U.S. Joint Fire Science Program


Date of this Version


Document Type



Final Report: JFSP Project Number 09-3-01-47


US government work.


More frequent fires under climate warming are likely to alter terrestrial carbon (C) stocks by reducing the amount of C stored in biomass and soil. However, the thresholds of fire frequency that could shift landscapes from C sinks to C sources under future climates and whether these are likely to be exceeded during the coming century are not known. We used the Greater Yellowstone Ecosystem (GYE) as a case study to explore the conditions under which future climate and fire regimes would result in tipping points of C source/sink dynamics. We asked: (1) How great a change in climate and fire regime would be required to shift each of the dominant vegetation communities in the GYE from a net C sink to a net C source? (2) Do current projections indicate that changes of this magnitude are likely to occur in the next century, and if so, where in the GYE do they occur? and (3) What are the integrated effects of changing climate, vegetation, and fire on spatial patterns of C flux across the GYE landscape as a whole? To answer these questions, we developed downscaled climate projections for the GYE for three general circulation models and used these projections in dynamic and statistical modeling approaches. Using the CENTURY ecosystem model, we simulated C storage for individual forest stands under three fire-event pathways (fires at 90, 60 or every 30 years) to year 2100 compared to a reference simulation (no fire, representing the historical fire interval) under both future and current climate scenarios. Our results show that fire intervals would need to be less than 90 years for lodgepole pine (Pinus contorta var. latifolia) forest stands to shift from a net C sink to a net C source because the time between fires would be less than the time required to recover 85% of the C lost to fire (Question 1). We also developed new statistical models to relate monthly climate data to the occurrence of large fires (> 200 ha) and area burned, evaluated these for the 1972-1999 time period, and then used these relationships to predict fire occurrence and area burned in the GYE through 2100 given the downscaled climate projections. Results showed that anticipated climate changes are likely to increase fire frequency and annual area burned over the next century compared to the observational record. However, the timing of these changes and the probability of future largescale 1988-type fires depended on the type of climate-fire model that was used, the accuracy of the simulated future climates, and to a small degree, the specific climate simulation. The climatefire frequency and climate-fire size models are extremely sensitive to temperature differences between the projected future climate and the 1961-1990 base period because the two large fire years that occurred in the 1972-1999 climate-fire model calibration period had relatively small temperature anomalies (0.5 to 1 °C) and the small sample size of the large fire years in the time series makes model building a challenge. Between now and 2050, where we have the most confidence in the model, all climate scenarios and both fire-climate model formulations projected at least two 1988 sized fires (range 2-6, fires projected to be > 300,000 ha). After 2050, climatic conditions are sufficiently outside the historic range of variability used to estimate statistical fire models that those models cannot be used to characterize the magnitude of extreme fire years. However, extreme fire years from 2050-2100 will almost certainly become more common then projected for 2010-2050, because temperature is projected to continue to increase while precipitation is projected to remain at historical levels. We note, however, that projected changes in temperature by the climate scenarios only reach the historical differences in temperature between a subalpine forest (with an historical fire return interval of > 100 years) and a montane forest (with an historical fire return interval of < 30 years) by the end of this century (5-6 °C). In the northern Rocky Mountains, large fire years have been driven historically by extreme climate conditions. Our results imply that fuel availability would become increasingly important for fire as weather conditions conducive to large fires become common. The capacity for fast post-fire regeneration of lodgepole pine from an aerial seedbank (serotinous cones) and the projected increase in lodgepole pine productivity under warmer climate conditions are unlikely to counter the anticipated reductions in fire-return interval. In all future climate scenarios, decreases in fire-return interval are likely to reduce the potential of the GYE landscape to store C (Question 3). The magnitude of this shift will depend on the future distribution of forest and nonforest ecosystems across the landscape, other constraints on fire patterns not considered here (fuels, ignition factors, and landscape management), and the accuracy of the fire-climate model as future climate diverges increasingly from the past. If past climate-fire relationships can predict the future, soon after 2050 climate conditions projected by all three general circulation models would likely result in more fire than the current conifer forest ecosystem in the GYE could sustain. Forest managers should be considering the potential for qualitative shifts in forest distribution and regional C storage to occur before 2100.