Different models, similar outcomes: projected loss of forest carbon in California under climate change

Abstract Changing climatic conditions and disturbance regimes are reshaping ecosystems globally, with profound implications for forest carbon dynamics. Because California’s forests represent a substantial carbon sink, understanding how climate-driven disturbances will alter their capacity to store and sequester carbon is critical for meeting the state’s emissions reduction and adaptation goals. Here we compared two structurally distinct landscape vegetation models, a state and transition model Land Use and Carbon Scenario Simulator (LUCAS) and mechanistic Second-generation Landscape Disturbance and Succession model (LANDIS-II), to evaluate projections of forest carbon dynamics under shared climate scenarios and a continuation of current management practices. Because model structure can substantially influence projected outcomes, comparing structurally distinct models under shared inputs is essential for characterizing uncertainty and informing confidence in projected trends. Both models were run with same initial conditions and utilized the same climate projections but were allowed to use their own respective wildfire model extensions. The model outputs compared were aboveground biomass, net ecosystem exchange, total carbon, and cumulative area burned. Despite structural differences, both models projected net carbon losses over the next four decades, with LUCAS estimating a 6% decline (145 Tg C) and LANDIS‑II a 4% decline (94 Tg C). Fire activity was projected to keep pace or increase depending on climate projection, with a future burned area equivalent to the past 30 years occurring within the next 15 to 25 years. While model outputs converged on broad trends, differences in soil carbon pools and post-fire recovery trajectories highlight structural uncertainties that influenced these projections. These findings suggest that California’s forests are at risk of transitioning from carbon sinks to sources, with significant implications for climate policy. By demonstrating that models built on fundamentally different structural assumptions converge on the same directional trend of forest carbon loss, this comparison provides greater confidence in the projected sink-to-source transition than any single model can, while quantifying the structural sources of uncertainty that will shape the reliability of future carbon projections used in state climate policy.