Ecology and ethnobotany were integrated to assess the impact of ancient Maya tree-dominated home gardens (i.e., "forest gardens"), which contained a diversity of tree species used for daily household needs, on the modern tree species composition of a Mesoamerican forest. Researchers have argued that the ubiquity of these ancient gardens throughout Mesoamerica led to the dominance of species useful to Maya in the contemporary forest, but this pattern may be localized depending on ancient land use. The tested hypothesis was that species composition would be significantly different between areas of dense ancient residential structures (high density) and areas of little or no ancient settlement (low density). Sixty-three 400-m2 plots (31 high density and 32 low density) were censused around the El Pilar Archaeological Reserve in northwestern Belize. Species composition was significantly different, with higher abundances of commonly utilized "forest garden" species still persisting in high-density forest areas despite centuries of abandonment. Subsequent edaphic analyses only explained 5% of the species composition differences. This research provides data on the long-term impacts of Maya forests gardens for use in development of future conservation models. For Mesoamerican conservation programs to work, we must understand the complex ecological and social interactions within an ecosystem that developed in intimate association with humans.