Directional Selection Rather Than Functional Constraints Can Shape the G Matrix in Rapidly Adapting Asexuals

Abstract

Genetic covariances represent a combination of pleiotropy and linkage disequilibrium, shaped by the population's history. Observed genetic covariance is most often interpreted in pleiotropic terms. In particular, functional constraints restricting which phenotypes are physically possible can lead to a stable G matrix with high genetic variance in fitness-associated traits, and high pleiotropic negative covariance along the phenotypic curve of constraint. In contrast, population genetic models of relative fitness assume endless adaptation without constraint, through a series of selective sweeps that are well described by recent traveling wave models. We describe the implications of such population genetic models for the G matrix when pleiotropy is excluded by design, such that all covariance comes from linkage disequilibrium. The G matrix is far less stable than has previously been found, fluctuating over the timescale of selective sweeps. However, its orientation is relatively stable, corresponding to high genetic variance in fitnessassociated traits and strong negative covariance-the same pattern often interpreted in terms of pleiotropic constraints but caused instead by linkage disequilibrium. We find that different mechanisms drive the instabilities along vs. perpendicular to the fitness gradient. The origin of linkage disequilibrium is not drift, but small amounts of linkage disequilibrium are instead introduced by mutation and then amplified during competing selective sweeps. This illustrates the need to integrate a broader range of population genetic phenomena into quantitative genetics.

Mutation bias can shape adaptation in large asexual populations experiencing clonal interference

“…We therefore also simulate the simultaneous evolution of two traits (k = 1, 2), with beneficial mutations of fitness effect s k appearing at rate U k , using an exten-220 sion of the method of Gomez et al [26]. We consider strong selection relative to population size, such that 1/N s k ≤ 1.…”

“…In the last step we set 245 abundances of deterministic classes (n * i,j > 1/Q i,j ) to n i,j (t + 1) = Round n * i,j (t) , while classes that grow stochastically (n * i,j ≤ 10/Q i,j ) are instead sampled from a Poisson distribution with mean n * i,j (t). We implemented the simulation method described 250 above using Matlab code originally developed by Pearce and Fisher [32], modified by Gomez et al [26] to deal with two traits, and here modified to allow those traits to have distinct mutation rates and selection coefficients.…”

“…Models that have previously been used to dismiss 105 mutation-biased adaptation when U N log(N s) ≥ 1 [3] are simple, and do not capture the multi-locus complexities included in more recent traveling wave models of population genetics [22,23,24,25]. Here we build on a recent two-dimensional travelling wave model [26] 110 to formally investigate the circumstances under which mutation-biased adaptation can occur.…”

Evolution of Microbial Growth Traits Under Serial Dilution

“…We therefore need a quantitative description of both between-population as well as within-population covariation of traits of microbial populations in this regime. In the present study, we focus on between-population covariation in growth traits, but recent work by Gomez et al (2019) provides insight into the case of within-population covariation. They showed that a tradeoff across individuals within a population evolves between two quantitative traits under positive, additive selection; this suggests that while growth rate and lag time will be negatively correlated across populations ( Figure 5, D and E), they should be positively correlated within populations.…”

Selection and drift determine phenotypic stasis despite genetic divergence

“…However, the Robertson (1961) model presupposes sequential, complete sweeps due to directional selection on initially rare QTL alleles, which may not be case when alleles are maintained at intermediate frequencies under stabilizing selection, as recently suggested with Drosophila experiments (Barghi et al, 2019) and likely to be the case here (Chelo and Teotónio, 2013). It is also possible, and perhaps probable in our setting, that small allele frequency and LD differences between replicate populations due to sampling of the domesticated population were accentuated by selection during the focal stage, resulting in variation between replicate populations of QTL effects on trait covariances (Bohren et al, 1966; Gomez et al, 2019). This process should be dampened by linked selection (Hill and Robertson, 1966; Zhang and Hill, 2005), although strong oligogenic sign epistasis we observed previously for fitness-related traits may exacerbate such sampling effects (Noble et al, 2017).…”