Reviewer #1 asked about the most significant parameter for improving aerosol distributions (Specific Comment #1). You respond by computing diversity, which is a normalized standard deviation. However, I don't think this quite addresses the comment. For example, for sulphate, its gaseous chemical production was highlighted as having the highest diversity (77%) among the 4 models that include this process. However, the standard deviation in absolute units if 0.26 Tg/yr. Compared to, say, the std dev for SO4 emissions (0.84 Tg/yr) or wet deposition (1.21 Tg/yr), it's clear that even if gas phase SO4 production was perfectly represented, its effect on SO4 concentrations is not substantial enough to matter very much. Thus, diversity is not a very good metric for what models must improve to more accurately represent aerosol. I suggest that you use metrics that directly addresses the question.

To this end, Table 1 seems to show that SO4 and dust are the main contributors to AOD, so whatever model parameters that are important for determing the concentrations of these aerosol types and exhibit the largest inter-model variability seem to be the ones that should be identified as most critical. It does not appear that the representation of OA or BC will matter nearly as much because they contribute little to AOD.

Response to Reviewer #1, Specific Comment #3: These are rather large discrepancies in the imaginary components of the various aerosol components. What are the consequences for modeled AOD, and hence for the model-obs comparisons of AOD? It seems like some discussion of this is warranted in the text.