Stemming biodiversity loss requires strategic conservation guided by well-articulated and achievable targets, whether they be proactive (e.g., protect diverse places) or reactive (e.g., protect threatened species). Both types of targets can be effective, but there are trade-offs, especially for broadly-distributed ecosystems or taxa, such as migratory species, a group for which conservation has been challenged by limited knowledge of distributions throughout the annual cycle. We combined novel spatiotemporal distribution models with population trend data to first examine focal areas for the conservation of Neotropical migratory birds (n=112 species) during the non-breeding period in the Western Hemisphere based on a proactive approach (highest diversity) versus a reactive approach (strongest declines) to conservation. For the focal areas, we then assessed the extent of recent anthropogenic impact, protected area status and projected future changes in land cover using three shared socioeconomic pathways (Sustainability=SSP1, Business-as-usual=SSP2, Regional nationalism=SSP3). Spatial priorities were strikingly different when targeting areas of high species diversity, emphasizing southern Mexico and northern Central America, versus areas with more severe declines across species, emphasizing the Andean cordilleras of South America. Only a fraction of the non-breeding region (1.4%) met targets for diversity and decline, mostly in southern Central America. Current levels of protection were similar for the two targets. Areas prioritized to conserve high species diversity have experienced less recent anthropogenic impact than areas prioritized for decline but are predicted to experience more rapid land conversion to less suitable open, agricultural landscapes in the next three decades under both an SSP1 and SSP2 scenario. Only the SSP3 scenario projected similar conversion rates for the two targets. Our findings indicate how even within taxa, efficient conservation efforts will depend on the careful consideration of desired targets combined with reliable predictions about the locations and types of land cover change under alternative socioeconomic futures.