ABSTRACT
Forest degradation accounts for ~70% of total carbon losses from tropical forests. Substantial emissions are from selective logging, a land-use activity that decreases forest carbon density. To maintain carbon values in selectively logged forests, climate change mitigation policies and government agencies promote the adoption of reduced-impact logging (RIL) practices. However, whether RIL will maintain both carbon and timber values in managed tropical forests over time remains uncertain. In this study, we quantify the recovery of timber stocks and aboveground carbon at an experimental site where forests were subjected to different intensities of RIL (4, 8, and 16 trees/ha). Our census data span 20 years postlogging and 17 years after the liberation of future crop trees from competition in a tropical forest on the Guiana Shield, a globally important forest carbon reservoir. We model recovery of timber and carbon with a breakpoint regression that allowed us to capture elevated tree mortality immediately after logging. Recovery rates of timber and carbon were governed by the presence of residual trees (i.e., trees that persisted through the first harvest). The liberation treatment stimulated faster recovery of timber albeit at a carbon cost. Model results suggest a threshold logging intensity beyond which forests managed for timber and carbon derive few benefits from RIL, with recruitment and residual growth not sufficient to offset losses. Inclusion of the breakpoint at which carbon and timber gains outpaced postlogging mortality led to high predictive accuracy, including out-of-sample R2 values >90%, and enabled inference on demographic changes postlogging. Our modeling framework is broadly applicable to studies that aim to quantify impacts of logging on forest recovery. Overall, we demonstrate that initial mortality drives variation in recovery rates, that the second harvest depends on old growth wood, and that timber intensification lowers carbon stocks.