Live carbon accumulation by trees in forest landscapes represents a major sink for atmospheric carbon. Local factors, such as disturbance history and topography, mediate aboveground live forest carbon (ALC; Mg/ha) accumulation dynamics, but data appropriate for analysis of landscape ALC change patterns are sparse. In this study, (1) we characterized ALC change for an eight-year period (2008-2016) based on vegetation plots and high resolution remote sensing (i.e., lidar) and (2) we related spatial variation in landscape variables to variation in ALC change in the 64-km2 H. J. Andrews Experimental Forest, a conifer-dominated forest landscape in western Oregon, USA. For 2008 and 2016, we predicted ALC for all 25-m pixels using a random forest model based on field observations of forest structure and lidar metrics characterizing vegetation height and vertical vegetation complexity. Using a sample of 1408 pixels, we developed a multiple regression model for the relative change in ALC (%) as a function of initial ALC in 2008, disturbance history (harvested vs. unharvested), slope, topographic position (difference between elevation of focal pixel compared to surrounding pixels), and elevation.
Results/Conclusions
Disturbance history, initial ALC, and elevation explained 74% of the variation in relative ALC change across the study area while slope and topographic position explained less than 2% of the variation. Within harvested and unharvested forests, the model explained 50% and 22% of the relative ALC change, respectively. In harvested forests, relative ALC change decreased, but remained positive, with initial ALC and increased with elevation. In contrast, relative ALC change in unharvested forests was hump-shaped, increasing when initial ALC was less than 90 Mg/ha and decreasing when initial ALC was above 90 Mg/ha, resulting in ALC losses in stands with the greatest initial ALC in the study area. Relative ALC change in unharvested forests did not vary substantially with elevation. The strong elevation effect on relative ALC change in unharvested forests implies a temperature constraint that is not obvious for unharvested forests, highlighting the increasing importance of tree mortality to ALC dynamics and a potential decoupling from environmental drivers as stands age. Our results imply that environment and disturbance history interact to constrain landscape-level biomass and carbon accumulation in coniferous forests of the Pacific Northwest.