the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 19 Feb 2025
High-Resolution Paleo-Storm Reconstruction from Eastern Canada Aligns with Late-Holocene Northwestern Atlantic Hurricane Records
Abstract. Atlantic Canada experiences frequent major storms, particularly tropical cyclones transitioning into post-tropical storms. Events such as Hurricane Fiona (2022), Dorian (2019), and Juan (2003) have caused significant damage, loss of life, and coastal erosion, exacerbated by sea level rise and warming waters. Despite this, centennial- to millennial-scale storm records in the region remain scarce. Existing studies in North America focus primarily on marine and coastal overwash records, with limited use of aeolian mineral inputs in ombrotrophic peatlands as storm proxies. Here, we address these gaps by analysing grain-size and geochemical data from two peatlands in Quebec, Canada’s Magdalen Islands.
Our two peat records reveal consistent storm signals over the past 4000 years, with three key periods of heightened activity: 800–550 BCE, 600–800 BCE, and 1300–1700 CE. These signals align with marine and overwash records spanning the past 2000 years across eastern Canada, the US, and the Bahamas, indicating low storm activity during the Medieval Climate Anomaly, followed by increased activity during the Little Ice Age. Our findings suggest that storm records in these regions are influenced by local climatic factors. Negative phases of the Atlantic Multidecadal Variability, which typically suppress hurricane activity in the North Atlantic, are associated with conducive hurricane formation and intensification north of the Bahamas. Additionally, the position of the Bermuda High seems to play a more significant role in directing storm tracks during different climatic phases. Our findings highlight the potential antiphase relationship in storm activity between regions north of the Bahamas and those in the Gulf of Mexico, suggesting broader climatic mechanisms that warrant further investigations.
Despite the similarities between our two sites, discrepancies in geochemistry and mineralogical profiles highlight the importance of site-specific conditions in interpreting the storm record from peatlands, namely the distance of the sites to the coast and source of aeolian sediment, as well as peatland size. Challenges also remain in calibrating peat-based proxies with historical storm records, as identifying specific events from the past 150 years remains difficult.
Competing interests: At least one of the (co-)authors is a member of the editorial board of Climate of the Past. Dr. Pierre Francus is a member of the editorial board of Climate of the Past.
Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this preprint. The responsibility to include appropriate place names lies with the authors.- Preprint
(8600 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2025-400', Anonymous Referee #1, 10 Mar 2025
General Comments
The manuscript by Lachance and coauthors presents two new records of past storm activity from northeastern Canada. Methods of Aeolian Sediment Influx (using a sieving method) and ITRAX XRF analysis are used to measure the past aeolian sand deposits to the two bogs, spanning the late Holocene. The manuscript is very well written, with clear language, well presented figures and appropriate numbers of citations in most parts. The statistical analysis of the data appears to me (a non expert) to be appropriate, although as detailed below I have questions about the use of unnormalized itrax count data in parts of the paper and the chosen method of detecting storm events. The discussion is particularly interesting and well written, and includes comparisons between the storm records and individual recent storms, as well as a very good comparison of records from a latitudinal transect. The conclusion includes interesting take home messages, but could be strengthened by detailing more of the results and findings.
Major Corrections
1) Unnormalized ITRAX count data: the authors do a good job with the statistical analysis of the ITRAX data, converting it to Centered Log Ratios before doing the statistical analysis, because ITRAX data is compositional data. Therefore there is an acknowledgement that changes in one element can affect the counts of another element, and hence the need to normalize it. It was therefore surprising that following the statistical analysis the Ti records that are shown and on which the storm records are based were raw counts per second (kcps) results, rather than the CLR results. Previous studies (e.g. Löwemark et al. 2011; Weltje and Tjallingii, 2008; Croudace et al. 2019) have highlighted that ITRAX results should be normalized because otherwise variations in the amount of undetected organic matter, water content, surface roughness and counts of other elements can all influence the results. Appendix D shows photos of the cores, which have a very rough surface – was the surface smoothed at all prior to ITRAX scan as this would surely influence the ITRAX results greatly? This makes it important that the results are normalized as the variability in the surface would have likely influenced the measured Ti kcps. The CLR results (or natural log ratios with a carefully selected denominator) should provide the basis for the storm records.
2) Potential misinterpretation and overreliance on the Ti results
- Why was Ti selected as the element most indicative of mineral content over other elements? I think more justification of this decision is needed. The Principle Component Analysis in figure 4 shows that for TLM Ti is not strongly correlated with PC1, when the ASI results and other elements are. Figure 7 which shows the ASI overlaying the Ti record does support the use of Ti and perhaps this comparison should be done earlier in the paper (and using the full records), as currently much of the ITRAX results are discounted with little explanation in the results section.
- The event frequency was calculated using the number of times that Ti exceeded a threshold. The data was detrended using a 10/30 year moving average window and peaks above a threshold were marked as storm events. The authors state ‘To avoid overestimating the number of events due to consecutive Ti measurements associated with a single Ti peak, consecutive measurements were grouped together, and only the maximum value within each group was retained’. I wonder however whether some of the identified storm events are represented by single data points or small groups, which could be outliers or erroneous data points rather than sand layers. As noted above, in data that has not been normalized other factors could cause an increase or decrease in the kcps that is unrelated to mineral content in the peat. The authors should check the Mean Squared Error and total counts per second measurements through the records to see if any of the peaks in Ti are associated with anomalies and should be excluded as artefacts (see Lowemark et al. 2019).
- A point related to this one about the detection of storm events, is that the moving window used may be too narrow and single large events may be identified as several small events. If you look at Figure 5b and the record of TAC the red line of the moving average threshold goes up over some of the increases in Ti as a result of the concentrations increasing over a >10 year period (depending on the accumulation rate 10 years could be just 1-2cm within the peat). My concern is that if there was a large deposit of sand on the bog by a single large magnitude storm, then sand may settle on the bog at different depths due to the uneven surface, with some falling into pools and sinking down, some at the roots, others on the top of plants etc.. Further mixing could occur with root bioturbation potentially. You acknowledge in the discussion ‘downward movement of sediments in the spongey peat matrix may have dated sediments associated to Hurricane Dorien to an older interval’ to explain the timing of an event, so a similar mechanism could spread the sand from a single large event over a few cm’s within the consolidated peat. Therefore, this could lead to overestimation of the number of events if these larger increases in sand are removed by detrending and then small variations in Ti on these peaks are counted as events. Either the method should be adjusted to address this or these limitations should be discussed, with perhaps more cautious statements about the frequency of events.
3) Exclusion of part of the TAC record
You should be consistent about whether or not the period between ~600 BCE and ~1000 CE in TAC is useful as a storm record, or not given you interpret it as being not ombrotrophic. In Figure 5 you shaded out the section but in Figure 8 it is included. In the discussion you also observe that the Ti and storm events records of the two cores are similar over this section. While you have concluded that TAC was not ombrotrophic at this time, that is not to say that the peaks in Ti and sand shown by the TAC record were not deposited by the same storms that caused the deposits in TLM. Perhaps better not to discount this section of the TLM record but to rather use it with the caveat that deposits may come from other sources in addition to storms.
Minor and Moderate Corrections
Title – I would put ‘storm reconstruction’ as plural, and not sure Late Holocene should be hyphenated
Line 36 – ‘in Atlantic Canada ever’, I would suggest that this is changed to ‘on record’ or ‘to date’ rather than ‘ever’
Line 46 – ‘Answering these questions is aided by a long-term perspective’, the wording of this is a little unusual, perhaps ‘A long term perspective can help to answer these questions,’
Line 65 – the sentence ending ‘cyclone strikes’ needs citations. Currently these are put after the following sentence, but they should go after the first sentence, or the two sentences could be merged.
