Abstract
Bird counts by community volunteers provide valuable information about the conservation needs of many bird species. The statistical modeling techniques commonly used to analyze these counts provide robust, long-term population trend estimates from heterogeneous community science data at regional, national, and continental scales. Here, we present a new modeling approach that increases the spatial resolution of trend estimates and reduces the computational burden of trend estimation, each by an order of magnitude. We demonstrate the approach with data for the American Robin (Turdus migratorius) from Audubon Christmas Bird Counts conducted between 1966 and 2017. We show that aggregate regional trend estimates from the proposed method aligned well with those from the current standard method, and that spatial variation in trends was associated with winter temperatures and human population densities as predicted by ecological energetics. This technique can provide reasonable large-scale trend estimates for users interested in general patterns, while also providing higher-resolution estimates for examining correlates of abundance trends at finer spatial scales, which is a prerequisite for tailoring management plans to local conditions.