Abstract Precise nitrogen (N) fertilization requires new indices of plants’ nutritional status. Non-structural carbohydrates (NSC) are the energetic currency of plants and can, thus, serve as a physiological indicator for their condition. Nevertheless, only a few records exist about NSC composition and allocation in crops, and their relationship with N uptake and the current methods to detect NSC compositions in plants are cumbersome and expensive, which limits their use. The current work aimed to associate the nutritional status of almond trees with the carbohydrate compositions in their roots, branches, and leaves by a high-throughput technique. We found that low N availability forces trees to allocate carbohydrates to their roots. High N availability, on the other hand, promoted above-ground vegetative growth, and minimized carbohydrate storage in the leaves. These observations implied that carbohydrate distribution could, indeed, serve as an indication of the nutritional status of the trees. To measure NSC content efficiently, we attempted to quantify soluble carbohydrates and starch in dried and powdered tissues by visible-to-shortwave infrared (VIS-NIR-SWIR; 350–2500 nm) reflectance spectroscopy, which is an inexpensive, safe, and non-destructive technique. We applied several multivariate statistical models based on the spectral datasets, including partial least squares-regression (PLS-R) and discriminant analysis (PLS-DA) as supervised registration and classification models. PLS-DA results of the N gradient in the roots and leaves showed an overall accuracy of 94% and 98%, respectively. PLS-R model performances of soluble carbohydrates and starch improved, in terms of the coefficient of determination (R2), if the leaf and root samples were integrated. Moreover, we found that the SWIR spectral region (1100–2500 nm) had unique reflectance features that revealed the carbohydrate composition and starch concentrations in the different plant tissues. The analyses also clustered the reflectance by tree part (root, branch, or leaf tissues) and N availability, forming a holistic model that can identify the nutritional status of trees. Conclusively, it is suggested that reflectance spectroscopy at the SWIR spectral region could guide precise fertilization by high-throughput identification of plants’ seasonal metabolism.

中文翻译：

---

通过碳水化合物的可见光到短波红外反射光谱评估杏仁树的氮状态

摘要 精确施氮（N）需要新的植物营养状况指标。非结构性碳水化合物 (NSC) 是植物的能量货币，因此可以作为植物状况的生理指标。然而，关于作物中 NSC 组成和分配的记录很少，而且它们与 N 吸收的关系以及目前检测植物中 NSC 组成的方法既繁琐又昂贵，限制了它们的使用。目前的工作旨在通过高通量技术将杏仁树的营养状况与其根、枝和叶中的碳水化合物成分联系起来。我们发现低氮可用性迫使树木将碳水化合物分配到它们的根部。另一方面，高氮可用性促进了地上营养生长，并最大限度地减少叶子中的碳水化合物储存。这些观察结果暗示碳水化合物的分布确实可以作为树木营养状况的指示。为了有效地测量 NSC 含量，我们尝试通过可见光到短波红外（VIS-NIR-SWIR；350-2500 nm）反射光谱来量化干燥和粉末组织中的可溶性碳水化合物和淀粉，这是一种廉价、安全且非-破坏性技术。我们应用了几种基于光谱数据集的多元统计模型，包括偏最小二乘回归 (PLS-R) 和判别分析 (PLS-DA) 作为监督注册和分类模型。根和叶中 N 梯度的 PLS-DA 结果分别显示总体准确度为 94% 和 98%。如果整合叶和根样品，可溶性碳水化合物和淀粉的 PLS-R 模型性能在决定系数 (R2) 方面得到改善。此外，我们发现短波红外光谱区域（1100-2500 nm）具有独特的反射特征，揭示了不同植物组织中的碳水化合物组成和淀粉浓度。分析还按树木部分（根、树枝或叶组织）和 N 可用性对反射率进行了聚类，形成了一个可以识别树木营养状况的整体模型。总之，表明短波红外光谱区域的反射光谱可以通过高通量识别植物季节性代谢来指导精确施肥。我们发现 SWIR 光谱区域（1100-2500 nm）具有独特的反射特征，可以揭示不同植物组织中的碳水化合物组成和淀粉浓度。分析还按树木部分（根、树枝或叶组织）和 N 可用性对反射率进行了聚类，形成了一个可以识别树木营养状况的整体模型。总之，表明短波红外光谱区域的反射光谱可以通过高通量识别植物季节性代谢来指导精确施肥。我们发现 SWIR 光谱区域（1100-2500 nm）具有独特的反射特征，可以揭示不同植物组织中的碳水化合物组成和淀粉浓度。分析还按树木部分（根、树枝或叶组织）和 N 可用性对反射率进行了聚类，形成了一个可以识别树木营养状况的整体模型。总之，表明短波红外光谱区域的反射光谱可以通过高通量识别植物季节性代谢来指导精确施肥。形成一个整体模型，可以识别树木的营养状况。总之，表明短波红外光谱区域的反射光谱可以通过高通量识别植物季节性代谢来指导精确施肥。形成一个整体模型，可以识别树木的营养状况。总之，表明短波红外光谱区域的反射光谱可以通过高通量识别植物季节性代谢来指导精确施肥。