Search

Farmed Atlantic salmon (Salmo salar) welfare and performance can be strongly influenced by stress episodes caused by handling during farming practices. To better understand the changes occurring after an acute stress response, we exposed a group of Atlantic salmon parr to an acute stressor, which involved netting and transferring fish to several new holding tanks. We describe a time-course response to stress by sampling parr in groups before (T0) and 10, 20, 30, 45, 60, 120, 240, 300, and 330 min post-stress. A subgroup of fish was also subjected to the same stressor for a second time to assess their capacity to respond to the same challenge again within a short timeframe (ReStressed). Fish plasma was assessed for adrenocorticotropic hormone (ACTH), cortisol, and ions levels. Mucus cortisol levels were analyzed and compared with the plasma cortisol levels. At 5 selected time points (T0, 60, 90, 120, 240, and ReStressed), we compared the head kidney transcriptome profile of 10 fish per time point. The considerably delayed increase of ACTH in the plasma (60 min post-stress), and the earlier rise of cortisol levels (10 min post-stress), suggests that cortisol release could be triggered by more rapidly responding factors, such as the sympathetic system. This hypothesis may be supported by a high upregulation of several genes involved in synaptic triggering, observed both during the first and the second stress episodes. Furthermore, while the transcriptome profile showed few changes at 60 min post-stress, expression of genes in several immune-related pathways increased markedly with each successive time point, demonstrating the role of the immune system in fish coping capacity. Although many of the genes discussed in this paper are still poorly characterized, this study provides new insights regarding the mechanisms occurring during the stress response of salmon parr and may form the basis for a useful guideline on timing of sampling protocols.

Dip nets are widely used to retrieve fish from the water but can cause injuries to the fish. The objective of this study was to document the potential effects of dip net mesh size with respect to external injuries, more specifically fin splits and scale losses, in Atlantic salmon (Salmo salar). The study included 273 post-smolt Atlantic salmon from two fish batches referred to as the Tank group (w¯¯¯ = 178 ± 36 g, n = 198) and the Cage group (w¯¯¯ = 1319 ± 334 g, n = 75). Four identical dip nets with mesh sizes of 5, 10, 15, and 20 mm, respectively, were used to net fish out of their enclosures before the external injuries were assessed by visual inspection. The results showed that the number and severity of fin splits increased with dip net mesh size, but no effect of mesh size on the percentage of scale losses was found. Dip-netting of five fish at a time instead of one with the 20 mm dip net, had no statistically significant effect on the examined injuries. It is concluded that the use of dip nets with smaller mesh sizes can be positive for fish welfare by reducing fin splitting.

Dip nets are widely used to retrieve fish from the water but can cause injuries to the fish. The objective of this study was to document the potential effects of dip net mesh size with respect to external injuries, more specifically fin splits and scale losses, in Atlantic salmon (

Squeak and rattle (S&R) are nonstationary annoying and unwanted noises in the car cabin that result in considerable warranty costs for car manufacturers. Introduction of cars with remarkably lower background noises and the recent emphasis on electrification and autonomous driving further stress the need for producing squeak- and rattle-free cars. Automotive manufacturers use several road disturbances for physical evaluation and verification of S&R. The excitation signals collected from these road profiles are also employed in subsystem shaker rigs and virtual simulations that are gradually replacing physical complete vehicle test and verification. Considering the need for a shorter lead time and the introduction of optimisation loops, it is necessary to have efficient and inclusive excitation load cases for robust S&R evaluation. In this study, a method is proposed to truncate and identify the important parts of the different road profiles that are often used for S&R physical verification and then merge them to develop one representative excitation load case. The criteria for signal truncation were based on the S&R risk and severity metrics calculated from the vibration response at the critical interfaces for S&R. the method was used in a case study involving the instrument panel of a passenger car. Results of the virtual simulation and the rig tests were compared with the complete vehicle test. The proposed synthesised signal generation strategy was validated by physical testing through measuring vibration signals. The results supported the possibility of replacing multiple S&R excitation signals with one single representative inclusive signal, while the quality of S&R risk prediction from the system response was maintained. The outcome of this work can lead to a more efficient physical and virtual S&R verification in the development process of passenger cars.

