Are elevated δ15N values in herbivores in hot and arid environments caused by diet or animal physiology?


Correspondence author. E-mail:


1. Mammalian species from hot and arid environments often have elevated nitrogen isotope values compared to animals from similar trophic levels in more temperate climates. This pattern has most often been explained as the result of a physiological response by animals to heat and water stress. However, a positive correlation between rainfall and the δ15N values of plants and herbivorous mammalian species suggests that diet may be responsible.

2. This study uses the horn keratin of desert-adapted bovids (Dorcas gazelle and Nubian ibex) to test whether the δ15N values of herbivore body tissues are determined by heat and water stress or by the isotopic composition of their diet. The δ15N values of horn keratin are compared against several climatic factors that affect heat and water stress to determine if a relationship exists. In addition, the range of δ15N values measured in desert vegetation is used to evaluate the contribution of the diet to the isotopic values of bovid body tissue.

3. The δ15N values of desert bovid horn keratin were correlated against individual climatic factors that induce water stress and were not found to be significant. When climatic factors were combined, a significant positive correlation was found between the δ15N values of Dorcas gazelles and temperature, humidity, and rainfall. This observation contradicts the physiological stress hypothesis that predicts a negative correlation between rainfall and humidity and δ15N values. Instead, this correlation is likely attributable to denitrification processes in the soil that directly affect the isotopic values of the plants. Values for δ15N of horn keratin fall within the range predicted by discrimination between diet and consumer (Δ = 4·1‰) which supports the diet hypothesis.

4. The results suggest that the isotopic composition of the diet is the dominant factor determining the δ15N values of herbivore body tissue.


The δ15N values of herbivore body tissue have been shown to correlate negatively with rainfall (Sealy et al. 1987; Gröcke, Bocherens & Mariotti 1997; Murphy & Bowman 2006; Pate & Anson 2008; but see also Cormie & Schwarcz 1996). In particular, high δ15N values have been observed in herbivores inhabiting hot and arid environments (Heaton et al. 1986; Schwarcz, Dupras & Fairgrieve 1999).

Three physiological and one dietary hypothesis have been proposed to explain the relationship between the δ15N values of herbivore body tissue and rainfall in arid regions (Ambrose & DeNiro 1986, 1987; Sealy et al. 1987; Sponheimer et al. 2003; Balter et al. 2006; Murphy & Bowman 2006). The first physiological hypothesis states that drought-tolerant animals consuming high protein diets improve water conservation through protein catabolism. The result is the enrichment of δ15N values in herbivore body tissues (Ambrose & DeNiro 1986, 1987). The second physiological hypothesis states that plants growing in arid environments have lower nitrogen content than those in wetter environments. Consequently, the herbivores that feed on these plants must increase their reliance on amino acids produced by symbiotic gastrointestinal bacteria. The δ15N value of herbivore tissue increases owing to the consumption of protein produced by bacteria that is one trophic level higher than plant protein (Sealy et al. 1987). The third physiological hypothesis states that increased δ15N values are likely a product of improved water recycling and reduced nitrogen intake (Balter et al. 2006). According to this hypothesis water stressed mammals incorporate elevated δ15N values from hydrolysed urea into amino acids that are then synthesized into body proteins.

In contrast, it has also been argued that the δ15N of body tissue of herbivores, in this case marsupials, is determined by diet particularly by the isotopic composition of the plants consumed (Murphy & Bowman 2006). This hypothesis is built upon the well demonstrated negative relationship between plant δ15N values and rainfall at both global and regional scales (Austin & Vitousek 1998; Handley et al. 1999; Swap et al. 2004). As plants provide the only exogenous source of nitrogen to primary consumers, it is expected that the relationship between rainfall and plant δ15N values will be transferred to herbivores although with a predictable and consistent trophic effect (Murphy & Bowman 2006).

If physiology is the dominant factor affecting the enrichment of 15N in herbivore body tissues, δ15N values should increase along an aridity gradient regardless of the isotopic composition of local plant communities. Even if the δ15N values of the plant diet independently negatively correlate with aridity, this should be reflected as increased isotopic spacing between diet and body tissue. A number of climatic variables can affect water stress in herbivores and should thus correlate with the δ15N values of herbivore body tissue if the physiological hypothesis is correct. In addition to low annual rainfall, high ambient temperature and low humidity increase evaporative water loss and thus can contribute to water stress in herbivores (Grenot 1992; Cain et al. 2006). The aridity index estimates water availability for crop production and is calculated by dividing mean annual rainfall by evapotranspiration (Middleton & Thomas 1997). As evaporative water loss also affects the water economy of vertebrates (Cortés, Rosenmann & Bozinovic 2000; Nagy 2004), an aridity index that accounts for rainfall, temperature and humidity as well as wind speed and solar radiation can serve as an independent predictor of water stress. In contrast, if diet is the primary determinant of δ15N values, the isotopic composition of herbivores from dry, hot regions should reflect the isotopic composition of the plants that they consume plus a constant trophic effect.

