the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 03 Jan 2024
Cenozoic pelagic accumulation rates and biased sampling of the deep sea record
Abstract. Global weathering is a primary control of the earth's climate over geologic time scales: converting atmospheric pCO2 into dissolved bicarbonate; with carbon sequestration by marine plankton as carbonate and organic carbon on the ocean floor. The accumulation rate of pelagic marine biogenic sediments are thus a measure of weathering history. Previous studies of Cenozoic pelagic sedimentation have yielded contrasting results, though most show a dramatic rise (up to 6 times) in rates over the Cenozoic. This contrasts with model expectations for approximate steady state in weathering, pCO2, and sequestration over time. Here we show that the Cenozoic record of sedimentation recovered by deep sea drilling has a strong, systematic bias towards lower rates of sedimentation with increasing age. When this bias is removed accumulation rates are shown to actually decline by ca 2 times over the Cenozoic. When accumulation area however is adjusted for changes in available deposition area, global weathering is shown to have nearly doubled at the Eocene-Oligocene boundary, but was otherwise essentially constant. Compilations of other metrics correlated to sedimentation rate (e.g. productivity, biotic composition) also must have a strong age bias, which will need to be considered in future paleoceanographic studies.
- Preprint
(1853 KB) - Metadata XML
-
Supplement
(673 KB) - BibTeX
- EndNote
Status: closed
-
RC1: 'Comment on egusphere-2023-3087', Adriana Dutkiewicz, 27 Jan 2024
This is an interesting and useful paper that describes a new approach to understanding the extent to which pelagic sediment accumulation rates in the global ocean have changed through the Cenozoic, and to what extent these observed changes are the result of drilling bias. The paper is novel and will be of interest to a wide readership using scientific ocean drilling data. The manuscript should be published after moderate revisions, that include clarification of the methodology, reconsideration of the CCD model used, better illustration of depositional area computations, and inclusion of new figures (locality map, methodology flow chart, seawater Sr curve). The methodology used in the analysis is very difficult to follow, and needs to be explained/visualised better. Some of the writing should also be tightened and figures improved, particularly in terms of font sizes (Figs 2 and 5, especially). See detailed comments below:
Abstract
The statement “The accumulation rate of pelagic marine biogenic sediments are thus a measure of weathering history” (line 3) is incorrect and should be reworded. Perhaps replace “thus a measure” with “an indication”, bearing in mind that not all continental weathering products end up on the seafloor (e.g., dissolution of carbonate) and not all weathering products come from continents (e.g., seafloor weathering).
In the abstract the authors state that “When accumulation area however is adjusted for changes in available deposition area, global weathering is shown to have nearly doubled at the Eocene-Oligocene boundary.” It would be useful to see maps of these time-dependent deposition areas, and a graph showing the time-dependence of these areas in individual ocean basins in the manuscript or supplement. The statement also assumes a straight forward correlation between deep-sea sedimentation and continental weathering, which is an oversimplification. The authors also need to explain why this would have happened at the EOT.
Other comments about the abstract:
L1: capitalize Earth (and elsewhere in the manuscript)
L2: pCO2 is written incorrectly
L3: change “are” to “is”
L6: change “deep sea drilling” to “deep-sea drilling” (do this elsewhere and for other compound adjectives in the manuscript)
Methodology
The hypothesis is that the older the sediments recovered are, the more likely they are to be represented only by low sedimentation rate sites, as penetration to older aged sediments is more likely when the local sedimentation rate is low. I have no doubt that this effect exists, but it is questionable whether this is the only effect driving the observation that sedimentation rates decrease with age.
The bias correction is based on this: "The inverse of the resulting modelled LSR vs geologic age curve is used as a correction factor on the NSB-based LSR and SAR compilation."
I don't understand why the correction for this effect is based on LSRs which are not corrected for compaction (unless I understand it incorrectly, and the modelled LSR with age includes a compaction correction). Compaction clearly plays an important role in decreasing apparent LSRs with increasing age, so the compaction effect needs to removed first. This automatically happens when calculating sediment accumulation rates (SARs), so isn't it logical to compute the bias-correction from SARs, as opposed to LSRs which themselves suffer from a bias from compaction, which in turn could bias a bias correction?
Next, the authors are following the assumption that the long-term LSR change with age is due to a drilling bias, and this is implemented by applying the inverse of this relationship as bias correction. The problem that I see is that if your assumption is incorrect, in the sense that there are long-term reductions in sedimentation rate that are not in fact due to drilling bias, then the "corrected" accumulation rates will be biased themselves.
I see some potential ways for further testing. One optional thing that could be done is to exclude all sites that were drilled in very high productivity areas, i.e., sites that were deliberately focused on Neogene very high sedimentation rate areas, including most sites in the equatorial Pacific, and sites from coastal upwelling regions (e.g., Benguela current sites offshore Africa etc). Likewise, regions of sediment focusing (contourite drifts) are also anomalous and would need to be excluded. What do the rest of the sites look like in terms of LSR with age? I realize that this would be a fair amount of work so perhaps the authors can assess if this is worthwhile doing.
Figure S2 shows that the accumulation rate data uncorrected for drilling bias (black curves) are quite different between the ocean basins. In this context it would be of interest to see the uncorrected LSR data subdivided into ocean basins as well, to get a sense of what the curves look like before the correction is applied (keeping in mind that the correction is a model that might perhaps be incorrect). For instance, in the Atlantic Ocean, the "raw" SARs are essentially constant at times before 15 Ma (long-term trend) while the Pacific Ocean shows a long-term decrease in SARs with increasing age, the Indian Ocean somewhere in between, and the Southern Ocean shows ups and downs entirely different from the other ocean basins. With so much inter-basin variability, I start wondering how meaningful a global correction for sedimentation rate with age is, and how valid the premise of the paper is (but again, this is difficult to judge without seeing the uncorrected data, which should be shown).
In Figs 7 and S1, are the linear sedimentation rates shown corrected for compaction? This is not clear from the captions. In the text, please provide an estimate of how much the sedimentation rate would have been affected by compaction.
Fig. 6 shows corrected linear sedimentation rates, and corrected SARs. The figure should show both uncorrected and corrected rates, as well as the correction function, to make this more transparent.
Lastly, the entire workflow is so complex that a casual reader of this paper currently would have no hope to understand what was actually done. It is quite possible that some of my comments above merely reflect that I didn't fully understand the methodology. This can be fixed by a flow chart that covers each step in the analysis, to make the paper clearer and more comprehensible to a general audience.
Deposition area computation
A simple global median curve based on ‘red clay’-free oceans (i. e., 74.2% of the Atlantic, 51% of the Pacific and 74.7% of the Indian Ocean) instead of their full area is shown in Fig. 7. But aren’t these percentages time dependent? This is problematic because the median curve is for the entire Cenozoic whereas the proportions are for the present-day. Not only has the area of the individual basins changed substantially since 66 Ma, but so has the pattern of sedimentation and lithology distributions. For example, the CCD has deepened substantially since 66 Ma, meaning that the global area available for carbonate sedimentation has also increased. As mentioned above, it would be helpful to have more information be supplied about what the deposition area through time actually looks like based on the authors’ choices of paleo-topography maps and CCD changes.
Have ‘red clay’ sections been removed from the LSR compilations in the 400+ drill sites? Following the logic above, failure to do so introduces a lithological bias, if not for the present-day then for some other time in the Cenozoic. We know that the CCD has deepened significantly since ~ 55 Ma (e.g., Palike et al., 2012) meaning that the accumulation rate of carbonate was a lot lower in the past thanks to dissolution, and that sedimentation was dominated by much more slowly accumulating pelagic clay. So older sections could simply correspond to more slowly accumulating sediments rather than a drilling bias. This point needs to be clarified and some of these ideas explored in more depth. A map showing the distribution of drill sites should be included in the manuscript.
Line 104: The Pälike et al. CCD for the equatorial Pacific is not representative of the global ocean. This is an upwelling region of high productivity, which is strikingly different (much deeper in terms of CCD) from the rest of the Pacific, and most of the Atlantic and Indian oceans. The authors should consider using a more appropriate model or models (e.g., global curve of Boss and Wilkinson, 1991), that also cover the entire Cenozoic.
Line 157: CCD trends in other ocean basins do not follow the trend for the Pacific CCD. This is evident in Campbell et al. (2018) and in Dutkiewicz and Muller (2022). The differences are quite striking. The statement should be corrected.
Other comments
Please refer to in-text figures in correct order. For example, Line 59 refers to Fig. 2 then Line 63 refers to Fig. 7. Fig.1 is first cited on Line 79 and there is no mention of Figs 3–6 prior to Line 63. This needs to be fixed as it’s greatly reducing the readability of an already complex manuscript.
Please improve the readability of some of the figures. In many cases the fonts are too small (Figs 2 and 5, in particular).
Line 159: “Global mean deep sea pelagic sediment accumulation rates have decreased slightly during the Cenozoic.” Relative to what? Please clarify.
Line 60: change “representative of pelagic sedimentation” to “representative of biogenic pelagic sedimentation”.
Line 133 onwards: state which figure this text is referring to.
