resumo
- Growing sweet chestnut (Castanea sativa) in high forest stands is a lucrative investment. Wood with medium and large dimensions can attain interesting profitability levels. Furthermore, forest producers are interested of new on this species and decision criteria for the best management are needed for landowners. So, accuracy is very important when evaluating volume in standing trees. Based on regression theory, it is developed predicting equations of wood volume as a function of tree variables easily measured (fast and cheap), such as diameter at breast height (DBH) and tree total height (h). Variable selection in volume equations was based on a stepwise modified procedure (linear models) and using generalized-F tests (nonlinear models). The resulting volume equation, with DBH and h as regressors, was confronted with the equation recommended by the Portuguese Forest Services. In order to evaluate wood volume at the lowest costs, the height to diameter relationship was also analyzed. Candidate regression equations were evaluated basing on fitting statistics and validation statistics, complemented with the use of generalized-F tests. The biological behavior of the equations was also considered.