Climate change has the potential to reduce surface-water supply by expanding the activity, density, or coverage of upland vegetation, although the likelihood and severity of this effect are poorly known. We quantified the extent to which vegetation and evapotranspiration (ET) are presently cold-limited in California's upper Kings River basin and used a space-for-time substitution to calculate the sensitivity of riverflow to vegetation expansion. We found that runoff is highly sensitive to vegetation migration; warming projected for 2100 could increase average basin-wide ET by 28% and decrease riverflow by 26%. Kings River basin ET currently peaks at midelevation and declines at higher elevation, creating a cold-limited zone above 2,400 m that is disproportionately important for runoff generation. Climate projections for 2085-2100 indicate as much as 4.1°C warming in California's Sierra Nevada, which would expand high rates of ET 700-m upslope if vegetation maintains its current correlation with temperature. Moreover, we observed that the relationship between basin-wide ET and temperature is similar across the entire western slope of California's Sierra Nevada, implying that the risk of increasing montane ET with warming is widespread.water resources | plant migration R oughly 4 billion people globally and 20 million people in the state of California rely on mountain runoff for freshwater, and there is growing concern these water resources will prove vulnerable to climate change (1-6). River flow (Q) is a function of precipitation (P) minus evapotranspiration (ET) (P−ET); increased montane ET with warming, either because of the direct effect of temperature on evaporative demand or the indirect effect of warming on vegetation density and distribution, would reduce Q (5, 7-9). However, hydrologic model projections for California's Sierra Nevada have discounted this possibility, indicating little or no effect of warming on annual . This result appears linked to two model assumptions: (i) models have often assumed the properties of montane vegetation will remain static, and (ii) models have often implicitly assumed that current annual montane ET is almost entirely limited by water availability and that warming will simply hasten the beginning and end of the growing season.Recent evidence calls both of these assumptions into question. Widespread increases in subalpine tree growth, tree-line altitude, and species distribution with elevation have been reported with recent climate trends in California and elsewhere, implying that rapid vegetation shifts are possible (14-16). Time series of Sierra Nevada forest greenness indicate a transition from water limitation at low elevation to cold limitation at high altitude, implying that upper elevation ET is sensitive to warming (17). Nonetheless, the extent to which annual montane ET is currently temperature-limited, as well as the sensitivity of large-scale ET to vegetation redistribution, remain largely unquantified.We used the upper Kings River basin in California's Sierra ...