Aminoisoscapes and palaeodiet reconstruction: New perspectives on millet-based diets in China using amino acid δ13C values

Foxtail millet (Setaria italica) and common millet (Panicum miliaceum) were important staple crops for the inhabitants of northern China since the Neolithic. The near exclusive consumption of these millets results in extremely elevated bulk collagen delta C-13 values (similar to-7 parts per thousand to -5 parts per thousand), which serve as natural isotopic tracers in palaeodiet studies. Here we report individual amino acid delta C-13 results (delta C-13(AA)) for humans (n = 12) and animals (n = 9) that consumed varying amounts of millets at the Proto-Shang period (2000-1600 BC) site of Nancheng, China. Using established delta C-13(AA) proxies (Delta C-13(Gly-Phe), Delta C-13(val-Phe), and plots of delta C-13(phe) vs. delta C-13(val), delta C-13(Lys) vs. Delta C-13(Giy-phe), and delta C-13(Lys) vs. Delta C-13(va1-phe)) and machine learning assisted principal component analysis (MLA-PCA), we compared the Nancheng data to published known archaeological C-3, C-4, marine and freshwater protein consumers. Exclusive millet-consuming humans and animals from Nancheng displayed highly C-13-enriched amino acid results, which were distinct from C-4 consumers of maize (Zea mays) in the Americas. Compared to delta C-13(AA) dietary proxies, MLA-PCA provides improved separation for all of the different dietary categories reviewed. Further, this method was able to distinguish additional dietary details, such as identifying brackish species. Increased application of MLA-PCA in palaeodiet research utilizing delta C-13(AA) measurements could create regional and global aminoisoscapes that can reveal unique dietary and environmental information that is otherwise hidden by bulk and existing delta C-13(AA) proxy isotopic analyses.

Automated summary

This study reported individual amino acid delta C-13 results for humans and animals at a Proto-Shang period site in China, revealing highly C-13 enriched amino acid results in millet consumers, and the ability of machine learning assisted principal component analysis to provide improved separation and distinguish additional dietary details. Increased application of this method in palaeodiet research utilizing delta C-13 measurements could reveal unique dietary and environmental information.