Effects of environmental hypoxia and hypercarbia on ventilation and gas exchange in Testudines

Animals

Adults of both sexes of T. scripta (body mass: M_B = 1.08 ± 0.10 kg; N = 8) and C. carbonarius (M_B = 3.77 ± 0.61 kg; N = 6) living under natural conditions were obtained from the Jacarezário, Univeridade Estadual Paulista “Júlio de Mesquita Filho”, Rio Claro, SP, Brazil, transported to the laboratory at the University of São Paulo in Ribeirão Preto, SP, and maintained for at least three months before experimentation to acclimate to laboratory conditions. Experiments were performed between November 2014 and February 2015 following approval by the Instituto Chico Mendes de Conservação da Biodiversidade (SISBIO; license number 35221-1) and Comissão de Ética no Uso de Animais (CEUA USP/Campus de Ribeirão Preto; protocol number 12.1.1541.53.0). Animals were maintained under a 12 h light/dark photoperiod cycle, in a temperature-controlled room at 25 ± 2 °C and received a mixed diet supplemented with amino acids, vitamins and minerals (Aminomix Pet, Vetnil^®, Louveira, Brazil) three times a week. T. scripta were housed in a box with a water reservoir for diving whereas C. carbonarius were housed in boxes whose bottom was covered with wooden chips.

Setup

Animals were submitted to open respirometry following Glass, Wood & Johansen (1978), Wang & Warburton (1995) and Silva et al. (2011) to measure ventilation and gas exchange. Individuals of T. scripta were placed in an aquarium with a single access to an inverted funnel, and each individual only needed to extend its neck and protrude its nostrils into the chamber for air breathing. C. carbonarius were placed in a plastic box and a mask was fitted to the head of each animal for respirometry and a collar was fixed to the neck to prevent head retraction. The dead space of the funnel or the mask was never larger than 40 ml. The exit of the funnel and the frontal tip of the mask were equipped with a pneumotach (Fleisch tube), which was connected to a spirometer (FE141, ADInstruments, Sydney, Australia). The gas inside the funnel or mask was sampled at 180  ml min⁻¹, dried, and pulled to a gas analyzer (ML206, ADInstruments, Sydney, Australia). Data were recorded and analyzed using PowerLab 8/35 and LabChart 7.0 (ADInstruments, Sydney, Australia).

Both the funnel and the mask were calibrated by injections of known volumes, using an Inspira ventilator (Harvard Apparatus, Cambridge, MA, USA), and concentrations of gas, supplied by a Pegas 4000MF gas mixer (Columbus Instruments Columbus, OH, USA). Air was used for the spirometer calibration with volumes ranging from 1 to 60 ml and different volumes and concentrations of O₂ were used to calibrate the gas exchange measurements. In all cases, calibrations resulted in linear regressions with R² >0.95.

Experimental protocols

Experimental temperature and photoperiod were the same as during maintenance, and all animals were fasted for three to seven days before experimentation to avoid the confounding effects of digestion on metabolism. The animals were weighed one day before the beginning of each experimental treatment. Before any measurements, animals were placed into the experimental setup or equipped with a mask at least 12 h before initiation of experiments. Experimentation started around 08:00, and ventilation and gas exchange were measured first under normoxic conditions, followed by progressively decreasing hypoxic (9, 7, 5, 3% O₂) or progressively increasing hypercarbic (1.5, 3.0, 4.5, 6.0% CO₂) exposures, and after that animals were exposed again to normoxia. The exposure time of each gas mixture, as well as normoxic conditions, was 2 h. Animals were first exposed to one randomly chosen progressive gas exposure and the following day to the other one.

