Boundary 2x N 51°26.674 W 116°12.532 3 3 – Marmot Basin 2x N 52°48.100 W 118°04.956 3 2 – Wilcox Creek 2x N 52°13.082 W 117°10.621 3 – – Fortress Mountain 2x N 50°49.534 W 115°12.052 3 4 – Moose Meadows 4x N 51°15.231 W 115°52.336 – – 3 Powderface Low 4x N 51°01.318 W 114°53.861 – – 3 Sibbald Mountain 4x N 51°03.160 W 114°56.847 – – 3 Barrier Lake 4x N 51°01.785 W 115°02.097 – – 3 Coleman Clearcut Mixed N 50°18.648 W 114°36.761 3 2 3 Rampart Creek Mixed N 52°02.511 W 116°51.767 3 2 3 Mount Kitchener Mixed N 52°16.461 W 117°18.538 – 2 – Total 18 15 18 Neotetraploids (4xNeo) are listed under the source population of their maternal parent. View Large To obtain newly synthesized tetraploid seed, diploid seedlings were treated with colchicine, resulting in genome duplication and conversion to tetraploids (i.e. neotetraploids, 4xNeo). The diploid seeds used originated from the diploid and mixed-ploidy populations indicated in Table 1. Seeds were sown in Petri dishes on moist filter paper and individually treated with 40 μL of 0.2 % colchicine solution after 10 d when cotyledons were expanded and the seedlings exhibited vertical growth. Eighteen hours after application, the colchicine was rinsed from the seedlings twice with deionized water, and seedlings were transplanted onto soil. Crosses were performed between 11 neotetraploid individuals surviving to maturity, but due to high rates of pollen sterility in the small number of successfully converted neotetraploids F1 crosses were performed both within and between population sources. A total of 15 neotetraploid individuals resulting from eight unique crosses were used in the current study. Neotetraploids are listed in Table 1 under the population source of their maternal parent. Diploid, naturally occurring tetraploid and neotetraploid seeds were sown on moist filter paper in Petri dishes and kept in the dark at 4 ºC for 24 h before moving to a Percival growth cabinet for germination at 22 ºC over 2 weeks. Seedlings were transplanted into a mixture of 5:1 Sunshine Mix to turface for a further 4 weeks of growth for seedling establishment. Individuals were then transplanted into HML Elite 1000 pots (8.83 L) with the same soil composition, and randomly placed on a greenhouse bench. This final pot size is an approx. 40 % larger volume than those used in the previous study (Baldwin and Husband, 2013) in order to ensure that the root systems were not severely root bound (coiling roots tend to influence the location of root bud elongation, W. E.Van Drunen, pers. obs.). The ploidy of all plants was confirmed using estimates of DNA content via a FACSCalibur flow cytometer (BD Bioscience, San Jose, CA, USA). Approximately 1 cm2 of dried leaf tissue per plant was chopped with a clean razor blade along with an equal amount of internal standard (Solanum lycopersicum) in 0.7 mL of modified DeLaat’s buffer (Kron and Husband, 2009) with 50 μg mL–1 of the DNA-selective fluorochrome propidium iodide and 50 μg mL–1 RNase, and then passed through a 30 μm filter. Samples were stained for a minimum of 20 min. Relative fluorescence was measured with the FL2 detector (585/42 nm) and DNA content was quantified with FL2-area (integrated fluorescence). Ploidy was determined by estimating the mean relative fluorescence of the nuclei for each sample using ModFit LT (Verity Software House Inc., San Jose, CA, USA). Sample fluorescence peaks violating quality thresholds [coefficient of variation (CV) >8, nuclei count <500] were run twice and excluded if the peak was not identifiable as diploid or tetraploid. Fluorescence peaks are generally clearly recognizable in C. angustifolium, as the mean fluorescence values for diploid, tetraploid and the internal standard S. lycopersicum are non-overlapping. Figure 1 shows the distribution of fluorescence values for the three cytotypes in this study relative to those of the internal standard. The relative fluorescence values for 98 % of the individuals used in this study fell within 10 % of the mean fluorescence value for diploids (0.748 ± 10 % = 0.673–0.823), natural tetraploids (1.488 ± 10 % = 1.339–1.637) and neotetraploids (1.498 ± 10% = 1.348–1.648). The single sample that fell outside of these ranges was assigned to the nearest ploidy. Peak nuclei counts, CVs and relative fluorescence means for each cytotype are shown in Table 2. Fig. 1. View largeDownload slide Distribution of DNA content for diploids (2x, red), neotetraploids (4xNeo, green) and naturally occurring tetraploids (4x, blue), determined through flow cytometry on dried leaf tissue. Overlaps between neotetraploid and naturally occurring tetraploid distributions are darker blue. DNA content is the relative fluorescence ratio of each sample peak to the internal standard. Fig. 1. View largeDownload slide Distribution of DNA content for diploids (2x, red), neotetraploids (4xNeo, green) and naturally occurring tetraploids (4x, blue), determined through flow cytometry on dried leaf tissue. Overlaps between neotetraploid and naturally occurring tetraploid distributions are darker blue. DNA content is the relative fluorescence ratio of each sample peak to the internal standard. Table 2. Flow cytometry on leaf tissue from diploids (2x), neotetraploids (4xNeo) and naturally occurring tetraploids (4x) Ploidy n Events per fluorescence peak Peak CV Relative fluorescence 2x 18 1278 ± 88 4.51 ± 0.27 0.748 ± 0.005 4xNeo 15 2073 ± 213 3.42 ± 0.15 1.488 ± 0.012 4x 18 1262 ± 138 3.91 ± 0.25 1.498 ± 0.012 Ploidy n Events per fluorescence peak Peak CV Relative fluorescence 2x 18 1278 ± 88 4.51 ± 0.27 0.748 ± 0.005 4xNeo 15 2073 ± 213 3.42 ± 0.15 1.488 ± 0.012 4x 18 1262 ± 138 3.91 ± 0.25 1.498 ± 0.012 Quality thresholds for peak CVs (CV <8) and nuclei counts (>500) were met for all samples. Values are means ± s.e. View Large Table 2. Flow cytometry on leaf tissue from diploids (2x), neotetraploids (4xNeo) and naturally occurring tetraploids (4x) Ploidy n Events per fluorescence peak Peak CV Relative fluorescence 2x 18 1278 ± 88 4.51 ± 0.27 0.748 ± 0.005 4xNeo 15 2073 ± 213 3.42 ± 0.15 1.488 ± 0.012 4x 18 1262 ± 138 3.91 ± 0.25 1.498 ± 0.012 Ploidy n Events per fluorescence peak Peak CV Relative fluorescence 2x 18 1278 ± 88 4.51 ± 0.27 0.748 ± 0.005 4xNeo 15 2073 ± 213 3.42 ± 0.15 1.488 ± 0.012 4x 18 1262 ± 138 3.91 ± 0.25 1.498 ± 0.012 Quality thresholds for peak CVs (CV <8) and nuclei counts (>500) were met for all samples. Values are means ± s.e. View Large Growth, biomass allocation and root bud traits Plants were grown for 12 weeks under greenhouse conditions within the normal range experienced by C. angustifolium in source populations during the flowering season (Thompson et al., 2014), with a daylength of approx. 16 h and a day temperature of 22 ºC. Plants were watered biweekly and given fertilizer (Plant Prod pH Reducer, 18-9-18) on a weekly basis. After 12 weeks, plants were moved outside for 1 week to stimulate the beginning of senescence, then harvested. Above- and below-ground measurements were taken for all plants. Plant height (from the soil surface to the top of the primary stem) and the number of axillary branches per plant were assessed. The above-ground biomass of each plant was harvested and stored in paper bags. The above-ground biomass was separated into shoots and inflorescences, where an inflorescence was defined as all shoot mass above the lowest flower produced on a stem or branch. Inflorescence mass was used as a proxy for investment in sexual reproduction. Above-ground shoots and inflorescences were dried at 65 ºC for at least 72 h before being weighed. After the above-ground biomass was harvested, root systems were gently excavated from the soil, washed, and stored at 4 ºC to inhibit further growth before root bud measurements could be taken. Root buds were identified by their light coloration, scaly appearance and vertical growth. Root buds were located along the roots, counted and their distances from the primary shoot recorded. Distances were measured on straightened roots, not on 3-D root topology as it grew in the pot, and thus are potential rather than realized distances. Root bud heights (base to tip) were measured using digital calipers. Root buds <1.5 mm in height have a similar appearance to small root initials, and as such were excluded from measurement. Root buds were removed from the roots and dried in silica for approx. 1 week before total root bud mass per plant was measured. Individual root bud mass was calculated as the total root bud mass divided by the number of root buds. Roots were placed in paper bags and dried at 65 ºC for at least 72 h before being weighed. Statistical analysis All analyses were performed in R (3.4.0; R Core Team, 2017). Linear mixed models were used to determine the effect of ploidy on plant growth, biomass allocation and root bud attributes. Models were formulated with the ‘lmer’ function in R (‘lme4’ package; Bates et al., 2015) and response variables were transformed where necessary in order to adhere to model assumptions (see Table 3). Plant growth traits included plant height and the number of axillary branches produced off the main stem of the plant. Total dry biomass was divided into four categories: above-ground shoot mass, inflorescence mass, root mass and root bud mass. Root bud attributes were the number of root buds, the average mass of an individual root bud, average root bud height, average bud distance along the root from the primary stem, and the minimum and maximum mean root bud distances per individual. The minimum and maximum root bud distances were included to provide information on the potential spatial extent and aggregation of a growing genet. To control for the effects of plant size on the measured traits, total biomass (growth and biomass attributes) or root mass (root bud attributes) were used as covariates in the models (see Table 3). To include variability between individual root buds more accurately, two additional models were used to quantify the effect of ploidy on root bud height and distance for all 1929 root buds identified. Significance testing for ploidy in all models was performed through likelihood ratio tests (LRTs) comparing full and reduced models for each response variable. Table 3. Results of linear mixed models for plant growth, biomass allocation and root bud characteristics TF COV Ploidy Random effects LRT χ2 d.f. P LRT χ2 d.f. P Growth attributes Plant height (cm) – TM 9.97 2 0.007 0 1 1 Branches – TM 13.66 2 0.001 0.01 1 0.917 Biomass allocation Total mass (g) – – 11.54 2 0.003 0 1 1 Shoot mass (g) – TM 19.14 2 <0.0001 3.05 1 0.080 Inflorescence mass (g) – TM 17.03 2 <0.001 3.95 1 0.047 Root mass (g) SR TM 0.85 2 0.652 0 1 1 Root bud mass (mg) SR TM 2.71 2 0.257 0.12 1 0.724 Root bud mass/inflorescence mass (mg g–1) FR TM 8.85 2 0.012 2.88 1 0.090 Root bud attributes Bud number SR RM 9.20 2 0.010 0.01 1 0.987 Individual root bud mass (mg) FR RM 2.49 2 0.288 0.01 1 0.904 Average bud height (mm) LOG RM 1.17 2 0.556 0 1 1 Average bud distance (cm) LOG RM 1.41 2 0.494 3.51 1 0.061 Minimum bud distance (cm) LOG RM 1.30 2 0.522 0.62 1 0.430 Maximum bud distance (cm) LOG RM 1.30 2 0.522 1.76 1 0.185 All root buds (n = 1929) Bud height (mm) LOG – 0.20 2 0.905 577.99 2 <0.0001 Bud distance (cm) LOG – 237.39 3 <0.0001 292.02 2 <0.0001 TF COV Ploidy Random effects LRT χ2 d.f. P LRT χ2 d.f. P Growth attributes Plant height (cm) – TM 9.97 2 0.007 0 1 1 Branches – TM 13.66 2 0.001 0.01 1 0.917 Biomass allocation Total mass (g) – – 11.54 2 0.003 0 1 1 Shoot mass (g) – TM 19.14 2 <0.0001 3.05 1 0.080 Inflorescence mass (g) – TM 17.03 2 <0.001 3.95 1 0.047 Root mass (g) SR TM 0.85 2 0.652 0 1 1 Root bud mass (mg) SR TM 2.71 2 0.257 0.12 1 0.724 Root bud mass/inflorescence mass (mg g–1) FR TM 8.85 2 0.012 2.88 1 0.090 Root bud attributes Bud number SR RM 9.20 2 0.010 0.01 1 0.987 Individual root bud mass (mg) FR RM 2.49 2 0.288 0.01 1 0.904 Average bud height (mm) LOG RM 1.17 2 0.556 0 1 1 Average bud distance (cm) LOG RM 1.41 2 0.494 3.51 1 0.061 Minimum bud distance (cm) LOG RM 1.30 2 0.522 0.62 1 0.430 Maximum bud distance (cm) LOG RM 1.30 2 0.522 1.76 1 0.185 All root buds (n = 1929) Bud height (mm) LOG – 0.20 2 0.905 577.99 2 <0.0001 Bud distance (cm) LOG – 237.39 3 <0.0001 292.02 2 <0.0001 Transformations (TFs) were performed on some variables to improve normality (SR – x2, FR – x4, LOG – log10). Total plant mass (TM) or root mass (RM) were used as covariates (COV) in the models. Test statistics from mixed models of the effect of ploidy and random effects on each characteristic are shown with likelihood ratio tests (LRTs) between full and reduced models along with the P-value associated with the LRT. Random effects for models for mean bud height and mean bud distance include population and plant nested within population, while all other models include population only. Significant test results are in bold. View Large Table 3. Results of linear mixed models for plant growth, biomass allocation and root bud characteristics TF COV Ploidy Random effects LRT χ2 d.f. P LRT χ2 d.f. P Growth attributes Plant height (cm) – TM 9.97 2 0.007 0 1 1 Branches – TM 13.66 2 0.001 0.01 1 0.917 Biomass allocation Total mass (g) – – 11.54 2 0.003 0 1 1 Shoot mass (g) – TM 19.14 2 <0.0001 3.05 1 0.080 Inflorescence mass (g) – TM 17.03 2 <0.001 3.95 1 0.047 Root mass (g) SR TM 0.85 2 0.652 0 1 1 Root bud mass (mg) SR TM 2.71 2 0.257 0.12 1 0.724 Root bud mass/inflorescence mass (mg g–1) FR TM 8.85 2 0.012 2.88 1 0.090 Root bud attributes Bud number SR RM 9.20 2 0.010 0.01 1 0.987 Individual root bud mass (mg) FR RM 2.49 2 0.288 0.01 1 0.904 Average bud height (mm) LOG RM 1.17 2 0.556 0 1 1 Average bud distance (cm) LOG RM 1.41 2 0.494 3.51 1 0.061 Minimum bud distance (cm) LOG RM 1.30 2 0.522 0.62 1 0.430 Maximum bud distance (cm) LOG RM 1.30 2 0.522 1.76 1 0.185 All root buds (n = 1929) Bud height (mm) LOG – 0.20 2 0.905 577.99 2 <0.0001 Bud distance (cm) LOG – 237.39 3 <0.0001 292.02 2 <0.0001 TF COV Ploidy Random effects LRT χ2 d.f. P LRT χ2 d.f. P Growth attributes Plant height (cm) – TM 9.97 2 0.007 0 1 1 Branches – TM 13.66 2 0.001 0.01 1 0.917 Biomass allocation Total mass (g) – – 11.54 2 0.003 0 1 1 Shoot mass (g) – TM 19.14 2 <0.0001 3.05 1 0.080 Inflorescence mass (g) – TM 17.03 2 <0.001 3.95 1 0.047 Root mass (g) SR TM 0.85 2 0.652 0 1 1 Root bud mass (mg) SR TM 2.71 2 0.257 0.12 1 0.724 Root bud mass/inflorescence mass (mg g–1) FR TM 8.85 2 0.012 2.88 1 0.090 Root bud attributes Bud number SR RM 9.20 2 0.010 0.01 1 0.987 Individual root bud mass (mg) FR RM 2.49 2 0.288 0.01 1 0.904 Average bud height (mm) LOG RM 1.17 2 0.556 0 1 1 Average bud distance (cm) LOG RM 1.41 2 0.494 3.51 1 0.061 Minimum bud distance (cm) LOG RM 1.30 2 0.522 0.62 1 0.430 Maximum bud distance (cm) LOG RM 1.30 2 0.522 1.76 1 0.185 All root buds (n = 1929) Bud height (mm) LOG – 0.20 2 0.905 577.99 2 <0.0001 Bud distance (cm) LOG – 237.39 3 <0.0001 292.02 2 <0.0001 Transformations (TFs) were performed on some variables to improve normality (SR – x2, FR – x4, LOG – log10). Total plant mass (TM) or root mass (RM) were used as covariates (COV) in the models. Test statistics from mixed models of the effect of ploidy and random effects on each characteristic are shown with likelihood ratio tests (LRTs) between full and reduced models along with the P-value associated with the LRT. Random effects for models for mean bud height and mean bud distance include population and plant nested within population, while all other models include population only. Significant test results are in bold. View Large All mixed models had source population as a random effect, and the two models, including all root bud height and distance measurements, contained both population and individual plant nested within population as random factors. To assess the contribution of random effects, the full model was compared with an analogous analysis of variance (ANOVA) model with no random component. We tested for a relationship between root bud growth (height) and location along the root (distance from the primary stem) for all root buds identified using linear regression and linear mixed models in the R function ‘lmer’. Linear regression was performed on the raw root bud height and distance data across all plants to investigate their relationship for all cytotypes. Root bud height was then set as the response variable in a linear mixed model, while distance along the root, ploidy and their interaction were fixed effects. Both root bud height and distance were log transformed to improve model residual normality. Source population and individual plant nested within population were included in the model as random effects. Significance testing for fixed and random factors was performed through LRTs as described above. RESULTS Growth, biomass allocation and root bud traits Cytotypes differed with respect to several growth traits, biomass allocation patterns and root bud attributes (Table 3; for raw variable distributions and means, see Supplementary Data Figs S1 and S2; Table S1). Tetraploids were significantly taller than both diploids and neotetraploids, while there was no difference in plant height between diploids and neotetraploids (Table 3). Neotetraploids had a significantly less branched growth pattern than diploids and tetraploids, while diploids and tetraploids did not differ in axillary branch number (Table 3). Diploids had a greater mean total biomass than both naturally occurring tetraploids and neotetraploids, whereas tetraploids and neotetraploids had comparable values (Table 3; Fig. 2B). Investment in shoot mass relative to total plant biomass was significantly higher in neotetraploids than in either diploids or tetraploids (68 % of total biomass vs. 54 and 58 %; Table 3; Fig. 2A, C). There were no differences in the relative investment in roots between cytotypes (Table 3; Fig. 2A, D). Mean inflorescence biomass in neotetraploids was significantly lower than in diploid and tetraploid plants (54 and 40 % lower), whereas tetraploid and diploid means were statistically indistinguishable (Table 3; Fig. 2E). Investment in inflorescence biomass relative to total plant biomass was lowest in neotetraploids; diploids and tetraploids invested 34 and 31 %, respectively, of their total biomass in inflorescences vs. only 21 % in neotetraploids (Fig. 2A). The raw mean root bud mass in diploids was greater than that in both neotetraploids and tetraploids (53 and 75 % higher; Supplementary Data Table S1), and neotetraploids had a higher mean total root bud mass than tetraploids (47 % higher; Supplementary Data Table S1), but no differences were significant due to high variability in total root bud mass (Table 3; Fig. 2F). The proportion of root bud mass relative to total plant biomass was similar across cytotypes (Fig. 2A). Fig. 2. View largeDownload slide Biomass allocation (A) for diploids (2x), neotetraploids (4xNeo) and naturally occurring tetraploids (4x). The percentage of the total biomass that a category represents for each cytotype is shown in white in each bar, and root bud biomass measures have been adjusted to the scale of the other biomass categories by adding 10 % of the average total biomass for each cytotype. (B–G) The adjusted least-square means (± s.e.) for each biomass category. Letters in (B–G) represent the Tukey’s HSD groupings for each cytotype according to the mixed models in Table 3. Fig. 2. View largeDownload slide Biomass allocation (A) for diploids (2x), neotetraploids (4xNeo) and naturally occurring tetraploids (4x). The percentage of the total biomass that a category represents for each cytotype is shown in white in each bar, and root bud biomass measures have been adjusted to the scale of the other biomass categories by adding 10 % of the average total biomass for each cytotype. (B–G) The adjusted least-square means (± s.e.) for each biomass category. Letters in (B–G) represent the Tukey’s HSD groupings for each cytotype according to the mixed models in Table 3. Largely due to differences in allocation to sexual structures, there was a marked difference in the root bud:inflorescence biomass ratio between neotetraploids and tetraploids. Neotetraploids had a 71 % higher root bud mass to inflorescence mass ratio than tetraploids, but diploids did not differ from either neotetraploids or tetraploids (Table 3; Fig. 2G). Neotetraploids produced a significantly higher mean number of root buds than both tetraploids and diploids (Table 3; Fig. 3A). Root bud production in neotetraploids was 29 % higher than in diploids and 49 % higher than in natural tetraploids (Supplementary Data Table S1). Diploids produced more root buds than tetraploids (27 % more; Supplementary Data Table S1), but this difference was not significant (Table 3; Fig. 3A). Fig. 3. View largeDownload slide Adjusted least-square mean values (± s.e.) of root bud attributes for diploids (2x), neotetraploids (4xNeo) and naturally occurring tetraploids (4x). (D) Minimum, mean and maximum root bud distances for each cytotype. Letters in (A–D) represent the Tukey’s HSD groupings for each cytotype according to the mixed models in Table 3. Fig. 3. View largeDownload slide Adjusted least-square mean values (± s.e.) of root bud attributes for diploids (2x), neotetraploids (4xNeo) and naturally occurring tetraploids (4x). (D) Minimum, mean and maximum root bud distances for each cytotype. Letters in (A–D) represent the Tukey’s HSD groupings for each cytotype according to the mixed models in Table 3. Diploids had a mean individual root bud mass approximately twice that of both tetraploids and neotetraploids (Supplementary Data Table S1), but these differences were not significant due to large variances in mass among the buds per individual (Table 3; Fig. 3B). No differences were found between the cytotypes for mean root bud height per individual or for all root buds measured (Table 3; Fig. 3C), though diploids tended to have higher raw mean values in comparison with tetraploids and neotetraploids (Supplementary Data Table S1). The mean average root bud distance per individual was not significantly different between the cytotypes (Table 3; Fig. 3D). However, when the distance measurements from all 1929 root buds were included, diploids had a higher average distance than neotetraploids (Table 3). There were no cytotype differences in mean minimum and maximum root bud distances (Table 3; Fig. 3D). Root bud height vs. distance There was no significant relationship between root bud height and distance along the root for any of the three cytotypes. Fitting linear models to the raw distance and height data for each root bud measured revealed that the linear line of best fit for neotetraploids and tetraploids had slopes that were not different from zero (Supplementary Data Fig. S3; 4xNeo = 0.006 ± 0.019 s.e., P > 0.05; 4x = 0.031 ± 0.040, P > 0.05). The diploid model had a slightly negative slope estimate, though it was also not statistically different from zero (Supplementary Data Fig. S3; 2x = –0.039 ± 0.020, P > 0.05). The results of the linear mixed model with ploidy as a covariate showed that neither root bud distance nor ploidy had a significant relationship with root bud height (distance, LRT χ2 = 5.80, d.f. = 3, P = 0.122; ploidy, LRT χ2 = 5.80, d.f. = 4, P = 0.215). The interaction between root bud distance and ploidy was marginally significant (distance × ploidy: LRT χ2 = 5.61, d.f. = 2, P = 0.061), indicating that there may be weak differences in the relationship between root bud height and distance between the three cytotypes. DISCUSSION In this study, newly synthesized neotetraploid C. angustifolium produced more but smaller root buds than diploids, and invested fewer resources in sexual vs. clonal reproduction. Naturally occurring tetraploids produced fewer root buds than neotetraploids, and allocated more biomass towards sexual structures than root bud production. Consequently, naturally occurring tetraploids were generally more similar to diploids in their patterns of clonal and sexual reproduction than they were to neotetraploids. Overall, we find that WGD induces immediate increases in root bud production in C. angustifolium, supporting the expectation that clonal reproduction is higher in polyploids. However, there is also evidence that evolutionary processes acting after the WGD event continue to reshape resource allocation between sexual and clonal reproduction in established tetraploids. Selection appears to be operating in a direction opposite to predictions, resulting in naturally occurring tetraploids that are less clonal than neotetraploids and have levels of clonal reproduction comparable with diploids. Immediate phenotypic shifts due to WGD The neotetraploids used in this study were synthesized using colchicine, a technique that induces genome duplication by inhibiting chromosome segregation during mitosis (Caperta et al., 2006). The application of chemicals to induce WGD has the potential to create lasting side effects affecting growth, morphology and reproduction in new polyploids. In a recent study, Husband et al. (2016) examined the effects of colchicine in C. angustifolium by comparing the phenotypes of unexposed diploids with that of exposed diploids that did not convert into neotetraploids. They found significant phenotypic differences between unexposed and exposed diploids in the exposed generation, but no differences in the F1 seed generated from those plants. Studies on Fragaria and arabidopsis show similarly weak long-term effects of colchicine between treated and untreated diploids (Kwok, 2013; A. Green and B. C. Husband, University of Guelph, Canada, unpubl. res.). In contrast, Münzbergová (2017) found differences in plant performance between natural and second-generation synthetic polyploids, and evidence that selection during colchicine treatment could influence phenotype. While we cannot completely dismiss the possibility of selection in our synthetic polyploids, we have endeavoured to minimize its impact by treating a large number of diploid seedlings to maximize variation in successfully converted plants, and by using all converted plants to generate the individuals used in this study. Because previous studies on C. angustifolium have found no transgenerational effects of colchicine in F1 plants, it is likely that lingering effects are minimal and the differences observed here between diploids and neopolyploids can be directly attributed to genome duplication. How might WGD directly affect root bud production? Adventitious root buds in C. angustifolium form endogenously from root tissue, often at the junction of lateral roots (Stocklin, 1992; Klimešová et al., 2009). Genetically regulated hormone production controls meristem growth and organogenesis in plants (Reinhardt et al., 2000; Horvath et al., 2003), and lateral root development is also governed by hormonal gradients (e.g. auxin and brassinosteroids; Vanstraelen and Benkova, 2012; Taylor-Teeples et al., 2016). The specific effects of genome duplication and polyploidy on these growth patterns are little researched (Levin, 2002), but a handful of studies demonstrate that meristem growth and the expression of hormone-controlling genes can be altered in polyploids (Hatano et al., 2012; Cheng et al., 2015; Dai et al., 2015). No studies have determined the immediate genetic or hormonal effects of WGD on adventitious root bud formation, or any other form of clonal reproduction. However, there is some evidence that WGD can cause abrupt changes in root size, root morphology and lateral root initiation (Kim et al., 2004; Tavan et al., 2015). It is plausible that the effect of WGD on gene expression, epigenetic interactions, chromosomal segregation and cell growth could have a large influence on across-plant hormonal balance and tissue development in C. angustifolium (Klimešová et al., 2009). If rates of lateral meristem formation below-ground increase due to altered hormone concentrations (Swarup et al., 2008), this could potentially result in an overproduction of adventitious root buds in neotetraploids. The precise developmental nature of root buds in C. angustifolium is unclear (Stocklin, 1992; Klimešová et al., 2009), but changes in gene expression could work either to stimulate higher root bud formation around lateral root sites (Baird et al., 1992; Sharma et al., 1993) or preferentially to ‘switch’ undetermined primordia into adventitious root buds vs. lateral root initials (Schirman and Zamora, 1978; Ellmore, 1981; Kirschbaum et al., 2004). Root bud production in neotetraploids may also be the result of a trade-off with sexual effort. We see two key shifts above-ground in neotetraploids compared with diploids: (1) they have fewer lateral branches; and (2) they have approximately half the dry inflorescence biomass (Fig. 2E; Table 3; Supplementary Data Table S1). As in the root system, WGD could influence meristem and shoot growth above-ground by altering gene expression and hormone production. Interestingly, WGD has resulted in increased root bud production below-ground but decreased lateral branch growth above-ground, implying that hormone-regulated pathways for root vs. shoot systems may be affected by WGD in separate and opposing ways (Christianson and Warnick, 1983; Taylor-Teeples et al., 2016). Fewer lateral branches and fewer structures with the ability to bear flowers present a morphological constraint on sexual reproduction in neotetraploids, and they subsequently have a much lower investment in sexual reproduction than diploids. Additionally, though not assessed here, sex function in neopolyploids is often negatively impacted by meiotic irregularities leading to gamete sterility and seed abortion (Stebbins, 1950; Levin, 2002; Cifuentes et al., 2010). A decrease in sexual structures or non-functional sexual reproduction could lead to a restructuring of resource allocation and a compensatory increase in clonal reproduction (e.g. Geber et al., 1992; Vallejo-Marin et al., 2010; Van Drunen and Dorken, 2012). Studies that have found increases in clonal reproduction in extant polyploids often involve sexually sterile odd ploidies (Bruneau and Anderson, 1988; Hroudová and Zákravský, 1993; Holmes et al., 2009), where clonal reproduction may be offsetting the lack of sexual reproduction. Further study of the phenology of flowering and root bud production in C. angustifolium may reveal whether resource trade-offs between reproductive modes are occurring in neotetraploids. Our data suggest that the reproductive shifts induced by WGD could considerably alter ecological interactions in neotetraploid C. angustifolium, and could impact their establishment potential. Diploid root buds elongated faster than neotetraploid root buds (Supplementary Data Figs S3 and S4), but were located further away from the parent shoot. Because neotetraploids produce more root buds, neotetraploid genets in natural populations may cover an area comparable with diploid genets but have a denser concentration of ramets closer to the parent stem. Large, aggregated genets would increase rates of geitonogamous selfing between ramets in neotetraploids (Vallejo-Marin et al., 2010) and, since inbreeding depression in neotetraploid C. angustifolium is much lower than that of diploids and naturally occurring tetraploids (Ozimec and Husband, 2011), we expect that the increase in same-cytotype mating will assist in overcoming MCE (Levin, 1975; Husband et al., 2008). Moreover, while fast growing diploid root buds could convey advantages in very competitive environments (e.g. old populations), higher root bud numbers could lead to rapid local habitat colonization for neotetraploids at population edges and contribute to polyploid persistence. Follow-up research conducted in natural populations to determine the consequences of higher root bud production in neotetraploid C. angustifolium is needed to understand further how phenotypic shifts in clonal investment and ramet location can impact the fitness and establishment dynamics of new polyploids. Creating synthetic neopolyploids to explore the early stages of polyploid evolution requires that neopolyploids be created from extant diploid plants. Present-day diploids might differ both genetically and phenotypically from the diploids that gave rise to the established polyploids of the present day (Ramsey, 2011), and the neotetraploid C. angustifolium generated in this study may not reflect the phenotypes of historical neotetraploids. Though it is difficult to estimate the impact of these differences when hypothesizing how phenotypic shifts could have affected original polyploid establishment, they could be mitigated in young autopolyploids where presumably little change in diploids has occurred. Though the age of C. angustifolium polyploids is unknown, there is evidence that naturally occurring tetraploids in C. angustifolium have arisen recurrently and frequently (Roy, 2008), so the phenotypes of newly synthesized polyploids are still highly relevant to current population processes in this species. Phenotypic change in established polyploids The differences in root bud production and inflorescence biomass between neotetraploids and naturally occurring tetraploids indicate strong selection to suppress clonality and increase sexual reproduction after the WGD event. Furthermore, the phenotypic similarities between naturally occurring tetraploids and diploids is at odds with the prevalent hypothesis that polyploids are more clonal than their diploid relatives. The habitat and ecological conditions that neotetraploid C. angustifolium would encounter in natural populations could play an important role in the shift away from clonal reproduction and towards sexual reproduction seen in naturally occurring tetraploids. Chamerion angustifolium is typically found in open areas with low competition (Mosquin, 1966), and is a rapid colonizer of disturbed areas (Myerscough and Whitehead, 1966; Stocklin, 1992). Diploids and naturally occurring tetraploids exhibit traits typical of such pioneers; sexual fruit and seed production are high (Stocklin, 1992), outcrossing is promoted (Husband and Schemske, 1997) and fast localized spread through clonal reproduction is an additional asset (Mosquin, 1966; Myerscough and Whitehead, 1966; Stocklin, 1992). While the skewed reliance on clonal reproduction seen in neotetraploids may initially aid in establishment, this allocation pattern may ultimately prove to be maladaptive in the common environment shared with their diploid parents. As tetraploids become naturalized, and perhaps sexual reproduction becomes more viable as chromosome segregation normalizes (Cifuentes et al., 2010; Hollister, 2015), we might expect to see a shift towards higher levels of sexual vs. clonal reproduction in order to optimize fitness. The establishment period for new polyploids offers many opportunities for selection to drive phenotypic change. The data from this study indicate that neotetraploid genets have the potential to be large and dense, and may experience increased rates of geitonogamous selfing in the early stages of establishment. Due to the low inbreeding depression of neotetraploids (Husband et al., 2008), the high rate of self-fertilization is predicted to have relatively minor effects on overall seed production and viability. However, previous research has shown that neotetraploid C. angustifolium subject to several generations of self-fertilization have increased inbreeding depression at a level similar to that of diploids and naturally occurring tetraploids (Husband et al., 2008; Ozimec and Husband, 2011). Even with minimal sexual output in neotetraploids, the increasingly negative effects of inbreeding would probably cause selection pressure to decrease root bud production and genet size until clonal reproduction in tetraploids is comparable with that in diploids. Indeed, we find that total root bud production in naturally occurring tetraploids is generally lower than, but statistically similar to, that in diploids. This conclusion corroborates that of Baldwin and Husband (2013), where tetraploids produced fewer root buds and surveys of eight mixed- and single-ploidy populations of C. angustifolium found that tetraploid genets had fewer ramets than diploid genets. CONCLUSIONS A recent phylogenetic comparative study showed a strong evolutionary correlation between clonal reproduction and polyploidy in the central European flora (Herben et al., 2017), while early across-species surveys by Müntzing (1936) and Gustafsson (1948) revealed higher chromosome numbers and incidence of polyploidy in perennial or ‘root-wandering’ species. The population-level ecological and evolutionary mechanisms behind these large-scale patterns remain unclear, as researchers often do not find that extant polyploids are more clonal than close diploid relatives in mixed-ploidy species (Keeler, 2004; Baldwin and Husband, 2013; Schulze et al., 2013; Hanzl et al., 2014; Martínková et al., 2015; but see Schlaepfer et al., 2010). Clonal species may be pre-disposed towards generating successful polyploid lineages by facilitating polyploid establishment and persistence (Stebbins, 1950; Freeling, 2017; Herben et al., 2017), even if no quantitative differences in clonal reproduction are found between natural cytotypes. If, as we find in C. angustifolium, WGD immediately causes phenotypic shifts resulting in increased clonal output in neopolyploids, polyploid establishment may be even more rapidly accomplished. This study is the first to measure the immediate influence of WGD and polyploidy on clonal reproduction using synthetic neopolyploids. To understand further how WGD can affect resource investment in clonality and sexual reproduction, and the implications of such changes on the evolutionary fate of neopolyploids, more research should be conducted on synthetic neopolyploids or polyploids of recent origin, and on species with different modes of clonal reproduction. Moreover, the realized consequences of clonality on the establishment potential of polyploids remains to be studied. Experiments determining how clonal reproduction affects selfing and outcrossing patterns in natural mixed-ploidy populations will provide valuable insight into how evolutionary interactions between clonal reproduction and polyploidy are operating within the angiosperms. SUPPLEMENTARY DATA Supplementary data are available online at https://academic.oup.com/aob and consist of the following. Figure S1: biomass allocations with raw data distributions. Figure S2: root bud attributes with raw data distributions. Figure S3: regression of root bud height vs. distance for diploids, neotetraploids and naturally occurring tetraploids. Figure S4: distribution of root bud heights per ploidy for all measured root buds. Table S1: raw means of growth, biomass and root bud traits for diploids, neotetraploids and naturally occurring tetraploids. ACKNOWLEDGEMENTS The authors thank M. Mucci and T. Slimmon at the University of Guelph Phytotron for greenhouse support; and P. Kron, A. Abbruzzese and J. Seery for technical and data collection assistance. This work was supported by the Natural Science and Engineering Research Council of Canada (NSERC) with a Canada Graduate Scholarship (CGSD3-438830-2013) to W.E.V.D. and a Discovery Grant (400434) to B.C.H. LITERATURE CITED Baack EJ . 2005 . To succeed globally, disperse locally: effects of local pollen and seed dispersal on tetraploid establishment . Heredity 94 : 538 – 546 . Google Scholar CrossRef Search ADS PubMed Baird JH , Dute RR , Dickens R . 1992 . Ontogeny, anatomy, and reproductive biology of vegetative reproductive organs of Diodia virginiana L. (Rubiaceae) . International Journal of Plant Sciences 153 : 320 – 328 . Google Scholar CrossRef Search ADS Baldwin SJ , Husband BC . 2013 . The association between polyploidy and clonal reproduction in diploid and tetraploid Chamerion angustifolium . Molecular Ecology 22 : 1806 – 1819 . Google Scholar CrossRef Search ADS PubMed Barringer BC . 2007 . Polyploidy and self-fertilization in flowering plants . American Journal of Botany 94 : 1527 – 1533 . Google Scholar CrossRef Search ADS PubMed Bates D , Maechler M , Bolker B , Walker S . 2015 . Fitting linear mixed-effects models using lme4 . Journal of Statistical Software 67 : 1 – 48 . Google Scholar CrossRef Search ADS Bretagnolle F , Lumart F . 1995 . Bilateral polyploidization in Dactylis glomerata L. subsp. lusitanica: occurrence, morphological and genetic characteristics of first polyploids . Euphytica 84 : 197 – 207 . Google Scholar CrossRef Search ADS Bruneau A , Anderson GJ . 1988 . Reproductive biology of diploid and triploid Apios americana (Leguminosae) . American Journal of Botany 75 : 1876 – 1883 . Google Scholar CrossRef Search ADS Caperta AD , Delgado M , Ressurreição F , et al. 2006 . Colchicine-induced polyploidization depends on tubulin polymerization in c-metaphase cells . Protoplasma 227 : 147 – 153 . Google Scholar CrossRef Search ADS PubMed Charpentier A . 2002 . Consequences of clonal growth for plant mating . Evolutionary Ecology 15 : 521 – 530 . Google Scholar CrossRef Search ADS Cheng S , Zhu X , Liao T , et al. 2015 . Gene expression differences between high-growth Populus allotriploids and their diploid parents . Forests 6 : 839 – 857 . Google Scholar CrossRef Search ADS Christianson ML , Warnick DA . 1983 . Competence and determination in the process of in vitro shoot organogenesis . Developmental Biology 95 : 288 – 293 . Google Scholar CrossRef Search ADS PubMed Cifuentes M , Grandont L , Moore G , Chèvre AM , Jenczewski E . 2010 . Genetic regulation of meiosis in polyploid species: new insights into an old question: research review . New Phytologist 186 : 29 – 36 . Google Scholar CrossRef Search ADS PubMed Comai L . 2005 . The advantages and disadvantages of being polyploid . Nature Reviews. Genetics 6 : 836 – 846 . Google Scholar CrossRef Search ADS PubMed Dai F , Wang Z , Luo G , Tang C . 2015 . Phenotypic and transcriptomic analyses of autotetraploid and diploid mulberry (Morus alba L.) . International Journal of Molecular Sciences 16 : 22938 – 22956 . Google Scholar CrossRef Search ADS PubMed Eckert CG . 2002 . The loss of sex in clonal plants . Evolutionary Ecology 15 : 501 – 520 . Google Scholar CrossRef Search ADS Ellmore GS . 1981 . Root dimorphism in Ludwigia peploides (Onagraceae): development of two root types from similar primordia . Botanical Gazette 142 : 525 – 533 . Google Scholar CrossRef Search ADS Freeling M . 2017 . Picking up the ball at the K/Pg boundary: The distribution of ancient polyploidies in the plant phylogenetic tree as a spandrel of asexuality with occasional sex . The Plant Cell 29 : 202 – 206 . Google Scholar CrossRef Search ADS PubMed Geber MA , Watson MA , Furnish R . 1992 . Genetic differences in clonal demography in Eichhornia crassipes . Journal of Ecology 80 : 329 – 341 . Google Scholar CrossRef Search ADS Guo M , Davis D , Birchler JA . 1996 . Dosage effects on gene expression in a maize ploidy series . Genetics 142 : 1349 – 1355 . Google Scholar PubMed Gustafsson Å . 1948 . Polyploidy, life-form and vegetative reproduction . Hereditas 34 : 1 – 22 . Google Scholar CrossRef Search ADS Hanzl M , Kola F , Novakova D , Suda J . 2014 . Nonadaptive processes governing early stages of polyploid evolution: insights from a primary contact zone of relict serpentine Knautia arvensis (Caprifoliaceae) . American Journal of Botany 101 : 935 – 945 . Google Scholar CrossRef Search ADS PubMed Hatano H , Mizuno N , Matsuda R , Shitsukawa N , Park P , Takumi S . 2012 . Dysfunction of mitotic cell division at shoot apices triggered severe growth abortion in interspecific hybrids between tetraploid wheat and Aegilops tauschii . New Phytologist 194 : 1143 – 1154 . Google Scholar CrossRef Search ADS PubMed Herben T , Suda J , Klimešová J . 2017 . Polyploid species rely on vegetative reproduction more than diploids: a re-examination of the old hypothesis . Annals of Botany 120 : 341 – 349 . Google Scholar CrossRef Search ADS PubMed Hollister JD . 2015 . Polyploidy: adaptation to the genomic environment . New Phytologist 205 : 1034 – 1039 . Google Scholar CrossRef Search ADS PubMed Holmes GD , James EA , Hoffmann AA . 2009 . Divergent levels of genetic variation and ploidy among populations of the rare shrub, Grevillea repens (Proteaceae) . Conservation Genetics 10 : 827 – 837 . Google Scholar CrossRef Search ADS Honnay O , Bossuyt B . 2005 . Prolonged clonal growth: escape route or route to extinction ? Oikos 108 : 427 – 432 . Google Scholar CrossRef Search ADS Horvath DP , Anderson JV , Chao WS , Foley ME . 2003 . Knowing when to grow: signals regulating bud dormancy . Trends in Plant Science 8 : 534 – 540 . Google Scholar CrossRef Search ADS PubMed Hroudová Z , Zákravsky P . 1993 . Ecology of two cytotypes of Butomis umbellatus II. Reproduction, growth, and biomass production . Folia Geobotanica et Phytotaxonomica 28 : 413 – 424 . Google Scholar CrossRef Search ADS Husband BC . 2000 . Constraints on polyploid evolution: a test of the minority cytotype exclusion principle . Proceedings of the Royal Society B: Biological Sciences 267 : 217 – 223 . Google Scholar CrossRef Search ADS Husband BC , Sabara HA . 2004 . Reproductive isolation between autotetraploids and their diploid progenitors in fireweed, Chamerion angustifolium (Onagraceae) . New Phytologist 161 : 703 – 713 . Google Scholar CrossRef Search ADS Husband BC , Schemske DW . 1997 . The effect of inbreeding in diploid and tetraploid populations of Epilobium angustifolium (Onagraceae): implications for the genetic basis of inbreeding depression . Evolution 51 : 737 – 746 . Google Scholar CrossRef Search ADS PubMed Husband BC , Schemske DW . 2000 . Ecological mechanisms of reproductive isolation between diploid and tetraploid Chamerion angustifolium . Journal of Ecology 88 : 689 – 701 . Google Scholar CrossRef Search ADS Husband BC , Ozimec B , Martin SL , Pollock L . 2008 . Mating consequences of polyploid evolution in flowering plants: current trends and insights from synthetic polyploids . International Journal of Plant Sciences 169 : 195 – 206 . Google Scholar CrossRef Search ADS Husband BC , Baldwin SJ , Sabara HA . 2016 . Direct vs. indirect effects of whole-genome duplication on prezygotic isolation in Chamerion angustifolium: implications for rapid speciation . American Journal of Botany 103 : 1259 – 1271 . Google Scholar CrossRef Search ADS PubMed Keeler KH . 2004 . Impact of intraspecific polyploidy in Andropogon gerardii (Poaceae) populations . American Midland Naturalist 152 : 63 – 74 . Google Scholar CrossRef Search ADS Kim YS , Hahn EJ , Murthy HN , Paek KY . 2004 . Effect of polyploidy induction on biomass and ginsenoside accumulations in adventitious roots of ginseng . Journal of Plant Biology 47 : 356 – 360 . Google Scholar CrossRef Search ADS Kirschbaum DS , Cantliffe DJ , Shaw NL , Liu JR . 2004 . Direct adventitious shoot formation on seedling radicles in seed cultures of strawberry . Journal of Plant Biology 47 : 160 – 162 . Google Scholar CrossRef Search ADS van K leunen M , Fischer M , Schmid B . 2002 . Experimental life-history evolution: selection on the allocation to sexual reproduction and its plasticity in a clonal plant . Evolution 56 : 2168 – 2177 . Google Scholar CrossRef Search ADS PubMed Klimešová J , De Bello F . 2009 . CLO-PLA: the database of clonal and bud bank traits of Central European flora . Journal of Vegetation Science 20 : 511 – 516 . Google Scholar CrossRef Search ADS Klimešová J , Pokorná A , Klimeš L . 2009 . Establishment growth and bud-bank formation in Epilobium angustifolium: the effects of nutrient availability, plant injury, and environmental heterogeneity . Botany 87 : 195 – 201 . Google Scholar CrossRef Search ADS Kolář F , Čertner M , Suda J , Schönswetter P , Husband BC . 2017 . Mixed-ploidy species: progress and opportunities in polyploid research . Trends in Plant Science 22 : 1041 – 1055 . Google Scholar CrossRef Search ADS PubMed Kron P , Husband BC . 2009 . Hybridization and the reproductive pathways mediating gene flow between native Malus coronaria and domestic apple, M. domestica . Botany 87 : 864 – 874 . Google Scholar CrossRef Search ADS Kwok A . 2013 . The role of polyploidy in the evolution of gender dimorphism: an experimental approach using Fragaria vesca . MSc Thesis, University of Guelph , Canada . Lande R , Schemske DW . 1985 . The evolution of self-fertilization and inbreeding depression in plants. I: genetic models . Evolution 39 : 24 – 40 . Google Scholar CrossRef Search ADS PubMed Levin DA , Kerster HW . 1974 . Gene flow in seed plants . Evolutionary Biology 7 : 139 – 220 . Levin DA . 1975 . Minority cytotype exclusion in local plant populations . Taxon 24 : 35 – 43 . Google Scholar CrossRef Search ADS Levin DA . 1983 . Polyploidy and novelty in flowering plants . American Naturalist 122 : 1 – 25 . Google Scholar CrossRef Search ADS Levin DA . 2002 . The role of chromosomal change in plant evolution . Oxford : Oxford University Press . Lim KY , Soltis DE , Soltis PS , et al. 2008 . Rapid chromosome evolution in recently formed polyploids in Tragopogon (Asteraceae) . PLoS One 3 : e3353 . doi: 10.1371/journal.pone.0003353 . Google Scholar CrossRef Search ADS PubMed Maherali H , Walden AE , Husband BC . 2009 . Genome duplication and the evolution of physiological responses to water stress . New Phytologist 184 : 721 – 731 . Google Scholar CrossRef Search ADS PubMed Martin SL , Husband BC . 2012 . Whole genome duplication affects evolvability in natural populations of a flowering plant . PLoS One 7 : e44784 . doi: 10.1371/journal.pone.0044784 Google Scholar CrossRef Search ADS PubMed Martínková J , Klimešová J , Doležal J , Kolář F . 2015 . Root sprouting in Knautia arvensis (Dipsacaceae): effects of polyploidy, soil origin and nutrient availability . Plant Ecology 216 : 901 – 911 . Google Scholar CrossRef Search ADS Mosquin T . 1966 . A new taxonomy for Epilobium angustifolium L. (Onagraceae) . Brittonia 18 : 167 – 188 . Google Scholar CrossRef Search ADS Müntzing A . 1936 . The evolutionary significance of autopolyploidy . Hereditas 21 : 263 – 378 . Münzbergová Z . 2017 . Colchicine application significantly affects plant performance in the second generation of synthetic polyploids and its effects vary between populations . Annals of Botany 120 : 329 – 339 . Google Scholar CrossRef Search ADS PubMed Myerscough PJ , Whitehead FH . 1966 . Comparative biology of Tussilago farfara L., Chamaenerion angustifolium (L.) Scop., Epilobium montanum L. and Epilobium adenocaulon Hausskn. I. General biology and germination . New Phytologist 65 : 192 – 210 . Google Scholar CrossRef Search ADS Osborn TC , Pires JC , Birchler JA , et al. 2003 . Understanding mechanisms of novel gene expression in polyploids . Trends in Genetics 19 : 141 – 147 . Google Scholar CrossRef Search ADS PubMed Oswald BP , Nuismer SL . 2010 . Neopolyploidy and diversification in Heuchera grossulariifolia . Evolution 65-6 : 1667 – 1679 . Otto SP , Whitton J . 2000 . Polyploid incidence and evolution . Annual Review of Genetics 34 : 401 – 437 . Google Scholar CrossRef Search ADS PubMed Ozimec B , Husband BC . 2011 . Effect of recurrent selfing on inbreeding depression and mating system evolution in an autopolyploid plant . Evolution 65 : 2038 – 2049 . Google Scholar CrossRef Search ADS PubMed Parisod C , Holderegger R , Brochmann C . 2010 . Evolutionary consequences of autopolyploidy: research review . New Phytologist 186 : 5 – 17 . Google Scholar CrossRef Search ADS PubMed R Core Team . 2017 . R: a language and environment for statistical computing . Vienna, Austria : R Foundation for Statistical Computing . https://www.R-project.org/ Ramsey J . 2011 . Polyploidy and ecological adaptation in wild yarrow . Proceedings of the National Academy of Sciences, USA 108 : 7096 – 7101 . Google Scholar CrossRef Search ADS Ramsey J , Schemske DW . 2002 . Neopolyploidy in flowering plants . Annual Review of Ecology and Systematics 33 : 589 – 639 . Google Scholar CrossRef Search ADS Reinhardt D , Mandel T , Kuhlemeier C . 2000 . Auxin regulates the initiation and radial position of plant lateral organs . The Plant Cell 12 : 507 – 518 . Google Scholar CrossRef Search ADS PubMed Robertson K , Goldberg EE , Igić B . 2011 . Comparative evidence for the correlated evolution of polyploidy and self-compatibility in Solanaceae . Evolution 65 : 139 – 155 . Google Scholar CrossRef Search ADS PubMed Rodriguez DJ . 1996 . A model for the establishment of polyploidy in plants: viable but infertile hybrids, iteroparity, and demographic stochasticity . Journal of Theoretical Biology 180 : 189 – 196 . Google Scholar CrossRef Search ADS Roulin A , Auer PL , Libault M , et al. 2013 . The fate of duplicated genes in a polyploid plant genome . The Plant Journal 73 : 143 – 153 . Google Scholar CrossRef Search ADS PubMed Roy Y . 2008 . The evolutionary history of polyploidy in the herbaceous plant Chamerion angustifolium . MSc Thesis, University of Guelph , Canada . Sabara HA , Kron P , Husband BC . 2013 . Cytotype coexistence leads to triploid hybrid production in a diploid–tetraploid contact zone of Chamerion angustifolium (Onagraceae) . American Journal of Botany 100 : 962 – 970 . Google Scholar CrossRef Search ADS PubMed Schirman R , Zamora BA . 1978 . Bud development in excised roots of rush skeletonweed (Chondrilla juncea) . Weed Science 26 : 582 – 584 . Google Scholar CrossRef Search ADS Schlaepfer DR , Edwards PJ , Billeter R . 2010 . Why only tetraploid Solidago gigantea (Asteraceae) became invasive: a common garden comparison of ploidy levels . Oecologia 163 : 661 – 673 . Google Scholar CrossRef Search ADS PubMed Schulze J , Erhardt A , Stoll P . 2013 . Reduced clonal reproduction indicates low potential for establishment of hybrids between wild and cultivated strawberries (Fragaria vesca × F. × ananassa) . Ecological Research 28 : 43 – 52 . Google Scholar CrossRef Search ADS Segraves KA , Thompson JN . 1999 . Plant polyploidy and pollination: floral traits and insect visits to diploid and autotetraploid Heuchera grossulariifolia . Evolution 53 : 1114 – 1127 . Google Scholar CrossRef Search ADS PubMed Sharma KK , Yeung EC , Thorpe TA . 1993 . Histology of shoot bud ontogeny from seedling root segments of Brassica napus L . Annals of Botany 71 : 461 – 466 . Google Scholar CrossRef Search ADS Soltis DE , Buggs RJ , Doyle JJ , Soltis PS . 2010 . What we still don’t know about polyploidy . Taxon 59 : 1387 – 1403 . Soltis DE , Misra BB , Shan S , Chen S , Soltis PS . 2016 . Polyploidy and the proteome . Biochimica et Biophysica Acta 1864 : 896 – 907 . Google Scholar CrossRef Search ADS PubMed Soltis PS , Soltis DE . 2016 . Ancient WGD events as drivers of key innovations in angiosperms . Current Opinion in Plant Biology 30 : 159 – 165 . Google Scholar CrossRef Search ADS PubMed Stebbins GL . 1950 . Variation and evolution in plants . New York : Columbia University Press . Stebbins GL . 1971 . Chromosomal evolution in higher plants . Reading, UK : Addison-Wesley . Stocklin J . 1992 . Differences in the location of subcotyledonary buds among Epilobium angustifolium L., E. dodonaei Vill. and E. fleischeri Hostst. (Onagraceae) and effects on architecture and population structure . Botanical Journal of the Linnean Society 108 : 35 – 47 . Google Scholar CrossRef Search ADS Swarup K , Benková E , Swarup R , et al. 2008 . The auxin influx carrier LAX3 promotes lateral root emergence . Nature Cell Biology 10 : 946 – 954 . Google Scholar CrossRef Search ADS PubMed Tavan M , Mirjalili MH , Karimzadeh G . 2015 . In vitro polyploidy induction: changes in morphological, anatomical and phytochemical characteristics of Thymus persicus (Lamiaceae) . Plant Cell, Tissue and Organ Culture 122 : 573 – 583 . Google Scholar CrossRef Search ADS Taylor-Teeples M , Lanctot A , Nemhauser JL . 2016 . As above, so below: auxin’s role in lateral organ development . Developmental Biology 419 : 156 – 164 . Google Scholar CrossRef Search ADS PubMed Thompson JD , Lumaret R . 1992 . The evolutionary dynamics of polyploid plants: origins, establishment and persistence . Trends in Ecology and Evolution 7 : 302 – 307 . Google Scholar CrossRef Search ADS PubMed Thompson KA , Husband BC , Maherali H . 2014 . Climatic niche differences between diploid and tetraploid cytotypes of Chamerion angustifolium (Onagraceae) . American Journal of Botany 101 : 1868 – 1875 . Google Scholar CrossRef Search ADS PubMed Vallejo-Marín M . 2012 . Mimulus peregrinus (Phrymaceae): a new British allopolyploid species . PhytoKeys 14 : 1 – 14 . Google Scholar CrossRef Search ADS Vallejo-Marín M , Dorken ME , Barrett SCH . 2010 . The ecological and evolutionary consequences of clonality for plant mating . Annual Review of Ecology, Evolution, and Systematics 41 : 193 – 213 . Google Scholar CrossRef Search ADS Van Drunen WE , Dorken ME . 2012 . Trade-offs between clonal and sexual reproduction in Sagittaria latifolia (Alismataceae) scale up to affect the fitness of entire clones . New Phytologist 196 : 606 – 616 . Google Scholar CrossRef Search ADS PubMed Van Drunen WE , van Kleunen M , Dorken ME . 2015 . Consequences of clonality for sexual fitness: clonal expansion enhances fitness under spatially restricted dispersal . Proceedings of the National Academy of Sciences 112 : 8929 – 8936 . Google Scholar CrossRef Search ADS Vanstraelen M , Benková E . 2012 . Hormonal interactions in the regulation of plant development . Annual Review of Cell and Developmental Biology 28 : 463 – 487 . Google Scholar CrossRef Search ADS PubMed Weiss-Schneeweiss H , Emadzade K , Jang TS , Schneeweiss GM . 2013 . Evolutionary consequences, constraints and potential of polyploidy in plants . Cytogenetic and Genome Research 140 : 137 – 150 . Google Scholar CrossRef Search ADS PubMed Wood TE , Takebayashi N , Barker MS , Mayrose I , Greenspoon PB , Rieseberg LH . 2009 . The frequency of polyploid speciation in vascular plants . Proceedings of the National Academy of Sciences, USA 106 : 13875 – 13879 . Google Scholar CrossRef Search ADS © The Author(s) 2018. Published by Oxford University Press on behalf of the Annals of Botany Company. All rights reserved. For permissions, please e-mail: [email protected]. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)
Evolutionary diversification of the African achyranthoid clade (Amaranthaceae) in the context of sterile flower evolution and epizoochory2018 Annals of Botany
doi: 10.1093/aob/mcy055pmid: 29688271
Abstract Background and Aims Many African genera of the Amaranthaceae exhibit unique inflorescences that include sterile flowers modified to hooks or spines. Considering that the abundance of large terrestrial herbivores increased on the African continent with the expansion of grassland and savannah ecosystems, modified sterile flowers could have been an innovation that boosted the diversification of an African achyranthoid clade of Amaranthaceae, with large animals serving dispersal. Methods We generated an extensively sampled phylogeny comprising 26 of the 31 achyranthoid genera as well as representatives of all other lineages of Amaranthaceae. Phylogenetic tree inference employed four genomic regions, using parsimony, likelihood and Bayesian inference methods. We estimated divergence times, evaluated trait-dependant changes and species diversification rates using state-dependent speciation and extinction models, and reconstructed ancestral character states for modified sterile flowers. Key Results The achyranthoids were found to be a major clade of the Amaranthaceae, comprising mostly African members. Phylogenetic relationships within this clade were well resolved and supported two main subclades. Several genera were found to be polyphyletic. Our results indicate that the achyranthoids started to diversify ~28 million years ago, and that modified sterile flowers evolved multiple times. An asymmetry in transition rates towards the gain of sterile flowers was observed, whereas no trait-dependent increase in species diversification rates was detected. Bayesian rate heterogeneity analyses indicated that the achyranthoids diversified without significant rate shifts. Conclusions The accumulation of modified sterile flowers within achyranthoids appears to result from the higher transition rates in favour of modified sterile flowers. Multiple gains suggest an adaptive value for this trait. However, epizoochory does not appear to fuel species diversification, possibly due to extensive gene flow through regularly migrating mammals, which limits the possibility of speciation by isolation. Africa, Amaranthaceae, BiSSE, Caryophyllales, character evolution, epizoochory, floral morphology, HiSSE, phylogeny, species diversification INTRODUCTION The Amaranthaceae constitute a nearly cosmopolitan family of flowering plants within the order Caryophyllales, forming a well-supported lineage with the Chenopodiaceae (Cuénoud et al., 2002; Schäferhoff et al., 2009). In the strict sense (excluding Chenopodiaceae), the Amaranthaceae contain ~800 species in 79 genera (Hernández-Ledesma et al., 2015), comprising a variety of life forms (herbs, shrubs, lianas and trees) that inhabit a wide range of habitats (from deserts to rainforests). Most of the diversity of the Amaranthaceae is found in the neotropics as well as in eastern and southern Africa (Townsend, 1993). Müller and Borsch (2005a) found the subfamily Amaranthoideae and most of the tribes and subtribes in the classification of Schinz (1934) and Townsend (1993) to be para- or polyphyletic. A newly recovered achyranthoid clade included several genera of the subtribe Aervinae, while the genera Aerva, Nothosaerva and Ptilotus were identified as part of a separate lineage (the ‘aervoid clade’; Müller and Borsch 2005b). The achyranthoid clade was estimated to comprise ~134–150 species in 31 genera, if currently accepted genera were upheld. However, adequate support for the clade was lacking due to limited character and taxon sampling. We assume that, next to the mostly neotropical gomphrenoids (Sánchez-del Pino et al., 2009), which comprise about 380 species, the achyranthoid clade constitutes the second main radiation within the Amaranthaceae. The majority of achyranthoid taxa occur in Africa, except the Hawaiian genus Nototrichium, species of Achyranthes from Asia and the Pacific, and certain species of Cyathula distributed in China and the Himalaya. Members of the achyranthoid clade (Müller and Borsch, 2005b) and of many Aervinae in the pre-phylogenetic circumscription of the subtribe possess cymose inflorescence structures. In several taxa, sterile modified flowers appear within partial florescences, which fall off completely at maturity as burr-like units (Acosta et al., 2009). The sterile flowers are modified to hooks, spines or barbs, which provide adhesive structures serving epizoochory. In some genera (e.g. Achyranthes), which do not exhibit sterile flowers, the midribs of bracteoles are thickened to spines, which also form adhesive appendages. Epizoochory occurs in <5 % of seed plants (Sorensen, 1986) but is frequent in achyranthoids (>80 % of the genera). Ridley (1930) suggested that the evolution of plants exhibiting burrs as dispersal units in Africa was primarily promoted through epizoochory by herds of ungulates covering large areas of the African continent between the Eocene and the Pleistocene. He reported species of Cyathula, Achyranthes and Pupalia to be animal-dispersed by adhesion and described the diaspores of Pupalia as ‘the most persistently adhesive burrs’ he knows. Agnew and Flux (1970) as well as Mori and Brown (1998) showed that several achyranthoid species had been dispersed over long distances in the fur of animals. Most African achyranthoids occur in Acacia–Commiphora wood- and bushlands, dry evergreen afromontane forests, grassland vegetation complexes or semi-desert scrublands, all of which are vegetation types with open spaces (White, 1983; Friis et al., 2011) and are thus suitable for long-distance dispersal via epizoochory. In the present investigation, we hypothesize that the modified sterile flowers present in many African Amaranthaceae evolved during time periods when African habitats changed to more open conditions and the local diversity of terrestrial mammals increased. We propose that the diversification of the African achyranthoid clade may have been fuelled by these adaptations to epizoochory. To evaluate this hypothesis, we conducted a series of phylogenetic analyses of the Amaranthaceae by following three primary aims. First, we aimed to generate a robust molecular phylogenetic hypothesis for the achyranthoids to evaluate the composition of the achyranthoid clade initially found by Müller and Borsch (2005b). Second, we aimed to estimate divergence times and to reconstruct the evolution of modified sterile flowers based on our new phylogenetic trees. This second aim would allow us to identify the number of origins of modified sterile flowers within the achyranthoids. Third, we aimed to evaluate the influence of morphological traits that may have an adaptive value for epizoochory in speciation and extinction in the achyranthoids. In this third goal, we specifically aimed to understand (1) whether speciation and extinction rates are significantly different in the presence of the character ‘modified sterile flowers’ (and the more inclusive functional trait of ‘adhesive appendages’, which can be considered as any part of a dispersal unit with an adhesive property and an adaptive value for epizoochory, regardless of whether a taxon possess sterile flowers) than in their absence, (2) whether transition rates between character states are strongly asymmetrical, and (3) whether net diversification is higher in the presence of these features than in their absence. The specific research questions of the third aim were evaluated through a trait-dependent as well as a trait-independent approach. In the trait-dependant approach, we estimated and compared speciation, extinction and character transition rates in relation to morphological traits with a potential adaptive value for epizoochory. In the trait-independent approach, we explored possible diversification rate shifts across the entire family Amaranthaceae. Taken together, these analyses will allow us to evaluate whether the diversification of the African achyranthoid clade may have been fuelled by the emergence of modified sterile flowers in this group. MATERIALS AND METHODS Collection of plant material and taxon sampling Plant material was primarily collected during field trips to Ethiopia and Kenya in 2012 and 2013. Taxon sampling focused on the whole subtribe Aervinae sensuSchinz (1934). We included almost half of all known species belonging to the achyranthoid clade in our analyses, representing 26 of 31 genera. Our sampling was thus guided to best represent the morphological diversity in the group, independently of the current classification into genera. Also, the very few species occurring outside Africa were represented in our sampling; sampling statistics are provided in Table 1. Taxa that were recovered with minimal support or in unexpected phylogenetic positions were supplemented by additional samples of the same species from different localities to verify the inferred positions. Species names, geographical origin and herbarium voucher information for all taxa under study are given in Supplementary Data Table S1. Table 1. Number and proportions of genera and species sampled for the present investigation Genera (approximate sampling proportions) Species (approximate sampling proportions) Estimated taxon numbers of achyranthoids 31 ~134–150 Taxa in dataset A 26 (84 %) 69 (51 %) Taxa in dataset B 26 (84 %) 65 (49 %) Taxa in dataset C 23 (74 %) 63 (47 %) Genera (approximate sampling proportions) Species (approximate sampling proportions) Estimated taxon numbers of achyranthoids 31 ~134–150 Taxa in dataset A 26 (84 %) 69 (51 %) Taxa in dataset B 26 (84 %) 65 (49 %) Taxa in dataset C 23 (74 %) 63 (47 %) View Large Table 1. Number and proportions of genera and species sampled for the present investigation Genera (approximate sampling proportions) Species (approximate sampling proportions) Estimated taxon numbers of achyranthoids 31 ~134–150 Taxa in dataset A 26 (84 %) 69 (51 %) Taxa in dataset B 26 (84 %) 65 (49 %) Taxa in dataset C 23 (74 %) 63 (47 %) Genera (approximate sampling proportions) Species (approximate sampling proportions) Estimated taxon numbers of achyranthoids 31 ~134–150 Taxa in dataset A 26 (84 %) 69 (51 %) Taxa in dataset B 26 (84 %) 65 (49 %) Taxa in dataset C 23 (74 %) 63 (47 %) View Large Construction of datasets For the present investigation, we generated a total of 396 new DNA sequences. In addition, we included 112 sequences from Müller and Borsch (2005a, b) in our datasets. Three different DNA matrices were assembled: (1) a combined plastid dataset (hereafter dataset A), which consisted of the chloroplast regions trnK/matK, rpl16 and trnL-F and comprised 130 samples with a focus on the achyranthoid clade but also a representative coverage of subfamily Amaranthoideae; (2) a broader plastid dataset (dataset B), which consisted of the chloroplast region trnK/matK only and comprised a total of 189 samples from across the Caryophyllales (134 samples of Amaranthaceae, 40 of Chenopodiaceae and 15 representing clades within the Caryophyllales); and (3) a nuclear ribosomal dataset (dataset C), which consisted of internal transcribed spacer (nrITS) sequences of 105 taxa with a focus on achyranthoids plus representatives of other main Amaranthaceae clades. We used dataset A to reconstruct detailed relationships of the achyranthoid clade; dataset B was employed for divergence dating, as it allowed (1) the placement of three fossil specimens assigned to the Chenopodiaceae as primary calibration points, and (2) the definition of three secondary calibration points by using age estimates of Bell et al. (2010). Dataset C allowed a comparison of tree topologies inferred from nuclear ribosomal data with those inferred from plastid DNA regions. DNA isolation, amplification and sequencing Total genomic DNA was isolated as described in Müller and Borsch (2005a). Primer sequences and details of the PCR protocols are listed in Supplementary Data Table S2. Upon amplification, all PCR products were sequenced by standard Sanger sequencing at Macrogen Europe (Amsterdam, The Netherlands). Chromatograms were inspected by eye, corrected for erroneous nucleotide calls and assembled using PhyDE 0.9971 (Müller et al., 2010). Final DNA sequences were submitted to ENA (http://www.ebi.ac.uk/ena/) upon conversion to checklist files with custom Python scripts (https://github.com/michaelgruenstaeudl/annonex2embl). ENA/GenBank accession numbers for all taxa under study are given in Supplementary Data Table S1. Sequence alignment, indel coding and model selection Initial alignment files generated with Muscle 3.6 (Edgar, 2004) were further refined under a motif-based approach (Löhne and Borsch, 2005) using PhyDe. Areas of uncertain homology were subsequently excluded from matrices. Insertions and deletions in datasets A and C were coded according to the ‘simple indel coding’ scheme (Simmons and Ochoterena, 2000) as implemented in SeqState 1.4.0 (Müller, 2005). Inversions were reverse-complemented and coded manually. Best-fitting models of nucleotide substitution were selected via the Akaike information criterion (AIC; Akaike, 1974) using the software jModeltest 2.1.7 (Darriba et al., 2012). The substitution model GTR+G was found to display the best fit for DNA sequences of matK, the trnL intron and the trnL-F intergenic spacer. The model GTR+G+I was found to display the best fit for DNA sequences of ITS1, ITS2, rpl16 and the trnK intron, and the model HKY+I for the 5.8S ribosomal DNA. Phylogenetic analysis Phylogenetic inference was performed via maximum parsimony (MP), maximum likelihood (ML) and Bayesian inference (BI). Parsimony analyses were conducted in PAUP* 4.0b10 (Swofford, 2002), with 10 000 jackknife (JK) replicates to calculate node support. Likelihood analyses were conducted in RAxML 7.4.2 (Stamatakis, 2006), with node support calculated via 1000 rapid bootstrap (BS) replicates, using the RAxMLGui 1.3 as input interface (Silvestro and Michalak, 2012). BI was performed with MrBayes 3.2.2 (Ronquist et al., 2012), with node support calculated as posterior probabilities. Four independent Markov chain Monte Carlo (MCMC) runs with four chains and 10 000 000 generations each were conducted, with trees sampled every 1000th generation. The burn-in was set to 50 %, resulting in a combined posterior tree distribution of 20 000 trees. Majority-rule consensus trees were calculated from the combined posterior tree distribution subsampled to 1000 trees. We rated BS/JK values of 70–84 % and posterior probability of 0.9–0.94 as moderate support, and BS/JK values of 85–100 % and posterior probability of 0.95–1 as strong support. For BI and ML analyses, datasets were partitioned to allow independent parameter estimation for each partition. For analyses in RAxML, the model MULTIGAMMAI was employed as the nearest approximation to the model GTR+I+G. The binary character model (Lewis, 2001) was applied to indel partitions. DNA sequence alignments and the main phylogenetic trees inferred were submitted to TreeBASE (number 21912). For all trait-dependent and -independent analyses of speciation and extinction rates as well as our ancestral character state reconstructions, we reduced multiple samples of the same species in our phylogenetic trees to a single representative by randomly selecting one sequence. In cases of non-monophyly of multiple samples per taxon, one representative per phylogenetic position was kept. This strategy ensured that taxa were properly represented and simultaneously prevented species duplicates biasing the final results. Molecular clock calibration To calibrate the molecular clock, three fossils were assigned to different clades of the Chenopodiaceae. These fossils were: Chenopodipollis multiplex, Salicornites massalongoi and Parvangula randeckensis. Chenopodipollis multiplex had been found in a marine influence assemblage alongside other marine organisms (Nichols and Traverse, 1971) and was assigned to Chenopodiaceae by Kadereit et al. (2003). Although habitats close to coastal environments are typical of Chenopodiaceae, this is rarely the case for Amaranthaceae or Caryophyllaceae with similar pollen. Thus, we implemented a conservative calibration point for this fossil and used it as a minimum constraint on the stem node of Chenopodiaceae (excluding subfamily Polycnemoideae). The fossil S. massalongoi had been described as a fragment of a plant stem (Principi, 1926; Collinson et al., 1993) and was since assigned to the Salicornioideae (Hohmann et al., 2006; Kadereit et al., 2012). Here, we used it as a minimum constraint for the stem node of Salicornioideae, assuming that the succulent stem features appeared with the split of Salicornioideae and have fostered its diversification into dry and saline habitats. The fossil P. randeckensis had been described as Chenopodium-like seeds (Gregor, 1982) but, considering seed evolution in the Chenopodioideae (Sukhorukov and Zhang 2013), these seeds may belong to a lineage after divergence of the Axyris–Krascheninnikovia clade. The fossil was placed in different positions by different authors (Kadereit et al., 2003, 2012; Hohmann et al., 2006). We designated this fossil as a minimum age constraint for the stem of the Axyris–Krascheninnikovia clade. Maximum ages for all three fossils were set to 125 million years (Ma), which represents the mean age of the core eudicots [corresponding to node 29 of Appendix S5 in Bell et al. (2010)]. The age distribution of each of these primary calibration points was modelled as an exponential prior distribution (Ho and Philips, 2009), with an offset equal to the minimum age of the respective fossil and a mean such that the median age of the exponential distribution equalled the maximum age of the fossil. A fossil calibration so constrained renders the age estimate of the fossil the likely divergence time of the calibrated node, but also allows the inference of older ages. A summary of the primary calibration points employed in this investigation as well as their respective age ranges is given in Table 2. Table 2. Fossils and their age ranges used for primary calibration of the molecular clock No. Fossil name Description Origin Age (Ma) Epoch Publication 1 Chenopodipollis multiplex Pollen record USA 65–56 Palaeocene Nichols and Traverse (1971) 2 Salicornites massalongoi Part of an axis Italy 35–23 Oligocene Principi (1926) 3 Parvangula randeckensis Chenopodium-Like seeds Germany 23–16 Early Miocene Gregor (1982) No. Fossil name Description Origin Age (Ma) Epoch Publication 1 Chenopodipollis multiplex Pollen record USA 65–56 Palaeocene Nichols and Traverse (1971) 2 Salicornites massalongoi Part of an axis Italy 35–23 Oligocene Principi (1926) 3 Parvangula randeckensis Chenopodium-Like seeds Germany 23–16 Early Miocene Gregor (1982) View Large Table 2. Fossils and their age ranges used for primary calibration of the molecular clock No. Fossil name Description Origin Age (Ma) Epoch Publication 1 Chenopodipollis multiplex Pollen record USA 65–56 Palaeocene Nichols and Traverse (1971) 2 Salicornites massalongoi Part of an axis Italy 35–23 Oligocene Principi (1926) 3 Parvangula randeckensis Chenopodium-Like seeds Germany 23–16 Early Miocene Gregor (1982) No. Fossil name Description Origin Age (Ma) Epoch Publication 1 Chenopodipollis multiplex Pollen record USA 65–56 Palaeocene Nichols and Traverse (1971) 2 Salicornites massalongoi Part of an axis Italy 35–23 Oligocene Principi (1926) 3 Parvangula randeckensis Chenopodium-Like seeds Germany 23–16 Early Miocene Gregor (1982) View Large Additionally, we defined three secondary calibration points by using age estimates of Bell et al. (2010): the age for the crown group of the Caryophyllales (their node 39); the age for the crown group of the polygonoid clade (their node 41); and the maximum estimated age of the entire tree, corresponding to Gunneridae (their node 31). The age distribution of each of these secondary calibration points was modelled as a normal prior distribution and was equal to the 95 % highest posterior density (HPD) interval of the ages inferred by Bell et al. (2010), which were based on lognormal prior age distributions. Estimation of divergence times Divergence times were estimated using a Bayesian relaxed molecular clock model in BEAST 1.8.0 (Drummond et al., 2012). A preliminary evaluation of the nucleotide substitution rates across dataset B indicated that the rates did not conform to a strict molecular clock (Peterson, 2006). Hence, a relaxed molecular clock with uncorrelated substitution rates sampled from an exponential distribution for the estimation of divergence times was applied (Drummond et al., 2006). Lineage diversification was modelled as a birth–death process, and a random starting tree was employed. Two MCMC runs with 50 000 000 generations each were performed for the estimation of divergence times, with trees sampled every 1000 generations. We confirmed adequate parameter sampling (effective sample size >200) and the mixing of Markov chains in both runs through visual inspection of the MCMC samples using Tracer 1.4 (Rambaut and Drummond, 2007). Given the results of our sampling and convergence diagnostics, the first 50 % of the MCMC generations were removed as burn-in. The remaining trees were subsampled to every fifth generation, generating a post-burn-in posterior tree distribution of 500 trees per run. The post burn-in MCMC samples of both runs were combined via LogCombiner 1.8.0 (Drummond et al., 2012), and the combined tree distribution was summarized as a maximum clade credibility (MCC) tree via TreeAnnotator 1.8.0 (Drummond et al., 2012). Analyses of state-dependent speciation and extinction To evaluate the influence of morphological traits with a potential adaptive value for epizoochory in speciation and extinction rates in the achyranthoids, we employed a trait-dependent analysis approach using two morphological characters: modified sterile flowers and the more inclusive trait of adhesive appendages (Fig. 1). Presence (noted as 1) and absence (0) of both characters was coded for the achyranthoid taxa included in dataset B, using the same specimens as in the molecular phylogenetic analyses. Additional assessment and confirmation of the traits in the study taxa were conducted via extraction of morphological descriptions from the literature (e.g. protologues of taxa, genus and species descriptions of floras, morphological treatments) or via the examination of herbarium material. The complete character matrix is provided in Supplementary Data Table S3. Fig. 1. View largeDownload slide Flower morphology of achyranthoids. (A, B) Inflorescences of Pupalia lappacea (Di Vincenzo et al., 301, B) and Cyathula cylindrica (Wondafrash et al., 3332, B) exhibiting modified sterile flowers. (C) Adhesive dispersal unit of Cyathula orthacantha (Di Vincenzo et al., 21, B) with sterile flowers modified to spines. (D) Inflorescence of Achyranthes aspera (Di Vincenzo et al., 9, B), with bracteoles having a thickened spine-like midrib. (E) Partial inflorescence of Cyathula uncinulata (Wondafrash et al., 3199, B) with terminal parts of sterile and fertile flowers modified to hooks. (F) Cyathula cylindrica (Wondafrash et al., 3332, B); cyme of one fertile flower and two lateral modified sterile flowers. (G) Inflorescence of Chionothrix latifolia (Di Vincenzo et al., 236, B) with hairs growing at maturity and serving wind dispersal. (H) Inflorescence of Psilotrichum gnaphalobryum (Di Vincenzo et al., 239B, B) with solitary fertile flowers. Scales bars: (A, B, D, G, H) = 1 cm; (C, E, F) = 2 mm. Fig. 1. View largeDownload slide Flower morphology of achyranthoids. (A, B) Inflorescences of Pupalia lappacea (Di Vincenzo et al., 301, B) and Cyathula cylindrica (Wondafrash et al., 3332, B) exhibiting modified sterile flowers. (C) Adhesive dispersal unit of Cyathula orthacantha (Di Vincenzo et al., 21, B) with sterile flowers modified to spines. (D) Inflorescence of Achyranthes aspera (Di Vincenzo et al., 9, B), with bracteoles having a thickened spine-like midrib. (E) Partial inflorescence of Cyathula uncinulata (Wondafrash et al., 3199, B) with terminal parts of sterile and fertile flowers modified to hooks. (F) Cyathula cylindrica (Wondafrash et al., 3332, B); cyme of one fertile flower and two lateral modified sterile flowers. (G) Inflorescence of Chionothrix latifolia (Di Vincenzo et al., 236, B) with hairs growing at maturity and serving wind dispersal. (H) Inflorescence of Psilotrichum gnaphalobryum (Di Vincenzo et al., 239B, B) with solitary fertile flowers. Scales bars: (A, B, D, G, H) = 1 cm; (C, E, F) = 2 mm. To estimate speciation and extinction rates in the achyranthoids in relation to the morphological traits, two types of statistical analysis were conducted via the application and comparison of two sets of state-dependent speciation and extinction (SSE) models: binary-state speciation and extinction (BiSSE) models (Maddison et al., 2007) as implemented in the R package ‘diversitree’, and hidden-state speciation and extinction (HiSSE) models (Beaulieu and O’Meara, 2016) as implemented via the R package ‘hisse’. Specifically, we conducted analyses of state-dependent speciation and extinction rates under (1) a detailed group of hierarchical BiSSE models and (2) a combined group of SSE models selected from the BiSSE and HiSSE concepts. The aim of applying model group (1) was to evaluate which of the modelled combinations of speciation, extinction and character transition rates could best explain character distribution and phylogenetic diversification of the achyranthoids. The aim of model group (2) was to evaluate whether the inclusion of an additional, unaccounted (i.e. hidden) character state may better explain the fit of the speciation, extinction and character transition rates than the BiSSE models alone. For model group (1), we estimated trait-dependent rates of speciation (λ), extinction (μ) and character transition (q) under 20 different BiSSE models and compared the fit of each model, given our inferred phylogenetic trees. Each of these models had six different parameters (i.e. the rates of speciation, extinction and character transition under character state 0 and character state 1) and differed in the number of constraints applied to the parameters. Two sets of models were hereby defined: models in which one or more parameters were constrained to be equal (e.g. λ0 = λ1; hereafter ‘BiSSE model set 1’), and models in which one or more parameters were constrained to be zero (e.g. λ0 = 0; hereafter ‘BiSSE model set 2’). The trait-dependent rates were estimated for both states of modified sterile flowers and adhesive appendages. In addition, we calculated the net effect for each model (de Vos et al., 2014), which represents the difference between speciation and extinction rate under character state 1 (i.e. the net diversification under state 1) minus the difference between speciation and extinction rate under character state 0 (i.e. the net diversification under state 0). Relative model fit was evaluated within each BiSSE model set but not among the two sets. All parameter estimations and comparisons of model fit also included a model with unconstrained parameters (i.e. λ, μ and q were allowed to vary; hereafter ‘unconstrained model’), which was used as a null hypothesis. A summary of all BiSSE models employed for the analysis of the character ‘modified sterile flowers’ is provided in Table 3 and a summary of all BiSSE models employed for the more inclusive trait ‘adhesive appendages’ is provided in the Supplementary Data Table S4. Table 3. Model fit of different binary state-dependent speciation and extinction models regarding the presence of modified sterile flowers (0 = absent; 1 = present). In the models evaluated, parameters were set as equal (BiSSE model set 1) and to zero (BiSSE model set 2) across character states. Relative model fit was evaluated via the AIC and AICc; a comparison of model fit between the unconstrained and each constrained model was performed via the LRT. Parameter estimates and test statistics were calculated on the MCC tree that summarizes 1000 post-burn-in MCMC trees of the posterior tree distribution inferred during the estimation of divergence times Model specification Model fit Speciation (λ) Extinction (μ) Character transition rate (q) d.f. lnLik AIC AICc LRT χ2 LRT P-value Unconstrained – all rates free 6 −251.656 515.311 516.494 NA NA BiSSE model set 1 free μ0 = μ1 Free 5 −252.745 515.491 516.324 2.179 0.140 free free q01 = q10 5 −251.877 513.753 514.587 0.442 0.506 λ0 = λ1 Free Free 5 −253.045 516.091 516.924 2.779 0.096 free μ0 = μ1 q01 = q10 4 −253.325 514.651 515.199 3.340 0.188 λ0 = λ1 free q01 = q10 4 −253.373 514.745 515.293 3.434 0.180 λ0 = λ1 μ0 = μ1 free 4 −253.340 514.679 515.227 3.368 0.186 λ0 = λ1 μ0 = μ1 q01 = q10 3 −253.380 512.760 513.084 (best) 3.448 0.328 BiSSE Model set 2 λ0 = 0 Free Free 5 −276.098 562.196 563.029 48.884 <0.01* free μ0 = 0 Free 5 −251.847 513.693 514.527 0.382 0.536 free μ1 = 0 Free 5 −254.647 519.295 520.128 5.983 0.014* free Free q01 = 0 5 −252.394 514.789 515.622 1.477 0.224 free Free q10 = 0 5 −252.754 515.508 516.341 2.196 0.138 λ0 = 0 μ0 = 0 Free 4 −273.762 555.525 556.073 44.214 <0.01* λ0 = 0 μ1 = 0 Free 4 −276.098 560.196 560.743 48.884 <0.01* λ0 = 0 Free q01 = 0 4 −286.815 581.629 582.177 70.318 <0.01* Free μ0 = 0 q01 = 0 4 −253.201 514.402 514.950 3.091 0.213 Free μ0 = 0 q10 = 0 4 −252.755 513.509 514.057 (best) 2.198 0.333 Free μ1 = 0 q01 = 0 4 −254.647 517.295 517.843 5.983 0.050 Free μ1 = 0 q10 = 0 4 −257.465 522.930 523.478 11.619 <0.01* Model specification Model fit Speciation (λ) Extinction (μ) Character transition rate (q) d.f. lnLik AIC AICc LRT χ2 LRT P-value Unconstrained – all rates free 6 −251.656 515.311 516.494 NA NA BiSSE model set 1 free μ0 = μ1 Free 5 −252.745 515.491 516.324 2.179 0.140 free free q01 = q10 5 −251.877 513.753 514.587 0.442 0.506 λ0 = λ1 Free Free 5 −253.045 516.091 516.924 2.779 0.096 free μ0 = μ1 q01 = q10 4 −253.325 514.651 515.199 3.340 0.188 λ0 = λ1 free q01 = q10 4 −253.373 514.745 515.293 3.434 0.180 λ0 = λ1 μ0 = μ1 free 4 −253.340 514.679 515.227 3.368 0.186 λ0 = λ1 μ0 = μ1 q01 = q10 3 −253.380 512.760 513.084 (best) 3.448 0.328 BiSSE Model set 2 λ0 = 0 Free Free 5 −276.098 562.196 563.029 48.884 <0.01* free μ0 = 0 Free 5 −251.847 513.693 514.527 0.382 0.536 free μ1 = 0 Free 5 −254.647 519.295 520.128 5.983 0.014* free Free q01 = 0 5 −252.394 514.789 515.622 1.477 0.224 free Free q10 = 0 5 −252.754 515.508 516.341 2.196 0.138 λ0 = 0 μ0 = 0 Free 4 −273.762 555.525 556.073 44.214 <0.01* λ0 = 0 μ1 = 0 Free 4 −276.098 560.196 560.743 48.884 <0.01* λ0 = 0 Free q01 = 0 4 −286.815 581.629 582.177 70.318 <0.01* Free μ0 = 0 q01 = 0 4 −253.201 514.402 514.950 3.091 0.213 Free μ0 = 0 q10 = 0 4 −252.755 513.509 514.057 (best) 2.198 0.333 Free μ1 = 0 q01 = 0 4 −254.647 517.295 517.843 5.983 0.050 Free μ1 = 0 q10 = 0 4 −257.465 522.930 523.478 11.619 <0.01* Asterisks indicate LRT results at P < 0.05. lnLik, natural logarithm of the likelihood. View Large Table 3. Model fit of different binary state-dependent speciation and extinction models regarding the presence of modified sterile flowers (0 = absent; 1 = present). In the models evaluated, parameters were set as equal (BiSSE model set 1) and to zero (BiSSE model set 2) across character states. Relative model fit was evaluated via the AIC and AICc; a comparison of model fit between the unconstrained and each constrained model was performed via the LRT. Parameter estimates and test statistics were calculated on the MCC tree that summarizes 1000 post-burn-in MCMC trees of the posterior tree distribution inferred during the estimation of divergence times Model specification Model fit Speciation (λ) Extinction (μ) Character transition rate (q) d.f. lnLik AIC AICc LRT χ2 LRT P-value Unconstrained – all rates free 6 −251.656 515.311 516.494 NA NA BiSSE model set 1 free μ0 = μ1 Free 5 −252.745 515.491 516.324 2.179 0.140 free free q01 = q10 5 −251.877 513.753 514.587 0.442 0.506 λ0 = λ1 Free Free 5 −253.045 516.091 516.924 2.779 0.096 free μ0 = μ1 q01 = q10 4 −253.325 514.651 515.199 3.340 0.188 λ0 = λ1 free q01 = q10 4 −253.373 514.745 515.293 3.434 0.180 λ0 = λ1 μ0 = μ1 free 4 −253.340 514.679 515.227 3.368 0.186 λ0 = λ1 μ0 = μ1 q01 = q10 3 −253.380 512.760 513.084 (best) 3.448 0.328 BiSSE Model set 2 λ0 = 0 Free Free 5 −276.098 562.196 563.029 48.884 <0.01* free μ0 = 0 Free 5 −251.847 513.693 514.527 0.382 0.536 free μ1 = 0 Free 5 −254.647 519.295 520.128 5.983 0.014* free Free q01 = 0 5 −252.394 514.789 515.622 1.477 0.224 free Free q10 = 0 5 −252.754 515.508 516.341 2.196 0.138 λ0 = 0 μ0 = 0 Free 4 −273.762 555.525 556.073 44.214 <0.01* λ0 = 0 μ1 = 0 Free 4 −276.098 560.196 560.743 48.884 <0.01* λ0 = 0 Free q01 = 0 4 −286.815 581.629 582.177 70.318 <0.01* Free μ0 = 0 q01 = 0 4 −253.201 514.402 514.950 3.091 0.213 Free μ0 = 0 q10 = 0 4 −252.755 513.509 514.057 (best) 2.198 0.333 Free μ1 = 0 q01 = 0 4 −254.647 517.295 517.843 5.983 0.050 Free μ1 = 0 q10 = 0 4 −257.465 522.930 523.478 11.619 <0.01* Model specification Model fit Speciation (λ) Extinction (μ) Character transition rate (q) d.f. lnLik AIC AICc LRT χ2 LRT P-value Unconstrained – all rates free 6 −251.656 515.311 516.494 NA NA BiSSE model set 1 free μ0 = μ1 Free 5 −252.745 515.491 516.324 2.179 0.140 free free q01 = q10 5 −251.877 513.753 514.587 0.442 0.506 λ0 = λ1 Free Free 5 −253.045 516.091 516.924 2.779 0.096 free μ0 = μ1 q01 = q10 4 −253.325 514.651 515.199 3.340 0.188 λ0 = λ1 free q01 = q10 4 −253.373 514.745 515.293 3.434 0.180 λ0 = λ1 μ0 = μ1 free 4 −253.340 514.679 515.227 3.368 0.186 λ0 = λ1 μ0 = μ1 q01 = q10 3 −253.380 512.760 513.084 (best) 3.448 0.328 BiSSE Model set 2 λ0 = 0 Free Free 5 −276.098 562.196 563.029 48.884 <0.01* free μ0 = 0 Free 5 −251.847 513.693 514.527 0.382 0.536 free μ1 = 0 Free 5 −254.647 519.295 520.128 5.983 0.014* free Free q01 = 0 5 −252.394 514.789 515.622 1.477 0.224 free Free q10 = 0 5 −252.754 515.508 516.341 2.196 0.138 λ0 = 0 μ0 = 0 Free 4 −273.762 555.525 556.073 44.214 <0.01* λ0 = 0 μ1 = 0 Free 4 −276.098 560.196 560.743 48.884 <0.01* λ0 = 0 Free q01 = 0 4 −286.815 581.629 582.177 70.318 <0.01* Free μ0 = 0 q01 = 0 4 −253.201 514.402 514.950 3.091 0.213 Free μ0 = 0 q10 = 0 4 −252.755 513.509 514.057 (best) 2.198 0.333 Free μ1 = 0 q01 = 0 4 −254.647 517.295 517.843 5.983 0.050 Free μ1 = 0 q10 = 0 4 −257.465 522.930 523.478 11.619 <0.01* Asterisks indicate LRT results at P < 0.05. lnLik, natural logarithm of the likelihood. View Large For model group (2), we estimated speciation, extinction and character transition rates under selected SSE models and again compared the fit of each model given our inferred phylogenetic trees. Parameters of the HiSSE models were set up according to the recommendations of Beaulieu and O’Meara (2016) for defining character-independent models (CIDs). Models with four diversification process parameters were set up, with the most parameterized CID model containing 13 free parameters (CID2 with unconstrained net turnover, extinction fraction and character transition rates among and between the two observed and the two hidden states) and the least parameterized CID model containing six free parameters (CID2 with unconstrained net turnover and extinction fraction parameters among and between the two observed and the two hidden states, but character transitions constrained to be equal). Relative model fit was evaluated among the selected SSE models, with the most parameterized CID and the most parameterized BiSSE model acting as null hypothesis. A summary of all CID and BiSSE models of model group (2) employed for the analysis of the character ‘modified sterile flowers’ is provided in Table 4, and a summary of all CID and BiSSE models of model group (2) employed for the more inclusive trait ‘adhesive appendages’ is provided in the Supplementary Data Table S4. Table 4. Comparisons of model fit of different CID models of the HiSSE model concept and of BiSSE models regarding the presence of modified sterile flowers. Two two-state and two four-state CID models as well as four BiSSE models were evaluated. Dual transitions between character states (e.g. q0A ↔ q1B; q0B ↔ q1A) were not included in any CID model. Relative model fit was evaluated via the AIC and AICc; the difference in model fit between the most parameterized CID or BiSSE model and all others was evaluated via the LRT Model specification Model fit Comparison with CID-2 unconstrained Comparison with BiSSE unconstrained Net turnover (τ) Extinction fraction (ε) Character transition rate (q) d.f. lnLik AIC AICc LRT χ2 LRT P-value LRT χ2 LRT P-value HiSSE models CID-2 Free Free Free 13 −212.593 449.185 455.185 NA NA 0.183 1.000 Free Free Equal 6 −218.105 446.209 447.226 0.0705 1.000 0.026 1.000 CID-4 Free Free Three rates 11 −215.528 453.057 458.038 0.0201 1.000 0.082 1.000 Free Free Equal 9 −215.826 449.651 452.924 0.0244 1.000 0.074 1.000 BiSSE models Free Free Free 7 −221.510 455.019 456.467 0.1832 1.000 NA NA Equal Free Free 6 −221.545 454.108 453.091 0.1846 1.000 <0.001 1.000 Free Equal Free 6 −221.714 454.445 453.428 0.1916 1.000 <0.001 1.000 Equal Equal Free 5 −221.504 451.675 451.008 0.1829 0.999 <0.001 1.000 Model specification Model fit Comparison with CID-2 unconstrained Comparison with BiSSE unconstrained Net turnover (τ) Extinction fraction (ε) Character transition rate (q) d.f. lnLik AIC AICc LRT χ2 LRT P-value LRT χ2 LRT P-value HiSSE models CID-2 Free Free Free 13 −212.593 449.185 455.185 NA NA 0.183 1.000 Free Free Equal 6 −218.105 446.209 447.226 0.0705 1.000 0.026 1.000 CID-4 Free Free Three rates 11 −215.528 453.057 458.038 0.0201 1.000 0.082 1.000 Free Free Equal 9 −215.826 449.651 452.924 0.0244 1.000 0.074 1.000 BiSSE models Free Free Free 7 −221.510 455.019 456.467 0.1832 1.000 NA NA Equal Free Free 6 −221.545 454.108 453.091 0.1846 1.000 <0.001 1.000 Free Equal Free 6 −221.714 454.445 453.428 0.1916 1.000 <0.001 1.000 Equal Equal Free 5 −221.504 451.675 451.008 0.1829 0.999 <0.001 1.000 lnLik, natural logarithm of the likelihood; NA, not applicable. View Large Table 4. Comparisons of model fit of different CID models of the HiSSE model concept and of BiSSE models regarding the presence of modified sterile flowers. Two two-state and two four-state CID models as well as four BiSSE models were evaluated. Dual transitions between character states (e.g. q0A ↔ q1B; q0B ↔ q1A) were not included in any CID model. Relative model fit was evaluated via the AIC and AICc; the difference in model fit between the most parameterized CID or BiSSE model and all others was evaluated via the LRT Model specification Model fit Comparison with CID-2 unconstrained Comparison with BiSSE unconstrained Net turnover (τ) Extinction fraction (ε) Character transition rate (q) d.f. lnLik AIC AICc LRT χ2 LRT P-value LRT χ2 LRT P-value HiSSE models CID-2 Free Free Free 13 −212.593 449.185 455.185 NA NA 0.183 1.000 Free Free Equal 6 −218.105 446.209 447.226 0.0705 1.000 0.026 1.000 CID-4 Free Free Three rates 11 −215.528 453.057 458.038 0.0201 1.000 0.082 1.000 Free Free Equal 9 −215.826 449.651 452.924 0.0244 1.000 0.074 1.000 BiSSE models Free Free Free 7 −221.510 455.019 456.467 0.1832 1.000 NA NA Equal Free Free 6 −221.545 454.108 453.091 0.1846 1.000 <0.001 1.000 Free Equal Free 6 −221.714 454.445 453.428 0.1916 1.000 <0.001 1.000 Equal Equal Free 5 −221.504 451.675 451.008 0.1829 0.999 <0.001 1.000 Model specification Model fit Comparison with CID-2 unconstrained Comparison with BiSSE unconstrained Net turnover (τ) Extinction fraction (ε) Character transition rate (q) d.f. lnLik AIC AICc LRT χ2 LRT P-value LRT χ2 LRT P-value HiSSE models CID-2 Free Free Free 13 −212.593 449.185 455.185 NA NA 0.183 1.000 Free Free Equal 6 −218.105 446.209 447.226 0.0705 1.000 0.026 1.000 CID-4 Free Free Three rates 11 −215.528 453.057 458.038 0.0201 1.000 0.082 1.000 Free Free Equal 9 −215.826 449.651 452.924 0.0244 1.000 0.074 1.000 BiSSE models Free Free Free 7 −221.510 455.019 456.467 0.1832 1.000 NA NA Equal Free Free 6 −221.545 454.108 453.091 0.1846 1.000 <0.001 1.000 Free Equal Free 6 −221.714 454.445 453.428 0.1916 1.000 <0.001 1.000 Equal Equal Free 5 −221.504 451.675 451.008 0.1829 0.999 <0.001 1.000 lnLik, natural logarithm of the likelihood; NA, not applicable. View Large Model fit was evaluated by two different procedures. First, we applied likelihood ratio tests (LRTs) between the unconstrained model and each constrained model to determine whether any of the constrained models was significantly different from the unconstrained model; this comparison was enabled by a hierarchical setup of the parameter constraints. Second, we evaluated the relative fit of each model by ML optimization, using the AIC as well as an AIC corrected for small sample size (AICc; Hurvich and Tsai, 1989) to identify best-fitting models. Parameter estimates for each model and test statistics were calculated and then averaged over 100 randomly selected trees of the posterior tree distribution generated during the estimation of divergence times in order to accommodate the phylogenetic uncertainty of the tree distribution. In addition to the comparison of model fit via ML optimization, we also applied MCMC sampling to estimate marginal posterior probability densities for the diversification rate parameters of the unconstrained model. We thus conducted MCMC sampling for 10 000 generations, with 95 % HPD intervals calculated for each parameter. To allow the application of a global sampling fraction for SSE models to correct for incomplete taxon sampling (Table 1), our analyses were conducted exclusively on the achyranthoid clade of our ultrametric trees inferred from dataset B. This clade comprised a total of 65 taxa. We bioinformatically extracted the achyranthoid clade from each of the 100 randomly selected trees of the posterior tree distribution using the R package ‘phyloch’ 1.5–5 (Heibl, 2008). For modified sterile flowers, we assumed a sampling proportion of 0.44/0.41 (absence/presence); for adhesive appendages we assumed a ratio of 0.63/0.34 (Supplementary Data Table S5). Ancestral character state reconstruction We conducted ancestral character state reconstruction (ACSR) for modified sterile flowers under three different BiSSE models. The second trait, adhesive appendages, was not reconstructed, as their functionality is achieved by different, non-homologous organs (sterile flowers and bracteoles). The same matrix for presence and absence of sterile flowers was applied for ACSR as for model testing (Supplementary Data Table S3). We performed ACSR with the function ‘asr-bisse’ of the R package diversitree 0.9–7 (FitzJohn et al., 2009). Reconstruction was performed under the unconstrained binary-state speciation and extinction model and those BiSSE models with the smallest AICc value out of BiSSE model sets 1 and 2 (Table 3). To average out phylogenetic uncertainty, final ACSR was conducted over the post-burn-in posterior tree distribution as inferred during divergence dating. Inferred ancestral character states were visualized on the MCC tree. To evaluate the significance of the reconstruction results, we followed Schluter et al. (1997) and used likelihood ratios as measures of relative weight of evidence and the ratio 1:7.4 (i.e. 0.86) as indication for significance of a particular character state. Analyses of shifts in diversification rates In the trait-independent approach, we evaluated rate shifts across the entire family Amaranthaceae. The software BAMM 2.5.0 (Rabosky, 2014) was employed to calculate the rate heterogeneity of speciation and extinction on the MCC tree inferred during the estimation of divergence times. To apply a more confident sampling fraction in BAMM, our analyses were conducted exclusively on the clade of Amaranthaceae. Sampling fractions for the main clades of the Amaranthaceae are listed in Supplementary Data Table S5. Analyses were conducted in four independent runs with 10 000 000 MCMC generations each, and results were sampled every 1000 generations. Post-run analyses and visualizations were performed using the R package BAMMtools 2.1.5 (Rabosky et al., 2014) and run convergence was tested with the R package CODA 0.19–1 (Plummer et al., 2006). Upon removal of the first 10 % of all sampled MCMC generations as burn-in, we inferred the posterior probabilities of rate-shift configurations under different numbers of rate shifts and calculated Bayes factors (BFs) to compare all rate-shift configurations with prior and posterior probabilities greater than zero with a model without shifts. We considered BF > 20 to be substantial evidence for one model over another, and BF > 50 was considered very strong evidence in support of the model under analysis. The 95 % credible set of distinct rate-shift configurations and the single best-shift configuration with the highest posterior probabilities were generated as final results. RESULTS Molecular datasets The aligned matrix of dataset A comprised 4747 bp (trnK/matK, 2474 bp; rpl16, 1069 bp; trnL-F, 1204 bp), of which 373 bp were excluded as mutational hotspots (trnK/matK, five hotspots with a total of 67 bp; rpl16, eight hotspots with a total of 158 bp; trnL-F, ten hotspots with a total of 148 bp). Simple indel coding of dataset A resulted in an additional matrix of 573 characters. Six inversions (in total 39 bp) were found in trnK/matK, two inversions (in total 58 bp) in rpl16 and three inversions (in total 34 bp) in trnL-F of dataset A. Dataset B included a total of 2474 aligned positions of the trnK/matK region, and five mutational hotspots with a total of 145 bp were excluded from this dataset. Eight inversions with a total of 61 bp were found in dataset B. Dataset C included a total of 706 aligned positions for nrITS, from which four mutational hotspots with a total of 83 bp were excluded. Simple indel coding of dataset C resulted in an additional matrix of 122 characters. Phylogenetic analyses of dataset A Reconstructions under BI, ML and MP yielded identical tree topologies, only differing in node support (Fig. 2). Monophyly of the achyranthoid clade was confirmed with high support. Five of the 26 genera were identified as polyphyletic: Psilotrichum was inferred as four separate lineages (in addition to P. ferrugineum outside the clade), Cyathula as six separate lineages, Sericocomopsis as three separate lineages, Achyranthes as three separate lineages, and Calicorema as two separate lineages. The genera Centrostachys, Chionothrix, Dasysphaera, Lopriorea, Mechowia, Nelsia, Polyrhabda and Volkensinia have been included in a molecular phylogenetic analysis for the first time and were identified as part of the achyranthoid clade. The two endemic Madagascan genera Henonia and Lagrezia were resolved within the celosioid clade, and Digera as well as Pleuropterantha within the amaranthoid clade. Fig. 2. View largeDownload slide Phylogeny of the Amaranthaceae with a focus on the achyranthoid clade inferred by Bayesian inference on dataset A. Posterior probabilities of the Bayesian inference are given above branches, jackknife percentages of the maximum parsimony (left) and bootstrap percentages of the maximum likelihood analyses (right) are given below branches. Fig. 2. View largeDownload slide Phylogeny of the Amaranthaceae with a focus on the achyranthoid clade inferred by Bayesian inference on dataset A. Posterior probabilities of the Bayesian inference are given above branches, jackknife percentages of the maximum parsimony (left) and bootstrap percentages of the maximum likelihood analyses (right) are given below branches. Furthermore, we found two strongly supported major subclades that diverged after a grade of three early diverging achyranthoid lineages (Fig. 2). These three lineages are: (1) Arthraerua plus Calicorema capitata; (2) Cyathula lanceolata plus the genus Volkensinia; and (3) the genus Chionothrix. Subclade I comprises the genera Polyrhabda, Pupalia, Dasysphaera, Sericocoma, Sericorema, Marcelliopsis, Kyphocarpa, Centema and Lopriorea, Centemopsis, the species Calicorema squarrosa and several lineages of Psilotrichum. Subclade II comprises the genera Nototrichium, Achyranthes, Achyropsis, Centrostachys, Sericocomopsis, Sericostachys, Nelsia, Pandiaka and Leucosphaera, and several lineages of Cyathula (except C. lanceolata). Phylogenetic analyses of dataset C Reconstructions under BI, ML and MP recovered the achyranthoid clade as well as most internal nodes also found in the plastid tree (Supplementary Data Fig. S1). The core of subclade II of achyranthoids was inferred with maximum support, whereas the genus Leucosphaera was inconsistently recovered as sister to Cyathula lanceolata rather than as sister to all other taxa in subclade II. The species that were recovered as subclade I under the plastid data were inferred as a grade of three independent lineages under nrITS data. Divergence dating The age of the crown group of the Amaranthaceae was estimated to be 51.25 Ma (95 % HPD 42.97–57.55 Ma) (node number 56 in Supplementary Data Fig. S7). The achyranthoid clade split from the gomphrenoid clade ~27.7 million years ago (Mya) (95 % HPD 21.42–33.86 Mya) and started to diversify 22.2 Mya (95 % HPD 15.79–27.89 Mya) (Fig. 3; node numbers 88 and 112 in Supplementary Data Fig. S7). The diversification of most subclades within the achyranthoids was inferred to have occurred within the past 10 Ma (Fig. 3). A list of all inferred ages including their 95 % HPD intervals is provided in Supplementary Data Fig. S7. Fig. 3. View largeDownload slide Chronogram of the Amaranthaceae–Chenopodiaceae alliance inferred from dataset C. Fossil (white triangles) and secondary (black triangles) calibration points as well as 95 % HPD intervals (grey bars) are visualized on the MCC tree. Epoch abbreviations: Lower, Lower Cretaceous; Upper, Upper Cretaceous; Pl., Pliocene; P., Pleistocene. Fig. 3. View largeDownload slide Chronogram of the Amaranthaceae–Chenopodiaceae alliance inferred from dataset C. Fossil (white triangles) and secondary (black triangles) calibration points as well as 95 % HPD intervals (grey bars) are visualized on the MCC tree. Epoch abbreviations: Lower, Lower Cretaceous; Upper, Upper Cretaceous; Pl., Pliocene; P., Pleistocene. Analyses of state-dependent speciation and extinction In our analyses of different state-dependent SSE models using model group (1), none of the models of BiSSE model set 1 was found to be significantly different from the unconstrained model regarding the character ‘modified sterile flowers’ (Table 3). Among the models of BiSSE model set 2, the LRTs indicated six models to be significantly different from the unconstrained model, but the fit of each of these models was worse than that of the unconstrained model. Comparisons of relative model fit via the AICc indicated that the model with all parameters constrained to be equal between state 0 and state 1 (λ0 = λ1, μ0 = μ1, q01 = q10) had the best fit among models of BiSSE model set 1. By contrast, the model that constrained the extinction rate under state 0 as well as the character transition from state 1 to state 0 to zero (μ0 = 0, q10 = 0) was found to have the lowest AICc value within BiSSE model set 2. Since no model was significantly better compared with the null model and AICc values were similar across most models, we calculated the parameter values for speciation, extinction and character state transitions as well as the net effect under the unconstrained model (Table 5). For all other models, parameter values are provided in Supplementary Data Table S8. In the unconstrained model, both speciation and extinction rates for taxa with modified sterile flowers were found to be higher than for taxa lacking sterile flowers. However, the net effect under the unconstrained model was slightly negative, pointing to lower net diversification rates for taxa with than without modified sterile flowers. Rates for gains of modified sterile flowers were higher than for their loss (Table 5, Fig. 4). The accumulation of modified sterile flowers within achyranthoid taxa therefore results from the higher transition rates in favour of modified sterile flowers. Table 5. Median rates of speciation (λ), extinction (μ) and character transition (q) and the net effect of diversification for modified sterile flowers (0 = absent; 1 = present) under the unconstrained model. Median rate values were calculated across 100 post-burn-in MCMC trees of the posterior tree distribution as inferred during the estimation of divergence times. Each median value is followed by the interquartile range in parentheses. A negative net effect indicates a lower diversification rate for taxa with modified sterile flowers Model specification Parameterization Net effect Speciation (λ) Extinction (μ) Character transition rate (q) λ0 λ1 μ0 μ1 q01 q10 (λ1 − μ1) − (λ0 − μ0) Unconstrained – all rates free 0.28 (0.26, 0.32) 0.94 (0.76, 1.14) 0.05 (0, 0.13) 0.87 (0.64, 1.09) 0.08 (0.02, 0.11) 0.03 (0, 0.05) −0.12 (−0.28, −0.06) Model specification Parameterization Net effect Speciation (λ) Extinction (μ) Character transition rate (q) λ0 λ1 μ0 μ1 q01 q10 (λ1 − μ1) − (λ0 − μ0) Unconstrained – all rates free 0.28 (0.26, 0.32) 0.94 (0.76, 1.14) 0.05 (0, 0.13) 0.87 (0.64, 1.09) 0.08 (0.02, 0.11) 0.03 (0, 0.05) −0.12 (−0.28, −0.06) View Large Table 5. Median rates of speciation (λ), extinction (μ) and character transition (q) and the net effect of diversification for modified sterile flowers (0 = absent; 1 = present) under the unconstrained model. Median rate values were calculated across 100 post-burn-in MCMC trees of the posterior tree distribution as inferred during the estimation of divergence times. Each median value is followed by the interquartile range in parentheses. A negative net effect indicates a lower diversification rate for taxa with modified sterile flowers Model specification Parameterization Net effect Speciation (λ) Extinction (μ) Character transition rate (q) λ0 λ1 μ0 μ1 q01 q10 (λ1 − μ1) − (λ0 − μ0) Unconstrained – all rates free 0.28 (0.26, 0.32) 0.94 (0.76, 1.14) 0.05 (0, 0.13) 0.87 (0.64, 1.09) 0.08 (0.02, 0.11) 0.03 (0, 0.05) −0.12 (−0.28, −0.06) Model specification Parameterization Net effect Speciation (λ) Extinction (μ) Character transition rate (q) λ0 λ1 μ0 μ1 q01 q10 (λ1 − μ1) − (λ0 − μ0) Unconstrained – all rates free 0.28 (0.26, 0.32) 0.94 (0.76, 1.14) 0.05 (0, 0.13) 0.87 (0.64, 1.09) 0.08 (0.02, 0.11) 0.03 (0, 0.05) −0.12 (−0.28, −0.06) View Large Fig. 4. View largeDownload slide Posterior probability distributions of speciation (λ), extinction (μ) and character transition rates (q) under the unconstrained BiSSE model regarding the presence (1) and absence (0) of modified sterile flowers (A–C) and adhesive flower appendages (D–F). Vertical lines represent mean values; horizontal bars and shaded areas represent the 95 % credibility intervals of posterior probability distributions. Fig. 4. View largeDownload slide Posterior probability distributions of speciation (λ), extinction (μ) and character transition rates (q) under the unconstrained BiSSE model regarding the presence (1) and absence (0) of modified sterile flowers (A–C) and adhesive flower appendages (D–F). Vertical lines represent mean values; horizontal bars and shaded areas represent the 95 % credibility intervals of posterior probability distributions. Concerning adhesive flower appendages as a more inclusive functional trait, our LRT evaluations indicated that none of the models of BiSSE model set 1 was found to be significantly different from the unconstrained model (Supplementary Data Table S4). Comparisons of relative model fit via the AICc indicated that the model that constrained extinction and character transition rates to be equal between state 0 and state 1 (μ0 = μ1, q01 = q10) had the best fit among models of BiSSE model set 1. Among the models of BiSSE model set 2, the LRTs indicated a total of seven models to be significantly different from the unconstrained model. However, none of these seven models had a better fit to our data than the null model according to the AICc values. Comparisons further indicated that the model that constrained the extinction rate under state 1 as well as the character transition from state 0 to state 1 (μ1 = 0, q01 = 0) was found to have the best relative model fit. Similar to the results for modified sterile flowers, our analyses regarding the presence of adhesive appendages indicated that no model was significantly better than the null model, and AICc values were quite similar across most BiSSE models. We calculated the parameter values for speciation, extinction and character transition as well as the net effect under all models (Supplementary Data Table S8). In the unconstrained model, we found higher speciation and extinction rates for taxa with adhesive appendages than for taxa without them. Here, the net effect was found to be positive, indicating that diversification rates for taxa with adhesive appendages were higher than for taxa without them. Character transition rates were found to be equal for both gains and losses (Supplementary Data Table S8, Fig. 4). In our analyses of different state-dependent SSE models using model group (2), none of the evaluated models was found to be significantly different from the unconstrained model (Table 4, Supplementary Data Table S4). This finding was true for both morphological traits evaluated. For both traits, neither the BiSSE models nor the constrained CID models of the HiSSE concept displayed a significantly better model fit to the data than the null CID model when evaluated using an LRT. Comparisons of relative model fit via the AICc indicated that the least parameterized CID model had the best relative fit among models of model group (2). Again, this result was observed for both morphological traits evaluated. In summary, the results of our analyses of different state-dependent SSE models using model group (2) indicated that the inclusion of an additional, hidden character state did not explain the fit of the speciation, extinction and character transition rates significantly better than the BiSSE models alone. Ancestral character state reconstruction Since model testing did not infer a single best BiSSE model, we reconstructed ancestral character states for modified sterile flowers under three different BiSSE models: the unconstrained model and the two best-fitting models from BiSSE model sets 1 and 2 as inferred through the AICc (Fig. 5). The best-fitting BiSSE models yielded very similar reconstructions for the majority of the early divergences in the achyranthoid clade but were distinctly different from the reconstructions under the unconstrained model. The best-fitting BiSSE models inferred the absence of modified sterile flowers in the most recent common ancestor (MRCA) of the achyranthoids as more likely than the presence (BiSSE model set 1, 16 % presence, 84 % absence; BiSSE model set 2, 0 % presence, 100 % absence) and, indicated multiple origins of this trait within the clade: 14 independent origins for modified sterile flowers are inferred within the achyranthoids under the best-fit model of BiSSE model set 1, and 15 independent origins under the best-fit model of BiSSE model set 2 (according to the significance level of 86 % for a particular character state). The unconstrained model, by contrast, inferred the presence of modified sterile flowers in the MRCA of the achyranthoids as more likely than the absence (59 % presence, 41 % absence). Similarly, the reconstruction results for the MRCA of subclades I and II of the achyranthoids under the best-fit BiSSE models indicated the absence of modified sterile flowers as more likely than their presence (subclade I, BiSSE model set 1, 6 % presence, 94 % absence; BiSSE model set 2, 0 % presence, 100 % absence; subclade II, BiSSE model set 1, 14 % presence, 86 % absence; BiSSE model set 2, 0 % presence, 100 % absence), whereas the reconstruction results under the unconstrained model were uncertain, indicating equal probability for both character states (subclade I, 51 % presence, 49 % absence; subclade II, 54 % presence, 46 % absence). Fig. 5. View largeDownload slide Reconstruction of ancestral character states for modified sterile flowers under three different BiSSE models (bars from left to right): the unconstrained model; the best fit model of BiSSE model set 1 with parameters set as equal across the character states; and the best fit model of BiSSE model set 2 with parameters set to zero across states. The asterisk marks a species that bears sterile flowers that are not modified to adhesive appendages, but instead are modified to long hairs that serve wind dispersal. Percentages in the magnified nodes represent character state probabilities. Fig. 5. View largeDownload slide Reconstruction of ancestral character states for modified sterile flowers under three different BiSSE models (bars from left to right): the unconstrained model; the best fit model of BiSSE model set 1 with parameters set as equal across the character states; and the best fit model of BiSSE model set 2 with parameters set to zero across states. The asterisk marks a species that bears sterile flowers that are not modified to adhesive appendages, but instead are modified to long hairs that serve wind dispersal. Percentages in the magnified nodes represent character state probabilities. Analyses of shifts in diversification rates Convergence of BAMM analyses was assessed in four independent runs via effective sample size (ESS) values (ESS log-likelihood, 481.59–699.76; ESS number of shifts, 824.84–1151.58). Our analyses via BAMM indicated that the zero-shift configuration (null model) had the highest prior and posterior probability (posterior probability = 0.31; Fig. 6). Model comparison via BFs did not show evidence in favour of a model with one or more shifts relative to the null model, as BFs for all models with one or more shifts were <20, ranging from 0.37 to 2.73. Computing the 95 % credible set of macroevolutionary rate shifts yielded a set of 11 distinct shift configurations, out of which three accounted for 72 % of the posterior probability (Supplementary Data Fig. S9). The first configuration accounted for 40 % of the probability of the data, the second for 26 % and the third for ~6 %. The null model with zero shifts was the one with the highest posterior probability (0.40). None of the other configurations had sampling frequencies >0.06 and are therefore not shown. A phylorate plot that displays the most frequently sampled configuration out of the credible shift set is given in Supplementary Data Fig. S9. DISCUSSION Monophyly and overall phylogenetic structure of the achyranthoid clade Under our increased sampling of taxa and molecular characters, the achyranthoid clade originally evidenced by Müller and Borsch (2005a, b) has been found strongly supported. In addition, the genera Centrostachys, Chionothrix, Dasysphaera, Lopriorea, Mechowia, Nelsia, Polyrhabda, Sericocomopsis and Volkensinia of the subtribe Aervinae sensuTownsend (1993), which had not been included in any phylogenetic study so far, have been identified as part of the achyranthoid clade. Compared with the sole analyses of trnK/matK (Müller and Borsch 2005a, b), the internal relationships of achyranthoids are much better resolved in this study, except for the first diverging lineages. While Cyathula lanceolata and Volkensinia appear in a medium-supported subclade, at least in the plastid tree (Fig. 2), their relationships to Chionothrix and the Calicorema capitata plus Arthraerua subclade remain unclear. However, the relationship of C. capitata with Arthraerua leubnitziae is confirmed with maximum support in plastid and nuclear trees (Fig. 2, Supplementary Data Fig. S6). The affinities of both South African taxa are further supported by similar pollen morphology (Müller and Borsch, 2005b). Our analysis also confirms the position of the achyranthoid clade as sister to the Gomphrenoideae, both of which are the sister to an aervoid clade (Müller and Borsch 2005a, b). Four other genera of the paraphyletic Amaranthoideae sensuTownsend (1993) that were newly included in our phylogenetic analysis appear outside of the achyranthoids: Digera and Pleuropterantha are close relatives of Amaranthus (extending the amaranthoid clade of Müller and Borsch 2005a). Henonia and Lagrezia belong to the celosioids (Fig. 2), indicating that tribe Celosieae in the circumscription of Schinz (1934) is not monophyletic unless Lagrezia [classified to Amaranthineae (Schinz, 1934; Townsend, 1993)] is included. The two lineages of achyranthoids initially found by Müller and Borsch (2005b) appear as two major subclades with our more representative taxon sampling (subclades I and II; Fig. 2). Plastid and nrITS trees are largely congruent. Topological differences only occur among samples of Psilotrichum in subclade I and concerning the placement of Cyathula orthacantha. Species of Psilotrichum are found in several different lineages. The Asian P. ferrugineum is confirmed in a position outside the achyranthoid clade (Müller and Borsch, 2005b) and P. scleranthum (= P. africanum) is now resolved as isolated sister to subclades I plus II in the combined plastid dataset (Fig. 2). Psilotrichum gnaphalobryum, P. gracilipes, P. gloveri and P. sericeum form a well-supported subclade in all trees (Fig. 2, Supplementary Data Fig. S6). Psilotrichum schimperi is also included in this subclade based on plastid data (Fig. 2), but its position is unresolved under nrITS (Supplementary Data Fig. S6). Another lineage with the remaining species of Psilotrichum (P. elliotii, P. cyathuloides) appears close to a Dasysphaera–Pupalia subclade. Interestingly, the Somalian monotypic genus Polyrhabda is suggested as a close relative of P. elliotii. The Dasysphaera–Pupalia subclade is highly supported in all analyses and is one of the youngest groups that have evolved modified sterile flowers with hooks forming adhesive diaspores (Figs 1 and 5). Calicorema squarrosa is now well supported as sister to Sericocoma (Fig. 2, Supplementary Data Fig. S6), and not to Sericorema as depicted in the Bayesian trnK/matK tree (posterior probability = 0.81; Müller and Borsch 2005b). In subclade II of the achyranthoids, the genus Achyranthes appears paraphyletic to Achyropsis and Nototrichium (Fig. 2, Supplementary Data Fig. S6). All species of Nototrichium are small trees, indicating insular gigantism in these Hawaiian endemics. Centrostachys aquatica is sister to all other species in this subclade. It morphologically resembles the herbaceous species of Achyranthes but has bigger flowers and occurs in swampy habitats. All four genera are characterized by spiciform inflorescences with solitary fertile flowers. In Centrostachys and Achyranthes (Fig. 1D) the flowers with their bracts and bracteoles are deflexed at maturity and bracteoles possess strongly thickened, spine-like midribs that serve for adhesion in diaspores. This feature was obviously lost in Achyropsis and Nototrichium. Furthermore, elaborate stamen tube appendages characteristic of Achyranthes and relatives are absent in Nototrichium. Different samples of Achyranthes aspera are found in at least three lineages, paralleling a considerable morphological variation (e.g. plant size from 10 to 150 cm, presence or absence of woodiness, dense or very sparse indumentum). The Asian Achyranthes bidentata is depicted close to Ethiopian and Kenyan plants of Achyranthes aspera, and the Hawaiian endemic Achyranthes splendens is found distantly from Nototrichium. Further taxon and character sampling is necessary to understand the complex evolutionary history in Achyranthes. The species of Cyathula appear scattered across subclade II (Fig. 2, Supplementary Data Fig. S6), except for C. lanceolata, which belongs to the early diverging achyranthoids. The deviating position of C. lanceolata in Müller and Borsch (2005b) is an artefact because their sequence (AC022) was mislabelled and in fact is C. prostrata. Cyathula orthacantha is isolated and diverges after Leucosphaera within subclade II (Fig. 2). Cyathula uncinulata (Fig. 1E), C. polycephala, C. tomentosa and C. capitata constitute a subclade of taxa with very dense and partly subglobose synflorescences that is sister to C. cylindrica (Fig. 1B) plus three samples of the polyphyletic Sericocomopsis hildebrandtii (Fig. 2, Supplementary Data Fig. S6). Cyathula natalensis, C. prostrata and Pandiaka form a lineage with elongate, lax inflorescences. Since the genus concept of Cyathula is based on the presence of sterile flowers (Townsend, 1993; Hérnandez-Ledesma et al., 2015), now shown as homoplastic, it is not surprising that the genus does not appear as a natural group. Sericocomopsis currently has two accepted species (Townsend, 1993), distinguished by simple versus stellate hairs. They are resolved in three different lineages (Fig. 2, Supplementary Data Fig. S6), indicating that the simple-haired individuals currently circumscribed as S. hildebrandtii constitute more than one species. Diversification of the achyranthoid clade through time Compared with the gomphrenoids, which are a Neotropical radiation (Sánchez-Del Pino et al., 2009) that started 26.4 Mya (95 % HPD 20.68–32.85 Mya) and resulted in the largest clade within the Amaranthaceae, we estimate the crown group age of the achyranthoids as 22.2 Ma (95 % HPD 15.79–27.89 Ma) (node numbers 89 and 112 in Supplementary Data Fig. S7). Most achyranthoids occur in open habitats, mainly semi-desert scrubland, Acacia–Commiphora wood- and bushland as well as dry evergreen afromontane forest and grassland complexes. Characteristic plant groups of these habitats are Acacia (Fabaceae, subfamily Mimosoideae), Commiphora (Burseraceae) and different C4 grasses, such as Sporobolus (subfamily Chloridoideae), Hyparrhenia, Heteropogon and Panicum (subfamily Panicoideae). As shown by Bouchenak-Khelladi et al. (2014), these C4 grass lineages originated around 35 Mya. The crown age of the genus Commiphora was estimated to be ~30 Ma by Becerra et al. (2012). The earliest split of an African Acacia clade with species occurring in open environments from its American sister group is dated to early Miocene times by Bouchenak-Khelladi et al. (2010) but was estimated to even have occurred from around 33 Mya onward by Miller et al. (2013). Since the above-mentioned genera are dominant in the vegetation in which achyranthoids occur, we may assume that those habitats evolved from ~30 Mya onward. These findings are in line with our age estimation for the achyranthoids. The inferred origins of all these plants, being indicators of open arid and semi-arid habitats, furthermore coincide with the assumed stepwise aridification of East Africa since Eocene–Oligocene times in conjunction with the uplift of the East African Rift system (e.g. Sepulchre et al., 2006). As shown in our chronogram (Fig. 3) the two major subclades of achyranthoids started to diversify in the early Miocene 16.6 Mya (95 % HPD 11.94–21.22 Mya) and 18.8 Mya (95 % HPD 14.05–23.87 Ma), respectively (node numbers 155 and 121 in Supplementary Data Fig. S7). Achyranthoid species are mostly herbs and small shrubs which are not competitive in broad-leaved forests. Thus, the opening of forests is likely to have offered new niches for achyranthoid species to diversify. Many achyranthoid taxa, such as the Dasysphaera–Pupalia lineage or the Cyathula species with dense inflorescences (node number 152 in Supplementary Data Fig. S7), evolved in Pliocene–Pleistocene times, shortly after C4 grasslands became ecologically dominant in large parts of East Africa (8–6 Mya; Bouchenak-Khelladi et al., 2014). Evolution of modified sterile flowers At the selected significance level, the reconstruction of ancestral character states suggests that modified sterile flowers originated at least 14 times independently in the achyranthoids (Fig. 5). Reconstruction results for deeper nodes under the unconstrained model are uncertain, indicating almost equal probabilities for both character states. The transition rates under the unconstrained model were found to favour gains of modified sterile flowers during the evolution of achyranthoids (Table 5). This is in line with our ACSR (Fig. 5), which suggests more gains than losses of this character over time. For the more inclusive functional trait of adhesive appendages, by contrast, transition rates are equal in both directions (Supplementary Data Table S8). This also suggests that a reversal of the bracteole midrib morphology [from thickened and thorn-like, as in Achyranthes (Fig. 1D), to the unmodified state] is more probable than the loss of sterile flowers, which would require a change in inflorescence architecture. The first occurrence of modified sterile flowers above the significance level was inferred to the early Pliocene under the constrained model of BiSSE model set 1. However, it seems more parsimonious regarding the number of independent origins that the first achyranthoid lineages with modified sterile flowers originated in the second half of the Miocene (once in the stem node incorporating Nelsia quadrangula and several species of Cyathula, and a second time in the ancestor of Kyphocarpa, Marcelliopsis and Sericorema). Several independent origins of this trait occurred during the Pleistocene in subclades I and II. Our reconstructions also indicated that sterile flowers in several early-diverging taxa (e.g. Volkensinia, Leucosphaera, Cyathula lanceolata and Cyathula orthacantha) could have evolved on terminal branches, and therefore dating the origin of the trait is not possible. We assume that the asymmetries in transition rates contributed highly to the prevalence of modified sterile flowers in many achyranthoid species. Species diversification and epizoochory in achyranthoid evolution To illustrate the effect of modified sterile flowers (and the more inclusive functional trait of adhesive appendages) on achyranthoid diversification, we evaluated the rates of speciation, extinction and net diversification by comparing lineages with and without these traits under several BiSSE models (Table 5, Supplementary Data Table S8). Speciation as well as extinction rates were found to be higher in the presence than in the absence of both traits (Fig. 4, Table 5, Supplementary Data Table S8). Thus, net diversification was only slightly different in the presence than in the absence of these traits. The reliability of analyses with SSE models on datasets with <300 terminals has been evaluated in several investigations (e.g. Davis et al., 2013; Rabosky and Goldberg, 2015). Even though the sample size in this investigation was <300, it is well in line with recently recommended sample sizes for SSE models (Gamisch, 2016). Beaulieu and O’Meara (2016) developed proper null models (CID-2 and CID-4), which we used to validate our data and which confirmed the lack of diversification signal in the observed traits. Additionally, we employed BAMM to test for trait-independent rate heterogeneity among the Amaranthaceae, specifically within certain subclades of the achyranthoids or the clade as a whole. The idea was to look for diversification rate shifts that, if present, could be correlated to the dated origins of modified sterile flowers. However, our results indicate the absence of diversification rate shifts within the diversification of the achyranthoids (Fig. 6, Supplementary Data Fig. S9). Fig. 6. View largeDownload slide Prior and posterior probability distributions of the number of rate shifts as inferred via BAMM on the MCC tree inferred during the estimation of divergence times. Fig. 6. View largeDownload slide Prior and posterior probability distributions of the number of rate shifts as inferred via BAMM on the MCC tree inferred during the estimation of divergence times. Beaulieu and Donoghue (2013) argued that ‘the origin of a trait may not, in itself, be sufficient to increase diversification rate, but rather, requires the right combination of traits’. This view is shared by different authors (e.g. de Queiroz, 2002; Maddison et al., 2007; Beaulieu and O’Meara, 2016). In achyranthoids there is no significant boost of diversification after the acquisition of traits that can serve epizoochory. Nevertheless, the high number of independent origins of sterile flowers, connected with biased transition rates, suggests an adaptive value. The evolution of open habitats in East Africa has probably been an ecological opportunity for the achyranthoids. Nearly all members of this clade occur in habitats with large animals (with a few exceptions, such as Sericostachys, which evolved a liana habit secondarily). Our inferred divergence times indicate that the evolution of achyranthoid diversity coincides with times in which habitats changed and appropriate dispersal agents became abundant (Fig. 3). Large terrestrial mammals started to diversify in the Oligocene. Fossil evidence pointed to a further increase in diversity of large herbivores (including bovids, which represent the largest group of herbivores in Africa today) during the Late Miocene (8–6 Mya) due to a well-established Eurasian connection that allowed massive migration to Africa. A peak of diversity was reached in the Pliocene due to the global transition from C3 to C4 vegetation (Bobe, 2006; Charles-Dominique et al., 2016). Several investigations developed models to measure and predict the adhesive ability of different diaspore types. Factors such as retention times and separation forces strongly influence the persistence of diaspores in fur and, by extension, dispersal distances (Gorb and Gorb, 2002; Will et al., 2007; Couvreur et al., 2008; Will and Tackenberg 2008; Bullock et al., 2011). Animals migrate between similar habitats, leading to directed dispersal of plants to habitats with optimal conditions for their offspring (Stebbins, 1971). Since most animals use the same route during migration and have been shown to be mobile links between fragmented and therefore isolated ecosystems (Couvreur et al., 2004), there is only a limited possibility for populations to become isolated due to an interruption in gene flow. In achyranthoids, epizoochory may therefore be a successful dispersal mechanism, but rapid diversification may be counteracted by regular interchange of gene pools. More work is needed to evaluate this possibility, particularly by analysing the phylogeography and population structure of widespread achyranthoid species. SUPPLEMENTARY DATA Supplementary data are available online at https://academic.oup.com/aob and consist of the following. Table S1: list of analysed taxa of (a) Amaranthaceae and (b) Caryophyllales. Table S2: (a) PCR protocol used for amplification of the plastid trnK/matK, rpl16, trnL-F and the nrITS genomic regions, and (b) primers used for amplification and sequencing of the plastid trnK/matK, rpl16, trnL-F and the nrITS genomic regions. Table S3: character matrix coding the presence and absence of traits within the achyranthoids. Table S4: (a) model fit of different binary state-dependent speciation and extinction models regarding the presence of adhesive flower appendages, and (b) comparisons of model fit of different CID models of the HiSSE model concept and of BiSSE models regarding the presence of adhesive flower appendages. Table S5: (a) global sampling fractions specified for BiSSE analyses, and (b) clade-specific sampling fractions specified for BAMM analyses and species numbers assumed for Amaranthaceae. Figure S6: phylogeny of the achyranthoid clade based on nrITS sequences inferred via BI. Figure S7: chronogram and summary of age estimates of the Amaranthaceae–Chenopodiaceae alliance inferred from dataset C via divergence dating. Table S8: median rates of speciation, extinction and character transition as well as the net effect of diversification for modified sterile flowers under all models. Figure S9: the shift configurations with the highest posterior probabilities out of the 95 % credible set of distinct shift configurations. ACKNOWLEDGEMENTS This study was carried out in partial fulfilment of a doctoral thesis by V.D.V. We thank the Ethiopian Biodiversity Institute, Itambo Malombe (East African Herbarium) and the Kenyan authorities for support and collection permits. We also acknowledge the assistance of Assefa Hailu (Addis Ababa University) and Mathias Mbale (National Museums of Kenya) during fieldwork. Additional assistance was provided by staff from the Botanical Garden und Botanical Museum Berlin, especially by Kim Govers, Bettina Giesicke, Julia Pfitzner and Doreen Weigel for laboratory work and by Susy Fuentes Bazan regarding the placement of Chenopodiaceae fossils, as well as by Michael Rodewald for graphically processing the photographic plate. This study was supported by computer resources from the HPC Service of ZEDAT of the Freie Universität Berlin. This work was supported by Deutsche Forschungsgemeinschaft (BO1815/1–4 to T.B. and S.D.) LITERATURE CITED Acosta JM , Perreta M , Amsler A , Vegetti AC . 2009 . The flowering unit in the synflorescences of Amaranthaceae . Botanical Review 75 : 365 – 376 . Google Scholar CrossRef Search ADS Agnew ADQ , Flux JEC . 1970 . Plant dispersal by hares (Lepus capensis L.) in Kenya . Ecology 51 : 735 – 737 . Google Scholar CrossRef Search ADS Akaike H . 1974 . A new look at the statistical model identification . IEEE Transactions on Automatic Control 19 : 716 – 723 . Google Scholar CrossRef Search ADS Beaulieu JM , Donoghue MJ . 2013 . Fruit evolution and diversification in campanulid angiosperms . Evolution 67 : 3132 – 3144 . Google Scholar CrossRef Search ADS PubMed Beaulieu JM , O’Meara BC . 2016 . detecting hidden diversification shifts in models of trait-dependent speciation and extinction . Systematic Biology 65 : 583 – 601 . Google Scholar CrossRef Search ADS PubMed Becerra JX , Noge K , Olivier S , Venable DL . 2012 . The monophyly of Bursera and its impact for divergence times of Burseraceae . Taxon 61 : 333 – 343 . Bell CD , Soltis DE , Soltis PS . 2010 . The age and diversification of the angiosperms re-revisited . American Journal of Botany 97 : 1296 – 1303 . Google Scholar CrossRef Search ADS PubMed Bobe R . 2006 . The evolution of arid ecosystems in eastern Africa . Journal of Arid Environments 66 : 564 – 584 . Google Scholar CrossRef Search ADS Bouchenak-Khelladi Y , Maurin O , Hurter J , van der Bank M . 2010 . The evolutionary history and biogeography of Mimosoideae (Leguminosae): an emphasis on African acacias . Molecular Phylogenetics and Evolution 57 : 495 – 508 . Google Scholar CrossRef Search ADS PubMed Bouchenak-Khelladi Y , Slingsby JA , Verboom GA , Bond WJ . 2014 . Diversification of C-4 grasses (Poaceae) does not coincide with their ecological dominance . American Journal of Botany 101 : 300 – 307 . Google Scholar CrossRef Search ADS PubMed Bullock JM , Galsworthy SJ , Manzano P , et al. 2011 . Process-based functions for seed retention on animals: a test of improved descriptions of dispersal using multiple data sets . Oikos 120 : 1201 – 1208 . Google Scholar CrossRef Search ADS Charles-Dominique T , Davies TJ , Hempson GP , et al. 2016 . Spiny plants, mammal browsers, and the origin of African savannas . Proceedings of the National Academy of Sciences of the USA 113 : E5572 – E5579 . Google Scholar CrossRef Search ADS PubMed Collinson ME , Boulter MC , Holmes PL . 1993 . Magnoliophyta (‘Angiospermae’) . In: The fossil record 2 . London : Chapman & Hall , 809 – 840 Couvreur M , Christiaen B , Verheyen K , Hermy M . 2004 . Large herbivores as mobile links between isolated nature reserves through adhesive seed dispersal . Applied Vegetation Science 7 : 229 – 236 . Google Scholar CrossRef Search ADS Couvreur M , Verheyen K , Vellend M , et al. 2008 . Epizoochory by large herbivores: merging data with models . Basic and Applied Ecology 9 : 204 – 212 . Google Scholar CrossRef Search ADS Cuénoud P , Savolainen V , Chatrou LW , Powell M , Grayer RJ , Chase MW . 2002 . Molecular phylogenetics of Caryophyllales based on nuclear 18S rDNA and plastid rbcL, atpB, and matK DNA sequences . American Journal of Botany 89 : 132 – 144 . Google Scholar CrossRef Search ADS PubMed Darriba D , Taboada GL , Doallo R , Posada D . 2012 . jModelTest 2: more models, new heuristics and parallel computing . Nature Methods 9 : 772 – 772 . Google Scholar CrossRef Search ADS PubMed Davis MP , Midford PE , Wayne Maddison W . 2013 . Exploring power and parameter estimation of the BiSSE method for analyzing species diversification . BMC Evolutionary Biology 13 : 38 . Google Scholar CrossRef Search ADS PubMed Drummond AJ , Ho SYW , Phillips MJ , Rambaut A . 2006 . Relaxed phylogenetics and dating with confidence . PLOS Biology , 4 : e88 . Google Scholar CrossRef Search ADS PubMed Drummond AJ , Suchard MA , Xie D , Rambaut A . 2012 . Bayesian phylogenetics with BEAUti and the BEAST 1.7 . Molecular Biology and Evolution 29 : 1969 – 1973 . Google Scholar CrossRef Search ADS PubMed Edgar RC . 2004 . MUSCLE: multiple sequence alignment with high accuracy and high throughput . Nucleic Acids Research 32 : 1792 – 1797 . Google Scholar CrossRef Search ADS PubMed FitzJohn RG , Maddison WP , Otto SP . 2009 . Estimating trait-dependent speciation and extinction rates from incompletely resolved phylogenies . Systematic Biology 58 : 595 – 611 . Google Scholar CrossRef Search ADS PubMed Friis I , Demissew S , van Breugel P . 2011 . Atlas of the potential vegetation of Ethiopia . Addis Ababa, Ethiopia: Addis Ababa University Press, Shama Books. Gamisch A . 2016 . Notes on the statistical power of the binary state speciation and extinction (BiSSE) model . Evolutionary Bioinformatics 12 : EBO.S39732 . Google Scholar CrossRef Search ADS Gorb E , Gorb S . 2002 . Contact separation force of the fruit burrs in four plant species adapted to dispersal by mechanical interlocking . Plant Physiology and Biochemistry 40 : 373 – 381 . Google Scholar CrossRef Search ADS Gregor HJ . 1982 . Die ‘Parvangulae’ und ‘Guttulae’ Hiltermann & Schmitz 1968 aus dem Randecker Maar – Samenreste von Centrospermae . Paläontologische Zeitschrift 56 : 11 – 18 . Google Scholar CrossRef Search ADS Heibl C . 2008 . PHYLOCH: R language tree plotting tools and interfaces to diverse phylogenetic software packages . http://www.christophheibl.de/Rpackages.html. Hernández-Ledesma P , Berendsohn WG , Borsch T , et al. 2015 . A taxonomic backbone for the global synthesis of species diversity in the angiosperm order Caryophyllales . Willdenowia 45 : 281 – 383 . Google Scholar CrossRef Search ADS Ho SYW , Phillips MJ . 2009 . Accounting for calibration uncertainty in phylogenetic estimation of evolutionary divergence times . Systematic Biology 58 : 367 – 380 Google Scholar CrossRef Search ADS PubMed Hohmann S , Kadereit JW , Kadereit G . 2006 . Understanding Mediterranean-Californian disjunctions: molecular evidence from Chenopodiaceae-Betoideae . Taxon 55 : 67 – 78 . Google Scholar CrossRef Search ADS Hurvich CM , Tsai C-L . 1989 . Regression and time series model selection in small samples . Biometrika 76 : 297 – 307 . Google Scholar CrossRef Search ADS Kadereit G , Borsch T , Weising K , Freitag H . 2003 . Phylogeny of Amaranthaceae and Chenopodiaceae and the evolution of C-4 photosynthesis . International Journal of Plant Sciences 164 : 959 – 986 . Google Scholar CrossRef Search ADS Kadereit G , Ackerly D , Pirie MD . 2012 . A broader model for C-4 photosynthesis evolution in plants inferred from the goosefoot family (Chenopodiaceae s.s.) . Proceedings of the Royal Society B Biological Sciences 279 : 3304 – 3311 . Google Scholar CrossRef Search ADS PubMed Lewis PO . 2001 . A likelihood approach to estimating phylogeny from discrete morphological character data . Systematic Biology 50 : 913 – 925 . Google Scholar CrossRef Search ADS PubMed Löhne C , Borsch T . 2005 . Molecular evolution and phylogenetic utility of the petD group II intron: a case study in basal angiosperms . Molecular Biology and Evolution 22 : 317 – 332 . Google Scholar CrossRef Search ADS PubMed Maddison WP , Midford PE , Otto SP . 2007 . Estimating a binary character’s effect on speciation and extinction . Systematic Biology 56 : 701 – 710 . Google Scholar CrossRef Search ADS PubMed Miller JT , Murphy DJ , Ho SYW , Cantrill DJ , Seigler D . 2013 . Comparative dating of Acacia: combining fossils and multiple phylogenies to infer ages of clades with poor fossil records . Australian Journal of Botany 61 : 436 – 445 . Google Scholar CrossRef Search ADS Mori SA , Brown JL . 1998 . Epizoochorous dispersal by barbs, hooks, and spines in a lowland moist forest in central French Guiana . Brittonia 50 : 165 – 173 . Google Scholar CrossRef Search ADS Müller K . 2005 . SeqState – primer design and sequence statistics for phylogenetic DNA data sets . Applied Bioinformatics 4 : 65 – 69 . Google Scholar CrossRef Search ADS PubMed Müller K , Borsch T . 2005a. Phylogenetics of Amaranthaceae based on matK/trnK sequence data – evidence from Parsimony, likelihood, and Bayesian analyses . Annals of the Missouri Botanical Garden 92 : 66 – 102 . Müller K , Borsch T . 2005b. Multiple origins of a unique pollen feature: stellate pore ornamentation in Amaranthaceae . Grana 44 : 266 – 281 . Google Scholar CrossRef Search ADS Müller J , Müller K , Neinhuis C , Quandt D . 2010 . PhyDE: Phylogenetic Data Editor v 0.9971 . www.phyde.de. Nichols DJ , Traverse A . 1971 . Palynology, petrology, and depositional environments of some early tertiary lignites in Texas . Geoscience and Man 3 : 37 – 48 . Google Scholar CrossRef Search ADS Peterson AT . 2006 . Application of molecular clocks in ornithology revisited . Journal of Avian Biology 37 : 541 – 544 . Google Scholar CrossRef Search ADS Plummer M , Best N , Cowles K , Vines K . 2006 . CODA: Convergence Diagnosis and Output Analysis for MCMC . R News 6 : 7 – 11 . Principi P . 1926 . La flora oligocenica de Chiavon e Salcedo . Memorie della Carta Geologica d’Italia 10 : 1 – 127 . de Queiroz A . 2002 . Contingent predictability in evolution: key traits and diversification . Systematic Biology 51 : 917 – 929 . Google Scholar CrossRef Search ADS PubMed Rabosky DL . 2014 . Automatic detection of key innovations, rate shifts, and diversity-dependence on phylogenetic trees . PLoS ONE 9 : e89543 . Google Scholar CrossRef Search ADS PubMed Rabosky DL , Goldberg EE . 2015 . Model inadequacy and mistaken inferences of trait-dependent speciation . Systematic Biology 64 : 340 – 355 . Google Scholar CrossRef Search ADS PubMed Rabosky DL , Grundler M , Anderson C , et al. 2014 . BAMMtools: an R package for the analysis of evolutionary dynamics on phylogenetic trees . Methods in Ecology and Evolution 5 : 701 – 707 . Google Scholar CrossRef Search ADS Rambaut A , Drummond AJ . 2007 . Tracer v1.4.http://tree.bio.ed.ac.uk/software/tracer/. Ridley HN . 1930 . The dispersal of plants throughout the world . Ashford, UK : Reeve & Co . Ronquist F , Teslenko M , van der Mark P , et al. 2012 . MrBayes 3.2: efficient Bayesian phylogenetic inference and model choice across a large model space . Systematic Biology 61 : 539 – 542 . Google Scholar CrossRef Search ADS PubMed Sánchez-Del Pino I , Borsch T , Motley TJ . 2009 . trnL-F and rpl16 sequence data and dense taxon sampling reveal monophyly of unilocular anthered Gomphrenoideae (Amaranthaceae) and an improved picture of their internal relationships . Systematic Botany 34 : 57 – 67 . Google Scholar CrossRef Search ADS Schäferhoff B , Müller KF , Borsch T . 2009 . Caryophyllales phylogenetics: disentangling Phytolaccaceae and Molluginaceae and description of Microteaceae as a new isolated family . Willdenowia 39 : 209 – 228 . Google Scholar CrossRef Search ADS Schluter D , Price T , Mooers AØ , Ludwig D . 1997 . Likelihood of ancestor states in adaptive radiation . Evolution 51 : 1699 – 1711 . Google Scholar CrossRef Search ADS PubMed Schinz H . 1934 . Amaranthaceae . In: Engler A , Prantl K , eds. Die natürlichen Pflanzenfamilien , Vol. 16c , 2nd edn . Leipzig : W. Engelmann . Sepulchre P , Ramstein G , Fluteau F , Schuster M , Tiercelin JJ , Brunet M . 2006 . Tectonic uplift and Eastern Africa aridification . Science 313 : 1419 – 1423 . Google Scholar CrossRef Search ADS PubMed Simmons MP , Ochoterena H . 2000 . Gaps as characters in sequence-based phylogenetic analyses . Systematic Biology 49 : 369 – 381 . Google Scholar CrossRef Search ADS PubMed Sorensen AE . 1986 . Seed dispersal by adhesion . Annual Review of Ecology and Systematics 17 : 443 – 463 . Google Scholar CrossRef Search ADS Silvestro D , Michalak I . 2012 . raxmlGUI: a graphical front-end for RAxML . Organisms Diversity & Evolution 12 : 335 – 337 . Google Scholar CrossRef Search ADS Stamatakis A . 2006 . RAxML-VI-HPC: maximum likelihood-based phylogenetic analyses with thousands of taxa and mixed models . Bioinformatics 22 : 2688 – 2690 . Google Scholar CrossRef Search ADS PubMed Stebbins GL . 1971 . Adaptive radiation of reproductive characteristics in angiosperms, II: Seeds and seedlings . Annual Review of Ecology and Systematics 2 : 237 – 260 . Google Scholar CrossRef Search ADS Sukhorukov AP , Zhang ML . 2013 . Fruit and seed anatomy of Chenopodium and related genera (Chenopodioideae, Chenopodiaceae/Amaranthaceae): implications for evolution and taxonomy . PLoS ONE 8 : e61906 . Google Scholar CrossRef Search ADS PubMed Swofford DL . 2002 . PAUP* Phylogenetic Analysis Using Parsimony (*and other methods). Version 4 . Sunderland, MA : Sinauer Associates . Townsend CC . 1993 . Amaranthaceae . In: Kubitzki K , Rohwer JG , Bittrich V , eds. Families and genera of vascular plants , Vol. 2 . Berlin : Springer . Google Scholar CrossRef Search ADS de Vos JM , Hughes CE , Schneeweiss GM , Moore BM , Conti EC . 2014 . Heterostyly accelerates diversification via reduced extinction in primroses . Proceedings of the Royal Society B Biological Sciences 281 : 20140075 . Google Scholar CrossRef Search ADS PubMed White F . 1983 . The vegetation of Africa. A descriptive memoir to accompany the UNESCO/AETFAT/UNSO vegetation map of Africa. Natural Resources Research, No. 20 . Paris : UNESCO . Will H , Maussner S , Tackenberg O . 2007 . Experimental studies of diaspore attachment to animal coats: predicting epizoochorous dispersal potential . Oecologia 153 : 331 – 339 . Google Scholar CrossRef Search ADS PubMed Will H , Tackenberg O . 2008 . A mechanistic simulation model of seed dispersal by animals . Journal of Ecology 96 : 1011 – 1022 . Google Scholar CrossRef Search ADS © The Author(s) 2018. Published by Oxford University Press on behalf of the Annals of Botany Company. All rights reserved. For permissions, please e-mail: [email protected]. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)
