Skip to main content
NCBI home page
Annals of Botany logoLink to Annals of Botany
. 2018 Aug 29;123(1):205–212. doi: 10.1093/aob/mcy162

On the adaptive value of monomorphic versus dimorphic enantiostyly in Solanum rostratum

Emiliano Mora-Carrera 1, Miguel Castañeda-Zárate 1, Juan Fornoni 1, Karina Boege 1, César A Domínguez 1,✉
PMCID: PMC6344091  PMID: 30184097

Abstract

Background and Aims

Enantiostyly is a reproductive system with heteromorphic flowers characterized by asymmetrical deflection of the style, either to the left or to the right of the floral axis. There are two types of enantiostyly. In monomorphic enantiostyly, plants produce the two types of flowers in the same individual. Dimorphic enantiostyly is restricted to only seven species and their populations consist of individuals producing either the right or the left flower type. It is hypothesized that the dimorphic form is derived from monomorphic ancestors because it functions as an outcrossing mechanism. We tested this latter hypothesis and investigated if monomorphic enantiostyly is resistant to invasion by individuals with dimorphic enantiostyly, because it functions as a reproductive assurance mechanism.

Methods

To determine the conditions favouring the invasion of dimorphic enantiostyly, measurements of reproductive success and outcrossing rates in 15 natural flowering patches of Solanum rostratum were made. To test if monomorphic enantiostyly provides a reproductive assurance mechanism, experimental plants with either manually created dimorphic or natural monomorphic reproductive systems were exposed to two different pollination scenarios (flower density treatments), and reproductive success and outcrossing rates were measured.

Key Results

Naturally flowering patches experienced severe pollination limitation, showed marked differences in reproductive success and had relatively high outcrossing rates. Plants in the experimental patches also showed pollination limitation and high outcrossing rates. Individuals with dimorphic enantiostyly expressed higher reproductive and outcrossing advantages under high-density conditions. These advantages disappeared in the low-density treatment, where the monomorphic form attained a higher reproductive success and no differences in outcrossing rates were detected.

Conclusions

Monomorphic enantiostyly should be resistant to invasion of the dimorphic form because the prevalent ecological conditions favour the maintenance of geitonogamous individuals that are able to take advantage of ecological heterogeneity and generalized pollination limitation.

Keywords: Enantiostyly, Solanum rostratum, reproductive assurance, adaptive value, pollen limitation, geitonogamy

INTRODUCTION

The diversity of plant reproductive systems has been the focus of naturalists and evolutionary biologists for more than two centuries (Sprengel, 1793; Darwin, 1862). This fascination derives from the complex combination of morphological (i.e. monoclinous to diclinous species, monomorphic to polymorphic systems), temporal (protandrous or protogynous species) and genetic (presence or not of self-incompatibility systems) phenomena building up the astonishing variation of sexual expressions exhibited by flowering plants (Richards, 1997; Geber et al., 1999). The seminal work of Charnov (1982) stressed how the different forms can be explained as the result of a single process, the optimal allocation of resources to each sexual function under a set of ecological and genetic conditions.

The evolution of enantiostyly, a stylar polymorphism characterized by deflection of the style to either the right or the left (hereafter, floral types R or L) from the perspective of the flower (Jesson and Barrett, 2003), is particularly interesting in the context of the evolution of plant reproductive systems. This stylar polymorphism is reported in at least 12 unrelated monocot and dicot families, suggesting that similar selective pressures drove its evolution (Endress, 2001; Jesson and Barrett, 2003). Because enantiostyly enforces reciprocal pollen transfer between R and L floral types, it has been visualized as an outcrossing enhancing mechanism (Jesson and Barrett, 2005; Gao et al., 2006; Almeida et al., 2013a). Such interpretation, however, is applicable only to species with dimorphic enantiostyly. Populations of these species are dimorphic regarding sex expression with some individuals producing one floral type (i.e. the Right type), while the rest produce the other type (i.e. the Left type; Jesson et al., 2003). Thus, species with dimorphic enantiostyly behave just as distylous systems in which reciprocal crosses occur between individuals with different morphs. In monomorphic enantiostyly, by contrast, all individuals within a population bear both floral types (usually in the same proportion), thus allowing the occurrence of geitonogamy (i.e. pollination among flowers of the same plant), which in turn favours the evolution of mixed mating systems with relatively lower outcrossing rates (Wang et al., 1998; Vallejo-Marín et al., 2013).

Although empirical and theoretical studies suggest that dimorphic enantiostyly may have a reproductive advantage over the monomorphic type, the former has been reported in only seven species [four genera (Wachendorfia, Babaretta, Hetheranthera and Cyanella) in three families (Haemodoraceae, Tecophilaeaceae, Pontederiaceae)] (Fenster, 1995; Jesson and Barrett, 2003, 2005), raising the question of whether selection favouring outcrossing is the main pressure involved in the evolution of enantiostyly. Given that dimorphic enantiostyly has clear similarities with distyly, it probably serves the same functions, i.e. the reduction or avoidance of selfing and the increase in the proficiency of pollen transfer between plants. This explanation, however, does not fully apply to monomorphic enantiostyly, because although it potentially reduces the probability of selfing, it still allows for geitonogamy and reduces the number of flowers for mating (in the same or different individuals). Therefore, here we argue that the ability of monomorphic enantiostyly to reproduce via geitonogamy under pollen limitation scenarios and still be capable of taking advantage of more favourable conditions may function as a reproductive assurance mechanism. Large flowering displays could increase pollinator visitation to self-compatible species facing a shortage of pollinators and/or mating partners (Klinkhamer and Jong, 1991; Lloyd and Schoen, 1992; de Jong et al., 1993), but incurs the costs imposed by higher rates of geitonogamy. Monomorphic enantiostyly could partially solve this problem by reducing the probability of geitonogamous crosses within an individual independently of the size of the floral display (Jesson and Barrett, 2003) and could also take advantage of high pollinator abundance and/or high availability of reproductive consorts, by attaining higher rates of outcrossing than self-compatible species without flower polymorphisms. Thus, monomorphic enantiostyly could be resistant to the invasion of dimorphic individuals if there is low inbreeding depression and temporal and spatial heterogeneity in the availability of pollinators and/or mates (Jesson et al., 2003). Such a dynamic would result in an evolutionary stable mixed mating system (Goodwillie et al., 2005). In contrast, if natural selection favours high rates of outcrossing (high levels of inbreeding depression) and plants face a relatively stable pollination environment, dimorphic enantiostyly is expected to invade a monomorphic population.

