the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 21 Sep 2026
The Global Database of Deep-time Sedimentary Mercury Concentrations and Isotope Compositions (DeepSedHg)
Abstract. Mercury (Hg) concentrations and isotope compositions preserved in sedimentary successions provide important information for investigating long-term Hg cycling, volcanism, environmental perturbations, and Earth-system evolution. During the past decade, the rapid expansion of sedimentary Hg research has generated a large volume of concentration and isotope data. However, these data remain dispersed across individual studies and are reported using heterogeneous stratigraphic frameworks, metadata conventions, and analytical formats, limiting their accessibility and large-scale reuse. Here we present DeepSedHg (Deep-time Sedimentary Mercury Concentrations and Isotope Compositions) database, a comprehensive global compilation of published sedimentary Hg concentrations and isotope records spanning approximately 2.7 billion years of Earth history. DeepSedHg version 1.0 integrates data from 199 peer-reviewed publications published between 2010 and February 2026 and contains 28264 sample-level records. The database includes 27,990 Hg measurements, of which 3,736 contain both concentration and isotope data. Each record is accompanied by standardized metadata describing stratigraphy, numerical age, lithology, depositional environment, geographic location, paleogeographic position, and associated geochemical parameters. All records were subjected to systematic data harmonization and quality-control procedures, including duplicate screening, metadata verification, chronostratigraphic standardization, coordinate validation, and consistency checks of geochemical variables. The resulting database provides broad temporal, spatial, and environmental coverage of sedimentary Hg records and is distributed in a machine-readable format together with comprehensive metadata documentation. DeepSedHg is publicly available through Zenodo (https://doi.org/10.5281/zenodo.20713398) and an online tool (https://deepsedhg.pages.dev, which is intended to support future synthesis studies, meta-analyses, data-driven investigations, and interdisciplinary research on Hg cycling and environmental change throughout Earth history.
- Preprint
(1514 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 28 Oct 2026)
- RC1: 'Comment on essd-2026-511', Anonymous Referee #1, 27 Sep 2026 reply
-
RC2: 'Comment on essd-2026-511', Anonymous Referee #2, 28 Sep 2026
reply
Review of “The Global Database of Deep-time Sedimentary Mercury Concentrations and Isotope Compositions (DeepSedHg)” by Yuchen Yang et al.
The authors have invested substantial effort in constructing DeepSedHg v1.0. They compiled 199 peer-reviewed publications into a standardized, machine-readable database and applied systematic quality control, duplicate screening, metadata verification, and chronostratigraphic and geographic harmonization. The resulting database is highly valuable and will support syntheses, meta-analyses, spatiotemporal comparisons, and data-driven investigations of Hg cycling, volcanism, and Earth-system evolution. This is a meaningful contribution well suited to ESSD.
The manuscript is generally well organized and readable. It follows a logical structure appropriate for an ESSD database paper. The authors generally succeed in explaining what the database contains, how it was assembled, and how it can be accessed. The figures are useful for showing temporal, spatial, and geochemical coverage.
I have a few minor comments that I hope will help improve the quality of the manuscript. I recommend acceptance after minor revision. My specific comments are provided below:
Abstract:
Lines 13-15: The text should clarify that the database includes 27,990 Hg concentration measurements and 3,969 Hg isotope measurements, of which 3,736 contain both concentration and isotope data.
Lines 19-21: The abstract claims the database provides "broad temporal, spatial, and environmental coverage." However, Section 4.2 later acknowledges strong Northern Hemisphere bias (~91% of sites) and uneven temporal coverage. The abstract should be more cautious or explicitly note these biases to avoid overstating coverage. Is it better to revise it to “The resulting database provides broad coverage of sedimentary Hg records across geological time, although spatial and temporal coverage remains uneven, with sampling concentrated in the Northern Hemisphere and in certain geological intervals.”.
Data compilation, harmonization, and validation:
Lines 120-125: For Hg/TOC, Hg/TS, and Hg/Al, the manuscript states that published values were retained when available and calculated otherwise. Please specify how many records were calculated and reported.
