Visibility bias in aerial survey: mark–recapture, line-transect or both?

Abstract

Mark–recapture and line-transect sampling procedures both provide estimators for visibility bias in aerial surveys, and have coexisted in the literature for decades. Mark–recapture estimators of abundance tend to be negatively biased in this context as a result of unmodelled heterogeneity. Line-transect sampling can also be negatively biased if detection probability on the line is less than 1.0. Numerous papers have described hybrid approaches using mark–recapture and line transect methods but there have been some subtle but important differences that may not be apparent to the practitioner. We have used wild horse survey data collected in south-eastern Australia and some imaginary data to highlight these subtle differences. We demonstrate the advantage of using the hybrid approach, which uses the strengths of both mark–recapture and line-transect procedures by fitting a detection function (with p(0) = 1) to the line-transect data to estimate the shape of the detection function, and uses a separate detection function for the mark–recapture data to estimate the intercept (p(0)).