Stand structure drives disparities in carbon storage in northern hardwood-conifer forests

Elements of forest structure are fundamentally associated with an array of ecosystem services and habitat characteristics. However, forest structure varies, in particular, through interactions with natural and human disturbances. Both variation in structural characteristics and associated relationships with ecosystem service outcomes have been poorly explored in mature, secondary forests redeveloped since 19th century agricultural abandonment in the eastern United States. Our study addressed this uncertainty focusing on carbon storage as an important climate regulation service. We conducted an inventory of 45 plots sharing similar land use history (i) to identify differences in forest structure, and (ii) to investigate links between stand structure and aboveground carbon storage. We derived 19 structural attributes and used these in Agglomerative Hierarchical Clustering (AHC) to categorize structurally different groups. Subsequently, we analyzed carbon density in each cluster and employed a random forest algorithm to derive partial effects of structural attributes on carbon storage. We found a distinctive disparity in forest structure inferred from two hardwood-dominated and one softwood-dominated clusters. Nine variables (cavity tree density, conifer ratio, foliage height index, gap area, live basal area, species diversity, variation in heights and diameters, and vertical shrub cover) explained significant differences between these clusters. Carbon storage varied markedly, and was highest in the softwood cluster. Structural complexity was overall positively associated with carbon storage, whereas this effect was more distinctive in hardwood compared to softwood-dominated forests. In particular, five variables exhibited a positive (conifer ratio, diameter variation, dead basal area, large live trees, and live basal area), one a negative (live tree density), and two (dead tree density and species diversity) a mixed relationship with carbon storage. Despite only moderate variation in climatic conditions across the investigated plots, we found a strong sensitivity of carbon storage to mean annual temperature. In contrast, annual precipitation and topography had no effect on carbon storage. The link between structural complexity and carbon storage suggests a high potential to actively increase forest carbon density. Based on our findings, a variety of options are available to enhance forest structure, and thus to improve carbon storage in managed forest ecosystems. These include, for instance, increasing the variability in tree dimensions and fostering large live trees which may improve niche cornplementarity. Increasing structural complexity in managed forest stands may thus improve buffering of potentially negative impacts of climate change on carbon stocks.