Using simulation to evaluate wildlife survey designs: polar bears and seals in the Chukchi Sea

• Published in: Royal Society Open Science
• Main Authors: Conn, Paul B., Moreland, Erin E., Regehr, Eric V., Richmond, Erin L., Cameron, Michael F., Boveng, Peter L.
• Other Authors: National Oceanic and Atmospheric Administration
• Format: Article in Journal/Newspaper
• Language: English
• Published: The Royal Society 2016
• Subjects: Arctic; Chukchi Sea; Chukchi; polar bear; Sea ice
• Online Access: http://dx.doi.org/10.1098/rsos.150561; https://royalsocietypublishing.org/doi/pdf/10.1098/rsos.150561; https://royalsocietypublishing.org/doi/full-xml/10.1098/rsos.150561

Logistically demanding and expensive wildlife surveys should ideally yield defensible estimates. Here, we show how simulation can be used to evaluate alternative survey designs for estimating wildlife abundance. Specifically, we evaluate the potential of instrument-based aerial surveys (combining infrared imagery with high-resolution digital photography to detect and identify species) for estimating abundance of polar bears and seals in the Chukchi Sea. We investigate the consequences of different levels of survey effort, flight track allocation and model configuration on bias and precision of abundance estimators. For bearded seals (0.07 animals km −2 ) and ringed seals (1.29 animals km −2 ), we find that eight flights traversing ≈7840 km are sufficient to achieve target precision levels (coefficient of variation (CV)<20%) for a 2.94×10 5 km 2 study area. For polar bears (provisionally, 0.003 animals km −2 ), 12 flights traversing ≈11 760 km resulted in CVs ranging from 28 to 35%. Estimators were relatively unbiased with similar precision over different flight track allocation strategies and estimation models, although some combinations had superior performance. These findings suggest that instrument-based aerial surveys may provide a viable means for monitoring seal and polar bear populations on the surface of the sea ice over large Arctic regions. More broadly, our simulation-based approach to evaluating survey designs can serve as a template for biologists designing their own surveys.