Random samples of individual fish for the cause of estimating body size average and variation in fish groups is a widespread practice in scientific experiment as well as in fish production. Sample accuracy and precision naturally increases with sample size, but may very well depend on other factors such as fish size dependent success rate of capture method and spatial-behavioural size segregation. In 4 case studies with Atlantic salmon (Salmo salar) where the entire population was crowded and measured for length and weight, we show how the estimated mean weight and condition factor, and their deviation from the true population mean, evolve with number of sampled fish: 1) sampling of 3400 pre-smolts (77 g) netted by hand from a 5 m tank, 2) 466 post-smolts (711 g) netted by hand from a 5 m sea cage, 3) 5569 post-smolts (394 g) netted by hand from a 12 m sea cage, 4) the same fish as in 3, but 9 months later (3367 g, n = 5288) and netted out by using a crane mounted tarpaulin container. Sampling of cases 1, 3 and 4 took place over consecutive days. Cumulative mean, i.e. the mean for each new fish sampled, of weight and condition factor were calculated for each sampling day. To simulate a perfectly random sampling to compare with that observed, the order in which the individuals in the entire population were sampled were randomized, and cumulative means calculated in the same manner. With the randomized sampling order the cumulative mean values rapidly approached true population mean and remained at that level for both weight and condition factor in all days on all sampling events. In contrast, the cumulative mean weight and/or condition factor of the observed sampling order generally deviated from the true mean even after a large number of fish had been sampled, and the patterns in how cumulative mean values developed was inconsistent between days.

Recreational fishing is a popular activity around the globe, and fish welfare issues related to the activity have received increasing attention in some countries, particularly in central and northern Europe and Australia. This chapter offers an introduction to recreational fishing, reviews literature on fish welfare in relation to recreational fishing and provides an overview of potential biological impacts and ways to reduce such impacts. We first focus on the question on how to reduce impacts on the welfare of the fish during recreational fishing. Second, we describe two case studies highlighting that practical implications of the fish welfare discourse may be disjointed from the scientific information base and be rather about fundamental moral questions about the ethical acceptability of the activity per se. We end by providing an outlook on the future of recreational fishing in the light of the current fish welfare discourse.

The introduction of catch welfare to commercial wild-capture fisheries will be challenging. In this chapter, we discuss how taking a science-based approach to understanding catch welfare in commercial fisheries could lead to practical solutions to improving welfare that will not only have ethical benefits, but may also have tangible benefits for the fishery, including improved sustainability, product quality and shelf life, and hence profitability. There has been little research to date specifically directed at the development of catch welfare in commercial fisheries. However, there is a substantial and growing body of literature on the fate and vitality of released animals from commercial fisheries—most recently catalysed by the introduction of the Landing Obligation in the EU. Furthermore, there is much to be learned from the aquaculture industry with regard to good welfare practices and product quality, particularly regarding catch handling and slaughter. This chapter utilises this available knowledge to develop a risk assessment-based framework for identifying capture-related stressors and suggests ways of mitigating their impact on the welfare of the catch, as well as on product quality. This framework is developed in context with four contrasting case study capture methods: trawl, purse seine, gill/trammel nets and pots. Finally, it concludes with a summary of the current research priorities and significant strategic challenges for developing welfare-conscious practices in commercial fisheries.

