Fig 1.
The study area covered by both the MC2 dynamic global vegetation model and the state-and-transition simulation models.
Table 1.
Changes in modal temperature and precipitation values projected by three global circulation models (GCMs) for the study area.
Fig 2.
Conceptual overview of modeling process.
Ellipses represent model input and rectangles represent models. Unbounded text indicates model outputs. In step1, future climate scenario and biogeographic information are input into the MC2 Dynamic Global Vegetation Model. Within MC2, three sub-models typically interact to project potential future vegetation, wildfire, and carbon, among other output. For this analysis, we included a species distribution model of forest zones within the biogeography sub-model. In step 2, MC2 wildfire and forest zone projections for each climate scenario are analyzed to develop forest transition average probabilities and both wildfire and forest zone trends. In step 3, the MC2-derived trends and probabilities are used to develop climate-informed state-and-transition simulation models (cSTSMs) for each climate scenario. Within a cSTSM, area is permitted to shift across previously developed state-and-transition simulation models (two simplified example models are shown) following a stand-replacing disturbance (i.e., when landscape inertia is broken). Figure adapted from [46].
Fig 3.
Projected wildfire trends by MC2 and a randomly selected cSTSM simulation iteration.
Differences in wildfire projections between MC2 and cSTSMs largely reflect distinct model assumptions about wildfire. Specifically, MC2 is deterministic in setting burn probabilities/events, whereas the stochastic nature of cSTSMs results in fewer wildfire opportunities (even with a slightly increasing trend in area burned with time under changing climate).
Fig 4.
Trends and variation in wildfire, as projected by cSTSMs.
Partial fire exclusion assumes a 50% reduction in area burned.
Table 2.
Average annual percent area burned and fire rotation projected by each model under different climate scenarios.
Fig 5.
MC2 forest zone projections across three GCMs (30-year modal values).
Maps represent vegetation potential based on biophysical properties and climate.
Fig 6.
Projected MC2 (top row) and average cSTSM (bottom row) forest zone trends.
Differences between MC2 and cSTSM trends reflect differing assumptions of landscape inertia: MC2 shows incremental change in potential distribution without accounting for inertia effects, while cSTSM only allows actual change to occur after high-severity fires.
Fig 7.
Trends and variation in projected cSTSM forest zones.
Partial fire exclusion assumes a 50% reduction in area burned. Note y-axes have different scales and some lines have near-complete overlap.
Fig 8.
Projected trends and variation in early-seral conditions under difference climate and fire exclusion assumptions.
Partial fire exclusion assumes a 50% reduction in area burned. Note y-axes have different scales and some lines have near-complete overlap. See Fig 7 for legend.
Fig 9.
Projected trends and variation in late-seral conditions under difference climate and fire exclusion assumptions.
Partial fire exclusion assumes a 50% reduction in area burned. Note y-axes have different scales and some lines have near-complete overlap. See Fig 7 for legend.