Unmanaged logging has the potential to be a widespread threat to tropical forests, reducing carbon storage and causing carbon emissions. On the other hand, tropical reforestation on logged or degraded land is a major, ongoing land cover change occurrence, with the potential to store carbon and conserve biodiversity. Comparative studies of how tropical forests re-establish post-disturbance are lacking, but needed due to large variation through space and time. We utilize 27 years of forest recovery, over 9 hectares, following a logging treatment experiment (Biomass and Nutrient Experiment, BIONTE) in the Central Amazon, and compare against a near-by old-growth forest dataset. The one-time logging treatments ranged from removing 16% to 28% of the plots basal area (BA). To complement the BIONTE dataset, we evaluated tropical forest recovery in two dynamic, demographic vegetation models (ZELIG-TROP and ELM-FATES), varying in representation of individual-based vs. cohort-based, and multiple ecosystem level processes. We test the hypothesis that (1) a central Amazon forest will take ~26 years to recovery from a 22% basal area logging treatment (based on previous literature), (2) post logging recovery will be attributed to growth of existing large trees, and (3) current demographic, dynamic vegetation models fail to capture this forest recovery process, due to exhibiting faster recovery.
Results/Conclusions
Upon evaluating the BIONTE dataset, with an average of ~22 % BA logged, biomass accumulation was almost doubled compared to control plots, and six out of the nine treatment plots fully recovery to pre-logging biomass values after 27 years. The average forest recovery rate after logging was 3.59 Mg ha-1 yr-1, and a faster, clear shift back to pre-treatment values (after 12-14 years) was observed for fluxes related to recovery trends (i.e. ingrowth and growth rates). Both demographic models predict a slower recovery rate compared to observed data (0.8 and 1.5 Mg ha-1 yr-1), opposite to our 3rd hypothesis. We aimed to understand what are the driving factors leading to biomass recovery and can they be replicated in the models? A multiple regression analysis found that increases in 1) stem density, 2) volume of large trees, and 3) carbon use efficiency (CUE) tended to be significant predictors of biomass change in the BIONTE plots. Original model simulations showed that CUE only slightly increased and volume of large trees actually decreased during recovery. Applying these two forest processes in the models, post-logging, was able to enhance the recovery trajectories to match field data.