## Philip C. Withers, Christine E. Cooper, Shane K. Maloney, Francisco Bozinovic, and Ariovaldo P. Cruz Neto

Print publication date: 2016

Print ISBN-13: 9780199642717

Published to Oxford Scholarship Online: November 2016

DOI: 10.1093/acprof:oso/9780199642717.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2017. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a monograph in OSO for personal use (for details see http://www.oxfordscholarship.com/page/privacy-policy).date: 17 January 2018

# Physiological Adaptations to Extreme Environments

Chapter:
(p.290) 4 Physiological Adaptations to Extreme Environments
Source:
Ecological and Environmental Physiology of Mammals
Publisher:
Oxford University Press
DOI:10.1093/acprof:oso/9780199642717.003.0004

# Abstract and Keywords

Chapter 4 describes how the plastic and flexible bauplan of mammals is adapted for extreme environments and extreme activities. It first examines how mammals survive in extreme cold environments, including polar regions, and then hot and dry environments, including the harshest deserts. Adaptations are described for mammals that live underground in often hypoxic, hypercapnic, and humid environments; at high altitudes in a hypoxic and cold environment; and underwater, where they must cope with extended apnoea and limited oxygen stores as well as high pressures in very deep ocean environments. How various mammals are adapted for specialized cursorial or brachiating locomotion, migration, and have occupied the aerial environment as gliders—and bats as powered fliers—is then described. Finally, the chapter explores how various mammals are able to exploit food sources that are generally difficult to digest (i.e. diets including keratin, bone, waxes, chitin, and plant cells containing cellulose, hemicellulose, and plant secondary metabolites).

The general bauplan of mammals, in particular their morphological and physiological characteristics, has proven to be extremely flexible. From a small, nocturnal, insectivorous ancestor, an amazing array of sizes, shapes, and physiologies have evolved, including species able to exploit some of the most extreme environments on Earth. Mammals (and birds) inhabit the coldest environments, including polar regions and high altitudes; hot and dry environments, including the harshest deserts; and live underground in fossorial environments. Many mammals are semi-aquatic or aquatic, and some, at least temporarily, occupy very deep underwater depths. Some mammals occupy the aerial environment; some as gliders and bats as powered fliers. Survival in these harshest of environments requires remarkable adaptations.

# 4.1 Cold Environments

The lowest recorded air temperature (Ta) at the Earth’s surface is $−89.2°C$, measured at Vostok station in Antarctica during the Austral winter of 1983 (Turner et al. 2009). Although this very low Ta occurred in a particularly extreme environment in a location with presumably no natural mammalian inhabitants, sub-zero temperatures occur frequently in a wide variety of habitats, even usually hot desert environments. Polar regions and high altitude locations typically experience winter temperatures $<−35°C$, and mean monthly winter temperatures of −50 to $−60°C$ have been recorded for some areas of eastern Siberia and north-western Canada (Scholander et al. 1950a). These regions may only rarely reach temperatures above freezing during summer, but mammals are a conspicuous element of the fauna of these very cold regions year-round. The endothermic bauplan of mammals provides the mechanism to maintain a high and stable body temperature (Tb), but at the energetic cost of sustaining a high metabolic heat production. For many mammals in moderately to very cold environments, their suite of physiological (p.291) and morphological characteristics, along with their ecological niche, enables them to remain active and endothermic at a sustainable metabolic cost year-round. However, for some mammals, this energetic cost can become temporarily too difficult to sustain if the required rate of heat production exceeds the maximal sustainable metabolic rate (e.g. at very low Ta or during periods of low food availability). The solution for non-migratory mammals under such circumstances is to become inactive in a more moderate refugia and/or enter periods of prolonged hibernation.

## 4.1.1 Endurers

Many mammals remain active during winter, even in extremely cold environments. Cold winter temperatures in high latitude and altitude habitats require enhanced metabolic heat production, at a time when food accessibility may be reduced. Although challenging for all endothermic homeotherms, small non-hibernating mammals can be most impacted by extreme winters, as small mammals have a large surface area to volume ratio, limited scope to increase fur thickness and therefore insulation, and high mass-specific energy requirements. However, a number of medium and large, as well as small mammals—such as voles, hamsters (Cricetidae), and even shrews (Soricinae; with body mass of < 6 g)—remain active at air temperatures $<−30°C$. Although small and large mammals have differing approaches to the seasonal acclimatization required for them to remain active in extreme cold environments, both are based on two general responses: increased thermogenesis and decreased heat loss (Heldmaier 1989).

Some small non-social mammals, such as voles (Clethrionomys and Microtus spp.), shrews (Cryptotis pava) and mice (Peromyscus leucopus and P. maniculatus), form social aggregations during winter. They inhabit communal subnivean nests that can be substantially warmer than ambient conditions and gain thermal and energetic advantages from huddling, reducing heat loss by decreasing their collective surface area to volume ratio (Marchand 2014). For example, nests of taiga voles (Microtus xanthognathus) containing five to ten individuals can be as much as 12°C warmer than the surrounding soil (−3 to $−5°C$) and 25°C warmer than surface air temperature (mean −5 to $−23°C$) in the field.

In the laboratory, the mean nest temperature of individual voles was lower than that of groups of voles. Voles in both the field and laboratory never all simultaneously vacate the nest, ensuring that nest temperature always remains above ground temperature (Wolf & Lidicker 1981). Winter sociality may also provide opportunities for cooperative defence against predation and guarding of food stores, but can be associated with costs such as increased conspicuousness to predators and higher parasite loads and disease (Wolf & Lidicker 1981; Marchand 2014). Seasonally social mammals may undergo substantial reductions in winter body mass, which presumably reduces total food requirements during winter and limits (p.292) intraspecific competition. Added insulation from nest building and from restricting activity to the subnivean microclimate also reduces energetic requirements (Chappell 1980). For example, voles build nests from stems and leaves, and the insulation provided by this nesting material reduces heat loss to the environment; a colony of some 19 Brandt’s voles accumulate approximately 940 g of winter nesting material (Benedict & Benedict 2001; Zhong et al. 2007). Remaining on the surface can increase energetic costs by 15–25% compared to remaining under the snow, and on cold, clear nights this cost can increase to 40–50%, as temperatures are lower, wind speeds higher, and radiative heat loss is enhanced on the surface compared to the subnivean environment (Chappell 1980).

Small mammals may hoard food to provide an energy supply during cold winter periods when food sources may be limited and ambient conditions become unfavourable for foraging (Marchand 2014). Food, consisting of leaves, grasses, stems, roots, rhizomes, tubers, fungi, and lichens, is stored in piles on the ground underneath snow cover, in underground chambers, amongst rocks or boulders, in tree cavities, and in birds’ nests (Benedict & Benedict 2001). For example, Brandt’s vole (Lasiopodomys brandtii), which inhabits grasslands and steppes of Mongolia, China, and Inner Mongolia, in regions where winter temperatures may be as low as $−40°C$, do not hibernate, but store food in the form of plant matter (primarily the prairie sagewort, Artemisia frigida) in underground chambers. Groups with a mean number of 16–22 individuals occupy a territory of about 25 m diameter, storing 4.4 to 6.3 kg of food during autumn, with larger groups storing more food (Zhong et al. 2007). A combination of reduced thermoregulatory demands from communal nesting and considerable accumulation of stored food limits winter foraging requirements. Energetic calculations for Benedict’s montane voles (Microtus montanus) indicate that a group accumulates sufficient food in 3 weeks to support one individual for > 22 days without any additional foraging (Marchand 2014).

Medium and large mammals also have behavioural adaptations that impact on their physiological response to cold and enhance their ability to withstand extremely cold environmental conditions. Amongst primates, huddling, basking, sheltering in caves or snow-covered trees, and postural adjustments such as pressing the furred limbs to the ventral surface of the body, have all been observed (Hori et al. 1977; Zhang et al. 2007). Japanese macaques (Macaca fuscata) are the most northerly distributed monkey species, surviving in snow-covered mountains and forests during winter, with some areas of their distribution reaching Ta as low as $−20°C$ (Hori et al. 1977). They have a thick winter pelt that provides sufficient insulation to maintain a sub-zero lower critical temperature when acclimatized to winter conditions. Famously, some Japanese macaques inhabiting the Jigokudani Monkey Park bathe in a man-made, hot-spring-fed pool with a water temperature of $38−40°C$. Bathing in the hot pool is more frequent in winter than summer (Zang et al. 2007; Figure 4.1), when the monkeys are not fed, and at lower Ta, evidence that this behaviour is indeed thermoregulatory. Dominant females and (p.293) their offspring bathe more frequently than subordinate females and their offspring, suggesting that hot water is a valued resource. Females are more likely to bathe than males and adolescents; females potentially obtain a greater benefit as they have a lower Tb, but also have a greater opportunity to learn this behaviour as they (but not males) maintain site fidelity (Zhang et al. 2007).

Figure 4.1 Use of hot pools by a population of Japanese macaques (Macaca fuscata) is greater in winter (black bars) than in summer (grey bars).

Modified from Zhang et al. (2007).

Some medium and large mammals are extraordinarily cold tolerant; they can reduce their lower critical temperatures to $<−30°C$ during winter (Scholander et al. 1950b). For example, Norwegian and Svalbard reindeer (Rangifer tarandus tarandus and R. t. platyrhynchus) reduced their lower critical temperatures from 0 to $−30°C$ and −15 to $−50°C$ from summer to winter, respectively (Nilssen et al. 1984). Much of this thermal acclimation is due to an increase in the insulation of their pelt and associated reduction in whole-body thermal conductance. The fur of large mammals provides excellent insulation in the cold. For example, muskox (Ovibos moschatus) in winter have surface temperatures approaching ambient (as low as $−40°C$), except for small areas around the eyes, snout, and hooves (Figure 4.2). Hart (1956) measured seasonal changes in pelt insulation of temperate and northern Arctic mammals of 12–52% of winter values.

Figure 4.2 Thermal image of muskox (Ovibos moschatus) in winter, showing near-ambient surface temperatures, reflecting effective thermal insulation, except around the eyes, snout, and hooves.

Photograph provided by A. J. Munn.

Moulting of the summer pelt occurs in autumn; it is replaced by a winter pelt that has different biophysical properties. The winter underpelt is denser, due to the reactivation of secondary derived hair follicles that regress during summer; more underhairs are therefore produced from shared canals during winter. Winter underfur may also be more crimped than that of a summer pelt, more effectively trapping a layer of insulating still air. Winter guard hairs tend to be finer, longer, and contain medullary air spaces that enhance the pelt’s insulative properties (Marchand 2014). Some species, such as the least weasel (Mustela nivalis), varying lemming (Dicrostonyx torquatus), and Arctic fox (Alopex lagopus), also change colour from a darker summer pelt to a white winter pelt, although the functional (p.294) significance of this colour change is likely related to crypsis rather than thermoregulation (Chappell 1980; see 3.2.3.4). For very thick, well-insulating fur, solar heat gain at the level of the skin is limited and colour has minimal impact on thermal balance (Dawson et al. 2013). Small mammals have a limited capacity to increase the insulation of their pelt, as pelt insulation is linearly related to pelt thickness (see 3.2.3.4) and there is a physical limit to how thick a small mammal’s pelt can be. Small Arctic mammals therefore have winter pelt insulation similar to that of tropical mammals. However, for Arctic mammals ranging from the size of an Arctic fox (approximately 5 kg) to that of a moose (Alces alces; approximately 550 kg), winter pelt insulation is independent of body mass. These medium-to-large mammals also have a much more dramatic seasonal increase in insulation (Scholander et al. 1950a; Hart 1956).

In water, fur pelts of many mammals lose most of their insulative properties, as water replaces air between the hairs. For example, heat loss through polar bear (Ursus maritimus) fur increases 20–25 times in ice water compared to air; this increases to 45–50 times for moving water (Scholander et al. 1950a). Despite this, many temperate and Arctic mammals spend all or considerable portions of their time in water with a temperature of $≤0°C$. Polar bears use temporal and regional heterothermy, as well as a large body mass (and thus high thermal inertia) and considerable metabolic heat generation to withstand prolonged periods of swimming in icy water. Their pelt provides limited insulation in water and they don’t use adipose tissue for insulation; enhanced fat accumulation appears to be only for the purposes of energy storage (Pond et al. 1992; Whiteman et al. 2015). (p.295) This presumably reflects their relatively recent exploitation of marine habitats (0.5 MYBP; Liwanag et al. 2012). During periods of swimming, Tb of free-living polar bears has been observed to fall to $<35°C$, presumably an active process achieved via vasoconstriction that provides a cool insulating shell for parts of the core (Whiteman et al. 2015; Figure 4.3).

Beaver (Castor canadensis), muskrat (Ondatra zibethicus), and North American mink (Mustela vision) are medium-sized, semiaquatic mammals that forage in water, even during temperate and subarctic winters. Beaver and muskrat construct lodges that provide a favourable micro-environment, and beaver store food, and fat in their tails, to reduce foraging demands (Marchand 2014). All lose considerable heat in water. The thermal conductance of a mink carcass in water is up to 7.9 times greater than in air, and submerged beaver fur has a 10 times higher thermal conductance compared with that in air (Scholander et al. 1950a; Williams 1986). For beavers, the pelt only accounts for 23.5% of whole animal insulation in cold water. Regional heterothermy, facilitated by countercurrent heat exchange in hind foot and caudal rete mirabilia (see 3.2.4.1) and in opposing veins and arteries in the thoracic and hind leg regions, reduce heat loss, although beavers still lose heat and experience a decline in Tb at a rate of when in water at . This limits their aquatic foraging time; free-living beavers reduce core Tb fluctuations to within 2°C by behaviourally restricting time spent in water (MacArthur & Dyck 1990).

Muskrats and mink also experience declines in Tb during immersion in cold water. For muskrats, dives of only 0.5 to 4 min resulted in a steady decline in Tb, with the degree of Tb reduction, post-dive metabolic rate, and recovery time all positively correlated with submersion time, and negatively correlated with water temperature (MacArthur 1984). For mink, heat loss exceeds heat production within 5 min of swimming, heat storage becomes negative ( at intermediate swimming speeds of 0.36 m s−1), and Tb declines, even at a moderate water (p.296) temperature of 20°C. Winter pelages, however, provide more effective insulation in the water than summer pelages (heat storage ), more efficiently trapping some insulating air within the pelt. Removing this air resulted in more rapid cooling. This insulation, together with regional heterothermy, particularly from vasoconstriction of blood vessels to the paws, and enhanced metabolic rate, permits mink to forage for short periods in cold water (Williams 1986).

Like mink, river otters spend time on land and in the water, although they forage exclusively in the water. The pelt of river otters is very dense, approximately four times that of a muskrat. It is characterized by guard hairs with a well-developed medulla and interlocking shaft scales, along with a dense and crimped underfur, that trap air. Sebaceous gland secretions coating the hair further limit water penetration to the skin. In addition, their high mustelid basal metabolic rate (BMR) presumably reflects a high capacity for metabolic heat production, and countercurrent exchange in the limbs, peripheral vasoconstriction, and a decrease in heart rate limits heat loss during diving (Marchand 2014).

Sea otters (Enhydra lutris) are the smallest marine mammals. Like semi-aquatic mammals, these exclusively marine mustelids rely on a combination of an insulating fur pelt and high metabolism to maintain Tb while almost permanently immersed in cold water (Costa & Kooyman 1982). Sea otters have a high metabolic rate (2.5–3 times allometric predictions; Yeates et al. 2007) and therefore high thermogenic capacity, even compared with other otter species (Kruuk & Balharry 1990). Use of fur rather than blubber for insulation reflects their relatively recent invasion of a marine environment (1.6 MYBP) and their small body size, which would make insulation via blubber impractical. Sea otter fur is exceptionally dense (737–2,465 hair bundles cm−2, with 19–91 hairs per bundle), twice that of river otters, and is coated by oily secretions of squalene (C30H50) from sebaceous glands that prevent water penetrating the pelt to the level of the skin and maintain an insulating layer of air within the pelt. Sea otter pelts have sparse guard hairs that have interlocking scales at the base, aiding in maintenance of the pelage structure. Underfur hairs also interlock, facilitated by their wavy form and scales, to trap still air. The sloping angle of the hairs $(61.9−84.3°)$ also apparently contributes to keeping water from the pelt (Williams et al. 1992).

Maintenance of the pelt requires considerable grooming, > 2 h day−1 (Yeates et al. 2007) to remove debris, restore pelt structure, and disperse glandular secretions throughout the pelt. Grooming consists of two phases: washing the fur and then drying it. At the end of a grooming session, sea otters blow air back into the fur or aerate it by churning the water with their forepaws (Kenyon 1969). Contamination of the pelt can lead to greatly increased thermoregulatory costs and hypothermia (Costa & Kooyman 1982). Interestingly, the pelts of European otters (Lutra lutra) lose their ability to trap insulating air after several sessions swimming in saltwater, and the otters become prone to hypothermia unless they are able to wash in freshwater and restore the insulating properties of the pelt (p.297) (Kruuk & Balharry 1990). This presumably occurs due to salt crystals forming in the fur and disrupting its structure, or the saltwater interfering with the secretions of sebaceous glands and their distribution, or both.

Sea otters rarely leave the water; therefore they are unlikely to have salt crystals form in drying pelts, and they spend considerably longer grooming per day than European or river otters; European otters groom only on land, not in the water (Kruuk & Balharry 1990; Marchand 2014). Use of fur rather than blubber confers several costs for sea otters: air trapped in the fur, although necessary to maintain the pelt’s insulative properties, increases buoyancy and therefore contributes to a high cost of diving compared to other marine mammals. Considerable energy—2.4 mJ day−1, or 15% of an otter’s daily energy expenditure—is also expended on the grooming necessary to maintain the insulative function of the pelt; the metabolic cost of grooming (29.4 ml O2 kg−1 min−1) is higher than that of feeding and is equivalent to that of swimming (Figure 4.4). Grooming also requires 9.1% of the otters’ time budget, but as sea otters spend approximately 40% of their time resting, energetic rather than time costs of grooming are probably more important (Yeates et al. 2007). Sea otter pups can neither swim nor groom despite being born and raised at sea, and so are reliant on their mothers for both floatation and grooming to prevent drowning and hypothermia. Female sea otters spend approximately (p.298) 13% of their time grooming their pups to ensure a waterproof pelt. The layer of air trapped in the pup’s pelt is also essential for the pup to maintain buoyancy while the female dives to forage; it is left floating on the ocean surface for these brief periods despite an inability to swim (Cortez et al. 2015).

Figure 4.3 The core body temperature of a wild, free-living polar bear (Ursus maritimus) decreases from resting (r, white symbols) during swimming in cold water (black symbols). The bear rewarms during walking (grey symbols) to return to normal Tb. Activity level is indicated by the accelerometer score.

Redrawn from Whiteman et al. (2015).

Pinnipeds, sirenians, and cetaceans spend considerable or all of their time in water. For harbour seals (Phoca vitulina), immersion in cold water increases the lower critical temperature from $<−10$ to $+10°C$ (Irving & Hart 1957), but core Tb of seals declines very little during diving at a range of water temperatures (e.g. Gallivan & Ronald 1979; Ponganis et al. 1993), and cetaceans obviously indefinitely maintain core Tb in an aquatic environment. This maintenance of Tb occurs despite restrictions on thermoregulatory heat production that occur due to metabolic adaption to oxygen limitations associated with diving (Boyd 2000). Seals are unusual amongst mammals in using a combination of both fur and blubber for insulation, presumably reflecting their intermediate evolutionary history of 29–23 MYBP in a marine environment, while more recently marine mammals (polar bears and sea otters, < 1.6 MYBP) rely on fur, and exclusively and long-term marine sirenians and cetaceans (50 MYBP) rely on blubber (Liwanag et al. 2012).

Blubber consists of a ‘continuous, subcutaneous layer of adipose tissue, reinforced by collagen and elastic fibres’ (Liwanag et al. 2012). It is a more effective insulator for deep and/or long-term submersion than fur, when the hydrostatic pressure associated with depth will force air out of the pelt. For example, the ringed seal (Phoca hispida) has a very thin pelt, which becomes completely saturated when immersed—but a thick layer of poorly vascularized blubber. Although a poor insulator in air, blubber is much more effective than wet fur in water; the insulation of a seal carcass may be only 5% less in ice water than in 0°C air, but a relatively greater thickness of blubber is required compared with fur (Scholander et al. 1950a; see Figure 3.11). The insulation of blubber can, however, be more readily altered than that of fur, by changing the degree of blood perfusion, and it has the added functions of providing an energy store, aiding buoyancy control, and contributing to attaining a streamlined body form. It has independently evolved in cetaceans, sirenians, and pinnipeds (Liwanag et al. 2012). Amongst pinnipeds, there are allometric, phylogenetic, and ecological patterns in the morphology, biochemistry, and insulative properties of blubber. Fur seals rely more on fur for insulation than sea lions, phocids, and walrus, and this is reflected in differences in fur density and blubber thickness and biochemistry. The various groups of pinnipeds also differ in the composition of the inner and outer layers of blubber: fur seals use their blubber more for storing energy, while phocids and sea lions use the inner layer of blubber for energy storage and the outer layer for thermoregulation, as indicated by fatty acid composition (Liwanag et al. 2012).

Overall, regional heterothermy across the body tissues, facilitated by blubber and vasomotor control, is probably the most important mechanism for maintaining Tb (p.299) for highly aquatic mammals (Boyd 2000); it can achieve levels of insulation similar to that of the pelt of large terrestrial mammals (Irving & Hart 1957). Indeed, for harbour and harp (Phoca groenlandica) seals in cold water, low skin temperatures and steep temperature gradients established over about 6 cm in the body tissues negate any thermoregulatory increase in metabolic rate. Heat loss across the skin of free-ranging Antarctic fur seals in water of $1.4−4°C$ ranged from 30 W m−1 during diving to more than 600 W m−1 while swimming at the surface, the mean being more than twice that recorded for dolphins in warmer tropical water. The skin-to-water thermal gradient of seals was highly variable during their time at sea, ranging from almost 0 to $>20°C$, and was labile over short time periods of < 1 h. This indicates that seals and presumably other largely aquatic mammals use skin temperature as an important thermoregulatory mechanism in aquatic environments (Boyd 2000).

Regional heterothermy that results in cooling of the extremities is common in medium and large cold-climate mammals (and birds) in winter. This reduces heat loss from appendages that have a high surface area to volume ratio. It is generally achieved by countercurrent heat exchange between warm arterial blood and cool venous blood, with anatomical specializations, such as closely opposed vessesls or rete, enhancing the effect. For example, Irving and Hart (1957) observed regional heterothermy in the flippers of harbour seals in water at 6°C. For seals with a core body temperature of 37°C, the tissue temperature amongst the metacarpals was 22°C at a distance of 2.5 cm from the body and $9−15°C$ at 5 cm; the subcutaneous temperature of the flipper web was $6−7.8°C$ (Irving & Hart 1957). It is essential, however, that this regional heterothermy is controlled, and the temperature of the extremities is regulated above the freezing point of tissues, even when the surface that the animal is contacting may be $>50°C$ colder. Tropical and non-acclimated temperate mammals generally use vasoconstriction to limit heat loss from the extremities, and undergo pulsatile increases in blood flow to these regions to delay freezing of the tissues.

Figure 4.4 Energetic cost (cross-hatched bars) of various activities for sea otters (Enhydra lutris), time spent engaged in each activity (white bars), and overall metabolic cost of each activity (black bars).

Data from Yeates et al. (2007).

Cold-acclimatized temperate mammals, such as wolves (Canis lupus) and Arctic foxes, use continuous proportional temperature control via regulated vasodilation that allows for warm blood to maintain the appropriate foot temperature (Henshaw et al. 1972). Anatomical specialization of vascular system of the foot allows warm blood to flow directly to the footpad surface, and differential flow of this blood regulates footpad temperature to prevent freezing. For example, adult wolves and Arctic foxes regulate footpad temperature at 3.9°C (variation $≤0.7°C$) at temperatures of $−38°C$ for periods of 7.5 h (Henshaw et al. 1972). This may also allow for enhanced heat loss from highly insulated animals during exercise. Indeed, infrared thermography of Arctic, red (Vulpes vulpes), and kit (Vulpes macrotis) foxes indicated that the lower legs and paws function as effective thermoregulatory surfaces (Klir & Heath 1992).

(p.300) Moving through snow dramatically increases the cost of transport (COT) for mammals. The deeper and denser the snow, and the faster the animal moves, the greater is the increase in COT. Deep snow may necessitate a change of gait to a bound that requires considerable energy for the complete vertical displacement of the body, and dense snow increases foot drag and prevents wading, requiring the animal to lift its legs free of the snow with each stride (Parker et al. 1984; Marchand 2014). Therefore, considerable energetic costs are associated with foraging for winter-active mammals. For example, energetic costs of simply standing compared to lying down for most ungulates increases metabolic rate by about 21–37%. COT is roughly linearly related to velocity, and increases for uphill and decreases for downhill locomotion. Relative increases of COT increase exponentially with snow depth for mule deer (Odocoileus hemionus) and elk (Cervus elaphus) walking in snow. In powdery snow with a density of 0.2 g cm−3, the relative increase in COT is approximately 40% at a sinking depth of 30% of brisket height, increasing to approximately 375% at a sinking depth of 90% of brisket height. These costs are further increased for snow of higher density—approximately 600% increase in COT at a sinking depth of 80% for wet snow with a density of 0.4 g cm−3 (Parker et al. 1984). An animal’s foot loading, leg length, and movement velocity also determine how much it will sink in snow and therefore impact on the energetics of locomotion. For example, powdery snow results in greater increases in COT for mule deer compared with elk, as deer have shorter legs, so sinking to the ground means they sink relatively deeper. However, in dense snow, elk have a greater COT increase compared with deer, as deer have proportionally longer legs and lower foot loading, reducing relative sinking depth (Parker et al. 1984). Mammals can also behaviourally mitigate some of the energetic costs of locomotion in snow by using preformed trails, selecting areas of packed snow, or modifying their gait. For example, coyotes in south-eastern Quebec select for shallower and harder snow to reduce their sinking depth, and during conditions of heavy snow preferentially travel on trails of artificially packed snow. These behavioural responses were calculated to reduce heart rate by approximately 5% and therefore presumably to result in energy savings (Crête & Larivière 2003).

Telfer and Kelsall (1984) developed morphological, behavioural, and combined snow-coping indices for a range of large North American temperature and sub-Arctic mammals (Table 4.1). Caribou and moose have the highest morphological snow-coping indices, followed by the predatory wolves, wolverines, and coyotes. Bison (Bison bison) and pronghorn antelope (Antilocapra americana) had the least favourable indices. Amongst ungulates, behavioural indices were similarly ranked (with the exception of white-tailed deer, Odocoileus virginianus, which appears to behaviourally compensate for limited morphological adaption to snow), and therefore so were overall snow-coping indices, with caribou the best snow-adapted species examined. These indices correlate with the severity of snow cover within the species’ distributions (Telfer & Kelsall 1984).

Table 4.1 Morphological ((chest height + foot loading)/200), behavioural (score of 0–5 for each of six behavioural traits/30), and overall mean snow-coping indices for large North American temperate and sub-arctic mammals. Data from Telfer and Kelsall (1984).

Category

Species

Morphological Index

Behavioural Index

Mean Overall Snow-coping Index

Ungulates

Caribou (Rangifer tarandus)

154

26

0.82

Moose (Alces alces)

140

19

0.67

White-tailed Deer (Odocoileus virginianus)

112

21

0.63

118

18

0.60

Dall Sheep (Ovis dalli)

121

17

0.59

Bighorn Sheep (Ovis canadensis)

114

16

0.55

Bison (Bison bison)

95

15

0.49

Pronghorn Antelope (Antilocapra americana)

81

13

0.42

Carnivores

Coyote (Canis latrans)

133

Wolf (Canis lupus)

135

Wolverine (Gulo gulo luscus)

135

(p.301) Snow (and ice) provide problems in addition to locomotion for mammals in extremely cold environments. Freezing of freshwater sources means that some mammals must ingest frozen water in the form of snow or ice to drink. Ingested frozen water must first be melted, which requires 334 J g−1 (latent heat of fusion) and then warmed to Tb, another 4.19 J g−1 °C−1 (specific heat capacity of water; Withers 1992). For small mammals active in subnivean environments, energetic costs associated with consuming frozen water range from 2% (red-backed voles; Clethrionomys rutilus) to 12.9% (meadow voles; Microtus pennsylvanicus) of their daily energy expenditure during winter (Whitney 1977; Holleman et al. 1982; Berteaux 2000). These energetic costs can be met for winter-active mammals by consuming additional food. However, the necessity to arouse and drink cold or frozen water for the hibernating mountain pygmy possum (Burramys parvus) places limitations on hibernation duration and therefore may influence overwinter survival. Ingesting 5% of body mass of cold (2°C) water and warming it to 35°C requires as much energy as 13 h of torpor, and ingesting frozen water requires the energetic equivalent of 45 h (Cooper & Withers 2014b). Over the entire hibernation season, drinking cold water may reduce the potential hibernation period (p.302) by 11.5 days and eating snow by 30 days, and costs are even higher for juvenile pygmy possums. Consuming cold or frozen food and warming it to Tb also has an energetic cost, which is dependent on food water content, energy content, and digestibility, as well as the Ta. However, these costs are generally low compared to drinking cold or frozen water.

Predators rarely experience cold-associated energetic costs of feeding, as freshly caught prey have a Tb similar to their own, and insectivores typically consume high-energy, readily digested food. Herbivores, however, may feed during the winter months on poor-quality, sometimes frozen plant matter that must be melted and warmed (Chappell 1980). These herbivores may preferentially select for plant tissues that avoid freezing at low Ta by accumulation of solutes and therefore avoid the high costs of melting frozen water. For dry plant matter, the energetic costs of warming cold food is relatively low, as cellulose has a specific heat capacity only about one-third that of water (1.3 J g−1 °C−1) and there is no latent heat of fusion (Berteaux 2000).

