Mitochondrial function and sexual selection: can physiology resolve the ‘lek paradox’?

Across many taxa, animals use sexually selected traits – such as elaborate ornaments and complex behavioural displays – to attract potential mates. Sexually selected traits have more often evolved in males than females as a result of the lower benefit of parental care for males (Fromhage and Jennions, 2016), and because anisogamy (see Glossary) creates an asymmetry in the benefits of increasing fertilization success when sperm outnumber eggs (Lehtonen et al., 2016). The question of why females prefer males with greater expression of elaborate sexual traits has been widely debated. In some species, females gain benefits from their mates in the form of material resources that increase their lifetime reproductive output (reviewed in Kokko et al., 2003). In these species, sexually selected traits are hypothesized to signal the benefits that females will receive from mates; thus, females benefit from choosing males with greater relative expression of sexually selected traits (Iwasa and Pomiankowski, 1999; Kirkpatrick, 1996). However, in many species, females do not appear to receive any material benefits from mate choice (Andersson, 1994; Moller, 1994). Therefore, a key challenge has been to explain the persistence of female mating preferences when there are no apparent material benefits to female choice (i.e. when males only transfer sperm to females).

In standard ‘Fisherian runaway’ scenarios, females that choose males with more attractive sexually selected traits produce sons that also have more attractive sexually selected traits as adults (i.e. ‘sexy sons’). Consequently, the sons of choosy females are more successful than average at acquiring mates. In this scenario, female mating preferences will be favoured through indirect selection if genes for a female mating preference co-occur with genes for preferred traits in males that are under direct selection (Kokko et al., 2006). In other words, non-random mating as a result of female choice increases the co-occurrence (linkage disequilibrium) of genes for mating preferences and genes for beneficial sexually selected traits (Kuijper et al., 2012). One challenge to explain the evolution of mating preferences is that female choice should generate strong directional selection on males that erodes additive genetic variation (see Glossary) in preferred sexually selected traits and, hence, reduces variation in trait expression (Fig. 1A–C). If males do not vary in the fitness benefits they provide to offspring, then females no longer gain a genetic benefit from their mating preferences and selection will eliminate female mating preferences if choosiness is costly (Bonduriansky and Day, 2013; Kokko et al., 2015; Kotiaho et al., 2008; Tobler et al., 2011; Vitousek et al., 2007). However, female mating preferences and heritable variation in sexually selected male traits persist in many species. This so-called ‘paradox of the lek’ is most obvious in lek mating systems (see Glossary) where, compared with other mating systems, female choice is strongest, male sexual traits are most elaborate and choosy females only appear to gain genetic benefits from males (Kirkpatrick and Ryan, 1991). The lek paradox has puzzled biologists since it was originally described by R. A. Fisher (Fisher, 1930; Kirkpatrick and Ryan, 1991; Petrie, 2021; Pomiankowski and Moller, 1995; Taylor and Williams, 1982), and it has inspired a plethora of hypotheses and experimental tests that seek to resolve it (e.g. Bonilla et al., 2016; Pomiankowski and Moller, 1995; Tomkins et al., 2004).

One challenge in distinguishing between competing hypotheses that try to resolve the lek paradox is a poor understanding of the physiological mechanisms that underlie the development and expression of sexually selected traits. Most hypotheses have focused on generic physiological traits that mediate sexual trait expression, such as body condition (see Glossary), immune function and hormone levels (e.g. Folstad and Karter, 1992; Gusdon et al., 2007; Hamilton and Zuk, 1982; McGraw et al., 2010; Morehouse, 2014; Tomkins et al., 2004). Broadly speaking, empirical studies have shown some associations between these physiological traits and sexual trait expression, but the direction of the associations is inconsistent, and the underlying mechanisms that regulate such relationships remain largely unresolved. Consequently, very few hypotheses have specified how the genetic basis of physiological mechanisms that affect sexually selected traits contributes to the maintenance of additive genetic variation for these traits (but see Hill, 2018, discussed below).

