Stable carbon isotopes in exhaled breath as tracers for dietary information in birds and mammals

During the past decades, ecologists have increasingly used stable isotopes to study trophic interactions and dietary preferences in animal food webs(reviewed in Fry, 2006). Animals assimilate stable isotopes from the diet into their tissue or release them in their excreta and exhaled breath(DeNiro and Epstein, 1978; DeNiro and Epstein, 1981). Usually, excreta are depleted in ¹³C in relation to the stable carbon isotope signature of the diet, whereas animal tissues are enriched in ¹³C in relation to the diet by enzymatic action called isotopic fractionation (DeNiro and Epstein,1978; DeNiro and Epstein,1981). Isotopes are assimilated into animal tissue according to organ-specific tissue turnover rates (e.g. Tieszen et al., 1981). Thus, stable isotope signatures in organs with different turnover rates shed light on the diet consumed by the animal during the corresponding time periods. The stable carbon isotope composition (δ¹³C) of bone is considered to integrate the stable isotope ratio of the food consumed during an animal's lifetime (e.g. Tieszen and Fagre,1993). The stable carbon isotope ratio of other tissues, such as muscle, fat and liver, provides information on shorter time periods (Tieszen et al., 1981). The isotopic incorporation rates into these tissues have been found to be variable (Dalerum and Angerbörn, 2005) and possibly linked to the mass of the mammal (Carleton and Martínez del Rio, 2005). The stable carbon isotope signature of exhaled breath(δ¹³C_breath) is considered to be the fastest recorder of dietary information, becauseδ ¹³C_breath should match the isotopic signature of the ingested and combusted substrate(Perkins and Speakman, 2001; Hatch et al., 2002a; Hatch et al., 2002b).

The earliest experiments conducted under controlled conditions supported the idea that δ¹³C_breath closely matches the isotopic composition of the recently ingested food(δ¹³C_diet)(DeNiro and Epstein, 1981; Klein et al., 1988; Tieszen and Fagre, 1993). Since δ¹³C_breath seems to provide a snapshot of the stable isotope signature of the substrate currently metabolized, it is increasingly used by experimental physiologists and behavioural ecologists to study metabolic substrate use in captive (e.g. Carleton et al., 2006) and free-ranging animals (e.g. Podlesak et al., 2005). Diet-switching experiments during which the isotopic composition of the diet is drastically changed at the onset of the experiment– from e.g. plant products of C3 plant origin to those of C4/CAM plant origin – have provided important insights into substrate combustion in birds (e.g. Hatch et al.,2002a; Hatch et al.,2002b; Carleton et al.,2006) and mammals (e.g. Jeuckendrup and Jentjens,2000; Voigt and Speakman,2007). During diet-switch experiments, breath samples are collected at regular time intervals after animals have started to consume a meal differing in stable carbon isotope composition from their previous diet. Such experiments have revealed that simple sugars are routed to combustion within less than 10 min in small birds and mammals(Voigt and Speakman, 2007; Welch et al., 2006; Welch et al., 2008; Voigt et al., 2008a) and in approximately 30 min in exercising humans(Jeuckendrup and Jentjens,2000). Complex macronutrients such as cellulose, starch and proteins are probably more difficult to digest than simple sugars, since the number of enzymatic processes is likely to increase with increasing complexity of the catabolized macronutrient. Therefore, animals ingesting complex macronutrients may incorporate exogenous substrate at slower rates than animals ingesting simple sugars.

In addition to differences in the time lag at which macronutrient combustion is reflected in δ¹³C_breath,δ ¹³C_breath may be affected by other factors such as isotopic fractionation during enzymatic or physical processes, or by the combined use of exogenous and endogenous substrates. Endogenous substrates are usually depleted in ¹³C by –0.6 to –8.4‰ in relation to the diet at the time of lipogenesis (e.g. DeNiro and Epstein, 1977;Podlesak et al., 2006). Accordingly, δ¹³C_breath of fasting animals is usually depleted in ¹³C in relation to the previous diet (e.g. Hatch et al.,2002a; Hatch et al.,2002b). In many animals, especially free-ranging animals, it may not be evident whether individuals combust exclusively endogenous substrates or a mixture of exogenous and endogenous substrates. In this case, the difference between δ¹³C_breath andδ ¹³C_diet, henceforth calledΔ _{diet–breath}, may be affected by the isotopic composition of both substrates and the ratio at which they are used. Following this line of argument, if endogenous substrates are not in isotopic equilibration with the most recent diet and if animals metabolize both types of substrate, Δ_{diet–breath} may largely deviate from the most recent δ¹³C_diet (e.g. Podlesak et al., 2005; Carleton et al., 2006). Continuous ingestion of new food items with a δ¹³C signature differing from that of the previous diet will eventually cause an exchange of isotopes in endogenous reserves over time. Thus,Δ _{diet–breath} should decrease with increasing equilibration of fat and glycogen to the newδ ¹³C_diet(Carleton et al., 2006).

