Grey zones of sexual selection: why is finding a modern definition so hard?

Abstract

Sexual selection has long been acknowledged as an important evolutionary force, capable of shaping phenotypes ranging from fascinating and unusual displays to cryptic traits whose function is only uncovered by careful study. Yet, despite decades of research, reaching a consensus definition of the term ‘sexual selection’ has proved difficult. Here we explore why arriving at a unifying definition of sexual selection is so hard. While some researchers have argued about whether sexual selection should be considered a form of natural selection, we concentrate on where the line between sexual selection and other forms of selection falls. We focus on identifying the ‘grey zones’ of sexual selection by illustrating cases in which application of the term ‘sexual selection’ would be considered controversial or ambiguous. We believe that clarifying why sexual selection is so difficult to define is an essential first step forward towards greater clarity, and if possible towards reaching a consensus definition. We suggest that a more nuanced perspective may be necessary, particularly one that specifies for cases of ‘sexual selection’ why the term is used or whether they fall into a grey zone.

1. Introduction

Sexual selection was proposed by Charles Darwin in part to explain the existence of traits that do not increase an individual's survival or fecundity [1,2]. These traits posed a puzzle because they were not easily explained by Darwin's initial formulation of the theory of evolution by natural selection. Darwin hypothesized that these otherwise puzzling traits might evolve because they increase an individual's success competing for or attracting mates, which he called sexual selection [1,2]. Perhaps the most classic example of a sexually selected trait is the striking train of male peacocks, which is clearly an impediment to escaping predators or finding food. Darwin argued that such traits could be favoured, even if they decrease survival, if males with larger and more colourful trains also experience increased attractiveness to females, and therefore also greater mating success (i.e. number of mates or mating rate).

While initially controversial, the existence and importance of sexual selection are supported empirically and theoretically [3–6]. It is the capacity to explain otherwise counterintuitive traits that makes the concept of sexual selection so powerful, not only within evolutionary biology but for our general understanding of biological diversity.

Darwin described sexual selection multiple times in his writing ([1,2], reviewed in [7]), rather than giving a single, diagnostic definition (see [8] for a discussion of the distinction between and importance of descriptions and definitions in science). At that time, it was not known how genes were transmitted or what mechanisms determined differential fertilization success. Thanks to extensive research, we now know that organisms exhibit far more complex and diverse traits related to mating and fertilization than could have been imagined at Darwin's time. Though these discoveries created a need for a more modern definition of sexual selection, finding a consensus definition has been challenging and controversial (e.g. [7,9–13]).

Various researchers have proposed definitions for sexual selection that are consistent with Darwin's original idea while addressing what we now know about the biology of fertilization and the complexities of mate choice and attraction (table 1). Existing definitions could be interpreted as differing in what ‘counts’ as sexual selection. We argue that an even more critical issue is that all existing definitions leave grey zones because they are ambiguous about what does and does not count as sexual selection. Thus, even if researchers agree on a definition, they may still come to a different conclusion about whether a specific trait is under sexual selection. This not only leads to confusion, but also makes synthesis challenging (e.g. [7]).

Table 1.

Illustrative definitions of sexual selection.

(1) Darwin (1859) [1] chapter 4, page 88: sexual selection ‘depends, not on a struggle for existence, but on a struggle between the males for possession of the females; the result is not death to the unsuccessful competitor, but few or no offspring’

(2) Darwin (1871) [2] part 2 chapter 2, page 256: sexual selection ‘depends on the advantage which certain individuals have over other individuals of the same sex and species, in exclusive relation to reproduction’

(3) Arnold and Wade (1984) [14] page 720: ‘selection arising from variance in mating success… We define mating success as the number of mates that bear progeny given survival of the mating organism to sexual maturity’

(4) Charlesworth et al. (1987) [15] pages 318: ‘competition between individuals for mates caused by traits which yield unequal fitness gains through male and female function, where the selective value of a trait is not due to its effect on survivorship of the sporophyte itself’

(5) Ryan and Rand (1993) [16] page 648: ‘variance in reproductive success that derives from variation in the ability to acquire mates’

(6) Andersson (1994) [3] page 7: ‘differences in reproductive success, caused by competition over mates, and related to the expression of the trait’ and page 31: ‘differences in reproduction that arise from variation among individuals in traits that affect success in competition over mates and fertilizations’

