the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 21 Jul 2026
Chlorophyll–particle backscattering displacement reveals multiple deep chlorophyll maximum structures in the North Pacific
Abstract. Subsurface chlorophyll-a (Chl) maxima are common features of stratified oceans, but a Chl maximum does not necessarily indicate a maximum in particle-associated optical structure. Here we use Biogeochemical-Argo (BGC-Argo) observations from the North Pacific during 2022–2025 to examine how the previously recognized separation between Chl maxima and particle-associated optical structure is expressed across a large set of vertically resolved in situ profiles. For 10,801 usable paired Chl and particulate backscattering at 700 nm (BBP700) profiles, we diagnosed the depths of the Chl DCM (deep chlorophyll maximum) and BBP700 maximum within the upper 300 dbar and classified profiles by their vertical displacement. Coupled profiles, in which the two maxima occurred within 25 dbar of each other, were the most frequent class and had median Chl DCM and BBP700 maximum depths of 52.5 and 47.5 dbar, respectively. In Chl deeper profiles, the Chl DCM was displaced downward relative to a shallower BBP700 maximum, consistent with a stronger contribution of photoacclimation or pigment adjustment to the Chl maximum. In Chl shallower profiles, a shallow Chl maximum coexisted with a deeper BBP700 maximum, demonstrating that Chl–BBP700 decoupling is not unidirectional. The three displacement classes showed coherent basin-scale spatial organization and distinct nitracline contexts, with the nitracline generally deeper than both optical maxima in Coupled and Chl deeper profiles, whereas Chl shallower profiles tended to show a deeper BBP700 maximum near the nitracline depth range. Together, these findings indicate that North Pacific Chl DCMs do not represent a single uniform vertical bio-optical structure. Instead, the previously recognized separation between Chl maxima and particle-associated optical structure is expressed in mature BGC-Argo observations as coherent vertical, spatial, and nitracline-related structures across the North Pacific.
- Preprint
(3705 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 01 Sep 2026)
- AC1: 'Author Comment: correction of an outdated profile count in the manuscript text', Taketo Hashioka, 22 Jul 2026 reply
-
RC1: 'Comment on egusphere-2026-3910', Emmanuel Boss, 02 Aug 2026
reply
Review of ‘Chlorophyll–particle backscattering displacement reveals multiple deep chlorophyll maximum structures in the North Pacific’ By Hashioka and Nita.
Reviewer: Emmanuel Boss, University of Maine.
The authors classify profiles of chlorophyll (chl) and particulate backscattering at 700nm (bbp) in the North Pacific to those that have maxima within 25m, those that have a chl maxima deeper than 25m relative to bbp and those that have chl maxima shallower by more than 25m compared to bbp. They find the majority to have chl maxima deeper or within 25m from bbp maxima with a much smaller fraction having bbp maxima deeper than the chl maxima. The authors find that in all cases the nutricline is deeper than the bbp maxima and chl maxima being closest to the bbp maxima when the chl maxima is shallower.
This paper is well written an clear, however I find it to provide a very limited and rather descriptive contribution to the literature on the subject the reasons for which I outline below. For this work to be a significant contribution, significant more work needs to be done.
- Two of the most important forcing functions are missing from the analysis: a. stratification (density structure) and b. light distribution. It is well known that there is a relationship between the distribution of chl and light availability, such that when chl is higher, light penetrates less deep, confining phytoplankton nearer to the surface. Using physical depth distances in the analysis, rather than (or together with) optical depths, limits our ability to build an understanding of the ‘coupled’ profiles, the largest class observed here. Stratification, and in particular the presence of a chlorophyll maxima within the ML (assuming NPQ correction is applied), is also an important diagnostic of the vertical structure. During the phytoplankton accumulation phase (‘bloom’) in sub-polar regions the maxima in biomass as well as chl are typically within this ML. Data of stratification is readily available for all profiles and the relevant light attenuation (Kd(PAR)), if not available from coincident light profiles (now available on some BGC-Argo), can be computed from Chl distributions.
- The justification for not doing this analysis worldwide or not using data from before 2022 is weak at best. If done, it could provide significant more insight on regions of the world’s ocean having similar biogeochemistry. In general, lumping profiles within biogeochemically consistent regions (biomes) would allow you to obtain consistent seasonal cycles (as Longhurst has done) from which deeper understanding could be gleaned regarding the underlying processes.
- No attempt has been done to classify profiles with respect to their shape (other than the depth of a maxima). There is a wide literature attempting to fit a vertical structure to chl profiles as these often have salient feature. Here there are no profiles displayed to the point where one wonder at the robustness of the analysis, for example to 2D processes which can result in multiple maxima, advective features associated with the base of the ML etc’.
- Also, given that spikes and noise can affect Argo profiles (due to a variety of causes, including zooplankton), one wonders to what degree they could have affected the results (particularly bbp deep).
I realize I am asking for significantly more work to convince me of the importance of your contribution. I also realize that not every published paper is such and that often papers provide a progress report. However, given the large literature on this subject, the current manuscript adds too little in my mind. I am attaching an annotated PDF with additional comments I wrote as I was reading the manuscript some of which are already mentioned above.
