Ecology not genetics explains correlated trait divergence during speciation

This archive comprises the scripts and codes used in the manuscript "Ecology not genetics explains correlated trait divergence during speciation" by de Carvalho et al.

The abstract of the  manuscript follows:

The formation of new species often involves the correlated divergence of multiple traits and genetic regions. However, the mechanisms by which such trait covariation builds up remain poorly understood. In this context, we consider two non-exclusive hypotheses. First, genetic covariance between traits can cause divergent selection on one trait to promote population divergence in correlated traits (a genetic covariation hypothesis). Second, correlated environmental pressures can generate selection on multiple traits, facilitating the evolution of trait complexes (an environmental covariation hypothesis). Here, we test these hypotheses using cryptic coloration (controlled by an incipient supergene) and chemical traits (i.e., cuticular hydrocarbons, CHCs) involved in desiccation resistance and mate choice in Timema cristinae stick insects. We first demonstrate that population divergence in color-pattern is correlated with divergence in some (but not all) CHC traits. We show that when correlated population divergence does occur, it is unlikely to be explained by genetic covariation because within-population genetic covariance between color-pattern and CHCs traits is weak. In contrast, we find that correlated variation in climate and host plant likely generates selection jointly on color-pattern and some CHC traits. This supports the environmental covariation hypothesis, likely via the effects of two correlated environmental axes selecting on different traits. Finally, we provide evidence that misalignment between natural and sexual selection also contributes to patterns of correlated trait divergence. Our results shed light into transitions between phases of speciation by showing that environmental factors can promote population divergence in trait complexes, even without strong genetic covariance.

The methods to obtain the data follow:

We used previously published data from Riesch et al. (2017) to estimate divergence in color-pattern and in different female CHC classes. For color-pattern, we computed the Euclidean distances from percentage of striped individuals. For CHCs, we used data of females from the same populations. We conducted principal component analyses (PCA) separately for each CHC class, following the methods described in Riesch et al. (2017), then estimated the pairwise Euclidean distances between populations. We used this data to estimate correlations. 

We estimated the genetic covariance between color-pattern and each of the CHCs classes using the results from genome-wide association studies (GWA) from Nosil et al. (2024). We performed new GWA for each female CHC trait. We extracted the breeding values for each trait, and then estimated Pearson correlations among them.

We estimated correlations between environmental variables affecting color-pattern and CHCs using data from Nosil et al. (2018). Finally, we estimated the regression relationship between divergence in CHC traits and the pairwise index of sexual isolation (Ipsi) between the studied populations. The Ipsi data was used from Riesch et al. (2017). We fitted Bayesian linear mixed models to estimate the degree of association between these factors. All the statistical analyses were performed in R v4.3.2.