Accuracy of the Multiple-Regression Method for Estimating Population-Density in Strip Transects

Abstract

Two experiments examined the accuracy of the multiple regression method for estimating population density. In experiment I , in a balanced design, an observer counted objectsin simulated strip transects. Multiple regression analyses yielded accurate estimates when true density was low, and overestimated density when true density was high. Regression equations calculated at each level of true density varied from linear to quadratic. A simple polynomial model accurately estimated true density. In experiment 2 an aerial survey of sheep showed that estimated density was highly significantly (P <0.001) different from true density. The results suggest that greater use should be made of established criteria for robust estimation of true density in transect studies.