Only studies that possessed both DEE and RMR were included in the analyses. Table 1 shows the mole-rat species for which raw data were available for analysis. Additionally, raw data for outgroup rodent species were available for this study, namely, the Cape ground squirrel (Xerus inauris; Scantlebury et al., 2007) and four-striped mouse (Rhabdomys pumilio; Scantlebury et al., 2006a). Table 2 presents the general biological, ecological and behavioral information about each rodent species used in this study. IBM SPSS 28 (IBM Corp., Armonk, NY, United States) was used for all statistical analyses and significance was assumed at p ≤ 0.05. The normality of the response variables (RMR, DEE, SusMS, WTO, WEI, and BF) was determined using Shapiro Wilk tests (S–W) and residual plots. Homogeneity of all dependent variables was confirmed with a Levene’s test. All non-normally distributed dependent variables were log-transformed in an attempt to obtain a normal distribution.

Climate data were gathered using the methods outlined by Wallace et al. (2021). All environmental conditions are presented in Table 3. Climate data for each site were retrieved from ERA5-Land of the European Centre for Medium-Range Weather Forecasts-the latest generation created by the Copernicus Climate Change Service (Muñoz-Sabater et al., 2021). The spatial (resolution is 0.1° by 0.1°). These data were used to calculate an annual aridity index (AI) (eq. (1)). Whereas total precipitation (tp) was directly obtained from ERA5-Land, potential evapotranspiration (PET) was calculated from the well-known Romanenko estimation (eq. (2)) (Romanenko, 1961). For eq. (2), relative humidity (RH) was calculated from ERA5-Land d2m (eq. (3)).

Each environmental condition (Table 3) was analyzed separately using a GLM, Gamma log-link distribution, and the site as the primary predictor. Subsequent post-hoc analyses were conducted using LSD using estimated marginal means (EMMs). All data are presented as mean ± SD.

Male B. suillus, on average, were 165.2% heavier than female BS (t = −5.406, p = 0.001, Figure 1A). When both sexes were combined, there was a strong effect of body mass on RMR; the linear fitted regression

significantly explained the variance (F = 16.8, p = 0.004, Figure 2A). Likewise, when both sexes were combined, there was a strong effect of body mass on DEE; the linear fitted regression

significantly explained the variance (F = 61.6, p = 0.004, Figure 2C).

Furthermore, when both sexes were combined, there was a strong effect of body mass on WTO; the linear fitted regression

significantly explained the variance (F = 135.2, p = 0.001, Figure 2E).

Neither sex (F = 9.78, p = 0.89, Figure 1C) nor body mass (F = 1291.1, p = 0.14) affected RMR. Interestingly, in females there was a strong effect of body mass on RMR;the linear fitted regression

significantly explained the variance (F = 18.9, p = 0.007, Figure 2B). While, body mass did not have a significant effect on RMR in males (F = 0.21, p = 0.73, Figure 2B) but did possess a negative relationship with RMR (Slope:−0.04, Figure 2B).

A similar trend was seen in DEE (Figure 1B) as in RMR, with increased DEE in adult male B. suillus compared to adult female B. suillus. Likewise, adult male B. suillus possess a higher SusMS than adult female B. suillus (Figure 1D). Similar values of WEI occurred for both male and female adult B. suillus (Figure 1F), while higher WTO was observed in adult male B. suillus compared to adult female B. suillus (Figure 1E). In addition, both male and female adult B. suillus DEE and WTO showed a positive relationship with body mass (Figures 2D,F). Contrastingly, BF of adult female B. suillus was higher than adult male B. suillus (Figure 1G).