Line 175: “Our results thus largely reconcile the discrepancy that until now has existed between 87Sr/86Sr estimates of Cenozoic weathering, and the deep sea pelagic accumulation rate record.” Please demonstrate this using your data and the seawater Sr curve. The curve will also provide an independent check of whether an increase in deep-sea accumulation rates corresponds to an increase in continental weathering.
References:
Boss, S.K., Wilkinson, B.H., 1991. Planktogenic/Eustatic Control on Cratonic/Oceanic Carbonate Accumulation. The Journal of Geology , 99, 497-513.
Campbell, S. M., Moucha, R., Derry, L. A., & Raymo, M. E. (2018). Effects of dynamic topography on the Cenozoic carbonate compensation depth. Geochemistry, Geophysics, Geosystems, 19, 1025–1034.
Citation: https://doi.org/10.5194/egusphere-2023-3087-RC1 -
AC1: 'Reply on RC1', Johan Renaudie, 14 Mar 2024
We thank all reviewers for their very helpful, constructive and detailed comments on our ms. There are many specific comments on ms contents which we will address as individual answers, and which we do not discuss here. We reply here to the main points of first two reviewer's comments, which are very similar.We are pleased that all reviewers see the work generally positively and, assuming successful revision, a useful addition to the literature. The first two reviewers also however note that the current version is very hard to follow, as the methods are not described at all clearly or in enough detail, and that the order of presentation is poor. We will indeed do a major rewrite of the introduction, materials and methods, improving organisation, and giving more detail on both the methods, and the reasoning that underlies our analyses. We briefly summarize the latter here to aid reviewers and the editor.Our ms is a bit unusual in that the analysis presented here is in fact composed of several sub-analyses carried out sequentially, with each building on the results of the prior sub-analysis. It is thus useful, before going into details in the ms, to not only summarize the methods, but also the results of each substudy in sequence, so that the reasoning and methods for each subsequent substudy is clear. We note that this presentation style does not strictly follow the conventional standard, used in the submitted ms, which gives all methods first, all results second, and discussion of results only last. Should the reviewers/editor find the following exposition useful, we ask permission to include this, suitably modified, as an additional section betweeen the Introduction, and Material and Methods parts of the revised ms. We would reduce some of the expository text currently elsewhere in the ms to prevent duplication.Substudy 1 - Impact of hiatus distribution on SAR[current ms Figures 4 and 5]In this sub-study we showed that hiatuses in deep sea pelagic sediment sections were relatively infrequent and mostly of short duration, nor do they increase in magnitude with increasing geologic age. Thus the apparent decline of accumulation with age could not be due to biassing by the cumulative effect of common, and with increasing scale of measurement, ever larger hiatuses on the calculation of apparent accumulation rates (aka the Sadler effect, as had been speculated by some prior authors). This result suggested that another cause, or set of causes must be responsible.Substudy 2 - By-section Analysis of declining accumulation rates with increasing geologic age[current ms Figure 1]We next tested a hypothesis of a uniform decline of average accumulation rate vs increasing age over the entire ocean. For this we examined the record of accumulation rate within a large number of individual sections. The accumulation rate for each segment of the age model for each section was expressed as the ratio of the mean rate for the whole section. The ratio value and midpoint age of each age model segment from each individual section were then compiled across all the sections in our study. This had the effect of removing any between-section differences in section average accumulation rate, leaving only the temporal trend in change in relative accumulation rate vs age. We were looking for the expected signal, at least in the composite global dataset, of major (ca 6X) declining average relative accumulation rate with increasing age, as had been reported by prior global summaries of deep sea sections in the literature. Instead, we discovered that, despite substantial short term variance of individual age model segments within the individual sections, the composite of relative change vs age for the set of all sections did not show any detectable trend towards lower average accumulation rate with increasing age.Substudy 3 - A Model of How Between Section Differences in Average Accumulation Rate Interact With Incomplete Recovery of Cenozoic Sections via Drilling to Create a Bias in Compiled Data[current ms Figures 3, 2 and 6 and 7]The near constant rates of relative accumulation found in our second sub-study, where we had removed between site differences in accumulation rate, suggested that the signal of declining accumulation rate vs age as reported in prior studies must in some way be connected with these differences between sections in accumulation rate, and/or in how the data for different sections was composited in these prior studies. In the course of both making age models for many sections over a period of many years, and during the above synthesis of relative accumulation vs age, it was apparent that a) most sections with high average accumulation rates ended in relatively young sediments, while sections reaching older sediments typically had low average accumulation rates; b) there was no obvious difference in the total depth (in meters below sea floor) reached by drilling between these different types of section; and c) only a few sections actually recovered the entire sedimentation record for a location (ie reached basement), or even recovered the entire Cenozoic time interval - most ended prior to base of the Cenozoic, constrained by limits imposed on total drilling depth from Leg priorities, bad weather, stuck drill bits etc. This led to the insights that: the age of recovered sediments in the global deep sea drilling data is determined in part by constraints from the drilling process, not just primary sedimentary record; and, the only ways (barring the occasional major hiatus) to reach older time intervals is to have chosen (deliberately or by chance) to drill at a location with a low average accumulation rate, and/or to drill an unusually deep section. Further, if one were to compile (i.e. bin) accumulation rate data across sections by time interval, younger time interval bins would have a mix of both high and low average rate sections, and thus higher bin-average accumulation rates, than older interval bins, where only sections with low average within section rates can normally be recovered.Our third sub-study explored the magnitude of this 'drilling bias' in a model, generating large numbers of model drill sections, and compiling (binning) the average accumulation rate across sections for 1 my time bins over the Cenozoic. Each model section was defined by just two main parameters: the average accumulation rate for the section, and the total depth drilled by the section. The model assumed for simplicity that any individual section had a constant rate of accumulation vs time. This was based on the results of our second sub-study on relative rates with time within single sections. There we had found, other than random variance (here treated as noise, and ignored), essentially no trend in the rate of accumulation vs time for any given section. The accumulation rate values for model sections were chosen by random sampling of the actual distribution of accumulation rates in 'recent' ie late Neogene sediment sections recovered by the deep sea drilling programs. Similarly, the total depth drilled was chosen by random sampling of the actual distribution of maximum drilling depth from the sections recovered by the deep sea drilling programs. All analyses were also adjusted to remove the effects of sediment compaction. Lastly, as hiatuses do exist in pelagic deep sea sections, our model included these as well, as occuring at random within a section, with a per time interval probability and time duration that match the actual data from the age model library of deep sea drilling. Our model thus simulated how, with constant actual accumulation rate vs time at any location, the variation of average accumulation rates between sections, and random variation in drilling depths between sections, affected the composite (binned) accumulation rate across sections. The magnitude of this drilling bias turned out to be quite large, and in fact accounted for the entire ca 6X change in apparent binned average accumulation rates across deep sea drill sections over the Cenozoic. When the effect of this bias was substracted from the 'raw' across section binned data, the resultant 'drilling bias corrected' accumulation rate over the Cenozoic showed only a slight (2X) decline - and thus in broad accordance with the near constant rates seen in our earlier (second) substudy of relative trends within sections vs time. We also looked at how different ocean basins might have different Cenozoic accumulation rate histories. These, when compiled together using a basin area adjustment matched as expected the global curve. There were however some differences in individual basin histories.Substudy 4[current ms Figure 8]In our fourth sub-analysis we applied our new, largely constant accumulation rates over the Cenozoic curve to the question of Cenozoic rates of global weathering. For this we examined only the dominant component of pelagic sediment output of weathering supplied dissolved nutrients - the calcium carbonate sequestered in pelagic sediments by marine plankton. This was also done as relatively little information is available on the Cenozoic distribution of the other main component of weathering - biogenic silica in sediments; and in older intervals, how it has been affected by diagenesis to chert. To convert the accumulation rates for sections for each geologic time interval from our earlier substudy 3 into global accumulation rates per time interval of carbonate material, we calculated the area of the ocean accumulating carbonate for each time interval. The area of the oceans above a given depth (the CCD) was calculated from paleotopographic maps for each time interval of the Cenozoic. To further simplify the analysis, we considered the global ocean accumulation rate of pelagic carbonate to be defined by the area of ocean accumulating carbonate above just the Pacific Ocean basin CCD. This was done as the Pacific basin is by far the largest basin, and the Pacific CCD history for the Cenozoic is the best defined by published studies.Another issue brought by the first two reviewers has to do with changes in CCD in two of the analyses:- In the computation of the SAR corrected by ocean basins area (in which we excluded the [modern] area occupied by red clays)- In the final estimation of total sediment flux in the Cenozoic, where we used the Equatorial Pacific history of CCD reconstructed by Pälike et al. 2006.Computing the total area above CCD as we did in the second case is relatively easy but splitting it into basins (with necessitates reconstructing the basin borders through time) is computationally more extensive. Additionally, not all basins have a complete Cenozoic published CCD history.What we can do for the first case is using the basin area without correction for red clays, as well as correcting for red clays using the global CCD estimate (such as Boss and Wilkinson, 1991 as proposed by the first reviewer), as a minimum-maximum estimates of that curve.We are also happy to try and calculate the total sediment flux using individual CCD histories for each ocean basin with the best published estimates, if possible, and present them alongside one computed using a global CCD history.Another recurring comment is the over-emphasis on the link between global weathering and pelagic sedimentation: we will make an effort to reword this.Lastly, we also apologise for not having checked the ms carefully enough prior to submission to insure such basic things as having the figure numbering in correct order. These, annoying if only cosmetic errors will of course also be fixed in our revision.-------------------------------------
In addition to the reply above addressing both RC1 and RC2 , we would like to answer some additional important comments from RC1 here:
In the abstract the authors state that “When accumulation area however is adjusted for changes in available deposition area, global weathering is shown to have nearly doubled at the Eocene-Oligocene boundary.” It would be useful to see maps of these time-dependent deposition areas, and a graph showing the time-dependence of these areas in individual ocean basins in the manuscript or supplement.
