The purpose of this study was to indentify the most appropriated restoration model by manipulating floristic composition and silvicultural practices, and to understand how the best development was explained by the factors that controls light use and light use efficiency at the canopy. A “Restoration models for Atlantic Forest disturbed lands aiming carbon sequestration” experiment was installed in 2004 using 20 Brazilian woody species and 8 forest restoration systems, in a factorial 23 design, with 3 factors and two levels per factor: i) Floristic composition with 67% and 50% of pioneer species; ii) Plant spacing (3x1m e 3x2m); and iii) Silvicultural management with usual (higher environmental stress level); and maximum (lower environmental stress level) silvicultural practices due to weed control and fertilization. Wood net primary production (WNPP), based on inventory surveys and wood density of all 20 species, leaf area index (LAI, with hemispherical pictures), absorbed photossintetically active radiation (APAR, on meteorological stations and LAI), light use efficiency (LUE=WNPP/APAR) and canopy nitrogen content was measured or estimated up to 42 months-old. The floristic composition proportion, with 50% and 67% of pioneers, were statistically equivalent, in all ages, for WNPP, LAI and LUE. The 3x1m spacing provided larger biomass stock at age of 42 months, comparatively to the 3x2m planting (13.0 Mg.ha-1 versus 10.3 Mg. ha-1), due a higher productivity until 24 months. However, WNPP of both spacing were equivalent after 36 months (5.5 Mg.ha-1year-1). The higher productivity of the 3x1m spacing was directly related to its greater leaf area index comparatively to the 3x2m spacing (0,98 m2m-2 versus 0.71 m2m-2). The maximum silvicultural practices provided a 4-fold increase on wood biomass at age 42 months, comparatively to the usual practices (18.6 Mg.ha-1 versus 4.6 Mg.ha-1). This larger stock was explained by a higher LAI (1.34 m2m-2 versus 0.34 m2m-2) leading to higher WNPP (8.5 Mg.ha-1year-1 versus 2.4 Mg. ha-1year-1, at 42 months). The Spacing x Management interaction was significant for LAI, LUE and WNPP, meaning that 3x2m and 3x1m spacing had distinct responses to environmental stress levels, showing that 3x2m depends on intensive silvicultural practices to express its maximum productivity. Finally, the WNPP was strongly related to LAI during summer months (at 36 months) (r2=0,96), and this relationship increased by the inclusion of canopy nitrogen content, showing that these two variables have a potential to be used to evaluate forest restoration quality and productivity. Thus, for the study conditions, the best model was: Composition 50%:50% of pioneer:non-pioneer species, 3x2m spacing, and the maximum silvicultural practices, with total weed control (Brachiaria decumbens) and fertilization.
