Workshop on ICES Reference Points (WKREF1)

The ICES Workshop on ICES reference points (WKREF1) was tasked to provide a thorough review of the ICES reference points system as a basis to re-evaluate the process for estimating, updating and communicating reference points in the context of the ICES advice. As part of the preparation leading to WKREF1 a large database of the most recent assessment outputs for 78 Category 1 stocks were collated in the form `FLStock` objects, which formed the basis for several components of the presented analyses.

The first part of the meeting involved a detailed overview of the history and basis of the ICES references points system, which was aligned with the results of an empirical review of the procedures and choices made to derive ICES reference points for category 1 stocks. The ICES procedures were then contrasted with those used in the USA, Canada, New Zealand and across tuna Regional Fishery Management Organizations. A limitation in terms of transparency of the ICES procedures is a lack of complete documentation of the settings used for deriving reference points using, e.g., the EQSIM software. In comparison to other international standards, the main differences identified include the absence of a target biomass reference point and inconsistent estimates of the limit biomass reference point B_lim, which is estimated to be below 10% of the unfished biomass (B0) for a high proportion of analysed stocks (around 50%). In addition, an important difference is that direct estimates of F_MSY are used in ICES (which can be unreliable), whereas elsewhere F_MSY is often replaced by more conservative biological proxies, such as Fspr% and FB%.

The second part focused on the robustness evaluation of the current ICES reference point system. Work presented included examples that demonstrate differences in reference point estimates between standard ICES procedures (EQSIM) and full Management Strategy Evaluation (MSE) simulations, and a study that highlighted the considerable uncertainty in estimating B_lim, with guidance on how to quantify uncertainty depending on length and contrast in the time-series. A large simulation experiment was conducted by applying a short-cut MSE approach to 68 Category 1 stocks, which revealed that the ICES MSY advice rule is the least robust of all tested generic approaches if assumptions about the typically highly uncertain stock recruitment relationship are violated. This led to poor performance of the ICES MSY advice rule associated with the lowest long-term yields, highest risks to fall below limit reference points and lowest probabilities of attaining biomass levels at MSY. Contributing factors to the poor performance were combinations of comparably high F_MSY estimates and low B_lim (<10% B0) and thus low MSY B_trigger (<14% B0) values. In cases where the precautionary FP.05 was invoked (mostly B_lim > 10% B0), performance improved notably. Results from a backtest using hindcasting with forecasts of 1-5 years reinforced the need to re-estimate reference points regularly at benchmark assessments, and that short-term forecasts should not exceed a three-year time span to account for time-varying biological traits. Finally, presented work highlighted the advantages (in terms of high consistency and accuracy) of estimating the stock-recruitment relationship or reference points internally in the assessment model, which was illustrated for Stock Synthesis and SAM, and supported by a comprehensive simulation study.

The key recommendations of WKREF1 were to: i) revise and simplify how B_lim is derived. An absolute B_lim should only be specified empirically in cases where there is sufficient contrast in the stock-recruit data to estimate a well-defined break-point. Alternatively, it is suggested that B_lim should be determined as a plausible ratio of B0 based on biological principles and the life history of the stock (e.g. 10-25% B0 depending on the stocks characteristic; Section 7). ; ii) FP.05 should be calculated without B_trigger; iii) to use biological proxies (F_brp) for deriving FMSY, and the resultant F_MSY proxy must not exceed FP.05; iv) to report biomass target (B_trg) that corresponds to the F_MSY proxy; and v) to set B_trigger as either a fraction of Btrg or multiplier of Blim. Specifications of setting reference points (e.g. F_brp) or Btrigger should be informed through further simulation testing to be presented at WKREF2.