High muscle mitochondrial volume and aerobic capacity in a small marsupial (Sminthopsis crassicaudata) reveals flexible links between energy-use levels in mammals

Marsupials (Metatheria) are the sister clade of placentals (Eutheria). Together they comprise the Theria or advanced mammals and they have many anatomical and physiological characteristics in common, which largely reflect ancestral traits that evolved prior to the divergence of these two groups about 148 million years ago (Bininda-Emonds et al., 2007). Differences between the two groups, such as their distinct reproductive features, reflect divergent evolutionary trajectories during their long, separate histories. Another area where differences seem to have occurred concerns metabolic patterns. Marsupials have relatively low basal metabolic rates (BMRs), generally about 75% of placental values, and historically were seen as being less competent at thermoregulation and also as ‘low energy’ mammals (Martin, 1902; Dawson, 1973). While distinctions regarding the thermal biology of these clades have been long discarded (Dawson, 1973; Dawson, 1989), debate about differences in metabolic capabilities of marsupials has persisted in some disciplines (e.g. McNab, 1980; McNab, 2005). Here, we expand on investigations of the aerobic capacities of marsupials and also focus on the functional structures that underlie the capacity for oxygen use in their muscles. Our aim was to put these into perspective relative to the features that have recently been established for the structure–function relationships underlying the aerobic capacities of the placental mammals (Weibel et al., 2004).

It has become apparent that some marsupials have substantial aerobic capabilities that result from a relatively large factorial aerobic scope (fAS), such that they achieve levels of maximum oxygen consumption () similar to those seen in placentals (Dawson and Dawson, 1982; Hinds et al., 1993). Recent data (Kram and Dawson, 1998; Dawson et al., 2004) for Macropus rufus (Family Macropodidae), the red kangaroo, are notable because, despite a relatively low BMR, its extreme fAS of 54 results in a or maximum metabolic rate (MMR) equivalent to the high levels reported in a group of placental mammals, such as dogs and horses, that were categorised as ‘athletic’ (Taylor et al., 1987; Weibel et al., 2004). Underlying this capability in M. rufus is a large mass of locomotor muscles that have comparatively high mitochondrial and capillary volumes (Dawson et al., 2004). Another hopping marsupial, Bettongia penicillata, a rat-kangaroo (Family Potoroidae), though much smaller (body mass, M_b, 1 kg), also shows an elevated fAS of 23 and a markedly high MMR (Seeherman et al., 1981; Webster and Dawson, 2012). Again, this is associated with a large skeletal muscle mass that has relatively high muscle mitochondrial volume densities and both a large total capillary volume and a large total capillary erythrocyte volume (Webster and Dawson, 2012). Overall, the muscle and cardio-respiratory structural features of M. rufus and B. penicillata are identical to those previously reported for ‘athletic’ placental mammals of equivalent sizes (Weibel et al., 2004). Notably, the ratio between MMR and total muscle mitochondrial volume (~5 ml O₂ min⁻¹ ml⁻¹) is, as initially proposed (Hoppeler and Lindstedt, 1985), consistent in placentals (Weibel et al., 2004) and macropodiform marsupials (Webster and Dawson, 2012).

Macropus rufus and B. penicillata belong to the specialised monophyletic suborder Macropodiformes (kangaroos, wallabies and rat-kangaroos) (Meredith et al., 2008), but do they differ aerobically from other marsupials? Evidence for ‘athletic’ level aerobic capacities in other marsupials is strong and comes from disparate studies of their cardio-respiratory features (e.g. Dawson and Needham, 1981; Hallam et al., 1989; Hallam et al., 1995; Agar et al., 2000; Dawson et al., 2003). The generality of this assertion is not certain because, while Hinds and colleagues (Hinds et al., 1993) measured comparatively high fAS during locomotion for several smaller species of marsupial, their reported MMR values did not reach ‘athletic’ placental levels. However, the MMR value obtained (Hinds et al., 1993) for B. penicillata was lower than those obtained on more extensively trained animals (Seeherman et al., 1981; Webster and Dawson, 2012). The only other marsupial for which comparable data are available is a South American opossum, Monodelphis domestica (Family: Didephidae) (M_b≈90 g). It also has a comparatively large fAS (Dawson and Olson, 1988; Schaeffer et al., 2003) but its MMR does not reach an ‘athletic’ level, though the mitochondrial volume densities of its skeletal muscles are not lower than generally seen in similar sized placentals (Schaeffer et al., 2003). However, there is support for relatively high aerobic capabilities extending to smaller marsupials. This comes from studies of field metabolic rate (FMR). Marsupials and placentals show distinctly different patterns of variation of FMR with M_b but, intriguingly, small marsupials (M_b<125 g) have higher FMRs than placentals of similar size (Nagy et al., 1999; Capellini et al., 2010). These differences are particularly marked among the smallest marsupials; at a M_b of ~15 g, the FMR of S. crassicaudata is almost double that of the predicted placental value (Nagy, 1988).

