Heritability, selection, and the response to selection in the presence of phenotypic measurement error: effects, cures, and the role of repeated measurements

Abstract

Quantitative genetic analyses require extensive measurements of phenotypic traits, which may be especially challenging to obtain in wild populations. On top of operational measurement challenges, some traits undergo transient fluctuations that might be irrelevant for selection processes. The presence of transient variability, denoted here as measurement error in a broader sense, is a possible cause for bias in the estimation of quantitative genetic parameters. We illustrate how transient effects with a classical measurement error structure may bias estimates of heritability, selection gradients, and the response to selection of a continuous trait, and propose suitable strategies to obtain error-corrected estimates. Unbiased estimates can be retrieved with repeated measurements, given that they were taken at an appropriate temporal scale. However, the fact that repeated measurements were previously used to isolate permanent environmental (instead of transient) effects requires that we rethink their information content. We propose a distinction into "short-term" and "long-term" repeats, whereas the former capture transient variability, and the latter the permanent effects. A simulation study illustrates how the inclusion of appropriate variance components in the models helps retrieving unbiased estimates of all quantities of interest. The methodology is then applied to data from a Swiss snow vole population.

Most published selection gradients are underestimated: Why this is and how to fix it

“…We deem our thesis important because a review of papers estimating standardized selection gradients published in Evolution from 2010-2019 inclusive (Supporting Information Text S1 and Table S1) demonstrates that most published studies fail to (properly) control for this form of bias (Table 1). Specifically, most published estimates are based on traits measured once (193 out of 325 estimates; 59%); these are attenuated under realistic residual within-individual error distributions (Ponzi et al 2018), the extent depending on the type of selection gradient and level of trait repeatability (see section "The Problem"; Table 2). Given that repeatability of most traits generally varies from 0.2 to 0.9 (Bell et al 2009;Holtmann et al 2017), bias in estimates and its effect on our ability to interpret patterns of selection is potentially huge.…”

“…Deriving nonlinear selection gradients from multivariate mixed-model approaches requires a simple extension, which we describe below. Second, errors-in-variables (or "measurement error") models have recently been introduced as an alternative solution (Ponzi et al 2018) and were not employed in any of the studies we reviewed. All three approaches strictly require repeated measures; we use simulations to study bias and precision associated with each approach, for both linear, quadratic, and correlational selection gradient analyses, and for trait repeatability (R) values that are either relatively low (0.3) or high (0.7).…”

The simulation extrapolation technique meets ecology and evolution: A general and intuitive method to account for measurement error

“…If measurement error in other variables is present at the same time, a possible strategy would be to apply the PSIMEX approach to a model that accounts for the (parametric) error in such covariates, for example to a Bayesian hierarchical error model (see e.g. Ponzi et al., ).…”

“…the error variance or the misclassification rate) must be specified prior to correcting for the error. In addition, some techniques require that latent (so‐called ‘exposure’) models specify the distributions of the unobserved (latent), true variables, in particular when errors are modelled in a Bayesian framework (Muff, Riebler, Held, Rue, & Saner, ; Ponzi, Keller, Bonnet, & Muff, ). However, the error‐generating mechanisms that blur true variables can be rather complex, and specifying a model for the unobserved variables may also not be very straightforward.…”

Keragaan dan Keragaman Genetik Genotipe-genotipe F2:3 Gandum (Triticum aestivum L.) di Dataran Tinggi Indonesia

“…Besaran angka diferensial seleksi menunjukkan tingkat keberhasilan dari kegiatan seleksi. Diferensial seleksi merupakan selisih nilai rata-rata populasi hasil seleksi dengan populasi dasarnya (Ponzi et al, 2018). Seleksi langsung pada famili F 2:3 gandum dapat meningkatkan bobot biji per tanaman sebesar 26.65% pada F 2:3 G2/Se hingga 59.29% pada famili F 2:3 G1/Se (Tabel 7).…”