The latest review suggests that the authors have not answered all the earlier questions posed. I still have some questions of my own, so I agree with the reviewer that more work needs to be done on this paper before it can be accepted.

The authors have attempted to address at least some of the concerns of the reviewers by adding a section based on the SODA reanalysis, which also shows a decrease in salinity along the 30°S line in the South Atlantic and suggests that there has been a general decrease in the flux of water from the Agulhas since about 2004 (although the Indian-Atlantic flux anomaly was still positive during the period from 2004-2011). Thus there is a general qualitative agreement between the IPRC Argo data, the WOCE data, and the model reanalysis, but since the reanalysis is presumably based at least partly on these data it would be surprising if they did not agree. Wind data also appear to confirm the suggestion that the westerly wind strength over the Indian Ocean has decreased since about 2004.

A major question that has not been answered is whether the observed changes are caused only by changes in the Agulhas leakage. Only a few Agulhas rings cross latitude 30° S west of the 0° meridian (Duncombe Rae, 1991; Biastoch et al 2008), and yet the salinity changes in the AAIW layer are apparently seen all the way across the South Atlantic. This suggests that there should be some contribution to the freshwater flux from another source, possibly the SE Pacific, as pointed out on p.2 of the manuscript, or maybe from the Indian Ocean south of the Agulhas retroflection. How well do the estimates of Agulhas leakage from the SODA reanalysis account for the change in salinity seen in the data? Are there other data that could help determine how much of the observed changes comes from each source?

Finally, the P-E data in Fig. 4 suggest that there may be a cycle in the salinity balance with a periodicity of 30-40 years that would presumably affect the salinity along 30°S. Yet the authors do not comment on this possibility. Are there any long-term modeling studies that could support this argument, which would be additional support for their ideas?

BIastoch, A, CW Bonning and JRE Lutjeharms (2008). Agulhas leakage dynamics affects decadal variability in Atlantic overturning circulation. Nature, 456, 489-492, DOI:10.1038/nature07426
Duncombe Rae, CM (1991) Agulhas retroflection rings in the South Atlantic Ocean: an overview, South African Journal of Marine Science, 11:1, 327-344, DOI:10.2989/025776191784287574