Accounting for Detection Probability in Fisheries Surveys Using Occupancy Modeling Approaches

Efforts to exam broad-scale patterns and changes in fish assemblages in an ecosystem over time are confounded by survey issues.  Survey gear may change over time, thereby affecting the catchability of some species.  Similarly, multiple agencies or institutions may have survey data available, but because different agencies likely use different gear types catchability may confound estimates of species composition.  In addition, rare or elusive species may be important indicators ecosystem change, but relative abundance estimates of these species may be unreliable. To explore changes in fish distributions in the Northwest Atlantic Coastal Shelf (NWACS), I applied occupancy modeling techniques to summary data sets of multiple fisheries trawl surveys conducted within regions of this ecosystem. This modeling approach provides information on regional occupancy probability and species-specific patterns in temporary colonization and local extinction rates in surveyed areas.  Occupancy modeling is frequently used in terrestrial systems where information on the occurrence of a rare or elusive species in a region is important for management decisions.  The data structure of long-term fisheries surveys is readily adapted for multiple-season occupancy models. These models combined with environmental covariates enable the evaluation of environmental factors that affect distributions.  Simultaneously estimating occupancy/occupancy dynamics parameters (e.g., regional colonization and extinction) and detection probability effectively removes the confounding issues of heterogeneity in surveys and gears used in the regions of this system.