Physiology and Behaviour of Free-Swimming Atlantic COD (Gadus Morhua) Facing Fluctuating Salinity and Oxygenation Conditions

In the preceding paper (Claireaux et al. 1995), we reported the effects of variations in water temperature on the physiology and behaviour of free-swimming Atlantic cod (Gadus morhua L.). In heterogeneous temperature fields, cod appear to ‘track’ the water temperatures to which they were previously acclimated. Moreover, acute temperature changes of less than a few degrees Celsius can lead to marked changes in metabolic rate, which may be counterproductive to further activity or foraging. In order to understand more extensively the various interactions existing between the fish and its physico-chemical environment, and so to have a broader view of the processes that regulate the distribution and migratory patterns of cod, the repercussions of fluctuating salinity and oxygenation conditions also had to be addressed. In nature, cod seldom encounter changes in environmental conditions which require full-scale physiological regulatory mechanisms. It follows that, in most instances, the full scope of these response mechanisms will not be the prime determinant of fish behaviour. Typically, Atlantic cod are faced with environmental perturbations which are well below the resistance threshold. Thus, in optimizing their position in the water column, cod take into account factors such as the presence of prey or predators, hierarchical organisation within the school, their own biological and nutritional status and other environmental cues such as depth and bottom structure. In the present experiment, we examine the effects of variations in water salinity and oxygenation in a further attempt to understand the sensitivity with which cod perceive their physico-chemical environment.

Atlantic cod acclimated to a temperature of 5 °C and a salinity of 30 ‰ were equipped with ultrasonic transmitters which allowed the continuous monitoring of their vertical movements and heart rate. Fish were then placed in a 125 m³ tower tank in which various environmentally relevant fluctuations in the salinity and oxygenation conditions were replicated. The physiological and behavioural responses of cod were followed in parallel to the environmental manipulations. To evaluate the impact of the water physico-chemical characteristics on their metabolism, some of the experimental conditions studied in the tower tank were also reproduced in a swimming respirometer where fish oxygen consumption and heart rate could be monitored at various levels of activity.

The fish holding and instrumentation procedures, the oxygen consumption measurement method, the ultrasonic transmitters and the computerized data acquisition system were as described in the accompanying paper (Claireaux et al. 1995).

The respirometry measurements presented here used the same individuals as were tested by Claireaux et al. (1995). Following the temperature experiments, and after an overnight recovery period at their temperature and salinity of acclimation (i.e. 5 °C and 30 ‰), the water salinity in the respirometer was quickly reduced (<30 min) to 26 ‰. Fish oxygen consumption and heart rate (fH) were then measured during an exercise sequence similar to the one previously described (i.e. at 7, 18, 27, 37, 48 and 57 cm s^-1). At the end of the swimming trial, water salinity was returned to 30 ‰ and the fish were allowed to recover. The next day, the effect of a decrease in water oxygenation level from full saturation (17.9 kPa) down to 9.0 kPa was then tested according to the same swimming protocol. To induce hypoxia, water flowed through a gas-exchange column bubbled with nitrogen before reaching the respirometer. and fH recorded during swimming in the respirometer were expressed as mean ± standard error of the mean (S.E.M.) and were compared with their value at the lowest water speed (reference condition) to assess the statistical significance of any changes (unpaired Student’s t-test; P<0.05).

The experimental conditions (water supply, lighting, number of fish) in the tower tank were identical to those described previously (Claireaux et al. 1995).

Adjustments in salinity and the stratification of the water column were achieved by flowing known volumes of full-strength or diluted sea water through inlets placed at various depths in the tank. Water temperature was controlled using two titanium heat-exchange coils immersed at different levels in the tank. Water oxygen content was controlled using an H-shaped ‘airlift’ immersed in the tank. This device, made of 10 cm diameter polyvinylchloride pipe, consisted of two 6 m long vertical sections connected at the top by a 0.5 m horizontal piece. When air or nitrogen was bubbled into one of the ‘legs’ of the H, the upward movement of the bubbles in the pipe caused a suction effect and caused water to ascend. When it reached the top of the pipe, the newly equilibrated water flowed through the horizontal portion and returned to its initial layer via the second vertical section. The intensity of the bubbling was adjusted to regulate the water flow in the airlift and to avoid mixing with adjacent water layers. Oxygen, salinity and temperature profiles of the water column were made periodically using YSI meters, models 57 DO and 33 SCT, respectively.

