the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 29 Jul 2026
Near-absence of surface zooplankton and pervasive anthropogenic particles along a zonal North Atlantic transect: a multi-vessel citizen-science image survey
Abstract. Climate policy treats carbon dioxide as the single lever on the climate. We suggest a second one has been overlooked: the marine biodiversity that builds and maintains the ocean surface microlayer (SML) and that pollution is now degrading. Water vapour, not CO₂, is the largest single contributor to the natural greenhouse effect and the strongest feedback on warming. The ocean supplies most of that vapour across the 71% of the planet it covers, and the thin biogenic lipid-and-surfactant film phytoplankton spread over the sea surface helps govern how readily it evaporates and forms aerosol. That same film gathers lipophilic "forever" chemicals, together with the microplastic and black-carbon soot they cling to, to levels far above the water beneath, where they turn toxic to the plankton that hold the ecosystem together.
An automated image pipeline sized and classified 51,547 particles larger than 20 μm in 352 georeferenced surface samples from eight sailing vessels crossing the trade-wind North Atlantic. Zooplankton were absent from 93.8% of stations. Black carbon appeared at 89.8%, microplastic particles at 69.9%, microfibres at 22.4% and phytoplankton at 81.8%; 99.2% of everything counted measured below 200 μm. A separate 110-image, GPS-tagged crossing returned zero zooplankton at every point on the track, and a plankton net towed at the surface by day and by night confirmed the scarcity directly, returning roughly 0.2 to 0.5 animals per cubic metre – far below any productive ocean. In short, this surface layer is stripped of animal plankton and steeped in combustion and plastic residue. We trace what that may mean for the microlayer, for water-vapour feedback and cloud formation, and for ocean pH, and argue that cutting pollution has to sit beside cutting carbon.
- Preprint
(2312 KB) - Metadata XML
-
Supplement
(655 KB) - BibTeX
- EndNote
Status: open (until 23 Sep 2026)
-
RC1: 'Comment on egusphere-2026-3979', Anonymous Referee #1, 17 Aug 2026
reply
-
AC1: 'Reply on RC1', Howard Dryden, 17 Aug 2026
reply
Author response to Referee Comment 1
Manuscript: egusphere-2026-3979, “Near-absence of surface zooplankton and pervasive anthropogenic particles along a zonal North Atlantic transect: a multi-vessel citizen-science image survey”
Authors: Howard Dryden and Diane Duncan
We are grateful for this review. It is detailed, it is plainly the work of someone who knows this ocean, and it has improved the manuscript already. It has also found an error in our reference list that we accept without reservation, along with several methodological limitations that we accept and will address.
It rests, in part, on an attribution of our zooplankton result to the wrong instrument, and we deal with that first because the referee’s central quantitative argument follows from it.
Our approach below is simple. Where the referee is right we say so and state what changes. Where we think he has gone wrong we give the evidence rather than the assertion. Where a question cannot be settled either way, we say that too.
0. What this survey is, and what it is not
Before the specific points, it may help to state the design intent, because several of the referee’s criticisms are answered by it and one of them is answered by his own text.
This is a citizen-science project built around what an ocean-going blue-water yacht can actually do. The sampling protocol is constrained by that: a small crew, underway, without winches, wire, laboratory space or power for depth-resolved work. Those constraints are real and we do not pretend otherwise. This survey does not replicate oceanographic sampling at depth, and we make no such claim. It cannot resolve the deep chlorophyll maximum, it does not retain picoplankton, and it does not substitute for a research cruise.
What it offers instead is coverage. Cruising yachts cross ocean that research vessels rarely visit, on tracks and at times that no funded programme covers, and they do so continuously and at negligible marginal cost. The referee makes this point for us: “Routine CPR sampling is unfortunately not carried out in this ocean area.” Nor is the AMT, which runs meridionally along about 20°W; ours is a zonal transect at right angles to it. The water we sampled is not under-described because it is uninteresting. It is under-described because nobody goes there with a ship.
One further point of context bears on the referee’s closing remarks about planning. To our knowledge this is the first survey to measure plankton and anthropogenic particles together along a zonal transect of the equatorial Atlantic, from North Africa to the Caribbean, and no prior near-surface survey of that water exists against which a protocol could have been designed. We did not know in advance what the sampling would find, and a protocol cannot be specified in full for conditions that have never been described. The sampling was also carried out by yacht skippers rather than by trained researchers, people giving their time on passages they were making anyway, and that places a hard ceiling on procedural complexity. A method that a volunteer cannot carry out reliably at sea, in a seaway, at the end of a watch, is not a method available to this project at all. Within those constraints the survey returned 352 stations across a basin, which we offer as a useful first return rather than as a finished protocol. Like any survey conducted for the first time it carries lessons, and the referee has identified several of them. The refinements set out below, namely spectroscopic confirmation, a flowmeter, procedural blanks from every crew and consistent depth terminology, are the second iteration of a method that did not exist before the first.
We would add that the referee’s own citations make this case better than we can. Simoniello et al. (2019), which he offers as a model of how citizen science should be done, is in part a study of plankton sampling from cruising yachts, and it observes that their routes “often cover tracts of ocean undersampled by traditional oceanographic cruises”, describing the conversion of ordinary yachts into in-situ monitoring platforms at a cost roughly twenty times below that of a research vessel. Hidalgo-Ruz and Thiel (2015), his other citizen-science reference, recommends this design explicitly, proposing projects “with the help of sailing clubs, where long-distance travelers can survey floating marine debris by direct observation at sea, to study the distribution, composition and degradation of marine litter in the open ocean”. We take his closing principle seriously. We also note that the two sources he cites for it endorse the approach he questions.
The trade is therefore explicit: depth resolution and analytical control are given up in exchange for spatial and temporal coverage that is otherwise unobtainable. Whether that trade yields anything useful is a fair question. It turns on whether a constrained, well-characterised measurement can be compared like-for-like against other constrained measurements of the same layer, not on whether it matches what a research vessel would have produced. We accept the referee’s closing principle without reservation: citizen science in this field has to be planned with rigour, its selectivity stated, and its comparisons made against the right baseline.
1. The zooplankton densities come from net tows, not from the 0.5 L imaging device
The referee writes that “the limited volume (0.5 litre; i.e. 0.0005 m³) of GOES device makes it highly unlikely that any single water sample will include zooplankton”, and compares it with the Continuous Plankton Recorder’s 3.2–3.8 m³ standard unit (Jonas et al., 2004), concluding that the CPR filters “6,000 to 7,000 times greater volume”.
The 0.5 L device is not the source of the zooplankton densities. The manuscript states this at line 178: “the zooplankton result comes instead from the surface net tow (Sect. 2.8, Table 2).” Section 2.8 describes that tow: a 120 µm mesh net of 0.30 m mouth diameter, towed at the surface at 1–2 knots for 10 minutes per haul, straining approximately 43,600 litres, or 43.6 m³, per haul (mouth area 0.071 m² × ~617 m towed), run in paired day and night sets.
Three consequences follow:
- The filtered volume behind the reported densities is roughly 11 to 14 times larger than the CPR standard unit, not thousands of times smaller. The comparison in the review is inverted.
- The mouth diameter is 300 mm, not the 50 mm the referee attributes to it, which changes the evasion argument materially.
- Diel vertical migration, raised by the referee as though unaddressed, is the explicit reason the tows were run by day and by night, and both sets are reported in Table 2.
On the aperture criterion specifically, we would ask the referee to apply it consistently. The Continuous Plankton Recorder he proposes as the comparison standard samples through a 1.25 × 1.25 cm aperture in its nose cone, on 270 µm silk, at about 10 m depth. Our tow net has a 300 mm mouth on 120 µm mesh at the surface; even the imaging device he criticises has a 50 mm aperture, four times the width and some sixteen times the area of the CPR’s. If a small aperture disqualified a sampler from quantifying larger, actively swimming zooplankton, it would disqualify the longest plankton time series in existence. It does not. The reason it does not is that a known and stated selectivity can still support quantitative comparison, provided like is compared with like, and that is the same reason we report our mesh, mouth diameter and filtered volume explicitly. We accept the selectivity; we do not accept that it is disqualifying here and not there.
The same principle applies to optical measurement. The Phytoplankton Colour Index, one of the CPR survey’s principal outputs, is a visual assessment of the silk’s greenness against a standard colour chart, described by the survey itself as a semi-quantitative estimate of biomass, and it exists precisely because the chloroplasts of broken cells and the smaller phytoplankton cannot be counted under the microscope. An optical, semi-quantitative index is therefore not disqualifying in itself, since one such index underpins the longest plankton record in existence. Our own counts are of a different kind, and the distinction is worth drawing. The imaging pipeline enumerates individual particles above a stated size floor within a known volume, which yields a concentration rather than an index. What is provisional in our case is the assignment of those counts to material classes, not the counting itself. We are careful about what that does and does not license. It does not excuse our own particle classification from spectroscopic confirmation, which we accept in full at point 4 below, because identifying a material is a different claim from estimating a biomass. It does mean that an objection to optical methods as such cannot be sustained.
There is a mechanical reason for the difference in what the two methods can see. The CPR is towed behind merchant vessels on their normal sailings, and Jonas et al. (2004), cited by the referee, establishes that the volume filtered falls as ship speed rises. Plankton meeting the silk at those speeds are pressed between two layers of mesh, which is the reason a colour index is needed for the fraction that survives as pigment rather than as countable cells. Our imaging step applies no filtration, no concentration and no tow of any kind. The 0.5 litre sample is a direct fill of the sample tube, and the whole of it is then examined across five apertures, so the organisms are never drawn across a mesh, never pressed between layers of silk and never subjected to the mechanical stress of a high-speed tow. They are imaged in the condition in which they were collected. The separate confirmation net, which is a different instrument, is towed at 1 to 2 knots rather than at service speed. We have not quantified specimen condition and we claim no figure for it. We will add an image-based condition score to the next campaign so that this comparison can rest on evidence rather than on inference.
The mesh comparison also bears on the particle result, and here it is decisive. The CPR silk is 270 µm. Our imaging pipeline sized and classified 51,547 particles larger than 20 µm across 352 stations, and abundance in that record is concentrated at the smallest resolvable size, with plastic fibres forming a distinct mode at a median length of 204 µm. The great majority of the particles reported in this manuscript are therefore smaller than the aperture of a CPR silk and would pass through it. We make no criticism of the CPR in saying so, since it was built for a different purpose and long predates the question. The point is only that the instrument the referee proposes as the standard could not have produced this result, which is part of why the result did not exist before now.
We would also be wary of treating any single filtered-volume figure as a fixed benchmark. Jonas et al. (2004), the source the referee cites, found that the volume filtered by a CPR falls as ship speed rises. Takahashi et al. (2006), in the same journal, found the correlation positive, attributed it to the flow meter itself, and reported reductions of up to 60% from clogging and up to 78% from clogging and ship speed combined. The CPR’s own sampled volume therefore carries a wide uncertainty, which is a further reason to compare methods by their stated selectivity rather than by a single number.
We should say plainly where this leaves the comparison, because we do not want the preceding paragraphs read as a criticism of the CPR. The two systems carry different limitations and different strengths, and we do not think either should be treated as the standard by which the other is judged. A CPR resolves taxonomy along routes that have been occupied for ninety years, which nothing else in oceanography can offer. A direct-fill imaging sample resolves the small particle fraction in water that no research vessel visits, which the CPR was never designed to do. Neither is a substitute for the other, and a direct comparison of the two obscures what each contributes. The more useful position is that different instruments extend the field of measurement in different directions, and that the marine ecosystem is described better by both together than by either alone. We would welcome the difference being taken that way, and we offer this reply in that spirit.
We do not offer any of this as a complaint. If a careful referee attributed the result to the wrong instrument, the manuscript did not separate the two methods clearly enough, and that is our responsibility to fix. In revision the net-tow method, its mesh and its filtered volume per haul will appear in the abstract and at the first mention of any zooplankton density, so that no reader can associate those numbers with the imaging device.
Two elements of the referee’s concern survive this correction and we accept them. A 120 µm mesh retains larger taxa while passing the smallest copepods and nauplii, and our filtered volume is computed from tow geometry rather than from a flowmeter. Both are already stated in the manuscript (lines 197–198); both will be given more prominence, and a flowmeter will be fitted for the next campaign.
2. The citation error is real, and we accept it
The referee states that Woodd-Walker et al. (2000), as we cite it, does not exist. He is correct, and we apologise for it. We note without complaint that the same appraisal appears in the referee’s own text as Loder and Gerdis (2013) and in his reference list as Löder and Gerdts (2015). Citations are treacherous things, and we are in no position to be smug about it.
The source we intended is Woodd-Walker, R. S., Ward, P., and Clarke, A. (2002), “Large-scale patterns in diversity and community structure of surface water copepods from the Atlantic Ocean”, Marine Ecology Progress Series 236, 189–203, doi:10.3354/meps236189, which draws on 259 samples from Atlantic Meridional Transect cruises between 60°N and 63°S and therefore does span the tropical Atlantic, rather than the 35 to 60°N band of Gallienne et al. The entry will be corrected, and every remaining reference in the manuscript has been re-checked against Crossref.
Correcting it, however, repairs only the geography, and we would rather set out the consequence than leave the referee to find it. Despite its title, that study sampled with a WP2 net through the top 200 m, so the objection the referee raises against Fernández de Puelles et al. applies to it equally. Taken together with his other two points, the position is this: none of the three sources we cited is a like-for-like surface measurement. Two integrate the epipelagic and one is a deep-sea synthesis.
On the two related points, we agree. Vereshchaka et al. (2017) is a deep-sea synthesis and is not an appropriate comparison for a surface stock. Fernández de Puelles et al. (2019) integrate to 200 m; we described this as the “epipelagic layer” (line 96), which is accurate but does not make the comparison like-for-like, because that integration includes the deep chlorophyll maximum. In revision the comparison will be restricted to surface-sampled datasets wherever such data exist, and where an integrated value is used it will be stated as an upper bound on the surface concentration rather than as a comparable measurement. We note in passing that the difficulty of finding a genuine surface comparison for this water is itself the coverage gap described in Section 0. The surface mesozooplankton stock of the tropical North Atlantic has scarcely been measured as a surface stock, which is why the comparison is hard to make and why we think it is worth making.
