The Relative Efficiency of Gaseous Exchange Across the Lungs and Gills of an African Lungfish Protopterus Aethiopicus

The habitat of Protopterus aethiopicus is similar to that of Lepidosiren, and, though it has long been assumed that the lungs were the principal respiratory organ, until very recently little experimental work has been carried out to verify this assumption. Prior to the completion of this manuscript, however, several workers have reported studies which verify this assumption. Jesse et al. (1968) suggest that both lungs and gills are important in respiration in juvenile Protopterus (species not certain) but report no result from adult animals. Johansen & Lenfant (1968), however, demonstrate that the lung is the principal organ in oxygen uptake in Protopterus aethiopicus. In this study experiments to elucidate and quantify the actual efficiency of lung and gill in gaseous exchange will be described and discussed.

Recordings were made of aerial and aquatic respiratory frequencies, both in animals immersed in well-aerated water with access to air and in animals which were confined either under water or in moist air. The animals were confined in a Perspex observation tank which could be perfused with either warmed aerated water or warmed humidified air. Ambient water was warmed by passage through a glass warming coil immersed in a thermostatically controlled water bath. The temperature in these experiments was maintained at 24+ 1°C.

Ventilation rates were recorded in three ways: (a) observed by the investigator and transferred immediately to an event recorder; (b) by the electromyograms of the respiratory muscles picked up incidentally by ECG probes; (c) by the pressures recorded by means of cannulae implanted in buccal and/or opercular cavities (McMahon, 1969).

In all the experiments animals were allowed to acclimatize to the experimental chamber for at least 4 hr. and usually over 20 hr. A record of the activity of the animals was taken at all times, but in fact acclimatized animals usually remained at rest on the floor of the chamber, moving only to take air. All recordings discussed here were taken from animals at rest.

Rates of gas exchange were measured by respirometry. The simultaneous measurement of aerial and aquatic gas exchange poses problems in the lungfish, as streams of respiratory air and water must be kept separate to avoid interdiffusion of gases.

Methods where the animal is partitioned using rubber membranes, as used by Van Dam (1938) and Berg & Steen (1965), are not easily adapted for use with a free-swimming animal breathing air or water at will, and may also impair the respiratory mechanism (Piiper & Schumann, 1967). In these experiments the rates were measured using a specially designed respirometer in which air and water streams flowing past the animal were kept separate except at the moment of breathing. This apparatus is illustrated in Fig. 1.

The respirometer consisted of a Perspex tube flanged at both ends which could be mounted diagonally in a rigid frame. The animal was persuaded to swim into the tube and the ends were sealed by Perspex discs equipped with ‘O’ ring seals. The discs held a number of taps which controlled the flow of respiratory media. Once the animal was installed in the respirometer a flow of water was allowed to pass through it. This water was collected under paraffin in 1 l. samples during the experimental periods and was preserved for subsequent analysis.

Air was introduced into the respirometer only during lung ventilation. (The animal always lifted the head prior to lung ventilation, and on this ‘signal’ air was allowed into the respirometer). Air was drawn in by lowering the water level in an accessory container outside the respirometer but connected to it (Fig. 1). As the water level fell in this outer container, so it fell in the respirometer, drawing in air from the perfusing air stream and allowing the animal to ventilate the lung. Immediately after the lung ventilation the original water level in the outside container was restored and the expired air was then forced out of the respirometer. This air passed into the perfusing air stream and was collected in a large bottle to await analysis. All gas and water interphases, with the exception of that in the respirometer itself, were protected with paraffin. The water samples were analysed for oxygen content using the Winkler method and for carbon dioxide content by the nomograph method of Dye (1951). The gas samples were analysed for both oxygen and carbon dioxide content using a Scholander I ml. Gas Analyser (Scholander, 1947). All experiments were carried out in a constant temperature room (24 ± 1° C), thus ensuring that gas and water samples were at the same temperature.

It was not possible to acclimatize the animals in the respirometer for long periods because continual manipulation by the operator was necessary to maintain the animals in the chamber. In fact, they were to some degree pre-acclimatized to the tubes (which were often inhabited by them when left in the ‘home’ tanks). This suggests that the animals were not seriously disturbed by their close confinement during the experimental periods. Once installed and settled into the chamber the animals very rarely showed any activity not associated with air breathing.

