Review of embedding and non-embedding techniques for quantitative wood anatomy

In recent decades, xylem anatomical traits have become increasingly important in dendrochronological research, as they offer the unique opportunity to assess eco-physiological drivers of tree growth at intra-annual resolution. However, standard protocols for generating such data are still missing, leading to methodological uncertainty, and complicating data exchange among laboratories. Here, we compare protocols for high-quality permanent slide preparation in dendroanatomy and address the effects of paraffin embedding vs. non-embedding approaches. Tests are conducted on both gymnosperm and angiosperm wood types of widely distributed European tree species, considering cell wall thickness (CWT), mean lumen area (MLA), and hydraulic diameter (Dh). Results indicate that non-embedding does not significantly alter the qualitative and quantitative characteristics of permanent slides compared to embedded samples. Whereas the mean chronologies of MLA and Dh and their non-embedded counterparts share substantial high-frequency variance, the CWT chronologies reveal slightly larger discrepancies at inter-annual scale. However, methodological differences do not exceed 11.1 % for any parameter. While these results show high similarity between the two approaches, we recommend adopting the non-embedding procedure, since it saves resources and therefore allows to produce larger datasets. Regardless of the protocol used to build wood anatomical datasets, assembling large-scale networks of wood anatomical data could transform our understanding of forest responses to global changes.