Applications of stable isotope techniques to the ecology of mammals

Authors


  • Editor: JS

K. Crawford. E-mail: kcrawford07@qub.ac.uk

ABSTRACT

  • 1Stable isotope analysis (SIA) has the potential to become a widespread tool in mammalian ecology, because of its power in resolving the ecological and behavioural characteristics of animals. Although applications of the technique have enhanced our understanding of mammalian biology, it remains underused. Here we provide a review of previous applications to the study of extant mammals, drawing when appropriate on examples from the wider ecological literature, to identify the potential for future development of the approach.
  • 2Stable isotope analysis has been applied successfully to understanding the basic foraging decisions of mammals. However, SIA generates quantitative data on a continuous scale meaning that the approach can be particularly powerful in the characterization of community metrics, such as dimensions of resource partitioning within species assemblages or nutrient dynamics in food chains. Resolving spatial and temporal patterns of individual, intraspecific and interspecific resource use is of fundamental importance in animal ecology and evolutionary biology and SIA will emerge as a critical tool in these fields.
  • 3Geographical differences in naturally occurring stable isotopes have allowed ecologists to describe large-scale mammal migrations. Several isotopic gradients exist at smaller spatial scales, which can provide finer resolution of spatial ecology.
  • 4A combination of foraging and movement decisions is of prime importance in the study of ecotoxicology, since this discipline requires quantitative understanding of exposure risk.

INTRODUCTION

Information on animal movements and foraging preferences is fundamental to our understanding of a suite of evolutionary, physiological, behavioural and conservation-related processes (Webster et al., 2002; West et al., 2006). In the field of mammalian ecology, conventional approaches used to study diet, for example analyses of faeces, stomach contents and direct observations, have provided the basis for much of our current understanding of foraging decisions. However, there are well-described limitations and biases associated with these methods (Putman, 1984). Tracking mammals has provided important insights (Harris et al., 1990; Webster et al., 2002), although this approach also has a number of drawbacks. For example, individual tags or marks rely on subsequent recapture and/or resighting, while satellite transmitters are currently very expensive or too bulky to fit to smaller species and radio transmitters only work well over relatively short distances (Rubenstein & Hobson, 2004).

These problems are by no means the preserve of ecologists studying mammals and since the 1980s researchers have developed the use of stable isotope techniques as a complementary means of investigating diets and movements. Many of the studies that established the basis of stable isotope approaches used mammals as models (DeNiro & Epstein, 1978, 1981; Tieszen et al., 1983) and in recent years, the application of stable isotope analysis (SIA) within ecological studies has advanced considerably. However, there are still many potential applications of stable isotope approaches within mammalian ecology, presenting something of an untapped opportunity. The breadth of this topic necessitates that we focus on recent applications of stable isotope technology to the ecology of extant mammals. In addition, we only consider those studies using measurements of the natural abundances of stable isotopes and thus tracer-based approaches, such as doubly labelled water methods, are not covered. We begin with the theoretical basis behind SIA, allowing the reader to understand the concept before presenting more detailed information on case studies and the most recent advances. We also highlight a number of potential future research opportunities for ecologists studying mammals, and cover common misconceptions and important caveats associated with SIA.

SOME PRINCIPLES OF STABLE ISOTOPE ANALYSES

Many elements have two or more naturally occurring stable isotopes, which have different masses, for example nitrogen exists most commonly as 14N, but has a heavier, less common stable isotope 15N. Stable isotopes differ from radioisotopes because they do not decay. Because of their difference in mass, these isotopes behave differently in many environmental and physiological processes, resulting in natural variations in their relative abundance. This process, known as isotopic fractionation, leads to predictable changes in isotopic ratios, which can be analysed in environmental samples including animal tissues, using an isotope ratio mass spectrometer (McKinney, 1950). Moreover, the advent of the continuous flow isotope ratio mass spectrometer (Preston & Owens, 1983; Brenna et al., 1997) means that today we can conduct SIA on large numbers of samples with relative ease and efficiency. Differences in the abundance of the isotopes (ratios) are measured relative to a common international standard and expressed in delta notation (δ) as parts per thousand or per mil (‰) according to the following equation:

image

Where X denotes the element, H is the mass of the heavy isotope, R is the ratio of the heavy to light isotope for X. Thus, the ratio of 13C:12C is expressed as δ13C and 15N:14N becomes δ15N. A sample, which has more of the heavier isotope in relation to the standard, is commonly referred to as being ‘enriched’, and one with proportionally less of the heavy isotope is referred to as ‘depleted’.

Naturally occurring variation in the stable isotope ratios of a number of elements is useful to ecologists since consumer tissues are synthesized from dietary nutrients and so reflect the composition of their food in a predictable manner. For example, stable hydrogen isotope ratios in Canadian rainfall differ markedly from those in South America and this pattern is reflected in the tissues of migrating bats that eat food in each region (Cryan et al., 2004). Similarly, the isotope ratios of stable nitrogen and carbon within tissue from brown bears Ursus arctos feeding on salmon Salmo salar are quite distinct from those feeding on berries (Ben-David, Titus & Beier, 2004).

Since different animal tissues turnover at different rates, they integrate dietary and habitat information at different temporal and, if the animal is moving, different spatial scales (Tieszen et al., 1983; Bearhop et al., 2003; Rubenstein & Hobson, 2004; Dalerum & Angerbjorn, 2005). Blood plasma has a rapid turnover rate, which will give dietary information spanning the few days prior to sample collection, whereas red blood cells may persist for several months, depending on the species in question, and so indicates dietary integration over a much longer period (Hilderbrand et al., 1996; Shaner, Bowers & Macko, 2007). Also, some keratinized tissues such as hair, claws and whiskers are metabolically inert after formation, and so preserve the isotopic record of the animal's diet at the time of tissue synthesis (Hobson & Schell, 1998; Greaves et al., 2004; Zhao & Schell, 2004; Lee et al., 2005). These features give stable isotope analyses a distinct advantage over conventional approaches, as information from several time windows can be accessed from a single sampling event. The technique can also provide dietary information from elusive stages within a mammal's life history, which may otherwise remain unknown (i.e. during months when mammals are away from breeding areas or are difficult to capture).

