| 12 Nov 2025
Selective accumulation of dissolved organic matter in the sea surface microlayer: Insights from CDOM and FDOM characterisation at a Mediterranean coastal site
Abstract. The sea surface microlayer (SML) is an extremely thin (~1 mm) boundary between the ocean and the atmosphere, forming a dynamic microenvironment that regulates air–sea gas exchange. Owing to its unique position, the SML is enriched with surface-active and hydrophobic organic compounds, typically more concentrated than in the underlying water (ULW), and can therefore modulate gas transfer across the interface. Despite its importance in ocean–atmosphere interactions, the processes governing its functioning remain insufficiently understood. This study investigates the biogeochemical coupling between the SML and ULW by examining the dynamics of dissolved organic matter (DOM)—including its chromophoric (CDOM) and fluorescent (FDOM) fractions—at a coastal Mediterranean site from 2016 to 2017. Twenty-two paired SML–ULW samples and fourteen rainwater samples were analyzed for dissolved organic carbon (DOC), UV–visible (250–700 nm) absorption spectra, and 3D fluorescence excitation emission matrices (EEMs). The SML was consistently enriched in DOC, CDOM, and FDOM relative to the ULW throughout the study. Enrichment factors (EFs) for long-wavelength absorption coefficients (a300, a370) exceeded 5. Relationships between a300 and the spectral slope (S275-295) indicated photodegradation in both layers, though more pronounced in the ULW. In the SML, photodegradation effects appeared to be partially counterbalanced by in situ production or aggregation of hydrophobic, optically active, higher-molecular-weight material. Parallel Factor Analysis (PARAFAC) identified four FDOM components: two humic-like (A/C, A/M) and two protein-like (T, B). Terrestrial humic-like (A/C) and tryptophan-like (T) fluorophores were dominant and strongly enriched in the SML (EF > 4). By introducing an FDOM/CDOM index, a decoupling between fluorescent and non-fluorescent chromophoric organic fractions was revealed: the SML exhibited higher fluorescence in the UV-C/UV-B regions (A, B, T peaks) but lower fluorescence in the UV-A/visible region (C peak), suggesting selective accumulation of absorbing but non-fluorescent CDOM. Potential drivers of this decoupling include biological transformations, rapid microlayer reorganization, and atmospheric inputs. Rainwater showed DOC concentrations and absorption features comparable to the SML but distinct fluorescence characteristics. PARAFAC modeling of rainwater did not resolve the tryptophan-like fluorophore and revealed blue-shifted humic-like components, consistent with photochemically aged, low molecular weight DOM of mixed marine–terrestrial origin. Overall, the results indicate that photodegradation, biological activity, rapid molecular reorganization, and atmospheric deposition collectively shape the wavelength-dependent enrichment of CDOM and FDOM in the SML, highlighting its active role in air–sea biogeochemical exchange.
General comments
The authors have presented a study that investigates sea-surface microlayer dynamics over one seasonal cycle during 2016-2017, in a coastal area of the eastern Mediterranean Sea. Specifically, the authors investigate biogeochemical de/coupling of the surface microlayer and corresponding underlying water, with comparison to rainwater samples from the study area, and quantify DOC, CDOM, FDOM and Chl-a, with the addition of introducing a novel FDOM/CDOM index.
The data presented are likely to be of interest to a broad biogeochemical audience, and the study itself is within the scope of BG, but I find the data interpretation lacks cohesion. The scientific question(s) that the study aims to answer are unclear and not outlined in the manuscript, and the Discussion section does not sufficiently contextualise the findings. In my opinion, the Conclusion section also does not fully explain the importance of the study or how the findings fit within the context of other published literature. The final concluding sentence (Lines 616-618) captures the main problem; that the study does not develop an existing or present a new concept effectively. This is not to say that this study cannot present a novel concept, but the presentation falls short of making such clear and the data are not structured as a narrative.
This study has the potential to contribute meaningfully to our understanding of SML dissolved organic matter dynamics, and I commend the authors’ efforts in developing the FDOM/CDOM index. However I suggest a reworking of the statistical analyses and presentation of data to ensure effectual communication of the narrative presented, which seems to get lost at the manuscript progresses. The consistent decoupling of some parameters between the SML and ULW is of particular interest, as is the different composition of rainwater to the SML. It is widely accepted that the primary source of organic matter to the SML is from the ULW, and studies such as this that demonstrate additional nuance to SML organic matter sources and dynamics are invaluable to our understanding of spatiotemporal variability in SML dynamics. If this is the authors’ intended message, then it is lost in the detail.
