Ventilation and partitioning of oxygen uptake in the frog Rana pipiens: effects of hypoxia and activity

Many adult anurans breathe both water and air, using both skin and lungs. Apnoea, submergence, air exposure, temperature and ambient respiratory gas levels may all affect the relative importance of the lungs and skin in gas exchange.

Regulation of lung ventilation in amphibians has not been extensively studied, in part because of analytical difficulties caused by their intermittent ventilation (Shelton & Boutilier, 1982; Glass & Wood, 1983). Generally, however, lung ventilation in amphibians is regulated in response to both oxygen and carbon dioxide. Either low oxygen or high carbon dioxide in inhaled gas increases the rate of lung ventilation (Smyth, 1939; Toews, 1971; Boutilier & Toews, 1977, 1981; McDonald, Boutilier & Toews, 1980; West & Burggren, 1982; Feder, 1983), but different responses to O₂ and CO₂ occur in different amphibians and the control mechanisms remain unclear. Lung ventilation frequency has not been measured simultaneously with oxygen uptake in unrestrained frogs, and few measurements of ventilation volume are available (West & Jones, 1975).

Little attention has been given to skin ventilation of floating or submerged amphibians, although Burggren & Feder (1986) found that experimental ventilation of the skin increased cutaneous oxygen uptake from water by almost 50 % in adult Rana catesbeiana, probably by disrupting a boundary layer of stagnant water next to the skin. Existing models of cutaneous gas exchange do not incorporate skin ventilation, although boundary layer effects are recognized to be a potential complicating factor of cutaneous gas exchange (Piiper, Gatz & Crawford, 1976; Piiper & Scheid, 1977).

The purposes of these experiments are (1) to establish the effects of activity (and therefore variation in oxygen uptake) and hypoxia on ventilation of both the lungs and skin, and (2) to interpret effects of activity and hypoxia on oxygen uptake partitioning in terms of ventilation and other processes of the lungs and skin.

Adult Rana pipiens pipiens (northern grass frogs) were obtained from a commercial supplier in two groups. One group (37–47 g) was used in the investigation of skin ventilation and activity. The other group (30–45 g) was used for measurement of lung ventilation and effects of hypoxia. All animals were maintained in the laboratory at 22–26°C for at least 1 week before experimentation. They were housed in aquaria with 5–7 cm of water and a platform above water level, and fed flies, crickets, mealworms or baby voles, as available.

All measurements were made at 25°C using flow-through respirometers in which the animals floated in water and had access either to air or to an air/nitrogen mixture. Oxygen uptake from gas was considered to be equivalent to pulmonary ⁠, and from water was considered to be cutaneous ⁠.

The effect of skin ventilation was investigated by comparing pulmonary and cutaneous in experimentally stirred respirometers to values in unstirred respirometers. Activity of the animal, visible through a window in the respirometers, was noted for each measurement of ⁠. Water entering the respirometer was air saturated. Frogs were acclimated to the respirometer overnight. A total of 17–21 measurements of pulmonary and cutaneous ⁠, separated by at least 30 min, was made over a 1-to 2-day period on each frog. After obtaining resting measurements, inactive animals were prodded intermittently with a flexible plastic-covered probe to ensure that some of the measurements were taken during activity. Prodding, when used, was maintained for 30-90 min. The of each frog was measured in separate experiments in both stirred and unstirred respirometers; half of the frogs were first measured in the stirred respirometer, the other half in the unstirred respirometer. The frogs were given 2–3 days to recover between experiments.

Lung ventilation and were measured in unstirred respirometers during combined aerial and aquatic normoxia and three levels of hypoxia; 97, 75 and 52mmHg⁠. All activity was spontaneous -the frogs were never prodded.