Data analysis

The last hour of each exposure was used to extract the following data: breathing frequency during breathing episodes (f_Repi, breaths episode⁻¹), frequency of breathing episodes (f_E, episodes.h⁻¹), duration of non-ventilatory period (T_{NV P}, s; defined as the time between the end of an inspiration and the beginning of the following expiration), duration of inspiration (T_INSP, s), duration of expiration (T_EXP, s), total duration of one ventilatory cycle (T_TOT = T_EXP + T_INSP, s), tidal volume (V_T, ml kg⁻¹), breathing frequency (f_R, breaths min⁻¹), and oxygen consumption (VO₂, mlO₂ kg⁻¹) (Fig. 1). During episodic breathing, due to the slow response time of the oxygen analyzer, oxygen consumption was determined by integrating the area above the oxygen trace for the entire episode, and then divided by the number of expirations to obtain mean oxygen consumption per breath. From the extracted data, the instantaneous breathing frequency (f′, breaths min⁻¹), the relative duration of expiration (T_EXP∕T_TOT), the relation between inspiration and expiration (T_INSP∕T_EXP), the expiratory flow rate (V_T∕T_EXP, ml s⁻¹), minute ventilation (${\dot{V}}_E$, ml kg⁻¹ min⁻¹), oxygen consumption ($\dot{V}{O}_2$, mlO₂ kg⁻¹ min⁻¹), and air convection requirement (${\dot{V}}_E$/ $\dot{V}{O}_2$, ml mlO${}_2^{-1}$) were calculated.

Data were analyzed using GraphPad Prism 6.0 and applying Repeated Measures ANOVA followed by a Tukey’s multiple comparison test. Values of P < 0.05 were considered significant.

To compare the results of the present study with previously published data, we searched for relevant publications using Pubmed, Web of Science, and Google Scholar databases using keywords such as ‘turtle’, ‘Testudines’, ‘hypoxia’, ‘hypercarbia’, ‘hypercapnia’, ‘ventilation’, ‘gas exchange’, etc. Values of respiratory variables of Testudines obtained under exposure to environmental hypoxia or hypercarbia (but not anoxia or hypoxic-hypercarbia) measured at temperatures between 20 and 30 °C were included (Table 1). Data from animals that had their trachea cannulated, or that had their respiratory system surgically manipulated, were not included. Values were directly obtained from the text or tables given, or by extracting values from published figures using the free software PlotDigitizer (version 2.6.2). To enable comparison among species, data were expressed as changes relative to normoxic values. Due to the very low number of chelonian species with a complete set of respiratory variables available and the varying experimental protocols applied at different temperatures, levels of hypoxia or hypercarbia, phylogenetically informed multivariate analysis was not possible.

Ventilation and oxygen consumption in T. scripta and C. carbonarius

During normoxia, both species showed an episodic breathing pattern with two to three ventilatory cycles interspersed by non-ventilatory periods (Fig. 2). The T_{NV P} was, on average, three to four times longer in T. scripta than in C. carbonarius. In T. scripta, hypoxia significantly increased f_E, V_T, ${\dot{V}}_E$ and ${\dot{V}}_E∕\dot{V}{O}_2$, whereas f_Repi, T_{NV P} and $\dot{V}{O}_2$ were significantly reduced (Figs. 3–6). Under hypoxic exposure, C. carbonarius significantly increased f_E, T_INSP, V_T, f_R, ${\dot{V}}_E$, and ${\dot{V}}_E∕\dot{V}{O}_2$ (Figs. 3–6). Once the hypoxic exposure ended, all variables returned to pre-hypoxic values within 1 hour, with the exception of f_R in C. carbonarius, which was significantly greater when compared to the pre-hypoxic value. Exposure to CO₂ increased ${\dot{V}}_E$ and ${\dot{V}}_E∕\dot{V}{O}_2$ significantly and decreased $\dot{V}{O}_2$ significantly in T. scripta, whereas in C. carbonarius T_INSP, T_TOT, V_T, f_R, ${\dot{V}}_E$, and ${\dot{V}}_E∕\dot{V}{O}_2$ significantly increased but T_{NV P} and f′ significantly decreased (Figs. 3–6). One hour after the withdrawal of CO₂, all variables had returned to pre-hypercarbic values. The relationships between T_EXP, T_INSP and T_TOT (i.e., T_EXP/T_TOT, and T_INSP/T_EXP respectively) were not significantly affected by either hypoxia nor hypercarbia, just as expiratory flow rate (V_T/T_EXP), but the latter did show a tendency to increase in both species with increasing levels of hypoxia and hypercarbia (Fig. 5).

