Transbranchial Ammonia Gradients and Acid-Base Responses to High External Ammonia Concentration in Rainbow Trout (Oncorhynchus Mykiss) Acclimated to Different Salinities

Teleost fish excrete the majority of their nitrogenous waste as ammonia (Smith, 1929). By far the largest proportion of this ammonia output occurs at the gills (Smith, 1929; Sayer and Davenport, 1987) and results from clearance of ammonia from the blood (Payan and Matty, 1975; Cameron and Heisler, 1983; Ogata and Murai, 1988) as it crosses the arterio-arterial circuit through the lamellae (Payan et al. 1984). Under normal conditions of low ambient [ammonia] the transbranchial ammonia gradients (⁠ and ΔNH₄⁺) are positive (from blood to water) and ammonia excretion can theoretically be achieved by a variety of mechanisms; transcellular and/or paracellular diffusion of NH₃, paracellular diffusion of NH₄⁺, apical or basolateral NH₄⁺/Na⁺(or H⁺) exchange, and basolateral NH₄⁺/Na⁺/2C1⁻ cotransport (see recent reviews by Cameron and Heisler, 1985; Evans and Cameron, 1986; Randall and Wright, 1987). Despite numerous attempts to quantify the importance of each of these transport mechanisms, the subject of branchial ammonia transfer across fish gills remains controversial.

Transbranchial ammonia gradients can be experimentally reversed by exposure to elevated external ammonia concentrations. When and ΔNH₄⁺ are reversed both freshwater and marine teleosts are able to maintain the total ammonia concentration (TAmm=NH₄⁺+NH₃) in their extracellular fluid below that of the external environment (Cameron and Heisler, 1983; Cameron, 1986; Claiborne and Evans, 1988). These authors attributed this to active exchange of NH₄⁺ for an external counterion (Na⁺ or H⁺), which is sufficient to balance the passive inward diffusion of ammonia.

The blood acid-base changes and the ammonia load induced by ammonia exposure will depend upon the relative permeability of the fish to NH₃ and NH₄⁺ and the mechanism used in the active excretion of NH₄⁺. In freshwater teleosts, when transbranchial ammonia gradients are reversed, the passive uptake of ammonia occurs predominantly via the inward diffusion of NH₃ (Fromm and Gillette, 1968; Cameron and Heisler, 1983; Cameron, 1986; Claiborne and Evans, 1988; Avella and Bornancin, 1989). This has generally been explained by the high diffusibility of lipid-soluble, gaseous NH₃ through cell membranes (Jacobs, 1940; Pitts, 1973) in contrast to the relatively low permeability of cell membranes and ‘tight’ epithelia (such as freshwater fish gills; Sardet, 1980; Pisam et al. 1987) to the hydrated, charged and poorly lipid-soluble ammonium ion. However, the gills of seawater-adapted teleosts have a higher ionic permeability than those of fresh water teleosts (by a factor of 10 or more; Girard and Payan, 1980; Evans, 1979, 1984), which may include an increased permeability to NH₄⁺ (Goldstein et al. 1982; Claiborne and Evans, 1988; Evans et al. 1989). In seawater-adapted fish the uptake of ammonia when gradients are reversed may, therefore, be augmented by a significant inward diffusion of NH₄⁺. To test this hypothesis we have compared the ammonia loading and blood acid-base response during exposure to high external ammonia concentration in rainbow trout acclimated to different salinities ranging from fresh water to full-strength sea water (35 ‰). Thus, for the first time, the influence of salinity on branchial ammonia transfer has been investigated within a single euryhaline species.

Rainbow trout, Oncorhynchus mykiss (Walbaum) (300-800g), were obtained from Zeals Fish Farm, Wiltshire. Following transportation to the ICI Environmental Laboratory, Brixham, Devon, they were transferred to 6001 grey fibreglass tanks and maintained in continuously flowing, dechlorinated Brixham tapwater ([Na⁺]≈0.6, [Cl²⁺]≈0.5, [Ca²⁺]≈0.8mequivl⁻¹; pH=7.0–7.8; T=10–18°C). Fish were kept on a maintenance diet (1 % body weight day⁻¹) of commercial trout pellets but were starved for 4 days prior to surgery (to avoid post-prandial changes in metabolic rate and excretion of nitrogenous waste during experiments; Jobling, 1981).

