Dissection of complex, fitness-related traits in multiple Drosophila mapping populations offers insight into the genetic control of stress resistance

We leverage two complementary Drosophila melanogaster mapping panels to genetically dissect starvation resistance, an important fitness trait. Using >1600 genotypes of the multiparental Drosophila Synthetic Population Resource (DSPR) we map numerous starvation stress QTL that collectively explain a substantial fraction of trait heritability. QTL effects further allowed us to estimate DSPR founder phenotypes, predictions that were correlated with the actual founder phenotypes. Starvation resistance has been linked to triglyceride level, and while we observe a modest phenotypic correlation between the traits in the DSPR, overlap among the QTL identified for each trait is low. Since we show that DSPR strains with extreme starvation phenotypes also differ in desiccation resistance and activity level, our data imply that multiple physiological mechanisms contribute to starvation variability. We also exploit the Drosophila Genetic Reference Panel (DGRP), identifying a number of sequence variants associated with starvation resistance. Consistent with prior work these sites rarely fall within QTL intervals mapped in the DSPR. Two other groups previously measured starvation resistance in the DGRP, offering a unique opportunity to directly compare mapping results across labs. We found strong phenotypic correlations among studies, but extremely low overlap in the sets of genomewide significant sites. Despite this, our analyses revealed that the most highly-associated variants from each study typically showed the same additive effect sign in independent studies, in contrast to otherwise equivalent sets of random variants. This consistency provides evidence for reproducible trait-associated sites in a widely-used mapping panel, and highlights the polygenic nature of starvation resistance.