Abstract Distyly is a ﬂoral polymorphism characterized by the presence of two discrete morphs with reciprocal positioning of anthers and stigmas in ﬂowers on different plants in the same population. For a distylous system to function effectively, the presence of suitable pollinators that transfer pollen from the two morphs (a short-styled S-morph and a long-styled L-morph) on separate locations of their body is required. In this study, we surveyed pollinator communities visiting flowers of the distylous Primula veris in two different natural habitats (grassland and forest). Previous research has shown differences in the positioning of the anthers and stigmas between flowers of forest and grassland populations, possibly impacting on patterns of pollen transfer and seed set. To test this hypothesis, we measured anther–stigma separation in the two habitats, assessed pollen uptake on the head and proboscis of each of the pollinator species observed, and compared stigmatic pollen deposition and subsequent seed set between short-styled and long-styled flowers of P. veris. More specifically, we tested the hypothesis that the ratio of pollen morph uptake by pollinators is related to the length of their proboscis and to differences in floral morphology and that this in turn is related to the efficiency with which legitimate pollen is deposited on S- and L-morph stigmas. The forest and grassland P. veris populations contained distinct pollinator communities. Pollen from L-morph flowers was more abundantly deposited on the proboscis than on the head of the pollinator, whereas the opposite was observed for pollen of short-styled flowers. Tongue length seemed to be a determining characteristic for the predisposition of a pollinator for pollen uptake of a certain pollen morph on the proboscis or head. Proboscis length was positively correlated with proportional uptake of pollen of long-styled flowers, but negatively correlated with the uptake of pollen of short-styled flowers on the head. Long-styled stigmas captured more pollen grains in total, but short-styled stigmas contained proportionally more legitimate pollen. Pollen proficiencies were higher in the grassland habitat, but seed set did not significantly differ between habitats. Overall, these results suggest that that long- and short-tongued insects complement each other in the legitimate pollination of a distylous plant species and that differences in floral morphology do not impact on reproductive success. INTRODUCTION Heterostyly is a floral polymorphism in which flowers exhibit spatial separation in anther–stigma positioning (Darwin, 1877). It has evolved independently in 28 different plant families, most probably as a strategy to promote cross-pollination and to avoid self-interference (Darwin, 1877; Lloyd & Webb, 1992; Stone & Thomson, 1994; Pérez-Barrales, Vargas & Arroyo, 2006). The reciprocal positioning of stigmas and anthers limits interference of pollen removal with pollen deposition within the same flower and morph type and promotes precise intermorph cross-pollen transfer (i.e. disassortative pollen transfer) (Darwin, 1877; Ganders, 1979; Barrett, 1990). Besides the reciprocal positioning of stigmas and anthers, heterostyly is also often accompanied by a sporophytically controlled, diallelic self-incompatibility system (Barrett & Shore, 2008; Brys & Jacquemyn, 2015) and a suite of ancillary polymorphisms related to specific characteristics of pollen and stigmas. In general, the self-incompatibility system precludes both self- and intramorph cross-fertilization and therefore only allows intermorph cross-fertilizations (Barrett, 1990; Kohn & Barrett, 1992), whereas morph-specific differences in pollen grain size and in the size and shape of the stigmatic papillae function as a lock and key mechanism to prevent illegitimate pollination and promote legitimate intermorph pollen capture. These physiological and mechanical barriers further safeguard against self-fertilization and the deleterious effects of inbreeding (Ganders, 1979) and thus maximize heterozygosity in the offspring (Darwin, 1877; Barrett & Shore, 2008). A prerequisite for the heterostylous syndrome to function effectively is that transfer of pollen between individuals with a different morph exceeds that of pollen between individuals of the same morph (Ganders, 1979; Lloyd & Webb, 1992). Ideally, this is realized when pollinators pick up pollen on different parts of their body coinciding with the placement of the anthers in the flower (Rosov & Screbtsova, 1958; Ganders, 1979; Stone & Thomson, 1994; Alexandersson & Johnson, 2002; Brys et al., 2008). The efficiency of reciprocal pollination (i.e. legitimate pollination) can thus be affected by several factors such as variation in the positioning of stigmas and anthers (Arroyo & Barrett, 2000), intermorph differences in pollen production (Ganders, 1979) and morph-specific differences in the accessibility of the anthers and/or stigmas to pollinators (Lau & Bosque, 2003). Distylous plants thus require pollinators that are able to obtain and deposit pollen at the two levels where the sexual organs are placed reciprocally inside the flower. Specifically, the proboscis has to be sufficiently long to reach the bottom of the corolla to collect nectar and passively take up pollen in a long-styled flower (Brys et al., 2008; Keller, Thomson & Conti, 2014; Costa et al., 2017). However, most heterostylous plant species are not pollinated by a single insect species, but rather by a broad array of insect species that may vary in numerous aspects and thus potentially influence the way pollen is collected on their body and are deposited on stigmas of the legitimate morph type. For example, flowers of the distylous Primula vulgaris Huds. are visited by at least 50 different insect species, including bees, bumblebees, butterflies and beetles (Jacquemyn et al., 2009). It is thus reasonable to assume that not all foraging insect species function as effective pollinators in heterostylous plant species. For example, research on the distylous Pulmonaria vulgaris Mérat has shown that its flowers are visited by several insect species, mainly bumblebees and bees, but that only the long-tongued Anthophora plumipes (proboscis length 14.4 mm) carried proportionally