Considering all these factors that may have a large and short-term impact on δ¹³C_breath and consequentlyΔ _{diet–breath}, the aim of our study was (1) to generate baseline data of stable isotope turnover andΔ _{diet–breath} for a protein-combusting animal and (2) to survey the literature for general patterns to improve our ability to interpret stable isotope breath data of free-ranging animals.

First, using small insectivorous greater mouse-eared bats (Myotis myotis, Vespertilionidae), we quantifiedΔ _{diet–breath} in fasting individuals, the rate at which exogenous substrates are incorporated into the pool of metabolized substrates after ingestion of a protein-rich diet and the plateauδ ¹³C_breath at which the animals' breath levelled off after equilibration to the new diet. We predicted that (1) the rate at which M. myotis would make use of dietary proteins for metabolism would be slower than in sugar-combusting bats of similar size, (2)δ ¹³C_breath should be depleted in ¹³C in relation to δ¹³C_diet in fasting animals, and (3)δ ¹³C_breath should level off close toδ ¹³C_diet in satiated animals.

Second, we reviewed the current knowledge on species-specificΔ _{diet–breath} values and fractional rates at which exogenous substrates are incorporated into the pool of metabolized substrates for birds and mammals. We predicted that (1) animals ingesting complex macronutrients, such as proteins and complex carbohydrates, would have a slower fractional incorporation rate of exogenous substrate into the pool of metabolized substrate than animals feeding on simple carbohydrates such as hexose or sucrose, (2) fasting birds and mammals would have, in general,negative Δ_{diet–breath} values, since endogenous substrates are usually depleted in ¹³C in relation to the diet, and(3) Δ_{diet–breath} should decrease in satiated animals with increasing equilibration of endogenous substrate to the isotopic signature of the diet.

We studied metabolic substrate use in six male adult Myotis myotisBorkhausen 1797 in a captive population housed in facilities at the Max-Planck-Institute for Ornithology in Seewiesen. M. myotis is an insectivorous bat that forages on carabid beetles in temperate zone forests(Arlettaz et al., 1997). Our study animals were kept under an inverted dark–light cycle. The dark phase, i.e. the activity period, started at 09:00 h and lasted until 17:00 h. As exclusive bat food for the duration of the experiment, we raised mealworms(larval stages of Tenebrio molitor) on corn from either Hungary or Ecuador for 7 weeks. Ten days before the breath sampling, bats were fed mealworms enriched in ¹³C with corn from Hungary. Previous experiments in other bats demonstrated that this time period is sufficient to equilibrate the endogenous reserves of bats isotopically to a new diet(Voigt and Speakman, 2007). During the experiment, bats were fed with mealworms that were enriched in ¹³C with corn from Ecuador. We collected several mealworms from each diet for later stable isotope analyses. Dead mealworms were dried to constant mass at 40°C in a drying oven and then a subsample from the tissue of inner organs was taken for stable carbon isotope analyses. The mealworms raised for 7 weeks on Hungarian corn had a δ¹³C of–24.1±0.8‰ (diet 1), while those raised on the Ecuadorian corn had a δ¹³C of –22.0±0.3‰ (diet 2),indicating that the isotopic signature of the mealworms was slightly more enriched in ¹³C than before.

We assessed the rate at which exogenous substrate was incorporated into the pool of metabolized substrates in a diet-switch experiment, i.e. we fed fasted bats that had been maintained over 6 or 7 days ad libitum on diet 1 with diet 2. Prior to the diet-switch experiment, bats fasted for at least 13 h, which is fully congruent with the bats' natural diurnal pattern of food intake. All experiments were performed during the activity period of the bats.

Before offering mealworms, we collected an initial breath sample from each bat. Following the ingestion of the first mealworm we collected breath samples after 10, 20, 40, 60, 80, 100 and 120 min (see below for the technique of breath collection). Bats were weighed to the nearest 0.1 g before and after the experiment using a hand-held balance (Pesola, Baar, Switzerland). We lost body mass data from two individuals for the time after the experiment and,therefore, can only provide body mass change data for four individuals.

