The role of ambient temperature and body mass on body temperature, standard metabolic rate and evaporative water loss in southern African anurans of different habitat specialisation

Study species

We chose three sympatric anuran species with different habitat specialisations, based on their modes of egg deposition and development (Mercurio, Böhme & Streit, 2009). The pipid, X. laevis, inhabits and breeds in permanent water bodies. This species is usually referred to as permanently aquatic because it possesses several key physiological adaptations (e.g. lateral line system, webbed hind feet etc.), suitable for an aquatic life style (see Measey, 2004). However, evidence also suggests that under severe drying conditions of permanent ponds, X. laevis frequently disperse overland (Measey, 2016). A. delalandii (family: Pyxicephalidae—previously known as A. angolensis and A. quecketti) breeds in and inhabits streams and flowing rivers. Adults are usually encountered on the water edge and on rocks along streams but are seldom encountered away from water bodies (Channing, 2004). S. capensis (previously known as Bufo rangeri and later Amietophrynus rangeri) is a member of the toad family Bufonidae. These toads generally breed in shallow, temporary water bodies and adult toads are adapted to a terrestrial mode of life (Cunningham, 2004).

Adult individuals of our three target species were caught around the area of Port Elizabeth, South Africa (Table 1). Experiments were undertaken under the animal research ethics clearance permit number A13-SCI-ZOO-007 issued by the Research Ethics Committee (Animal) at the Nelson Mandela University. Animals were collected under permit number CRO41/14CR, issued by the Department of Economic Development, Environmental Affairs and Tourism, Eastern Cape Province. We caught between 20 and 30 individuals (of varying but overlapping interspecific body size classes) excluding juveniles and gravid females. We attempted to maintain M_b comparable among species as much as possible, although we found this difficult for A. delalandii. We included both male and female individuals to account for variation that might be a result of sexual dimorphism. Sexual dimorphism is particularly pronounced in anurans, with females generally larger than males (see Table 1 for mass differences, with X. laevis showing the largest differences between males and females). Animals were kept for a maximum period of 2–4 months in the lab under environmentally enriched conditions (see below), until all experiments were completed. After completing the experiments, animals were returned to their respective sites of capture.

Amietia delalandii were maintained in 110 L plastic boxes, with sand and small logs for cover, at low densities (five individuals per box). S. capensis were kept in terraria made from paddling pools (d × h: 2.16 × 0.45 m) with sand, water and small logs and bark to provide cover. X. laevis were kept in a freshwater tank (l × b × h: 3.55 × 0.9 × 0.63 m) with stacked-bricks and stones to provide adequate cover and fed a diet of ox-heart. Both A. delalandii and S. capensis were fed mealworms and crickets, dusted with calcium (ReptiCalcium; Zoo Med Laboratories Inc, San Luis Obispo, CA, USA). All species had food available ad libitum, and the holding rooms were maintained at 20 °C, on a 12:12 photoperiod. Atmospheric air was circulated to maintain a constant temperature in the holding rooms, thus ambient humidity levels fluctuated with outside air. Following Hillman et al. (2009), feeding of individuals ceased 3 days prior to experiments to ensure that individuals were post-absorptive during experimental sessions. Water was provided throughout the duration of the study (including periods when food was withheld) to prevent dehydration stress. We further ensured that the terraria in which we held S. capensis and A. delalandii, were sprayed with water every 3 days to dampen the sand and we provided a bowl with water daily where amphibians could rehydrate. Mass was recorded to the nearest ±0.01 g before and after each experimental run.

Gas exchange measurements

Standard metabolic rates and EWL measurements were conducted on the three anuran species following Lighton (2008) and Steyermark et al. (2005), also see Gomes et al. (2004) for a range of respirometry methods used, particularly for ectotherms. We used an open-flow respirometry system operated on a push through mechanism on post-absorptive, non-reproductive individuals, at rest (Sinclair et al., 2013). Experiments were conducted at temperatures ranging from 5 to 35 °C at 5 °C intervals (Dunlap, 1971). The order of experimental temperature runs was randomised to reduce the effects of experimental acclimation to any directional shift in temperature. Eight individuals of each species (four males: four females) were randomly selected for trials at each temperature and a single trial was conducted on each individual per temperature.