Dear authors, I am often wrong. If you feel that my comments are misplaced feel free to contact me and if convinced, I will be more than happy to change them. All the best, Emmanuel Boss
-
AC2: 'Reply on RC1', Taketo Hashioka, 14 Aug 2026
reply
AUTHOR COMMENT
0. Preamble
We thank the reviewer for a detailed reading, and in particular for the annotations, which identify specific places where the manuscript claims more than it establishes.
This comment is a first response rather than a complete one. Where we already have the measurement, we give the number here. Where we do not, we describe what the revised version will do, in terms specific enough that the reviewer can judge whether it answers the point. We have not waited until all of it is done, because several of the comments bear on how the analysis should be organized, and we would rather agree on that before rebuilding the figures.
Two things have changed as a result of this reading. The numbers in this comment come from a corrected processing series; we describe the correction where it arises, in Sect. 4. That correction also supersedes our own author comment of 22 July 2026, which gave the classification population as 10,740 and attributed the difference from 10,801 to the basin-mask step; both statements are withdrawn. We have also moderated several statements, including in the abstract, and identify those changes where they arise below.
We also say at the outset what we take the contribution to be, since several annotations question it. We do not claim the separation of the chlorophyll and particle maxima as new; Sect. 5 says so against the specific literature, which we have read and will cite. What the manuscript adds is a paired-profile diagnostic that resolves the relative vertical organization of the two optical maxima, and their position with respect to the nitracline measured on the same profile, across a large in situ population and at basin scale. One indication that this diagnostic carries additional information is reported in Sect. 5: the fraction of profiles in which the nitracline lies strictly between the two maxima differs by nearly an order of magnitude between the classes, and that ordering is preserved at every displacement threshold we examined.
We accept most of what the reviewer asks. We do not accept a global extension of the analysis, and we give our reasons in Sect. 2; separately, we accept the request that the classes be related to ecological structure rather than presented as a map.
Summary of the main changes.
- All figures are recomputed on a single processing series; the corrected panel sample sizes are 7,279, 2,956 and 566 (Sect. 4).
- Our author comment of 22 July 2026 is superseded; the classification population is 10,801 (Sect. 4).
- Terminology is changed: within the Chl shallower class we write "shallow chlorophyll maximum" rather than "Chl DCM", and "demonstrating" becomes "indicating" (Sect. 4 and 5).
- The novelty claim is narrowed, and the earlier literature that reports the same configuration is cited (Sect. 5).
- Stratification context is added for every class, and optical context is reported on the subset with measured photosynthetically available radiation (Sect. 1).
- The analysis is extended in time to every year with usable paired profiles; a global extension is declined, with reasons (Sect. 2).
- The classes are stratified by North Pacific biome and season, and a float-block permutation test of spatial aggregation is added (Sect. 2).
- Spikes are addressed by an explicit spike criterion rather than by wider smoothing, and the sensitivity to the smoothing and to the prominence threshold is reported (Sect. 4).
- Figures of individual profiles and of the attrition of the profile population are added (Sect. 3 and 5).
- Advection and vertical shear are stated as alternatives, and the implication that the classes correspond to distinct local mechanisms is removed (Sect. 3).1. Stratification and light
The reviewer asks that the displacement classes be placed in a physical context — the stratification that supports a subsurface maximum, and the light field that sets its depth. We accept that the manuscript describes vertical structure without either, which currently limits its physical and mechanistic interpretation. We treat the two requests separately, because what we can measure differs.
Stratification is directly available. Temperature and salinity are recorded on every profile in our population, so the mixed layer depth and the buoyancy frequency can be computed for all 10,801 profiles without additional data. The revised version will report, for each displacement class, the mixed layer depth, the depth of maximum stratification, and the position of the chlorophyll and BBP700 maxima relative to both. This also answers a question the present classification leaves open: how many of the shallow chlorophyll maxima in the Chl shallower class lie within the mixed layer rather than below it. Preliminary analysis indicates that this class is associated with the deepest mixed layers ; the revised analysis will test this explicitly rather than infer the mechanism in advance. The prominence of both maxima will be reported by class alongside the quantities above, so that this can be judged from the data rather than assumed.
The light field is less straightforward, and we prefer to state the difficulty rather than present a diagnostic that conceals it. Radiometers are carried by a minority of the floats in this population, so an optical depth computed for all profiles would have to be derived from the chlorophyll profile itself through a chlorophyll-to-Kd(PAR) model of the kind the reviewer suggests. Normalizing a chlorophyll-derived depth diagnostic by a chlorophyll-derived optical depth would not provide a fully independent optical diagnostic, because both quantities would ultimately depend on the chlorophyll profile. We will therefore report the optical context on the subset of profiles with measured photosynthetically available radiation, where the euphotic depth is observed rather than modeled, and use that subset to assess whether the modeled quantity behaves consistently before applying it more widely. If it does not, we will report the stratification context alone and say so.
2. Regional scope
The reviewer asks why the analysis is confined to the North Pacific and whether the displacement classes could be organized into a province-type synthesis of the kind associated with Longhurst. We treat these as two distinct issues — geographic extent and ecological organization — because our response differs.
Global extent
On geographic extent, a global BGC-Argo assessment of deep chlorophyll maxima already exists. Cornec et al. (2021), which we cite, characterized the occurrence, drivers and characteristics of DCMs across the global float array and distinguished maxima associated with particle structure from those associated with photoacclimation. A biogeographical partition of the global ocean built on the same two variables also exists: Bock et al. (2022) applied empirical orthogonal function and cluster analysis to annual cycles of chlorophyll fluorescence and particulate backscattering profiles and identified six biomes, four of which are characterized by the depth and frequency of deep chlorophyll maximum formation. We will add this reference. A global extension would therefore substantially overlap with existing global assessments and would move the manuscript away from its intended North Pacific basin-scale contribution.