There are a number of studies that have measured turnover rates in tissues (see Table 1). In general, there is a substantial amount of variation in these rates among tissue types among species, for example prey sources with varying C : N ratio (Miron et al., 2006) and so these values should be treated with caution, particularly if the precise timing of tissue synthesis is crucial to the question being addressed (Dalerum & Angerbjorn, 2005). In these instances, there is no substitute for a controlled experiment, manipulating the diet of the species of interest to calibrate turnover rates (Miron et al., 2006). However, care should be taken if applying exponential models to estimate isotopic turnover rates in tissue (Voigt et al., 2003). If tissues are not in equilibrium with the new diet, i.e. a curve has not reached an asymptote, and/or the number of tissue measurements over time is too low (Voigt et al., 2003), then fitting inappropriate models, albeit with high regression coefficients, can lead to substantial overestimates of isotopic turnover times. Such methodological errors may be one of the reasons why estimates of turnover rates vary markedly among studies. Moreover, Cerling et al. (2007) have highlighted the difficulties of fitting exponential models to tissues with multiple pools. They use the reaction progress variable, which allows the isotopic time series to be treated as multiple linear functions and can produce more accurate estimates of isotopic turnover. However, this approach still requires an adequate number of measurements over time.

Table 1.  Isotopic turnover rates [average half-life (t1/2 days)] of δ13C and δ15N for various tissues within mammalian species
 MammalDietTissueδ13Ct1/2
(days)
δ15Nt1/2
(days)
Reference
  1. Values in parentheses represent fractions of either single or multiple pools.

  2. RBC, red blood cell.

RodentGerbilC3 wheat+supplementsMuscle27.6 Tieszen et al. (1983)
GerbilC3 wheat+supplementsLiver6.4 Tieszen et al. (1983)
GerbilC3 wheat+supplementsHair47.5 Tieszen et al. (1983)
MouseMixed diet (casein, beet sugar, soybean oil)Whole blood17.315.4MacAvoy et al. (2006)
RatMixed diet (casein, beet sugar, soybean oil)Whole blood24.827.7MacAvoy et al. (2006)
CarnivoreBlack bearPlant/meat/fishRBC2828Hilderbrand et al. (1996)
Black bearPlant/meat/fishPlasma/serum44Hilderbrand et al. (1996)
HerbivoreDomestic horseC3/C4 grassHair0.5 (0.41); 4.3 (0.15), 136 (0.44) Ayliffe et al. (2004)
AlpacaC4 grassBreath CO22.8 [0.4 (0.46); 10 (0.19); 26 (0.35)] Sponheimer et al. (2006)
AlpacaC4 grassLiver37.3 [11 (0.45); 200 (0.55)] Sponheimer et al. (2006)
AlpacaC4 grassMuscle178.7 [10 (0.03); 215 (0.97)] Sponheimer et al. (2006)
NectarivoreNectarivorous batSoyaWhole blood24.325.6Miron et al. (2006)
Nectarivorous batAmaranthWhole blood39.725.0Miron et al. (2006)
Nectarivorous batC4 plants ie. Corn syrupHair537 Voigt et al. (2003)
Nectarivorous bats (2 spp.)C4 plants ie. Corn syrupWing membrane102/134329/821Voigt et al. (2003); Voigt & Matt (2004)
Nectarivorous bats (2 spp.)C4 plants ie. Corn syrupWhole blood113/120274/514Voigt et al. (2003); Voigt & Matt (2004)
Omnivorous marsupialLong-nosed bandicootC3/C4Blood plasma9–109–10Klaassen et al., (2004)
Long-nosed bandicootC3/C4Blood cells∼90∼90Klaassen et al. (2004)

COMMONLY USED STABLE ISOTOPES AND THEIR APPLICATIONS

The variable roles of elements in physical and biochemical processes mean that different isotopes can be used to address a range of ecological questions (Rubenstein & Hobson, 2004; Fry, 2006). So far, the most commonly used stable isotopes in ecological studies have been carbon, nitrogen and hydrogen; however, the use of other isotopes, for example oxygen, sulphur and strontium, has the potential to provide further resolution in some systems.

Stable carbon isotope ratios (δ13C) can be very effective tracers of different carbon sources within a food web. In terrestrial systems, δ13C varies mostly according to the photosynthetic metabolism of plants. For example, the δ13C values of C4 and CAM plants are enriched by about 12–14‰, when compared with terrestrial C3 plants (Lajtha & Michener, 1994). This gradient is commonly utilized, as shown when studying the significance of C3 and C4 plants to the diet of baboons Papio ursinus within two areas of South Africa (Codron et al., 2006). C3 trees had approximately δ13C of −26‰ whereas C4 grasses δ13C of −12‰; therefore, it was possible to estimate that baboons within one area were eating approximately 25% more C4 plants than those within the other area. δ13C can also be used to discriminate between marine and terrestrial carbon sources, since primary producers in terrestrial systems fix carbon from atmospheric CO2, whereas in marine systems dissolved inorganic carbon is a main contributor (Fry, 2006). Carbon isotope ratios also show alignment with ecological divisions within aquatic systems: inshore sources tend to have enriched stable carbon isotope ratios in comparison with offshore regions; and similarly benthic sources have higher δ13C values compared with pelagic sources (Rubenstein & Hobson, 2004; Fry, 2006). δ13C also varies predictably with latitude, with values becoming more depleted with increasing latitude (Kelly, 2000; Rubenstein & Hobson, 2004; Fry, 2006), which is largely a consequence of changes in the relative proportions of C3 and C4 plants.