The links between leaf hydraulic vulnerability to drought and key aspects of leaf venation and xylem anatomy among 26 Australian woody angiosperms from contrasting climates2018 Annals of Botany
doi: 10.1093/aob/mcy051pmid: 29668853
Abstract Background and Aims The structural properties of leaf venation and xylem anatomy strongly influence leaf hydraulics, including the ability of leaves to maintain hydraulic function during drought. Here we examined the strength of the links between different leaf venation traits and leaf hydraulic vulnerability to drought (expressed as P50leaf by rehydration kinetics) in a diverse group of 26 woody angiosperm species, representing a wide range of leaf vulnerabilities, from four low-nutrient sites with contrasting rainfall across eastern Australia. Methods For each species we measured key aspects of leaf venation design, xylem anatomy and leaf morphology. We also assessed for the first time the scaling relationships between hydraulically weighted vessel wall thickness (th) and lumen breadth (bh) across vein orders and habitats. Key Results Across species, variation in P50leaf was strongly correlated with the ratio of vessel wall thickness (th) to lumen breadth (bh) [(t/b)h; an index of conduit reinforcement] at each leaf vein order. Concomitantly, the scaling relationship between th and bh was similar across vein orders, with a log–log slope less than 1 indicating greater xylem reinforcement in smaller vessels. In contrast, P50leaf was not related to th and bh individually, to major vein density (Dvmajor) or to leaf size. Principal components analysis revealed two largely orthogonal trait groupings linked to variation in leaf size and drought tolerance. Conclusions Our results indicate that xylem conduit reinforcement occurs throughout leaf venation, and remains closely linked to leaf drought tolerance irrespective of leaf size. Leaf hydraulic vulnerability, xylem anatomy, leaf venation, vein density, xylem reinforcement, leaf size, drought INTRODUCTION The anatomical and architectural features of leaf venation strongly influence plant productivity and survival across species and environments. Given that the efficiency of water transport through leaf veins is a major determinant of maximum rates of photosynthesis (Brodribb et al., 2007; Sack and Scoffoni, 2013), venation traits that influence water transport efficiency, such as xylem vessel width (Aasamaa et al., 2001), vessel perforation-plate anatomy (Feild and Brodribb, 2013) and vein density (Sack and Frole, 2006; Brodribb and Feild, 2010; Walls, 2011; Buckley et al., 2015; Gleason et al., 2016), have been examined across large numbers of species. Recent studies have also identified several venation traits related to the ability of leaves to resist hydraulic decline under increasing levels of drought stress (Cochard et al., 2004; Brodribb and Holbrook, 2005; Blackman et al., 2010; Scoffoni et al., 2011, 2017b; Nardini et al., 2012). Quantifying these traits offers a potentially useful approach for screening leaf drought tolerance thresholds in extant species, as well as those in the fossil record (Sack and Scoffoni, 2013). As soils dry out during drought, tension (water potential, in MPa) within the leaf xylem increases. Under relatively mild drought conditions, this process can cause leaf water transport capacity (Kleaf) to decline as a result of turgor loss and leaf shrinkage (Scoffoni et al., 2014, 2017a; Trifilo et al., 2016). Under more severe drought conditions, further increases in xylem tension can exceed species hydraulic safety thresholds, causing Kleaf to decline as a result of embolism formation (air blockages) in the water-conducting xylem (Johnson et al., 2009; Brodribb et al., 2016b). If drought continues, this process can lead to complete leaf hydraulic failure and even plant death (Brodribb and Cochard, 2009; Scholz et al., 2014). The ability of leaves to resist hydraulic decline during drought is typically characterized by their hydraulic vulnerability, measured as the water potential associated with 50 % loss in hydraulic conductance, or P50leaf. Recent studies indicate that P50leaf varies widely across species from environments with contrasting rainfall (Brodribb and Hill, 1999; Blackman et al., 2014) and temperature (Nardini and Luglio, 2014), and represents a major determinant of species distributional limits (Blackman et al., 2012; Nardini et al., 2012). Leaf hydraulic vulnerability to drought is an integrated trait derived from different structural and functional characteristics of the leaf water transport pathway. Recent cross-species studies have reported close linkages between variation in P50leaf and specific aspects of leaf vein anatomy and venation design. These studies suggest that angiosperm species with low hydraulic vulnerability (i.e. more negative P50leaf) tend to have leaves with narrow xylem conduits that help minimize the spread of drought-induced embolism (Nardini et al., 2012; Scoffoni et al., 2017b) and high major vein density that provides multiple pathways for water movement around air-filled conduits (hydraulic redundancy) (Scoffoni et al., 2011; Nardini et al., 2014). A strong correlation has also been found between leaf hydraulic vulnerability and the ratio of conduit wall thickness (t) to lumen breadth (b) in leaf minor veins of conifer (Cochard et al., 2004; Brodribb and Holbrook, 2005) and angiosperm (Blackman et al., 2010) species. These findings suggest xylem conduit reinforcement provides a degree of safety from vessel wall collapse during drought. However, it remains unknown to what degree xylem reinforcement occurs throughout the leaf venation network, and whether the scaling of t and b varies across species from different habitats. Euler buckling theory suggests that t should scale proportionately with b to prevent collapse as the breadth of conduits increases (Hacke et al., 2001; Brodribb and Holbrook, 2005). Scaling less than proportionately with b would indicate stronger xylem reinforcement in smaller vessels. If the th–bh scaling exponent shifts across habitats with species operating at lower water potentials displaying a coefficient closer to unity, then this would indicate an increasing cost to constructing leaves with large vessels in arid habitats. As far as we are aware, these possibilities have not previously been assessed in leaves. Although relationships between P50leaf and different venation and xylem anatomy traits have been examined across small groups of ecologically diverse species, it remains unknown whether specific leaf venation traits can become decoupled from P50leaf due to their intrinsic link to leaf size. Leaf size is closely linked to major vein density (Scoffoni et al., 2011) as a consequence of vein packing constraints during leaf development (Sack et al., 2012), and to petiole vessel size (McCulloh et al., 2009; Gleason et al., 2016, 2018), for optimal leaf water transport efficiency (Sack et al., 2003). The link to P50leaf helps explain the propensity of small-leaved species to occupy more arid environments (Scoffoni et al., 2011). However, leaf size can be influenced by multiple environmental factors, including rainfall, temperature, light and nutrient conditions (Givnish, 1987; Cunningham et al., 1999; Fonseca et al., 2000; Tozer et al., 2015), whereas P50leaf is most strongly influenced by site water availability (Brodribb and Cochard, 2009; Blackman et al., 2014; Nardini and Luglio, 2014; Scholz et al., 2014). For species where leaf size is strongly constrained by selection pressures other than rainfall, venation traits intrinsically linked to leaf size might be expected to become decoupled from P50leaf. Here, we tested the level of coordination among different leaf xylem anatomy and venation traits, leaf size, leaf mass per unit area (LMA) and leaf hydraulic vulnerability to drought (P50leaf) across a phylogenetically diverse group of eastern Australian temperate and sub-tropical woody angiosperms. We collected leaves from 26 species that varied strongly in leaf hydraulic vulnerability from four sites characterised by different rainfall, but similarly poor nutrient conditions. For each species, we measured hydraulically weighted diameter (bh), wall thickness (th) and an index of implosion resistance (t/b)h of xylem vessels within the petiole, midrib, 2° veins and leaf minor veins. We also measured leaf major and minor vein density, as well as leaf size and LMA. P50leaf values were sourced from previously published vulnerability curves (Blackman et al., 2014). We asked: (1) What are the venation traits most strongly linked to P50leaf across species? (2) Does the scaling of vessel wall thickness to vessel lumen breadth depart from proportionality within individual leaves, across species, or among habitats? (3) Assuming that in our species group leaf size is constrained by both low rainfall and low soil nutrients, can P50leaf vary independently from leaf venation traits that are intrinsically linked to leaf size? MATERIALS AND METHODS Study sites and species Twenty-six species representing ten families were sampled from four sites across coastal and inland eastern Australia (Table 1; Supplementary Data Table S1). Three of these sites (Warm-Wet, Warm-Dry and Warm-Arid) were associated with a strong east–west aridity gradient in New South Wales, while the fourth site (Hot-Dry) was located in seasonally dry eucalypt woodland in northern Queensland (for more detailed site climate descriptions see Gleason et al., 2012). All four sites were characterized by late successional vegetation with eucalyptus (senso lato) occurring on weathered oligotrophic soils, low in phosphorous (Gleason et al., 2012). The sites varied strongly in rainfall from 383 mm annually at the Warm-Arid site to 1210 mm at the Warm-Wet site (Table 1). Six or seven dominant shrub and/or tree species were sampled from each site. The sample group contained a variety of simple leaf types including flat, revolute, terete and phyllodinous leaves. All species were evergreen except for Planchonia careya from the Hot-Dry site, which was drought-deciduous. In addition to the mean annual precipitation (MAP) for each site, MAP data were downloaded for cleaned occurrence records of each species from the Atlas of Living Australia (http://www.ala.org.au) and used to calculate the MAP across each species distribution. Sampling at each site occurred in 2012–2013, outside of the hot summer months. The same individuals of each species were used for measurements of leaf hydraulics and leaf anatomy. Table 1. The geographical, climatic and edaphic details of each of the four sites sampled from in this study Site Habitat Latitude (°S) Longitude (°E) MAP (mm) MAT (°C) Soil P (mg kg−1) Species (n) P50leaf (MPa) LA (cm2) Kur-ing-gai Warm-Wet 33.68 151.15 1210 17.0 54.8 6 −2.7 ± 0.3a 4.5 ± 3.7a Yengo Warm-Dry 32.78 150.92 779 16.6 223 6 −3.1 ± 0.3a 12.4 ± 6.0ab Round Hill Warm-Arid 32.98 146.16 383 17.9 106 7 −5.5 ± 0.6b 4.9 ± 1.7ab Princess Hills Hot-Dry 18.29 145.49 1139 21.3 157 7 −3.0 ± 0.2a 26.5 ± 4.1b Site Habitat Latitude (°S) Longitude (°E) MAP (mm) MAT (°C) Soil P (mg kg−1) Species (n) P50leaf (MPa) LA (cm2) Kur-ing-gai Warm-Wet 33.68 151.15 1210 17.0 54.8 6 −2.7 ± 0.3a 4.5 ± 3.7a Yengo Warm-Dry 32.78 150.92 779 16.6 223 6 −3.1 ± 0.3a 12.4 ± 6.0ab Round Hill Warm-Arid 32.98 146.16 383 17.9 106 7 −5.5 ± 0.6b 4.9 ± 1.7ab Princess Hills Hot-Dry 18.29 145.49 1139 21.3 157 7 −3.0 ± 0.2a 26.5 ± 4.1b MAP, mean annual precipitation; MAT, mean annual temperature. Climate data were sourced from the Atlas of Living Australia (2017), while soil P was sourced from a previous study (Gleason et al 2012). Also included are site means (plus standard errors) for leaf hydraulic vulnerability (P50leaf) and leaf size (LA); significant differences (P < 0.05) between sites, using pairwise comparisons, are denoted by different superscript letters. View Large Table 1. The geographical, climatic and edaphic details of each of the four sites sampled from in this study Site Habitat Latitude (°S) Longitude (°E) MAP (mm) MAT (°C) Soil P (mg kg−1) Species (n) P50leaf (MPa) LA (cm2) Kur-ing-gai Warm-Wet 33.68 151.15 1210 17.0 54.8 6 −2.7 ± 0.3a 4.5 ± 3.7a Yengo Warm-Dry 32.78 150.92 779 16.6 223 6 −3.1 ± 0.3a 12.4 ± 6.0ab Round Hill Warm-Arid 32.98 146.16 383 17.9 106 7 −5.5 ± 0.6b 4.9 ± 1.7ab Princess Hills Hot-Dry 18.29 145.49 1139 21.3 157 7 −3.0 ± 0.2a 26.5 ± 4.1b Site Habitat Latitude (°S) Longitude (°E) MAP (mm) MAT (°C) Soil P (mg kg−1) Species (n) P50leaf (MPa) LA (cm2) Kur-ing-gai Warm-Wet 33.68 151.15 1210 17.0 54.8 6 −2.7 ± 0.3a 4.5 ± 3.7a Yengo Warm-Dry 32.78 150.92 779 16.6 223 6 −3.1 ± 0.3a 12.4 ± 6.0ab Round Hill Warm-Arid 32.98 146.16 383 17.9 106 7 −5.5 ± 0.6b 4.9 ± 1.7ab Princess Hills Hot-Dry 18.29 145.49 1139 21.3 157 7 −3.0 ± 0.2a 26.5 ± 4.1b MAP, mean annual precipitation; MAT, mean annual temperature. Climate data were sourced from the Atlas of Living Australia (2017), while soil P was sourced from a previous study (Gleason et al 2012). Also included are site means (plus standard errors) for leaf hydraulic vulnerability (P50leaf) and leaf size (LA); significant differences (P < 0.05) between sites, using pairwise comparisons, are denoted by different superscript letters. View Large Leaf hydraulic vulnerability Leaf vulnerability curves for the current group of species were sourced from a previous study published by our lab group (Blackman et al., 2014). In brief, each curve was generated using a modified rehydration technique (Brodribb and Cochard, 2009), whereby leaves or small shoots were excised underwater from branches (three branches per species) dried down over 2–4 d to a range of water potentials and connected to a flow meter. We ensured water potential was equilibrated before each ‘rehydration’ experiment by placing branches into opaque plastic bags for up to 1 h. Measurements were conducted under normal light conditions in the lab or in the field under a shade tent. For most species, the response of leaf hydraulic conductance (Kleaf) to increasing water potential (in MPa) was sigmoidal, with Kleaf not declining over an initial range of water potentials, then declining once species hydraulic safety thresholds were reached (Fig. S1). For each rehydration experiment, leaves were connected to the hydraulic apparatus within 2 s and Kleaf was calculated from the flow rate recorded within the first 4–6 s following leaf connection to the flow meter. These initial flow rates were assumed to be influenced predominantly by the hydraulic resistance of the xylem pathway, and thus we considered the decline in Kleaf to be driven primarily by the formation and spread of xylem embolism (see Nolf et al., 2015; Skelton et al., 2015, 2017a; Brodribb et al., 2016a, b). However, we acknowledge that the decline in Kleaf can also be influenced by measurement light intensity, which has been shown to affect hydraulic processes in leaf tissues beyond the xylem (Guyot et al., 2012; Trifilo et al., 2016). The influence of light intensity on the decline in Kleaf has been demonstrated using the evaporative flux technique (Sack et al., 2002), as well as the timed rehydration kinetics technique, devised by Brodribb and Holbrook (2003), where leaves were allowed to absorb water under high or low light for 15–45 s (Scoffoni et al., 2008). Thus, although we cannot entirely exclude the influence of outside xylem processes, we considered our measurements of leaf hydraulic vulnerability to represent the water potential associated with 50 % loss in hydraulic conductance (P50leaf) driven primarily by embolism formation in the leaf xylem. This contrasts with the evaporative flux method (Sack et al., 2002; Scoffoni et al., 2008) where measurements include mesophyll and stomatal conductance and hence are responsive to factors including light intensity. Across species, P50leaf varied substantially from −1.9 MPa in Banksia serrata to −7.8 MPa in Melaleuca uncinata (Table S1). Leaf xylem anatomy and venation traits Fully expanded sun-leaves were collected from three individuals from each field site at the time leaf hydraulic vulnerability measurements were made. Between five and ten sample leaves from each of three individuals per species were sealed in zip-lock bags with moist paper-towel and placed inside an insulated cool-box. Samples were transported back to the laboratory within 3 d of collecting and fixed in FAA (formalin acetic acid) solution and stored. Measurements of xylem anatomy in petioles, midribs, 2° veins and minor veins were made in one leaf from two to three individuals per species, with the exception of vessels in 2° veins of Pultenaea scabra which were calculated from a single leaf (see Table S1). Leaf area was measured using a flat-bed scanner (Scan Maker i900, Microtek International, China) before sectioning. Transverse sections of vein xylem were made using a vibratome (VT1000s, Leica Microsystems, Germany). Sections were made half-way along the length of the petiole, mid-rib and 2° veins (minor vein anatomy was generally captured within lower order vein sections). Small (<°1 cm2) pieces of leaf, each containing a target vein, were cut out and individually suspended in 6 % agarose blocks. Each agarose block was shaped with a razor-blade and then mounted onto the vibratome stage ensuring that the target vein was perpendicular to the cutting edge of the blade. Several transverse sections were cut at a thickness of between 10 and 20 µm. Sections were stained in dilute 1 % methylene blue before mounting onto glass slides in phenol glycerine jelly. The xylem anatomy of each vein order was photographed using a digital camera (DXM1200F, Nikon, Japan) attached to a light microscope (Bx50, Olympus Optical, Japan). Magnification of each vein depended on vessel size; petioles, mid-ribs and 2° veins were photographed at 40× or 100×, while minor veins were photographed at 100×. From each image, lumen breadth (b) and wall thickness (t) was measured using ImageJ software (National Institutes of Health, USA) from a representative sample of between ten and 100 hydraulically functional vessels. Due to the typically low number of vessels in minor veins, b and t were measured from two or three different minor veins per leaf. For all veins, care was taken to avoid cell-types such as fibre cells and xylem parenchyma that were deemed to provide functional roles beyond water transport. Also, minor veins were identified as the smallest veins in cross section with a clearly defined vascular bundle (xylem and phloem), and were carefully distinguished from free vein endings, which were often enlarged and represented sclereids and/or tracheids in some species. Because vessels were often elliptical in shape, b was measured along the short and long axes of each vessel, and then transformed to the circular equivalent diameter (Choat et al., 2007). Hydraulically weighted vessel diameter (bh) was calculated according to the formula bh = Σ(b4/n)0.25, which weights the vessels (n) within each vein order by their hydraulic contribution to total vein conductance (Tyree and Zimmermann, 2002). For each hydraulically weighted diameter we estimated its wall thickness from the ordinary least-squares relationship between t and b measured across a subsample of 10–15 vessels within each vein order. For each cell, b was measured as described above, while t was measured as the single-thickness of a clearly defined radial wall. The level of xylem reinforcement of hydraulically weighted vessels was then calculated as the ratio of wall thickness (th) and lumen breadth (bh), (t/b)h. We also examined the shape and slope of the th–bh relationship across vein orders and species. Euler buckling theory suggests that th should scale proportionately with bh (i.e. an expected log–log slope of 1) to maintain a constant crushing tension as vessel radii narrow from petioles to minor veins (Hacke et al., 2001; Brodribb and Holbrook, 2005). Less than proportional scaling between these two vessel traits would suggest that large leaves [i.e. with large vessels (McCulloh et al., 2009; Gleason et al., 2018)] represent a savings in network construction costs; for example, a doubling leaf size would result in a somewhat less than doubling of network construction costs. Furthermore, the slope of this relationship (i.e. the th and bh ratio assuming a y-intercept of zero) may also differ across species and habitats. A change in the slope (but not the shape) of the function would indicate greater carbon investment (thicker vessel walls) at all points throughout the network. To test if the slope or shape of the th–bh relationship differed across species or habitats, we plotted th–bh on log10-transformed axes and compared the log–log slopes (i.e. departure from proportionality; scaling exponents), as well as the log–log intercepts (i.e. the logged arithmetic slopes; normalization constants) among species and habitats. Log–log intercepts were only compared if there was no difference in slope among species or habitats. The ‘sma’ function in the SMATR package for R was used for these analyses (Warton et al., 2006). Leaf venation architecture was characterized using one leaf from each of three individuals per species. For species with flat or revolute leaves, we used a protocol described by Scoffoni et al. (2011) for leaf clearing and quantifying vein density. In brief, leaves were chemically cleared with 5 % NaOH, put through a dehydration series in ethanol, stained with saffranin and counter-stained with fast green. Leaves were mounted in water on transparency film and scanned at high resolution using a flatbed scanner (Scan Maker i900, Microtek International, China). The leaf area and lengths of midribs and 2° veins were measured using ImageJ. To ensure that 3° and higher order veins were visible, we exposed the veins prior to leaf clearing by cutting a small window (<1 cm2) through the epidermis and top layers of mesophyll. For large leaves (>10 cm2), three vein windows were made, located centrally in the top, middle and bottom thirds of the leaf. The lengths of 3° and minor veins were measured (using ImageJ) from photographs of these vein windows taken with a digital camera (DXM1200F, Nikon) attached to a light microscope (Bx50, Olympus Optical) at 4× and 10× magnification, respectively. Vein density was calculated for each vein order as the length of vein per unit leaf area. In large leaves, 3° and minor vein densities were averaged across the three exposed windows. The major vein density (Dvmajor) was determined as the sum of 1°, 2° and 3° order vein densities, and minor vein density (Dvminor) as the total length per unit area of 4° and higher order veins. For the four species with terete leaves with parallel leaf venation, vein orders were distinguished by size class in transverse section (see above for leaf sectioning protocol). The vein density of each vein order was then calculated as the sum of the number of veins within each vein order, multiplied by leaf length (assumed to be equivalent to vein length) and divided by projected leaf area. Leaf structural traits Leaf area (LA) was determined for each species from the same three leaves used for quantifying xylem anatomy traits in cross section. Leaves were imaged on a standard flatbed scanner (Epson Perfection V33, Australia). Projected leaf area was calculated from these images using ImageJ software (National Institutes of Health). Leaf mass per unit area (LMA) values for each species were taken from a previous study conducted at the same sites (Gleason et al., 2012). Statistical analysis Bivariate relationships were fit with ordinary least squares (OLS) or standard major axes (SMA) models using the ‘smatr’ package in R (Warton et al., 2006). Differences among sites and vein orders may manifest as different slope or intercept coefficients. When relationships were well approximated by power models (e.g. th ~ a.bhb), scaling exponents were evaluated among groups by testing the log–log slope coefficient (b; scaling exponent) as well as the intercept coefficient (a; scaling constant) when slopes were statistically similar among sites or vein orders. Principal components analysis (PCA) (‘prcomp’ function in R; R Core Team, 2015) was used to determine the dominant axes of variation among a selection of traits, including petiole vessel th, bh and (t/b)h, respectively, as well as major and minor vein density, LA, LMA and P50leaf. All variables were scaled to unit variance in the PCA. RESULTS Substantial variation in venation architecture, leaf xylem anatomy and gross morphology was observed across species (Table S1). Major vein density varied ~15-fold from 0.58 to 9.2 mm mm−2, while among the xylem anatomy traits petiole vessel bh varied ~6.5-fold from 3.3 to 24.4 µm, petiole vessel th varied ~2.5-fold from 0.67 to 2.7 µm, and petiole vessel (t/b)h varied ~2-fold from 0.08 to 0.25. Across species, LMA varied nearly 5-fold, ranging from 75 to 447 g m−2, and leaf area varied more than 300-fold, ranging from 0.12 to 40.0 cm2. Among sites, leaf size differed significantly between the Warm-Wet and the Hot-Dry sites (pairwise comparisons; Table 1), while leaf size was unrelated to site rainfall and weakly correlated with soil phosphorus (r2 = 0.15, P = 0.05; Fig. S2), and was unrelated to species mean annual rainfall (Table 2; Fig. S3). In contrast, P50leaf was significantly different among species from the Warm-Arid site compared to species from the other three sites (Table 1), while P50leaf was strongly correlated with site rainfall (r2 = 0.48, P< 0.001) but not soil phosphorus (Fig. S2), and was correlated with species mean annual rainfall (Table 2; Fig. S3). Table 2. Pearson correlation r values among climate (species mean annual rainfall) and leaf, vein and petiole vessel traits measured in 26 woody angiosperm species MAP P50leaf Pet_t–bh Pet_bh Pet_th Dvmajor Dvminor LMA LA MAP P50leaf −0.76 Pet_t–bh −0.66 0.73 Pet_bh 0.33 −0.34 −0.73 Pet_th −0.06 0.11 −0.23 0.83 Dvmajor −0.20 0.11 0.27 −0.60 −0.63 Dvminor −0.19 0.08 0.01 0.15 0.21 −0.03 LMA −0.29 0.25 0.10 −0.05 0.01 0.37 −0.27 LA 0.18 −0.06 −0.42 0.86 0.89 −0.64 0.14 −0.03 MAP P50leaf Pet_t–bh Pet_bh Pet_th Dvmajor Dvminor LMA LA MAP P50leaf −0.76 Pet_t–bh −0.66 0.73 Pet_bh 0.33 −0.34 −0.73 Pet_th −0.06 0.11 −0.23 0.83 Dvmajor −0.20 0.11 0.27 −0.60 −0.63 Dvminor −0.19 0.08 0.01 0.15 0.21 −0.03 LMA −0.29 0.25 0.10 −0.05 0.01 0.37 −0.27 LA 0.18 −0.06 −0.42 0.86 0.89 −0.64 0.14 −0.03 All data were log-transformed for analysis. Values in bold type are significant at P < 0.05. View Large Table 2. Pearson correlation r values among climate (species mean annual rainfall) and leaf, vein and petiole vessel traits measured in 26 woody angiosperm species MAP P50leaf Pet_t–bh Pet_bh Pet_th Dvmajor Dvminor LMA LA MAP P50leaf −0.76 Pet_t–bh −0.66 0.73 Pet_bh 0.33 −0.34 −0.73 Pet_th −0.06 0.11 −0.23 0.83 Dvmajor −0.20 0.11 0.27 −0.60 −0.63 Dvminor −0.19 0.08 0.01 0.15 0.21 −0.03 LMA −0.29 0.25 0.10 −0.05 0.01 0.37 −0.27 LA 0.18 −0.06 −0.42 0.86 0.89 −0.64 0.14 −0.03 MAP P50leaf Pet_t–bh Pet_bh Pet_th Dvmajor Dvminor LMA LA MAP P50leaf −0.76 Pet_t–bh −0.66 0.73 Pet_bh 0.33 −0.34 −0.73 Pet_th −0.06 0.11 −0.23 0.83 Dvmajor −0.20 0.11 0.27 −0.60 −0.63 Dvminor −0.19 0.08 0.01 0.15 0.21 −0.03 LMA −0.29 0.25 0.10 −0.05 0.01 0.37 −0.27 LA 0.18 −0.06 −0.42 0.86 0.89 −0.64 0.14 −0.03 All data were log-transformed for analysis. Values in bold type are significant at P < 0.05. View Large Across species, variation in P50leaf was strongly correlated with vessel (t/b)h of petioles (r2 = 0.53, P < 0.001), midribs (r2 = 0.58, P < 0.001), second-order veins (r2 = 0.46, P < 0.001) and minor veins (r2 = 0.72, P < 0.001) (Fig. 1), indicating that greater resistance to leaf hydraulic dysfunction (more negative P50leaf) is linked to greater xylem vessel reinforcement across all vein orders. Concomitantly, neither log–log slope nor log–log elevation of the relationship between th and bh were significantly different among vein orders (Fig. 2A; slope P = 0.17; elevation P = 0.08), suggesting similar scaling exponents and scaling constants throughout the networks. However, the log–log slope of the th–bh relationship across vein orders was significantly shallower than 1 (P < 0.001), indicating that the ratio of wall thickness to lumen breadth decreased as vessels became wider. Comparing across sites, species from the Warm-Wet site tested as having significantly shallower log–log slope than the other three sites (0.52 vs 0.72 in common for the other three sites; P < 0.025). The suggestion (Fig. 2B) is that larger vessels at this wet site tended to have relatively thinner walls than at the dry and arid sites, but minor veins did not. However, considering the graph (Fig. 2B), we do not attach strong weight to this apparent difference in slope. More clear-cut is that among the three sites exhibiting a common log–log slope (Warm-Dry, Warm-Arid, Hot-Dry), the elevation of the th–bh relationship varied significantly (P < 0.001), with species from the drier sites having larger values, meaning vessel wall thickness in these species exhibited stronger reinforcement at a given lumen diameter than species from wetter sites (Fig. 2B).We note that although petiole (t/b)h was correlated with species MAP (r2 = 0.44, P < 0.001), neither of the individual components of the th/bh ratio were related to rainfall (Table 2). Fig. 1. View largeDownload slide The log–log relationship between P50leaf and hydraulically weighted vessel (t/b)h at each vein order across species: black filled circles = petioles (solid line, r2 = 0.53***); dark-grey filled circles = midribs (dotted line, r2 = 0.58***); light-grey filled circles = 2° veins (dashed line, r2 = 0.46***); white filled circles = minor veins (dash-dot line, r2 = 0.72***). Level of significance: ***P < 0.001. Fig. 1. View largeDownload slide The log–log relationship between P50leaf and hydraulically weighted vessel (t/b)h at each vein order across species: black filled circles = petioles (solid line, r2 = 0.53***); dark-grey filled circles = midribs (dotted line, r2 = 0.58***); light-grey filled circles = 2° veins (dashed line, r2 = 0.46***); white filled circles = minor veins (dash-dot line, r2 = 0.72***). Level of significance: ***P < 0.001. Fig. 2. View largeDownload slide (A) The log–log relationship between vessel wall thickness (th) and lumen breadth (bh) at each vein order across species: black filled circles = petioles (solid line, r2 = 0.69***, slope = 0.7, elevation = −0.61); dark-grey filled circles = midribs (dotted line, r2 = 0.64***, slope = 0.76, elevation = −0.64); light-grey filled circles = 2° veins (dashed line, r2 = 0.60***, slope = 0.62, elevation = −0.65); white filled circles = minor veins (dash-dot line, r2 = 0.31**, slope = 0.99, elevation = −0.58). Neither slope nor elevation of the log–log th–bh relationship varied across vein orders (SMATR analysis: slope P = 0.17, elevation P = 0.08). (B) The log–log th–bh relationship across all vein orders for species separated by site: blue filled circles = Warm-Wet (solid line, r2 = 0.72***, slope = 0.52, elevation = −0.66), green filled circles = Warm-Dry (dashed line, r2 = 0.94***, slope = 0.78, elevation = −0.59); red filled circles = Warm-Arid (dotted line, r2 = 0.68***, slope = 0.69, elevation = −0.51); orange filled circles = Hot-Dry (dash-dot line, r2 = 0.94***, slope = 0.69, elevation = −0.64). Slope of the log–log th–bh relationship differed only for species from the Warm-Wet site (SMATR analysis: P = 0.02). The red dashed line in both plots represents a proportional th–bh scaling slope of 1, where (t/b)h = 0.24. Level of significance: ***P < 0.001; **P < 0.01. Fig. 2. View largeDownload slide (A) The log–log relationship between vessel wall thickness (th) and lumen breadth (bh) at each vein order across species: black filled circles = petioles (solid line, r2 = 0.69***, slope = 0.7, elevation = −0.61); dark-grey filled circles = midribs (dotted line, r2 = 0.64***, slope = 0.76, elevation = −0.64); light-grey filled circles = 2° veins (dashed line, r2 = 0.60***, slope = 0.62, elevation = −0.65); white filled circles = minor veins (dash-dot line, r2 = 0.31**, slope = 0.99, elevation = −0.58). Neither slope nor elevation of the log–log th–bh relationship varied across vein orders (SMATR analysis: slope P = 0.17, elevation P = 0.08). (B) The log–log th–bh relationship across all vein orders for species separated by site: blue filled circles = Warm-Wet (solid line, r2 = 0.72***, slope = 0.52, elevation = −0.66), green filled circles = Warm-Dry (dashed line, r2 = 0.94***, slope = 0.78, elevation = −0.59); red filled circles = Warm-Arid (dotted line, r2 = 0.68***, slope = 0.69, elevation = −0.51); orange filled circles = Hot-Dry (dash-dot line, r2 = 0.94***, slope = 0.69, elevation = −0.64). Slope of the log–log th–bh relationship differed only for species from the Warm-Wet site (SMATR analysis: P = 0.02). The red dashed line in both plots represents a proportional th–bh scaling slope of 1, where (t/b)h = 0.24. Level of significance: ***P < 0.001; **P < 0.01. Across species, variation in P50leaf was not related to either th or bh at any vein order, with the exception of minor vein vessel bh (r2 = 0.17, P = 0.04; Table S2). P50leaf was also unrelated to major vein density (Dvmajor), minor vein density (Dvminor), leaf size (LA) and leaf mass per unit area (LMA) (Table 2). Nonetheless, leaf size was significantly and negatively correlated with major vein density (r2 = 0.42, P < 0.001) (Fig. 3), but was unrelated to minor vein density (Table 2). Leaf size was positively correlated with vessel bh and th in leaf petioles [bh, r2 = 0.75, P < 0.001 (Fig. 3); th, r2 = 0.79, P < 0.001], midribs (bh, r2 = 0.75, P < 0.001; th, r2 = 0.80, P < 0.001) and 2° veins (bh, r2 = 0.61, P < 0.001; th, r2 = 0.43, P < 0.001), but not in leaf minor veins (see Table S2). Leaf size was also related to vessel (t/b)h at the petiole and 2° veins, but not at the midrib or minor veins (Tables 2 and S2). Strong negative correlations were observed between major vein density and vessel bh and th in lower order veins (Table S2). Fig. 3. View largeDownload slide Log–log relationships across species between leaf area and (A) major vein density (r2 = 0.42, P < 0.001) and (B) hydraulically weighted vessel diameter in leaf petioles (r2 = 0.75, P < 0.001). Species values are separated by site (blue filled circles = Warm-Wet, green filled circles = Warm-Dry, red filled circles = Warm-Arid, orange filled circles = Hot-Dry). Fig. 3. View largeDownload slide Log–log relationships across species between leaf area and (A) major vein density (r2 = 0.42, P < 0.001) and (B) hydraulically weighted vessel diameter in leaf petioles (r2 = 0.75, P < 0.001). Species values are separated by site (blue filled circles = Warm-Wet, green filled circles = Warm-Dry, red filled circles = Warm-Arid, orange filled circles = Hot-Dry). PCA of leaf, vein and petiole vessel traits identified two major axes, which cumulatively explained 66 % of the total variation among the traits (Fig. 4). The first principal component (PC) accounted for 46 % of the total variation and was dominated by leaf size, major vein density, and both petiole vessel bh and th, while the second PC (21 % of the total variation) was associated with P50leaf and petiole (t/b)h. Importantly, (t/b)h was closely aligned with P50leaf, whereas the components of this ratio, th and bh, aligned primarily with leaf size and major vein density, and were largely orthogonal to P50leaf and (t/b)h (Fig. 4). Fig. 4. View largeDownload slide Principal components plot of leaf, vein and petiole vessel properties, including petiole vessel th, bh and (t/b)h, major and minor vein density, LA and LMA. Individual points denote individual species. PC1 and PC2 account for 45 and 21 % of the variation among species and sites, respectively. Vector loadings are represented with arrows. All data were log-transformed and scaled to unit variance prior to analysis. Fig. 4. View largeDownload slide Principal components plot of leaf, vein and petiole vessel properties, including petiole vessel th, bh and (t/b)h, major and minor vein density, LA and LMA. Individual points denote individual species. PC1 and PC2 account for 45 and 21 % of the variation among species and sites, respectively. Vector loadings are represented with arrows. All data were log-transformed and scaled to unit variance prior to analysis. DISCUSSION The strong correlations between more negative P50leaf and greater vessel (t/b)h at each vein order across our sample group of 26 species highlight the functional link between xylem conduit reinforcement throughout the leaf venation and the ability of leaves to resist drought-induced hydraulic dysfunction (Blackman et al., 2010), probably caused by xylem embolism in our measurements using direct flow rehydration. The strength of this relationship is robust given that our species group spanned a very wide range of leaf vulnerabilities, with P50leaf values ranging from −1.9 MPa in the most vulnerable species Banksia serrata to −7.8 MPa in Melaleuca uncinata, which to the best of our knowledge is among the most negative P50leaf values recorded for angiosperm species (see also Skelton et al., 2017b). Our sample of species also spanned a wide range of values for each of the leaf venation, xylem anatomy and leaf morphology traits, which increased the likelihood of detecting significant trait–trait relationships across species. Even so, we note that large leaves >41 cm2 were not present in our species group, although compared to previous studies it did contain species with small leaves <1 cm2. The ratio of wall thickness to lumen breadth (t/b) has been used widely as an index of the implosion resistance of conduit walls under tension (Hacke et al., 2001; Sack and Scoffoni, 2013). Hacke et al. (2001) first observed a strong correlation between the ratio of conduit double-wall thickness and lumen breadth and drought resistance in woody stems. They argued that xylem conduit reinforcement should increase with increasing drought resistance on the basis that drought-resistant plants tend to experience stronger internal loads (water potential) in the field. This argument has been supported by studies highlighting the adaptive link between xylem conduit wall reinforcement and embolism resistance (P50) in plant stems (Jacobsen et al., 2005, 2007; Pittermann et al., 2006). In the current study, vessel (t/b)h was higher in species from more arid environments and was strongly correlated with P50leaf across species. This suggests that a similar level of coordination between xylem reinforcement and embolism resistance may be present in leaves. Although we cannot preclude the possibility that cell collapse occurs in angiosperm leaves under tension (see Zhang et al., 2016), the weight of recent evidence suggests that leaf hydraulic decline during severe drought is caused by xylem embolism (Johnson et al., 2009; Nolf et al., 2015; Brodribb et al., 2016a, b; Scoffoni et al., 2017b; Skelton et al., 2017a). Thus, we suggest xylem reinforcement in leaves has evolved to provide a degree of safety from vessel collapse under tension, while embolism spread via air-seeding could potentially relate to the size and structure of pores in pit membranes (Jansen et al., 2009), nucleation from hydrophobic surfaces (Tyree et al., 1994) or, as hypothesized recently, conduit/fluid properties that influence the expansion of nanobubbles (Schenk et al., 2015). We acknowledge that changes in the hydraulic properties of the extra-xylary pathway due to turgor loss and leaf shrinkage (Scoffoni et al., 2014, 2017a; Trifilo et al., 2016), and associated changes in aquaporin expression (Kim and Steudle, 2007), play a major role in the decline of leaf hydraulic conductance measured from petiole to external atmosphere. These processes tend to occur during the early stages of drought, at water potentials preceding those associated with xylem embolism (Scoffoni et al., 2017a), and furthermore are sensitive to measurement light conditions. Indeed, we note that our estimate of maximum Kleaf and thus P50leaf for each species may be underestimated on the basis that our rehydration measurements were conducted under low light and thus did not allow for the influence of outside xylem processes in driving the response of Kleaf during drought (Scoffoni et al., 2008, 2017a). However, our approach of measuring Kleaf under low light is consistent with the approach of other studies that have shown strong correspondence between the decline in Kleaf during drought, measured using the rehydration technique (Brodribb and Cochard, 2009), and the accumulation of xylem embolisms detected acoustically (Nolf et al., 2015) and visually via a recently developed optical technique (Brodribb et al., 2016a, b) and X-ray micro-computed tomography (Skelton et al., 2017a). Furthermore, in contrast to the evaporative flux technique (Sack et al., 2002) and the timed rehydration kinetics technique (Brodribb and Holbrook, 2003), we calculated Kleaf from the initial flow rate into the leaf through the petiole, which is more likely to be influenced by within-xylem rather than outside-xylem processes. Thus, it is reasonable to assume that the decline in Kleaf observed in our vulnerability curves was largely driven by embolism formation, although we note that further studies across diverse species are required to test the relative influence of xylem and outside xylem processes on the decline in Kleaf during drought measured using different measurement techniques under a range of conditions. Nevertheless, if embolism is the primary driver of the decline in Kleaf in vulnerability curves generated using the rehydration technique, then our study provides important mechanistic insights into the linkages between P50leaf and leaf venation and xylem anatomy traits related to xylem embolism resistance. Previous studies have observed strong links between xylem wall reinforcement in leaf minor veins and both leaf vulnerability to drought (Blackman et al., 2010) and species climatic limits (Jordan et al., 2013). In the current study, we observed for the first time strong relationships between P50leaf and vessel (t/b)h at each leaf vein order, with the scaling of vessel wall thickness to lumen breadth being consistent across vein orders. These results indicate that xylem conduit reinforcement occurs throughout the leaf venation network, from petioles to minor veins. Nonetheless, across species we found that the level of xylem vessel reinforcement was higher in smaller vessels (i.e. the t–b scaling exponent was significantly less than 1). This suggests that larger vessels perhaps require less reinforcement than smaller vessels, and/or that tension in vessels (negative water potential) increases from petiole to minor veins. Furthermore, it is consistent with findings that smaller minor veins are the most resistant to embolism within the leaf vein network (Brodribb et al., 2016a, b; Scoffoni et al., 2017b). Across habitats, we hypothesized that increasing the log–log t–b scaling exponent towards unity (Euler buckling exponent) would result in lower ‘network’ efficiency as a consequence of increasing carbon costs per unit increase in hydraulically weighted vessel diameter, which might play a role in constraining vessel size in species from more arid sites. Unfortunately, our data are ambiguous on this point. The least arid site did have a shallower scaling exponent than the other sites, but the ‘dry’ sites and ‘arid’ site had similar scaling exponents, which do not support this idea. However, among these dry-land sites the th–bh constant (i.e. elevation in Fig. 2B) varied significantly and increased with increasing site aridity. This suggests that natural selection acts predominantly on the scaling constant rather than the scaling exponent, which results in thicker walls relative to lumen diameter in drier habitats. Across our set of species, leaf hydraulic vulnerability to drought was unrelated to major vein density. This contrasts with previous studies (Scoffoni et al., 2011; Nardini et al., 2012) that found significant relationships between high major vein density and low hydraulic vulnerability in leaves. These authors suggested high major vein density confers increased drought tolerance on the basis that it provides more alternative pathways for water movement around vein embolisms. We acknowledge that high major vein density and hydraulic redundancy may play a role maintaining leaf hydraulic function under drought. Nonetheless, our results suggest that the functional link between higher major vein density and greater resistance to embolism-induced hydraulic decline may not hold across sets of species where leaf size is strongly influenced by factors in addition to water availability. In the current study, we observed a strong negative relationship between leaf size and major vein density and a strong positive relationship between leaf size and hydraulically weighted petiole diameter, consistent with the intrinsic links and scaling relationships between these traits (McCulloh et al., 2009; Sack et al., 2012; Gleason et al., 2018). These bivariate relationships were supported by PCA, which grouped together vessel and venation traits linked to leaf size. However, our analyses showed that leaf size varied independently of P50leaf. Although all species with large leaves (>10 cm2) had P50leaf values less negative than −5 MPa, species with small leaves (<1 cm2) were characterized by both high and low vulnerability (i.e. P50leaf values ranged from roughly −2 to −8 MPa). This implies that the relationship between P50leaf and leaf size – and thus the relationship between P50leaf and both major vein density and petiole vessel diameter – can break down across species from different habitats, especially where leaf size is constrained by multiple environmental factors. Indeed, in contrast to global trends (Wright et al., 2017), across our four temperate–tropical sample sites, mean leaf size was unrelated to climate indices of site water availability. In contrast, P50leaf is known to vary more systematically with rainfall across these sites (Blackman et al., 2014). Decoupling of leaf size and site water availability (and P50leaf) is consistent with poor leaf size–climate relationships reported for Australian vegetation (Peppe et al., 2011; Tozer et al., 2015), indicating that leaf size variation can be shaped by additional environmental filters, including low soil nutrients (Cunningham et al., 1999; Fonseca et al., 2000). Nevertheless, our findings across a morphologically diverse set of species suggest that smaller leaves and higher vein density are not necessarily conferring strongly negative P50leaf. The results of this study clearly indicate that the ability of leaves to resist hydraulic dysfunction under drought is closely related to the degree of leaf xylem reinforcement throughout the venation network. Unlike traits such as major vein density and petiole vessel diameter, the degree of xylem vessel reinforcement was unrelated to leaf size, which in turn was unrelated to P50leaf. These results point strongly to the usefulness of measuring xylem reinforcement in leaf veins when examining variation in leaf hydraulic vulnerability to drought across ecologically and morphologically diverse species. SUPPLEMENTARY DATA Supplementary data are available online at https://academic.oup.com/aob and consist of the following. Table S1: Species information and trait values measured for each species. Table S2: Pearson correlation r values among selected leaf traits. Figure S1: Leaf hydraulic vulnerability curves for each of the 26 species included in this study. Figure S2: The log–log relationship between leaf traits and site climate and soil characteristics. Figure S3: The log–log relationship between leaf traits and species mean annual precipitation. ACKNOWLEDGEMENTS This research was supported by an Australian Research Council fellowship awarded to M.W. We thank the Macquarie University Microscopy Unit for their help with leaf sectioning and imaging. Thanks are also given to New South Wales National Parks and Wildlife Service, and the Queensland Department of National Parks, Recreation, Sport and Racing for allowing access to field sites. C.J.B., S.M.G. and M.W. planned and designed the research. C.J.B., S.M.G., A.M.C., Y.C. and C.A.L. carried out field sampling, leaf sectioning/clearing and image analysis. C.J.B and S.M.G carried out data analysis. C.J.B, S.M.G. and M.W. wrote the manuscript. LITERATURE CITED Aasamaa K , Sober A , Rahi M . 2001 . Leaf anatomical characteristics associated with shoot hydraulic conductance, stomatal conductance and stomatal sensitivity to changes of leaf water status in temperate deciduous trees . Australian Journal of Plant Physiology 28 : 765 – 774 . Blackman CJ , Brodribb T , Jordan GJ . 2010 . Leaf hydraulic vulnerability is related to conduit dimensions and drought resistance across a diverse range of woody angiosperms . New Phytologist 188 : 1113 – 1123 . Google Scholar CrossRef Search ADS PubMed Blackman CJ , Brodribb TJ , Jordan GJ . 2012 . Leaf hydraulic vulnerability influences species’ bioclimatic limits in a diverse group of woody angiosperms . Oecologia 168 : 1 – 10 . Google Scholar CrossRef Search ADS PubMed Blackman CJ , Gleason SM , Chang Y , Cook AM , Laws C , Westoby M . 2014 . Leaf hydraulic vulnerability to drought is linked to site water availability across a broad range of species and climates . Annals of Botany 114 : 435 – 440 . Google Scholar CrossRef Search ADS PubMed Brodribb TJ , Cochard H . 2009 . Hydraulic failure defines the recovery and point of death in water-stressed conifers . Plant Physiology 149 : 575 – 584 . Google Scholar CrossRef Search ADS PubMed Brodribb TJ , Feild TS . 2010 . Leaf hydraulic evolution led a surge in leaf photosynthetic capacity during early angiosperm diversification . Ecology Letters 13 : 175 – 183 . Google Scholar CrossRef Search ADS PubMed Brodribb T , Hill RS . 1999 . The importance of xylem constraints in the distribution of conifer species . New Phytologist 143 : 365 – 372 . Google Scholar CrossRef Search ADS Brodribb TJ , Holbrook NM . 2003 . Stomatal closure during leaf dehydration, correlation with other leaf physiological traits . Plant Physiology 132 : 2166 – 2173 . Google Scholar CrossRef Search ADS PubMed Brodribb TJ , Holbrook NM . 2005 . Water stress deforms tracheids peripheral to the leaf vein of a tropical conifer . Plant Physiology 137 : 1139 – 1146 . Google Scholar CrossRef Search ADS PubMed Brodribb TJ , Feild TS , Jordan GJ . 2007 . Leaf maximum photosynthetic rate and venation are linked by hydraulics . Plant Physiology 144 : 1890 – 1898 . Google Scholar CrossRef Search ADS PubMed Brodribb TJ , Bienaime D , Marmottant P . 2016a . Revealing catastrophic failure of leaf networks under stress . Proceedings of the National Academy of Sciences of the United States of America 113 : 4865 – 4869 . Google Scholar CrossRef Search ADS Brodribb TJ , Skelton RP , McAdam SAM , Bienaime D , Lucani CJ , Marmottant P . 2016b . Visual quantification of embolism reveals leaf vulnerability to hydraulic failure . New Phytologist 209 : 1403 – 1409 . Google Scholar CrossRef Search ADS Buckley TN , John GP , Scoffoni C , Sack L . 2015 . How does leaf anatomy influence water transport outside the xylem ? Plant Physiology 168 : 1616 – 1635 . Google Scholar CrossRef Search ADS PubMed Choat B , Sack L , Holbrook NM . 2007 . Diversity of hydraulic traits in nine Cordia species growing in tropical forests with contrasting precipitation . New Phytologist 175 : 686 – 698 . Google Scholar CrossRef Search ADS PubMed Cochard H , Froux F , Mayr FFS , Coutand C . 2004 . Xylem wall collapse in water-stressed pine needles . Plant Physiology 134 : 401 – 408 . Google Scholar CrossRef Search ADS PubMed Cunningham SA , Summerhayes B , Westoby M . 1999 . Evolutionary divergences in leaf structure and chemistry, comparing rainfall and soil nutrient gradients . Ecological Monographs 69 : 569 – 588 . Google Scholar CrossRef Search ADS Feild TS , Brodribb TJ . 2013 . Hydraulic tuning of vein cell microstructure in the evolution of angiosperm venation networks . New Phytologist 199 : 720 – 726 . Google Scholar CrossRef Search ADS PubMed Fonseca CR , Overton JMC , Collins B , Westoby M . 2000 . Shifts in trait-combinations along rainfall and phosphorus gradients . Journal of Ecology 88 : 964 – 977 . Google Scholar CrossRef Search ADS Givnish TJ . 1987 . Comparative studies of leaf form: assessing the relative roles of selective pressures and phylogenetic constraints . New Phytologist 106 : 131 – 160 . Google Scholar CrossRef Search ADS Gleason SM , Butler DW , Zieminska K , Waryszak P , Westoby M . 2012 . Stem xylem conductivity is key to plant water balance across Australian angiosperm species . Functional Ecology 26 : 343 – 352 . Google Scholar CrossRef Search ADS Gleason SM , Blackman CJ , Chang Y , Cook AM , Laws CA , Westoby M . 2016 . Weak coordination among petiole, leaf, vein, and gas-exchange traits across Australian angiosperm species and its possible implications . Ecology and Evolution 6 : 267 – 278 . Google Scholar CrossRef Search ADS PubMed Gleason SM , Blackman CJ , Gleason ST , McCulloh KA , Ocheltree TW , Westoby M . 2018 . Vessel scaling in evergreen angiosperm leaves conforms with Murray’s law and area-filling assumptions: implications for plant size, leaf size, and cold tolerance . New Phytologist . https://doi.org/10.1111/nph.15116 . Guyot G , Scoffoni C , Sack L . 2012 . Combined impacts of irradiance and dehydration on leaf hydraulic conductance: insights into vulnerability and stomatal control . Plant Cell and Environment 35 : 857 – 871 . Google Scholar CrossRef Search ADS Hacke UG , Sperry JS , Pockman WT , Davis SD , McCulloh KA . 2001 . Trends in wood density and structure are linked to prevention of xylem implosion by negative pressure . Oecologia 126 : 457 – 461 . Google Scholar CrossRef Search ADS PubMed Jacobsen AL , Ewers FW , Pratt RB , Paddock III WA , Davis SD . 2005 . Do xylem fibers affect vessel cavitation resistance ? Plant Physiology 139 : 546 – 556 . Google Scholar CrossRef Search ADS PubMed Jacobsen AL , Agenbag L , Esler KJ , Pratt RB , Ewers FW , Davis SD . 2007 . Xylem density, biomechanics and anatomical traits correlate with water stress in 17 evergreen shrub species of the Mediterranean-type climate region of South Africa . Journal of Ecology 95 : 171 – 183 . Google Scholar CrossRef Search ADS Jansen S , Choat B , Pletsers A . 2009 . Morphological variation of intervessel pit membranes and implications to xylem function in angiosperms . American Journal of Botany 96 : 409 – 419 . Google Scholar CrossRef Search ADS PubMed Johnson DM , Meinzer FC , Woodruff DR , McCulloh KA . 2009 . Leaf xylem embolism, detected acoustically and by cryo-SEM, corresponds to decreases in leaf hydraulic conductance in four evergreen species . Plant, Cell and Environment 32 : 828 – 836 . Google Scholar CrossRef Search ADS Jordan GJ , Brodribb TJ , Blackman CJ , Weston PH . 2013 . Climate drives vein anatomy in Proteaceae . American Journal of Botany 100 : 1483 – 1493 . Google Scholar CrossRef Search ADS PubMed Kim YX , Steudle E . 2007 . Light and turgor affect the water permeability (aquaporins) of parenchyma cells in the midrib of leaves of Zea mays . Journal of Experimental Botany 58 : 4119 – 4129 . Google Scholar CrossRef Search ADS PubMed McCulloh KA , Sperry JS , Meinzer FC , Lachenbruch B , Atala C . 2009 . Murray’s law, the ‘Yarrum’ optimum, and the hydraulic architecture of compound leaves . New Phytologist 184 : 234 – 244 . Google Scholar CrossRef Search ADS PubMed Nardini A , Luglio J . 2014 . Leaf hydraulic capacity and drought vulnerability: possible trade-offs and correlations with climate across three major biomes . Functional Ecology 28 : 810 – 818 . Google Scholar CrossRef Search ADS Nardini A , Peda G , La Rocca N . 2012 . Trade-offs between leaf hydraulic capacity and drought vulnerability: morpho-anatomical bases, carbon costs and ecological consequences . New Phytologist 196 : 788 – 798 . Google Scholar CrossRef Search ADS PubMed Nardini A , Ounapuu-Pikas E , Savi T . 2014 . When smaller is better: leaf hydraulic conductance and drought vulnerability correlate to leaf size and venation density across four Coffea arabica genotypes . Functional Plant Biology 41 : 972 – 982 . Google Scholar CrossRef Search ADS Nolf M , Creek D , Duursma R , Holtum J , Mayr S , Choat B . 2015 . Stem and leaf hydraulic properties are finely coordinated in three tropical rain forest tree species . Plant Cell and Environment 38 : 2652 – 2661 . Google Scholar CrossRef Search ADS Peppe DJ , Royer DL , Cariglino B , et al. 2011 . Sensitivity of leaf size and shape to climate: global patterns and paleoclimatic applications . New Phytologist 190 : 724 – 739 . Google Scholar CrossRef Search ADS PubMed Pittermann J , Sperry JS , Hacke UG , Wheeler JK , Sikkema EH . 2006 . Inter-tracheid pitting and the hydraulic efficiency of conifer wood: the role of tracheid allometry and cavitation protection . American Journal of Botany 93 : 1265 – 1273 . Google Scholar CrossRef Search ADS PubMed Sack L , Frole K . 2006 . Leaf structural diversity is related to hydraulic capacity in tropical rain forest trees . Ecology 87 : 483 – 491 . Google Scholar CrossRef Search ADS PubMed Sack L , Scoffoni C . 2013 . Leaf venation: structure, function, development, evolution, ecology and applications in the past, present and future . New Phytologist 198 : 983 – 1000 . Google Scholar CrossRef Search ADS PubMed Sack L , Melcher PJ , Zwieniecki MA , Holbrook NM . 2002 . The hydraulic conductance of the angiosperm leaf lamina: a comparison of three measurement methods . Journal of Experimental Botany 53 : 2177 – 2184 . Google Scholar CrossRef Search ADS PubMed Sack L , Cowan PD , Jaikumar N , Holbrook NM . 2003 . The ‘hydrology’ of leaves: co-ordination of structure and function in temperate woody species . Plant Cell and Environment 26 : 1343 – 1356 . Google Scholar CrossRef Search ADS Sack L , Scoffoni C , McKown AD , et al. 2012 . Developmentally based scaling of leaf venation architecture explains global ecological patterns . Nature Communications 3 : 837 . Google Scholar CrossRef Search ADS PubMed Schenk HJ , Steppe K , Jansen S . 2015 . Nanobubbles: a new paradigm for air-seeding in xylem . Trends in Plant Science 20 : 199 – 205 . Google Scholar CrossRef Search ADS PubMed Scholz FG , Bucci SJ , Goldstein G . 2014 . Strong hydraulic segmentation and leaf senescence due to dehydration may trigger die-back in Nothofagus dombeyi under severe droughts: a comparison with the co-occurring Austrocedrus chilensis . Trees-Structure and Function 28 : 1475 – 1487 . Google Scholar CrossRef Search ADS Scoffoni C , Pou A , Aasamaa K , Sack L . 2008 . The rapid light response of leaf hydraulic conductance: new evidence from two experimental methods . Plant Cell and Environment 31 : 1803 – 1812 . Google Scholar CrossRef Search ADS Scoffoni C , Rawls M , McKown A , Cochard H , Sack L . 2011 . Decline of leaf hydraulic conductance with dehydration: relationship to leaf size and venation architecture . Plant Physiology 156 : 832 – 843 . Google Scholar CrossRef Search ADS PubMed Scoffoni C , Vuong C , Diep S , Cochard H , Sack L . 2014 . Leaf shrinkage with dehydration: coordination with hydraulic vulnerability and drought tolerance . Plant Physiology 164 : 1772 – 88 . Google Scholar CrossRef Search ADS PubMed Scoffoni C , Albuquerque C , Brodersen CR , et al. 2017a . Outside-xylem vulnerability, not xylem embolism, controls leaf hydraulic decline during dehydration . Plant Physiology 173 : 1197 – 1210 . Google Scholar CrossRef Search ADS Scoffoni C , Albuquerque C , Brodersen CR , et al. 2017b . Leaf vein xylem conduit diameter influences susceptibility to embolism and hydraulic decline . New Phytologist 213 : 1076 – 1092 . Google Scholar CrossRef Search ADS Skelton RP , West AG , Dawson TE . 2015 . Predicting plant vulnerability to drought in biodiverse regions using functional traits . Proceedings of the National Academy of Sciences of the United States of America 112 : 5744 – 5749 . Google Scholar CrossRef Search ADS PubMed Skelton RP , Brodribb TJ , Choat B . 2017a . Casting light on xylem vulnerability in an herbaceous species reveals a lack of segmentation . New Phytologist 214 : 561 – 569 . Google Scholar CrossRef Search ADS Skelton RP , Brodribb TJ , McAdam SAM , Mitchell PJ . 2017b . Gas exchange recovery following natural drought is rapid unless limited by loss of leaf hydraulic conductance: evidence from an evergreen woodland . New Phytologist 215 : 1399 – 1412 . Google Scholar CrossRef Search ADS R Core Team . 2015 . R: a language and environment for statistical computing . Vienna : R Foundation for Statistical Computing . Tozer WC , Rice B , Westoby M . 2015 . Evolutionary divergence of leaf width and its correlates . American Journal of Botany 102 : 367 – 378 . Google Scholar CrossRef Search ADS PubMed Trifilo P , Raimondo F , Savi T , Lo Gullo MA , Nardini A . 2016 . The contribution of vascular and extra-vascular water pathways to drought-induced decline of leaf hydraulic conductance . Journal of Experimental Botany 67 : 5029 – 5039 . Google Scholar CrossRef Search ADS PubMed Tyree MT , Zimmermann MH . 2002 . Xylem dysfunction: when cohesion breaks down . In Xylem Structure and the Ascent of Sap . Berlin : Springer , 89 – 141 . Google Scholar CrossRef Search ADS Tyree MT , Davis SD , Cochard H . 1994 . Biophysical perspectives of xylem evolution - is there a tradeoff of hydraulic efficiency for vulnerability to dysfunction . IAWA Journal 15 : 335 – 360 . Google Scholar CrossRef Search ADS Walls RL . 2011 . Angiosperm leaf vein patterns are linked to leaf function in a global-scale data set . American Journal of Botany 98 : 244 – 253 . Google Scholar CrossRef Search ADS PubMed Warton DI , Wright IJ , Falster DS , Westoby M . 2006 . Bivariate line-fitting methods for allometry . Biological Reviews 81 : 259 – 291 . Google Scholar CrossRef Search ADS PubMed Wright IJ , Dong N , Maire V , et al. 2017 . Global climatic drivers of leaf size . Science 357 : 917 – 921 . Google Scholar CrossRef Search ADS PubMed Zhang YJ , Rockwell FE , Graham AC , Alexander T , Holbrook NM . 2016 . Reversible leaf xylem collapse: a potential ‘circuit breaker’ against cavitation . Plant Physiology 172 : 2261 – 2274 . Google Scholar CrossRef Search ADS PubMed © The Author(s) 2018. Published by Oxford University Press on behalf of the Annals of Botany Company. All rights reserved. For permissions, please e-mail: [email protected]. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)
Gain-of-function mutation of AtDICE1, encoding a putative endoplasmic reticulum-localized membrane protein, causes defects in anisotropic cell elongation by disturbing cell wall integrity in Arabidopsis2018 Annals of Botany
doi: 10.1093/aob/mcy049pmid: 29659701