more pollen of short-styled flowers on its head and more pollen of long-styled flowers on its proboscis (Brys et al., 2008). In contrast, the shorter tongued bumblebees (Bombus terrestris, B. pascorum, B. pratorum) did not show disassortative pollen pick-up, which was related to the inadequate length of their proboscis. Similarly, A. plumipes stratified pollen from short-styled and long-styled flowers on different parts of its proboscis and transferred pollen reciprocally between floral morphs of Primula elatior Hill and P. vulgaris (Keller et al., 2014). Given that pollinator assemblages are known to vary across plant populations and environmental features (Potts et al., 2003; Zhu et al., 2015), pollen uptake and pollination efficiency can therefore vary between populations according to the presence or absence of effective pollinators or changes in their foraging behaviour and therefore can have a strong impact on legitimate pollen flow patterns and subsequent seed set (Wolfe & Barrett, 1989; Lloyd & Webb, 1992; Arroyo & Dafni, 1995; Brys et al., 2008; Meeus, Honnay & Jacquemyn, 2013; Zhu et al., 2015). In this study, we assessed variation in pollinator community composition in natural populations of the distylous Primula veris L. in southern Belgium. Previous research has shown that in this area the species can be found in two distinct habitats (forest and grasslands) (Deschepper et al., 2017) and that the positioning of anthers and stigmas differed between these habitats, with a clear disruption of distyly in the long-styled flower morph in forest populations and no such differences in anther–stigma positioning in grassland populations (Brys & Jacquemyn, 2015). We hypothesize that the ratio of pollen morph deposition on the frontal body parts (proboscis and head) by different pollinators is correlated with their proboscis length and the positioning of anthers and stigmas, and that this in turn is related to the efficiency with which legitimate pollen is deposited on S- and L-morph stigmas. Specifically, long-tongued pollinators are hypothesized to accumulate more pollen of long-styled flowers, especially on their tongue, whereas short-tongued insects may be more likely to have more pollen of short-styled flowers on their frontal body parts. Additionally, we expect that the disruption of clear distyly in flowers of the long-styled morph in forest populations has further impacts on patterns of pollen transfer within forest populations, ultimately translating into significant differences in seed set between the two habitats in which P. veris occurs. MATERIAL AND METHODS Study species Primula veris (cowslip) is a herbaceous, spring- flowering perennial that usually grows in calcareous grasslands, but can also be present in old-growth forest and hedgerows (Brys & Jacquemyn, 2009). This rosette-forming hemicryptophyte can be found throughout most of temperate Europe and Britain and Ireland to the western Russian border (Brys & Jacquemyn, 2009). In early spring, P. veris forms a rosette of leaves and produces flowers that grow in umbels. Flowering occurs about 1 month earlier in forests than in grasslands and lasts for 3–4 weeks (Brys & Jacquemyn, 2015). Selfing is prevented by a diallelic self-incompatibility system in combination with the floral polymorphism, with two reciprocal flower morphs (a long-styled pin morph and a short-styled thrum morph) (Van Rossum, De Sousa & Triest, 2006; Brys & Jacquemyn, 2015). Flowers produce copious nectar and attract a variety of insects (Brys & Jacquemyn, 2009). Generally, pollination efficiency is assumed to be most successful for long-tongued pollinators that are able to reach and consume the nectar that is exposed at the bottom of the narrow and deep corolla tube (Proctor & Yeo, 1973). Pollen of S-morph plants (on average 29.5 µm) is larger than pollen of L-morph plants (on average 18.1 µm), which makes discrimination between pollen coming from the two morphs possible (Brys & Jacquemyn, 2015). Additionally, stigmatic surfaces of the two morphs differ in the size of their papillae with short-styled stigmas having papillae two to three times shorter than those of long-styled stigmas, giving them a smoother surface (Darwin, 1862; Brys & Jacquemyn, 2009). Study area and populations This research was conducted in the Calestienne region, which is located in south-western Belgium in the Namur province. This area is characterized by the occurrence of Devonian limestone hills, which were historically covered by large stretches of calcareous grasslands, and deciduous forests on limestone. Elevation in the study area ranges from 150 to 250 m a.s.l. (Adriaens, Honnay & Hermy, 2006). For this research, four study populations were selected that occurred in forest patches or calcareous grasslands that were located in the valley of the river Viroin. The forest populations were located south of Dourbes in Le Franc forest and north of Treignes in the forest of Matignolle and under closed forest canopy consisting of oak (Quercus robur L.) and hornbeam (Carpinus betulus L.). These forests generally show strong affinities with the Galio–Carpinetum association. Grassland populations were located on south-facing slopes of the Viroin River east of Dourbes and east of Mazée, with no forests within 200 m. Vegetation in calcareous grasslands is characterized by short and herbaceous flowering plants adapted to dry, basic soils (Butaye et al., 2005). Calcareous grasslands of the Calestienne region belong to the Festuco–Brometea phytosociological group, but show affinities with other classes, such as Sedo-Scleranthetea, Trifolio-Geranietea and Molinio-Arrhenatheretea (Butaye et al., 2005; Adriaens et al., 2006). Vouchers for specimens collected at the grassland and forest site used for assessment of pollen deposition were deposited in the national herbarium of the Botanic Garden Meise (Meise, Belgium), with the voucher number BR0000025296587 for the specimen from grassland habitat east of Dourbes and BR0000025296594 for the specimen from the forest of Matignolle. Field pollinator survey and quantification of pollen on proboscis and head Community composition of insects visiting P. veris was assessed in each of the four study populations during the peak of their flowering periods, which was