Frogs seek hypoxic microhabitats that accentuate metabolic depression during dormancy

A number of animals enter a state of dormancy (e.g. aestivation) to tolerate seasonally arid environmental conditions. During aestivation, animals undergo a suite of physiological modifications to cope with the challenges associated with arid environments, including limited food and water availability (Pinder et al., 1992). Among the more striking physiological changes is the strong depression of metabolic rate to conserve limited endogenous energy reserves in the absence of external food sources (for reviews, see Guppy and Withers, 1999; Storey, 2002; Storey and Storey, 2012). Here, we use the term metabolic depression to describe a reduction in metabolic rate below the routine value. The extent to which different animals suppress metabolic rate during aestivation falls along a continuum, but a considerable metabolic depression is necessary for animals that remain dormant for many months (Guppy and Withers, 1999). The premature exhaustion of endogenous reserves can result in the atrophy of critical muscles and tissues (Hudson and Franklin, 2002; Mantle et al., 2009; Secor and Lignot, 2010) and the impairment of reproduction when animals resume activity (Pusey, 1990), and may be ultimately fatal (Horne, 1979; Etheridge, 1990). In the extreme, aestivating populations may be threatened with extirpation if all individuals completely exhaust endogenous reserves before the return of environmental conditions favourable for active life (van Beurden, 1980).

The extent of metabolic depression in aestivating ectotherms can be influenced by the microhabitat. For example, microhabitats with elevated temperatures can increase metabolic rate and, consequently, expedite substrate utilisation (Young et al., 2011), whereas microhabitat conditions that constrain aerobic metabolism may promote metabolic depression. Hypoxic conditions, for example, have been reported to depress metabolism in a number of ectothermic animals (e.g. Boutilier et al., 1997; Hicks and Wang, 1999; St Pierre et al., 2000). Hypoxic hypometabolism is accomplished by downregulating physiological processes involved in ATP turnover (Hochachka et al., 1996; Boutilier, 2001), thereby conserving energy and limiting the accumulation of toxic metabolic end-products (e.g. lactate) in animals during conditions of limited oxygen availability (Hochachka and Somero, 1984). For animals that aestivate for many months, hypoxic hypometabolism may be beneficial if it accentuates metabolic depression and considerably slows the rate of substrate utilisation.

Several animals aestivate within underground burrows, where they may encounter hypoxic conditions. Burrowing has been reported in annelids (Bayley et al., 2010), molluscs (Kotsakiozi et al., 2012; Osborne and Wright, 2018), fishes (Smith, 1931; Chew et al., 2004), amphibians (Ruibal and Hillman, 1981; Etheridge, 1990; Booth, 2006) and reptiles (Kennett and Christian, 1994) as a strategy to avoid desiccation, elevated temperatures and predation while in an aestivating state. Although burrows dug in sandy soils are typically well ventilated (Hanks and Thorp, 1956), heavy clay soils pose a significant barrier for gas exchange, and thus hypoxic conditions within the burrow can develop (Chew et al., 2004; Shams et al., 2005). The influence of hypoxic microhabitats on the metabolic rate of aestivating amphibians has not been thoroughly investigated. However, we have recently demonstrated that amphibious fish (Kryptolebias marmoratus) seek hypoxic microhabitats during prolonged air exposure that accentuate metabolic depression (Rossi and Wright, 2020). Do other animals also seek hypoxic microhabitats to enhance hypometabolism during dormancy?

Using the green-striped burrowing frog (Cyclorana alboguttata), we tested the hypothesis that aestivating animals select hypoxic microhabitats that accentuate metabolic depression during aestivation. During periods of prolonged drought, C. alboguttata burrow underground and aestivate within a cocoon of shed skin and mucus (Lee and Mercer, 1967). As C. alboguttata dig down into wet clay, they form a subterranean chamber, effectively becoming completely encased/entombed in the mud with no direct connection with the surface. Consequently, hypoxic conditions may develop in the underground cavities occupied by C. alboguttata during aestivation (Booth, 2006). The hypoxic habitat hypothesis predicts that frogs will preferentially occupy hypoxic rather than normoxic microhabitats in response to habitat drying. We placed frogs in custom-built experimental choice chambers that maintained an oxygen (O₂) gradient (hypoxic to normoxic) to determine the preferred partial pressure of O₂ (P_O2) of frogs in response to drying habitat conditions. The hypoxic habitat hypothesis also predicts that hypoxia will accentuate metabolic depression in aestivating C. alboguttata. Thus, we compared the rate of O₂ consumption in frogs maintained in either normoxia (control) or hypoxia during the transition from an active to an aestivating state.

