The Genome-wide Signature of Short-term Temporal Selection

Abstract

Despite evolutionary biology's obsession with natural selection, few studies have evaluated multi-generational series of patterns of selection on a genome-wide scale in natural populations. Here, we report on a nine-year population-genomic survey of the microcrustacean Daphnia pulex. The genome-sequences of > 800 isolates provide insights into patterns of selection that cannot be obtained from long-term molecular-evolution studies, including the pervasiveness of near quasi-neutrality across the genome (mean net selection coefficients near zero, but with significant temporal variance about the mean, and little evidence of positive covariance of selection across time intervals), the preponderance of weak negative selection operating on minor alleles, and a genome-wide distribution of numerous small linkage islands of observable selection influencing levels of nucleotide diversity. These results suggest that fluctuating selection is a major determinant of standing levels of variation in natural populations, challenge the conventional paradigm for interpreting patterns of nucleotide diversity and divergence, and motivate the need for the development of new theoretical expressions for the interpretation of population-genomic data.

Keywords: Daphnia pulex; effective population size; fluctuating selection; genetic diversity; linked selection effects; natural selection; population genomics; quasi-neutrality.

Figure 1.

Features of site-specific selection coefficients, across genomic sites. A) Distribution of time-averaged (mean) selection coefficients, $\overline{s}$, for the full set of genomic sites, given for different windows of minor-allele frequencies (MAFs); the table to the right gives averages over all MAFs for different functional categories of sites. B) Means of site-specific $\overline{s}$, the standard deviations of $\overline{s}$ among sites after removal of sampling variance (red points), and the average temporal variance of site-specific s, again after removing variance associated with sampling (blue points) as a function of minor-allele frequency. Note that in all cases the standard errors of the estimates are on the order of the width of the plotted points or smaller. C) The distribution of the site-specific estimates of temporal standard deviations of s (after removing the contribution from sampling error).

Figure 2.

Average temporal covariance of selection coefficients for individual sites over four different time intervals (T = 1, 2, 3, 4). Results are given for the averages over all sites, and separately for the averages conditional on being positive or negative. The range narrows to the right owing to the reduction in sampling variance when minor-allele frequencies are high. The dashed line is the reference for a covariance of zero. In all cases, the standard errors of the estimates arc on the order of the width of the plotted points or smaller.

Figure 3.

Genome-wide temporal covariance of allele-frequency changes across intervals with increasing length. With eight consecutive annual sampling points, there arc seven ways to compute singlc-ycar changes in adjacent years, six ways to compare single-year changes separated by two years, etc., and the plotted values denote the means and SEs of such sets. (There is only one way to compare changes across a seven year interval, the change between the first two years vs. that between the final two years, and so the SE is only reported for the full-genome analysis, based on the SE of the means of the window-specific estimates). Results are given for the full set of polymorphic sites as well as for narrower windows of minor-allele frequencies (denoted in the inset). For each time interval, the results for the different MAF windows are offset slightly on the x axis for visualization purposes.

Figure 4.

Chromosome-wide scans of window-specific intensities of selection. A) For the set of SNPs within each window, the average estimate of absolute values of selection coefficients was compared with the probability distribution based on samples of randomized sites, with P denoting the probability of achieving the observed measure by chance. The dashed red lines denote the critical cutoff points for significance after correcting for multiple comparisons. Horizontal blue bars denote the approximate locations of centromeric regions based on multiple laboratory crosses (Molinier et al. 2021). B) Decline of the correlation of window-specific selection strengths with increasing distance, reported as both the squared correlation coefficients and the slopes of the regressions of pairs of windows at the window distance designated on the x-axis, where x = 1 denotes adjacent (nonoverlapping) windows. C) Cumulative frequency distributions for the lengths of ISS (islands of strong selection) blocks residing on chromosome arms vs. windows.

Figure 5.

Response of expected within-population measures of nucleotide diversity as a function of the temporal standard deviation of the selection coefficient (σ_s) and the population size (N). A) Nucleotide diversity at a linked silent site (π_s) relative to the neutral expectation at drift-mutation equilibrium under free recombination. B) Ratio of diversity at the selected site to that at a completely linked neutral site, π_N/π_s. Color legend in the upper panel designates various levels of the temporal standard deviation of the selection coefficient s; inset in the lower panel designates the symbol shapes used for three levels of the average selection coefficient $\overline{s}$.