Environmental drivers of varying selective optima in a small passerine: A multivariate, multiepisodic approach

Abstract

In changing environments, phenotypic traits are shaped by numerous agents of selection. The optimal phenotypic value maximizing the fitness of an individual thus varies through time and space with various environmental covariates. Selection may differ between different life‐cycle stages and act on correlated traits inducing changes in the distribution of several traits simultaneously. Despite increasing interests in environmental sensitivity of phenotypic selection, estimating varying selective optima on various traits throughout the life cycle, while considering (a)biotic factors as potential selective agents has remained challenging. Here, we provide a statistical model to measure varying selective optima from longitudinal data. We apply our approach to analyze environmental sensitivity of phenotypic selection on egg‐laying date and clutch size throughout the life cycle of a white‐throated dipper population. We show the presence of a joint optimal phenotype that varies over the 35‐year period, being dependent on altitude and temperature. We also find that optimal laying date is density‐dependent, with high population density favoring earlier laying dates. By providing a flexible approach, widely applicable to free‐ranging populations for which long‐term data on individual phenotypes, fitness, and environmental factors are available, our study improves the understanding of phenotypic selection in varying environments.

The evolution of phenotypic plasticity when environments fluctuate in time and space

“…; see also Gamelon et al. ). An alternative method, using population‐level spatiotemporal data, is also able (with caveats) to estimate plasticity and the environmental sensitivity of selection (Phillimore et al.…”

Short‐term insurance versus long‐term bet‐hedging strategies as adaptations to variable environments

“…Common examples are thermal performance curves (Angilletta ), optimal clutch or litter sizes (Mountford ; Boyce and Perrins ; Gamelon et al. ), and reproductive benefits versus viability costs of sexually selected ornaments (Andersson and Iwasa ). In these types of scenarios, uncertainty across instances in any component determining individual fitness will cause the optimal trait value to differ from the trait value at the peak of the fitness function (Yoshimura and Shields ; Parker and Smith ).…”

A time‐series model for estimating temporal variation in phenotypic selection on laying dates in a Dutch great tit population

“…() and Gamelon et al. () in several new ways. First, instead of assuming a fixed zero‐inflation parameter for modelling the number of fledglings as in Chevin et al.…”

“…Second, in addition to random effects on the peak and location of the fitness optimum as in Gamelon et al. (), we also allow the width of the fitness function to vary between years, with all three properties of the Gaussian fitness function jointly following a vector autoregressive process. Such variation in the width is of theoretical importance for the evolution of the phenotypic variance (Zhang & Hill, ) and for the evolutionary stability of the additive genetic variance‐covariance matrix (Revell, ).…”

“…As in Gamelon et al. (), we implement our method using TMB (Kristensen et al., ), an R package providing a comprehensive framework for fast fitting nonlinear, complex, latent variable models.…”