Comparative studies and evolutionary models support the hypothesis that dimorphic enantiostyly evolved from the monomorphic form (Jesson and Barrett, 2003; Jesson et al., 2003). However, no attempts have been made to explore the ecological scenarios and selective pressures favouring invasion of the dimorphic form over the monomorphic form of enantiostyly. Such an endeavour would allow us to distinguish whether monomorphic enantiostyly (and its associated mixed mating system) functions as a reproductive assurance mechanism (i.e. is maintained because of its functional advantages), or if there is a developmental constraint precluding the evolution of the dimorphic form.

In this study, we experimentally tested whether the monomorphic enantiostyly of Solanum rostratum has functional advantages reducing or impeding the invasion of a dimorphic enantiostylous reproductive system. To this end, we simulated the evolution of a population of plants with dimorphic enantiostyly and compared the magnitude of two fitness components (seed production and individual outcrossing rates) with those of the ‘ancestral’ monomorphic enantiostylous form. Moreover, because we hypothesized that monomorphic enantiostyly is advantageous under high temporal and spatial heterogeneity conditions, we simulated two contrasting pollination environments by manipulating flower density. Based on the assumption that monomorphic enantiostyly functions as a reproductive assurance mechanism because of its ability to deal with heterogeneous pollen/pollinator availability scenarios (Lloyd, and Schoen, 1992; Delmas et al., 2015), our experimental setting allowed us to test the following two predictions: (1) dimorphic enantiostyly should express higher outcrossing rates and seed production than the monomorphic form under high pollen availability conditions; (2) in contrast, dimorphic but not monomorphic enantiostyly should be disadvantageous under conditions of pollen or pollinator scarcity. Therefore, provided that monomorphic enantiostyly functions as a reproductive assurance mechanism, we predict that the average fitness across pollination environments should be higher for those individuals with monomorphic enantiostyly. Finally, because our hypothesis relies on the existence of temporal and spatial heterogeneity in the ecological conditions determining individual reproductive success, we first characterized individual fitness and outcrossing rates in 15 naturally flowering patches.

MATERIAL AND METHODS

Study species and study site

Solanum rostratum is a self-compatible annual shrub with monomorphic enantiostyly and yellow, hermaphroditic, zygomorphic flowers exhibiting heteranthery (flowers bear two sets of anthers each specialized in either pollination or feeding). These floral attributes result in the relatively high outcrossing rate (t) observed in S. rostratum populations (t = 0.70 ± 0.03; Vallejo-Marín et al., 2013). Flowers usually open at 0800 h, close at 1800 h and remain active for 2–3 d (up to 5 d under glasshouse conditions). The main pollinators are buzz-pollinating bees belonging to the genera Bombus and Xylocopa. The daily floral display may range from one to 50 flowers depending on the size of the plant with almost all individuals exhibiting a 1: 1 proportion of R and L flowers. The species is native to central and northern Mexico, and is currently distributed across USA, the southern part of Canada and has invaded some parts of Asia and Europe (Zhao et al., 2013). All experiments and observations from this study were performed at the Pedregal de San Angel ecological preserve, which is located in central Mexico (19°18′21″–19°20′11″N, 99°10′15″–99°12′4″W) within the natural distribution range of S. rostratum, during summer 2015.

Natural variation in reproductive success and outcrossing rates

To characterize the amount of natural variation in reproductive success and outcrossing rates, we selected 15 flowering patches of S. rostratum within the ecological preserve (n = 73 plants). Because it is not rare to observe isolated reproductive individuals within the preserve, we included four ‘patches’ consisting of only one individual (mean individuals per patch = 4.8 ± 5.4, range: 1–20). Natural patches were found in open spaces and side roads with individual plants growing in close proximity to each other (0.3–0.5 m). Mean distance among natural patches was 215.8 m (s.e. ± 85.5). All reproductive plants within each flowering patch were marked and flower, fruit and seed production were counted throughout the reproductive season. The probability of seed germination was also measured for each individual. Individual reproductive success was estimated as the multiplicative fitness of fruit-set, mean seeds per fruit and the probability of germination (methods as outlined below). These data were given relative to WiW¯, where Wi is the multiplicative fitness of the ith individual, and W¯ is the average fitness of all individuals in the sample. Multilocus outcrossing rates for each individual were estimated following the methods outlined below (see Estimation of outcrossing rates). To explore whether the variation in fitness and outcrossing rates was explained by the magnitude of the flowering display, we used these two components of fitness as dependent variables in two independent regression models. In both cases, the number of flowers per plant (FI) and the number of flowers in the reproductive neighbourhood (FN; i.e. the number of flowers produced in a given patch minus the flowers produced by the ith individual) were included as independent variables. Because plants in extreme flowering displays may fail to reproduce because of pollination limitation (Knight et al., 2005), we included the quadratic and interaction terms in the models.

Reproductive advantages of monomorphic vs. dimorphic enantiostyly

Experimental design.

