Ventilatory Mechanisms of the Amphibian, Xenopus Laevis; The Role of the Buccal Force Pump

Adult anuran amphibians utilize skin, lungs, and perhaps buccal membranes for respiratory exchange (Krogh, 1904). The lungs are usually the most important for oxygen transfer (Emilio & Shelton, 1974; Rahn & Howell, 1976) although they are ventilated only intermittently. The buccal cavity is used as a pump to fill the lungs, and in ranids and bufonids two ventilation patterns, renewing gas in the lungs or in the buccal cavity alone, are known to occur (Martin, 1878; Willem, 1919; Cherian, 1956). The mechanisms underlying these patterns have been described (dejongh & Gans, 1969; West & Jones, 1975). It is generally considered that ventilation is ineffective in preventing some admixture of fresh air and used lung gases in the buccal cavity before lung inflation (Foxon, 1964), though it has been suggested that separate buccal flow streams may reduce the mixing (Gans, dejongh & Farber, 1969).

Xenopus laevis is an entirely aquatic, aglossal anuran which differs from ranids and bufonids in that it does not actively ventilate the buccal cavity between lung ventilations (dejongh, 1972). In spite of the substantial differences in habitat and ventilation patterns, the systemic blood oxygen concentrations in Xenopus during ventilation are comparable to those in Rana ridibunda and Rana esculenta (Emilio & Shelton 1972; Emilio, 1974; Jones, 1972). The present study analyses the mechanism of lung ventilation in Xenopus and compares it with those of the more terrestrial Anura.

Experiments were carried out on 31 adult female Xenopus laevis during the spring and summer months, at 20 °C (± 1 °C). Animals were obtained from a commercial supplier or were bred in the laboratory, and ranged in weight from 58 to 110 g (mean 80·5 g). They were maintained in large freshwater tanks at a temperature of 21 °C (± 2 °C), and were fed on a diet of fresh liver except during the experimental period.

Anaesthesia was induced by immersion in solutions of tricaine methane sulphonate (MS-222, Sandoz) at concentrations of 500 mg/1 or 700 mg/1. If the animal showed signs of recovery during the operative procedure, MS-222 at a concentration of 1 g/1 was dripped on to the animal to prolong anaesthesia. Recovery in fresh water took 1–2 h. Recording commenced when the animal began to dive and surface unassisted and recording continued for at least 3–4 h. Some chronically implanted cannulae remained intact for up to 3 days, thus allowing measurements after a longer recovery period.

To measure pressure, the buccal cavity and one lung were cannulated with polypropylene tubing (Portex PP. 200). For the buccal cavity cannulation, a 19G hypodermic needle was inserted into the buccal cavity from the dorso-lateral side, just behind the skull, and a 40 cm length of tubing, cut obliquely at the distal end to form a sharp tip, was passed out of the hole formed by the needle. A short collar of vinyl tubing (heat flared at one end) on the buccal cavity side of the cannula, and a similar collar on the exterior, clamped the cannula into position and sealed the hole (Jones, 1970). The cannula was stitched to the skin of the flank and to the roof of the buccal cavity, with the tip of the cannula about 5 mm posterior to the internal nares.

For the lung cannulation, the posterior tip of the lung was first located by inflating the lungs through a small tube inserted into the glottis. The cannula was slightly heat flared at one end and a series of small holes was made along the first 1·5 cm to prevent it being blocked by small quantities of blood and body fluid. The cannula was then bent into a right angle, about 6 cm from its proximal end, and about half this length was inserted through the exposed, cut tip of the lung. The perforation of the lung was sewn around the cannula which was then stitched to the body musculature and to the skin as the incision was closed.

Pressures in the lung and buccal cavity were measured with the air-filled cannulae connected to air-filled transducers (Sanborn differential, Model 270; Devices differential, Type UP2; or Elcomatic, Model Em75o) whose outputs were amplified with Sanborn carrier preamplifiers. Transient pressure tests (Gabe, 1972) showed damping to be around 24% of critical and the natural frequency of the system was 40–86 Hz. The most rapid transients in the phenomena recorded during these experiments had a frequency of about 10 Hz. Calibration of the pressure system was carried out with manometer U-tubes filled with Brodie’s solution and connected via an air-filled tube to the pressure transducers. Small amounts of body fluid sometimes collected in both lung and buccal cannulae during pressure measurements, so injections of less than 0·1 ml of air were used to clear the cannulae.