(7) Kokko et al. (2006) [9] page 44: ‘selection generated by differential access to opposite-sex gametes (or mates)’

(8) Clutton-Brock (2007) [17] page 1885: ‘Sexual selection is now commonly defined as a process operating through intrasexual competition for mates or mating opportunities, with the result that selection pressures arising from intrasexual competition between females to conceive or rear young are generally excluded and sexual selection is, by definition, a process that is largely confined to males. An unfortunate consequence of this is that characteristics that increase the competitive ability of individuals are likely to be attributed to sexual selection if they occur in males—but not if they occur in females. As a result, it may be helpful to return to a broader definition of sexual selection as a process operating through intrasexual competition for reproductive opportunities, providing a conceptual framework that is capable of incorporating the processes leading to the evolution of secondary sexual characters in both sexes’

(9) Shuker (2010) [12] page e12: ‘selection of traits associated with competition for mates. As such we can say that sexual selection arises from competition for mates [3]. To avoid confusion, I define ‘mate’ as a reproductive partner with which one or more zygotes are formed (thereby allowing post-copulatory processes)’

(10) Rosenthal (2017) [18] page 503: ‘differential reproductive success due to the ability to secure matings and/or fertilizations’

It is not our goal to propose a definition. Instead, we focus on identifying why it has been hard to find a consensus definition for sexual selection and where the differences in understanding and interpretation arise. We first consider why (and whether) having a clear, consensus definition of sexual selection is important. Next, we identify areas where existing definitions agree and mention some concepts they omit. Finally, we identify ‘grey zones’ of sexual selection, which we define as areas where there is confusion about whether the observed pattern represents sexual selection. We believe greater clarity regarding where these grey zones lie is an essential first step towards a consensus definition of sexual selection.

2. Is defining sexual selection important?

The question above has an obvious answer; as scientists we strive for clarity so ideally all researchers would use the same, clearly diagnostic definition for important concepts. There are, however, objectives for which a consensus definition matters more or matters less. For example, when simply measuring the net fitness of a trait, it may not be important to distinguish precisely between sexual selection, viability selection (arising from variance in survival), or fecundity selection (arising from variance in offspring number independent of any interaction with the opposite sex). Similarly, if one wishes to partition fitness between reproduction and survival it may not matter whether reproductive fitness is attributable to sexual selection or fecundity selection. Both forms of selection can explain the existence of traits that initially seem counterintuitive because they do not maximize survival.

On the other hand, the degree of nuance in sexual selection studies has grown vastly over the past few decades as researchers have attempted to understand details of the sexual selection process. Answering these more subtle questions requires distinguishing the workings of sexual selection more specifically than just obtaining a measure of reproductive fitness.

Current definitions of sexual selection all involve the fitness of one sex depending on the opposite sex (see examples in table 1). This raises a way in which sexual selection may differ from other types of selection. Because sexual selection involves interactions between different individuals in ways that depend upon their phenotypes or genotypes it is necessarily frequency-dependent—that is it will vary in strength, or even direction, depending on the frequency of other genotypes in the population. This is not generally true of fecundity selection, which is usually a property only of the focal individual's phenotype or genotype (though it can be affected by social interactions or population composition, e.g. via competition for resources, [17,19]). This frequency-dependence can affect the evolution of mating preferences and traits (e.g. [20,21]), highlighting one way in which the evolutionary expectations from sexual selection may be different from those of fecundity selection.

In addition to invoking sexual selection to explain the existence of elaborate traits, researchers attribute many other evolutionary outcomes and phenomena to the process (e.g. [22]). Sexual selection has been argued to facilitate adaptation and prevent extinction (by purging deleterious mutations, e.g. [23–25]), or, in other situations, to increase extinction risk (by decreasing population mean fitness, [26], but see [27]). Successful colonization of a new habitat by niche expansion may also be facilitated by sexual selection through condition dependence, whereby females prefer males that are in the highest condition [28]. Finally, much work has been done on whether sexual selection facilitates (e.g. [29,30]), or retards (e.g. [31], see [32]) the processes of speciation and diversification. We would like to be able to draw generalizations about the role that sexual selection plays in such processes. This task is made even more challenging without a consensus and diagnostic definition of what we mean by ‘sexual selection’.

