Abstract. To improve the simulation of vegetation-climate feedbacks in the high latitudes, three new circumpolar Plant Functional Types (PFTs) were added in the ORCHIDEE land surface model, namely non-vascular plants (NVPs) representing bryophytes and lichens, arctic shrubs, and arctic C3 grasses. Non-vascular plants are assigned no stomatal conductance, very shallow roots, and can desiccate during dry episodes and become active again during wet periods, which gives them a larger phenological plasticity compared to grasses and shrubs. Shrubs have a specific carbon allocation scheme, and differ from trees by their larger survival rates in winter, due to protection by snow. Arctic C3 grasses have the same equations than in the original ORCHIDEE version, but different parameter values, optimized from in-situ observations of biomass and NPP in Siberia. In situ observations of living biomass and productivity from Siberia were used to calibrate the parameters of the new PFTs using a Bayesian optimization procedure. With the new PFTs, we obtain a lower Net Primary Productivity (NPP) by 31 % (from 55° N), as well as a lower roughness length (−41 %), transpiration (+33 %) and a higher winter albedo (by 3.6 %) due to a larger snow cover. A simulation of the water balance and runoff and drainage in the high northern latitudes using the new PFTs results in an increase of fresh water discharge in the Arctic ocean by 11 % (+140 km−3 y−1), owing to less evapotranspiration. Future developments should focus on the competition between these three PFTs and boreal trees PFTs, in order to simulate their area changes in response to climate change, and the effect of carbon-nitrogen interactions.