Two alternative hypotheses have been advanced to explain the demise of about half of the mammalian genera exceeding 5 kg in body mass in the later Pleistocene. One hypothesis invokes climatic change and resulting habitat transformations. This fails to predict the increased likelihood of extinctions with increasing body size, greater severity in both North and South America than in Eurasia or Australia, lack of simultaneous extinctions in Africa and tropical Asia, and the absence of extinctions at the end of previous glacial periods. The other hypothesis invokes human predation as the primary cause. This fails to explain the simultaneous extinctions of a number of mammalian and avian species not obviously vulnerable to human overkill. I propose a “keystone herbivore” hypothesis, based on the ecology of extant African species of megaherbivore, (i.e., animals exceeding 1,000 kg in body mass). Due to their invulnerability to non-human predation on adults, these species attain saturation densities at which they may radically transform vegetation structure and composition. African elephant can change closed woodland or thicket into open grassy savanna, and create open gaps colonized by rapidly-regenerating trees in forests. Grazing white rhinoceros and hippopotamus transform tall grasslands into lawns of more nutritious short grasses. The elimination of megaherbivores elsewhere in the world by human hunters at the end of the Pleistocene would have promoted reverse changes in vegetation. The conversion of the open parklike woodlands and mosaic grasslands typical of much of North America during the Pleistocene to the more uniform forests and prairie grasslands we find today could be a consequence. Such habitat changes would have been detrimental to the distribution and abundance of smaller herbivores dependent upon the nutrient-rich and spatially diverse vegetation created by megaherbivore impact. At the same time these species would have become more vulnerable to human predation. The elimination of megaherbivore influence is the major factor differentiating habitat changes at the end of the terminal Pleistocene glaciation from those occurring at previous glacial-interglacial transitions.
