Responses of an Amazonian Teleost, the Tambaqui (Colossoma macropomum), to Low pH in Extremely Soft Water

Our goal was to compare the internal physiological responses to acid challenge in an acidophilic tropical teleost endemic to dilute low-pH waters with those in nonacidophilic temperate species such as salmonids, which have been the subject of most previous investigations. The Amazonian tambaqui (Colossoma macropomum), which migrates between circumneutral water and dilute acidic “blackwater” of the Rio Negro, was exposed to a graded low-pH and recovery regime in representative soft water (Na+ = 15, Cl− = 16, Ca2+ = 20 μmol L−1). Fish were fitted with arterial catheters for repetitive blood sampling. Water pH was altered from 6.5 (control) to 5.0, 4.0, 3.0, and back to 6.5 (recovery) on successive days. Some deaths occurred at pH 3.0. Throughout the regime, there were no disturbances of blood gases (O2 and CO2 tensions and contents) or lactate levels, and only very minor changes in acid-base status of plasma and red cells. However, erythrocytic guanylate and ade-nylate levels increased at pH's less than or equal to 5.0. Down to pH 4.0, plasma glucose, cortisol, and total ammonia levels remained constant, but all increased at pH 3.0, denoting a stress response. Plasma Na+ and Cl− levels declined and plasma protein concentration increased at pH 3.0, indicative of ionore-gulatory and fluid volume disturbance, and neither recovered upon return to pH 6.5. Cortisol and ammonia elevations also persisted. Transepithelial potential changed progressively from highly negative values (inside) at pH 6.5 to highly positive values at pH 3.0; these alterations were fully reversible. Experimental elevations in water calcium levels drove the transepithe-lial potential positive at circumneutral pH, attenuated or prevented changes in transepithelial potential at low pH, and reduced Na+ and Cl− loss rates to the water during acute low-pH challenges. In general, tambaqui exhibited responses to low pH that were qualitatively similar but quantitatively more resistant than those previously documented in salmonids.

1. Dissolved organic carbon concentration was 2.05 mg L 01 , g for 2 min, and the red blood cell (RBC) pellet was immediately frozen in liquid N 2 for measurement of intracellular pH pH was about 6.5, and temperature was 28Њ -30ЊC. This very dilute water, close to natural blackwater apart from its lower (RBC pHi). Extracellular pH (pHe) that was usually the pH of arterial blood (pHa), RBC pHi, the partial pressure of O 2 organic content, was used in all experiments.
In preparation for all experiments except series 3 (see be-in arterial blood (PaO 2 ), the total concentration of O 2 in arterial blood (CaO 2 ), the total concentration of CO 2 in arterial plasma low), the fish were anaesthetized with MS-222 (0.5 g L 01 , neutralized to pH Å 6.0 with KOH), placed on an operating table, (CaCO 2 ), hematocrit (hct), and hemoglobin concentration (Hb) were measured immediately. Plasma Na / , K / , Ca 2/ , Cl 0 , and fitted with an indwelling catheter for repetitive blood sampling and/or transepithelial potential measurements. The very protein, total ammonia, glucose, and cortisol were measured after storage at 070ЊC. Blood samples (100 mL) for determina-small size of the mouth precluded the traditional dorsal aortic catheterization technique, so instead the caudal artery (or in tion of lactate and RBC nucleoside phosphates (ATP, ADP, AMP, GTP, GDP, and GMP) were immediately deproteinized a few cases the caudal vein) was cannulated in the tail region. This involved surgically separating the hypaxial and epaxial in 200 mL ice-cold 8% perchloric acid for several hours. The slurry was spun briefly, and then the supernatant was decanted, muscle masses so as to expose the hemal arch, and then inserting the catheter under a vertebral spine and advancing it neutralized with 6 mol L 01 KOH, and then stored at 070ЊC for later analysis. several centimeters into the vessel. PE50 tubing (Clay-Adams, Parsippany, N.J.) was used, filled with Cortland saline (Wolf 1963) heparinized at 50 i.u. mL 01 with lithium heparin (Sigma, Series 2. This series used fish of intermediate size (190 -440 g; N Å 18) to examine the acute responses of transepithelial St. Louis). The wound was dusted with oxytetracycline (Sigma) and tightly closed with silk sutures. potential to alterations in water [Ca 2/ ] and pH. Cannulated tambaqui were placed overnight in small containers of known The fish were allowed to recover for 36 h in their experimental chambers, separate light-shielded polyethylene containers volume (generally 3.5 L) with individual flow