Stronger net selection on males across animals

  1. Lennart Winkler
  2. Maria Moiron
  3. Edward H Morrow
  4. Tim Janicke

  • Curated by eLife

    eLife logo

    Evaluation Summary:

    This study addresses an interesting and important question in evolutionary biology: how does the variance in fitness (components) vary between the sexes? In particular, it aims to evaluate whether there is a larger sex difference in systems with strong sexual selection. This study will be of considerable interest to researchers working on sexual coevolution and the role of sexual selection in promoting adaptation. However, there are some concerns regarding the limitations of the data and methods in support of the conclusions.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #3 agreed to share their name with the authors.)

Reviewed by

Read the full article

Listed in

Log in to save this article

Abstract

Sexual selection is considered the major driver for the evolution of sex differences. However, the eco-evolutionary dynamics of sexual selection and their role for a population’s adaptive potential to respond to environmental change have only recently been explored. Theory predicts that sexual selection promotes adaptation at a low demographic cost only if sexual selection is aligned with natural selection and if net selection is stronger on males compared to females. We used a comparative approach to show that net selection is indeed stronger in males and provide preliminary support that this sex bias is associated with sexual selection. Given that both sexes share the vast majority of their genes, our findings corroborate the notion that the genome is often confronted with a more stressful environment when expressed in males. Collectively, our study supports one of the long-standing key assumptions required for sexual selection to bolster adaptation, and sexual selection may therefore enable some species to track environmental change more efficiently.

Article activity feed

  1. Version published to 10.7554/elife.68316 on eLife
  2. eLife

    Evaluation Summary:

    This study addresses an interesting and important question in evolutionary biology: how does the variance in fitness (components) vary between the sexes? In particular, it aims to evaluate whether there is a larger sex difference in systems with strong sexual selection. This study will be of considerable interest to researchers working on sexual coevolution and the role of sexual selection in promoting adaptation. However, there are some concerns regarding the limitations of the data and methods in support of the conclusions.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #3 agreed to share their name with the authors.)

  3. eLife

    Reviewer #1 (Public Review):

    This study addresses an interesting and important question in evolutionary biology: how does the variance in fitness (components) vary between the sexes? In particular, it aims to evaluate whether there is a larger sex difference in systems with strong sexual selection. Using estimates from the literature, this study compares the (CV) between the sexes for two major fitness components: reproductive success and lifespan. Taxa are categorized as polygamous or socially monogamous. Despite my concerns below, I was glad to have read this manuscript.

    The main finding, illustrated in Figure 1, is that both phenotypic and genetic CV is greater for males than females in polygamous species but not monogamous ones; there is no difference for lifespan. I would have parsed their study as two questions (Q1) Is CV greater for males than females, on average? (Q2) Is the extent of that difference greater for polygamous than socially monogamous species. Starting with the manuscript's title, the authors have heavily emphasized Q2, which is too bad because I think they are on particularly shaky ground there.

    Figure 1 is striking! However, the multitude of data points shown is somewhat misleading. Unfortunately, the authors are limited by the available data in the literature. They were able to find 101 estimates for reproductive success but these come from just 26 species. Further, only 6 of species are classified as socially monogamous. Perhaps, most concerningly, all 6 of these socially monogamous species are vertebrates (5 birds + humans) and the data set contains only 6 polygamous vertebrate species. Within this vertebrate phylogeny there are only two transitions to social monogamy. At some level, their contrast amounts to a comparison of 2 P versus 2 M or 6 P versus 6 M, depending on your level of concern about phylogenetic confounds. The authors analyze all 101 estimates from the 26 species (including the 14 invertebrates at are all P). Unfortunately, the data are not presented in a manner that allows readers to see how CV values relate to taxonomic groups. The authors attempt to control for phylogenetic confounds through the magic of PGLMM and their results are statistically significant. I am not an expert in PGLMMs but I know that those who have serious concerns about their potential to mislead (e.g., Uyeda et al. 2018). This feels like a case where there is the potential for a fancy (but imperfect) statistical method overriding common sense. At the very least, I would have liked to see what the 6 P vs the 6 M vertebrates looked like. I would like to see the distribution of logCVR for the 12 vertebrate species (6M v 6P). There should be just 12 points (use mean of log CVR if you have multiple estimates from certain species). If there is a clear pattern, I will be much more comfortable than I am now.

    I would have appreciated some more information about the data itself. In particular, I would like to know what "reproductive success" means. The methods simply describe it as "number of offspring". I can imagine there is large heterogeneity among studies about what this means. I have several questions. Is this life time in some studies but only from a single breeding season in others. Perhaps, most importantly, how is offspring number measured for each sex? Is it measured in the same way / with the same certainty? I assume offspring number is reasonably straightforward to measure for females but what about for males. Are ALL the mates of males known and are the offspring of these mates genotyped to assign paternity? This is not simply an issue of more measurement error in males than females; in some cases it could lead to a problematic bias. For example, if females always mate with a single male (or only use the sperm of a single male), then at least the phenotypic variance of males must be greater than for females if there is any variance in number of mates per male; this is not true if there is multiple mating. If it is assumed a male's reproductive success is his mate number times the average fecundity of those mates (even if multiple mating occurs), then the variance of males will be inflated. Even for evaluating Q1, I would like to be convinced the authors have been diligent and thoughtful about whether "reproductive success" has been measured in equivalent and accurate ways between the sexes. In addition to more information in the Methods, there should be a supplementary material that provides a brief description of the data from each of the 55 studies used here. I would be concerned if counts were done at different stages (e.g., eggs vs juveniles) for assessing female vs. male paternity or if there are strong paternity assumptions employed in estimating male numbers. For Q2, I would like to be reassured that there are not important methodological differences in measuring reproductive success between their 6 socially monogamous species and the remainder.