Most small, non-hibernating mammals decrease body mass during winter, and some actually reduce body size (a phenomenon known as the Dehnel effect; see 3.2.7), which functions to reduce overall energetic requirements by reducing absolute metabolic rates and therefore food requirements (Heldmaier 1989; Lovegrove 2005). Small species appear to have the most extreme seasonal mass changes; the smallest species may decrease mass by as much as 50% (Lovegrove 2005). Larger species generally increase winter body mass as a consequence of fat storage that may be important for withstanding winter food shortages (Heldmaier 1989), although Lovegrove (2005) found no effects on body size or latitude associated with the magnitude of seasonal body mass change for large mammals in general. There are, however, exceptions: body mass of reindeer decreased 8.6 and 3.8% in winter compared to summer (Nilssen et al. 1984). Mass changes may relate to the severity of the local winter conditions. For example, winter body mass reductions were most extreme for the common shrew (Sorex araneus) in Finland, were intermediate in Poland, and were smallest in the Rhine valley. Similar patterns are evident for voles (Heldmaier 1989). Lovegrove (2005) found a significant correlation of body mass changes with latitude for mammals after accounting for body size and phylogenetic history.

Many less-well insulated mammals, particularly smaller species, do have to increase their metabolic heat production during winter months. For placental mammals, an increase in brown fat enhances their capacity for non-shivering thermogenesis, and higher myoglobin concentration in skeletal muscle also increases the capacity for oxygen transfer and storage, and thus shivering thermogenesis (Heldmaier 1989; Marchand 2014). Despite these increases in thermogenic capacity, significant decreases in body mass generally result in reduced absolute BMR for small mammals during winter (Lovegrove 2005). Other species decrease their metabolic rate, presumably to reduce food requirements when feed may be limited (p.303) during winter. For example, the resting metabolic rate of Svalbard and Norwegian reindeer decreased from 2.15 to 1.55 and 2.95 to 2.05 W kg−1, summer to winter, respectively. This was accompanied by a decrease in food intake of 57 and 55%, respectively (Nilssen et al. 1984).

## 4.1.2 Avoiders

Some mammals withstand cold winter conditions and associated food limitations by avoiding them. Some migrate to more equable climates (see 4.6.3), but others use long-term hibernation and remain in a secure hibernaculum and reduce Tb and metabolic rate—often to as low as 0°C and 1% of BMR—and await milder spring or summer conditions, when they typically arouse to reproduce. Hibernacula often have more favourable microclimates than the extreme external climatic conditions. For example, the winter subnivian hibernaculum of the mountain pygmy possum remains at a constant temperature of $1.5−2.5°C$, despite ambient air temperature fluctuating from −8 to 20°C. This constancy occurs due to the insulating effects of overlying rocks, soil, and in particular snow, and is essential for successful overwintering of the possums (Körtner & Geiser 1998). Early snowmelt eliminates the insulative effects of snow and reduces hibernacula temperatures. This results in possums expending more energy during torpor, undergoing more frequent and longer periodic arousals, and undertaking their final arousal before the availability of their spring food, all of which can result in substantial population declines (Smith & Broome 1992; Broome 2001).

Some monotremes, marsupials, and placentals hibernate seasonally for extended periods (although all long-term, cold-climate hibernators undergo periodic arousals during this hibernation period; see 3.2.4.2). Short-beaked echidnas (Tachyglossus aculeatus) in cold regions such as the Australian Alps and Tasmania hibernate for up to seven months (Grigg et al. 1992; Nicol & Andersen 2002). Generally, they enter hibernation 68 days after the summer solstice; males arouse about 5 days before the winter solstice; reproductive females arouse about 30 days later; and non-reproductive females may continue to hibernate for another 2 months (Nicol & Andersen 2002). During hibernation, Tb declines considerably (typically < 10°C), but there are periodic arousals throughout the hibernation period (see 3.2.4.2).

Amongst marsupials, the Australian pygmy possums (Burramyidae) and feathertail glider (Acrobates pygmaeus; Acrobatidae), as well as the South American monito del monte (Dromiciops gliroides; Microbiotheria) undergo long-term seasonal hibernation to avoid cold winter conditions (Bozinovic et al. 2004; Riek & Geiser 2014). The marsupial mountain pygmy possum is a typical hibernator, hibernating for up to 7 months, beginning 130 days after the summer solstice and arousing 100 days after the winter solstice (Nicol & Anderson 2002). The longest recorded hibernation period for any mammal is for the eastern pygmy possum (Cercartetus (p.304) nanus) in the laboratory. Pygmy possums were able to extend their hibernation period to last on average 310 days without food, with one individual lasting 367 days, the only report to date of a mammal extending hibernation for more than 1 year (Geiser 2007).

Placental mammals of the orders Insectivora, Chiroptera, and Rodentia (particularly the families Sciuridae, Cricetidae, and Gliridae) also use seasonal hibernation in cold environments (Ruf & Geiser 2015). Indeed, hibernation was first described by Aristotle (384–322 BC) for the edible dormouse (Glis glis; Cooper & Withers 2010). Placental seasonal hibernators typically hibernate for some 7 or so months; sciurids and marmots enter hibernation between 62–72 days after the summer solstice and finally arouse 112–130 days after the winter solstice (Kenagy et al. 1990; Armitage 1998). The longest recorded hibernation season observed for a free-living mammal was for dormice in Austria, in a year of European beech (Fagus sylvatica) mast failure. Individuals with a large body mass entered hibernation early, in late July, forgoing reproduction to extend the hibernation season up to 11.4 months (Hoelzl et al. 2015).

Figure 4.5 Body temperature during the hibernation period of a male Arctic ground squirrel (Urocitellus parryii). Insets show absence of periodicity of body temperature cycles in darkness at the end of hibernation but 24-h entrained rhythms after the squirrel leaves the hibernaculum and is exposed to daylight.

Redrawn from Williams et al. (2012). Reproduced with permission of Biology Letters, The Royal Society.

Long-term cold-climate hibernators show predictable circannual hibernation patterns (and other seasonal responses, such as activity, body mass, and reproduction) that are intrinsically entrained, and do not rely on external environmental cues. For example, it has been well established that ground squirrels maintain an annual cycle of activity and hibernation even when maintained at constant Ta and photoperiod (Pengelley & Fisher 1963). This level of circannual entrainment reflects their seasonally predictable environments and allows them to still (p.305) undertake required activities, even when environmental cues are unavailable (e.g. within sealed, underground hibernacula; Körtner & Geiser 2000).

Unlike circadian clocks, the location and underlying mechanisms that determine the function of the circannual clock are still unknown, which hinders understanding of its role in hibernation (Williams et al. 2014a). Despite evidence for intrinsic entrainment of circannual hibernation patterns, changes in the environment require some degree of entrainment to external cues to ensure the circannual cycle remains synchronized to environmental conditions. For example, ground squirrels and marmots hibernate irrespective of changing photoperiod or pinealectomy prior to hibernation, but long-term changes to photoperiod, such as a change in hemisphere or pinealectomy well before hibernation, do interfere with the timing of hibernation, suggesting there is a short period in summer during which the circannual clock of these hibernators is sensitive to photoperiod (Körtner & Geiser 2000). Observations that different populations of ground squirrels have varying hibernation characteristics related to environmental variables, such as altitude, aspect, and snow cover, are further evidence of some plasticity or adaptability in the timing of entrained hibernation (Williams et al. 2014a). Ta appears to have some impact on the circannual cycle of hibernators, with excessively cold spring and autumn temperatures influencing the length of the hibernation period. This presumably has important ecological consequences, allowing hibernating mammals to time their emergence to best exploit or avoid variations in food abundance (Körtner & Geiser 2000).

Hibernators generally rely on fat stores to sustain their reduced metabolic rate during hibernation (although some such as chipmunks, Tamias, feed on cached food during interbout arousals), and many will not hibernate or reduce their hibernation period if they have not attained a suitable body mass (Geiser 2013). (p.306) For example, only dormice with high body fat extended their hibernation period by entering hibernation early in years of beechmast failure (Hoelzl et al. 2015). A reduction in respiratory quotient (RQ) of hibernators from 1 (carbohydrate metabolism) to 0.7 (fat metabolism) is consistent with hibernators metabolizing fat rather than carbohydrate during hibernation, and pre-hibernation fattening is accompanied by changes in circulating leptin concentrations and leptin sensitivity (Kronfeld-Schor et al. 2000). Age, sex, and reproductive status also impact hibernation; for example, adult male echidnas arise from hibernation much earlier than females, and seek out and mate with females still torpid. Non-reproductive females may hibernate for 2–3 months longer (Nicol & Anderson 2002; Morrow & Nicol 2009). Male ground squirrels also arouse from hibernation and exit their hibernacula earlier than females (Williams et al. 2012), presumably in response to hypothalamus-pituitary-gonadal axis endocrine changes that precede the reproductive season. Social hibernators such as marmots are much more synchronized (Williams et al. 2014a).

Extreme seasonal hibernation requires mammals to undergo a suite of physiological, biochemical, and morphological changes that would prove fatal for other mammals, but the cellular and molecular basis for these changes are not well understood. For example, there is still no clear understanding of the molecular basis for initiating and terminating hibernation, despite extensive research (Frerichs et al. 1998; Carey et al. 2003), although there is some evidence of a role for thermoregulatory neurotransmitters such as serotonin, histamine, and opioids on central nervous system control of hibernation (Sallman et al. 1999).

Examples of seasonal hibernators are found in all three mammalian linages, and heterothermia is believed to be a plesiomorphic trait amongst mammals, so the ability to hibernate is most likely due to patterns in gene expression, rather than to the presence or absence of particular genes (Srere et al. 1992; Carey et al. 2003; Xu et al. 2013). For example, mRNA for genes encoding for thyroxine-binding globulin, apolipoprotein A1, cathepsin H, CIRBP, and α‎2-macroglobin are upregulated in ground squirrels (Spermophilus) during the hibernation season, presumably to enhance function during hibernation; α‎2-macroglobin, for instance, plays a role in the increased blood clotting times of hibernators (Srere et al. 1992).

The molecular response during hibernation is complex, involving a suite of modifications in gene expression occurring in a variety of tissues and organs, and is probably regulated by metabolic status (van Breukelen & Martin 2015). Some recognized molecular changes during hibernation include differential gene expression that regulates switching from carbohydrate to fatty acid metabolism, with downregulation of mRNA and associated proteins of glyceraldehyde-3-phosphate dehydrogenase and acetyly CoA carboxylase, and upregulation of pyruvate dehydrogenase kinase isoenzyme 4, pancreatic triacylglycerol lipase, hormone-sensitive lipase, and transcription factor PPAR.

(p.307) Changes in heat production capacity associated with the necessity for periodic arousals are reflected in upregulation of mRMA for uncoupling proteins 2 and 3, and heart- and adipose-type fatty acid binding proteins in tissues including white and brown adipose tissue, skeletal muscle, heart, kidneys, and liver, and in increased myoglobin proteins in skeletal muscle (Carey et al. 2003; Xu et al. 2013). Downregulated mRNAs include those associated with activity, such as prostaglandin D2 in the brain and glycerabdehyde-3 phosphate dehydrogenase in muscle. Other changes in gene expression that occur during hibernation include upregulation of nRMA and proteins associated with protecting the body of hibernators from negative impacts of long-term cold exposure, such as increased risk of ischaemia leading to redox shifts and oxidative stress; mitochondrial proteins such as cytochrome-c oxidase subunit1, ATP synthase 6/8, transcription factor nuclear factor-kB, and glucose-regulated protein 75 (Carey et al. 2000; Hittel & Storey 2002). Despite these reported changes in gene expression, mRNA and proteins associated with the majority of genes remain at normothermic levels during the hibernation season, presumably to facilitate function during periodic arousals (Carey et al. 2003).

As for many physiological processes, numerous cellular and molecular processes are slowed, or cease altogether, during long-term seasonal hibernation (van Breukelen & Martin 2015). RNA transcription is an energetically expensive process; it is reduced during hibernation via reductions or cessation of initiation and elongation processes, consistent with a Q10 effect of reduced Tb (van Breukelen & Martin 2002). Translation of mRNA into protein is also affected by hibernation. Protein synthesis is reduced in hibernating compared to normothermic animals, and there are significant losses of polyribosomes in numerous organs of hibernators. Unlike transcription, there is evidence of active suppression of this even more energetically costly process beyond temperature effects (Carey et al. 2003). Brain extracts from hibernating individuals have significantly less translational activity than those from normothermic individuals, but placing polyribosomes from hibernating animals in extracts from normothermic individuals overcomes this suppression (Frerichs et al. 1998). DNA synthesis and cell division are also impacted during hibernation, with DNA synthesis occurring at about 4% of the rate of normothermic animals and mitosis ceasing altogether; it remains unclear if there is regulated suppression of these processes, or if these responses are entirely a consequence of low temperature. They do, however, return to or even exceed pre-hibernation rates during interbout arousals (Carey et al. 2003). Changes in the ultrastructure of cells and mitochondrial respiration also reflect cellular and molecular alternations during torpor.

Seasonal hibernators tend to increase the proportion of monounsaturated and polyunsaturated fatty acids (MUFAs and PUFAs, respectively) in their tissues, which have relatively low melting points compared to saturated fatty acids (SFAs) and may play an important role in maintaining fluidity of body lipids. Lipids can (p.308) be accessed for metabolism only if they are in a liquid state. Increases in the proportion of PUFAs in cell membrane phospholipids are also observed in deep hibernators, presumably to aid in maintenance of normal membrane function at low Tb by maintaining the usual liquid-crystalline state (Geiser 1993; Florant 1998; Munro & Thomas 2004). Indeed, golden-mantled ground squirrels on a high PUFA diet in the laboratory had a higher hibernation frequency, higher survival, and attained lower Tb than those on diets with lower PUFA concentrations (Frank 1992). Mammals are unable to synthesize PUFAs and therefore must obtain them from their diet; these dietary PUFAs then influence the fatty acid composition of the tissues (Geiser 1990).

Meta-analysis of the impacts of experimental and natural diets on hibernation reveals that low PUFA diets, with about 11% PUFA concentrations, do limit hibernation, with mammals maintaining higher minimum Tb and torpor metabolic rates and having shorter hibernation durations. Providing higher PUFA diets enables mammals to reduce minimum Tb and torpor metabolic rate by incorporating more PUFA into their lipids. However these changes are small (about $1−1.2°C$ for a change in PUFA concentration from 11.2 to 54.4%) and occur at moderate as well as low torpor Tbs, suggesting that factors other than lipid state impact the interaction between PUFA availability and hibernation characteristics, possibly by influencing the Tb setpoint or temperature perception (Geiser 1993; Munro & Thomas 2004).

Some mammals show little or no selection for dietary PUFAs, and for some species, diets high in UFAs have no significant impact on hibernation characteristics (e.g. monito del monte; Contreras et al. 2014). Species such as echidnas and other insectivores, which have low concentrations of PUFAs in their natural diet, may substitute MUFAs such as oleic acid in their adipose tissue and membranes, retain and selectively ingest the limited available dietary PUFAs, and employ cholesterol to aid in maintaining lipid fluidity (Schalk & Brigham 1995; Falkenstein et al. 2001). Despite their apparent benefit to lipid and membrane function at low Tb, PUFAs are vulnerable to auto-oxidation. During hibernation, mammals are already exposed to enhanced oxidative stress, and therefore the ideal PUFA concentration for a hibernator may involve a trade-off between minimizing potential auto-oxidative effects and maintaining lipid and membrane fluidity. Indeed, evidence suggests that mammals with access to high PUFA diets optimize rather than maximize their PUFA intake; if PUFA levels are too high, hibernation may be limited to restrict negative oxidate effects (Munro & Thomas 2004). For example, the PUFA content of pre-hibernation diets of free-living Arctic ground squirrels vary more than threefold, although no individuals appear to select low PUFA diets. Those ground squirrels that consume an intermediate level of PUFAs (33–74 mg g−1) have higher overwinter persistence, longer torpor bouts, fewer arousals, and more torpid days during the hibernation season than those with high PUFA diets (> 74 mg g−1), demonstrating the impact of PUFAs on hibernation for wild herbivores (Frank et al. 2008).

(p.309) The changes in Tb accompanying torpor and hibernation have an impact on acid–base status, because of the effect of temperature on dissociation of water (see 2.5.6), the increased CO2 solubility at low Ta, and physiological control of ventilation relative to metabolic rate (MR). For example, daily torpor of the little pocket mouse (Perognathus longimembris) is associated with a decrease in MR to 0.05 ml O2 g−1 h−1 at $Ta=10°C$ (from 7.04 ml O2 g−1 h−1 for normothermia), a decrease in minute ventilation (VI) to 6 ml air g−1 h−1 (from 329), and an increase in VI/MR to 120 ml air ml $O2−1$ (from 47). This change in VI/MR is reflected in a decreased blood pCO2 during hibernation to 1.9 kPa (from 4.8) and increased pH of 7.51 (from 7.28); this is a ∆pH/∆T change of about $−0.0085U°C–1$, which is intermediate between that expected from the ionization of water (α‎-stat hypothesis, $−0.017$) and constancy of pH (pH-stat hypothesis, 0). The blood $[HCO3−]$ did not change from 17.3 to 18.8 mmol l−1. For hibernating hamsters (Cricetus cricetus), pCO2 also declined (6.0 to 4.4 kPa), but $[HCO3−]$ increased (28.2 to 53.8 mmol l−1), and pH increased (7.40 to 7.57), with $ΔpH/ΔT=−0.006;$ for hibernating marmots (Marmota marmota), pCO2 declined (5.5 to 4.9 kPa), $[HCO3−]$ increased (29 to 52.8 mmol l−1), and pH increased (7.45 to 7.57), with $ΔpH/ΔT=−0.004$ (Malan et al. 1973). Kreienbühl et al. (1976) reported similar changes for hibernating dormice (Glis glis) in pCO2 (5.1 to 3.7 kPa) and pH (7.24 to 7.44), with $ΔpH/ΔT=−0.006.$

These results for hibernators are similar to those for the daily torpidator, although blood $[HCO3−]$ increased during hibernation, indicating a slightly different acid–base balance, perhaps reflecting the longer temporal scale for hibernators. The reduction of MR during torpor and hibernation is caused by inhibition of cold-induced thermogenesis, the Q10 effect on MR due to the decreased Tb, and possibly a further metabolic depression caused by CO2 retention and tissue acidosis (Malan 1982, 1988, 2014; see Figure 3.16). Hyperventilation during arousal from hibernation provides further evidence that acidosis might have a metabolic depression effect that needs to be removed for arousal. Geiser (2004) calculated that hibernating mammals have a considerably higher Q10 for MR (3–6) than daily heterotherms (1.5–3.5) (cf. typical Q10 values of about 2.5 for mammals; Guppy & Withers 1999).

The Tb of at least one mammalian hibernator, the Arctic ground squirrel, drops below 0°C during hibernation, and therefore there is a risk of body tissues freezing (Barnes 1989). They have a Tb as low as $−2.9°C$ while in their hibernacula surrounded by soil temperatures of $−6°C$, but they do not freeze. Plasma solute concentrations of ground squirrels can account for depressing of the freezing point to $−0.6°C$, but below this, the ground squirrels supercool. Supercooling occurs when an absence of a nucleating agent allows for cooling below the freezing point without freezing. There is as yet no evidence of antifreeze proteins in ground squirrel blood. Supercooling to Tbs as low as $−5°C$, with subsequent rewarming and survival, has been achieved for small mammals in the laboratory, but only for (p.310) periods of 50 to 70 min (Kenyon 1961). Nevertheless, Arctic ground squirrels are able to remain in a supercooled state for periods of approximately 3 weeks, between interbout arousals.

Bears (Ursus) are an interesting example of mammals that avoid extreme cold winter conditions by becoming inactive, as their physiological characteristics during this seasonal period of dormancy differ considerably from those of other hibernating mammals. This has led to a long-running debate concerning the nature of their winter quiescence: do they hibernate like other smaller mammals, or do they just enter a state of extreme inactivity and prolonged fasting (Hellgren 1995)? Bears fatten considerably in the months preceding the winter hibernation season, and then den for up to 7 months during which they do not eat, urinate, or defaecate. They do, however, maintain some limited activity during this time—standing, drinking, and arranging their nesting material every 1–2 days, for an average of 24 min per day (Tøien et al. 2011; Robbins et al. 2012; but see Folk et al. 1976), and can be easily roused to an active state (Nelson et al. 1983).

Unlike other hibernators that drop their Tb setpoint during hibernation to close to Ta, often approaching (or even dropping below) 0°C, bears maintain a relatively high Tb, only a few degrees below their normothermic Tb of $37−38°C$. For example, black bears (Ursus americanus) overwintering in dens with a Ta of 0 to $−20°C$ maintained mean Tbs of $31.7−34°C$ during mid-hibernation, and as high as $36−37°C$ towards the end of the denning period (Tøien et al. 2011; Figure 4.6). This high Tb is likely a consequence of a large body mass, which limits cooling rates and makes the energetic costs and rate of arousal from very low Tb prohibitive. Bears do not undergo the periodic arousals to normothermia characteristic of all other cold-climate seasonal hibernators, presumably because their relatively high Tb does not result in the same degree of perturbation of homeostasis as the very low Tb of other hibernators, as observed for the tropical hibernating tenrec (Tenrec ecaudatus; Lovegrove et al. 2015). However Tøien et al. (2011) did observe (p.311) a 1.6–7.3 day cycle of Tb, with an amplitude of $2−6°C$ that was accompanied by shivering and increased MR, that may reflect periodic arousals, although the circadian cycle appears to have been abolished, as for smaller hibernators.

Figure 4.6 Body and ambient temperatures during overwinter denning by black bears (Ursus americanus).

Redrawn from Tøien et al. (2011).

Metabolic rate of denning bears over winter is reduced to 30–75% of their BMR (Hellgren 1995). For polar bears, the metabolic rate is 1.7–2.2 times higher for fasting but active bears than it is for denning individuals (Robins et al. 2012), which together with observations for black bears of MR reduced well below that expected from the Q10 effect of reduced Tb, and maintenance of low BMR even after spring exit from the den at normothermic Tb, suggests that bears may indeed use metabolic depression to enter a form of hibernation (Tøien et al. 2011). Denning winter bears also have reduced heart rates, from a normal resting rate of 40–60 beats min−1 to as low as 8–27 beats min−1 (Folk et al. 1970). Denning bears generally lose 15–37% of their body mass during the hibernation period. Their overwinter MR is supported by metabolism of adipose stores rather than lean body mass, evidenced by observations of RQs of < 0.73. For example, 74–99% of the body energy content lost by polar bears was from adipose tissue. However, the quantity of adipose stores at the initiation of fasting appears to determine the ability of polar bears to conserve lean body mass (Atkinson et al. 1996).

Amongst hibernating black and grizzly bears, there is no evidence of development of ketosis, as would be expected for a mammal undergoing severe fasting. Fasting but active polar bears in summer have similar blood chemistry to denning black bears in winter (Nelson et al. 1983). Bears suppress urination during the denning period, but a post-hibernating grizzly bear produced only 181 ml of urine after 4.5 months, containing 0.98 g of urea and 72 mg of ammonia, compared to a control sample collected in the bladder over 24 hours, which had a volume of 2,080 ml and contained 55.05 g of urea and 1,785 mg of ammonia (Folk et al. 1976). Remarkably for a mammal, denning winter bears produced urine daily, but the water and solutes were reabsorbed through the bladder wall (Nelson et al. 1975). There were no or very small net changes in blood amino acids, protein, urea, uric acid, or ammonia due to effective protein anabolism rather than due to starvation or cessation of urea production. Regulation of urea balance appears essential for maintenance of hibernation; experimental disruption results in disruption of hibernation (Nelson et al. 1975). Protein metabolism of denning bears increases three to five times, allowing for amino acids to be incorporated into proteins rather than forming urea, and possibly contributing to thermogenesis and the relatively high Tb of overwintering bears (Hellgren 1995).

After spring emergence, bears generally continue to fast for several weeks and their MR remains depressed below BMR, even if food is available (Nelson et al. 1983; Tøien et al. 2011), despite a return to normothermic Tb. Bears will not undertake hibernation during the summer months when they are usually active, even when subjected to simulated winter environmental conditions and fasted. Fasting bears during this time use muscle as an energy source and become (p.312) dehydrated and uremic without access to water (Nelson et al. 1975). In autumn, the energy intake of bears increases more than threefold, and free-living individuals may feed as much as 20 h day−1 to build up the adipose supplies necessary to sustain them for the next hibernation period.

Clearly, bears share some aspects of winter hibernation with smaller mammals that undergo classical seasonal hibernation: seasonally restricted dormancy with pre-dormancy fattening, abolishment of circadian patterns, reduced MR and heart rate, and long-term survival without food. However, their winter inactivity is characterized by some profound differences, including a smaller reduction in Tb; maintenance of limited activity and responsiveness; no periodic interbout arousals to normothermia; and unique protein and urea management, with an absence of urination for the hibernation period.

# 4.2 Hot Environments

Deserts are traditionally considered to be hot and dry environments. The record air temperature, measured by a standard meteorological weather station (Stevenson screen) is 58.0°C, measured in El Aziza (Libya), with Death Valley (USA) a close second at 56.7°C (Mildrexler et al. 2006). However, the hottest land surface temperature (remotely measured radiometric temperature) is considerably higher because of direct solar radiation and the relatively low conductivity of air; the hottest records are 70.7°C in the Lut Desert (Iran) and 69.3°C in Queensland (Australia). Deserts can not only be hot; they can also be extremely dry. Arica (Atacama Desert, Chile) is reputably the driest place on Earth, with an average annual precipitation of 0.8 mm, and no recorded rainfall for 14 consecutive years (Krause & Flood 1997). Arid zones, which we traditionally synonymize with deserts, are generally defined by a combination of temperature and rainfall characteristics. For example, Köppen’s classification scheme defines climate zones in part by the mean annual precipitation (Pann; mm), mean annual Ta measured at 2 m height (Tann; °C−1), and dryness threshold (Pth; mm); Pth is 2 Tann if two-thirds of the annual precipitation occurs in winter, if $23$ of the annual precipitation occurs in summer, and otherwise (Kottek et al. 2006). The arid zone (labelled type B) has $Pann<10Pth$, and is subdivided into steppe (BS) if $Pann>5Pth$ and desert (BW) if $Pann≤Pth$. Both BS and BW are divided into hot $(h,Tann≥18°C)$ and cold $(c,Tann<18°C)$ subregions (e.g. hot desert is BWh and cold desert is BWc). We focus here on the physiological ecology of mammals in hot deserts.

The physiology of mammals provides a number of mechanisms to maintain a stable Tb at high Ta in hot deserts. In moderately warm environments, their suite of physiological and morphological characteristics enables them to dissipate their metabolic heat load, which is substantial given their endothermic bauplan. (p.313) Essentially two mechanisms are available to dissipate their endogenous heat load: increasing thermal conductance to facilitate heat loss by conduction and convection, and evaporative heat loss. In very warm environments, mammals have to not only dissipate their metabolic heat load; they must also dissipate any additional environmental heat load. Heat storage is important, and evaporative heat loss becomes the only mechanism for dissipating a metabolic plus environmental heat load, but at the cost of sustaining a high rate of water loss. These physiological mechanisms, along with how mammals exploit their ecological niche, enables them to remain active and endothermic at a sustainable water cost year-round.

There are some mammals, particularly small species, for which the water cost of maintaining activity in the heat can become temporarily too difficult to sustain if the required rate of heat loss taxes their body water balance (e.g. at very high Ta for periods of low water availability). One solution for non-migratory mammals under such circumstances is to avoid the hot conditions; they become inactive and seek refuge in more equable burrows (for small mammals) or caves, or limit their exposure to high ambient temperatures by being active nocturnally when environmental temperatures are usually lower. Many large mammals are unable to escape hot environmental conditions—they must endure them. This is facilitated by their large size, low thermal conductance, considerable thermal inertia, and high dehydration tolerance (Cain et al. 2006). Some large mammals decrease their daytime activity at the hottest time of year; some compensate with increased nocturnal feeding activity (Maloney et al. 2005; Hetem et al. 2012).

## 4.2.1 Endurers

Large mammals in particular must be able to endure the harsh climate of hot deserts, because they are generally unable to escape; shade under large trees is probably their main refuge, if available. Their maintenance of homeostasis in hot environments depends on being able to achieve heat balance or being able to sustain heat storage during the day. Small mammals (e.g. ground squirrels) are able to endure the harsh climate for short periods, but must periodically seek refuge to avoid overheating (see 4.2.2).

When environmental temperature is lower than Tb, the temperature differential $(Tb−Ta)$ can contribute to heat loss by conduction, convection, and radiation, but is not necessarily sufficient to dissipate all the metabolic heat production, and there is usually a requirement for some enhanced evaporative heat loss (EHL) to dissipate some of the metabolic heat. When environmental temperature exceeds Tb, the temperature differential causes heat gain by conduction, convection, and radiation from the environment to the mammal (see 2.4), so heat storage is important and EHL is the only possible mechanism for heat loss; EHL must be sufficient to dissipate metabolic heat plus any environmental heat gain for Tb to remain in steady state. Consequently, the various avenues for evaporation, hence EHL, becomes paramount for endurers: salivation, sweating, and panting.

(p.314) Many mammals increase saliva production when they are exposed to heat, and spread it onto the body surface, generally by licking their fur or naked patches of skin, to facilitate EHL when the saliva evaporates. Because salivation is a common thermoregulatory response of many mammals once considered to be ‘primitive’, such as opossums (Higginbotham & Koon 1955), it has often been considered to be a ‘primitive’ thermoregulatory response. However, it is clearly a controlled and quite a sophisticated thermoregulatory response of many mammals, particularly rodents, which lack sweat glands. It involves the coordinated physiological and behavioural responses of an increase in the production of a protein-poor saliva, the behavioural response of licking to spread the saliva, and often an increase in blood flow to the skin beneath the licked surface. Rodents lack the ability to pant or sweat, and laboratory rats increase their salivation when exposed to temperatures higher than Tb (Toth 1973). Rats acclimated to high Ta produce more saliva at a lower core temperature, and the ligation of the salivary glands makes the animals vulnerable to rapid hyperthermia (Horowitz et al. 1983). Kangaroos (Macropus) will salivate in the heat (unless they are dehydrated) and lick their forearms to evaporatively cool (Dawson 1973b; Figure 4.7). Their mandibular salivary glands may preferentially contribute to saliva production for thermoregulation compared to the parotid salivary glands (see 4.8.2). Vascular casting of the forearms reveals a dense superficial network of fine blood vessels underlying the forelimb skin (Needham et al. 1974). In a marriage of physiology, morphology, and behaviour, an increase in the core temperature of kangaroos results in a 3–4 times increase in blood flow to the forearms and consequent loss of heat by evaporative cooling of saliva (Needham et al. 1974).