Trout used for freshwater experiments were transferred to 6001 acclimation tanks. Water temperature was adjusted to 15°C in steps of 2°C per day and then maintained at that temperature for at least 2 weeks prior to experiments.

A continuous supply of 33 % sea water was obtained by mixing dechlorinated tapwater and Torbay sea water at a ratio of 2:1 in a flow-through system. Freshwater trout were transferred to acclimation tanks and the salinity was raised to 12 ‰ ([Na⁺]≈170, [Cl⁻] ≈200mequivl⁻¹) over a period of 24 h. The temperature of the acclimation water was adjusted to 15°C in steps of 2°C per day, and the fish were allowed a minimum acclimation period of 2 weeks once the desired salinity and temperature had been reached.

Fish were transferred to acclimation tanks and the salinity of the incoming water was increased in steps to full-strength Torbay sea water (35 ‰, [Na⁺]≈460, [Cl⁻] ≈550mequivl⁻¹) over a period of 10 days. During this time, water temperature was adjusted to 12°C and the trout were then given a minimum of 4 weeks to acclimate to this salinity and temperature before use in experiments (the seawater temperature could not be maintained at 15 °C as in the previous two regimes, and 12°C proved to be the highest constant temperature attainable). Daily water samples were taken for monitoring of salinity (Kent Industrial Instruments Salinometer; model MC5).

To enable repeated blood sampling, fish were anaesthetised with a 100 mg I⁻¹ solution of MS222 (Sigma) dissolved in a medium of the appropriate salinity (buffered to approximately pH7.5 with NaHCO₃ in fresh water). Once anaesthetised, fish were fitted with a chronic indwelling dorsal aortic catheter (Soivio et al. 1972) whilst the gills were irrigated with a lower concentration of oxygenated anaesthetic (60 mg I⁻¹ MS222). Following surgery, catheters were filled with sodium-heparinised trout saline (Perry et al. 1984; heparin=50i.u. ml⁻¹) and fish were transferred to individual, darkened, Perspex respirometer tubes (40 cm × 10 cm) supplied with aerated water of the appropriate salinity and allowed to recover for a minimum of 36 h.

The experimental set-up was designed to allow exposure of fish to elevated levels of ammonia in a pH-statted flow-through system. Respirometers received water at a rate of 500–1000 ml min⁻¹ during experiments (according to fish size). When required, a stock solution of 250 mmol I⁻¹ ammonium sulphate was dosed into mixing cells situated immediately upstream from the respirometers using peristaltic pumps (Watson-Marlow) at 1/500 of the water flow rate to produce the desired water concentration of l.0mmoll⁻¹ total ammonia. Water pH was nominally maintained at 7.9 throughout all regimes by controlled addition of either 3 mol I⁻¹ HC1 (to sea water) or 2 mol I⁻¹ NaOH (to fresh water). Water pH was monitored using Corning pH and reference electrodes in conjunction with a Kent Industrial pH meter. The output from the meter controlled the speed of a programmable peristaltic pump (Watson-Marlow, 202U/AA), which dosed the acid or base. In addition to this control system, water pH was monitored in the mixing cells (Corning electrodes and meter) to give more accurate readings for the pH of water entering the respirometers.

Catheterised trout were exposed to l.0 mmoll⁻¹ [T_Amm] at pH7.9 for a single 24-h period, in fresh water, 33 % sea water (both at 15°C) and full-strength sea water (12°C). Approximately 1 h before exposure to high ammonia concentrations a single ‘control’ blood sample was taken. To follow the changes in acid-base status, plasma ammonia and ions, blood samples were subsequently taken after 1, 5 and 24 h of exposure to high ammonia.