Specific comments
1. References
The number of references is sufficient, but the number of recent references (i.e. within the last 10 years) is minimal (28 out of 102 citations). Considering that SML research is a very active global field and to enable relevant discussion and contextualisation of this study, I would expect to see a greater number of recent studies included in this study’s literature review and discussion. Suggestions, but not an exhaustive list, of recent studies that explore CDOM (and/or FDOM or surfactants or other OM) dynamics in the SML that are missing from this study:
2. Data and Statistical Analyses
A more unified approach to data analysis would benefit this study; currently there is a combination of correlation analysis, regression analysis, and Mann-Whitney U and Wilcoxon signed-rank tests, which appear to hunt for significance; with specific reliance on p-values, as these are all that are stated in text, without further scrutiny of the data. The p-values alone tell the reader nothing without f/t-statistics or the specific test used. I suggest using GLMs/GAM to include multiple variables in one model and test which have an effect on a response variable. This would streamline the narrative and ensure robustness of the suggested SML dynamics.
Additionally, with the current text presented, the authors negate from reporting data appropriately, for example, average values throughout the manuscript are reported without standard deviations, there are inconsistent spaces between values and units, and where p-values are stated there is no further information about the test being reported in text.
3. FDOM/CDOM index
While the development of the novel FDOM/CDOM index is a strength of the paper the authors could improve the robustness of their suggested measure by including some validation steps. Given that the index is based on both modelled and transformed data (e.g. the normalisation (would also be good to know what exact transformation was used here)) the ratio of FDOM/CDOM might not vary linearly. Can the authors provide some evidence that the ratio is robust to any transformations used? You could include compare the transformed and untransformed data. It would also be useful to see a comparison of index values generated from the raw data to cross validate the smoothed modelled data. Providing these validation checks for the novel index will make this a much more robust and useful value/measure for the field and result in a more highly cited, impactful paper.
4. Manuscript narrative flow
The manuscript as whole seems to lose steam as it progresses. The authors have evidently undertaken a considerable amount of work for this study, and as such the manuscript is lengthy, but it should be condensed. Appropriate structuring of a clear research question would help select where condensing text will be appropriate, including which data to showcase and which to use as contextual data. For example, there is one mention of FDOM in the Introduction (Line 76), but FDOM makes up a considerable portion of the manuscript; ultimately the FDOM portion, with the novel FDOM/CDOM index, could be a standalone and potentially more impactful manuscript. Further to this, Chl-a data are presented, but not mentioned in the Introduction or Conclusion sections. There is a substantial portion of the Introduction that talks about surfactants and air-sea gas exchange, and while these parameters are central in SML studies they are not measured in this study. Consequently, the narrative and flow is not cohesive, and the manuscript would benefit from restructuring, with a clear focus and conclusions.
Technical corrections / Line-by-line
L14: Some surfactants are hydrophobic yes, but some surfactants are amphiphilic; the SML is made up of a complex gelatinous matrix of particulate and dissolved surface-active organic matter.
L15-16: The typical enrichment of surface-active substances in the SML does not automatically equate to them being a control on air-sea gas transfer. More their specific physico-chemical properties that impede molecular diffusion across the interface.
L18: ‘including’ is misleading, as CDOM and FDOM were the only DOM fractions presented in this study.
L40-41: This study does not quantify any air-sea gas exchange parameter. I suggest rephrasing this sentence to be representative of what this study does highlight, such as microscale SML OM dynamics. Outlining clear research questions will help to address this.
L45: The SML is the boundary interface between the atmosphere and ocean, it does not 'represent' the boundary.
L47: I would lean toward saying ‘historically’ rather than ‘generally’. The general thinking now is that the composition of the surfactant pool in the SML changes with windspeed, with in/soluble surfactants dominant at different thresholds.
L49: What are the authors classifying as ‘high wind speeds’ and how was this threshold determined?
L53: See Specific Comments re updating/expanding with recent studies.
L58: 15-50% reduction relative to what?
L66: I advise caution with the use of ‘primarily’ here, and suggest rephrasing.
L74-82: References to support these statements?
L76: This is the only mention of FDOM throughout the whole Introduction section. Given the prevalence of FDOM throughout the remainder of this study, I suggest integrating an introduction of FDOM into this section.
L98: Yes, but where there are slicks there is more OM overall. Have any of the studies cited found a correlation between CDOM or FDOM and surfactant concentrations in the SML?
L99-101: Agreed, but there have been more studies within the last 10 years, which are missing from this study that have explored this (See Specific Comments). I suggest integrating such studies.
102-107: The authors have introduced the SML and CDOM, but are yet to explain why they are undertaking this study specifically. What question(s) will it answer and/or what value will it add to the field? Outlining this clearly will help to strengthen the structure of the Discussion and Conclusion sections.
L117-119: Figure 1: The map area is very large relative to the study area. I suggest zooming in further to the study area so that it is more obvious to the reader, so that individual sampling sites can be marked. Also, use a colour that is easier to see clearly than yellow.
L123: Suggest ‘were’ instead of ‘have been’.
L128: Should be ‘ULW’.
L130: Table S?
L133: This is the first mention of Chl-a so far in the manuscript. Please also integrate into Introduction.
L134-135: Were only the Chl-a samples 0.22 µm filtered, or where all DOC, CDOM, FDOM and Chl-a filtered? This reads like only the Chl-a. Rephrase for clarity. If the DOC, CDOM and FDOM were not filtered then the parameters measured are of the total, not the dissolved phase.
