The Amazon delivers a fifth of the global continental runoff and riverine dissolved organic carbon (DOC) to the ocean. Intensified biogeochemical processes are expected at the junction of the Amazon's major blackwater tributary, the Rio Negro, and its parent, the Rio Solimbes, due to large gradients in pH, conductivity, DOC and particle load. Dissolved organic matter (DOM) plays a major role in aquatic biogeochemical processes which are poorly understood on the molecular level. To gain insights into the potential role of DOM in non-conservative processes, we assessed dynamics of Cu, Fe and DOM by ultrahigh resolution mass spectrometry in: (1) endmembers, (2) regional samples and (3) laboratory mixing experiments under presence/absence of natural particles (>0.2 p.m). The relative abundances of 3600 DOM molecular formulae were interpreted via multivariate statistics which revealed major dynamics in the DOM molecular composition. >40% of molecular formulae displayed conservative behavior even in the presence of natural particles, agreeing with bulk DOC behavior, but opposing the oftenpresumed non-conservative behavior of DOM. Another 16-27% of formulae fluctuated in FT-ICRMS signal intensity during mixing, but did not show consistent non-conservative behavior. Both rivers left a clear molecular imprint within the DOM of the Amazon, each being linked to >800 molecular formulae. Characteristic for the Rio Negro was a dominance of phenolics with a wide molecular mass range (centered at similar to 400 Da), and for the Rio Solimbes more saturated but lower-molecular mass compounds (centered at similar to 300 Da). Both Fe and Cu showed distinct non-conservative mixing patterns under particle presence. In the controlled mixing experiments including original particles at natural concentration, up to 0.5 ttg/L Cu was released from the particles into solution at 20-40% blackwater contribution. Our molecular analysis revealed distinct DOM compositional changes in polyphenol- and nitrogencontaining formulae paralleling this release, suggesting links to desorption of potential ligands and charge-induced effects at particle surfaces caused by pH and conductivity changes in the course of mixing. (C) 2019 Elsevier Ltd. All rights reserved.