The Effects of Branchial Denervation and Pseudobranch Ablation on Cardioventilatory Control in an Air-Breathing Fish

Amia calva is a primitive fish with the ability to breathe both air and water. It possesses a swimbladder adapted to function as an air-breathing organ and gills for water-breathing. Relatively little is known about control of reflex ventilatory and cardiovascular responses to respiratory gases in Amia or air-breathing fish in general. Air-breathing and gill ventilation are stimulated by aquatic hypoxia in Amia (Johansen et al. 1970; Randall et al. 1981), and ventilatory responses to blood hypercapnia and acidosis do not occur unless they are associated with a reduction in blood O₂ content (McKenzie et al. 1991). This indicates that ventilatory control in Amia is similar to that observed in purely water-breathing fish, where blood and water O₂ status are the primary stimuli for cardioventilatory reflex responses (Randall and Jones, 1973; Smith and Jones, 1981; Randall, 1982).

There is much evidence indicating that hypoxic reflexes in fish are mediated by O₂-sensitive chemoreceptors in the gills and pseudobranch, innervated by cranial nerves VII, IX and X. Nerve activity sensitive to internal and external O₂ levels has been recorded from nerve X to the first gill arch of tuna (Milsom and Brill, 1986) and nerve IX to the first gill arch of trout (Burleson and Milsom, 1990).

Section of cranial nerves IX and/or X to the gill arches abolishes reflex bradycardia in response to aquatic hypoxia or NaCN (a potent chemoreceptor stimulant) in various teleost species (Smith and Jones, 1978; Fritsche and Nilsson, 1989; Burleson and Smatresk, 1990a). In elasmobranchs, nerves V and VII innervating the buccal cavity also require sectioning (Butler et al. 1977).

Complete branchial denervation abolishes ventilatory reflex responses to hypoxia in anaesthetised, spontaneously ventilating Ictalurus punctatus (Burleson and Smatresk, 1990a) and Lepisosteus oculatus, an air-breathing fish (Smatresk, 1991). The failure of branchial denervation to abolish ventilatory reflex responses completely in various other species (Hughes and Shelton, 1962; Saunders and Sutterlin, 1971) has been attributed to the existence of an extrabranchial insensitive chemoreceptor (Bamford, 1974) or to afferent information arising from the pseudobranch and/or nerves serving one particular gill arch, which were left intact (Smatresk, 1991). In channel catfish, O₂ sensitivity is located diffusely throughout the gills, and one branchial nerve left intact unilaterally is sufficient to stimulate a ventilatory response (Burleson and Smatresk, 1990a).

Recent evidence, however, suggests that ventilatory responses to hypoxia and acidosis in dogfish and trout may be mediated humorally by the circulating catecholamines NA and A, possibly via a direct effect on central respiratory motoneurones (Randall and Taylor, 1989; Taylor and Randall, 1988; Aota et al. 1990). There is evidence to suggest that circulating catecholamines play a role in gill ventilatory responses to acidosis in Amia (McKenzie et al. 1991).

The present study in Amia employed total branchial denervation (branchial branches of cranial nerves IX and X) and pseudobranch ablation to investigate the role of O₂-sensitive chemoreceptors in the gills and pseudobranch in reflex cardiovascular, air-breathing and gill ventilatory responses to hypoxia, sodium cyanide (NaCN) and catecholamines, and also to investigate the role of circulating catecholamines in reflex gill ventilatory responses to hypoxia. Sodium cyanide was used as a stimulant for O₂-sensitive chemoreceptors (Lahiri et al. 1970; Burleson and Smatresk, 1990a; Smatresk, 1986). A conscious animal preparation was used to avoid the effects of anaesthesia, which may compromise cardiovascular and ventilatory reflex responses.

