the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 17 Jul 2026
Tracing terrestrial organic matter dynamics in the subarctic North Pacific using sedimentary ancient DNA metabarcoding
Abstract. The subarctic North Pacific receives and preserves substantial amounts of terrestrial organic matter, yet the limited taxonomic resolution of traditional isotope and biomarker proxies constrains our ability to identify its source taxa and source regions, leaving its provenance and transport dynamics poorly understood. Here we use land-plant sedimentary ancient DNA (sedaDNA) metabarcoding to trace terrestrial plant organic matter in marine sediment cores from off-Kamchatka and the Bering Sea. We achieved unprecedented taxonomic resolution for marine sedaDNA records, identifying 328 terrestrial plant taxa, including 173 resolved to species level, across the two marine cores. Comparison with Siberian and Alaskan lake sedaDNA records showed that terrestrial plant assemblages recovered from the marine cores broadly reflected regional vegetation patterns and their temporal changes, while contrasts in relative taxon abundances indicated taphonomic filtering in marine sedaDNA records. Compositional relationships between the marine and lacustrine records, together with Siberian- and Alaskan-associated indicator taxa (e.g., Spiraea salicifolia and Shepherdia canadensis), provided taxonomic evidence for regional source attribution. Across the glacial–interglacial transition, the two marine records preserved distinct regional vegetation histories. Diversity patterns differed between the sites, with the Bering Sea record more strongly influenced by deglacial terrigenous delivery and the off-Kamchatka record more closely associated with postglacial warming. Our results demonstrate the potential of targeted sedaDNA metabarcoding as a high-resolution tool for tracing terrestrial organic matter in high-latitude oceans and highlight the need for expanded DNA reference databases and improved understanding of marine sedaDNA taphonomy.
- Preprint
(32579 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 14 Nov 2026)
- RC1: 'Comment on egusphere-2026-3670', Anonymous Referee #1, 04 Oct 2026 reply
-
RC2: 'Comment on egusphere-2026-3670', Anonymous Referee #2, 15 Oct 2026
reply
The subject of this paper, i.e. using land plant DNA metaB to trace taxonomic composition and continental source region of terrestrial OM in marine records, is very interesting and original.
Nevertheless, I have a few critics and comments.
The preservations biases of sedaDNA are not discussed enough. There are two very interesting paragraphs at the end of the discussion (from l. 586), which should be moved at the beginning of that section to shed light on the possible limitations of this proxy before discussing the results more in depth. In the M&M section, it is said that the cores were all sampled in 2009. Were they immediately subsampled for DNA extraction, if not when? At what temperature were the cores stored? Same question for the lacustrine samples. All this information is very important to have an idea of sedaDNA preservation.
Another question coming to mind that you discuss rather late in the manuscript is the nature of the organism parts carrying DNA, was it macro-rests, pollens, spores, free DNA? A sentence or paragraph about that earlier in the discussion may be interesting, particularly about the transport of these different parts. Pollens or fern spores being lighter may travel by air and deposit directly in the ocean for example. Is there an over-representation of taxa dispersed by wind?
There is a strong focus on the comparison between lacustrine and marine records, whereas the fate of OM in the ocean is taking much less space. Could it be possible to spend less space for the comparison and have a better equilibrium between both topics?
L. 95-98: What is the distance from the coast for the second core? This is an interesting info to compare both cores.
L. 105-106: How many grams of sediment extracted?
L. 125-126: Did you sequence these lacustrine cores yourselves? The info is in the table but say it also in the text.
Fig. 1: Temperate conifer forest not visible on the map, better to remove the caption, maybe precise that it is a present-day map.
L. 138: By individually do you mean that you did separate analyses for each group (lacustrine/marine) or each core?
L. 294: It is difficult to speak about relative abundance as there are many biases involved with metaB data (number of cells per organism, gene copy numbers,…).
Fig. 4: Relative abundances -> is the number of reads per sample shown somewhere?
Fig. 6: Is there any moss or tree in the graph, if not, better to remove the caption and say they are absent in the figure text caption.