To clarify the factors underlying this disparity and to further our understanding of the metabolic patterns and aerobic capabilities among small marsupial species from different phylogenetic clades, we studied S. crassicaudata (Family: Dasyuridae). This species (M_b≈15 g) is an active, quadrupedal insectivore that has often been used as a model for Australian dasyurid marsupials and was among the species examined by Hinds and colleagues (Hinds et al., 1993). We focused first on gaining an accurate determination of its MMR and then ascertained how MMR related to its muscle content and muscle mitochondrial features such as volume density. Because of its high MMR levels, we predicted that it would share characteristics that match what has previously been found in the Macropodiformes, thereby indicating a generally higher aerobic capability among marsupials, matching that seen in placentals designated as ‘athletic’ (see Weibel et al., 2004). If so, it would point to a fundamental structure–function relationship for oxygen delivery to muscles evolving in or before the earliest mammals. Further, such information provides the opportunity to examine the presumed relationships between BMR, MMR and FMR in mammals. This is significant in view of the idea that BMR is a good predictor of energy budgets, which is based on the notion that the allometric relationship between M_b and BMR locks in other metabolic levels (West et al., 1997; West et al., 1999; Brown et al., 2004).

Fat-tailed dunnarts, S. crassicaudata (Gould 1844) of the Family Dasyuridae (Krajewski et al., 2012) are mouse-sized insectivorous marsupials that inhabit the surface of open habitats, usually in semi-arid and arid regions of Australia. They are active nocturnal predators, catching relatively large, invertebrate prey, such as crickets and beetles (Morton, 1978). Our investigations were in two parts. (1) An analysis of the mitochondrial characteristics of locomotor muscles of S. crassicaudata, carried out at the University of New South Wales, Sydney. The animals used in this study were derived from colonies maintained at the University of Adelaide and University of Wollongong. (2) A study of aerobic capacity of S. crassicaudata during running, undertaken at the University of Wollongong, using animals from their breeding colony that was established 3 years earlier from free-living animals collected in western Queensland. This study was carried out under approval given by the University of New South Wales and University of Wollongong Animal Care and Ethics Committees (project approval number 00-17 and 04/06, respectively).

During investigation and prior to killing, animals were kept at an air temperature (T_a) of 23±0.4°C, with a 12 h:12 h light:dark cycle (lights on at 06:00 h). They were housed individually in clear plastic containers (55×38×20 cm), which were fitted with wire tops; bedding of straw and shredded paper was provided. A mixture of dried cat food (moistened) and canned dog food was provided ad libitum; this was supplemented with live crickets and vitamin drops (Penta-vite infant vitamins, Bayer, Pymble, NSW, Australia). Water was available at all times.

To assess the muscle mitochondrial parameters of the skeletal muscle of the whole body, we followed a sampling procedure comparable to that developed previously (Hoppeler et al., 1984) and used in other studies of marsupials (Dawson et al., 2004; Webster and Dawson, 2012). The musculature of S. crassicaudata was divided into five functional regions, these being head/neck, foreleg, trunk, hindleg and tail. Animals were killed by gassing (carbon dioxide) and weighed to the nearest 0.01 g on an electronic balance (Sartorius AG, Goettingen, Germany) directly before dissection. Four animals were dissected to estimate total skeletal muscle mass and the proportions of muscle in the five body regions. The contributions to M_b of skin, heart and the digestive tract were also determined. In a further five animals the heart plus seven skeletal muscles, including the diaphragm because of its role in ventilation, were then dissected out and small blocks sampled for electron microscopy. The skeletal muscles used were randomly selected, one coming from each region except the hindleg, where two muscles were selected. These were, m. trapezius (head/neck), m. deltoid (forelimb), m. pectoralis minor (trunk), m. multifidi lumborum (trunk), m. gluteus maximus and m. quadriceps (hindleg), and m. multifidi lumborum (tail).

