The hypothesis that the lungless salamander Desmognathus fuscus responds actively to hypoxia was tested. Patterns of buccal movements [apneic period duration, the duration (min h−1) of buccal pumping and buccal pumping frequency], heart rate and metabolic rate (rates of oxygen uptake and carbon dioxide output) were determined during a control period (21 % oxygen), a hypoxic period (2, 5, 6.5, 8 or 10 % oxygen) and a recovery period (21 % oxygen). Hypoxic salamanders maintained their rate of oxygen uptake at control levels until a critical oxygen level between 10 and 8 % oxygen was reached. The rate of carbon dioxide output remained constant across all oxygen levels, except for a significant increase during exposure to 5 % oxygen. The buccal activity of lungless salamanders was responsive to environmental hypoxia, with a significant stimulation during exposure to 6.5 % and 5 % oxygen. Buccal pumping frequency was inhibited at 2 % oxygen. Heart rate was stimulated at all hypoxic levels except 2 % O2. During recovery, metabolic rate and heart rate returned to control levels within 20 min after all hypoxic exposures. The durations of apneic periods increased significantly compared with the hypoxic value during recovery from exposure to 10 %, 6.5 % and 5 % oxygen. Overall, the animals responded actively to hypoxia by increasing the duration of buccal activity as oxygen levels decreased. The ability of these changes to facilitate oxygen uptake is not known. However, the response of the dusky salamander to low levels of oxygen is analogous to the hypoxic ventilatory response observed in lunged vertebrates.

Adult salamanders of the family Plethodontidae are lungless; the only surfaces available for the exchange of respiratory gases are the skin and the inside surfaces of the mouth and pharynx. Although lungs are absent, the process of gas exchange is aided in the mouth and pharynx by the movement of the skin on the floor of the mouth; i.e. buccopharyngeal (or buccal) pumping (Whitford and Hutchison, 1965). The decreased total respiratory surface area available to Plethodontids compared with lunged salamanders and their inability to maintain gas exchange through increased lung ventilation, while adequate during normoxia, could impact negatively on their ability to tolerate hypoxia.

A considerable amount of work has been performed on the properties of cutaneous gas exchange and diffusion limitation using the lungless salamander model (Gatz et al., 1975; Piiper et al., 1976; Feder et al., 1988). Although the contribution to gas exchange by the buccal capillaries is probably small, the percentage of total capillaries located in the buccal cavity is greater in plethodontids (approximately 10 %) than in lunged salamanders (1–5 %) (Czopek, 1961), and changes in buccal pumping patterns may be important during times of environmental stress. In lunged and lungless amphibians, buccal activity (measured as the total amount of buccal pumping per minute) has been observed to increase with increasing temperature (Desmognathus quadramaculatus and Taricha granulosa; Whitford and Hutchison, 1965), during hypoxia (Xenopus laevis; Feder and Wassersug, 1984) and during hypercapnia (Cryptobranchus alleganiensis; Boutilier and Toews, 1981). In all cases, rates of buccal activity were measured only as the overall rate of buccal pumping per minute. However, buccal pumping typically occurs in a burst-like pattern, with apneic periods interspersed between periods of activity.

We tested the hypothesis that the pumping of the buccal pouch might itself be responsive to hypoxia and constitute a ‘ventilatory’ response to hypoxia. Since lungless salamanders are derived from lunged ancestors (Ruben and Boucot, 1989), they could retain this response even if its potential to aid in gas exchange is very small. Our study looks at the pattern of these active and apneic periods, rather than the overall average rate of buccal pumping, to examine changes in pumping patterns in response to lowered oxygen levels. We also tested the effects of hypoxia on heart rate and metabolic rate, other potential compensatory responses to hypoxia.