and aeration. In one experiment, the flow was stopped and the water changed (volume of 8 L in series 1, 3.5 L in series 2) served with individual aeration and a flow of 1.5 L min 01 of recirculated to groundwater that had purposely not been supplemented with Ca(NO 3 ) 2 in order to start the series at a very low [Ca 2/ ] water. The fish chambers were housed on a wet table that drained back into the recirculation reservoir. Total volume of (9 mmol L 01 ). After a 10-min settling period, transepithelial potential was measured, and then known volumes of a concen-the system, serving a maximum of 12 fish, was 760 L; 90% of the water was replaced twice a day. trated stock solution of Ca(NO 3 ) 2 were added at 15-min intervals so as to sequentially raise the water [Ca 2/ ] from approximately 10 to approximately 10,000 mmol L 01 in half-log-unit Experimental Series steps. The pH was maintained at about 6.5. At each step, transepithelial potential was measured via the catheter after 10 Series 1. This series used the largest fish available (490 -1,640 g; N Å 12). Repetitive blood sampling was used to monitor min, and a water sample was taken at this time in order to determine the exact [Ca 2/ ]. In a second experiment with a the internal responses of cannulated tambaqui to a graded acid exposure; H 2 SO 4 was used to lower water pH. On each sam-similar protocol, water Ca 2/ was set to approximately 20, 200, or 10,000 mmol L 01 by Ca(NO 3 ) 2 addition at pH 6.5, and then pling day at about 1000 hours, after an overnight period at a given pH, the water in the reservoir was replaced with new pH was sequentially lowered to pH 3.0 by addition of H 2 SO 4 at 15-min intervals. water of the same pH, without interruption of flow to the fish chambers. Blood and water samples were taken and transepithelial potentials were measured between 1400 and 2000 hours Series 3. This series used small tambaqui (99 -288 g; N Å 24) to examine the effect of acute exposure to pH 3.5 on net whole-on successive days at pH's of 6.5 (control), 5.0, 4.0, 3.0, and 6.5 again (recovery). On each day, the water was changed again body Na / and Cl 0 flux rates at different water [Ca 2/ ] values.
Noncannulated fish were allowed to settle for several hours in at about 2000 hours and gradually adjusted to the new pH over the next 4 h. Thus fish had experienced each pH for about one of six 3.5-L chambers served with individual aeration and ecirculation flow from a 200-L reservoir of water with a [Ca 2/ ] 17 h at the time of each sampling.
At each sample time, a water sample was drawn from in of 20 mmol L 01 , pH Å 6.5. The flow was then stopped for 60 min, during which the individual chambers were operated as front of the fish's mouth for measurement of water O 2 tension (PwO 2 ). Transepithelial potential was measured via the cathe-closed systems to make control flux measurements. Water samples (20 mL) were taken at the beginning and end of the hour ter, and then blood samples (1,000 mL) were drawn into two gastight, ice-cold 500-mL Hamilton syringes and immediately for analysis of [Na / ] and [Cl 0 ]. In the reservoir, water pH was then lowered to pH 3.5 with H 2 SO 4 , and water [Ca 2/ ] was apportioned for analyses. Approximately 200 mL of blood (recovered from electrodes) plus 800 mL of Cortland saline were either kept at 20 mmol L 01 or raised to 100 or 700 mmol L 01 with Ca(NO 3 ) 2 in different experiments. The chambers were reinfused into the fish after sampling to restore blood volume. Plasma was separated by centrifugation of an aliquot at 10,000 flushed for 10 min with this low-pH water from the reservoir  Cameron (1971); and RBC pHi was measured Flux rates of Na / and Cl 0 were calculated from measured by the freeze-thaw method of Zeidler and Kim (1977). Water changes in the concentration of the ion in the closed external pH was monitored with a GK2401C glass combination elecwater over the flux period, factored by the mass of the fish, trode; whole-blood pHe and RBC pHi were determined with chamber volume, and time. an E5021a ''gun'' micro-electrode system; and PwO 2 and PaO 2 Most data have been expressed as means { 1 SEM (N), were determined with an E5036 electrode. The Tucker and where N is the number of fish. As each fish was used as its Cameron chambers were fitted with E5036 and E5046 elecown control, all data points were compared to the original trodes, respectively. All electrodes were thermostatted to the control values using Student's two-tailed paired t-test at P experimental temperature, except those in the Cameron and°0 .05, with the Bonferroni correction for multiple compari-Tucker chambers, which were operated at 40ЊC.