Two small blocks, no greater than 2 mm thick, were randomly cut from each muscle whilst being bathed in a drop of cold glutaraldehyde fixative solution (2.5% in 0.1 mol l⁻¹ sodium cacodylate buffer, pH 7.4). Sample blocks were transferred to vials containing the buffered glutaraldehyde fixative solution for proper immersion fixation for a minimum of 4 h. Sample preparation thereafter followed the method of previous studies (Dawson et al., 2004; Webster and Dawson, 2012), with the blocks ultimately being embedded in Spurr's resin (a slow cure, low viscosity epoxy) over a long infiltration period (3–4 days) and cured at 60°C for 48 h. Ultra-thin sections of ~60–80 nm were cut for each muscle sample using glass knives mounted on a Reichert-Jung Ultracut microtome (Leica Microsystems, Vienna, Austria). The sections were placed onto copper grids (200 square mesh) and were immediately stained with uranyl acetate in 50% ethanol for 10 min.

Grids were viewed at 10,000× magnification with either a Hitachi 7000 (Tokyo, Japan) or JEOL 1400 (Tokyo, Japan) transmission electron microscope (TEM). Ten grid squares were selected per sample block using a systematic random sampling method (Howard and Reed, 1998). Digital micrographs were taken in the top left corner of the grid squares using an Olympus SQ (Tokyo, Japan) digital camera and software package AnalySIS (attached to the Hitachi 7000 TEM) or a Gatan (Pleasanton, CA, USA) digital camera and software package Gatan Digital Micrograph (attached to the JEOL 1400). For each animal, a total of 160 micrographs were taken (10 micrographs per block × two blocks per muscle × eight muscles).

To ensure that maximum aerobic capacity (MMR) was achieved, we followed procedures published elsewhere (Seeherman et al., 1981); such procedures were used in comparable studies on placental mammals (see Weibel et al., 2004). The essence of these procedures was extensive treadmill training (running) that ensured an accurate and reproducible MMR. Seeherman and colleagues found that at least 2–6 weeks of training were needed for this to be achieved for most of the species that they investigated (Seeherman et al., 1981). We trained S. crassicaudata for treadmill running for 6–8 weeks by exercising them at speeds of up to 1.5 m s⁻¹, generally on alternate days. The highest training speed at which an animal could maintain 5 min of constant running, following an initial speed adjustment period of 30 s, was used during the measurement of MMR; such speeds ranged between 1 and 1.5 m s⁻¹. The MMR obtained was the highest 2 min period of instantaneous oxygen consumption when an animal ran for at least 5 min. The method for obtaining instantaneous oxygen consumption (Bartholomew et al., 1981) involved initially determining the washout characteristics of the chamber, at the flow rate used, by tracking the dynamics of a sudden pulse of O₂-depleted air followed by an immediate return to room air.

For actual measurement, one S. crassicaudata was contained within an inverted 1.2 l rectangular plastic container on a stationary treadmill belt. The treadmill speed was then adjusted to that required, ie. the highest training speed for that individual. A constant airflow of 2.0 l min⁻¹ was aspirated through the container at all treadmill speeds. Air entered through two small holes in the front of the chamber and also through the bottom edges of the chamber in contact with the belt. Flow rate was monitored with a Sierra Top-Trak mass-flow meter (Sierra, Monterey, CA, USA). Oxygen content of inlet and outlet air was measured using a Sable Systems FC-1 oxygen analyser (Las Vegas, NV, USA), with a detection sensitivity of 0.0005%. Water and CO₂ were removed from sampled air prior to gas analysis using Drierite and soda lime, respectively. The O₂ throughout this exercise period was determined using appropriate corrections for the system configuration (Hill, 1972). Values were adjusted for variations in chamber air leakage at different treadmill speeds. Air leakage was determined by delivering a gas mix into the chamber via a mass flow controller with a percentage O₂ similar to that while a S. crassicaudata was running. Readings were first taken while the treadmill belt was stationary and then recorded at each belt speed used in the MMR determinations. Corrections ranged from 7% at the lowest running speed to 14% at the highest.