Line 68 – sentence ending ‘centennial data’. As with the above point, the citations should go after the first sentence rather than the following one.
Line 82 – NAO needs writing in full the first time it is mentioned
Line 97 – ‘composed of quartz and being the main sand source’, this is phased in an unusual way. I would suggest ‘ …quartz, which is the main sand source..’
Line 100 – ‘steady winds from all directions’. Remove this, as in a following sentence you discuss wind directions being seasonal and show this with the results in the appendix.
Line 111 – it depends on the journal guidelines perhaps, but shouldn't the ages be 'ka before present' and ka BP rather than just ka?
Figure 1d and 1e - TLM is in the center of the island so could have sand from a few directions. Perhaps also include a couple more arrows with distances to beaches. Also the distance of TAC to the cliffs may not be as relevant to highlight as the distance to sand sources.
Line 182 – I don’t think Bjorck has an ‘l’ at the end
Line 202 – in equation 2, were the denominator measurements all the elements measured, or just those terrigenous mineral elements listed above.
Figure 3 – some of the y axis labels are too close together and are overlapping
Line 345 – The statement that the storm records are based solely on the Ti results needs more justification
Figure 5 - 1) Could the colored dots on the storm events be made smaller? It is hard to see the results. 2) While the results on this figure are convincing, as they show similar patterns, I would like to also see the ASI results presented, as these results are surely also as relevant as Ti to showing past storminess. 3) The AMV results should be included on 8, rather than 5, as this is where they are discussed.
Line 402 – this sentence says that sand layers in TAC at 600BCE and 810 CE were not visible in TLM. But there was an increase in identified storm events coinciding with these times in TLM, so could it be that there was an enhanced sand deposition related to storms at these times, but just more sand reached TAC?
Line 403 – The sentence says that there was no abrupt contacts between the sand and overlying and underlying peat, supporting a gradual accumulation of sand over time. I am not sure about this interpretation as peat bog environments don’t seem to often have sharp boundaries in the same way as lakes. The surface of bogs are uneven and I could see that when sand blows in a single event over the surface it could fall into pools, sit on top of plants but also land or wash down to the base of the plant and so be incorporated within different depths giving a gradual boundary even for a single large event.
Line 406 – for the sentence about the likely explanation is the in-situ concentration of mineral matter I think you should say that this interpretation is based on the hiatus shown by the age model, and any other evidence you have. I was confused at first why this would be a more likely explanation than the proximity of the sand sources.
Line 413 - I would not include the Netherlands example, as the Netherlands are so far away and not relevant to the regional climate. Lots of other sites around the world probably show no change at this time.
Line 422 – the sentence suggests the proximity of TAC to the cliffs is the reason for the larger particles at TAC. But are the cliffs the sediment source during storms? Are sand sized particles being eroded from the cliffs and transported in land during storms? It looks like TAC is slightly closer to the beach to the west than TLM is, but the satellite photos seem to show the environment around TAC is dryer, so would exposed soil potentially be providing a source of minerals to TAC?
Line 426 – in this sentence again the assumption is that the windblown sands came from the cliffs and/or beaches. I expect the soil on the island would have a similar elemental composition to the bedrock, so could wind blown soil also be a contributor and spatial variations in vegetation a factor?
Line 431 – remove typo (sand written twice)
Line 432 - ‘Ti emerged as the common aeolian sand indicator in both TLM and TAC.’ I think this needs more justification earlier in the paper.
Line 446 – remove the link to Fig 6c because it doesn’t show evidence that supports the use of ASI and Ti
Line 458 – ‘winter westerlies storms’, maybe 'storms associated with the winter westerlies' would sound better here
Line 461 – recent calendar years need CE throughout this section
Line 467 – ‘despite it coming ashore’, its not clear what this means - did waves come ashore?
Line 526 – ‘active hurricane period’ perhaps
Section 5.3 – this is a really nice comparison of the different records along the transect
Figure 8 – 1) you could just have CE on the x axis label. 2) in the caption you write that there are similarities between the Ti record from TAC and the storm activity in TLM. The fact that you have two records close to each other showing similar patterns increases confidence in the records and I think this point should be made more prominent in the text. It also gives some support that the non-ombrotrophic section of TAC is still capturing a storm signal.
Line 560 - a citation is needed here
Conclusion – this part highlights some interesting challenges and lessons but doesn’t go into much detail about the results of the paper, in the same way as the abstract does for example. I would like to see the results summarized in the conclusions either at the start or incorporated in. An example is the paragraph at line 631, where there could be another sentence covering the point you make about regional versus tropical Atlantic SST's being important for hurricane impacts in different regions.
Appendix A - 30-40, 40-50 with hyphens might make this clearer
Citation: https://doi.org/10.5194/egusphere-2025-400-RC1 -
CC1: 'Reply on RC1', Antoine Lachance, 12 Apr 2025
We sincerely thank Reviewer 1 for their thoughtful and constructive reviews and comments. Below, we provide a detailed response for each of the main concerns regarding (1) data processing and normalization, (2) the use and interpretation of Ti as a proxy for storm activity, and (3) the interpretation of the TAC record. We also address several minor comments. Suggested corrections and modifications are being incorporated into the revised manuscript. We have taken steps to clarify our decision-making process, improve the presentation of the results and discussion, and highlight key findings more effectively. We greatly appreciate these suggestions, which have helped improve the overall quality of the manuscript.
Major corrections
- XRF Data processing and normalization
Reviewer comment:
The authors do a good job with the statistical analysis of the ITRAX data, converting it to Centered Log Ratios before doing the statistical analysis, because ITRAX data is compositional data. Therefore there is an acknowledgement that changes in one element can affect the counts of another element, and hence the need to normalize it. It was therefore surprising that following the statistical analysis the Ti records that are shown and on which the storm records are based were raw counts per second (kcps) results, rather than the CLR results. Previous studies (e.g. Löwemark et al. 2011; Weltje and Tjallingii, 2008; Croudace et al. 2019) have highlighted that ITRAX results should be normalized because otherwise variations in the amount of undetected organic matter, water content, surface roughness and counts of other elements can all influence the results. Appendix D shows photos of the cores, which have a very rough surface – was the surface smoothed at all prior to ITRAX scan as this would surely influence the ITRAX results greatly? This makes it important that the results are normalized as the variability in the surface would have likely influenced the measured Ti kcps. The CLR results (or natural log ratios with a carefully selected denominator) should provide the basis for the storm records.
Author response:
We appreciate the reviewer’s concerns and comments regarding the use of raw Ti counts in the storm reconstructions rather than CLR-transformed values. The reviewer points out that compositional data, such as ITRAX-derived elemental data, require appropriate normalization to account for the matrix effect (the influence that the overall composition of a sample has on the accuracy of an element’s measurements). We would like to clarify our rationale for not applying CLR – or any other log-ratio transformation – to the Ti data used specifically for the storm event reconstructions, and propose an alternative normalization.
First, we can confirm that the core surface was smoothed prior to ITRAX scanning, reducing the impact of surface roughness on CPS. Pictures taken during the XRF analysis can be provided if need be (pictures in the Appendix are from the fieldwork).
We agree that log-ratios, including CLR, are valuable for addressing many issues inherent to compositional data, as outline by Löwemark et al. 2011, Weltje et al. 2008, 2015, and Croudace 2015. The CLR transformation is particularly useful in the context of multivariate analysis, like PCA and correlations analysis, because it controls for the matrix effect and allows these analyses to focus on the relative proportions of elements. This transformation corrects for compositional constraints, ensuring that the results reflect the relationships between elements without being confounded by the “absolute” concentration of the elements in the sample, and reduce impacts of spurious correlations.
