The Publications Warehouse does not have links to digital versions of this publication at this time
Most of the surveys presently used to estimate population trends on a large geographic scale depend upon repeated visits to a number of randomly selected routes or monitoring points. As these surveys cannot be analyzed by modeling annual mean densities among routes within a region, no natural annual index of population density exists for the region. We discuss two possible methodologies for estimating annual indices of abundance. In the context of the route-regression methodology, in which trends are estimated for each route and regional population trends are estimated as weighted averages of route trends, it is possible to find average residual distances between the predicted trends on each route and the actual data points. Adding these average residuals to the regional predicted values provides a measure of average distance from the actual data points to the predicted trends. A linear model approach can also be used to estimate annual indices, in which a regional slope parameter can be fit to the data in combination with annual effects. Bootstrapping can be used to provide some measure of the variability of these annual effects. These methods provide similar results in an example using Breeding Bird Survey data for scissor-tailed flycatcher (Tyrannlls forficatus) trends in Arkansas and Oklahoma.
Additional Publication Details
Federal Government Series
Estimation of annual indices from roadside surveys