Animal Mating Preference

Abstract

Animal mating preference is a type of animal sexual behavior: the differential propensity to mate with individuals bearing particular traits. It is the behavioral substrate of intersexual selection—the choice half of Darwin's theory—and the engine that drives ornaments and displays. This article separates the preference itself from the choosiness that acts on it and the choice that results, a distinction that governs how preferences are measured. It sets out the four accounts of why a preference evolves—direct benefits, Fisherian runaway, good genes, and pre-existing sensory bias—and develops the preference function, the mapping from a male trait to its mating success, as the quantity those accounts predict. A worked example computes the selection differential a preference imposes, and three demonstrations develop the Fisherian coevolution of preference and trait, a preference function's shape, and the assortative mating that similarity preferences produce.

Keywords: animal mating preference, mate choice, sexual selection

A female frog sits in a chorus of calling males and moves toward one of them. What she is doing looks like a single act, but it hides three separable things. There is the ordering she would impose on the males if she could evaluate them all at once—her preference. There is how much effort, risk, and time she is willing to spend to act on that ordering—her choosiness. And there is the male she actually ends up mating with—her choice, the only one of the three an observer sees directly. A preference can exist and leave no trace in the choices if choosiness is zero; a choice can be made with no underlying preference if the males were sampled at random. Animal mating preference is the first of these—the internal ordering—and the central problem of the field is to infer it from the choices it only partly determines.

Mating preference matters because it is the behavioral cause of one of Darwin's two mechanisms of sexual selection. Where competition among males favours weapons and size, the preferences of females favour ornaments and displays—the peacock's train, the frog's call, the bowerbird's bower. These traits exist because preferences exist, so explaining the trait reduces to explaining the preference: why does a female order males this way, and why has selection built that ordering into her? The answers divide into whether the preference pays off in the female's own reproduction, in her offspring's genes, or in neither.

What Animal Mating Preference Is

Animal mating preference is the differential propensity to mate with individuals bearing particular traits. In the Medical Subject Headings vocabulary it is filed as a kind of animal sexual behavior, beside copulation and the pair bond, and it is the behavioral substrate of what Darwin called the second mechanism of sexual selection: intersexual selection, or mate choice (Darwin, 1871). Darwin's insight was that a trait can spread not because it helps its bearer survive but because the other sex prefers it, and that a preference, acting consistently over generations, is a selective force as real as any predator. The preference is the cause; the ornament is the effect.

A preference is defined at the level of the individual chooser and need not be shared, fixed, or adaptive. Reviewing the evidence, Michael Jennions and Marion Petrie documented how widely mating preferences vary—between species, between individuals of the same species, and within an individual across its condition, age, and context—so that a population's mating pattern is an aggregate of heterogeneous and shifting preferences rather than the expression of one species-typical rule (Jennions & Petrie, 1997). That variation is not noise to be averaged away: it is the raw material selection acts on, and much of the modern study of preference is the study of why it varies. The sections below take up, in turn, how a preference is distinguished from the behavior that expresses it, why preferences evolve, how a preference is summarized as a function, and what mating patterns preferences produce.

Preference, Choosiness, and Choice

The single most important conceptual move in the field is to separate three things that a casual description runs together. Preference is the chooser's internal ordering of potential mates—which she would take if evaluation and acceptance were free. Choosiness is the effort she invests in acting on that ordering: how many males she samples, how long she searches, how much risk and energy she spends, how far she will pass over an acceptable male in hope of a better one. Choice is the observable outcome—the male she actually mates with. Dustin Edward set out this three-part decomposition and showed why it is indispensable: the same choice data are consistent with many different preferences depending on the choosiness that generated them, so preference and choosiness cannot be read off the choices without a model that separates them (Edward, 2014). A female that mates with a middling male may have a weak preference, or a strong preference and no choosiness, or a strong preference frustrated by a poor sample.

