The Influence of Calcium on the Physiological Responses of the Rainbow Trout, Salmo Gairdneri, to Low Environmental pH

Acid precipitation resulting from industrial emissions of sulphur and nitrogen oxides now occurs over wide areas of Northern Europe and North America (Cogbill & Likens, 1974; Oden, 1976; Dillon et al. 1978). Lakes susceptible to acid input are typically soft waters with low ionic strength and low buffer capacity. Extensive surveys (Harvey, 1975; Schofield, 1976; Leivestad et al. 1976) have documented the progressive acidification of many such lakes and the concomitant loss of fish populations. Fish stocks are affected if the pH is less than 5·5 (Leivestad et al. 1976) and lakes below pH 4·3 are normally completely devoid of even the most acid-tolerant species (Harvey, 1979). While decreased recruitment of young fish has been widely cited as the primary factor in this disappearance (e.g. Schofield, 1976), massive fish kills brought on by episodic excursions of water pH to toxic levels (pH ≤4·3) have been reported in a number of instances (Jenson & Snekvik, 1972; Leivestad & Muniz, 1976; Harvey, 1979).

Several studies have been made of the physiological mechanisms of acid toxicity. Major disturbances in blood oxygen transport (Vaala & Mitchell, 1970; Packer, 1979), blood acid-base state (Lloyd & Jordan, 1964; Packer & Dunson, 1970, 1972 Neville, 1979; Packer, 1979) and ionic balance (Packer & Dunson, 1970, 1972; Mudge & Neff, 1971; Leivestad & Muniz, 1976; McWilliams & Potts, 1978) have now been reported, but the primary cause of death remains unknown. A major complicating factor is that the ionic concentrations of test waters have varied from those typical of acidified soft water lakes (e.g. Leivestad et al. 1976) to levels at least 10-fold higher (e.g. Neville, 1979). In both field and laboratory studies (Packer & Dunson, 1972; Leivestad et al. 1976) increased ion levels have been shown to improve the survival of fish exposed to low pH. Ameliorative effects apparently result from the elevation of either NaCl (Packer & Dunson, 1972; Leivestad et al. 1976) or Ca and Mg salts (Leivestad et al. 1976). Possibly the nature of the mechanism of acid toxicity also varies with the nature of the ionic environment.

In the present investigation, we have examined the physiological responses of the rainbow trout during chronic exposure to low environmental pH in hard and soft water. Repetitive blood sampling from chronically catheterized animals has been employed to monitor blood acid-base status (arterial pH, [HCO₃⁻] and ⁠), ionic status (plasma [Na+], [Cl⁻], [K+], [Ca²+]) and the extent of anaerobic metabolism (blood lactate levels), the latter serving as an index of the adequacy of blood O₂ transport.

Rainbow trout (Salmo gairdneri) of both sexes were obtained from Spring Valley Trout Farm, Petersburg, Ontario and held at least 1 week prior to experimentation in large fibreglass tanks continuously supplied with well-aerated, dechlorinated tap water at 7–12 °C. One- to two-year-old trout (90–442 g) were used in physiological studies, and fingerlings (2–5 g) in toxicity tests. The animals were acclimated for 2 weeks to water of the ionic composition (see below) and temperature (11 ±2 °C) appropriate for the experimental tests. The animals were fed regularly with commercial trout pellets but starved during the last week of acclimation and the subsequent experimental period.

To allow chronic blood sampling in the physiological studies, trout were anaesthetized in MS-222 (1:10000 dilution in the acclimation water) and surgically implanted with dorsal aortic catheters (Smith & Bell, 1964). The fish were allowed to recover ia acclimation water for 24–48 h. Control (i.e. day 0) blood samples were then drawn from each fish and analysed for acid-base state and ion levels as described below. Subsequently fish were either exposed to acidified water (pH4·0–4·5; experimental series) or allowed to remain in the acclimation water (pH 7·0–7·5; control series). Blood samples were drawn daily under each condition for at least 5 subsequent days. The control series were employed to separate the effects of the experimental protocol alone (e.g. catheterization, repetitive blood sampling, acclimation time) from the effects of acid exposure on acid-base state and ion levels.

