Fish kills at Neotropical hydroelectric power plants, particularly those driven by total dissolved gas (TDG) supersaturation, are difficult to diagnose and quantify due to data scarcity and rapid carcass decomposition. This study proposes a novel framework to address these challenges by (1) inferring TDG saturation from more readily available Dissolved Oxygen (DO) data, using thermodynamic principles and (2) estimating true daily mortality through a Bayesian multistate model that accounts for decomposition stages and catchability. Applied to a fish kill event at an Amazonian dam, the models showed that TDG levels can be accurately inferred from DO data and that direct carcass counts severely underestimate the magnitude of mortality events due to low detection probabilities. Furthermore, these results revealed a significant temporal lag between the actual mortality peak and carcass recovery. This integrated methodology offers a robust framework for quantifying impacts, expanding monitoring capacity, and supporting more rigorous environmental licensing processes in the Neotropical region.
Alves et al. (Sat,) studied this question.