Metabolic rate and environmental productivity: Well-provisioned animals evolved to run and idle fast

Sixty-four Peromyscus mice of five species (10–15 individuals each of P. californicus, P. eremicus, P. leucopus, P. maniculatus, and P. melanophrys) were bred from ancestral stocks previously captured at five North American locations characterized by very different rainfalls and productivities (Table 1). Parent mice were obtained from the Peromyscus Genetic Stock Center (Univ. of South Carolina, Columbia), where the stocks had been held for 10–40 generations in captivity. We bred and maintained our mice in the Univ. of California, Los Angeles (UCLA), Health Science Vivarium, in a single room at 27°C, on a 16-h/8-h light/dark cycle. All mice were provided with ad libitum quantities of water and of the same diet [a commercial purified rodent diet based on maltose and casein and manufactured by Dyets (Pittsburgh)]. To further minimize sources of variation, we made measurements only on adult males between 8 and 15 months old, maintaining constant body mass. When mice were not being used for experimental measurements, we housed them in same-sex pairs in large plastic rodent cages (40 × 25 × 20 cm) that allowed room for running and jumping. During experimental measurements of food intake and digestibility we housed single mice in cages appropriate to their body size: 25 × 15 × 12 cm for the smaller P. eremicus, P. leucopus, and P. maniculatus, 40 × 25 × 20 cm for the larger P. californicus and P. melanophrys. Metabolic rate was measured on individual mice placed in round plastic chambers 20 cm in diameter and 10 cm high. All cages contained environmental enrichment in the form of cotton nesting material and PVC (polyvinyl chloride) pipes for hiding places. Mice were kept on aspen shaving bedding except during digestibility trials, when they were placed on wire grids to enable collection of feces. All techniques were approved by the UCLA Animal Research Committee and met federal standards for animal care and welfare.

Values for terrestrial NPP were calculated for 32 months (1997–2000) by a model (modified CASA: Carnegie–Ames–Stanford Approach) that integrates absorbed photosynthetically active radiation (APAR) from satellite-based remote sensing measurements, with values for efficiency of conversion of radiation to biomass based on physiological and ecological information such as water availability and suitability of temperature for plant growth (17, 23). From maps and collectors' field records, we determined as closely as possible the coordinates of the site where the founders of each laboratory breeding population of the five mouse species had been captured, then we determined the average annual NPP for a square 18 km on a side and centered on this location (Table 1).

Our species differed over a factorial range of 2.4 in body mass, but differences in body mass are by themselves a major cause of differences in metabolic rate. To correct the measurements of metabolic rate to effectively the same body mass for all five species, we analyzed metabolic rate measurements by analysis of covariance (ANCOVA) with body mass as the covariate (JMP statistical software, SAS Institute, Greenville, SC). Alternatively, we also analyzed metabolic rate measurements by scaling them according to the accepted allometric coefficient for Peromyscus (0.73; ref. 12), but that analysis yielded results statistically equivalent to the ANCOVA and so will not be reported separately.

As emphasized by many authors (e.g., refs. 24–26), phylogeny is a potentially confounding factor in any analysis in which one compares species differing with respect to some factor to detect a putative relation between that factor and the measured quantity. For instance, in our comparison of metabolic rates of five Peromyscus species from environments of differing NPP, our measured metabolic rates might be similar or different among species not because of similarities or differences among environments in NPP (the variable of interest to us) but because of the constraints imposed by phylogeny. Each of the five species is more or less closely related to the other four species phylogenetically, and differences in metabolic rate might merely reflect differences in phylogenetic inheritance of shared ancestral metabolic rates. We factored out such effects of phylogeny by the method of phylogenetically independent contrast analysis, determined by using the program COMPARE 4.4 (ref. 27; http://compare.bio.indiana.edu), which uses Martins' (28) adaptation of Felsenstein's (26) method of analysis. Our phylogeny for these five species was developed from published information on Peromyscus systematics (29), with all branch lengths set arbitrarily to 1.0. It turned out that the statistical strength of all five of our principal results (the relations between BMR and NPP, DMI and NPP, DMI and BMR, DDMI and NPP, and DDMI and BMR) remained unchanged after factoring out effects of phylogeny, so we shall not discuss the contrast analysis further. That is, these relationships are not merely a legacy of phylogeny.

After we had adjusted for species differences in body mass by means of ANCOVA, a 1.4-fold range of BMRs among the five species remained. Mass-adjusted BMR was closely correlated with NPP, the productivity of each species' habitat of origin (Fig. 1, P = 0.02, R² = 0.88): that is, NPP accounted for 88% of the variation in BMR.

