122 / 2024-09-09 09:01:37
Unraveling the Dynamics of Alexandrium minutum Blooms: Integrating Metabarcoding and Machine Learning in a semi-enclosed tropical costal lagoon (Malaysia, South China Sea)
Alexandrium, Harmful Algal Blooms (HABs), Metabarcoding, machine learning, plakton community
Session 12 - Alleviating the impact of emerging Harmful Algal Blooms (HABs) to coastal ecosystems and seafood safety for a sustainable and healthy Ocean
Abstract Accepted
The species Alexandrium minutum, a key species responsible for Paralytic Shellfish Poisoning (PSP), poses significant threats to human health and socio-economic activities worldwide. Despite the critical impacts, studies focusing on early detection and prediction of A. minutum blooms remain scarce. This study utilized metabarcoding to investigate plankton community shifts in Geting Lagoon between 2018 and 2020. Amplicon sequence variants (ASVs) of dominant phytoplankton and zooplankton, combined with environmental parameters (dissolved oxygen, temperature, salinity, pH, nitrate, nitrite, ammonia, phosphate, silica, and chlorophyll-a), were analyzed through a machine learning stacking model (XGBoost and Random Forest) to assess the relative influence of bottom-up and top-down mechanisms on A. minutum bloom dynamics. A total of 206 micro-phytoplankton species were identified, with A. minutum, Chaetoceros sp., Skeletonema subsalsum, Chaetoceros diadema, Unruhdinium minimum, and Cyclotella striata emerging as dominant ASVs throughout the study period. Alexandrium minutum blooms occurred between August and October 2020, with ASV reads ranging from 457 to 52,437. Additionally, 22 orders of zooplankton were detected, with bivalvia, copepods, gastropods, and rotifers dominating. Machine learning predictions (MSE = 3.96, R² = 73.6%, cross-validation R² = 85.9%) highlighted the importance of biotic factors, such as Cyclotella striata, Skeletonema ardens, and rotifers, suggesting that top-down controls—via competition and grazing—play a pivotal role in regulating A. minutum abundance. However, bottom-up influences, including nutrient levels and salinity, also contributed to the bloom dynamics, indicating that both top-down and bottom-up processes are at play. These findings emphasize the multifaceted nature of A. minutum bloom dynamics and demonstrate the potential of metabarcoding and machine learning for early detection and prediction, which could be crucial for management and mitigation efforts.