Line 129: "ICS GTS v2024/12" and later "GTS v202412" (Line 203). Keep consistent
Database structure:
Lines 175–180: The manuscript states that the database was designed to support FAIR principles, but it does not explicitly map DeepSedHg to the Findable, Accessible, Interoperable, and Reusable criteria. Figure 1 shows the database architecture and field categories, but not how each FAIR principle is satisfied. Please add a short paragraph or table clarifying, for example: Findable through DOI, standardized field names, and metadata; Accessible through public CSV, Zenodo, and the online tool; Interoperable through machine-readable formats, standardized units, WGS84 coordinates, and ICS chronostratigraphy; Reusable through provenance, README, citation information, and versioning.
Line 213: Depositional-setting distinguish among marine, terrestrial systems, lacustrine, and other environmental settings (Such as ???).
Data coverage and characteristics:
Line 293: End Permian mass extinction (ETME) should be “End Permian mass extinction (EPME)”
Lines 321-323: In the Phanerozoic, the Cambrian forms the most distinct high-Hg interval, with a median of 200 ppb, whereas Mesozoic records are generally lower, with median values of 23 ppb in the Triassic and 27 ppb in both the Jurassic and Cretaceous. Should also give the IQR values.
Line 324: "Raw Hg concentration are" should be "Raw Hg concentrations are"
Line 336: Precambrian and Phanerozoic records have similar median values of -1.07‰ and -1.01‰, respectively. Should give the IQR values, also the below.
Lines 338-339: Should be “Around the major geological events such as the “five big” mass extinctions, δ202Hg values exhibit substantial fluctuations in response to massive environmental changes.” In addition, the statement that δ202Hg values “exhibit substantial fluctuations” is vague. How exactly are these fluctuations expressed? Please specify their magnitude, direction, and stratigraphic or temporal distribution.
Line 361: "comparatively narrower range" should be "a comparatively narrower range.".
Lines 361-367: Slope and R2 values in Figure 8 treat all data points equally, regardless of age or source. Given the large temporal and spatial heterogeneity, these regressions are descriptive at best. Please state this limitation.
Figure 6: the LOWESS-smoothed curves are based on age-specific median values. Given the highly irregular sampling density across geological time, these curves may be misleading. Please either remove them or clearly state that they are for visualization only and should not be over-interpreted.
Figure 8: For the regression, please report the uncertainties of the slopes and the corresponding p-values.
Citation: https://doi.org/10.5194/essd-2026-511-RC2
Data sets
DeepSedHg version 1.0: A global database of deep-time sedimentary mercury concentrations and isotope compositions Yuchen Yang; Songjing Li; Wang Zheng; Yi Liu; Jiubin Chen; Ruoyu Sun https://zenodo.org/records/20713398
Model code and software
DeepSedHg figure-generation code Yuchen Yang; Songjing Li; Wang Zheng; Yi Liu; Jiubin Chen; Ruoyu Sun https://github.com/TJUYyc/DeepSedHg
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 312 | 85 | 27 | 424 | 24 | 24 |
- HTML: 312
- PDF: 85
- XML: 27
- Total: 424
- BibTeX: 24
- EndNote: 24
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Review of “The Global Database of Deep-time Sedimentary Mercury Concentrations and Isotope Compositions (DeepSedHg)”
General comments
This manuscript presents DeepSedHg, a global compilation of sedimentary mercury concentrations and isotope compositions spanning approximately 2.7 billion years of Earth history. The database contains 28,264 sample-level records compiled from 199 peer-reviewed publications and integrates Hg measurements with stratigraphic, chronological, lithological, depositional and geographic information.
I found this to be a useful and timely contribution. Sedimentary Hg research has expanded considerably over the past decade, and bringing these dispersed datasets into a standardised and openly accessible resource will be valuable for the Hg and broader palaeoenvironmental communities. I particularly appreciate the effort involved in harmonising such a large number of records and retaining traceability to the original publications.
