Abstract
Background and Aims: In contrast to the wealth of information on sap flow characteristics of forest trees at low-elevation sites, knowledge is scarce for trees within the treeline ecotone. Thus, our aim was to identify environmental controls on water loss in P. cembra trees at the treeline.
Methods: Three isolated individuals were equipped with thermal dissipation probes for monitoring sap flow density (QS ). A Jarvis-type model and multiple linear regression, with irradiance and vapour pressure deficit as explanatory variables, were applied to compare predicted and measured sap flow density.
Results: Both models successfully predicted measured QS . The Jarvis model tended to underestimate QS values by 25% on average, as compared with an underestimate of 8% by multiple linear regression analysis. The multiple linear regression analysis also allowed discrimination between effects of irradiance (RS ) and vapour pressure deficit (D) on QS , indicating that RS (ß-coefficient = 0.70) had a greater effect on QS than D (ß-coefficient = 0.25). Whole-tree water loss scaled to ground surface area (1.9 ± 0.5 mm d−1) was within the range reported for adjacent low-stature vegetation.
Conclusions: The results suggest that the multiple linear regression approach is superior to the non-linear Jarvis-type model, as this approach allows discrimination between the effects of RS and the effects of D on QS .
Acknowledgements
This work was funded by the Austrian Science Fund Project (FWF P 22206-B16) ‘Transpiration of conifers in contrasting environments’. Thanks are due to two anonymous reviewers for helpful suggestions on an earlier draft of this manuscript.