Knighton James, Sanchez-Martinez Pablo, Anderegg Leander
Department of Natural Resources and the Environment, University of Connecticut, Storrs, Connecticut, USA.
School of GeoSciences, University of Edinburgh, Edinburgh, UK.
Sci Data. 2024 Dec 18;11(1):1336. doi: 10.1038/s41597-024-04254-4.
We present a dataset of plant hydraulic and structural traits imputed for 55,779 tree species based on TRY plant trait dataset observations and phylogenetic relationships. We collected plant trait values for maximum stomatal conductance (gs), xylem pressure at 12%, 50%, and 88% conductance loss (P12, P50, P88), maximum observed rooting depth (rd), photosynthetic Water Use Efficiency (WUE), maximum plant height (height), Specific Leaf Area (SLA), and leaf Nitrogen content (LeafN). We demonstrated that each of these traits exhibited remarkably large phylogenetic signals across all land plants. Based on the strength of this signal we then developed random forest (RF) models trained on TRY trait data to impute the traits of previously unstudied tree species using Phylogenetic Eigenvector Maps. We quantified imputed trait uncertainty by fitting RF model test dataset residuals to skew exponential power distributions accounting for heteroscedasticity, demonstrating encouraging lack of biases in the imputed dataset. The resulting dataset of imputed trait values can support global analyses of plant trait variations and species-level parameterization of earth systems models.
我们展示了一个基于TRY植物性状数据集观测结果和系统发育关系推算出的55779种树的植物水力和结构性状数据集。我们收集了植物性状值，包括最大气孔导度（gs）、导度损失12%、50%和88%时的木质部压力（P12、P50、P88）、观测到的最大生根深度（rd）、光合水分利用效率（WUE）、最大株高（height）、比叶面积（SLA）和叶片氮含量（LeafN）。我们证明，在所有陆地植物中，这些性状中的每一个都表现出显著的系统发育信号。基于该信号的强度，我们随后开发了基于TRY性状数据训练的随机森林（RF）模型，以利用系统发育特征向量图推算先前未研究树种的性状。我们通过将RF模型测试数据集残差拟合到考虑异方差性的偏态指数幂分布来量化推算性状的不确定性，结果表明推算数据集中令人鼓舞地没有偏差。由此产生的推算性状值数据集可支持植物性状变异的全球分析以及地球系统模型的物种水平参数化。