Indeed this is a good idea, we will produce such maps and add them to the SOM.
I don't understand why the correction for this effect is based on LSRs which are not corrected for compaction (unless I understand it incorrectly, and the modelled LSR with age includes a compaction correction).
[and further comments]SAR is computing by multiplying LSR with the actual, measured density of the cores, thus already correcting for compaction. Using the compaction-corrected model output to correct the compiled SAR would thus over-correct for compaction.
I see some potential ways for further testing. One optional thing that could be done is to exclude all sites that were drilled in very high productivity areas, i.e., sites that were deliberately focused on Neogene very high sedimentation rate areas, including most sites in the equatorial Pacific, and sites from coastal upwelling regions (e.g., Benguela current sites offshore Africa etc). Likewise, regions of sediment focusing (contourite drifts) are also anomalous and would need to be excluded. What do the rest of the sites look like in terms of LSR with age? I realize that this would be a fair amount of work so perhaps the authors can assess if this is worthwhile doing.
The first part might be doable indeed in a reasonable time-frame. We will see if we can do that and add it at the very least in the SOM.
Fig. 6 shows corrected linear sedimentation rates, and corrected SARs. The figure should show both uncorrected and corrected rates, as well as the correction function, to make this more transparent.
Indeed. We will modify the figure accordingly.
Line 175: “Our results thus largely reconcile the discrepancy that until now has existed between 87Sr/86Sr estimates of Cenozoic weathering, and the deep sea pelagic accumulation rate record.” Please demonstrate this using your data and the seawater Sr curve. The curve will also provide an independent check of whether an increase in deep-sea accumulation rates corresponds to an increase in continental weathering.
We could indeed add such a curve to Figure 7.
Citation: https://doi.org/10.5194/egusphere-2023-3087-AC1
-
AC1: 'Reply on RC1', Johan Renaudie, 14 Mar 2024
-
RC2: 'Comment on egusphere-2023-3087', Sophie Westacott, 14 Feb 2024
Overview
This work is a valuable contribution to the ongoing discussion of pelagic sedimentation rates through the Cenozoic. It considers a possible bias created by drilling depth limitations resulting in a lower likelihood of reaching older sediments at sites with higher sedimentation rates, and thus most older-sediment sites being from areas of low sedimentation. It does this using randomized sampling of age models from the Neptune (NBS) database, and goes on to estimate global sediment accumulation rate through the Cenozoic by factoring in CCD depth, red-clay-free area, ocean basin, intra- vs inter-hole variability, age model quality, hiatus length and frequency, and compaction. It will be useful for anyone working on whole-Cenozoic comparisons or trends in pelagic sediments, including microfossils.
However, the manuscript requires considerable revision before it will be readily digested by readers, as it is currently very challenging to follow. The explanation of the central method—using a null model to test for the predicted drilling bias—is particularly confusing, and as detailed below, the logic behind some of the interpretations of the results also needs further elaboration. The manuscript is also rather unpolished: the figures are out of order, the axes on most of them are too small, a few of them (e.g. Figs. 1, 2 and 5) are rough drafts rather than finished products, the text has grammatical errors, and one or two references to figures don’t make sense—none of which helps with following what was done. The content of the paper seems worthy of a tighter, neater form that will allow others to fully engage with it.
Broad Comments
-The inter- vs intra-hole variability is a key finding, and the argument for a drilling bias hangs on it: if higher sedimentation rates nearer the modern do not relate to higher sedimentation rates earlier in the Cenozoic within each hole, then the drilling bias explanation doesn’t hold. I think the overall thesis of the paper would thus be made stronger by exploring this finding more, and by clarifying the (currently very confusing) part on ‘by-section normalization’.
-The use of red-clay-free area is an interesting new approach, but I have the same question as the first reviewer—what about the fact that the amount of red-clay-free has shifted over the Cenozoic? The maps in Wade et al. (2020), for example, show a much smaller red-clay-free area of seafloor in the early Cenozoic than later. Why is the basin median and mean calculated from the modern red-clay-free area, but then the CCD used in compiling the SAR curve? Although it would include a large unknown, estimating the red-clay-free area over time and using that estimate in the SAR calculations would likely be more accurate than assuming the same size of red-clay-area over time. Furthermore, if the same amount of carbonate and opal that is buried today were to be buried in the much smaller red clay free area of the early Cenozoic, apparent sedimentation rates would need to be substantially higher in the earlier Cenozoic than in the modern, no? Is that reflected in the raw LSR and SAR data? It isn’t clear to me in Figure 1 that it is, but it’s hard to tell from the way the data is depicted. Using the CCD combined with topographic data makes sense, and seems like a worthwhile approach, but I do wonder about opal sedimentation rates, particularly in the Southern Ocean. While these may be more minor in terms of global sedimentation rates, could they be useful in understanding at least the Southern Ocean history? A more detailed discussion of the topic would be helpful.
-The manuscript emphasizes a very direct link between weathering and pelagic sedimentation, a claim that needs more support than it’s given. Lines 39-41 say “If the deep sea pelagic sedimentary record is not (strongly) biased by hiatus-driven measurement scale artifacts, then the major increase in apparent rate over Cenozoic means either that geochemical proxies are somehow wrong and rate has changed exponentially, or that some other factors are biasing the Cenozoic deep sea record.” The debate over the interpretation of those geochemical proxies is alluded to, but since the discrepancy between weathering and sedimentation rates is set up as the driving question this study is aiming to resolve, more detail on the current state of that debate seems called for, including more recent literature (e.g., Caves Rugenstein, Ibarra and von Blanckenburg, 2019; Katchinoff et al. 2021; Pogge Von Strandmann, Kasemann and Wimpenny, 2020). Line 175-176 makes the claim that “Our results thus largely reconcile the discrepancy that until now has existed between 87Sr/86Sr estimates of Cenozoic weathering, and the deep sea pelagic accumulation rate record.” This certainly needs a discussion of at least the Sr record along with a visual comparison of the timeline of the strontium data with the updated SAR curve. Is the EOT SAR step-change shown in Fig. 8 reflected in the Sr data? What about the non-directional but still substantial SAR variation in the Neogene?
-Additionally, shallow marine settings are rejected as a possible alternative to pelagic settings for the burial of earlier Cenozoic weathering products, but the two citations listed—particularly Ridgewell and Hargreaves (2007)--do not by themselves provide an obvious explanation for why this is the case. Elaboration would be helpful here, and updated literature--e.g., for silica, Treguer et al. (2021), which substantially updates the Treguer and de la Rocha (2013) estimates of shallow water Si burial, and Rahman et al. (2017). (Although it would mean more work and is perhaps beyond the scope of the paper, it would be interesting to see a quantification of the magnitude of variation in shallow marine carbonate and silica burial that would be required to accommodate a 2-6x shift in pelagic sedimentation without a change in total weathering product sequestration.)
-Similarly, several authors have suggested that biogenic pelagic sedimentation became more spatially concentrated over the Cenozoic (e.g., Dunlea et al 2017; Barron and Baldauf 1990). Some discussion of this hypothesis and how this study’s results fit in with it would be beneficial, given that one of the exciting novelties of this work is its spatial specificity.
Broadly speaking, to extend the conclusions of these analyses beyond pelagic sedimentation (interesting in its own right) to summed global weathering and make the claim that this solves a lynchpin of the Cenozoic weathering debate, there should be a more in-depth discussion of the abundant empirical (e.g., stable and cosmogenic isotopic, geophysical, terrestrial sedimentary, paleo-CO2) and model-based work that has been done on the subject and its conflicting interpretations, as well as alternative hypotheses.
Specific Comments
As mentioned above, there are numerous minor errors and style issues throughout the text that, although I have not listed them here, should be cleaned up before publication. The abstract and the last paragraph of the introduction in particular need some serious textual work.
L42: the way ‘decline’ is used in the last sentence of the introduction, while correct, is confusing in the context of ‘increase’ in the paragraph before.
L55: Is it a running median with an overlapping 10 ky window, or a median (and mean) taken at 10 ky intervals starting centered at 5 ky? The former would make more sense for the purpose, I would think. It would also be helpful to understand why 10 ky was selected as the window size.
L60: Given the wide range in sampling coverage between basins (12-220 sites), I wonder what a sort of rarefaction/bootstrap type analysis might show? In other words, what heterogeneity do you find if you subsample the Atlantic in groups of 12 sites with replacement? Just curious…
L63: Figure 7 is out of order, and Figure 1 isn’t called out in the text until much later... Please order the figures correctly.
L58: Figure 2 is cited here, but it doesn’t do much to show how either LSR or SAR are calculated. It would be helpful to have a different figure showing the process and/or more clarifying text.
