Reserve selection algorithms (RSAs) are a powerful tool for marine conservation, but need biological realistic constraints. A common constraint is that the RSA aims to include the same percentage area of each habitat in the reserve network. The example of Caribbean reefs is used to argue that there is significant additional scope to set specific areal targets for each habitat type, as in many terrestrial case studies. Ten Caribbean habitats are weighted by data on their biodiversity, functional value to ecosystem processes and ecosystem services. These weightings vary depending on the criteria used, and are particularly distinct for ecosystem services. Furthermore, a percentage goal for a habitat can be achieved in multiple ways by including different combinations of habitat patches from across a seascape. RSAs may treat these combinations equally, but some habitat patches are more ‘valuable’ than others because, for example, high abundances of grazers make the benthic community resilient. Incorporating some of these considerations into RSAs is relatively easy, while others are more difficult. However, establishing how to integrate patterns of biodiversity, and ecological processes, functions, goods and services across seascapes into RSAs offers opportunities to improve the design of networks of reserves.