Comparisons between muscles were analysed using one-way analysis of variance (ANOVA). A Student–Newman–Keuls (SNK) multiple-range test was applied when significant differences were indicated by the ANOVA (using Statistica for Mac, StatSoft, Tulsa, OK, USA). Values are given as means ± s.d. Regression analyses were carried out using Microsoft Excel (Microsoft, Redmond, WA, USA).

The mean body mass of S. crassicaudata in this investigation was 15.0 g (Table 1), which is similar to the mass of wild-caught animals. The contribution of skeletal muscle to body mass in S. crassicaudata was estimated to be 32.3±1.96% (Table 1). The size of the heart and the contributions to body mass of some other major components, such as skin and the digestive system, are also shown in Table 1.

In S. crassicaudata, Sv(im,mt) varied little between the muscle tissues investigated. Values were: heart, 34.0±7.8 m² cm⁻³; diaphragm, 36.8±5.8 m² cm⁻³; m. gluteus maximus, 35.8±7.3 m² cm⁻³; m. deltoid, 35.0±7.5 m² cm⁻³.

The content of mitochondria in heart and a range of skeletal muscles from S. crassicaudata is shown in Table 2 as Vv(mt,f). The heart and diaphragm contained significantly higher densities of mitochondria than the skeletal muscles; that of the heart was 33.9±2.7%, with that of the diaphragm being 21.1±2.9% (F_7,1=47.15, P=0.0001). While values for the other muscles ranged from 14.5±2.3% for the m. multifidi lumborum of the trunk and tail to 10.6±1.4% for the m. pectoralis muscle, the differences between them were not significant (F_5,1=2.116, P=0.1).

The muscle content throughout the body showed significant differences between regions (Table 3; F_4,1=51.86, P=0.0001). Both the hindleg and trunk had significantly more muscle than other regions and together comprised 60.6% of the total skeletal muscle mass. The foreleg and head/neck regions equally made up most of the residual body muscle mass, whereas the tail contained little muscle. Although Table 3 shows that there are significant differences in Vv(mt,f) in muscle regions (F_4,1=4.849, P=0.01), the differences are relatively small and the regional mitochondrial volumes V(mt,m) largely reflect regional muscle masses. There is a significant difference in V(mt,m) across muscle regions (F_4,1=137.1, P=0.0001). The trunk has a significantly larger volume (Table 3), with the hindleg also having more than each of the remaining regions. The percentage of total muscle mitochondrial volume contained by the regions follows a similar pattern of significant differences (F_4,1=234.1, P=0.0001). The total volume of mitochondria in the skeletal muscle, V(mt), was 0.68±0.064 ml (Table 3).

The mean MMR of S. crassicaudata determined from sustained treadmill running was 4.09 ml O₂ min⁻¹, or 272 ml O₂ min⁻¹ kg⁻¹ (N=8 animals) at an average speed of 1.2 m s⁻¹ (Table 4). The BMR reported in a previous study (Dawson and Hulbert, 1970) was 0.320 ml O₂ min⁻¹ and thus the fAS was 12.8.

The muscle characteristics and aerobic capacity of the marsupial S. crassicaudata (Table 4) mark it as a mammal of high aerobic capacity in relation to other studies (Weibel et al., 2004; Weibel and Hoppeler, 2005). It compares favourably with Apodemus sylvaticus, the European wood mouse, a similar sized placental previously studied in detail (Hoppeler et al., 1984). Apodemus sylvaticus is grouped with the ‘athletic’ as against the ‘sedentary or normal’ mammals by those examining the structure–function relationships that underpin the aerobic capabilities of placental mammals (Weibel et al., 2004; Weibel and Hoppeler, 2005). In the phylogenetically disparate species S. crassicaudata and A. sylvaticus, both the MMR and the total muscle mitochondrial volume, V(mt), are alike (Table 4) but there are differences in the way the two species achieve their high aerobic capacities (MMRs). Notably, these are in the relative volumes of muscle and the Vv(mt,f).