The frogs were allowed to recover overnight in the respirometer after implantation of electrodes for measurement of lung ventilation (see below). Three measurements of ⁠, separated by at least 30 min, were taken at each oxygen level. Physiograph records of ventilation were taken during a 10-min period bracketing the measurement. The order of presentation of oxygen levels was from normoxia to progressively greater hypoxia. The animals were returned to normoxia for at least 1 h between hypoxic exposures. Further measurements of were made during these recovery periods, so that more measurements were made during normoxia than at any one level of hypoxia.

Flow-through respirometers containing 1·81 of water and 50–75 ml of gas were used. Oxygen uptake was calculated from the difference between incurrent and excurrent water and gas, the capacitances of water and gas for oxygen, and the flow rates of water and gas through the respirometers. The flow of water was 50–100 ml min^-1, resulting in an incurrent – excurrent difference of 3–10 mmHg. Gas flow was 5–15 ml min^-1, also with an incurrent – excurrent difference of 3–10 mmHg. The electrodes were calibrated whenever the measurement of incurrent departed significantly from 152·0 mmHg in normoxia.

Lung ventilation was measured by detecting changes in impedance between electrodes implanted subcutaneously on either side of the chest, a placement which gives an impedance signal that is linearly related to lung volume (West & Jones, 1975; Brett & Shelton, 1979). The impedance electrodes were thin copper disks 2–3 mm in diameter, soldered to insulated 0·1 mm diameter copper leads. In a tracheally cannulated frog, impedance was highly correlated with the volume of air injected into the lungs (r = 0·99, N = 14, P< 0·001).

The impedance leads were led out of the respirometer through a small tube sealed with wax and connected to a Narco Mark IV Physiograph through Narco impedance converters. Lung ventilation rate (fL) was measured by counting all rapid changes in lung volume over a 10-min period (except those caused by movement of the animal, which were usually easily distinguished from lung ventilation). Ventilation volume was measured by summing all vertical displacements of the trace (reflecting lung volume changes) over the same 10-min period using a Numonics Corporation electronic planimeter with a precision of 0·25 mm.

To analyse the effect of activity on ⁠, data from the skin ventilation experiment were separated into measurements taken while the frogs were resting or measurements taken while frogs were highly active, according to the visual observations of activity taken simultaneously with the measurements. Three measurements each of resting and active were averaged to give a single measurement of resting and active per animal under each condition. Where more than three observations were available (the usual case), the three measurements to be used were chosen using a random number generator. Each individual animal thus yielded measurements of active and resting pulmonary, cutaneous and total ⁠, in both stirred and unstirred respirometers.

Unless otherwise noted, individual measurements were treated as independent observations. The number of measurements available depended on the analysis being made. Pulmonary and cutaneous were always measured together. Movement artifacts, damaged leads, and electrodes that had shifted after implantation caused obvious difficulties in interpreting the traces so that not all recordings of ventilation could be analysed for fL and ventilation volume. Thus, not all measurements of had an associated measurement of fL. Measurement of ventilation volume was attempted only for normoxia because of a change in ventilatory pattern in hypoxia.

Because lung ventilation traces were not calibrated, ventilation is in arbitrary units. To compare results from different animals the ventilation volumes were normalized as follows. The sum of the vertical displacements of the recorded trace (ventilation volume) over a 10-min period was divided by fL for that time period, giving a ratio of volume/ventilation movement (stroke volume), in arbitrary units. All stroke volumes (3–9 measurements per animal) for each individual were averaged, and the measured ventilation volume was then divided by the average stroke volume. This procedure sets the average stroke volume for each animal to 1·0, and therefore allows comparison between different animals.

The product–moment correlation coefficient, r, was used to assess the significance of correlations. Least-squares linear regression slopes are reported; slopes were compared using the analysis of covariance technique given in Sokal & Rohlf (1981). Mean values are given ± standard error of the mean. Differences between means were analysed with pair t-tests. A minimum significance of P<0·05 was used for all significance tests. Statistical significance is noted on figures with asterisks: * = P<0·05, ** = P<0·01, *** = P<0·001.

