Improved bathymetry leads to >4,000 new seamount predictions in the global ocean – but beware of phantom seamounts!

Minor edits made in response to constructive comments of 3 reviewers. 

Specific points raised by reviewers and our responses are detailed below (review comments in italics - line numbers refer to revision 2 of this manuscript):

Reviewer: Cherisse Du Preez
Link to original review: https://doi.org/10.14293/S2199-1006.1.SOR-GEO.ACTAJL.v1.RYVRIC

Flow: The main story of the title and the discussion is “new seamounts identified,” which is different than the “phantom seamounts” focus of the abstract, methods, and results.

- title amended to bring in the 'phantom seamounts' (previously removed due to reviewer request). Also the section of the discussion covering phantom seamounts has been expanded to give a greater focus to this issue.

Line 11: While “hotpots” made me chuckle, I think it’d be better to change the word to “hotspots.”

- Corrected!

Line 13: In large numbers, help the reader by using commas. “10,000 to more than 60,000”—consider changing throughout the article (used sometimes).

- Text revised to consistently use ","

Line 13-14: Provide more detail, “Seamount locations can be estimated by extracting conical shaped features [that meet other criteria (e.g., elevation)] from bathymetry grids.”

- edited

Line 41: Mention deep-sea mining here. This data was used at the first ISA REMP workshop for deep-sea mining on seamounts (a marine spatial planning meeting for the International Seabed Authority; report still in progress) to start to inventory seamounts in the North Pacific Area, to identify seamounts that are (i) contracted for exploration and (ii) could/should be considered for protection. Please consider mentioning mining or that fishing isn’t the only threat (e.g., climate change impacts too).

- line added

Line 49: If complete list, add: Kitchingman, A., and Lai, S. 2004. Inferences on potential seamount locations from mid-resolution bathymetric data. Seamounts: biodiversity and fisheries 12: 7-12.

- Added

Line 54: 1.5km diameter, right? Or height?

- height - detail added

Figure 1 / Line 82-87: Is the grey “sub-figure profile” line meant to show on the map where the inset profile data is from? The lines on the map are so close, and the colour difference so subtle that I can’t tell—poor quality. Extract profile and present separately to avoid cluttered and help with figure readability.

- Figure simplified. The subfigure has been removed as it is not mentioned in the text.

Figure 1 / Line 82-87: Not sure what is meant by “Chart symbol: No bottom detected at 183 m.” Delete since the line below clearly states, “No bottom detected on 2016 survey.”

- Subfigure removed

Lines 83-87 (Figure 1): Include/explain figure labels A & B when describing the location of each site in the caption: “...NW of the Great Chagos Bank (site A)…Area 40km north of this (site B)…”

- Reference added in the caption

Lines 89-91 (Figure 2): Text jumps between “Seamount A…B” and “site A…B”. Go with one.

- Agreed. site is better as there are no seamounts here!

Line 95: “were” not “where”

Line 119: I’m happy to see the authors mention that new bases can encompass multiple “old dataset” peaks, but it makes it sound like the new dataset doesn’t suffer the same issue of individually identifying multiple peaks on the same seamount—in reviewing the shapefiles I see this is still the case. Please see the comment for lines 144-155 below.

- amended

Figure 4 / Line 139: Remove table inset and present separately to avoid clutter. I am having a really hard time with readibility of the figures (poor quality).

- Figures have been remade at higher dpi. Table has been split out.

Lines 144-155: In my experience, these models are incredibly helpful in marine spatial planning--especially when assessed altogether--but I have witnessed the pitfall/danger in counting the predictions as the “number of seamounts” instead of the number of peaks (e.g., justification for allowing harmful activities on dozens of seamounts because models illustrate there are supposedly hundreds within the region--when in actual fact more than half of the predicted points are just peaks on the same seamount). It’d be beneficial for the authors to provide this word of warning regarding peaks vs. counts. I don’t think the high number of replicate predictions is unique to my study regions, but if the authors want to review overlapping bases and replicate predictions, I would suggest the NW and NE Pacific seamounts.

