Due to anthropogenic activity, atmospheric CO₂ has reached a current concentration of 422 ppm (Dlugokencky and Tans, 2021), and could reach ~1000 ppm by the end of this century (IPCC, 2021). This is causing ocean warming and acidification. According to the intermediate emission scenario (SSP2-4.5), an increase between 2.1 to 3.5°C in the global sea surface temperature as well as a decrease of up to 0.3 pH units in the global sea surface pH or up to 0.4 in the North Atlantic Ocean are expected by the end of the century compared to 1850-1900 (Caldeira and Wickett, 2003; Orr et al., 2005; IPCC, 2021).

Although most marine teleosts have long been considered efficient acid-base regulators, coping with high CO₂ and elevated temperature may lead to downstream physiological defects (Munday et al., 2012; Heuer and Grosell, 2014; Esbaugh, 2018; Cominassi et al., 2019; Howald et al., 2019; Cohen-Rengifo et al., 2022a) including those related to sensory behaviour in some species (Clements and Hunt, 2015; Ashur et al., 2017; Tresguerres and Hamilton, 2017; Domenici et al., 2019). Several studies identified a range of behavioural traits that were severely impacted when the sensory channels of olfaction (Dixson et al., 2010; Devine et al., 2012; Devine and Munday, 2013; Munday et al., 2016; Porteus et al., 2018; Velez et al., 2019; Roggatz et al., 2022), vision (Ferrari et al., 2012; Chung et al., 2014; Steckbauer et al., 2018), audition (Simpson et al., 2011; Bignami et al., 2013; Rossi et al., 2016; Radford et al., 2021) or mechano-audition (Allan et al., 2014) were primarily stimulated under acidification alone or in combination with warming (Watson et al., 2018). These impacted traits may have ecological implications on recruitment (Munday et al., 2010; Stiasny et al., 2016), homing (Devine et al., 2012), habitat selection (Devine and Munday, 2013), learning (Chivers et al., 2014; Paula et al., 2019a), and interspecific relationships such as mutualism (Paula et al., 2019a; Paula et al., 2019b) and avoidance of a simulated predator (Ferrari et al., 2012; Allan et al., 2014; Munday et al., 2016; Steckbauer et al., 2018; Draper and Weissburg, 2019) or a real predator (Allan et al., 2013; Allan et al., 2015; Allan et al., 2017). Deleterious responses of acidification observed on olfaction (Nilsson et al., 2012; Hamilton et al., 2014; Jiahuan et al., 2018) and vision (Chung et al., 2014) result from brain dysfunction at the level of the GABAergic system where an inversion of the Cl⁻/HCO₃⁻ flux occurs across neuronal membranes. Conversely, other studies stimulating the sensory channels of olfaction (Manciocco et al., 2015; Sundin and Jutfelt, 2016; Sundin et al., 2017; Clark et al., 2020a), vision (Jarrold et al., 2020) and mechano-audition (Manciocco et al., 2015; Näslund et al., 2015; Nasuchon et al., 2016; Jarrold and Munday, 2018) under acidification (Näslund et al., 2015; Sundin and Jutfelt, 2016; Sundin et al., 2017; Clark et al., 2020a; Jarrold et al., 2020) and warming (Manciocco et al., 2015) either independently or interactively (Nasuchon et al., 2016; Jarrold and Munday, 2018) revealed no impacts, minor impacts or both. This highlights the complexity of interpretation as the impacts are trait-specific, even in closely related traits such as lateralization, an indicator of brain function. For instance, in both Gobiusculus flavescens (Sundin and Jutfelt, 2018) and Ctenolabrus rupestris (Sundin and Jutfelt, 2016), relative lateralisation was impacted by acidification whereas absolute lateralisation was not. The opposite occurred in Acanthochromis polyacanthus (Jarrold and Munday, 2018). In Ctenolabrus rupestris, even if lateralisation was impacted, most activity metrics and olfactory choice metrics were not impacted by acidification (Sundin and Jutfelt, 2016). A similar pattern was observed in Gasterosteus aculeatus exposed to acidification, as altered absolute lateralization was not accompanied by altered escape response (Näslund et al., 2015). Escape response to predator olfactory cues also varied within species of the same genus and showed overall conflicting results across studies (Dixson et al., 2010; Munday et al., 2010; Clark et al., 2020a). All these results evidence an underlying variation probably related to differences in the experimental conditions, system and methodology that may partially account for conflicting results. Furthermore, this puts forward the importance of standardised methods, data transparency, replicability and reproducibility (Clark et al., 2020a; Clark et al., 2020b; Munday et al., 2020; Williamson et al., 2020).

