Water economy in rodents : evaporative water loss and metabolic water production

Studies on water balance in desert-dwelling granivorous rodents use evaporative water loss (EWL) and metabolic water production (MWP) to evaluate the efficiency of water regulation, expressed by the model Ta @ MWP = EWL, defined by an ambient temperature (Ta) value at(@) which MWP EWL = l. Here we evaluate and apply this model (1 r2) determining water regulation efficiency, based on the energetic cost (MR) to maintain water balance (WB ), that is MRWB. To test the model, EWL was measured and MWP estimated in nine species of rodents from different localities of northern and north-central Chile (II and IV Regions): Octodon degus (Od) and 0. lunatus (Olu) (Octodontidae), Abrothrix olivaceus (Ao ),A. longipilis (Al),A. andinus (Ad), Phyllotis darwini (Pd), P. rupestris (Pr), P. magister(Pm), Oligoryzomys longicaudatus (01) (Muridae) and Chinchilla lanigera (Cl) (Chinchillidae). Literature information on rodents from xeric and mesic habitats was also analyzed. Results indicate that: 1) Cl has the lowest EWL (0.58 mg H g h), followed by Od < Olu < Ao < Pd <Pm< Pr <Ad< Al < 01. 2) EWL and body mass (M) are related through independent equations considering two distinctive habitats: EWL (xeric)= 5.968 M-0416 (r = -0.89; P < 0.001) andEWL (mesic) = 17.272 M 0·532 (r = -0.85; P < 0.001). 3) MWP and M are related through the equation: MWP = 14.256 M_-0-539 (r =0.98; P< 0.001). 4) At the intraspecific level, MWP/EWL and Ta are related through a negative exponential equation: MWP/EWL=a JO-bTa (r = -0.95; P< 0.001). 5) Ta@ MWP = EWL and M are related through the equation: T@ (MWP = EWL) = 26.799 M-0 142, (r =0.49, P < 0.02). 6) MR-WB and M, are related through independent equations according to the prevailing animal's habitat: MRWB (xeric) = 34.627 M 0339 (r =0.93; P < 0.001) and MR-WB (mesic) = 68.132 M 0·381 (r = -0.86; P < 0.001). These last two equations have comparative advantages to the previous ones because they include rodents with different dietary habits, are able to discriminate patterns in the water regulation efficiency as a function of different habitats (xeric and mesic ), and enable to evaluate the energetic cost of water balance.


