Modelling recruitment in a spatial context: A review of current approaches, simulation evaluation of options, and suggestions for best practices.

• An increasing number of stock assessments are based on spatially-structured models. • Six formulations for spatially-structured assessment models are outlined. • Estimation of migration rates within assessments leads to improved performance. • Guidelines are given for estimating recruitment in spatially-structured assessments. Many, if not most, stocks of fish and invertebrates subject to commercial, recreational or subsistence harvesting exhibit some degree of spatial heterogeneity. This can be due to a lack of complete movement within a single biological population or due to the assessed region consisting of multiple sub-stocks. In principle, assessments for stocks for which the assumption of homogeneity is violated should be based on population dynamics models that are spatially-explicit. There are, however, relatively few stocks for which multi-stock/multi-area assessments are conducted. However, the number of assessments in which space is explicitly represented in the population dynamics model has been increasing in recent years and such models are available for fish stocks such as New Zealand hoki and canary rockfish off the US west coast, as well as invertebrates such as the rock lobsters off southern Australia and New Zealand. A challenge within such spatially-explicit assessments pertains to how recruitment and movement are modelled. This paper reviews assessments for fish and invertebrate stocks that are based on spatially-explicit models, identifying the challenges associated with modelling recruitment caused by increased model and parameter estimation complexity, and the available solutions. A set of illustrative simulations are undertaken to evaluate the consequences of different assumptions regarding spatial recruitment and movement, and hence develop some recommendations for best practice guidelines. [ABSTRACT FROM AUTHOR]