Complex cytogeographical patterns reveal a dynamic tetraploid–octoploid contact zone

Complex cytogeographical patterns reveal a dynamic tetraploid–octoploid contact zone Research Article Complex cytogeographical patterns reveal a dynamic tetraploid–octoploid contact zone 1 1,2 3 4 1 Mariana Castro *, Sílvia Castro , Albano Figueiredo , Brian Husband and João Loureiro CFE, Centre for Functional Ecology, Department of Life Sciences, University of Coimbra, Calçada Martim de Freitas, 3000-456 Coimbra, Portugal Botanic Garden of the University of Coimbra, Calçada Martim de Freitas, 3000-456 Coimbra, Portugal CEGOT, Departamento de Geografia e Turismo, Faculdade de Letras, Universidade de Coimbra, Largo da Porta Férrea, 3004-530 Coimbra, Portugal Department of Integrative Biology, University of Guelph, Guelph, ON N1G 2W1, Canada Received: 13 October 2017 Editorial decision: 27 January 2018 Accepted: 13 February 2018 Published: 14 February 2018 Associate Editor: Adrian C. Brennan Citation: Castro M, Castro S, Figueiredo A, Husband B, Loureiro J. 2018. Complex cytogeographical patterns reveal a dynamic tetraploid–octoploid contact zone. AoB PLANTS 10: ply012; doi: 10.1093/aobpla/ply012 Abstract. The distribution of cytotypes in mixed-ploidy species is crucial for evaluating ecological processes involved in the establishment and evolution of polyploid taxa. Here, we use flow cytometry and chromosome counts to explore cytotype diversity and distributions within a tetraploid–octoploid contact zone. We then use niche model- ling and ploidy seed screening to assess the roles of niche differentiation among cytotypes and reproductive inter- actions, respectively, in promoting cytotype coexistence. Two cytotypes, tetraploids and octoploids, were dominant within the contact zone. They were most often distributed parapatrically or allopatrically, resulting in high geo- graphic isolation. Still, 16.7 % of localities comprised two or more cytotypes, including the intermediate hexaploid cytotype. Tetraploids and octoploids had high environmental niche overlap and associated with similar climatic environments, suggesting they have similar ecological requirements. Given the geographical separation and habi- tat similarity among cytotypes, mixed-ploidy populations may be transitional and subject to the forces of minor- ity cytotype exclusion which lead to pure-ploidy populations. However, seed ploidy analysis suggests that strong reproductive barriers may enforce assortative mating which favours stable cytotype coexistence. High cytogenetic diversity detected in the field suggests that unreduced gamete formation and hybridization events seem frequent in the studied polyploid complex and might be involved with the recurrent polyploid formation, governing, as well, the gene flow between cytogenetic entities. Keywords: Contact zone; distribution patterns; Gladiolus communis; hexaploid; hybridization; niche modelling; niche overlapping; octoploid; tetraploid. et al. 2011) and sympatric speciation (Otto and Whitton Introduction 2000; Soltis et al. 2010). Based on recent molecular and Polyploidization, the duplication of complete chromo- fossil studies, polyploidy has been linked with radia- some sets, is widely considered an important mech- tions in species diversity throughout evolutionary his- anism of plant evolution (Soltis and Soltis 1999; Jiao tory (Soltis et  al. 2009) and associated with 15  % of *Corresponding author’s e-mail address: mcastro@uc.pt © The Author(s) 2018. Published by Oxford University Press on behalf of the Annals of Botany Company. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/ licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. AoB PLANTS https://academic.oup.com/aobpla © The Author(s) 2018 1 Downloaded from https://academic.oup.com/aobpla/article-abstract/10/2/ply012/4857208 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Castro et al. – Cytogeographical patterns in a tetraploid–octoploid complex speciation events in extant angiosperms (Wood et  al. which detect extensive cytogenetic diversity and, in 2009). Consequently, polyploidy is pervasive in flower - - several cases, occurrence of mixed-ploidy popula ing plants. The standing incidence of polyploid species is tions (e.g. Baack 2004; Kolář et al. 2009; Trávníček et al. estimated at 35 % (Wood et al. 2009), with higher values 2010; Castro et  al. 2012; Zozomová-Lihová et  al. 2015; being observed in specific geographic regions such as Wefferling et  al. 2017), rare cytotypes (e.g. Kolář et  al. the Mediterranean basin (ranging between 37 and 47 %; 2009; Trávníček et  al. 2010), production of unreduced Marques et al. 2017) and the Arctic region (69 and 87 %; gametes (e.g. Maceira et  al. 1992; Burton and Husband Brochmann et al. 2004). 2001; Ramsey 2007; Castro et  al. 2016a) or recurrent The geographic distribution of polyploids is useful for occurrence of gene flow (e.g. Husband 2004; Kolář et al. inferring mechanisms of polyploid evolution, coexist- 2009; Castro et  al. 2011). Particularly interesting are ence and divergence. The spatial arrangement of cyto- polyploid complexes with higher ploidies, such as dip- types in situ is the result of several interacting processes loid–hexaploid (e.g. Aster amellus; Castro et  al. 2012) operating in natural populations including formation or tetraploid–octoploid complexes (e.g. Gymnadenia and migration; ecological preferences, and competitive conopsea; Jersáková et  al. 2010), that can produce and dispersal abilities; and reproductive interactions, even-ploidy hybrids, which are potentially more stable among others (Petit et  al. 1999; Levin 2002; Lexer and and lead to highly dynamic contact zones. Regardless of van Loo 2006). Cytotype distributions can be character- the increasing number of studies at contact zones, the ized as sympatric, parapatric or allopatric depending on available information is still scarce and insufficient for whether the different cytotypes grow intermixed, adja- many plant groups and regions (Soltis et al. 2010, 2016; cent or disjunct, respectively (Petit et al. 1999; and illus- Marques et al. 2017). trated in Fig. 2 of Mallet et al. 2009, which can be applied Gladiolus communis is a Mediterranean polyploid com- to polyploid complexes). Theoretical models predict that plex with high morphological variation (Alonso and Crespo within zones of sympatry, mixed-ploidy populations are 2010). Multiple ploidy levels have been described for the expected to be rare and evolutionarily unstable because complex, namely tetraploids (2n = 4x = 60 chromosomes; frequency-dependent selection will drive the exclusion Fernandes et al. 1948; Fernandes 1950; Nilsson and Lassen of the minority cytotype (Levin 1975; Rodriguez 1996; 1971; Queirós 1980; Fernández et  al. 1985) and octop- Husband and Schemske 2000). Still, numerous studies loids (2n = 8x  =  120; Fernandes and Queirós 1971; Löve have documented mixed-ploidy populations (reviewed and Kjellqvist 1973; Queirós 1980), although hexaploids in Husband et al. 2013; and examples below). The pres- (2n = 6x  =  90) and duodecaploids (2n  =  12x  =  180) have ence of multiple cytotypes in the same population can also been occasionally reported in the Mediterranean basin reflect either a transitory stage, in which neopolyploids (Darlington and Wylie 1955). The Iberian Peninsula seems are recurrently formed, or a persistent stage such as to harbour this diversity (Fernandes et al. 1948; Fernandes when cytotypes are ecologically and reproductively and Queirós 1971; Queirós 1979) and areas of close con- isolated on a small spatial scale (e.g. Kolář et  al. 2009; tact between tetraploids and octoploids have been Jersáková et al. 2010). In this context, assessing the dis- detected, for example, in calcareous regions from Central tribution of cytotypes within and among natural popula- Portugal (Castro et  al. 2016b). Occasionally, G.  communis tions is crucial to build and test hypotheses that account grows with another congeneric species, namely G. italicus, for the successful establishment of polyploids. which, in the Iberian Peninsula, is represented by duode- Contact zones, areas with two or more cytotypes caploid individuals (Queirós 1979; Pérez and Pastor Díaz growing in close proximity, are thus considered natural 1994; although octoploids have also been described in the laboratories within which to study evolutionary transi- Mediterranean basin, e.g. Susnik and Lovka 1973; Strid and tions through polyploidy. In recent years, an increas- Franzen 1981; van Raamsdonk and De Vries 1989; Kamari ing number of studies have provided insights into et al. 2001). The high morphological variation of the group ploidy-mediated processes occurring in contact zones has led taxonomists to accept multiple taxonomic entities (e.g. Husband et  al. 2013; Ramsey and Ramsey 2014). within the G. communis complex (e.g. Gussone 1832; van Significant advances in this field have been largely Raamsdonk and De Vries 1989), although morphologic- fuelled by the ability to rapidly and easily screen thou- ally intermediate forms are found in natural populations, sands of individuals using flow cytometry (Kron et  al. and many characters used to distinguish each taxon are 2007). This approach has resulted in a proliferation of extremely variable and largely overlap, even within popula- cytogeographical studies (e.g. Baack 2004; Kolář et  al. tions (Hamilton 1980; revised in Alonso and Crespo 2010). 2009; Ståhlberg 2009; Trávníček et  al. 2010; Castro Consequently, recent morphological reviews and prelimin- ary molecular analyses failed to support the previous taxo- et  al. 2012; Zozomová-Lihová et  al. 2015; Wefferling et  al. 2017; reviewed in Ramsey and Ramsey 2014), nomic delimitations and the species is currently accepted 2 AoB PLANTS https://academic.oup.com/aobpla © The Author(s) 2018 Downloaded from https://academic.oup.com/aobpla/article-abstract/10/2/ply012/4857208 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Castro et al. – Cytogeographical patterns in a tetraploid–octoploid complex as a complex formed by three ploidy levels (Buchanan bulb, relatively thick roots, a cylindrical glabrous stem 2008; Alonso and Crespo 2010). Regardless of the variabil- and linear leaves with typical parallel ribs. The pink bisex- ity detected in the species, nothing is known about the role ual flowers are zygomorphic and usually grouped in one of genome duplications generating diversity within this spiked inflorescence per individual. A  second Gladiolus polyploid complex. Exploring cytotype diversity and dis- species, G. italicus, is found on the Iberian Peninsula and tribution patterns, especially at contact zones, is thus cru- occurs in sympatry with G.  communis in some places. cial to understand ecological processes, such as ecological Although very similar morphologically, these two spe- preferences and reproductive interactions, driving current cies are easily distinguished based on inflorescence diversity patterns at natural contact zones. architecture, anther and filament lengths, and seed In this study, we explore in detail the cytotype diver- morphology. Gladiolus communis has a unilateral inflor - sity and distribution patterns in a tetraploid–octoploid escence, anthers equalling or shorter than the filaments, G. communis contact zone. In particular, we pose the fol- and broadly winged seeds, while G.  italicus usually has lowing specific questions: (i) what are the dominant cyto - a weakly distichous inflorescence, anthers longer than types and how are they distributed in the contact zone? the filaments, and polyhedric apterous seeds (Hamilton (ii) do cytotypes coexist and at which spatial scale? (iii) 1980; Alonso and Crespo 2010). is coexistence facilitated by differences in environmen- In the Iberian Peninsula, G.  communis is recog- tal associations between cytotypes? And finally, (iv) is nized as a polyploid complex comprising tetraploids there evidence for the production of unreduced gametes (2n = 4x = 60 chromosomes), hexaploids (2n = 6x = 90) and/or cytotype hybridization? To address our questions, and octoploids (2n = 8x = 120) (e.g. Fernandes et al. 1948; cytotype diversity was studied at several spatial scales, Fernandes and Queirós 1971; Alonso and Crespo 2010) namely (i) across the contact zone, to characterize the with duodecaploids being described elsewhere in the most dominant cytotypes and their environmental pref Mediterranean region (Darlington and Wylie 1955). The erences within areas of contact; (ii) within mixed-ploidy high morphological resemblance among G.  communis populations, to measure microhabitat segregation; and cytotypes (Alonso and Crespo 2010; Cantor and Tolety (iii) among offspring from plants in pure- and mixed- 2011) suggests a putative autopolyploid origin. The spe- ploidy populations, to detect cytotype diversity at early cies is common in the calcareous regions from Central stages. Flow cytometric analyses complemented with Portugal, where preliminary field sampling revealed the chromosome counts were used to assess ploidy levels presence of tetraploid and octoploid populations grow- of all the sampled individuals. The reproductive success ing in close proximity. This study focused on this contact of pure- and mixed-ploidy populations was also quanti- zone, an area extending from 39.3° to 40.6° in latitude, fied in natural conditions to depict fitness differences and from 7.8° to 9.4° in longitude. This territory is domi- between cytotypes. The spatial arrangement of cyto- nated by calcareous rocks and presents a Mediterranean types in the contact zone was analysed with niche mod- climate that exhibits a strong influence from the Atlantic elling tools to determine if differences in environmental Ocean, an attribute identified on the significant values requirements could explain cytotype distribution. If cyto- of annual precipitation (1000–1300 mm). However, the types differ in environmental requirements, we expect dominance of poor soils determines a low water storage a mosaic contact zone with tetraploids and octoploids capacity, which, combined with a long and hot summer, fairly isolated within a given spatial scale and with plants determines the dominance of evergreen vegetation. growing in different habitats or microhabitats. If no envir- Allied to such climatic conditions, human pressure con- onmental differences are observed, we expect a tension tributed to current dominance of shrubby communities zone where sympatric cytotype co-occurrence is possible, in the landscape, and constrained forests (evergreen where intermediate cytotypes are detected and where and semi-deciduous) to very small patches, favouring other processes such as reproductive barriers, competi- the wide presence of open habitats. These open habi- tion or dispersal abilities are expected to play major roles tats are also characterized by the presence of limestone in driving distribution patterns. outcrops exposed to stressful ecological conditions that limit the installation of higher vegetation covers. Although not exhaustive, additional sampling was Methods extended beyond this area to determine the dominant cytotypes within the species. Also, because G. communis Study system and studied region coexist with G. italicus, hybridization might occur and gen- erate additional cytogenetic diversity, the duodecaploid Gladiolus communis is a perennial species that is wide- G. italicus (Queirós 1979; Pérez and Pastor Díaz 1994) was spread on the Iberian Peninsula and throughout the Mediterranean basin. The species produces an ovoid also sampled whenever growing with G. communis. AoB PLANTS https://academic.oup.com/aobpla © The Author(s) 2018 3 Downloaded from https://academic.oup.com/aobpla/article-abstract/10/2/ply012/4857208 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Castro et al. – Cytogeographical patterns in a tetraploid–octoploid complex ploidy were sampled to determine reproductive success Field sampling and screen ploidy of the seeds (see section Reproductive Field collections were carried out during the flowering success in natural populations). and fruiting seasons (mid-April to July) of G.  commu- nis from 2012 to 2015. Individual plants or clusters of Chromosome counts plants were easily detected when blooming because Chromosome counts were used to calibrate genome of the tall inflorescences growing above the remaining size estimates, obtained using flow cytometry, to a vegetation. We sampled 81 populations across the con- given ploidy level. For this, the plants grown from bulbs tact zone where both tetraploid and octoploid cytotypes collected in the selected natural populations and main- have been previously detected in close proximity. An tained in the common garden were used simultaneously additional group of 27 populations covering the west- for genome size estimates and chromosome counting. ern distribution of the species around the contact zone For chromosome counts, we followed the protocol of was also sampled to depict the dominant cytotypes [see Goldblatt and Takei (1993), with some adjustments. Supporting Information—Table S1]. In each of the 108 Briefly, actively growing root tips were harvested and populations, we collected ~3  cm of fresh leaf of up to pretreated in 0.002 M aqueous 8-hydroquinoline at 53 individuals of G.  communis (with an average of 20 room temperature for 1630  h, and fixed in 95  % etha - individuals per locality, excluding two particularly large nol and glacial acetic acid (in a ratio of 3:1) for at least localities where more intensive sampling was done, with 48 h at 4 °C. Roots tips were hydrolysed in 1 M hydrogen 106 and 454 plants being screened), and of G.  italicus chloride at 60 °C in a sand bath for 40 min, submerged whenever detected growing with G.  communis (up to in Schiff reagent (Greilhuber and Ebert 1994) for 1330 h, 32 individuals, averaging 13 plants per locality). The washed in sulphur water for three periods of 10 min and sampled individuals were randomly selected, covering finally squashed under a glass cover in a drop of acetic the extension of the population. Leaves were stored in orcein 2  %. Chromosome spreads were observed using labelled hermetic bags and maintained at 4 °C for later a Nikon Eclipse 80i light microscope and photographed flow cytometric analysis (see section Genome size and using a Nikon Plan Apo VC 100×/1.40 oil-immersion lens DNA ploidy level estimates). Geographic coordinates of with a Q Imaging Retiga 2000R Fast 1394 digital camera the population were recorded. Bulbs of nine localities and Q-Capture Pro v.7 software. A  total of 40 individu- identified in preliminary surveys as DNA tetraploid, DNA als from nine populations were used to access chromo- hexaploid and DNA octoploid populations (Castro et  al. some number and genome size: 4x—populations MC147 2016b) were also collected, potted and maintained at (n = 10 individuals), MC193 (n = 1), MC195 (n = 4), MC201 the common garden for chromosome counts (see sec- (n = 1) and MC212 (n = 2); 6x—population MC211 (n = 4); tion Chromosome counts). 