This review focuses on calcified tissues, since bones and teeth are by far the most common vertebrate tissues surviving into the archaeological record, although keratinous tissues such as hair and nails occasionally survive in more recent, special circumstances. The patterned isotope distributions described below are archived and analysed in bones and teeth, which are composite tissues made up of complex organic molecules and minerals. Collagen is the main protein in bone and dentine that provides the source for organic carbon (13C/12C),1 nitrogen (15N/14N) and, to a lesser extent, oxygen (18O/16O), sulphur (34S/32S) and, most recently, hydrogen (D/H). It is composed of multiple helical peptide fibrils stippled with a fine, poorly crystalline ‘cement’ of mineral. Bone mineral and enamel are highly substituted biological calcium phosphate apatites, differing subtly in their chemistry and properties, from which 18O/16O can be determined from phosphate, and 13C/12C and 18O/16O from substituted carbonate. The timespan captured in these tissues differs. Since bone is a living tissue that turns over regularly within the life of an individual, isotope values reflect long-term averages that depend on the age of the individual. A recent study based on radiocarbon showed that turnover slows dramatically after full maturity is achieved (Hedges et al. 2007). In contrast, tooth enamel and dentine are incremental tissues that form during a limited, mostly juvenile, period of an individual's life, with the exception perhaps of the third molar. Consequently, isotope values reflect conditions at that time, although there is a little secondary dentine formation, and the nature of amelogenesis and primary and maturation mineralization in enamel means that time intervals are not discrete (see below).
Natural abundances of stable isotopes in foodwebs
The stable isotopes of an element differ slightly in their nuclear mass as a result of differences in the number of neutrons, leading to small but significant differences in their thermodynamic and kinetic properties (Sharp 2007). Molecules containing the higher-mass, rarer isotope tend to accumulate in the thermodynamically most stable component of a system—for instance, in the liquid rather than gaseous phase—or are slower to react in mass-sensitive kinetic reactions. In equilibrium, and incomplete or multidirectional physical and biochemical processes, the result is fractionation or partitioning. The principles governing physico-chemical fractionation are relatively well understood theoretically and empirically, and thus they provide a means of tracking the pathways of the ‘life’ elements through a complex series of chemical transformations.
The largest source of carbon isotope variability occurs in primary producers on land and in the oceans. In land plants, the two dominant photosynthetic pathways, C3 and C4 (after the number of carbon atoms fixed in the first product), differ in their net discrimination against 13C during fixation of CO2 (Smith and Epstein 1971; O’Leary 1981; Farquhar et al. 1989). In C3 photosynthesis, strong discrimination against 13C during CO2 fixation by ribulose biphosphate carboxylase/oxygenase (RUBISCO) results in more negative δ13C values in virtually all trees, woody shrubs, herbs and temperate or shade-loving grasses. Because plants following the C4 pathway (tropical grasses and many sedges) concentrate CO2 in bundle-sheath cells prior to release into the RUBISCO cycle, and as all of it is converted, fractionation is not expressed. C4 photosynthesis is believed to be a relatively recent adaptation for lower pCO2 and high solar radiation in the growing season (Ehleringer et al. 1997), so distribution of C4 plants is confined to environments with such conditions. In C3 plants δ13C varies widely, from about −24 to −36‰ (global mean −26.5‰) depending on light intensity, temperature, humidity, moisture and recycling of CO2 (O’Leary 1981; Farquhar et al. 1989; van der Merwe and Medina 1991). C4 plant δ13C (global mean −12.5‰) is less variable. Economically important C3 cereals include wheat, barley, oats and rice, as well as all root staples such as potato, manioc and yam, while important C4 plants include maize, sorghum, millet and cane sugar. In general, marine primary producers (e.g., phytoplankton, algae, diatoms and radiolaria) are enriched in 13C compared to those in terrestrial C3 ecosystems, because the source of carbon is mainly dissolved bicarbonate, which has relatively high δ13C compared to atmospheric CO2. The mean δ13C is about −20‰ (Smith and Epstein 1971), but values vary.
Plant δ13C values are reflected in the tissues of consumers. In the first controlled feeding experiment, DeNiro and Epstein (1978) showed that δ13C of the whole animal is very similar to that of its diet (where it is possible to measure the whole organism), but there is partitioning among tissues according to their chemistry and biosynthetic pathways. Thus isotopic differences, often expressed as Δ (difference) or ɛ (enrichment factor), vary between diet and particular tissues. The difference between diet and collagen δ13C is generally about +5‰, as first observed by van der Merwe and Vogel (1978), based on their values for humans in a mono-isotopic C3 biome. This offset is supported by many later studies of free-ranging herbivores (e.g., Sullivan and Krueger 1981; Lee-Thorp et al. 1989). Two well-controlled dietary experiments showed that the relationship is largely between dietary protein and collagen, because dietary amino acids are preferentially utilized for collagen tissue construction (Ambrose and Norr 1993; Tieszen and Fagre 1993) (see discussion below). A small trophic-level effect of about +1 to 2‰ is observed in subsequent steps, among omnivores and carnivores, and including humans.
