Projections of climate change impacts on forest succession for local land management using a new vegetation model, CV-STM

Background/Question/Methods

The processes that change climate, succession and land-use occur at different time scales and therefore contribute to complex feedbacks between human activities and ecosystems. Understanding the interactions among these processes is essential to evidence-based land-use decision making. We developed a new vegetation model, CV-STM, to simulate fine-grained changes in vegetation on an annual time step under projected climate change and in conjunction with simulations of changes in land-use and management. CV-STM uses information on biogeography and biogeochemistry from a dynamic global vegetation model (DGVM). In its current implementation, CV-STM relies on output from the DGVM MC1; our approach, however, could be effective with any DGVM. CV-STM provides information on the species composition and structural stage of stands of trees, at a spatial resolution as fine as 0.5 ha, by using a state-and-transition model to track successional changes. We conducted simulations within a 1,000 km² study area in the southern Willamette Valley, Oregon, USA. CV-STM can operate with mechanistic models of disturbance, e.g., fire or insects, although for the present study we implemented stand-replacing disturbances randomly, at three different levels across the landscape. CV-STM can also operate as a module within Envision, an agent-based model of land-use change. Within Envision, CV-STM can stochastically simulate vegetation under a variety of land-use policy scenarios, and is therefore essential to land-use planning across a wide range of future scenarios.

Results/Conclusions

Simulations in a 1,000 km² study area showed changes in successional trajectories under alternative climate change and disturbance scenarios. Climate and disturbance scenarios also showed strong interactions. CV-STM responded dynamically and stochastically to the generic, random disturbance that we implemented. In some regards, output from CV-STM matched output from MC1 – climate scenarios that produced more rapid and larger changes in vegetation in MC1 resulted in similar changes in vegetation in CV-STM. This result indicates that our approach successfully used the DGVM MC1 to train and adjust transition probabilities within an STM. Changes in CV-STM, however, were generally slower than in MC1, due to key ecologically appropriate and conservative constraints that we built into the model. Our results indicate that DGVMs may be overestimating the rates of vegetation change, especially in the absence of stand-replacing disturbances. Because CV-STM can work in conjunction with models of land-use change, modeling tools like CV-STM are part of the necessary next steps to develop effective, adaptive policies that can help our landscapes and societies adapt to rapidly changing climate.