Quota allocation for stocks that span multiple management zones: analysis with a vector autoregressive spatiotemporal model

Allocating quotas among stakeholders requires an agreed-upon formula. If the stock unit spans multiple management jurisdictions, the formula may require updated biomass estimates of the stock's spatial distribution with respect to those jurisdictions. Data for calculating stock biomass often come from fishery-independent surveys. While stratified random sampling is a common design, strata boundaries may not always align with state or national jurisdictions, requiring post hoc stratification and imputation to calculate area-specific biomass. The vector autoregressive spatiotemporal (VAST) model was explored as a tool for calculating fish biomass within subareas of a defined stock unit for three different stocks jointly managed by the United States and Canada on Georges Bank in the Northwest Atlantic Ocean. VAST estimated proportions of stock biomass in each nation's waters were compared with an existing allocation algorithm that utilises a loess smooth through the average design-based swept area biomass from three fishery-independent surveys. The ability of VAST to impute biomass when no tows occur in a subarea of a survey stratum was demonstrated, as well as the relatively smoother biomass trend compared with design-based estimates, which may be desirable if the intent is to avoid large inter-annual swings in allocated quota.