Abstract Background and Aims Anisotropic cell elongation depends on cell wall relaxation and cellulose microfibril arrangement. The aim of this study was to characterize the molecular function of AtDICE1 encoding a novel transmembrane protein involved in anisotropic cell elongation in Arabidopsis. Methods Phenotypic characterizations of transgenic Arabidopsis plants mis-regulating AtDICE1 expression with different pharmacological treatments were made, and biochemical, cell biological and transcriptome analyses were performed. Key Results Upregulation of AtDICE1 in Arabidopsis (35S::AtDICE1) resulted in severe dwarfism, probably caused by defects in anisotropic cell elongation. Epidermal cell swelling was evident in all tissues, and abnormal secondary wall thickenings were observed in pith cells of stems. These phenotypes were reproduced not only by inducible expression of AtDICE1 but also by overexpression of its poplar homologue in Arabidopsis. RNA interference suppression lines of AtDICE1 resulted in no observable phenotypic changes. Interestingly, wild-type plants treated with isoxaben, a cellulose biosynthesis inhibitor, phenocopied the 35S::AtDICE1 plants, suggesting that cellulose biosynthesis was compromised in the 35S::AtDICE1 plants. Indeed, disturbed cortical microtubule arrangements in 35S::AtDICE1/GFP-TuA6 plants were observed, and the cellulose content was significantly reduced in 35S::AtDICE1 plants. A promoter::GUS analysis showed that AtDICE1 is mainly expressed in vascular tissue, and transient expression of GFP:AtDICE1 in tobacco suggests that AtDICE1 is probably localized in the endoplasmic reticulum (ER). In addition, the external N-terminal conserved domain of AtDICE1 was found to be necessary for AtDICE1 function. Whole transcriptome analyses of 35S::AtDICE1 revealed that many genes involved in cell wall modification and stress/defence responses were mis-regulated. Conclusions AtDICE1, a novel ER-localized transmembrane protein, may contribute to anisotropic cell elongation in the formation of vascular tissue by affecting cellulose biosynthesis. Arabidopsis, anisotropic cell elongation, cellulose biosynthesis, cell wall integrity, endoplasmic reticulum, isoxaben, poplar, secondary wall thickening, vascular tissue formation INTRODUCTION Cell elongation is a fundamental process for optimal growth and development of plants by directing proper cell morphology. Cell elongation is initiated by vacuolar turgor pressure and limited by a strong but flexible cell wall that permits directional cell elongation. Plant cell walls are essential not only for protecting cells against external stresses but also the cell elongation process (Baskin, 2005). The cell wall is composed mainly of polysaccharides (e.g. cellulose, hemicelluloses and pectins) interwoven among a relatively small amount of highly glycosylated protein (Somerville et al., 2004; Carpita, 2011). The strongest component in the cell wall is a network of cellulose microfibrils. While cellulose microfibrils are synthesized by cellulose synthase (CesA) complexes (CSCs) at the plasma membrane (PM) and are extruded directly into the cell wall (Geisler et al., 2008; McFarlane et al., 2014), other polysaccharides are synthesized in the endoplasmic reticulum (ER) or Golgi and delivered to the cell wall through the intracellular secretory pathway (Gibeaut and Carpita, 1994). CesA subunits are synthesized in the ER, and CSC rosettes are assembled in the Golgi, transported via the trans-Golgi network (TGN) and inserted into the PM through cortical microtubule-assisted vesicle trafficking (Paredez et al., 2006; McFarlane et al., 2014; Zhang et al., 2016). Cellulose microfibrils, as load-bearing components of the cell wall, are important in controlling anisotropic cell elongation in plants. Cellulose microfibrils are aligned perpendicular to the axis of elongation, thus forming a spring-like structure that reinforces the cell crosswise and favours longitudinal expansion in most growing cells (Baskin, 2005; Lei et al., 2014). Furthermore, because the newly synthesized cellulose microfibrils are aligned by the cortical microtubules through microtubule–CSC interacting proteins, such as CSI1/POM2, CSI3 and CC, it has been suggested that the shape of plant cells is determined by the orientation of cortical microtubules (Green, 1962; Bringmann et al., 2012; Endler et al., 2015). Thus, perturbation of either microtubules or cellulose microfibril organization leads to defects in cell elongation (Fagard et al., 2000; Lane et al., 2001; Schindelman et al., 2001; Pagant et al., 2002; Caño-Delgado et al., 2003; Somerville, 2006; Wang et al., 2006). The dwarf phenotype of cellulose-deficient mutants is therefore commonly interpreted as resulting from the loss of anisotropic cell elongation by weakened cell walls. Many genes involved in either cell expansion or cellulose biosynthesis have been identified through genetic screening and subsequent cloning. For example, mutation of RSW1/CesA1, RSW5/CesA3, QUILL/PROCUSTE/CesA6, POM-POM1/CHITINASE LIKE1 (POM1/CTL1), KORRIGAN/LION’S TAIL1/RADIALLY SWOLLEN2 (KOR/LIT/RSW2), COBRA and KOBITO/ELONGATION DEFECTIVE1 (KOB1/ELD1) was associated with epidermal cell swelling, restricted elongation of root and hypocotyl, and reduced cell wall cellulose content (Hauser et al., 1995; Arioli et al., 1998; Nicol et al., 1998; Fagard et al., 2000; Lane et al., 2001; Schindelman et al., 2001; Pagant et al., 2002; Wang et al., 2006). During cell expansion, the integrity of the cell wall is vital. Once the integrity of the cell wall is disrupted, a series of compensatory reactions, such as ectopic accumulation of lignin and callose, modification of cell wall composition, and enhanced stress/defence responses, follow. For example, mutations or treatment with chemical inhibitors of cellulose biosynthesis lead to ectopic lignification and/or callose deposition, and change the composition of the cell wall matrix (Desprez et al., 2002; Pagant et al., 2002). In both eli1 (ECTOPIC LIGNIFICATION1) (Caño-Delgado et al., 2000) and cev1 (CONSTITUTIVE EXPRESSION OF VSP1) (Ellis and Turner, 2001) mutants, which were identified as different mutant alleles of a cellulose synthase gene CesA3, the jasmonic acid (JA) and ethylene-signalling pathways were highly activated and stress responsive genes were up-regulated. Also, cob-5, a loss of function mutant of COBRA, exhibits abnormal cell growth with a massive accumulation of stress response chemicals, such as anthocyanins and callose. Furthermore, cob-5 over-accumulates JA and induces defence and stress-related genes coordinately (Ko et al., 2006b). These reports suggest that plant cells initially sense biotic stresses at the level of the integrity of the cell wall (Ko et al., 2006b; Wolf et al., 2012). In this study, we report the functional characterization of a gene named AtDICE1 (DEFECT IN CELL ELONGATION1) from Arabidopsis. A gain-of-function mutation of AtDICE1 caused severe dwarfism with defects in anisotropic cell elongation. Further anatomical, molecular, biochemical and whole transcriptome analyses suggest that AtDICE1 is a novel ER-localized transmembrane protein contributing to the proper anisotropic cell elongation process in the vascular tissue through participation in cell wall formation. MATERIALS AND METHODS Plant materials and growth conditions Arabidopsis thaliana, ecotype Columbia (Col-0), was used in both the wild-type (WT) and transgenic experiments. Plants were growth on soil or on MS-agar plates [0.5× MS, 2 % sucrose, 0.8 % (w/v) agar] in a growth chamber (16-h light/8-h dark) at 23 ± 2 °C, after stratification. For the growth measurement of young seedlings, MS-agar plates were vertically orientated. For isoxaben (36138, Sigma-Aldrich, St Louis, MO, USA) treatment, isoxaben was dissolved in methanol and added to the A. thaliana seedling cultures to final concentrations of 1, 5 or 10 nm. Seedlings were grown in 0.5× MS medium containing 2 % (w/v) sucrose, 0.8 % (w/v) agar and isoxaben at different concentrations and observed after 7 d. Histological analysis The stem area located immediately above the rosette (basal level) was cross-sectioned by hand and stained with 2 % phloroglucinol/HCl or 0.05 % toluidine blue O for 1 min. Microtome (Leica RM2025, Leica, http://www.leica.com) sectioning after paraffin embedding was used to observe the detailed structure of hypocotyls and stems. For confocal laser scanning microscopy, a Zeiss PASCAL microscope (Jena, Germany), with a 488-nm excitation mirror, a 560-nm emission filter and a 505–530-nm emission filter, was used to record images. Image analysis was performed using laser scanning microscope PASCAL LSM version 3.0 SP3 software. For scanning electron microscopy (SEM), a tabletop microscope (Hitachi TM3000, Tokyo, Japan) was used to visualize hypocotyls from 7-d-old seedlings of WT and 35S::AtDICE1 (7-5 line) plants using the default conditions without any pretreatments. Plasmid vector construction and generation of transgenic Arabidopsis plants The full-length cDNA of AtDICE1 (At2g41610) was amplified by PCR and inserted downstream of the 35S promoter in the pB2GW7 or pB7BWIWG2(II) vectors (Karimi et al., 2002) using the Gateway cloning system to produce 35S::AtDICE1 or 35S::AtDICE1-RNAi constructs, respectively. The N-terminal deletion fragment of AtDICE1 (ΔNT-AtDICE1) was amplified by PCR using the 35S::AtDICE1 construct as a template and inserted downstream of the 35S promoter in the pB2GW7 vector to produce the 35S::ΔNT-AtDICE1 construct. For the AtDICE1 promoter::GUS construct, the 1.0-kb XbaI/BamHI genomic fragment flanking the 5′ end of the AtDICE1 coding sequence was amplified using genomic DNA extracted from Arabidopsis leaf tissue as template, and subcloned into the pCB308 vector (Xiang et al., 1999). To construct the 35S::GFP:AtDICE1 plasmid, the Gateway destination vector pEarleyGate103 was used to translationally fused GFP:AtDICE1. All the vector constructs produced were verified by DNA sequencing and introduced into Agrobacterium tumefaciens (C58), which was used to transform Arabidopsis (Col-0) or the GFP-TuA6 background using the floral-dip method described by Clough and Bent (1998). Extraction of alcohol-insoluble cell wall residue Stem tissues (5 cm from rosette level) from 60-d-old WT and 35S::AtDICE1 plants were ground to a fine powder. The ground material (approx. 1 g) was washed in 15 mL of 70 % ethanol and heated for 15 min at 70 °C to inactivate endogenous enzymes and remove the cell contents. Samples were centrifuged for 10 min at 4000 g, and the pellets were washed twice with 100 % ethanol and once with 100 % acetone. The remaining pellet was considered to be the alcohol-insoluble cell wall residue (AIR) and was dried at 70 °C. Analysis of cell wall monosaccharide composition Cell wall sugars (as alditol acetates) were determined using the procedure reported by Hoebler et al. (1989). Briefly, AIRs (3 mg) were incubated with 70 % sulfuric acid at room temperature for 30 min, followed by the addition of inositol as the internal standard and dilution with water to a final concentration of 6 % sulfuric acid. After heating for 120 min at 105 °C, the solution was treated with 25 % ammonium solution. After reduction with sodium borohydride in dimethyl sulfoxide, the solution was heated for 90 min at 40 °C, followed by sequential treatment with glacial acetic acid, acetic anhydride, 1-methylimidazole, dichloromethane and water. The organic layer containing the alditol acetates of the hydrolysed cell wall sugars was washed three times with water, and sugars were analysed on a gas–liquid chromatograph (model 6890; Hewlett-Packard, http://www.hp.com) equipped with a 30 m × 0.25 mm (i.d.) silica capillary column DB 225 (Alltech Associates Inc., Deerfield, IL, USA). Statistical analysis Data were analysed with a two-tailed, unpaired t-test by using GRAPHPAD PRISM 7.03 (Graphpad Software, San Diego, CA, USA). P-values of <0.05 were considered significant. Histochemical GUS staining Histochemical GUS staining of transgenic plants was performed as described by Jefferson et al. (1987) with slight modifications as described by Nguyen et al. (2016). Briefly, samples were incubated at 37 °C for 18–24 h in GUS reaction buffer and visualized after removing chlorophyll by rinsing with 70 % ethanol. Intracellular localization of AtDICE1 Each 35S::GFP and 35S::GFP:AtDICE1 construct was introduced into Agrobacterium tumefaciens (strain C58) by the freeze–thaw method (Höfgen and Willmitzer, 1988). Transient expression in Nicotiana benthamiana leaf epidermal cells was determined by the agroinfiltration method (Di Sansebastiano et al., 2004). The N. benthamiana leaves were further grown for 12 h, and the fluorescence was observed using a Zeiss PASCAL confocal laser scanning microscope as described above. RNA extraction and semi-quantitative RT-PCR analysis Total RNAs were extracted using Trizol reagent (Gibco-BRL, http://www.invitrogen.com) according to the manufacturer’s instructions. Total RNAs were reverse transcribed using Superscript II reverse transcriptase (Invitrogen) in 20-µL reactions. Semi-quantitative reverse transcriptase PCR (RT-PCR) was carried out using 1 µΛ of the reaction products as a template. Amplified DNA fragments were separated on a 1 % agarose gel and stained with ethidium bromide. The primers used for RT-PCR analysis are shown in Supplementary Data Table S3. Whole transcriptome microarray analysis Total RNAs were isolated from 7-d-old seedlings of 35S::AtDICE1 (7-5, 10-1, 11-6) and WT grown on MS agar plates. An Agilent Bioanalyzer (Agilent, Santa Clara, CA, USA) was used to check RNA quality. Nucleic acid labelling was performed according to the manufacturer’s recommended procedures for single-colour arrays. Labelled RNA was hybridized to Agilent 4x44K Arabidopsis Gene Expression (V4) Microarrays (Agilent G2519F-021169). Arrays were scanned with an Agilent G2565B Array Scanner, and images were analysed using the Feature Extractor v9.5 default protocol GE1-v5_95_Feb07. The resulting hybridization intensity values (i.e. signal intensity) of each spot reflect the abundance of a given mRNA relative to the total mRNA population and were used in all subsequent analyses. All data normalization and selection of genes showing fold-changes were performed using GeneSpringGX 7.3.1 (Agilent). Genes were filtered by removing flag-out genes in each experiment. Intensity-dependent normalization [locally weighted scatterplot smoothing (LOWESS)] was performed, where the ratio was reduced to the residual of the LOWESS fit of the intensity versus ratio curve. The averages of normalized ratios were calculated by dividing the average of the normalized signal channel intensity by the average of normalized control channel intensity. Differentially expressed genes were identified using the Single t-test, which assumes unequal variances between groups. The criterion for significant genes was P < 0.05. The raw microarray data were deposited in the Gene Expression Omnibus database together with details of the protocol. RESULTS Overexpression of AtDICE1 results in a severe dwarf phenotype Previously, using a global comparative transcriptome analysis, we found a gene network regulating secondary xylem development in Arabidopsis (Ko et al., 2006a). Among the genes in this network, we were interested in a completely unknown gene locus, At2g41610, which we named AtDICE1. AtDICE1 is specifically expressed in vascular tissues or cells (Supplementary Data Fig. S1) and the Potri.014G108300 gene (named as PtrDICE1), the closest homologue of AtDICE1 in poplar, is also preferentially expressed in the wood-forming tissues of poplar (Fig. S1C). Consistently, our promoter–GUS reporter analysis showed that AtDICE1 promoter activity was detected mainly in the vascular tissues of cotyledon and hypocotyl, and in the roots of young seedlings as early as 2 d old (Fig. S1D). In cross-sections of stems and roots, activity of the AtDICE1 promoter was found mainly in vascular bundles, especially in cambium and/or xylem parenchyma cells (Fig. S1E). Further sequence analysis identified that AtDICE1 is a plant-specific and transmembrane protein (Fig. S2). Both TMHMM (http://www.cbs.dtu.dk/services/TMHMM-2.0/) and SOSUI (http://bp.nuap.nagoya-u.ac.jp/sosui/) analyses predicted that AtDICE1 has seven transmembrane domains with the N-terminus outside and C-terminus inside (Fig. S2B, C). Interestingly, the putative external N-terminal motif is highly conserved in all plant species examined (Fig. S2A). To characterize the molecular function of AtDICE1, we first generated gain-of-function transgenic Arabidopsis plants by overexpressing AtDICE1 under control of the cauliflower mosaic virus (CaMV) 35S promoter (i.e. 35S::AtDICE1 plants). This was done because T-DNA knock-out mutants were unfortunately not available. A total of 26 T1 transgenic plants were examined and we found that most of them were dwarfed compared to WT Arabidopsis plants (data not shown). To perform additional detailed analyses, three T3 homozygote transgenic plants (i.e. 7-5, 10-1 and 11-6 lines of 35S::AtDICE1) based on phenotypic severity were selected (Fig. 1). The dwarfing phenotypes of 35S::AtDICE1 plants were evident not only in young seedlings but also at the adult stage (Fig. 1; Table S1). Furthermore, the degree of dwarfing was consistent with the expression levels of the AtDICE1 gene (Fig. 1B). In adult plants, 35S::AtDICE1 plants showed extremely small rosette leaves, very short inflorescence stems, small flowers and reduced apical dominances (Figs 1C, D and S3). Furthermore, overexpression of PtrDICE1 reproduced almost the identical phenotype observed in 35S::AtDICE1 plants (Fig. 1E). This result suggests that PtrDICE1 is orthologous to AtDICE1. Fig. 1. View largeDownload slide Gain-of-function mutation of AtDICE1 resulted in a severe dwarfing phenotype. (A) Seven-day-old seedlings of wild-type (WT) and 35S::AtDICE1 plants. (B) Semi-quantitative RT-PCR analysis for AtDICE1 transcript accumulation in each of three T3 homozygote 35S::AtDICE1 plants. Actin8 was used as a control. (C) Growth features of 35S::AtDICE1 (line 7-5) and WT plants on soil at 10 d (upper) and 33 d (lower). (D) Forty-day-old WT and 35S::AtDICE1 (7-5) plants grown in pots. (E) Forty-day-old 35S::AtDICE1 (7-5) (left) and 35S::PtrDICE1 (right) plants. Scale bars represent 1 cm, except 10 cm in D. Fig. 1. View largeDownload slide Gain-of-function mutation of AtDICE1 resulted in a severe dwarfing phenotype. (A) Seven-day-old seedlings of wild-type (WT) and 35S::AtDICE1 plants. (B) Semi-quantitative RT-PCR analysis for AtDICE1 transcript accumulation in each of three T3 homozygote 35S::AtDICE1 plants. Actin8 was used as a control. (C) Growth features of 35S::AtDICE1 (line 7-5) and WT plants on soil at 10 d (upper) and 33 d (lower). (D) Forty-day-old WT and 35S::AtDICE1 (7-5) plants grown in pots. (E) Forty-day-old 35S::AtDICE1 (7-5) (left) and 35S::PtrDICE1 (right) plants. Scale bars represent 1 cm, except 10 cm in D. We produced RNA interference (RNAi) suppression lines of AtDICE1 (i.e. 35S::AtDICE1-RNAi) to analyse the loss-of-function phenotype (Fig. S4). Among a total of seven independent T3 homozygous lines, two of them (7-1, 12-1) showed significantly decreased expression of AtDICE1 (Fig. S4A). However, we were unable to observe any significant alterations in phenotype in seedlings or adult plants (Fig. S4B, C). 35S::AtDICE1 plants exhibit defects in anisotropic cell elongation In young seedlings of 35S::AtDICE1 plants, epidermal cell swelling is evident in both hypocotyls and cotyledons; this was never seen in WT seedlings (Fig. 2). Analysis of a cross-section of hypocotyls from 7-d-old seedlings revealed that the cell swelling was found not only in epidermal cells but also in cells inside, which caused disarray of the cell arrangement in 35S::AtDICE1 plants (Fig. 2A). SEM clearly showed epidermal cell enlargements, which were distorted during the imaging process due to vacuum application (Fig. 2B). This isotropic cell expansion was found persistently in adult plants (Fig. 2C), which probably caused the severe dwarfism of 35S::AtDICE1 plants. To verify this phenotype, we produced additional transgenic Arabidopsis plants in which AtDICE1 expression can be induced. For inducible gene expression, the Gateway Destination vector pMDC7 (Curtis and Grossniklaus, 2003) was used. This vector contains the XVE inducible promoter for 17-β-estradiol inducible expression in plants (Fig. 3). Our results clearly demonstrated that treatment with 17-β-estradiol at concentrations as low as 10 αm successfully induced AtDICE1 expression (Fig. 3A) and resulted in epidermal cell swelling as early as in 2-d-old seedlings (Fig. 3B). Fig. 2. View largeDownload slide 35S::AtDICE1 plants showed defects of anisotropic cell expansion. (A) Epidermal cell swelling in cotyledons (upper), hypocotyls (middle) and hypocotyl cross-sections (lower) of 7-d-old 35S::AtDICE1 seedlings compared to WT. Arrowheads indicate swollen cells. Scale bars represent 1 mm (top), 100 µm (middle and bottom). (B) Scanning electron micrographs of hypocotyls of 7-d-old plants. (C) Petioles of rosette leaves from 30-d-old plants. Arrowheads indicate swollen cells. Scale bars represent 0.5 cm. Fig. 2. View largeDownload slide 35S::AtDICE1 plants showed defects of anisotropic cell expansion. (A) Epidermal cell swelling in cotyledons (upper), hypocotyls (middle) and hypocotyl cross-sections (lower) of 7-d-old 35S::AtDICE1 seedlings compared to WT. Arrowheads indicate swollen cells. Scale bars represent 1 mm (top), 100 µm (middle and bottom). (B) Scanning electron micrographs of hypocotyls of 7-d-old plants. (C) Petioles of rosette leaves from 30-d-old plants. Arrowheads indicate swollen cells. Scale bars represent 0.5 cm. Fig. 3. View largeDownload slide Inducible expression of AtDICE1 reproduced the cell swelling phenotype of 35S::AtDICE1 plants. (A) Semi-quantitative RT-PCR analysis performed to confirm AtDICE1 induction by 17-β-estradiol treatment of 10-d-old seedlings of T3 homozygous AtDICE1/pMDC7 plants (line 8-4). (B) Estradiol treatment of AtDICE1/pMDC7 plants (line 8-4). AtDICE1/pMDC7 plants were grown on MS-agar plates with (+Estradiol) or without 17-β-estradiol (+DMSO). Red arrowheads indicate epidermal cell swelling. Scale bars represent 1 mm. Fig. 3. View largeDownload slide Inducible expression of AtDICE1 reproduced the cell swelling phenotype of 35S::AtDICE1 plants. (A) Semi-quantitative RT-PCR analysis performed to confirm AtDICE1 induction by 17-β-estradiol treatment of 10-d-old seedlings of T3 homozygous AtDICE1/pMDC7 plants (line 8-4). (B) Estradiol treatment of AtDICE1/pMDC7 plants (line 8-4). AtDICE1/pMDC7 plants were grown on MS-agar plates with (+Estradiol) or without 17-β-estradiol (+DMSO). Red arrowheads indicate epidermal cell swelling. Scale bars represent 1 mm. Cellulose biosynthesis is impaired in 35S::AtDICE1 plants Perturbations in either cellulose biosynthesis or cortical microtubule arrangements are known to cause cell swelling (Fagard et al., 2000; Lane et al., 2001; Schindelman et al., 2001; Pagant et al., 2002; Caño-Delgado et al., 2003; Somerville, 2006; Wang et al., 2006). To test our hypothesis, isoxaben, a well-known inhibitor of cellulose biosynthesis, was applied to both WT and 35S::AtDICE1 plants. Treatment of WT plants with isoxaben successfully mimicked the epidermal cell swelling phenotype of 35S::AtDICE1 plants at concentrations as low as 5 nm (Fig. 4A). Because isoxaben treatment is known to prevent the insertion of CSCs into the PM (Paredez et al., 2006; Gutierrez et al., 2009), the cell swelling phenotype was probably caused by defects in cellulose biosynthesis in the cell wall. Fig. 4. View largeDownload slide 35S::AtDICE1 resulted in defects in cellulose biosynthesis. (A) Treatment with isoxaben in WT plants. Plants were grown for 7 d on MS-agar plates containing isoxaben as indicated. Scale bars represent 0.5 mm. (B) Cell wall monosaccharide composition analysis. Inflorescence stems of 60-d-old soil-grown plants were used. AIR, alcohol-insoluble residue. Neutral monosaccharides in the sulfuric acid-soluble fraction were determined by quantification of their corresponding alditol acetates by gas chromatography. Rha, rhamnose; Fuc, fucose; Ara, arabinose; Xyl, xylose; Man, mannose; Gal, galactose; Glu, glucose. Error bars indicate s.e. using triplicate experiments (*P < 0.05; **P < 0.01, unpaired t-test). (C) Observation of cortical microtubule arrangement by confocal laser microscopy. Cortical microtubules were visualized in normal cells from GFP-TuA6 plants (upper) and swollen cells from 35S::AtDICE1/GFP-TuA6 plants (lower). Scale bars indicate 25 αµ. Fig. 4. View largeDownload slide 35S::AtDICE1 resulted in defects in cellulose biosynthesis. (A) Treatment with isoxaben in WT plants. Plants were grown for 7 d on MS-agar plates containing isoxaben as indicated. Scale bars represent 0.5 mm. (B) Cell wall monosaccharide composition analysis. Inflorescence stems of 60-d-old soil-grown plants were used. AIR, alcohol-insoluble residue. Neutral monosaccharides in the sulfuric acid-soluble fraction were determined by quantification of their corresponding alditol acetates by gas chromatography. Rha, rhamnose; Fuc, fucose; Ara, arabinose; Xyl, xylose; Man, mannose; Gal, galactose; Glu, glucose. Error bars indicate s.e. using triplicate experiments (*P < 0.05; **P < 0.01, unpaired t-test). (C) Observation of cortical microtubule arrangement by confocal laser microscopy. Cortical microtubules were visualized in normal cells from GFP-TuA6 plants (upper) and swollen cells from 35S::AtDICE1/GFP-TuA6 plants (lower). Scale bars indicate 25 αµ. To further confirm whether cellulose biosynthesis was affected in 35S::AtDICE1 plants, we analysed the monosaccharide composition of cell wall materials prepared from stem tissues of WT and 35S::AtDICE1 plants. To our surprise, the glucose content was significantly decreased in 35S::AtDICE1 plants (Fig. 4B). Because the AIRs of the cell wall were used in this analysis, most of the glucose content came from the degradation of cellulose. Thus, this result suggests that the cellulose content of the cell walls of 35S::AtDICE1 plants was reduced. Next, to determine whether there were any changes in the cortical microtubule arrangement of 35S::AtDICE1 plants, we overexpressed AtDICE1 in GFP-TuA6 plants. These plants constitutively express a green fluorescent protein (GFP)-fused tubulin A6, and thus cortical microtubules can be visualized using confocal microscopy (Ueda and Matsuyama, 2000). Interestingly, we were able to observe abnormal arrangements of cortical microtubules in swollen cells from 35S::AtDICE1/GFP-TuA6 plants compared to GFP-TuA6 plants, which showed arrangements of microtubules perpendicular to the elongation direction (Fig. 4C). AtDICE1 protein is probably localized in the ER Our amino acid sequence analysis of AtDICE1 suggested that AtDICE1 is a putative transmembrane protein having seven transmembrane domains (Fig. S2). To investigate the intracellular localization of AtDICE1, a transient expression analysis was performed using tobacco epidermal cells. The vector construct expressing the AtDICE1 protein fused to the C-terminus of GFP driven by the CaMV 35S promoter (35S::GFP:AtDICE1) was introduced into the leaves of tobacco using Agrobacterium and the epidermal cells were observed under a confocal laser microscope. As shown in Fig. 5, the fluorescent signal of GFP:AtDICE1 was observed exclusively in the reticulate tubular networks and in sheets typical of the ER in the cortical regions of cells, indicating that AtDICE1 is probably localized in the ER. The same result was obtained by using the 35S:: AtDICE1:GFP construct (data not shown). By contrast, in a control experiment, the fluorescent signal of GFP alone was found in the plasma membrane, cytoplasm and nucleus of the cells (Fig. 5). Fig. 5. View largeDownload slide Intracellular localization of the AtDICE1 protein. Intracellular localization of either GFP alone (left) or GFP:AtDICE1 (right) driven by the 35S promoter was observed by confocal imaging of transiently transformed tobacco epidermal leaf cells. Single confocal optical sections visualizing images of GFP fluorescence (500–520 nm) were photographed using dark field (upper) and bright field (middle) and then merged (lower). Scale bars represent 50 αµ. Fig. 5. View largeDownload slide Intracellular localization of the AtDICE1 protein. Intracellular localization of either GFP alone (left) or GFP:AtDICE1 (right) driven by the 35S promoter was observed by confocal imaging of transiently transformed tobacco epidermal leaf cells. Single confocal optical sections visualizing images of GFP fluorescence (500–520 nm) were photographed using dark field (upper) and bright field (middle) and then merged (lower). Scale bars represent 50 αµ. The N-terminal conserved domain is necessary for AtDICE1 function Because AtDICE1 has a putative external N-terminal motif, which is highly conserved in the plant species examined (Fig. S2A), we attempted to determine the functional significance of the motif using deletion analysis. We generated transgenic Arabidopsis plants overexpressing an N-terminal deleted AtDICE1 gene (35S::ΔNT-AtDICE1) (Fig. 6A, B). However, we did not observe any phenotypic changes in either the young seedlings or adult plants of the 19 independent T3 homozygous transgenic plants we examined (Fig. 6C, D). This suggests that the highly conserved N-terminal external domain of AtDICE1 might play an important role in AtDICE1 function. Fig. 6. View largeDownload slide The N-terminal conserved domain is necessary for AtDICE1 function. (A) Schematic diagram of full length (left) and N-terminal conserved domain-deleted (right; ΔNT-AtDICE1) AtDICE1. The N-terminal conserved domain is indicated by the red circle. (B) Semi-quantitative RT-PCR analysis to confirm the expression of ΔNT-AtDICE1 in 35S::ΔNT-AtDICE1 plants. (C) Seven-day-old 35S::ΔNT-AtDICE1 seedlings compared to WT. (D) Phenotypic consequences in a 33-d-old 35S::ΔNT-AtDICE1 plant compared to both WT and 35S::AtDICE1 plants. Scale bars indicate 1 cm. Fig. 6. View largeDownload slide The N-terminal conserved domain is necessary for AtDICE1 function. (A) Schematic diagram of full length (left) and N-terminal conserved domain-deleted (right; ΔNT-AtDICE1) AtDICE1. The N-terminal conserved domain is indicated by the red circle. (B) Semi-quantitative RT-PCR analysis to confirm the expression of ΔNT-AtDICE1 in 35S::ΔNT-AtDICE1 plants. (C) Seven-day-old 35S::ΔNT-AtDICE1 seedlings compared to WT. (D) Phenotypic consequences in a 33-d-old 35S::ΔNT-AtDICE1 plant compared to both WT and 35S::AtDICE1 plants. Scale bars indicate 1 cm. Transcription phenotype of 35S::AtDICE1 plants is highly correlated with its physiological phenotype To examine the transcription phenotype of 35S::AtDICE1 plants, a whole-transcriptome analysis was performed using total RNAs extracted from 7-d-old seedlings of three independent T3 homozygous lines of 35S::AtDICE1 (i.e. 7-5, 10-1, 11-6) and WT plants grown on MS agar plates (see Materials and Methods). Gene expression profiles were obtained using Agilent’s Whole Arabidopsis Gene Expression Microarray (G2519F-021169, V4, 4x44K), which contains a total of 43 505 probe sets covering the entire Arabidopsis transcriptome (http://www.genomics.agilent.com). In the subsequent gene expression analysis, we found a total of 90 probe sets (representing 72 genes) which were significantly upregulated (three-fold as a threshold) in all three lines of 35S::AtDICE1 plants compared to WT plants. Meanwhile only 17 probe sets (representing 16 genes) were downregulated (Fig. 7A, B; Table S2). Fig. 7. View largeDownload slide Whole transcriptome analyses of 7-d-old 35S::AtDICE1 plants compared to WT. (A and B) Venn diagrams showing differentially expressed probe sets from three independent lines (i.e. 7-5, 10-1, 11-6) of 35S::AtDICE1 plants compared to WT. The number of probe sets in each line is displayed within a circle. The number of common probe sets is shown in the intersections between lines. Upregulated (A) or downregulated (B) probe sets in each line compared to WT using a three3-fold change as the threshold. (C) Gene Ontology (GO) analysis performed using REVIGO (http://revigo.irb.hr/). The representative GO terms of the 90 upregulated probe sets of A are visualized in semantic similarity-based scatterplots. (D) Functional classification generated using the Classification SuperViewer (http://bar.utoronto.ca/ntools/cgi-bin/ntools_classification_superviewer.cgi). The Classification SuperViewer generates an overview of functional classification of the 90 upregulated probe sets of A based on the GO database. The input set was bootstrapped 100 times to provide an idea as to over- or underrepresentation reliability. Gene categories are shown on the left, and their normalized frequencies are found on the x-axis and are calculated as follows: (Number_in_Classinput_set/Number_Classifiedinput_set)/(Number_in_Classreference_set (25k)/Number_Classifiedreference_set). Fig. 7. View largeDownload slide Whole transcriptome analyses of 7-d-old 35S::AtDICE1 plants compared to WT. (A and B) Venn diagrams showing differentially expressed probe sets from three independent lines (i.e. 7-5, 10-1, 11-6) of 35S::AtDICE1 plants compared to WT. The number of probe sets in each line is displayed within a circle. The number of common probe sets is shown in the intersections between lines. Upregulated (A) or downregulated (B) probe sets in each line compared to WT using a three3-fold change as the threshold. (C) Gene Ontology (GO) analysis performed using REVIGO (http://revigo.irb.hr/). The representative GO terms of the 90 upregulated probe sets of A are visualized in semantic similarity-based scatterplots. (D) Functional classification generated using the Classification SuperViewer (http://bar.utoronto.ca/ntools/cgi-bin/ntools_classification_superviewer.cgi). The Classification SuperViewer generates an overview of functional classification of the 90 upregulated probe sets of A based on the GO database. The input set was bootstrapped 100 times to provide an idea as to over- or underrepresentation reliability. Gene categories are shown on the left, and their normalized frequencies are found on the x-axis and are calculated as follows: (Number_in_Classinput_set/Number_Classifiedinput_set)/(Number_in_Classreference_set (25k)/Number_Classifiedreference_set). To gain functional insight into the 72 upregulated genes in 35S::AtDICE1, we performed Gene Ontology (GO) analyses (Fig. 7C, D). In the REVIGO (http://revigo.irb.hr) analysis, which provides a visual summary of the long list of GO terms as a scatterplot by removing redundant GO terms, the responses to stress, response to organic substance (e.g. JA) and defence response categories plotted most clearly (Fig. 7C). In addition, using the Classification SuperViewer tool (http://bar.utoronto.ca/ntools/cgi-bin/ntools_classification_superviewer.cgi), genes related to the abiotic/biotic stimulus, response to stress and cell wall categories were found to be over-represented (Fig. 7D). Accordingly, more than half of the 72 upregulated genes (37 genes, 51.4 %) were involved in the stress and/or defence mechanisms (Table 1). In addition, many cell wall formation and modification-related genes were significantly upregulated, which probably reflects the defects in the cell elongation phenotype of 35S::AtDICE1 plants (Fig. 7D and Table 1). These results clearly show that the transcription phenotype of 35S::AtDICE1 plants is highly correlated with its physiological phenotype. Table 1. List of genes upregulated in 35S::AtDICE1 plants AGI* WT F† 7-5 F 10-1 F 11-6 F 7-5/WT‡ 10-1/WT 11-6/WT cob-5 /WT§ Gene description St/Def¶ AT1G72260 6.7 A 7978.0 P 2015.0 P 2907.6 P 1185.5 299.3 431.9 106.4 thionin 2.1 YES AT4G15100 15.9 A 1239.8 P 156.3 P 88.0 P 78.2 9.9 5.5 12.5 serine carboxypeptidase- like 30 AT2G39330 267.0 P 8145.8 P 4897.3 P 4446.0 P 30.5 18.3 16.7 43.1 jacalin-related lectin 23 YES AT2G41610 408.1 P 8791.5 P 6760.8 P 3807.9 P 21.5 16.6 9.3 0.7 AtDCE1 AT1G54020 166.3 P 3426.4 P 1788.4 P 1949.7 P 20.6 10.8 11.7 53.9 GDSL esterase/lipase YES AT1G52030 805.8 P 14800.8 P 9016.5 P 10847.8 P 18.4 11.2 13.5 n/a myrosinase-binding protein 2 YES AT1G52040 181.9 P 2666.8 P 2117.9 P 2019.6 P 14.7 11.6 11.1 27.6 myrosinase-binding protein 1 YES AT1G33760 41.1 P 525.4 P 394.8 P 295.1 P 12.8 9.6 7.2 20.2 ERF022 YES AT4G01985 16.9 A 215.6 P 298.5 P 118.7 P 12.7 17.6 7.0 n/a hypothetical protein AT3G28220 3471.2 P 38 894.9 P 28 177.1 P 36 552.9 P 11.2 8.1 10.5 13.6 TRAF-like family protein AT1G66370 43.3 P 478.5 P 254.8 P 224.4 P 11.1 5.9 5.2 7.4 MYB113 YES AT4G27160 17.4 A 167.7 P 53.5 P 162.2 P 9.6 3.1 9.3 0.3 seed storage albumin 3 AT1G35140 1734.1 P 15 869.4 P 13 207.7 P 17 206.2 P 9.2 7.6 9.9 3.5 PHI-1/EXL1 YES AT5G28237 4.9 A 41.3 P 51.1 P 26.7 P 8.4 10.3 5.4 n/a tryptophan synthase-like protein AT4G27170 18.9 A 157.0 P 57.5 P 103.7 P 8.3 3.0 5.5 0.5 seed storage albumin 4 AT3G27810 7.2 A 53.6 P 57.1 P 45.0 P 7.4 7.9 6.2 1.6 MYB21 YES AT4G17920 25.1 A 184.6 P 247.3 P 222.6 P 7.3 9.8 8.9 n/a RING-H2 finger protein ATL29 AT1G61120 21.4 A 150.9 P 180.1 P 321.5 P 7.0 8.4 15.0 71.5 terpene synthase 04 YES AT1G17380 222.5 P 1471.2 P 982.5 P 1118.6 P 6.6 4.4 5.0 21.7 JAZ5/TIFY 11A YES AT4G34410 35.5 P 230.8 P 121.3 P 148.0 P 6.5 3.4 4.2 0.3 ERF109 YES AT1G52400 11461.6 P 72 721.5 P 44 118.3 P 51 794.5 P 6.3 3.8 4.5 3.1 BGLU18/beta glucosidase 18 YES AT1G52410 3933.8 P 23 332.2 P 16 344.8 P 15 134.4 P 5.9 4.2 3.8 21.9 TSK-associating protein 1 YES AT3G25180 11.3 A 66.3 P 70.7 P 131.2 P 5.9 6.3 11.7 2.9 CYP82G1 YES AT1G76640 174.4 P 999.9 P 660.1 P 551.5 P 5.7 3.8 3.2 19.2 calcium-binding protein CML39 AT2G34600 244.6 P 1319.1 P 798.1 P 1062.2 P 5.4 3.3 4.3 1.7 JAZ7/TIFY 5B YES AT5G41750 622.9 P 3355.5 P 2809.3 P 2847.5 P 5.4 4.5 4.6 n/a TIR-NBS-LRR disease-resistant protein YES AT5G59310 1746.9 P 9175.7 P 24 203.0 P 17 251.5 P 5.3 13.9 9.9 1.7 non-specific lipid-transfer protein 4 YES AT5G41740 354.1 P 1794.6 P 1501.2 P 1671.3 P 5.1 4.2 4.7 4.7 TIR-NBS-LRR disease-resistant protein YES AT1G66610 16.7 A 83.5 P 91.2 P 54.5 P 5.0 5.4 3.3 1.5 E3 ubiquitin-ligase SINA-like 1 AT1G66860 6.8 A 33.6 P 48.6 P 30.9 P 5.0 7.2 4.6 2.6 Class I glutamine amidotransferase-like AT4G28790 21.5 A 100.4 P 172.9 P 175.4 P 4.7 8.1 8.2 n/a transcription factor bHLH23 AT5G59580 17.2 A 77.4 P 66.9 P 66.7 P 4.5 3.9 3.9 0.8 UDP-glucosyl transferase 76E1/UGT76E1 AT5G24780 7476.2 P 33 245.3 P 35 412.6 P 39 064.4 P 4.4 4.7 5.2 4.8 vegetative storage protein 1 YES AT4G11911 15.3 A 67.4 P 132.8 P 84.5 P 4.4 8.7 5.5 n/a hypothetical protein AT5G13220 260.0 P 1124.2 P 815.6 P 886.6 P 4.3 3.1 3.4 11.0 JAZ10/TIFY 9 YES AT1G77640 338.2 P 1433.9 P 1595.4 P 1283.1 P 4.2 4.7 3.8 2.4 ERF013 AT3G27415 20.9 A 87.1 P 146.4 P 111.3 P 4.2 7.0 5.3 n/a hypothetical protein AT5G57560 5382.4 P 22 307.3 P 21 813.0 P 21 676.0 P 4.1 4.1 4.0 1.3 TCH4/XTH22 AT4G23215 17.7 A 73.1 P 89.4 P 74.1 P 4.1 5.0 4.2 n/a hypothetical protein AT4G29570 16.6 A 68.0 P 175.5 P 185.8 P 4.1 10.6 11.2 2.7 cytidine/deoxycytidylate deaminase-like AT2G32200 259.6 P 1056.0 P 838.7 P 1025.1 P 4.1 3.2 3.9 n/a unknown protein AT1G03880 38.8 P 155.0 P 118.8 P 557.6 P 4.0 3.1 14.4 2.2 cruciferin 2 YES AT4G10265 208.0 P 824.1 P 850.3 P 994.6 P 4.0 4.1 4.8 n/a putative wound-responsive protein YES AT5G24770 10375.6 P 40 351.9 P 65 197.1 P 55 171.0 P 3.9 6.3 5.3 n/a vegetative storage protein 2 YES AT4G12735 11.8 A 45.1 P 40.4 P 41.3 P 3.8 3.4 3.5 n/a hypothetical protein AT5G30341 9.5 A 36.2 P 43.5 P 40.7 P 3.8 4.6 4.3 n/a hypothetical protein AT5G11410 447.9 P 1698.6 P 1865.2 P 1458.5 P 3.8 4.2 3.3 18.9 protein kinase family protein AT1G52000 482.5 P 1782.7 P 1549.3 P 1480.3 P 3.7 3.2 3.1 7.1 jacalin-like lectin domain protein YES AT4G16590 33.7 P 124.4 P 247.1 P 397.6 P 3.7 7.3 11.8 4.2 cellulose synthase-like A01 AT5G52760 107.3 P 395.5 P 503.6 P 354.6 P 3.7 4.7 3.3 7.9 copper transport family protein AT1G21910 2375.9 P 8713.1 P 7190.5 P 7939.6 P 3.7 3.0 3.3 2.5 ERF012/DREB26 YES AT5G58400 52.8 P 193.5 P 184.9 P 170.7 P 3.7 3.5 3.2 1.2 peroxidase 68 YES AT1G69880 1654.5 P 6054.5 P 6756.3 P 6254.5 P 3.7 4.1 3.8 11.6 thioredoxin H8 YES AT1G19610 305.4 P 1104.3 P 1089.4 P 1019.9 P 3.6 3.6 3.3 9.3 PDF1.4/defensin-like protein 19 YES AT2G38250 121.0 P 435.5 P 681.0 P 363.1 P 3.6 5.6 3.0 3.6 DNA-binding protein AT5G49520 284.6 P 998.3 P 991.5 P 900.4 P 3.5 3.5 3.2 3.7 WRKY48 YES AT3G57260 74.7 P 260.4 P 800.4 P 1107.0 P 3.5 10.7 14.8 8.4 BGL2/β-1,3-glucanase 2/PR2 YES AT1G68290 172.0 P 599.1 P 532.4 P 625.9 P 3.5 3.1 3.6 2.4 endonuclease 2 AT2G17040 1983.1 P 6898.2 P 6486.4 P 5957.2 P 3.5 3.3 3.0 5.0 ANAC036 YES AT4G02170 45.9 P 157.5 P 188.2 P 172.0 P 3.4 4.1 3.7 2.3 hypothetical protein AT4G02330 4488.5 P 15 318.2 P 18 712.9 P 21 081.7 P 3.4 4.2 4.7 10.0 pectinesterase/ATPMEPCRB AT3G50770 504.3 P 1692.4 P 1544.9 P 2176.5 P 3.4 3.1 4.3 2.1 calcium-binding protein CML41 AT1G72520 438.8 P 1431.8 P 1316.7 P 2118.1 P 3.3 3.0 4.8 8.2 lipoxygenase 4 YES AT2G35290 468.8 P 1491.9 P 1571.7 P 1476.1 P 3.2 3.4 3.1 0.8 SAUR79 AT3G02550 111.9 P 351.8 P 411.3 P 461.1 P 3.1 3.7 4.1 1.2 LOB41 YES AT4G32060 15.2 A 47.6 P 54.6 P 86.8 P 3.1 3.6 5.7 0.7 calcium-binding EF hand protein AT4G25810 2394.0 P 7487.3 P 8327.9 P 7563.7 P 3.1 3.5 3.2 5.2 XTH23 AT1G63750 1816.6 P 5667.8 P 6791.9 P 7526.2 P 3.1 3.7 4.1 1.7 TIR-NBS-LRR disease-resistant protein YES AT2G14560 72.0 P 222.7 P 521.3 P 253.2 P 3.1 7.2 3.5 3.8 LURP1 YES AT2G05580 32.8 P 100.1 P 130.3 P 115.9 P 3.1 4.0 3.5 0.2 glycine-rich protein AT4G14365 591.9 P 1784.3 P 2416.5 P 2205.6 P 3.0 4.1 3.7 4.9 E3 ubiquitin-protein ligase XBAT34 YES AT2G23100 19.5 A 58.7 P 61.5 P 87.2 P 3.0 3.2 4.5 1.5 cys/his-rich C1 domain protein AGI* WT F† 7-5 F 10-1 F 11-6 F 7-5/WT‡ 10-1/WT 11-6/WT cob-5 /WT§ Gene description St/Def¶ AT1G72260 6.7 A 7978.0 P 2015.0 P 2907.6 P 1185.5 299.3 431.9 106.4 thionin 2.1 YES AT4G15100 15.9 A 1239.8 P 156.3 P 88.0 P 78.2 9.9 5.5 12.5 serine carboxypeptidase- like 30 AT2G39330 267.0 P 8145.8 P 4897.3 P 4446.0 P 30.5 18.3 16.7 43.1 jacalin-related lectin 23 YES AT2G41610 408.1 P 8791.5 P 6760.8 P 3807.9 P 21.5 16.6 9.3 0.7 AtDCE1 AT1G54020 166.3 P 3426.4 P 1788.4 P 1949.7 P 20.6 10.8 11.7 53.9 GDSL esterase/lipase YES AT1G52030 805.8 P 14800.8 P 9016.5 P 10847.8 P 18.4 11.2 13.5 n/a myrosinase-binding protein 2 YES AT1G52040 181.9 P 2666.8 P 2117.9 P 2019.6 P 14.7 11.6 11.1 27.6 myrosinase-binding protein 1 YES AT1G33760 41.1 P 525.4 P 394.8 P 295.1 P 12.8 9.6 7.2 20.2 ERF022 YES AT4G01985 16.9 A 215.6 P 298.5 P 118.7 P 12.7 17.6 7.0 n/a hypothetical protein AT3G28220 3471.2 P 38 894.9 P 28 177.1 P 36 552.9 P 11.2 8.1 10.5 13.6 TRAF-like family protein AT1G66370 43.3 P 478.5 P 254.8 P 224.4 P 11.1 5.9 5.2 7.4 MYB113 YES AT4G27160 17.4 A 167.7 P 53.5 P 162.2 P 9.6 3.1 9.3 0.3 seed storage albumin 3 AT1G35140 1734.1 P 15 869.4 P 13 207.7 P 17 206.2 P 9.2 7.6 9.9 3.5 PHI-1/EXL1 YES AT5G28237 4.9 A 41.3 P 51.1 P 26.7 P 8.4 10.3 5.4 n/a tryptophan synthase-like protein AT4G27170 18.9 A 157.0 P 57.5 P 103.7 P 8.3 3.0 5.5 0.5 seed storage albumin 4 AT3G27810 7.2 A 53.6 P 57.1 P 45.0 P 7.4 7.9 6.2 1.6 MYB21 YES AT4G17920 25.1 A 184.6 P 247.3 P 222.6 P 7.3 9.8 8.9 n/a RING-H2 finger protein ATL29 AT1G61120 21.4 A 150.9 P 180.1 P 321.5 P 7.0 8.4 15.0 71.5 terpene synthase 04 YES AT1G17380 222.5 P 1471.2 P 982.5 P 1118.6 P 6.6 4.4 5.0 21.7 JAZ5/TIFY 11A YES AT4G34410 35.5 P 230.8 P 121.3 P 148.0 P 6.5 3.4 4.2 0.3 ERF109 YES AT1G52400 11461.6 P 72 721.5 P 44 118.3 P 51 794.5 P 6.3 3.8 4.5 3.1 BGLU18/beta glucosidase 18 YES AT1G52410 3933.8 P 23 332.2 P 16 344.8 P 15 134.4 P 5.9 4.2 3.8 21.9 TSK-associating protein 1 YES AT3G25180 11.3 A 66.3 P 70.7 P 131.2 P 5.9 6.3 11.7 2.9 CYP82G1 YES AT1G76640 174.4 P 999.9 P 660.1 P 551.5 P 5.7 3.8 3.2 19.2 calcium-binding protein CML39 AT2G34600 244.6 P 1319.1 P 798.1 P 1062.2 P 5.4 3.3 4.3 1.7 JAZ7/TIFY 5B YES AT5G41750 622.9 P 3355.5 P 2809.3 P 2847.5 P 5.4 4.5 4.6 n/a TIR-NBS-LRR disease-resistant protein YES AT5G59310 1746.9 P 9175.7 P 24 203.0 P 17 251.5 P 5.3 13.9 9.9 1.7 non-specific lipid-transfer protein 4 YES AT5G41740 354.1 P 1794.6 P 1501.2 P 1671.3 P 5.1 4.2 4.7 4.7 TIR-NBS-LRR disease-resistant protein YES AT1G66610 16.7 A 83.5 P 91.2 P 54.5 P 5.0 5.4 3.3 1.5 E3 ubiquitin-ligase SINA-like 1 AT1G66860 6.8 A 33.6 P 48.6 P 30.9 P 5.0 7.2 4.6 2.6 Class I glutamine amidotransferase-like AT4G28790 21.5 A 100.4 P 172.9 P 175.4 P 4.7 8.1 8.2 n/a transcription factor bHLH23 AT5G59580 17.2 A 77.4 P 66.9 P 66.7 P 4.5 3.9 3.9 0.8 UDP-glucosyl transferase 76E1/UGT76E1 AT5G24780 7476.2 P 33 245.3 P 35 412.6 P 39 064.4 P 4.4 4.7 5.2 4.8 vegetative storage protein 1 YES AT4G11911 15.3 A 67.4 P 132.8 P 84.5 P 4.4 8.7 5.5 n/a hypothetical protein AT5G13220 260.0 P 1124.2 P 815.6 P 886.6 P 4.3 3.1 3.4 11.0 JAZ10/TIFY 9 YES AT1G77640 338.2 P 1433.9 P 1595.4 P 1283.1 P 4.2 4.7 3.8 2.4 ERF013 AT3G27415 20.9 A 87.1 P 146.4 P 111.3 P 4.2 7.0 5.3 n/a hypothetical protein AT5G57560 5382.4 P 22 307.3 P 21 813.0 P 21 676.0 P 4.1 4.1 4.0 1.3 TCH4/XTH22 AT4G23215 17.7 A 73.1 P 89.4 P 74.1 P 4.1 5.0 4.2 n/a hypothetical protein AT4G29570 16.6 A 68.0 P 175.5 P 185.8 P 4.1 10.6 11.2 2.7 cytidine/deoxycytidylate deaminase-like AT2G32200 259.6 P 1056.0 P 838.7 P 1025.1 P 4.1 3.2 3.9 n/a unknown protein AT1G03880 38.8 P 155.0 P 118.8 P 557.6 P 4.0 3.1 14.4 2.2 cruciferin 2 YES AT4G10265 208.0 P 824.1 P 850.3 P 994.6 P 4.0 4.1 4.8 n/a putative wound-responsive protein YES AT5G24770 10375.6 P 40 351.9 P 65 197.1 P 55 171.0 P 3.9 6.3 5.3 n/a vegetative storage protein 2 YES AT4G12735 11.8 A 45.1 P 40.4 P 41.3 P 3.8 3.4 3.5 n/a hypothetical protein AT5G30341 9.5 A 36.2 P 43.5 P 40.7 P 3.8 4.6 4.3 n/a hypothetical protein AT5G11410 447.9 P 1698.6 P 1865.2 P 1458.5 P 3.8 4.2 3.3 18.9 protein kinase family protein AT1G52000 482.5 P 1782.7 P 1549.3 P 1480.3 P 3.7 3.2 3.1 7.1 jacalin-like lectin domain protein YES AT4G16590 33.7 P 124.4 P 247.1 P 397.6 P 3.7 7.3 11.8 4.2 cellulose synthase-like A01 AT5G52760 107.3 P 395.5 P 503.6 P 354.6 P 3.7 4.7 3.3 7.9 copper transport family protein AT1G21910 2375.9 P 8713.1 P 7190.5 P 7939.6 P 3.7 3.0 3.3 2.5 ERF012/DREB26 YES AT5G58400 52.8 P 193.5 P 184.9 P 170.7 P 3.7 3.5 3.2 1.2 peroxidase 68 YES AT1G69880 1654.5 P 6054.5 P 6756.3 P 6254.5 P 3.7 4.1 3.8 11.6 thioredoxin H8 YES AT1G19610 305.4 P 1104.3 P 1089.4 P 1019.9 P 3.6 3.6 3.3 9.3 PDF1.4/defensin-like protein 19 YES AT2G38250 121.0 P 435.5 P 681.0 P 363.1 P 3.6 5.6 3.0 3.6 DNA-binding protein AT5G49520 284.6 P 998.3 P 991.5 P 900.4 P 3.5 3.5 3.2 3.7 WRKY48 YES AT3G57260 74.7 P 260.4 P 800.4 P 1107.0 P 3.5 10.7 14.8 8.4 BGL2/β-1,3-glucanase 2/PR2 YES AT1G68290 172.0 P 599.1 P 532.4 P 625.9 P 3.5 3.1 3.6 2.4 endonuclease 2 AT2G17040 1983.1 P 6898.2 P 6486.4 P 5957.2 P 3.5 3.3 3.0 5.0 ANAC036 YES AT4G02170 45.9 P 157.5 P 188.2 P 172.0 P 3.4 4.1 3.7 2.3 hypothetical protein AT4G02330 4488.5 P 15 318.2 P 18 712.9 P 21 081.7 P 3.4 4.2 4.7 10.0 pectinesterase/ATPMEPCRB AT3G50770 504.3 P 1692.4 P 1544.9 P 2176.5 P 3.4 3.1 4.3 2.1 calcium-binding protein CML41 AT1G72520 438.8 P 1431.8 P 1316.7 P 2118.1 P 3.3 3.0 4.8 8.2 lipoxygenase 4 YES AT2G35290 468.8 P 1491.9 P 1571.7 P 1476.1 P 3.2 3.4 3.1 0.8 SAUR79 AT3G02550 111.9 P 351.8 P 411.3 P 461.1 P 3.1 3.7 4.1 1.2 LOB41 YES AT4G32060 15.2 A 47.6 P 54.6 P 86.8 P 3.1 3.6 5.7 0.7 calcium-binding EF hand protein AT4G25810 2394.0 P 7487.3 P 8327.9 P 7563.7 P 3.1 3.5 3.2 5.2 XTH23 AT1G63750 1816.6 P 5667.8 P 6791.9 P 7526.2 P 3.1 3.7 4.1 1.7 TIR-NBS-LRR disease-resistant protein YES AT2G14560 72.0 P 222.7 P 521.3 P 253.2 P 3.1 7.2 3.5 3.8 LURP1 YES AT2G05580 32.8 P 100.1 P 130.3 P 115.9 P 3.1 4.0 3.5 0.2 glycine-rich protein AT4G14365 591.9 P 1784.3 P 2416.5 P 2205.6 P 3.0 4.1 3.7 4.9 E3 ubiquitin-protein ligase XBAT34 YES AT2G23100 19.5 A 58.7 P 61.5 P 87.2 P 3.0 3.2 4.5 1.5 cys/his-rich C1 domain protein *AGI, Arabidopsis Gene Index. †F, flag calling (presence, absence, marginal). ‡Fold change. §Data from Ko et al. (2006b). ¶Stress/defence-related genes. View Large Table 1. List of genes upregulated in 35S::AtDICE1 plants AGI* WT F† 7-5 F 10-1 F 11-6 F 7-5/WT‡ 10-1/WT 11-6/WT cob-5 /WT§ Gene description St/Def¶ AT1G72260 6.7 A 7978.0 P 2015.0 P 2907.6 P 1185.5 299.3 431.9 106.4 thionin 2.1 YES AT4G15100 15.9 A 1239.8 P 156.3 P 88.0 P 78.2 9.9 5.5 12.5 serine carboxypeptidase- like 30 AT2G39330 267.0 P 8145.8 P 4897.3 P 4446.0 P 30.5 18.3 16.7 43.1 jacalin-related lectin 23 YES AT2G41610 408.1 P 8791.5 P 6760.8 P 3807.9 P 21.5 16.6 9.3 0.7 AtDCE1 AT1G54020 166.3 P 3426.4 P 1788.4 P 1949.7 P 20.6 10.8 11.7 53.9 GDSL esterase/lipase YES AT1G52030 805.8 P 14800.8 P 9016.5 P 10847.8 P 18.4 11.2 13.5 n/a myrosinase-binding protein 2 YES AT1G52040 181.9 P 2666.8 P 2117.9 P 2019.6 P 14.7 11.6 11.1 27.6 myrosinase-binding protein 1 YES AT1G33760 41.1 P 525.4 P 394.8 P 295.1 P 12.8 9.6 7.2 20.2 ERF022 YES AT4G01985 16.9 A 215.6 P 298.5 P 118.7 P 12.7 17.6 7.0 n/a hypothetical protein AT3G28220 3471.2 P 38 894.9 P 28 177.1 P 36 552.9 P 11.2 8.1 10.5 13.6 TRAF-like family protein AT1G66370 43.3 P 478.5 P 254.8 P 224.4 P 11.1 5.9 5.2 7.4 MYB113 YES AT4G27160 17.4 A 167.7 P 53.5 P 162.2 P 9.6 3.1 9.3 0.3 seed storage albumin 3 AT1G35140 1734.1 P 15 869.4 P 13 207.7 P 17 206.2 P 9.2 7.6 9.9 3.5 PHI-1/EXL1 YES AT5G28237 4.9 A 41.3 P 51.1 P 26.7 P 8.4 10.3 5.4 n/a tryptophan synthase-like protein AT4G27170 18.9 A 157.0 P 57.5 P 103.7 P 8.3 3.0 5.5 0.5 seed storage albumin 4 AT3G27810 7.2 A 53.6 P 57.1 P 45.0 P 7.4 7.9 6.2 1.6 MYB21 YES AT4G17920 25.1 A 184.6 P 247.3 P 222.6 P 7.3 9.8 8.9 n/a RING-H2 finger protein ATL29 AT1G61120 21.4 A 150.9 P 180.1 P 321.5 P 7.0 8.4 15.0 71.5 terpene synthase 04 YES AT1G17380 222.5 P 1471.2 P 982.5 P 1118.6 P 6.6 4.4 5.0 21.7 JAZ5/TIFY 11A YES AT4G34410 35.5 P 230.8 P 121.3 P 148.0 P 6.5 3.4 4.2 0.3 ERF109 YES AT1G52400 11461.6 P 72 721.5 P 44 118.3 P 51 794.5 P 6.3 3.8 4.5 3.1 BGLU18/beta glucosidase 18 YES AT1G52410 3933.8 P 23 332.2 P 16 344.8 P 15 134.4 P 5.9 4.2 3.8 21.9 TSK-associating protein 1 YES AT3G25180 11.3 A 66.3 P 70.7 P 131.2 P 5.9 6.3 11.7 2.9 CYP82G1 YES AT1G76640 174.4 P 999.9 P 660.1 P 551.5 P 5.7 3.8 3.2 19.2 calcium-binding protein CML39 AT2G34600 244.6 P 1319.1 P 798.1 P 1062.2 P 5.4 3.3 4.3 1.7 JAZ7/TIFY 5B YES AT5G41750 622.9 P 3355.5 P 2809.3 P 2847.5 P 5.4 4.5 4.6 n/a TIR-NBS-LRR disease-resistant protein YES AT5G59310 1746.9 P 9175.7 P 24 203.0 P 17 251.5 P 5.3 13.9 9.9 1.7 non-specific lipid-transfer protein 4 YES AT5G41740 354.1 P 1794.6 P 1501.2 P 1671.3 P 5.1 4.2 4.7 4.7 TIR-NBS-LRR disease-resistant protein YES AT1G66610 16.7 A 83.5 P 91.2 P 54.5 P 5.0 5.4 3.3 1.5 E3 ubiquitin-ligase SINA-like 1 AT1G66860 6.8 A 33.6 P 48.6 P 30.9 P 5.0 7.2 4.6 2.6 Class I glutamine amidotransferase-like AT4G28790 21.5 A 100.4 P 172.9 P 175.4 P 4.7 8.1 8.2 n/a transcription factor bHLH23 AT5G59580 17.2 A 77.4 P 66.9 P 66.7 P 4.5 3.9 3.9 0.8 UDP-glucosyl transferase 76E1/UGT76E1 AT5G24780 7476.2 P 33 245.3 P 35 412.6 P 39 064.4 P 4.4 4.7 5.2 4.8 vegetative storage protein 1 YES AT4G11911 15.3 A 67.4 P 132.8 P 84.5 P 4.4 8.7 5.5 n/a hypothetical protein AT5G13220 260.0 P 1124.2 P 815.6 P 886.6 P 4.3 3.1 3.4 11.0 JAZ10/TIFY 9 YES AT1G77640 338.2 P 1433.9 P 1595.4 P 1283.1 P 4.2 4.7 3.8 2.4 ERF013 AT3G27415 20.9 A 87.1 P 146.4 P 111.3 P 4.2 7.0 5.3 n/a hypothetical protein AT5G57560 5382.4 P 22 307.3 P 21 813.0 P 21 676.0 P 4.1 4.1 4.0 1.3 TCH4/XTH22 AT4G23215 17.7 A 73.1 P 89.4 P 74.1 P 4.1 5.0 4.2 n/a hypothetical protein AT4G29570 16.6 A 68.0 P 175.5 P 185.8 P 4.1 10.6 11.2 2.7 cytidine/deoxycytidylate deaminase-like AT2G32200 259.6 P 1056.0 P 838.7 P 1025.1 P 4.1 3.2 3.9 n/a unknown protein AT1G03880 38.8 P 155.0 P 118.8 P 557.6 P 4.0 3.1 14.4 2.2 cruciferin 2 YES AT4G10265 208.0 P 824.1 P 850.3 P 994.6 P 4.0 4.1 4.8 n/a putative wound-responsive protein YES AT5G24770 10375.6 P 40 351.9 P 65 197.1 P 55 171.0 P 3.9 6.3 5.3 n/a vegetative storage protein 2 YES AT4G12735 11.8 A 45.1 P 40.4 P 41.3 P 3.8 3.4 3.5 n/a hypothetical protein AT5G30341 9.5 A 36.2 P 43.5 P 40.7 P 3.8 4.6 4.3 n/a hypothetical protein AT5G11410 447.9 P 1698.6 P 1865.2 P 1458.5 P 3.8 4.2 3.3 18.9 protein kinase family protein AT1G52000 482.5 P 1782.7 P 1549.3 P 1480.3 P 3.7 3.2 3.1 7.1 jacalin-like lectin domain protein YES AT4G16590 33.7 P 124.4 P 247.1 P 397.6 P 3.7 7.3 11.8 4.2 cellulose synthase-like A01 AT5G52760 107.3 P 395.5 P 503.6 P 354.6 P 3.7 4.7 3.3 7.9 copper transport family protein AT1G21910 2375.9 P 8713.1 P 7190.5 P 7939.6 P 3.7 3.0 3.3 2.5 ERF012/DREB26 YES AT5G58400 52.8 P 193.5 P 184.9 P 170.7 P 3.7 3.5 3.2 1.2 peroxidase 68 YES AT1G69880 1654.5 P 6054.5 P 6756.3 P 6254.5 P 3.7 4.1 3.8 11.6 thioredoxin H8 YES AT1G19610 305.4 P 1104.3 P 1089.4 P 1019.9 P 3.6 3.6 3.3 9.3 PDF1.4/defensin-like protein 19 YES AT2G38250 121.0 P 435.5 P 681.0 P 363.1 P 3.6 5.6 3.0 3.6 DNA-binding protein AT5G49520 284.6 P 998.3 P 991.5 P 900.4 P 3.5 3.5 3.2 3.7 WRKY48 YES AT3G57260 74.7 P 260.4 P 800.4 P 1107.0 P 3.5 10.7 14.8 8.4 BGL2/β-1,3-glucanase 2/PR2 YES AT1G68290 172.0 P 599.1 P 532.4 P 625.9 P 3.5 3.1 3.6 2.4 endonuclease 2 AT2G17040 1983.1 P 6898.2 P 6486.4 P 5957.2 P 3.5 3.3 3.0 5.0 ANAC036 YES AT4G02170 45.9 P 157.5 P 188.2 P 172.0 P 3.4 4.1 3.7 2.3 hypothetical protein AT4G02330 4488.5 P 15 318.2 P 18 712.9 P 21 081.7 P 3.4 4.2 4.7 10.0 pectinesterase/ATPMEPCRB AT3G50770 504.3 P 1692.4 P 1544.9 P 2176.5 P 3.4 3.1 4.3 2.1 calcium-binding protein CML41 AT1G72520 438.8 P 1431.8 P 1316.7 P 2118.1 P 3.3 3.0 4.8 8.2 lipoxygenase 4 YES AT2G35290 468.8 P 1491.9 P 1571.7 P 1476.1 P 3.2 3.4 3.1 0.8 SAUR79 AT3G02550 111.9 P 351.8 P 411.3 P 461.1 P 3.1 3.7 4.1 1.2 LOB41 YES AT4G32060 15.2 A 47.6 P 54.6 P 86.8 P 3.1 3.6 5.7 0.7 calcium-binding EF hand protein AT4G25810 2394.0 P 7487.3 P 8327.9 P 7563.7 P 3.1 3.5 3.2 5.2 XTH23 AT1G63750 1816.6 P 5667.8 P 6791.9 P 7526.2 P 3.1 3.7 4.1 1.7 TIR-NBS-LRR disease-resistant protein YES AT2G14560 72.0 P 222.7 P 521.3 P 253.2 P 3.1 7.2 3.5 3.8 LURP1 YES AT2G05580 32.8 P 100.1 P 130.3 P 115.9 P 3.1 4.0 3.5 0.2 glycine-rich protein AT4G14365 591.9 P 1784.3 P 2416.5 P 2205.6 P 3.0 4.1 3.7 4.9 E3 ubiquitin-protein ligase XBAT34 YES AT2G23100 19.5 A 58.7 P 61.5 P 87.2 P 3.0 3.2 4.5 1.5 cys/his-rich C1 domain protein AGI* WT F† 7-5 F 10-1 F 11-6 F 7-5/WT‡ 10-1/WT 11-6/WT cob-5 /WT§ Gene description St/Def¶ AT1G72260 6.7 A 7978.0 P 2015.0 P 2907.6 P 1185.5 299.3 431.9 106.4 thionin 2.1 YES AT4G15100 15.9 A 1239.8 P 156.3 P 88.0 P 78.2 9.9 5.5 12.5 serine carboxypeptidase- like 30 AT2G39330 267.0 P 8145.8 P 4897.3 P 4446.0 P 30.5 18.3 16.7 43.1 jacalin-related lectin 23 YES AT2G41610 408.1 P 8791.5 P 6760.8 P 3807.9 P 21.5 16.6 9.3 0.7 AtDCE1 AT1G54020 166.3 P 3426.4 P 1788.4 P 1949.7 P 20.6 10.8 11.7 53.9 GDSL esterase/lipase YES AT1G52030 805.8 P 14800.8 P 9016.5 P 10847.8 P 18.4 11.2 13.5 n/a myrosinase-binding protein 2 YES AT1G52040 181.9 P 2666.8 P 2117.9 P 2019.6 P 14.7 11.6 11.1 27.6 myrosinase-binding protein 1 YES AT1G33760 41.1 P 525.4 P 394.8 P 295.1 P 12.8 9.6 7.2 20.2 ERF022 YES AT4G01985 16.9 A 215.6 P 298.5 P 118.7 P 12.7 17.6 7.0 n/a hypothetical protein AT3G28220 3471.2 P 38 894.9 P 28 177.1 P 36 552.9 P 11.2 8.1 10.5 13.6 TRAF-like family protein AT1G66370 43.3 P 478.5 P 254.8 P 224.4 P 11.1 5.9 5.2 7.4 MYB113 YES AT4G27160 17.4 A 167.7 P 53.5 P 162.2 P 9.6 3.1 9.3 0.3 seed storage albumin 3 AT1G35140 1734.1 P 15 869.4 P 13 207.7 P 17 206.2 P 9.2 7.6 9.9 3.5 PHI-1/EXL1 YES AT5G28237 4.9 A 41.3 P 51.1 P 26.7 P 8.4 10.3 5.4 n/a tryptophan synthase-like protein AT4G27170 18.9 A 157.0 P 57.5 P 103.7 P 8.3 3.0 5.5 0.5 seed storage albumin 4 AT3G27810 7.2 A 53.6 P 57.1 P 45.0 P 7.4 7.9 6.2 1.6 MYB21 YES AT4G17920 25.1 A 184.6 P 247.3 P 222.6 P 7.3 9.8 8.9 n/a RING-H2 finger protein ATL29 AT1G61120 21.4 A 150.9 P 180.1 P 321.5 P 7.0 8.4 15.0 71.5 terpene synthase 04 YES AT1G17380 222.5 P 1471.2 P 982.5 P 1118.6 P 6.6 4.4 5.0 21.7 JAZ5/TIFY 11A YES AT4G34410 35.5 P 230.8 P 121.3 P 148.0 P 6.5 3.4 4.2 0.3 ERF109 YES AT1G52400 11461.6 P 72 721.5 P 44 118.3 P 51 794.5 P 6.3 3.8 4.5 3.1 BGLU18/beta glucosidase 18 YES AT1G52410 3933.8 P 23 332.2 P 16 344.8 P 15 134.4 P 5.9 4.2 3.8 21.9 TSK-associating protein 1 YES AT3G25180 11.3 A 66.3 P 70.7 P 131.2 P 5.9 6.3 11.7 2.9 CYP82G1 YES AT1G76640 174.4 P 999.9 P 660.1 P 551.5 P 5.7 3.8 3.2 19.2 calcium-binding protein CML39 AT2G34600 244.6 P 1319.1 P 798.1 P 1062.2 P 5.4 3.3 4.3 1.7 JAZ7/TIFY 5B YES AT5G41750 622.9 P 3355.5 P 2809.3 P 2847.5 P 5.4 4.5 4.6 n/a TIR-NBS-LRR disease-resistant protein YES AT5G59310 1746.9 P 9175.7 P 24 203.0 P 17 251.5 P 5.3 13.9 9.9 1.7 non-specific lipid-transfer protein 4 YES AT5G41740 354.1 P 1794.6 P 1501.2 P 1671.3 P 5.1 4.2 4.7 4.7 TIR-NBS-LRR disease-resistant protein YES AT1G66610 16.7 A 83.5 P 91.2 P 54.5 P 5.0 5.4 3.3 1.5 E3 ubiquitin-ligase SINA-like 1 AT1G66860 6.8 A 33.6 P 48.6 P 30.9 P 5.0 7.2 4.6 2.6 Class I glutamine amidotransferase-like AT4G28790 21.5 A 100.4 P 172.9 P 175.4 P 4.7 8.1 8.2 n/a transcription factor bHLH23 AT5G59580 17.2 A 77.4 P 66.9 P 66.7 P 4.5 3.9 3.9 0.8 UDP-glucosyl transferase 76E1/UGT76E1 AT5G24780 7476.2 P 33 245.3 P 35 412.6 P 39 064.4 P 4.4 4.7 5.2 4.8 vegetative storage protein 1 YES AT4G11911 15.3 A 67.4 P 132.8 P 84.5 P 4.4 8.7 5.5 n/a hypothetical protein AT5G13220 260.0 P 1124.2 P 815.6 P 886.6 P 4.3 3.1 3.4 11.0 JAZ10/TIFY 9 YES AT1G77640 338.2 P 1433.9 P 1595.4 P 1283.1 P 4.2 4.7 3.8 2.4 ERF013 AT3G27415 20.9 A 87.1 P 146.4 P 111.3 P 4.2 7.0 5.3 n/a hypothetical protein AT5G57560 5382.4 P 22 307.3 P 21 813.0 P 21 676.0 P 4.1 4.1 4.0 1.3 TCH4/XTH22 AT4G23215 17.7 A 73.1 P 89.4 P 74.1 P 4.1 5.0 4.2 n/a hypothetical protein AT4G29570 16.6 A 68.0 P 175.5 P 185.8 P 4.1 10.6 11.2 2.7 cytidine/deoxycytidylate deaminase-like AT2G32200 259.6 P 1056.0 P 838.7 P 1025.1 P 4.1 3.2 3.9 n/a unknown protein AT1G03880 38.8 P 155.0 P 118.8 P 557.6 P 4.0 3.1 14.4 2.2 cruciferin 2 YES AT4G10265 208.0 P 824.1 P 850.3 P 994.6 P 4.0 4.1 4.8 n/a putative wound-responsive protein YES AT5G24770 10375.6 P 40 351.9 P 65 197.1 P 55 171.0 P 3.9 6.3 5.3 n/a vegetative storage protein 2 YES AT4G12735 11.8 A 45.1 P 40.4 P 41.3 P 3.8 3.4 3.5 n/a hypothetical protein AT5G30341 9.5 A 36.2 P 43.5 P 40.7 P 3.8 4.6 4.3 n/a hypothetical protein AT5G11410 447.9 P 1698.6 P 1865.2 P 1458.5 P 3.8 4.2 3.3 18.9 protein kinase family protein AT1G52000 482.5 P 1782.7 P 1549.3 P 1480.3 P 3.7 3.2 3.1 7.1 jacalin-like lectin domain protein YES AT4G16590 33.7 P 124.4 P 247.1 P 397.6 P 3.7 7.3 11.8 4.2 cellulose synthase-like A01 AT5G52760 107.3 P 395.5 P 503.6 P 354.6 P 3.7 4.7 3.3 7.9 copper transport family protein AT1G21910 2375.9 P 8713.1 P 7190.5 P 7939.6 P 3.7 3.0 3.3 2.5 ERF012/DREB26 YES AT5G58400 52.8 P 193.5 P 184.9 P 170.7 P 3.7 3.5 3.2 1.2 peroxidase 68 YES AT1G69880 1654.5 P 6054.5 P 6756.3 P 6254.5 P 3.7 4.1 3.8 11.6 thioredoxin H8 YES AT1G19610 305.4 P 1104.3 P 1089.4 P 1019.9 P 3.6 3.6 3.3 9.3 PDF1.4/defensin-like protein 19 YES AT2G38250 121.0 P 435.5 P 681.0 P 363.1 P 3.6 5.6 3.0 3.6 DNA-binding protein AT5G49520 284.6 P 998.3 P 991.5 P 900.4 P 3.5 3.5 3.2 3.7 WRKY48 YES AT3G57260 74.7 P 260.4 P 800.4 P 1107.0 P 3.5 10.7 14.8 8.4 BGL2/β-1,3-glucanase 2/PR2 YES AT1G68290 172.0 P 599.1 P 532.4 P 625.9 P 3.5 3.1 3.6 2.4 endonuclease 2 AT2G17040 1983.1 P 6898.2 P 6486.4 P 5957.2 P 3.5 3.3 3.0 5.0 ANAC036 YES AT4G02170 45.9 P 157.5 P 188.2 P 172.0 P 3.4 4.1 3.7 2.3 hypothetical protein AT4G02330 4488.5 P 15 318.2 P 18 712.9 P 21 081.7 P 3.4 4.2 4.7 10.0 pectinesterase/ATPMEPCRB AT3G50770 504.3 P 1692.4 P 1544.9 P 2176.5 P 3.4 3.1 4.3 2.1 calcium-binding protein CML41 AT1G72520 438.8 P 1431.8 P 1316.7 P 2118.1 P 3.3 3.0 4.8 8.2 lipoxygenase 4 YES AT2G35290 468.8 P 1491.9 P 1571.7 P 1476.1 P 3.2 3.4 3.1 0.8 SAUR79 AT3G02550 111.9 P 351.8 P 411.3 P 461.1 P 3.1 3.7 4.1 1.2 LOB41 YES AT4G32060 15.2 A 47.6 P 54.6 P 86.8 P 3.1 3.6 5.7 0.7 calcium-binding EF hand protein AT4G25810 2394.0 P 7487.3 P 8327.9 P 7563.7 P 3.1 3.5 3.2 5.2 XTH23 AT1G63750 1816.6 P 5667.8 P 6791.9 P 7526.2 P 3.1 3.7 4.1 1.7 TIR-NBS-LRR disease-resistant protein YES AT2G14560 72.0 P 222.7 P 521.3 P 253.2 P 3.1 7.2 3.5 3.8 LURP1 YES AT2G05580 32.8 P 100.1 P 130.3 P 115.9 P 3.1 4.0 3.5 0.2 glycine-rich protein AT4G14365 591.9 P 1784.3 P 2416.5 P 2205.6 P 3.0 4.1 3.7 4.9 E3 ubiquitin-protein ligase XBAT34 YES AT2G23100 19.5 A 58.7 P 61.5 P 87.2 P 3.0 3.2 4.5 1.5 cys/his-rich C1 domain protein *AGI, Arabidopsis Gene Index. †F, flag calling (presence, absence, marginal). ‡Fold change. §Data from Ko et al. (2006b). ¶Stress/defence-related genes. View Large To verify the microarray results, we performed a semi-quantitative RT-PCR analysis of 15 selected genes, including six that were upregulated, four that were downregulated and five that had no significant changes in gene expression in 35S::AtDICE1 plants. The band intensities of the PCR products were quite comparable to the microarray data (Fig. S5). Abnormal secondary wall thickenings in pith cells of the inflorescent stem in 35S::AtDICE1 plants Because the growth of inflorescent stems in 35S::AtDICE1 plants was severely inhibited and the surface of the stems was bumpy, the anatomy of the inflorescent stems of 35S::AtDICE1and WT plants was examined using microtome sectioning. Surprisingly, we found abnormal secondary wall thickenings in the pith cells of 35S::AtDICE1 plants in addition to the epidermal cell swelling/enlargements (Fig. 8). This phenotype was also observed in the stem sections of 35S::PtrDICE1 plants (Fig. S7). Cross-sections along the stem length indicate that these abnormal secondary wall thickenings are regional or sporadic (Fig. S7B). Interestingly, this result is reminiscent of the eli1 mutant, which has defects in cell expansion due to mutation of the CesA3 gene (Caño-Delgado et al., 2000, 2003). Fig. 8. View largeDownload slide 35S::AtDICE1 plants showed abnormal secondary wall thickenings in pith of inflorescent stems. Stem cross-sections of 50-d-old 35S::AtDICE1 and WT plants stained with 0.05 % toluidine blue O (TBO) and 2 % phloroglucinol/HCl (Phloroglucinol) as indicated. Pith cells having secondary wall thickenings are indicated by arrows and swollen cells by arrowheads. Scale bars represent 100 µm. Fig. 8. View largeDownload slide 35S::AtDICE1 plants showed abnormal secondary wall thickenings in pith of inflorescent stems. Stem cross-sections of 50-d-old 35S::AtDICE1 and WT plants stained with 0.05 % toluidine blue O (TBO) and 2 % phloroglucinol/HCl (Phloroglucinol) as indicated. Pith cells having secondary wall thickenings are indicated by arrows and swollen cells by arrowheads. Scale bars represent 100 µm. DISCUSSION Precise regulation of cell expansion is essential for cellular morphogenesis and differentiation and consequently directs typical plant growth and development. It is well known that cellulose microfibrils strengthen cell walls and define the direction of anisotropic cell growth (Cosgrove, 2005; Lloyd and Chan, 2006). In this report, we attempted to characterize a gene locus, AtDICE1, which encodes a novel transmembrane protein probably localized to the ER. Upregulation of AtDICE1 caused abnormal cell expansion Upregulation of AtDICE1 caused severe growth and developmental defects at the whole plant level (Figs 1, 2 and S3, Table S1). The prominent cell swelling phenotype, caused by defects in anisotropic cell expansion, probably accounted for the dwarf phenotype of 35S::AtDICE1 plants (Figs 2 and 3). This phenotypic consequence was reproduced by overexpression of PtrDICE1, a poplar homologue of AtDICE1 (Fig. 1E) and was further verified by inducible expression of AtDICE1 which resulted in a clear phenotype as early as in 2-d-old Arabidopsis seedlings (Fig. 3). We hypothesized that the cell swelling phenotype of 35S::AtDICE1 plants may result from alterations in cell wall integrity, especially the decrease in cellulose content estimated from our cell wall monosaccharide analysis (Fig. 4). This is further supported by the following results: (1) treatment with isoxaben, a well-known cellulose synthase inhibitor, mimicked the cell swelling phenotype of 35S::AtDICE1 plants in WT plants (Fig. 4A), and (2) cortical microtubule arrangements, which direct proper cellulose deposition in cell walls via an association with CesA interacting1/CSI1 (Paredez et al., 2006; Bringmann et al., 2012; Li et al., 2012), were severely altered in 35S::AtDICE1/GFP-TuA6 plants (Fig. 4C). Previously, loss-of-function mutants in several genes, including RSW1/CesA1, RSW5/CesA3, QUILL/PROCUSTE/CesA6, POM-POM1/CHITINASE LIKE1 (POM1/CTL1), KORRIGAN/LION’S TAIL1/RADIALLY SWOLLEN2 (KOR/LIT/RSW2), COBRA and KOBITO/ELONGATION DEFECTIVE1 (KOB1/ELD1), involved in cellulose biosynthesis or cell wall modification were reported to have cell swelling and dwarfing phenotypes similar to that observed in 35S::AtDICE1 plants (Hauser et al., 1995; Arioli et al., 1998; Nicol et al., 1998; Fagard et al., 2000; Lane et al., 2001; Schindelman et al., 2001; Pagant et al., 2002; Wang et al., 2006). The korrigan mutant showed abnormal cell expansion phenotypes in roots and hypocotyls, and was found to be defective in an endo-1,4-d-glucanase (Nicol et al., 1998). The cobra mutant exhibited abnormal cell expansion and cellulose deposition in root cells resulting from a failure of orientated cell expansion (Roudier et al., 2002). Furthermore, our whole transcriptome analysis revealed the relevant transcription profile of 35S::AtDICE1 plants responsible for the defects in anisotropic cell expansion (Table 1). For example, expression of the MYB21 and ANAC036 transcription factors, which are known to negatively regulate cell elongation, were up-regulated up to 7.9-fold in 35S::AtDICE1 plants. Overexpression of an MYB21-GR fusion protein resulted in decreased hypocotyl cell expansion, shorter stems, and smaller and narrower leaves (Shin et al., 2002). Overexpression of ANAC036 also resulted in a semi-dwarf phenotype (Kato et al., 2010). In addition, many cell wall formation/modification-related genes were upregulated, including xyloglucan endotransglucosylase/hydrolase (XTH22/TCH4 and XTH23), pectinesterase, cellulose synthase-like A01, lipid transfer proteins and peroxidase (Passardi et al., 2004; Nieuwland et al., 2005) (Table 1). Upregulation of AtDICE1 invokes a cellular defence mechanism Interference with cell wall integrity results in various compensatory reactions, such as the reactive oxygen species (ROS) response, ectopic lignin deposition (Caño-Delgado et al., 2003; Denness et al., 2011), altered pectin methyl-esterification status (Manfield et al., 2004), elevated callose deposition (Desprez et al., 2002), and increased JA and ethylene production with an associated resistance to pathogens (Ellis et al., 2002). Thus, the finding that a high proportion (51.4 %) of the upregulated genes in 35S::AtDICE1 plants were genes involved in stress and defence responses can be explained as the response of plant cells to changes in cell wall integrity and abnormal cell shape (Fig. 7 and Table 1). In particular, JA biosynthesis, signalling and responsive genes were massively upregulated. For example, Thionin 2.1 (AT1G72260), a cysteine-rich protein with antimicrobial properties, was dramatically upregulated (up to 1185-fold). Myrosinase-binding proteins (MBP1, AT1G52040; and MBP2, AT1G52030), involved in metabolizing defence compounds to protect against herbivory, were upregulated (up to 18-fold). Additionally, vegetative storage proteins (VSP1, AT5G24780; and VSP2, AT5G24770), known to be induced by JA, were upregulated (up to six-fold). It has been reported that COBRA, a glycosyl phosphatidyl inositol (GPI)-anchored protein, is targeted to the plasma membrane and involved in orientated cell expansion in Arabidopsis (Roudier et al., 2005). cob-5, a loss-of-function mutant of COBRA with an epidermal cell swelling phenotype similar to 35S::AtDICE1, over-accumulates JA and coordinately induces stress- and defence-related genes (Ko et al., 2006b). Interestingly, our whole transcriptome analysis revealed that the upregulated genes in 35S::AtDICE1 plants overlapped dramatically with those upregulated in cob-5. Specifically, 32 of the 72 upregulated genes were also upregulated more than three-fold in cob-5 and, among them (i.e. 32 genes), 23 genes (~72 %) were involved in the stress/defence response (Table 1). This suggests that the changes in cell wall integrity caused by cell swelling in 35S::AtDICE1 plants induce cellular defence responses as in cob-5. Taken together, these results further support the hypothesis that plant cells initially perceive biotic stress at the cell surface (Ko et al., 2006b). While the overexpression of AtDICE1 induced conspicuous phenotypes, the suppression of AtDICE1 resulted in no distinguishable phenotypic changes compared to WT (Fig. S4). However, expression of Thionin 2.1, which was dramatically upregulated in 35S::AtDICE1 plants, was significantly downregulated in AtDICE1_RNAi plants (Fig. S6). Thus, it is possible that the suppression of AtDICE1 in AtDICE1_RNAi plants was not strong enough to cause phenotypic changes. Generating a loss-of-function mutant using CRISPR/Cas9 technology may be required to address this hypothesis. Functional significances of AtDICE1 Amino acid sequence analysis predicted that AtDICE1 is a membrane protein having seven transmembrane domains, and containing a putative di-lysine (KK) motif at the C-terminus, which is required for ER targeting (Fig. S2; Teasdale and Jackson, 1996). Indeed, the results of our transient expression analysis of GFP:AtDICE1 suggest that AtDICE1 is probably localized to the ER (Fig. 5). The ER is a cytoplasmic network of flattened, membrane-enclosed sacs or tube-like structures known as cisternae that play important roles in the folding of protein molecules and the transport of synthesized proteins in vesicles to the Golgi apparatus (Chen et al., 2012). Correct folding of newly synthesized proteins is made possible by several ER chaperone proteins and only properly folded proteins are transported from the rough ER to the Golgi apparatus. We hypothesize that AtDICE1 might be involved in the processing of some proteins responsible for cell wall formation, such as CesA proteins, in the ER via a currently unknown mechanism. Previously, mutants of the PEANUT1 gene encoding an ER-localized mannosyltransferase required for synthesis of the GPI anchor showed radially swollen embryos and decreased crystalline cellulose in the cell wall (Gillmor et al., 2005). Thus, it might also be plausible that the overly increased AtDICE1 protein population in the ER might disturb other proteins such as PEANUT1. If this is the case, the N-terminal external domain is obligatory because overexpression of ΔNT-AtDICE1 was not associated with any phenotypic changes (Fig. 6). In addition, because AtDICE1 was identified as a member of a gene network regulating secondary xylem development in Arabidopsis (Ko et al., 2006a) and confirmed to be expressed specifically in vascular tissues (Fig. S1), it might be plausible that AtDICE1 may function in vascular tissue formation by regulating anisotropic cell elongation. It remains uncertain how cell elongation is integrated with secondary wall formation, which occurs irreversibly in some specialized cells such as xylem cells and fibres after cessation of cell elongation (Boerjan et al., 2003; Déjardin et al., 2010). Previously, the eli1 mutant exhibited a stunted phenotype and xylem cell disorganization in stem pith, which was caused by altered cell expansion with inappropriate initiation of secondary wall formation (Caño-Delgado et al., 2000). Because ectopic secondary wall formation or lignification phenotypes were observed in many other cell expansion mutants (e.g. rsw1, korrigan1/lit, det3), a mechanism that senses cell size and induces subsequent secondary wall formation may exist (Caño-Delgado et al., 2000). Subsequently, it was shown that eli1 mutants occur in the cellulose synthase gene CesA3 (Caño-Delgado et al., 2003). Recently, AtC3H14, a plant-specific tandem CCCH zinc-finger protein, was suggested to be a coordinate regulator of cell elongation and secondary wall formation during xylogenesis because overexpression of AtC3H14 or PtC3H17, a poplar homologuw of AtC3H14, suppressed cell elongation but increased secondary wall thickening (Kim et al., 2014; Chai et al., 2015). In this report, we found abnormal secondary wall thickenings in the pith cells of 35S::AtDICE1 and 35S::PtrDICE1 plants (Figs 8 and S7). This result further supports the hypothesis that a cell size sensing mechanism exists, leading to secondary wall formation. However, because these abnormal secondary wall thickenings in stem tissues of 35S::AtDICE1 plant are patchy (Fig. S7B), we assumed that the overall cellulose contents in adult stems of 35S::AtDICE1 plants were lower than in the WT, as shown in Fig. 4B. Furthermore, expression of AtC3H14 was not significantly altered in 35S::AtDICE1 plants and, furthermore, no changes in AtDICE1 were observed in 35S::AtC3H14 plants (data not shown), suggesting that it is unlikely that a mechanistic connection exists between them. In conclusion, our data suggest that AtDICE1 is a novel putative ER-localized transmembrane protein contributing to a proper anisotropic cell elongation process through participation in cell wall formation via a currently unknown mechanism. Because AtDICE1 was identified as a member of a gene network regulating secondary xylem development in Arabidopsis (Ko et al., 2006a) and confirmed to be expressed specifically in secondary wall-forming tissues (Fig. S1), it is possible that AtDICE1 might be involved in the anisotropic cell elongation during xylogenesis. Further in-depth investigations, including loss-of-function mutation employing CRISPR/Cas9 technology, immunohistochemistry and protein–protein interactions with cell-wall-forming proteins (e.g. AtDICE1 with CESA complexes), will be required to elucidate AtDICE1 function in detail. SUPPLEMENTARY DATA Supplementary data are available online at https://academic.oup.com/aob and consist of the following. Table S1: Growth features of 35S::AtDICE1. Table S2: List of genes differentially expressed in 35S::AtDICE1. Table S3: Primers used in this study. Figure S1: AtDICE1 (At2g41610) is specifically expressed in vascular tissues/cells. Figure S2: AtDICE1 is a plant-specific transmembrane protein. Figure S3: Floral organs of 35S::AtDICE1 plants are smaller than in WT. Figure S4: Phenotypic characterization of AtDICE1_RNAi plants. Figure S5: Confirmation of the microarray data using semi-quantitative RT-PCR analysis. Figure S6: Expression of the Thionin 2.1 gene was significantly downregulated in AtDICE1_RNAi plants. Figure S7: 35S::PtrDICE1 resulted in abnormal secondary wall thickenings in pith cells as in 35S::AtDICE1 plants. ACKNOWLEDGEMENTS We are grateful to Prof. Kim (Youngrok) at Kyung Hee University for technical assistance with SEM analysis. This work was supported by the Forest Resources Genome Project (2014071G10-1722-AA04) and by a grant from the Basic Science Research Program through the National Research Foundation of Korea (NRF-2015R1D1A1A01060807). LITERATURE CITED Arioli T , Peng L , Betzner AS , et al. 1998 . Molecular analysis of cellulose biosynthesis in Arabidopsis . Science 279 : 717 – 720 . Google Scholar CrossRef Search ADS PubMed Baskin TI . 