General comments

Sedimentary phosphorus, which can be remobilized by diagenetic processes and reach the water column, thus increasing the risk of eutrophication, is essentially found in two reactive forms, P associated with organic matter and P associated with iron oxides. Iron oxide-bound P is released when iron oxides are reduced and the iron is not rapidly reoxidized. This occurs in sediments that come into contact with anoxic bottom water, or at least, free of oxidants for Fe(II), bearing in mind that dissolved oxygen is not the only possible oxidant. The P associated with organic matter is released when the organic matter is mineralized. The rate of OM mineralization depends on several factors. The presence of strong oxidants such as oxygen or nitrate promotes efficient mineralization.

In the study presented here, the authors emphasize a counter-intuitive fact, but one that is based on correct reasoning. Indeed, the authors describe the evolution of benthic P fluxes in relation to the oxygen concentration in bottom waters of areas of the Baltic Sea that have been studied for a long time. It is known that the anoxic periods that characterize the bottom of the Baltic Sea favour benthic P fluxes. In some years, oxygenated waters from the North Sea invade the Baltic Sea bottom. The expected effect is a trapping of P due in particular to iron oxides which become stable, and thus a decrease in the benthic fluxes of P. However, the authors show by flux monitoring data following an oxygen supply episode, that the fluxes of P have increased. They explain this by the fact that the oxygen supply significantly increases the rate of OM mineralization and that it is this process that increases the P fluxes. This is supported by dissolved inorganic carbon (DIC) flux data that point in this direction.

The conclusions are important, the idea is elegant. However, I would suggest that the authors provide some clarification on elements for discussion, and some clarification on the presentation of the results.

specific comments

1) When the bottom water becomes oxic, only a very fine fringe of sediment also becomes oxidized. Most of the sediment column (probably here from the 1st cm below the water-sediment interface) is anoxic and does not change its redox state when the supernatant water undergoes redox oscillations. The oxides present in the fine oxidized fringe of the sediment most probably have their adsorption sites very quickly saturated and do not trap the P that diffuses from the sediment. This nuance should be mentioned.

2) Fig. 1 : the 'viscious cycle': The diagram shows P entering the cycle, but no exit route for this P. In this cycle, it seems that 100% of the P that reaches the sediment is then recycled. Shouldn't we consider a fraction that remains in the sediment in an authentic form for example?

3) line 47: "Fe and Mn oxides adsorb DIP" OK for Fe oxides. It is less obvious for Mn oxides.

4) section 2.1 or section 3: nothing is said about the nature of the sediments: grain size, POC content, porosity... This is general information necessary for a better understanding of the system (and for checking the inventory calculations).

5) line 131 "The sediment was transferred into centrifuge tubes and pore water was collected using Rhizon samplers" I don't understand: the interstitial waters were collected by centrifugation or with rhizon? or both?

6) line 177: the authors refer to dissolved silica data which are unfortunately not shown here. This is unfortunate, because the increase in Si fluxes could be an indicator of the increase in bioturbation which is only briefly discussed here. The article cited below co-authored by one of the co-authors of the present manuscript is probably the best example of this: van der Loeff, M. M. R., Anderson, L. G., Hall, P. O. J., Iverfeldt, Å., Josefson, A. B., Sundby, B., & Westerlund, S. F. G. (1984). The asphyxiation technique: An approach to distinguishing between molecular diffusion and biologically mediated transport at the sediment—water interface. Limnology and Oceanography, 29(4), 675-686. doi:10.4319/lo.1984.29.4.0675

7) Figure 3 and the other figures: the shades of colour to represent the different sampling periods are not contrasted enough. An effort must be made by the reader to locate the dates. The quality of the figures needs to be improved.

