Sloth metabolism may make survival untenable under climate change scenarios

Ethics

This research was approved by the Swansea University Animal Welfare & Ethical Review Process Group (AWERP), and the Costa Rican government and associated departments (MINAE, SINAC, ACLAC) permit number: R − 033 − 2015

Resting metabolic rate (temperature manipulation in the metabolic chamber)

Measurement of body temperature

A miniature temperature logging device (iButton^®, Thermochron, Dallas Semiconductors; Maxim Integrated Products, Inc., Sunnyvale, CA, USA) (model DS1922L (±0.0625 °C)) was inserted into the rectum of nine of the sloths using a gloved digit and lubricant. The logger was calibrated prior to use by immersion into a temperature-controlled water bath and programmed to record temperature every 30 min (Cliffe et al., 2018). Sloths defecate only once a week, storing faeces in an anal pouch. Rectal insertion of the temperature logger was therefore deemed the least-invasive, non-surgical method of obtaining accurate core temperature values. If faecal pellets were found in the anal pouch of the animal, then these were removed prior to logger insertion to ensure the most accurate temperature readings.

Measurement of resting metabolic rate (RMR)

Metabolic data were collected as previously described in (Cliffe et al., 2018). Specifically, prior to measurements, all sloths were weighed (E-PRANCE^® Portable Hanging Scale (±0.01g)). They were then placed in an 87-L Perspex^® metabolic chamber (55 cm long × 45 cm high × 35 cm wide). The chamber was placed in a temperature-controlled water bath which was covered with a polystyrene lid. The water bath (95 cm × 85 cm × 75 cm), also made from Perspex^®, was lined with black plastic sheeting and supported with an exterior metal frame. Within the metabolic chamber, there was a branch for the animal to hold on to, and from which it could comfortably suspend itself upside down. There was a small window in the plastic sheeting (a ‘peep’-hole) through which the sloth could be observed without it being disturbed by the observer.

Oxygen consumption (VO₂) was measured using an open-flow system with an upstream flow meter. Fresh air from outside was pumped into the chamber (AIR CADET^® Barnant, model 420-1902; Barnant, Barrington, IL, USA), via a copper coil submerged in the water bath, at rates of between 4 and 12 L/min. Flow rate was adjusted to the mass of the sloth to ensure that the depression in oxygen concentration within the chamber remained in the range 0.2–0.8% (Speakman, 2013). The flow was measured using a flow meter (ICEhte10 platon flow meter 1-12L/min; ICEoxford Limited, Oxford, UK) which was factory calibrated and checked prior to use using a mass-flow generator (Sable Systems Flowkit 100; Las Vegas, NV, USA). The incurrent air flow rate was measured before drying. The system was checked for leaks using a dilute solution of soapy water. The air inlet was located on the opposite side of the chamber to the air outlet to ensure an adequate mixing of air within the chamber. Air leaving the chamber was subsampled at 200 ml/min and then dried (using Drierite) before entering an oxygen and carbon dioxide analyser (FoxBox Field Gas Analysis System, Sable Systems International, Las Vegas, NV, USA). The length of tubing leading from the metabolism chamber to the gas analysers was 0.5 m. The lag time for the analyser reading to equilibrate when the tubing was placed into the chamber to subsample the gasses was less than 1 min. The analyser was factory calibrated and set to 20.95% oxygen before each animal was measured. Fresh air readings were recorded at the start and the end of each run to correct for analyser drift. Any drift in the analyser was assumed to be linear for baseline correction. An acclimatization period of ∼150 min was allowed at the beginning of each experiment for any sedation to wear off, for each sloth to become accustomed to the chamber, for  T_b to adjust to the chamber temperature and for the chamber gases to equilibrate (McClune et al., 2015). The animals were observed continuously through the peep hole (for welfare reasons and to make sure they weren’t showing any signs of stress). During measurement periods (i.e., following temperature adjustment periods and when gas concentrations had stabilised), oxygen and carbon dioxide concentrations were recorded manually at two-minute intervals. A total of 12 experimental runs were made (Table S2). An ‘experimental run’ refers to a series of measurements from one animal, taken during the course of a day.

