Abstract
<jats:p>Previous molecular archaeological research has made the surprising discovery that almost all studied European walrus finds dating between the 11th and 14th centuries CE seemed to be traded from the Norse colony of Greenland. But why has almost no walrus ivory from the Barents Sea region been detected in medieval Europe, despite its closer proximity and being mentioned in historical accounts? Here, we show that the full spatial extent of medieval Atlantic walrus hunting has been underestimated in previous analyses that targeted modified walrus skulls (rostra) with which pairs of attached tusks were traded, because destructive sampling of ivory artefacts is not always practicable. By analysing workshop offcuts from medieval Sigtuna, Sweden, we directly compare the geographical sources of walrus ivory with rostra, and show that isolated tusks were traded in different ways. We first resolve the genome-wide trans-Atlantic walrus population structure, discovering significant genome-wide nuclear differentiation between western and eastern Atlantic walrus populations. By employing spatially diagnostic nuclear SNPs (n = 144,986), we then use low-coverage sequence data to classify historical walrus rostrum and tusk specimens to either western or eastern Atlantic origin, and thus overcome previous limitations when provenancing walruses based on mitochondrial DNA alone. We find that all medieval walrus rostra are assigned exclusively towards the western Atlantic. In contrast, half of the medieval ivory specimens from Sigtuna are assigned to eastern Atlantic sources including Iceland and the Barents Sea region. Moreover, the objects of eastern origin precede those from Norse Greenland in time, suggesting sequential exploitation. Our observations resolve a discrepancy between earlier molecular inference and historical evidence and imply a significantly broader extent and impact of medieval ecological globalisation in the Arctic.</jats:p>