Origin of mound-field landscapes: a multi-proxy approach combining contemporary vegetation, carbon stable isotopes and phytoliths

Seasonally flooded South American savannas harbor different kinds of mound-field landscapes of largely unknown origin. A recent study used soil carbon-isotope depth profiles and other proxies to infer vegetation history in murundu landscapes in Brazil. Results suggested that differential erosion, not building-up processes (e.g., termite mounds), produced mounds. We tested this approach to inferring mound origin in a mound-field landscape in French Guiana. We examined carbon-isotope depth profiles of soil organic matter, phytolith profiles and contemporary vegetation composition in mounds and inter-mounds. Relative abundance of C3 and C4 plants across habitats was very different from that in murundu landscapes; C3 plants were better represented in inter-mounds than on mounds. Habitat differences in C3/C4 distribution were subtler than in murundu landscapes, limiting inference of vegetation history based on carbon isotopes. Still, carbon-isotope and phytolith depth profiles gave similar pictures of vegetation history, both favoring a building-up hypothesis, corroborating other evidence that these mounds are vestiges of ancient agricultural raised fields. Carbon-isotope depth profiles are unlikely to be adequate for deciphering origin of mound-field landscapes from vegetation history in seasonally flooded savannas. Including data on current vegetation and phytoliths makes inferences more robust.