Tooth enamel sampling strategies for stable isotope analysis: Potential problems in cross-method data comparisons

• Bulk and serial enamel sampling methods are evaluated through enamel δ18O and δ13C.
• Bulk and serial sampling methods show good reproducibility as standalone techniques.
• Comparisons of bulk and mean serial δ values show systematic offsets.
• Sampling methodology alone cannot explain these offsets.
• Cross-method data comparisons may be problematic.

The stable isotope composition of a given sample of tooth enamel is a function of the mineralisation process, the isotopic composition of the input signal, and sampling geometry. As the dominant axis of tooth growth is orientated from the occlusal surface to the enamel–root junction, the isotopic composition of an enamel sample that spans the full length of this axis is often taken to represent a whole-tooth or ‘bulk’ average, while serial (or ‘intra-tooth’) samples taken along the same axis are considered to track isotopic variation across the period of tooth formation. A ‘bulk’ sample of enamel should therefore approximate the mean of the ‘serial’ samples, and in teeth that form over a complete annual cycle ‘bulk’ and ‘serial’ sampling methods are often assumed to possess an annual–seasonal relationship. However, this does not take into account the effect of the mineralisation process and sampling geometry on the derived isotope signals. Here we investigate the reproducibility of both the ‘bulk’ and ‘serial’ sampling methods as independent, stand-alone techniques and then explore the assumed relationship between the two. Each method shows good reproducibility in the oxygen (δ18O) and carbon (δ13C) isotope results across paired left and right molars. However, cross method comparisons reveal a systematic offset between bulk and serial results within a single population, with 19 out of 23 δ18O bulk-tooth values being lower, and 22 out of 23 δ13C bulk-tooth values being higher, than the means of their respective intra-tooth sequences. In 4 out of 19 cases for δ18O and 6 out of 22 cases for δ13C the bulk-tooth results fall completely out of the range observed in the corresponding intra-tooth sequence. This finding has implications for cross-method data comparisons and questions the assumed temporal relationship between ‘serial’ and ‘bulk’ isotope data.