Extant sloths present an evolutionary conundrum in that the two living genera are superficially similar (small-bodied, folivorous, arboreal) but diverged from one another approximately 30 million years ago and are phylogenetically separated by a radiation of medium to massive, mainly ground-dwelling, taxa. Indeed, the species in the two living genera are among the smallest, and perhaps most unusual, of the 50+ known sloth species, and must have independently and convergently evolved small size and arboreality. In order to accurately reconstruct sloth evolution, it is critical to incorporate their extinct diversity in analyses. Here, we used a dataset of 57 species of living and fossil sloths to examine changes in body mass mean and variance through their evolution, employing a general time-variable model that allows for analysis of evolutionary trends in continuous characters within clades lacking fully-resolved phylogenies, such as sloths.

Our analyses supported eight models, all of which partition sloths into multiple subgroups, suggesting distinct modes of body size evolution among the major sloth lineages. Model-averaged parameter values supported trended walks in most clades, with estimated rates of body mass change ranging as high as 126 kg/million years for the giant ground sloth clades Megatheriidae and Nothrotheriidae. Inclusion of living sloth species in the analyses weakened reconstructed rates for their respective groups, with estimated rates for Megalonychidae (large to giant ground sloths and the extant two-toed sloth) were four times higher when the extant genus Choloepus was excluded.

Analyses based on extant taxa alone have the potential to oversimplify or misidentify macroevolutionary patterns. This study demonstrates the impact that integration of data from the fossil record can have on reconstructions of character evolution and establishes that body size evolution in sloths was complex, but dominated by trended walks towards the enormous sizes exhibited in some recently extinct forms.