Background/Question/Methods Climate change may shift vegetation latitudinally or elevationally, altering ecosystems and harming human well-being. Spatial data on the vulnerability of ecosystems to vegetation shifts could improve the ability of natural resource management organizations to identify priority areas for adaptation measures. We identified areas vulnerable to vegetation shifts and potential refugia through spatial analyses of three sets of global data at 50 km spatial resolution: (1) observed changes of 20^th century temperature and precipitation, (2) projected changes of 21^st century vegetation using the MC1 dynamic global vegetation model under three Intergovernmental Panel on Climate Change (IPCC) emissions scenarios, and (3) the overlap of results from (1) and (2). We classified areas into vulnerability classes based on IPCC treatment of uncertainty and analyzed the human population of vulnerable areas. In addition, we conducted a systematic search of published literature for cases of field research that examined long-term trends of biomes in areas where climate (not land use change or other factors) predominantly determined the distribution of vegetation.

Results/Conclusions Field observations in boreal, temperate, and tropical ecosystems have detected biome changes in the 20^th century in 15 of 19 identified cases. In the 21^st century, one-tenth to one-half of the world's land area may be highly (confidence ~0.80) to very highly (confidence ~0.95) vulnerable to vegetation shifts at the biome level. Temperate mixed forest, boreal conifer, and tundra and alpine biomes show the highest vulnerability, whereas tropical evergreen broadleaf forest and desert biomes show the lowest vulnerability. A mismatch between the patterns of vulnerability that we found and the geographic priorities that natural resource organizations have established suggests the need for adaptation of management plans. Approximately a billion people live in highly vulnerable areas.