Improved Removal-Based Abundance Estimates By Accounting For Fish Behavior

Fish abundance or density is often one of the key variables for basing management decisions or evaluating research results. The removal method is one of the most common statistical tools for estimating abundance in streams and lakes. However, many authors have found removal estimates to be biased, which could lead to misleading results of a scientific study or inappropriate management actions. Two key assumptions in the removal method are constant catchability over samples and a closed population (no immigration or emigration). However, fine-scale fish behaviours make these two assumptions unlikely. Possible behaviors include: territoriality which may preclude some fish from being available to be captured; fish being behaviorally inactive at some times; density dependent catchability; or catchability varying across individuals. We explore various models to explain these behaviors and examine the bias in abundance that may occur due to incorrectly assuming constant catchability and abundance. We present the most robust model and fit to a typical dataset exhibiting changes in mean catchability. We show that the model that correctly characterizes dominant fish behaviour patterns can be unbiased, resulting in proper assessment of abundance and reliable metrics for basing management decisions.