Thanks for reading our manuscript and pointing out details that can be further improved. We have copied-in your comments in bold and italics, and our reply is following directly after each comment.

Beyond the cosmetic issues, I found the manuscript somewhat difficult to follow in that, while there is a stated research aim of identifying natural and anthropogenic nitrogen sources over the past 5000 years, the discussion started with a section of sedimentary signatures of major events in the Black Sea. The abstract also mentioned the difficulty of determining a pristine reference state for nutrient reduction, but this idea was not fully developed in the manuscript. I would recommend breaking down the overarching research aim into smaller sub-aims or sub-questions to help organize the discussion more clearly.

We agree to restructure the discussion to increase the readability.

I also miss some reference to the novelty of the study. There has already been a lot of work published on nitrogen inputs into the Black Sea and eutrophication of the Danube. Indeed, in the conclusions, the authors acknowledge that many of the results are confirming previous findings. I do think that the finding of early eutrophication of the Danube is novel but that this can be more clearly stated.

We agree, that the novelty of our study needs to be stated more clearly. and we will expand the discussion accordingly. We will state that our observations from the Danube to Black Sea transition enables to plan future sampling locations on the romanian shelf depending on whether riverine signals or basin-wide signals are to be studied. This includes to compare our results with results from e.g. Fulton et al. 2012 as a reference. We can further define which locations are unsuitable to sample.

Finally, I have some methodological concerns that I will describe in detail in the “Specific comments.”

Specific comments:

Abstract:

Line 30: “This raises the question of what point in time could serve as a realistic reference for nutrient reduction goals, given that the Danube has not had pristine nutrient levels for at least 800 years.” - This is an interesting point but is not brought up in the manuscript at all. If defining a pristine reference state is one of the study aims, I think that this should be discussed further.

We agree and will expand the discussion acordingly. The discussion will reference the European Water Frame Directive, which aims to achive „good ecological status“ for all european water bodies. The key starting point is the „high ecological status (pristine conditions with no or minimal human influence)“ against which change must be measured and eventually rectified (e.g. Moss 2008, Science of The Total Environment, https://doi.org/10.1016/j.scitotenv.2008.04.029). We will point out that human activity may have altered large river systems such as Danube since centuries and that it therefore might be impossible to define a pristine condition as a target for restoration efforts. This conclusion adds to the critique of Moss 2008. However, we will not attempt to define the pristine conditions of River Danube as this is well beyond the scope of this study.

Introduction:

Lines 72-83: Citations needed here.

We will add the additional references to support this section and to enable readers to find additional informations.

Methods:

Line 131-133: I am assuming by the reference to Stuiver et al. 1998 that the IntCal98 was used for the calibration. However, in Table 2, it seems that for all the calibrated dates, 500 was just subtracted from the 14C age. Was that really the result of the calibration? Additionally, what was the rationale for using IntCal98? As there has been many updates to the curve, I do not think it makes sense not to use the most recent version, i.e., IntCal20. However, it would be more appropriate to use Marine20 as that is specifically for marine sediments.

Furthermore, given that linear interpolation assumes constant sedimentation, using a Bayesian approach, such as that of Bacon, could provide a more robust chronology. Given that the manuscript relies heavily on the chronology, the methods here should be brought in line with current practice.

We agree, the method of 14C dating is not in line with current practice, and we will change the method according to your suggestions. Specifically, we will use the Marine20 calibration curve in conjunction with the calib.org database to determine the ΔR value of samples from the Black Sea shelf. We will use the Calib tool to execute the calibration according to Stuiver, M., and Reimer, P.J., 1993, Radiocarbon, 35, 215-230. All figures, tables, and plots will be updated.

We further agree that the Bacon method of Blaauw and Christen (2011) generally provides more plausible age models with less artifacts such as abrupt changes of the modelled sedimentation rate. However, we have only a very limited number of 14C dates- Core 4 has only 3 data points to support an age model. We thus chose to use simple, step-wise linear interpolation as it does not rely on assumptions, which are not supported by the data. To adress possible artifacts that follow from the simple linear interpolation, we propose to point out these artifacts (such as abrupt changes in the modelled sedimentation rate) in the corresponding part of the discussion.

Results:

Line 207: Were these zones identified solely based on observation or were any statistics used?

The zones were identfied on observation without additional statistics. The underlying assumption was, that periods when a particular process dominated then the trend in d15N vs [N] is roughly constant. We will clarify this in the method section.

Line 241-243: The description of the model development should be in the methods section.

This part is meant as a brief recapitulation to help the reader with the change of subject. If this is a serious flaw then we will remove this part.

Technical comments:

Introduction:

Line 40: Replace “as displayed in” with “as seen”

We agree and will edit the manuscript accordingly.

Line 61: Replace “represents” with “is”

We agree and will edit the manuscript accordingly.

Line 64: Should be Maselli & Trincardi, 2013? 

