Interannual variability in the timing of New England shellfish toxicity and relationships to environmental forcing

Routine monitoring along the coast of the Gulf of Maine (GoM) reveals shellfish toxicity nearly every summer, but at varying times, locations, and magnitudes. The responsible toxin is known to be produced by the dinoflagellate Alexandrium fundyense, yet there is little apparent association between Alexandrium abundance and shellfish toxicity. One possibility is that toxic cells are persistent in offshore areas and variability in shellfish toxicity is caused not by changes in overall abundance, but rather by variability in transport processes. Measurements of offshore Alexandrium biomass are scarce, so we bypass cell abundance as an explanatory variable and focus instead on the relations between shellfish toxicity and concurrent metrics of GoM meteorology, hydrology, and oceanography. While this yields over two decades (1985-2005) of data representing a variety of interannual conditions, the toxicity data are gappy in spatial and temporal coverage. We address this through a combination of parametric curve fitting and hierarchical cluster analysis to reveal eight archetypical modes of seasonal toxicity timing. Groups of locations are then formed that have similar interannual patterns in these archetypes. Finally, the interannual patterns within each group are related to available environmental metrics using classification trees. Results indicate that a weak cross-shore sea surface temperature (SST) gradient in the summer is the strongest correlate of shellfish toxicity, likely by signifying a hydrological connection between offshore Alexandrium populations and near-shore shellfish beds. High cumulative downwelling wind strength early in the season is revealed as a precursor consistent with this mechanism. Although previous studies suggest that alongshore transport is important in moving Alexandrium from the eastern to western GoM, alongshore SST gradient is not an important correlate of toxicity in our study. We conclude by discussing the implications of our results for designing efficient and effective shellfish monitoring programs along the GoM coast.