Here, we argue that understanding the proximate mechanisms that determine mitochondrial function could help explain the maintenance of additive genetic variation in sexually selected traits, despite the erosive effect of directional female mate choice. Mitochondria play a central role in the conversion of food into cellular energy, and variation in their efficiency affects the amount of energy that eukaryotic organisms can allocate to competing life processes (Box 1). Recent theoretical work has proposed that mitochondrial function affects the expression of sexually selected traits and could play a critical role in the evolution of mating preferences (Hill, 2014, 2018; Koch and Hill, 2018; Koch et al., 2017). We combine this theoretical work with an idea proposed by Miller and Moore (2007) that indirect genetic effects maintain variation in the expression of sexually selected traits through maternal effects (see Glossary).

Specifically, we propose that greater expression of sexually selected traits indicates superior mitochondrial function. Given that mitochondrial function is often influenced by the maternal environment (see below), a male with superior sexual trait expression provides information to females about his mother's quality (which will be inherited by a choosy female's daughters if there is additive genetic variation in maternal effects). In this scenario, phenotypic variation in male sexually selected traits arises from two non-mutually exclusive sources that are both under female control: (1) variation in maternal effects that influence mitochondrial function in sons; and (2) the combination of mitochondrial and nuclear genes that a female provides to her offspring through her own genetic contribution and by her choice of mate (i.e. the ‘mitonuclear compatibility model of sexual selection’; Hill and Johnson, 2013; see below). In both cases, choosy females can gain genetic benefits because being choosy improves mitochondrial function of their offspring in ways that increase: (a) the expression of sexually selected traits in their sons and (b) the fitness of their daughters, whose own offspring receive better maternal effects.

In this Commentary, we describe how indirect genetic effects for good ‘maternal’ genes can maintain genetic variation in condition-dependent, sexually selected male traits despite strong female preferences (Miller and Moore, 2007). We then discuss how mitochondrial function can link maternal effects to the expression of sexually selected traits and review empirical studies that link mitochondrial function to the expression of sexually selected ornaments (carotenoid coloration) and displays (bird song). We highlight how maternal effects and interactions between the mitochondrial and nuclear genomes drive variation in mitochondrial function. We then discuss the implications of these processes for mate choice and consider whether including measurements of mitochondrial function in models of sexual selection can help to resolve the lek paradox. Finally, we propose directions for future studies to test the strength of the proposed links between mitochondrial function, maternal effects, sexually selected traits and female mating preferences.

Sexually selected traits often exhibit more phenotypic variation than other traits under natural selection (Pomiankowski and Moller, 1995), even though directional sexual selection due to female mate choice should reduce additive genetic variation. All else being equal, a decline in phenotypic variation in sexual trait expression should make it harder for females to discriminate among males (Fig. 1A–C). Phenotypic and additive genetic variation in sexually selected traits could be maintained more easily if genes at many loci contribute to the expression of sexual traits. With more loci, there is a greater likelihood that mutations generate phenotypic variation and a lower likelihood that alleles reach fixation as a result of directional selection (Houle, 1992; Pomiankowski and Moller, 1995). The maintenance of additive genetic variation in this way is particularly likely for sexually selected traits whose expression is dependent on condition. Theory proposes that condition-dependent sexually selected traits are costly to produce and/or maintain (e.g. Jennions et al., 2001; but see Kotiaho, 2001). Consequently, individuals in good condition that pay smaller marginal costs can afford to invest more heavily in these traits compared with individuals in poor condition (Kuijper et al., 2012). An individual's condition is likely to depend on many loci (Houle, 1992; Tomkins et al., 2004). Therefore, additive genetic variation in sexually selected traits is more likely to be maintained, despite strong directional selection, if they are condition dependent, because a large proportion of the genome affects condition, creating a bigger ‘mutational target’. Consequently, mutations in genes that affect condition are ‘captured’ and additive genetic variation in sexually selected traits is maintained (the ‘genic capture’ hypothesis; Rowe and Houle, 1996; Tomkins et al., 2004; Fig. 1D–F).

In most models of mate choice evolution that invoke body condition, it is assumed that sexually selected traits signal additive genetic variation that directly elevates a male's condition and, thus, his ability to express and maintain elaborate ornaments or displays. Miller and Moore (2007) put a twist on this logic by pointing out that mothers can affect the phenotype of their offspring beyond their own genetic contribution through maternal effects (Mousseau and Fox, 1998). Their rationale is clever and, in their own words, ‘simple’: a mother can affect the environment her offspring experience during development and her ability to do so can be influenced by her genes. Specifically, Miller and Moore (2007) acknowledged that mothers often play a fundamental role in shaping the developmental environment, which can have strong effects on the body condition of their offspring (e.g. Cunningham and Russell, 2000; Love et al., 2005; but see Merila et al., 2001). Maternal effects can thus affect the expression of sexually selected traits in her sons through developmentally induced changes that influence their condition as adults. Crucially, the magnitude of maternal effects can vary, in part, as a result of genes that the mother bears (e.g. genes for better parental care or more egg yolk). Therefore, when a female chooses a male based on sexually selected traits, she obtains genes for daughters who are better mothers that provide a higher-quality developmental environment for their offspring and not just genes for sexy sons. A choosy female will therefore have fitter grandsons (i.e. increased sexual trait expression in grandsons) because of the superior maternal environment that her daughters provide.

A shared component of the genic capture hypothesis and the ideas proposed by Miller and Moore (2007) is that variation in condition among males generates variation in sexually selected traits. Historically, ‘body condition’ has been defined as the total pool of resources available to an individual to allocate to fitness-related traits (Miller and Moore, 2007; Rowe and Houle, 1996; Tomkins et al., 2004). A fundamental challenge to empirical studies that seek to link variation in body condition to variation in the expression of sexually selected traits is that body condition (as defined above) is difficult to quantify and often obligates the use of morphometric indices (reviewed in Jakob et al., 1996; Kraft et al., 2019; Peig and Green, 2010). Many common body condition indices are reflective of body fat (Kraft et al., 2019). However, they may not be biologically relevant depending on the natural history of the focal species, the environmental conditions individuals experience or the life history stage of individuals. As such, they do not always predict fitness (Barnett et al., 2015; Sánchez et al., 2018; Wilder et al., 2016).

Recent theoretical work that discusses condition in terms of not just energy availability but also energy utilization may offer a resolution (e.g. Hill, 2011). In eukaryotes, energy utilization depends on cellular respiration, including the process of oxidative phosphorylation (see Glossary) where mitochondria convert substrates derived from food into adenosine triphosphate (ATP; Box 1). Importantly, mitochondrial structure and function are determined by gene products that are encoded by both the mitochondrial and nuclear genomes (Poyton and McEwen, 1996). It has been proposed that mitochondrial function provides the most fundamental information about a potential mate's quality to females because of the role of mitochondria in the production of ATP and overall organismal function (Hill, 2011; Hill and Johnson, 2013). However, females cannot directly evaluate the mitochondrial function of their potential mates. Instead, sexually selected traits that are energetically expensive to produce and maintain are hypothesized to signal a potential mate's mitochondrial function (Hill, 2011; Hill and Johnson, 2013; Koch and Hill, 2018). Including energy utilization in the definition of condition provides a physiological mechanism (mitochondrial function) that links genes to condition and, ultimately, to the expression of sexually selected traits (Fig. 1G).

Empirical evidence supports a mechanistic link between mitochondrial function and the expression of sexually selected traits. For example, male house finches (Haemorhous mexicanus) with redder feathers are more attractive to females (Hill, 1990, 1991), and Hill et al. (2019b) found that plumage redness was positively correlated with the efficiency of liver mitochondria. Specifically, redder males had higher respiratory control ratios (see Glossary), increased mitochondrial membrane potential and less proton leak than duller-coloured birds (Hill et al., 2019b). Other studies have found that drug treatments that increase mitochondrial antioxidants (see Glossary) increase bill redness (Cantarero and Alonso-Alvarez, 2017; Cantarero et al., 2020), further supporting a mechanistic link between mitochondrial function and carotenoid-coloured ornaments. Mitochondrial function has also been invoked to explain the evolution of sexually selected behaviours, such as bird song (Koch and Hill, 2018). For example, male zebra finches (Taeniopygia castanotis) with lower mitochondrial respiration capacity and less efficient mitochondria have lower-frequency songs (a trait important for mate choice; Mikula et al., 2021) than males with greater mitochondrial performance (Crino et al., 2022). Mitochondrial function could affect behavioural displays through effects on neurogenesis that determine an individual's ability to learn and perform complex cognitive tasks, or by limiting the energy available to perform displays or to allocate to morphological structures used in displays (Crino et al., 2022; Koch and Hill, 2018).