INTRODUCTION
Adaptations of rodents to life in deserts may include different combinations of morphological, physiological, behavioral, and ecological characteristics (Bozinovic & Contreras 1990).Traditionally, low energy expenditure linked to dietary specialization on seeds is considered adaptive for energy and water conservation in xeric environments because it results in lower food requirements and reduced water turnover in variable and unproductive habitats (Hinds & MacMillen 1985).Granivorous desert rodents display physiological features that favor body water conservation, such as an efficient kidney function, low fecal water content, and comparatively lower evaporative water loss (EWL) (MacMillen & Lee 1967, MacMillen 1972, Schmidt-Nielsen 1979, Christian 1979, Withers et al. 1982).In contrast, omnivorous, herbivorous, and insectivorous desert rodents are unable to survive under the same conditions because they depend on exogenous water (Bozinovic & Contreras 1990).Desert colonization and survival in xeric habitats of rodents lacking physiological specialization to desert life are based on behavioral strategies such as nocturnal and/or crepuscular activity, and selection offavorable microhabitats (MacMillen 1972, Schmidt-Nielsen 1979, MacMillen & Hinds 1983, Degen et al. 1986).
Xeric habitats have existed in South America for a long time and it is intriguing why present day South American desert rodents do not show an array of adaptive traits to life in deserts comparable to that found in other continents.Several authors pointed out that South American desert rodents lack physiological and energetic specialization for energy and water conservation, postulating that their success essentially owes to behavioral and ecological strategies (Mares 197 5, Meserve 1978, Streilein 1982).Nevertheless, McNab ( 1982) noted that such lack of adaptations may, to some extent, be more apparent than real, because only few studies have been conducted in South American desert environments.In fact, Bozinovic et al. (1995), and Cortes et al. (1988, 1990) reported several energetic, renal and pulmocutaneous features favoring water conservation in some species of north-central Chile, resembling those previously described for "typical" desert rodents (e.g.Schmidt-Nielsen 1979).Cortes et al. (1988Cortes et al. ( , 1990) ) reported maximum urine concentration capabilities over 4,000 mOsm/kg, similar to those found among desert rodents of North America and Australia (Schmidt-Nielsen & O'Dell 1961, MacMillen & Lee 1967, 1969, Abbott 1971, Maxon & Morton 1974, Schmidt-Nielsen 1979).Also, Cortes et al. (1988Cortes et al. ( , 1990) ) documented species-specific variability in EWL and differences in nasal passages that account for variations in the dependency to free water among sympatric species in north-central Chile.In addition, Bozinovic & Rosenmann (1988a, 1988b ), Caviedes-Vidal et al. (1990), Bozinovic & Marquet ( 1991) reported mechanisms of energy conservation such as torpor in South American rodents.
Theoretically, some Ta exists at which a rodent can survive exclusively on a dry diet and be independent of exogenous water (MacMillen & Hinds 1983).By analyzing the amount of common variation between two variables (r 2 ), here we evaluate MacMillen and Hinds' model and introduce new indices to assess water regulation efficiency.We search for the energetic cost (MR) necessary to maintain water balance (WB ), which is expressed by MR-WB, condition that is met when MWP/EWL = 1.To test our model, we measured EWL and estimated MWP (at different Ta' s) at intra-and inter-specific levels among rodents differing in food habits and inhabiting different localities of northern and north-central Chile (Regions 11 and IV).We also used literature information on EWL and Ta @ MWP = EWL of rodents from xeric and mesic habitats (MacMillen & Hinds 1983, Hinds & MacMillen 1985).To eliminate potential methodological distortions, we restricted our analysis to original data and published information in which all variables were studied with similar methods.

Evaporative Water Loss, Metabolic Water Production and Water Regulation Efficiency
Evaporative water loss was measured at different ambient temperatures (Ta) in an open-flow system (Cortes et al.1990).Air flow passed through the system at a constant rate of 3 Llmin.Moisture in the air was removed by columns of silica gel and Drierite desiccants.To avoid temperature changes dried air was admitted through copper coils submerged in a thermoregulated bath.After passage through a flowmeter (La Motte), air was introduced into a glass chamber containing the animal.This chamber was equipped with a 25 mm mesh wire floor placed above 2.5 inch of mineral oil to prevent evaporation from urine and feces.To monitor the ambient temperature, a stainless steel thermistor probe was inserted into the chamber.Evaporative water loss in the airstream leaving the animal chamber was recovered in a preweighed (± 0.1 mg) glass U-tube containing Drierite.All EWL measurements were carried out after reaching thermal equilibrium (45 min).Measurements were conducted every 5 min during 2-3 h.Metabolic water production was estimated from oxygen consumption values, assuming that in average, 1 m! of 0 2 yields 0.62 mg of metabolic water (Schmidt-Nielsen 1979).Oxygen consumption was calculated from the equations (metabolic rates versus Ta) reported by Rosenmann (1977) and Bozinovic & Rosenmann (1988a).Finally, water regulation efficiency was estimated according to Hinds & MacMillen (1985) and also by the proposed efficiency index MR-WB.

Statistical analyses
Results are reported as mean ± 1 standard deviation for each species.Regression equations were calculated by the least-square method.Slopes of the regression equations were compared with the equality test for two slopes.Statistical analyses were performed using ST A TISTICA ® ( 1997) statistical package for Windows, 95 program.

