Forecasting assemblage-level responses to climate change remains one of the greatest challenges in global ecology [1, 2]. Data from the marine realm are limited because they largely come from experiments using limited numbers of species , mesocosms whose interior conditions are unnatural , and long-term correlation studies based on historical collections . We describe the first ever experiment to warm benthic assemblages to ecologically relevant levels in situ. Heated settlement panels were used to create three test conditions: ambient and 1°C and 2°C above ambient (predicted in the next 50 and 100 years, respectively ). We observed massive impacts on a marine assemblage, with near doubling of growth rates of Antarctic seabed life. Growth increases far exceed those expected from biological temperature relationships established more than 100 years ago by Arrhenius. These increases in growth resulted in a single “r-strategist” pioneer species (the bryozoan Fenestrulina rugula) dominating seabed spatial cover and drove a reduction in overall diversity and evenness. In contrast, a 2°C rise produced divergent responses across species growth, resulting in higher variability in the assemblage. These data extend our ability to expand, integrate, and apply our knowledge of the impact of temperature on biological processes to predict organism, species, and ecosystem level ecological responses to regional warming.