Water-electrolyte balance in goldfish Carassius auratus, under constant and diurnally cycling temperature conditions

The goldfish Carassius auratus L. tolerates temperatures ranging from approximately o to 40 °C, and between these extremes its standard rate of oxygen consumption rises by at least one order of magnitude (Beamish & Mookherjii, 1964). The responsive increases in ventilatory flow, cardiac output and effective gill area used to amplify oxygen uptake inevitably lead to increases in endosmosis and rates of branchial and urinary ion loss (Evans, 1969; Isaia, 1972; Maetz, 1972; Motais & Isaia, 1972; Randall, Baumgardner & Malyusz, 1972; Houston, 1973; Mackay, 1974; Cameron, 1976). Nevertheless, increases in water content and reductions in electrolyte concentrations are usually seen only at near-lethal high temperatures. Goldfish, like many freshwater temperate zone fishes, exhibit only modest changes in water-electrolyte status over relatively broad temperature ranges (Prosser, Mackay & Kato, 1970; Murphy & Houston, 1974; Mackay, 1974).

In common with most studies on the thermoacclimatory process, however, those concerned with its ionoregulatory manifestations have been based primarily upon long-term exposure of test animals to precisely controlled temperatures. Implicit is the assumption that temperature per se is the critical variable. For many species constancy of temperature represents an unusual, if not abnormal condition. The relatively shallow, slow-flowing waters in which goldfish are locally abundant, for example, undergo substantial daily temperature fluctuations during the summer months; often under conditions which prevent the expression of avoidance behaviours. Under these circumstances rate and direction of temperature change may be as important as the actual temperature. Accordingly, the present study was undertaken to examine ionic responses to constant temperature in relation to those which develop under the more ecologically realistic conditions provided by a diurnal temperature cycle.

Goldfish (32·6+ 1·11 g, mean weight +1 S.E.M.) were obtained locally and maintained in 200 1 recirculating fibreglass tanks provided with high capacity suction filters ( ∼ 5 1 min⁻¹). The St Catharines city water used in the study following dechlorination and ageing was of moderate total hardness (∼ 140 mg 1⁻¹, as CaCO₃) and alkalinity (∼ 95 mg 1⁻¹, as CaCO₃). Conductivity ranged from 500 to 515 μmho cm⁻¹, and pH from 7·2 to 7·6. Each tank was provided with a light-tight hood, and all groups were held on a 12 h light: 12 h darkness schedule initiated at 07.00 using time switches accurate to + 10 min.

Tanks were aerated by jets from their filter pumps and by laterally positioned air bars, an arrangement which provided uniform circulation. Oxygen levels varied inversely with temperature, but routinely exceeded 80% saturation. Constanttemperature tanks were regulated to within ±0·15 °C of set-point as measured by U.S. National Bureau of Standards certified thermometer at 12 fixed tank positions. A low-frequency sine wave generator was used to drive a 24 h temperature cycle accurate to ± 4 min day⁻¹ and to within + 0·5 °C on successive cycles. The tempera ture cycle was designed to mimic local conditions without exposing the animals to acute thermal stress. Thermal tolerance, scope-for-activity and growth rate data indicate that the optimum temperature of the goldfish is close to 25 °C. This was accordingly selected as the cycle midpoint. The sine wave cycle provided equal periods of exposure to temperatures above and below the midpoint. Cycle peak (30 °C) and minimum (20 °C) temperatures corresponded to summer variations in regional waters where this species is abundant.

Following transfer to acclimation tanks, animals were held at 22–24 °C 10 days. Temperatures were then adjusted by about 1 °C day⁻¹ until desired levels had been established. Fish to be exposed to cycling temperature conditions were first brought to the mid-range temperature. Cycle amplitude was then increased by approximately + 1 °C day⁻¹ to 25 ± 5 °C. All groups were held for at least three weeks before use. Feeding was carried out twice daily during acclimation using a finely ground commercial pellet, with excess food and faeces being removed and water levels adjusted after 30 min. There were no deaths, and the animals appeared to be healthy throughout the study.

