A method to improve tree water use estimates by distinguishing sapwood from heartwood using Electrical Resistivity Tomography

Estimating whole-plant water use in trees requires reliable and accurate methods. Measuring sap velocity and extrapolating to tree water use is seen as the most commonly used. However, deducing the tree water use from sap velocity requires an estimate of the sapwood area. This estimate is the highest cause of uncertainty, and can reach more than 50 % of the uncertainty in the estimate of water use per day. Here, we investigate the possibility of using Electrical Resistivity Tomography to evaluate the sapwood area distribution in a plantation of Pinus elliottii. Electric resistivity tomographs of Pinus elliottii show a very typical pattern of electrical resistivity, which is highly correlated to sapwood and heartwood distribution. To identify the key factors controlling the variation of electrical resistivity, cross sections at breast height for ten trees have been monitored with electrical resistivity tomography. Trees have been cut down after the experiment to identify the heartwood/sapwood boundaries and to extract wood and sap samples. pH, electrolyte concentration and wood moisture content have then been analysed for these samples. Results show that the heartwood/sapwood patterns are highly correlated with electrical resistivity, and that the wood moisture content is the most influencing factor controlling the variability of the patterns. These results show that electric resistivity tomography could be used as a powerful tool to identify the sapwood area, and thus be used in combination with sapflow sensors to map tree water use at stand scale. However, if Pinus elliottii shows typical patterns, further work is needed to identify to see if there are species - specific characterictics as shown in previous works (<i.e>, electrolyte gradients from the bark to the heartwood). Also, patterns of high resistivity in between needles positions, which are not correlated with either wood moisture content or sapwood, appear to be artifacts. Thus, inversion methods have also to be improved to take into account these measurements issues.