Investigating Observer Bias in Aerial Survey by Simultaneous Double-Counts

Publisher Website
Google Scholar
Abstract
We adjusted aerial counts of feral horses (Equus caballus) and donkeys (E. asinus) in the Northern Territory of Australia for observer bias in the visible population by an adaptation of the Peterson estimate (Seber 1982:59) based on visual captures. We developed a method of simultaneous double-counts using tandem observers counting groups of animals in the same transect. Observers overlooked 12.8-43.8% of the groups of visible animals in their field of view. The mean probability of seeing a solitary animal was 0.66 and 0.91 for a group of 8. Observed group size increased with height, interpreted as a declining probability of seeing small groups with increasing height. The chances of seeing a group appear to be influenced more by the number of its members than member size. We compared the method with the logically equivalent model of Cook and Jacobson (1979). Both are simple to apply with potential for investigating the cause of observer bias or as practical alternatives to conventional aerial survey methods.