Maintaining a steady increase of yields requires knowledge of plant stress physiology and modern techniques of quantitative data collection and analysis. Here, the chlorophyll fluorescence parameters are used for modelling of relative water content (RWC) in field-grown wheat cultivars. RWC is commonly used for the detection of plant tolerance to temporary droughts, but its determination is laborious and does not meet the requirements of a mass test like fluorescence detection. The paper presents a beta generalized linear mixed model (GLMM) fitted for RWC prediction based on chlorophyll fluorescence data repeatedly measured over time. The nature of fluorescence parameters with the strong correlations between them leads to the use of a multilevel principal component analysis to overcome this issue prior to the fitting of the model. Furthermore, a beta generalized estimating equation (GEE) model is fitted for identifying population-average effects of the parameters used. Finally, highly significant results in terms of prediction with the use of 10-fold cross-validation (R_Pearson-CV = 0.86, MAE_CV = 0.0365, RMSE_CV = 0.048) were obtained. Moreover, the population-average effects provide important information for the parameters used in RWC prediction. The beta GLMM can provide good predictions combined with important cultivar-specific information. Conclusively, these implementations can be a useful tool for drought tolerance improvement.

中文翻译：

---

基于快速叶绿素荧光瞬变的β混合回归模型定量估计田间小麦水分状况：一种耐旱性估计方法

保持产量的稳定增长需要了解植物胁迫生理学和定量数据收集和分析的现代技术。在这里，叶绿素荧光参数用于模拟大田小麦品种的相对含水量 (RWC)。RWC常用于检测植物对暂时性干旱的耐受性，但其测定费力，不符合荧光检测等大规模检测的要求。本文提出了一种适用于 RWC 预测的 beta 广义线性混合模型 (GLMM)，该模型基于随时间重复测量的叶绿素荧光数据。荧光参数的性质与它们之间的强相关性导致在模型拟合之前使用多级主成分分析来克服这个问题。此外，拟合了一个 beta 广义估计方程 (GEE) 模型，用于确定所用参数的总体平均效应。最后，使用 10 折交叉验证在预测方面取得了非常显着的结果（获得 R _Pearson-CV  = 0.86，MAE _CV  = 0.0365，RMSE _CV  = 0.048)。此外，总体平均效应为 RWC 预测中使用的参数提供了重要信息。beta GLMM 可以提供良好的预测，并结合重要的品种特定信息。总之，这些实施可以成为提高耐旱性的有用工具。