Detecting and interpreting information about resources and dangers and behaving flexibly and effectively are essential for survival and welfare. Fish, as well as other organisms, access their surroundings through their sensory systems. They must also have a cognitive system capable of integrating and interpreting the sensory input in relation to earlier experiences, and eventually act according to this input. Learning ability and memory enable fish to detect regularities and associations and construct mental images, categories and concepts. In this way, they can adapt their behaviour to the dynamic environment and predict the near future and the consequences of their behaviour. Numerous studies have shown that many fish species have evolved good cognitive abilities, and can construct internal maps, cope with complex social relationships and retain memories for long periods. Some fish can even innovate and use tools. However, the enormous diversity within the piscine world demands that we view learning, cognition and welfare from an ecological perspective. The cognitive capacity of individual species depends on the environmental and social complexity they encounter, and also differs between populations, coping styles and sexes. It is reasonable to believe that fish are conscious and have emotions and feelings, although their subjective experiences must be very different from ours and also vary between species. Their cognitive capacity and the behavioural flexibility that enables them to cope with aquaculture environments and procedures are essential for the welfare of farmed fish.

The present study shows that permanent melanophore spot patterns in Atlantic salmon Salmo salar make it possible to use images of the operculum to keep track of individual fish over extended periods of their life history. Post-smolt S. salar (n = 246) were initially photographed at an average mass of 98 g and again 10 months later after rearing in a sea cage, at an average mass of 3088 g. Spots that were present initially remained and were the most overt (largest) 10 months later, while new and less overt spots had developed. Visual recognition of spot size and position showed that fish with at least four initial spots were relatively easy to identify, while identifying fish with less than four spots could be challenging. An automatic image analysis method was developed and shows potential for fast match processing of large numbers of fish. The current findings promote visual recognition of opercular spots as a welfare-friendly alternative to tagging in experiments involving salmonid fishes.

The aim of this study was to evaluate the effects of pumping height and repeated pumping on the generalized stress response and gross injuries in harvest sized Atlantic salmon. Fish pumped from a net pen at either high (5.2 m) or low (3.6 m) pumping heights showed an elevated, but not severe physiological stress response (pH, pCO2, lactate, potassium, haematocrit, and sodium) compared to fish netted (not pumped), while effects of different pumping heights were overall not demonstrated. Repeated pumping (either 3 or 6 times) also caused an increase in stress response (pH, pCO2, pO2, lactate, potassium and sodium) compared to control fish, and a positive dose-response relationship was found for lactate. No fish died as a result of pumping, nor were injuries observed that could exclusively be attributed to pumping. In conclusion, although elevated from the control groups, the stress response following increasing pumping height and repeated pumping as conducted in these experiments were not indicative of causing severe stress or injuries.

Selection for faster growth to market size has a long history in selective breeding programs for aquaculture species. In Atlantic salmon, selection for faster growth has been carried out for more than four decades, i.e. about ten generations. Genetic improvement of growth rate is indicated as favourable correlated to improved feed utilization and, from an economic perspective, it can be argued that improved feed utilization is the main breeding objective: improved fish production per unit of nutrients (protein or energy) provided. However, also reduced production time in challenging grow-out environments, i.e. reduced exposure to sea lice and other disease agents, may be highly valuable from an animal welfare and environmental perspective. Reduced production time in challenging grow-out environments may also be economically favourable by reducing the need for mechanical or chemical treatments of the fish. Body weights at stocking and harvest are commonly recorded in breeding programs for Atlantic salmon. Selection for faster growth rate is usually based on recorded harvest weights, which represent the accumulated growth during the entire production cycle. Alternatively, selection could be based on other growth parameters, i.e. Thermal Growth Coefficient (TGC), which specify growth in specific production periods. Production of Atlantic salmon in open sea cages entails large variation of environmental parameters, i.e. water temperature, water quality, salinity, day light etc., that influences the growth. In this study, growth of individual Atlantic salmon in an open sea cage were investigated (as a part of the project BEHAVEGENES, NFR 2014–2017) to describe potential GxE for growth in different production periods. Test fish were transported to sea in autumn, 8 months after first feeding, and harvested after 16 months at a mean harvest weight of 5.5 kg. Individual body weights were recorded at five different time points allowing estimation of growth during a total of four subsequent production periods. Sexual growth dimorphism was discovered well before onset of sexual maturation; already 4 months after sea transfer, at average population body weight ~500 g, males were measured on average 2.8% heavier than females. Heritability estimates for body weight were all high (0.42–0.72), while heritability estimates expressed by growth calculated as TGC were mediate to high (0.35–0.56). Genetic correlations between growth (TGC) in different production periods varied between −0.20 and 0.82. The correlation between growth during an 8-month summer period (average body weights increase from 500 g to 3.9 kg) and a 4.5-month winter period (average body weight increased from 3.9 kg to 5.8 kg) was not significant (rg 0.16 ± 0.17), suggesting genetic variation for growth pattern and that different genes are involved in different growth periods.