Ventilatory gas flow through the nares was measured with a pneumotachograph. This apparatus consisted of a cylindrical breathing chamber (2·5 cm in diameter and 2·5 cm in height) to which the surfacing animal was directed by an inverted 25 cm diameter plastic funnel or a plastic mesh guide. A plastic tube, 1·2 cm in diameter and containing the pneumotach screen, led out of the breathing chamber. Pressure differences on either side of the screen were detected by a Sanborn differential pressure transducer (Model 270) and were proportional to ventilatory flow. Both inhalant and exhalant flow rates down to 50 ml/min could be detected. Water-saturated air was continually introduced into the chamber through a small hole in the top at a constant rate of 300 ml/min monitored by a flow meter. Calibration of the system was carried out by varying this flow.

Volume changes of the lung were estimated with an impedance pneumograph (Geddes & Baker, 1968; West & Jones, 1975). Two 1 cm × 1 cm silver electrodes were sewn on to the body musculature under the skin on the dorso-lateral aspects of the animal, directly above the main body of the lungs, to form one arm of a bridge circuit. This type of system has been criticized because of its intrinsic non-linearity (Pacela, 1966) and in the present application a larger signal was observed for volume changes in the deflated lung than for equivalent changes in the inflated lung. Impedance also varied from one surfacing and breathing sequence to the next. However, an approximate calibration of the system was carried out in restrained animals by injecting known volumes of air into the lung through the lung cannula.

Patterns of muscle activity were investigated using electromyograms (EMGs) recorded between pairs of electrodes. Thin (01 mm) 25 cm lengths of enamelled copper wire, bared for 1 mm at the tip were formed into hook-shaped electrodes around the points of 25G hypodermic needles for insertion into larger muscles (Basmajian & Stecko, 1962) but with smaller muscles, the hooked electrodes were pushed into the muscle itself without the aid of the needle. The electrodes inserted into superficial muscles were individually stitched to the skin of the animal. Electrodes in muscles of the hyoid apparatus were stitched to the roof of the buccal cavity and brought out through the mouth and also individually stitched to the skin. Signals were amplified with Tektronix Type 122 low-level preamplifiers, displayed on an oscilloscope screen, and recorded. Immediately after the last EMG recording, the animal was heavily anaesthetized in the recording chamber by immersion in a 10 g/1 solution of MS-222 and the position of the electrodes was checked by dissection. Several post-mortem checks of electrode position were also made by passing current through the electrodes followed by staining for copper deposit with potassium ferricyanide.

A Bolex H16 Reflex 16 mm cine camera was used to film the opening and closing of the external nares in one animal (106.5 g), whilst simultaneously recording lung and buccal pressures. Film speed was 24 frames per second. Film and pressure records were synchronized by including a flashing (2·5 Hz) neon bulb in the field of the camera. The neon was powered by pulses from a Grass S4HR stimulator which were simultaneously recorded with the pressures.

In all experiments, a maximum of four variables could be simultaneously recorded on a Racal Store 4 instrumentation tape recorder and a Devices pen recorder. At the end of an experiment, selected parts of the taped records were replayed to a Medeled For-4 recorder, a S.E. 3006 UV recorder or a 4 channel Sanborn pen recorder.

All experiments were performed on unrestrained, free-swimming Xenopus in a 1·5l perspex box, except that impedance pneumograph techniques were applied to both unrestrained and restrained animals. Animals were free to surface, breathe, and dive at will as all cannulae and electrodes were long enough to allow three complete turns in the tank.