However, we believe that using CLR-transformed Ti data for the extreme value analysis in storm event detection would not be appropriate. Log transformations, such as CLR, tend to reduce the variance and smooth out extreme values in the data. Since extreme value analysis relies on detecting peaks or outliers in the data, applying CLR would likely diminish the prominence of extreme values, which may correspond to significant storm events.
Nevertheless, we agree with the reviewer that normalization is important to address some of the compositional nature of the Ti data. To strengthen our analysis, we propose to follow Bertrand et al. (2023)’s normalization guidelines by normalizing the Ti data to the total counts of lithogenic elements (equivalent to a back-transformation of CLR data): Ti/[Ti+K+Fe+Si+Mn+S+Si]. This will ensure that we measure Ti extreme values relative to the other lithogenic elements, rather than having Ti influenced by confounding elements, including those that aren’t detected by ITRAX, such as carbon. Preliminary results using this method show patterns consistent with our original approach, while providing additional robustness and preserving the original data distribution (see Figure 1 in the supplementary PDF). We would like to refer to this study which uses a similar approach: https://link.springer.com/article/10.1007/s10933-024-00345-9#ref-CR18.
- Interpretation and reliance on Ti.
Reviewer comment:
Why was Ti selected as the element most indicative of mineral content over other elements? I think more justification of this decision is needed. The Principal Component Analysis in figure 4 shows that for TLM Ti is not strongly correlated with PC1, when the ASI results and other elements are. Figure 7 which shows the ASI overlaying the Ti record does support the use of Ti and perhaps this comparison should be done earlier in the paper (and using the full records), as currently much of the ITRAX results are discounted with little explanation in the results section.
Author response:
Regarding our use of Ti as the primary indicator of aeolian mineral input and the basis for storm reconstruction: We agree that the current presentation would benefit from a clearer and more detailed justification for both the selection and interpretation of Ti. We would like to outline proposed revisions to improve the clarity, robustness, and transparency of our approach.
First, we acknowledge the reviewer’s concern about the relatively weak association between Ti and PC1 in TLM. We think that this pattern is primarily driven by the skewed distribution of ASI in the TLM core, where Zone 1 (the lower, older portion) exhibits high ASI. This Zone is dominated by Fe and S relative to Ti (evident in Fig. 3: Fe and S increases in Zone 1, and PC1, for which Fe and S show the strongest , indicating a relatively important presence of Fe and S compared to other elements). Thus Ti, as a relatively minor contributor to the total ASI content of TLM, appears to be more weakly correlated to PC1, and appears to be a weak proxy for ASI.
However, we think that this issue could be addressed by dealing with ASI differently. At present, we use the term “ASI” uniformly when describe all the zones of the cores. This may have caused confusion, as the mineral content in the oldest parts of the cores (fen/wetland stages) is likely not entirely of aeolian origin (probably more from groundwater and surface water processes). We can only assume that the mineral content is of aeolian origin in the ombrotrophic portions of the cores. With that in mind, we propose the following revision:
- Clarify figure content and terminology: We will remove references to ASI from Figure 3 and 4, as well as from sections 4.2 and 4.3. These sections and figures will instead focus on a general description of the core’s stratigraphy and geochemistry, setting the stage for our interpretation of the sites’ development (e.g., timing of the fen-bog transition) and our selection of specific zones showing ombrotrophic conditions for storm reconstruction. To reinforce this, section 4.1 (the age-depth model) to 4.3 (geochemistry) will be grouped under a new sub-heading titled “Core Description”, the emphasize the focus on whole-core characterization rather than aeolian input.
- Introduce a dedicated results section on aeolian proxies: We will add a new section to the results, placed between section 4.3 and 4.4 (Storm reconstructions), titled “Aeolian sand influx and proxies”. This section will focus exclusively on the ombrotrophic zones (i.e., those selected for storm reconstruction) and present new analyses to justify the choice of Ti as an ASI proxy over other potential element proxies such as Fe and K. We will provide side-by-side downcore trend graphs of ASI and geochemical variables (Ti, K, Fe) and a supplementary PCA limited to ombrotrophic intervals and including the ASI. This analysis demonstrates that both Ti and ASI consistently co-vary with PC1 in the ombrotrophic zones of TLM and TAC, and more so than the other elements, including K. A quick draft of the proposed figures and analysis is provided in the supplementary PDF (Figure ).
Reviewer comment:
The event frequency was calculated using the number of times that Ti exceeded a threshold. The data was detrended using a 10/30 year moving average window and peaks above a threshold were marked as storm events. The authors state ‘To avoid overestimating the number of events due to consecutive Ti measurements associated with a single Ti peak, consecutive measurements were grouped together, and only the maximum value within each group was retained’. I wonder however whether some of the identified storm events are represented by single data points or small groups, which could be outliers or erroneous data points rather than sand layers. As noted above, in data that has not been normalized other factors could cause an increase or decrease in the kcps that is unrelated to mineral content in the peat. The authors should check the Mean Squared Error and total counts per second measurements through the records to see if any of the peaks in Ti are associated with anomalies and should be excluded as artefacts (see Lowemark et al. 2019).
Author response:
We agree with the reviewer that short-lived peaks in Ti may result from isolated data points or measurement noise rather than discrete storm events. First, we confirm that we applied quality control to the raw XRF data (following the method in Using ITRAX data in R from Thomas Bishops, https://tombishop1.github.io/itraxBook/index.html), but this process was not sufficiently described in the Methods section. We will indeed add this citation to our manuscript. We will revise this section to clarify that all element measurement corresponding to a CPS or MSE values outside the normal distribution were deleted.
Our earlier suggestion to introduce a dedicated results section on aeolian proxies will certainly strengthen our storm event identification analysis and confidence in Ti anomalies relating to storms. We propose overlaying the ASI and Ti measurements in Figure 5 to show the correspondence between the two records, and note where peaks co-occurrence occurs.
Reviewer comment:
A point related to this one about the detection of storm events, is that the moving window used may be too narrow and single large events may be identified as several small events. If you look at Figure 5b and the record of TAC the red line of the moving average threshold goes up over some of the increases in Ti as a result of the concentrations increasing over a >10 year period (depending on the accumulation rate 10 years could be just 1-2cm within the peat). My concern is that if there was a large deposit of sand on the bog by a single large magnitude storm, then sand may settle on the bog at different depths due to the uneven surface, with some falling into pools and sinking down, some at the roots, others on the top of plants etc.. Further mixing could occur with root bioturbation potentially. You acknowledge in the discussion ‘downward movement of sediments in the spongey peat matrix may have dated sediments associated to Hurricane Dorien to an older interval’ to explain the timing of an event, so a similar mechanism could spread the sand from a single large event over a few cm’s within the consolidated peat. Therefore, this could lead to overestimation of the number of events if these larger increases in sand are removed by detrending and then small variations in Ti on these peaks are counted as events. Either the method should be adjusted to address this or these limitations should be discussed, with perhaps more cautious statements about the frequency of events.
Author response:
We appreciate the reviewer’s concern that the 10/30-year detrending window may fragment large events into multiple smaller ones, by flattening out the large and long-term peak in Ti and therefore focusing on short term increases, especially if sand from a single event is spread vertically due to post-depositional processes (e.g., bioturbation or uneven settling). If aeolian sands settle at different depth, it could therefore be measured multiple times, overestimating the number of events identified. However, we think that downward movement of particles following deposition may be negligible. While we are not aware of studies addressing the downward movement of mineral sediments after deposition on peat, similar challenges have been encountered in research on microplastic deposition in peat. For instance, Allen et al. (2020) (https://doi.org/10.1021/acs.estlett.1c00697) measured microplastic deposition over the past 100 years (~ 45 cm of peat) from a peat core in the Pyrenees, and compared it to a nearby lake record. The peat archive showed a remarkable correspondence between the age-depth model, microplastic trends, and European plastic production trends, showing that microplastic particles do not significantly move downward after deposition, and supporting the assumption that aeolian sand, which is of similar size to microplastic, may behave similarly. Although our resolution for Ti is 1 mm (compared to 1 cm in the Allan et al. study), this provides a useful starting point to constrain the potential downward movement of particles.