The respirometry experiments described here were conducted as part of the experimental series described in the preceding paper (Claireaux et al. 1995) and thus concern the same individual cod (see Materials and methods). The relationships between heart rate (fH), oxygen consumption and swimming speed established in reference conditions (i.e. a temperature of 5 °C, a salinity of 30 ‰) are presented in Fig. 1A. Fig. 1B shows the same relationships following an abrupt (within 30 min) decrease in water salinity from 30 to 26 ‰. When compared with reference conditions, similar values were measured in resting fish (7 cm s^-1) exposed to dilute sea water, although the mean fH was higher, 22.1±0.6 beats min^-1 in reference conditions and 24.0± 0.5 beats min^-1 in low-salinity water. During the subsequent exercise period, the metabolic responses observed in both salinity conditions (Fig. 1A,B) were of comparable amplitude. Furthermore, as in the reference conditions, no significant changes in mean heart rate were observed during exercise in low-salinity water.

When cod were exposed to decreased water oxygen tension ⁠, a bradycardia was observed, as fish resting fH dropped from 22.1±0.6 beats min^-1 in reference conditions to 19.9±0.8 beats min^-1 in the lower-oxygen environment. Simultaneously, their resting decreased from 26.8± 2.1 mol min^-1 kg^-1 in normoxia to 19.5±0.9 mol min^-1 kg^-1 in hypoxia. During exercise, however, did not change noticeably with increasing water velocity, but fH rose significantly, reaching 25.8±1.2 beats min^-1 at 57 cm s^-1.

When these fH and data were plotted irrespective of the swimming speed at which they were measured, more interesting features were observed. Following exposure to diluted sea water (Fig. 2), no significant changes in the relationship between the two variables were observed, suggesting that a 4 ‰ fluctuation in the water salinity had no major physiological consequences for the animal. In hypoxic conditions, however, we noticed a downward shift of the regression line (Fig. 3). This deviation, presumably related to an overall decrease in the scope of the aerobic metabolism, suggested that fish were operating on new bioenergetic bases in order to satisfy their energetic requirements.

Following a 24 h post-surgery recovery period, Atlantic cod, acclimated to 5 °C and 30 ‰ salinity, were placed in the tower tank and their vertical movements and heart rate were monitored from their ultrasonic tags for 3 days. During this period, water temperature and salinity conditions remained homogeneous across the tank (5 °C, 30 ‰). On the fourth day, diluted sea water was slowly added at the top of the tank, resulting in a layering of the water column which was completed after 2 days. Fig. 4A shows the behavioural and metabolic responses recorded from one fish during the second day of this stratification procedure. As during the previous 4 days, this animal spent the dark hours at the bottom of the tank (9.5–10.5 m). After daybreak (06:30–06:45 h), however, a marked change in the animal’s behaviour was observed. Instead of patrolling the bottom 1 m of the tank, the cod undertook periodic and rapid excursions into the upper layers of the water column. Ultimately (between 12:00 and 15:00 h), presumably after having become accustomed to their new environmental conditions, all fish present in the tank voluntarily moved into the low-salinity upper half of the tower. Moreover, from then on cod were never observed at salinities above 27 ‰, excluding their feeding period (see Figs 7 and 8). After a week, the reverse experiment was conducted as the water column was homogenized over a 2 day period by flowing full-strength sea water into the bottom of the tank. Fig. 4B shows the behavioural response of the individual previously followed (Fig. 4A). As high-salinity layers rose in the tower tank, this animal, togather with the rest of the school, moved up into increasingly shallower water, presumably to avoid being exposed to the higher-salinity water mass.

Heart rate recordings made during these two experiments also displayed intriguing features. Irrespective of the environmental conditions faced by the animal, daytime was always associated with sudden, short-lived periods of tachycardia. These cardiac events were never observed during the night and occurred at increasing frequency during periods of environmental perturbations. During stable conditions, in contrast, their frequency of occurrence diminished irrespective of the exact nature of the conditions imposed. For example, before and after the stratification of the water column (Fig. 4A), spikes in the fish heart beat occurred approximately 5–10 times per day. During the stratification procedure (Fig. 4A), or as the water column was being homogenized (Fig. 4B), these brief episodes of tachycardia were observed on approximately 20 occasions per day.