The manuscript is generally well organised and clearly describes the motivation, structure and potential applications of the database. I have several comments that I believe would help improve the clarity and long-term usability of DeepSedHg. Most relate to providing additional information about data provenance, uncertainties and the interpretation of some of the database-wide patterns.
Overall, I am supportive of publication following revision.
Main comments
1. Lines 14–23 – Clarification of the number of Hg and isotope records
The Abstract reports 27,990 Hg measurements, including 3,736 records containing both concentration and isotope data. Elsewhere in the manuscript, slightly different numbers are provided for Hg isotope measurements and records containing isotope parameters.
These numbers may simply refer to different subsets of the database, but I found them somewhat difficult to reconcile. It would be helpful to provide a small summary table showing the number of records containing Hg concentration data, any Hg isotope data, both concentration and isotope data, and the individual isotope parameters. This would make the overall structure of the database immediately clear to readers.
2. Lines 75–95 – Literature search
The literature compilation appears extensive and represents a considerable amount of work. To make the database easier to reproduce and update in the future, I suggest providing slightly more detail about the literature-search procedure.
For example, the authors could provide the main search strings, dates of the searches and perhaps a supplementary list of the publications considered. A simple flow diagram showing the number of studies identified, screened and included could also be useful.
Because the search included names of specific geological intervals and environmental events, it may also be worth briefly acknowledging that the database reflects, to some extent, where the research community has concentrated its efforts.
3. Lines 90–92 – Data obtained from figures
The authors mention that values were extracted from figures when raw data were unavailable. It would be useful to briefly describe how these values were digitised.
I also suggest including a simple provenance field indicating whether individual data were obtained from a manuscript table, supplementary information, repository or digitised figure. This would be valuable for users wishing to select only directly reported numerical data.
4. Lines 110–119 – Analytical uncertainty
I appreciate that the authors have retained published 2σ uncertainties where available and have not populated missing uncertainty fields. essd-2026-511
I would slightly revise the sentence suggesting that missing uncertainties “can be estimated” according to iGOS4M. Analytical uncertainty can vary among laboratories and analytical procedures. It may therefore be clearer to state that users may consult iGOS4M for guidance regarding typical analytical performance, while missing uncertainty values in DeepSedHg remain unspecified.
The statement that Hg concentration uncertainties were “mostly within ±10%” could also either be supported with a reference or slightly softened.
5. Lines 120–126 – Hg normalisation
The inclusion of Hg/TOC, Hg/TS and Hg/Al is very useful because these parameters are commonly used when interpreting sedimentary Hg records.
It would be helpful to specify the equations and units used when the authors calculated these ratios themselves. I also suggest including a field indicating whether each ratio was reported by the original authors or calculated during construction of DeepSedHg.
This is a relatively small addition that would improve transparency and reproducibility.
6. Lines 127–132 – Age information
The chronostratigraphic standardisation is an important strength of the database. At the same time, the numerical ages in the source studies are likely to have quite different levels of precision.
Where possible, it would be useful to distinguish ages based on direct dating from ages derived through interpolation, biostratigraphy or assignment to a stratigraphic interval. If this information is already included in the 109 database fields, it would be helpful to make this clearer in the manuscript.
For future versions of the database, age uncertainties or minimum/maximum ages would also be valuable where they are available in the original studies.
7. Lines 134–140 – Geographic and paleogeographic coordinates
The reconstruction of coordinates for studies that did not report numerical locations is a useful addition. I suggest adding a simple field identifying whether coordinates were reported in the original publication or estimated from maps/locality descriptions.
Similarly, it would be useful to clearly distinguish modern reported coordinates from model-derived paleocoordinates. The manuscript already appropriately notes the increasing uncertainty in paleogeographic reconstruction for older records. essd-2026-511
8. Lines 147–164 – Quality control
The quality-control procedure appears thorough. A little more information would nevertheless be useful.
For example, the authors mention “randomized secondary checks”. Could they indicate approximately what proportion of the database was checked in this way and whether this checking was undertaken independently by another member of the team?
