The boreal forests of Interior Alaska are experiencing a climate-induced biome shift and increased frequency and size of wildfires. Effective management of forests and wildlife requires basic understanding of the interactions between vegetation, climate, and wildlife, especially during periods of rapid ecological change. Caribou depend upon fire-sensitive lichens as a winter food source. To assess the impact of fire and climate on boreal forests and caribou habitat, we randomly sampled over 200 sites in an area greater than 2.2 million acres (White Mountains National Recreation Area and Steese National Conservation Area). Lichen growth has been linked to forest stand ages and fire return intervals in many studies. Using a variety of vegetation, dendrochronological, and fire history datasets we address the relationships between lichen abundance, fire return intervals, and forest stand ages using regression trees and machine learning (Stochastic Gradient Boosting). The scale of recent wildfires and drought-related tree mortality in lowland boreal forests make it essential to improve our understanding of these systems and consider the potential impacts to the two caribou herds which use the area.
Results/Conclusions
Forest stand ages were greater at higher elevations and northern aspects or where ground cover of lichens was below 30% and moss cover was greater than 40%. These spatial and vegetation characteristics typify areas that burn less frequently and may represent islands of higher quality winter habitat for caribou, especially when lowlands are more prone to fire and a climate-induced shift from coniferous forest. Boosted regression trees poorly predicted lichen cover, especially when non-lichen ground cover variables were removed from the model. Forest stand age influenced lichen cover by only 7%, but lichen cover was greatest before and after two centuries of tree growth. The decrease in lichens around 200 years may be due to a high probability of fire occurrence within this timeframe or a closing of the canopy and development of a dense litter or moss layer. Number of years since last fire was not a good predictor of lichens, possibly due to fire return intervals not being known for many sites. Fire scar data is being incorporated into the analysis to address this issue. Additional model refinement and subsequent spatial extrapolation using Random Forests is planned to determine potential caribou habitat and assist management of public lands.