Interactions between microenvironment, selection and genetic architecture drive multiscale adaptation in a simulation experiment

Abstract

When environmental conditions differ both within and among populations, multiscale adaptation results from processes at both scales and interference across scales.We hypothesize that within-population environmental heterogeneity influences the chance of success of migration events, both within and among populations, and maintains within-population adaptive differentiation. We used a simulation approach to analyse the joint effects of environmental heterogeneity patterns, selection intensity and number of QTL controlling a selected trait on local adaptation in a hierarchical metapopulation design. We show the general effects of within-population environmental heterogeneity: (i) it increases occupancy rate at the margins of distribution ranges, under extreme environments and high levels of selection; (ii) it increases the adaptation lag in all environments; (iii) it impacts the genetic variance in each environment, depending on the ratio of within-to between-populations environmental heterogeneity; (iv) it reduces the selection-induced erosion of adaptive gene diversity.Most often, the smaller the number of QTL involved, the stronger are these effects.We also show that both within-and between-populations phenotypic differentiation (Q ST ) mainly results from covariance of QTL effects rather than QTL differentiation (F STq ), that within-population QTL differentiation is negligible, and that stronger divergent selection is required to produce adaptive differentiation within populations than among populations. With a high number of QTL, when the difference between environments within populations exceeds the smallest difference between environments across populations, high levels of within-population differentiation can be reached, reducing differentiation among populations. Our study stresses the need to account for within-population environmental heterogeneity when investigating local adaptation.

Common microgeographical selection patterns revealed in four European conifers

“…One consequence of this finding is that microgeographical divergence appears to contribute to maintaining genetic diversity in populations through the existence of multiple optima (Delph & Kelly, 2014), in a kind of self‐sustained dynamic process. This finding reinforces the evidence of the strength of microgeographical adaptation because, in a hierarchical environmental heterogeneity pattern, when many quantitative trait loci (QTL) are involved in an adaptive trait, microgeographical adaptation is expected to rely mainly on the covariances among QTL and strong selection intensity is required to affect individual QTL frequencies as we studied here (Cubry et al, 2022). The estimated values of scaled selection partially match theoretical expectations for the intensity of selection needed to maintain divergence with gene flow, as derived by Yeaman and Whitlock (2011) and Yeaman and Otto (2011), and to which we can attempt a simple comparison as follows.…”

Rolling down that mountain: microgeographical adaptive divergence during a fast population expansion along a steep environmental gradient in European beech

“…Population-genetic theory predicts that fewer, larger-effect loci should underlie local adaptation relative to global adaptation (Yeaman 2022). Therefore, it is not surprising that we identified few outliers, seemingly undergoing strong selection, even though a compact genetic control should limit adaptation to very divergent local conditions (Cubry et al 2022). It is worth noticing that large-effect mutations that confer advantage in new environments can actually favour population expansion (Gilbert and Whitlock 2017); such loci could therefore actually have facilitated recolonisation.…”

“…In addition to this, the use of SFS-derived statistics to infer demography has been criticised because of its high sensitivity to fluctuations in SFS (Myers et al 2008;Lapierre et al 2017). Another limitation of our work is that we mostly focused on allele frequency shifts rather than changes in allele frequency covariances, which are known to be more involved in the early stages of adaptation (Le Kremer 2003, 2012;Cubry et al 2022). This may explain the low number of outliers detected here.…”

A demo‐genetic model shows how silviculture reduces natural density‐dependent selection in tree populations

“…The evolutionary rate was higher, and the reduction of additive variance and QTL diversity was lower, with 50 QTL than with 10 QTL. This difference may be attributable to the higher contribution of covariance between QTL to the selection response with 50 QTL than with 10 QTL (Cubry et al., 2022; Le Corre & Kremer, 2012). The evolutionary predictions of the model were much less sensitive to the distribution of allelic effects and allelic frequencies with 50 QTL than with 10 QTL.…”

“…The response to selection depends on the initial genetic variance and, for a given initial variance, changes in genetic mean and variance depend on the genetic architecture of the trait, defined by the number of QTL, the distribution of allelic effects, and the allele frequencies. To test the sensitivity of the model to assumptions made on the genetic architecture of the variable trait (the genetic architecture is generally unknown), we fixed the initial additive variance of vigor and considered two contrasted genetic architectures, either 10 or 50 QTL, which are expected to result in slightly different micro‐evolutionary outcomes (Cubry et al., 2022). We considered two cases of initial population: a reference with no phenotypic variation in vigor, i.e., no QTL effects and no environmental variance, and a panmictic population with genetic and environmental variation.…”