Which cues are sexy? The evolution of mate preference in sympatric species reveals the contrasted effect of adaptation and reproductive interference

Abstract

Mate preferences may target traits (a) enhancing offspring adaptation and (b) reducing heterospecific matings. Because similar selective pressures are acting on traits shared by different sympatric species, preference-enhancing offspring adaptation may increase heterospecific mating, in sharp contrast with the classical case of so-called "magic traits." Using a mathematical model, we study which and how many traits will be used during mate choice, when preferences for locally adapted traits increase heterospecific mating. In particular, we study the evolution of preference toward an adaptive versus a neutral trait in sympatric species. We take into account sensory trade-offs, which may limit the emergence of preference for several traits. Our model highlights that the evolution of preference toward adaptive versus neutral traits depends on the selective regimes acting on traits but also on heterospecific interactions. When the costs of heterospecific interactions are high, mate preference is likely to target neutral traits that become a reliable cue limiting heterospecific matings. We show that the evolution of preference toward a neutral trait benefits from a positive feedback loop: The more preference targets the neutral trait, the more it becomes a reliable cue for species recognition. We then reveal the key role of sensory trade-offs and the cost of choosiness favoring the evolution of preferences targeting adaptive traits, rather than traits reducing heterospecific mating. When sensory trade-offs and the cost of choosiness are low, we also show that preferences targeting multiple traits evolve, improving offspring fitness by both transmitting adapted alleles and reducing heterospecific mating. Altogether, our model aims at reconciling "good gene" and reinforcement models to provide general predictions on the evolution of mate preferences within natural communities.

Keywords: good gene; mate preference; preference for multiple traits; speciation; species recognition; theory.

Figure 1

Schematic description of the model. (A) Life cycle. The evolution of the preference may depend on the interactions between species and natural selection acting on the preferred traits. Here, we assume two sympatric species, $A$ and $B$, depicted in this scheme as butterfly species with different wing shapes. The individuals can display two traits, $T_1$ and $T_2$, represented by the forewing and hindwing colors as an example. We study the coevolution of the trait values (0 or 1, shown as intense versus light color of the wings) and the preference in species $A$. We assume that all individuals in species $B$ displayed the trait value $1$ (intense color) at both traits. We assume a selection step, promoting trait value 1 at both traits in species $A$, increasing similarity with species $B$, where value 1 is fixed for both traits. We then assume a reproduction step, where the mating success of the different individuals in species $A$ depends on the traits and preferences carried by males and females. In particular, females of species $A$ may attempt costly and unfertile sexual interactions with males of species $B$ depending on their preferences. (B) Genetic basis of preference. Depending on the genotype at locus $M$, females modulate their level of attention toward either trait displayed by males (relative preference weighting$\gamma$). We also assume that the level of attention on one trait diminishes the attention on the alternative one. We investigate several shapes of this trade-off tuned by the parameter $a$.

Figure 2

Illustration of the mating process. We assume that females can mate at most once. We assume that each female meets sequentially random males, which can be either conspecifics or heterospecifics. At each encounter, the female accepts or rejects the male with a probability depending on the female’s preference and the male’s traits. After an encounter, a female accepts a conspecific (resp. heterospecific) male with probability $T$ (resp. $T_{\mathrm{ri}}$). If a female engages in heterospecific mating, she produces totally unviable offspring. Because females mate at most once, a female engaged in a heterospecific mating cannot recover the associated fitness loss. Moreover, we assume that females refusing a mating opportunity can encounter another male with a probability of $1-c$. We interpret $c$ as the cost of choosiness.

Figure 3

Evolution of relative preference weighting toward a trait under selection $T_1$ or a neutral trait $T_2$ (${\gamma }^{*}$), depending on the strength of natural selection acting on trait $T_1$ ($s_1$) and the strength of reproductive interference ($c_{\mathrm{ri}}$) when $T_2$ is neutral ($s_2=0$).

Figure 4

Decomposition of the fitness gradient into terms associated with (A) offspring survival ($S_{\mathrm{os}}$), (B) offspring reproductive success ($S_{\mathrm{or}}$), (C) reproductive interference ($S_{\mathrm{ri}}$), and (D) cost of choosiness ($S_{\mathrm{c}}$), depending on the resident relative preference weighting value (${\gamma }_{\mathrm{r}}$). We fixed the level of reproductive interference ($c_{\mathrm{ri}}=0.0025$) and assume that trait $T_1$ is under natural selection ($s_1=0.02$), while trait $T_2$ is neutral ($s_2=0$). We assume the cost of choosiness ($c=0.001$). When the line is green (resp. purple), the corresponding evolutionary force (offspring survival, offspring reproductive success, reproductive interference, and cost of choosiness) promotes the evolution of preference toward the neutral selection $T_2$ (resp. the trait under selection $T_1$). The more intense the color, the more intense the selection.

Figure 5

Evolution of relative preference weighting toward a trait under selection $T_1$ or a neutral trait $T_2$ (${\gamma }^{*}$) depending on the shape of the cognitive trade-off function (through the parameter $a$) and the cost of choosiness $c$ for different ancestral preferences. We assume ancestral preference targeting: (A) mainly trait $T_1$ (${\gamma }_{{\mathrm{t}}_0}=0.01$), (B) equally both traits (${\gamma }_{{\mathrm{t}}_0}=0.5$), and (C) mainly trait $T_2$ (${\gamma }_{{\mathrm{t}}_0}=0.99$). We assume reproductive interference ($c_{\mathrm{ri}}=0.0025$), that trait $T_1$ is under natural selection ($s_1=0.02$) and trait $T_2$ is neutral ($s_2=0$).

Figure 6

Evolution of relative preference weighting toward traits $T_1$ or $T_2$ (${\gamma }^{*}$), depending on the strength of natural selection acting on trait $T_1$ and $T_2$ ($s_1$ and $s_2$), for different strengths of reproductive interference ($c_{\mathrm{ri}}$).