Ecological speciation can be driven by divergent natural and/or sexual selection. The relative contribution of these processes to species divergence, however, is unknown. Here, we investigate how sexual selection in the form of male and female mate preferences contributes to divergence of body size. This trait is known be under divergent natural selection and also contributes to sexual isolation in species pairs of threespine sticklebacks (Gasterosteus aculeatus). We show that neither female nor male size preferences contribute to body size divergence in this species pair, suggesting that size-based sexual isolation arises primarily through natural selection.
Ecological speciation occurs when divergent selection reduces gene flow between populations that exploit different ecological niches (Schluter 2000, Rundle & Nosil 2005). Natural selection can drive ecological speciation when adaptation to different environments alters traits also used in mate recognition, leading to assortative mating among populations (Schluter 2000). Alternatively, sexual selection can cause divergence when different environments alter sexual selection on male sexual displays and/or female preferences for these displays (Endler 1992, Boughman 2002).
While it is clear that both natural and sexual selection play a role in ecological speciation, their relative importance remains a major outstanding question in speciation research (Nosil et al. 2007). They may interact to determine the direction and magnitude of evolutionary response in reproductive isolation (Blows 2002), and sexual selection may be most effective when operating alongside natural selection (Ritchie 2007). Most studies investigating how sexual selection causes speciation have focused on secondary sexual traits where natural selection can be excluded as a primary cause of divergence. However, we need to explore how natural and sexual selection operate on the same traits to produce reproductive isolation, to fully understand this process.
Species divergence in body size can result from natural selection due to differences in predator regime (Nosil & Crespi 2006), foraging resources (Schluter 1993) or parasites (Maccoll 2009). However, sexual selection may also influence the evolution of body size (Andersson 1994). Body size is often an important criterion in mate choice, and can influence success during intrasexual mate competition (Hunt et al. 2009). Thus, both sexual and natural selection may be important in driving species divergence in body size, and the resulting size divergence can lead to sexual isolation.
We examine the role of mate choice in body size divergence in a species pair of threespine stickleback (Gasterosteus aculeatus). Limnetic and benthic sticklebacks co-occur in a number of lakes throughout British Columbia (McPhail 1994). Both natural and sexual selection have been implicated in the evolution of sexual isolation between these species pairs (Rundle et al. 2000; Boughman 2001). Natural selection for improved feeding efficiency has led to divergence in body size and shape. Limnetics have small, slender bodies suited to foraging in open water, and benthics have large deep bodies suited to foraging on prey attached to the sediment or vegetation (Schluter 1993). Body size differences between species are important to reproductive isolation; the probability of hybridization is higher between similar-sized individuals (Nagel & Schluter 1998; Boughman et al. 2005). This pattern is also seen in stream-anadromous species pairs of sticklebacks (McKinnon et al. 2004), indicating the importance of body size in maintaining reproductive isolation throughout the stickleback species complex. However, whether sexual selection contributes to size-based sexual isolation is unknown. Variation in body size preference among threespine stickleback populations exists with some populations showing no size preference (Kraak et al. 1999; Baube 2008). Such differences in size-based mate choice between species could cause both divergent selection on body size and size-based sexual isolation. Here, we ask whether limnetic and benthic sticklebacks differ in male or female size preferences. Because benthics are larger than limnetics, we expect benthics to prefer larger mates and limnetics to prefer smaller mates. We infer the direction of evolutionary change by including the anadromous ancestors.
MATERIAL AND METHODS
Size preferences of limnetic, benthic and anadromous sticklebacks were assessed in no-choice mating trials. No-choice trials are standard for measuring mating preferences in sticklebacks (Nagel & Schluter 1998) and are useful for assessing the strength of preferences (Wagner 1998). For details on experimental setup, see the electronic supplementary material.
Courtship behaviours recorded included male zig-zagging, biting and leading and female following (described in Rowland 1994). We calculated female preference as the proportion of male leads that a female followed to account for variation in male courtship behaviour. For male courtship intensity, we calculated the rate of zig-zagging and biting per second. To estimate the probability of spawning, we recorded whether the female entered the male's nest.
Standard length of males and females was measured with vernier calipers accurate to 0.2 mm. Males were measured before and after behavioural trials and measurements were averaged for analyses. Females were measured after trials.
We measured courtship, preferences and mating success of 60 male and female pairs from each population, but excluded trials where females or males were unresponsive. Consequently, we obtained a sample size of 56 anadromous, 45 benthic and 48 limnetic trials. We used only one observation for each fish.
All analyses were performed using SPSS 16.0. Male and female mate preferences were square root transformed when appropriate to ensure normality and analysed using ANCOVA. The probability of spawning was analysed using logistic regression; we estimated effect sizes in the same manner as for other variables (see below). We investigated the effects of species, male length, female length and all interaction terms on male and female mate preferences and on spawning probability. For all models, non-significant terms were sequentially removed until only significant sources of variation remained. Divergent sexual selection based on differences in male and female size preferences would be evident as an interaction between the species and the size of the chosen fish. To examine mate preferences within species when such interactions exist, we visualize data using thin-plate splines and conduct separate regressions for each species.