the end of March and the end of April for the forest and grassland populations, respectively (Brys & Jacquemyn, 2015). Forest populations were visited on bright and sunny days, on the last day of March and the first day of April (2014), whereas the grassland populations were visited on sunny days at the end of April. On each day, observations of pollinating insects on flowers of P. veris were recorded for 12 20-min sessions, with a total of 4 h of observation per population. Six sessions took place before noon and six sessions after noon. As many pollinators as possible were captured and stored in plastic tubes containing ethyl acetate to asphyxiate the insect. Following identification of each pollinator, we measured the size and length of their head and proboscis, after which both body parts were dissected and separately stored in 1.5-mL Eppendorf tubes. The hairs on the head of the pollinator prevent detachment of pollen when dissecting the head. The proboscis needs to be handled with greater care because pollen does not adhere to this organ well. Next, we added 400 µL Tween and 100 µL lactophenol cotton blue in each tube to detach the pollen grains from the respective body parts and to stain the pollen grains. Each tube was then vortexed for 30 s to obtain a homogenized suspension of the pollen. Pollen deposition on each body part was assessed by taking three volumes of 20 µL and counting the number of pollen grains in each replicate under a microscope. Finally, the counts of those three replicates were averaged and multiplied by 25 to obtain a proxy of the total number of pollen grains in the sample. Pollen production, stigmatic pollen deposition and herkogamy in forest and grassland To assess pollen production per flower, ten flowers per morph and habitat were collected just before anthesis from each of the study populations. In the laboratory, anthers were dissected from the flower and placed in a micro-tube with 75 µL Tween solution and 75 µL lactophenol cotton blue. Each tube was then vortexed to achieve homogenization. From each sample we counted the total number of pollen grains in three subsamples of 20 µL under the microscope to obtain a proxy of total pollen production per flower. To assess stigmatic pollen deposition patterns in the grassland and forest populations, we collected 20 flowers from each morph type in each habitat (in total 40 flowers per habitat). Flowers were dissected longitudinally and anther and stigma height were measured using ImageJ software (Rasband, 2011) and the style was extracted. Herkogamy (i.e. stigma–style separation) was calculated by subtracting the height of the anthers from the stigma height (Ganders, 1979; Brys & Jacquemyn, 2015). Harvested styles were softened in 8 M HCl for 24 h, after which they were rinsed three times with tap water. Next, we added 400 µL Tween and 100 µL lactophenol cotton blue in 1.5-mL Eppendorf tubes to detach the pollen grains from the stigma and to stain the pollen grains. Each tube was then vortexed for 30 s to homogenize pollen in the solution. Pollen grains were counted in three volumes of 20 µL and the mean value was multiplied by 25. The number of flowers produced was also assessed for each plant of which we collected a style. We estimated the probability of a single pollen grain coming from a given flower morph being deposited on the stigma of a flower with the same morph as the pollen donor (illegitimate pollen transfer) and on the stigma of a flower with the opposite morph as the pollen donor (legitimate pollen transfer) by calculating pollen transfer proficiencies (Tij) following Lloyd & Webb (1992): Tij=(average stigma load)ij × (number of flowers)j(pollen production per flower)i × (number of flowers)i where i and j are the two floral morphs and (average stigma load)ij is the average number of pollen grains of morph i on the stigmatic surface of morph j. To investigate female reproductive success under natural conditions, we used the same 40 flowering plants (20 S-morph and 20 L-morph individuals) in each of the habitats. For each plant, we counted the total number of flowers and, when fruits were mature (end of June), three closed fruits per plant were harvested and brought to the laboratory. For each of these fruits, the total numbers of ovules and seeds were counted and averaged per plant. For each fruit, proportional seed set was calculated as the ratio of the number of developed ovules on the total number of viable ovules. Data analysis We used a generalized linear model (GLM) with a Poisson distribution of errors to investigate whether morph type, habitat or their interaction affected the number of flowers produced per individual. A multivariate analysis of variance (MANOVA) was used to test the hypothesis that the number of pollen of both the S- and the L-morph deposited on the proboscis and head of pollinators differed between pollinators. The amounts of pollen coming from short- and long-styled flowers were used as dependent variables and insect species and body part as predictor variables. This analysis allowed us to investigate whether the amounts of pollen of a particular morph differed between the proboscis and head. Additionally, with this analysis we could determine whether the amount of deposited pollen of each morph on the proboscis and head differed between insect species. A logistic regression was performed to test the relationship between proboscis length and the proportion of pollen of L-morph flowers deposited on this organ. Similarly, a logistic regression was performed to examine the relationship between proboscis length and proportion of pollen of short-styled flowers deposited on the head. The average proportion of legitimate and illegitimate pollen on the proboscis and head of the pollinators in the grassland and forest habitats was calculated and weighted respecting the abundance of each pollinator species in the two habitats. Staphilinae, Gonepteryx rhamni and Melanostoma mellinum were omitted from the analyses because they carried too few pollen grains (fewer than five pollen grains in total). Twenty-three and 18 pollinators in forest and grassland habitat, respectively, were used in the analyses. Finally, GLMs were used to investigate whether legitimate pollen deposition, pollen