Abstract Sap flow measurements are a crucial tool for studies of plant water relations, but upscaling to tree or stand transpiration requires accounting for the radial distribution of sap flow across the stem. Because radial sap flow profiles cannot be measured with single-point measuring methods, predicting their shape based on easier to measure structural or functional traits is a potential route to cost-efficiently improve water-use estimates. We used heat field deformation (HFD) sensors to measure sap flux density at different sapwood depths on 38 trees of eight tropical tree species in secondary dry forest patches in north-west Costa Rica. Based on a Bayesian hierarchical modeling framework, we developed a flexible model for radial profiles that expresses the average profile depth and the spread of the profile around this depth as functions of tree height, wood density and stem growth rate, while allowing for random tree and species effects. From the model output, we scale up to approximate whole-tree daily water use (DWU). Our model revealed pronounced tree height effects, with taller trees having narrower sap flow profiles peaking closer to the cambium. While it explained 96% of the variance in the dataset, large fractions were attributed to random species and tree individual effects. Including functional traits as predictors improved the accuracy of predictions considerably both for new trees and new species. DWU responded positively to tree height, but not to annual stem increment and wood density. Extrapolating the average sap flux in the outermost 2 cm over the entire sapwood area overestimated DWU on average by 26% compared to estimates based on radial profiles. Our model provides a starting point for future studies aiming to improve landscape-scale water-use estimates integrating single-point sap flow measurements and radial profiles from a subset of trees.

中文翻译：

---

树高预测哥斯达黎加热带干旱森林树种径向树液流剖面的形状

摘要 树液流量测量是研究植物水分关系的重要工具，但升级到树木或林分蒸腾作用需要考虑树液流量的径向分布。由于无法使用单点测量方法测量径向树液流剖面，因此根据更容易测量的结构或功能特征预测其形状是经济高效地改进用水估算的潜在途径。我们使用热场变形 (HFD) 传感器测量哥斯达黎加西北部次生干旱森林斑块中 8 种热带树种的 38 棵树上不同边材深度的树液通量密度。基于贝叶斯分层建模框架，我们为径向剖面开发了一个灵活的模型，该模型将平均剖面深度和剖面在该深度附近的分布表示为树木高度、木材密度和茎生长速率的函数，同时允许随机的树木和物种效应。从模型输出中，我们放大到近似全树每日用水量 (DWU)。我们的模型揭示了明显的树高效应，较高的树具有更窄的树液流剖面，其峰值更靠近形成层。虽然它解释了数据集中 96% 的差异，但很大一部分归因于随机物种和树木个体效应。将功能性状作为预测因子，大大提高了对新树和新物种的预测准确性。DWU 对树高有积极响应，但对年度茎干增量和木材密度没有响应。与基于径向剖面的估计值相比，外推整个边材区域最外层 2 厘米的平均树液通量平均高估了 DWU 26%。我们的模型为未来的研究提供了一个起点，旨在改进景观尺度的用水估计，整合单点树液流测量和来自树木子集的径向剖面。