|| Checking for direct PDF access through Ovid
A decline in species richness moving from equatorial regions to polar regions is a common, but not universal, macroecological pattern. Many studies have focused on this pattern, but few have focused on how the vital rates responsible for species richness patterns, local rates of species extinction and turnover, vary with latitude. We examine patterns of richness, turnover and extinction in North American avian communities inhabiting three ecoregions, using methods that account for failure to detect all species present. We use breeding bird point count data from > 1000 routes in the Breeding Bird Survey collected from 1982 to 2001 to estimate richness, extinction probability and turnover rates. Our analyses differ from others in 1) the use of annual estimates derived at specific locations rather than index data accumulated over numbers of years, 2) the use of estimators that incorporated detection probabilities and 3) a focus on dynamical processes (colonization, extinction) in addition to static patterns (species richness). We find average species richness estimates (48 to 135 species) increasing with latitude for all three regions, contradicting predictions based on the latitudinal diversity gradient. The estimated rates of extinction and turnover declined with latitude across the three ecoregions. We speculate that higher richness might be linked to periods of superabundant food supply in northern areas that support greater numbers of resident and migrant species. Our primary ecological conclusions are that the latitudinal gradient in species richness is reversed for North American birds in the studied ecoregions, and that both local extinction and turnover decrease from southern to northern latitudes. Thus, the vital rates that determine richness show evidence of greater stability and reduced dynamics in northern areas of higher richness. We recommend additional studies examining patterns of colonization, extinction and turnover in communities, that use clearly defined estimators that deal with detection probability.