Extending the paired in situ analysis globally is also not a matter of relaxing a geographic filter. The present analysis requires a Chl and a BBP700 profile from the same cycle, both passing quality control and both resolving the upper 300 dbar, and Fig. 1 shows how recently that combination became available even within a single basin: the number of grid cells with at least three usable paired profiles in the North Pacific rose from roughly 100 in 2010–2019 to roughly 650 in 2022–2025. Establishing and verifying an equivalent sampling basis for all ocean basins would require a substantially broader analysis, which we regard as future work.
Temporal extension
The reviewer's request that we use more of the available data has a second part, about time rather than space, and here we accept it. We restricted the analysis to 2022–2025 because that is the period in which paired Chl and BBP700 profiles are dense enough to grid the basin, as Fig. 1 shows, and not because we regard the earlier observations as unusable or the ocean as having changed between periods. The revised version will not be limited to that window: we will use every year for which usable paired profiles exist in the North Pacific study domain, and we will report the displacement classes for the full record as well as for the recent period separately. Because the earlier years are spatially much more sparsely sampled, the basin-scale gridded maps will continue to rely primarily on the recent period. The extended record will instead test whether the class fractions and depth characteristics obtained for 2022–2025 are also evident in the earlier observations. If they are not, we will report that rather than presenting the full record as a single population.
Ecological organization
On ecological organization we accept the reviewer's point. The manuscript presents the spatial distribution of the classes but does not relate it to any ecological partition, and the revision will add that context within the North Pacific. The revised version will stratify the displacement classes by North Pacific biome — subtropical, transition and subarctic — and by season, and will report class fractions, depth medians and nitracline context within each stratum. This addresses the substance of the request, which is that the classes should be shown to correspond to recognized ecological settings rather than to appear as an unstructured map.
Part of this is already measured and we report it here. The Chl shallower class is not distributed uniformly with latitude. Its median latitude is 37.5°N, and 43.8 % of these profiles lie poleward of 40°N compared with 18.1 % of all profiles; the class fraction rises from 0.3–2.2 % between 10 and 25°N to 11.3–14.2 % between 40 and 55°N. These profile-level statistics, which do not pass through the gridded cells, indicate that the class is enriched at higher latitudes while not being restricted to them: the tenth percentile of its latitude distribution is 4.8°N, and 63 of these profiles lie between 0 and 5°N.
We also accept that the manuscript asserts coherent basin-scale spatial organization on the basis of visual inspection of Fig. 4, without a test. The revised version will report a test of spatial aggregation of class fractions using float-block permutation, since the 10,801 profiles were collected by 149 floats and are not independent at the profile level. One annotation suggests that the profiles may be regarded as independent realizations of the ocean state; they do sample distinct states, but they are not independent for a test of spatial aggregation. Because individual floats contribute repeated profiles along their trajectory, we will permute at the float rather than profile level, which is the conservative choice. We will calibrate the wording of the Abstract and Sect. 4.3 to whatever that test supports.
Adopting Bock et al. (2022) also obliges us to engage with an alternative interpretation it offers. They attribute the contrast between vertically homogeneous bbp and unimodal chlorophyll profiles in oligotrophic subtropical waters to the dominance of the backscattering signal by non-algal particles within the euphotic zone. That structure corresponds to our Chl deeper class, for which we invoke photoacclimation. Kitchen and Zaneveld (1990) reached the same interpretation from the other direction, concluding that the vertical non-correlation of chlorophyll and particle concentration follows from the light dependence of cellular pigment content rather than from a difference between layers in the origin of the particles. We will state both interpretations in Sect. 4.2, note that the older literature favours the second, and note that our diagnostic does not distinguish between them, since a Chl maximum displaced below a shallower BBP700 maximum is consistent either with pigment adjustment at depth or with a backscattering signal set largely by non-algal material.
Summary of scope
In summary, we do not extend the analysis globally because such an extension would substantially overlap with existing global assessments and would shift the manuscript away from its intended contribution: a basin-scale assessment of the relative vertical organization of chlorophyll and particle backscattering in the North Pacific, where paired profiles provided sufficient spatial coverage for basin-scale analysis. Separately, we accept the request for ecological organization, and address it by relating the displacement classes to biomes and seasons within the study basin. We regard a global in situ extension, including comparison with the existing global biome partition, as the natural next study rather than as a revision of this one.
3. Vertical shape and example profiles
The reviewer asks twice to be shown the shapes, and separately notes that our framing assumes a one-dimensional balance while advection and vertical shear also generate vertical structure. We take these together, because the same omission underlies both: the manuscript shows composites and class fractions but never an individual profile.
The revised version will add a figure of individual profiles, showing chlorophyll and BBP700 on the same axes for profiles drawn from each displacement class, with the nitracline and the mixed layer depth marked. The selection will be stated rather than chosen for appearance: for each class we will show the profile closest to the class median in both maximum depths, together with the profiles at the tenth and ninetieth percentiles of the displacement within that class, so that the range within a class is visible and not only its center. The selection rule and the class sizes will be given in the caption. We note what this figure can and cannot show. It will show whether the two maxima are separable by eye in individual profiles, and whether the within-class range of the two depths overlaps between classes; it cannot establish that the classes correspond to distinct processes, and we do not intend it to.