8x—populations MC032 (n  =  8), MC143 (n  =  3), MC190 In addition, we sampled in mixed-ploidy popula- (n = 4), MC193 (n = 2) and MC201 (n = 1) [see Supporting tions more intensively to test for microhabitat seg- Information—Table S1]. regation. Three mixed-ploidy populations containing tetraploids, hexaploids and/or octoploids were revisited Genome size and DNA ploidy level estimates and all adult, individuals (both vegetative and repro- ductive individuals) were mapped with x/y coordinates, To estimate genome size and DNA ploidy levels, fresh tagged and sampled for ploidy level analyses using flow leaves collected in natural populations were analysed cytometry (see section Genome size and DNA ploidy using flow cytometry. Nuclear suspensions were pre - level estimates). To delimit the clusters of plants grow- pared following Galbraith et  al. (1983) by chopping the ing in sympatry, screenings for Gladiolus plants were plant material of the sampled species together with leaf made around a radius of over 150 m around the cluster tissue of an internal reference standard. In the case of of plants initially detected or until an anthropogenic or Gladiolus nuclear suspensions, 100  mg of leaf tissue or natural barrier was observed. Additional mixed-ploidy 2–5 seeds were co-chopped with 50 mg of leaf of Solanum populations were not sampled because they were dis- lycopersicum ‘Stupické’ (2C = 1.96 pg; Doležel et al. 1992) turbed by grazing or human activities. or Pisum sativum ‘Ctirad’ (2C  =  9.09 pg; Doležel et  al. Finally, we screened offspring from plants in pure- 1998). Solanum lycopersicum was used as the internal and mixed-ploidy populations to examine the produc- standard in most cases, except when unavailable, with tion of unreduced gametes and/or hybridization events P.  sativum being used in those situations. Sample and by the detection of rare cytotypes that might not reach standard were co-chopped in 1 mL of WPB buffer (WPB: the adult stage. For this, four tetraploid, two hexaploid, 0.2 M Tris–HCl, 4  mM MgCl ·6H O, 1  % Triton X-100, 2 2 four octoploid and one mixed tetraploid–octoploid pop- 2  mM EDTA Na ·2H O, 86  mM NaCl, 10  mM metabisul- 2 2 ulations were revisited and individual plants with known fite, 1 % PVP-10, pH adjusted to 7.5 and stored at 4 °C; 4 AoB PLANTS https://academic.oup.com/aobpla © The Author(s) 2018 Downloaded from https://academic.oup.com/aobpla/article-abstract/10/2/ply012/4857208 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Castro et al. – Cytogeographical patterns in a tetraploid–octoploid complex Loureiro et  al. 2007) using a razor blade. The resulting that of the endosperm. Consequently, the interpretation nuclear suspension was filtered through a 50  µm nylon of each histogram was made with particular caution, −1 filter and 50  µg mL propidium iodide (Fluka, Buchs, determining the ploidy levels of all the peaks obtained in −1 Switzerland) and 50 µg mL RNAse (Fluka) were added to the histogram. Preliminary analyses revealed that hexa- the sample, to stain the DNA and avoid staining of dou- ploid populations presented higher variability and thus ble-stranded RNA, respectively. After 5 min of incubation, only two seeds were pooled, in order to unambiguously DNA fluorescence of the sample was analysed using a assign the DNA ploidy levels of each seed. Partec CyFlow Space flow cytometer (532 nm green solid- The holoploid genome size (2C in pg; sensu Greilhuber state laser, operating at 30 mW; Partec GmbH, Görlitz, et al. 2005) was obtained using the following formula: Germany). Using Partec FloMax software v2.4d (Partec Holoploid reference GmbH, Münster, Germany) the following four histograms G.communisG peak mean genome = × standard were obtained: fluorescence pulse integral in linear scale referencestanda ardG peak mean size(pg) 1 genome size (FL); forward light scatter (FS) vs. side light scatter (SS), both in logarithmic (log) scale; FL vs. time; and FL vs. SS in log scale [see Supporting Information—Fig. S1]. To Based on the chromosome counts obtained in this study digitally remove some of the debris, the FL histogram and respective genome sizes, as well as the four chro- was gated using a polygonal region defined in the FL vs. mosome numbers described in the literature for G. com- SS histogram (see R1 in Supporting Information—Fig. munis and G. italicus, DNA ploidy levels were inferred for S1), and was further applied to all the other graphics. At each sample and individual. Populations were then char- least 1300 nuclei in both sample and standard G peaks acterized according to their DNA ploidy composition. were analysed per sample (Suda et al. 2007). Only coef- Descriptive statistics of holoploid genome size were ficient of variation (CV) values of G peak of G. communis calculated for each cytotype and species (mean, stand- below 5  % were considered acceptable (see examples ard deviation of the mean, coefficient of variation of the in Supporting Information—Fig. S1), otherwise a new mean, maximum and minimum values) based on the sample was prepared and analysed until this quality individual flow cytometric estimates. Mean and stand - standard was achieved (Greilhuber et al. 2007). ard deviation of the mean were also calculated for the Genome size was estimated in 41 populations by monoploid genome size (1Cx; holoploid genome size analysing three plants per population and cytotype indi- divided by inferred DNA ploidy level; sensu Greilhuber vidually (rarely less, unless there were no more plants et  al. 2005). Differences in holoploid and monoploid in the locality, while in a few populations up to 30 indi- genome sizes among species and cytotypes were inves- viduals were analysed for genome size) [see Supporting tigated using linear models (hereafter LM) performed Information—Table S2]. For the remaining individu- in R software version 3.0.1 (R Core Development Team als and populations, only DNA ploidy level information 2016), using the packages ‘car’ for Type-III analysis of was gathered following the pooled sample strategy variance (Fox et  al. 2015), ‘lme4’ for generalized linear (5–6 individuals plus the reference standard). A total of models (GLMs; Bates et  al. 2014) and ‘multcomp’ for 108 natural populations of G. communis and 2665 indi- multiple comparisons after Type-III analysis of variance viduals were sampled and analysed [see Supporting (Hothorn et al. 2017). Information—Table S1]. The geographical isolation index (GI) between the two We used flow cytometry to measure DNA ploidy of dominant cytotypes (i.e. tetraploids and octoploids) at offspring produced by plants of known ploidy. A total of the contact zone was calculated according to the fol- 1252 seeds from 178 individuals from four tetraploid, lowing formula (Husband et al. 2016), where only pure- two hexaploid and four octoploid pure-ploidy popula ploidy and mixed-ploidy populations of tetraploids and tions and one tetraploid–octoploid mixed population octoploids from the contact zone were considered: were analysed. We sampled 10–15 seeds per mater- nal individual, and 7–15 individuals per population and no. mixed-ploidy populations GI=- 1 cytotype. For pure-ploidy populations of tetraploids and total no. of populations octoploids and mixed-ploidy population, five seeds were chopped simultaneously with the internal reference Environmental preferences standard (pooled sample strategy) following the proto- col described above, producing easy to interpret histo- The environmental associations of the two dominant grams. When analysing the seeds, at least two peaks cytotypes were evaluated through GLM, and spatial pre- (plus the peak of the internal standard) were always dictive models were produced based on niche modelling obtained, corresponding to the peak of the embryo and tools, aiming to assess niche overlap. To explore niche AoB PLANTS https://academic.oup.com/aobpla © The Author(s) 2018 5 Downloaded from https://academic.oup.com/aobpla/article-abstract/10/2/ply012/4857208 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Castro et al. – Cytogeographical patterns in a tetraploid–octoploid complex overlapping, two approaches were used considering two was explored using Pearson coefficient for continuous different spatial scales: (i) one with an extension encom- variables and Spearman’s ρ for categorical variables, to passing the contact zone in Central Portugal; and (ii) the assist variable selection by removing variables with cor- other extension encompassing the entire territory of relation values higher than 0.7. The final set of variables mainland Portugal. selected included the following four which were also Variables were extracted from the following sources important descriptors of the type of habitat where the with a resolution of ~111 m: (i) bioclimatologi- species grows: mean annual total precipitation, mean cal data from http://home.isa.utl.pt/~tmh/aboutme/ temperature of the hottest month, soil texture and pH Informacao_bioclimatologica.html (methodology to (highlighted in bold in Table 1). obtain variables in Monteiro-Henriques et  al. 2016); Spatial predictive models were calibrated based on and (ii) data for soil conditions from: http://epic-web presence/absence records collected in the field and gis-portugal.isa.ulisboa.pt/. Values for climatic and soil the selected environmental and soil variables (Table 1). variables were extracted for all the surveyed popula- For the tetraploid data set, tetraploid populations were tions using the R package ‘dismo’ (Hijmans et al. 2017). recorded as presences and octoploid populations as Then, GLMs were used to explore climatic and soil vari- absences, and vice versa for the octoploid data set. ables and assess differences between tetraploid and Mixed tetraploid–octoploid populations were consid- octoploid populations (Table 1), namely for climatic vari- ered as presences for both cytotypes. For the contact ables [mean annual total precipitation (PP), mean tem- zone (Central Portugal) we used data from 76 sampling perature of the hottest month of the year (Tmax), mean points (including 33 tetraploid, 40 octoploid and 3 tetra- temperature of the coldest month of the year (Tmin), ploid–octoploid populations), corresponding to all the mean maximum temperature of the coldest month of known occurrences of G. communis with a minimum dis- the year (M), mean minimum temperature of the coldest tance between populations of 600 m.  For the territory month of the year (m)], bioclimatic indexes [continental- of Portugal, and aiming to reduce the bias effect of spa- ity index (IC), compensated thermicity index (ITC), sum- tial clustering associated with our intense screening in Table 1. Characterization of the climatic and soil variables for tetraploid and octoploid populations of Gladiolus communis in the contact zone of Central Portugal. The mean, standard error of the mean (SE) and statistical tests (comparison between cytotypes) are provided for each variable and cytotype. Significance levels: ***P < 0.01; *0.05 < P < 0.01; n.s., non-significant. In bold the variables used in niche modelling are highlighted. Variables were extracted from the following sources with a resolution of ~111 m: (i) bioclimatological data from http://home.isa. utl.pt/~tmh/aboutme/Informacao_bioclimatologica.html (methodology to obtain variables in Monteiro-Henriques et  al. 2016); and (ii) data for soil conditions from: http://epic-webgis-portugal.isa.ulisboa.pt/. Variables Code Tetrapoid Octoploid ANOVA F value 1, 78 Mean ± SE, n = 43 Mean ± SE, n = 36 a a Precipitation PP 1096.11 ± 21.97 1106.89 ± 21.74 0.12 n.s. a b Mean temperature of the hottest month of the year Tmax 20.51 ± 0.13 20.95 ± 0.19 4.04* a a Mean temperature of the coldest month of the year Tmin 9.06 ± 0.10 8.91 ± 0.12 1.10 n.s. a a Mean max. temp. of the coldest month of the year M 13.52 ± 0.11 13.42 ± 0.12 0.38 n.s. a a Mean min. temp. of the coldest month of the year m 4.61 ± 0.08 4.50 ± 0.09 0.84 n.s. a b Continentality index IC 11.44 ± 0.13 12.04 ± 0.19 6.95* a a Compensated thermicity index ITC 327.12 ± 2.78 325.81 ± 3.22 0.10 n.s. a a Summer ombrothermic index Ios3 1.10 ± 0.03 1.09 ± 0.03 0.05 n.s. a a Soil texture Texture 2.14 ± 0.29 2.25 ± 0.13 0.11 n.s. a b Soil pH pH 308.60 ± 46.51 99.44 ± 19.69 15.02*** a a Altitude Alt 198.61 ± 18.63 169.23 ± 17.38 0.94 n.s. a a Latitude Lat −8.47 ± 0.05 −8.58 ± 0.03 3.14 n.s. a a Longitude Long 39.98 ± 0.05 40.01 ± 0.04 0.18 n.s. mer ombrothermic index (Ios3)], soil conditions [texture the contact zone, only occurrences that had a minimum (txt) and pH] and altitude. Correlation between variables distance of 10 km between them were selected, based 6 AoB PLANTS https://academic.oup.com/aobpla © The Author(s) 2018 Downloaded from https://academic.oup.com/aobpla/article-abstract/10/2/ply012/4857208 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Castro et al. – Cytogeographical patterns in a tetraploid–octoploid complex on radon selection, resulting in a subset of 66 sampling (Broennimann et al. 2012). In niche similarity (similarity points (including 35 tetraploid, 19 octoploid and 6 tetra- test), we evaluate if the environmental niches of the two ploid–octoploid populations). cytotypes were distinguishable from each other. In this Environmental niche modelling (ENM) of tetraploids case, the comparison was between the points of one and octoploids was created using R package ‘biomod2’ cytotype and random points from the geographic range (Thuiller et  al. 2016). Final model for each cytotype is of the other cytotype. As in the identity test, the pro- based on the combination of results from different mod- cess was repeated 100 times and D values were calcu- elling techniques, each one replicated 30 times after lated. The results revealed if niche overlap between the data splitting into training (70  %) and testing (30  %) cytotypes is greater (niche conservation) or lower (niche subsets based on random selection, aiming to reduce divergence) than expected, according to the geographic uncertainty and to produce robust models (Phillips et al. region of the other cytotype. All the models and analy- - ses were performed in R software version 3.0.1 (R Core 2006; Araújo and New 2007). In the resampling repli cation, each specific occurrence was used only once in Development Team 2016). each run, as training or as test without replacement, Reproductive success in natural populations making all replicates statistically independent (Phillips 2008). Models were evaluated based on the independ- The reproductive success of each cytotype was evalu- ent accuracy measure AUC of ROC (area under the curve ated in 11 natural populations, namely 10 pure-ploidy of the receiver operating characteristic), and only those populations (including four tetraploid, two hexaploid with AUC > 0.7 were used in the ensemble forecast- and four octoploid populations) and one mixed-ploidy ing procedure, the approach used to produce the final population composed by tetraploid and octoploid indi- model for each cytotype. viduals (MC201). In each population, 11–20 individuals Model evaluation revealed high ROC values (contact of known ploidy level were labelled and infructescences zone: 4x—0.79  ±  0.01 and 8x—0.79  ±  0.01; Portugal: collected in individually labelled bags. The number of 4x—0.77  ±  0.01 and 8x—0.76  ±  0.01) and relatively fruits was counted for each inflorescence and fruit set low omission rates (contact zone: 4x—0.19  ±  0.02 and calculated as the proportion of flowers that developed 8x—0.28  ±  0.02; Portugal: 4x—0.23  ±  0.01 and 8x— into fruit. The number of morphologically viable seeds 0.28 ± 0.01). However, when considering the binary pro- (based on their size and shape) was assessed in all fruits, jections, the omission rates decrease to 0.10 and 0.09 and the seed:ovule ratio (S:O ratio) was calculated by for the tetraploid and octoploid models in the contact dividing the number of morphologically viable seeds by zone, respectively, and 0.17 and 0.04 in Portugal (tetra- the number of ovules. The total reproductive success of ploids and octoploids, respectively), demonstrating that populations and cytotypes was also calculated by multi- the models were able to predict the occurrences with plying the S:O ratio by the fruit set. Descriptive statistics high accuracy, namely for octoploids. The binary projec- were calculated for each population type. tion produced by the final model of each cytotype was Differences in fruit set, S:O ratio and total reproduc- used to calculate niche overlap. tive success between the three cytotypes (tetraploids, Cytotype niche overlap was quantified through the hexaploids and octoploids) within pure-ploidy popula- metric of proportional similarity of the distribution of tions, differences between tetraploids and octoploids in both cytotypes, using Schoener’s D (a measure of niche the mixed-ploidy population, and differences between similarity; Schoener 1970). This metric ranges from zero pure- and mixed-ploidy populations (excluding hexa- (no overlap) to one (complete overlap). The ‘ecospat’ ploid ones) were assessed using GLM. Mixed models with (Broennimann et  al. 2012) and ‘raster’ (Hijmans et al. individual and population as random factors were ini- 2017) packages were used to perform niche identity and tially used, but the random factors were further removed similarity tests (Warren et  al. 2008; Broennimann et  al. due to low variance in comparison with residuals (Bolker 2012). In niche equivalency (identity test), the points et al. 2009). A binomial distribution with a logit link func- of both cytotypes were pooled and randomly split in tion was used for fruit set, and a Gaussian distribution two groups according to size of the original data set. with an identity link function was used for S:O ratio and This new data set was used in D calculation, and the total reproductive success after transformation with the process was repeated 100 times (to obtain confidence arcsine of the square root. When significant differences intervals that enable evaluation of the null hypothesis). were obtained, post hoc tests for multiple comparisons The resulting D values (simulated values) were com- were performed. pared with the observed D value, and cytotype niches All analyses were performed in R software ver- were considered equivalent if the observed D value fell sion 3.0.1 (R Core Development Team 2016), using within the 95th percentile of the simulated D values the packages ‘car’ for Type-III analysis of variance AoB PLANTS https://academic.oup.com/aobpla © The Author(s) 2018 7 Downloaded from https://academic.oup.com/aobpla/article-abstract/10/2/ply012/4857208 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Castro et al. – Cytogeographical patterns in a tetraploid–octoploid complex (Fox and Weisberg 2015), ‘lme4’ for GLMs (Bates et  al. segregation pattern of cytotype arrangement in space 2014) and ‘multcomp’ for multiple comparisons after was observed: tetraploids seem to occur across the Type-III analysis of variance (Hothorn et al. 2017). entire area surveyed, and octoploids in the centre and south of the surveyed area, forming broad contact zones. Minority cytotypes were also detected, namely Results hexaploids, which were observed growing with other cytotypes and occasionally found forming pure popula- Genome size and cytogenetic diversity tions (Fig. 3). A few nonaploids in a mixed-ploidy popu- Based on chromosome counts and flow cytometric anal - lation harbouring all cytotypes of G. communis were also yses, we detected three ploidy levels in G.  communis: detected (Fig.  3) [see Supporting Information—Table tetraploids with 2n = 4x = 60 chromosomes (Fig. 1A) and S3]. Most populations were cytogenetically uniform (i.e. an average genome size of 2.69  ±  0.06 pg/2C (mean ± pure-ploidy populations, 86.1 %) and, in the majority of SD); hexaploids with 2n = 6x = 90 chromosomes (Fig. 1B) cases, were composed of either tetraploid or octoploid and an average genome size of 4.07 ± 0.07 pg/2C; and individuals (43.5 and 39.8  %, respectively). Hexaploids octoploids with 2n = 8x = 120 chromosomes (Fig. 1C) and were detected growing alone in three locations (2.8 %) an average genome size of 5.42  ±  0.14 pg/2C (Table  2; (Fig.  3). Populations harbouring two or more cytotypes Fig. 2A and B) [see Supporting Information—Table S2]. (i.e. mixed-ploidy populations) represented 13.9  % of Genome size estimates also suggest the occurrence of all sampled populations. The mixed-ploidy populations nonaploid G.  communis individuals, characterized by presented different cytotype compositions: tetraploids genome sizes with nine times the monoploid genome size and hexaploids (4.6 %), in which the former is more fre- (1Cx) values obtained for the other ploidy levels, and had quent than the latter; tetraploids and octoploids (5.6 %) mean genome size of 6.10 ± 0.18 pg/2C (Table 2; Fig. 2C). again, in which tetraploids are generally more abundant These individuals were rare and we were unable to con- than octoploids, except in one population; tetraploids, firm their ploidy using chromosome counts. Gladiolus hexaploids, octoploids and nonaploids (0.9  %; one italicus had a higher genome size (2C = 7.27 ± 0.17 pg) population), where octoploids are the dominant cyto- than G.  communis, consistent with duodecaploids, as type; and hexaploids and octoploids (2.8  %), in which described for the species (Table  2; Fig.  2D). The holop- octoploids are dominant, except in one location where loid genome sizes (2C) of the five cytotypes differed sig - only two plants, one of each cytotype, were found [see nificantly (F  = 7691.3, P < 0.001; Table 2). Monoploid Supporting Information—Table S3]. 4, 175 genome size values were conserved within G. communis, Within the contact area (Fig. 3B), most localities con- with no significant differences being observed between tained a single ploidy of either tetraploids (42.0  %), cytotypes (F  = 7691.3, P = 0.5272; Table 2). However, octoploids (44.4  %) or rarely hexaploids (2.5  %). These 3, 155 monoploid genome size of G. communis (0.67 ± 0.03 pg) populations were distributed mostly in parapatry; still, was significantly higher than for G.  italicus (0.61 ± 0.01 cytotypes were found growing in sympatry in some loca- pg; F  = 7691.3, P < 0.001). tions (11.1 %) (Fig. 3B). Octoploid populations occur from 1, 178 north to south, resulting in cytogenetically diverse con- Geographic distribution of cytotypes tact zones with tetraploids to the east, south and south- Tetraploids and octoploids were prevalent across the west. At these contact zones, areas with different types geographic area sampled, both occurring in pure- of mixed-ploidy populations were detected. Hexaploids and in mixed-ploidy populations (Fig.  3). No marked were frequent in the contact zones between tetraploids Figure 1. Gladiolus communis chromosome counts. (A) tetraploid (2n = 4x = 60 chromosomes), (B) hexaploid (2n = 6x = 90) and (C) octoploid (2n = 8x = 120) individuals. Bar = 20 µm. 8 AoB PLANTS https://academic.oup.com/aobpla © The Author(s) 2018 Downloaded from https://academic.oup.com/aobpla/article-abstract/10/2/ply012/4857208 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Castro et al. – Cytogeographical patterns in a tetraploid–octoploid complex Table 2. Genome size and DNA ploidy level estimations for Gladiolus communis and G. italicus. Holoploid genome size (G.s.; 2C) is provided for each species and cytotype as mean and standard deviation (SD) in picograms (pg), followed by coefficient of variation (CV, %), DNA range (minimum, Min, and maximum, Max, genome size values); mean and standard deviation of the mean is also provided for the monoploid genome size (1Cx); total number of individuals (N total) and populations (N pop) analysed for genome size are also provided for each ploidy level. Chromosome numbers (Chr. number) for each species based on chromosome counts of this study and bibliographic records are also provided. DNA ploidy levels: tetraploid (4x), hexaploid (6x), octoploid (8x), nonaploid (9x) and duodecaploid (12x). Different letters denote 1 2 statistically significant differences at P  <  0.05. Chromosome numbers detected in this study; Chromosome counts documented in the bibliography; DNA ploidy level extrapolated based on the genome size values obtained here and on the chromosome counts available from other ploidy levels. Species DNA ploidy level Chr. Holoploid G.s. (2C, pg) Monoploid G.s. N total N pop number (1Cx, pg) Mean SD CV (%) Min Max Mean SD 1,2 a a G. communis 4x 60 2.69 0.06 2.28 2.58 2.86 0.67 0.02 57 16 1,2 b a G. communis 6x 90 4.07 0.07 1.76 3.93 4.19 0.68 0.01 9 3 1,2 c a G. communis 8x 120 5.42 0.14 2.53 5.13 5.73 0.68 0.02 91 21 3 d a G. communis 9x 135 6.10 0.18 2.89 5.98 6.23 0.68 0.02 2 1 2 e b G. italicus 12x 180 7.27 0.17 2.31 6.97 7.55 0.61 0.01 21 10 and octoploids, although they were also detected in diversity (MC193) revealed to be dominated by octoploid other places of the screened area, growing with tetra- individuals with a few tetraploid, hexaploid and nona- ploid individuals. Tetraploids and octoploids, the two ploid plants growing intermingled (Fig. 4C). While MC201 main cytotypes, were observed growing together in four and MC193 were located in the contact zones, MC232 is locations out of the 81 populations at the contact zone located in an otherwise tetraploid zone (Fig. 3). (4.9  %), resulting in a total GI of 0.95, with tetraploids Environmental preferences and octoploids presenting a similar individual geograph- Niche geographic overlap between tetraploids and ical isolation index (GI  = 0.90, GI  = 0.91). 