Discrimination between healthy and tomato spotted wilt virus infected tomato plants by means of temperature-modulated gas sensors

Plant virus diseases represent a major issue in agriculture, with relevant economic impact. The correlation between the plant health status and the emission of characteristics volatiles has been well established in the literature of analytical chemistry. However, the development of a cost-affordable technology suitable to work in-field for the automatic detection of these olfactive signals remains a major challenge in gas sensing. In this work, we investigated the use of metal oxide gas sensors working with temperature modulation to detect the outbreak of tomato spotted wilt virus infection in tomato plants, chosen as a relevant case study. To handle sensor-to-sensor reproducibility issues and plants’ individualities we designed experiments comprising 8 plants per time, each equipped with a dedicated sensor. Considering the complex time dependence of plants’ emissions and infection-independent interfering effects, we adopted an analysis method exploiting the signals from the whole network. Results indicates that the proposed technology detects the outbreak of infection through the detection of an anomalous signals from individual sensors with respect to the common behavior of the network, and is potentially suitable for autonomous, non-invasive, in-field operation. Polymerase Chain Reaction and Gas-Chromatography Mass-Spectroscopy characterizations suggest that the sensing system respond mainly to volatiles of plant tissues damage. These results indicate the potentialities of this technology for the development of a distributed network aimed at a non-invasive monitoring of plants health status in greenhouses. • A network of cheap and portable gas sensors is proposed to track the health status of tomato plants. • Sensors are based on metal oxide semiconductors working with temperature modulation. • Using Tomato Spotted Wilt Virus (TSWV) as target, the network detects infection within a few days. • Sensors data correlates with volatiles released upon infection-induced tissue damage. • The technology is suitable to perform a non-invasive detection, working on-field in green houses.