Likewise, it would be useful to report approximately how many duplicate records were identified and removed during the process. These details would further demonstrate the considerable validation work that has gone into the database.
9. Sample, section and publication structure
Because DeepSedHg contains sample-level data, many individual samples may originate from the same stratigraphic section and publication. I suggest making this hierarchy particularly clear in the database documentation.
Ideally, users should be able to identify:
publication - site/section/core - sample.
This will allow future users to account for differences in sampling density among studies when conducting meta-analyses. It would also be useful to mention this issue briefly in the limitations section, because geological intervals represented by high-resolution sections naturally contribute more individual observations to the database.
10. Lines 305–351 – Figures 6 and 7
I found the temporal overview figures useful for demonstrating the scope of the database. I would simply recommend being slightly cautious about interpreting the LOWESS curves as representing global changes in the Hg cycle.
As the authors already explain in the limitations section, temporal sampling density is uneven. essd-2026-511 Some geological intervals and individual sections contain many more measurements than others.
It may therefore be helpful to describe the LOWESS curves explicitly as summaries of the available records in DeepSedHg. A sentence noting that they are intended primarily to visualise the database rather than provide a formal reconstruction of the global Hg cycle would be sufficient.
11. Lines 321–325 – Hg enrichment around geological events
Related to the previous point, I suggest replacing “significant Hg enrichments” with “elevated Hg concentrations” unless statistical significance has been formally evaluated.
It may also be useful to remind readers that raw Hg concentrations can be influenced by host phases and sediment composition, which is precisely why the inclusion of Hg/TOC, Hg/TS and Hg/Al in DeepSedHg is valuable.
12. Lines 353–366 – Relationships among Hg isotopes
Figure 8 provides a useful overview of the isotope data contained in DeepSedHg.
For the relationship between Δ²⁰¹Hg and Δ¹⁹⁹Hg, the manuscript states that most observations cluster near the 1:1 reference line. It would be useful to explain briefly why the 1:1 relationship is shown and how it should be interpreted relative to the fitted relationship.
For Δ²⁰⁰Hg, I would also mention that some of the variation is relatively small compared with the other isotope parameters and that users should consider analytical uncertainty when interpreting these relationships.
13. Lines 401–419 – Limitations
I appreciate that the authors have included a clear limitations section. I suggest adding two further points.
First, individual sample records are not necessarily statistically independent because many samples can come from the same stratigraphic section.
Second, there may be a research-effort or publication bias towards major geological events, particularly mass extinctions and intervals associated with LIP activity.
These are not shortcomings of the database itself, but they are useful considerations for researchers who use DeepSedHg for future quantitative analyses.
14. Lines 401–419 – Geographic coverage
The strong Northern Hemisphere representation is worth emphasising. The database shows that approximately 91% of sampling sites are in the Northern Hemisphere, while regions including South America, Africa and Australia are comparatively poorly represented. essd-2026-511
I see this as an interesting outcome of the compilation rather than simply a limitation. DeepSedHg helps identify major geographic gaps in the existing sedimentary Hg literature and therefore provides useful guidance for future research.
The authors might consider highlighting this point slightly more strongly in the Discussion.
Minor comments
Line 43: “advance in analytical techniques” → “advances in analytical techniques”.
Lines 55–57: “the need for a comprehensive data principles and standardized database” needs grammatical revision. Perhaps: “the need for a comprehensive and standardised database consistent with FAIR data principles”.
Lines 94–95: The authors describe DeepSedHg as “the most comprehensive collection”. I suggest either briefly supporting this statement or using the slightly more cautious “a comprehensive global collection”.
Lines 115–119: EPA Method 7473 involves thermal decomposition, amalgamation and atomic absorption spectrophotometry. It would be useful to use the full description here.
Line 121: “depositional environment” should be “depositional environments”.
Lines 166–175: I strongly encourage the authors to ensure that the data dictionary provides a clear definition, unit and description for all 109 fields. essd-2026-511
Throughout: Please ensure that terms such as “records”, “measurements”, “samples”, “sites” and “sections” are used consistently, as these represent different levels of the database.