Using self-organizing maps to identify the drivers of Florida red tide (Karenia brevis) dynamics (in preparation)

Abstract

Harmful algal bloom (HAB) states such as initiation, proliferation, and termination occur across highly variable spatial and temporal scales, complicating efforts to link bloom dynamics to environmental drivers. In this study, we present a novel approach to identify state-dependent drivers of Karenia brevis (red tide) blooms, which occur nearly annually along the southwest Florida coast. Percent coverage of K. brevis alongside potential environmental drivers including both contemporaneous and antecedent conditions, were clustered simultaneously using a self-organizing map (SOM) to define distinct temporal environmental regimes. By reducing the dimensionality of this multivariate, time-lagged dataset, we delineate seven red tide states: no bloom, bloom initiation, expanding medium bloom, expanding widespread bloom, contracting widespread bloom, contracting medium bloom, and bloom termination. Each state is characterized by a unique combination of environmental conditions. Our results demonstrate that the influence of managed freshwater discharges is state dependent and must be evaluated within the broader context of the magnitude and trajectory of co-occurring environmental conditions. This methodology provides a new framework for identifying, interpreting, and potentially forecasting HABs and broader ecological dynamics using temporally resolved environmental states.