Breeding bird communities in spruce—fir forests of the southern Appalachian mountains are considerably less diverse than those of similar northern forests. Censuses also show more widespread coexistence of congeneric and confamilial species in the north. Long—distance migrants dominate the northern forests while resident species dominate in the south. In a 4—yr study of these bird communities, I used observations of foraging strategies and resource partitioning to test hypotheses that might explain this probematic latitudinal gradient in species diversity. I conducted quantitative studies of foraging behavior of arboreal insectivorous birds at sites with very similar plant communities in the Great Smoky Mountains of North Carolina and in northwestern Maine. Six species that breed in both north and south had consistently more generalized foraging in the south, as measured by spatial distribution of foraging, variety of techniques used, directionality in probing and flights, and patchiness of habitat use. These north—south differences in foraging behavior result in part from greater behavioral diversity of each individual. Five exclusively northern species showed greater foraging generalization at the northern site than their sympatric congeners that breed in the south as well. The 11 species in the north occupied a smaller total foraging niche volume than the 6 species in the south. Each species overlapped more broadly with others in the north, and maximum overlap was greater. The southern avifauna is a nearly totally included subset of the northern. Northern diversity is produced by the addition of very similar (congeneric), widely overlapping generalists. This "stacking" of generalists differs considerably from tropical patterns of community diversification. Ecological, biogeographic, and evolutionary hypotheses explaining tropical diversity and island diversity are inadequate to account for the reversal of this latitudinal gradient and the species—packing characteristics of northern diversification. Populations in the depauperate south apparently released from competition with exclusively northern congeners did not show stronger generalization than populations of species that are the sole representatives of their families north and south, contrary to expectation. This is better explained by a common response, predicted by foraging optimization theory, to lower food levels in the south than by ecological release. Seasonal compression of productivity in northern spruce—fir forests, suggested by geophysical and climatological data, produces stronger pulses of food production than in the south, as shown by arthropod sampling. Compressed productivity and winter limitation of both northern residents and migrants probably produce higher food levels relative to consumer biomass in the north. Seasonal diversity is probably enhanced in northern forests by strong, predictable resource oscillations that are difficult for consumers to track Cyclic relaxation of limits to ecological similarity through seasonally abundant resources creates a higher diversity peak than under weaker seasonality. In the more seasonally constant south, residents are probably better able to track resources, and dominance by these groups is enhanced. Predictable seasonal peaks of production have probably promoted the evolution of opportunistic migrants and reduced their competition with winter—limited residents. I discuss this hypothesis in light of a theory proposing that variable conditions may often lead to greater diversity than more stable conditions. This study and others indicate the importance of seasonal species and temporal complexity in natural communities.