Figure 4.7 Thermal image of an eastern grey kangaroo (Macropus giganteus) in the heat, showing the forelimbs cooled by licking.

Photograph provided by A. J. Munn.

For many large mammals (e.g. humans, horses, some antelope), sweating is the primary means of evaporative heat loss, but small mammals tend to rely more on salivation (see earlier) or panting (see later). Many rodents and lagomorphs lack sweat glands, so sweating is not an option for them. Aquatic mammals such as (p.315) cetaceans and hippopotamus also lack sweat glands, presumably as EHL is redundant in an aquatic environment. The total sweat gland number can range from as few as 20–30 cm−2 for pigs to more than 2,000 cm−2 for zebu cattle. African dik-dik (Rhynchotragus) are relatively small African ungulates (about 2–8 kg) that have a moderate density of sweat glands, about 190 cm−2, and a relatively low sweating rate when heat challenged, about 19 g m−2 h−1; they rely more on panting to dissipate heat, like larger African ungulates (Maloiy 1973). Sweat glands can also have functions other than thermoregulation, including reducing friction (palms and eyelids), excreting wastes, providing antibacterial protection (e.g. the antibacterial peptide dermicidin in human sweat; Schitteck et al. 2001), and for interspecific communication (Jenkinson 1973; see 2.2.4.1).

Sweat is produced by specialized glands in the skin (see 2.2.4.1). In some taxa, the glands occur only in specialized regions (e.g. the footpads of felids and canids), while in other taxa they are distributed all over the body surface. Atrichial (eccrine) sweat glands are not associated with hair follicles; they are found in humans, moles, and those species with plantar glands. Most other mammals have epitrichial (apocrine) sweat glands, associated with hair follicles, and some taxa, such as primates, have both atrichial and epitrichial sweat glands. Sweat glands consist of a fundus and a duct. The fundus varies from a simple sac-like structure to a highly coiled tube, and is formed by two layers of epithelium: an inner secretory layer and an outer myoepithelium. Sweat gland ducts are usually fairly straight, with those of atrichial glands opening directly onto the skin surface, and those of epitrichial glands opening in the pilosebaceous canal, allowing for mixing of sweat and sebum (Jenkinson 1973). Sweat begins as a filtrate of plasma in the sweat gland, the filtrate being modified as it travels down the sweat duct to the skin. Sweating is stimulated by catecholamines in bovids (Robertshaw 1975), while in primates sweating is cholinergically stimulated; cholinergic postganglionic sympathetic nerves lie close to the sweat gland. Marsupials, at least kangaroos, sweat profusely when they exercise but do not sweat during heat exposure (Dawson et al. 1974). Sweating is presumably a more appropriate mechanism to dissipate internal heat, because in a hot environment a consequent reduction in skin temperature would promote heat gain, whereas panting is advantageous to dissipate an external heat load because it dissipates core body heat and does not cool the skin.

Panting can markedly increase respiratory evaporative water loss, and is an important mechanism for heat dissipation in many small and large mammals. It involves a coordinated increase in ventilation combined with an increase in blood flow to the upper respiratory tract; the upper respiratory tract is kept wet, and air flow across its surface is increased by an elevated ventilation rate (Vesterdorf et al. 2011) while the respiratory mucosa is kept cool by limiting its blood flow (Jessen 2001a). The specialized respiratory turbinates that are present in the nose of mammals (and birds) are thought to have evolved for precisely the opposite reason—to minimize evaporative water loss by recovering water vapour from the (p.316) expired air (Hillenius 1992; see 3.6.2). So, when enhanced respiratory evaporative water loss is required, as in panting, the process of recuperative heat exchange is short-circuited by a large increase in arterial blood flow to the mucosa to prevent condensation and cooling of the expired air (Hales 1973). Recuperative heat exchange is sometimes further reduced by expiration via the oral cavity rather than the respiratory chamber (Schmidt-Nielsen et al. 1970b). For example, reindeer (Rangifer tarandus) alternate abruptly between closed- to open-mouth panting, with more open-mouth panting at higher Ta (Aas-Hansen et al. 2000). The heat gained by the air on inspiration is thereby exhaled, and the cooled venous blood draining the mucosa cools the general body.

The increase in ventilation required to achieve respiratory heat loss is not without physical consequence, because ventilation rate is linked to gas exchange and the control of pH, pO2, and pCO2 (see 2.5.6). An increase in alveolar ventilation (VA) results in a washout of CO2 and a decrease in arterial pCO2 (paCO2), as shown by the alveolar gas equation: $paCO2=VCO2K/VA,$ where VCO2 is the CO2 production rate and K is the proportionality constant. During mild hyperthermia, increased air flow in the upper respiratory tract occurs without any major increase in alveolar ventilation; respiratory rate (fR) increases but there is a decrease in tidal volume (VT), so the increased ventilation is of the respiratory dead space rather than alveolar ventilation (VA; Hales 1976). This ventilatory pattern is Type I panting. Under more severe heat stress, mammals switch to Type II panting: VT increases, VA increases, and PaCO2 decreases, leading to acid–base disturbances (Hales 1976). Panting involves extra work by respiratory muscles, so the added heat production associated with panting can also become important; the laboured breathing associated with Type II panting has been estimated to be responsible for 11% of the increased heat production in severe heat (Hales & Findlay 1968).

Overall, salivation, sweating, and panting are the main evaporative responses to heat for different mammals. However, the mechanism for evaporation varies from species to species, and also with acclimation. Important in these considerations is the maximum evaporative heat loss that the mammal can achieve, because a higher heat loss ability will allow thermal equilibrium at larger heat loads. In general, the magnitude of heat loss possible by sweating is greater than by panting, and more energetic and larger species tend to depend on sweating (for example, humans and horses).

Taylor and Lyman (1972) reported that a large running antelope (Thomson’s gazelle Gazella thomsoni) produced large amounts of internal heat when running and Tb increased considerably $(+4.6°C)$, but brain temperature (Tbr) increased more slowly than Tb and could be up to 2.7°C lower. This selective brain cooling was thought to be an important adaptation to heat stress, protecting a thermally vulnerable brain from heat damage. Laboratory studies showed that selective brain cooling was activated above a threshold core Tb via the control of venous blood perfusing the cavernous sinus where there is heat exchange in the carotid rete between (p.317) warm arterial blood going to the brain and cool venous blood returning from the nasal mucosa (Figure 4.8; see 3.4.4). While some early studies reported magnitudes of selective brain cooling (the extent to which Tbr was lower than arterial blood temperature) of nearly 4°C, later studies showed that this large magnitude was probably an artefact of rapidly changing Tb during exercise (Maloney et al. 2009); the maximum magnitude of selective brain cooling in steady state is $1−1.5°C$ (Fuller et al. 2007; Hetem et al. 2012). It is worth noting that an increase in core Tb of 1°C is sufficient to considerably elevate respiratory evaporative loss (e.g. by about six times in the goat; Kuhnen & Jessen 1994).

Figure 4.8 Schematic of heat exchange in the carotid rete of a goat, between warm arterial blood flowing through the cavernous sinus to the brain, and cool blood returning from the nasal cavity to the body.

Modified from Daniel et al. (1953) and Taylor and Lyman (1972).

Selective brain cooling is activated above a threshold core Tb when cool blood is diverted to the cavernous sinus; at low core Tb, venous blood bypasses the sinus and no heat exchange occurs in the rete (Kuhnen & Jessen 1994). However, the first measurements of Tb and Tbr for a free-living mammal, the black wildebeest (Connochaetes gnou), indicated that selective brain cooling was not always activated when Tb increased, and that during exercise, when Tbr was the highest $(41−42°C)$, the wildebeest did not use selective brain cooling at all, casting doubt on its protective role (Jessen et al. 1994). Further, the control of selective brain cooling was much more varied than had been found in laboratory studies, with a large range of core Tb where selective brain cooling may or may not be activated.

An alternative role proposed by Jessen (1998) is that the adaptive significance of selective brain cooling is to modulate the use of water for thermoregulation, since it reduces the temperature of the hypothalamus, where the neural control centres (p.318) for heat defence, including evaporative water loss (EWL), are located. Kuhnen (1997) had shown that manipulation of selective brain cooling in goats altered their EWL during heat exposure. Two studies offer solid support for the role proposed by Jessen (1998). First, if selective brain cooling does modulate water use, then an osmotically stressed mammal could reduce water use by augmenting selective brain cooling. Sheep that are exposed to heat and deprived of drinking water for five days increase selective brain cooling (Fuller et al. 2007). Later, Strauss et al. (2015) showed that the individual sheep that used selective brain cooling more during dehydration and heat exposure used less water than those individuals that used selective brain cooling less, concluding that the activation of selective brain cooling leads to a reduction of water use for evaporative cooling during heat exposure. It has been argued that the evolution of the carotid rete and the ability of artiodactyls to use selective brain cooling and modulate respiratory water loss is what facilitated the increased diversity of artiodactyls since the Eocene in both hot and cold environments, in contrast to the contraction in diversity of perissodactyls, which do not have a carotid rete (Mitchell & Lust 2008).

Another mechanism for mammals to reduce water requirements in the heat is to allow Tb to increase, storing the heat that would normally need to be dissipated by evaporation, and also increasing the temperature differential $(Tb−Ta)$ for heat loss if Ta is slightly lower than Tb, or reducing the gradient for heat gain from a hot environment when $Tb. In a landmark study in the 1950s, Schmidt-Nielsen and colleagues showed that this is exactly what happens for camels when they are water-deprived and exposed to high environmental temperatures (Figure 4.9; Schmidt-Nielsen et al. 1957). The amplitude of the daily rhythm of core Tb increased from the normal 2°C to more than 6°C when camels were deprived of drinking water and exposed to Ta exceeding 40°C. When hydrated, the considerable EHL and minor heat storage balance metabolic heat production (MHP) and (p.319) heat gain from the environment. When dehydrated, the reduced evaporative heat loss and increased heat storage balance MHP and heat gain from the environment; note that the total heat gain is considerably reduced for dehydrated camels. Later, Taylor (1968) reported that the amplitude of the daily rhythm of core Tb for several species of large African ungulates increased when they were exposed to heat, and more so when they were deprived of drinking water.

Figure 4.9 Heat balance of camels (Camelus dromedarius) when hydrated and dehydrated, and daily Tb amplitude.

Modified from Schmidt-Nielsen et al. (1957).

These seminal studies became the basis for the widely held view that the survival of large mammals in hot environments depends on the relaxation of homeothermy and the adoption of a more heterothermic pattern of core Tb. However, studies using bio-logging of mammals free-ranging in their natural habitat consistently find that the amplitude of the daily rhythm of core Tb is independent of environmental temperature, as long as the animals have access to drinking water (Fuller et al. 2014). It seems that under normal conditions, large mammals activate EHL in the face of environmental heat load, and that heat loss is sufficient to maintain core Tb (but see later). When water intake is limited, osmotic stress causes changes in thermoregulation that may be due to the integration of signals from osmo-sensitive and thermally sensitive neurons in the hypothalamus, or may be due to the activation of selective brain cooling that reduces hypothalamic temperature and suppresses the activation of EHL, leading to a larger amplitude of core Tb (Fuller et al. 2014).

In hot, humid environments, sweating and panting might be stimulated, but the high humidity is not conducive for evaporation. In humid tropical environments, mammals generally have a very high heat loss capacity and usually other adaptations that facilitate heat balance. For example, the predominant cattle breeds used for meat and milk production are based on Bos taurus taurus, but these breeds generally do not cope well with the tropics; breeds based on Bos taurus indicus cope better with tropical conditions. The essential characteristics of the indicus breed that makes it better in the tropics are higher sweating capacity (more and larger sweat glands) and a lower metabolic heat production (Beatty et al. 2006). The lower metabolic heat production is an advantage in the heat, but conflicts with the intention of animal production, which is to produce more animals. Attempts to meld the productivity of taurus with the heat tolerance of indicus have not generally been successful; any increase in productivity is usually associated with a decrease in heat tolerance (Cunningham 1991).

A reduction in metabolic rate can also be achieved by a reduction in food intake, which should assist with heat balance in hot environments. Many mammals do decrease feed intake in the heat. Indeed, rats forced to maintain the same level of food intake in the heat suffer mortality (Hamilton 1976). For some grazing species, reduced food intake may be an indirect effect of shade-seeking in hot conditions, but even in the absence of radiant heat, an effect of heat load on food intake is well described (Yousef et al. 1968). Metabolic heat production can also be reduced during prolonged exposure to heat by lowered thyroid activity (Yousef et al. 1967). (p.320) Thyroid activity decreases when food intake is restricted (Blincoe & Brody 1955) and so it is logical to reason that the reduction in food intake causes a decrease in circulating thyroid hormone, with a subsequent decrease in metabolism. However, several studies have shown that thyroid activity and heat production decrease even if animals are force-fed to maintain the thermoneutral level of feed intake during heat exposure (Yousef & Johnson 1966; Kibler et al. 1970), so it seems likely that Tb directly affects thyroid hormone release and metabolism—evidence for a direct effect of Tb is equivocal. Andersson (1963) suppressed feed intake in goats by heating their hypothalamus, but Spector et al. (1968) and Hamilton and Ciaccia (1971) showed that brain heating increased feed intake in rats. Effects of Tb on rumen function might also contribute to reduced appetite of ruminant mammals (Collier et al. 1982).

## 4.2.2 Avoiders

Small mammals, in particular, are able to seek refuge from hot environmental conditions. Many are nocturnal, avoiding harsh daytime temperatures, but some are diurnal and some are flexible with respect to timing of activity (MacMillen 1972; Walsberg 2000). For example, two species of spiny mice coexist in Israeli desert areas; the common spiny mouse (Acomys carinus) is nocturnal whereas the golden spiny mouse (Acomys russatus) is more behaviourally flexible and is frequently diurnal (Shkolnik 1971). Diurnally active desert mammals generally avoid exposure to direct sunlight, but can resort to physiological mechanisms to cope with solar heat loads. The white-tailed antelope ground squirrel (Ammospermophilus leucurus) has an exceptionally labile Tb when diurnally active, and especially in summer they rely on periodic bouts of hyperthermia to store heat when surface-foraging, then they dump the stored heat by thigmothermy or retreat to their burrow to dump heat to cool soil (Figure 4.10). For these small diurnal desert mammals, hyperthermia is a short-term strategy of heat storage then dumping, compared to the day-long strategy of large mammals (such as camels; see Figure 4.9), because of their much higher surface area to volume ratio and lower thermal inertia, which means they heat and cool much faster.

Figure 4.10 A: An antelope ground squirrel (Ammospermophilus leucurus) dumping heat to the substrate by thigmothermy.

Photograph by C. E. Cooper. B: Body temperature of an antelope ground squirrel during a summer day, compared with environmental temperature (standard operative temperature, Tes) in sun and shade. Modified from Chappell and Bartholomew (1981b).

Aestivation, or ‘summer dormancy’, is a common response of many vertebrate and invertebrate animals during which metabolic rate and water loss are depressed (Withers & Cooper 2010). For ectothermic vertebrates, the metabolic depression associated with aestivation is intrinsic, apparently unrelated to body temperature, pO2, or body water changes (Withers & Cooper 2010). For endothermic mammals (and a few birds), aestivation is essentially the ecological equivalent of winter torpor and hibernation (see 4.1.2), but in hot and dry conditions. Aestivation is phylogenetically widespread amongst mammals (Geiser 2010). Some mammalian aestivators, such as the cactus mouse (Peromyscus eremicus; MacMillen 1965) aestivate on a daily basis, whereas others such as the Mohave ground squirrel (Citellus mohavensis) aestivate for weeks to months (Bartholomew & Hudson 1960).

(p.321) The pattern of change in body temperature and metabolic rate with Ta is the same for aestivation as for torpor and hibernation. In fact, we cannot discriminate on physiological grounds between single-day aestivation, multi-day aestivation, single-day winter torpor, and multi-day hibernation (Wilz & Heldmaier 2000; Van Breukelen & Martin 2015) other than aestivation occurs in the summer, at higher ambient temperatures, under dryer conditions. So, summer aestivation presents different challenges than winter torpor and hibernation, with high ambient temperatures and low water availability being as or even more challenging than lack of food (Geiser 2010). For aestivating cactus mice, Tb declines with Ta, becoming closer to Ta at lower temperatures; for example, $(Tb−Ta)$ is < 1°C at $Ta=10°C$, and about 3°C at Ta of 20–30°C (MacMillen 1965; Figure 4.11). Since aestivation commonly occurs at relatively high Ta, there is presumably a lesser reduction (p.322) of metabolic rate and water loss rate than torpor/hibernation at lower Ta, as Tb must remain above Ta. Furthermore, aestivation presumably results in a poorer relative water economy $(RWE=MWP/EWL)$ than for normothermy (as for torpor; see 3.2.4.2), because aestivation reduces MR more than EWL, and MWP is likely less than EWL (i.e. $RWE<1$).

Figure 4.11 Body temperature and metabolic rate of cactus mice (Peromyscus eremicus) when normothermic and aestivating.

Modified from MacMillen (1965).

Aestivation can be induced by food restriction, water restriction, or both. For example, some cactus mice enter aestivation in response to water restriction, but some individuals do not (MacMillen 1965). Some (20%) least gerbils (Gerbillus pusillus) enter aestivation when water-deprived, but more (88%) do so when food- and water-deprived (Buffenstein 1985). The stripe-faced dunnart (Sminthopsis macroura) does not increase its use of torpor when water-deprived if it has access to moist food, but does when only dry food is provided (Song & Geiser 1997). The tendency of mammals to enter aestivation in response to water restriction seems somewhat equivocal, and their response to water deprivation with food available depends on the potential of the food to provide sufficient water. Many species restrict food intake in response to water deprivation. Nevertheless, food and water restriction are powerful stressors than induce aestivation.

Little is known of the physiological role of aestivation in the field. Two rodents found on rocky outcrops in the Namib Desert (pygmy rock mouse, Petromyscus collinus; Namaqua rock rat, Aethomys namaquensis) normally have a Tb of 33.6 and 34.0°C, respectively, but during aestivation at Ta of 18–22°C, their Tb is reduced to 18.0–23.6°C (Withers et al. 1980). In the field, both species have remarkably low (p.323) water turnover rates, of 0.8 ± 0.1 and 2.2 ± 0.2 ml day−1 respectively, presumably reflecting the use of aestivation. Interestingly, their water turnover rates increased in the field after an advective fog, to 1.4 ± 0.2 and 3.2 ± 0.2 ml day−1, respectively.

# 4.3 Underground Environments

A fossorial (subterranean) existence has developed in several mammalian taxonomic orders, including marsupials, rodents, insectivores, and edentates, and includes species that live almost exclusively underground and only rarely come to the surface. The convergent evolution of fossorial mammals is a fascinating and puzzling evolutionary phenomenon. Convergent morphological features for burrowing include compact bodies, short tails and necks, microphthalmic eyes, and large and powerful forefeet, pectoral girdles, and associated muscles (McNab 1966; Nevo 1999; Warburton et al. 2003; Warburton 2006). The abiotic microenvironment inhabited by fossorial mammals is relatively humid, hypoxic and hypercapnic, and thermally constant (McNab 1966; Arieli 1979; Cooper & Withers 2005). Living underground has resulted in a number of physiological adaptations to these conditions, including respiratory responses to hypoxia and hypercapnia, mechanisms for maintaining heat balance in warm and humid environments, and sustaining the metabolic cost of digging burrows.

Many mammals live both underground and above ground, often on a daily cycle. These semi-fossorial species are generally not so specialized for an underground existence with respect to their morphology or physiology, because they are also active above ground. Nevertheless, many semi-fossorial species show similar adaptations to an underground existence as fully fossorial species, albeit often less extreme.

## 4.3.1 Hypercapnic Hypoxia

An important constraint faced by fossorial mammals is potentially low O2 availability (hypoxia) and excess CO2 (hypercapnia) underground (Arieli 1979, 1990). The depletion of O2 in burrow environments due to animal (or soil microbe) metabolism is directly correlated with an increase in CO2, so this form of hypoxia is termed ‘hypercapnic hypoxic’ (cf. ‘hypoxic hypoxia’ with altitude; 4.4.1). For example, levels of CO2 as high as 6.1% and O2 as low as 7.2% were recorded in the breeding mounds of a blind mole rat (S. carmeli) in a flooded, poorly drained field of heavy clay soil with very high volumetric water content (Shams et al. 2005). Gaseous interchange between burrows and the atmosphere depends on the gas permeability properties of the soil (Wilson & Kilgore 1978; Withers 1978; Arieli 1979), and some air ventilation caused primarily by animal movements (Buffenstein 2000) or wind-induced convection (Vogel et al. 1973). These factors mean that burrow gas (p.324) composition can differ considerably from atmospheric air, and any activity of the inhabitants increases such differences. Models of diffusive gas exchange (Withers 1978) and experimental data show that, unless the soil is completely devoid of biotic substances, burrow atmospheres will always be hypoxic and hypercapnic relative to the surface atmosphere. Faced with low pO2 and the potential CO2 perturbation of their acid–base balance in the burrow atmosphere, burrow-dwelling mammals would be expected to have physiological mechanisms to avoid excessive energy expenditure.

Tomasco et al. (2010) reviewed studies of the respiratory responses of mammals to underground environments. The primary respiratory driver for mammals is CO2 rather than O2 (see 3.3.3), so hypoxia is often less significant than hypercapnia for non-fossorial mammals. An attenuated ventilatory response to hypoxia is not a general characteristic of semi-fossorial or fossorial mammals. Many semi-fossorial species, such as the Syrian hamster (Mesocricetus auratus), woodchuck (Marmota monax), golden-mantled (Spermophilus lateralis) and Columbian (S. columbianus) ground squirrels, do not differ in their response to hypoxia to similar-sized non-fossorial species.

However, some burrowing species do have an attenuated sensitivity to hypoxia; the hyperventilatory response to hypoxia of the echidna, armadillo (Dasypus novemcincus), and hairy-nosed wombat (Lasiorhinus latifrons) is depressed (Frappell et al. 2002). The Chilean fossorial coruro (Spalacopus cyanus) and semi-fossorial degu (Octodon degus) both have a low sensitivity to hypoxia, but the coruro has a more acute response to hypoxia than the degu (Tomasco et al. 2010). These findings support the conclusion that persistent changes in the neural control system for respiratory ventilation are generated based on prior experience (Mortola 2004). Chronic sustained hypoxia elicits plasticity in the carotid body chemoreceptors, with delayed effects on the central neural integration of carotid chemo-afferent neurons that become more prominent as the duration of hypoxia is extended. In the case of fossorial coruros, hypoxia may last for long periods inside closed burrows, and their enhanced ventilatory response is the result of the potentiation of the carotid chemoreflex to hypoxia. Degus, however, might tolerate intermittent hypoxia, experiencing hypoxia only while resting at night; this may require plasticity via central neural mechanisms of respiratory control (Ling et al. 2001).

Burrowing mammals often encounter significant hypercapnia, and generally have a reduced sensitivity and response to CO2 compared to non-fossorial (and non-diving) species (Boggs et al. 1984; Tomasco et al. 2010). The most CO2-insensitive species are the fossorial pocket gopher (Thomomys bottae; Darden 1972) and Middle East blind mole rat (Spalax ehrenbergi; Arieli & Ar 1979). Unlike hypoxia, which seems to result in adaptations via sensory input, severe hypercapnia is often associated with long-lasting depression of respiratory motor output. During hypercapnia, respiratory activity initially increases, but then decreases (p.325) progressively when the hypercapnia is sustained. Interestingly, marsupials have a low CO2 sensitivity, and there would appear to be little difference between above-ground and burrowing species (Frappell & Baudinette 1995; Frappell et al. 2002; Figure 4.12). A reduction in the gain of the ventilatory control system in marsupials to a level akin to that of burrowing placentals could reflect potential exposure to hypercapnia during development—burrowing animals within the burrow and marsupials within the pouch. In fact, exposure to hypercapnia during development causes long-lasting attenuation of the acute hypercapnic ventilatory response so hypercapnia-induced developmental plasticity may play a role in the reduced hypercapnic ventilatory responses commonly observed in fossorial mammals.

Figure 4.12 Respiratory responses to hypercapnia (expressed as percentage change in minute volume per kPa partial pressure of CO2) for non-fossorial placental mammals, fossorial or diving placental mammals, marsupials in general, and a semi-fossorial marsupial (wombat). Dashed lines are averages ± standard deviation for non-fossorial, and burrowing and diving placental mammals.

Modified from Frappell et al. (2002). Reproduced with permission of the University of Chicago Press.

## 4.3.2 Temperature and Energetics

Burrowing mammals can have a reduced capacity for convective heat loss, and evaporative heat loss can be low, because of the relatively high Ta and ambient humidity in burrows (McNab 1966, 1979). Many fossorial mammals have a slightly lower-than-expected $Tb(35−37°C)$, which would reduce their metabolic heat production and reduce the risk of overheating. The naked mole rat (Heterocephalus glaber) is exceptional in this regard, having a Tb of only 32°C, which is close to soil temperature (about 31°C), and being essentially ectothermic even when huddling in groups (McNab 1966; Withers & Jarvis 1980; Buffenstein & Yahav 1991). (p.326) Fossorial mammals also often have a high capacity for non-evaporative heat dissipation, reflecting the limited capacity for evaporation in a humid burrow. For example, the Talas tuco-tuco (Ctenomys talarum), which shelters in a sealed burrow but forages above ground (Baldo et al. 2015), has a considerable capacity to increase dry thermal conductance (by about fourfold) to facilitate non-evaporative heat dissipation at $Ta=35°C$ compared to lower Tas, compared with only about a 1.3 times increase in EHL.

Fossorial mammals generally have to dig their own burrow, so they must cope with the high energy costs of digging as well as foraging (Ebensperger & Bozinovic 2000). Digging can be an energetically demanding process, and requires some anatomical specializations of the limbs or incisors. Loosened soil must also be transported back along the burrow and disposed of, often via a lateral tunnel to the surface (Vleck 1979). The metabolic cost of digging is therefore high relative to other costs of transport for mammals, and varies with soil density, cohesiveness, and burrow size and structure. Digging can be 360 to 3,400 times more energy-consuming than moving the same distance by walking (Vleck 1979). The cost of transport has been found to be similarly high for a range of burrowing mammals (Figure 4.13).

Figure 4.13 Net cost of transport for burrowing mammals (black symbols) and sand-swimming mammals in sand (grey symbols) and walking (open symbols), and walking and running mammals (regression line).

Data from Vleck (1979), Du Toit et al. (1985), Lovegrove (1989), Seymour et al. (1998), and Withers et al. (2000). Modified from Withers et al. (2000) and Xu et al. (2014).

A different ‘burrowing’ strategy is sand-swimming by small fossorial mammals that live in loose, aeolian sand dunes. Sand-swimming is ‘swimming’ through the (p.327) loose sand using limb movements and an undulatory body motion to push the sand away from the front of the animal; there is no need to move spoil, as must a burrowing mammal, because loose sand fills the space left behind the sand-swimmer. The Namib Desert golden mole (Eremitalpa granti namibensis) is an adept sand-swimmer (Holm 1969). It is nocturnally active on the dune face, foraging and periodically ‘dipping’ its head into the sand (to sense food); it rests under the sand during the day. Its metabolic cost of sand-swimming is considerably less than that of burrowers in compacted soil, but nevertheless substantially higher than when walking on the surface (Seymour et al. 1998; Figure 4.13). The Australian marsupial mole (Notoryctes caurinus) is a remarkably convergent sand-swimmer that, when sand-swimming or walking, has a similar net cost of transport as the Namib Desert golden mole (Withers et al. 2000).

Rates of metabolism have been linked to different biotic and abiotic factors as evidence of metabolic adaptation to environmental conditions and geographic distribution (Bozinovic 1992; Spicer & Gaston 1999; McNab 2002). The observation that fossorial mammals have lower than allometrically expected BMR (McNab 1979; Contreras 1983; Lovegrove 1986) has been explained as an adaptation to hypoxic conditions within the burrow and the energetic costs of digging. Several competing hypotheses have also been suggested to explain how physical microenvironmental conditions and underground life affect the energetics of fossorial mammals. Two of these are the thermal-stress hypothesis and the cost-of-burrowing hypothesis (but see Lovegrove and Wissell, 1988, and Lovegrove, 1989, for alternative ecological hypotheses, such as the aridity-food distribution hypothesis).

The thermal-stress hypothesis posits that a lower mass-independent BMR reduces overheating in burrows where the capacity for convective and evaporative heat loss is low (McNab 1966, 1979). The cost-of-burrowing hypothesis states that a lower mass-independent BMR compensates for the extremely high energetic cost of digging during foraging activity (Vleck 1979, 1981; see 4.3.3). Using phylogenetic and conventional allometric analyses, White (2003) examined both hypotheses for approximately 100 species of fossorial and semi-fossorial species in a biogeographic scenario. He concluded that mammalian species from mesic habitats support the thermal-stress hypothesis, but species from arid habitats support the cost-of-burrowing hypothesis. BMRs of large (> 77 g) fossorial (i.e. truly subterranean) mammals from mesic and arid habitats are not different from BMRs of their semi-fossorial (i.e. less adapted to digging) counterparts despite expected differences in their foraging costs, a result consistent with the thermal-stress hypothesis (White 2003). On the other hand, small (< 77 g) fossorial mammals from arid habitats have lower BMRs than their similarly sized but semi-fossorial counterparts, a result consistent with the cost-of-burrowing hypothesis (White 2003). These results led to the conclusion that the two hypotheses are not mutually exclusive. Bozinovic et al. (2005) tested the thermal-stress and cost-of- burrowing (p.328) hypotheses at an intraspecific level. They compared seven populations of the coruro from different geographic localities with contrasting habitat conditions. Their results did not support the thermal stress or the cost-of-burrowing hypotheses. Coruros from habitats with contrasting climatic and soil conditions had similar basal and digging metabolic rates when measured under similar semi-natural conditions. It is possible that S. cyanus originated in Andean locations where it adapted to relatively hard soils, and later, when populations dispersed into coastal areas characterized by softer soils, they may have retained the original adaptation without further phenotypic changes.

