The spread of an ensemble of weather predictions initialized from an ensemble Kalman filter may grow slowly relative to other methods for initializing ensemble predictions, degrading the skill of probabilistic predictions. Several possible causes of the slow spread growth were evaluated in perfect- and imperfect-model experiments with a 2-layer primitive equation spectral model of the atmosphere. The causes examined were covariance localization, the additive noise used to stabilize the model and parameterize the system error, and model error itself. For this experiment, the flow-independent additive noise was the biggest factor in constraining spread growth. Pre-evolving additive noise perturbations was tested as a way to make the additive noise more flow dependent. This improved the data assimilation and ensemble predictions both in the 2-layer model results and in a brief test of the assimilation of real observations into a global spectral primitive equation model.