The reviewer also asks why we did not fit a vertical shape, and points to the literature that does. A parametric fit — a Gaussian on a background, or a sigmoid — imposes a single maximum, and the profiles we most need to characterize are those for which that assumption is least safe. Our diagnostic has the mirror-image weakness: it returns one maximum per profile and is silent about profiles carrying two of comparable size, which is what two-dimensional structure would produce. We will report, for each class, how many profiles carry a secondary maximum within a stated fraction of the primary one, and we will include such profiles in the new figure. A full shape-based classification would constitute a separate extension beyond the scope of the present revision.
Part of what the reviewer asks for can be reported now. The manuscript reports the median depth of each maximum but not its spread. In the corrected series the interquartile range of the chlorophyll maximum depth is 27.5 to 87.5 dbar in the Coupled class, 77.5 to 122.5 dbar in the Chl deeper class and 7.5 to 37.5 dbar in the Chl shallower class; for the BBP700 maximum the corresponding ranges are 22.5 to 77.5, 22.5 to 67.5 and 47.5 to 112.5 dbar. The interquartile ranges of adjacent classes overlap in both depths, and in the BBP700 maximum depth the Coupled and Chl deeper classes are barely distinguishable, so the Chl deeper class is characterized by a deep chlorophyll maximum rather than by a shallow backscattering maximum. This reinforces the need to show individual profiles, because the classes are defined by the relative displacement of the two maxima rather than by either depth alone.
On the framing, we have written as though the vertical arrangement of the two maxima reflects local processes. It need not. A displacement between the chlorophyll and backscattering maxima can also arise from differential advection of water masses with different optical properties, or from vertical shear acting on an initially aligned structure, and our diagnostic cannot distinguish these from local photoacclimation or from local particle production. We will state this in Sect. 4.6, and remove the implication that the classes correspond to distinct local mechanisms. The classification identifies recurrent vertical configurations and shows that they are organized in space and with respect to the nitracline; it does not identify the processes that generate them.
4. Spikes, prominence, and the depth of the BBP700 maximum
We agree with the reviewer that isolated backscattering spikes must be addressed explicitly. We therefore quantified their contribution to the Chl shallower class, and tested the sensitivity of the classification to alternative smoothing treatments.
In preparing this response we extended the verification upstream of the archived source tables. In the series used for Figs. 3 and 5, the smoothing step assigned values to pressure bins containing no observation, which in sparsely sampled profiles could place the reported BBP700 maximum at a depth where no measurement existed. All figures will be recomputed on the single diagnostic described in Sect. 2.2, in which such bins remain undefined. This affects 5.6 % of profiles and 6.4 % of class assignments; the Chl shallower class comprises 566 profiles, against the 730 printed in Fig. 3. In the corrected series the median BBP700 maximum depth is 42.5, 47.5 and 67.5 dbar for the Coupled, Chl deeper and Chl shallower classes, respectively. We note that the preprint gives 47.5 dbar as the Coupled value in the Abstract and as the Chl deeper value in the Results, and we therefore give the corrected medians for all three classes rather than a difference against a single published number.
This also supersedes our author comment of 22 July 2026. The correct classification population is 10,801; the previously reported value of 10,740 was the sum of class sizes generated by the earlier processing series. The corrected panel sample sizes are 7,279, 2,956 and 566, which sum to 10,801. The affected figures and source tables will be replaced, whereas the three-class structure and the ordering of the classes by frequency are unchanged.
The reviewer's concern bears on the choice of vertical smoothing. Omitting the smoothing gives 1,363 Chl shallower profiles; the adopted five-bin rolling median gives 566; four stronger treatments give 611, 515, 501 and 404. The count is therefore not monotone in smoothing strength, and the adopted setting lies in the interior of the range rather than at either end. Although class size varies, the three-class structure and the frequency ordering hold under every variant. Across the four stronger treatments the profile-level classification agrees with the adopted setting in 86.4 to 91.6 %, compared with 76.0 % when smoothing is omitted. Robustness to the width of the smoothing is therefore not the same as robustness to its presence, and we do not treat the two as one.
We did not adopt a wider window because stronger smoothing removes vertical displacement in general rather than spike-like maxima in particular. The wider filter in question — a seven-point median followed by a five-point mean — is the bin-space analogue of the one Bock et al. (2022) apply to bbp profiles on the float's native grid; our adopted setting is the five-bin rolling median. Stepping from the adopted setting to the wider filter, 141 profiles leave the Chl shallower class and 90 enter it, a net loss of 51; of the 141 that leave, 126 are not flagged by our spike test, defined below, and their BBP700 maxima lie at a median depth of 67.5 dbar, which is the median of the class as a whole. Over the same step the Chl deeper class loses 186 profiles on net and the Coupled class gains 237, so both directions of displacement are absorbed together. The Supplement gives the step in full. Accordingly we address spikes by an explicit spike criterion rather than by widening the smoothing window.
