Methylmercury, cadmium and plastron length topologically modulate multimetallic bioaccumulation in Colombian Slider (Trachemys callirostris Gray, 1856): A Bayesian structure learning.

Heavy metal bioaccumulation in chelonians remains relatively under-researched. Due to their high sensitivity to such type of pollution, chelonians represent ideal bioindicators for assessing aquatic ecosystems health, especially in zones with historical records of multimetallic pollution. Here, we quantified total mercury (THg), methylmercury (MeHg), lead (Pb), cadmium (Cd), copper (Cu), and zinc (Zn) in muscle and liver of Trachemys callirostris and examined significant differences (p < 0.05) between their concentrations in both tissues. As well, their partial-conditional bioaccumulative correlations along with their uncertainty in relation to carapace length (CL), plastron length (PL) and turtles' weight were determined through a Bayesian Undirected Graphical Model (BUGM). The topological co-occurrence in the inferred correlations was modelled using network analysis through a weighted and undirected graph. Results reveal that median THg concentrations in muscle and liver (157.47 µg/kg and 331.78 µg/kg) were within the referenced hematotoxicity range (100 µg/kg - 350 µg/kg). MeHg was the only pollutant that showed no significant inter-tissue differences. The BUGM inferred 13 key positive associations (12.38 % of total) with a high posterior probability of correlation structure (S) (p = 0.654). Network analysis revealed a highly modular, heterogeneous, sparse and disassortative topology, where MeHg and Cd in muscle with PL acted as the most central and prominent nodes, bridging five well-differentiated communities. This work represents the first Bayesian approach applied to model structural bioaccumulation networks in chelonians, offering a novel framework to approach the nature and uncertainty of complex interactions between pollutants and morphometry in water ecosystems.