Genomic Basis of Host Specialization in Cordyceps: a Mosaic Selection Pattern primarily affects Effectors and Secondary Metabolite Clusters

The entomopathogenic fungal genus Cordyceps is a key model for understanding insect host specialization and ecological adaptability. Nevertheless, the genomic and adaptive mechanisms driving the wide host range of the pathogen are poorly defined. The present study conducted a comprehensive comparative genomic analysis of 12 species within Cordyceps, providing high quality genome assemblies and a robust phylogenomic framework based on 5,123 single-copy orthologs. Analyses revealed a dynamic secondary metabolite landscape, where conserved core biosynthetic gene clusters (BGCs) contrast sharply with heterogeneous, species-specific BGCs. While a core enzymatic toolkit was conserved, effector repertoires exhibited substantial lineage-specific diversification with only few core effectors widely conserved. Importantly, genome-wide analyses revealed that adaptive evolution was widespread but mosaic, specifically targeting functional hotspots that mediate host-pathogen interactions. Apart from CAZymes such as chitinases, penetration-associated effectors, interestingly only single catalytic steps in BGCs were affected, indicating modular fine-tuning of specific moieties after core structures evolve rather than complete pathway rewiring. The current study revealed a multi-layered adaptive mechanism where a conserved core is functionally refined via focused, intermittent positive selection on effectors and BGCs. This is not only a convergent evolutionary pattern with diverse plant and human pathogens, but the modular selection on BGC is surprisingly similar to the human strategy to modify moieties of core structures in antibiotics optimisation, rather than trying out entirely new structures. Our findings thus offer a pivotal genomic framework for dissecting the molecular basis of host range expansion and exploiting the vast metabolite diversity of Cordyceps for therapeutic and ecological applications.