Cardiorespiratory responses to hypercarbia in tambaqui Colossoma macropomum: chemoreceptor orientation and specificity

Renewed interest in CO₂/H⁺ chemoreception in fish over the last few years has resulted in considerable advances. It is now clear that exposure to environmental hypercarbia (elevated water P_CO2) initiates a host of cardiorespiratory adjustments in fish that do not simply reflect impairment of blood O₂ transport,as earlier work suggested (Randall,1982; Smith and Jones,1982). Rather, changes in CO₂ and/or pH can elicit cardiorespiratory responses directly, through interaction with specific CO₂/H⁺ chemoreceptors. Interspecific variation in the particulars of such cardiorespiratory responses is high, and cardiovascular responses to hypercarbia, in particular, have been assessed in relatively few species. However, most fish examined to date exhibit a striking increase in ventilation (e.g. Dejours,1973; Janssen and Randall,1975; Smith and Jones,1982; reviewed by Gilmour,2001), together with a fall in heart rate(Perry et al., 1999; Reid et al., 2000; Sundin et al., 2000; McKendry et al., 2001; Perry and McKendry, 2001). Depending on the species, the hypercarbia-induced hyperventilation and bradycardia may also be accompanied by changes in blood pressure, cardiac output and systemic resistance (Perry et al., 1999; Reid et al.,2000; Sundin et al.,2000; McKendry et al.,2001; Perry and McKendry,2001).

These cardiorespiratory responses to hypercarbia appear to be triggered by peripheral chemoreceptors that are located primarily, although probably not exclusively in at least some species (Reid et al., 2000; Milsom et al.,2002), on the gills. Bilateral denervation of the gills was sufficient to abolish most or all of the hypercarbia-induced changes in ventilation and/or cardiovascular variables in spiny dogfish Squalus acanthias (McKendry et al.,2001), channel catfish Ictalurus punctatus(Burleson and Smatresk, 2000),traira Hoplias malabaricus (Reid et al., 2000) and tambaqui Colossoma macropomum(Sundin et al., 2000; Milsom et al., 2002; Florindo et al., 2004). Similarly, bilateral extirpation of the first gill arch prevented or greatly attenuated ventilatory and cardiovascular responses to hypercarbia in rainbow trout Oncorhynchus mykiss, pointing to the first gill arch as the chief location of chemoreceptors involved in initiating cardiorespiratory responses to hypercarbia in this species(Perry and Reid, 2002).

The branchial chemoreceptors in rainbow trout, Atlantic salmon Salmo salar and dogfish appear to respond primarily to changes in water CO₂ tension specifically; neither alteration of water pH nor manipulation of blood P_CO2 were effective in triggering cardiorespiratory responses (Perry and McKendry, 2001; Perry and Reid, 2002). Although denervation studies have identified the gills as the principal location of the chemoreceptors that initiate cardiorespiratory responses to hypercarbia in tambaqui(Sundin et al., 2000; Milsom et al., 2002; Florindo et al., 2004), the orientation of these receptors (whether they preferentially detect water or blood) as well as their sensitivity to CO₂vsH⁺, remain uncertain. The tambaqui is a hypoxia-tolerant(P₅₀=2.4 mmHg; Brauner et al., 2001) and hypercarbia-tolerant neotropical fish species that is found throughout the Amazon basin, often in floodplain lakes that are subject to large variations in O₂, CO₂ and pH. Owing to the frequent occurrence of hypercarbic conditions in its natural environment(water total dissolved CO₂ may range from 0.82 to 1.79 mmol l^-1 depending on season and depth; Reid et al., 2000), and an anatomy that renders this species amenable to surgical sectioning of nerves innervating selected chemosensory areas, the tambaqui has been the focus of a concerted research effort to identify the locations and roles of chemoreceptors involved in respiratory reflexes to both hypoxia and hypercarbia (Sundin et al.,2000; Milsom et al.,2002; Reid et al.,2003; Florindo et al.,2004). Previous studies have focused primarily on reflex changes in ventilation and heart rate during hypoxic or hypercarbic exposures without monitoring blood gas or acid-base status. Blood gas and acid-base data were reported by Wood et al.(1998), but in the context of acid-base regulation in response to an acid challenge, and so without cardiorespiratory data. Thus, the objective of the present study was to characterize more fully blood gas and acid-base status, as well as the cardiovascular responses to hypercarbia, while testing the hypothesis that cardiorespiratory responses to hypercarbia in tambaqui are triggered by externally oriented branchial chemoreceptors that react specifically to changes in water CO₂ tension.

