Established in the Fram Strait in 1999, the LTER (Long-Term Ecological Research) observatory HAUSGARTEN enables us to study ecological changes on the deep Arctic seafloor. Repeated deployments of a towed camera system (Ocean Floor Observation System) along the same tracks allowed us to build a time series longer than a decade (2004–2015). Here, we present the first time-series results from a northern and the southernmost station of the observatory (N3 and S3, ~2650 m and 2350 m depth respectively) obtained via the analysis of still imagery. We assess temporal variability in community structure, megafaunal densities and diversity, and use a range of biotic factors, environmental sediment parameters and habitat features to explain the patterns observed. There were significant temporal differences in megafaunal abundances, diversity and habitat features at both stations. A particularly high increase in megafaunal abundance was recorded at N3 from 12.08 (±0.39; 2004) individuals m−2 to 35.21 (±0.97; 2007) ind. m−2 alongside a ten-fold increase in (drop-)stones. At S3, megafaunal densities peaked in 2015 (22.74±0.61 ind. m−2) following a general increase since 2004 (12.44±0.32 ind. m−2). Sea cucumbers showed particularly striking temporal differences: densities of the small holothurian Elpidia heckeri rose ten-fold from 0.31 ind. m−2 (±0.04; 2004) to 3.74 ind. m−2 (±0.14; 2015) at S3, and 24-fold from 0.09 ind. m−2 (±0.02; 2004) to 2.20 ind. m−2 (±0.10; 2015). Initially entirely absent from N3, densities of the larger holothurian Kolga hyalina peaked in 2007 (5.87±0.22 ind. m−2) and declined continuously since then. Overall diversity (γ) increased at both stations over the course of the study, however, with varying contributions of α and β diversities. Our results highlight the importance of time-series studies as megafaunal community structure is characterised by continuous changes. This indicates that epibenthic communities from the deep seafloor are reactive and dynamic, with no consistent community state. To continue to monitor them is therefore crucial in understanding natural and anthropogenic impacts in an area exposed to the effects of climate change.