Water-corrected food intake (DMI) was even more closely correlated with NPP than was BMR (P = 0.01, R² = 0.90) (Fig. 2). Because BMR and DMI were both closely correlated with NPP, they were also closely correlated with each other (P = 0.02, R² = 0.88). Apparent dry-matter digestive efficiency equaled 89% ± 2% and did not vary significantly among the five species. (That is, 89% of the ingested dry matter was actually absorbed by mice, and only 11% passed out undigested in the feces). Hence not only daily food intake, but also digestible dry matter intake (DDMI), was closely correlated with NPP (P = 0.03, R² = 0.82) and with BMR (P = 0.02, R² = 0.86).

In the steady state, for an animal in mass balance as were our mice at the time of study, daily food intake must equal daily energy expenditure when both are expressed in the same units (e.g., calories). What could account for the observed differences in daily food intake or energy expenditure among our five mouse species?

Total daily energy expenditure (DEE) of our caged animals (which is analogous to FMR of a free-living animal) may be partitioned into three components: energy expenditure while resting and not digesting (BMR); the energetic costs of digestion (termed heat increment of feeding, HIF); and the energetic costs of activity. For our mouse species, measured BMR values were 53–68% of total energy input as estimated from food intake times the caloric content of food, when all values are expressed as metabolizable energy. We did not determine HIF for our mice, but measurements for other rodent species eating just enough to maintain body mass yield values equivalent to about 12% of DEE (30, 31). The remainder of DEE, constituting about 20–35% of it [100% − (53–68%) − 12%], must be the cost of activity. (Of course, for a wild animal running around outdoors and not living in a small cage, DEE and activity would be much higher, and BMR would be a much lower percentage and activity a much higher percentage of DEE).

We found BMR and total energy expenditure (equal to total energy intake, measured as food intake) to differ in virtually the same rank order among our species. Because our five species have similar natural diets and similar gut anatomies, it seems unlikely that the percentage contribution of HIF to total energy expenditure varies greatly among the species. Instead, one would expect most of the species differences in DEE unaccounted for by differences in BMR to arise from species differences in activity.

This expectation accords well with our qualitative observations of activity in the course of handling and caring for the mice. The two species with the highest DEE and BMR (P. leucopus and P. maniculatus) were jumpy, bit frequently and forcefully, escaped readily from their cages, and required a net for recapture in the vivarium. The two species with the lowest DEE and BMR (P. eremicus and P. melanophrys) were placid and calm in the hand. P. californicus was intermediate in both DEE and BMR and in its activity. Our observations accord with those of previous researchers, who noted that P. maniculatus (our species with the highest DEE and BMR) tends to be “more restless” (32) and “nervous” (10) than other Peromyscus species. This is not a surprising observation, because food requirements of “nervous” or “high-strung” domestic animals as well tend to exceed those of their more placid cage-mates or stall-mates.

One would expect an animal eating more food to have greater intestinal capacity for absorbing food nutrients. Hence we measured intestinal brush-border d-glucose uptake capacity as the integrated product of small intestinal mass times glucose uptake rate per unit mass of intestinal tissue, integrated over the length of the small intestine. Species differences in intestinal mass, uptake rate, and uptake capacity all fell in approximately the same sequence as species differences in NPP, BMR, and DDMI. The strongest relations were between small intestinal mass (P = 0.03, R² = 0.75) or uptake capacity (P = 0.03, R² = 0.76) and BMR. That is, species with higher food intake and basal metabolisms tend to have larger small intestines and higher glucose uptake rates.

Two previous laboratory studies (8, 10) of Peromyscus populations from different habitats found a trend for metabolic rate to be lower for mice from arid environments than for mice from mesic environments. A literature review (6) of basal metabolic rates of 497 mammal species, assigned to six terrestrial zoogeographic zones defined mainly by latitude and rainfall, similarly found mammals of xeric environments to have significantly lower BMRs than mammals of mesic environments. Our study agrees with these conclusions but makes the argument more compelling in two respects.

First, we related metabolic rates to a continuously varying habitat measure (NPP) rather than to a coarser assignment of habitats to one of several categories. Our conclusion is that the relevant feature of our habitats associated with low metabolic rate is low productivity.

Second, instead of compiling from the literature metabolic measurements obtained in different studies by different methods, we measured all of the data ourselves by identical methods, using animals born in our laboratory and maintained under identical conditions, including identical diets. Thus, extraneous sources of variation (including nongenetic effects of differing environments and diets in studies of wild-caught animals) were minimized. This minimization of extraneous variation may have contributed to the very high correlation of metabolic rate with environmental productivity that we obtained: NPP explained 82–90% of the variance in metabolic rate after removing effects of body mass. During the 10–40 generations that our mouse stocks had been bred in captivity at the Peromyscus Genetic Stock Center, selection for adaptation to their original natural environments was undoubtedly relaxed. First-generation offspring of wild-caught individuals would presumably have yielded even higher correlations of metabolic rate with environmental productivity.