Yes, that is a typo. It is indeed Maselli & Trincardi 2013. We will edit the manuscript accordingly.

Line 68: Replace “measurements” with “measures”

We agree and will edit the manuscript accordingly.

Methods:

Line 100: Station 2 depth is 27 m in Table 1 but 22 m here. Which is it?

That’s a typo, and will be corrected.

Line 118: Replace “better” with “less”

We agree and will edit the manuscript accordingly.

Line 127 and later: Italicize scientific names

We agree and will edit the manuscript accordingly.

Discussion:

Line 312: Should be Fulton et al. 2012?

Yes, that’s a typo, and will be corrected.

References:

Ensure all references have the same style of formatting

We agree and will edit the manuscript accordingly.

Thanks for reading our manuscript and pointing out details that can be further improved. We have copied-in your comments in bold and italics, and our reply is following directly after each comment

>> Fig. 1: It would be helpful to include a sub-map that shows the study area within a broader geographical context.

We agree and will add a sub map to make it easier to put the study area into a broader geographical context.

>> Line 194: seems TOC/N ratio is higher than 8 at 35 cm depth, better check it

We agree, this is indeed an error. We will change the sentence to “The molar TOC / N ratio decreased from 15 at the sediment surface to 8 at 18 cm sediment depth, and slightly increased to 10 at 35 cm.”

>> Line 200: at station 6, ‘the organic carbon and nitrogen content decreases in the upper 3 to 7 cm of the cores’, which is not the case from the figure. Better check it.

We agree that this part is not very clear. We will change it to: “At stations 4 and 6, the organic carbon and nitrogen content decreased from the surface down to 7 cm depth in core 4, and down to 3 cm depth in core 6, respectively (Fig. 3). The TOC content of these cores was in the range of 1.1 to 4.0 %, and the total N content was in the range of 0.10 to 0.42 %. (Fig. 3)”

>> Line 203: It’s hard to see the trends, better to provided detailed numbers or fitting lines

We would like to not add additional elements such as trend lines to the already quite condensed Figure 3. Instead, we suggest to change the text to: “In contrast, the molar TOC / N ratios decreased with sediment depth at station 4 from a TOC/N ratio of 12.1 at the surface to 8.5 in 27 cm sediment depth. At station 6, the TOC/N ratio slightly increased from a value of 11.0 at the surface to a value of 12.4 at 25 cm depth (Fig. 3 K, L).”

>> Line 210: I recommend the authors mark the four zones in Fig. 3, which would be helpful to understand Fig. 4

We agree and will mark the four Zones in Fig. 3 by adding dashed horizontal lines and corresponding labels to the sub-plots. However, this will also include to present Figure 4 and the associated text first, where we introduce the 4 Zones. If it is not an issue that we present derived data such as the four Zones (Fig. 4) before we present the profiles data first as an introduction, then we can make the changes as suggested.

Thanks for reading our manuscript and pointing out details that can be further improved. We have copied-in your comments in bold and italics, and our reply is following directly after each comment.

>> Containing large anoxic water body, Black Sea is likely affected by water column denitrification (WCD) and other N loss processes. WCDs can critically affected N isotope signals and nutrient pool and is likely to affect N sources analysis. The authors should include Black Sea WCD in the Introduction and Discussion section.

We will add water column denitrification as a significant process of nitrogen turnover to the discussion. We will further add references to the studies of Moebius & Daehnke 2015 (Limnology & Oceanography, https://doi.org/10.1002/lno.10068) who demonstrated that most of the Danube N load is assimilated into plankton biomass and not transported as DIN within the Danube River Plume. As part of organic matter, the assimilated N is much more stable, especially in oxygen-deficient environments. We thus argue that water column denitrification had no dominant impact on the isotopic composition of the organic matter in the shelf sediment.

>> Another concern is the novelty. Fulton et al., 2012 did some similar analysis regarding N fixation analysis in Black Sea with similar proxies. Authors should clarify what is the major novelty compared with Fulton et al., 2012 and prior research.

It is correct that Fulton et al. 2012 did a similar analysis. In our study we use Fulton et al. 2012 as a reference for our own results to ensure that our own results are consistent with previous observations. A novelty of our study is that we sampled the upper shelf at the Danube Delta while Fulton et al. have sampled the deep basins. While the sediment record of the deep basins integrates signals over the whole Black Sea, this is not necessarily the case in our study area where the riverine signals can be dominant over basin wide signals. The most significant novelty is the reconstruction of Danube N loads with a high temporal resolution for the past 200 years, which was not attempted by Fulton et al. 2012. And since we present a gradient from high influence of Danube to low influence, it is now possible to select future study sites depending on whether signals from the Danube or the Black Sea are to be observed.

We will make this clearer in the respective sections of introduction and discussion.

>> Line 73-75 Define delta N isotope notation first.

We agree and will add a short introduction of the notation first.

>> Section 2.4 and 2.5: The age calculation methods based on Pb and Cs isotope should also be included in methods sections.

