Preliminary stock assessment of Indian ocean swordfish (Xiphias gladius) using the Bayesian state-space surplus production model JABBA

Bayesian State-Space Surplus Production Models were fitted to Indian Ocean swordfish (Xiphias gladius) catch and CPUE data using the ‘JABBA’ R package. This document presents details on the model diagnostics and stock status estimates for a single ‘Reference’ model with a prior distribution for r of log(r) ~ N(log(0.42),0.4) and a fixed input value of B_MSY/K = 0.4 (Pella-Tomlinson model type). Generally, the CPUE indices were consistent in showing a period of decline from early 1990s until mid-2000s, thereafter stabilizing and even increasing in the northeast and northwest areas. The model fit the CPUE data reasonably well (RMSE = 20.7%) with marginal data conflict between the CPUE indices in the last 5 years. The MSY estimate was 30,630 metric tons, which is very similar to the current catch (catch₂₀₁₈: 30,686 metric tons) and the current estimate of biomass as a proportion of “pristine” biomass was B₂₀₁₈/K = 0.47. Results from the Reference Model indicate that there is an 81% probability that the swordfish stock status currently falls within the green quadrant of the Kobe biplot (B₂₀₁₈>B_MSY and F₂₀₁₈<F_MSY). A retrospective analysis indicated a negligible retrospective pattern, and hindcasting cross-validation results suggested that the model has good prediction skill (MASE = 0.72). Various scenarios of CPUE input data were explored using a sensitivity analysis, and the trends in biomass and stock status estimates were fairly insensitive to variations in CPUE input data. Notably, the inclusion of all of the available CPUE indices produced estimates that were the most similar to the Reference Model, while the inclusion of the Indonesian CPUE index produced the most pessimistic results.