Dear editor and reviewers,

First of all, thank you very much for your support and constructive comments on our manuscript. We respond to all comments below. However, there are some critical points that were raised by several reviewers. These relate to ambiguities in the a) selection and b) characterization of source water types, the c) robustness of our OMP analysis with regard to the oxygen and nutrient definition of source water types (SWTs), d) long-term changes and e) title of the manuscript. Furthermore, we were asked to follow a uniform and comprehensible approach to resolve these ambiguities.

Below, we respond to these overarching points and hope to thereby resolve the issues raised by the reviewers. This is followed by point-by-point responses.

a) Selection of source water types:

There are two main reasons for our decision to focus on the water masses that form near the surface of the Indian Ocean: 1. These water masses are essentially responsible for the ventilation of the OMZs, which takes place at depths between 100 and 1000 m. 2. These are the water masses that carry the influence of global warming on the properties of source water types into the depths of the ocean. The OMP analyses carried out with data from the same region but for different time periods can then in turn be used to estimate the influence of mixing on the development of the OMZ in response to global warming and its effects on circulation in the ocean.

The number of SWTs we considered was reduced to six by means of an exclusion process, as will be explained in more detail later, for methodological reasons, as suggested by reviewer 1. However, in addition to four water masses that outcrop at the surface in the Indian Ocean, the six water masses  also include the Indonesian Throughflow water (ITFW) and modified Antarctic bottom water ( mAABW). These two water masses are exceptions in that they are not formed at the surface of the Indian Ocean but in the Indonesian Seas and the deep sea, respectively. Unlike reviewer 1, however, we also consider the Bay of Bengal water (BOBW) to be a source water formed at the surface of the Indian Ocean (Shetye et al., 1996). In any case, both the ITFW and the mAABW are important water masses for the ventilation of the OMZ in the Indian Ocean. We have therefore taken them into account and assumed that they influence the ventilation of the OMZ due to changing mixing ratios. We will explain these reasons for the selection of source water masses and the exclusion process in more detail in the revised manuscript to make our approach more comprehensible to readers.

b) Characterization of source water types

We have compiled the published properties of all SWTs relevant to the OMZ of the Indian Ocean in Table 1, which shows that there is no uniform definition of source water properties. This is due to the fact that the properties of the source water types were defined at different times in different locations (i.e., geographical region and water depth) and are based on different data sources. Taking these variabilities into account, we agree with the reviewer that a well-founded, systematic and, above all, comprehensible approach is necessary to achieve a consistent picture that enables us to compare different time periods. Critical elements brought up by the reviewer to achieve this, are 1) PGW and RSW defined downstream within the Arabian Sea rather than formation or entry point.  2) the selection of the database.

b.1) PGW and RSW defined downstream within the Arabian Sea

In defining source water properties, it is important to note that water masses are not exported to depth continuously from the surface, and therefore surface mean properties are often not representative of the water that ultimately ventilates the ocean interior. For this reason, source water types are commonly characterized just below the surface mixed layer, where waters begin their subsurface pathways. In the case of PGW and RSW, both the outcropping regions and the depths at which these waters descend vary spatially, and source properties are therefore typically defined at their entry into the subsurface Indian Ocean, rather than at their exact formation sites. We agree that Persian Gulf water (PGW) and Red Sea water (RSW) properties are defined downstream within the Arabian Sea rather than at their exact formation sites or entry point. According to the pathways described by Tchernia (1980), we have revised the definition regions for PGW and RSW to around the Oman slope and sill, respectively. These are the entry points into the Arabian Sea. The defined source water types represent slightly modified primary water masses. This choice is intentional as properties defined here therefore represent the ventilating PGW and RSW as they actually enter and influence the Arabian Sea OMZ. As a result, oxygen concentrations increase and nutrient concentrations decrease, i.e., water masses more closely resemble the properties of surface water, as suggested by the reviewers. We have revised the methods section of the manuscript to clarify source water types selection and definition.

b.2) Database

With regard to the database, we have chosen the World Ocean Atlas (WOA) because it is freely accessible and summarizes all available and quality-checked data that can be used to calculate the properties of water masses for specific regions and depth intervals. For temperatures and salinity, there are average values for different decades, so we have data available for the periods from 1995 to 2004 and from 2015 to 2022 with high density of observations in source regions according to WOA “number of observations” documentation.

c) Robustness of our OMP analysis

One major concern by the referee was the robustness of our eOMP solution to variability in oxygen and nutrient properties of the SWTs, since we assumed constant oxygen and nutrient values for 1995 and later years.

There are no decadal averages for oxygen and nutrients in the WOA database, as the data base is presumably insufficient. However, the distribution of nutrients is strongly influenced by the biological carbon pump, which ensures that nutrients that reach the euphotic zone are also exported again. As far as we know, there is no evidence to date that the carbon storage of the biological pump has changed, so we assume that also the nutrient concentration in the SWTs have not changed significantly between 1995 and 2016. This assumption is also based on the fact that the defined SWTs are not the product of mixtures of different water masses.

