Introduction & Theoretical Foundations
10 April 2026
Stable Isotope Mixing Models for Dietary Reconstruction | Module 1
This module establishes the conceptual and mathematical foundations on which all subsequent mixing-model work rests.
By the end of this session you should be able to:
Isotopes are atoms of the same element differing in neutron number and therefore atomic mass.
Stable isotopes do not undergo radioactive decay — unlike 14C in radiocarbon dating, they are permanent constituents of the biosphere.
| System | Light isotope | Heavy isotope | Abundance (heavy) |
|---|---|---|---|
| Carbon | 12C | 13C | 1.11 % |
| Nitrogen | 14N | 15N | 0.37 % |
| Sulphur | 32S | 34S | 4.25 % |
The 1 Da (atomic) mass difference is sufficient to produce measurable fractionation at every trophic transfer and biochemical transformation involving bond formation or breakage.
Heavier isotopes form slightly stronger bonds & react slightly more slowly.
At every bond-breaking or bond-forming event the lighter isotope is preferentially processed.
The heavier isotope becomes:
Because this repeats at every trophic step, the isotopic signal accumulates predictably up a food chain.
Absolute heavy-isotope abundances are tiny (~1 %), so ratios are reported as per-mil (‰) deviations from an international standard:
\[ \delta X\ (\text{‰}) = \left[\frac{R_{\text{sample}}}{R_{\text{standard}}} - 1\right] \times 1000 \]
where \(R\) = heavy/light isotope abundance ratio.
δ13C — reference: Vienna Pee Dee Belemnite (VPDB)
Values: −35 ‰ (C3 plants) to −7 ‰ (marine fish)
δ15N — reference: atmospheric N2 (AIR = 0 ‰)
Values: ~+2 ‰ (legumes) to +20 ‰+ (apex marine predators)
Stable isotope ratios are measured by Isotope Ratio Mass Spectrometry (IRMS).
In continuous-flow (CF-IRMS) configurations, sample combustion and MS introduction are integrated:
±0.1–0.2 ‰ precision is sufficient to resolve trophic-level differences, which are typically 3–5 ‰ per level for nitrogen.
δ13C discriminates:
δ15N tracks trophic level:
δ13C and δ15N place individuals in a 2-D isotope space reflecting the main axes of dietary variation — and both can be measured simultaneously from a few milligrams of collagen in a single IRMS run.
Sulphur (δ34S) adds a third isotopic axis to dietary mixing models, which is particularly valuable because its fractionation pathway is largely independent of C and N metabolism — meaning it responds to different environmental and dietary signals. It’s especially useful for distinguishing marine from terrestrial protein sources (since oceanic sulphur has a distinct signature around +20‰ compared to terrestrial values nearer 0–10‰), and for separating foodwebs that overlap in δ¹³C/δ¹⁵N space but occupy geologically distinct landscapes with different bedrock sulphur inputs.
Equilibrium fractionation
Heavy and light isotopes have slightly different bond energies → temperature-dependent → primary driver of 13C partitioning during carbonate precipitation.
Kinetic fractionation
Lighter isotopes react faster → drives trophic enrichment: every time an enzyme breaks or forms a bond, the lighter isotope is processed preferentially, leaving the heavier isotope enriched in consumer tissue.
For most archaeological and ecological applications, kinetic fractionation dominates. Its predictable, stepwise nature is what makes stable isotopes such a powerful dietary tool.
When an organism synthesises tissue from diet, fractionation produces systematic isotopic enrichment. These offsets are TEFs (also: discrimination factors), denoted Δ (capital delta):
\[\Delta^{13}C = \delta^{13}C_{\text{tissue}} - \delta^{13}C_{\text{diet}}\]
\[\Delta^{15}N = \delta^{15}N_{\text{tissue}} - \delta^{15}N_{\text{diet}}\]
When an animal eats food and builds tissue from it, the biochemical processes involved (transamination, deamination, protein synthesis etc.) are not isotopically neutral. Lighter isotopes tend to be preferentially lost or excreted, meaning the tissue ends up slightly enriched in the heavier isotope relative to what was consumed. The TEF (or Δ) simply quantifies that offset.
| Source | Δ¹³C (‰) | Δ¹⁵N (‰) |
|---|---|---|
| Ambrose & Norr 1993 | +1.0 | +3.4 |
| Bownes et al. 2017 | 1+1.0 | +5.5±0.5 |
| O’Connell et al. 2012 | n/a | ~+6.0 |
| Fernandes et al. 2015 | +4.8 ± 0.5 | +5.5±0.5 |
| Common mixing-model default | +4.8 ± 0.5 | +5.5±0.5 |
TEF values are not universal constants. They vary with tissue type, metabolic state, protein quality, latitude, and season. Always justify your choice, propagate its uncertainty using the reported SD, and conduct sensitivity analyses before drawing dietary conclusions.
At each trophic level, δ15N increases ~3–5 ‰ because:
| Dietary group | Approx. δ¹⁵N range (‰) |
|---|---|
| Legumes (N2-fixing) | 0 – 2 |
| C3 terrestrial herbivores | 4 – 8 |
| Omnivorous humans (mixed plant/animal) | 8 – 12 |
| Marine or freshwater fish consumers | 13 – 18 |
Elevated δ15N is not always diagnostic of high-status meat consumption:
Isotopic routing = the differential biochemical fate of dietary macronutrients — proteins, lipids, and carbohydrates — during tissue synthesis.
