Polygenic adaptation to an environmental shift: temporal dynamics of variation under Gaussian stabilizing selection and additive effects on a single trait

Abstract

1Predictions about the effect of natural selection on patterns of linked neutral variation are 2 largely based on models involving the rapid fixation of unconditionally beneficial mutations. 3 However, when phenotypes adapt to a new optimum trait value, the strength of selection 4 on individual mutations decreases as the population adapts. Here, I use explicit forward 5 simulations of a single trait with additive-effect mutations adapting to an optimum shift. 6 Detectable "hitch-hiking" patterns are only apparent if i. the optimum shifts are large with 7 respect to equilibrium variation for the trait, ii. mutation rates to large-effect mutations 8 are low, and iii., large-effect mutations rapidly increase in frequency and eventually reach 9 fixation, which typically occurs after the population reaches the new optimum. For the pa-10 rameters simulated here, partial sweeps do not appreciably affect patterns of linked variation, 11 even when the mutations are strongly selected. The contribution of new mutations versus 12 standing variation to fixation depends on the mutation rate affecting trait values. Given the 13 fixation of a strongly-selected variant, patterns of hitch-hiking are similar on average for the 14 two classes of sweeps because sweeps from standing variation involving large-effect mutations 15 are rare when the optimum shifts. The distribution of effect sizes of new mutations has little 16 effect on the time to reach the new optimum, but reducing the mutational variance increases 17 the magnitude of hitch-hiking patterns. In general, populations reach the new optimum prior 18 to the completion of any sweeps, and the times to fixation are longer for this model than for 19 standard models of directional selection. The long fixation times are due to a combination 20 of declining selection pressures during adaptation and the possibility of interference among 21 weakly selected sites for traits with high mutation rates. 22 24 lations by analyzing genome-wide patterns of polymorphism. The interpretation of observed 25 patterns relies heavily on mathematical models, accompanied by various simulation methods, 26 which make concrete predictions about the effect of evolutionary forces (natural selection, 27 demographic events, etc.) on patterns of variation.28The models of natural selection used to interpret data come primarily from what we may 29 call "standard population genetics" models. In these models, mutations have a direct effect 30 on fitness (a "selection coefficient"). The fitness effects of mutations are most often assumed 31 to be constant over time. For example, background selection is a model of unconditionally 32 deleterious mutations resulting in strong purifying selection (Charlesworth et al., 1993, 33 1995; Hudson and Kaplan, 1995; Cvijović et al., 2018). The model of a selective sweep 34 from a new mutation similarly posits that the variant is unconditionally beneficial with a 35 constant effect on fitness over time (Maynard-Smith and Haigh, 1974; Kaplan et al., 36 1989; Braverman et ...

Low concordance of short‐term and long‐term selection responses in experimental Drosophila populations

“…Sweep simulations result in haplotype blocks with many selection targets (Barghi & Schlötterer, 2020), which in turn influences the number of candidate SNPs. Simulations of polygenic adaptation can become quite complex and rich in parameters (Thornton, 2019), which limits the power of computer simulations to scrutinize this result further. Rather, experimental validation of selection targets in secondary E&R studies (Burny et al., 2020) may be used to confirm selection signatures beyond statistical testing and could serve an important role to test early selection signatures that cannot be confirmed at later time points.…”

Low concordance of short-term and long-term selection responses in experimentalDrosophilapopulations

“…Sweep simulations result in haplotype blocks with many selection targets (Barghi and Schlötterer, 2020), which in turn influences the number of candidate SNPs. Simulations of polygenic adaptation can become quite complex and rich in parameters (Thornton, 2019), which limits the power of computer simulations to scrutinize this result further. Rather, experimental validation of selection targets in secondary E&R studies (Burny et al, 2020) may be used to confirm selection signatures beyond statistical testing and could serve an important role to test early selection signatures that cannot be confirmed at later time points.…”

“…All statistical tests, which are evaluating a parallel selection signature across replicates, are more likely to detect selection signatures shared across replicates, even with only moderate allele frequency changes. This "enhanced" parallelism could result in wrong or incomplete conclusions about the underlying genetic architecture (Jain and Stephan, 2017a;Höllinger et al, 2019;Thornton, 2019). The analysis of selection signatures in replicated experiments running for only a moderate number of generations is more likely to detect parallel than replicate specific selection signatures.…”

Polygenic Adaptation in a Population of Finite Size

“…Recognizing this problem, the studies published in the past few years [ 33 , 34 , 35 , 36 , 37 ] include genetic drift (due to finite population size) into their polygenic models. Simons et al (2018) [ 35 ] proposed a model of selection that simultaneously acts on multiple traits (pleiotropy).…”

“…For a single trait their model is identical to the one we discuss below. Stetter et al (2018) [ 36 ] and Thornton (2019) [ 37 ] used extensive forward simulations to analyze a model (though with relatively few selected loci) that also includes neutral loci linked to selected ones. This enables them to study genetic hitchhiking in a (potentially) polygenic context.…”