Correlations between structure and function in the design of the bat lung: a morphometric study

Bats are unique among mammals in their capacity for continuous flapping flight (Greenhall & Paradiso, 1968; Wimsatt, 1970; Thomas & Suthers, 1972; Dawson, 1975; Yalden & Morris, 1975; Thomas, 1975, 1980; Walker, 1975; Jürgens, Bartels & Bartels, 1981). Bats have thus been able to exploit widely dispersed ecological niches, and consequently show a remarkable adaptive radiation giving rise to about 1000 out of the 4200 mammalian species (Wimsatt, 1970; Thomas & Suthers, 1972; Yalden & Morris, 1975). Windtunnel experiments (Tucker, 1972; Carpenter, 1975; Thomas, 1975) have shown that flapping flight is energetically a very demanding form of exercise, being substantially beyond the energetic capacity of terrestrial mammals of similar size during maximum exercise. For example, a budgerigar flying horizontally at its most economical speed increases its oxygen consumption to about 13 times its standard metabolic rate, which is about 1·5 times the oxygen consumption of a mouse running hard in an exercise wheel (Tucker, 1968a,b). In turbulent air or when ascending the bird increases its oxygen consumption for short periods by about 20–30 times, whereas even a good human athlete can achieve for a few minutes an increase of only 15–20 times (Tucker, 1969). Bats have been shown to increase their oxygen consumption during sustained flight by at least 20 times (Bartholomew, Leitner & Nelson, 1964; Carpenter, 1975). In general, the energetics of bat flight appear to be comparable to that of birds (Studier & Howell, 1969; Thomas & Suthers, 1972; Carpenter, 1975; Dawson, 1975; Thomas, 1975; Jürgens et al. 1981).

The general anatomy of bats is, of course, fundamentally similar to that of terrestrial mammals (Yalden & Morris, 1975), and in at least one species the general microscopic structure of the pulmonary parenchyma is known to resemble that of other mammals (aina, King & King, 1982b). In contrast, birds have evolved voluminous air sacs which continuously and undirectionally ventilate the compact, rather rigid lung (for review of anatomy see King, 1966; Duncker, 1971; for physiology see Schmidt-Nielsen, 1975; Scheid, 1979). Moreover the blood-gas relationship in the avian parabronchus is cross-current as opposed to the uniform pool in the pulmonary alveoli of bats and all other mammals (Scheid & Piiper, 1970; Piiper & Scheid, 1973). The cross-current design is functionally superior to the uniform pool, since a higher degree of arterialization can be attained for an equal degree of ventilation (Scheid & Piiper, 1970; Schmidt-Nielsen, 1975; Piiper & Scheid, 1973). Furthermore, examination of the structure of the exchange tissue of the avian lung by morphometric methods indicates that pulmonary parameters such as the thinness of the blood-gas barrier and the surface area of the barrier per gram body weight are superior in birds as compared with mammals generally, leading to a higher anatomical diffusing capacity for oxygen per gram body weight in birds than in mammals (Maina & Settle, 1982).

Despite having an essentially mammalian type of lung, bats nevertheless appear to have an exercise capacity as good as that of birds. The question then arises, might the bat lung possess subtle structural adaptations such as can only be disclosed by morphometric methods? This conclusion was indeed reached by Maina & Nicholson (1982) and Maina et al. (19826) from quantitative observations on the lung of a fruit that (Epomophorus wahlbergi). However, this is the only exhaustive published morphometric study of the bat lung and was limited to a single species.

The present investigation expands our previous work on Epomophorus by extending morphometric techniques to the lungs of five additional bat species of widely differing body weights and modes of life. Comparisons are made with corresponding data from the lungs of birds and terrestrial mammals.

The lungs of 24 wild specimens of adult bats of mixed sexes from five species have been analysed by morphometric techniques. The bats were killed soon after capture by an intraperitoneal injection with barbiturate, and weighed at once. The subsequent procedure for fixation, processing and analysis of the lungs was essentially similar to that adopted by Maina et al. (1982b), but the main points were as follows.