Juvenile tambaqui Colossoma macropomum Cuvier (1271±53 g,mean ± s.e.m.; N=17) were obtained from CAUNESP (Aquaculture Centre of the São Paulo State University - UNESP), Jaboticabal, SP, Brazil, and transported to the Federal University of São Carlos. These fish were third- or fourth-generation descendents of native tambaqui taken from the Amazon in 1993 and introduced into the south-eastern region of Brazil for aquaculture. Tambaqui were maintained in large fibreglass aquaria supplied with aerated water from an artesian well. The tanks were located outdoors and so exposed to a natural photoperiod. Temperature was maintained at 25°C, and fish were fed to satiation every second day.

Surgery was carried out on fish anaesthetized by immersion in an aerated solution of benzocaine (ethyl-p-aminobenzoate; 100 mg l^-1)and then transferred to an operating table where the gills were irrigated continuously with a more dilute anaesthetic solution (50 mg l^-1). For monitoring of blood gas and acid-base status using an extracorporeal circulation (Thomas, 1994),the caudal artery and caudal vein were cannulated(Axelsson and Fritsche, 1994);in addition, the caudal artery cannula was used for measurements of blood pressure (Pda) while the caudal vein cannula also served for saline or drug administration. Following exposure of the haemal arch by means of a lateral incision at the level of the caudal peduncle, flexible polyethylene tubing (Clay-Adams PE50, Becton-Dickenson and Co., Sparks, MD,USA) was inserted into the vessels in the anterior direction. Cannulae were filled with heparinized (100 i.u. ml^-1 ammonium heparin) modified(4.5 mmol l^-1 NaHCO₃) Cortland saline(Wolf, 1963) and flushed daily. To measure cardiac output, a 3S ultrasonic flow probe (Transonic Systems, Ithaca, NY, USA) was placed around the ventral aorta, which was accessed via an incision through the overlying epithelium within the opercular chamber. The operculum was reflected forward, and a small (∼1.5 cm) incision was made parallel to the ventral aorta in the epithelium near the isthmus. Blunt dissection exposed the ventral aorta, and the flow probe was then placed around the vessel using lubricating jelly (K-Y Personal Lubricant;Johnson and Johnson, Montréal, Canada) as an acoustic couplant. With this approach, disruption of the pericardium was avoided. Both incisions were closed, and the cannulae and flow probe lead were secured to the skin, with silk sutures. Ventilation was assessed by suturing brass plates (1 cm²) to the external surface of each operculum to measure breath-by-breath displacement of the opercula with an impedance converter. Finally, two holes were drilled through the snout between the nostrils using a Dremel tool, and a flared cannula (PE 160) was fed through each hole and secured in place with a cuff. These cannulae were used to deliver CO₂-equilibrated or acidified water into the flow of inspired ventilatory water. After surgery, fish were revived and transferred to individual holding boxes of opaque acrylic provided with flowing, aerated water for at least 24 h of recovery before experimentation.

Experiments commenced with a 5 min `pre' period of recording baseline ventilation and cardiovascular parameters. Tambaqui were then subjected to a series of injections of CO₂-equilibrated or acidified water (into the inspired water stream), and CO₂-equilibrated saline (into the caudal vein), with 4 or 6 min intervals between injections. Water injections(50 ml kg^-1) were delivered over a 20 s period into a snout cannula, and included aerated but otherwise untreated water (control) and water pre-equilibrated with 1%, 3%, 5% or 10% CO₂ in air. Measurement of pH for water equilibrated to each of these CO₂levels revealed values of 6.3, 5.6, 5.3 and 4.9 pH units, respectively. To differentiate between CO₂-induced cardiorespiratory effects and those triggered by the concomitant elevation of H⁺, each injection of CO₂-equilibrated water was followed by an injection of water titrated with HCl to the corresponding pH value; vigorous aeration before and after addition of HCl ensured removal of CO₂. Saline (control), or saline equilibrated with 5% or 10% CO₂ in air, was delivered as a bolus (2 ml kg^-1 over 20 s) into the caudal vein. Because CO₂-enriched saline injections were without effect (see Results),there was no further attempt to dissect the relative roles of CO₂vs H⁺.

Following the injection series, the extracorporeal blood circulation (see below) was initiated. Once the extracorporeal blood shunt was established and the measured variables had stabilized, baseline conditions for blood gases and acid-base status as well as ventilation and cardiovascular parameters under normoxic normocarbic conditions were recorded over a 5 min `pre' period. Blood pressure was monitored at regular (2-5 min) intervals by briefly (for 10-15 s)switching the caudal artery cannula from the extracorporeal blood loop to a pressure transducer using a T-junction and three-way valve. The water supplying the fish box was then rendered progressively hypercarbic by gassing a water equilibration column with 1%, 3%, and then 5% CO₂ in air(Cameron flowmeter model GF-3/MP, Port Aransas, TX, USA) for 15 min at each level. Using this protocol, the final water P_CO2(Pw_CO2) values achieved at each step were 7.2±0.5,15.5±0.9 and 26.3±2.5 mmHg, respectively (mean ± s.e.m., N=11). At the end of the final hypercarbic exposure, Pw_CO2 within the experimental chamber was rapidly returned to normocarbic conditions ('washed out') by increasing the flow of air-equilibrated water to the fish box.

