For several decades, the idea of a virgin Amazon rainforest has been questioned by anthropologists, archaeologists, and geographers (Clement, 2015). The presence of soils of anthropogenic origin and the accounts of the first Europeans of dense populations led these researchers to propose much higher population estimates and to postulate the existence of complex societies before the demographic collapse produced by Old World diseases at the first contacts around 1500. Some have gone so far as to postulate that the space of the present dense and evergreen forest was open parkland. In addition, the concentration of certain useful species near archaeological sites has prompted the postulate that pre-Columbian societies have profoundly modified the floristic composition of the forest. It is considered to consist of manufactured landscapes. These theses continue to be bitterly contested by research teams of paleoecologists, climatologists, and botanists (Barlow et alii, 2012). Addressing this problem as a historical geographer involves both a geographer's work on historical sources, but also the mobilization of conceptual contributions from geohistory. As for the approaches of the first written sources of the 16th century for the reconstitution of the population, the transformation of the environment by pre-Columbian societies, the physiognomy of plant formations, they are not very concerned with the study of the spatial differentiation of the Amazon rainforest, and this is only partially due to the difficulty of localizing accurately the observations of the first explorers (Rostain, 2017). The idea of a heavily populated Amazon rainforest also contradicts certain achievements of geohistory, making densification over large areas a function of the long duration, the antiquity of the population, despite phases of demographic decline (Grataloup, 2007). This paper aims to examine what a geohistorical approach can can contribute to this debate.