The accurate representation of cloud droplet number concentration (Nd) is crucial for predicting future climate. However, models often underestimate Nd over the Southern Ocean (SO), where natural sources dominate, and aerosols are composed primarily of marine biogenic sulfate and sea spray. This study uses a range of diverse datasets to evaluate and untangle biases in E3SMv2 simulated clouds, aerosols, and sulfur species. The default E3SMv2 underestimates Nd over SO by a factor of 2 when compared with observations in 3km-resolution simulations. Updating the dimethyl sulfide (DMS) emission and chemistry leads to a better agreement between the model and the observations in Nd and boundary layer aerosols, but low biases persist in the free tropospheric aerosol concentrations larger than 70 nm, possibly attributable to insufficient particle growth. Preliminary evaluation also reveals biases in simulated sulfur species, including overestimation in DMS at high latitudes, and in simulated sulfate mass concentration, highlighting the necessity for further efforts to improve the model treatment of relevant processes.