Using a Multi-Gear Occupancy Model to Estimate Detection Probability and Index the Distribution of Red Snapper in the Southeast United States

Fishery-independent monitoring programs often neglect important sources of error when developing indices of abundance.  One often overlooked source of error is imperfect detection, which is the inability of survey gears to detect all individuals or species present in a surveyed area.  There has been significant progress in developing monitoring methods that account for imperfect detection, but these methods have rarely been utilized in marine environments.  We develop a novel multi-gear occupancy model for red snapper Lutjanus campechanus using presence-absence data from two gears deployed simultaneously: chevron fish traps and attached underwater video cameras.  Red snapper occupancy models were constructed such that the underlying distribution across the landscape was modeled separately from detection.  Year, depth, latitude, habitat type, and temperature influenced the distribution of red snapper, while water temperature influenced trap detection probability.  Using a goodness-of-fit test based on parametric bootstrapping, there was no evidence that the red snapper occupancy model fit poorly, nor did simulations suggest that potential non-independence between gears was a concern.  Developing occupancy models for exploited marine fish species may be advantageous because imperfect detection can be explicitly accounted for, and indices of distribution (or abundance) based on two combined gears will likely be more robust than indices from separate gears.