Traditional richness- and abundance-based indices primarily quantify alpha diversity but do not directly represent spatial compositional distinctiveness. We developed a novel Combination Index (CI) that quantifies the rarity of species co-occurrence patterns using presence–absence data, without requiring abundance information, and is less directly dependent on species richness than conventional richness-based metrics. Using records of 46 native fish species from 818 sites across 12 Ecological Regions (ERs) in Japan, we examined relationships among the CI, Shannon–Weaver diversity, Local Contribution to Beta Diversity (LCBD), and environmental and boundary-related variables. Pearson's correlation detected no significant relationship between the CI and Shannon–Weaver diversity across all sites, whereas the Maximal Information Coefficient detected a significant association. The CI was positively correlated with LCBD across all sites, although regional relationships varied. Both the CI and LCBD were significantly higher in estuarine sections despite lower mean species richness, whereas Shannon–Weaver diversity was lower. The CI also decreased significantly with distance from riverbed gradient change points across all sites and within six ERs. Random Forest and generalized additive model analyses supported the importance of these gradient changes for the CI, while local habitat variables related to flow velocity, water depth, and substrate conditions were more frequently associated with Shannon–Weaver diversity and LCBD. These findings suggest that the CI complements alpha- and beta-diversity measures by identifying distinctive assemblages associated with certain riverine transitions, although its relationships with boundary-related factors vary geographically.