Sensory-mediated behaviours regulate inter-specific interactions including predator-prey interactions. Particularly, the escape response is considered a fundamental trait in neurobiology because when fish are confronted with a predator cue, they are able to display a wide array of behavioural and cognitive traits including decision-making (Domenici and Hale, 2019). While temperature is the primary environmental parameter governing sensory-mediated behaviours and interactions, littles is known concerning the effects of warming alone or its interactive effects with ocean acidification, notably on temperate pelagic fish of ecological and socio-economic importance (Domenici et al., 2014; Draper and Weissburg, 2019). Yet, the study of independent and interactive effects of warming and acidification provide understanding on how stressors can operate (i.e. additively, synergistically or antagonistically) in a near-future ocean where warming and acidification are expected to act simultaneously (Vinebrooke R et al., 2004; Nowicki et al., 2012; Pimentel et al., 2016; Allan et al., 2017; Domenici and Hale, 2019).

The European sea bass (Dicentrarchus labrax) is a large pelagic predatory fish of ecological relevance and has an established aquaculture value and a developing capture value in the north-eastern Atlantic Ocean (Vandeputte et al., 2019; FAO Yearbook, 2021). It is found in highly hydrodynamic habitats from Morocco though the Mediterranean and the Northeast Atlantic up to Norway (Jennings and Pawson, 1992). D. labrax juveniles inhabit nursery areas for their first four to five years (Jennings and Pawson, 1992). Habitat connectivity and complexity and high food availability make these areas ideal for growth and recruitment (Pawson et al., 2007; Sheaves et al., 2015) even if piscivorous fish and birds exert a certain degree of predation pressure (Baker and Sheaves, 2006; Sheaves et al., 2015). Juveniles must be able to recognize predators using their sensory systems in order to survive since overflying birds can suddenly undertake a swoop attack. For example, Gasterosteus aculeatus dart away and abruptly remain still to avoid being seen by an avian predator; a response that is believed to be innate, as it was observed in predator-naïve laboratory fish (Giles, 1984).

If warming and acidification affect the sensory ability of juvenile fish to detect a predator, the risk of being predated might be higher and consequently, the movement of fish from nursery areas to adjacent adult habitats may be reduced. Therefore, community-level impacts might be expected with carry over socio-economic implications. Understanding how warming and acidification may impact sensory behaviour-mediated predator-prey interactions is therefore of utmost importance. The aim of this study was to evaluate whether exposing juvenile D. labrax to independent and interactive warming and acidification for a mid-duration exposure (60-92 days) could alter their sensory behaviour, which is a mediator of predator detection and escape response. The sensory channels of vision and mechano-audition were assessed in separate behavioural tests, and for the first time both were evaluated under independent and combined warming and acidification. We used automated video tracking to calculate kinematic swimming variables and evaluate escape response.

The present study highlights a robust perception and escape response to independent and combined effects of warming and acidification in Dicentrarchus labrax in the face of danger, such as that mimicked by either the shadow of an overflying bird (visual cue) or the attack of a swooping bird (mechano-acoustic cue). Behavioural resilience to OA has been previously observed in ~60 days old juveniles D. labrax (Duteil et al., 2016) and the present study extends escape response resilience to late juveniles of 9-10 months. Our study also highlights a large variability in individual responses observed before and after either cue.

Some of the observed variability may arise from technical variability (Hecksteden et al., 2015) such as the differences in acclimation time in the bucket, fish being familiar with the arena upon their second challenge, different exposure duration to stressors or short acclimation period in the arenas. Regarding acclimation duration in D. labrax of similar body lengths when exposed to a mechano-acoustic challenge, it seems that if we had granted a longer acclimation duration before the trials, variability would not necessarily have been reduced. Indeed, 15 min acclimation was accompanied by 75 to 90% of reactivity (Benhaïm et al., 2012) whereas 2 h acclimation was accompanied by 50% of reactivity (Millot et al., 2009). In addition, although we conducted the experiment under apparently similar conditions, the environment of the experimental infrastructure changes every day, and it was technically difficult to identified every source of variability. We should therefore not neglect the importance of unidentified background noise as this can modulate fish environment and translate into differences in fish behaviour on a daily basis. Another portion of the observed variability can be explained by intra- or interindividual variability (Hecksteden et al., 2015) whether it is induced by the environmental and technical parameters or by genetic endowment. Within-subject variability manifested as single fish showing different activities that ranged from immobility, sustained swimming to erratic fast one-off movements. These activities varied in frequency, duration and layout between fish on a daily basis and likely contribute to greater between-subject variability.