Adult Cyclorana alboguttata (Günther 1867) (mean±s.e.m. body mass=20.8±0.7 g, mixed sex) were collected from wet roads in non-protected areas near Lake Broadwater in Dalby, Queensland, Australia, in December 2018 (n=21). Frogs were individually placed into large Ziploc plastic bags for transport to The University of Queensland, where they were maintained for several months prior to experimentation. Frogs were housed individually in either small (235×170×120 mm) or large (265×235×12 mm) well-ventilated clear plastic containers. Each housing container was lined with paper towels saturated with chemically aged water (dilution 1:4000; VitaPet, NSW, Australia), and contained a half PVC pipe for shelter. Frogs were fed vitamin-dusted crickets (Acheta domesticus) once per week, and the housing containers were cleaned weekly. The photoperiod was maintained on a 12 h:12 h light:dark cycle and room temperature was kept constant at 23°C. All animals were collected with approval of the Queensland Department of Environment and Heritage Protection (SPP WA0011256), and all experiments were carried out with the approval of The University of Queensland Animal Welfare Committee (SBS/502/18).

The P_O2 within natural C. alboguttata aestivation cavities has never been measured. The underground chambers are extremely difficult to locate from above ground because an air tunnel conduit is not maintained. Thus, we collected clay soil from the C. alboguttata collection site to create artificial chambers in which P_O2 could be measured. We buried three hollow, perforated rubber balls (95 mm diameter) at approximately 10 cm depth in individual soil-filled buckets (6 litres) to simulate the underground cavity in which frogs reside during aestivation (Fig. S1). We maintained access to the aestivation cavity through a plastic tube (27 mm diameter) that protruded from the soil. The soil in each container was saturated with water to promote microbial activity as would be present when C. alboguttata excavate their burrows. We monitored P_O2 within the artificial aestivation cavity over a 2 week period (between 10:00 and 14:00 h) using an O₂-sensing optode and Presens Optical Oxygen Sensor (Precision Sensing, Regensburg, Germany) inserted down the plastic tube and into the perforated ball. The plastic tube was closed to the external environment when O₂ measurements were not being taken. The artificial burrows remained outdoors during the 2 week period, where temperatures ranged from 14°C (night) to 27°C (day), and were typical of temperatures that C. alboguttata would experience during aestivation.

To determine whether C. alboguttata select hypoxic microhabitats in response to drying habitat conditions, we built custom experimental choice chambers (750×240×120 mm) through which frogs could move freely (Fig. 1A). The choice chambers were lined with paper towels saturated with aged water. We individually placed frogs in the centre of an experimental choice chamber and then generated an O₂ gradient (10.9±0.1 to 18.8±0.1 kPa; Fig. 1) by introducing a gentle stream of air at one end of the chamber, and a gentle stream of N₂ at the other. We periodically monitored the O₂ levels along the gradient during each behavioural trial using four evenly spaced O₂-sensing optodes glued to the choice chamber, and a Presens Optical Oxygen Sensor. The O₂ gradient remained stable over time (Fig. S2). The gradient direction (low to high O₂) was reversed (high to low O₂) between trials. Frogs remained in the experimental choice chambers for 24 h, during which time the paper towel dried gradually. We continuously video-recorded frogs during the 24 h experimental period using infrared surveillance cameras (Eonboom Electronics Limited, and K Guard Security, New Taipei City, Taiwan), and a 16-channel H.264 Digital Video Recorder (DVR) system. We chose a 24 h experimental period because preliminary behavioural trials lasting 48 h revealed that frogs were relatively immobile after 24 h. We conducted all O₂ choice trials in the dark to simulate the burrowing conditions of C. alboguttata in the wild. All monitoring of the O₂ gradient was performed under red light to minimise disturbance to the frogs. All frogs were fasted for at least 48 h prior to choice experiments.

For analysis, we binned the choice chambers into three O₂ zones of equal size: the ‘low O₂ zone’ (10.9±0.1 to 15.1±0.1 kPa), the ‘medium O₂ zone’ (15.1±0.1 to 17.4±0.1 kPa) and the ‘high O₂ zone’ (17.4±0.1 to 18.8±0.1 kPa). We calculated the proportion of time frogs spent in each O₂ zone after they had sampled every zone. We also recorded the first O₂ zone in which frogs remained immobile for at least 1 h and assumed a water conserving posture (Fig. S3) as a proxy for a suitable burrowing location. The 1 h time period was chosen because it takes C. alboguttata approximately 1 h to burrow in wet clay substrates (Booth, 2006). One frog did not sample all O₂ zones, and was consequently excluded from all behavioural analyses. All animals were weighed before and after each experimental trial to determine whether any dehydration occurred owing to habitat drying.

