Quantifying the causal structure of natural and sexual selection

• Applicant: Dr. Jonathan Henshaw
• Subject Area: Evolution, Anthropology ; Bioinformatics and Theoretical Biology
• Term: since 2021
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 456626331

Causal thinking is essential to a nuanced understanding of natural selection. At the coarsest explanatory level, selection and its evolutionary consequences can be treated as purely statistical phenomena. However, explaining why selection favours certain traits requires explicitly causal analyses of the interrelationships among phenotypic traits, fitness components, and fitness. Causality is also inherent to a family of key evolutionary concepts - sexual selection, fecundity selection, longevity selection, and so on – that can be understood as selection via particular fitness components (e.g. sexual selection arises when traits affect mating or fertilisation success, which in turn affect fitness). Remarkably, we still lack a general quantitative approach for teasing apart how various causal pathways contribute to natural selection. Existing approaches depend on assumptions such as linearity and additivity, which fail to capture the complexities of real world causal effects. This is particularly detrimental to the field of sexual selection, which consequently lacks rigorous operational definitions of its core concepts.I aim to solve these problems by leveraging recent developments in the mathematical theory of causal processes. I will first develop a general framework for quantifying how various causal pathways contribute to natural selection and its evolutionary consequences. This framework will accommodate arbitrary causal relationships among variables (e.g. non-linearity, non additive interactions) and will be compatible with a broad range of regression techniques for estimating such relationships. I will then apply this framework to the problems of separating sexual from non-sexual selection, and of disentangling the various causal processes that generate sexual selection. These include pre-mating sexual selection (acting via mate number or mate quality) and post-mating sexual selection (acting via fertilisation success). Lastly, I will make these methods accessible to the evolutionary community by developing an R software package.The proposed framework will be a fundamental theoretical advance and of interest to any biologist wishing to understand how selection operates in their study system. The framework will be applicable to observational data if particular causal assumptions hold. This feature is desirable, because the multi-step causal processes that determine fitness are often difficult to manipulate experimentally. The framework’s flexibility will ensure that – unlike in existing approaches – complex causal processes can be modelled using appropriate statistical assumptions, which is crucial to generating meaningful estimates and predictions. The application to sexual selection will contribute to the disambiguation of core concepts in this field. Further, general methods for quantifying sexual selection and its components are essential to resolving many current controversies (e.g. over the primary mechanisms of sexual selection).

• DFG Programme: Research Grants