Prior to each experimental session, we recorded T_b (using a Fluke 80PK-1 probe, Type K thermocouple −40 to 260 °C, to the nearest 0.1 °C), then patted frogs dry to remove excess water from the skin and we then recorded M_b to the nearest 0.01 g. All experiments were conducted between 07:00 and 18:00 h, always within the light cycle of the 12:12 photoperiod, when animals were less likely to be active (Gomes et al., 2004). Frogs were placed individually in suitably-sized air tight glass metabolic chambers of three sizes: 341 mL for small-, 476 mL for medium and 978 mL for large frogs, depending on the size of the individual. A similar approach was followed by Young et al. (2005) in order to minimise large amounts of unoccupied spaces inside the metabolic chamber (see also Gatten, 1987; Gomes et al., 2004). We found that individual frogs were agitated when they experienced respirometry procedures for the first time. Prior to the respirometry experiments, we performed a training session on each individual frog by placing them inside metabolic chambers for 15 min at 20 °C. Frogs were noticeably quiescent during subsequent respirometry runs.

A 0.5 cm layer of mineral oil was added to each chamber to prevent evaporation of excreted materials. Inside the chamber, a frog was placed on a plastic mesh platform (with sufficiently large holes for faeces to fall through), suspended at least two cm above the oil layer (Smit & McKechnie, 2010). Air temperatures inside the metabolic chamber were recorded using a thermocouple probe (Fluke 54IIB; Fluke Corporation, Everett, Washington, D.C., USA) that was inserted inside the chamber. Once the animal was placed in the chamber, we used ‘cling-wrap’ (GLAD, Johannesburg, South Africa) before sealing with a glass lid. After placing the lid, we placed Prestik, ‘Blu-Tac’ type material (Bostik, Cape Town, South Africa) around the lid of the metabolic chamber to minimise air leaks. Two metabolic chambers, one containing an animal, and the other an empty reference chamber (serving as a chamber to determine baseline levels) were placed in a custom-made environmental chamber made from a 100 L cooler box with the interior lined with copper tubing. Baseline levels were recorded for 30 min before each trial (Smit & McKechnie, 2010; Van de Ven, Mzilikazi & McKechnie, 2013). The temperature inside the cooler box was controlled by pumping temperature-controlled water through the copper tubing using a circulating water bath (FRB22D; Lasec, Cape Town, South Africa; see Van de Ven, Mzilikazi & McKechnie, 2013). A small fan was used to ensure air circulation inside the environmental chamber.