Arterial blood samples (1.0 ml) were anaerobically drawn into chilled Hamilton gas-tight syringes. A 400 μl subsample was immediately centrifuged for 3 min at 13500g (MSE Microcentaur) for later plasma T_Amm and ion analysis. The remainder of the sample was used for the determination of whole-blood ⁠, haematocrit (Het), haemoglobin content ([Hb]), and the pH and total of both whole blood and plasma. The blood used for measurement (approximately 250 μl) was then returned to the animal followed by infusion of enough saline to replace the net blood volume removed (approximately 750μl). Whole-blood and true plasma pH were measured using a Radiometer G279/G2 glass capillary electrode coupled with a K497 calomel reference electrode. Both electrodes were thermostatted to the experimental temperature (15 or 12°C) and used in conjunction with a pH/blood gas monitor (Radiometer PHM73). Total CO₂ was measured on 50μl subsamples by the method of Cameron (1971) using a electrode (Radiometer E5037) connected to the same pH/blood gas monitor as above. For both plasma pH and measurements, plasma was taken from blood centrifuged anaerobically in microhaematocrit tubes used for Het determination (see below). Arterial blood oxygen tension and water were measured using oxygen electrodes (Radiometer E5046), both thermostatted to experimental temperature, and oxygen meters (Strathkelvin). For each blood sample Hct was determined in duplicate, using 80μl sodium-heparinised microhaematocrit tubes centrifuged at 12000g in a haematocrit centrifuge (Hawksley) for 2 min.

[Hb] was determined in duplicate on 20 μl samples of blood using the cyanomethaemoglobin method (Sigma kit no. 525). Absorbance of the sample plus reagent solution was measured on a dual-beam spectrophotometer at 540 nm (Kontron Uvikon 860). Plasma [T_Amm] was measured on 200μl subsamples using a specific enzymatic assay (G1DH/NADH; Sigma 170-UV at 340 nm) and a dualbeam spectrophotometer (Kontron Uvikon 860). Another 40μl subsample of plasma was immediately diluted 100 times in deionised water and frozen (–20°C) for later measurement of the plasma cations Na⁺, K⁺, Ca²⁺ and Mg²⁺ by atomic absorption spectrophotometry (Pye Unicam SP9). Plasma [Cl⁻] was measured by amperometric titration on 50μl samples of undiluted plasma (Aminco-Cotlove automatic chloride titrator).

The water [T_Amm] was measured using an ammonia electrode (Philips IS 570-NH3 connected to an Orion specific ion meter, model 407A) after alkalisation of 50 ml samples with 1ml of 10 mol I⁻¹ NaOH to convert all ammonia to the free base form (NH₃). High-salinity samples were first treated with 1.5 mol l^−l EDTA (tetrasodium salt; 1ml added to each 50 ml sample) to chelate metal ions that would otherwise form hydroxide precipitates on addition of NaOH.

Blood and plasma bicarbonate values were calculated from measurements of plasma and whole-blood pH using a rearrangement of the Henderson-Hasselbalch equation and values for CO₂ solubility and pK’ derived from Boutilier et al. (1984).

The and NH₄⁺ in plasma and fresh water were similarly calculated from their respective pH and T_Amm values using the rearranged Henderson-Hassel-balch equation and values of pK’ and solubility determined by Cameron and Heisler (1983). For the 33% sea water and full sea water media, values for pK’ were calculated from their NaCl concentrations and experimental temperatures (15 and 12°C, respectively) using the nomogram of Cameron and Heisler (1983) (9.605 for 33% sea water and 9.776 for 100% sea water), which both agree well with the values obtained from the data of Whitfield (1974) for salinities of 12 ‰ (at 15°C) and 35% (at 12°C).

It is important to note that the pK’ and solubility values of ammonia both increase as salinity increases and temperature decreases (Thurston et al. 1979). As a result, the fraction of ammonia present as NH₃ was progressively lower at the higher salinities, despite constant water pH and T_Amm (see Table 1).

The concentration of metabolic (or non-respiratory) protons added to or removed from the blood plasma during the interval between any two successive blood samples was calculated according to the formula of McDonald et al. (1980) using non-bicarbonate buffer values estimated from the blood [Hb] and the regression relationship of Wood et al. (1982). The net load of acidic equivalents in blood plasma at any sample time was then calculated by summing the values, signs considered, for each period from the control sample onwards.