- Text added on this point. "We also note that these totals are really counts of seamount peaks, some of which may be linked together into seamount chains which could be regarded as a single feature. This potential double-counting may become more prevalent as these features are mapped in greater detail and smaller peaks on larger structures are identified. It was to address this issue that Yesson et al. (2011) introduced an optional filter to remove spatially adjacent features, and we recommend always examining the filtered and unfiltered predictions with this in mind."
 

Reviewer: Xiaoyun Wan
Link to original review: https://doi.org/10.14293/S2199-1006.1.SOR-GEO.AKXABB.v1.RVEAQH

Although this study found new seamounts, however, there are also phantom seamounts. Hence, I think the author should point out that some erroneous predictions may exist in the new predictions.

- A major emphasis of this study is to show that not all of these predictions can be relied upon. We have added a warning in the conclusion.

If the new results can be verified by sounding data, it would be better

- Agreed, but it is beyond the scope of this study.

It would be more helpful for the international research community if the authors could use mathematical equations to break down the method used in addition to the content of lines 115 to 119. It can be added as an appendix.

- These are previously published methods that are fully described in the references provided. For clarity we have revised the methods to give more detail: “The prediction algorithm of Yesson et al. (2011), which identifies seamounts as cone-shaped features rising more than 1000m above the surrounding seabed, was run on SRTM30+ V11, creating a new set of seamount predictions based solely on the new bathymetry”. We have also expanded the methods describing how the old and new datasets were compared.

“Seamounts were defined as present in the old dataset if the base of a seamount in the new dataset spatially overlapped with a seamount summit in the old seamount dataset (i.e. both datasets have a predicted seamount in approximately the same location). Seamount bases are the area covered by the ‘cone’ of the seamount, and are delimited by 8 radii 45° apart, radiating from the seamount summit point, that extend outwards from this point until the downward slope levels off, up to a maximum distance of 20km from the summit (thus the maximum base area is ~1,131 km²).  These seamount bases can, and often do, encompass multiple seamount peaks in both the old and new datasets, but a new seamount has to overlap with just one seamount in the old dataset to count as being a consistent prediction.”

Line 64, give the geographic extent (corner coordinates) of BIOT Seamount Survey.

- The survey was multi-discriplinary. The seamount portion of the survey moved around the Great Chagos bank, spanning approx 5-7S and 71-73E.

Line 117, insert ‘where’ after ‘terminate at the point’.

- rewritten

Line 118, “km2”->”km²”

- Corrected

It would be better to give the units of the data in the inset table of Figure 4.

- Details added to caption

Line 162, change ‘removed’ to ‘resolved’.

- Corrected

Reviewer: Junjun Yang
Link to original review: https://doi.org/10.14293/S2199-1006.1.SOR-GEO.AOD2B9.v1.RZJKAA

...However, the two stories are not well combined whereas usually only one theme is expected in a paper. This is the largest problem of this manuscript, as evidenced by the disordered structure.

- Previous iterations of this paper were focussed on the 'phantom seamounts' theme (rather than on the updated predictions), but this was repeatedly rejected as being insufficient so the scope was expanded at request of previous reviewers. We opt to maintain both of these themes in this short paper.

But, the motivation of this study is not clearly presented. What is the gap in our current knowledge? What are the objectives of this paper?

- The aim in the abstract has been revised to "This study presents an update of seamount predictions based on SRTM30 PLUS global bathymetry version 11 and examine a potential source of error in these predictions"

- Further edits have been made to keep the focus on these aims.

The findings reported in the “BIOT Seamount Survey” section is very interesting, but its connection with the last and next paragraphs is not well established, making it seems abrupt and standalone. How does these findings contribute to your seamount census? 

- We have added a linking section between the main background and the introduction of the BIOT survey.

This study describes issues with seamount predictions stemming from the use of historical sounding records, based on the findings of a seamount survey in the Indian Ocean. It presents an update of previous seamount predictions and examines whether this erroneous use of historical data persists.

- We present a method to detect these no-bottom-sounding-based seamounts and quantify how many persist in the latest predictions.

