Accuracy of some aerial survey estimators: contrasts with known numbers

Abstract

Density estimates are seldom examined against actual population size, hence the ability of estimators to correct for bias is unknown. Studies that compare techniques are difficult to interpret because of the uncertainty of adherence to their respective assumptions. Factors influencing detection probability, estimators that correct for bias, the validity of their assumptions and how these relate to true density are important considerations for selecting suitable methods. Here we contrasted five estimates of feral goat (Capra hircus) densities obtained from aerial surveys (strip counts, Petersen, stratified Petersen, Chao, Alho) against known densities derived from total counts. After correcting for recounting, the Alho and stratified Petersen estimators applied to helicopter surveys were the most accurate (bias = 0.08 and –0.09 respectively), which suggests that estimates were improved by correcting individual observations according to the characteristics of each observation. An approach using modified Horvitz–Thompson equations for unequal-sized units is described and is recommended to allow for this. Both the Chao (bias = 0.35) and Petersen (bias = 0.22) estimators were positively biased, which is likely to be a consequence of averaging detection probability across all observations. Helicopter survey using capture–recapture with multiple observers is recommended for estimating the density of wildlife populations. However, adjustment for the factors that influence detection probability is required.