FORECASTING FOREST HEALTH: BUILDING A QUANTITATIVE BAYESIAN MODLEING APPROACH

Abstract

In the face of increased climate change and anthropogenic effects on the environment, quantifying forest and ecosystem health is a critical undertaking. The Montreal Process Criteria (MPC) is a globally supported forest health framework, developed by an international collaborative of twelve countries, that uses seven criteria and fifty four indicators to assess the health of temperate and boreal forests. This project used the MPC to develop a statistical modeling framework to predict changes in forest health based on climate and land use data. Few existing studies use the MPC framework, suggesting that a statistical modeling framework addressing all MPC would be a novel contribution to the field. Seven indicators were selected to quantify forest health: Alpha Diversity, Beta Diversity, Functional Diversity, Carbon Storage, Tree Mortality, Air Quality Index and Soil Quality Index. A total of 37 hierarchical Bayesian models were built and tested using public long-term field, climate, and land use data covering the six most common forest types in the Mid-Atlantic United States. Twelve models for five forest types produced acceptable predictions for beta diversity, carbon storage, tree mortality, air and soil quality. Using these twelve models, future forest health indicator values were predicted using projected climate and land use data from 2025 to 2080. Trends varied by forest health indicator, forest type group and future change scenario, but locations with large projected change emerge as candidates for continued monitoring and possible management intervention. The goal of this forest health modeling framework is to inform future science, management and decision making to preserve forest health in response climate change and human disturbance and land conversion.