Abstract

Sap flow sensors measure the movement of sap (J_s, sap flux density) at discrete points within a tree trunk. Upscaling these measurements to estimate the water consumption of an entire tree (SF, whole-tree sap flow) typically requires assuming the trunk’s cross section is perfectly circular. This assumption will overestimate the area and underestimate the perimeter of a real trunk’s cross section. But how might it bias sap flow estimates in trees with irregularly shaped trunks?

We digitized trunk cross sections from Pisonia grandis, a tropical canopy tree that exhibits diverse trunk shapes, and compared sap flow from the traditional circular model (SF_circ) to a spatially explicit reference model (SF_ref) for a wide array of radial sap flux density profiles. Depending on the profile, we found that the circular assumption may underestimate sap flow by more than 25% or overestimate it by over 50% in the most irregular Pisonia trunks.

To mitigate these biases, we recommend measuring a trunk’s perimeter in addition to its diameter at breast height (DBH). These measurements enable the calculation of the trunk’s perimeter convexity (c_p), which strongly predicted sap flow bias in Pisonia (r² = 0.961) from the circular trunk assumption. Simulations show that for radial sap flux profiles typical of dicots, bias should remain within ±5% when c_p ≥ 0.96. For trees with c_p < 0.96, we introduce two new upscaling models that can significantly decrease sap flow biases. The most sophisticated model reduced bias to within ±5% for 31 of 33 tested trees with c_p values as low as 0.68 and requires just one additional field measurement, the depth of the deepest trunk concavity. These corrected sap flow models are available in a new R package, sapscaleR.

Full text.