There are no doubts long-term observatories provide unique insight on ecosystems trajectories. Can we use earliest data to set restoration goals? We take the example of Lake Geneva, for which descriptions of the ecosystem are available for as soon as the late 19 th and early 20 th century. Forel writes about how the luxuriant growth of plant communities provided important habitat for aquatic animals, as well as trapping nutrients and affecting water currents. It can be hard to believe Forel is referring to the same lake as present-day Lake Geneva; however, without continuous monitoring, this qualitative description can hardly be compared to recent observations. We resorted to paleolimnology to quantify the changes in plankton communities, as a proxy of general ecological changes, over the past 1,500 years. Our results show that from 563 AD (beginning of the record) to the 20 th century, the cladoceran assemblage remained stable, despite important amplitude of climate variability (3°C).
Trajectory of LakeGeneva shifted for the first time in 1946. Online dynamic linear models revealed the following transition, in 1958-1961, transition was critical, i.e., the ecosystem changed state. Littoral associated species were totally lost, and the assemblage is now dominated by pelagic species. The shift took place around the beginning of the long-term monitoring program, when local perturbations (eutrophication) were escalading. Our result raises the vexing observation that the historical dataset, one of the longest records in the world, may not provide a baseline for Lake Geneva's condition.