Modelling airborne inoculum and plant-pathogen interactions for enhancing potato late blight support systems
(2026)
Files
Thèse_VLeVourch.pdf
Embargo Access until 2027-02-06 - Adobe PDF
- 3.32 MB
Details
- Authors
- Supervisors
- Legrève, Anne
- Abstract
- Potato late blight (PLB), caused by Phytophthora infestans, remains one of the most destructive diseases affecting potato production worldwide and represents a major constraint for European agriculture. In the current context of the European Green Deal, aiming to drastically reduce pesticide use, improving the precision, sustainability, and biological relevance of late blight forecasting tools is essential. This thesis combines aerobiology, epidemiology, machine learning, and host–pathogen interaction studies to refine risk assessment frameworks and contribute to next-generation decision support systems (DSS) for managing PLB. First, a four-year aerobiological monitoring network was deployed across multiple potato-growing regions in Wallonia to characterize the spatiotemporal dynamics of P. infestans airborne inoculum. Airborne propagules were detected at all sites and in all years, but with marked heterogeneity driven by local meteorology, microclimate, and landscape structure. Inoculum was consistently detected before the first field symptoms and occasionally during winter, challenging the assumption that the pathogen becomes absent outside the cropping season. These findings indicate that airborne inoculum constitutes a reliable early indicator of epidemic onset and demonstrate the epidemiological value of aerobiological surveillance for improving DSS. Building on these results, artificial neural network (ANN) models were developed to predict early-season airborne inoculum using only meteorological variables. The models performed best at higher detection thresholds, with strong discriminatory power for significant inoculum events. Key predictive variables reflected biologically meaningful processes, including minimum temperature, relative humidity, leaf wetness, and short-term rainfall patterns. Despite these promising outcomes, predictive power remained constrained by coarse temporal resolution and limited aerobiological datasets. The work highlights the potential for integrating artificial intelligence-driven inoculum forecasting into operational DSS, while emphasizing the need for finer-scale environmental data, expanded monitoring networks, and real-time detection technologies. The final part of the thesis investigates interactions between isolates and cultivars through inoculation assays using increasing concentrations of eight P. infestans genotypes representative of the Belgian population on ten potato cultivars. Strong genotype-by-cultivar effects were observed, revealing substantial intra-lineage phenotypic diversity and demonstrating that pathogen fitness cannot be inferred from aggressiveness alone. Several isolates displayed high infection efficiency at very low inoculum doses, a trait with major implications for epidemic initiation and field risk assessment. Incorporating these traits, together with clonal lineage frequencies, into modelling frameworks markedly improved predictions of field late blight severity over nine years of data. These results confirm that controlled-environment assays can predict field outcomes when pathogen population structure and inoculum levels are explicitly integrated. Overall, this thesis provides new insights into the epidemiology of P. infestans and offers methodological advances for the development of more accurate, ecologically grounded forecasting tools. By combining aerobiology, machine learning, and host–pathogen interaction studies, it contributes a unified framework for reducing prophylactic fungicide use while maintaining robust crop protection. These findings support the transition toward sustainable disease management strategies required under evolving regulatory, economic, and environmental constraints, and propose concrete pathways for enhancing DSS performance at field and regional scales.
- Affiliations
Citations
Le Vourch, V. (2026). Modelling airborne inoculum and plant-pathogen interactions for enhancing potato late blight support systems. https://hdl.handle.net/2078.5/273589