Evaporative water loss
Evaporative water loss has often been ignored in the comparative analysis of South American desert rodents.This variable has fundamental implications on thermoregulatory mechanisms, but it is also an important factor in the maintenance of water balance (Blackwell & Pivorun 1979).Our measured values of EWL obtained at Ta below the lower critical temperature were relatively constant at the intraspecific level.However, significant differences were observed between some species (Table 2).The two larger species studied, Chinchilla lanigera and 0. degus had the lowest rates of EWL (0.58 and 0.89 mg Hp! g h, respectively).The rate of EWL in C. lanigera was equivalent to 2/3 of that in 0. degus, 1/2 of that in 0. lunatus, 1/3 of the rates in both A. olivaceus (IV Region) and P. darwini, 114 of those shown by A. olivaceus (11 Region), A. andinus, A. longipilis, P. rupestris and P. magister, and only 115 of that in 0. longicaudatus (Table 2).
The relationship between EWL (mgHp!g h) and body mass (M in g) for all species of rodents examined (current data and literature sources) inhabiting xeric habitats (see Table 2) is shown in Fig. 1, and is represented by the following allometric equation:

Metabolic Water Production
Estimated values of MWP are shown in Table 3. Same as in the EWL comparison, C. lanigera showed the lowest MWP value (0.54 mg Hp!g h), followed by the Octodontids 0. degus and 0. lunatus (0.98 and 0.97 mg Hp /g h, respectively).Marked differences were found when comparing these three low values with the much higher MWP rates of the murids.Nevertheless, the observed variations in MWP may be explained by differences in body size (Fig. 2).Equation 3 expresses this relationship: On the basis of equation 1 it may be expected that differences in EWL among species should partly be explained by the effects of body size.Nevertheless, comparing measured values ofEWL of the nine species against those expected from equation l (± Syx 0.272), only C. lanigera and 0. degus present values that fall within the range predicted by this equation.In contrast, the rest of the examined species (0. lunatus, P. darwini, A. olivaceus, A. longipilis, 0. longicaudatus, P. -..c a  Body mass (g) Fig. 2. Double logarithmic relationship between metabolic water production estimated at Ta = l5°C and body mass of rodents inhabiting Chilean habitats (see Table 2 for species identification).
rupestris, P. magister, A. olivaceus and A. andinus) had higher EWL values.Consequently, C. lanigera and 0. degus appear as the only two species (thus far reported) with EWL rates similar to those of "typical" desert rodents.
Phyllotis rupestris, P. magister and A. andinus, had relatively higher rates of EWL as compared to rodents from xeric habitats.This physiological trait was unexpected, considering that they live at 2,300 to 3,360 m above sea level, in arid high-Andean environments of the Atacama desert, where extreme climatic conditions prevail.In spite of the stressful environmental condition, these rodents maintain moderate water balance, judging from their summer urine concentrations ranging between 2,500 to 3,000 mOsm/kg (A.Cortes unpublished data).On the other hand, we have observed that P. rupestris and P. magister are chiefly nocturnal, while A. andinus is mostly crepuscular.Further, they select microhabitats where shrub vegetation is relatively abundant, along to water courses, and with rock piles.Consequently, the observed physiological constraint on water conservation by the pulmocutaneous route, appears to be compensated by behavioral traits, as happens among desert rodents of North America such as Neotoma lepida (MacMillen 1972).
Slopes of equations 1 and 2 are significantly different (t = 4.86, P < 0.001).Thus, the magni- tude of EWL of the species studied (Table 2), is not only affected by body size, but it is also related to the type of microhabitat used.Consequently, EWL appears as an important variable to evaluate the degree of specialization of the pulmocutaneous route in species that occupy different habitats (xeric versus me sic).Equation 3indicates that the magnitude of the MWP of the species studied depends only on the body size, suggesting that the ratio MWP/ EWL may be a variable that reflects the efficiency in water regulation of rodents as a function of the habitat in which they live.
The relationship between MWP/EWL and Ta is described by exponential equations with negative slopes (Fig. 3 ), similar to those reported by MacMillen & Hinds (1983).These relationships are highly significant (r = 0.97; P << 0.01).Regression equations and Ta @ MWP = EWL for rodents in Chile and for North American heteromyids as reported in the literature (MacMillen & Hinds 1983) are listed in Table 4.
Regarding the values of the index Ta @ MWP = EWL, A. olivaceus and 0. degus are the Chilean rodents that had the highest values (18.6 and 16.6°C, respectively).Based on the model of MacMillen & Hinds (1983), A. olivaceus could be said to have a greater efficiency in water economy than that of 0. degus.Nevertheless, studies of Environmental temperature (°C)  tolerance to water deprivation, maximum urine concentration and EWL show the opposite (Cortes et al. 1990).The relationships between T@ MWP = EWL and M for the species listed in Table 4 is described by the following allometric equation (Fig. 4): Ta@ MWP = EWL = 26.799M• 0142 (4) r = -0.49;P < 0.02 Non-explained variance (1-r 2 ) of the model (4) was 76%, suggesting that the index Ta@ MWP = EWL had low ability to detect differences in water regulation efficiency of species with regard to the habitat type used.

