Statistical aspects of point count sampling

Abstract

The dominant feature of point counts is that they do not census birds, but instead provide incomplete counts of individuals present within a survey plot. Considering a simple model for point count sampling, we demon-strate that use of these incomplete counts can bias estimators and testing procedures, leading to inappropriate conclusions. A large portion of the variability in point counts is caused by the incomplete counting, and this within-count variation can be confounded with ecologically meaningful varia-tion. We recommend caution in the analysis of estimates obtained from point counts. Using; our model, we also consider optimal allocation of sampling effort. The critical step in the optimization process is in determining the goals of the study and methods that will be used to meet these goals. By explicitly defining the constraints on sampling and by estimating the relationship between precision and bias of estimators and time spent counting, we can predict the optimal time at a point for each of several monitoring goals. In general, time spent at a point will differ depending on the goals of the study.