Values are expressed as mean±one standard error (N) throughout the text. Time-dependent changes during each exposure regime were tested against individual pre-exposure control values, using a Student’s two-tailed t-test (paired) at 5% and 1% levels of significance. The mean values for 33% seawater and seawater trout, at each sample time, were also compared with the corresponding mean values from freshwater trout using a Student’s unpaired i-test, again at 5 % and 1 % levels of significance.

Control values for [T_Amm] in arterial plasma averaged 43±5 (6) in FW trout, 85±20 (8) in 33%SW trout and 86±12 (5) μmoll⁻¹ in SW trout (Fig. 1). The twofold higher plasma [T_Amm] from the 33 %SW and SW trout (P<0.05) coincided with the higher [T_Amm] in Brixham sea water (10–30μmol I⁻¹ compared with less than 10μmoll⁻¹ in fresh water). The resultant transbranchial ammonia gradients (ΔNH₄⁺ and ⁠) were always positive (from blood to water) during the control periods with no significant differences between the three groups (Fig. 2).

At the start of each exposure, water [T_Amm] was raised to a nominal value of 1000μmoll⁻¹ (995±9 in FW, 974±11 in 33%SW and 989±25μmoll⁻¹ in SW). This caused an immediate reversal of both and ΔNH₄⁺ to large negative values. However, the magnitude of the initial reversal was dependent on the ambient during high-ammonia treatments (Table 1). Accordingly, the initial reversals were 22 and 40% smaller in 33%SW and SW trout, respectively, when compared with FW trout. In contrast the Δ [NH₄⁺] reversals were effectively the same in all three media (Fig. 2).

Reversal of the transbranchial ammonia gradients caused rapid and substantial increases in plasma P_Amm in all three groups within the first hour (Fig. 1). The general pattern of hyperammoniaemia was the same at all three salinities (a rapid initial rise over the first few hours followed by a stabilisation of plasma T_Amm). Although in all cases plasma T_Amm remained well below the external T_Amm, the magnitude of the resultant plasma ammonia load increased substantially with salinity. After 24 h the plasma P_Amm levels in FW, 33%SW and SW trout were 311±15, 418±56 and 537±l7μmolI⁻¹, respectively. Both the 33%SW and SW trout had accumulated significantly more ammonia (34 % and 73 % more) than the FW trout by the end of the exposure (P<0.05).

In all cases and ΔNH₄⁺ remained negative throughout the high-ammonia treatment. However, FW trout maintained a more negative AP_NH3 than SW trout both as an absolute value (–50.5 mPa vs -11.2 mPa at 24 h; Fig. 2) and when compared with their respective reversals at time zero (⁠ values after 24h were 87% and 34% of the original in FW and SW trout, respectively). Although the original (time zero) reversal of ΔNH₄⁺ was virtually the same in all media, after 24 h Δ [NH₄⁺] was significantly smaller in the 33 %SW and SW trout (Fig. 2).

The pre-exposure pH and [HCO₃⁻] values of SW trout were lower than those of the FW and 33 %SW groups, but arterial was essentially the same at all three salinities (Fig. 3). Exposure to ammonia caused no significant blood pH changes in FW trout, and only a slight alkalosis after 1 h in the 33 %SW trout. In contrast, the SW trout experienced a gradual increase in blood pH as the exposure continued which became significant after 24 h (an increase of 0.1 pH units over the control value; Fig. 3). There was a small reduction in arterial in all groups during the exposure, which occurred more rapidly at the two lower salinities (significant after just 1 h). The trend for a reduction in continued for 24 h in FW and SW trout. The only significant changes in plasma [HCO₃⁻] were small decreases observed in the FW trout after 1 and 24 h.

Analysis of the non-respiratory component of the acid-base changes elicited during high-ammonia treatment revealed opposite responses in FW and SW trout (Fig. 4). After 24h, FW trout had accumulated a significant acid load whereas SW trout had developed a significant acid deficit (i.e. a base load; mequiv l⁻¹, N=5;P<0.05). No significant change was observed in the 33% SW group.

