Incorporating detection probability to estimate pheasant density

Abstract

ABSTRACTIndices of abundance, such as point counts, commonly are used to monitor trends in bird populations. In some circumstances, however, an index of abundance provides insufficient information for making management decisions and accurate density estimates are necessary. Wild ring‐necked pheasants (Phasianus colchicus) were translocated to 10 study areas in Pennsylvania from 2007 to 2014 with the goal of establishing female densities of 3.86 pheasants/km2. We developed a population density estimator that used 3‐minute crowing counts adjusted for probability of detection to estimate male pheasant density and flushing surveys to estimate the female:male ratio. To account for detection probability, we estimated the probability a pheasant was available to be detected by monitoring crowing frequency of male pheasants fitted with radio‐transmitters and the probability an observer was able to detect a crowing pheasant at distances from 0 to 0.93 km. We found the probability a pheasant crowed during 3 minutes decreased linearly over our survey period from 0.66 in mid‐April to 0.46 by the end of May. At the farthest distance we were able to accurately detect a crowing pheasant. We estimated the probability of detecting a pheasant at 0.80 km to be 0.019 ± 0.005 (SE), which means that we could not assume any fixed distance beyond which crowing birds could not be detected. Therefore, we replaced the probability of detection in the standard distance sampling estimator with the effective area of detection. The estimation of the effective area of detection is robust to choice of radius of the point and did not require observers to estimate the distance to crowing pheasants. We estimated the female:male ratio to be 1.02:1, despite the ratio of released pheasants being 4.46:1. Only 1 study area achieved the female density goal ( = 4.16); the maximum density at all other study areas was <2 females/km2. The estimator we developed incorporated multiple detection probabilities to provide density estimates and simplified the crowing count protocol by eliminating the need for observers to estimate their distance from a detected bird, which makes the estimator useful for estimation of population abundance when explicit population density objectives must be evaluated. © 2018 The Wildlife Society.