The ripest fruit and the youngest leaves grow at the ends of branches, which is exactly where a branch stops being able to hold you. For a 40-gram mouse lemur this is not a problem. For a 40-kilogram orangutan it is the central problem of daily life. You need those calories, in part because your brain is expensive, and the only way to get them is to distribute your weight across whatever the canopy offers while your hands stay busy holding on. A fall from that height kills a large-bodied primate. The forensic case for Australopithecus afarensis AL 288-1, the specimen known as Lucy, is that this is precisely how she died.
Adriano Lameira’s new paper in Biology Letters1 argues that this predicament, and not the demands of social life, is what set the ceiling on how many different sounds a primate species can make.
That claim runs directly at a result that has been load-bearing for two decades. In 2005, Karen McComb and Stuart Semple published a phylogenetically controlled analysis in the same journal showing that evolutionary increases in vocal repertoire size among nonhuman primates tracked increases in both group size and time spent grooming. It fit neatly inside the social brain hypothesis, Robin Dunbar’s proposal that the cognitive load of tracking many relationships is what drove primate brain expansion. More relationships, more brain, more to say. Language, on this account, is what happens when a lineage gets social enough.
Lameira rebuilt the dataset. He surveyed the literature for new repertoire descriptions published since 2005, ending up with 86 studies covering 77 species, taking the maximum reported repertoire estimate where multiple studies described the same species. Body mass and habitat classification came from Estrada and Marshall’s 2024 review of ground use across the primate order, which sorts species as arboreal (less than 10% of time on the ground), mixed (10 to 50%), or terrestrial (more than 50%). Brain masses came from six published sources, with gaps filled by allometric regression on log body mass; a leave-one-out validation on species with known brain masses returned a cross-validated R² of 0.936. Group sizes were compiled from field accounts and databases, using midpoints of reported ranges and typical party size for fission-fusion species. He then ran phylogenetic generalized least squares models over a consensus tree from 10kTrees, with log repertoire size as the response.
Brain size predicted nothing (t = -0.51, p = 0.61). Group size predicted nothing either, and the coefficient leaned negative (t = -1.74, p = 0.086). What came out significant was the interaction between body size and habitat use (p = 0.027), and it was driven entirely by the arboreal species. Among primates living in the canopy, the slope was 0.352: a 10% increase in body mass corresponds to a 3.5% larger repertoire, and a doubling of body mass to a 27.6% increase. Across a group spanning three orders of magnitude in mass, that is an enormous amount of variation accounted for by weight alone. In mixed-habitat and terrestrial species the relationship was not detectable at all.
The residuals tell a second story. After accounting for brain, body, and group size, species in mixed habitats carried repertoires roughly 17% above expectation, arboreal species about 8% above, and terrestrial species about 6% below. (The paper’s abstract gives slightly different figures for the arboreal and terrestrial categories, +6% and -3%, though the rank order is the same in both places.) The ordering that emerges is mixed, then arboreal, then terrestrial. The species that move between two structurally different worlds talk the most.
One caution before the interpretation: brain and body mass are tightly correlated across primates, as they are across mammals generally, so the finding is not quite that brains are irrelevant. It is that when body mass and brain mass compete to explain repertoire size, body mass wins, and by a reasonable margin (ΔAIC ≈ 5 in favor of the model retaining body size).