Thermal treatment has become the most used delousing method in salmonid aquaculture. However, concerns have been raised about it being painful for the fish. We studied the behavioural response of Atlantic salmon acclimated to 8 °C when transferred to temperatures in the range 0–38 °C. Exposure time was 5 min or until they reached the endpoint of losing equilibrium and laying on their side, a sign of imminent death. At temperatures below 28 °C, none of the fish reached endpoint within the 5-min maximum. At 28 °C four of five fish reached endpoint, and fish reached endpoint more rapidly as temperature increased further. Fish transferred to temperatures above 28 °C had higher swimming speed immediately after transfer and maintained a high swimming speed until just before loss of equilibrium. Their behaviour was from the start characterised by collisions into tank walls and head shaking. Just before loss of equilibrium they started breaking the surface of the water, swimming in a circle pattern and in some instances displayed a side-wise bending of their body. In other words, salmon transferred to temperatures above 28 °C showed instant behavioural responses indicative of nociception or pain. Keywords: Behavioural response, Fish, Heated water, Nociception, Pain, Thermal delousing

Individual dynamics within salmon sea cages are poorly understood. Large inter- and intra-individual variations in swimming depth and higher average body weight deeper in the cage have been observed. Sampling of fish for inspection purposes and estimates of body weight distributions based on automatic measures at a limited depth interval may thus be skewed. The present study investigates how 335 randomly PIT tagged Atlantic salmon of 3.4 ± 0.96 kg (bled weight, mean ± SD) swam near square PIT antennas (0.6 m) fixed at either 5 or 9 m depth within a 14 m deep cage holding a total of 3750 individuals. The individual variation in registration frequency was large, with 76 individuals never registered and 12 individuals registrered 30 times or more. Larger individuals were much overrepresented at 9 m depth, resulting in 8.5% overestimation of average weight at this depth when repeated registrations were not accounted for. No overrepresentation of any size class was found at 5 m depth. Half of the individuals registered at 9 m depth were also registered at 5 m depth, and 82% of the individuals registered at 5 m depth were also registered at 9 m depth. We interpret the result as fish of all sizes using a wide depth range, but larger individuals spending more time at deeper water than smaller individuals. Therefore, size estimates at a limited depth interval should be avoided.

The possibility to prepare for and respond to challenges in a proper manner is essential to cope with a changing environment, and learning allows fish to up or downregulate the stress response based on experience. The regulation of the response to predicted needs should be easier in more predictable environments. We exposed salmon parr to chasing of either 15 s (weak stressor) or 5 min (strong stressor) twice daily for a 7-day learning period, with chasing either announced by a 30 s light signal (conditioned) or not announced (unconditioned). The behavioural response to the light signal was different between the conditioned and unconditioned groups, demonstrating that conditioned groups associated the signal with chasing. We could, however, not demonstrate any effect on the stress response of anticipation. The fish habituated to repeated stress exposures with a similar decrease in oxygen hyperconsumption in all groups. Due to habituation, possible effects of predictable announcement of a stressor on the physiological stress response may not have been expressed in this study. Plasma cortisol concentrations 1h after light signal and chasing the day after the training period was moderate in all groups although higher after 5 min chasing (13 ng ml(-1)) than 15 s chasing (7 ng ml(-1)). There was no physiological stress response after exposure to the light signal only after the learning period. We argue that the benefit of predictability of stressors is limited when the fish have no way to avoid the stressor.