Unlike nutrient concentration, changes in oxygen concentration can be assumed, as the solubility of oxygen depends on temperature and salinity. We have now calculated the change in oxygen solubility based on WOA temperatures and salinity for the two selected decades, the difference was added and subtracted from the oxygen concentration that we calculated from WOA data and we used the resulting oxygen values for each source water type to test the robustness of the eOMP solution to increase or decrease in oxygen values of SWTs in the later years, as explained later. The oxygen variability test confirms the robustness of our eOMP solution and conclusions. This is not perfect, but in our opinion, it is a comprehensible approach that represents the influence of warming on oxygen concentration in a representative manner. Additionally, the changes to the definition of source water types proposed by the reviewers have had only minor effect on the results, which is a clear indication of the robustness of our OMP analysis. 

d) Long-term trends versus snapshots at different points in time

In the manuscript, we do indeed refer to long-term trends, but we compared two periods (1995 and 2016/2018) with each other. We agree with the reviewers that such a comparison does not indicate a trend. However, the evaluation of the literature, in line with our previous results, clearly shows that the OMZ was in a phase of expansion in 1995 (Banse et al., 2014; Rixen et al., 2014, 2020; Goes et al., 2020). Argo float data show now that the OMZ in the Arabian Sea has been shrinking massively since around 2013 (Liu et al., 2024). The GLODAP database enables us to conduct OMP analysis for the years 1995 and 2016, representing periods of OMZ expansion and shrinking in the Arabian Sea..

Furthermore, the compared periods do not represent persistently opposite or extreme ENSO or IOD phases, but rather a mixture of weak, moderate, and neutral conditions, suggesting that pronounced climate anomalies are of little significance for the differences observed. Differences between 1995 and 2016 thus allow conclusions to be drawn about processes that could explain a phase transition from expansion to contraction of OMZs. We have also included a paragraph in the discussion section to clarify the influence of interannual and decadal variability. Our conclusion that underlying processes is a zonal circulation that weakens in favor of a meridional circulation, are also largely consistent with model results (Ditkovsky et al., 2023), which reinforces confidence in OMP and the models. However, we can just conclude that such a transition seems to have occurred and not that this is a trend which would point to an increasing contraction in future. We will express this clearer in the revised versions of the manuscript.

e) Title

As suggested by the reviewers we have revised the title of the manuscript to reflect the new contributions of the article to: “Recent changes in dissolved oxygen concentration in the Indian Ocean Oxygen Minimum Zones driven by reorganization of water masses”

We hope that this has addressed the main points of criticism raised by the reviewers.

Please find below the point-by-point response.

POINT-BY-POINT RESPONSE

1.  COMMENTS ON eOMP METHOD:

Thank you for this constructive comment. Regarding the formulation of the eOMP analysis. We agree that, as originally presented, the inclusion of 8 source water types (SWTs) plus a remineralization term leads to an underdetermined system when constrains are only 6 (temperature, salinity, oxygen, phosphate, nitrate, and mass). 

To address this concern and ensure a determined set of equation, we have added a new constraint (silicate) and methodologically reduced the SWTs to 6. This brings the number of unknown to 7 and number of constraints to 7.

In the paragraph text below we explained the procedure used to reduce the SWTs, we have added this paragraph to the method section of the revised manuscript:

“We performed a preliminary underdetermined eOMP analysis using the entire (8) SWTs. Thereafter, we separated the SWTs into major SWTs (PGW, STSW, mAABW, ITFW) and minor SWTs (AAIW, ASHSW, RSW and BoBW) based on their contributions in the Arabian Sea OMZ. Then, we performed 4 sets of overdetermined eOMP analysis using 5 SWTs and 7 constraints (temperature, salinity, oxygen, phosphate, nitrate, silicate and mass). The 5 SWTs included 4 constant major SWTs plus 1 minor SWTs at a time. Using this method, we identify PGW, RSW, STSW, mAABW and ITFW as the major contributors to the OMZ with the least residuals, and the combination of PGW, RSW, STSW, mAABW, ITFW and AAIW as the major contributors in the entire western Indian Ocean transect with the least residuals (see figures in supplementary document). Since these six SWTs resulted in even smaller residuals in the Eastern Indian Ocean transect, we used them in the final eOMP analysis.” 

2. COMMENT ON DEFINITION OF PGW:

We earlier explained (kindly refer to the summary above) that we agree that the PGW used in the original analysis represents a higher nutrient form of PGW because its properties were defined closer to the Arabian Sea. We have revised the definition of PGW parameters to better reflect outflow properties. We  re-defined the properties of PGW using the World Ocean Atlas (WOA) data around the Oman slope at depth between 100-300m, latitude between 23.6 -27.1°N and longitude 55.5 -57.5°E according to the PGW pathway described by Tchernia (1980).  Defining the properties around this region helps us capture climate driven changes in the source region and outflow properties as observed in the increase in temperature value. Defining PGW properties closer to their entry regions therefore provides a more realistic representation of the oxygen and nutrient characteristics that actually ventilate the Indian Ocean/OMZs.