Finally, the carbonic anhydrase inhibitor acetazolamide was used to investigate the cardiorespiratory effects of external vs internal hypercarbia. After the 5 min `pre' period of recording baseline conditions,acetazolamide (30 mg kg^-1) was administered via the caudal vein cannula and blood acid-base and cardiorespiratory parameters were monitored for 30 min. Fish were then exposed to hypercarbia by gassing the water equilibration column with 3% CO₂ in air (final Pw_CO2=13.6±0.5, N=7), and at the end of the 15 min hypercarbic exposure, CO₂ was again rapidly washed out of the system by increasing the flow of air-equilibrated water to the fish box. Acetazolamide was prepared by dissolving the drug in saline with added NaOH and then slowly titrating the pH down to a level as close as possible to physiological (final pH of the acetazolamide solution was approximately 8.5).

An extracorporeal blood circulation(Thomas, 1994) was used to continuously monitor blood gas and acid-base variables. Blood was withdrawn at a rate of 0.5 ml min^-1 from the caudal artery cannula using a peristaltic pump, and passed through an external circuit (of ∼1 ml volume)containing P_O2, P_CO2 and pH electrodes before being returned to the fish via the caudal vein cannula. To prevent clotting, the circuit was rinsed with heparinised (540 i.u. ml^-1) saline for 10-15 min prior to initiating blood flow. Arterial blood pH (pHa), P_CO2 (Pa_CO2) and P_O2 (Pa_O2) were measured using Metrohm(model 6.0204.100, Brinckman Instruments, Canada, Ltd., Mississuaga, ON,Canada; pH) and Cameron Instruments (CO₂, O₂) electrodes housed in thermostatted cuvettes and connected to a blood gas analyser (BGM 200; Cameron Instruments). A second peristaltic pump was used to withdraw water at a rate of 3.5 ml min^-1 from the mouth of the fish via one of the two buccal cannulae. This water was passed through thermostatted cuvettes containing pH (Metrohm model 6.0204.100) and P_CO2 (Cameron Instruments) electrodes connected to a blood gas analyser (Cameron Instruments) for the measurement of water pH (pHw) and Pw_CO2. Before each experiment, the pH electrodes were calibrated by pumping precision buffer solutions through the circuits until stable readings were recorded. A similar procedure was used to calibrate the O₂ and CO₂ electrodes with a zero solution (2 g l^-1 sodium sulphite; O₂ electrode only) and/or water equilibrated with appropriate gas mixtures (supplied by a GF-3/MP gas mixing flowmeter; Cameron Instruments).

Blood pressure was measured by connecting the caudal artery cannula to a pressure transducer (Model 1050BP, UFI, Morro Bay, CA, USA) linked to an amplifier (Biopac DA 100, Santa Barbara, CA, USA). The pressure transducer was calibrated daily against a static column of water. Blood flow was determined by attaching the factory-calibrated ultrasonic flow probe to a blood flowmeter(model T106; Transonic Systems, Ithaca, NY, USA). The frequency and amplitude of opercular displacements were monitored as indices of ventilation using an impedance converter (model 911, Biocom Inc; Culver City, CA, USA) that detected and quantified the changes in impedance between the brass plates attached to the opercula (Peyraud and Ferret-Bouin, 1960).

A data acquisition system (Biopac Systems) with Acknowledge data acquisition software (sampling rate set at 40 Hz) and a personal computer were used to convert all analogue signals (blood gases and pH, water P_CO2 and pH, blood pressure, blood flow and impedance recordings) to digital data. With this system, continuous data recordings were obtained for Pa_CO2, Pa_O2, pHa, Pw_CO2, pHw, mass-specific blood flow(V̇b), heart rate(fh; automatic rate calculation from the pulsatile V̇b (trace), mean Pda (arithmetic mean), systemic vascular resistance(R_s; Pda/V̇b and ventilation amplitude (V_AMP; the difference between maximum and minimum impedances). Note that Pda and R_s were recorded only intermittently in experiments where the extracorporeal blood shunt was utilised. In addition, cardiac stroke volume (Vs) was calculated by dividing mass-specific blood flow by heart rate, and ventilation frequency (fv) was determined from the impedance trace.