In inactive frogs floating in well-aerated, unstirred water at 25 °C, common natural conditions for these frogs, cutaneous provides 23 % of total ⁠. Partitioning of between the skin and lungs is highly variable, however, and is affected by skin ventilation, increases of due to physical activity, and environmental hypoxia.

Cutaneous (i.e. from water) increases when the skin is ventilated, whether ventilation is caused by stirring the respirometer or by physical activity of the animal itself. Physical activity has the additional consequence of increasing total ⁠. Stirring of the respirometer does not affect total ⁠, either in resting or active animals. When the skin is not ventilated by stirring the respirometer, cutaneous increases with increasing total (Fig. 1A), while in a stirred respirometer cutaneous is high and independent of total (Fig. IB). This difference is also apparent in individual animals (Table 1); cutaneous increases with increasing total in unstirred respirometers but there is generally no correlation between cutaneous and total in stirred respirometers.

Much of the variation in total is attributable to variable physical activity. This is obvious from comparing ‘active’ with ‘resting’ animals in either stirred or unstirred respirometers (Table 2; Fig. 1). Active frogs have a total about double their resting (P< 0·001 in stirred respirometer, P<0·01 in unstirred respirometer). [Note: this difference should not be construed as the aerobic scope, because ‘active’ is not maximum ⁠, both because the animals were not prodded to maximal activity and because the procedure used in choosing ‘active’ did not necessarily choose the highest from each animal -see Fig. 1.]

Activity thus increases both total and cutaneous in the unstirred respirometer, resulting in the significant correlation of cutaneous with total in Fig. 1A. When there is no skin ventilation from either activity or stirring (i.e. resting animals in unstirred respirometers), cutaneous is reduced by 34% compared to resting animals in stirred respirometers (P <0·05). As a result, cutaneous is 35 % of the total in the stirred respirometer but only 23 % of the total in the unstirred respirometer (comparing resting animals). The contribution of cutaneous decreases to 19 % of total in both stirred and unstirred respirometers during activity.

Individual variation in cutaneous is apparent in the slopes of the regression of cutaneous on total (m = 0·62−0·476), and from an analysis of variance (ANOVA), comparing measurements of cutaneous of individual animals in stirred respirometers (P<0·001). Different types of activity (e.g. bursts of struggling or steady, slow movement) will provide different amounts of skin ventilation per unit ⁠. Thus, the relationship of cutaneous to total is expected to be variable depending on the activity patterns of individual frogs.

Lung ventilation frequency (fL) is directly proportional to pulmonary (Fig. 2). Although there is considerable short-term variability in fL (Fig. 3A), and probably in because of bursts of activity, fL and pulmonary are highly correlated when short-term variations are averaged out over longer time periods.

Lung ventilation volume is directly proportional to fL, at least in normoxia (Fig. 4). Stroke volume does not change with fL; the slope of the regression line for pooled data is 0·996, not significantly different from 1·0. Slopes of regression lines calculated for individual frogs are also not significantly different from 10.

In normoxia, ventilation is very irregular (Fig. 3A). Most ventilatory movements consist of an exhalation immediately followed by an inhalation of similar magnitude. Some ventilatory movements, however, are all expiratory or all inspiratory. The period between breaths, during which lung volume remains relatively fixed, is variable.

Hypoxia both decreases total and increases fL, resulting in a shift of partitioning towards the lung. Lung ventilation frequency relative to (slopes of the regression lines in Fig. 2) increases as ambient decreases. fL slightly more than doubles for a given when the ambient is reduced from 150 mmHg to 52 mmHg.

The pattern of lung ventilation is also different during hypoxia, characterized by regular large oscillations in lung volume involving several ventilatory movements (Fig. 3B). The lungs are deflated in one or a few exhalations, then reinflated with a series of several inhalations. Exhalation and inhalation are less frequently coupled into a single ventilatory movement compared to the normoxic pattern.