The most commonly utilized characteristic of stable nitrogen isotopes is the predictable increase in δ15N (usually between 2‰ and 4‰) at each trophic level (Kelly, 2000; Fry, 2006). This feature can potentially allow direct inferences to be made about the diet of consumers, but can also be used to provide insights into community level phenomena such as trophic cascades, food chain length and resource partitioning (Post, 2002; Roemer, Donlan & Courchamp, 2002). Moreover, nitrogen isotope ratios can vary along several other potentially useful gradients (Rubenstein & Hobson, 2004). For example, δ15N was used to determine age at weaning of two species of otariids, whereupon a change in the lipid content of the diet from pup to adult created an N enrichment gradient which was detected within bone collagen (Newsome et al., 2006). Other commonly used gradients exist including variation in soil nitrogen (due to fertiliser application) which can generate very small scale spatial gradients across terrestrial systems or within aquatic systems were anthropogenic inputs, such as sewage outfalls lead to nitrogen enrichment within specific areas (Anderson & Cabana, 2006).

Stable hydrogen isotopes (2H : 1H, expressed as δ2H or δD) and, less commonly, stable oxygen isotopes (18O : 16O, expressed as δ18O) are often used in studies of animal movements, particularly in terrestrial systems, as they both vary predictably with latitude, altitude, distance from sea, season and the amount of precipitation (Hobson & Wassenaar, 1997; Bowen, Wassenaar & Hobson, 2005; Levin et al., 2006) due to isotopic discrimination during phase changes (e.g. evaporation or condensation). This variation is reflected in consumer tissues via both diet and drinking water (Hobson, Atwell & Wassenaar, 1999b). Stable hydrogen isotope ratios have therefore become the tool in studies of long-distance migrations, particularly for many bird populations (Kelly et al., 2002; Rubenstein & Hobson, 2004).

In some circumstances, the most commonly used isotopes do not provide enough discriminatory power to address the questions of interest. In these circumstances, the addition of other elements to the stable isotope profile may often provide further resolution. For example, in questions relating to differential use of marine v terrestrial habitats the addition of stable sulphur isotopes (34S : 32S expressed as δ34S) may aid interpretation. There is limited fractionation when sulphur compounds are being incorporated into the food chain and therefore in certain circumstances δ34S provides enhanced resolution (Nino-Torres et al., 2006). Finally, stable strontium isotopes (87Sr : 86Sr expressed as δ87Sr) vary with the age of the underlying geology of a region and have most frequently been used to monitor the movements of fish populations (Milton & Chenery, 2005), although they have also been used to investigate the migration of birds (Chamberlain et al., 1997; Hobson, 1999).

INFERRING DIET/HABITAT SELECTION USING ISOTOPE MIXING MODELS

Once isotopic values are obtained from both the consumer's tissue and that of its potential prey, it is often possible to quantify the consumer's intake of the various prey items or its relative use of different habitats, using one of an expanding selection of isotopic mixing models. However, there are several assumptions that need to be met before the relative contributions from each dietary component/habitat (source) can be determined. First, each prey category or source has to be isotopically distinct; if this is not the case, then it may be required to combine different sources in isotopically distinct categories. In cases where the number of prey categories is small, it is often possible to use simple linear mixing models. These have been applied in studies of mammals, but some caution is advised. For example, they were used to demonstrate spatial variation in the amount of salmon that wolves Canis lupus were consuming in different parts of Alaska (Szepanski, Ben-David & Van Ballenberghe, 1999). However, the approach applied in this instance is based on Euclidean distances and as such tends to underestimate the importance of common prey items and overestimate the importance of rarely used prey items. Phillips & Gregg (2001) pointed out that the errors associated with this are not insubstantial and proposed an alternative (IsoError software available at http://www.epa.gov/wed/pages/models/stableIsotopes/isotopes.htm). The other drawback with these models is that there is no unique solution in cases where there are greater than n + 2 dietary items, where n is equal to the number of stable isotope ratios measured.

Adding extra isotopes to the analysis is a potential solution to the problem of ‘coping with too many sources’, but this can substantially increase cost, without guaranteeing any extra resolution. This limitation lead Phillips & Greg (2003) to develop IsoSource (web details above); a probabilistic mixing model which calculates all potential dietary combinations that could produce a consumer tissue value. This allows the range and frequency of all possible dietary contributions to be determined and is an excellent qualitative and semiquantitative tool (Urton & Hobson, 2005; Benstead et al., 2006; Cardona et al., 2007). However, dealing with the IsoSource outputs statistically is by no means easy, since there is no reason to prefer one solution over another. This can be a problem in cases where specific questions on the consequences of dietary variability are of interest. In such situations, it may be possible to rank some measures of the distribution or carry out a posteriori aggregations (Phillips, Newsome & Greg, 2005) or potentially use alternative approaches such as Moore Penrose pseudoinverse matrices (Hall-Aspland, Hall & Rogers, 2005).

WHY USE STABLE ISOTOPES?

When it comes to inferring diet, stable isotope approaches have several attributes that make them attractive to ecologists. First, since different tissues integrate diet over different temporal scales, several distinct sources of information can be accessed from a single sampling event. For example, carbon and nitrogen isotopes were measured in the blood and fur of the red fox Vulpes vulpes and coyote Canis latrans within an agriculture setting in Illinois (Lavin et al., 2003). Blood (the short-term indicator) was used to show a localized habitat effect showing the young fox pups foraging opportunistically around the den, whereas fur (the long-term indicator) provided the necessary temporal period to show that the adult foxes had adjusted their niche width, as a direct response to the additional competition for resources from the coyote. To achieve this insight conventionally would have been practically impossible and at the very least require a huge amount of time and sampling effort.