Data are reported as means ± 1 s.e.m. For experiments involving water or saline injections, mean ventilatory and cardiovascular data were compiled for 10 s intervals over the 20 s before and 100 s after the injection, except during the injection itself. For experiments employing the extracorporeal blood circulation, mean blood gas, water gas, acid-base, ventilatory and cardiovascular data were compiled over 2 min periods at selected intervals,apart from Pda and R_s, for which data were compiled only during the 10-15 s periods of recording. Data were analysed for statistical significance by one-way repeated measures analysis of variance(RM-ANOVA) followed by post hoc multiple comparisons using the Holm-Sidak method, as appropriate. Where assumptions of normality or equal variance were violated, equivalent non-parametric analyses were employed. The commercial package SigmaStat v3.0 (SPSS Inc.) was used to carry out statistical analyses, and the fiducial limit of significance in all cases was 5%.

Exposure of tambaqui (N=11) to three stepwise increases in water CO₂ tension to achieve Pw_CO2 values of 7.2±0.5 (1% CO₂ delivered to the equilibration column),15.5±0.9 (3% CO₂ delivered to the equilibration column) and 26.3±2.5 mmHg (5% CO₂ delivered to the equilibration column)within 15 min elicited a significant respiratory acidosis(Fig. 1D,F), together with a marked hyperventilation (Fig. 1B,C) accompanied by bradycardia(Fig. 2E). The changes were in general P_CO2-dependent, with V_AMPincreasing to a greater extent than fv (e.g. 128%vs 42% increases, respectively, at the highest Pw_CO2). The significant increases in Pa_O2 at the higher water CO₂ tensions(Fig. 1E) likely reflected this hyperventilation. Although heart rate was lowered (by 18-35%) during hypercarbia, blood flow was maintained(Fig. 2D) because of a significant rise in cardiac stroke volume (17-62%; Fig. 2F). Neither arterial blood pressure nor systemic vascular resistance were affected by exposure to hypercarbia (Fig. 2B,C).

Rapid removal of CO₂ from the water resulted in the equally rapid return of ventilation (Fig. 1B,C), heart rate (Fig. 2E) and stroke volume (Fig. 2F) towards pre-exposure levels. Blood flow increased significantly during the rapid washout, by 16-23%(Fig. 2D), owing to the persistent elevation of cardiac stroke volume. The greater blood flow was accompanied by a significant lowering of systemic resistance(Fig. 2C). Although blood gas and acid-base variables also recovered during the lowering of water CO₂ (Fig. 1D-F),these changes were slower to occur, such that even 20 min after initiation of the rapid washout, Pa_CO2 remained significantly (Wilcoxon signed rank test, P=0.008) elevated over the pre-exposure value by 6.62±2.20 mmHg (N=8), while pHa was depressed by 0.18±0.04 units (N=8). Notably, neither ventilation amplitude(Wilcoxon signed rank test, P=0.469) nor frequency (paired Student's t-test, P=0.094) was significantly different from the baseline value at this time, suggesting that ventilation, at least, was more sensitive to changes in Pw_CO2 than to those in Pa_CO2.

This observation was confirmed by administration of the carbonic anhydrase inhibitor acetazolamide, a treatment that generated a significant respiratory acidosis (Fig. 3D,F) in the absence of any change in Pw_CO2(Fig. 3A). A single bolus injection of acetazolamide caused Pa_CO2 to increase approximately threefold over the subsequent 30 min, at the same time lowering pHa by 0.26±0.02 units (N=8), yet was without significant effect on ventilation (Fig. 3B,C) or most cardiovascular variables(Fig. 4); systemic resistance(Fig. 4C) and cardiac stroke volume (Fig. 4F) did differ statistically from the pre-injection value at single points in each case. By contrast, exposure of acetazolamide-treated fish to 15 min of hypercarbia(first broken line), in which Pw_CO2 reached 13.6±0.5 mmHg, produced changes in ventilation and cardiovascular variables that were virtually identical to those observed in untreated fish exposed to a similar level of hypercarbia, even though the respiratory acidosis was more profound in treated fish; V_AMP and fv rose by 110% and 24%, respectively, and heart rate fell by 20% (Fig. 4E), but blood flow was maintained (Fig. 4D)owing to the concomitant 19% increase in stroke volume. A small and transient increase in blood pressure also occurred(Fig. 4B); this response was not observed in untreated fish.