8) line 2015-225: Here is a major point; The authors compare the fluxes with the inventories of P present in the sediment. They show that the inventory of P associated with iron oxides cannot explain the fluxes. However, since the authors hypothesize an increase in benthic P fluxes related to OM mineralization, it is interesting to compare these fluxes with the inventory of P associated with OM. I took the liberty of doing so based on the OM-associated P content of Figure A4 (about 15 µmol/g) and a porosity of 0.8. I find that the inventory of P associated with OM is 400 mmol/m2 for a sediment thickness of 5 cm. The measured flux of P is 2 mmol/m2/day, which means that this flux would correspond to a total mineralization of P associated with organic matter (and thus OM) in 200 days over a thickness of 5 cm. This is not consistent with the data and poses a major problem with regard to the representativeness of the measured fluxes. The measured P flux of 2 mmol/m2/day gives the impression of extracting all the P from the sediment. This is not the case if, in addition to the outgoing fluxes, there are equivalent incoming fluxes via sedimentation, which must then be described. This quantitative control is necessary.

9) line 291-296: the role of bioturbation is too quickly dismissed here. It is written that there is no sign of colonization by animals. However, the increase of Si fluxs is a possible explanation (see the reference cited above). In lines 302 to 304, an ad hoc explanation is given to describe the Si fluxes (that we would like to see). I think bioturbation is a better explanation. Moreover, the colonization of the sediment by fauna as a result of oxygenation of the waters makes sense and is in line with the authors' conclusions: an increase in mineralization of OM.

10) Dissolved oxygen concentrations are low, even after North Sea inflow events. The environment is hypoxic. Regarding OM mineralization, oxygen is a powerful oxidant, but so is nitrate. Nitrate is also an oxidant for Fe(II). In this type of eutrophic environment, nitrate concentrations are probably very high. Have they been measured? Do the dynamics of nitrate follow that of oxygen in the chronicle presented here, before, during and after the oxygenation event?

The study "Deep-water inflow event increases sedimentary phosphorus release on a multi-year scale" presents a thought-provoking hypothesis concerning the effect of deep water oxygenation on DIP fluxes from the sediments of the Eastern Gotland Basin (EGB), Baltic Sea. In a nutshell, the authors propose that the repeated inflows of 2014-2017, which led to persistently oxygenated bottom water conditions at their study sites, promoted organic matter remineralization in the surface sediments, leading to an overall enhancement in the flux of DIP to the water column relative to that observed during the typical anoxic conditions. According to the authors' calculations, the post-oxygenation flux of DIP exceeded that predicted from the re-release of P trapped immediately following the initial oxygenation, hence leading to the conclusion that an additional P source (OM remineralization) is required to explain the results. The observations run contrary to a long-held paradigm concerning retention of DIP by sediments following oxygenation. Several mechanisms are proposed to explain the results, including changes in organic matter supply and enhanced rates of remineralization under oxic conditions.

The main body of data supporting the core hypothesis is derived from lander-based flux measurements of DIP, DIC, DIN and DSi. This dataset is a very high-value component of the paper. The level of replication is good, and the parallel trends in the fluxes of all biogenic parameters during the period 2015-2018 appear robust. The more problematic aspects of the study concern the attempted closed-sum budgets of P cycling, including the upscaling of the data to larger areas of the EGB. Some of the interpretations and assumptions here need to be better elucidated. In general I am supportive of the study and especially the robust flux, porewater and sediment data, but I feel that the further treatment of that data needs improvement to be ready for Biogeosciences.

Major comments:

