OOS 32-6 - Study design considerations for integrated population models: Improving conservation and management of polar bears

Thursday, August 10, 2017: 9:50 AM
Portland Blrm 254, Oregon Convention Center
Nathan J. Hostetter, USGS Patuxent Wildlife Research Center, Laurel, MD, Sarah J. Converse, USGS Washington Cooperative Fish and Wildlife Research Unit, University of Washington, Seattle, WA and Eric V. Regehr, Polar Science Center - Applied Physics Laboratory, University of Washington, Seattle, WA

Population surveys are an important component of wildlife conservation, providing the data required to estimate population size, monitor demographic rates, and evaluate responses to management actions and environmental factors. Surveying species of conservation concern is often expensive and logistically challenging, especially for species that are rare, occur in low densities, or range across large or remote areas. As such, it is important to evaluate possible study designs prior to data collection and verify results will be defensible (i.e. unbiased), useful (i.e. acceptable precision), and efficient. Recent developments in integrative modeling provide new possibilities for optimal study design, where the use of multiple data sources can reduce bias, improve precision, and provide the potential to estimate vital rates when data are unavailable. We present a simulation-based approach to compare and evaluate different integrative study designs for the estimation of abundance, survival, and reproductive parameters. Motivation for this study is based on monitoring and recovery efforts for polar bears, which are currently listed as Threatened under the US Endangered Species Act.


Simulation-based approaches provide a framework to evaluate integrative study designs that include different levels of survey effort and allocation. We emphasize the potential benefits of mark-recapture study designs, especially the collection of single-season, robust-design, or spatial capture-recapture data. Study design performance is evaluated through bias and precision of abundance and vital rate estimates. Additionally, we describe how parameter identifiability varies across study designs, a particularly important consideration when recovery goals require estimation of specific demographic rates. We argue that (1) evaluation of integrative study designs should be conducted prior to data collection, (2) integrative modeling approaches increase the number of identifiable parameters, including those vital to conservation management, and (3) there are important trade-offs in survey effort, allocation, and objectives. Overall, simulation-based approaches provide an analytical method to evaluate study designs for integrative models and can serve as a template for practitioners designing their own studies.