The recovery of ventilation and cardiovascular variables in response to the rapid lowering of Pw_CO2 (second broken line) was particularly striking in acetazolamide-treated fish because it occurred on the backdrop of a sustained, severe, respiratory acidosis (Figs 3 and 4). The significant decreases in ventilation variables (Fig. 3B,C) and stroke volume (Fig. 4F) as well as the rise in heart rate(Fig. 4E) tracked changes in Pw_CO2 rather than Pa_CO2(Fig. 4A). The lack of correspondence between V_AMP and Pa_CO2is illustrated by a representative data recording for an individual fish(Fig. 5). As in untreated fish,blood flow increased significantly with the onset of rapid water CO₂ washout (Fig. 4D), but the extent of the blood flow increase (7-13%) was smaller in acetazolamide-treated tambaqui and was not accompanied by a significant decrease in systemic resistance (Fig. 4C).

As a final test of the potential for changes in blood CO₂tension to elicit cardiorespiratory responses, tambaqui were injected with saline equilibrated with 5% or 10% CO₂ in air. Assuming complete mixing of the saline bolus (2 ml kg^-1 delivered over 20 s) with venous blood, estimating venous P_CO2 to be approximately 4 mmHg (based on the measured Pa_CO2 of ∼3mmHg), and using the cardiac output measured under resting conditions (21 ml min^-1 kg^-1), these internal injections of 5% or 10%CO₂-equilibrated saline would be expected to yield transient P_CO2 values of 11.5 or 20mmHg, respectively, in the blood at the gill. Increases in blood pressure (7-11%), blood flow (20-24%) and systemic resistance (6-11%) occurred in response to these injections but were attributable to volume loading, since similar increases (9%, 31% and 4%,respectively) were also observed upon injection of air-equilibrated saline(Table 1). No specific effect of internally injected CO₂ was detected for any measured variable(Table 1).

The effects of injecting CO₂-enriched water into the flow of ventilatory water were compared with reactions to the injection of acidified water to distinguish between the roles of CO₂ and H⁺ in eliciting cardiorespiratory responses to hypercarbia. A marked, P_CO2-dependent bradycardia accompanied the injection of CO₂-enriched water into the buccal cavity of tambaqui(Fig. 6A-D). At the peak of the response (∼20-30 s after beginning the injection), fhwas decreased by 20-49% for injections of water equilibrated with 1-10%CO₂. Because cardiac stroke volume was maintained or increased only slightly (data not shown), the bradycardia resulted in significant 14-45%reductions in cardiac output with injection of all levels of CO₂-enriched water (Table 2). The fish also exhibited significant P_CO2-dependent increases in systemic resistance(Fig. 6E-H), which were reflected in significant increases in blood pressure(Table 2). The different responses of systemic resistance to bolus injections of CO₂-enriched water (Fig. 6E-H) vs exposure to hypercarbic water (Figs 2C, 4C) were striking. By contrast with the effect of CO₂, injection of acidified water was generally without significant effect, apart from a small (11%) and transient depression of heart rate that occurred only in response to injection of the most acidic(pH 4.9) water (Fig. 6C), and a correspondingly slight 5% depression of cardiac output(Table 2). Ventilatory responses to CO₂-enriched water injections were more sporadic; V_AMP increased significantly only with injection of 10%CO₂, while significant frequency responses were limited to injections of 5% and 10% CO₂(Table 2). Responses to the parallel injections of acidified water were either insignificant or of substantially lower magnitude (Table 2). Injection of air-equilibrated water was without significant effect in all cases (data not shown).

Previous studies of CO₂/pH chemoreception in tambaqui focused on determining the location and innervation of the receptors that mediate the cardiorespiratory responses to hypercarbia(Sundin et al., 2000; Milsom et al., 2002; Florindo et al., 2004). Tambaqui, like other exclusively water-breathing fish (reviewed by Milsom, 2002), appear to lack central CO₂/pH chemoreceptors(Milsom et al., 2002). Using the elimination of hypercarbic cardiorespiratory responses upon denervation as the criterion for involvement of a particular population of peripheral receptors, a predominantly branchial location for CO₂/pH chemoreceptors was revealed (Sundin et al., 2000; Milsom et al.,2002; Florindo et al.,2004). Cardiac and ventilation rate responses, in particular, were attributed to receptors that were exclusively branchial, with the hypercarbic bradycardia mediated by receptors confined to the first gill arch, and the increase in breathing frequency by receptors distributed across all gill arches (Sundin et al., 2000). In contrast, the persistence of blood pressure and, to some extent,ventilation amplitude responses to hypercarbia in tambaqui subjected to total gill denervation suggested the involvement of extrabranchial receptors(Sundin et al., 2000; Florindo et al., 2004). The findings of the present study add to this existing information on receptor location by characterizing the orientation of the CO₂/pH chemoreceptors in tambaqui as well as their specificity for CO₂vs H⁺.

