Causes of between-individual differences in mating preferences

Many species show substantial between-individual variation in mating preferences, but studying the causes of such variation remains a challenge. For example, the relative importance of heritable variation versus shared early-environmental effects (like sexual imprinting) on mating preferences has never been quantified in a population of animals. Here, we estimate the heritability of and early-rearing effects on mate choice decisions in zebra finches based on the similarity of choices between pairs of genetic sisters raised apart and pairs of unrelated foster sisters. We found a low heritability of preference functions and no evidence for early-rearing effects. A literature review confirms that heritable variation is low in most species and that there is no good evidence for sexual imprinting on continuous variation in any animal (except humans). This highlights that within populations, heritable variation and sexual imprinting play a limited role in explaining between-individual variation in preferences. While effects on preference functions were weak, we found strong individual consistency in choice behaviour and part of this variation was heritable. It seems likely that variation in choice behaviour (choosiness, responsiveness, sampling behaviour) would produce patterns of non-random mating and this might be the most important source of between-individual differences in mating patterns. Uunderstanding the origin of mating preferences is a critical issue for understanding sexual selection (Darwin 1859; Andersson & Simmons 2006). Mating biases can arise from several sources and it is conceptually important to distinguish between them (Jennions & Petrie 1997). Most importantly, mating preferences can be separated into preference functions, i.e. the ranking order of stimuli, and choosiness, i.e. the investment into mating with the preferred stimulus (Jennions & Petrie 1997; Brooks & Endler 2001). To understand sexual selection it is important to disentangle these different sources and to study their transmission mechanisms. Estimates of the heritability of mating preferences are particularly desirable, since this is a critical assumption of 18 Chapter 1: Heritability of and early-rearing effects on preferences runaway selection (Fisher 1930; Lande 1981). Mating preferences might also be transmitted by cultural processes, in particular by sexual imprinting (Aoki et al. 2001), which so far has received mixed support (see below). Substantial support for a genetic basis of preference functions comes from betweenpopulation differences in preferences and within-populations differences in preferences for dichotomous traits (Majerus et al. 1982; Houde & Endler 1990; Bakker & Pomiankowski 1995; Velthuis et al. 2005). However, it is not always clear how betweenpopulation differences translate into variation of preference functions within a single population (Chenoweth & Blows 2006). Within populations, there is very good evidence for heritable variation in choosiness (the range of stimuli that are accepted) or sexual responsiveness (the strength of response to sexual stimuli) (Collins & Cardé 1990; Bakker 1993; Bakker & Pomiankowski 1995; Brooks & Endler 2001; Brooks 2002; Rodrı́guez & Greenfield 2003). Although this might result in mating biases and is thus relevant for mate choice and sexual selection, it is not the same as quantifying the heritability of preference functions, since in order to measure preference functions, it is necessary to measure aspects of mate discrimination (Jennions & Petrie 1997). The number of studies addressing the genetic basis of preference functions within populations is very limited. Evidence for heritable variation comes from selection lines in stalk-eyed flies and in guppies (Wilkinson & Reillo 1994; Brooks & Couldridge 1999). Other studies have tested for genetic effects on preferences and have found (Moore 1989; Charalambous et al. 1994; Houde 1994) or have not found significant effects (Johnson et al. 1993; Breden & Hornaday 1994; Ritchie et al. 2005). These studies, however, do not quantify the amount of heritable variation. A careful search for studies that estimated the heritability of preference functions revealed only seven studies that present heritability estimates (Table 1.1): Three of them showed significant heritability (point estimates for h between 0.14 and 0.51), while the others were non-significant and mostly very low. Importantly, Qvarnström et al. (2006b) found significant heritability of female realised pairing with respect to a male trait in the collared flycatcher, although this heritability was very low (h = 0.026). However, such similarities between related females might also have arisen from sexual imprinting. Sexual imprinting is another form of vertical transmission of preferences, where preferences are formed during early development usually by using the own parents as models (Grant & Grant 1997; Aoki et al. 2001). There is very good evidence for sexual imprinting on heterospecific foster parents, morphs or novel-ornaments (e.g. Immelmann 1975; ten Cate & Bateson 1988; Qvarnström et al. 2004; Burley 2006), traits that are probably related to species recognition. The evidence for sexual imprinting on continuous variation within a single population (and thus not involving the categorisation of individuals into distinct classes) is limited and ambiguous (Bereczkei et al. 2004; Schielzeth et al. 2008b). Between-individual differences in mating preferences 19 Table 1.1: Published studies that present within-population heritability estimates of preference functions. We included only studies that analyse preference functions for traits that vary continuously within a population (although often only preferences for extremes were tested) and that estimate the heritability for the discrimination of mating stimuli (excluding estimates for the strength of a response to stimuli without discrimination). Manipulation Choice of preferred Heritability Study Species Preference for method trait (± SE) Remarks Collins & Pink bollpheromone sequenyes 0.14± 0.05 Cardé (1989) worm composition tial (three component ratios) Jang & Moth pulse rate and simultyes 0.21± 0.13 Greenfield asynchrony anuous (2000) interval of calls Iyengar Arctiid body size simultsamples of 0.51± 0.11 sex chromoet al. (2002) moth anuous defined some linked differences inheritance Gray & Field pulses per trill simultyes 0.34± 0.17 Cade (1999) cricket in calls anuous Simmons Field long chirp simultyes < 0.00 (2004) cricket relative to shortanuous chirp song elements Hall et al. Guppy male attractivesimultsamples −0.07± 0.13 selection (2004) ness (as measanuous including lines (direct ured in choice extremes and indirect chamber) selection) Brooks & Guppy colouration simultno 0.10± 0.11 maximal Endler (2001) & size (several anuous heritability measures) max from a large for brightness range of contrast tests Most studies, in particular those on the heritable variation of preference functions, have focused on specific traits and have used manipulation (Collins & Cardé 1989; Gray & Cade 1999; Jang & Greenfield 2000; Simmons 2004; Ritchie et al. 2005) or have chosen extreme phenotypes (Houde 1994; Wilkinson & Reillo 1994; Brooks & Couldridge 1999; Hall et al. 2004) to increase the variance along a specific axis of ornamentation. This valuable approach, however, does not allow an understanding of sources of varia20 Chapter 1: Heritability of and early-rearing effects on preferences tion in preferences for potential mates in their full multidimensionality (Candolin 2003; Fawcett & Johnstone 2003). Hence, we have employed an experimental design that allows testing for the similarity in preferences between (a) genetic sisters and (b) foster sisters in a population of zebra finches, where males are likely to vary along multiple axes. A full-individual cross-fostering scheme ensured that all individuals grew up with only unrelated nest siblings and were raised by unrelated foster parents. This enabled us to disentangle genetic and early-rearing effects. For the first time in a population of animals, we quantify the genetic and early-rearing effects on mating preferences simultaneously. We focus on preference functions, but at the same time present results on between-individual variation in choice behaviour.