3. COMMENTS ON SURFACE ORIGINATED WATER MASSES: 

As earlier explained (kindly refer to summary above), the major aim of selecting our source water types was to take the effects of global warming into account.

We intentionally defined Source water type (SWT) properties at the regions where these waters enter and influence the Indian Ocean/OMZs, rather than at their distant outcrop or formation sites. For SWTs formed within the Indian Ocean, Arabian Sea and Bay of Bengal (e.g., STSW, ASHSW, BoBW), we defined their properties from the WOA data based on formation region and pathways described in literature. SWTs whose outcrop region/formation sites are not within the Indian Ocean (i.e., PGW, RSW, mAABW, ITFW), were defined in regions where there enter into the Indian Ocean/Arabian Sea based on their pathways described in literature. We agree that SWTs are much more oxygen saturated at their outcrop region, however, defining SWT properties closer to their entry regions therefore provides a more realistic representation of the oxygen and nutrient characteristics that actually ventilate the Indian Ocean/OMZs.

Our approach is consistent with previous Indian Ocean OMP and water-mass studies, which commonly define water masses within the basin or at entry points, rather than strictly at outcrop locations (e.g., Tomczak & Large, 1989; You, 1998; Hupe & Karstensen, 2000; Acharya & Panigrahi, 2016). These studies similarly show that water masses experience substantial modification between formation and ventilation of the northern Indian Ocean, particularly with respect to oxygen concentration of SWTs. Furthermore, our defined SWT properties is quite similar to the values in literature when you compare Table 1 and 2.

In the revised manuscript, we have replaced this term “surface-originated water masses” with “primary source water type” the meaning of this term has also been defined within the manuscript (i.e., source water types which have not undergone rigorous mixing) and are distinct from secondary mixtures  such as Arabian Sea Water (ASW) which is a strong mixture of Red Sea Water (RSW) and Persian Gulf Water (PGW), or Indian Central Water (ICW), which is a mixture of Antarctic Intermediate Water (AAIW) and Subtropical Surface Water (STSW). For example, we defined the PGW and RSW around the slope of Oman and around the sill respectively, where they have not undergone much mixing within the Arabian Sea, we also defined STSW around the formation region and AAIW around the entrance region into the Indian Ocean.

4. COMMENTS ON INFERRED CLIMATE-DRIVEN CHANGES AND VARIABILITY:

As explained earlier (kindly refer to summary above) evidence suggests that the observed differences between the analyzed periods likely reflect a longer-term shift in the mean state rather than short-lived variability. The compared periods do not represent persistently opposite or extreme ENSO or IOD phases, but rather a mix of weak, moderate, and neutral conditions, suggesting that the observed changes reflect longer-term circulation variability rather than short-lived climate events. According to the Japan Meteorological Agency (JMA) IOD and ENSO events record (https://ds.data.jma.go.jp/tcc/tcc/products/elnino/iodevents.html), the year 1995 coincided with a negative IOD event, occurring alongside a developing La Niña in autumn, conditions that can influence regional circulation and productivity but were limited in duration. The later period exhibits greater variability: 2016 followed the strong 2015 El Niño and was associated with a negative IOD, with neutral ENSO conditions, while 2017 experienced a weak positive IOD under neutral ENSO conditions. In 2018, a weak positive IOD occurred from summer to autumn concurrently with a weak El Niño. Overall, this sequence reflects alternating and moderate ENSO–IOD states rather than a persistent bias toward either extreme phase. Consequently, the observed changes in dissolved oxygen and water mass composition are unlikely to be driven by individual ENSO or IOD events alone, but instead point to circulation changes and basin-scale reorganization as the dominant controls on recent OMZ variability.

We have also discussed the influence of inter-annual and decadal variability in the revised manuscript and included a limitation section, to discuss limitations of the methodology.

5. COMMENTS ON STRUCTURE, PRESENTATION AND TITLE:

We agree that the original title and parts of the Introduction could give the impression of a broad review rather than a focused research article. To address this, we have revised the Introduction and the Study area to be more precise and change the title to: 

“Recent changes in dissolved oxygen concentration in the Indian Ocean Oxygen Minimum Zones driven by reorganization of water masses”

6. COMMENTS ON SEPARATING RESULTS AND DISCUSSION: 

We have revised the manuscript by separating the 'Results' and 'Discussion' sections, with the aim of clarifying the conclusions.

7. COMMENTS ON MANUSCRIPT FIGURES:

We have included more lines of discussion for figure 4 and 5 while figure 7 would be moved to the supplementary figures.