Salmon fish farmers are seeking methods to reduce infestations of salmon lice. The highest concentrations of salmon lice copepodites are typically found in the upper meters of the water column and preventing this water from passing through the salmon cages has been proposed as a way to reduce salmon lice infestation. One simple implementation proposed by the industry is to place a permanent tarpaulin skirt around the upper part of the sea cages. Currently, there is little evidence that a skirt will block the entrance of surface water to the cages and thereby reduce infestation. However, this pilot study shows that putting a tarpaulin skirt around a full scale commercial sea cage may seriously decrease the oxygen saturation levels available for the fish inside the skirt. Further investigations into the effectiveness of skirts as a means of reducing the exposure of farmed salmon to sea lice must therefore take precautional measures.

Among other applications, self-feeding has been used to study food preferences in fish allowing them to choose between feeders with different food content. Preference tests assume that (i) trigger actuations are motivated by appetite, (ii) fish can learn which feeder contains which food and discriminate between feeders solely on the basis of their content, and (iii) in groups of fish, the triggering preferences is representative for the individuals of the group. We studied individual triggering behaviour in four groups of 14 Atlantic cod (length of 34 +/- 2 cm, weight of 424 +/- 102 g, mean +/- SE, water temperature comprised between 7-8 degrees C) that were first given the choice between two self-feeders with identical content (Period 1 of 14 days) and subsequently with one feeder full and the other empty (Period 2 of 14 days). In all four groups, one or two individuals performed the majority of the actuations, and in three groups the high triggering fish was a female high-ranked for size and growth rate. Cod displayed a preference for one of the two feeders despite their identical content. When the preferred feeder was emptied, the preference switched after one to eight days but both feeders were still actuated throughout the experiment. In conclusion, the assumption that actuation frequency reflects food preference and is representative for the individuals of the group may not be true, at least for Atlantic cod. If aiming at determining preferences representative for the whole population multiple representative fish should be kept isolated in separate tanks, with self-feeders containing each food option, on each tank.

The present study investigates vertical distribution of fish size in three 12 m × 12 m wide and 14 m deep sea cages stocked with Atlantic salmon of average weight from 3.5 to 3.7 kg, at commercial densities between 15.6 to 16.2 kg m-3 , in Norway. For each cage, individual fish weight were estimated by three 0.6 m × 0.6 m measuring frames as fish swam through. The frames were positioned at 3, 6 and 9 m depth. Recordings were carried out over five days per cage in succession, and during relatively stable environmental conditions in late autumn 2010. In all cages, measured fish were 15−25% smaller at 3 m compared to the average weights at 6 and 9 m depth. Largest average weight difference between depths within one cage was 0.995 kg. The average weight at 6 and 9 m was higher at night-time compared to daytime. Fish at the lower end of the weight spectrum were predominantly registered at 3 m, while fish at the higher end of the spectrum were mostly registered at the greater depths. Fish of average size were well represented at all three depths. The fact that smaller fish swam shallower may be ascribed to natural behavioural traits and introduce an important consideration in representative sampling within commercial sea cages. In biomass estimations and sea lice counts size-dependent vertical stratification needs to be accounted for.

Using reward conditioning to change the cognitive interpretation of stressors has been suggested as a method to reduce stress in farmed fish. We studied the modification of behavioural and physiological responses of Atlantic cod to a stressor. Eight groups of cod were exposed to a moving dip net six times daily over 7 days. Half were rewarded with food 10 s after dip net exposure (trace reward conditioning), while the remaining groups were fed 30 min after dip net exposure (unrewarded). The swimming speed and oxygen consumption responses after dip net exposure in the unrewarded groups declined at a rate of 13% and 12% trial−1, respectively. The swimming speed response was lost completely, while 37% of the initial oxygen hyper consumption response remained after habituation. The rewarded groups learnt to associate dip net exposure with food delivery. When the dip net stopped moving, they approached it before gathering in the feeding area. Due to this anticipatory behaviour, swimming speed and oxygen consumption remained elevated after dip net exposure in the rewarded groups. Plasma cortisol and glucose levels remained low after dip net exposure on the last day in both groups. The experiment demonstrates that a stimulus that initially evokes a fright response can become a positive signal of a forthcoming reward.

