Evolution of floral traits: from ecological contex to functional integration
MetadataVis full innførsel
- Institutt for biologi 
Whether macroevolution, that is, evolution above the population level, can be explained by micro-evolutionary processes happening at the population level remains debated. In this thesis, we tested two hypotheses concerning the link between micro- and macro-evolution of the blossom of the Neotropical vine Dalechampia scandens. The first hypothesis posits that patterns of population differentiation reflect local variation in selection generated by pollinators and antagonistic interactors. To test this hypothesis, we first estimated natural selection mediated by pollinators and seed predators on various blossom traits. We then survey 22 populations of D. scandens in Costa-Rica to test whether genetic differentiation among populations reflected the patterns of selection inferred from the variation in pollinators and predators interacting with D. scandens. We found evidence of pollinator mediated selection on specific traits, especially under low pollinator abundance. In contrast, seed predators did not generate selection on specific blossom traits. At the macro-evolutionary scale, population divergence in blossom traits appeared mostly affected by the pollinator reliability, that is, the abundance and the type of pollinator present in each population. The second hypothesis suggests that patterns of genetic variance within population may constrain evolution at higher taxonomic scale (i.e. among populations and species). We tested this hypothesis by comparing the predicted response to selection with the response observed in an artificial-selection experiment on various blossom traits conducted on two species of the D. scandens species complex. We also compared patterns covariance among traits at various taxonomic levels to test whether patterns observed at the among-populations and among species levels were congruent with the patterns of genetic covariance (within-population). After four episodes of artificial selection, predicted and observed responses were mostly congruent for the selected traits. The response of the correlated trait was less predictable and differed between the two species. Furthermore, despite strong genetic correlation observed between seed size and blossom size, the two traits did not covary at the among population level. These results suggest that the genetic architecture may not strongly constrain pattern of macroevolution in our study system.