In terms of adult tree mortality, harvesting is the most prevalent disturbance in northeastern United States forests. Previous studies have demonstrated that stand structure and tree species composition are important predictors of harvest. We extend this work to investigate how social factors further influence harvest regimes. By coupling the Forest Inventory and Analysis database to U.S. Census and National Woodland Owner Survey (NWOS) data, we quantify social and biophysical variation in the frequency and intensity of harvesting throughout a 20-state region in the northeastern United States. Among social factors, ownership class is most predictive of harvest frequency and intensity. The annual probability of a harvest event within privately owned forest (3%/yr) is twice as high as within publicly owned forests (1.5%/yr). Among private owner classes, the annual harvest probability on corporate-owned forests (3.6%/yr) is 25% higher than on private woodlands (2.9%/yr). Among public owner classes, the annual probability of harvest is highest on municipally owned forests (2.4%/ yr), followed by state-owned forests (1.6%/yr), and is lowest on federal forests (1%/yr). In contrast, corporate, state, and municipal forests all have similar distributions of harvest intensity; the median percentage of basal area removed during harvest events is approximately 40% in these three owner groups. Federal forests are similar to private woodlands with median harvest intensities of 23% and 20%, respectively. Social context variables, including local home prices, population density and the distance to a road, help explain the intensity, but not the frequency, of harvesting. Private woodlands constitute the majority of forest area; however, demographic data about their owners (e.g., their age, educational attainment, length of land tenure, retired status) show little relationship to aggregate harvest behavior. Instead, significant predictors for harvesting on private woodlands include live-tree basal area, forest type, and distance from roads. Just as with natural disturbance regimes, harvest regimes are predictable in terms of their frequency, intensity, and dispersion; and like their natural counterparts, these variables are determined by several important dimensions of environmental context. But in contrast to natural disturbance regimes, the important dimensions of context for harvesting include a combination of social and biophysical variables.

Fragmentation transforms the environment along forest edges. The prevailing narrative, driven by research in tropical systems, suggests that edge environments increase tree mortality and structural degradation resulting in net decreases in ecosystem productivity. We show that, in contrast to tropical systems, temperate forest edges exhibit increased forest growth and biomass with no change in total mortality relative to the forest interior. We analyze >48,000 forest inventory plots across the north-eastern US using a quasi-experimental matching design. At forest edges adjacent to anthropogenic land covers, we report increases of 36.3% and 24.1% in forest growth and biomass, respectively. Inclusion of edge impacts increases estimates of forest productivity by up to 23% in agriculture-dominated areas, 15% in the metropolitan coast, and +2% in the least-fragmented regions. We also quantify forest fragmentation globally, at 30-m resolution, showing that temperate forests contain 52% more edge forest area than tropical forests. Our analyses upend the conventional wisdom of forest edges as less productive than intact forest and call for a reassessment of the conservation value of forest fragments.