Understanding how climate and vegetation influence fire regimes is crucial for interpreting ecological feedbacks and predicting future changes. Such knowledge is particularly relevant in boreal forest and tundra ecosystems, where increased fire activity could release stored soil carbon and accelerate climatic warming. We quantified decadal-scale fire-climate-vegetation relationships using 2-km resolution fire (1950-2010), climate (1971-2000), and vegetation data from Alaskan boreal forest and tundra ecosystems. Using boosted regression trees (BRTs) we estimated the decadal probability of fire occurrence as a function of summer temperature and precipitation, annual growing degree days and moisture deficit, and vegetation type. One hundred BRTs were calibrated from random subsets of the total available data and validated against subsets of un-sampled observations. BRT accuracy was quantified using the area under the receiver operating characteristic curve (AUC), where values >0.5 indicate increasingly accurate classifications of fire occurrence. These BRTs were then used to predict fire occurrence across Alaska during the historic period, and for periods throughout the 21stcentury, based on climate projections from downscaled general circulation models (GCMs).
Results/Conclusions
AUC values range from 0.75 to 0.88 for predicting fire presence/absence, indicating significant model accuracy. Summer temperature has the highest relative influence of all explanatory variables on fire occurrence, while vegetation type has the lowest relative influence. These results indicate that summer temperature is the dominant factor controlling the presence/absence of fire in Alaska, and the direct influence of vegetation cannot be decoupled from climate, at least at large spatial scales. Fire-climate relationships are non-linear in Alaskan boreal forest and tundra, with both displaying a threshold for fire occurrence near a mean summer temperature of 12°C. Future projections reflect general warming across Alaska and the strong influence of summer temperature in BRT models highlights the potential of increased flammability in Alaskan boreal forest and tundra during the 21st century. Four out of five GCMs suggest an increased probability of fire occurrence in boreal forest and tundra in the mid-21st century, with the most extreme projections suggesting a doubling of the decadal probability of fire in the boreal forest, from 3.5% to 7%, and tripling in tundra, from 0.3% to 1%. By providing spatially-explicit projections, these results can help scientists and managers plan for long-term changes in boreal-forest and tundra ecosystems.