Low turnover rates of carbon isotopes in tissues of two nectar-feeding bat species

Stable isotopes of carbon occur at varying ratios as a consequence of the specific enzymatic route of CO₂-fixation (C₃ or C₄/CAM) in plants and in animals according to their diet (DeNiro and Epstein, 1978, 1981). These observations are the basis for studies on dietary preferences (e.g. Ben-David et al., 1997; Hobson et al., 2000) and migratory movements of animals (e.g. Fleming et al., 1993; Marra et al., 1998; Hobson, 1999). Based on carbon and hydrogen isotope data, Rubenstein et al.(2002) showed, for example,that black-throated blue warblers (Dendroica caerulescens) wintering on western Caribbean islands originate from the northern end of the species'range, whereas birds wintering on eastern Caribbean islands are primarily from the southern breeding range. In a similar study, Kelly et al.(2002) used hydrogen isotope data to demonstrate that southern populations of Wilson's warbler(Wilsonia pusilla) migrate to the northern range of the wintering habitat, whereas northern populations migrate in a leapfrog pattern to southern wintering areas. It has long been proposed that the analysis of stable isotopes of several tissues from the same animal can serve as a window to different dietary periods, because the turnover rates of isotopes may vary between tissues (Hobson and Sealy,1991). In captive rodents, for example, turnover rates of carbon isotopes were high in fat and liver, intermediate in muscles and low in brain and hair (Tieszen et al.,1983). Stable isotope composition in fat and liver can give short-term dietary information, that of muscles medium-term dietary information, and brain and hair long-term dietary information. This comparison can be valuable in the study of migration because it helps to distinguish between individuals that have newly arrived at the wintering habitat from those who arrived earlier (Hobson,1999). In addition, it may help to track changes in feeding habits or metabolic states (e.g. catabolizing versus fat-depositing animals)in dietary studies.

Tieszen et al. (1983)showed that the isotope turnover rates in tissues are related to the tissue-specific turnover rates. On a whole-animal basis, it could be argued that animals with a high mass-specific energy turnover rate should have a high isotope turnover rate. Nectar-feeding bats (Glossophaginae, Phyllostomidae)have mass-specific metabolic rates that exceed those of most other eutherians by a factor of two (von Helversen and Winter, in press). Thus, we were interested to determine at what rate stable carbon isotopes are exchanged in tissues of nectar-feeding bats. We postulated that the isotope turnover of nectar-feeding bats should be higher than in other eutherian mammals.

Stable isotopes have been used in the study of bats and especially in nectar-feeding bats (Herrera et al., 1993, 1998, 2001; Fleming et al., 1993; Fleming, 1995). Fleming et al.(1993) showed that migratory Leptonycteris curasoae, within their northern distribution range,feed extensively or, in some cases even exclusively, on the nectar of CAM-plants (Cactaceae and Agavaceae) with a carbon isotope ratio of approximately -10‰. By contrast, in the southern parts of their distributional range in Mexico, which corresponds to the wintering habitat of northern populations, Leptonycteris feeds primarily on nectar of C₃ plants that secrete sugar with a carbon isotope ratio of-25‰. By analyzing the carbon isotope abundance in tissues of L. curasoae from different geographical regions, Fleming et al.(1993) tracked the seasonal patterns of dietary preferences and migratory movements of this species, and concluded that L. curasoae depends on a spatio—temporal nectar corridor of cacti and agave flowers during migration.

We performed a diet-switching experiment with two glossophagine bat species(Leptonycteris curasoae and Glossophaga soricina), during which we changed the carbon isotope ratio of the diet by almost 14‰. Diet-switching experiments are facilitated in nectar-feeding animals because the isotope ratio of the diet can be changed without changing the overall composition of the diet, as both diets are based on sugar water. We collected three types of tissue (hair, wing membrane and blood) that caused minimal harm to the animals. We selected blood as our standard target tissue, assuming that it would show the highest turnover rate. We chose small pieces of wing membrane because we expected it to exhibit an intermediate turnover rate relative to blood and hair. Biopsies of wing membrane are routinely taken for DNA extraction during genetic studies (Worthington Willmer and Barratt, 1996)and, based on our experience, punctured wing membranes regenerate within 3-4 weeks. We chose hair because of its expected low turnover rate(Tieszen et al., 1983). As a metabolically inert tissue at maturity, isotope ratios in hair should reflect the isotope composition of food consumed only during the time of hair growth.