Because the relationship between the δ15N values of plants and rainfall is highly variable in arid regions (Hartman & Danin 2010), these areas provide a natural laboratory to test the effects of the physiological and dietary hypotheses. In the Eastern Mediterranean, the δ15N values of desert plants in the arid and hyper-arid regions (<150 mm year−1) are driven largely by topography which determines both water and nitrogen availability (Hartman & Danin 2010). Plants growing in dry washes have significantly higher δ15N values than plants growing on the slopes or exposed ridges of desert landscapes in association with the combined effect of water availability and high temperatures (Swap et al. 2004; Hartman & Danin 2010). The high δ15N values of dry wash plants fit the relationship between rainfall and the δ15N values of plants across the Eastern Mediterranean rainfall gradient (75–1000 mm) (Hartman & Danin 2010). In contrast, the δ15N values of plants growing on exposed ridges are typically depleted in 15N because of the absence of denitrification processes in the soil and because N-fixing desert crust microorganisms and aeolic deposition are the major nitrogen sources for plant growth (West 1990; Böhlke, Ericksen & Revesz 1997). The δ15N values of desert plants in arid regions thus vary widely among microhabitats regardless of regional climatic conditions that are expected to increase water stress in herbivores.

This study uses δ15N data from desert herbivore horn keratin (Dorcas gazelle and Nubian ibex) and plants from known locations in the Eastern Mediterranean to investigate whether diet or animal physiology best explains the relationship between rainfall and herbivore δ15N values (Fig. 1). The δ15N values of bovids are correlated against local climatic conditions associated with water stress to test the physiological hypotheses. The δ15N values of bovids are then plotted against the expected ranges of δ15N values based on observed plant data assuming a constant positive shift between diet and horn keratin values to test if diet is the primary factor controlling horn keratin values. This study will also examine whether δ15N values of C3 and C4 desert plants typically consumed by bovids is translated to their tissues.

Figure 1.

 Map of the desert regions (shaded in dark grey) of the Eastern Mediterranean including the Sinai Peninsula, Negev Desert and Judean Desert and the collection locations of the bovids used in this study; Dorcas gazelle (•) and Nubian ibex (○).

Materials and methods

This study measured the δ15N values of herbivore body tissue by sampling the horn keratin from Dorcas gazelle and Nubian ibex from arid regions of the Eastern Mediterranean (for the species ecology see Appendix S1, Supporting Information; for collection location and climatic data for individual specimens see Table S1, Supporting information). Keratin rather than bone collagen or tooth dentine was selected for sampling because keratin grows continuously and thus provides finer (seasonal) resolution and can be sampled with minimal damage to curated specimens. Despite differences in tissue turnover and amino acid composition, this study assumed that there is no significant difference between the δ15N values of bone collagen and keratin tissues of the same animal. This assumption was supported by the results of a study presented in Table S2 (Supporting Information) that compared the δ15N values of hoof keratin and rib bone collagen from seven mountain gazelle (Gazella gazella). To assess the contribution of physiology and diet to herbivore δ15N values, the relationship between keratin δ15N values and climatic factors (rainfall, relative humidity, temperature and aridity index) in each animal’s habitat of origin were first assessed. Next, the δ15N values of herbivores were compared to that of plants from the same region.

There is general agreement about the trophic spacing (∼ +4‰) between the δ15N values of herbivores that were not subjected to extreme environmental conditions (i.e. diet quality, heat and water stress) and the δ15N values of their plant diet (Steele & Daniel 1978; Ambrose 1991; Jenkins et al. 2001; Sponheimer et al. 2003; Murphy & Bowman 2006; Mannel, Auerswald & Schnyder 2007). It is expected that similar spacing will exist between herbivores feeding in mesic environments and the plants that they consume (Δ15Ndiet–consumer). To test this hypothesis and to measure isotopic variability within a single population of animals, the δ15N of animals from a closely related domesticated bovid species (domestic goat, Capra hircus) were also studied. The comparative specimens included goats from three organic dairy farms in Israel that foraged on a mixture of natural pasture and supplemented forage. This part of the study estimated within-group variability in the δ15N values of herded herbivores that foraged in the same location.