VO₂ (ml/min⁻¹) was calculated as: (1)${\displaystyle V{O}_2=\frac{\mathrm{FR}\cdot \left(\left(F_i{O}_2-F_e{O}_2\right)-F_e{O}_2\cdot \left(F_{e}C{O}_2-F_{i}C{O}_2\right)\right)}{\left(1-F_e{O}_2\right)}}$ where FR is the flow rate, F_iO₂ is the fractional amount of O₂in the chamber incoming air, F_eO₂ is the fractional amount of O₂ in the outgoing air, F_iCO₂ is the fractional amount of CO₂ in the incoming air and F_eCO₂ is the fractional amount of CO₂ in the outgoing air (Lighton, 2008). Values were corrected for standard temperature and pressure. Metabolic rates were calculated using a conversion factor of 20.1 joules per millilitre of oxygen, which is correct for an obligate herbivore such as the sloth (Schmidt-Nielsen, 1997).

Values for resting metabolic rate (RMR) were compared with allometrically predicted values for terrestrial mammals as cited in (Kleiber, 1961; White & Seymour, 2003).

Temperature manipulation

Temperatures within the chamber were manipulated following the protocol described by (Cliffe et al., 2018). This was achieved by varying the temperature of the water bath which contained two electric water heaters (Grant water bath heater circulator) and two water fans which stirred the water in a clockwise direction around the metabolic chamber. The temperature within the chamber was measured using a copper-constantan thermocouple and monitored on a Tecpel 307P Dual Input Digital Thermometer (0.1 °C). Chamber temperature was recorded at four-minute intervals throughout the duration of each experimental run. The first three experimental runs were undertaken with the chamber maintained at constant temperature. The remaining 9 experimental runs had the chamber temperature directly manipulated. Following the initial ∼150-minute acclimatization period, the temperature of the metabolic chamber was increased incrementally in 2-degree steps i.e.: 16−19 °C, 20−23 °C, 24−26 °C, 27−29 °C, 30−32 °C, and 33−35 °C by varying the temperature of the water bath. These temperature brackets were selected as they encompass the most extreme range of ambient temperatures to which Choloepus sloths are naturally exposed in the wild.

The length of time animals spent at each temperature increment was sufficient to allow both equilibrations of gases within the chamber, and for the animal T_b to adjust to the new T_a. Typically, animals spent 60 min adjusting to each 2-degree temperature increment. Following the c.60-min adjustment period, when sloths were seen to be at rest and the gas concentrations had stabilised, RMR readings took place and recordings were taken every 2 min for a further 10 min. RMR values were then calculated from the mean of these 5 values. In nearly every case, the sloths were inactive, apart from slow postural adjustments. As a control, the empty chamber was taken through 5 different temperature increments on three separate occasions prior to testing with animals. During these control tests, temperatures were recorded from twelve different locations within the chamber (Cliffe et al., 2018).

The effect of natural fluctuations in T_a on T_b and posture

Thirty-four C. Hoffmanni sloths (seventeen males, seventeen females, six high-altitude, twenty-eight low-altitude) had pre-calibrated iButton^® temperature loggers inserted rectally. No sedation was necessary, and all logger insertions were carried out without removing any sloths from the enclosures. The loggers were programmed to record temperature every 30 min.

All sloths were housed in individual standardised enclosures measuring 5.3 m² with a shelf (114 cm by 61 cm) and 13 horizontal climbing bars. Sloths were fed twice daily at 7am and 2pm. The enclosures were outdoors, exposing the animals to natural fluctuations in T_a although, to ensure uniform temperatures and minimise possible microclimate differences, all enclosures were covered by a metal roof to prevent access of rain or direct sunlight. Although levels of non-visible light such as ultraviolet (UV) were not monitored in this study, the metal roofing should have standardised and minimised these effects. Three further temperature loggers were uniformly distributed throughout the enclosures in order to measure T_a.