Following the O₂ choice experiments, we randomly assigned frogs to one of two 7 week aestivation treatments: normoxia (control; n=9) or hypoxia (n=10). All frogs were fasted for 72 h in their housing containers, then individually placed in a 500 ml respirometry chamber lined with two 15 cm² pieces of paper towel saturated with aged water. The normoxia (21.0 kPa) and hypoxia (10.5 kPa) exposures were accomplished by continuously flushing the respirometry chambers with a gentle stream of humidified air or air–N₂ mixture. We verified the O₂ levels within the respirometry chambers periodically throughout the 7 week experimental period using O₂-sensing optodes glued to the inside of each respirometry chamber, and a Presens Optical Oxygen Sensor. Aestivation was induced by allowing the paper towel to dry gradually, as well as maintaining frogs in the dark throughout the experimental period (24°C) (Flanigan et al., 1991).

The mass-specific rate of O₂ consumption in C. alboguttata was measured weekly using closed-system respirometry. Frogs were given 24 h to acclimate to their respirometry chamber and respective O₂ exposure before the week 0 measurements. The O₂ consumption measurements were made by sealing respirometry chambers for a 2–8 h period and measuring the decline in P_O2 (longer measurement periods were required as the rate of O₂ consumption declined throughout the experimental period). Week 0 measurements consisted of three trials (repeated measures) to establish a mean ‘resting’ O₂ consumption rate for each frog. In order to minimise disturbance during the onset of aestivation, only one O₂ consumption measurement was made per animal in the remaining weeks as previously described (Young et al., 2011). All O₂ consumption measurements were performed in the dark under red light to minimise disturbance to the frogs. When necessary, 1 ml of aged water was discreetly added to the respirometry chambers to prevent complete desiccation. Two control frogs died during the 7 week experimental period and were thus excluded from the analysis. One hypoxia-exposed frog was also excluded because it did not assume the water conserving posture by the end of the experimental period. Frogs were weighed before and after the experimental period to assess hydration status.

We determined the preferred P_O2 of C. alboguttata using methods for compositional data, because the proportion of time frogs spend in any one O₂ zone is dependent upon the time spent in other O₂ zones. All proportions were isometric log-ratio (ILR) transformed and then analysed using an ordinary least squares (OLS) regression. We subsequently back-transformed the coefficients for meaningful interpretation of the results (van den Boogaart and Tolosana-Delgado, 2013). We performed a chi-squared goodness-of-fit test to determine which O₂ zone C. alboguttata selected as the first suitable ‘burrowing’ location in response to habitat drying. A two-way repeated-measures ANOVA with a Greenhouse–Geisser correction was used to compare the rate of O₂ consumption in normoxia- and hypoxia-exposed frogs over time. We performed a Dunnett's many-to-one comparison test to determine when the rate of O₂ consumption within each P_O2 treatment group differed from that of active frogs (all reported P-values are adjusted for multiple comparisons). Paired t-tests were used to compare the body mass of C. alboguttata before and after use in the behavioural and O₂ consumption experiments. Prior to analysis, we assessed all data for normality of residuals and homogeneity of variance. All results were considered significant at P<0.05.

The P_O2 in the artificial aestivation cavities varied considerably across our three replicates, and over time. We found that the P_O2 within the aestivation chambers became hypoxic (P_O2 as low as 8.9 kPa) while the clay soil was saturated with water (Fig. 2). As the soil began to dry and crack several days later, the P_O2 within the aestivation cavities returned to relatively normoxic levels.

The proportion of time C. alboguttata spent in each O₂ zone was not equally distributed. We found that frogs spent 57.4±6.9% of the time in the low O₂ zone, and only 23.9±5.6% and 18.7±5.2% in the medium and high O₂ zones, respectively (OLS: P<0.001; Fig. 3A). Cyclorana alboguttata did not seek any one zone as their preferred ‘burrowing’ location (chi-squared: P=0.12), although 55% of frogs selected the low O₂ zone as their first suitable ‘burrowing’ location, and only 20% and 25% selected the medium and high O₂ zones, respectively. The body mass of C. alboguttata after the O₂ choice experiments was 7% lower than that before (t-test: P<0.001).