Figure 3: Frequency would, I think, be more useful to the reader than density in these histograms. Also, panel labels (letters or numbers) here and throughout would make the figures easier to relate to the text and caption, rather than ‘top left’, ‘bottom right’, etc.
L77 cites a density plot of SAR vs time span in the Supplementary Material, but I can’t see one there. Does it mean Figure 5?
L77: “We then calculate by-section normalized change in sedimentation rates vs geologic age in intervals with sedimentation to see if older time intervals, corrected for between site differences in local rates, show a decline in relative rate with increasing geologic age (see Figure 1).” I’m not sure I follow what was done here. Clarification of what is meant by ‘by-section normalized’ would be helpful.
L78-80: I’m confused by the wording here. What is being corrected for?
Figure 4: Hiatus frequency (bottom left panel) – does this refer to the proportion of the holes that include a given age and don’t have sediment of that age? In other words, are they true gaps or do they include cores that simply don’t reach the older ages? It would be helpful if this were made clearer in the caption.
Figure 5: Very hard to read, should be cleaned up.
Figure 7 caption: It’s unclear to me what ‘area unweighted’ means here. Is that corrected for CCD depth, red-clay-free area, basin area…?
Figure 7: The way the y axes run into each other makes them difficult to read, particularly as they’re log scale and don’t start at the same point. It would be easier for the reader to compare them if the axes used the same scale and limits.
L111-112: Figure 4 doesn’t show that as far as I can tell – but it would be an interesting thing to show.
L115: It seems from Figure 4 that hiatus frequency does decrease with age at both the start of the Cenozoic and approaching the modern, unless I’m missing something? It would also be useful to see a running median of hiatus length in the Figure 4 top left panel.
Figure 1: It’s almost impossible to see the black dotted line in the top panel, and there’s no mention of it in the caption. This seems particularly important to show because the top and bottom panels actually look quite similar at first glance.
L130: Is that supposed to be a citation at the end of the sentence?
Figure 6 is cited out of order. Please order the figures correctly.
References
B.S. Wade, et al., Evolution of deep-sea sediments across the Paleocene-Eoceneand Eocene-Oligocene boundaries. Earth-Science Rev. 211, 103403 (2020).
Caves Rugenstein, J. K., Ibarra, D. E. and von Blanckenburg, F. (2019) ‘Neogene cooling driven by land surface reactivity rather than increased weathering fluxes’, Nature. Springer US, 571(7763), pp. 99–102. doi: 10.1038/s41586-019-1332-y.
Katchinoff, J. A. R. et al. (2021) ‘Seawater Chemistry and Hydrothermal Controls on theCenozoic Osmium Cycle’, Geophysical Research Letters, 48(20), pp. 1–11. doi:10.1029/2021gl095558.
Pogge Von Strandmann, P. A. E., Kasemann, S. A. and Wimpenny, J. B. (2020) ‘Lithium and Lithium Isotopes in Earth’s Surface Cycles’, Elements, 16(4), pp. 253–258. doi:10.2138/GSELEMENTS.16.4.253.
Tréguer, P. et al. Reviews and syntheses: The biogeochemical cycle of silicon in the modern ocean. Biogeosciences 18, 1269–1289 (2021).
P. Tréguer, C. L. De La Rocha, The World Ocean Silica Cycle. Ann. Rev. Mar. Sci.5, 477–501 (2013).
Rahman, S., Aller, R. C. and Cochran, J. K. (2017) ‘The Missing Silica Sink: Revisiting the Marine Sedimentary Si Cycle Using Cosmogenic 32Si’, Global Biogeochemical Cycles, 31, pp. 1559–1578. doi: 10.1002/2017GB005746.
A.G. Dunlea, R. W. Murray, D. P. Santiago Ramos, J. A. Higgins, Cenozoic global cooling and increased seawater Mg/Ca via reduced reverse weathering. Nat. Commun. 8, 844 (2017).
Baldauf, J. G. and Barron, J. A. (1990) ‘Evolution of Biosiliceous Sedimentation Patterns — Eocene Through Quaternary: Paleoceanographic Response to Polar Cooling’, Geological History of the Polar Oceans: Arctic versus Antarctic, pp. 575–607. doi:10.1007/978-94-009-2029-3_32.
Citation: https://doi.org/10.5194/egusphere-2023-3087-RC2 -
AC2: 'Reply on RC2', Johan Renaudie, 14 Mar 2024
In addition to the main reply in which we discuss the criticisms common to all review (see Reply to RC1), we would like to answer some additional comments here:
The debate over the interpretation of those geochemical proxies is alluded to, but since the discrepancy between weathering and sedimentation rates is set up as the driving question this study is aiming to resolve, more detail on the current state of that debate seems called for, including more recent literature
Indeed we will add a small review of the recent literature on the subject.
Additionally, shallow marine settings are rejected as a possible alternative to pelagic settings for the burial of earlier Cenozoic weathering products, but the two citations listed—particularly Ridgewell and Hargreaves (2007)--do not by themselves provide an obvious explanation for why this is the case. Elaboration would be helpful here, and updated literature--e.g., for silica, Treguer et al. (2021), which substantially updates the Treguer and de la Rocha (2013) estimates of shallow water Si burial, and Rahman et al. (2017).
OK we will discuss this more.
(Although it would mean more work and is perhaps beyond the scope of the paper, it would be interesting to see a quantification of the magnitude of variation in shallow marine carbonate and silica burial that would be required to accommodate a 2-6x shift in pelagic sedimentation without a change in total weathering product sequestration.)
It would indeed be interesting but the issue here is the very low number of shallow marine carbonate sites for which we have data for, which is why we centered the debate on the deep-sea pelagic setting.
Similarly, several authors have suggested that biogenic pelagic sedimentation became more spatially concentrated over the Cenozoic (e.g., Dunlea et al 2017; Barron and Baldauf 1990). Some discussion of this hypothesis and how this study’s results fit in with it would be beneficial, given that one of the exciting novelties of this work is its spatial specificity.
Indeed we can add a small discussion about this.
Citation: https://doi.org/10.5194/egusphere-2023-3087-AC2
-
AC2: 'Reply on RC2', Johan Renaudie, 14 Mar 2024
-
RC3: 'Comment on egusphere-2023-3087', Jakub Witkowski, 20 Feb 2024
The manuscript by Johan Renaudie and David Lazarus is an insightful study on how objective and pragmatic factors contribute to our understanding of how sediment accumulation rate estimates are biased with increasing geological time. It addresses the fundamental questions that paleoceanographic studies usually aim to answer, and thus it is very likely to become a highly-cited, influential paper.
In my opinion, one important motivation for future readers of this article, will be pragmatic, i.e., what can I do to correct my data for the temporal bias in sedimentation rates? For this reason, I think it would be worthwile to include some kind of recommendations for future studies on how to handle the issues of compaction and drilling bias. Can a site or hole-specific correction factor be established from the data currently present in NSB, or included as supplementary data (see comment below)? Is it feasible/meaningful to compute such correction factor for individual sites? when global correction factor is applied, as in the figures in manuscript, but to a single site - will the results be meaningful? Speaking from my experience, I think expanding the discussion to include such issues would be highly beneficial for the readers.
I do have several questions relating to the online supplementary materials. The supplement is supposed to include six items: the text, supplementary figures, Python code, and three additional data items. I failed to find the latter three. Were they included with the submission?
I was also interested to actually run the code provided by the authors. The formatted text used in the supplement file, however, makes it difficult to copy the code to a code editor. As the authors apparently care for readers not familiar with programming (as revealed by the final part of the model description), this issue may be a nuisance to such readers. I also looked if the authors made the code available in an online repository (like the NSB code), but failed to find one.
Citation: https://doi.org/10.5194/egusphere-2023-3087-RC3 -
AC3: 'Reply on RC3', Johan Renaudie, 14 Mar 2024
In addition to the main reply in which we discuss the criticisms common to all review, we would like to answer some additional comments here:
In my opinion, one important motivation for future readers of this article, will be pragmatic, i.e., what can I do to correct my data for the temporal bias in sedimentation rates? For this reason, I think it would be worthwile to include some kind of recommendations for future studies on how to handle the issues of compaction and drilling bias. Can a site or hole-specific correction factor be established from the data currently present in NSB, or included as supplementary data (see comment below)? Is it feasible/meaningful to compute such correction factor for individual sites? when global correction factor is applied, as in the figures in manuscript, but to a single site - will the results be meaningful? Speaking from my experience, I think expanding the discussion to include such issues would be highly beneficial for the readers.
We will add a paragraph in the conclusions clarifying that the biases discussed here only affects compilations of SAR/LSR (and thus any interpretation of regional/global patterns) and not on single-site studies.
I do have several questions relating to the online supplementary materials. The supplement is supposed to include six items: the text, supplementary figures, Python code, and three additional data items. I failed to find the latter three. Were they included with the submission?
This is a good catch, thank you for alerting us of that: something must have happened while transferring this paper between CP and BG and the missing SOMs were not uploaded correctly. We apologize for this oversight.
I was also interested to actually run the code provided by the authors. The formatted text used in the supplement file, however, makes it difficult to copy the code to a code editor. As the authors apparently care for readers not familiar with programming (as revealed by the final part of the model description), this issue may be a nuisance to such readers. I also looked if the authors made the code available in an online repository (like the NSB code), but failed to find one.
