The authors present a dataset from the shallow coastal waters in the eastern English Channel and southern bight of the North Sea collected from 1992-2001, primarily for the purpose of water quality monitoring, The data are reported to include 281 taxa and 3687 samples spread over 10 stations. 

Readers would greatly benefit from a careful and thorough editing of the text, in particular the abstract and introduction. At present ideas are sometimes out of order or disjointed and the purpose of the manuscript is not clear.

The methods and results are quite clear. The overviews of the physical-chemical variables are very helpful. I was surprised to see no indication of dinoflagellates in Fig. 2.  Is it possible that there were no phytoplankton dinoflagellates (autotrophs or mixotrophs)? This seems unlikely. I wonder if a seasonal version of the count distribution would be helpful.  Are any size data or approximate biomasses of the enumerated taxa available? It might also be interesting to know something of the distribution of species richness, even just the average richness per sample.

I thank the authors for their efforts to disseminate and describe their wonderful dataset. Unfortunately, from my perspective the data are inadequately described. Since this is a data paper in a data journal, this is a problem.

I have examined the data at https://doi.org/10.17882/50832

The data appear to be encoded in latin1 as a csv file with a semi-colon separator. It only takes a few guesses to work this out, but ideally the reader would not need to guess.

I did not see a data dictionary or other description of the contents of the file anywhere. The variable headings are written in French. I understand the desire to work in one’s preferred language, but the language of publication is English, so a translation should be provided.

I was able to read a data table of 61 variables and 99,006 observations. The number of observations and variables should be reported in the metadata so that the user can be sure the data were received as expected.

There are 6 variables which are missing for all observations. I don’t understand the need to include undesribed variables with no data.

There were some (809) zero counts for taxonomic abundance, but very few (<1%). Please clarify the reason for including these zeros. Was a consistent taxonomic list used for all stations and times? Can the reader infer that the taxa recorded at some stations but not others have zero abundance when not reported?

It would be helpful to provide latitude and longitude of the stations; these can be read approximately from Fig. 1, but they do not appear to be in the data file. I was unable to decode the station location data: cordonnees passage min and max for x and y.

The values I computed for Table 2 did not match the authors’, likely because I misinterpreted something; incomplete description of the data makes this easy to do unfortunately. I suggest additional details to clarify the differences.

Is there a problem with station SRN Somme mer 1 (Mer1?) resulting in it not being reported in Table 2?

Number of observations (samples) per station:

z1 |> count(lieu_de_surveillance_libelle, passage_identifiant_interne, passage_date) |> count(lieu_de_surveillance_libelle)

# A tibble: 11 × 2

   lieu_de_surveillance_libelle     n

 1 At so                          479

 2 Bif                            414

 3 Mimer                          272

 4 Point 1 Boulogne               508

 5 Point 1 Dunkerque              406

 6 Point 2 SRN Boulogne           401

 7 Point 3 SRN Boulogne           395

 8 Point 3 SRN Dunkerque          324

 9 Point 4 SRN Dunkerque          309

10 SRN Somme mer 1                301

11 SRN Somme mer 2                391

The total number of samples in the dataset is 4200. This does not match Table 2 (3687) even if all the observations from SRN Somme mer 1 are removed. My calculations showed that 2007 had the most observations (179), but this does not agree with the number in the abstract (184, line 12). It would be more representative to report the mean or median number of observations (142, 140) or the range (100 to 179).

Number of species:

z1 |> count(resultat_nom_du_taxon_referent, lieu_de_surveillance_libelle) |> count(lieu_de_surveillance_libelle)

# A tibble: 11 × 2

   lieu_de_surveillance_libelle     n

 1 At so                          224

 2 Bif                            206

 3 Mimer                          197

 4 Point 1 Boulogne               215

 5 Point 1 Dunkerque              221

 6 Point 2 SRN Boulogne           208

 7 Point 3 SRN Boulogne           188

 8 Point 3 SRN Dunkerque          202

 9 Point 4 SRN Dunkerque          191

10 SRN Somme mer 1                167

11 SRN Somme mer 2                197

Has the taxonomy be standardized in any way? Was a database such as marinespecies.org used? I suspect the count of species in table 2 in fact refers to some level of taxonomic resolution and not species. In addition to species, the data table reports many genera, some higher classifications, and size or shape features. Some taxa are fusions of several species or genera, e.g., “Chaetoceros densus + eibenii + borealis + castracanei”. Some classifications can be guessed, but are incomplete, e.g., “Centriques”, “Pennées”.

I did not see any spelling errors in the taxonomic identifications; I congratulate the data curators for this success!

Phaeocystis globosa was repeatedly identified in the manuscript, but does not appear in the dataset. Only the genus-level id Phaeocystis is reported in the data. This is a serious oversight or inconsistency. It would be helpful to note in the data if the counts refer to cells, colonies, or a mixture.

I was able to read the physical-chemical data and interpret it. The general concerns above copy over here about documentation, encoding, location information. I did not see any information about the units of measurement of the various quantities (temperature, chl-a, nitrate, nitrite, phosphate, silicate; some can be guessed, but the nutrients could easily be in mass or mol units and there is no way to tell.) Guessing is not ideal in a documented dataset. The metadata indicated phaeopigments, suspended matter organic and mineral are reported, but I did not see any observations of these quantities in the data. These are oversights that should be corrected.