Speakman (1997a) suggested that field metabolic rate (FMR), the DEE of animals in the wild or “field”, is affected by environmental conditions. For example ambient temperature affects DEE with the trend of animals inhabiting a cooler climate (a characteristic of a mesic environment) possessing a higher FMR to animals that persist in hotter climates (a characteristic of an arid environment). A similar pattern found in other desert-dwelling vertebrates (see Nagy et al. (1999) for review). However, a contrasting pattern was observed within the African mole-rat family, particularly in the solitary genus Bathyergus. Bathyergus janetta, which inhabits an arid and hot region of South Africa, possessed a greater DEE than its mesic-dwelling cousin, B. suillus. Indeed, B. janetta possessed the highest DEE compared to other arid dwelling mole-rat species, including F. damarensis, and arid dwelling aboveground species, namely X. inauris and R. pumilio. African mole-rats need to forage for geophytes through the energetically expensive method of digging (Vleck, 1979), which becomes energetically more expensive the drier the soils become (Lovegrove, 1989). It would, therefore, be expected that the arid-dwelling species, such as B. janetta and F. damarensis, would possess the highest DEE. Our results suggest that B. janetta follows this expected trend, but the social F. damarensis does not, as F. damarensis appears to possess similar DEE to the mesic dwelling solitary B. suillus and the social C. h. natalensis. As proposed by the aridity food distribution hypothesis (Jarvis et al., 1994), the evolution of social living may negate or ameliorate the need for increased physiological adaptations to an arid environment (Scantlebury et al., 2006a,2008). In species that form large groups, such as F. damarensis, animals will not have to expend as much energy daily, even in arid environments, as the increased group size allows for the efficient locating and retrieval of food. Group-living allows for such substantial reduction of DEE in F. damarensis that their DEE is similar to aboveground rodent species, both X. inauris and R. pumilio, who do not need to dig to forage and thus should possess a lower DEE. The benefit of group living is absent in the arid dwelling (solitary) B. janetta, in an increased DEE to allow them to locate and retrieve the supply of sufficient resources. By comparison, even though B. suillus lives solitarily, it exists in much softer soils and with access to a more abundant food resources, resulting in less energy being spent daily foraging (Bennett and Faulkes, 2000).

Both Speakman (1997a) and Nagy et al. (1999) revealed a strong correlations between body mass and DEE. As expected, in this study, body mass played a significant role in predicting an animal’s DEE, with larger animals expending more energy per day when performing their natural behaviors compared to smaller animals. Interestingly, G. capensis was the only species not to show a significant relationship with body mass. Unlike G. capensis, both B. janetta (Scantlebury et al., 2006b) and B. suillus (this study) exhibit sexual dimorphism in body mass that leads to sexual dimorphism in DEE, lending added support to the body-maintenance hypothesis for sexual dimorphism (see Key and Ross (1999)). The lack of sexual dimorphism in G. capensis may be the leading cause for the absence of association between body mass and DEE, as even though equal numbers of males and females were captured, they were all adults (Bennett et al., 1991) and thus, all individuals were of similar mass resulting in no effect of body mass on DEE (Scantlebury et al., 2006b). We would expect that if smaller G. capensis were captured and compared with larger individuals, an effect of body mass would be seen.

It has long been thought that with increasing aridity (decreasing A.I.), there is an adaptive advantage for a lower RMR (Haim, 1987; Haim Skinner and Robinson, 1987; Tieleman et al., 2003; Williams et al., 2004; Van Sant et al., 2012; Klüg-Baerwald and Brigham, 2017), as the adaptive value of reduced metabolic rates in more arid dwelling animals is thought to be related to increased water and energy savings under low and unpredictable resource availability (Lovegrove, 2000; Tieleman et al., 2003). Along this line of thinking, the arid dwelling mole-rat species (F. damarensis, B. janetta) are predicted to possess lower RMR than the mesic dwelling species (G. capensis; B. suillus, C. h. natalensis). Alternatively, African mole-rat are known to possess low RMRs in response to inhabiting an environment where they are exposed to hypercapnic and hypoxic conditions, whereby their low RMRs allows an equilibrium to be established between the partial pressures of oxygen and carbon dioxide in their blood and tissues, and in their external environment (Ar et al., 1977; Lovegrove, 1986; Ivy et al., 2020; Wallace et al., 2021). If this hypothesis were true, we would expect the aboveground foraging species, namely X. inauris and R. pumilio, and African mole-rat species with poor hypoxia tolerance, indicating that they are not routinely exposed to hypoxia in their burrow system, namely B. suillus (see Logan et al. (2020) for review), to possess the highest RMR. Nevertheless, both these predictions were not supported in this study. In this study, most mole-rat species possessed similar RMRs to one another. While the less hypoxic-tolerant African mole-rat species, B. suillus, possessed a similar RMR to both other solitary (G. capensis) and social (F. damarensis) mole-rat species. Therefore, other explanations, such as physiological and behavioral thermoregulatory strategies (Mota-Rojas et al., 2021), may be driving the pattern of RMR seen within these rodent species (see Speakman (1999) for review).