During SMR and EWL measurements, we used Mass Flow System pumps (MFS-2; Sable Systems, Las Vegas, NV, USA) to pump atmospheric air scrubbed of water vapour (using a Drierite column (98% CaSO₄, 2% CoCl₂, Sigma-Aldrich, Darmstadt, Germany) at a flow-rates of 100–600 mL min⁻¹, through the metabolic chambers. We calibrated the MFS-2 pumps using a flow-bubble meter (calibrated flow-rates were used in subsequent equations, see below). We scrubbed the air of water vapour to have better control of ambient humidity levels in the respirometry chamber (actual vapour pressures recorded for atmospheric air in Port Elizabeth varied greatly among days during our study period). Air from the metabolic chambers was sequentially sub-sampled, using Subsampler (SS3; Sable Systems, Las Vegas, NV, USA) and a Multiplexer (V3; Sable Systems, Las Vegas, NV, USA) was programmed though Expedata (Sable Systems, Las Vegas, NV, USA) to record gas concentrations for each chamber at 20-min intervals, recording an air sample every second. Subsampled air was first pulled through a water vapour analyser (RH-300; Sable Systems, Las Vegas, NV, USA) to measure water vapour pressure (WVP). We were mainly interested in measuring total EWL, not different components of water loss such as boundary layer and cutaneous resistance; hence we did not use agar models in our approach (Buttemer, 1990). Air samples then passed through a carbon dioxide analyser (Ca-10a; Sable Systems, Las Vegas, NV, USA) and finally through to an oxygen analyser (Fc-10a; Sable Systems, Las Vegas, NV, USA). The behaviour of the frogs was monitored during the trials using live video feed for the duration of the trial. The thermocouple probe was used to measure T_air (air temperature inside the metabolic chamber). During a trial, frogs experienced one controlled temperature at a time. A trial was considered completed when the water vapour pressure and temperature trace was stable for 20 min or if the animal appeared too distressed to continue with measurements. Although we acknowledge that different sized frogs show differences in cooling rates (Wygoda, 1988b, 1989), we assume that the final 20 min interval of stable vapour pressure and temperature trace suggest that at this point, each individuals’ T_b (irrespective of size) had reached equilibrium with the desired test temperature. Trials did not last longer than 2 h (following Dunlap, 1971). After each trial, we removed the frog from the chamber, recorded cloacal temperature (T_b) within 30 s of removal (using a Fluke 80PK-1 probe, Type K thermocouple −40 to 260 °C) and the final M_b. Species lost on average between 3% and 5% of their initial M_b (see Table 1). This loss in M_b was comparable to other studies looking at water loss and metabolic rates in anurans (see Steyermark et al., 2005). In this study, it was concluded that the average loss of 5% in M_b suggested that the animals remained adequately hydrated throughout the experimental trial (Steyermark et al., 2005). After each trial, frogs were individually placed in temporary holding facilities. X. laevis were kept in 20 L buckets, half-filled with water at 20 °C and fed ox heart. Both S. capensis and A. delalandii were kept in a 0.5 L plastic container lined with a wet lab paper and were fed mealworms and crickets, respectively. After each trial, individuals were eligible for selection for another trial run only after 3 days.

Data extraction

Once all experiments were complete, we used Expedata software to extract oxygen, carbon dioxide and WVP traces from data files. We selected the most stable 20 min trace in each run, when the animal was at rest. Standard metabolic rates can either be calculated using the rate of oxygen consumption or carbon dioxide production rates (Gatten, Miller & Full, 1992; Lighton, 2008; Withers, 2001). We used rates of oxygen consumption (VO₂) at each temperature following Gomes et al. (2004). Moreover, flow rate was calibrated using a flow-bubble meter. To determine the rates of EWL, we converted rates of WVP to water vapour density, which we subsequently converted to rates of EWL (see Lighton, 2008). Lastly, we calculated saturation WVP at each test temperature to determine vapour pressure deficit (VPD) as recent evidence suggests that VPD directly drives rates of EWL in amphibians (Riddell & Sears, 2015). We calculated VPD by estimating the saturation vapour pressure at each air temperature we studied from using the equations in Campbell & Norman (1998). We converted the absolute WVP in the animal chamber to kPa and calculated the VPD as the difference between saturation vapour pressure and absolute WVP. However, our experimental design did not allow us to test for VPD directly (e.g. scrubbing water from incurrent air) but we tested if VPD was a better predictor of EWL in our test species (see Table S1). We found that VPD was a better predictor of EWL in both A. delalandii and S. capensis but not in X. laevis.

