Answers are grounded in this paper and cite their sources.
Although pore morphometrics in benthic foraminifera serve as a proxy for dissolved oxygen (DO) and nitrate (NO3-) reconstruction in oxygen minimum zones (OMZs), their applicability in marginal marine settings such as lagoons, estuaries and fjords is not well understood. The largest lagoonal system in Asia, Chilika Lagoon (Odisha, India), provides an ideal setting for this research. Here, we propose a novel approach for the quantitative reconstruction of DO using machine learning (ML) models on the pore dataset of benthic foraminifera, Ammonia parkinsoniana. ML models, random forest regression (RFR) and support vector regression (SVR) were trained on a pore dataset of living A. parkinsoniana (Rose Bengal stained) from 37 recent core-top samples. With an 87% prediction accuracy rate, the SVR model performs marginally better than the RFR (85%). Based on the core study, our results suggest that the DO level in the lagoon was 5.68–6.25 mg/L (high oxic) over the past 82 years (2023–1941 CE). The DO results from the model have been verified with multi-proxy records from foraminifera and geochemistry of the sediments. Our findings show that the main controlling factor related to DO is pore density (r = 0.84, p = < 0.001). The correlation between DO and pore parameters is consistent with the pore-proxy hypothesis. The outcomes of this research also include the development of an interactive web application, ForCAST, for the ML models. However, models trained on a larger dataset may yield higher prediction accuracy.
No takes yet. Share an insight, caveat, or question.
Subba et al. (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: