Click on “Download PDF” for the PDF version or on the title for the HTML version. If you are not an ASABE member or if your employer has not arranged for access to the full-text, Click here for options. Application of Chlorophyll a Fluorescence in Analysis and Detection of Bacterial Wilt in Tomato PlantsPublished by the American Society of Agricultural and Biological Engineers, St. Joseph, Michigan www.asabe.org Citation: Journal of the ASABE. 65(2): 347-356. (doi: 10.13031/ja.14696) @2022Authors: Xin Wang, Wei Yang, Yu Yang, Min Huang, Ya Guo, Qibing Zhu Keywords: Chlorophyll a fluorescence, Extreme learning machine, Genetic programming, Tomato bacterial wilt. Highlights Eight JIP-test parameters most relevant to tomato bacterial wilt were identified. A novel model was developed for detection of tomato bacterial wilt. The overall detection accuracy of the developed model was 88.42%. Abstract. Bacterial wilt seriously threatens global tomato yield. Timely and accurately identification of plants infected with bacterial wilt is crucial to the implementation of disease management practices, but such detection methods are lacking. In this study, chlorophyll a fluorescence (ChlF) was used in the analysis and detection of tomato bacterial wilt. ChlF induction curves were collected from the leaves of control and infected plants after different days-post-inoculation (DPI), and eight JIP-test parameters most relevant to tomato bacterial wilt were selected from 22 JIP-test parameters through statistical analysis. A novel detection model, multidimensional multiclass genetic programming with multidimensional populations extreme learning machine (M3GP-ELM), was developed to identify tomato plants infected with bacterial wilt based on the selected JIP-test parameters. The M3GP-ELM model used a genetic programming algorithm to perform linear and/or non-linear transformations on the selected eight variables and then used the classification accuracy of ELM as a fitness function to evaluate the performance of the transformed variables. The results of the experiment indicated that the differences in the ChlF induction curves and the eight selected JIP-test parameters between the infected group and the control group became more obvious with increased time after inoculation. Compared with partial least squares discriminant analysis (PLS-DA), support vector machine (SVM), and ELM, the M3GP-ELM model achieved the best detection performance with an overall accuracy of 88.42% and an accuracy of 82.83% at the early stage (1 to 5 DPI). Therefore, ChlF technology combined with M3GP-ELM has the potential to detect tomato bacterial wilt. (Download PDF) (Export to EndNotes)
|