Machine learning assessment of dredging impacts on the phytoplankton community on the Brazilian equatorial margin: A multivariate analysis.
Journal:
Environmental pollution (Barking, Essex : 1987)
PMID:
39800153
Abstract
Dredging in estuarine systems significantly impacts phytoplankton communities, with suspended particulate matter (SPM) and dissolved aluminum (Al) serving as indicators of disturbance intensity. This study assessed the effects of dredging in the São Marcos Estuarine Complex (SMEC), Brazil, over three distinct events (2015, 2017, 2020), involving varying sediment volumes and climatic influences. Prolonged dredging operations and increased sediment volumes led to a pronounced 43.81% reduction in species diversity, with diatoms and dinoflagellates being the most affected. Climatic variability, particularly El Niño events, exacerbated environmental dispersion, amplifying the complexity of ecosystem responses. Despite these losses, certain centric diatoms persisted, reflecting resilience mechanisms within this tropical macrotidal estuary. Machine learning approaches, specifically Random Forest (RF) models, revealed SPM and dissolved Al as critical stressors influencing species diversity. Additionally, river discharge and salinity were identified as key predictors of phytoplankton biomass. Generalized Additive Models (GAMs) confirmed that chlorophyll-a concentrations responded negatively to elevated SPM and Al levels but were less sensitive to dredging than diversity metrics. This study provides novel insights into the compounded effects of dredging and climatic variability, emphasizing the utility of RF and GAM models for predicting ecosystem responses and guiding management strategies. Recommendations include optimizing operations to reduce biodiversity impacts, minimizing sediment resuspension, and integrating predictive tools to mitigate long-term disturbances. These findings offer a data-driven framework for sustainable dredging in sensitive estuarine ecosystems.