All physiological experiments were conducted in temperature-controlled (11 ± 1 °C) recirculating water systems, each supplying eight darkened plexiglass chambers (2 I volume) in which individual fish were isolated. Each chamber received a flow of at least 500 ml/min and separate aeration. The recirculating systems consisted of either one 300 1 reservoir (control experiments) or two 100 1 reservoirs (acid exposure experiments). In the latter, one reservoir supplied water at normal pH (pH 7·o–7·5), while the other supplied acidified water (initially at pH 4·0, see below). The switchover from one reservoir to the other could be effected without disturbance to the fish. The systems contained no internal metal components, so as to minimize the risk of heavy metal poisoning from acid leaching of metal ions. Heavy metal ion levels were determined on acidified water resident in the recirculating system for 6 days (Table 1). With the exception of cadmium, the concentrations were below the water quality objectives (International Joint Commission, 1976). The cadmium level, while above the objective, was less than 6% of the 10-day lethal threshold concentration for trout at 12 °C (Roch & Maly, 1979).

In all series, both control and experimental, water was decarbonated prior to use in order to avoid possible complicating effects of high (Lloyd & Jordan, 1964; Neville, 1979) that might otherwise accompany acid titration of water bicarbonate. Decarbonation was effected by HC1 acidification to pH 2·5, 24 h aeration, and back titration to the appropriate pH with NaOH (KOH in the high [Ca²⁺], low [salt] water series). Water was always less than 1 mmHg during the experiments. HC1 was used for water acidification in all experiments rather than H₂SO₄, the more common mineral acid pollutant in the wild, to avoid possible toxic effects of sulphate anion (Maetz, 1973). A nominal pH of 4·0 was chosen for the experimental series. The median survival time of rainbow trout in soft water is in excess of 5 days at this pH (Leivestad et al. igyty, thus allowing adequate time in which to examine the development of deleterious physiological effects. The pH of the acidified water rose when being circulated through the fish chambers. To counter this, the water pH was returned to 4·0 at twice-daily intervals during the experiment by addition of HC1. The actual mean pH over the duration of the experiments was about 4·3 (Table 1).

Experiments were carried out in either ‘hard’ water, ‘soft’ water or high [Ca²⁺], low [salt] water. The chemical compositions of these waters are listed in Table 1. ‘Hard’ water was dechlorinated Hamilton, Ontario tap water. ‘Soft’ water was prepared by a 10-fold dilution of tap water with distilled water, and contained ions at concentrations approximating those found in acidified natural waters (Table 1). The mean effective dilution was slightly less than 10-fold for [Na+] and [Ca²⁺], and non-existent for [K⁺] because of the continual addition of these ions to the water by the fish during the course of an experiment (Table 1). High [Ca²⁺], low [salt] water was prepared by the addition of Ca(NO₃)₂ to soft water so as to raise [Ca²⁺] to levels approximating those in hard water (Table 1). Four experimental treatments were employed:

Acclimated hard water. Fish were acclimated and tested in hard water. Control (N= 8) and experimental (N = 21) series were run.

D. G. MCDONALD, H. HOBE AND C. M. WOOD

1. Acclimated soft water. Fish were acclimated and tested in soft water. Control = 6) and experimental (N = 8) series were run.

2. Unacclimated soft water. Fish were acclimated to hard water but were tested in soft water. Control (N = 7) and experimental (N = 8) series were run.

3. Acclimated high [Ca²⁺], low [salt] water. An experimental series only (N = 7) was run.

Blood samples (∼ o-6 ml) were drawn anaerobically using chilled, gas-tight Hamilton syringes. An equal volume of heparinized (100 i.u./ml) Cortland saline (Wolf, 1963) was then returned to the animal via the catheter. Each blood sample was analyzed for the following (except where noted in Results): haematocrit, pHa, total CO₂ content (C_a,C02, whole blood and plasma), lactate, plasma [Na⁺], plasma [Cl⁻], plasma [K+], plasma [Ca²⁺] and plasma [protein].

