We show that sapflow is a useful tool for studies of water fluxes in forest ecosystems, because (i) it gives access to the spatial variability within a forest stand, (ii) it can be used even on steep slopes, and (iii) when combined with eddy correlation measurements over forests, it allows separation of individual tree transpiration from the total water loss of the stand. Moreover, sapflow techniques are quite easy to implement.
Four sapflow techniques currently coexist, all based on heat diffusion in the xylem. We found a good agreement between three of these techniques. Most results presented here were obtained using the radial flow meter (Granier 1985).
Tree sapflow is computed as sap flux density times sapwood area. To scale up from trees to a stand, measurements have to be made on a representative sample of trees. Thus, a number of trees in each circumference class is selected according to the fraction of sapwood they represent in the total sapwood area of the stand. The variability of sap flux density among trees is usually low (CV. 10–15%) in close stands of temperate coniferous or deciduous forests, but is much higher (35–50%) in a tropical rain forest. It also increases after thinning or during a dry spell.
A set of 5–10 sapflow sensors usually provides an accurate estimate of stand transpiration. Transpiration measured on two dense spruce stands in the Vosges mountains (France) and one Scot's pine plantation in the Rhine valley (Germany) showed that maximum rate was related to stand LAI and to local climate. Preliminary results comparing the sapflow of a stand of Pinus banksiana to the transpiration of large branches, as part of the BOREAS programme in Saskachewan, Canada showed a similar trend.
For modelling purposes, tree canopy conductance (gc) was calculated from Penman-Monteith equation. In most experiments, calculated canopy conductance was dependent on global radiation (positive effect) and on vapour pressure deficit (negative effect) in the absence of other limiting factors. A comparison of the vapour pressure deficit response curves of gc for several tree species and sites showed only small differences among spruce, oak and pine forests when including understorey. Tropical rainforests exhibited a similar behaviour.