This increased lung ventilation results in a shift in partitioning towards the lung. Total is reduced in hypoxia (Fig. 5), mostly due to a reduction in cutaneous ⁠, which decreases 63 % compared to normoxia at an ambient of 52 mmHg, while pulmonary decreases only 20 % over the same range of ambient ⁠. The proportion of total taken through the lungs therefore increases from 63 % in normoxia to 77 % at 52 mmHg (P< 0·01). These figures are for spontaneously active frogs, with all values from each frog, regardless of activity level, averaged together for each level. This ‘average’ activity level results in an in normoxia midway between ‘active’ and ‘resting’ values measured in the skin ventilation experiment, and slightly higher than ‘standard’ or ‘resting’ values measured by Seymour (1973) and Carey (1979) for R. pipiens at 25°C.

Hypoxia and increased resulting from increased activity both increase the proportion of taken up through the lungs (Fig. 6). Thus, pulmonary oxygen uptake is at its most important in active animals and during hypoxia, whereas cutaneous oxygen uptake is at its most important in resting animals in well-aerated water.

The predominant adjustment by Rana pipiens to increased O₂ demand in these experiments is to increase pulmonary ventilation and ⁠. Lung ventilation is proportional to in mammals (Dejours, 1981 ; Puper, 1982) and reptiles (Jackson, 1978; Glass & Wood, 1983). Although this is thought to be true in amphibians as well (Withers & Hillman, 1983; Gottlieb & Jackson, 1976), there are few measurements in amphibians which actually show this relationship. If ventilation is measured over a longer period, as in this study, then analytical problems associated with shortterm variability in ventilation pattern (Shelton & Boutilier, 1982) are decreased, and it becomes evident that lung ventilation rate (fL) is indeed directly proportional to in R-pipiens (Fig. 2).

Pulmonary ventilation volume has not been measured in relation to in any amphibian, although alveolar ventilation has been calculated for R. pipiens (Withers & Hillman, 1983). Tidal volume in R. pipiens (the stroke volume of the buccal pump), although highly variable from breath to breath, does not change with fL, so ventilation volume increases in direct proportion with fL. Thus, the air convection requirement (the volume of air ventilated for a given remains constant, as it does in mammals and reptiles during exercise (Jackson, 1978; Dejours, 1981).

Cutaneous in R. pipiens decreases when the skin is not ventilated (Fig. 1 ; Table 2), as found by Burggren & Feder (1986) in Rana catesbeiana. Unlike the experiments of Burggren & Feder (1986), in which the frogs were immobilized, the frogs in the present study were free to move. Thus the skin could be ventilated either by voluntary locomotor movements or, during periods of inactivity, by experimental stirring. Movement within the respirometer (swimming, diving and struggling) simultaneously increases both the total and skin ventilation. Thus cutaneous (dependent on skin ventilation) increases with increasing total (dependent on the amount of movement). The dependence of cutaneous on total must be largely due to skin ventilation, rather than some other change associated with activity, because in stirred respirometers cutaneous is high and increases little with activity.

Although skin ventilation is a major factor affecting cutaneous of frogs in water, it is not the only one. Two out of seven frogs tested in both stirred and unstirred respirometers showed significant correlations of cutaneous with total even in a stirred respirometer. Also, cutaneous in resting animals is significantly lower than in active animals in both stirred and unstirred respirometers (although there is much less difference between them in the stirred respirometer than in the unstirred respirometer). An increase in the gradient across the skin at higher values, as suggested by Gottlieb & Jackson (1976), could account for at least some of the correlation. An increase in cutaneous diffusing capacity, as demonstrated in Rana catesbeiana during submergence in hypoxic water (A. Pinder, in preparation) or with changes in skin ventilation (Burggren & Feder, 1986), could also be a factor..

