Understanding the effects of topoedaphic characteristics on site quality in a Bertholletia excelsa Bonpl. plantation in Amazonas.

This work aimed to assess if nonlinear models accurately predict the wood productive capacity of B. excelsa plantations in Amazonia and identify topoedaphic characteristics more linked to growth variations in these plantations. Nonlinear models were tested to classify the productive capacity of B. excelsa stands of different ages (14 to 21 years), by algebraic difference method. We used a database of 75 temporary plots and complete stem analyses of 30 trees. Soil sampling was performed at depths of 0–20 cm and 20–40 cm, for chemical analyses ( pH H 2 O , pH_KCl, C, P, K⁺, Ca²⁺, Mg²⁺, Al³⁺, Fe²⁺, Mn²⁺, Zn²⁺ and Na⁺) and physical (soil density, penetration resistance, granulometry). Topographic information was obtained by GPS and digital elevation image. Site classes were spatialized in the study area by method ordinary point kriging. Relationships between topoedaphic variables and site index were evaluated by correlation matrix to identify major characteristics limiting site productivity, and regression models were generated by stepwise backward method to evaluate whether topoedaphic characteristics explain growth variation in B. excelsa stands. The Chapman-Richards model presented good statistical performance, good distribution of residues, and resulted in consistent polymorphic curves. Polymorphism evaluation indicated differentiated growth patterns of specie among sites. Soil texture (sand content) and topography (elevation) were the main drivers of site productivity, followed by some soil chemical variables (K⁺, Mn²⁺ and pH_KCl). These results can support planning of activities related to installation and management of B. excelsa stands in the Amazon, besides identifying potential areas of expansion of plantations of this species. [ABSTRACT FROM AUTHOR]