As with DEE, body mass seems to be the primary factor determining RMR in the current set of rodents, rather than the behavior or the micro- or macro-habitat of the animal. This conclusion is clearly evident by the strong relationship between RMR and body mass overall and within species.

It has long been hypothesized that vertebrates that inhabit arid regions would work less (SusMS) than those inhabitating mesic regions to reduce energy expenditure and increase water and energy savings under low and unpredictable resource availability (Tieleman et al., 2003). However, as with DEE, the African mole-rat family does not follow this trend. Even though beneficial in some circumstances, an exclusively subterranean lifestyle imposes extremely high energetic demands on the animals inhabiting these ecotopes (Vleck, 1979). These energetic digging demands become greater as soils become drier and thus harder (Lovegrove, 1989). These demands are one of the fundamental suggested pressures that have resulted in some African mole-rat species evolving group-living that leads to them sharing the workload amongst group members; this sharing is especially important in more arid environments as posited by the aridity food distribution hypothesis (Jarvis et al., 1994; Lacey and Sherman, 1997).

The current analysis suggests that the aridity food distribution hypothesis proposes selective pressures for group living in African mole-rats (Jarvis et al., 1994; Lacey and Sherman, 1997). Regardless of environmental conditions and digging method, all solitary mole-rat species possessed equally high SusMS compared to individuals of a colony-forming subterranean species (C. h. natalensis; F. damarensis) and fossorial and terrestrial group-living species (X. inauris; R. pumilio). This finding suggests that if an animal is solely responsible for underground foraging and burrow construction, whether the animal is digging in an arid (B. janetta) or mesic (G. capensis; B. suillus) environment, or with teeth (G. capensis), or a combination of teeth and forelimbs (B. suillus; B. janetta), they will be working harder than animals that live in groups in similar environments and/or use similar digging methods. For example, F. damarensis and B. janetta inhabit similar climatic conditions (see Table 3), but F. damarensis lives in large groups and thus possesses a lower SusMS.

Group living, however, does not mitigate all pressures placed upon subterranean animals by their environment. Group living mole-rat species that inhabit arid areas, namely F. damarensis, with drier soils still possess a higher SusMS than those group-living species living in mesic conditions, namely C. h. natalensis. Lovegrove (1989) observed that animals digging in moist soils use less energy than those digging in drier soils. As expected, the outgroup species, X. inauris and R. pumilio, that foraged aboveground (Table 3) possessed some of the lowest SusMS as predicted by Vleck (1979). Interestingly, C. h. natalensis, under the added benefits of group living and inhabiting a mesic environment, possessed equally low SusMS to the above ground-foragers, indicating that group living in a mesic environment could overcome the energetic demands of a subterranean lifestyle. The aridity food distribution hypothesis is often contested as several species of the genus Fukomys and Cryptomys species are found in mesic conditions (Bennett and Faulkes, 2000). These mesic conditions are predicted to be a driving force in selecting a solitary lifestyle, yet, some species still persist as group-forming mole-rat species. Using the patterns investigated in this current study, we hypothesize that the ancestors of mesic-dwelling mole-rat species may have been arid-dwelling social species that moved into a mesic environment and since there is an energetic benefit to remaining as group-forming species, as indicated by the similar SusMS to above ground-foragers, they retained a social lifestyle.

