«Dissertation a¨ Fakult¨ t fur Biologie ¨ Ludwig-Maximilians-Universit¨ t Munchen a ¨ durchgefuhrt am ¨ Max-Planck-Institut fur Ornithologie ...»
in mating preferences
Fakult¨ t fur Biologie
Ludwig-Maximilians-Universit¨ t Munchen
Max-Planck-Institut fur Ornithologie
Erstgutachten: Prof. Dr. Bart Kempenaers
Zweitgutachten: Prof. Dr. Susanne Foitzik
Eingereicht am: 17.12.2008
Tag der mundlichen Prufung: 28.05.2009
Chapter 1 Heritability of and early-environmental effects on variation in mating preferences................................. 17 Chapter 2 Sexual imprinting on continuous variation: do female zebra ﬁnches prefer or avoid unfamiliar sons of their foster parents?....... 29 Chapter 3 Assortative versus disassortative mating preferences of female zebra ﬁnches based on self-referent phenotype matching......... 41
Chapter 4 Variation in personality traits and their relevance for sexual selection:
a study on captive zebra ﬁnches..................... 53 General discussion.................................... 73 References......................................... 83 Summary.......................................... 107 Acknowledgments.................................... 109 Curriculum vitae..................................... 111 General introduction Sexual selection, mate choice and mating preferences exual selection, the intra-speciﬁc selection for successful reproduction, is a mecha- S nism that helps explaining the evolution of exaggerated characters that appear coun- terintuitive, when seen from the natural selection perspective alone (Darwin 1859). Dar- win himself proposed sexual selection as an additional force of evolution and dedicated an entire book to it (Darwin 1871). Although conceptually important, there is no strict distinction between natural and sexual selection, since both are, ultimately, about the representation of an individual’s own alleles in future generations (Andersson 1994).
In this sense, ’survival of the ﬁttest’ (Darwin 1859) always pertains to the ability of an individual to leave copies of it’s own alleles.
Sexual selection can be conceptually separated into intra-sexual competition and mate choice (Andersson 1994). Intra-sexual competition describes the competition for access to individuals of the other sex and can take different forms (e.g. contest, scramble, endurance rivalry or coercion, Andersson & Iwasa 1996). Competition is usually most intense in the sex that shows greater variance in reproductive success (Bateman’s principle, Bateman 1948), which is usually also the sex that invests less in each individual offspring (Trivers 1972). Mate choice, the inter-sexual component of sexual selection, describes the selection of phenotypes imposed by one sex (usually the one with lower variance in reproductive success) on the other (Andersson 1994). The variance in reproductive success is generally higher in systems with multiple matings and hence lower in strictly monogamous species. However, even in socially monogamous species there is often a fair amount of extra-pair paternity, which tends to increase the betweenindividual variance in reproductive success (Reynolds 1996; Petrie & Kempenaers 1998).
General introduction 4 Terminology In a very broad sense, mate choice or mating biases can be deﬁned as ’a process leading to the tendency of members of one sex to mate non-randomly with respect to one or more varying traits in members of the other sex’ (Kokko et al. 2003). This deﬁnition does not require any active choice and includes ’passive acceptance of the ﬁrst conspeciﬁcs encountered’ (Kokko et al. 2003). Such a deﬁnition, however, encompasses all of sexual selection and is hence too broad to help understanding the mechanisms that lead to non-random mating. A narrower deﬁnition of mate choice requires the existence of one or more traits in the choosing sex that lead to non-random mating (Heisler et al. 1987;
Maynard Smith 1987). This sets the focus on characteristics of the choosing sex, while still covering active forms of choice as well as resistance. Mate choice can be seen as the outcome of matings that arise from mating preferences, i.e. the traits that induce non-random mating (Jennions & Petrie 1997). Throughout this thesis, I have followed this deﬁnition and have focused on preferences as a characteristic of the choosing sex.
It is further useful to divide mating preferences 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; Widemo & Sæther 1999). Widemo & Sæther (1999) have additionally separated the sampling strategy (the ’how to choose’) from the ’how much to invest’ (choosiness) and ’what to choose’ (preference functions). Throughout my thesis, the distinction between preference functions and other aspects inﬂuencing mate choice is most important.
