Effects of sexual dimorphism on pollinator behaviour in a dioecious species

Floral traits often display sexual dimorphism in insect-pollinated dioecious plant species, with male individuals typically being showier than females. While this strategy is theorized to be optimal when pollinators are abundant, it might represent a risk when they become scarce, because the disproportionately high number of visits on the most attractive sex, males, might preclude efficient pollen transfer from males to females. Here, the effect of sexual dimorphism on pollination efficiency was assessed in experimental arrays of dioecious Silene dioica that were exposed to one frequent visitor of the species, Bombus terrestris, and that differed in the magnitude of sexual dimorphism for either flower number or flower size. While flower size dimorphism did not impact pollination efficiency, we found that flower number dimorphism negatively affected the number of visits on female plants, on female flowers and on the number of female flowers visited after a male flower. However, flower number dimorphism had no effect on the number of pollen grains deposited per stigma, presumably because the decrease in the number of visits to female flowers was compensated by a higher number of pollen grains deposited per visit.


34
One common feature in plants with separate sexes is sexual dimorphism, both in terms of 35 vegetative and reproductive traits (reviewed in Barrett and Hough 2013). In insect-pollinated dioecious 36 species, male plants often display more conspicuous floral phenotypes than females, at least in temperate 37 species, with males typically producing larger floral displays and/or larger flowers (Guitián 1995, Delph 38 et al. 1996, Pailler et al. 1998, Eckhart 1999 an abrupt decrease in pollinator abundance could have deleterious demographic effects well before any 54 evolutionary response could take place. Accordingly, the modelling study by Vamosi and Otto (2002) 55 predicts that pollinator preference towards one sex, in this case males, could lead to population 56 extinction when pollinators are scarce and when plants display a high degree of sexual dimorphism. 57 To date, only a few experimental studies have examined the effect of sexual dimorphism on 58 pollinator behaviour at the plant population level. By manipulating the distribution of floral size between 59 males and females in experimental arrays of the dioecious Sagittaria latifolia, Glaettli and Barrett (2008)

Experimental plants and bumblebee species 97
Silene dioica is a common perennial herb of north-western Europe, flowering from April to June (Kay 98 et al. 1984). This species has a generalist insect-pollination system, with Bombus species and syrphids 99 as most common visitors (Baker 1947). Female flowers have five stigmas and male flowers ten stamens. 100 Males typically carry showier displays than females: on average, they produce over ten times more 101 flowers over the course of the flowering season, with corollas that are 20% wider than in females 102 (Moquet et al. 2020). Both male and female flowers secrete nectar at the base of the corolla (Vogel 1998, 103 Comba et al. 1999, with male flowers producing less abundant but more concentrated nectar than 104 female flowers (Kay et al. 1984, Comba et al. 1999. We used the same experimental material as in 105  Bombus terrestris is a frequent visitor of S. dioica (Baker 1947, Kay et al. 1984, Goulson and Jerrim 111 1997. We used colonies of B. terrestris from BioBest Biological Systems (Westerlo, Belgium) in which 112 we removed access to the factory-supplied nectar reservoir during the experiment to encourage foraging. 113 Pollen was delivered ad-libidum directly to the nests. 114

Experimental design 115
To test whether the occurrence of sexual dimorphism in flower number and/or size impacted visitation 116 rates, pollinator behavior, and pollen transfer from male flowers to female stigmas, we ran experiments The day before the experiment, plants were placed in a greenhouse to exclude nectar-feeding insects and 133 allow nectar replenishment. Flower buds were marked with a water-based marker pen in order to identify 134 the virgin flowers that were used to estimate pollen loads after each observation session. bumblebees probing a flower was recorded as a visit. Bumblebees were captured and marked after they 8 visited ten flowers. In the rare cases where bumblebees did visit fewer than ten flowers and stopped 148 foraging for more than 10 minutes, the trial was stopped and a sixth trial was performed. Each session 149 consisted of exactly 50 visits (five bumblebees × ten visits) in the flight cage. Because the maximum 150 number of flowers per session was 50 for all treatments, all flowers had a chance to be visited at least 151 once. Some flowers were visited more than once, thus, the amount of reward of an individual flower 152 could decline during each session. This experiment thus mimics common situations in which nectar 153 secretion rates are weak in comparison to visitation rates (Real 1981

Does sexual dimorphism affect the distribution of visits on male and female plants? 208
When comparing the number of visits received per plant between the "flower number dimorphism" and 209 the "no dimorphism" treatment, we detected an effect of both plant sex (χ 1 2 = 9.604, P = 0.002) and the 210 interaction between plant sex and treatment (χ 1 2 = 39.593, P < 0.001). In the sessions with no sexual 211 dimorphism, female plants received slightly more visits than male plants, and this pattern was reversed 212 when male plants carried more flowers ( Fig. 2A). Only plant sex had a significant effect (χ 1 2 = 5.816, P 213 = 0.016) on the number of visits per flower, with higher visitation rates on females in both treatments 214 Regarding the comparison between the "flower size dimorphism" and the "no dimorphism" treatments, 216 although the number of visits per plant and per flower was similar in males and females when males 217 produced larger flowers, the interaction between plant sex and treatment was not significant (Fig. 2C  218 and 2D). 219

Does sexual dimorphism impact the characteristics of visitation sequences? 220
Bumblebee behavior was found to be statistically similar between the "flower size dimorphism" and the 221 "no dimorphism" treatments for all tested variables. On the contrary, when comparing bumblebee 222 behavior between the "flower number dimorphism" and the "no dimorphism" treatment, we found 223 significant differences for several metrics (Fig. 3) The preference for females in the "no dimorphism" treatment resulted in a higher number of 288 sequences starting by visits on one or several female flowers. This was mainly due to a probabilistic 289 effect: with more female flowers in the sequence, the probability of starting by a female flower increases. 290 In addition, we found that some sequences featured less females visited after a male than expected under 291 random behaviour, which could be driven by a slight active preference for female flowers. As a result, 292 for 38% of visited female flowers in the « no dimorphism » treatment, the pollination sequence did not 293 feature an upstream visit to a male flower, highlighting the fact that a higher attractivity of females might 294 negatively impact pollination efficiency. 295 296 Flower size dimorphism had no effect on pollinator visitation patterns 297 Flower size has been shown to play a role in pollinator attraction in many other systems (e.g. Bell 298 1985, Galen and Newport 1987, Stanton and Preston 1988). This trait can indeed directly impact the 299 intensity of the visual signal and/or the quantity of reward for pollinators in systems where flower size 300 is associated with reward abundance (Blarer et al. 2002, Brunet et al. 2015. Since no significant 301 preference for males was found when they carried larger flowers, the effect of this trait on pollinator 302 attraction was not supported in our study. One possible explanation is that, if sugar quantity was on This result is consistent with the theoretical predictions made by Vamosi and Otto (2002), who argued 350 that decreasing the relative attractivity of female plants should diminish the number of pollinating visits. 351 One other effect of this higher number of visited male flowers could be that pollinators accumulate a 352 high number of pollen grains on their body. Pollen load on insects should depend on various attributes 353 of both pollen grains and insect morphology and behaviour, which was not be measured during the 354 current study. Pollen carryover is generally considered to be high in flowering plants (Robertson 1992). 355 This could explain why no difference was found in the average number of pollen grains deposited on 356 stigmas among treatments. We can suggest that, while the number of potentially pollinating visits 357 decreased with sexual dimorphism, each visit might have been more efficient in terms of pollen deposit.