Extracting Long-Term Patterns of Population Changes from Sporadic Counts of Migrant Birds

Declines of many North American birds are of conservation concern. Monitoring their population changes has largely depended on formally structured Breeding Bird Surveys, and Migration Monitoring Stations, although some use has been made of lists by birders. For almost 40 years, birders have kept daily counts of migrant landbirds during visits to Seal Island, of Nova Scotia's south tip. Here we present results for several common migrants using day-counts made between August 15 and November 15. Most existing analyses have used linear models to extract trends and other variables from such long-term data sets. Instead we applied Generalized Additive Models (GAMs) to extract the continuous trend functions and patterns of influence of observer number, wind speed, wind direction on count nights and prior nights, and moon phase. The results suggest that GAMs are a powerful way of dealing with such "noisy" data of the sort collected by birders in their recreational pursuits. In addition, it is possible to analyse groups of species (related taxonomically or ecologically) simultaneously with the potential of determining overall more general trends.