Fish exposed to cycling conditions were sampled at 03.00 (20 °C), 09.00 (25 °C), 15.00 (30 °C) and 21.00 h (25 °C). Sampling at 03.00 and 21.00 h was carried out under dim red light to reduce disturbance. Specimens at constant temperature were simultaneously sampled to provide appropriate controls, i.e. 20 °C at 03.00, 25 °C at 09.00 and 21.00 and 30 °C at 15.00 h. Meier, Lynch & Garcia (1973) and Spence, Meier & Dietz (1977) have recently reported that significant diurnal variations in the ionic composition of Fundulus occur under reasonably constant temperature conditions. To examine this in goldfish, the sampling schedule for the constant temperature groups was expanded to include the previously noted periods. The acclimation and sampling programme was repeated three times, and all data pooled to give samples of 15–25 specimens at each temperature and sampling period.

Sampling of plasma and epaxial muscle and analytical procedures have been outlined in previous reports (e.g. Houston & Smeda, 1979), and differed in this study only in that a Perkin-Elmer model 327 atomic absorption spectrophotometer was used rather than the SP-90 instrument employed in earlier work. Muscle water distribution was calculated on the basis of chloride space values since recent studies (Houston & Mearow, 1979) have shown that there are no significant differences between these, and volumes obtained with administered [¹⁴C]PEG-4000. Estimates of cellular compartmental volume and ionic composition were carried out as described by Houston & Mearow (1979).

Data were analysed by single-classification analysis of variance. All values were routinely transformed to base-10 logarithms, and significance attributed to differences at the 0·05 level or better.

All plasma ions other than magnesium exhibited significant diurnal variation at one or more temperatures (Table 1). These variations were not, however, consistent in the sense that maxima and (or) minima were associated with specific sampling times.

This was also true of plasma water. On the other hand, peak and minimum chloride concentrations were observed during light and dark periods respectively. Similarly, minimum potassium concentrations regularly occurred at either 03.00 or 21.00, while maximum plasma water was encountered during light periods. Thus, there is some evidence that plasma composition in this species, as in rainbow trout (Salmogairdneri), is subject to photoperiodic influence (Murphy & Houston, 1977).

More uniform variations were seen in relation to temperature. Chloride concentrations increased with increases in acclimation temperature, while sodium levels tended to decline. Magnesium was consistently lower at 25 °C than at 20° and .30 °C. Variations in other ions and plasma water were sometimes significant, but presented an inconsistent pattern in relation to temperature.

Muscle ions other than potassium also exhibited significant diurnal changes at at least one temperature, and most commonly at 20 °C (Table 2). Again, these tended to be irregular in occurrence. However, maximum chloride and magnesium concentrations were consistently observed at 09.00 and 21.00 respectively. Interestingly, peak levels of four of the five ions considered occurred during dark periods in animals held )t 20 °C. At 30 °C, on the other hand, maxima were more prevalent at 09.00 and 15.00.

Progressive changes in concentration with temperature were not observed. Magnesium levels were, however, consistently higher at 25 °C than at either 20 or 30 °C. Except at 03.00 h, this was the case with sodium, potassium and calcium as well.

Diurnal variations in muscle water content were also observed, with minima most common at 03.00 h (Table 3). In general, however, the differences observed were modest, and only occasionally significant. Much the same was true of the volumes of the extracellular and cellular compartments. Significant variations were apparent, but inconsistent in times of occurrences. Except at 03.00 h, water content tended to rise at higher acclimation temperatures, and this was true of cellular phase volume as well. Chloride space volumes, on the other hand, declined as temperature increased.

By comparison with fish held at constant temperature and sampled at corresponding times and temperatures plasma sodium, chloride, potassium and water levels in animals exposed to cycling temperatures were substantially and, for the most part, significantly reduced (Fig. 1). Magnesium concentrations, however, were stable Calcium concentrations tended to increase with temperature under these conditions J a tendency also apparent in relation to chloride and potassium. Concentrations tended to be higher in fish sampled during dark than light periods. This was particularly noticeable at 25 °C since samples were taken at both 09.00 and 21.00 h.