Relative changes in respiratory variables

Both hypoxia and hypercarbia increased ventilation. This increase was achieved by increasing the number of breathing episodes, caused by decreasing the non-ventilatory period (Fig. 7). T_{NV P} at 3% O₂, for example, consistently represented about 20% of the T_{NV P} seen during normoxia in all species investigated, whereas 6% CO₂ roughly reduced T_{NV P} by 50%. Interestingly, hypercarbia about doubled f_E, with the exception of P. geoffroanus, and slightly increased f_Repi (exceptions P. geoffroanus and C. carbonarius), whereas hypoxia caused a greater increase in f_E, but slightly decreased f_Repi (exception C. carbonarius). Neither hypoxia nor hypercarbia drastically altered T_INSP, T_EXP, T_TOT, and f′ (Fig. 8), as well as T_EXT/T_TOT and T_INSP/T_EXP (Fig. 9).

V_T/T_EXP increased two to five-fold under hypoxic and hypercarbic conditions (Fig. 9) in all species studied, which was mainly caused by an about two to three-fold increase in V_T at severe levels of hypoxia and hypercarbia (Fig. 10). C. carbonarius, showing a 12-fold, and P.  geoffroanus, showing a six-fold increase in V_T, were the only species showing much larger increases in V_T. Several species increased f_R during hypercarbia about six to seven-fold, but many species only doubled or tripled f_R (Fig. 10). The only species that increased f_R more than three-fold during hypoxia were C. picta at 30 °C (Glass, Boutilier & Heisler, 1983) and P. geoffroanus at 25 °C (Cordeiro, Abe & Klein, 2016). The product of V_T and f_R, minute ventilation, showed the greatest relative increases, with P. geoffroanus increasing ${\dot{V}}_E$ 42 times and C. carbonarius about 30 times, both at 6% CO₂. The relative increase at 6% CO₂ ranged from four to 12 times, whereas at 3% O₂ the increase in ${\dot{V}}_E$ ranged between three and six or between 12 and 17 for C. picta, C. carbonarius and P. geoffroanus.

With the exception of P. geoffroanus, both under hypoxia and hypercarbia, and of P. unifilis under hypercarbia, $\dot{V}{O}_2$ decreased or remained unaltered during both exposures (Fig. 11). The resulting air convection requirement, however, increased about 10 to 30-fold in T. scripta (Jackson (1973) and Lee & Milsom (2016) versus this study, respectively), in C. picta (3% O₂; Glass, Boutilier & Heisler, 1983), and in C. carbonarius (4.5 and 6% CO₂; this study) (Fig. 11). In the remaining species, ${\dot{V}}_E∕\dot{V}{O}_2$ increased about three to 12 times under both hypoxic and hypercarbic conditions.

Ventilation and oxygen consumption in T. scripta and C. carbonarius

Breathing pattern of both species followed the general reptilian behavior of intermittent lung ventilation. Burggren (1975) and Glass, Burggren & Johansen (1978) observed intermittent ventilation in Testudo graeca and T. pardalis, respectively, but in both species breathing pattern consisted of just one ventilatory cycle interspersed by short and regular non-ventilatory periods. In the present study, both, T. scripta and, unexpectedly, C. carbonarius, showed more than one ventilatory cycle per breathing episode, but the mean duration of the non-ventilatory periods was lower in C. carbonarius when compared to T. scripta. Vitalis & Milsom (1986a) consider episodic breathing an adaptive mechanism that decreases the energetic cost of ventilation in ectotherms, and Randall et al. (1981) consider such a breathing behavior advantageous for aquatic species, since it reduces the energetic cost to surface and also reduces the exposure time at the surface, possibly lessening risks of predation. Since episodic breathing with long non-ventilatory periods leads to a significant change in arterial blood gases, decreasing P_aO₂ and pH and increasing P_aCO₂ (Glass, Burggren & Johansen, 1978), as well as decreasing the efficiency of pulmonary CO₂ excretion (Malte, Malte & Wang, 2013), it should be more advantageous for a terrestrial species to ventilate regularly and thereby maintain homeostasis of arterial blood gases. It is therefore interesting to ask why the terrestrial C. carbonarius employs episodic breathing under normoxic conditions, thereby possibly increasing variation in arterial blood gases instead of maintaining a regular breathing pattern, such as seen in this species only under severe levels of hypoxia or hypercarbia (Fig. 2). C. carbonarius does frequently seek shelter in shallow burrows or other small spaces and remains non-ventilatory for long periods (AS Abe, pers. obs., 2000), possibly explaining the episodic breathing seen in this terrestrial species, but currently a physiological explication for this behavior is lacking. Interestingly, other ectothermic terrestrial species such as varanid (Thompson & Withers, 1997) and agamid lizards (Frappell & Daniels, 1991) also breathe intermittently. However concomitant blood gas analyses have not been performed in these species to verify accompanying variations in blood gases or pH.

