PhD Defense by Vania Henríquez Tribino

Date: Thursday, October 24, 2024
Time: 9:30 AM
Location: Over Zoom (Link TBA).

Evaluating approaches to model time-varying natural mortality and predation in stock assessment models

Over the past two decades, persistent declines in various fish stocks in the Atlantic and Pacific Oceans have been observed. This has led to significant reductions in fishing effort, expecting rapid stock recovery. Some stocks have recovered, but for many others, this has not happened. Marine mammal predation has been hypothesized to be a key factor, but understanding how predation affects the rate of natural mortality (M) requires further research. Reliable stock assessment estimation of M is needed to understand variation in its magnitude and trends, and provide better informed management advice.

I developed robust and straightforward methodologies to complement existing stock assessment models that estimate non-stationary M. I applied Virtual Population Analysis (VPA), Statistical-Catch-at-Age models (SCA), and Stock Reduction Analysis (SRA) with configurations to model temporal and age-specific changes in M. VPA and SCA models were applied to an Atlantic cod (Gadus morhua) stock, while VPA and SRA models were applied to five Pacific herring (Clupea pallasii) stocks. The modeling focused on stocks that have not recovered (cod and some herring stocks) despite reduced fishing effort, considering the possibility that this is due to changes in M. VPA and SRA models accounted for predation by grey seals (cod) and Steller sea lions (herring).

The results showed increases in M in recent years for all analyzed stocks, regardless of the modeling approach. Depensatory M (i.e., high M at low abundance) is impacting Atlantic cod and Pacific herring by weakening the compensatory response of per capita surplus production rates. High M in cod was not entirely attributable to seal predation, while time-varying M in herring could be associated with Steller sea lion predation.

Simulation-estimation procedures showed that VPA estimated M more accurately than SCA. However, estimating changes across ages and time often resulted in biased estimates.
Future work should integrate these models into management procedures (MPs) under different predation scenarios and apply them to stocks with persistent low biomass, fishery closures, and suspected predation impacts on recovery.

This dissertation addresses knowledge gaps on time-varying M and improves the integration of predation in existing stock assessment models.

Committee members

Supervisor: Dr. Murdoch McAllister
Supervisory Committee Member: Dr. Carl Walters
Supervisory Committee Member: Dr. Villy Christensen
Supervisory Committee Member: Dr. Yanjun Wang
University Examiners: Dr. Scott Hinch, Dr. Dave Rosen