Abstract
Population movements constitute a significant driver of cultural change in prehistoric societies. In recent years, sulfur isotopes have emerged as a valuable approach for distinguishing human/animal provenance. However, the scarcity of sulfur isotope studies and the lack of baseline maps predicting their variations in the landscape limit our current knowledge about mobility behaviours. Here, we first present the δ34S isotope values of 142 human and animal bone collagen samples from coastal and inland funerary sites located in northern Iberia. Second, to apply a multivariate machine-learning regression and a random forest model to predict sulfur isotope variations across Iberia, we compiled the sulfur isotope data from 554 specimens of 41 archaeological locations from Holocene contexts. Our research demonstrated that population movement between coastal and inland locations is observable through differences in the δ34S isotope values of individuals linked to their respective environments, suggesting migrations on both sides of the Cantabrian mountain range. The resulting isoscape model demonstrates that sulfur isotope patterns are highly predictable, with 82% of the sulfur isotope variation explained by only four variables: elevation, Bouguer anomaly, distance from the coast, and strontium isotope values. While the model is highly accurate for regions with large amounts of data, such as northern Iberia, Central and Eastern Iberia still require more sulfur isotope data to predict isoscapes.