Drift effects on the multivariate floral phenotype ofCalceolaria polyrhizaduring a post‐glacial expansion in Patagonia

Abstract

Quaternary environmental changes substantially impacted the landscape and promoted rapid evolutionary changes in many species; however, analyses of adaptive phenotypic variation in plants have usually neglected the underlying historical context. Here, we associate phylogeography and phenotypic evolution by analysing the divergence of Calceolaria polyrhiza multivariate floral phenotype after a Pleistocene post-glacial expansion in Patagonia. Phenotypic matrix (P) properties (size, shape, orientation and phenotypic integration) of six refugium and six recent populations from two different phylogroups were compared following different approaches. We found that P-matrix shape and orientation remained stable despite the strong phylogeographic footprint of post-glacial expansion. However, average proportional reductions in matrix size supported the expectation that drift had a significant effect on the floral phenotype in the northern phylogroup. When phylogeographic history was not included in the analyses, the results overestimated phenotypic differences, whereas under explicit phylogeographic control, drift appeared as the best explanation for matrix differences. In general, recent populations showed a larger phenotypic divergence among them, but a lower overall phenotypic variation than refugium populations. Random Skewers analyses indicated a lower potential response to selection in recently colonized populations than in refugium populations. We discuss that the combination of phylogeographic analyses with geographical distribution of functional phenotypic (genotypic) variation is critical not only to understand how historical effects influence adaptive evolution, but also to improve field comparisons in evolutionary ecology studies.

Natural selection acting on integrated phenotypes: covariance among functional leaf traits increases plant fitness

“…We used the R script from Maubecin et al . () to evaluate the correlation between responses to selection.…”

“…The average correlations range from zero, if matrices are completely different, to one, if both matrices are similar. The magnitude of the response vectors was calculated as the vectors' length ratio (Maubecin et al, 2016). This ratio assesses the similarity in matrix size; if two matrices are equal, the magnitude of the response is expected to be the same for both matrices, and the ratio will be equal to one.…”

“…This ratio assesses the similarity in matrix size; if two matrices are equal, the magnitude of the response is expected to be the same for both matrices, and the ratio will be equal to one. We used the R script from Maubecin et al (2016) to evaluate the correlation between responses to selection.…”

Evolutionary transition between bee pollination and hummingbird pollination inSalvia: Comparing means, variances and covariances of corolla traits

“…Our results provide some insights about the evolution of variances and covariances among floral traits. First, proportional changes in matrix size accounted for a significant amount of variation among P ‐matrices (changes in the total amount of phenotypic variance) suggesting a role of drift (Jones et al., ; Maubecín et al., ; Roff, ). Second, simulations also suggested that variation among species in the magnitude of corolla integration occurred independently from the transition from bee to hummingbird pollination.…”

“…Finally, we performed a series of numerical simulations to examine the potential role of drift and selection on the observed variation of the major axes of the variance-covariance space. Drift is expected to influence the total amount of variance of a matrix (Jones, Arnold, & Bürger, 2003;Roff, 2000), whereas stabilizing and correlational selection may influence phenotypic integration (Maubecín, Cosacov, Sérsic, Fornoni, & Benitez-Vieyra, 2016).…”