The thermal-stress hypothesis neglects potential behavioural thermoregulatory adjustments. There are daily and seasonal patterns in the timing of daily activity for small ground-dwelling mammals (e.g. Chappell & Bartholomew 1981a,b). Coruros, for instance, adjust their surface activity according to diurnal and seasonal changes in environmental temperature (Rezende et al. 2003). They decrease their surface activity (digging, foraging, and vigilance) during the warmest time of day (from midday to early afternoon) in summer, but not during winter (Rezende et al. 2003). Thus, coruros might cope with thermal constraints inside burrows by shifting their activity according to changes in environmental temperature in time and space.

Low metabolic rates of fossorial small mammals may also be interpreted as adaptation to the low pO2 of underground environments (see Nevo 1999 and references therein), which would predict a similar lower BMR at high altitude (although this may be confounded by adaptation to lower Ta at higher altitude). Future study of Ctenomys species might provide a suitable test of this hypothesis, as fossorial ctenomyids include Andean species living above 3,500 m. Indeed, there are some parallels in the physiology of fossorial and high altitude species with respect to haematological adaptations to low pO2 (see 4.4.1). It has been postulated small erythrocytes have a larger relative surface area and hence faster diffusional exchange. For example, the fossorial Palestine mole rat (Spalax ehrenbergi), which is very tolerant of hypoxia and hypercapnia (Arieli et al. 1977), has small erythrocytes (Shams et al. 2005). This allows a reduced capillary diameter, which in turn allows a high tissue capillary density and mitochondrial volume. Its erythrocyte count and lung diffusion capacity are high, and it has functional specializations of its haemoglobin and myoglobin, all of which presumably facilitate gas exchange under hypoxic conditions.

# 4.4 High Altitude Environments

Mt Everest (8,848 m; Himalayas, Tibet/Nepal) is the highest mountain on Earth while Mt Aconcagua (6,961 m; Andes, Argentina-Chile) is the highest mountain outside Asia. The peaks of these and other high mountains are above the (p.329) highest elevations inhabited by mammals (see McNab 2002). The yak (Bos grunniens) is the mammal that inhabits the highest elevations, living at up to 5,500 m, but domesticated pack yaks can work intermittently at up to 7,200 m (Wiener et al. 2006). They, along with ibex (Capra [ibex] sibirica), Tibetan gazelle (Procapra picticaudata), and vicuna (Vicugna vicugna), are the most eurybaric mammals, tolerant of low barometric pressures. Many stenobaric mammals (less tolerant of low pressure), such as rabbits, pikas, felids, and mountain goats and sheep, occur in less extreme but nevertheless high altitude environments (an elevation of 2,500 m defines the lower limit of ‘high altitude’).

High altitude habitats are characterized by low pO2, cold and significantly variable Ta, low humidity, high solar radiant energy, high winds, and low primary productivity. As such, high altitude is a multiple stress environment for mammals (Monge & León-Velarde 1991). Although the identification of specific physiological traits has been of evident value, it is the overall metabolic response to hypoxia that gives an integrated measure of the extent of adaptation to high altitudes (Rosenmann & Morrison 1975). In this vein, adaptation to high altitude involves many steps in the O2 cascade (see 2.5.1) from the atmosphere to its final utilization in the cell (Dill 1938; Morrison 1964). Mammals can function to some extent during hypoxia with no change in their MR, but at some threshold pO2 their MR decreases. Classical studies in comparative physiology demonstrated that this critical pressure (PC) is a measure of the hypoxic sensitivity of the species, but is also related to the level of oxygen demand by the individual (Rosenmann & Morrison 1974b). As pO2 is lowered below PC, the MR decreases to a lethal level.

## 4.4.1 Hypoxic Hypoxia

The most important factor that may limit the overall altitudinal range of mammals is a lack of oxygen, a consequence of the decrease in barometric pressure (Pb) with increased elevation (Dill et al. 1964). Altitudinal hypoxia (or ‘hypoxic hypoxia’) is low O2 partial pressure, resulting from the altitudinal decline in Pb; the partial pressure of CO2 is also reduced because of the lowered Pb (whereas pCO2 is elevated in the ‘hypercapnic hypoxia’ of burrows).

Barometric pressure declines with increasing elevation (Table 4.2), from ‘standard atmospheric pressure’ of 101.3 kPa at sea level to 55 kPa at the highest human habitation (La Rinconada, Peru, 5,100 m), 43 kPa at 7,010 m (the usual air-breathing ceiling for humans, where arterial saturation is only 50%), 15 kPa at 14,326 m (the ceiling for humans breathing pure O2, where arterial saturation is again only 50%), to 6.3 kPa at 18,190 m, which is the saturation water vapour pressure for water at 37°C, meaning that water boils at 37°C at elevations above 18,190 m. The physiological effect of this decrease in Pb with elevation arises from the concomitant decline in the partial pressure of oxygen; , where FO2 is the fractional content of air that is O2. At sea level, FO2 is 0.2095, and this (p.330) value does not change significantly with elevation, so therefore pO2 at sea level is and at La Rinconada (5,100 m) is . The decline in ambient pO2 with elevation causes a decline in alveolar pO2, and other levels through the O2 cascade to the mitochondria, and compromises O2 delivery.

Table 4.2 Effect of elevation on barometric pressure and partial pressure of oxygen.

Location

Elevation (m)

Barometric pressure (Pb, kPa)

Oxygen partial pressure* (pO2, kPa)

Sea level

0

101.3

21.2

Mt Kosciuszko, Australia

2,228

78

16.3

Lower limit of ‘high altitude’

2,500

76

15.9

Lower limit of ‘very high altitude’

3,658

66

13.8

Mt Blanc, France

4,810

57

11.9

Mt Kilimanjaro

4,877

57

11.9

La Rinconada, Peru (highest human habitation)

5,100

55

11.5

Lower limit of ‘extremely high altitude’

5,500

53

11.1

Mt Elbrus

5,642

52

10.9

Mt McKinley

6,190

48

10.1

Mt Aconcagua

6,961

43

9.0

Human air ‘ceiling’ +

7,010

43

9.0

Mt Everest

8,848

34

7.1

Human pure O2 ‘ceiling’ + +

14,326

15

3.1

Water boils (at 37°C)

18,190

6.3

1.3

(*) pO2 = 0.2095 Pb;

(+) limit of consciousness when breathing air;

(+ +) limit of consciousness when breathing pure O2

Humans lose consciousness if their arterial percentage O2 saturation is less than about 50%, which occurs at an elevation of 7,010 m, where Pb is 43 kPa and pO2 is 9.0 kPa; this defines the elevation ‘ceiling’ above which a human breathing air will lose consciousness (Table 4.2). This ceiling is below the top of Mt Everest, so in theory mountaineers should not be able to summit Mt Everest if they breathe ambient air. However, breathing pure O2 raises the ceiling to , which facilitated the first summit of Mt Everest by Tenzing Norgay and Edmund Hillary, aided by the physiological research of Griffith Pugh, who demonstrated that the weight of oxygen tanks was more than compensated for by the extra aerobic capacity that oxygen breathing afforded. These limitations apply to un-acclimated humans; two highly trained (p.331) and altitude-acclimated mountaineers (Reinhold Messner and Peter Haberler) proved the fallacy of this widely held dogma concerning altitudinal limits when they ascended Mt Everest without breathing pure O2 (Oelz et al. 1986).

There are two basic types of physiological responses of animals to reduced pO2: (1) for oxyconformers, metabolism is linearly dependent on ambient pO2; (2) for O2-regulators, such as mammals, metabolism is constant with decreasing pO2 to the Pc, below which MR decreases linearly with pO2. Their Pc is modified by several exogenous and endogenous factors, including Ta (Rosenmann 1987). Thus, mammals that live or evolved in environments with lower pO2 have a lower VO2 and lower Pc than mammals living at higher pO2 (Rosenmann & Morrison 1975). Consequently, Pc can be used as a measure of sensitivity to hypoxia in different mammal species inhabiting high and low altitudes (Rosenmann & Morrison 1974).

Overall, mammals are much more sensitive to and dependent on environmental pO2 than birds, and consequently their altitudinal distribution is more restricted. So, although birds live and nest at high altitudes , migrate to even higher altitudes (Bouverot 1985), and can fly still higher (a griffin vulture collided with an aircraft at 11,300 m; Laybourne 1974), only a few mammals can attain 6,000 m (see McNab 2002). Unlike birds (which have highly efficient cross-current lungs with unidirectional airflow; Bicudo et al. 2010), mammals have an alveolar lung with a tidal (inspiration-expiration flow change). In functional terms, this and the alveolar dead space limit the maximum alveolar pO2, which can only be increased by increased respiratory ventilation. Indeed, the first response of mammals faced with experimental hypoxia is increased respiratory frequency (fR; Rosenmann & Morrison 1974; Nice et al. 1980). Rodent species that usually live at low altitudes generally have a much greater increase in fR in response to hypoxia than high altitude species (Rosenmann & Morrison 1975). A similar pattern can be seen with Tb, which decreases more markedly in lowland than high altitude species exposed to similar levels of hypoxia. Thus, mammals generally have differential tolerance to hypoxia, which is closely related to the altitude at which they live. Accordingly, the metabolic response to hypoxia (Pc) of mammals depends on the altitude of their evolutionary point of origin (Novoa et al. 2002). Data for 27 rodent species from high and low altitudes show that metabolic demand greatly influences the magnitude of Pc (Table 4.3). Overall, native high altitude species have a significantly lower Pc (14.9 kPa) than native low altitude species (16.2 kPa).

Table 4.3 Critical pressure of oxygen (Pc) in rodents from low and high altitude. Table modified from Novoa et al. (2002); data from Rosenmann and Morrison (1975) and Novoa et al. (2002).

Species

Body mass (g)

Critical pO2 Pc* (kPa)

Low altitude species

• Phyllotis darwini limatus

56

18.1

• Octodon degus

181

18.5

• Acomys cahirinus

49

17.5

• Citellus undulatus

472

17.5

• Glaucomys volans

67

17.1

• Baiomys taylori

8

16.9

• Microtus oeconomus

36

16.4

• Meriones unguiculatus

48

16.4

• Dicrostonyx g.stevensoni

52

16.4

• Clethrionomys rutilus

33

16.3

• M. musculus (white)

35

15.5

• M. musculus (feral)

17

14.9

• Calomys callosus

48

14.8

• Peromyscus m. bairdi

21

14.5

• Akodon olivaceus

27

14.1

• Microtus pennsylvanicus

29

12.9

High altitude species

• Phyllotis darwini chilensis

29

15.9

• Eligmodontia puerulus

21

15.9

• Akodon andinus (Farellones)

29

15.9

• Cavia porcellus

481

15.6

• Mus musculus (feral)

19

15.5

• Akodon boliviensis

26

15.1

• Akodon andinus (Parinacota)

27

14.5

• Calomys ducilla

18

14.4

• Phyllotis darwini posticalis

78

14.3

• Ochotona rufescens

176

13.3

• Auliscomys boliviensis

87

13.3

(*) Pc corrected for different metabolic loads according to and an increase in Pc of 1.76 kPa per metabolic load (after Rosenmann & Morrison 1975).

Mammals have a reduced maximal metabolic rate (MMR) at low Pb (< 75 kPa; Rosenmann & Morrison 1975), and this impairs their capacity for sustained activity and thermoregulation in the cold. Consequently, some native high altitude mammals have specific adaptations that maximize gas exchange and thermogenesis. There is strong evidence that evolutionary adaptation of hypoxia tolerance is associated with the O2 content of blood leaving the lungs, which in turn is highly dependent on haemoglobin (Hb), the number and size of red blood corpuscles, haematocrit (Hct), and Hb-O2 affinity (P50). For example, high altitude adaptations of yaks (p.332) (p.333) include increased red corpuscle count and decreased mean corpuscular volume (Ding et al. 2014). Yaks also have large lungs and heart relative to their mass, and their haemoglobin has a high O2 affinity (Wiener et al. 2006). Juveniles of two South American Andean rodents (Abrothrix and Phyllotis) born in captivity at sea level retain the high Hb and Hct of adults, maintaining their functional architecture, and suggesting evolutionary adaptation to high altitude (Ruiz et al. 2005).

The affinity of Hb for oxygen and its relationship with altitude and hypoxia have been widely examined as physiological adaptations, as have the size and number of erythrocytes. One solution to low pO2 at altitude is a high Hb-O2 affinity to facilitate O2 loading in the lungs. However, a high affinity would not be beneficial to overcome the problem of delivery of O2 to tissues (see Figure 3.21). It seems that in hypoxic environments it is O2 loading that is paramount as without adequate arterial O2 saturation, facilitating O2 delivery to tissues becomes less relevant. Physiologists initially considered the decreased Hb-O2 affinity and high Hb and Hct of humans to be beneficial adaptations to high altitude, whereas they now acknowledge that these responses may not be generally adaptive for high altitude mammals. It is now generally acknowledged that high Hb-O2 affinity is characteristic of hypoxia-tolerant mammals, and this pattern has been observed for widely divergent vertebrate species. Nevertheless, it is important to interpret ‘adaptations’ to altitude hypoxia in an evolutionary perspective. For instance, the high Hb-O2 affinity (low P50) of South American camelids was initially considered to be an adaptation to high altitudes. Studies of Old World camels and dromedaries now indicate that high Hb-O2 affinity is a common trait in the family, being present in the lineage before the colonization of the Andes, suggesting that camelids in South America exploited their low haemoglobin P50 as an exaptation to life at high altitude (see Rezende et al. 2005).

The respiratory functions of Hb are a product of both its intrinsic O2-binding affinity and interactions with allosteric effectors such as H+, Cl, CO2, and organic phosphates (e.g. 2,3-diphosphoglycerate, DPG). For many high altitude mammals, it is possible to identify specific mechanisms of Hb adaptation to hypoxia. Indeed, functional studies of human Hb mutants also suggest that there is ample scope for evolutionary adjustments in Hb-O2 affinity.

At a molecular level, Storz et al. (2009) showed that adaptive modifications of heteromeric proteins (such as haemoglobin) can involve genetically based changes in single subunit polypeptides or parallel changes in multiple genes that encode distinct, interacting subunits. Their evolutionary and functional analysis of duplicated globin genes showed that natural populations of deer mice are adapted to different elevational areas. The adaptation of the haemoglobin of P. maniculatus to high altitude involves parallel functional differentiation at multiple unlinked gene duplicates, two α‎-globin paralogs, and two β‎-globin paralogs. Differences in the O2-binding affinity of the alternative β‎-chain haemoglobin isoforms are entirely attributable to allelic differences in sensitivity to DPG, a classical allosteric cofactor that (p.334) stabilizes the low-affinity, deoxygenated conformation of the haemoglobin tetramer. The two-locus β‎-globin haplotype that predominates at high altitude is associated with suppressed DPG sensitivity and therefore increased haemoglobin-O2 affinity and enhanced pulmonary O2 loading under hypoxia.

Deer mice (Peromyscus maniculatus) have an array of Hb polymorphisms, which are inherited as two different haplotypes—similar to what is expected with Mendelian inheritance of a single locus with two alleles—and haplotype frequencies are correlated with altitude (Snyder et al. 1988). Individuals with the high altitude haplotype had increased Hb-O2 affinities and MMR during cold exposure when measured at high altitudes, whereas the opposite was observed for the low altitude haplotype, suggesting that the correlation between allelic frequencies and altitude may reflect local adaptation.

For humans, the evolutionary adaptation of Tibetans to hypoxic high altitudes on the Tibetan plateau is remarkable, as it provides evidence of incorporation of genetic material from another species into the human genome (Bigman 2010; Yi et al. 2010). The hypoxia pathway gene (EPAS1) was identified as having the most extreme signature of positive selection, being associated with phenotypic differences in Hb concentration at high altitude. This gene has an extraordinary haplotype structure that can only be explained by introgression of DNA from Denisovan-related individuals (fossil hominins) into humans (Huerta-Sánchez et al. 2014), with the selected haplotype found only in Denisovans and Tibetans, suggesting that admixture with other hominin species has provided genetic variation that facilitated humans adapting to new environments.

(p.335) A very different metabolic strategy for adapting to altitude hypoxia is to minimize the amount of O2 required for metabolism by modifying metabolic pathways. Hochachka (1985) suggested that increased reliance on carbohydrates as the substrate for metabolism requires less O2, since the chemical stoichiometry of carbohydrate metabolism requires less O2 per kJ than lipid or protein metabolism (see Table 2.5). Schippers et al. (2012) found that two species of Andean rodent (Phyllotis andium and P. xanthopygus) use more carbohydrate than low altitude counterparts (P. amicus and P. limatus) during activity, suggesting that the high altitude species have indeed adapted their fundamental metabolic physiology in response to hypoxia (Figure 4.14). Cardiac muscle of the high altitude species also had enhanced oxidative capacity, facilitated by higher citrate synthase and isocitrate dehydrogenase (although skeletal muscle had no striking differences in enzyme levels).

Figure 4.14 Comparison of carbohydrate metabolism at 75% of maximal metabolic rate, showing a higher respiratory exchange ratio for high altitude Andean rodents (bold font) compared to low altitude species, suggesting the independent evolution of high carbohydrate substrate use (high RER) as an adaptation to hypoxia.

Modified from Schippers et al. (2012).

## 4.4.2 Thermal Balance

Cold stress is the second major physiological challenge at high altitude. The typical response of mammals to cold stress is increased thermogenic capacity (see 4.1), but the hypoxia associated with altitude compromises gas exchange, hence metabolism (p.336) and thermogenesis. Increased insulation is consequently an especially viable strategy for counteracting the cold at high altitude. For yaks, temperature is the most important factor determining their survival and distribution (Wiener et al. 2006). They prefer an average ambient Ta of about 5°C, and survive well at even $−40°C$. Their primary thermal strategy is to conserve heat, not thermogenesis. They have a thick fur coat, a thick layer of subcutaneous fat, and thick skin, all of which aid heat conservation. They almost completely lack functional sweat glands.

Thermogenesis is nevertheless important for some high altitude mammals, especially small species for which increased insulation is not a viable option. The plateau pika (Ochotona curzoniae; 3,000–5,000 m) has an unusually high resting metabolic rate and capacity for non-shivering thermogenesis (Li et al. 2001). Hayes and O’Connor (1999) found some evidence for natural selection of maximum thermogenic capacity (VO2,max) amongst high altitude (3,800 m) deer mice in the field, which should facilitate thermoregulation in the cold and also allow sustained maximal activity. In one of two years, there was strong selection for high VO2,max in the deer mice, suggesting that a high thermogenic capacity is important in the cold (see 4.1); there was also weak selection for reduced body mass (counter to predictions from Bergmann’s rule).

# 4.5 Aquatic Environments

Diving mammals have a suite of adaptations that range from the molecular and cellular level, through morphology and physiology, to behaviour and ecology. The adaptive consequence of many of these is to facilitate their swimming and to extend their breath-hold time, hence depth of diving and foraging duration (aerobic dive limit, ADL). However, breath-hold time and foraging duration are only some of the problems faced by aquatic mammals; thermal balance can also be problematic for diving mammals. In cold water, considerable heat is lost to water, so heat retention is important; in warm water, they can experience the opposite problem of having sufficient heat loss to maintain constant Tb. For marine mammals, water and solute balance can also be problematic, especially if they are consuming food that is isosmotic with seawater (e.g. most marine invertebrates, particularly very hydrated ones, such as jellyfish). Seawater is generally their only source of drinking water, so this exacerbates their salt load if they do indeed drink. Since mammals lack specialized salt glands (unlike marine birds; Bicudo et al. 2010), they rely on kidneys for maintenance of body water volume and solute balance. Finally, deep-diving marine mammals also experience considerable hydrostatic pressures (pressure increases by 1 atm every 10 m of water depth). For the deepest-diving mammals (e.g. Cuvier’s beaked whale, Ziphius cavirostris; 2,992 m), this corresponds to a hyperbaria of about 29,652 kPa. Even though water is essentially incompressible, such high pressures can have physical and biochemical effects on cells, as well as (p.337) extreme effects on air spaces (e.g. lung volume) and the dynamics of gas exchange (hence problems such as the ‘bends’).

Semi-aquatic mammals, such as beaver, muskrat, and sea otters, have some adaptations for swimming and extending their ADL (Williams & Worthy 2009; Berta et al. 2015; Davis 2014). They have webbed feet for swimming and typically have a high buoyancy, which is related to the volume of air trapped in their fur. For example, the pelage air layer of muskrat is 21% of body volume, which reduces its density to 0.79. Although such a high buoyancy is good for floating at the surface and contributes greatly to pelt insulation (see 4.1.1), it increases the energetic cost of diving for these semi-aquatic mammals (Williams et al. 1992; Fish 2000; Fish et al. 2002), and they do reach considerable depths; sea otters dive up to 100 m (Berta et al. 2015).

Sirenians (dugong, Dugong dugon and manatees, Trichechus spp.) are permanently aquatic and forage on sea grasses in shallow water. They have high-density bone that facilitates neutral buoyancy while foraging in shallow water. Nevertheless, sirenians have been reported to dive for 6 min to 600 m for manatee and 8 min to 400 m for dugong (Berta et al. 2015). However, it is the deep-diving marine mammals, such as many cetaceans and seals, that rely in extremis on many levels of adaptation to achieve their remarkable breath-hold time and capacity for deep diving. For example, phocid seals dive up to 1,530 m and for up to 120 min; otariid seals, 482 m and 18 min; mysticete cetaceans, 500 m and 80 min; and odontocete cetaceans, 3,000 m and 138 min (Berta et al. 2015). Cuvier’s beaked whale dives up to 2,992 m depth and for up to 137.5 min; deep dives on average were to 1,401 m depth and 67.4 min duration (Schorr et al. 2014).

An obvious morphological adaptation of diving mammals is their streamlined profile, with limbs modified for swimming (flippers of seals, flukes of cetaceans). Streamlining provides a marked energetic advantage for swimming (Davis 2014). A fineness ratio (body length/diameter) of about 4.5 is optimal for drag reduction; pressure drag increases at lower ratios and frictional drag increases at higher ratios (Vogel 1994). The high aspect ratio (span2/area) of flippers and flukes aids thrust production and swimming efficiency (Fish 1998). Surface swimming increases the energy cost of swimming by up to five times because of the formation of a swimming wave, but swimming three body diameters below the surface eliminates the wave energy dissipation (Fish 2000). Not surprisingly, most marine mammals minimize their time swimming at the surface, but they do have to occasionally breathe. Deep-diving marine mammals have little hair, so pelage buoyancy is not an issue; rather, they have blubber (20–30% of body mass), which provides some buoyancy as well as thermal insulation, enhances streamlining, provides energy reserves, and has lower maintenance costs (Fish 2000; see also 4.1.1). Changes in dynamics of buoyancy of these mammals with diving are related more to lung volume and lung compression at depth. For them, changes in swimming behaviour seem to be the main compensation for changes in buoyancy during diving. (p.338) Hydrodynamic lift from the flippers or flukes can counterbalance the reduced buoyancy at depth due to compression of the air spaces. Deep-diving mammals typically exhale upon diving to facilitate negative buoyancy and to limit exposure to high partial pressures of gases while submerged. Their negative buoyancy at depth conserves O2 stores and facilitates extended feeding durations. They can also exploit intermittent swimming to reduce locomotory costs by gliding during decent (Skrovan et al. 1999; Fish 2000).

## 4.5.1 Diving Response

The obvious problem for diving mammals of extending breath-hold time (hence foraging duration) meant that the physiology of breath-holding (apnoea), or the diving response, was one of the first adaptations investigated for diving. The concept of the diving response was based on early observations by Bert and Richet (see Castellini 2012 and Davis & Wiliams 2012 for historical reviews). The physiology of the diving response was examined in the 1930s by Scholander and Irving using simple (if somewhat crude) laboratory experiments involving forced submersion of various mammals and birds, to examine the hypothesis that during diving there were circulatory adjustments that confined O2 delivery to critical tissues such as heart and brain. These early experiments clearly established the cardiovascular sequelae of diving apnoea. There is a rapid decrease in heart rate (bradycardia) upon cessation of breathing (Figure 4.15A) that is associated with reduced distribution of blood flow to many organs (e.g. muscle, gut, liver, kidney, skin) but not the more essential aerobic organs (e.g. heart, brain, lungs, adrenals). The decrease in heart rate combined with the peripheral vasoconstriction maintains central blood pressure (Figure 4.15B). In fact, this cardiovascular adjustment is a general and primitive response of mammals (even non-diving species) and other vertebrates to hypoxic challenge (Panneton 2013). Humans also have a pronounced diving response to breath-holding and water contacting the face (Gooden 1994). Scholander (1963) described this dramatic cardiovascular response to apnoea and hypoxia as the ‘master switch of life’.

Figure 4.15 Diving response for a common harbour seal (Phoca vitulina) during a forced dive. A: Heart rate pre-dive, at various stages during a 15-min dive (numbers indicates minutes into the dive), and post-dive (‘15 RECOVERY’ indicates artefact from struggling, not heart rate). B: Central arterial (femoral) and peripheral (toe), and peripheral venous (toe) blood pressure during diving; vertical line indicates the end of the dive for each pressure trace.

From Irving et al. (1942). Reproduced with permission of the American Physiological Society.

Although these early investigators of the diving response appreciated that there may be physiological differences between forced and voluntary diving, it was not until the 1970s that technology became available to study the physiology of mammals and birds diving voluntarily. Kooyman and Campbell (1972) showed that the diving response was generally less dramatic for voluntary dives. This and other pioneering studies of Kooyman further showed that seals would voluntarily make both shallow and short dives, and deep and long dives, and that there were differences, such as more pronounced bradycardia for the latter (even at the initiation of diving). More recent studies of free-diving cetaceans indicate that the bradycardia response to diving is quite flexible, being modulated by behaviour and the nature of their activity. For example, free-swimming bottlenose dolphins (Tursiops (p.339) truncatus) have a lower but variable heart rate while underwater compared to when resting at the surface (105 beats min−1, compared to their maximum heart rate of 128 min−1), submerged resting (40 min−1), horizontal swimming (37 min−1), and head bobbing (56 min−1; Noren et al. 2012). Thus, Scholander’s concept of a ‘master switch’ was in reality more of a graded cardiovascular response (‘dimmer switch’) that anticipated the extent of the dive.

The next conceptual leap in diving physiology was addressing the question of whether diving metabolism was primarily aerobic or anaerobic (Davis 2014; see 2.3). Diving mammals have cardiac and microvascular adaptations that conserve essential nutrients (especially oxygen), and isolate lactate that is produced in active muscle while diving underwater (Irving et al. 1942; Zapol et al. 1979). In a forced dive, the extreme diving bradycardia is associated with a major redistribution of blood flow away from many peripheral tissues (gut, kidneys, muscle) that helps to conserve oxygen for hypoxia-sensitive tissues such as the heart and brain. (p.340) The hypoxic tissues will rely more on anaerobic metabolism during the dive, and accumulate sometimes very high levels of lactate.

The aerobic organs of diving mammals also have their own particular adaptations (Davis 2014). The heart has a high myoglobin concentration and lactate dehydrogenase (LDH) activity, and a high glycogen content, and the kidneys can have a high mitochondrial volume density. The splanchnic organs (liver, stomach, intestine) have high mitochondrial volume density and LDH levels. If the ADL is not exceeded, then kidney and splanchnic blood flow are similar to resting levels, and function is sustained. If the ADL is exceeded, blood flow and organ function is dramatically reduced, but is rapidly re-established after O2 availability is restored.

Diving-induced ischaemia/reperfusion produces reactive oxygen species (ROS), and ROS production is higher for long- than short-duration divers (Zenteno-Savín et al. 2012). Oxidative damage is not limited by restricting ROS production, but is limited by the synergistic effects of higher tissue activities of antioxidant enzymes and non-enzymatic compared to tissues of non-diving mammals. High levels of antioxidant enzymes have been reported in tissues of marine mammals; for example, superoxide dismutase (SOD), glutathione peroxidase (GPx), glutathione reductase (GR), and glucose-6-phosphate dehydrogenase (G6PDH). High levels of a non-enzymatic antioxidant, γ‎-L-glutamyl-L-cysteinyl glycine (GSH), have been reported for various terrestrial animals that routinely experience hypoxia, including hibernation and aestivation, environmental hypoxia and diving (Hermes-Lima & Zenteno-Savín 2002; Zenteno-Savín et al. 2012). The (p.341) antioxidant systems of diving mammals and birds are effective in that their tissues do not have higher oxidative damage than non-divers, and GSH seems to have an important role in both, but the details of antioxidant patterns are different between seals and penguins.

## 4.5.2 O2 Stores and Aerobic Dive Limit

Diving mammals are isolated from an air supply while underwater, hence any use of O2 for energy generation while diving must rely on O2 stored within the body. There are three main potential O2 storage sites during diving: lung O2 stores, blood O2 stores, and muscle myoglobin O2 stores (Table 4.4).

Table 4.4 Lung, blood, and muscle oxygen store parameters for various diving mammals, and cattle as a terrestrial comparison. DLV, diving lung volume; BV, blood volume; Hb, haemoglobin concentration; Mb, myoglobin concentration. Data from Ponganis (2011), Gerlinsky et al. (2013) and Davis (2014).