sons (Nemenyi et al. 1977). Note that in series 1, of the 12 Hct was determined by centrifugation at 5,000 g for 5 min, tambaqui present at the start of the experiment, catheters failed and Hb was measured by the cyanmethemoglobin method in two animals, and another two died before measurement at (Sigma reagents). Plasma and water [Na / ], [Ca 2/ ], and [K / ] the pH 3.0 step of the regime. To avoid bias from these missing were determined with a CELM model FC108 flame photometer data, means given in figures and tables, and their statistical (CELM, Rio de Janeiro), and [Cl 0 ] was measured by the coloriassessment, were tabulated for only those fish (N Å 8) that metric assay of Zall et al. (1956). Plasma total ammonia, glusupplied complete data sets down to pH 3.0. Three additional cose, and whole-blood lactate concentrations were determined fish died before the recovery measurement, so the recovery enzymatically using Sigma kits (nos. 171, 16, and 826, respecmeans (and their statistical assessment relative to control) are tively). Cortisol was measured by an [ 125 I] radioimmunoassay based on N Å 5 and are joined to the rest of the data by a (ICN Immunocorp, Montreal) using standards diluted to the dashed line in the figures to indicate this change in sample protein concentrations found in tambaqui plasma and analyzed size. Two of the eight fish providing data through pH 3.0 and on a Canberra-Packard Minaxi 5000 gamma counter (Downers one of the five providing data on the recovery day had caudal Grove, Ill.).
vein rather than caudal artery catheters. Use of the venous RBC adenylates (ATP, ADP, and AMP) and guanylates sampling site had no appreciable effect on most parameters, (GTP, GDP, and GMP) were measured by high-pressure liquid but it clearly provided different blood gas and pH values. Thus, chromatography (HPLC) using an LKB 2152 HPLC controller only arterial data were used for the blood gas and pH means and 2150 titanium pump coupled to a 2220 recording integand statistics. Therefore, for these assessments only, N Å 6 rator (LKB, Turku, Finland). The separation was performed through pH 3.0, and N Å 4 at recovery. on an Aquapore (PMI Products, Ithaca, N.Y.) AX-300 7-mm weak anion exchanger eluting at 2 mL min 01 according to the methods of Val et al. (1994).
Results Transepithelial potential was determined by means of 3-mol L 01 KCl-agar bridges connected via Ag/AgCl electrodes to a Series 1: Internal Responses to a Graded Low-pH Regime high-impedance voltmeter (Perry and Wood 1985). The reference electrode was placed in the water in the fish chamber, Exposure to pH's as low as 3.0 had negligible influence on and the measurement electrode was connected to the blood blood oxygenation and transport in Colossoma macropomum. via the catheter. The system was calibrated to zero potential At an inspired PwO 2 of about 110 Torr (where 1 Torr Å 133.322 Pa), tambaqui exhibited a relatively low PaO 2 of approximately by placing both electrode KCl-agar tips in the water.
9g16$$no01 09-25-98 13:03:36 pzal UC: PHYS ZOO Both hct and mean cell Hb concentration, an index of red blood cell size, remained unchanged at control levels (20.9% { 2.1% and 0.3312 { 0.0106 g Hb mL 01 , respectively, N Å 8) throughout the exposure (data not shown). However, there were pronounced changes in RBC nucleoside phosphate levels. Guanylates clearly dominated over adenylates by about fivefold at all times, and this difference was consistent for tri-, di-, and monophosphates (Fig. 3). ATP and AMP each made up about 40% of the RBC adenylate pool, while ADP constituted the remaining 20%. GTP made up 50% -60% of the guanylate pool, with the remainder being made up of approximately equal contributions from GDP and GMP. The total erythrocytic adenylate pool (per unit Hb) increased progressively throughout the regime, almost doubling by the end. The elevation was significant at pH 4.0, pH 3.0, and at recovery. This reflected increases in both ATP (significant at all points throughout the regime) and AMP (significant at pH Å 3.0 and at recovery). The total erythrocytic guanylate pool also showed a tendency to increase at low pH, although this was not significant. However, the GTP component was significantly elevated by about 40% at pH 5.0, 4.0, and recovery.