To evaluate if monomorphic enantiostyly (ME) has functional advantages over the dimorphic form, we used a split-plot design aimed to compare the reproductive success and outcrossing rates of individuals belonging to monomorphic or dimorphic enantiostyly (reproductive system, hereafter RS) under two contrasting pollination scenarios (floral density, hereafter FD). To simulate dimorphic enantiostyly (DE), individual plants were maintained in 2-litre pots and manipulated to bear only four L or R flowers per day, while plants assigned to the ME reproductive system maintained two L flowers and two R flowers each (hereafter R/L). In this way, all experimental plants were identical in terms of flowering display. Thus, depending on the availability of flowering plants, each experimental patch consisted of nine or 12 experimental individuals (i.e. three or four plants of each floral type). Because we aimed to simulate scenarios favouring (or not) outcrossing, we established two treatments differing in flower density. A high-density (HD) flowering patch is expected to favour outcrossing because of its attractiveness to pollinators (higher flowering display and increased reward availability and more potential consorts). In contrast, low-density patches (LD) should attract fewer pollinators, more visits per flower by a single pollinator and increased geitonogamy. HD conditions were attained by adding 6–10 peripheral individuals with 6–10 flowers per individual. Accordingly, mean (± s.e.) flower density in HD and LD patches was 112.0 m−2 (± 1.6) and 44.1 flowers m−2 (± 0.6), respectively. These values were chosen on the basis of the natural variation observed among flowering patches within the Pedregal preserve. In total, 16 experimental patches (eight HD and eight LD) were used for this study. Mean distance among pairs of experimental patches was 304.2 m (± 77.1 m). All plants used in this experiment (n = 192) were randomly assigned to each treatment and location within a given patch. Experimental patches were located in open spaces within the preserve simulating natural patches. Once a flowering patch was set in the field, it remained there for 3 d exposed to natural pollination. For each experimental patch, observations of pollinator visitation and estimations of individual reproductive success were performed. After this period, experimental plants were moved to an enclosure to avoid additional pollination visitation. Because we were interested in the effects of the pollination environment on fitness, reproductive success (W) was estimated by multiplying the number of flowers per individual (experimentally set to four), fruit-set and the mean number of seeds per fruit. To evaluate the potential effects of inbreeding depression on fitness, the previous analysis was repeated by using a slightly different estimate of reproductive success including the proportion of germinated seeds.

Flower visitation rate.

To test if floral density has an effect on patch and flower visitation rate, we performed a series of pollinator censuses. Observations were carried out for 6 d during late September 2015 between 0900 and 1400 h. Each observation day consisted of six 20-min observation periods with a 5-min break between censuses. Observations on HD and LD patches were carried out simultaneously by independent observers. A total of 24 h of observation were completed (12 h for each floral density treatment). Data on the number of pollinators visiting each patch and the number of flowers visited per pollinator per patch were collected. Visitations by bees from the genus Apis were not taken into account because they have been reported as pollen thieves of S. rostratum (Solís-Montero et al., 2015). Because we were unable to distinguish pollinator visitation to plants belonging to different RS treatments within a flowering patch, we only analysed differences between flower density treatments. The rate of pollinator visitation (number of visits per flower per hour) and the total number of pollinators per patch were compared between flower density treatments by means of Wilcoxon tests.

Estimation of outcrossing rates.

Outcrossing rates were estimated for each experimental plant by genotyping 5–30 seedlings per mother plant. Seeds from each plant were randomly selected, treated with 100 ppm of gibberellic acid, incubated for 24 h at 20 °C with 12: 12 h light–dark periods to break dormancy and sowed in germination pots with a 1: 1 mix of soil and soil improver (Agrolita). Germination pots were maintained in a controlled glasshouse (22 °C day, 16 °C night). Vegetative tissue (leaves) was collected 8 weeks after germination.

DNA extractions were performed using a KingFisher Pure Plant DNA kit with a KingFisher Flex automated nucleic acid purification system (Thermo Fisher Scientific, Waltham, MA, USA) following the manufacturer’s instructions. Outcrossing rates were estimated by using 13 microsatellites previously reported for S. rostratum (Vallejo-Marín et al., 2011). All 13 microsatellites were amplified in a single reaction using a Multiplex PCR kit (Qiagen, Valencia, CA, USA) in a TC-512 thermocycler (TECHNE, USA). We followed the previously reported microsatellite amplification program (Vallejo-Marín et al., 2013). Fragment analysis for genotyping was carried out in the UIUC Core Sequencing Facility (University of Illinois, USA). Analysis of fragment data was performed with the GeneMarker (SoftGenetics, State College, PA, USA) program. Multi- (Tm) and single- (Ts) locus outcrossing rates were calculated using the program MLTR (Ritland, 2002). Outcrossing rates were estimated for 44 mother plants, from an average of 15.93 (s.e. = 0.79) seedlings per plant. Overall, outcrossing rates were estimated for 22 plants from HD and 22 LD experimental patches with 13 plants in the ME and 31 in the DE treatment (14 and 17 R and L, respectively).

Statistical analysis.

We used generalized linear mixed models (GLMMs) to analyse differences in fitness (W) and outcrossing rates (Tm) between individuals belonging to the monomorphic (ME) or dimorphic (DE) reproductive system under two flower density conditions (HD and LD patches). In both cases, we included reproductive system, floral density and the interaction term as main factors. To account for the split-plot design, we treated patch (whole plot), treatment within patch (subplot) and individual as random effects. We used the lme4 package (Bates et al., 2015) in R 3.2.5. (R Core Team, 2016) using a Poisson distribution for the W analysis. Because only integer values can be used with this distribution, we used the raw values of fitness and decimals were rounded to the closest integer value. In the Tm analysis, we used a binomial distribution. In all cases, Akaike’s information criterion (AIC) was used for selecting between alternative models. Significance of main factors in these analyses was tested by means of Wald’s χ2 tests (type III) with the car package (Fox and Weisberg, 2011). Multiple contrasts analyses were made with the multcomp package (Bretz et al., 2010).