Statistical analysis

To determine the effect of T_a on T_b, EWL and SMR of anurans, we used linear mixed-effect models with the R package ‘nlme’ (Pinheiro et al., 2014) due to repeated measurements on individuals. We also included M_b and sex as fixed effects in models. Within ectotherm, T_b is highly correlated to T_a and we thus ran a repeated measures correlation using the R package ‘rmcorr’ (Bakdash & Marusich, 2017a, 2017b). We found a strong, significant positive correlation between T_a and T_b (r₍₉₆₎ = 0.9917; 95% CI [0.988–0.994], P < 0.001) for all three species (see Fig. 1). We included sex in the model because of the pronounced sexual dimorphism in anurans. Furthermore, we wanted to determine if species differed in their responses to changes in T_a. However, this analysis was restricted to species with comparable M_b (i.e. S. capensis and X. laevis), because it is known to be a significant contributor to observed physiological difference. We also ran a repeated measure correlation between flow rate and EWL to determine if the variation in flow rate had any effect on EWL. We found a non-significant relationship (r₍₉₇₎ = 0.157, 95% CI [−0.044 to 0.345], P = 0.122) between the two variables and thus did not include flow rate as a fixed variable in subsequent models. We determined the proportion of the variance explained by the model (coefficient of determination R² for mixed models following a procedure by Nakagawa & Schielzeth (2013) and Nakagawa, Johnson & Schielzeth (2017). Furthermore, we also determined the relative importance of each of the fixed variables in the model using the semi-partial R² as a measure of effect size (see Jaeger et al., 2017) using the ‘r2glmm’ package, implemented in R (Jaeger & R Core Team, 2017). All analysis were undertaken in R 3.4.2 (R Development Core Team, 2017).

African clawed frog: Xenopus laevis

We found that T_b increased with an increase in T_a (F_(6,23) 1,050.629; P < 0.0001), although it did not affect sex (F_(1,23) 3.440; P = 0.0765) and M_b (F_(1,23) 0.238; P = 0.631; see Table S2 for semi-partial R² values). Whole-animal EWL increased at high T_a (F_(1,22) = 16.201; P < 0.0001; see also Table S2 for semi-partial R² values on the different temperature levels). We did not find any significant effect of EWL on sex (F_(1,23) 3.056; P = 0.094) and M_b (F_(1,23) 0.198; P = 0.661). Furthermore, we found that whole-animal SMR increased with an increase in T_a (F_{(6, 23)} = 10.659; P < 0.0001), although we did not find any significant difference among sex (F_(1,23) 3.354; P = 0.080) and M_b (F_(1,23) 0.007; P = 0.934).

Common river frog: Amietia delalandii

We found that T_b increased with an increase in T_a (F_(6,24) = 1,374.112; P < 0.0001). Furthermore, M_b had a significant positive effect on T_b (F_(1,24) = 8.196; P < 0.01; semi-partial R² = 0.160), such that T_b matched T_a closer in larger individuals while smaller individuals maintained T_b below T_a. However, we did not find a significant effect of sex (F_(1,24) = 0.748; P = 0.396) on T_b. Whole-animal EWL (F_(6,24) = 6.612; P < 0.001, Fig. 2) and SMR (F_(6,24) = 5.711; P < 0.0001; Fig. 3) increased with an increase in T_a. We did not find a significant effect of M_b in both whole-animal EWL (F_(1,24) = 1.341; P = 0.258) and whole-animal SMR (F_(1,24) = 2.372; P = 0.137), respectively (see also Table S3). Moreover, sex did not influence whole-animal EWL (F_(1,24) = 0.205; P = 0.655) and SMR (F_(1,24) = 0.497; P = 0.490).

Raucous toad: Sclerophrys capensis

Sclerophrys capensis T_b was positively correlated with T_a (F_(1,28) = 420.726; P < 0.0001; Fig. 1). We also found that M_b had a significant positive effect on T_b (F_(1,28) = 6.045; P < 0.05, semi-partial R² = 0.099) such that T_b matched T_a in larger individuals while smaller individuals generally maintained T_b below T_a, although sex did not have an effect on T_b (F_(1,19) = 0.020; P = 0.890, semi-partial R² = 0.036). Furthermore, we found that an increase in T_a lead to a significant increase in whole-animal EWL (F _(6,28) = 15.055; P < 0.0001; Fig. 3), although M_b (F_(1,28) = 3.708; P = 0.064) and sex (F_(1,19) = 1.190; P = 0.289) did influence whole-animal EWL (see Table S4 for semi-partial R²). Whole-animal SMR also increased at high T_a (F_(6,28) = 11.639; P < 0.0001; Fig. 3). In addition, we found that M_b had a positive significant effect on whole-animal SMR (F_(1,28) = 5.184; P < 0.05, semi-partial R² = 0.086), although we found a non-significant sex effect (F_(1,19) = 0.228; P = 0.638).

