Scholars have lavished attention on wetland agriculture in ancient Mesoamerica, but upland terrace agriculture was a more significant form of intensive farming and had a profound influence on today's sustainable agriculture (Wilken 1969; Siemens 1983; Sluyter 1994; Beach and Dunning 1995; Dunning and Beach 2000). Terracing probably originated in the earliest agricultural intensification in the Fertile Crescent, and it remains an effective way to limit soil erosion, improve water management, and provide a planting surface on steep slopes (Hurni 1988; Zurayk 1994). Across the ancient Maya Lowlands, terracing made steep escarpments and areas of thin soils usable. Authors emphasize the heterogeneous Maya adaptations to diverse environments, but only with terracing could the ancient Maya bring the region into long-term production (Whitmore and Turner 1992; Fedick 1996; Dunning and others 1998). In this article we examine ancient terraces in the Three Rivers region of Belize, speak to the extent and intent of terra cing in this tropical lowland, and discuss the history of terracing, from its origins in the Early Classic period (A.D. 250-550), to its florescence in the Late Classic period (A.D. 550-850), and to its abandonment in the Postclassic period (A.D. 900-1500). Hugh Hammond Bennett argued that soil depletion caused the ancient Maya to collapse (1926), and we know that major disturbances affected these soils in the Maya Preclassic and Late Classic periods. We, however, find much more evidence that the complex and widespread agroengineering features across the Maya Lowlands were a successful accommodation to steep tropical environments.
Agricultural terracing, like canal irrigation in Mesoamerica (Doolittle 1995), is a classic element of cultural ecology because it provides direct evidence of human adaptation of the land to sustain production. The tropical lowlands of southern Mexico, Belize, Honduras, and Guatemala are a major, active focus of ancient-terrace study within the broader realm of work on ancient environmental change and on settlement. Much scholarship has been driven by a contrast between the extensively used (more "natural") landscapes of today and the intensive agricultural uses of the past that these terraces embody. We have investigated ancient Maya terraces, soil, and agriculture across this region since 1990 to study the ancient Maya relationship with the land, to understand the age and architecture of terraces, and to track the relationships between terraces and environmental change. Many of these terraces still function after a thousand years under tropical forests and milpas, thereby implying a practical technology wi th modern, "grass roots" utility (Bocco 1991; Beach and Dunning 1995).
TERRACING IN THE MAYA LOWLANDS
Space permits only an overview of the rich literature on indigenous terracing in the New World. Three recent books and two earlier works synthesize information on the indigenous agriculture of the Americas, including important sections on terracing and terrace cultivation (Wilken 1969; Donkin 1979; Doolittle 2000; Denevan 2001; Whitmore and Turner 2001). The broader literature on the southwestern United States and northern Mexico is important, voluminous, and relevant to Mesoamerica. Notable works include those by Charles DiPeso (1984), Robert Schmidt and Rex Gerald (1988), Jonathan Sandor, Paul Gersper, and John Hawley (1986, 1990), and Laurance Herold and Reuben Miller (1993). DiPeso develops a watershed model for water and soil engineering through terraces and canals at the eleventh-century Casas Grandes site (1984). Schmidt and Gerald, however, use precipitation and bedrock patterns to argue that terraces were the product of localized adaptation (1988; see also Doolittle 2000, 300-301). In the same region , Herold and Miller show that terrace sequences maintained local soil moisture longer than did nonterraced land (1993). …