Several stress indicators demonstrated negligible effects down to pH 4.0, but a substantial disturbance at pH 3.0. Total plasma ammonia approximately doubled at pH 3.0 from control values of about 140 mmol L 01 and continued to rise during remained very low (õ1 mmol L 01 ) and did not change throughout the regime (Fig. 4B), confirming that the tambaqui 35 Torr, which did not change during acid exposure and recovencountered no O 2 delivery problem down to pH 3.0. ery (Fig. 1A) 1B). This constancy was reflected in stable levels and 4 mmol L 01 , respectively) were all relatively high under of CaO 2 throughout the regime (Fig. 1B). Arterial O 2 saturation control conditions and remained unchanged at pH 5.0 and (70.7% { 2.2%, N Å 8 under control conditions) also re-4.0. The stress response at pH 3.0 coincided with the onset of mained constant (data not shown). Venous PO 2 was 1 -6 Torr, osmoregulatory dysfunction (Fig. 5). Plasma Na / and Cl 0 both providing a venous O 2 saturation of less than 30% (two fish fell significantly by about 20% relative to control values, with only; data not shown).
no evidence of restoration at the recovery measurement (Fig. Blood acid-base status was also barely affected by these low-5A). Plasma protein concentration, an inverse index of changes pH exposures, despite the fact that at pH 3.0, external [H / ] in blood volume (McDonald et al. 1980), increased by about was nearly five orders of magnitude greater than that in the 20% at pH 3.0, again with no evidence of subsequent recovery blood. The pHa was about 7.80 initially and fell marginally at (Fig. 5B). There were no significant changes in plasma [Ca 2/ ] pH 5.0 and 4.0, but was not significantly different from the or [K / ] levels throughout the exposure regime (data not control value at pH 3.0 or at recovery ( Fig. 2A). RBC pHi, shown). about 7.3 initially, did not change significantly down to an Transepithelial potential between the body fluids and the external pH of 4.0, but fell by about 0.1 units at pH 3.0 and external water (with the latter taken as the zero reference point) did not recover when water pH was raised ( Fig. 2A). PaCO 2 changed dramatically in response to low pH (Fig. 6). Transepiwas relatively high, about 5 Torr under control conditions (Fig. thelial potential rose from a highly negative value (023 mV) 2B), where PwCO 2 , as estimated from water pH and titratable at control pH 6.5 to a highly positive value (/35 mV) at pH 3.0, alkalinity measurements, was about 2 Torr. PaCO 2 remained with a complete reversal back to the control level at recovery. stable throughout the regime. CaCO 2 also remained stable (data Crossover from negative to positive potential occurred between not shown), so arterial plasma [HCO 0 3 ] (7 -9 mmol L 01 ) did not change significantly throughout the exposure (Fig. 2C). However, at low pH, net ion loss rates were very sensitive to water [Ca 2/ ] (Fig. 9). Acute reductions of water pH to 3.5 for Series 2: Responses of Transepithelial Potential to Acute 1 h caused substantial net losses of both ions, with flux rates Changes in Environmental Ca 2/ and pH of about 01,500 and 01,000 mmol kg 01 h 01 for Na / and Cl 0 , At control pH (about 6.5), transepithelial potential was sensirespectively, at a [Ca 2/ ] of approximately 20 mmol L 01 . These tive to water [Ca 2/ ], changing from about 030 mV at 10 loss rates were reduced by about 50% when the experiment mmol L 01 to about /10 mV at 10,000 mmol L 01 (Fig. 7). The was performed at a [Ca 2/ ] of about 100 mmol L 01 , and by relationship between transepithelial potential and log [Ca 2/ ] about 67% at a [Ca 2/ ] of about 700 mmol L 01 (for Na / only). was hyperbolic, such that [Ca 2/ ] changes in the low, natural When LaCl 3 rather than low pH was used as the challenge at range had a much greater influence than at the upper range, a concentration (20 mmol L 01 ) equimolar to water [Ca 2/ ], where an asymptote was approached.