3. Oligotrophy, the deep chlorophyll maximum, and what we do and do not claim
We accept the substance of the referee’s framing point. The North Atlantic subtropical gyre is strongly stratified and nutrient-depleted; pico- and nanophytoplankton dominate carbon fixation there, at 84% of primary production in the western tropical Atlantic on the referee’s own source (Tilstone et al., 2017); the chlorophyll maximum lies well below our sampling depth (Rees et al., 2015); and our sampling reaches roughly the top 1 m. We did not sample the DCM.
One detail of that framing needs correcting, and we raise it only because the depth bears on the comparison. The referee places the deep chlorophyll maximum at 50 to 200 m and states that productivity there increases by one to two orders of magnitude, citing Rees et al. (2015). That paper is a seven-page retrospective of the AMT programme. It reports no primary production data and contains no such statement. Its Figure 2 places the tropical Atlantic chlorophyll maximum at roughly 45 to 100 m, at values a few times the surface concentration rather than ten to a hundred times it. The maximum is real, it lies below our sampling, and we accept the point. Its depth and its magnitude are both smaller than the review states.
On phytoplankton the position needs stating precisely. The imaging device does count a phytoplankton class. What is reported in the manuscript is this: pigmented aggregates present at 81.8% of stations, median 10 per litre (Sect. 2.11). Every count in this survey is subject to a stated size floor of 20 µm (line 236). Pico- and nanophytoplankton fall below that floor, so the referee is right that they are neither retained nor identified by our method.
We are not certain, though, what the criticism amounts to, and we would welcome the referee’s clarification. If the point is that we drew conclusions about productivity from a size-limited surface count, we accept it without reservation, and the rewrite described below removes them. If the point is that a 20 µm floor is itself a defect, we do not follow the argument. Every plankton method carries a size floor, the CPR’s is 270 µm, and ours is declared in the manuscript rather than uncovered in review. A stated selectivity is a parameter of a method and not a fault in it, which is the same principle we applied to the CPR at point 1 above.
The consequence is a limit on interpretation rather than an absence of data, and we will make it explicit. Our phytoplankton class is an optical count of pigmented particles of 20 µm and larger. It is not a measure of phytoplankton biomass, of chlorophyll, or of primary production, and no inference about productivity in this region should be drawn from it. What it can support is a comparison of the ≥20 µm particle field between stations along the transect, and that is the use we make of it.
We would note that we are aware of the region’s oligotrophy: lines 94–96 frame the comparison explicitly against oligotrophic subtropical gyres, name the Malaspina circumnavigation as the comparison, and record that its 300 µm net undercounts the smallest copepods. The claim under test is not “productivity here is low”, which is not in dispute and would not be worth reporting. It is that the surface mesozooplankton stock is far below what the surface literature reports for comparably unproductive water. We can settle this now, and the answer is that there is no surface data. The thousandfold figure comes from setting our surface value of 0.2 to 0.5 animals per m³ against epipelagic values of upward of 200 per m³. That sets a surface stock against a 200 m integration, so it is not like-for-like, and we cannot make it like-for-like because a surface-only mesozooplankton measurement for this water does not exist. None of the three sources we cited provides one, and we have not found one elsewhere. In revision we will say precisely that, give the ratio against integrated values with the integration named in the same sentence, and withdraw the thousandfold figure as a statement of surface deficit. The absence of a surface baseline is not a weakness in the argument. It is the gap the survey was designed to fill.
On line 419–420 the surrounding paragraph concerns Prochlorococcus, numerically the most abundant photosynthetic organism in the ocean and the organism whose sensitivity to plastic leachates (Tetu et al., 2019) motivates the passage, rather than the chlorophyll maximum. But “where the ocean’s most abundant photosynthetic organisms live” reads as a claim about where production occurs, and in this region that is subsurface. The sentence will be rewritten to say more clearly what it means: that the near-surface layer we sampled carries a high particle load and is inhabited by picocyanobacteria, without any implication that it is the region’s productivity maximum.
4. Particle identification
Procedural blanks were run, by the lead author, and they showed no obvious contamination (lines 226 to 229). What we did not do was ask each volunteer crew to run blanks of their own, and that was a deliberate design decision rather than an oversight. Every additional procedure asked of a skipper underway is a procedure that will be performed inconsistently or skipped altogether, so we minimised what the crews were asked to do and relied instead on blanks run under controlled conditions together with a single common microscope design across all vessels. There is also a practical difficulty with a blank at sea that is worth naming, because it bears on how useful per-crew blanks would have been. A procedural blank is only as good as the water used to make it, and a cruising yacht carries no certified particle-free supply. The water available is tank water that has passed through the vessel’s own plumbing and filters, and it may carry fibres and particles of its own, so a blank prepared from it risks adding error rather than measuring it. Any remedy also has to fit the vessel, and stowage on a cruising yacht is scarce, so carrying prepared blank water for every station is not practical. Two measures that do fit are written into the next campaign. The first is a small number of sealed blank-water samples per voyage rather than per station. The second is an exposure blank taken at the point of analysis, a prepared slide left open beside the working sample, which requires no water at all, occupies no space, and tests the airborne fibre pathway that is the main contamination risk we identified in the manuscript.
The manuscript already flags the limitation (lines 226–231): the optical “black carbon” class awaits spectroscopic confirmation and the plastic-versus-phytoplankton partition is provisional. The difficulty of identifying microplastics reliably by eye is well documented (Löder and Gerdts, 2015), and his point stands. We would note, for the record rather than in objection, that the 1.4% confirmation rate he quotes from that chapter comes from North Sea sediment in which 96% of the particles examined were quartz sand grains left behind by an incomplete density separation, an outcome the authors themselves describe as surprising and use to argue for routine FTIR. Sand is not a candidate class in an open-ocean surface sample. We expect our own confirmed fraction to be higher, and we will report it whatever it proves to be.
We should be straight about what can and cannot be done now. Spectroscopic confirmation cannot be completed for this manuscript. It needs material collected under a protocol designed for it, which means the next sampling campaign rather than a reanalysis of this one. We would rather say that than promise a measurement we cannot deliver in revision.
What we will do in revision is withdraw the inference that confirmation would have supported. The optical classes will be labelled as optical classes throughout, with no claim about material composition, in the abstract and in every table and figure caption as well as in the text. Sargassum fragments will be named as a candidate contributor rather than assumed absent, which is the referee’s suggestion and a good one. FTIR or Raman confirmation is written into the next campaign, and the confirmed fraction will be reported whatever it turns out to be.
5. The meaning of “surface”
Accepted. The manuscript does use “surface” for the microlayer, for the top metre and, in the comparison literature, for the upper 200 m, and the referee is right that steep physical, chemical and biological gradients sit between those layers in stratified tropical water. In revision we will define three terms at first use, the microlayer (<1 mm), the near-surface layer (0 to 1 m, which is the layer we sampled), and the epipelagic (0 to 200 m), and we will use them consistently, including in the figure captions and table headings.
6. The microlayer, evaporation and climate
For context, the manuscript as first drafted was confined to the sampling; the microlayer and evaporation material was added when an earlier external referee asked us to widen the scope beyond the survey itself. We note that only to explain how it came to be here.
On the substance, we ask the referee to distinguish two claims, because we think the review conflates them, and we accept part of what he says.
What we do not claim. We do not have a magnitude for the effect of microlayer condition on ocean evaporation, and we are not aware of a published quantification of it. To the extent that our framing implies a quantified climate effect, that is an overstatement and we will correct it. What we would maintain is narrower: a surfactant-rich microlayer is expected to alter the rate of evaporation, and with it the humidity supplied to the atmosphere above, because that is what organic films do at an air-water interface. The major syntheses of microlayer science quantify its suppression of gas exchange in detail and offer no comparable treatment of the water flux (Engel et al., 2017; Wurl et al., 2017), so whether the effect at sea is large enough to matter for climate remains unestablished, in our work and in the wider literature.
What the gap itself tells us. We would go further and offer that absence as a finding rather than as an embarrassment. Every water molecule that evaporates from the 71% of the planet’s surface covered by ocean passes through the microlayer, and that same layer is known to suppress the transfer of gas across it by a large fraction. That a layer of such position, and of such demonstrated influence on one exchange, has attracted no quantitative study of its influence on the other is a gap worth naming rather than passing over in silence. We would rather the manuscript be read as flagging that neglect than as claiming to have measured its consequences.
What we would resist. The referee describes the effect as “trivial, with temperature, wind-speed and ocean area being much more important by several orders of magnitude”, while also stating that “detailed evaluation of this aspect of the MS is beyond the expertise of this reviewer”. We accept the second statement in good faith and note that it applies to the first: the several-orders-of-magnitude figure is an expectation, not a measurement, and the same objection we accept against our framing applies to it.
Why we regard the question as legitimate rather than idle. The retardation of evaporation by insoluble organic monolayers is established physics, studied since Langmuir and Schaefer and applied, with well-documented practical difficulty, to water storages; the documented weakness of those films is precisely that wind and wave action disrupt them, which is the referee’s point and a fair one. Separately, and more directly relevant, natural surfactants in the sea-surface microlayer are now known to suppress the in-situ CO₂ transfer velocity by a large fraction across the Atlantic (Pereira et al., 2018). Gas transfer and evaporation are not the same resistance and we do not conflate them. But a demonstration that natural marine films exert a large control on one exchange across the air-sea interface is the reason we consider the question about the other exchange worth asking, rather than answered in advance. There is a further reason, and it bears directly on the referee’s point about wind. He treats wind and wave action as the thing that destroys an organic film, and for a monolayer on a reservoir that is correct. At sea it is also the mechanism by which the microlayer enters the atmosphere. Bubbles bursting through the microlayer inject its organic material into the marine boundary layer as sea spray aerosol, and that aerosol carries a biogenic organic fraction which contributes to the cloud condensation nuclei over the remote ocean (O’Dowd et al., 2004). Whether this amounts to a regulating feedback on cloud albedo is disputed (Quinn and Bates, 2011), and we rest nothing on the stronger version of that claim. What it does establish is that the condition of the microlayer already has a documented pathway to cloud formation, so a question about its influence on the water flux is not an idle one.
What we propose. The retardation of evaporation by organic films at an air-water interface is established physics. That the microlayer suppresses gas transfer at sea by a large fraction is measured (Pereira et al., 2018). That microlayer organics reach the atmosphere as aerosol and contribute to marine cloud condensation nuclei is measured (O’Dowd et al., 2004). None of that is hypothesis. What is hypothesis, and the only part of it that is, is the magnitude: how much the condition of the microlayer alters the water flux in the open ocean, and whether that is large enough to matter for climate. That single question will be labelled as open, with the observation that would settle it, namely paired eddy-covariance latent heat flux measurements inside and outside natural slicks set against measured surfactant coverage. The manuscript should not be read as claiming that this survey measured a climate effect, because it did not. What it sampled was particles and plankton. We would not accept, however, that the microlayer is climatically inert, and the manuscript says why. That layer already exerts two documented influences on climate. It suppresses the transfer of carbon dioxide across the ocean surface by a large fraction (Pereira et al., 2018), and it supplies the organic material that helps seed cloud over the remote ocean (O’Dowd et al., 2004). A layer through which every molecule of ocean evaporation must pass, which already throttles one gas flux and already feeds cloud formation, is not plausibly without effect on the water flux as well. What is genuinely unknown is the extent of that effect, and the extent is what the manuscript will flag.
We do not propose to remove this material, and we would ask the referee to consider why. It is the reason the survey was designed to sample the near-surface layer rather than any other. The pathway from the microlayer to cloud formation is documented rather than speculative. And the absence of any quantification of the water flux, in a literature that has quantified the gas flux carefully, is a finding this manuscript is well placed to record. A full quantitative treatment will need dedicated flux measurements and will be a separate paper, which we intend to pursue. Stating the question, and stating that nobody has yet answered it, does not require that paper to exist first.
7. A single survey cannot establish a trend
We did not claim that this survey represents a season, and we did not claim a measured trend. Where the manuscript sets our counts against older surveys it says the contrast “may reflect a genuine change in the equatorial Atlantic over the intervening two decades rather than sampling differences alone”, which is a hedge and not an assertion.
The limitation behind the referee’s point is nonetheless real and we accept it. One occupation cannot separate seasonal variability, episodic enrichment from dust deposition or storms, and long-term change, and nothing in the manuscript should be read as separating them. In revision the results will be presented as a baseline against which repeat occupations can be compared, and every comparison with a historical survey will be labelled as such rather than as a measured change. The long-term programmes the referee names, the AMT (Rees et al., 2015) and BATS (Madin et al., 2001), are the context against which a repeat occupation should eventually be judged. Neither samples this zonal transect, which is the gap the survey was designed to address and the reason a repeat occupation is planned.
References cited in this response
Engel, A., Bange, H. W., Cunliffe, M., Burrows, S. M., et al.: The ocean’s vital skin: toward an integrated understanding of the sea surface microlayer, Frontiers in Marine Science, 4, 165, https://doi.org/10.3389/fmars.2017.00165, 2017.
Fernández de Puelles, M. L., Gazá, M., Cabanellas-Reboredo, M., Santandreu, M. del M., Irigoien, X., González-Gordillo, J. I., Duarte, C. M., and Hernández-León, S.: Zooplankton abundance and diversity in the tropical and subtropical ocean, Diversity, 11, 203, https://doi.org/10.3390/d11110203, 2019.
Gallienne, C. P., Robins, D. B., and Woodd-Walker, R. S.: Abundance, distribution and size structure of zooplankton along a 20° west meridional transect of the northeast Atlantic Ocean in July, Deep-Sea Research II, 48, 925–949, https://doi.org/10.1016/S0967-0645(00)00114-4, 2001.
Hidalgo-Ruz, V. and Thiel, M.: The contribution of citizen scientists to the monitoring of marine litter, in: Marine Anthropogenic Litter, edited by: Bergmann, M., Gutow, L., and Klages, M., Springer, 429–447, https://doi.org/10.1007/978-3-319-16510-3_16, 2015.
