Basic density of eucalyptus wood: variability, sampling strategies and optimization
Name: GUSTAVO JASKE DA CONCEIÇÃO
Publication date: 27/02/2026
Examining board:
Name |
Role |
|---|---|
| EMANUELLA MESQUITA PIMENTA | Examinador Externo |
| GRAZIELA BAPTISTA VIDAURRE DAMBROZ | Presidente |
| MARIA NARUNA FÉLIX DE ALMEIDA | Examinador Externo |
| VANIELE BENTO DOS SANTOS | Examinador Externo |
Summary: Wood density is recognized as one of the main quality parameters of Eucalyptus spp., directly influencing industrial performance and the efficiency of conversion processes. Despite being widely studied, uncertainties remain regarding the representativeness of base-to-top stem positions and the sampling intensity required to estimate wood basic density (BD) with statistical robustness. This study evaluated longitudinal (base–top) variation and optimized sampling points along the stem to estimate the mean tree BD, comparing two sample types—wedges and discs—in terms of representativeness, longitudinal behavior, and statistical performance. A total of 23 commercial clones distributed across 15 locations in Brazil were analyzed, comprising 390 trees aged between 5.5 and 8 years. Wedge and disc samples were collected at 12 positions along the stem: 0%, DBH (diameter at breast height), 12.5%, 25%, 33%, 37.5%, 50%, 62.5%, 66%, 75%, 87.5%, and 100%. Statistical analyses included mean comparison tests (t-test and Dunnett test), evaluation of data distribution and base–top profiles, assessment of variability within and between trees, multiple regression using Exhaustive Search, and comparison of sampling strategies commonly adopted in the forestry sector. Longitudinal variation indices were proposed and applied, including the Base–Top Directional Index (BTDI), the Base–Top Absolute Variation Index (BTAV), the Absolute Difference between DBH and Mean Density (BD), the Percentage Representativeness of DBH (BD%), and the between-tree (BTA) and within-tree (WTA) amplitudes. The BD results demonstrated statistical equivalence between wedges and discs, with an average difference of only 2 kg m³ between sample types, and a similar base–top pattern characterized by greater homogeneity at intermediate stem positions (33% and 37.5%) and higher variability at the stem extremes. Models including one to three sampling positions explained between 88% and 99% of BD variation, with a statistical saturation effect observed as additional positions were included. The best single positions were 33% (wedges) and 37.5% (discs). Although the traditional strategy (0%, DBH, 25%, 50%, 75%, and 100%) yielded the highest adjusted R² values (>98%), simplified strategies with three positions maintained efficiencies above 93%, indicating the feasibility of reducing sampling effort with minimal loss of accuracy. The use of DBH as a single sampling point showed lower explanatory power and underestimated the mean tree BD, with differences ranging from 17 to 37 kg m³, highlighting its limitation as a sole reference position. Between-tree amplitude confirmed high genetic heterogeneity, exceeding 100 kg m³ within the same density class. It can be concluded that the mean stem density can be accurately estimated using a reduced number of strategically selected sampling positions, thereby reducing operational costs and laboratory effort. Wedges and discs can be used interchangeably, provided that consistent sampling and preparation protocols are adopted.
Keywords: Statistical modeling; Wood quality; Longitudinal variation; Eucalyptus spp.