Horn keratin collection

Keratin horn sheaths from modern Dorcas gazelle (= 27) and Nubian ibex (= 14) specimens curated in the zoological collections at Tel Aviv University and the Hebrew University of Jerusalem were sampled for δ13C and δ15N analysis. The specimens were sampled by shaving approximately 2 mg of keratin off the rough surface at the proximal edge of the horn sheath after the surface was swabbed with a solvent (ethanol 70%). Date of collection, location of collection and sex were recorded for each sampled gazelle and ibex. This sampling strategy was designed to determine the range of isotopic variability (δ13C and δ15N) in Dorcas gazelle and Nubian ibex populations from the arid and hyper-arid regions of Israel and Sinai (Egypt) and its relationship with climatic and plant isotope data from the same regions.

Isotopic variability within individual animals was studied by sequentially sampling the length of the horn sheaths of randomly selected Dorcas gazelles (two males and one female) and Nubian ibexes (three males). The first sample was taken from the base of the horn and subsequent samples were taken at regular intervals away from the base on the dorsal aspect of the horn (= 7). Although there are estimates for Spanish ibex (Capra pyrenaica) (Fandos 1995; Barbosa et al. 2009), no published data on the growth rate of horn keratin for Dorcas gazelle or Nubian ibex exist. Samples were instead taken at regular intervals along the horn sheath (10 mm in Dorcas gazelles and 30 mm in Nubian ibex) taking into account average horn length for both species and the fact that gazelle horns attain full size when the animals reach 2 years of age (Yom-Tov, Mendelssohn & Groves 1995).

In the laboratory, 1 mg of horn sheath keratin from each sample was weighed into a tin boat and sent for isotopic analysis at Boston University’s stable isotope laboratory.

Plant data

The plant data used in this study were derived from Hartman & Danin (2010) and are only briefly reviewed here. Eastern Mediterranean plant communities (= 29) were sampled across rainfall gradients spanning 75–1000 mm year−1 in both the dry (July 2006) and wet season (March 2007). The sample included 12 desert sites (rainfall <150 mm year−1) where samples were taken from exposed ridges or slopes (= 6) and others from dry washes (wadis) (= 6). The desert sampling strategy was designed to determine the range of δ15N values that are expected in plants growing in arid environments and to compare the isotopic values of plants growing on exposed ridge and dry wash sites.

To study the contribution of the plant diet to the δ15N values of herbivore keratin, the relationship between rainfall and the mean δ15N values of C3 and C4 plants was assessed. A comparison was then made between the δ15N values of herbivores and plants that originate from similar environments (i.e. climate, soil types and plant species composition) and receive similar annual rainfall. As mentioned before, this analysis is part of a previous study (Hartman & Danin 2010). The regression equations required to predict the potential contribution of desert plants to the δ15N values of desert herbivores are presented in Table S3 (Supporting information).

Goat farm data collection

The isotopic spacing between herbivores and their diet was studied by sampling domestic goats (= 18; Capra hircus) that forage on natural pasture from three organic dairy farms in the Mediterranean hill region of Israel (dry season, July 2006). Goat herds were followed to the pasture and the plants consumed by mature goats were sampled. Supplemental feed given to the goats was also sampled. The supplemental feed combined carbohydrate (maize) and protein rich (soybean feed mix) food sources. In all of the farms, the supplement feed was offered ad libitum to the goats while in the pen on a daily basis to increase milk yield (Table 1) (Landau, Vecht & Perevolotsky 1993). Hair samples were collected from six mature females from each farm. Like horn sheaths, hair is an inert keratinous tissue that grows incrementally and thus represents the isotopic composition of the nitrogen that was available to the goat when the keratin was formed. Locks of hair were cut about 15 mm from their proximal end (towards the skin) to ensure that the most recent hair was collected. Ten hairs from each goat were selected for analysis. In the laboratory, the hairs were cleaned with a solvent (ethanol 70%) and pulverized to homogenize the sample. One milligram of hair was weighed into tin boats for isotopic analysis. Feed supplement samples were initially lyophilized then ground using a mortar and pestle. Two milligrams of the ground feed was then weighed into tin boats for isotopic analysis.

Table 1.   The isotopic spacing between plant diet and domestic goat hair keratin (= 18; six animals from each of three farms). All values are reported in ‰ (±1SD). For detailed analysis of the goats, plants and supplemental diets, see Table S4 (carbon and nitrogen isotope values and C/N ratios)
Goat farmPlant diet mean δ15NHair keratin meanδ15NΔ15Nplant diet–keratinSupplemented feed δ15N
Har Haruach−0·6 ± 0·52·92 ± 0·223·522·40 ± 1·71
Sataf−1·4 ± 1·32·63 ± 0·254·031·77 ± 1·00
Shabi−0·4 ± 1·14·23 ± 0·434·634·63 ± 1·06
Mean Δ15N plant diet–keratin  +4·06 ± 0·56 