Second, if the phenology of growth is known, collection of keratinized tissues can provide dietary information from elusive stages of a mammal's life history, which may otherwise remain unknown (i.e. during months when mammals are away from breeding areas or are difficult to capture). This feature has been used to great effect in understanding mammalian behaviour. For example, researchers have analysed the stable isotope ratios of fur to reveal hoary bat Lasiurus cinereus migration (Cryan et al., 2004); pinniped vibrissae to track temporal shifts in their diets (Hirons, Schell & Finney, 2001); and baleen to track the migrations of whales (Hobson & Schell, 1998).

Third, stable isotopic characterization of diet generates a quantitative and continuously distributed variable, which makes for easier statistical analyses and construction of predictive models (Felicetti et al., 2003; Urton & Hobson, 2005; Inger et al., 2006). This becomes a particularly powerful tool when resolving issues of specialized diets (McIlwee & Johnson, 1998; Felicetti et al., 2003), especially useful in cases where the particular significance of scarce protein sources has a direct impact on species conservation. For example, it was known that the endangered Yellowstone grizzly bear Ursus arctos horribilis fed upon the nuts of whitebark pine Pinus albicaulis, itself an increasingly vulnerable species. However, the significance of this scarce source of dietary protein to the bears was unknown. These pine nuts have a distinct sulphur isotope signature, which was utilized in this study as a tracer of its nutritional importance to bear diet (Felicetti et al., 2003). It was shown using mixing models (Phillips & Gregg, 2003) that in years of good pine nut production this food resource provided over 51% of their assimilated sulphur and nitrogen, to the majority of bears sampled (approximately 67%). This proportion was significantly reduced in years of low pine production, whereupon their main protein source was animal protein. The stable isotope technique applied in this study highlighted the importance of pine nuts and meat to individual species, providing a means to understanding the population dynamics as individuals change their food sources. These three basic principles underpin the use of stable isotope technology as an effective and efficient ecological recorder (West et al., 2006) particularly where conventional approaches alone have provided less resolution.

STABLE ISOTOPES AS DIETARY TRACERS IN MAMMALIAN ECOLOGY

As described previously, in many studies of mammal feeding behaviour, stable isotopes have simply been used to provide a quantitative description of diet. This has been especially effective when understanding the influence of marine allochthonous inputs to the dynamics of terrestrial mammals and in particular quantifying the impact of invasive species within island systems (Hobson, Drever & Kaiser, 1999a; Drever et al., 2000; Stapp & Polis, 2003a,b; Major et al., 2007). For example, this technique provided evidence for several studies that introduced Norway rats Rattus norvegicus were dependent on the input of energy from marine sources, which included seabird tissues (Hobson et al., 1999a; Major et al., 2007), a finding that indicated the rats might be responsible, at least in part, for the decline of some of the local seabird populations. The extent of this type of foraging specialization has been shown to be correlated to the size of the island (Stapp & Polis, 2003b) and weather conditions (Stapp & Polis, 2003a), showing that when rodents were subjected to limited feeding resources their population declined. However, within a different island community, Norway rats have been shown to have a highly variable diet and therefore due to this plasticity, they are expected to survive and flourish even when their preferred diet choice is extirpated (Major et al., 2007).

The ability to use SIA to quantify prey items over differing temporal and spatial scales has allowed greater resolution of prey switching, scavenging or seasonal changes in foraging behaviour (Darimont & Reimchen, 2002; Roth, 2003; Ben-David et al., 2004). Obtaining empirical evidence to determine prey switching and the dynamics of energy flow within ecosystems has greatly improved our understanding of the interactions within communities and how these change over time. This work has led to further research more focused on how introduced prey affects the competition for resources among top predators. An example of the power of SIA in elucidating some of the more subtle impacts of predation is demonstrated by a study of predator prey systems on the California Channel Islands (Roemer et al., 2002). The recent colonization of Santa Cruz Island by golden eagles Aquila chrysaetos was largely a consequence of the introduction of pigs Sus scrofa, which provided them with a non-limiting food supply. The colonization event altered trophic relationships on the island, leading to hyperpredation on the native fox Urocyon littoralis by golden eagles, which were heavily subsidized by the pigs. This hyperpredation caused a decline in the fox population and a subsequent increase of its direct competitor, the spotted skunk Spilogale gracilis amphiala. Stable carbon and nitrogen isotope values, provided empirical evidence of the prey consumed and dietary overlap of the eagle, fox and skunk and, in combination with energetic models, provided unique evidence of trophic reorganization.

The factors that determine niche breadth and the consequences of variation in this parameter have been of considerable interest to ecologists for some time (Bolnick et al., 2002; Matthews & Mazumder, 2004). An important component of the expression of population niche breadth at the population level is the extent to which individuals are generalists or specialists (Bolnick et al., 2002). Again SIA offers a number of opportunities for addressing this type of question. Individual specialization has been shown within a generalist herbivore, the dusky-footed woodrat Neotoma fuscipes, using stable carbon and nitrogen isotope ratios to parameterise multiple-source mixing models (McEachern et al., 2006). Two populations of woodrats specialized on different local plant species, Juniperus occidentalis and Calocedrus decurrens, respectively, even though these were in connected habitats. This type of behaviour was possibly driven by continued exposure to specific plant toxins, creating individual specialists within this generalist population. This example used SIA of a single tissue to identify population level specializations.

Bearhop et al. (2004) have proposed the determination of trophic niche width using either serial tissue sampling or sampling different tissues with varying turnover rates as a mode of investigating both individual and population level specializations. The assumption here is that isotopic variance observed within a slow growing tissue from a single individual (i.e. along a whisker), or among samples of the same tissue collected serially from the same individual (e.g. blood plasma) is a product of the animal foraging on isotopically distinct (different) prey items over time. This type of approach was used to understand the extent of variation among the diet of individuals within different populations of grey wolf Canis lupus found in the boreal forest system of Saskatchewan Canada (Urton & Hobson, 2005). It was expected that wolves, which forage cooperatively, would have similar stable carbon and nitrogen isotopes within slow-growing guard hairs; however, there was a great deal of individual variation. This could have been affected by pack stability or variable social status of the individuals; for example, lower ranking individuals, who only get leftovers of pack kills, may subsist their diet with other small mammals causing an increase in their isotopic niche width and hence greater individual variation.

