Coastal ecosystems are subject to intense anthropogenic pressures, requiring robust and efficient monitoring tools to assess fish assemblages and inform conservation strategies. This study evaluated and compared the performance of Baited Remote Underwater Video Systems (BRUVS) and a recently developed Mapping-Tracked Roaming Transect visual census (M-UVC) in a mixed-bottom, multiple-use Mediterranean coastal area in southeastern Spain. Surveys were conducted across sandy bottoms, rocky reefs, and Posidonia oceanica meadows. A total of 28 taxa were recorded across nine paired deployments. M-UVC detected 27 taxa, significantly outperforming BRUVS (17 taxa) in terms of species richness and frequency of occurrence. Rarefaction analyses indicated that BRUVS consistently underestimated total richness, particularly for reef-associated species. Assemblage composition differed markedly between methods: M-UVC was dominated by reef and pelagic species, whereas BRUVS primarily recorded soft-bottom taxa. Importantly, BRUVS proved more effective for detecting certain threatened elasmobranchs, as the critically endangered common guitarfish (Rhinobatos rhinobatos) and the endangered spiny butterfly ray (Gymnura altavela), likely reflecting avoidance behavior toward divers. Spatially explicit data from M-UVC revealed strong habitat associations. Overall, M-UVC provided broader assemblage characterization and high-resolution spatial information, while BRUVS complemented these results by improving detection of shy, soft-bottom–associated chondrichthyans. Our findings highlight the importance of method selection according to target taxa and habitat type and support the complementary use of visual census and baited video techniques for monitoring.