Section 3.2: did you also take the distance from the sea (particularly the location of both marine cores) when comparing composition of lacustrine cores?
L. 383-386: Why are ferns better represented in marine cores, is it because their spores are lighter than pollens and travel more easily (e.g. by air)?
L. 423-428: Meaning that 11 candidates are left for Siberia and 9 for Alaska? Is it enough?
L. 455: Moderate correlation between taxon abundances in marine and lacustrine cores -> due to biases in record? Size, strongness… of plant remains?
L. 559-561: Is it due to their respective localisation or a different distance to the coast?
L. 568-571: Signal of ecological turnover synchronised between lake and marine cores? How long does it take to transport DNA from land to sea bottom? You may discuss that too.
L. 572-574: Is it due to an increased distance between source and deposit (because of sea level rise)?
L. 586-602: Very good discussion about sedaDNA biases, it may be better to move it at the beginning of the discussion as some of the topics discussed before are prone to this kind of biases.
L. 631: Incompleteness of reference DNA databases? In the rest of the article, you often claim that this database is very complete. You may need to balance these two claims.
L. 638-639: This aspect (transport- and degradation related bias) is not discussed enough before.
L. 662-662: This problem is only emerging at the end of the discussion; it is important to talk about it already before.
L. 755: It is possible to deposit the data in ENA and get the accession number before releasing the dataset to the public, why did you not do that?
Citation: https://doi.org/10.5194/egusphere-2026-3670-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 266 | 130 | 32 | 428 | 37 | 31 |
- HTML: 266
- PDF: 130
- XML: 32
- Total: 428
- BibTeX: 37
- EndNote: 31
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Review of "Tracing terrestrial organic matter dynamics in the subarctic North Pacific using sedimentary ancient DNA metabarcoding" by Lu et al.
The manuscript by Lu et al. presents an ambitious dataset, applying sedimentary ancient DNA (sedaDNA) and tnrl P6 loop metabarcoding to trace terrestrial plant communities into marine sinks. The identification of regional indicators like Sheperdia canadensis and the marine control Zostera are really interesting findings. However, before the manuscript can be considered for publication, it must address critical concerns in the methodology and statistics.
Therefore, the manuscript requires major revisions. Below are my major comments, followed by minor revisions and finally, small typography and formatting corrections.
General comments
Material and Methods
102-106. The methods rightly mention the use of the DNeasy PowerMax Soil Kit but do not specify the physical dimensions of the cores, core slices and the exact sediment mass processed per extraction. Reporting the core lengths, thickness of slices (I guess 1 or 2cm) and the input mass (e.g., 5 g, 10 g) of each slice is necessary to allow the reproducibility for sediment DNA extraction.
108. Using a Broad Range assay kit is unusual for degraded aDNA, which normally requires a High Sensitivity kit. Could you detail your approach and justify your choice in the manuscript?
113-114. Running PCRs for 50 cycles is quite high for those kinds of studies and increases the risk of chimera formation, primer-dimers, and index-hopping. Explaining your approach behind this unusually high cycle count is necessary. Moreover, detail the bioinformatic thresholds used in OBITools to filter the potential artifacts generated by this level of amplification.
118-119. The authors state that extraction blanks and NTCs were sequenced despite being negative in the PCRs. Though the authors rightly included those blanks in the sequencing run, sequencing reads for these blanks are not reported. Gel electrophoresis alone is not sensitive enough to confirm a blank is clean. Please report the total number of reads and taxa recovered from these negatives. If there was any contamination, the bioinformatic threshold for decontamination or the subtraction method applied to the samples must be detailed.
Results
155, 208, 217, 233. A mix of several data transformations has been applied throughout the manuscript, i.e., Hellinger for PCR replicates, CLR for temporal trends, untransformed relative abundances for the main Bray-Curtis NMDS, and a 0.125 power transformation for mean percentages. Untransformed Bray-Curtis allows highly abundant taxa such as Saliceae to dominate the ordination. Therefore, I would recommend standardizing the approach for the main community ordinations, e.g. Aitchinson distance, to better account for the species compositions. On the opposite, if authors want to keep these data transformation, a justification for each choice must be detailed in the revised manuscript.