Because preference is latent, everything turns on how the choices are elicited, and the experimental design is not a neutral window onto the preference. In a meta-analysis of mate-choice experiments, Liam Dougherty and David Shuker found that whether animals are tested with mates presented one at a time (a no-choice design) or several at once (a simultaneous design) systematically changes the preference the experiment appears to reveal, because the two designs engage choosiness differently (Dougherty & Shuker, 2014). The lesson is methodological but it cuts to the concept: a measured preference is always a preference-as-expressed-through-a-design, and comparing results across designs without accounting for this conflates the ordering with the machinery that acted on it (Figure 1).

Why Mating Preferences Evolve

Before asking why a preference takes a particular form, it helps to ask why one sex evolves to express preferences at all. The usual answer is the asymmetry in reproductive investment: because A. J. Bateman's Drosophila experiments found male reproductive success rising steeply with mate number while female success saturated, the sex that gains less from extra matings—typically the female—is the one for which being choosy pays (Bateman, 1948). Robert Trivers generalized this into the theory of parental investment: the sex investing more per offspring becomes the limiting resource and can afford to choose, while the sex investing less competes for access (Trivers, 1972). The foundation is not beyond dispute—a careful repetition of Bateman's study by Patricia Gowaty and colleagues found his original data could not reliably support the conclusion once methodological biases were corrected, so the asymmetry is best read as a prediction rather than a law (Gowaty et al., 2012). Granting that a choosy sex exists, the question becomes why its preference is built the way it is.

A preference imposes costs—sampling takes time and exposes the chooser to predators—so a consistent preference must be built by selection, and there are four accounts of how. They are not rivals so much as mechanisms that can each operate, often together, and the live empirical question for any species is which dominates (Table 1). The first is the direct-benefit account: the preferred trait signals something the chooser gains in her own reproduction—a nuptial gift, a good territory, parental care, or freedom from parasites—so the preference pays off immediately and needs no appeal to genetics. Direct benefits are the least contentious route to a preference and, where they exist, the strongest, because selection on them is not diluted through the offspring (Kokko et al., 2003).

The remaining three are indirect: the preferred trait carries no direct benefit, and the payoff, if any, is in the genes of the offspring. In Fisher's runaway process the preference and the preferred trait become genetically correlated, because a female who prefers the trait bears sons who carry it and daughters who carry the preference; the two then reinforce each other and can escalate together until the trait's survival cost checks it (Fisher, 1930). Russell Lande made this quantitative, showing that for polygenic traits the coupled dynamics produce not a single equilibrium but a line of equilibria, along which a population can drift and from which it can run away (Lande, 1981). Mark Kirkpatrick's explicit genetic model made the logic unmistakable: a preference allele, carrying no viability benefit of its own, can hitchhike to high frequency purely through its statistical association with the trait it favours—a preference can evolve for a trait that does nothing but attract (Kirkpatrick, 1982).

The third account is good genes, or the handicap principle: a costly trait is a reliable signal of heritable quality precisely because only a high-quality male can bear the cost and survive, so a female preferring it endows her offspring with good genes (Zahavi, 1975). The good-genes logic is harder to demonstrate than it looks. Amanda Achorn and Gil Rosenthal argued that much of the evidence for it mismeasures offspring quality and quietly equates a son's attractiveness with his viability, so the case for good genes should be made with far more care than the classic verbal model implies (Achorn & Rosenthal, 2020). The difficulty is partly theoretical: James Fry re-examined the formal arguments and confirmed that indirect selection on a preference—whether Fisherian or good-genes—is generally expected to be weak relative to even slight direct selection, so a small direct cost or benefit can overwhelm the indirect force the classic models emphasize (Fry, 2022).

The fourth account dispenses with benefit altogether. In the sensory bias (or sensory-exploitation) model, the preference exists first, as a by-product of how the chooser's perceptual system is already tuned—for finding food, say, or detecting predators—and a male trait evolves later to exploit that pre-existing bias. Alexandra Basolo's swordtail experiment is the canonical demonstration: females of a swordless species preferred males given an artificial sword, showing the preference predated the trait and so could not have coevolved with it (Basolo, 1990). Michael Ryan generalized this into a synthesis of receiver biases, arguing that the properties of sensory and cognitive systems, evolved for other reasons, set the stage on which sexual signals evolve, so that many preferences are better explained by the receiver's biology than by any benefit the choice confers (Ryan, 1998). Reviewing the mechanisms together, Malte Andersson and Leigh Simmons stressed that these accounts are not mutually exclusive and that disentangling them is the central empirical programme of mate-choice research (Andersson & Simmons, 2006). The first demonstration develops the Fisherian case, the hardest to see without a model: sliders set the genetic variances and the cost of the trait, and the display traces whether the preference and trait settle onto the line of equilibria or run away together.