3. Existing definitions of sexual selection: commonalities and omissions

Here we briefly review some existing definitions of sexual selection (table 1). We do not aim to give an exhaustive review or argue for one definition over others. Instead, we summarize various ways sexual selection has been defined to identify where there is clarity versus ambiguity. For example, all definitions of sexual selection focus on variation in fitness arising from some aspect of reproduction. They therefore all clearly distinguish sexual selection from natural selection arising owing to variation in adult survival (e.g. definitions 1 and 4). General agreement also exists that selection owing to variation in mate number or mating frequency is clearly sexual selection (e.g. definitions 3, 5 and 6). Some definitions implicitly assume that selection is occurring (e.g. definitions 1 and 7) or that there is variance in mating success (definitions 3 and 5) but do not explicitly mention genes or traits. Some also explicitly include selection arising from differential fertilization success (e.g. definitions 6 and 10). Similarly, a few definitions specify that sexual selection results from competition among individuals of the same sex for access to the opposite sex and/or their gametes (e.g. definitions 7 and 9). The potential for sexual selection on females is rarely explicit, and classic sex roles are often assumed in the way that sexual selection is defined (but see definition 8).

Importantly, current definitions also omit discussion of some key points. For example, few definitions clarify which if any forms of post-mating selection would be considered sexual selection; post-mating and post-fertilization processes that involve cryptic female choice and differential allocation in response to a female's mate are not addressed. Current definitions also omit or fail to specify how sexual selection can be differentiated from other forms of natural selection arising during reproduction (e.g. see definition 8). For example, when does variation in gamete production drive sexual selection and when does it cause a form of natural selection? Finally, existing definitions do not state explicitly whether or not variation in mate and/or offspring quality constitutes sexual selection.

These inconsistencies between and omissions in existing definitions have caused confusion about where to draw the line between sexual selection and other forms of selection, both conceptually and empirically. Yet, as argued above, there are many reasons other than semantics to define sexual selection clearly. It is thus important to identify and address any ‘grey zones of sexual selection’ explicitly. Without this clarity, it is challenging if not impossible to answer big questions about sexual selection, such as when and why it arises, how important it is as a driver of adaptation, and its role in generating and maintaining biodiversity.

In the next sections, we identify and explore some important grey zones of sexual selection. Specifically, we identify illustrative cases in which there could be arguments both for and against considering them examples of sexual selection. Some of these cases ‘feel’ like sexual selection but breaking down the sources of selection makes this label questionable. In other cases, the example seems to fall under ‘sexual selection’ in some scenarios and under a different form of selection such as fecundity selection in others. By exploring each case, we highlight common themes that make a strict definition of sexual selection challenging. One issue that arises repeatedly is where to draw the line between fecundity selection and sexual selection, or whether they are sometimes inextricably linked. Another issue concerns parental care; there are situations where care is a target of mate choice, or is induced by mate choice, in ways that cause confusion as to whether the term sexual selection should be applied.

Below we divide these cases into ones in which a trait may be under sexual selection and ones where sexual selection may instead be thought of as acting on a mating preference. In the examples below we use sex-specific language for choosers and targets of mate choice. While we agree with other authors (e.g. [33,34]) that there are many good reasons to use sex-neutral language in the discussion of sexual selection, our attempts towards this goal significantly reduced the clarity of our arguments.

4. Grey zones of sexual selection on male and female traits

(a). Case 1. The trait affects the acquisition of mating partners with greater current reproductive effort (e.g. mating partners differ in parental effort or fecundity)

Our first case is a grey zone principally because it is unclear where, or even whether, the line between fecundity selection and sexual selection should be drawn. Imagine that males, for example, do not differ in mating or fertilization success (e.g. monogamy with equal mating success), but males with a particular trait mate disproportionally more often with females with high fecundity (e.g. [35–37]). Access to high quality mates and mate identity therefore affect reproductive success (by which we mean the number of surviving offspring).

While this may initially sound like a straightforward example of sexual selection, this process requires a pre-existing correlation such that calling it sexual selection is not always clear. Specifically, the association of a trait with the acquisition of a high fecundity partner could occur in one of two ways. First, females with a mating preference (for males with the trait) could have high fecundity—i.e. there would have to be an existing association between female preference, or the exercise thereof, and female fecundity. This seems plausible if only high condition females are able to exercise choice. Alternatively, males with the trait could have a preference for females with high fecundity. Sexual selection, if we call it such, would then act on the trait in males, even though males, not females, express a mating preference. (Note that if males with the trait were more likely to win these females through male–male competition such males might also have higher mating success, which would be a separate avenue for sexual selection.)

