Distance sampling with independent observers: Reducing bias from heterogeneity by weakening the conditional independence assumption

Using replicated surveys of a known population of wooden stakes in a sagebr ush habitat, I illustrate the application of surveys which combine sight-re sight with line transect sampling. This example demonstrates that the usual assumption of full conditional independence (FCI) for all distances (x gre ater than or equal to 0) is unnecessarily strict and can be weakened to ass uming conditional independence on the trackline (x = 0). I present a likeli hood for combined sight-resight and distance sampling based on trackline co nditional independence (TCI) and use a likelihood ratio test for comparison of the TCI and FCI models. For the wooden stake surveys, in most cases the TCI model was superior to the FCI model because of heterogeneity in detect ion probability created by variation in vegetation density for stakes at x > 0. Equivalent factors in marine mammal surveys, such as sea state and clo ud cover, will create heterogeneity in detection probability that will bias abundance estimates developed with combined distance and sight-resight sur veys under the FCI assumption.