The mean proportion of skeletal muscle in the body of placental mammals, M_m/M_b (%), is 36–38% (Lindstedt and Schaeffer, 2002; Weibel et al., 2004). Sminthopsis crassicaudata with a M_m/M_b of 32.3±1.96% and A. sylvaticus with M_m/M_b of 42.5% (Table 4) fall on either side of this mean, with A. sylvaticus having one of the highest M_m/M_b values in the data set of Weibel and colleagues (Weibel et al., 2004). These authors found that M_m/M_b was independent of body mass, but was consistently higher in the ‘athletic’ group of species. The pronghorn (Antilocapra americana) at 45% had the highest value for a placental; however, the marsupial red kangaroo (M. rufus) has an M_m/M_b value of 47% (Dawson et al., 2004). The lower M_m/M_b of S. crassicaudata compared with A. sylvaticus, however, is offset by its relatively higher Vv(mt,f), which is 14% versus 11% in A. sylvaticus (Table 4). The very similar total heart mitochondrial volumes in the two species reflect this balance. This trait is a reliable predictor of the MMR of equivalent sized species among marsupials (Dawson et al., 2003) and placentals (Karas et al., 1987). The heart masses were 0.79% and 0.78% of M_b, respectively, for S. crassicaudata and A. sylvaticus, while Vv(mt,f) in the hearts of both species was approximately 34%.

The surface area of the inner mitochondrial membranes, S(im,m), has been consistently correlated with the activity of the terminal respiratory chain enzymes in vertebrate groups (Else and Hulbert, 1981), and appears to be functionally linked with aerobic metabolic capacity. The surface density of inner mitochondrial membranes per unit volume of mitochondria, Sv(im,mt), in the muscles of S. crassicaudata is ~35 m² cm⁻³, which is similar to that of other marsupials (Dawson et al., 2004; Webster and Dawson, 2012) and placentals including A. sylvaticus (Hoppeler et al., 1981; Hoppeler et al., 1984; Schwerzmann et al., 1989). As S(im,m) equals V(mt,m) multiplied by Sv(im,mt) (Eqn 2), mitochondrial volume in skeletal muscle can be used as a proxy for S(im,m). The high overall Vv(mt,f) of S. crassicaudata relative to that of A. sylvaticus and those of other placentals compiled by Weibel and colleagues (Weibel et al., 2004) results from high mitochondrial volume densities in all muscles across the body (Tables 2, 3). The Vv(mt,f) of individual muscles, except for the diaphragm, did not vary through the body (Table 2); this would reflect S. crassicaudata's active quadrupedal lifestyle. The pattern differs in the more specialised kangaroos, whereby muscle Vv(mt,f) is markedly higher in the region of the pelvis and lower back where the bulk of the skeletal muscle is also found (Dawson et al., 2004).

These data from S. crassicaudata considerably extend our understanding of the overall aerobic capacities of marsupials relative to those of placentals. Initially, an investigation of the cardio-respiratory allometry in marsupials (Dawson and Needham, 1981) identified them as having the capability for a considerable aerobic capacity. Dawson and Dawson further challenged the notion that marsupials, with their low BMRs, were ‘low energy’ mammals (Dawson and Dawson, 1982). Two small marsupials species that they exposed to cold had generally larger fAS values, 8–9 as against 4–6 for similar-sized placental species, and aerobic capabilities equivalent to those of the placentals. Data from Hinds and colleagues further highlighted relatively high fAS values in a range of marsupials (Hinds et al., 1993). In response to cold, marsupials and placentals were able to increase aerobic metabolism above BMR by 8.3 and 5.1 times, respectively; values during locomotion were almost twice those observed in the cold (Hinds et al., 1993) and fAS values were again higher in marsupials (17) than in placentals (13.5). However, subsequent locomotor investigations indicate that these fAS values were underestimated in marsupials (see below).

The aerobic factorial scope of M. rufus is of the order of 54 (Dawson et al., 2004). How could this be so much greater than the value of 17 given by Hinds and colleagues (Hinds et al., 1993) for the fAS of marsupials during locomotion? The answer comes from investigations on placentals (see Weibel et al., 2004; Weibel and Hoppeler, 2005). Allometric equations from these studies show that MMR is more loosely associated with BMR than was previously considered. When MMR is plotted against M_b for placentals, two distinct patterns occur (Weibel et al., 2004). One group of species has a relatively high MMR while most other species tend to have a distinctly lower MMR; the former was designated ‘athletic’ and the latter ‘sedentary’ (Fig. 1). Furthermore, these MMR patterns vary with M_b in a different manner from that of BMR. While the allometic equations usually used for BMR have an exponent of 0.75, the exponents found for MMR of ‘athletic’ and ‘sedentary’ placentals were much steeper, 0.942 and 0.849, respectively (Fig. 1). Weibel and co-workers have shown that MMR is largely set by the energy needs of active cells, primarily those in muscle, during maximal work and that total skeletal muscle mitochondrial volume, V(mt), is a superior proxy for this (Weibel et al., 2004). ‘Athletic’ species had greater V(mt) than ‘sedentary’ species, which was due to either greater M_m/M_b and/or higher Vv(mt,f). Overall, as initially proposed (Hoppeler and Lindstedt, 1985), there was a strong and consistent correlation between MMR and V(mt) (Fig. 2). Our previous studies (Dawson et al., 2004; Webster and Dawson, 2012) have shown that both the hopping marsupials have MMRs that fall within the ‘athletic’ grouping in relation to M_b (Fig. 1), and that the relationship between MMR and V(mt) is indistinguishable from that of placentals (Fig. 2). Given the large evolutionary distance and the disparity in body form between modern placentals and the kangaroos and rat-kangaroos, we were somewhat surprised to find comparable relationships. The volume of muscle, its total mitochondrial content and its overall vascular supply were essentially identical in the Macropodiformes to values seen in ‘athletic’ placental mammals.