Animals

Dusky salamanders (Desmognathus fuscus (Rafinesque)) (mean mass 1.92±0.10 g, mean ± S.E.M., N=50) were collected at West Branch Reservoir (Portage County, Ohio, USA), between April and October in 1998 and 1999. The salamanders were housed in the laboratory at 10 °C on a 12 h:12 h light:dark cycle for a minimum of 3 days and a maximum of 14 days before being studied. They were provided with a diet of wingless fruit flies ad libitum. A different group of salamanders was exposed to each level of hypoxia, rather than re-exposing the same animals to all oxygen levels, which facilitated pairwise comparisons between normoxia and hypoxia. This method allowed each animal to act as its own control and also avoided the possible effects of multiple hypoxic exposures. The numbers of salamanders (N) exposed to each level of hypoxia (% oxygen) were as follows: 2 %, N=7; 5 %, N=13; 6.5 %, N=10; 8 %, N=10; 10 %, N=10.

Experimental protocol

For hypoxia testing, individual animals were placed in a 15 ml glass chamber inside a temperature-controlled cabinet (model PTC-1, Sable Systems, Henderson, Nevada, USA) at 15 °C. Each animal was exposed to a control period of 21 % oxygen overnight (at least 16 h), and all experiments were started between 09:00 and 10:00 h the following morning. First, control values were recorded at 21 % oxygen, followed by exposure to a single hypoxic gas (2, 5, 6.5, 8 or 10 % oxygen) for 90 min, and then a 90 min recovery period at 21 % oxygen. Gas compositions were controlled using a gas-mixing flowmeter (model GF-3/MP, Cameron Instruments, Port Aransas, Texas, USA), and gas mixtures were humidified by passing them through a water-filled flask. The flow rate of gas through the chamber was maintained at 60 ml min−1 using a mass flow meter (model 840, Sierra Instruments, Monterey, California, USA) and controller (version 1.0, Sable Systems, Henderson, Nevada, USA). Animals were exposed to each gas for approximately 30 min before metabolic rate data were collected. At each oxygen level, oxygen uptake and carbon dioxide output were measured using flow-through respirometry (FC-1 oxygen analyzer with signal conditioner, Sable Systems, Henderson, Nevada, USA; resolution of 0.0005 %; LI-6251 carbon dioxide analyzer, Li-Cor, Lincoln, Nebraska, USA; resolution of 0.00002 %), and mean values were calculated over a period of 1 h. Data were collected and analyzed using Datacan V Software (Sable Systems, Henderson, Nevada, USA). A repeated-measures analysis of variance (ANOVA) was used to compare control values with hypoxic mean values. The respiratory quotient (RQ) (the ratio of oxygen uptake to carbon dioxide output), was calculated at each oxygen level. Because RQs were not normally distributed, a logarithmic conversion was performed before comparing values using a repeated-measures ANOVA.

Before recording metabolic rate measurements at each level of oxygen (i.e. control, exposure and recovery), we used an infrared activity detector (model AD-1, Sable Systems, Henderson, Nevada, USA) under the glass respirometry chamber to measure buccal cavity movements and heart rate. A 15 min trace was recorded at each oxygen level, after allowing 15 min of exposure to a new gas mixture. From the data trace, the following variables could be obtained: apnea duration, respiratory burst duration, buccal pumping frequency during each respiratory burst and heart rate during the apneic periods (Fig. 1). An overall estimate of buccal activity (A in min h−1) was calculated using the respiration data as follows:

Fig. 1.

A 4 min sample trace (A) of buccal activity and heart rate measured with an infrared activity detector from which four variables could be determined: buccal pumping frequency within a burst of activity, heart rate during an apneic period, respiratory burst duration and apneic period duration. The trace in B is a 12 s recording of the region marked in A showing how heart rate and buccal frequency were determined. The ordinate is the intensity of the infrared signal.

Fig. 1.

A 4 min sample trace (A) of buccal activity and heart rate measured with an infrared activity detector from which four variables could be determined: buccal pumping frequency within a burst of activity, heart rate during an apneic period, respiratory burst duration and apneic period duration. The trace in B is a 12 s recording of the region marked in A showing how heart rate and buccal frequency were determined. The ordinate is the intensity of the infrared signal.

formula
where Dburst is respiratory burst duration (s) and t is total observation time (900 s). Means for each buccal variable and for heart rate (over the 15 min period) were compared using a repeated-measures ANOVA with a Tukey’s multiple-comparison post-hoc test.