Environmental and climatic data

The contributions of four climatic factors that may potentially influence water stress and diet quality in herbivores were examined: mean annual rainfall, mean daily temperature, relative humidity, and the combined aridity index. If water and heat stress are responsible for the elevated δ15N values measured in desert herbivores, the values should be negatively correlated with rainfall and humidity and positively correlated with temperature and aridity index values (see below). The data for this study were obtained from the Hebrew University of Jerusalem, GIS Centre data base covering the years 1961–1990, excluding the relative humidity data that were taken from Rosenan & Gilead (1985). Rainfall data in the GIS data base were interpolated using inverse distance weighting (IDW). It should be noted that annual rainfall values can vary erratically and can differ markedly from reported mean values in desert regions. Temperature data were interpolated using a combination of regression predictions that accounts for local environmental conditions such as the distance from the Mediterranean Sea in addition to IDW, and Spline methods (Kurtzman & Kadmon 1999).

The aridity index (AI R/ET0) is the ratio of mean annual rainfall (R) and reference evapotranspiration (ET0) (Middleton & Thomas 1997). ET0 was calculated based on the Penman–Monteith equation (Allen et al. 1998) using Aquastat climate information tool ( from the Food and Agriculture of the United Nations. The data for Aquastat climate information tool originated from the CRU CL 2.0 data set (New et al. 2002) and covers observations made from 1961 to 1990.

The contribution of C4 plant diet to the δ15N values of horn keratin

C4 plants are unlikely to be more than dietary supplements in the arid regions of the Eastern Mediterranean given the prevalence of C3 vegetation, and the fact that bovids tend to avoid chenopod species that often have high chlorite and oxalate salt contents (Vogel, Fuls & Danin 1986; Yom-Tov, Mendelssohn & Groves 1995; Palgi, Vatnick & Pinshow 2005). Nevertheless, the consumption of C4 plants may influence herbivore δ15N values (Appendix S1), because the average δ15N value of C4 plants from a given site is approximately 2‰ higher than C3 plants (Hartman & Danin 2010). C3 and C4 plants form fully segregated isotopic groups based on their δ13C values; values for C4 plants are consistently more positive than those measured for C3 plants. Thus, the consumption of C4 plants is expected to increase horn keratin δ15N and δ13C values simultaneously, and will be evaluated by determining whether both are positively correlated.

Statistical analysis

To test the effect of climatic and environmental factors that are most likely to cause physiological stress in herbivores, simple linear and multiple linear regression models were used. The interaction between the δ15N values of horn sheath keratin and the following climatic factors were tested independently and then together: rainfall [natural log transformed (Ln) mean mm year−1]; relative daily humidity (%) and mean daily temperature (ºC). Because the aridity index incorporates a wide range of climatic factors including relative humidity, mean daily temperature and rainfall, it had to be treated separately. The contribution of C4 plants to the diet (δ13C‰) was also treated separately. In all cases the null hypothesis (physiology) is that increased aridity (i.e. a decrease in rainfall and humidity and an increase in temperature and the aridity index) will lead to a corresponding increase in keratin δ15N.

Isotopic analyses

Keratin and bone samples were analysed using an automated continuous-flow isotope ratio mass spectrometer (Preston & Owens 1983). The samples were combusted in a EuroVector Euro EA elemental analyser at Boston University’s stable isotope laboratory. The combustion gases (N2 and CO2) were separated on a GC column, passed through a GVI (GV Instruments Ltd., Manchester, UK) diluter and reference gas box, and introduced into the GVI IsoPrime isotope ratio mass spectrometer. Water was removed using a magnesium perchlorate water trap. Ratios of 13C/12C and 15N/14N were expressed as the relative per mil (‰) difference between the samples and international standards (Vienna PDB carbonate and N2 in air) where:




The sample isotope ratio is compared to a secondary gas standard with an isotope ratio that was calibrated to international standards. For 13CV-PDB the gas was calibrated against NBS 20 (Solnhofen Limestone, −1·05 ± 0·02‰). For 15Nair the gas was calibrated against atmospheric N2 and IAEA standards N-1, N-2 (ammonium sulphate, 0·4 ± 0·2‰ and 20·3 ± 0·2‰ respectively) and NO-3 (potassium nitrate, 4·7 ± 0·2‰). All international standards were obtained from the National Institute of Standards and Technology in Gaithersburg, Maryland, USA.


Nitrogen isotopic spacing between domestic goat hair keratin and their plant diet

There was little variability in the δ15Nplant diet–keratin of the domestic goats with a mean of 4 06 ± 0 56‰ (±1SD) and a range of 3·52–4·63‰ (Table 1). This isotopic spacing agrees with other studies on medium- to large-sized herbivorous mammals (using both bone collagen and hair keratin) that range between 3 and 5‰. The δ15N values of the supplemental feed were higher and more variable across goat farms than those measured for the natural plant pasture (for detailed results see Table S4, Supporting information).