Amia weighing 400–1000 g were netted in Lake Erie and air-freighted to UBC, where they were maintained in a large outdoor fibreglass tank supplied with dechlorinated Vancouver tapwater (pH approx. 6.5, 7–12°C). Fish were fed a diet of live goldfish and salmon fry. Before experimentation, Amia were placed in Plexiglas holding tanks and the water temperature was raised to 20°C, over a 3- to 5-day period, to increase air-breathing frequency (Johansen et al. 1970).

Animals were anaesthetized in a tricainemethanesulphonate (MS-222) solution at a concentration of 0.01%, buffered with sodium bicarbonate. Following transfer to an operating table, fish were artificially ventilated with a buffered MS-222 solution at 0.005%, oxygenated with 100% O₂ to prevent hypoxia during surgery when only half of the gills were ventilated. A dorsal aortic cannula (PE 50, Intramedic) was implanted using the technique of Soivio et al. (1972). A buccal cannula was fitted using flared PE 50 (Intramedic) passed through a small hole in the roof of the mouth and secured with a cuff and sutures. An opercular cannula was fitted using flared PE 190 (Intramedic) passed through a small hole in the operculum and secured with a cuff and sutures. The operculum was reflected forward and a small incision made in the epithelium dorsal and posterior to the fourth gill arch. This allowed access to the branchial branches of cranial nerve X, serving all four gill arches. These were gently dissected free of connective tissue and sectioned with iris scissors, care being taken not to damage gill vasculature or musculature. Cardiac and visceral branches (including that to the swimbladder) of nerve X were left intact. The incision was closed with absorbable sutures (Ethicon 2.0). Cranial nerve IX serving the first gill arch was exposed by making a small incision at the base of the gill filaments where the arch meets the roof of the opercular cavity, and dissecting the nerve free of connective tissue. The nerve was then sectioned with iris scissors. The pseudobranch in Amia is glandular and situated in the roof of the mouth just anterior to the first gill arch. It was exposed via a small incision and removed by cautery. The same procedure was followed on the other side so that denervation and pseudobranch ablation were bilateral. Surgery took between 30 and 40 min and all nerve sections were confirmed post mortem, by dissection. Sham-operated animals had gill nerves exposed as described, but not sectioned.

Following surgery, fish were transferred to individual darkened Plexiglas chambers, ventilated with water until ventilatory movements resumed and then allowed to recover for 48h with a continuous flow of aerated water (approx. 900 ml kg⁻¹ min⁻¹) through the chamber. The anterior end of the chamber had an air space to allow air-breathing. The dorsal aortic cannula was flushed twice daily with heparinised saline.

Following recovery, oxygen consumption $(V˙O2)$ was measured in both denervated and sham-operated fish (see below for analytical methods). Subsequently, the water-filled opercular cannula was attached to a pressure transducer (Statham P23BB), allowing measurement of ventilation rate (fG, beats min⁻¹) and opercular pressure amplitude (POP, kPA). Opercular pressure amplitude was used as an index of ventilatory effort. The saline-filled dorsal aortic cannula was attached to a pressure transducer (Statham P23Db) for measurement of heart rate (fH, beats min⁻¹) and dorsal aortic blood pressure (PDA, kPA). The output from both transducers was displayed on a pen recorder (Gould 8188-2202-XX). Air-breathing frequency (fAB, breaths h⁻¹) was visible as large pressure excursions on the opercular trace, associated with changes in fH and PDA. Air breaths were all verified visually through a small hole in a screen separating experimenter and fish.

Ventilatory and cardiovascular responses to a variety of different stimuli were measured in sham-operated and denervated animals. Experiments on any one fish were performed over a 2 day period, with overnight recovery between days. No experiments were initiated until ventilatory and cardiovascular variables had remained stable for 30min. Animals were exposed to the following treatments, assigned randomly.

