That's what Andrea Taylor and Carel van Schaik conclude in this paper:
Setting aside P. p. pygmaeus, for which data on behavioral ecology and life history are lacking, the trend revealed in this study is for female P. p. morio to have the smallest brains and for P. abelii to have the largest brains. This contrast coincides with previously documented differences in resource availability and diet quality (MacKinnon, 1974, Rodman, 1988, Leighton, 1993, Knott, 1998 and Wich et al., in press). Longer droughts, lower soil productivity, and more mast-fruiting species converge to produce an environment for orangutans of eastern Borneo that is at times seriously resource-limited. In northern Sumatra, where the environment is more productive and the effects of El Niño less pronounced, mean productivity of ripe fruit is higher and variance is lower (Wich et al., in press). Indeed, Sumatran orangutan densities, which are two times as high as in similar habitats on Borneo (Delgado and van Schaik, 2000 and Marshall et al., 2006), have been linked to the greater productivity of Sumatran forests.
How to test?
The general scenario supported by these results, then, is that an increase in the frequency of uncontrollable periods of low energy intake in one part of the orangutan's geographic range selected for a reduction in brain size, thereby ameliorating the effects of inadequate energy intake on mortality. Reduced brain size in P. p. morio, through its developmental effects, would have shortened the period of development and increased adult reproductive rates. This scenario predicts that P. p. morio also has an earlier onset of maturity and shorter maximum lifespan than the other Pongo taxa, a hypothesis that remains to be tested in the wild.
Well, not exactly, since a relatively faster development might still be slowed in the resource-stress environment. The real test would be to compare the two subspecies in captivity where they presumably have similar (and sufficient) diets.
The way the paper ends is provocative, at least to me, and worth some thought.
Taylor and van Schaik finish with a generalization that low or variable resource availability may select for smaller brain size, and mention that this may include selection for profoundly smaller brain size in Homo on Flores. That's a simple, straightforward interpretation of their result, and it ought to be maddening for some.
Consider: variable resource availability and climatic instability have been suggested (by Rick Potts, among others) as factors underlying the increase in brain size in hominids! Something doesn't click here.
I suppose one might resolve this paradox by suggesting that a maintenance of high diet quality under resource stress requires a larger (i.e. smarter) brain, while a sacrifice of diet quality with dependence on low-energy fallback foods selects for a smaller (i.e., lower energy cost) brain.
Or, one might suggest that hominids increased in brain size under resource stress because they increased their reliance on social learning, while orangutans decreased in brain size under resource stress because they abandoned their social organization. Certainly, the stereotypical solitary orangs are the low-population-density orangs on Borneo with the periodic shortfalls. But that's hardly satisfactory -- why should orangutans abandon social groupings at low population densities, while hominids strengthened them?
Saying that the hominid adaptation was just different from the orang adaptation is begging the question.
To my mind, any hypothesis of difference between the hominids and orangs here is ad hoc. True, they eat different diets, have different locomotor patterns, and live in different habitats. But explain why that matters! If the important variable is energy, and less energy constrains brain size, then energy shortfalls for hominids should have constrained their brain size.
Keep in mind, we have no idea what the ancestral state for orangutans might be. Did the small-brained subspecies shrink, or did all the other orangs grow? Did encroachment by humans (or some other continental species) over the last few hundreds (or thousands) of years make a difference? Is a sample size of < 20 really sufficient?
Taylor AB, van Schaik CP. 2007. Variation in brain size and ecology in Pongo. J Hum Evol 52:59-71. doi:10.1016/j.jhevol.2006.07.010