This is a good point. We will create a repository on github/Zenodo for this code, as well as the R code used for the rest of the analysis.
Citation: https://doi.org/10.5194/egusphere-2023-3087-AC3
-
AC3: 'Reply on RC3', Johan Renaudie, 14 Mar 2024
Status: closed
-
RC1: 'Comment on egusphere-2023-3087', Adriana Dutkiewicz, 27 Jan 2024
This is an interesting and useful paper that describes a new approach to understanding the extent to which pelagic sediment accumulation rates in the global ocean have changed through the Cenozoic, and to what extent these observed changes are the result of drilling bias. The paper is novel and will be of interest to a wide readership using scientific ocean drilling data. The manuscript should be published after moderate revisions, that include clarification of the methodology, reconsideration of the CCD model used, better illustration of depositional area computations, and inclusion of new figures (locality map, methodology flow chart, seawater Sr curve). The methodology used in the analysis is very difficult to follow, and needs to be explained/visualised better. Some of the writing should also be tightened and figures improved, particularly in terms of font sizes (Figs 2 and 5, especially). See detailed comments below:
Abstract
The statement “The accumulation rate of pelagic marine biogenic sediments are thus a measure of weathering history” (line 3) is incorrect and should be reworded. Perhaps replace “thus a measure” with “an indication”, bearing in mind that not all continental weathering products end up on the seafloor (e.g., dissolution of carbonate) and not all weathering products come from continents (e.g., seafloor weathering).
In the abstract the authors state that “When accumulation area however is adjusted for changes in available deposition area, global weathering is shown to have nearly doubled at the Eocene-Oligocene boundary.” It would be useful to see maps of these time-dependent deposition areas, and a graph showing the time-dependence of these areas in individual ocean basins in the manuscript or supplement. The statement also assumes a straight forward correlation between deep-sea sedimentation and continental weathering, which is an oversimplification. The authors also need to explain why this would have happened at the EOT.
Other comments about the abstract:
L1: capitalize Earth (and elsewhere in the manuscript)
L2: pCO2 is written incorrectly
L3: change “are” to “is”
L6: change “deep sea drilling” to “deep-sea drilling” (do this elsewhere and for other compound adjectives in the manuscript)
Methodology
The hypothesis is that the older the sediments recovered are, the more likely they are to be represented only by low sedimentation rate sites, as penetration to older aged sediments is more likely when the local sedimentation rate is low. I have no doubt that this effect exists, but it is questionable whether this is the only effect driving the observation that sedimentation rates decrease with age.
The bias correction is based on this: "The inverse of the resulting modelled LSR vs geologic age curve is used as a correction factor on the NSB-based LSR and SAR compilation."
I don't understand why the correction for this effect is based on LSRs which are not corrected for compaction (unless I understand it incorrectly, and the modelled LSR with age includes a compaction correction). Compaction clearly plays an important role in decreasing apparent LSRs with increasing age, so the compaction effect needs to removed first. This automatically happens when calculating sediment accumulation rates (SARs), so isn't it logical to compute the bias-correction from SARs, as opposed to LSRs which themselves suffer from a bias from compaction, which in turn could bias a bias correction?
Next, the authors are following the assumption that the long-term LSR change with age is due to a drilling bias, and this is implemented by applying the inverse of this relationship as bias correction. The problem that I see is that if your assumption is incorrect, in the sense that there are long-term reductions in sedimentation rate that are not in fact due to drilling bias, then the "corrected" accumulation rates will be biased themselves.
I see some potential ways for further testing. One optional thing that could be done is to exclude all sites that were drilled in very high productivity areas, i.e., sites that were deliberately focused on Neogene very high sedimentation rate areas, including most sites in the equatorial Pacific, and sites from coastal upwelling regions (e.g., Benguela current sites offshore Africa etc). Likewise, regions of sediment focusing (contourite drifts) are also anomalous and would need to be excluded. What do the rest of the sites look like in terms of LSR with age? I realize that this would be a fair amount of work so perhaps the authors can assess if this is worthwhile doing.
Figure S2 shows that the accumulation rate data uncorrected for drilling bias (black curves) are quite different between the ocean basins. In this context it would be of interest to see the uncorrected LSR data subdivided into ocean basins as well, to get a sense of what the curves look like before the correction is applied (keeping in mind that the correction is a model that might perhaps be incorrect). For instance, in the Atlantic Ocean, the "raw" SARs are essentially constant at times before 15 Ma (long-term trend) while the Pacific Ocean shows a long-term decrease in SARs with increasing age, the Indian Ocean somewhere in between, and the Southern Ocean shows ups and downs entirely different from the other ocean basins. With so much inter-basin variability, I start wondering how meaningful a global correction for sedimentation rate with age is, and how valid the premise of the paper is (but again, this is difficult to judge without seeing the uncorrected data, which should be shown).
In Figs 7 and S1, are the linear sedimentation rates shown corrected for compaction? This is not clear from the captions. In the text, please provide an estimate of how much the sedimentation rate would have been affected by compaction.
Fig. 6 shows corrected linear sedimentation rates, and corrected SARs. The figure should show both uncorrected and corrected rates, as well as the correction function, to make this more transparent.
Lastly, the entire workflow is so complex that a casual reader of this paper currently would have no hope to understand what was actually done. It is quite possible that some of my comments above merely reflect that I didn't fully understand the methodology. This can be fixed by a flow chart that covers each step in the analysis, to make the paper clearer and more comprehensible to a general audience.
Deposition area computation
A simple global median curve based on ‘red clay’-free oceans (i. e., 74.2% of the Atlantic, 51% of the Pacific and 74.7% of the Indian Ocean) instead of their full area is shown in Fig. 7. But aren’t these percentages time dependent? This is problematic because the median curve is for the entire Cenozoic whereas the proportions are for the present-day. Not only has the area of the individual basins changed substantially since 66 Ma, but so has the pattern of sedimentation and lithology distributions. For example, the CCD has deepened substantially since 66 Ma, meaning that the global area available for carbonate sedimentation has also increased. As mentioned above, it would be helpful to have more information be supplied about what the deposition area through time actually looks like based on the authors’ choices of paleo-topography maps and CCD changes.
Have ‘red clay’ sections been removed from the LSR compilations in the 400+ drill sites? Following the logic above, failure to do so introduces a lithological bias, if not for the present-day then for some other time in the Cenozoic. We know that the CCD has deepened significantly since ~ 55 Ma (e.g., Palike et al., 2012) meaning that the accumulation rate of carbonate was a lot lower in the past thanks to dissolution, and that sedimentation was dominated by much more slowly accumulating pelagic clay. So older sections could simply correspond to more slowly accumulating sediments rather than a drilling bias. This point needs to be clarified and some of these ideas explored in more depth. A map showing the distribution of drill sites should be included in the manuscript.
Line 104: The Pälike et al. CCD for the equatorial Pacific is not representative of the global ocean. This is an upwelling region of high productivity, which is strikingly different (much deeper in terms of CCD) from the rest of the Pacific, and most of the Atlantic and Indian oceans. The authors should consider using a more appropriate model or models (e.g., global curve of Boss and Wilkinson, 1991), that also cover the entire Cenozoic.
Line 157: CCD trends in other ocean basins do not follow the trend for the Pacific CCD. This is evident in Campbell et al. (2018) and in Dutkiewicz and Muller (2022). The differences are quite striking. The statement should be corrected.
Other comments
Please refer to in-text figures in correct order. For example, Line 59 refers to Fig. 2 then Line 63 refers to Fig. 7. Fig.1 is first cited on Line 79 and there is no mention of Figs 3–6 prior to Line 63. This needs to be fixed as it’s greatly reducing the readability of an already complex manuscript.
Please improve the readability of some of the figures. In many cases the fonts are too small (Figs 2 and 5, in particular).
Line 159: “Global mean deep sea pelagic sediment accumulation rates have decreased slightly during the Cenozoic.” Relative to what? Please clarify.
Line 60: change “representative of pelagic sedimentation” to “representative of biogenic pelagic sedimentation”.
Line 133 onwards: state which figure this text is referring to.
Line 175: “Our results thus largely reconcile the discrepancy that until now has existed between 87Sr/86Sr estimates of Cenozoic weathering, and the deep sea pelagic accumulation rate record.” Please demonstrate this using your data and the seawater Sr curve. The curve will also provide an independent check of whether an increase in deep-sea accumulation rates corresponds to an increase in continental weathering.
References:
Boss, S.K., Wilkinson, B.H., 1991. Planktogenic/Eustatic Control on Cratonic/Oceanic Carbonate Accumulation. The Journal of Geology , 99, 497-513.
Campbell, S. M., Moucha, R., Derry, L. A., & Raymo, M. E. (2018). Effects of dynamic topography on the Cenozoic carbonate compensation depth. Geochemistry, Geophysics, Geosystems, 19, 1025–1034.