there was no effect on ion loss rates (data not shown). At the control water [Ca 2/ ] (about 20 mmol L 01 ), the large effects of acute reductions in water pH on transepithelial potential (Fig. 8) were very similar to those seen during the graded Discussion pH regime of series 1 (Fig. 6). The relationship was attenuated at [Ca 2/ ] of approximately 200 mmol L 01 , but the potential Basic Physiology of the Tambaqui still increased substantially from a slightly negative value (04 mV) at pH 6.5 to /28 mV at pH 3.0. However, at very high Colossoma macropomum is now a species of immense importance for commercial fishing and aquaculture in the Amazon external [Ca 2/ ] (about 10,000 mmol L 01 ), transepithelial potential became insensitive to water pH, remaining constant at basin (Goulding and Carvalho 1982;Roubach and Saint-Paul 1994;Val and Honczaryk 1995), but until recently, very little about /12 mV from pH 6.5 down to 3.0 (Fig. 8).
9g16$$no01 09-25-98 13:03:36 pzal UC: PHYS ZOO Figure 3. The influence in tambaqui of exposure to a graded low-the estimated PwCO 2 of 2 Torr (Fig. 2B). We hypothesize that pH and recovery regime (1-d steps) on the concentrations of nuthe presence of these high partial pressure gradients between cleoside phosphates in the RBCs. All concentrations are expressed blood and water indicates a low gill diffusing capacity, despite in micromoles of nucleoside phosphate per micromole of Hb to the fact that total gill surface area is unusually large (Saintcorrect for possible changes in RBC volume. The bar represents Paul 1984). Low branchial diffusing capacity would thereby the sum of the three components with each fraction indicated (mean / SEM; N Å 8 at control, pH 5.0, 4.0, and 3.0; N Å 5 at minimize ionic losses and ionoregulatory work load in dilute, recovery). Asterisks indicate means significantly different (P acidic environments. Indeed, plasma ion levels under control°0 .05) from the respective control mean for each fraction; dagconditions were high for freshwater teleosts (Fig. 5A). Most gers indicate adenylate or guanylate totals significantly different other blood parameters appeared to be within the normal range (P°0.05) from the respective control total. (McDonald and Milligan 1992).
has been known about its basic ionoregulatory and respiratory physiology. Most interest has centred on its exceptional toler-The Graded Low-pH Regime ance to environmental hypoxia, with perfect regulation of oxygen consumption down to environmental O 2 levels of about Most previous studies on the responses of fish to environmental acidity have involved large stepwise reductions in pH (e.g., 7.0 30% air saturation (Saint-Paul 1984). Below this point, the tambaqui exhibits a remarkable ability to ''grow'' a large ''lip'' to 4.0). Arguably, such a change could represent a sudden rainstorm or snowmelt event associated with acidic precipita-within several hours of exposure to environmental hypoxia (Braum and Junk 1982). The tambaqui does not breathe air, tion, but it would have little relevance to the environmental situation in the Amazon basin. A few studies have employed but this lip mechanically facilitates ''skimming'' of the more O 2 -rich surface waters (''aquatic surface respiration''; Rantin a more gradual acidification regime and have concluded that in general it is better tolerated (Stuart and Morris 1985;Wen-and Kalinin 1996), and its appearance is accompanied by simultaneous biochemical adjustments in erythrocytes and white delaar Bonga et al. 1987;Van Dijk et al. 1993). Therefore, in the present study, we employed a graded regime in which the muscle that aid hypoxia tolerance (reviewed by Val and Almeida-Val [1995]). Recently, a pronounced Root effect (Val fish were exposed to progressively lower pH's on successive days. This might represent the situation of a tambaqui migrat-and Almeida-Val 1995) and a capacity for adrenergic pHi regulation in the RBCs ) have been reported. The ing voluntarily from circumneutral water to lower-pH blackwater and then to an extremely acidic forest stream for feeding present study augments this picture by demonstrating very low resting levels of PaO 2 , approximately 75 Torr below PwO 2 (Fig. (see, e.g., Goulding 1980;Goulding and Carvalho 1982). The recovery measurements at the end of the regime were per-1A). PaCO 2 was correspondingly elevated in accord with the difference in solubility coefficients, being about 3 Torr above formed to check which physiological responses were readily 9g16$$no01 09-25-98 13:03:36 pzal UC: PHYS ZOO reversible and which were not, with the latter being indicative of damage to the physiological mechanism(s) involved.

