Abstract
The recent influx of genomic data has provided greater insights into the molecular basis for regressive evolution, or vestigialization, through gene loss and pseudogenization. As such, the analysis of gene degradation patterns has the potential to provide insights into the evolutionary history of regressed anatomical traits. We specifically applied these principles to the xenarthran radiation (anteaters, sloths, armadillos), which is characterized by taxa with a gradation in regressed dental phenotypes. Whether the pattern among extant xenarthrans is due to an ancient and gradual decay of dental morphology or occurred repeatedly in parallel is unknown. We tested these competing hypotheses by examining 11 core dental genes in most living species of Xenarthra, characterizing shared inactivating mutations and patterns of relaxed selection during their radiation. Here we report evidence of independent and distinct events of dental gene loss in the major xenarthran subclades. First, we found strong evidence of complete enamel loss in the common ancestor of sloths and anteaters, suggested by the inactivation of five enamel-associated genes (AMELX, AMTN, MMP20, ENAM, ACP4). Next, whereas dental regression appears to have halted in sloths, presumably a critical event that ultimately permitted adaptation to an herbivorous lifestyle, anteaters continued losing genes on the path towards complete tooth loss. Echoes of this event are recorded in the genomes of all living anteaters, being marked by a 2-bp deletion in a gene critical for dentinogenesis (DSPP) and a putative shared 1-bp insertion in a gene linked to tooth retention (ODAPH). By contrast, in the two major armadillo clades, genes pertaining to the dento-gingival junction and amelogenesis appear to have been independently inactivated prior to losing all or some enamel. These genomic data provide evidence for multiple pathways and rates of anatomical regression, and underscore the utility of using pseudogenes to reconstruct evolutionary history when fossils are sparse.