In the Fisherian runaway model a female preference evolves only by riding along with the male trait it favours. The mean trait z and mean preference y co-evolve; the dashed line is the balance between sexual and natural selection. Set the genetic covariance between preference and trait and watch the population either settle onto that line or spiral away from it.

Table 1. The four accounts of why a mating preference evolves. The first pays off in the chooser's own reproduction; the next two in the genes of her offspring; the last in nothing at all. They are not mutually exclusive, and the empirical task is to measure their relative contributions in a given species (Andersson & Simmons, 2006).

The Preference Function

To make a preference quantitative, it is summarized as a preference function: a mapping from a male trait value to the relative mating success it confers, holding choosiness and sampling fixed. The function's shape carries the biology. An open-ended (directional) function rises monotonically—more of the trait is always better—and drives the trait ever upward until survival costs intervene. An absolute (stabilizing) function peaks at some preferred value and falls away on either side, so there is a most-preferred trait rather than an unlimited one. A relative function peaks at a value set by the chooser's own trait, producing the similarity preferences that generate assortative mating. The width of the peak is the chooser's tolerance, the inverse of how sharply she discriminates, and it is a property of the preference distinct from where the peak sits.

The preference function is not fixed within a species or even within an individual. Nancy Burley and colleagues showed experimentally that when males differ on several traits at once, the preference is not a simple sum of single-trait functions: traits combine synergistically, and the decision rule itself shifts with context, so the function measured on one trait in isolation can misrepresent the preference that operates when all the traits are present (Burley et al., 2018). Preferences also track the environment in which the choice is made. Megan Head and colleagues found that an environmental change altered mate choice for an extended-phenotype signal—a structure the male builds—while leaving choice for the male's intrinsic quality intact, showing that a single preference function can be environment-dependent for one component of a signal and robust for another (Head et al., 2016). This conditional, multivariate character is why a preference is better treated as a decision process than as a parameter: Gil Rosenthal's synthesis reframes mate choice as a chain from the signals a chooser can perceive, through the perceptual and cognitive machinery that weighs them, to the ecological cost of searching, so that the preference function is the output of a decision embedded in a sensory and ecological context rather than a fixed property of the trait (Rosenthal, 2017). The second demonstration draws the three function shapes over a trait axis and reports the selection differential each imposes on a normally distributed male population.

A preference function maps a male trait onto his relative mating success; multiplied by the trait’s distribution it fixes the selection differential S — how far choice shifts the mated males’ mean trait. The male trait here is Gaussian with mean 10 and standard deviation 2. Choose a preference shape and read off S.

Assortative Mating

When the preference is relative—when choosers prefer mates similar to themselves on some trait—the population-level consequence is positive assortative mating: mated pairs are more alike than random pairing would make them. In a meta-analysis across animals, Yuexin Jiang, Daniel Bolnick, and Mark Kirkpatrick found positive assortative mating to be common and, on average, moderately strong, making it one of the most widespread of all mating patterns, while negative assortment (mating with dissimilar partners) is comparatively rare (Jiang et al., 2013). The pattern is not only a curiosity of who pairs with whom. Assortative mating on a heritable trait builds a positive correlation between the genes of mates, which inflates the genetic variance of the trait in the next generation and can, when the assortment is strong and the trait is tied to ecological divergence, contribute to the reproductive isolation that begins speciation—the link Lande's models drew between mate preference and the splitting of lineages (Lande, 1981).