Comparing our data with previous studies on the effect of hypoxia or hypercarbia on ventilation and gas exchange in Trachemys scripta, Frankel et al. (1969) found values for T_TOT about three times larger during normoxia, hypoxia and hypercarbia when compared to our study. However animals in their study had their tracheas cannulated which may have influenced the length of the ventilatory cycle, since T_TOT values reported by Vitalis & Milsom (1986b) (calculated from their f′: 1.7 s during normoxia and 4% O₂ and 1.8 s during 3–5% CO₂) are similar to ours. Reyes & Milsom (2009) report similar values for f_E as in the present study (from 8.4 ± 1.6 in normoxia during winter up to 37.1 ± 2.3 episodes.h⁻¹ in summer), but found considerable variation in f_Repi through different seasons, ranging from 3.6 ± 0.4 breaths.episode⁻¹ in normoxia during winter up to 26.1 ± 5.4 breaths.episode⁻¹ in hypoxic-hypercarbia during autumn, thereby demonstrating considerable seasonal variation in breathing pattern in T. scripta. Lee & Milsom (2016) report nearly identical values as in the present study for f_Repi and f_E during normoxia and hypoxia, and Frankel et al. (1969) report a comparable f_Repi during normoxia. Johnson & Creighton (2005), on the other hand, report greater values of f_Repi during both normoxia and hypercarbia, and Frankel et al. (1969) found f_Repi at 10–12% CO₂ to be 5.6 ± 1.0 at 28 °C.

More data are available regarding V_T, f_R, ${\dot{V}}_E$, $\dot{V}{O}_2$, and ${\dot{V}}_E∕\dot{V}{O}_2$ during both, hypoxic and hypercarbic exposures. In general, data obtained in the present study for normoxia are similar to the ones obtained by other authors, such as $\dot{V}{O}_2$, which at 25 °C varies from 0.82 (Hicks & Wang, 1999; this study) to 1.1 mlO₂ kg⁻¹ min⁻¹ (24 °C; Jackson & Schmidt-Nielsen, 1966), whereas the values given by Vitalis & Milsom (1986b) for V_T and ${\dot{V}}_E$ are the lowest ones reported for T. scripta exposed to hypoxia or hypercarbia. The overall changes observed in the ventilatory responses of T. scripta to hypoxia and hypercarbia are also comparable between the present study and data from the literature. Only ${\dot{V}}_E∕\dot{V}{O}_2$ in the present study, both during hypoxia and hypercarbia, was greater when compared to data from the literature. This difference was caused by a much lower $\dot{V}{O}_2$ during hypoxic and hypercarbic exposures when compared to data from other authors, since ${\dot{V}}_E$ was very similar to data obtained by others at similar temperatures (Jackson, Palmer & Meadow, 1974; Lee & Milsom, 2016). The oxygen consumption measured by us during hypercarbia was similar to the one obtained by Jackson, Palmer & Meadow (1974) at 10 °C, a 15 °C difference, that may represent a variation in chemosensivity seen in this species during different seasons (Reyes & Milsom, 2009), as we found similarly low $\dot{V}{O}_2$ values during hypoxic exposures. Interestingly, our normoxic $\dot{V}{O}_2$ values were well within the range for T. scripta at 25 °C reported in the literature (Hicks & Wang, 1999; Jackson & Schmidt-Nielsen, 1966). A significant drop in oxygen consumption during hypoxia has also been described before (Jackson & Schmidt-Nielsen, 1966; Jackson, 1973; Lee & Milsom, 2016), whereas other studies did not find a pronounced fall in metabolism during hypercarbia (Hicks & Wang, 1999; Jackson, Palmer & Meadow, 1974). One motive for the observed variations could lie in the significant seasonal variations in metabolism, gas exchange, and, consequently, ventilation found in T. scripta (Reyes & Milsom, 2009), variations that possibly were not eliminated by maintaining the animals at a constant temperature of 25 °C. Furthermore, exposing animals for 2 hours to each gas mixture may not have been sufficient to reach a physiological steady-state, as suggested by Malte, Malte & Wang (2016). Another reason for this discrepancy could be the species physiological phenotypic plasticity, since animals used in the previous studies were native to the North American continent and thereby subject to a more temperate climate than the animals used in the present study, that have been bred under the subtropical climate of southeastern Brazil.