For breath sampling, bats were transferred singly into cotton bags (17 cm×25 cm) that were individually put into a larger plastic bag (volume 0.5 ml; Ziplock™, Racine, WI, USA). Ambient air was washed of CO₂ using NaOH and flushed through the bag via a plastic tube (diameter 3 mm) at a minimal flow through rate of 1.4 l min^–1. The outlet of the plastic bag consisted of a small slit of 4 cm (width 0.2 cm). A small tube was positioned with one end close to the bat's head inside the bag (diameter 1 mm, length 4 cm). We fused a needle hermetically to the other end of the tube outside the bag. For CO₂accumulation, we sealed the plastic bag for 1 min. M. myotis have a resting metabolic rate of approximately 25 ml O₂h^–1 (Hanu,1959). Therefore, we expected CO₂ to accumulate to approximately 0.5% during this time span. We then sucked air from the bag including the bat's breath into a vacutainer (Labco, Buckinghamshire, UK) by penetrating the Teflon membrane of the vacutainer with the needle. After each breath collection the plastic bag was unsealed again and CO₂-free air was flushed through the bag. Breath collection was repeated after 10, 20,40, 60, 80, 100 and 120 min following the first feeding event, since we expected an exchange of stable carbon isotopes in exhaled CO₂during this time period (Voigt &Speakman, 2007; Voigt et al.,2008a). Bats were fed repeatedly after 20, 40 and 60 min following the first feeding event to ensure that the bat's breath was equilibrated isotopically to the new diet. We define isotopic equilibration of an animal's tissue or breath as the status in which the isotopic composition of animal tissue or breath does not change any more with continuous consumption of the same food items.

Subsamples from the bats' diet (inner organs of the mealworms, excluding the cuticula) were dried at 40°C in a drying oven to constant mass,weighed on a Sartorius microbalance (Satorius AG, Göttingen, Germany) and loaded into tin capsules. All samples were combusted and analysed with a Flash elemental analyser and a Conflo II, coupled to a Delta-Advantage isotope ratio mass spectrometer (FisherThermo, Bremen, Germany) at the Stable Isotope Laboratory of the Leibniz-Institute for Zoo and Wildlife Research, Berlin(Germany). Samples were analysed in combination with internal standards that had previously been characterized relative to an international ¹³C standard (NBS22). All ¹³C/¹²C data were expressed relative to the international standard in the δ notation (‰)using Eqn 1. Precision was better than ±0.03‰ (1 s.d.).

For comparative reasons, we calculated the time at which 50% of carbon isotopes were exchanged in the animal's breath (t₅₀)according to the following equation: t₅₀=–ln(0.5)k, with ln representing the natural logarithm and 0.5 the exchange of 50% of isotopes. All values are given as means ± 1 s.d. and all statistical tests were performed two tailed unless otherwise stated.

We surveyed the literature for experimental studies that usedδ ¹³C_breath as a predictor forδ ¹³C_diet in birds and mammals. For each study, we obtained the following six parameters: (a) taxonomic group, (b) body mass in grams, (c) t₅₀ (min), or the fractional incorporation rate at which 50% of carbon atoms in exhaled CO₂ have been exchanged with carbon atoms from the exogenous substrate, (d)Δ _{diet–breath} for animals having fasted over a prolonged time period, (e) Δ_{diet–breath} for animals having ingested a diet that was more enriched in ¹³C in relation to the previous diet (exogenous substrate with a C4 signature and endogenous substrate with a C3 signature and, thus, endogenous substrate not in isotopic equilibrium with the new diet), and (f) Δ_{diet–breath}for animals having ingested a diet that was isotopically identical to the previous diet (endogenous substrate in isotopic equilibrium with the new diet). In cases where we obtained two or more values for one of these parameters, we calculated an arithmetic mean for these data to yield a single species-specific data point. We calculated a Wilcoxon matched-pairs signed-rank test for comparison of Δ_{diet–breath} between scenarios d and e, and d and f, and a Mann-Whitney U-test for a comparison of Δ_{diet–breath} between scenarios e and f. All three tests were performed one tailed, as we expected that the exclusive or additional combustion of endogenous substrates would cause a depletion of ¹³C in exhaled breath. We calculated a Mann–Whitney U-test for comparison of the t₅₀ values in animals digesting complex substrates, such as starch- or protein-rich diets,and animals digesting simple substrates, such as hexose or sucrose. This analysis was only performed for mammals and birds weighing less than 1 kg, as we expected a large impact of passage rate and endosymbiontic fermentation on t₅₀ values in large mammals (e.g. in ruminants).