proficiency values and proportional seed set were affected by habitat type (grassland or forest), morph type (S- or L-morph) or their interaction (R Core Team, 2015). All analyses were conducted in R Studio v.1.0.136 (RStudio Team, 2016). RESULTS Pollinator communities in two contrasting habitats of Primula veris Pollinator community composition differed between forest and grassland populations (Fig. 1), with several pollinator species showing a strong preference for a certain habitat type. Anthophora plumipes was encountered six times and was strictly found in the grassland habitat where it was the principal visitor of flowers of P. veris. On the other hand, the butterfly Gonepteryx rhamni and small rove beetles (species of Staphylinae) were only found in the forest habitat (Fig. 1). The genus Bombus was represented by three species in the forest site (B. terrestris, B. lapidarius and B. pratorum), where it was the dominant group of pollinators, apart from the small rove beetles (Staphylinae). Bombus terrestris and B. lapidarius were also found in the grassland site with B. pascuorum. The last species was not present in the forest site. Forty-five and 19 individual pollinating insects were observed in forest and grassland habitats, respectively, during 24 20-min periods of observation per habitat. Figure 1. View largeDownload slide Abundance of insect species visiting flowers of Primula veris in forest and grassland habitats. Figure 1. View largeDownload slide Abundance of insect species visiting flowers of Primula veris in forest and grassland habitats. Pollen deposition on the proboscis and head of pollinators Pollen loads on the proboscis and head of the pollinators varied greatly across the different species (on average 25 ± 41 and 31 ± 29 pollen grains were found on the proboscis and head of the pollinator, respectively). Anthophora plumipes carried the largest amounts of pollen on the proboscis and head (93 ± 38 and 94 ± 38 pollen grains in total, respectively). Compared to the other observed pollinators, this species was also characterized by a long proboscis (14.4 mm); only Bombylius major had a longer proboscis (20.0 mm; Fig. 2). There was a significant positive relationship (P = 0.017, t = 3.09, d.f. = 7 with r = 0.76) between mean proboscis length of the different pollinators and the proportion of pollen from long-styled flowers on the proboscis (Fig. 3A). Proboscis length was negatively related to the proportion of pollen of short-styled flowers deposited on the head of the pollinator (P = 0.044, t = −2.45, d.f. = 7 with r = −0.70) (Fig. 3B). The MANOVA revealed that the amount of pollen of short- and long-styled flowers on the head and proboscis of pollinators was significantly determined by the pollinator species (P ≤ 0.01, F = 2.09, d.f. = 108). This analysis also showed that morph-specific pollen deposition was associated with the body part of the insect upon which they were present (head or proboscis) (P ≤ 0.01, F = 8.05, d.f. = 53), with pollen of short-styled flowers being significantly more likely to be deposited on the head than on the proboscis and vice versa for pollen of long-styled flowers. The average proportion of legitimate pollen on proboscis (pollen of long-styled flowers) and head (pollen of short-styled flowers) weighted with respect to pollinator species abundances in both grassland and forest habitat also exceeded that of illegitimate pollen in all cases (legitimate pollen on proboscis: 62.41 and 57.51 in grassland and forest, respectively; legitimate pollen on the head: 59.23 and 62.22 in grassland and forest, respectively) (Fig. 4A). Figure 2. View largeDownload slide Insect species visiting flowers of Primula veris sorted by the length of their proboscis. Pie charts next to the head and proboscis reflect the proportion of pollen of short-styled (green) and long-styled (orange) flowers on these body parts. The number next to the pie charts indicates the total average number of pollen on proboscis and head. An illustration of the average long-styled and short-styled form of P. veris shows the depth of the corolla tube in relation to the proboscis length of the depicted pollinators. Figure 2. View largeDownload slide Insect species visiting flowers of Primula veris sorted by the length of their proboscis. Pie charts next to the head and proboscis reflect the proportion of pollen of short-styled (green) and long-styled (orange) flowers on these body parts. The number next to the pie charts indicates the total average number of pollen on proboscis and head. An illustration of the average long-styled and short-styled form of P. veris shows the depth of the corolla tube in relation to the proboscis length of the depicted pollinators. Figure 3. View largeDownload slide (A) Relationship between the proportion of pollen of long-styled flowers on the proboscis and the average length of the proboscis across all pollinator species. (B) Relationship between the proportion of pollen of short-styled flowers on the head and the average length of the proboscis across all pollinator species. Each data point in Fig. 3A and Fig. 3B represents the relationship between pollen from one flower morph deposited on the proboscis and the proboscis length for one of the 9 pollinator species that are also used in Fig. 2. The line is a logarithmic trendline. Figure 3. View largeDownload slide (A) Relationship between the proportion of pollen of long-styled flowers on the proboscis and the average length of the proboscis across all pollinator species. (B) Relationship between the proportion of pollen of short-styled flowers on the head and the average length of the proboscis across all pollinator species. Each data point in Fig. 3A and Fig. 3B represents the relationship between pollen from one flower morph deposited on the proboscis and the proboscis length for one of the 9 pollinator species that are also used in Fig. 2. The line is a logarithmic trendline. Figure 4. View largeDownload slide Patterns of pollen deposition and seed set in Primula veris populations in grassland and forest habitats. (A) Proportion of legitimate and illegitimate pollen on the proboscis and head of a pollinator weighted with respect to their abundances in both habitats, i.e. the proportion of legitimate and illegitimate