To pool or not to pool data? Applying a generalized depletion model to assess American eel elver Anguilla rostrata fisheries from multiple rivers in Nova Scotia, Canada

Spatial scales are important for examining health of exploited fishery stocks and guiding management actions. However, information about the optimal spatial scale is still unclear for assessment of transit fisheries, such as elver fisheries of the American eel Anguilla rostrata. We applied a generalized depletion model to assess catch and effort data from three nearby rivers (within 50 km) to test the hypothesis that modeling on pooled and separate data from nearby rivers would give similar estimates of abundance and exploitation rate. Overall, pooling data from rivers within 50 km did not result in large differences (<20% in relative difference) in estimates of abundance and exploitation rate with close mean abundance estimates and similar temporal trends in abundance, exploitation rate, and relative escapement. Pooling nearby river systems can greatly reduce modeling effort, at the cost of ignoring fine-scale variability in elver recruitment and having coarser spatial scale for the management. When only an index of annual recruitment and exploitation rate are of interest, pooling data may be practical from different locations up to 50 km.