We included in our model both technical variability as the random factors bucket acclimation and floating cage, and interindividual variability as the categorical factor rapidity, assuming that pre-cue swimming velocity will modulate post-cue swimming velocity. Contrary to our assumption, even slow fish were able to swim as fast as fast fish after either cue, suggesting that regardless of initial responsiveness, fish were able to escape from dangerous situations where their survival could be threatened. Although rapidity improved model fit, there is yet a part of the variance that is not fully explained by our models, and together with the large behavioural variability, perhaps precluded statistical identification of a potential impact of treatment. Indeed, round or treatment had no statistical impact on the cumulative distance travelled or the mean swimming velocity indicating that differences in 32-day exposure to warming and/or ocean acidification had no impact on the escape response to simulated predator cues in juvenile D. labrax.

Concerning the visual test, swimming velocity was significantly different during than before the cue, indicating that the shadow elicited a response. In addition, swimming velocity did not vary between the dark and post-cue periods denoting a sustained shadow-induced response that can last a few minutes after the danger has passed. Rather than a strong escape response, the shadow cue elicited the maintenance of a precautionary response where fish swam conservatively faster, regardless of rapidity and treatment. The few studies addressing the response to a visual cue under acidification revealed negative impacts. In early juvenile Pomacentrus amboinensis, anti-predator response was reduced after 6 days of exposure (Ferrari et al., 2012), while in juvenile Acanthochromis polyancanthus retinal flickering was reduced after 7-day exposure (Chung et al., 2014), though it is unknown if this can alter fish ability to react to visual cues. On the other hand, the only study addressing both warming and acidification found no impact on lateralization and visual response in early juvenile Seriola lalandi, though vision was only tested under acidification because fish exposed to warming were too fast (Jarrold et al., 2020).

Concerning the mechano-acoustic test, swimming velocity significantly increased after the swoop attack in fish of every treatment. After the cue, slow fish swam slower that fast fish, yet their swimming velocities were still highly indicative of a strong escape response. However, it is necessary to experimentally validate whether the lower swimming velocity of slow fish leads to reduced escape success. Our results suggest a robust response to the predator swoop attack under warming and acidification. This is consistent with related work in which late juvenile D. labrax exposed 21 days to +4°C (Manciocco et al., 2015) and adult Gasterosteus aculeatus exposed 20 days to acidification (Näslund et al., 2015) showed unaltered overall anti-predator response. However, signs of neuronal dysfunction were found in both species and a longer latency to escape was observed in D. labrax. Compared to these studies, the absence of effect we observed might partially be explained by a longer exposure duration to both stressors that might allow our fish to acclimate over 3 months. In adult Engraulis japonicus, 1-month exposure to both warming and acidification, did not impact a range of escape response traits (Nasuchon et al., 2016). However, longer transgenerational exposure to acidification in the coral reef species Amphiprion melanopus (15 dph) restore only a few escape response traits that were impacted by acidification (Allan et al., 2014). Interestingly, multi-stressor experiments performed on early life stages revealed that a range of escape traits are more impacted by warming than by acidification in 25-dph Seriola lalandi (Watson et al., 2018) and 28-49-dph Acanthochromis polyacanthus (Jarrold and Munday, 2018). On the other hand, the auditory sensory channel alone, is also sensitive in early life stages exposed to acidification, as observed in 17-20-dph Amphiprion percula that did not avoid recordings of daytime predator-rich reef noise (Simpson et al., 2011) and in 42-dph Chrysophyrs auratus that showed diminished hearing sensitivity (Radford et al., 2021).