Such intra-population variability in niche has also been linked to the fitness of individual animals and their susceptibility to predation (Darimont, Paquet & Reimchen, 2007) and differing resource partitioning between sexes (Lewis et al., 2006). In the former study, isotope approaches have revealed the relationship between individual variation in the dietary niches of black-tailed/mule deer Odocoileus hemionus and the risk of wolf predation. Those individuals foraging in a nutritionally rewarding area were more likely to be killed by wolves, with those opting for a more generalized less rewarding foraging strategy having greater survivorship, indicating resource-specific fitness (Darimont et al., 2007).

Of course, using SIA as a trophic marker is not limited to studies of populations and is readily extended to levels of the community. Resource partitioning among species within an assemblage is proposed to be a major driver of speciation (Van Valen, 1978; Grant & Grant, 1994). However, testing this theory using conventional approaches, such as comparative anatomy (Grant & Grant, 1994; Ernest, 2005) or analysis of food remains (McDonald, 2002), has its limitations and SIA has certainly highlighted some of these (Stewart et al., 2003; Cerling, Harris & Passey, 2003; Cerling, Hart & Hart, 2004; Codron & Brink, 2007). For example, preliminary work on resource partitioning among three sympatric herbivores using conventional approaches suggested significant feeding niche overlap (Stewart et al., 2003). However, δ13C and δ15N analyses of faeces demonstrated that there was dietary segregation among the three species with minimal habitat partitioning. Mule deer, the smallest species, showed greater variability and foraged opportunistically on poorer-quality food sources found in xeric areas, while North American elk Cervus elaphus and cattle Bos taurus had more specialized diets, focusing on forbs and grasses, respectively, more commonly found in mesic areas.

The above examples provide evidence of how stable isotopes can be used to provide a suite of foraging information, from basic foraging behaviour through to complex community level interactions, often providing insights that would be extremely difficult to obtain by alternative means. Indeed, very recent work has indicated that SIA may be able to provide us with a number of further community metrics that can provide quantitative measures of trophic structure (Layman et al., 2007). These measures of isotopic niche space, niche compactness and trophic redundancy will potentially yield exciting new insights into the factors shaping community structure.

APPLICATIONS USING STABLE ISOTOPES IN ECOTOXICOLOGICAL STUDIES

Given that many contaminants in animal tissues are ingested via the diet, SIA offers considerable potential for understanding the dynamics of these materials within and among individuals/species and potentially provide critical information on how they enter foodwebs. In this respect, the stepwise enrichment of δ15N with trophic level is particularly useful, since many of these contaminants are thought to bioamplify or biomagnify along food chains and δ13C can potentially provide information on the contaminant sources, particularly in marine systems.

For example, the rate of biomagnification through the food web can be determined by comparing the stable nitrogen isotopic value and contaminant value regression slopes, of different species (including mammals) within a food chain (Borga et al., 2004; Campbell et al., 2005). Much of the mammal-related research has focused on the Arctic, as it tends to be a sink for many contaminants, coupled with the potentially detrimental effect to indigenous human populations. Thus, mercury has been shown to biomagnify within the Northwater Polynya Arctic marine food web, shown through positive correlations between stable nitrogen and the total mercury levels (Campbell et al., 2005), with ringed seals Phoca hispida foraging at high trophic levels and having very high total mercury burdens in their body tissues; a finding consistent with other studies in the region (Atwell, Hobson & Welch, 1998). However, other potentially harmful heavy metals, such as cadmium, showed no relationships with δ15N, with piscivorous ringed seals having similar profiles to planktivorous little auks Alle alle indicating that biomagnification is not a major driving force of variation in the levels of these metals in animal tissues (Campbell et al., 2005; Dehn et al., 2006). Stable carbon isotopes can also be used to determine foraging strategies, which can then be linked to contaminant levels. For example, within the marine environment, stable carbon δ13C values were used to infer inshore vs. offshore foraging of the harbour porpoise Phocoena phocoena found within the Black Sea, showing that those feeding offshore had lower contaminant levels than those found inshore (Das et al., 2004).

However, there are potential confounding factors in studies of this nature (Jardine, Kidd & Fisk, 2006). For example, catabolism of body proteins as a consequence of starvation or reallocation of resources during reproduction can result in shifts in tissue δ15N signatures (Fuglei et al., 2007). These shifts (particularly during starvation) will often be correlated with changes in the amount of contaminants in tissues, relative to body mass (as mass is lost during starvation). This would increase the strength of the correlation between δ15N and a particular contaminant, irrespective of any trophic level effect. Examination of C : N ratios (higher C : N ratios would suggest that N is in short supply and thus indicate potential nutrient limitation) or coupling the bulk SIA with compound-specific SIA (Macko et al., 1987; Popp et al., 2007), may provide greater resolution in these situations.

USING STABLE ISOTOPES AS SPATIAL MARKERS

There are well-described spatial gradients in the stable isotope ratios of several elements. These gradients exist at a variety of spatial scales and have been used most extensively by avian ecologists, particularly those investigating the migration of small songbirds that are difficult to track by other means (Hobson, 1999, 2005; Webster et al., 2002; West et al., 2006). The starting point for any study of this nature is to ensure that the isotopic gradient exists over the scale of the migratory or dispersal movement (for some elements such as H and O, maps are available at http://www.waterisotopes.org). This application of the stable isotope technique has also been used to track other animal groups from invertebrates (Morgan et al., 2006), through fish (Litvin & Weinstein, 2004; Guelinckx et al., 2006) to migratory mammals such as whales and bats (Best & Schell, 1996; Hobson & Schell, 1998; Cryan et al., 2004).

