Detection of Domestication Signals through the Analysis of the Full Distribution of Fitness Effects

Domestication is a process marked by complex interactions between demographic changes and selective pressures, which together shape genetic diversity. While the phenotypic outcomes of domestication are well documented, its genetic basis—particularly the dynamics of selection— remain less well understood. To investigate these dynamics, we performed simulations designed to approximate the demographic history of large domestic mammals. These simulations used selection coefficients as a modeling tool to represent changes in selection pressures, recognizing that such coefficients are abstractions rather than direct representations of biological reality. Specifically, we analyzed site frequency spectra (SFS) under varying distributions of fitness effects (DFE) and proportions of mutations with divergent selective pressures. Our results show that the discretized deleterious DFE can be reliably inferred from the SFS of a single population, but reconstructing the beneficial DFE and demographic history remains challenging, even when using the joint SFS of both populations. We further developed a novel joint DFE inference model to estimate the proportion of mutations with divergent selection coefficients (p_c), although we found that signals of classic hard sweeps can mimic increases inp_c, complicating interpretation. These findings underscore both the utility and limitations of DFE inference and highlight the need for caution when interpreting demographic histories in domesticated populations based on such modeling assumptions.