As raised by the first reviewer, there is an issue concerning the closed-sum budget of in-situ P pools in the sediments and the estimated DIP fluxes, that requires further elaboration in the text. The range of measured DIP fluxes at station F during 2016-2018 is 2-4 mmol/m2/d (Fig. 3a, b). The first reviewer calculates that a 2 mmol/m2/d flux would exhaust the P-org supply of the upper 5 cm of the sediment column in 200 days, assuming porosity 0.8 and P-org = 15 umol/g. I repeated this calculation with a porosity of 0.9 (probably closer to the true value in the EGB) and of course the duration is even shorter due to the lower volumetric P-org inventory in the sediments (less than 100 days, exact value depends on assumed sediment density). The flux data are correct, and so are the sediment profiles, so to me the only explanation can be that the turnover of biogenic material at the very surface of the sediments (e.g. in the “fluffy layer”) must be extremely rapid, therefore dictating the high fluxes. This is very important because it has an implication for the further interpretations. If the observed DIP flux cannot be sustained by the in-situ pool of P-org (as measured in the sediment profiles), it must be sustained by an incoming flux of organic material to the sediment surface. The higher fluxes in 2016-2018 therefore indicate first and foremost that the input of organic material has increased at station F. I suggest that the authors use their own porosity data, if available, to carry out this exercise for themselves and present the results and implications in a revised version.

Following on from this line of thought, I am concerned that the study interprets the enhanced fluxes (and especially DIP fluxes) at these study locations too simply. For example, section 3.4 is devoted to implications of DIP release for coastal deoxygenation and is set up in a way that extrapolates the observed fluxes to a larger area, implying overall enhancement of DIP fluxes and therefore a possible feedback to eutrophication/hypoxia. On the contrary, I would challenge the authors that what is being observed here is an internal rearrangement of the distribution of organic material in the deep areas of the EGB after the inflow, leading to changes in remineralization rates at different locations. Therefore the enhanced fluxes at one location are presumably matched by lower fluxes at another location where the same organic matter would be sedimented under normal conditions. The authors refer to several recent papers describing possible physical mechanisms such as gravity currents and particle aggregation following oxygenation, which could modify the spatial distribution of sedimentation of organic material. I think these are critical to understanding the results. For example, I would ask the question whether we are looking primarily at an enhancement in particle shuttling into deeper areas following the inflow? (it is interesting to note that the deeper site F shows the strongest effects on fluxes, see e.g. Fig. A2, despite a similar oxygenation timeline to the shallower site E, see e.g. Fig. A1).

I do not deny that remineralization of OM under oxic conditions can be more rapid (at least for certain compounds; I refer to the cited Arndt et al. 2013 ESR paper for the concept) and hence that the observed effects on fluxes can be influenced directly by the presence of oxygen. However the fact remains that the in-situ pool of OM is insufficient to support the temporally-extrapolated fluxes as presented in Fig. 5, under either oxic or anoxic conditions. So the authors must emphasize the role of variable carbon inputs (donor control in the terminology of Arndt et al.) in their interpretation.

It is of course a fascinating result that there is a genuine increase in the flux of DIP from sediments to the water column in 2016-2018, despite the oxic bottom water conditions. This does indeed run contrary to the often-cited paradigm of P retention with Fe oxides under oxic conditions (e.g. early Einsele and Mortimer papers). This begs the question, why should the P not be retained after release from OM during remineralization? Is it simply a question of the relative availability of Fe-Mn oxides with respect to P, or do the authors have an alternative hypothesis? Note that Fe:P ratios are often used in limnological studies to estimate potential P retention in sediments (Jensen et al Hydrobiol. 235, 1992). If the authors have bulk sediment Fe and Mn contents (or even e.g. CDB-Fe, CDB-Mn data) these could be added to aid this investigation. At the very least, further discussion is required on this important topic.

Minor comments:

Line 61, 340: The reference to Jilbert and Slomp (2013) is not ideal for this concept. That paper focused more on the role of particle shuttling in stimulating authigenic mineral formation. I suggest to find an alternative.  

Line 195-200 and Fig 3: It took me a while to understand the message here, which is that an impact of Fe-P dissolution on the flux data would be seen as a DIC:DIP ratio even lower than the mean value of 40 observed in 2016-2018. I think the figure can be annotated to improve this concept. Perhaps add another line for the ratio of 40, or a segment for the ratio 20-70 as referred to in the text, and modify the text to link better to the figure. Also label the lines in the figure (including the existing Redfield line) with the corresponding ratios.

Line 231: It is not clear what is meant by “change over time in response to external factors”. Please clarify.