The production of salmonids in sea-cages has been developed for monoculture of the target species. However, we show here for the first time, that wild fish may enter sea-cages used for farming of Atlantic salmon (Salmo salar L.) in Norway, out-grow the mesh size, and thereafter become permanently trapped. Within seven different sea-cages located in western Norway, eight different species of wild fish were identified; European eel (Anguilla anguilla), sea trout (Salmo trutta L.), cod (Gadus morhua), haddock (Melanogrammus aeglefinus), saithe (Pollachius virens), pollack (Pollachius pollachius), hake (Merluccius merluccius) and whiting (Merlangius merlangus). In the two most extreme cases, a 5 × 5 × 7 m cage with 311 farmed salmon (903 g) also contained 542 whiting (79 g), 77 haddock (43 g), and 5 cod (26 g), and a 12 × 12 × 15 m cage with 1695 farmed salmon (559 g) also contained 1196 haddock (35 g), 1115 whiting (31 g), 46 cod (23 g), 23 saithe (48 g), 15 pollock (22 g), 5 sea trout (54 g), and 2 hake (29 g). The present study thus demonstrates that aquaculture cages designed for monoculture may attract and effectively ‘trap’ wild fish. We did not investigate the frequency of this occurrence, and the ecological significance of these observations remains unclear. However, with the ever-increasing number of sea-cages used for global aquaculture, this is clearly a topic for further research.

In this study, we demonstrate how an event that is initially frightening to Atlantic salmon is turned to a positive stimulus through habituation and associative learning. The study was carried out in four commercial sized tanks (5 m3) with near industry densities (>550 fish, 16 kg m−3), using a delay conditioning procedure with an aversive flashing light as the conditioned stimulus and food reward as the unconditioned stimulus. By using video image analysis of the distribution of the fish in the tanks, the changes in behaviour from trial to trial could be documented in great detail. The current study documents the change in behaviour across the individual conditioning trials, clearly showing the step-by-step nature of the transition. The salmon needed more than 26 trials to become fully habituated to the flashing light but showed clear anticipatory behaviour already after about 19 trials. This demonstrates that the learning process is a combination of habituation and associative learning.

The present study investigated the habituation rate of Atlantic salmon parr to a daily sudden transition from darkness to light during the 6-week “winter signal” period of alternating light (L12, D:12) in commercial smolt production. Increase in oxygen consumption when the light was turned on was used as a measure of activity, and habituation rate was estimated from the day to day decrease of this hyper-consumption. A non-linear regression model was fitted to the data, where an initial response of 4.36 mg O 2 kg − 1 min − 1 declined at a rate corresponding to 24% day − 1 , and approached a response level of 1.33 mg O 2 kg − 1 min − 1 over the course of the experiment. The results show that parr effectively habituated to the strong stimulus, but that 31% of the initial response was sustained over weeks and not subject to habituation.

This study describes how three individual fish, Atlantic cod (Gadus morhua L.), developed a novel behaviour and learnt to use a dorsally attached external tag to activate a self-feeder. This behaviour was repeated up to several hundred times, and over time these fish fine-tuned the behaviour and made a series of goal-directed coordinated movements needed to attach the feeder’s pull string to the tag and stretch the string until the feeder was activated. These observations demonstrate a capacity in cod to develop a novel behaviour utilizing an attached tag as a tool to achieve a goal. This may be seen as one of the very few observed examples of innovation and tool use in fish.