Within horn sheath isotopic variability

The average range of the δ15N values measured along the horns of Dorcas gazelles and Nubian ibex were 0·9‰ and 2·75‰ respectively. The range in the δ15N values of horn keratin collected from a single animal was significantly lower for Dorcas gazelle than Nubian ibexes (t-test, = 0·013) (Table 2; for complete data set see Table S5, Supporting information).

Table 2.   Variation in δ15N values of keratin from individual Dorcas gazelle and Nubian ibex horns. All values are reported in ‰ (±1SD)
SpecimensMean δ15NRangeMean range
G. dorcas
3930 ♀12·38 ± 0·461·190·90 ± 0·30
3931 ♂9·23 ± 0·320·92
6739 ♂9·47 ± 0·230·59
C. ibex nubiana
440 ♂7·49 ± 0·862·782·75 ± 0·70
438 ♂8·10 ± 1·203·43
439 ♂7·03 ± 0·702·04

The relationship between climate and the δ15N values of the horn keratin of desert herbivores

A weak yet significant correlation was found between the δ15N values of Dorcas gazelle horn keratin and mean daily temperature (Fig. 2c). In contrast, the δ15N values of Nubian ibex horn keratin were not correlated with temperature. There was no correlation between the δ15N values of the keratin of either species or rainfall, relative humidity and the aridity index when analysed using simple linear regression analysis (Fig. 2). Multiple regression analyses that combined temperature and rainfall, and temperature and humidity on the δ15N values of horn keratin were significant for gazelle but not for ibex (Fig. 2e,f). When the two species were analysed together, none of the climatic variables were significantly correlated with the δ15N values of horn keratin (Table S6, Supporting information).

Figure 2.

 The relationship between climatic factors and the δ15N values of gazelle and ibex. Gazelle horn keratin (diamonds) indicated by solid regression line; and ibex (x) indicated by dashed regression line. The relationship between the δ15N values of horn keratin and (a) rainfall; (b) humidity; (c) temperature and (d) the aridity index. The relationship between observed δ15N values and the expected values predicted by multiple regression analysis – (e) temperature and humidity and (f) temperature and rainfall. See Table S6 for statistics.

The contribution of desert plant diet to the δ15N values of herbivores

Figure 3 shows the range of δ15N values in the plants potentially available to desert herbivores against rainfall (Fig. 3). The range encapsulates the δ15N values included within 1 standard error (SE) of the regression line formed between rainfall and desert plants (<150 mm of rain per year) (Table S3). A standard Δ15Ndiet–body tissue of 4·1‰ was added to the plant regression equation intercepts to describe the predicted range of herbivore δ15N values (Table S3; Steele & Daniel 1978; Ambrose 1991; Jenkins et al. 2001; Sponheimer et al. 2003; Murphy & Bowman 2006; Mannel, Auerswald & Schnyder 2007).

Figure 3.

 (a) Regressions of δ15N values of plants from dry wash and exposed ridge habitats on rainfall, and the predicted regressions and ranges for desert bovids in the Eastern Mediterranean. The solid grey regression lines depict the relationship between mean annual rainfall and the δ15N values of C3 and C4 plants for dry wash and C3 plants for exposed ridge sites in arid regions (<150 mm year−1). Dashed lines represent the predicted δ15N values of idealized herbivores that consumed plants from the exposed ridge and dry wash sites assuming constant fractionation (+4·1‰) between plant diet and herbivore body tissue based on farm pasture (grey triangles) and domestic goats (open triangles). The shaded grey polygons represent the potential variability expected in the bovid values by applying 1 standard error above and below (±1SE) the predicted regression lines that were calculated for the plant data (see Table S3 for statistics). (b) The shaded area represents the expected range of δ15N values of herbivores consuming a hypothetical mixed diet and is extrapolated from the predicted regression lines ± 1 standard error for the C3 and C4 plant data depicted in part (a). (c) The relationship between δ15N values of horn keratin and rainfall and between predicted δ15N values of horn keratin according to the diet hypothesis assuming that diet is the primary factor determining bovid values. Dorcas gazelle (diamonds), Nubian ibex (x). The error bars represent the isotopic range from individual animals that were sampled multiple times.

The relationship between the δ15N values of herbivore horn keratin and their diet

The observed herbivore horn keratin δ15N values are plotted against the range predicted in Fig. 3 to test whether diet determines herbivore body tissue composition. If water conservation is the dominant factor affecting body tissue δ15N values then a uniform negative relationship between horn keratin δ15N and rainfall is expected. If body tissue δ15N values are solely determined by the isotopic composition of the plant diet, the horn keratin δ15N values should vary within the wide range predicted by the model.