To address prominence, we evaluated it in two independent ways for the 154 profiles (27 % of 566) in which the BBP700 maximum lies at or below 100 dbar: the ratio of the binned maximum to the profile median, computed from the same product as the depth diagnostic, and the ratio of the raw peak to a despiked local baseline, computed directly from the Sprof files within 7.5 dbar of the diagnosed depth. The second does not depend on the depth diagnostic and is defined in Sect. 2.2 of the revised Methods. The two are normalized to different references and rank the profiles similarly while differing in level; both indicate that most of these maxima are weak protrusions rather than sharp spikes, and the Supplement reports the comparison in full. Throughout this comment, spike-flagging is on the second measure at a prominence threshold of 2, and the spike fractions given below are computed on it.
More informative than any single fraction is where these profiles lie in the water column. Spike-like maxima are concentrated at the base of the profiling range: the fraction classified as spikes is 10 % of the 80 profiles between 100 and 150 dbar, 0 % of the 23 between 150 and 200 dbar, 32 % of the 19 between 200 and 250 dbar and 47 % of the 32 between 250 and 300 dbar. Twenty-one of the 154 maxima fall at or below 290 dbar, that is, at the lower limit of the search range. In the interval where displacement below the Chl maximum bears on our argument, 100–200 dbar, spike-derived maxima account for 8 of the 103 profiles in that interval, which is 1.4 % of the 566 profiles in the Chl shallower class.
The reviewer names two specific causes. Spikes from a zooplankton layer would be expected at depth, and the depth distribution above is consistent with that: the spike fraction rises from 10 % in the shallowest band to 47 % in the deepest, and 21 of the 154 maxima sit at the lower limit of the search range. Near-surface bubbles act in the uppermost few metres, whereas the BBP700 maxima that bear on our argument are deep — in the Chl shallower class the interquartile range of the BBP700 maximum depth is 47.5 to 112.5 dbar — so bubble contamination is not a candidate explanation for them; we will nevertheless report the number of profiles whose BBP700 maximum falls in the uppermost bins. Neither cause is excluded by these numbers, but the one that bears on the deep BBP700 maxima the reviewer asks about is the first, and in the interval where our argument rests it accounts for 1.4 % of the Chl shallower class.
The classification does not rest on the affected profiles under any of these definitions. Removing all spike-flagged profiles leaves 537 of 566; removing every profile whose BBP700 maximum lies below 250 dbar leaves 534; removing both leaves 520. The spike fraction itself depends on where the prominence threshold is placed: at thresholds of 1.2, 1.5, 2, 2.5, 3, 4 and 5 it is 12.4 %, 7.2 %, 5.1 %, 4.4 %, 3.2 %, 2.8 % and 1.8 % of the class, so that even at the most permissive threshold we examined the class retains 496 of 566 profiles. We report the range rather than a single value.
Following the reviewer's request, we will no longer describe these features as "distinct deep BBP700 maxima". We will describe them as "BBP700 maxima located below the Chl maximum, often of modest prominence", and replace "demonstrating" with "indicating" where the Chl shallower class is discussed.
In the revised version, Sect. 2.2 will state the spike criterion explicitly, the Supplement will carry the full smoothing sensitivity together with the depth stratification above, and the sensitivity of the spike fraction to the threshold will be reported alongside it. We note two things we do not claim. First, we do not use the number of observations within the peak window as evidence of spike character: most floats in this dataset sample at 5 dbar above 100 dbar and 10 dbar below, so a narrow window necessarily contains one or two points irrespective of the underlying structure. Second, our spike test is a quality-control step, and we do not interpret the flagged features beyond noting that their form is consistent with isolated particles; their origin and composition are outside the scope of this paper.
5. Novelty, the nitracline ordering, and specific annotations
Novelty and framing. Several annotations state that the seasonal and regional patterns of near-surface and subsurface chlorophyll maxima are already established in the subpolar, subtropical and equatorial regimes, and question the spatial organization of our classes. We agree that these individual patterns are known and do not claim them as new. Sect. 4.1 says this, and we will state it earlier in the manuscript and in the Abstract. The contribution is not the discovery of Chl-particle separation but the construction and basin-scale application of a paired-profile diagnostic that quantifies the relative vertical organization of the two maxima across a large in situ population, and that allows us to test whether recurrent vertical configurations are associated with distinct ecological settings, stratification regimes and nitracline positions. A framing has to earn its place by what it separates, and we accept that this is the standard the paper should be judged against. It is met first by what we already report below, and by a measurement that neither of the two studies discussed in the next paragraph could make, since neither carries nitrate on the same profile: the fraction of profiles in which the nitracline lies strictly between the two optical maxima differs by nearly an order of magnitude between the classes — 18.5 % of Chl shallower profiles against 1.4 % of Coupled and 2.3 % of Chl deeper — and that ordering is preserved at every displacement threshold we examined. The two tests described above extend the standard to ecological and spatial structure: the stratification of the classes by North Pacific biome and season, and the float-block permutation test of spatial aggregation. We will calibrate the framing and claims to the results of these analyses, and will report either outcome, for the Chl deeper class and for the Chl shallower class alike.
