Models for leaf area estimation of three forest species in a short coppice rotation

Abstract

It was obtained statistical models to estimate the leaf area (LA) based in the length (L) and the width (W) of Bambusa vulgaris, two different eucalypt clones, AEC-144 (spontaneous hybrid of Eucalyptus urophylla) and LW07 (Eucalyptus urophylla x Eucalyptus grandis), and Salix nigra leaves. The trees or clumps were provbrided from a short rotation coppice (SRC) for bioenergy, mainly characterized by the high tree density, in Botucatu, Sao Paulo, Brazil. It was collected, by chance, more than 4000 leaves that represented a quarter of the coppices. The bamboo and AEC-144 clone were, at the time, 22 months old, while the willow and LW07 clone were 18 months old. Young, intermediate and old leaves were mixed and measured. The measured leaves were correlated to obtain the simple linear eqs. (LA in function of L and W) and multiple linear regression (LA in function of L × W), to each species. All the species shown a positive correlation coefficient (r) to L (r = 0.75 to 0.95), W (r = 0.70 to 0.82) e L × W (r = 0.87 to 0.95), significative to p ≤ .05. The multiple linear models, that used L × W, are the most appropriated once it had better adjustments with the determination coefficients (R2) between 0.76 and 0.91 with exception in the case of S. nigra (willow), the R2 of the simple linear regression using L was similar to the multiple linear regression, 0.90 and 0.91 respectively, showing that it is possible to estimate the LA in willow just using length.