Aim: Mesophotic ecosystems, found at the limit of light penetration in the ocean, are rich in biodiversity and harbor unique ecological communities. However, they remain among the least studied habitat zones on earth due to high costs and technological limitations. Here, we characterize mesophotic communities in two marine reserves across a range of habitat types, depths, and temperatures using submersible technologies, with the goal of understanding the processes that structure these communities across biogeographic regions. Location: The Bay of La Paz and the Revillagigedo Archipelago, Mexico. Taxa: Fish and algal species. Methods: We used a small and inexpensive Remotely Operated Vehicle (ROV) to conduct roving-swim surveys of major habitat types in depths from 20–100 m. With the resulting binary data of fish species presence, we used generalized linear mixed models and canonical correspondence analysis and to determine whether biogenic habitat, depth, and/or temperature best explained species richness and community structure across reef and non-reef substrate. Results: We identified 81 species, including new depth records for nine fish species and a new geographic record for one fish species. Our surveys included large undocumented rhodolith beds (free-living coralline algae) and mesophotic algal communities, in addition to diverse communities of soft corals and sponges. Fish species richness was positively associated with rocky substrate and warmer water, and reef fish communities differed significantly by depth, temperature, and biogenic habitat. Main conclusion: Our results highlight the importance of biogenic habitat in structuring communities across gradients of depth and temperature. We also demonstrate the effectiveness of a small and economical ROV for conducting mesophotic surveys in remote regions. Our methods and results provide a framework that can be used to greatly increase the biogeographic and taxonomic scope of mesophotic research, especially for readily identifiable taxa such as fish.

All surveys were conducted using a small (roughly 1 m³ and 11 kg) and highly portable ROV (BlueRobotics BlueROV2) capable of diving to 100 m and equipped with four 1500 lumen lights, two cameras that filmed in 1080 p resolution (one 8MP Raspberry Pi v2 Camera and one GoPro HERO3 or HERO4), a Measurement Specialties MS5837-30BA sensor that measured depth and temperature, and a 250 m tether (see Appendix S1 in Supporting Information for image of ROV). At each site, the ROV operator would descend the ROV directly to the benthos and begin roving swim surveys. The duration of surveys ranged from 7 to 65 minutes, depending on weather and ocean conditions, battery duration, and cruise schedules. Video, depth, and temperature were continuously recorded and saved for later analysis. Footage from the GoPro and Raspberry Pi cameras were synchronized for each dive and cut into 5-minute transects, starting at the time the ROV reached the benthos. In each video, all fish were identified to the lowest possible taxonomic level using Robertson and Allen (2015). Biotic and abiotic habitat characteristics of each transect were also noted. Biogenic habitat represented the dominant foundation species group (sensuDayton 1972) that contributed to three-dimensional habitat, including coral, unidentified filamentous algae, rhodolith beds (also termed maërl, comprised of colonies of slow-growing, free-living coralline algae [Foster 2001]), turf (broadly encompassing encrusting organisms such as algae, bryozoans, hydroids, and crustose coralline algae), and burrows in sand (see Appendix S2 for representative images of habitat classifications). Substrate classifications included reef (continuous hard substrate or large semi-continuous boulders), and non-reef (small patches of unconsolidated hard substrate over sand or sand). Depth (m) and temperature (°C) for each transect were averaged and rounded to the nearest integer.

This dataset only records the presence of a given species in a transect, not the abundance.