COS 308-3 - Integrating mark-resight and count data to estimate effective sampling area and fish density

Thursday, August 10, 2023

4:00 PM – 4:15 PM PDT

Location: B114

Presenting Author(s)

VZ

Viviane Zulian

Postdoctoral Researcher
North Carolina Cooperative Fish and Wildlife Research Unit, Department of Applied Ecology, North Carolina State University, Raleigh, North Carolina, USA.
Raleigh, North Carolina, United States

Coauthor(s)

KP

Krishna Pacifici

Department of Forestry and Environmental Resources, North Carolina State University, Raleigh, North Carolina, USA., North Carolina, United States
NB

Nathan M. Bacheler

Southeast Fisheries Science Center, National Marine Fisheries Service, Beaufort, North Carolina, USA., North Carolina, United States
JB

Jeffrey A. Buckel

Department of Applied Ecology, Center for Marine Sciences and Technology, North Carolina State University, Morehead City, North Carolina, USA., North Carolina, United States
WP

William F. Patterson III

University of Florida, School of Forest, Fisheries, and Geomatics Sciences, Gainesville, Florida, USA., Florida, United States
BR

Brian J. Reich

Department of Statistics, North Carolina State University, Raleigh, North Carolina, USA, North Carolina, United States
KS

Kyle W. Shertzer

Southeast Fisheries Science Center, National Marine Fisheries Service, Beaufort, North Carolina, USA., North Carolina, United States
NH

Nathan Hostetter

USGS North Carolina Cooperative Fish and Wildlife Research Unit, Department of Applied Ecology, North Carolina State University, Raleigh, North Carolina, USA., North Carolina, United States

Abstract: Conservation and management of populations benefit from accurate estimates of abundance. N-Mixture models are commonly used to estimate abundance from count data based on the assumptions of a closed population among visits and a known sampling area. Populations of mobile individuals often violate the closure assumption, leading to an unknown effective sampling area (ESA) and inability to estimate density from N-mixture models. Here we describe an approach to estimate the ESA and density from repeated counts of unmarked and previously marked individuals within a modified Marked-N-mixture framework. Our work was motivated by a 3-dimensional telemetry study of red snapper (Lutjanus campechanus), an economically and ecologically important fish species, at a reef in the Atlantic Ocean off North Carolina, USA. We marked 16 individual fish with external, acoustic tags and subsequently deployed 31 baited cameras across the reef (1.6 km²). Cameras recorded observations of unmarked and marked individuals every 30-seconds for 12-minutes, while acoustic data indicated the number of tagged fish on the reef during that period. We developed a modified Marked-N-Mixture model to estimate abundance, ESA, detection probability, and covariate relationships, and we evaluated model performance in a case study and using simulations under a range of scenarios of data quality and animal movement. Estimated ESA varied as a function of current direction and a camera-specific random effect, with lower ESA occurring when the current was toward a camera (i.e., individuals were likely behind the camera). The estimated density of red snapper was 102 (95% CI: 99-165) individuals/km². Simulation results demonstrated the Marked-N-Mixture model improved the precision of parameter estimates relative to the standard N-Mixture model, even under scenarios of poor data (e.g., low detection probability). In addition, the Marked-N-Mixture model provided unbiased estimates under multiple scenarios of animal movement, while bias in the standard N-Mixture model estimates increased as space use of individuals increased. This approach demonstrates a flexible alternative to estimating density in geographically open populations and will be particularly useful for studies where some animals are marked but unique identification is not possible during resighting efforts (e.g., batch marks, non-field readable tags).