A hierarchical model combining distance sampling and time removal to estimate detection probability during avian point counts
Courtney L. Amundson,J. Andrew Royle,Colleen M. Handel
DOI: https://doi.org/10.1642/auk-14-11.1
2014-10-01
The Auk
Abstract:Imperfect detection during animal surveys biases estimates of abundance and can lead to improper conclusions regarding distribution and population trends. Farnsworth et al. (2005) developed a combined distance-sampling and time-removal model for point-transect surveys that addresses both availability (the probability that an animal is available for detection; e.g., that a bird sings) and perceptibility (the probability that an observer detects an animal, given that it is available for detection). We developed a hierarchical extension of the combined model that provides an integrated analysis framework for a collection of survey points at which both distance from the observer and time of initial detection are recorded. Implemented in a Bayesian framework, this extension facilitates evaluating covariates on abundance and detection probability, incorporating excess zero counts (i.e. zero-inflation), accounting for spatial autocorrelation, and estimating population density. Species-specific characteristics, such as behavioral displays and territorial dispersion, may lead to different patterns of availability and perceptibility, which may, in turn, influence the performance of such hierarchical models. Therefore, we first test our proposed model using simulated data under different scenarios of availability and perceptibility. We then illustrate its performance with empirical point-transect data for a songbird that consistently produces loud, frequent, primarily auditory signals, the Golden-crowned Sparrow (Zonotrichia atricapilla); and for 2 ptarmigan species (Lagopus spp.) that produce more intermittent, subtle, and primarily visual cues. Data were collected by multiple observers along point transects across a broad landscape in southwest Alaska, so we evaluated point-level covariates on perceptibility (observer and habitat), availability (date within season and time of day), and abundance (habitat, elevation, and slope), and included a nested point-within-transect and park-level effect. Our results suggest that this model can provide insight into the detection process during avian surveys and reduce bias in estimates of relative abundance but is best applied to surveys of species with greater availability (e.g., breeding songbirds).