Mitochondria vary in the rate and efficiency of ATP production and in the production of metabolic by-products such as reactive oxygen species (ROS). One source of variation in mitochondrial function arises from the conditions that individuals experience during development through maternal effects. In support of this claim, there is a rich biomedical literature from studies with mammals that link maternal diet and body condition during gestation to long-term metabolic consequences for offspring due to changes in mitochondrial function (reviewed in Armitage et al., 2005; Bruce and Hanson, 2010; Fernandez-Twinn and Ozanne, 2010). In addition to maternal nutrition, studies have identified other developmental conditions that induce long-term effects on mitochondrial function. For example, Stier et al. (2022) found that adult Japanese quail (Coturnix japonica) that experienced an elevated incubation temperature (38.4°C) had higher mitochondrial respiration rates than quail that were incubated at a cooler temperature (37.0°C). Similarly, studies in zebra finches found that exposure to elevated temperatures during development has sustained effects on mitochondrial function that persist into adulthood (Pacheco-Fuentes et al., 2023; Ton et al., 2021; Udino et al., 2021). Exposure to elevated temperatures during development could cause sustained changes in mitochondrial function through epigenetic modifications that affect the expression of mitochondrial uncoupling proteins (see Glossary) (Argyropoulos and Harper, 2002; Gao et al., 2021). Alternatively, higher temperatures could cause changes in mitochondrial function through elevation of glucocorticoid hormones. Glucocorticoids are steroid hormones that regulate mitochondrial function and play a role in the vertebrate ‘stress’ response (McEwen and Wingfield, 2003; Picard et al., 2018). Nestling birds exposed to elevated rather than control levels of glucocorticoids have less-efficient mitochondria as adults (Casagrande et al., 2020; Crino et al., 2022), and these effects are thought to be regulated through sustained changes in the neuroendocrine pathway that regulates glucocorticoid release, which, in turn, affects mitochondrial function into adulthood (Crino et al., 2022). Together, these studies demonstrate that conditions during development can have long-term effects on mitochondrial function.

Another potential source of variation in mitochondrial function arises from variable compatibility between nuclear and mitochondrial genes. Processes such as mitochondrial gene replication, transcription and translation, and the generation of ATP through oxidative phosphorylation depend on interactions between mitochondrial and nuclear genes (Burton and Barreto, 2012). For example, mitochondrial genes code for proteins that combine with proteins encoded by nuclear genes to form complexes that are integral to the process of oxidative phosphorylation. These mitonuclear complexes (see Glossary) must ‘fit’ together precisely for mitochondria to function efficiently. Incompatibilities between mitochondrial and nuclear genes reduce the fit of mitonuclear complexes and, consequently, the ability of electrons to move through the electron transport system (Hill, 2019; Lane, 2015). A poorly fitting complex reduces ATP production and increases the production of ROS (Brand and Nicholls, 2011). Empirical studies suggest that selection strongly favours compatible mitonuclear complexes (Healy and Burton, 2020; Pereira et al., 2021). A female contributes all her mitochondrial genome to her offspring but only half of her nuclear genome, and she can potentially increase offspring fitness by choosing a mate whose nuclear genes are more likely to be compatible with her mitochondrial genes (Hill, 2018; Hill and Johnson, 2013).

If condition-dependent sexual traits signal mitochondrial function, a key question that follows is how this information might benefit choosy females. The mitochondrial genome is transmitted from mothers to their offspring, so a father's mitochondrial genes will not affect his offspring. However, we propose that females can benefit from choosing a mate that demonstrates superior mitochondrial function in scenarios where mitochondrial function is influenced by maternal effects and/or indicative of compatible mitochondrial and nuclear genes (i.e. the ‘mitonuclear compatibility model of sexual selection’).

Miller and Moore (2007) proposed that when maternal effects have a heritable basis, a female will benefit from choosing males with high-quality sexually selected traits because her daughters are likely to inherit genes for superior maternal care. Briefly, they point out that sexual traits can signal the quality of the maternal environment that a male received. Factors that maintain additive genetic variation in maternal effects therefore also maintain variation in the genetic benefits signalled by sexual traits. We expand on this idea by suggesting that mitochondrial function is the key physiological mechanism that links maternal effects to the expression of sexual traits. Given that females are more likely than males to reproduce, regardless of their quality (Arnold, 1994; Janicke et al., 2016), there will be persistent variation in the sexual traits of sons due to variation in maternal effects. Even so, sons of low-quality mothers are unlikely to breed because of their low attractiveness, thereby eliminating some genes for poor maternal effects. In contrast, daughters of low-quality mothers will still tend to reproduce, so the consequences of genes for poor maternal effects are partially hidden from selection, thereby slowing erosion of additive genetic variation in sexual traits.