Samples of pulmonary gas, and both inhalant and exhalant branchial water were obtained from chronically implanted cannulae. The pulmonary cannula was inserted into the anterior median sac of the lung, and the other cannulae into the buccal and opercular cavities as in McMahon (1969). Samples of 0·5−1·0 ml. of pulmonary gas were withdrawn into a syringe, the dead space of which was filled with a solution which was non-absorbent for gases. A volume of gas equivalent to the cannula volume was withdrawn and rejected before each sample. Samples were taken at intervals through a number of air-breathing cycles and stored immersed in cooled, non-absorbent solution for periods of 2−4 hr. to await analysis. Tests indicated that no significant change occurred in samples stored in this way. In some of the experiments the of the samples was taken on withdrawal and this provided a check on the subsequent analysis.

Water samples were withdrawn from the buccal and opercular cavities to determine the extent of the gas exchange ocurring over the gill surface. During slow branchial irrigation (less than 1/min.) very small samples (0·2−0·3 ml.) were taken into a syringe attached to the cannula. The samples were taken very slowly to avoid drawing water over the gills artificially, and were withdrawn at varying intervals after the last branchial respiratory movement. During faster branchial irrigation larger samples (0·5−1·5 c.c.) were withdrawn over a period covering several branchial respiratory movements (0·5−1·5 min.). In all cases a volume equivalent to the cannula volume was taken and discarded before a sample was taken. The samples were analysed for oxygen and carbon dioxide tension using an ‘Eschweiler’ gas analyser. The electrodes were calibrated with gas mixtures of known oxygen and carbon dioxide tension several times during each experiment. The oxygen electrode system was very stable and gave repeatable results at 1 % level of accuracy. The carbon dioxide electrode, however, had a slow response time at this temperature and was liable to drift from the calibration settings. These factors may have introduced a level of inaccuracy (< 5 %) into the results.

In the first series of experiments the frequencies of branchial irrigation and lung ventilation were monitored in animals before, during and after confinement in either air or water. Even in the resting animals the rates were very irregular and for this reason the rates were usually recorded for periods of at least 30 min. and the results expressed as an average over this period.

The graphs in Fig. 2 show the results of typical experiments. Fig. 2 A shows the effect of protracted confinement under water. Branchial and aerial respiratory rates were both low before confinement. A considerable increase in activity was seen immediately following confinement as the animals struggled to reach the surface, and reliable estimates of the resting branchial rate could not be made. After 30−60 min., however, the animals’ activity decreased and long periods were spent at rest at the bottom of the chamber. During the rest of the submergence the animals were quiet but initially showed a marked increase in the rate of branchial irrigatory movements. This increase continued for the first 3−5 hr. of confinement and the branchial respiratory rate reached levels of 25−30 beats/min. This high level was maintained throughout the rest of the submergence.

Both pressure recordings from the buccal and opercular cavities and direct observation of the confined animals indicated that the amplitude of the branchial movements increased together with the frequency of respiration. The increase in amplitude of the recorded pressure waveforms was not constant but decreased almost to the resting level at times. As no change in the observed amplitude of the breathing movements was seen at this time the decrease in the pressures recorded was apparently due either to a change of gill resistance or a failure of the mouth or opercular flaps to close at the correct phase of the cycle. Adult animals have been kept submerged for periods of up to 24 hr. With the exception of one animal (in poor general health), which died after only 7 hr., all the animals survived the submergence without harm. It was noticeable, however, that after submergence for over 15−20 hr. the animals’ respiratory movements became very forceful and irregular, and they began to show signs of loss of equilibrium. At the onset of the latter symptoms the animals were allowed access to air.

Immediately following the first lung ventilation the rate of branchial irrigatory movements fell rapidly, usually reaching the pre-submergence level in 30−60 min. The rate of lung ventilation following the submergence period was considerably greater than had been seen before confinement, and this rate decreased more slowly to the resting level at a rate dependent on the severity of the previous confinement.

All the experiments quoted above were carried out on adult (over 200 g.) animals. Table 1 shows the results of preliminary experiments on smaller animals. Larval Protopterus (3−4 weeks old) with functional external gills were not affected by confinement under well-aerated water even for periods in excess of 14 days. No branchial respiratory movements were seen in these animals before or during the submergence.

In juvenile animals the branchial hyperirrigation response develops gradually, with increasing size. As the animals’ size increases, however, the length of time the animal can survive without lung ventilation decreases.