Blood pH was determined on 20 μl aliquots injected into a Radiometer microelectrode connected to a Radiometer PHM-72 acid-base analyzer and thermostatted to the experimental temperature. Before each measurement the pH electrode calibration was checked with Radiometer precision buffers (S 1500 and S 1510). Haematocrits were measured by centrifuging 80 μl blood samples in heparinized capillary tubes at 5000 g for 5 min. Blood and plasma s (50 μl samples) were assayed by the micro-method of Cameron (1971), using a teflon membrane on the electrode for greater stability and bracketing each unknown with NaHCO₃ standards to increase the precision of this technique. Lactate analyses were performed on 150 [A of blood immediately deproteinized in 300 μl of ice-cold perchloric acid (8%, w/v) and then centrifuged at 5000 g for 10 min. The supernatant was analyzed 1-lactate with Sigma reagents (Sigma, 1976). Chloride determinations were made on 20 μl plasma samples using a Radiometer CMT-10 chloride titrator. Sodium (20 μ l plasma sample), potassium (100 μl) and calcium (100 μl) concentrations were determined after dilution (1:500, 1:30, and 1:100, respectively) on an Eel (Na⁺ and K⁺) or Coleman 20 (Ca²⁺) flame photo-meter. Appropriate swamping was employed to eliminate the interfering effect of Na⁺ on Ca²⁺ and K+ emissions. Plasma samples (10 μl) were analysed for plasma protein concentration using a Goldberg refractometer (American Optical).

1. [HCO₃^-] and P_o002

2. ΔH⁺_b

Final blood volumes in acid-exposed fish were estimated from the changes in haematocrit and plasma protein concentration which occurred over the course of the experiment (cf. Fig. 10). The calculations (see below) assume that the initial blood volume was 5-0 ml/100 g (Stevens, 1968) in all fish, that the blood volume was unaffected by repetitive blood sampling and that the amounts of erythrocytes and plasma protein lost from the blood per 100 g due to sampling were the same for acid-exposed and control fish. The last assumption is true only when the body weights of control and acid-exposed fish are similar. This condition existed for all groups except the acclimated hard-water group exposed to acid. The overall mean weight for the former groups was 305 ± 8 g and there were no significant differences among the groups, whereas the body weights for the latter were significantly lower (140 ± 11 g). Therefore the data from these fish were excluded from the analysis.

Means ± one S.E.M. are reported throughout. These means (Figs. 1-10) exclude values from fish that succumbed during the 5-day acid exposure. The final values recorded prior to death for these fish are presented separately (Table 3). Significant differences (P < 0·05) within each acid series were tested with Student’s two-tailed t test (paired design) using each fish for its own ‘control’ values (the day o values). Significant differences (P < 0·05) among acid series were tested with Student’s two-tailed t test (unpaired design).

The relative toxicity of HC1 in hard ([Ca²⁺] = 3·3 m-equiv/1) and soft ([Ca²⁺] = o·2 m-equiv/1) water was assessed in the classical manner (Sprague, 1969) by 7-day lethality tests on fingerling trout (2–5 g) acclimated to water of the appropriate composition. Water was prepared as in the physiological studies. The tests were conducted in 801 polyethylene recirculatory systems. A range of pH levels from 3·0 to 48 in 0·2 unit increments was employed; pH control was accurate within ±0·05 unit up to 4·4, and ±0·1 unit at 4·6 and 4·8. The fish were tested at each pH, and individual mortality times recorded for each fish. Median lethal times, 95 % confidence limits, and significance of differences (P < 0·05) were estimated by log-probit analysis (Litchfield, 1949).

In hard-water-acclimated trout, the five-day exposure to low external pH resulted in a fall in arterial pH (Fig. 1 C) and [HCO₃⁻] (Fig. 1 B). The acidosis was apparently not due to the endogenous generation of respiratory or metabolic acids since there was no concomitant change in and lactate levels (Fig. 1 A, D), but rather represented an invasion of the extracellular fluid compartment by external H⁺ ions.

In Fig. 1C it can be seen that the acidosis was virtually fully developed by day 2. This may also be seen in the calculated ΔH⁺_b (Fig. 2 A) and was further borne out by measurements made on four fish until day 9. Net ΔH⁺₆ was 5·610·7 m-equiv/1 by day 2, and 5·9 ± 0·7 m-equiv/1 by day 5 (Fig. 2 A). By day 9 net ΔH⁺₆ had further greased by 23 ± 6 % only.

Accompanying this blood acid-base disturbance was a relatively minor plasma ionic disturbance (Fig. 3). There was a significant fall in plasma [Na+] which was apparently complete by day 3. Plasma [K+] showed only a very slight increase, significant on day 3 only, while plasma [Ca²⁺] and [Cl⁻] showed no significant variation.

The hard water control series clearly indicated that the experimental procedure had no effect on acid-base (Fig. 1) or ionic parameters (Fig. 3).

In soft-water-acclimated fish, day o values for ⁠, [HCO₃⁻] and [K⁺] were significantly lower than in fish acclimated to hard water (Table 2). However, pH₂ was unaffected, as were plasma Na+, Cl⁻ and Ca²⁺ levels (Table 2) despite the much lower concentrations of these ions in the water (Table 1).

Exposure to low external pH resulted in only a small drop in pHa (Fig. 4C); a third that in the hard-water-acclimated fish. Furthermore, some recovery from this mild acidosis was indicated, for day 4 and day 5 values were not significantly different from the pH_a at day o. Plasma [HCO₃⁻] showed a small significant decline on day 1 but was otherwise unchanged until day 5 when there was also a rise in blood lactate (Fig. ₄D). was unaffected (Fig. 4A).

The net ΔH⁺_b by day 5 (i.e. the sum of the daily values in Fig. 2B) was 2·0 ± 0·9 m- quiv/1. A large fraction of this quantity can be accounted for by the increase in blood lactate during this period (1-30 ± 0-35 m-equiv/1; Fig. 4D). Thus the net AH+₆ of external origin by day 5 (i.e., A lactate subtracted) amounted to 0 69 + 0-71 m-equiv/1, a value only 12% of that in acclimated hard water fish.

Plasma [Na⁺] and [Cl⁻] both declined progressively during the acid exposure in soft water (Fig. 5) to lower levels than the corresponding values in hard water (Fig. 3). The increase in [K+] was also more pronounced (Fig. 5C) than in hard water (Fig. 3C). Plasma [Ca²⁺] declined significantly from day 2 onwards, but a similar trend was seen in the control group (Fig. 5 D). There were no other significant changes within the control group (Figs. 4, 5).

These experiments were designed to test whether the influence of soft water on the effects of acid exposure, as observed above, would be accentuated or reduced by acute exposure to soft water. A second objective was to determine the time course of soft-water acclimation, by looking for changes in the control series relative to the two preceding control series (in which all parameters, with the possible exception of Ca²⁺, were unaffected by repetitive sampling).

The day o values for acid-base parameters in soft-water-unacclimated fish closely resembled those in soft-water-acclimated animals (Table 2). However, plasma [Na⁺] and [Cl⁻] were both significantly depressed and [K⁺] significantly elevated. Plasma [Ca²⁺] was unaffected (Table 2).

Environmental acid caused no significant blood acidosis over 5 days in these unacclimated fish (Fig. 6C). Indeed, on day 2 they exhibited a significant metabolic alkalosis relative to controls (Fig. 6C). However, [HCO₃⁻] did fall slightly on day and markedly on day 5 (Fig. 6B); pH_a was maintained by a corresponding drop in ⁠. Overall, the pattern of AH⁺₆ variation was similar to that in acclimated soft-water fish (Fig. 2B, C).

Plasma [Na⁺] and [Cl⁻] fell markedly during acid exposure (Fig. 7 A, B) and the day 5 values were virtually identical to those in soft-water-acclimated fish (Fig. 7 A, However, the decrease in [Na⁺] was perhaps underestimated, since the level rose in controls during this period (Fig. 7B). Changes in [K+] and [Ca²⁺] (Fig. 7C, D) were similar to those seen in the acclimated soft-water group (Fig. 5 C, D). Thus the ionic disturbances caused by external acid were at least as great in trout acutely exposed to soft water as in those acclimated to it for 2 weeks.

The unacclimated soft-water control trial indicated that a 2-week acclimation period was adequate. After only 5 days, all parameters except [Ca²⁺] (Figs. 6, were identical to those seen in acclimated fish (Table 2).

These experiments were designed to test whether the differences in physiological effect that were observed between hard and soft acidified water were due to the difference in [Ca²⁺]. The fish were acclimated to water in which [Ca²⁺] approximated the level found in hard water, whereas other ion levels were similar to those in soft water (Table 1).

After acclimation, values for acid-base parameters resembled those in fish acclimated to hard rather than soft water (Table 2). However, plasma [Na⁺] and [Cl⁻] were significantly lower than in either group.

The overall response to acid exposure in high [Ca²⁺], low [salt] water was very similar to that seen in hard water and very different from that seen in soft water. The fish developed a blood acid-base disturbance (Fig. 8) that was substantially larger than that seen in the soft-water experimental series (Figs. 4, 6). The pattern of acid accumulation (Fig. 2 D) was very similar to that in hard water (Fig. 2 A), with the bulk occurring over the first 2 days although the net loading by day 5 was higher, being 9·3 + o·8 m-equiv/1. The decrease in plasma [Cl⁻] (Fig. 9 A) was similar to that seen in the hard water series while the decrease in [Na⁺] (Fig. 9B) was significantly less. Plasma [K+] and [Ca²⁺] (Fig. 9C, D) also followed the hard water pattern (Fig. 3C, D). The fact that in high [Ca²⁺], low [salt] water the acid-base disturbance was even greater and the ionic disturbance even less than in acidified hard water may be related to the higher environmental Ca²⁺ in the former (2·7 vs 1·6 m-equiv/1, Table 1).

Haematocrit and plasma protein concentration decreased in the control groups (Fig. 10), as might be expected as a result of repetitive blood sampling. In acid-exposed groups, however, there was substantially less decline in haematocrit (Fig. 10A), and an increase in plasma protein (Fig. 10B).

Day 5 blood volumes (BV_f) in acid-exposed fish, calculated according to haematocrit changes (see Methods), were lower than the assumed initial blood volume of 5-0 ml/ 100 g. Data from the acidified hard-water series could not be analysed (see Methods) but in the high [Ca²⁺], low [salt] series, B K_fs averaged 2·94 ± 0·23 ml/100 g. A similar figure for BV_f (3·19 ± 0·28 ml/100 g) was obtained in the acidified soft-water series (acclimated plus unacclimated experiments). However, these BV_f estimates were quite variable within each series, probably partly because haematocrit can vary independently of blood volume. The BV_fs calculated from changes in plasma protein Xa.ee Methods) were much more consistent. In fish exposed to acid in high [Ca²⁺], Bv [salt] water, the average BV_f was 4·13 ±0·14 ml/100 g (N = 7), significantly higher than the average BV_f in acclimated soft-water fish (3·14 ± 0·17 ml/100 N = 7). These estimates represent a reduction in blood volume of 17·5 ± 2·7% and 37·4 ± 3·4%, respectively, from the assumed BV_iof 5·0 ml/100 g.

An additional eleven fish died during acid exposure. The data from these fish have been excluded from Figs. 1–10. While these fish exhibited a normal acid-base and ionic state on day o, the physiological disturbances they exhibited during acid exposure were generally accentuated relative to the survivors. Since these disturbances may more clearly indicate the key toxic mechanism(s) the final measurements prior to death for each mortality are recorded in Table 3. In the light of the preceding results the data are grouped as deaths in either high [Ca²⁺] (i.e., hard-water-acclimated and high [Ca²⁺], low [salt] acclimated series) or low [Ca²⁺] water (i.e. soft-water-acclimated and -unacclimated series). In high [Ca²⁺] water, 3 out of 27 fish (11 %) died within 6 days (one additional death was documented on Day 9). In low [Ca²⁺] water, 8 out of 16 fish (50%) died within the same period.

Within each group the terminal data are listed in Table 3 in order of the approximate time before death at which the final sample was taken. While these data are somewhat variable, they do indicate, when compared with control values (Table 2) the major physiological disturbances accompanying death at low pH. First, the major difference in the extent of the acid-base disturbance between high and low [Ca²⁺] water is clearly substantiated. In acidified high [Ca²⁺] water, blood pH_as within 20 h of death were at least 0 4 unit below normal and accompanied by a substantial depression of plasma [HCO₃⁻]. In low [Ca²⁺] water on the other hand, blood acid-base balance was little affected until 2 h from death. Secondly, in all low [Ca²⁺] animals, except fish F, plasma [Na+] and [Cl^-] were substantially depressed (on average by 37 and 45 m-equiv/1, respectively) below acclimated soft-water control levels (Table 2). The one high [Ca²⁺] animal in which terminal ion measurements were obtained showed a similar depression, but the data are too limited to draw any conclusions. Thirdly, in both high and low [Ca²⁺] water, the terminal haematocrits and plasma protein concentrations (where measured) were, for the most part, substantially higher than control values. Furthermore, a trend of increasing haematocrit with increasing proximity to death was apparent. These results suggest a major and progressive loss of plasma water as death approaches. Fourthly, K+ levels tended to rise as much as twofold above normal while Ca²⁺ levels tended to fall. Finally, there was no particular evidence of a major disturbance in gas exchange or transport. Arterial was elevated in only three fish and then only to levels roughly corresponding to normal venous levels (C. M. Wood, unpublished data). Similarly, the elevation in blood lactate which would indicated tissue hypoxia and impaired blood O₂ transport was relatively minor (cf. Kobayashi & Wood, 1980) in those animals in which this parameter was measured.

At acutely toxic environmental pH levels (3·0, 3·2), median lethal resistance times for fingerling trout were significantly greater in hard water ([Ca²⁺] = 3·3 m-equiv/1) than in soft water ([Ca²⁺] = 0·2 m-equiv/1) (Fig. 11). At slightly higher pH levels (3·4, 3·6) the toxicity curves coincided, while at still higher levels the situation was reversed. Trout survived significantly longer in soft water than in hard water at pH 3·8 and 4·0. The same was true at pH 4·2 and 4·4, though statistical tests could not be performed because the majority of the soft-water animals survived beyond the end of the 7-day experimental period. This should not be interpreted as indefinite survival; indeed the results provide no indication that a true incipient lethal threshold (Sprague, 1969) was defined within the 7-day period.

There were no consistent differences between hard and soft water trials at comparable pH levels in the slope functions of the log time vs probit mortality curves (Litchfield, 1949) from which the data in Fig. 11 are derived. However, within each water type, the slope functions increased progressively and significantly with pH, from about I·I at pH 3·0 to 1·4 at pH 4·0.

At a mean pH of 4·3, the physiological disturbances in the rainbow trout are clearly dependent on the level of calcium in the environment, and not the levels of other ions (at least over the ranges tested here). High [Ca²⁺] was associated with a marked blood acidosis and a relatively small plasma ionic disturbance as found previously in trout (Neville, 1979) and carp (N. Heisler, personal communication). The reverse was found with low [Ca²⁺]. The acclimated soft-water series ([Ca²⁺] = 0·3 m-equiv/1), the acclimated hard water series ([Ca²⁺] = 1·6 m-equiv/1) and the acclimated high [Ca²⁺], low [salt] water series ([Ca²⁺] = 2·7 m-equiv/1) form a sequence of increasing calcium levels in which this phenomenon is well illustrated (Figs. 1, 2, 3, 4, 5, 8, 9). Low [Ca²⁺] also seemed to be associated with a greater reduction in blood volume during acid exposure. Clearly, these results indicate that the effect of low external pH on ion regulation in fish cannot be considered in isolation from its effect on acid-base regulation, and neither effect can be considered in isolation from the influence of external [Ca²⁺]. Such interrelations are to be expected, since it is known that ionoregulation and acid-base regulation are’ coupled to some degree by electroneutral exchanges at the gills (Na⁺ vs, H⁺ or NH₄+; Kerstetter, Kirschner & Rafuse, 1970; Maetz, 1973; Payan, 1978 and Cl⁻ vs HC0₃⁻ or OH⁻; Kerstetter & Kirschner, 1972; DeRenzis & Maetz, 1973; DeRenzis, 1975) and that Ca²⁺ is an important modulator of branchial ion and water permeability in fish (Isaia & Masoni, 1976).

A number of studies have shown that exposure of fish to acidic external pH (Ca²⁺ level unspecified; Packer & Dunson, 1970, 1972; Maetz, 1973; 1-om-equiv/1 Ca²⁺; McWilliams & Potts, 1978) results in an inhibition of active Na⁺ influx and a stimulation of its diffusional efflux, leading to net Na+ loss and reduction of body Na+ levels. Reductions in plasma [Na⁺] occurred in all acid-exposed groups in the present study (Figs. 3B, 5B, 7B, 9B). The smaller reduction in plasma [Cl⁻] than [Na⁺] in the trout at pH 4-3 in high [Ca²⁺] water (Figs. 3, 9) suggests that Cl diffusional losses and/or active transport were less affected than Na⁺ fluxes, at least under high [Ca²⁺] conditions. Chloride fluxes under acidic conditions have not previously been examined to a similar extent but Maetz (1973) showed that Cl^- net flux was not affected by a small reduction in external pH (from 7·2 to 6·1) whereas Na⁺ influx and net flux were significantly reduced.

A reduction in environmental [Ca²⁺] has been shown to increase the passive efflux from fish of both Na+ (Potts & Fleming, 1971; Cuthbert & Maetz, 1972; Eddy, 1975) and Cl⁻ (Eddy, 1975). In solutions near neutrality the increased effluxes were compensated by increased influxes so that no net loss of either ion occurred (Cuthbert & Maetz, 1972; Eddy, 1975). This probably explains the observation (Table 2) that plasma Na+ and Cl⁻ levels were initially reduced by acute soft-water exposure but were entirely compensated after 2 weeks acclimation. The more extensive losses of both plasma Na⁺ and Cl⁻ which occurred with acid exposure in low [Ca²⁺] water (Fig. 5) relative to high [Ca²⁺] water (Figs. 3, 9) must reflect a greater imbalance between influxes and effluxes.

In addition to raising diffusive salt losses, low [Ca²⁺] can be expected to increase the passive penetration of protons at the gills during acid exposure (McWilliams Potts, 1978). However, the blood acidosis accompanying low external pH was neg⁵ ligible in low [Ca²⁺] relative to high [Ca²⁺] (Figs. 1, 2, 4). This suggests that a major elevation of proton excretion took place during acid exposure in low [Ca²⁺] water. This could be linked to Na+ uptake (see above) which is elevated in low [Ca²⁺] water (Cuthbert & Maetz, 1972; Eddy, 1975). An elevation of Na+/H+ exchange could be the reason for the increase in plasma Na⁺ level observed in the unacclimated soft-water controls (Fig. 7) and for the metabolic alkalosis observed on day 2 in the acid-exposed group (Fig. 6).

Implicit in the above arguments are three conditions. First, activation of Na+ transport by low [Ca²⁺] (Eddy, 1975) must predominate over the inhibition of Na⁺ transport by low pH (Packer & Dunson, 1970; Maetz, 1973; McWilliams & Potts, 1978). Secondly, the stimulation of Na+ efflux by low pH (Packer & Dunson, 1970; McWilliams & Potts, 1978) must be greater than the stimulation of Na⁺ influx by low [Ca²⁺] to explain the large depression in plasma Na⁺ levels (Figs. 5B, 7B). Thirdly, any activation of Cl⁻/base exchange at low Ca²⁺ levels and acid pH must be less than that of Na⁺/proton exchange. The latter appears to be the case during the correction of an acidosis associated with salt depletion in goldfish (DeRenzis & Maetz, 1973). It is also supported by the significantly greater depression of plasma [Cl⁻] by day 5 of acid exposure in both low [Ca²⁺] series (Figs. 5, 7).

In addition to the above mechanisms it seems highly probable that kidney function will also play a significant role in the overall compensation to acid stress (Wood & Caldwell, 1978; Kobayashi & Wood, 1980). Clearly, direct and simultaneous measurements of branchial and renal ion and acid fluxes together with continuous assessment of blood acid-base and electrolyte status will be required to fully identify the complex interactions between water [Ca²⁺], water pH, ionoregulation and acid-base regulation. Nevertheless, the present findings do establish the critical importance of environmental [Ca²⁺], usually the major component of water hardness, in determining the nature of the response to acid exposure.

As Sprague (1971) has pointed out, any environmental stress, such as low pH, will produce a host of physiological disturbances, but most will be within the range of adaptation of the animal and therefore of little influence on individual survival. If the effects are to be used for predictive purposes (e.g. sublethal bioassays, field surveys) it will be necessary to understand which effects cause death, under a variety of conditions.

At critically low pH levels (< 4·0) where mortality is 100% and death occurs within hours rather than days, a failure of O₂ delivery to the tissues is probably of primary importance. A pronounced accumulation of mucus on the gills (Plonka & Neff, 1969; Daye & Garside, 1976; Ultsch & Gros, 1979) and a sloughing of gill epithelial tissue (Daye & Garside, 1976) may severely impair branchial O₂ diffusion. This, combined with a marked reduction in blood O₂ capacity (Root effect), due to massive acidosis (Packer & Dunson, 1970, 1972; Packer, 1979; N. Heisler, personal communication), results in eventual cellular anoxia. However, such pH levels are rarely encountered by fish in the wild. Much more common will be chronic exposal pHs in the range 4·5–6·0 which occurs when poorly buffered soft-water lakes are gradually titrated to the endpoint (∼ 4-5) of the HCO₃⁻/CO₂ buffer curve by longterm acid input. This may be accompanied by episodic excursions down to ∼ pH 4·0 during rainstorms and snow melt (Leivestad et al. 1976; Harvey, 1979).