Mitochondrial function can also be influenced by mate choice that results in compatible mitonuclear complexes for offspring. The mitonuclear compatibility model of sexual selection shifts the focus of female choice away from ‘good genes’ (see Glossary) to nuclear genes that create compatible mitonuclear complexes (Hill, 2018; Hill and Johnson, 2013). The logic is reminiscent of non-additive models of sexual selection that propose that females seek out ‘genetically compatible’ mates, such that the combination of maternal and paternal genes increases offspring fitness (i.e. non-additive genetic effects; see Glossary; Aguirre et al., 2016; Zeh and Zeh, 1996, 1997). Here, the expression of sexually and naturally selected traits in offspring, and hence their fitness, is determined by interactions between paternal and maternal genes (e.g. increased heterozygosity can elevate fitness; Fromhage et al., 2009; Mays and Hill, 2004). Non-additive models of sexual selection propose that variation in sexual traits can be maintained by interactions between alleles at a locus (dominance) or through epistasis (see Glossary) when population size, mutation rates and allele frequencies are low (Neff and Pitcher, 2008). In these scenarios, non-additive gene actions can explain persistent genetic variation in sexually selected traits (Neff and Pitcher, 2008, 2009; Puurtinen et al., 2009). The mitonuclear compatibility model of sexual selection proposes that genetic variation in sexually selected traits can be maintained because new combinations of mitonuclear genes arise each generation, providing an extensive source of variation despite directional selection (Hill, 2018). This hypothesis proposes a novel explanation for how mitochondrial and nuclear genes interact at an individual level to generate genetic variation in sexually selected traits. Future studies that use mathematical models could test what factors determine the extent to which mitonuclear compatibility maintains genetic variation in sexually selected traits.

Female choice for compatible mitonuclear complexes is only useful if males with better mitochondrial function transfer nuclear genes that increase the likelihood of siring offspring with above-average mitochondrial function. A male may signal that his mitonuclear complexes are highly compatible through the expression of sexually selected traits, but that does not guarantee that his nuclear genes will be compatible with a choosy female's mitochondrial and nuclear genes, and result in offspring with highly compatible complexes. However, there are far fewer mitochondrial than nuclear genes, which suggests that most of the variation in mitochondrial function is due to nuclear rather than mitochondrial genes (Hill, 2019). As such, a female who chooses a male that demonstrates highly compatible mitonuclear complexes is more likely to receive nuclear genes for her offspring that produce compatible complexes than is a female who mates randomly. One way to envisage this is to take the extreme example of a species with only two mitochondrial haplotypes. In this scenario: (a) a female is more likely to transmit the more common haplotype to her offspring; and (b) a male is more likely to have this haplotype. As such, a male whose ornaments reveal that he has superior mitochondrial function is signalling that he has nuclear genes that are likely to be compatible with the mitochondrial haplotype that the female will transmit to her offspring. By extension of this argument, as mitochondrial genetic variation increases, females are less likely to receive nuclear genes from a mate that are compatible with her mitochondrial genes.

An alternative possibility is that females increase the chance that some of their offspring have highly compatible mitonuclear complexes through an adaptive bet-hedging strategy (see Glossary). For example, if a female mates with multiple males, some males will provide nuclear genes that are more compatible with her mitochondrial genes than others, even if all the males have high-quality sexual traits. In this scenario, some of her offspring will perform well and some will not. Conversely, if a female mates singly, then either all or none of her offspring will receive nuclear genes that produce highly compatible mitonuclear complexes and all or none of her offspring will perform well. In this way, mating multiply increases variation in mitochondrial function within a female's offspring and may decrease her fitness compared with mating singly. However, selection will favour such a bet hedging strategy compared with an invariant strategy if it lowers variance in fitness among females and increases long-term fitness across generations (Philippi and Seger, 1989; Seger and Brockmann, 1987; Starrfelt and Kokko, 2012). In general, models that test the adaptive value of bet-hedging strategies, such as multiple mating, find that the parental fitness gains from lower variation in offspring fitness are small, present under restricted circumstances (e.g. small populations, high environmental variation) and can easily be outweighed by even small costs imposed by bet hedging (e.g. sexual diseases in the case of multiple matings; Holman, 2016; Starrfelt and Kokko, 2012; Yasui, 2001). However, the potential benefits of a bet hedging strategy in relation to mitonuclear compatibility have yet to be formally modelled.