O2 Store

Lung Stores

Blood Stores

Muscle Stores

Species

LV (ml kg–1)

BV ml (kg–1)

Hb (g dl–1)

Mb (g kg–1)

Muscle (g kg–1)

Cattle

66

71

12

0.4

330

Bottlenose dolphin

90

71

14

3.3

Manatee

37

80

15

0.4

350

Sea otter

207

91

17

2.6

30

Steller sea lion

55

97

120

2.7

Hooded seal

40

106

23

9.5

270

Walrus

58

106

16

3.0

300

Pacific white-sided dolphin

108

17

3.5

Northern fur seal

72.5

109

17

3.5

300

California sea lion

58

120

18

5.4

370

Beluga whale

57.8

128

21

3.4

300

Harbour seal

39

132

21

5.5

300

Dall porpoise

143

20

Harp seal

168

23

8.6

250

Baikal seal

177

27

6.0

300

Australian sea lion

178

19

2.7

Sperm whale

28

200

22

5.4

340

Weddell seal

27

210

26

5.4

350

Elephant seal

20

216

25

6.5

280

(p.342) Lung O2 stores depend on the volume of air in the lungs (where gas exchange occurs) on commencement of diving. For near-surface swimmers, the inspiratory volume is assumed to be available as an O2 store for manatees, 60% of lung volume for sea otters and 50% for pinnipeds, compared to 100% of lung volume for cetaceans (Ponganis et al. 2011). Further, an extraction of 15% lung O2 is assumed, to calculate the effective lung O2 store. The relatively large lung size of delphinid and phocoenid dolphins, which are rapid breathing, short-duration, shallow divers, may enable their lung to function as a site of respiratory gas exchange throughout a dive (Piscitelli et al. 2010). For blood O2 stores, essentially all the O2 is bound to haemoglobin; an insignificant amount is dissolved in the plasma. Consequently, the O2 store is proportional to the blood haemoglobin concentration, saturation levels of arterial/venous blood, and the total blood volume.

Not surprisingly, marine mammals typically have much higher blood haemoglobin concentrations than terrestrial mammals (see 3.4.1); there is no scope for adaptive variation in how much O2 can be bound to haemoglobin, as all vertebrate haemoglobins typically bind 1.34 ml O2 per gram (Withers 1992). It is usually assumed that one-third of the blood is arterial and two-thirds is venous, initial arterial blood is 95% saturated and end arterial saturation is 20%, initial venous content is 5 ml dl−1 lower than 95% arterial blood saturation, and end venous blood is 0% saturated (Ponganis et al. 2011). Muscle O2 stores rely on intracellular myoglobin, which is essentially a single subunit of tetrameric haemoglobin, and typically binds 1.34 ml O2 per gram (like haemoglobin). Myoglobin stores are assumed to be 100% saturated prior to diving, and as low as 0% saturated at the end of a dive. Not surprisingly, marine mammals typically have much higher muscle myoglobin concentrations than terrestrial mammals (see later).

The suite of physiological adjustments of diving mammals that increase total body O2 stores, and morphological and locomotory adaptations that maximize the energetic efficiency of diving (e.g. streamlined shape, prolonged gliding), extend their ADL (Kooyman et al. 1980; Williams et al. 2000; Ponganis et al. 2011). Although the ADL can be measured experimentally by measuring blood lactate levels, it is most often determined as the ratio of calculated O2 stores to diving metabolic rate. Most marine mammals are thought to dive within their ADL, but the specialized deep and long-diving beaked whales (e.g. Ziphius cavirostris) and high-speed sprinting fin whales (e.g. Globiocephala macrorhynchus) might challenge this point of view. Both groups of whales have high myoglobin, as expected, and muscle fibre specializations that extend their ADL (Velten et al. 2013). Beaked whales have mainly fast-twitch fibres that decrease the metabolic cost of diving, and fin whales have some rare slow-twitch oxidative fibres to support high swimming activity. Calculated ADLs suggest that both groups of whales have sufficient O2 stores to sustain aerobic dive limits.

Muscle myoglobin concentration ([Mb]) has increased independently in a variety of mammalian divers. Mirceta et al. (2013) reconstructed the evolutionary (p.343) history of [Mb], using the net surface charge density of myoglobin as a proxy for maximal [Mb]. Their findings not only show the pattern of independent evolution of high Mb levels in various extant mammalian diving lineages (Figure 4.16), but imply an aquatic ancestry of related but now non-aquatic mammals, such as echidnas (cf. platypus), talpid moles (cf. star-nosed moles), and hyrax and elephants (cf. sirenians).

Figure 4.16 Evolutionary reconstruction of myoglobin net surface charge (a proxy for myoglobin concentration) for terrestrial and aquatic mammals. The myoglobin net surface charge increases in all lineages of mammalian divers with an extended aquatic component, and is used to infer the diving capacity of extinct species representing stages during mammalian land-to-water transitions.

From Mirceta et al. (2013). Reproduced with permission from Science (AAAS).

For shallow semi-aquatic divers (and diving birds), and deep divers prior to lung collapse, the O2 in the lung air is a substantial store (Figure 4.17). However, the lung O2 store of deep divers is compromised at depth because of the hyperbaric compression of particularly their lung (Ponganis 2011; see 4.5.3), so they typically have a very low lung O2 store. For most divers, blood O2 is the main store. Muscle O2 stores can range from being a relatively small store to the main store, depending on the muscle myoglobin content of particular species.

Figure 4.17 Lung, blood, and muscle O2 stores in non-diving mammals (human, cow), shallow-diving marine mammals (manatee, sea otter), various seals and cetaceans, and some diving birds (italics) for comparison. Deep-diving marine mammals (in bold, far right) typically have low lung O2 stores, especially compared to diving birds and the surface-diving sea otter.

Redrawn using data from Ponganis (2011), Ponganis et al. (2011), Davis (2014), Stephenson et al. (1989), and Croll et al. (1992)

## 4.5.3 Hyperbaria: Diving under Pressure

Three physiological syndromes are experienced by human divers when breathing air at considerable depth, hence pressure: decompression sickness, N2 narcosis, and O2 toxicity. These are all consequences of the fact that pressure increases (p.344) dramatically with depth underwater, by about 1 atmosphere pressure (101.3 kPa) per 10 m depth (Figure 4.18). For an air space underwater (e.g. the respiratory system), the increase in pressure collapses the volume of air (unless there is sufficient physical strength of the vessel wall to resist collapse; e.g. the hull of a submarine), and the partial pressures of all gases are increased in proportion to the pressure. For example, at 10 m depth, the pressure is 202.6 kPa (2 atm), and for normal air the pO2 is 42.4 kPa (compared to 21.2 at 1 atm) and the pN2 is 160.2 kPa (compared to 80.1).

Figure 4.18 Effect of water depth on barometric pressure, and use of various gases in diving to counteract decompression sickness, nitrogen narcosis, and high pressure nervous syndrome (HPNS).

Modified from Pocock et al. (2013). Reproduced with permission of Oxford University Press.

Breathing high partial pressures of gases increases the amount of those gases dissolved in the body fluid and fat (Henry’s law; see 2.5.4). Over an extended time at even moderate depth (e.g. 10 m), more and more O2 and N2 dissolve into the body fluids and fat. When the diver returns to the surface, these gases become supersaturated and can bubble out of solution; such bubbles can cause pain, tissue damage, and even death; this is decompression sickness (the ‘bends’, or caisson disease; Ponganis 2011; Pocock et al. 2013). Consequently, divers must ascend slowly, adhering to strict decompression schedule tables. Otherwise, they must be re-compressed at the surface (or by returning underwater) for long enough that gases in the body fluids and fat are ‘blown off’.

Free-diving (without compressed air) occurs to relatively limited depth, for short periods, so insufficient gases dissolve into the body fluids and fat to cause (p.345) decompression sickness, although continuous intense free-diving (e.g. pearl and sponge divers) can apparently rarely experience mild decompression sickness (Wong 2000). High pO2 has direct physiological effects, resulting in O2 toxicity, or high pressure nervous syndrome (HPNS; Ponganis 2011; Pocock et al. 2013). Even breathing pure O2 at 1 atm pressure has physiological sequelae (pulmonary congestion, impaired mental activity), and above 2 atm leads to O2 toxicity (nausea, dizziness, feeling of intoxication, tremors, convulsions). A rapid rate of increase in pressure exacerbates the extent of HPNS; humans typically experience HPNS at about 190 m depth. High pN2 also has physiological effects at depths below (p.346) about 50 m when breathing normal air, resulting in nitrogen narcosis. Symptoms are mental confusion and poor motor coordination, accompanied by a sense of euphoria (‘raptures of the deep’).

These various syndromes can be alleviated by breathing special gas mixtures (Figure 4.18). Breathing helox (or heliox; about 21% O2 and 79% He) largely overcomes the problems of nitrogen narcosis, and its lower solubility than N2 reduces the problems of decompression sickness. However, the high thermal conductivity of helium can lead to thermal balance problems, and its lower density makes vocal communication difficult (the voice becomes very high pitched). Trimix, a mixture of O2 (0.8%), N2 (9.2%), and He (90%), uses the narcotic effect of N2 to reduce O2 toxicity. Diving mammals might be equivalent to free-diving humans, but their extended dive times and considerably greater depths and pressures potentially make them as, or even more, susceptible to these syndromes (Ponganis 2011). Many cetaceans and pinnipeds exceed the 190 m HPNS onset depth experienced by humans, and should potentially experience nitrogen narcosis and decompression sickness, so they have particular adaptations to avoid these problems.

First, diving to considerable depth exposes the mammal to physical barotrauma, resulting from compression of air-filled spaces (Ponganis 2011). In some diving mammals, the aortic arch is relatively compliant whereas the aorta is relatively incompliant (Lillie et al. 2013). The compliant arch section acts as a windkessel, maintaining arterial blood pressure during the extended bradycardia. Extreme differences in stiffness of the arch and the aorta in species such as fin whales may be an adaptation to avoid haemodynamic consequences of rapid depth-related changes in pressure across the aortic wall, between intravascular blood and the thoracic cavity. Remarkable collapsibility of the chest wall and lung of various seals can provide nearly unlimited compensation for high pressures. Diving sea lions experience lung collapse at about 225 m, although the depth for collapse depends on the maximum dive depth; this suggests that sea lions inhale more deeply for anticipated deeper dives (McDonald & Ponganis 2012). Scholander (1940) noted that the excised lung of fin whales was able to collapse nearly completely with hyperbaria, with little remaining residual volume, and speculated that lung collapse avoided gas exchange at depth, hence avoided decompression sickness (Kooyman 2015).

Lung collapse not only reduces the uptake of N2 at depth (and therefore reduces the risk of decompression sickness and nitrogen narcosis) but also preserves the lung O2 store for use on ascent. Beaked whales, which dive deeply more regularly than other cetaceans, might be more susceptible to decompression sickness, and have additional adaptations that reduce decompression bubble nucleation such as specialized endothelial structures or elevated levels of nitric oxide, together with behavioural management of acent rates and recompression with subsequent dives (Hooker et al. 2009). Indeed, the adaptations of deep-diving marine mammals might be more to manage the N2 load during ascent rather than to minimize N2 loading during diving (Hooker et al. 2011). It has been suggested that human (p.347) activities (e.g. sonar) might increase the risk of decompression sickness and stranding, by modifying acent behaviour (e.g. causing whales to acend more rapidly) or sonar directly interacting with tissues supersaturated with nitrogen (e.g. Weilgart 2007) although this idea remains controversial (Kvadsheim et al. 2012).

## 4.5.4 Vision and Echolocation

Light does not penetrate water, even clear water, very well. For turbid water, light penetration can be almost zero. Consequently, much of the aquatic environment is dark even during the daytime, so vision becomes difficult. Nevertheless, vision remains an important sense for marine mammals, and most have relatively large eyes, although walrus and sirenians have relatively small eyes (Berta et al. 2015). Marine mammals generally have a high visual acuity, facilitated by their nearly circular lens, thick retina with a high proportion of rod photoreceptors, and a well-developed tapetum lucidum to reflect light from the back of the retina back to the photoreceptors. Not surprisingly, sea otters and polar bears have more terrestrial-type eyes, relying more on aerial vision and other senses (e.g. chemoreception).

Marine mammals (dolphins, seals, manatee) have a higher carbohydrate concentration in their tears than terrestrial mammals (humans, camelids; Davis & Argüeso 2014), presumably to help maintain the hydration of the cornea in seawater (Tarpley & Ridgway 1991). The tear film of dolphins (Tarpley & Ridgway 1991) and seals (Davis et al. 2013) lacks lipids, unlike typical terrestrial mammals, which have a reduced ocular EWL because of the lipid component of their tear film (see 3.6.2). Cetaceans lack a nictitating membrane, so a lubricating role of the tear film is less important than for mammals that have a nictitating membrane (e.g. seals, polar bear).

Deep-diving marine mammals must locate their prey in a dark environment, unless they have echolocation (see later), although the Guiana dolphin (Sotalia guianensis) has cutaneous electroreceptors for feeding underwater (Czech-Damal et al. 2012b; see 3.7.2.6), and the platypus (Ornithorhynchus anatinus) has electroreceptors on its bill for feeding underwater (with its eyes closed; Grant 1989). The vision of many marine mammals is adapted to low intensity light with a peak sensitivity at 485 nm, which matches the wavelength of bioluminescence produced by a large range of marine organisms, including myctophid fish (Vacquié-Garcia et al. 2012). Bioluminescence thus likely is a key to predator-prey interactions in deep, dark marine environments.

Hearing is an important sense for marine mammals (Berta et al. 2015). Polar bears and sea otters vocalize and hear much like terrestrial mammals, but pinnipeds have a number of specializations for underwater and terrestrial hearing. Many pinnipeds vocalize considerably on land, for various social interactions, including mother-pup identification calls that are important for re-establishing contact when (p.348) the mother returns from feeding (other senses such as olfaction, vision, and spatial awareness are also important). Various species also have underwater vocalizations for general social interactions. Male walrus emit strange ‘knocking’ sounds during the breeding season; these ‘songs’ seem to reinforce dominance status. Some pinnipeds (e.g. harbour seal, Phoca vitulina; ringed seal, Pusa hispida; harp seal, Pagophilus groenlandicus; grey seal, Halichoerus grypus; hooded seal, Cystophora cristata) emit underwater clicks, but it is unclear whether these are used for echolocation. The ear of pinnipeds is generally similar to that of terrestrial mammals, although some modifications amplify auditory signals. The cavernous venous sinuses in tissues of the middle and outer ear may enhance sound transmission to the inner ear and also provide volume compensation for the hyperbaric effect of deep diving (Nummela 2008).

Cetaceans have a sophisticated, high-frequency echolocation system (sonar) for orientating and capturing prey underwater (Berta et al. 2015). Sound travels about five times faster in water (about 1,450–1,559 m s−1) than air (about 340 m s−1), and acoustic signals are useful, especially in turbid water or darkness, for feeding, sensing environmental features, and often remarkably sophisticated communication. Cetaceans emit both high- and low-frequency acoustic signals, with different roles. The source of the acoustic signals for small odontocetes is the complex nasal sac system, located just inside the blow hole. This is coupled to the role of the melon, a low-density, lipid-filled structure that sits above the upper jaw. The melon functions as an acoustic lens that creates a focused and highly directional acoustic beam emitted in front of the animal.

Toothed whales are a morphologically and ecologically diverse group of predators; sperm whales (Physeter macrocephalus) are deep-sea squid predators, dusky dolphins (Lagenorhynchus obscurus) prey on oceanic fish schools, and shallow water river dolphins (Platanistoidea) feed on individual prawns and fish (Jensen et al. 2013). Their acoustic calls are equally diverse, of four general types. The echolocation clicks of most odontocetes are short and relatively broadband, about , with a range of about 100 m (Berta et al. 2015). Sperm whale echolocations are highly directional, low frequency , and high intensity (235 dB). Killer whales (Orcinus orca) produce low- , mid- (10 kHz) and high- frequency signals, for object detection (low frequencies) and prey discrimination and compass bearing direction (high frequencies). They also have group ‘dialects’ for pod recognition. Some odontocetes use high-intensity sounds to stun their prey. Beaked whales have low-intensity, frequency-modulated clicks (about 45 kHz). Whistling dolphins have higher frequency , short, loud (220 dB) broadband clicks. Dolphins also produce a more harmonic and narrow frequency band whistle (or squeal). These are used for individual or group recognition. A range of porpoises, pygmy sperm whales, and other unrelated delphinids have narrowband clicks (about 130 kHz). River dolphins use a high repetition of relatively low frequency and low intensity (180–195 dB) (p.349) clicks compared to similar-size marine delphinids, perhaps reflecting their lesser need for long-distance detection.

The inverse scaling of echolocation frequency with body mass seems to be a major driver of echolocation call diversity in cetaceans (compared to habitat and diet variation, which drives much of the acoustic diversity in bats; see 4.7.4). Standard allometry of echolocation frequency and body mass for whales is strong $(R2=0.86−0.93)$, suggesting an allometric constraint on frequency, but there is a strong phylogenetic component to this pattern, as it is considerably weaker $(R2=0.27)$ after phylogenetic correction (May-Collardo et al. 2007). Mysticetes are again less understood than odontocetes with respect to their uses of acoustic signals. Many emit broadband clicks that may be used for echolocation and for communication; for instance, the complex, low-frequency ‘songs’ of humpback whales (Megaptera novaeangliae) are used for long-distance communication, over hundreds or even thousands of kilometers.

The cetacean auditory system is considerably modified for underwater hearing (Berta et al. 2015). The external auditory canal is very narrow, and in mysticetes is plugged and may not be functional. For odontocetes, echoes of sound reflected from objects in the environment are detected primarily by the dense bone of the lower jaw (‘jaw hearing’; Norris 1964) and fat bodies that channel sound to the middle ear. The two bony parts of the ear, the tympanic and periotic bones, are very dense bone; they differ markedly from the typical terrestrial mammal ear. Inner ear cochlear structure appears modified for high-frequency hearing (e.g. the basilar membrane is relatively narrow and thick). The acoustic system of mysticete cetaceans is less understood but likely to be generally similar to odontocetes. They lack vocal cords but have a U-fold in the larynx that seems homologous in function to vocal cords, and distribution of sound into the environment seems to involve cranial sinuses (Berta et al. 2015).

Less is known of vocalization and auditory reception by sirenians (Berta et al. 2015). They emit ‘chirp-squeaks’ presumably from vocal folds in the larynx, and may be able to use their nasal cavity and fat pads to direct these sounds (like their proboscidean relatives; Landrau-Giovanetti et al. 2014). The chirp-squeaks are short, frequency-modulated sounds used for communication (e.g. female-young bonding, individual identity, aggression, and perhaps the size of the signaller). Dugongs and manatees lack external ears and have a much-reduced auditory meatus and ear canal leading to the highly ossified tympano-periotic complex (like cetaceans).

# 4.6 Extreme Terrestrial Locomotion

Terrestrial locomotion is very diverse in mammals, reflecting both size and ecological niche, from slow ambulatory mammals, such as echidnas, pangolins (Manis (p.350) spp.), aardvarks (Orycteropus afer), and elephants, to super-fast predators (e.g. cheetah, wolves), to their nearly-as-fast prey (e.g. various antelopes and pronghorn). High speed cursoriality is the extreme, most metabolically demanding, example of terrestrial locomotion, involving many anatomical and physiological adaptations. Brachiation (‘arm swinging’) through trees, similarly requires anatomical and physiological specializations. Migrating, often for considerable distances, is a different kind of extreme terrestrial locomotion. Each of these aspects of extreme terrestrial locomotion is examined later. Other extreme forms of mammalian locomotion include burrowing underground, which can be extremely metabolically costly (see 4.3.3); flight, which is costly in terms of the immediate metabolic cost but relatively economical in terms of cost of transport (see 2.8.2, 4.6.1); and diving, which is not so metabolically costly but entails many other physiological challenges (see 2.8.3, 4.5).

## 4.6.1 Cursorial Locomotion

The running speed of terrestrial mammals depends largely on body mass; cursorial mammals are those that run fast (typically artiodactyls, perissodactyls, and carnivores), with associated morphological adaptations such as long distal limb segments, shortened proximal segments, and change in foot stance from plantigrade to digitigrade or unguligrade (Garland & Janis 1993). Speed relative to body length (BL) ranges from < 5 BL s−1 in elephants (Loxodonta africana, Elephas maximus), hippopotamus (Hippopotamus amphibious), rhinoceros (Ceratotherium simum, Diceros bicornis), camels (Camelus dromedarius), kouprey (Bos sauveli), giraffe (Giraffa camelopardalis), and polar bear to > 40 BL s−1 for small marsupials, rabbits and hares, and various rodents (Iriarte-Díaz 2002). However, the allometric scaling exponent varies with mass; for small mammals , , for intermediate mass (10 kg < M < 100 kg), , and for large mammals (> 100 kg), . These differing scaling exponents seem to reflect differences in the allometry of bone-bending stress scope ; larger mammals have less scope for bending stress (a lower safety factor).

In terms of absolute speed, many small marsupials and rodents are slow (< 5 m s−1), whereas the cheetah (Acinonyx jubatus) is the fastest (30 m s−1), closely followed by the pronghorn (Antilocapra americana; 28 m s−1) and numerous African artiodactyls (25–28 m s−1). Lovegrove (2004) showed that maximum running speed (MRS) of mammals differed systematically with locomotor limb anatomy; speeds increased with mass for plantigrade (m s−1 α‎ M0.124), digitigrade (m s−1 α‎ M0.194), and lagomorph (hopping; m s−1 α‎ M0.319) mammals, but decreased with mass for unguligrade mammals (; Figure 4.19). Lovegrove and Mowoe (2014) described a micro-cursorial digitigrade locomotor mode for elephant shrews (Elephantulus) based on their very high metatarsal:femur ratio (of 1.07) that is typical of unguligrade cursors; their MRS is at the high end of the range (p.351) for unguligrade mammals, relative to their body mass. The low speeds of plantigrade mammals may constrain the evolution of large body mass because of risks and costs of predation, unless they evolve protective armour (Lovegrove 2001).

Figure 4.19 Relationship between maximum running speed and body mass, showing the different allometries for five anatomical locomotor types: plantigrade, micro-cursorial (Elephantulus), digitigrade, lagomorph hopping, and unguligrade.

Modified from Iriarte-Díaz (2002), Lovegrove (2004), and Lovegrove and Mowoe (2014).

Whether fast cursorial mammals achieve their extreme performance by having an exceptional aerobic capacity and high locomotor muscle mass, or anatomical adaptations that makes running more efficient, differs for different cursors. Pronghorn antelope, which have a maximal running speed of about 28 m s−1 (100 km h−1) and exceptional endurance (e.g. running 11 km in 10 min), have an exceptionally high aerobic capacity (Lindstedt et al. 1991). Pronghorns running at about 10 m s−1 up an 11% incline had maximal O2 uptake of about 18 ml O2 g−1 h−1, which is over three times the maximum predicted for a similar mass mammal (and about that for a 10-g mouse compared to 32 kg for the pronghorn).

In contrast to pronghorn, running cheetah (the fastest land mammal) have a lower metabolic rate at 5 m s−1 of about 3.2 ml O2 g−1 h−1, and the expected net cost of transport of 0.14 ml O2 g−1 km−1 for a running mammal (Taylor et al. 1974). These values are similar to those of one of their prey (Gazella gazella) of 3.4 ml O2 g−1 h−1 and 0.16 ml O2 g−1 km−1. The remarkable running speed of cheetah is due in part to their considerable spinal flexion that increases stride length (Figure 4.20); this probably contributes about 1.3 m s−1 at 27 m s−1 (Hildebrand 1959). Wilson et al. (2013a,b) have recently demonstrated elegant variation of the running strategy of cheetah during a hunt, using miniature data loggers that (p.352) (p.353) recorded speed and acceleration (see 5.4). Top speed was recorded as 25.9 m s−1 but most hunts were at more moderate speeds (Wilson et al. 2013a). Cheetah accelerated at up to 7.5 m s−1 (cf. gravitational acceleration is 9.8 m s−1) and reached hunting speeds up to 18.94 m s−1 at the start of a hunt, but reduced their speed about 5–8 s before the end, to make rapid turns in response to their prey’s evasion tactics (Wilson et al. 2013b).

Figure 4.20 Top: Running cheetah with schematics showing range of spinal flexion. Adapted from Hildebrand (1959) and McDonald (2010). Bottom: Body temperature of free-ranging cheetah for a successful hunt, showing the initial small hyperthermia from metabolic heat accumulation when running, followed by a stress hyperthermia, compared to the minimal metabolic hyperthermia of an unsuccessful hunt.

Modified from Hetem et al. (2013).

Lions (Panthera leo), in contrast to cheetah, have a strong social structure and often do not hunt alone. They have a high net cost of transport (0.36 ml O2 g−1 km−1) compared to predicted (0.11 ml O2 g−1 km−1), but why is not clear—it is not associated with a peculiar limb morphology (Chassin et al. 1976). Puma (Felis concolor) have an entirely different hunting strategy in very different environments from cheetah and lion. They are low-energy-cost, sit-and-wait cryptic predators in diverse, rugged habitats that precisely match their pouncing force to prey size (Williams et al. 2014). Nevertheless, their energetic cost of free-ranging hunting is, like that of other large mammalian carnivores, about 2.3 times higher than that predicted by laboratory studies for routine cost of transport based on mass, and 3.8 times higher than the minimum cost of transport. Hunting is clearly an energetically expensive feeding strategy.

The high metabolic cost of running fast can create a thermal problem from accumulation of metabolic heat. It is impossible for a fast runner to dissipate all its metabolic heat production, so there is a continual storage of heat in the body, causing Tb to rise. For example, cheetahs running at 28 m s−1 produce more than 60 times as much heat as at rest, and have to store most of this heat by an increase in Tb. Taylor and Rowntree (1973) determined that cheetah running on a treadmill at 5 m s−1 stored 90% of their metabolic heat; since cheetah refuse to run if Tb exceeded 40.5°C, they calculated that cheetah could only run for about 1 km, which is about their pursuit distance in nature. Consequently, they speculated that cheetah’s sprinting might be limited by their capacity to store heat. However, Hetem et al. (2013) found that free-hunting cheetah had a Tb of only about 38.4°C after hunting, so heat storage did not compromise hunting. This suggests that perhaps sprinting is limited by the cheetah’s aerobic capacity, and associated anaeroblic limits and lactate accumulation. There is actually a greater hyperthermia for cheetah after successful hunts (Tb increases by 1.3°C) than unsuccessful hunts (Tb increases by 0.5°C; Figure 4.20), likely reflecting a stress response to the danger of being attacked by other large predators (e.g. leopard, lion) prior to their commencing to feed on the carcass. Amazingly, Hetem et al. (2013) found that a lame male cheetah, which joined his female sibling after her successful hunt but did not participate in the hunt, showed the same stress hyperthermia as the hunter! Cheetahs, like many other predators, experience kleptoparasitism, the theft of their prey by other animals. About 25% theft of prey requires cheetahs to hunt for about an extra 1.1 h per day, increasing their daily energy expenditure by about 12% (Scantlebury et al. 2014). The stress of kleptoparasitism and predation by (p.354) other predators has a major physiological impact on cheetah, and is presumably a major evolutionary cost.

Kangaroos and wallabies are large bipedal hoppers, and have a remarkable uncoupling of the metabolic cost of transport and speed, reflecting elastic storage of energy in tendons and tensed muscles during hopping. The classic study of Dawson and Taylor (1973) showed that the metabolic rate of red kangaroos (Macropus rufa) increased linearly with speed when locomoting pentapedally (using forelimb, hind limbs, and tail) at low speed, but actually decreased at higher speeds after the transition (at about 1.8–1.9 m s−1) to bipedal hopping. Stride frequency showed essentially the same pattern with speed, but stride length increased linearly with speed; hopping faster is accomplished by jumping further per hop.

This remarkable conservation in the metabolic cost of locomotion over varying speeds reflects the elastic storage of energy in tendons, ligaments, and active muscle on footfall, and its release on take-off to decrease the metabolic cost of hopping. The tammar wallaby (Macropus eugenii) shows the same remarkable pattern for metabolic rate and hopping speed (Baudinette et al. 1992). The smaller potoroo lineage probably diverged from macropods about 20–30 MYBP, but it is unclear whether their pattern of hopping energetics is the same as kangaroos. The metabolic rate of the long-nosed potoroo (Potorous tridactylus) increases linearly with speed, but it is primarily quadrupedal, only using bipedal hopping intermittently and for short bursts at high speed (Baudinette et al. 1993). The rat-kangaroo lineage of marsupials is even older than the kangaroos and wallabies (diverged about 26 MYBP), and the rat-kangaroo Bettongia penicillata has a linear increase in metabolic rate with hopping speed, albeit at a lower-than-predicted rate of increase (Webster & Dawson 2003). Other hopping mammals, weighing less than 5 kg, have the typical linear relationship between metabolic rate and running speed (Thompson et al. 1980). So, the remarkable independence of metabolic rate on speed currently appears unique to the larger macropods. However, the use of elastic storage in locomotion is not unique to macropods, and is relatively common in many terrestrial runners (e.g. Biewener 1998), but is insufficient to have such an impact of the metabolic cost of transport as is seen for macropods.

## 4.6.2 Brachiation and Climbing

Brachiation, or ‘arm swinging’, is a form of arboreal locomotion whereby the body is supported by the arms, as for primates such as gibbons (Hylobatidae), siamangs (Symphalangus syndactylus), orangutans (Pongo), atelines (Cebidae), and suspensory quadrupeds (e.g. loris, Lorisinae; and sloths, Choloepus didactylus; Bradypus variegatus). It is associated with anatomical specializations such as long forelimbs, high shoulder joint mobility, and modified finger/elbow flexion musculature (Nyakatura & Andrada 2013). In gibbons, there are two brachiation gaits: (1) continuous contact brachiation is where at least one hand is always in contact with (p.355) the substrate and is low speed, and (2) ricochetal brachiation has a flight (no-hand contact) phase and is higher speed (Michilsens et al. 2012). Continuous contact brachiation is a continuum of four transition types, which presumably confers efficient determination of locomotor cost at varying speed, like the variable gait of terrestrial mammals such as horses (Hoyt & Taylor 1981) and lemurs (O’Neill 2012). A mechanical cost of brachiating by primates is the substantial dynamic forces resulting from swinging from branches. Gibbon’s hindlimbs also show adaptations to bipedalism. Their Achilles tendon has biomechanical properties suitable for storage and release of elastic energy, acting as an elastic ‘spring’ to reduce the locomotor cost of bipedalism (like humans and kangaroos), whereas the high stiffness of their patellar tendon might enhance leaping performance (Vereecke & Channon 2013).