We randomly selected ten Leptonycteris curasoae Miller (7 males and 3 females) and ten Glossophaga soricina Pallas (3 males and 7 females; Phyllostomidae, Glossophaginae) from a captive breeding colony maintained in greenhouse facilities at the University of Erlangen-Nürnberg, Germany. Female bats were neither pregnant nor lactating. All bats were banded with coloured and numbered plastic bands on the forearm (size XCL for G. soricina and XL for L. curasoae; A. C. Hughes, Hampton Hill, UK) for individual identification. The animals were introduced into two indoor flight enclosures (4 m×3 m×2.5 m), one for each species, and the two groups were maintained over a period of 70 days with food provided ad libitum at three artificial feeders. All bats were returned to the breeding colony after the experiment. The room temperature was set to approximately 23°C, the ambient humidity to 70% and the light:dark regime to 12 h:12 h. The two groups were habituated to the flight enclosure over a 1-week period and fed with the same diet as in the breeding colony. The food provided during the initial phase of the experiment was identical in isotopic composition with the diet the bats received in the greenhouse facility where they were held prior to and after the experiment. This food originated from plants using the C₃photosynthetic pathway (Table 1). The stock solutions (Tables 1, 2) were dissolved in water to a final concentration of 18% sugar water (mass/mass, Atego refractometer;accuracy 0.2%) and the diluted sugary waters were provided from separate feeders.

After 7 days on the initial diet, we switched the diet of plants to one with a carbon isotope composition of the C₄ and CAM photosynthetic pathways (Table 2), referred to as day 1 of the experiment. As before, the stock solutions were diluted to a sugar concentration of 18% (mass/mass). To complement the sugary solution of the bats, we added several mg of vitamins and mineral supplements to the diet each day.

Before and after each night, the sugar water was weighed to an accuracy of 1 g and we refer to ingested food as the difference between the two measurements. The concentration of ¹³C in the ingested food was estimated by calculating the proportion of each of the three main food sources that were ingested each day and by multiplying this value with the corresponding concentration of ¹³C in the food items (Tables 1 and 2).

At the end of the first week and during each subsequent sampling event,bats were weighed to the nearest 0.01 g on an electronic balance (Mettler PM-100, Columbus, OH, USA) and three types of samples were taken from each bat. We took two tissue samples from the wing membrane using a 3 mm diameter biopsy punch. Next, we drew approximately 30 μl of blood from the propatagial vein using a small sterile needle. Finally, we removed hair from an area of approximately 0.25 cm² from the back of each bat. The hair was cut with scissors at the base close to the skin and from varying regions of the back but from the same spot at a given day. All samples were placed into Eppendorf tubes, labelled and transferred immediately into a drying oven, where they were dried to constant mass at 60°C. Subsequently,samples were stored in a freezer below 0°C. After changing the diets, we took blood, wing membrane and hair samples at the end of the second, fourth,sixth and eighth weeks.

To remove external contaminants from skin and hair samples, we washed the samples with a chloroform/methanol solvent (1:1). To test for any differences between washed and unwashed samples, we collected samples of hairs from eight Leptonycteris and divided each sample into two parts. The first part was washed in the solvent and the second part remained untreated. We then measured the carbon (δ¹³C, ‰) and nitrogen isotopes(δ¹⁵N, ‰) in the hair samples following the procedure described below and performed pair-wise comparisons between washed and unwashed samples. We found a significant carbon enrichment of 1‰ in treated versus untreated samples but no difference inδ ¹⁵N (Table 3).

We used the carbon isotope ratio of Vienna Pee Dee Belemnite limestone and the nitrogen isotope ratio of air as standards.