The δ15N values of both Dorcas gazelle and Nubian ibex keratin varied widely regardless of mean annual rainfall. All keratin specimens plot within the range predicted from the isotopic values of C3 and C4 plants (Fig. 3c). Most of the herbivore specimens fall within the range expected if their nitrogen isotope values were determined by a diet comprised primarily of C3 plants (Fig. 3). The remaining five animals also plot within the range expected if these animals also incorporated some C4 plants into their diet (Fig. 3).

C4 plant consumption and δ15N values of horn keratin

A significant positive correlation was found between the δ15N and δ13C values of horn keratin of both Dorcas gazelle and Nubian ibex (G. dorcas: R2 = 0·225, < 0·05; C. ibex: R2 = 0·508, < 0·01). As the slopes and intercepts of the linear regression lines are similar for the two species they were combined and treated as a single population (t-test comparing slopes: = −0·51, = 0·612; intercepts: = −0·10, = 0·890; combined regression equation: δ15N = 0·88 δ13C+26·68, R2 = 0·32, < 0·001) (Fig. 4). Previous analysis indicated that the δ15N values of C4 plants are significantly higher than that of C3 plants from the same sites (Hartman & Danin 2010). Increased C4 plant consumption (δ13C) may thus be responsible for the increase in the δ15N values of horn keratin beyond the range expected for exclusive foraging on C3 plants. Animals consuming a pure C4 plant diet (plant δ13C = −13·5 ± 0·9‰; Hartman & Danin 2010) are thus expected to have horn keratin δ13C = ∼−10·4‰, based on Δ13Ndiet–keratin of ∼ +3·14 ‰ (adjusted from O’Connell et al. 2001; see also Table S2). The δ13C of keratin in this study (up to −16·5‰) indicates that C4 plants are only secondary dietary source for desert bovids (Fig. 4).

Figure 4.

 The relationship between the δ13C and δ15N values of desert site plants and bovid horn keratin. The dashed polygon indicates the isotopic range (δ13C and δ15N) of desert plant sites: dry wash sites (grey triangles), exposed ridge sites (open triangles). Plant data from Hartman & Danin (2010). The solid-lined polygon represents the corresponding predicted isotopic range of bovid keratin assuming constant spacing between diet and body tissue composition (Δ13C = +3·1‰; Δ15N = +4·1‰; grey arrows) mixed of C3 and C4 plants. The Dorcas gazelles (diamonds) and Nubian ibexes (x) are plotted onto the predicted range.


The relationship between water stress, physiology and horn keratin δ15N values

The physiological hypothesis put forward by Ambrose & DeNiro (1986, 1987) was later tested by Ambrose (2000). The δ15N values of the body tissues of laboratory-reared rats subjected to different dietary, water and heat regimes failed to show a relationship with water stress, heat stress and protein content in the diet (Ambrose 2000). This study came to a similar conclusion. According to the physiological hypothesis water conservation in the Dorcas gazelle and Nubian ibex species will become more pronounced as aridity increases. Consequently, an increase in water stress should lead to δ15N values that greatly exceed the typical Δ15Nplant diet–body tissue (Ambrose & DeNiro 1986). When the relationship between climatic variables potentially associated with water stress and δ15N values was examined, only temperature was weakly correlated with the δ15N values of gazelle horn keratin (Fig. 2c; Table S6). The frequently cited relationship between rainfall and the δ15N values of herbivore body tissue (Schwarcz, Dupras & Fairgrieve 1999; Murphy & Bowman 2006) may be complicated in arid and hyper-arid environments where mean annual rainfall drops below 150 mm year−1.

Multiple regression analyses examining the combined effect of temperature and rainfall and temperature and humidity on the δ15N values of Dorcas gazelle horn keratin, however, were significant. Nevertheless, the multiple regression analyses fail to support the physiological hypothesis as the relationship between both rainfall and humidity and the δ15N values of horn keratin is positive – indicating that wetter rather than the expected drier conditions are associated with elevated δ15N values (Fig. 2e,f; Table S6). This observation contradicts the expectation of the physiological hypotheses that rainfall and the δ15N values of herbivore body tissue will be negatively correlated. This result is likely related to the synergistic relationship between hot and wet conditions that trigger denitrification processes in desert soils and positively affect the δ15N values of local vegetation (Aranibar et al. 2004). If that is true, then the positive correlation between high temperature and humidity or rainfall seen in δ15N values of desert bovids merely reflects the relationship between environmental conditions and plant diet.