Because of the non-normal distribution of the apneic period durations, we also analyzed these values visually using log-survivorship curves. This type of analysis is routinely performed (Tyler, 1979) to assess the distributional changes in a behavior that lasts for various durations (in this case, the apneic period or breath-hold duration). Changes in the shapes of the log-survivorship curves give an easy assessment of changes in the distribution of a non-normally distributed variable.

Control (normoxic) values for all variables tested did not differ signifcantly among treatment groups (P>0.05 using a one-way ANOVA). This result allows us to assume that all groups had similar responses to control conditions, even though each group was composed of different individual salamanders.

Metabolic responses to hypoxia

The rate of carbon dioxide output increased significantly in response to 5 % oxygen (P=0.014), but showed no change at the other oxygen levels tested (Fig. 2). The rate of oxygen uptake was unchanged from the control value (21 % oxygen) at 10 % oxygen exposure, but dropped significantly (P<0.05 in all cases) during exposure to 8 %, 6.5 %, 5 % and 2 % oxygen (Fig. 2). Recovery values for carbon dioxide output and oxygen uptake did not differ from control values at any oxygen level tested. The mean respiratory quotient was calculated at each level and increased significantly (P<0.05 in all cases) at 6.5 %, 5 % and 2 % oxygen exposure (Fig. 2).

Fig. 2.

Rate of oxygen uptake, rate of carbon dioxide output and respiratory quotient versus the oxygen level (%). Asterisks (P<0.05) and double daggers (P=0.014) indicate a significant difference compared with control (normoxia, 21 % O2) values. Values are means ± S.E.M. for all animals (see Materials and methods for values of N) at the given oxygen level.

Fig. 2.

Rate of oxygen uptake, rate of carbon dioxide output and respiratory quotient versus the oxygen level (%). Asterisks (P<0.05) and double daggers (P=0.014) indicate a significant difference compared with control (normoxia, 21 % O2) values. Values are means ± S.E.M. for all animals (see Materials and methods for values of N) at the given oxygen level.

Heart rate response to hypoxia

Heart rate increased significantly during exposure to 10 %, 8 %, 6.5 % and 5 % oxygen. No change was observed during exposure to 2 % oxygen (Fig. 3). Heart rate returned to control levels after each hypoxic exposure, demonstrating no latent post-hypoxic effects (Fig. 3).

Fig. 3.

Normoxic, hypoxic and recovery heart rate responses to hypoxia in lungless salamanders: (A) 2 % O2 experiment, (B) 5 % O2 experiment, (C) 6.5 % O2 experiment, (D) 8 % O2 experiment, (E) 10 % O2 experiment. Values are means ± S.E.M. over a 15 minute period for individual animals. Within each treatment, groups with different letters are significantly different (P<0.05). (F) Summary of heart rate responses at each experimental level of oxygen; values are means ± S.E.M. for all animals tested. Asterisks indicate a significant difference (P<0.05) from the control value.

Fig. 3.

Normoxic, hypoxic and recovery heart rate responses to hypoxia in lungless salamanders: (A) 2 % O2 experiment, (B) 5 % O2 experiment, (C) 6.5 % O2 experiment, (D) 8 % O2 experiment, (E) 10 % O2 experiment. Values are means ± S.E.M. over a 15 minute period for individual animals. Within each treatment, groups with different letters are significantly different (P<0.05). (F) Summary of heart rate responses at each experimental level of oxygen; values are means ± S.E.M. for all animals tested. Asterisks indicate a significant difference (P<0.05) from the control value.

Buccal responses to hypoxia

Buccal activity and buccal pumping frequency during respiratory bursts (Figs 4, 5) did not change during exposure to 10 % and 8 % oxygen. There was a significant increase in buccal activity (Fig. 4) during exposure to 6.5 % (P=0.008) and 5 % oxygen (P=0.0005) as a result of a decrease in apneic period duration (Fig. 7), but no change in buccal frequency (Fig. 5).

Fig. 4.

