G-matrix stability in clinally diverging populations of an annual weed

Abstract

How phenotypic and genetic divergence among populations is influenced by the genetic architecture of those traits, and how microevolutionary changes in turn affect the within-population patterns of genetic variation, are of major interest to evolutionary biology. Work on Ipomoea hederacea, an annual vine, has found genetic clines in the means of a suite of ecologically important traits, including flowering time, growth rate, seed mass, and corolla width. Here we investigate the genetic (co)variances of these clinally varying traits in two northern range-edge and two central populations of Ipomoea hederacea to evaluate the influence of the genetic architecture on divergence across the range. We find 1) limited evidence for clear differentiation between northern and southern populations in the structure of G, suggesting overall stability of G across the range despite mean trait divergence and 2) that the axes of greatest variation (gmax) were unaligned with the axis of greatest multivariate divergence. Together these results indicate the role of constraint on the divergence among populations across the range.

Strong selection is poorly aligned with genetic variation inIpomoea hederacea: implications for divergence and constraint

“…The copyright holder for this preprint this version posted November 20, 2022. ; https://doi.org/10. 1101/2022…”

“…G is expected to be aligned with divergence as it deflects the response to selection toward the axis of greatest variation (Schluter 1996). In Ipomoea hederacea, when previous estimates of G across four populations were compared with the axis of multivariate clinal divergence (Henry and Stinchcombe 2022) the predicted selection gradients were not aligned with the axis of greatest variation in G (gmax) for any population. Two possible explanations exist which could explain the discordance in the observed patterns: 1) that there was strong, directional selection acting in a direction largely orthogonal to G, and thus generating divergence unaligned with G or, 2) that there were traits missing from our estimates of G and divergence, such that if the missing traits had been included, greater alignment would have been detected (Henry and Stinchcombe 2022).…”

“…We used 303 lines gathered by Campitelli and Stinchcombe (2013), from which we selected lines that had sufficient seeds and to maximize geographic localities. To this, we added 40 maternal lines, randomly chosen from Henry and Stinchcombe's (2022) collections that had sufficient seeds. We note that Stock et al (2014) also studied a subset of (Campitelli and Stinchcombe 2013a) lines, and that our sample includes 19 of the 20 populations studied by Stock et al (2014).…”

“…Third, we used our field estimates of 𝛽 and 𝜸 to determine how much the genetic covariances described by Henry and Stinchcombe (2022) slow the rate of adaptation. We used the metrics described by (Agrawal and Stinchcombe 2009), which compare the rate of adaptation with estimated genetic covariances to a scenario in which covariances are set to zero.…”

“…The copyright holder for this preprint this version posted November 20, 2022. ; https://doi.org/10. 1101/2022 Selection, constraint, and divergence in Ipomoea hederacea 2 interpreted as adaptive (cf. Vasemägi 2006), natural selection acts in conjunction with other evolutionary forces such as genetic drift and gene flow, which can also contribute to divergence or act as homogenizing forces.…”