Summary Aluminum (Al) toxicity occurs frequently in natural aquatic ecosystems as a result of acid deposition and natural weathering processes. Detrimental effects of Al toxicity on aquatic organisms are well known and can have consequences for survival. Fish exposed to Al in low pH waters will experience physiological and neuroendocrine changes that disrupt homeostasis and alter behavior. To investigate the effects of Al exposure to both brain and behavior, Atlantic salmon (Salmo salar) kept in water treated with Al (pH 5.7, 0.37±0.04 µmol 1-1 of Al) for 2 weeks were compared to fish kept in a control condition (pH 6.7

We compared behavioural and physiological responses and recovery time after different acute environmental challenges in groups of salmon parr. The fish were prior to the study conditioned to a flashing light signalling arrival of food 30 s later to study if the strength of Pavlovian conditioned food anticipatory behaviour can be used to assess how salmon parr cope with various challenges. The effect on anticipatory behaviour was compared to the effect on feed intake and physiological responses of oxygen hyper-consumption and cortisol excretion. The challenges were temperature fluctuation (6.5C° over 4 h), hyperoxia (up to 380% O(2) saturation over 4 h), and intense chasing for 10 min. Cortisol excretion was only elevated after hyperoxia and chasing, and returned to baseline levels after around 3 h or less. Oxygen hyper-consumption persisted for even shorter periods. Feed intake was reduced the first feeding after all challenges and recovered within 3 h after temperature and hyperoxia, but was reduced for days after chasing. Food anticipatory behaviour was reduced for a longer period than feed intake after hyperoxia and was low at least 6 h after chasing. Our findings suggest that a recovery of challenged Atlantic salmon parr to baseline levels of cortisol excretion and oxygen consumption does not mean full recovery of all psychological and physiological effects of environmental challenges, and emphasise the need for measuring several factors including behavioural parameters when assessing fish welfare.

When animals associate a stimulus with food, they may either direct their response towards the stimulus (sign-tracking) or towards the food (goal-tracking). The direction of the conditioned response of cod was investigated to elucidate how cod read cue signals. Groups of cod were conditioned to associate a blinking light (conditioned stimulus, CS) with a food reward (unconditioned stimulus, US), with the CS and the US located at opposite sides of the tank. Two groups were trained in a delay conditioning procedure (CS = 60 s, interstimulus interval = 30 s) and two groups were trained in a trace conditioning procedure (CS = 12 s, trace interval = 20 s). The response pattern was similar for the delay- and trace-conditioned groups. The initial main response at the onset of the CS was approaching the blinking lights, i.e. sign-tracking. In the early trials, the fish did not gather in the feeding area before the arrival of food. In the later trials, the fish first approached the blinking lights, but then moved across the tank and gathered below the feeder before the food arrived, i.e. sign-tracking followed by goal-tracking within each trial. These two responses are interpreted as reflecting two learning systems, i.e. one rapid, reflexive response directed at the signal (sign-tracking) and one slower, more flexible response based on expectations about time and place for arrival of the food (goal-tracking). The ecological significance of these two learning systems in cod is discussed.

Basic knowledge about learning capacities and awareness in fish is lacking. In this study we investigated which temporal gaps Atlantic cod could tolerate between two associated events, using an appetitive trace-conditioning paradigm with blinking light as conditioned stimulus (CS) and dry fish food as unconditioned stimulus (US). CS-US presentations were either temporally overlapping (delay conditioning, CS duration 24 s, interstimulus interval 12 s) or separated by 20, 60, or 120 s (trace conditioning, CS duration 12 s) or 2 h (control, CS duration 12 s). The percentage of fish in the feeding area increased strongly during CS presentation in all delay, 20 s, and 60 s trace groups and in one out of two 120 s trace groups, but not in the control groups. In the 20 and 60 s trace procedures, the fish crowded together in the small feeding area during the trace interval, showing strong anticipatory behaviour. In all the conditioned groups, the fish responded to the CS within eight trials, demonstrating rapid learning. At 88 and 70 days after the end of the conditioning experiments, the delay and 20 s trace groups, respectively, were presented the CS six times at 2-h intervals without reward. All groups responded to the light signal, demonstrating memory retention after at least 3 months. This study demonstrates that Atlantic cod has an impressively good ability to associate two time-separated events and long time retention of learnt associations.