Marine Protected Areas (MPAs) are increasingly used for biodiversity conservation and the management of sustainable fisheries. The use of taxonomic surrogates in routine monitoring of the reserve effects on fish assemblages may represent a promising method due to its substantial technical and economic benefits. However, higher taxonomic ranks should be used as surrogates with caution, especially for fish, where the approach is still virtually undocumented. This study aims to shed light on relationships between taxonomic relatedness and ecological similarity, which is crucial to assess the relevance of species surrogacy to reflect species-level information and detect changes in fish related to protection regimes. By analyzing data from a Mediterranean MPA, we show that rather being related to taxonomic relationships, the ability of higher taxa to reflect species-level patterns was explained in terms of aggregation level and distribution of species within taxa. Null models using random aggregations of species were created to identify the best surrogates able to depict changes in responses of assemblages to protection observed at the species-level. Comparison of null model predictions with the more classical higher-taxon approach revealed that the latter was not reliable because, unlike null model outcomes, surrogates determined empirically were not relevant for other subsequent independent monitoring. The effectiveness of species surrogates to depict changes in responses of assemblages to protection observed at the species-level depended (1) on the numerical resolution of the aggregation and (2) the magnitude of differences between protected and unprotected locations. Such findings suggest that surrogacy approaches may be applied for routine monitoring of Mediterranean reef fish communities, although mere empirical determinations of sufficient taxonomy seems to be not reliable, legitimizing the use of alternative methods based on null models. Guidelines for the careful use of species surrogacy in the ecological evaluation of MPAs on fish assemblages are provided