Coping with variation in aerial survey protocol for line-transect sampling

Abstract

Altitude and flight speed affect detection probability and they typically vary during the course of most aerial surveys. We demonstrate how these sources of variation can be accommodated with covariates in a line-transect analysis using data from a pronghorn (Antilocapra americana) survey in Wyoming and a survey of Antarctic ice seals (Lobodon carcinophaga, Leptonychotes weddellii, Hydrurga leptonyx, Ommatophoca rossii). We also show how the likelihood for binned distance data can be modified to allow for variation in altitude. As an alternative, we develop an estimator for aerial line-transect sampling based on vertical angles rather than distance. With a small simulation study, we show that our estimators are unbiased and are preferable to using biased estimators based on fixed-distance intervals derived from average altitude.