Assesment ofthe"new model"
Table 4 shows the average values of water regulation efficiency expressed by the index MR-WB for the rodents studied and for heteromyids of North America.The relationship between MR-WB and M, values taking into account the type of habitat (Table 2), yields two allometric equations (see Fig. Equations 5 and 6 have similar slopes, but different intercepts, indicating that when body size effects are considered, variations in MR-WB are basically explained by the type of habitat that the species are using (Fig. 5).Out of all species of north-central Chile that we studied, only C. lanigera and 0. degus resembled the pattern of MR-WB observed among heteromyid rodents of xeric habitats of North America (MacMillen & Hinds 1983, Hinds & MacMillen 1985).The remainder of the species studied are similar to heteromyids inhabiting mesic habitats, such as Liomys salvani, L. irroratus or Heteromys desmarestianus (Hinds & MacMillen 1983, MacMillen & Hinds 1985).As expected, energy cost to maintain water balance in rodents inhabiting xeric habitats was lower than that of rodents from mesic habitats.The lower energy cost may be adaptive for energy and water conservation because it permits a reduced water turnover in a variable, unpredictable, or unproductive environment.
The rodents we studied have different food habits: 0. degus, 0. lunatus and C. lanigera are herbivorous (Meserve 1981, A. Cortes, unpublished data), A. olivaceus, A. andinus and A. longipilis are insectivorous/omnivorous (Meserve 1978, Meserve 1981, Contreras & Rosi 1981), P. darwini and 0. longicaudatus are frugivorous/ granivorous (Meserve 1978(Meserve , 1981) ) and P. magister is herbivorous/insectivorous (Pizzimenti & DeSalle 1980).In contrast, the 13 species of North American heteromyids in the present study are highly granivorous (MacMillen & Hinds 1983, Hinds & MacMillen 1985).Thus, I) different food habits of Chilean and North American rodents do not seems to be important in the validation of the water regulation efficiency model,i.e. the index of MR-WB, or 2) because we maintained rodents with the same diet, they responded in a similar way; however, this last aspect seems unlikely since rodents were maintained in the lab for a short time.In contrast, one of the assumptions in the model of MacMillen & Hinds (1983) is that granivory is associated with the aridity and seasonality of precipitations in arid environments, a major factor in the evolutionary history of the family Heteromyidae.Consequently, the water regulation efficiency (Ta @ MWP = EWL) of heteromyids is negatively related to body mass, apparently because these rodents depend on seeds, which sustain both their energetic and hydric requirements.However, we found a low r-value when the MacMillen & Hinds relationship, log Ta @ MWP = EWL, was plotted against log M.An additional inconvenient of this approach is that, at present, it does not enable for a clear discrimination between different patterns of habitat use.
Finally, we think that our new model has some advantages, because it is able to discriminate patterns of water regulation efficiency as a function of the habitat (xeric versus me sic), independent of dietary habits and allows to evaluate variations in the energetic cost of water balance.

Fig. 1 .
Fig. 1.Double logarithmic relationships between evaporative water loss and body mass in rodents from xeric and mesic habitats (seeTable 2 for species identification).

Fig. 3 .
Fig. 3. Relationship between the ratio of metabolic water production to evaporative water loss (MWP/ EWL) and environmental temperature in rodents from Chilean habitats.

TABLE I Number
of individuals (n), body mass (mean± SD) and collection sites (habitats in