Assessing Chilling Injury in Cucumber Seedlings Using Chlorophyll Fluorescence Based on a Quantum Genetic Algorithm and Support Vector Regression Model

Abstract

HighlightsA method is proposed to analyze chilling injury in cucumber seedlings.A mutual information method was adopted to determine model output.Different modeling approaches were compared for improving accuracy.The optimal model input type was determined and discussed.Abstract. Low temperature limits photosynthesis and leads to reduced production in plants. Accurate monitoring of the physiological state of cucumber (Cucumis sativus L.) seedlings and assessing their chilling injury are of great importance for the off-season cucumber industry. Dark chlorophyll fluorescence parameters have been proven to be efficient for revealing environmental stress on plant growth. We used seedlings of cucumber cultivar Bonai 14-3 to test a nested set of fluorescence parameters to assess the response characteristics of dark chlorophyll fluorescence parameters during eight days at low temperatures. Mutual information analysis showed that the dark chlorophyll fluorescence parameter Fv/Fo, an indicator of the potential activity of photosystem II, gave the strongest correlation with stage of chilling injury. We compared three Fv/Fo prediction models with different inputs, constructed using a quantum genetic algorithm support vector regression method. The model showed the highest accuracy when the inputs were Fv and Fo measured before low-temperature treatment, coupled with temperature, and duration. The coefficient of determination of the training and testing sets were both greater than 0.98. The root mean square error and the mean absolute error were both less than 0.15. This model was able to achieve accurate prediction of Fv/Fo for seedlings in a low-temperature environment, providing a potential method for non-destructive analysis and diagnosis of cucumber chilling injury. Keywords: Chlorophyll fluorescence parameters, Low temperature, Mutual information, Prediction.