This is a useful summary of a major data compilation of paleoecological data in Korea.  To my knowledge, this is the first such compilation of these data.  The data compilation is somewhat small in volume and spatial scope, but it represents an important step forward in building a culture of data sharing and FAIR data in Korea.  Papers like this are really helpful for showing the data now available and as a starting point for future community-building efforts, building open databases, and conducting large-scale scientific research.

1. Of the three datasets built (see Fig. 2), only one of them has a detailed description of variable names, in Table 1. The other two datasets need a similarly detailed table.  Ideally these two new tables would be put in the main text, but if there’s not enough space, the two new tables could go into supplementary information.  Specific comments about the data schema shown in Figure 2 and Table 1
   1. ‘Chron-Depth Collection’ is an awkward name… recommend naming to ‘Age-Depth Data’ or ‘Geochronology Data’.
   2. The Chron-Depth Collection table lacks a field for the depth of each age control, which is essential data/metadata; for age-depth models to be built or rebuilt, each age control must have a depth and time coordinate.
   3. The Chron-Depth Collection table also should include a field indicating whether a date is in radiocarbon years or calendar/calibrated years. Ideally should also indicate whether the age datum is 1950 CE (which is the standard for radiocarbon years) or something else.
   4. In Site Inventory, what does ‘Year’ store and is this information redundant with the ‘Year’ field in the Publication table? A standard principle of good database design is that each piece of information should be stored in only one location.
   5. In the Site Inventory, it is a little strange, or at least not internally consistent to store age as Oldest, Youngest endpoints, while depth is stored as an interval. Ideally Depth would also be stored as Top, Bottom endpoints. 
   6. Elevation (m): Elevation above sea level? 
   7. Recommend renaming ‘Specifications on environment’ to ‘Site Description’
   8. For Sample type, recommend renaming ‘Surface pollen’ to ‘Surface sample’ so that other kinds of surface data can be stored in the database.
   9. For Sample type, note that what this database calls ‘Trench’, Neotoma calls ‘Section’. ‘Section’ is a little more general, because ‘Trench’ implies a dug trench, while ‘Section’ can be any stratigraphic section or outcrop, e.g. including trenches, riverbanks, or rock outcrops.  See also comment on L179 below.
   10. Could probably rename ‘Written language’ as ‘Language’. Also, there seems to be a confusion here, because ‘Language’ in Fig. 1 is part of the Publication table, but Table 1 is listed as describing the variables in the Site table.
   11. Table 1 lists a category called ‘Proxy’ and fields including Pollen, Diatom, etc. However, these fields aren’t shown in Figure 2’s entity relationship diagram.  In general, Table 1 and Figure 2 should be checked to ensure that there is a 1:1 match between all variable names shown in each location.
2. The paper needs to be very careful in how it handles and reports ages, because it is using at least three different time scales, creating major potential for confusion. First, it uses the standard AD/BC (or CE/BCE) timescale for information such as publication year.  Second, it uses the radiocarbon timescale when reporting radiocarbon dates, which are expressed in years before present, uses 1950 CE as ‘present’ and has a non-linear relationship to the calendar-year timescale.  Third, the paper uses the calendar-year timescale for OSL and U/Th dates, which also are expressed in years before present and can use various years as ‘present’.
   1. For all variables that store year, the paper should clearly establish what time scale is being used.
   2. For all geochronological datasets, the dataset should formally store the datum used for ‘present’ or the paper should establish that in all cases, ‘present’ is treated as 1950 CE.
   3. For all radiocarbon dates, they should be reported in their original radiocarbon-year dates with 1sigma error. The use of the radiocarbon timescale should be explicitly noted as a metadata field.  The paper optionally could also include calendar-year conversions for each radiocarbon date as additional fields, along with information about the calibration curve and program used to convert from radiocarbon years to calendar years. 
   4. For all other geochronological dates, the use of the calendar-year timescale should be explicitly noted as a metadata field.
3. In the Summary, this paper does a good job of describing how this site inventory could be used for future work. That noted, I think the paper would be strengthened if its closing provided more of a forward-looking vision and potential future directions, in at least three ways.  Papers like this can set the tone for a community and inspire others.
   1. Communicate more of a scientific vision (this could be done in both Intro and Conclusions/Summary). What kinds of research questions can be enabled when large open paleoecological databases are available?  Cite more of the recent papers that have used large paleoecological datasets and discuss how these approaches could be used to advance questions in Korean paleoecology
   2. I know that the politics of data sharing and openness can be sensitive, but I think it would still be good to add a gentle call to authors to contribute their data and thereby further enhance the database by obtaining and adding the actual proxy datasets. This call follows naturally from the scientific vision… what scientific research questions are enabled by the inventory in its current form?  What new questions would be possible if all datasets were also available?
   3. It would also be good to talk about the longer-term vision / plan for the GEK. One option is to keep it as a standalone database; another would be for it to join Neotoma as a Constituent Database.  There are pros and cons to each option.  A standalone database allows for more autonomy and local control but raises questions of sustainability; small databases find it hard to persist over time.  You could point to other regional / national efforts such as the LAPD, APD, and Chinese Pollen Database as various options.  The first two have been merged into Neotoma and the latter has remained a standalone database.  The Japanese Pollen Database is somewhat in between, with some but not all of these data in Neotoma.  You could discuss options without reaching a firm decision on which to pursue, so as to leave options open.  Could also here talk about the importance of data governance and making sure that people feel like they have a voice in how their data are curated… this could e.g. lead to a working group, South Korean Pollen Council, or other group… describe options with invitations for others to join this effort.  Williams et al. (2018) provides a good description of Neotoma data governance.  Note of course that the authors have a much better sense of academic culture and norms in South Korea, so are best positioned to figure out how best to gently advance community-oriented data governance, while not pushing too hard.

