Impact of different management practices on tree biomass and carbon dynamics 30 years after logging in eastern Amazon

Reduced-impact logging (RIL) techniques are crucial for enhancing the sustainability of high-quality timber production in tropical regions. This study evaluates whether improved forest management techniques (RIL), compared to conventional logging (CL), significantly alter above-ground biomass (AGB) stocks 30 years after logging. The experiment took place in a forest area near Paragominas, Pará, Brazil. Tree diameters at 1.3 m aboveground (dbh) were measured twelve times between 1993 and 2023 across two selective management systems (RIL and CL) and an unlogged control plot. Each system was implemented on a single 24.5 ha plot. Biomass stock per area (Mg.ha^-1) was estimated for (i) the forest as a whole, (ii) logged species, (iii) species with logging potential, and (iv) species with no timber value. The Gompertz model was fitted for all treatments, and its equations were evaluated using RMSE, RMSE%, AIC, R², and the Box-Ljung (temporal autocorrelation), Shapiro-Wilk (normality), and White (heteroscedasticity) tests. After 30 years, the biomass balance was 70.68, -11.35, and 0.12 Mg ha^-1 for RIL, CL, and control plots, respectively. RIL (353.42 Mg ha^-1; 95 % CI: 263.41-443.43 Mg ha^-1) showed statistically higher maximum stocks from the adjusted Gompertz model than CL (192.91 Mg ha^-1; 95 % CI: 175.02-210.81 Mg ha^-1) and control (178.60 Mg ha^-1; 95 % CI: 175.30-181.90 Mg ha^-1). This trend was consistent across all species groups. An evaluation of the biomass stock structure, based on ecological indicators for the CL system, revealed a decrease in biomass stocks (46.6 %, 43.9 %, and 41.4 % in years 0, 30, and 60, respectively). These percentages were calculated from equations fitted with 30 years of data and then extrapolated to 60 years, representing an approximation of a climax forest (409 Mg ha^-1, as reported in the literature). Conversely, for RIL, this trend was reversed, showing an increase over time (45.1 %, 61.7 %, and 66.3 % in years 0, 30, and 60, respectively). The application of RIL demonstrated superior modeled biomass recovery relative to CL, specifically for the first logging cycle under site-specific conditions.