Baobab species are representative of the high biodiversity and endemism rates that place Madagascar in the top three of the countries with the highest biodiversity in the world. In this study, we estimated the vulnerability of three endangered Malagasy baobab species (Adansonia grandidieri Baill., Adansonia perrieri Capuron and Adansonia suarezensis H. Perrier) to climate change and the effectiveness of the protected area network (PAN) for the future conservation of these species. To estimate the environmental niche of the species, we used an original data-set based on satellite image analysis to detect species presence and an ensemble modelling approach using three species distribution models (GLM, GAM and MaxEnt). We projected the species distribution in 2050 and 2080 using an ensemble forecasting approach combining the three species distribution models and three global circulation models for climate projections. Measures of connectivity were employed to assess the present and future effectiveness of the existing PAN. Among the three baobab species studied, two are severely threatened by climate change (A. perrieri and A. suarezensis), in part because the present PAN does not overlap with future species distribution areas. Recently, strong efforts have been made in designing an optimised PAN to conserve Madagascar outstanding biodiversity. Nevertheless, its future effectiveness is questioned by the potential shifts in species distributions due to predicted changes in climate. In the context of climate change, alternative strategies such as ecological restoration would also have to be adopted to conserve biodiversity in Madagascar.