On The Estimation Of Detection Probabilities For Sampling Stream Dwelling Fishes

Author :
Publisher :
ISBN 13 :
Total Pages : 68 pages
Book Rating : 4.:/5 (13 download)

DOWNLOAD NOW!

Download or read book Estimating Detection Probabilities for Fishers Using Non-invasive Methods and Implications for Survey Protocols written by and published by . This book was released on 2009 with total page 68 pages. Available in PDF, EPUB and Kindle. Book excerpt: Conservation concern for the fisher in the western United States has become heightened in the last decade, culminating in the listing in 2004 as 'warranted but precluded' under the U. S. Endangered Species Act. Occupancy surveys using non-invasive detection devices have, and continue to, serve a major role in the research and management of fishers in the western U.S. However, the standard device-specific survey protocols have not been subjected to quantitative evaluation using the recently developed methods for occupancy estimation. These methods involve the estimation of two parameters (P and S|) from the collection of detection histories, the sequences of 0s and 1s representing detection or non-detection of a fisher, over the duration of a survey at each sample unit. P refers directly to the ability of the survey protocol to detect a fisher given its presence in the sample unit, which directly refers to the characteristics (e.g., number of stations, number of visits, survey duration) of the survey protocol. S| refers to the probability that one or more fishers occupies the area surveyed, which is directly related to the ecological characteristics of the locations surveyed (e.g., habitat structure, carnivore community). The focus of this paper is to address issues affecting P, not probability of occupancy, thus we did not attempt to explore habitat or community covariate effects on S|. We refer to this enterprise as occupancy estimation and address the application of these methods for two specific survey objectives: (1) assessing the occurrence of fishers at a single location and (2) a non-spatial method for assessing distribution. We used data from standardized occupancy survey protocols using track plates, remote cameras, and hair snares to detect fishers in the western United States to: (1) estimate their respective detection probabilities (2) investigate factors causing detection heterogeneity, and (3) evaluate the effects of spatial or temporal changes to survey protocols. We found three significant sources of detection heterogeneity: (1) within sample unit visit dependency, (2) visit-specific detection probabilities, and (3) survey season. Incorporating covariates to account for these sources of detection heterogeneity significantly increased the accuracy of estimates and inferences from previous analyses. Visit dependency was accounted for by adding a Markov chain persistence factor. The probability of detection varied by visit, generally increasing throughout the survey duration. Fishers had a lower probability of detection during the summer season (July-September) compared to the rest of the year. We developed new protocol recommendations largely on the basis of the analysis of the track plate data but we assume it would apply also to the use of cameras and hair snares. Our primary conclusion is that to achieve the same probability of detection, surveys in the summer (July-September) require longer durations than non-summer seasons (October-June). Based on our analysis and review of the literature, we propose that survey effort (# stations x survey duration) be used to define quantitative thresholds for non-invasive detection device protocols. Distributional surveys or monitoring efforts should use protocols to achieve a 0.80 detection probability and surveys to assess the occurrence of fishers at a single-location (e.g., pre-project surveys) use protocols to achieve a 0.95 detection probability. Non-summer season protocols need 50 and 60 functional survey days of survey effort to achieve probabilities of detection of at least 0.80 and 0.95, respectively. Summer season protocols need 120 and 200 functional survey days of survey effort to achieve detection probabilities of at least 0.8 and 0.95, respectively. These recommendations assume the use of 6 track plate stations or 2 camera stations per sample unit, with minimal spacing of 500m and 1.6 km between stations, respectively. We recommend that the minimally valid survey effort for evaluating previous survey results be set to achieve a detection probability of at least 0.95, resulting in a minimum of 200 and 60 survey days needed per sample unit during the summer and non-summer seasons, respectively.