The establishment of protected areas to face global diversity declines has mainly prioritized taxonomic diversity, leaving aside phylogenetic and functional diversities, which determine ecosystem functioning and resilience. Furthermore, the assessment of protected areas' effectiveness is mainly done using short-duration surveys (\textless2 h), which may undermine the detection of rare species. Through a long-duration video approach, reef fish taxonomic, phylogenetic and functional facets of diversity were assessed for 3 days within a fully protected area and a nearby poorly protected area in Mayotte Island (Western Indian Ocean). We found that temporally rare species contributed to more than 60% of the taxonomic facet and 85% of the functional facet of biodiversity found on each site. Those rare species, which harbour the most distinct trait values, also make reef fish diversity particularly vulnerable to their loss. Taxonomic, phylogenetic and functional richness were similar between the fully protected area and the poorly protected area, while the species, lineage and trait compositions were markedly different. These results pinpoint the importance of considering taxonomic, functional and phylogenetic dissimilarities while assessing protected areas' effectiveness, instead of using only richness. In addition, benefits of the fully protected area were detected only using more than 15 h of video survey, which emphasizes the importance of long-duration remote approaches to capture the within- and between-day temporal variations.