Estimation of inbreeding depression

The amount of inbreeding depression (δ) in S. rostratum was estimated through comparison of the germination rate and survival to reproduction of plants derived from self- and outcross pollination. Fifty individuals of S. rostratum from the REPSA population were randomly selected and maintained in 2-litre pots until reproduction. For each individual, half of the flowers were hand-pollinated with pollen from the same individual, while the rest were pollinated with a mixture of pollen from five to ten other plants. Intra- and inter-floral type crosses were made indistinctly because no difference in fruit production was detected (M. Castañeda-Zárate, pers. observ.). Mature seeds were pooled by treatment and individual, and 50 were randomly chosen and sowed in germination pots. Pots were maintained under glasshouse conditions and the location of each pot within the greenhouse was randomly assigned. Inbreeding depression (δ) for each individual was estimated as 1−(WsWo) (Hedrick, 2000), where Ws and Wo are the mean performance (estimated as the probability of germination, and survival to reproduction) of self- and outcross progeny, respectively. To have a broader perspective of the amount of inbreeding depression experienced by S. rostratum populations, we also estimated δ from published population genetic data (Vallejo-Marín et al., 2013) following the procedures outlined by Goodwillie et al. (2005). To test whether these estimates differed from the threshold value of δ = 0.5 (Lande and Schemske, 1985), we performed three independent t tests in R 3.2.5 (R Core Team, 2016).

RESULTS

Natural variation in reproductive success and outcrossing rates

The studied population of S. rostratum in the Pedregal de San Angel showed a marked heterogeneity in the size of the flowering display among patches. The average number of flowers per patch was 130.0 (± 52.8) with most patches expressing a relatively low flowering display (mode = 19). In fact, there were only three patches with flowering displays of more than 200 flowers. In accordance, relative fitness (relative number of germinated seeds per plant) showed marked differences among natural flowering patches of S. rostratum. There were two orders of magnitude between the patches with the lowest and highest reproductive success (0.01 to 2.86, respectively, Fig. 1A). The same trend was observed in the average multilocus outcrossing rate (Fig. 1B), which expressed a fivefold difference between individuals (range in Tm from 0.2 to 1) with an average value of 0.94 (± 0.015). Regression analyses showed that relative fitness (but not outcrossing rates, results not shown) was significantly associated with both the number of flowers per individual and those in the reproductive neighbourhood (Supplementary Data Table S1). As indicated by the significance of the quadratic and interaction terms, intermediate values of both the number of flowers per individual (FI) and the number of flowers in the reproductive neighbourhood (FN) maximized reproductive success (141.12 and 134.82, respectively). Thus, in accordance with our expectations, the above results show a marked heterogeneity in the ecological conditions determining individual reproductive success and outcrossing rates.

Fig. 1.

Fig. 1.

Open in a new tab

Natural variation of (A) mean relative reproductive success ± s.e. and (B) mean multilocus outcrossing rates (Tm) ± s.e. of individuals in 15 natural patches of Solanum rostratum of the REPSA population (N = 73). Reproductive success (W) is the multiplicative product of fruit-set, mean seeds per fruit and probability of germination; individual W were given relative to the mean of the population.

Floral visitation in experimental settings

Xylocopa tabaniformis subsp. mexicana was the main flower visitor observed in the experimental settings, although visitation by Apis bees was occasionally observed. As predicted, flowers in HD patches had a statistically significantly higher visitation rate (number of visits per flower per hour) than those in LD patches (χ2(1) = 13.5, P = 0.0002; Fig. 2). This result is due to the combination of a higher, but non-significant (χ2(1) = 2.58, P = 0.11), number of pollinators arriving at HD than LD patches (15.43 ± 4.6 and 8.14 ± 3.1 pollinators, respectively), which in turn visited more flowers (17.7 ± 1.64 flowers per foraging bout in HD patches and 13.1 ± 1.82 in LD; χ2(1) = 3.93, P = 0.047).

Fig. 2.

Fig. 2.

Open in a new tab

Boxplot showing the effects of high (HD) and low (LD) flower density treatments on pollinator visitation rate (visits per flower per hour). Wilcoxon test revealed differences between treatments (χ2(1) = 13.5, P = 0.0002), based on 24 h of observation. Only visits of Xylocopa tabaniformis subsp. azteca were included.

Differences in reproductive success and outcrossing rates in experimental conditions

Preliminary analyses showed that most plants in the experimental settings failed to set any fruit. Accordingly, we adjusted a logistic model with a split plot design to determine whether the probability of setting a fruit differed between floral density or reproductive system treatments. Plants were categorized into two categories (successful reproduction [1] or failed reproduction [0]). Results revealed that there were no differences in the probability of reproducing between density or reproductive system treatments (χ2(1) = 1.15, P = 0.28; χ2(1) = 0.20, P = 0.65, respectively), nor of the interaction term (χ2(1) = 1.71, P = 0.19). Thus, analyses of reproductive success were performed only including plants that were able to reproduce.

Reproductive success showed significant differences between reproductive systems (χ2(1) = 5.38, P = 0.02) and floral densities (χ2(1) = 4.31, P = 0.03). Plants in the ME treatment had a higher multiplicative fitness than those in the DE treatment, while plants in LD conditions outperformed those in HD (Fig. 3A). This latter result is explained by the significance of the interaction term (χ2(1) = 18.92, P < 0.001), indicating that plants in the ME treatment had higher fitness at low floral density (Fig. 3A). When this analysis was repeated with the reproductive success estimate including early-life inbreeding depression, qualitatively equivalent results were obtained, but differences in the patterns of significance emerged. As indicated by the significance of the interaction term, plants in the DE treatment showed reproductive advantages in HD conditions, while no differences between reproductive systems were observed in LD conditions (χ2(1) = 7.82, P = 0.005; Fig. 3B). Comparison between the former and the latter results suggests that DE advantages are expressed under restrictive conditions (high floral density – high pollination visitation) through the action of inbreeding depression.