Jonas, T. D., Walne, A., Beaugrand, G., Gregory, L., and Hays, G. C.: The volume of water filtered by a Continuous Plankton Recorder sample: the effect of ship speed, Journal of Plankton Research, 26, 1499–1506, https://doi.org/10.1093/plankt/fbh137, 2004.
Löder, M. G. J. and Gerdts, G.: Methodology used for the detection and identification of microplastics: a critical appraisal, in: Marine Anthropogenic Litter, edited by: Bergmann, M., Gutow, L., and Klages, M., Springer, 201–227, https://doi.org/10.1007/978-3-319-16510-3_8, 2015.
Madin, L. P., Horgan, E. F., and Steinberg, D. K.: Zooplankton at the Bermuda Atlantic Time-series Study (BATS) station: diel, seasonal and interannual variation in biomass, 1994–1998, Deep-Sea Research II, 48, 2063–2082, https://doi.org/10.1016/S0967-0645(00)00171-5, 2001.
O’Dowd, C. D., Facchini, M. C., Cavalli, F., Ceburnis, D., Mircea, M., Decesari, S., Fuzzi, S., Yoon, Y. J., and Putaud, J.-P.: Biogenically driven organic contribution to marine aerosol, Nature, 431, 676–680, https://doi.org/10.1038/nature02959, 2004.
Pereira, R., Ashton, I., Sabbaghzadeh, B., Shutler, J. D., and Upstill-Goddard, R. C.: Reduced air-sea CO₂ exchange in the Atlantic Ocean due to biological surfactants, Nature Geoscience, 11, 492–496, https://doi.org/10.1038/s41561-018-0136-2, 2018.
Quinn, P. K. and Bates, T. S.: The case against climate regulation via oceanic phytoplankton sulphur emissions, Nature, 480, 51–56, https://doi.org/10.1038/nature10580, 2011.
Rees, A., Robinson, C., Smyth, T., Aiken, J., Nightingale, P., and Zubkov, M.: 20 years of the Atlantic Meridional Transect (AMT), Limnology and Oceanography Bulletin, 24, 101–107, https://doi.org/10.1002/lob.10069, 2015.
Simoniello, C., Jencks, J., Lauro, F. M., Loftis, J. D., et al.: Citizen-science for the future: advisory case studies from around the globe, Frontiers in Marine Science, 6, 225, https://doi.org/10.3389/fmars.2019.00225, 2019.
Takahashi, K., Kuwata, A., Saito, H., and Ide, K.: Continuous Plankton Recorder flow rates revisited: clogging, ship speed and flow meter design, Journal of Plankton Research, 28, 847–855, https://doi.org/10.1093/plankt/fbl020, 2006.
Tetu, S. G., Sarker, I., Schrameyer, V., Pickford, R., Elbourne, L. D. H., Moore, L. R., and Paulsen, I. T.: Plastic leachates impair growth and oxygen production in Prochlorococcus, the ocean’s most abundant photosynthetic bacteria, Communications Biology, 2, 184, https://doi.org/10.1038/s42003-019-0410-x, 2019.
Tilstone, G. H., Lange, P. K., Misra, A., Brewin, R. J. W., and Cain, T.: Micro-phytoplankton photosynthesis, primary production and potential export production in the Atlantic Ocean, Progress in Oceanography, 158, 109–129, https://doi.org/10.1016/j.pocean.2017.01.006, 2017.
Vereshchaka, A. L., Abyzova, G., Lunina, A., and Musaeva, E.: The deep-sea zooplankton of the North, Central, and South Atlantic: biomass, abundance, diversity, Deep-Sea Research II, 137, 89–101, https://doi.org/10.1016/j.dsr2.2016.06.017, 2017.
Woodd-Walker, R. S., Ward, P., and Clarke, A.: Large-scale patterns in diversity and community structure of surface water copepods from the Atlantic Ocean, Marine Ecology Progress Series, 236, 189–203, https://doi.org/10.3354/meps236189, 2002.
Wurl, O., Ekau, W., Landing, W. M., and Zappa, C. J.: Sea surface microlayer in a changing ocean: a perspective, Elementa: Science of the Anthropocene, 5, 31, https://doi.org/10.1525/elementa.228, 2017.
Summary of changes
- Net-tow method, mesh and filtered volume per haul stated in the abstract and at first mention of any density; imaging device and net tow separated unambiguously throughout.
- Woodd-Walker et al. reference corrected to Woodd-Walker, Ward and Clarke (2002), MEPS 236, 189–203, with a note that this repairs the geography only, since that study also integrates the top 200 m; all references re-verified against Crossref.
- The comparison restated as a ratio against depth-integrated values, with the integration named in the same sentence; the thousandfold figure withdrawn as a statement of surface deficit; the absence of any surface-only mesozooplankton measurement for this water stated explicitly, since that absence is itself the coverage gap; Vereshchaka et al. (2017) removed from the comparison.
- Line 419–420 rewritten; no implication that the sampled layer is the region’s productivity maximum. The 20 µm size floor restated wherever the phytoplankton class appears, with an explicit statement that the class is an optical count and not a measure of biomass or production.
- Spectroscopic confirmation deferred to the next campaign, since it cannot be completed for this manuscript; optical classes labelled as optical throughout, with no claim about material composition, in the abstract and in every table and figure caption; Sargassum fragments named as a candidate contributor rather than assumed absent.
- Microlayer, near-surface and epipelagic defined at first use and applied consistently.
- Evaporation and climate material restated with the status of each part made explicit, the established mechanism separated from the open question of magnitude, that question labelled as open and given its test, and the quantification pursued in a separate paper.
- Every comparison with a historical survey labelled as a comparison rather than as a measured change; results presented as a baseline against which repeat occupations can be judged.
- Flowmeter to be fitted for the next campaign; blanks extended to every crew by means that fit a small vessel, namely a few sealed blank-water samples per voyage and an exposure blank at the point of analysis; an image-based specimen condition score added, so that sample condition is measured rather than inferred.
- A short methods comparison added, setting the tow net’s mouth, mesh, depth and filtered volume against those of the CPR, so that the selectivity of each is explicit.
- The design rationale and its constraints stated explicitly in the introduction: what a yacht-based protocol can and cannot do, and the coverage it buys in return.
We thank the referee for the time this review evidently took. The citation error in particular should not have survived our own checking, and we would rather it was caught here than after publication. We thank too the skippers and crews who collected these samples on their own passages, and we hope this exchange makes the next campaign a better one than the first.
Howard Dryden and Diane Duncan GOES Foundation, Seahorse Point Nature Laboratory, Bocas del Toro, Panama
-
AC2: 'Reply on AC1', Howard Dryden, 18 Aug 2026
reply
Author comment: addendum to AC1
Manuscript: egusphere-2026-3979
Authors: Howard Dryden and Diane Duncan
Since posting our reply to Referee 1 (AC1), we have completed a full re-analysis of the survey’s image archive, prompted by both referees’ questions about the reliability of the automated particle classification. This addendum corrects the statements in AC1 that the re-analysis supersedes, because we would rather flag them ourselves than leave them standing.
What was done. The image pipeline used for the first version of the preprint had not been preserved in runnable form and its output could not be reproduced. We rebuilt the analysis as a deterministic, fully documented pipeline: a curation stage that classified every one of the 1,495 archived frames as a sample micrograph or not (270 were excluded with a logged reason each, including duplicate frames, screen captures, photographs of scale bars and of printed documents), detection and sizing of every particle above 20 µm in the 1,225 curated micrographs, classification by stated rules, and individual visual inspection of every object of 250 µm and above. The scripts, per-image output, curation list and station tables are published as version 2 of the data record (https://doi.org/10.5281/zenodo.21477682) and every number in the revised manuscript can be reproduced from them end to end.
What changes relative to AC1.
- AC1 quoted the original inventory of 51,547 sized and classified particles. The curated, reproducible figure is 34,609 particles across 317 station-level sampling events (122.5 litres). The pervasiveness result is unchanged: black carbon at 92% of stations, microplastic particles at 94%, phytoplankton at 87%.
- AC1 discussed the 44 image-derived zooplankton objects. The re-analysis could not locate them in the archived images, and every candidate in the animal size range proved on inspection not to be an animal. The 44 is withdrawn in the revised manuscript. The corrected result is stronger than the original claim: no zooplankton was confirmed in any filter sample, and the surface zooplankton measurement rests entirely on the net tow, in which the animals are stained, photographed and individually identifiable.
- AC1 quoted a median plastic-fibre length of 204 µm from the original sizing. The re-derived value from the deposited per-particle output is 188 µm; the mesh-comparison argument made in AC1, that most particles reported here would pass a 270 µm CPR silk, is unaffected and slightly strengthened.
- AC1 stated phytoplankton presence at 81.8% of stations with a median of 10 per litre. The curated figures are 87% presence and a median of 25 per litre across stations where the class is present. The point that the same sampled volume abundantly detected phytoplankton while detecting no animals stands with more force than before.
None of the physical or oceanographic arguments in AC1, on the net tow, the mesh sizes, the surface comparison against the Tara Pacific datasets, the deployment and contamination questions, is affected by the re-analysis; the net-tow result is untouched. The revised manuscript (version 2) states all corrected figures and discloses the re-analysis in its Sections 2.2, 2.3 and 2.7.
Howard Dryden and Diane Duncan GOES Foundation, Seahorse Point Nature Laboratory, Bocas del Toro, Panama
-
RC3: 'Reply on AC2', Anonymous Referee #1, 18 Aug 2026
reply
Reviewer RC1 greatly appreciates the authors’ constructive, detailed and speedy replies (AC1, AC2, AC3) to referee comments made on their MS. The preparation of >8,000 words of clarification and additional information relating to both reviews, together with a re-analysis of the image archive, within ~24 hours is an impressive achievement. The major revisions to the MS that are planned will go a long way to allaying the concerns of this reviewer.
In particular, this reviewer welcomes the intention to much more explicitly state the limitations of the GOES sampling device and the particle characterisation process, also the greater emphasis on actions taken to minimise contamination and on the importance of using complementary zooplankton sampling devices, based on larger volumes. Those changes will greatly improve the MS.
However, major uncertainties regarding particle identification and data interpretation will remain, and cannot be resolved since the original samples have not been retained. It would therefore seem highly desirable to re-frame the MS as “lessons learnt” or a “preliminary study” (e.g. included in the title?), with its conclusions being a list of proposed improvements to the methodologies – to include preservation of the GOES filters, for subsequent land-based re-analysis (of at least a sub-sample) of particles considered to be black carbon and microplastics. With regard to the statement that “sand is not a candidate in an open ocean surface sample”, NASA estimates that around 800 million tonnes of desert dust blows over the tropical North Atlantic every year (https://science.nasa.gov/earth/earth-observatory/dust-traverses-the-atlantic-ocean-146871/) – so that possibility cannot be ruled out.
The potential for further development of the GOES sampler for open-ocean pollution monitoring would seem to be the most useful scientific contribution of the study. In addition, the desirability of much finer resolution of the vertical distribution of zooplankton in oligotrophic epipelagic waters could be highlighted as a research need, even if it is not directly investigated by the authors or by further citizen-science initiatives.
On that issue, the lack of like-for-like comparisons for near surface zooplankton abundances is indeed a major research gap. Whilst comparisons with Pacific data could be made, Hopkins (1982) would seem the only other study to provide geographically- relevant depth distributions, with a 0-15m sampling category in the eastern Gulf of Mexico, showing marked daytime near-surface depletion for many species. In addition to relatively low food abundance at the near-surface, the most plausible explanations for the observed ‘scarcity’ of meso-zooplankton in the top 1m would seem:
- Predation risk, since such zooplankton would be strongly silhouetted in high-clarity water
- The occurrence of damagingly-high UV levels (Johnsen & Widder, 2001; Al-Aidaroos et al., 2015), with UV penetration depths being unusually high in the tropical Atlantic (Tedetti & Sempere, 2007).
There would therefore seem strong evolutionary pressure for light avoidance in tropical oligotrophic waters, hence the very low abundance levels found. Whilst light gradients would be much less at nightime, there would seem no biological reason for purposeful vertical migration at night. The higher zooplankton abundances found in night-time tows could thertefore result from passive water mixing (from the DCM), rather than active swimming to the near-surface.
With regard to the linkages made in the MS to the surface microlayer and climate, the authors say they do not intend to remove such material – yet also state that it was only included on the suggestion of another referee. That was presumably in the context of a different submission: there is no record of such advice in the open discussion on the current MS. Whilst this referee has given careful consideration to the rationale for inclusion given in AC1, the justification is still found unconvincing. Furthermore, no novel data is presented on this aspect. Currently half the Abstract is given to the topic: that would seem misleading and highly disproportionate. At most a short paragraph, or preferably just a sentence or two, in the Discussion would suffice, with a couple of references. Otherwise such material is a speculative distraction, weakening not strengthening, the MS as a whole.
NB: Apologies for my own text mis-referencing regarding Gerdis (2013) - that should, of course, have been Gerdts (2015), as in the reference list.
Additional references cited above:
Al-Aidaroos, A.M., El-Sherbiny, M.M., Satheesh, S., Mantha, G., Agustī, S., Carreja, B. and Duarte, C.M., 2015. Strong sensitivity of Red Sea zooplankton to UV-B radiation. Estuaries and Coasts, 38, 846-853.
Hopkins, T.L., 1982. The vertical distribution of zooplankton in the eastern Gulf of Mexico. Deep Sea Research Part A. Oceanographic Research Papers, 29, 1069-1083.
Johnsen, S. and Widder, E.A., 2001. Ultraviolet absorption in transparent zooplankton and its implications for depth distribution and visual predation. Marine Biology, 138, 717-730.
Tedetti, M. and Sempéré, R., 2007. Penetration of ultraviolet radiation in the marine environment. A review. Photochemistry and Photobiology, 82, 389-397.
Citation: https://doi.org/10.5194/egusphere-2026-3979-RC3 -
AC4: 'Reply on RC3', Howard Dryden, 19 Aug 2026
reply
Author response to Referee 1’s second comment (RC3)
Manuscript: egusphere-2026-3979, “Zooplankton, phytoplankton and pervasive anthropogenic particles along a zonal North Atlantic transect: a multi-vessel citizen-science image survey”
Authors: Howard Dryden and Diane Duncan
We thank the referee for this second reading, for the generous assessment of the pace and scope of AC1, AC2 and the addendum, and for continuing to push the manuscript toward something more defensible than the version first submitted. Several of the points below identify real gaps; we accept those in full and describe the changes made. Two points needed correction on our side rather than the manuscript’s, and we say so plainly.
On the sand/dust point
The referee is right, and we were wrong to state, as a blanket rule, that “sand is not a candidate class in an open-ocean surface sample.” We wrote that line in AC1 while arguing that a North Sea sediment study’s 1.4% microplastic-confirmation rate, driven by quartz sand grains, was not informative for our water. That argument about the North Sea study’s relevance stands, but the general claim does not, and NASA Earth Observatory’s figure of some 800 million tonnes of Saharan dust crossing the tropical Atlantic every year (Hansen, 2020) confirms dust cannot be dismissed a priori.
We can say more than that, though, because we saw it happen. Passing North Africa on the way to Cape Verde, the sails turned dark brown with Saharan dust. We looked for it under the microscope and found the grains hard to pick out: Saharan dust is round and almost completely clear, unlike the dark, colourless particles our pipeline scores as black carbon. We have added this to Sect. 2.7 rather than leave it as a reply-only claim: the manuscript now states that dust was directly observed during the crossing, that its optical properties (round, near-transparent) are unlike the black-carbon class as defined, so we do not expect it to be a significant contributor to the classes reported here, and that its low contrast against the filter background is instead flagged as a possible blind spot of the imaging method for the next campaign’s spectroscopic confirmation to test. This is a more precise statement than either the original blanket claim or a generic “cannot be excluded”: we have direct field evidence for what the dust looks like and why it should not be confused with the classes we scored, rather than only a reason it might be.
On reframing as a preliminary study / lessons learnt
Accepted, and implemented. The abstract now opens by describing the work as “a preliminary citizen-science survey” rather than asserting the SML framing as established motivation, and the Conclusion adds a paragraph stating plainly that this is offered as “a preliminary, lessons-learnt first crossing rather than a finished protocol,” naming the retained-filter gap as the clearest lesson for the next iteration alongside the flowmeter, extended blanks and condition-scoring commitments already in Sect. 2.7.
The title has, in fact, already been changed in this revision for exactly this reason, and the tracked-changes manuscript shows it: “Near-absence of surface zooplankton and pervasive anthropogenic particles along a zonal North Atlantic transect…” is now “Zooplankton, phytoplankton and pervasive anthropogenic particles along a zonal North Atlantic transect: a multi-vessel citizen-science image survey.” The reasoning is worth stating plainly. The near-absence result itself is not fragile: 317 stations, each sampling a full 0.5-litre cartridge draw for a combined 158.5 litres of surface water, returned zero confirmed zooplankton, and an independent surface net tow (0.2–0.5 animals per cubic metre, day and night) corroborates a genuinely thin near-surface stock rather than a small-sample artefact: the two methods agree. We had expected, even knowing the water was oligotrophic, to find more sub-300 µm zooplankton than we did; finding none was the observation that motivated the original title. Where this review has changed our thinking is on what that scarcity actually demonstrates. Nobody has measured what surface-layer (0–1 m) zooplankton density should be in this water, and with no pre-existing baseline there is no basis for treating our result as a departure from a known normal, or for a title that carries the inference that something is wrong. The near-absence we measured may simply be the natural state of the equatorial Atlantic away from upwelling. The revised title therefore just names what we measured, without asserting that any of it is anomalous or that one part explains another; the survey now presents itself as setting a first baseline that future campaigns can validate against, and the hedging added in Sect. 5, Sect. 7.1 and the Conclusion (offering the pollution-grazer link as a candidate mechanism, not a demonstrated cause) carries the same discipline through the text. Phytoplankton has also been added to the title in this round. It was present at 87% of stations and is one of the four particle classes this survey actually classified and reported, so leaving it out of the title while naming zooplankton and the anthropogenic particles understated what the crossing measured.
Specifically on preserving material for re-analysis: the referee’s suggestion is right, and we have added it to Sect. 7 (Recommendations) as a stated protocol lesson, retaining at least a sub-sample of the physical filter material recovered as black carbon and microplastic, so that land-based spectroscopic re-analysis is possible after the fact, which this survey cannot offer because the original filters were not kept. We have also now stated why they were not kept, since the reason matters for how the lesson gets implemented: marine plankton samples degrade very quickly, especially at tropical temperatures, and no preservation step was available on board the volunteer vessels. Retention is therefore not simply a matter of deciding to keep the filters next time. Sect. 7 now states that a practical preservation method, most plausibly drying the filters and sealing them in an airtight container, will itself need to be evaluated before retained-filter analysis can become part of the volunteer protocol.
On the GOES sampler’s development potential
We agree, and have added this to Sect. 7: we now state explicitly that further development of the GOES sampler itself, extended to a stratified near-surface series and paired with retained-filter spectroscopy, is plausibly the most useful scientific contribution this programme can make to open-ocean pollution monitoring going forward, alongside the standardised monitoring commitment already there.
While revising this point we also tightened Sect. 2.8’s account of the net tow itself. The plankton trawl was towed by the lead author alone, as a one-off confirmation that the imaging protocol was not simply missing animals present in the water, and it is not equipment we expect other citizen-science crews to carry or that we intend as part of the replicable multi-vessel method. We have added a sentence stating this explicitly and cross-referencing Sect. 7, so the manuscript is clear that the GOES cartridge, not the net, is the instrument intended for wider volunteer deployment, which is consistent with, and reinforces, the referee’s point above.
On finer vertical resolution as a research need
Also added to Sect. 7: we now state the need for much finer vertical resolution of near-surface zooplankton than existing depth categories allow, ideally a stratified series within the top 1 m, as a research need this survey has identified but not answered, independent of whether it is pursued by further citizen-science deployments or by conventional means. We have strengthened the case for that research need with three further points, added to Sect. 2.3 and Sect. 7. First, the near-surface layer is not inherently uninhabitable: neustonic zooplankton do live in the top metre and in the microlayer itself, and the distinct blue carotenoid pigmentation of pontellid copepods is a recognised adaptation to exactly the ultraviolet exposure and visual predation pressure of that niche (Rahlff et al., 2018), so adapted surface taxa exist elsewhere, which makes the near-absence we recorded more informative than a trivial expectation of emptiness rather than less. Second, very few surveys have sampled the top metre of the ocean anywhere, and we are aware of none dedicated to the equatorial Atlantic specifically; the gap RC1 and RC2 both pushed us to be explicit about is a real one, not a rhetorical concession. Third, we now cite the paper that first characterised the sea-surface microlayer itself as a living, gelatinous biofilm rather than a passive film (Cunliffe and Murrell, 2009), alongside the Cunliffe et al. (2013) synthesis already cited, since it is that biological character of the SML, not just its chemistry, that makes biogenic particles there potentially important for air–sea exchange and aerosol formation, and Sect. 7 now states that this makes surveying the top metre and its atmospheric interface a worthwhile objective in its own right, independent of the zooplankton question.
On Hopkins (1982) and the near-surface comparison
Added to Sect. 2.3. We had missed this reference in the surface-comparison literature search behind AC1 and AC2, and the referee is right that it is the closest geographically relevant precedent available: a 0–15 m near-surface category in the eastern Gulf of Mexico showing marked daytime depletion of many mesozooplankton species relative to deeper strata. We have been careful in how we use it, and the revised text now states the limits on the comparison explicitly. First, its shallowest category still integrates 0–15 m rather than the top 1 m this survey targets. Second, the semi-enclosed Gulf of Mexico is a different hydrographic setting from the open equatorial Atlantic: warmer, strongly influenced by terrestrial runoff and, over much of its shelf, by pronounced eutrophication, which carries further physicochemical consequences. Eutrophication-driven respiration lowers pH and aragonite saturation (Cai et al., 2011), and higher organic loading tends to lower both the redox potential of the water (Breitburg et al., 2018) and the zeta potential of suspended particles (Parrella et al., 2024).
We also checked whether the resulting change in micro-viscosity says anything about zooplankton body size in the Gulf of Mexico relative to this survey, since size structure bears directly on whether Hopkins’ community is comparable to the sub-300 µm fraction targeted here. The one study we found that measured viscosity effects on zooplankton swimming speed directly (Larsen et al., 2008) points a specific direction: of their three study animals, the smallest (a rotifer) was the most sensitive to increased viscosity and the largest (a calanoid copepod) the least, which is the opposite of a naive size-drag argument. We have added this to Sect. 2.3: if that pattern holds more generally, the Gulf of Mexico’s higher micro-viscosity would disadvantage smaller-bodied zooplankton preferentially, shifting its community toward larger body sizes than the near-surface fraction this survey targets, a further, specific reason the two settings are not a like-for-like match. As with the other physicochemical points above, we are careful to state that none of these parameters was measured in either study and we do not know how, or whether, they bear on zooplankton abundance in practice; we offer the viscosity point as a documented, cited possibility, not a demonstrated mechanism. So we take from Hopkins (1982) only the qualitative pattern of daytime near-surface depletion, not any quantitative expectation for the oligotrophic water sampled here. It is a precedent for the pattern, not a benchmark for the numbers; the quantitative comparison of record remains the true-surface (0–1 m) Tara Pacific datasets.
This same reasoning points the other way for the water in this survey. The equatorial Atlantic is open, is cooler, and has lower micro-viscosity than the Gulf of Mexico. If the pattern in Larsen et al. (2008) holds more generally, that lower viscosity should favour small-bodied zooplankton over large ones, the opposite of what we describe for the Gulf of Mexico above. We therefore expected to find more small-bodied zooplankton in this water, not fewer. Finding almost none was not the result we expected, and that surprising result is a genuine concern, not a neutral observation. However, we cannot say it points to one specific cause. We also cannot say it marks a change from an earlier, normal state, because no earlier baseline exists for surface zooplankton density in the equatorial Atlantic. This survey is the first attempt anyone has made to measure that density. We have added this same point to Sect. 2.3 of the manuscript, directly after the viscosity discussion above, so the reasoning is stated in both places.
On predation risk and UV avoidance as candidate mechanisms
Added to Sect. 2.3, immediately alongside the Hopkins citation. We now name both mechanisms explicitly: visual predation risk in the high-clarity water of this transect, and avoidance of damagingly high ultraviolet exposure, which penetrates unusually deep in the clear water of the tropical Atlantic, with the three citations the referee supplied (Johnsen and Widder, 2001; Tedetti and Sempéré, 2006; Al-Aidaroos et al., 2015). Consistent with the co-occurrence-not-causation discipline we have applied throughout this revision, we present both as candidate explanations for future testing rather than as demonstrated causes of the scarcity we observed; this survey was not designed to distinguish between them, or from a genuine absence, or from residual sampling limitations.
On the night/day tow difference
Accepted: Sect. 2.8 previously attributed the roughly twofold higher night-time net-tow density to diel vertical migration without considering the alternative the referee raises. We now state that we cannot rule out passive night-time mixing of animals up from the deep chlorophyll maximum as an alternative to active swimming, since the light gradients relevant to both visual predation and UV avoidance are much reduced after dark and we see no obvious biological reason for purposeful upward migration under those conditions.
On the surface-microlayer and climate material
We accept the criticism and have made the cut. In responding to the first review we stated that the wider implications material had been added because we were asked to broaden the paper beyond a narrow survey report, without being able to document that claim at the time, and it was fair for the referee to press us on it. We can now substantiate it properly. Before this manuscript was opened for public review, the handling editor’s own screening decision required exactly that broadening, stating plainly that we needed to spell out what is learnt from the data set and what it implies for our understanding of the ocean, and that a paper confined to reporting the survey itself would be better suited to a data journal than to Ocean Science. That instruction is why the manuscript discusses pollution sources, ecosystem consequences and recommendations alongside the survey itself, rather than stopping at the method and its raw results. It does not, on its own, excuse the specific length or certainty of the surface-microlayer and climate material in Section 4, and we still accept that this particular section overreached on both counts. The broader scope of the paper, however, was not an unexplained addition on our part. Without meeting that condition, the paper would not have been sent out for review.
Ocean Science’s own published aims and scope make the same point in stronger terms than the screening decision did. The journal states that it “publishes studies with important implications for our understanding of the state and behaviour of the ocean, its impacts on and responses to climate, including coupling at the ocean–atmosphere, ocean–sea ice–ice sheet and ocean–coast interfaces” (Ocean Science, 2026). This transect sits inside that remit in a way it did not when the manuscript was a narrower survey report. The equatorial Atlantic, and its top metre specifically, has rarely been measured at all, and we are aware of no comparable near-surface survey of this water. What the crossing adds beyond the zooplankton result is a picture of plastic and partially combusted carbon co-occurring in the same near-surface water, at the same stations, in a combination we do not believe has been reported this way before. We then connect that picture to the sea-surface microlayer, whose biological character is already established in the literature (Cunliffe and Murrell, 2009; Cunliffe et al., 2013), and to its documented role in ocean–atmosphere coupling: air–sea gas exchange, aerosol formation and evaporation control. That role sits close to the centre of the journal’s own stated scope, yet it remains a surprisingly under-studied part of ocean science given how directly it bears on climate regulation. Our contribution is not a new mechanism; the mechanism is already understood and cited. It is the first evidence that the pollution and the microlayer occupy the same place, in this specific and consequential water. We think this observation opens a line of enquiry that ocean science has rarely pursued.
The cut itself: the abstract’s opening paragraph, previously some eight sentences motivating the survey through the microlayer’s climate role, is now one sentence stating the theme without asserting the argument. Section 4, previously roughly a thousand words with its own figure (ERA5 water-vapour and cloud-amplitude reanalysis, redox/zeta-potential speculation, the CLAW hypothesis, and named 2022 storm attribution studies), is now two sentences: that the microlayer’s suppression of air-sea CO₂ exchange and its contribution to cloud condensation nuclei are measured facts, that its effect on evaporation is an open, testable hypothesis this survey cannot answer, and that we intend to pursue that quantification separately, with dedicated eddy-covariance flux measurements, in a paper of its own. The removed material, namely the ERA5 reanalysis, the figure, and the fuller discussion, is marked as a tracked deletion in the revision rather than silently discarded, so the full prior text remains visible in the redline for the record. This also resolves a loose end from AC1: we had earlier promised that the fuller quantitative treatment would be “a separate paper, which we intend to pursue,” but never actually stated that commitment in the manuscript text itself. It now sits in the two sentences that replace Section 4.
We want to be precise about what we are and are not conceding, because the cut was about proportion, not about whether the underlying mechanism is real. The relationship linking phytoplankton productivity, the zooplankton that graze it and the resulting stability of the surface microlayer is established in the literature, not speculative; Cunliffe et al. (2013), already cited in the manuscript, is a synthesis of exactly that biological control. What is genuinely novel, and to that extent still speculative, is not the mechanism but its application here: nobody appears to have taken these measurements, in this combination, in this water, before. That absence is itself worth naming, because the water in question is not an arbitrary stretch of ocean. This transect crosses the tropical Atlantic’s Main Development Region, the corridor of African easterly waves from which most Atlantic hurricanes affecting the Caribbean and the US Gulf and southeast coasts draw their initial organisation (Goldenberg and Shapiro, 1996). An established mechanism and a specific, consequential body of water that nobody has measured together are, on their own, a good reason to look, even before either party has proof. We have added one paragraph to the Conclusion making exactly this case in the same hedged register as the rest of the revision (established mechanism, unmeasured application, first baseline rather than demonstrated result), without reopening the evaporation, cloud or CLAW material the referee asked us to cut. We think this is a fair middle ground: we accept that Section 4 overreached in length and certainty, and we have cut it accordingly, but we do not think the underlying reason this water is worth studying should be cut along with it.
On Gerdis/Gerdts
No correction needed on our side: the manuscript’s reference list already reads “Gerdts, G.” (2015), which is correct, and the referee’s own typo in the review does not appear anywhere in our text. We are grateful for the flag regardless.
Summary of changes
- 2.7: windborne Saharan dust (Hansen, 2020; NASA Earth Observatory) added, correcting the overstated claim in AC1 that sand cannot be a candidate in an open-ocean sample. Direct field observation (sails visibly darkened with dust off North Africa) and microscopy (round, near-transparent grains, unlike the dark, colourless black-carbon class) are both now stated; we do not expect dust to be a significant contributor to the classes reported here, but flag its low optical contrast as a possible detection blind spot.
- Abstract and Conclusion: explicit preliminary/lessons-learnt framing added; the Conclusion now states this is a first crossing rather than a finished protocol, naming the retained-filter gap as the clearest lesson for the next iteration. Title changed in this revision (tracked): “Near-absence of surface zooplankton…” → “Zooplankton, phytoplankton and pervasive anthropogenic particles…”, on the reasoning given above. The near-absence result is robust (158.5 L across 317 stations, corroborated by the independent net tow), but with no pre-existing baseline for this water we no longer present it as evidence that something is wrong, only as a first baseline measurement, and the title no longer carries that inference.
- 7: filter/sample preservation for land-based spectroscopic re-analysis added as a stated protocol lesson, with the reason the filters were not kept (rapid degradation of marine plankton samples, especially at tropical temperatures, with no preservation step available on board) and the consequent need to evaluate a practical preservation method (e.g. drying and sealing in an airtight container) before retention can join the volunteer protocol.
- 7: further development of the GOES sampler itself named as plausibly the programme’s most useful scientific contribution to open-ocean pollution monitoring.
- 7: finer vertical resolution of near-surface zooplankton (ideally a stratified series within the top 1 m) stated as an identified but unanswered research need. Sect. 2.3 and Sect. 7 further strengthened with: neustonic zooplankton (blue-pigmented pontellid copepods) as evidence the near-surface layer is habitable and adapted to, not inherently empty (Rahlff et al., 2018); the observation that almost no surveys have sampled the top metre anywhere, and none we are aware of in the equatorial Atlantic specifically; and the sea-surface microlayer’s characterisation as a living, gelatinous biofilm (Cunliffe and Murrell, 2009), stated as the reason biogenic particles at the air–sea interface are a worthwhile survey target independent of the zooplankton question.
- 2.3: Hopkins (1982) Gulf of Mexico near-surface (0–15 m) depth-distribution data added as the closest available geographically relevant precedent, with the limits of the comparison stated explicitly. These include the 0–15 m depth integration versus this survey’s top-1-m target, and the Gulf’s different hydrographic setting (semi-enclosed, warmer, terrestrially influenced and in large part eutrophic) versus the open oligotrophic equatorial Atlantic, including the eutrophication’s physicochemical consequences (lower pH and aragonite saturation, Cai et al., 2011; lower redox potential, Breitburg et al., 2018; lower particle zeta potential, Parrella et al., 2024) and, specifically, higher micro-viscosity’s likely size-selective effect toward larger-bodied zooplankton in that setting (Larsen et al., 2008). Used as a qualitative pattern precedent only, not a quantitative benchmark. A further sentence is now added to Sect. 2.3 stating the reverse expectation for our own water: lower viscosity here should favour small-bodied zooplankton, so we expected more of them, not fewer. Finding almost none was a genuine surprise, but it cannot be linked to a specific cause because no earlier baseline exists for surface zooplankton density in the equatorial Atlantic.
- 2.3: predation risk and UV avoidance added as candidate explanations for the observed near-surface scarcity (Johnsen and Widder, 2001; Tedetti and Sempéré, 2006; Al-Aidaroos et al., 2015), framed as hypotheses rather than demonstrated causes. (In-text citation year for Tedetti and Sempéré corrected from 2007 to 2006 to match the reference-list entry and the publisher’s own record, a transcription slip on our part, caught while auditing the reference list for this response.)
- 2.8: passive night-time mixing from the deep chlorophyll maximum added as an alternative to active diel vertical migration for the observed day/night net-tow difference.
- Abstract paragraph 1 cut from eight sentences to one; Section 4 cut from roughly a thousand words and one figure to two sentences with two references, with the removed material retained as a tracked deletion; the AC1 commitment to a separate, dedicated quantitative treatment is now stated in the manuscript text itself rather than only in correspondence. One new paragraph added to the Conclusion stating that the phytoplankton-zooplankton-microlayer mechanism is established, not speculative (Cunliffe et al., 2013), and that this transect’s location in the Atlantic hurricane Main Development Region (Goldenberg and Shapiro, 1996) is part of why the water is worth studying, framed as motivation for a first baseline rather than a demonstrated result. The provenance of the paper’s broader scope, previously withdrawn for lack of documentation, is now substantiated: the handling editor’s pre-review screening decision required the paper to discuss the data’s wider implications rather than stand as a narrow survey report, on pain of being redirected to a data journal instead, which explains the manuscript’s coverage of pollution sources, ecosystem consequences and recommendations beyond the citizen-science method itself, though it does not excuse Section 4’s specific length or certainty. We have also set that broader scope against the journal’s own published aims and scope, which states that Ocean Science publishes studies with important implications for the ocean’s behaviour and its coupling with climate; the transect’s near-surface measurements of plastic and partially combusted carbon, connected to the already-established biology of the sea-surface microlayer, sit inside that remit and, we argue, open a line of enquiry the field has not much pursued.
- Sand/dust claim in AC1 corrected as above; no manuscript change required for the Gerdis/Gerdts point, since the reference list already reads Gerdts (2015) correctly.
- 2.8: the net-tow’s status clarified. It was towed by the lead author alone as a one-off confirmation of the imaging protocol, not equipment intended for the replicable multi-vessel method, and the GOES cartridge is stated as the instrument intended for wider volunteer deployment (cross-referenced to Sect. 7).
We are grateful for this second, equally careful reading. Between RC1, RC2 and this second comment, the manuscript has been substantially strengthened; we think it is now a materially more honest account of what a first citizen-science crossing of this kind can and cannot show.
Howard Dryden and Diane Duncan GOES Foundation / Seahorse Point Nature Laboratory, Bastimentos, Bocas del Toro, Panama
References cited in this response
Al-Aidaroos, A. M., El-Sherbiny, M. M., Satheesh, S., Mantha, G., Agustí, S., Carreja, B., and Duarte, C. M.: Strong sensitivity of Red Sea zooplankton to UV-B radiation, Estuaries and Coasts, 38, 846–853, https://doi.org/10.1007/s12237-014-9857-7, 2015.
Breitburg, D., Levin, L. A., Oschlies, A., Grégoire, M., Chavez, F. P., Conley, D. J., Garçon, V., Gilbert, D., Gutiérrez, D., Isensee, K., Jacinto, G. S., Limburg, K. E., Montes, I., Naqvi, S. W. A., Pitcher, G. C., Rabalais, N. N., Roman, M. R., Rose, K. A., Seibel, B. A., Telszewski, M., Yasuhara, M., and Zhang, J.: Declining oxygen in the global ocean and coastal waters, Science, 359, eaam7240, https://doi.org/10.1126/science.aam7240, 2018.
Cai, W.-J., Hu, X., Huang, W.-J., Murrell, M. C., Lehrter, J. C., Lohrenz, S. E., Chou, W.-C., Zhai, W., Hollibaugh, J. T., Wang, Y., Zhao, P., Guo, X., Gundersen, K., Dai, M., and Gong, G.-C.: Acidification of subsurface coastal waters enhanced by eutrophication, Nature Geoscience, 4, 766–770, https://doi.org/10.1038/ngeo1297, 2011.
Cunliffe, M., Engel, A., Frka, S., Gašparović, B., Guitart, C., Murrell, J. C., Salter, M., Stolle, C., Upstill-Goddard, R., and Wurl, O.: Sea surface microlayers: a unified physicochemical and biological perspective, Progress in Oceanography, 109, 104–116, https://doi.org/10.1016/j.pocean.2012.08.004, 2013.
Cunliffe, M. and Murrell, J. C.: The sea-surface microlayer is a gelatinous biofilm, ISME Journal, 3, 1001–1003, https://doi.org/10.1038/ismej.2009.69, 2009.
Goldenberg, S. B. and Shapiro, L. J.: Physical mechanisms for the association of El Niño and West African rainfall with Atlantic major hurricane activity, Journal of Climate, 9, 1169–1187, https://doi.org/10.1175/1520-0442(1996)009<1169:PMFTAO>2.0.CO;2, 1996.
Hansen, K.: Dust Traverses the Atlantic Ocean, NASA Earth Observatory, https://science.nasa.gov/earth/earth-observatory/dust-traverses-the-atlantic-ocean-146871/, 2020.
Hopkins, T. L.: The vertical distribution of zooplankton in the eastern Gulf of Mexico, Deep Sea Research Part A, 29, 1069–1083, https://doi.org/10.1016/0198-0149(82)90032-0, 1982.
Johnsen, S. and Widder, E. A.: Ultraviolet absorption in transparent zooplankton and its implications for depth distribution and visual predation, Marine Biology, 138, 717–730, https://doi.org/10.1007/s002270000499, 2001.
Larsen, P. S., Madsen, C. V., and Riisgård, H. U.: Effect of temperature and viscosity on swimming velocity of the copepod Acartia tonsa, brine shrimp Artemia salina, and rotifer Brachionus plicatilis, Aquatic Biology, 4, 47–54, https://doi.org/10.3354/ab00093, 2008.
Ocean Science: Aims and Scope, Copernicus Publications, https://ocean-science.net/about/aims_and_scope.html, last access: 19 August 2026, 2026.
Parrella, F., Brizzolara, S., Holzner, M., and Mitrano, D. M.: Impact of heteroaggregation between microplastics and algae on particle vertical transport, Nature Water, 2, 541–552, https://doi.org/10.1038/s44221-024-00248-z, 2024.
Rahlff, J., Ribas-Ribas, M., Brown, S. M., Mustaffa, N. I. H., Renz, J., Peck, M. A., Bird, K., Cunliffe, M., Melkonian, K., and Zappa, C. J.: Blue pigmentation of neustonic copepods benefits exploitation of a prey-rich niche at the air–sea boundary, Scientific Reports, 8, 11510, https://doi.org/10.1038/s41598-018-29869-7, 2018.
Tedetti, M. and Sempéré, R.: Penetration of ultraviolet radiation in the marine environment. A review, Photochemistry and Photobiology, 82, 389–397, https://doi.org/10.1562/2005-11-09-IR-733, 2006.
-
RC4: 'Reply on AC4', Anonymous Referee #1, 19 Aug 2026
reply
The authors are thanked for their constructive and detailed response to the further comments made and queries raised by this referee There is now much greater clarity regarding what has been achieved, also the uncertainties - and the need for follow-up studies.
Citation: https://doi.org/10.5194/egusphere-2026-3979-RC4 -
AC5: 'Reply on RC4', Howard Dryden, 19 Aug 2026
reply
Reply to Referee 1’s closing comment
Manuscript egusphere-2026-3979We thank the referee for this careful and generous review. Their comments were constructive throughout, and directly improved the manuscript at several points. We agree that this survey sits in new territory: combined surface monitoring of plankton and anthropogenic particles across the equatorial Atlantic has not been attempted before, by anyone, anywhere, at this scale.
Citizen science also has a great deal to offer in collecting samples of this kind. We were surprised to find that no one had sampled across the equatorial Atlantic in this way before us. Blue-water yachts sail across every ocean, including places research vessels rarely visit. We believe the additional information these vessels can gather will help build the understanding needed for informed decisions.
The oceans cover 71% of the planet, and they form its life support system and the planet’s primary regulator of weather. Yet the surface layer where air meets water, the thinnest and most exposed part of that system, remains one of the least monitored. This survey found that layer nearly empty of animal life and heavily loaded with combustion and plastic particles, in the same water, at the same time. That combination deserves far more attention than it currently receives.
We believe the task ahead has two parts, not one. Scientists must understand the processes now unfolding at the ocean surface, though understanding alone will not be enough. Research must also find practical ways to regenerate marine life, and, where possible, to soften the worst effects of climate disruption. The ocean will not wait for a perfect dataset before it changes. Neither should our effort to understand and protect it wait either.
Howard Dryden and Diane Duncan
GOES Foundation, Seahorse Point Nature Laboratory, Bocas del Toro, PanamaCitation: https://doi.org/10.5194/egusphere-2026-3979-AC5
-
AC5: 'Reply on RC4', Howard Dryden, 19 Aug 2026
reply
-
AC1: 'Reply on RC1', Howard Dryden, 17 Aug 2026
reply
-
RC2: 'Comment on egusphere-2026-3979', Anonymous Referee #2, 17 Aug 2026
reply
The manuscript claims a near-absence of surface zooplankton alongside a ubiquity of anthropogenic particles, suggesting a causal link between widespread pollution and declining zooplankton concentrations.
I appreciate the authors' initiative and efforts to leverage citizen science to support ocean monitoring. There is a continuous need for in situ data, particularly quantitative surveys of zooplankton concentration. However, the push for increased data collection should not come at the expense of data quality.
Below is a non-exhaustive list of critical methodological issues that call into question the validity of this study:
- Insufficient Sampling Volume: A 500 mL sample volume is far too small to quantitatively sample zooplankton (or even phytoplankton) in open-ocean surface oligotrophic waters. Sampling volume must scale inversely with target particle concentration. For context, quantitative sampling in oligotrophic waters typically requires several cubic meters for particles between 20 and 200 m (i.e., microphytoplankton) and tens of cubic meters for mesozooplankton. Given the tiny volume, the near-absence of zooplankton in these samples is expected, but it does not indicate a near-absence of zooplankton in the ocean.
- Missing and Inaccessible Sampling Details: A detailed methodology explaining the self-contained cartridge and the specific filter used is missing from the main manuscript. The explanation provided in Appendix A is vague and confusing, and the external link meant to provide additional details is broken (returning an "HTTP ERROR 403: Access Denied" message). Complete transparency regarding this equipment is crucial to support the authors' claims.
- Avoidance Behaviors and Intake Geometry: Zooplankton can easily avoid intake orifices with such a small aperture diameter—which is precisely why standard zooplankton nets feature much larger mouth diameters. Consequently, even if the sample volume were increased, this intake design would remain non-quantitative for zooplankton.
- Self-Contamination from the Vessel: Every vessel acts as a significant point source of local contamination. Main engine and generator exhausts, hydraulic oil traces, paints, lines, sails, and degrading composite, textile, or plastic components are continuously flushed from the deck and hull into the surrounding surface layer. Because the cartridge was towed directly behind the vessel, this setup likely sampled water already contaminated by the vessel itself. Furthermore, the blanking procedures only addressed airborne contamination rather than vessel-derived waterborne contaminants, making this methodology unsuitable for estimating true environmental pollution.
- Disregard for Validation Red Flags (Confirmation Bias): In Section 2.7 ("Validation and limitations of the classification"), the authors state:"We cross-checked the pipeline against one vessel leg counted by hand. The dominant signals agreed — zero zooplankton and a load dominated by carbon and fine unresolved material — while the split between plastic and phytoplankton diverged, and we therefore treat that partition as provisional. Because visual identification of these small, degraded particles by non-specialists is necessarily subjective, we do not rely on the crew hand-logs quantitatively."This passage suggests that the authors discarded validation results simply because they failed to align with expected outcomes, dismissing manual cross-checks rather than addressing the clear divergence in classification.
- Misinterpretation of Net-Tow Controls: The authors use independent net-tow results to argue that the low zooplankton counts were not an artifact of the cartridge's small sample volume. However, the zooplankton concentrations estimated from the validation net tows fall within the expected range for the studied region. Because the net tows detected normal zooplankton levels while the cartridge did not, these control measurements actually prove that the small sample volume and intake method of the cartridge present severe methodological limitations.
Conclusion
While I acknowledge the significant commitment and effort required to execute such a large-scale citizen science project, the manuscript in its current form suffers from fundamental flaws. The analysis reflects a degree of confirmation bias, wherein data contradicting the core hypothesis were dismissed, and severe methodological limitations were overlooked. Ultimately, the data presented do not support the conclusion that surface zooplankton are absent or that a causal link exists with anthropogenic pollution.
Citation: https://doi.org/10.5194/egusphere-2026-3979-RC2 -
AC3: 'Reply on RC2', Howard Dryden, 18 Aug 2026
reply
Author response to Referee Comment 2
Manuscript: egusphere-2026-3979, “Near-absence of surface zooplankton and pervasive anthropogenic particles along a zonal North Atlantic transect: a multi-vessel citizen-science image survey”
Authors: Howard Dryden and Diane Duncan
We thank the referee for a careful and clearly expert reading, and for the opening remark that in situ quantitative zooplankton data are needed. We agree with his principle that more data must not come at the cost of data quality, and we accept several of his points without argument.
Two of his six points rest on descriptions of our method that do not match the manuscript, and one is an allegation about our conduct. We answer each point below. But the referee’s underlying challenge, that the image classification could not be trusted, deserved more than argument, and we begin with what we did about it.
We re-analysed the entire image archive from scratch. The pipeline used for the first version of this preprint had not been preserved in a runnable form and its output was not linked to the source images, so its numbers could not be reproduced, by the referee or by us. Rather than defend figures we could not reproduce, we rebuilt the analysis as a deterministic, fully documented pipeline: a curation stage that classifies every one of the 1,495 archived frames as a sample micrograph or not (270 were excluded, with a logged reason for each), detection and sizing of every particle above 20 µm in the 1,225 curated micrographs, classification by stated rules, and individual visual inspection of every object of 250 µm and above. The scripts, the per-image output, the curation list and the station-level tables are published as version 2 of the data record (https://doi.org/10.5281/zenodo.21477682), so every number in the revised manuscript can be reproduced end to end by anyone.
The re-analysis reproduced the particle inventory to within about ten percent (34,609 particles against 51,547; the difference is largely the excluded non-sample frames) and confirmed the pervasiveness of black carbon (92% of stations), microplastic (94%) and phytoplankton (87%). It could not substantiate the zooplankton class: none of the 44 reported animal-class objects could be located in the archived images, and every candidate in the animal size range proved on inspection to be a plastic fragment, debris, an aperture artefact or, in frames the curation stage now excludes, printed matter. The 44 is withdrawn in the revision. The corrected result is an absence: no zooplankton was confirmed in any filter sample, and the surface zooplankton measurement rests entirely on the net tow, where the animals are stained, photographed and individually identifiable. The referee’s scepticism about the image-based zooplankton identification was, in short, correct, and it has made the paper stronger; the central finding it questioned is now supported by better evidence than before.
0. What this survey is, and what it is not
This is a citizen-science project built around what an ocean-going blue-water yacht can do, with the limitations and the advantages that entails. The protocol is constrained by a small crew, underway, without winches, wire, laboratory space or power for depth-resolved work, and we do not pretend otherwise. Against that, cruising yachts cross water that research vessels rarely visit, on tracks and at times that no funded programme covers, and they do so continuously and at negligible marginal cost.
The project was built around a single instrument, the 0.5 litre GOES cartridge, used by every yacht in every ocean from the high latitudes to the equator. That is a deliberate choice and it is the main advantage of the system. When many vessels sample in many places, a fixed volume and a fixed filter make the counts directly comparable between them, which matters more for a distributed fleet than absolute sensitivity at any one station. The cost is that the volume is fixed, and in oligotrophic water it is small. A protocol that changed gear by water type would give better absolute numbers in each and no comparability across the fleet.
The Continuous Plankton Recorder makes the point for us, and we make it without criticism of that survey. The CPR is not an ideal sampler either. It filters through 270 µm silk, behind merchant ships at service speed, at about 10 m depth, so the smaller plankton pass through it, and so does most microplastic and combustion carbon, which it was never designed to retain or to classify. Its value does not rest on being the best available gear. It rests on ninety years of the same instrument deployed the same way, so that change over time can be read without a change of method confounding it. That is the model we have in mind: the same discipline applied globally, and extended to the particles the CPR was not built to see, the soot and the plastic entering the ocean alongside the plankton. A first crossing cannot deliver that. It can establish whether the instrument and the fleet are capable of it.
The instrument is also not limited to yachts. The same cartridge has been used from piers and from fishing vessels, in rivers, and in rainwater. That range is not incidental to the question. Microplastic and partially combusted carbon reach the ocean from land, by river and through the atmosphere, and they cannot be understood from open-ocean samples alone. A single sampler usable at every point along that path, from rainfall to river to shelf to gyre, yields one internally comparable record of the whole route rather than four records that cannot be set against each other.
There is also a practical reason why this particular layer is under-sampled, and it is not neglect. Ship time is the scarcest resource in oceanography, and monitoring a global surface boundary layer from research vessels is not affordable at the coverage the question needs. It is also not the natural platform. A large hull, its bow wave and its wake disturb the top metre, which is the layer the sample is meant to represent, and microlayer work is for that reason commonly done from small craft launched clear of the ship rather than from the ship itself. A small sailing vessel drawing a single fixed volume is a more practical platform for that layer than a research ship, although, as we accept at point 4, it raises contamination questions of its own. Neither the economics nor the hydrodynamics favours doing this work from a ship. Nor does the geography. A research vessel cannot be where these samples were taken. The Atlantic Meridional Transect runs a single meridional line once or twice a year, the CPR follows commercial routes, and neither crosses the trade-wind belt zonally. Cruising yachts cross it in numbers, every season, on passages they are making anyway.
1. Sampling volume
The referee is right that 0.5 litre cannot quantify mesozooplankton in oligotrophic surface water. We agree, and the manuscript agrees. Two things follow that we should have made clearer.
The zooplankton result does not come from the 0.5 litre cartridge. The manuscript states this at line 178: “the zooplankton result comes instead from the surface net tow (Sect. 2.8, Table 2).” Section 2.8 describes a 120 µm mesh net of 0.30 m mouth diameter, towed at the surface at 1 to 2 knots for 10 minutes per haul, straining approximately 43,600 litres, or 43.6 m³, per haul, in paired day and night sets. That is within the “tens of cubic metres” the referee gives as the requirement for mesozooplankton, and it is the reason the net tow was run at all.
The first referee made the same attribution. When two expert readers reach the same wrong conclusion, the fault lies with our presentation and not with their attention. In revision the net-tow method, its mesh and its filtered volume will appear in the abstract and at the first mention of any zooplankton density, and the cartridge will be described only as what it is, a particle sampler.
The cartridge targets a size fraction the referee’s argument does not address. His scaling rule, that volume must rise as concentration falls, is correct, and applied to mesozooplankton it gives the tens of cubic metres he states. The cartridge does not sample mesozooplankton. It resolves everything above 20 µm, and the fraction that matters here is the one between 20 and 200 µm: copepod nauplii, early copepodites and other small metazoan stages. Essentially every previous measurement of these waters used a net of 200 to 330 µm, which passes that entire fraction. The surface layer of the tropical Atlantic has scarcely been examined down to 20 µm at all.
We expected low counts. We did not expect none. Mesozooplankton are scarce in oligotrophic surface water and we knew that before sailing. What we expected, sampling down to 20 µm, was that the smaller fraction would appear well above what we in fact recorded. Instead the deposited re-analysis confirms no zooplankton in any curated sample: 122.5 litres examined at 20 µm resolution across 317 station-level sampling events, and not one animal. This is not a single small sample but the sum of 1,225 of them. In revision that expectation is stated in the introduction, with the size ranges named, so that the reader knows which fraction the claim concerns.
The same sample volume detected everything else. This is the point we would ask the referee to weigh, because it is an internal control rather than a comparison with someone else’s ocean. The identical 0.5 litre cartridge, the identical filter and the identical 20 µm size floor classified 34,609 particles across the curated archive. Phytoplankton were present at 87% of stations, black carbon at 92%, microplastic particles at 94%, with the busiest stations reaching several hundred per litre. All of these are at or near the numbers we expected. Zooplankton were present at no station at all.
A sampled volume too small to find zooplankton should also have been too small to find phytoplankton, and it found them at nearly nine tenths of all stations. Whatever explains the absence of animals has to explain why the same water, drawn through the same filter and imaged to the same threshold, yielded abundant counts of three particle classes and none at all of the fourth. Sampled volume alone does not do that.
The plankton trawl was run because of that shortfall, not to confirm it. The obvious explanation for detecting almost nothing was that the cartridge was the wrong instrument, which is precisely the referee’s argument. So we put a 120 µm net of 0.30 m mouth diameter in the water and strained 43.6 m³ per haul, by day and by night. The net returned 0.2 to 0.5 animals per m³. That is the test his objection calls for, and it was run for his reason rather than against it.
A note on the figures used as our benchmark. His volume requirements, and the expected range he invokes at point 6, are given without a reference, a location, a sampling depth or a mesh size. Those four attributes settle the number completely. Mesh alone spans three orders of magnitude in the same water, from tens of thousands per cubic metre at 20 µm to a few per cubic metre at 330 µm. Depth matters as much, for the reasons set out at point 6 below. Without those four attributes an expected range cannot be evaluated, and we would genuinely welcome the source: if it is a surface measurement at a comparable mesh in comparable water, it is the comparison the literature otherwise lacks, and we would adopt it.
We would add that his stated requirement of several cubic metres for the 20 to 200 µm fraction does not match routine practice, which quantifies that fraction from discrete water samples of one to a few litres rather than from cubic metres of filtered water. If we have misunderstood the basis of that figure we would be glad to be corrected.
2. Missing and inaccessible method detail
Accepted in full, and the inaccessible link is the worst of it. The referee reports an HTTP 403 on the field guide that carries the build details, which means he could not check the instrument at the centre of the paper. That is our fault and it is being fixed. We note that the first referee did reach the same page, since he cites the deployment video hosted on it, so the failure appears to be intermittent rather than a dead address. That is not a mitigation. A specification a referee cannot open when he needs it is a specification we have not published.
In revision the full cartridge specification will appear in the main text rather than in an appendix: dimensions, filter plate geometry, aperture count and diameter, filter medium and pore size, fill mechanism, and the deployment procedure step by step. The field guide link will be repaired and, because a live URL is not a durable citation, the guide will also be deposited with a DOI so that it cannot break again.
3. Intake geometry and avoidance
There is a geometry error here, and it is ours rather than his.