To determine the cause of the alterations in the cardiac rhythm, we examined the vertical movements of the fish and their heart rate on a more appropriate time scale (Fig. 5). It then emerged that each distinguishable cardiac event was associated with a sudden burst in cod activity. Indeed, each time the animal initiated a ‘round-trip’ to the bottom of the tank, a slight, short-lived bradycardia was observed, followed by a more prolonged tachycardia. Although the leisurely shuttle down and up the tank usually occupied less than 2–3 min, 5–6 min was needed for the heart rate to recover. Interestingly, these periods of relatively higher heartbeat frequency were also characterized by little or no beat-to-beat variability in the heart rate. The animal response depicted in Fig. 5 was selected because it displayed the most obvious cardiac responses. However, although less clear-cut, these features were observed in several other fish, including those monitored during the temperature experiments (Claireaux et al. 1995).

Fig. 6 shows the behavioural response recorded from one fish monitored when the water oxygen partial pressure was reduced over a 2 day period. During the first 12 h, the fish followed the daily cycle it had conformed to during the preceding days, i.e. swimming into shallower waters at night and into deeper water layers during the day. However, as started to decrease at the bottom of the tank, an alteration of fish behaviour was observed, the animal initiating its upward migration 3–4 h earlier than usual. On the next day, the downward migration associated with daybreak still occurred, but its amplitude was then greatly reduced as the fish avoided being exposed to hypoxic water (i.e. p.p.m.). All the animals (N=12) present in the tank at the time displayed the same behaviour. During the following days, fish were occasionally observed swimming through the less-oxygenated water layers. However, these incursions were transitory and animals always returned to the normoxic layers within 2–3 min.

These quick intrusions into ‘inauspicious’ waters were frequently observed during feeding periods (Figs 7 and 8). Fig. 7 illustrates how the introduction of food at the bottom of the tank impacts on cod behaviour. The water column was stratified in salinity and, before food was introduced into the tank, the fish occupied the upper half of the tower (salinity >27 ‰). As soon as the animals perceived the presence of food at the bottom, they immediately moved into the high-salinity layer and remained there for as long as food could be found (45 min). When no more food was available, fish eventually returned above the ‘chemocline’ (in this case a halocline) symbolised by the shaded area. When the bottom layer was made hypoxic, the feeding behaviour changed again, illustrating the great accuracy of the fish’s perception of its environment (Fig. 8). Following the addition of food, fish moved from their initial position down to the vicinity of the chemocline and from there made swift ‘round-trips’ to the bottom and back of the tank to grab food. When all the food was eaten, the fish swam away from the chemocline and eventually returned to their original depth. Note that the values presented in Figs 7 and 8 are the averaged depth measured over 30 s at 2 min intervals. In the case of Fig. 8, fish shuttle-trips below the transition zone were so fast that we could not precisely individualize each of them.

We first examined, in a swim-tunnel respirometer, certain aspects of the physiology of the Atlantic cod in situations of changing salinity or oxygenation. The results showed that, whereas a 4 ‰ decrease in water salinity had no major impact on the fish metabolic and cardiac performances, a reduction in ambient oxygenation level from saturation to 9.0 kPa induced significant physiological readjustments.

Apart from some initial differences in the resting heart rate immediately following exposure to diluted sea water, no additional discrepancies between the two salinity treatments were observed in the relationships connecting fH, and swimming speed (Figs 1A,B and 2). While many studies have investigated the changes in oxygen consumption in fish acclimated to various salinities, very few have actually examined the impact of acute salinity changes on the metabolic rate. Davenport and Vahl (1979) studied, in the euryhaline species Blennius pholis, the effect of salinity fluctuation in the range 0–34 ‰ at tidal periodicity. They found that increased at low and rising salinities but dropped back to routine level at high or falling salinities. These authors also reported no significant changes in the heart rate of blenny during these salinity oscillations. Atlantic cod is also considered to be a euryhaline fish (Odense et al. 1966; Fletcher, 1978a,b). For example, Dutil et al. (1992) showed that cod directly exposed to ambient salinities as low as 7 ‰ do not experience severe osmoregulatory disturbances, even during the first 24 h in diluted sea water. Similarly, Claireaux and Dutil (1992) showed that acclimation of cod to salinities ranging from 28 to 7 ‰ had no repercussion on their ability to tolerate low oxygen levels. In the present study, the lack of any significant cardiac and metabolic perturbations following a 4 ‰ decrease in ambient salinity may relate to this osmo-and iono-regulatory ability. However, as only heart rate was measured, conclusions concerning the cardiac responses at rest, and also during swimming, must be treated carefully. Priede and Tytler (1977) showed that the range of possible values for a given heart rate was very large in cod, substantiating the statement by Randall (1968) that, in fish, the stroke volume can vary extensively, giving a wide range of cardiac output for any given heart rate. This is illustrated by the fact that, in the respirometer, fH changed little with increasing swimming speed (Fig. 1A,B), whereas in the tower tank, increased activity was accompanied by marked changes in fH (Fig. 5).