That literature is closer to this manuscript than our reference list acknowledges. Kitchen and Zaneveld (1990), working in the North Pacific, showed that the vertical structures of chlorophyll and of particulate scattering are not correlated and have no theoretical reason to be, and attributed the offset to the rise in pigment content per unit biomass as irradiance falls with depth, following Kiefer and Kremer (1981); Pak et al. (1988) reached the same conclusion from two meridional transects of the same basin, reporting that the vertical distributions of chlorophyll and of particle concentration are not parallel, and attributing this to the light-driven increase in pigment per particle with depth. Their particle measure is beam attenuation rather than backscattering. We will cite both, and we do not claim either the configuration in which the chlorophyll maximum lies below the particle maximum or its explanation by photoacclimation as new. Two things distinguish what we report. The first is the observation itself. Kitchen and Zaneveld obtained their backscattering profiles from Mie calculations rather than from measurement, noting that few backscattering measurements had been made together with chlorophyll, and the particle maximum in that literature is defined by beam attenuation, which weights the particle size distribution differently from backscattering at 700 nm. We report measured BBP700 paired with chlorophyll from the same cycle, for 10,801 profiles. The second is a configuration those studies do not describe. Kitchen and Zaneveld report the chlorophyll maximum below the particle maximum and do not report the reverse. Pak et al. describe the particle profile as two-layered, with a distinct subsurface particle maximum only at some stations: above the chlorophyll maximum in the central gyre, and at or near it in the equatorial and subarctic zones. Neither study describes a particle maximum lying below the chlorophyll maximum, which is the configuration defining the Chl shallower class, which comprises 566 profiles here, and which is present under every smoothing treatment we examined (Sect. 4). Pak et al. also caution that the coincidence or non-coincidence of the two maxima does not by itself indicate a different mechanism of formation. We agree, and say in Sect. 3 that our diagnostic cannot establish that the classes correspond to distinct processes; what it establishes is that these configurations recur and are organized in space and with respect to the nitracline.
The nitracline ordering. One annotation summarizes Sect. 3.5 as showing that the nitracline was always deeper. On the corrected series that summary holds for the large majority of every class: the nitracline lies deeper than both optical maxima in 94.6 % of Chl deeper profiles, 91.1 % of Coupled profiles and 79.1 % of Chl shallower profiles. What differs between the classes, and by much more, is how often it lies between them, and that is why we report the ordering by class rather than in aggregate. Two changes affect the numbers given in Sect. 3.5, and we separate them. The population is redistributed by the corrected classification: the nitracline subset comprises 6,335 Coupled, 2,608 Chl deeper and 459 Chl shallower profiles, the same 9,402 profiles as before, in place of the 6,150, 2,651 and 601 published. The definition of "between" is also tightened: we now count only profiles in which the nitracline lies strictly between the shallower and the deeper of the two maxima, because at the 5 dbar resolution of the diagnostic a nitracline that coincides with one of the maxima is at a maximum and not between them. Under the published definition, which admits those coincidences, the fractions are 6.5 % of Coupled, 4.2 % of Chl deeper and 20.5 % of Chl shallower profiles; under the strict definition they are 1.4 %, 2.3 % and 18.5 %. The tightening is not neutral between the classes, since coincidences account for 78 % of the Coupled profiles counted as between under the published definition, 46 % of the Chl deeper and 10 % of the Chl shallower, and it therefore widens rather than narrows the contrast we report. We give both definitions and will use the strict one in the revision. The configuration in which the chlorophyll maximum lies below the backscattering maximum is the one that the light-driven pigment argument of Kitchen and Zaneveld (1990) accounts for, and we return to this in Sect. 4.2.
Terminology. Two of the reviewer's annotations concern what we are entitled to call a deep chlorophyll maximum. Our operational definition (Sect. 2.2) takes the maximum of the smoothed Chl profile within 0–300 dbar and does not require separation from the surface layer, and we state this. The consequence is that part of the Chl shallower class has a near-surface rather than a subsurface chlorophyll maximum: in the corrected series its median Chl maximum depth is 7.5 dbar, and among the 154 profiles of that class whose BBP700 maximum lies at or below 100 dbar, 30 have a Chl maximum shallower than 20 dbar. Calling these deep chlorophyll maxima is not warranted by our own definition. The two figures just given refer to different parts of the class. The median is that of the class as a whole; the 30 profiles are 19.5 % of the deep subset on which the displacement argument rests. The change of terminology is required by our own definition and does not bear on the configuration the class describes. In the revised version we retain the classification but change the wording: within the Chl shallower class we write "shallow chlorophyll maximum" rather than "Chl DCM", in the Abstract, Sect. 3.3, Sect. 4.2 and the caption of Fig. 3.
The fluorescence-to-chlorophyll conversion. Two annotations note that the conversion from fluorescence to chlorophyll concentration is not unique but varies with biome, season and community composition (Roesler et al., 2017). We accept this and will state it. It bears less on our diagnostic than on a concentration-based one, because we use the depth of the maximum rather than its magnitude, and the depth of the maximum is unchanged by any factor that scales a profile as a whole. What would affect it is a bias that varies with depth within a profile. Non-photochemical quenching is the known case of this and is addressed separately in the revised Methods; a depth-dependent community or packaging effect cannot be excluded from these observations, and we will say so rather than assume that the profile shape is unaffected.
Figure 5 colors. The reviewer notes that the BBP700 maximum and the nitracline are not distinguishable under color-vision deficiency. Fig. 5 will be redrawn with a color-vision-safe palette and with the two lines additionally differentiated by line style, so that they remain separable in greyscale as well.