The intake is 50 mm in diameter, not 5 mm. The 5 mm holes the referee refers to are in the two acrylic plates at the base of the tube, downstream of collection, where the filter papers sit. They are the deposition and viewing apertures, not the orifice through which water is drawn. The 0.5 litre enters through the 50 mm mouth of the tube, and everything it carries is deposited onto the five 5 mm filter sections, which are then examined under the microscope. Those holes size-select nothing. They are where the sample ends up, not where it enters.
The first referee read the same drawing the same way, describing a 50 mm aperture but treating the filtration geometry as the limiting one. Two expert readers reaching the same conclusion about the same figure means Appendix A is not clear enough, and that is ours to fix. In revision the water path will be described step by step, and Figure A1 will label the intake and the filtration plates separately so that the two cannot be confused.
The referee’s underlying point survives in part, and we accept it in part. It also depends on how the cartridge fills, which the manuscript states only in part. The methods do record that the tube is flushed three times with ambient water before the sample is taken. What they do not say is that the tube is then dropped into the water, so that the sample is scooped by immersion rather than drawn by suction. No sustained intake flow exists for an animal to detect, and none for it to swim clear of before the tube is full.
A larger, strong-swimming animal may still have scope to escape a scoop, and for mesozooplankton we make no claim that the cartridge is quantitative. That is precisely why the zooplankton number comes from a net of 0.30 m mouth diameter, and why the net was put in the water at all. But avoidance requires an organism to sense the disturbance and to out-swim it, and that capacity falls away steeply with body size. It is not a plausible explanation for the 20 to 300 µm fraction, which is the fraction the near-absence claim actually concerns. In revision the filling mechanism will be described explicitly alongside the flushing procedure, since an intake that draws and one that scoops raise different questions.
4. Contamination from the vessel
The referee’s specific mechanism does not apply, but his underlying concern is sound and we accept part of it.
The cartridge is not a towed net. Appendix A describes a self-contained tube which, because it hangs from its line, filters vertically, and which takes a fixed 0.5 litre on each fill, independent of the vessel’s speed through the water. It fills once, it is recovered, and the filter stays capped from collection until analysis. It does not strain water continuously behind the hull in the way the referee’s description implies.
A correction to Appendix A. The text describes the cartridge as suspended on a Kevlar cord. During this survey it was suspended on stainless steel wire. Kevlar is the intended replacement, chosen for durability, and the appendix was written against the current design rather than the one that took these samples. That is our error and the appendix will be corrected to state stainless steel wire for the reported survey, with the change to Kevlar noted for future campaigns. It happens to bear on the referee’s point, since steel sheds no fibres, but we would be disclosing it either way.
Deployment. The sample is taken off the stern or the leeward side of the vessel. The manuscript does not say so, and it should. In revision the deployment will be specified exactly: the position relative to the hull, whether the vessel is making way or stopped, and the depth of immersion. We note plainly that stern water has passed the hull, so as it stands this is not a geometry that disposes of the referee’s concern. We will therefore specify the leeward side only for future sampling, so that the question does not arise.
We would also put the scale of the expected effect on the record. These are small sailing yachts working under sail, with no engine running while the sample is taken. The wetted area is a small fraction of a research vessel’s, there is no cooling-water discharge, no exhaust plume and no propeller wash, and the volume of water disturbed is correspondingly small. On that reasoning we expect any vessel-derived contribution to be very small. It is reasoning and not measurement, and we offer it as such, since we have not put a number on it.
Handling. The risk we identified as the largest was the operator rather than the vessel: synthetic clothing worn by whoever takes the sample. The protocol addresses it directly, and both controls are already in the methods. The tube is flushed three times with ambient water before the sample is taken, which displaces anything carried from the deck or retained from the previous station, and it is capped as soon as it leaves the water and uncapped only immediately before examination, so the sample spends no time open on deck. Since the referee has read the paper closely and still reached the contamination question, these controls are evidently not prominent enough where they sit. In revision they will be gathered into a single contamination-control paragraph rather than distributed through the method description.
What the fibre sizes say. Cordage is the near-field source the referee would expect to dominate, and a fibre shed from a rope in use is of the order of millimetres. The fibres we recovered are far smaller than that, with plastic fibres forming a distinct mode at a median length of 204 µm, which is the wrong size distribution for rope shedding. Ropes that are not under load or running are in any case not shedding. None of this is proof, since a fibre can fragment, but the sizes run against the mechanism rather than for it.
Antifouling. Cruising yachts increasingly carry hard antifouling coatings rather than ablative ones, and a hard coating is not designed to slough. We do not treat hull coating as the leading contamination risk for this reason. That is as far as we can take it, since we did not test the coatings.
Blanks. The blanks were run by the lead author rather than by the crews, and they found no contamination of the samples. They tested for airborne contamination, which is all we claim for them.
What we accept without reservation:
- The blanks addressed airborne contamination, not waterborne contamination from the vessel itself.
- Antifouling paint, deck runoff, degrading cordage and sailcloth, and exhaust deposition are all plausible near-field sources, and we did not test for them.
- We cannot presently quantify what fraction of the particle load is vessel-derived.
We would add one observation about the principle rather than the case. The referee writes that every vessel is a significant point source of local contamination, and we agree. That applies with more force to a research ship than to a sailing yacht, since a large vessel carries more machinery, more coatings, more paint and more waste streams, and it disturbs a far larger volume of the layer being sampled. This is not an argument that our samples are clean. It is an argument that the problem he identifies is general to shipborne sampling of the surface layer, and it is not solved by using a bigger vessel.
In revision this will be stated as a limitation in its own right rather than folded into the general caveats, and the particle concentrations will be described as an upper bound on the environmental load. The next campaign will carry a vessel-source test: paired samples taken upwind and upstream of the vessel while hove to and while under way, with paint, cordage and sailcloth reference spectra held for comparison against the confirmed particle spectra. If a vessel signature exists, that design will find it.
5. The hand-count comparison
This point began as an allegation about our conduct, and the honest answer to it turned out to be larger than the sentence the referee quoted. In re-analysing the archive we examined the crew hand-logs in full, and they disagree with the image analysis in both directions: two crews logged totals of 973 and 1,148 “zooplankton” that the archived micrographs do not support at any size, while other crews logged zero. Visual identification of small, degraded particles by non-specialists is not reliable, and the first version of the manuscript should not have implied that a single hand-counted leg constituted validation. The revision reports the hand-logs for what they are, qualitative context retained in the deposited workbook, and rests the zooplankton measurement on the net tow, where identification is not in doubt. We did not bury inconvenient validation; but we did lean on a validation that could not bear weight, and the referee was right to press on it.
6. What the net tow shows
This is the substantive scientific disagreement and we set out the evidence.
Before doing so we correct something in our own earlier reply. In responding to the first referee we stated that no surface-only mesozooplankton measurement exists for this water. Further searching since has located one, the Tara Pacific surface datasets described below, and we correct that statement here rather than leave it standing. It was our slip and we would rather flag it ourselves than have someone else find it.
The referee states that our net-tow concentrations “fall within the expected range for the studied region” and that “the net tows detected normal zooplankton levels”. No source is given for that expectation, and we do not think the literature supports it.
It is worth noting what is not in dispute. The referee accepts the net-tow measurement itself; his argument is about what 0.2 to 0.5 animals per m³ means, not about whether we measured it. That narrows the disagreement to a single question, which is what counts as normal at the surface.
The comparison has to be surface against surface. In stratified oligotrophic water the animals are not in the top metre. Comparing our surface tow against a depth-integrated value is the error the first referee identified in our own manuscript, and we have accepted it there. Taking published surface data against a published 0 to 200 m integration for the same trophic regime, the top decimetre to metre of the daytime oligotrophic ocean holds of order 1 to 6% of the mean 0 to 200 m concentration. A 0 to 200 m value of about 200 ind m⁻³ (Fernández de Puelles et al., 2019, 300 µm, oligotrophic tropical and subtropical ocean) therefore scales to a daytime surface expectation of roughly 2 to 12 ind m⁻³, not to 200.
Against true surface data, our values are low and not ordinary. The Tara Pacific expedition published quantitative datasets of the top 0 to 1 m sampled in daylight across the Pacific and North Atlantic (Mériguet et al., 2025). The values below are computed by us from the archived data rather than quoted from the paper, which prints no summary statistics; the datasets are open at SEANOE (https://doi.org/10.17882/102336 and https://doi.org/10.17882/102537) and the computation is reproducible.
With the 330 µm high-speed net, the median total metazoan mesozooplankton concentration in the North Atlantic subtropical gyre is 2.20 ind m⁻³ and the lowest station is 0.76; across 130 tropical and subtropical surface stations the median is 3.82 ind m⁻³. The authors describe that net as semi-quantitative and report it undersampling by roughly a factor of four relative to a conventional neuston net, which is why we also give the second gear: the 333 µm Manta neuston net on the same expedition returns a tropical median of 11.4 ind m⁻³ with a minimum of 2.23. Correcting the high-speed values for the stated bias moves them towards the Manta figures rather than towards ours.
Our value is 0.2 to 0.5 ind m⁻³, and it was obtained with a 120 µm mesh. Those Tara nets are 330 and 333 µm. A finer mesh retains more, not less, so a like-for-like comparison would raise our number relative to theirs rather than lower it. On the only published true-surface dataset for tropical and subtropical water, our concentrations sit at or below the bottom of the range, measured with gear that should have read higher.
What we cannot claim. No published study reports total mesozooplankton in the top few metres with a 100 to 200 µm mesh in the oligotrophic tropical North Atlantic. The exact measurement that would settle this does not exist, which is why the survey was undertaken. We therefore say that our values are low against the closest available surface data, and we do not say that a comparison of record has been made.
In revision the comparison will be rebuilt on surface data only, with Tara Pacific as the principal reference, the mesh difference stated in the same sentence as the ratio, and the depth-integrated values used only after the surface fraction has been applied. The wording “far below densities typical of productive open ocean” will go, because productive open ocean is the wrong baseline and we should not have used it.
On the conclusion
The referee’s closing charge is that the data do not support a near-absence of surface zooplankton or a causal link with pollution. On the second we agree, and we will say so more clearly: the manuscript reports a co-occurrence and does not establish causation, and any wording implying otherwise will be removed.
On the first, the re-analysis has made the evidence simpler than it was. What the revision claims is what the data carry: a reproducible surface particle inventory across a basin that has not been surveyed this way before; no zooplankton confirmed in any of 1,225 curated filter samples; and a net-tow surface mesozooplankton concentration at or below the lowest values in the published surface record, measured with a finer mesh than those records used.
That concession is about what this survey can prove, and not about whether the question is a reasonable one to ask. The mechanism is not ours and it is not speculative. Zooplankton ingest microplastic, which was shown directly by Cole et al. (2013) across thirteen taxa. Plastic leachate suppresses growth and oxygen production in Prochlorococcus, the most abundant photosynthetic organism in the ocean (Tetu et al., 2019). Plastic surfaces adsorb lipophilic organic contaminants and carry them, and the affinity and the leachate toxicity both rise as the particle weathers (Liu et al., 2020; Ferrari et al., 2024). Every step of that chain is in the literature. What is missing is a measurement of exposure in the surface layer where the particles and the animals actually meet, and that gap is the one this survey was built to close. A monitoring survey establishes the exposure. It does not establish the effect, and we will not write as though it does.
A word on why the survey was designed this way, since it bears on the referee’s view of what the data are worth. Plastic particle analysis was the primary purpose of the crossing, and the buoyant fraction can only be measured at the surface. Density decides this, and it does not follow the everyday division between soft and hard plastics. Polyethylene at roughly 0.91 to 0.97 g cm⁻³ and polypropylene at roughly 0.90 to 0.92 float in seawater of about 1.025, and those two are the largest-volume polymers ever produced (Geyer et al., 2017). Expanded polystyrene floats. Polyester and PET at about 1.38, PVC from 1.16 upwards, acrylic at about 1.18 and ABS at about 1.05 all sink, so they leave the surface layer and are lost to a surface survey from the start. Biofouling eventually takes down much of what floats as well, which makes the surface a residence time rather than a permanent home for the buoyant fraction. Whatever is going to be seen there has to be caught there.
The vertical structure of that fraction is measured, not assumed. Buoyant plastic concentrates in the top few centimetres and falls away with depth as the wind mixes it down, so a sample taken below the immediate surface, or taken in a blow, returns a lower number than the water actually holds (Kukulka et al., 2012; Reisser et al., 2015). Add the mesh. The conventional instrument is a manta trawl of about 333 µm towed at the surface or a net worked deeper, and everything smaller than the mesh passes through it. Our filters resolve to 20 µm. Between the depth and the mesh, the great majority of the particles reported here would not appear in a conventional survey at all, which is the reason the survey was run rather than a defence of it after the fact.
There is a second point in the same area which we think matters more, and it has had almost no attention. Our filters recovered partially combusted carbon particles alongside the plastic, and the carbon, not the plastic, dominated the load. That is the one classification the hand count and the automated pipeline agreed on without qualification. Almost no marine plastics survey counts these particles, and most sampling designs would not retain them in the first place. Yet black carbon is among the strongest sorbents of hydrophobic organic compounds known in the environment, binding them one to two orders of magnitude more strongly than amorphous organic matter (Cornelissen et al., 2005). The comparison is worth setting beside Koelmans et al. (2016), who concluded from modelling that the plastic vector is modest relative to other exposure pathways. We cite that paper knowing it cuts against the strongest version of the plastic argument, because it also points somewhere uncomfortable: if plastic is a modest sorbent and is nevertheless the particle the community counts, then the stronger sorbent in the same water is going almost entirely unmeasured.
We are not claiming to have demonstrated that soot is toxic to marine life, and we have run no experiment that could. We are saying that a particle class with a documented affinity for exactly the compounds that concern us is present in the surface ocean in quantity, that hardly anyone measures it, and that the question of what it does there is open and answerable. In revision we will state that as a finding of the survey in its own right, since a gap of that size in marine monitoring is a result and not an aside.
A last word on where we think this leads, because it bears on how the paper should be weighed. The value of the approach is not this one crossing. It is the access an extended network could give the wider community: a large and continuously growing body of water and marine science data, from places and times that funded programmes do not reach, available to any academic who wants to use it. That only works on two conditions, and both referees have effectively named them. The filter has to stay simple and cheap enough that volunteers can use it everywhere, from a rain gauge to a mid-ocean passage. And the sampling procedure has to be repeatable to a standard that lets one crew’s numbers be set against another’s, and this year’s against next year’s. Read that way, the criticisms in this review are a specification for the second version of the protocol, and we are grateful for them.