See line-by-line comments below and attached PDF for more detailed comments.  My edits to the PDF are fairly extensive but are mainly focused on readability and brevity.  I’ve tried to keep substantive changes out of the PDF edits, but have flagged below in line-by-line comments any wording edits that might create a substantive change in tone or emphasis.

LINE-BY-LINE COMMENTS

L24-25:  This closing sentence in the abstract about data sharing initiatives is good, but it would also be good to talk about scientific vision and future questions / opportunities that are being enabled.

L50:  Should also add mention of open-source statistical packages, e.g. those developed by Mottl et al for rate-of-change analyses or Simpson for hierarchical GAMs.

L53 and elsewhere:  note that ‘assumptions’ is often being used when ‘analyses’ or other wording would be better.  Usually a scientist makes assumptions in the absence of data… the assumptions are precursors to formulation of hypotheses and the analysis of data.  Much of the paper is focusing on this later stage of hypothesis testing and analysis.

L62:  See edits to PDF, have revised text to clarify how regional databases can affiliate with Neotoma and advantages.

L64:  See proposed rewording

L85 and Fig 1: 

*Clarify what the blue bands represent, i.e. they are intended to highlight Korea and the absence of data

*Possibly could truncate the map and associated panel B at 65S and 85N given the absence of sites at these latitudes

L115 and Fig 2:

*typo in the second table: ‘depositionoanal’

L134-139:  This paragraph needs to be written for a broad ESSD audience, who won’t e.g. know what the space-for-time substitution method is.  Could cite Chevalier et al. (2020).  Also ‘paleo-site’ here and elsewhere reads as jargony, suggest instead ‘fossil site’ or ‘paleoecological record’.  And recommend here and elsewhere ‘surface sample’ instead of ‘surface site’ or ‘surface pollen’.

L150:  What is meant by ‘distinct geographic coordinates’?

L164 and elsewhere:  Avoid using a hyphen to indicate numeric ranges, because it can be easily confused for a negative sign.  Rewrite here to ’51 to 1,305 m’

Fig 6 / L215 and elsewhere:  Change ‘Others’ to ‘Other’ when using it as a categorical name.

L172:  Note that what the paper calls a sample (surface pollen, core, trench), Neotoma calls a collection unit.  In Neotoma’s naming system, a site (such as a lake) can have multiple collection units (e.g. multiple cores per lake) and multiple samples per core, distributed by depth.  Understood that this paper’s data schema doesn’t have to closely align with Neotoma… still, aligning language now will avoid much confusion later.

