LAUSR

Abstract Fieldwork and desk-based research in the western Amazon basin has led to an explosive growth in the state of knowledge surrounding the pre-Columbian archaeology of this region. Previously thought to be a sparsely occupied environment, archaeologists have recorded hundreds of geometric earthworks between the Purus and Acre rivers in recent years, spurring renewed interest in understanding the distribution, age, and function of these structures. A challenge has been to identify possible relationships between sites and to place them in their broader landscape setting. The precise spatial scale, relative importance of different factors, and strength of any relationships that contributed to shaping their distributions remain an open question. This paper develops and applies an explicitly spatial framework to address this problem, drawing on a rich body of recent research in Acre state (Brazil) and advanced point process modelling. The analytical approach, which is fully documented and reproducible with the accompanying code, infers the factors affecting geometric earthwork distribution at multiple spatial scales. This enables the first robust predictions of territorial integration in the region, which is discussed context of extant archaeological models. The findings support the interpretation that non-stratified societies likely occupied Acre during the late pre-Columbian period.