Line 233: I think that the SMHI data referred to here is presented in Fig. A1. If so please cite the figure.

Line 318-320: Should the units in the text be the same as those in Fig. 5 (mmol/m2/d)?

Review of Hylen et al. Deep-water inflow event increases sedimentary phosphorus release on a multi-year scale

I found this paper to be clearly communicated and I appreciate the targeted focus of the study. I think the ability to capture a sediment-water flux response to a series of inflow events in the Baltic Sea is unique and the results are highly relevant to our understanding of the biogeochemical responses to eutrophication and oxygen depletion. The combination of in-situ sediment-water fluxes and porewater analysis (and supplemented by available monitoring) provides a sufficient dataset to explore the questions proposed by the authors. I do think some critical questions should be addressed in a revision of the paper, notably with respect to the impact of the seasonality of sediment-water fluxes, the ratio of DIC to DIP fluxes, and the balance between the inflows causing elevated organic deposition versus stimulating additional remineralization through oxygenation. I provide some specific and general comments below:

• I think Figure 1 is unnecessary. The main features of the feedbacks have been previously well described in the literature and the authors description of the feedback in the text is clear and adequate.
• Figure 2 – please specify the depths included to compute “bottom water” oxygen
• I think it could be stated in the paragraph in Line 120 that the prior fluxes (2008, 2010, and 2015) were performed using the same lander system as in the more recent fluxes.
• I would like the authors to enhance the discussion around the differences in sampling season between the earlier and later sediment-water fluxes. This is an important feature of the interpretation of the flux enhancement over time. Annual cycles of sediment-water fluxes in many temperate estuaries involve substantial seasonal variation. Is there a reason that a deep, typically anoxic basin should be different? Are the other datasets to cite from the literature? What about seasonal temperature effects? Could a reactive pool of organic material that may have accumulated during colder months be available by April, where if a spring bloom was important, perhaps this material would have been exhausted by late summer, when some of the other fluxes were measured?
• I appreciate that all of the flux rates (DIC, DIP, DSi, NH4) increased together in 2016-2017 at station F. But in reference to the Redfield ration, there were substantial excess DIP fluxes in 2016 to 2018. The authors consider this DIP excess but do not provide a clear explanation as I can see. The later discussion on elevated Mn oxide deposition might be relevant, as would Mn oxides scavenge P from the water-column preferentially? They also don’t explore sufficiently why the DIC flux exceeds the DIN flux from a Redfield perspective, which perhaps is due to denitrification?
• In your conclusions, I think it is important to make a clear distinction between the inflows causing elevated POM flux and the inflows providing oxygen to enhance remineralization. If the former is more important, it would suggest that the inflow enhanced POM flux to sediments is the reason for the elevated sediment-water fluxes, and the oxygenation is simply associated with the inflow/flux in time. If this were true, it would dampen the argument that oxygenation might actually cause a transient enhancement of the “Vicious cycle”
• I was hoping the authors might come back to the idea of the “viscous cycle” in a discussion paragraph and clearly put their results in the context of that hypothesis. It has been shown that DIP concentration in bottom-waters is related to oxygen concentration in the Baltic – and this is cited as indirect evidence of the enhancement of sediment-water DIP flux under low oxygen. The supplementary data you provide on bottom-water DIP concentrations doesn’t appear to show any enhancement of water-column DIP, even with the elevated rates you measured. Is this because the DIP-enhancement of sediment-water flux you measured only occurred in a limited area (although the up-scaling you did suggests it could be meaningful over larger scales). What if the April measurements are not representative of most of the year, as they are clearly higher than the previous years when measurements were made during summer? If the DIP flux enhancement is due to oxygenation, do your measurements suggest a timescale of this enhancement (1-2 years) given reoxygenation of an area? I realize this is a lot to digest, but I think a paragraph in the discussion that evaluates your results in the greater context of the cycle, citing the limitations of your measurements, would help the paper.