The values for minute ventilation in T. scripta at 8% CO₂ found by Hitzig & Nattie (1982) seem somewhat low, when compared to the values found by Johnson & Creighton (2005) at the same CO₂ concentration at a different temperature (20 versus 27–28 °C, respectively), but are somewhat similar to the values found by Jackson, Palmer & Meadow (1974) at 20 °C and 6% CO₂ (135.0 versus 215 ml kg⁻¹ min⁻¹, respectively). The general response of T. scripta to reducing oxygen concentrations can be described by a moderate, when compared to the response during hypercarbia, increase in minute ventilation, mainly caused by increasing V_T, and a reduction in oxygen consumption, thereby increasing the air convection requirement. These changes are generally more pronounced below 5% O₂. The response to hypercarbia also includes an increase in ventilation due to an increase in V_T and f_R. In T. scripta neither hypoxia nor hypercarbia caused significant changes in T_INSP, T_EXP, T_TOT, f′, and f_Repi, whereas f_E and T_{NV P}, increased and decreased significantly, respectively.

In respect to C. carbonarius during hypoxic or hypercarbic exposures, only data on V_T, f_R, ${\dot{V}}_E$, and $\dot{V}{O}_2$ are available for other terrestrial Testudines belonging to the Emydidae and Testudinidae. Despite comparing different species, the ventilatory variables are similar among the terrestrial species studied, with the exception of the normoxic $\dot{V}{O}_2$ value given by Altland & Parker (1955) for Terrapene carolina carolina, possibly indicating that animals in their study may not have been resting quietly during normoxia. However, their $\dot{V}{O}_2$ value reported for 3–5% O₂ is identical to the values from other studies at similar oxygen concentrations. V_T in C. carbonarius is on the lower end of data available for terrestrial Testudines, which may have been influenced by the relative large amount of bone tissue present in adult individuals of this species (AS Abe, pers. obs., 2005). Breathing frequency and ${\dot{V}}_E$, on the other hand, were very similar to the data obtained on other terrestrial Emydidae and Testudinidae (Altland & Parker, 1955, Benchetrit, Armand & Dejours, 1977, Burggren, Glass & Johansen, 1977, Glass, Burggren & Johansen, 1978 and Benchetrit & Dejours, 1980).

Ultsch & Anderson (1988), studying Gopherus polyphemus and Terrapene carolina, found values of oxygen consumption very similar to those of C. carbonarius during both normoxia and hypoxia. Interestingly, G. polyphemus spends a significant amount of time in burrows that may show hypoxia as well as hypercarbia, and whose critical oxygen level (percentage of O₂ where $\dot{V}{O}_2$ starts decreasing) can be found at approximately 1.5% O₂, whereas the exclusively terrestrial T. carolina shows a somewhat larger critical oxygen tension of 3.5% O₂ (Ultsch & Anderson, 1988). Since C. carbonarius did not show any significant changes in $\dot{V}{O}_2$during hypoxia down to 3% O₂, the critical oxygen level of this species seems to be similar to the one seen in the former two species, but $\dot{V}{O}_2$ was consistently lower at any oxygen concentration when compared to G. polyphemus and T. carolina and e.g., at 3% O₂ (0.08 mlO₂ kg⁻¹ min⁻¹) was similar to the lowest $\dot{V}{O}_2$ given for G. polyphemus (0.05 mlO₂ kg⁻¹ min⁻¹) and T. carolina (0.08 mlO₂ kg⁻¹ min⁻¹) at less than 1% O₂ (Ultsch & Anderson, 1988). C. carbonarius is not known to use burrows and therefore may not show a critical oxygen level as low as G. polyphemus, but Chelonoidis chilensis has been reported to use shallow burrows for retreat during cold days (Pritchard, 1979) and therefore other species of the Testudinidae may possess a similarly low oxygen level as the testudinidid G. polyphemus.