Immediately after death an incision was made just caudal to the xiphisternum and the diaphragm carefully perforated (avoiding puncturing the lungs) on both sides of the mediastinum; this created a pneumothorax which caused the lungs to collapse. The trachea was then cannulated and the lungs fixed in situ by intratracheal infusion with 2·3% glutaraldehyde buffered in sodium cacodylate (pH 7·4, total osmolarity 350 mosmol l⁻¹). The infusion was done by gravity from a height of 25 cm above the supine body. When the fixative stopped flowing the trachea was ligated caudal to the cannula. The volumes of the left and right lungs were estimated by the water displacement method of Scherle (1970).

One complete transverse slice was taken from each of the lobes of the lung and diced into small pieces (about 2 mm³) for electron microscopy. These pieces were postfixed in 2% osmium tetroxide, block stained in 2% uranyl acetate in maleic acid, and dehydrated in graded ethanol (starting at 50 %) and acetone before infiltration and embedding in Taab resin. Two blocks were taken at random from a group derived from each of the lobes and trimmed to eliminate non-parenchymal tissue; ultrathin sections were cut, counterstained with lead citrate, and examined electron microscopically. Five electron micrographs (from each block) were taken from predetermined corners of the grid squares (to avoid bias) of the first technically adequate section at a primary magnification of × 3000, and the subsequent analysis was performed at about × 7500 with a superimposed quadratic lattice grid. This magnification was found to be optimal as it gave a relatively large field but still enabled the requisite parenchymal components to be identified and quantified. A total of 50 micrographs was analysed for each animal. Point counting was employed to estimate volumes, intersection counting to estimate surface areas, and intercept length measurement to estimate harmonic mean thicknesses (Weibel, 1970/71). The arithmetic mean thickness of the tissue barrier was estimated by a random short line test grid (Weibel & Knight, 1964). The surface density of the tissue barrier was calculated from its surface area and parenchymal volume. The surface area of the plasma layer was estimated as the mean of the surface area of the capillary endothelium (Sc) and that of the erythrocytes (Se) (Weibel, 1970/71). In case of an overlap of erythrocytes only the intersections of the test grid with the outer surface of the erythrocytes (the surface directly adjacent to the endothelium) were considered, this avoids an overestimate of Se and consequently Sp. As plasma is more of a layer than a surface it was thought reasonable to estimate its surface area (Sp) from the areas of its two boundaries, the endothelium and erythrocytes. The morphometric pulmonary diffusing capacities (conductances) for oxygen of the tissue barrier ⁠, plasma ⁠, and erythrocyte were calculated using the physical coefficients for oxygen permeation of the tissue barrier and plasma layer ⁠, and the coefficient for oxygen uptake by whole blood ⁠, as estimated for mammalian tissue and cited by (Weibel, 1970/71). The value of was adjusted for a venous haematocrit of 60 % reported in a bat by Thomas & Suthers (1972). The membrane diffusing capacity for oxygen and the total morphometric pulmonary diffusing capacity were calculated from the individual serial conductances, the model used being that developed by Weibel (1970/71).

The remaining parts of the lobes were cut in 7 µm serial transverse paraffin sections. Three sections were taken from each lobe at predetermined equidistant intervals, stained with haematoxylin and eosin and analysed for the volume density of the parenchyma (tissue involved in gas exchange) and of the non-parenchyma by point counting using a 100 point Zeiss integrating graticule at × 200 magnification. The absolute volumes of these two components could then be calculated from the volume of the lungs.

The results are summarized in Tables 2–8. In Cheiromeles torquatus and Cynopterus brachyotis the alveoli constituted 91% and 92% respectively of the parenchyma, while the lowest value (81 %) was observed in Pipistrellus pipistrellus, the mean for all five species being 87 %. The mean volume densities of the pulmonary blood capillaries and the tissue of the interalveolar septa in the five species of bat were 7% and 6% respectively. The components of the parenchyma, and the blood gas barrier are shown in Fig. 1.

The surface areas of the resistance barriers constituting the air haemoglobin pathway, namely the blood–gas (tissue) barriers (St), alveolar epithelium (Sa), capillary endothelium (Sc), and the erythrocytes (Se), are shown in Table 4. St is functionally a more meaningful parameter than Sa as it takes into account only those areas of the alveolar epithelium where gas exchange actually takes place. It was felt that St, which can be directly determined, was a much more accurate estimate of the morphometric pulmonary diffusing capacity rather than estimating St as the mean of Sa and Sc (Weibel, 1970/71). In this study the values of St, Sa and Sc are given in Table 5 and the difference in estimating St both ways is clearly apparent. It consistently appears in this study, and our previous ones, that the indirect way of estimating St leads to an overestimate of the surface area available for gas exchange and consequently that of the diffusing capacity. Obviously Sa will always be higher than St when these values are estimated together on the same preparations.