Overall response of buccal activity (min h−1) to hypoxia: (A) 2 % O2 experiment, (B) 5 % O2 experiment, (C) 6.5 % O2 experiment, (D) 8% O2 experiment, (E) 10 % O2 experiment. Within each treatment, groups with different letters are significantly different (P<0.05). Each point is the calculated value for an individual animal. (F) Summary of buccal activity responses at each experimental level of oxygen; values are means ± S.E.M. for all animals tested. Asterisks indicate a significant difference (P<0.05) from the control value.

Fig. 4.

Overall response of buccal activity (min h−1) to hypoxia: (A) 2 % O2 experiment, (B) 5 % O2 experiment, (C) 6.5 % O2 experiment, (D) 8% O2 experiment, (E) 10 % O2 experiment. Within each treatment, groups with different letters are significantly different (P<0.05). Each point is the calculated value for an individual animal. (F) Summary of buccal activity responses at each experimental level of oxygen; values are means ± S.E.M. for all animals tested. Asterisks indicate a significant difference (P<0.05) from the control value.

Fig. 5.

Response of buccal pumping frequency (Hz) to hypoxia: (A) 2 % O2 experiment, (B) 5 % O2 experiment, (C) 6.5 % O2 experiment, (D) 8% O2 experiment, (E) 10 % O2 experiment. Within each treatment, groups with different letters are significantly different (P<0.05). Values are means ± S.E.M. over a 15 minute period for individual animals. (F) Summary of buccal pumping frequency responses at each experimental level of oxygen; values are means ± S.E.M. for all animals tested. Asterisks indicate a significant difference (P<0.05) from the control value.

Fig. 5.

Response of buccal pumping frequency (Hz) to hypoxia: (A) 2 % O2 experiment, (B) 5 % O2 experiment, (C) 6.5 % O2 experiment, (D) 8% O2 experiment, (E) 10 % O2 experiment. Within each treatment, groups with different letters are significantly different (P<0.05). Values are means ± S.E.M. over a 15 minute period for individual animals. (F) Summary of buccal pumping frequency responses at each experimental level of oxygen; values are means ± S.E.M. for all animals tested. Asterisks indicate a significant difference (P<0.05) from the control value.

Fig. 7.

Log-survivorship plots of apnea period durations during normoxia (N), hypoxia (H) and recovery (R) in normoxia: (A) 2 % O2 experiment, (B) 5 % O2 experiment, (C) 6.5 % O2 experiment, (D) 8 % O2 experiment, (E) 10 % O2 experiment. Recorded apnea period durations (t) are plotted against the percentage of periods whose duration is greater than or equal to the given duration for all animals exposed to a given oxygen level. The apneic period duration needed to account for 90 % of the periods recorded (i.e. where 90 % of the apneic periods are ⩽t) is marked on each graph. A leftward shift in the graph represents a shift to shorter breath-holds, and a rightward shift represents a greater proportion of longer breath-holds.

Fig. 7.

Log-survivorship plots of apnea period durations during normoxia (N), hypoxia (H) and recovery (R) in normoxia: (A) 2 % O2 experiment, (B) 5 % O2 experiment, (C) 6.5 % O2 experiment, (D) 8 % O2 experiment, (E) 10 % O2 experiment. Recorded apnea period durations (t) are plotted against the percentage of periods whose duration is greater than or equal to the given duration for all animals exposed to a given oxygen level. The apneic period duration needed to account for 90 % of the periods recorded (i.e. where 90 % of the apneic periods are ⩽t) is marked on each graph. A leftward shift in the graph represents a shift to shorter breath-holds, and a rightward shift represents a greater proportion of longer breath-holds.

Exposure to 2 % oxygen resulted in a significant (P<0.0054) decrease in buccal pumping frequency (Fig. 5). There was no change in buccal activity (Fig. 4) or in the length of apneic periods (Fig. 6) compared with the normoxic controls.

Fig. 6.