pollen on the proboscis and head of an average pollinator in the grassland and forest habitat. (B) Proportional number of legitimate and illegitimate pollen on both stylar morphs in both habitats with mean numbers above each bar. (C) Legitimate and illegitimate pollen proficiency for both stylar morphs in both habitats. (D) Proportional seed set per fruit for both stylar morphs in both habitats. Figure 4. View largeDownload slide Patterns of pollen deposition and seed set in Primula veris populations in grassland and forest habitats. (A) Proportion of legitimate and illegitimate pollen on the proboscis and head of a pollinator weighted with respect to their abundances in both habitats, i.e. the proportion of legitimate and illegitimate pollen on the proboscis and head of an average pollinator in the grassland and forest habitat. (B) Proportional number of legitimate and illegitimate pollen on both stylar morphs in both habitats with mean numbers above each bar. (C) Legitimate and illegitimate pollen proficiency for both stylar morphs in both habitats. (D) Proportional seed set per fruit for both stylar morphs in both habitats. Herkogamy, stigmatic pollen load, pollen proficiency and fruit set in two contrasting habitats of Primula veris An average of 8.13 ± 4.35 flowers were produced per plant. Flower production did not differ between habitats or morph type (P = 0.87, z = −0.17, d.f. = 78 and P = 1.0, z = 0.00, d.f. = 77). Total pollen production per flower varied significantly between habitats and morph type. L-morph flowers produced 114656 and 51000 pollen grains in the forest and grassland population, respectively, whereas S-morph flowers only produced 24375 and 11156 pollen grains for the forest and grassland habitat, respectively. In the grassland habitat, flowers showed clear herkogamy (−7.68 ± 0.25 and 7.43 ± 0.29 for the S- and L-morph, respectively). In the forest habitat, on the other hand, P. veris showed deviations in stigma–anther separation in L-morph flowers (herkogamy: –9.42 ± 0.38 and 4.82 ± 0.85 for the S- and L-morph, respectively). Significantly more pollen was deposited on stigmas of forest plants than on stigmas of grassland plants (P ≤ 0.01, z = −11.89, d.f. = 78). Total pollen deposition was significantly higher (P ≤ 0.01, z = −42.70, d.f. = 77) on the stigmas of long-styled flowers (332 ± 178 and 403 ± 229 pollen grains in the grassland and forest habitat, respectively) than on the stigmas of short-styled flowers (145 ± 77 and 168 ± 78 pollen grains captured for the grassland and forest habitat, respectively). Similar numbers of legitimate pollen were deposited in the grassland habitat, with on average 80 ± 53 and 79 ± 31 legitimate pollen grains deposited on the stigmas of S- and L-morph flowers, respectively. On average, 96 ± 64 and 64 ± 42 legitimate pollen grains were deposited on the stigmas of S- and L-morph flowers, respectively, in the forest habitat (Fig. 4B). The proportion of legitimate pollination was significantly higher on short-styled than on long-styled stigmas (0.59 ± 0.04 and 0.41 ± 0.04, respectively, P ≤ 0.001, t = 6.28, d.f. = 78) (Fig. 4B). Legitimate pollen proficiency was significantly higher in the grassland habitat than in the forest habitat (4.3 × 10–3 ± 2.8 × 10–4 and 1.8 × 10–3 ± 1.7 × 10–4, respectively, P ≤ 0.001, t = 6.12, d.f. = 76) and was higher for the L-morph than for the S-morph (4.9 × 10–3 ± 3.5 × 10–4 and 1.2 × 10–3 ± 1.9 × 10–4, respectively, P = 0.01, t = −2.65, d.f. = 77) (Fig. 4C). Proportional seed set, in terms of ovules developing into seeds, was significantly higher in the grassland population (P = 0.029, t = 2.23, d.f. = 78), but was not different between the flower morphs (P = 0.66, t = 0.44, d.f. = 77) (Fig. 4D). DISCUSSION Pollinator communities of Primula veris Although their abundance was generally low, a considerable number of insect species was observed visiting flowers in the four studied populations of P. veris. These results largely confirm previous results of Brys & Jacquemyn (2009), who showed that, based on the available literature, P. veris is pollinated by a large number of insect species. The low abundance of the observed pollinators may reflect their rarity in early spring (end of March), a time when only insects that are well adapted to the fluctuating spring temperatures can be active. The array of insects pollinating P. veris flowers was different between the two habitats in terms of pollinator species and abundances (Fig. 1). Whereas the two Halictus spp. and Bombus terrestris were the most dominant pollinators in forests, Anthophora plumipes was the most important pollinator in grasslands. The latter species was totally absent in forests (Fig. 1). This is not surprising because A. plumipes is a solitary bee that is known to visit more open areas such as grasslands, gardens and parks (Stone, 1994). Gonepteryx rhamni and Halictus sp. 2, on the other hand, were absent from grassland populations. Gonepteryx rhamni only has two natural host plants (Gutiérrez & Thomas, 2000) of which Frangula alnus Mill. (= Rhamnus frangula L.) is often encountered in or adjacent to old-growth forests. The observed insects differed substantially in the length of their proboscis, which varied between 4.0 and 20.0 mm. Given that the corolla tube of flowers of P. veris is on average 16 mm deep (Brys & Jacquemyn, 2015), this implies that several of these insects can be expected to be inappropriate for efficiently entering the flower and reaching the nectar with their proboscis alone. Pollen uptake by pollinators It is generally assumed that deep-probing insects are the most effective pollinators in distylous plant species that are characterized by tube-like flowers in which nectar is concealed deep in the flower (Lloyd & Webb, 1992; Simón-Porcar, Santos-Gally & Arroyo, 2014; Costa et al., 2017). Our results are in line with this as the absolute number and the proportion of legitimate and illegitimate pollen deposited on the head and proboscis of the observed pollinators differed significantly between insect species and depended on the length of their proboscis (Fig. 3). Moreover, the ratio of pollen coming from short- and long-styled flowers deposited on the proboscis and head of the pollinator largely correlated with the likelihood