Summary:

This manuscript uses statistical tests to show the historical trend of the streamflow in Australian, which is intriguing. The results would be useful for the future study to learn how climate change, evapotranspiration, rainfall pattern or other factors affect the streamflow pattern.

Major Comments:

1. For several statistical tests, the 0.1 of P value were used. Nowadays, many research use p value of 0.05, could you give us more detail or explanation why 0.1 has been used.
2. Line 675, “abrupt shift in streamflows across water year”, it would be better to show how abrupt percentage is based on the region (spatial) and year (year), like in Table.
3. How does the long drought or extreme event affect the statistical test, like linear trend? E.g. after the long drought years, will it become increasing trend?
4. I recommend have a table or paragraph to show the raw data format like time interval, and explain how to deal with the raw data (take out extreme abnormal data, delete zero value). This is really important input for the following statistical tests.
5. Climate change could make extreme events more often, but it could be possible that the average yearly precipitation will not change. In this case, it might have increasing trend for the wet season, decreasing trend for the dry season and no trend for the annual basis. Do you find any station having this similar situation?

Minor Comments:

1. Line 39, please add reference to support the statement
2. Line156, 23,2846 km separator is wrong. It should be 232,846
3. Figure 1, north arrow is missing
4. Line 409, how do you convert GL/year to mm/year
5. In section 4.2, the authors mention the Figure 3 several times. Did you mean by Figure 5?
6. Line 706, where is the Figure 14?

This work presents trends in annual and seasonal mean streamflow since 1950 across an expanded Australia-wide network of reference gauges. It was found that streamflow has mostly decreased, which can be thought of as a step change in the 1990s, except for some areas in the far north that saw increasing streamflow. The authors highlight the widespread interannual persistence of streamflow anomalies in Australia, and make use of statistical trend tests that account for autocorrelation. This is a valuable contribution and in my view should be published, subject to minor revisions.

There are a few unclear sentences, such as at line 383: "The main objective is to assess whether the number of locations with significant trends occur at a regional scale or not" and 540: "It is least sensitive to outliers, and skewed distribution makes it most suitable for the analysis of streamflow data". These should be rephrased. In general some proofreading is needed.

The terminology "linear trend" is confusing at times (e.g. Section 3.1), as the MK test is for monotonic but not necessarily linear changes (including step changes).

In Section 3.3, mention that the test for regional significance is actually conservative, as it is based on a null hypothesis of independent trends across stations, when the trends within a region are actually positively correlated.

Check units -- e.g., at line 409, should it be 1.8 mm/year per year?

In Section 5.3, mention the possible role of CO₂ increase in reducing vegetation evapotranspiration rate, which could increasing streamflow in certain climatic and geomorphic settings, offsetting the increased evaporation rate due to warming -- cf. for example my 2008 HESS paper "Mapping and attribution of change in streamflow in the coterminous United States".

In Section 5.4, consider mentioning that, given the decadal persistence in streamflow regimes, it will be useful to add the available information on streamflows before 1950 to a future analysis to better seperate trends from oscillations.