Ratios of some parameters of the components of the parenchyma are shown in Table 6. Pipistrellus pipistrellus and Tadarida mops had the highest values (respectively 63 and 56 cm² g⁻¹) for the surface area of the blood-gas (tissue) barrier per unit body weight (St/W); the lowest values were observed in Cheiromeles torquatus (33-3cm²g⁻¹) and Cynopterus brachyotis (30cm²g⁻¹). The ratio S_v(t,p) expresses the surface density of the blood-gas (tissue) barrier and thus indicates the relative alveolar diameter (see Discussion). The highest value of S_v(t,p) was observed in a specimen of Pipistrellus pipistrellus (153 mm² mm⁻³), and the lowest (31 mm² mm⁻³) in a specimen of Cynopterus brachyotis. The ratio Vc/Sa has been defined as capillairy loading (Perry, 1978; Gehr et al. 1980) and expresses the degree of exposure of blood to air, a lower ratio indicating a higher degree of exposure (Perry, 1978). The lowest value of Vc/Sa was found in Pipistrellus pipistrellus (mean 0·32 cm³ m⁻²). Relatively high values of Vc/Sa (means 0·80 and 0·71 cm³m⁻²) were observed in Tadarida mops and Miniopterus minor respectively.

Table 6 gives the harmonic and arithmetic mean thicknesses of the blood-gas (tissue) barrier and the plasma layer. The thinnest harmonic mean thickness of the tissue barrier, τht, (0·184µm) was found in a specimen of Pipistrellus pipistrellus ; the thickest barrier (0·313µm) was found in a specimen of Cynopterus brachyotis. The lowest and the highest values of the arithmetic mean thickness of the blood-gas (tissue) barrier (1·12 and 1·81 µm) were again observed respectively in a specimen of Pipistrellus pipistrellus and a specimen of Cynopterus brachyotis.

In Table 7 are shown the morphometric pulmonary diffusing capacities of the three barriers constituting the air-haemoglobin pathway; also shown are the membrane diffusing capacity and the total morphometric pulmonary diffusing capacity. The highest mean values of the specific (weight normalized) total morphometric pulmonary diffusing capacity, (mlO₂min⁻¹ mmHg⁻¹ kg⁻¹), were found in Pipistrellus pipistrellus and Tadaridamops (11·53 and 10·2 mlO₂ min⁻¹ mmHg⁻¹ kg⁻¹ respectively). The lowest mean values of were found in Cheiromeles torquatus (4·18) and Cynopterus brachyotis (4·19mlO₂min⁻¹ mmHg⁻¹ kg⁻¹).

The structure of the lung is adapted to meet the oxygen demands of an animal, which in turn will reflect various factors such as body weight and mode of life. Bats are excellent fliers in terms of speed, distance and manoeuvrability (Hartman, 1963; Krzanowski, 1964; Griffin, 1970; Vaughan, 1970; Schmidt-Nielsen, 1975, 1979; Yalden & Morris, 1975; Walker, 1975; Norberg, 1976; Snyder, 1976; Bouverot, 1978). It has been shown that bats have developed specializations of the oxygen transport system in the form of a relatively large heart, high haematocrit, high haemoglobin concentration, a high oxygen capacity and low oxygen affinity of the blood (Hartman, 1963; Thomas & Suthers, 1972; Snyder, 1976; Jürgens et al. 1981). However, Snyder (1976) observed that the haematological parameters of Cynopterus brachyotis are within the range of the values reported for non-flying mammals and even lower than those of the laboratory mouse; he nevertheless observed that this species of bat had a heart weight 60% greater than that expected of a non-flying mammal and comparable to that of a bird. In the present study it will be shown that the bat lung is morphologically well adapted for gas exchange ; furthermore there are indications that individual species seem to be adapted according to their small or large size and their energetic or non-energetic life style.

