Response to Referee #1

We thank Referee #1 for insightful and constructive comments that help improve the manuscript.

Referee has several Specific Comments, which are well justified, and which we will address as follows.

1. We will add more discussion about previous studies on the flocculation dynamics of organic carbon and metals in estuaries in Introduction.
2. We will add a mention of the bucket material used. Metal flocculation was studied in the flocculator experiment that used glass beakers, so the bucket material is not important for metal adsorption here.
3. We will clarify that ICP-OES was used for Al and Fe, and ICP-MS was used for Co, Cu and Mn, for the highest analytical precision.
4. Referee #1 is right that the section is only one sentence. However, we find it is good, for the sake of clarity, that the initial compositions of the river waters are considered separately from the experiment results. Because this is a structural matter, we expect Editor to provide us with guidance if necessary.
5. Unfortunately, high-frequency optical sensors for determining DOC concentration are not available. Instead, sensors for measuring DOM fluorescence are commercially available and are time-tested technology. DOM fluorescence has been used as a proxy for DOC concentration widely for the past decades, and we will provide relevant references making this point in case.
6. Referee #1 is right that the first paragraph of Discussion briefly restates the study aim and methodological approach. The purpose of the paragraph is to help the busy reader who may wish to jump to Discussion section and, therefore, we would like to keep it as it does not take up much space. Because this is a structural matter, we expect Editor to provide us with guidance if necessary.
7. Climate change is likely to increase the acidic runoff and metal loading from AS soils to the Gulf of Bothnia as shown by the previous studies. We expect this effect to influence Co, Cu and Mn, which remained largely in the dissolved phase in our experiments and may, therefore, be transported out to the sea. Aluminum and Fe transferred strongly to the particulate phase, so they are expected to be deposited close to the river mouths, which is also confirmed by our previous study of seafloor metal distribution (Virtasalo et al., 2020). We will clarify the reasoning in the paragraph in question.

Kind regards on behalf of all co-authors,

Joonas Virtasalo, Geological Survey of Finland

Response to Referee #2

We thank Referee #2 for insightful and constructive comments that help improve the manuscript.

Referee’s comments are well justified and we will address them as follows.

Referee #2 correctly points out the difference between experiment replication and analysis replication. We will clarify this in the text.

Other comments.

Lines 87-91: We will include a hypothesis related to the differences between the studied rivers.

Lines 134-151: We will add the missing details about the bucket experiment: the initial volume of river water, the rate of seawater addition, and the total duration of the experiment.

Line 335: We will state the relevant 0-2 salinity range.

Lines 391-402: We have other optical parameters (such as CDOM absorbance at 440 nm and protein-like DOM fluorescence), that show clearly how humic and aromatic compounds are removed most efficiently from the DOM pool. We did not include this data in the original submission, but we will add new figures to supplementary material and elaborate this in the text.

Line 430: We agree. We will clarify the experiment duration in Methods.

Line 436: Referee probably means line 456. The determination of the local background level is described in the paper cited in that sentence. We will add a mention of the approach used in the sentence.

Figures: We will add panel letters to all the figures and use them when referring to the figures.

Figure 4: We thank Referee for pointing this out. The results in the figure indeed are from the flocculator experiment. We will correct this.

Figure 9: We had a blank subtraction error regarding manganese, resulting in values slightly below zero for two samples with very low concentration. This will be corrected, and does not affect the overall results or interpretation. It is correct that when mean values are close to zero, large standard deviation will cause error bars to indicate negative values. This is of course not true, as reviewer correctly points out. This is however a plotting issue, and we will update the figure by removing error bars altogether and showing each observation as points instead of means and error bars.

Kind regards on behalf of all co-authors,

Joonas Virtasalo, Geological Survey of Finland