OTHER POINTS

8. SUGGESTION ON USING SOURCE WATER TYPES:

Thank you for this suggestion, we have adopted the term “source water types” for source waters across the manuscript.

9. COMMENTS ON EQUATION 4:

Equation 4 :  oxygen consumption = measured oxygen – oxygen supply + oxygen residual

This equation represents biological oxygen consumption due to remineralization, and it is exactly equal to the remineralization term calculated within the eOMP framework which can be retrieved with the equation below:

 oxygen consumption2 = -155*1 * phosphate remin. 

-155*1 is the conversion ratio of oxygen/phosphate

And phosphate remin. is the phosphate remineralization values from the eOMP solution.

Both equations give exactly the same value for oxygen consumption, we have included the residual in equation 4 for this reason to make it equal to the term calculated using phosphate remineralization. The difference however without the residual is very small and negligible.

10. ADDING REFERENCE TO FIGURE 1:

We have added a reference to Figure 1. And highlighted the region where the source water types were measured in this figure as well.

11. COMMENTS ON LINE 668: 

What we intend to imply is that despite increased oxygen consumption, there is an overall increase in dissolved oxygen. We have revised the section of the manuscript to read as follows:

“…the enhanced inflow of oxygen-rich southern waters increases physical oxygen supply. Although this is accompanied by higher biological oxygen consumption, the net result is an overall increase in dissolved oxygen...”

12. COMMENTS ON LINE 680:  

We have separated the results and discussion sections and emphasized the speculative nature of our connection. The text has been revised as follows:

“While the present analysis cannot resolve multi-centennial variability, the observed changes are conceptually consistent with paleo and historical evidence for long-term variability in Indian Ocean OMZ intensity. Importantly, our study compares two periods (1995 and 2016/2018) in which the AS OMZ expanded and contracted as discussed earlier. By comparing these contrasting periods, our analysis provides insight into how a phase transition in OMZ behavior may have occurred. The current reversal is driven by circulation reorganization rather than sea-level or glacial forcing, indicating that the present phase transition operates under a different climatic background than in the past.”

Thanks, your suggestions will be considered in our revised manuscript.

1. Thanks for your comments, we have revised the manuscript title to reflect our findings:

“Recent changes in dissolved oxygen concentration in the Indian Ocean Oxygen Minimum Zones driven by reorganization of water masses”

2. Thanks for this suggestion about the abstract. We have streamlined the abstract in the revised manuscript to focus on main findings.

3. Thanks for your recommendation. We will be adding the suggested distribution maps.

4. Thank you for this comment about Figure 6, we have clarified the calculation and interpretation in the methods as well as in the Figure caption.

The measured oxygen values represent average dissolved oxygen within the OMZ core (O₂ < 20 µmol kg⁻¹) over the same depth ranges used in the eOMP analysis (Arabian Sea: 150–1000 m; Bay of Bengal: 150–800 m). These spatial boundaries and the averaging method are now explicitly stated to ensure consistency with the oxygen supply estimates.

Oxygen supplied by mixing is calculated from the eOMP-derived water-mass fractions and their source oxygen concentrations, averaged over the same OMZ volume. Biological oxygen consumption is then estimated as the difference between mixing-derived oxygen supply and measured oxygen, following standard eOMP practice. Local mixing and diffusive processes are implicitly included in the mixing term, while unresolved small-scale effects are assumed to be minor relative to the basin-wide changes observed.

MINOR COMMENTS AND RESPONSES (Comments in bold letters)

        5. Line 44: does not only affect -> not only affects.

           Changes implemented in the revised manuscript

1. Line 60: are the semi-enclosed basins …->  are semi-enclosed basins …

              Changes implemented in the revised manuscript

1. Lines 59-65: This paragraph suffers from major language issues and an unclear conceptual framework. The definitions of hypoxia and microbial hypoxia are currently confusing and partially incorrect, and the sentence “Hypoxia defines the oxygen threshold…”is grammatically incorrect. Please revise for both language and conceptual clarity.

               The paragraph has been revised to improve clarity.

1. Line 68: Nevertheless, …-> In particular, …

              Changes implemented in the revised manuscript

1. Line 75: OMZ -> OMZs

              Changes implemented in the revised manuscript

1. Line 89: close link between …to …-> a close link between … with …

                Changes implemented in the revised manuscript

1. In Figure 3, the seven bar charts have very similar visual appearance, while the panel sizes and alignment are inconsistent. A reorganization of the figure layout, along with clear indication of the study region in each subpanel, would substantially improve clarity.

                Thanks for your comment, the figure has been revised to improve clarity as suggested.

