Deep-sea benthic ecosystems remain poorly studied in South Africa, limiting understanding of community biodiversity patterns and their environmental drivers. This is one of the first studies to (i) visually investigate marine epifaunal community patterns and their environmental drivers along the Agulhas ecoregion outer shelf, shelf edge and upper slope to support marine ecosystem classification and mapping, and (ii) to determine the conservation targets for selected national marine ecosystem types to inform improved management of the marine environment, through Marine Spatial Planning processes. Visual surveys of the seabed were conducted to quantify epifauna during the ACEP Deep Secrets Cruise in 2016, using a towed benthic camera system. Twenty-nine sites were sampled, ranging from 120-700 m in depth and spanning the shelf-slope transition from the western edge of the Agulhas Bank to offshore of the Kei River mouth. A total of 855 seabed images were processed, and 173 benthic taxa quantified. Corresponding environmental variables were used to determine potential drivers of observed biodiversity patterns. Data were analysed using multivariate analyses, including CLUSTER, MDS and DistLM, in PRIMER v6 with PERMANOVA. Ten different epifaunal communities were classified and described with key characteristic taxa identified. Communities found in habitats that comprised mostly hard rocky substrata generally exhibited higher in species richness and were most commonly characterized by stalked crinoids, various corals and bryozoans, whereas communities found in habitats comprising unconsolidated sediment were lower in species richness and commonly characterized by polychaetes, cerianthids and brittle stars. Communities found in habitats comprising both hard and soft substrata had a mix of the above-mentioned epifauna. The distribution of these communities was mostly influenced by substratum type, longitude, trawling intensity, depth, and presence of visible particulate organic matter. The combined interactions of topography, substratum and the unique hydrodynamic conditions along the Agulhas ecoregion shelf-slope transition are likely responsible for the observed patterns. The observed community patterns were also compared to the existing classification of marine ecosystem types from the 2018 National Biodiversity Assessment. Fine-scale heterogeneity was revealed within the examined marine ecosystem types, particularly with substratum type and associated community variability and should be recognized and incorporated into future iterations of the national marine ecosystem classification and map. Species-area curves were used to calculate conservation targets for three ecosystem types, defined by the 2018 National Biodiversity Assessment, namely the Agulhas Coarse Sediment Shelf Edge, South West Indian Upper Slope, and the Agulhas Rocky Shelf Edge. Considering the epifaunal species richness (using the bootstrap estimator) and area, per image and per ecosystem type, the rate of accumulation of species was calculated and used to estimate the percentage of species expected to be represented by any given percentage of protected ecosystem type area. Between 20 and 30% of the area within these ecosystem types will need to be protected to represent 80% of the species. This study has shown that an integration of environmental parameters together with biodiversity measures to better understand and classify offshore benthic ecosystems has worked well. However, to improve the resolution of the national marine ecosystem classification and map, there needs to be greater input of fine-scale biological and environmental sampling and mapping of substratum types across the Agulhas ecoregion shelf-slope transition zone. This work is contributing to improvements in the national marine ecosystem classification and map and hence the spatial assessment and planning processes that rely on these products.