Thus, our data for S. crassicaudata provide a wide size range over which marsupials have aerobic capabilities that are essentially similar to those of ‘athletic’ placentals (Fig. 1), despite the significantly lower BMR of these marsupials (Table 4). Notably, while V(mt) is high in S. crassicaudata, the relationship between MMR and V(mt) follows the general mammalian pattern (Fig. 2). Does a pattern of ‘athleticism’ pertain for most other marsupials or do marsupials also have variable aerobic potentials, as do placentals (Weibel et al., 2004; Weibel and Hoppeler, 2005)? Available information is equivocal in regard to this question. The only marsupial for which comparable information is also available on MMR and on muscle and muscle mitochondria volumes (Table 4) is the grey short-tailed opossum (Monodelphis domestica, Family: Didelphidae) (Schaeffer et al., 2003). The MMR of M. domestica is relatively low (Table 4, Fig. 1) and falls in line with those of the ‘sedentary’ placentals, not with ‘athletic’ small mammals such as the placental A. sylvaticus and the marsupials, S. crassicaudata and B. penicillata. This is somewhat surprising given its fAS at 13.6 is relatively large, but this mostly reflects it having a BMR that is low, even for a marsupial (Table 4). The BMR of M. domestica is approximately 70% of the value predicted for a marsupial of its mass (including other didelphids) from allometric equations (Dawson and Hulbert, 1970; Withers et al., 2006). However, the relationship between MMR and V(mt) in M. domestica is similar to that of mammals generally (Fig. 2) and its Vv(mt,f) is also relatively low (Table 4).

Apart from the results for M. domestica, other data suggest that high aerobic capacity may be a general characteristic of marsupials. For example, marsupials tend to have larger hearts than placentals (Dawson et al., 2003), a trait that benefits attaining high MMR. Also, the data collected by Hinds and colleagues (Hinds et al., 1993) corroborates the greater aerobic potential of marsupials when it is examined in detail. The MMR of marsupials during locomotion that they report are from untrained animals, but still, with their expanded fAS values, they mostly exceed those of trained ‘sedentary’ placentals (Fig. 1). Full treadmill training presumably would increase the MMR of many of these species to ‘athletic’ levels, as we found for S. crassicaudata (Fig. 1). A clear supporting framework for these abilities is apparent in other species so far examined. As in ‘athletic’ placentals (Weibel et al., 2004; Weibel and Hoppeler, 2005), the peak aerobic demands associated with maximum energy output by muscle are met via the commensurate, matched oxygen supply system from the lungs to the muscle mitochondria via the expanded supply of erythrocytes. This is the concept of symmorphosis (Weibel, 2000) and it also pertains to M. rufus and B. penicillata (Dawson et al., 2004: Webster and Dawson, 2012). Following this concept, there are numerous other studies that lend support for a generally high MMR in marsupials. These examined lung structure and function of the respiratory system (Dawson and Needham, 1981; Hallam et al., 1989; Chappell and Dawson, 1994; Dawson et al., 2000b), heart structure and capacities (Dawson and Needham, 1981; Dawson et al., 2003), blood oxygen affinities (Hallam et al., 1995) and relative haematocrit levels (Agar et al., 2000).