Dear editor and reviewers,

First of all, thank you very much for your support and constructive comments on our manuscript. We respond to all comments below. However, there are some critical points that were raised by several reviewers. These relate to ambiguities in the a) selection and b) characterization of source water types, the c) robustness of our OMP analysis with regard to the oxygen and nutrient definition of source water types (SWTs), d) long-term changes and e) title of the manuscript. Furthermore, we were asked to follow a uniform and comprehensible approach to resolve these ambiguities.

Below, we respond to these overarching points and hope to thereby resolve the issues raised by the reviewers. This is followed by point-by-point responses.

a) Selection of source water types:

There are two main reasons for our decision to focus on the water masses that form near the surface of the Indian Ocean: 1. These water masses are essentially responsible for the ventilation of the OMZs, which takes place at depths between 100 and 1000 m. 2. These are the water masses that carry the influence of global warming on the properties of source water types into the depths of the ocean. The OMP analyses carried out with data from the same region but for different time periods can then in turn be used to estimate the influence of mixing on the development of the OMZ in response to global warming and its effects on circulation in the ocean.

The number of SWTs we considered was reduced to six by means of an exclusion process, as will be explained in more detail later, for methodological reasons, as suggested by reviewer 1. However, in addition to four water masses that outcrop at the surface in the Indian Ocean, the six water masses also include the Indonesian Throughflow water (ITFW) and modified Antarctic bottom water ( mAABW). These two water masses are exceptions in that they are not formed at the surface of the Indian Ocean but in the Indonesian Seas and the deep sea, respectively. Unlike reviewer 1, however, we also consider the Bay of Bengal water (BOBW) to be a source water formed at the surface of the Indian Ocean (Shetye et al., 1996). In any case, both the ITFW and the mAABW are important water masses for the ventilation of the OMZ in the Indian Ocean. We have therefore taken them into account and assumed that they influence the ventilation of the OMZ due to changing mixing ratios. We will explain these reasons for the selection of source water masses and the exclusion process in more detail in the revised manuscript to make our approach more comprehensible to readers.

b) Characterization of source water types

We have compiled the published properties of all SWTs relevant to the OMZ of the Indian Ocean in Table 1, which shows that there is no uniform definition of source water properties. This is due to the fact that the properties of the source water types were defined at different times in different locations (i.e., geographical region and water depth) and are based on different data sources. Taking these variabilities into account, we agree with the reviewer that a well-founded, systematic and, above all, comprehensible approach is necessary to achieve a consistent picture that enables us to compare different time periods. Critical elements brought up by the reviewer to achieve this, are 1) PGW and RSW defined downstream within the Arabian Sea rather than formation or entry point.  2) the selection of the database.

b.1) PGW and RSW defined downstream within the Arabian Sea

In defining source water properties, it is important to note that water masses are not exported to depth continuously from the surface, and therefore surface mean properties are often not representative of the water that ultimately ventilates the ocean interior. For this reason, source water types are commonly characterized just below the surface mixed layer, where waters begin their subsurface pathways. In the case of PGW and RSW, both the outcropping regions and the depths at which these waters descend vary spatially, and source properties are therefore typically defined at their entry into the subsurface Indian Ocean, rather than at their exact formation sites. We agree that Persian Gulf water (PGW) and Red Sea water (RSW) properties are defined downstream within the Arabian Sea rather than at their exact formation sites or entry point. According to the pathways described by Tchernia (1980), we have revised the definition regions for PGW and RSW to around the Oman slope and sill, respectively. These are the entry points into the Arabian Sea. The defined source water types represent slightly modified primary water masses. This choice is intentional as properties defined here therefore represent the ventilating PGW and RSW as they actually enter and influence the Arabian Sea OMZ. As a result, oxygen concentrations increase and nutrient concentrations decrease, i.e., water masses more closely resemble the properties of surface water, as suggested by the reviewers. We have revised the methods section of the manuscript to clarify source water types selection and definition.

b.2) Database

With regard to the database, we have chosen the World Ocean Atlas (WOA) because it is freely accessible and summarizes all available and quality-checked data that can be used to calculate the properties of water masses for specific regions and depth intervals. For temperatures and salinity, there are average values for different decades, so we have data available for the periods from 1995 to 2004 and from 2015 to 2022 with high density of observations in source regions according to WOA “number of observations” documentation.

c) Robustness of our OMP analysis

One major concern by the referee was the robustness of our eOMP solution to variability in oxygen and nutrient properties of the SWTs, since we assumed constant oxygen and nutrient values for 1995 and later years.

There are no decadal averages for oxygen and nutrients in the WOA database, as the data base is presumably insufficient. However, the distribution of nutrients is strongly influenced by the biological carbon pump, which ensures that nutrients that reach the euphotic zone are also exported again. As far as we know, there is no evidence to date that the carbon storage of the biological pump has changed, so we assume that also the nutrient concentration in the SWTs have not changed significantly between 1995 and 2016. This assumption is also based on the fact that the defined SWTs are not the product of mixtures of different water masses.

