Harmful algal blooms of the dinoflagellate Alexandrium catenella occur nearly every year in the Gulf of Maine and cause shellfish closures due to the threat of paralytic shellfish poisoning. A 25-year hindcast of A. catenella blooms is evaluated for model skill and diagnosed for sources of interannual variability. The physical-biological model builds on previous studies, but for the first time we link simulated cell concentrations with shellfish toxicity at coastal shellfish monitoring stations using a toxin uptake-and-depuration submodel. Comparisons of modeled and observed toxicity quantify skill within and among years. Results show consistency between the model and observations in the years with the greatest toxicity and most extensive closures. Calculating toxicity based on uptake and depuration outperforms toxicity based on cell concentrations alone, and model simulations have greater skill than long-term climatology. The influence of benthic cysts of A. catenella, which initiate the blooms, is evaluated by comparing simulations using observed interannual variability in cyst abundance versus the long-term mean cyst distribution. The differences in model skill for coastal toxicity are negligible, suggesting that cysts are not the dominant source of interannual variability in the long-term record. Water temperature increases growth rates and overall bloom magnitude, but toxicity is greatest in years when downwelling-favorable winds transport cells toward the coast. The wind conditions and cyst abundance are just two of many potential factors (including growth conditions) affecting blooms and coastal toxicity, and these results highlight the importance of time series observations paired with mechanistic models to diagnose the underlying processes.