Response of apnea period duration (s) to hypoxia: (A) 2 % O2 experiment, (B) 5 % O2 experiment, (C) 6.5 % O2 experiment, (D) 8 % O2 experiment, (E) 10 % O2 experiment. Within each treatment, groups with different letters are significantly different (P<0.05). Values are means ± S.E.M. over a 15 minute period for individual animals. (F) Summary of apnea period duration responses at each experimental level of oxygen; values are means ± S.E.M. for all animals tested.

Fig. 6.

Response of apnea period duration (s) to hypoxia: (A) 2 % O2 experiment, (B) 5 % O2 experiment, (C) 6.5 % O2 experiment, (D) 8 % O2 experiment, (E) 10 % O2 experiment. Within each treatment, groups with different letters are significantly different (P<0.05). Values are means ± S.E.M. over a 15 minute period for individual animals. (F) Summary of apnea period duration responses at each experimental level of oxygen; values are means ± S.E.M. for all animals tested.

Apneic periods were significantly longer during recovery from 10 %, 6.5 % and 5 % hypoxia (P<0.05 for all) than during exposure to these oxygen levels (Fig. 6). The non-normal distribution and the consistently longer apneic periods during recovery led us to construct log-survivorship plots for apneic period duration at each oxygen level tested (Fig. 7) using data combined from all animals exposed to a given oxygen level. During exposure to 10 %, 8 %, 6.5 % and 5 % hypoxia (Fig. 7B–E), a greater percentage of the apneic periods recorded were of shorter duration (a leftward shift in the log-survivor curve) compared with pre-exposure data recorded at 21 % oxygen. During recovery from any level of oxygen, apneic period durations returned to control levels, although longer apneic periods (a rightward shift in the log-survivor curve) were more frequent during recovery from exposure to 10 % and 6.5 % oxygen than pre-exposure. For example, during normoxic control readings before exposure to 6.5 % oxygen (Fig. 7C), 33.5 % of the apneic periods are longer than 14 s. During exposure to 6.5 % oxygen, that value drops to 10 %. During recovery from hypoxia, apneic period duration increases, and 43.7 % of the apneic periods are longer than 14 s.

The most important finding of this study is that the buccal activity and heart rates of lungless salamanders are responsive to environmental hypoxia. The threshold for the buccal response occurs between 8 % and 6.5 % oxygen. There is a significant stimulation during exposure to 6.5 % and 5 % oxygen and no response, or an inhibitory response, to 2 % oxygen. Heart rate increases in response to hypoxia at all tested O2 levels, except 2 % oxygen.

The buccal and heart rate responses coincide with the metabolic response to hypoxia. When encountering hypoxia, most animals maintain their rate of oxygen uptake down to a critical oxygen level (oxyregulation), below which the rate of oxygen uptake conforms (oxyconformation) to the availability of oxygen (Pörtner and Grieshaber, 1993). This property has been observed in lunged and in naturally or artificially lungless animals (Beckenbach, 1975; Pinder, 1987; Tattersall and Boutilier, 1999), demonstrating that the regulation of oxygen uptake during moderate hypoxia can be accomplished without lungs. In our study, the critical oxygen level for oxygen uptake occurs between 10 % and 8 % oxygen. Once below the critical oxygen level, the lungless salamander may still have some ability to facilitate oxygen uptake, as observed by the increase in buccal activity during exposure to 6.5 % and 5 % oxygen. Although increased buccal activity at 6.5 % and 5 % oxygen did not allow metabolic rates to be maintained at control values, it may result in a rate of oxygen uptake that is higher than would otherwise be possible. By continuously replenishing the gaseous environment of the buccal cavity, diffusion across the buccal respiratory surface of the salamander should be enhanced (Gatz et al., 1975).

The increase in carbon dioxide output but not in oxygen uptake observed at 5 % oxygen, together with significant changes in RQ, suggests that the greater buccal activity may result in increased release of carbon dioxide from a relative hyperventilation. An increase in blood flow through cutaneous capillaries (through increased capillary recruitment and/or blood flow rate) in response to hypoxia could also increase carbon dioxide output, because loss of carbon dioxide is perfusion-limited (Burggren and Moalli, 1984). Carbon dioxide output did not continue to increase during exposure to 2 % oxygen, but oxygen uptake decreased, and the resulting high respiratory quotient indicates that the carbon dioxide lost through perfusion does not have a completely metabolic source. A possibility for increased production of carbon dioxide in this situation is the titration of bicarbonate as a result of a decrease in pH caused by the formation of lactic acid.