The absence of a direct relationship between environmental factors that cause water stress in bovids and the δ15N values of their horn keratin is not surprising given the physiological response of desert animals to water stress. Mammals eliminate the toxic effect of nitrogenous oxidation products by excreting excess nitrogen in the urine (Singer 2003). Desert adapted species concentrate urine more efficiently than mammalian species from wet environments because they have a higher percentage of juxtamedullary nephrons in the kidney that greatly improve water resorption from the filtrate prior to excretion (Schmidt-Nielsen 1997). Balter et al. (2006) hypothesize that water-stressed animals reach a negative nitrogen balance because of the reduced ingestion of dietary nitrogen and the increased recycling of nitrogen leading to elevated δ15N values in body tissue. This hypothesis is partially based on documented increases in the δ15N values of body tissue of species that experience negative nitrogen balance mainly due to nutritional stress (Hobson, Alisauskas & Clark 1993; Polischuk, Hobson & Ramsay 2001; Fuller et al. 2005). The process of water conservation however, does not involve protein catabolism. Instead, the retention of urea in the kidneys of water-deprived desert goats actually improves nitrogen balance, by increasing nitrogen recycling and decreasing its loss in the urine (Brosh et al. 1987). Furthermore, desert bovids also conserve nitrogen by reducing their resting metabolic rate when fed on low quality diet or when water stressed (Choshniak et al. 1995; Ostrowski, Mesochina & Williams 2006).

Nutritional stress, specifically the lack of nitrogen in plant diets, has also been argued to explain elevated δ15N values in the bone collagen of herbivores inhabiting arid regions (Sealy et al. 1987). In contrast, Sponheimer et al. (2003) demonstrated the opposite trend (a decrease in Δ15Nplant diet–keratin) in herbivores fed a low protein diet. In this case, the herbivores most likely maintained a balance between nitrogen influx and efflux, which has also been reported in pregnant women and reindeer (Fuller et al. 2004; Barboza & Parker 2006). A decrease in the isotopic spacing between body tissue and protein-deficient diets in pregnant mammals is primarily caused by minimizing the oxidation of amino acid nitrogen into urea (Barboza & Parker 2008). Another potential cause for the observed decrease is the recycling of nitrogen from urea. Nitrogen is salvaged from the urea by bacteria after it is transferred through the blood stream to the gastrointestinal tract. The bacteria first hydrolyse urea into ammonia (NH3), which can be directly reabsorbed back into the bloodstream. Alternatively, the bacteria convert urea into amino acids and peptides that bovids can incorporate into newly formed body tissues (Stewart & Smith 2005). The metabolic process that ends in the excretion of urea involves negative nitrogen isotope fractionation relative to dietary values (Steele & Daniel 1978). Negative fractionation may also take place in the gastrointestinal tract when ammonia is used as a substrate by bacteria to biosynthesize amino acids (Sutoh, Koyama & Yoneyama 1987; Wattiaux & Reed 1997; for host–parasite analogy see Pinnegar, Campbell & Polunim 2001). Protein poor diets are compensated by improving nitrogen salvaging, perhaps by incorporating the 15N-depleted protein produced by intestinal bacteria, and ultimately decreasing Δ15Nplant diet–keratin. The contribution of internally recycled nitrogen can potentially dampen fluctuations in the δ15N values of newly formed body tissue following seasonal shifts in plant diet quality. Feeding studies have shown that Nubian ibex and domesticated desert goats fed on protein poor diets recycle 70–90% (vs. 22% in the control group) of their urea (Choshniak & Arnon 1985; Silanikove 2000). The low variation in Dorcas gazelle keratin δ15N values sampled from individual animals here (Table 2) implies that potential periods of nutritional stress (i.e. dry season) have little effect on newly formed keratin δ15N values (Barboza & Parker 2006). Variability in the δ15N of individual Nubian ibex horn sheaths is significantly higher and may be associated with more dramatic seasonal shifts in foraging locations (Baharav & Meiboom 1981; Hakham & Ritte 1993). These observations require further testing. In summary, the hypothesis that elevated δ15N values in the body tissue of arid-adapted animals results from a physiological response to seasonal periodicity in water-stressed conditions is not supported.