Jürgens et al. (1981), on the basis of measurements of lung weight, concluded that bats had proportionately larger lungs than small terrestrial mammals. Our morphometric observations confirm that bats do indeed have immensely large lungs compared with those of both birds and terrestrial mammals (Fig. 2, Table 5). The mean lung volume in the five species of bats examined here per gram body weight (VL/W) was 0·074 cm³ g⁻¹. In contrast, the mean VL/W for eight species of shrews examined by Gehret al. (1980) was only 0·030 cm³ g⁻¹; the VL/W in the violet-eared hummingbird was 0·043 cm³g⁻¹ (Dubach, 1981). Thus, the specific lung volume (i.e. volume per gram body weight) of our bats was 2·5 times greater than that of the shrews and almost twice that of the hummingbird. Even more remarkable was the fruit be Epomophorus uahlbergi, which had a specific lung volume of 0·13 cm³g⁻¹ (Maina al. 19826), i.e. about 4·3 times greater than the shrews.

Tenney & Remmers (1963), however, in their plot of lung volume against body weight in a large number of mammalian species, included a value for an unnamed species of bat which fell slightly below the common regression line. This discrepancy could possibly be due to species differences or to the technique used for fixing and measuring lung volume. Tenney & Remmers removed the lungs immediately after death and kept them inflated at a pressure of 20 cm of water until they were dry. Unfortunately this technique gives no assurance that the volume of the dried lungs was similar to their volume in situ. Comparison of their values with those obtained here is therefore inconclusive.

Three of our five species of bats exhibited relatively high values of the surface area of the blood-gas (tissue) barrier per gram body weight (St/W). Thus the mean values of St/W obtained in Pipistrellus pipistrellus, Miniopterus minor and Tadarida mops were respectively 63, 50 and 56 cm² g⁻¹. These values are substantially higher than the values of 26, 26, 43, 29 and 46cm²g⁻¹ obtained by Maina (1982) and Maina & King (1982) in anseriform, charadriiform, columbiform, piciform and passeriform avian species respectively. They are also much higher than the mean value of 33cm²g⁻¹ reported by Gehr et al. (1980) for eight species of shrew. On the other hand, the relatively low values of St/W in Cynopterus brachyotis (30cm²g⁻¹) and Cheiromeles torquatus (33 cm²g⁻¹) are closely similar to the mean values for the shrews (Gehr et al. 1980; Weibel, 1979). The value of St/W found in the violet-eared hummingbird was 87cm²g⁻¹ (Dubach, 1981), and this exceeds all of the five bat species in the present study. However, the highest values of St/W reported so far in any vertebrate (138cm²g⁻¹) occurred in the fruit-bat Epomophorus uuahlbergi (Maina et al. 19826). The plot (Fig. 3) suggests that bats may prove to have a more extensive blood-gas (tissue) barrier than birds in general or terrestrial mammals of comparable size, though undoubtedly examination of more species of bats is called for.

The relatively high values of St/W in Tadarida mops could be associated with the capacity for fast and sustained flight in molossid bats (Vaughan, 1966, 1970; Yalden & Morris, 1975). The values of St/W in Pipistrellus pipistrellus are higher than those of Tadarida though Vaughan (1966) observed that vespertilionid bats exhibit slower flight close to the ground. The insectivorous bats like Pipistrellus pipistrellus have, however, to be more manoeuvrable and fly continuously in search of food (Vaughan, 1970). Pipistrellus pipistrellus is one of the smallest bats (Yalden & Morris, 1975). This, and the need for sustained flight when feeding, may account for the high value of St/W in this species of bat.

Tenney & Remmers (1963) reported the value of the alveolar surface area in an unnamed species of bat to be similar to that of terrestrial mammals. To estimate alveolar surface areas they used relatively thick sections 100 µm, which were photographed at low power (presumably using a light microscope). Analysis of tissues at low magnifications gives an underestimate due to lack of resolution (Dunnill, 1962). The value for the bat obtained by Tenney & Remmers (1963) is not therefore comparable with our study.