4x 8x octoploids at both the contact zone (Schoener’s D met- The detailed screening of three selected mixed-ploidy populations revealed variable patterns of cytotype dis- ric, D = 0.03) and Portugal (D = 0.01) was low (Table  3). However, and despite little geographical overlap, there tribution within each population (Fig.  4). In the tetra- was no statistical evidence that the environmental ploid–octoploid population (MC201), cytotypes were distributed in two well-defined clusters separated by >20 niches differed, i.e. neither niche equivalency nor niche similarity was rejected (Table  3). This indicates that m, with tetraploids being restricted to the north-east environmental niche of the dominant cytotypes was side and octoploids to the south-west of the population equivalent within the suitable ranges of both tetraploids (Fig.  4A). The mixed-ploidy population with tetraploids and octoploids, and that environmental niche of each and hexaploids (MC232) was dominated by tetraploid cytotype was similar to the suitable range of the other individuals, with a few hexaploids growing intermingled cytotype. At the contact zone, the selected climatic and (Fig.  4B). The population with the highest cytogenetic Figure 2. Flow cytometric histograms of relative propidium iodide fluorescence from nuclei isolated from fresh leaves of Solanum lycopersi- cum ‘Stupické’ (S.l.) and different cytotypes and/or species of Gladiolus: (A) tetraploid (4x) and hexaploid (6x), (B) octoploid (8x) and (C) nona- ploid (9x) individuals of G. communis, and (D) duodecaploid (12x) individual of G. italicus. AoB PLANTS https://academic.oup.com/aobpla © The Author(s) 2018 9 Downloaded from https://academic.oup.com/aobpla/article-abstract/10/2/ply012/4857208 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Castro et al. – Cytogeographical patterns in a tetraploid–octoploid complex Figure 3. Gladiolus communis cytotype screening: (A) all studied area (Portugal); and (B) detail of the contact zone studied (Central Portugal). White, grey and black circles represent pure tetraploid, hexaploid and octoploid populations, respectively. Mixed-ploidy populations are rep- resented by a grey diamond and each population is accompanied by a pie diagram reflecting cytotype composition. One sole population harbouring also two nonaploid individuals (not included in the pie diagram) is denoted by a dotted grey diamond, namely population MC193. Populations identified with ID code correspond to the populations where all the individual plants were sampled in detail (see Fig.  4). DNA ploidy levels: tetraploid (4x), hexaploid (6x), octoploid (8x). soil variables explained 62.98  % of the variance in the having lower fruit set than tetraploids. S:O ratio and distribution (Fig. 5A), and a high environmental overlap reproductive success were similar in tetraploids and of a given cytotype within the niche of the opposite cyto- octoploids (P > 0.05), but significantly higher than type was observed (74.87 and 61.95  % for tetraploids for hexaploids (P < 0.05; Fig.  6B and C). Within the and octoploids, respectively; Fig.  5B). A  similar pattern mixed-ploidy population, no significant differences was observed for Portugal, although the climatic and were observed between the cytotypes for any of the soil variables explained higher variance than at the reproductive variables (fruit set: F  = 0.27, P = 0.603; 1, 79 contact zone (74.78  %; Fig.  5C). A  high environmental S:O ratio: F  = 0.01, P = 0.934; reproductive success: 1, 37 overlap between cytotypes was also observed (91.51 F  = 0.15, P = 0.698). 1, 79 and 47.96 % for tetraploids and octoploids, respectively; The analyses of offspring ploidy revealed that tetra- Fig. 5D). ploid and octoploid individuals, in both pure-ploidy and mixed-ploidy populations, produced seeds with the same Reproductive success in natural populations and ploidy as the mother plants (Fig. 6C). Tetraploid plants in offspring cytogenetic composition pure populations produced a few aneuploids (<1  % of Plants in all the natural populations successfully the offspring; Fig.  6C) [see Supporting Information— formed fruits and seeds. However, the success differed Table S4]. In contrast, the flow cytometric analyses of according to the cytotype and population type. Pure- the seeds from hexaploid individuals pointed out highly ploidy populations (excluding the hexaploid popula- variable genome sizes, the analyses of the genome size tions) had higher reproductive success compared to estimates suggest the following DNA ploidy levels: 62 % the mixed-ploidy population for all parameters (fruit of seeds were aneuploid, 20  % were pentaploids and set: F   =  15.51, P  <  0.001; S:O ratio: F   =  4.62, 18  % were hexaploid, although further confirmation is 1,  1033 1,  706 P  =  0.032; reproductive success: F   =  21.04, needed. 1,  1033 P < 0.001; Fig. 6). Within pure-ploidy populations, sig- nificant differences between cytotypes were observed Discussion for all the variables (fruit set: F   =  4.96, P  =  0.007; 2,  1087 S:O ratio: F   =  100.18, P  <  0.001; reproductive This study corroborates the existence of high cytogenetic 2,  770 success: F   =  28.34, P  <  0.001), with octoploids diversity within the G. communis polyploid complex. Two 2,  1087 10 AoB PLANTS https://academic.oup.com/aobpla © The Author(s) 2018 Downloaded from https://academic.oup.com/aobpla/article-abstract/10/2/ply012/4857208 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Castro et al. – Cytogeographical patterns in a tetraploid–octoploid complex Figure 4. Fine-scale distribution of Gladiolus communis individuals within three mixed-ploidy populations: (A) tetraploid and octoploid mixed- ploidy population (MC201), (B) tetraploid and hexaploid mixed-ploidy population (MC232); and (C) tetraploid, hexaploid, octoploid and nona- ploid mixed-ploidy population (MC193). Each point represents one individual plant mapped in a x/y system where distance is given in metres (m): tetraploids (4x), hexaploids (6x), octoploids (8x) and nonaploids (9x) individuals are represent by white, grey, black and dark grey points, respectively. dominant cytotypes, tetraploids and octoploids, were occurred throughout the sampling area, although they observed along with two minority cytotypes, mostly were more common in the north and central regions hexaploids, and rarely nonaploids. Tetraploids and of Portugal. Octoploids occurred in south and central octoploids have been well documented on the Iberian regions of Portugal, but not in the north, notwithstand- Peninsula through chromosome counts (Fernandes ing the fact that more extensive surveys are needed to et  al. 1948; Fernandes 1950; Fernandes and Queirós confirm this pattern. Although several mixed-ploidy pop - 1971; Nilsson and Lassen 1971; Löve and Kjellqvist 1973; ulations were found, the geographical isolation index Queirós 1980; Fernández et  al. 1985). Also, hexaploids between tetraploids and octoploids is high, reflecting have been previously reported in the Mediterranean the fact that most of the populations contain a single basin (Darlington and Wylie 1955). We observed them cytotype. These populations distribute in space allopat- rically or parapatrically, forming several contact zones in 11  % of the sampled localities (12 of 105 localities), commonly growing with one of the dominant cytotypes between tetraploids and octoploids. However, despite and occasionally in pure-ploidy populations. Nonaploids that tetraploids and octoploids have non-overlapping are reported here for the first time and were detected in distributions, they can inhabit similar environmental the most diverse mixed-ploidy population. niches. Niche identity and similarity tests showed that Despite the cytogenetic diversity reported in G.  com- tetraploids and octoploids occupy similar niches and are munis, almost nothing was known about the geographic not differentiated in their environmental niches, show- distribution of the cytotypes, or the presence and struc- ing niche conservation. These results contrast with other ture of its contact zones. Based on our survey, tetraploids polyploid complexes for which niche differentiation, driven either by the direct effects of polyploidy or by Table  3. Niche analyses in Gladiolus communis. For each region subsequent selection, underlies the spatial separation of studied, equivalency (D and P values) and similarity (P value) tests cytotypes and allows them to escape the minority cyto- for suitable habitat are given. type disadvantage (e.g. Glennon et al. 2012; Thompson Suitable Equivalence test Similarity test (P values) et al. 2014; Visger et al. 2016; Muñoz-Pajares et al. 2017). habitat Still, the absence of environmental niche differences D value P value Tetra -> Octo Octo -> Tetra might not be completely unexpected as polyploids might not differ from their lower ploidy ancestors, either Contact 0.034 0.960 0.406 0.337 zone because they have been formed recently and the new polyploids did not have time to diverge from their pro- Portugal 0.009 0.515 0.535 0.515 genitors, because genome duplications did not generate AoB PLANTS https://academic.oup.com/aobpla © The Author(s) 2018 11 Downloaded from https://academic.oup.com/aobpla/article-abstract/10/2/ply012/4857208 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Castro et al. – Cytogeographical patterns in a tetraploid–octoploid complex Figure 5. Results of ecological niche models for Gladiolus communis polyploid complex at (A, B) the contact zone in Central Portugal, and (C, D) Portugal. (A) and (C) represent the contribution of climatic and soil variables in the first two axes of the principal component analyses (PCA) and the percentage of variance explained by each axis. (B) and (C) represent the environmental niche of each cytotypes based on the PCA of selected variables; coloured areas represent suitable habitats as follows: light grey—tetraploids; dark grey—octoploids; and green—overlap- ping areas between tetraploids and octoploids; the continuous line corresponds to the whole climatic space, while the dashed line indicates the 75th percentile. significant direct physiological changes, and/or because in several polyploid complexes (e.g. Baack 2004, 2005; they might have been subjected to recurrent gene flow Pannell et  al. 2004; Baack and Stanton 2005; Godsoe (Godsoe et al. 2013; Laport et al. 2016). Also, the effect et  al. 2013; Münzbergová et  al. 2013; Wefferling et  al. of other environmental parameters on the distribution 2017). Contact zones are generated by direct emer- patterns observed in G. communis cannot be completely gence of neopolyploids in lower ploidy parental pop- ruled out, nor the fact that niche differentiation might ulations or through secondary contact of previously occur at a special resolution higher than that used in our allopatric distributions in which cytotypes colonized study, although we did not find any clear evidence of dif - the area separately in dissimilar ways and at differ- ferentiation in the field, namely considering the type of ent timings (Petit et al. 1999; Lexer and van Loo 2006). vegetation or the type of substrate in the mixed-ploidy Although we still do not know the origin of G. commu- populations detected (M. Castro, field observations). nis contact zones, the different cytotype compositions Considering that G. communis cytotypes do not differ found in natural populations provide significant insights in suitable habitat, there should be historical processes into the processes that might be occurring at these and other ecological determinants shaping their distri- areas (e.g. Husband and Schemske 1998; reviewed in butional patterns, similarly to what has been observed Husband et al. 2013; Suda et al. 2013). One of the main 12 AoB PLANTS https://academic.oup.com/aobpla © The Author(s) 2018 Downloaded from https://academic.oup.com/aobpla/article-abstract/10/2/ply012/4857208 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Castro et al. – Cytogeographical patterns in a tetraploid–octoploid complex Figure  6. Reproductive fitness of natural pure- and mixed-ploidy populations of Gladiolus communis: (A) fruit set; (B) S:O ratio (number of viable seeds divided by the number of ovules); and (C) reproductive success (fruit set multiplied by S:O ratio). In (C) the proportion of DNA ploidy levels detected in the offspring is also given. DNA ploidy levels: tetraploid (4x), pentaploid (5x), hexaploid (6x), octoploid (8x); seeds with genome size values out of the range of variation of each ploidy levels were assumed as aneuploids (An.). Different letters correspond to statistically significant differences as follows: (i) differences between population type (pure- vs. mixed-ploidy populations, excluding 6x) are denoted by upper case letters; and (ii) differences between ploidy levels within population type (among 4x, 6x and 8x from pure populations, and between 4x and 8x from the mixed-ploidy population) are denoted by lower case letters (Tukey HSD; P < 0.05); n.s. correspond to non- significant differences (P > 0.05). observations is the fairly few mixed-ploidy populations isolation observed between G.  communis cytotypes (10 vs. 90  % of mixed- and pure-ploidy populations), suggests that the mixed-ploidy populations might be all composed of unbalanced number of tetraploid and transitory because strong frequency-dependent selec- octoploid plants (either dominated by tetraploid or by tion is expected to eliminate the minority cytotype as octoploids). In the absence of environmental differ- a result of fitness disadvantage generated by its lower ences, and regardless of the origin of the contact zone, number. This selection will ultimately drive the occur- G.  communis mixed tetraploid–octoploid populations rence of pure-ploidy populations at contact zones are expected to be more common at contact areas (Levin 1975; Husband 2000). than detected here (4.9  % in the contact zone and However, tetraploid–octoploid populations may per- 6.5  % from the total), since cytotypes might disperse sist in nature. The regular production of unreduced to areas of the other cytotype and/or new cytotypes gametes and the presence of reproductive barriers pro- might be formed. Consequently, the high geographical moting assortative mating might lessen the magnitude AoB PLANTS https://academic.oup.com/aobpla © The Author(s) 2018 13 Downloaded from https://academic.oup.com/aobpla/article-abstract/10/2/ply012/4857208 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Castro et al. – Cytogeographical patterns in a tetraploid–octoploid complex of frequency-dependent selection and enable cytotype unreduced gamete formation is an important path- coexistence (e.g. Felber 1991; Segraves and Thompson way for new polyploid emergence and has been shown 1999; Husband 2004; Husband and Sabara 2004; to be common in nature (Felber 1991; Bretagnolle and Kennedy et al. 2006). Octoploids might emerge directly Thompson 1995; Ramsey and Schemske 1998; Husband in tetraploid populations through the union of two unre- 2004; Ramsey 2007). This might explain the detection duced gametes (n = 4x) or might result from seed disper- of hexaploid plants frequently found in otherwise tetra- sal from neighbouring octoploid populations. Unreduced ploid populations. Alternatively, hexaploids may form gamete production has been detected in controlled pol- as a result of hybridization events between tetraploid linations in tetraploid G.  communis (M. Castro et  al., in and octoploid G.  communis individuals. Gladiolus com- preparation) and in screenings in natural populations munis is pollinated by generalist pollinators that seem through the detection of hexaploid individuals (see to have no cytotype preferences and might move pol- below). The rates at which unreduced gametes are pro- len within mixed-ploidy populations or between popula- duced might feed the population of octoploids enabling tions in close proximity (M. Castro et al., in preparation). their maintenance within tetraploid populations (Felber Additionally, controlled pollinations between tetraploid 1991; Husband 2004). Additionally, seed ploidy analyses and octoploid plants were also successful in produc- in a tetraploid–octoploid population suggest that strong ing hexaploid offspring (M. Castro et al., in preparation). reproductive barriers may enforce assortative mating, Either one of these pathways, i.e. unreduced gamete for- further favouring cytotype coexistence. Reproductive mation or hybridization, may operate in natural popula- barriers driven, for example, by phenological and/or mor- tions, being difficult to distinguish them without genetic phological mismatch, different pollinator assemblages markers. However, the relative abundance of tetraploid– or preferences, and/or gametic isolation will, thus, play a hexaploid populations and paucity of tetraploid–hexa- major role for overcoming minority cytotype exclusion in ploid–octoploid populations suggests that the majority mixed-ploidy populations. Therefore, the fate of octop- of the hexaploids are formed through unreduced gam- loids might depend not only on the rates of unreduced etes in tetraploid populations. Additionally, unreduced gamete formation but also on the reproductive isolation gamete production has been frequently detected in con- levels. Additionally, differences in other traits, such as trolled pollinations involving tetraploid G. communis (M. perenniality or asexual reproduction, could compensate Castro et al., in preparation), supporting it as a probable for the minority cytotype disadvantage (e.g. Rodriguez pathway for new cytotype emergence in natural popula- 1996; Kao 2007; Castro et al. 2016a). In other polyploid tions. Quantifying unreduced gamete production in nat- complexes, traits such as the production of bulbs repre- ural populations will provide significant insights on how sented an advantage, enabling new cytotypes to persist frequent this process could be involved with hexaploid at initial stages and spread within lower ploidy popula- emergence. tions (e.g. Allium oleraceum; Duchoslav et al. 2010; G. × Interestingly, hexaploid individuals were also found sulistrovicus; Szczepaniak et  al. 2016). If, through some forming pure-ploidy populations, showing that this of these traits, the number of octoploids can surpass the cytotype can successfully establish and spread beyond number of tetraploids, at some time octoploids might parental populations, although their sexual repro- even outcompete tetraploids and exclude them from ductive fitness was revealed to be lower in compari - the population, as observed in other polyploid com- son with tetraploids and octoploids. Regardless of the plexes (e.g. Buggs and Pannell 2007). Indeed, octop- lower fitness, recurrent unreduced gamete formation loids were observed as the dominant cytotype in some and asexual reproduction might enable to compensate mixed-ploidy populations of the contact zone. Future for this disadvantage (e.g. Husband 2004; Kao 2007; studies on the contribution of all the above-mentioned Castro et  al. 2016a). The successful establishment of processes, and on the relative contribution of sexual vs. hexaploid plants further contributes to the diversifica - asexual reproduction for the maintenance of the popu- tion of the complex. Ultimately, contact zones result lations, are needed to fully understand the dynamics of from the combination of several factors, including mixed-ploidy populations. historical factors, unreduced gamete formation, pol- The cytotype composition of G.  communis natural len flow and hybridization events, and seed dispersal, populations also revealed that hexaploid plants might among others (Petit et al. 1999; Levin 2002; Lexer and be more common than previously thought. These hexa- van Loo 2006). Future studies reconstructing the his- ploids might have originated through two different tory of the complex and quantifying unreduced gamete pathways. Hexaploids may originate from tetraploids production, and its ability to hybridize, would provide through the union of reduced (n = 2x) and unreduced significant insights on the dynamics of the distribution (n = 4x) gametes (Ramsey and Schemske 1998). Indeed, of G. communis. 14 AoB PLANTS https://academic.oup.com/aobpla © The Author(s) 2018 Downloaded from https://academic.oup.com/aobpla/article-abstract/10/2/ply012/4857208 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Castro et al. – Cytogeographical patterns in a tetraploid–octoploid complex The genome size of G. italicus suggests that this spe- mixed-ploidy populations were expected to be fre- cies is duodecaploid in the studied area, which is in quent; however, a high geographical isolation index was accordance with chromosome counts for the Iberian obtained. The high geographical isolation observed in Peninsula (Queirós 1979; Pérez and Pastor Díaz 1994), nature, along with habitat similarity, suggests that the and contrasts with the dominance of the octoploids cytotype distribution in G.  communis reflects histor - elsewhere in the Mediterranean basin (Susnik and Lovka ical patterns of migration and colonization, and further 1973; Strid and Franzen 1981; van Raamsdonk and de selection against minority cytotype, and does not result Vries 1989; Kamari et  al. 2001). Interestingly, the vari- from different environmental requirements, creating a ation in monoploid genome size within G.  communis tension zone of contact. Still, in areas of contact, repro- cytotypes was very low and differed significantly from ductive barriers might mediate assortative mating and that of G. italicus (~9 %). Given the magnitude of the dif- enable cytotype coexistence. Nevertheless, the high ferences between G.  italicus and G.  communis, both in cytogenetic diversity detected in the field suggests that ploidy levels and in monoploid genome sizes, holoploid unreduced gamete formation and hybridization events genome size might be an important tool to detect hybrid- seem frequent in this complex and might be involved ization (e.g. Kolář et al. 2009; Agudo-García 2017). In our with recurrent polyploid formation and with gene flow study, G.  italicus and G.  communis were found growing between cytogenetic entities. Future studies involving in sympatry in 13 % of localities; however, all the G. itali- reciprocal transplants will provide significant insights cus individuals were duodecaploid. In most of the cases, into the dynamics of this polyploid complex. the duodecaploid G.  italicus was found growing with the octoploid G. communis (12 out of 14 localities); still, Sources of Funding no decaploids were observed in these localities. When growing with the tetraploid G.  communis, no octoploid This research was supported by POPH/FSE funds by the individuals with lower genome size resulting from the Portuguese Foundation for Science and Technology (FCT) hybridization between the two species (~5.00 pg based with a doctoral grant to M.C. (SFRH/BD/89910/2012) on the monoploid genome sizes of each species) were and a starting grant and exploratory project to S.C. observed. Although hybridization has been suggested to (IF/01267/2013), and by Project RENATURE financed by occur in these and in other Gladiolus species (e.g. van the ‘Programa Operacional Regional do Centro 2014–2020 Raamsdonk and de Vries 1989; Mifsud and Hamilton (Centro2020) - CENTRO-01-0145-FEDER-000007’. 