In all experiments where the animals were confined in moist air the branchial respiratory rate was always low even in the pre-confinement period. Branchial respiratory movements were usually only recorded when the animals rose to the surface to breathe air, or when they were otherwise active. One branchial movement has been shown to be an integral part of the air-breathing cycle (McMahon, 1969), but after 1 hr. exposure to air no branchial irrigatory movements are seen, even this obligatory flushing stage of the air-breathing cycle having been suppressed (Fig. 3). The animals were usually restless when first exposed to air and no measurements of the breathing rates were made in the first hour of each exposure. The animals’ activity decreased after this time and recordings showed that the lung ventilation rate had been markedly increased (average of ten experiments, 6 ×) despite the abundance of air. This increased rate was maintained throughout the air exposure, which was limited to 5 hr. or less, as it is well known that Protopterus can survive long periods of air exposure as long as it is kept moist. On termination of the air exposure a dramatic but transitory increase in the branchial respiratory rate was observed as soon as the animal was able to submerge the mouth (Fig. 2B).

Samples of pulmonary gas were removed from the lungs at intervals throughout a number of natural and artificially prolonged submergences. Immediately following a lung ventilation the oxygen concentration in the lung gas was high, generally over 15% (> 110 mm. Hg ⁠), and the carbon dioxide concentration was low, rarely more than 1 ·5−2·5% (11−22 mm. Hg ⁠) During the length of an average submergence (20-25 min., at rest in the home aquarium) the oxygen concentration fell rapidly (Fig. 4) to 4−5 % (30 mm. Hg ⁠). If the submergence period was prolonged by denying the animal access to the surface the oxygen concentration continued to decrease, but much more slowly, reaching a level of 0·3−0·5% (3 mm. Hg after 150 min.

The carbon dioxide concentration rose rapidly in the first 5−10 min. after lung ventilation, reaching a level of 4−5 % (25−30 mm. Hg ⁠). Little or no further increase occurred, however, and pulmonary carbon dioxide concentration exceeded 5 % in only one animal after a 150 min. submergence (Fig. 4). The original level of both gases was always restored at the next lung ventilation after a natural submergence, but a second ventilation was often needed to restore the levels fully when the submergence had been prolonged for more than one hour.

Samples were removed from the buccal and opercular cavities by means of implanted cannulae. The water in the buccal cavity varied little from that in the ambient water, but changes in the concentration of both oxygen and carbon dioxide occurred in passage over the gills and were detected in the exhalant branchial water ⁠. These changes are plotted as a function of respiratory rate in Fig. 5. It can be seen that both oxygen and carbon dioxide are exchanged at this site, and that the rate of this exchange varies with the rate of branchial respiration. Unfortunately, no estimate of ventilation volume could be made.

The actual consumption of oxygen and production of carbon dioxide were measured in the specially designed respirometer illustrated in Fig. 1. Because of the difference in size of the experimental animals (150−600 g.), and perhaps because of differences in physiological state, some variation of the individual rates of gas exchange was seen. Mean figures for oxygen and carbon dioxide exchange have been calculated from the data obtained from nine experiments on four different animals. These figures are presented in histogram form in Fig. 6.

The first column shows the results obtained from animals free to breathe air or water. The average figure for total oxygen consumption was 62·5 c.c. oxygen/kg./hr. (range of variation 27·8−86·5 ml./kg.^-1/hr.^-1). The average figure for carbon dioxide production was 47·4 c.c. carbon dioxide/kg./hr. (range of variation = 19·9−56·3 ml./ kg.^-1/hr.^-1 CO₂). The calculated total RQ for the averaged results is 0·755, a reasonable figure for a carnivorous animal. The results show that the exchange ratio was very different at the two respiratory rates. In every experiment the oxygen consumption over the pulmonary surface was much greater than over the gills. Average aerial oxygen consumption was 91·7% of the total (range 86·5−94·0%). It is evident that under these conditions the lung is the principal site for oxygen exchange. Examination of the figures, however, showed that only 32·5 % of the carbon dioxide is excreted via the aerial route, the remainder passing via the aquatic route.

In five experiments the animals were denied access to the surface for periods of about 60 min. during the course of the experiment. Branchial respiratory rate was increased by up to four times and increased gaseous exchange was seen across the gill surface (Fig. 6, second column). Carbon dioxide production via the gills was increased so that up to 100% (average 85 %) of the pre-confinement production now passed by this route. The aquatic oxygen consumption, however, though considerably increased, could provide only 17% (11·8−30·0%) of the animals’ total pre-confinement oxygen requirements.

Rates of gas exchange were also measured for 1 hr. after the end of the confinement. A very marked increase in the total oxygen consumption was seen (49·8−72·6 % above the pre-confinement level.) The rate of lung ventilation also increased and all the additional oxygen was consumed by this route. The increased lung ventilation also affected the carbon dioxide production ratio, as on the average 57 % of the total carbon dioxide produced was now eliminated via the lungs, while the amount passing via the gills was correspondingly reduced. Before evaluating the results of these experiments it must be mentioned that the method of estimation of carbon dioxide concentration in the water samples was accurate to 5 % only. This level of inaccuracy, while high, was not sufficient to influence the conclusions drawn from these results.