Internal Responses of Tambaqui to Low pH
The present study is the first to report the internal responses to low environmental pH of a member of the order Characiformes or, indeed, of any Amazonian fish that naturally inhabits blackwater. In general, the picture that emerges is qualitatively similar to but quantitatively different from that of earlier studies on nonacidophilic species of the Northern Hemisphere, such as salmonids, when these fish were exposed to low pH in soft water (reviewed by Fromm [1980]; Wood and McDonald [1982]; McDonald [1983]; Potts and McWilliams [1989]; Wood [1989]; Reid [1995]). In such species, stress responses and ionoregulatory disturbance become evident at a pH thresh- Difference approach (Stewart 1978(Stewart , 1983. From this view-Each symbol represents a different fish; N Å 7. point, the marked blood acidosis commonly seen in trout at high water [Ca 2/ ] was due to a large measured influx of acidic ing plasma Na / and Cl 0 and increasing plasma protein levels equivalents, constrained by an excess of net Na / over net Cl 0 (Fig. 5).
loss. At low water [Ca 2/ ], although both Na / and Cl 0 loss These effects correlate well with data from a parallel series rates increased, Cl 0 loss became equal to or slightly greater of flux experiments performed on a separate batch of tambathan Na / loss, thereby preventing net acid uptake or promoting qui with a similar graded low-pH exposure regime (R. W. slight net acid excretion. However, in the tambaqui and other Wilson, C. M. Wood, R. G. Gonzalez, M. L. Patrick, H. L. Amazonian blackwater species, net flux measurements showed Bergman, A. Narahara, and A. L. Val, unpublished results).
that net Na / loss consistently exceeded net Cl 0 loss at low pH These experiments show that initial net Na / and Cl 0 losses in very soft water, both in the present investigation ( Fig. 9) to the water at pH 4.0 were corrected during continued expoand in the study by Gonzalez et al. (1998). This suggests that sure, whereas larger and persistent losses occurred at pH 3.5.
other (unmeasured) cations are entering or anions are leaving Interestingly, while there was no mortality and full recovery the fish. Clearly, it would be informative in future studies to of net ion uptake upon return to control pH after 1 d at pH measure net acidic equivalent and ion flux rates, together with 3.5, there was some mortality and no restoration of plasma blood acid-base status, during exposure of tambaqui to low Na / and Cl 0 levels in the present study after 1 d at pH 3.0 pH in both high-and low-[Ca 2/ ] water. (Fig. 5). This suggests that the threshold for lethal damage to A notable finding was the complete absence of respiratory gill transport mechanisms and/or permeability in tambaqui disturbance at low pH, whereas nonacidophilic species generlies between 3.0 and 3.5.
ally suffer a suffocation response because of gill structural dam-An important point of agreement with earlier studies on age, edema, and mucification at pH's below 4.0 (reviewed by species such as the rainbow trout (see McDonald et al. 1980;Fromm [1980]; Ultsch et al. [1981]; Wood and McDonald Wood 1989) is the almost complete absence of blood acid- [1982]). Even though some measurements were taken in tambase disturbance, despite a 5-pH-unit difference between water baqui close to death at pH 3.0, there was no evidence of any and blood pH (Fig. 2). In the trout, this stability of acid-base disturbance of arterial blood PO 2 or PCO 2 levels (Figs. 1, 2), status has been related to the fact that when exposures are lactate elevation (Fig. 4B), or visible signs of hyperventilation. performed in soft water very low in [Ca 2/ ] (õ200 mmol L 01 , Presumably, the gill surface of the tambaqui is structurally as in the present study), there is negligible net uptake of acidic resistant and/or does not show an inflammatory response at equivalents from the external environment (or even slight net severely low pH. base uptake, equivalent to acid excretion at the gills).