I did not examine the data at the following sites as they did not seem to be the primary target of this data paper:

https://doi.org/10.17882/85178 , https://doi.org/10.17882/47248 , https://doi.org/10.17882/47251. 

A bit more information at line 347 about the relationship between these data would be helpful. (Are they completely distinct, partially overlapping, etc.?)

I was unable to use the R package TTAinterfaceTrendAnalysis. It requires X windows and Tk application software which many users will not have installed. It might be helpful to indicate something of the software requirements in a brief note. These requirements are a bit unusual for modern software packages and will likely limit the usage of their package.

Detailed comments

Abstract

The abstract does not clearly describe the dataset, which is the main purpose of this paper. I suggest informing the reader of the years covered by the data and the total number of observations.

The abstract is written about an “historical” dataset, yet is largely written in the present tense (SRN collects, objectives … are, regular acquisition of data…) This is a bit confusing, indeed the paper both describes an ongoing program and presents data from 30 years of collection. A bit of smoothing of the exposition and clarification of the goals of the manuscript would help the reader.

Line 8. Define acronym SRN before it is used (It is defined repeatedly below)

Line 8. Give location of Iframer (Brest)

Line 8. What makes the data historical? Has data collection ceased? Are modern data excluded?

Line 12. 184 samples in one year would be quite intense sampling. Is this phrasing correct? Samples per station and the number of stations might be more informative.

Line 14. Are continental inputs, development and management policy metadata part of the time series? Are these data described in this manuscript? Available publicly? 

Introduction

Line 30: “others cause excessive organic matter inputs”. I think I know what you mean here, but this is a relatively unusual observation, so an example taxon or citation could help make your point more clearly.

Line 59. What does “address” mean here? It’s a fairly indirect verb.

Line 74. Presumably French should be capitalized here.

Purpose

Line 97. Should “propose” be “present”?

This is an interesting paper, describing a set of measurements conducted by the SRN network in the Eastern English Channel, and the Southern Bight of the North Sea.

The dataset described in this paper consists of measurements of hydrological and ecological variables, going back to 1992. Such historical datasets contribute to our understanding of long term ecological processes, and improve our ability to monitor and preserve sensitive ecosystems. This is especially important in areas subject to strong anthropogenic pressure, as the one addressed here, making the described dataset interesting and valuable.

The paper is well organized and clear. The authors do well in describing the geographical, hydrological and ecological context, emphasizing the datset’s importance and relevance. The technical aspects of data collection, analysis methods and quality control are, in general, well-described, although some important corrections/clarifications should be made (see comments below). In addition, the authors provide useful description of observed trends, discuss possible interpretation of observed variability patterns, and provide simple codes for data handling.

Overall this is a well written article that describes an important dataset and merits publication in Earth System Science Data. However, I have two concerns that should be addressed before acceptance:

While horizontal aspects of the sampling strategy are well described, it is not clear to me what is the vertical configuration of the sampling strategy. The authors should describe whether the measurements were taken at a single or multiple depths, what the sampling depths are, and what is the rationale behind their selection. This is an important limitation that has to be addressed both when describing the data collection, and when discussing the observed trends

In the Discussion and Conclusion section the authors provide a review of scientific works conducted using SRN data. Although interesting by itself, to my understanding such a review is not in the scope of a data description paper, and should not be included here.

The authors propose a synthesis of data contained in databases over nearly 30 years. The interest is obvious, but the globalisation of the parameters leads to a loss of sight, and ends up pushing open doors. The authors need to go into much more detail with corelations (more original, finer and more precise... to be found) for this kind of article to be useful and also to correspond to the title (which I find particularly well selling but misleading).



In detail, not wanting to repeat what the two firstreferees have already pointed out:

- Line 30: not all toxicity phenomena for humans are through shellfish consumption,

- Line 34/35: "...major effects on the biodiversity of higher trophic levels". Need a solid reference to back this up,

- Line 60: "...abnormal increase...", "...naturally occurring...". I think these considerations are no longer in the way of thinking... and without getting into philosophical debates!

- Line 71/72: Pseudo-N needles stick into Phaeocystis colonies irritate filter feeders. Is this proven? Do they irritate more or less than in the isolated planktonic state?

- Line 89: Phaeopigment. They are not used afterwards.

- Line 193/196: These seem to me to be generalities that deserve to be detailed or referenced otherwise they do not belong here.

- Line 208: What is "Phytoplantonic taxonomic productivity"?

- Line 213: P. globosa is not a prymnesiophyceae but a coccolithophyceae in the current systematics. (idem in the legend of fig. 2)

- Lines 218-220: Potentially toxic but no toxin detected. OK but isn't that a bit short on explanation.

- Fig 2: Only bacillariophyceae are taken into account? Why are not all diatoms considered?

- Line 259: Why use the term dinoflagellates when other algal groupings use taxonomic ranks?

Line 266: The 3 diatoms mentioned are not bacillariophyceae. Idem for the following lines, there is a mishmash of terms.

Line 386: Carpentier, Martin & Vaz: This is grey literature.