TESTES DE IDENTIDADE DE MODELOS APLICADOS NA CONSTRUÇÃO DE CURVAS DE SÍTIO E NA PROJEÇÃO DA PRODUÇÃO DE POVOAMENTOS DE Eucalyptus grandis W. Hill. Ex. Maiden

Abstract

This study aimed to apply the identity tests of model in the construction of curves site and in the projection in the production of population of Eucalyptus grandis. The data used in the modeling are from plantations located on farms in two regions of the state of São Paulo, arising from permanent plots, with ages ranging from 17 to 96 months. Mean dominant height, total height, quadratic mean diameter, individual volume, basal area and bark volume per plot estimates were provided by a company linked to the forest sector. The sampled area totaling 14,955 and 4,066 hectares in the regions of Paulista and Agudos, respectively. The spacing between the farms ranged from 4.50 m? (2222.2 plant / ha) 6.15 m? (1,626.0 plant / ha). Models were tested to estimate dominant height as a function of age for each farm individually in both regions. The fit of the models was performed by evaluating the Coefficient of Determination Adjusted and Standard Error of the Estimate, plus the distribution graphs of waste. The test Regazzi (2003) was applied to examine whether a single site model could be used to represent the growth of farms in each region. For the modeling process of growth and production, it was used the model of Clutter (1963). The initial basal area was estimated from models and arithmetic mean in each specified location. After the estimates of growth and production it has been found the behavior of these variables in each medium site, in order to ascertain whether a single equation could be used for each region using the Chow test (1960). To check the influence on production through the construction of curves of site by farm or grouping curves site we calculated the average percentage error of the intermediate site at ages 48, 60 and 72 months, and both were compared with volumes obtained by the permanent plots with the Dunnett`s test ( p - value ? 0.05). The dominant height models showed appropriate fit and a single equation is not sufficient to represent the growth of farms in each region. When applying the test, separately between the farms, it was observed that some of the data may be grouped without loss of precision. The Clutter model (1963) showed adequate projections of basal area and volume. From the Chow test (1960), estimates of growth and yield in the average site were generated individually for each farm, indicating that some farms could be grouped and others should not be grouped. The projection of the production with site curves by farm and by grouping them, the curves sites have provided projections of volume per unit area statistically similar. However, the comparison of projections from the mean error, the volume of wood per area was better designed with curves site per farm. However, the small difference between the two volumetric projections can be very important from an economic standpoint by making inferences in the average growth for large planted areas, it is recommended, therefore, the use of curves site per farm.