Attrition of the profile population. Several annotations ask how many profiles were excluded and at which step, and object to the word "successfully" in "successfully diagnosed". Both are fair. The manuscript reports the retained population but not the path to it. The revised version will add a figure showing the attrition from the initial geographic window to the 10,801 paired profiles used in the main analysis, and separately to the 9,402 profiles with a nitracline diagnosis, with the count removed at each criterion. We already report that the classification population is 10,801 after the Pacific mask and the requirement that both maxima be diagnosable; the cell counts retained by the gridded maps at the three-profile threshold are recomputed on the corrected series and given in the attrition figure. Where "successfully diagnosed" now stands we will state the criterion that was met and the number of profiles that failed it, and we will remove the word "finite" where it stands without a stated range.
The 25 dbar criterion. The reviewer objects that 25 dbar is an absolute threshold. We agree: a fixed threshold is a convention, and it does not scale with the depth of the feature. We keep it in the revised version, for two reasons that we make explicit rather than assume.
First, the threshold is not what generates the structure. The distribution of the displacement itself is unimodal at zero and does not increase away from zero out to at least 25 dbar, so the classes are cuts through a continuum rather than separated modes; we will show this distribution in the Supplement. The reviewer asks what other thresholds we examined. The manuscript reports none, and in preparing this response we repeated the classification at 10, 15, 20, 30, 35 and 40 dbar, which is every threshold other than the adopted 25 dbar that the 5 dbar grid of the diagnostic resolves over that range. Three results hold across all of them. The frequency ordering of the three classes is unchanged, from 5,000 Coupled, 4,694 Chl deeper and 1,107 Chl shallower profiles at a 10 dbar threshold to 8,720, 1,755 and 326 at 40 dbar. The median chlorophyll maximum depth of the Chl shallower class is 7.5 dbar at every threshold, so the change of terminology described above does not follow from where the cut is placed. And within the nitracline subset the class ordering of the fraction lying between the two maxima is preserved throughout, the Chl shallower fraction exceeding the Chl deeper fraction by a factor of 4.8 at a 10 dbar threshold and 9.7 at 40 dbar. What does change is the magnitude of the separation the classes describe: the median BBP700 maximum depth of the Chl shallower class moves from 52.5 dbar at a 10 dbar threshold to 82.5 dbar at 40 dbar, and its nitracline-between fraction from 11.3 % to 25.9 %. We retain the threshold and state what the sweep supports: the existence of the classes and their ordering are not produced by the threshold, while the depth statistics and the fractions reported within each class are conditional on it. The Supplement will report this sensitivity in full. A relative threshold would move the boundaries but would not create or remove the continuum; we note also that on a 5 dbar grid a threshold is resolved only in 5 dbar steps, so that in practice our criterion assigns to the displaced classes those profiles whose two maxima differ by 30 dbar or more.
Second, the two depths are not redundant. Their standard deviations are 41 and 37 dbar, and they correlate at r = 0.64 (r² = 0.41), so that about 60 % of the variance of each is not accounted for by the other. The displacement therefore contains substantial information not represented by either depth alone. We will also test the reviewer's specific expectation that the displacement covaries with mixed-layer chlorophyll, and report the resulting relationship.
We will nevertheless report the classification against depth-relative references in the Supplement, using the mixed layer depth and the nitracline depth, both of which we already diagnose. This addresses the reviewer's point without introducing the circularity that an optical-depth normalization would carry here, since our optical depth would itself be derived from the chlorophyll profile.
Further annotations. Several remaining comments are accepted without qualification and are answered in the Methods and the Supplement rather than here. The 5 dbar binning and the median smoothing reduce the vertical resolution of the diagnostic and remove isolated high values; Sect. 2.2 will state this and give the reason, and we note that the diagnostic uses the depth of the maximum and not its magnitude. Chlorophyll denotes pigment and not biomass, and the caution the reviewer applies to backscattering as a biomass proxy applies to it as well; we will state this where the classes are interpreted. Whether individual grid cells hold enough profiles to span a seasonal cycle is a fair question of the class-fraction maps, and the attrition figure and the stratification by biome and season described above are where we answer it. Finally, a single cell of Fig. 2 aggregates profiles from different seasons and water masses, so that the figure describes the joint distribution of the two maximum depths and not a process; we will say so in its caption and rely on the stratified version for interpretation.
6. Summary of planned revisions
Applied in this comment, and to be carried into the manuscript:
- Within the Chl shallower class, "shallow chlorophyll maximum" replaces "Chl DCM" in the Abstract, Sect. 3.3, Sect. 4.2 and the caption of Fig. 3.
- "Demonstrating" becomes "indicating" where that class is discussed.
- Kitchen and Zaneveld (1990), Pak et al. (1988), Roesler et al. (2017) and Bock et al. (2022) added, with the alternative interpretation offered by Bock et al. and the light-driven interpretation of Kitchen and Zaneveld stated in Sect. 4.2 alongside our own.
- Fig. 5 redrawn with a color-vision-safe palette and differentiated line styles.
- "Mature" removed from the Abstract, Sect. 4.1 and the conclusions; the state of the observing network is given by the paired-profile counts of Fig. 1 rather than by an adjective.
- All figures recomputed on a single processing series, as described in Sect. 4.
- The Abstract sentence stating that Chl shallower profiles tended to show a deeper BBP700 maximum near the nitracline depth range is replaced by the measured ordering given in Sect. 5: the nitracline lies deeper than both maxima in 79.1 % of Chl shallower profiles and strictly between them in 18.5 %, against 1.4 % of Coupled and 2.3 % of Chl deeper profiles.
