Sexual selection is selection
I wrote this short essay during a book course on Evolutionary Quantitative Genetics by S. J. Arnold (2023). I also gave a short presentation under the same title.
Sexual selection arises from competition within a sex for access to mates and fertilisations (Andersson 1994). As an idea, sexual selection emerged with Darwin (1871) and Wallace (1889) as the proposed cause of elaborate animal traits that did not appear to increase survival or fecundity, and thus could not be explained by natural selection. Sexual selection research has since bloomed into a diverse field of study that is often found occupying the interface between evolutionary biology and many other fields (Jones et al. 2013). Such a cross-disciplinary nature has undoubtably benefited the field by providing multiple, mutually-beneficial lines of enquiry. However, this comes at the cost of an increased risk of disconnect between researchers supposedly studying the same phenomenon. Consequently, it is important to stress that sexual selection is selection, and as such is best studied in that context (Arnold and Wade 1984). The school of thought presented in Arnold (2023) provides a rigorous framework in which sexual selection thinking has been both formalised and clarified.
Much of modern sexual selection research has it’s roots in the work of the geneticist Angus John Bateman. In his classic paper, Bateman (1948) found that in experimental populations of Drosophila melanogaster (\(1\)) variance in reproductive success (number of offspring) and (\(2\)) variance in mating success (number of sexual partners) was greater in males than in females, and that (\(3\)) male reproductive success increased with mating success, while for females it generally did not. These three results have since become known as Bateman’s principles (Dewsbury 2005). Ultimately, Bateman proposed anisogamy to be the main driver of sex differences in the dependency of reproductive success on mating success, and suggested his results would likely apply to all sexually reproducing anisogamous organisms (Bateman 1948). He reasoned that males (with abundant small gametes) should be limited in reproduction by their mating success, and not on their investment in each gamete, and that females (with less numerous large gametes) should be limited by their ability to produce and invest in viable gametes, not on the probability that their gametes get fertilised. This line of reasoning would eventually expand into what is called the “Darwin-Bateman paradigm” (Dewsbury 2005).
Bateman’s principles have since been formalised and connected to selection theory, providing clarity and a common currency by which the potential for sexual selection can be measured and compared (Klug et al. 2010; Anthes et al. 2017). Intra-sexual variance in reproductive success (\(1\)) reflects an upper limit on the magnitude of selection, and is termed the opportunity for selection (\(I\)) (Crow 1958; Wade 1979; Arnold and Wade 1984). In other words, if \(I=0\), then selection cannot act, as there is no variation in fitness in the population. Likewise, intra-sexual variance in mating success (\(2\)) reflects an upper limit on sexual selection, termed the opportunity for sexual selection (\(I_s\)), which again in the case where \(I_s=0\), there can be no correlation between a trait and mating success, as there is no variation in mating success in the population (Wade 1979; Wade and Arnold 1980). These variance based metrics, while clearly connected to selection theory, are limited in the picture they can paint of how sexual selection can act. For example, in a species where fecund females attract more mates, but not benefit from additional matings, female \(I_s\) would be high, yet there would be no selection for increased mating success (Lorch 2005).
Arnold (1994) formalised Bateman’s insight that the relationship between mating success and reproductive was indicative of the strength of sexual selection (\(3\)), and did so by placing it in the individual selection surface (ISS) framework, with mating success as the focal trait, and fitness measured in reproductive success. The linear component of this surface (\(\beta_{ss}\)) is what has since become known as the Bateman gradient or the sexual selection gradient, which measures the strength of directional selection on mating success. A simple interpretation of this slope is the expected gain in number of offspring from each additional mating, although how reproductive success and mating success are measured may alter this interpretation. Likewise, the (often not estimated) quadratic component (\(\gamma_{ss}\)) of the ISS captures the strength of stabilising or disruptive selection on mating success. A priori, we might expect this term to take a negative value (stabilising selection), as in most species we do not expect the gains from increased matings to be linear indefinitely. The Bateman gradient represents the path to fitness for all sexually selected traits, such that for any trait correlated with mating success to be selected, an increase in mating success must also causally result in an increase in reproductive success (\(\beta_{ss}>0\)). However, if a single proxy for the potential for sexual selection is desired, then the metric proposed by Jones (2009), \(s'_{max}\), likely has better properties. Jones recognised that \(\beta_{ss}\) largely ignores variance in mating success, and proposed incorporating the opportunity for sexual selection (\(I_s\)), to give \(s'_{max}=\beta_{ss}\sqrt{I_{s}}\), which captures the upper limit of selection on a trait due to the trait’s effect on mating success. In a simulation study by Henshaw et al. (2016), \(s'_{max}\) outperformed \(I\), \(I_s\) and \(B_{ss}\) in capturing the true strength of sexual selection.
While the Bateman gradient can be discussed as a selection gradient, it has one notable difference: unlike the majority of phenotypic traits studied under this framework, “mating success” is not only the product of an individual, and is instead the outcome of an individual’s (relative) competitive ability, the ecological setting in which the competition takes place, and the social environment (Emlen and Oring 1977). As such, it’s response may be poorly predicted when used in the Breeder’s equation (Jones 2009; Jones et al. 2013).
By formalising the Darwin-Bateman paradigm into selection theory, sexual selection researchers could both place their work in a greater context, and draw on a rich body of thought. This clarified the thinking in the field, and provided a rigourous means by which sexual selection could be studied. The “common currency” of metrics from selection theory has allowed for comparisons in the potential for sexual selection between sexes, populations and species, and subsequently for testing the paradigm’s assertions across diverse taxa (Janicke et al. 2016). While I only cover the cause of sexual selection, selection theory and adaptive landscape thinking have also contributed greatly to our understanding of sexually selected traits, and will continue to do so for the foreseeable future (Jones et al. 2013).