Abstract
Sustainable control of microbial pathogens requires alternatives to chemical agents. However, the efficacy of physical methods like Ultraviolet-C (UVC) radiation is often inconsistent due to poorly understood, pathogen-specific resistance mechanisms. To address this, we investigated the differential responses of cacao-infecting fungi (Colletotrichum gloeosporioides and the more resistant Pestalotiopsis sp.) to UVB (305 nm) and UVC (275 nm) radiation. We developed an integrated framework using quantitative morphology, hyperspectral imaging (HSI), and machine learning to dissect the physiological underpinnings of UV sensitivity. UVC proved significantly more potent than UVB; for example, a 4-min UVC exposure achieved a similar level of inactivation on a sensitive isolate as a 30-min UVB exposure. After 30 min of UVC, the resistant Pestalotiopsis sp. maintained an 89% survival rate, whereas C. gloeosporioides isolates were almost completely inactivated (< 8% survival). HSI revealed that this resistance correlated with physiological stability, while sensitive isolates exhibited significant biochemical disruption. Machine learning models successfully classified isolates based on their UV-induced phenotypes with over 73% accuracy. This understanding enabled targeted strategies, such as synergistic treatment with sonication, which overcame the high resistance of Pestalotiopsis sp. Our work provides a mechanistic basis for optimizing physical pathogen controls by linking non-invasively measured physiological states to UV resistance.