Costs and beneﬁts of mate choiceCosts of mate choice
Mate choice necessarily has some costs associated with it. Discrimination against some phenotypes involves the resistance to mating when harassed by members of the other sex (Andersson & Iwasa 1996; Kokko et al. 2003). More active forms of mate choice entail searching for mates, which involves investment in terms of time and energy (Alatalo et al. 1988; Vitousek et al. 2007; Booksmythe et al. 2008). Costs of mate choice can be low, if distances to travel between potential mates are short, traits are easy to evaluate and when time and energy constraints are low (Byers et al. 2005, 2006; Vitousek et al. 2007).
Under other circumstances costs might be large and this tends to decrease choosiness (Milinski & Bakker 1992; Backwell & Passmore 1996; Booksmythe et al. 2008). The costs associated with mate choice are mainly determined by the ecology of a species or population. It is difﬁcult to quantify the costs of mate choice, because it is often difﬁcult to unambiguously assign some behaviour to the mate choice context. This seems to be Between-individual differences in mating preferences 5 the main reason why we are largely lacking estimates of mate choice costs, although such estimates would be highly valuable (Jennions & Petrie 1997; Kokko et al. 2003).
Marginal costs might differ between individuals and this might lead to conditiondependent mate choice. This should mainly affect the choosiness and hence the range of accepted stimuli rather than the preference function itself. However, there are two reasons why it might also affect preference functions: First, if there is strong assortative mating for quality (either because of mutual mate choice or competition within the choosing sex), individuals might beneﬁt from adjusting their preference to stimuli that are achievable (Johnstone 1997). Such anticipated competition would induce conditiondependent variation in preference functions (Burley & Foster 2006). Second, if attractiveness is multidimensional because there are different beneﬁts to be optimised and there is condition-dependent variation in needs (e.g. direct versus indirect beneﬁts), this would also affect how individuals rank opposite-sex stimuli, because they might shift the importance of attractiveness axes. The effect can be strong if these axes are relatively independent, while a correlation between them would diminish this effect.
Beneﬁts of mate choice
For investment into mate choice to be an overall successful strategy, beneﬁts of choice have to outweigh the costs. These beneﬁts can be broadly classiﬁed into direct and indirect beneﬁts (Kirkpatrick & Ryan 1991; Kokko et al. 2003). Direct beneﬁts help reducing the investment into the current breeding event and hence to increase an individual’s prospects for future reproduction. The most obvious forms of direct beneﬁts are nuptial gifts and help in parental care (Badyaev & Hill 2002; Nakagawa et al. 2007). But also a high quality territory or protection from harassment by other members of the opposite sex can be considered direct beneﬁts.
Indirect beneﬁts are somewhat less obvious and harder to understand. Nevertheless, they have been an important research focus in the ﬁeld of sexual selection (Møller &
Alatalo 1999). Two main ideas have been put forward by Fisher (1930):
Good-gene indicator hypothesis: Choosing individuals should prefer potential mates that carry alleles for high viability, since this promises increased offspring viability and hence higher (long-term) ﬁtness. Empirically, these effects are small, but signiﬁcant (Møller & Alatalo 1999). Since viability depends on a large number of loci, goodgene indicators are expected to reﬂect genome-wide quality (Andersson & Iwasa 1996).
Some mechanism has to ensure honesty of the signal, since the invasion of ’cheating’ mutations (that produce a strong signal, but are otherwise not associated with genetic quality) will diminish the indicator value of a trait. Honesty is most easily ensure by condition-dependent trait expression (e.g. by indicating resistance to parasites, Hamilton & Zuk 1982) or by constituting a handicap to the carrier (Zahavi 1975).
