Hyperspectral Imaging and Selected Biological Control Agents for the Management of Fusarium Head Blight in Spring Wheat.

Abstract

Fusarium spp. are important pathogens on cereals, capable of causing considerable yield losses and significantly reducing the quality of harvested grains due to contamination with mycotoxins. The European Union intends to reduce the use of chemical-synthetic plant protection products (csPPP) by up to 50% by the year 2030. To realize this endeavor without significant economic losses for farmers, it is crucial to have both precise early detection of pathogens and effective alternatives for csPPP. To investigate both the early detection of Fusarium head blight (FHB) and the efficacy of selected biological control agents (BCAs), a pot experiment with spring wheat (cv. 'Servus') was conducted under semi-field conditions. Spikes were sprayed with different BCAs prior to inoculation with a mixture of F. graminearum and F. culmorum conidia. While early detection of FHB was investigated by hyperspectral imaging (HSI), the efficiency of the fungal ( Trichoderma sp. T10, T. harzianum T16, T. asperellum T23 and Clonostachys rosea CRP1104) and bacterial ( Bacillus subtilis HG77 and Pseudomonas fluorescens G308) BCAs was assessed by visual monitoring. Evaluation of the hyperspectral images using linear discriminant analysis (LDA) resulted in a pathogen detection nine days post inoculation (dpi) with the pathogen, and thus four days before the first symptoms could be visually detected. Furthermore, support vector machines (SVM) and a combination of LDA and distance classifier (DC) were also able to detect FHB symptoms earlier than manual rating. Scoring the spikes at 13 and 17 dpi with the pathogen showed no significant differences in the FHB incidence among the treatments. Nevertheless, there is a trend suggesting that all BCAs exhibit a diminishing effect against FHB, with fungal isolates demonstrating greater efficacy compared to bacterial ones.

Plant phenotyping relevance

ハイパースペクトル画像と機械学習によるコムギ赤かび病の早期検出が中心的な方法的貢献であり、植物病害状態の推定を技術的に評価している。

Code and data availability

The paper's hyperspectral imaging datasets and analysis outputs are not publicly deposited; the Data Availability Statement states they are available only upon request from the corresponding author. No public code, models, or image data URLs are provided.

No evidence-backed public reproduction asset is currently recorded.