|| Checking for direct PDF access through Ovid
We modelled the evolution of virulence when co-infections were permitted and compared patterns generated by our modelling approach with those of other models. In support of current models, we found that strains with virulences above the maximum basic reproductive rate (Ro) were competitively superior, and that both high host density (as indexed by host encounter rate) and high parasite infectiousness favoured evolution of strains with high virulence. However, we found that co-existence typically did not continue indefinitely and that parasites with high virulences, and Ro values approaching unity, often did not persist. We examined the extent to which processes such as stochasticity and positive frequency dependence influenced patterns generated by our model. Also, we examined commonly used indices of parasite fitness [i.e. Ro and rate of spread (ROS)], and found that only ROS was positively related to competitive ability when co-infections were permitted. However, there was considerable variation in competitive ability that was not explained by variation in ROS. We conclude that our modelling approach can significantly influence patterns generated and that conclusions from single models or conclusions based on current indices of parasite fitness should be viewed with caution. We suggest empirical tests that distinguish our model from other models and further examine the impact of mechanisms such as positive frequency dependence on the evolution of virulence.