INTRODUCTION Inherent in traditional views of ape origins is the idea that, like living apes, early large-bodied apes lived in tropical forests. In response to constraints related to locomoting in forest canopies, it has been proposed that early apes evolved their quintessential upright torsos and acrobatic climbing and suspensory abilities, enhancing their locomotor versatility, to distribute their weight among small supports and thus reach ripe fruit in the terminal branches. This feeding and locomotor transition from a quadruped with a horizontal torso is thought to have occurred in the Middle Miocene due to an increasingly seasonal climate and feeding competition from evolving monkeys. Although ecological and behavioral comparisons among living apes and monkeys provide evidence for versions of terminal branch forest frugivory hypotheses, corroboration from the early ape fossil record has been lacking, as have detailed reconstructions of the habitats where the first apes evolved. RATIONALE The Early Miocene fossil site of Moroto II in Uganda provides a unique opportunity to test the predictions of terminal branch forest frugivory hypotheses. Moroto II documents the oldest [21 million years ago (Ma)] well-established paleontological record of ape teeth and postcranial bones from a single locality and preserves paleoecological proxies to reconstruct the environment. The following lines of evidence from Moroto II were analyzed: (i) the functional anatomy of femora and a vertebra attributed to the ape Morotopithecus ; (ii) dental traits, including molar shape and isotopic profiles of Morotopithecus enamel; (iii) isotopic dietary paleoecology of associated fossil mammals; (iv) biogeochemical signals from paleosols (ancient soils) that reflect local relative proportions of C 3 (trees and shrubs) and C 4 (tropical grasses and sedges that can endure water stress) vegetation as well as rainfall; and (v) assemblages of phytoliths, microscopic plant-derived silica bodies that reflect past plant communities. RESULTS A short, strong femur biomechanically favorable to vertical climbing and a vertebra indicating a dorsostable lower back confirm that ape fossils from Moroto II shared locomotor traits with living apes. Both Morotopithecus and a smaller ape from the site have elongated molars with well-developed crests for shearing leaves. Carbon isotopic signatures of the enamel of these apes and of other fossil mammals indicate that some mammals consistently fed on water-stressed C 3 plants, and possibly also C 4 vegetation, in a woodland setting. Carbon isotope values of pedogenic carbonates, paleosol organic matter, and plant waxes all point to substantial C 4 grass biomass on the landscape. Analysis of paleosols also indicates subhumid, strongly seasonal rainfall, and phytolith assemblages include forms from both arid-adapted C 4 grasses and forest-indicator plants. CONCLUSION The ancient co-occurrence of dental specializations for leaf eating, rather than ripe fruit consumption, along with ape-like locomotor abilities counters the predictions of the terminal branch forest frugivory hypotheses. The combined paleoecological evidence situates Morotopithecus in a woodland with a broken canopy and substantial grass understory including C 4 species. These findings call for a new paradigm for the evolutionary origins of early apes. We propose that seasonal, wooded environments may have exerted previously unrecognized selective pressures in the evolution of arboreal apes. For example, some apes may have needed to access leaves in the higher canopy in times of low fruit availability and to be adept at ascending and descending from trees that lacked a continuous canopy. Hominoid habitat comparisons. Shown are reconstructions of a traditionally conceived hominoid habitat ( A ) and the 21 Ma Moroto II, Uganda, habitat ( B ).