2013; Szczepaniak et  al. 2016), we were not able to detect hybrids between G.  italicus and G.  communis. Contributions by the Authors This suggests that, in the studied range, hybridization S.C., B.H.  and J.L.  designed the research experiment; between them might be less common, either because of assortative mating or hybrid offspring inviability. M.C., S.C.  and J.L.  developed the field and laboratory Monoploid genome size also suggests a close relation- work; M.C.  performed the statistical analyses; and ship between the cytotypes of G. communis, pointing to M.C.  and A.F.  performed the niche modelling analyses; an autopolyploid origin of the complex in the studied all the authors were involved in the discussion of the area. This is also supported by the high morphological results and writing of the manuscript, approving the resemblance between G.  communis cytotypes (Alonso final document. and Crespo 2010; Cantor and Tolety 2011) and by the lack of evidence supporting hybridization between Conflict of Interest G. communis and G. italicus in this region. Still, the origin of G. communis polyploid complex needs to be properly None declared. evaluated in future studies. Acknowledgements Conclusions The authors are thankful to several researchers that have assisted in field collections, in particular to J. Costa, In this study, we find a complex cytogeographical pat - tern in G.  communis, which opens several hypotheses L.  Mota, D.  Tavares, A.  Martins, J.  M. Costa, C.  Silva and that might explain the formation and maintenance of M.  C. Duarte. We are also thankful to A.  Caperta and L. Morais for all the helpful methodological suggestions its tetraploid–octoploid contact zone. According to our results, tetraploids and octoploids do not differ in their on chromosome count techniques, and to two anony- environmental requirements, potentially growing in simi- mous reviewers for their insightful comments on previ- ous versions of the manuscript. lar habitats. Without differences in habitat requirements, AoB PLANTS https://academic.oup.com/aobpla © The Author(s) 2018 15 Downloaded from https://academic.oup.com/aobpla/article-abstract/10/2/ply012/4857208 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Castro et al. – Cytogeographical patterns in a tetraploid–octoploid complex Buggs RJ, Pannell JR. 2007. Ecological differentiation and diploid Supporting Information superiority across a moving ploidy contact zone. Evolution The following additional information is available in the 61:125–140. online version of this article— Burton TL, Husband BC. 2001. Fecundity and offspring ploidy in mat- ings among diploid, triploid and tetraploid Chamerion angustifo- Figure S1. Flow cytometry graphics analysed for each lium (Onagraceae): consequences for tetraploid establishment. sample. Heredity 87:573–582. Table S1. Geographic information of sampled Gladiolus Cantor M, Tolety J. 2011. Gladiolus In: Kole C. eds. Wild crop relatives: populations. genomic and breeding resources - Plantation and ornamental Table S2. Genome size variation in Gladiolus communis. crops. Berlin: Springer-Verlag Berlin Heidelberg, 133–160. Table S3. Mixed-ploidy populations of Gladiolus Castro S, Castro M, Ferrero V, Costa J, Tavares D, Navarro L, Loureiro communis. J. 2016a. Invasion fosters change: independent evolutionary Table S4. DNA ploidy levels of the offspring of pure- and shifts in reproductive traits after Oxalis pes-caprae L. introduc- mixed-ploidy populations of Gladiolus communis. tion. Frontiers in Plant Science 7:874. Castro M, Loureiro J, Husband BC, Castro S. 2016b. Can we live together? The role of reproductive barriers mediating cytotype Literature Cited coexistence in Gladiolus communis contact zones. XII Encontro Agudo-García A. 2017. Evolution in anacyclus L.  (Anthemideae, Nacional de Biologia Evolutiva, 16 December, University of Aveiro. Asteraceae). Analysis of the contact zone between A.  clava- Castro S, Loureiro J, Procházka T, Münzbergová Z. 2012. Cytotype tus (Desf.) Pers. and A.  valentinus L. PhD Thesis, Universidad distribution at a diploid-hexaploid contact zone in Aster amellus Autonoma de Madrid, Spain. (Asteraceae). Annals of Botany 110:1047–1055. Alonso MÁ, Crespo MB. 2010. Gladiolus L. In: Crespo MB, Herrero Castro S, Münzbergová Z, Raabová J, Loureiro J. 2011. Breeding A, Quintanar A, eds. Flora iberica. Madrid: Real Jardín Botánico, barriers at a diploid–hexaploid contact zone in Aster amellus. CSIC, 185:485–491. Evolutionary Ecology 25:795–814. Araújo MB, New M. 2007. Ensemble forecasting of species distribu- Darlington CD, Wylie AP. 1955. Chromosome atlas of flowering tions. Trends in Ecology & Evolution 22:42–47. plants. London, UK: George Allen and Unwin Ltd. Baack EJ. 2004. Cytotype segregation on regional and micro- Doležel J, Greilhuber J, Lucretti S, Meister A, Lysák MA, Nardi L, geographic scales in snow buttercups (Ranunculus adoneus: Obermayer R. 1998. Plant genome size estimation by flow Ranunculaceae). American Journal of Botany 91:1783–1788. cytometry: inter-laboratory comparison. Annals of Botany Baack EJ. 2005. To succeed globally, disperse locally: effects of local 82(suppl_1):17–26. pollen and seed dispersal on tetraploid establishment. Heredity Doležel J, Sgorbati S, Lucretti S. 1992. Comparison of three DNA 94:538–546. fluorochromes for flow cytometric estimation of nuclear DNA Baack EJ, Stanton ML. 2005. Ecological factors influencing tetra - content in plants. Physiologia Plantarum 85:625–631. ploid speciation in snow buttercups (Ranunculus adoneus): Duchoslav M, Safárová L, Krahulec F. 2010. Complex distribution niche differentiation and tetraploid establishment. Evolution patterns, ecology and coexistence of ploidy levels of Allium 59:1936–1944. oleraceum (Alliaceae) in the Czech Republic. Annals of Botany Bates D, Maechler M, Bolker B, Walker S. 2014. lme4: linear mixed- 105:719–735. effects models using Eigen and S4. http://CRAN.R-project.org/ Felber F. 1991. Establishment of a tetraploid cytotype in a diploid package=lme4 (21 April 2016). population: effect of relative fitness of the cytotypes. Journal of Bolker BM, Brooks ME, Clark CJ, Geange SW, Poulsen JR, Stevens MH, Evolutionary Biology 4:195–207. White JS. 2009. Generalized linear mixed models: a practical Fernandes A. 1950. Sobre a cariologia de algumas plantas da Serra guide for ecology and evolution. Trends in Ecology & Evolution do Gerês. Agronomia Lusitana 12:551–600. 24:127–135. Fernandes A, Garcia J, Fernandes R. 1948. Herborizações nos Bretagnolle FA, Thompson JD. 1995. Gametes with the somatic domínios da fundação da Casa de Bragança. I-Vendas Novas. chromosome number: mechanisms of their formation and Memórias da Sociedade Broteriana 4:5–89. role in the evolution of autopolyploid plants. New Phytologist Fernandes A, Queirós M. 1971. Sur la caryologie de quelques plan- ème 129:1–22. tes récoltéess pendant la III réunion de botanique péninsu- Brochmann C, Brysting AK, Alsos IG, Borgen L, Grundt HH, Scheen lare. Memórias da Sociedade Broteriana 21:343–385. AC, Elven R. 2004. Polyploidy in arctic plants. Biological Journal Fernández I, Díez MJ, Pastor J. 1985. Números cromosómicos para of the Linnean Society 82:521–536. la flora española, 373–381. Lagascalia 13:299–302. Broennimann O, Fitzpatrick MC, Pearman PB, Petitpierre B, Pellissier Fox J, Weisberg S, Adler D, Bates D, Baud-Bovy G, Ellison S. 2015. L, Yoccoz NG, Thuiller W, Fortin MJ, Randin C, Zimmermann NE, car: companion to applied regression. http://CRAN.R-project. Graham CH. 2012. Measuring ecological niche overlap from org/package=car. (21 April 2016). occurrence and spatial environmental data. Global Ecology and Galbraith DW, Harkins KR, Maddox JM, Ayres NM, Sharma DP, Biogeography 21:481–497. Firoozabady E. 1983. Rapid flow cytometric analysis of the cell Buchanan A. 2008. The taxonomic status of Gladiolus illyricus cycle in intact plant tissues. Science 220:1049–1051. (Iridaceae) in Britain. Doctoral dissertation, Department of Life Glennon KL, Rissler LJ, Church SA. 2012. Ecogeographic isolation: a Sciences and the Natural History Museum, Imperial College reproductive barrier between species and between cytotypes in London, London. Houstonia (Rubiaceae). Evolutionary Ecology 26:909–926. 16 AoB PLANTS https://academic.oup.com/aobpla © The Author(s) 2018 Downloaded from https://academic.oup.com/aobpla/article-abstract/10/2/ply012/4857208 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Castro et al. – Cytogeographical patterns in a tetraploid–octoploid complex Godsoe W, Larson MA, Glennon KL, Segraves KA. 2013. Jiao Y, Wickett NJ, Ayyampalayam S, Chanderbali AS, Landherr L, Polyploidization in Heuchera cylindrica (Saxifragaceae) did not Ralph PE, Tomsho LP, Hu Y, Liang H, Soltis PS, Soltis DE, Clifton result in a shift in climatic requirements. American Journal of SW, Schlarbaum SE, Schuster SC, Ma H, Leebens-Mack J, Botany 100:496–508. dePamphilis CW. 2011. Ancestral polyploidy in seed plants and Goldblatt P, Takei M. 1993. Chromosome cytology of the African angiosperms. Nature 473:97–100. genus Lapeirousia (Iridaceae-Ixioideae). Annals of the Missouri Kao RH. 2007. Asexuality and the coexistence of cytotypes. The Botanical Garden 80:961–973. New Phytologist 175:764–772. Greilhuber J, Dolezel J, Lysák MA, Bennett MD. 2005. The origin, Kamari G, Blanche C, Garbari F. 2001. Mediterranean chromosome evolution and proposed stabilization of the terms ‘genome number reports: 11. Flora Mediterranea 11:435–483. size’ and ‘C-value’ to describe nuclear DNA contents. Annals of Kennedy BF, Sabara HA, Haydon D, Husband BC. 2006. Pollinator- Botany 95:255–260. mediated assortative mating in mixed ploidy populations Greilhuber J, Ebert I. 1994. Genome size variation in Pisum sativum. of Chamerion angustifolium (Onagraceae). Oecologia 150: Genome 37:646–655. 398–408. Greilhuber J, Temsch EM, Loureiro J. 2007. Nuclear DNA content Kolář F, Stech M, Trávnícek P, Rauchová J, Urfus T, Vít P, Kubesová measurement. In: Dolezel J, Greilhuber J, Suda J, eds. Flow M, Suda J. 2009. Towards resolving the Knautia arvensis agg. cytometry with plant cells: analysis of genes, chromosomes and (Dipsacaceae) puzzle: primary and secondary contact zones genomes. Weinheim: Wiley-VCH, 67–101. and ploidy segregation at landscape and microgeographic Gussone G. 1832. Supplementum ad Florae Siculae Prodromum, vol. scales. Annals of Botany 103:963–974. 1. Regia Typographia, Neapoli. Kron P, Suda J, Husband BC. 2007. Applications of flow cytometry to Hamilton AP. 1980. Gladiolus L. Flora Europaea 5:101–102. evolutionary and population biology. Annual Review of Ecology, Hijmans RJ, van Etten J, Cheng J, Mattiuzzi M, Sumner M, Evolution, and Systematics 38:847–876. Greenberg JA, Lamigueiro OP, Bevan A, Racine EB, Shortridge Laport RG, Minckley RL, Ramsey J. 2016. Ecological distributions, A, Ghosh A, 2017. Package ‘raster’. https://cran.r-project.org/ phenological isolation, and genetic structure in sympatric and package=raster. (21 May 2017). parapatric populations of the Larrea tridentata polyploid com- Hijmans RJ, Phillips S, Leathwick J, Elith J. 2015. dismo: species distribu- plex. American Journal of Botany 103:1358–1374. tion modeling. R package version 1.0-12. Vienna: The R Foundation Levin DA. 1975. Minority cytotype exclusion in local plant popula- for Statistical Computing. http://cran.r-project.org. (21 May 2017). tions. Taxon 24:35–43. Hothorn T, Bretz F, Westfall P, Heiberger RM. 2017. Multcomp: sim- Levin DA. 2002. The role of chromosomal change in plant evolution. ultaneous inference for general linear hypotheses. http://CRAN. Oxford: Oxford University Press. Rproject.org/package=multcomp. (21 April 2016). Lexer C, van Loo M. 2006. Contact zones: natural labs for studying Husband BC. 2000. Constraints on polyploid evolution: a test of the evolutionary transitions. Current Biology 16:R407–R409. minority cytotype exclusion principle. Proceedings of the Royal Loureiro J, Rodriguez E, Dolezel J, Santos C. 2007. Two new nuclear Society of London B: Biological Sciences 267:217–223. isolation buffers for plant DNA flow cytometry: a test with 37 Husband BC. 2004. The role of triploid hybrids in the evolutionary species. Annals of Botany 100:875–888. dynamics of mixed-ploidy populations. Biological Journal of the Löve A, Kjellqvist E. 1973. Cytotaxonomy of Spanish plants. II. Linnean Society 82:537–546. Monocotyledons. Lagascalia 7:147–182. Husband BC, Baldwin SJ, Sabara HA. 2016. Direct vs. indirect Maceira NO, Haan AD, Lumaret R, Billon M, Delay J. 1992. Production effects of whole-genome duplication on prezygotic isolation of 2n gametes in diploid subspecies of Dactylis glomerata in Chamerion angustifolium: implications for rapid speciation. L.  1. Occurrence and frequency of 2n pollen. Annals of Botany American Journal of Botany 103:1259–1271. 69:335–343. Husband BC, Baldwin SJ, Suda J. 2013. The incidence of polyploidy Mallet J, Meyer A, Nosil P, Feder JL. 2009. Space, sympatry and spe- in natural plant populations: major patterns and evolutionary ciation. Journal of Evolutionary Biology 22:2332–2341. processes. In: J. Greilhuber, J. Dolezel, and J. F. Wendel eds. Marques I, Loureiro J, Draper D, Castro M, Castro S. 2017. How much Plant genome diversity. Vienna: Springer Vienna, 255–276. do we know about the frequency of hybridization and polyploidy in Husband BC, Sabara HA. 2004. Reproductive isolation between the mediterranean region? Plant Biology. doi:10.1111/plb.12639. autotetraploids and their diploid progenitors in fireweed, Mifsud S, Hamilton AP. 2013. Preliminary observations from long- Chamerion angustifolium (Onagraceae): research review. New term studies of Gladiolus L. (Iridaceae) for the Maltese Islands. Phytologist 161:703–713. Webbia 68:51–56. Husband BC, Schemske DW. 1998. Cytotype distribution at a diploid- Monteiro-Henriques T, Martins MJ, Cerdeira JO, Silva PC, Arsénio P, tetraploid contact zone in Chamerion (Epilobium) angustifolium Silva Á, Bellu A, Costa JC. 2016. Bioclimatological mapping tack- (Onagraceae). American Journal of Botany 85:1688–1694. ling uncertainty propagation: application to mainland Portugal. Husband BC, Schemske DW. 2000. Ecological mechanisms of repro- International Journal of Climatology 36:400–411. ductive isolation between diploid and tetraploid Chamerion Muñoz‐Pajares AJ, Perfectti F, Loureiro J, Abdelaziz M, Biella P, Castro angustifolium. Journal of Ecology 88:689–701. M, Castro S, Gómez JM. 2017. Niche differences may explain the Jersáková J, Castro S, Sonk N, Milchreit K, Schödelbauerová I, geographic distribution of cytotypes in Erysimum mediohispani- Tolasch T, Dötterl S. 2010. Absence of pollinator-mediated pre- cum. Plant Biology 20(Suppl. 1):139–147. mating barriers in mixed-ploidy populations of Gymnadenia Münzbergová Z, Surinová M, Castro S. 2013. Absence of gene flow conopsea sl (Orchidaceae). Evolutionary Ecology 24: between diploids and hexaploids of Aster amellus at multiple 1199–1218. spatial scales. Heredity 110:123–130. AoB PLANTS https://academic.oup.com/aobpla © The Author(s) 2018 17 Downloaded from https://academic.oup.com/aobpla/article-abstract/10/2/ply012/4857208 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Castro et al. – Cytogeographical patterns in a tetraploid–octoploid complex Nilsson O, Lassen P. 1971. Chromosome numbers of vascular Dactylorhiza maculata s.l. (Orchidaceae). Evolutionary Ecology plants from Austria, Mallorca and Yugoslavia. Botaniska Notiser 23:295. 124:270–276. Strid A, Franzen R. 1981. In Chromosome number reports LXXIII. Otto SP, Whitton J. 2000. Polyploid incidence and evolution. Annual Taxon 30:829–842. Review of Genetics 34:401–437. Suda J, Kron P, Husband BC, Trávníček P. 2007. Flow cytometry Pannell JR, Obbard DJ, Buggs RJ. 2004. Polyploidy and the sexual and ploidy: applications in plant systematics, ecology and evo- system: what can we learn from Mercurialis annua? Biological lutionary biology. In: Doležel J, Greilhuber J, Suda J, eds. Flow Journal of the Linnean Society 82:547–560. cytometry with plant cells: analysis of genes, chromosomes and Pérez E, Pastor Díaz JE. 1994. Contribución al estudio cariológico genomes. Weinheim: Wiley-VCH, 103–130. de la familia Iridaceae en Andalucía Occidental. Lagascalia Suda J, Oberlander KC, Dreyer LL. 2013. Two new species of Oxalis 17:257–272. (Oxalidaceae) from the Greater Cape Floristic Region. Phytotaxa Petit C, Bretagnolle F, Felber F. 1999. Evolutionary consequences of 124:13–24. diploid-polyploid hybrid zones in wild species. Trends in Ecology Susnik F, Lovka EH. 1973. Reports. IOPB Chromosome numbers & Evolution 14:306–311. reports XLI. Taxon 22:462–463. Phillips SJ. 2008. Transferability, sample selection bias, and back- Szczepaniak M, Kamiński R, Kuta E, Słomka A, Heise W, Cieślak E. ground data in presence-only modelling: a response to Peterson 2016. Natural hybridization between Gladiolus palustris and et al. (2007). Ecography 31:272–278. G. imbricatus inferred from morphological, molecular and repro- Phillips SJ, Anderson RP, Schapire RE. 2006. Maximum entropy mod- ductive evidence. Preslia 88:137–161. eling of species geographic distributions. Ecological Modelling Thompson KA, Husband BC, Maherali H. 2014. Climatic niche differ- 190:231–259. ences between diploid and tetraploid cytotypes of Chamerion Queirós M. 1979. Números cromossómicos para a flora Portuguesa: angustifolium (Onagraceae). American Journal of Botany 16–37. Boletim da Sociedade Broteriana Série 2 53:15–28. 101:1868–1875. Queirós M. 1980. Números cromossómicos para a flora portuguesa: Thuiller W, Georges D, Engler R, Breiner F, Georges MD, Thuiller 38–63. Boletim da Sociedade Broteriana Série 2 54:47–64. CW. 2016. Package ‘biomod2’. https://cran.r-project.org/ R Core Development Team. 2016. A language and environment package=biomod2. for statistical computing. Vienna: R Foundation for Statistical Trávníček P, Kubátová B, Curn V, Rauchová J, Krajníková E, Jersáková Computing. http://www.R-project.org/. (21 April 2016). J, Suda J. 2010. Remarkable coexistence of multiple cytotypes Ramsey J. 2007. Unreduced gametes and neopolyploids in natural of the Gymnadenia conopsea aggregate (the fragrant orchid): populations of Achillea borealis (Asteraceae). Heredity 98:143–150. evidence from flow cytometry. Annals of Botany 107:77–87. Ramsey J, Ramsey TS. 2014. Ecological studies of polyploidy in the Van Raamsdonk LWD, De Vries T. 1989. Biosystematic studies in 100 years following its discovery. Philosophical Transactions of the European species of Gladiolus (Iridaceae). Plant Systematics and Royal Society B: Biological Sciences. doi:10.1098/rstb.2013.0352. Evolution 165:189–198. Ramsey J, Schemske DW. 1998. Pathways, mechanisms, and rates Visger CJ, Germain-Aubrey CC, Patel M, Sessa EB, Soltis PS, Soltis of polyploid formation in flowering plants. Annual Review of DE. 2016. Niche divergence between diploid and autotetraploid Ecology and Systematics 29:467–501. Tolmiea. American Journal of Botany 103:1396–1406. Rodriguez DJ. 1996. A model for the establishment of polyploidy in Warren DL, Glor RE, Turelli M. 2008. Environmental niche equiva- plants. The American Naturalist 147:33–46. lency versus conservatism: quantitative approaches to niche Schoener TW. 1970. Nonsynchronous spatial overlap of lizards in evolution. Evolution 62:2868–2883. patchy habitats. Ecology 51:408–418. Wefferling KM, Castro S, Loureiro J, Castro M, Tavares D, Hoot SB. Segraves KA, Thompson JN. 1999. Plant polyploidy and pollination: 2017. Cytogeography of the subalpine marsh marigold polyploid floral traits and insect visits to diploid and tetraploid Heuchera complex (Caltha leptosepala s.l., Ranunculaceae). American grossulariifolia. Evolution 53:1114–1127. Journal of Botany 104:271–285. Soltis DE, Albert VA, Leebens-Mack J, Bell CD, Paterson AH, Zheng C, Wood TE, Takebayashi N, Barker MS, Mayrose I, Greenspoon PB, Sankoff D, Depamphilis CW, Wall PK, Soltis PS. 2009. Polyploidy and Rieseberg LH. 2009. The frequency of polyploid speciation angiosperm diversification. American Journal of Botany 96:336–348. in vascular plants. Proceedings of the National Academy of Soltis DE, Buggs RJ, Doyle JJ, Soltis PS. 2010. What we still don’t Sciences 106:13875–13879. know about polyploidy. Taxon 59:1387–1403. Zozomová-Lihová J, Malánová-Krásná I, Vít P, Urfus T, Senko D, Soltis DE, Soltis PS. 1999. Polyploidy: recurrent formation and gen- Svitok M, Kempa M, Marhold K. 2015. Cytotype distribution ome evolution. Trends in Ecology & Evolution 14:348–352. patterns, ecological differentiation, and genetic structure in a Ståhlberg D. 2009. Habitat differentiation, hybridization and gene diploid-tetraploid contact zone of Cardamine amara. American flow patterns in mixed populations of diploid and autotetraploid Journal of Botany 102:1380–1395. 18 AoB PLANTS https://academic.oup.com/aobpla © The Author(s) 2018 Downloaded from https://academic.oup.com/aobpla/article-abstract/10/2/ply012/4857208 by Ed 'DeepDyve' Gillespie user on 16 March 2018 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png AoB Plants Oxford University Press

Complex cytogeographical patterns reveal a dynamic tetraploid–octoploid contact zone

Free
18 pages

Loading next page...
