R coding for insect flight muscle tracheole scaling

The statistical code was executed within R. Mean tracheolar volume density and body mass data were log10-transformed and their relationship analysed by an ordinary least-squares (OLS) regression model, and then by a phylogenetic generalized least-squares (PGLS) regression model using the caper v1.0.1 package within R v4.5.1. The phylogeny used for the PGLS regression was built from a published species-level insect supertree, with missing species substituted for those in the same genus (in the first instance) or family (in the second instance). A measure of phylogenetic correlation, Pagel’s λ, was estimated by fitting PGLS regressions with different values of λ and identifying by log likelihood that which maximized the model fit. We performed Hartigans' dip test of unimodality using the diptest v0.76-0 package in R to assess unimodality and potential clustering in the residuals of the PGLS regression. To account for the potential effect of different types of flight muscle (synchronous or asynchronous) on the relationship between mean tracheolar volume density and body mass, the PGLS regression was also run with this trait as an additional predictor. The candidate models were compared using the Akaike Information Criterion (AIC).