Escape response is known to vary according to the perceived level of threat (Domenici and Batty, 1997; Domenici and Hale, 2019). We observed that perception of the predator cue elicited different escape responses in fish according to the sensory channel stimulated. Stimulation of the mechano-acoustic sensory channel led to exacerbated distance travelled and swimming velocity, indicating a stronger escape behaviour after the swoop attack than after the shadow. Likewise, the proportion of CT fish travelling distances greater than 25 bl was higher in the mechano-acoustic tests (72%), than in the visual tests (55%), and this was also valid in the other treatments (Table S1). This probably reflects a higher sensitivity of the mechano-acoustic sensory channel related to the higher threat to survival of a direct bird attack compared to that of an overflying bird shadow. Also, fish exposed to the visual cue probably slowed down to reduce the chances of being perceived by a bird, a behaviour observed in both wild and captive three-spined sticklebacks (Giles, 1984). Furthermore, the mechano-acoustic test stimulated simultaneously three sensory channels: mechanosensation and audition, but also vision, and fish responses were likely stronger than if the sensory channels had been stimulated separately. It was not intended to separate these sensory channels, since during a swoop attack, fish are able to hear the bird hitting the water, feel the vibrations of the hit, and see the bird in the water. Besides, it is technically difficult to evaluate separately mechanosensation or audition, because sound, in addition to being detected by the inner ears, is also detected by the same organ that detects vibrations, the lateral line (Braun and Sand, 2014).

In the wild, D. labrax juveniles aggregate in schools; a behaviour known to benefit anti-predator performance in other species, and which is further enhanced by the level of social recognition between members of a school (Domenici and Batty, 1997; Nadler et al., 2021). Although escape response varied whether measured in a single-fish or group situation (Domenici and Batty, 1997; Domenici and Hale, 2019), our single-fish approach puts forward the individual responsiveness to a danger which is ecologically relevant for schooling dynamics. Indeed, anti-predator responses of the entire school is governed by individual responses of the leader fish that initiates the escape (Marras and Domenici, 2013). Individual responses are therefore critical for the success of anti-predator response and consequently for survival and fitness (Fuiman et al., 2006). Thus, the variation of responses between individuals is crucial to understand different ways of responding to a predator and drives natural selection and population persistence across generations.

Predator detection in the wild is mediated by the use of multiple senses, so that if a given sensory channel is compromised by an environmental stressor, another less sensitive sensory channel can mitigate negative effects through trade-offs or sensory redundancy (Lönnstedt et al., 2013). However, in experiments in which all the sensory channels of both predatory fish (Pseudochromis fuscus) and prey (Pomacentrus wardi) were stimulated simultaneously under warming and acidification (Allan et al., 2015; Allan et al., 2017), a range of kinematic escape traits were impacted in a greater way by warming than by acidification, suggesting that P. wardi was not able to compensate (Allan et al., 2017). These experiments highlight that predator’s behaviour may also be impacted by climate change which can regulate the success of prey escape response in complex ways. Even if warming and acidification did not directly impact escape response in juvenile D. labrax after stimulation of the visual and mechano-auditory sensory channels, it in unknown how D. labrax would react in a real-predator situation where all the senses are stimulated simultaneously, and if predators are also exposed to the stressors.

As for the near-future ocean, facilitation of cue transmission through the natural environment, also depends on the quality of the transmission medium in which predator-prey sensory processes operate. For example, warming-induced algal blooms can reduce prey visual cues through increased turbidity that impeds predation (Strod et al., 2008), while habitat fragmentation may enhance the transduction of visual cues and facilitates predation (McCormick and Löonnstedt, 2013). The impacts of warming and acidification on the sensory transduction pathway including cue emission, perception and processing by fish may be modulated by the interaction with other natural or anthropogenic factors as well as by the organism adaptive response after transgenerational acclimation (Allan et al., 2014). Whether this may promote resiliency or plasticity of predator-prey interactions, is still an open question that deserves ideally head-to-head predator-prey challenges in a climate change mesocosm using both single-fish and school approach.

The datasets generated and analysed in this article are available in the SEANOE public repository https://doi.org/10.17882/87244.

The animal study was approved by Commission nationale pour la protection des animaux utilisés à des fins scientifiques (CNEA). The study was conducted in accordance with the local legislation and institutional requirements (authorization number APAFIS#10745).

All authors conceived the scientific questions. MC-R & M-LB designed the behavioural experiments. MC-R conditioned fish to warming and acidification, performed the behavioural tests, analysed the videos and conceived and run the statistical models. M-LB & DM provided valuable advice and supervision. All authors contributed significantly to the revision of the manuscript and approved the submitted version.