L179:  Given these (useful) definitions, I’d argue that ‘Section’ or ‘Stratigraphic Section’ should be used instead of ‘Trench’.  ‘Trench’ usually implies a ditch dug by human action.  The manuscript describes instead naturally occurring outcrops, which aren’t trenches.

L191 and elsewhere:  edit ‘Open-coastal zone’ to ‘Open Coastal Zone’  and edit ‘Volcanic cone’ to ‘Volcanic Cone’, capitalizing these names because you are proposing them as formally named categories. 

L191-197:  Can remove all quotation marks, because the capitalized names will be enough to establish these as defined entities.

L200:  Given that so many samples are soil samples, recommend adding ‘Soil’ as a category and reducing the scope of ‘Other’.

L213/Fig. 7: Typo ‘coatal’

L222:  Entering ages and depths as zero for the surface samples is not a good idea, because no measurements actually exist.  Setting ages to zero is a particular problem, because in the radiocarbon timescale, 0 = 1950 CE, which would be an incorrect age for these samples.  For depth, recommend leaving blank, and for age, recommend using the sampling year

L222-223:  Somewhere in this paper, state explicitly a) that you are using the radiocarbon timescale for all ages and b) remind ESSD reader that on this timescale, 1950 CE = 0.  Also, because some time-related variables use the AD/BC calendar year (e.g. for publications) while others us the radiocarbon timescale (e.g. for ages), be very explicit about which systems are being used for which variables. 

L226:  The statement that 96% of sites use only 14C dating doesn’t appear to match Fig. 8B, where the proportion looks more like 85%. 

L230 / Fig 8:  I don’t really understand the histogram plot shown in Fig. 8B.  How do the three histograms differ from each other?  They appear to contain overlapping information.

L236-237:  I don’t understand what this parenthetical about -73 years is trying to convey.

L238 & Fig. 9:  Recommend renaming ‘last glacial-interglacial cycle’ to ‘Pleistocene and Holocene’.  The last glacial-interglacial cycle is usually reserved for sites that span the last 100,000 years, i.e. a full glacial-interglacial cycle.  This renaming would also better align the other two categories ‘Pleistocene-only’ and ‘Holocene-only’

L275 & Fig 9: 

*typo: ‘Interglcacial’

*x-axis title in Panel B is incorrect, it should indicate thousands of years ago, not years ago

*Panel D:  Check the math used to calculate the number of dates per 1 cm depth interval.  The reported values for the Pleistocene sites (anywhere from 1 date per cm to 2.5 dates per cm) is unimaginably high.   

*The use of multiple scale breaks in the x-axis is non-traditional.  Recommend either just one scale break or perhaps experimenting with a log scale. 

L280 & Fig 10:

*Panel D legend:  delete word ‘record’

*Panel D:  the colors in the legend don’t appear to match the colors used in the map

L292-293:  Per comments above, because radiocarbon dates and U/Th use different timescales, need XX

L300 and Fig. 11: 

*What does ‘pMC’ stand for?

*1950 what?  CE?

*Note that this figure does a good job of clearly differentiating calendar years vs. 14C years by indicating 14C kyr BP for some plots and kyr BP for others.  This model should be adopted throughout ms.  Be sure to define kyr abbreviation somewhere.

*L311:  Please report the Pearsons correlation coefficient and significance test for this regression.

This paper and dataset compiles and examines records of Korean Quaternary paleoenvironmental proxy records and associated modern datasets. The authors performed a literature search to locate potential records, imposed some light quality controls to filter out some records, then assembled a database of metadata about the datasets. They then examined the dataset for different measures of comprehensiveness and bias, related to different paleoecological and geographic attributes.  Overall, the authors have compiled a very nice dataset that highlights, and then starts to solve, a gap in data accessibility for Korean Quaternary data. As noted by the authors, there are currently no dataset from Korea in the Neotoma Paleoecology Database and the region is underrepresented in global syntheses, despite there being a lot of good work on Korean Quaternary paleoenvironments in the literature. This paper represents a strong starting point for making Quaternary data from Korea more accessible.

I have four primary comments, and then some minor comments.