For nitrogen isotopes, variability in ecosystems reflects the balance between biological nitrogen fixation, complex recycling within the biosphere, and re-release of N2 (Robinson 2001). Atmospheric N2 is globally uniform in isotope composition, with a low δ15N composition (0‰ by definition). On land, soils and plants are slightly higher in 15N compared to atmospheric N2 (Delwiche and Steyn 1970); soil and plant δ15N is typically about +1–4‰ subject to variability caused by environmental aridity, leaching (with high precipitation), anoxia and salinity (Shearer et al. 1978; Heaton 1987; Handley and Raven 1992). A general ‘rule of thumb’ is that soil δ15N is weakly inversely related to rainfall (Handley et al. 1999), but in practice the relationship that holds, although still variably, where mean annual rainfall is <400 mm (Heaton 1987). In the oceans, the most abundant form of nitrogen available to primary producers is recycled nitrate, with an average δ15N value of about +5–6‰ in the productive upwelling centres along ocean margins (Liu and Kaplan 1989).
Nitrogen isotopes vary with trophic level, and a stepwise trophic shift of +2–6‰ in δ15N from plants to herbivores, and from herbivores to carnivores, has been widely documented in marine and terrestrial foodwebs (DeNiro and Epstein 1981; Minigawa and Wada 1984; Schoeninger and DeNiro 1984; Sealy et al. 1987). In long marine foodchains, the effect is a stepwise enrichment in 15N, resulting in distinct high δ15N values in most marine foods and consumers (Minigawa and Wada 1984), compared to terrestrial foods (Schoeninger and DeNiro 1984). Freshwater ecosystems behave similarly to marine systems, so that freshwater foods also have high δ15N, although their δ13C does not follow the same pattern as the marine system (Dufour et al. 1999). The trophic shift is probably the result of loss of 15N-depleted excretion products (urea in the case of most animals; Ambrose 1991). However, as summarized by Hedges and Reynard (2007), there is considerable diet to tissue variability amongst species with different physiologies—we do not know what it is for humans, it may not be linear, and the effects of high- or low-protein diets are not well understood. Logically, if the process of urea loss/body 15N-enrichment continues, values in animal tissues should become progressively higher with age. This is not observed, however, and it may be the case that isotope effects leading to trophic enrichment in 15N are more marked in certain phases of maturation.
Biochemical processes induce minimal sulphur isotope fractionation in plants (Trust and Fry 1992) and in higher foodweb levels according to a single controlled feeding study (Richards et al. 2003b), so their distribution is largely governed by variations in underlying geology on land (ranging in δ34S from −22 to +22‰), and the contrast with the uniform composition of the oceans (δ34S =+20‰). Krouse pioneered the application of sulphur isotopes to studies of location and human diets (e.g., Krouse et al. 1987), but earlier applications were limited by the large amounts of bone collagen required until improvements in continuous flow methods for sulphur isotopes emerged. Given the uniform oceanic composition, marine diets are detectable, but because of a strong sea-spray effect, marine-like δ34S also reflects coastal or even island residence. Therefore δ34S must be applied in combination with δ13C and δ15N (Richards et al. 2003b).
It is well known that the global distribution of hydrogen and oxygen isotopes is largely bound up with their behaviour in water (Dansgaard 1964). In animal tissues, however, analysis and interpretation of the two isotopes is separated because of the nature of the tissue archives. Studies of δ18O in vertebrate bone and enamel mineral have a long history, with attention directed towards exploring δ18O in apatite phosphate or carbonate as a palaeoclimate indicator (Longinelli 1984; Luz et al. 1984; Luz and Kolodny 1985). Dietary ecology is implicated, since water and oxygen in food contribute to body water δ18O, to a degree influenced by an animal's drinking habits and thermophysiology (Luz and Kolodny 1985; Bocherens et al. 1996; Kohn 1996; Sponheimer and Lee-Thorp 2001). Because hydrogen isotopes exchange rapidly and readily, studies have focused on non-exchangeable, tightly bound hydrogen in organic molecules. This work is in its infancy, in part related to the analytical difficulties. One recent study demonstrated that in addition to providing indications of ambient climate conditions, trophic-level effects are observed (Reynard and Hedges 2008). The implications of these variations for human diets have not yet been explored.