Date of Award
2025
Degree Type
Dissertation
Degree Name
Doctor of Philosophy in Oceanography
Specialization
Biological Oceanography
Department
Oceanography
First Advisor
Jeremy Collie
Abstract
Commercial fishing is a highly valuable industry in the United States and globally, and as such, managing based on the best available science is necessary to maintain important fish stocks. Some parameters in stock assessment models that are assumed to be constant over age and time may actually vary to an extent that stock assessment modeling results could be improved by allowing the parameters to vary in the models. In the present dissertation, two stock assessment parameters that are typically held constant are explored for potential time or age variance. Potential environmental drivers of nonstationary parameters are also explored. Productivity, or per-capita recruitment rate, was found to be time-varying in 50 out of 84 United States commercial fish stocks. Generalized linear modeling suggested that stocks with higher contrast, a higher range of low to high values, in spawning stock biomass over time, were more likely to have time-varying productivity than stocks with low contrast. Of the stocks with time-varying productivity, 35 stocks were found to have at least one environmental driver of productivity that improved model fit, either based on the values of the environmental variable itself, or on the variable’s rate of change. Simulation tests indicated that if the true system has a climate driver, most of the time a climate-enhanced model will detect it. However, the different model variants may have similar statistical support, such that model selection may be unable to distinguish the ecological hypotheses of the climate-recruitment interaction. Another parameter that may vary with time or age is natural mortality rate (M). Using Southern New England/Mid-Atlantic (SNEMA) winter flounder as a case study, the Woods Hole Assessment Model (WHAM) was configured to explore potential nonstationarity and environmental drivers of M. Model results suggest that under current conditions, the natural mortality rate may be higher than the current assumed rate of 0.3 for SNEMA winter flounder. Furthermore, M may increase with age, with time, and with temperature. Simulation tests suggest that when M varies with time and age, including this variation in the modeling framework can reduce bias in the M estimates. Both the analyses of productivity and natural mortality rate in the present dissertation indicate that parameters typically held constant in stock assessment models may actually vary with time or age, and stock assessment models may provide more reliable management advice if these variations are accounted for. While further stock-specific research is necessary before the results of the present dissertation can be applied to fisheries science and management, these results provide a valuable starting point for future study into model specification and stock-specific dynamics, and for management action.
Recommended Citation
Marshall, Rachel Catherine, "INCORPORATING TIME-VARYING PARAMETERS AND CLIMATE DRIVERS INTO FISHERIES MODELS" (2025). Open Access Dissertations. Paper 4517.
https://digitalcommons.uri.edu/oa_diss/4517