Phenomics‐driven insights into zoysiagrass drought resistance using small unmanned aircraft systems (sUAS)‐based hyperspectral images

Abstract

Abstract The application of small unmanned aircraft systems (sUAS)‐based high‐throughput phenotyping in plant breeding has advanced significantly over the past decade. Hyperspectral images and machine learning approaches offer potential to enhance drought resistance screening in turfgrass. However, large‐scale field applications remain limited, and the transition from controlled environments to real‐world phenotyping is not well understood. This study aimed to develop an sUAS‐based hyperspectral image workflow to monitor changes in turfgrass canopy reflectance during drought, validate previously reported indices from controlled environment studies in a large‐scale field study, and estimate visual turfgrass quality (TQ) from hyperspectral images. Images were collected from a zoysiagrass ( Zoysia spp.) mapping population at three dates under varying soil moisture conditions. Vegetation indices (VIs) related to light use efficiency, leaf pigments, senescence, water status, and green vegetation were computed and compared. Top‐performing genotypes under drought exhibited greater absorption in blue and red wavelengths and higher near‐infrared reflectance than poor‐performing ones. The photochemical reflectance index and plant senescence reflectance index were highly correlated with TQ ( r = 0.84 and −0.76), showed higher coefficient of variation (range 18%–37%), and had higher broad‐sense heritability (0.73–0.74) than normalized difference vegetation index (0.69), warranting their use in large‐scale field study. Machine learning models estimated TQ with a mean absolute error of 0.46. These findings highlight the importance of integrating VIs related to light use efficiency, leaf pigments, senescence, and water status to gain deeper insights into turfgrass drought response and support breeding for stress tolerance.

Plant phenotyping relevance

sUASハイパースペクトル画像によるキャノピー形質取得ワークフローを開発し、指標を検証して芝草品質を推定しており、フェノタイピング手法が中心である。

Code and data availability

The paper's data availability statement points to a Zenodo-hosted dataset of spectral reflectance measurements from the zoysiagrass mapping population under drought, which directly reproduces this paper's phenotyping measurements. No author analysis code or trained models were identified.