Isotopic gradients tend to be better characterized within terrestrial systems when compared with those within marine systems. This is generally because of oceanic mixing, which leads to much more spatially homogeneous signatures and therefore stable isotopes tend to be less effective tracers of the movement of marine animals. However, there are some exceptions, most notably the discontinuities in δ13C at oceanic fronts and the gradual decrease in δ13C with increasing latitude (Cherel & Hobson, 2007). Thus, it has been possible to use stable isotopes to infer the movements of mammals (Webster et al., 2002), in both marine (Best & Schell, 1996; Hobson & Schell, 1998) and terrestrial (Fleming, Nunez & Sternberg, 1993; Cryan et al., 2004) ecosystems. Migration of cetaceans has been demonstrated very effectively using stable isotopes (Schell, Saupe & Haubenstock, 1989; Best & Schell, 1996; Hobson & Schell, 1998; Mitani et al., 2006). In most cases, researchers have sampled sequentially along the length of baleen plates. These plates growing continuously and thereby effectively provide a time series of the animal's movements over many years. All of the studies above have utilized predictable and consistent isotopic gradients in δ13C that exist across the migratory regions. For example, the McKenzie River Delta, a brackish area, and the Bering and Chukchi Seas, pure marine areas, each have distinct δ13C signatures, which have allowed researchers to confirm the seasonal migrations of Arctic Bowhead whales Balaena mysticetus (Schell et al., 1989; Hobson & Schell, 1998). These movements are reflected in the fluctuations of isotope signatures along the length of baleen plates. Isotopic studies have also revealed that the majority of the annual food requirements of adult Bowheads are acquired during the winter period from the Bering-Chukchi Sea, whereas sub-adults show a slightly more variable diet, foraging also within the McKenzie River area (Lee et al., 2005). Similarly, Fleming et al., (1993) used analysis of δ13C in the muscle of migratory nectivorous bats Leptonycteris curasoae to prove their migration was timed to coincide with the flowering of columnar cacti (Fleming, Nunez & Sternberg, 1993).

Stable hydrogen ratios of precipitation are often used as indices of migratory origins within terrestrial systems and isotopic maps have been produced and utilized for Europe (Hobson et al., 2004; Bearhop et al., 2005), America (Bowen et al., 2005) and Africa (Yohannes et al., 2005). However, such maps while intuitively appealing, in most instances, should be used only as indicators of potential gradients and it is important that patterns are ‘ground truthed’ before embarking on any study, via collection of appropriate tissues across the geographical region(s) of interest.

Stable hydrogen isotope ratios have provided evidence that California and Mexican populations of hoary bats are migrating up to 2000 km north during spring (Cryan et al., 2004). Moulted hair was collected throughout their migratory range, which determined that there was a highly significant relationship between δD of precipitation (δDP) and the δD of bat hair (δDH) sampled in the same region, with δDP explaining around 60% of the δDH variance during the moult period. The difference between the δDH collected outside the moult period and the corrected δDP of the region where the sample was collected provided evidence of their long distance migration. Understanding the migratory nature of many bat species' has proved difficult due to the weight and size of satellite transmitter technology, making it almost impossible to track these animals over long distances and although banding has provided some qualitative information the probability of recovery is extremely low. Thus, at this point in time stable hydrogen isotope analysis is one of the few viable techniques that can reveal these aspects of their behaviour.

This approach has been taken beyond simply tracking animals, to address important questions on how events across the entire annual cycle interact upon fitness (Marra, Hobson & Holmes, 1998; Bearhop et al., 2005; Hobson, 2005; Lee et al., 2005). For example, during the winter months, migratory American redstarts Setophaga ruticilla exhibit a behaviour known as dominance-mediated habitat segregation, where dominant individuals (mostly males) exclude subordinates from high-quality habitats (Marra, Hobson & Holmes, 1998). These habitats have quite distinct δ13C signatures, most likely as a consequence of differences in the proportions of C4 plants. Individuals occupying high-quality areas had lower δ13C signatures and tended to arrive at breeding colonies earlier than their conspecifics, obtaining a significant advantage over those wintering in scrub areas, which arrived later. Similar small-scale gradients can be exploited within certain areas where stable δ15N signatures vary with aridity. Chronological δ15N signatures found within African elephant Loxodonta africana hair were initially used to determine ranging behaviour of these mammals; however, this was coupled with δ13C signatures (C3/C4 gradient), determining foraging behaviour (Cerling et al., 2006). In this instance, crop signatures were distinct from the other vegetation types used by the elephants, and thus it was possible to determine the frequency of crop raiding events, which is becoming an increasingly important issue in many parts of Africa.

There is clearly considerable scope for extending applications of this nature to the movements of other mammals. Although only a few mammal species undertake lengthy migrations compared with those of birds, the latter examples demonstrate how the isotopic gradients that exist over much smaller scales are well within the range of the dispersal movements of many mammal species. As such, it may be possible to investigate the causes and consequences of dispersal movements in a range of mammals with relative ease.

SOME CAVEATS OF STABLE ISOTOPE APPROACHES

No review of this type would be complete without a note of caution. There are a number of caveats associated with this approach and such is the appeal of the technique, these are often over looked. The most common sources of error include variability in the isotopic fractionation values across different combinations of diets and tissues/species, unquantified temporal or spatial variation in prey isotopic values and variation caused by routing of particular dietary nutrients into particular tissues (Gannes, O'Brien & delRio, 1997; Bearhop et al., 2002; Howland et al., 2003).

Although numerous controlled studies (Table 2) have generally found fractionation values very close to the commonly used 1‰ for carbon and 3‰ for nitrogen, there are several factors that can cause substantial deviations from these. These deviations can cause problems, particularly when using mixing models to estimate dietary contributions, as errors in the estimation of diet-tissue fractionation factors are propagated leading to incorrect outputs. If possible, controlled experiments should be conducted on the species of interest, by feeding animals an isotopically consistent diet to obtain species-specific fractionation values (Roth & Hobson, 2000; see Vanderklift & Ponsard, 2003). If this is not possible, then the sensitivity of any conclusion/model outputs should be assessed by varying diet-tissue fractionation factors.