Arterial was unaffected by ammonia exposure apart from a slight increase in the 33%SW trout after 5h. There were no significant differences between the three groups, with average values within the range 12.5-14.2 kPa.

Plasma ion concentrations also remained unchanged during ammonia exposure, regardless of salinity. FW and 33%SW trout had plasma ion levels typical of previously published values for freshwater trout (Na⁺ = 148±3, Cl⁻ = 126±2, K⁺=2.6±0.1, Ca²⁺=4.4±0.2, Mg²⁺ = 1.8±0.1 mequivl⁻¹). However, SW trout had significantly higher plasma levels of all the measured ions except magnesium (Na⁺ = 159±4, Cr=152±2, K⁺=3.1±0.3, Ca²⁺=5.8±0.5, Mg²⁺ = 1.9±0.2 mequivl⁻¹).

The changes in plasma ammonia levels are consistent with those of previous reversed-gradient studies using freshwater rainbow trout (Cameron and Heisler, 1983), freshwater channel catfish Ictalurus punctatus (Cameron, 1986) and the marine long-horned sculpin Myoxocephalus octodecimspinosus (Claiborne and Evans, 1988). In all cases, plasma T_Amm increased rapidly, but approached a new steady state within 3–5 h, at a point where both Δ [NH₄⁺] and were either negative or negligible. Previous reversed-gradient studies have employed exposure times of 6 h or less. It is clear from the present investigation that trout can maintain reversed ammonia gradients for much longer periods (we have continued exposures for up to 48 h) and, indeed, some of the relevant acid-base changes only become apparent after at least 5h of exposure, subsequent to the stabilisation of plasma [T_Amm].

It is unlikely that any change in the form of the total excreted waste nitrogen occurred (e.g. a switch from mainly ammonia to urea excretion), since we have found that during similar ammonia exposure regimes changes in urea excretion play no significant role in the restoration of nitrogenous waste excretion in rainbow trout (R. W. Wilson and R. S. Munger, unpublished results). If we also assume that no change in the tissue ammonia production rate occurred, then the maintenance of negative and [NH₄⁺] gradients requires that an active NH₄⁺ extrusion mechanism was operating, which was sufficient to counterbalance the influx of ammonia in addition to excreting the endogenously produced ammonia.

A likely mechanism for the active extrusion of NH₄⁺ is via branchial Na⁺/NH₄⁺ exchange. Many authors have provided evidence that supports the existence of a carrier-mediated Na⁺/NH₄⁺ exchange across the fish gill epithelium (Maetz and Garcia-Romeu, 1964; Maetz, 1973; Payan and Matty, 1975; Evans, 1977; Payan, 1978; Wright and Wood, 1985; Balm et al. 1988; McDonald and Prior, 1989). When transbranchial ammonia gradients were reversed in the freshwater trout, Cameron and Heisler (1983) suggested that Na⁺/NH₄⁺ exchange could be operating to counterbalance the diffusive uptake of NH₃. However, Cameron (1986) more recently found that exchange of internal NH₄⁺ for external H⁺ was the only plausible mechanism that could be fitted to his data from the channel catfish to explain the maintenance of reversed NH₃ and NH₄⁺ gradients (e.g. the changes in titratable acidity, net apparent H⁺ efflux and lack of change in Na⁺ uptake).

