Bias of animal population trend estimates

A computer simulation study of the population trend estimator used for the Mourning Dove Call-Count Survey, Woodcock Singing Ground Survey, Breeding Bird Survey and other surveys concluded that the estimator had negligible bias in most situations but that observer covariables should not be used with less than five years of data. With rare species (e.g. two birds per route), at least five years should be used. The estimator is seriously biased towards not detecting population changes with very rare species (e.g. 0.3 birds per route). Other technical recommendations are made.