The cardiorespiratory responses elicited by hypercarbia may be initiated by branchial chemoreceptors that detect changes in water CO₂/pH,and/or by receptors that monitor blood CO₂/pH levels, because exposure to elevated ambient CO₂ causes a rise in blood P_CO2 and a concomitant fall in blood pH(Fig. 1). Thus, three experimental approaches were employed to discern between external and internal orientation of the branchial CO₂/pH receptors. First, tambaqui were treated with acetazolamide to inhibit red blood cell carbonic anhydrase activity. Assuming that CO₂ excretion in tambaqui follows the pathway mapped out for teleost fish in general (e.g. Perry, 1986; Tufts and Perry, 1998),carbonic anhydrase will catalyze the interconversion of CO₂ and HCO₃^- within the red blood cell, a step that is critical to the transfer of CO₂ from tissues to blood, and from blood to ventilatory water. In acetazolamide-treated fish, this step would be slowed to the uncatalyzed rate, thereby causing CO₂ retention(Henry and Heming, 1998). As expected, treatment of tambaqui with acetazolamide evoked the classic response(e.g. Hoffert and Fromm, 1973)of a profound respiratory acidosis, in which Pa_CO2approximately tripled in 30 min while pHa fell by 0.26 units(Fig. 3D,F). Yet despite this marked internal hypercapnia, the typical cardiorespiratory responses to CO₂/pH of hyperventilation and bradycardia (Figs 1B,C, 2E) were not observed until the acetazolamide-treated tambaqui were exposed to external hypercarbia (Figs 2B,C, 3E). These findings argue strongly in favour of branchial CO₂/pH chemoreceptors with a solely external orientation. Observations from the two additional experimental approaches were in agreement with this conclusion, allaying concerns about any non-specific side effects of the drug treatment. Chemoreceptor impairment, in particular, was considered a possibility because carbonic anhydrase plays a role in CO₂ chemoreception in both invertebrates and vertebrates(e.g. Iturriaga et al., 1991; Swenson and Hughes, 1993; Erlichman et al., 1994; Coates et al., 1998; see also review by Iturriaga, 1993),although the very similar responses of acetazolamide-treated and untreated tambaqui to Pw_CO2 values of 14-15 mmHg rendered this prospect unlikely.

Externally oriented branchial CO₂/pH chemoreceptors would account for the close correspondence between cardiorespiratory adjustments and changes in water P_CO2 during the rapid washout of CO₂ from the water following hypercarbic exposures, as well as the independence of these responses from arterial P_CO2. The CO₂ electrode response time was likely faster for water than for blood measurements, owing to the different viscosities of these liquids. However, any concern that the apparent tracking of cardiorespiratory responses to water rather than blood P_CO2 simply reflected different response times was alleviated by the particularly marked differences in the time courses of Pw_CO2 and cardiorespiratory variable changes during washout in acetazolamide-treated tambaqui(Fig. 5).

The observation that injection of CO₂-laden water into the buccal cavity triggered cardiorespiratory reactions(Fig. 6, Table 2) that were not detected in response to the injection of a bolus of CO₂-enriched saline into the caudal vein (Table 1) was also consistent with an external orientation for the branchial CO₂/pH chemoreceptors. In the latter experiment, it was assumed that externally oriented receptors would be preferentially stimulated by injecting CO₂-enriched water into the mouth, whereas CO₂-enriched saline injections would preferentially stimulate internally oriented receptors. Although exclusive stimulation of internally or externally oriented receptors is likely to be impossible with this approach owing to the potential for CO₂ diffusion across the gill epithelium, the extent of activation of blood-oriented receptors by external(water) relative to internal (saline) injections was probably trivial. Similarly, the absence of response to internal injection suggested that water-oriented receptors were activated to a trivial extent by CO₂-enriched saline. Alternative explanations for the lack of response to CO₂-enriched saline injection are that the P_CO2 increase achieved by the injection of CO₂-equilibrated saline was insufficient (in length or magnitude)to trigger internally oriented CO₂ receptors, or that the CO₂ was converted to HCO₃^- and/or excreted during transit through the circulation and gills. These possibilities cannot be ruled out, but the most parsimonious explanation of the data in the context of the results for the two other experimental approaches is that the cardiorespiratory responses to hypercarbia in tambaqui are linked to the activation of branchial CO₂/pH chemoreceptors that are oriented only towards the external (water) milieu. In dogfish(Perry and McKendry, 2001),Atlantic salmon (Perry and McKendry,2001) and rainbow trout(McKendry and Perry, 2001; Perry and Reid, 2002),externally oriented branchial CO₂/pH chemoreceptors were also deduced from data generated using experimental approaches similar to those of the present study. The existence of internally oriented receptors that detect changes in blood P_CO2 and/or pH was suggested by earlier studies in which indirect correlative relationships between ventilation and blood P_CO2 or acid-base status were constructed (e.g. Heisler et al., 1988; Graham et al., 1990; Wood and Munger, 1994; see also review by Gilmour, 2001). Increasingly, however, the weight of evidence from experiments designed to distinguish directly between internal and external stimuli suggests that reflex cardiorespiratory responses to hypercarbia are mediated by externally oriented chemoreceptors. In this regard, the situation for CO₂/pH sensing differs from that for O₂, in that a population of internally oriented O₂ chemoreceptors exists; these receptors are distributed over all gill arches and linked specifically to ventilatory reflexes (Burleson et al.,1992; Burleson,1995).