Surprisingly, all studies using DLW techniques on African mole-rats have not investigated WTO or WEI. As a family, the African mole-rats occupy a range of environmental conditions, from hyper humid to hyper-arid, and as eutherian mammals, one would expect a trend in WEI ratios as seen in Nagy (2004). Nagy (2004) observed that eutherian mammals that occupy arid regions require less water per day (WTO) to achieve a water balance compared to their mesic dwelling relatives. Once Nagy (2004) adjusted for their metabolic rates (i.e., the WEI ratio), which is often lower in arid dwelling mammals (Nagy et al., 1999), arid-dwelling eutherians had significantly lower WEI compared to non-arid dwelling species. One might interpret this finding as arid-dwelling mammals using less water to perform their daily behaviors, which is expected as arid environments are water restricted. Reduced metabolism and physiological (lower evaporation, increased water retaining abilities of kidneys) or behavioral (less time spent actively above ground, being relatively more nocturnal) adaptations may be responsible for the reduction in WEI ratios of Nagy’s (2004) above ground eutherian mammals.

African mole-rats do not share this pattern as the more arid dwelling species possessed higher WEI than the more mesic dwelling species. A more surprising fact is that African mole-rats do not drink freestanding water but obtain all water requirements from underground geophytes, which become more dispersed in arid regions (Bennett et al., 1994; Bennett and Jarvis, 1995; Spinks et al., 1999; Malherbe et al., 2003). African mole-rats need to forage for geophytes through the energetically expensive method of digging (Vleck, 1979), which becomes energetically more expensive the drier the soils become (Lovegrove, 1989). It would, therefore, be expected that arid-dwelling species, such as F. damarensis, would require a lower WEI to survive their water restricted environment as water is less available and therefore they would need to conserve water the best they could. Yet, in this study, we see that F. damarensis possesses the highest WEI ratio of all species tested. The decreases in water availability and aridity seem to drive the WEI ratio higher in subterranean mole-rat species.

Several reasons for this surprising reversal in pattern could be at play. Physiological differences between arid-dwelling small mammals and mesic-dwelling small mammals that allow arid-dwelling small mammals to survive and thrive in an arid environment, such as lower RMR, body temperature and rates of evaporative water loss (Haim, 1987; Haim Skinner and Robinson, 1987; Tieleman et al., 2003; Williams et al., 2004; Van Sant et al., 2012; Klüg-Baerwald and Brigham, 2017), have been lost in African mole-rats species due to the shared constraints of their microclimate (burrow system) (Wallace et al., 2021). This hypothesis is supported by similar high water turnover rates in the arid dwelling subterranean species of Spalax ehrenbergi and Thomomys bottae (Gettinger, 1984; Yahav et al., 1989).

Furthermore, as proposed by the aridity food distribution hypothesis, the evolution of social living may negate the need for physiological adaptations to an arid environment (Jarvis et al., 1994). In species that form large groups, such as F. damarensis, animals can increase WEI as water is not a limited resource as the increased group size allows for the efficient locating and retrieval of food and water (behavioral osmoregulation). To date, no comprehensive multi-species study has been conducted on African mole-rats osmoregulatory abilities; however, anecdotal evidence has pointed to no differences in urine concentrating abilities of kidneys in different mole-rats species from different environments, again pointing toward behavioral osmoregulation to allow for the survival of social mole-rats in arid environments (M. van Dyk and D.W. Hart unpublish data).