The anaesthetic and surgical procedures used in these experiments had no detectable effect on the ventilation patterns. Similar results were obtained 2 h after the surgery as after 2 days. Under slight anaesthesia, the breathing was atypical and loss of motor control of limb muscles prevented the animals from surfacing. Removing an animal from the water only affected the absolute levels of lung pressure, though not the relationships between buccal pressure and lung pressure. Restraining an unanaesthetized animal on a board, even if the animal was placed in water at an angle comparable to the normal breathing position, caused reduced lung pressures and exaggerated buccal pressures. The adverse effects of anaesthesia, restraint, and confinement were avoided in the experiments reported below, although there was, inevitably, some general disturbance imposed upon the animal by the rigours of the experiment.

Pressure in the buccal cavity before ventilation was always above atmospheric, usually 50 – 100 Pa above (133 Pa equals about 1 mmHg), and pressure in the lungs before ventilation ranged from 350 to 800 Pa above atmospheric (Fig. 1). The range of pre-ventilation lung pressures was variable both within animals and between animals, and depended on size of animal, depth of water and position of cannula. Slight regular oscillations of about 6 Pa, in both buccal cavity and lungs, were recorded during apnoea, and these persisted when the nares were submerged. They coincided with heart beat and had no function in ventilation. Larger, erratic single oscillations of buccal pressure also occurred between ventilation series when the animal was at the surface, but these also continued when the nares were under water, and none was associated with flow through the nares. They may have been a response to the buccal cannula, a feeding response, or a redistribution of air within the respiratory system.

Buccal pressure changes during the ventilation cycle were more complex than those in the lung. For simplicity, a single ventilation cycle, consisting of a single exhalation and a single inhalation, will be described, and this cycle is most readily divided into three phases (I, II, and III, Fig. 1) which will later be shown to correspond to exhalation of air, inhalation of air into the buccal cavity, and lung filling respectively.

During Phase I (which lasted 500 – 700 ms) pressure in the buccal cavity initially increased (Figs. 2a, 5) although in some cases this was preceded by a transient (less than 200 ms) decrease (Figs. 1, 5). Buccal pressure continued to rise gradually to peak between about 100 and 200 Pa, a level that was below lung pressure. The end of Phase I was denoted by a drop in buccal pressure to atmospheric. In a very few instances, buccal pressure remained at atmospheric for the duration of Phase I (Fig. 7).

Buccal pressure stayed below atmospheric for the duration of Phase II, for about 300 – 600 ms. Generally, pressure decreased rapidly to less than 100 Pa below atmospheric and either remained at this level or decreased slowly at a constant rate. The end of Phase II was marked by an abrupt rise in buccal pressure to atmospheric.

In Phase III, buccal cavity pressure increased rapidly until it equalled or just exceeded lung pressure. Following this, both lung and buccal pressures continued to rise, but at a slower rate, peaking above 400 Pa. These were the highest pressures recorded in the buccal cavity during a lung ventilation. Phase III ended with a fast drop in buccal pressure to a level comparable to that found before the start of Phase I, and this level was maintained until the start of the next lung ventilation.

Lung pressure always decreased during Phase I, and the onset of this decrease was coincident with the increase in buccal cavity pressure (Fig. 1). Lung pressure reached its lowest value at the end of Phase I, about 200 Pa less than the pre-ventilation level, as buccal pressure fell towards atmospheric. In Phase II, lung pressure slowly rose at a constant rate, to increase about 50 Pa by the end of the Phase. At the start of Phase III, lung pressure dropped momentarily to equilibrate with buccal pressure, and then increased simultaneously with buccal pressure. Lung pressure remained slightly less than buccal pressure during this portion of the cycle. However, at the end of Phase III as buccal pressure dropped, lung pressure remained at the elevated level, sometimes decreasing slightly, and always levelling off to a constant value.

The changes in lung volume as indicated by the impedance pneumograph (although not quantified) were correlated with lung pressure changes in all three phases (Fig. 1). Thus there was an apparent volume decrease in Phase I, a slight apparent volume increase in Phase II, and an additional volume increase in Phase III.