Stress responses (elevated plasma cortisol, glucose, and am-In trout, Wood (1989) related the acid-base responses occurring at low environmental pH to the difference between net monia levels; Fig. 4 the gills yet favour O 2 unloading at the tissues. The role of this strategy during low pH exposure is unclear, but it again confirms that these fish had no difficulty with branchial O 2 loading. We cannot eliminate the possibility that these were responses to blood sampling rather than to low pH, because similar responses have been demonstrated after repetitive blood removal, resulting in anemia in other species; the immature erythrocytes that are recruited have higher activities of oxidative phosphorylation (reviewed by Val et al. [1994]). However, it is important to note that blood sampling did not cause anemia in these acid-exposed tambaqui, probably because of simultaneous hemoconcentration (see, e.g., Figs. 1C, 5B) and/ or mobilization of erythrocytes stored in the spleen (Milligan and Wood 1982).

Transepithelial Potential Responses of Tambaqui to Low pH
There was a marked and persistent reversal of transepithelial potential in tambaqui from highly negative inside at pH 6.5 to highly positive inside at pH 4.0, in very soft water (Figs. 6, 7). These responses were very similar, both qualitatively and quantitatively, to the detailed observations of McWilliams and Potts (1978) on brown trout in Ca 2/ -free water. The pattern also parallels less detailed results on rainbow trout at low pH Most workers agree (though not all; see below) that at cirmmol L 01 ; N Å 3 at Ç200 mmol L 01 ; and N Å 5 at Ç10,000 mmol cumneutral pH, the inside negative transepithelial potential L 01 Ca 2/ ).
represents a diffusion potential due to the differential permethreshold, were fairly typical of those seen in other teleosts during acid exposure. The well-known effect of cortisol in promoting proteolysis (Anderson et al. 1991;van der Boon et al. 1991) was undoubtedly an important contributor to the persistent elevation in plasma ammonia concentration. The role of cortisol in glucose mobilization remains controversial (see, e.g., Anderson et al. 1991). It is quite possible that catecholamine mobilization may also have occurred, as it does in rainbow trout at pH 4.0 (Ye et al. 1991), and made an important contribution to glucose elevation, either alone or in combination with cortisol (Reid et al. 1992;Perry and Reid 1993). The general levels of nucleoside phosphates in the erythrocytes of tambaqui were comparable to those recorded previously in this species, with guanylates clearly dominating over adenylates (Val and Almeida-Val 1995;. However, to our knowledge, the present data (Fig. 3) are the first Figure 9. The influence in tambaqui of three different water to record the effects of low pH exposure on RBC nucleoside [Ca 2/ ]'s on the net flux rates of sodium (J Na/ net) and chloride (J Cl0 net) to the external water during a 1-h acute exposure to pH phosphate levels in any teleost. They reveal a marked elevation 3.5 (N Å 6 for all treatments). The control flux data shown at pH of both guanylates and adenylates, with changes in GTP and ability of the gills to Na / versus Cl 0 (Na / permeability ú Cl 0 fying their permselectivity to Na / versus Cl 0 in a similar fashion while having very different effects on absolute permeability; see, e.g., Potts [1984]). McWilliams and Potts (1978) and Potts and McWilliams (1989) argued that the trans-permeabilities (i.e., H / increasing and Ca 2/ decreasing absolute permeabilities). Simultaneous electrical and radioisotopic (uni-epithelial potential became positive at low pH because of a very high permeability of the gills to H / ions and, therefore, directional) flux measurements, as opposed to the ''cold'' net flux measurements of the present study ( Fig. 9) will be required a large net entry of H / ions from the acidic environment. However, direct H / flux measurements (discussed earlier) have to test these ideas.
Nevertheless, the present net flux measurements do confirm demonstrated that this is not the case, at least in salmonids (Wood 1989). McWilliams and Potts (1978) based their calcu-a marked protective effect of elevated water [Ca 2/ ] against the net losses of Na / and Cl 0 that occur at low pH (Fig. 9). Again, lations on the assumption that Cl 0 permeability does not change at low pH, whereas unidirectional flux measurements the action of [Ca 2/ ] appears similar to that seen in salmonids (McDonald 1983;Freda and McDonald 1988;Wood 1989) have shown that Cl 0 permeability in fact increases greatly in salmonids (Wood 1989). A more reasonable explanation would and differs from the insensitivity to [Ca 2/ ] seen in both the acidophilic perch (Freda and McDonald 1988) and three black-be that Cl 0 permeability increases to a greater extent than Na / permeability at low pH, causing reversal of the potential -that water species collected from the Rio Negro (Gonzalez et al. 1998). Direct comparison tests showed that all three of the is, a simple pH-dependent modification (Na / permeability õ Cl 0 permeability) of the diffusion potential.