Body temperature

The concept of thermal inertia suggests that although larger individuals take longer to warm up, they also take longer to cool down (Carey, 1978). A study looking at how M_b affects T_b in different toad species concluded that larger individuals had higher thermal inertia than small sized individuals (Carey, 1978). In addition, Newman & Dunham (1994) found that Scaphiopus couchii toadlets that metamorphosed at larger sizes took longer to reach critical dehydration levels compared to small-sized toadlets. Collectively, these results highlight the importance of M_b as a significant factor influencing key physiological processes such as T_b through its effect on rates of heating and cooling (Wygoda, 1988a) particularly in vertebrate ectotherms. In this study, we found that changes in M_b had a significant effect of T_b in A. delalandii and S. capensis. These results suggest that the effect of M_b on heat flux may be more beneficial for species spending a large proportion of their time on land, where changes in temperature are more pronounced and may be sudden.

Our results seem to suggest that although M_b is a key factor affecting T_b in amphibians, this may not be the case across a broad range of available environmental temperatures and species ecologies. Furthermore, we found no significant relationship between M_b and T_b in X. laevis despite the species showing the largest difference in sexual dimorphism (see Table 1). Wygoda (1988a) suggested that prolonged cooling may be adaptive as it assists anurans to maintain a higher T_b that is essential for performance under decreasing temperature conditions. Hence amphibians are expected to have larger body sizes in more temperate, cooler environments (Ashton, 2002) and with altitude (Measey & Van Dongen, 2006), although the generality of this assertion has been challenged, particularly in largely aquatic urodeles (Olalla-Tárraga & Rodríguez, 2007) and in anurans (Gouveia et al., 2019).

Evaporative water loss

We found that whole-animal EWL increased with an increase in T_a for all our species. We expected this result because amphibians use EWL to reduce heat gain, and water loss increases with an increase in metabolic rates (Hillman et al., 2009). Although our study was not designed to specifically test for the effect of VPD on EWL, we did show that T_a was strongly related to EWL while accounting for VPD, suggesting that elevations in EWL play a role in evaporative cooling in two of the three species considered (see Table S1). We also wanted to determine whether different species differ in their ability to regulate water loss (Wygoda, 1984; Young et al., 2005). Both Wygoda (1984) and Young et al. (2005) concluded that species that adopt an arboreal lifestyle show significantly reduced levels of water loss. In addition to arboreal lifestyle, Tracy, Christian & Tracy (2010) concluded that both high cutaneous resistance to water loss (R_c), and larger M_b were important in reducing desiccation time in amphibians, although high R_c may not be beneficial to large-bodied frogs for thermoregulation (due to possibility of overheating at high T_a). In this study however, M_b and sex did not have an influence on rates of whole-animal EWL in the three-species considered. While studying the relationship between desiccation tolerance and body size, Schmid (1965), found no relationship between the two variables despite suggestion that small sized frogs’ loose water more readily as compared to larger ones (Heatwole et al., 1969). Furthermore, there seems to be a difference in the onset of evaporative cooling such that species with high R_c or atypical frog species only employing EWL at high T_a compared low and moderate R_c or typical frogs (Tracy et al., 2008). Species with high R_c are reported to have reduced EWL and have been observed to increase their T_b above ambient as an adaptation to terrestrial habitats (Buttemer, 1990; Tracy & Christian, 2005).