The relationship between plant diet and herbivore body tissue composition

The physiological and biochemical processes that contribute to the isotopic spacing between diet and the body tissues of animals are complex and thus most measures of isotopic spacing can only be considered estimates (Post 2002; Olive et al. 2003; Vanderklift & Ponsard 2003; Balter et al. 2006). It is thus remarkable that a trophic level increase of ∼ +4‰ has frequently been measured between the δ15N values of diet and the body tissues of medium to large herbivores (Steele & Daniel 1978; Ambrose 1991; Jenkins et al. 2001; Sponheimer et al. 2003; Murphy & Bowman 2006; Mannel, Auerswald & Schnyder 2007). The Δ15Nplant diet–keratin measured in the domestic goats studied here agrees with these results. The lack of isotopic spacing between the δ15N values of domestic goat hair keratin and the supplemental feed (Table 2) suggests that the supplemental feed made only a negligible contribution to the nitrogen in goat body tissue (see comparable results in Peltier & Barboza 2003; Mannel, Auerswald & Schnyder 2007). Although the hair samples were collected in the dry season when supplemental feed is important for milk production, the hair was formed in the spring season when natural pasture is highly nutritious. Supplemental feed was likely used sparingly during the spring when the sampled hair was grown, and thus the contribution of nitrogen from supplement feed is too small to be observed. Information on the nutritional value of natural pasture and of supplementary feed from one of the goat farms (Sataf) is published elsewhere (Landau, Vecht & Perevolotsky 1993; Perevolotsky et al. 1998). Given the agreement between the results of the isotopic spacing study undertaken here and previous studies, a Δ15Nplant diet–keratin value of +4‰ was used to estimate the range of horn keratin values in desert bovids in tests of the dietary hypothesis.

Most of the variability in the δ15N values of the horn keratin of desert bovids fall within the range expected if C3 plant diet determined nitrogen isotope composition (Fig. 3). A few horn keratin values exceeded the predicted upper range for the exclusive consumption of C3 plants, but still fell within the range for animals that also consumed some C4 plants. Most notable is the divergence of the δ15N values in arid regions – in particular the low δ15N values from exposed ridge sites cannot be described by the same relationship as that shown between δ15N values and rainfall in less arid regions. According to the physiological hypothesis, horn keratin δ15N values should have increased uniformly in relation to a decrease in mean annual rainfall regardless of plant isotope composition. In addition, the δ15N values of bovid keratin fall within the predicted dietary range and there is no observable increase in the Δ15Nplant diet–keratin. Together, these results support the dietary hypothesis whereas simultaneously refuting the physiological hypotheses (Ambrose & DeNiro 1986; Balter et al. 2006). Likewise, Murphy & Bowman (2006) recently found that the δ15N values of macropod bone collagen correlates with their potential plant diet and with rainfall along aridity gradients in Australia.

Because of the δ15N values of C4 plants are significantly higher than those of C3 plants in the Eastern Mediterranean region (Table S3, Hartman & Danin 2010), the relative contribution of C4 plants to the diet of desert bovids can be estimated from the horn keratin δ13C values (Fig. 4). The positive relationship between δ13C and δ15N values of both Dorcas gazelle and Nubian ibex horn keratin indicates that elevated δ15N values are at least partially attributable to the consumption of C4 plants (Fig. 4). As C4 plants only supplement C3-dominated plant diets in the Eastern Mediterranean, they are unlikely to cause the level of nutritional stress expected to alter Δ15Nplant diet–keratin. Nevertheless, Aranibar et al. (2004) reported the opposite trend in the δ15N values of C4 plants (Graminae) which were significantly lower than those of C3 plants from the same site. Given this result, the conclusions drawn in this study should be independently established in different regions, and for different plant communities.


The results of this study support Murphy & Bowman’s (2006) argument that the δ15N values of herbivore body tissue are directly derived from their diet. Furthermore, the Δ15Nplant diet–keratin does not increase regardless of the degree of aridity. These results have wider implications. First, in regions where there is a strong relationship between rainfall and plant δ15N values, this relationship should be reflected in the tissues of the herbivores that consume them. When preserved, the organic remains in herbivore hard tissues can provide a rich source of data for reconstructing paleorainfall and past environments. Second, in locations where plant δ15N values diverge from a linear relationship with rainfall, as is the case in arid and hyper-arid environments of the Eastern Mediterranean, isotopic differences among habitats can be used to reconstruct local herbivore foraging ecology.


This study was supported by an NSF dissertation improvement grant (#0643645), a Harvard ASPR Summer Research Grant and a Cora DuBois Fellowship. The author would like to thank Dan Lieberman, Noreen Tuross, Ofer Bar-Yosef and Natalie Munro for discussions and comments on earlier drafts of the article. I would also like to thank Yoram Yom-Tov from Tel-Aviv University, Rivka Rabinovich from the Hebrew University, and Guy Bar-Oz from Haifa University, for allowing me to sample their herbivore collections. Thanks also to Ori Adler from the Israel Bio-Organic Agriculture Association for coordinating the sampling in the goat farms; and the Har Eitan (Sataf), Har Haruach and Halav im haruach (Shabi) goat farms. Finally, thanks to Ofir Altstein from the Hebrew University for his assistance with GIS data and map production, and to Bob Michener from Boston University Stable Isotope Laboratory for sample analysis. Finally, I would like to thank Brett Murphy and an anonymous reviewer for their constructive comments.