Microscopic-scale processes significantly influence benthic marine ecosystems such as coral reefs and kelp forests. Due to the ocean's complex and dynamic nature, it is most informative to study these processes in the natural environment yet it is inherently difficult. Here we present a system capable of non-invasively imaging seafloor environments and organisms in situ at nearly micrometre resolution. We overcome the challenges of underwater microscopy through the use of a long working distance microscopic objective, an electrically tunable lens and focused reflectance illumination. The diver-deployed instrument permits studies of both spatial and temporal processes such as the algal colonization and overgrowth of bleaching corals, as well as coral polyp behaviour and interspecific competition. By enabling in situ observations at previously unattainable scales, this instrument can provide important new insights into micro-scale processes in benthic ecosystems that shape observed patterns at much larger scales.

Thwaites Glacier is one of the fastest-changing ice-ocean systems in Antarctica^1-3. Much of the ice sheet within the catchment of Thwaites Glacier is grounded below sea level on bedrock that deepens inland⁴, making it susceptible to rapid and irreversible ice loss that could raise the global sea level by more than half a metre^2,3,5. The rate and extent of ice loss, and whether it proceeds irreversibly, are set by the ocean conditions and basal melting within the grounding-zone region where Thwaites Glacier first goes afloat^3,6, both of which are largely unknown. Here we show-using observations from a hot-water-drilled access hole-that the grounding zone of Thwaites Eastern Ice Shelf (TEIS) is characterized by a warm and highly stable water column with temperatures substantially higher than the in situ freezing point. Despite these warm conditions, low current speeds and strong density stratification in the ice-ocean boundary layer actively restrict the vertical mixing of heat towards the ice base^7,8, resulting in strongly suppressed basal melting. Our results demonstrate that the canonical model of ice-shelf basal melting used to generate sea-level projections cannot reproduce observed melt rates beneath this critically important glacier, and that rapid and possibly unstable grounding-line retreat may be associated with relatively modest basal melt rates.