Marine Pollution Bulletin increasingly applies machine learning and explainable AI to pollutant and shellfish poisoning risk, exemplified by PCA-based source apportionment and SHAP-based feature attribution. However, linear PCA may misrepresent structure in inherently nonlinear environmental data, and existing studies often treat model-derived feature importances as evidence of true associations without assessing consistency or dose-response relationships. This paper clarifies that supervised models possess two distinct accuracies: prediction and feature importance, and only prediction can be validated against ground truth. Using a Basque coastal dataset (8195 instances, 14 features) with chlorophyll-a as a proxy for paralytic shellfish poisoning risk, we introduce a leave-top1-out procedure to test ranking stability. Random Forest and XGBoost with and without SHAP show pronounced instability, indicating biased, model-dependent importances. In contrast, unsupervised and non-target-prediction methods yield perfectly stable rankings while matching or exceeding supervised performance, supporting routine stability, consistency, dose-response, and linearity checks in environmental ML studies.
This study assessed the proximate composition and heavy metal (HM) contamination of the jellyfish Rhizostoma octopus collected from the Cox's Bazar coast, Bangladesh. To minimize post-mortem alterations, three specimens were collected alive from fishermen's nets near the shoreline. HM concentrations were determined using atomic absorption spectroscopy. The mean concentrations of metals and minerals decreased in the order: Na > Mg > K > Ca > Fe > Zn > Mn > Cu > Pb > Ni > Co > Cd > Cr > As. Proximate composition analysis (sun-dried weight basis) showed that ash was predominant (46.84-54.58%), followed by protein (25.79-27.76%), moisture (16.47-18.73%), carbohydrate (0.21-9.74%), and fat (0.46-0.79%). Principal component analysis identified two major components explaining most of the variance in metal contents, while hierarchical cluster analysis revealed similar accumulation patterns among metals. Health risk assessment suggested that individual target hazard quotient values for all metals were below 1, and the overall hazard index was also below 1, suggesting no significant non-carcinogenic risk. However, carcinogenic risk estimates were higher for children than adults, with Cd was estimated to exceed the USEPA acceptable threshold for children, showing a value of 2.48E-04. Despite the advantage of live sampling, the small sample size and limited spatial and temporal coverage restrict the representativeness of the results. Therefore, larger-scale and long-term studies are required to validate the safety of R. octopus for human consumption and to support its potential as a sustainable food resource within the blue economy.
This study presents a 3D hydrodynamic simulation of coastal circulation coupled with copper advection emanating from Quintero Bay, in the Valparaíso region. The bay is home to an important industrial complex and has a history of heavy metal contamination due to a copper smelter located in its northern sector. This study aims to simulate the distribution of contamination from existing sources and identify potential risk scenarios associated with copper exposure in water and sediment. A simple data assimilation method has been developed to capture elements of coastal wind not well represented by existing models used for wind forcing. The simulation period represents an austral summer, and several circulation patterns have been identified in response to external forcings. Two different point sources are considered, one within the bay (source 1), and the other just outside the south side of the bay (source 2). Simulations show advection of copper towards coastal areas north of the bay in trace concentrations from both sources, with instances at certain points within the bay when recommended thresholds for chronic and acute exposure are surpassed. Source 1 shows advection predominantly leaving the north side of the bay in surface waters, while source 2 shows advection in bottom waters periodically entering the south side of the bay. These results are consistent with existing local studies of copper concentrations in the water and sediment.