Aquatic hypoxia

Using a three-way valve at the inflow of the Plexiglas chamber, animals were acutely exposed to hypoxic water [O₂ partial pressure $(PO2)=6.31±0.16kPa]$ obtained by bubbling 100% N₂ counter-current to water flowing through a stripping column. Fish were exposed for 15 min, and then normoxic flow was resumed. Immediately prior to exposure, and following 5 min of exposure, arterial blood samples (700μl) were withdrawn anaerobically from the dorsal aortic cannula, via a three-way stopcock at the pressure transducer. Blood samples were replaced with an equal volume of heparinised saline. 500 μl of blood was immediately centrifuged, the plasma decanted and frozen in liquid nitrogen for subsequent analysis of plasma catecholamine levels. Arterial blood pH (pHa), $PO2(PaO2)$ and O₂ content $(CaO2)$ were all measured on the remainder of the blood sample (see below for analytical methods). Animals were allowed a minimum of 2h recovery following hypoxic exposure.

Either 1ml of saline (control) or 1mg of sodium cyanide (NaCN), dissolved in 1 ml of saline (experimental), was given as a bolus injection into the buccal cavity via the buccal cannula.

Either 300 μl of heparinised saline (control) or 300 μg of NaCN, dissolved in 300μl of saline (experimental), was delivered as a bolus injection via the dorsal aortic (DA) cannula.

0.5mlkg⁻¹ saline (control), 0.5 ml kg⁻¹ 10⁻⁵moll⁻¹ noradrenaline hydrochloride (Sigma) in saline (experimental) or 0.5 ml kg⁻¹ 10⁻⁵moll⁻¹ adrenaline bitartrate (Sigma) in saline (experimental) was infused via the DA cannula.

In all treatments, control and experimental injections or infusions were performed in random order. Animals were given a minimum of 30 min to recover between injections.

Oxygen consumption was calculated for both aquatic and aerial phases. Steady-state measurements of $PO2$ of inflow and outflow water were made using a Radiometer PHM72 acid-base analyser and Radiometer O₂ electrode thermo-statted to 20 °C. Oxygen consumption in the aquatic phase was then calculated by the Fick equation. Measurements of the decline in $PO2$ in the sealed forward airspace over a 3h period were used to calculate $V˙O2$ by air-breathing. Plexiglas chambers used for O₂ consumption were those used by Randall et al. (1981). In that study, it was demonstrated that there was no exchange of gases between aquatic and aerial phases within the chamber.

Arterial blood pH and $PaO2$ were measured using a Radiometer PHM72 acid-base analyser and associated electrodes, thermostatted to 20°C. Arterial blood O₂ content was measured on 30 μl blood samples using the method of Tucker (1967) and a Radiometer electrode thermostatted to 20°C. Arterial plasma noradrenaline and adrenaline concentrations ([NA] and [A], respectively) were measured on alumina-extracted samples by high performance liquid chromatography with electrochemical detection, using a Waters plasma catecholamine reverse-phase column, Waters M460 electrochemical detector and Waters 510 solvent delivery pump (Waters/Milhpore), as described by Woodward (1982), with peaks generated on a chart recorder. Catecholamine concentrations were calculated by integrating the area under the peaks with Sigmascan (Jandel Scientific) and an Olivetti M24 computer, and comparing with appropriate standards.

Ventilatory and cardiovascular responses were analysed for a control period and at designated intervals following each experimental intervention. Ventilation and heart rate were counted for 30 s in each minute, and POP and PDA were averaged from six measurements within that period; for 2 min control and at 1,2.5,5,10 and 15 min following intervention. Air-breathing frequency was averaged for the whole measurement period following intervention. When a blood sample was withdrawn, cardiovascular measurements were made at 4 min.

Cardiovascular and gill ventilatory responses were normalised as percentage change from control and, following arcsine transformation, compared at each time interval with an analysis of variance (ANOVA). In those cases where the ANOVA indicated a significant difference amongst time intervals, each time interval was compared a posteriori with the control value. Gill ventilatory responses to NA and A infusion were also assessed by comparing control and peak responses with a paired t-test, using the normalised transformed values. Normalised responses were used for graphical display.