We found three measures of metabolic rate to be correlated quantitatively (BMR, DMI, and DDMI), and one to be correlated anecdotally (behavioral activity), with NPP. Interspecies comparisons of FMR (approximately equal to DDMI in the steady state) and BMR have shown a correlation of these two measures with each other in rodents (33). That is, animals with access to abundant food have evolved to “run fast” (high FMR and DDMI) and also to “idle fast” (high BMR). The link between BMR and FMR arises because BMR is in effect the maintenance cost of the biological machinery (muscles, heart, intestine, liver, kidneys, etc.) that enables animals to sustain high metabolic rates while active.

Arid environments tend to be characterized not only by low rainfall and low NPP, but also by high variability of rainfall within and between years (34). One might reason that, all other things being equal, animals from highly variable environments should have evolved lower metabolic rates than animals from less variable environments with similar time-averaged productivity, so as to be able to idle their metabolism at a rate low enough that the environment could constantly support it, even at unproductive times. Several authors have suggested that environmental variability rather than environmental productivity is the primary factor responsible for low metabolic rates of arid-zone species (e.g., refs. 6 and 35).

We tested this idea as follows. Data adequate to calculate the long-term variability of NPP itself were not available, but long-term rainfall records were available, and rainfall patterns are a primary driver of variation in NPP (16, 36). Hence we calculated 30-year coefficients of variation (CV) of annual rainfall (see Table 1) for the capture sites of our five Peromyscus populations from data provided by the North Carolina State Climatologist's office, Michigan State Office of Climatology, and National Oceanic and Atmospheric Administration. In our data set CV of rainfall did tend to decrease with NPP, but the relation was very loose and not statistically significant (P = 0.12, R² = 0.57). Whereas BMR (P = 0.28, R² = 0.36) and DDMI (P = 0.4, R² = 0.24) both tended to decrease with CV of rainfall, both relations fell far short of statistical significance, and CV of rainfall explained far less of the variation in BMR or DDMI than did NPP (R² = 0.88 or 0.82, respectively).

That is, in our limited data set of five species, time-averaged mean productivity rather than its variability is the more important driving variable: possibly because our five species were from habitats encompassing a 13-fold range in NPP. We anticipate that a larger data set might make it possible to demonstrate a separate effect of variability, as well as of NPP itself.

The preceding three paragraphs illustrate a qualification worth stressing explicitly. Just because we have succeeded in demonstrating a high correlation between metabolic rate and NPP among five Peromyscus species does not prove that NPP is the only important factor underlying all residual variation in metabolic rate (i.e., variation other than that arising from body mass and higher-level taxonomic affiliation). We selected five species especially suitable for testing the postulated importance of NPP: five generalist omnivorous species very similar to each other in their diets and in their gut morphology. Species with different diets and different gut morphologies may yield different results.

For example, we have made preliminary measurements on another Peromyscus species, P. aztecus, which differs from our other five species in having a very large fermentative cecum, suggesting a dietary strategy of specialized herbivory (37). P. aztecus proved to have lethargic behavior and low BMR and DMI, much lower than expected from the NPP of its natural environment (highly productive subtropical cloud forest). These observations of P. aztecus are in accord with previous arguments that animals consuming diets of low energy density tend to have low metabolic rates (3, 38).

Whereas P. aztecus deviates from predictions based on NPP alone and requires invoking a seemingly separate factor (low-energy diet), both of these factors actually exemplify one common factor: availability of food or of calories. For related species of similar dietary habits but of different habitats, NPP may be a good measure of the differences that they encounter in food availability. But NPP measures the sum of all plant biomass production in a habitat. Species specialized on different diets may encounter different productivities of their particular food classes in the same habitat. For instance, species such as P. aztecus, specialized on low-energy diets, experience the world as poorer than do generalist omnivores living in the same habitat.

As we mentioned in the Introduction, some previous studies have noted apparent correlations of metabolic rate with the latitude, altitude, or temperature of an animal's habitat of origin. Productivity and food availability vary in complex ways among habitats differing with respect to each of these three variables. It remains to be seen whether variation in productivity or food availability can account for these apparent correlations of metabolic rate with latitude, altitude, or temperature. Such tests will be best carried out by the two methods used in this paper: calculating NPP for the actual habitat of origin, and measuring metabolic rates of laboratory-maintained stocks of closely related species or populations differing in their habitat of origin but maintained identically in the laboratory. These two conditions did not apply to the previous studies, making it difficult to interpret them. If such future studies yield results similar to those of our study, then three factors may largely account for the enormous range of species variation in metabolic rate: body mass, higher-level taxonomic affiliation, and the conditions of environmental productivity or food availability under which an animal has evolved.