Model predicts coastal paralytic shellfish toxicity in mussels from in situ phytoplankton concentrations.

Harmful algal blooms (HABs) of Alexandrium catenella occur annually in the Gulf of Maine, leading to accumulation of paralytic shellfish toxins (PSTs) in bivalve mollusks and closure of harvesting areas. State monitoring programs in the region rely on weekly shellfish sampling and toxin analysis, requiring extensive effort with limited temporal resolution. With the development of autonomous in situ detection devices such as the Environmental Sample Processor (ESP), this study aimed to model shellfish toxicity using ESP-derived A. catenella counts to complement traditional monitoring programs using mussels (Mytilus edulis). We developed a one-compartment model of PST accumulation using A. catenella concentrations from nearshore ESP deployments. Multi-year and multi-site data were used to test four approaches for estimating mussel toxin uptake and depuration parameters, that is, local, global, or fixed literature values. When compared with direct PST measurements in nearby shellfish, the models achieved high correlations (R 2 = 0.69-0.90), with no false negatives. Model fit varied mainly with toxin uptake parameters and was influenced by Alexandrium concentrations. The impact of other environmental and biological factors on uptake and depuration parameters remained elusive, likely due to the mussels' physiological plasticity, spatial or hydrographic separation between ESP and toxicity measurements, and microhabitat conditions. This ESP-based model offers a simple, effective tool for translating offshore, automated cell counts into daily shellfish toxicity estimates and is promising for improvement of monitoring capacity in remote or under-sampled areas. Incorporation of this approach into physical-biological models and application in early warning systems with other biosensors remain to be tested.