The surface area of the blood-gas (tissue) barrier per unit volume of parenchyma, S_v(t,p), shows how much of the barrier has been packed into a unit containing volume. A large diffusing surface area can be achieved by an increase in either the lung volume, or the lung partitioning, or both (Tenney & Remmers, 1963; Geelhaar & Weibel, 1971 ; Weibel, 1979). The value S_v(t,p) is an indicator of the relative sizes of the terminal gas exchange units. The values of S_v(t,p) in mammalian lungs are in general relatively lower than those of the avian species (Fig. 4). For instance the mean value of S_v(t,p) in the eight species of shrew examined by Gehr et al. (1980) was 121 mm² mm⁻³ ; this value was the same as that found in the lung of the metabolicals active Japanese waltzing mouse (Mus wagneri) by Geelhaar & Weibel (1971). In birds, the value of Sv(t,p) in Gallus domesticus was 179·5 mm²mm⁻³ (Abdalla et al. 1982). The hummingbird Coribri coruscans has a value as high as 389 mm² mm⁻³ (Dubach, 1981). The air capillaries of the avian lung are extremely small, being only 3–10 µm in diameter (Duncker, 1972), whereas the diameters of mammalian alveoli range from about 28 µm in a bat to about 1800 pm in the dugong (Tenney & Remmers, 1963). Maina et al. (19826) suggested that on the basis of data horn Epomophorus, of which the mean value of S_v(t,p) was 121 mm² mm⁻³, bats may generally have relatively small alveoli. Although this generalization agrees with the observations of Tenney & Remmers (1963), it is not upheld by the values of S_v(t,p) obtained from the larger population of bats examined here (Fig. 4). Thus, althoughPipistrellus had a value of 125 mm² mm⁻³, the mean for all five species was only 78 mm² mm⁻³. This aspect needs to be investigated further, either by direct measurements of alveolar diameters or estimation of S_v(t,p) in more species of bat.

This parameter (τht) is the most appropriate estimator of the barrier conductance to oxygen (Weibel & Knight, 1964; Weibel, 1970/71, 1973; Hughes, 1980). The thinner the barrier, the greater will be its conductance to oxygen. The smallest Tht hitherto reported in a mammalian lung was 0·230µm in a specimen of the shrew (Suncus etruscus) by Gehr et al. (1980).

In the bats examined here the smallest τht (0·184 µm) was observed in one of the specimens of Pipistrellus pipistrellus, this value being the thinnest barrier reported in any mammalian lung so far. It is apparent from Fig. 5 that the bats have a thicker τht than the active birds, but a thinner τht than the terrestrial mammals. The τht in the lungs of flightless species of birds was generally thicker than that of the bat. Thus, τht in Gallus domesticus was 0·314 pµm (Abdalla et al. 1982), 0·320 µm in Numida meleagris (Abdalla & Maina, 1981), and 0·385 µm in Meleagris gallopavo (Dubach, 1981).

The volume of the pulmonary capillary blood is an important parameter since it influences the total morphometric pulmonary diffusing capacity for oxygen. Fig. 6 indicates that a bat does have a larger amount of blood in the gas exchange tissue (parenchyma) than a terrestrial mammal of the same body weight. However, the volumes of the pulmonary capillary blood in the bat and bird appear to be much the same. Duncker (1973) remarked that the relatively large volume of blood in the avian lung may be a possible reason for its high degree of efficiency; such an observation may also apply to the bat lung.

The total morphometric pulmonary diffusing capacity for oxygen estimates the maximum possible conductance in the lung under perfect conditions of ventilation and perfusion over the entire barrier (Weibel, 1970/71; Siegwart, Gehr, Gil & Weibel, 1971), although such conditions are seldom realised even during strenuous exercise. The physiologically-estimated pulmonary diffusing capacity is thus always lower than that estimated morphometrically. Apparently the physiological pulmonary diffusing capacity has not been estimated for any bat.