The two suggested mechanisms for active excretion of ammonium ions (NH₄⁺/Na⁺ and NH₄⁺/H⁺ exchange) should have opposite acid-base consequences if operating to counteract the inward diffusion of ammonia. If the majority of ammonia enters the fish as NH₃, then an internal alkalosis might be expected, since at the pH of the fish plasma at 12–15°C (approximately 7.9) any inwardly diffusing molecular NH₃ would associate with protons to form NH₄⁺. Any ammonia entering the fish as NH₄⁺ would have no acid-base effect. However, transbranchial ammonia movements will only cause changes in blood pH if the respiratory component of blood acid-base status remains constant. For this reason one must use the non-respiratory component of the blood acid-base status when discussing the acid-base consequences of branchial ammonia movements. The schematic diagram in Fig. 5A shows that operation of NH₄⁺/Na⁺ exchange would complete an H⁺ ‘shuttle’, exporting protons ‘trapped’ in NH₄⁺ out of the blood and would therefore be expected to cause an acid deficit or ‘base load’ (−⁠). In contrast, operation of NH₄⁺/H⁺ exchange (Fig. 5B) would counteract the alkalising effect of passive NH₃ influx, by taking up H⁺ from the ambient water in exchange for each NH₄⁺ exported. In this respect, NH₄⁺/H⁺ exchange would be both electrically and acid-base neutral. However, if any ammonia were to diffuse into the blood as NH₄⁺ then a small acid load would be expected. In our FW trout, blood pH remained constant but a significant non-respiratory acid load had accumulated after 24 h. Although the gills of freshwater fish are considered to be relatively impermeable to NH₄⁺, the large diffusion gradients during the present experimental regime would probably result in some inward diffusion of NH₄⁺ (McDonald and Prior, 1989). If even a small proportion of the ammonia influx in these freshwater fish was the result of NH₄⁺ diffusion, then any subsequent NH₄⁺/H⁺ exchange would cause an acid load resulting from the import of external H⁺ in exchange for the inwardly diffusing NH₄⁺ (see Fig. 5B). It would therefore appear that the NH₄⁺/H⁺ exchange proposed by Cameron (1986) to explain how freshwater channel catfish maintain negative transbranchial ammonia gradients will also explain the acid-base response in FW trout under similar conditions. In contrast, the SW trout developed an acid deficit or ‘base load’ ⁠. Although the inward diffusion of NH₄⁺ probably does contribute to the plasma hyperammoniaemia in SW trout (see below), any subsequent exchange of NH₄⁺ with external Na⁺ would be acid-base neutral. The acid deficit in the SW trout can be explained by the inward diffusion of NH₃ followed by extrusion of the subsequently formed NH₄⁺via NH₄⁺/Na⁺ exchange (see Fig. 5A).

On the basis of the differential acid-base changes observed in FW and SW trout, we suggest that NH₄⁺/Na⁺ exchange predominates in SW-adapted trout but that NH₄⁺/H⁺ is the primary exchange utilised by FW-adapted trout. Given the very high concentration of Na⁺ in sea water (approximately 460mequivl⁻¹), the former seems likely purely from the aspect of counterion availability. In terms of acid-base balance, operation of NH₄⁺/H⁺ exchange under these conditions in fresh water is actually more appropriate than NH₄⁺/Na⁺ exchange if one assumes that the majority of the ammonia influx is via NH₃ diffusion, as removal of NH₄⁺ by this exchange is acid-base neutral and would not involve an increased rate of Na⁺ uptake that might cause a Na⁺ load (plasma [Na⁺] did not change). In seawater teleosts the unidirectional ion fluxes are an order of magnitude greater than in freshwater fish (Evans, 1984). Increasing the unidirectional Na⁺ uptake (through NH₄⁺/Na⁺ exchange) for the purpose of active NH₄⁺ excretion would probably have little effect on net sodium uptake in SW trout. This is presumably why plasma [Na⁺] also remained unchanged in the SW trout exposed to a high ammonia concentration.