Relative changes in respiratory variables

Analyzing the respiratory variables available in the literature for chelonians exposed to hypoxia and hypercarbia (Figs. 7–11), one notices the discrepancy in data availability between commonly studied parameters such as V_T, f_R, ${\dot{V}}_E$, and $\dot{V}{O}_2$, and less frequently reported ones such as T_EXP, T_TOT, or f_E, for example. Furthermore, only very few terrestrial species have been studied, when compared to the wealth of data available for T.  scripta and C. picta. Based on the data analyzed, it seems clear that the breathing pattern of terrestrial chelonians does not significantly differ from aquatic or semi-aquatic species when considering the responses to hypoxia and hypercarbia. With few exceptions, both hypoxia and hypercarbia elicit similar respiratory responses, showing variation mainly in the magnitude of the species’ responses. The different patterns seen in f_E and f_Repi during hypoxia and hypercarbia may suggest varying degrees of chemosensivity between species and towards different gas exposures. Previous experimental manipulations transforming episodic breathing into continuous single ventilations in T. scripta were vagotomy (Vitalis & Milsom, 1986b) and dissection of the spinal cord (Johnson & Creighton, 2005). Recently, Johnson, Krisp & Bartman (2015) changed episodic breathing in T. scripta from episodic to singlet breathing through pharmacological manipulation of serotonin 5-HT₃ receptors. Studying the participation of serotonin in central chemoreception under hypoxia and hypercarbia in phylogenetically distant species, as well as species occupying different habitats, might help to elucidate the varying responses to hypoxia and hypercarbia seen in chelonian breathing pattern, since the switch from episodic to singlet breathing under hypoxia has been suggested to be caused by an increased respiratory drive (Fong, Zimmer & Milsom, 2009). However, under hypercarbia nearly all species increase the number of breaths per episode, and do not decrease f_Repi as under hypoxia, suggesting that CO₂ exposure increases respiratory drive by different regulatory pathways than under hypoxia. Interestingly, Herman & Smatresk (1999) demonstrated that in T. scripta hypoxia and hypercarbia cause different changes in pulmonary ventilation and perfusion. During hypoxia, lung ventilation and perfusion increased, whereas under hypercarbia only lung ventilation increased, but not pulmonary perfusion, resulting in a ventilation/perfusion mismatch during exposure to CO₂. Burggren, Glass & Johansen (1977) found a similar cardiovascular response in Pelomedusa subrufa and in Testudo pardalis, suggesting a common testudine response, but its importance or relation with the differences observed in f_Repi remains unclear.