of contact between these body parts and lower and higher placed anthers. Overall, pollen from L-morph flowers was deposited proportionally more on the proboscis than on the head of the pollinator and vice versa for pollen from S-morph flowers (Fig. 4A). These results confirm earlier studies that have shown stratification of pollen coming from different morphs on the upper and lower parts of the proboscis of Bombus terrestris in the distylous Pulmonaria obscura Dumort. (Olesen, 1979) and on the tongue of Anthophora plumipes in two species of Primula L. (Keller et al., 2014). For the long-tongued A. plumipes our findings largely confirm these observations, but for the whole range of pollinators visiting P. veris we found that reciprocal pollen deposition on the heads and proboscises of the pollinators depended significantly on proboscis length (Fig. 2). Moreover, the proportional number of pollen from long-styled flowers deposited on the proboscis increased significantly with increasing proboscis length. This is in line with the findings of Brys et al., (2008) who studied hymenoptera with different proboscis lengths in a pollination study on the distylous Pulmonaria officinalis. They also showed that the long-tongued A. plumipes was the only pollinator for which the amount of pollen of long-styled flowers exceeded that of pollen of short-styled flowers on the proboscis when compared to shorter tongued pollinators. Since stigmas of short-styled flowers rely heavily on long-tongued pollinators, the hairy A. plumipes can therefore be considered as the most efficient pollinator to transfer pollen from anthers of long-styled flowers to styles of short-styled flowers. Indeed, A. plumipes exceeded every other pollinator species in the amount of pollen coming from long-styled flowers accumulated on its proboscis (Fig. 2). Only one observed pollinator species, Bombylius major, had a longer proboscis, but this species has a more slender and less hairy proboscis than A. plumipes, resulting in less pollen capture. The presence of A. plumipes could explain the slightly higher proficiency for S-morph flowers in the grassland habitat (Fig. 4C), because this species carries plenty of pollen coming from long-styled flowers on its proboscis, even more than Bombus terrestris, a species that is more abundant in the forest habitat. Furthermore, the higher pollen production in forest populations reduces the chances of an individual pollen grain being picked up and legitimately transferred when the number of active pollinators is not increasing accordingly. On the other hand, species with a shorter tongue were less able to pick up pollen of long-styled flowers, but carried proportionally more pollen of short-styled flowers on their heads (Fig. 3B). Hence, short- and long-tongued insects pick up pollen of the different flower morphs in different proportions on their tongue and proboscis. This reveals a remarkable complementarity between long- and short-tongued insects in the pollination of P. veris and probably even other distylous species. Short-tongued species tend to proportionally transfer more pollen from the S-morph anthers to the L-morph stigma. By contrast, long-tongued insects can reach the anthers of L-morph flowers and can thereby also transfer pollen of L-morph flowers to S-morph stigmas. These findings are in partial fulfilment of the Darwinian hypothesis, which states that clear segregation of pollen morphs on the body of the insect is needed to avoid intramorph pollination (Darwin, 1877; Weller, 2009). It seems that tongue lengths plays a crucial role in the ability of the pollinator to pick up pollen of long-styled flowers. As a result, short-tongued pollinators hardly accumulate any pollen of long-styled flowers on their proboscis and are accordingly not well suited to accumulate and segregate pollen of the two flower morphs on their frontal body parts. Consequently, only long-tongued pollinators could be regarded as good candidates for the Darwinian interpretation of a suitable pollinator in a distylous system. Ultimately, this pollinator- mediated pollen flow diminishes the chances for illegitimate pollination and also prevents pollen wastage. Short-tongued species, such as syrphid flies, feed on pollen and are known to act as pollen-thieves that may reduce the availability of legitimate pollen for long-styled flowers (Santos-Gally et al., 2013; Simón-Porcar et al., 2014). The legitimate pollination on long-styled stigmas was indeed lower than on short-styled stigmas, but this effect was most probably caused by the lower production of pollen by S-morph flowers. In our survey, we encountered three different species that are known to consume pollen (Melanostoma mellinum and two Halictus spp.). Halictus spp. are known to collect plenty of pollen on their hind legs which is then no longer available for pollination (Brittain & Newton, 1933). A more detailed study on the foraging behaviour of short-tongued species could shed more light on the relative importance of pollen robbing. Future studies should examine more pollinators resulting from a longer observation time as the number of pollinators is rather low in early spring. Pollen deposition on stigmas and seed set There was a clear distinction between the two morphs regarding proportional legitimate stigmatic pollen deposition. Although stigmas of long-styled flowers captured more pollen grains in total, stigmas of short-styled flowers captured significantly more legitimate pollen. This is in agreement with results for Primula vulgaris (Keller et al., 2014), Pulmonaria officinalis (Brys et al., 2008) and Limonium vulgare Mill. (Costa et al., 2016). This observation can be explained by the fact that long-styled flowers produced nearly four times as many pollen grains than short-styled flowers, which makes legitimate pollen for short-styled stigmas far more abundant (Brys et al., 2008). The higher pollen production in forests translated into more pollen deposition on stigmas in this habitat. Nevertheless, seed set was highest in the grassland habitat. which could be related to differences in pollen quality (Ashman et al., 2004), but this needs further investigation. Although more pollen was deposited on stigmas of short-styled flowers relative to the amount