Additionally, given the rapid accumulation rates at the peat surface (e.g., 1.5-2 cm per year at TLM; 0.45 cm per year at TAC), we suggest that sand would need to travel a significant distance downward to span multiple depths over several years. Additionally, as peat becomes buried and densified, downward movement becomes increasingly unlikely, though not impossible due to bioturbation. Ti, however, is a highly conservative element and normally remains stable in the peat column.
To address this, we propose expanding the discussion of this issue in section 5.2 on modern storm attribution, emphasizing the potential for downward movement of sand in the peat matrix distorting the vertical expression of storm layers, while nuancing with the above explanation.
- Exclusion of part of the TAC record:
Reviewer comment:
You should be consistent about whether or not the period between ~600 BCE and ~1000 CE in TAC is useful as a storm record, or not given you interpret it as being not ombrotrophic. In Figure 5 you shaded out the section but in Figure 8 it is included. In the discussion you also observe that the Ti and storm events records of the two cores are similar over this section. While you have concluded that TAC was not ombrotrophic at this time, that is not to say that the peaks in Ti and sand shown by the TAC record were not deposited by the same storms that caused the deposits in TLM. Perhaps better not to discount this section of the TLM record but to rather use it with the caveat that deposits may come from other sources in addition to storms.
Author response:
We appreciate the reviewer’s comment on maintaining consistency in interpreting this section of the TAC record. We would like to point out that, in section 4.4 of the manuscript, we do not conclude that TAC is not ombrotrophic, rather than we cannot confirm that it is ombrotrophic, as it lacks clear ombrotrophic conditions. We will make sure that this distinction is clear in the revised manuscript.
We thank the reviewer for pointing out that this section is shaded in Figure 5, but not in Figure 8. For consistency, we suggest shading this section in Figure 8 as well, and use this part of the TAC record while clearly accounting for its caveats, allowing readers to interpret the evidence with this context.
Minor comments
Reviewer comment:
Line 402 – this sentence says that sand layers in TAC at 600BCE and 810 CE were not visible in TLM. But there was an increase in identified storm events coinciding with these times in TLM, so could it be that there was an enhanced sand deposition related to storms at these times, but just more sand reached TAC?
Author response:
We agree with the reviewer that the sand layers at TAC could indicate heightened sensitivity to storm-related deposition, rather than being due to independent in-situ processes, as is currently implied in the discussion. While the exact source and mechanism for these layers remain unclear, the synchronicity with increased depositional activity in TLM points to a regional, possibly climatic or storminess signal. Therefore, we propose expanding the discussion 5.1.1 to include other potential mechanisms and hypothesis.
In particular, we would emphasize that our interpretation of in-situ mineral concentration is informed, in part, by the age model, which suggests a hiatus or very low accumulation rate during this period, and the presence of detritus peat suggesting local hydrological shifts (dryer conditions) that may have enhanced in situ mineral accumulation. These local factors could have been combined with increased sand availability due to a dryer climate, and possibly an increased in storm activity, as suggested by the TLM record, resulting in more deposition both at TAC and TLM.
We would also clearly acknowledge the limitations of our current age-model, with significant chronological uncertainty during this period.
Reviewer comment:
Line 403 – The sentence says that there was no abrupt contacts between the sand and overlying and underlying peat, supporting a gradual accumulation of sand over time. I am not sure about this interpretation as peat bog environments don’t seem to often have sharp boundaries in the same way as lakes. The surface of bogs are uneven and I could see that when sand blows in a single event over the surface it could fall into pools, sit on top of plants but also land or wash down to the base of the plant and so be incorporated within different depths giving a gradual boundary even for a single large event.
Author response:
We think it is unlikely that these layers are the results of a single event due to the fact that they are not pure mineral sediments (they have between 32 % and 55 % mineral content); in contrast, overwash layers in lakes usually appear as distinct, inorganic layers. Additionally, the elevation of TAC, at ~20 m above sea level, and lack of directly proximate beach, precludes the idea of an overwash type event at TAC; a tremendous amount of sand would have needed to be transported through aeolian process to leave such a layer as a single event, and we believe that such an event would most probably be visible in TLM as well.
Reviewer comment:
Line 422 – the sentence suggests the proximity of TAC to the cliffs is the reason for the larger particles at TAC. But are the cliffs the sediment source during storms? Are sand sized particles being eroded from the cliffs and transported in land during storms? It looks like TAC is slightly closer to the beach to the west than TLM is, but the satellite photos seem to show the environment around TAC is dryer, so would exposed soil potentially be providing a source of minerals to TAC?
Line 426 – in this sentence again the assumption is that the windblown sands came from the cliffs and/or beaches. I expect the soil on the island would have a similar elemental composition to the bedrock, so could wind blown soil also be a contributor and spatial variations in vegetation a factor?
Author response:
We would like to point here that, in section 5.1.2, when describing the distinct aeolian sediment patterns between TAC and TLM, we are stating that “ASI values at TLM were an order of magnitude smaller than those at TAC", meaning the quantity of aeolian – and not necessarily the grain-size – is significantly higher at TAC compared to TLM. We will rephrase this part of the discussion to make this point clearer. We will also add a clear statement about the friability of the sandstone sediments, which can easily be eroded by winds during storms. Additionally, while the tops of the cliffs are nearly at the same elevation as the bog, the beaches are much lower, and therefore sediment would have a greater vertical distance to travel before settling at TAC compared to sediments from the cliffs.
Additionally, we propose mentioning that the surrounding environment is now drier and more open due to farmland, which may have enhanced recent aeolian activity. However, we would acknowledge that these land-use changes are recent (~200 years) and unlikely to have influenced sediment delivery during earlier periods.
Citation: https://doi.org/10.5194/egusphere-2025-400-CC1 -
AC2: 'Reply on RC1', Antoine Lachance, 13 May 2025
We sincerely thank Reviewer 1 for their thoughtful and constructive reviews and comments. Below, we provide a detailed response for each of the main concerns regarding (1) data processing and normalization, (2) the use and interpretation of Ti as a proxy for storm activity, and (3) the interpretation of the TAC record. We also address several minor comments. Suggested corrections and modifications are being incorporated into the revised manuscript. We have taken steps to clarify our decision-making process, improve the presentation of the results and discussion, and highlight key findings more effectively. We greatly appreciate these suggestions, which have helped improve the overall quality of the manuscript.
Major corrections
- XRF Data processing and normalization
Reviewer comment:
The authors do a good job with the statistical analysis of the ITRAX data, converting it to Centered Log Ratios before doing the statistical analysis, because ITRAX data is compositional data. Therefore there is an acknowledgement that changes in one element can affect the counts of another element, and hence the need to normalize it. It was therefore surprising that following the statistical analysis the Ti records that are shown and on which the storm records are based were raw counts per second (kcps) results, rather than the CLR results. Previous studies (e.g. Löwemark et al. 2011; Weltje and Tjallingii, 2008; Croudace et al. 2019) have highlighted that ITRAX results should be normalized because otherwise variations in the amount of undetected organic matter, water content, surface roughness and counts of other elements can all influence the results. Appendix D shows photos of the cores, which have a very rough surface – was the surface smoothed at all prior to ITRAX scan as this would surely influence the ITRAX results greatly? This makes it important that the results are normalized as the variability in the surface would have likely influenced the measured Ti kcps. The CLR results (or natural log ratios with a carefully selected denominator) should provide the basis for the storm records.