When cod were exposed to hypoxic conditions, significant metabolic and cardiac adjustments occurred, resulting in the decline of both fH and in resting conditions and in the amplitude of the metabolic response during swimming (Fig. 1A,C). In contrast, the tachycardia observed during exercise in hypoxic water was more pronounced, fH measured at high swimming speed reaching the level recorded under normoxic conditions. This resulted in a downward shift of the fH versus relationship (Fig. 3). As in many other species (Randall and Daxboeck, 1984; Taylor, 1985), the hypoxic bradycardia in cod occurs with essentially no change in cardiac output (Fritsche and Nilsson, 1989) owing to a simultaneous increase in the stroke volume of the heart. According to Randall (1982), these cardiac adjustments promote lamellar recruitment in situations of limited oxygen availability. It is generally accepted that the metabolic response of fish exposed to low oxygen levels can be partitioned into two phases. In moderate hypoxia, adjustments in the fish respiration and circulation allow the oxygen consumption to be sustained (regulatory phase), whereas under more severe hypoxic conditions decreases (conforming phase). The threshold between these two phases is species-specific. Our data suggest that, in Atlantic cod, modification of the metabolic rate according to the amount of oxygen available may take place at as high as 9 kPa (Figs 1C, 3). Sundnes (1957) reported 8.0 kPa to be the below which became dependent upon water oxygen content. Saunders (1963) also reported that the oxygen consumption of small cod (1.14 kg) tended to decrease in situations of decreasing ambient oxygen content. In larger individuals (2.33 kg), however, was apparently maintained constant over the range of tested (3–10 p.p.m.). To explain this difference in the metabolic response with size, Saunders drew a parallel between the observed changes in metabolic rate and the fact that the increase in the respiratory volume was from 7 to 27 l h^-1 (threefold) in small fish and from 18 to 90 l h^-1 (fivefold) in large fish. Conversely, Claireaux and Dutil (1992) reported no significant changes in the concentrations of lactate, pyruvate and glucose measured in the white muscle, the liver and the heart of cod exposed to mild hypoxia (8.0 kPa) for 6 h.

Our objective in comparing swim-tunnel respirometry experiments with observations made on free-swimming animals was to characterize the potential ties existing between the physiological and behavioural processes taking place in cod. For example, even though the environmental conditions tested in the tower tank were well within the range of the perturbations ordinarily experienced by cod, they nevertheless induced marked behavioural responses. This is a clear example where inferences about the behavioural response of free-ranging animals facing a given set of ambient conditions based only on their physiological ability to endure these conditions could be very misleading. For instance, the data presented in Fig. 2 and the studies by Odense et al. (1966), Fletcher (1978a,b) and Dutil et al. (1992) indicate that Atlantic cod can withstand a wide range of ambient salinities. It might therefore be tempting to conclude that, in the range 30–7 ‰, water salinity has no major impact on the distribution of this species in its natural environment. Similarly, it could easily be concluded from our respirometry data and from the study by Claireaux and Dutil (1992) that ambient oxygen level is not a predominant environmental factor influencing cod distribution and migration patterns. However, the telemetry experiments clearly reveal that this is not the case and that, as for temperature (Claireaux et al. 1995), small changes in salinity or oxygenation levels are meaningful when trying to understand cod behaviour.