of developing ovules, there was no significant difference in seed set between the two morphs. Fertilization of ovules with illegitimate pollen is controlled by the self-incompatibility system of P. veris, which hampers the formation of pollen tubes after illegitimate pollination (Richards & Ibrahim, 1982; Brys & Jacquemyn, 2009). There was no deficiency in legitimate pollination, but a considerable number of ovules did not develop into seeds, indicating pollen limitation in respect of pollen quality (Ashman et al., 2004). Furthermore, our results point to a disruption in the reciprocal placement of stigma and anthers in the long-styled flowers of the forest habitat, which agrees with the findings of Brys & Jacquemyn (2015), who showed that variation in stigma height causes this observed deviation from perfect reciprocity. Anther height is conserved in both flower morphs (Brys & Jacquemyn, 2015) and therefore we can assume that the pattern of pollen deposition on a pollinator depends largely on the anatomy of the pollinator and is only minimally affected by the habitat in which a plant grows. Stigmatic pollen deposition on a flower, on the other hand, is probably explained by the level of herkogamy and the present pollinator community because pollen–stigma contact depends largely on the way a pollinator picks up pollen on its body and how the sexual organs are positioned. Because of the different pollinator communities in the habitats and differences in herkogamy, it is difficult to estimate the relative contribution of these two factors in the pollination of P. veris. Further research is needed to disentangle these two components that are involved in the pollination of this distylous plant. CONCLUSIONS Our data have shown that natural pollinator communities of P. veris varied between grassland and forest populations. They also indicate that pollinators with different proboscis lengths may complement each other in legitimately pollinating distylous plant species, with long-tongued species functioning as more appropriate legitimate pollinators for S-morph flowers and short-tongued pollinators more appropriately transporting legitimate pollen for L-morph flowers. These observations thus provide only partial proof of Darwin’s hypothesis of pollen stratification of different morphs on different body parts of a pollinator (Darwin, 1877). Furthermore, our findings also confirm the asymmetry in legitimate stigmatic pollen deposition between the long-styled and short-styled morphs in distylous species with more illegitimate pollen capture in L-morph flowers. Short-tongued insects mainly forage on the pollen stock of high-level anthers, whereas insects with a longer tongue such as A. plumipes can reach the concealed nectar in the bottom of the flower and thereby pick up pollen from low-level anthers. In this way, pollen flow is directed mainly to go from high-level anthers to the higher placed stigmas of the L-morph and vice versa for pollen from low-level anthers. However, more research is needed to unravel the contribution of legitimate pollen flow to total pollen flow for each insect species separately. ACKNOWLEDGEMENTS We would like to thank the FWO (Flemish Fund for Scientific Research) and Rik Sanders for his contribution in surveying and collecting insects. Two reviewers and the Associate Editor provided useful comments on a previous draft of the manuscript. The authors declare no conflicts of interest. REFERENCES Adriaens D, Honnay O, Hermy M. 2006. No evidence of a plant extinction debt in highly fragmented calcareous grasslands in Belgium. Biological Conservation 133: 212– 224. Google Scholar CrossRef Search ADS Alexandersson R, Johnson SD. 2002. Pollinator-mediated selection on flower-tube length in a hawkmoth-pollinated Gladiolus (Iridaceae). Proceedings of the Royal Society B 26: 631– 636. Google Scholar CrossRef Search ADS Arroyo J, Barrett SCH. 2000. Discovery of distyly in Narcissus (Amaryllidaceae). American Journal of Botany 87: 748– 751. Google Scholar CrossRef Search ADS PubMed Arroyo J, Dafni A. 1995. Variations in habitat, season, flower traits and pollinators in dimorphic Narcissa tazetta L. (Amaryllidaceae) in Israel. New Phytologist 129: 135– 145. Google Scholar CrossRef Search ADS Ashman T, Knight TM, Steeds JA, Amarasekare P, Burd M, Campbell DR, Dudash MR, Johnston MO, Mazer SJ, Mitchell RJ, Morgan MT, Wilson WG. 2004. Pollen limitation of plant reproduction: ecological and evolutionary causes and consequences. Ecology 85: 2408– 2421. Google Scholar CrossRef Search ADS Barrett SCH. 1990. The evolution and adaptive significance of heterostyly. Trends in Ecology and Evolution 5: 144– 148. Google Scholar CrossRef Search ADS PubMed Barrett SCH, Shore JS. 2008. New insights on heterostyly: comparative biology, ecology and genetics. In: Barrett SCH, Shore JS, Self-incompatibility in flowering plants: evolution, diversity, and mechanisms . Berlin: Springer, 3– 32. Google Scholar CrossRef Search ADS Brittain WH, Newton DE. 1933. A study in the relative constancy of hive bees and wild bees in pollen gathering. Canadian Journal of Research 9: 334– 349. Google Scholar CrossRef Search ADS Brys R, Jacquemyn H. 2009. Biological flora of the British Isles: Primula veris L. Journal of Ecology 97: 581– 600. Google Scholar CrossRef Search ADS Brys R, Jacquemyn H. 2015. Disruption of the distylous syndrome in Primula veris. Annals of Botany 115: 27– 39. Google Scholar CrossRef Search ADS PubMed Brys R, Jacquemyn H, Hermy M, Beeckman T. 2008. Pollen deposition rates and the functioning of distyly in the perennial Pulmonaria officinalis (Boraginaceae). Plant Systematics and Evolution 273: 1– 12. Google Scholar CrossRef Search ADS Butaye J, Honnay O, Adriaens D, Delescaille L, Hermy M. 2005. Phytosociology and phytogeography of the calcareous grasslands on Devonian limestone in southwest Belgium. Belgian Journal of Botany 138: 24– 38. Costa J, Castro S, Loureiro J, Barrett SCH. 2016. Experimental insights on the function of ancillary pollen and stigma polymorphisms in plants with heteromorphic incompatibility. Evolution 71: 121– 134. Google Scholar CrossRef Search ADS PubMed Costa J, Castro S, Loureiro J, Barrett SCH. 2017. Experimental insights on Darwin’s cross-promotion hypothesis in tristylous