Table 2.  Diet, tissues sampled and mean δ13C and δ15N fractionation factors (Δ‰) for various species of mammal
 MammalDietTissueΔδ13C
(‰)
Δδ15N (‰)Reference
RodentsMouse*Wayne Laboratory-Blox F6Hair−1.13.1DeNiro & Epstein (1978, 1981)
Mouse*Wayne Laboratory-Blox F6Liver1.34.1DeNiro & Epstein (1978, 1981)
Mouse*Wayne Laboratory-Blox F6Muscle1.62.7DeNiro & Epstein (1978, 1981)
Mouse*JAX 911AHair23.2DeNiro & Epstein (1978, 1981)
Mouse*JAX 911ALiver1.44.7DeNiro & Epstein (1978, 1981)
MouseMixed diet (casein, beet sugar, soybean oil)Blood1.23.0MacAvoy et al. (2006)
VoleAlfafa-basedEnamel11.5 Passey et al. (2005)
VoleAlfafa-basedBreath0.3 Passey et al. (2005)
VoleAlfafa-basedEnamel11.5 Passey et al. (2005)
RatMixed diet (casein, beet sugar, soybean oil)Blood1.62.9MacAvoy et al. (2006)
Marine mammalSeal (Harp, Harbour & Ringed)HerringSkin2.82.3Hobson et al. (1996)
Seal (Harp, Harbour & Ringed)HerringWhisker3.22.8Hobson et al. (1996)
Seal (Harp, Harbour & Ringed)HerringNail2.82.3Hobson et al. (1996)
Seal (Harp, Harbour & Ringed)HerringHair2.83.0Hobson et al. (1996)
Seal (Harp, Harbour & Ringed)HerringBlood1.71.7Hobson et al. (1996)
Seal (Harp, Harbour & Ringed)HerringMuscle1.32.4Hobson et al. (1996)
Seal (Harp, Harbour & Ringed)HerringLiver0.63.1Hobson et al. (1996)
Northern Fur Seal(Herring & Capelin)RBC1.354.1Kurle (2002)
Northern Fur Seal(Herring & Capelin)Plasma1.05.2Kurle (2002)
Terrestrial carnivoreRed foxPellet foodSerum0.64.2Roth & Hobson (2000)
Red foxPellet foodRBC0.72.6Roth & Hobson (2000)
Red foxPellet foodLiver0.43.6Roth & Hobson (2000)
Red foxPellet foodMuscle1.13.6Roth & Hobson (2000)
Red foxPellet foodHair2.63.4Roth & Hobson (2000)
MinkBeefPlasma 4.0Ben-David & Schell (2001)
Polar bearRinged sealPlasma 3.8Hobson & Welch (1992)
Terrestrial herbivoreAlpaca*Alfalfa (high P)Hair3.26.4Sponheimer et al. (2003a,b)
Alpaca*CBG (low P)Hair 3.7Sponheimer et al. (2003b)
Cattle*Alfalfa (high P)Hair2.74Sponheimer et al. (2003a,b)
Cattle*CBG (low P)Hair 2.8Sponheimer et al. (2003b)
SteerDifferent C3/C4 switchesBreath2.9 Passey et al. (2005)
SteerDifferent C3/C4 switchesEnamel14.6 Passey et al. (2005)
PigDifferent C3/C4 switchesBreath1.8 Passey et al. (2005)
PigDifferent C3/C4 switchesEnamel13.3 Passey et al. (2005)
PigMixed dietCollagen2.9 Howland et al. (2003)
PigMixed dietApatite10.2 Howland et al. (2003)
PigMixed dietLipids−2.4 Howland et al. (2003)
Goat*Alfalfa (high P)Hair3.25.1Sponheimer et al. (2003a,b)
Goat*CBG (low P)Hair 2.7Sponheimer et al. (2003b)
Horse*Alfalfa (high P)Hair 4.5Sponheimer et al. (2003b)
Horse*CBG (low P)Hair 1.9Sponheimer et al. (2003b)
Llama*Alfalfa (high P)Hair3.54.6Sponheimer et al. (2003a,b)
RabbitAlfalfa-basedEnamel12.8 Passey et al. (2005)
Rabbit*Alfalfa (high P)Hair3.42.8Sponheimer et al. (2003a,b)
RabbitAlfalfa-basedBreath1.0 Passey et al. (2005)
InsectivoreNectarivorous batSoya (low C : N ratio)Whole blood0.13.3Miron et al. (2006)
Nectarivorous batAmaranth (high C : N ratio)Whole blood2.04.4Miron et al. (2006)
Nectarivorous bat (2 spp.)*C3 plants (high C : N ratio)Whole blood2–1.63.0–3.2Voigt et al. (2003); Voigt & Matt (2004)
Nectarivorous bat (2 spp.)*C3 plants (high C : N ratio)Wing membrane3.2–2.74.7–4.0Voigt et al. (2003); Voigt & Matt (2004)
Nectarivorous bat (2 spp.)*C3 plants (high C : N ratio)Hair3–2.6Voigt et al. (2003); Voigt & Matt (2004)
Omnivorous marsupialLong-nosed bandicootC3/C4Blood plasma1.42.8Klaassen et al. (2004)
Long-nosed bandicootC3/C4Blood cells−0.22.1Klaassen et al. (2004)

One of the major drivers of this variability is the difference in the extent to which dietary isotope ratios are fractionated as they are allocated to different tissues (DeNiro & Epstein, 1978; Cerling & Harris, 1999; Klaassen, Thums & Hume, 2004; Sponheimer et al., 2006). Other factors that may lead to variability in diet-tissue fractionation include differences in nutritional intake (Roth & Hobson, 2000; Kurle, 2002; Voigt & Matt, 2004; Robbins, Felicetti & Sponheimer, 2005; Miron et al., 2006), age (Roth & Hobson, 2000; Jenkins et al., 2001), metabolic rate (MacAvoy, Arneson & Bassett, 2006) and reproductive state (Kurle, 2002). Also, different external environmental pressures or nutritional content of food sources, as examples, can lead to differences in the assimilation and storage of nutrients among tissue types; a process referred to as metabolic routing (Schwarcz, 1991). For example, a controlled dietary experiment on pigs found that when fed various diets, with different compositions of amino-acids, etc., diet-tissue fractionation factors for pig bone collagen varied from 0.5‰ to 6.1‰ (Howland et al., 2003).

