Leaf chlorophyll content is an important indicator of the growth and photosynthesis of maize under water stress. The promotion of maize physiological growth by (AMF) has been studied. However, studies of the effects of AMF on the leaf chlorophyll content of maize under water stress as observed through spectral information are rare. In this study, a pot experiment was carried out to spectrally estimate the leaf chlorophyll content of maize subjected to different durations (20, 35, and 55 days); degrees of water stress (75%, 55% and 35% water supply) and two inoculation treatments (inoculation with Funneliformis mosseae and no inoculation). Three machine learning algorithms, including the back propagation (BP) method, least square support vector machine (LSSVM) and random forest (RF) method, were used to estimate the leaf chlorophyll content of maize. The results showed that AMF increased the leaf chlorophyll content, net photosynthetic rate (A), stomatal conductance (gs), transpiration rate (E), and water use efficiency (WUE) of maize but decreased the intercellular carbon dioxide concentration (Ci) of maize and atmospheric vapor pressure deficit (VPD) regardless of the water stress duration and degree. The first-order differential spectral data can better reflect the correlation between leaf chlorophyll content and spectrum of inoculated maize when compared with original spectral data. The BP model performed bestin modeling the maize leaf chlorophyll content, yielding the largest R 2-values and smallest root mean square error (RMSE) values, regardless of stress duration. These results provide a reliable basis for the effective monitoring of the leaf chlorophyll content of maize under water stress.
Keywords: arbuscular mycorrhizal fungi; leaf chlorophyll content; machine learning algorithms; spectral reflectance; water stress.
Copyright © 2021 Sun, Yang, Yang, Wei, Li, Guo and Kong.