The strength of the resulting correlation follows directly from the width of the similarity preference. A chooser who insists on a near-identical partner (a narrow tolerance) produces strong assortment; a chooser who accepts a wide range produces weak assortment, and in the limit of infinite tolerance the pairing is random. The third demonstration makes this precise: a tolerance slider sets how sharply choosers discriminate on a single trait, and the display shows the mated pairs as a scatter of male against female trait values together with the correlation the preference induces.

When choice favours a similar partner, the population mates assortatively: mated pairs resemble each other, and the regression slope of one partner’s trait on the other’s measures the strength of assortment. Narrow the similarity window and watch the cloud of pairs tighten toward the diagonal.

Worked Example: A Preference Function

Consider a population of males whose trait value z—say, the length of an ornament—is normally distributed with mean μ = 10 and standard deviation σ = 2, so the variance is σ² = 4. A preference function ψ(z) assigns each male a relative mating weight, and the mean trait among the males who actually mate is the population mean weighted by ψ. The selection differential the preference imposes is the difference between that mated mean and the population mean, S = z̄_mated − μ. This is the quantity that translates a preference into evolutionary change.

Take first an open-ended preference of exponential form, ψ(z) = exp(βz), with β = 0.25: every extra unit of ornament multiplies a male's mating success by a constant factor, so more is always better. For a normal trait distribution this weighting shifts the mean by exactly βσ², giving a mated mean of 10 + 0.25 × 4 = 11.0 and a selection differential of S = βσ² = 1.0. The preference drags the trait distribution upward by one unit in a single episode of selection; nothing in the function itself halts the climb, which is why open-ended preferences are the ones that run away.

Now take an absolute preference of Gaussian form peaked at a preferred value θ = 14 with tolerance ν = 2 (so ν² = 4): males are judged against a target, and a male far from θ in either direction is penalized. Weighting a normal population by a Gaussian preference yields a mated mean that is the precision-weighted average of the population mean and the preferred value, z̄_mated = (μ/σ² + θ/ν²) / (1/σ² + 1/ν²) = (10/4 + 14/4) / (1/4 + 1/4) = 6.0 / 0.5 = 12.0, for a selection differential of S = 2.0. The preferred value pulls the mated mean toward θ, but only partway, because the supply of males thins out toward the preferred extreme—the preference wants a 14 but the population offers mostly 10s, and the realized mean settles in between. A Monte Carlo simulation drawing four million males and weighting them by each preference function reproduces these values (11.0 and 12.0) to three significant figures, confirming the closed forms the PreferenceFunctionDemo uses (Edward, 2014).

Discussion

The study of mating preference has been organized by two clarifications that are easy to state and hard to honour. The first is conceptual: a preference is not a choice. The internal ordering, the effort spent acting on it, and the mate obtained are three different things, and because only the last is observed, every claim about a preference is an inference through a model of choosiness and a particular experimental design (Edward, 2014; Dougherty & Shuker, 2014). The second is causal: a preference need not be good for anything. It may pay off directly, pay off only through the genes of offspring, or—on the sensory-bias account—pay off not at all, being a by-product of a perceptual system built for another task (Basolo, 1990; Ryan, 1998). Treating a preferred trait as automatic evidence of mate quality is the error both clarifications warn against.

The balance among the mechanisms has shifted as the theory has been scrutinized. The indirect-benefit accounts that once dominated—Fisherian runaway and good genes—now look weaker than the direct-benefit route, both because the formal arguments show indirect selection to be easily overwhelmed by slight direct selection (Fry, 2022) and because the evidence for good genes rests on measurements that often conflate attractiveness with viability (Achorn & Rosenthal, 2020). None of this retires the indirect models; it relocates them from the default explanation to hypotheses that must clear a higher bar. The most durable reframing treats a preference not as a parameter but as the output of a decision process, multivariate and conditional on environment and context, which is where the field's measurement effort now concentrates (Burley et al., 2018; Rosenthal, 2017). The population-level consequences of preference—above all the pervasive positive assortative mating that preferences for similarity produce—then connect individual choice to the structure of genetic variation and to the origins of species (Jiang et al., 2013; Lande, 1981).

