Causal inference clarifies the roles of background selection and mutation rate variation in shaping human genetic diversity
Year: 2026
Type: preprint
Abstract: Decades of theoretical and empirical work have struggled to reconcile competing views on how evolution shapes patterns of genetic variation. We now understand the genome as a mosaic molded by both neutral and selective forces, but quantifying their relative contributions remains an open challenge. A major obstacle has been the tendency to analyze each evolutionary process in isolation. But different processes may leave similar signatures on genetic variation, making it challenging to draw conclusions from correlations alone. To address this gap, we make predictions of the landscape of diversity based on background selection and mutation rate variation. We then develop structural equation models describing how mutation, recombination and selection jointly shape the genomic landscape of diversity. This approach offers a more realistic representation of biological interactions and enables rigorous evaluation of hypothesized causal structures, marking both conceptual and practical improvements over previous studies. Analyses of human data reveal large variation in the explanatory power of candidate models across chromosomes. We find that previous studies likely overestimated the predictive accuracy of background selection; although it emerges as the overall main driver of diversity, in some chromosomes mutation rate variation has a comparable impact. We show that recombination increases diversity more through its influence on background selection than its mutagenic effect, resolving a longstanding debate. This work demonstrates that modeling variation inherent in genome biology substantially improves our ability to explain human genetic diversity. At the same time, our results suggest inadequacies in available genomic maps or in background selection as the unique mode of linked selection in humans. Significance Understanding the mechanisms that create and maintain genetic diversity is fundamental to explaining how populations evolve. Early work emphasized the role of linked selection, but this focus has obscured the contributions of other processes. To overcome this limitation, we introduce structural equation models that encode the causal relationships among mutation, recombination, and selection. We find that background selection is the primary driver of genetic diversity, but its predictive power has likely been overstated in previous studies. Meanwhile, mutation rate variation substantially impacts diversity in some chromosomes. We also identify mismatches between model expectations and observed data that point to limitations in available genomic maps or to the insufficiency of background selection models to fully describe linked selection in humans.
Language: English
Subfield: Genetics
Domain: Life Sciences
Open Access status: green
OpenAlex ID: W7163417742