The experimental evidence presented here demonstrates clearly that gaseous exchange at the gill surface accounts for very little of the total oxygen uptake in the adult Protopterus. This is in agreement with the work of Lenfant & Johansen (1968). The rate of branchial respiration is very low in animals at rest and the percentage utilization is very low when compared with the figures published for other fishes in Table 2. Percentage utilization seen in Neoceratodus compares with that seen in the other fishes but that of Protopterus is only 50 % efficient at very low irrigation rates. If the animal increases the rate of branchial irrigation, as seen in response to protracted submersion, the percentage utilization falls to a very low level. Oxygen consumption thus falls by over 80% when adult Protopterus are confined underwater for protracted periods. The oxygen consumption also falls in Neoceratodus similarly confined but by only 20−25% (Grigg, 1965 c). Direct evidence as to the inefficiency of the gills in oxygen uptake is given by the respirometry experiments, where increases of branchial respiratory rate of up to four times in confined animals could provide only 10-30% of the animals’ oxygen requirements. The animals were forced into oxygen debt when prevented from breathing air, and this oxygen debt was paid off by the marked hyperventilation seen once access to air was possible. The increase in aquatic oxygen uptake during long submergence is, however, of value in allowing even adult animals to remain submerged for relatively long periods.

Carbon dioxide excretion occurs over both lung and gill surfaces, with perhaps the major part passing aquatically in the resting animal. The percentage of carbon dioxide passing via either route can be increased by hyperventilation, but, whereas moderately increased branchial respiration was able to remove all the carbon dioxide during aquatic confinement, a very marked increase in aerial respiration was needed to remove the carbon dioxide accumulated during exposure to air.

If the data obtained from direct measurement of the respiratory media are plotted in the form of O₂/CO₂ diagrams, as first used by Willmer (1934) and more recently by Rahn & Fenn (1955), confirmation of the results expressed above can be seen. Figures 7A, B show O₂/CO₂ plots for pulmonary air and expired water respectively. In Fig. 7B a regression line has been calculated from the plotted data to show the relationship between aquatic oxygen and carbon dioxide exchange. Theoretical gas exchange lines where R = unity are also drawn for aerial and aquatic routes. The slope of the plotted regression line (R = 4·9) is considerably greater than R = 1 for water, and this figure is in good agreement with the exchange ratio of 5·4 calculated from the respirometry data. A high aquatic gas exchange ratio indicates that much more carbon dioxide than oxygen is being exchanged across the gill surface. This imbalance could be explained by a prior oxygenation of the blood reducing the amount of possible oxygen uptake, or could be due to a thickening of the gill epithelium—such as is seen in Lepidosiren (Fullarton, 1931)—which would reduce the possible exchange for the less soluble oxygen while having much less effect on carbon dioxide. As very little has yet been published on the degree of separation occurring in the partially divided heart of Protopterus, the extent of the former is difficult to estimate. The lengthening of the diffusion path would be of adaptive benefit in an animal where the blood passing through the gills may have a higher than that of the ambient water. Under these conditions a short diffusion path would result in the loss of oxygen to the ambient water and a consequent loss of efficiency of the lung.

When the pulmonary gas data are plotted on an O₂/CO₂ diagram all the points are seen to lie beneath the R = 1 line for air, indicating that more oxygen uptake than carbon dioxide elimination takes place in the lung. This conclusion is in agreement with the exchange ratio deduced from respirometry data (R = 0·27). Above carbon dioxide tensions of 20−25 mm. Hg the slope of the relationship approximates to zero, indicating that above this level no further carbon dioxide is eliminated into the lung, though oxygen is still being removed. The time course for gas exchange in the lung (Fig. 4) shows that this does, in fact, occur. This indicates that the level of carbon dioxide in the lung is very low immediately following lung ventilation, but that the tension rises very quickly as carbon dioxide diffuses into the lung from the pulmonary blood stream. Equilibration quickly occurs between the blood and gas and they remain in equilibrium while any further carbon dioxide produced by the animal is eliminated through the gills. This explanation presupposes a very much higher tension of carbon dioxide in the circulating blood than has been shown to occur in the blood of the fishes studied to date (3·3 mm. Hg in dogfish venous blood (Piiper & Schumann, 1967); 5·7 in trout ventral aortic blood (Stevens & Randall, 1967); 7·7 mm. Hg in the venous blood of Neoceratodus (Lenfant et al. 1966). A high level of circulating carbon dioxide might, however, be expected in the blood of an obligatory airbreathing form. In fact, Lenfant & Johansen (1968) show levels of up to 30 mm. Hg in the dorsal arterial blood of this animal.