2005 . Anisotropic expansion of the plant cell wall . Annual Review of Cell and Developmental Biology 21 : 203 – 222 . Google Scholar CrossRef Search ADS PubMed Boerjan W , Ralph J , Baucher M . 2003 . Lignin biosynthesis . Annual Review of Plant Biology 54 : 519 – 546 . Google Scholar CrossRef Search ADS PubMed Bringmann M , Li E , Sampathkumar A , Kocabek T , Hauser MT , Persson S . 2012 . POM-POM2/CELLULOSE SYNTHASE INTERACTING1 is essential for the functional association of cellulose synthase and microtubules in Arabidopsis . Plant Cell 24 : 163 – 177 . Google Scholar CrossRef Search ADS PubMed Caño-Delgado AI , Metzlaff K , Bevan MW . 2000 . The eli1 mutation reveals a link between cell expansion and secondary cell wall formation in Arabidopsis thaliana . Development 127 : 3395 – 3405 . Google Scholar PubMed Caño-Delgado AI , Penfield S , Smith C , Catley M , Bevan M . 2003 . Reduced cellulose synthesis invokes lignification and defense responses in Arabidopsis thaliana . The Plant Journal 34 : 351 – 362 . Google Scholar CrossRef Search ADS PubMed Carpita NC . 2011 . Update on mechanisms of plant cell wall biosynthesis: how plants make cellulose and other (1→4)-β-d-glycans . Plant Physiology 155 : 171 – 184 . Google Scholar CrossRef Search ADS PubMed Chai G , Kong Y , Zhu M , et al. 2015 . Arabidopsis C3H14 and C3H15 have overlapping roles in the regulation of secondary wall thickening and anther development . Journal of Experimental Botany 66 : 2595 – 2609 . Google Scholar CrossRef Search ADS PubMed Chen J , Doyle C , Qi X , Zheng H . 2012 . The endoplasmic reticulum: a social network in plant cells . Journal of Integrative Plant Biology 54 : 840 – 850 . Google Scholar PubMed Clough SJ , Bent AF . 1998 . Floral dip: a simplified method for Agrobacterium-mediated transformation of Arabidopsis thaliana . The Plant Journal 16 : 735 – 743 . Google Scholar CrossRef Search ADS PubMed Cosgrove DJ . 2005 . Growth of the plant cell wall . Nature Reviews Molecular Cell Biology 6 : 850 – 861 . Google Scholar CrossRef Search ADS PubMed Curtis MD , Grossniklaus U . 2003 . A gateway cloning vector set for high-throughput functional analysis of genes in planta . Plant Physiology 133 : 462 – 469 . Google Scholar CrossRef Search ADS PubMed Déjardin A , Laurans F , Arnaud D , Breton C , Pilate G , Leplé JC . 2010 . Wood formation in Angiosperms . Comptes Rendus Biologies 333 : 325 – 334 . Google Scholar CrossRef Search ADS PubMed Denness L , McKenna JF , Segonzac C , et al. 2011 . Cell wall damage-induced lignin biosynthesis is regulated by a reactive oxygen species- and jasmonic acid-dependent process in Arabidopsis . Plant Physiology 156 : 1364 – 1374 . Google Scholar CrossRef Search ADS PubMed Desprez T , Vernhettes S , Fagard M , et al. 2002 . Resistance against herbicide isoxaben and cellulose deficiency caused by distinct mutations in same cellulose synthase isoform CESA6 . Plant Physiology 128 : 482 – 490 . Google Scholar CrossRef Search ADS PubMed Di Sansebastiano GP , Renna L , Piro G , Dalessandro G . 2004 . Stubborn GFPs in Nicotiana tabacum vacuoles . Plant Biosystems 138 : 37 – 42 . Google Scholar CrossRef Search ADS Ellis C , Turner JG . 2001 . The Arabidopsis mutant cev1 has constitutively active jasmonate and ethylene signal pathways and enhanced resistance to pathogens . Plant Cell 13 : 1025 – 1033 . Google Scholar CrossRef Search ADS PubMed Ellis C , Karafyllidis I , Wasternack C , Turner JG . 2002 . The Arabidopsis mutant cev1 links cell wall signaling to jasmonate and ethylene responses . Plant Cell 14 : 1557 – 1566 . Google Scholar CrossRef Search ADS PubMed Endler A , Kesten C , Schneider R , et al. 2015 . A mechanism for sustained cellulose synthesis during salt stress . Cell 162 : 1353 – 1364 . Google Scholar CrossRef Search ADS PubMed Fagard M , Desnos T , Desprez T , et al. 2000 . PROCUSTE1 encodes a cellulose synthase required for normal cell elongation specifically in roots and dark-grown hypocotyls of Arabidopsis . Plant Cell 12 : 2409 – 2424 . Google Scholar CrossRef Search ADS PubMed Geisler DA , Sampathkumar A , Mutwil M , Persson S . 2008 . Laying down the bricks: logistic aspects of cell wall biosynthesis . Current Opinion in Plant Biology 11 : 647 – 652 . Google Scholar CrossRef Search ADS PubMed Gibeaut DM , Carpita NC . 1994 . Biosynthesis of plant cell wall polysaccharides . The FASEB Journal 8 : 904 – 915 . Google Scholar CrossRef Search ADS PubMed Gillmor CS , Lukowitz W , Brininstool G , et al. 2005 . Glycosylphosphatidylinositol-anchored proteins are required for cell wall synthesis and morphogenesis in Arabidopsis . Plant Cell 17 : 1128 – 1140 . Google Scholar CrossRef Search ADS PubMed Green PB . 1962 . Mechanism for plant cellular morphogenesis . Science 138 : 1404 – 1405 . Google Scholar CrossRef Search ADS PubMed Gutierrez R , Lindeboom JJ , Paredez AR , Emons AMC , Ehrhardt DW . 2009 . Arabidopsis cortical microtubules position cellulose synthase delivery to the plasma membrane and interact with cellulose synthase trafficking compartments . Nature Cell Biology 11 : 797 – 806 . Google Scholar CrossRef Search ADS PubMed Hauser MT , Morikami A , Benfey PN . 1995 . Conditional root expansion mutants of Arabidopsis . Development 121 : 1237 – 1252 . Google Scholar PubMed Hoebler C , Barry JL , David A , Delort-Laval J . 1989 . Rapid acid hydrolysis of plant cell wall polysaccharides and simplified quantitative determination of their neutral monosaccharides by gas-liquid chromatography . Journal of Agricultural and Food Chemistry 37 : 360 – 367 . Google Scholar CrossRef Search ADS Höfgen R , Willmitzer L . 1988 . Storage of competent cells for Agrobacterium transformation . Nucleic Acids Research 16 : 9877 . Google Scholar CrossRef Search ADS PubMed Jefferson RA , Kavanagh TA , Bevan MW . 1987 . GUS fusions: beta-glucuronidase as a sensitive and versatile gene fusion marker in higher plants . The EMBO Journal 6 : 3901 – 3907 . Google Scholar PubMed Karimi M , Inzé D , Depicker A . 2002 . GATEWAY vectors for Agrobacterium-mediated plant transformation . Trends in Plant Science 7 : 193 – 195 . Google Scholar CrossRef Search ADS PubMed Kato H , Motomura T , Komeda Y , Saito T , Kato A . 2010 . Overexpression of the NAC transcription factor family gene ANAC036 results in a dwarf phenotype in Arabidopsis thaliana . Journal of Plant Physiology 167 : 571 – 577 . Google Scholar CrossRef Search ADS PubMed Kim WC , Kim JY , Ko JH , Kang H , Kim J , Han KH . 2014 . AtC3H14, a plant-specific tandem CCCH zinc-finger protein, binds to its target mRNAs in a sequence-specific manner and affects cell elongation in Arabidopsis thaliana . The Plant Journal 80 : 772 – 784 . Google Scholar CrossRef Search ADS PubMed Ko JH , Beers EP , Han KH . 2006a. Global comparative transcriptome analysis identifies gene network regulating secondary xylem development in Arabidopsis thaliana . Molecular Genetics and Genomics 276 : 517 – 531 . Google Scholar CrossRef Search ADS PubMed Ko JH , Kim JH , Jayanty SS , Howe GA , Han KH . 2006b. Loss of function of COBRA, a determinant of oriented cell expansion, invokes cellular defence responses in Arabidopsis thaliana . Journal of Experimental Botany 57 : 2923 – 2936 . Google Scholar CrossRef Search ADS PubMed Lane DR , Wiedemeier A , Peng L , et al. 2001 . Temperature-sensitive alleles of RSW2 link the KORRIGAN endo-1,4-beta-glucanase to cellulose synthesis and cytokinesis in Arabidopsis . Plant Physiology 126 : 278 – 288 . Google Scholar CrossRef Search ADS PubMed Lei L , Li S , Bashline L , Gu Y . 2014 . Dissecting the molecular mechanism underlying the intimate relationship between cellulose microfibrils and cortical microtubules . Frontiers in Plant Science 5 : 90 . Google Scholar CrossRef Search ADS PubMed Li S , Lei L , Somerville CR , Gu Y . 2012 . Cellulose synthase interactive protein 1 (CSI1) links microtubules and cellulose synthase complexes . Proceedings of the National Academy of Sciences USA 109 : 185 – 190 . Google Scholar CrossRef Search ADS Lloyd C , Chan J . 2006 . Not so divided: the common basis of plant and animal cell division . Nature Reviews Molecular Cell Biology 7 : 147 – 152 . Google Scholar CrossRef Search ADS PubMed Manfield IW , Orfila C , McCartney L , et al. 2004 . Novel cell wall architecture of isoxaben-habituated Arabidopsis suspension-cultured cells: global transcript profiling and cellular analysis . The Plant Journal 40 : 260 – 275 . Google Scholar CrossRef Search ADS PubMed McFarlane HE , Döring A , Persson S . 2014 . The cell biology of cellulose synthesis . Annual Review of Plant Biology 65 : 69 – 94 . Google Scholar CrossRef Search ADS PubMed Nguyen VP , Cho JS , Choi YI , Lee SW , Han KH , Ko JH . 2016 . Evaluation of a novel promoter from Populus trichocarpa for mature xylem tissue specific gene delivery . Plant Physiology and Biochemistry 104 : 226 – 233 . Google Scholar CrossRef Search ADS PubMed Nicol F , His I , Jauneau A , Vernhettes S , Canut H , Höfte H . 1998 . A plasma membrane-bound putative endo-1,4-β-d-glucanase is required for normal wall assembly and cell elongation in Arabidopsis . The EMBO Journal 17 : 5563 – 5576 . Google Scholar CrossRef Search ADS PubMed Nieuwland J , Feron R , Huisman BA , et al. 2005 . Lipid transfer proteins enhance cell wall extension in tobacco . Plant Cell 17 : 2009 – 2019 . Google Scholar CrossRef Search ADS PubMed Pagant S , Bichet A , Sugimoto K , et al. 2002 . KOBITO1 encodes a novel plasma membrane protein necessary for normal synthesis of cellulose during cell expansion in Arabidopsis . Plant Cell 14 : 2001 – 2013 . Google Scholar CrossRef Search ADS PubMed Paredez AR , Somerville CR , Ehrhardt DW . 2006 . Visualization of cellulose synthase demonstrates functional association with microtubules . Science 312 : 1491 – 1495 . Google Scholar CrossRef Search ADS PubMed Passardi F , Penel C , Dunand C . 2004 . Performing the paradoxical: how plant peroxidases modify the cell wall . Trends in Plant Science 9 : 534 – 540 . Google Scholar CrossRef Search ADS PubMed Roudier F , Schindelman G , DeSalle R , Benfey PN . 2002 . The COBRA family of putative GPI-anchored proteins in Arabidopsis. A new fellowship in expansion . Plant Physiology 130 : 538 – 548 . Google Scholar CrossRef Search ADS PubMed Roudier F , Fernandez AG , Fujita M , et al. 2005 . COBRA, an Arabidopsis extracellular glycosyl-phosphatidyl inositol-anchored protein, specifically controls highly anisotropic expansion through its involvement in cellulose microfibril orientation . Plant Cell 17 : 1749 – 1763 . Google Scholar CrossRef Search ADS PubMed Schindelman G , Morikami A , Jung J , et al. 2001 . COBRA encodes a putative GPI-anchored protein, which is polarly localized and necessary for oriented cell expansion in Arabidopsis . Genes & Development 15 : 1115 – 1127 . Google Scholar CrossRef Search ADS PubMed Shin B , Choi G , Yi H , et al. 2002 . AtMYB21, a gene encoding a flower-specific transcription factor, is regulated by COP1 . The Plant Journal for Cell and Molecular Biology 30 : 23 – 32 . Google Scholar CrossRef Search ADS PubMed Somerville C . 2006 . Cellulose synthesis in higher plants . Annual Review of Cell and Developmental Biology 22 : 53 – 78 . Google Scholar CrossRef Search ADS PubMed Somerville C , Bauer S , Brininstool G , et al. 2004 . Toward a systems approach to understanding plant cell walls . Science 306 : 2206 – 2211 . Google Scholar CrossRef Search ADS PubMed Teasdale RD , Jackson MR . 1996 . Signal-mediated sorting of membrane proteins between the endoplasmic reticulum and the golgi apparatus . Annual Review of Cell and Developmental Biology 12 : 27 – 54 . Google Scholar CrossRef Search ADS PubMed Ueda K , Matsuyama T . 2000 . Rearrangement of cortical microtubules from transverse to oblique or longitudinal in living cells of transgenic Arabidopsis thaliana . Protoplasma 213 : 28 – 38 . Google Scholar CrossRef Search ADS Wang J , Howles PA , Cork AH , Birch RJ , Williamson RE . 2006 . Chimeric proteins suggest that the catalytic and/or C-terminal domains give CesA1 and CesA3 access to their specific sites in the cellulose synthase of primary walls . Plant Physiology 142 : 685 – 695 . Google Scholar CrossRef Search ADS PubMed Wolf S , Hématy K , Höfte H . 2012 . Growth control and cell wall signaling in plants . Annual Review of Plant Biology 63 : 381 – 407 . Google Scholar CrossRef Search ADS PubMed Xiang C , Han P , Lutziger I , Wang K , Oliver DJ . 1999 . A mini binary vector series for plant transformation . Plant Molecular Biology 40 : 711 – 717 . Google Scholar CrossRef Search ADS PubMed Zhang Y , Nikolovski N , Sorieul M , et al. 2016 . Golgi-localized STELLO proteins regulate the assembly and trafficking of cellulose synthase complexes in Arabidopsis . Nature Communication 7 : 11656 . Google Scholar CrossRef Search ADS © The Author(s) 2018. Published by Oxford University Press on behalf of the Annals of Botany Company. All rights reserved. For permissions, please e-mail: [email protected]. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)
Variations in leaf growth parameters within the tree structure of adult Coffea arabica in relation to seasonal growth, water availability and air carbon dioxide concentration2018 Annals of Botany
doi: 10.1093/aob/mcy042pmid: 29659697
Abstract Background and Aims Dynamics in branch and leaf growth parameters, such as the phyllochron, duration of leaf expansion, leaf life span and bud mortality, determine tree architecture and canopy foliage distribution. We aimed to estimate leaf growth parameters in adult Arabica coffee plants based on leaf supporter axis order and position along the vertical profile, considering their modifications related to seasonal growth, air [CO2] and water availability. Methods Growth and mortality of leaves and terminal buds of adult Arabica coffee trees were followed in two independent field experiments in two sub-tropical climate regions of Brazil, Londrina-PR (Cfa) and Jaguariúna-SP (Cwa). In the Cwa climate, coffee trees were grown under a FACE (free air CO2 enrichment) facility, where half of those had been irrigated. Plants were observed at a 15–30 d frequency for 1 year. Leaf growth parameters were estimated on five axes orders and expressed as functions of accumulated thermal time (°Cd per leaf). Key Results The phyllochron and duration of leaf expansion increased with axis order, from the seond to the fourth. The phyllochron and life span during the reduced vegetative seasonal growth were greater than during active growth. It took more thermal time for leaves from the first- to fourth-order axes to expand their blades under irrigation compared with rainfed conditions. The compensation effects of high [CO2] for low water availability were observed on leaf retention on the second and third axes orders, and duration of leaf expansion on the first- and fourth-order axes. The second-degree polynomials modelled leaf growth parameter distribution in the vertical tree profile, and linear regressions modelled the proportion of terminal bud mortality. Conclusions Leaf growth parameters in coffee plants were determined by axis order. The duration of leaf expansion contributed to phyllochron determination. Leaf growth parameters varied according the position of the axis supporter along the vertical profile, suggesting an effect of axes age and micro-environmental light modulations. Axis order, bud mortality, Coffea arabica, dynamic multiscale tree graphs, FACE, growing degree-days, irrigation, leaf expansion, life span, metamer, phyllochron, polynomial function INTRODUCTION Plant development and growth are characterized by the repeated formation, expansion and continuous senescence of the phytomer (metamer), a basic vegetative architectural unit (White, 1979). In this regard, the position and fate of each metamer will define the plant form. The leaf growth parameters (leaf appearance rate, phyllochron, duration of leaf expansion, life span and leaf fall) contribute to the construction of the whole plant architecture (Harmer, 1992) in combination with shoot bud mortality (Alla et al., 2013). The leaf appearance rate represents the number of leaves emerging in a unit of time (Xue et al., 2004), while the phyllochron was originally defined as the time interval between the appearance of two successive leaves on the same shoot (Erickson and Michelini, 1957). It means that one metamer is added to the growing shoot during each phyllochron (Wilhelm and McMaster, 1995). The phyllochron is modified by soil moisture, temperature (Baker et al., 1992; Eggers et al., 2004), water stress (Cutforth et al., 1992), light quality (Zhu et al., 2014), light quantity and daylength (Rosa et al., 2011; Baldissera et al., 2014), or air CO2 supply (Baker et al., 1992). The phyllochron may change with plantation management, e.g. plant population density in potato (Dellai et al., 2005) or sowing dates in oat (Chaves et al., 2017). The effect of temperature on the phyllochron can be calculated as a function of thermal time, expressed in growing degree days [GDDs (°Cd per leaf)] (Garcia-Huidobro et al., 1982) rather than chronological time (days or hours). Leaf expansion is the central process in which plants colonize space, and it is controlled by water and carbon availability (Pantin et al., 2011). It represents the phase of the leaf’s life span where the construction cost is paid (Jurik and Chabot, 1986). Leaf life span is a property of individual leaves (Kikuzawa, 1991). Despite low variation between short and long daylengths in the tropics, leaf fall in diverse species of this region is signalled by photoperiod (Garcia et al., 2017), and its rate in those species is related to water availability (Borchert, 1998). The values of leaf growth parameters and their susceptibility to environmental impacts vary with plant ontogeny (Lemaire et al., 2009; Rosa et al., 2011; Pantin et al., 2011). The phyllochron can change between genotypes in salak palm (Ashari, 2002). In peach, the phyllochron can vary over the growing season (Davidson et al., 2015) and can differ within the same tree, relative to branch morphism and branch position (Kervella et al., 1995). Due to their close relationship to plant architecture, leaf area formation and distribution, leaf growth parameters are widely used in mechanistic models, such as in CANON (Hargreaves and McMaster, 2009), GreenLab (Kang et al., 2008; Jullien et al. 2011), SIRIUS (Jamieson et al., 1998), SORKAM (Narayanan et al., 2014), L-PEACH (Da Silva et al., 2011) and other functional–structural and growth models. Coffea arabica L. (Arabica coffee) is currently the most important tropical tree in the agronomic sense. In Brazil, the largest worldwide coffee producer, Arabica coffee is cultivated mainly in plantations, with density varying from low (approx. 800 plants ha–1) to high (up to 14 000 plants ha–1; Androcioli-Filho, 2002). It is an evergreen, short-day species, characterized by continuous growth. Its architecture is described as a Roux’s model (Hallé et al., 1978), characterized by the existence of branch dimorphism. The main axis is orthotropic, i.e. erect, with opposite leaves in decussate phyllotaxy. The lateral branches are plagiotropic, i.e. tending to a horizontal orientation. The first level of plagiotropic branches are called the ‘primaries’ in the coffee research community (Bustos et al., 2008). According to the concept of plant architecture, ‘primaries’ are considered second-order axes. The second-order axes have great longevity. From these are born plagiotropic axes of the third to fifth order, (i.e. secondaries, tertiaries and quaternaries). Usually, only the dynamics of first-order axes (orthotropic axes) and growth dynamics of sampled second-order axes are observed in order to evaluate the growth of whole adult Arabica coffee trees (Silva et al., 2004; Ghini et al., 2015). The synchronized growth among first- and second-order axes is defined as a linear regression, where the equation parameters vary according to plant age, planting density and arrangement (Matsunaga et al., 2016). About 78 % of the length of second-order axes in adult coffee plants is produced in the warm, rainy season (October to March, active growth), while 22 % is produced during the cold-dry season (April to September, reduced growth) in South-east Brazil (Silva et al., 2004), with flowering taking place from September to December and berry harvesting from May to July (Camargo and Camargo, 2001). Despite the huge agronomical importance of Arabica coffee, the leaf growth parameters and the dynamics in bud mortality have never been quantified in this species. The first hypothesis of our study was that the phyllochron, leaf expansion and leaf life span could vary within the hierarchy of the order of axes in Arabica coffee trees, and those leaf growth parameters could be modified by growth season and climate conditions. The second hypothesis was that air CO2 concentration and water availability would cause modifications of leaf parameters and bud mortality, considering the metabolic and hydraulic roles of those environmental factors. Thus, we aimed to estimate the phyllochron, duration of leaf expansion, and leaf and bud life span in adult field-grown Arabica coffee plants depending on growth season, axis order and position of emerged leaves in coffee trees, with and without water limitations, and under actual and elevated [CO2] to permit accurate leafy branch growth reconstructions. MATERIALS AND METHODS Location and agronomic aspects of experiments Two independent field experiments were carried out on initially 3- and 4-year-old Arabica coffee trees in two sub-tropical Brazilian regions, Southern and South-eastern. The first experiment was conducted at the Agronomic Institute of Paraná (IAPAR), Londrina (23°18ʹS, 51°17ʹW, altitude 620 m, Köpen-Geiger climate type Cfa, sub-tropical without occurrence of a long dry period; Supplementary Data Fig. S1), Paraná. In the Londrina region, the minimum and maximum daylight durations over the year are 10 h 42 min and 13 h 34 min, respectively. Our experiment was managed during the period from October 2013 to November 2014 and considered an Ethiopian accession, ‘E083’. Ethiopian accessions were seeded in the nursery in 2009, and seedlings were transplanted to the field in March 2010, in a planting design 2 m × 0.5 m. The rows were East–West orientated. The limiting factors for Arabica coffee growth in the Cfa climate are defined by low autumn and winter temperatures (Meireles et al., 2009). The second experiment was conducted at Embrapa Environment, Jaguariúna (22º42ʹS, 46º59ʹW, altitude 570 m, Köpen-Geiger climate type Cwa, sub-tropical with occurrence of dry winters; Supplementary Data Fig. S2), São Paulo. In the Jaguariúna region, the minimum and maximum daylight durations over the year are 10 h 43 min and 13 h 32 min, respectively. The limiting factor for Arabica coffee vegetative growth in the Cwa climate of South-east Brazil is defined by low autumn and winter precipitations (Chapa and Rao, 2004). Our experiment was managed during the period from July 2015 to July 2016 and considered the cultivar ‘Catuaí Vermelho IAC 144’. Seedlings of ‘Catuaí Vermelho IAC 144’ were transplanted to the field in March 2011, in a planting design 3.5 m × 0.60 m. The plants were grown under a free air CO2 enrichment (FACE) facility (Ghini et al., 2015). The studied plants were delimited by 10 m diameter octagon plots (rings), with each ring containing four North–South-oriented rows, with a total of 44 plants for each ring. The addition of CO2 to air began on 25 August 2011 and stopped on 30 June 2016. The actual air CO2 concentration (a[CO2]) at the beginning of the experiment was about 390 μL CO2 L–1. The direct injection of pure CO2 allowed the elevation of the air [CO2] to 200 μL CO2 L–1 above the a[CO2] during the daylight hours, which represented a treatment named e[CO2]. The dynamics of [CO2] during our experiment are shown in Supplementary Data Fig. S2C. Considering the existence of a dry period in the Cwa climate, half of the coffee trees under the FACE facility received drip irrigation (IRR), while all other plants were cultivated under rainfed conditions (NI). The irrigation started in October 2015. The need for irrigation was calculated using the soil water balance method to achieve a soil water storage capacity within the plant’s root zone of about 130 mm m–2 month–1. NPK fertilization was carried out with 1000 kg ha–1 year–1 (20:5:15 NPK formulation) split into four applications in the Cfa region, and 1750 kg ha–1 year–1 (20:5:15 NPK formulation) split into three applications in the Cwa region. Zinc sulphate (0.6 %), potassium chlorite (0.5 %) and boric acid (24 kg ha–1 year–1) were also applied. Plant abstraction Coffee plants were observed at a 15–30 d frequency and codified in dynamic multiscale tree graphs (dynamic MTGs; Godin and Caraglio, 1998). The dynamic MTG codification followed the VPlants methodology (Pradal et al., 2008), with three topological scales, i.e. plants, axes and metamers, as in a previous study of coffee plant structure modelling and reconstruction (Rakocevic and Androcioli-Filho, 2010). A coffee metamer is composed of an internode, two leaves and 4–5 buds, which are present in a linear sequence in axil of each of two leaves in a pair (Majerowicz and Söndahl, 2005). The most developed bud of this series can originate a new lateral branch or an inflorescence. In the coding process, we distinguished orthotropic from plagiotropic metamers. North–South cardinal orientation was considered the x-axis, and East–West orientation was considered the y-axis. Four cardinal orientations of branches and leaves were attributed (N, S, E and W) using the planting lines as the reference. In both experiments, the identification of each fifth metamer on decomposed axes was marked with coloured thread, to help the field identification of the metamer position for data collection. The dynamics of various morphological characteristics were recorded: the length of each internode on orthotropic and plagiotropic axes; the position and orientation of followed axes; length, width and orientation of each leaf; and leaf presence on followed axes. From one date to the next, only the differences in observed characters were recorded. In the experiment in the Cfa climate, the leaf growth dynamics were followed on four plants, tagging 12 axes of each of second to fifth axis orders, totalling about 135 axes. Some of the plants did not have all 12 axes of the fifth order. All tagged axes were described at the metamer scale, including the four first-order axes. Plants were observed on 26 dates. In the experiment in the FACE facility in the Cwa climate, plants were codified following the methodology of sampling, considering a detailed description of leaf dynamics of the first-order axes and four second-order axes (each one oriented to one cardinal point) per 50 cm thick layer along a vertical plant profile (Matsunaga et al., 2016). These second-order axes were described at the metamer scale (as detailed in the Cfa experiment), as well as their lateral axes of the third to fifth order. All the other second-order axes were described by their position along the orthotropic trunks, total length of the living branch part (up to the most distant branching), stage of the terminal meristem (active, senescence or dormant) and cardinal orientation. Plants were observed on 22 dates. As the main aim of this study was to analyse leaf growth parameters, the presence of reproductive components has not been followed. In both experiments, flowering occurred from September–October to November–December, but neither the inflorescence position and intensity, nor berry setting were codified, despite the fact that it has recently been shown that the intensity of branch growth is inversely related to fruit load in coffee (Bote and Vos, 2016). Phyllochron, duration of leaf expansion, leaf life span and terminal bud senescence The phyllochron, duration of leaf expansion and life span were expressed as a function of accumulated thermal time [i.e. the sum of growing degree-days (GDDs), °Cd). The growth dynamics of about 400 axes and histories of >4000 leaves were analysed. Daily minimum and maximum air temperatures and pluviometry for the Cfa experiment were provided by the Meteorological System of Paraná, Simepar (Supplementary Data Fig. S1) and by a local meteorological station for the Cwa experiment, including CO2 concentrations (Supplementary Data Fig. S2). GDDs were calculated as GDD = (Tmean – Tb), where Tmean is the mean air temperature calculated as the average of daily minimum and maximum air temperatures, and Tb is the base temperature for each species. The calculated Tb = 10.2 °C (Pezzopane et al., 2008) was assumed for all calculations. The phyllochron was defined as the thermal time, expressed in GDDs, required for the emergence of a leaf. It was calculated as 1/b from the linear regression y = bx + a (Streck et al., 2005), where b is the slope coefficient between the number of produced metamers on the plant axis (y-axis of the equation) and accumulated thermal time (x-axis of the equation). The portions of each leafy coffee axis formed during the active and reduced seasonal periods of growth were distinguished. The regular phyllochron was assumed for all leaves on the same axis during each seasonal growth. The R2 values of regressions were always >0.9. The duration of leaf expansion was calculated as the sum of GDDs from the observed date of metamer appearance to the date of final size acquisition of each leaf in a pair. The life span was calculated as the sum of GDDs from the observed date of metamer appearance to the date of leaf fall, separately for each leaf in a pair. The proportion of preserved leaves on each observed axis was expressed as a percentage of the observed leaves compared with the maximum potential number [(number of leaves ×100)/(number of metamers × 2)]. During the experimental period, some of the followed axes lost the terminal apex and had not been followed in sequence. New axes of different branching orders appeared and were included in observations. In the FACE experiment, the fate of terminal buds was followed, considering second- to fourth-order axes. In coffee plants, the branch is not obligatorily completely senesced when the terminal bud is dead, e.g. when the second-order terminal bud mortality occurred (branch top) due to natural causes or trauma by management. The branch can be functional from the insertion on the orthotropic axis (axis bottom) to the metamer bearing the latest third-order branch toward the branch top, preserving the phloem and xylem continuum, and senescing the whole zone from the last branching to the terminal bud. The average proportion of terminal bud mortality was based on the number of living or dead terminal buds of each axis order on each position over the orthotropic axes, in a population of four plants for each of four treatments. The proportion of terminal bud mortality was determined for each plant observed on the following temporal observation dates: 31 July 2015 (reduced growth), 18 February 2016 (active growth) and 3 July 2016 (active growth). The probability of terminal bud mortality of the second-order axes was modelled as the linear regression related to the position of each second-order axis along the ranks of the orthotropic axis, for the population of four plants for each of four treatments. Data extraction The basic data of morphological traits for definition of leaf life dynamics were extracted from dynamic MTGs under AMAPStudio (Griffon and Coligny, 2014), differentiating the period (active and reduced vegetative growth) of metamer appearance, leaf growth and fall, and order of axis supporter (first to fifth). In dynamic MTGs, each axis had the attributed identification that allowed the extraction of its components relative to sequential dates. The active seasonal growth in the sub-tropical Cfa region occurred in two periods, from October 2013 to March 2014 and from October to November 2014. Those two active periods were pooled to avoid the impact of intercalated reduced growth on phyllochron estimation, leaf expansion rate and life span of active growth. A similar procedure was executed for two periods of reduced growth (from July to September 2015 and from April to July 2016) in the Cwa region experiment, under the FACE facility. Statistical analyses and modelling of leaf and bud life dynamics Analysis of variance (ANOVA) of leaf growth parameters (phyllochron, duration of leaf expansion and leaf life span) were estimated by mixed model fit using the ‘lme’ function in R version 3.3.0 (2017), considering plant repetitions, axes repetitions and branch cardinal orientations as fixed effects and analysing the effects of axis order, seasonal growth, irrigation and [CO2]. The rough distributions of phyllochron, duration of leaf expansion, life span and bud fate over the orthotropic rank and different seasonal growth (reduced vs. active), and the impact of irrigation and [CO2] were modelled under SciLab software (2017). The modelled second-degree polynomial functions were estimated by the ordinary least squares method (Burden and Faires, 2003), accepting R2 >0.5. This method consists of the determination of the second-degree polynomial function P(x) = a2 + bx + c to approximate the distance between real f(x) and calculated values of P(x), by minimizing the truncation error determined by |f(x) – P(x)|2, where the P(x) coefficients are calculated by the linear systems equation solved by Gauss elimination. The polynomial equations are more efficient for programming and application in simulation models, and the second degree was chosen to make an efficient numerical approximation of data sets. RESULTS Axis order and seasonal growth impacts on leaf growth parameters in two climatic regions The coffee plants, grown under rainfed conditions in two sub-tropical regions, showed variations in leaf growth parameters, more in Cwa than in the Cfa region, over the axes orders and growth seasons (Figs 1 and 2; Tables 1 and 2). During reduced seasonal growth, the phyllochron and leaf expansion durations were longer than during the active growth season, while the proportion of preserved leaves was higher during the active seasonal growth. The phyllochron and thermal time needed for leaf expansion increased from the second to the fourth axis order in the Cwa region, while in the Cfa region higher GDDs for one leaf to attain its final dimensions were noted only on second-order axes during the active, but not during the reduced seasonal growth. In both regions, the phyllochrons of first-order axes did not differ between two growth seasons, indicating the continuous carbon investment in the plant height components. The leaf life span was highly impacted by the seasonal growth period in the Cwa region and by axis order in the Cfa region. The following sections serve to illustrate these main points. Table 1. ANOVA P-values for effects of five axes orders and two types of seasonal growth (active and reduced) on leaf growth parameters (GDD, °Cd per leaf) in Coffea arabica estimated for two independent experiments Effect Phyllochron Duration of leaf expansion Leaf life span Order Season Order × season Order Season Order × season Order Season Order × season Order Sub-tropical Cfa region All <0.0001 <0.0001 0.1588 0.0037 0.0036 0.4319 0.0091 0.0933 0.1978 1 0.2571 0.9840 0.8557 2 <0.0001 0.0050 0.1804 3 <0.0001 0.5986 0.3712 4 <0.0001 0.5622 0.0096 5 0.2182 0.6244 0.0650 Sub-tropical Cwa region All 0.0163 <0.0001 0.0648 <0.0001 0.0600 0.9865 0.3079 <0.0001 0.5520 1 0.1758 0.9460 0.4539 2 <0.0001 <0.0001 <0.0001 3 <0.0001 <0.0001 <0.0001 4 0.2023 0.0069 0.0053 Effect Phyllochron Duration of leaf expansion Leaf life span Order Season Order × season Order Season Order × season Order Season Order × season Order Sub-tropical Cfa region All <0.0001 <0.0001 0.1588 0.0037 0.0036 0.4319 0.0091 0.0933 0.1978 1 0.2571 0.9840 0.8557 2 <0.0001 0.0050 0.1804 3 <0.0001 0.5986 0.3712 4 <0.0001 0.5622 0.0096 5 0.2182 0.6244 0.0650 Sub-tropical Cwa region All 0.0163 <0.0001 0.0648 <0.0001 0.0600 0.9865 0.3079 <0.0001 0.5520 1 0.1758 0.9460 0.4539 2 <0.0001 <0.0001 <0.0001 3 <0.0001 <0.0001 <0.0001 4 0.2023 0.0069 0.0053 The Cfa sub-tropical region has occurrences of low minimum daily winter temperatures and the Cwa sub-tropical region has an accentuated dry winter period. P-values <0.05 were considered significant and are marked in bold. View Large Table 1. ANOVA P-values for effects of five axes orders and two types of seasonal growth (active and reduced) on leaf growth parameters (GDD, °Cd per leaf) in Coffea arabica estimated for two independent experiments Effect Phyllochron Duration of leaf expansion Leaf life span Order Season Order × season Order Season Order × season Order Season Order × season Order Sub-tropical Cfa region All <0.0001 <0.0001 0.1588 0.0037 0.0036 0.4319 0.0091 0.0933 0.1978 1 0.2571 0.9840 0.8557 2 <0.0001 0.0050 0.1804 3 <0.0001 0.5986 0.3712 4 <0.0001 0.5622 0.0096 5 0.2182 0.6244 0.0650 Sub-tropical Cwa region All 0.0163 <0.0001 0.0648 <0.0001 0.0600 0.9865 0.3079 <0.0001 0.5520 1 0.1758 0.9460 0.4539 2 <0.0001 <0.0001 <0.0001 3 <0.0001 <0.0001 <0.0001 4 0.2023 0.0069 0.0053 Effect Phyllochron Duration of leaf expansion Leaf life span Order Season Order × season Order Season Order × season Order Season Order × season Order Sub-tropical Cfa region All <0.0001 <0.0001 0.1588 0.0037 0.0036 0.4319 0.0091 0.0933 0.1978 1 0.2571 0.9840 0.8557 2 <0.0001 0.0050 0.1804 3 <0.0001 0.5986 0.3712 4 <0.0001 0.5622 0.0096 5 0.2182 0.6244 0.0650 Sub-tropical Cwa region All 0.0163 <0.0001 0.0648 <0.0001 0.0600 0.9865 0.3079 <0.0001 0.5520 1 0.1758 0.9460 0.4539 2 <0.0001 <0.0001 <0.0001 3 <0.0001 <0.0001 <0.0001 4 0.2023 0.0069 0.0053 The Cfa sub-tropical region has occurrences of low minimum daily winter temperatures and the Cwa sub-tropical region has an accentuated dry winter period. P-values <0.05 were considered significant and are marked in bold. View Large Table 2. ANOVA P-values for effects of axes orders and seasonal growth (active and reduced) on the proportion of preserved leaves per axis of Coffea arabica grown in the Cfa climate region, observed from October 2013 to November 2014 Axes order Effects Order Season Order × season All <0.0001 <0.0001 <0.0001 Order 1 0.0072 Order 2 <0.0001 Order 3 <0.0001 Order 4 <0.0001 Order 5 <0.0001 Axes order Effects Order Season Order × season All <0.0001 <0.0001 <0.0001 Order 1 0.0072 Order 2 <0.0001 Order 3 <0.0001 Order 4 <0.0001 Order 5 <0.0001 P-values <0.05 were considered significant and are marked in bold. View Large Table 2. ANOVA P-values for effects of axes orders and seasonal growth (active and reduced) on the proportion of preserved leaves per axis of Coffea arabica grown in the Cfa climate region, observed from October 2013 to November 2014 Axes order Effects Order Season Order × season All <0.0001 <0.0001 <0.0001 Order 1 0.0072 Order 2 <0.0001 Order 3 <0.0001 Order 4 <0.0001 Order 5 <0.0001 Axes order Effects Order Season Order × season All <0.0001 <0.0001 <0.0001 Order 1 0.0072 Order 2 <0.0001 Order 3 <0.0001 Order 4 <0.0001 Order 5 <0.0001 P-values <0.05 were considered significant and are marked in bold. View Large Fig. 1. View largeDownload slide Mean and s.e. for leaf growth parameters (GDD, °Cd per leaf) in Coffea arabica estimated from two experiments, one in the sub-tropical Cfa region and the other in the sub-tropical Cwa region, during active and reduced seasonal growth. (A) Phyllochron, (B) duration of leaf expansion and (C) leaf life span of five axes orders. Fig. 1. View largeDownload slide Mean and s.e. for leaf growth parameters (GDD, °Cd per leaf) in Coffea arabica estimated from two experiments, one in the sub-tropical Cfa region and the other in the sub-tropical Cwa region, during active and reduced seasonal growth. (A) Phyllochron, (B) duration of leaf expansion and (C) leaf life span of five axes orders. Fig. 2. View largeDownload slide The dynamics of preserved leaf proportion of five axes orders in Coffea arabica plants grown in the sub-tropical Cfa region, observed from October 2013 to November 2014, during active and reduced seasonal growth. Fig. 2. View largeDownload slide The dynamics of preserved leaf proportion of five axes orders in Coffea arabica plants grown in the sub-tropical Cfa region, observed from October 2013 to November 2014, during active and reduced seasonal growth. Phyllochron variations within the tree structure during active and reduced vegetative seasonal growth The phyllochron was roughly 30–40 % (Cfa region) to 300 % (Cwa) higher during the reduced compared with the active seasonal growth, on the second- to fourth-order axes (Fig. 1A; Table 1). Among the five axes orders, the first- and second-order axes exhibited the lowest phyllochron (on average 315 °Cd per leaf during active growth in the Cfa region). The phyllochron increased with increasing axis order. The fourth- and fifth-order axes showed the highest phyllochron (about 660 °Cd per leaf in the active seasonal growth); Fig. 1A. In both regions, the phyllochrons of first-order axes did not differ between seasonal growths. During the active seasonal growth, the phyllochrons of first- and second-order axes did not differ between the two regions; however, they were lower in Cwa than in Cfa for the third- and fourth-order axes. On the other hand, during the reduced seasonal growth, the phyllochrons of second- and fourth-order axes were significantly higher in Cwa than in the Cfa region. Leaf expansion duration and life span variations within the tree structure during active and reduced vegetative seasonal growth In the Cfa region, the duration of leaf expansion of first- and second-order axes was similar (about 730 °Cd per leaf during active growth, and about 670 °Cd per leaf during reduced growth), increasing on higher axes orders to 820 °Cd per leaf (Fig. 1B; Table 1). In the Cwa region, the thermal time needed for leaf blade expansion increased from second to fourth axis order. The first-order axes did not differ in duration of leaf expansion between the active and reduced periods, in both regions (Fig. 1B; Table 1). In the Cfa region, higher GDDs for one leaf to attain its final dimensions were noted on second-order axes during the active, but not during the reduced seasonal growth (Fig. 1B; Table 1), while in the Cwa region the leaf expansion duration was much longer during reduced than during active growth for second- to fourth-order axes (Fig. 1B; Table 1). Throughout the reduced growth in the Cwa region, the average duration of leaf expansion did not differ between the first-, second- and third-order axes, with the highest GDDs recorded on fourth-order axes (about 1085 °Cd per leaf). In the Cwa region, leaves on second-order axes needed the lowest GDD to expand during active growth, but doubled the thermal time for leaf expansion during the reduced growth (Fig. 1B; Table 1). Life span did not differ between the active and reduced seasonal growth in the Cfa region, except on fourth-order axes (Fig. 1C; Table 1), while life span was higher during the reduced growth within the second- to fourth-order axes in the Cwa region (Fig. 1C; Table 1). The longest life span was observed on fourth- and fifth-order axes in Cfa, on average 2250 °Cd per leaf (Fig. 1C), and the shortest on leaves on first- and second-order axes, on average 1900 °Cd per leaf, in the same region. Under the Cwa conditions, the life span was much shorter during the active than during the reduced seasonal growth (average overall value of 1046 for the active and 1972 °Cd per leaf for the reduced period) and did not differ among the axis orders (Table 1; Fig. 1C). Proportion of preserved leaves on axes within the tree structure during active and reduced vegetative seasonal growth The proportion of preserved leaves at each axis compared with the potential number was strongly influenced by the seasonal growth (Fig. 2; Table 2). A higher leaf proportion was preserved during the active than during the reduced seasonal growth (Fig. 2), with the exception of the first-order axes, where this proportion was higher during the reduced seasonal growth (interaction order × period in Table 2). The third-, fourth- and fifth-order axes preserved leaves in a higher proportion than the second-order axes during the whole period of the experiment. The lowest proportion of preserved leaves was found on first-order axes (Fig. 2) related to the greatest total metamer number on this axis order. Modifications on leaf growth parameters with irrigation and elevated air [CO2] Between all observed leaf growth parameters, duration of leaf expansion and leaf retention benefited most from irrigation (Figs 3 and 4; Tables 3 and 4). The leaves of first- to fourth-order axes of irrigated plants cultivated under e[CO2] expanded their blades during a longer thermal time than under a[CO2]. The e[CO2] showed the compensation effects for lower water availability (NI) on leaf expansion duration of first- and fourth-order axes; the e[CO2] permitted a longer thermal time for blade expansion than under a[CO2] under rainfed conditions. Similar compensation effects of e[CO2] under low water availability were observed on leaf retention of the second- and third-order axes in the fifth year of growth under FACE. The life span was not impacted by water and carbon supply during this experimental period, representing the most stable leaf growth parameter. The following details illustrate these main points. Table 3. ANOVA P-values for effects of air [CO2] and irrigation (IRR) treatments on leaf growth parameters (GDD, °Cd per leaf) in Coffea arabica estimated during the seasons of active and reduced growth on four axes orders Effect Phyllochron Duration of leaf expansion Leaf life span Season <0.0001 <0.0001 <0.0001 Order 0.0240 <0.0001 0.0003 Season × order 0.0008 0.0001 0.0017 [CO2] IRR [CO2] × IRR [CO2] IRR [CO2] × IRR [CO2] IRR [CO2] × IRR Order Season of active growth 1 0.0120 0.0408 0.0694 0.1471 0.4308 0.0010 0.2063 0.1111 0.1867 2 0.1012 0.2266 0.0550 <0.0001 <0.0001 0.6596 0.1891 0.2717 0.9305 3 0.8779 0.8186 0.6735 <0.0001 0.0471 0.4211 0.3825 0.1536 0.5745 4 0.4933 0.7549 0.1014 0.7099 0.0170 0.1729 0.3771 0.1727 0.3290 Order Season of reduced growth 1 0.8630 0.8630 0.4656 0.7213 0.5860 0.2116 0.1180 0.0543 0.1641 2 0.6293 0.9427 0.2496 0.3962 0.3372 0.1269 0.5301 0.2290 0.3486 3 0.0001 0.9941 0.1655 0.0045 0.0135 0.3194 0.1273 0.7394 0.4253 4 0.1739 0.7209 0.1940 0.1169 0.0644 0.0011 0.6106 0.8029 0.6578 Effect Phyllochron Duration of leaf expansion Leaf life span Season <0.0001 <0.0001 <0.0001 Order 0.0240 <0.0001 0.0003 Season × order 0.0008 0.0001 0.0017 [CO2] IRR [CO2] × IRR [CO2] IRR [CO2] × IRR [CO2] IRR [CO2] × IRR Order Season of active growth 1 0.0120 0.0408 0.0694 0.1471 0.4308 0.0010 0.2063 0.1111 0.1867 2 0.1012 0.2266 0.0550 <0.0001 <0.0001 0.6596 0.1891 0.2717 0.9305 3 0.8779 0.8186 0.6735 <0.0001 0.0471 0.4211 0.3825 0.1536 0.5745 4 0.4933 0.7549 0.1014 0.7099 0.0170 0.1729 0.3771 0.1727 0.3290 Order Season of reduced growth 1 0.8630 0.8630 0.4656 0.7213 0.5860 0.2116 0.1180 0.0543 0.1641 2 0.6293 0.9427 0.2496 0.3962 0.3372 0.1269 0.5301 0.2290 0.3486 3 0.0001 0.9941 0.1655 0.0045 0.0135 0.3194 0.1273 0.7394 0.4253 4 0.1739 0.7209 0.1940 0.1169 0.0644 0.0011 0.6106 0.8029 0.6578 P-values <0.05 were considered significant and are marked in bold. View Large Table 3. ANOVA P-values for effects of air [CO2] and irrigation (IRR) treatments on leaf growth parameters (GDD, °Cd per leaf) in Coffea arabica estimated during the seasons of active and reduced growth on four axes orders Effect Phyllochron Duration of leaf expansion Leaf life span Season <0.0001 <0.0001 <0.0001 Order 0.0240 <0.0001 0.0003 Season × order 0.0008 0.0001 0.0017 [CO2] IRR [CO2] × IRR [CO2] IRR [CO2] × IRR [CO2] IRR [CO2] × IRR Order Season of active growth 1 0.0120 0.0408 0.0694 0.1471 0.4308 0.0010 0.2063 0.1111 0.1867 2 0.1012 0.2266 0.0550 <0.0001 <0.0001 0.6596 0.1891 0.2717 0.9305 3 0.8779 0.8186 0.6735 <0.0001 0.0471 0.4211 0.3825 0.1536 0.5745 4 0.4933 0.7549 0.1014 0.7099 0.0170 0.1729 0.3771 0.1727 0.3290 Order Season of reduced growth 1 0.8630 0.8630 0.4656 0.7213 0.5860 0.2116 0.1180 0.0543 0.1641 2 0.6293 0.9427 0.2496 0.3962 0.3372 0.1269 0.5301 0.2290 0.3486 3 0.0001 0.9941 0.1655 0.0045 0.0135 0.3194 0.1273 0.7394 0.4253 4 0.1739 0.7209 0.1940 0.1169 0.0644 0.0011 0.6106 0.8029 0.6578 Effect Phyllochron Duration of leaf expansion Leaf life span Season <0.0001 <0.0001 <0.0001 Order 0.0240 <0.0001 0.0003 Season × order 0.0008 0.0001 0.0017 [CO2] IRR [CO2] × IRR [CO2] IRR [CO2] × IRR [CO2] IRR [CO2] × IRR Order Season of active growth 1 0.0120 0.0408 0.0694 0.1471 0.4308 0.0010 0.2063 0.1111 0.1867 2 0.1012 0.2266 0.0550 <0.0001 <0.0001 0.6596 0.1891 0.2717 0.9305 3 0.8779 0.8186 0.6735 <0.0001 0.0471 0.4211 0.3825 0.1536 0.5745 4 0.4933 0.7549 0.1014 0.7099 0.0170 0.1729 0.3771 0.1727 0.3290 Order Season of reduced growth 1 0.8630 0.8630 0.4656 0.7213 0.5860 0.2116 0.1180 0.0543 0.1641 2 0.6293 0.9427 0.2496 0.3962 0.3372 0.1269 0.5301 0.2290 0.3486 3 0.0001 0.9941 0.1655 0.0045 0.0135 0.3194 0.1273 0.7394 0.4253 4 0.1739 0.7209 0.1940 0.1169 0.0644 0.0011 0.6106 0.8029 0.6578 P-values <0.05 were considered significant and are marked in bold. View Large Table 4. ANOVA P-values for effects of [CO2] and irrigation on the dynamics of the proportion of preserved leaves on four axes orders of Coffea arabica, observed from the end of July 2015 to the beginning of July 2016, during active and reduced seasonal growth Effects Order Season Order × season <0.0001 0.3692 0.2290 Effects [CO2] IRR [CO2] × IRR Order Season of active growth Order 1 <0.0001 0.1329 0.1983 Order 2 0.0148 0.0496 <0.0001 Order 3 <0.0001 <0.0001 0.0001 Order 4 0.6062 0.0031 0.2116 Order Season of reduced growth Order 1 0.7131 0.7859 0.2712 Order 2 0.2398 0.2770 0.0011 Order 3 0.0041 <0.0001 0.2932 Order 4 0.2071 0.7361 0.1126 Effects Order Season Order × season <0.0001 0.3692 0.2290 Effects [CO2] IRR [CO2] × IRR Order Season of active growth Order 1 <0.0001 0.1329 0.1983 Order 2 0.0148 0.0496 <0.0001 Order 3 <0.0001 <0.0001 0.0001 Order 4 0.6062 0.0031 0.2116 Order Season of reduced growth Order 1 0.7131 0.7859 0.2712 Order 2 0.2398 0.2770 0.0011 Order 3 0.0041 <0.0001 0.2932 Order 4 0.2071 0.7361 0.1126 P-values <0.05 were considered significant and are marked in bold. View Large Table 4. ANOVA P-values for effects of [CO2] and irrigation on the dynamics of the proportion of preserved leaves on four axes orders of Coffea arabica, observed from the end of July 2015 to the beginning of July 2016, during active and reduced seasonal growth Effects Order Season Order × season <0.0001 0.3692 0.2290 Effects [CO2] IRR [CO2] × IRR Order Season of active growth Order 1 <0.0001 0.1329 0.1983 Order 2 0.0148 0.0496 <0.0001 Order 3 <0.0001 <0.0001 0.0001 Order 4 0.6062 0.0031 0.2116 Order Season of reduced growth Order 1 0.7131 0.7859 0.2712 Order 2 0.2398 0.2770 0.0011 Order 3 0.0041 <0.0001 0.2932 Order 4 0.2071 0.7361 0.1126 Effects Order Season Order × season <0.0001 0.3692 0.2290 Effects [CO2] IRR [CO2] × IRR Order Season of active growth Order 1 <0.0001 0.1329 0.1983 Order 2 0.0148 0.0496 <0.0001 Order 3 <0.0001 <0.0001 0.0001 Order 4 0.6062 0.0031 0.2116 Order Season of reduced growth Order 1 0.7131 0.7859 0.2712 Order 2 0.2398 0.2770 0.0011 Order 3 0.0041 <0.0001 0.2932 Order 4 0.2071 0.7361 0.1126 P-values <0.05 were considered significant and are marked in bold. View Large Fig. 3. View largeDownload slide Mean and s.e. for leaf growth parameters (GDD, °Cd per leaf). (A) Phyllochron, (B) duration of leaf expansion and (C) leaf life span of four axes orders in Coffea arabica, estimated on plants grown under elevated air [CO2] (e[CO2]) and actual [CO2] (a[CO2]), subjected to drip irrigation (IRR) or rainfed conditions (NI) during active and reduced seasonal growth. Fig. 3. View largeDownload slide Mean and s.e. for leaf growth parameters (GDD, °Cd per leaf). (A) Phyllochron, (B) duration of leaf expansion and (C) leaf life span of four axes orders in Coffea arabica, estimated on plants grown under elevated air [CO2] (e[CO2]) and actual [CO2] (a[CO2]), subjected to drip irrigation (IRR) or rainfed conditions (NI) during active and reduced seasonal growth. Fig. 4. View largeDownload slide The impact of air [CO2] and irrigation on dynamics of the proportion of preserved leaves on four axes orders of Coffea arabica, observed from the end of July 2015 to the beginning of July 2016, during active and reduced seasonal growth. Plants were grown under: (A) elevated [CO2] with irrigation, (B) elevated [CO2] and rainfed conditions, (C) actual [CO2] with irrigation and (D) actual [CO2] and rainfed conditions. Fig. 4. View largeDownload slide The impact of air [CO2] and irrigation on dynamics of the proportion of preserved leaves on four axes orders of Coffea arabica, observed from the end of July 2015 to the beginning of July 2016, during active and reduced seasonal growth. Plants were grown under: (A) elevated [CO2] with irrigation, (B) elevated [CO2] and rainfed conditions, (C) actual [CO2] with irrigation and (D) actual [CO2] and rainfed conditions. Phyllochron, leaf expansion and life span modifications under elevated [CO2] and irrigation During active growth, e[CO2] and irrigation decreased the phyllochron of the first-order axes (Fig. 3A; Table 3). Throughout reduced seasonal growth, the third-order axes of plants grown under e[CO2] showed a decreased phyllochron compared with those grown under a[CO2]. It took more thermal time for leaves of first-order axes to expand their blades in rainfed-grown plants under a[CO2] than under e[CO2] during the active seasonal growth (Fig. 3B; Table 3). The leaves on second- and third-order axes needed a longer thermal time to attain their final size under irrigation than under rainfed conditions, in both active and reduced seasonal growth. It took more thermal time for expansion of leaves on second- and third-order axes with CO2 fertilization than in those under a[CO2] during the active seasonal growth, Finally, the fourth-order axes of IRR plants needed a shorter thermal time for leaf expansion than NI plants, throughout active growth, while CO2 fertilization under rainfed conditions shortened the GDDs throughout the reduced growth of fourth-order axes. The leaf life span of the investigated plants had not been modified by water and [CO2] treatments (Table 3; Fig. 3C), but was strongly impacted by growth season and axis order (Table 3; Fig. 3). The proportion of preserved leaves on axes under e[CO2] and irrigation The first-order axes preserved the lowest proportion of leaves among all axes orders (Table 4; Fig. 4), which was related to the highest metamer number with only a few of the newest metamers bearing pairs of leaves. During the active growth, e[CO2] promoted the retention of leaves on first-order axes (Table 4; Fig. 4A–D). The preserved leaf proportions on second-order axes in e[CO2] did not differ between two water treatments during the active growth. The e[CO2] compensated the effect of NI on second-order axes during both the active and reduced growth seasons. During the active seasonal growth, the third-order axes retained more leaves under a[CO2] than under e[CO2]; leaf retention of third-order axes was positively impacted by irrigation only in plants grown under e[CO2], but not under a[CO2]. During the reduced seasonal growth, the third-order axes of plants grown under IRR preserved a higher proportion of leaves, and e[CO2] impacted on a lower proportion of retained leaves than a[CO2]. Throughout the active growth, the highest proportion of leaves on fourth-order axes was preserved on plants grown under e[CO2] and irrigation (Table 4; Fig. 4A–D); during the reduced growth, no impact of water and carbon supply was observed on the proportion of retained leaves on fourth -order axes (Table 4). Modelling the leaf life dynamics in the vertical profile of plants The distributions of leaf growth parameters on second, third and fourth axes order along the orthotropic rank were estimated by second-degree polynomial functions. The complete list of coefficients for equations defining the distribution of the phyllochron, duration of leaf expansion and leaf life span in the vertical profile and the generated R2 for all treatments and axes orders can be seen in Supplementary Data Tables S1–S3. A concave upward parabola modelled the phyllochron distributions along the orthotropic axis during the active growth (Fig. 5A–C) and a concave downward parabola modelled the phyllochrons during the reduced seasonal growth (Fig. 5D–F). This means that during the active growth, the second-order axes in the low and the upper layers were characterized by a higher phyllochron than those in the middle layer (Fig. 5A–C), while the situation among the plant layers was inverted during the reduced seasonal growth. Fig. 5. View largeDownload slide Distribution of phyllochron in the vertical plant profile. (A) Second-, (B) third- and (C) fourth-order axes during active seasonal growth and (D) second-, (E) third- and (F) fourth-order axes during reduced seasonal growth. The phyllochron was expressed in GDD (°Cd per leaf), and four treatments were considered: e[CO2] with irrigation, e[CO2] under rainfed conditions, a[CO2] with irrigation and a[CO2] under rainfed conditions. Fig. 5. View largeDownload slide Distribution of phyllochron in the vertical plant profile. (A) Second-, (B) third- and (C) fourth-order axes during active seasonal growth and (D) second-, (E) third- and (F) fourth-order axes during reduced seasonal growth. The phyllochron was expressed in GDD (°Cd per leaf), and four treatments were considered: e[CO2] with irrigation, e[CO2] under rainfed conditions, a[CO2] with irrigation and a[CO2] under rainfed conditions. A concave upward parabola modelled the leaf expansions along the orthotropic rank during both active and reduced seasonal growth (Fig. 6), with the exception of leaves of third-order axes under e[CO2]–IRR treatment (Fig. 6B, E). This means that leaves needed higher GDDs to expand when localized in low and upper layers than in the middle layer. Only the third-order axes under high CO2 and water supply expanded with higher GDDs in the middle layer than in the low and upper layers. Fig. 6. View largeDownload slide Distribution of the duration of leaf expansion in the vertical plant profile. (A) Second-, (B) third- and (C) fourth-order axes during active seasonal growth and (D) second-, (E) third- and (F) fourth-order axes during reduced seasonal growth. The duration of leaf expansion was expressed in GDD (°Cd per leaf), and four treatments were considered: e[CO2] with irrigation, e[CO2] under rainfed conditions, a[CO2] with irrigation and a[CO2] under rainfed conditions. Fig. 6. View largeDownload slide Distribution of the duration of leaf expansion in the vertical plant profile. (A) Second-, (B) third- and (C) fourth-order axes during active seasonal growth and (D) second-, (E) third- and (F) fourth-order axes during reduced seasonal growth. The duration of leaf expansion was expressed in GDD (°Cd per leaf), and four treatments were considered: e[CO2] with irrigation, e[CO2] under rainfed conditions, a[CO2] with irrigation and a[CO2] under rainfed conditions. A concave upward parabola modelled the distribution of leaf life span of second-order axes in the vertical profile of coffee plants for both active and reduced seasonal growths (Fig. 7A, D). A concave downward parabola modelled the distribution of leaf life span of third-order axes for both the active and reduced seasonal growths (Fig. 7B, E). This means that leaves on second-order axes had a longer life span when localized in low and upper layers than in the middle layer, while the opposite was observed on leaves of third-order axes, characterized by a lower life span in the low and upper layers. A concave upward parabola modelled the distribution of leaf life span only during the active seasonal growth on fourth-order axes under a[CO2] (Fig. 7C), due to lack of a data set for e[CO2] and reduced seasonal growth. It means that leaves on fourth-order axes situated in low layers had a longer life span than those in the middle layer. Fig. 7. View largeDownload slide Distribution of leaf life span in the vertical profile. (A) Second-, (B) third- and (C) fourth-order axes during the active seasonal growth and (D) second- and (E) third-order axes during reduced seasonal growth. Leaf life span was expressed in GDD (°Cd per leaf) and four treatments were considered: e[CO2] with irrigation, e[CO2] under rainfed conditions, a[CO2] with irrigation and a[CO2] under rainfed conditions. Fig. 7. View largeDownload slide Distribution of leaf life span in the vertical profile. (A) Second-, (B) third- and (C) fourth-order axes during the active seasonal growth and (D) second- and (E) third-order axes during reduced seasonal growth. Leaf life span was expressed in GDD (°Cd per leaf) and four treatments were considered: e[CO2] with irrigation, e[CO2] under rainfed conditions, a[CO2] with irrigation and a[CO2] under rainfed conditions. Terminal bud mortality in the vertical profile The probability of terminal bud mortality of second-order axes in the vertical profile of plants grown under two [CO2] treatments and two water regimes was calculated (Fig. 8). In the beginning of the fifth year under the FACE facility, on 31 July 2015, e[CO2] induced a higher incidence of second-order mortality in the low vertical profile layer (Fig. 8A, B) than a[CO2] (Fig. 8C, D), considering the y-axis intersection. Non-zero probability of the incidence of second-order mortality was encountered more often under e[CO2] than under a[CO2] in the upper layer. In all treatments, the maximum second-order bud mortality occurred between 31 July 2015 and 18 February 2016, because of the previous four dry seasons without irrigation. The mortality of second-order terminal buds significantly diminished between 18 February 2016 and 3 July 2016, from a rainy to a dry period. Fig. 8. View largeDownload slide The probability of mortality of terminal buds in the second-order axes in the vertical profile of plants grown under: (A) e[CO2] with irrigation, (B) e[CO2] and rainfed conditions, (C) a[CO2] with irrigation and (D) a[CO2] and rainfed conditions. Three distinct dates were included: 31 July 2015 (reduced growth), 18 February 2016 (active growth) and 3 July 2016 (reduced growth). Fig. 8. View largeDownload slide The probability of mortality of terminal buds in the second-order axes in the vertical profile of plants grown under: (A) e[CO2] with irrigation, (B) e[CO2] and rainfed conditions, (C) a[CO2] with irrigation and (D) a[CO2] and rainfed conditions. Three distinct dates were included: 31 July 2015 (reduced growth), 18 February 2016 (active growth) and 3 July 2016 (reduced growth). The proportion of terminal bud mortality in relation to the total number of axes was calculated for the third and fourth orders (Supplementary Data Table S4). The percentage of third- and fourth-order terminal bud mortality was lower on 31 July 2015, mainly in a[CO2]–NI treatment. This probability increased from July 2015, when the highest rate of bud mortality occurred, mainly under the e[CO2] treatment. The bud mortality proportion decreased on 3 July 2016, because of the appearance of new living third- and fourth-order axes. DISCUSSION Variation in leaf growth parameters over the axes order hierarchy in Arabica coffee trees, and their modification by growth season and climate conditions The initiation of primordia depends on carbon supply coming from adult leaves and from reserves stored in branch structures (Fanwoua et al., 2014). Shade reduces leaf carbon supply (Felippe and Dale, 1973). In Arabica coffee, the fourth- and fifth-order axes exhibited the highest phyllochron. Those high axes orders were found in more self-shaded positions over the plant canopy. High values of their phyllochrons suggest that the intensity of a botanical event, such as leaf appearance, was modified by low irradiance and the light quality change (lower red:far-red ratio), as noted in maize (Birch et al., 1998; Zhu et al., 2014). The dry season (reduced growth season) increased the phyllochron of second- and fourth-order axes in the Cwa region much more than in the Cfa region. In spite of idealized continuous growth of coffee plants (Hallé et al., 1978), Amaral et al. (2006) define the seasonal vegetative growth in Arabica coffee for South-east Brazil, attributing the reduction in branch growth to the minimum air temperature and higher stomatal resistance in autumn/winter (reduced growth) than in spring/summer (active growth) season. ‘In situ’ measured leaf photosynthesis rates diminish during the reduced growth season compared with the active growth season, especially in self-shaded leaves of low plant layers (Rakocevic et al., 2015). Thus, the reduced photosynthesis and high stomatal resistance of self-shaded leaves during the season of drought and low minimum daily temperatures would have an impact on lower carbon balance and higher phyllochron of higher axes orders placed in low plant layers under low light conditions. The leaf growth parameters differed greatly between the two sub-tropical regions, due to climate effects. Two different coffee genotypes, one Ethiopian accession (‘E083’) and ‘Catuaí 144’, used in two experiments, could also contribute to expressed differences. The ‘E083’ accession could have specific genetic responses, as it was the only survivor after one strong frost in 2013, among 63 FAO accessories from Ethiopia and two test cultivars (‘IAPAR 59’ and ‘Catuaí IAC 99’) planted in 2010. Leaves on second-order axes finished their expansion very quickly during the reduced growth season in the Cfa region (‘E083’) and attained a smaller size than during active growth (data not shown), while in the Cwa region the leaf blades of ‘Catuaí IAC 144’ needed comparatively doubled thermal time to attain their final size during the reduced seasonal growth. This means that in the Cfa region the acclimation to temperatures occurred within the same tree of wild-type ‘E083’, in duration of leaf expansion on leading second-order axes, as a response to autumn/winter dry conditions and low minimum daily temperatures. The acclimation to temperatures of leaf growth indicators is observed in maize, comparing some cultivating fields under tropical and temperate climates (Birch et al., 1998). An increase in leaf expansion duration could shift the time of leaf emergence, and this could contribute to the phyllochron increase (Zhu et al., 2014). The values of the phyllochron and expansion time were very similar in coffee plants, which indicates that leaf expansion contributed to the determination of the phyllochron. The leaf life span of all axes orders was similar and much higher during the reduced than during the active seasonal growth in the Cwa climate, while small differences between the seasons and the highest life span were noted on fourth- and fifth-order axes in the Cfa climate. High values for life span on high-order axes indicate that it was longer in self-shaded layers, as was observed for the phyllochron. The proportion of preserved leaves per axis also diminished during the reduced compared with the active seasonal growth in Cfa. The proportion of preserved leaves depended on axis order, local shading conditions and metamer number per axis; the first-order axes had about 70–80 metamers at the end of the experiment, the second-order axes were generally shorter, dependent on their insertion rank on the orthotropic axis, while the third- to fifth-order axes had an even lower number of metamers. Finally, the general picture of the proportion of preserved leaves within the plant structure showed an increase with increased axis order, first < second < third = fourth = fifth. The air [CO2] and water availability modify leaf growth parameters and mortality of buds on the terminal axes During the active seasonal growth, the e[CO2] diminished the phyllochron of the first-order coffee axes, while during reduced growth the e[CO2] decreased the phyllochron of third-order axes. High [CO2] diminishes the leaf emission rate in rice (Baker et al., 1992), while doubling the ambient CO2 concentration has no effect on the plastochron in young peach trees (Davidson et al., 2016). The lack of any responses in peach could be attributed to the short-term experiment of only 38 d in one tree species. The coffee leaf expansion was more sensitive to carbon and water variations than the phyllochron. This sensitivity can be related to the fact that during the expansion, new leaves depend on carbon that is assimilated and imported from the older leaves (Fanwoua et al., 2014). Carbon may also be supplied from starch reserves stored in branch structures. This carbon is especially important for initial axis growth and for buffering any deficit in leaf carbon supply (De Schepper et al., 2013). So, leaf expansion is dependent on carbon fluctuations, and vulnerable to water deficit (Van Volkenburgh, 1999). CO2 fertilization showed the compensation effect for a lower water supply in leaf expansion duration of the first- and fourth-order axes allowing a shorter thermal time for blade expansion than under a[CO2]. During the individual leaf ontogeny, the predominant control of leaf expansion switches from metabolic, i.e. carbon and light availability, to hydraulic, i.e. water availability (Pantin et al., 2011). It may be deduced that under CO2 fertilization, the duration of expansion of leaves on first-order axes, localized under high light availability, was shortened during an initial metabolic control, while the expansion of fourth-order leaves, localized in low and shaded layers, was shortened over the later period of expansion, characterized by predominant control of water availability. The life span of Arabica coffee leaves was not impacted by external water and carbon variations but increased with axis order and showed higher values during the reduced seasonal growth. A longer life span will permit longer assimilation of the same leaf and will decrease the total plant costs. Leaf longevity is related to net photosynthetic rate and construction costs of the leaf (Kikuzawa, 1991). By increasing the life span by one-third, the profit from invested construction costs in leaves can increase by about 80 % in low-light leaves and by about 50 % in high-light leaves in strawberry (Jurik and Chabot, 1986). In this sense, the generally longer leaf life span of second-order coffee axes, together with longer expansion of their blades in the low and upper layers compared with the middle layer, could reflect the coffee tree strategy of higher profit of high- and low-light leaves of this axis order, especially during the reduced growth season. Generally, a decrease in branching intensity of any plant species corresponds to the tuning of growth to resource availability, at the whole-plant scale (Diepenbrock, 2000). In regions with dry winters, the usual drops of leaves of fourth- and fifth-order axes are expected as normal botanic events during the winter in coffee (Camargo and Camargo, 2001), which was expressed through their high terminal bud mortality, and high values of the phyllochron in this season. Berry load and fruit maturing on productive axes (Bote and Vos, 2016), which happen at the beginning of reduced seasonal growth, could be a biological factor that induced high values of phyllochron and leaf drop. In C. canephora, greater leaf retention and longevity might be partially associated with improved growth of irrigated plants at the end of a dry season (Silveira and Carvalho, 1996), but in our experiment the leaf longevity was not impacted by irrigation. The new branches appeared at the end of reduced growth and at the beginning of active growth, which explained the elevated proportion of leaves on second- to fourth-order axes. The first- and second-order axes retained more leaves during the active than during the reduced growth, while the third- and fourth-order axes retained more leaves during the reduced seasonal growth. The latter could be a response of plant rejuvenation after fruit collection (in May and the middle of June), when emergence of new third- and fourth-order axes occurred with 100 % leaf presence on a branch, together with the lower phyllochron of those axes orders under e[CO2]. The compensation effects of e[CO2] for low water availability was observed as leaf retention of the second- and third-order axes during the active and reduced seasonal growths. The mitigation effect of e[CO2] under drought in the reduced growth season is observed in the leaf assimilation rate (Ghini et al., 2015; Rakocevic et al., 2018a). Our data could help to paint a more complete picture of possible morphophysiological responses in Arabica coffee under global climate changes. The phyllochron, duration of leaf expansion and life span of second-order axes were higher in low and upper layers compared with the middle layer during active vegetative growth. Variations in blade elongation duration are also shown to be bell-shaped over the phytomer ranks of maize plants (Zhu et al., 2014). The bell-shaped leaf growth parameter distributions over the orthotropic rank in coffee plants suggest the age effects of the axes and environmental light modulations. The responses in coffee leaf growth parameters could be generalized and applied to younger and older plants, but the modulations with age effects could be expected (Matsunaga et al., 2016). In young plants, less complex synchronization of leaf growth parameters between axes orders could be expected, when only two axes orders are present (Rakocevic et al., 2018b) that share the same phyllochron (Matsunaga et al., 2016). Our intention is to include the equations regarding leaf growth parameters with water and carbon availability during the seasonal growth in the CoffePlant3D software (Matsunaga et al., 2016). This software provides a reconstruction and visualization of the structure of several varieties of Arabica coffee, using an available data set modelled by mathematical and statistical methods and random algorithms. The inclusion of the equation parameters can provide a first step toward a functional–structural plant modelling of coffee tree growth, similarly to models such as L-PEACH (Da Silva et al., 2011), which considered water transport, carbohydrate allocation and physiological functions at the peach organ level. The integration of polynomial functions into CoffeePlant3D could be facilitated by their generalization as second-order functions, changing only the coefficients, according to the leaf growth parameters, which can be stored in a relational database with GDD information. The phyllochron value resulting from processing of the equation would help in decision-making about metamer emission, the duration of leaf expansion would be associated with the rate of increase in the leaf area considering its initial and final size, while leaf life span would be linked to the leaf drop and the GDD information stored in a database. This application would facilitate the generation of 3-D reconstructions of coffee plants in different growing seasons, without requiring manual, daily and exhaustive measurements. CONCLUSIONS The novelty of this work is that the phyllochron, leaf expansion and leaf life span in Arabica coffee plants vary within the tree structure and axes orders and that they were seasonally modified by the environment. This work could raise the question of the impact of branching order in other species. The seasonal environmental impacts on leaf growth parameters distinguished among regions of coffee production could be attributed to differences in minimum daily autumn/winter temperatures and water availability. The e[CO2] showed the compensation of the low water availability effect on duration of leaf expansion and leaf retention, while leaf life span was the least sensitive to carbon and water variations among the observed leaf growth parameters. Furthermore, the probability of occurrence of a terminal bud mortality event in the low layer was higher under e[CO2] than under a[CO2]. On the other hand, in the upper layer the probability of non-occurrence of a terminal bud mortality event was higher under e[CO2] than under a[CO2]. Investments in leaves under high carbon and water availability promoted the leaf permanence of the second- and fourth-order axes in low and upper layers during active seasonal growth, while during the dry period higher dynamics in leaf growth parameters were observed in the upper layer. Those findings could help to paint a more complete picture of possible morphophysiological responses in Arabica coffee under global climate changes. The dynamics and distribution of plant leaf area of Arabica coffee were shown to be the result of synchronization in the phyllochron, duration of leaf expansion, leaf life span, emergence of axes and senescence within the branching structure. SUPPLEMENTARY DATA Supplementary data are available online at https://academic.oup.com/aob and consist of the following. Figure S1: meteorological data in the Cfa region. Figure S2: meteorological data in the Cwa region. Table S1: coefficients for equations defining the distribution of phyllochron in the vertical profile and the generated R2 for all treatments and axes orders. Table S2: coefficients for equations defining the distribution of leaf expansion duration in the vertical profile and the generated R2 for all treatments and axes orders. Table S3: coefficients for equations defining the distribution of leaf life span in the vertical profile and the generated R2 for all treatments and axes orders. Table S4: proportion of terminal bud mortality of third- and fourth-order axes in the vertical profile related to the total number of analysed axes of respective order. ACKNOWLEDGEMENTS This work was supported by the Consórcio Pesquisa Café [02.09.20.008.00.03 and 02.13.02.042.00.03], Agronomic Institute of Paraná (Londrina-PR) and Embrapa Environment (Jaguariúna-SP). M.R. thanks the Consórcio Pesquisa Café for a fellowship. We thank undergraduate students Laís Escorcio Correia and Carolina Antônio Alvim for their technical help, and the editors and three anonymous referees for all helpful criticisms and suggestions. LITERATURE CITED Alla AQ , Camarero JJ , Palacio S , Montserrat-Martí G . 