Fig. 3.

Fig. 3.

Open in a new tab

Estimated marginal means (± s.e.) of reproductive success (A) excluding and (B) including probability of germination. Mean multilocus outcrossing rates (± s.e.), estimated with 13 microsatellites. Dimorphic enantiostyly (DE; either L or R) or monomorphic enantiostyly (ME) with 1: 1 floral type ratio (L/R); and across high and low floral density patches (HD and LD, respectively).

Individual outcrossing rates ranged from 0.29 to 1 (mean = 0.898 ± 0.03), with 17 individuals expressing a Tm of 1. Both floral density (χ2(1) = 84.21, P < 0.001) and reproductive system (χ2(1) = 120.2, P < 0.001) showed significant differences in outcrossing rates. Tm in the HD treatment was significantly higher (0.927 ± 0.04) than in the LD treatment (0.870 ± 0.05). As expected, the DE treatment had a higher Tm (mean = 0.931 ± 0.02) than ME (mean = 0.817 ± 0.08). These results are explained by the significance of the interaction term (RS × FD; χ2(1) = 39.65, P < 0.001; Fig. 3C), indicating that plants assigned to the DE treatment had a higher outcrossing rate than those in ME under HD, but not under LD conditions.

Estimation of inbreeding depression

Inbreeding depression of S. rostratum estimated herein was 0.46 ± 0.03 for the probability of germination and 0.46 ± 0.04 for survival to reproduction. Independent estimations based on population genetic data were equivalent or even slightly higher than these values. The average inbreeding depression expressed by four other Mexican populations was 0.514 ± 0.06 and ranged from 0.48 to 0.61. Neither our own estimations of inbreeding depression nor those obtained from Vallejo-Marín et al. (2013), differed significantly from δ = 0.5 (P > 0.14, in all cases).

DISCUSSION

Results from this study demonstrated that S. rostratum faces heterogeneous ecological conditions determining the probability of successful reproduction. This heterogeneity seems to be determined by the number and size of the reproductive individuals constituting a flowering patch, which in turn results in the size of the flowering display. Pollinator censuses performed on experimental settings revealed that more pollinators arrived at HD patches, and once there, visited more flowers, thus resulting in a higher flower visitation rate in experimental HD patches. Our experiments also showed that the DE treatment had reproductive advantages (higher fitness and outcrossing rates) only under high density – high pollination conditions. These advantages, however, disappeared at low flower density where individuals with an ME system outperformed the dimorphic form and still attained relatively high outcrossing rates.

Overall, the results from this study indicate that monomorphic enantiostyly should be resistant to the invasion of dimorphic enantiostyly because (1) reproductive success (averaged through reproductive systems) was higher in the LD than in the HD treatment; (2) the advantages associated with the dimorphic form are attained only under restrictive ecological conditions (high density – high pollination conditions); (3) natural flowering patches are commonly restricted to a few individuals with relatively small flowering displays; (4) both large individuals or patches (with many flowers) seem to experience pollination limitation in natural conditions; and (5) only plants in the ME treatment were able to take advantage of the poor pollination environment generally experienced by S. rostratum individuals in the Pedregal de San Angel ecological conditions.

It has been suggested that pollination environment is a major determinant of the reproductive success of plants, and therefore, a major driver of the evolution of plant reproductive systems (Knight, 2003; Ashman et al., 2004; Knight et al., 2005). In this study, we found that the magnitudes of the flowering display of both plants and patches affect the floral visitation rate (experimental settings) and plant reproductive success (natural conditions and experimental settings) (Kunin, 1997; Grindeland et al., 2005). It has been found that larger flowering displays correlate with higher rates of pollinator visitation (Thomson, 1988; Klinkhamer and Jong, 1991; Makino et al., 2007; Zhang and Lou, 2015) and increased reproductive success (Jennersten, 1988; Steffan-Dewenter and Tscharntke, 1999; Knight, 2003; Dauber et al., 2010), which in turn promote increased outcrossing rates (Murawski and Hamrick, 1991). Although our results partially support these statements, we found that the relationship between flower production, the size of the reproductive neighbourhood and reproductive success is more complex than a simple linear association. Although the number of flowers produced by a given plant and the size of the flowering display of a given patch correlated with individual reproductive success in natural conditions, we also found that extreme values of both variables resulted in reduced fitness. This pattern could be explained if positive density dependence (Knight et al., 2005; Zhang and Lou, 2015) operates in small plants (and/or patches), and pollinator limitation–satiation occurs in larger plant (and/or patches) (Coombs et al., 2009; Spigler and Chang, 2009; Delmas et al., 2015; Menge et al., 2017). Such behaviour has been observed in other plant species in the context of pollination and seed dispersal (Gryj and Dominguez, 1996; Domínguez et al., 1997). Consequently, our results indicate that intermediate values of flowering display resulting from the additive effect of plant and patch size (in terms of flower number) maximize individual reproductive success. These findings suggest that pollinator limitation operating in large plants or flowering patches may preclude the potential advantages of plants with dimorphic enantiostyly. This conclusion is further supported by experimental pollination experiments revealing that the average fruit-set after hand pollination (0.74 ± 0.02) is more than three times higher than that observed in natural conditions (0.21 ± 0.03; F(1,207) = 110.6, P < 0.0001).

