I'm reading through the volume Integrative Paths to the Past (Corruccini and Ciochon, eds.) because of a piece of work I've been doing, and I came across this interesting passage in the contribution by Elizabeth Vrba, titled "An hypothesis of heterochrony in response to climatic cooling and its relevance to early hominid evolution."
Conversely, I suggest that acceleration and hypomorphism often evolve in warmer environments. In fact, the celebrated correlation of dwarfing of mammals on islands may well have less to do with the absence of predators and resource depletion (e.g., Lomolino 1985) than with the fact that island refugia for large mammals come into being in times of global warming and sea level rise--namely, maximal warming periods--and islands at all times enjoy a more mesic climate than the mainland uplands of the ancestors. Prothero and Sereno's (1982) results for North American fossil rhinoceroses is relevant: Dwarf species were associated with mesic forest-swamp "climatic islands" on the Miocene land mass, surrounded by savanna uplands on which larger rhinoceros taxa lived (Vrba 1994:355-356).
In other words, Vrba says that the reason we find dwarf elephants, hippopotamus, and other large mammals on the Mediterranean islands during the past few million years may be climatic. I'm not sure this accounts for the presence of dwarf mammoths on Wrangell Island, or the Catalina Islands for that matter, since full-size mammoths were on the adjacent mainland. On the other hand, migration is a factor affecting the evolution of continental taxa that simply isn't an issue for species trapped on islands. So there may be a combination of climate and the necessity for long-distance movement that makes sense.
In any event, this kind of explanation appears to be closer to the truth for phyletic dwarfs in continental regions than other explanations. Consider the Pygmies of West Africa: their small body size has been variously interpreted as a result of nutritional restrictions, inability to thermoregulate efficiently in the humid atmosphere, the need to maintain small mass for effective climbing, or sexual selection. It is not obvious that any of these explanations are applicable to other human populations with small body sizes, such the Negritos from Southeast Asia. There is obviously much thinking to do here, but I'm not sure that we have a very good explanation for dwarfism in human populations. Vrba's remarks lead to believe that we don't have a very good explanation for dwarfism in mammal populations in general.
Vrba ES. 1994. An hypothesis of heterochrony in response to climatic cooling and its relevance to early hominid evolution. In Corruccini RS and Ciochon RL, eds., Integrative paths to the past: paleoanthropological advances in honor of F. Clark Howell. Prentice Hall, Englewood Cliffs, NJ. pp. 345-376.