purple loosestrife (Lythrum salicaria). American Journal of Botany 104: 616– 626. Google Scholar CrossRef Search ADS PubMed Darwin CR. 1862. On the two forms, or dimorphic condition, in the species of Primula and on their remarkable sexual relations. Proceedings of the Linnean Society of London (Botany) 6: 77– 96. Google Scholar CrossRef Search ADS Darwin CR. 1877. The different forms of flowers on plants of the same species . London: J. Murray. Google Scholar CrossRef Search ADS Deschepper P, Brys R, Fortuna MA, Jacquemyn H. 2017. Analysis of spatial genetic variation reveals genetic divergence among populations of Primula veris associated to contrasting habitats. Scientific Reports 7: 8847. Google Scholar CrossRef Search ADS PubMed Dulberger R. 1974. Structural dimorphism of stigmatic papillae in distylous Linum species. American Journal of Botany 61: 238– 243. Google Scholar CrossRef Search ADS Ganders FR. 1979. The biology of heterostyly. New Zealand Journal of Botany 17: 607– 635. Google Scholar CrossRef Search ADS Gutiérrez D, Thomas CD. 2000. Marginal range expansion in a host-limited butterfly species Gonepteryx rhamni. Ecological Entomology 25: 165– 170. Google Scholar CrossRef Search ADS Jacquemyn H, Endels P, Brys R, Hermy M, Woodell SRJ. 2009. Biological flora of the British Isles: Primula vulgaris Huds. (P. acaulis (L.) Hill). Journal of Ecology 97: 812– 833. Google Scholar CrossRef Search ADS Keller B, Thomson JD, Conti E. 2014. Heterostyly promotes disassortative pollination and reduces sexual interference in Darwin’s primroses: evidence from experimental studies. Functional Ecology 28: 1413– 1425. Google Scholar CrossRef Search ADS Kohn JR, Barrett SCH. 1992. Experimental studies on the functional significance of heterostyly. Evolution 46: 43– 55. Google Scholar CrossRef Search ADS PubMed Lau P, Bosque C. 2003. Pollen flow in the distylous Palicourea fendleri (Rubiaceae): an experimental test of the disassortative pollen flow hypothesis. Oecologia 135: 593– 600. Google Scholar CrossRef Search ADS PubMed Lloyd DG, Webb CJ. 1992. The selection of heterostyly. In: Barrett SCH ed. Evolution and function of heterostyly . Berlin: Springer, 179– 207. Google Scholar CrossRef Search ADS Lowry DB, Rockwood RC, Willis JH, Lowry B. 2012. Ecological reproductive isolation of coast and inland races of Mimulus guttatus. Evolution 62: 2196– 2214. Google Scholar CrossRef Search ADS Meeus S, Honnay O, Jacquemyn H. 2013. Differences in fine-scale spatial genetic structure across the distribution range of the distylous forest herb Pulmonaria officinalis (Boraginaceae). BMC Genetics 14: 101. Google Scholar CrossRef Search ADS PubMed Olesen JM. 1979. Floral morphology and pollen flow in the heterostylous species Pulmonaria obscura (Boraginaceae). New Phytologist 82: 757– 764. Google Scholar CrossRef Search ADS Pérez-Barrales R, Vargas P, Arroyo J. 2006. New evidence for the Darwinian hypothesis of heterostyly: breeding systems and pollinators in Narcissus sect. Apodanthi. New Phytologist 171: 553– 567. Google Scholar CrossRef Search ADS PubMed Peter CI, Johnson SD. 2014. A pollinator shift explains floral divergence in an orchid species complex in South Africa. Annals of Botany 113: 277– 288. Google Scholar CrossRef Search ADS PubMed Potts SG, Vulliamy B, Dafni A, Neeman G, Willmer P. 2003. Linking bees and flowers: how do floral communities structure pollinator communities? Ecology 84: 2628– 2642. Google Scholar CrossRef Search ADS Proctor M, Yeo P. 1973. The pollination of flowers . London: Collins. Rasband WS. 2011. ImageJ . Bethesda: US National Institutes of Health. Available at: http://imagej.net R Core Team 2015. R: a language and environment for statistical computing . Vienna: R Foundation for Statistical Computing. Available at: http://www.R-project.org. Richards AJ, Ibrahim HB. 1982. The breeding system in Primula veris L. New Phytologist 90: 305– 314. Google Scholar CrossRef Search ADS RStudio Team 2016. RStudio: integrated development for R . Boston: RStudio, Inc. Available at: http://www.rstudio.com. Rosov SA, Screbtsova ND. 1958. Honey bees and selective fertilization of plants. XVII International Beekeeping Congress 2: 494– 501. Santos-Gally R, Pérez-Barrales R, Simón VI, Arroyo J. 2013. The role of short-tongued insects in floral variation across the range of a style-dimorphic plant. Annals of Botany 111: 317– 328. Google Scholar CrossRef Search ADS PubMed Simón-Porcar VI, Santos-Gally R, Arroyo J. 2014. Long-tongued insects promote disassortative pollen transfer in style-dimorphic Narcissus papyraceus (Amaryllidaceae). Journal of Ecology 102: 116– 125. Google Scholar CrossRef Search ADS Stone G. 1994. Activity patterns of females of the solitary bee Anthophora plumipes in relation to temperature, nectar supplies and body-size. Ecological Entomology 19: 177– 189. Google Scholar CrossRef Search ADS Stone JL, Thomson JD. 1994. The evolution of distyly: pollen transfer in artificial flowers. Evolution 48: 1595– 1606. Google Scholar CrossRef Search ADS PubMed Van Rossum F, De Sousa SC, Triest L. 2006. Morph-specific differences in reproductive success in the distylous Primula veris in a context of habitat fragmentation. Acta Oecologica 30: 426– 433. Google Scholar CrossRef Search ADS Weller SG. 2009. The different forms of flowers – what have we learned since Darwin? Botanical Journal of the Linnean Society 160: 249– 261. Google Scholar CrossRef Search ADS Wolfe LM, Barrett SCH. 1989. Patterns of pollen removal and deposition in tristylous Pontederia cordata L. (Pontederiaceae). Biological Journal of the Linnean Society 36: 317– 329. Google Scholar CrossRef Search ADS Zhu XF, Jiang XF, Li L, Zhang ZQ, Li QJ. 2015. Asymmetrical disassortative pollination in a distylous primrose: the complementary roles of bumblebee nectar robbers and syrphid flies. Scientific Reports 5: 7721. Google Scholar CrossRef Search ADS PubMed © 2018 The Linnean Society of London, Botanical Journal of the Linnean Society
Botanical Journal of the Linnean Society – Oxford University Press
Published: Mar 1, 2018
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
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
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.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera