0306~4492/92 $5.00+ 0.00 0 1992Pergamon Press plc
Camp. Biochem. Physiol. Vol. IOlC, No. 2, pp. 275-282, 1992
Printed in Great Britain
ACUTE TOXICITY OF AMMONIA SALMON (SALMO S&AR)
TO ATLANTIC PARR
MAI BRITT KNOPH Institute of Aquaculture and Fish Diseases, The Norwegian College of Veterinary Medicine, Box 8146 Dep., 0033 Oslo 1, Norway 13 May 1991)
(Received
Abstract-l. Atlantic salmon (Salmo s&r) parr were exposed to (NH&SO, solutions in static systems with aerated, soft water for 96 hr. The 96 hr-LcSOfor un-ionized ammonia (expressed as mg/l NH,-N) ranged from 0.031 (2.1”C) to 0.111 (17.1”C) at pH 6.0 and from 0.030 (1.8”C) to 0.146 (12S”C) at pH 6.4. 2. No mortality was found in a KC1 solution and a physiological salt solution with chloride concentrations approximately equivalent to the chloride concentration in a NH&l solution giving 45% mortality, and to the ammonia concentration in a (NH&SO., solution giving 35% mortality, all solutions tested at pH 6.0 and 2°C
INTRODUCTION
In this paper un-ionized ammonia is represented as NH3, the ionized form as ammonium (NH,+), and their sum as total ammonia (TA). Ammonia is an unspecific term meaning NH,, NH,+ or both. Concentrations are reported in terms of mg N/l, and referred data not in this form were converted by
multiplying by the appropriate factor (NH,-N = 0.8235 NH3). Acute ammonia intoxication may occur during transport of fish (Amend et al., 1982) or in polluted waters (Ruffier et al., 1981; Tarazona et al., 1987). High density stocking and/or water re-use in fish farms may also lead to ammonia toxicity, acute (Robinette, 1976), or more often chronic (Burrows, 1964; Klontz et al., 1985). Recent reviews of ammonia toxicity on fish are given by Meade (1985) and U.S. EPA (1985, 1989). The production of Atlantic salmon in Norway in 1990 was about 158,000 tons, and the production of Atlantic salmon is increasing in other countries as well. Information on ammonia toxicity to Atlantic salmon is sparse, but indicates that Atlantic salmon is among the most vulnerable fish species. Authors describing ammonia toxicity to Atlantic salmon are Herbert and Shurben (1965), Samylin (1969), Alabaster et al. (1979, 1983) and Morrison and Piper (1988). Earlier investigations have consistently demonstrated that un-ionized ammonia toxicity decreases with increasing temperature (Brown, 1968; Ministry of Technology, 1968; Thurston et al., 1983; Erickson, 1985). Toxicity of NH, is also known to decrease with increasing pH (Thurston et al., 1981b; Thurston and Russo, 1983; Broderius et al., 1985; Sheehan and Lewis, 1986). Increased Ca2 + concentration in rearing water is shown to increase the tolerance of channel catfish (Zctahrus punctatus) to ammonia (Tomasso et al., 1980). Bradley and Rourke (1985) found that the addition of external Na + when rearing rainbow trout (Salmo gairdneri Richardson) in low external Na+
concentrations resulted in a decrease in plasma NH,+ and an increase in plasma Na+ concentration. No experiments with ammonia toxicity on salmonids in the extremely soft waters commonly found in Norway are known. The aim of the study described below was to find levels of ammonia acutely toxic to Atlantic salmon parr in soft water with hardness < 10 mg/l as CaCO,, low Na + concentration and a pH of 6.0-6.5. The effect of temperature on ammonia toxicity at these conditions was given special attention by testing several temperatures in the range 2-17°C. Toxicity of NH, to a salmonid in this water type, especially at the lowest temperatures, is expected to be high. Sheehan and Lewis (1986) found an apparent increase in toxicity of NH, to channel catfish at lower pH to be due to osmotic effects of test salt, rather than specifically to NH4+. A similar test was included in the present study to see if this was also the case for the Atlantic salmon.
MATERIALSAND
METHODS
The experiments were conducted at the Norwegian Institute for Water Research (NIVA) in Oslo. The tests were performed according to the guidelines for acute toxicity tests given by Parrish (1985). Atlantic salmon parr of the Drammen river and Sunndalsarra strain were obtained from the fish farm of Oslo Angling Administration (OFA) in Ssrkedalen, Oslo. The fish were acclimated for at least 2 weeks to the control water quality and stocking conditions before being used in experiments, and fed at least three times a day with a commercial dry salmon feed. No fish died during the transport or acclimation period. Fish from both strains were aged 1 + , and fish length ranged from 4.8 to 9.2 cm. The temperature of the rearing water increased from 1°C in February to 15°C in late June. Daily water temperature fluctuations were less than 0.5”C. Minimum and maximum pH observed in the rearing water was 6.15 and 6.40. The tests were conducted in twelve 40-l glass aquaria filled with 30 1 of test solution. By aerating the oxygen saturation were kept at 90-100%. The aquaria were placed in a walk-in environmental chamber which made it possible to keep the temperature of the test solutions within f 1.0%. Continu-
275
276
MAI BRITT KNOPH
ous low light was used to avoid abrupt on-and-off switching of light at inspection and sampling. The fish were acclimated to test temperature in separate trays. During temperature acclimation the water temperature was not changed more than SC/24 hr. The fish were held at the test temperature for minimum 24 hr before the start of a test. The fish were starved during temperature acclimation and the 96-hr test period, which gave a total of 67 days starvation. A test temperature series of 2,7, 12 and 17°C was selected. Two parallels of five test concentrations and two control treatments were used in each test. The test salt used in the 96 hr-LcSO tests was (NH&SO, from Merck. After addition of test salts to the aquaria, the pH of the solutions was corrected by adding drops of concentrated NaOH or H,SO,. In some tests it was necessary to repeat this pH correction at 24 hr intervals to maintain the pH within acceptable limits. All chemicals used were of pro analysis quality. Chemical and physical data of test solutions/test water are listed in Table 1. The loading of fish in aquaria was in the range of 0.4 to 1 g/l. At the start of a test 10 fish were transferred directly from temperature-acclimation trays to each of the 12 aquaria containing the test solutions. Dead fish were counted and removed every 24 k 1 hr from the beginning of a test until termination. Temperature, pH and dissolved O2 were measured at start of a test and then every 24 hr in each aquarium. TA-N was measured 2 or 3 times in each aquarium. Fish were considered dead if they showed no respiratory movements or reaction to gentle prodding. Most chemical analyses were performed according to the standard methods of APHA ef al. (1985). Total ammonia was measured with an Orion Research 901 microprocessor ionanalyser and an Orion 95-12 ammonia-specific probe. The reproducibility of ammonia measurements was tested on a sample of a test solution with 200 mg/l TA-N, and the relative average deviation from the mean found to be + 2% (N = 20). pH was measured with a Radiometer pH-meter model 29 and a Radiometer GK240lC combined electrode and temperature measured with a mercury thermometer. Salt tests Mortality in a KC1 solution, a physiological salt solution, a NH&l solution and a (NH&SO., solution was tested. The
chloride and ammonium concentrations of the solutions were approximately equivalent to the ammonium concentration in a (NH&SO, solution giving about 50% mortality in test A (Table 3). Two parallels of each solution and two controls were tested. Test methods were as described earlier, except for a duration of 24 hr. Dead fish were counted and removed and temperature, pH, TA-N and dissolved 0, measured at 0, 12 and 24 hr. All solutions were tested at pH 6.0 (5.956.10) and 2°C (1.8-2.2) with fish of the Sunndalsera strain. Data treatment Averaging of pH was accomplished by converting the pH values to their corresponding hydrogen ion concentrations, averaging these, and then reconverting the averages to pH units. Un-ionized ammonia concentrations were calculated based on the theory of Stumm and Morgan (1981), using the mean TA-N concentration in a treatment and pH and temperature at specific time as input. LC~~S with 95% confidence intervals were estimated with the trimmed Spearman-Kirber method (Hamilton et al., 1977). Twoway analysis of variance, with replication and regression analysis, was performed with the Minitab computer program. The adequacy of the simple linear model for 96 hrLC~~ vs temperature regressions was tested with the method of Afifi and Azen (1979). The coefficient of determination, r2, was estimated according to the method of Sokal and Rohlf (1980). Confidence intervals of 95% for the regression lines and 95% prediction intervals were calculated according to Sokal and Rohlf (1980) using the method for more than one value of Y for each value of X. The equality of estimated LC~,,S (logarithmically transformed values) was tested with test for equality of the means of two samples whose variances are assumed to be unequal (Sokal and Rohlf, 1980). Sample size was taken to be the number of concentrations used for calculating the LC~~ of a test, and was accordingly corrected for trimming. Standard error of the In (Lc,,)-estimate was calculated from its logarithmic 95% confidence interval which was estimated with the trimmed Spearman-Klrber method. A significance level of 0.05 was consequently used. Range -finding tests
Table 1. Chemical and physical characteristics of test solutionssamples taken from the aquaria with the highest test concentration in each series at the time of 100% mortality MeaIl value
Range
0.86 0.261.90 TA-N* w/l 0.10 0.091-0.108 Alkalinityt mg/l taco, 5.1 4.g6.8 Hardness mg/l C&O, Conductivity* mS/m 3.5 0.71 0.35-1.0 Turbidity FTU < 1.27 1.09-s 1.44 mg/l CO& CO2 IO.43
Camp. Biochem. Physiol. Vol. IOlC, No. 2, pp. 275-282, 1992
Printed in Great Britain
ACUTE TOXICITY OF AMMONIA SALMON (SALMO S&AR)
TO ATLANTIC PARR
MAI BRITT KNOPH Institute of Aquaculture and Fish Diseases, The Norwegian College of Veterinary Medicine, Box 8146 Dep., 0033 Oslo 1, Norway 13 May 1991)
(Received
Abstract-l. Atlantic salmon (Salmo s&r) parr were exposed to (NH&SO, solutions in static systems with aerated, soft water for 96 hr. The 96 hr-LcSOfor un-ionized ammonia (expressed as mg/l NH,-N) ranged from 0.031 (2.1”C) to 0.111 (17.1”C) at pH 6.0 and from 0.030 (1.8”C) to 0.146 (12S”C) at pH 6.4. 2. No mortality was found in a KC1 solution and a physiological salt solution with chloride concentrations approximately equivalent to the chloride concentration in a NH&l solution giving 45% mortality, and to the ammonia concentration in a (NH&SO., solution giving 35% mortality, all solutions tested at pH 6.0 and 2°C
INTRODUCTION
In this paper un-ionized ammonia is represented as NH3, the ionized form as ammonium (NH,+), and their sum as total ammonia (TA). Ammonia is an unspecific term meaning NH,, NH,+ or both. Concentrations are reported in terms of mg N/l, and referred data not in this form were converted by
multiplying by the appropriate factor (NH,-N = 0.8235 NH3). Acute ammonia intoxication may occur during transport of fish (Amend et al., 1982) or in polluted waters (Ruffier et al., 1981; Tarazona et al., 1987). High density stocking and/or water re-use in fish farms may also lead to ammonia toxicity, acute (Robinette, 1976), or more often chronic (Burrows, 1964; Klontz et al., 1985). Recent reviews of ammonia toxicity on fish are given by Meade (1985) and U.S. EPA (1985, 1989). The production of Atlantic salmon in Norway in 1990 was about 158,000 tons, and the production of Atlantic salmon is increasing in other countries as well. Information on ammonia toxicity to Atlantic salmon is sparse, but indicates that Atlantic salmon is among the most vulnerable fish species. Authors describing ammonia toxicity to Atlantic salmon are Herbert and Shurben (1965), Samylin (1969), Alabaster et al. (1979, 1983) and Morrison and Piper (1988). Earlier investigations have consistently demonstrated that un-ionized ammonia toxicity decreases with increasing temperature (Brown, 1968; Ministry of Technology, 1968; Thurston et al., 1983; Erickson, 1985). Toxicity of NH, is also known to decrease with increasing pH (Thurston et al., 1981b; Thurston and Russo, 1983; Broderius et al., 1985; Sheehan and Lewis, 1986). Increased Ca2 + concentration in rearing water is shown to increase the tolerance of channel catfish (Zctahrus punctatus) to ammonia (Tomasso et al., 1980). Bradley and Rourke (1985) found that the addition of external Na + when rearing rainbow trout (Salmo gairdneri Richardson) in low external Na+
concentrations resulted in a decrease in plasma NH,+ and an increase in plasma Na+ concentration. No experiments with ammonia toxicity on salmonids in the extremely soft waters commonly found in Norway are known. The aim of the study described below was to find levels of ammonia acutely toxic to Atlantic salmon parr in soft water with hardness < 10 mg/l as CaCO,, low Na + concentration and a pH of 6.0-6.5. The effect of temperature on ammonia toxicity at these conditions was given special attention by testing several temperatures in the range 2-17°C. Toxicity of NH, to a salmonid in this water type, especially at the lowest temperatures, is expected to be high. Sheehan and Lewis (1986) found an apparent increase in toxicity of NH, to channel catfish at lower pH to be due to osmotic effects of test salt, rather than specifically to NH4+. A similar test was included in the present study to see if this was also the case for the Atlantic salmon.