In both operated and non-operated Xenopus, the simple style of ventilation consisting of a single exhalation and inhalation was less common than a more complex cycle consisting of an exhalation followed by two inhalations (Fig. 5). The exhalation and the first inhalation were followed by a second buccal pressure drop below atmospheric while lung pressure remained constant or increased slightly, as in the previous Phase II. This was followed by a repeat of Phase III when buccal pressure rapidly increased and lung pressure rose to a new level that equalled or was greater than the original pre-breath pressure (Fig. 5).

The nares of Xenopus open for short periods of time during lung ventilation. At all other times the nares remain tightly closed, even during prolonged periods at the surface and between breaths. In a pithed or in an anaesthetized animal, the nares are always closed, and they will remain closed if the mouth is opened.

The cine film of Xenopus breathing at the water surface (Fig. 2) showed that the nares always opened during Phase I, but the timing in relation to lung and buccal pressure changes varied from breath to breath. These variations were of three general types. In the first type, the opening of the nares coincided with the initial fall in buccal cavity pressure towards atmospheric, when lung pressure remained at the preventilation level (Fig. 1). In the second variation, narial opening occurred as lung pressure began to decrease and buccal pressure began to rise (Fig. 3). In the last variation, narial opening was delayed until after buccal pressure began to increase while lung pressure stayed constant (Fig. 5).

The opening of the nares during Phase I was not an off/on event. The narial valves did not immediately open to their maximum during this initial phase of ventilation, but instead they dilated progressively (Fig. 3 and compare frame 3 and frame 5 of Fig. 2) to open maximally by the time buccal pressure had peaked. The nares stayed fully open for the rest of Phase I, remained open for Phase II and closed abruptly at the beginning of Phase III as buccal pressure increased. If a second inhalation followed, the nares totally dilated at the commencement of the buccal pressure decrease and closed suddenly at the start of the rapid buccal pressure increase (Fig. 3).

EMG recordings from the m. dilator laryngis (Fig. 4) indicated the timing of glottal opening. This muscle always fired during the lung pressure drop in Phase I, and actively ceased only when lung pressure began the slow increase of Phase II. Spikes were recorded from the m. dilator laryngis again when lung pressure dropped to buccal pressure levels in Phase III and this activity continued until lung pressure reached its high level at the end of Phase III.

A deflexion of the pneumotachograph flow profile in Phase I indicated the start of exhalant flow (Fig. 5), coinciding with the opening ot the nares. The commencement of exhalant flow was confirmed by measuring temperature differences at the nares with a bead thermistor (West & Jones, 1975). The initial component of exhalation was gradual and corresponded with the initial decrease in buccal pressure (Fig. 5 second series of ventilations). Exhalant flow continued as lung pressure decreased in Phase I and reached maximum rate at the time of lowest lung pressure, immediately before closure of the glottis. The rate of exhalation decreased after closure of the glottis, but how was often maintained for 50 – 100 ms, as long as buccal pressure remained above atmospheric.

The first inhalation commenced immediately after exhalation, as demonstrated by the rapid change in direction of the flow profile. The start of inhalation was concurrent with the drop of buccal cavity pressure to levels below atmospheric pressure at the beginning of Phase II. Inhalant flow peaked and slowly began to decrease as buccal pressure increased towards ambient. No exhalant or inhalant flow was recorded during Phase III.

This biphasic flow profile, consisting of a single exhalation followed by a single inhalation, was the simplest form observed. However, this simple form was most commonly followed by a second inhalation (Fig. 5). The start of the second inhalation was delayed until buccal pressure fell below atmospheric, and the flow profile was similar to that of the preceding inhalation. Again, inhalant flow stopped when buccal pressure began to rise sharply towards atmospheric. Other more complex flow profiles, consisting of a number of exhalations and inhalations following in series, occurred rarely.