Citation: https://doi.org/10.5194/egusphere-2023-3087-RC1 -
AC1: 'Reply on RC1', Johan Renaudie, 14 Mar 2024
We thank all reviewers for their very helpful, constructive and detailed comments on our ms. There are many specific comments on ms contents which we will address as individual answers, and which we do not discuss here. We reply here to the main points of first two reviewer's comments, which are very similar.We are pleased that all reviewers see the work generally positively and, assuming successful revision, a useful addition to the literature. The first two reviewers also however note that the current version is very hard to follow, as the methods are not described at all clearly or in enough detail, and that the order of presentation is poor. We will indeed do a major rewrite of the introduction, materials and methods, improving organisation, and giving more detail on both the methods, and the reasoning that underlies our analyses. We briefly summarize the latter here to aid reviewers and the editor.Our ms is a bit unusual in that the analysis presented here is in fact composed of several sub-analyses carried out sequentially, with each building on the results of the prior sub-analysis. It is thus useful, before going into details in the ms, to not only summarize the methods, but also the results of each substudy in sequence, so that the reasoning and methods for each subsequent substudy is clear. We note that this presentation style does not strictly follow the conventional standard, used in the submitted ms, which gives all methods first, all results second, and discussion of results only last. Should the reviewers/editor find the following exposition useful, we ask permission to include this, suitably modified, as an additional section betweeen the Introduction, and Material and Methods parts of the revised ms. We would reduce some of the expository text currently elsewhere in the ms to prevent duplication.Substudy 1 - Impact of hiatus distribution on SAR[current ms Figures 4 and 5]In this sub-study we showed that hiatuses in deep sea pelagic sediment sections were relatively infrequent and mostly of short duration, nor do they increase in magnitude with increasing geologic age. Thus the apparent decline of accumulation with age could not be due to biassing by the cumulative effect of common, and with increasing scale of measurement, ever larger hiatuses on the calculation of apparent accumulation rates (aka the Sadler effect, as had been speculated by some prior authors). This result suggested that another cause, or set of causes must be responsible.Substudy 2 - By-section Analysis of declining accumulation rates with increasing geologic age[current ms Figure 1]We next tested a hypothesis of a uniform decline of average accumulation rate vs increasing age over the entire ocean. For this we examined the record of accumulation rate within a large number of individual sections. The accumulation rate for each segment of the age model for each section was expressed as the ratio of the mean rate for the whole section. The ratio value and midpoint age of each age model segment from each individual section were then compiled across all the sections in our study. This had the effect of removing any between-section differences in section average accumulation rate, leaving only the temporal trend in change in relative accumulation rate vs age. We were looking for the expected signal, at least in the composite global dataset, of major (ca 6X) declining average relative accumulation rate with increasing age, as had been reported by prior global summaries of deep sea sections in the literature. Instead, we discovered that, despite substantial short term variance of individual age model segments within the individual sections, the composite of relative change vs age for the set of all sections did not show any detectable trend towards lower average accumulation rate with increasing age.Substudy 3 - A Model of How Between Section Differences in Average Accumulation Rate Interact With Incomplete Recovery of Cenozoic Sections via Drilling to Create a Bias in Compiled Data[current ms Figures 3, 2 and 6 and 7]The near constant rates of relative accumulation found in our second sub-study, where we had removed between site differences in accumulation rate, suggested that the signal of declining accumulation rate vs age as reported in prior studies must in some way be connected with these differences between sections in accumulation rate, and/or in how the data for different sections was composited in these prior studies. In the course of both making age models for many sections over a period of many years, and during the above synthesis of relative accumulation vs age, it was apparent that a) most sections with high average accumulation rates ended in relatively young sediments, while sections reaching older sediments typically had low average accumulation rates; b) there was no obvious difference in the total depth (in meters below sea floor) reached by drilling between these different types of section; and c) only a few sections actually recovered the entire sedimentation record for a location (ie reached basement), or even recovered the entire Cenozoic time interval - most ended prior to base of the Cenozoic, constrained by limits imposed on total drilling depth from Leg priorities, bad weather, stuck drill bits etc. This led to the insights that: the age of recovered sediments in the global deep sea drilling data is determined in part by constraints from the drilling process, not just primary sedimentary record; and, the only ways (barring the occasional major hiatus) to reach older time intervals is to have chosen (deliberately or by chance) to drill at a location with a low average accumulation rate, and/or to drill an unusually deep section. Further, if one were to compile (i.e. bin) accumulation rate data across sections by time interval, younger time interval bins would have a mix of both high and low average rate sections, and thus higher bin-average accumulation rates, than older interval bins, where only sections with low average within section rates can normally be recovered.Our third sub-study explored the magnitude of this 'drilling bias' in a model, generating large numbers of model drill sections, and compiling (binning) the average accumulation rate across sections for 1 my time bins over the Cenozoic. Each model section was defined by just two main parameters: the average accumulation rate for the section, and the total depth drilled by the section. The model assumed for simplicity that any individual section had a constant rate of accumulation vs time. This was based on the results of our second sub-study on relative rates with time within single sections. There we had found, other than random variance (here treated as noise, and ignored), essentially no trend in the rate of accumulation vs time for any given section. The accumulation rate values for model sections were chosen by random sampling of the actual distribution of accumulation rates in 'recent' ie late Neogene sediment sections recovered by the deep sea drilling programs. Similarly, the total depth drilled was chosen by random sampling of the actual distribution of maximum drilling depth from the sections recovered by the deep sea drilling programs. All analyses were also adjusted to remove the effects of sediment compaction. Lastly, as hiatuses do exist in pelagic deep sea sections, our model included these as well, as occuring at random within a section, with a per time interval probability and time duration that match the actual data from the age model library of deep sea drilling. Our model thus simulated how, with constant actual accumulation rate vs time at any location, the variation of average accumulation rates between sections, and random variation in drilling depths between sections, affected the composite (binned) accumulation rate across sections. The magnitude of this drilling bias turned out to be quite large, and in fact accounted for the entire ca 6X change in apparent binned average accumulation rates across deep sea drill sections over the Cenozoic. When the effect of this bias was substracted from the 'raw' across section binned data, the resultant 'drilling bias corrected' accumulation rate over the Cenozoic showed only a slight (2X) decline - and thus in broad accordance with the near constant rates seen in our earlier (second) substudy of relative trends within sections vs time. We also looked at how different ocean basins might have different Cenozoic accumulation rate histories. These, when compiled together using a basin area adjustment matched as expected the global curve. There were however some differences in individual basin histories.Substudy 4[current ms Figure 8]In our fourth sub-analysis we applied our new, largely constant accumulation rates over the Cenozoic curve to the question of Cenozoic rates of global weathering. For this we examined only the dominant component of pelagic sediment output of weathering supplied dissolved nutrients - the calcium carbonate sequestered in pelagic sediments by marine plankton. This was also done as relatively little information is available on the Cenozoic distribution of the other main component of weathering - biogenic silica in sediments; and in older intervals, how it has been affected by diagenesis to chert. To convert the accumulation rates for sections for each geologic time interval from our earlier substudy 3 into global accumulation rates per time interval of carbonate material, we calculated the area of the ocean accumulating carbonate for each time interval. The area of the oceans above a given depth (the CCD) was calculated from paleotopographic maps for each time interval of the Cenozoic. To further simplify the analysis, we considered the global ocean accumulation rate of pelagic carbonate to be defined by the area of ocean accumulating carbonate above just the Pacific Ocean basin CCD. This was done as the Pacific basin is by far the largest basin, and the Pacific CCD history for the Cenozoic is the best defined by published studies.Another issue brought by the first two reviewers has to do with changes in CCD in two of the analyses:- In the computation of the SAR corrected by ocean basins area (in which we excluded the [modern] area occupied by red clays)- In the final estimation of total sediment flux in the Cenozoic, where we used the Equatorial Pacific history of CCD reconstructed by Pälike et al. 2006.Computing the total area above CCD as we did in the second case is relatively easy but splitting it into basins (with necessitates reconstructing the basin borders through time) is computationally more extensive. Additionally, not all basins have a complete Cenozoic published CCD history.What we can do for the first case is using the basin area without correction for red clays, as well as correcting for red clays using the global CCD estimate (such as Boss and Wilkinson, 1991 as proposed by the first reviewer), as a minimum-maximum estimates of that curve.We are also happy to try and calculate the total sediment flux using individual CCD histories for each ocean basin with the best published estimates, if possible, and present them alongside one computed using a global CCD history.Another recurring comment is the over-emphasis on the link between global weathering and pelagic sedimentation: we will make an effort to reword this.Lastly, we also apologise for not having checked the ms carefully enough prior to submission to insure such basic things as having the figure numbering in correct order. These, annoying if only cosmetic errors will of course also be fixed in our revision.-------------------------------------
In addition to the reply above addressing both RC1 and RC2 , we would like to answer some additional important comments from RC1 here:
In the abstract the authors state that “When accumulation area however is adjusted for changes in available deposition area, global weathering is shown to have nearly doubled at the Eocene-Oligocene boundary.” It would be useful to see maps of these time-dependent deposition areas, and a graph showing the time-dependence of these areas in individual ocean basins in the manuscript or supplement.
Indeed this is a good idea, we will produce such maps and add them to the SOM.
I don't understand why the correction for this effect is based on LSRs which are not corrected for compaction (unless I understand it incorrectly, and the modelled LSR with age includes a compaction correction).
[and further comments]SAR is computing by multiplying LSR with the actual, measured density of the cores, thus already correcting for compaction. Using the compaction-corrected model output to correct the compiled SAR would thus over-correct for compaction.
