Species extinctions typically occur following dramatic population declines; detecting these declines is therefore crucial for biodiversity conservation. Attempts to quantify local and global population declines are complicated by the fact that abundance at a single time period can be defined in terms of total abundance, local density, and occupancy. Traditional approaches to distribution-abundance patterns rely on plotting mean values for abundance against occupancy. However, mean values obscure variation in abundance across occupied sites. We used a new method called Rank Occupancy Abundance Profiles (ROAPs) to estimate declines in North American bird species. Using data from the long-term, large-scale North American Breeding Bird Survey, we quantify the magnitude of declines for 213 bird species. ROAPs allow us to partition losses in regional abundance according to the proportion of individuals lost in populations of different densities, as well as the proportion of the regional abundance lost due to local extinctions.

Results/Conclusions

Nearly 20% of bird species surveyed showed statistically significantly declines in regional abundance between 1970 and 2005. The 20 bird species to experience the highest proportional losses showed declines between 48% and 89% of their initial regional population sizes. Patterns of decline varied according to initial regional abundance of the species: very abundant species experienced large proportional drops in local density, while species with low regional population sizes experienced proportionally higher losses due to local extinctions. Based on patterns we observed across species, as well as those observed in time-series data within species, we propose a new model depicting phases of decline a single species experiences as it nears extinction. We suggest that two time lags exist as a species approaches rarity: 1) species lose individuals from intermediate-density sites prior to losing individuals from sites with the highest densities, and 2) species lose individuals from the highest density sites before local extinctions ensue (i.e. declines in occupancy). We also detected threshold regional abundances below which local populations can no longer persist. Our results emphasize the interplay between local and regional abundance as species decline.  Early detection of declines could alert conservationists before the largest populations diminish. Further, monitoring schemes that record presence-absence data only will not capture declines until local extinctions occur—which we propose is at a late stage of decline.