The energetic cost of brachiating does not appear to be lower than for terrestrial locomotion. The metabolic rate of spider monkeys (Ateles sp.) when hanging motionless by the arms is about two times their resting rate, and increases linearly with speed when brachiating; their energetic cost of brachiation is higher than when walking, regardless of speed (Parsons & Taylor 1977). So, the advantage of brachiation is unlikely to be a low energetic cost of transport per se, but there is likely a reduced effective cost of locomotion by being able to move more directly by brachiating through forests rather than climbing up and down trees. In comparison, the energetic cost of climbing is the same as for terrestrial locomotion for small primates (< 0.5 kg), which presumably contributes to their ecological success in finely branching arboreal environments, and the success of early primates (Hanna & Schmitt 2011). In contrast, the cost of climbing is about two times terrestrial locomotion for larger primates.

Slow loris (Nycticebus coucang) are suspension quadrupeds that essentially walk slowly, upside down; their metabolic rate when suspended from a rope, or standing on top, is about 1.4 times resting, and increases linearly with speed (Parsons & Taylor 1977). The red slender loris (Loris tardigradus) has the same metabolic cost of transport when walking and climbing (Hanna & Schmitt 2011). Sloths brachiate slowly, using relatively extended limbs, with ‘trot-like’ sequences (Nyakatura & Andrada 2012). Their ‘slow-motion’ movements reduce the risks of breaking contact with the substrate, which is important because they are slow-moving and unlikely to be able to respond quickly to avoid potentially lethal falls—but they lose the dynamic mechanical benefits of primate pendular mechanics. Slow movement is also advantageous for remaining cryptic and avoiding predation, and is consonant with their low metabolic expenditure associated with their use of a low-energy food source (see 4.8.2).

## 4.6.3 Migration

Many mammals migrate seasonally or shorter term. Migration is the regular, two-way movement of animals, generally on an annual cycle, between feeding grounds (p.356) and breeding sites. Dispersal, in contrast, is a one-way movement, generally by juveniles away from their natal origin to a new home range. The reasons for migration vary, but include avoidance of unsuitable weather, short (or no) photoperiod, and exploitation of heterogeneous food or water resources. Costs of migration include the high energetic cost of moving, potential for exhaustion, and a substantially increased mortality from predation. Migration distance varies, with a strong effect of body mass (Figure 4.21) for both terrestrial and marine migrators, with terrestrial species having lower mass and shorter migration distances. However, small and highly mobile bats have migration distances that rival those of marine mammals weighing six orders of magnitude more.

Figure 4.21 Migration distance for terrestrial (white symbols), marine (black symbols), and flying mammals (bats; light and dark grey symbols), as a function of body mass.

Data in part from Fleming and Eby (2003) and Hein et al. (2012).

Various terrestrial mammals migrate, typically seasonally; Harris et al. (2009) list 24 large terrestrial mammal species (all ungulates) that are considered to migrate. Maximum migration distance varies from relatively short (e.g. oryx, Oryx dammah, 100 km; American bison, Bison bison, 160 km; elk, Cervus elephas, 200 km); intermediate (blue wildebeest, Connochaetes taurinus, 600 km; chiru, Pantholops hodgsoni, 600 km); and very long (Siberian roe deer, Capreolus pygargus, 1,000 km; Mongolian gazelle, Procapra gutterosa, 1,000 km; saiga, Saiga tatarica, 2,400 km; caribou/reindeer, Rangifer tarandus, 3,031 km). These migration distances are being adversely limited by human activities such as fencing of agricultural and conservation areas. Various smaller terrestrial mammals migrate over much smaller distances in absolute terms, but similar distances relative to their body length.

(p.357) Most baleen whales migrate annually from productive feeding grounds in the higher latitudes to more tropical breeding grounds. The considerable cost of swimming such long distances indicates that there must be a substantial evolutionary advantage. One of the longest mammalian migrations is that of the humpback whale (Megaptera novaeangliae) that moves up to 8,500 km annually from feeding areas off Antarctica to more favourable breeding areas along the Pacific coast of Central America (Rasmussen et al. 2007). Calves that develop in warmer water may have higher survival and grow into larger adults, providing an evolutionary advantage that counters the energy cost of such a long migration. Other possible advantages include a lower metabolic cost of thermoregulation in warmer waters and avoidance of killer whales. Recently, (Silva et al. 2013) have used satellite-linked radio transmitter tags (see 5.4) to track fin (Balaenoptera physalis) and blue whales (B. musculus) to monitor their annual migration from Greenland/Iceland (feeding grounds) to the Azores (breeding ground). Interestingly, some whales feed mid-latitude during their migration, and even suspend their migration for extended periods. Northern elephant seals (Mirounga angustirostris) are a good example of migration by pinnipeds. Satellite tracking (see 5.4) has revealed the extent of their migrations as well as ocean physiographic conditions at various depths. They are at sea for most of the year, and range over the eastern North Pacific Ocean on double annual migrations between Californian breeding grounds and distant foraging grounds (Brillinger & Stewart 1998).

Bats, being smaller but more mobile by virtue of flight than other migrating mammals, move remarkably long distances (Figure 4.21), rivalling some migratory birds. Migration by bats has evolved independently in a number of lineages, particularly for temperate species; less is known for tropical species (Moussy et al. 2013). Many bats move annually, typically between summer feeding grounds to winter hibernacula sites, but some move regionally, over shorter distances. Cryan et al. (2004) used stable H analysis of hair (see 5.6) from live and museum specimens of hoary bats (Lasiurus cinereus) to determine their latitudinal movements; the δ‎Dfur (difference in deuterium and hydrogen isotopes of fur) reflects the δ‎Dpre of the precipitation where the fur was grown, which varies latitudinally. They showed that hoary bats are long-distance migrants, moving > 2,000 km between Central America and Canada (e.g. one individual migrated from Chihuahua, Mexico, to north of the Canadian border). Tri-coloured bats (Perimyotis subflavus) also range from Central America to Canada. Fraser et al. (2012) found that these bats also show regional movements as well as north-south annual migration. Nathusius’s pipistrelle (Pipistrellus nathusii) migrates annually from north-eastern to south-western Europe; individuals fly more than 2,000 km one way. Bats have a lesser capacity than birds to store energy for their annual migration as fat. Voigt et al. (2012) found that when migrating Nathusius’s pipistrelles consume insects, they preferentially oxidize the protein and carbohydrate content for energy, but use the lipid to replenish their (meagre) body fat stores. In contrast, non-migratory bat (p.358) species oxidize the lipid as well as protein carbohydrate in their diet. Bats accumulate especially C14 and C18 saturated and unsaturated lipid as a mass-effective energy store, with lipid indices (g lipid per g lean dry mass) of 0.38–0.87 (Blem 1980).

# 4.7 Flying Mammals

Flight is divided into two categories: gliding, an unpowered and passive mode of flight where aerodynamic forces are passively exerted on a membranous structure stretched between a range of body parts (the patagium); and flapping, an active and powered flight mode, where muscular power is used to generate aerodynamic forces (Norberg 1990). Gliding is found in more than 60 species of mammals, from six families of three distantly related orders (Dermoptera, Rodentia, and Diprotodontia) and in four extinct lineages (Byrnes & Spence 2011; Jackson & Schouten 2012). In marsupials, gliding occurs in the feathertail gliders (Acrobatidae, Acrobates), gliding possums (Petauridae, Petaurus), and the greater glider (Pseudocheiridae, Petauroides). Gliding in rodents is observed in flying squirrels (family Sciuridae; e.g. Glaucomys, Iomys, Petinomys, Hylopeles, Petaurista) and in scaly tailed squirrels (Family Anomaluridae; e.g. Anomalurus, Idiurus, Zenkerella). The quintessential examples of gliding mammals are two dermopterans of the family Cynocephalidae (flying lemurs or colugos; Cynocephalus and Galeopterus). Byrnes and Spence (2011) reviewed the multiple hypotheses proposed to explain the evolution of gliding in mammals. They suggested that the selective pressure(s) that led to gliding may have been to decrease predation, increase foraging efficiency, and/or control landing forces. Using a phylogenetically based comparative analysis, they argued that gliding evolved in folivorous, frugivorous, or exudivorous groups, suggesting that it was a response to poor-quality (especially protein-deficient) diets that require extensive foraging.

Amongst vertebrates, powered flight independently evolved in three lineages: bats, birds, and the extinct pterosaurs. Bats are the only extant mammals that use powered flight (Figure 4.22). It has been suggested that Old World megachiropteran fruit bats (Pteropodidae) independently evolved flight from microchiropteran insectivorous bats (Pettigrew et al. 1989), but the current consensus is that bats are a monophyletic group, so that powered flight evolved only once in mammals (Teeling et al. 2005; Simmons et al. 2008). The earliest known bat, Onichonycteris finney, from the early Eocene period (ca 52 MYBP), was capable of powered flight but, unlike modern bats, its digits had tiny claws, which suggests that it could climb trees (Simmons et al. 2008).

Figure 4.22 General body plan of a microchiropteran bat (diadem leaf-nosed bat, Hipposideros diadema).

From Macdonald (2009). Reproduced with permission of Oxford University Press.

Bats probably evolved powered flight from an arboreal species that could glide (Norberg 1990; Dudley et al. 2007). The transition from gliding to powered flight by bats required extensive changes from the ancestral glider bauplan. Although we lack transitional fossil records to accurately describe the steps involved in this (p.359) transition, several authors analysed the possible steps involved using a combination of aerodynamics, biochemical, and morphological analysis (Norberg 1985; Swartz et al. 1992, 2006; Thewissenn & Babcock 1992; Dudley 2002; Dudley et al. 2007). Bishop (2008) provided an integrative analysis of these results and proposed that the following steps were involved: the evolution of a wing capable of acting as an aerofoil; the evolution of enough power acting on this wing to generate thrust while maintaining lift; liberation of the forelimbs from other functional constraints so that the wings could be fully developed; and, finally, the evolution of physiological machinery able to supply enough fuel and O2 to sustain the metabolic costs of powered flight.

The evolution of flight requires extensive, and intertwined, modifications in virtually all organ systems and biological process, mostly notable in the sensory, locomotory, digestive, circulatory, and respiratory systems (Maina 2000a). Some of these modifications, especially in the digestive and circulatory systems, are remarkably convergent between bats and birds (see 4.7.3). Others, however, are strikingly different (e.g. respiration), reflecting the independent evolution of flight in these two quite different ecologically and divergent vertebrate groups (the ancestral mammals diverged from the ancestral birds about 310 MYBP; Kumar & Hedges 1998). For example, bats, like birds and pterosaurs, evolved a number (p.360) of adaptations of the skull to become lighter but, at the same time, strong enough to sustain the large aerodynamics forces imposed by flight (Swartz et al. 1992). In bats and birds, the main flight muscle is the pectoral muscle, which in bats has a relatively smaller mass than in birds. However, in birds, only one set of muscles (the supracoracoideus) is involved in flight, whereas in bats the flight musculature is more complex and involves at least 17 muscles that are active during the downstroke and upstroke of the wings (Norberg 1990). Bats, like their putative ancestral mammal, are nocturnal, whereas birds are primarily diurnal. Competition with birds, increased risk of predation, and hyperthermia have been invoked to explain why bats are evolutionary trapped in a nocturnal niche (Rydell & Speakman 1995; Speakman 1995, 2005; Voigt & Lewanzik 2010).

An evolutionary commitment to a nocturnal niche required profound changes in the sensory system of bats. Chief amongst these changes was the evolution of a sophisticated echolocation system (see 4.7.4). Not all bats echolocate (e.g. some members of the more diurnally active and visually oriented members of the Pteropodidae), and those that do can produce broadband (frequency-modulated), narrowband, or long constant-frequency calls. The larynx produces these calls, but brief broadband calls can also be emitted by tongue clicking (Jones & Teeling 2006). Laryngeal echolocation calls are more sophisticated and complex, and are regarded to be the ancestral mode of echolocation in bats (Springer et al. 2001; Jones & Teeling 2006). Echolocation calls are energetically expensive to produce for resting bats, but these costs are negligible during flight (Speakman & Racey 1991; Voigt & Lewanzik 2012) due to the 1:1 synchronization while flying between the respiratory and the flight muscles (Lancaster et al. 1995; see also 4.7.3).

The body mass range and the maximum body mass attainable by bats are much smaller than those reported for birds capable of powered flight. The body mass of bats ranges from 2 g (bumblebee bat, Craseonycteris thonglongyai) to 1.6 kg (golden-crowned flying fox, Acerodon jubatus), whereas body mass ranges for birds from 2 g (bee hummingbird, Mellisuga helenae) to 12–14 kg bustards (Ardeotis kori), California condor (Gymnogypus californinanus), and mute swan (Cygnus olor). There is a complex interplay between the scaling relationship of wingbeat frequency and flight muscle mass, patterns of echolocation calls, wing morphology, and the capacity to delivery fuel and O2 to the muscles during flight, that placed constraints on the maximum body mass that bats are able to achieve (Arita & Fenton 1997; Bullen & McKenzie 2002, 2004; Norberg & Norberg 2012; Ruxton 2014). As for other mammals, body mass is the most important factor affecting life history traits of bats, so the evolution of flight, by imposing constraints on the range of body mass of bats, has also impacted their evolution of reproductive patterns (Barclay & Harder 2003).

In bats, unlike birds, the wings evolved from a gliding membrane, and the ancestral bats had to develop a different range of structural modifications to fully (p.361) develop and support a wing capable of sustained, powered flight (Sears et al. 2006; Cretekos et al. 2008; Konow et al. 2015). The wingbeat kinematics of bats is different from those of birds, a direct consequence of the evolutionary constraints imposed by the more compliant skin membrane of bats compared to the more rigid feather membrane of birds (Swartz et al. 2006; Hedenström & Johansson 2015). The aerodynamic forces acting on a bat’s wing are the same as those for a glider (Figure 2.22), but a bat uses the power of the wing downstroke to maintain horizontal flight (Figure 4.23). The two common measurements that describe the size and shape of wings are wing loading (WL) and aspect ratio (AR; see 4.7.4). WL is the mass of the bat divided by the area of the wing elements; aspect ratio (AR) is the square of the wingspan (distance from the tip of the wing to the central axis of the body, multiplied by two) divided by the area of the wing elements (Norberg & Rayner 1987). There is a positive linear relationship between AR and (p.362) WL, with the combination of these characteristics determining aerodynamics of bat flight; AR impacts the cost of transport while WL influences agility and manoeuvrability (Fullard et al. 1991). Bats with a high AR and high WL have long, thin wings, resulting in poor manoeuvrability but a low cost of transport, whereas bats with a low AR and low WL have short, thick wings, with good agility but a high cost of transport (Norberg & Rayner 1987). So, in general (as for birds) high WL characterizes species adapted for fast flight but at the expense of a reduced manoeuvrability, while a high AR characterizes species capable of flying with reduced energetic cost. A particular combination of WL and AR can be interpreted as a compromise between speed, endurance, economy, and agility, which, in turn, can be linked to the food habits, habitat use and, for insectivore bats, to the structure of echolocation calls (see 4.7.4; Bogdanowicz et al. 1999; Jennings et al. 2004; Siemers & Schnitzler 2004; Mancina et al. 2012; Voigt & Holderied 2012; Falk et al. 2014; Marinello & Bernard 2014). Wing tip proportions (pointed vs rounded) also impact aerodynamics, particularly manoeuvrability, and can be described by the proportions by which the hand wing and arm wing areas contribute to the total wing area; a rounded wing is characterized by a short, broad hand wing (Aldridge & Rautenbach 1987).

Not surprisingly, wing morphology correlates with the foraging niche of insectivorous bats. Broadly speaking, insectivorous bats forage in three locations: in closed environments (gleaning, within stands of trees or amongst the canopy), beside vegetation (beside or on the edge of stands of trees, just above or under the canopy), and in the open (well above the canopy; Fullard et al. 1991). Bats that primarily forage in open environments tend to have high AR and WL and pointed wing tips; those in closed environments have low AR and WL, with rounded wing tips; while bats foraging beside vegetation have intermediate wing morphologies (Aldridge & Rautenbach 1987; Norberg & Rayner 1987). The aerodynamic principles governing the evolution of the wing morphology for insectivorous bats extend to bats with other diet types. Piscivores and carnivores snatch prey from the ground or water, and therefore require slow flight with a low WL, although a high AR improves flight economics, and long wings are possible in an uncluttered environment. Nectarivores and frugivores often have to travel long distances between food sources, so generally have high AR and WL, although this structure may be somewhat tempered by the requirement to hover and manoeuvre amongst vegetation to access food (Norberg & Rayner 1987). Although some bats are generalists and forage in a variety of habitat types, adaption to the more extreme niches may severely limit the potential to exploit other environments. For example, the long wings of bats that typically forage in open areas mean that they are unable to negotiate the cluttered environment within vegetation; similarly, there are flight constraints on closed-environment bats foraging in the open, particularly in adverse environmental conditions such as high wind speeds (Fullard et al. 1991).

Figure 4.23 Aerodynamic forces acting on a bat wing in horizontal flight; forward movement in concert with wing downstroke creates the upward- and backward-directed incoming air flow. The aerodynamic force is resolved into lift (perpendicular to incoming airstream) and drag (parallel to incoming airstream), and horizontal thrust to counterbalance the other drag forces acting on the bat, and an upward force to counterbalance body weight; α‎ is the angle of attack. Inset shows a Peter’s disk-winged bat Thyroptera discifera;

reproduced with permission of Oxford University Press.

## (p.363) 4.7.1 Metabolic Cost of Flight

Of the two basic modes of flight, gliding is less costly than powered flight. For birds, the metabolic rate while gliding is only twice the BMR (Baudinette & Schmidt-Nielsen 1974; Sapir et al. 2010). For bats, powered flight is up to 17 times BMR (see later). A gliding mammal uses the potential energy gained by its previous climbing, and, because the power required to control and manoeuvre is negligible, the only costs are those necessary to keep the gliding membrane rigid to resist deformation by the aerodynamic forces (Norberg 1985). In fact, it has been postulated that the energetic economy achieved during gliding was one of the main selective forces responsible for the evolution of this mode of flight (the ‘energetic-economy hypothesis’; Dudley et al. 2007; Byrnes & Spence 2011). Mathematical models suggest that the economy achieved by gliding seems to be high at intermediate body mass, but that large gliders must glide a much longer distance than smaller gliders before a substantial economy can be accrued (Dial 2003b; Scheibe et al. 2006). Byrnes et al. (2011) compared the costs of gliding with those of moving horizontally in the canopy for the Malayan colugo (Galeopterus variegatus) and concluded that gliding is energetically inexpensive. However, because the initiation of gliding requires climbing to a certain height, the overall cost of gliding (climbing + gliding) was higher than what it would be if the colugo moved horizontally through the canopy. Nevertheless, since colugos spend a small fraction of their time budgets engaged in gliding, the impact of these costs to their daily energy budget was small.

The basic aerodynamic principles for bat flight are the same as for gliders (see 2.8.2, Figure 4.23), except that the powered downstroke of the wing provides sufficient thrust and vertical force to overcome the drag and weight of the bat, enabling it to fly horizontally indefinitely, whereas the glider eventually descends to the ground. The power requirements for powered flight are complex (Norberg 1990; Rayner 1999). In theory, the total aerodynamic power required for flight (Ptot) is composed of three power requirements, parasite power (Ppar), profile power (Ppro), and induced power (Pin); Ptot = Ppar + Ppro + Pin. Parasite power (Ppar) is the power required to overcome the aerodynamic drag of the body; this increases with flight velocity. Profile power (Ppro), the power required to overcome the aerodynamic drag of the wings, is much lower than Ppar because the wings are more streamlined than the body; however, Ppro also increases with velocity. Induced power (Pin) is the drag associated with the wings having to deflect oncoming air to generate an aerodynamic force. Pin decreases with velocity, because at high velocity relatively little deflection is required to obtain the momentum change needed to counterbalance the body weight; conversely, at low velocity, the slower moving air needs to be deflected downwards more to achieve the same momentum change. Consequently, induced power decreases with speed. The net effect is that the total power required (p.364) for flight $(Paero=Ppar+Ppro+Pin)$ has a U-shaped curve, being high at low and high flight velocities, and lowest at intermediate velocities (Figure 4.24). The flight velocity for maximum range (Vmr) is higher than the velocity for minimum cost (Vmin), equivalent to the velocity where a line from the origin makes a tangent with the parabolic curve.

Figure 4.24 Schematic of the partitioning of total cost of flight (Ptot) as a function of flight velocity, into parasite power (Ppar), profile power (Ppro), and induced power (Pin), with the velocity for minimum cost of flight (Vmin) and maximum range velocity (Vmr).

Modified from Withers (1992).

Another power requirement is for inertial power (Pi), the power necessary to overcome inertial forces (joint friction, internal tissue viscosity) and move the wings up and down. At medium to high speed, Pi is thought to be negligible because the wing inertia is converted into aerodynamic work at the end of each stroke (Norberg 1990). However at very low speed and, especially for stationary (hovering) flight, Pi can be high. In stationary flight, there is no horizontal air flow, so that all air flowing over the wings must be supplied by forces generated by the muscle activity necessary to beat them. This explains the high power requirement of hovering flight compared to horizontal flight. For the nectar-feeding bat Glossophaga soricina, $Ptot(=Pi+Paero)$ was estimated to be 0.34 W for hovering, with Pi accounting for more than 60% of the cost (Norberg et al. 1993). In comparison, the metabolic cost for forward flight near the minimum-cost flight speed (see later) is about 0.14 W.

Each of the individual power requirements in this model embraces a suite of aerodynamic components associated with body and wing morphology, aerodynamic ‘cleanliness’ attributes, and dynamic factors that must be quantified if the earlier equation is to accurately predict the aerodynamic power required for animal flight. However, measurement of such parameters is challenging (Bullen & McKenzie 2001, 2002, 2008; Norberg & Norberg 2012). This classical model assumes a fixed wing and a steady-state flight (Norberg 1990; Rayner 1999), but kinematic studies (p.365) show that the compliant membranous wings of bats have more degrees of change than a bird wing, being capable of pronounced deformation under different aerodynamic loads (Swartz et al. 2006; Cheney et al. 2014). Additionally, unsteady effects have been reported for bats during flapping flight, especially at low speeds (Muijres et al. 2008). Furthermore, this model indicates only the mechanical power necessary for flight, but not the metabolic power actually used during flight.

Basic thermodynamic laws dictate that the conversion of chemical energy to mechanical energy is far less than 100%, so the mechanical costs calculated by aerodynamic models represent only a small part of the total metabolic costs. In fact, the estimated mechanical costs derived from aerodynamic models are nearly five times lower than the actual metabolic costs (Norberg et al. 1993; Speakman & Thomas 2003; Bullen et al. 2014). Mechanical efficiency (η‎), the effectiveness of converting chemical energy to mechanical work, has been indirectly assessed by comparing measurements of metabolic rate during flight with mechanical power derived from the classical power-speed curve. Under the assumption that η‎ is similar at different speeds, these results gave mean η‎ values for bats of 20% (range 10–28%; Speakman & Thomas 2003), which are similar to those calculated for birds (range 12–40%; Ward et al. 2001). However, these are probably overestimates. Bullen et al. (2014), using a quasi-steady model, calculated η‎ for eight species of bats to vary between 3.8 and 18.2%, and reported that this efficiency varies allometrically and with flight speed (from 4 to 10 m s−1). At speeds close to Vmin (5–8 m s−1), η‎ varied from 4.3 to 10.8%. von Busse et al. (2014) used a more complete analysis for Seber’s short-tailed bat, Carollia perspicillata. Combining isotopic measurements of flight costs with simultaneous measurements of mechanical power output, they calculated η‎ at 5.6–11.3%, which, like those of Bullen et al. (2014), overlapped with the values calculated for nectar-feeding bats (Voigt & Winter 1999).

The metabolic cost of flight has been measured for 14 species of bat (Speakman & Thomas 2003; von Busse et al. 2013), at different flight speeds, and using different methods (doubly labelled water, 13C-labelled NaHCO3, mass-balance experiments, and mask respirometry). The U-shaped curve expected for the classical power-speed curve derived from fixed-wing aerodynamic models was observed for Seber’s short-tailed bat (Von Busse et al. 2013), which is more in concert with those reported for birds (Ellington 1991; Bundle et al. 2007). Other studies for the greater spear-nosed bat (Phyllostomus hastatus) and the black flying fox (Pteropus gouldi) suggested a J- or L-shaped curve (Carpenter 1985, 1986; Thomas 1975). Data for a nectar feeding bat (Pallas’s long-tongued bat, Glossophaga soricina) showed that hovering flight increased metabolic rate by 1.2 times above the cost of forward flight, which suggests a shallow U-shape curve (Winter 1998). Differences in body mass and methodology, by affecting the range of flight velocities used in these studies, probably accounts for these differences. Large bats, such as the greater spear-nosed bat and the black flying fox, cannot fly at low velocities, and the use of masks (p.366) to measure flight metabolic rate also prevented measurements at high flight speeds for these species. Von Busse et al. (2013) reported that the minimum metabolic cost of flight for Seber’s short-tailed bat was at a flight speed (4 m s−1) close to the minimum power predicted by the fixed-wing aerodynamic model , but different from the power requirement at Vmr.

As with birds and running mammals, the metabolic cost of flight increases with body mass in bats (Figure 4.25). As noted by Speakman and Thomas (2003), this relationship is surprisingly tight for bats $(r2>0.99)$ if we consider that the data came from results that used different methodologies and included costs that, expect for hovering flight, do not necessarily reflect maximum costs. The highly significant allometric relationship precludes analysis of residual variation to test for the effects of morphological variables, such as aspect ratio, on the metabolic cost of flight in bats (Speakman & Thomas 2003). When the comparison is restricted to the same mass range, the allometric exponent of this relationship (0.74) is similar to those reported for BMR of bats (0.79; Cruz-Neto & Jones 2006). Flight by bats increases their metabolic rate up to 17 times above BMR, but this scope is probably too high; it is unlikely that the basic physiological processes that determine BMR (e.g. Tb) would not increase during flight.

Figure 4.25 Comparison of metabolic rate for flying bats (0.233 M0.939) with flying birds (0.340 M0.754), terrestrial mammals (0.121 M0.795), and BMR of bats (0.135 M0.790).

Data from Hails (1979), Bartholomew and Lighton (1986), Suarez et al. (1991, 1997), Speakman and Thomas (2003), Cruz-Neto and Jones (2006), and Dlugosz et al. (2013).

The scaling exponent for bats is also very similar to those reported for running mammals and birds with a similar mass range (0.75–0.79). As expected (Norberg 1990; Bishop 1999), flight is more expensive than running (2 times in bats and 2.8 times in birds). Birds are able to fly faster and more efficiently than bats (Muijres et al. 2012), so we might expect that the cost of transport would be higher for bats than birds. However, Winter and von Helversen (1998) suggested that bats (p.367) fly more cheaply than birds, and more recent analyses (Speakman & Thomas 2003; Hedenström et al. 2009) suggest that the costs of flight for bats and birds are not significantly different (1.0–1.3). Although the effects of flight speed were not accounted for, these results suggest that aerodynamic considerations and kinematics of flight should be more important in determining the supposed differences in the costs of transport between birds and bats. During hovering, the costs of flight also do not differ between bats and birds (Voigt & Winter 1999), which probably reflects the similarities in aerodynamics and the enzymatic capacity of the flight muscles observed between these two groups (Suarez et al. 1997, 2009).

## 4.7.2 Thermal Balance

Bats are considerably thermolabile mammals. Normothermic resting Tb of bats varies with body mass and ranges from 31 to 38°C (Clarke & Rothery 2008), but the low average Tb of bats of 35.3°C compared to most other placental mammals might reflect methodological difficulties with measuring their normothermic Tb (Willis & Cooper 2009; see 3.2.1; Table 3.3). During hibernation or daily torpor, Tb of bats can drop to $5−25°C$ depending on Ta and M (Geiser 2004; McNab 2012).

The capacity of bats to tolerate an increase in Ta to levels close or even higher than their normothermic Tb is quite variable. Early studies suggested that bats become hyperthermic when exposed to higher Ta, but survival is only possible for a short period of time (Licht & Leitner 1967a,b). At high $Ta(>35°C)$, EHL accounted for more than 50% of MHP for Gould’s long-eared bat (Nyctophilus gouldi; Morris et al. 1994). For the Angolan free-tailed bat (Mops condylurus), a species that thrives in roosts that often exceed 40°C, virtually all their heat production is dissipated by evaporative cooling at Tas approaching roost temperature (Maloney et al. 1999). Bats use behavioural and physiological mechanisms such as wing-fanning to increase evaporative cooling and increase blood flow to the peripheral parts of the body, especially to the wings and to naked and vascularized regions on the side of the body (Ochoa-Acuna & Kunz 1999; Makanya & Mortola 2007; Reichard et al. 2010a).

