Despite more than 60 yr of coral reef research using scuba diving, mesophotic coral ecosystems (MCEs) between 30 and 150 m depth remain largely unknown. This study represents the first underwater visual census of reef fish communities in the Greater Caribbean on MCEs at depths up to 80 m in Bermuda and 130 m in Curaçao. Sampling was performed using mixed-gas closed-circuit rebreathers. Quantitative data on reef fish communities were obtained for four habitats: coral reefs (45-80 m), rhodolith beds (45-80 m), ledges (85-130 m) and walls (85-130 m). A total of 38 species were recorded in Bermuda and 66 in Curaçao. Mesophotic reef fish communities varied significantly between the two localities. MCEs in Bermuda had lower richness and abundance, but higher biomass than those in Curaçao. Richness, abundance and biomass increased with depth in Bermuda, but decreased in Curaçao. A high turnover of species was found among depth strata and between Bermuda and other Caribbean upper MCEs (45-80 m), indicating that depth was an important driver of community structure at all localities. However, local and evolutionary factors (habitat and endemism) are likely the main factors shaping communities in isolated locations such as Bermuda. High fishing pressure is evident in both localities, as total biomass of apex predators was generally low, and thus may be driving a ''refugia'' scenario in Bermuda, as the abundance and biomass of macro-carnivores increased with depth and distance from the coast.

The detrimental effects of invasive lionfishes (Pterois volitans and Pterois miles) on western Atlantic shallow reefs are well documented, including declines in coral cover and native fish populations, with disproportionate predation on critically endangered reef fish in some locations. Yet despite individuals reaching depths >100 m, the role of mesophotic coral ecosystems (MCEs; reefs 30-150 m) in lionfish ecology has not been addressed. With lionfish control programs in most invaded locations limited to 30 m by diving restrictions, understanding the role of MCEs in lionfish distributions remains a critical knowledge gap potentially hindering conservation management.Here we synthesise unpublished and previously published studies of lionfish abundance and body length at paired shallow reef (0-30 m) and MCE sites in 63 locations in seven western Atlantic countries and eight sites in three Indo-Pacific countries where lionfish are native. Lionfish were found at similar abundances across the depth gradient from shallow to adjacent MCEs, with no difference between invaded and native sites. Of the five invaded countries where length data were available three had larger lionfish on mesophotic than shallow reefs, one showed no significant difference, while the fifth represented a recently invaded site. This suggests at least some mesophotic populations may represent extensions of natural ontogenetic migrations. Interestingly, despite their shallow focus, in many cases culling programs did not appear to alter abundance between depths. In general, we identify widespread invasive lionfish populations on MCE that could be responsible for maintaining high densities of lionfish recruits despite local shallow-biased control programs. This study highlights the need for management plans to incorporate lionfish populations below the depth limit of recreational diving in order to address all aspects of the local population and maximise the effectiveness of control efforts.