The body mass of fasted M. myotis averaged 23.1±1.9 g 1 h after the onset of the dark cycle. The meanδ ¹³C_breath of fasted M. myotis equalled–29.9±0.7‰, which was –5.8‰ lower than theδ ¹³C of the diet (–24.1‰) that the bats fed on during the 6–7 days preceding the day of breath collection; this difference was significant (Wilcoxon signed-rank test: T+=0, T–=–21, P=0.0156). After being fed continuously with mealworms (δ¹³C=–22.0‰),δ ¹³C_breath became enriched in ¹³C(Fig. 1). The body mass of bats increased until the end of the feeding experiment by on average 4.4±1.6 g due to the ingestion of mealworms. The regression model calculated for the fractional incorporation rate of exogenous substrate into the pool of metabolized substrate reads as follows:δ ¹³C_breath(t)=–24.6(±0.7)–5.5(±1.1)e^{–t/31.2(±15.2)}. The mean time required to exchange 50% of the carbon atoms in the exhaled CO₂ with carbon atoms of the recently ingested protein source was 21.6±10.5 min (Table 1). The δ¹³C_breath levelled off at–24.6±0.7‰ after 120 min, which was –2.6‰lower than the δ¹³C of the new diet (Wilcoxon signed-rank test: T+=0, T–=–21, P=0.0313).

The fractional rates of incorporation of exogenous substrates into the pool of metabolized substrates ranged from a few minutes in nectar-feeding small birds and bats to almost 3 days in hay-ingesting alpacas (see Table 2). In species below 1 kg body mass, animals had higher t₅₀ values when digesting complex substrates such as a starch- or protein-rich diet than when digesting simple substrates such as hexose or sucrose (Mann–Whitney U-test: U=0.0, U′=25, N₁=5, N₂=5, P=0.0079). In fasting birds and mammals,mean δ¹³C_breath was depleted in relation toδ ¹³C_diet by –3.2±2.0‰(N=4 bird and N=5 mammal species; Table 2, Fig. 2). In animals that had recently fed on a diet with a C4 signature but still carried endogenous reserves with a C3 signature, δ¹³C_breath was depleted in ¹³C by –0.6±2.3‰ in relation to theδ ¹³C of the new diet (N=5 bird and N=7 mammal species). After continuously feeding on the new diet, the endogenous reserves (fat and glycogen) became isotopically equilibrated to the new diet.δ ¹³C_breath was, on average, enriched in ¹³C by +0.5±1.8‰ in relation to the diet after isotopic equilibration of endogenous substrate with the diet (N=6 species; Fig. 2). Depletion of ¹³C in exhaled breath in relation toδ ¹³C_diet was stronger in fasting animals than in animals that had recently fed on an isotopically contrasting diet (Wilcoxon matched-pairs signed-rank test: T+=7, T–=–38, N=9 pairs, P=0.0371; scenario d–e) and also more pronounced than in animals carrying endogenous substrates already isotopically equilibrated to the diet (Mann–Whitney U-test: U=2.0, U′=38, N₁=9, N₂=5, P=0.0062; Fig. 2;scenario d–f). On average, Δ_{diet–breath} was not significantly different between satiated animals with endogenous substrates equilibrated to the new diet and satiated animals with endogenous substrates not equilibrated to the new diet (Mann–Whitney U-test: U=19, U′=36, N₁=12, N₂=5, P=0.377; Fig. 2; scenario e–f).

We fed fasting vespertilionid bats (Myotis myotis) mealworms and monitored changes in the bats' δ¹³C_breath over a 2 h period. The δ¹³C_breath of the bats converged quickly on the stable carbon isotope signature of the newly ingested food items. M. myotis required twice as much time to exchange 50% of carbon atoms in exhaled CO₂ with those of the last meal as similar-sized Carollia perspicillata (family Phyllostomidae) feeding on simple sugars (Voigt et al.,2008a), but only 70% of the time similar-sized vampire bats need to metabolize recently ingested blood (Voigt et al., 2008). Mice (Mus musculus) feeding on corn, which includes mainly complex carbohydrates,i.e. starch, had almost identical t₅₀ values to M. myotis (Perkins and Speakman,2001). The lower t₅₀ values of sugar-ingesting bats may be caused by the faster enzymatic processing of small molecules, such as hexose or sucrose, than more complex molecules, such as starch and proteins. Our literature survey supports the notion that animals require more time to digest and combust complex macronutrients. However, t₅₀ values may be affected not only by the type of combusted macronutrient but also by the animal's size and phylogenetic background (Carleton and Martínez del Rio, 2005). Currently, we lack sufficient data to shed light on the effects of phylogenetic inertia on t₅₀ values. Data for t₅₀ listed in Table 2 suggest that large animals have slower incorporation rates (t₅₀ values) than small animals. Despite being large, alpacas and horses make use of complex carbohydrates such as cellulose by endosymbiontic fermentation. This mode of digestion may additionally slow down the speed at which exogenous substrates are available to fuel metabolism. Our survey also highlighted the fact that t₅₀ values were lower when the focus organisms were exercising [see reports on humans (Jeukendrup and Jentjens, 2000) and bats(Welch et al., 2008)]. However, the effect of exercise may become negligible in small animals ingesting simple sugars, as t₅₀ values were almost identical for resting and flying nectar-feeding bats feeding on a sugar water solution (Voigt and Speakman,2007; Welch et al.,2008).