The present study also included an experiment to discern between the specific effects of CO₂vs H⁺ in the initiation of cardiorespiratory responses to hypercarbia; namely, a comparison of responses to the injection of CO₂-enriched water into the buccal cavity with those elicited by CO₂-free water acidified to the pH of the corresponding CO₂-laden water injection. The results clearly demonstrated that CO₂ itself was the key factor controlling cardiorespiratory function (Fig. 6, Table 2). Although the injection of acidified water was accompanied by significant changes in heart rate, blood flow, ventilation amplitude and ventilation frequency, these effects were in general restricted to injection of the most acidic water (pH 4.9) and were in all cases of much smaller magnitude (11%,5%, 15% and 16%, respectively) than those produced by injection of the corresponding CO₂-laden water (49%, 45%, 44% and 34%,respectively). The data imply that minor cardiorespiratory reactions may occur with strongly acidic stimuli, but that the chemoreceptors respond predominantly to CO₂ rather than to protons. The small magnitude of the effects, coupled with the need for intense acidic stimuli, probably accounted for the absence of response to acid injection in a previous study on tambaqui (Sundin et al.,2000). Acid injections into the inspired water were also found to be without significant effect in traira(Reid et al., 2000) and dogfish (Perry and McKendry,2001), while the slight impact of acid injection on ventilation amplitude in Atlantic salmon was attributed to CO₂ formed when HCO₃^- ions in seawater were titrated by the added H⁺ (Perry and McKendry,2001). Earlier studies similarly reported that environmental acidification in the absence of elevated Pw_CO2 had little effect on ventilation in rainbow trout(Janssen and Randall, 1975; Neville, 1979; Thomas and Le Ruz, 1982) and taken as a whole, these data suggest that the cardiorespiratory responses to hypercarbia are mediated by externally oriented branchial chemoreceptors that respond specifically to CO₂. Nevertheless, protons produced by the hydration of CO₂ that diffuses into the cell probably play a role in signal transduction within chemoreceptor cells in fish, as in mammals (e.g. Iturriaga et al., 1991, 1993; Gonzalez et al., 1994).

In addition to contributing information on receptor orientation and stimulus modality to the existing data on chemoreceptor localization in tambaqui, the present study more fully characterizes the cardiovascular reflexes of tambaqui to hypercarbia, as previous work focused on changes in heart rate and ventilation (Sundin et al.,2000; Milsom et al.,2002; Florindo et al.,2004). In the presence of ambient CO₂ tensions elevated to at least 7 mmHg (lower CO₂ tensions were not tested), tambaqui consistently exhibited bradycardia, greater cardiac stroke volume and hyperventilation, marked by increases of both frequency and amplitude. Arterial blood pressure rose in some cases, presumably in response to increased systemic resistance, and despite simultaneous reductions in cardiac output. This pattern of cardiorespiratory reflexes to hypercarbia emphasises equally the relatively conserved nature of some responses and the highly variable nature of others. For example, hyperventilation is a common response to hypercarbia, both among the studies on tambaqui(Sundin et al., 2000; Milsom et al., 2002; Florindo et al., 2004) and within fish in general (see review by Gilmour, 2001). In some species, including tambaqui (this study; Florindo et al., 2004) and the elasmobranchs examined to date (Randall et al., 1976; Graham et al.,1990; Perry and Gilmour,1996; McKendry et al.,2001; Perry and McKendry,2001), increases in ventilation amplitude are more important contributors to the hyperventilatory response than are frequency adjustments,but overall there is a high degree of interspecific variation in the relative importance of frequency vs amplitude changes(Gilmour, 2001). Bradycardia also is a common response to hypercarbia (see review by Perry and Gilmour, 2002), yet while tambaqui responded to higher CO₂ tensions (5%) in all studies with bradycardia (Sundin et al.,2000; Milsom et al.,2002; Florindo et al.,2004), conflicting results were obtained using lower levels of CO₂. For example, Florindo et al.(2004) observed increases in heart rate at 1-2.5% CO₂, in contrast to the bradycardia observed in the present study (Fig. 2E)and the lack of heart rate response observed by Sundin et al.(2000) at similar CO₂ tensions. It is possible that this discrepancy reflects differences in the length of hypercarbic exposure, as the first measurement time utilized by Florindo et al.(2004) was 60 min, while the hypercarbic period in the present study was limited to 15 min.