The present results clearly show that O₂ delivery failure is not the cause of death at these more moderate pH levels. Mucus accumulation on the gills was not seen and blood levels remained normal. Blood lactate levels were either unaffected (hard water; Fig. ID) or marginally elevated (soft water; Fig. 4D) after 5 days exposure. Terminal measurements on dying fish (Table 3) were well within the normal range of variation (Kobayashi & Wood, 1980; C. M. Wood, unpublished results). These findings would also seem to rule out possible circulatory failure associated with reductions in blood volume.

In those few fish which succumbed in high [Ca²⁺] water in the present study, the terminal blood acid-base disturbance was slightly more severe than in the survivors (Table 3; Figs. 1, 9). The extent of this pH depression (about 0 ·4 units) is within the short-term tolerance of the trout (Holeton & Randall, 1967; Kiceniuk & Jones, 1977; Kobayashi & Wood, 1980) but chronic acidosis of this magnitude may be more serious, particularly if combined with an ionic disturbance. A severe ionic disturbance was noted in one fish that died at pH 4 ·3 in high [Ca²⁺]. Although these data are too limited to be conclusive we have recently noted (D. G. McDonald & C. M. Wood, unpublished results) similar ionic disturbances associated with deaths in hard water acidified with H₂SO₄.

In low Ca²⁺ water the results (Table 3) more clearly suggest that ionoregulatory failure is the sole toxic mechanism of low pH; a conclusion supported by field studies (Leivestad & Muniz, 1976; Leivestad et al. 1976). In the present study, percentage mortality was markedly increased relative to high [Ca²⁺] water at the same acidic pH (50% vs 11 % over 6 days). In both surviving and dying fish, blood acid-base status was little affected (Figs. 4, 6; Table 3) but plasma ion levels were severely disturbed (Figs. 5, 7; Table 3). Prior to death (Table 3), plasma Na+ and Cl ⁻ concentrations were uniformly depressed by about 25% and 30% respectively while [K+] showed an almost 1 ·6-fold increase.

Toxicity trials with fingerling trout were performed to determine whether these same differences in acid toxicity and toxic mechanism could also be detected by the classical lethality bioassay. In these tests, differences in toxicity are detected as differences in median lethal resistance times while changes in the slope functions of log time vs probit mortality relationships are interpreted as changes in toxic mechanisms (Sprague, 1969). Progressive changes in slope function were found with increasing pH in both hard and soft water, which tends to confirm that a gradual transition from one key toxic mechanism (e.g. oxygen delivery failure) to another (e.g. acid-base and/or ion regulatory failure) may have occurred. However, there were no consistent differences in the slope functions between hard and soft water indicative of a difference in modes of toxicity between the two environments. Furthermore, while [Ca²⁺] prolonged survival at acutely toxic pH levels (3 ·0, 3 ·2) the reverse occurred at more moderate levels (3-8-4’4). Very similar data were resented by Lloyd & Jordan (1964) in their comparison of an even wider range calcium levels than that examined here. Although a number of factors (variations in age, size, degree of restraint, metabolic rate, etc.) may have contributed to t differences between the toxicological and physiological studies, the fact that they gave fundamentally different results emphasizes that extreme caution must be used in extrapolating from one type of study to the other. Nevertheless, there is some common ground in so far as both methods indicate the critical importance of environmental calcium levels. Relatively minor variations in water [Ca^a+] may thus spell the difference between survival and eventual extinction for fish populations under acid stress in the wild.

We thank C. L. Milligan and M. Graham for excellent technical assistance, Dr G. P. Harris for the loan of equipment and Dr P. V. Hodson of the Canada Center for Inland Waters for heavy metal analyses. Financial support was provided by a strategic grant in environmental toxicology from the Natural Sciences and Engineering Research Council of Canada and by grants from the Canadian National Sportsmen’s Fund and Fisheries and Oceans Canada.