I see some potential ways for further testing. One optional thing that could be done is to exclude all sites that were drilled in very high productivity areas, i.e., sites that were deliberately focused on Neogene very high sedimentation rate areas, including most sites in the equatorial Pacific, and sites from coastal upwelling regions (e.g., Benguela current sites offshore Africa etc). Likewise, regions of sediment focusing (contourite drifts) are also anomalous and would need to be excluded. What do the rest of the sites look like in terms of LSR with age? I realize that this would be a fair amount of work so perhaps the authors can assess if this is worthwhile doing.
The first part might be doable indeed in a reasonable time-frame. We will see if we can do that and add it at the very least in the SOM.
Fig. 6 shows corrected linear sedimentation rates, and corrected SARs. The figure should show both uncorrected and corrected rates, as well as the correction function, to make this more transparent.
Indeed. We will modify the figure accordingly.
Line 175: “Our results thus largely reconcile the discrepancy that until now has existed between 87Sr/86Sr estimates of Cenozoic weathering, and the deep sea pelagic accumulation rate record.” Please demonstrate this using your data and the seawater Sr curve. The curve will also provide an independent check of whether an increase in deep-sea accumulation rates corresponds to an increase in continental weathering.
We could indeed add such a curve to Figure 7.
Citation: https://doi.org/10.5194/egusphere-2023-3087-AC1
-
AC1: 'Reply on RC1', Johan Renaudie, 14 Mar 2024
-
RC2: 'Comment on egusphere-2023-3087', Sophie Westacott, 14 Feb 2024
Overview
This work is a valuable contribution to the ongoing discussion of pelagic sedimentation rates through the Cenozoic. It considers a possible bias created by drilling depth limitations resulting in a lower likelihood of reaching older sediments at sites with higher sedimentation rates, and thus most older-sediment sites being from areas of low sedimentation. It does this using randomized sampling of age models from the Neptune (NBS) database, and goes on to estimate global sediment accumulation rate through the Cenozoic by factoring in CCD depth, red-clay-free area, ocean basin, intra- vs inter-hole variability, age model quality, hiatus length and frequency, and compaction. It will be useful for anyone working on whole-Cenozoic comparisons or trends in pelagic sediments, including microfossils.
However, the manuscript requires considerable revision before it will be readily digested by readers, as it is currently very challenging to follow. The explanation of the central method—using a null model to test for the predicted drilling bias—is particularly confusing, and as detailed below, the logic behind some of the interpretations of the results also needs further elaboration. The manuscript is also rather unpolished: the figures are out of order, the axes on most of them are too small, a few of them (e.g. Figs. 1, 2 and 5) are rough drafts rather than finished products, the text has grammatical errors, and one or two references to figures don’t make sense—none of which helps with following what was done. The content of the paper seems worthy of a tighter, neater form that will allow others to fully engage with it.
Broad Comments
-The inter- vs intra-hole variability is a key finding, and the argument for a drilling bias hangs on it: if higher sedimentation rates nearer the modern do not relate to higher sedimentation rates earlier in the Cenozoic within each hole, then the drilling bias explanation doesn’t hold. I think the overall thesis of the paper would thus be made stronger by exploring this finding more, and by clarifying the (currently very confusing) part on ‘by-section normalization’.
-The use of red-clay-free area is an interesting new approach, but I have the same question as the first reviewer—what about the fact that the amount of red-clay-free has shifted over the Cenozoic? The maps in Wade et al. (2020), for example, show a much smaller red-clay-free area of seafloor in the early Cenozoic than later. Why is the basin median and mean calculated from the modern red-clay-free area, but then the CCD used in compiling the SAR curve? Although it would include a large unknown, estimating the red-clay-free area over time and using that estimate in the SAR calculations would likely be more accurate than assuming the same size of red-clay-area over time. Furthermore, if the same amount of carbonate and opal that is buried today were to be buried in the much smaller red clay free area of the early Cenozoic, apparent sedimentation rates would need to be substantially higher in the earlier Cenozoic than in the modern, no? Is that reflected in the raw LSR and SAR data? It isn’t clear to me in Figure 1 that it is, but it’s hard to tell from the way the data is depicted. Using the CCD combined with topographic data makes sense, and seems like a worthwhile approach, but I do wonder about opal sedimentation rates, particularly in the Southern Ocean. While these may be more minor in terms of global sedimentation rates, could they be useful in understanding at least the Southern Ocean history? A more detailed discussion of the topic would be helpful.
-The manuscript emphasizes a very direct link between weathering and pelagic sedimentation, a claim that needs more support than it’s given. Lines 39-41 say “If the deep sea pelagic sedimentary record is not (strongly) biased by hiatus-driven measurement scale artifacts, then the major increase in apparent rate over Cenozoic means either that geochemical proxies are somehow wrong and rate has changed exponentially, or that some other factors are biasing the Cenozoic deep sea record.” The debate over the interpretation of those geochemical proxies is alluded to, but since the discrepancy between weathering and sedimentation rates is set up as the driving question this study is aiming to resolve, more detail on the current state of that debate seems called for, including more recent literature (e.g., Caves Rugenstein, Ibarra and von Blanckenburg, 2019; Katchinoff et al. 2021; Pogge Von Strandmann, Kasemann and Wimpenny, 2020). Line 175-176 makes the claim that “Our results thus largely reconcile the discrepancy that until now has existed between 87Sr/86Sr estimates of Cenozoic weathering, and the deep sea pelagic accumulation rate record.” This certainly needs a discussion of at least the Sr record along with a visual comparison of the timeline of the strontium data with the updated SAR curve. Is the EOT SAR step-change shown in Fig. 8 reflected in the Sr data? What about the non-directional but still substantial SAR variation in the Neogene?
-Additionally, shallow marine settings are rejected as a possible alternative to pelagic settings for the burial of earlier Cenozoic weathering products, but the two citations listed—particularly Ridgewell and Hargreaves (2007)--do not by themselves provide an obvious explanation for why this is the case. Elaboration would be helpful here, and updated literature--e.g., for silica, Treguer et al. (2021), which substantially updates the Treguer and de la Rocha (2013) estimates of shallow water Si burial, and Rahman et al. (2017). (Although it would mean more work and is perhaps beyond the scope of the paper, it would be interesting to see a quantification of the magnitude of variation in shallow marine carbonate and silica burial that would be required to accommodate a 2-6x shift in pelagic sedimentation without a change in total weathering product sequestration.)
-Similarly, several authors have suggested that biogenic pelagic sedimentation became more spatially concentrated over the Cenozoic (e.g., Dunlea et al 2017; Barron and Baldauf 1990). Some discussion of this hypothesis and how this study’s results fit in with it would be beneficial, given that one of the exciting novelties of this work is its spatial specificity.
Broadly speaking, to extend the conclusions of these analyses beyond pelagic sedimentation (interesting in its own right) to summed global weathering and make the claim that this solves a lynchpin of the Cenozoic weathering debate, there should be a more in-depth discussion of the abundant empirical (e.g., stable and cosmogenic isotopic, geophysical, terrestrial sedimentary, paleo-CO2) and model-based work that has been done on the subject and its conflicting interpretations, as well as alternative hypotheses.
Specific Comments
As mentioned above, there are numerous minor errors and style issues throughout the text that, although I have not listed them here, should be cleaned up before publication. The abstract and the last paragraph of the introduction in particular need some serious textual work.
L42: the way ‘decline’ is used in the last sentence of the introduction, while correct, is confusing in the context of ‘increase’ in the paragraph before.
L55: Is it a running median with an overlapping 10 ky window, or a median (and mean) taken at 10 ky intervals starting centered at 5 ky? The former would make more sense for the purpose, I would think. It would also be helpful to understand why 10 ky was selected as the window size.
L60: Given the wide range in sampling coverage between basins (12-220 sites), I wonder what a sort of rarefaction/bootstrap type analysis might show? In other words, what heterogeneity do you find if you subsample the Atlantic in groups of 12 sites with replacement? Just curious…
L63: Figure 7 is out of order, and Figure 1 isn’t called out in the text until much later... Please order the figures correctly.
L58: Figure 2 is cited here, but it doesn’t do much to show how either LSR or SAR are calculated. It would be helpful to have a different figure showing the process and/or more clarifying text.
Figure 3: Frequency would, I think, be more useful to the reader than density in these histograms. Also, panel labels (letters or numbers) here and throughout would make the figures easier to relate to the text and caption, rather than ‘top left’, ‘bottom right’, etc.
L77 cites a density plot of SAR vs time span in the Supplementary Material, but I can’t see one there. Does it mean Figure 5?
L77: “We then calculate by-section normalized change in sedimentation rates vs geologic age in intervals with sedimentation to see if older time intervals, corrected for between site differences in local rates, show a decline in relative rate with increasing geologic age (see Figure 1).” I’m not sure I follow what was done here. Clarification of what is meant by ‘by-section normalized’ would be helpful.
L78-80: I’m confused by the wording here. What is being corrected for?
Figure 4: Hiatus frequency (bottom left panel) – does this refer to the proportion of the holes that include a given age and don’t have sediment of that age? In other words, are they true gaps or do they include cores that simply don’t reach the older ages? It would be helpful if this were made clearer in the caption.