The consequences of applying inaccurate fractionation factors are exemplified by a study of the foraging ecology of a group of bat species. In this study, carbon and nitrogen isotopes were used to determine the nutritional importance of fruit and insects, within the blood of two species of frugivorous bats Artibeus jamaicensis and Sturnira lilium and one species of insectivorous bat Pteronotus parnellii (Herrera et al., 2001). A dual-isotope mixing model was used to investigate this and standard fractionation values were used (1‰ for carbon and 3‰ for nitrogen). The model estimated that the frugivorous bats consumed a significantly higher proportion of insects than fruits, with the insectivorous bat consuming more fruit than insects. Behavioural data indicated that these results could not be correct and were most likely a consequence of incorrect diet-tissue fractionation factors. The problem was overcome by combining the two sources of information in a sensitivity analysis, varying fractionation values in order to determine the most likely solution.

Variations in diet-tissue fractionation can also make for equivocal interpretations of isotopic data. The eastern Arctic bowhead whale population is found within Northern Hudson Bay, eastern Lancaster Sound and Baffin Bay area. SIA of samples collected along baleen plates showed little isotopic difference in δ13C but there were distinct oscillations in the δ15N values (Hobson & Schell, 1998). This left the researchers with three equally plausible interpretations: first, the population might migrate from areas that are similar in δ13C values, but distinct in δ15N; second, the whales switch to prey of differing trophic levels across seasons; third, the shift in δ15N may be linked to seasonal fasting, where body stores are being remobilized to fuel metabolism. In the absence of any other information, none of these hypotheses can be favoured over the others. Compound-specific isotopic analyses of essential and non-essential amino acids offer considerable scope with respect to resolving such issues. For example, while δ15N signatures of non-essential amino acids of consumer tissues increase with trophic level, there appears to be little change in the δ15N of essential amino acids (Popp et al., 2007). Thus, by comparing the δ15N of essential amino acids to the δ15N of bulk tissue, it is possible to discriminate between dietary changes and shifts in nutrient regimes.

Inadequate sampling of the consumer's diet is another problem that often makes for equivocal interpretation of stable isotope signatures. If some dietary items are missed, then mixing models will often not find any feasible solutions, or produce spurious results. Moreover, the isotopic signatures of prey items may often vary temporally (e.g. as the diet of prey shifts). As such, in some situations, it may be important to sample prey from the time period that the consumer tissue will integrate over. Thus, if the tissue chosen or available is blood plasma, then ideally prey items should be collected over 3–7 days prior to consumer tissue sampling. Clearly this is not always possible, but some attempt should be made to try and account for such variation, even if just to remove it as a possible explanation for the patterns observed in consumer tissues.

The use of stable hydrogen isotopes as spatial markers presents several unique problems. First, unlike other stable isotopes, a proportion of the stable hydrogen isotopes found in keratin-based tissues freely exchange with variable ambient hydrogen. This problem of exchangeable hydrogen leads to the development of a global protocol (Wassenaar & Hobson, 2003) for standardizing this variability with the aim of allowing direct comparison of values across studies. Second, in addition to the large-scale patterns described earlier, there are a number of small-scale heterogeneities. Stable hydrogen isotope ratios also vary with altitude (Hobson et al., 2003, 2004), and localized anomalies in water sources, for example deep aquifers (Boronina et al., 2005) or evaporative environments. In addition, physiological processes may also generate unwanted variation. For example, the δD signatures in feathers of wood thrushes Hylocichla mustelina are higher than would be expected based on predictions from precipitation values (Powell & Hobson, 2006). It is thought that these apparent anomalies may be linked to the birds losing large amounts of 1H during bouts of evaporative cooling during periods of heat stress. It is therefore evident that sampling the isotopic ratios of the tissue of an individual will not necessarily provide the evidence needed to assign it to a geographically distinct area (Wunder et al., 2005) and therefore careful project and sampling designs are required.

CONCLUSIONS

The application of stable isotope techniques to mammalian ecology has met with considerable success and offers considerable scope for the future, with the potential to allow us to gain new insights to a suite of ecological processes. Sampling different tissues from a single individual gives a unique opportunity to quantify dietary inputs over varying temporal and spatial scales, from one sampling episode. The studies discussed here emphasize the strengths of using stable isotope analyses to gain a better understanding of individuals, populations and ecosystems. The opportunities that exist for this technique in the fields of mammalian ecology are substantial and are likely to stimulate new lines of investigation. The ease of quantifying diet and dietary variability using SIA means that it is particularly amenable to investigations of individual and population level niche specialization and ultimately the analysis of resource partitioning within species assemblages. With respect to more applied problems, it is clear that ability to understand and quantify the impact of introduced species is a major focus for conservation biologists and SIA has the potential to be an important tool in the modelling of future scenarios. While there is a clear need for further controlled experiments, published details of specific protocols and archiving geographical isotopic values for future ecological and forensic studies (West et al., 2006), the unique information that this technique can provide on foraging ecology, migratory behaviour, pollution ecology and population dynamics, means that SIA is likely to become an increasingly important addition to the mammalian ecologist's toolbox.

ACKNOWLEDGEMENTS

We thank Richard Inger for helpful discussions on applications of SIA and an anonymous referee for many useful comments. Kerry Crawford is supported by a research studentship from the Department of Agriculture and Rural Development, Northern Ireland.

Ancillary