Quantifying patterns of mountain pine beetle (MPB) outbreak through spatio-temporal analysis can potentially reveal factors related to risk and resilience to this disturbance vector. MPB outbreaks are historically an agent of healthy forest disturbance, which maintains diversity in forest age and structure. MPB outbreak has risen to unprecedented levels differing from historical infestations in their intensity, extent, and synchronicity with other outbreaks across western North America. Measuring spatio-temporal patterns is relatively new to the field of forest entomology and invites further innovation in techniques and applications. We mapped percent tree mortality over 11 years and used the results to analyze spatial patterns of beetle spread using a spatial logistic model that incorporated local and landscape-level covariates. We modeled the probability of infestation from one year to the next by examining patterns in newly infested and disease-free pixels within our study area in the Rocky Mountains of northern Colorado and southern Wyoming.
Results/Conclusions
Our maps of percent tree mortality resulted in an overall pseudo-median difference of <1% between observed and predicted values and a mean absolute deviation of <8%. Our spatially explicit regression analysis predicted probability of infestation over the study area for each year. Covariates included elevation, slope, aspect, temperature, precipitation, forest cover type, percent forest cover, and previous infestation within the neighborhood both as a binary and continuous value. Previous infestation was evaluated both isotropically and anisotropicly at multiple scales. Second order correlation and directionality in this correlation was important due to the contagious nature of MPB and its potential for long-range, wind-directed dispersal. Our results can help inform mitigation and management strategies for MPB outbreaks in the future.