Key stress indicators from chlorophyll fluorescence in five desert plant species

• Published in: Ecological indicators 2022-12, Vol.145, p.109679, Article 109679
• Main Authors: Jin, Chuan, Zha, Tianshan, Bourque, Charles P.-A., Liu, Peng, Jia, Xin, Tian, Yun, Li, Xinhao, Liu, Xinyue, Guo, Xiaonan, Xu, Mingze, Kang, Xiaoyu, Guo, Zifan, Wang, Ning
• Format: Article
• Language: English
• Subjects: Chlorophyll fluorescence; Desert species; Dryland; Environmental stress; Photosystem II photochemistry; Random forest
• ISSN: 1470-160X; 1872-7034
• DOI: 10.1016/j.ecolind.2022.109679

•We studied the prospect of creating new daytime indicators in desert plant species.•Parametric and non-parametric methods were used to test the feasibility.•The y-intercept and slope of ФPSII-to-PAR-based regressions provided improvement.•The new method provided greater discrimination at the interspecific level. Pulse-amplitude modulated chlorophyll fluorescence (ChlF) is widely used to measure environmental stress in plants. Yet, its continuous, long-term usage in situ is challenged due to the lack of appropriate daytime indicators. We investigated the prospect of creating new daytime indicators based on an application of linear regression on daily measurements of photochemical efficiency (ФPSII) and photosynthetically active radiation acquired in situ at 30-minute intervals. Daily regression parameters, thus generated, were subsequently used in a broader context of desert-plant response to environmental change. Here, we compared parametric and non-parametric methods to test the feasibility of daytime-based regression parameters, i.e., maximal quantum yield of photosystem II photochemistry (Fv/Fm) and daily mean ФPSII, in gauging physiological response in three shrub and two herb species as influenced by eight environmental variables. Results demonstrated that: (i) Random Forest (RF; a non-parametric method) provided the best assessment of interaction between ChlF-based parameters and environmental variables, compared with multiple linear regression (MLR; a conventional parametric method); (ii) variable importance and partial dependence plots indicated that the daily regression parameters of the ФPSII-to- photosynthetic photon flux density relationship (y-intercept and slope) were superior to those of ФPSII and Fv/Fm; and (iii) compared with Fv/Fm, the y-intercept and slope improved discrimination of interspecific differences (p