The method section is too simple, lacking of formulas and step-by-step description of how you produce the seamount census. This makes readers hard to duplicate your results. 

- There are no new methods presented in this paper. We provide citations to the method descriptions. We believe this meets the criteria of reproducibility. We do not feel it is appropriate to republish previously published methodology in this way. An overview of the method has been provided to give a better indication of what is being done.

You could consider adding more text to the discussion section, e.g., histogram of the seamount heights, evaluation of the reliability of results, known insufficiency in the present method that needs to be improved in the future, etc.; please refer to Wessel (2001) and Harris et al. (2014) for ways of extending the discussion. Besides, adding something specific about how your seamount census improves fishery will be helpful.

- Many of these examinations of seamount predictions have been accomplished by previous studies. We do not believe there is much value in repeating these. While there are new avenues that could be explored, we are aiming for a short update paper and feel that what we have written is sufficient to meet this objective. 

We have included a new histogram documenting seamount characteristics for all datasets. We have also expanded the discussion to cover some of these issues.

Line 1: 4000 is not compatible with the number in the abstract.

- The number in the abstract is 4,437, we rounded for a neater title. Title changed to ">4,000"

Line 51, Line 67: Do you mean SRTM30_PLUS? Note that SRTM30 and SRTM_PLUS are two different models. Only the land and ice topography of SRTM30_PLUS comes from SRTM30.

- Yes, we mean srtm30 plus. Text has been corrected to be consistent througout and we now consistently use srtm30 plus to refer to this grid.

Line 82: GEBCO 2020 has been released. Why do not you use the latest version?

- We did this work back in 2019, but it has taken a really long time to get this reviewed. We are not prepared to rerun & rewrite at this stage. 

Line 84: Add the text “Great Chagos Bank” onto Figure 1 to improve the readability.

- Corrected

Lines 117-118: This sentence is hard to understand. Please consider rewriting it.

- reworded

Line 135: In the legend, the marker for “New Seamounts” is hardly visible.

- Corrected

Line 139: Define the EEZ shown in the legend.

- Added.

Line 147: I did not find (Costello et al., 2010) in the reference list.

- Added

Line 152: Do you mean SRTM15+? Please use the name given by the author.

- Corrected

Line 162: Substitute “that be” by “it is”.

- Corrected.

Seamounts are important marine habitats that are hotspots of species diversity. Relatively shallow peaks, increased productivity and offshore locations make seamounts vulnerable to human impact and difficult to protect. Present estimates of seamount numbers vary from anywhere between 10,000 to more than 60,000. Seamount locations can be estimated by extracting large, cone-like features from bathymetry grids (based on criteria of size and shape). These predicted seamounts are a useful reference for marine researchers and can help direct exploratory surveys. However, these predictions are dependent on the quality of the surveys underpinning the bathymetry. Historically, quality has been patchy, but is improving as mapping efforts step up towards the target of complete seabed coverage by 2030.This study presents an update of seamount predictions based on SRTM30 PLUS global bathymetry version 11 and examine a potential source of error in these predictions. This update was prompted by a seamount survey in the British Indian Ocean Territory in 2016, where locations of two putative seamounts were visited. These ‘seamounts’ were targeted based on previous predictions, but these features were not detected during echosounder surveys. An examination of UK hydrographic office navigational (Admiralty) charts for the area showed that the summits of these putative features had soundings reporting “no bottom detected at this depth” where “this depth” was similar to the seabed reported from the bathymetry grids: we suspect that these features likely resulted from an initial misreading of the charts. We show that 15 phantom seamount features, derived from a misinterpretation of no-bottom sounding data, persist in current global bathymetry grids and updated seamount predictions. Overall, we predict 37,889 seamounts, an increase of 4,437 from the previous predictions derived from an older global bathymetry grid (SRTM30 PLUS v. 6). This increase is due to greater detail in newer bathymetry grids as acoustic mapping of the seabed expands.The new seamount predictions are available at https://doi.pangaea.de/10.1594/PANGAEA.921688.