When considering the decreases in oxygen uptake as the severity of hypoxia increases, it is difficult to differentiate between an active downregulation (i.e. hypometabolism) and an uncontrolled shut-down of the system due to a lack of energy supplies. In another lungless salamander, Desmognathus quadramaculatus, Booth and Feder (1991) suspected that the reduction in oxygen uptake during hypoxia was partially due to a reversible metabolic downregulation, a hypothesis that has recently been gaining favor in the field (Boutilier et al., 1997; Hicks and Wang, 1999).

Once hypoxia becomes severe (between 5 % and 2 % oxygen), the salamanders respond with no change or with a decrease in respiratory activity and heart rate. This may be the point at which the salamanders begin to rely on anaerobic metabolism (Gatz and Piiper, 1979). Although oxygen uptake begins to fall after reaching the critical oxygen tension, plethodontids may rely on anaerobiosis only after passing a lower threshold below which they cannot further decrease energetic demand. Evidence for this is seen in other functionally lungless animals. For example, bullfrogs Rana catesbeiana treated with curare rely entirely on skin breathing. In 5 °C water, these frogs do not begin to accumulate lactate until the of the water is below 50 mmHg (6.67 kPa), even though aerobic metabolic rate decreases at values below 80 mmHg (10.66 kPa) (Pinder, 1987). The inhibition of buccal activity during severe hypoxia may reflect the biphasic (excitation–inhibition) response to graded hypoxia. It may also be adaptive. When the lungless salamander is exposed to extremely low oxygen concentrations, decreasing buccal movements may slow the diffusive loss of oxygen to the environment, at least upon initial exposure.

The observed tachycardia in response to low oxygen levels probably reflects an increase in sympathetic tone and could support increased perfusion of the cutaneous respiratory surface (Boutilier et al., 1986). Heart rate did not increase with exposure to 2 % oxygen. More severe hypoxia (5 mmHg; 0.67 kPa) has previously been observed to decrease heart rate in the dusky salamander (Gatz et al., 1974). In extremely hypoxic conditions, sympathetic activity may be inhibited, or low oxygen levels may directly inhibit heart rate. Buccal pumping may also be inhibited in severe hypoxia, as seen in the decrease in buccal pumping frequency at 2 % oxygen.

In general, the buccal response to moderate hypoxia in the dusky salamander is similar to those of other lunged amphibians and reptiles. The hypoxic buccal response in these lungless salamanders is accomplished mostly by an increase in the overall amount of buccal pumping (Fig. 5) and, in some cases, by a decrease in the duration of the apneic period (Fig. 7), but not by a change in the actual buccal pumping frequency. In most amphibians, the ventilatory response to hypoxia is achieved by a reduction in the duration of the non-ventilatory periods (apneas) and a consequent overall increase in lung ventilation (Boutilier, 1990). This phenomenon is demonstrated in the log-survivorship plots (Fig. 7).

At most levels of hypoxia, the apneic period durations appeared to be shorter than during normoxia, demonstrating a shift to shorter breath-holds and away from longer apneic periods (Fig. 7). A similar pattern has been observed in amphibians responding to hypercapnia (Boutilier, 1984). It is unclear whether the shift in buccal breathing patterns during hypoxia is a chemosensory-driven reflex or whether the buccal activity is controlled by higher centers in the brain (Milsom et al., 1997). However, the almost constant frequency of individual buccal movements across all levels of oxygen implies that a functional central rhythm generator (see Milsom et al., 1999) exists and that it is not modulated by peripheral chemosensory input. As a result, changes in buccal pumping activity can be accommodated only by alterations in the duration of breathing and apneic periods. To complicate matters further, the buccal breathing patterns of lungless salamanders appear to be subject to multiple levels of control (e.g. olfaction and chemosensitivity; Burggren and Just, 1992). In fact, we found that direct observations, small noises and vibrations could bring about increases in buccal pumping that appeared to have little to do with ventilatory requirements.

