Fixed and mixed effect models for fishery data on depth distribution of Georges Bank yellowtail flounder

Abstract

Fishermen reported that Georges Bank yellowtail flounder (Limanda ferruginea) migrated to deeper waters during 2000–2004 and 2006–2010. To test this hypothesis, we analyzed fishery data from otter trawl vessels targeting a mixed groundfish complex over the 10 year period, using a statistical linear model with catch-per-unit-effort weighted depth as the response variable, and abiotic (e.g., bottom water temperature) and biotic (e.g., skate and dogfish catch) data as predictor variables. We considered mixed as well as fixed effect models to account for dependence or correlation in catches among hauls within a trip. Yellowtail flounder shifted to deeper waters during the 10 years. Bottom water temperature had a greater influence on the movement than the distribution of skate or dogfish. Optimal water temperature was about 6.8°C from the fixed effect model and about 7.1°C from the mixed effect model. Skate distribution affected yellowtail flounder depth more than dogfish distribution. The mixed effect model was more parsimonious than the fixed effect model, although the latter fitted the data better and performed better under cross validation.