 
/lp/ou_press/complex-cytogeographical-patterns-reveal-a-dynamic-tetraploid-O6Vv3V00PL
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of the Annals of Botany Company.
eISSN
2041-2851
D.O.I.
10.1093/aobpla/ply012
Publisher site
See Article on Publisher Site

Abstract

Research Article Complex cytogeographical patterns reveal a dynamic tetraploid–octoploid contact zone 1 1,2 3 4 1 Mariana Castro *, Sílvia Castro , Albano Figueiredo , Brian Husband and João Loureiro CFE, Centre for Functional Ecology, Department of Life Sciences, University of Coimbra, Calçada Martim de Freitas, 3000-456 Coimbra, Portugal Botanic Garden of the University of Coimbra, Calçada Martim de Freitas, 3000-456 Coimbra, Portugal CEGOT, Departamento de Geografia e Turismo, Faculdade de Letras, Universidade de Coimbra, Largo da Porta Férrea, 3004-530 Coimbra, Portugal Department of Integrative Biology, University of Guelph, Guelph, ON N1G 2W1, Canada Received: 13 October 2017 Editorial decision: 27 January 2018 Accepted: 13 February 2018 Published: 14 February 2018 Associate Editor: Adrian C. Brennan Citation: Castro M, Castro S, Figueiredo A, Husband B, Loureiro J. 2018. Complex cytogeographical patterns reveal a dynamic tetraploid–octoploid contact zone. AoB PLANTS 10: ply012; doi: 10.1093/aobpla/ply012 Abstract. The distribution of cytotypes in mixed-ploidy species is crucial for evaluating ecological processes involved in the establishment and evolution of polyploid taxa. Here, we use flow cytometry and chromosome counts to explore cytotype diversity and distributions within a tetraploid–octoploid contact zone. We then use niche model- ling and ploidy seed screening to assess the roles of niche differentiation among cytotypes and reproductive inter- actions, respectively, in promoting cytotype coexistence. Two cytotypes, tetraploids and octoploids, were dominant within the contact zone. They were most often distributed parapatrically or allopatrically, resulting in high geo- graphic isolation. Still, 16.7 % of localities comprised two or more cytotypes, including the intermediate hexaploid cytotype. Tetraploids and octoploids had high environmental niche overlap and associated with similar climatic environments, suggesting they have similar ecological requirements. Given the geographical separation and habi- tat similarity among cytotypes, mixed-ploidy populations may be transitional and subject to the forces of minor- ity cytotype exclusion which lead to pure-ploidy populations. However, seed ploidy analysis suggests that strong reproductive barriers may enforce assortative mating which favours stable cytotype coexistence. High cytogenetic diversity detected in the field suggests that unreduced gamete formation and hybridization events seem frequent in the studied polyploid complex and might be involved with the recurrent polyploid formation, governing, as well, the gene flow between cytogenetic entities. Keywords: Contact zone; distribution patterns; Gladiolus communis; hexaploid; hybridization; niche modelling; niche overlapping; octoploid; tetraploid. et al. 2011) and sympatric speciation (Otto and Whitton Introduction 2000; Soltis et al. 2010). Based on recent molecular and Polyploidization, the duplication of complete chromo- fossil studies, polyploidy has been linked with radia- some sets, is widely considered an important mech- tions in species diversity throughout evolutionary his- anism of plant evolution (Soltis and Soltis 1999; Jiao tory (Soltis et  al. 2009) and associated with 15  % of *Corresponding author’s e-mail address: mcastro@uc.pt © The Author(s) 2018. Published by Oxford University Press on behalf of the Annals of Botany Company. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/ licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. AoB PLANTS https://academic.oup.com/aobpla © The Author(s) 2018 1 Downloaded from https://academic.oup.com/aobpla/article-abstract/10/2/ply012/4857208 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Castro et al. – Cytogeographical patterns in a tetraploid–octoploid complex speciation events in extant angiosperms (Wood et  al. which detect extensive cytogenetic diversity and, in 2009). Consequently, polyploidy is pervasive in flower - - several cases, occurrence of mixed-ploidy popula ing plants. The standing incidence of polyploid species is tions (e.g. Baack 2004; Kolář et al. 2009; Trávníček et al. estimated at 35 % (Wood et al. 2009), with higher values 2010; Castro et  al. 2012; Zozomová-Lihová et  al. 2015; being observed in specific geographic regions such as Wefferling et  al. 2017), rare cytotypes (e.g. Kolář et  al. the Mediterranean basin (ranging between 37 and 47 %; 2009; Trávníček et  al. 2010), production of unreduced Marques et al. 2017) and the Arctic region (69 and 87 %; gametes (e.g. Maceira et  al. 1992; Burton and Husband Brochmann et al. 2004). 2001; Ramsey 2007; Castro et  al. 2016a) or recurrent The geographic distribution of polyploids is useful for occurrence of gene flow (e.g. Husband 2004; Kolář et al. inferring mechanisms of polyploid evolution, coexist- 2009; Castro et  al. 2011). Particularly interesting are ence and divergence. The spatial arrangement of cyto- polyploid complexes with higher ploidies, such as dip- types in situ is the result of several interacting processes loid–hexaploid (e.g. Aster amellus; Castro et  al. 2012) operating in natural populations including formation or tetraploid–octoploid complexes (e.g. Gymnadenia and migration; ecological preferences, and competitive conopsea; Jersáková et  al. 2010), that can produce and dispersal abilities; and reproductive interactions, even-ploidy hybrids, which are potentially more stable among others (Petit et  al. 1999; Levin 2002; Lexer and and lead to highly dynamic contact zones. Regardless of van Loo 2006). Cytotype distributions can be character- the increasing number of studies at contact zones, the ized as sympatric, parapatric or allopatric depending on available information is still scarce and insufficient for whether the different cytotypes grow intermixed, adja- many plant groups and regions (Soltis et al. 2010, 2016; cent or disjunct, respectively (Petit et al. 1999; and illus- Marques et al. 2017). trated in Fig. 2 of Mallet et al. 2009, which can be applied Gladiolus communis is a Mediterranean polyploid com- to polyploid complexes). Theoretical models predict that plex with high morphological variation (Alonso and Crespo within zones of sympatry, mixed-ploidy populations are 2010). Multiple ploidy levels have been described for the expected to be rare and evolutionarily unstable because complex, namely tetraploids (2n = 4x = 60 chromosomes; frequency-dependent selection will drive the exclusion Fernandes et al. 1948; Fernandes 1950; Nilsson and Lassen of the minority cytotype (Levin 1975; Rodriguez 1996; 1971; Queirós 1980; Fernández et  al. 1985) and octop- Husband and Schemske 2000). Still, numerous studies loids (2n = 8x  =  120; Fernandes and Queirós 1971; Löve have documented mixed-ploidy populations (reviewed and Kjellqvist 1973; Queirós 1980), although hexaploids in Husband et al. 2013; and examples below). The pres- (2n = 6x  =  90) and duodecaploids (2n  =  12x  =  180) have ence of multiple cytotypes in the same population can also been occasionally reported in the Mediterranean basin reflect either a transitory stage, in which neopolyploids (Darlington and Wylie 1955). The Iberian Peninsula seems are recurrently formed, or a persistent stage such as to harbour this diversity (Fernandes et al. 1948; Fernandes when cytotypes are ecologically and reproductively and Queirós 1971; Queirós 1979) and areas of close con- isolated on a small spatial scale (e.g. Kolář et  al. 2009; tact between tetraploids and octoploids have been Jersáková et al. 2010). In this context, assessing the dis- detected, for example, in calcareous regions from Central tribution of cytotypes within and among natural popula- Portugal (Castro et  al. 2016b). Occasionally, G.  communis tions is crucial to build and test hypotheses that account grows with another congeneric species, namely G. italicus, for the successful establishment of polyploids. which, in the Iberian Peninsula, is represented by duode- Contact zones, areas with two or more cytotypes caploid individuals (Queirós 1979; Pérez and Pastor Díaz growing in close proximity, are thus considered natural 1994; although octoploids have also been described in the laboratories within which to study evolutionary transi- Mediterranean basin, e.g. Susnik and Lovka 1973; Strid and tions through polyploidy. In recent years, an increas- Franzen 1981; van Raamsdonk and De Vries 1989; Kamari ing number of studies have provided insights into et al. 2001). The high morphological variation of the group ploidy-mediated processes occurring in contact zones has led taxonomists to accept multiple taxonomic entities (e.g. Husband et  al. 2013; Ramsey and Ramsey 2014). within the G. communis complex (e.g. Gussone 1832; van Significant advances in this field have been largely Raamsdonk and De Vries 1989), although morphologic- fuelled by the ability to rapidly and easily screen thou- ally intermediate forms are found in natural populations, sands of individuals using flow cytometry (Kron et  al. and many characters used to distinguish each taxon are 2007). This approach has resulted in a proliferation of extremely variable and largely overlap, even within popula- cytogeographical studies (e.g. Baack 2004; Kolář et  al. tions (Hamilton 1980; revised in Alonso and Crespo 2010). 2009; Ståhlberg 2009; Trávníček et  al. 2010; Castro Consequently, recent morphological reviews and prelimin- ary molecular analyses failed to support the previous taxo- et  al. 2012; Zozomová-Lihová et  al. 2015; Wefferling et  al. 2017; reviewed in Ramsey and Ramsey 2014), nomic delimitations and the species is currently accepted 2 AoB PLANTS https://academic.oup.com/aobpla © The Author(s) 2018 Downloaded from https://academic.oup.com/aobpla/article-abstract/10/2/ply012/4857208 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Castro et al. – Cytogeographical patterns in a tetraploid–octoploid complex as a complex formed by three ploidy levels (Buchanan bulb, relatively thick roots, a cylindrical glabrous stem 2008; Alonso and Crespo 2010). Regardless of the variabil- and linear leaves with typical parallel ribs. The pink bisex- ity detected in the species, nothing is known about the role ual flowers are zygomorphic and usually grouped in one of genome duplications generating diversity within this spiked inflorescence per individual. A  second Gladiolus polyploid complex. Exploring cytotype diversity and dis- species, G. italicus, is found on the Iberian Peninsula and tribution patterns, especially at contact zones, is thus cru- occurs in sympatry with G.  communis in some places. cial to understand ecological processes, such as ecological Although very similar morphologically, these two spe- preferences and reproductive interactions, driving current cies are easily distinguished based on inflorescence diversity patterns at natural contact zones. architecture, anther and filament lengths, and seed In this study, we explore in detail the cytotype diver- morphology. Gladiolus communis has a unilateral inflor - sity and distribution patterns in a tetraploid–octoploid escence, anthers equalling or shorter than the filaments, G. communis contact zone. In particular, we pose the fol- and broadly winged seeds, while G.  italicus usually has lowing specific questions: (i) what are the dominant cyto - a weakly distichous inflorescence, anthers longer than types and how are they distributed in the contact zone? the filaments, and polyhedric apterous seeds (Hamilton (ii) do cytotypes coexist and at which spatial scale? (iii) 1980; Alonso and Crespo 2010). is coexistence facilitated by differences in environmen- In the Iberian Peninsula, G.  communis is recog- tal associations between cytotypes? And finally, (iv) is nized as a polyploid complex comprising tetraploids there evidence for the production of unreduced gametes (2n = 4x = 60 chromosomes), hexaploids (2n = 6x = 90) and/or cytotype hybridization? To address our questions, and octoploids (2n = 8x = 120) (e.g. Fernandes et al. 1948; cytotype diversity was studied at several spatial scales, Fernandes and Queirós 1971; Alonso and Crespo 2010) namely (i) across the contact zone, to characterize the with duodecaploids being described elsewhere in the most dominant cytotypes and their environmental pref Mediterranean region (Darlington and Wylie 1955). The erences within areas of contact; (ii) within mixed-ploidy high morphological resemblance among G.  communis populations, to measure microhabitat segregation; and cytotypes (Alonso and Crespo 2010; Cantor and Tolety (iii) among offspring from plants in pure- and mixed- 2011) suggests a putative autopolyploid origin. The spe- ploidy populations, to detect cytotype diversity at early cies is common in the calcareous regions from Central stages. Flow cytometric analyses complemented with Portugal, where preliminary field sampling revealed the chromosome counts were used to assess ploidy levels presence of tetraploid and octoploid populations grow- of all the sampled individuals. The reproductive success ing in close proximity. This study focused on this contact of pure- and mixed-ploidy populations was also quanti- zone, an area extending from 39.3° to 40.6° in latitude, fied in natural conditions to depict fitness differences and from 7.8° to 9.4° in longitude. This territory is domi- between cytotypes. The spatial arrangement of cyto- nated by calcareous rocks and presents a Mediterranean types in the contact zone was analysed with niche mod- climate that exhibits a strong influence from the Atlantic elling tools to determine if differences in environmental Ocean, an attribute identified on the significant values requirements could explain cytotype distribution. If cyto- of annual precipitation (1000–1300 mm). However, the types differ in environmental requirements, we expect dominance of poor soils determines a low water storage a mosaic contact zone with tetraploids and octoploids capacity, which, combined with a long and hot summer, fairly isolated within a given spatial scale and with plants determines the dominance of evergreen vegetation. growing in different habitats or microhabitats. If no envir- Allied to such climatic conditions, human pressure con- onmental differences are observed, we expect a tension tributed to current dominance of shrubby communities zone where sympatric cytotype co-occurrence is possible, in the landscape, and constrained forests (evergreen where intermediate cytotypes are detected and where and semi-deciduous) to very small patches, favouring other processes such as reproductive barriers, competi- the wide presence of open habitats. These open habi- tion or dispersal abilities are expected to play major roles tats are also characterized by the presence of limestone in driving distribution patterns. outcrops exposed to stressful ecological conditions that limit the installation of higher vegetation covers. Although not exhaustive, additional sampling was Methods extended beyond this area to determine the dominant cytotypes within the species. Also, because G. communis Study system and studied region coexist with G. italicus, hybridization might occur and gen- erate additional cytogenetic diversity, the duodecaploid Gladiolus communis is a perennial species that is wide- G. italicus (Queirós 1979; Pérez and Pastor Díaz 1994) was spread on the Iberian Peninsula and throughout the Mediterranean basin. The species produces an ovoid also sampled whenever growing with G. communis. AoB PLANTS https://academic.oup.com/aobpla © The Author(s) 2018 3 Downloaded from https://academic.oup.com/aobpla/article-abstract/10/2/ply012/4857208 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Castro et al. – Cytogeographical patterns in a tetraploid–octoploid complex ploidy were sampled to determine reproductive success Field sampling and screen ploidy of the seeds (see section Reproductive Field collections were carried out during the flowering success in natural populations). and fruiting seasons (mid-April to July) of G.  commu- nis from 2012 to 2015. Individual plants or clusters of Chromosome counts plants were easily detected when blooming because Chromosome counts were used to calibrate genome of the tall inflorescences growing above the remaining size estimates, obtained using flow cytometry, to a vegetation. We sampled 81 populations across the con- given ploidy level. For this, the plants grown from bulbs tact zone where both tetraploid and octoploid cytotypes collected in the selected natural populations and main- have been previously detected in close proximity. An tained in the common garden were used simultaneously additional group of 27 populations covering the west- for genome size estimates and chromosome counting. ern distribution of the species around the contact zone For chromosome counts, we followed the protocol of was also sampled to depict the dominant cytotypes [see Goldblatt and Takei (1993), with some adjustments. Supporting Information—Table S1]. In each of the 108 Briefly, actively growing root tips were harvested and populations, we collected ~3  cm of fresh leaf of up to pretreated in 0.002 M aqueous 8-hydroquinoline at 53 individuals of G.  communis (with an average of 20 room temperature for 1630  h, and fixed in 95  % etha - individuals per locality, excluding two particularly large nol and glacial acetic acid (in a ratio of 3:1) for at least localities where more intensive sampling was done, with 48 h at 4 °C. Roots tips were hydrolysed in 1 M hydrogen 106 and 454 plants being screened), and of G.  italicus chloride at 60 °C in a sand bath for 40 min, submerged whenever detected growing with G.  communis (up to in Schiff reagent (Greilhuber and Ebert 1994) for 1330 h, 32 individuals, averaging 13 plants per locality). The washed in sulphur water for three periods of 10 min and sampled individuals were randomly selected, covering finally squashed under a glass cover in a drop of acetic the extension of the population. Leaves were stored in orcein 2  %. Chromosome spreads were observed using labelled hermetic bags and maintained at 4 °C for later a Nikon Eclipse 80i light microscope and photographed flow cytometric analysis (see section Genome size and using a Nikon Plan Apo VC 100×/1.40 oil-immersion lens DNA ploidy level estimates). Geographic coordinates of with a Q Imaging Retiga 2000R Fast 1394 digital camera the population were recorded. Bulbs of nine localities and Q-Capture Pro v.7 software. A  total of 40 individu- identified in preliminary surveys as DNA tetraploid, DNA als from nine populations were used to access chromo- hexaploid and DNA octoploid populations (Castro et  al. some number and genome size: 4x—populations MC147 2016b) were also collected, potted and maintained at (n = 10 individuals), MC193 (n = 1), MC195 (n = 4), MC201 the common garden for chromosome counts (see sec- (n = 1) and MC212 (n = 2); 6x—population MC211 (n = 4); tion Chromosome counts). 