No obvious differences in the muscle sodium content of animals held under the two temperature regimes were apparent (Fig. 2). Potassium, magnesium and particularly calcium levels were reduced under cycling conditions, while chloride tended to be more stable with respect to temperature and time than was the case in goldfish held at constant temperature. Potassium, however, exhibited a marked peak coincident with the peak cycle temperature. Opposed trends were apparent in some instances. Chloride, for example, increased at higher temperatures under cycling conditions, but declined in constant-temperature animals. Consistent with earlier observations on rainbow trout exposed to cycling temperatures (Toews & Hickman, 1969), reciprocal variation in muscle sodium and potassium levels was apparent. The relationship between the two was best fitted by the simple function [Na⁺] = 7·9e^−0·01 [K⁺] (r = —0·864, P < 0·05). By contrast, a direct relationship between sodium and potassium characterized fish held at constant temperature, i.e. [Na⁺] = 62·8 In [K⁺]–270·2 (r = 0·602, P < 0·05).

Differences in water content and distribution were also apparent under the two regimes (Fig. 3). Animals exposed to cycling temperatures were notable for reduced overall water content, larger extracellular phase volumes and correspondingly smaller cellular volumes. Furthermore, chloride space under cycling conditions tended to increase with increasing temperature, while the volume of the cellular compartment decreased. The converse was true of fish held at constant temperature.

Mean estimated cellular sodium, potassium, magnesium and calcium concentrations are summarized in Fig. 4. For obvious reasons, use of chloride space as an indicator of extracellular phase volume precludes calculation of cellular chloride concentrations. Both sodium and calcium levels tended to be substantially lower under cycling conditions than was the case in fish at constant temperature and, on a diurnal basis, both exhibited less variation. This was true of magnesium as well. This cation varied little with time or temperature under cycling conditions, whereas, at constant temperature, magnesium concentrations increased at higher acclimation temperatures. Potassium, on the other hand, rose sharply at the peak of the temperature cycle; a distinct contrast with the situation observed at constant temperature. Again, reciprocal variation between sodium and potassium was apparent.

Virtually all studies on the nature and consequences of the thermoacclimatory process have been conducted at constant temperature. Few north temperate-zone freshwater fishes actually live under such circumstances, although, given access to a suitable range of temperatures, many practise some degree of behavioural thermoregulation (Reynolds & Casterlin, 1979; Beitinger & Fitzpatrick, 1979). Equally, however, other species, including some of relatively limited thermal tolerance, impose substantial temperature changes upon themselves. Beitte & Geen (1980), for example, have recently reported that in Babine Lake, British Columbia, sockeye salmon, Oncorhynchus nerka, are regularly exposed to temperature changes of up to 10 °C as a result of their daily vertical feeding migrations.

The present study indicates that water-electrolyte metabolism in goldfish exposed to a diurnal temperature cycle differs in several respects from that of animals held at constant temperature. Ionic concentrations are regulated at different levels, and variations in concentration with temperature differ, sometimes distinctly, in character and relationships between ion pairs. Consequently, these findings generally confirm the conclusions reached by Toews & Hickman (1969) following studies on rainbow trout exposed to a temperature cycle similar to that of a known trout habitat.

Other differences between fish held under constant and cycling temperature conditions have also been described. Incipient lethal temperatures and median resistance times increase following exposure to cycling temperature conditions (Heath, 1963; Feldmeth, Stone & Brown, 1974). The concentrations of specific goldfish haemoglobins are altered, as are erythrocytic concentrations of ions known to effect haemoglobin-oxygen affinity (Houston & Rupert, 1976; Koss & Houston, 1981). The thermal sensitivity of red cell carbonic anhydrase is reduced following acclimation of this species to cycling temperatures (Beaumont, Koss & Houston, 1981). In short, exposure to the more ecologically realistic circumstances provided by a cycling temperature regime appears to prompt significant changes at a variety of organizational levels. If this proves to be a general phenomenon, one must question the extent to which the practice of long-term acclimation to constant temperature conditions has provided a realistic picture of thermal relationships in species normally exposed to variable temperatures.