    In the Introduction the authors put a lot of emphasis on the value of the genetic variance in these fitness components and argue that such a difference reflects stronger selection. I have two caveats here. First, there could be more "male limited" genes, i.e., genes that affect males but have no fitness effects (perhaps because they are not expressed) in females. These genes are, of course, more strongly selected in males than females but they are also irrelevant to the context the author's have used to motivate this study. The authors are studying fitness components not fitness. Offspring number should, more or less, be directly proportional to fitness but longevity less so (i.e., an individual that lives twice as long as another will typically not be twice as fit). Another way to think of it is there is positive selection on fitness components but not necessarily purely linear selection. Moreover, it is easy to imagine scenarios where this selection differs considerably between the sexes. This means, for example, that even if the genetic variance (or CV) for longevity is the same for the two sexes, the resulting genetic variance (or CV) for fitness could be different.

  4. eLife

    Reviewer #2 (Public Review):

    This study aims to test the long-held assumption that males typically experience stronger selection than females, and that strong sexual selection on males can therefore promote the removal of deleterious mutations from the gene pool and facilitate adaptation. The authors compared 101 estimates of male vs. female genetic variance in reproductive success and lifespan across 26 animal species, and found that genetic variance was generally greater in males for reproductive success but not for lifespan. They also found that this pattern held for polygamous species but not for monogamous ones, as expected if males of monogamous species experience weak sexual selection.

    I am generally convinced by the evidence of stronger selection on male reproductive success, but not entirely convinced that this would tend to enhance female fitness and promote adaptive evolution. If sexual conflict plays a major role--such that females and males benefit by expressing different genetic alleles, or male adaptations directly harm females--then stronger selection on males will tend to reduce female fitness. The comparison of "polygamous" vs. "monogamous" categories is also not convincing. I explain my concerns more fully below.

    The existence of sexual conflict challenges the authors' interpretation and conclusions, and this issue needs to be addressed more fully. This problem comes up in several places in the manuscript. For example:

    L68: "In contrast to the solid support for an alignment of sexual and natural selection in many species..." This statement seems to gloss over the problem of intra-locus sexual conflict (mentioned by the authors earlier in the Introduction).If intra-locus conflict is indeed pervasive, strong sexual selection on males would not enhance mean female fitness or facilitate ecological adaptation. Indeed, strong selection on males would tend to have the opposite effect: it would tend to pull female traits away from their viability-selected optima. Are the authors assuming that this form of sexual conflict is relatively unimportant? If so, they should state that clearly and explain why they feel that this assumption is justified.

    L368: This broad conclusion likewise glosses over the problem of sexual conflict. Intra-locus sexual conflict directly undermines this conclusion because strong selection on males will tend to pull females away from their viability optimum. Moreover, inter-locus sexual conflict is also a potential problem, because male-benefit traits will tend to harm females directly. Strong selection on harmful male-benefit traits (such as toxic accessory gland proteins) could therefore reduce mean female fecundity and perhaps impede ecological adaptation as well. The authors need to deal with the problem of sexual conflict, or at least to acknowledge the assumptions that they're making in this regard (i.e., that sexual conflict tends to be weak).

    In addition, the comparison of "polygamous" vs. "monogamous" species is taxonomically biased and unconvincing. Instead of dichotomizing the data into monogamous vs. polygamous, it would be better to use a continuous metric of the strength of sexual selection. The authors argue that relevant data for such a metric are lacking in most of these species, but perhaps useful proxies of relative sexual selection strength could be identified? At any rate, this analysis as it stands is not very convincing. I'm worried about the monogamous/polygamous dichotomy not only because it's a somewhat artificial dichotomy, but also because it's taxonomically biased. The "monogamous" category consists entirely of birds and one dubious mammalian example (Homo sapiens). A continuous metric of sexual selection strength could get around this problem of taxonomic bias. If it's not possible to develop such a metric then perhaps the monogamous/polygamous comparison should be de-emphasized and placed in the supplementary material.

  5. eLife

    Reviewer #3 (Public Review):

    The authors investigate whether net selection is generally stronger on males than females - as has been hypothesised, and which is a requirement of the theory that (sexual) selection on males can "purge the genome" of low-fitness alleles. They do this by compiling 101 estimates of the genetic variance in fitness in both sexes using phylogenetic models, from an initial sample of >3000 studies.

    I found the literature search and statistical methods to be excellent throughout, and all the key details are present (e.g. PRISMA diagram). The paper could be further enhanced by depositing the annotated R code that was used to analyse the data along with all the key files (e.g. the phylogeny), as well as all the data collected for the meta-analysis, once the paper is accepted. The authors could check the R package workflowr and the code report of Cally et al 2020 (Nature Comms) for ideas of what is possible when documenting R code from a meta-analysis.

    I think the paper will be widely cited, as it has at least 3 interesting conclusions:

    - There is partial support for the notion of the 'phenotypic gambit': phenotypic variance was not well-correlated with genetic variance in reproductive success, though it was correlated in the case of lifespan
    - There was a male bias in phenotypic and genetic variance in reproductive success, but not lifespan, as was tentatively predicted in the Introduction based on evolutionary theory
    - This result held true especially in non-monogamous species, as predicted because these species typically have stronger sexual selection. This is a striking result given that it's a second-order hypothesis (i.e. it involves an interaction between sex and mating system) and statistical power/precision is expected to be low.

  6. Version published to 10.1101/2021.04.16.440171v1 on bioRxiv