Résumé
Combining different components of quantitative resistance in new banana hybrids has potential for developing efficient and durable control of black leaf streak disease (BLSD) caused by the fungus Pseudocercospora fijiensis. This study aimed to identify such components in six banana progenitors used in genetic improvement programmes and five elite hybrids from crosses between some of these progenitors. Parameters related to several resistance components were estimated from natural infections in the field and under controlled conditions, using classic and new methods of phenotyping based on deep-learning image analysis. Field assessments highlighted two parameters that may reflect two groups of resistance components. The first one notably linked the percentage of leaves without necrosis (INSL) with resistance to asexual sporulation. The second one linked disease severity to resistance to sexual sporulation. In a complementary approach, other parameters, including the infection efficiency, were evaluated from inoculation under controlled conditions with several P. fijiensis isolates. Genotype-by-genotype (G × G) interactions between P. fijiensis and banana genotypes were detected from these inoculations. Some banana genotypes exhibited broad resistance across multiple components, whereas specific genotypes excelled in one component, such as resistance to asexual sporulation. This study emphasises the need for multitrait resistance assessments in banana breeding, revealing gaps in current methodologies and advocating the integration of advanced phenotyping tools for high-throughput resistance assessments. The findings provide valuable data for parameterising epidemiological models to identify which combination of resistance components could be the most efficient and durable to control BLSD.