8x—populations MC032 (n  =  8), MC143 (n  =  3), MC190 In addition, we sampled in mixed-ploidy popula- (n = 4), MC193 (n = 2) and MC201 (n = 1) [see Supporting tions more intensively to test for microhabitat seg- Information—Table S1]. regation. Three mixed-ploidy populations containing tetraploids, hexaploids and/or octoploids were revisited Genome size and DNA ploidy level estimates and all adult, individuals (both vegetative and repro- ductive individuals) were mapped with x/y coordinates, To estimate genome size and DNA ploidy levels, fresh tagged and sampled for ploidy level analyses using flow leaves collected in natural populations were analysed cytometry (see section Genome size and DNA ploidy using flow cytometry. Nuclear suspensions were pre - level estimates). To delimit the clusters of plants grow- pared following Galbraith et  al. (1983) by chopping the ing in sympatry, screenings for Gladiolus plants were plant material of the sampled species together with leaf made around a radius of over 150 m around the cluster tissue of an internal reference standard. In the case of of plants initially detected or until an anthropogenic or Gladiolus nuclear suspensions, 100  mg of leaf tissue or natural barrier was observed. Additional mixed-ploidy 2–5 seeds were co-chopped with 50 mg of leaf of Solanum populations were not sampled because they were dis- lycopersicum ‘Stupické’ (2C = 1.96 pg; Doležel et al. 1992) turbed by grazing or human activities. or Pisum sativum ‘Ctirad’ (2C  =  9.09 pg; Doležel et  al. Finally, we screened offspring from plants in pure- 1998). Solanum lycopersicum was used as the internal and mixed-ploidy populations to examine the produc- standard in most cases, except when unavailable, with tion of unreduced gametes and/or hybridization events P.  sativum being used in those situations. Sample and by the detection of rare cytotypes that might not reach standard were co-chopped in 1 mL of WPB buffer (WPB: the adult stage. For this, four tetraploid, two hexaploid, 0.2 M Tris–HCl, 4  mM MgCl ·6H O, 1  % Triton X-100, 2 2 four octoploid and one mixed tetraploid–octoploid pop- 2  mM EDTA Na ·2H O, 86  mM NaCl, 10  mM metabisul- 2 2 ulations were revisited and individual plants with known fite, 1 % PVP-10, pH adjusted to 7.5 and stored at 4 °C; 4 AoB PLANTS https://academic.oup.com/aobpla © The Author(s) 2018 Downloaded from https://academic.oup.com/aobpla/article-abstract/10/2/ply012/4857208 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Castro et al. – Cytogeographical patterns in a tetraploid–octoploid complex Loureiro et  al. 2007) using a razor blade. The resulting that of the endosperm. Consequently, the interpretation nuclear suspension was filtered through a 50  µm nylon of each histogram was made with particular caution, −1 filter and 50  µg mL propidium iodide (Fluka, Buchs, determining the ploidy levels of all the peaks obtained in −1 Switzerland) and 50 µg mL RNAse (Fluka) were added to the histogram. Preliminary analyses revealed that hexa- the sample, to stain the DNA and avoid staining of dou- ploid populations presented higher variability and thus ble-stranded RNA, respectively. After 5 min of incubation, only two seeds were pooled, in order to unambiguously DNA fluorescence of the sample was analysed using a assign the DNA ploidy levels of each seed. Partec CyFlow Space flow cytometer (532 nm green solid- The holoploid genome size (2C in pg; sensu Greilhuber state laser, operating at 30 mW; Partec GmbH, Görlitz, et al. 2005) was obtained using the following formula: Germany). Using Partec FloMax software v2.4d (Partec Holoploid reference GmbH, Münster, Germany) the following four histograms G.communisG peak mean genome = × standard were obtained: fluorescence pulse integral in linear scale referencestanda ardG peak mean size(pg) 1 genome size (FL); forward light scatter (FS) vs. side light scatter (SS), both in logarithmic (log) scale; FL vs. time; and FL vs. SS in log scale [see Supporting Information—Fig. S1]. To Based on the chromosome counts obtained in this study digitally remove some of the debris, the FL histogram and respective genome sizes, as well as the four chro- was gated using a polygonal region defined in the FL vs. mosome numbers described in the literature for G. com- SS histogram (see R1 in Supporting Information—Fig. munis and G. italicus, DNA ploidy levels were inferred for S1), and was further applied to all the other graphics. At each sample and individual. Populations were then char- least 1300 nuclei in both sample and standard G peaks acterized according to their DNA ploidy composition. were analysed per sample (Suda et al. 2007). Only coef- Descriptive statistics of holoploid genome size were ficient of variation (CV) values of G peak of G. communis calculated for each cytotype and species (mean, stand- below 5  % were considered acceptable (see examples ard deviation of the mean, coefficient of variation of the in Supporting Information—Fig. S1), otherwise a new mean, maximum and minimum values) based on the sample was prepared and analysed until this quality individual flow cytometric estimates. Mean and stand - standard was achieved (Greilhuber et al. 2007). ard deviation of the mean were also calculated for the Genome size was estimated in 41 populations by monoploid genome size (1Cx; holoploid genome size analysing three plants per population and cytotype indi- divided by inferred DNA ploidy level; sensu Greilhuber vidually (rarely less, unless there were no more plants et  al. 2005). Differences in holoploid and monoploid in the locality, while in a few populations up to 30 indi- genome sizes among species and cytotypes were inves- viduals were analysed for genome size) [see Supporting tigated using linear models (hereafter LM) performed Information—Table S2]. For the remaining individu- in R software version 3.0.1 (R Core Development Team als and populations, only DNA ploidy level information 2016), using the packages ‘car’ for Type-III analysis of was gathered following the pooled sample strategy variance (Fox et  al. 2015), ‘lme4’ for generalized linear (5–6 individuals plus the reference standard). A total of models (GLMs; Bates et  al. 2014) and ‘multcomp’ for 108 natural populations of G. communis and 2665 indi- multiple comparisons after Type-III analysis of variance viduals were sampled and analysed [see Supporting (Hothorn et al. 2017). Information—Table S1]. The geographical isolation index (GI) between the two We used flow cytometry to measure DNA ploidy of dominant cytotypes (i.e. tetraploids and octoploids) at offspring produced by plants of known ploidy. A total of the contact zone was calculated according to the fol- 1252 seeds from 178 individuals from four tetraploid, lowing formula (Husband et al. 2016), where only pure- two hexaploid and four octoploid pure-ploidy popula ploidy and mixed-ploidy populations of tetraploids and tions and one tetraploid–octoploid mixed population octoploids from the contact zone were considered: were analysed. We sampled 10–15 seeds per mater- nal individual, and 7–15 individuals per population and no. mixed-ploidy populations GI=- 1 cytotype. For pure-ploidy populations of tetraploids and total no. of populations octoploids and mixed-ploidy population, five seeds were chopped simultaneously with the internal reference Environmental preferences standard (pooled sample strategy) following the proto- col described above, producing easy to interpret histo- The environmental associations of the two dominant grams. When analysing the seeds, at least two peaks cytotypes were evaluated through GLM, and spatial pre- (plus the peak of the internal standard) were always dictive models were produced based on niche modelling obtained, corresponding to the peak of the embryo and tools, aiming to assess niche overlap. To explore niche AoB PLANTS https://academic.oup.com/aobpla © The Author(s) 2018 5 Downloaded from https://academic.oup.com/aobpla/article-abstract/10/2/ply012/4857208 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Castro et al. – Cytogeographical patterns in a tetraploid–octoploid complex overlapping, two approaches were used considering two was explored using Pearson coefficient for continuous different spatial scales: (i) one with an extension encom- variables and Spearman’s ρ for categorical variables, to passing the contact zone in Central Portugal; and (ii) the assist variable selection by removing variables with cor- other extension encompassing the entire territory of relation values higher than 0.7. The final set of variables mainland Portugal. selected included the following four which were also Variables were extracted from the following sources important descriptors of the type of habitat where the with a resolution of ~111 m: (i) bioclimatologi- species grows: mean annual total precipitation, mean cal data from http://home.isa.utl.pt/~tmh/aboutme/ temperature of the hottest month, soil texture and pH Informacao_bioclimatologica.html (methodology to (highlighted in bold in Table 1). obtain variables in Monteiro-Henriques et  al. 2016); Spatial predictive models were calibrated based on and (ii) data for soil conditions from: http://epic-web presence/absence records collected in the field and gis-portugal.isa.ulisboa.pt/. Values for climatic and soil the selected environmental and soil variables (Table 1). variables were extracted for all the surveyed popula- For the tetraploid data set, tetraploid populations were tions using the R package ‘dismo’ (Hijmans et al. 2017). recorded as presences and octoploid populations as Then, GLMs were used to explore climatic and soil vari- absences, and vice versa for the octoploid data set. ables and assess differences between tetraploid and Mixed tetraploid–octoploid populations were consid- octoploid populations (Table 1), namely for climatic vari- ered as presences for both cytotypes. For the contact ables [mean annual total precipitation (PP), mean tem- zone (Central Portugal) we used data from 76 sampling perature of the hottest month of the year (Tmax), mean points (including 33 tetraploid, 40 octoploid and 3 tetra- temperature of the coldest month of the year (Tmin), ploid–octoploid populations), corresponding to all the mean maximum temperature of the coldest month of known occurrences of G. communis with a minimum dis- the year (M), mean minimum temperature of the coldest tance between populations of 600 m.  For the territory month of the year (m)], bioclimatic indexes [continental- of Portugal, and aiming to reduce the bias effect of spa- ity index (IC), compensated thermicity index (ITC), sum- tial clustering associated with our intense screening in Table 1. Characterization of the climatic and soil variables for tetraploid and octoploid populations of Gladiolus communis in the contact zone of Central Portugal. The mean, standard error of the mean (SE) and statistical tests (comparison between cytotypes) are provided for each variable and cytotype. Significance levels: ***P < 0.01; *0.05 < P < 0.01; n.s., non-significant. In bold the variables used in niche modelling are highlighted. Variables were extracted from the following sources with a resolution of ~111 m: (i) bioclimatological data from http://home.isa. utl.pt/~tmh/aboutme/Informacao_bioclimatologica.html (methodology to obtain variables in Monteiro-Henriques et  al. 2016); and (ii) data for soil conditions from: http://epic-webgis-portugal.isa.ulisboa.pt/. Variables Code Tetrapoid Octoploid ANOVA F value 1, 78 Mean ± SE, n = 43 Mean ± SE, n = 36 a a Precipitation PP 1096.11 ± 21.97 1106.89 ± 21.74 0.12 n.s. a b Mean temperature of the hottest month of the year Tmax 20.51 ± 0.13 20.95 ± 0.19 4.04* a a Mean temperature of the coldest month of the year Tmin 9.06 ± 0.10 8.91 ± 0.12 1.10 n.s. a a Mean max. temp. of the coldest month of the year M 13.52 ± 0.11 13.42 ± 0.12 0.38 n.s. a a Mean min. temp. of the coldest month of the year m 4.61 ± 0.08 4.50 ± 0.09 0.84 n.s. a b Continentality index IC 11.44 ± 0.13 12.04 ± 0.19 6.95* a a Compensated thermicity index ITC 327.12 ± 2.78 325.81 ± 3.22 0.10 n.s. a a Summer ombrothermic index Ios3 1.10 ± 0.03 1.09 ± 0.03 0.05 n.s. a a Soil texture Texture 2.14 ± 0.29 2.25 ± 0.13 0.11 n.s. a b Soil pH pH 308.60 ± 46.51 99.44 ± 19.69 15.02*** a a Altitude Alt 198.61 ± 18.63 169.23 ± 17.38 0.94 n.s. a a Latitude Lat −8.47 ± 0.05 −8.58 ± 0.03 3.14 n.s. a a Longitude Long 39.98 ± 0.05 40.01 ± 0.04 0.18 n.s. mer ombrothermic index (Ios3)], soil conditions [texture the contact zone, only occurrences that had a minimum (txt) and pH] and altitude. Correlation between variables distance of 10 km between them were selected, based 6 AoB PLANTS https://academic.oup.com/aobpla © The Author(s) 2018 Downloaded from https://academic.oup.com/aobpla/article-abstract/10/2/ply012/4857208 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Castro et al. – Cytogeographical patterns in a tetraploid–octoploid complex on radon selection, resulting in a subset of 66 sampling (Broennimann et al. 2012). In niche similarity (similarity points (including 35 tetraploid, 19 octoploid and 6 tetra- test), we evaluate if the environmental niches of the two ploid–octoploid populations). cytotypes were distinguishable from each other. In this Environmental niche modelling (ENM) of tetraploids case, the comparison was between the points of one and octoploids was created using R package ‘biomod2’ cytotype and random points from the geographic range (Thuiller et  al. 2016). Final model for each cytotype is of the other cytotype. As in the identity test, the pro- based on the combination of results from different mod- cess was repeated 100 times and D values were calcu- elling techniques, each one replicated 30 times after lated. The results revealed if niche overlap between the data splitting into training (70  %) and testing (30  %) cytotypes is greater (niche conservation) or lower (niche subsets based on random selection, aiming to reduce divergence) than expected, according to the geographic uncertainty and to produce robust models (Phillips et al. region of the other cytotype. All the models and analy- - ses were performed in R software version 3.0.1 (R Core 2006; Araújo and New 2007). In the resampling repli cation, each specific occurrence was used only once in Development Team 2016). each run, as training or as test without replacement, Reproductive success in natural populations making all replicates statistically independent (Phillips 2008). Models were evaluated based on the independ- The reproductive success of each cytotype was evalu- ent accuracy measure AUC of ROC (area under the curve ated in 11 natural populations, namely 10 pure-ploidy of the receiver operating characteristic), and only those populations (including four tetraploid, two hexaploid with AUC > 0.7 were used in the ensemble forecast- and four octoploid populations) and one mixed-ploidy ing procedure, the approach used to produce the final population composed by tetraploid and octoploid indi- model for each cytotype. viduals (MC201). In each population, 11–20 individuals Model evaluation revealed high ROC values (contact of known ploidy level were labelled and infructescences zone: 4x—0.79  ±  0.01 and 8x—0.79  ±  0.01; Portugal: collected in individually labelled bags. The number of 4x—0.77  ±  0.01 and 8x—0.76  ±  0.01) and relatively fruits was counted for each inflorescence and fruit set low omission rates (contact zone: 4x—0.19  ±  0.02 and calculated as the proportion of flowers that developed 8x—0.28  ±  0.02; Portugal: 4x—0.23  ±  0.01 and 8x— into fruit. The number of morphologically viable seeds 0.28 ± 0.01). However, when considering the binary pro- (based on their size and shape) was assessed in all fruits, jections, the omission rates decrease to 0.10 and 0.09 and the seed:ovule ratio (S:O ratio) was calculated by for the tetraploid and octoploid models in the contact dividing the number of morphologically viable seeds by zone, respectively, and 0.17 and 0.04 in Portugal (tetra- the number of ovules. The total reproductive success of ploids and octoploids, respectively), demonstrating that populations and cytotypes was also calculated by multi- the models were able to predict the occurrences with plying the S:O ratio by the fruit set. Descriptive statistics high accuracy, namely for octoploids. The binary projec- were calculated for each population type. tion produced by the final model of each cytotype was Differences in fruit set, S:O ratio and total reproduc- used to calculate niche overlap. tive success between the three cytotypes (tetraploids, Cytotype niche overlap was quantified through the hexaploids and octoploids) within pure-ploidy popula- metric of proportional similarity of the distribution of tions, differences between tetraploids and octoploids in both cytotypes, using Schoener’s D (a measure of niche the mixed-ploidy population, and differences between similarity; Schoener 1970). This metric ranges from zero pure- and mixed-ploidy populations (excluding hexa- (no overlap) to one (complete overlap). The ‘ecospat’ ploid ones) were assessed using GLM. Mixed models with (Broennimann et  al. 2012) and ‘raster’ (Hijmans et al. individual and population as random factors were ini- 2017) packages were used to perform niche identity and tially used, but the random factors were further removed similarity tests (Warren et  al. 2008; Broennimann et  al. due to low variance in comparison with residuals (Bolker 2012). In niche equivalency (identity test), the points et al. 2009). A binomial distribution with a logit link func- of both cytotypes were pooled and randomly split in tion was used for fruit set, and a Gaussian distribution two groups according to size of the original data set. with an identity link function was used for S:O ratio and This new data set was used in D calculation, and the total reproductive success after transformation with the process was repeated 100 times (to obtain confidence arcsine of the square root. When significant differences intervals that enable evaluation of the null hypothesis). were obtained, post hoc tests for multiple comparisons The resulting D values (simulated values) were com- were performed. pared with the observed D value, and cytotype niches All analyses were performed in R software ver- were considered equivalent if the observed D value fell sion 3.0.1 (R Core Development Team 2016), using within the 95th percentile of the simulated D values the packages ‘car’ for Type-III analysis of variance AoB PLANTS https://academic.oup.com/aobpla © The Author(s) 2018 7 Downloaded from https://academic.oup.com/aobpla/article-abstract/10/2/ply012/4857208 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Castro et al. – Cytogeographical patterns in a tetraploid–octoploid complex (Fox and Weisberg 2015), ‘lme4’ for GLMs (Bates et  al. segregation pattern of cytotype arrangement in space 2014) and ‘multcomp’ for multiple comparisons after was observed: tetraploids seem to occur across the Type-III analysis of variance (Hothorn et al. 2017). entire area surveyed, and octoploids in the centre and south of the surveyed area, forming broad contact zones. Minority cytotypes were also detected, namely Results hexaploids, which were observed growing with other cytotypes and occasionally found forming pure popula- Genome size and cytogenetic diversity tions (Fig. 3). A few nonaploids in a mixed-ploidy popu- Based on chromosome counts and flow cytometric anal - lation harbouring all cytotypes of G. communis were also yses, we detected three ploidy levels in G.  communis: detected (Fig.  3) [see Supporting Information—Table tetraploids with 2n = 4x = 60 chromosomes (Fig. 1A) and S3]. Most populations were cytogenetically uniform (i.e. an average genome size of 2.69  ±  0.06 pg/2C (mean ± pure-ploidy populations, 86.1 %) and, in the majority of SD); hexaploids with 2n = 6x = 90 chromosomes (Fig. 1B) cases, were composed of either tetraploid or octoploid and an average genome size of 4.07 ± 0.07 pg/2C; and individuals (43.5 and 39.8  %, respectively). Hexaploids octoploids with 2n = 8x = 120 chromosomes (Fig. 1C) and were detected growing alone in three locations (2.8 %) an average genome size of 5.42  ±  0.14 pg/2C (Table  2; (Fig.  3). Populations harbouring two or more cytotypes Fig. 2A and B) [see Supporting Information—Table S2]. (i.e. mixed-ploidy populations) represented 13.9  % of Genome size estimates also suggest the occurrence of all sampled populations. The mixed-ploidy populations nonaploid G.  communis individuals, characterized by presented different cytotype compositions: tetraploids genome sizes with nine times the monoploid genome size and hexaploids (4.6 %), in which the former is more fre- (1Cx) values obtained for the other ploidy levels, and had quent than the latter; tetraploids and octoploids (5.6 %) mean genome size of 6.10 ± 0.18 pg/2C (Table 2; Fig. 2C). again, in which tetraploids are generally more abundant These individuals were rare and we were unable to con- than octoploids, except in one population; tetraploids, firm their ploidy using chromosome counts. Gladiolus hexaploids, octoploids and nonaploids (0.9  %; one italicus had a higher genome size (2C = 7.27 ± 0.17 pg) population), where octoploids are the dominant cyto- than G.  communis, consistent with duodecaploids, as type; and hexaploids and octoploids (2.8  %), in which described for the species (Table  2; Fig.  2D). The holop- octoploids are dominant, except in one location where loid genome sizes (2C) of the five cytotypes differed sig - only two plants, one of each cytotype, were found [see nificantly (F  = 7691.3, P < 0.001; Table 2). Monoploid Supporting Information—Table S3]. 4, 175 genome size values were conserved within G. communis, Within the contact area (Fig. 3B), most localities con- with no significant differences being observed between tained a single ploidy of either tetraploids (42.0  %), cytotypes (F  = 7691.3, P = 0.5272; Table 2). However, octoploids (44.4  %) or rarely hexaploids (2.5  %). These 3, 155 monoploid genome size of G. communis (0.67 ± 0.03 pg) populations were distributed mostly in parapatry; still, was significantly higher than for G.  italicus (0.61 ± 0.01 cytotypes were found growing in sympatry in some loca- pg; F  = 7691.3, P < 0.001). tions (11.1 %) (Fig. 3B). Octoploid populations occur from 1, 178 north to south, resulting in cytogenetically diverse con- Geographic distribution of cytotypes tact zones with tetraploids to the east, south and south- Tetraploids and octoploids were prevalent across the west. At these contact zones, areas with different types geographic area sampled, both occurring in pure- of mixed-ploidy populations were detected. Hexaploids and in mixed-ploidy populations (Fig.  3). No marked were frequent in the contact zones between tetraploids Figure 1. Gladiolus communis chromosome counts. (A) tetraploid (2n = 4x = 60 chromosomes), (B) hexaploid (2n = 6x = 90) and (C) octoploid (2n = 8x = 120) individuals. Bar = 20 µm. 8 AoB PLANTS https://academic.oup.com/aobpla © The Author(s) 2018 Downloaded from https://academic.oup.com/aobpla/article-abstract/10/2/ply012/4857208 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Castro et al. – Cytogeographical patterns in a tetraploid–octoploid complex Table 2. Genome size and DNA ploidy level estimations for Gladiolus communis and G. italicus. Holoploid genome size (G.s.; 2C) is provided for each species and cytotype as mean and standard deviation (SD) in picograms (pg), followed by coefficient of variation (CV, %), DNA range (minimum, Min, and maximum, Max, genome size values); mean and standard deviation of the mean is also provided for the monoploid genome size (1Cx); total number of individuals (N total) and populations (N pop) analysed for genome size are also provided for each ploidy level. Chromosome numbers (Chr. number) for each species based on chromosome counts of this study and bibliographic records are also provided. DNA ploidy levels: tetraploid (4x), hexaploid (6x), octoploid (8x), nonaploid (9x) and duodecaploid (12x). Different letters denote 1 2 statistically significant differences at P  <  0.05. Chromosome numbers detected in this study; Chromosome counts documented in the bibliography; DNA ploidy level extrapolated based on the genome size values obtained here and on the chromosome counts available from other ploidy levels. Species DNA ploidy level Chr. Holoploid G.s. (2C, pg) Monoploid G.s. N total N pop number (1Cx, pg) Mean SD CV (%) Min Max Mean SD 1,2 a a G. communis 4x 60 2.69 0.06 2.28 2.58 2.86 0.67 0.02 57 16 1,2 b a G. communis 6x 90 4.07 0.07 1.76 3.93 4.19 0.68 0.01 9 3 1,2 c a G. communis 8x 120 5.42 0.14 2.53 5.13 5.73 0.68 0.02 91 21 3 d a G. communis 9x 135 6.10 0.18 2.89 5.98 6.23 0.68 0.02 2 1 2 e b G. italicus 12x 180 7.27 0.17 2.31 6.97 7.55 0.61 0.01 21 10 and octoploids, although they were also detected in diversity (MC193) revealed to be dominated by octoploid other places of the screened area, growing with tetra- individuals with a few tetraploid, hexaploid and nona- ploid individuals. Tetraploids and octoploids, the two ploid plants growing intermingled (Fig. 4C). While MC201 main cytotypes, were observed growing together in four and MC193 were located in the contact zones, MC232 is locations out of the 81 populations at the contact zone located in an otherwise tetraploid zone (Fig. 3). (4.9  %), resulting in a total GI of 0.95, with tetraploids Environmental preferences and octoploids presenting a similar individual geograph- Niche geographic overlap between tetraploids and ical isolation index (GI  = 0.90, GI  = 0.91). 