Author response:
We appreciate the reviewer’s concerns and comments regarding the use of raw Ti counts in the storm reconstructions rather than CLR-transformed values. The reviewer points out that compositional data, such as ITRAX-derived elemental data, require appropriate normalization to account for the matrix effect (the influence that the overall composition of a sample has on the accuracy of an element’s measurements). We would like to clarify our rationale for not applying CLR – or any other log-ratio transformation – to the Ti data used specifically for the storm event reconstructions, and propose an alternative normalization.
First, we can confirm that the core surface was smoothed prior to ITRAX scanning, reducing the impact of surface roughness on CPS. Pictures taken during the XRF analysis can be provided if need be (pictures in the Appendix are from the fieldwork).
We agree that log-ratios, including CLR, are valuable for addressing many issues inherent to compositional data, as outline by Löwemark et al. 2011, Weltje et al. 2008, 2015, and Croudace 2015. The CLR transformation is particularly useful in the context of multivariate analysis, like PCA and correlations analysis, because it controls for the matrix effect and allows these analyses to focus on the relative proportions of elements. This transformation corrects for compositional constraints, ensuring that the results reflect the relationships between elements without being confounded by the “absolute” concentration of the elements in the sample, and reduce impacts of spurious correlations.
However, we believe that using CLR-transformed Ti data for the extreme value analysis in storm event detection would not be appropriate. Log transformations, such as CLR, tend to reduce the variance and smooth out extreme values in the data. Since extreme value analysis relies on detecting peaks or outliers in the data, applying CLR would likely diminish the prominence of extreme values, which may correspond to significant storm events.
Nevertheless, we agree with the reviewer that normalization is important to address some of the compositional nature of the Ti data. To strengthen our analysis, we propose to follow Bertrand et al. (2023)’s normalization guidelines by normalizing the Ti data to the total counts of lithogenic elements (equivalent to a back-transformation of CLR data): Ti/[Ti+K+Fe+Si+Mn+S+Si]. This will ensure that we measure Ti extreme values relative to the other lithogenic elements, rather than having Ti influenced by confounding elements, including those that aren’t detected by ITRAX, such as carbon. Preliminary results using this method show patterns consistent with our original approach, while providing additional robustness and preserving the original data distribution (see Figure 1 in the supplementary PDF). We would like to refer to this study which uses a similar approach: https://link.springer.com/article/10.1007/s10933-024-00345-9#ref-CR18.
- Interpretation and reliance on Ti.
Reviewer comment:
Why was Ti selected as the element most indicative of mineral content over other elements? I think more justification of this decision is needed. The Principal Component Analysis in figure 4 shows that for TLM Ti is not strongly correlated with PC1, when the ASI results and other elements are. Figure 7 which shows the ASI overlaying the Ti record does support the use of Ti and perhaps this comparison should be done earlier in the paper (and using the full records), as currently much of the ITRAX results are discounted with little explanation in the results section.
Author response:
Regarding our use of Ti as the primary indicator of aeolian mineral input and the basis for storm reconstruction: We agree that the current presentation would benefit from a clearer and more detailed justification for both the selection and interpretation of Ti. We would like to outline proposed revisions to improve the clarity, robustness, and transparency of our approach.
First, we acknowledge the reviewer’s concern about the relatively weak association between Ti and PC1 in TLM. We think that this pattern is primarily driven by the skewed distribution of ASI in the TLM core, where Zone 1 (the lower, older portion) exhibits high ASI. This Zone is dominated by Fe and S relative to Ti (evident in Fig. 3: Fe and S increases in Zone 1, and PC1, for which Fe and S show the strongest , indicating a relatively important presence of Fe and S compared to other elements). Thus Ti, as a relatively minor contributor to the total ASI content of TLM, appears to be more weakly correlated to PC1, and appears to be a weak proxy for ASI.
However, we think that this issue could be addressed by dealing with ASI differently. At present, we use the term “ASI” uniformly when describe all the zones of the cores. This may have caused confusion, as the mineral content in the oldest parts of the cores (fen/wetland stages) is likely not entirely of aeolian origin (probably more from groundwater and surface water processes). We can only assume that the mineral content is of aeolian origin in the ombrotrophic portions of the cores. With that in mind, we propose the following revision:
- Clarify figure content and terminology: We will remove references to ASI from Figure 3 and 4, as well as from sections 4.2 and 4.3. These sections and figures will instead focus on a general description of the core’s stratigraphy and geochemistry, setting the stage for our interpretation of the sites’ development (e.g., timing of the fen-bog transition) and our selection of specific zones showing ombrotrophic conditions for storm reconstruction. To reinforce this, section 4.1 (the age-depth model) to 4.3 (geochemistry) will be grouped under a new sub-heading titled “Core Description”, the emphasize the focus on whole-core characterization rather than aeolian input.
- Introduce a dedicated results section on aeolian proxies: We will add a new section to the results, placed between section 4.3 and 4.4 (Storm reconstructions), titled “Aeolian sand influx and proxies”. This section will focus exclusively on the ombrotrophic zones (i.e., those selected for storm reconstruction) and present new analyses to justify the choice of Ti as an ASI proxy over other potential element proxies such as Fe and K. We will provide side-by-side downcore trend graphs of ASI and geochemical variables (Ti, K, Fe) and a supplementary PCA limited to ombrotrophic intervals and including the ASI. This analysis demonstrates that both Ti and ASI consistently co-vary with PC1 in the ombrotrophic zones of TLM and TAC, and more so than the other elements, including K. A quick draft of the proposed figures and analysis is provided in the supplementary PDF (Figure ).
Reviewer comment:
The event frequency was calculated using the number of times that Ti exceeded a threshold. The data was detrended using a 10/30 year moving average window and peaks above a threshold were marked as storm events. The authors state ‘To avoid overestimating the number of events due to consecutive Ti measurements associated with a single Ti peak, consecutive measurements were grouped together, and only the maximum value within each group was retained’. I wonder however whether some of the identified storm events are represented by single data points or small groups, which could be outliers or erroneous data points rather than sand layers. As noted above, in data that has not been normalized other factors could cause an increase or decrease in the kcps that is unrelated to mineral content in the peat. The authors should check the Mean Squared Error and total counts per second measurements through the records to see if any of the peaks in Ti are associated with anomalies and should be excluded as artefacts (see Lowemark et al. 2019).
Author response:
We agree with the reviewer that short-lived peaks in Ti may result from isolated data points or measurement noise rather than discrete storm events. First, we confirm that we applied quality control to the raw XRF data (following the method in Using ITRAX data in R from Thomas Bishops, https://tombishop1.github.io/itraxBook/index.html), but this process was not sufficiently described in the Methods section. We will indeed add this citation to our manuscript. We will revise this section to clarify that all element measurement corresponding to a CPS or MSE values outside the normal distribution were deleted.
Our earlier suggestion to introduce a dedicated results section on aeolian proxies will certainly strengthen our storm event identification analysis and confidence in Ti anomalies relating to storms. We propose overlaying the ASI and Ti measurements in Figure 5 to show the correspondence between the two records, and note where peaks co-occurrence occurs.