In the six species of bat which have so far been examined, the highest value of (20mlO₂min⁻¹ mmHg⁻¹ kg⁻¹) was found in Epomophorus (Maina et al. 1982b). In the species investigated in the present study, the highest value was found in a specimen of Pipistrellus pipistrellus (13 mlO₂min⁻¹ mmHg⁻¹ kg⁻¹). Relatively high values of were also found in Tadarida mops, the mean for this species being 10 mlO₂min⁻¹ mmHg⁻¹ kg⁻¹. The lowest values of in our five species (mean 4·2mlO₂min⁻¹ mmHg⁻¹ kg⁻¹) were found both in Cynopterus brachyotis and Cheiromeles torquatus. However, the values of even in Cynopterus and Cheiromeles are slightly higher than the mean value of 4mlO₂min⁻¹ mmHg⁻¹ kg⁻¹ in the eight species of shrew examined by Gehr et al. (1980). Fig. 7 clearly shows that the of the bats so far examined is higher than that of both birds and terrestrial mammals. This is mainly as a result of the extensive and thin blood–gas (tissue) barrier, and the higher pulmonary capillary blood volume in the bats. When these morphological features of the bat lung are combined with the haematological adaptations which have been found in several species of bat, they appear to constitute effective adaptations for transferring the large amount of oxygen required by flight.

The molossid bats tend to be regarded as the most advanced bats (Griffin, 1970; Yalden & Morris, 1975). These insectivorous bats are speedy and enduring flyers, a chiropteran version of swallows and swifts (Vaughan, 1970). It might be expected that the energetic demands of this mode of life would entail an advanced level of respiratory specialization. Among our bats, the molossid Tadarida mops did have a relatively high morphometric diffusing capacity for oxygen. On the other hand the lowest value occurred in Cheiromeles torquatus, another molossid species. Among the six species of bat which we have now examined, by far the most outstanding respiratory adaptations have been found in the fruit bat Epomophorus wahlbergi. The big fruit bats have often been regarded as relatively ‘primitive’. However, the body conformation of Epomophorus has been adapted to house lungs which are extraordinarily large and have a most extensive area for gas exchange, features which culminate in a morphometric diffusing capacity for oxygen of exceptional magnitude; on the other hand, Cynopterus brachyotis had a very low ⁠. Both of these pteropodid species are said to travel long distances to and from their feeding sites (Yalden & Morris, 1975), but Cynopterus seems to be entirely incapable of hovering (Snyder, 1976). The haematological parameters for Cynopterus differ very little from those of a terrestrial mammal (Snyder, 1976). Jepsen (1970) cast doubt on the concept of ‘primitiveness’, which includes being ‘less highly organized anatomically’, and noted that the debate ends to be ‘short on information’. We hope that data on the respiratory characteristics of bat species may broaden the basis for discussion.

It is notable that the volume proportions of the parenchyma and non-parenchyma in the bats which we have investigated are fairly constant and are similar to those of the lungs of the terrestrial mammals (Burri & Weibel, 1971 ; Gehr & Erni, 1980). The volume proportions of the main components of the parenchyma (alveoli, blood capillaries and the tissue of the interalveolar septa) in the bat lung are also similar to those reported for terrestrial mammals (Gehr, Bachofen & Weibel, 1978 ; Gehr et al. 1980; Gehr & Erni, 1980). In birds it has been found (Maina, 1982; Maina, Abdalla & King, 1982a) that the volume proportion of the exchange tissue is higher in the more active species of bird. The results of the present study of the bat lung are consistent with the suggestion by Maina et al. (19826) that these basic proportions of the mammalian lung have become uniformly optimized in the course of evolution, whereas the avian lung has been sufficiently adaptable to permit even these basic proportions to be varied. Both orders possess enough flexibility for detailed refinements of the exchange surface. But for bats, confronted by immense energetic demands imposed by flight, this alone must have been insufficient: their sole remaining resource was greatly to increase the size of the lung as a whole.

The authors are indebted to the British Council for financial support which made our collaboration possible. We gratefully acknowledge the help received from many colleagues. In Kenya: Professor F. A. Mutere, Kenyatta University College, Nairobi, in the course of writing this paper; Mrs G. Hinga and Miss B. Waweru for skilfully typing the manuscript. In the Universiti Pertanian Malaysia: Dr R. Stuebing, Dr Bertha Allison, Dr Lloyd Whitten and Dr M. Vidyadaran, for invaluable aid with material. In the University of Aberdeen: Dr Paul Racey, for generous help with material and background information. In the University of Liverpool: D. Zoe King for assistance with material in Malaysia, electron microscopy and criticism of the manuscript; Julie Henry for extensive and expert electron microscopy.