We did not present the method for age calculation based on Pb and Cs for two reasons: 1) We used the results from Constantinescu et al. (2023) for the stations 1 and 2, and chose to refer to the original study to keep our manuscript concise. 2) We did not attempt to apply an age model for the results of core 4 as the sedimentation rate is apparently already so low that the unsupported, short-lived isotope Pb-210 is basically decayed within the first few cm sediment layer. We thus use the low penetration depth of unsupported Pb-210 and Cs-137 as an indication that the deposition of isotopically enriched N started much earlier than the start of industrialization and that the apparent deep occurrence of enriched N is not simply the result of burial of modern N by e.g. bioturbation, because otherwise modern Pb-210 and Cs-137 would also be present and those depths. In the revision, we will make this clearer.

>> Section 3.1: I did not see the age results from the Pb/Cs isotopes. I think the dating results for station 1 and 2 should also be included in Figure/Table. 

We did not present explicit age results for core 4 as we did not attempt to derive an Pb/Cs based age model for core 4. The reason is that the modern, unsupported Pb-210 and Cs-237 are decayed already in the uppermost sediment layer, and we concluded that an age model would not be feasible. Instead, we use the low penetration depth of unsupported Pb-210 and Cs-137 as an indication that the deposition of isotopically enriched N started much earlier than the start of industrialization and that the apparent deep occurrence of enriched N is not simply the result of burial of modern N by e.g. bioturbation, because otherwise the co-occurring modern Pb-210 and Cs-137 would also be present and those depths. In the revision, we will make this clearer.

We chose to not present the age results from cores 1 and 2 as we used the results of Constantinescu et al. (2023) as stated in the method section and the results section. We would like to not show the results of Constantinescu et al. (2023) to prevent any confusion of the reader and to not evoke the impression that these are our own results.

>> Section 3.2: I suggest adding more subsections here.

We agree and will restructure the section by adding more subsections to increase the readability.

>> Line 234 Figure 5: How is the f calculated? Should be included in method section.

From the subset of the date that we have assigned to Zone 2, we used the largest value of the total N content as the reference for the calculation of f, which consequently plots at the coordinate origin (0, 0) of the plots. We will add a brief explanation to the methods section to make this clearer.

>> Section 3.3: I think this section is more suitable as a part of discussion section.

We disagree because the results of the correlation analysis and the resulting plots have to be presented. However, our usage of the word “reconstruct” in the section (L 243, L 245) does indeed signal an interpretation and thus belong to the discussion section. We will split this section and shift the following part to the discussion:

“The N content-based Models 1 and 2 reconstruct the observed DIN loads of the 1955 – 2015 period reasonably well (Fig. 6 B, D). The δ15N-based Model 3 reconstructs the observed DIN load less accurately (Fig. 6 F). For the period 1800 – 1950, all three models reconstructed that the Danube DIN load was 236 to 318 kt / yr in 1800 CE”

The last sentence “For the period 1800 – 1950, all three models reconstructed that the Danube DIN load was 236 to 318 kt / yr in 1800 CE and increased gradually with 0.2 to 0.5 kt / yr2” will remain in the results part.

>> Section 4.1: This section is particularly hard to follow. The raw data in Figure 3 are hard to connect to these events described here. I think at least a time series plot should be presented here (e.g., like fig.3 in Fulton et al., 2012). And the major events/periods should be marked in figures.

We think that this is a good idea and will add a timeseries figure with marks for the events we mention in this discussion section.

>> Line 265: Consider briefly explaining what 'Unit IIb' and other jargons refers to, as not all readers may be familiar with this regional stratigraphic nomenclature.

We agree and will follow this suggestion.

>> Line 276: Why did no detect of H2S agree with time frame of zone 2 samples?

Organic matter is remineralised faster in oxic conditions than in anoxic conditions, and the Rayleigh plots in Fig. 5 indicate kinetic fractionation that is characteristic for oxic remineralisation. The absence of a H2S proxy supports this conclusion as the presence of H2S in the photic zone would also imply oxygen-free bottom water. We will add this explanation to the discussion.

>> Section 4.2 This is a very long section discussing about N sources. It could be broken down into subsection according to time periods. And I am wondering the N loss processes’ impacts on N isotopic signals.

We agree to split this section according to time periods as suggested. With respect to the effect of N-loss processes on the N isotopic signal, we already discuss how oxygen exposure enhances remineralisation (e.g.line 341 ff) and how euxinic conditions preserve biomass and the N-signature within (e.g. line 304 ff).

>> Section 4.3 The section aims to discussion the age offsets in Inorganic 14C and has little connection in core topic. I think these contents can be moved to age result section.

We agree and will move this section to the age results part.

>> Typo note: There are a lot of typos. E.g., subscript/superscript typos in lines 130, 197, 206, 210 and so on.

These typos unfortunately escaped our attention, and we will correct the typos in the revised manuscript.