Dimorphic enantiostyly has been regarded as a mechanism promoting outcrossing (Fenster, 1995; Jesson and Barrett, 2005) because the presence of enantiomorphic flowers increases pollen flow between different floral types (i.e. Chamescrita flexulosa, Almeida et al., 2013b) and is expected to augment outcrossing when these flowers are located in different individuals (Jesson and Barrett, 2005). Although our results supported this hypothesis because we found a significantly higher outcrossing rate in plants assigned to the DE (0.93 ± 0.02) treatment in comparison with those in ME (0.82 ± 0.08), both estimations fell within the range of outcrossing species (range for predominantly outcrosser species 0.8–1.0, sensuLande and Schemske, 1985). Thus, even though dimorphic enantiostyly actually promotes outcrossing, monomorphic enantiostyly may attain high values of outcrossing even under stressful ecological scenarios (low availability of pollen or mating partners). This conclusion is further supported by the relatively high outcrossing rates observed for the Pedregal natural population (0.90 ± 0.03, this study) and for four other Mexican populations of S. rostratum (0.70 ± 0.03; Vallejo-Marín et al., 2013). Accordingly, for dimorphic enantiostyly to invade a monomorphic population, the potential increase in the outcrossing rate should provide a strong competitive advantage. Such an advantage can only be attained when the deleterious effects of inbreeding depression are relatively intense (>0.5, sensuLande and Schemske, 1985), and there is no pollen limitation.

A theoretical study by Jesson et al. (2003) concluded that monomorphic enantiostylous populations with low inbreeding depression and restricted pollen flow would be resistant to invasion by dimorphic enantiostylous mutants. Under these circumstances, plants with dimorphic enantiostyly are expected to experience pollen and mate limitation because of their practical inability to reproduce via selfing. Our results supported this expectation; the putative advantages of this reproductive system (higher fitness and outcrossing rates) were only apparent under HD conditions, but disappear under LD conditions. This result is a consequence of ME plants being able to take full advantage of LD conditions (higher fitness than DE plants and equivalent outcrossing rates). A comparison between DE and ME (Table S2) revealed that the disadvantages of ME under high density conditions are due to slight differences in all three fitness components, suggesting the presence of mild inbreeding depression acting on the first stages of the life cycle. In contrast, more stringent conditions (LD treatment) may favour the maintenance of self-compatible individuals through geitonogamous selfing (Vaughton and Ramsey, 2010; Delmas et al., 2015) because of the advantages of geitonogamous selfing as a reproductive assurance mechanism (Pannell, 2015). In fact, none of the estimated values of inbreeding depression (δ = 0.51 ± 0.06, range 0.48–0.61 for four Mexican populations, and of 0.46 ± 0.04 for the Pedregal population) was significantly higher than the threshold value of 0.5 determining the evolution of outcrossing (Lande and Schemske, 1985). Hence, the monomorphic enantiostyly of S. rostratum in the Pedregal population seems to have characteristics conferring resistance to the invasion of a dimorphic reproductive system: (1) the ability to attain relatively high outcrossing rates even under restricted pollen flow, (2) relatively low inbreeding depression and (3) a generalized pollen-limitation ecological scenario.

In accordance with the above arguments, the low frequency of dimorphic enantiostylous species (seven species in four genera and three families; Jesson and Barrett, 2003) seems to be associated with an ecological constraint related to the prevalence of pollen limitation. In this sense, Knight et al. (2005) reported that fruit-set was significantly pollen-limited in 63 % of a sample of 482 data records. Moreover, it is probable that the pollen limitation constraint is more accentuated in annual than in perennial species. Dimorphic enantiostylous species with annual life cycles may be more susceptible to shortages of pollen or pollinators because they lack the reproductive assurance mechanism provided by geitonogamy. In fact, six out of the seven dimorphic enantiostylous species have a perennial life cycle [Barberetta aurea, Wachendorfia brachyandra, W. thyrsifolia and W. paniculata (Haemodoraceae); Cyanella alba (Tecophilaeaceae); and Saintpaulia ionantha (Gesneriaceae)] (Ornduff and Dulberger, 1978; Dulberger and Ornduff, 1980; Simpson, 1990; Helme and Linder, 1992; Salvador et al., 2005). Heteranthera mulfiflora (Pontederiaceae) is the only annual plant with dimorphic enantiostyly (Horn, 2004).

Finally, it is clear that a thorough analysis of the conditions maintaining monomorphic enantiostyly requires the inclusion of the fitness benefits obtained through both male and female fitness (Charnov, 1982; Campbell, 1998, 2000; Goodwillie et al., 2005). Nonetheless, the effect of pollinator limitation and/or the scarcity of mating partners is expected to be more severe on male- (pollen export to conspecific stigmas) than on female-fitness, thus reducing the opportunities of dimorphic enantiostyly to invade. We therefore predict that the conditions favouring the evolution of dimorphic enantiostyly would be even more stringent once male fitness is taken into account (Jesson et al., 2003).

SUPPLEMENTARY DATA

Supplementary data are available online at https://academic.oup.com/aob and consist of the following. Table S1: Results of the regression analyses of phenological traits and reproductive success of individuals in a natural population of Solanum rostratum. Table S2: Mean (± s.e.) reproductive success and individual components of reproductive success for each floral density and reproductive system treatments.

Supplementary Table S1
Click here for additional data file. (31KB, doc)
Supplementary Table S2
Click here for additional data file. (45KB, doc)

ACKNOWLEDGEMENTS

We thank Rubén Perez-Ishiwara for help in the field and laboratory, and Santiago Benitez-Vieyra for statistical advice. This work was supported by the Programa de Apoyo a Proyectos de Investigación e Innovación Tecnológica (PAPIIT - UNAM) [IN212314]. We thank Posgrado en Ciencias Biológicas UNAM and CONACYT for scholarship to E.M.C.