Figure 5: Very hard to read, should be cleaned up.
Figure 7 caption: It’s unclear to me what ‘area unweighted’ means here. Is that corrected for CCD depth, red-clay-free area, basin area…?
Figure 7: The way the y axes run into each other makes them difficult to read, particularly as they’re log scale and don’t start at the same point. It would be easier for the reader to compare them if the axes used the same scale and limits.
L111-112: Figure 4 doesn’t show that as far as I can tell – but it would be an interesting thing to show.
L115: It seems from Figure 4 that hiatus frequency does decrease with age at both the start of the Cenozoic and approaching the modern, unless I’m missing something? It would also be useful to see a running median of hiatus length in the Figure 4 top left panel.
Figure 1: It’s almost impossible to see the black dotted line in the top panel, and there’s no mention of it in the caption. This seems particularly important to show because the top and bottom panels actually look quite similar at first glance.
L130: Is that supposed to be a citation at the end of the sentence?
Figure 6 is cited out of order. Please order the figures correctly.
References
B.S. Wade, et al., Evolution of deep-sea sediments across the Paleocene-Eoceneand Eocene-Oligocene boundaries. Earth-Science Rev. 211, 103403 (2020).
Caves Rugenstein, J. K., Ibarra, D. E. and von Blanckenburg, F. (2019) ‘Neogene cooling driven by land surface reactivity rather than increased weathering fluxes’, Nature. Springer US, 571(7763), pp. 99–102. doi: 10.1038/s41586-019-1332-y.
Katchinoff, J. A. R. et al. (2021) ‘Seawater Chemistry and Hydrothermal Controls on theCenozoic Osmium Cycle’, Geophysical Research Letters, 48(20), pp. 1–11. doi:10.1029/2021gl095558.
Pogge Von Strandmann, P. A. E., Kasemann, S. A. and Wimpenny, J. B. (2020) ‘Lithium and Lithium Isotopes in Earth’s Surface Cycles’, Elements, 16(4), pp. 253–258. doi:10.2138/GSELEMENTS.16.4.253.
Tréguer, P. et al. Reviews and syntheses: The biogeochemical cycle of silicon in the modern ocean. Biogeosciences 18, 1269–1289 (2021).
P. Tréguer, C. L. De La Rocha, The World Ocean Silica Cycle. Ann. Rev. Mar. Sci.5, 477–501 (2013).
Rahman, S., Aller, R. C. and Cochran, J. K. (2017) ‘The Missing Silica Sink: Revisiting the Marine Sedimentary Si Cycle Using Cosmogenic 32Si’, Global Biogeochemical Cycles, 31, pp. 1559–1578. doi: 10.1002/2017GB005746.
A.G. Dunlea, R. W. Murray, D. P. Santiago Ramos, J. A. Higgins, Cenozoic global cooling and increased seawater Mg/Ca via reduced reverse weathering. Nat. Commun. 8, 844 (2017).
Baldauf, J. G. and Barron, J. A. (1990) ‘Evolution of Biosiliceous Sedimentation Patterns — Eocene Through Quaternary: Paleoceanographic Response to Polar Cooling’, Geological History of the Polar Oceans: Arctic versus Antarctic, pp. 575–607. doi:10.1007/978-94-009-2029-3_32.
Citation: https://doi.org/10.5194/egusphere-2023-3087-RC2 -
AC2: 'Reply on RC2', Johan Renaudie, 14 Mar 2024
In addition to the main reply in which we discuss the criticisms common to all review (see Reply to RC1), we would like to answer some additional comments here:
The debate over the interpretation of those geochemical proxies is alluded to, but since the discrepancy between weathering and sedimentation rates is set up as the driving question this study is aiming to resolve, more detail on the current state of that debate seems called for, including more recent literature
Indeed we will add a small review of the recent literature on the subject.
Additionally, shallow marine settings are rejected as a possible alternative to pelagic settings for the burial of earlier Cenozoic weathering products, but the two citations listed—particularly Ridgewell and Hargreaves (2007)--do not by themselves provide an obvious explanation for why this is the case. Elaboration would be helpful here, and updated literature--e.g., for silica, Treguer et al. (2021), which substantially updates the Treguer and de la Rocha (2013) estimates of shallow water Si burial, and Rahman et al. (2017).
OK we will discuss this more.
(Although it would mean more work and is perhaps beyond the scope of the paper, it would be interesting to see a quantification of the magnitude of variation in shallow marine carbonate and silica burial that would be required to accommodate a 2-6x shift in pelagic sedimentation without a change in total weathering product sequestration.)
It would indeed be interesting but the issue here is the very low number of shallow marine carbonate sites for which we have data for, which is why we centered the debate on the deep-sea pelagic setting.
Similarly, several authors have suggested that biogenic pelagic sedimentation became more spatially concentrated over the Cenozoic (e.g., Dunlea et al 2017; Barron and Baldauf 1990). Some discussion of this hypothesis and how this study’s results fit in with it would be beneficial, given that one of the exciting novelties of this work is its spatial specificity.
Indeed we can add a small discussion about this.
Citation: https://doi.org/10.5194/egusphere-2023-3087-AC2
-
AC2: 'Reply on RC2', Johan Renaudie, 14 Mar 2024
-
RC3: 'Comment on egusphere-2023-3087', Jakub Witkowski, 20 Feb 2024
The manuscript by Johan Renaudie and David Lazarus is an insightful study on how objective and pragmatic factors contribute to our understanding of how sediment accumulation rate estimates are biased with increasing geological time. It addresses the fundamental questions that paleoceanographic studies usually aim to answer, and thus it is very likely to become a highly-cited, influential paper.
In my opinion, one important motivation for future readers of this article, will be pragmatic, i.e., what can I do to correct my data for the temporal bias in sedimentation rates? For this reason, I think it would be worthwile to include some kind of recommendations for future studies on how to handle the issues of compaction and drilling bias. Can a site or hole-specific correction factor be established from the data currently present in NSB, or included as supplementary data (see comment below)? Is it feasible/meaningful to compute such correction factor for individual sites? when global correction factor is applied, as in the figures in manuscript, but to a single site - will the results be meaningful? Speaking from my experience, I think expanding the discussion to include such issues would be highly beneficial for the readers.
I do have several questions relating to the online supplementary materials. The supplement is supposed to include six items: the text, supplementary figures, Python code, and three additional data items. I failed to find the latter three. Were they included with the submission?
I was also interested to actually run the code provided by the authors. The formatted text used in the supplement file, however, makes it difficult to copy the code to a code editor. As the authors apparently care for readers not familiar with programming (as revealed by the final part of the model description), this issue may be a nuisance to such readers. I also looked if the authors made the code available in an online repository (like the NSB code), but failed to find one.
Citation: https://doi.org/10.5194/egusphere-2023-3087-RC3 -
AC3: 'Reply on RC3', Johan Renaudie, 14 Mar 2024
In addition to the main reply in which we discuss the criticisms common to all review, we would like to answer some additional comments here:
In my opinion, one important motivation for future readers of this article, will be pragmatic, i.e., what can I do to correct my data for the temporal bias in sedimentation rates? For this reason, I think it would be worthwile to include some kind of recommendations for future studies on how to handle the issues of compaction and drilling bias. Can a site or hole-specific correction factor be established from the data currently present in NSB, or included as supplementary data (see comment below)? Is it feasible/meaningful to compute such correction factor for individual sites? when global correction factor is applied, as in the figures in manuscript, but to a single site - will the results be meaningful? Speaking from my experience, I think expanding the discussion to include such issues would be highly beneficial for the readers.
We will add a paragraph in the conclusions clarifying that the biases discussed here only affects compilations of SAR/LSR (and thus any interpretation of regional/global patterns) and not on single-site studies.
I do have several questions relating to the online supplementary materials. The supplement is supposed to include six items: the text, supplementary figures, Python code, and three additional data items. I failed to find the latter three. Were they included with the submission?
This is a good catch, thank you for alerting us of that: something must have happened while transferring this paper between CP and BG and the missing SOMs were not uploaded correctly. We apologize for this oversight.
I was also interested to actually run the code provided by the authors. The formatted text used in the supplement file, however, makes it difficult to copy the code to a code editor. As the authors apparently care for readers not familiar with programming (as revealed by the final part of the model description), this issue may be a nuisance to such readers. I also looked if the authors made the code available in an online repository (like the NSB code), but failed to find one.
This is a good point. We will create a repository on github/Zenodo for this code, as well as the R code used for the rest of the analysis.
Citation: https://doi.org/10.5194/egusphere-2023-3087-AC3
-
AC3: 'Reply on RC3', Johan Renaudie, 14 Mar 2024
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 513 | 179 | 42 | 734 | 42 | 49 | 32 |
- HTML: 513
- PDF: 179
- XML: 42
- Total: 734
- Supplement: 42
- BibTeX: 49
- EndNote: 32
Viewed (geographical distribution)
| Country | # | Views | % |
|---|---|---|---|
| United States of America | 1 | 312 | 41 |
| Germany | 2 | 115 | 15 |
| China | 3 | 48 | 6 |
| United Kingdom | 4 | 36 | 4 |
| Netherlands | 5 | 25 | 3 |
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
- 312