Following temperature logger insertion, visual surveys were completed on all sloths at 2-hour intervals for 48 h. Posture was graded on a scale of 1–6 (1 = tight ball, 6 = all limbs spread) (Cliffe et al., 2018; Muramatsu et al., 2022). Temperature loggers were collected opportunistically when the sloths defecated. The mean time that the temperature loggers were retained in the rectum was 3.1 days. Six temperature loggers were never retrieved and were presumed to have been washed away during cleaning of the enclosures. Consequently, data presented are from twenty-eight sloths (fifteen males, thirteen females, three of these being high-altitude sloths, twenty-five low-altitude sloths).

Statistical analysis

The effect of natural fluctuations in T_a on T_b and posture

Rectal T_b and natural T_a were recorded at 30-minute intervals. Due to the high likelihood of temporal autocorrelation in temperature data, a generalized additive mixed model (GAMM) was used to test the relationship between T_a and T_b. The GAMM included Julian day and time, as well as mass, sex, and altitude as covariates using the “gamm4” package (Wood & Fabian, 2022). The ‘acf’ function was used to test for autocorrelation in the temperature data; where autocorrelation was found, the GAMM model was refitted to correct for autocorrelation using the ‘corAR1’ function. The ‘mgcv’ package was used to test whether the relationships between T_a and T_b, T_b and time of day, were linear or polynomial and the best fit model was used to analyse relationships. The standard smooth function set in the models was a cubic regression spline with automatically set knots. Differences in T_b between high- and low-altitude sloths were examined using a Mann–Whitney test after conducting a Shapiro–Wilk test to assess the normality distribution of the data. The mean time lag between ambient and core body temperature was determined by calculating the average time between maximum/minimum ambient temperature and maximum/minimum core body temperature for each individual. The effect of T_a and T_b on posture was examined using a GAMM—two different models were tested given the correlation between T_a and T_b; the best fit model was identified as the one with the lowest AICc score.

Projected impacts of climate change

To estimate the projected impacts of climate change on the body temperature (T_b) of sloths, we employed a bootstrap method adjusted for sample size differences to account for the uncertainty around our predictions. The rate of change of T_b ( °C/min) for both high- and low-altitude sloths was plotted against the difference between T_b and T_a. We used the resultant regression equations to model the projected T_b increase for high- altitude and low-altitude sloths if the climate warmed by an average of 2 °C. We simulated 1,000 predictions for each projected Tb value by drawing from a normal distribution centred on the mean of the projection with a standard deviation equal to the standard error of the model predictions. We derived 95% confidence intervals for the predicted T_b values by taking the 2.5th and 97.5th percentiles of the bootstrap distributions (Efron & Tibshirani, 1994) (Fig. S2).

To estimate the projected impacts of climate change on the RMR of sloths, we used a similar bootstrap approach adjusted for sample size (Table S3). Individual regression equations were calculated for high- and low-altitude sloth RMR as a function of T_a (calculated for temperature brackets: 19−23 °C; 23−27 °C; 27−29 °C; 27−29 °C; 29−32 °C; 32−34 °C) and the intercepts from these regressions were used to calculate daily RMR on a minute-by-minute basis for current T_a’s, and to estimate the effect of climate change (from 5 °C below, to 3 °C above current T_a’s) on the projected RMR for high- and low-altitude sloths.

Resting metabolic rate (temperature manipulation in the metabolic chamber)

Mean body mass across the 12 sloths was 5.33 ± 0.67 kg (SD) and mean RMR over all temperatures was 118.26 ±  36.76 kJ/kg/day (Table S2). Mean RMR values were 39% lower than the general mammalian allometric prediction of Kleiber (1961), and 13% lower than the prediction of White & Seymour (2003) which includes variation due to factors such as body temperature and digestive state. Neither body mass (p = 0.85) nor sex (p = 0.73) had a significant effect/interaction on RMR (Table 1). There was no significant difference in RMR for sloths that had (121.59 kJ/kg/day ± 17.21 kJ/kg/day) and had not (117.58 kJ/kg/day ± 39.59 kJ/kg/day) been sedated prior to entering the metabolic chamber (w = 939, p = 0.329).

The LMMs showed that ambient temperature (represented as a continuous variable in LMM₁ [t = 8.70, p < 0.001]) had a significant effect on RMR (Table 1) (Fig. 1, Fig. S3A). There was also a significant interaction between ambient temperature (represented as a categorical variable in LMM₂) and altitude on RMR (t = 3.72, p < 0.001) and a significant effect of body temperature (t = 5.22, p < 0.001) on RMR (Table 1) (Fig. 1, Fig. S3B). A significant effect and interaction of body temperature on RMR can be seen in Fig. S3C. As there were multiple measurements taken from each individual, individual effects were accounted for in the model using estimates of the random effect (Fig. S4).

RMR of high-altitude sloths (n = 4) (126.25 ±  40.84 kJ/kg/day) was significantly higher than RMR of low-altitude sloths (n = 8) (110.70 ±  30.94 kJ/kg/day) when all the data were considered together (w = 1763, p = 0.038) (Fig. S3A). At T_a’s ≥32 °C, RMR values of high-altitude sloths (n = 4) (162.71 ±  52.03 kJ/kg/day) were significantly higher than those of low-altitude sloths (n = 5) (103.72 ±  34.69 kJ/kg/day) (t = 3.08, df = 17.52, p = 0.007) (Fig. S3B). There was no significant difference in RMR between high- and low-altitude sloths at T_a’s <32 °C (w = 1013, p = 0.44). Metabolic rates were lowest at 16 °C–19 °C (high-altitude: 90.90 ±  23.75 kJ/kg/day, low-altitude: 83.53 ±  21.74 kJ/kg/day) and increased with increasing T_abefore plateauing at temperatures between 23 °C–32 °C (high-altitude: 134.19 ±  27.42 kJ/kg/day, low-altitude: 127.21 ±  24.73 kJ/kg/day). At T_a’s above 32 °C, high-altitude sloth RMR increased sharply, while low-altitude sloth RMR decreased (Fig. S3 B). In high-altitude sloths, T_b at T_a’s ≥32 °C (35.87 ±  0.76) was significantly higher than T_b at T_a’s <32 °C (34.51 ±  0.73) (t = 5.31, df = 15.23, p < 0.001).

The effect of natural fluctuations in T_a on T_b and posture

There was a significant effect of T_a on T_b (F = 25.98, p < 0.001) (Table 2), and a significant effect of altitude origin on T_b (t =  − 44.95, p < 0.001) (Fig. 2). There was a significant effect of Julian day (t =  − 7.11, p < 0.001) and time of day (F = 81.99, p < 0.001) on T_b (Table 2) after controlling for temporal autocorrelation. The relationship between T_a and T_bwas best described using a linear model (delta AICc = 0; polynomial: delta AICc = 0.58) (Fig. 2). Mean T_awithin the enclosures was 26.90 °C ± 1.93 °C (overall recorded minimum: 24.56 °C, maximum: 33.11 °C). Rectal temperatures averaged 34.84 °C ± 0.88 °C across individuals, ranging from an overall recorded minimum of 33.43 °C to a maximum of 37.28 °C. The mean T_b range within each individual was 2.60 °C. There was a mean lag of 2.5 h between the maximum/minimum daily ambient temperature and the maximum/minimum sloth T_b(Fig. 3). T_b of high-altitude sloths (35.70 ±  0.61 °C) (n = 4) was significantly higher than that of low-altitude sloths (n = 8) (34.34 ±  0.56 °C) (w = 19604, p < 0.001) (Figs. 1 and 2). T_a (F = 10.33, p < 0.001) and time of day (F = 1.54, p = 0.01) had a significant effect on sloth body posture. T_b also had a significant effect on sloth posture with both high- and low-altitude sloths adopting spread out postures more frequently at higher temperatures (Table 2) (F = 4.30, p = 0.04); however, model selection showed that GAMM₂ with T_a was the best fit model (delta AICc = 0; GAMM₃: delta AICc = 1.38).

