Genome-wide association study of wheat chlorophyll dynamics under drought and irrigation using multispectral UAV phenotyping

Yukun Cheng¹Wanlong He¹Fuxing Zheng¹Feifei Zhang¹Bin Bai²Na Sun³Wei Wang⁴Hongwei Geng^1*

• ¹College of Agriculture, Xinjiang Agricultural University, Urumqi, Xinjiang, China
• ²Wheat Research Institute, Gansu Academy of Agricultural Sciences (CAAS), Lanzhou, Gansu Province, China
• ³Yili Institute of Angricultural Science, Yili, China
• ⁴School of Computer Science and Information Engineering, Anyang Institute of Technology, Anyang, Henan, China

High-throughput phenotypic analyses using multispectral unmanned aerial vehicle (UAV) technology are essential for enhancing gene mining in both quantity and quality. This study was conducted with winter wheat at the Zepu and Manas experimental bases in fall 2019. We compared manually obtained chlorophyll content data during the Heading, Flowering, and Grain Filling Stages with multispectral data collected by drones. Values from manual measurements and UAV multispectral data were used to construct a model for predicting chlorophyll contents, which facilitated a genome-wide association analysis of correlated data.The findings support the feasibility of gene mining using remote sensing techniques. Artificial chlorophyll content ranged from 43.96 to 65.85 under normal conditions and from 45.00 to 66.33 under drought stress. UAV-predicted chlorophyll values ranged from 47.59 to 62.29 (normal) and 47.83 to 65.89 (drought). Both measured and predicted chlorophyll content increased across growth stages: heading < flowering < filling. Under normal conditions, correlation coefficients between measured and predicted values were 0.90-0.93 (heading), 0.91-0.92 (flowering), and 0.88-0.90 (filling). Under drought stress, correlations were 0.57-0.70, 0.89-0.91, and 0.94-0.96, respectively. The neural network model showed high accuracy in retrieving chlorophyll content.A genome-wide association analysis identified 308 loci, with predicted values revealing 206 loci across 21 chromosomes explaining 7.58%-19.58% of the phenotypic variation. Correlation analysis of measured values identified 102 loci across 21 chromosomes, accounting for 9.31%-15.83% of the variation. Two rounds of association analysis detected 18 overlapping loci on chromosomes 1A, 1B, 2B, 3B, 4B, 5A, 5B, 5D, 6B, 6D, 7A, and 7B, explaining 7.58%-19.58% of the overall phenotypic variation. Correlation analysis indicated that site P had lower predicted values, stronger correlations, higher variation rates, and superior prediction quality.After examining loci from both predicted and measured values, 21 genes potentially related to chlorophyll content were identified, including those encoding a chlorophyll a/b-binding protein, aquaporin, and chlorophyll kinase. This study confirms that chlorophyll content inversion via UAV multispectral modeling is reliable for genome-wide association analysis, offering researchers high-throughput, efficient, and accurate chlorophyll phenotyping.