Three superficial muscles, the m. submentalis, m. submaxillaris, and m. subhyoideus, form a continuous layer on the ventral surface of the head, spanning the lower jaw from one mandible to the other, and when active tend to elevate the buccal floor and thus decrease the volume of the buccal cavity. Electrical activity in all three muscles was correlated with phases of the ventilation cycle (Figs. 6 and 7). Activity in the m. submentalis, the m. submaxillaris, and sometimes in the m. subhyoideus began either just before, or coincident with, the rise in buccal pressure at the beginning of Phase I, and continued until the glottis closed. In three out of twelve animals, EMG activity continued until buccal pressure reached atmospheric at the end of Phase I, but in all other cases, activity was not observed as buccal pressure fell to atmospheric. The amplitude of the EMGs of the m. submentalis during the initial stages of Phase I was not as great as that activity observed towards the end of Phase I. In two out of five animals, activity in the m. subhyoideus continued after activity in the m. submaxillaris and m. submentalis had ceased, through to Phase II (Fig. 7). Electrical activity in the m. submaxillaris and m. submentalis was never recorded during Phase II. The second major period of activity in all muscles of the floor of the mouth occurred during the rapid buccal pressure rise of Phase III although the start of major activity in the m. subhyoideus was sometimes slightly delayed. All activity ceased when buccal pressure had peaked.

The m. geniohyoideus externus and internus are both paired muscles originating from the region of the mandibular symphysis and inserting on to the postero-lateral processes of the hyoid cartilages, as in Rana. The geniohyoids decrease the volume of the buccal cavity and pull the hyoid anteriorly. In the majority of animals, activity was recorded in the m. geniohyoideus externus during the buccal pressure rises associated with exhalant flow in Phase I, but not from the m. geniohyoideus internus during this Phase. The m. geniohyoideus internus and externus both showed activity during the rapid buccal pressure increase of Phase III, although activity did not begin until after the start of the pressure rise (Fig. 8). All activity ended at the peak buccal pressure. Activity in the m. petrohyoideus, a muscle which connects the hyobranchial skeleton to the skull, was only correlated with the buccal pressure rise after the glottis had re-opened at the beginning of Phase III (Fig. 9).

The m. sternohyoideus arises from the sternum and combines with the m. rectus abdominis and m. transversus in Xenopus to attach extensively to the hyoid cartilage (Grobbelaar, 1924), unlike Rana where the m. sternohyoideus is a discretely separate muscle (Ecker, 1889). These muscles act to increase the volume of the buccal cavity by pulling the hyoid posteriorly and ventrally. The m. sternohyoideus was the only muscle in which activity was consistently observed during Phase II of the ventilatory cycle (Fig. 9). Activity began about 50 ms before the buccal pressure had reached atmospheric and continued until the start of the buccal pressure increase at the end of Phase II. In some animals, the amplitude of firing of the m. sternohyoideus and the decrease in buccal pressure below atmospheric were related -higher amplitude peaks were recorded when the buccal pressure fell to lower levels (Fig. 10). Electrodes placed in the m. sternohyoideus at its origin on the sternum or in the m. sternohyoideus/m. rectus abdominis/m. transversus complex ventral to the anterior section of the hyoid cartilage showed the same activity.

Activity in the m. temporalis and m. pterygoideus was not consistently related to any portion of the ventilation sequence, and recordings from the m. hyoglossus could not be related to any discrete portion of the ventilation cycle. Xenopus lacks a m. omohyoideus and any muscular tissue which would correspond to the m. genioglossus of Rana.

The lung ventilation cycle in Xenopus differs in several very important respects from that in more terrestrial amphibians.

Exhalation occurs for the entire duration of Phase I (Fig. 11). The nares and glottis open and the lungs empty, as indicated by the concurrent decreases in lung pressure and lung volume. Pressure in the buccal cavity usually increases at first, but a small pressure decrease is observed if the nares open before the glottis. Increase in buccal pressure is due to a combination of the rapid outflow of gas from the lungs following glottal opening, and to activity in the expiratory musculature gradually increasing from the time of narial opening. During this phase gas flow through the partially open nares increases as the pressure gradient from buccal cavity to the exterior goes up. Later in the exhalation phase the nares open fully, buccal pressure begins to fall, and the glottis closes (at Ic in Fig. 11) halting the decrease in lung pressure and volume. The expiratory musculature is still active during this period and gas continues to flow out through the nares, though at decreasing velocity.

