Estimation of population abundance from samples has inherent practical challenges. Moreover, analytical methods to estimate abundance may vary in statistical assumptions and prediction uncertainties. I evaluated the performance of design-based and model-based methods to estimate Canada geese (Branta canadensis) abundance based on aerial fixed-width transect surveys in the Hudson Bay Lowlands, Canada. I evaluated Empirical Bayesian Kriging (EBK), areal interpolation and a ratio estimator on the basis of accuracy and precision using spatial point simulations. Untransformed EBK was the most accurate and precise, due in part, to its inherent handling of nonstationary distributions. The ratio estimator followed the same trends as EBK and, in some cases, had higher precision. Consideration of alternative analytical methods and their strengths and weaknesses is an important step in generating reliable information for population monitoring. Geostatistical approaches such as EBK have the benefit of providing spatially explicit mapping of abundance and reliable population estimates.
Author Keywords: Areal interpolation, Design-based inference, Empirical Bayesian Kriging, Geostatistics, Model-based inference, Ratio estimator