224-229, & Figures 6 and B4. The 2D NMDS ordinations exhibit high stress values, i.e., > 0.15 and close to 0.20, which means that they poorly represent the multi-dimensional relationships in your data. Furthermore, to compare marine and lake samples, the authors ran a joint NMDS followed by a Procrustes analysis to prove the addition of marine samples did not distort the lake data's structure. Though the reasoning is pertinent, the approach is over-complexified. I would recommend switching either to a 3D NMDS to reduce stress or to a metric ordination (PCA, PCoA) that would allow marine samples to be passively projected directly into the lake dataset without distortion.
229-232. SIMPER inherently favors the most abundant taxa, which certainly highlights Saliceae and Dryas as the main dissimilarity drivers. Could the authors consider replacing or supplementing their SIMPER analysis with an Indicator Species Analysis (e.g., IndVal index through the indicspecies package in R)? It would allow the identification of species which are statistically significant ecological indicators, regardless of their baseline abundance.
520-525 (Discussion) & Figure 10 (Results). Plotting compositional DNA data alongside absolute mass accumulation rates of n-alkanes may not be the best approach as relative sequence reads do not scale linearly with absolute physical sediment fluxes. Authors should reframe this as a qualitative approach.
Discussion
124-127 (Mat & Meth) & 488-491. From my perspective, the comparison between the off-Kamchatka core with a pooled dataset of Siberian and Alaskan inland lakes represent quite a significant ecological stretch. Please detail your approach and add a dedicated paragraph acknowledging that the maritime climate of Kamchatka and volcanic soils likely exhibit a different flora that is not fully captured by interior tundra lakes.
Specific comments
Abstract
16-17. Precise if the 328 terrestrial plant taxa represent the “raw” unique assigned taxa or the final count after you applied the filtering and scaling with ranked subsampling.
Introduction
141-143. Specify the proportion of the total plant reads discarded due to the strict 100% identity threshold you applied during the reference database matching.
194-198. Controlling for unequal sample sizes by drawing 18 samples from each lake core is a nice approach choice. However, combining these to form a 162-sample pooled inventory means the marine cores are being compared against a vastly larger lake sample pool. The authors should briefly clarify if this sample imbalance inflates the 'lake-unique' taxa count and negatively influence the Sørensen similarity index.
281-282 & Figure B1. The Simpson index heavily weights dominant taxa; therefore, the high relative abundance of Saliceae (i.e., 71.8%) mathematically decreases the community evenness which does not obviously imply a biological negative interaction. Clarifying in the manuscript that this negative contribution of Saliceae may be a mathematical artefact is required.
Figures (Results)
Figure 3. One of the smoothing parameters of the GAM (k or sp) should be reported, either in the caption or methods to ensure the reproducibility of the curve fitting.
Moreover, it would make more sense to rotate your Figure to get your Age axis as the Y-axis such as in Figure 4 for constituency and for a better understanding of core dimensions
Figure 4 & L. 333-338. Plotting low abundances <0.2% at a visible 1% threshold using grey points creates a visual exaggeration of trace taxa. Explicitly define this choice in the main text when discussing these taxa so readers understand it indicates presence rather than 1% abundances.
Figures 6, 8, B3 and B4. There is an overlapping text issue on these figures. I guess that authors already used the ggrepel function in R, therefore, I recommend to subset labels to only show the most significant taxa.
Discussion
385. The detection of Zostera is an awesome result. Authors should highlight it even more as it validates the marine depositional context by being absent from terrestrial lake records.
444-445. Detecting S. canadensis in both cores at the same period is an interesting finding too. Authors could shortly expand on what may facilitate this “trans-basin” transport (e.g., oceanography, currents).
Technical corrections
905. There is a formatting error in the author list for Huang et al., (2021), i.e., "Pestryakova, Luidmila. A."