To test for differences in mean isotope enrichment between (1) the tissues and (2) the diets, we performed one-way analysis of variance (ANOVA). We ran post-hoc Tukey HSD tests for pair-wise comparisons to evaluate differences in mean values.

We calculated mean isotopic values for all sample periods. In theory,changes in isotopic composition should follow an exponential curve (e.g. Tieszen et al., 1983). Hence,equations of the type y=a+be^ct were fitted to theδ ¹³C data from each tissue and each bat species. In this equation, a represents the asymptotic δ¹³C value for the tissue equilibrated on a C₄/CAM-diet, b equals the overall change in isotope ratio, c is the turnover rate of carbon isotopes in the tissues, and y the mean carbon isotope ratio in the tissue at the time t. For the reasons of simplicity, we refer to c as the regression coefficient in the exponential model.

We assumed that the different tissues equilibrate to an isotope ratio close to the value measured for the C₄/CAM-diet plus the average fractionation value found for that specific tissue. Thus, a equalled the average isotope ratio of carbon isotopes in the C₄/CAM-diet plus the difference between δ¹³C of the C₃ diet and the tissue caused by fractionation. Additionally, we assumed that b equalled the overall change in isotopic composition in the two diets, which was 13.7‰ in Leptonycteris curasoae and 13.6‰ in Glossophaga soricina. Estimation of c was performed on an iterative basis starting with a value of 0.05. The iteration was stopped after changes in the sum of squares were smaller than 1.0×10^-8. To estimate the half time of the carbon isotope exchange in the different tissues, we calculated t₅₀ using the following equation: t₅₀=log_e(0.5)/c, where t₅₀ is the time in days in which half of the carbon isotopes were exchanged in the corresponding tissue, and 0.5 represents the exchange of 50% isotopes. For reasons of simplicity we will describe the half life of carbon isotopes (=t₅₀) in hair, although we are aware that the isotope composition of hair reflects only the period of growth,as hair is an isotopically inert tissue.

In a preliminary experiment with five Leptonycteris curasoae, we observed that within 22 days following the diet switch bats had lost some body mass. To evaluate a possible bias in our experiment due to loss of body mass,we calculated exponential exchange curves on an individual basis. We then tested whether the rate of increase c of the exponential functions was related to the change in body mass. We predicted that the regression coefficient c decreased with increasing loss of body mass if carbon isotopes of catabolized fat or protein of C₃ origin mixed with the ingested carbon isotopes of C₄/CAM origin. We expected such effects to be most apparent in tissues with a relatively high turnover and thus performed the statistical analysis only for the data set of wing membrane and blood. The level of significance was Bonferroni-corrected to 2.5%, because two data sets were tested for each individual.

Values are expressed as means ± 1 S.D. In general, two-tailed tests were performed. We used SPSS (version 9.0) for all statistical analysis and regression models.

At the end of the initial phase of the experiment (i.e. before day 1), mean values of isotopic composition differed significantly between the first diet and the three target tissues (Fig. 1; ANOVA: Leptonycteris curasoae: F_3,33=66, P<0.001; Glossophaga soricina: F_3,33=98, P<0.001). The difference between carbon isotope ratios of the initial diet and in the animal tissues averaged 2.8±0.4‰ for L. curasoae and 2.6±0.5‰ for G. soricina. In L. curasoae,blood was more enriched in heavy carbon isotopes than wing membrane(Fig. 1: Tukey's test, P=0.007). In G. soricina, blood contained more ¹³C than wing membrane and hair(Fig. 1: Tukey's test, P<0.001).