Furthermore, the preference alone would not be enough to generate sexual selection, as this case assumes equal mating success. For the trait to be favoured, fecundity selection would also have to be occurring, such that males with the trait have more offspring that survive owing to greater female investment. Development of a mathematical model of this scenario would have no differential reproductive success due to the male trait at the ‘mating’ step of the model; such a benefit would only arise if fecundity selection were added as a separate step (figure 1). It is therefore unclear whether to attribute the benefit of the male trait to sexual selection, fecundity selection, or both.

Figure 1.

Life cycles with the possible placement of sexual selection circled. (a) ‘Standard’ scenario of sexual selection owing to differential mating success with regard to a trait. Sexual selection occurs during the non-random mating step of the life cycle because not every individual of the chosen sex gets to mate. (b) Case 1, in which there is no differential mating success during non-random mating, but individuals of the chosen sex procure more fecund mates. In this case it is unclear where in the life cycle sexual selection would be said to occur (if it occurs at all). No selection occurs during the non-random mating step, so sexual selection does not occur in that step alone, as in panel (a). Note that this assumption (non-random mating with equal mating success) can arise in ‘grouping’ models of sympatric speciation, where it is generally assumed to mean there is no sexual selection, regardless of what occurs later in the life cycle (e.g. [38]). Fecundity selection without non-random mating would also not constitute sexual selection because it would not affect the reproductive success of the chooser. Sexual selection could, however, potentially be said to be applied by the combination of non-random mating and fecundity selection because males, for example, could have differential access to eggs even in the absence of direct male–male competition for the more fecund females. (Online version in colour.)

A useful exercise is to flip the sex roles and see if our arguments hold. If we were instead considering sexual selection on a female trait, a parallel discussion would apply. Females could gain a fecundity benefit if high fecundity males preferred females with a certain trait, even if each female had equal mating success. This necessitates a correlation between preference and fecundity in males. Alternatively, females that have the trait could prefer to mate with high fecundity males, although it is unclear how such a correlation could develop.

This case highlights that sexual selection and fecundity selection can be inextricably linked. The description of this case in the sub-header above, when read quickly, certainly ‘feels’ like sexual selection, but if fecundity selection is not occurring, neither is sexual selection, which seems peculiar.

(b). Case 2. The trait affects the acquisition of mating partners with better genes

This case will not initially sound like a grey zone for many readers because it bears a superficial resemblance to the familiar scenario of ‘good genes’—a canonical example of sexual selection. In a good genes scenario, there is differential mating success based on whether males possess an indicator of high condition; this differential success places sexual selection on the indicator trait. In our Case 2, however, males, for example, possess a trait that causes them to attract females with genes for high condition (i.e. the genes for condition are in the sex with the preference, not the sex with the trait, e.g. [37]).

This case constitutes a grey zone for many of the same reasons as Case 1. This is most apparent when the trait does not affect mating success, but only affects mate identity. Referring to figure 1b, the process that we would potentially consider ‘sexual selection’ would still include ‘non-random mating without differential mating success’, but there would now be no fecundity selection. Instead there must be viability selection in the next generation (the next turn of the life cycle), whereby offspring of the individuals with the trait would have higher fitness. ‘Sexual selection’ would thus constitute two different processes occurring across generations.

Additionally, parallel to Case 1, there would have to be a pre-existing correlation between the preference and condition in order for individuals with the trait to acquire ‘good genes’ for their offspring. This is feasible if only individuals with high condition have the resources to exercise a preference, though the ability to express a preference would be expected to then be plastic.

(c). Case 3. The trait affects the attraction of reproductive (mating or non-mating) partners that vary in their investment in offspring care

This example is a subset of Case 1, and suffers from the same issue of the entanglement of fecundity selection and sexual selection (as well as requiring the correlations discussed in Case 1). Here, however, parental care plays a prominent role. Parental care in particular can be problematic as a source of high fecundity because it can be given by a partner that is not the biological parent of the offspring. It is unclear how to characterize this case because a continuum exists in degrees of potential parentage. In birds, for example, a provider can sire all offspring in the nest, to half, to no offspring in the nest. Thus, if we call this case sexual selection a logical extension would be to conclude that sexual selection on a trait in either sex can be applied by a partner that is not even a mating partner, which is quite an unusual idea. To extend this further, sexual selection could then be applied by cooperative breeding partners of the same sex, or by offspring from a previous mating that are helpers-at-the-nest. This certainly does not ‘feel’ like sexual selection anymore, but it is unclear where a strict line can or should be drawn.