Frogs exposed to combined aerial and aquatic hypoxia increase lung ventilation, increase the proportion of from the lungs, and decrease total ⁠. Increased fL during hypoxia in amphibians has been noted previously in adult Bufo marinus (Boutilier & Toews, 1977) and tadpoles of Rana berlandieri and R. catesbeiana (Feder, 1983; West & Burggren, 1982). Withers & Hillman (1983) also found an increase in ventilation in hypoxia when they calculated ventilation volume in adult R. pipiens exposed to aerial hypoxia. The fL of R. pipiens in the present study more than doubled at a of 52 mmHg compared to the rate in normoxia (Fig. 2). An increase in ventilation during hypoxia should decrease the difference between alveolar and ambient ⁠, thereby partially offsetting a decrease in ambient ⁠.

Although fL increased during hypoxia, the most dramatic and immediately noticeable effect is a change in ventilation from an extremely irregular pattern in normoxia to a more regular hypoxic pattern consisting of large oscillations in lung volume over the course of several buccal movements (compare Fig. 3A with Fig. 3B). This pattern of ventilation during hypoxia in R. pipiens is similar to the lung inflation cycles described by Boutilier & Toews (1977) in Bufo marinus during hypoxia. The effect of this change in ventilation pattern, which consists largely of an increase in depth and a decrease in frequency of ventilation, may be to reduce functional dead space by reducing mixing of inhaled and exhaled gas within the buccal cavity (West & Jones, 1975; de Jongh & Gans, 1969).

Lung hyperventilation affects partitioning during hypoxia, although ventilation is not the only reason for changes in partitioning. Most of the decrease in total seen in hypoxia is in cutaneous ⁠. Pulmonary changed little (Fig. 5), probably because of lung hyperventilation; as a result, the pulmonary contribution to total increases. An increase in the proportion of taken through the actively ventilated gas exchange organ has also been observed in studies of partitioning between skin and gills of anuran larvae during aquatic hypoxia (Feder, 1983; Wassersug & Feder, 1983). In these studies, partitioning between skin and gills (the lungs were found to be of lesser importance) shifted towards the gills during hypoxia, largely because gill ventilation was increased to compensate for a decreased ambient ⁠.

In addition to differences in ventilation mechanisms, lungs and skin differ in that cutaneous gas exchange is primarily diffusion limited, whereas pulmonary gas exchange is much less so (Piiper & Scheid, 1977; Burggren & Moalli, 1984; Feder & Burggren, 1985). One result of the diffusion limitation of the skin is that the decrease in ambient during hypoxia is probably reflected in a decrease in the gradient across the skin, and therefore a decreased cutaneous ⁠. The gradient across the skin may also be affected by a Bohr shift of haemoglobin-oxygen affinity due to lung hyperventilation and decreased blood ⁠.

Hypoxia and activity interact in their effects on cutaneous and the partitioning of (Fig. 6). Cutaneous is reduced in hypoxia for the reasons stated above. In addition, cutaneous is affected by skin ventilation at all ambient values, as it is in normoxia. As a result of these two effects, the maximum cutaneous (at high activity levels and therefore high ⁠) decreases with hypoxia, and the slope of the regression line of cutaneous on total is concomitantly reduced.

Partitioning of ⁠, is thus extremely variable, and the lungs can potentially account for anything between zero (in the case of a submerged frog) and 100 % (in a frog in hypoxic water with access to air) of the total ⁠. A complete understanding of gas exchange partitioning must include knowledge of internal conditions such as haemoglobin-oxygen affinity, partial pressures of blood gases, acid-base status and distribution of blood flow to lungs, skin and body. These, in turn, will be related to ventilation of lungs and skin, ambient and total oxygen consumption of the animal.

This work was supported by an NSERC (Canada) predoctoral fellowship (AP) and NSF grant PCM-8309404 (WB). We thank Dr Don Jackson, Gordon Wyse, John Roberts and Willy Bemis for their comments and criticism of earlier drafts.