On average, individuals of L. curasoae ingested more sugar water per day (20.3±2.9 ml day^-1) than did individuals of G. soricina (19.2±1.8 ml day^-1) (Student t-test, t₁₁₀=2.6, P=0.014). Daily rate of food intake did not vary between diets 1 and 2 (L. curasoae: Student t-test: t₆₁=1.95, P=0.06, G. soricina: Student t-test, t₆₁=0.23, P=0.82). In addition,the mean nutritional intake rate remained constant throughout the experiment(ANCOVA: L. curasoae: F_4,58=1.4, P=0.25; G. soricina: F_4,58=0.97, P=0.43). In both species, mean body mass decreased significantly during the course of the experiment (ANOVA for repeated measures: L. curasoae, interval: F_4,39=37, P<0.001; G. soricina,interval: F_4,39=63, P<0.001). In addition,mean body mass was significantly different between individuals (L. curasoae, F_9,39=26, P<0.001; G. soricina,individual, F_9,39=43, P=0<0.001). On day 1 of the experiment, the mean body mass of L. curasoae was 23.6±2.1 g and G. soricina, 10.2±0.7 g. At the end of the experiment,mean body mass of L. curasoae had decreased to 21.7±2.3 g and of G. soricina to 9.4±1.0 g. Thus, both species lost on average 8% of their initial body mass during the course of the experiment. Body mass changes of individuals of both species are plotted in Fig. 2A,B.

After switching the diet to plant products of C₄/CAM-origin, the enrichment of heavy carbon isotopes increased in all three types of tissues sampled in both species (Fig. 3A,C,E for Leptonycteris curasoae and Fig. 3B,D,F for Glossophaga soricina). At the end of the experiment, at day 60, none of the tissues had equilibrated to the expected point of carbon isotopic enrichment (first numerical value in the exponential regression equations plotted in the graphs). Based on the calculated exponential equations, we estimated the half life of the carbon isotopes in tissues (see t₅₀ values in Figs 3 and 4). In both species, wing membrane and blood had very similar turnover rates of approximately 116 days(range: 102-134 days, Table 4). Based on a paired t-test, mean regression coefficients of wing membrane and blood were not significantly different within species (L. curasoae: t₉=1.0, P=0.32, G. soricina: t₉=1.3, P=0.22). In addition,neither mean regression coefficients of blood nor those of wing membrane were different between species (Student t-test; blood: t₁₈=0.1, P=0.91; wing membrane: t₁₈=1.4, P=0.17). Compared to blood and wing membrane, estimated t₅₀ values for hair were higher by a factor of five, averaging 537 days in L. curasoae(Table 4). We could not detect a significant change in δ¹³C in hairs of G. soricina.

On an individual basis, we tested whether body mass loss was related to the regression coefficient of the exchange curve of carbon isotopes in Leptonycteris curasoae (Fig. 4A) and Glossophaga soricina(Fig. 4B). In blood samples of G. soricina, the regression coefficients increased significantly with increasing loss of body mass of the corresponding individual(Table 5). Thus, contrary to our expectation, turnover rates were faster in those animals that lost body mass than those with a constant body mass, at least in blood of G. soricina. When controlling for the effect of loss of body mass in G. soricina, t₅₀ of blood was 126 days (see legend of Fig. 4).

During biochemical processes, enzymes discriminate against heavy isotopes of carbon and nitrogen (DeNiro and Epstein, 1978, 1981). In addition to the processes of kinetic fractionation, this leads to an average enrichment of heavy nitrogen of 3.4‰ between trophic levels(Eggers and Jones, 2000). The discrimination function is less intense for carbon isotopes than for nitrogen isotopes and, thus, the enrichment of ¹³C from diet to tissue is smaller: on average only 1‰ or 2‰(DeNiro and Epstein, 1981). Hobson and Clark (1992b)reported values from 0.2‰ to 2.7‰ in tissues of Japanese quails(Coturnix japonica). In our study, the fractionation effect against ¹³C at 2.8‰ and 2.7‰ was consistent within both species and similar to the findings of Hobson and Clark.

We found a low rate of growth in hair of Leptonycteris curasoaeand no significant change in the isotope composition of Glossophaga soricina hair. Because hair is a metabolically inert tissue, only the basal, growing part should reflect the isotope ratio of the current diet. Thus far, all bats studied with respect to hair growth and molt showed a seasonal pattern (Constantine, 1957, 1958; Kunz, 1974; Mazak, 1963, 1965). Thus, if both study species were in a phase of reduced hair growth, only negligible amounts of carbon with a C₄ isotope signature would be included. Since the patterns of hair growth are unknown for glossophagine bats, we cannot evaluate the effects of seasonal hair growth in our study species.

