We conducted a study with sample units of conglomerate structure using cross-database of forest inventory 100% of Annual Production Unit (APU-08), located at the Top Pacajá Martins farm, municipality of Portel-PA to study the effects of variation between and within clusters with reference to three response variables they were: i) VT = Total volume of wood of 77 tree species found in the study area, ii) VO = Total Volume of the 10 most frequent tree species and volume in the study area and, iii) Vt = Total volume of 10 forest species Cikel more commercialized by the company we aimed to determine the ideal size and distance of plots from the Method of Maximum Curvature, once defined the size, its area was fixed and then there was tested five different values to determine the ideal distance from the center of the conglomerate to the plots and between them. Results showed that the size of 0.10 ha plot, the distance of 50 m between plots, and cross-structure with 8 subplots are efficient for estimating the parameters.
Forest inventory; cluster sampling; size of subplot; intraclass correlation; sample structure