Although hyperventilation and bradycardia are common responses to hypercarbia, they are by no means universal. White sturgeon Acipenser transmontanus, for example, responded to hypercarbia with a tachycardia(Crocker et al., 2000), while a number of species, including channel catfish(Burleson and Smatresk, 2000)and brown bullhead (see table 2 in Gilmour, 2001; table 1 in Perry and Gilmour, 2002) were resistant to hypercarbia, either failing to change heart rate or ventilation,or exhibiting very attenuated responses. To some extent, this variation may reflect species differences in sensitivity to CO₂. For example,neither the European eel Anguilla anguilla nor the closely related American eel Anguilla rostrata responded to CO₂ tensions of 5-6mmHg (McKenzie et al.,2002; see table 2 in Gilmour,2001; table 1 in Perry and Gilmour, 2002), but adjustments of both heart rate and ventilation were exhibited by European eels exposed to Pw_CO2 values of 10-80 mmHg (McKenzie et al.,2002). Similarly, it was necessary to raise water P_CO2 to 14 mmHg before hyperventilatory responses appeared in carp Cyprinus carpio (Soncini and Glass, 2000), and to ∼38 mmHg to observe significant hypercarbic responses in traira (Reid et al., 2000). Like eel, carp and traira, the results of earlier studies indicated that tambaqui are relatively insensitive to changes in water CO₂ tension (Sundin et al.,2000). The results of the present study and that of Florindo et al. (2004), however, indicated that tambaqui are more sensitive to CO₂ than originally thought. Variation in the methods used to expose fish to hypercarbia may account for this difference.

The remaining cardiovascular adjustments to hypercarbia, including blood pressure, systemic resistance, cardiac output and stroke volume, exhibit a high degree of interspecific variation among fish, encompassing essentially all possible response patterns (see table 2 in Perry and Gilmour, 2002). The increases in arterial blood pressure and systemic resistance, coupled with constant or slightly reduced cardiac output observed for tambaqui under some conditions in the present study (Fig. 6, Table 2), were reminiscent of the responses of the salmonid fish that have been examined to date (Perry et al., 1999; McKendry and Perry, 2001; Perry and McKendry, 2001). The appearance of these responses, however, was dependent upon the method of CO₂ delivery (injection of a bolus of CO₂-enriched water into the mouth vs exposure to hypercarbic water), perhaps because of preferential stimulation of different receptor populations. In addition,Milsom et al. (2002) reported the existence in tambaqui of receptors sensitive to CO₂ that had an inhibitory influence on ventilation frequency. The chemoreceptor control of cardiorespiratory function during hypercarbia in fish is clearly complex,likely involving multiple receptor populations in a variety of locations(although with a branchial concentration) that are linked to different cardiorespiratory parameters with positive and/or negative influences. Resolving this complexity is an ongoing challenge, particularly because the patterns of chemoreceptor control as well as the cardiorespiratory responses to hypercarbia vary among fish species for reasons that remain elusive. Despite the variability, CO₂ has emerged as an important modulator of cardiorespiratory function in fish, with responses mediated, in all species that have been examined to date, by receptors that are oriented towards the external environment with sensitivity specifically to CO₂.

List of symbols
 
• fh

  heart rate

 
• fv

  ventilation frequency

 
• Pa_CO2

  arterial P_CO2

 
• P_CO2

  partial pressure of CO₂

 
• Pda

  pressure in dorsal aorta

 
• pHa

  arterial blood pH

 
• pHw

  water pH

 
• P_O2

  partial pressure of oxygen

 
• Pw_CO2

  water P_CO2

 
• R_s

  systemic vascular resistance

 
• V_AMP

  ventilation amplitude

 
• V̇b

  mass-specific blood flow

 
• Vs

  stroke volume

Thanks are extended to Dr Elizabeth Urbinati (CAUNESP, Jaboticabal, SP) for providing the tambaqui. This study was supported by NSERC of Canada operating and equipment grants to K.M.G., W.K.M. and S.F.P., and FAPESP(Fundação de Amparo à Pesquisa do Estado de São Paulo) and CNPq (the Brazilian National Research Council for Development of Sciences and Technology) grants to F.T.R. S.G.R. is currently supported by a Parker B. Francis Fellowship (Francis Families Foundation).