The δ¹³C_breath of M. myotis fasting for 13 h was depleted in ¹³C by almost –6‰ in relation to the δ¹³C_diet of the animal's previous diet. In general, Δ_{diet–breath} averaged–3.2±2.0‰ for nine bird and mammal species that had fasted over an extended period of time (Table 2, Fig. 2). Thus, a depletion of ¹³C in relation to the isotopic signature of the diet(Δ_{diet–breath}) seems to be a common phenomenon in fasting animals and is probably caused by the strong fractionation of stable carbon isotopes during lipogenesis (DeNiro and Epstein, 1977).

After we fed fasting M. myotis with mealworms (diet 2) with a stable carbon isotope signature that was enriched by +2‰ in relation to the bats' previous food (diet 1), the bats'δ ¹³C_breath converged to a new plateau, which was depleted by –2.6‰ in relation to the δ¹³C of the new diet. By contrast, δ¹³C_breath of blood-feeding vampire bats (Desmodus rotundus) levelled off +2‰ above theδ ¹³C of their recently ingested blood meal(Voigt et al., 2008b). Given that both bat species process and metabolize protein, and assuming similar fractionation factors during protein combustion, M. myotis probably used more ¹³C-depleted endogenous substrate to fuel its metabolism than D. rotundus.

Many previous studies have looked at the relative level of the animals'δ ¹³C_breath plateau values after a diet switch (see Table 2). During most diet-switch experiments animals used a combination of exogenous and endogenous substrates to fuel their metabolism and in most cases experimental animals were labelled with carbon atoms from a C3 food web (low δ¹³C)at the beginning of the experiment. Since fat and glycogen stores of these experimental animals were built from carbon atoms of a C3 food source,Δ _{diet–breath} may deviate from zero throughout the experiment, because the exogenous substrate combustion in relation to the endogenous substrate combustion may vary over time (e.g. Carleton et al., 2006). On average, Δ_{diet–breath} equalled–0.6±2.3‰ in 12 bird and mammal species that had recently ingested food with a C4 signature, but still carried endogenous reserves with a C3 signature (Table 2, Fig. 2). After isotopic equilibration of endogenous substrate with the new diet, averageΔ _{diet–breath} equalled +0.5±1.8‰(N=6 species; Fig. 2). Thus, Δ_{diet–breath} decreased with increasing equilibration of endogenous substrates with the stable carbon isotope signature of the new diet.

Our survey highlights the fact that δ¹³C_breathmay be affected not only by the stable carbon isotope signature of the recently ingested food but also by the time elapsed since an isotopic shift in the diet, the turnover rate of endogenous substrates and the ratio at which endogenous and exogenous substrates are used for metabolism, e.g. when active or resting. Thus, animals that have recently changed to food items with an isotopic composition different from their previous diet may exhibit the largest range in Δ_{diet–breath}. Interpretation ofδ ¹³C_breath may prove difficult if no alternative measure of substrate combustion is available, e.g. the respiratory quotient of exhaled breath or metabolic products in blood plasma. We, therefore, advise caution in the interpretation of δ¹³C_breath from free-ranging animals without further validation studies under controlled experimental conditions, preferably in the same animal species as investigated in the field. The sensitivity of δ¹³C_breath to various factors such as fractionation, mixed combustion of exogenous and endogenous substrates and others may provide a powerful tool for physiologists and behavioural ecologists, once the underlying mechanisms are clearly understood.

Peter Thompson and Dr Paula Redman analysed the samples at the Aberdeen Centre for Energy Regulation and Obesity, and Karin Sörgel at the Stable Isotope Laboratory of the IZW. The experiments complied with the current laws of Germany; animals were caught and kept under licence of the responsible regulatory authorities; Regierungspräsidium Freiburg, licence no. 55-8852.44/1095, Veterinäramt Starnberg). We thank Carlos Martínez del Rio, Scott Carleton and Detlev H. Kelm for commenting on an earlier version of this manuscript.