Given the differences observed, questions regarding mechanism and significance immediately arise. Unfortunately little is known as yet concerning the effects of cycling temperatures upon the transport systems involved in the maintenance of ionic balance. Reference was made to the reduced thermosensitivity of red cell carbonic anhydrase (Beaumont et al. 1981). This may have some bearing on the stabilization of chloride levels, since Kerstetter & Kirschner (1972) have provided evidence that erythrocyte-generated bicarbonate plays an important role in chloride uptake by exchange.

Three features seen in fish exposed to cycling temperatures appear to be adaptively appropriate. First, plasma ion levels in these animals were typically below those of animals at constant temperature. This was most apparent in the case of the principal plasma electrolytes, sodium and chloride, where differences of ∼ 5 and ∼ 10 mM were commonly seen. Corresponding reductions in the gradients driving the diffusion of these ions from the gills would be expected. Accordingly, a reduction in the rate of expenditure of metabolic energy in active uptake from the medium would presumably take place.

Differences in the ratio of plasma (∼ interstitial fluid) and estimated cellular potassium concentrations were also of interest. Resting membrane potentials in excitable tissues are frequently approximated by the potassium equilibrium potential : (RT/zF) (In [pK]/[cK]), where R = gas constant, T = absolute temperature, z = valency, F = Faraday, pK = plasma [K] and cK = estimated cellular [K]. In animals at constant temperature In ([pK]/[cK]) increases steadily as acclimation temperature rises. By contrast, under cycling conditions In ([pK]/[cK]) is relatively stable. If it can be assumed that threshold potential is not greatly altered during acclimation, this suggests that muscular excitability under the cycling regime would be less influenced by temperature than is the case in fish maintained at constant temperatures.

Finally, the importance of inorganic cations in metabolic regulation is increasingly appreciated, and has been the subject of several recent reviews (e.g. Bygrave, 1967; Freidberg, 1974; Wacker, 1980). Magnesium, for example, is associated with a variety of reactions involving phosphates, and is a critical element in control of the pyruvate dehydrogenase complex linking glycolysis to the Krebs cycle. Calcium is frequently antagonistic to magnesium-activated enzymes which function as metal-enzyme complexes (e.g. pyruvate kinase, enolase, phosphoglucomutase). Comparable antagonisms between potassium and sodium are known in relation to specific biochemical activities, as are marked differences in their activating or inhibiting effects upon others. Wacker (1980) has employed plots of (log [Mg]/[Ca]) versus (log [K]/[Na]) to examine relationships between the generally stimulatory and generally inhibitory effects of these ions. In fish exposed to cycling temperature conditions (log [Mg]/ [Ca]) = 1·41–0·79 (log [K]/[Na]). The correlation coefficient, r = 0·984, was significant at the 0·05 level. The corresponding relationship for those at constant temperature was (log [Mg]/[Ca]) = 0·82 (log [K]/[Na]) —0·615 (r = +-0·784, P < 0·01). This suggests that the stimulatory effects of magnesium in relation to calcium, and potassium in relation to sodium, would be additive at higher temperature in systems sensitive to this form of control in animals held at constant temperature. The opposite was true of fish held under cycling temperature conditions. Magnesium declined in relation to calcium as (log [K]/[Na]) increased. Presumably, any activating effects of increased (log [K]/[Na]) would, in this instance, be opposed by reductions in the [Mg] : [Ca] ratio. The specific metabolic implications are not obvious at present. Quite possibly, however, variations of this kind could play some role in the stabilization of metabolic activity under diurnally cycling temperature conditions.

Financial support of this project through NSERC and Shaver Research Foundation grants is gratefully acknowledged. Mr John Rustenberg of the University Technical Services Laboratory designed and built the temperature controller used in this study.