You are here:
Hidden Markov models for estimating animal mortality from anthropogenic hazards
Citation:
Etterson, M. Hidden Markov models for estimating animal mortality from anthropogenic hazards. ECOLOGICAL APPLICATIONS. Ecological Society of America, Ithaca, NY, 23(8):1915-1925, (2013).
Impact/Purpose:
Carcasses searches are a common method for studying the risk of anthropogenic hazards to wildlife, including non-target poisoning and collisions with anthropogenic structures. Typically, numbers of carcasses found must be corrected for scavenging rates and imperfect detection. Parameters for these processes (scavenging and detection) are often estimated using carcass-distribution trials in which researchers place carcasses in the field at known times and locations. In this manuscript a discrete time analog of a commonly used estimator for the number of animals killed at wind turbines is derived and used to make explicit the assumptions underlying use of the estimator. A series of increasingly general estimators is developed, culminating in one that relies only on accidental kills, obviating the need for carcass distribution trials. Simulations are used to evaluate the bias and sampling variance associated with the use of specific estimators against more general alternatives. All estimators considered are shown to be essentially unbiased when the assumptions underlying them are met, but that even seemingly minor violations of assumptions might result in substantial bias in estimating the numbers of animals actually killed.
Description:
Carcasses searches are a common method for studying the risk of anthropogenic hazards to wildlife, including non-target poisoning and collisions with anthropogenic structures. Typically, numbers of carcasses found must be corrected for scavenging rates and imperfect detection. Parameters for these processes (scavenging and detection) are often estimated using carcass-distribution trials in which researchers place carcasses in the field at known times and locations. In this manuscript a discrete time analog of a commonly used estimator for the number of animals killed at wind turbines is derived and used to make explicit the assumptions underlying use of the estimator. A series of increasingly general estimators is developed, culminating in one that relies only on accidental kills, obviating the need for carcass distribution trials. Simulations are used to evaluate the bias and sampling variance associated with the use of specific estimators against more general alternatives. All estimators considered are shown to be essentially unbiased when the assumptions underlying them are met, but that even seemingly minor violations of assumptions might result in substantial bias in estimating the numbers of animals actually killed.
