Evaluating habitat selection with radio-telemetry triangulation error

Radio-telemetry triangulation errors result in the mislocation of animals and misclassification of habitat use. We present analytical methods that provide improved estimates of habitat use when misclassification probabilities can be determined. When misclassification probabilities cannot be determined, we use random subsamples from the error distribution of an estimated animal location to improve habitat use estimates. We conducted Monte Carlo simulations to evaluate the effects of this subsampling method, triangulation error, number of animal locations, habitat availability, and habitat complexity on bias and variation in habitat use estimates. Results for the subsampling method are illustrated using habitat selection by redhead ducks (Aythya americana ). We recommend the subsampling method with a minimum of 50 random points to reduce problems associated with habitat misclassification.