Increasing the connectivity of protected areas is an urgent need to ensure the conservation of forest species and help them to shift their ranges due to anthropogenic drivers. However, efforts to do so considering the joint effects of habitat fragmentation and climate change are still scant. Here, we aimed to outline a framework that incorporates spatial, temporal and multi-taxa criteria to pinpoint locations that connect protected areas in the eastern Amazon. We analyzed three mosaics of protected areas, and data on 603 species (bees, birds, bats) and developed two models using species movement flow (MF; through circuit theory) and habitat suitability (HS; through species distribution models). Considering only the MF, northward areas are the main candidates for corridors, most of which presenting forest cover (68% of the 928,379 ha). This result changes when we analyze the HS, since the corridors are mostly positioned in a different direction (westward) and less than half have forest cover (45% of the 925,058 ha). Candidate areas for both approaches totaled 135,171 ha, with 86% still covered by forest. Our results rely on methodological and taxonomic redundancy (to depict a range of movement and/or habitat requirements) for an efficient strategy to prioritize areas for connectivity. Dynamic restoration simulations showed that the location and order of restoration are important to ensure increased availability of habitat. Our approach can help address two important biodiversity threats (habitat loss and climate change) and maximize the selection of the best corridors to protect species in a rapidly changing world.