Unlike nutrient concentration, changes in oxygen concentration can be assumed, as the solubility of oxygen depends on temperature and salinity. We have now calculated the change in oxygen solubility based on WOA temperatures and salinity for the two selected decades, the difference was added and subtracted from the oxygen concentration that we calculated from WOA data and we used the resulting oxygen values for each source water type to test the robustness of the eOMP solution to increase or decrease in oxygen values of SWTs in the later years, as explained later. The oxygen variability test confirms the robustness of our eOMP solution and conclusions. This is not perfect, but in our opinion, it is a comprehensible approach that represents the influence of warming on oxygen concentration in a representative manner. Additionally, the changes to the definition of source water types proposed by the reviewers have had only minor effect on the results, which is a clear indication of the robustness of our OMP analysis. 

d) Long-term trends versus snapshots at different points in time

In the manuscript, we do indeed refer to long-term trends, but we compared two periods (1995 and 2016/2018) with each other. We agree with the reviewers that such a comparison does not indicate a trend. However, the evaluation of the literature, in line with our previous results, clearly shows that the OMZ was in a phase of expansion in 1995 (Banse et al., 2014; Rixen et al., 2014, 2020; Goes et al., 2020). Argo float data show now that the OMZ in the Arabian Sea has been shrinking massively since around 2013 (Liu et al., 2024). The GLODAP database enables us to conduct OMP analysis for the years 1995 and 2016, representing periods of OMZ expansion and shrinking in the Arabian Sea..

Furthermore, the compared periods do not represent persistently opposite or extreme ENSO or IOD phases, but rather a mixture of weak, moderate, and neutral conditions, suggesting that pronounced climate anomalies are of little significance for the differences observed. Differences between 1995 and 2016 thus allow conclusions to be drawn about processes that could explain a phase transition from expansion to contraction of OMZs. We have also included a paragraph in the discussion section to clarify the influence of interannual and decadal variability. Our conclusion that underlying processes is a zonal circulation that weakens in favor of a meridional circulation, are also largely consistent with model results (Ditkovsky et al., 2023), which reinforces confidence in OMP and the models. However, we can just conclude that such a transition seems to have occurred and not that this is a trend which would point to an increasing contraction in future. We will express this clearer in the revised versions of the manuscript.

e) Title

As suggested by the reviewers we have revised the title of the manuscript to reflect the new contributions of the article to: “Recent changes in dissolved oxygen concentration in the Indian Ocean Oxygen Minimum Zones driven by reorganization of water masses”

We hope that this has addressed the main points of criticism raised by the reviewers.

Please find below the point-by-point response.

POINT-BY-POINT RESPONSE

1. RESPONSE TO MAJOR COMMENT #1. METHODOLOGICAL LIMITATIONS OF eOMP ANALYSIS UNDER NON-STATIONARY CONDITIONS:

In addition to our explanation in the above (refer to summary above), our analysis already considered changes in temperature of Source Water Types (SWTs) which accounts for global warming. We have gone further to consider possible changes in oxygen caused by temperature changes based on oxygen solubility.

We performed oxygen variability test, using higher and lower oxygen scenarios. The result from this test agrees with our conclusions. Below is a paragraph that describes the variability test and the paragraph has been added to methods section in the revised manuscript.

“We performed oxygen variability test to demonstrate the robustness of our eOMP result to expected temporal variability in SWT properties. Oxygen saturation was calculated from temperature and salinity of each SWT using the formulation of Benson and Krause (1984) and converted from µmol L⁻¹ to µmol kg⁻¹ using seawater density derived from the UNESCO equation of state at atmospheric pressure. Differences in oxygen saturation between 1995 and 2018 were interpreted as oxygen change. The calculated oxygen change was added and subtracted from the 1995 oxygen values to derive a higher oxygen and lower oxygen scenarios respectively for each SWT. We performed the eOMP analysis oxygen variability test using the higher oxygen and lower oxygen scenarios. The resulting variations in derived SWT fractions and oxygen supply due to mixing were not significantly different from the constant state oxygen scenario”

In conclusion our findings remain valid as results from the variability test do not alter the direction or significance of our main findings ((i.e., the enhanced contribution of oxygen-rich southern water masses and the concurrent reduction of local/equatorial water masses).

Furthermore, while we agree that dissolved oxygen can exhibit significant temporal variability, in addition to previous explanation, large-scale changes in nutrient concentrations of source water types are not expected to be smaller over the considered timescale. Nutrient distributions in the ocean interior are primarily controlled by long-term biological remineralization and large-scale circulation, processes that typically operate on centennial to millennial timescales (Broecker & Peng, 1982; Sarmiento & Gruber, 2006). The mean sequestration times in the Indian and Pacific Ocean basins are relatively uniform at 100–200 years, with longer sequestration times associated with areas of sub-tropical mode water formation and the poorly-ventilated shadow zone in the Arabian Sea (DeVries et al., 2012). We therefore assume a constant nutrient state in our analysis.