The mass-specific rate of O₂ consumption in C. alboguttata was significantly influenced by P_O2 exposure and time (two-way ANOVA: P=0.002, P<0.001; Fig. 4). On average, the O₂ consumption rate in hypoxia-exposed frogs was 27.7±0.1% lower than that of normoxic (control) frogs over the 7 weeks. The rate of O₂ consumption in hypoxia-exposed frogs during weeks 2–7 was significantly different from that of week 0 (Dunnett's: P<0.01). In contrast, it took longer for metabolic depression to occur in normoxic frogs (weeks 4–7 were different from week 0; Dunnett's: P<0.05). By the end of the experimental period (week 7), hypoxia-exposed frogs exhibited a 59% reduction in the rate of O₂ consumption from active (week 0) counterparts (from 41.4±4.6 to 21.1±1.8 µl O₂ g⁻¹ h⁻¹; Dunnett's: P<0.001), whereas control frogs exhibited a 51% reduction (from 34.2±1.8 to 14.2±1.1 µl O₂ g⁻¹ h⁻¹; Dunnett's: P>0.001). We found no change in the body mass of C. alboguttata before (22.0±0.8 g) and after (22.2±1.0 g) the 7 week experimental period (t-test; P=0.71).

We used C. alboguttata to test the hypothesis that animals seek hypoxic microhabitats that accentuate metabolic depression during dormancy. Indeed, we found that C. alboguttata spent a significantly greater proportion of time in the low O₂ zone compared with the medium and high O₂ zones in response to drying habitat conditions. We also found that hypoxic microhabitats had a significant influence on the rate of O₂ consumption in C. alboguttata during the transition from an active to an aestivating state. Hypoxia exposure accelerated the onset of metabolic depression in C. alboguttata by 2 weeks, and resulted in a more profound metabolic depression than could be achieved under normoxic conditions. Taken together, our findings suggest that some animals may seek microhabitats that maximally depress metabolic rate during aestivation, and that microhabitat O₂ availability can have significant implications for energy metabolism.

Cyclorana alboguttata chose to spend more time in hypoxia compared with normoxia in response to drying habitat conditions, and one possible benefit is metabolic depression. Our laboratory findings are consistent with the observed habitat preference of C. alboguttata in the wild, as frogs have only been found burrowed in heavy clay soils despite the presence of sandy soils within their distribution range (Lee and Mercer, 1967; Booth, 2006). Interestingly, a previous laboratory study reported that C. alboguttata preferentially burrowed in wet sand rather than wet clay, likely because of the reduced energetic cost of burrowing in friable substrates (Booth, 2006). However, in the Booth (2006) study, frogs were not provided with hypoxic soils, which may affect their burrowing behaviour. In the wild, burrowing in clay soils may also provide aestivating frogs with benefits beyond the presence of environmental hypoxia. For example, clay soils dry much slower than sandy soils because of the smaller spaces between soil particles (Hubble, 1984), which would delay the onset of desiccating conditions during aestivation. Similarly, the soils from which C. alboguttata were collected smell of ‘rotten eggs’ (C. E. Franklin, R. L. Cramp, personal observations), suggesting that hydrogen sulphide (H₂S) may be present. Because H₂S is a potent inhibitor of aerobic metabolism (Smith et al., 1977), it may also promote metabolic depression in aestivating frogs. Numerous aestivating animals spend the vast majority of their life in subterranean burrows (Shoemaker, 1988; Abe, 1995), yet the factors that contribute to microhabitat selection remain poorly understood.

We showed that hypoxic microhabitats can have a significant influence on the metabolic rate of C. alboguttata entering aestivation. Hypoxia exposure accelerated the onset of metabolic depression in C. alboguttata by 2 weeks. Although C. alboguttata may aestivate for several months of the year (Withers and Richards, 1995), any reduction in metabolic rate during the first weeks of aestivation – when O₂ consumption rates are highest – may result in considerable energy savings. Environmental hypoxia has previously been shown to accelerate metabolic depression. For example, partially submerged freshwater turtles (Pseudemys scripta) with access to hypoxic air depressed metabolic rate five times faster than those with access to normoxic air (Jackson and Schmidt-Nielsen, 1966). Surprisingly, very few studies have examined how the rate at which metabolic depression is achieved impacts the endogenous fuel stores of dormant animals, but this is a fascinating area for future investigation.

where L is the mass of the lipid body (g), M is the total mass of the frog (g) and V·_O2 is the O₂ consumption rate of the frog (μl O₂ g^–1 h^–1). Based on the assumption of Klieber (1961), the only substrate utilized was lipid, and 2.02 l of oxygen is required to oxidise 1 g of lipid. Remarkably, C. alboguttata aestivating in normoxia can depress metabolic rate to less than 20% of the normal resting value after 10 weeks, thereby extending survival time to several years (Kayes et al., 2009). It is unknown whether there is a limit to the extent frogs can depress metabolic rate during aestivation. Regardless, hypoxia exposure significantly accelerated and accentuated metabolic depression in C. alboguttata and may therefore delay the exhaustion of critical endogenous energy reserves during periods of prolonged drought.

