Evaluating the threat of phytoplasma disease emergence in agroecosystems and natural habitats

Outbreaks of phytoplasma diseases annually cause billions of dollars in crop losses worldwide. A few efforts have been made to predict disease outbreaks and management continues to focus primarily on reducing pathogen spread following an outbreak. This study leverages machine learning to assess the global risk of emerging phytoplasma diseases using data from the literature on previous phytoplasma outbreaks in agroecosystems, combined with newly documented occurrences of phytoplasma-positive insects (potential vectors) in natural areas worldwide. By applying supervised machine learning on these datasets, key predictors of vector-host-phytoplasma interactions were identified and their importance in facilitating disease outbreaks was evaluated. The model highlights critical differences between two types of ecosystems and establishes a foundation for predicting new phytoplasma-host associations. These findings pave the way for targeted interventions to mitigate the risk of future outbreaks.