Blood measurements taken during aquatic hypoxia exposure were compared with measurements obtained immediately prior to hypoxia exposure with a paired t-test. Air-breathing frequency under control normoxic conditions was compared with fAB under hypoxic conditions using a paired t-test. For NaCN and catechol-amine infusions, fAB following control (saline) injection was compared with fAB following the associated experimental injection using paired t-tests. Air-breathing frequency was compared between sham-operated and denervated fish, for each experimental intervention, using unpaired t-tests. P<0.05 was taken as the fiducial limit of significance.

Gill denervation and pseudobranch ablation resulted in a significant reduction in $V˙O2$⁠, fAB and $PaO2$ and a significant increase in the level of circulating NA. At 20°C, in aquatic normoxia, sham-operated Amia obtained 0.1% of their total O₂ uptake by air-breathing. Following gill denervation and pseudobranch ablation, animals (denervates) had a significantly reduced O₂ consumption rate (30% lower than sham-operated Amia), and there was no O₂ uptake by air-breathing (Table 1). Mean control values for PDA, fH, POP and/G are also given in Table 1. There was no significant difference between sham-operated Amia and denervates for these variables, although denervates appeared to have a higher POP. During normoxia, fAB was very low in the sham-operated Amia, usually zero, and only one denervate breathed air, on two occasions (Table 2). In normoxia, denervates showed no differences in pHa, $CaO2$ and [A] as compared with sham-operated Amia, but $PaO2$ was significantly lower and [NA] significantly higher in denervates (Table 3).

In sham-operated Amia, acute exposure to aquatic hypoxia elicited significant cardiovascular and ventilatory responses (Fig. 1). A gradually developing brady-cardia was evident, with/H significantly reduced following 15 min of exposure. Gill ventilation increased, with significant changes in POP and fG at 5 min that were sustained until the end of hypoxic exposure. Air-breathing frequency increased significantly (Table 2), with most of the air-breaths occurring in the first 5 min of hypoxic exposure. At 5 min of exposure, pHa and $CaO2$ were maintained at pre-exposure values, but $PaO2$ decreased significantly. Arterial NA concentrations increased, but there was no significant effect on [A] (Table 3).

In denervates, the response to aquatic hypoxia was distinctly different (Fig. 1). The bradycardia response was abolished. Gill ventilation did not increase until 10 min of exposure, with a sustained increase in/G and a transient increase in POP that was no longer evident at 15 min. There was no change fAB during hypoxia, and air-breathing responses were abolished (Table 2). At 5 min of exposure, pHa was not changed from pre-exposure values. Arterial blood O₂ content was not maintained at the pre-exposure value but decreased significantly, as did $PaO2$⁠. The concentration of circulating NA increased, but there was no significant effect on [A] (Table 3).

In sham-operated Amia, bolus injection of NaCN into the buccal cavity (Fig. 2) elicited a transient bradycardia and a transient increase in POP. Arterial blood pressure and/G did not change significantly from control values. In two out of six fish, external NaCN immediately stimulated an air-breath (Table 2) but, overall, NaCN in the buccal cavity had no significant effect on air-breathing. It is of interest to note that partial denervates (i.e. animals with a branchial branch of IX or X and/or the pseudobranch intact) showed an immediate and significant increase in fAB within 5 min following external NaCN (Table 2). External saline control injections had no effect on any variable (Fig. 2).

In denervates (Fig. 2), all cardiovascular and ventilatory responses to NaCN in the buccal cavity were abolished, with no significant change in any variable over time. One animal took an air-breath in response to both external saline and NaCN, but in the latter case at 13 min post-injection, which indicates that it was not in response to the NaCN (Table 2). Injection of a saline bolus into the buccal cavity had no effect on any variable (Fig. 2).

Bolus injection of NaCN into the dorsal aorta of sham-operated Amia had no significant effect on PDA and/H, but significantly stimulated POP and/G (Fig. 3). POP increased transiently at 2.5 min post-injection, and/G was elevated at 1 and 2.5 min. Internal NaCN had no effect on fAB (Table 2), and injection of a saline bolus into the DA had no effect on any variable (Fig. 3).

