What is the evidence for the Keystone Herbivore Hypothesis?
The hypothesis relies on modern observational evidence in places where megaherbivores still remain. It has a very strong co-evolution component in that it maintains that the success of open-country foragers was very tightly linked to the presence of significant numbers of environment-modifying megaherbivores.
One of the very interesting aspects of the keystone herbivore hypothesis is Owen-Smith's recognition of the different ecological roles of herbivorous species. Although he sees merit in the idea that the large proboscideans and a few other species could have been hunted to extinction, he strongly disagrees that smaller, fleeter, faster-reproducing species could have been brought low by overhunting alone. His hypothesis offers a way to view their extinction as a consequence of other losses, not as a sole consequence of human predation.
Another interesting aspect of his argument is that it supplies a reason for the survival of some large species. A strong explanation of the megafaunal extinctions has to account not only for those species that disappeared, but also ones that survived that were (according to size, at least) not unlike those that disappeared. Moose, elk, and deer--which make up the bulk of the dozen or so surviving megafauna of modern North America--are closed-woodland browsers. These are not the kinds of species that should have come to grief when the megaherbivores became extinct, and the fact that this group has survived, at least in part, adds to the plausibility of this argument.
With respect to Africa and Asia, Owen-Smith is sympathetic to Martin's behavioral co-evolution argument: because of the continuing co-evolution of potential predator and prey, no abrupt episode of extinction occurred until very recently. He also notes that the distinctive basin-and-swell topography of Africa on a continental scale insures environmental heterogeneity, insulating species from deleterious impacts.