The spatial dynamics of phenotypic selection in the wild

Background/Question/Methods

Our current understanding of general features of natural selection in the wild has focused on determining major patterns in the form and strength, as well as the temporal dynamics of selection. One of the remaining challenges is to understand the spatial dynamics of selection. Spatial variation in environmental features is a ubiquitous feature of the landscape, suggesting that selection should often vary spatially. Understanding this variation is important for a number of reasons. For example, geographic variation in the form or direction of selection can cause divergent selection facilitating speciation, is a critical feature structuring the coevolutionary process, and may account for patterns of evolutionary stasis. Identifying how selection varies spatially may also provide information on what drives variation in selection; that is, what causes selection. For instance, there may be common features of the environment that correlate with spatial variation in selection. To begin understanding the major features of temporal variation in selection we assembled a database of spatially replicated studies of selection in the wild.

Results/Conclusions

We identified 57 studies, resulting in a database containing 3201 estimates of linear selection gradients, 1090 estimates of quadratic gradients, 3036 linear selection differentials, and 1389 quadratic selection differentials. Overall, spatial variation in both the strength and direction of selection estimates among populations appears common, which is consistent with evidence of local adaptation for many populations. These major features of selection are similar when comparing selection gradients (which represent direct selection on a trait by controlling for correlations among included traits) and selection differentials (which represent total selection on a trait). In combination with a previously published database consisting of 89 temporally replicated studies and more than 5500 estimates of selection, it is apparent that selection among populations is typically greater than temporal variation in selection within populations, which would accentuate population divergence. We will also discuss the results of new metaanalytic models which directly take into account sampling error and how results of these models affect our interpretation of the major features of spatial variation in selection. Finally, we examine how spatial variation in selection differs among fitness components and among different types of traits. Results of these metaanalyses reveal that spatial variation in selection is widespread in nature.