When a low-salinity water layer was made available to the fish, they all moved into it and were then reluctant to leave. When full-strength sea water was then added to the tank, fish distribution was altered again as they tried to escape from the approaching higher-salinity water mass. However, if food was introduced into the high-salinity layer, fish did not hesitate to dive in to feed, but they always returned to the low-salinity layer soon afterwards. Numerous reasons can be proposed to explain the behavioural choices made by cod, but the extent of their osmoregulatory ability is clearly not one of them. Likewise, the marked behavioural response we observed as hypoxia developed at the bottom of the tank was probably not directly driven by any sort of physiological requirements related to respiration. Indeed, Jobling (1988) mentioned 1.7 p.p.m. as being the limit below which cod asphyxiate. The lowest level measured at day 1 of the hypoxia experiments was still above 6.0 p.p.m. (Fig. 6). It seems probable that, in these environmental conditions, the need to meet respiratory criteria was higher in the fish’s priority list and thus weighed more heavily than other factors in determining certain aspects of their daily behaviour (e.g. the amplitude of their daily vertical migration or feeding behaviour). Schurmann and Steffensen (1992) also described interactions between the behaviour of cod and the oxygenation conditions in their environment. They showed that, as the amount of oxygen available decreased, cod tended to select colder environments (i.e. behavioural hypothermia), increasing their chance of survival by reducing their metabolic rate.

Several workers have tried to relate marine fish abundance with the physico-chemical properties of the surrounding water and found no reproducible patterns (Jean, 1964; Scott, 1982; Tremblay and Sinclair, 1985; Perry and Neilson, 1988; Smith et al. 1991). Our data suggest that, when establishing correlations between distribution patterns and environmental factors, one should also take into account the various options available to the fish at the time. In some instances, fish may simply be constrained into choosing the lesser of two evils.

Most estimates of energy expenditures in wild fish have been made by extrapolating laboratory measurements of metabolic rate to the field situation. However, as pointed out by Lucas et al. (1991), in order to formulate a bioenergetic model which fits with field estimates, information concerning short-term changes in the metabolic rate are essential. The heart rate recordings presented in Figs 4 and 5 exemplify this necessity for Atlantic cod. Indeed, in addition to the problem already discussed of the loose relationship between heart rate and cardiac output in fish, the data reported in Figs 4 and 5 show that the heart rate of a free-ranging animal, and presumably its metabolic rate, can be extremely variable (Figs 2 and 3). Overlooking the relevance of such short-term oscillations will inevitably result in an underestimation of the true energy expenditure of the fish. In this example, the beat-to-beat variability was related to swimming activity and confirmed the existence of a daily cycle in the fish movements. Sureau and Lagardère (1991) reported similar peaks in the heart rate of sole Solea solea in relation to activity, although in this case they mainly occurred at night. Such short-term cardiac adjustments with bursts of animal activity were also observed in pike Esox lucius by Lucas et al. (1991).

Great caution must be taken when extrapolating laboratory measurements to the field, since different time scales necessarily apply. Priede (1985) discussed the problem of choosing the most appropriate time scale when measuring metabolic processes. Priede argued that fluctuations in oxygen uptake at the gill are regulated on a time-base of about a minute and that the immediate energetic objective of a fish is simply to regulate its metabolism on a minute-by-minute basis. Accordingly, the most significant changes in the heart rate of our telemetered cod occurred within a time frame of 5–6 min and they were the result of periods of fish activity which lasted no longer than 2–3 min.

In conclusion, the present study, together with that of Claireaux et al. (1995), indicates that the distribution and movements of Atlantic cod are determined by complex interactions, behavioural and physiological, between the animal and its environment. The studies also clearly illustrate that such interactions cannot be satisfactorily described if only the physiological regulatory ability of this species is taken into account. These data also shed new light on the growing amount of information concerning the heterogeneity of cod distribution (Jean, 1964; Scott, 1982; Tremblay and Sinclair, 1985; Perry and Neilson, 1988; Smith et al. 1991), the variability of its migratory patterns (Rose and Leggett, 1988a,b, 1989), the genetic variation throughout its range (Mork et al. 1985) and the cyclic variations in its abundance (Sutcliffe et al. 1983; Koslow 1984; Koslow et al. 1987). Obviously, cod are exquisitely sensitive to their surroundings, and it follows that the productivity of their stocks will reflect these sensitivities.

We are indebted to Todd Bishop for his technical assistance and to Norval Balch and the staff of the Aquatron (Dalhousie University, Halifax) for providing us with exemplary working conditions. We also thank VEMCO Engineering Ltd for their excellent heart rate and depth transmitters. Funding for this research has been provided by OPEN, one of the fifteen Network of Centres of Excellence supported by the Government of Canada, to S.R.K. and R.G.B.