Abstract
The influence of nutrient availability on transpiration is not well understood, in spite of the importance of transpiration to forest water budgets. Soil nutrients have the potential to affect tree water use through indirect effects on leaf area or stomatal conductance. For example, following addition of calcium silicate to a watershed at Hubbard Brook, in New Hampshire, streamflow was reduced for 3 years, which was attributed to a 25% increase in evapotranspiration associated with increased foliar production. The first objective of this study was to quantify the effect of nutrient availability on sap flux density in a nitrogen, phosphorus, and calcium addition experiment in New Hampshire in which tree diameter growth, foliar chemistry, and soil nutrient availability had responded to treatments. We measured sap flux density in American beech (Fagus grandifolia, Ehr.), red maple (Acer rubrum L.), sugar maple (Acer saccharum Marsh.), white birch (Betula papyrifera Marsh.), or yellow birch (Betula alleghaniensis Britton.) trees, over five years of experiments in five stands distributed across three sites. In 2018, 3 years after a calcium silicate addition, sap flux density averaged 36% higher in trees in the treatment than the control plot, but this effect was not very significant (p = 0.07). Our second objective was to determine whether this failure to detect effects with greater statistical confidence was due to small effect sizes or high variability among trees. We found that tree-to-tree variability was high, with coefficients of variation averaging 39% within treatment plots. Depending on the species and year of the study, the minimum difference in sap flux density detectable with our observed variability ranged from 46% to 352%, for a simple ANOVA. We analyzed other studies reported in the literature that compared tree water use among species or treatments and found detectable differences ranging from 16% to 78%. Future sap flux density studies could benefit from power analyses to guide sampling intensity. Including pretreatment data, in the case of manipulative studies, would also increase statistical power.