However, this evidence begs the question of how B. janetta, possessing the second-highest WEI value, persists in their arid environment, while possessing a similar WEI to their mesic-dwelling cousin B. suillus, is not group living. On a recent research field trip (2021), a noticeable reduction in the B. janetta population in the areas of succulent Karoo (Kamieskroon, Garies and Groenrivier) was observed by one of the authors who have been working on B. janetta for the last 30 years (N.C. Bennett per comm). It is believed that due to the effects of climate change, which has seen the areas of Kamieskroon and Garies regions in the Northern Cape area receive below-average rainfall and increased temperatures resulting in desertification (increased aridity) (see Supplementary Figure 1), in combination with poor water retention abilities of solitary African mole-rats species, this has driven this population decline. Even though lacking behavioral osmoregulatory benefits of social living, the solitary African mole-rats that inhabit mesic areas (G. capensis; B. suillus) still have fewer water demands placed on them as food is more abundant and equally dispersed in the softer and wetter soils of their environment. At the same time, C. h. natalensis possesses similar WEI to the solitary mole-rat species due to living in a hyper-mesic environment (Colantoni et al., 2015).

Body fat composition (BF) of an animal can be linked to its resource availability and amount of work (DEE and SuSMS) it does. Thus, animals which inhabitat arid regions are often expected to possess lower body conditions, including lower BF. Once again, the African mole-rat family does not fit that trend. Only one study (before the current study) on African mole-rats using the DLW technique has reported BF. Scantlebury et al. (2006c), using BF, discovered distinct physiological castes in a social mammal, F. damarensis, which solidified the concept that F. damarensis could be classed as a eusocial organism with similar traits to bees and termites. Scantlebury et al. (2006c) showed that infrequent workers in an F. damarensis colony also possess higher BF than frequent workers (particularly in the dry season), which suggested that infrequent workers constitute a physiologically distinct dispersing caste who contribute less to cooperative behaviors in the colony and focus more on building up their body reserves in preparation for dispersal and reproduction when environmental conditions are suitable.

In the current study, all social mole-rat species (C. h. natalensis; F. damarensis) and aboveground foragers (R. pumilio; X. inauris) possessed similar BF to one another. It should be noted that within this current study, all physiological castes, namely breeding individuals, and infrequent and frequent workers, were combined. The Cape mole-rat, G. capensis possessed the highest BF of all African mole-rat investigated a surprising find considering G. capensis is a solitary species and needs to forage (indicted by their high SusMS and DEE) and thermoregulation alone which should result in a lower BF. Surprisingly, this finding also includes the solitary B. suillus. Georychus capensis and B. suillus occur sympatrically, but B. suillus lives in much softer, easier to dig sandy soils while G. capensis inhabit a harder clay soil in vlei areas (Bennett and Faulkes, 2000). This should lead to the hypothesis that B. suillus should possess a higher BF than G. capensis, mainly because this species shows a significantly lower DEE than G. capensis as digging in softer sands is less energetically demanding (Lovegrove, 1989; this study). However, a possible lack of sexual dimorphism in G. capensis (Scantlebury et al., 2006b) and an apparent sexual dimorphism in B. suillus (this study) may cause the differences in BF between species. Bathyergus suillus males show indications of increased DEE and SusMS, suggesting that males are far more active than females resulting in less stored fat in males. This sex bias is possibly due to the larger male B. suillus body size, which increases energy expenditure while constructing larger tunnels and nest complexes. Furthermore, males have been documented to search for female tunnel systems and dig into them, searching for a mate (Hart et al., 2006; Thomas et al., 2009); this searching would require more activity and energy expenditure in the males, while the females are less active and store fat in preparation for offspring care.

A finding of great concern is that the solitary arid-dwelling B. janetta possessed the lowest BF of all species tested in this study, indicating low resource reserves and possibly poor body condition. The WEI already indicated that the possible decline of the B. janetta population observed in 2021 is possibly due to desertification due to climate change and overgrazing by domestic animals on land in their distributional range (Supplementary Figure 1). The BF adds to this claim, as in 2006, when these data were collected, to 2021 there has been a drastic increase in aridity (a decrease in AI) (see Supplementary Figure 1). If B. janetta already showed reduced body fat reserves in 2006, possibly due to being a large, solitary subterranean mole-rat in an arid environment with dispersed food sources. Then as the area becomes more arid and food becomes scarcer due to geophytes drying out as well as the ground becoming harder, B. janetta may no longer obtain enough food to maintain a healthy body condition, which could drive local extinction of this species.