"Since the mid 1970s, Interior Alaska white spruce trees experienced markedly lower growth than during the 19th and early 20th centuries. This observation raises the question of forest persistence on certain sites of lowland central and eastern Alaska. We analyzed white spruce growth across a 36-site network (540 trees) on three major river floodplains in boreal Alaska along a longitudinal gradient from eastern Interior to the southwest tree limit to test for the presence of tree growth patterns and climate sensitivities. Chronologies are compared for temperature sensitivity at both stand and individual tree levels, using data from Bethel, McGrath, and Fairbanks NWS stations during the common period of 1952–2001. Cross-dated stand-level ring width chronologies indicate three regions of common signal in tree growth across the gradient. Temperature sensitivity of stand- and individual-tree chronologies is spatially coherent. Most downriver chronologies correlate positively with spring mean monthly temperatures (MMT) at Bethel, mid- and upriver chronologies correlate negatively with MMT of May and previous year July at either McGrath or Fairbanks, and an area in between is a mixed population of positive and negative responders. In downriver positive responders, recent increases from suboptimal cool temperatures accelerated tree growth, while in mid- and upriver negative responders, recent increases from optimal or above-optimal temperatures decreased growth. Fairbanks negative responders are also negatively correlated with a 200-yr index of recorded and reconstructed Fairbanks summer temperatures, and recent sustained record high summer temperatures are associated with the lowest relative growth. Until the 1940s, absolute growth rate of negative responders was greater than positive responders, but from the 1970s the positive responders grew more. These results explain why northern ring width samples can display opposite temperature sensitivity and contribute to understanding recent “divergence” or loss of temperature sensitivity in a changing climate. We find that July MMT and annual precipitation at Fairbanks are now outside the limits that previously characterized the North American distribution of white spruce, and are near the reported physiological limits of the species. Our results of the spatial and temporal change of white spruce temperature sensitivity provide strong empirical evidence of previously proposed early stage biome shift in boreal Alaska due to clear climatic causes. Already, western Alaska, previously extending to tree limit, has become the optimum climate region for the species. With modest additional warming widespread tree death will be unavoidable on warmer lowland interior sites, where persistence of white spruce is unlikely. " (source éditeur)