Identification of multivariate phenotypes most influenced by mutation: Drosophila serrata wings as a case study

The distribution of mutational effects, especially pleiotropic effects, significantly impacts phenotypic evolution and population persistence. However, our empirical understanding of these effects remains quite limited. Here, we interrogated multivariate wing shape data from 12 replicated samples of the same mutation accumulation experiment in Drosophila serrata to characterize both the variation in wing shape introduced by mutations, and the among-sample heterogeneity in estimates of that variation. Consistent with previous studies and expectations from statistical theory, there was considerable heterogeneity among replicate estimates of the same parameter (median ratio of largest to smallest estimate of mutational variance per trait: 7.6; estimates of the same pairwise trait correlation ranged up to a maximum of the theoretical limits: −1 to 1). However, multivariate analyses identified a trait combination associated with relatively similar magnitude of mutational variance in all population samples (ratio of largest to smallest estimate: 1.96). This major axis of mutational variance aligned closely with the major axis of among-line variance within most of the population samples, suggesting that each sample adequately estimated the predominant pattern of correlated mutational effects, despite the marked heterogeneity of individual (co)variance parameters. Mutation predominantly introduced variation along an axis of shorter, wider vs. longer, narrower wing tips. This was qualitatively the same mutational wing shape variance identified in independent studies of Drosophila wings, using different experimental designs and analytical approaches. Our study suggests that while estimation error can independently influence variances of individual traits and covariances among traits, multivariate analyses may be effective at revealing clear signals higher dimensional patterns of variation.