Reviewer comment:
A point related to this one about the detection of storm events, is that the moving window used may be too narrow and single large events may be identified as several small events. If you look at Figure 5b and the record of TAC the red line of the moving average threshold goes up over some of the increases in Ti as a result of the concentrations increasing over a >10 year period (depending on the accumulation rate 10 years could be just 1-2cm within the peat). My concern is that if there was a large deposit of sand on the bog by a single large magnitude storm, then sand may settle on the bog at different depths due to the uneven surface, with some falling into pools and sinking down, some at the roots, others on the top of plants etc.. Further mixing could occur with root bioturbation potentially. You acknowledge in the discussion ‘downward movement of sediments in the spongey peat matrix may have dated sediments associated to Hurricane Dorien to an older interval’ to explain the timing of an event, so a similar mechanism could spread the sand from a single large event over a few cm’s within the consolidated peat. Therefore, this could lead to overestimation of the number of events if these larger increases in sand are removed by detrending and then small variations in Ti on these peaks are counted as events. Either the method should be adjusted to address this or these limitations should be discussed, with perhaps more cautious statements about the frequency of events.
Author response:
We appreciate the reviewer’s concern that the 10/30-year detrending window may fragment large events into multiple smaller ones, by flattening out the large and long-term peak in Ti and therefore focusing on short term increases, especially if sand from a single event is spread vertically due to post-depositional processes (e.g., bioturbation or uneven settling). If aeolian sands settle at different depth, it could therefore be measured multiple times, overestimating the number of events identified. However, we think that downward movement of particles following deposition may be negligible. While we are not aware of studies addressing the downward movement of mineral sediments after deposition on peat, similar challenges have been encountered in research on microplastic deposition in peat. For instance, Allen et al. (2020) (https://doi.org/10.1021/acs.estlett.1c00697) measured microplastic deposition over the past 100 years (~ 45 cm of peat) from a peat core in the Pyrenees, and compared it to a nearby lake record. The peat archive showed a remarkable correspondence between the age-depth model, microplastic trends, and European plastic production trends, showing that microplastic particles do not significantly move downward after deposition, and supporting the assumption that aeolian sand, which is of similar size to microplastic, may behave similarly. Although our resolution for Ti is 1 mm (compared to 1 cm in the Allan et al. study), this provides a useful starting point to constrain the potential downward movement of particles.
Additionally, given the rapid accumulation rates at the peat surface (e.g., 1.5-2 cm per year at TLM; 0.45 cm per year at TAC), we suggest that sand would need to travel a significant distance downward to span multiple depths over several years. Additionally, as peat becomes buried and densified, downward movement becomes increasingly unlikely, though not impossible due to bioturbation. Ti, however, is a highly conservative element and normally remains stable in the peat column.
To address this, we propose expanding the discussion of this issue in section 5.2 on modern storm attribution, emphasizing the potential for downward movement of sand in the peat matrix distorting the vertical expression of storm layers, while nuancing with the above explanation.
- Exclusion of part of the TAC record:
Reviewer comment:
You should be consistent about whether or not the period between ~600 BCE and ~1000 CE in TAC is useful as a storm record, or not given you interpret it as being not ombrotrophic. In Figure 5 you shaded out the section but in Figure 8 it is included. In the discussion you also observe that the Ti and storm events records of the two cores are similar over this section. While you have concluded that TAC was not ombrotrophic at this time, that is not to say that the peaks in Ti and sand shown by the TAC record were not deposited by the same storms that caused the deposits in TLM. Perhaps better not to discount this section of the TLM record but to rather use it with the caveat that deposits may come from other sources in addition to storms.
Author response:
We appreciate the reviewer’s comment on maintaining consistency in interpreting this section of the TAC record. We would like to point out that, in section 4.4 of the manuscript, we do not conclude that TAC is not ombrotrophic, rather than we cannot confirm that it is ombrotrophic, as it lacks clear ombrotrophic conditions. We will make sure that this distinction is clear in the revised manuscript.
We thank the reviewer for pointing out that this section is shaded in Figure 5, but not in Figure 8. For consistency, we suggest shading this section in Figure 8 as well, and use this part of the TAC record while clearly accounting for its caveats, allowing readers to interpret the evidence with this context.
Minor comments
Reviewer comment:
Line 402 – this sentence says that sand layers in TAC at 600BCE and 810 CE were not visible in TLM. But there was an increase in identified storm events coinciding with these times in TLM, so could it be that there was an enhanced sand deposition related to storms at these times, but just more sand reached TAC?
Author response:
We agree with the reviewer that the sand layers at TAC could indicate heightened sensitivity to storm-related deposition, rather than being due to independent in-situ processes, as is currently implied in the discussion. While the exact source and mechanism for these layers remain unclear, the synchronicity with increased depositional activity in TLM points to a regional, possibly climatic or storminess signal. Therefore, we propose expanding the discussion 5.1.1 to include other potential mechanisms and hypothesis.
In particular, we would emphasize that our interpretation of in-situ mineral concentration is informed, in part, by the age model, which suggests a hiatus or very low accumulation rate during this period, and the presence of detritus peat suggesting local hydrological shifts (dryer conditions) that may have enhanced in situ mineral accumulation. These local factors could have been combined with increased sand availability due to a dryer climate, and possibly an increased in storm activity, as suggested by the TLM record, resulting in more deposition both at TAC and TLM.
We would also clearly acknowledge the limitations of our current age-model, with significant chronological uncertainty during this period.
Reviewer comment:
Line 403 – The sentence says that there was no abrupt contacts between the sand and overlying and underlying peat, supporting a gradual accumulation of sand over time. I am not sure about this interpretation as peat bog environments don’t seem to often have sharp boundaries in the same way as lakes. The surface of bogs are uneven and I could see that when sand blows in a single event over the surface it could fall into pools, sit on top of plants but also land or wash down to the base of the plant and so be incorporated within different depths giving a gradual boundary even for a single large event.
Author response:
We think it is unlikely that these layers are the results of a single event due to the fact that they are not pure mineral sediments (they have between 32 % and 55 % mineral content); in contrast, overwash layers in lakes usually appear as distinct, inorganic layers. Additionally, the elevation of TAC, at ~20 m above sea level, and lack of directly proximate beach, precludes the idea of an overwash type event at TAC; a tremendous amount of sand would have needed to be transported through aeolian process to leave such a layer as a single event, and we believe that such an event would most probably be visible in TLM as well.
Reviewer comment:
Line 422 – the sentence suggests the proximity of TAC to the cliffs is the reason for the larger particles at TAC. But are the cliffs the sediment source during storms? Are sand sized particles being eroded from the cliffs and transported in land during storms? It looks like TAC is slightly closer to the beach to the west than TLM is, but the satellite photos seem to show the environment around TAC is dryer, so would exposed soil potentially be providing a source of minerals to TAC?
Line 426 – in this sentence again the assumption is that the windblown sands came from the cliffs and/or beaches. I expect the soil on the island would have a similar elemental composition to the bedrock, so could wind blown soil also be a contributor and spatial variations in vegetation a factor?
Author response:
We would like to point here that, in section 5.1.2, when describing the distinct aeolian sediment patterns between TAC and TLM, we are stating that “ASI values at TLM were an order of magnitude smaller than those at TAC", meaning the quantity of aeolian – and not necessarily the grain-size – is significantly higher at TAC compared to TLM. We will rephrase this part of the discussion to make this point clearer. We will also add a clear statement about the friability of the sandstone sediments, which can easily be eroded by winds during storms. Additionally, while the tops of the cliffs are nearly at the same elevation as the bog, the beaches are much lower, and therefore sediment would have a greater vertical distance to travel before settling at TAC compared to sediments from the cliffs.
Additionally, we propose mentioning that the surrounding environment is now drier and more open due to farmland, which may have enhanced recent aeolian activity. However, we would acknowledge that these land-use changes are recent (~200 years) and unlikely to have influenced sediment delivery during earlier periods.