Projected impacts of climate change

There was a projected T_b increase of 1.53 °C and 2.13 °C, respectively, for high-altitude and low-altitude sloths if the climate warmed by an average of 2 °C (Fig. 4). For low-altitude sloths, the mean projected T_b was 36.57 ±  0.01 °C (95% CI [36.564 °C–36.576 °C]) indicating high precision in the estimates due to the larger sample size (n = 25). For high-altitude sloths, the mean projected T_b was 37.27 ± 0.01 °C (95% CI [37.258 ° C–37.283 °C]) reflecting greater variability and less precision due to the smaller sample size (n = 3) (Fig. S2). As climate change causes an increase in average daily T_a, the RMR of both high- and low-altitude sloths is projected to increase accordingly (Fig. 5). As the increase in average daily T_a exceeds 2 °C above current T_a’s, low-altitude sloth RMR is projected to plateau, while high-altitude sloth RMR continues to escalate (Fig. 5). The 95% confidence intervals (Table S3) reflect the greater variability in the estimates for high altitude sloths due to the smaller sample size.

Projected impacts of climate change

How organisms obtain, convert and expend energy is directly related to the T_a of their environment (Brown et al., 2004; Levy et al., 2017) and this is one of the reasons why climate change is projected to have an extensive effect on the global energetic requirements of organisms (Dillon, Wang & Huey, 2010; Parmesan, 2006; Parmesan & Yohe, 2003; Root et al., 2003; Shokri et al., 2022). What might the consequences be for sloths?

While future climatic predictions for the South and Central American rainforests are variable, all point towards these regions becoming hotter and drier, with current estimates forecasting a 2–6 °C increase in average daily air temperatures by the year 2100 (Marengo et al., 2014; Nũez, Solman & Cabré, 2009; Romero & J, 2022). From the data presented in this paper, we were able to create a simplistic model to predict the possible effect of climate change-associated temperature increases on the T_b and RMR of sloths originating from both high- and low-altitude regions.

As climate change causes an increase in average daily T_a, the RMR of both high- and low-altitude sloths is projected to increase accordingly. As the increase in average daily T_a exceeds 2 °C above current T_a’s, the capacity of low-altitude sloths to invoke metabolic depression limits any further increase in RMR (Angilletta, 2009; Dillon, Wang & Huey, 2010). This physiological plasticity should be accentuated by the ability of lowland populations to shift distribution ranges along climatic gradients to higher elevations (Parmesan & Yohe, 2003; Perry et al., 2005; Root et al., 2003), thereby providing a degree of flexibility when faced with a warming climate. On the other hand, sloths originating from high-altitude mountain-top locations appear to lack the metabolic and geographic plasticity of their lowland counterparts and consequently are likely to be more constrained in their ability to adapt to a continuously warming climate.

An increased rate of energy expenditure must be balanced by an increased rate of energy intake. This option appears biologically implausible for sloths due to their slow digestive rate and constantly full stomach, restricting food intake and imposing intrinsic energy processing limitations (Cliffe et al., 2015; Montgomery & Sunquist, 1978; Nagy & Montgomery, 1980). Indeed, estimates for digesta passage time for sloths range from 150–1,200 h (Foley, Engelhardt & Charles-Dominique, 1995; Montgomery & Sunquist, 1978; Vendl et al., 2016), some 3–24 times slower than similar sized arboreal folivores (Espinosa-Gómez et al., 2013), with the primary reason for this believed to be linked to the time required to detoxify the food plants (McNab, 1978). While it is plausible that the increase in metabolic activity with environmental temperature may increase the rate of food passage (Doucette et al., 2023), and therefore intake, it is unlikely that the sloth’s digestive tract has the capacity to process food much faster. Although the model presented here is rudimentary in its omission of error and uncertainty considerations, and further research is needed to fully understand the sloth’s metabolic response to temperature, we predict that a comparatively small increase in ambient temperature could see high-altitude sloths pushed into a situation where it is impossible to make their energy consumption tie in with their energy budget.