Tooth enamel nitrogen isotope composition records trophic position: a tool for reconstructing food webs

The study addresses whether nitrogen isotopes preserved in mammalian tooth enamel reliably record dietary composition and trophic position, enabling reconstruction of ancient food webs where collagen is degraded. Traditional stable isotope analyses (e.g., carbon and oxygen in bioapatite; trace elements like Sr/Ca, Ba/Ca) have informed paleodietary reconstructions, and δ15N of organic matter is a widely used trophic proxy in modern ecosystems with an average consumer–diet enrichment of ~3–4‰. However, nitrogen-based reconstructions in fossils have been constrained by poor preservation of N-bearing organics in bone and dentin, restricting analyses to relatively young, well-preserved samples. Enamel, due to its dense, crystalline hydroxyapatite structure and low porosity, is more diagenetically resistant and can encapsulate organic matter, but the very low N content has limited δ15N measurements with conventional EA-IRMS. Building on improved oxidation–denitrification approaches that enable precise δ15N measurements from small enamel masses, prior work established in controlled feeding experiments and a single African ecosystem that enamel δ15N reflects diet. This study expands to multiple modern ecosystems and a Late Pleistocene fossil assemblage to test whether enamel δ15N robustly captures trophic level and preserves dietary signals through diagenesis.

Past research has shown consistent trophic enrichment of 15N (~3–4‰) between diet and consumer tissues across diverse ecosystems. Applications to fossil vertebrates have relied primarily on collagen, which is susceptible to diagenetic loss, limiting temporal reach. New non-traditional proxies (e.g., Ca and Zn isotopes) are promising but less baseline-calibrated. Enamel’s high mineral density and crystalline structure promote preservation of occluded organics, but low N content historically hindered δ15N measurements using standard EA-IRMS or nano-EA methods, which required large samples and yielded lower precision. An oxidation–denitrification method with rigorous pre-treatment and low blanks improved analytical precision at nanomole N levels. Controlled feeding experiments on ferrets demonstrated that enamel δ15N tracks diet, and a case study from Gorongosa National Park indicated higher enamel δ15N in carnivores than herbivores, though with limited carnivore representation and no paired collagen data. Additional work showed enamel N and δ15N are resilient under oxidative and thermal stress, and enamel δ15N from shark teeth tracks trophic position over tens of millions of years, highlighting enamel’s preservation potential. Whether mammalian enamel (hydroxyapatite) preserves trophic δ15N comparably to shark enamel (fluorapatite) in terrestrial settings remained an open question.

Sample sets: (1) Modern African mammals (n = 54) spanning multiple ecosystems and dietary guilds (herbivores: browsers, grazers, mixed feeders; omnivores; carnivores). For a subset (n = 33), mandibular bone collagen was analyzed for δ15N and δ13C to enable direct comparison with enamel from the same individuals. (2) Late Pleistocene fossil mammals (n = 10; 8.4–13.4 ka) from Tam Hay Marklot (THM) Cave, Laos, characterized by excellent enamel preservation but poor collagen preservation. Four fossil dentin collagen isotopic data points (from previous work) were available for comparison. Analytical approach for enamel δ15N: Tooth enamel aliquots (5–7 mg) were cleaned using an oxidative pre-treatment to remove exogenous organics, then analyzed using an oxidation–denitrification method coupled to GC–EA–IRMS at the Max Planck Institute for Chemistry (Mainz). This method yields high precision (<0.2‰ at ~5 nmol N) from low-N samples with low procedural blanks, isolating endogenous enamel-bound nitrogen. Carbon isotopes: Enamel δ13C was measured to assess C3 vs C4 resource use and potential habitat signals. For the modern subset and limited fossils with preserved collagen, δ13C of collagen was also measured to compare with enamel carbonates. Collagen extraction: Mandibular bone powder (~150 mg) was demineralized in 0.5 M HCl at 4 °C, rinsed to pH ~2–3, gelatinized at 70 °C for 48 h, filtered, and freeze-dried. Collagen quality (yield, C:N) was assessed prior to IRMS analysis. δ15N and δ13C of collagen were measured with EA–IRMS with analytical precision better than ±0.25‰ (1σ). Statistics: Data distributions and homoscedasticity were evaluated. Group differences were tested using ANOVA (with Tukey–Kramer HSD) or non-parametric Kruskal–Wallis with Dunn’s test and Bonferroni correction as appropriate. Relationships between paired isotopic measures (enamel vs collagen; δ13C enamel vs δ13C collagen) used Pearson or rank correlations. Significance threshold α = 0.05. Software: PAST v4.03 and IBM statistical software.