The core of the lek paradox is that choosy females obtain genes that enhance mean offspring fitness through a combination of naturally and sexually selected benefits for sons and/or daughters by mating with males that invest more in sexually selected traits (Kokko et al., 2006). This benefit of choosiness requires that there is genetic variation in fitness even though choosiness should deplete variation. Here, we have proposed that additive genetic variation in sexually selected traits can be maintained through indirect selection on maternal effects that influence the expression of sexually selected traits through changes in mitochondrial function. We have expanded on proposals that variation in mitochondrial function underlies variation in the expression of sexually selected traits (Hill, 2011, 2014, 2018, 2022; Hill and Johnson, 2013; Koch and Hill, 2018; Koch et al., 2017). Variation in mitochondrial function among individuals can arise both from the quality of the environment experienced during development (i.e. maternal effects) and from mitonuclear gene interactions. Both sources of variation are strongly influenced by the maternal genotype. In species where most females breed, some females provide a poor-quality developmental environment and/or transfer mitochondrial and nuclear genes that increase the likelihood of incompatible mitonuclear complexes. Males produced by such mothers will have reduced mitochondrial function and, consequently, lower expression of sexually selected traits.

The pathway we propose provides a partial resolution to the lek paradox that captures the salient points of other widely accepted hypotheses and suggests avenues for future research. First, mitochondrial function determines an individual's ability to convert food resources to cellular energy and is therefore intrinsically linked to condition-dependent sexually selected traits which underpin the ‘genic capture’ hypothesis (Houle, 1992). Second, we are building on ideas proposed by Miller and Moore (2007) by identifying a physiological mechanism (mitochondrial function) that links maternal genes (i.e. genes for good maternal effects) to the expression of sexually selected traits in sons. Finally, these ideas are compatible with hypotheses that invoke genetic compatibility to explain the maintenance of female preferences (i.e. non-additive genetic benefits; Aguirre et al., 2016; Zeh and Zeh, 1996, 1997), by noting that females should choose mates with nuclear genes that increase mitonuclear compatibility (and, hence, mitochondrial function) in their offspring (Hill, 2018). Overall, our ideas highlight the need to look more closely at the sources of variation in mitochondrial function in model study systems for sexual selection.

There has been a recent boom in evolutionary and ecological studies of mitochondrial bioenergetics (Koch et al., 2021), and there is compelling evidence that links mitochondrial function to the expression of sexually selected traits (e.g. Crino et al., 2022; Hill et al., 2019b). Even so, this area of research is in its infancy. Future studies need to test the extent to which mitochondrial function affects a wider range of sexually selected ornaments and displays. Additionally, we need to disentangle the different physiological pathways that influence mitochondrial function. Studies that experimentally alter mitochondrial function using drugs such as 2,4-dinitrophenol (DNP), mitoquinone mesylate (MitoQ; e.g. Cantarero and Alonso-Alvarez, 2017; Stier et al., 2014) or mitochondria-targeted antioxidants (Jiang et al., 2020; Zinovkin et al., 2023) could be particularly informative. New methodologies now allow for non-terminal sampling of mitochondrial function from small animals (Stier et al., 2017). Such advances allow for longitudinal studies in many model systems and open promising avenues of research. Finally, formal mathematical models are needed to validate verbal arguments that sexually selected traits in males signal additive genetic variation in nuclear genes that affect mitochondrial function, be this through the nuclear and mitochondrial genes they express or because of maternally inherited genes that affect the expression of the bearer's sexual traits. Identifying the benefits females gain from choosing males with specific traits is crucial to understanding the process of sexual selection. Thus, we will improve our understanding of fundamental principles in sexual selection by considering the role that females play in driving the expression of sexually selected traits in sons through effects on mitochondrial function.

We thank three anonymous reviewers for suggestions that improved the quality of this manuscript. O.L.C. thanks H. Jane Brockman and the participants in her Animal Behaviour reading group at the University of Florida from 2005 to 2008, especially Matt D. Smith.