Radial stem variations – a source of tree physiological information not fully exploited yet

Time series of stem radius (SR) variations offer information about radial stem growth and tree water relations in unmatched quality and resolution (Steppe et al. 2006; Zweifel et al. 2006). However, the task of turning raw SR displacement readings into physiologically reasonable measures is more complicated than it appears at first sight. There is thus great potential still waiting to be discovered in terms of interpreting dendrometer readings. Chan et al. (2015) have added another important piece of knowledge to this topic with their approach presented in this issue of Plant, Cell & Environment. The main difficulty to mechanistically interpreting radial stem variations (measured over living bark) is due to potentially co-occurring and partially opposing processes. On the one hand, there is the irreversible stem expansion of growing cells, namely, the radial increase because of dividing and enlarging wood and bark cells in the cambium (termed growth, GRO). On the other hand, there is the reversible, tree water deficit-induced shrinking and swelling of the stem (in the former called TWD), caused by imbalances between transpiration and root water uptake (Zweifel et al. 2005), and processes altering osmotic water potentials, for example, sugar loading and unloading in the phloem (Mencuccini et al. 2013). GRO is a one-directional process that only ever increases SR, whereas changing water potentials is bi-directional and can either lead to increasing or decreasing SR. Consequently, radial shrinking of the stem is always clearly attributable to decreasing water potentials and thus increasing TWD under the assumption that there is no structural degradation of the existing stem tissue structure. Radial increase, however, can either be induced by returning water and therefore swelling tissues or by GRO (Drew & Downes 2009). Consequently, any partitioning approach for SR time-series data needs a concept to define potential growth processes during periods of contracted stems, because this process is not a priori deducible from a single dendrometer measurement. Chan et al. (2015) solved this problem with an approach that is based on two SR readings measured in parallel over the bark and on the xylem. They combine these measurements with a model calculating xylem water potentials as the main driver of diurnal stem radius fluctuations. The difference between the modelled and measured fluctuations is interpreted as growth (cambial activity) and osmotic pressure changes. The model needs (only) two parameters to be determined and is therefore ranked in about the middle of approaches ranging from very complex, multi-parameter, tree water relations and carbon transport models (De Schepper & Steppe 2010; Sevanto et al. 2011; Mencuccini et al. 2013) to very simple approaches separating growth-related and tree water-related fractions of stem radius readings without the need of a model or additional measurements (Zweifel et al. 2005; Deslauriers et al. 2007). The entire range of approaches has their specific pros and cons and faces different levels of difficulties and most likely also inaccuracies. Of all the various types of dendrometer applications, the over bark measurement is by far the most frequently applied worldwide. The abundance of tree physiological information in this SR time series is still poorly exploited, and one reason for that may be the difficulties mentioned in turning displacement readings into physiologically realistic measures. Other reasons may be of amore technical or practical nature, because models to separate SR data into GRO and TWD often require additional physiological data to parameterize the model, particularly data that might not be available or are difficult to measure over longer time periods (Zweifel et al. 2014), such as the on-xylem SR measurements needed for the method of Chan et al. (2015). Simple approaches for disentangling GRO and TWD without a modelling component, however, may not be accurate enough to allow for a proper separation of the two main fractions determining SR fluctuations. Nevertheless, a simple SR disentangling approach is needed that has high physiological accuracy to allow separating the two main fractions of SR time series from a single dendrometer, from studies measuring SR over bark. Is such an approach just wishful thinking?Maybe yes, but let me put forward a hypothesis that could help solve the problem: