Dissecting genetic correlation and pleiotropy through a genetic cross

Abstract

Genetic correlation represents an important class of evolutionary constraint, which is itself evolvable. Empirical studies have found mixed results on whether genetic correlations change rapidly or slowly. This uncertainty challenges our ability to predict the outcome of selection. Despite the tremendous diversity and complexity of life forms, there are certain forms of life that are never observed. This might be because of developmental biases that restrict how organisms can evolve, or because they have low fitness in any environment yet available on Earth. Given that both developmental bias and selection can generate similar phenotypes, it is difficult to distinguish between the two causes of evolutionary stasis among related taxa. For example, remarkably invariant traits are observed spanning million years, such as wing shape in Drosophila wherein qualitative differences are rare within genera. Here, we ask whether the absence of certain combinations of traits, as indicated by genetic correlation, reflects developmental bias. However, much confusion and controversy remain over definitions of developmental bias, and probing it is challenging. We thus present a novel approach aiming to dissect genetic correlations and estimate the relative contribution of developmental bias in maintaining genetic correlations. We do so by leveraging a common but under-utilized type of data: genetic crosses. Through empirical analyses, we find that our approach can distinguish whether genetically correlated traits are developmentally constrained to covary. We also find that our developmental bias metric is an indicator of genetic correlation stability across conditions. Our framework presents a feasible way to dissect the mechanisms underlying genetic correlation and pleiotropy.