(d). Case 4. The trait causes reproductive partners to invest more resources in offspring

This case is also closely related to Case 1 and like Case 3 has a shared focus on fecundity selection and parental effort. However, here the extra fecundity benefit is not an inherent property of the female, but is induced in the female (potentially in any female regardless of their genotype) by the male with the trait. Therefore, the fecundity benefit, while owing to the energy input of the female, could be interpreted as a property of the male, because only certain males can extract it from females [39]. This complicates an interpretation of sexual selection because the male trait simultaneously both applies to the selection (the fecundity selection, which is part of sexual selection in this case, see figure 1), and is selected upon. This case applies equally well to female traits that allow the female to extract parental effort from a male partner.

One extreme example of this case would be selective abortion of zygotes in both plants [40,41] and animals [42], if it is a property of the male that causes the female to invest in the offspring rather than abort it. Selective abortion is often thought of as cryptic female choice occurring after fertilization (after [43]), but sexual selection can be thought to act on a male trait if it reduces the chances of abortion.

(e). Case 5. A female trait causes males to invest more in their ejaculate

This case of induced investment by a partner is related to Case 4, in the situation where the trait is expressed in females and induces increased investment by males. Consider a female trait that causes males to produce, release or transfer more sperm. Similarly, males might invest more in other components of their ejaculate, increasing for example the peptides or other compounds they produce or transfer in response to variation in a trait of the female. Assuming the female trait is favoured by selection, is this a form of sexual selection on females? We argue this is in a grey zone because the interpretation depends on whether female fitness is limited by competition for sperm (to fertilize existing eggs) or access to resources (to produce more or higher quality eggs).

For example, if female reproductive success increases because males transfer a larger ejaculate with more resources for egg production, as in the case of ‘nuptial gifts’, this would cause fecundity selection on females in a fashion similar to Case 4, and in turn Case 1 (see the discussion there about whether fecundity selection plus non-random mating can constitute sexual selection). It is therefore ambiguous whether this should be considered sexual selection. On the other hand, if female fitness is limited by fertilization owing to sperm limitation, a female trait that increases female access to sperm (by causing males to produce, release or transfer more sperm) could be considered a sexually selected trait. In this latter instance, females compete for sperm and are limited by fertilization success rather than egg production. It is worth noting that if females with these traits are both more able to produce eggs (because the male supplies resources that increase egg production) and to fertilize their eggs (because they receive more sperm), then the female trait will be favoured by both fecundity selection and sexual selection.

(f). Case 6. A trait affects competition for a reproductive territory or other resource (such as food). Territories/resources are either required for reproduction or variation in territory/resource quality affects mating success

This case can either be interpreted as sexual selection or fecundity selection. Consider the following contrasting examples.

• 1.In some cases, such as when males compete for food that acts as a nuptial gift, it seems natural to state that sexual selection can act on a trait that controls this food acquisition. A more ambiguous case is one where males vary in their ability to obtain food, which limits the energy available for competition for reproductive territories. Males with more energy are either more able to obtain a territory or obtain a higher quality territory, thereby having higher mating success. This could be called sexual selection on traits related to food acquisition, although this is up for debate ([11], see [44]). Note that the same issues arises if the food was necessary to bring the male into reproductive condition in the first place. These scenarios could also apply to females under sex role reversal.
• 2.Females vary in their ability to obtain food, which affects their ability to make more eggs and/or rear more surviving offspring. Females are limited by their access to food, but not by their access to mates, therefore we call this fecundity selection. The same is true of males who need a territory of sufficiently high quality to raise their offspring. For example, if males that vary in their ability to obtain food are more able to provide paternal care, this is fecundity selection.

(g). Case 7. The trait affects whether an individual is chosen earlier or more quickly as a mate

This case addresses fecundity selection when its source is environmentally dependent. Specifically, consider the case where pairs that form early have a fecundity benefit. Kirkpatrick et al. [35] included this situation under the heading of ‘sexual selection’ in a quantitative-genetic model of monogamy in which females that breed early both have greater reproductive success and tend to mate with more preferred males. There are several reasons, however, why this may fall in a ‘grey zone’ with regard to whether it constitutes sexual selection.

