Assessment of Bird Band Recovery Data
Approximately 1.2 million birds are banded in North America annually and around 85,000 encounters/recoveries of banded birds are reported each year. These data provide valuable information on survival rates and migration patterns for species that are routinely recovered. Banding data can help to identify population concentration areas and areas where humans interact with birds (e.g. hunting and monitoring).
We analyzed bird band recovery data from 1960 through 2010 to identify migration corridors that could be important areas to avoid for wind tower development. We recognize that inferences from these data should be made cautiously since hotspots are, to varying degrees, also indicative of human activity. Data were obtained through the USGS Bird Banding Laboratory for Waterfowl and Non-game bird species. Recovery locations were mapped by creating point locations in ArcMap 9.3 using longitude and latitude from recovered birds. Points were assigned to 10 minute grid cells covering the lower 48 states to determine frequency of point occurrences in grid cells. Lastly, Hot Spot Analysis (Getis-Ord Gi Statistic) was used to create maps by calculating a Z score.
Analysis of waterfowl and non-game birds were separated due to the difference in number and frequency of recoveries, dispersal and movement patterns, landscapes preferences, and management strategies. Waterfowl comprised 80% of all recoveries. Non-game species were combined to increase number of observations. Non-game records were made up of 10% raptors, 34% shore and waterbirds, and 56% passerines and other birds.
Inclusion of banding station data results in significant recovery spikes in close proximity to banding stations. To eliminate bias, observations were removed from the dataset that consisted of bander submissions, recapture, and of juvenile waterfowl reports. To increase sample size of non-game bird reports, juvenile reports were kept in the dataset.
Migration patterns are easily detected in the waterfowl map and difficult to determine in the non-game map. Map differences are likely a result of variations in sample size, human actions leading to band encounters, and bird behavior. Band reports from hunting efforts produce large numbers of band returns, which enables researchers to learn rapidly and significantly more about waterfowl than other species.
Banding data fails to capture observations of birds that migrate quickly, use few stop over points, use limiting habitats, and travel through areas with a lower probability of encounters with humans and those where there is little banding effort taking place.
Despite the need for information on migration corridors of birds prone to strikes on wind towers and other development, banding data does not sufficiently represent migration corridors of nongame birds and should not be used, at least independently, to target conservation actions.