Evaluating the Potential for Freshwater Bycatch Using Likelihood Methods and Fishery-Independent Data

Effective management of fishery resources is an activity often limited by poor data availability, particularly with regards to uncertainties in stock distribution and fishing effort. Despite such uncertainties, quantifying the spatiotemporal distribution of bycatch is critical to assess the likelihood of incidental harvest, possible impacts, and effectiveness of management strategies (such as spatial or temporal restrictions) that may be necessary to protect spawning stocks or critically imperiled fishes.  Recent studies have identified the possibility of population-level mortality associated with non-retention in fishery gear. Such impacts emphasize the potential for underestimates of bycatch risk when projections are based solely upon fishery-dependant data.  To accounts for such issues and illustrate the paucity of studies estimating freshwater bycatch risk, we demonstrate a probabilistic modeling approach that implicitly incorporates the spatial uncertainty of harvest activity when fishery dependant data are unattainable or unsuitable for interpretation. Simulation models were based on fishery-independent data from a live-removal bait fishery in Ontario, Canada. The likelihood of distributional co-occurrences between target fishes and game, imperiled, invasive and other non-target fish bycatch was determined.  Generally, spatial overlap of target and non-target stocks was relatively common, but the likelihood of co-occurrences varied substantially depending on the fishery type and species.  Model simulations of harvest activity implied substantial uncertainty of bycatch at low harvest effort, regardless of bycatch group, species and harvest strategy.  Greater certainty of bycatch was associated with high harvest effort.  Simulations implied that within the Lake Erie nearshore harvest, greatest bycatch likelihood was associated with invasive fishes; whereas, game bycatch was prevalent for Great Lakes tributary harvests. Results emphasized the variability of bycatch risk associated with fisheries composed of many, discrete stocks. Likelihood approaches may provide methodological advantages when managers are forced to make decisions to mitigate bycatch risk in light of substantial uncertainty of the spatial distribution, and intensity, of harvest activity.  In the absence of fishery-dependant data, such approaches may be used to estimate the relative risk of incidental harvest as a necessary first step to inform stock management of non-target fishes.