The cyclical excitation of respiratory muscles, and control of the nares and glottis by central nervous mechanisms (Ito & Watanabe, 1962; West & Jones, 1975) clearly follows similar sequences in successive ventilations. However, slight variations in timing give rise to many differences in the detail of flow and pressure profiles as observed in Xenopus during exhalation. Clearly the control mechanisms are not producing precisely stereotyped outputs from breath to breath.

There are several major differences between the exhalation phase of Xenopus and that of other Anura. Before ventilation, buccal pressure never falls below atmospheric in Xenopus as it does in some ranids, and exhalation starts the ventilation cycle without any preceding buccal movement. The initial changes, with pressures increasing in the buccal cavity and decreasing in the lungs, are the same as those observed in other frogs. However, the increase in buccal pressure in Rana is associated with an increase in buccal cavity volume mediated by the m. sternohyoideus (Scholten, 1942; Cherian, 1956; Das & Srivastava, 1956; dejongh & Gans, 1969). In Xenopus, muscular contraction decreases the volume of the buccal cavity. The muscles involved (m. submentalis, m. submaxillaris, and m. subhyoideus) have not been recorded as active during exhalation in any other amphibian. In comparison to the ranids, buccal pressure peaked and decreased during exhalation in Xenopus, and lung pressure rarely equilibrated with buccal pressure. The active decrease in buccal cavity volume forces gas out of the nares during exhalation and, as lung pressure is always greater than buccal pressure in this phase, flow must also occur from lungs to buccal cavity. The continued exhalant flow shown by the pneumotach after glottis closure is further evidence for the active decrease in volume of the buccal cavity.

The nares of Xenopus open only during exhalation and inhalation and, unlike those of the ranids, do not open between lung ventilations (West & Jones, 1975; dejongh & Gans, 1969), suggesting an active control of narial opening for Xenopus. The results obtained in these experiments tend to confirm this hypothesis -in anaesthetized and dead animals, the nares are not open. However, muscular tissue, around the cartilages and bones of the nasal passages, corresponding to the m. lateralis narium and m. dilator narium of Rana (Ecker, 1889), could not be found in Xenopus and no EMG activity could be recorded from this region. EMG activity in the m. submentalis of the lower jaw occurs when the nares are either open or closed which shows that this muscle does not play a role in closing or opening the nares as has been suggested (Grobbelaar, 1924) although it may help to maintain closure. It is also difficult to understand how the action of the m. submentalis pushing up on the fused premaxillae could open the nares (Noble, 1931) because manual pressure on the ventral side of the premaxillae tends to close the nares. A possible site for muscular tissue involved in opening the nares is near the articulation of the upper jaw where stimulation (50 Hz) produced narial dilation. Further investigations of the anatomy and control of the nares are required.

At least one inhalation (Phase II, Fig. 11) always immediately follows exhalation in Xenopus. Inhaled gas passes only into the buccal cavity (nares fully dilated and glottis closed) through the action of the m. sternohyoideus and the complex of m. rectus abdominis/m. transversus/m. sternohyoideus. Together, these muscles pull on the hyoid apparatus to lower it and increase the volume of the buccal cavity.

During inhalation into the buccal cavity, the glottis is closed and total lung pressure increases slightly. This increase could result from a damped elastic recoil of the lung, hydrostatic pressure compressing the lung, or contractions of lung muscles (Keith, 1905; Brett and Shelton, in preparation): all forces which would act to decrease lung volume. The change indicated by the impedance pneumograph could reflect a change in the shape of the lung between the recording electrodes rather than an increase in total lung volume. After deflation of the lungs, a slower responding component in the walls of the lung and in the surrounding viscera continues to adjust volume and shape of the lungs (the non-elastic substructures, Taglietti & Casella, 1968) which perhaps also widens the lungs at the point of volume measurement by the impedance pneumograph.

The major difference between inhalation of gas into the buccal cavity in Xenopus and the same act in other anurans is its appearance in mid-ventilation cycle, after expiration of gases from the lungs. Inhalation of fresh air in other anurans is mainly accomplished by the oscillatory cycles of buccopharyngeal ventilation which occur between lung ventilations and before expiration (dejongh & Gans, 1969; West & Jones, 1975; MacIntyre & Toews, 1976). In some amphibians, the start of a ventilatory cycle is marked by a pre-ventilation decrease in buccal cavity pressure which draws a larger volume of fresh air into the buccal cavity (dejongh & Gans, 1969) but again these inhalations always precede exhalation.