- Modern enamel δ15N ranged 3.5–14.9‰ (n = 54) and differed significantly by diet (F(2,51) = 26.05, p < 0.001): herbivores 6.2 ± 1.6‰ (n = 36), carnivores 9.9 ± 2.0‰ (n = 14), omnivores 7.2 ± 4.0‰ (n = 4). Herbivores vs carnivores differed (p < 0.001). Differences also among herbivore feeding modes (F(4,49) = 15.18, p < 0.001). - Modern collagen δ15N ranged 1.2–12.6‰ (n = 33) with diet effects (F(2,39) = 11.6, p < 0.001): herbivores 6.7 ± 2.0‰ (n = 21), carnivores 9.8 ± 1.4‰ (n = 12). Among herbivores, browsers 6.0 ± 1.2‰ (n = 6), grazers 6.4 ± 1.1‰ (n = 12), mixed feeders 9.3 ± 2.6‰ (n = 3). - Enamel vs collagen δ15N (paired, same individuals) showed a strong positive correlation (Pearson r(31) = 0.865, p < 0.001) with regression: δ15Ncollagen = 0.88 × δ15Nenamel + 0.96 (95% CI slope: 0.66–1.11); no consistent directional offset. - Trophic enrichment (Δ15N = carnivore − herbivore) was evident in both materials: enamel Δ15N ≈ 3.7‰; collagen Δ15N ≈ 3.1‰. - Modern enamel δ13C ranged −15.9 to +1.9‰ (n = 54) and differed by diet (χ2(4) = 41, p < 0.001). Browsers had the lowest values (−13.0 ± 1.8‰), mixed feeders intermediate (−9.3 ± 3.3‰), grazers higher (−8.9 ± 1.5‰). Omnivores −8.9 ± 1.5‰. Enamel δ13C correlated positively with collagen δ13C (Spearman R(31) = 0.867, p < 0.001); offsets larger in herbivores (~8.2‰) than carnivores (~4.5‰). - Fossil enamel N content (4–10 nmol/mg; mean 6.4 ± 1.6 nmol/mg; n = 10) overlapped modern enamel (2–10 nmol/mg; mean 4.8 ± 2.0 nmol/mg; n = 54). No significant correlation between enamel N content and δ15N, indicating preservation without exogenous N addition or endogenous N loss trends. - THM fossil enamel δ15N preserved trophic structure: herbivores low, omnivores intermediate, carnivore (Panthera pardus) highest. Enamel δ13C distinguished grazers (higher δ13C, consistent with C4 resources; e.g., Asian water buffalo and an indet. bovid near −1 to 0‰) from taxa with lower δ13C (∼−15‰) indicative of C3 habitats (e.g., rhinoceros, serow). Limited paired fossil collagen data showed consistency with enamel-based inferences. - Ecological nuances in modern carnivores: leopards and foxes had lower δ15Nenamel than lions and spotted hyenas; elevated δ15N in some hyenas likely reflects nursing signals. Combined δ15N and δ13C resolved niche differences linked to prey preferences (C3- vs C4-consuming prey).

The findings demonstrate that nitrogen isotopes preserved in tooth enamel record trophic position in a manner consistent with the canonical collagen-based proxy. The clear trophic enrichment signal (~3.7‰) between herbivores and carnivores in enamel across multiple African ecosystems, and the strong correlation between paired enamel and collagen δ15N, validate enamel-bound nitrogen as a robust dietary tracer. Carbon isotope data from enamel, and its correlation with collagen δ13C, further contextualize dietary resources (C3 vs C4) and habitat use, enabling finer ecological interpretations such as niche partitioning among carnivores. Critically, the THM fossil enamel exhibits N contents and δ15N patterns comparable to modern enamel, supporting the diagenetic resilience of enamel-bound organics and the preservation of in vivo dietary signals even where collagen is lost. Together, these results extend nitrogen-based trophic reconstructions to older and more diagenetically altered assemblages, opening new avenues for reconstructing ancient food webs and dietary transitions in vertebrate evolution.

This study establishes tooth enamel δ15N as a powerful and diagenetically robust proxy for reconstructing diet and trophic position in both modern and fossil mammals. Enamel δ15N captures expected trophic enrichment and correlates strongly with collagen δ15N from the same individuals. Fossil enamel from the Late Pleistocene preserves original dietary signals despite collagen loss, with enamel N contents comparable to modern samples. Coupled enamel δ15N and δ13C enable refined ecological interpretations, including prey base (C3 vs C4) and niche partitioning. Future research should expand taxonomic and geographic coverage, increase fossil sample sizes across depositional contexts and ages, and further calibrate enamel–collagen offsets and potential influences of physiology and environment on enamel-bound nitrogen.

- Uneven sample sizes across dietary groups (e.g., few omnivores; limited carnivore representation in some locales) and ecosystems may influence variance and limit within-guild comparisons. - The modern dataset integrates multiple regions with potential baseline δ15N variation; baseline effects were not explicitly modeled. - Potential confounding abiotic and biotic factors affecting δ15N (diet quality, physiology, microbiota, climate) were not directly tested. - Fossil dataset size is small (n = 10), with only four specimens yielding collagen for limited comparisons. - Some reported values (e.g., subset sizes for δ13Ccollagen) are constrained by preservation and may reduce statistical power.