Influence of environmental and biological factors on mercury accumulation in fish from the Atrato River Basin, Colombia

Abstract

Understanding variations in total mercury (T-Hg) levels in fish is crucial for protecting aquatic biota and human health. This article evaluates the influence of environmental factors (temperature, pH) and biological variables (feeding habits, trophic level, total length, total weight), on T-Hg concentrations in fish from the Atrato River basin, Colombia. Utilizing a robust secondary data set of 842 fish samples from 16 species collected in 2019, we conducted a comprehensive analysis of these influences. We examined differences in T-Hg accumulation rates by habitat type (pelagic, benthopelagic and demersal) and probabilistically classified species based on their feeding habits and trophic levels. Our analysis identified a hierarchy of variables influencing T-Hg levels: feeding habits > total length > estimated total weight > trophic level > water temperature > pH, with temperature being the only predictor exerting a negative influence. Together, these variables accounted for over 60% of the variability in T-Hg accumulation in fish muscle tissue. Furthermore, fish in the Atrato River exhibited differential T-Hg based on habitat type, grouping into three distinct subpopulations stratified by feeding habits and trophic levels. These findings suggest that observed T-Hg accumulation patterns are driven by the functional ecology of the organisms, phenological characteristics, metabolism, contamination patterns, biogeography, land use, and the spatial and chemical configuration of the environmental matrices of the basin. Our results emphasize the importance of understand how biological and environmental factors influence T-Hg concentrations in fish, as these factors vary across aquatic systems. This knowledge is crucial for developing effective biodiversity management strategies. While we used a machine learning approach to identify key predictors of T-Hg accumulation, we also caution against potential biases in modeling T-Hg concentrations for aquatic biota management.

Keywords: Bioaccumulation; Fish; Gold mining; Machine learning; Mercury.