4x 8x octoploids at both the contact zone (Schoener’s D met- The detailed screening of three selected mixed-ploidy populations revealed variable patterns of cytotype dis- ric, D = 0.03) and Portugal (D = 0.01) was low (Table  3). However, and despite little geographical overlap, there tribution within each population (Fig.  4). In the tetra- was no statistical evidence that the environmental ploid–octoploid population (MC201), cytotypes were distributed in two well-defined clusters separated by >20 niches differed, i.e. neither niche equivalency nor niche similarity was rejected (Table  3). This indicates that m, with tetraploids being restricted to the north-east environmental niche of the dominant cytotypes was side and octoploids to the south-west of the population equivalent within the suitable ranges of both tetraploids (Fig.  4A). The mixed-ploidy population with tetraploids and octoploids, and that environmental niche of each and hexaploids (MC232) was dominated by tetraploid cytotype was similar to the suitable range of the other individuals, with a few hexaploids growing intermingled cytotype. At the contact zone, the selected climatic and (Fig.  4B). The population with the highest cytogenetic Figure 2. Flow cytometric histograms of relative propidium iodide fluorescence from nuclei isolated from fresh leaves of Solanum lycopersi- cum ‘Stupické’ (S.l.) and different cytotypes and/or species of Gladiolus: (A) tetraploid (4x) and hexaploid (6x), (B) octoploid (8x) and (C) nona- ploid (9x) individuals of G. communis, and (D) duodecaploid (12x) individual of G. italicus. AoB PLANTS https://academic.oup.com/aobpla © The Author(s) 2018 9 Downloaded from https://academic.oup.com/aobpla/article-abstract/10/2/ply012/4857208 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Castro et al. – Cytogeographical patterns in a tetraploid–octoploid complex Figure 3. Gladiolus communis cytotype screening: (A) all studied area (Portugal); and (B) detail of the contact zone studied (Central Portugal). White, grey and black circles represent pure tetraploid, hexaploid and octoploid populations, respectively. Mixed-ploidy populations are rep- resented by a grey diamond and each population is accompanied by a pie diagram reflecting cytotype composition. One sole population harbouring also two nonaploid individuals (not included in the pie diagram) is denoted by a dotted grey diamond, namely population MC193. Populations identified with ID code correspond to the populations where all the individual plants were sampled in detail (see Fig.  4). DNA ploidy levels: tetraploid (4x), hexaploid (6x), octoploid (8x). soil variables explained 62.98  % of the variance in the having lower fruit set than tetraploids. S:O ratio and distribution (Fig. 5A), and a high environmental overlap reproductive success were similar in tetraploids and of a given cytotype within the niche of the opposite cyto- octoploids (P > 0.05), but significantly higher than type was observed (74.87 and 61.95  % for tetraploids for hexaploids (P < 0.05; Fig.  6B and C). Within the and octoploids, respectively; Fig.  5B). A  similar pattern mixed-ploidy population, no significant differences was observed for Portugal, although the climatic and were observed between the cytotypes for any of the soil variables explained higher variance than at the reproductive variables (fruit set: F  = 0.27, P = 0.603; 1, 79 contact zone (74.78  %; Fig.  5C). A  high environmental S:O ratio: F  = 0.01, P = 0.934; reproductive success: 1, 37 overlap between cytotypes was also observed (91.51 F  = 0.15, P = 0.698). 1, 79 and 47.96 % for tetraploids and octoploids, respectively; The analyses of offspring ploidy revealed that tetra- Fig. 5D). ploid and octoploid individuals, in both pure-ploidy and mixed-ploidy populations, produced seeds with the same Reproductive success in natural populations and ploidy as the mother plants (Fig. 6C). Tetraploid plants in offspring cytogenetic composition pure populations produced a few aneuploids (<1  % of Plants in all the natural populations successfully the offspring; Fig.  6C) [see Supporting Information— formed fruits and seeds. However, the success differed Table S4]. In contrast, the flow cytometric analyses of according to the cytotype and population type. Pure- the seeds from hexaploid individuals pointed out highly ploidy populations (excluding the hexaploid popula- variable genome sizes, the analyses of the genome size tions) had higher reproductive success compared to estimates suggest the following DNA ploidy levels: 62 % the mixed-ploidy population for all parameters (fruit of seeds were aneuploid, 20  % were pentaploids and set: F   =  15.51, P  <  0.001; S:O ratio: F   =  4.62, 18  % were hexaploid, although further confirmation is 1,  1033 1,  706 P  =  0.032; reproductive success: F   =  21.04, needed. 1,  1033 P < 0.001; Fig. 6). Within pure-ploidy populations, sig- nificant differences between cytotypes were observed Discussion for all the variables (fruit set: F   =  4.96, P  =  0.007; 2,  1087 S:O ratio: F   =  100.18, P  <  0.001; reproductive This study corroborates the existence of high cytogenetic 2,  770 success: F   =  28.34, P  <  0.001), with octoploids diversity within the G. communis polyploid complex. Two 2,  1087 10 AoB PLANTS https://academic.oup.com/aobpla © The Author(s) 2018 Downloaded from https://academic.oup.com/aobpla/article-abstract/10/2/ply012/4857208 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Castro et al. – Cytogeographical patterns in a tetraploid–octoploid complex Figure 4. Fine-scale distribution of Gladiolus communis individuals within three mixed-ploidy populations: (A) tetraploid and octoploid mixed- ploidy population (MC201), (B) tetraploid and hexaploid mixed-ploidy population (MC232); and (C) tetraploid, hexaploid, octoploid and nona- ploid mixed-ploidy population (MC193). Each point represents one individual plant mapped in a x/y system where distance is given in metres (m): tetraploids (4x), hexaploids (6x), octoploids (8x) and nonaploids (9x) individuals are represent by white, grey, black and dark grey points, respectively. dominant cytotypes, tetraploids and octoploids, were occurred throughout the sampling area, although they observed along with two minority cytotypes, mostly were more common in the north and central regions hexaploids, and rarely nonaploids. Tetraploids and of Portugal. Octoploids occurred in south and central octoploids have been well documented on the Iberian regions of Portugal, but not in the north, notwithstand- Peninsula through chromosome counts (Fernandes ing the fact that more extensive surveys are needed to et  al. 1948; Fernandes 1950; Fernandes and Queirós confirm this pattern. Although several mixed-ploidy pop - 1971; Nilsson and Lassen 1971; Löve and Kjellqvist 1973; ulations were found, the geographical isolation index Queirós 1980; Fernández et  al. 1985). Also, hexaploids between tetraploids and octoploids is high, reflecting have been previously reported in the Mediterranean the fact that most of the populations contain a single basin (Darlington and Wylie 1955). We observed them cytotype. These populations distribute in space allopat- rically or parapatrically, forming several contact zones in 11  % of the sampled localities (12 of 105 localities), commonly growing with one of the dominant cytotypes between tetraploids and octoploids. However, despite and occasionally in pure-ploidy populations. Nonaploids that tetraploids and octoploids have non-overlapping are reported here for the first time and were detected in distributions, they can inhabit similar environmental the most diverse mixed-ploidy population. niches. Niche identity and similarity tests showed that Despite the cytogenetic diversity reported in G.  com- tetraploids and octoploids occupy similar niches and are munis, almost nothing was known about the geographic not differentiated in their environmental niches, show- distribution of the cytotypes, or the presence and struc- ing niche conservation. These results contrast with other ture of its contact zones. Based on our survey, tetraploids polyploid complexes for which niche differentiation, driven either by the direct effects of polyploidy or by Table  3. Niche analyses in Gladiolus communis. For each region subsequent selection, underlies the spatial separation of studied, equivalency (D and P values) and similarity (P value) tests cytotypes and allows them to escape the minority cyto- for suitable habitat are given. type disadvantage (e.g. Glennon et al. 2012; Thompson Suitable Equivalence test Similarity test (P values) et al. 2014; Visger et al. 2016; Muñoz-Pajares et al. 2017). habitat Still, the absence of environmental niche differences D value P value Tetra -> Octo Octo -> Tetra might not be completely unexpected as polyploids might not differ from their lower ploidy ancestors, either Contact 0.034 0.960 0.406 0.337 zone because they have been formed recently and the new polyploids did not have time to diverge from their pro- Portugal 0.009 0.515 0.535 0.515 genitors, because genome duplications did not generate AoB PLANTS https://academic.oup.com/aobpla © The Author(s) 2018 11 Downloaded from https://academic.oup.com/aobpla/article-abstract/10/2/ply012/4857208 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Castro et al. – Cytogeographical patterns in a tetraploid–octoploid complex Figure 5. Results of ecological niche models for Gladiolus communis polyploid complex at (A, B) the contact zone in Central Portugal, and (C, D) Portugal. (A) and (C) represent the contribution of climatic and soil variables in the first two axes of the principal component analyses (PCA) and the percentage of variance explained by each axis. (B) and (C) represent the environmental niche of each cytotypes based on the PCA of selected variables; coloured areas represent suitable habitats as follows: light grey—tetraploids; dark grey—octoploids; and green—overlap- ping areas between tetraploids and octoploids; the continuous line corresponds to the whole climatic space, while the dashed line indicates the 75th percentile. significant direct physiological changes, and/or because in several polyploid complexes (e.g. Baack 2004, 2005; they might have been subjected to recurrent gene flow Pannell et  al. 2004; Baack and Stanton 2005; Godsoe (Godsoe et al. 2013; Laport et al. 2016). Also, the effect et  al. 2013; Münzbergová et  al. 2013; Wefferling et  al. of other environmental parameters on the distribution 2017). Contact zones are generated by direct emer- patterns observed in G. communis cannot be completely gence of neopolyploids in lower ploidy parental pop- ruled out, nor the fact that niche differentiation might ulations or through secondary contact of previously occur at a special resolution higher than that used in our allopatric distributions in which cytotypes colonized study, although we did not find any clear evidence of dif - the area separately in dissimilar ways and at differ- ferentiation in the field, namely considering the type of ent timings (Petit et al. 1999; Lexer and van Loo 2006). vegetation or the type of substrate in the mixed-ploidy Although we still do not know the origin of G. commu- populations detected (M. Castro, field observations). nis contact zones, the different cytotype compositions Considering that G. communis cytotypes do not differ found in natural populations provide significant insights in suitable habitat, there should be historical processes into the processes that might be occurring at these and other ecological determinants shaping their distri- areas (e.g. Husband and Schemske 1998; reviewed in butional patterns, similarly to what has been observed Husband et al. 2013; Suda et al. 2013). One of the main 12 AoB PLANTS https://academic.oup.com/aobpla © The Author(s) 2018 Downloaded from https://academic.oup.com/aobpla/article-abstract/10/2/ply012/4857208 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Castro et al. – Cytogeographical patterns in a tetraploid–octoploid complex Figure  6. Reproductive fitness of natural pure- and mixed-ploidy populations of Gladiolus communis: (A) fruit set; (B) S:O ratio (number of viable seeds divided by the number of ovules); and (C) reproductive success (fruit set multiplied by S:O ratio). In (C) the proportion of DNA ploidy levels detected in the offspring is also given. DNA ploidy levels: tetraploid (4x), pentaploid (5x), hexaploid (6x), octoploid (8x); seeds with genome size values out of the range of variation of each ploidy levels were assumed as aneuploids (An.). Different letters correspond to statistically significant differences as follows: (i) differences between population type (pure- vs. mixed-ploidy populations, excluding 6x) are denoted by upper case letters; and (ii) differences between ploidy levels within population type (among 4x, 6x and 8x from pure populations, and between 4x and 8x from the mixed-ploidy population) are denoted by lower case letters (Tukey HSD; P < 0.05); n.s. correspond to non- significant differences (P > 0.05). observations is the fairly few mixed-ploidy populations isolation observed between G.  communis cytotypes (10 vs. 90  % of mixed- and pure-ploidy populations), suggests that the mixed-ploidy populations might be all composed of unbalanced number of tetraploid and transitory because strong frequency-dependent selec- octoploid plants (either dominated by tetraploid or by tion is expected to eliminate the minority cytotype as octoploids). In the absence of environmental differ- a result of fitness disadvantage generated by its lower ences, and regardless of the origin of the contact zone, number. This selection will ultimately drive the occur- G.  communis mixed tetraploid–octoploid populations rence of pure-ploidy populations at contact zones are expected to be more common at contact areas (Levin 1975; Husband 2000). than detected here (4.9  % in the contact zone and However, tetraploid–octoploid populations may per- 6.5  % from the total), since cytotypes might disperse sist in nature. The regular production of unreduced to areas of the other cytotype and/or new cytotypes gametes and the presence of reproductive barriers pro- might be formed. Consequently, the high geographical moting assortative mating might lessen the magnitude AoB PLANTS https://academic.oup.com/aobpla © The Author(s) 2018 13 Downloaded from https://academic.oup.com/aobpla/article-abstract/10/2/ply012/4857208 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Castro et al. – Cytogeographical patterns in a tetraploid–octoploid complex of frequency-dependent selection and enable cytotype unreduced gamete formation is an important path- coexistence (e.g. Felber 1991; Segraves and Thompson way for new polyploid emergence and has been shown 1999; Husband 2004; Husband and Sabara 2004; to be common in nature (Felber 1991; Bretagnolle and Kennedy et al. 2006). Octoploids might emerge directly Thompson 1995; Ramsey and Schemske 1998; Husband in tetraploid populations through the union of two unre- 2004; Ramsey 2007). This might explain the detection duced gametes (n = 4x) or might result from seed disper- of hexaploid plants frequently found in otherwise tetra- sal from neighbouring octoploid populations. Unreduced ploid populations. Alternatively, hexaploids may form gamete production has been detected in controlled pol- as a result of hybridization events between tetraploid linations in tetraploid G.  communis (M. Castro et  al., in and octoploid G.  communis individuals. Gladiolus com- preparation) and in screenings in natural populations munis is pollinated by generalist pollinators that seem through the detection of hexaploid individuals (see to have no cytotype preferences and might move pol- below). The rates at which unreduced gametes are pro- len within mixed-ploidy populations or between popula- duced might feed the population of octoploids enabling tions in close proximity (M. Castro et al., in preparation). their maintenance within tetraploid populations (Felber Additionally, controlled pollinations between tetraploid 1991; Husband 2004). Additionally, seed ploidy analyses and octoploid plants were also successful in produc- in a tetraploid–octoploid population suggest that strong ing hexaploid offspring (M. Castro et al., in preparation). reproductive barriers may enforce assortative mating, Either one of these pathways, i.e. unreduced gamete for- further favouring cytotype coexistence. Reproductive mation or hybridization, may operate in natural popula- barriers driven, for example, by phenological and/or mor- tions, being difficult to distinguish them without genetic phological mismatch, different pollinator assemblages markers. However, the relative abundance of tetraploid– or preferences, and/or gametic isolation will, thus, play a hexaploid populations and paucity of tetraploid–hexa- major role for overcoming minority cytotype exclusion in ploid–octoploid populations suggests that the majority mixed-ploidy populations. Therefore, the fate of octop- of the hexaploids are formed through unreduced gam- loids might depend not only on the rates of unreduced etes in tetraploid populations. Additionally, unreduced gamete formation but also on the reproductive isolation gamete production has been frequently detected in con- levels. Additionally, differences in other traits, such as trolled pollinations involving tetraploid G. communis (M. perenniality or asexual reproduction, could compensate Castro et al., in preparation), supporting it as a probable for the minority cytotype disadvantage (e.g. Rodriguez pathway for new cytotype emergence in natural popula- 1996; Kao 2007; Castro et al. 2016a). In other polyploid tions. Quantifying unreduced gamete production in nat- complexes, traits such as the production of bulbs repre- ural populations will provide significant insights on how sented an advantage, enabling new cytotypes to persist frequent this process could be involved with hexaploid at initial stages and spread within lower ploidy popula- emergence. tions (e.g. Allium oleraceum; Duchoslav et al. 2010; G. × Interestingly, hexaploid individuals were also found sulistrovicus; Szczepaniak et  al. 2016). If, through some forming pure-ploidy populations, showing that this of these traits, the number of octoploids can surpass the cytotype can successfully establish and spread beyond number of tetraploids, at some time octoploids might parental populations, although their sexual repro- even outcompete tetraploids and exclude them from ductive fitness was revealed to be lower in compari - the population, as observed in other polyploid com- son with tetraploids and octoploids. Regardless of the plexes (e.g. Buggs and Pannell 2007). Indeed, octop- lower fitness, recurrent unreduced gamete formation loids were observed as the dominant cytotype in some and asexual reproduction might enable to compensate mixed-ploidy populations of the contact zone. Future for this disadvantage (e.g. Husband 2004; Kao 2007; studies on the contribution of all the above-mentioned Castro et  al. 2016a). The successful establishment of processes, and on the relative contribution of sexual vs. hexaploid plants further contributes to the diversifica - asexual reproduction for the maintenance of the popu- tion of the complex. Ultimately, contact zones result lations, are needed to fully understand the dynamics of from the combination of several factors, including mixed-ploidy populations. historical factors, unreduced gamete formation, pol- The cytotype composition of G.  communis natural len flow and hybridization events, and seed dispersal, populations also revealed that hexaploid plants might among others (Petit et al. 1999; Levin 2002; Lexer and be more common than previously thought. These hexa- van Loo 2006). Future studies reconstructing the his- ploids might have originated through two different tory of the complex and quantifying unreduced gamete pathways. Hexaploids may originate from tetraploids production, and its ability to hybridize, would provide through the union of reduced (n = 2x) and unreduced significant insights on the dynamics of the distribution (n = 4x) gametes (Ramsey and Schemske 1998). Indeed, of G. communis. 14 AoB PLANTS https://academic.oup.com/aobpla © The Author(s) 2018 Downloaded from https://academic.oup.com/aobpla/article-abstract/10/2/ply012/4857208 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Castro et al. – Cytogeographical patterns in a tetraploid–octoploid complex The genome size of G. italicus suggests that this spe- mixed-ploidy populations were expected to be fre- cies is duodecaploid in the studied area, which is in quent; however, a high geographical isolation index was accordance with chromosome counts for the Iberian obtained. The high geographical isolation observed in Peninsula (Queirós 1979; Pérez and Pastor Díaz 1994), nature, along with habitat similarity, suggests that the and contrasts with the dominance of the octoploids cytotype distribution in G.  communis reflects histor - elsewhere in the Mediterranean basin (Susnik and Lovka ical patterns of migration and colonization, and further 1973; Strid and Franzen 1981; van Raamsdonk and de selection against minority cytotype, and does not result Vries 1989; Kamari et  al. 2001). Interestingly, the vari- from different environmental requirements, creating a ation in monoploid genome size within G.  communis tension zone of contact. Still, in areas of contact, repro- cytotypes was very low and differed significantly from ductive barriers might mediate assortative mating and that of G. italicus (~9 %). Given the magnitude of the dif- enable cytotype coexistence. Nevertheless, the high ferences between G.  italicus and G.  communis, both in cytogenetic diversity detected in the field suggests that ploidy levels and in monoploid genome sizes, holoploid unreduced gamete formation and hybridization events genome size might be an important tool to detect hybrid- seem frequent in this complex and might be involved ization (e.g. Kolář et al. 2009; Agudo-García 2017). In our with recurrent polyploid formation and with gene flow study, G.  italicus and G.  communis were found growing between cytogenetic entities. Future studies involving in sympatry in 13 % of localities; however, all the G. itali- reciprocal transplants will provide significant insights cus individuals were duodecaploid. In most of the cases, into the dynamics of this polyploid complex. the duodecaploid G.  italicus was found growing with the octoploid G. communis (12 out of 14 localities); still, Sources of Funding no decaploids were observed in these localities. When growing with the tetraploid G.  communis, no octoploid This research was supported by POPH/FSE funds by the individuals with lower genome size resulting from the Portuguese Foundation for Science and Technology (FCT) hybridization between the two species (~5.00 pg based with a doctoral grant to M.C. (SFRH/BD/89910/2012) on the monoploid genome sizes of each species) were and a starting grant and exploratory project to S.C. observed. Although hybridization has been suggested to (IF/01267/2013), and by Project RENATURE financed by occur in these and in other Gladiolus species (e.g. van the ‘Programa Operacional Regional do Centro 2014–2020 Raamsdonk and de Vries 1989; Mifsud and Hamilton (Centro2020) - CENTRO-01-0145-FEDER-000007’. 