Using LiDAR to quantify, map, and conserve late-successional and old-growth forest in Maine, USA

Abstract

The world continues to lose late-successional and old-growth (LSOG) forest as the human population and demand for wood fiber grow. However, older forest age classes provide structural and compositional characteristics important to biodiversity that are often not present in forests managed for timber. LSOG forest conservation has been constrained by our inability to accurately and cost-effectively map it over large spatial scales. The advent of light detection and ranging (LiDAR) provides a potential solution to this problem. We used publicly available LiDAR to map LSOG forest for approximately 4.2 million ha of the unorganized territories of Maine, USA. We built a classification model using known-class forest types and eight canopy structure metrics derived from the LiDAR point cloud. The model was 93% accurate in distinguishing “Not LSOG” forest from three LSOG classes (transitioning late successional, late successional, and old growth). At a finer level, the model was also effective at distinguishing transitioning late successional from late-successional forest, but less effective at differentiating old growth from late-successional forest. We evaluated the amounts and rates of LSOG forest loss across landowner types, such as public lands, private commercial forests, and private conservation forests. Private commercial forest had the most LSOG forest of any owner type but it was losing it faster (−2.2%/year) than other owner types. Tens of thousands of distinct LSOG parcels were identified throughout the study area, the most common size between 1 and 5 ha. The advent of this precise, accurate, spatially explicit LSOG map is being used by conservation planners and facilitating a larger social conversation about how much LSOG to conserve and how it should be distributed across the landscape.