Citation: https://doi.org/10.5194/egusphere-2025-400-AC2
-
CC1: 'Reply on RC1', Antoine Lachance, 12 Apr 2025
-
RC2: 'Comment on egusphere-2025-400', Anonymous Referee #2, 24 Mar 2025
SUMMARY: The authors present a detailed multi-proxy reconstruction of climate and environmental change over late Holocene timescales based on two cores from two different peat bogs on the Magdalen Islands, eastern Canada, and compare these to recorded events of storms in the past in order to facilitate a storminess reconstruction for the last 4000 years. The analyses involve a description of lithology, 14C- and 210Pb-based chronologies, XRF-core scanning, as well as measurements of LOI and grain size. Single elements identified in the XRF data, the inferred mineral content and the relative contribution of “sand” (based on the “Aeolian Sand Index”, Björck & Clemmensen, 2004) for the fractions <63µm and <125µm are used as proxies for storminess. Overall, the manuscript is well-written and thoroughly prepared and builds on sound and timely methods. There is a lot of positive things about this manuscript, but my main concern is the (uncontested) validity of the data as a storm record. This is for three main reasons: (1) The authors do not discuss local landscape evolution and potential changes regarding the sources of the allochthonous material blown into the bogs by storms or other dynamics that could have influenced the amounts deposited (such as exposure or storm direction). (2) The used grain sizes for ASI differ from other applications (which of course is not an issue), but are neither explained nor argued for, in terms of source properties or the processes determining transport (i.e., storminess). (3) There are a couple of issues with the chronologies (which is not uncommon for a natural system), but my critique is that the inferred age models are used a little too uncritical in the reconstruction of storminess and the discussion of climate in the big picture. That said, a huge strength of the manuscript is innovative combination of different parameters and methods (to my surprise very little GIS/geodata), which is extremely useful for depositional environments at the land-ocean interface and has a lot of potential. An ideal correspondence between all parameters cannot be expected, but I would regardless of that strongly encourage the authors to revise the manuscript thoroughly and argue for their data set more carefully and convincingly. I have added a couple of comments, both major and minor, below (ordered by line number).
L79: Small typo here and again further below. The author’s name is “Björck”.
L131ff: A little later on you mention a peat dome. Is there anything for the field site that is worth mentioning regarding to local geomorphological context (do you have access to LiDAR data or another type of digital elevation model for the island)? A quick look at Google Earth reveals that there is a prograded beach-ridge system located to the NE of your site TLM. Is the age of the system known? Its presence suggests that beaches have been as close to the sampling site as ~1200 m at some point in the past. Is there anything known about local coastal evolution at TLM and TAC (or for the island as a whole)? (There are a couple of really interesting studies by Billy et al. for the nearby island of Miquelon). The youthern coast around TAC appears to be a rock coast mostly, but there are smaller (pocket) beaches nearby. Have beaches ever been wider here? In analogy to that: Is there anything known about the past vegetation or land use history of the Magdalene Island (e.g., have the forests ever been cleared and has the peat ever been harvested for fuel)?
L144ff: How long are the cores overall and what’s their surface elevation? Did you reach the bottom of the bog? If a moraine, a marine sequence or something similar was reached, this would be very useful as supporting information of the chronology.
L149: A bit of unnecessary information here. Important is that they were stored cold.
L150: What is the “correction factor”? A percentage? This is a bit unclear.
L161ff: I am not sure to understand. This means that there is no valid Pb-based chronology for the TLM monolith, right?
L173ff: I would suggest to stick with a simpler nomenclature for the LOI processing stages and to use dry weight (DW or DW105) and LOI550 for combusted samples. Also, I feel that drying peat samples at 105°C overnight appears a little short.
L179: Unlike LOI, the ASI approach is not a very well established as a method and does require a more detailed description or more thorough and precise referencing. Did you consider the whole grain-size distribution of the samples or was there a significant amount of larger minerogenic particles present? It would be worthwhile to reference a bit more here and argue for the approach. Also, it reads here as if the fraction <125 µm is used as a proxy, but in the discussion, it is clearly stated that >125 µm is used. What did you use? Also, I would suggest to separate LOI and ASI, or change the title to “Variations in mineral content and ASI”. LOI is often used to identify the content of organic material in a sample, and I was a little confused that you went straight to the residual here.
L192: Often (but not always) LOI correlates well with measured values of TOC based on element analysis, so arguably LOI is an absolute measure (with a bit of uncertainty and the occasional outlier). Would the results look much different would you skip the data processing as described?
L228: You mention the identification of (non-)aeolian processes here. Did you by any chance measure full grain-size distributions on the samples from the core or on a few of the samples from the bottom of the cores at least? This would be very useful to pinpoint the differences in sources and processes.
L229f: The last sentence is very interesting, but does not really fit into the methods and would be more suitable for the discussion.
L243ff: In a similar fashion, the interpretation of the Ti outliers should be moved to the discussion.
L249ff: This sounds very promising and is a huge effort on your part! I look forward to read, how that integrates into indications from the lead-dated monoliths.
L261f: The sand layers at the bottom of the core are relatively thin. How can you be certain that they are a basal sand layer that marks the bottom of the peat sequence (and not just an interbed, e.g. from overwash)?
L265: Arguably, the marked drop in PAR could be interpreted as a hiatus rather than a stage of very slow accumulation (which slightly overstretches the ability of the 14C chronology). Also, there is a lot happening in the core at about the same time (or slightly thereafter). The peat looks much brighter in the core images and the ASI picks up up-core from here. Wouldn’t it be advisable to add an uncertainty to the rates and to subdivide a little more, especially in the case of TAC?
L272: Figure 2. I do not find the top panels above the age-depth models particularly informative (also, the last bit of text to the right is cut off) and you should consider to remove them. It would be helpful for the readers understanding to have a table of the radiocarbon ages with uncalibrated and calibrated ages (ideally as ranges), the 2-sigma probabilities, and info on the material dated and possibly the lithology of the unit dated. (NB: Sorry! I just found the table in the appendix – it could be useful to have this integrated into the main text). While the age model for TLM looks fairly straight forward, there is a tricky hiatus (or continued deposition followed by erosion?) at 150 cm. Would there be a good way (visually) to scale the y-axes the same way? The fact that the oldest peat deposits have a surprisingly similar age is easily overlooked!
L280ff: I feel it could be sensible to exclude the ASI (as a separate proxy) from the description of the lithology, which is (mostly) descriptive otherwise.
L300: Figure 3. I find it difficult to compare the sites visually. Would it be an option to display the data scaled for age and displayed with a linear axis (even distance of the tick-marks between e.g. centuries)? In a similar way, the values for ASI and mineral content should be scaled the same way and you could consider to use a log scale to make the small variations more visible. Some of the labels are not entirely visible in the current version (e.g., Si, ASI).
L347: I am a little surprised, that zone 3 of the fore from TAC was not excluded. The hiatus in the record (and the lack of information to fully explain it) is a bit puzzling, but if your storm proxy is the input of allochthonous material into the peat bog, a hiatus can not be ignored, as the entire zone 3 is potentially the time-integrated results of almost 2kyr of all kinds of processes (limited plant growth, a higher ground water table, flooding with marine water, etc.) potentially paired with a near constant influx of sand.
L365: I would suggest to remove the “and hence storminess” but here in the results section. The result is that you detected periods of enhanced mineral inputs, the rest is an interpretation and should ideally be discussed.
L383: There is a very nice match between the two reconstructed event densities for the 880-550 BCE window, that unfortunately falls into the hiatus at TAC (and hence in a time window, where the age model may be highly unreliable. It is very interesting regardless, but I really feel that you should discuss this more. For the 600-800 CE window, I don’t see a way the age model can be trusted for TAC. The phase is nicely visible in the Ti record for TLM, but bare visible in the ASI record.