In the wing membrane of bats, the estimated half life of carbon isotopes was 100 and 130 days. In contrast to this finding, biopsies of wing membranes heal at a remarkably rapid rate (see also Worthington Wilmer and Barratt,1996). In nectar-feeding bats, punctured wing membranes caused by a biopsy punch heal and completely close within 3-4 weeks. However, the regeneration time of damaged bat wings is probably not representative of the regeneration time of the whole tissue. Quite likely, special biochemical processes are triggered after a wing membrane is damaged and, as a consequence, regeneration time of wounds in the membrane is higher than the overall turnover time in that tissue. Wing membranes comprise an extremely reduced dermis that is sandwiched between the dorsal and ventral layer of the epidermis (Quay, 1970). To increase elasticity and stiffness, bat wings consist of a two-dimensional network of large, macroscopic collagen—elastin fibre bundles(Holbrook and Odland, 1978). Palaeontologists use bone collagen as an indicator of yearly or lifetime diets of organisms because collagen is known to regenerate at a low rate (e.g. t₅₀=173 days in quails: Hobson and Clark, 1992a). High concentrations of collagen and elastin in wing membranes and their possible low rate of regeneration could explain our findings of a low carbon isotope turnover rate in the wing membrane. Bat wings are also cooler than the core body temperature of bats. Lancaster et al.(1997) showed that the temperature of the wing membrane is approximately 1-2°C above the ambient temperature. Thus, in the present study, biochemical processes in the wing membrane probably took place at a temperature of 24-25°C. The specific molecular composition and the relatively low temperature could explain the low turnover rates of carbon isotopes in bat wings.

Blood was estimated to have a t₅₀ value of 120 and 126 days. Because this is the first measurement of blood turnover in bats, we compared our data with those from mammals other than bats. In humans,erythrocytes have a t₅₀ value of 60 days(Schmidt and Thews, 1995). In avian blood, the carbon isotope half life was 11 days in Japanese quails(Coturnix japonica; Hobson and Clark, 1992a) and approximately 30 days in young, growing American crows (Corvus brachyrhynchos; Hobson and Clark, 1993). Our results indicate that the estimated isotope half-life of approximately 120 days in bat blood is higher than corresponding values from other animals,including humans, and is also in contrast to our prediction.

We discuss this unexpected result in the context of three different scenarios. (1) The rate of isotope turnover was slowed down because animals were torpid or lethargic. (2) The bats accumulated a nutritional deficit,which contributed to a loss of body mass. In the latter situation, the mobilization of fat and/or body proteins would affect the estimate of stable isotope turnover rates, so we evaluated the possibility that carbon from sugar and carbon from body tissues are used for different purposes: energy versus tissue synthesis. Finally, we consider the possibility that the measured turnover rates of carbon isotopes are representative for the actual tissue turnover despite the presence of body mass loss.

Many bat species are known for their propensity to enter torpor under low ambient temperatures or unfavourable food regimes. Torpor or lethargic conditions have been mostly studied in bats of temperate zones, namely vespertilionid species (see review by Speakman and Thomas, in press). Most tropical bats do not enter torpor on a regular basis,although studies are scarce on this phenomenon in tropical species. Cruz-Neto and Abé (1997) found that captive G. soricina reduced their body temperature when deprived of food. Facultative torpor has also been observed in other nectar-feeding bats, such as L. curasoae and Choeronycteris mexicana in captivity (C. Voigt, personal observation). Thus, some phyllostomids do reduce their body temperature under certain conditions. It could be argued that the low turnover rates in tissues of the two species of nectar-feeding bats could be due to prolonged periods of torpor, but two facts argue against this hypothesis. First, the bats of this study were always ready to fly from their roost when they were disturbed and torpid or lethargic bats are incapable of flight. Second, the nectar-uptake rate did not change during the course of the experiment, which would have been expected if bats used torpor after the switch in diets. Thus, torpor is an unlikely explanation for the observed low turnover rates.

