Improving sustainable yield estimates for tropical reef fisheries

Fishing sustainably is a fundamental problem in tropical regions where diverse fisheries and scarce fisheries information challenges efforts to make reliable estimates and associated policies. To improve evaluations and decisions, we compared the predictions of six surplus production models calibrated using various permutations of fisheries‐dependent data with a benchmark model. The benchmark model was built from fisheries‐independent estimates of r and K , tested against rates of change in Kenyan reef fisheries and found to be accurate. Comparisons with the benchmark model were made with fisheries‐dependent equilibrium and non‐equilibrium models, fixing or not fixing r and K , pooled versus site averaged solutions, and rising, falling, and pooled fishing effort over time. Evaluations indicate high variability in MMSY predictions and notable overestimates of MMSY (~75%) and effort (~210%) for Fox and Schaefer equilibrium models. Non‐equilibrium models had high failure rates (~25%) but successful fits performed better and indicated smaller overestimates (16%) for site‐level evaluations. The Pella–Tomlinson model was most accurate (MMSY = 5.6 ± 0.60 ( SD ) tonnes/km 2 /year) and best‐fit r ‐ K relationships also aligned well with ecoregional data on K and short‐term yields. Future efforts are advised to pool site data, use conservative recruitment values (z = 0.8), and collect data across times of both rising and falling effort. Recommended methods and subsequent adjustments of the benchmark model should improve local and ecoregional scale MMSYs. The benchmark model was calibrated to estimate MMSY in fished seascapes, but to conserve species with slower life histories, we suggest modifications to limit MMSYs to between 1.8 and 3.2 tonnes/km 2 /year.