Current Directions

The live programme on the mechanism side is the measurement of preference as a multivariate, context-dependent function rather than a single coefficient. Experiments that vary several male traits at once show that preferences combine traits synergistically and that decision rules shift with the set of options present, so the function estimated from one trait in isolation can mislead (Burley et al., 2018). Work on environmental dependence pushes in the same direction, dissociating the components of a signal a preference tracks robustly from those it tracks only under particular conditions (Head et al., 2016). The organizing idea is to treat the chooser as an information processor whose preference emerges from perception, cognition, and the ecological cost of sampling (Rosenthal, 2017).

On the theoretical side, the reassessment of indirect selection is unfinished. Re-derivations of the classic models continue to find indirect selection on preference fragile relative to direct selection, which raises the question of how the elaborate preferences seen in nature are maintained if their indirect payoff is so easily swamped (Fry, 2022). Reconciling the apparent strength of real preferences with the apparent weakness of the indirect forces once invoked to explain them—whether through overlooked direct benefits, sensory biases, or the coupling of preference to other selected traits—is the open question the current theory is working toward.

Common Misconceptions

A preference is the same as a choice.

It is not. A preference is the chooser's internal ordering of mates; a choice is the mate actually obtained, which also depends on choosiness and on which mates were sampled. The same choices are consistent with many different preferences, so preference must be inferred, not read off (Edward, 2014).

A preferred trait must signal mate quality.

No. On the Fisherian account the trait signals nothing but its own attractiveness, and on the sensory-bias account the preference is a by-product of a perceptual system tuned for another task, with no quality information at all (Kirkpatrick, 1982; Basolo, 1990).

Good-genes benefits are well established.

They are harder to demonstrate than the verbal model suggests. Much of the evidence mismeasures offspring quality and equates a son's attractiveness with his viability, and the formal arguments show indirect selection to be weak relative to direct selection (Achorn & Rosenthal, 2020; Fry, 2022).

A species has one fixed mating preference.

Preferences vary between individuals and within an individual across condition, context, and environment, and they combine multiple traits non-additively. A single species-typical preference function is an idealization, not the usual reality (Jennions & Petrie, 1997; Burley et al., 2018).

Glossary

Animal sexual behavior.

The set of behaviors directed at reproduction—courtship, competition, mate choice, copulation, and pair bonding—under which MeSH files mating preference as a subtype.

Assortative mating.

A departure from random pairing in which mates resemble each other (positive) or differ (negative) on a trait; positive assortative mating, produced by similarity preferences, is the common case.

Choosiness.

The effort a chooser invests in acting on a preference—how many mates are sampled, how long the search lasts, how much risk is accepted; distinct from the preference itself.

Direct benefits.

Advantages a preference confers on the chooser's own survival or fecundity—a nuptial gift, a good territory, parental care, or a mate free of parasites—rather than on the genes of offspring.

Fisherian runaway.

A process in which a preference and the preferred trait become genetically correlated, so each reinforces the other and the two escalate together until checked by the trait's survival cost.

Good genes.

The hypothesis that a costly trait reliably signals heritable viability, so a preference for it endows offspring with good genes; also the handicap principle.

Intersexual selection.

Sexual selection arising from the choice of mates by one sex, usually females; the force that favours ornaments and displays, with mating preference as its behavioral substrate.

Mate choice.

The observed pattern of mating that results when a preference is acted on with some degree of choosiness; the behavioral output from which the latent preference is inferred.

Mating preference.

The differential propensity to mate with individuals bearing particular traits; the chooser's internal ordering of potential mates, latent and inferred from choices.

Preference function.

The mapping from a male trait value to the relative mating success it confers; its shape—open-ended, absolute, or relative—determines the selection a preference imposes.

Selection differential.

The difference between the mean trait value among mated individuals and the population mean; the quantity that translates a preference function into evolutionary change in the trait.

Sensory bias.

A preference that exists as a by-product of a perceptual system tuned for another task, such as finding food; on this account the male trait evolves later to exploit the pre-existing bias.

Sexual selection.

The differential reproductive success arising from competition for mates and from being chosen; Darwin's second selective process, comprising intrasexual competition and intersexual choice.

Tolerance.