The initially rapid increase in plasma [T_Amm] is likely to be at least partly due to inward diffusion of NH₃ down its partial pressure gradient since the calculated gill diffusion coefficient for NH₃ is about the same as that for CO₂ in rainbow trout (Cameron and Heisler, 1983; Avella and Bornancin, 1989). However, the hyperammoniaemia developed more rapidly and was far more pronounced (73 % higher) in the SW trout where the inward P_NH₃ gradient was 40 % smaller at the start (–35.4 mPa in seawater vs. –60.4 mPa in freshwater fish) and 78 % smaller after 24h (–11.2 mPa in seawater vs a freshwater value of –50.5 mPa at 24h). Assuming that the gill permeability to gaseous NH₃ does not change with salinity, and that ammonia production/excretion rates are the same in freshwater and seawater fish (the transbranchial gradients supporting excretion were not significantly different), then there are two possible explanations for this: (i) SW trout may have a greater permeability to NH₄⁺ than their FW-acclimated counterparts, and/or (ii) SW trout have a reduced capacity to excrete NH₄⁺ against a negative gradient. The latter seems extremely unlikely given that SW trout are known to have a fivefold higher capacity than FW trout for the excretion of acidic equivalents (H⁺ and NH₄⁺) following exhaustive exercise (Tang et al. 1989), which is correlated with the availability of Na⁺ as an external counterion (McDonald et al. 1989). In the present study the level of Na⁺ available for active NH₄⁺ excretion is about 660 times greater in sea water. The most plausible explanation for the more rapid and greater accumulation of ammonia in SW trout (which were subjected to a much smaller reversal of ⁠, but a near identical reversal of Δ [NH₄⁺] compared to the FW trout) would appear to be that an increased permeability to NH₄⁺ accompanies the adaptation from fresh water to sea water. This is in keeping with the notion that marine teleosts have higher ionic permeabilities than freshwater fish (Girard and Payan, 1980; Evans, 1979, 1984). Branchial permeability to NH₄⁺ has previously been shown to be important in two marine teleost species, the long-horned sculpin Myoxocephalus octodecimspinosus and the Gulf toadfish Opsanus beta (Goldstein et al. 1982; Claiborne and Evans, 1988; Evans et al. 1989). Evans et al. (1989) reported some preliminary results from experiments using perfused heads of toadfish acclimated to a reduced salinity (5 % sea water) that showed the same correlation between salinity and NH₄⁺ permeability. However, this is the first time that a complete comparison has been reported for animals of the same species acclimated to both fresh water and full strength sea water.

It is of course the electrochemical gradient that provides the driving force behind the passive diffusion of ions across the gill. Analysis of branchial NH₄⁺ diffusion should take into account the gill transepithelial potential (TEP). The TEP of most euryhaline fish in sea water is positive (inside relative to outside) and generally in the range of +10 to +35 mV (Potts, 1984), which will promote the outward diffusion of NH₄⁺ under normal conditions (i.e. a low external ammonia concentration). Although NH₄⁺ diffusion is not considered important in fresh water fish, the greater permeability coupled with a positive TEP in seawater teleosts suggests that NH₄⁺ diffusion may have a substantial role in branchial ammonia transfer under normal conditions. Evans et al. (1989) determined that passive NH₄⁺ diffusion contributed at least 21 % to the overall rate of ammonia excretion in the perfused head of marine Opsanus beta. Future research should seek to clarify and quantify the role of this pathway in seawater fish by simultaneous measurements of gill NH₄⁺ and NH₃ gradients, ammonia fluxes and TEP.

Acute ammonia toxicity occurs when the internal ammonia concentration reaches a critical level causing impairment of cerebral energy metabolism and nerve function (see Smart, 1978; Randall and Wright, 1987). This is obviously dependent on the rate of influx of ammonia during ammonia exposure which, in turn, will be determined by the permeability to NH₃ and NH₄⁺. Ammonia toxicity to fish is usually described in terms of the ambient un-ionised ammonia concentration, since this is the variable best correlated with toxicity (Alabaster and Lloyd, 1980; Thurston and Russo, 1983; Thurston et al. 1984). This is not surprising considering that the above criterion is based almost exclusively upon data obtained from freshwater fish, which are considered to be relatively impermeable to NH₄⁺. Since seawater-adapted teleosts appear to have enhanced NH₄⁺ permeability, the use of un-ionised ammonia concentrations to describe toxicity may prove misleading when considering tests on marine species.

This research was conducted under the tenure of a Science and Engineering Research Council CASE studentship (R.W.W.) in cooperation with ICI Environmental Research Laboratories, Brixham, Devon. The authors wish to thank R. S. Thompson and D. Taylor of ICI Environmental Research Laboratories for their invaluable assistance in this project.