The occurrence of a high level of carbon dioxide in the blood of Protopterus is of considerable interest. Rahn (1966) considers that the lungs of the emergent tetrapods, though efficient in oxygen exchange, were much less efficient carbon dioxide exchange mechanisms than were the gills of the ancestral aquatic forms. This, in fact, is the case in Protopterus (B. R. McMahon, in preparation). Rahn postulates that a cutaneous carbon dioxide exchange route was needed in the first terrestrial animals to complement the lung exchange and thus prevent dangerously increased carbon dioxide levels in the blood. He considers that the next step would have been the evolution of a tolerance of high carbon dioxide which would have rendered the lung tidal ventilation mechanism sufficient for all gas exchange, and would allow the cutaneous exchanger with its additional problems of water loss to be abandoned.

The presence of high circulating carbon dioxide levels in the aquatic Protopterus, however, indicates that the early air-breathing fishes, including the Rhipidistia, may have already evolved a degree of tolerance of high carbon dioxide levels, partly due to the development of an aerial respiratory mechanism but mostly in response to the presence of high carbon dioxide levels frequently found in the environment. In this case the ancestral tetrapods may well have been pre-adapted to the terrestrial habitat in this respect and the evolution of an intermediate cutaneous carbon dioxide exchanger may not have been essential.

No measurement has been made of the respiratory exchange occurring across the skin of Protopterus. Though the importance of cutaneous carbon dioxide exchange has been demonstrated in Lepidosiren (Cunningham, 1934), the skin of the adult Protopterus is neither particularly thin nor particularly vascular and would not appear to be an efficient exchange surface. The proven efficiency of the fish gill in gaseous exchange indicates that the skin of Protopterus is unlikely to be important in this role in the submerged animal, though it may be important during aestivation when the gills are collapsed in air (Lenfant & Johansen, 1968).

It has been demonstrated that adult Protopterus aethiopicus obtain 90% of their oxygen consumption from the aerial exchange occurring in the lungs, even when the animals are immersed in well-aerated water. If the animals are prevented from ventilating the lungs for long periods, branchial hyperirrigation is seen. This response cannot provide the whole of the animals’ oxygen requirement, but the additional oxygen consumption, though small, is of importance in prolonging the possible submergence time. The results obtained by Jesse et al. (1968) for either P. aethiopicus or P. dolloi (not specified) would suggest that the gills were of greater importance than is indicated here. These workers used juvenile specimens, however, in which the degree of dependence on aerial respiration is less well developed (B. R. McMahon, in preparation).

Carbon dioxide excretion can occur via either aquatic or aerial routes, and although the major part normally passes over the gills the fraction passing aerially can be increased by hyperventilation. As the animal can utilize aerial oxygen and is tolerant of high external carbon dioxide concentrations (B. R. McMahon in preparation), it is extremely well suited to its periodically hypoxic and hypercarbic environment. Protopterus is thus more similar to Lepidosiren than to Neoceratodus, both in habitat and in the degree of dependence on aerial respiration. If we imagine the rhipidistian fishes as having been similarly adapted to their rather similar environment, then they were eminently pre-adapted to colonize the terrestrial habitat.

1. The efficiency of gas exchange over the lung and gill surfaces of Protopterus has been investigated.

2. Animals confined in water or in air showed an increased respiratory frequency in the remaining medium, indicating that both routes were important in the total gas exchange.

3. Direct measurement of the oxygen and carbon dioxide tensions of pulmonary air and inspired and expired branchial water showed gas exchange ratios (R) of 0·2 for the lung and 5·0 for the gills approximately, demonstrating that more oxygen was consumed via the lungs and more carbon dioxide excreted via the gills.

4. Oxygen consumption and carbon dioxide production were measured directly in a respirometer in which respiratory air and water streams could be kept separate except during lung ventilation. At least 90% of the animals’ oxygen consumption occurred in the lung, while 60% of the carbon dioxide excreted passed via the aquatic route.

5. The results are discussed with reference to the animals’ adaptation to its environment and with reference to the evolution of the terrestrial vertebrates.

I am greatly indebted to Professor G. M. Hughes, Department of Zoology, University of Bristol, in whose laboratory and under whose supervision this work was carried out, and to the Science Research Council who provided financial support.