Climate change poses an existential threat to trees, given our understanding of the importance of climate in shaping their geographic distributions. Climate envelope models are commonly used to predict how species will respond to climate change. These models give rise to the leading edge-trailing edge paradigm for range change: populations at the cool edge of a species’ distribution are expected to benefit from warming, whereas populations at the warm edge are expected to decline. We challenged this paradigm with a spatial network of 1,633 tree-ring time series on Pinus edulis collected in 937 forest inventory plots encompassing this species’ entire geographic distribution. A Bayesian multiple regression analysis of ring widths showed that growth of P. edulis increases with temperature variation across space (warmer sites have higher growth rates, especially at relatively wet, warm sites) but that trees everywhere grow less in warmer-than-average years. This ubiquitous negative response to interannual temperature variability implies that instead of expecting “leading edge” populations to prosper with warming, we should expect all populations to experience decreased performance (and fitness), which may require either assisted evolution or assisted migration to mitigate. We question the underlying assumption of climate envelope models that global (species-wide) climatic tolerances can be translated directly into projections of range change, ignoring individual-level plasticity and local adaptation. We conclude that predictions of the effects of climate change on trees’ geographic distributions should be re-evaluated with spatial networks of longitudinal (time series) data.