Nonindependent mate choice in monogamy

Individuals usually mate with opposite-sex others based on their own assessment of prospective mates’ suitability (reviews in Kempenaers 2007; Roberts and Little 2008), but this assessment can also be modulated by observing decisions of others—so-called nonindependent mate choice. We have proposed the term ‘‘mate quality bias’’ to describe the type of nonindependent mate choice that occurs when a female biases her own evaluation of a male in accordance with his mate’s quality (Vakirtzis and Roberts 2009). This type of nonindependent choice should be expected to occur in monogamous or relatively monogamous species where, due to mutual choice, there will usually exist a high correlation between a male’s quality and his mate’s quality (Trivers 1972; Burley 1977; Johnstone 1997). In these species, the most desirable males will tend to mate with the most desirable females and less desirable individuals will be left to mate among themselves (Burley 1983; Jones and Ratterman 2009). In principle, this selective mapping between male and female quality will not obtain in promiscuous and polygynous species, where, due to minimal male choice and high male mating skew, the top males will mate with all willing females, whereas lower quality males will achieve fewer or no mating opportunities. In these species, it is thus unlikely that an observing female will deduce reliable information about the male from the quality of his mate; rather, the frequency of partners and/or copulations may be a more useful cue. Use of such cues in mate assessment is known as mate choice copying (Pruett-Jones 1992; Dugatkin 1998). This is not to say that observing females in these nonmonogamous species should copy the choices of all females in the population. For example, focal females should ignore the choices of model females who are young and sexually inexperienced and may thus have an intolerably high error component in their choice of mates (Dugatkin and Godin 1993; Nordell and Valone 1998). The outcome of such facultative copying will be that the males with whom these immature females mate experience a conditional probability of choice that approximately equals their absolute probability (for definitions, see Pruett-Jones 1992). In other words, their probability of being chosen after being observed with the immature female should neither increase nor decrease in comparison to what it would be had they been observed alone (Dugatkin and Godin 1993; Amlacher and Dugatkin 2005; Vukomanovic and Rodd 2007). Mate quality bias and this type of facultative mate choice copying therefore both involve aspects of female quality (Witte and Godin 2010). However, the idea of mate quality bias is based on a particular set of assumptions and conditions that will usually be found in monogamy, where, in contrast to more promiscuous mating systems, mate choice copying is unlikely to evolve (see Table 1). Because mate choice copying and mate quality bias start from different background assumptions, they inevitably lead to different predictions (though there will be cases where they produce superficially similar results). To see this, let a male enjoy an absolute probability of choice by a given female equal to Pb, a constant between 0 and 1. In mate quality bias, we move from independent choice to a limited set of contexts in which the female, instead of assessing the quality of the male, evaluates instead the more easily assessable quality of his mate (for examples of these contexts, see Vakirtzis and Roberts 2009 and below). This results in a conditional probability that is, theoretically, a continuous and monotonically increasing function h of female quality that ranges from 0 to 1, with h#(x) . 0 for all x. By the intermediate value theorem, there must be a value x0 of female quality for which h(x0) = Pb. Moreover, because the function is monotonically increasing, x0 must be unique. All values of female quality larger than x0 will therefore satisfy h(x) . Pb, whereas h(x) , Pb for every x , x0. There exists therefore in mate quality bias the potential for a male’s mating success (with a certain range of females) to actually lower his probability of choice, a possibility that is absent from mate choice copying, facultative or not. Recall that in copying the only way a male can lower his probability of choice is if he is rejected by a female (Pruett-Jones 1992; Witte and Ueding 2003); in mate quality bias, this will happen when a male is accepted by a female. This analysis leads to the following prediction: In mate quality bias, the larger the male’s absolute probability of choice, the larger the proportion of the entire female population that, when mated to the male, decreases his probability of choice. (Equivalently, the smaller the male’s absolute probability of choice, the larger the proportion of the entire female population that, when mated to the male, increases his probability of choice.) Put another way, the higher the quality of the male the higher the corresponding ‘‘parity value’’ of female quality over which the male starts to increase his probability of choice and under which he starts to lower it. Symbolically, for every Ph . Pb, there must exist an x1 . x0 for which h(x1) = Ph, and letting F represent the cumulative distribution function of female quality in the population, it follows that F(x0) , F(x1) (see Figure 1). Is there empirical support for this prediction? Yes. Waynforth (2007) had a sample of men’s facial photographs rated for attractiveness by female subjects (a measure of absolute probability). Two weeks later, the female subjects re-rated the male images, but this time they also were simultaneously shown a facial image of each man’s supposed girlfriend (conditional probability). The female stimuli that were used as girlfriends had been randomly assigned to each male stimulus and covered a wide attractiveness range from very attractive to very unattractive. Waynforth initially tested whether the attractiveness ratings of men presented with a girlfriend increased compared with when they had been presented alone, as a mate choice copying hypothesis would predict. He could find no change, a negative result which corroborates earlier studies that had used different methodologies (Uller and Johansson 2003; Milonoff et al. 2007). A meaningful pattern in the data only emerged when Waynforth examined the effect of the