P. geoffroanus and C. carbonarius seem to be more sensitive regarding T_INSP, T_EXP, T_TOT, f′, T_EXT/T_TOT, and T_INSP/T_EXP, with the former species increasing these variables mainly during hypercarbia, but the latter one increasing all variables with increasing levels of hypoxia and hypercarbia. Such increases in T_INSP and T_EXP have been interpreted by Johnson, Krisp & Bartman (2015) as a stronger respiratory drive from central respiratory neurons, whose intensity, however, seems to vary among species. The absolute and relative decrease in instantaneous breathing frequency seen in C. carbonarius implies that breathing mechanics may be more variable than previously anticipated for Testudines, since Vitalis & Milsom (1986b) found f′ to be unaffected by either hypoxia or hypercarbia in T. scripta and suggested (Vitalis & Milsom, 1986a; Vitalis & Milsom, 1986b) that T. scripta breathes at combinations of volume and frequency to keep the mechanical work of breathing at a minimum. In the present study, f′ in T. scripta, as well as in C. carbonarius, did show larger variations than reported for T. scripta in earlier studies (Frankel et al., 1969; Vitalis & Milsom, 1986b). Vitalis & Milsom (1986a) found, based on mechanical analyses of the respiratory system of T. scripta, that the mechanical work of breathing is minimal at ventilation frequencies of 35 to 45 cycles min⁻¹ for different levels of minute pump ventilation (100, 200, 300 ml min⁻¹), meaning that for a minute pump ventilation of 200 ml min⁻¹. Animals should therefore ventilate at a frequency of 40 breaths min⁻¹ and a tidal volume of 5 ml to ventilate the respiratory system with the lowest mechanical work, but such a breathing pattern would result in severe alkalosis due to increased CO₂ excretion (Vitalis & Milsom, 1986b). In the present study, however, T. scripta reached the greatest level of minute ventilation (215.9 ml min⁻¹ kg⁻¹) at 6% CO₂, using a tidal volume of 57.2 ml kg⁻¹ and an instantaneous breathing frequency of 18.3 breaths min⁻¹ (f_R = 3.0 breaths min⁻¹), values much different from mechanical predictions. The significance of this variation in breathing pattern versus the mechanical predictions of work of breathing needs to be investigated to better understand the mechanical work of breathing of the Testudines respiratory system, since mechanical work of breathing increases markedly with increasing tidal volume, e.g., from 57 to 272 ml cmH₂O min⁻¹ kg⁻¹ at 6.2 ml kg⁻¹ and 3.0 breaths min⁻¹ in undisturbed T. scripta versus 34.2 and 0.8 breaths min⁻¹ in vagotomized T. scripta, each at 4% O₂ (Vitalis & Milsom, 1986b).

The relatively large increases seen in V_T/T_EXP of C. carbonarius and P. geoffroanus can be explained by very low values of V_T under normoxic conditions. P. geoffroanus (3.1 ml kg⁻¹; Cordeiro, Abe & Klein, 2016) and C. carbonarius (3.98 ml kg⁻¹; this study) show much smaller tidal volumes during normoxia than other chelonians (mostly between 10 and 20 ml kg⁻¹), resulting in relatively larger increases in V_T during hypoxia and hypercarbia than the other species. Both species also showed relatively larger increases in ${\dot{V}}_E$, which are again attributable to the low values seen in f_R and V_T under normoxic conditions.

Whereas ${\dot{V}}_E$ increases largely in all species, $\dot{V}{O}_2$ remains unaltered or even decreases under both hypoxia and hypercarbia in nearly all species investigated. The relative increase seen in P. geoffroanus under both hypoxia and hypercarbia can be explained by the very low oxygen consumption under normoxic conditions, which could be a consequence of significant extra-pulmonary gas exchange or hypometabolism in this species (Cordeiro, Abe & Klein, 2016). Increases in ${\dot{V}}_E∕\dot{V}{O}_2$ have been linked both under hypoxia (e.g., Glass, Boutilier & Heisler, 1983) and hypercarbia (e.g., Funk & Milsom, 1987) to regulation of arterial PO₂, PCO₂, and pH, as all turtles investigated maintain control of these variables under varying environmental conditions.

The chelonian respiratory system shows significant variations in lung structure, as well as in associated structures such as the post-pulmonary septum (PPS; Perry, 1998). The PPS is a membrane that partially or completely separates the lungs from the other viscera (Lambertz, Böhme & Perry, 2010). As a testudinid, C. carbonarius possesses a complete post-pulmonary septum (W Klein, pers. obs., 2015), when compared to the incomplete PPS of the emydid T. scripta (Lambertz, Böhme & Perry, 2010). The presence or absence of a PPS may significantly influence the mechanics of the respiratory system, as has been shown for the post-hepatic septum of the lizard Salvator (Tupinambis) merianae, whose static breathing mechanics was significantly affected by the removal of their post-hepatic septum (Klein, Abe & Perry, 2003). Similarly, a complete PPS in Testudinidae could alter the mechanics of the respiratory system by reducing the impact of the viscera onto the lungs, when compared to species with an incomplete PPS such as T. scripta.