B23B-0604
Developing a Long-Term Forest Gap Model to Predict the Behavior of California Pines, Oaks, and Cedars Under Climate Change and Other Disturbance Scenarios

Tuesday, 15 December 2015
Poster Hall (Moscone South)
Samantha L Davis and Emily Moran, University of California Merced, Merced, CA, United States
Abstract:
Many predictions about how trees will respond to climate change have been made, but these often rely on extrapolating into the future one of two extremes: purely correlative factors like climate, or purely physiological factors unique to a particular species or plant functional group. We are working towards a model that combines both phenotypic and genotypic traits to better predict responses of trees to climate change.

We have worked to parameterize a neighborhood dynamics, individual tree forest-gap model called SORTIE-ND, using open data from both the USDA Forest Service Forest Inventory & Analysis (FIA) datasets in California and 30-yr old permanent plots established by the USGS. We generated individual species factors including stage-specific mortality and growth rates, and species-specific allometric equations for ten species, including Abies concolor, A. magnifica, Calocedrus decurrens, Pinus contorta, P. jeffreyi, P. lambertiana, P. monticola, P. ponderosa, and the two hardwoods Quercus chrysolepis and Q. kelloggii.

During this process, we also developed two R packages to aid in parameter development for SORTIE-ND in other ecological systems. MakeMyForests is an R package that parses FIA datasets and calculates parameters based on the state averages of growth, light, and allometric parameters. disperseR is an R package that uses extensive plot data, with individual tree, sapling, and seedling measurements, to calculate finely tuned mortality and growth parameters for SORTIE-ND. Both are freely available on GitHub, and future updates will be available on CRAN.

To validate the model, we withheld several plots from the 30-yr USGS data while calculating parameters. We tested for differences between the actual withheld data and the simulated forest data, in basal area, seedling density, seed dispersal, and species composition. The similarity of our model to the real system suggests that the model parameters we generated with our R packages accurately represent the system, and that our model can be extended to include changes in precipitation, temperature, and disturbance with very little manipulaton. We hope that our examples, R package development, and SORTIE-ND module development will enable other ecologists to utilize SORTIE-ND to predict changes in local and important ecoystems around the world.