2. RESPONSE TO MAJOR COMMENT #2. SELECTION OF SOURCE WATER MASSES AND PHYSICAL CONSISTENCY:

Our selection of source water types (SWTs) was intentional and designed to focus on primary source water types (i.e SWTs that are weakly modified by mixing) and therefore better account for temperature change due to global warming, rather than secondary mixtures such as ICW and SAMW.

Although ICW and SAMW are not explicitly included, their influence is fully represented through their primary source components i.e., the Subtropical Surface Water (STSW) and Antarctic Intermediate Water (AAIW). These water masses together form the ventilating branches of the Indian Central and Subantarctic Mode Waters (Tomczak & Large, 1989; You, 1998; Schott & McCreary, 2001; Sokolov & Rintoul, 2002; Vianello et al., 2017). Therefore, including both the ICW and the SAMW separately would probably introduce redundancy and potentially destabilize the eOMP solution.

In our study, the Modified Antarctic Bottom Water (mAABW) does not indicate that new deep bottom water is directly reaching the OMZ cores. Instead, it acts as a tracer showing the growing influence of deep southern-origin waters in the Indian Ocean. This pattern agrees with recent studies showing a weaker Indonesian Throughflow (ITF) and a stronger transport of southern-sourced water toward the north (Ditkovsky et al. 2023). 

Although Southern Ocean waters have long transit times, the observed changes could reflect redistribution of existing southern-origin waters within the Indian Ocean driven by circulation changes, rather than solely the arrival of new deep waters.

This interpretation fits well with both model results and observations (P. Vallivattathillam et al. 2023, Liu et al. 2024) and is further supported by the low residuals in our eOMP analysis, which indicate a stable and realistic solution.

We have included the text below in the method section of the revised manuscript to improve the clarity:

“Our selection of source water types (SWTs) designed to focus on primary source water types rather than secondary mixtures such as Indian Central Water (ICW) and Subantarctic Mode Water (SAMW). Although ICW and SAMW are not explicitly included, their influence is fully represented through their parent source components i.e., the Subtropical Surface Water (STSW) and Antarctic Intermediate Water (AAIW). These water masses together form the ventilating branches of the Indian Central and Subantarctic Mode Waters (Tomczak & Large, 1989; You, 1998; Schott & McCreary, 2001; Sokolov & Rintoul, 2002; Vianello et al., 2017). Therefore, including both ICW and SAMW separately, would probably introduce redundancy and potentially destabilize the eOMP solution. This approach allows us to capture the key source water types with outcropping history, making them better suited for tracking temperature changes associated with global warming and for assessing their influence on the Indian Ocean and its OMZs”.

3. RESPONSE TO MAJOR COMMENT #3, DISTINCTION BETWEEN DECADAL VARIABILITY AND LONG-TERM TRENDS:

We agree that comparisons between discrete periods must be interpreted with caution, particularly in a system where internal climate variability can generate oxygen fluctuations on interannual to multi-decadal timescales, however, as earlier discussed in the summary (kindly refer to summary above), several lines of evidence suggest that the observed differences between the analyzed periods likely reflect a longer-term shift in the mean state rather than short-lived variability. 

We emphasized that the long-term trend is only suggestive based on the available data and included a paragraph in this regard in the limitation section of the revised manuscript. We have also included a paragraph in the discussion section to clarify the influence of interannual and decadal variability.

4. RESPONSE TO MAJOR COMMENT #4, STUDY OBJECTIVES, SCOPE AND PRESENTATION:

Thank you for the constructive comments, we have revised the manuscript to clearly defined the scope and objectives and primary objective of the study in the last paragraph of the introduction. Secondly, we agree that the original title give the impression of a broad review rather than a focused research article. To address this, we have revised the title to: 

“Recent changes in dissolved oxygen concentration in the Indian Ocean Oxygen Minimum Zones driven by reorganization of water masses”

The description of the eOMP method has been expanded and clarified. We now explicitly define properties A and B as oxygen and nutrient tracers used to represent remineralization processes. The choice of parameter weights has been more thoroughly justified with reference to prior applications in the Arabian Sea and Indian Ocean (e.g., Acharya and Panigrahi, 2016). Additional explanation/justification has been added on the selection of source water types.

The spatial extents of the Arabian Sea (AS) and Bay of Bengal (BoB) OMZs, as well as the portions of the transects used to estimate OMZ oxygen budgets has been indicated using Figure 2.

RESPONSE TO SPECIFIC COMMENT

5. RESPONSE TO SPECIFIC COMMENT P4, LINE 111:

We appreciate the referee’s comment on the placement of Section 1.1. and will redesign the text accordingly.

6. RESPONSE TO SPECIFIC COMMENT P6, LINE 189:

As earlier discussed in  response to major comment #1, we performed oxygen variability test, using higher and lower oxygen scenarios. Our conclusions remain valid. We would revise the text to ensure consistency. 