General introduction6 Sexy-son hypothesis: The sexy-son hypothesis puts the focus more on good-genes for attractive sons and less on viability. The main conceptual difference between goodgenes for viability and good-genes for attractive sons, is that the trait does not have to signal genome-wide quality, since the indirect beneﬁts are realised via the mating advantage of the (male) offspring (Andersson & Iwasa 1996). Hence, if there is a populationwide preference for a particular trait, novel mutations that increase the strength of the signal will be selected for independent of their inﬂuence on viability. If preference and trait are genetically correlated (e.g. via linkage disequilibrium due to assortative mating), this can lead to runaway selection, where the preferences for and the expression of the trait co-evolve and are driven to extremes (Fisher 1930; Lande 1981). Since sexy-son traits signal quality by promising high ﬁtness via high mating success in male offspring, it has been suggested that the good-gene indicator and the sexy son-hypothesis are not as different as often suggested and might be combined to a ’Fisher-Zahavi hypothesis’ of indirect beneﬁts (Eshel et al. 2000; Kokko et al. 2002, 2003).
The good-gene indicator and the sexy-son hypothesis predict a unifying overall best solution to mate choice. The sexy-son hypothesis requires a heritability of preferences and if preferences are heritable, both hypotheses predict a genetic correlation between preferences for and expression of a trait.
Other causes of mating preferences
Preferences might have been selected for in contexts other than mating and hence be due to sensory bias (Ryan 1998; Fuller et al. 2005). For example, a preference for orange spots in guppies and sticklebacks seems to have been shaped by a preference for colourful food items (Rodd et al. 2002; Smith et al. 2004). Mating preference might also originate from apparently non-adaptive sensory biases (Burley et al. 1982; Ryan et al. 1990; Ryan & Rand 1990; Basolo 1995). Holland & Rice (1998) have proposed an antagonistic chaseaway model of sexual selection, in which members of the chosen sex exaggerate traits for purposes of sensory exploitation, while the choosing sex develops a resistance to these stimuli (since they may convey costs, but no beneﬁts) and this can lead to an acceleration of the process very much like in a Fisherian runaway scenario.
Between-individual differences in mating preferences Most of the ideas described above suggest that all individuals should aim at mating with the same superior individual(s). Empirical studies, however, have shown that this is often not the case (Bakker & Pomiankowski 1995; Jennions & Petrie 1997; Widemo & Sæther 1999; Brooks & Endler 2001; Brooks 2002; Forstmeier & Birkhead 2004). This is at ﬁrst glance a puzzling situation. However, there are potential adaptive explanations Between-individual differences in mating preferences 7 for between-individual differences in preferences (Figure 1). Furthermore, it is useful to study the proximate causes in order to understand the origins of this variation (Figure 2).
Adaptive explanations If there is non-additive genetic variation in ﬁtness-relevant traits, the interaction between the parental genomes matters for offspring ﬁtness (Zeh & Zeh 1996; Neff & Pitcher 2005). Special cases of genetic compatibility that are well-documented in natural systems are heterozygote advantage (Kempenaers 2007), inbreeding depression (Charlesworth & Charlesworth 1987; Pusey & Wolf 1996) and outbreeding depression (LeBas 2002; Price & Bouvier 2002; Peer & Taborskyi 2005). It is unclear, however, how important mate choice for compatible alleles is independent of an outbreedinginbreeding axis (i.e. different degrees of relatedness, Tregenza & Wedell 2000). Best evidence comes from MHC-disassortative preferences (Wedekind et al. 1995; Penn 2002) and from studies showing that extra-pair young are more dissimilar than within-pair young (Johnsen et al. 2000; Blomqvist et al. 2002; Foerster et al. 2003).
The most likely situation in which mate choice for compatibility might be relevant between unrelated individuals within single populations is when the recombination in chromosomal regions relevant for compatibility is suppressed (Tregenza & Wedell 2000). Additionally, the polymorphism has to be phenotypically expressed, so that it can be detected by the chooser. In mice, for example, females heterozygous for a t allele avoid smells of males also carrying a t allele (which is genetically incompatible), while wild-type homozygotes do not (Williams & Lenington 1993). Zeh & Zeh (1996) suggest that by mating multiply, females adopt a bet-hedging strategy against the ’cumulative toll of genetic incompatibility’. Chapters 1, 2 and 3 of this thesis address mating preferences in relation to genetic compatibilities.