Understanding how development changes the genetic covariance of complex phenotypes is fundamental for the study of evolution. If the genetic covariance changes dramatically during postnatal ontogeny, one cannot infer confidently evolutionary responses based on the genetic covariance estimated from a single postnatal ontogenetic stage. Mammalian skull morphology is a common model system for studying the evolution of complex structures. These studies often involve estimating covariance between traits based on adult individuals. There is robust evidence that covariances changes during ontogeny. However, it is unknown whether differences in age-specific covariances can, in fact, bias evolutionary analyses made at subadult ages. To explore this issue, we sampled two marsupials from the order Didelphimorphia, and one precocial and one altricial placental at different stages of postnatal ontogeny. We calculated the phenotypic variance-covariance matrix (P-matrix) for each genus at these postnatal ontogenetic stages. Then, we compared within genus P-matrices and also P-matrices with available congeneric additive genetic variance-covariance matrices (G-matrices) using Random Skewers and the Krzanowsky projection methods. Our results show that the structural similarity between matrices is in general high (> 0.7). Our study supports that the G-matrix in therian mammals is conserved during most of the postnatal ontogeny. Thus it is feasible to study life-history changes and evolutionary responses based on the covariance estimated from a single ontogenetic stage. Our results also suggest that at least for some marsupials the G-matrix varies considerably prior to weaning, which does not invalidate our previous conclusion because specimens at this stage would experience striking differences in selective regimes than during later ontogenetic stages.

The Drosophila wing is a structure shared by males and females with the main function of flight. However, in males, wings are also used to produce songs, or visual displays during courtship. Thus, observed changes in wing phenotype depend on the interaction between sex-specific selective pressures and the genetic and ontogenetic restrictions imposed by a common genetic architecture. Here, we investigate these issues by studying how the wing has evolved in twelve populations of Drosophila buzzatii raised in common-garden conditions and using an isofemale line design. The between-population divergence shows that sexual dimorphism is greater when sex evolves in different directions. Multivariate Qst-Fst analyses confirm that male wing shape is the target for multiple selective pressures, leading males' wings to diverge more than females' wings. While the wing blade and the wing base appear to be valid modules at the genetic (G matrix) and among-population (D matrix) levels, the reconstruction of between-population adaptive landscapes (Ω matrix) shows selection as an integrative force. Also, cross-sex covariances reduced the predicted response to selection in the direction of the extant sexual dimorphism, suggesting that selection had to be intensified in order to circumvent the limitations imposed by G. However, such intensity of selection was not able to break the modularity pattern of the wing. The results obtained here show that the evolution of D. buzzatii wing shape is the product of a complex interplay between ontogenetic constraints and conflicting sexual and natural selections.