2013 . Revisiting the fate of buds: size and position drive bud mortality and bursting in two coexisting Mediterranean Quercus species with contrasting leaf habit . Trees 27 : 1375 – 1386 . Google Scholar CrossRef Search ADS Amaral JAT , Rena AB , Amaral do JFT . 2006 . Crescimento vegetativo sazonal do cafeeiro e sua relação com fotoperíodo, frutificação, resistência estomática e fotossíntese . Pesquisa Agropecuaria Brasileira 41 : 377 – 384 . Google Scholar CrossRef Search ADS Androcioli-Filho A . 2002 . Café adensado – Espaçamentos e cuidados no manejo da lavoura, Circular No. 121 . Londrina : IAPAR . Ashari S . 2002 . On the agronomy and botany of Salak (Salacca zalacca) . PhD Thesis, Wageningen University , The Netherlands . Baker JT , Allen LH Jr , Boote KJ . 1992 . Temperature effects on rice at elevated CO2 concentration . Journal of Experimental Botany 43 : 959 – 964 . Google Scholar CrossRef Search ADS Baldissera TC , Frak E , Carvalho PF , Louarn G . 2014 . Plant development controls leaf area expansion in alfalfa plants competing for light . Annals of Botany 113 : 145 – 157 . Google Scholar CrossRef Search ADS PubMed Birch CJ , Vos J , Kiniry J , Bos HJ , Elings A . 1998 . Phyllochron responds to acclimation to temperature and irradiance in maize . Field Crops Research 59 : 187 – 200 . Google Scholar CrossRef Search ADS Borchert R . 1998 . Responses of tropical trees to rainfall seasonality and its long-term changes . Climatic Change 39 : 381 – 393 . Google Scholar CrossRef Search ADS Bote AD , Vos J . 2016 . Branch growth dynamics, photosynthesis, yield and bean size distribution in response to fruit load manipulation in coffee trees . Trees 30 : 1275 – 1285 . Google Scholar CrossRef Search ADS Burden RL , Faires JD . 2003 . Numerical analysis . Pioneira Thomson Learning . Bustos PA , Pohlan HAJ , Schulz M . 2008 . Interaction between Coffee (Coffea arabica L.) and intercropped herbs under field conditions in the Sierra Norte of Puebla, Mexico . Journal of Agriculture and Rural Development in the Tropics and Subtropics 109 : 85 – 93 . Camargo AP , Camargo MBP . 2001 . Definição e esquematização das fases fenológicas do cafeeiro arábica nas condições tropicais do Brasil . Bragantia 60 : 65 – 68 . Google Scholar CrossRef Search ADS Chapa SR , Rao VB . 2004 . Annual cycle of precipitation and moisture characteristics over Brazil . Available at http://mtcm15.sid.inpe.br/col/cptec.inpe.br/walmeida/2004/10.14.16. Chaves GG , Cargnelutti AF , Alves BM , et al. 2017 . Phyllochron and leaf appearance rate in oat . Bragantia 76 : 73 – 81 . http://www.scielo.br/pdf/brag/v76n1/0006-8705-brag-1678-4499090.pdf Google Scholar CrossRef Search ADS Cutforth HW , Jame YW , Jefferson PG . 1992 . Effect of temperature, vernalization and water stress on phyllochron and final main-stem leaf number of HY320 and Neepawa spring wheats . Canadian Journal of Plant Science 72 : 1141 – 1151 . Google Scholar CrossRef Search ADS Da Silva D , Favreau R , Auzmendi I , Dejong TM . 2011 . Linking water stress effects on carbon partitioning by introducing a xylem circuit into L-PEACH . Annals of Botany 41 : 433 – 447 . Davidson AM , Da Silva D , Quintana B , DeJong TM . 2015 . The phyllochron of Prunus persica shoots is relatively constant under controlled growth conditions but seasonally increases in the field in ways unrelated to temperature or radiation . Scientia Horticulturae 184 : 106 – 113 . Google Scholar CrossRef Search ADS Davidson AM , Da Silva D , Saa S , Mann P , DeJong TM . 2016 . The influence of elevated CO2 on the photosynthesis, carbohydrate status, and plastochron of young peach (Prunus persica) trees . Horticulture, Environment and Biotechnology 57 : 364 – 370 . Google Scholar CrossRef Search ADS Dellai J , Trentin G , Bisognin DA , Streck NA . 2005 . Phyllochron at different plant densities in potato . Ciência Rural 35 : 1269 – 1274 . Google Scholar CrossRef Search ADS De Schepper V , De Swaef T , Bauweraerts I , Steppe K . 2013 . Phloem transport: a review of mechanisms and controls . Journal of Experimental Botany 64 : 4839 – 4850 . Google Scholar CrossRef Search ADS PubMed Diepenbrock W . 2000 . Yield analysis of winter oilseed rape (Brassica napus L.): a review . Field Crops Research 67 : 35 – 49 . Google Scholar CrossRef Search ADS Eggers L , Cadenazzi M , Boldrini II . 2004 . Phyllochron of Paspalum notatum Fl. and Coelorhachis selloana (Hack.) Camus in natural pasture . Scientia Agricola 61 : 353 – 357 . Google Scholar CrossRef Search ADS Erickson RO , Michelini FJ . 1957 . The plastochron index . American Journal of Botany 44 : 297 – 305 . Google Scholar CrossRef Search ADS Fanwoua J , Bairam E , Delaire M , Buck-Sorlin G . 2014 . The role of branch architecture in assimilate production and partitioning: the example of apple (Malus domestica) . Frontiers in Plant Science 5 : 338 . doi: 10.3389/fpls.2014.00338 Google Scholar CrossRef Search ADS PubMed Felippe GM , Dale JE . 1973 . Effects of shading the first leaf of barley plants on growth and carbon nutrition of the stem apex . Annals of Botany 37 : 45 – 56 . Google Scholar CrossRef Search ADS Garcia LC , Barros FV , Lemos-Filho JP . 2017 . Environmental drivers on leaf phenology of ironstone outcrops species under seasonal climate . Annals of the Brazilian Academy of Sciences 89 : 131 – 143 . Google Scholar CrossRef Search ADS Garcia-Huidobro J , Monteith JL , Squire GR . 1982 . Time, temperature and germination of pearl millet (Pennisetum typhoides S & H). I. Constant temperature . Journal of Experimental Botany 33 : 288 – 296 . Google Scholar CrossRef Search ADS Ghini R , Torre-Neto A , Dentzien AFM , et al. 2015 . Coffee growth, pest and yield responses to free-air CO2 enrichment . Climatic Change 132 : 307 – 320 . Google Scholar CrossRef Search ADS Godin C , Caraglio Y . 1998 . A multiscale model of plant topological structures . Journal of Theoretical Biology 191 : 1 – 46 . Google Scholar CrossRef Search ADS PubMed Griffon S , de Coligny F . 2014 . AMAPstudio: an editing and simulation software suite for plants architecture modeling . Ecological Modelling 290 : 3 – 10 . Google Scholar CrossRef Search ADS Hallé F , Oldeman RAA , Tomlinson PB . 1978 . Tropical trees and forests: an architectural analysis . Berlin : Springer . Hargreaves JNG , McMaster GS . 2009 . CANON: a canonical composition for building plant shoots from the bottom up . In: Cao W , White JW , Wang E , eds. Crop modeling and decision support . Dordrecht : Springer , 59 – 70 . Google Scholar CrossRef Search ADS Harmer R . 1992 . Relationships between shoot length, bud number and branch production in Quercus petraea (Matt.). Liebl . Forestry 65 : 61 – 72 . Google Scholar CrossRef Search ADS Jamieson PD , Semenov MA , Brooking IR , Francis GS . 1998 . Sirius: a mechanistic model of wheat response to environmental variation . European Journal of Agronomy 8 : 161 – 179 . Google Scholar CrossRef Search ADS Jullien A , Mathieu A , Allirand JM , et al. 2011 . Characterization of the interactions between architecture and source–sink relationships in winter oilseed rape (Brassica napus) using the GreenLab model . Annals of Botany 107 : 765 – 779 . Google Scholar CrossRef Search ADS PubMed Jurik TW , Chabot BF . 1986 . Leaf dynamics and profitability in wild strawberries . Oecologia 69 : 296 – 304 . Google Scholar CrossRef Search ADS PubMed Kang MZ , Cournède PH , de Reffye P , Auclair D , Hu BG . 2008 . Analytical study of a stochastic plant growth model: application to the GreenLab model . Mathematics and Computers in Simulation 78 : 57 – 75 . Google Scholar CrossRef Search ADS Kervella J , Pagès L , Génard M . 1995 . Growth context and fate of axillary meristems of young peach trees. Influence of parent shoot growth characteristics and of emergence date . Annals of Botany 76 : 559 – 567 . Google Scholar CrossRef Search ADS Kikuzawa K . 1991 . A cost–benefit analysis of leaf habit and leaf longevity of trees and their geographical pattern . American Naturalist 138 : 1250 – 1263 . Google Scholar CrossRef Search ADS Lemaire S , Maupas F , Cournède PH , de Reffye P . 2009 . A morphogenetic crop model for sugar-beet (Beta vulgaris L.) . In: Cao W , White JW , Wang E , eds. Crop modeling and decision support . Dordrecht : Springer , 116 – 129 . Google Scholar CrossRef Search ADS Majerowicz N , Söndahl MR . 2005 . Induction and differentiation of reproductive buds in Coffeea arabica L . Brazilian Journal of Plant Physiology 17 : 247 – 254 . Google Scholar CrossRef Search ADS Matsunaga FT , Tosti JB , Androcioli-Filho A , Brancher JD , Costes E , Rakocevic M . 2016 . Strategies to reconstruct 3D Coffea arabica L. plant structure . SpringerPlus 5 : 2075 . doi: 10.1186/s40064-016-3762-4 . Google Scholar CrossRef Search ADS PubMed Meireles EJL , Camargo MBP , Pezzopane JRM , et al. 2009 . Fenologia do Cafeeiro: condições agrometeorológicas e balanço hídrico do ano agrícola 2004–2005 . Embrapa Informação Tecnológica . Narayanan S , Aiken RM , Prasad PVV , Xin Z , Paul G , Yu J . 2014 . A simple quantitative model to predict leaf area index in sorghum . Agronomy Journal 106 : 219 – 226 . Google Scholar CrossRef Search ADS Pantin F , Simonneau T , Rolland G , Dauzat M , Muller B . 2011 . Control of leaf expansion: a developmental switch from metabolics to hydraulics . Plant Physiology 156 : 803 – 815 . Google Scholar CrossRef Search ADS PubMed Pezzopane JRM , Júnior MJP , de Camargo MBP , Fazuoli LC . 2008 . Heat requirements of Mundo Novo coffee for the flowering-harvest phenological stage . Ciência e Agrotecnologia 32 : 1781 – 1786 . Google Scholar CrossRef Search ADS Pradal C , Boudon F , Nouguier C , Chopard J , Godin C . 2008 . PlantGL: a Python-based geometric library for 3D plant modelling at different scales . Graphical Models 71 : 1 – 21 . Google Scholar CrossRef Search ADS Rakocevic M , Androcioli-Filho A . 2010 . Morphophysiological characteristics of Coffea arabica L in different arrangements: lessons from a 3D virtual plant approach . Coffee Science 5 : 154 – 166 . Rakocevic M , Scholz MBS , Charmetant P . 2015 . Leaf photosynthesis in four coffee genotypes as response to the irrigation during the biennial period . In: IX Simpósio de Pesquisa dos Cafés do Brasil , Curitiba-PR , 89 : 1 – 6 . Rakocevic M , Ribeiro RV , Marchiori PER , Filizola HF , Batista ER . 2018a. Structural and functional changes in coffee trees after 4 years under free air CO2 enrichment . Annals of Botany 121 : 1065 – 1078 . Google Scholar CrossRef Search ADS PubMed Rakocevic M , Scholz MBS , Kitzberger CSG . 2018b. Berry distributions on coffee trees cultivated under high densities modulate the chemical composition of respective coffee beans during one biannual cycle . International Journal of Fruit Science 18 : 117 – 137 . Google Scholar CrossRef Search ADS Rosa HT , Walter LC , Streck NA , Andriolo JL , da Silva MR , Langner JA . 2011 . Base temperature for leaf appearance and phyllochron of selected strawberry cultivars in a subtropical environment . Bragantia 70 : 939 – 945 . Google Scholar CrossRef Search ADS Scilab Enterprises . 2017 . SciLab, open source software for numerical computation . Last accessed 15 May 2017 . Silva EA , DaMatta FM , Ducatti C , Regazzi AJ , Barros RS . 2004 . Seasonal changes in vegetative growth and photosynthesis of arabica coffee trees . Field Crops Research 89 : 349 – 357 . Google Scholar CrossRef Search ADS Silveira JSM , Carvalho CHS . 1996 . Efeito da época de irrigação sobre o crescimento do ramo plagiotrópico e da longevidade foliar do café conilon . In: Anais de 22º Congresso Brasileiro de Pesquisas Cafeeiras, PROCAFÉ, Água de Lindóia , 99 – 100 . Streck NA , Tibola T , Lago I , et al. 2005 . Estimating the plastochron in muskmelon (Cucumis melo L.) grown inside plastic greenhouse at different planting dates . Ciência Rural 35 : 1448 – 1450 . Google Scholar CrossRef Search ADS Van Volkenburgh E . 1999 . Leaf expansion – an integrating plant behavior . Plant, Cell and Environment 22 : 1463 – 147 . Google Scholar CrossRef Search ADS White J . 1979 . The plant as a metapopulation . Annual Review of Ecology and Systematics 10 : 109 – 145 . Google Scholar CrossRef Search ADS Wilhelm WW , McMaster GS . 1995 . Importance of the phyllochron in studying development and growth in grasses . Crop Science 35 : 1 – 3 . Google Scholar CrossRef Search ADS Xue Q , Weiss A , Baenziger PS . 2004 . Predicting leaf appearance in field-grown winter wheat: evaluating linear and non-linear models . Ecological Modelling 175 : 261 – 270 . Google Scholar CrossRef Search ADS Zhu J , Vos J , Werf van der W , Putten van der PEL , Evers JB . 2014 . Early competition shapes maize whole-plant development in mixed stands . Journal of Experimental Botany 65 : 641 – 653 . Google Scholar CrossRef Search ADS PubMed © The Author(s) 2018. Published by Oxford University Press on behalf of the Annals of Botany Company. All rights reserved. For permissions, please e-mail: [email protected]. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)
Hornwort stomata do not respond actively to exogenous and environmental cues2018 Annals of Botany
doi: 10.1093/aob/mcy045pmid: 29897395
Abstract Backgrounds and Aims Because stomata in bryophytes occur on sporangia, they are subject to different developmental and evolutionary constraints from those on leaves of tracheophytes. No conclusive experimental evidence exists on the responses of hornwort stomata to exogenous stimulation. Methods Responses of hornwort stomata to abscisic acid (ABA), desiccation, darkness and plasmolysis were compared with those in tracheophyte leaves. Potassium ion concentrations in the guard cells and adjacent cells were analysed by X-ray microanalysis, and the ontogeny of the sporophytic intercellular spaces was compared with those of tracheophytes by cryo-scanning electron microscopy. Key Results The apertures in hornwort stomata open early in development and thereafter remain open. In hornworts, the experimental treatments, based on measurements of >9000 stomata, produced only a slight reduction in aperture dimensions after desiccation and plasmolysis, and no changes following ABA treatments and darkness. In tracheophytes, all these treatments resulted in complete stomatal closure. Potassium concentrations are similar in hornwort guard cells and epidermal cells under all treatments at all times. The small changes in hornwort stomatal dimensions in response to desiccation and plasmolysis are probably mechanical and/or stress responses of all the epidermal and spongy chlorophyllose cells, affecting the guard cells. In contrast to their nascent gas-filled counterparts across tracheophytes, sporophytic intercellular spaces in hornworts are initially liquid filled. Conclusions Our experiments demonstrate a lack of physiological regulation of opening and closing of stomata in hornworts compared with tracheophytes, and support accumulating developmental and structural evidence that stomata in hornworts are primarily involved in sporophyte desiccation and spore discharge rather than the regulation of photosynthesis-related gaseous exchange. Our results run counter to the notion of the early acquisition of active control of stomatal movements in bryophytes as proposed from previous experiments on mosses. Abscisic acid, dehiscence, evolution of land plants, potassium flux, plasmolysis, spore discharge, stomata, sporangia, X-ray microanalysis INTRODUCTION Stomata are a key innovation that enabled freshwater algae to colonize Earth’s land masses >500 million years ago (Morris et al., 2018). A long-held hypothesis posits that stomata evolved once in the common ancestor of land plants and that their role and regulation are conserved across all land plant lineages (Raven, 2002). However, this major assumption has been based on scarce or no structural and functional evidence for phylogenetically key groups at the base of the land plant tree: bryophytes (hornworts, mosses), the most basal clade, and pteridophytes (lycophytes, ferns), the sister group to the seed plants. Recent studies focusing on these groups have presented conflicting evidence either supporting the evolutionary theory of early acquisition of active stomatal control (Chater et al., 2011, 2013, 2017; Ruszala et al., 2011; Xu et al., 2016; Hõrak et al., 2017; Merilo et al., 2017) or rejecting it in favour of the alternative hypothesis of gradual acquisition of active control mechanisms (Brodribb et al., 2009; Brodribb and McAdam, 2011, 2013, 2017; Pressel et al., 2014; Field et al., 2015; Villarreal and Renzaglia, 2015; Sussmilch et al., 2017; Duckett and Pressel, 2017a; Merced and Renzaglia, 2017; Renzaglia et al., 2017). Following recent demonstrations of key stomatal developmental stages unique to bryophytes (Pressel et al., 2014; Villarreal and Renzaglia, 2015; Merced and Renzaglia, 2017), here we investigate stomatal behaviour in hornworts – the bryophyte clade currently considered sister to the vascular plants (Liu et al., 2014; Qiu et al., 2006, 2007). Stomata occur in all embryophyte groups except liverworts. Unlike tracheophytes where stomata abound on vegetative organs, especially leaves, of the sporophyte, those in bryophytes are restricted to the sporangia. As monosporangiates, bryophytes are characterized by the occurrence of a single sporangium on a matrotrophic sporophyte (Ligrone et al., 1993). Undoubtedly, stomata on these sporangia have experienced a suite of evolutionary constraints independent from those in tracheophytes, potentially leading to fundamental differences in structure and function across these groups. The hornwort sporophyte is unparalleled in any living or extinct group in that it is an elongating sporangium with asynchronous meiosis. Such a system requires precise co-ordination of all portions of the sporophyte to ensure protection of all stages, from meristematic cells to dehiscence of mature spores. Much like the apical meristem of vegetative organs, it affords a dynamic experimental system in which to examine spatial and temporal changes in a static structure. It is perplexing therefore that, in spite of the multitude of physiological studies on tracheophyte stomata, no conclusive analysis has been conducted on the regulation of stomatal function in hornworts. Conflicting reports exist on regulation of movement in hornwort stomata. Responses to environmental stimuli and exogenous abscisic acid (ABA) were reported by Bopp and Werner (1993), Hartung (2010) and Hartung et al. (1987, 1994). However, Lucas and Renzaglia (2002) found no evidence for stomatal closure in response to ABA or darkness, with some response to desiccation, in line with previous studies by Paton and Pierce (1957). Staining in the guard cells (GCs) with Fast Violet B was associated with pore opening, thus suggesting organic acid accumulation (e.g. malate), but cobaltous nitrate staining for potassium produced equivocal results between open and closed stomata. Lucas and Renzaglia (2002) concluded that hornwort stomata open only once, through osmotic changes, thereafter remaining open with a dual role in providing a passageway for photosynthesis-based gaseous exchange and facilitating sporophyte dehydration. Consequent on a recent study of early stomatal ontogeny in hornworts, Pressel et al. (2014) came to the same conclusion about function (Lucas and Renzaglia, 2002), a view echoed most recently by Villarreal and Renzaglia (2015), Merced and Renzaglia (2017) and Renzaglia et al. (2017). An unexpected discovery by Pressel et al. (2014) was that in vascular plants, the schizogenous intercellular spaces associated with stomata are gas filled from their inception whereas those in hornworts are initially fluid filled. Opening of the stomatal pores is a prerequisite for drying the spaces and replacing the liquid with gas. The importance of this process and the ultimate drying of the sporophyte for spore release would seem to preclude any requirement for regulation of water loss through stomatal movements. Further additional pieces in the jigsaw of stomatal function and evolution are that asymmetric stomata along the sporophyte sutures in hornworts, originally considered abnormal (Pressel et al., 2014), are in fact an integral component in the dehiscence mechanism (Villarreal and Renzaglia, 2015) and that, in hornworts and mosses, apertures and numbers do not respond to different atmospheric CO2 concentrations (Field et al., 2015). In contrast, in vascular plants, species with passive stomatal control have reduced stomatal densities when grown in raised CO2 (Brodribb et al., 2009; Brodribb and McAdam, 2011, 2013, 2017), whereas those with active regulation respond by aperture reductions (Haworth et al., 2015; Lake et al., 2016). It should also be noted that, whilst some authors strongly promote early acquisition of active control on the basis of changes in stomatal number and density in response to both ABA and CO2 in ferns and angiosperms under different laboratory growth conditions (Xu et al., 2016; Hõrak et al., 2017; Merilo et al., 2017), more broadly based and critical evaluations by others make a strong case for gradual evolution (Sussmilch et al., 2017; Brodribb and McAdam, 2017). There are few studies of moss stomatal structure and physiology (Sack and Paolillo, 1983a, b, c, 1985; Duckett et al., 2009; Merced and Renzaglia 2013, 2014; Merced, 2015). In Funaria hygrometrica and the closely related model moss Physcomitrella patens, the GCs are slightly sensitive to ABA, the fungal toxin fusicoccin that stimulates proton pumping across plant cell membranes, darkness and CO2 concentrations (Chater et al., 2011). Expanding green capsules of Funaria have only a short period of up to 5 d during which stomata are responsive to environmental stimuli (Garner and Paolillo, 1973), following which they remain open if hydrated. Merced and Renzaglia (2014) demonstrated that in Funaria, GC walls are thinner, contain higher pectin concentration and are flexible during this period when pores first form. With capsule maturation, GC walls thicken, contain reduced pectin and become less flexible. Stomatal function and homology have been called into question in Sphagnum, the sister group to other mosses (Newton et al., 2000; Cox et al., 2004; Shaw and Renzaglia, 2004). In Sphagnum, the paired GC-like structures adorning the capsule walls never open into intercellular spaces and there is no evidence of potassium fluxes between these and the adjacent epidermal cells (Beerling and Franks, 2009; Duckett et al., 2009). The conclusion from these and a more recent ultrastructural and immunocytochemical study of wall architecture (Merced, 2015) is that these structures in Sphagnum facilitate capsule desiccation leading to spore discharge, but their relationship to true stomata remains an open question. Echoing this notion, Chater et al. (2016) state that delayed dehiscence in stomata-less mutants of Physcomitrella implies that stomata might have promoted dehiscence in the first complex land-plant sporophytes, though this conclusion is difficult to reconcile with the fact that Physcomitrella sporophytes are cleistogamous (McDaniel et al., 2009). Most recently it has been reported that hornwort stomata are short lived, with the GCs dying prior to drying of other sporophytic cells (Merced and Renzaglia, 2017; Renzaglia et al., 2017). These stomata subsequently collapse and remain locked with the apertures wide open over the sub-stomatal cavity. GC collapse precedes the final dehydration of the sporogenous tissue, which is surrounded by pectin-containing mucilage. Thus, as in mosses, hornwort stomata have a short window of potential physiological response to their environment in the green lower portions of the sporophyte. Against the background of new information on stomatal and intercellular space development in hornworts, we report here an extensive experimental programme to determine whether or not: (1) aperture dimensions, immediately after initial opening, are subsequently capable of changing in response to the same external factors that affect mosses and vascular plants (i.e. ABA, desiccation, darkness and plasmolysis); and (2) opening is associated with a potassium flux into the GCs. MATERIALS AND METHODS Nomenclature of hornworts follows Duff et al. (2007) and Glenny (1998). Electron microscopy observations were made on a range of freshly collected hornwort species: Anthoceros punctatus L. [UK]; Dendroceros javanicus (Nees) Nees [Peninsular Malaysia]; Folioceros fuciformis (Mont.) D.C. [Bharadwaj]; Notothylas levieri Schiffn. Ex Steph. [India]; Phaeoceros carolinianus (Michx.) Prosk. [USA]; P. laevis (L.) Prosk. [Wakehurst Place, Sussex, UK]; and Phaeomegaceros fimbriatus (Gottsche) Duff et al. [Panama]. Voucher specimens are in the herbarium at the Natural History Museum, London. For all the experimental treatments to measure stomatal responses to exogenous and environmental cues (exogenous ABA, desiccation and darkness), we used wild-collected UK plants of A. punctatus and P. laevis. We also used the same plants to study the effects of loss of turgor pressure in hornwort GCs by plasmolysing these in 1.0 m sucrose. Since plasmolysis in seed plants results in closure of stomata due to loss of turgidity in GCs, open apertures following plasmolysis in hornwort stomata would be indicative of mechanical maintenance of open pores simply via cell wall properties. The experiments on the wild plants were conducted over a period of 1–3 d after collection, i.e. as quickly as was feasible to record the large numbers of apertures needed. During this time, plants were maintained on their nutrient-poor native substrates in a growth cabinet with a constant irradiance of 100 μmol m–2 s–1 at 18 °C. We used constant illumination to compensate for the higher irradiances experienced by the plants in nature (typically up to 600–900 μmol m–2 s–1). As well as rigorous hornwort controls and, as a further check on our procedures, during our hornwort experiments we also observed the effects of ABA and sucrose on angiosperms (arabidopsis, Lactuca and Hedera) kept in the same growth cabinets. Unlike the hornworts, their stomata closed completely. Because hornwort stomata open at or just above the involucres (Pressel et al., 2014) and moss stomata are considered to be responsive to stimuli for only a few days after pore formation (Chater et al., 2011; Merced and Renzaglia, 2014), plus the recent demonstration that apertures become fixed open on older parts of sporophytes (Renzaglia et al., 2017), we restricted observations to stomata in the first 1 cm of the sporophytes above involucres. As a further precaution to minimize the risk of measuring possible dying or dead stomata, we selected sporophytes at least 3 cm long with the change in colour of the developing spores to yellow or black in Phaeoceros and Anthoceros, respectively, usually >2.5cm above the involucres. For the ABA treatments and desiccation, measurements were taken from ten sporophytes; for darkness and plasmolysis they were taken from five sporophytes. Sporophytes were excised from the subjacent thalli level with the bases of the involucres and sliced longitudinally at right angles to the dehiscence grooves. The two halves were mounted in water, epidermis upwards, and digital images were taken of all the stomata using a Zeiss Axioscop 2 microscope equipped with an AxioCam MRc digital camera. Aperture dimensions (width, length and surface area – calculated with the equation for an ellipse) were measured using the associated Axiovision software. For the desiccation treatment, sporophytes were mounted in immersion oil to avoid rehydration. Control (hydrated) sporophytes were also mounted in immersion oil to discern any possible effect of the oil on aperture dimensions. There was no evidence that any of our manipulations led to the death of the GCs as the plasmolysed sporophytes in sucrose rapidly recovered when placed in water. For the exogenous ABA treatment, we followed the protocol of Hartung (1983) and Hartung et al. (1987). Intact sporophytes and others sectioned longitudinally to optimize uptake were exposed to a range of ABA concentrations (100, 10 and 1 µm), dissolved either in water (unbuffered) or in a medium containing 10 mm KCl in 10 mm MES/NaOH (buffered) (Hartung et al., 1987) for 1, 2, 4 and 24 h in the controlled-environment cabinet. For the buffered treatment, sporophytes were placed in the medium for 1 h prior to addition of ABA as per Hartung et al. (1987). Control treatments were kept in water or in the same medium for corresponding lengths of time. The buffered and unbuffered ABA treatments, for intact or sectioned sporophytes, and irrespective of ABA concentration used and length of exposure, all gave similar results. Therefore, here we report only the results for the intact sporophytes in unbuffered 100 µm after 4 h exposure. The same ABA treatments were repeated using leaves of Lactuca sativa L, Arabidopsis thaliana (L.) Heyn. and Hedera helix L. as vascular plant controls. For the desiccation treatment, sporophyte-bearing thalli were allowed to dry under natural conditions – after approx. 72 h, the dried thalli bearing flaccid sporophytes had lost >50 % of their original water (Fig. 1D) and were used for imaging and aperture measurements. Fig. 1. View largeDownload slide Cryo-scanning (A–G, K) and transmission (H–J) electron micrographs of Anthoceros punctatus (A–F), Phaeoceros laevis (G, J, K), Phaeomegaceros fimbriatus (H) and Folioceros fuciformis (I). (A) Portion of sporophyte just above the involucre with young stomata at the bottom (arrowed) and stomata with fully open apertures above. (B) Closed stoma with smooth walls. (C) Fully opened stoma. (D) Open stomata in a dehydrated sporophyte. (E) Stoma in the wall of a sporophyte well beyond the dehiscence point. (F) Cryo-fractured sporophyte with a mixture of liquid- (*) and gas-filled (arrowed) intercellular spaces. Note the liquid investing the developing spores and elaters. (G) Above the level where the stomata open the liquid is replaced by air, with liquid dried down to a thin layer lining the cell junctions (arrowed). (H, I) Liquid is also visible by TEM: (H) liquid-filled intercellular space (*) and (I) dried out liquid along the cell junctions in the assimilatory tissue (arrowed). (J) Section through stomatal guard cells; note the thickened inner and outer walls of the guard cells (G), the large starch-filled plastid and the dried down liquid in the corners of the intercellular spaces (arrowed). (K) An open stoma after ion milling. Scale bars: (A, D) 500 µm; (F) 100 µm; (B, C, E, J, K) 20 µm; (H, I) 10 µm. Fig. 1. View largeDownload slide Cryo-scanning (A–G, K) and transmission (H–J) electron micrographs of Anthoceros punctatus (A–F), Phaeoceros laevis (G, J, K), Phaeomegaceros fimbriatus (H) and Folioceros fuciformis (I). (A) Portion of sporophyte just above the involucre with young stomata at the bottom (arrowed) and stomata with fully open apertures above. (B) Closed stoma with smooth walls. (C) Fully opened stoma. (D) Open stomata in a dehydrated sporophyte. (E) Stoma in the wall of a sporophyte well beyond the dehiscence point. (F) Cryo-fractured sporophyte with a mixture of liquid- (*) and gas-filled (arrowed) intercellular spaces. Note the liquid investing the developing spores and elaters. (G) Above the level where the stomata open the liquid is replaced by air, with liquid dried down to a thin layer lining the cell junctions (arrowed). (H, I) Liquid is also visible by TEM: (H) liquid-filled intercellular space (*) and (I) dried out liquid along the cell junctions in the assimilatory tissue (arrowed). (J) Section through stomatal guard cells; note the thickened inner and outer walls of the guard cells (G), the large starch-filled plastid and the dried down liquid in the corners of the intercellular spaces (arrowed). (K) An open stoma after ion milling. Scale bars: (A, D) 500 µm; (F) 100 µm; (B, C, E, J, K) 20 µm; (H, I) 10 µm. Possible effects of darkness on aperture dimensions were studied on sporophytes kept in the dark for 15–24 h, and measurements were taken straight away, within 5 min of mounting these. While exposure to darkness was longer than that experienced by plants under natural conditions, the additional time reflects experimental time constraints and the fact that we had to record a large number of stomata. Sporophyte plasmolysis was brought about by submersion of intact sporophytes in 1.0 m sucrose for 15 min and compared with the leaves of a range of identically treated vascular plants (L. sativa, Osmunda regalis L. and Scolopendrium vulgare Sm.). The protocol for transmission electron microscopy (TEM) follows Ligrone and Renzaglia (1990), and the cryo-scanning electron microscopy (cryo-SEM) protocol follows Duckett et al. (2009). For comparisons on intercellular space ontogeny in hornworts, we used representatives of all the major tracheophyte lineages: Huperzia selago (L.) Bernh. ex Schrank & Mart. (lycophyte), Scolopendrium vulgare (monilophyte), Ginkgo biloba L., Podocarpus nivalis Hook., Pinus mugo Turra (gymnosperms) and H. helix (angiosperm). Element X-ray spectra were obtained for both GCs of at least five stomata plus adjacent epidermal cells from the following: (1) stomata inside involucres; (2) exposed stomata below and (3) above the dehiscence point; (4) exposed stomata in plants that had been allowed to dry out for 72 h; (5) exposed stomata on plants kept in the dark for 15–24 h (these were exposed to light for only 5 min, the time between mounting the specimens and freezing); and (6) plants treated with 1, 10 or 100 µm ABA for 1–24 h. Thallus epidermal cells provided controls for potassium concentrations in somatic cells. Spectra were obtained from pits in the cells produced by ion milling and directly from the lumina of fractured cells. Since both procedures produced closely similar results, these are not separated in Table 1. For comparison with hornworts, element spectra sets were obtained from five open stomata and adjacent epidermal cells immediately following removal of leaves from well-watered arabidopsis plants. Other detached leaves were allowed to dry out in the laboratory and five further sets of spectra were collected as soon as the stomata closed (after approx. 1 h and 10 % water loss). Table 1. Means from 8–9 readings of the percentage mass of potassium from X-ray microanalysis Species Inside involucre Exposed Splitting Desiccated Dark 4 h of 100 µm ABA Phaeoceros laevis Guard cells 2.1 ± 0.18 0.9 ± 0.07 3.8 ± 0.42 1.6 ± 0.12 2.4 ± 0.16 2.4 ± 0.20 Stomatal condition Closed Open Open Open Open Open Epidermis 1.7 ± 0.12 1.3 ± 0.10 3.9 ± 0.31 1.5 ± 0.09 2.1 ± 0.18 2.0 ± 0.21 Ratio of K in guard cells to that in epidermal cells 1.24 0.69 0.97 1.06 1.14 1.2 Guard cell K relative to epidermis Higher Lower Same Same Higher Same Thallus control 0.98 ± 0.07 Anthoceros punctatus Guard cells 1.2 ± 0.06 0.9 ± 0.05 1.3 ± 0.05 1.1 ± 0.06 1.0 ± 0.04 6.0 ± 0.82 Stomatal condition Closed Open Open Open Open Open Epidermis 1.0 ± 0.02 1.1 ± 0.02 1.3 ± 0.04 1.2 ± 0.04 0.6 ± 0.03 5.8 ± 0.74 Ratio of K in guard cells to that in epidermal cells 1.2 0.82 1.0 0.92 1.67 1.03 Guard cell K relative to epidermis Higher Lower Same Same Higher Same Arabidopsis* Turgid Wilted Guard cells 3.36 ± 0.36 1.56 ± 0.12 Stomatal condition Open Closed Epidermis 3.24 ± 0.45 3.91 ± 0.64 Ratio of K guard cells to epidermal cells 1.04 0.40 Ratio of K in guard cells to that in epidermal cells Same Lower Species Inside involucre Exposed Splitting Desiccated Dark 4 h of 100 µm ABA Phaeoceros laevis Guard cells 2.1 ± 0.18 0.9 ± 0.07 3.8 ± 0.42 1.6 ± 0.12 2.4 ± 0.16 2.4 ± 0.20 Stomatal condition Closed Open Open Open Open Open Epidermis 1.7 ± 0.12 1.3 ± 0.10 3.9 ± 0.31 1.5 ± 0.09 2.1 ± 0.18 2.0 ± 0.21 Ratio of K in guard cells to that in epidermal cells 1.24 0.69 0.97 1.06 1.14 1.2 Guard cell K relative to epidermis Higher Lower Same Same Higher Same Thallus control 0.98 ± 0.07 Anthoceros punctatus Guard cells 1.2 ± 0.06 0.9 ± 0.05 1.3 ± 0.05 1.1 ± 0.06 1.0 ± 0.04 6.0 ± 0.82 Stomatal condition Closed Open Open Open Open Open Epidermis 1.0 ± 0.02 1.1 ± 0.02 1.3 ± 0.04 1.2 ± 0.04 0.6 ± 0.03 5.8 ± 0.74 Ratio of K in guard cells to that in epidermal cells 1.2 0.82 1.0 0.92 1.67 1.03 Guard cell K relative to epidermis Higher Lower Same Same Higher Same Arabidopsis* Turgid Wilted Guard cells 3.36 ± 0.36 1.56 ± 0.12 Stomatal condition Open Closed Epidermis 3.24 ± 0.45 3.91 ± 0.64 Ratio of K guard cells to epidermal cells 1.04 0.40 Ratio of K in guard cells to that in epidermal cells Same Lower *From Duckett et al. (2009). View Large Table 1. Means from 8–9 readings of the percentage mass of potassium from X-ray microanalysis Species Inside involucre Exposed Splitting Desiccated Dark 4 h of 100 µm ABA Phaeoceros laevis Guard cells 2.1 ± 0.18 0.9 ± 0.07 3.8 ± 0.42 1.6 ± 0.12 2.4 ± 0.16 2.4 ± 0.20 Stomatal condition Closed Open Open Open Open Open Epidermis 1.7 ± 0.12 1.3 ± 0.10 3.9 ± 0.31 1.5 ± 0.09 2.1 ± 0.18 2.0 ± 0.21 Ratio of K in guard cells to that in epidermal cells 1.24 0.69 0.97 1.06 1.14 1.2 Guard cell K relative to epidermis Higher Lower Same Same Higher Same Thallus control 0.98 ± 0.07 Anthoceros punctatus Guard cells 1.2 ± 0.06 0.9 ± 0.05 1.3 ± 0.05 1.1 ± 0.06 1.0 ± 0.04 6.0 ± 0.82 Stomatal condition Closed Open Open Open Open Open Epidermis 1.0 ± 0.02 1.1 ± 0.02 1.3 ± 0.04 1.2 ± 0.04 0.6 ± 0.03 5.8 ± 0.74 Ratio of K in guard cells to that in epidermal cells 1.2 0.82 1.0 0.92 1.67 1.03 Guard cell K relative to epidermis Higher Lower Same Same Higher Same Arabidopsis* Turgid Wilted Guard cells 3.36 ± 0.36 1.56 ± 0.12 Stomatal condition Open Closed Epidermis 3.24 ± 0.45 3.91 ± 0.64 Ratio of K guard cells to epidermal cells 1.04 0.40 Ratio of K in guard cells to that in epidermal cells Same Lower Species Inside involucre Exposed Splitting Desiccated Dark 4 h of 100 µm ABA Phaeoceros laevis Guard cells 2.1 ± 0.18 0.9 ± 0.07 3.8 ± 0.42 1.6 ± 0.12 2.4 ± 0.16 2.4 ± 0.20 Stomatal condition Closed Open Open Open Open Open Epidermis 1.7 ± 0.12 1.3 ± 0.10 3.9 ± 0.31 1.5 ± 0.09 2.1 ± 0.18 2.0 ± 0.21 Ratio of K in guard cells to that in epidermal cells 1.24 0.69 0.97 1.06 1.14 1.2 Guard cell K relative to epidermis Higher Lower Same Same Higher Same Thallus control 0.98 ± 0.07 Anthoceros punctatus Guard cells 1.2 ± 0.06 0.9 ± 0.05 1.3 ± 0.05 1.1 ± 0.06 1.0 ± 0.04 6.0 ± 0.82 Stomatal condition Closed Open Open Open Open Open Epidermis 1.0 ± 0.02 1.1 ± 0.02 1.3 ± 0.04 1.2 ± 0.04 0.6 ± 0.03 5.8 ± 0.74 Ratio of K in guard cells to that in epidermal cells 1.2 0.82 1.0 0.92 1.67 1.03 Guard cell K relative to epidermis Higher Lower Same Same Higher Same Arabidopsis* Turgid Wilted Guard cells 3.36 ± 0.36 1.56 ± 0.12 Stomatal condition Open Closed Epidermis 3.24 ± 0.45 3.91 ± 0.64 Ratio of K guard cells to epidermal cells 1.04 0.40 Ratio of K in guard cells to that in epidermal cells Same Lower *From Duckett et al. (2009). View Large Statistical analysis Because in our experiments we start from the proposition that experimental treatments have no effect when compared with the controls (reference value) and since ‘absence of evidence is not evidence of absence’, here we take a different approach from the usual statistical analyses, e.g. Student’s t-tests, to avoid using the null hypothesis significance test (NHST) and its well-known defects, including giving only a binary result. Instead we define a zone of indifference (ZoI) on either side of the reference value, and plot (T – R)/R, and its 95 % confidence interval (CI95); where R is the reference mean, and T the treatment mean. This plot (see Fig. 5) has a central y-axis value of 0, with the symmetrical ZoI lines above and below set at 0.1 and 0.2, because we judge that a difference of up to ±10 % of the control indicates a small treatment effect that we suggest is, in the circumstances, of little biological importance since it would correspond to a change in aperture width of less than about 1 and 2 µm against a mean of about 8–16 µm for Anthoceros and Phaeoceros, respectively (see Figs 4 and 5). We suggest that a 20 % difference (ZoI ± 0.2) is, again in the circumstances, of some but small biological importance. RESULTS Stomatal architecture and ontogeny of intercellular spaces In sporophytes of this length and stage of maturation, the first few (1–5) stomata immediately above the involucre are recently formed and unopened (Fig. 1A, B; Supplementary Data Fig. S1a). Further up the sporophyte, stomata open (Supplementary Data Fig. S1b, c) with full aperture dimensions in stomata 1 mm above the involucre (Fig. 1A, C; Supplementary data Fig. S1d, e); see also Pressel et al. (2014). Whereas the outer walls in GCs of unopened stomata (Fig. 1B) are smooth, additional wall materials are progressively deposited around pores and over GCs with maturation (Supplementary Data Fig. S1f). Above the point of sporophyte dehiscence, the desiccated epidermal cells are deeply ridged and GCs remain broad in surface view with apertures remaining wide open (Fig. 1E) (see also Villarreal and Renzaglia, 2015; Renzaglia et al., 2017). Cryo-sections of sporophytes within involucres of the genera with stomata we examined (Anthoceros, Folioceros, Phaeoceros and Phaeomegaceros) invariably show liquid-filled intercellular spaces (data not shown, but see Pressel et al., 2014, Fig. 7A). In TEM micrographs, liquid is also visible as a finely granular network (Fig. 1H). Above the involucres, the liquid is gradually replaced by gas (Fig. 1F) and the assimilatory cell chloroplasts become positioned next to the gas-filled spaces (Fig. 1I). This gradual drying process is illustrated by the presence higher up the sporophytes of liquid that has dried down to a thin layer lining the cell junctions (Fig. 1G, I). In sporophytes of hornworts that lack stomata such as Nothoceros (Fig. 2A, B), Dendroceros and Notothylas (data not shown), intercellular spaces are absent in the assimilatory tissues (Fig. 2A), which dry down and collapse post-dehiscence (Fig. 2B). Fig. 2. View largeDownload slide Cryo-scanning electron micrographs of Notothylas levieri (A, B) and Hedera helix (C–E). (A, B) Cryo-fractured sporophytes of the estomate hornwort species Notothylas levieri, young (A) and mature (B); note the complete absence of intercellular spaces in the assimilatory layers (A) which collapse and dry out at maturity (B, arrowed). (C–E) Cryo-fractured young leaves. (C) General aspect of a leaf approximately one-tenth its final size with nascent intercellular spaces. (D) Detail of an intercellular space from (C). (E) Spongy mesophyll and lower epidermis in a leaf approximately a quarter of its final size. Scale bars: (C) 100 µm; (A, B) 50 µm; (E) 20 µm; (D) 10 µm. Fig. 2. View largeDownload slide Cryo-scanning electron micrographs of Notothylas levieri (A, B) and Hedera helix (C–E). (A, B) Cryo-fractured sporophytes of the estomate hornwort species Notothylas levieri, young (A) and mature (B); note the complete absence of intercellular spaces in the assimilatory layers (A) which collapse and dry out at maturity (B, arrowed). (C–E) Cryo-fractured young leaves. (C) General aspect of a leaf approximately one-tenth its final size with nascent intercellular spaces. (D) Detail of an intercellular space from (C). (E) Spongy mesophyll and lower epidermis in a leaf approximately a quarter of its final size. Scale bars: (C) 100 µm; (A, B) 50 µm; (E) 20 µm; (D) 10 µm. In contrast to hornworts, the intercellular spaces in the leaves from representatives of all major tracheophyte groups (lycophytes, ferns, gymnosperms and angiosperms) are gas filled from the outset, as illustrated here in H. helix (Fig. 2C–E). In all stages of development in the array of species we examined, thin pectic strands (Carr et al., 1980) extend across intercellular spaces with no remnants of liquid lining the cell junctions around spaces (Fig. 2D). X-ray microanalysis Fig. 1K shows a cryo-SEM image of ion beam milling prior to X-ray microanalysis in an open stoma and adjacent epidermal cells. The milling holes are precisely positioned to penetrate into the vacuoles of the guard and epidermal cells (Fig. 1J). The mass-proportion of potassium, expressed as a percentage, detected by the probe in GCs of P. laevis and A. punctatus at different developmental stages and in adjacent epidermal cells, is summarized in Table 1. Thallus epidermal cells serve as controls. Typical spectra from which the numerical data were obtained are illustrated in Fig. 3. Potassium was not detected in the sporophytic intercellular liquid or on the surface of the thalli. Peaks (Fig. 3A) and mass-proportions from gametophytic cells were lower than in the sporophytes (Fig. 3B, C). Fig. 3. View largeDownload slide X-ray spectra from Phaeoceros laevis (A–D) and Anthoceros punctatus (E, F). Three vertical arrows indicate peaks attributable to potassium. (The peaks at 0.5 keV are gallium contamination from the ion milling.) (A) Low potassium peaks in a thallus epidermal cell (control). (B, C) The guard cell from an unopened stoma inside an involucre (B) has more pronounced potassium peaks than those in a newly opened stoma (C). (D) High potassium peaks in a guard cell of an open stoma after 4 h exposure to 100 µm ABA. (E) Guard cell in an unopened stoma. (F) High potassium peaks in a guard cell of an open stoma after 4 h exposure to 100 µm ABA. Fig. 3. View largeDownload slide X-ray spectra from Phaeoceros laevis (A–D) and Anthoceros punctatus (E, F). Three vertical arrows indicate peaks attributable to potassium. (The peaks at 0.5 keV are gallium contamination from the ion milling.) (A) Low potassium peaks in a thallus epidermal cell (control). (B, C) The guard cell from an unopened stoma inside an involucre (B) has more pronounced potassium peaks than those in a newly opened stoma (C). (D) High potassium peaks in a guard cell of an open stoma after 4 h exposure to 100 µm ABA. (E) Guard cell in an unopened stoma. (F) High potassium peaks in a guard cell of an open stoma after 4 h exposure to 100 µm ABA. In newly opened stomata of both species, the potassium mass-proportion in GCs is lower than that in epidermal cells. Thereafter it remains the same in both cell types even when sporophytes are desiccated. Also, potassium mass-proportions of unopened stomatal GCs are higher (Fig. 3B) than those of newly opened stomata (Fig. 3B). The mass-proportions of potassium were higher in Anthoceros after treatment with ABA (Fig. 3D–F) and were the same whether or not the pores were open or closed. After periods in darkness, potassium mass- proportions in GCs of Anthoceros appeared slightly higher than in the epidermal cells. In striking contrast, when stomata of arabidopsis close, the GC potassium mass-proportion falls significantly from parity between open GCs and epidermis in the turgid open state (see Duckett et al., 2009). Stomatal responses to environmental stimuli Aperture dimensions (widths and lengths) for the first centimetre of sporophytes above the involucre in control plants of A. punctatus and P. laevis are shown in Supplementary Data Fig. S2. Results from the experimental treatments are summarized in Fig. 4, and indicate that stomatal aperture dimensions in both Anthoceros and Phaeoceros do not decrease in response to exposure to exogenous ABA or darkness, while desiccation and plasmolysis appear to elicit a slight reduction, but never the complete closure observed in ABA-treated (data not shown) and plasmolysed vascular plant stomata (Supplementary Data Fig. S3). A similar decrease in size (width only) was also elicited by desiccation in hornwort sporophyte non-stomatal epidermal cells. After 1–2 h in water, both stomata and epidermal cells regained their pre-desiccation dimensions (data not shown). Fig. 4. View largeDownload slide Box plots of measurements (width, length and surface area) on Anthoceros punctatus (A) and Phaeoceros laevis (B) in the experimental treatments and corresponding controls. The central bar is the median; the box defines the interquartile range (IQR); dashed ‘whiskers’ extend 1.5 IQR beyond the box; beyond that individual measurements are circles. The two parallel horizontal lines near the median define the 95 % confidence limits on the mean. To the left are controls which were kept moist and in light; to the right are experiment treatments: ABA, 100 μm for 4 h; DES, desiccated; DRK, in the dark for 15–24 h; PLA, plasmolysed; OIL, measurements made under oil (Anthoceros only); DCS, non-stomatal epidermal cells desiccated (Anthoceros only; cell widths only). Missing items are shown by an asterisk. Fig. 4. View largeDownload slide Box plots of measurements (width, length and surface area) on Anthoceros punctatus (A) and Phaeoceros laevis (B) in the experimental treatments and corresponding controls. The central bar is the median; the box defines the interquartile range (IQR); dashed ‘whiskers’ extend 1.5 IQR beyond the box; beyond that individual measurements are circles. The two parallel horizontal lines near the median define the 95 % confidence limits on the mean. To the left are controls which were kept moist and in light; to the right are experiment treatments: ABA, 100 μm for 4 h; DES, desiccated; DRK, in the dark for 15–24 h; PLA, plasmolysed; OIL, measurements made under oil (Anthoceros only); DCS, non-stomatal epidermal cells desiccated (Anthoceros only; cell widths only). Missing items are shown by an asterisk. Plotting mean values of aperture sizes (width and length) of treated hornwort stomata and non-stomatal epidermal cells (width only) (CI95) against the reference (control) value (Fig. 5) confirms that both ABA and darkness treatments have no or only a trivial effect on aperture sizes, as values fall well within a ZoI of ±10 %. Desiccation and plasmolysis, on the other hand, have an effect at ZoI ±10 %, especially on aperture width sizes and, for desiccation only, on the lumen width of non-stomatal epidermal cells. However, these values do fall within a ZoI set at ±20 %. The results for the oil treatment on Anthoceros indicate that mounting sporophytes in immersion oil does not affect stomatal aperture sizes when compared with the same mounted in water. Fig. 5. View largeDownload slide Anthoceros punctatus (A) and Phaeoceros laevis (B). The 95 % confidence interval (CI95) of differences in means of ‘experimental’ treatment and ‘control’, expressed as a proportion of the ‘control’ mean, i.e. (x – c)/c. This standardized measure allows comparison of different measures. The ‘controls’ were kept moist and in light. Treatments: ABA, 100 μm for 4 h; DES, desiccated; DRK, in the dark for 15–24 h; PLA, plasmolysed; OIL, measurements made under oil (Anthoceros only); DCS, non-stomatal epidermal cells desiccated (Anthoceros only; cell widths only). Missing items are shown by an asterisk. Fig. 5. View largeDownload slide Anthoceros punctatus (A) and Phaeoceros laevis (B). The 95 % confidence interval (CI95) of differences in means of ‘experimental’ treatment and ‘control’, expressed as a proportion of the ‘control’ mean, i.e. (x – c)/c. This standardized measure allows comparison of different measures. The ‘controls’ were kept moist and in light. Treatments: ABA, 100 μm for 4 h; DES, desiccated; DRK, in the dark for 15–24 h; PLA, plasmolysed; OIL, measurements made under oil (Anthoceros only); DCS, non-stomatal epidermal cells desiccated (Anthoceros only; cell widths only). Missing items are shown by an asterisk. DISCUSSION The present study reveals major physiological differences between hornwort stomata and those in both mosses and vascular plants. Hornwort stomata do not respond to external stimuli and there are no potassium fluxes between epidermal cells and GCs. Identical to hornworts, there is no evidence of potassium fluxes associated with moss stomata (Duckett and Pressel, 2017a). Developmental data (see also Pressel et al., 2014; Renzaglia et al., 2017) and responses to desiccation and exogenous ABA show that in hornworts, stomatal apertures open in early development and are not subsequently actively regulated. Hornwort stomata thus appear to be different from those in the mosses Physcomitrella and Funaria where ABA, darkness and reduced CO2 all decrease aperture dimensions (Chater et al., 2011). These authors also demonstrate, through cross-species complementation and knockout experiments, that stomata in the moss Physcomitrella behave like those in seed plants. However, Merced and Renzaglia (2017) and Renzaglia et al. (2017), drawing attention to the frequent incidence of pore occlusion by waxes and additional wall materials in bryophytes, illustrate nearly mature capsules of Physcomitrella with blocked open pores and/or liquid-filled intercellular spaces. The present physiological data, coupled with liquid-filled sporophytic intercellular spaces, point to the primary role of hornwort stomata as facilitation of sporophyte drying leading to spore discharge, a conclusion reinforced by the early death of the GCs and locking of the apertures in an open position (Renzaglia et al., 2017). The same might well be true in mosses since recent observations over the entire life cycles of the sporophytes of five common mosses revealed that the stomata are invariably open regardless of the prevailing environmental conditions (Duckett and Pressel, 2017b). Functional considerations X-ray microanalysis in hornworts demonstrates that potassium mass-proportion is lower in GCs of newly opened stomata than in the adjacent epidermal cells, and that similar potassium mass-proportions occur in GCs and epidermal cells in stomata on older regions, desiccated sporophytes and those treated with ABA. These findings eliminate the possibility that potassium fluxes are involved in stoma formation and any putative opening or closing mechanism. We view the slightly higher GC potassium mass-proportion in dark-grown specimens as simply reflecting the ability of sporophytes to grow in the dark (Ligrone and Fioretto, 1987). In the absence of potassium fluxes, it seems plausible that stoma formation is the result of uneven and specialized wall deposition in GCs coupled with mobilization of abundant starch reserves in the plastids of young GCs (Pressel et al., 2014). The same arguments are equally applicable to moss stomata (Duckett and Pressel, 2017a). Another major difference that separates hornwort stomata from those in other plant groups (Chater et al., 2011) is their lack of responsiveness to the stress hormone ABA, to darkness and to exogenous/environmental cues widely known to close apertures in seed plants and to result in decreased aperture size in mosses (Chater et al., 2011). Even though we followed the same protocol with far more samples, the results from our ABA treatment contrast with those reported previously by Hartung et al. (1987) who showed a dose-dependent ABA response in the stomata of A. punctatus. In trying to find an explanation for this disparity, we noticed that the two images of control and ABA-treated stomata in Hartung et al. (1987; see figs 5 and 6) seem to represent very different developmental stages unsuitable for comparison (compare with our Supplementary Data Fig. S1a, b and e), especially given the conspicuous difference in aperture wall thickness between the two stomata. Hartung et al. (1987) had no knowledge of the major ontogenetic changes subsequently described in hornwort stomata (Pressel et al., 2014; Villarreal and Renzaglia, 2015; Renzaglia et al., 2017) and which formed the basis for choices in the present work. There are no indications in Hartung et al. (1987) that they chose particular regions or ages of sporophytes to examine. The slight reductions in aperture dimensions after desiccation and plasmolysis in hornwort stomata also contrast with the much more pronounced responses typical of vascular plant stomata to the same cues (Supplementary Data Fig. S3). The absence of closure following plasmolysis is completely at odds with the notion that stomatal movement is modulated by turgor pressure. This finding is consistent with the early development of GC walls that mechanically restrain the apertures of dead stomata in an open configuration as sporophytes dry out (Renzaglia et al., 2017). Similarly, the slight reduction in stoma dimensions after desiccation is most likely to be related to physical, not physiological events, resulting from a general lateral shrinkage of all the sporophyte epidermal cells rather than a stoma-specific response (Merced and Renzaglia, 2017; Renzaglia et al., 2017). The width reductions associated with plasmolysis and desiccation also indicate that the vast majority of our measurements were on living stomata. Dead stomata would not respond. It should be noted that the duration of our darkness treatment (15–24 h), a necessity of our experimental protocol, exceeds what plants experience in the wild and therefore might have caused carbon starvation in hornwort sporophytes potentially leading to stress-induced stomatal opening. However, it is extremely unlikely that 15–24 h, rather than 12 h, could have had such dramatic effects on our experimental plants; further, our results are in line with previous observations of a lack of response to darkness in hornwort stomata (Lucas and Renzaglia, 2002). Although, the dimension reductions following desiccation and plasmolysis were small and not per se significant statistically, the remote possibility that these might be significant functionally still requires close scrutiny in the context of relationships between conductance of water vapour and CO2 which are determined by aperture sizes. The smaller the apertures the greater are the effects of changes in their sizes (Parlange and Waggoner, 1970; Brown and Escombe, 1900; Kaiser, 2009). For example, a width reduction of just 1 µm in a small 4 µm aperture width produces a 25 % reduction in conductance, whereas the corresponding figures for hornworts with larger mean aperture widths of between 6 and 10 µm (Fig. 4) are only 15 and 5 % reductions, respectively. The possibility that these last two small differences might have any real functional significance is even less likely because the stomatal densities in hornworts are an order of magnitude or more lower than those in most angiosperms (Willmer and Fricker, 1996; Field et al., 2015). This observation also appears to rule out in hornworts possible interstomatal diffusion interference, an important parameter in vascular plant leaves particularly in the context of density changes related to atmospheric CO2 concentrations (Woodward, 1987; Beerling et al., 2001; Beerling and Royer, 2002) to which bryophyte stomata are unresponsive (Field et al., 2015). Interstomatal diffusion interference becomes negligible when spacing is over three times stomatal length (Parlange and Waggoner, 1970). Less than 20 % of the stomata of hornworts, even with their very large GC sizes, lie outside this range, in contrast to the much more closely spaced stomata in many mosses (see Field et al., 2015; Table 1; Fig. 3). Similar mechanical constraints in guard and subsidiary cell walls that restrict stomatal movements have also been noted in tracheophytes (Franks and Farquhar, 2007; Shtein et al., 2017a). The interpretation that hornwort stomata are structures facilitating sporophyte desiccation and spore dispersal finds parallel in anthers. Stomata on young anthers may open and close, but permanent maturational opening is thought to facilitate drying out and dehiscence leading to faster pollen release (Schmid, 1976; Heslop-Harrison et al., 1987; Lersten, 2004). Permanently open stomata are also a feature of nectaries, where they act as secretory pores rather than in gas exchange (Davis et al., 1986), of hydathodes, where the open pores may become occluded with granular, crystalline and waxy materials (Stevens, 1956; Takeda et al., 1991), and in aquatic angiosperms where loss of function is associated with altered patterns of cellulose crystallinity (Shtein et al., 2017b). Evolution of stomata The lack of stomata in four hornwort genera represents two independent losses and may be readily understood in terms of this postulated role in sporophyte desiccation (Pressel et al., 2014; Renzaglia et al., 2017). All four taxa fill ecological niches where drying via stomata is not adaptive. In the ephemeral Notothylas, the spores are usually water dispersed from highly reduced sporophytes that are, in many species, immersed in involucres throughout development (Hasegawa, 1980; Schuster, 1984; Singh, 2002). In the Megaceros/Nothoceros/Dendroceros lineage, the loss of stomata is consistent with the habitats they occupy. Megaceros and Nothoceros occur in highly humid regions with long growing seasons where they do not experience seasonal cycles of desiccation as do most temperate taxa. In the epiphytic and epiphyllous Dendroceros, the only desiccation-tolerant hornwort genus, cycles of rapid drying and re-wetting are the norm. Long involucres in these three genera provide additional protection of vulnerable tissue in environments where spores are dispersed over long growing seasons. The absence of any experimental evidence for regulation of aperture movements and liquid-filled subjacent intercellular spaces supports the conclusion that the primary role of hornwort stomata is sporophyte desiccation. This conclusion points to a rethink of the possible course of stomatal evolution in land plants. Although the positions of mosses sister to hornworts and hornworts sister to vascular plants (Qiu et al., 2006, 2007; Liu et al., 2014) are consistent with a single origin for stomata (Raven, 2002), this hypothesis infers that the absence of stomatal regulation, as shown in the present study, was lost in hornworts following an earlier acquisition in mosses (Chater et al., 2011). On the other hand, phylogenetic analyses placing hornworts sister to other land plants (Cox et al., 2014; Wickett et al., 2014) are in line with the evolution of the stomatal regulatory molecular toolkit described by Chater et al. (2011, 2013, 2016, 2017) after the split of hornworts from the remaining land plants. A recent analysis of possible stomatal gains and losses (Rensing, 2018) shows the most parsimonious configuration to be one in which stomata were independently gained during land plant evolution, with a mosses-liverworts clade as sister to all other land plants and hornworts sister to the vascular plants (Puttick et al., 2018). All other configurations involve more losses and/or gains. When the hornwort genome becomes available, thus enabling the rapid identification of stomatal genes, we predict that a host of both stomatal patterning and identity genes will be shared across stomatophytes. However, the function of those genes will require further genetic and experimental analyses. Pertinent to these evolutionary considerations are contrasting views on stomatal physiology between seed plants and early divergent tracheophytes (Brodribb and McAdam, 2011, 2012, 2017; Chater et al., 2011; McAdam and Brodribb, 2011; Ruszala et al., 2011). In seed plants, control of water balance is by active metabolic regulation of stomatal aperture via ion-pumping processes and closure induced by ABA. On the basis of molecular data and physiological experiments showing small changes in aperture and stomatal conductance, Ruszala et al. (2011) conclude that stomatal responses in Selaginella to both CO2 and the physiologically active enantiomer of ABA are similar to those in arabidopsis. They also report differences in potassium mass-proportion between GCs of light- and dark-grown Selaginella, although the cobaltinitrite staining method they used produced largely qualitative data. From their experiments, they argue that physiologically active stomatal control originated as least as far back as the origin of the lycophytes approximately 420 million years ago. From Physcomitrella studies, Chater et al. (2011, 2013) posit the same for mosses; they contend that mosses possess the same core regulatory components as flowering plants in GC ABA signalling, and that these originated in a common ancestor of bryophytes and tracheophytes. Phylogenomic studies of stomatal genes that include Physcomitrella support the single origin of GCs across land plants as Physcomitrella possesses GC patterning and development genes found in arabidopsis (Bergmann and Sack, 2007; Peterson et al., 2010; Rychel and Peterson, 2010; Caine et al., 2016; Chater et al., 2017). In contrast, Brodribb and McAdam’s (2011) finding of only a 15 % reduction in stomatal apertures in the presence of ABA in a lycophyte and further extensive studies on pteridophytes (reviewed by Brodribb and McAdam, 2017; Sussmilch et al., 2017) points to a lack of or very limited active control of water loss. These authors suggested that early land plants possessed passive stomatal control with the absence of an active ion-pumping mechanism, much like that demonstrated here in hornworts, and the gradual acquisition of active control mechanisms occurred as plants diversified. However, differently from hornworts, desiccation and plasmolysis did cause rapid stomatal closure in some ferns and lycophytes. The general lack of stomatal response in hornworts clearly conforms to the notion of gradual acquisition of stomatal regulation (Brodribb and McAdam, 2017; Sussmilch et al., 2017). Added to the present results is the demonstration that stomatal density and aperture in both hornworts and mosses do not respond to atmospheric CO2 concentrations (Field et al., 2015). The single study (Chater et al., 2011) based on actual measurements of stomata rather than analysis of stoma-related genes provides the only direct evidence for the early acquisition of active regulation. That study showed small responses rather than complete closure in two mosses with unusual pores in single GCs (Funaria and Physcomitrella). Merced and Renzaglia (2017) highlight that studies on Physcomitrella stomata are challenging; ‘they might not reflect the reality of the great majority of mosses because of its reduced cleistocarpic capsule with few stomata, small apertures and poorly developed intercellular spaces.’ It is difficult to envisage how absence of stomata might delay the dehiscence of cleistocarpous capsules with this suite of characters (Chater et al., 2016). Physiological and genetic studies including aperture measurements now need to be replicated in additional mosses with structurally elaborate capsules that contain well-developed apophyses and a system of intercellular spaces leading to regularly spaced stomata with two guard cells. No matter how many more stomata-related genes are discovered to be common across land plants, future considerations of structural and functional continuity must now take into account the recent demonstration of the multiple evolution of intercellular spaces in bryophytes (Duckett and Pressel, 2017a). Whereas this integral component of the stomatal apparatus is present at the foot of the hornwort tree (Leiosporoceros), it is strikingly absent from the basal moss clades, Sphagnum, Takakia and Andreaea. SUPPLEMENTARY DATA Supplementary data are available online at https://academic.oup.com/aob and consist of the following. Figure S1: light micrographs of living sporophytes illustrating the ontogenetic changes in stomatal apertures in Anthoceros punctatus. Figure S2: Anthoceros punctatus and Phaeoceros laevis scatter plots of stomatal width and length for the four experimental treatents and corresponding controls. Figure S3: effects of plasmolysis on Phaeoceros laevis, Lactuca sativa, Osmunda regalis and Scolopendrium vulgare stomata. ACKNOWLEDGEMENTS We are grateful for skilled assistance in operating the cryo-SEM provided by Ken M. Y. P’ng, NanoVision Centre, School of Engineering and Materials Science, Queen Mary University of London E1 4NS, UK, and by Integrated Microscopy and Graphics Expertise staff for use of electron microscope facilities at Southern Illinois University. Support from DEFRA (Darwin Initiative), The New Phytologist Trust, the Royal Society (UK), a Leverhulme Emeritus Fellowship (J.G.D.) and a Leverhulme Early Career Fellowship (S.P.) enabled J.G.D. and S.P. to collect specimens used in this study from Chile and New Zealand. We thank Brian Butterfield (University of Canterbury) and David Glenny (Landcare, New Zealand) for help with collecting, and the New Zealand Department of Conservation for collecting permits. This project was supported by two NSF Tree of Life initiatives (DEB-0228679, DEB-0531751). LITERATURE CITED Beerling DJ , Franks PJ . 2009 . Evolution of stomata in ‘lower’ land plants . New Phytologist 183 : 921 – 925 . Google Scholar CrossRef Search ADS PubMed Beerling DJ , Royer DL . 2002 . Reading a CO2 signal from fossil stomata . New Phytologist 153 : 387 – 397 . Google Scholar CrossRef Search ADS Beerling DJ , Osborne CP , Chaloner WG . 2001 . Evolution of leaf-form in land plants linked to atmospheric CO2 decline in the Late Palaeozoic era . Nature 410 : 352 – 354 . Google Scholar CrossRef Search ADS PubMed Bergmann FC , Sack FD . 2007 . Stomatal development . Annual Review of Plant Biology 58 : 163 – 181 . Google Scholar CrossRef Search ADS PubMed Bopp M , Werner O . 1993 . Abscisic acid and desiccation tolerance in mosses . Botanica Acta 106 : 103 – 106 . Google Scholar CrossRef Search ADS Brodribb TJ , McAdam SAM . 2011 . Passive origins of stomatal control in vascular plants . Science 331 : 582 – 585 . Google Scholar CrossRef Search ADS PubMed Brodribb TJ , McAdam SAM . 2012 . Fern and lycophyte guard cells do not respond to endogenous ABA . The Plant Cell 24 : 1510 – 1521 . Google Scholar CrossRef Search ADS PubMed Brodribb TJ , McAdam SAM . 2013 . Unique responsiveness of angiosperm stomata to elevated CO2 explained by calcium signalling . PLoS One 8 : e82057 . doi: 10.1371/journal.pone.0082057 . Google Scholar CrossRef Search ADS PubMed Brodribb TJ , McAdam SA . 2017 . Evolution of the stomatal regulation of plant water content . Plant Physiology 174 : 639 – 649 . Google Scholar CrossRef Search ADS PubMed Brodribb TJ , McAdam SAM , Jordan GJ , Feild TS . 2009 . Evolution of stomatal responsiveness to CO2 and optimization of water-use efficiency among land plants . New Phytologist 183 : 839 – 847 Google Scholar CrossRef Search ADS PubMed Brown H , Escombe F . 1900 . Static diffusion of gases and liquids in relation to the assimilation of carbon and translocation in plants . Philosophical Transactions of the Royal Society B: Biological Sciences 193 : 223 – 291 . Google Scholar CrossRef Search ADS Caine R , Chater CC , Kamisugi Y , et al. 2016 . An ancestral stomatal patterning module revealed in the non-vascular plant Physcomitrella patens . Development 143 : 3306 – 3314 . Google Scholar CrossRef Search ADS PubMed Carr DJ , Oates K , Carr SGM . 1980 . Studies on intercellular pectic strands of leaf palisade parenchyma . Annals of Botany 45 : 403 – 413 . Google Scholar CrossRef Search ADS Chater C , Kamisugi Y , Movahedi M , et al. 2011 . Regulatory mechanism controlling stomatal behavior conserved across 400 million years of land plant evolution . Current Biology 21 : 1025 – 1029 . Google Scholar CrossRef Search ADS PubMed Chater C , Gray JE , Beerling DJ . 2013 . Early evolutionary acquisition of stomatal control and development gene signalling networks . Current Opinion in Cell Biology 16 : 638 – 646 . Google Scholar CrossRef Search ADS Chater C , Caine RS , Tomek M , et al. 2016 . Origin and function of stomata in the moss Physcomitrella patens . Nature Plants 2 : 16179 . doi: 10.1038/nplants.2016.179 . Google Scholar CrossRef Search ADS PubMed Chater C , Caine RS , Fleming AJ , Gray JE . 2017 . Origins and evolution of stomatal development . Plant Physiology 174 : 624 – 638 . Google Scholar CrossRef Search ADS PubMed Cox CJ , Goffinet B , Shaw AJ , Boles SB . 2004 . Phylogenetic relationships among the mosses based on heterogeneous Bayesian analysis of multiple genes from multiple genomic compartments . Systematic Botany 29 : 234 – 250 . Google Scholar CrossRef Search ADS Cox CJ , Li B , Foster PG , Embley TM , Civáň P . 2014 . Conflicting phylogenies for early land plants are caused by composition biases among synonymous substitutions . Systematic Biology 63 : 862 – 878 . Google Scholar CrossRef Search ADS PubMed Davis AR , Peterson RL , Shuel RW . 1986 . Anatomy and floral nectaries of Brassica napus (Brassicaceae) . Canadian Journal of Botany 64 : 2508 – 2516 . Google Scholar CrossRef Search ADS Duckett JG , Pressel S . 2017a . The evolution of the stomatal apparatus: intercellular spaces and sporophyte water relations—two ignored dimensions . Philosophical Transactions of the Royal Society B: Biological Sciences 373 : doi: 10.1098/rstb.2016.0498 . Duckett JG , Pressel S . 2017b . The colorful phenology of five common terricolous mosses in London, England . Bryophyte Diversity and Evolution 39 : 44 – 56 . Google Scholar CrossRef Search ADS Duckett JG , Pressel S , P’ng KMY , Renzaglia KS . 2009 . Exploding a myth: the capsule dehiscence mechanism and the function of pseudostomata in Sphagnum . New Phytologist 183 : 1053 – 1063 . Google Scholar CrossRef Search ADS PubMed Duff JE , Villarreal JC , Cargill DC , Renzaglia KS . 2007 . Progress and challenges toward developing a phylogeny and classification of the hornworts . Bryologist 110 : 214 – 243 . Google Scholar CrossRef Search ADS Field KJ , Duckett JG , Cameron DD , Pressel S . 2015 . Stomatal density and aperture in non-vascular land plants are non-responsive to atmospheric CO2 concentrations . Annals of Botany 115 : 915 – 922 . Google Scholar CrossRef Search ADS PubMed Franks PJ , Farquhar GD . 2007 . The mechanical diversity of stomata and its significance in gas exchange control . Plant Physiology 143 : 78 – 87 . Google Scholar CrossRef Search ADS PubMed Garner DLB , Paolillo DJ . 1973 . On the functioning of stomates in Funaria . Bryologist 76 : 423 – 427 . Google Scholar CrossRef Search ADS Glenny D . 1998 . A revised checklist of New Zealand liverworts and hornworts . Tubingia 10 : 119 – 149 . Hartung W . 1983 . The site of action of abscisic acid at the guard cell plasmalemma of Valerianella locusta . Plant, Cell and Environment 6 : 427 – 428 . Google Scholar CrossRef Search ADS Hartung W . 2010 . The evolution of abscisic acid (ABA) and ABA function in lower plants, fungi and lichen . Functional Plant Biology 37 : 806 – 812 . Google Scholar CrossRef Search ADS Hartung W , Weiler EW , Volk OH . 1987 . Immunochemical evidence that abscisic acid is produced by several species of Anthocerotae and Marchantiales . Bryologist 90 : 393 – 400 . Google Scholar CrossRef Search ADS Hartung W , Hellwege EM , Volk OH . 1994 . The function of abscisic acid in bryophytes . Journal of the Hattori Botanical Laboratory 76 : 59 – 65 . Haworth M , Killi D , Materassi A , Raschi A . 2015 . Coordination of stomatal physiological behaviour and morphology with carbon dioxide determines stomatal control . American Journal of Botany 102 : 677 – 688 . Google Scholar CrossRef Search ADS PubMed Hasegawa J . 1980 . Taxonomic studies on Asian Anthocerotae. II. Some Asian species of Dendroceros . Journal of the Hattori Botanical Laboratory 47 : 287 – 309 . Heslop-Harrison JS , Heslop-Harrison Y , Reger BJ . 1987 . Anther filament extension in Lilium: potassium ion movement and some anatomical features . Annals of Botany 59 : 505 – 525 . Google Scholar CrossRef Search ADS Kaiser H . 2009 . The relation between stomatal aperture and gas exchange under consideration of pore geometry and diffusional resistance in the mesophyll . Plant, Cell and Environment 32 : 1091 – 1098 . Google Scholar CrossRef Search ADS Hõrak H , Kollist H , Ebe Merilo E . 2017 . Fern stomatal responses to ABA and CO2 depend on species and growth conditions . Plant Physiology 174 : 672 – 679 . Google Scholar CrossRef Search ADS PubMed Lake JA , Walker H , Cameron DC , Lomax BH . 2016 . A novel root-to-shoot stomatal response to very high CO2 levels in the soil: electrical, hydraulic and biochemical signalling . Physiologia Plantarum 159 : 433 – 444 . Google Scholar CrossRef Search ADS PubMed Lersten N . 2004 . Flowering plant embryology . London : Wiley . Google Scholar CrossRef Search ADS Ligrone R , Fioretto A . 1987 . Chloroplast development in dark grown sporophytes of Phaeoceros laevis L. Prosk . New Phytologist 105 : 301 – 308 . Google Scholar CrossRef Search ADS Ligrone R , Renzaglia KS . 1990 . The sporophyte–gametophyte junction in the hornwort Dendroceros tubercularis Hatt. (Anthocerophyta) . New Phytologist 114 : 497 – 505 . Google Scholar CrossRef Search ADS Ligrone R , Duckett JG , Renzaglia KS . 1993 . The gametophyte–sporophyte junction in land plants . Advances in Botanical Research 19 : 231 – 317 . Google Scholar CrossRef Search ADS Liu Y , Cox CJ , Wang W , Goffinet B . 2014 . Mitochondrial phylogenomics of early land plants: mitigating the effects of saturation, compositional heterogeneity and codon-usage bias . Systematic Biology 63 : 862 – 878 . Google Scholar CrossRef Search ADS PubMed Lucas JR , Renzaglia KS . 2002 . Structure and function of hornwort stomata . Microscopy and Microanalysis 8 ( Suppl 2 ): 1090 – 1091 . McAdam SAM , Brodribb TJ . 2011 . Stomatal innovation and the rise of seed plants . Ecology Letters 14 : 1 – 8 . Google Scholar CrossRef Search ADS PubMed McDaniel SF , von Stackelberg M , Richardt S , Quatrano R , Reski R , Rensing SA . 2009 . The speciation history of the Physcomitrium–Physcomitrella species complex . Evolution 64 : 217 – 231 . Google Scholar CrossRef Search ADS PubMed Merced A . 2015 . Novel insights on the structure and composition of pseudostomata in Sphagnum . American Journal of Botany 102 : 329 – 335 . Google Scholar CrossRef Search ADS PubMed Merced A , Renzaglia KS . 2013 . Moss stomata in highly elaborated Oedipodium (Oedipodiaceae) and highly reduced Ephemerum (Pottiaceae) sporophytes are remarkably similar . American Journal of Botany 100 : 2318 – 2327 . Google Scholar CrossRef Search ADS PubMed Merced A , Renzaglia KS . 2014 . Developmental changes in guard cell wall structure and pectin composition in the moss Funaria: implications for function and evolution of stomata . Annals of Botany 114 : 1001 – 1010 . Google Scholar CrossRef Search ADS PubMed Merced A , Renzaglia KS . 2017 . Structure, function and evolution of stomata from a bryological perspective . Bryophyte Diversity and Evolution 39 : 007 – 020 . Google Scholar CrossRef Search ADS Merilo E , Yarmolinsky D , Jalakas P , et al. 2017 . Origin and role of ABA in stomatal regulation . Plant Physiology On the Inside doi: 10.1104/pp.17.00912 . Morris JL , Puttick MN , Clark J , Edwards D , Kenrick P , Pressel S , Wellman CH , Yang Z , Schneider H , Donoghue PCJ . 2018 . The timescale of early land plant evolution . Proceedings of the National Academy of Sciences of the USA 115 : E2274 – E2283 . Google Scholar CrossRef Search ADS PubMed Newton AE , Cox CJ , Duckett JG , Goffinet B , Hedderson TAJ , Mishler BD . 2000 . Evolution of the major moss lineages: phylogenetic analyses based on multiple gene sequences and morphology . Bryologist 103 : 187 – 211 . Google Scholar CrossRef Search ADS Parlange J-Y , Waggoner PE . 1970 . Stomatal dimensions and resistance to diffusion . Plant Physiology 46 : 337 – 342 . Google Scholar CrossRef Search ADS PubMed Paton JA , Pearce JV . 1957 . The occurrence, structure and functions of the stomata in British bryophytes. II. Functions and physiology . Transactions of the British Bryological Society 3 : 242 – 259 . Google Scholar CrossRef Search ADS Peterson KM , Rychel AL , Torii KU . 2010 . Out of the mouths of plants: the molecular basis of the evolution and diversity of stomatal development . The Plant Cell 22 : 296 – 306 . Google Scholar CrossRef Search ADS PubMed Pressel S , Goral T , Duckett JG . 2014 . Stomatal differentiation and abnormal stomata in hornworts . Journal of Bryology 36 : 87 – 103 . Google Scholar CrossRef Search ADS Puttick MN , Morris J , Williams TA , Cox CJ , Edwards D , Kernick P , Pressel S , Wellman CH , Schneider H , Pisani D , Donoghue PCJ . 2018 . The interrelationships of land plants and the nature of the ancestral embryophyte . Current Biology 28 : 733 – 745 . Google Scholar CrossRef Search ADS PubMed Qiu Y-L , Li L , Wang B , et al. 2006 . The deepest divergences in land plants inferred from phylogenomic evidence . Proceedings of the National Academy of Sciences, USA 103 : 15511 – 15516 . Google Scholar CrossRef Search ADS Qiu Y-L , Li L , Wang B , et al. 2007 . Nonflowering land plant phylogeny inferred from nucleotide sequences of seven chloroplast, mitochondrial, and nuclear genes . International Journal of Plant Science 168 : 691 – 708 . Google Scholar CrossRef Search ADS Raven JA . 2002 . Selection pressures on stomatal evolution . New Phytologist 153 : 371 – 386 . Google Scholar CrossRef Search ADS Rensing SA . 2018 . Plant evolution: phylogenetic relationships between the earliest land plants . Current Biology 28 : R208 – R231 . Google Scholar CrossRef Search ADS PubMed Renzaglia KS , Villarreal JC , Piatkowski BT , Lucas JR , Merced A . 2017 . Hornwort stomata: architecture and fate shared with 400 million year old fossil plants without leaves . Plant Physiology 174 : 788 – 797 . Google Scholar CrossRef Search ADS PubMed Ruszala EM , Beerling DJ , Franks PJ , et al. 2011 . Land plants acquired active stomatal control early in their evolutionary history . Current Biology 21 : 1030 – 1035 . Google Scholar CrossRef Search ADS PubMed Rychel AL , Peterson KM . 2010 . Plant twitter: ligands under 140 amino acids enforcing stomatal patterning . Journal of Plant Research 123 : 275 – 280 . Google Scholar CrossRef Search ADS PubMed Sack F , Paolillo DJ . 1983a . Stomatal pore and cuticle formation in Funaria . Protoplasma 116 : 1 – 13 . Google Scholar CrossRef Search ADS Sack F , Paolillo DJ . 1983b . Protoplasmic changes during stomatal development in Funaria . Canadian Journal of Botany 61 : 2515 – 2526 . Google Scholar CrossRef Search ADS Sack F , Paolillo DJ . 1983c . Structure and development of walls in Funaria stomata . American Journal of Botany 70 : 1019 – 1030 . Google Scholar CrossRef Search ADS Schmid R . 1976 . Filament histology and anther dehiscence . Botanical Journal of the Linnean Society 73 : 303 – 315 . Google Scholar CrossRef Search ADS Schuster RM . 1984 . Morphology, phylogeny and classification of the hornworts . In: Schuster RM , ed. New manual of bryology , Vol. 2 . Nichinan, Japan : Hattori Botanical Laboratory , 1071 – 1092 . Shaw AJ , Renzaglia KS . 2004 . Phylogeny and diversification of bryophytes . American Journal of Botany 10 : 1557 – 1581 . Google Scholar CrossRef Search ADS Shtein I , Shelef Y , Marom Z , et al. 2017a . Stomatal cell wall composition: distinctive structural patterns associated with different phylogenetic groups . Annals of Botany 119 : 1021 – 1033 . Google Scholar CrossRef Search ADS Shtein I , Popper ZA , Harpaz-Saad S . 2017b . Permanently open stomata of aquatic angiosperms display modified cellulose crystallinity patterns . Plant Signaling and Behavior 12 : e1339858 . doi: 10.1080/15592324.2017.1339858 Google Scholar CrossRef Search ADS Singh DK . 2002 . Notothylaceae of India and Nepal: a morpho-taxonomic revision . Dehradun, India : Bishen Singh Mahendrapal Singh . Stevens ABP . 1956 . The structure and development of the hydathodes of Caltha palustris L . New Phytologist 55 : 339 – 345 . Google Scholar CrossRef Search ADS Sussmilch FC , Brodribb TJ , McAdam SAM . 2017 . What are the evolutionary origins of stomatal responses to abscisic acid (ABA) in land plants ? Journal of Integrative Plant Biology 59 : 240 – 260 . Google Scholar CrossRef Search ADS PubMed Takeda F , Wisniewski ME , Glenn DM . 1991 . Occlusion of water pores prevents guttation in older strawberry leaves . Journal of the American Society of Horticultural Science 116 : 1122 – 1125 . Villarreal JC , Renzaglia KS . 2015 . The hornworts: important advancements in early land plant evolution . Journal of Bryology 37 : 157 – 170 . Google Scholar CrossRef Search ADS Wickett NJ , Mirarab S , Nguyen N , et al. 2014 . Phylotranscriptomic analysis of the origin and early diversification of land plants . Proceedings of the National Academy of Sciences, USA 111 : E4859 – E4868 . Google Scholar CrossRef Search ADS Willmer CM , Fricker M . 1996 . Stomata , 2nd edn . London : Chapman & Hall . Google Scholar CrossRef Search ADS Woodward FI . 1987 . Stomatal numbers are sensitive to increases in CO2 from pre-industrial levels . Nature 327 : 617 – 618 . Google Scholar CrossRef Search ADS Xu Z , Jiang Y , Jia B , Zhou G . 2016 . Elevated-CO2 response of stomata and its dependence on environmental factors . Frontiers in Plant Science 7 : 657 . doi: 10.3389/fpls.2016.00657 . Google Scholar PubMed © The Author(s) 2018. Published by Oxford University Press on behalf of the Annals of Botany Company. All rights reserved. For permissions, please e-mail: [email protected]. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)
Plant Cuttings2018 Annals of Botany
doi: 10.1093/aob/mcy129pmid: 29961809
Inspiring the Botanists of the future View largeDownload slide View largeDownload slide Part of the goal of Plant Cuttings items is to share news of botanical research with the wider plant-minded community, the better to advertise that wonderful example of human scientific endeavour. And that’s fine for promoting the work of established plant scientists. But what about the not unimportant – i.e. very important – matter of trying to ensure ‘continuity of supply’? How do we enthuse the new botanists to replace those who will eventually retire, etc. (and whose own future discoveries and contributions to botanical knowledge may one day be shared via a Plant Cutting)? To help with that, and another aim of these items, is to inform the current crop of plant practioners of ‘tools’ and resources that they can use to inspire the next generation of plant biologists. So, here’s a round-up (no glyphosate[1]-related pun intended; these items are intended to ‘cause to flourish’ rather than kill…) of some that caught my eye recently. First – although in no particular order of importance– is the initiative of Prof. Lena Struwe (of the USA’s Rutgers University[2])*. Called Botany Depot[3,4], this resource aspires to be “a global website for creative ideas and materials for teaching botany in the 21st century for all ages and levels”. Whilst the focus of most of those resources is on existing students and inspiring plant knowledge and understanding within that important audience, it is equally necessary to reach out to the general public and help them to appreciate the importance of plant science and plants more generally. Developing projects and activities to achieve that admirable aim[5] was part of the outcomes of the ASPB Conviron Scholars Program[6,7]. Organised by the ASPB (the American Society of Plant Biology[8]), the programme’s participants were drawn from a global pool of talented and aspiring plant scientists from the USA, the United Kingdom, Nigeria, and Belgium[6]. Many of the projects are suitable not only for informing but also enthusing the general public with plant science. They should also work well with students (who, after all, until they commit to a plant science career, are also members of the general public…), and can be explored here[5]. A personal interest of mine is exploring the inter-relatedness of plants and people. That is catered for by Herbaria 3.0[9]. Although it’s arguably less ‘academic’ in focus than some of the other resources considered in this item, it encourages the sharing of stories about plants and people, especially those that cause us to recall and reflect upon the important role that plants play in all our lives. The more such stories are shared, the more people might realise how important plants are. This in turn might also help to inspire a desire to study them further. All of the resources mentioned have the goal of sharing the excitement and joy that one gets from knowing about plants. If that helps to enthuse, inspire, and create the next generation of plant biologists, that is a job well done. At the other end of the spectrum of ‘outreach’ and spreading the message that plants are cool (too!… [10]) and worthy of study, a quick mention of Plant Roots and Light[11], a blog by Dr Kasper van Gelderen (Universiteit Utrecht, The Netherlands[12]). Kasper is an established plant scientist whose research focusses on the integration of shade avoidance signals from the shoot to the root and vice versa. His blog concentrates on those aspects of his professional life, with the intention to introduce this work to a broader audience. Real plant scientists blogging about their work (or using other social media platforms), with passion, is another great way of inspiring more and future plant scientists – and is another free resource to use and share. Blog on, Kasper – et al.[13]!** * Regular readers of Plant Cuttings might recognise this name[14], for it’s the same person that runs the Better Botanical Business Bureau, which appears as the Botanical Accuracy blog site[15]. The latest item to appear there [when this Cutting was written] was a surgical dissection of a press release from the University of Bristol (Bristol, UK)[16]. Entitled “Plants colonized the earth 100 million years earlier than previously thought”, it purports to report the science behind Jennifer Morris et al.’s paper “The timescale of early land plant evolution”[17]. However, rather than just itemise the perceived deficiencies and inaccuracies in the press release (and explain why they are deficient and/or inaccurate…), Prof. Struwe also helpfully provides a reworked version of the press release. In that way she is attempting to educate those who report on the work of botanists. After all, it is important to have the important work of botanists reported in the most accurate way possible. ** One was tempted to say ‘High Five’[18] to Kasper, as an appreciative pun that alludes to his work with the HY5 transcription factor[19]. But that might be far too specialist for the more generalist audience a Plant Cutting item is trying to reach. So I resisted the temptation… Image from: Wikimedia Commons References [1] https://ec.europa.eu/food/plant/pesticides/glyphosate_en [2] https://www.rci.rutgers.edu/~struwe/ [3] https://botanydepot.com/ [4] http://sebsnjaesnews.rutgers.edu/2018/02/lena-struwe-launches-botany-depot-a-global-botanical-education-resource/ [5] https://community.plantae.org/article/4917253407145723467/articles-and-projects-by-conviron-scholars [6] https://blog.aspb.org/announcing-the-inaugural-class-of-the-aspb-conviron-scholars-program/ [7] https://aspb.org/awards-funding/aspb-awards/aspb-conviron-scholars-program/ [8] https://aspb.org/about/ [9] https://herbaria3.org/about-herbaria-3-0-2/ [10] https://www.bucknell.edu/x78387.xml [11] https://plantrootsandlight.wordpress.com/ [12] https://www.uu.nl/staff/KvanGelderen/0 [13] https://blog.feedspot.com/botany_blogs/ [14] https://www.botany.one/2016/05/whither-botanical-accuracy/ [15] http://www.botanicalaccuracy.com/ [16] http://www.bris.ac.uk/news/2018/february/plant-evolution.html [17] PNAS ; http://www.pnas.org/cgi/doi/10.1073/pnas.1719588115 [18] https://dictionary.cambridge.org/dictionary/english/high-five [19] https://plantrootsandlight.files.wordpress.com/2018/04/18-01-04-kvg-poster-ispp2018_v2.pdf When plant biology meet physics… View largeDownload slide View largeDownload slide Great things are possible when disciplines that may be studied separately and distinctly are brought together. For example, and famously, when botany, zoology, bacteriology, mycology, protistology, virology, chemistry, physics, and anthropology (and maybe a few more ‘-ologies’ and non-ologies…) come together we get the new(-ish) discipline of ecology. More modestly, this item is concerned just with two sciences, botany and physics*. And its sole declared intention is to alert the readers of Plant Cuttings – who are a switched-on plant-minded bunch – to a special issue of Physics World. Although this is a journal that may not be on their radar as far as plant-related reading goes, April 2018’s issue featured many articles that take a physics perspective on plant matters[1]. And, because Plant Cuttings is about service to the botanical community, I’ve done the hard work for you (and it took quite a while to do…) and tracked down freely-available copies of that issue’s plant physics articles. So, you can now read about: Cornell University (USA) botanist Karl Niklas[2,3]**; the issues of growing plants in space[4]; nano-strategies used by flowers for colour and pollinator-attraction[5]; discover the connection between transpiration and cooling vehicles that travel at hypersonic speeds[6]***; gain insights into how electric fields can affect root growth and regeneration[7]****; and discover whether – or not – photosynthesis is ‘quantum-ish’[8]. Happy to help put some ‘fizz’ back into your botany. * More examples of ‘when botany meets physics’ can be found in the following Cuttings item, Flowers (it’s what angiosperms are all about!). For a good source of plant (and other lifeforms) and physics investigations, we recommend the Journal of the Royal Society Interface[9], which publishes “cross-disciplinary research at the interface between the physical and life sciences”. ** For a video of Prof. Niklas talking about plants and physics (and thereby promoting his book Plant Physics, co-authored with Hanns-Christof Spatz[10]), visit [11]. *** Since transpiration is primarily a xylem-related phenomenon, in the interests of balance we shouldn’t neglect that other long-distance vascular transport pathway – the phloem. For an update on the physics of phloem we recommend Kaare Jensen’s article[12]. **** For more about the phenomenon of electrical signals originating in the root of vascular plants, see Javier Canales et al.[13]. Image from: Wikimedia Commons References [1] https://physicsworld.com/a/plant-physics-the-april-2018-issue-of-physics-world-is-now-out/ [2] http://live.iop-pp01.agh.sleek.net/2018/03/26/a-flowering-success/ [3] http://labs.plantbio.cornell.edu/niklas/ [4] https://physicsworld.com/a/rocket-for-rocketeers/ [5] https://physicsworld.com/a/a-flowers-nano-powers/ [6] http://live.iop-pp01.agh.sleek.net/2018/03/26/top-tips-from-tree-tops/ [7] http://live.iop-pp01.agh.sleek.net/2018/03/26/rooted-in-physics/ [8] http://live.iop-pp01.agh.sleek.net/2018/03/26/is-photosynthesis-quantum-ish/ [9] http://rsif.royalsocietypublishing.org/ [10] http://press.uchicago.edu/ucp/books/book/chicago/P/bo12400940.html [11] https://www.youtube.com/watch?v=yj6aKaC5Auk [12] Current Opinion in Plant Biology 43 : 96 – 100 , 2018 ; https://doi.org/10.1016/j.pbi.2018.03.005 CrossRef Search ADS PubMed [13] Front. Plant Sci . 8 : 2173 ; doi: https://doi.org/10.3389/fpls.2017.02173 CrossRef Search ADS PubMed Flowers (it’s what angiosperms are all about!) View largeDownload slide View largeDownload slide Although one shouldn’t, it is easy to accept that flowers (the defining feature of the angiosperms, the flowering plants[1]) are ‘just there’ and get on with life in their quiet, seemingly unremarkable way. If one subscribed to that view, hardly any study of floral biology would be carried out, and we’d miss a lot of really interesting stuff. To demonstrate just what we might have been missing, this item showcases several insights into aspects of the biology of flowers that have surfaced so far in 2018. The intimate association between flowers and their pollinators provides many examples of ways in which flowers are adapted to specific pollinating organisms. In a study of field bean (Vicia faba, faba bean[2,3]), Emily Bailes et al.[4] investigated – among other aspects – the ‘operative strength’ of a flower. The operative strength is equivalent to the force a pollinator needs to exert to ‘trip’ a flower so that it can gain access to the pollen-containing interior. For the bean lines studied it ranged between 17.1 and 20.1 mN. Although those values might not mean all that much to the uninitiated it does imply that only certain insects will be powerful enough to open the flowers and therefore act as pollinators of this species. So, while this flower-accessing feat should be easily achieved by bees such as Bombus spp. (bumble-bees[5]) – which can exert over 200 mN force – it might prove problematic for weaker individuals of Apis mellifera (the honey bee[6]) which can only generate approx. 26 mN of force, and other smaller – and less powerful – bee species. This analysis therefore introduces another factor to bee (yes, ‘typo’ intended…) considered specifically in breeding field bean lines and varieties to suit available pollinators to maximise crop yield. It also has relevance more generally for other crops where insects must physically open the flowers to participate in pollinating activity. But, having allowed a suitable pollinator to access the flower, what’s the best way to ensure the visitor gets coated with pollen, the better to pollinate the next flower it visits? Callin Switzer et al. examined this phenomenon in mountain laurel (Kalmia latifolia[7,8]). This plant releases pollen in an explosive fashion when the anthers are triggered by appropriate insects. Although the pollen moves at only approx. 8 mph, its acceleration to achieve that is 400 times the acceleration due to gravity (!)[9]. Importantly, pollen-release appears only to be activated by insects such as bumble and honey bees, which are able to effectively transfer that pollen to other flowers. Combined with other aspects of the investigation, this study appears to settle the question of whether this pollen-release mechanism is for insect pollination (yes) or wind-dispersal of the pollen (apparently not). Once flowers have been pollinated, and fertilised, and seed formed, there’s the issue of how to jettison the seed so it lands sufficiently far away from the parent to have a chance of establishing itself as a new individual. To achieve this feat, the wild petunia (Ruellia ciliatiflora[10]) also employs an explosive release mechanism to launch its seeds at velocities exceeding 30 mph[11], and which land up to 7 m away from the parent plant. But there’s more to this phenomenon than that, as Eric Cooper et al.[12] have revealed. In particular, using high-speed video of the seeds’ flight, they show that the seeds spin at 1600 revolutions a second. This ‘backspin’ stabilises the flight of the seeds in such a way that it reduces the energy costs for their dispersal by up to a factor of five. Not only that, but the spinning reduces drag enabling the seeds to travel further from the parent plant than if they didn’t spin. Finally, we must appreciate that flowers are such a precious and all-important part of the angiosperm’s life-cycle that they need to be protected from those organisms that would eat them. We began this item with flower opening; we come full circle now and end with an example of floral closure (and, coincidentally, a third example of rapid movement in a plant related to floral biology). Investigating Drosera tokaiensis[13] – a sundew[14], which group of insectivorous plants are probably better-known for their mechanically-stimulated tentacles and leaves whose movements help to trap and wrap insect prey[15,16] – Kazuki Tagawa et al. report that its petals close rapidly in response to mechanical stimulation[17]. Petal closure was recognised and brought about artificially – by humans touching the flower with a pair of tweezers (e.g. [18]). However, the authors speculate that this phenomenon may function in nature as a defence against specialist florivores (organisms that consume flowers prior to seed coat formation[19]) that would eat the flowers rather than play a role in their pollination. Whether any such specialist sundew flower-eating organisms exist was not mentioned in the paper, but this remains an intriguing hypothesis that is ripe for testing. It could be further speculated, that sudden petal closure might startle or dislodge the florivore so that it falls off the flower on to the equally mechano-sensitive insect-trapping leaves of the plant and ends up as lunch itself; the would-be plant-predator plummets to be predated by the plant. Flowers, much more than meets the eye. [Ed. – Mindful that there might be ‘communications’ complaining that what’s described above doesn’t reflect any single flower that’s known to exist in nature, we would like to emphasise that the above account is not based on a flower of any known single flowering plant species, but is a compendium of insights into aspects of floral biology from several different species (as indicated by the different taxa specified).] Image from: Wikimedia Commons References [1] http://www.theplantlist.org/browse/A/ [2] http://powo.science.kew.org/taxon/urn:lsid:ipni.org:names:524737-1 [3] https://www.feedipedia.org/node/4926 [4] Ecology and Evolution 6 : 1 – 11 , 2018 ; doi: https://doi.org/10.1002/ece3.3851 [5] http://www.bumblebee.org/ [6] https://www.rspb.org.uk/birds-and-wildlife/wildlife-guides/other-garden-wildlife/insects-and-other-invertebrates/bees-wasps-ants/honey-bee/ [7] https://www.fs.fed.us/database/feis/plants/shrub/kallat/all.html [8] The American Naturalist ; doi: https://doi.org/10.1086/697220 [9] https://news.harvard.edu/gazette/story/2018/02/harvard-researchers-study-flower-that-catapults-pollen/ [10] https://www.inaturalist.org/taxa/168211-Ruellia-ciliatiflora [11] https://www.nytimes.com/2018/03/09/science/hairyflower-wild-petunia-seeds.html [12] J. R. Soc. Interface 15 : 20170901 .http://dx.doi.org/10.1098/rsif.2017.0901 [13] http://www.carnivorousplants.org/cp/taxonomy/Droseraspatulata [14] https://www.littleshopofhorrors.co.uk/sundews [15] https://www.youtube.com/watch?v=CUT7iZAilWE [16] Simon Poppinga et al. ( 2012 ) PLoS ONE 7 ( 9 ): e45735 ; https://doi.org/10.1371/journal.pone.0045735 CrossRef Search ADS PubMed [17] Plant Species Biology 33 : 153 – 157 , 2018 ; doi: https://doi.org/10.1111/1442–1984.12203 CrossRef Search ADS [18] https://uk.rs-online.com/web/p/tweezers/2386227/ [19] Andrew McCall and Rebecca Irwin , Ecology Letters 9 : 1351 – 1365 , 2006 ; doi: https://doi.org/10.1111/j.1461-0248.2006.00975.x3 CrossRef Search ADS PubMed © The Author(s) 2018. Published by Oxford University Press on behalf of the Annals of Botany Company. All rights reserved. For permissions, please e-mail: [email protected]. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)
Depletion of sucrose induces changes in the tip growth mechanism of tobacco pollen tubes2018 Annals of Botany
doi: 10.1093/aob/mcy043pmid: 29659664
Abstract Background and Aims Pollen tubes are rapidly growing, photosynthetically inactive cells that need high rates of energy to support growth. Energy can derive from internal and external storage sources. The lack of carbon sources can cause various problems during pollen tube growth, which in turn could affect the reproduction of plants. Methods We analysed the effects of energy deficiency on the development of Nicotiana tabacum pollen tubes by replacing sucrose with glycerol in the growth medium. We focused on cell growth and related processes, such as metabolite composition and cell wall synthesis. Key Results We found that the lack of sucrose affects pollen germination and pollen tube length during a specific growth period. Both sugar metabolism and ATP concentration were affected by sucrose shortage when pollen tubes were grown in glycerol-based media; this was related to decreases in the concentrations of glucose, fructose and UDP-glucose. The intracellular pH and ROS levels also showed a different distribution in pollen tubes grown in sucrose-depleted media. Changes were also observed at the cell wall level, particularly in the content and distribution of two enzymes related to cell wall synthesis (sucrose synthase and callose synthase). Furthermore, both callose and newly secreted cell wall material (mainly pectins) showed an altered distribution corresponding to the lack of oscillatory growth in pollen tubes. Growth in glycerol-based media also temporarily affected the movement of generative cells and, in parallel, the deposition of callose plugs. Conclusion Pollen tubes represent an ideal model system for studying metabolic pathways during the growth of plant cells. In our study, we found evidence that glycerol, a less energetic source for cell growth than sucrose, causes critical changes in cell wall deposition. The evidence that different aspects of pollen tube growth are affected is an indication that pollen tubes adapt to metabolic stress. Nicotiana tabacum, pollen tube, sucrose, glycerol, callose, cellulose, sucrose synthase, cell wall, tip growth INTRODUCTION Pollen grains, the male gametophyte of flowering plants, are unusual cells because they contain additional cells (the generative or sperm cells), which are in turn delimited by their own cell wall and plasma membrane. Pollen grains and pollen tubes are responsible for delivering sperm cells to the embryo sac, an essential process for sexual reproduction (Malho et al., 2006). Pollen tubes grow exclusively at the apex using a specialized form of polar growth, known as ‘tip growth’. Active intracellular streaming carries the cytoplasmic content (including gametes) towards the tube tip. To support this extremely rapid growth, pollen tubes require energy (Selinski and Scheibe, 2014). Carbohydrates and other storage compounds initially present in mature pollen are sufficient to support pollen survival and germination (Rodriguez-Garcia et al., 2003; Nashilevitz et al., 2009; Obermeyer et al., 2013; Zienkiewicz et al., 2013), but pollen tubes also require the uptake of external carbohydrate to support growth (Reinders, 2016). Since pollen tubes are fast-growing, photosynthetically inactive cells, an active respiratory and/or fermentative metabolism is required to support fast growth. The metabolism of pollen tubes (such as those of lily) is characterized by at least three different phases of respiration. A first, rapid phase occurs before germination and takes ~30 min; the second phase is characterized by decreasing respiration rate during germination, while the third phase is characterized by a further increase in the respiration rate (Rounds et al., 2011b). During the pre-germination step, cells showed a high rate of conversion of sugar to starch, suggesting that cells are preparing for the high energy demand of the next phases. Respiration is not likely to be sufficient because pollen tube growth is also supported by fermentation (Obermeyer et al., 2013). In fact, when respiration is inhibited, growth is briefly stopped but restarted using aerobic fermentation as energy source (Rounds et al., 2010). In the pollen of tobacco (Nicotiana tabacum), high respiration rates correspond to a significant level of ethanolic fermentation (Mellema et al., 2002). The energy demand of pollen tubes is initially maintained by internal storage (mainly starch), but subsequently it is supported by intake of external carbohydrates, mainly sucrose. Sucrose may enter pollen tubes through two routes. It can be catabolized into fructose and glucose through invertase (Goetz et al., 2016) or it enters using specific transporters (Lemoine et al., 1999; Stadler et al., 1999; Hirose et al., 2010). Monosaccharides can be directed towards glycolysis and then respiration and/or fermentation, while sucrose can also be converted into UDP-glucose through the enzymatic action of sucrose synthase (Persia et al., 2008). In the pollen tube, sucrose synthase is present in both the cytoplasm and the plasma membrane (in addition to the cell wall). Distribution of the enzyme is likely dependent on its phosphorylation status as well as on sucrose level, which affects the binding of sucrose synthase to actin filaments (Persia et al., 2008). Production of UDP-glucose (which can also be obtained from glucose) is important because this metabolite is the precursor of callose and cellulose. Callose, the most abundant cell wall component of pollen tubes, is produced by callose synthase, which is initially accumulated in Golgi-derived membranes and then delivered to the plasma membrane (Cai et al., 2011). The enzyme is inserted in the plasma membrane in an inactive state, but it is activated by proteolytic events (Brownfield et al., 2008). The role of callose is most likely to strengthen the cell wall of pollen tubes by forming a non-deformable sheath that counterbalances the turgor pressure (Williams, 2008; Winship et al., 2011). Turgor pressure is most probably a consequence of membrane potential maintained by the activity of transmembrane proton ATPases (Pertl et al., 2010; Lang et al., 2014). The latter are most likely also responsible for regulating the pH gradient at the apex and are likely involved in the deposition of callose plugs (Certal et al., 2008). Energy production (in the form of ATP) is also required for the synthesis of other cell wall components, such as cellulose and pectins. Pectins are synthesized in the Golgi bodies and then secreted via secretory vesicles that fuse specifically at the pollen tube apex (Wang et al., 2013), thereby allowing extension at the tip (Gu and Nielsen, 2013). Transport of Golgi and secretory vesicles, driven by the cooperation between actin filaments and microtubules (and associated proteins), is an additional energy-demanding process (Cai et al., 2015). The flow of energy, from uptake of sugars to their catabolism and then cell wall synthesis, requires a regulatory mechanism that must be necessarily interfaced with external signals and with the predetermined developmental programme of pollen tubes to precisely adjust the growth process both spatially and temporally. A primary role is played by the precise balance of ions and pH that accumulate preferentially in specific areas, thereby contributing to the polarization of pollen tubes (Winship et al., 2017). Moreover, a different distribution of ROS and pH values between the tube apex and subapex are considered essential for the correct growth of pollen tubes. Generally, protons are distributed unevenly, forming an intracellular gradient with slightly acidic pH values at the extreme apex and probably an alkaline band in the subapical domain (Feijò et al., 1999). The specific role of the proton gradient is still unclear, but it might be correlated with the activity of actin-depolymerizing factors (Chen et al., 2002). The pH gradient in pollen tubes is presumably the result of the influx of protons at the apex and their efflux at the subapex due to plasma membrane-associated H+-ATPase, the activity of which is also connected with the periodic deposition of callose plugs (Certal et al., 2008). Changes in extracellular proton concentration might also regulate the activity of pectin-methyl esterase (PME), the enzyme that converts methyl-esterified pectins into acid pectins (Li et al., 2002). Reactive oxygen species (ROS) are an inevitable consequence of aerobic metabolism, but are also generated in a controlled manner and used for a variety of functions, including pathogen defence and cell signalling. They are probably involved in the germination and growth of pollen tubes (Speranza et al., 2012; Smirnova et al., 2014). Reactive oxygen species are produced by enzymes such as NADPH oxidase at the apex of the pollen tube, and their production rate is strongly affected by Ca2+ (Potocky et al., 2007, 2012) and other factors, such as polyamines (Aloisi et al., 2015). The in vitro growth of pollen tubes is therefore an ideal model system for studying the metabolic pathways of plant cells. Because pollen tubes do not have chloroplasts (hence the metabolic contribution of respiration cannot be confused with photosynthesis), they are heterotrophic cells and import large quantities of sugars from extracellular environments. The simplicity of this cell model makes it possible to study the effects of carbon depletion by monitoring the effects on the germination and growth of pollen tubes and on levels of metabolites such as ATP and sugars. In this work, we have studied how the replacement of sucrose with less metabolizable energy compounds (glycerol) can affect the development of pollen tubes. The aim was to investigate how pollen tubes adapt from a metabolic and cellular point of view to the lack of energy. We analysed the distribution of pH and ROS as well as cell wall assembly by monitoring specific polysaccharides and the cytological distribution of two critical enzymes (sucrose synthase and callose synthase). In addition, we tested how metabolic stress affects the primary function of pollen tubes, i.e. the transport of gametes. MATERIALS AND METHODS Pollen growth in different media Seeds of Nicotiana tabacum (tobacco) were from the Seed and Plant Collection of the Botanical Garden at Siena University. Tobacco plants were grown under standard conditions in greenhouses and pollen was collected from opening flowers from June to August; pollen was dehydrated and stored at −20 °C. When needed, pollen was hydrated at room temperature over night in a moist chamber, then the pollen was placed in four different growth media: (1) BKS, BK medium supplemented with 12 % sucrose (Brewbaker and Kwack, 1963); (2) BKP, BK medium with 13 % polyethylene glycol 3350 (PEG); (3) BKM, BK medium with 11 % maltose; (4) BKG, BK medium with 3.3 % glycerol. We measured the germination rate and length of at least 100 pollen tubes in all assays to statistically evaluate both parameters. Growth media were chosen to maintain the osmolarity when compared with control (i.e. BK with 12 % sucrose). We referred to data presented in Zerzour et al. (2009). First, the growth rate was tested for each selected medium within 150 min; images of growing pollen tubes were taken at different intervals. For each image, we measured the length of germinated tubes using ImageJ software (https://imagej.nih.gov/ij/index.html) to evaluate the growth rate. Subsequently, we measured the growth rate of pollen tubes grown in BKS and BKG (the growth media selected) for 7 h, measuring the pollen tube length at intervals of 60 min. Protein extraction from cytosol, membranes and cell wall We extracted proteins of pollen tubes from three different intracellular compartments: the cytosol, the membranes and the cell wall. The extraction protocol was performed for pollen grown in BKS and BKG at three different germination times: 4, 5 and 6 h (the logic behind the choice of these three intervals is explained later). After germination, 500 mg of pollen was collected and centrifuged at 135 g for 5 min, then washed with HEMS or HEMG buffer (50 mm HEPES pH 7.5, 2 mm EGTA, 2 mm MgCl2, with either 12 % sucrose [HEMS] or 3.3 % glycerol [HEMG]) After washing, samples were centrifuged again at 135 g for 5 min, the supernatants were discarded and lysis buffer (50 mm HEPES pH 7.5, 2 mm EGTA, 2 mm MgCl2, protease inhibitors, 1 mm dithiothreitol) was added to the pellets. Mechanical lysis of pollen was carried out in a cold room (4 °C) using a Potter-Elvehjem homogenizer with 40 strokes. Samples were then centrifuged at 500 g for 10 min at 4 °C; pellets and supernatants were subjected to different processes. Pellets, containing cell wall proteins, were washed three times with HEM buffer by centrifuging at 15 000 g for 10 min at 4 °C after each wash. The last pellet (referred to as the ‘cell wall sample’) was resuspended with Laemmli sample buffer (Laemmli, 1970) for 1-D electrophoresis. The supernatants, containing both cytosol and membrane proteins, were centrifuged at 100 000 g, 4 °C, for 45 min. The resulting pellets (referred to as the ‘membrane sample’) were resuspended with Laemmli sample buffer; the corresponding supernatants were supplemented with 4 volumes of 20 % TCA in cold acetone (with 0.07 % 2-mercaptoethanol) to allow protein precipitation overnight at −20 °C. After precipitation, samples were centrifuged at 15 000 g for 15 min at 4 °C, the supernatants were discarded and the pellets were washed twice with 100 % cold acetone (with 0.07 % 2-mercaptoethanol) and resuspended in Laemmli sample buffer. The final sample is referred to as the ‘cytosol sample’. Determination of protein concentration The protein concentration of samples was determined using a commercial kit (2-D Quant Kit, GE HealthCare). The protocol was performed exactly as described in the instruction manual using BSA as reference. Each sample was analysed in three replicates using a Shimadzu UV-160 spectrophotometer at 480 nm. 1-D electrophoresis Separation of proteins by 1-D electrophoresis was performed on precast gel (Criterion Tris–HCl PreCast Gel, 10 %, Bio-Rad) using a Criterion cell (Bio-Rad) equipped with a Bio-Rad PowerPac 300 at 200 V for ~55 min. Gels were run according to Laemmli (1970) using TGS (25 mm Tris–HCl pH 8.3, 192 mm glycine, 0.1 % SDS) as running buffer. Gels were stained with Bio-Safe Coomassie blue (Bio-Rad) as described in the instruction protocol. Western blotting and image analysis Transfer of proteins from gels to nitrocellulose membranes was performed using a Trans-Blot Turbo Transfer System (Bio-Rad) according to the manufacturer’s instructions. Quality of blotting was determined by checking the transfer of precision pre-stained molecular standards (Bio-Rad). After blotting, membranes were blocked overnight at 4 °C in 5 % ECL Blocking Agent (GE HealthCare) in TBS (20 mm Tris pH 7.5, 150 mm NaCl) plus 0.1 % Tween-20. After washing with TBS, membranes were incubated with the primary antibodies for 1 h at room temperature. The antibody to sucrose synthase was used at the dilution of 1:1000 (Persia et al., 2008) while the antibody to callose synthase was diluted 1:300 (Cai et al., 2011). Subsequently, membranes were washed several times with TBS and then incubated for 1 h with anti-rabbit immunoglobulin G peroxidase-conjugated secondary antibodies, purchased from Bio-Rad and diluted 1:5000. Blots were finally incubated with the Clarity reagents (Bio-Rad). Images of gels and blots were acquired using a Fluor-S apparatus (Bio-Rad) and analysed with Quantity One software (Bio-Rad). Exposure times were 30–60 s for blots and 5–7 s for Coomassie-stained gels. Analysis of blots was performed with Quantity One software. All blots were developed using identical conditions from substrate incubation to exposure time. All images were processed correspondingly using the Autoscale command (to improve the quality of gels and blots) and the Background Subtraction command (to remove the background noise). The relative intensity of single spots was calculated with the Volume tool of Quantity One software. Blots were performed in triplicate. Results were exported and graphed with Microsoft Excel. Kymograph analysis of pollen tubes Growing pollen tubes were observed using an inverted microscope (Diaphot TMD) with a 40× objective (Nikon). Video sequences of growing pollen tubes were captured using a CCD camera (C2400-75i, Hamamatsu Photonics) connected to an Argus-20 processor (Hamamatsu) and then to a video capture system (Cai et al., 2000). Video clips of pollen tubes were captured as MPEG-2 files at a resolution of 720 × 576 pixels using the software PCTV Center. MPEG-2 files were converted to AVI (MJPEG compression) with the software VirtualDub (http://virtualdub.org/) and then opened in ImageJ software (http://rsbweb.nih.gov/ij/index.html).Video sequences were analysed with the plugin Kymograph (written by Jens Rietdorf, FMI, Basel, and Arne Seitz, EMBL, Heidelberg) to measure the speed of moving objects in a series of images. During kymograph analysis, we analysed and measured the grey values in each region of interest (ROI) selected manually for each frame in the image series. A new image (a kymograph or space–time graph) was generated, in which the x-axis is the time axis (the unit is represented by the frame interval) while the y-axis indicates the movement rate of the ROI (the unit of measurement is the distance in pixels travelled by the object). The speed of objects was measured directly by the plugin. At least 20 pollen tubes for each sample were analysed. Immunofluorescence microscopy Indirect immunofluorescence microscopy in pollen tubes was performed according to standard procedures (Cai et al., 2011). After fixation with 3 % paraformaldehyde in PM buffer (50 mm PIPES pH 6.9, 1 mm EGTA, 0.5 mm MgCl2) for 30 min, 50 mg of pollen sample was washed with PM for 10 min and then incubated with 1.5 % Cellulysin (Sigma) for 7 min in the dark. After two washes with PM buffer, samples were incubated with the primary antibodies. The anti-sucrose synthase antibody (Heinlein and Starlinger, 1989; Persia et al., 2008) was used at 1:100 dilution while the anti-callose synthase antibody was used at 1:50. Antibodies were incubated at 4 °C overnight. Following two washes with PM buffer, samples were incubated with a goat anti-rabbit secondary antibody conjugated to Alexa Fluor 488 (Invitrogen) diluted 1:150 for 45 min in the dark. Antibodies JIM5 (against acid pectins) and JIM7 (against methyl-esterified pectins) were purchased from PlantProbes (http://www.plantprobes.net). Both antibodies were used at 1:5 dilution for 3 h at room temperature. For JIM5 and JIM7 antibodies, incubation with Cellulysin was not carried out. As secondary antibody, a goat anti-rat antibody conjugated with Alexa 594 was used, diluted 1:50 and incubated for 45 min at 37 °C. After two washes in PM buffer, samples were placed on slides and covered with a drop of Citifluor. Observations were made using a Zeiss Axio Imager microscope with a 63× objective. Images were acquired with an AxioCam MRm camera using the software AxioVision. Images of higher quality were obtained using structured illumination. In controls, primary antibodies were omitted. For the measurement of fluorescence intensity in the cell border, original ZVI files generated by the AxioVision software were imported into ImageJ and the signal intensity was measured with a segmented line sufficiently wide to cover the signal (~2 μm) by starting from the apex of pollen tubes in the direction of the pollen grain. The obtained measure was normalized against the background by selecting a large area outside the pollen tube. To quantify the intracytoplasmic fluorescence signal, individual images were imported into ImageJ and a segmented line was depicted to cover the ROI. The line thickness was such that it largely occupied the cell cytoplasm. The signal was measured with the Analyze > Plot profile command. Analysis of newly secreted cell wall material, generative cell movement, pH and ROS Labelling of newly secreted cell wall material was performed using propidium iodide (PI). Rounds et al. (2011a) showed that levels of PI fluorescence matched those of GFP-labelled pectin methyl esterase. The authors also showed competitive reduction of PI fluorescence after binding of Ca2+ to pectins (indicating that PI fluorescence reflects pectin binding). Despite some evidence that PI may label pectins, it is more significant that PI labels the area where new cell wall material is secreted. Measurement of PI fluorescence was performed along the cell edge using the Segmented Line tool of ImageJ with a selection width of ~2 μm. After subtracting the background (as measured in representative pollen tubes), we calculated the average values and standard deviation. The generative cell test was performed using 4′,6-diamidino-2-phenylindole (DAPI) staining in pollen tubes grown in both sucrose- and glycerol-based media. We analysed the generative cell movement for 4 consecutive hours, after 3 h of germination. Pollen tubes were placed on glass slides and then stained with DAPI; after a few minutes of incubation, observations were made using a Zeiss Axio Imager fluorescence microscope equipped with a 63× objective. Images were captured with an MRm AxioCam video camera using AxioVision software. ZVI files were opened in ImageJ, which was used to measure the total length of pollen tubes and the distance covered by generative cells from the pollen grain. To statistically evaluate our observations, we performed two consecutive tests: first, we analysed the data by analysis of variance (ANOVA); secondly, we performed Student’s t-test. In both cases, we used the Data Analysis module of Excel software. The BCECF-AM [2′,7′-bis-(2-carboxyethyl)-5-(and-6)-carboxyfluorescein, acetoxymethyl] ester probe was used to visualize the proton levels (i.e. pH) in tobacco pollen tubes (Fricker et al., 1997). A final concentration of 5 µm was obtained from a dimethyl sulphoxide (DMSO) stock solution of 1 mm; the required volume was directly added to both the sucrose and the glycerol germination medium. The cytosolic pH was immediately visualized (within 5 min) after addition of the probe to prevent uptake of the pH probe by organelles. We detection ROS as described by Aloisi et al. (2015) using the fluorescent ROS indicator dye 2′,7′-dichlorodihydrofluorescein diacetate (DCFH2-DA; Molecular Probes). Samples were observed with a Zeiss Axio Imager fluorescence microscope equipped with a 63× objective, an MRm AxioCam video camera and structured illumination. Control of the ROS probe involved the absence of the probe itself (no signal except for the autofluorescence of pollen grains) and the addition of fluorescein diacetate (in which case we observed a signal uniformly diffused throughout the pollen tube) (not shown). To highlight the differences in fluorescence intensity between different regions of pollen tubes, when necessary single greyscale images were transformed into pseudocoloured images using ImageJ software, specifically the command Image > Lookup tables > 16 colours. Immunogold electron microscopy Immunogold labelling of tobacco pollen tubes was performed according to the protocol described in Li et al. (1995). The antibody to callose synthase was used at the dilution of 1:100 in 50 mm Tris–HCl pH 7.6, 0.9 % NaCl, 0.1 % Tween 20, 0.2 % BSA. The goat anti-rabbit secondary antibody was conjugated with 15 nm gold particles (BioCell). Images were captured with a Philips Morgagni 268 D transmission electron microscope (TEM) set at 80 kV and equipped with a MegaView II CCD camera (Philips Electronics, Eindhoven, The Netherlands). Samples were incubated with 5 % normal goat serum (Invitrogen) for 20 min at room temperature to prevent binding to non-specific sites. Sections were incubated with the primary antibody for 1 h and then washed (three or four times) in 50 mm Tris–HCl pH 7.6, 0.9 % NaCl, 0.1 % Tween 20 for 30 min. After drying, samples were incubated with the gold-conjugated secondary antibody for 15 min at room temperature. After washing for 30 min and in H2O for 10 min, sections were counterstained with 2 % uranyl acetate in H2O for 10–20 min, carefully washed in H2O for 15 min and then counterstained with lead citrate for 5–10 min. Analysis of ATP/ADP and sugars by high-performance liquid chromatography (HPLC) We analysed ATP and ADP by HPLC (Perkin Elmer Series 200) following the method tested by Liu et al. (2006). Fifty milligrams of pollen was collected by centrifugation at 135 g for 5 min from both sucrose and glycerol germination media at regular intervals of 1 h and resuspended in boiling water (1 mL). Complete disintegration and rupture of cells were performed with a Potter-Elvehjem homogenizer with 40 strokes per sample. The homogenate was centrifuged at 15 000 g for 15 min at room temperature. The supernatants were transferred to vials and 20 μL of sample was injected into a solid stationary-phase C18 column (75 × 4.6 mm, particle size 5 μm). The mobile phase was a binary mobile phase gradient (solvent A, 10 mm phosphate buffer pH 7; solvent B, acetonitrile) with the following gradient: 0 min, 100 % solvent A, 0 % solvent B; 2 min, 95 % A, 5 % B; 4 min, 80 % A, 20 % B; 5.3 min, 75 % A, 25 % B; 6 min, 100 % A, 0 % B. The following parameters were used: flow rate 0.3 mL min−1; room temperature; approximate elution time 6 min for ATP and 7 min for ADP. Identification of different components was obtained by programming a DAD 235C spectrophotometric detector with excitation wavelength 254 nm. Sugar analysis was performed by lysing pollen as described above; the final supernatants were examined by isocratic HPLC analysis with a Waters Sugar-Pak I ion-exchange column (6.5 × 300 mm) at a temperature of 90 °C and using a Waters 2410 refractive index detector. MilliQ grade water (pH 7) was used as a mobile phase with a flow rate of 0.5 ml min−1; an injection loop of 20 μL was used for all samples and standards (sucrose, glucose and fructose). RESULTS Germination rate and length of pollen tubes are affected by growth medium To understand whether replacement of sucrose with another carbon source can affect pollen metabolism and growth, we initially measured the germination rate and length of pollen tubes (Fig. 1) while growing in four different germination media. First, we measured the pollen tube length after growing for 150 min (Fig. 1A). We found that pollen grains germinated normally in the presence of sucrose (as standard condition), while use of maltose and PEG did not show relevant differences in pollen tube length. On the contrary, we found a significant reduction in the length of pollen tubes grown in glycerol-supplemented media (Fig. 1A). Reduction in length was clear after 120 min of growth and statistical analysis showed that length was significantly different from 100 min of growth onwards (asterisks). Because we were looking for pollen tubes metabolically inhibited by scarcity of carbon source, we continued the analysis only using sucrose- and glycerol-based media. Fig. 1. View largeDownload slide Pollen tube growth under standard conditions and in the presence of different osmotic compounds. (A) Pollen tube growth in BK medium supplemented with different osmotic molecules (sucrose, glycerol, maltose, PEG). The analysis was carried out for 150 min. Statistically significant differences were observed between growth in sucrose- and glycerol-based media (in all the panels the asterisks correspond to P <0.05). Bars indicate standard deviation. (B) Analysis of pollen tube length after growth in BK medium supplemented with either sucrose or glycerol. The analysis was carried out for 7 h. Statistically significant differences were determined after 4, 5 and 6 h of germination (asterisks). For both analyses, trend lines and corresponding R2 values are reported. (C) Analysis of germination of pollen tubes in BK medium supplemented with either sucrose or glycerol. This parameter was monitored for 6 h. Asterisks indicate the time points when germination in BK medium plus sucrose was statistically higher than that in BK medium plus glycerol. Fig. 1. View largeDownload slide Pollen tube growth under standard conditions and in the presence of different osmotic compounds. (A) Pollen tube growth in BK medium supplemented with different osmotic molecules (sucrose, glycerol, maltose, PEG). The analysis was carried out for 150 min. Statistically significant differences were observed between growth in sucrose- and glycerol-based media (in all the panels the asterisks correspond to P <0.05). Bars indicate standard deviation. (B) Analysis of pollen tube length after growth in BK medium supplemented with either sucrose or glycerol. The analysis was carried out for 7 h. Statistically significant differences were determined after 4, 5 and 6 h of germination (asterisks). For both analyses, trend lines and corresponding R2 values are reported. (C) Analysis of germination of pollen tubes in BK medium supplemented with either sucrose or glycerol. This parameter was monitored for 6 h. Asterisks indicate the time points when germination in BK medium plus sucrose was statistically higher than that in BK medium plus glycerol. We proceeded by testing the growth of pollen tubes for longer times (up to 7 h). While pollen tubes grown in sucrose-based medium exhibited a typical tube elongation rate (up to 350 μm after 7 h), the length of pollen tubes grown in glycerol-based medium was significantly reduced. Statistical analysis showed that the length of pollen tubes grown in glycerol-based medium was different after 4 h of growth until 6 h of growth. After 7 h of growth, pollen tubes in glycerol-based medium increased their growth rate. The difference between sucrose- and glycerol-based media was also confirmed by analysis of linear regression as reported in Fig. 1B. In sucrose-based medium, pollen tube length increased linearly (R2 = 0.9945) while in glycerol-based medium the value of R2 was typically around 0.948. The germination rate of pollen grains also changed significantly; while in sucrose-based medium we observed a higher percentage of germinated pollen (around 85–90 %) after 6 h, in glycerol-based medium we found that pollen grains germinated faster than the control after 1 h of incubation, but germination rate slowed down significantly after 4 h because the percentage of germinated grains remained stable but lower than in the control (Fig. 1C). The growth rate of pollen tubes was also subjected to kymographic analysis. In the standard condition (Fig. 2A) the speed of growth was constant, as shown by the linearity of the kymograph profile. In the example illustrated, the pollen tube showed a growth speed of ~1.37 μm min−1. In addition, it was possible to observe the typical oscillation of growth, i.e. slower phases of growth preceded/followed by peaks of fast growth (the oscillation between slow and rapid growth phases is shown by the magnification of a kymograph of a control pollen tube in Fig. 2C). When pollen grains were grown in glycerol-based medium (Fig. 2B) the speed of growth was different, as shown by the kymograph profile. We observed different speeds of growth because pollen tubes were growing like the controls during the first 2–3 h (1.3–1.5 μm min−1) but their growth rate decreased significantly compared with controls soon after (to 0.6 μm min−1). After 6 h of growth, the growth rate increased again, to 1.3 μm min−1. In addition, we did not detect any oscillation in the growth rate, as depicted in the magnification of the kymograph of a metabolically stressed pollen tube in Fig. 2D. The difference in growth rate is more appreciable in Fig. 2E. In glycerol-based medium, the growth rate of pollen tubes decreased significantly after 180–200 min but increased again after 6 h of growth. Interestingly, around 500 min (~8 h of growth) the elongation rate was comparable to controls in sucrose-based medium. Fig. 2. View largeDownload slide Growth of pollen tubes as analysed by kymography in BK medium supplemented with either sucrose or glycerol. (A) Typical growth profile of pollen tubes in BK medium plus sucrose. The velocity reported in the image refers to a frequently observed average value. Time and distance bars are shown at the top left. (B) Kymographic analysis of pollen tubes grown in BK medium plus glycerol. The growth pattern was arbitrarily divided into segments in which speed was relatively constant. The values shown indicate average speeds for each segment. Note the marked decrease in growth after the third hour of germination. Time and distance bars are shown at the top left. (C, D) Detail of the growth profile of pollen tubes in BK medium plus sucrose (C) or plus glycerol (D). Note the oscillatory profile in C as opposed to the linear profile in D. (E) Analysis of average growth speed as calculated by kymography. Note the minimum growth value around the fifth hour in BK medium plus glycerol. Fig. 2. View largeDownload slide Growth of pollen tubes as analysed by kymography in BK medium supplemented with either sucrose or glycerol. (A) Typical growth profile of pollen tubes in BK medium plus sucrose. The velocity reported in the image refers to a frequently observed average value. Time and distance bars are shown at the top left. (B) Kymographic analysis of pollen tubes grown in BK medium plus glycerol. The growth pattern was arbitrarily divided into segments in which speed was relatively constant. The values shown indicate average speeds for each segment. Note the marked decrease in growth after the third hour of germination. Time and distance bars are shown at the top left. (C, D) Detail of the growth profile of pollen tubes in BK medium plus sucrose (C) or plus glycerol (D). Note the oscillatory profile in C as opposed to the linear profile in D. (E) Analysis of average growth speed as calculated by kymography. Note the minimum growth value around the fifth hour in BK medium plus glycerol. ATP/ADP and sugar concentrations changed in different growth conditions In order to get a first assessment of the metabolic state of pollen tubes grown in the two different media, we analysed the concentrations of ATP/ADP and the main sugars (glucose, fructose and UDP-glucose) in pollen tubes. All metabolites were analysed using HPLC during the first 6 h of growth. We found clear differences in ATP concentrations between pollen tubes grown in sucrose- and glycerol-based media. The ATP concentration was relatively constant for the first 4 h in sucrose-based medium and slightly decreased after the fifth and sixth hours of growth. On the contrary, in glycerol-based medium the ATP concentration decreased specifically during the first 2 h of growth and progressively stabilized at lower levels at the end of treatment. The concentration of ADP was