Broadly then, clades of mammals, both placental and marsupial, have evolved elevated aerobic capacities that can be sustained by small species for at least several minutes and for relatively longer periods in larger species. The evolutionary forces behind such elevated capabilities are likely to be diverse, but the predator–prey ‘arms race’ (Vermeij, 1987) initially comes to mind. The MMR that a mammal can attain is clearly determined by the functional characteristics of muscle mitochondria, for which V(mt) is an appropriate proxy (Fig. 2). V(mt) results from various mixes of M_m/M_b ratios and V(mt,m) levels of individual muscles, which can also vary markedly. In regard to the link between MMR and BMR, it seems to be much more loose than previously accepted. The patterns differ considerably between marsupials and placentals, as indicated by their differing fAS values. Although fAS shows much plasticity (consider the value of 54 for M. rufus), there is an apparent upper limit to MMR based at the ‘athletic’ level (see Weibel et al., 2004; Weibel and Hoppeler, 2005), which we have shown is also reached by marsupials. Some mammalian groups may have evolutionarily varied their energetic profile by varying their basic energetic structure, i.e. their BMR. For example, a relatively low BMR in M. domestica is reflected in a low MMR (Table 4), which is seen in the converse in red-toothed shrews of the subfamily Soricinae such as Blarina brevicauda and Sorex araneus (Dawson and Olson, 1987; Poppitt et al., 1993). However, underlying patterns are common to marsupials and placentals, and indicate that the basic structure–function framework for mammalian aerobic capabilities is ancient. It must at least predate the divergence of the therians.

The high MMR of S. crassicaudata does not completely explain the unusual patterns in the allometry of FMR in marsupials and placentals, whereby small marsupials have higher FMRs than placentals (Koteja, 1991; Nagy et al., 1999; Cooper et al., 2003; Capellini et al., 2010). The BMR of S. crassicaudata is 75% of that predicted for a similar sized placental, yet its FMR at ~7 times BMR (Nagy, 1988) is almost double the predicted FMR for a placental (Fig. 3). In the context of its high MMR, via a fAS of 13, S. crassicaudata has ample aerobic capacity for such a FMR (Fig. 3). The reasons behind the high FMRs of small marsupials are conjectural, but the fact that most small marsupials are insectivores/carnivores could be an underlying feature. Note, that while A. sylvaticus has a high MMR, this omnivorous rodent has a FMR of only ~3.2 times BMR (Speakman, 1997). Nagy and co-authors highlight the plasticity of FMR in mammals and point to numerous causes (Nagy et al., 1999; Nagy, 2005).

The fact that FMR and MMR, with its clear connection to V(mt), may not be closely linked via BMR in mammals is highlighted by energetic profiles displayed among marsupials. As with placentals, these show marked impacts associated with M_b (Fig. 3, Table 5) and it is instructive to compare the overall data for S. crassicaudata with that for the large kangaroo, M. rufus (Fig. 3). While the BMR of M. rufus is ~75% of that of a placental, its FMR is low, 50% of that predicted for a placental (Munn et al., 2008). That M. rufus, with a fAS of 54, has one of the highest mammalian MMRs highlights the looseness in connections between the energy ‘levels’ of mammals. The patterns in the levels of energy use among mammals that we have clarified also robustly contest the proposal that design features of the O₂ transport system lock in an allometric exponent of 0.75 for the relationship between M_b and BMR (West et al., 1997; West et al., 1999) that extends mechanistically to MMR and FMR, as in the ‘metabolic theory of ecology’ (Brown et al., 2004).

Staff of the University of New South Wales Electron Microscope Unit provided much instruction on processing samples for electron microscopy and the use of two models of transmission electron microscopes. Mrs Sigrid Fraser of the UNSW Electron Microscope Unit performed sample processing not performed by the authors.

LIST OF SYMBOLS AND ABBREVIATIONS
 
• d

  density of muscle

 
• fAS

  factorial aerobic scope

 
• M_b

  body mass

 
• M_m

  muscle mass

 
• S(im,m)

  total surface area of inner mitochondrial membranes

 
• Sv(im,mt)

  surface density of inner mitochondrial membranes per unit volume of mitochondria

 
• V(mt,m)

  mitochondrial volume of individual muscles (or muscle regions)

 
• V(mt)

  total mitochondrial volume of skeletal muscle

 
• 

  maximal aerobic oxygen consumption

 
• Vv(f,m)

  volume fraction of muscle occupied by muscle fibres

 
• Vv(mt,f)

  volume fraction of mitochondria