In denervates (Fig. 3) the ventilatory responses to internal NaCN were abolished, with no significant changes in POP or fG. Cardiovascular variables showed a similar trend to those of sham-operated Amia-, the changes over time were not statistically significant. There was no air-breathing response to internal NaCN in the denervates. Injection of a saline bolus had no effect on any variable (Fig. 3). Fig. 4 shows representative traces of cardiovascular and ventilatory responses to NaCN in sham-operated Amia and denervates.

In sham-operated Amia, NA infusion (Fig. 5) significantly increased PDA, fH and POP. There was no significant effect on/G, and no stimulation Of fAB. Arterial blood pressure showed a transient increase at 2.5 min and fH was elevated at 2.5, 5 and 10min following infusion. Opercular pressure amplitude was significantly elevated at 2.5 min post-infusion, and remained elevated throughout the remainder of the measurement period. Adrenaline infusion had similar effects to NA infusion on cardiovascular variables (Fig. 5), significantly increasing PDA from 1 min post-infusion until the end of the measurement period, but only transiently stimulating fH, at 2.5 min. There was no statistically significant stimulation of POP or fG by A, as assessed by ANOVA. However, a paired t-test showed a significant increase in mean POP and/G at 2.5min post-infusion. There was no stimulation of fAB (Table 2). Control saline infusion had no effect on PDA, fH, POP, fG or fAB (Fig-5).

In denervated fish (Fig. 5), NA had effects on PDA and fH very similar to those seen in sham-operated Amia, but there was no statistically significant effect on POP or fG, when measured by ANOVA. Adrenaline also stimulated PDA and/H in a manner similar to the response of the sham-operated fish. There was no significant effect on POP following A infusion, but /G increased at 2.5, 5 and 10min postinfusion, when measured by ANOVA. Whilst NA and A showed no statistically significant effect on POP as measured by ANOVA, a comparison of denervate mean control POP and/G with mean POP and fG at 2.5 min post-infusion, by paired t-test, shows a significant increase in both ventilatory variables at 2.5 min (this was not the case for NaCN exposure in denervates). Saline control infusion had no effect on any measured variable (Fig. 5).

The present study demonstrates that in Amia, section of all branchial branches of cranial nerves IX and X and extirpation of the pseudobranch abolish air-breathing and cardiac reflex responses to hypoxia and NaCN. Gill ventilatory responses to NaCN are abolished. Gill ventilatory responses to hypoxia and catecholamines are significantly attenuated. These results indicate that air-breathing and cardiac reflex responses to hypoxia are mediated by O₂-sensitive chemoreceptors in the gills and pseudobranch, innervated by cranial nerves VII, IX and X, but that gill ventilatory reflex responses are mediated to some extent via an extrabranchial pathway.

Gill denervation did not change any resting cardiovascular or gill ventilatory variables significantly. This indicates that the motor pathways to the buccal and opercular pumps were not damaged by surgery, and also that Amia is different from L. oculatus, where section of the branchial branches of cranial nerve IX significantly attenuates resting POP and fG (Smatresk, 1991). A reduction or abolition of afferent information about water and blood O₂ levels and abolition of all efferent vascular and postural motor control of the gill arches probably combined to result in the measured reduction in $PaO2$ and $V˙O2$ seen in denervated Amia as compared with sham-operated fish. The denervates, however, had similar $CaO2$ and pHa values to sham-operated Amia, indicating that the former were not hypoxaemic and were not accumulating lactic acid as a result of anaerobic metabolism. Elevated circulating NA levels in denervated Amia indicate that they were more stressed than sham-operated Amia, as catecholamines are released into the circulation in response to stress in fish (Nakano and Tomlinson, 1967).