found to be relatively constant in both treatments; in fact we did not find any significant change, and generally the amount of ADP was lower in glycerol-based medium than in sucrose-based medium (Fig. 3A). Fig. 3. View largeDownload slide Analysis of metabolites in pollen tubes grown in BK medium supplemented with sucrose or glycerol. The analysis was performed after each hour of germination for a period of 6 h. (A) Measurement of the concentrations of ATP and ADP in pollen tubes grown in BK plus sucrose (BKS) or plus glycerol (BKG). (B) Concentration analysis of UDP-glucose in the two experimental conditions. (C) Analysis of the concentration of glucose and fructose in pollen tubes grown in BK medium plus glycerol. No signal was found at the first hour because values were below the instrumental detection limit. (D) Analysis of glucose and fructose in pollen tubes grown in BK medium plus sucrose. In all cases, bars indicate the standard deviation. Fig. 3. View largeDownload slide Analysis of metabolites in pollen tubes grown in BK medium supplemented with sucrose or glycerol. The analysis was performed after each hour of germination for a period of 6 h. (A) Measurement of the concentrations of ATP and ADP in pollen tubes grown in BK plus sucrose (BKS) or plus glycerol (BKG). (B) Concentration analysis of UDP-glucose in the two experimental conditions. (C) Analysis of the concentration of glucose and fructose in pollen tubes grown in BK medium plus glycerol. No signal was found at the first hour because values were below the instrumental detection limit. (D) Analysis of glucose and fructose in pollen tubes grown in BK medium plus sucrose. In all cases, bars indicate the standard deviation. We also investigated the concentrations of three important sugars of pollen tubes, namely UDP-glucose, glucose and fructose. The concentration of UDP-glucose during pollen tube growth in both media remained relatively constant for 6 h; nevertheless, the relative concentration of UDP-glucose in pollen tubes grown in sucrose-based medium was consistently and significantly higher than that observed in pollen tubes grown in glycerol-based medium (Fig. 3B). However, no specific trend of variation could be deduced from the measurements. Further analysis concerned the relative quantity of glucose and fructose in pollen tubes grown in two different media. In standard conditions, the relative concentrations of both monosaccharides increased during growth (Fig. 3C) probably because of progressive cleavage of sucrose. On the contrary, in glycerol-based medium the concentrations of both sugars oscillated similarly during growth, with a peak of concentration after 3–4 h of growth. At the first hour of analysis, the relative concentration of these monosaccharides was below the limit of instrumental detection. In any case, the most striking difference concerned the relative concentrations between sucrose- and glycerol-based media because in the latter concentration the values were at least a few hundred times lower than those measured in sucrose-based medium (Fig. 3D). Distribution of pH and ROS is affected by growth medium Variations in pH can affect the expansion of apical-growing cells. Therefore, to understand whether different media might affect pH values in pollen tubes, we analysed the pH distribution in pollen tubes grown for 5 h (the time during which major effects on growth were observed) in both sucrose- and glycerol-based medium. We found that pH in pollen tubes grown under standard conditions was low (acid) at the apex and increased (thus more basic) in the subapex and shank of pollen tubes (Fig. 4A; left panel, bright field; right panel, fluorescence). On the contrary, pollen tubes grown in glycerol-based medium showed a relevant difference because pH was relatively constant along the tube and no clear differences could be appreciated (Fig. 4B; left panel, bright field; right panel, fluorescence). To confirm such observations, we also measured the fluorescence signals from the very tip down along the tube axis. The corresponding graph, as obtained by analysing the fluorescence signal of several pollen tubes and by normalization against the background, illustrates the differences in pH distribution between the two growth media (Fig. 4C). In both graphs, the measured values were compared with the highest observed value (set as 100 %). Fig. 4. View largeDownload slide Analysis of pH distribution in pollen tubes grown for 5 h in sucrose- and glycerol-based media. (A) The pH in pollen tubes grown under standard conditions is relatively low (acid) at the apex but increases (becomes more basic) in the subapex and shank of pollen tubes. (B) In pollen tubes grown in glycerol-based medium, the pH is relatively constant along the tube. (C) Measurement of the fluorescence signal from the apex along the axis of pollen tubes. The graph illustrates the differences in pH distribution in pollen tubes grown in different growth media. Values were normalized against the background and referred to the highest value, which was set as 100 %. Scale bars = 10 μm. Fig. 4. View largeDownload slide Analysis of pH distribution in pollen tubes grown for 5 h in sucrose- and glycerol-based media. (A) The pH in pollen tubes grown under standard conditions is relatively low (acid) at the apex but increases (becomes more basic) in the subapex and shank of pollen tubes. (B) In pollen tubes grown in glycerol-based medium, the pH is relatively constant along the tube. (C) Measurement of the fluorescence signal from the apex along the axis of pollen tubes. The graph illustrates the differences in pH distribution in pollen tubes grown in different growth media. Values were normalized against the background and referred to the highest value, which was set as 100 %. Scale bars = 10 μm. Results obtained after analysis of ROS distribution were substantially like those obtained for the pH analysis. Pollen tubes grown for 5 h in sucrose-based medium were characterized by a high concentration of ROS in the apical region (the first 10–15 µm) while the subapical and shank regions were poor in ROS (Fig. 5A; left panel, bright field; right panel, fluorescence). Conversely, the distribution of ROS in pollen tubes grown for 5 h in glycerol-based medium was substantially different because it was quite uniform throughout the pollen tube (Fig. 5B; left panel, bright field; right panel, fluorescence). A statistical analysis of the data confirmed the visual interpretation; pollen tubes grown in sucrose-based medium exhibited a remarkable peak of ROS at the apex (Fig. 5C, solid line) while pollen tubes grown in glycerol-based medium showed a very homogeneous distribution of ROS (Fig. 5C, dashed line). Fig. 5. View largeDownload slide Analysis of the presence of ROS in pollen tubes after 5 h of growth in BK medium plus sucrose or glycerol. (A) A pollen tube (left) and the corresponding fluorescence signal of the ROS probe (pseudocoloured, right) after 5 h of growth in BKS. (B) A pollen tube (left) and the corresponding fluorescent signal of ROS (pseudocoloured, right) after 5 h of growth in BKG. (C) Distribution profile of ROS as measured from the apex of pollen tubes grown in BKS (solid line) and BKG (dashed line). The grey areas along the two profiles indicate the standard deviation. Fig. 5. View largeDownload slide Analysis of the presence of ROS in pollen tubes after 5 h of growth in BK medium plus sucrose or glycerol. (A) A pollen tube (left) and the corresponding fluorescence signal of the ROS probe (pseudocoloured, right) after 5 h of growth in BKS. (B) A pollen tube (left) and the corresponding fluorescent signal of ROS (pseudocoloured, right) after 5 h of growth in BKG. (C) Distribution profile of ROS as measured from the apex of pollen tubes grown in BKS (solid line) and BKG (dashed line). The grey areas along the two profiles indicate the standard deviation. Acidic and methyl-esterified pectins are differently localized in pollen tubes grown in different media To deepen our understanding of the way in which pollen tubes grow, we monitored the secretion of new cell wall material at the apex. Initially we used labelling with PI. After 5 h of growth, the cell wall material stained with PI accumulated mainly in the apical cell wall and, to a lesser extent, in the subapical cell wall of pollen tubes grown in sucrose-based medium (arrow in Fig. 6A). Unlike controls, pollen tubes grown for 5 h in glycerol-based medium showed a significant difference because the cell wall labelled with PI was homogeneous (Fig. 6B) and relatively similar between the apex/sub-subapex and the shanks of pollen tubes. This suggests the absence of preferential secretion/accumulation of newly synthesized pectins. To further investigate the pattern of pectin secretion, we labelled methyl-esterified pectins with JIM7 antibody and acidic pectins with JIM5 antibody. In sucrose controls, methyl-esterified pectins accumulated abundantly at the apex of pollen tubes and much less in distal regions (Fig. 6C). In contrast, the distribution of methyl-esterified pectins in glycerol-grown pollen tubes was relatively similar in different regions of pollen tubes (Fig. 6D). This difference in pectin distribution between sucrose- and glycerol-grown cells was confirmed by the analysis of acidic pectins. In pollen tubes grown in sucrose the antibody JIM5 provided a classic ring pattern (typical of pulsed growing pollen tubes) (Fig. 6E); this pattern of distribution was made clear by image analysis along the pollen tube axis, which highlighted the regularity of signal peaks of acidic pectins (Fig. 6F). In pollen tubes grown in glycerol, the ring pattern was maintained (Fig. 6G) but with two major differences, as shown by image analysis (Fig. 6H). Firstly, there was an intense signal at the apex of pollen tubes; secondly, signal peaks were much less intense, sometimes not discernible from background noise. This suggests that acidic pectins are more evenly distributed than sucrose controls. Fig. 6. View largeDownload slide Deposition of pectins in the cell wall. (A) Typical pollen tube grown in BKS medium and labelled with PI. Note the strong signal in the apical region of the pollen tube (arrow). (B) A pollen tube grown in BKG and labelled with the PI probe. The signal is homogeneous along the tube edge. (C) A typical pollen tube grown in BKS and labelled with JIM7 antibody; the apex is strongly stained. (D) A typical pollen tube grown in BKG and labelled with JIM7; the signal is homogeneous along the cell wall. (E) A pollen tube grown in BKS and labelled with JIM5 antibody; acidic pectins are distributed at regular intervals, as shown by the graph in (F). (G) A pollen tube grown in BKG and labelled with JIM5; as shown by the analysis in (H), acidic pectins accumulate at the apex and are more homogeneous along the pollen tube. Scale bars = 10 μm. Fig. 6. View largeDownload slide Deposition of pectins in the cell wall. (A) Typical pollen tube grown in BKS medium and labelled with PI. Note the strong signal in the apical region of the pollen tube (arrow). (B) A pollen tube grown in BKG and labelled with the PI probe. The signal is homogeneous along the tube edge. (C) A typical pollen tube grown in BKS and labelled with JIM7 antibody; the apex is strongly stained. (D) A typical pollen tube grown in BKG and labelled with JIM7; the signal is homogeneous along the cell wall. (E) A pollen tube grown in BKS and labelled with JIM5 antibody; acidic pectins are distributed at regular intervals, as shown by the graph in (F). (G) A pollen tube grown in BKG and labelled with JIM5; as shown by the analysis in (H), acidic pectins accumulate at the apex and are more homogeneous along the pollen tube. Scale bars = 10 μm. Lack of sucrose changes the accumulation and localization of sucrose synthase in pollen tubes As a next step in evaluating the adaptation of pollen tubes to new growing conditions, we analysed sucrose synthase as a critical enzyme in the allocation of carbon between respiration and cell wall synthesis. To get information on changes in protein content, we performed SDS–PAGE analysis of proteins extracted from pollen tubes grown for 4, 5 and 6 h in both sucrose- and glycerol-based media. Proteins were extracted from three different cell compartments, namely the cytosol (Fig. 7A), the cell wall (Fig. 7B) and the membranes (Fig. 7C). A preliminary visual analysis showed no significant difference in the protein pattern between different growth conditions or between different growth times within each cellular fraction. To investigate the accumulation rates of sucrose synthase, we analysed its relative content by immunoblot analysis of all fractions (Fig. 7D). The results confirmed that the relative content of sucrose synthase could be affected by metabolic stress. Specifically, we found that in the standard condition (sucrose-based medium) the accumulation of sucrose synthase was not homogeneous in the cytosol within the time of analysis, with a significant reduction after 6 h of growth (Fig. 7D, lanes 2–4). Comparatively, an uneven accumulation pattern could also be observed in the cell wall protein fraction (lanes 8–10), whereas membrane-associated sucrose synthase appeared to reach the highest amount after 5 h of growth in sucrose-based medium (lanes 14–16). These data indicate that the amount of sucrose synthase oscillated during the growth of pollen tubes. When the levels of sucrose synthase were analysed in the protein fractions of pollen tubes grown in glycerol-based medium, an uneven pattern of accumulation could be observed in the protein fractions from cytosol (lanes 5–7), cell wall (lanes 11–13) and membranes (lanes 17–19). To clarify whether oscillation in the levels of sucrose synthase was comparable between pollen tubes grown in either sucrose- or glycerol-based media, we compared different blot results quantitatively. In pollen tubes grown in sucrose medium, the amount of sucrose synthase in the cytosol and membranes decreased progressively from 4 to 6 h of growth, resulting in the accumulation of sucrose synthase in the cell wall protein fraction (Fig. 7E). This kind of behaviour was not observed in pollen tubes grown in glycerol medium. In this case, the levels of sucrose synthase in the cytosol slightly decreased while the relative amount of cell wall-associated sucrose synthase remained constant and the levels of enzyme in the membrane fraction increased slightly. Fig. 7. View largeDownload slide Relative accumulation of sucrose synthase in the cytosol, membranes and cell walls of pollen tubes grown in BKS and BKG. (A) Electrophoretic analysis of proteins extracted from the cytosol of pollen tubes grown for 4, 5 and 6 h in the two different media. Molecular weight standards (kDa) are indicated on the left. (B) Profile of proteins extracted from the cell wall of pollen tubes grown for 4, 5 and 6 h in BKS and BKG. (C) Electrophoresis of proteins extracted from the membrane fraction of pollen tubes grown under the same conditions. (D) Analysis by immunoblotting with antibody to sucrose synthase in the same protein samples described above: cytosol (left), cell wall (centre) and membrane fraction (right). (E) Relative quantification of signals after immunoblotting with antibodies to sucrose synthase in the three fractions described above: cytosol (black bars), cell wall (white bars) and membrane fraction (grey bars). S, sucrose; G, glycerol. Measurements are reported as percentages of total signals. Fig. 7. View largeDownload slide Relative accumulation of sucrose synthase in the cytosol, membranes and cell walls of pollen tubes grown in BKS and BKG. (A) Electrophoretic analysis of proteins extracted from the cytosol of pollen tubes grown for 4, 5 and 6 h in the two different media. Molecular weight standards (kDa) are indicated on the left. (B) Profile of proteins extracted from the cell wall of pollen tubes grown for 4, 5 and 6 h in BKS and BKG. (C) Electrophoresis of proteins extracted from the membrane fraction of pollen tubes grown under the same conditions. (D) Analysis by immunoblotting with antibody to sucrose synthase in the same protein samples described above: cytosol (left), cell wall (centre) and membrane fraction (right). (E) Relative quantification of signals after immunoblotting with antibodies to sucrose synthase in the three fractions described above: cytosol (black bars), cell wall (white bars) and membrane fraction (grey bars). S, sucrose; G, glycerol. Measurements are reported as percentages of total signals. Because sucrose synthase is present in different cellular sites of pollen tubes, we focused our attention on its distribution in the presence (control) and absence (medium with glycerol) of sucrose. Analyses were performed by immunofluorescence microscopy 5 h after germination (Fig. 8). Analysis of control samples (Fig. 8A) showed the typical distribution of the enzyme, which was very abundant in the apical region and relatively less present in the distal regions. Quantitative analysis confirmed the data by highlighting a significant signal in the first 15–20 μm of pollen tubes and a loss of signal in the following areas (black line in the graph of Fig. 8D). On the contrary, localization of sucrose synthase in pollen tubes grown in glycerol-based medium changed radically. While observing a consistent signal in the apex, the most evident change was an intense signal in the distal areas (Fig. 8B, C). In these regions, sucrose synthase was distributed both in the cell edges and within the cytoplasm with a dot-like pattern, suggesting a possible association/accumulation of the enzyme within cell membranes. This peculiar distribution was particularly evident 40–50 μm from the apex. The different localization was more appreciable after graphical quantitation of the signal (grey line in the graph of Fig. 8D). Fig. 8. View largeDownload slide Distribution of sucrose synthase in pollen tubes grown for 5 h in BKS and BKG. (A) A typical pollen tube labelled with antibody against sucrose synthase. Note the accumulation of signal in the apical region and along the pollen tube edge. Scale bar = 10 µm. (B, C) Two typical pollen tubes grown for 5 h in BKG and labelled with antibody against sucrose synthase. Note the substantial accumulation of signal in distal and internal regions of the tubes. Scale bars = 10 µm. (D) Measurement of sucrose synthase signal along the axis of pollen tubes grown in BKS (black line) and BKG (grey line), starting from the tube apex. (E) Immunogold labelling of sucrose synthase in pollen tubes grown in BKS. In the subapical and distal regions, sucrose synthase is localized in association with the plasma membrane (arrows). Scale bar = 500 nm. (F) The enzyme signal is also found in intracellular membranes. Scale bar = 500 nm. (G) In pollen tubes grown in BKG, the actual amount of sucrose synthase in the subapical region is much smaller and only a few gold particles are observed in association with the plasma membrane. Scale bar = 500 nm. (H) Other subapical sections indicate a very weak signal, suggesting that the amount of sucrose synthase associated with the plasma membrane decreases considerably in BKG. Scale bar = 500 nm. Fig. 8. View largeDownload slide Distribution of sucrose synthase in pollen tubes grown for 5 h in BKS and BKG. (A) A typical pollen tube labelled with antibody against sucrose synthase. Note the accumulation of signal in the apical region and along the pollen tube edge. Scale bar = 10 µm. (B, C) Two typical pollen tubes grown for 5 h in BKG and labelled with antibody against sucrose synthase. Note the substantial accumulation of signal in distal and internal regions of the tubes. Scale bars = 10 µm. (D) Measurement of sucrose synthase signal along the axis of pollen tubes grown in BKS (black line) and BKG (grey line), starting from the tube apex. (E) Immunogold labelling of sucrose synthase in pollen tubes grown in BKS. In the subapical and distal regions, sucrose synthase is localized in association with the plasma membrane (arrows). Scale bar = 500 nm. (F) The enzyme signal is also found in intracellular membranes. Scale bar = 500 nm. (G) In pollen tubes grown in BKG, the actual amount of sucrose synthase in the subapical region is much smaller and only a few gold particles are observed in association with the plasma membrane. Scale bar = 500 nm. (H) Other subapical sections indicate a very weak signal, suggesting that the amount of sucrose synthase associated with the plasma membrane decreases considerably in BKG. Scale bar = 500 nm. Comparative information was also obtained after immunogold labelling with antibody to sucrose synthase. In pollen tubes grown in sucrose-based medium, sucrose synthase was found mainly in association with the plasma membrane (Fig. 8E, arrows) at the level of the subapical region. In addition to signal associated with the plasma membrane, sucrose synthase was also detected internally, presumably in the cytosol and in association with membranes (Fig. 8F). When pollen tubes were grown in glycerol-based medium, the amount of sucrose synthase in the subapex (as estimated by visual inspection) was very low in comparison with pollen tubes grown in sucrose-based medium (Fig. 8G). Only a few gold particles were observed in association with the plasma membrane. Most of the sections showed a very relative poor signal, indicating that the amount of sucrose synthase in association with the pollen tube plasma membrane decreased considerably (Fig. 8H). Accumulation and localization of callose synthase are affected by growth media Callose synthase is involved in the deposition of callose during pollen tube growth. We analysed the accumulation and localization of this enzyme in pollen tubes growing both in the standard condition (with sucrose) and under metabolic stress (with glycerol). We first proceeded with immunoblot analysis to evaluate the quantitative changes in the enzyme during different growth conditions (Fig. 9A). Blot analysis was carried out on protein extracts from membrane compartments of pollen tubes grown for 4, 5 and 6 h in either sucrose- or glycerol-based medium. In the standard condition (with sucrose) we found a higher accumulation of callose synthase at the fourth and fifth hour of growth but a subsequent decrease at the sixth hour of growth (Fig. 9A, blot in bottom panel). This observation was particularly evident and confirmed by measurement and graphing of several blots (Fig. 9B, black bars). In the stressed condition (glycerol medium) we found a different trend of accumulation of callose synthase. Unlike the control condition, we observed a smaller amount of enzyme at the fourth hour then a further decrease at the fifth hour but an increase in accumulation at the sixth hour of growth (Fig. 9A, blot in bottom panel and graph in Fig. 9B, grey bars). In general, the levels of callose synthase in pollen tubes grown in glycerol-based medium were lower than the corresponding controls in sucrose-based medium. Fig. 9. View largeDownload slide Analysis of the accumulation and distribution of callose synthase in pollen tubes grown in BKS and BKG for 4, 5 and 6 h. (A) Electrophoresis (top) and immunoblotting (bottom) analysis with antibody to callose synthase (CalS) in different growth conditions. S, sucrose; G, glycerol. (B) Measurement of the immunoblot signal. Black bars, pollen tubes in BKS; grey bars, pollen tubes in BKG. *P < 0.05; **P < 0.01. (C) Analysis by immunofluorescence microscopy of callose synthase in pollen tubes grown in BKS after 5 h of germination. The enzyme is distributed predominantly in the apical region (asterisk) and, to a lesser extent, in more distal parts of the pollen tubes (arrow). Insert: single focal section approximately at the median level of a pollen tube apex. (D) In pollen tubes grown in BKG, the enzyme accumulates substantially in the cytoplasm of distal regions concomitantly to reduction of signal in the tube apex. Insert: distal portion of a pollen tube with clear intracytoplasmic accumulation of the enzyme. Scale bars = 10 µm. Fig. 9. View largeDownload slide Analysis of the accumulation and distribution of callose synthase in pollen tubes grown in BKS and BKG for 4, 5 and 6 h. (A) Electrophoresis (top) and immunoblotting (bottom) analysis with antibody to callose synthase (CalS) in different growth conditions. S, sucrose; G, glycerol. (B) Measurement of the immunoblot signal. Black bars, pollen tubes in BKS; grey bars, pollen tubes in BKG. *P < 0.05; **P < 0.01. (C) Analysis by immunofluorescence microscopy of callose synthase in pollen tubes grown in BKS after 5 h of germination. The enzyme is distributed predominantly in the apical region (asterisk) and, to a lesser extent, in more distal parts of the pollen tubes (arrow). Insert: single focal section approximately at the median level of a pollen tube apex. (D) In pollen tubes grown in BKG, the enzyme accumulates substantially in the cytoplasm of distal regions concomitantly to reduction of signal in the tube apex. Insert: distal portion of a pollen tube with clear intracytoplasmic accumulation of the enzyme. Scale bars = 10 µm. In parallel with the analysis of accumulation, we also tested the localization of callose synthase in pollen tubes grown in different germination media by immunofluorescence microscopy. The analysis was performed only at the fifth hour after germination. In the control, we found that callose synthase was prevalently distributed in the apical region and, to a lesser extent, in more distal parts of the pollen tubes (arrow in Fig. 9C). The inset in Fig 9C (representing a single focal section at the median level of a pollen tube apex) highlights the specific labelling at the level of the apical plasma membrane; this is most likely the cell domain where the enzyme is secreted and inserted. Callose synthase is poorly present in the cell cytoplasm. In pollen tubes grown in the presence of glycerol (Fig. 9D), we identified two substantial changes with respect to standard growing conditions. First, we found a considerable accumulation of the enzyme in the cytoplasm of distal regions; second, we noted a significant reduction in signal at the tube apex. The inset in Fig. 9D shows a distal portion of a pollen tube, in which intracytoplasmic accumulations of callose synthase are evident. After examining the distribution of callose synthase by immunofluorescence microscopy, we deepened the analysis by using immunoelectron microscopy in the cell areas indicated by the sketches in Fig. 10 (the top sketch refers to BKS analyses, the bottom sketch to BKG analyses). In the apex/subapex region (Fig. 10A), gold particles were found mostly in association with the plasma membrane and in the cell wall. The signal was also evident in association with the vesicular material normally present at the apex of pollen tubes (arrows), suggesting events of exocytosis. In distal regions (Fig. 10B), gold particles were associated with the plasma membrane and sporadically with the inner layer of the cell wall. In the underlying cytoplasm of distal regions, association between callose synthase and microtubular cytoskeleton could be observed. More specifically, gold particles were found in association with longitudinal cortical microtubules (Fig. 10C, arrow), thus suggesting again that microtubules participate in the localization of callose synthase in distal regions. Fig. 10. View largeDownload slide Analysis of the distribution of callose synthase by immunoelectron microscopy. (A) In the subapical region of pollen tubes grown in sucrose-based medium, callose synthase is strongly detected in association with the plasma membrane and cell wall. The enzyme is also found in association with vesicles in the underlying cytoplasm (arrows). Scale bar = 500 nm. (B) In the most distal regions, the enzyme is still detected in association with the plasma membrane and sporadically with the inner layer of the cell wall. Scale bar = 500 nm. (C) Association between callose synthase and longitudinal cortical microtubules as observed in the underlying cytoplasm (arrow). Scale bar = 150 nm. (D) In pollen tubes grown in glycerol-based medium for 5 h, the amount of enzyme decreases considerably at the apex/subapex level. Scale bar = 500 nm. (E) In the shank region, callose synthase is still predominantly associated with the plasma membrane and the initial layer of the cell wall. Scale bar = 500 nm. (F) Detail of the cytoplasmic region with strong signal of callose synthase. Scale bar = 500 nm. Fig. 10. View largeDownload slide Analysis of the distribution of callose synthase by immunoelectron microscopy. (A) In the subapical region of pollen tubes grown in sucrose-based medium, callose synthase is strongly detected in association with the plasma membrane and cell wall. The enzyme is also found in association with vesicles in the underlying cytoplasm (arrows). Scale bar = 500 nm. (B) In the most distal regions, the enzyme is still detected in association with the plasma membrane and sporadically with the inner layer of the cell wall. Scale bar = 500 nm. (C) Association between callose synthase and longitudinal cortical microtubules as observed in the underlying cytoplasm (arrow). Scale bar = 150 nm. (D) In pollen tubes grown in glycerol-based medium for 5 h, the amount of enzyme decreases considerably at the apex/subapex level. Scale bar = 500 nm. (E) In the shank region, callose synthase is still predominantly associated with the plasma membrane and the initial layer of the cell wall. Scale bar = 500 nm. (F) Detail of the cytoplasmic region with strong signal of callose synthase. Scale bar = 500 nm. The distribution of callose synthase was partially different when we analysed pollen tubes grown for 5 h in BKG. First, images at apex/subapex level showed that the amount of enzyme decreased significantly (Fig. 10D). Although the density of gold particles was significantly lower than in the controls, callose synthase was still predominantly associated with the plasma membrane. Another significant difference concerned the distribution of callose synthase in distal regions (Fig. 10E). Here, callose synthase was still found mainly in the plasma membrane and the inner layer of the cell wall. However, a consistent signal was also detected in the cytoplasm of pollen tubes (Fig. 10F), sometimes in association with membrane-like structures, suggesting internalization or lack of secretion of the enzyme. Distribution of callose changed in pollen tubes grown in different media The distribution of callose in the subapical region of pollen tubes grown in glycerol-based medium differed in comparison with pollen tubes in standard medium. In pollen tubes grown in sucrose-based medium, the distribution of callose was not significantly different during the transition from 4 to 6 h of germination. In all three cases (4, 5 and 6 h), the relative amount of callose was almost undetectable at the apex, then increased gradually and stabilized at ~15 μm from the apex (Fig. 11A). Distribution of callose was rather different when pollen tubes were grown in glycerol-based medium. After 4 and 6 h of growth, callose was not detected at the very tip region but the aniline blue signal increased rapidly and stabilized around 10–15 µm from the tip. The pattern was different after 5 h of growth because the callose signal was again undetectable at the very tip but increased and stabilized around 20–25 µm from the tip (Fig. 11B). This was not immediately visible after staining with aniline blue (Fig. 11C–E) but became apparent after measurement of callose in different pollen tubes. Consequently, the distribution of callose in 5-h-grown pollen tubes was markedly different from that in both 4- and 6-h-grown pollen tubes. Fig. 11. View largeDownload slide Distribution of callose in the subapical region of pollen tubes grown in medium with sucrose or glycerol. In all cases, fluorescence was measured starting from the apex down to 30 μm. (A) In medium with sucrose, distribution of callose did not differ during the transition from 4 to 6 h of germination. (B) Distribution of callose differed when pollen tubes were grown in medium with glycerol, especially after 5 h of growth, because the callose signal stabilized only around 20–25 μm from the apex. (C, D, E) Visualization of callose in pollen tubes grown in medium with glycerol, respectively after 4, 5 and 6 h of germination. Scale bars = 10 μm. Fig. 11. View largeDownload slide Distribution of callose in the subapical region of pollen tubes grown in medium with sucrose or glycerol. In all cases, fluorescence was measured starting from the apex down to 30 μm. (A) In medium with sucrose, distribution of callose did not differ during the transition from 4 to 6 h of germination. (B) Distribution of callose differed when pollen tubes were grown in medium with glycerol, especially after 5 h of growth, because the callose signal stabilized only around 20–25 μm from the apex. (C, D, E) Visualization of callose in pollen tubes grown in medium with glycerol, respectively after 4, 5 and 6 h of germination. Scale bars = 10 μm. The composition of growth media affects both generative cell movement and callose plug deposition An additional feature we evaluated was the relationship between generative cell movement and composition of the growth medium. Using DAPI staining, we analysed the movement of generative cells during growth of pollen tubes in both sucrose- and glycerol-based medium, starting from the third hour of germination. The data are reported as the ratio between the distance covered by generative cells from the grain and the length of corresponding pollen tubes. Data were analysed by two different statistical tests. First, we used ANOVA to prove that some data sets were statistically different. Subsequently, we applied a t-test comparative analysis between pairs of data sets obtained in sucrose- and glycerol-based media at the same time of growth. We found that data sets were statistically different only during the fifth hour of growth (P < 0.05), as indicated by the asterisk in Fig. 12A. At this specific growth point, the generative cell covered a smaller distance than the controls in sucrose-based medium; in addition, the larger standard deviation suggested a greater range of movement inhomogeneity. Fig. 12. View largeDownload slide Effects of germination medium on the translocation of the generative cell and on the deposition of callose plugs. (A) Ratio between the movement of the generative cell from the grain and the corresponding pollen tube length as measured from 3 to 7 h after germination in growth medium containing either sucrose (S, black bars) or glycerol (G, grey bars). Standard deviation is indicated at the top of the bars. *P < 0.05. (B) Ratio between positions of the first and second callose plugs and the corresponding pollen tube length as determined after 7 h of germination. The first callose plug is the one closest to the pollen grain. Bars indicate standard deviation. (C) A pollen tube grown in BKS for 7 h, in which two callose plugs (arrows) can be observed. Scale bar = 50 μm. (D) A pollen tube grown for 7 h in BKG containing only the first callose plug (arrow). Scale bar = 50 μm. Fig. 12. View largeDownload slide Effects of germination medium on the translocation of the generative cell and on the deposition of callose plugs. (A) Ratio between the movement of the generative cell from the grain and the corresponding pollen tube length as measured from 3 to 7 h after germination in growth medium containing either sucrose (S, black bars) or glycerol (G, grey bars). Standard deviation is indicated at the top of the bars. *P < 0.05. (B) Ratio between positions of the first and second callose plugs and the corresponding pollen tube length as determined after 7 h of germination. The first callose plug is the one closest to the pollen grain. Bars indicate standard deviation. (C) A pollen tube grown in BKS for 7 h, in which two callose plugs (arrows) can be observed. Scale bar = 50 μm. (D) A pollen tube grown for 7 h in BKG containing only the first callose plug (arrow). Scale bar = 50 μm. In addition to the measurement of generative cell movement, we also determined the frequency of callose plug deposition after 7 h of growth in both sucrose- and glycerol-based medium. Again, the data were reported as the ratio between the distance of callose plugs (as measured from the grain) and the length of corresponding pollen tubes. In the controls, the first callose plug (the one closest to the grain) was formed with an average ratio of <0.4 while in pollen tubes grown in glycerol-based media the average ratio was around 0.2. On the contrary, the second callose plug (the one closest to the tube apex) was formed in both cases with the same ratio, around 0.6 (Fig. 12B). Apparently, the presence of glycerol-based medium affected the speed of pollen tube elongation during the fourth and fifth hours of growth and, in turn, this determined the early formation of the first callose plug. After 7 h of germination, when the growth rate of pollen tubes was very similar in both experimental conditions, the second callose plug was also formed in the same position relative to pollen tube length. A characteristic and interesting aspect concerned the percentage of deposition of the second callose plug, which, in pollen samples grown in BKG, did not form consistently but at lower percentages (about 32.4 % compared with 63.6 % in pollen tubes grown in BKS). In Fig. 12 panels C and D illustrate, respectively, a representative pollen tube grown for 7 h in BKS (with two callose plugs) and a pollen tube grown in BKG (only the first callose plug is visible). DISCUSSION In this article we have analysed the effects of carbon availability on the development of pollen tubes by replacing sucrose with glycerol in the growth medium. Glycerol, unlike sucrose, cannot be efficiently converted into energy. Therefore, when internal stores are depleted, pollen tubes experience an energy deficit. In yeast, glycerol can be used as a carbon source under specific conditions and the catabolic pathway involves phosphorylation by a glycerol kinase and oxidation by a mitochondrial glycerol phosphate-ubiquinone oxidoreductase (Nevoigt and Stahl, 1997). Plant cells can also likely use glycerol as a carbon source and glycerol-3-phosphate derived from phosphorylation of glycerol is introduced into the glycolic pathway by a cytosolic and/or plastidial NAD-linked glycerol phosphate dehydrogenase. However, high concentrations of glycerol-3-phosphate prevent the flowing back of carbon from triose phosphates to glucose-6-phosphate. This is confirmed by the evidence that glycerol-3-phosphate is a competitive inhibitor of glucose-6-phosphate isomerases and, after sucrose starvation, addition of glycerol to the cell culture does not trigger a rapid accumulation of glucose-6-phosphate in the cytoplasmic compartment and prevents the reloading of intracellular pools of carbohydrates (starch and sucrose) (Aubert et al., 1994). This means that glycerol might enter the metabolism at a later stage and, most likely, it cannot contribute efficiently to the production of UDP-glucose. As a first analysis, we found that the presence of glycerol substantially changes the length and growth speed of pollen tubes; this is not unexpected because tube growth requires high levels of energy molecules. We monitored the effects of energy deficiency during 6–7 h of growth, during which internal carbohydrate stores would have to be depleted. Surprisingly, the most discernible effect occurred at the fifth hour of growth as a consistent reduction in the growth rate; after that, the growth rate was restored. This specific phase likely corresponds to the depletion of internal energy deposits (autotrophic phase) and the transition to heterotrophic phase, i.e. the consumption of carbohydrates taken up from the extracellular environment. As partial confirmation, low levels of sucrose or starch do not necessarily affect pollen tube growth in tomato, and pollen tubes can apparently use other energy molecules (Garcia et al., 2012). Pollen tubes might import sugars through two distinct mechanisms. The first involves the cleavage of sucrose by cell wall-associated invertase and the transport of monosaccharides. Glucose and fructose are then readily available in the cytoplasm of pollen tubes (Goetz et al., 2016). This metabolic pathway is probably active when the availability of sucrose is high (i.e. when there is no shortage of carbon). The second import mechanism requires the presence of sucrose transporters, which carry sucrose directly within the cytoplasm (Lemoine et al., 1999; Stadler et al., 1999). According to this alternative metabolic pathway, sucrose can then be used to produce fructose and UDP-glucose by the activity of cytoplasmic sucrose synthase. This second pathway is generally preferred when the availability of carbon is reduced because it saves a greater amount of energy. This could reasonably lead to reduction in the concentration of glucose and fructose in pollen tube cytoplasm because a small amount of fructose would be immediately directed to glycolysis. The ability of pollen tubes to modify their metabolism under metabolic stress is not surprising because similar effects have already been described when the mitochondrial electron transport chain is blocked by specific inhibitors. In such conditions, pollen tubes exhibit a reduction in growth rate and then switch to fermentative metabolism. Soon after, pollen tubes resume their normal growth rate, suggesting metabolic bypass and plasticity in these cells (Rounds et al., 2010). The metabolic adaptation of pollen is also demonstrated by the rapid changes in the content of metabolites that are observed during the transition from pollen to pollen tube, and even more when the mitochondrial electron transport chain is inhibited (Obermeyer et al., 2013). The growth of pollen tubes through the style requires considerable amounts of energy. Pollen tubes of lily are estimated to synthesize ~4.4 × 10−14 moles of ATP grain−1 s−1 during the normal growth process (Rounds et al., 2011b). In this study we found a different concentration of ATP between the two growth media. In glycerol-based medium, we found a significant reduction in ATP concentration throughout the analysed period compared with the concentration of ATP in the sucrose-based medium. This could be explained by a different rate of ATP production during metabolic stress, and it is confirmed by the observation that other stresses (such as cold treatment) cause reduction of ATP synthesis in pear pollen tubes (Gao et al., 2014). However, the reduction in ATP synthesis cannot be the only reason for the observed phenomena; in fact, the reduction in ATP concentration is constant throughout the growth period, while most of the effects are observable in a precise time span. Apparently, pollen tubes succeed in balancing the reduced production of ATP to adequately support the growth process. It is reasonably expected that switching to a different metabolism can affect many collateral processes that directly or indirectly depend on the way energy is produced. In our case, we found significant changes in cell wall structure, such as at the level of pectins. An incorrect deposition of pectins is a characteristic trait after metabolic stress and it may result from either a reduced fusion rate of secretory vesicles at the apex or from a lower content of pectins within vesicles. The synthesis of pectins occurs in the Golgi apparatus and most likely requires the availability of UDP-glucose. A complex metabolic network (Caffall and Mohnen, 2009) is critical for the interconversion of UDP-glucose into energy-rich monosaccharides required for pectin synthesis, such as the conversion of UDP-glucose to UDP-galactose (Huang et al., 2016) and the conversion of UDP-glucose to UDP-glucuronic acid (Klinghammer and Tenhaken, 2007). We suggest that the Golgi-associated sucrose synthase we previously found (Persia et al., 2008) may be important in fuelling the Golgi-bound UDP-glucose transporters (Munoz et al., 1996). Therefore, an altered activity/distribution of sucrose synthase can affect pectin synthesis by reducing the levels of UDP-glucose within Golgi membranes. The suggested modification of pectin levels is also accompanied by an altered distribution of pectins. Labelling with both PI and, more specifically, JIM7 antibody has shown that pollen tubes grown in BKG have a different, more homogeneous distribution of methyl-esterified pectins than the controls. In parallel and to support this, the profile of acidic pectins also changes. Although the latter are always distributed with regular periodicity, they accumulate at the apex, their levels along the pollen tube are more homogeneous and the ring pattern is much less accentuated. Data on pectins indicate a lower rate of secretion of methyl-esterified pectins, which results in higher levels of acidic pectin at the apex and more homogeneity along tubes. This new distribution pattern mirrors the new type of pollen tube growth, which is more continuous and not oscillatory. Such growth usually occurs when pollen tubes have a more homogeneous cell wall composition (Kroeger et al., 2011). The differential type of growth (continuous versus oscillatory) was observed both when the growth rate was similar to controls and when it decreased drastically. Therefore, continuous growth is not directly related to different growth rates, but is characteristic of pollen tubes that grow with low carbon availability. Because the growth of pollen tubes depends on the relationship between turgor pressure and cell wall stiffness, one of the two factors is responsible for the different growth type. If turgor pressure does not change, the new growth medium likely affects cell wall composition. The literature documents changes in the expression of genes coding for cell wall enzymes of pollen tubes growing in different growth media (different sucrose concentrations), indicating that the composition of the germination medium affects the cell wall structure (da Costa et al., 2013). Even a different osmolarity of the growth medium can have effects on the wall composition, as observed when pollen tubes are grown in low-osmolarity media (Biagini et al., 2014). The lack of standard (oscillatory) growth has already been reported during other stressful conditions, such as in the case of tobacco pollen tubes subjected to heat stress (Parrotta et al., 2015) or when pollen tubes grow in hypotonic media (osmotic stress) (Biagini et al., 2014). However, there are some exceptions. For example, when the electron transport chain is inhibited in lily pollen tubes, the growth of pollen tubes stops temporarily before resuming (as also in our study), but it continues to oscillate (Rounds et al., 2010). Data on lily pollen tubes suggest that oscillatory growth is not directly related to the type of metabolism. In our case, we can assume that the loss of oscillatory growth depends on the amount of energy that pollen tubes take up externally and on the direction of energy flow. Because glycerol may feed glycolysis but not UDP-glucose synthesis, sucrose depletion modifies the way the apical cell wall is assembled. The current data confirm the results of Persia et al. (2008) indicating that pollen tubes grown in glycerol-based medium for 3 h show both a different distribution of sucrose synthase and a deficit of growth, and suggest again that the external availability of sucrose can affect the distribution (and perhaps the activity) of sucrose synthase. In fact, we observed that a scarcity of sucrose causes a lower accumulation of sucrose synthase in the membrane protein fraction as well as in the cell wall protein fraction. In support of this observation, a differential distribution of sucrose synthase in cotton has been proposed in relation to the different use of sucrose. Plausibly, when plant cells need to elongate, sucrose is necessary to fuel the synthesis of cellulose and sucrose synthase is therefore mainly associated with the plasma membrane; when, instead, plant cells suffer sucrose shortage, they need to save energy and sucrose synthase is therefore in the cytosolic form (Haigler et al., 2001). In our case, when pollen tubes are grown in glycerol-based medium, sucrose synthase is likely to accumulate in the cytosol and only part of it moves to the apical plasma membrane or cell wall, perhaps accumulating in intracellular membranes. This redistribution could affect the ability to adequately support the synthesis of cellulose and callose through production of UDP-glucose, thereby affecting cell wall assembly. Although the association between sucrose synthase and cellulose synthase or callose synthase has not yet been established precisely, some work suggests this possibility (Fujii et al., 2010; Brill et al., 2011). Accordingly, the slight but significant changes in the deposition of callose, as observed after 5 h of growth in glycerol-based medium, are not unexpected because the less energetic contribution of glycerol may affect the synthesis of callose. Changes in callose deposition have a counterpart in the altered distribution of callose synthase. During the fifth hour of growth the content of callose synthase in the membrane fraction decreased significantly. In addition, we observed a significantly lower accumulation of callose synthase in the apical region that resulted in intracellular accumulation. In contrast, during the fourth and sixth hours of growth, the content of callose synthase in the membrane fraction was comparable to controls. At the cytological level, the distribution of the enzymes was also comparable to that described in the literature (Cai et al., 2011). These observations indicate that the transition between the use of internal stores and the consumption of glycerol also affects the distribution of callose synthase. Since the insertion and removal of callose synthase from the plasma membrane is closely related to the vesicular flow (Brownfield et al., 2008), changes in exocytosis/endocytosis may alter the enzyme levels in the apical plasma membrane. It should be noted that similar effects have also been obtained after treatment of tobacco pollen tubes with brefeldin A, a vesicle trafficking inhibitor (Cai et al., 2011). Reduced levels of callose synthase in the apical plasma membrane and its accumulation in the cytoplasm are also supported by immunoelectron microscopy analysis. Therefore, the combination of lower levels of UDP-glucose with incorrect distribution of callose synthase could reasonably determine an altered deposition of callose at the fifth hour of growth. Since callose levels and distribution of callose synthase are re-established during the sixth hour of growth, pollen tubes must necessarily have put in place mechanisms of recovery of callose synthesis in the presence of reduced levels of UDP-glucose. Other significant effects were observed on proton levels (i.e. intracellular pH values) and ROS. Since the correct distribution of both H+ and ROS contributes to regulation of the growth of pollen tubes, their anomalous redistribution can be linked to the new type of growth. In pollen tubes, H+-ATPases are the primary pump responsible for creating proton gradients that are used for myriad active transport processes (such as sucrose transport). These enzymes are also likely important for maintaining a correct pH at the pollen tube tip (Feijò et al., 1999). The pH homeostasis probably depends on the influx of H+ at the apex and efflux of H+ as achieved by membrane ATPases (the so-called alkaline band) (Feijò et al., 1999). We found that tobacco pollen tubes grown in sucrose-based media are characterized by an apical acid region, followed by a decrease in proton concentration. Conversely, when pollen tubes are grown in glycerol-based medium, proton levels are more homogeneous. Studies on lily pollen tubes during normal oscillatory growth have revealed that both the acidic tip and the alkaline band oscillate with the same growth period (Lovy-Wheeler et al., 2006). Because in our experiment we observed disappearance of oscillatory growth, this reasonably relates to disappearance of the proton gradient. Reactive oxygen species are an inevitable consequence of aerobic metabolism but are also produced in a controlled manner and used for a variety of functions, including defence against pathogens and cell signalling (Mittler et al., 2011). In our work, we determined that pollen tubes grown in standard media are characterized by a relatively higher concentration of ROS in the apical region. However, when pollen tubes are grown in glycerol-based medium, ROS show a uniform distribution along the growth axis. In pollen tubes, the presence of ROS is correlated to the calcium gradient since ROS most likely act as positive feedback for the accumulation of calcium ions, probably by modulating the opening of Ca2+ channels at the tube apex (Potocky et al., 2012; Kaya et al., 2014; Lassig et al., 2014; Aloisi et al., 2017). It has also been reported that the synthesis of ROS through NAPDH oxidase is activated by Ca2+, thus forming a positive feedback circuit that reinforces the tip polarity (Potocky et al., 2007). Although a link between the different processes is not immediately clear, it is possible to trace a metabolic pattern from the entrance of energy molecules (sucrose or glycerol) to the effects on pollen tube growth (Fig. 13). The thickness of the arrows in Fig. 13 indicates the hypothetical intensity of each process. Fig. 13. View largeDownload slide Schematic illustration of the main metabolic pathways occurring in pollen tubes grown in either sucrose or glycerol. (A) When the growth medium contains sucrose, it feeds glycolysis (and thus respiration) and UDP-glucose synthesis. This provides a normal growth process. (B) When the growth medium contains glycerol, synthesis of UDP-glucose is likely to be reduced; ATP levels are also lower, and this affects the several processes analysed, such as the homeostasis of pH and ROS, and production/secretion of pectins. The growth process is thus altered. The thickness of the arrows indicates the hypothetical intensity of each individual process. CalS, callose synthase; IMP, importer; INV, invertase; SUS, sucrose synthase. Fig. 13. View largeDownload slide Schematic illustration of the main metabolic pathways occurring in pollen tubes grown in either sucrose or glycerol. (A) When the growth medium contains sucrose, it feeds glycolysis (and thus respiration) and UDP-glucose synthesis. This provides a normal growth process. (B) When the growth medium contains glycerol, synthesis of UDP-glucose is likely to be reduced; ATP levels are also lower, and this affects the several processes analysed, such as the homeostasis of pH and ROS, and production/secretion of pectins. The growth process is thus altered. The thickness of the arrows indicates the hypothetical intensity of each individual process. CalS, callose synthase; IMP, importer; INV, invertase; SUS, sucrose synthase. A striking effect after growing pollen tubes in BKG was the change in the deposition of callose plugs. The first callose plug is deposited in advance of the control, while the second plug is deposited as in the controls. The low growth rate observed after glycerol treatment has therefore substantial effects on the deposition of callose plugs, as confirmed by the fact that glycerol treatment does not allow an adequate synthesis of UDP-glucose, the substrate for callose synthesis. The reduction in growth rate is also accompanied by a significant reduction in the movement of generative cells, so much so that the two events could be linked. If the first callose plug is properly deposited during growth in BKG, this could lead to the isolation of generative cells (if they move at slower speeds). Therefore, incorrect deposition of the first callose plug could be interpreted as an attempt not to isolate the generative cell. When the growth of pollen tubes resumes at a speed like that occurring in controls, the second callose plug is deposited in a way comparable to that in controls. This suggests the presence of a link between the growth of pollen tubes, the deposition of callose plugs and the movement of generative cells. Few works in the literature suggest such a relationship; in particular, microtubules have been indicated as connectors between the two processes (Laitiainen et al., 2002). In the present work we did not analyse the organization of microtubules following metabolic stress. We obtained some preliminary indications in actin filaments, which, in pollen tubes grown in BKG, seemed to be organized exactly as in the BKS controls (data not shown). Monitoring the cytoskeleton of pollen tubes in glycerol-based medium could be an interesting research topic to focus on later. ACKNOWLEDGEMENTS We are grateful to Mr Massimo Guarnieri (Department of Life Sciences, University of Siena) for technical assistance during HPLC analyses; we also thank the employees of the Botanical Garden of the University of Siena for kindly supporting us and for growing tobacco plants. This work was supported by PRIN 2015 ISIDE (Investigating Self Incompatibility Determinants in fruit trees) (http://prin.miur.it/) to S.D.D. and G.C. Conflict of Interest: All authors of the manuscript declare that they have no potential sources or conflict/financial interest. The research involves neither human participants nor animals. LITERATURE CITED Aloisi I , Cai G , Tumiatti V , Minarini A , Del Duca SD . 2015 . Natural polyamines and synthetic analogues modify the growth and the morphology of Pyrus communis pollen tubes affecting ROS levels and causing cell death . Plant Science 239 : 92 – 105 . Google Scholar CrossRef Search ADS PubMed Aloisi I , Cai G , Faleri C , Navazio L , Serafini-Fracassini D , Del Duca S . 2017 . Spermine regulates pollen tube growth by modulating Ca2+-dependent actin organization and cell wall structure . Frontiers in Plant Science 8 : 1701 . Google Scholar CrossRef Search ADS PubMed Aubert S , Gout E , Bligny R , Douce R . 1994 . Multiple effects of glycerol on plant cell metabolism. Phosphorus-31 nuclear magnetic resonance studies . Journal of Biological Chemistry 269 : 21420 – 21427 . Google Scholar PubMed Biagini G , Faleri C , Cresti M , Cai G . 2014 . Sucrose concentration in the growth medium affects the cell wall composition of tobacco pollen tubes . Plant Reproduction 27 : 129 – 144 . Google Scholar CrossRef Search ADS PubMed Brewbaker JL , Kwack BH . 1963 . The essential role of calcium ion in pollen germination and pollen tube growth . American Journal of Botany 50 : 859 – 865 . Google Scholar CrossRef Search ADS Brill E , van Thournout M , White RG , et al. 2011 . A novel isoform of sucrose synthase is targeted to the cell wall during secondary cell wall synthesis in cotton fiber . Plant Physiology 157 : 40 – 54 . Google Scholar CrossRef Search ADS PubMed Brownfield L , Wilson S , Newbigin E , Bacic A , Read S . 2008 . Molecular control of the glucan synthase-like protein NaGSL1 and callose synthesis during growth of Nicotiana alata pollen tubes . Biochemical Journal 414 : 43 – 52 . Google Scholar CrossRef Search ADS PubMed Caffall KH , Mohnen D . 2009 . The structure, function, and biosynthesis of plant cell wall pectic polysaccharides . Carbohydrate Research 344 : 1879 – 1900 . Google Scholar CrossRef Search ADS PubMed Cai G , Romagnoli S , Moscatelli A , et al. 2000 . Identification and characterization of a novel microtubule-based motor associated with membranous organelles in tobacco pollen tubes . Plant Cell 12 : 1719 – 1736 . Google Scholar CrossRef Search ADS PubMed Cai G , Faleri C , Del Casino C , Emons AMC , Cresti M . 2011 . Distribution of callose synthase, cellulose synthase and sucrose synthase in tobacco pollen tube is controlled in dissimilar ways by actin filaments and microtubules . Plant Physiology 155 : 1169 – 1190 . Google Scholar CrossRef Search ADS PubMed Cai G , Parrotta L , Cresti M . 2015 . Organelle trafficking, the cytoskeleton, and pollen tube growth . Journal of Integrative Plant Biology 57 : 63 – 78 . Google Scholar CrossRef Search ADS PubMed Certal AC , Almeida RB , Carvalho LM , et al. 2008 . Exclusion of a proton ATPase from the apical membrane is associated with cell polarity and tip growth in Nicotiana tabacum pollen tubes . Plant Cell 20 : 614 – 634 . Google Scholar CrossRef Search ADS PubMed Chen CY , Wong EI , Vidali L , et al. 2002 . The regulation of actin organization by actin-depolymerizing factor in elongating pollen tubes . Plant Cell 14 : 2175 – 2190 . Google Scholar CrossRef Search ADS PubMed da Costa ML , Pereira LG , Coimbra S . 2013 . Growth media induces variation in cell wall associated gene expression in Arabidopsis thaliana pollen tube . Plants 2 : 429 – 440 . Google Scholar CrossRef Search ADS PubMed Feijò JA , Sainhas J , Hackett GR , Kunkel JG , Hepler PK . 1999 . Growing pollen tubes possess a constitutive alkaline band in the clear zone and a growth-dependent acidic tip . Journal of Cell Biology 144 : 483 – 496 . Google Scholar CrossRef Search ADS PubMed Fricker MD , White NS , Obermeyer G . 1997 . pH gradients are not associated with tip growth in pollen tubes of Lilium longiflorum . Journal of Cell Science 110 : 1729 – 1740 . Google Scholar PubMed Fujii S , Hayashi T , Mizuno K . 2010 . Sucrose synthase is an integral component of the cellulose synthesis machinery . Plant and Cell Physiology 51 : 294 – 301 . Google Scholar CrossRef Search ADS PubMed Gao YB , Wang CL , Wu JY , et al. 2014 . Low temperature inhibits pollen tube growth by disruption of both tip-localized reactive oxygen species and endocytosis in Pyrus bretschneideri Rehd . Plant Physiology and Biochemistry 74 : 255 – 262 . Google Scholar CrossRef Search ADS PubMed Garcia CC , Guarnieri M , Pacini E . 2012 . Tomato pollen tube development and carbohydrate fluctuations in the autotrophic phase of growth . Acta Physiologiae Plantarum 34 : 2341 – 2347 . Google Scholar CrossRef Search ADS Goetz M , Guivarc’h A , Hirsche J , et al. 2016 . Metabolic control of tobacco pollination by sugars and invertases . Plant Physiology 173 : 984 – 997 . Google Scholar CrossRef Search ADS PubMed Gu F , Nielsen E . 2013 . Targeting and regulation of cell wall synthesis during tip growth in plants . Journal of Integrative Plant Biology 55 : 835 – 846 . Google Scholar CrossRef Search ADS PubMed Haigler CH , Ivanova-Datcheva M , Hogan PS , et al. 2001 . Carbon partitioning to cellulose synthesis . Plant Molecular Biology 47 : 29 – 51 . Google Scholar CrossRef Search ADS PubMed Heinlein M , Starlinger P . 1989 . Tissue- and cell-specific expression of the two sucrose synthase isoenzymes in developing maize kernels . Molecular and General Genetics 215 : 441 – 446 . Google Scholar CrossRef Search ADS Hirose T , Zhang Z , Miyao A , Hirochika H , Ohsugi R , Terao T . 2010 . Disruption of a gene for rice sucrose transporter, OsSUT1, impairs pollen function but pollen maturation is unaffected . Journal of Experimental Botany 61 : 3639 – 3646 . Google Scholar CrossRef Search ADS PubMed Huang JH , Kortstee A , Dees DCT , Trindade LM , Schols HA , Gruppen H . 2016 . Alteration of cell wall polysaccharides through transgenic expression of UDP-Glc 4-epimerase-encoding genes in potato tubers . Carbohydrate Polymers 146 : 337 – 344 . Google Scholar CrossRef Search ADS PubMed Kaya H , Nakajima R , Iwano M , et al. 2014 . Ca2+-activated reactive oxygen species production by Arabidopsis RbohH and RbohJ is essential for proper pollen tube tip growth . Plant Cell 26 : 1069 – 1080 . Google Scholar CrossRef Search ADS PubMed Klinghammer M , Tenhaken R . 2007 . Genome-wide analysis of the UDP-glucose dehydrogenase gene family in Arabidopsis, a key enzyme for matrix polysaccharides in cell walls . Journal of Experimental Botany 58 : 3609 – 3621 . Google Scholar CrossRef Search ADS PubMed Kroeger JH , Zerzour R , Geitmann A . 2011 . Regulator or driving force? The role of turgor pressure in oscillatory plant cell growth . PLoS One 6 : e18549 . Google Scholar CrossRef Search ADS PubMed Laemmli UK . 1970 . Cleavage of structural proteins during the assembly of the head of bacteriophage T4 . Nature 227 : 680 – 685 . Google Scholar CrossRef Search ADS PubMed Laitiainen E , Nieminen KM , Vihinen H , Raudaskoski M . 2002 . Movement of generative cell and vegetative nucleus in tobacco pollen tubes is dependent on microtubule cytoskeleton but independent of the synthesis of callose plugs . Sexual Plant Reproduction 15 : 195 – 204 . Google Scholar CrossRef Search ADS Lang V , Pertl-Obermeyer H , Safiarian MJ , Obermeyer G . 2014 . Pump up the volume – a central role for the plasma membrane H(+) pump in pollen germination and tube growth . Protoplasma 251 : 477 – 488 . Google Scholar CrossRef Search ADS PubMed Lassig R , Gutermuth T , Bey TD , Konrad KR , Romeis T . 2014 . Pollen tube NAD(P)H oxidases act as a speed control to dampen growth rate oscillations during polarized cell growth . Plant Journal 78 : 94 – 106 . Google Scholar CrossRef Search ADS PubMed Lemoine R , Burkle L , Barker L , Sakr S , et al. 1999 . Identification of a pollen-specific sucrose transporter-like protein NtSUT3 from tobacco . FEBS Letters 454 : 325 – 330 . Google Scholar CrossRef Search ADS PubMed Li YQ , Faleri C , Geitmann A , Zhang HQ , Cresti M . 1995 . Immunogold localization of arabinogalactan proteins, unesterified and esterified pectins in pollen grains and pollen tubes of Nicotiana tabacum L . Protoplasma 189 : 26 – 36 . Google Scholar CrossRef Search ADS Li YQ , Mareck A , Faleri C , Moscatelli A , Liu Q , Cresti M . 2002 . Detection and localization of pectin methylesterase isoforms in pollen tubes of Nicotiana tabacum L . Planta 214 : 734 – 740 . Google Scholar CrossRef Search ADS PubMed Liu H , Jiang Y , Luo Y , Jiang W . 2006 . A simple and rapid determination of ATP, ADP and AMP concentrations in pericarp tissue of litchi fruit by high performance liquid chromatography . Food Technology and Biotechnology 44 : 531 – 534 . Lovy-Wheeler A , Kunkel JG , Allwood EG , Hussey PJ , Hepler PK . 2006 . Oscillatory increases in alkalinity anticipate growth and may regulate actin dynamics in pollen tubes of lily . Plant Cell 18 : 2182 – 2193 . Google Scholar CrossRef Search ADS PubMed Malho R , Liu Q , Monteiro D , Rato C , Camacho L , Dinis A . 2006 . Signalling pathways in pollen germination and tube growth . Protoplasma 228 : 21 – 30 . Google Scholar CrossRef Search ADS PubMed Mellema S , Eichenberger W , Rawyler A , Suter M , Tadege M , Kuhlemeier C . 2002 . The ethanolic fermentation pathway supports respiration and lipid biosynthesis in tobacco pollen . Plant Journal 30 : 329 – 336 . Google Scholar CrossRef Search ADS PubMed Mittler R , Vanderauwera S , Suzuki N , et al. 2011 . ROS signaling: the new wave ? Trends in Plant Science 16 : 300 – 309 . Google Scholar CrossRef Search ADS PubMed Munoz P , Norambuena L , Orellana A . 1996 . Evidence for a UDP-glucose transporter in Golgi apparatus-derived vesicles from pea and its possible role in polysaccharide biosynthesis . Plant Physiology 112 : 1585 – 1594 . Google Scholar CrossRef Search ADS PubMed Nashilevitz S , Melamed-Bessudo C , Aharoni A , Kossmann J , Wolf S , Levy AA . 2009 . The legwd mutant uncovers the role of starch phosphorylation in pollen development and germination in tomato . Plant Journal 57 : 1 – 13 . Google Scholar CrossRef Search ADS PubMed Nevoigt E , Stahl U . 1997 . Osmoregulation and glycerol metabolism in the yeast Saccharomyces cerevisiae . FEMS Microbiology Reviews 21 : 231 – 241 . Google Scholar CrossRef Search ADS PubMed Obermeyer G , Fragner L , Lang V , Weckwerth W . 2013 . Dynamic adaption of metabolic pathways during germination and growth of lily pollen tubes after inhibition of the electron transport chain . Plant Physiology 162 : 1822 – 1833 . Google Scholar CrossRef Search ADS PubMed Parrotta L , Faleri C , Cresti M , Cai G . 2015 . Heat stress affects the cytoskeleton and the delivery of sucrose synthase in tobacco pollen tubes . Planta 243 : 43 – 63 . Google Scholar CrossRef Search ADS PubMed Persia D , Cai G , Del CC , Faleri C , Willemse MT , Cresti M . 2008 . Sucrose synthase is associated with the cell wall of tobacco pollen tubes . Plant Physiology 147 : 1603 – 1618 . Google Scholar CrossRef Search ADS PubMed Pertl H , Pockl M , Blaschke C , Obermeyer G . 2010 . Osmoregulation in Lilium pollen grains occurs via modulation of the plasma membrane H+ATPase activity by 14-3-3 proteins . Plant Physiology 154 : 1921 – 1928 . Google Scholar CrossRef Search ADS PubMed Potocky M , Jones MA , Bezvoda R , Smirnoff N , Zarsky V . 2007 . Reactive oxygen species produced by NADPH oxidase are involved in pollen tube growth . New Phytologist 174 : 742 – 751 . Google Scholar CrossRef Search ADS PubMed Potocky M , Pejchar P , Gutkowska M , et al. 2012 . NADPH oxidase activity in pollen tubes is affected by calcium ions, signaling phospholipids and Rac/Rop GTPases . Journal of Plant Physiology 169 : 1654 – 1663 . Google Scholar CrossRef Search ADS PubMed Reinders A . 2016 . Fuel for the road – sugar transport and pollen tube growth . Journal of Experimental Botany 67 : 2121 – 2123 . Google Scholar CrossRef Search ADS PubMed Rodriguez-Garcia MI , M’rani-Alaoui M , Fernandez MC . 2003 . Behavior of storage lipids during development and germination of olive (Olea europaea L.) pollen . Protoplasma 221 : 237 – 244 . Google Scholar PubMed Rounds CM , Hepler PK , Fuller SJ , Winship LJ . 2010 . Oscillatory growth in lily pollen tubes does not require aerobic energy metabolism . Plant Physiology 152 : 736 – 746 . Google Scholar CrossRef Search ADS PubMed Rounds CM , Lubeck E , Hepler PK , Winship LJ . 2011a. Propidium iodide competes with Ca(2+) to label pectin in pollen tubes and Arabidopsis root hairs . Plant Physiology 157 : 175 – 187 . Google Scholar CrossRef Search ADS PubMed Rounds CM , Winship LJ , Hepler PK . 2011b. Pollen tube energetics: respiration, fermentation and the race to the ovule . AoB Plants 2011 : lr019 . Google Scholar CrossRef Search ADS Selinski J , Scheibe R . 2014 . Pollen tube growth: where does the energy come from? Plant Signaling and Behaviour 9: e977200 . Smirnova AV , Matveyeva NP , Yermakov IP . 2014 . Reactive oxygen species are involved in regulation of pollen wall cytomechanics . Plant Biology 16 : 252 – 257 . Google Scholar CrossRef Search ADS PubMed Speranza A , Crinelli R , Scoccianti V , Geitmann A . 2012 . Reactive oxygen species are involved in pollen tube initiation in kiwifruit . Plant Biology 14 : 64 – 76 . Google Scholar PubMed Stadler R , Truernit E , Gahrtz M , Sauer N . 1999 . The AtSUC1 sucrose carrier may represent the osmotic driving force for anther dehiscence and pollen tube growth in Arabidopsis . Plant Journal 19 : 269 – 278 . Google Scholar CrossRef Search ADS PubMed Wang L , Wang W , Wang YQ , et al. 2013 . Arabidopsis galacturonosyltransferase (GAUT) 13 and GAUT14 have redundant functions in pollen tube growth . Molecular Plant 6 : 1131 – 1148 . Google Scholar CrossRef Search ADS PubMed Williams JH . 2008 . Novelties of the flowering plant pollen tube underlie diversification of a key life history stage . Proceedings of the National Academy of Sciences of the USA 105 : 11259 – 11263 . Google Scholar CrossRef Search ADS PubMed Winship LJ , Obermeyer G , Geitmann A , Hepler PK . 2011 . Pollen tubes and the physical world . Trends in Plant Science 16 : 353 – 355 . Google Scholar CrossRef Search ADS PubMed Winship LJ , Rounds C , Hepler KP . 2017 . Perturbation analysis of calcium, alkalinity and secretion during growth of lily pollen tubes . Plants 6 : 3 . Google Scholar CrossRef Search ADS Zerzour R , Kroeger J , Geitmann A . 2009 . Polar growth in pollen tubes is associated with spatially confined dynamic changes in cell mechanical properties . Developmental Biology 334 : 437 – 446 . Google Scholar CrossRef Search ADS PubMed Zienkiewicz A , Zienkiewicz K , Rejon JD , Rodriguez-Garcia MI , Castro AJ . 2013 . New insights into the early steps of oil body mobilization during pollen germination . Journal of Experimental Botany 64 : 293 – 302 . Google Scholar CrossRef Search ADS PubMed © The Author(s) 2018. Published by Oxford University Press on behalf of the Annals of Botany Company. All rights reserved. For permissions, please e-mail: [email protected]. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)