Oral Presentation Australian Society for Fish Biology Conference 2017

Spatial structure in stock assessments: results of simulation-estimation experiements (#138)

Andre Punt 1 , Malcolm Haddon 1 , Geoff Tuck 1 , Richard Little 1
  1. CSIRO, Hobart, TASMANIA, Australia

Most stock assessments are spatially-aggregated owing to a lack of data and the lack of a platform for conducting spatially-structured stock assessments. Moreover, it is unclear how poor estimates of quantities of management interest are when either spatial structure is ignored when conducting stock assessments or the assumptions or the assessment model underlying the assessment are mis-specified. A simulation-estimation experiment based on the stock assessment package Stock Synthesis is undertaken to explore these issues. The operating model considers three regions and is based on pink ling, Genypterus blacodes, off the east coast of Australia. Fishing pressure varies spatially for pink ling and it is unlikely that there is substantial post-recruitment movement, which implies that the sub-populations of pink ling will differ in their demographic structure spatially. A range of estimation models ranging from aggregating data spatially, to applying the areas-as-fleets approach to handling spatial heterogeneity, to applying fully spatially-structured models is considered. In addition, simulation scenarios consider the possibility that one of the region is closed to exploitation. Non-spatial assessment configurations that aggregate spatially-structured data provide more precise, but nevertheless biased estimates of initial and final spawning biomass, as well as biased estimates of the ratio between initial and final spawning biomass. A spatially-structured assessment configuration that correctly matches the structure of the model used to generate the simulated datasets is unbiased but imprecise. The bias in estimates of spawning stock biomass associated with spatially-aggregated assessment methods increases in the presence of closed areas while these biases can be reduced (or even eliminated) by applying appropriately constructed spatially-structured stock assessments. The performance of spatially-aggregated assessments when estimating spawning stock biomass is found to depend on the interactions among spatial variation in growth, in exploitation rate, and in knowledge of the spatial areas over which growth and exploitation rate are homogeneous.