Unidirectional response to bidirectional selection on body size II. Quantitative genetics

Abstract

Anticipating the genetic and phenotypic changes induced by natural or artificial selection requires reliable estimates of trait evolvabilities (genetic variances and covariances). However, whether or not multivariate quantitative genetics models are able to predict precisely the evolution of traits of interest, especially fitness-related, life history traits, remains an open empirical question. Here, we assessed to what extent the response to bivariate artificial selection on both body size and maturity in the medaka Oryzias latipes, a model fish species, fits the theoretical predictions. Three lines (Large, Small, and Control lines) were differentially selected for body length at 75 days of age, conditional on maturity. As maturity and body size were phenotypically correlated, this selection procedure generated a bi-dimensional selection pattern on two life history traits. After removal of nonheritable trends and noise with a random effect ("animal") model, the observed selection response did not match the expected bidirectional response. For body size, Large and Control lines responded along selection gradients (larger body size and stasis, respectively), but, surprisingly, the Small did not evolve a smaller body length and remained identical to the Control line throughout the experiment. The magnitude of the empirical response was smaller than the theoretical prediction in both selected directions. For maturity, the response was opposite to the expectation (the Large line evolved late maturity compared to the Control line, while the Small line evolved early maturity, while the opposite pattern was predicted due to the strong positive genetic correlation between both traits). The mismatch between predicted and observed response was substantial and could not be explained by usual sources of uncertainties (including sampling effects, genetic drift, and error in G matrix estimates).

Keywords: Bayesian mixed models; G matrix; animal model; artificial selection; asymmetric response; bivariate selection; evolvability.

FIGURE 1

Illustration of the selection response on body size in the medaka Oryzias latipes after five episodes of artificial selection. This photomontage displays on the same scale individuals from the F6 generation that are very close to the average length of the Large line (20.95 mm, top), Control line (19.80 mm, middle), and Small line (19.63 mm, bottom)

FIGURE 2

Illustration of the selection protocol. From generation F₁, three selected lines (Large, Control, and Small lines) were kept independently, with 15 full‐sib families per line (in practice: mean ± std. dev. = $15.4\pm 2.7$ families per selected line). The density was normalized at 15 fish per tank (in practice: $13.9\pm 5.8$). Fish were pictured, measured, and sex‐determined at 60 and 75 days. Selection happened in two stages, among families at 60 days, and within families at 75 days. At 60 days, 10 families out of 15 were preselected based on two criteria: density (tanks with low fish counts were discarded), and average size. Families discarded from the breeding pool were kept and measured at 75 days, and were thus considered when computing the population mean. At 75 days, breeders were picked within preselected families based on two criteria: maturity (immature fish, which sex could not be determined, were never selected) and size (large fish in the Large line, small fish in the Small line, and random fish in the Control line). Fifteen pairs of fish were formed in each line, in a pattern that minimizes inbreeding. The offspring of each pair then constitutes the next generation

FIGURE 3

(a) Bi‐dimensional cumulative selection gradients in all three lines. Gradients were calculated from selection differentials that take into account both within‐ and among‐family selection. (b) Cumulative selection differentials versus cumulative selection response for Body size in Large and Small lines, centered on the Control line. The regression coefficient (calculated by GLS procedure independently for both lines) is an estimate of heritability $h^2(\pm$std. err.)

FIGURE 4

Response to selection for body length (a and b), and maturity (c and d) for all three lines (red: Large, blue: Small, gray: Control). (a and c) Phenotypic means, (b and d) control‐centered responses. Generations F₀ and F₁ (common to all lines) are drawn in black. Error bars stand for standard errors of the means

FIGURE 5

Graphical representation of the posterior distribution of the (co)variance matrix for genetic, aquarium, and residual effects (full model). Thin lines represent individual iterations of the MCMC chain, while the thick line is the mean posterior. The binomial nature of the maturity trait fixes the residual variance to 1

FIGURE 6

Estimated dynamics of genetic (left and center) and generation (right) effects from the full model. The figure represents the median and 95% support interval over MCMC replicates

FIGURE 7

Estimation of the genetic space that is reachable by genetic drift along the selection response (95% support interval: hyphenated lines) (simulated pedigrees with G matrices sampled from the posterior distribution) versus estimated genetic trends (plain lines)

FIGURE 8

Predicted versus realized selection responses. The expected selection response was simulated by applying the Lande–Arnold equation Δz = Gβ recursively over six generations, using the selection gradients β estimated in Figure 3 and G matrices from the posterior distribution of the animal model applied on data from the Control line (so that the realized selection response was not part of the G matrix estimate). Genetic drift was simulated by applying a random deviation of variance G/N_e with $N_e=25$ every generation. (a and b) predicted time series for body size and maturity, respectively. Plain lines: predicted responses (Large line in red, Small line in blue); open symbols: actual phenotypic response; hyphenated lines and filled symbols: median genetic response estimated from Figure 6. Shaded areas represent the 95% support interval. (c) bivariate average gradients (dotted lines), average predicted responses (plain lines) and average genetic responses (hyphenated lines)