Recent scientific evidence for fish sentience has stressed the need for novel sentience-based detection tools of fish welfare impairment in commercial farms. In order to mimic a wellcharacterised stress situation, rainbow trout (Oncorhynchus mykiss) were exposed to poor water quality (hypoxia combined with high ammonia level) for three weeks (stressed group, S) and compared to a non-stressed control group (NS). After a return to water quality standard, emotional responses were assessed in fish subjected to two potentially threatening situations: (i) social isolation in a novel environment and (ii) human presence. In addition, we used an appetitive-conditioning paradigm to determine whether previous chronic deterioration of water quality disturbs cognitive abilities in fish. Spontaneous behaviour in the tanks was also recorded during the environmental challenge as a reference for fish activity. We observed that in S fish, plasma cortisol levels were increased before and after social isolation in a novel environment compared to the plasma cortisol levels in the NS group, despite the absence of a behavioural difference between the two groups. Under deteriorated water quality, fish locomotor activity was globally reduced and this reduction was correlated to increased

Recent scientific evidence for fish sentience has stressed the need for novel sentience-based detection tools of fish welfare impairment in commercial farms. In order to mimic a wellcharacterised stress situation, rainbow trout (Oncorhynchus mykiss) were exposed to poor water quality (hypoxia combined with high ammonia level) for three weeks (stressed group, S) and compared to a non-stressed control group (NS). After a return to water quality standard, emotional responses were assessed in fish subjected to two potentially threatening situations: (i) social isolation in a novel environment and (ii) human presence. In addition, we used an appetitive-conditioning paradigm to determine whether previous chronic deterioration of water quality disturbs cognitive abilities in fish. Spontaneous behaviour in the tanks was also recorded during the environmental challenge as a reference for fish activity. We observed that in S fish, plasma cortisol levels were increased before and after social isolation in a novel environment compared to the plasma cortisol levels in the NS group, despite the absence of a behavioural difference between the two groups. Under deteriorated water quality, fish locomotor activity was globally reduced and this reduction was correlated to increased 2 shoaling behaviour. Farmers can use these first behavioural modifications as a sentinel detector for fish welfare impairment. More importantly, we demonstrated that reactivity to a human presence in a home-environment and food-anticipatory behaviour were both inhibited in the S group. We consider that these two sentience-based tests are highly relevant for fish welfare assessment at the group level and are easy to use in the aquaculture industry.