1. I would suggest adding references to the first two paragraphs of the paper, to bolster the literature basis of this introduction and help point readers from outside the subdiscipline to relevant literature. As the introduction stands, the authors make a number of assertions about the benefits of making data FAIR, but it would be very helpful to ground these assertions in specific examples or cases where compiling data across sites, or across proxy types, led to better or different inferences than a single site study alone.

2. Data accessibility:

The abstract says “To enhance accessibility, we have developed GeoEcoKorea, an online platform archiving raw data of the compiled studies or linking to it through our metadata, site inventory, and chron-depth datasets if the data is made available elsewhere.” The relationship between the database and the viewer seems clear on the surface, but when digging more into the viewer (see comment #3) and other parts of the text, there appear to be some points of disconnect.

Specifically, the GeoEcoKorea viewer is stated to provide access to archived raw data in some cases. But I don’t see these links in either the database description or in the viewer. Database description: Is there somewhere in the three linked tables that provides links to the raw data or a field/variable storing the link? The ERD in Figure 2 does not appear to include any fields that are linked to an original dataset (maybe through DOI in the Publication Metadata table) or linked to the raw data. GeoEcoKorea viewer: the description of GeoEcoKorea in section ‘4, Data Availability’, does not mention links to raw data.

Perhaps this is a place where “raw data” needs to be defined? I took this phrase to mean the base proxy data, like pollen or diatom counts, but perhaps this simply refers to the chron-depth dataset specifically, which is provided in some cases? Regardless, some additional clarity here would be helpful.

3. GeoEcoKorea viewer

a. First, building the viewer in addition to the database is a huge effort. The authors have developed a nice, simple display that is easy to navigate and understand.

b. The authors might consider coloring the markers differently between the surface sample and sedimentary records, for ease of location and navigation.

c. Related to my comment about data accessibility (comment #2), in GeoEcoKorea on page https://geoecokorea.org/itnl_pubs, there is a statement that “International data must first be uploaded to the Neotoma DB, after which GEK will establish a link to it.” Where would this link be stored in the database? I don’t see any specific fields. Are there any links currently established? If so, you could perhaps highlight one of them.

4. Links to Neotoma Paleoecology Database.

• For transparency, I want to note that I am the Associate Chair of the Neotoma Paleoecology Database (and am very supportive of this effort to make Korean data more visible!).
• In the Introduction, the authors mention the Neotoma database and it is also mentioned in GeoEcoKorea (see #3c). For example, the Introduction says “The Neotoma community also helps regional databases independently manage their own servers and affiliate with Neotoma for global networking.” However, then in section ‘5 Summary and Dataset Reuse Potential’, the future plans for the Korean database do not mention any affiliation with Neotoma, though it’s clear from browsing the GeoEcoKorea that they are planned (see item c in my comments about the viewer). It may be helpful to elaborate in Section 5 or elsewhere on how the authors envision (or actually are) linking to Neotoma.
• While this is absolutely not necessary for publication, if the plan is to continue enhancing links with Neotoma, consider adding a table that provides a cross-walk between the variables stored in your database and the variable names in Neotoma and providing that on the GeoEcoKorea website and/or in the FigShare supporting files. This could help enhance inclusion of Korean data into syntheses that use Neotoma for the majority of their data searches.

Minor comments:

• Figure 6. There is a note here that “The location of Jeju Island is shown in Figure 3A.” But I think the figure number needs to be updated – I think Jenu Island is shown first in Figure 4.
• Figure 9:
  • In panel B, the color of the horizontal lines seems to vary slightly – different shades of red, different shades of blue. This may be purely an individual color screen issue! But if there is indeed color shading, then perhaps figure out how to more clearly differentiate some of the blue shades from the black/gray ‘Pleistocene-only’ sites.
  • I had a difficult time interpreting the LIC acronym in panel D. I originally thought it meant “last interglacial”, but then realized it essentially means “encompassing portions of both the Pleistocene and the Holocene”. And also, in examining panel B, it’s clear that at least some sites extend into the penultimate glacial-interglacial transition as well. So perhaps change this to “late Quaternary”?

• Figure 11: I would define in the caption what ‘pMC’ means, noted in red in panel A.
• In the FigShare files, the “Publication Metadata” file does not contain a tab with the Column descriptions, whereas the other two files do. I would add this to the Publication Metadata file.
•