Denisse Fierro Arcos , Stuart Corney, Amelie Meyer, Hakase Hayashida, Andrew E. Kiss, Petra Heil

Southern Ocean organisms are uniquely adapted to the extreme environmental conditions that characterise this region, making them especially vulnerable to climate change. Alterations to the physical environment have already been linked to alterations in the structure and functioning of entire ecosystems, and ecological disruptions are expected to continue to occur. Although our understanding of the physical processes driving ecological change in the Southern Ocean has improved in recent years, signi?ficant knowledge gaps remain largely as a result of insufficient observational data being available. High resolution ocean models are an important tool that can help us overcome data scarcity. However, models generally contain biases that may affect their ability to accurately represent environmental conditions in the region of interest. Thus, their outputs must be evaluated before they can be used to answer questions about ecological impacts. Here, we examined the suitability of ACCESS-OM2-01, a high-resolution coupled ocean-sea ice model, for ecological applications. We provide a template for testing the suitability of model outputs for ecological applications, as well as quantitative estimates of changes in key environmental variables for the Southern Ocean over the recent past. Our results highlight the heterogeneous nature of the mean state of the environmental variables examined and their trends across the Southern Ocean. Our assessment shows that the ACCESS-OM2-01 model performance differs across variables, but overall, it does a reasonable job in reproducing the observed seasonal cycle and broad baseline climatological conditions of the mixed layer depth and sea ice variables for the Southern Ocean over the half century. Model performance also varies across space and time, which reflect gaps in our understanding of how different atmospheric and oceanographic mechanisms interact to drive change in the variables examined. These results emphasise the importance of understanding the capabilities and shortcomings of models within the boundaries of the area of interest prior to using model outputs in ecological applications.