7. RESPONSE TO SPECIFIC COMMENT P11, LINE 310:

To defined the source water types, we used the World Ocean Atlas (WOA) annual statistical mean data, which represent average of all unflagged interpolated values at each standard depth level for each variable in each ¼° or 1° square which contains at least one measurement for the given oceanographic variable, this often offers higher accuracy and reliability compared to extrapolated data.  

We agree that oxygen and nutrient data in some Indian Ocean source regions, particularly the Red Sea and Persian Gulf, are sparse /missing in the WOA Database. However, further analysis of WOA “number of observations” data shows high density of observations in the region where PGW and RSW were defined, as well as other source water types.

We have included the text below in the methods section of the revised manuscript to clarify our methods.

“Water-mass properties were derived from World Ocean Atlas (WOA) annual statistical mean, represent average of all unflagged interpolated values at each standard depth level for each variable in each ¼° or 1° square which contains at least one measurement for the given oceanographic variable. We acknowledge that data interpolation can add some uncertainty. However, analysis of WOA “numbers of observation” dataset shows high density of observations in regions where the SWTs were defined.”

8. RESPONSE TO SPECIFIC COMMENT P12, TABLE 2: 

We agree that the PGW used in the original analysis represents a higher nutrient form of PGW because it was calculated closer to the Arabian Sea. We have revised the definition of PGW parameters to better reflect outflow properties. We  re-defined the properties of PGW using the World Ocean Atlas (WOA)  data around the Oman slope at depth between 100-300m, latitude between 23.6 -27.1°N and longitude 55.5 -57.5°E according to the PGW pathway described by Tchernia (1980).  Defining the properties around this region helps us capture climate driven changes in the source region and outflow properties as observed in the increase in temperature value. Defining PGW properties closer to their entry regions therefore provides a more realistic representation of the oxygen and nutrient characteristics that actually ventilate the Indian Ocean/OMZs. (We have updated Table 2 accordingly.). Kindly refer to summary above on characterization of source water types.

9. RESPONSE TO SPECIFIC COMMENT P13, TABLE 3:  

As earlier discussed in the summary above, we have adjusted the region where RSW and PGW were defined. The defined source water types represent slightly modified primary water masses. This choice is intentional as properties defined here therefore represent the ventilating PGW and RSW as they actually enter and influence the Arabian Sea OMZ. We have revised the methods section of the manuscript to clarify source water types selection and definition.

We also agree some SWT definitions were too shallow, we have therefore revised the definitions of Arabian Sea High Salinity Water (ASHSW), Subtropical Surface Water (STSW) and Bay of Bengal Water (BoBW) based on existing literatures. Table 3 has been adjusted likewise.

10. RESPONSE TO SPECIFIC COMMENT LINES 373–375: 

 We thank the referee for pointing out this ambiguity. We agree that the phrase “similar climate anomalies” was unclear. In the revised manuscript, we now explicitly discuss the role of inter-annual and decadal variability between the two periods, with specific reference to ENSO and Indian Ocean Dipole (IOD) events. 

11. RESPONSE TO SPECIFIC COMMENT P14, LINES 385–387:  

As earlier discussed in response to major comment #1, we performed oxygen variability test, using higher and lower oxygen scenarios and our conclusions remain valid. With respect to nutrients, changes are likely to be much smaller over the considered timescale as also explained earlier.

12. RESPONSE TO SPECIFIC COMMENT P18, LINE 501:  

Our results reflect the relative contribution of sourced waters in the Indian Ocean interior as they were in 1995 and ~2016. Changes of the relative contribution of sourced waters have been derived by comparing the results from 1995 and ~2016. Our explanation of these changes refers - as indicated by the reviewer - first of all on processes in upper ocean: an enhanced inflow of STSW balancing the reduced contribution by the Indonesian Throughflow, PGW etc. The enhance contribution of deep water would primarily be seen as a consequence of the redistribution of water masses in the upper part of the water column. We will describe this better in the revised manuscript.

13. RESPONSE TO SPECIFIC COMMENT P20, LINES 501–541:  

We agree that comparisons between discrete periods does not indicate a trend, However, as previously explained in the response summary (refer to summary above), Several lines of evidence suggest that the observed differences between the analyzed periods likely reflect a longer-term shift in the mean state rather than short-lived variability.

We have revised the paragraph in line 540-541 as well as other areas of the manuscript to emphasize the suggestive nature of the long-term trend and included the text below in the Limitation section of the revised manuscript.

“This study is based on comparisons between discrete observational periods rather than a continuous time series, which limits the ability to formally quantify trends. Nevertheless, the data analyzed represents opposite trend in the OMZ development and our results align with independent observational and modeling studies, thus suggesting that they reflect a longer-term shift in the mean state rather than short-term variability.”