The width of the peak of an absolute or relative preference function; the inverse of how sharply a chooser discriminates among trait values, and the parameter that sets the strength of assortative mating.

Key Researchers

A. J. Bateman

(John Innes Horticultural Institution). Geneticist whose 1948 Drosophila experiments quantified the sex difference in the variance of reproductive success, giving the anisogamy-based rationale for why females are typically the choosier sex—the empirical root of mating-preference asymmetry. Wikipedia

Charles Darwin

(independent naturalist, Down House). Naturalist who introduced sexual selection in The Descent of Man (1871) and first argued that female choice—a mating preference—could be a selective force shaping male ornament, the founding idea of the field. Wikipedia

Ronald A. Fisher

(University College London; Rothamsted Experimental Station). Statistician and geneticist whose Genetical Theory (1930) proposed the runaway process by which a preference and the preferred trait become genetically coupled and coevolve—the central theoretical mechanism of preference evolution. Wikipedia

Mark Kirkpatrick

(University of Texas at Austin). Evolutionary geneticist whose 1982 model gave the first explicit genetic demonstration that a preference allele can hitchhike to high frequency with the trait it favours, formalizing Fisherian runaway, and whose later work quantified assortative mating across animals. ORCID

Hanna Kokko

(University of Mainz). Evolutionary theoretician whose models unified the direct and indirect benefits of mate choice, showing how choosiness, the operational sex ratio, and population feedbacks jointly shape the evolution of mating preferences. ORCID

Gil G. Rosenthal

(University of Padua and Texas A&M University). Evolutionary biologist whose synthesis Mate Choice (2017) reframed mating preference as an information-processing problem spanning perception, cognition, and environment, integrating mechanism with evolutionary function. ORCID

Michael J. Ryan

(University of Texas at Austin). Evolutionary biologist whose work on túngara frogs and the sensory-exploitation hypothesis showed that mating preferences can arise as by-products of pre-existing perceptual biases rather than as adaptations for mate quality. ORCID

Frequently Asked Questions

What is animal mating preference?

It is the differential propensity to mate with individuals bearing particular traits—a chooser's internal ordering of potential mates. It is the behavioral substrate of intersexual selection, the mate-choice half of Darwin's theory of sexual selection (Darwin, 1871).

How is a preference different from a choice?

A preference is the internal ordering of mates; a choice is the mate actually obtained. The choice also depends on choosiness—the effort spent acting on the preference—and on which mates were sampled, so the same choices can reflect very different preferences (Edward, 2014).

Why do mating preferences evolve?

Through some combination of four routes: direct benefits to the chooser, Fisherian runaway, good-genes signalling, and pre-existing sensory bias. They are not mutually exclusive, and which dominates is an empirical question for each species (Andersson & Simmons, 2006; Kokko et al., 2003).

Does a preferred trait always indicate a good mate?

No. On the Fisherian account the trait signals only its own attractiveness, and on the sensory-bias account the preference is a by-product of perception with no quality information at all. A preferred trait is not automatic evidence of mate quality (Kirkpatrick, 1982; Basolo, 1990).

What is a preference function?

It is the mapping from a male trait value to the relative mating success it confers. Its shape—rising without limit (open-ended), peaked at a preferred value (absolute), or peaked at the chooser's own value (relative)—determines the selection the preference imposes (Edward, 2014).

What is assortative mating?

It is a departure from random pairing in which mates resemble each other on a trait. Positive assortative mating, produced by preferences for similar partners, is one of the most common mating patterns in animals and inflates the genetic variance of the trait (Jiang et al., 2013).

Are the good-genes benefits of mate choice well established?

Less firmly than once thought. Much of the evidence mismeasures offspring quality, and the formal theory shows indirect selection on a preference to be weak relative to direct selection, so good-genes claims require careful measurement (Achorn & Rosenthal, 2020; Fry, 2022).

Is a species' mating preference fixed?

No. Preferences vary between individuals and within an individual across condition and context, combine several traits non-additively, and can depend on the environment in which the choice is made (Jennions & Petrie, 1997; Head et al., 2016).

References