<strong>Abstract</strong> Assessing the impact of global changes and protection effectiveness is a key step in monitoring marine fishes. Most traditional census methods are demanding or destructive. Nondisturbing and nonlethal approaches based on video and environmental DNA are alternatives to underwater visual census or fishing. However, their ability to detect multiple biodiversity factors beyond traditional taxonomic diversity is still unknown. For bony fishes and elasmobranchs, we compared the performance of eDNA metabarcoding and long-term remote video to assess species’ phylogenetic and functional diversity. We used 10 eDNA samples from 30 L of water each and 25 hours of underwater videos over 4 days on Malpelo Island (pacific coast of Colombia), a remote marine protected area. Metabarcoding of eDNA detected 66% more molecular operational taxonomic units (MOTUs) than video. We found 66 and 43 functional entities with a single eDNA marker and videos, respectively, and higher functional richness for eDNA than videos. Despite gaps in genetic reference databases, eDNA also detected a higher fish phylogenetic diversity than videos; accumulation curves showed how 1 eDNA transect detected as much phylogenetic diversity as 25 hours of video. Environmental DNA metabarcoding can be used to affordably, efficiently, and accurately census biodiversity factors in marine systems. Although taxonomic assignments are still limited by species coverage in genetic reference databases, use of MOTUs highlights the potential of eDNA metabarcoding once reference databases have expanded. <strong>Methods</strong> We sampled around the Sanctuary of Fauna and Flora in Malpelo, a remote oceanic island 490 km off the Colombia in the eastern tropical Pacific (Fig. 1), for 4 days (25-28 March 2018) at 1 site (El Arrecife). Malpelo is surrounded by deep water and fishing activities are prohibited in the surrounding 8,757 km2 (Edgar et al. 2011). The reef ecosystem around the island is influenced by major oceanic currents (Rodríguez-Rubio et al. 2003) and local upwelling, and the benthos is bare rock with low coral cover (Quimbayo et al. 2017). Malpelo Island is one of the most pristine and vulnerable reef ecosystems in the tropical eastern Pacific. Fish biomass and biodiversity are high (&gt;250 vertebrates species) and provide a baseline for undisturbed assemblages in this marine province (Quimbayo et al. 2017). We deployed 1 long-duration remote underwater video system (RUV) (Extrem-Vision, Rivesaltes, France) that films up to 12 hours (screenshots in Appendix S1). The camera was 40 cm above the seafloor (13 m deep, 04.00600°, -81.60433°) and had a 90° field of view in which benthic and pelagic areas were recorded over 10 m2. Resolution was 1920 x 1080 pixels, and 30 frames/second were shot. Recording occurred on 25 (day and night) and 28 March (day) (Fig 1c). Cameras filmed 24 hours and 50 minutes of video. At night 2 dive lights illuminated the camera’s view. A Hero 5 (GoPro, San Mateo, California) was mounted on top the RUVs to film in the opposite direction for the first 2 hours of deployment of each daylight recording. Three hours and 30 minutes were recorded with the GoPros. During video recordings, we sampled eDNA above the camera in round surface transects. We did 5 identical transects at different times, corresponding to 10 samples (i.e., eDNA filters) (Fig. 1) because we collected 2 samples/transect. Transects sampled from a boat, and we pumped 30 L of water/sampled. We used an Athena peristaltic pump (Proactive Environmental Products, Bradenton, Florida) (nominal flow 1.0 L/minute) on each side of the boat to filter water through a VigiDNA 0.20 μm cross-flow filtration capsule (SPYGEN, le Bourget du Lac, France). To avoid contamination, we used only disposable sterile tubing and gloves for each filtration capsule. Immediately after filtration, the filter units were filled with CL1 Conservation buffer (SPYGEN, le Bourget du Lac, France) and stored at room temperature (20-25° C) for 5.5 months until DNA extraction. For video processing, two frames/second were extracted from all videos. Fishes were identified at the lowest taxonomic level possible, following Fishbase taxonomy (Froese &amp; Pauly 2000), by trained personnel, who recorded the first occurrence of each species in each of the videos (i.e., number of individuals per species was not recorded). <strong>Usage Notes</strong> * eDNA Pre and post-processed eDNA metabarcoding data for the three markers used in the study (teleo, chon, vert01) alongside metadata. * Videos Recording of each fish species' first occurence. Please see the README.md file for more information on the files.