2013; Szczepaniak et  al. 2016), we were not able to detect hybrids between G.  italicus and G.  communis. Contributions by the Authors This suggests that, in the studied range, hybridization S.C., B.H.  and J.L.  designed the research experiment; between them might be less common, either because of assortative mating or hybrid offspring inviability. M.C., S.C.  and J.L.  developed the field and laboratory Monoploid genome size also suggests a close relation- work; M.C.  performed the statistical analyses; and ship between the cytotypes of G. communis, pointing to M.C.  and A.F.  performed the niche modelling analyses; an autopolyploid origin of the complex in the studied all the authors were involved in the discussion of the area. This is also supported by the high morphological results and writing of the manuscript, approving the resemblance between G.  communis cytotypes (Alonso final document. and Crespo 2010; Cantor and Tolety 2011) and by the lack of evidence supporting hybridization between Conflict of Interest G. communis and G. italicus in this region. Still, the origin of G. communis polyploid complex needs to be properly None declared. evaluated in future studies. Acknowledgements Conclusions The authors are thankful to several researchers that have assisted in field collections, in particular to J. Costa, In this study, we find a complex cytogeographical pat - tern in G.  communis, which opens several hypotheses L.  Mota, D.  Tavares, A.  Martins, J.  M. Costa, C.  Silva and that might explain the formation and maintenance of M.  C. Duarte. We are also thankful to A.  Caperta and L. Morais for all the helpful methodological suggestions its tetraploid–octoploid contact zone. According to our results, tetraploids and octoploids do not differ in their on chromosome count techniques, and to two anony- environmental requirements, potentially growing in simi- mous reviewers for their insightful comments on previ- ous versions of the manuscript. lar habitats. Without differences in habitat requirements, AoB PLANTS https://academic.oup.com/aobpla © The Author(s) 2018 15 Downloaded from https://academic.oup.com/aobpla/article-abstract/10/2/ply012/4857208 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Castro et al. – Cytogeographical patterns in a tetraploid–octoploid complex Buggs RJ, Pannell JR. 2007. Ecological differentiation and diploid Supporting Information superiority across a moving ploidy contact zone. Evolution The following additional information is available in the 61:125–140. online version of this article— Burton TL, Husband BC. 2001. Fecundity and offspring ploidy in mat- ings among diploid, triploid and tetraploid Chamerion angustifo- Figure S1. Flow cytometry graphics analysed for each lium (Onagraceae): consequences for tetraploid establishment. sample. Heredity 87:573–582. Table S1. Geographic information of sampled Gladiolus Cantor M, Tolety J. 2011. Gladiolus In: Kole C. eds. Wild crop relatives: populations. genomic and breeding resources - Plantation and ornamental Table S2. Genome size variation in Gladiolus communis. crops. Berlin: Springer-Verlag Berlin Heidelberg, 133–160. Table S3. Mixed-ploidy populations of Gladiolus Castro S, Castro M, Ferrero V, Costa J, Tavares D, Navarro L, Loureiro communis. J. 2016a. Invasion fosters change: independent evolutionary Table S4. DNA ploidy levels of the offspring of pure- and shifts in reproductive traits after Oxalis pes-caprae L. introduc- mixed-ploidy populations of Gladiolus communis. tion. Frontiers in Plant Science 7:874. Castro M, Loureiro J, Husband BC, Castro S. 2016b. Can we live together? The role of reproductive barriers mediating cytotype Literature Cited coexistence in Gladiolus communis contact zones. XII Encontro Agudo-García A. 2017. Evolution in anacyclus L.  (Anthemideae, Nacional de Biologia Evolutiva, 16 December, University of Aveiro. Asteraceae). Analysis of the contact zone between A.  clava- Castro S, Loureiro J, Procházka T, Münzbergová Z. 2012. Cytotype tus (Desf.) Pers. and A.  valentinus L. PhD Thesis, Universidad distribution at a diploid-hexaploid contact zone in Aster amellus Autonoma de Madrid, Spain. (Asteraceae). Annals of Botany 110:1047–1055. Alonso MÁ, Crespo MB. 2010. Gladiolus L. In: Crespo MB, Herrero Castro S, Münzbergová Z, Raabová J, Loureiro J. 2011. Breeding A, Quintanar A, eds. Flora iberica. Madrid: Real Jardín Botánico, barriers at a diploid–hexaploid contact zone in Aster amellus. CSIC, 185:485–491. Evolutionary Ecology 25:795–814. Araújo MB, New M. 2007. Ensemble forecasting of species distribu- Darlington CD, Wylie AP. 1955. Chromosome atlas of flowering tions. Trends in Ecology & Evolution 22:42–47. plants. London, UK: George Allen and Unwin Ltd. Baack EJ. 2004. Cytotype segregation on regional and micro- Doležel J, Greilhuber J, Lucretti S, Meister A, Lysák MA, Nardi L, geographic scales in snow buttercups (Ranunculus adoneus: Obermayer R. 1998. Plant genome size estimation by flow Ranunculaceae). American Journal of Botany 91:1783–1788. cytometry: inter-laboratory comparison. Annals of Botany Baack EJ. 2005. To succeed globally, disperse locally: effects of local 82(suppl_1):17–26. pollen and seed dispersal on tetraploid establishment. Heredity Doležel J, Sgorbati S, Lucretti S. 1992. Comparison of three DNA 94:538–546. fluorochromes for flow cytometric estimation of nuclear DNA Baack EJ, Stanton ML. 2005. Ecological factors influencing tetra - content in plants. Physiologia Plantarum 85:625–631. ploid speciation in snow buttercups (Ranunculus adoneus): Duchoslav M, Safárová L, Krahulec F. 2010. Complex distribution niche differentiation and tetraploid establishment. Evolution patterns, ecology and coexistence of ploidy levels of Allium 59:1936–1944. oleraceum (Alliaceae) in the Czech Republic. Annals of Botany Bates D, Maechler M, Bolker B, Walker S. 2014. lme4: linear mixed- 105:719–735. effects models using Eigen and S4. http://CRAN.R-project.org/ Felber F. 1991. Establishment of a tetraploid cytotype in a diploid package=lme4 (21 April 2016). population: effect of relative fitness of the cytotypes. Journal of Bolker BM, Brooks ME, Clark CJ, Geange SW, Poulsen JR, Stevens MH, Evolutionary Biology 4:195–207. White JS. 2009. Generalized linear mixed models: a practical Fernandes A. 1950. Sobre a cariologia de algumas plantas da Serra guide for ecology and evolution. Trends in Ecology & Evolution do Gerês. Agronomia Lusitana 12:551–600. 24:127–135. Fernandes A, Garcia J, Fernandes R. 1948. Herborizações nos Bretagnolle FA, Thompson JD. 1995. Gametes with the somatic domínios da fundação da Casa de Bragança. I-Vendas Novas. chromosome number: mechanisms of their formation and Memórias da Sociedade Broteriana 4:5–89. role in the evolution of autopolyploid plants. New Phytologist Fernandes A, Queirós M. 1971. Sur la caryologie de quelques plan- ème 129:1–22. tes récoltéess pendant la III réunion de botanique péninsu- Brochmann C, Brysting AK, Alsos IG, Borgen L, Grundt HH, Scheen lare. Memórias da Sociedade Broteriana 21:343–385. AC, Elven R. 2004. Polyploidy in arctic plants. Biological Journal Fernández I, Díez MJ, Pastor J. 1985. Números cromosómicos para of the Linnean Society 82:521–536. la flora española, 373–381. Lagascalia 13:299–302. Broennimann O, Fitzpatrick MC, Pearman PB, Petitpierre B, Pellissier Fox J, Weisberg S, Adler D, Bates D, Baud-Bovy G, Ellison S. 2015. L, Yoccoz NG, Thuiller W, Fortin MJ, Randin C, Zimmermann NE, car: companion to applied regression. http://CRAN.R-project. Graham CH. 2012. Measuring ecological niche overlap from org/package=car. (21 April 2016). occurrence and spatial environmental data. Global Ecology and Galbraith DW, Harkins KR, Maddox JM, Ayres NM, Sharma DP, Biogeography 21:481–497. Firoozabady E. 1983. Rapid flow cytometric analysis of the cell Buchanan A. 2008. The taxonomic status of Gladiolus illyricus cycle in intact plant tissues. Science 220:1049–1051. (Iridaceae) in Britain. Doctoral dissertation, Department of Life Glennon KL, Rissler LJ, Church SA. 2012. Ecogeographic isolation: a Sciences and the Natural History Museum, Imperial College reproductive barrier between species and between cytotypes in London, London. Houstonia (Rubiaceae). Evolutionary Ecology 26:909–926. 16 AoB PLANTS https://academic.oup.com/aobpla © The Author(s) 2018 Downloaded from https://academic.oup.com/aobpla/article-abstract/10/2/ply012/4857208 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Castro et al. – Cytogeographical patterns in a tetraploid–octoploid complex Godsoe W, Larson MA, Glennon KL, Segraves KA. 2013. Jiao Y, Wickett NJ, Ayyampalayam S, Chanderbali AS, Landherr L, Polyploidization in Heuchera cylindrica (Saxifragaceae) did not Ralph PE, Tomsho LP, Hu Y, Liang H, Soltis PS, Soltis DE, Clifton result in a shift in climatic requirements. American Journal of SW, Schlarbaum SE, Schuster SC, Ma H, Leebens-Mack J, Botany 100:496–508. dePamphilis CW. 2011. Ancestral polyploidy in seed plants and Goldblatt P, Takei M. 1993. Chromosome cytology of the African angiosperms. Nature 473:97–100. genus Lapeirousia (Iridaceae-Ixioideae). Annals of the Missouri Kao RH. 2007. Asexuality and the coexistence of cytotypes. The Botanical Garden 80:961–973. New Phytologist 175:764–772. Greilhuber J, Dolezel J, Lysák MA, Bennett MD. 2005. The origin, Kamari G, Blanche C, Garbari F. 2001. Mediterranean chromosome evolution and proposed stabilization of the terms ‘genome number reports: 11. Flora Mediterranea 11:435–483. size’ and ‘C-value’ to describe nuclear DNA contents. Annals of Kennedy BF, Sabara HA, Haydon D, Husband BC. 2006. Pollinator- Botany 95:255–260. mediated assortative mating in mixed ploidy populations Greilhuber J, Ebert I. 1994. Genome size variation in Pisum sativum. of Chamerion angustifolium (Onagraceae). Oecologia 150: Genome 37:646–655. 398–408. Greilhuber J, Temsch EM, Loureiro J. 2007. Nuclear DNA content Kolář F, Stech M, Trávnícek P, Rauchová J, Urfus T, Vít P, Kubesová measurement. In: Dolezel J, Greilhuber J, Suda J, eds. Flow M, Suda J. 2009. Towards resolving the Knautia arvensis agg. cytometry with plant cells: analysis of genes, chromosomes and (Dipsacaceae) puzzle: primary and secondary contact zones genomes. Weinheim: Wiley-VCH, 67–101. and ploidy segregation at landscape and microgeographic Gussone G. 1832. Supplementum ad Florae Siculae Prodromum, vol. scales. Annals of Botany 103:963–974. 1. Regia Typographia, Neapoli. Kron P, Suda J, Husband BC. 2007. Applications of flow cytometry to Hamilton AP. 1980. Gladiolus L. Flora Europaea 5:101–102. evolutionary and population biology. Annual Review of Ecology, Hijmans RJ, van Etten J, Cheng J, Mattiuzzi M, Sumner M, Evolution, and Systematics 38:847–876. Greenberg JA, Lamigueiro OP, Bevan A, Racine EB, Shortridge Laport RG, Minckley RL, Ramsey J. 2016. Ecological distributions, A, Ghosh A, 2017. Package ‘raster’. https://cran.r-project.org/ phenological isolation, and genetic structure in sympatric and package=raster. (21 May 2017). parapatric populations of the Larrea tridentata polyploid com- Hijmans RJ, Phillips S, Leathwick J, Elith J. 2015. dismo: species distribu- plex. American Journal of Botany 103:1358–1374. tion modeling. R package version 1.0-12. Vienna: The R Foundation Levin DA. 1975. Minority cytotype exclusion in local plant popula- for Statistical Computing. http://cran.r-project.org. (21 May 2017). tions. Taxon 24:35–43. Hothorn T, Bretz F, Westfall P, Heiberger RM. 2017. Multcomp: sim- Levin DA. 2002. The role of chromosomal change in plant evolution. ultaneous inference for general linear hypotheses. http://CRAN. Oxford: Oxford University Press. Rproject.org/package=multcomp. (21 April 2016). Lexer C, van Loo M. 2006. Contact zones: natural labs for studying Husband BC. 2000. Constraints on polyploid evolution: a test of the evolutionary transitions. Current Biology 16:R407–R409. minority cytotype exclusion principle. Proceedings of the Royal Loureiro J, Rodriguez E, Dolezel J, Santos C. 2007. Two new nuclear Society of London B: Biological Sciences 267:217–223. isolation buffers for plant DNA flow cytometry: a test with 37 Husband BC. 2004. The role of triploid hybrids in the evolutionary species. Annals of Botany 100:875–888. dynamics of mixed-ploidy populations. Biological Journal of the Löve A, Kjellqvist E. 1973. Cytotaxonomy of Spanish plants. II. Linnean Society 82:537–546. Monocotyledons. Lagascalia 7:147–182. Husband BC, Baldwin SJ, Sabara HA. 2016. Direct vs. indirect Maceira NO, Haan AD, Lumaret R, Billon M, Delay J. 1992. Production effects of whole-genome duplication on prezygotic isolation of 2n gametes in diploid subspecies of Dactylis glomerata in Chamerion angustifolium: implications for rapid speciation. L.  1. Occurrence and frequency of 2n pollen. Annals of Botany American Journal of Botany 103:1259–1271. 69:335–343. Husband BC, Baldwin SJ, Suda J. 2013. The incidence of polyploidy Mallet J, Meyer A, Nosil P, Feder JL. 2009. Space, sympatry and spe- in natural plant populations: major patterns and evolutionary ciation. Journal of Evolutionary Biology 22:2332–2341. processes. In: J. Greilhuber, J. Dolezel, and J. F. Wendel eds. Marques I, Loureiro J, Draper D, Castro M, Castro S. 2017. How much Plant genome diversity. Vienna: Springer Vienna, 255–276. do we know about the frequency of hybridization and polyploidy in Husband BC, Sabara HA. 2004. Reproductive isolation between the mediterranean region? Plant Biology. doi:10.1111/plb.12639. autotetraploids and their diploid progenitors in fireweed, Mifsud S, Hamilton AP. 2013. Preliminary observations from long- Chamerion angustifolium (Onagraceae): research review. New term studies of Gladiolus L. (Iridaceae) for the Maltese Islands. Phytologist 161:703–713. Webbia 68:51–56. Husband BC, Schemske DW. 1998. Cytotype distribution at a diploid- Monteiro-Henriques T, Martins MJ, Cerdeira JO, Silva PC, Arsénio P, tetraploid contact zone in Chamerion (Epilobium) angustifolium Silva Á, Bellu A, Costa JC. 2016. Bioclimatological mapping tack- (Onagraceae). American Journal of Botany 85:1688–1694. ling uncertainty propagation: application to mainland Portugal. Husband BC, Schemske DW. 2000. Ecological mechanisms of repro- International Journal of Climatology 36:400–411. ductive isolation between diploid and tetraploid Chamerion Muñoz‐Pajares AJ, Perfectti F, Loureiro J, Abdelaziz M, Biella P, Castro angustifolium. Journal of Ecology 88:689–701. M, Castro S, Gómez JM. 2017. Niche differences may explain the Jersáková J, Castro S, Sonk N, Milchreit K, Schödelbauerová I, geographic distribution of cytotypes in Erysimum mediohispani- Tolasch T, Dötterl S. 2010. Absence of pollinator-mediated pre- cum. Plant Biology 20(Suppl. 1):139–147. mating barriers in mixed-ploidy populations of Gymnadenia Münzbergová Z, Surinová M, Castro S. 2013. Absence of gene flow conopsea sl (Orchidaceae). Evolutionary Ecology 24: between diploids and hexaploids of Aster amellus at multiple 1199–1218. spatial scales. Heredity 110:123–130. AoB PLANTS https://academic.oup.com/aobpla © The Author(s) 2018 17 Downloaded from https://academic.oup.com/aobpla/article-abstract/10/2/ply012/4857208 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Castro et al. – Cytogeographical patterns in a tetraploid–octoploid complex Nilsson O, Lassen P. 1971. Chromosome numbers of vascular Dactylorhiza maculata s.l. (Orchidaceae). Evolutionary Ecology plants from Austria, Mallorca and Yugoslavia. Botaniska Notiser 23:295. 124:270–276. Strid A, Franzen R. 1981. In Chromosome number reports LXXIII. Otto SP, Whitton J. 2000. Polyploid incidence and evolution. Annual Taxon 30:829–842. Review of Genetics 34:401–437. Suda J, Kron P, Husband BC, Trávníček P. 2007. Flow cytometry Pannell JR, Obbard DJ, Buggs RJ. 2004. Polyploidy and the sexual and ploidy: applications in plant systematics, ecology and evo- system: what can we learn from Mercurialis annua? Biological lutionary biology. In: Doležel J, Greilhuber J, Suda J, eds. Flow Journal of the Linnean Society 82:547–560. cytometry with plant cells: analysis of genes, chromosomes and Pérez E, Pastor Díaz JE. 1994. Contribución al estudio cariológico genomes. Weinheim: Wiley-VCH, 103–130. de la familia Iridaceae en Andalucía Occidental. Lagascalia Suda J, Oberlander KC, Dreyer LL. 2013. Two new species of Oxalis 17:257–272. (Oxalidaceae) from the Greater Cape Floristic Region. Phytotaxa Petit C, Bretagnolle F, Felber F. 1999. Evolutionary consequences of 124:13–24. diploid-polyploid hybrid zones in wild species. Trends in Ecology Susnik F, Lovka EH. 1973. Reports. IOPB Chromosome numbers & Evolution 14:306–311. reports XLI. Taxon 22:462–463. Phillips SJ. 2008. Transferability, sample selection bias, and back- Szczepaniak M, Kamiński R, Kuta E, Słomka A, Heise W, Cieślak E. ground data in presence-only modelling: a response to Peterson 2016. Natural hybridization between Gladiolus palustris and et al. (2007). Ecography 31:272–278. G. imbricatus inferred from morphological, molecular and repro- Phillips SJ, Anderson RP, Schapire RE. 2006. Maximum entropy mod- ductive evidence. Preslia 88:137–161. eling of species geographic distributions. Ecological Modelling Thompson KA, Husband BC, Maherali H. 2014. Climatic niche differ- 190:231–259. ences between diploid and tetraploid cytotypes of Chamerion Queirós M. 1979. Números cromossómicos para a flora Portuguesa: angustifolium (Onagraceae). American Journal of Botany 16–37. Boletim da Sociedade Broteriana Série 2 53:15–28. 101:1868–1875. Queirós M. 1980. Números cromossómicos para a flora portuguesa: Thuiller W, Georges D, Engler R, Breiner F, Georges MD, Thuiller 38–63. Boletim da Sociedade Broteriana Série 2 54:47–64. CW. 2016. Package ‘biomod2’. https://cran.r-project.org/ R Core Development Team. 2016. A language and environment package=biomod2. for statistical computing. Vienna: R Foundation for Statistical Trávníček P, Kubátová B, Curn V, Rauchová J, Krajníková E, Jersáková Computing. http://www.R-project.org/. (21 April 2016). J, Suda J. 2010. Remarkable coexistence of multiple cytotypes Ramsey J. 2007. Unreduced gametes and neopolyploids in natural of the Gymnadenia conopsea aggregate (the fragrant orchid): populations of Achillea borealis (Asteraceae). Heredity 98:143–150. evidence from flow cytometry. Annals of Botany 107:77–87. Ramsey J, Ramsey TS. 2014. Ecological studies of polyploidy in the Van Raamsdonk LWD, De Vries T. 1989. Biosystematic studies in 100 years following its discovery. Philosophical Transactions of the European species of Gladiolus (Iridaceae). Plant Systematics and Royal Society B: Biological Sciences. doi:10.1098/rstb.2013.0352. Evolution 165:189–198. Ramsey J, Schemske DW. 1998. Pathways, mechanisms, and rates Visger CJ, Germain-Aubrey CC, Patel M, Sessa EB, Soltis PS, Soltis of polyploid formation in flowering plants. Annual Review of DE. 2016. Niche divergence between diploid and autotetraploid Ecology and Systematics 29:467–501. Tolmiea. American Journal of Botany 103:1396–1406. Rodriguez DJ. 1996. A model for the establishment of polyploidy in Warren DL, Glor RE, Turelli M. 2008. Environmental niche equiva- plants. The American Naturalist 147:33–46. lency versus conservatism: quantitative approaches to niche Schoener TW. 1970. Nonsynchronous spatial overlap of lizards in evolution. Evolution 62:2868–2883. patchy habitats. Ecology 51:408–418. Wefferling KM, Castro S, Loureiro J, Castro M, Tavares D, Hoot SB. Segraves KA, Thompson JN. 1999. Plant polyploidy and pollination: 2017. Cytogeography of the subalpine marsh marigold polyploid floral traits and insect visits to diploid and tetraploid Heuchera complex (Caltha leptosepala s.l., Ranunculaceae). American grossulariifolia. Evolution 53:1114–1127. Journal of Botany 104:271–285. Soltis DE, Albert VA, Leebens-Mack J, Bell CD, Paterson AH, Zheng C, Wood TE, Takebayashi N, Barker MS, Mayrose I, Greenspoon PB, Sankoff D, Depamphilis CW, Wall PK, Soltis PS. 2009. Polyploidy and Rieseberg LH. 2009. The frequency of polyploid speciation angiosperm diversification. American Journal of Botany 96:336–348. in vascular plants. Proceedings of the National Academy of Soltis DE, Buggs RJ, Doyle JJ, Soltis PS. 2010. What we still don’t Sciences 106:13875–13879. know about polyploidy. Taxon 59:1387–1403. Zozomová-Lihová J, Malánová-Krásná I, Vít P, Urfus T, Senko D, Soltis DE, Soltis PS. 1999. Polyploidy: recurrent formation and gen- Svitok M, Kempa M, Marhold K. 2015. Cytotype distribution ome evolution. Trends in Ecology & Evolution 14:348–352. patterns, ecological differentiation, and genetic structure in a Ståhlberg D. 2009. Habitat differentiation, hybridization and gene diploid-tetraploid contact zone of Cardamine amara. American flow patterns in mixed populations of diploid and autotetraploid Journal of Botany 102:1380–1395. 18 AoB PLANTS https://academic.oup.com/aobpla © The Author(s) 2018 Downloaded from https://academic.oup.com/aobpla/article-abstract/10/2/ply012/4857208 by Ed 'DeepDyve' Gillespie user on 16 March 2018

Journal

AoB PlantsOxford University Press

Published: Mar 1, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off