Sham-treated Amia did not breathe air as much during aquatic normoxia as noted by Johansen et al. (1970) and Randall et al. (1981). Denervated animals did not breathe air at all, except for one individual that did so on two occasions within 1 h. This is similar to L. oculatus, where complete branchial denervation severely reduced resting fAB (Smatresk, 1991). This indicates that air-breathing behaviour in both Amia and L. oculatus is mainly, but not entirely, dependent on afferent input from the gills.

Abolition of the fH response to hypoxia and external NaCN following gill denervation indicates that reflex bradycardia in Amia is controlled by O₂-sensitive chemoreceptors in the gills, as is the case in exclusively water-breathing teleosts, e.g. trout (Smith and Jones, 1978), cod (Fritsche and Nilsson, 1989) and channel catfish (Burleson and Smatresk, 1990a), where bradycardia is abolished by section of cranial nerves IX and X to the gills.

In sham-operated Amia, stimulation of a bradycardia by NaCN given in the buccal cavity, but not by NaCN infused into the dorsal aorta, supports previous studies that indicate that O₂-sensitive chemoreceptors mediating bradycardia are externally oriented (Saunders and Sutterlin, 1971; Burleson and Smatresk, 1990b). Cardiovascular responses by Amia to external and internal NaCN are similar to the responses of channel catfish (Burleson and Smatresk, 1990a,b) but contrast with the responses of the air-breathing fish Lepisosteus osseus (Smatresk, 1986; Smatresk et al. 1986). In L. osseus, heart rate is either not changed by external or internal NaCN (Smatresk, 1986) or internal NaCN produces a bradycardia (Smatresk et al. 1986).

In sham-operated Amia and denervates, stimulation of PDA and /H by catecholamines is probably a result of direct effects on the myocardium and peripheral vasculature (Wood and Shelton, 1980; Farrell et al. 1986).

Increases in POP, fG and fAB following hypoxic exposure in sham-operated Amia were similar to the responses of most air-breathing fish (Smatresk, 1988). Denervates did not air-breathe during hypoxia, providing evidence that insensitive chemoreceptors stimulating this behaviour are located in the gills or pseudobranch. This is similar to the African lungfish Protopterus aethiopicus (Lahiri et al. 1970), where partial gill denervation attenuated the air-breathing response to hypoxia, and to L. oculatus, where complete branchial denervation abolished air-breathing and gill ventilatory responses to hypoxia (Smatresk, 1991).

Sham-operated Amia required large doses of NaCN to elicit ventilatory responses, as compared with L. osseus (Smatresk, 1986). In L. osseus, external NaCN stimulated air-breathing but had no significant effect on gill ventilation (Smatresk, 1986), whereas in Amia there was an increase in gill ventilation but no significant effect on/AB. L. osseus have reduced gills compared with purely water-breathing fish (Smatresk and Cameron, 1982) but Amia do not (Daxboeck et al. 1981). These facts suggest that under natural conditions Amia relies on air-breathing for O₂ uptake less than does L. osseus, and demonstrate the variation in air-breathing strategies amongst fish.

Injections of NaCN into the dorsal aorta of sham-operated Amia stimulated a similar response to that seen in L. osseus (Smatresk, 1986), where internal NaCN significantly stimulated gill ventilation but not air-breathing. This is unlike the response of P. aethiopicus, where internally administered NaCN stimulated air-breathing (Lahiri et al. 1970). It is unknown whether internal and external NaCN injections stimulated the same or different groups of ventilatory reflex receptors in Amia.

It is interesting that hypoxia significantly stimulated air-breathing in sham-operated Amia but external and internal NaCN did not. It is possible, however, that two groups of O₂-sensitive receptors, oriented externally and internally, exist in Amia and that information from both receptor groups is integrated to produce a final air-breathing pattern. This is known to be the case in L. osseus, where internally oriented receptors set the level of hypoxic drive and externally oriented receptors set the balance of air-breathing vs gill ventilation (Smatresk et al. 1986). In Amia, during hypoxia, both groups may have been stimulated, leading to air-breathing. External or internal NaCN only stimulated one group of receptors, and thus only sometimes stimulated air-breathing. In incomplete denervates (i.e. animals with a branchial nerve and/or pseudobranch intact) external NaCN consistently stimulated air-breathing. Partial denervation may have affected the balance of information from both groups and led to more frequent air-breaths.

