| Summary: | The techniques of stereology were applied to the microscopic transverse sections of 22 Brazilian tropical hardwood species. A minimum of 4 different trees, one slide per tree and 20 or 30 sampling fields per slide for each species were studied. Simple point fractions (P<sub>p</sub>), intersection counts (P<sub>L</sub>), and number of anatomical elements per unit area or feature count (N<sub>A</sub>) were converted to cell area and size distribution parameters such as mean tangential (d̅<sub>T</sub>) and mean radial cell diameters (d̅<sub>R</sub>), mean chord intercept (λ̅<sub>MCI</sub>) and mean free path (λ̅<sub>MFP</sub>) values. In addition, area percentages and two moments of cell diameter distributions (M₁ and M₂) were calculated. Analysis of variance (ANOVA) and Duncan's multiple range test were used to compare species. Within species differences were also tested.
The results indicated large differences among the species studied in all quantitative parameters calculated. Some of the differences appeared to be large enough to suggest strong species specificity. No strong similarity occurred among species of the same family. Most within species differences were statistically non-significant. ANOVA and associated Duncan's multiple range test showed highly significant species specificity for all quantitative anatomical characteristics. These tests could be used for successive elimination of species to determine their identity.
Bending and compression properties in the green condition were related to density. Residuals from those relationships were then related to six anatomical properties. Up to 80% of the variations in strength properties could be accounted for by density and an additional eight percent by quantitative anatomical parameters. === Master of Science
|
|---|