LITERATURE CITED

  1. Almeida NM, Castro CC, Leite AV, Novo RR, Machado IC. 2013a. Floral polymorphism in Chamaecrista flexuosa (Fabaceae-Caesalpinioideae): a possible case of atypical enantiostyly?Annals of Botany 112: 1117–1123. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Almeida NM, Castro CC, Leite AV, Novo RR, Machado IC. 2013b. Enantiostyly in Chamaecrista ramosa (Fabaceae-Caesalpinioideae): floral morphology, pollen transfer dynamics and breeding system. Plant Biology 15: 369–375. [DOI] [PubMed] [Google Scholar]
  3. Ashman TL, Knight TM, Steets JA, et al. 2004. Pollen limitation of plant reperoduction: Ecological and evolutionary causes and consequences. Ecology 86: 2408–2421. [Google Scholar]
  4. Bates D, Mächler M, Bolker BM, Walker SC. 2015. Fitting linear mixed-effects models using lme4. Journal of Statistical Software 67: 1–48. [Google Scholar]
  5. Bretz F, Hothorn T, Westfall P. 2010. Multiple Comparisons Using R. Boca Raton, FL: Chapman & Hall/CRC Press. [Google Scholar]
  6. Campbell DR. 1998. Variation in lifetime male firness in Ipomopsis aggregata: test of sex allocation theory. The American Naturalist 152: 338–353. [DOI] [PubMed] [Google Scholar]
  7. Campbell DR. 2000. Experimental tests of sex-allocation theory in plants. Trends in Ecology and Evolution 15: 227–232. [DOI] [PubMed] [Google Scholar]
  8. Charnov E. 1982. The Theory of Sex Allocation. Princeton, NJ: Princeton University Press. [Google Scholar]
  9. Coombs G, Peter CI, Johnson SD. 2009. A test for allee effects in the self-incompatible wasp-pollinated milkweed Gomphocarpus physocarpus. Austral Ecology 34: 688–697. [Google Scholar]
  10. Darwin CR. 1862. On the Various Contrivances by which British and Foreign Orchids are Fertilised by Insects. London: John Murray. [Google Scholar]
  11. Dauber J, Biesmeijer JC, Gabriel D, et al. 2010. Effects of patch size and density on flower visitation and seed set of wild plants: A pan-European approach. Journal of Ecology 98: 188–196. [Google Scholar]
  12. Delmas CEL, Escaravage N, Cheptou PO, et al. 2015. Relative impact of mate versus pollinator availability on pollen limitation and outcrossing rates in a mass-flowering species. Plant Biology 17: 209–218. [DOI] [PubMed] [Google Scholar]
  13. Domínguez CA, Avila-Sakar G, Vázquez-Santana S, Márquez-Guzmán J. 1997. Morph-biased male sterility in the tropical distylous shrub Erythroxylum havanense (Erythroxylaceae). American Journal of Botany 84: 626–632. [PubMed] [Google Scholar]
  14. Dulberger R, Ornduff R. 1980. Floral morphology and reproductive biology of four species of Cyanella (Tecophilaeaceae). New Phytologist 86: 45–56. [Google Scholar]
  15. Endress PK. 2001. Evolution of floral symmetry. Current Opinion in Plant Biology 4: 86–91. [DOI] [PubMed] [Google Scholar]
  16. Fenster CB. 1995. Mirror image flowers and their effect on outcrossing rate in Chamaecrista fasciculata (Leguminosae). American Journal of Botany 82: 46–50. [Google Scholar]
  17. Fox J, Weisberg S. 2011. An R Companion to Applied Regression. Thousand Oaks, CA: Sage. [Google Scholar]
  18. Gao JY, Ren PY, Yang ZH, Li QJ. 2006. The pollination ecology of Paraboea rufescens (Gesneriaceae): a buzz-pollinated tropical herb with mirror-image flowers. Annals of Botany 97: 371–376. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Geber MA, Dawson TE, Delph LF. 1999. Gender & Sexual Dimorphism in Flowering Plants. Heilderberg: Springer-Verlag. [Google Scholar]
  20. Goodwillie C, Kalisz S, Eckert CG. 2005. The evolutionary enigma of mixed mating: occurrence, theoretical explanations, and empirical evidence. Annual Review of Ecology, Evolution, and Systematics 36: 47–79. [Google Scholar]
  21. Grindeland JM, Sletvold N, Ims RA. 2005. Effects of floral display size and plant density on pollinator visitation rate in a natural population of Digitalis purpurea. Functional Ecology 19: 383–390. [Google Scholar]
  22. Gryj EO, Dominguez CA. 1996. Fruit removal and postdispersal survivorship in the tropical dry forest shrub Erythroxylum havanense: ecological and evolutionary implications. Oecologia 108: 368–374. [DOI] [PubMed] [Google Scholar]
  23. Hedrick P. 2000. Genetics of Populations. Mississauga: Jones and Bartlett Publishers, Inc. [Google Scholar]
  24. Helme NA, Linder HP. 1992. Morphology, evolution and taxonomy of Wachendorfia (Haemodoraceae). Bothalia 22: 59–75. [Google Scholar]
  25. Horn C. 2004. Poaceae-Rubiaceae. In: Berry PE, Yatskievych PE, Holst BK, eds. Flora of the Venezuelan Guayana. St Louis, MO: Missouri Botanical Garden Press, 371–376. [Google Scholar]
  26. Jennersten OL. 1988. Pollination in Dianthus deltoides (Caryophyllaceae): effects of habitat fragmentation on visitation and seed set. Conservation Biology 2: 359–366. [Google Scholar]
  27. Jesson LK, Barrett SCH. 2003. The comparative biology of mirror-image flowers. International Journal of Plant Sciences 164: 237–249. [Google Scholar]
  28. Jesson LK, Barrett SCH. 2005. Experimental tests of the function of mirror-image flowers. Biological Journal of the Linnean Society 85: 167–179. [Google Scholar]
  29. Jesson LK, Barrett SCH, Day T. 2003. A theoretical investigation of the evolution and maintenance of mirror-image flowers. The American Naturalist 161: 916–930. [DOI] [PubMed] [Google Scholar]
