Application of Cohort Analysis to Large Terrestrial Mammal Harvest Data

Abstract

Cohort analysis (also known as virtual population analysis) is a method of population reconstruction from age-specific harvest data. Because cohort analysis requires data over a whole life span to reconstruct a population for a single year, this method is impracticable for longer-lived animals. Three models are routinely combined by fisheries scientists to make cohort analysis more cost effective and to provide real-time estimates of population size; these models may be applied to large terrestrial mammal harvest data. Each model has unique assumptions about hunting mortality rates or age distributions, and the reliability of estimates depends on meeting these assumptions. In this study, we first tested previously used assumptions for these models through an analysis of long-term moose (Alces alces) harvest data, followed by an examination of the robustness of estimates for each moose age class. We developed practical ways to achieve more realistic assumptions for two of three models and showed that meeting these assumptions was more important in estimations of large terrestrial mammal population parameters than for fish population parameters. Therefore, we recommend compliance with assumptions of the three models for more reliable population estimates of large terrestrial mammals.