Since the last few decades, conservation of biodiversity emerged as an unavoidable societal expectation. As an example, the European Union, with the Water Framework directive (FWD), proposes to restore a good ecological status for river systems (mainly assessed on the basis of their biotas) in order to guarantee a balance and sustainable water use for human. The implementation of the WFD has set off numerous debates especially concerning the ways and targets considered for an efficient long term restoration of river ecosystems and their biological communities. Historical approaches are able to worthily light up such debates. By reconstructing trajectories of ecosystems on a long term perspective (i.e. several decades and centuries) it is possible to understand how human activities on watercourses and theirs catchments affected river ecosystems. Furthermore, they inform on the possible interaction between human impacts and much more uncontrolled influences linked to hydroclimatic fluctuations. Lastly they allow assessing if historical data on river biological communities, particularly those prior to the industrial revolution, could represent valuable references or targets for future rivers restoration planning. This paper focus on the Seine River basin in France, a catchment strongly and long standing impacted by human activities. The aim is to improve knowledge on the past ecological state of an aquatic ecosystem since the end of the 18th century, especially with a focus on fish communities. More precisely, the objective is to draw past fish communities' distribution under long term modifications of human pressures. Thus, we developed models predicting fish species distribution in the river network based on different environmental variables related to natural river features, climate and human activities. Coupling these models with a GIS we predict modifications of fish communities on the whole Seine River basin in a retrospective view point, considering changes of environmental conditions from the mid-18th century to nowadays. By comparing predictions with historical fish data, we assess capacity of our models to reconstruct past river ecosystems' trajectories and identify major human pressures driving these trajectories.