Both G. soricina and L. curasoae lost approximately 8% of their initial body mass during the experiment. Nutritional deficits due to insufficient energy, mineral or nitrogen supplies, are possible causes for such a trend. Both species ingested approximately 20 ml of sugar water each day. As the bats were fed with 18% sugar water, this can be converted into a daily energy intake rate of approximately 60 kJ day^-1. This value falls into the range of previously reported rates of energy expenditure for both species (von Helversen and Winter, in press). In addition, Horner et al.(1998) estimated the field metabolic rate of L. curasoae as at least 40 kJ day^-1 and daily energy intake rates of non-reproductive G. soricina ranged from 7 to 68 kJ day^-1 in a captive study(Winter, 1993; Voigt, in press). We find it unlikely that the study bats suffered from an insufficient supply of energy. Other forms of nutritional stress may have occurred despite our intention to provide all essential nutrients such as vitamins and minerals.

In the present work, we simulated the extent to which fat mobilization affects the overall isotope ratio of the carbon pool that is available to the bat for homeostasis. In this simulation, we assumed that the isotopic fractionation between exogenous foods and tissues is the same as that for the metabolism of tissues and endogenous reserves. Assuming that (1) loss of body mass was solely due to the mobilization of fat and (2) body fat contained approximately 20% water, loss of 1.6 g body mass would be equivalent to approximately 1.3 g dry fat in a L. curasoae weighing 20 g. Assuming also that a mammalian fat molecule consists of two C₁₆ chains and one C₁₄ chain, fat molecules would consist of 42 carbon, 100 hydrogen and 6 oxygen atoms. Taking the mass of carbon, hydrogen and oxygen into account, 1.3 g fat consists then of approximately 0.94 g carbon(¹²C), 0.18 g hydrogen (¹H) and 0.18 g oxygen(¹⁴O). Following this, 0.94 g carbon with an isotopic signature of approximately -22‰ δ¹³C should have mixed with the ingested sugar molecules. L. curasoae ingested on average 20.4 ml sugar water per day, or 1.2 l sugar water during the 2-month period of the experiment. For reasons of simplicity, we have calculated all following values for glucose (C₆H₁₂O₆), which was the predominant sugar molecule in the diet. 1 ml of 18% sugar water contains 180 mg sugar. As nectarivorous bats absorb sugar in the intestine with an efficiency of almost 100% (Winter,1998), each Leptonycteris is likely to have ingested 216 g sugar during the 2-month period. This is equivalent to approximately 86.4 g carbon, 14.5 g hydrogen and 115.1 g oxygen, respectively. Following this simulation, the ingestion of 216 g sugar results in the uptake of 86.4 g carbon from sugar (δ¹³C=-10‰), in contrast to 0.94 g carbon from fat (δ¹³C=-22‰). Thus, the overall isotope ratio of the carbon pool available for homeostasis would be -10.02‰,compared to -10.00‰ without fat mobilization. Even if fat was more depleted in heavy isotopes by approximately 3‰, a fractionation factor found in fat tissues of other animals, the overall δ¹³C value would change only slightly to -10.05‰. Therefore, fat mobilization alone cannot explain the unexpectedly low estimates of isotope turnover rates seen when carbon isotopes are used for homeostasis, irrespective of their origin.

