Disentangling a genome-wide mosaic of conflicting phylogenetic signals in Western Rattlesnakes.

Abstract

Species tree inference is often assumed to be more accurate as datasets increase in size, with whole genomes representing the best-case-scenario for estimating a single, most-likely speciation history with high confidence. However, genomes may harbor a complex mixture of evolutionary histories among loci, which amplifies the opportunity for model misspecification and impacts phylogenetic inference. Accordingly, multiple distinct and well-supported phylogenetic trees are often recovered from genome-scale data, and approaches for biologically interpreting these distinct signatures are a major challenge for evolutionary biology in the age of genomics. Here, we analyze 32 whole genomes of nine taxa and two outgroups from the Western Rattlesnake species complex. Using concordance factors, topology weighting, and concatenated and species tree analyses with a chromosome-level reference genome, we characterize the distribution of phylogenetic signal across the genomic landscape. We find that concatenated and species tree analyses of autosomes, the Z (sex) chromosome, and mitochondrial genome yield distinct, yet strongly supported phylogenies. Analyses of site-specific likelihoods show additional patterns consistent with rampant model misspecification, a likely consequence of several evolutionary processes. Together, our results suggest that a combination of historic and recent introgression, along with natural selection, recombination rate variation, and cytonuclear co-evolution of nuclear-encoded mitochondrial genes, underlie genome-wide variation in phylogenetic signal. Our results highlight both the power and complexity of interpreting whole genomes in a phylogenetic context and illustrate how patterns of phylogenetic discordance can reveal the impacts of different evolutionary processes that contribute to genome-wide variation in phylogenetic signal.