Testing the sexy son hypothesis--a research framework for empirical approaches

Sexual selection by direct and/or indirect benefits as well as sexual conflict determines the evolution of animal mating systems (e.g., Andersson and Simmons 2006). In polygynous mating systems, the female preference of apparently useless or deleterious male characteristics can be understood if a correlation between preferred male characteristics and direct benefits for the female exists (see polygyny threshold model, e.g., Orians 1969), such as improved paternal care or territorial quality (e.g., Price et al. 1993; Iwasa and Pomiankowski 1999). This issue is particularly interesting in facultative polygynous mating systems with biparental care because the female’s choice of a polygynous male as a mate is most often accompanied by a reduction in paternal care by this mate (for a review, see Ligon 1999). Several studies (e.g., Johnson et al. 1994; Pribil and Searcy 2001; Moreno et al. 2002; but see Both 2002) have observed a reduction in offspring number and/or offspring performance in females mated with a polygynous male, thus raising doubt in the relevancy of the polygyny threshold model (see, e.g., Orians 1969) in these instances. More generally, females of most species do not seem to take any appreciable direct benefits of their mate choice (Kirkpatrick and Ryan 1991; Andersson 1994). Since the 1970s, indirect benefits are discussed as a possibility to explain female choice (Kirkpatrick and Ryan 1991; Andersson and Simmons 2006). Weatherhead and Robertson (1979) proposed the sexy son hypothesis (SSH) as an additional explanation for the establishment of polygyny. According to this hypothesis, females are assumed to compensate for their direct inferior breeding success by the increased number of descendents produced by their sexy—and therefore polygynous—sons (see Table 1). The idea of focusing on fitness advantages as a result of male attractiveness is similar to the ‘‘Fisherian runaway selection process’’ (see, e.g., Andersson and Simmons 2006). The ‘‘good genes’’ model (see Table 1), that is, the idea of an indirect benefit due to the male’s genetic quality, is a second, complementary hypothesis that is discussed as a possibility to explain female choice by indirect benefits (Andersson and Simmons 2006; Charmantier and Sheldon 2006). Both hypothesized mechanisms, SSH and good genes, belong to the same continuum of indirect selection (Kokko et al. 2002; Radwan 2002). SSH as well as good genes postulate that indirect benefits due to male characteristics in a specific species will be similar for all females of this species (see also Table 1). Compatible genes, on the other hand, try to explain intraspecific differences in female mate choice (see Table 1). According to this theory, each individual female tries to enhance its fitness by choosing a male with a genetic pattern that is compatible with its own genetic configuration (e.g., Neff and Pitcher 2005). Another approach to understand mating behavior that results in deleterious female reproductive success, such as offspring number or performance, is the idea of sexual conflict (see Table 1; Gavrilets et al. 2001; Cameron et al. 2003; Arnqvist and Rowe 2005; Andersson and Simmons 2006; Lessells 2006; Parker 2006; Rice et al. 2006). Sexual conflict is driven by the opposed reproductive interests in females and males that result in sexually antagonistic behavior and coevolving adaptations. Whereas the traditional hypotheses of sexual selection, such as the polygyny threshold model and SSH, discuss the importance of direct versus indirect benefits, sexual conflict considers the possibility that polygynous males are not beneficial to females, neither directly nor indirectly. For example, Arnqvist and Kirkpatrick (2005) demonstrated that extrapair copulations, which do not lead to any direct advantages such as additional food resources or shelter, in socially monogamous birds are in line with sexual conflict assumptions as indirect benefits are much smaller than the direct negative selection caused by depressed paternal investment by social males. The present study will focus on SSH because SSH as well as the good genes assumption are based on the relative magnitude of direct and indirect effects of mate choice and thus have become relevant to the recent debate of whether sexual conflict can result in sexually antagonistic selection. SSH developed by Weatherhead and Robertson (1979), further called ‘‘narrow-sense SSH,’’ refers to mating systems with biparental care. In these mating systems, females mated with a polygynous male normally receive less assistance than females mated with a monogamous male (see, e.g., Ligon 1999) and thus suffer from direct fitness consequences that have to be (over)compensated by their sexy sons. On the other hand, ‘‘broad-sense SSH’’ encompasses polygyny and promiscuous mating systems with and without biparental care. Alatalo (1998) argues that the costs of any additional choice may be so minor that female choice for honestly signaling males, that is good genes, may evolve even if the indirect benefits on offspring quality are small. A similar argument can be given for SSH if mates of attractive males do not suffer any direct fitness consequences. Since it was first proposed by Weatherhead and Robertson (1979), SSH has been discussed vividly. Several theoretical studies either support (Weatherhead 1994; Kokko et al. 2002; McNamara et al. 2003) or attempt to argue against SSH (e.g., Kirkpatrick 1985; Wagner 1994; Alatalo and Ratti 1995). Recent sophisticated empirical studies dealing with SSH have reached contrasting results (compare Gwinner and Schwabl 2005; Head et al. 2005; Orteiza et al. 2005; Gustafsson and Qvarnström 2006; Huk and Winkel 2006). However, these studies differ in terms of the explicit research questions they ask and thus highlight the importance of questioning precisely what factors should be measured. The aim of the present paper is to shed light on the contrasting empirical results by analyzing the appropriateness of various research questions in empirical tests of SSH and in so doing to differentiate between ‘‘critical predictions’’ and ‘‘compatible predictions’’ (see Table 2). Critical predictions should be studied in future work as they address topics that are at the core of the predictions of SSH. Critical predictions are therefore able to validate or invalidate SSH as a hypothesis explaining female choice. Compatible predictions are aspects that can be discussed in the light of this hypothesis without directly addressing the critical predictions. Thus, compatible predictions only test for ‘‘by-product’’ predictions that can neither validate nor invalidate SSH. Behavioral Ecology doi:10.1093/beheco/arm150 Advance Access publication 18 January 2008