Date of Award
8-2026
Document Type
Dissertation
Degree Name
Doctor of Philosophy (PhD)
Department
Forestry and Environmental Conservation
Committee Chair/Advisor
DR. Bo Song (Committee chair)
Committee Member
Dr. David Coyle (Committee co-chair)
Committee Member
Dr. Shaowu Bao
Committee Member
Professor. Carle Brewster
Abstract
Southern pine beetle (SPB; Dendroctonus frontalis) outbreaks are among the most damaging biological disturbances affecting pine forests of the southeastern United States, capable of causing extensive tree mortality, economic losses, and reduced ecosystem resilience. This dissertation evaluated and advanced predictive approaches for SPB outbreak forecasting by integrating systematic review, geospatial modeling, comparative model evaluation, and artificial intelligence-based prediction. First, a systematic review of SPB predictive tools from 1978 to 2024 showed that modeling has evolved from statistical, climatic, and mechanistic approaches to remote sensing, GIS, machine learning, hybrid modeling, and decision-support systems, although many tools remain limited by scale mismatch and weak operational application. To address the problem of SPB outbreaks through predictive modeling, the Southern Pine Beetle Outbreak Model version 1 (SPBOM1) was developed by the U.S. Forest Service. SPBOM1, a Python-based statistical analysis tool based on its prediction outputs, proved useful for geospatial forecasting of SPB outbreaks at the county level. The model identifies prior-year SPB infestation spots, April maximum temperature and growing-season evapotranspiration as key predictors of outbreak risk. To ascertain the predictive capability of SPBOM1, it was compared to the Zero-Inflated Poisson (ZIP) model using observed county-level infestation data from 2020 to 2025, revealing complementary strengths for early-season planning and trap-informed refinement. To improve the prediction accuracy of SPBOM1, we proposed an Artificial intelligence (AI)- based model using Random Forest, Extreme Gradient Boosting (XGBoost), and Support Vector Machine (SVM), which identified previous-year infestation as the strongest predictor of outbreak magnitude while capturing localized hotspots. The AI-based models show a significant increase in SPB prediction accuracy. Artificial intelligence-based models can improve the predictive performance of SPBOM1 and the ZIP model, thereby strengthening early-warning systems for proactive southern pine beetle outbreak management.
Recommended Citation
Anozie, Ebere L., "Regional Forecasting of Southern Pine Beetle Outbreaks in the Southeastern USA Using SPBOM1 With Artificial Intelligence-Based Modeling" (2026). All Dissertations. 4296.
https://open.clemson.edu/all_dissertations/4296