L137: Should be ‘caps’ not ‘cups’.
L139: Is the 10% HCl stated by volume or concentration? Molar concentration would be clearer.
L150-151: Why was a lower volume filtered for the SML than the ULW? Is this an established or tested technique? Please outline which here, and cite accordingly.
L167-175: No backscatter correction done? See Kitidis et al. (2006); DOI: 10.1016/j.dsr2.2006.05.009
L176: The absorption coefficient in marine waters is often calculated using 300 nm (see references already cited and those listed above in Specific Comments) but also often using 350 nm. Can you justify further why you selected 300 nm for this study area?
L185: Should be ‘275-295’ not ‘275-275’.
L218: I suggest a stand-alone section for the FDOM/CDOM Index method description.
L239: EF 1 does not indicate depletion; EF < 1 indicates depletion.
L241-249: See above in Specific Comments section.
L260-261: End of sentence trails off, can you also be quantitative for the April-May period?
L266: An EF of 1.2 is not typically considered minimal. This would be an excellent opportunity to contextualise this with other literature.
L267: Is this unexpected considering that two phytoplankton blooms are expected per year, as outlined earlier in the manuscript?
L274: Figure 2c: The decoupling of DOC between the SML and ULW is very interesting and certainly worth further exploration in the text.
L275: How have you categorised a ‘low’ windspeed? And dust particles were elevated ‘relative’ to what? Were these parameters measured throughout the whole ‘study period’ or just on sampling days? Please be clear with this.
L283: Should be ‘ULW’ not ‘UWL’.
L287-289: This is where enrichment factors are useful. I suggest reporting DOC EFs in parallel to DOC concentrations.
L280-291: I disagree, there is clear decoupling evident from the seasonal cycle shown in Figure 2c, and from the immediately previous sentences in this paragraph.
L290-293: The decoupling is evident from Figure 2c, and of much more interest and inherent value to the narrative of this study than it is given here. I suggest a more thorough review of the SML DOC literature.
L297: What statistical test was applied here?
L299-301: Figure 3: The colour bar would be more informative to show seasonality of samples, while a 1:1 EF line can be plotted on each figure (above shows SML accumulation and below show depletion). Otherwise the two EF colour bars should be of the same scale.
L307: Is the correlation between DOC and a300 shown?
L310-313: This has been found in other studies – some already cited in this manuscript and some listed above in Specific Comments.
L314-315: Correlation does not indicate that one variable influences another variable, only that they covary.
L315-316 and Table 1: The text refers to correlation analysis while the Table caption refers to regression analysis. Which is it?
L317 and L318: DOC is not bulk DOM.
L318-319: Again, referring to recently published studies here would be useful and informative. It could also be that the processes in the SML and ULW are different, or that the DOC and CDOM pools are compositionally different in the two layers to begin with.
L326: Please explain how they are ‘disproportionately’ enriched? In context of other studies?
L343-346: I don’t understand what these sentences are saying, please clarify.
L346-347: Studies have looked at CDOM photodegradation in the SML specifically, or shown data that suggest a difference between the SML and ULW: e.g. Rickard et al., 2022; Sabbaghzahdeh et al., 2017.
L347-349: There is considerable overlap between the SML and ULW (i.e. S275-295 > ~0.025 on Figure 4). Can you provide evidence to support that they are in fact different, to support your suggestion of ‘other processes’? The fact that you have fit one exponential decay equation to both depths together is also contradictory to this statement.
L369: Slicks were only observed? Or actually sampled from? It would be informative to mark these samples of the figures presented in the manuscript, and highlight if any of the extreme EFs reported were (or were not) from slick sampling.
L425: There is no temporal component presented in Fig. 7, so ‘during most months’ cannot be ascertained by the reader. As with Fig. 3, a seasonal colour bar would be more informative than an EF colour bar. There can be no direct comparison between specific samples/components if the temporal component is not presented.
L425-438: The language in this paragraph is imprecise, with frequent use of non-specific words, such as ‘high’, ‘considerably lower’, ‘exceptionally strong’, without the in-text data to illustrate the narrative. Suggest condensing this paragraph, and/or including respective data to strengthen the text throughout (i.e. state value of outlier as well as sample date: L431).
L446: ‘Appear’ most enriched? They either are or they aren’t.
L449-450: So what does this suggest?
L454-457: Already described in Methods section.
Lines 467 and 470: Use of ‘appears’ – ditto above comment (L446).
L591-618: Conclusion section: The authors frequently refer to DOM instead of DOC; yes, DOC is a dominant component of DOM but the two are not interchangeable terms, and DOC was measured in this study. Please correct the terminology. Additionally, for this section as a whole the text does not indicate why any of the findings matter in the broader context; the final sentence is very general and does not indicate anything new (it is already widely accepted that the SML is an important and dynamic environmental compartment).
L686-688: Reference incomplete. 2018 paper?