Approaches to projecting recruitment in toothfish assessment models

The current CCAMLR toothfish Decision Rules to determine catch limits for assessed toothfish fisheries are based on a stock projection forward for 35 years from the most recent year included in the assessment. As a result, the outcomes rely heavily on recruitment assumptions as the stock and catch at the end of the projection period will be dominated by fish that are not yet born. 

We investigate the effects of alternative recruitment assumptions on the estimated SSB status of the stock over a 35-year projection period and show how alternative recruitment assumptions effect the precautionary catch limits using the CCAMLR toothfish Decision Rules.

The recruitment trends estimated from the four integrated toothfish assessment models differed in both the nature of the trend and the degree of autocorrelation between successive years. There is evidence for some stocks, that the estimated recruitment trends in the stock assessments differed from those observed in the research surveys. 

The expected stock status at the end of the 35-year projection can be highly influenced by the recruitment assumptions, and equivalently that the catch limit calculated from the current Decision Rules are highly dependent on those recruitment assumptions. Based on this work, we recommend that WG-SAM consider:

1. Comparing model estimates of recruitment to observations of recruitment from data such as research surveys and identifying where differences in trends may reflect model misspecification or observation bias.
2. That where there are trends in model estimates of recruitment YCS, stock assessors should explore model sensitivity runs that investigate whether the sources of this trend could be related to model misspecification including spatial changes in the fishery.
3. Where there is substantial evidence of a drop in recruitment, recent recruitment (e.g., the most recent estimated 10 years of recruitments) should be used in projections to determine the precautionary catch limits for the CCAMLR toothfish decision rules. Further work is needed to explore the rationale for identifying a suitable number of years (i.e. time frame) from which to estimate recent recruitment. 
4. Evaluations of the Decision Rules and alternative harvest control rules should include a range of recruitment scenarios, including those investigated in this paper.