To determine effect sizes, we calculated partial eta squared and its confidence intervals (ηp2 ± 95% CI) using ESCI (http://psychology.anu.edu.au/people/smithson/details/CIstuff/CI.html) for non-significant and significant terms in the models. Effect sizes were calculated from F values either when the term was removed from the model (non-significant terms) or from the final model (significant terms).
Female response to males did not depend on male size for any species. The probability of a female following a male did depend on her own size (figure 1), with larger females more likely to follow males. However, this effect accounted for less than 3 per cent of the total variability in female follows per lead (ηp2 = 0.0278).
Male zig-zag courtship did not depend on either male or female size (figure 1) despite differences in the overall rate of zig-zagging between species (figure 1). Limnetic and benthic males zig-zagged at a higher rate (untransformed mean ± s.e.: limnetic = 0.079 ± 0.014; benthic = 0.043 ± 0.007) than anadromous males (0.025 ± 0.004) (Tukey's test: limnetic versus benthic p = 0.091, limnetic versus anadromous p < 0.001 benthic versus anadromous p < 0.000).
Male biting courtship showed a significant three-way interaction between species, female length and male length (figures 1 and 2). This effect is due to a weak but significant interaction between male and female length for limnetics (β = −0.002, t = −2.182, p = 0.035), but not for benthic or anadromous sticklebacks (benthic: β = −0.000, t = −0.237, p = 0.814; anadromous: β = 0.002, t = 1.123, p = 0.266).
Despite size-related differences in male and female courtship behaviour, the probability of spawning was not dependent on male length, female length or their interaction. Spawning probability did, however, depend on species (figure 1); anadromous pairs spawned more often (50%) than benthics (20%) or limnetics (27%) (logistic regression: anadromous versus benthic p = 0.003, anadromous versus limnetic p = 0.019, benthic versus limnetic p = 0.423).
Benthic and limnetic sticklebacks differ markedly in size and this size difference is critical to sexual isolation between species (Nagel & Schluter 1998; Boughman et al. 2005). We find no evidence for size-based assortative mating within species, indicating that mate preferences within species are not responsible for size-based sexual isolation.
Neither benthic nor limnetic females showed size-based mating preferences. The strongest size effect was increased male biting with increased female size, which was especially pronounced for small limnetic males. However, this is unlikely to contribute to divergence in body size. First, this effect is too small to explain large size differences between species. Second, previous research has shown that males tailor their courtship to the species of female. Limnetic males court benthic females more aggressively than limnetic females (Kozak et al. 2009). Our results show that this adjustment in male behaviour extends to limnetic females that are in the size range of benthic females. This could be caused by mate recognition errors or by males tailoring courtship towards female display preferences. Either way, spawning probability does not depend on either male or female size, indicating that mate preferences do not drive size divergence between species.
This contrasts with earlier findings on another key contributor to sexual isolation in this system: male breeding coloration. Sexual selection has caused divergence in male breeding coloration due to varying conspicuousness of this trait in different light environments, and limnetic and benthic females differ in colour preference (Boughman 2001). The correlation between preference strength and male colour suggests that divergent colour preferences drive differences in male colour between species, and differences in both preference and colour cause sexual isolation (Boughman 2001). It appears that within species mate preferences are integral to colour-based sexual isolation (Boughman et al. 2005). Thus, sexual selection plays a different role in colour divergence from that in size divergence.
While we can rule out divergent size preferences, sexual selection may contribute to size differences and size-based sexual isolation through two other mechanisms: male–male competition or mate preferences for a correlated trait. Male–male competition may play a role in speciation (Seehausen & Schluter 2004), but is often overlooked. Body size has been shown to influence success in male–male competition in many taxa (Hunt et al. 2009) including sticklebacks (Kraak et al. 1999). Differences between benthic and limnetic breeding habitats (McPhail 1994) may influence the intensity of male–male competition; limnetics are known to be more aggressive towards rivals than benthics (Ridgway & McPhail 1987). Thus, differences in male–male competition could exert divergent selection on body size in the two species.
Mating preferences may, however, influence body size divergence if they act on correlated traits. A likely correlated trait is body shape. Shape differences between these species are well established (McPhail 1994) and could provide a reliable cue to species identity. Whether body shape is used in mate choice is unknown, but is worth investigating, especially given theoretical predictions that preferences for locally adapted mates should evolve and contribute to sexual isolation (Servedio 2004).
Theoretical predictions and empirical observations highlight the interplay between sexual and natural selection and their dual roles in the evolution of sexual isolation. It appears natural selection is key in determining body size divergence in this threespine stickleback species pair. However, we need more work on male–male competition and preferences for correlated traits before sexual selection can be completely ruled out. An interesting avenue for further work is why natural and sexual selection act so differently on the two primary traits that isolate these species.
Thanks to John Irwin for fish maintenance. Thanks to Iain Barber, Genny Kozak and Leeann Reaney for comments on the manuscript. Funding was provided by US National Science Foundation grants to J.W.B.