latter Rio Negro species lost far less Na / , K / , and Cl 0 when acutely challenged with low pH than did the tambaqui (Gonza-In contrast, Kirschner (1994) described an analogous situation in freshwater crayfish, presented unidirectional flux evi-lez et al. 1998). Losses by the tambaqui at pH 3.5 were comparable to those of the perch (Freda and McDonald 1988; Mc-dence against the diffusion potential hypothesis, and argued that an electrogenic mechanism responsible for inward Ca 2/ Donald et al. 1991).
The responses of the tambaqui to [Ca 2/ ] at low pH also transport might provide the origin of the transepithelial potential. Alternatively, Randall et al. (1996) have suggested that the differed qualitatively from those of other acidophilic species such as the banded sunfish. The sunfish shows an extremely negative potential inside is an electrogenic potential, and that its reversal at low pH is due to inhibition of the driving force, high affinity of the gill for [Ca 2/ ], with protective effects against low pH complete by 100 -125 mmol L 01 (Gonzalez and Dunson an H / -extruding pump on the apical surface of the gill cells. To date, available evidence suggests that an H / pump may 1987, 1989a). In the present study, protective effects were seen at [Ca 2/ ]'s up to 700 mmol L 01 (Fig. 9), comparable to the exist at the gill surface and contribute a potential across the apical cell membrane, but it is unclear whether it could create range seen in acid-sensitive species such as trout and shiners (Freda and McDonald 1988) and the moderately acid-tolerant a potential across the whole epithelium (Potts 1994;Lin and Randall 1995).
Amazonian blackskirt tetra (Gonzalez et al. 1997). Unfortunately, the picture is further clouded by the observation that lanthanum, thought to be a powerful antagonist that displaces The Influence of [Ca 2/ ] on the Responses Ca 2/ from tight junctions (Weiss 1974), had no effect on Na / of Tambaqui to Low pH or Cl 0 losses in the tambaqui, whereas it augments loss rates in trout, shiner (Freda and McDonald 1988), and blackskirt The present results provide abundant evidence that the gills of tambaqui are extremely sensitive to water [Ca 2/ ]. Log scale tetra (Gonzalez et al. 1997). These results underscore both the diversity of strategies for living at low environmental pH in increases in [Ca 2/ ] drove the transepithelial potential to positive values at circumneutral pH (Figs. 7, 8), a phenomenon various species and the incomplete nature of our current understanding of gill [Ca 2/ ], [H / ], and ion flux relationships. that has now been documented in many other freshwater fish and crustaceans (reviewed by Kirschner [1994]). Depending Nevertheless, it is clear that many of the basic principles with respect to responses to low pH first elucidated in salmonids on one's view of the origin of the transepithelial potential (see above), this could be explained as an effect of [Ca 2/ ] on one apply in tambaqui, and that most differences are quantitative rather than qualitative. or more of three properties: on the Na / permeability to Cl 0 permeability ratio, on Ca 2/ transport, or on the H / -pump. Elevated [Ca 2/ ] also attenuated or prevented the marked Acknowledgments changes in transepithelial potential that occur at low pH (Fig.  8), results very similar to those reported by McWilliams and This work was supported by a Natural Sciences and Engineering Research Council of Canada research grant to C.M.W., Potts (1978) on brown trout. At present, we favour the simplest explanation, that low external pH (high [H / ]) and high exter-a Royal Society research grant to R.W.W., a grant from the University of Wyoming to H.L.B., a Brazilian Research Council nal [Ca 2/ ] both act to shift the Na / permeability to Cl 0 permeability ratio below unity, thereby reversing and stabilizing the (CNPq/Brasil) research grant to A.L.V., and a bequest from the Two Minutes for Looking So Good Foundation. A.L.V. diffusion potential. For example, both agents could act by titrating negative charge on paracellular channels, thereby modi-was the recipient of a research fellowship from CNPq/Brasil.