Reliability and contradictions of indicators of

4.2.4 Reliability and contradictions of indicators of consciousness

high variation in signal amplitude is that signals from muscle activity (electromyography, EMG) may also be picked up on the EEG. It is thus advantageous to analyze visual evoked responses (VERs) as interference from muscular activity is minimized. In study IV, the method to induce VERs was refined even further as a purpose-built LED-light that delivered a much shorter light stimulus was used to evoke responses.

However, the results of study IV revealed that also two of the most commonly used EEG-indicators of unconsciousness in fish; absence of VERs and presence of an epileptic-like seizure, can result in conflicting conclusions. A 1 s stun with sufficient current density and electric field strength induced an epileptic-like insult visible as abnormal EEG-voltage fluctuations, indicative of a hyper-stimulated brain. The EEG during the insults recorded in study IV resembles previously published images and descriptions of epileptic-like insults in earlier studies (Daskalova et al., 2015;

Lambooij et al., 2008; Lambooij et al., 2012; Lambooij et al., 2008;

Lambooij et al., 2010). Other research groups have instead used VERs to determine unconsciousness in fish, e.g. (Brijs et al., 2020; Jung-Schroers et al., 2020; Kestin et al., 1991; Retter et al., 2018; Robb et al., 2000). It was therefore both surprising and concerning to observe recovery of VERs before the epileptic-like insult had ended, i.e. VERs were present during the period when the brain was assumed to be hyper-stimulated and the fish unconscious (see Fig. 4 in study IV). The conflict between these indicators is puzzling and something that has, to the best of my knowledge, not previously been reported in fish. These findings would then suggest that either i) what was observed on the EEG was in fact not a hyper-stimulated brain or ii), absence of VERs is a too strict criteria to be used when determining the onset of unconsciousness.

If the first alternative is true, it would mean that the observed epileptic-like insult is not what in humans is referred to as a grand mal seizure but instead may be an absence seizure (formerly known as a petit mal seizure).

The difference between these two types of seizures is, based on studies in humans, that the a grand mal seizure is indicative of unconsciousness but that unconsciousness may be very short lasting during an absence seizure (Bancaud et al., 1981; Panayiotopoulos, 2008; Sadleir et al., 2009). A similar phenomenon has been reported in chickens following electrical stunning,

where the authors propose that a petit mal seizure can occur following electrical stunning (Gregory & Wotton, 1987). The absence of an iso-electric phase and the rapid recovery of swimming following the tonic-clonic phase raises further questions regarding the actual nature of the epileptic-like insult that was observed in the electrically stunned rainbow trout. If what was observed in the rainbow trout in study IV in fact was not a grand mal seizure, it raises concerns also for other fish species, as the epileptic-like seizure I observed did in no obvious way differ from those described in previous studies, where they have been used to determine the onset of unconsciousness in fish (Daskalova et al., 2015; Lambooij et al., 2008;

Lambooij et al., 2012; Lambooij et al., 2008; Lambooij et al., 2010).

The alternative explanation is that absence of VERs is an overly conservative indicator when determining the onset of unconsciousness and may be present even if the fish is unaware of its surroundings. This possibility has previously been mentioned by Kestin et al. (1995). A second surprising finding in study IV, when the fish were exposed to the electric field for a longer time (15-60 s), further complicate things. With such long stun application times, no epileptic-like insult was observed on the EEG.

Instead, VERs were lost immediately after the stun but often returned within the first minute, only to disappeared again after approximately 1 min (Fig.

18). The same phenomenon was also observed in an unpublished study where I measured brain activity in rainbow trout using a head-to-body dry electrical stunning system. In those experiments, similar patterns were observed in some individuals, with VERs appearing during the first min post stun, only to disappear shortly after and the fish later died. In other individuals, VERs reappeared again after a few min with subsequent recovery of equilibrium and swimming movements (unpublished observations). The consequences of this phenomenon for fish welfare remains unknown, but it is possible that the short and transient period of VERs is not necessarily an indicator of consciousness. If the fish is insensible to stress, pain and fear during this transient period with VERs, the conclusions drawn in study IV may need to be revised.

Both explanations to the unexpected results of study IV highlight some of the knowledge gaps that currently prevents us from safeguarding the welfare of fish during time of slaughter. The million dollar question is still

what indicator provides the most accurate representation of unconsciousness in fish. Even vocal animals such as mammals cannot be asked what they feel, so we have to rely on assessment of physiological and behavioral responses.

This may not be an issue when percussive stunning is used as the physical disruption or destruction of the brain is both instant and irreversible and VERs can thus safely be used to determine unconsciousness. However, using slow (e.g. CO2) or reversible (e.g. electrical) methods require a reliable indicator to determine the moment when consciousness is lost or recovered.

Absence of VERs can be used to determine unconsciousness in a laboratory for most, if not all, stunning methods, but the search for a reliable indicator that is also practical to use in the abattoir must continue. Nevertheless, the results from study III and IV clearly show a high level of complexity between self-initiated behaviours, reflexes and neurophysiological indicators following stunning, and that potential correlation between indicators seems to vary depending on the stunning method and potentially also among species.

The findings presented here provide novel insights on how to monitor and assess responses to some of the issues that pose a serious threat to the welfare of the billions of farmed fish that are reared and slaughter in aquaculture each year.

The conclusions drawn from study I and II show that physiological responses can provide quantitative information on how fish in aquaculture perceive their environment and can be used to identify welfare hazards. Heart rate bio-loggers are useful tools to investigate stress in fish and, if used correctly, they can be used without any major negative effects on the health and welfare of the experimental animals. Furthermore, it is emphasized that stress sensitivity and indicators of secondary stress responses may vary even between relatively closely related fish species, which must be considered when evaluating stress responses of different species in their respective aquaculture settings. For many other fish species beyond salmonids, little is known about how abiotic and anthropogenic factors affect the health and wellbeing of the individual, which is crucial knowledge to maintain good fish welfare. Thus, the introduction of new farming species to the expanding aquaculture industry will require extensive evaluation of rearing conditions and other farm-related routines to ensure that they are consistent with needs of that specific species and does not lead to detrimental chronic stress. It may sound like a cliché, but more research is necessary to safeguard the welfare for all the farmed fish species. To date, most research has focused on economically important species such as salmonids. Nonetheless, my findings show that there are existing knowledge gaps regarding welfare also for the well-investigated rainbow trout.