The history of the Amazon River is a much-discussed subject, and the timing of the development of a transcontinental system in particular is a matter of some controversy, with estimations varying between the Early Miocene and the Pliocene or even the Pleistocene. To shed further light on this, we studied the sediment provenance of an Oligocene to Late Pleistocene marine sedimentary section from the Ceará Rise (ODP Site 925), a topographic high in the central Atlantic Ocean, using major element concentrations and Nd isotopic composition in 85 samples. In addition, the carbon isotopic composition of bulk organic matter and changes in the distribution of glycerol dialkyl glycerol tetraethers (GDGTs) were used to identify periods of increased river outflow. On the basis of these results, we suggest that the history of the development of the Amazon River is characterized by specific steps. During the late Oligocene/Early Miocene (30-18.3 Ma), the terrigenous mass accumulation rates (TARs) were high, and sediment and GDGT compositions suggest that a large river system existed, which at times received weathering products from a younger and probably Andean sediment source. A shift to a younger Andean sediment provenance after 8.7 Ma indicates that the Amazon River became permanently connected with the Andes. Between 18.3 and 4.5 Ma, TARs were generally low, and GDGTs were derived for the most part from in situ production in marine waters. Around 4.5 Ma, the river expanded, probably due to ongoing tectonic activity, and uplift in the Andes increased Andean rock erosion. This led to a strong increase in terrigenous sediment deposition and enhanced organic matter preservation on the Ceará Rise, and the delivery of terrestrial (both soil and riverine) branched GDGTs to the Ceará Rise.

In order to understand the impact of hydrological changes of the Amazon River on sedimentary organic matter (OM) composition in Amazonian floodplain lakes, three sediment cores were collected from Lake Maracá (eastern Amazonia) along a transect from the Amazon River main channel to inland. The cores were dated with 14 C accelerator mass spectrometry (AMS) and studied by x-ray, mineralogical composition, total organic carbon (TOC) and total nitrogen (TN) contents, stable isotopic composition of TOC and TN ( δ 13 OC Cand δ 15 N) and glycerol dialkyl glycerol tetraether (GDGT) distributions. Two distinctive sedimentary depositional phases were identified based on the mineralogical composition and the geochemical characteristics of sedimentary OM. During the early-mid Holocene (~13,000-3200 cal. yr BP), low values of TOC followed by a break in sedimentation suggest a complete drying of the lake caused by drier climatic conditions. Between 3600 and 3200 cal. yr BP, this lake received a reduced influence of the Amazon River main stem. This induced a predominant deposition of C 3 -plant-derived OM supplied by surface erosion and runoff of acidic soil. A distinct connection of Lake Maracá to the Amazon River began after 3200 cal. yr BP and became permanently established, with its modern characteristics, at 1880 cal. yr BP. This change provoked an increased contribution of phytoplankton and semi-aquatic C 4 macrophytes as well as C 3 plant derived more alkaline soil OM to the sedimentary OM pool. Consequently, our study demonstrates that the source of sedimentary OM in the Amazon floodplain lakes was strongly linked to the Amazon River hydrodynamics during the late Holocene.

No abstract