Restauração florestal como processo multidimensional: variação espacial da estrutura e efeitos de traços funcionais ao longo de 17 anos

Abstract

Forest restoration should be understood as a long-term process of ecological reorganization in which vegetation components and species respond at different rates to environmental changes following intervention. This thesis evaluated the development of a restoring tropical forest over 17 years around the reservoir of the Anhanguera Small Hydroelectric Power Plant in northeastern São Paulo State, Brazil. The restoration covered 70.33 ha and was established in 2008 with approximately 156,000 seedlings of 105 native tree species. The study was based on annual monitoring of 56 permanent 100 m² plots between 2008 and 2025 and, at the end of the period, comparison with 16 plots located in a local forest remnant. The thesis was based on the premise that time since planting alone does not adequately represent forest recovery because ecosystem development depends on temporal changes, spatial variation in environmental conditions, and functional differences among species. The first chapter investigated trajectories of aboveground dry tree biomass, canopy height, tree density, and vertical forest stratification. Gompertz models were used to describe and compare the trajectories of biomass, canopy height, and stratification, whereas density was analyzed directly because its dynamics were not compatible with a simple sigmoidal curve. Results showed strong asynchrony among structural components. Stratification reached its maximum rate of increase at approximately 1.2 years, followed by canopy height, with maximum growth at about 2.5 years. Biomass increased more gradually and reached its maximum increment at approximately 10.4 years. Mean biomass increased from 0.04 Mg/ha at the beginning of monitoring to 230.18 Mg/ha in year 17, accompanied by a marked increase in variation among plots. Density declined after planting because of initial mortality and later increased through the recruitment of naturally regenerating individuals, reaching 4,425 individuals/ha in 2025. At the end of monitoring, 54.9% of all individuals were recruits, although they accounted for only about 1% of total biomass, indicating that biomass accumulation remained concentrated in surviving planted trees. Spatial variation became progressively more evident during forest development, particularly for biomass. In year 17, a structural equation model showed that understory light intensity was positively associated with biomass and canopy height, whereas soil moisture was positively associated with biomass, density, and stratification. Light and moisture were negatively correlated, revealing a local environmental gradient associated with the structural organization of the forest. Because these variables were measured only at the end of the monitoring period, the relationships were interpreted as spatial associations describing the final state of the forest rather than as causal evidence for trajectories accumulated over 17 years. Comparison with the local remnant indicated partial structural convergence: stratification had reached values similar to the reference forest, but biomass corresponded to approximately 75% of that observed in the remnant, canopy height remained about 1 m lower, and density represented approximately 63% of the reference value. Thus, different attributes indicated distinct stages of recovery within the same forest. The second chapter examined whether functional traits measured at the seedling stage could anticipate differences in species performance over nearly two decades. A total of 1,161 individuals belonging to 32 tree species were followed. Specific leaf area, stem specific density, leaf and root mass fractions, and seed mass were summarized into three functional axes that explained 80.14% of total trait variation. The first axis mainly represented a gradient of biomass allocation between roots and leaves; the second was associated with stem specific density and seed mass; and the third was dominated by specific leaf area. Discrete-time hazard models were used to investigate annual mortality, and a structural equation model assessed direct and indirect effects of functional dimensions on biomass in the first and seventeenth years. At the end of monitoring, 871 individuals remained alive, corresponding to 75% of the initial cohort. Mortality risk followed a nonlinear trajectory, with a first peak between years 1 and 2, a prolonged period of lower mortality, and a second increase between years 13 and 15. Local plot biomass and the root-leaf allocation axis were not detectably associated with annual mortality risk. The axis dominated by specific leaf area was associated with a 35.4% reduction in mortality risk per unit increase, with no evidence that this relationship changed through time. In contrast, the axis related to stem specific density and seed mass showed a time-varying association: species with higher scores had greater mortality risk during part of the early and intermediate periods but lower risk during a later stage of forest development. For biomass, both the root-leaf allocation axis and the specific-leaf-area-dominated axis were positively associated with biomass in the first year. Initial biomass was the main predictor of biomass in year 17, and the specific-leaf-area-dominated axis also had a direct effect on final biomass. Therefore, differences established shortly after planting persisted for many years, but functional traits did not influence growth and mortality in the same way. Together, the two chapters demonstrate that restoration followed a multidimensional trajectory characterized by structural asynchrony, increasing spatial variation, and interspecific responses that depended on the performance component and, in some cases, on the stage of forest development. The results do not support a simple opposition between acquisitive and conservative strategies and do not allow the late increase in mortality to be attributed to a single environmental filter. From an applied perspective, restoration assessments should integrate multiple structural indicators, local environmental conditions, and complementary functional dimensions, using long-term monitoring to guide adaptive decisions. After 17 years, the forest shows a consolidated vertical structure, active natural recruitment, and substantial biomass accumulation, but it still differs from the local forest remnant and leaves open questions regarding functional replacement of planted species. Restoration therefore remains an ecological process under construction, in which growth, survival, recruitment, and functionality advance at different rates.