It has profound implications for interpreting collagen isotope data and for constructing mixing models, because collagen does not reflect bulk diet uniformly across all macronutrient classes.
Bone collagen is synthesised almost entirely from amino acids, and these are preferentially sourced from dietary protein:
Practical consequence: end-member values in a collagen-based mixing model should reflect the protein-derived signal from each food source, not the bulk food δ13C. Especially critical for protein-poor, high-carbohydrate diets (millet- or maize-dependent communities).
Different tissues record dietary signals over different timescales — directly analogous to the integration window concept in radiocarbon dating.
| Tissue | Integration period | Dietary window |
|---|---|---|
| Bone collagen | 10–30+ years (remodelling-dependent) | Long-term adult average |
| Bone carbonate | Similar to collagen | Long-term whole-diet average |
| Dentine collagen | Childhood; minimal turnover | Diet during crown formation (age 0–12 yr) |
| Enamel carbonate | Childhood; no remodelling | Diet during enamel formation |
| Hair keratin | ~1–3 months per cm | Recent / seasonal diet |
| Nail | ~6 months (fingernail) | Recent diet |
| Muscle (soft tissue) | ~180 days | Recent diet (rarely preserved) |
Comparing adult bone collagen with dentine collagen from the same individual provides a longitudinal perspective on dietary change across the life course — useful for studying weaning, migration, and social transitions.
| Year | Milestone |
|---|---|
| 1978 | van der Merwe & Vogel — δ13C as a diet tracer; maize adoption in North America |
| 1984 | Schoeninger & DeNiro — global reference dataset; marine vs. terrestrial signatures |
| 1993 | Ambrose & Norr — controlled diet experiments; isotopic routing to collagen |
| 2008 | Parnell et al. — Bayesian mixing models (SIAR) |
| 2010s– | MixSIAR, FRUITS, ReSources — probabilistic frameworks with formal uncertainty |
| Research question | Primary signal | Context |
|---|---|---|
| C3 vs. C4 plant staple adoption | δ13C collagen + apatite | Agricultural transitions |
| Marine vs. terrestrial protein | δ13C + δ15N collagen | Coastal/island populations |
| Trophic level & animal protein quantity | δ15N collagen | Social differentiation studies |
| Breastfeeding and weaning age | δ15N in infant/juvenile dentine | Infant burials |
| Freshwater fish reliance | δ13C + δ15N (aquatic reservoir) | Riparian settlements |
| Geographic mobility & migration | δ87Sr + δ18O + δ13C combined | Multi-isotope provenance models |
| Social status & dietary differentiation | δ13C + δ15N + δ34S | Cemetery analyses |
The Marine Reservoir Effect connection
Stable isotope data are routinely collected from specimens submitted for 14C dating — the δ13C used to correct for isotopic fractionation in AMS dating comes from the same collagen extract.
High marine protein intake (elevated δ13C and δ15N) is directly implicated in the Marine Reservoir Effect, which can cause apparent age offsets of several centuries if unaccounted for.
Mixing-model dietary estimates are increasingly used to quantify the proportion of marine protein in diet and thereby calculate a site-specific 14C offset (ΔR correction).
Before any mixing model can be constructed, collagen extract quality must be assessed. Degraded collagen introduces systematic bias because isotope ratios shift with diagenetic alteration.
Standard acceptance criteria:
| Criterion | Acceptable range | Reference |
|---|---|---|
| C:N atomic ratio | 2.9 – 3.6 | DeNiro 1985 |
| Carbon content (%C) | 30 – 45 % | van Klinken 1999 |
| Nitrogen content (%N) | 11 – 16 % | van Klinken 1999 |
| Collagen yield | ≥ 1 % of bone dry weight | Longin 1971 |
The C:N ratio is the most widely used single criterion in practice.
Ambrose, S.H. & Norr, L. (1993) Experimental evidence for the relationship of carbon isotope ratios of whole diet and dietary protein to those of bone collagen and carbonate. In: Lambert & Grupe (eds), Prehistoric Human Bone, pp. 1–37.
DeNiro, M.J. (1985) Postmortem preservation and alteration of in vivo bone collagen isotope ratios. Nature, 317, 806–809.
Hedges, R.E.M. & van Klinken, G.J. (1992) A review of current approaches in the pretreatment of bone for radiocarbon dating by AMS. Radiocarbon, 34, 279–291.
Parnell, A.C., Inger, R., Bearhop, S. & Jackson, A.L. (2010) Source partitioning using stable isotopes: coping with too much variation. PLoS ONE, 5, e9672.
Schoeninger, M.J. & DeNiro, M.J. (1984) Nitrogen and carbon isotopic composition of bone collagen from marine and terrestrial animals. Geochim. Cosmochim. Acta, 48, 625–639.
van der Merwe, N.J. & Vogel, J.C. (1978) 13C content of human collagen as a measure of prehistoric diet. Nature, 276, 815–816.
van Klinken, G.J. (1999) Bone collagen quality indicators for palaeodietary and radiocarbon measurements. J. Archaeol. Sci., 26, 687–695.
Module 1: Introduction & Theoretical Foundationss