First, we follow Kirkpatrick et al. [35] by considering a male trait. In general, greater reproductive success of males with the preferred trait would occur not because these males obtain more mates, but because their mates may have higher fecundity; this case thus shares the reasons that Case 1 is classified in a ‘grey zone’. Additionally, a trait that causes a male to breed early with no associated fecundity advantage would not constitute sexual selection, if males otherwise have equal mating success.

Furthermore, this case requires that females with a preference mate earlier than females without a preference. By contrast, discussions of female preferences and search costs assume that having a preference can cause females to skip mating opportunities and thus mate later than females with no preference (see Case 10), so there is an interesting mismatch between these two ideas.

For this scenario to exert sexual selection on a female trait, females would have to be mate-limited such that their ability to mate early was determined by whether a male was willing to mate with them, depending on whether they expressed the focal trait. If they were themselves able to control whether they mated early (again depending on whether they had the trait) then this would constitute fecundity selection on the trait rather than sexual selection. The distinction between fecundity selection and sexual selection is once again critical, in this case.

(h). Case 8. A trait aids in the production of a greater number of gametes

This case details why the roles of the sexes may be sometimes be critical when trying to distinguish between fecundity selection and sexual selection. When the gametes of multiple males compete to fertilize the eggs of a single female, males that produce and release more sperm often fertilize more eggs. This is typically considered a form of post-mating (e.g. post-copulatory) sexual selection arising owing to differential fertilization success, favouring male traits associated with increased sperm production [4,45]. It is important, however, to clarify why this might be considered sexual selection on male traits associated with greater gamete production, given that selection on similar female traits associated with increased egg production are typically considered to be fecundity selection. In both cases, the traits are favoured because they enhance gamete production and therefore either male or female fitness. The main distinction we see is that competition among male gametes leads to traits being favoured by sexual selection because male gametes typically compete for limited female gametes. By contrast, females that produce more eggs are not typically limited by access to sperm to fertilize these eggs. It is worth noting however that if individuals produce more gametes because they are better at foraging or obtaining resources required for gamete production, this falls into a grey zone of sexual selection (see Case 6).

5. Grey zones of sexual selection on preference

While there can be significant ambiguities regarding whether sexual selection can be said to act on a trait or mating cue, it is perhaps even less clear whether we can say that sexual selection is acting on a mating preference. While mating preferences are certainly involved in sexual selection, can we say that they evolve because of sexual selection, or do they simply cause sexual selection on the preferred trait? While we think that sexual selection can be said to be acting on a preference in certain cases, we list these as grey zones because they are not uncontroversial.

(a). Case 9. ‘Indirect’ sexual selection on a preference—a preference changes owing to its association with a trait under sexual selection

As demonstrated in mathematical models of sexual selection, even an evolutionarily neutral preference will change in frequency, but not because individuals expressing it have differential mating success. Specifically, preferences will evolve because they form statistical associations (genetic covariance or linkage disequilibrium) with evolving traits; preferences can evolve solely owing to this ‘indirect’ selection [20,21]. The question is, can we say, semantically, that the preference is evolving ‘owing to sexual selection’ in this case?

We think that we can, but only if we qualify this statement carefully. Consider, for example, a male trait that is selectively neutral except that it is preferred by females; the male trait would only evolve owing to sexual selection (the same would hold if the sex roles were reversed and a female trait were evolving owing to the presence of a male preference). It would thus be this sexual selection that, as the only selective force in the system, would ultimately be responsible for evolution of the preference via the statistical association that would automatically form between the preference and trait owing to non-random mating. One could then say that the preference was evolving owing to ‘indirect sexual selection’. We think that it would be misleading to say that the preference is evolving simply owing to ‘sexual selection’, however, because this implies that the preference is altering the mating success or fertilization success of the females that carry it, which is not the case in this example. We also note that the concept of ‘indirect sexual selection’ may be the best terminology to use when other traits evolve via a correlation with a trait under sexual selection, although distinguishing between direct and indirect sexual selection is an empirical challenge (see, e.g. [46]). These semantics are, however, open to debate, and we thus consider the concept of indirect sexual selection on preferences to be in the ‘grey zone’.

