Evolutionary lability of sexual selection and its implications for speciation and macroevolution

Sexual selection is widely hypothesized to facilitate speciation and phenotypic evolution, but evidence from comparative studies has been mixed. Many previous studies have relied on proxy variables to quantify the intensity of sexual selection, raising the possibility that inconclusive results may reflect, in part, the imperfect measurement of this evolutionary process. Here, we test the relationship between phylogenetic speciation rates and indices of the opportunity for sexual selection drawn from populations of 82 vertebrate taxa. These indices provide a much more direct assessment of sexual selection intensity than proxy traits and allow straightforward comparisons among distantly related clades. We find no correlation between the opportunity for sexual selection and speciation rate, and this result is consistent across many complementary analyses. In addition, widely used proxy variables – sexual dimorphism and dichromatism – are not correlated with indices employed here. Moreover, we find that the opportunity for sexual selection has low phylogenetic signal and that intraspecific variability in selection indices for many species approaches the range of variation observed across all vertebrates as a whole. Our results potentially reconcile a major paradox in speciation biology at the interface between microevolution and macroevolution: sexual selection can be important for speciation, yet the evolutionary lability of the process over deeper timescales restricts its impact on broad-scale patterns of biodiversity.

Methods
Meta-analytic procedure used for data collection

Estimates of the opportunity for sexual selection and Bateman gradients were obtained after an extensive search in the literature that follows meta-analytic procedures (see more details in Macedo-Rego et al 2024. 

While screening the literature, studies were selected whether they provided at least one estimate of φIs or φIf and one estimate of φI for a given sex (see main text). The search was performed in two databases: ISI Web of Science (all databases) and Scopus, and its step-by-step PRISMA design (see Nakagawa et al, 2017) is illustrated in Supplementary figure 2. The following keywords were used:"reproductive success" AND "mating success" OR "fitness" AND "mating success" OR "paternity" AND "mating success" OR "offspring" AND "mating success" OR "litter" AND "mating success" OR "fertilization success" AND "mating success" OR "breeding success" AND "mating success" OR "fecundity" AND "mating success" OR "reproductive rate" AND "mating success" OR "post-mating sexual selection" OR "post-mating selection" OR "Bateman*" OR "opportunit* for selection" OR "opportunit* for sexual selection" OR "selection gradient*" OR "Morisita index" OR "monopolization index for reproductive success" OR "Jones index" OR "copulation success" OR "opportunit* for natural selection" OR "intensit* of sexual selection" OR "mating success" AND "survival rate" OR "reproductive success" AND "number of mat*" OR "mixed paternity" OR "mating and reproductive success" OR "opportunit* for natural selection and sexual" OR "natural and sexual selection" OR "sexual and natural selection".

Combined, Scopus and Web of Science provided 7,624 studies. After reading titles and abstracts of each study, 1,659 were selected for further evaluation. Rejections were due to various reasons: studies not on sexual selection (54.89%), studies that focus only on pre-mating events of sexual selection (15.22%), studies on sexual selection but with no measurement of mating and reproductive success (13.23%), study on non-animal species (7.82%), studies that focus only on post-mating events of sexual selection (4.79%), studies on humans (3.08%), and studies whose complete versions were not found (0.97%).

Before/while reading texts in full, 185 additional studies were obtained from other sources (e.g., while checking reference lists). Consequently, 1,844 studies were read in full. At this step, studies were rejected due to various reasons: study on sexual selection but that does not provide at least one estimate of φIs or φIf and one estimate of φI (females: 41.88%; males: 38.88%), studies that focus only on pre-mating events of sexual selection (females: 18.41%; males: 20.31%), studies that focus only on post-mating events of sexual selection (females: 17.80%; males: 18.92%), studies not on sexual selection (females: 11.41%; males: 11.49%), studies that focus solely on males (males: 3.14%), or on females (males: 3.02%), studies whose complete versions were not found (both sexes: 3.02%), and other (both sexes: 4.35%).

After this second round of filtering, estimates of sexual selection intensity were extracted from 153 studies. Original studies either (I) directly provided calculations of φIs, φIf, φI and the Bateman gradient or (II) provided the raw data needed (i.e., mean mating success, mean fertilization success and mean reproductive success per individual and/or mean estimates per population, with the respective standard deviations). From this initial dataset, estimates on non-Osteichthyes (1 study) and turtles (1 study) were excluded, as well as estimates based on studies run under mesocosm (14 studies) or laboratory (15 studies) conditions. Finally, while running this project, three studies that met inclusion criteria were found sporadically in the literature, and were thus added to the final version of the dataset.

In total, 101 studies provided 634 estimates of the intensity of sexual selection (conting all indices for all sexes).

The methods used to obtain data other than indices for the opportunity for sexual selection are described in the main text of the article that this dataset refers to