Tracy & Christian (2005) concluded that species with low EWL had low variance in T_b as a result of a negligible influence of evaporative cooling on thermoregulation, although this may not be beneficial, particularly at high T_a because reduced skin resistance at high T_a shortens desiccation time (Tracy et al., 2008). Coincidentally, species with the smallest variance in T_b were also some of the largest (see Tracy & Christian, 2005). In the present study, we found that S. capensis had the smallest variance in T_b between 25 and 35 °C (±4 °C as opposed to approximately 7 °C for both A. delalandii and X. laevis; see Fig. 1). Perhaps, maintaining small variations in T_b particularly at high T_a is beneficial for life on land. Despite T_b and T_a being strongly correlated, we found that T_b was typically slightly lower than T_a for all our species (4–7 °C below T_a, depending on the species), particularly at high T_a suggesting that all species were using evaporative cooling at high T_a. Indeed, while studying toads, Tracy (1978) postulated that toads possess an ability to withstand higher T_b for longer periods, provided that the skin remains moist. Although the ability to use evaporative cooling while managing water loss is important for amphibians, other factors such as the ability to rehydrate quickly and absorb water from a variety of substrates (e.g. burrowing species) may have been as important for amphibians to occupy such a variety of terrestrial habitats through the course of evolution (Cartledge et al., 2006; Prates & Navas, 2009), as well as being of significance to current invasions (Vimercati, Davies & Measey, 2018).

Standard metabolic rates

Whole-animal SMR increases with an increase in T_a as a result of the increase in kinetic energy and reaction rate required at high temperatures (Brown et al., 2004; Clarke, 2006; Gillooly et al., 2001). We expected S. capensis to have comparatively low rates of SMR (particularly at high T_a), as an adaptation to terrestrial life because: (i) actively searching for food (insect prey) is coupled with water loss in terrestrial habitats and terrestrial specialists should adopt a sit-and weight foraging strategy, although see Pough & Taigen (1990), or (ii) only be active nocturnally or at low to intermediate levels of T_a to reduce rates of EWL (Peterman & Semlitsch, 2014). Indeed, we found that S. capensis had significantly lower rates of metabolism at 15, 25 and 35 °C. Furthermore, we found whole-animal SMR increased with an increase in M_b only in S. capensis. During species comparisons, we also found that increases in M_b led to an increase in whole-animal EWL and SMR only at the highest tested T_a (35 °C). This result is particularly interesting, suggesting that in Africa’s temperate South, large bodies confer an advantage in delaying warming rates.

Evidence on the influence of ecology on SMR has not been clearly articulated in the literature as it may have been for EWL. Where there has been support, this has largely focused on comparing cold adapted and warm adapted species (Clarke & Johnston, 1999; Addo-Bediako, Chown & Gaston, 2002). While looking at insects at a global scale, Addo-Bediako, Chown & Gaston (2002) found support for the metabolic cold adaptation hypothesis. Furthermore, Reinhold (1999) found that flying and highly vocal insect species had significantly high SMR than non-flying and less vocal insect species, respectively (but see also Hodkinson, 2003). However, Clarke & Johnston (1999) did not find evidence to support the hypothesis on teleost fish. While studying different but closely related anuran species, Navas (1996) found that high elevation species had both high metabolic rates and aerobic scope compared to their low elevation congeners. Lastly, Navas (1997) concluded that several factors contribute differently to adaptations to cold environments in anurans such that changes in physiology may be more important for nocturnal and terrestrial frog species.

Ecological specialisations usually occur as a result of adaptation to a finite set of environments encountered (Navas, 1997; Poisot et al., 2011). Our results suggest that although variation in T_a is important in determining T_b, EWL and SMR in amphibians, not all amphibians are affected in a similar fashion. Although M_b and sex (given the pronounced sexual dimorphism in anurans) have been identified as key factors affecting physiological traits in anurans, we suggest that perhaps this result should be considered in the context of each species’ prevailing ecology and habitat specialisation with more emphasis on M_b. However, certain caveats to this assertion need to be considered. First, we acknowledge that the observed differences could have been because the species that we considered represent very divergent groups (three different anuran families) so that the differences are a function of the divergent evolutionary history as opposed to different ecologies (but see Tracy & Christian, 2005). Second, our set-up did not allow us to directly test for the relative importance of VPD on anurans, we did find that VPD was a better predictor of EWL as opposed to T_b for two of the three species. X. laevis is a fully aquatic species and perhaps EWL may be somewhat independent of VPD over the test conditions we exposed the animals to in our study, compared to more terrestrial species. Although we welcome these new developments with VPD driving water loss rates in anurans, we also suggest that their generality should be tested against species habitat specialisations.