(b). Case 10. A preference affects mating success owing to the availability of individuals with the preferred trait (search costs)

There is one potentially important and common case where preferences could be said to be under sexual selection, though we have not seen this term commonly applied in this situation. When a preference affects a female's ability to settle on a mate (search costs), it technically causes differential female mating success, and would therefore fall under many definitions of sexual selection. This interpretation is in a grey zone, however, because it is possible that choosy females do not find a mate simply because searching increases the risk of predation. In this case the preference is selected against because it has a viability cost, which in turn leads to differential female mating success in that some (dead) females will not mate at all. Reversing the sexes, it is also possible that choosy males may pay search costs that lead to differential mating success by the same mechanisms.

To complicate this further, a mathematical model of search costs could use identical equations, occurring before the ‘mating’ step of the life cycle, that could be interpreted either as viability selection against the preference or as females with that preference not successfully choosing a mate and therefore being removed from the mating pool (e.g. [47]). These equations could even be frequency-dependent such that females with a preference pay a cost (of either interpretation) weighted by the frequencies or densities of preferred versus unpreferred males in the environment.

(c). Case 11. A female trait affects the ability of females to reduce or avoid mating

Sexual conflict over mating and fertilization has been widely documented and shapes the evolution of both male and female reproductive traits (e.g. [48]). While there is significant overlap between sexual selection and sexual conflict [49], it is worth exploring when selection on traits owing to conflict over mating and fertilization will also represent a form of sexual selection. When males and females experience divergent selection on mating rate (i.e. because mating is beneficial to male fitness but detrimental to female fitness), male persistence could be considered to be sexually selected, however, when mating reduces female fitness through decreased survival or fecundity, it seems natural to categorize the form of selection that favours traits that enhance a female's ability to reduce her mating rate or avoid mating altogether as natural selection. Female traits that reduce mating rate or increase control over mating and fertilization (because they favour offspring quality or control over paternity owing to good genes or direct benefits of mate choice) might be categorized as either sexual selection, viability selection or fecundity selection depending on the specifics of how mating affects female fitness. Finally, if selection disfavours female traits that reduce female mating rate (because they have the potential to prevent a female from successfully mating altogether or mating enough to fertilize all of her eggs), then these traits would clearly be under sexual selection, in a way parallel to search cost in Case 10.

6. Conclusion

Extensive research has documented the existence of sexual selection, demonstrated its power to drive trait evolution, and explored many ways in which sexual selection affects other biological patterns and processes. Despite its fundamental importance, a generally accepted modern definition of sexual selection does not exist, and as we reviewed in detail above, existing definitions all leave some grey zones. We argue here that identifying and discussing these grey zones is an essential next step towards ultimately reaching consensus; we must first clearly identify situations in which there is agreement and the zones in which there is either confusion or disagreement about whether sexual selection is operating.

The grey zones we identified share some common features. Many cases that fall into a grey zone involve selection arising from variation in offspring quality or quantity which depends on mate identity. These situations involve a grey zone between fecundity selection and sexual selection, or perhaps even the presence of the latter depending on the former. Another set of grey zones exist when mates vary in parental investment because of a trait in the opposite sex. This grey zone again involves fecundity selection, sexual selection or a combination of the two. Finally, selection on mate preferences can constitute a grey zone of sexual selection when selection on preferences and/or the preferred trait involve some combination of sexual, survival and fecundity selection.

How do we resolve the problem of these many grey zones? As a first step, we recommend being more specific about the ‘flavour’ of sexual selection in specific uses of the term. For example, we should not just say that some trait of interest is under ‘sexual selection’. Instead, we should specify that it is under, for example, ‘sexual selection based on variation in mating rate’ or ‘sexual selection based on attracting mates with higher quality genes’. While this may seem overly specific, it is similar to the need to use language that identifies not only that sexual conflict exists but also that specifies whether the sexual conflict exists over mating, fertilization, survival or parental effort [50]. Greater clarity of what we mean when we call something ‘sexual selection’ may or may not ultimately lead to a consensus definition, but it is a necessary first step if a consensus is ever to be reached. If we wish to resolve big questions about how sexual selection influences broader topics such as speciation and rates of evolutionary change, we need to know what we do and do not mean when we say sexual selection.

Data accessibility

This article has no additional data.

Authors' contributions

Both authors contributed equally to the ideas and arguments contained in the review as well as to the writing of the final paper.

Competing interests

We declare we have no competing interests.

Funding

We received no funding for this study.