In summary, the response of the dusky salamander to decreasing oxygen levels is biphasic, showing first an increase in buccal pumping, accomplished by shortening apnea duration and increasing buccal activity periods and, therefore, increasing minutes per hour of respiratory activity. This is achieved by changing the amount, but not the frequency, of buccopharyngeal pumping. Below a lower limit (between 5 % and 2 % oxygen), buccal activity is no longer increased. These changes in buccal activity may aid oxygen uptake during moderate hypoxia and may decrease metabolic costs and curtail the loss of oxygen during severe hypoxia. We cannot address the effect of this biphasic response on metabolic rate on the basis of our data. Presumably, the ancestral lunged salamanders from which plethodontids were derived had a hypoxic drive for ventilation, as do modern amphibians (Kruhøffer et al., 1987). However, if there had been no pressure for selection against a buccal response, the presence of a hypoxic drive in these lungless animals could have been retained even if the benefit were very small and required no current utility. An interesting idea, which remains to be tested, is that the variation in response between animals might be a sign of diminished selection pressure on the hypoxic ventilatory response after the loss of lungs. In conclusion, the buccal response to graded hypoxia is consistent with a ventilatory function; i.e. a hypoxic ventilatory response, which appears to be similar to that of lunged vertebrates.

This research was supported by NIH grant HL40537. Thanks to Dr Lowell Orr for help with animal collection and to Shannon Beham for laboratory help.

