Assessing wood formation phenology with dendrometers: opportunities and pitfalls

Comprehending the development, function, and adaptation of trees to changing environments requires a comprehensive understanding of the phenology of wood formation. In recent decades, numerous studies have investigated the impact of climate change on wood formation, with a focus on the start, end, and duration of the growing season. While some studies use microcores to directly observe wood phenology on anatomical sections, which requires a great deal of laboratory work, many others use indirect observations of variations in trunk circumference measured using dendrometers. However, it has not yet been properly assessed whether dendrometer records can provide reliable estimates of the phenology of wood formation. To answer this question, we collected weekly data from band dendrometers and wood microcores of six important European species (European beech, European larch, Norway spruce, Scots pine, silver fir, and pedunculate oak) growing at five different study sites located in eastern France over a period of three years or more. Classical logistic growth curves were used to fit dendrometer measurements to determine the days of the year when 5% and 95% of total annual growth are completed. Almost direct observations of the beginning and end of wood growth were obtained by processing and analyzing microcores. At the beginning of the growing season, dendrometer estimates and microcore observations were in agreement for fir and spruce trees. However, for larch, beech, pine, and oak trees, dendrometer estimates were consistently delayed by approximately 10, 13, 15, and 25 days, respectively. At the end of the growing season, the dendrometer approach did not yield significant differences for fir, larch, spruce, and oak trees. However, it provided earlier estimates for beech and later estimates for pine trees. Reduced major axis regressions indicated significant linear relationships between dendrometer estimates and microcore observations for fir, spruce, pine and beech at the beginning of the growing season and for fir, beech and oak at the end. The study explored the impact of tree species' life traits on the deviation level between the dendrometer and microcore approaches. The results indicate that deviation decreases with an increase in growth rate, but increases with greater heterogeneity of tree-ring structure (from diffuse-porous to conifer and ring-porous tree-rings) and roughness of the bark (from smooth or scaly to fissured bark types). The study revealed that deviation decreases with elevation for larch trees in the southern Alps. Conversely, for beech, fir, pine, spruce, and oak trees in the northeast of France, deviation increases with spring precipitation. This study highlights the challenges associated with using band dendrometers to estimate wood formation phenology. While the 'cheap and fast' band dendrometer approach may seem like an attractive alternative to the 'expensive and labor-intensive' microcore approach, it is important to consider the limitations of this method. Indeed, our results indicate that the accuracy of band dendrometers estimates depends on tree growth rate, species life traits, site conditions, and climate variability. Therefore, it may be challenging to use them to investigate tree adaptation to climate changes and changes in woody carbon sequestration.