MATERIALSAND
METHODS
The experiments were conducted at the Norwegian Institute for Water Research (NIVA) in Oslo. The tests were performed according to the guidelines for acute toxicity tests given by Parrish (1985). Atlantic salmon parr of the Drammen river and Sunndalsarra strain were obtained from the fish farm of Oslo Angling Administration (OFA) in Ssrkedalen, Oslo. The fish were acclimated for at least 2 weeks to the control water quality and stocking conditions before being used in experiments, and fed at least three times a day with a commercial dry salmon feed. No fish died during the transport or acclimation period. Fish from both strains were aged 1 + , and fish length ranged from 4.8 to 9.2 cm. The temperature of the rearing water increased from 1°C in February to 15°C in late June. Daily water temperature fluctuations were less than 0.5”C. Minimum and maximum pH observed in the rearing water was 6.15 and 6.40. The tests were conducted in twelve 40-l glass aquaria filled with 30 1 of test solution. By aerating the oxygen saturation were kept at 90-100%. The aquaria were placed in a walk-in environmental chamber which made it possible to keep the temperature of the test solutions within f 1.0%. Continu-
275
276
MAI BRITT KNOPH
ous low light was used to avoid abrupt on-and-off switching of light at inspection and sampling. The fish were acclimated to test temperature in separate trays. During temperature acclimation the water temperature was not changed more than SC/24 hr. The fish were held at the test temperature for minimum 24 hr before the start of a test. The fish were starved during temperature acclimation and the 96-hr test period, which gave a total of 67 days starvation. A test temperature series of 2,7, 12 and 17°C was selected. Two parallels of five test concentrations and two control treatments were used in each test. The test salt used in the 96 hr-LcSO tests was (NH&SO, from Merck. After addition of test salts to the aquaria, the pH of the solutions was corrected by adding drops of concentrated NaOH or H,SO,. In some tests it was necessary to repeat this pH correction at 24 hr intervals to maintain the pH within acceptable limits. All chemicals used were of pro analysis quality. Chemical and physical data of test solutions/test water are listed in Table 1. The loading of fish in aquaria was in the range of 0.4 to 1 g/l. At the start of a test 10 fish were transferred directly from temperature-acclimation trays to each of the 12 aquaria containing the test solutions. Dead fish were counted and removed every 24 k 1 hr from the beginning of a test until termination. Temperature, pH and dissolved O2 were measured at start of a test and then every 24 hr in each aquarium. TA-N was measured 2 or 3 times in each aquarium. Fish were considered dead if they showed no respiratory movements or reaction to gentle prodding. Most chemical analyses were performed according to the standard methods of APHA ef al. (1985). Total ammonia was measured with an Orion Research 901 microprocessor ionanalyser and an Orion 95-12 ammonia-specific probe. The reproducibility of ammonia measurements was tested on a sample of a test solution with 200 mg/l TA-N, and the relative average deviation from the mean found to be + 2% (N = 20). pH was measured with a Radiometer pH-meter model 29 and a Radiometer GK240lC combined electrode and temperature measured with a mercury thermometer. Salt tests Mortality in a KC1 solution, a physiological salt solution, a NH&l solution and a (NH&SO., solution was tested. The
chloride and ammonium concentrations of the solutions were approximately equivalent to the ammonium concentration in a (NH&SO, solution giving about 50% mortality in test A (Table 3). Two parallels of each solution and two controls were tested. Test methods were as described earlier, except for a duration of 24 hr. Dead fish were counted and removed and temperature, pH, TA-N and dissolved 0, measured at 0, 12 and 24 hr. All solutions were tested at pH 6.0 (5.956.10) and 2°C (1.8-2.2) with fish of the Sunndalsera strain. Data treatment Averaging of pH was accomplished by converting the pH values to their corresponding hydrogen ion concentrations, averaging these, and then reconverting the averages to pH units. Un-ionized ammonia concentrations were calculated based on the theory of Stumm and Morgan (1981), using the mean TA-N concentration in a treatment and pH and temperature at specific time as input. LC~~S with 95% confidence intervals were estimated with the trimmed Spearman-Kirber method (Hamilton et al., 1977). Twoway analysis of variance, with replication and regression analysis, was performed with the Minitab computer program. The adequacy of the simple linear model for 96 hrLC~~ vs temperature regressions was tested with the method of Afifi and Azen (1979). The coefficient of determination, r2, was estimated according to the method of Sokal and Rohlf (1980). Confidence intervals of 95% for the regression lines and 95% prediction intervals were calculated according to Sokal and Rohlf (1980) using the method for more than one value of Y for each value of X. The equality of estimated LC~,,S (logarithmically transformed values) was tested with test for equality of the means of two samples whose variances are assumed to be unequal (Sokal and Rohlf, 1980). Sample size was taken to be the number of concentrations used for calculating the LC~~ of a test, and was accordingly corrected for trimming. Standard error of the In (Lc,,)-estimate was calculated from its logarithmic 95% confidence interval which was estimated with the trimmed Spearman-Klrber method. A significance level of 0.05 was consequently used. Range -finding tests
Table 1. Chemical and physical characteristics of test solutionssamples taken from the aquaria with the highest test concentration in each series at the time of 100% mortality MeaIl value
Range
0.86 0.261.90 TA-N* w/l 0.10 0.091-0.108 Alkalinityt mg/l taco, 5.1 4.g6.8 Hardness mg/l C&O, Conductivity* mS/m 3.5 0.71 0.35-1.0 Turbidity FTU < 1.27 1.09-s 1.44 mg/l CO& CO2 IO.43