Total evaporative water loss (EWLtot) is made up of both cutaneous and respiratory components. Respiratory evaporative water loss (EWLresp) is higher than cutaneous evaporative water loss (EWLcut) for bats at high Ta (Minnaar et al. 2014; Muñoz-Garcia et al. 2016). The increase in EWLtot at high Ta places a burden on the capacity of bats to maintain water balance, so we might expect that there are mechanisms to minimize EWLtot and maintain water balance. Different species of bats show differences in their EWLcut and/or EWLresp responses (Muñoz-Garcia et al. 2012a,b, 2016), as well as the capacity of their kidneys to concentrate urine (Studier & Wilson 1983; Schondube et al. 2001; Casotti et al. 2006; Gopal 2013) that are related to body mass, habitat, and diet. (p.368) However, these capacities are not markedly different from what would be expected for similar-sized non-flying mammals having the same diet and living in a similar habitat (Webb 1995; Beuchat 1990, 1996; Al-kahtani et al. 2004; Van Sant et al. 2012). Differences in the capacity of different bat species to minimize EWLtot may partially explain why some bats use torpor in hot, desert areas (Muñoz-Garcia et al. 2016), and also the differential susceptibility of bats to heat waves (Welbergen et al. 2008; Bondarenco et al. 2014; Dey et al. 2015).

During flight, MHP of bats increases by up to 17 times above resting MHP (see 4.7.1), which poses challenges for their capacity to maintain thermal and water balance. To maintain a constant Tb during flight, this excess heat must be dissipated; flight-induced air convention, radiative heat loss, EWLcut, and EWLresp all play role in dissipation. For the mid-sized spear-nosed bat (Phyllostomus hastatus; 90–100 g), MHP and Tb increase with Ta during flight (Thomas et al. 1991). The grey-headed flying fox (Pteropus poliocephalus; 700–800 g) becomes hyperthermic when flying at Ta above 25°C (Carpenter 1985, 1986). EWLresp increases during flight but, as for birds (Torres-Bueno 1978; Giladi & Pinshow 1999; Michaeli & Pinshow 2001), the heat dissipated by this route accounts for only a small fraction (10–14%) of the MHP during flight in bats (Carpenter 1985, 1986; Thomas et al. 1991). The 1:1 synchronization between wing beat and pulmonary ventilation might result in an inability to match EWLresp to increasing MHP (Boggs 2002), with the capacity to increase evaporation more limited than metabolism.

Modulation of EWLcut and heat loss by radiation and convection, thus, might be very important for bats during flight. These possibilities were investigated by Reichard et al. (2010a,b) for thermoregulatory adjustments during natural flight for the Brazilian free-tailed bat (Tadarida brasiliensis), a bat capable of fast flight and high endurance. Thermographic IR images showed that surface temperature (Tsurf, measured at the centre of the body) was always higher than wing temperature (Twing) at , but Twing was always lower than Ta within this range. This suggests that the Brazilian free-tailed bat is able to shunt blood away from the wings. Although this might also be viewed as a strategy to avoid an increase in EWLcut, such regional hypothermia reduces convective heat loss from the wings during flight. The reduction seems to be partially compensated for by an increase in radiative heat loss and by the presence of thermal windows at the flanks of the body and pelvic region, which are capable of dumping any excess heat generated during flight, especially at high Ta. Finally, this bat can fly continuously at high altitudes (up to 2,000 m) for long distances (up to 100 km; McCracken et al. 2008). Birds that fly at high altitudes for long distances may remain in water balance and avoid dehydration (Torres-Bueno 1978; Landys et al. 2000), so it is possible that bats might also be able to maintain water balance during long flights at high elevation.

The possibility that renal function, by controlling sensible water loss, is correlated with flight capacity and, hence, affects flight performance has been (p.369) investigated by Happold and Happold (1988) for an assemblage of insectivorous bats in Africa. Although they found some correlations between habitat use, flight capacity, and renal function, the causal link underlying these mechanisms remains unclear. Nevertheless, some of the putative mechanisms suggested by Happold and Happold (1988) might explain the broad differences observed between studies of Carpenter (1985, 1986) and Thomas et al. (1991), on one hand, and that of Reichard et al. (2010), on the other. For example, the Brazilian free-tailed bat seems to have adaptations that reduce the amount of water loss during flight without jeopardizing its ability of maintain Tb at constant levels. As also expected from its insectivorous diet, this bat has a low BMR (and hence low EWLtot during rest) and kidneys with a higher relative medullary thickness and a higher capacity for concentrating urine than the frugivorous flying-fox and the omnivorous spear-nosed bat (Geluso 1978; McNab 1992; Casotti et al. 2006). Furthermore, Brazilian free-tailed bats are able to fly fast, high, and for long periods—an ability not shared by the other two frugivorous species. This interplay between renal morphology, flight capacity, and mechanisms to adjust EWL may thus explain the greater capacity of Brazilian free-tailed bats to maintain water and thermal balance during flight.

## 4.7.3 Digestion, Respiration, and Circulation

The gross anatomy of the digestive system of bats does not depart widely from that of other mammals (Stevens & Hume 1995). One exception is that, as for birds, the large intestine tends to be short, with less nominal surface area in bats compared to similar-sized non-flying mammals (Caviedes-Vidal et al. 2007; Price et al. 2015a). The reduction in intestinal volume, and hence mass of food carried, would be advantageous, because the costs of flight increase with load carried, and because takeoff and manoeuvrability are diminished at heavier mass (Kvist et al. 2002). One drawback of this reduction in the large intestine length and surface area is that food retention time is reduced for bats (Price et al. 2015a).

To compensate for reduced large intestine length and surface area, bats show some digestive adaptations that are convergent with those of birds. Unlike non-flying mammals, but similar to birds, bats usually rely on passive, paracellular absorption of water-soluble nutrients, such as glucose and amino acids, to a greater extent than active-transport transcellular uptake (up to 70% paracellular; Tracy et al. 2007; Caviedes-Vidal et al. 2008; Fasulo et al. 2013; Price et al. 2013, 2015b; Brun et al. 2014). Evidence suggests that the high paracellular absorption of bats results from an increased number of tight junctions in the intestine, which have a high permeability to nutrient-sized molecules (Brun et al. 2014; Price et al. 2015a). Compared to other mammals, bats also have more villous area and a higher number of enterocytes per cm2 of nominal surface area (Zhang et al. 2014), which may provide a mechanistic explanation for their high reliance on paracellular (p.370) absorption. All these differences are related to the costs of flight; a high reliance on paracellular absorption has also been observed for birds (Lavin et al. 2008; Karasov et al. 2012), presumably as a mechanism to overcome the reduction in intestinal length and area.

One functional consequence associated with the high reliance of bats on paracellular absorption is that they are able to meet their high energetic demands of flight using dietary nutrients. Non-flying mammals, which rely more on transcellular absorption, usually meet increased energy demands by mobilizing endogenous substrates such as fats or glycogen. Bats, on the other hand, are capable of channelling the substrates of recently eaten food to meet their energetic requirements (Voigt & Speakman 2007; Welch et al. 2008; Amitai et al. 2010; Voigt et al. 2010, 2012). Again, this is convergent with birds. For example, nectar-feeding birds, such as hummingbirds, also rely heavily on paracellular absorption and are able to meet the extremely high costs of hovering flight using exogenous sources (McWhorther et al. 2006; Welch et al. 2008; Chen & Welch 2014).

The capacity of bats to mobilize exogenous energy sources might also alleviate the costs of long-distance migration (see 4.6.3). Unlike birds, bats cannot mobilize fat stores to support the costs of long-distance migrations, and do not have the equivalent physiological modifications such as fattening, observed for birds prior to migration. Voigt et al. (2012) showed that the migratory Nathusius’s pipistrelle (Pipistrellus nathusii), in addition to using endogenous fatty acids, is able to oxidize proteins directly from the insects captured en route, and to mobilize the fatty acids of its diet to replenish body reserves. Reliance on exogenous sources may decrease the energetic costs associated with lipogenesis and gluconeogenesis, which, coupled with the energetic savings that accrue during torpor, may reduce the total energetic costs associated with migrations in bats (Guglielmo 2010).

The respiratory system of bats, like their digestive system, does not depart widely from the typical mammalian bauplan, but it does show some refinements associated with flight. Bats have relatively larger lung volume and heavier lungs compared to similar-size non-flying mammals (Jürgens et al. 1981; Canals et al. 2005a, 2011), and compared also to the volume of lung parenchyma in birds (Maina 2000a,b). The allometric relationship between lung volume (VL; ml) and M (g) for nine species of bats, ranging from the 3-g Kalinowski’s mastiff bat (Mormopterus kalinowski) to the 928-g grey-headed flying fox (Canals et al. 2005a), is $VL=0.0714Mb0.903±0.019$

Besides having a large lung, bats show modification of some parts of the respiratory system, which is thought to enhance its capability for gas exchange and, to some extent, to reduce the costs of breathing. For example, breathing work depends on the airways resistance of all components of the respiratory system, especially the proximal airway, to air convective flow (Zakynthinos & Roussos 1991). Canals et al. (2005b) showed that the Brazilian free-tailed bat (Tadarida (p.371) brasilensis), when compared to rodents, had extensive modifications in the geometry of its bronchial tree that would reduce the cost of breathing and minimize the energy cost of flight. Other respiratory structural parameters seem also to be refined in bats. For example, the alveoli of bats are smaller (Maina 2000a; Maina et al. 1991), which, combined with their high lung volume, means that the total respiratory surface area available for gas exchange is higher for bats (Canals 2005b; Maina 2000a,b). In fact, the highest mass-specific respiratory surface area reported thus far for a vertebrate (138 cm3 g−1) was for Wahlberg’s epauletted fruit bat (Epomophorus wahlbergi; Maina et al. 1991). In addition, the alveoli are highly vascularized (see later) and the thickness of the alveolar-capillary barrier is greatly reduced in bats compared to non-flying mammals (Maina et al. 1991).

Bats rely on the basic mammalian pattern of a bidirectional convective flow to ventilate their large lungs. The highly vascularized flight membrane of bats is a suitable site for the passive diffusion of gases, but under resting conditions it accounts for only 2–10% of the total O2 uptake (Herreid et al. 1968; Makanya & Mortola 2007). It is highly unlikely that cutaneous O2 uptake, even if it increased during flight, would be enough to accommodate the extra metabolic demand for flight. Rather, the respiratory accommodation of the metabolic cost of flight requires adjustment of respiratory minute volume (VI), which is the product of breathing frequency (fR) and tidal volume (VT). At resting, thermoneutral conditions, VI is usually higher for bats (and birds) than for similar-sized non-flying mammals, due to their larger VT and fR (Table 4.5). The amount of O2 extracted per volume of air that reaches the lung (EO2) at rest is available for only three species, and does not allow for meaningful comparisons. The greater spear-nosed bat (Phyllostomus hastatus) has an EO2 of 29%, and the other two species (black flying fox Pteropus gouldii and lesser bulldog bat Noctilio albiventris) of 16 and 18% respectively, all within the range measured for other mammals and also birds.

Table 4.5 Body mass (M, g), breathing frequency (fR, min–1), tidal volume (VT, ml), minute volume (VI, ml min–1), and oxygen extraction (EO2,%) for bats at rest (or near the lower critical limit of the thermoneutral zone). Numbers in parentheses denote the ratio of observed value with those expected from allometric equation for mammals and birds (in bold). All respiratory values are in standard temperature and pressure units.

Species

Mg

fR min–1

VT ml

VI ml min–1

EO2%

Scotorepens balsoni1

8

68.0

0.05

4.5

(0.36; 0.77)

(0.99; 0.4)

(1.41; 1.36)

Plecotus auritus2

8.5

159.2

(0.82; 1.85)

Leptonycteris sanborni3

22.0

30

(0.20; 0.48)

Noctilio albiventris4

40.0

91.5

0.20

17.5

18.3

(0.74; 1.80)

(0.74; 0.28)

(0.61; 0.46)

(1.1; 0.88)

Phyllostomus discolour5

43.1

170.5

0.39

61.05

(1.40; 3.44)

(1.33; 0.51)

(1.99; 1.52)

Phyllostomus hastatus6

110

81.6

0.88

72.05

29

(0.86; 2.27)

(1.14; 0.42)

(1.11; 1.89)

(1.81; 1.38)

Macroderma gigas7

150

55

(0.63; 1.69)

Pteropus dasymullus1

379

56.0

4.81

272.8

(0.81; 2.36)

(1.71; 0.60)

(1.56; 1.42)

Pteropus gouldii8

777

44.8

9.8

436.7

16.0

(0.78; 2.42)

(1.56; 0.56)

(1.41; 1.36)

(1.1; 0.84)

Allometric equations for mammals from Stahl (1967) and for birds from Frappell et al. (2001; conventional equations). Expected data (and ratio) for EO2 taken directly from the source. Sources:

During flight, the metabolic scope of bats increased dramatically, reaching values that are higher than those observed during thermoregulation in the cold, and similar to those reported for flying birds. To accommodate this increase in O2 demand during flight, bats can either increase VI, via changes in fR or VT, or both, and/or increase EO2. VI tends to increase by 10–17 times for bats during steady flight (Thomas 1981; Thomas et al. 1994). For all bats studied to date, fR increases by up to sixfold (Thomas & Suthers 1972; Thomas 1981; Carpenter 1985, 1986; Thomas et al. 1994).

It is difficult, however, to generalize about the relative contribution of fR to the increase in VI during flight, as only two studies concomitantly measured fR, VT, and VI. In the black flying fox (Pteropus alecto), VT and fR increased in roughly the same proportion (Thomas 1981), while in the greater spear-nosed bat, fR increased by more than fivefold while VT increased only 1.8 times above resting levels (p.372) (Thomas et al. 1994). For both species, as observed during resting conditions, the increase in fR during flight was higher than that observed for birds. This is expected, as bats shows a rigid 1:1 synchronization between wingbeat and respiratory cycle, while only few species of birds display such a rigid match (Carpenter 1985; Boggs 2002). VT during flight, however, was lower for these two bats compared to birds, but similar to what was observed for active non-flying mammals. This is also expected based on the functional constraints imposed by a typical mammalian lung (Maina 2000a). It is worth noting that EO2 during flight decreased for both species, from 29 to 22% for P. hastatus and from 16 to 10% for P. gouldii (Thomas 1981; Thomas et al. 1984). Why EO2 differs between these two species is unclear, but the overall decrease in EO2 suggests that bats might hyperventilate their lungs during flight (Thomas 1984).

(p.373) These morphological refinements of the respiratory systems accommodate the increase in O2 demand during flight (Canals et al. 2011); refinements of the cardiovascular system might also contribute to ensuring an adequate supply of O2 to the tissues in bats. As with birds, bats have a proportionally larger heart compared to similar-sized non-flying mammals (Jürgens et al. 1981; Bishop 1997; Canals et al. 2005a). This is mostly due to an increased mass of the right atrium and ventricle, which presumably accommodates an increased venous return during flight and pumps this blood for oxygenation in the lungs (Kallen 1977). For bats, heart mass (Mh; g) scales with M (g) as follows: $Mh=0.0173Mb0.741±0.023$ (data from Table 4.6).

Table 4.6 Body mass (M, g), heart mass (Mh, g), relative heart mass $(RHM=Mh/M)$, Mh_e (expected heart mass based on the allometric equation $Mh=0.0173M0.741$), and the ratio between observed/expected Mh (Mh/Mh_e) for various species of bat.

Species

Mb

Mh

RHM

Mh_e

Mh/Mh_e

Mormopterus kalinowski3

3.10

0.057

0.018

0.040

1.423

4.30

0.010

0.002

0.051

0.19

Pipistrellus pipistrellus2

4.85

0.061

0.013

0.056

1.09

Myotis chiloensis3

6.88

0.096

0.014

0.072

1.32

Lasiurus borealis1,3

7.87

0.069

0.009

0.080

0.86

Mormoops blainvilli1

8.30

0.015

0.002

0.083

0.18

Monophyllus redmani1

8.60

0.280

0.033

0.085

3.28

Eptesicus fuscus1

9.30

0.090

0.010

0.090

0.991

Histiotus macrotus3

9.65

0.166

0.017

0.093

1.787

9.80

0.178

0.018

0.094

1.893

Molossus molossus3

11.50

0.320

0.028

0.106

3.02

Histiotus montanus3

12.50

0.272

0.022

0.113

2.41

Lasiurus cinerus1

12.76

0.173

0.014

0.114

1.51

Erophylla sezerkoni1

13.20

0.038

0.003

0.117

0.32

Myotis myotis2

20.60

0.202

0.010

0.163

1.23

Molossus ater2

38.20

0.371

0.010

0.258

1.43

Artibeus jamaiscensis1

39.40

0.158

0.004

0.264

0.59

Brachyphylla cavernarum1

43.50

0.109

0.003

0.284

0.38

Phyllostomus discolor2

45.20

0.425

0.009

0.292

1.45

Noctilio leporinus1

56.60

0.130

0.002

0.345

0.37

Rousettus aegyptiacus2

146.00

1.226

0.008

0.696

1.76

(p.374) The relatively low predictive power of this equation means that other variables besides Mb can account for the variability in Mh. For example, Rodriguez-Duran and Padilla-Rodriguez (2008) showed that wing loading accounted for much of the variability in Mh for bats. This in turn suggests that factors such as flight style and diet can also explain the variability in Mh. For instance, the highest relative Mh (as well as the largest positive deviations from the expected value based on the allometric equation) was for the Greater Antillean long-tongued bat (Monophyllus redmani), a bat capable of energetically challenging hovering flight.

Cardiac output (Q) is the product of heart rate (fH) and stroke volume (VS). No data are available for Q or Vs in bats. The fH of resting bats is comparable to those of similar-sized non-flying mammals (Carpenter 1985), but shows a capacity for rapid and large increases to accommodate changes in O2 demand. For example, bats can change fH from less than 10 min−1 during torpor to more than 800 min−1 while thermoregulating at low Ta (Currie et al. 2014). During flight, fH increases up to sixfold above pre-flight levels (Studier & Howell 1969; Carpenter 1985, 1986), which is higher than that of running mammals. The capacity of bats to vary fR might be related to their capacity for increased venous return. Venous return occurs through two venae cavae, with the more muscular posterior part helping to store blood while at rest and then quickly releasing this blood in flight (Kallen 1977). Vs is expected to vary as a function of Mh (Bishop 1997), and therefore one might expect it to be higher in bats than non-flying mammals and birds. However, like Q, no studies have actually measured VS for bats. It is likely that the increase in Q to accommodate the changes in O2 demands during flight is mostly regulated by increases in fH in bats.

Kallen (1977) presented a detailed overview of the general circulatory pattern in bats. Relative to the increased O2 demands of flight, the most extensive modifications can be seen at the level of blood flow and pressure in capillaries supplying the lungs and pectoral muscles. Although bats have some interesting aspects associated with blood circulation to the wings (Kallen 1977; Davis 1988), these probably have more to do with regulation of heat transfer than with O2 uptake during flight (see also section 4.7.2). Lungs and pectoral muscles in bats are highly vascularized, and a great portion of Q must be diverted to supply these tissues as bats commence flight. Pulmonary blood flow is higher in bats than in non-flying mammals, and the thinner interface between the highly branched capillary bed perfusing the lungs and the alveoli facilitates O2 uptake (Maina 2000). The capillary bed supplying the pectoral muscle is also highly branched (Kallen 1977), and the number of capillaries per muscle fibre—compared to the bat hindlimb or the rat soleus muscle—is also high (Mathieu-Costello et al. 1992; Mathieu-Costello 1993). This facilitates delivery of O2 to the flight muscles at a rate that is comparable to those observed in birds.

(p.375) A suite of haematological changes further enhances the capacity of bats to deliver O2 to their tissues. Blood volume and blood Hb-O2 affinity (P50) of bats are not different from those of other similar-size mammals (Snyder 1976; Jürgens et al. 1981; Boggs et al. 1999). However, erythrocytes of bats are smaller, and the number of erythrocytes per volume of blood is higher for bats than for non-flying mammals (see 3.4.1). The mean Hct of bats (around 60%) is far higher than those reported for non-flying mammals and birds, with some insectivorous species (Pipistrellus pipistrellus, Molossus sinaloae, and Molossus bondae) reaching 64–65% (Jürgens et al. 1981; Rodriguez-Duran & Padilla-Rodriguez 2008; Schinnerl et al. 2011). Although this high Hct increases blood viscosity, the negative consequences for blood flow might be counterbalanced by the high Mh and fH observed in bats. In concert with the elevated number of red blood corpuscles, bats also have a high Hb (180–240 g l−1), which gives a blood oxygen carrying capacity of up to 30%, values that are higher than for non-flying mammals and birds (Thomas 1987; Jürgens et al. 1981; Canals et al. 2011).

In summary, bats have evolved a myriad of changes (or refinements) of their digestive and cardiorespiratory systems to secure an adequate rate of nutrient acquisition—and supply and delivery of oxygen to their tissues—to meet the high energetic demands associated with flight (Table 4.7). Some of these changes, such as those observed in the respiratory system, were made within the design constraints imposed by the basic mammalian bauplan. Other modifications, such as those observed in the digestive, and especially at the cardiovascular, systems, were (p.376) more extensive; and some of them represent classical cases of convergent evolution with birds.

Table 4.7 Summary of the main digestive, respiratory, and circulatory adaptations of bats as compared to non-flying mammals. After Maina (2000) and Canals et al. (2011).

Shorter intestine*

Increased lung volume

Increased heart mass*

Reduced food transit time*

Small alveoli

Increased heart rate*

Increased villus area

Thin blood-gas barrier

Development of right side of the heart

Increased number of enterocytes

High O2 diffusing capacity

Increased venous return

High paracellular absorption*

High respiratory frequency*

Increased capillary density in muscle*

High use of exogenous fuel*

Changes in the proximal airway

High hematocrit*

High haemoglobin concentration*

High O2 transport capacity*

(*) Denotes traits whose magnitude are thought to be similar to or even higher than in birds.

## 4.7.4 Echolocation

Bats generally have sophisticated echolocation capabilities, for orientation and prey location and even interspecific communication (Kunz & Fenton 2003; Jones & Holderied 2007; Voigt-Heucke et al. 2010). Although many bat calls are quite loud (e.g. 128–133 dB) to maximize the echo strength, ‘whispering’ bats, which listen for the faint rustling noises of their prey (e.g. scorpions), have lower call intensity (e.g. 82 dB). Frequencies vary from 11 kHz (e.g. Euderma maculatum) to 212 kHz (Cloeotis percivali), with most insectivorous species 20–60 kHz (Jones & Holderied 2007). Low frequencies are avoided because the echo is weak for wavelengths longer than about an insect-wing length, and very high frequencies are avoided because they suffer from severe atmospheric attenuation.

The often complex acoustic structure of different bat calls has adaptive significance. In general, long narrowband calls are used for long-range detection, whereas short broadband calls are used for accurate localization and precision. For example, calls of Myotis nattereri range from 16 to 135 kHz; this enables the bat to discriminate its prey from background clutter. The acoustic characteristics of bats’ echolocation calls can, therefore, be directly related to foraging environment and are closely related to wing morphology, mediated by the requirements of their foraging niche, which has placed strong selection pressure on the characteristics of both these important components of bat ecology (Norberg & Rayner 1987). Cluttered vs uncluttered environments require quite different acoustics for successful echolocation, and as a consequence of similar selection pressures, wing morphometrics and echolocation calls are closely correlated (Aldridge & Rautenbach 1987; Norberg & Rayner 1987). Echolocation calls can be characterized by their duration, intensity, peak and minimum frequency, and bandwidth, with peak frequency probably the most important determinant of echolocatory ability, as this appears to correlate most strongly with habitat and wing morphometric characteristics (Fullard et al. 1991).

Bats that forage in closed environments must contend with background acoustic contamination. Short, faint, high-frequency calls with wide bandwidth are most beneficial in these cluttered habitats, but these calls are impacted by atmospheric attenuation. Bats foraging in the open have more intense, longer calls, with lower frequency and narrower bandwidth, as these will attenuate less and provide better long-distance resolution, although they are more impacted by clutter (Aldridge & Rautenbach 1987; Norberg & Rayner 1987; Fullard et al. 1991). Consequently, bats foraging in open habitats have high WL and AR together with low-frequency, low-bandwidth calls, adapted for energetically efficient flight and long-range prey detection. Those in more cluttered environments, although morphologically and (p.377) physiologically more variable (as a consequence of subtle variations in microhabitat and their associated demands), have low WL and AR combined with higher-frequency, broader-bandwidth calls (Aldridge & Rautenbach 1987; Norberg & Rayner 1987; Figure 4.26).

Figure 4.26 Relationship between wing loading (which is correlated with aspect ratio), peak call frequency, and foraging location relative to vegetation for a guild of bats in Kruger National Park, South Africa.

Data from Aldridge and Rautenbach (1987).

For example, the fast flying molossids have the highest WL amongst bats (and similar to birds of the same body mass and diet), but their decreased agility associated with a high WL, which elevates the costs of manoeuvrability, restricts these bats to forage in open spaces above the vegetation. These bats use narrowband echolocation calls of long duration, which are ideal for locating targets at long distances in open spaces. On the other hand, Vespertilionid bats, which glean for insects amongst foliage, need to be more manoeuvrable, with wings of lower WL. These bats use short, broadband calls, which are well suited to locate prey at short distances. The need to be more manoeuvrable also explains the comparatively low WL of fruit- and nectar-feeding bats, which have to find their food amidst vegetation. For these bats, however, especially for hovering nectar-feeding bats, selective pressure was probably directed more towards an increase in endurance and economy, so their wings also have a high AR.

# 4.8 Difficult Digestion

The role of digestion is to provide the nutrients and energy required by mammals for maintenance and growth (Barboza et al. 2009). Mammals that ingest easy-to-digest material, such as vertebrate flesh and fruits, generally have a simple digestive tract, consisting of a stomach and short intestine (see 2.6, 3.5). However, many mammals have specialized diets, and consume less digestible (p.378) foods including proteins, such as keratin (in skin, nails, hooves, horns); bones; lipids, such as waxes (in many marine crustaceans, such as krill); the polysaccharide chitin (a nitrogenous polysaccharide found mainly in insect cuticle); and plant wall materials (glucose polymers, including cellulose, hemicellulose, and lignins). Digesting these materials requires specialized gut morphologies and physiologies.

## 4.8.1 Keratin, Bone, Wax, and Chitin

Keratin, in the form of skin, hooves, and horns, is particularly indigestible, and most carnivores avoid ingesting these components of carcasses. However, hypercarnivores such as hyaenas consume essentially all parts of a carcass, including skin, hooves, horns, and bones, and digest much of it. Nevertheless, hyaenas regurgitate much of the skin, hooves, and horns (Bearder 1977). For example, remains of skin are present in 35% of regurgitated material compared to 9% of scats, intact hooves in 27% of regurgitations compared to 0.4% of scats, and ceratin (horn) remains in 41% of regurgitations and 3% of scats. Hair is present in almost all scats, with 24% of scats having 10–20% hair content, and 15% of scats containing 80–100% hair. It is not clear how much, if any, energy is derived from ingested skin, hooves, horn, or hair.

Mammal bones have a very high mineral content (about 37% of wet mass) relative to their organic (primarily bone marrow; about 31%) and water (about 32%) content (Houston & Copsey 1994). Nevertheless, bones have a higher energy content per wet weight (6.7 kJ g−1) than muscle (5.8 kJ g−1; Brown 1988 cited by Houston & Copsey 1994), partly reflecting their high lipid content, so osteophagy (‘bone-cracking’) is an energetically viable, if morphologically demanding, strategy. Bone is generally a relatively small fraction of the diet for mammalian carnivores (e.g. hyaenas), but it is 70–90% of the diet for bearded vultures (Gyptaeus barbatus), which have a mean digestibility of 50% for bones (Houston & Copsey 1994). Digestion and mechanical breakdown for passage through the gut is presumably facilitated by the low stomach pH of bearded vultures, and mammalian scavengers also have a lower stomach pH (1.9) than other mammals (3.7), with herbivorous foregut-fermenting mammals having the highest pH (6.1; Beasley et al. 2015).

Consumption and digestion of bones requires the mechanical capacity to crack the bones prior to ingestion, hence a robust dentition and high bite force (Wroe et al. 2005; Tseng 2013). The killing bite of carnivores, using the canines, also requires maximal bite force. Wroe et al. (2005) compare the bite force of different mammalian carnivores from an allometric and phylogenetic perspective. Using a bite force quotient (BFQ), calculated as the residual of the allometric relationship for bite force (i.e. mass-adjusted bite force) normalized to 100, they found that the extant mammal with the highest BFQ was the Tasmanian devil (Sarcophilus (p.379) harrissi; 181), with the highest placental carnivore being the African hunting dog (Lycaon pictus; 142). Including fossil species, two extinct marsupial lions had higher BFQs (Thylacaleo carnifex, 194; Priscileo roskellyae, 196), and the highest placental was the dire wolf (Canis dirus, 163). In general, hypercarnivores, which prey on animals larger than themselves, have a higher BFQ (120 ± 8) than other carnivores (86 ± 7). Surprisingly, the osteophagous (bone-cracking) hyaenas did not have particularly high BFQs (Hyaena hyaena, 113; Crocuta crocuta, 117), despite their very robust dentition, although the Tasmanian devil is also osteophagous. The ability to crack bones is perhaps related more to dynamic than static bite forces (related to BFQ), and to unilateral biting using the carnassial to fracture bones. Not surprisingly, the specialized termitivorous aardwolf (Proteles cristatus), which is closely related to hyaenas, has a low BFQ (77).

Mammals are generally able to easily digest most lipids, but digestion of waxes is problematic. Waxes are esters of long-chain fatty acids and mono-hydroxylic alcohols (Stevens & Hume 1995), each of which is readily digestible separately but not when esterified. Waxes are found in plant cuticle and beehives, but are especially predominant in marine ecosystems (as is chitin; see later), in marine invertebrates such as planktonic crustaceans (e.g. krill), fishes, and the spermaceti of whales. Waxes supplement or supplant triglycerides for energy storage. Planktonic crustaceans are very ecologically important in marine food chains, and up to 50% of their lipid synthesis is waxes (Lee et al. 1971).