Inhalation in Xenopus and in other anurans is powered by the same muscles, namely those depressor muscles which act to increase the volume of the buccal cavity, especially the m. sternohyoideus. Muscles of the floor of the mouth and the m. geniohyoideus externus and internus are inactive during inspiration and their relaxation allows the buccal cavity to expand passively. Dejongh & Gans (1969) have stressed the importance of gravity during inhalation as a force acting on the relaxed muscles of the floor of the mouth to increase buccal cavity volume but no experiments have been attempted which would confirm or quantify this hypothesis.

The last phase of the lung ventilation cycle is a pumping of air from the buccal cavity into the lungs, occurring with the nares shut and the glottis open fully (Phase III, Fig. 11). All the muscles of the floor of the mouth, together with the petrohyoids and the geniohyoids, contribute to this action and it is typical of the buccal forcepump in other anuran amphibians (dejongh & Gans, 1969; West & Jones, 1975). As the glottis opens, lung pressure falls to equilibrate with the rising buccal pressure; lung pressure and buccal pressure then increase together, with buccal pressure always slightly greater than lung pressure. This gradient between the two chambers means air flows from buccal cavity to lung and lung volume increases. The pressure differences between lung and buccal cavity are not great, as has been similarly demonstrated for Rana pipiens (West & Jones, 1975), suggesting that the glottis is widely open and offers very little resistance to flow.

The end of the lung ventilation cycle is marked by a passive decrease in buccal cavity pressure as the compressor muscles of the buccal cavity relax, and by the closure of the glottis to maintain re-established lung pressure. A second inhalation and lung filling sequence may follow and act to increase the end lung pressure and volume. The factors promoting a double inhalation sequence, as versus a single inhalation, are likely to be complex but we did observe that a second lung filling occurred if lung pressure was less than pre-ventilation levels at the end of the first inhalation.

In Rana and Bufo, some expired gas must remain in the buccal cavity after expiration and be returned to the lungs during the force filling which follows expiration. In Xenopus, however, the mixing of expired and inspired gas is largely prevented by an expiration which precedes inspiration. At the start of the lung ventilation cycle, gas is expired from both lung and buccal cavity. After the lung has been closed, used gas is still expired from the buccal cavity as the contraction of the m. submentalis, m. submaxillaris, and m. subhyoideus is maintained. Fresh air is then inhaled into the buccal cavity and the only mixing which can occur is with the small volume of used gases left in the buccal cavity and with used gases not expired from the lungs. In a purely aquatic animal which can only acquire fresh air when at the surface, this method of ventilation ensures a more complete renewal of lung gas.

In Rana catesbeiana, it has been suggested that streams of expired and inspired gases are separated in the buccal cavity during lung ventilation to prevent mixing of used gases and fresh air (Gans et al. 1969), but this jet-streaming of gases is not required in Xenopus because expiration occurs before inspiration. The morphology of the buccal cavity of Xenopus also argues against jet-streaming, as the glottis is positioned horizontally in the buccal cavity, and the floor of the buccal cavity is flat not pocketed, as in R. catesbeiana.

Inspiration of air into the buccal cavity before expiration and lung filling, as occurs in the ranids, has been considered to be primary (Gans, 1970) and reflective of the lungfish style of ventilation (McMahon, 1969). The method of lung ventilation in Xenopus lends no support to this concept since Xenopus is phylogenetically more primitive than the bufonids and ranids. We suggest that the Xenopus air flow sequence, of expiration before inspiration in the buccal cavity, is best thought of as an adaptation to a totally aquatic existence in an animal that does not vocalise. Similar sequences are thought to occur in some urodeles (Guimond & Hutchison, 1973, 1974). They are clearly effective in reducing the time that these aquatic animals spend at the water surface.

This research was supported by grants from the National Research Council of Canada and the University of East Anglia.