Summary of changes
- The entire image analysis re-run as a deterministic, deposited pipeline (Zenodo v2 of 10.5281/zenodo.21477682): frame curation with a logged reason for every excluded file, detection, sizing, classification, and individual inspection of every object ≥250 µm. The 44 animal-class objects of the first version are withdrawn; the corrected result is no zooplankton confirmed in any filter sample. All particle totals, Table 1, Table 3 and Figures 2 to 4 regenerated from the curated archive.
- Net-tow method, mesh and filtered volume stated in the abstract and at first mention of any density; the cartridge described only as a particle sampler, with an explicit statement that no zooplankton abundance is derived from it.
- Full cartridge specification moved into the main text, with the water path described step by step and Figure A1 labelling the 50 mm intake and the 5 mm filtration plates separately; the field guide link repaired and the guide deposited with a DOI.
- Deployment geometry specified exactly, with leeward-side sampling specified for future campaigns, and the contamination controls already in the methods, the three ambient flushes and the capping of the tube on leaving the water, gathered into a single paragraph where a reader will find them. Appendix A corrected to state stainless steel wire for the reported survey, with the change to Kevlar noted for future campaigns. Vessel-derived contamination stated as a limitation in its own right; particle concentrations described as an upper bound on the environmental load; a vessel-source test written into the next campaign.
- Section 2.7 rewritten to disclose the re-analysis in full and to report the crew hand-logs as qualitative context that disagrees with the imagery in both directions; the zooplankton measurement assigned to the net tow alone.
- The zooplankton comparison rebuilt on published surface data, with the mesh difference stated alongside every ratio and depth-integrated values used only after the surface fraction is applied.
- “Far below densities typical of productive open ocean” removed.
- All wording implying a causal link between the particle load and the zooplankton concentration removed.
- The near-absence of combustion carbon from marine particle monitoring stated as a finding of the survey in its own right, with the sorption literature that makes it consequential.
- The rationale for a true-surface, 20 µm sampling design stated explicitly in the introduction: polymer density, the measured vertical distribution of the buoyant fraction, and the mesh of the conventional instrument.
We are grateful for the review. The challenge to the image classification led directly to the deposited re-analysis, which is the single largest improvement this paper has received, and the broken link should never have reached a referee.
Howard Dryden and Diane Duncan GOES Foundation, Seahorse Point Nature Laboratory, Bocas del Toro, Panama
References cited in this response
Cole, M., Lindeque, P., Fileman, E., Halsband, C., Goodhead, R., Moger, J., and Galloway, T. S.: Microplastic ingestion by zooplankton, Environmental Science & Technology, 47, 6646–6655, https://doi.org/10.1021/es400663f, 2013.
Cornelissen, G., Gustafsson, Ö., Bucheli, T. D., Jonker, M. T. O., Koelmans, A. A., and van Noort, P. C. M.: Extensive sorption of organic compounds to black carbon, coal and kerogen in sediments and soils: mechanisms and consequences for distribution, bioaccumulation and biodegradation, Environmental Science & Technology, 39, 6881–6895, https://doi.org/10.1021/es050191b, 2005.
Fernández de Puelles, M. L., Gazá, M., Cabanellas-Reboredo, M., Santandreu, M. del M., Irigoien, X., González-Gordillo, J. I., Duarte, C. M., and Hernández-León, S.: Zooplankton abundance and diversity in the tropical and subtropical ocean, Diversity, 11, 203, https://doi.org/10.3390/d11110203, 2019.
Ferrari, M., Laranjeiro, F., Sugrañes, M., Oliva, J. M., and Beiras, R.: Weathering increases the acute toxicity of plastic pellets leachates to sea-urchin larvae, a case study with environmental samples, Scientific Reports, 14, https://doi.org/10.1038/s41598-024-60886-x, 2024.
Geyer, R., Jambeck, J. R., and Law, K. L.: Production, use, and fate of all plastics ever made, Science Advances, 3, e1700782, https://doi.org/10.1126/sciadv.1700782, 2017.
Koelmans, A. A., Bakir, A., Burton, G. A., and Janssen, C. R.: Microplastic as a vector for chemicals in the aquatic environment: critical review and model-supported reinterpretation of empirical studies, Environmental Science & Technology, 50, 3315–3326, https://doi.org/10.1021/acs.est.5b06069, 2016.
Kukulka, T., Proskurowski, G., Morét-Ferguson, S., Meyer, D. W., and Law, K. L.: The effect of wind mixing on the vertical distribution of buoyant plastic debris, Geophysical Research Letters, 39, L07601, https://doi.org/10.1029/2012GL051116, 2012.
Liu, P., Zhan, X., Wu, X., Li, J., Wang, H., and Gao, S.: Effect of weathering on environmental behavior of microplastics: properties, sorption and potential risks, Chemosphere, 242, 125193, https://doi.org/10.1016/j.chemosphere.2019.125193, 2020.
Mériguet, Z., et al.: Quantitative imaging datasets of surface micro- to mesoplankton communities and microplastic across the Pacific and North Atlantic oceans from the Tara Pacific expedition, Earth System Science Data, 17, 2761–2792, https://doi.org/10.5194/essd-17-2761-2025, 2025.
Reisser, J., Slat, B., Noble, K., du Plessis, K., Epp, M., Proietti, M., de Sonneville, J., Becker, T., and Pattiaratchi, C.: The vertical distribution of buoyant plastics at sea: an observational study in the North Atlantic Gyre, Biogeosciences, 12, 1249–1256, https://doi.org/10.5194/bg-12-1249-2015, 2015.
Tetu, S. G., Sarker, I., Schrameyer, V., Pickford, R., Elbourne, L. D. H., Moore, L. R., and Paulsen, I. T.: Plastic leachates impair growth and oxygen production in Prochlorococcus, the ocean’s most abundant photosynthetic bacteria, Communications Biology, 2, https://doi.org/10.1038/s42003-019-0410-x, 2019.
-
RC5: 'Comment on egusphere-2026-3979', Anonymous Referee #3, 20 Aug 2026
reply
I have had very similar concerns as have the other reviewers and see that the authors agree that the manuscript needs some serious modification to address those comments. I am delighted to review the revised manuscript as soon as it has been submitted but do not think it very useful for me to make comments on material which is about to change.
Citation: https://doi.org/10.5194/egusphere-2026-3979-RC5
Video supplement
Sargassum in middle of Atlantic Howard Dryden https://youtu.be/TFcTT6AgExs
s/v Copepod in Atlantic trade winds about 500 nm from Grenada Howard Dryden https://youtu.be/zoW-CoUPbps
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 308 | 80 | 11 | 399 | 24 | 17 | 19 |
- HTML: 308
- PDF: 80
- XML: 11
- Total: 399
- Supplement: 24
- BibTeX: 17
- EndNote: 19
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
This MS provides results and interpretation from a new, low-cost method for quantifying the near-surface abundance of plankton and pollutant particles, with focus on the tropical North Atlantic. An overall framing is the hypothesis that pollution-induced reductions in plankton abundance will deleteriously affect the ocean’s surface microlayer, significantly changing the rate of water evaporation to the atmosphere and hence the rate of climate warming. Samples for analysis were collected by ‘citizen science’ from sailing yachts, with on-board camera-based characterisation of particles and abundance measurements.
The study is well-intentioned, addressing a significant data gap in our knowledge of plankton biogeography and open-ocean pollution measurements for this part of the global ocean. Unfortunately the MS is flawed in many regards. An overall weakness is an apparent lack of awareness - with implications for planning, implementing and interpreting the study - of the fact that the North Atlantic tropical gyre is one of the most oligotrophic (least productive) ocean areas in the world, due to strong, year-round thermal stratification and nutrient depletion in the upper ocean. This feature is clearly shown on satellite maps of ocean productivity (e.g. https://en.wikipedia.org/wiki/Marine_primary_production). The following consequences are directly relevant to this MS:
Since the depth-distributions of zooplankton are, not surprisingly, closely linked to their food supply of phytoplankton, the GOES sampling device is inappropriate for quantitative assessment of the abundance of either group of organisms. Furthermore:
The main finding reported in the MS – the lack of zooplankton - is therefore not a novel result. It could, however, be predicted from existing knowledge of the biology of the North Atlantic subtropical gyre.
This reviewer has other significant concerns regarding the scientific quality of the MS. In particular:
1. Invalid comparisons are made with other zooplankton abundance data for the tropical North Atlantic. Three papers are cited to justify the statement (lines 92-93) that observed near-surface zooplankton densities are two to three orders of magnitude below published values for the region: Woodd-Walker et al., (2000); Vereshchaka et al., (2017); and Fernandez de Puelles et al., (2019). However, the first of these papers does not exist; if it is intended to be Gallienne et al (2001), their sampling region was much further north (35- 60 degN), in a different biogeochemical province; the second of these papers also avoids direct sampling of the oligotrophic gyre. Whilst the third is potentially relevant, their ‘surface’ sampling integrates over water depth of 200m; i.e. including the DCM, where zooplankton abundances are much greater.
2. Insufficient evidence is provided to assess the validity of the abiotic particle characterisation. The detection and identification of microplastics and black carbon is not straightforward, with many problems associated with reliable quantification (Mutuku et al, 2024). In the MS, the statement is made that particle identification is “awaiting spectroscopic confirmation” (line 230); such validation is essential for the credibility of the results presented. For example, Loder & Gerdis (2013) found that only 1.4% of the particles visually representing microplastics were of synthetic origin. The MS notes that floating Sargassum was frequently encountered: could a proportion of the particles be degraded macro-algal fragments?
3. Throughout the MS there is ambiguity regarding the meaning of “surface ocean”: sometimes this refers to the surface microlayer, less than 1 mm thick, at other times to the upper ~200m. Whilst there is a lack of standardised definition in the wider scientific literature (and also for the terms “upper ocean” and “epipelagic”), care is needed to distinguish whether the microlayer, the top 1-5 m or other meanings are intended. That is because there can be steep gradients in water physics, chemistry and biology occurring within and between those water layers, particularly in highly-stratified tropical waters.
4. As initially noted above, the overall framing of the MS is provided by the hypothesis that the ocean microlayer significantly influences the global climate by affecting evaporation rate (and hence water vapour concentrations in the atmosphere). The evidence provided to support that argument is unconvincing: according to conventional physics, the scale of that effect is trivial, with temperature, wind-speed and ocean area being much more important by several orders of magnitude. Detailed evaluation of this aspect of the MS is beyond the expertise of this reviewer; however, the potential importance of this effect - if the MS authors’ hypothesis were correct, albeit that seems unlikely – is such that it requires a separate publication, focussed only on that issue, and subject to specialist scrutiny.
5. Inferences are made about changes in plankton abundances (and pollution levels) on the basis of a single survey. Whilst such changes may have occurred, and/or currently be underway, a time series – preferably of a decade or longer - is needed to distinguish seasonal variability, episodic events that may temporarily increase productivity (due to dust-deposition or hurricanes) and long- term trends due to climate change or direct anthropogenic forcing. The AMT transects (Rees et al., 2015) and Bermuda Atlantic Time Series Study (Madin et al, 2001) provide such context, particularly when linked to remotely-sensed data; however, those analyses do not support the conclusions of this MS.
Taken together, the above comments strongly question the suitability of this MS for publication in a reputable scientific journal. Whilst there can be circumstances for citizen science to contribute to marine pollution monitoring (e.g. Hidalgo-Ruz & Thiel, 2015) and oceanography more widely (Simoniello et al, 2019), such studies need to be planned with scientific rigour. Those leading and carrying out such research need to either have relevant expertise themselves or consult/collaborate more widely for valid results to be obtained: otherwise there is high risk of wasted effort.
ADDITIONAL REFERENCES GIVEN ABOVE excluding those included in MS
Gallienne, C.P., Robins, D.B. and Woodd-Walker, R.S., 2001. Abundance, distribution and size structure of zooplankton along a 20 west meridional transect of the northeast Atlantic Ocean in July. Deep Sea Research Part II: Topical Studies in Oceanography, 48, 925-949.
Hidalgo-Ruz, V. and Thiel, M., 2015. The contribution of citizen scientists to the monitoring of marine litter. Chapter 16 in Marine Anthropogenic Litter, (ed: M Bergmann et al), Springer, pp 429-447.
Jonas, T.D., Walne, A., Beaugrand, G., Gregory, L. and Hays, G.C., 2004. The volume of water filtered by a Continuous Plankton Recorder sample: the effect of ship speed. Journal of Plankton Research, 26), 1499-1506.
Löder, M.G. and Gerdts, G., 2015. Methodology used for the detection and identification of microplastics—a critical appraisal. Chapter 8 in Marine Anthropogenic Litter (ed: M Bergmann et al), Springer, pp 201-227.
Madin, L.P., Horgan, E. F., and Steinberg, D.K., 2001. Zooplankton at the Bermuda Atlantic Time-series Study (BATS) station: diel, seasonal and interannual variation in biomass, 1994-1998. Deep Sea Research II, 48, 2063-2082.
Mutuku, J., Yanotti, M., Tocock, M. and Hatton MacDonald, D., 2024, June. The abundance of microplastics in the world’s oceans: A systematic review. Oceans, 5, 398-428.
Rees, A., Robinson, C., Smyth, T., Aiken, J., Nightingale, P. and Zubkov, M., 2015. 20 years of the Atlantic Meridional Transect-AMT. Limnology and Oceanography Bulletin, 24, 101-107.
Simoniello, C., Jencks, J., Lauro, F.M., Loftis, J.D., Weslawski, J.M., Deja, K., Forrest, D.R., Gossett, S., Jeffries, T.C., Jensen, R.M. and Kobara, S., 2019. Citizen-science for the future: Advisory case studies from around the globe. Frontiers in Marine Science, 6, 225.
Tilstone, G.H., Lange, P.K., Misra, A., Brewin, R.J. and Cain, T., 2017. Micro-phytoplankton photosynthesis, primary production and potential export production in the Atlantic Ocean. Progress in Oceanography, 158, 109-129.