Terrestrial mammals do not seem to have a particularly effective pancreatic lipase for wax digestion (< 50% digestibility), or particularly high levels of bile salts in the gall bladder (21–358 mM) to emulsify fats, but many seabirds have much higher digestive efficiencies (> 90%) and high bile salt concentration (469–507 mM; Place 1992). Seabirds can also return gastric and duodenal contents to their gizzard for further mechanical processing, and overall have about equivalent utilization of waxes and triglycerides. Unlike terrestrial mammals, Minke whales (Balaenoptera acutorostrata) have a high wax digestibility (94%), equivalent to seabirds, as would be expected, because wax esters are 21% of the total energy and 47% of the total lipids in their krill diet (Nordøy 1995). Swaim et al. (2009) estimated from models of ingestion and defaecation that North Atlantic right whales could digest more than 99% of the about 58 kg of wax esters ingested per day (by a 40,000-kg animal).

Chitin is a structural carbohydrate (polymer of n-acetyl-glucosamine) that is present in many animals; it is up to 60–85% of dry mass for arthropods, and is present in fungal cell walls as well as in jaws, chaetae, exoskeletons, and the like of many marine invertebrates (Stevens & Hume 1995). It is broken down to chitobiose by the enzyme chitinase, then to glucosamine by chitobiase. Some mammals produce endogenous chitinase to digest dietary chitin (e.g. rodents, monkeys, pigs, and especially some insectivore-consuming carnivores and bats; Cornelius et al. 1975; Stevens & Hume 1995; Strobel et al. 2013).

(p.380) There are approximately 22 species of myrmecophagous mammal: highly specialized consumers of ants and/or termites (e.g. placental armadillos, Cabassous and Tolypeutues spp.; silky anteater, Cyclopes didactylus; giant anteater, Myrmecophaga tridactyla; tamandua, Tamandua tetradactyla; pangolins; aardvark; aardwolf, Proteles cristatus; the marsupial numbat, Myrmecobius fasciatus; and the echidna; Redford & Dorea 1984; Redford 1987). Despite representing very different evolutionary linages, myrmecophagous mammals are highly convergent and share a suite of anatomical and physiological characteristics that can be interpreted as adaptations and preadaptations to their diet. Typically, these mammals have structurally reduced, peg-like teeth (e.g. aardwolf), no teeth at all (e.g. echidna, giant anteater), and sometimes supernumerary teeth (e.g. numbat; reflecting the lack of selection pressure on its dentition); a long, extendible, vermiform tongue, which is effective in reaching the inner recesses of ant and termite nests; enlarged salivary glands, which produce copious sticky saliva to facilitate the capture of social insects with the tongue; an elongate snout and palate; anomalous stomachs; and digging-adapted forelimbs (Griffiths 1968). Myrmecophagous mammals also share a low-energy physiology, characterized by low Tb and low BMR (McNab 1984; 2000). Although it is unclear if these shared physiological characteristics are really dietary adaptations or have arisen as a consequence of the phylogeny and other characteristics (e.g. semi-fossorial, armoured) of myrmecophages, it is apparent that this low-energy approach is necessary for specialization on a diet of ants and termites (Cooper & Withers 2002). Despite the localized abundance of ants and termites, a myrmecophagous diet has a low energy yield due its temporally and spatially patchiness, short feeding bouts (due to the prey’s physical and chemical defences), low energy density (due to the inevitable ingestion of indigestible debris while feeding), and low digestibility (due to the chitin content of the prey; Redford & Dorea 1984; McNab 1984; Cooper & Withers 2004).

Myrmecophages don’t appear to digest the chitin of their prey; rather, they digest the tissues and eliminate the chitinous exoskeleton in their faeces (Cooper & Withers 2004). Many species, such as aardvarks, pangolins, anteaters, and armadillos, have specialized stomachs with a large muscular wall, cornifed stratified epithelium, and keratinized ‘teeth’ that grind termites (and presumably expose the tissues to digestion) and manage the large quantities of abrasive dirt ingested during feeding (Griffiths 1968). However, apparent digestibility does not appear to differ from those species that do not have these digestive specializations, being about 64–81% of dry matter (Cooper & Withers 2004).

Another significant group of chitin-consuming mammals are the baleen (mysticete) whales, which filter zooplankton and small crustaceans (Macdonald 2010). They lack teeth (except as embryonic vestigial buds), but have extensive baleen plates, which are keratinous, fringed filters in the oral cavity that strain the contents of large volumes of seawater. Some mysticetes, like right whales (Eubalaena), skim (p.381) the ocean surface to feed, whereas others, like Sei whales (Balaenoptera borealis), suck seawater into their extremely large oral cavity, then squeeze it out though the baleen. The filtered zooplankton and crustaceans are then scraped off the baleen using the tongue, and swallowed. Not surprisingly, these marine mammals have chitinolytic bacteria to digest dietary chitin (Souza et al. 2011). Minke whales, for example, have bacterial chitinase in their aglandular forestomach that dissolves chitinous crustacean exoskeletons (Olsen et al. 1999). The digestibility by Minke whales of chitin is 93% (joule content of krill is 23.8 kJ g−1); the crabeater seal, which also consumes krill but has a single-chambered stomach, has a slightly lower digestive efficiency of 84% (Mårtensson et al. 1994).

## 4.8.2 Plant Fermentation

The Earth has experienced profound changes in climate, hence plant diversity, since about 65 MYBP, and these changes have been accompanied by the diversification of mammals, especially the herbivorous mammals (Stevens & Hume 1995). It is these herbivorous mammals that have evolved the greatest specializations for digestion, because their plant food is often high in carbohydrates such as cellulose and hemicellulose that are difficult or impossible for normal vertebrate digestive enzymes to hydrolyze.

In terrestrial ecosystems, the primary structural carbohydrates are plant cell wall components, such as cellulose, hemicellulose, and lignin (Withers 1992; Stevens & Hume 1995). Cellulose is the main constituent of plant cell walls, being 20–40% of the dry matter. Cellulose is straight polymeric chains of glucose (as is animal glycogen), but the glucose units are joined by a different three-dimensional bond arrangement; the linkages between glucose subunits are described as α‎-1,4 linkages in glycogen and β‎-1,4 linkages in cellulose. This three-dimensional difference in α‎ and β‎ bonds renders enzymes that can hydrolyze α‎ bonds ineffective for β‎ bonds, and vice versa. Hemicellulose is a branching polymer of polysaccharides (e.g. xylose, glucose, mannose, arabinose), often based on xylose subunits, joined by β‎-1,4 linkages; it is covalently bound to lignin, making it less water soluble. Lignin is not a polysaccharide; rather it is a phenyl-propane polymer with cutin, tannins, proteins, and silica. This difference in α‎ and β‎ three-dimensional linkages renders cellulose impervious to the action of animal amylases; cellulase is required for cellulose breakdown. In fact, three different cellulases are required for the complete breakdown of cellulose: endo-β‎-gluconases split β‎-linkages; exo-β‎-gluconases split glucose or cellobiose from the end of polysaccharides; and β‎-glucosidases hydrolyze cellobiose to glucose (Withers 1992). Some invertebrate animals endogenously produce cellulase, and consequently are able to digest cellulose (e.g. some crustaceans, silverfish, snails, and wood-boring beetles), but no vertebrate animals are able to endogenously produce a cellulase enzyme. Nevertheless, many herbivorous mammals are able to digest cellulose and various other plant cell (p.382) wall structural carbohydrates, via the fermentative action of symbiont microorganisms (see later).

There have been various changes in the digestive anatomy and physiology of herbivores to utilize the ingested plant cell wall biomass (Batzli & Hume 1994; Stevens & Hume 1995). The maximum volume of the digestive tract is up to about 25% of body mass for herbivores (cf. < 10% for carnivores and omnivores), and the retention time for food in the gut is correspondingly increased (Barboza et al. 2009). In general, the gut becomes more complex for more difficult-to-digest foods. The simple digestive tract of carnivores, by contrast, is essentially a simple tube with digesta flowing through it in a pulsatile manner.

Herbivorous mammals adopted essentially four strategies to exploit plant food sources. First, some (e.g. bears, Ursus americanus) have an elongated midgut for greater digestion of nutrients, and possibly for accommodating some microorganismal fermentation. Second, some (e.g. rabbits, Oryctolagus cuniculus; koalas, Phascolarctos cinereus) have an enlarged caecum for microorganismal fermentation, and many use coprophagy or caecophagy to facilitate uptake of the products of this fermentation (see later). Third, perissodactyls (e.g. horses, rhinoceros, tapirs), elephants, and primates rely on an enlarged hindgut for fermentation. Fourth, many diverse groups of herbivores have a complex, often multi-chambered stomach that acts as a fermentation chamber; these are foregut fermenters. Ruminant artiodactyls (e.g. bovids, deer, antelope, camels, llamas, giraffes) have a complex four-chambered stomach, and regurgitate stomach contents into their oral cavity to rechew the digesta; this is rumination and the reason they are called ruminant mammals. Other artiodactyls (pigs, peccaries, and hippopotamus) do not ruminate. Kangaroos, sloths, and langur and colobid monkeys rely on an enlarged and generally compartmentalized stomach. Artiodactyls and perissodactyls largely evolved on separate continents (Africa/Eurasia and North America, respectively) with the global spread of grasslands. The current predominance of artiodactyls over perissodactyls suggests that foregut fermentation has advantages over hindgut fermentation (as does brain cooling; see 3.4.4, 4.2.1).

Ruminant mammals, the artiodactyls, have a characteristically large and multi-chambered stomach (Stevens & Hume 1995; Dijkstra et al. 2005; Figure 4.27). Ingested food is first physically processed in the oral cavity and swallowed into the largest stomach compartment, the rumen, where conditions (temperature, moisture) are ideal for microorganismal digestion (by bacteria, ciliates, flagellates, and fungal sporangia). Digesta are mixed in the rumen, and can be regurgitated into the oral cavity for rechewing; this process is termed rumination (Kennedy 2005). The rumen has a large surface area for absorption of microorganismal digestive products, primarily volatile fatty acids (VFAs; France & Dijkstra 2005). Digesta are then passed from the rumen into the reticulum, which is also a fermentation chamber but with a honeycomb lining; here, material is separated into coarse particles for return to the rumen, and fine particles to be passed to the omasum.

Figure 4.27 Schematic of the four-chambered stomach of a ruminant mammal (cow), showing the flow of digesta through the rumen (with internal mixing and regurgitation for rechewing), reticulum, omasum, and abomasum, thence to the small intestine.

Figure 4.28 Schematic of microorganismal fermentation of cellulose, starch, soluble sugars, and hemicellulose to pentoses and hexoses (e.g. glucose), then pyruvate, then microorganismal metabolic end products (volatile fatty acids, VFAs), and finally the combination of CO2 with H2 to form methane.

Modified from France and Dijkstra (2005).

(p.383) In the omasum, which has highly folded ‘leaves’ like book pages, digesta are further physically mixed and ground into smaller particles, before being passed to the abomasum, which is the ‘true stomach’ with the normal functions of the mammalian stomach, such as H+ secretion and protein digestion. The ruminant system converts food to short-chain VFAs, methane, ammonia, and sometimes lactic acid (Russell & Strobel 2005). For a ruminant mammal such as a sheep digesting alfalfa, about 41% of the ingested energy is lost as faeces; 5% is eructed as methane; 33% is converted to VFAs; 18% is converted to microorganismal growth in the stomach; 18% is absorbed in the abomasum and the small intestine; and 3% is lost as heat from microorganismal metabolism (Withers 1992). Cetaceans are phylogenetically closely associated with artiodactyls, as the Cetartiodactyla (Table 1.1; Appendix). Interestingly, although some cetaceans have a single-compartment stomach others have a stomach with two or three compartments; the forestomach houses bacteria responsible for chitin and wax digestion (Nordøy 1995; Olsen et al. 1999; Swaim et al. 2009). Stones in the forestomach might enable it to function as a grinding gizzard.

Pseudoruminant mammals include a variety of other mammals that have convergently evolved a similar fermentative digestive strategy using the stomach, but do not ruminate (Stevens & Hume 1995). Placental pseudoruminants include the non-ruminant artiodactyls (pigs, peccaries, hippopotamus), sloths, colobid monkeys, and langur monkeys. The domestic pig and warthog have a simple stomach, but some suids (e.g. babyrousa, Babyrousa celebensis) have a compartmentalized stomach. This hippopotamus has a very complex, compartmentalized stomach, but no caecum and a very short large intestine.

Sloths are arboreal folivores, and have a complex, three-compartment stomach: the first compartment is large, divided into two sacs; the second is small, with a (p.384) groove running to the third large and tubular compartment. Langur (Presbytis) and colobus (Colobus) monkeys also have a complex, three-compartment stomach. Macropod marsupials (e.g. Macropus, Potorous) are also pseudoruminants. They have a complex tubular stomach divided into three haustra. The forestomach has a sacciform anterior sac and a tubiform sac, whereas the hindstomach sac is the ‘true’ mammalian stomach, with the gastric and pyloric mucosa. Macropods chew their food more thoroughly than do ruminants, but they show a similar action of regurgitating and rechewing digesta, termed merycism (Barker et al. 1963; Hume 1982). Even some carnivorous marsupials undertake merycism (Archer 1974).

Sheep and kangaroos in Australia provide an informative comparison of a ruminant and a pseudoruminant mammal. Where sheep and red kangaroos coexist, kangaroos are more selective in their diet (Munn et al. 2009b). Sheep have a higher dry matter intake and a longer retention time than kangaroos (Hume 1982) but achieve fairly similar digestibilities (59 and 52%, respectively), despite their different fermentation strategies; sheep feed in short bouts whereas kangaroos have longer bouts (Munn et al. 2010b). Kangaroos can meet their daily energy requirements at lower dry matter intakes and higher fibre contents than sheep, reflecting (p.385) their lower energy metabolism (Munn et al. 2009b). Field metabolic rate (FMR) was 16,664 kJ day−1 for sheep $(mass=50.2kg)$ and 4,872 kJ day−1 for kangaroos $(mass=23.6kg)$; when corrected for mass differences, a kangaroo is energetically equivalent to 0.7 of a sheep (Munn et al. 2009b). Interestingly, sheep and kangaroos differ markedly in water requirements: red kangaroos use only 13% of the water that a sheep uses, and have more concentrated urine (1,852 compared to 1,000 mOsm kg−1). Since red kangaroos are not so dependent on water, their impact on rangelands is less and more broadly distributed.

For foregut fermenters, digesta encounter the fermenting microorganisms before the mammal’s ‘normal’ digestive tract. Therefore, one disadvantage of foregut fermentation is that the microorganisms not only hydrolyze the structural carbohydrates such as cellulose and hemicellulose to monosaccharides (e.g. glucose), but also anaerobically metabolize these monosaccharides as their energy source, producing VFAs as their metabolic waste product. Although these VFAs are an important energy source for the host, there is little glucose remaining for the host’s metabolism (Brockman 2005); for example, soluble sugars disappear from the forestomach of kangaroos, and there are low levels of disaccharases in the small intestine (Hume 1982). This low soluble carbohydrate availability is a serious physiological consequence because blood glucose concentration is low (compared to non-foregut fermenters), but glucose is essential for brain metabolism. To provide glucose to the body, the liver of ruminants and pseudoruminants synthesizes glucose from non-carbohydrate precursors, amino acids, by gluconeogenesis (Ballard et al. 1969). The liver of sheep and kangaroos releases glucose into the blood regardless of whether they have recently fed or are fasted, whereas the liver of a non-foregut fermenter, such as the dog, absorbs glucose after feeding but releases glucose when fasted.

An advantage of pre-peptic fermentation is that the rumen microorganisms can use the N waste of the host (urea) as a nitrogen source, thereby recycling N to the host through microorganismal protein synthesis (Nolan & Dobos 2005). In ruminants, rumen ammonia production can be 17–84% of the dietary N intake and the source for 69–80% of microorganismal protein; in macropods, ammonia is the N source for about 63% of microorganismal protein (Hume 1982). Urea is transferred to the rumen contents via saliva and diffusion from the blood. For kangaroos, the urea concentration of parotid and mandibular saliva (each produced at 0.05–4.5 ml min−1) is correlated with flow rate, from 1.04 to 0.92 (relative to plasma) for parotid, and 0.84 to 0.08 for mandibular salivary glands, respectively. Parotid saliva flow is continuous, and has been suggested to be the main route for salivary urea recycling to the stomach, whereas mandibular saliva is likely the primary source of water for evaporative cooling (e.g. via forelimb licking; Beal 1987). Another fermentative strategy of mammals is to use sites after the stomach as the fermentation chamber (i.e. the caecum, hindgut, or both). In general, small mammals (< 5 kg) use the caecum, whereas large species (> 50 kg) use the (p.386) hindgut; intermediate-size species use either or both (Hume 1982; Hume 1989; Stevens & Hume 1995; Gibson & Hume 2002). In general, hindgut fermenters have a relatively high passage time and a low digestibility; they achieve equivalent daily energy intakes as foregut fermenters by having a higher dry matter intake, particularly on poor-quality diets, so long as food is not limiting (Ilius & Gordon 1992). An advantage of hingut fermentation is that the digesta is subjected to normal digestive processes before fermentation so any digestible materials, such as glucose, are directly available to the mammal. However, hindgut fermenters suffer the disadvantage of having microorganismal fermentation located after the stomach and small intestine; the useful products of fermentation (VFAs, microorganismal production) are not subject to the normal mammalian stomach digestion and small intestine absorption functions.

Caecum fermenters include many rodents and lagomorphs (rabbits, hares, pika). The koala (Phascolarctos cinereus) has a relatively enormous caecum and hindgut, and many possums (e.g. brushtail, Trichosurus vulpecula) have a large caecum and hindgut. Coprophagy (or caecophagy) involves the elimination of specialized faecal pellets containing the product of microorganismal fermentation. These soft faeces, which are quite different from the normal harder and drier faecal pellets, are ingested to recycle the products of fermentation to the stomach for digestion, rather than simply eliminating these nutrient-rich materials, thereby overcoming the nutritional problems of having fermentation occur after the stomach (Figure 4.29). Coprophagy is best known for lagomorphs (rabbits, hares, pikas), but also occurs in rodents, shrews, and a folivorous prosimian (Lepilemur; Kenagy & Hoyt 1980). Coprophagy has been reported for ringtail possums and young koalas, presumably for inoculation with symbionts as well as nutrition. For the chisel-toothed kangaroo rat (Dipodomys microps), about one-quarter of the faecal pellets are of caecal origin, and are ingested over about 8 h during the day when not foraging (Kenagy & Hoyt 1980); their caecal pellets have more nitrogen and water than the normal faecal pellets. The herbivorous California vole Microtus californicus also ingests about one-quarter of its faeces, but in one-to-several-hour bouts, reflecting their continuous day and night foraging pattern. The diurnal degu rodent (Octodon degu) consumes its caecal pellets primarily at night (Kenagy et al. 1999).

Figure 4.29 Schematic of the fates of ingested food and reingested faeces for a post-gastric fermenter with coprophagy (rabbit). Gut schematic

A final aspect of microorganismal-based fermentation as an herbivorous strategy for mammals (and other animals) that has global significance is the incidental production of hydrogen, and particularly methane, by fermentation. Methane (CH4) is an important greenhouse gas, and ruminant livestock are the largest single source of methane, about 28% of global production (Klieve 2009). Archaean microbacteria in the rumen and forestomach of other fermentative herbivores, such as kangaroos and sloths, reduce H2 from fermentation to CH4; this reaction is the main sink for H2 in the stomach and is of considerable importance, as H2 must be removed from the rumen for continued efficient fermentation. Methane (p.387) is eliminated via the mouth by eructation. Methane production of ruminants and equids increases isometrically with mass ( and 0.18 M0.97, respectively), compared to M0.75 for metabolic rate, so methanogenic energy loss increases with mass and could limit the maximal mass of ruminants but not equids (Franz et al. 2010; Vendl et al. 2015a; Figure 4.30). Methane production of various non-ruminant herbivores scales the same as for equids (; Franz et al. 2011), but domesticated pigs are lower (0.07 M0.99; Franz et al. 2010). Kangaroos are similar to or lower than non-ruminants (Vendl et al. 2015a); sloths are slightly higher (Vendl et al. 2015b), as are peccary and pygmy hippopotamus (Vendl et al. 2016). Although informative, these allometries of absolute methane production need to be carefully interpreted in terms of methanogenesis per metabolic activity of the herbivore. Methane production of macropod marsupials increases compared to placental mammals when expressed per dry matter intake or metabolic rate, generally falling between that of ruminant and non-ruminant placentals, and especially increases for sloths, to about that of ruminants (Vendl et al. 2015a). Despite the independent evolution of foregut fermentation in various mammals (and birds; e.g. the hoatzin), the convergent role of methanogenesis has resulted in convergence of the microbial communities in these (p.388) phylogenetically distant herbivores, even to the level of cow and hoatzin rumen/crop communities being more similar than their respective colon/ceca communities (Godoy-Vitorino et al. 2012).

Figure 4.30 Methane production of herbivorous mammals: ruminants, macropods, sloths, hyrax, rabbit, guinea pigs, peccary, camelids, elephant, equids, and non-ruminants.

Data from Franz et al. (2010, 2011), Vendl et al. (2015a,b), and Vendl et al. (2016).

Despite microorganismal-assisted digestion, digestibility of plant matter is generally relatively low compared to other diets. Digestive efficiency (e.g. percentage dry matter digestion) varies with digestive strategy (which varies with body mass) and diet, rather than with the phylogenetic affiliation of species (Table 4.8). As a consequence, digestibility is higher for small herbivores and ruminants than for large hindgut fermenters.

Table 4.8 Comparison of dry matter intake and digestibility for various ruminants differing in body mass, diet, and digestive strategy.

Species

Mass (kg)

Diet

DMI (kg d–1)

DM digestibility (%)

Pika (Ochotona princeps)1

0.14

High tannin leaves + pelleted food

0.052

60

Caecum fermentation; coprophagic; high tannin diet

Brushtail possum (Trichosurus vulpecula)2

2.49

Eucalyptus leaves

0.074

51

Caecum + hindgut fermentation; high phenolic diet

Koala (Phascolarctos cinereus)3

6.56

Eucalyptus leaves

0.17

54

Caecum + hindgut fermentation; high phenolic diet

Reedbuck (Redunca redunca)4

14

Grass mixture

0.24

49

Ruminant fermentation

Grant’s gazelle (Nanger granti)5

80

Free-ranging

2.5

61

Ruminant fermentation

Cape Buffalo (Syncerus caffer)4

235

Grass mixture

7.2

46

Ruminant fermentation

Zebra (Equus burchellii)5

270

Free-ranging

7.2

42

Hindgut fermentation

Hippopotamus (Hippopotamus amphibius)4

1,200

Elephant grass

≈ 8

68

Non-ruminant artiodactyl

Asian elephant (Elephas maximus)6

2,427

Hay + pelleted food

37.9

36

Hindgut fermentation

African elephant (Loxodonta africana)7

≈ 4,578

Free-ranging

54–67

30–45

Hindgut fermentation

There is a general relationship between the body mass of herbivores and the quality of food consumed (as well as location of the fermentation chamber; see earlier). For example, amongst macropod marsupials, the small musky rat-kangaroo (Hypsiprymnodon moschatus; 1.8 kg) consumes fruit and insects; rat-kangaroos (Potorous, Bettongia; 3.8 kg) consume fungi, tubers, and insects; tree kangaroos (Dendrolagus; 11 kg) consume leaves and fruit; swamp wallabies (Wallabia bicolor; 14 kg) browse on shrubs; wallaroos (Macropus robustus; 22 kg) are mixed browsers/grazers; and large kangaroos (Macropus spp; 25–57 kg) graze on grass and herbs (Hume 2014). Rock wallabies (Petrogale; 8.6 kg) and hare-wallabies (Lagorchestes, Lagostrophus; 3 kg) are unexpectedly small grazers.

Similar patterns are evident for some assemblages of large African herbivores, but this simple pattern is not universal and is ‘blurred’ by differences in species’ bauplans (Clauss et al. 2013). For example, large herbivores of Kruger National Park vary markedly in diet, as evidenced by the percentage of C4 plants in their diet (Codron et al. 2006), which is related to their δ‎C13 isotope ratio (see 5.7.2; Figure 4.31). The percentage (%) C4 plants and ‰ δ‎C13 are remarkably consistent over geography (north and south of the park) and seasons (wet and dry season) for (p.389) (p.390) large browsers (about 8% and −27‰) and large grazers (about 93% and −14‰). A mixed browser/grazer (impala; Aepyceros melampus) has intermediate %C4 and ‰ δ‎C13, which is lower in the dry season (and lower in the north of the park) than in the wet season (with no geographic difference). However, the elephant, a very large herbivore, is a browser in the south of the park in the dry season, and a mixed browser/grazer in the north; in the wet season, they are even more mixed browser/grazer, with no geographic difference.

Figure 4.31 Dietary percentage of C4 plants and ‰ δ‎C13 content of the diet for African browsers (giraffe, Giraffa camelopardalis; greater kudu, Tragelaphus strepsiceros), grazers (Burchell’s zebra, Equus birchellii; African buffalo, Syncerus caffer; blue wildebeest, Connochaetes taurinus), and mixed feeders (impala, Aepyceros melampus; elephant, Loxodonta africana), in the north and south of Kruger National Park, in the wet and dry seasons.

Data from Codron et al. (2006).

Large body size has been considered to offer nutritional advantages to mammals, because of their lower mass-specific metabolic rates, relatively large gut volume, and longer passage times, but there is a limit to increasing body mass for optimization of fermentative digestion (Clauss et al. 2003b, 2013; Barboza et al. 2009). Large foregut fermenters are designed to slow the passage of ingesta and maximize physical and chemical digestion and fermentation, but this inherently limits their passage rate, hence amount of material potentially available for digestion (e.g. the hippopotamus). In contrast, large hindgut fermenters have a relatively fast passage time and low digestibility and are more adaptable to large body mass. This hypothesis is supported by the fossil record, which suggests that fossil foregut fermenters did not generally exceed the highest body mass of extant foregut fermenters, and the very large fossil mammalian herbivores are thought to have been hindgut fermenters. On various continents, there was an increase in maximum mass after the Cretaceous/Paleogene boundary (66 MYBP) that levelled off (p.391) at about 10,000 kg, around 40 MYBP (Smith et al. 2010); these largest terrestrial mammals included hindgut-fermenting perissodactyls (Indricotherium) and proboscideans (Deinotherium). While digestive constraints, such as the allometry of mean retention time, might explain the body mass limit for herbivory, they cannot alone be considered the evolutionary constraint on body mass—other factors such as fasting endurance or running speed may be involved (Clauss et al. 2007, 2013).

There are also various ecological consequences of herbivory, including how herbivores overcome plant defences. The evolutionary ‘arms race’ between plants and herbivores, like that between prey and predators, has resulted in continuing adaptation and counter-adaptation of physical and chemical ‘weapons’. Herbivorous mammals also affect plant diversity: regions where mammal herbivory is more important than insect herbivory tend to have lower plant diversity (Becerra 2015).

Many plants have physical defences against mammalian herbivores, such as spines, stinging hairs, or highly sclerotized leaves. Although spines are effective against some herbivores, others have responded with adaptations to circumvent the deterrence of spines. For example, the white-throated woodrat (Neotoma albigula) is an Opuntia cactus specialist; it is adept at clipping off the spines before transporting the cactus to its den for consumption, and actually prefers to collect spiny cacti over (artificially) de-spined cacti (Kohl et al. 2014).

Many plants use chemical defences against herbivory, in the form of plant secondary metabolites (PSMs). PSMs are an incredibly diverse assemblage of plant metabolites that are not just waste products, but also defend plants against herbivory and abiotic stress, and they are major contributors to community dynamics with effects that cascade through ecosystems (Iason et al. 2012). Many PSMs have marked effects on diet selection, digestion, and reproductive success of herbivorous mammals through aversive, detrimental, or toxic effects (DeGabriel et al. 2014; Moore et al. 2015). PSMs can impart a noxious taste to foliage; for example, many tannins are very astringent and potentially deter herbivores from their consumption. Some PSMs have adverse effects, such as reducing digestion of specific dietary components; for example, tannins bind to dietary and endogenous proteins and inhibit digestive enzymes (e.g. α‎-amylase and α‎-glucoamylase; Barrett et al. 2013). Nevertheless, many herbivorous mammals ingest PSMs, often in sufficient amounts that they have marked physiological effects.

Phenolics are a common toxic PSM; for example, eucalypt leaves (Eucalyptus punctata) have a phenolic content of about 28% dry weight (Cork et al. 1983); E. urograndis leaves contain about 0.04% hydroxybenzoic acid (a derivative of benzoic acid; Chapuis-Lardy et al. 2002). Terpines are a diverse group of volatile unsaturated hydrocarbons found in plant essential oils. For example, E. camaldulensis leaves contain about 1.1% dry weight cineole, a terpinoid oxide, and 2.05% total terpinoids (Stone & Bacon 1994).

As many PSMs have toxic effects on their consumers, detoxification is the consumers’ counter-response. Detoxification is a series of biochemical reactions that (p.392) convert the PSM into a suitable form for excretion. Herbivores can either hydrolyze, oxidize, or reduce the PSM for excretion (Phase I reactions), or conjugate the PSM with glucuronic acid, glycine, or sulphate for excretion (Phase II reactions), or both (Marsh et al. 2006). Phenolics are detoxified by conjugation with glucuronic acid for urinary excretion; cineole is detoxified by oxidation and conjugation with glucuronic acid; benzoic acid is mainly conjugated with glycine (Marsh et al. 2006). Glucuronic acid excretion accounts for about 40% of urinary energy loss of woodrats consuming creosote (Larrea tridentata) resin (about 1.9% of total metabolizable energy; Mangione et al. 2004). For brushtail possums (Trichosurus vulpecula), consumption of diets high in benzoate incurred a considerable cost in protein, about 30% of the daily N intake (Au et al. 2013).

Some mammals subsist entirely on the highly defended leaves of plants (e.g. the koala consumes only Eucalyptus leaves). Their stomach is small, as is their small intestine, but their colon and caecum are very long and wide (the koala has the longest caecum relative to body size of any mammal). The microorganisms present in a koala’s hindgut seem to detoxify essential Eucalyptus oils (Eberhard et al. 1975), and glucuronic acid excretion accounts for more than 7% of their urinary energy excretion (but only about 1% of the total metabolizable energy; Cork et al. 1983).