Speaker
Description
The elemental composition of wood reflects a combination of physiological processes, environmental conditions, and soil geochemistry, making it a valuable source of information for provenance studies, biodiversity research, and conservation science. Energy-dispersive X-ray fluorescence (EDXRF) offers important advantages such as being rapid, multielemental, nondestructive with minimal sample preparation. In this study, an exploratory chemometric evaluation of EDXRF data was conducted to investigate whether elemental fingerprints can reveal systematic differences between wood samples collected from two major Brazilian biomes: the Amazon Rainforest and the Atlantic Forest.
The dataset comprised 126 measurements representing 54 biological trees: 49 Amazon trees and five Atlantic Forest trees, distributed among ten and four species, respectively. Because the Atlantic Forest dataset included repeated radial and directional measurements from the same trees, inferential analyses were conducted at the tree level by summarizing repeated positions using the median. Values below the detection limit were retained as missing during data assembly; remaining tree-level missing values were median-imputed calculated separately for each elemental variable, and the elemental variables were standardized as z-scores.
Principal component analysis was used for exploratory visualization. The first two principal components explained 54.0% of the total variance, with PC1 accounting for 32.1% and PC2 for 21.9%. The score was distributed with partial overlap between the two sampled groups, although Atlantic Forest measurements tended to occupy more positive PC1 values. Positive PC1 scores were associated mainly with K, Ca, and Cl, whereas Cr, Mn, and Co contributed in the opposite direction. PC2 was influenced positively mainly by Cr and, to a lesser extent, by Cl, K, Co, Ca, and Mn, while P contributed toward negative PC2 scores. The strongest pairwise correlations were observed for K–Ca and Cr–Co, both with r = 0.66, followed by Cl–K (r = 0.54) and Cr–Mn (r = 0.52).
A tree-level permutational multivariate analysis of variance based on Euclidean distances and 9,999 permutations indicated a significant difference between the multivariate centroids of the sampled biomes (pseudo-F = 4.454, R² = 0.0789, p = 0.0017). Biome grouping therefore accounted for approximately 7.9% of the standardized multielement variation. Standard PERMDISP did not detect a significant difference in within-group dispersion (F = 2.484, p = 0.1958).
Results provide exploratory evidence that the sampled Amazon and Atlantic Forest wood sets differ in their XRF elemental profiles, particularly along gradients involving K, Ca, Cl, Cr, Mn, and Co. Thus, standardized non-destructive XRF analysis can serve as a preliminary screening in wood provenance studies, with broader validation.
| Keywords | X-ray fluorescence; wood provenance; Amazon; Atlantic Forest; dendrochemistry; multivariate analysis. |
|---|---|
| Proceedings Publication Intent | Yes |
| Student Award Eligibility | no |
| Financial Support Request | no |