Early warning through video monitoring: Dissolved hydrogen sulphide (H2S) affects Atlantic salmon swimming behavior in recirculating aquaculture systems

Abstract

Hydrogen sulphide (H2S) poses a major threat in marine land-based recirculating aquaculture systems (RAS) leading to acute mortality in sensitive fish species such as Atlantic salmon (Salmo salar). To date, little is known about the effects of H2S on the physiology and behavior of the species. The present study analyzed Atlantic salmon swimming behavior in response to H2S in a controlled trial. The setup consisted of two Recirculating Aquaculture Systems (RAS) in parallel. The control RAS comprised of one fish tank (800 L, 10 kg fish/m3 (≈ 70 fish)), while the exposure RAS included two fish tanks (800 L; 10 and 30 kg fish/ m3 (≈ 70 and 200 fish)). Swimming behavior was monitored via a submerged custom-built stereo camera system and an overhead surveillance camera. Fish (smolt, ≈ 114 g) were exposed once a day for 10 consecutive days to increasing H2S concentrations, from ≈ 1- up to ≈ 68 μg/L (2 μM). Continuous measurements of dissolved H2S, O2 and CO2 were taken using a real-time monitoring system. Three swimming parameters were extrapolated from video recordings using machine learning algorithms: i) speed, ii) pattern (representing whether the fish swim in a straight or zigzagging direction) and iii) dispersion (indicative of schooling behavior). The results showed that fish reacted rapidly to H2S, with a stress response characterized by faster swimming speed, erratic pattern, and loss of schooling behavior. The response was concentration-dependent, increasing linearly up to 30–40 μg/L, above which a clear threshold was observed. Notably, concentrations around 40–50 μg/L, induced significantly greater behavioral changes compared to lower concentrations, and further increases in H2S did not lead to additional changes in behavior. Swimming parameters quickly returned to basal levels, comparable to the one's prior exposure, once H2S was no longer present in the water. This study provides new insights on the sensitivity of Atlantic salmon to acute H2S exposure and highlights the potential behind the use of machine vision as an early warning tool for poor water quality in RAS.