The complete abolition of all ventilatory responses to NaCN following branchial denervation and pseudobranch ablation clearly indicates that these reflexes are mediated by O₂-sensitive chemoreceptors situated in the gills and pseudobranch, innervated by cranial nerves VII, IX and X, and that following denervation Amia do not retain O₂-sensitive-chemoreceptor-mediated ventilatory reflexes. Thus, the POP and fG responses seen in hypoxic denervates are not mediated by an extrabranchial O₂-chemoreceptor (Bamford, 1974).

Infusion of catecholamines into sham-operated Amia produced ventilatory effects similar to those seen in intact fish, with increases in POP and fG but no change in fAB (McKenzie et al. 1991). Following denervation, the ventilatory response to catecholamines was attenuated. Denervation of aortic and carotid bodies in mammals causes a reduction in ventilatory sensitivity to catecholamine infusion (Dempsey et al. 1986), suggesting that, in Amia, stimulation of gill ventilation by catecholamines may be largely via stimulation of O₂-sensitive chemoreceptors in the gills. It is also possible, however, that the elevated endogenous NA levels in denervated animals reduced ventilatory sensitivity to exogenous catecholamine infusion. The existence, however, of a ventilatory response to catecholamine infusion in denervates clearly indicates that catechol-amines stimulate ventilation via an extrabranchial pathway, possibly centrally. Catecholamines can cross the blood-brain barrier in fish (Nekvasil and Olson, 1986) and micro-injection of NA into the fourth ventricle stimulates ‘fictive’ ventilation in curarised dogfish (Taylor and Randall, 1988). It has been suggested that circulating catecholamines mediate ventilatory responses to hypoxia in trout and dogfish (Randall and Taylor, 1989; Taylor and Randall, 1988; Aota et al. 1990). The absence of an air-breathing response to catecholamine infusion in intact and denervated Amia indicates that catecholamines stimulate ventilation at a site, be it central or peripheral, that is not involved in controlling air-breathing. This suggests that some sites controlling reflex gill ventilatory responses are spatially and/or pharmacologically separate from sites controlling air-breathing reflexes.

Both sham-operated Amia and denervates showed a significant increase in [NA] following hypoxic exposure, which supports the view that circulating catecholamines are involved in ventilatory control. The complete abolition of all ventilatory responses to hypoxia following branchial denervation in anaesthetised, spontaneously breathing I. punctatus (Burleson and Smatresk, 1990a) and L. oculatus (Smatresk, 1991) may have occurred because these species are less sensitive to circulating NA than is Amia, or because the use of an anaesthetised preparation compromised ventilatory responses to catecholamines.

In conclusion, the removal of afferent information from the gills reduces resting $V˙O2$ in. Amia possess two distinct groups of branchial O₂-sensitive chemoreceptors. Reflex bradycardia is controlled by an externally oriented O₂-sensitive chemoreceptor group, whereas ventilatory reflexes are mediated by both internal and external O₂-sensitive receptors, or by a group sensitive to both internal and external conditions. Reflex air-breathing responses to hypoxia are mediated exclusively by O₂-sensitive chemoreceptors in the gills, and Amia possess no extrabranchial O₂-sensitive-chemoreceptor-mediated ventilatory reflexes. The evidence indicates that in denervated animals gill ventilatory responses to hypoxia may be mediated centrally by circulating catecholamines.

The authors would like to thank Mr S. Munger and Dr C. Wood for supplying the Amia. This research was supported by a Natural Sciences and Engineering Research Council of Canada grant to D.J.R., and a MacLean-Fraser Memorial Summer Fellowship to M.L.B.