  30. de Jong TJ, Waser NM, Klinkhamer PGL. 1993. Geitonogamy: the neglected side of selfing. Trends in Ecology and Evolution 8: 321–325. [DOI] [PubMed] [Google Scholar]
  31. Klinkhamer PGL, Jong TJ. 1991. Attractiveness to pollinators: a plant’s dilemma. Oikos 66: 180–184. [Google Scholar]
  32. Knight TM. 2003. Floral density, pollen limitation, and reproductive success in Trillium grandiflorum. Oecologia 137: 557–563. [DOI] [PubMed] [Google Scholar]
  33. Knight TM, Steets JA, Vamosi JC, et al. 2005. Pollen limitation of plant reproduction: pattern and process. Annual Review of Ecology, Evolution, and Systematics 36: 467–497. [Google Scholar]
  34. Kunin WE. 1997. Population size and density effects in pollination: pollinator foraging and plant reproductive success in experimental arrays of Brassica kaber. Journal of Ecology 85: 225–234. [Google Scholar]
  35. Lande R, Schemske DW. 1985. The evolution of self-fertilization and inbreeding depression in plants.I. Genetic models. Evolution 39: 24–40. [DOI] [PubMed] [Google Scholar]
  36. Lloyd DG, Schoen DJ. 1992. Self- and cross-fertilization in plants. I. Functional dimensions. International Journal of Plant Sciences 153: 358–369. [Google Scholar]
  37. Makino TT, Ohashi K, Sakai S. 2007. How do floral display size and the density of surrounding flowers influence the likelihood of bumble bee revisitation to a plant?Functional Ecology 21: 87–95. [Google Scholar]
  38. Menge EO, Greenfield ML, McConchie CA, Bellairs SM, Lawes MJ. 2017. Density-dependent reproduction and pollen limitation in an invasive milkweed, Calotropis procera (Ait.) R. Br. (Apocynaceae). Austral Ecology 42: 61–71. [Google Scholar]
  39. Murawski DA, Hamrick JL. 1991. The effect of the density of flowering individuals on the mating systems of nine tropical tree species. Heredity 67: 167–174. [Google Scholar]
  40. Ornduff R, Dulberger R. 1978. Floral enantiomorphy and the reproductive system of Wachendorfia paniculata (Haemodoraceae). New Phytologist 80: 427–434. [Google Scholar]
  41. Pannell JR. 2015. Evolution of the mating system in colonizing plants. Molecular Ecology 24: 2018–2037. [DOI] [PubMed] [Google Scholar]
  42. R Core Team. 2016. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. [Google Scholar]
  43. Richards AJ. 1997. Plant Breeding Systems. Cambridge: Cambridge University Press. [Google Scholar]
  44. Ritland K. 2002. Extensions of models for the estimation of mating systems using n independent loci. Heredity 88: 221–228. [DOI] [PubMed] [Google Scholar]
  45. Salvador ED, Minami K, Jadoski SO. 2005. Evaluation of different substrates on African violet (Saintpaulia ionantha Wendl.) growth. Acta Horticulturae 697: 125–132. [Google Scholar]
  46. Simpson MG. 1990. Phylogeny and classification of the Haemodoraceae. Annals of the Missouri Botanical Garden 77: 722–784. [Google Scholar]
  47. Solís-Montero L, Vergara CH, Vallejo-Marín M. 2015. High incidence of pollen theft in natural populations of a buzz-pollinated plant. Arthropod-Plant Interactions 9: 599–611. [Google Scholar]
  48. Spigler RB, Chang S-M. 2009. Pollen limitation and reproduction varies with population size in experimental populations of Sabatia angularis (Gentianaceae). Botany 87: 330–338. [Google Scholar]
  49. Sprengel CK. 1793. Das entdeckte Geheimniss der Natur im Bau und in der Befruchtung der Blumen I. Berlin: Friedrich Vieweg. [Google Scholar]
  50. Steffan-Dewenter I, Tscharntke T. 1999. Effects of habitat isolation on pollinator communities and seed set. Oecologia 121: 432–440. [DOI] [PubMed] [Google Scholar]
  51. Thomson JD. 1988. Effects of variation in inflorescence size and floral rewards on the visitation rates of traplining pollinators of Aralia hispida. Evolutionary Ecology 2: 65–76. [Google Scholar]
  52. Vallejo-Marín M, Solis-Montero L, Bacles CFE, Lepais O. 2011. Thirteen microsatellites developed by SSR-enriched pyrosequencing for Solanum rostratum (Solanaceae) and related species. American Journal of Botany 98: 296–299. [DOI] [PubMed] [Google Scholar]
  53. Vallejo-Marín M, Solís-Montero L, Souto Vilaros D, Lee MYQ. 2013. Mating system in Mexican populations of the annual herb Solanum rostratum Dunal (Solanaceae). Plant Biology 15: 948–954. [DOI] [PubMed] [Google Scholar]
  54. Vaughton G, Ramsey M. 2010. Pollinator-mediated selfing erodes the flexibility of the best-of-both-worlds mating strategy in Bulbine vagans. Functional Ecology 24: 374–382. [Google Scholar]
  55. Wang G, Yamasue Y, Itoh K, Kusanagi T. 1998. Outcrossing rates as affected by pollinators and the heterozygote advantage of Monochoria korsakowii. Aquatic Botany 62: 135–143. [Google Scholar]
  56. Zhang LJ, Lou AR. 2015. Pollen limitation in invasive populations of Solanum rostratum and its relationship to population size. Journal of Plant Ecology 8: 154–158. [Google Scholar]
  57. Zhao J, Solís-Montero L, Lou A, Vallejo-Marín M. 2013. Population structure and genetic diversity of native and invasive populations of Solanum rostratum (Solanaceae). PLoS ONE 8: 1–9. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Table S1
Click here for additional data file. (31KB, doc)
Supplementary Table S2
Click here for additional data file. (45KB, doc)

Articles from Annals of Botany are provided here courtesy of Oxford University Press

RESOURCES