A reconstruction of the Indigenous maize farming niche of Utah

• Machine learning reconstructs the Indigenous maize farming niche of Utah. • Early Utah farmers balanced elevation-driven precipitation and the growing season. • Farmers mitigate this trade-off by diverting perennial streams in drier regions. • Early Utah farmers settled in an ideal distribution model way. Maize ( Zea mays L . ) was one of the most widespread domesticated plants in the Americas before European colonization. Despite its widespread distribution, explaining how and why ancient maize farming spread into the northern reaches of the American Southwest remains a central research question in archaeology. To understand how ancient maize spread, we need a comprehensive suitability model for maize agriculture that incorporates multiple ecological variables to accurately predict where maize farming was suitable. In this paper, we construct a species distribution model for maize, using a novel machine learning approach from ecology, to produce a suitability model for Indigenous maize agriculture in Utah (ca. 3200–500 calBP). We then explore the variation in the suitability of maize farming settlements over time. Results suggest that Indigenous maize farmers selected settlement locations to minimize the distance to perennial streams and to balance the elevation-driven trade-off between growing degree days and precipitation. By comparing locations best suited for maize agriculture over time and relative to estimates of past population density, we suggest Indigenous maize farmer settlement patterns throughout the northwestern limit of ancestral maize farming follow predictions from an ideal distribution model, where increasing competition drove farmers into less suitable farming habitats. These results help elucidate the environmental and demographic patterns that influenced the spread of Indigenous maize farming.