Preferential use of carbon from fat or proteins for tissue synthesis and carbon from sugar for energy synthesis could, however, change the above picture. Hobson and Stirling(1997) addressed this problem of selective metabolic pathways or differential isotopic routing. In a dietary study on polar bears, the authors found that carbon isotope ratios in blood were not distinguishable between offshore populations feeding on seals and inland populations feeding on berries, although seals as a fat- and protein-rich food source and berries as a carbohydrate-rich food source differed by approximately 9‰ in carbon isotope ratios. This led to the conclusion that inland populations probably burned the carbohydrates from berries directly instead of incorporating them into tissues. This is a likely explanation in the specific case. In contrast to polar bears, nectar-feeding bats are dietary specialists depending strongly and sometimes even exclusively on nectar as their food source. Bats may supplement their diet with pollen,fruits and to some extent with insects, but the main carbon source of nectar-feeding bats is nectar. Therefore, nectar-feeding bats ultimately have to incorporate nectar carbon into their tissue. However, it is likely that in our study that carbon isotopes from mobilized body substrates were incorporated into the blood, thus mixing with the carbon of assimilated sugar. Obviously, this effect should be most apparent in starving animals and absent in animals with constant body mass. Hatch et al.(1995) showed that the catabolic states of rooster chicks and adult hens are indicated by increased levels of δ¹³C in haemoglobin. In both species of our study,we find individuals that lost body mass and individuals that maintained an almost constant body mass during the experiment(Fig. 2). We expected that the regression coefficient c of the estimated exchange curve would be underestimated in catabolizing animals and that the coefficient should reflect true values in individuals with constant body mass. We tested for a significant positive relationship between the regression coefficients c and body mass loss in four tissues, but found only a negative correlation for G. soricina blood(Fig. 4). Therefore, loss of body mass did not result in an overestimate of the isotope turnover half life in these specific cases. Possibly, additional fractionation effects obscured the presumed positive correlation between loss of body mass and regression coefficient, or may even have reversed this trend.

Bat blood is unusual in several ways. First, blood from bats contains more erythrocytes per ml than in other mammals (26×10⁶erythrocytes ml^-1 blood in bats versus approximately 18×10⁶ erythrocytes ml^-1 blood in Rodentia or Insectivora; summarized by Neuweiler,1993). Secondly, hemoglobin concentrations are higher in bats than in other mammals or even birds (0.24 g ml^-1 blood in a pipistrelle bat versus 0.18 g ml^-1 blood in hummingbirds; summarized by Neuweiler, 1993). These extreme values might be an adaptation to the high energy demands of flight(Voigt and Winter, 1999; Voigt, 2000). In birds, the oxygen capacity of blood is maximized by unidirectional lungs, resulting in an optimal oxygen uptake. Assuming that similar sized bats and birds are producing erythrocytes at similar rates, we hypothesize that bats reach a similar oxygen capacity of blood by having larger amounts of smaller,longer-lived erythrocytes. Such a mechanism could compensate for the less efficient bidirectional lung of bats. Thus, long-lived erythrocytes may be an adaptation of bats to the high energy demands of flight. By using the same experimental setup and separating the cellular fraction of blood from plasma by centrifugation (see also Hobson and Clark, 1993), it will be possible to test this idea.

In summary, the estimated turnover rates of isotopes in the target tissues were consistently low for both study species. In L. curasoae, we could trace small amounts of carbon isotopes from C₄-sugar in the hair, but no significant amounts from C₄-sugar could be found in hair of G. soricina. This finding is probably explained by the seasonal growth of hair. The rate of isotope turnover in wing membrane was low, with t₅₀ values ranging from 100 to 130 days. The particular composition and design of the wing membrane and its relatively low temperature are likely explanations for this result. In blood, the turnover rate of carbon isotopes was unexpectedly low. The carbon isotope half life was 120-126 days. We suggest that long-lived erythroycytes are a special adaptation of bats for maintaining a high oxygen capacity of blood, which is a prerequisite for enduring, aerobic flight performance. Further investigations are needed in the field of bat blood physiology.

We wish to express our thanks to Christiane Stemmer for preparing the samples for isotope analysis and to Michael Joachimski (stable isotope laboratory, department of geology, University of Erlangen-Nürnberg,Germany) for analysing part of the samples. Many thanks go to Otto von Helversen for allowing us to work with the bats and to Monika Otter for maintaining the captive colony. We thank Heribert Hofer, Roland Frey and two anonymous reviewers for valuable discussion.