A Hierarchical Community Occurrence Model for North Carolina Stream Fish. Issue 5 (30th August 2014)

• Record Type: Journal Article
• Authors: Midway, Stephen R.; Wagner, Tyler; Tracy, Bryn H.
• Is Part Of: Transactions of the American Fisheries Society. Volume 143:Issue 5(2014)
• Page Start: 1348
• Page End: 1357
• Publication Date: 2014-08-30
• Subjects: Fish culture -- Periodicals; Fishery management -- Periodicals; Pisciculture -- Périodiques; 639.2
• DOI: 10.1080/00028487.2014.931745
• Languages: English
• ISSNs: 0002-8487

Abstract

Abstract: The southeastern USA is home to one of the richest—and most imperiled and threatened—freshwater fish assemblages in North America. For many of these rare and threatened species, conservation efforts are often limited by a lack of data. Drawing on a unique and extensive data set spanning over 20 years, we modeled occurrence probabilities of 126 stream fish species sampled throughout North Carolina, many of which occur more broadly in the southeastern USA. Specifically, we developed species‐specific occurrence probabilities from hierarchical Bayesian multispecies models that were based on common land use and land cover covariates. We also used index of biotic integrity tolerance classifications as a second level in the model hierarchy; we identify this level as informative for our work, but it is flexible for future model applications. Based on the partial‐pooling property of the models, we were able to generate occurrence probabilities for many imperiled and data‐poor species in addition to highlighting a considerable amount of occurrence heterogeneity that supports species‐specific investigations whenever possible. Our results provide critical species‐level information on many threatened and imperiled species as well as information that may assist with re‐evaluation of existing management strategies, such as the use of surrogate species. Finally, we highlight the use of a relatively simple hierarchical model that can easily be generalized for similar situations in which conventional models fail to provide reliable estimates for data‐poor groups. Received January 13, 2014; accepted May 5, 2014