Beckenbach
,
A. T.
(
1975
).
Influence of body size and temperature on the critical oxygen tension of some plethodontid salamanders
.
Physiol. Zool
.
48
,
338
347
.
Booth
,
D. T.
and
Feder
,
M. E.
(
1991
)
Formation of hypoxic boundary layers and their biological implications in a skin-breathing aquatic salamander, Desmognathus quadramaculatus
.
Physiol. Zool
.
64
,
1307
1321
.
Boutilier
,
R. G.
(
1984
).
Characterization of the intermittent breathing pattern in Xenopus laevis
.
J. Exp. Biol
.
110
,
281
309
.
Boutilier
,
R. G.
(
1990
).
Control and co-ordination of gas exchange in bimodal breathers
.
In Advances in Comparative and Environmental Physiology
, vol.
6
(ed.
R. G.
Boutilier
), pp.
280
345
.
Berlin
:
Springer-Verlag
.
Boutilier
,
R. G.
,
Donohoe
,
P. H.
,
Tattersall
,
G. J.
and
West
,
T. G.
(
1997
).
Hypometabolic homeostasis in overwintering aquatic amphibians
.
J. Exp. Biol
.
200
,
387
400
.
Boutilier
,
R. G.
,
Glass
,
M. L.
and
Heisler
,
N.
(
1986
).
The relative distribution of pulmocutaneous blood flow in Rana catesbeiana: effects of pulmonary or cutaneous hypoxia
.
J. Exp. Biol
.
126
,
33
39
.
Boutilier
,
R. G.
and
Toews
,
D. P.
(
1981
).
Respiratory, circulatory and acid–base adjustments to hypercapnia in a strictly aquatic and predominantly skin-breathing urodele, Cryptobranchus alleganiensis
.
Respir. Physiol
.
46
,
177
192
.
Burggren
,
W. W.
and
Just
,
J. J.
(
1992
).
Developmental changes in physiological systems
.
In Environmental Physiology of the Amphibians
(ed.
M. E.
Feder
and
W. W.
Burggren
), pp.
467
530
.
Chicago
:
University of Chicago Press
.
Burggren
,
W. W.
and
Moalli
,
R.
(
1984
).
‘Active’ regulation of cutaneous gas exchange by capillary recruitment in amphibians: Experimental evidence and a revised model for skin respiration
.
Respir. Physiol
.
55
,
379
392
.
Czopek
,
J.
(
1961
).
Vascularization of respiratory surfaces in some Plethodontidae
.
Zool. Pol
.
11
,
131
150
.
Feder
,
M. E.
,
Full
,
R. F.
and
Piiper
,
J.
(
1988
).
Elimination kinetics of acetylene and Freon 22 in resting and active lungless salamanders
.
Respir. Physiol
.
72
,
229
240
.
Feder
,
M. E.
and
Wassersug
,
R. J.
(
1984
).
Aerial versus aquatic oxygen consumption in larvae of the clawed frog, Xenopus laevis
.
J. Exp. Biol
.
108
,
231
245
.
Gatz
,
R. N.
,
Crawford
,
E. C.
, Jr
and
Piiper
,
J.
(
1974
).
Metabolic and heart rate response of the plethodontid salamander, Desmognathus fuscus to hypoxia
.
Respir. Physiol
.
20
,
43
49
.
Gatz
,
R. N.
,
Crawford
,
E. C.
and
Piiper
,
J.
(
1975
).
Kinetics of inert gas equilibration in an exclusively skin-breathing salamander, Desmognathus fuscus
.
Respir. Physiol
.
47
,
151
164
.
Gatz
,
R. N.
and
Piiper
,
J.
(
1979
).
Anaerobic energy metabolism during severe hypoxia in the lungless salamander Desmognathus fuscus (Plethodontidae
).
Respir. Physiol
.
38
,
377
384
.
Hicks
,
J. W.
and
Wang
,
T.
(
1999
).
Hypoxic hypometabolism in the anesthetized turtle, Trachemys scripta
.
Am. J. Physiol
.
46
,
R18
R23
.
Kruhøffer
,
M.
,
Glass
,
M. L.
,
Abe
,
S. A.
and
Johansen
,
K.
(
1987
).
Control of breathing in an amphibian Bufo paracnemis: effects of temperature and hypoxia
.
Respir. Physiol
.
69
,
267
275
.
Milsom
,
W. K.
,
Harris
,
M. B.
and
Reid
,
S. G.
(
1997
).
Do descending influences alternate to produce episodic breathing?
Respir. Physiol
.
110
,
307
317
.
Milsom
,
W. K.
,
Reid
,
S. G.
,
Meier
,
J. T.
and
Kinkead
,
R.
(
1999
).
Central respiratory pattern generators in the bullfrog Rana catesbeiana
.
Comp. Biochem. Physiol
.
124A
,
253
264
.
Piiper
,
J.
,
Gatz
,
R. N.
and
Crawford
,
E. C.
(
1976
).
Gas transport characteristics in an exclusively skin-breathing salamander, Desmognathus fuscus (Plethodontidae
).
In Respiration of the Amphibious Vertebrates
(ed.
G. M.
Hughes
), pp.
339
356
.
London
:
Academic Press
.
Pinder
,
A. W.
(
1987
).
Cutaneous diffusing capacity increases during hypoxia in cold submerged bullfrogs (Rana catesbeiana
).
Respir. Physiol
.
70
,
85
95
.
Pörtner
,
H.-O.
and
Grieshaber
,
M. K.
(
1993
).
Critical (s) in oxyconforming and oxyregulating animals: gas exchange, metabolic rate and the mode of energy production
.
In The Vertebrate Gas Transport Cascade: Adaptations to Environment and Mode of Life
(ed.
J. E. P. W.
Bicudo
), pp.
330
357
.
Boca Raton, FL
:
CRC Press
.
Ruben
,
J. A.
and
Boucot
,
A. J.
(
1989
).
The origin of the lungless salamander (Amphibia: Plethodontidae
).
Am. Nat
.
134
,
161
169
Tattersall
,
G. J.
and
Boutilier
,
R. G.
(
1999
).
Behavioural oxy-regulation by cold-submerged frogs in heterogeneous oxygen environments
.
Can. J. Zool
.
77
,
843
850
.
Tyler
,
S.
(
1979
).
Time-sampling: a matter of convention
.
Anim. Behav
.
27
,
801
810
.
Whitford
,
W. G.
and
Hutchison
,
V. H.
(
1965
).
Gas exchange in salamanders
.
Physiol. Zool
.
38
,
228
242
.