In a context of scarce financial and human resources, the allocation of conservation efforts needs to be optimized. Our analysis attempts to draw conclusions on the integration of regional and local conservation assessments, specifically, with regard to the acquisition of fine-scale data to complement the regional assessment. This study undertaken in Réunion Island (Indian Ocean) assessed how biodiversity surrogates targeted at a regional scale represented other biodiversity surrogates at a local scale. Biodiversity surrogates at both scales consisted of species, habitats and processes. Habitats and processes at regional scale were defined using a coarser scale of thematic resolution than at local scale. The surrogacy was tested in terms of incidental representation of local-scale features in the regional assessments, and correlation of irreplaceability values between scales. Near-minimum sets and irreplaceability values were generated using MARXAN software. Our results revealed that conservation targets for processes at local scale were never met incidentally, while threatened species and fragmented habitats were also usually under-represented. More specifically, requiring only 12% of the local planning domain, the application of species as surrogates at regional scale was the least effective option at representing biodiversity features at local scale. In contrast, habitats at a coarse scale of thematic resolution achieved a significant proportion of conservation targets incidentally (67%) and their irreplaceability values were well correlated with the irreplaceability values of surrogates at local scale. The results highlighted that all three types of biodiversity surrogates are complementary for assessing overall biodiversity. Because of the cost of data acquisition, we recommended that the most efficient strategy to develop nested regional/local conservation plans is to apply habitats and processes at a coarse scale of thematic resolution at regional scale, and threatened species and degraded habitats at local scale, with their fine-scale mapping limited to highly transformed areas.