Beeckman, H. & Yin, Y. 2024. ‘Digitizing a xylarium beyond shared meta-data: the importance of quantitative wood anatomy, images and chemistry’. 26th IUFRO World Congress: T5.16 IAWA-IUFRO Symposium: Advancing Methods and Applications of Wood Identification. Book of abstracts. Stockholm : IUFRO.

Résumé de colloque

Xylaria are big institutional collections of wood samples that serve as scientific reference material. They are a result of long-term collecting efforts of multiple expeditions in the past, often in regions that are nowadays difficult to explore for logistical, conservational, or political reasons. Xylarium specimens have been key for comparative wood anatomy and establishing databases supporting botanical identification of traded timber. Strong tendencies in actual state-of-the-art research include the rising importance of quantitative approaches and the substantially improved means for producing, sharing, and analyzing images. At the same time there is a growing interest in combining microscopic observations of wood with chemical assessments. These trends urge for coordinated efforts of digitizing the invaluable information held in xylarium samples, apart from sharing metadata that are available for many xylaria. Quantitative features typically show variability, contrary to most of the qualitative characteristics commonly used in wood identification. There is indeed variability within species, but also within a single tree, e.g. vessel element or tracheid dimensions vary from pith-to-bark, in the height direction and according to different organs (trunk, roots, branches). More emphasis on quantitative features implies a statistical approach that typically needs study material consisting of more specimens than a classical wood anatomical description. This is a good reason for strengthening the network efforts between xylaria. There are also engaging perspectives for a bigger role of image-based assessments. Capturing a maximum of wood anatomical information in a collection of images involves taking pictures from the three principal angles, at different magnifications, different fields of view and using as well optical light, electron beams and X-rays. This digitizing process results in a multitude of images. Handling high resolution images in a standardized and efficient way is still challenging, but necessary to fully exploit the opportunities of deep learning. Beyond wood anatomy, several chemical analyses are also relevant for identification and provenance determination, including measurements of metabolites, stable isotopes and elemental assessments. Concerted actions will be key to share data that are maximally Findable, Interoperable and Reusable (FAIR principles)