The body mass of C. alboguttata was differentially affected by our experimental procedures. On the one hand, frogs exhibited a 7% reduction in body mass after the 24 h behavioural trials, probably owing to loss of body water in the rapidly drying choice chambers. Although many frogs can maintain water balance during aestivation, even in desiccating soils (e.g. Neobatrachus aquilonius; Cartledge et al., 2006), complete substrate drying is unlikely to occur in natural aestivation cavities over a 24 h period. On the other hand, these same frogs exhibited no further change in body mass following the 7 week O₂ consumption experiment. Although the respirometry chambers dried gradually over several weeks, we prevented complete desiccation by periodically adding 1 ml of water to each chamber, which may have helped frogs to maintain hydration. Furthermore, C. alboguttata can absorb and store up to 20% of their body mass in water as dilute urine in preparation for aestivation (Booth, 2006), and the closely related water-holding frog (Cyclorana platycephalus) can similarly store more than 50% of their body mass in water (van Beurden, 1984). The gradual drying in the respirometry chambers may have allowed for water uptake and storage that was not possible in the choice chambers, which dried far more rapidly.

The ecological significance of our study is reinforced by the empirical evidence of hypoxia in the aestivation substrates of C. alboguttata. Although frogs were not present within the artificial aestivation cavities, we showed that the P_O2 declined to hypoxic levels (P_O2 as low as 8.9 kPa) in the days following soil saturation. Soils often become hypoxic when saturated with water owing to rapid O₂ consumption by soil microbes (Drew, 1992), and because the presence of water between soil particles creates a significant diffusion barrier for O₂ (Arieli, 1979; Maclean, 1981). The presence of C. alboguttata within natural aestivation cavities would likely exacerbate the hypoxia because of their respiratory processes. Consequently, the P_O2 levels we measured in the artificial aestivation cavities were likely an overestimate of those experienced by C. alboguttata in the wild. Furthermore, the rate of drying in the field would be conceivably much slower than in our artificial aestivation cavities, suggesting that hypoxia could persist for longer than observed in our artificial cavities. A previous study has suggested that the clay soils inhabited by C. alboguttata during aestivation in the wild can remain wet for several months following heavy rains (Booth, 2006). However, as wet clay soils dry and crack, re-oxygenation can occur. We found that the P_O2 within some of the aestivation cavities returned to relatively normoxic levels, particularly when cracks started to form in the drying substrate. Overall, we suggest that C. alboguttata likely experience several months of hypoxia during aestivation, which ultimately may help frogs to depress metabolism and economise on endogenous reserves.

Hypoxia avoidance behaviour has been reported in virtually all major animal phyla, yet we demonstrated here hypoxia-seeking behaviour in C. alboguttata. At the proximate level, seeking hypoxic microhabitats during dormancy may ensure that endogenous reserves do not become limiting to an individual's survival. Although uncommon, there are some reports of hypoxia-seeking behaviour in other animals. For example, the ocean quahog (Artica islandica) burrows into hypoxic substrates during winter months when food is scarce as an energy-saving strategy (Strahl et al., 2011). Is strategic microhabitat selection widespread among aestivating and hibernating taxa? Exploring the microhabitat selection strategies of dormant animals remains an exciting avenue for future investigation.

Moving from proximate to ultimate questions, does the persistence of dormant populations over evolutionary time depend on microhabitat conditions that help animals economise on endogenous reserves? If so, then climatic changes (e.g. chronically elevated temperatures) that result in higher metabolic rates and, consequently, faster rates of substrate utilisation, may necessitate strategic microhabitat selection. Overall, we suggest that microhabitat selection strategies are at least as important as the physiological mechanisms used by dormant animals to tolerate environmental extremes. Moreover, the microhabitats animals occupy during dormancy may modulate, in part, their ecological and evolutionary success.

We thank Ed Meyer for animal collection and Argelia Rodríguez-Contreras, Nicholas Wu and Niclas Lundsgaard for animal care. We also thank the reviewers for helpful commentary that improved the manuscript.