Developing a new spatio-temporal framework for assessment of the Nova Scotia Inshore Sea Scallop (Placopecten magellanicus) Fishery
Date
2020-08-18T13:16:32Z
Authors
McDonald, Raphael
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
The current stock assessment model model used to assess areas of the Nova Scotia Inshore Sea Scallop (Placopecten Magellanicus) Fishery is a delay-difference model. However, this model has been found to be unable to reliably model the Scallop Production Area 3. The focus of this work is to incorporate spatial information into this assessment model in two steps. The first step involves reconsidering abundance indices to reduce the amount of necessary pre-processing and directly model all intra-annual variability, while simultaneously accounting for the large number of zeroes. The second step involves explicitly modeling the location of survey tows and commercial fishing by modifying the error structure of the model by using Gaussian Markov Random Fields such that a spatio-temporal model results.
The new framework for abundance indices is shown to better capture population changes and can be viewed as a hybrid between a traditional temporal model and a spatio-temporal version. The full spatio-temporal stock assessment framework is further able to capture both local population changes reliably and population trends for the entire area of interest. This novel framework shows promise to improve the reliability of scientific advice given to fisheries managers while opening up new possibilities for spatial management.
Description
The Nova Scotia Inshore Sea Scallop (Placopecten Magellanicus) Fishery is the fourth most valuable fishery in the province and has been in existence since the 1960s. Covering the Bay of Fundy and areas close to the shore of Nova Scotia, it is split up into multiple Scallop Production Areas (SPA) and Scallop Fishing Areas (SFA). Most areas are assessed separately with a different yearly Total Allowable Catch (TAC) set to limit maximum landings. SPA 3, encompassing St-Mary’s Bay and a good portion of the western shore of Nova Scotia, has been difficult to assess due to its strong spatial patterns in both biological characteristics and fishing effort.
The current model typically used to assess areas of this fishery is a delay-difference model, a type of biomass dynamics model which only requires an index of abundance and commercial landings. However, even after recasting it into a frequentist framework, this model has been found to be unable to reliably model SPA 3. The focus of this work is to incorporate spatial information into this assessment model in two steps. The first step involves reconsidering abundance indices to reduce the amount of necessary pre-processing and directly model all intra-annual variability, while simultaneously accounting for the large number of zeroes. The second step involves explicitly modeling the location of survey tows and commercial fishing by modifying the error structure of the model by using Gaussian Markov Random Fields such that a spatio-temporal model results.
The new framework for abundance indices is shown to better capture population changes and can be viewed as a hybrid between a traditional temporal model and a spatio-temporal version. The full spatio-temporal stock assessment framework is further able to capture both local population changes reliably and population trends for the entire area of interest. This novel framework shows promise to improve the reliability of scientific advice given to fisheries managers while opening up new possibilities for spatial management.
Keywords
stock assessment, fisheries, spatial, random fields, delay-difference model, scallops