- The class median BBP700 maximum depths given in the Abstract are taken from the corrected series: 42.5 dbar for the Coupled class, 47.5 dbar for the Chl deeper class and 67.5 dbar for the Chl shallower class.To be added to the Methods:
- An explicit spike criterion, with its definition (Sect. 2.2).
- The quality-control flags used, the treatment of empty bins after smoothing, the tie rule, and the display grid of the composite figures.
- The treatment of non-photochemical quenching, and the nitracline definition stated as ours rather than as the only one in use.To be added to the Results and Supplement:
- Stratification context for each class: mixed layer depth, depth of maximum stratification, the position of both maxima relative to them, and the prominence of both maxima, reported for the 2022–2025 population of 10,801 profiles and, separately, for the extended record of Sect. 2.
- Optical context on the subset with measured photosynthetically available radiation, with the circularity of the modeled alternative stated.
- Stratification of the classes by North Pacific biome and season.
- A test of spatial aggregation using float-block permutation, and the wording of the Abstract and Sect. 4.3 calibrated to it.
- The raw distribution of the displacement; the sensitivity of the classification to the smoothing strength in both directions and to the tie rule; the depth stratification of the spike-flagged profiles; and the dependence of the spike fraction on the prominence threshold.
- The standard deviations of the two maximum depths and their mutual correlation.
- The sensitivity of the classification to the displacement threshold, at 5 dbar intervals from 10 to 40 dbar.
- The displacement classes for every year with usable paired profiles in the study domain, reported separately from the 2022–2025 period (Sect. 2).
- The classification against depth-relative references, using the mixed layer depth and the nitracline depth (Sect. 5).
- A test of whether the displacement covaries with mixed-layer chlorophyll, reported whether or not it supports the fixed threshold (Sect. 5).
- The number of profiles in each class carrying a secondary maximum within a stated fraction of the primary (Sect. 3), and the number whose BBP700 maximum falls in the uppermost bins (Sect. 4).
- A figure of the attrition of the profile population, and a figure of individual profiles with a stated selection rule.To be removed or weakened:
- The implication that the three classes correspond to distinct local mechanisms; advection and vertical shear stated as alternatives in Sect. 4.6.
- "Successfully diagnosed" replaced by the criterion met and the number that failed it; "finite" removed where it stands without a range.An updated version of the archived data and code package will accompany the revision, since the source tables for Figs. 3 and 5 change. The concept DOI is unaffected.
Citation: https://doi.org/10.5194/egusphere-2026-3910-AC2 -
RC2: 'Reply on AC2', Emmanuel Boss, 14 Aug 2026
reply
Dear authors,
I an looking forward to your revised manuscript. Given your reply, I think it is likely to be novel and interesting.
All the best, Emmanuel
Citation: https://doi.org/10.5194/egusphere-2026-3910-RC2
-
RC2: 'Reply on AC2', Emmanuel Boss, 14 Aug 2026
reply
Data sets
Processed BGC-Argo data for chlorophyll–particle backscattering displacement analysis in the North Pacific Taketo Hashioka and Maki N. Aita https://doi.org/10.5281/zenodo.20873450
Model code and software
Python analysis code for BGC-Argo chlorophyll–particle backscattering displacement analysis Taketo Hashioka and Maki N. Aita https://doi.org/10.5281/zenodo.20873450
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 129 | 35 | 17 | 181 | 16 | 15 |
- HTML: 129
- PDF: 35
- XML: 17
- Total: 181
- BibTeX: 16
- EndNote: 15
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Dear Editor and Referees,
While re-checking our archived figure source tables after posting of the preprint, we identified an inconsistency in the manuscript text, which we wish to report proactively.
The text states that the analysis used 10,801 usable paired Chl-BBP700 profiles (Abstract; Sect. 2.1; Sect. 3.2; Sect. 3.4). The correct number is 10,740. This is the population underlying all figures and all other reported statistics: the three displacement classes comprise 6,973 Coupled, 3,037 Chl deeper and 730 Chl shallower profiles (Fig. 3), which sum to 10,740.
The value 10,801 predates the final Pacific basin-mask correction, in which profiles lying outside the connected Pacific domain (Gulf of Mexico and Caribbean side) but inside the initial rectangular longitude-latitude window were removed. All figures, source tables and derived statistics were recomputed after that correction, but the total profile count was not updated in the four places listed above.
We have re-verified every other quantitative statement in the manuscript against the archived source tables in the Zenodo package (DOI: 10.5281/zenodo.20873450), and all of them are consistent with the corrected population: median Chl DCM / BBP700 maximum / displacement values of 52.5 / 47.5 / +5.0 dbar (Coupled), 107.5 / 47.5 / +50.0 dbar (Chl deeper) and 17.5 / 97.5 / -55.0 dbar (Chl shallower); class sample sizes of 6,973 / 3,037 / 730; 663 grid cells containing 10,630 profiles in the gridded class-fraction summary; and 9,402 profiles with a successful nitracline diagnosis, comprising 6,150 Coupled, 2,651 Chl deeper and 601 Chl shallower profiles.
No figure, no source table and no conclusion is affected. The correction concerns the reported total profile count only, and we will implement it throughout the revised manuscript.
We apologise for the oversight and thank the Editor and Referees for their time.
Taketo Hashioka and Maki N. Aita