Some pests and diseases pose such a serious threat to plants and trees that they have to be prevented from spreading right away. In the case of quarantine pests that can in-flict major damage in agriculture or forestry, the infestation is subject to mandatory re-porting because of the risk that they could spread into new areas. In order to identify an infection, trained personnel carry out a visual assessment that involves an examina-tion of each individual tree and plant. These field inspections by plant health services are time-consuming and require considerable staff resources, particularly since sus-pected samples have to be sent to a molecular biology lab for confirmation. Working with RLP AgroScience GmbH, researchers at the Fraunhofer Institutes for Factory Oper-ation and Automation IFF and for Biomedical Engineering IBMT have now significantly accelerated this lengthy process through the PhenoTruckAI project (see below). The PhenoTruck® is a mobile lab that will enable quarantine pest identification to be car-ried out exactly where growers need it, namely on site.

“As a result of climate change and global trade, dangerous non-native plant pests are continuously finding their way to Germany. Quarantine pests pose a particularly serious threat to fruit farming and viticulture,” says Fraunhofer IFF researcher Bonito Thielert. “Taking phytoplasma diseases in fruit farming and viticulture as an example, we’re vali-dating pest detection using the PhenoTruck®—an off-road laboratory vehicle developed in this project that is equipped with specialized tools for monitoring plant diseases.” In-itial tests and measurement campaigns have already been carried out in wine-growing regions in Rhineland-Palatinate and Italy, focusing on the quarantine diseases Flavescence dorée (FD), Palatinate grapevine yel-lows (PGY) and the disease known as Bois noir (BN), all of which affect grapevines. The mobile lab also enables researchers to analyze widely occurring diseases such as apple proliferation and pear decline, which can potentially result in complete crop loss in af-fected orchards.

Innovative early diagnostics directly on site

The project partners combined several technologies so as to be able to detect disease symptoms early on. RLP AgroScience GmbH contributed its agricultural expertise and performed the molecular biological analyses, while Fraunhofer IFF provided expertise in AI-optimized monitoring and hyperspectral analysis. Fraunhofer IBMT was responsible for the design, development, integration and implementation of the mobile Pheno-Truck® platform.

As a first step, drones equipped with multispectral sensors are deployed for automated, large-scale monitoring, with the multispectral images enabling the survey of large culti-vation areas. In addition, affected zones are documented using a specially developed field assessment app. Symptomatic leaves are then analyzed in the PhenoTruck® by means of hyperspectral cameras that rapidly detect and localize suspicious discolora-tion. Thielert: “Early leaf discoloration is a key indicator of phytoplasma diseases. When a plant shows symptoms in a leaf sample in the lab, hyperspectral analysis reveals these in defined wavelength ranges with greater clarity and consistency than would be possi-ble based on visible color changes alone.”

AI as a key technology for analyzing disease symptoms

The next step involves the use of artificial intelligence methods for fast and reliable analysis of the captured data. “Our AI models are able to detect phytoplasmosis with a high degree of certainty, achieving accuracy rates of 95 to 99 percent. One particular feature is the AI’s ability to process datasets automatically.” The researchers also trained the AI models to differentiate between the relevant grapevine phytoplasma dis-eases, namely Flavescence dorée (FD), Bois noir (BN) and Palatinate grapevine yellows (PGY). This is important because BN and PGY are not as dangerous as FD: Unlike FD they do not spread epidemically from vine to vine. Although the symptoms of these diseases are very similar and often confused with one another, BN and PGY pose a lower risk due to their slower and more limited spread. The phytoplasma diseases were distinguished with 80 percent accuracy.

The preliminary selection is made based on hyperspectral analysis, and samples sus-pected of infection are then examined directly on site by means of molecular biological testing in the PhenoTruck® lab. This is because phytoplasma diseases in grapevines and fruit trees can only be confirmed with certainty using molecular methods. A rapid test developed specifically for the mobile lab (LAMP) takes one hour to complete, so it is somewhat faster than a PCR analysis. Given the platform’s wide range of applications, the aim is to use the PhenoTruck® for research and development in the future.