L430: Not sure, I can follow the argument. Earlier you argued that the XRF values are relative and (unless I missed it) no absolute values were measured. In the data displayed in Figure 3, I do not see an elevated Fe content in the TAC core. The highest relative values are in fact found in the bottom section of core TLM.
L430: A depletion in minerals other than quartz is not a necessity and highly dependent on e.g., local properties (lithology of the source rock), the sediment budget of a site (and hence the time of exposure in the system prior to deposition), and transport distance. In order to make this understandable for all readers, the local properties need to be described more.
L433: It would be useful to show the correlation between the two different (!) ASI values and elements in the manuscript or the supplement.
L440ff: My feeling is, that this section is a little too sloppily written for the strong conclusion made in the end. You should lay out more of your considerations and explain the acting processes.
L441: “relatively unconsolidated“ is difficult to understand as a description. If the sources allow for transport of sediment to a distal site during storms, it would be interesting to know what happens at a more proximal location. Did you observe the formation of primary dunes on the cliff top or the beaches and what are the grain-size properties here? In other words, what happens in the source areas during other conditions than storms?
L449ff: It would be useful to know more about the typical wind directions during these different storm events. This does have a huge a huge impact on how the sources and your storminess archives connect.
L456: I like to use dashes in more informal contexts, like emails, but think that commas would be more appropriate here.
L462: This is news! My understanding up to this point was, that the beaches NE of TLM would be the major source of sediment to the site or those immediately to the south.
L448ff: This is a very interesting section and you are really working at the limits of the ASI approach here. You are considering a time window of ca. 170 years with an averaged accumulation rate on the order of 3 mm year (much less for TLM based on the age model shown in Fig. C1). It would be really interesting to hear your opinion here: Is it at all possible to use the mineral content of peat to get to a storminess reconstruction (ideally with the attribution of single events) for these time scales? Have you considered to lump the data (e.g., to decades)?
L507-608: This section is very important, but I struggle a bit with three things here: (1) The storminess reconstruction stands a bit on shaky grounds for a couple of reasons (understandable reasons, not necessarily flaws!), but the identified storm periods are here taken as fact and I miss the careful arguing a bit, (2) Your results are placed in general climate context here, but while reading through this I repeatedly wondered, if the purpose is to argue for the validity of ASI (and your age model for TLM) or to show the added value of your records, and (3) Several different scales are naturally mixed here and your hypothesis is stuck between the coast of West Africa and distance relationships between source and sink in sediment transport. To summarize, I like the section overall, but I am not sure of its purpose.
Citation: https://doi.org/10.5194/egusphere-2025-400-RC2 -
AC1: 'Reply on RC2', Antoine Lachance, 12 May 2025
We thank Reviewer 2 for their thoughtful review and for highlighting that our storm record appears robust and comparable to other reconstructions for eastern North America, but that the validity of our data feels uncontested. This is an observation that was also raised by Reviewer 1, and, while we do believe in the validity of our methods and results, we agree that a stronger argument is needed to justify our methodological choices and main findings.
In this response, we aim to address the major concerns raised by Reviewer 2, particularly those related to: 1) discussing and incorporating information on the local landscape morphology, its evolution, and related sediment sources; 2) our use and justification of the ASI proxy, especially in terms of linking it to source properties and depositional processes; and 3) the robustness of our chronology. We responded to the major and minor comments by line number, and have already edited the minor corrections and typos directly in the manuscript.
Major and minor comments:
Reviewer comment (section 2.1: Location of the study):
L131ff: A little later on you mention a peat dome. Is there anything for the field site that is worth mentioning regarding to local geomorphological context (do you have access to LiDAR data or another type of digital elevation model for the island)? A quick look at Google Earth reveals that there is a prograded beach-ridge system located to the NE of your site TLM. Is the age of the system known? Its presence suggests that beaches have been as close to the sampling site as ~1200 m at some point in the past. Is there anything known about local coastal evolution at TLM and TAC (or for the island as a whole)? (There are a couple of really interesting studies by Billy et al. for the nearby island of Miquelon). The southern coast around TAC appears to be a rock coast mostly, but there are smaller (pocket) beaches nearby. Have beaches ever been wider here? In analogy to that: Is there anything known about the past vegetation or land use history of the Magdalene Island (e.g., have the forests ever been cleared and has the peat ever been harvested for fuel)?
Authors response:
We fully agree with the reviewer that the local geomorphological and landscape context is critical and should be more clearly incorporated into the manuscript, especially in section 2.1 (Location of the study).
First, we do have access to LiDAR and DEM data for the Magdalen Islands and will revise Figure 1 to include one of these datasets. This addition will help readers better visualize the landscape, its topography, and understand sediment sources and transport dynamics.
We appreciate the references to the work by (Billy et al., 2013; Billy et al., 2018; Billy et al., 2014) on St. Pierre and Miquelon, which provide a strong example of integrated geological, geomorphological, and chronological approaches to understanding barrier formation. While comparable studies are lacking for the Magdalen Islands, there is important work that informs the region’s Holocene evolution. For example, Holocene sea-level trends have been established through studies such as Rémillard et al. (2017) and Barnett et al. (2017), and modern geomorphological processes have been investigated by Bernatchez et al. (2012) and reviewed by Hétu et al. (2020). However, we are not aware of any studies that explicitly link framework geology, coastal geomorphology, and dating techniques to reconstruct barrier evolution in the Magdalen Islands with the same level of detail as Billy et al.’s work.
Regarding the beach ridge system northeast of TLM, Rémillard et al. (2015) provide OSL dates of 550 ± 60 and 400 ± 100 years BP, with older beach ridges located further inland. These dates suggest that around 500 years ago, the beach was indeed situated closer to the TLM site. More broadly, prograded ridge systems across the islands, such as Les Sillons on Havre-aux-Maisons, dated to at least 2600 years ago, appear to have formed despite relative sea-level rise. This was possible through substantial sediment input from the nearby erodible sandstone cliffs. Sediment transport analyses show negative net sand migration rates along the rocky cliffs located south of TAC (Bernatchez et al., 2012), which is in part composed of sandstones (see response to the next comment below), providing sediment for the eastern beaches such as Du Cap and La Grande Échouerie. Over the past 500 years, this would have translated to coastal cliff retreat (cliffs getting closer to TAC) and expanding beaches (for instance, beaches getting farther away from TLM, but possibly counterbalanced by an overall increase in beach sediment availability along the eastern coast). Therefore, the small pocket of beaches nearby TAC were probably less important in the past.
As for human impacts, there is no record of peat harvesting on the islands, and no evidence of permanent settlement prior to the 18th century. Roads were constructed only in the 1930s. We will verify and incorporate more specific land-use history in the site description.
In summary, we will expand Section 2.1 to include a clearer description of the late Holocene coastal evolution, particularly beach ridge development around 500 years ago. While the section currently emphasizes early to mid-Holocene postglacial sea-level change, we agree that the late Holocene landscape evolution—especially as it relates to sediment redistribution and geomorphological change—is underdeveloped.
Reviewer comment (section 3.1: Core sampling):
L144ff: How long are the cores overall and what’s their surface elevation? Did you reach the bottom of the bog? If a moraine, a marine sequence or something similar was reached, this would be very useful as supporting information of the chronology.
Authors response:
Information about the length of the cores, including information on the bottom sediment layers interpreted as the bottom of the bog, is described in section 4.1 (Chronostratigraphy). The information on the surface elevation is provided in section 2.2 (Study sites). While we think that stratigraphic information belongs in the results section, we can add a mention of the lengths of the cores, and that we cored until reaching bottom sediments, in section 3.1 (Core sampling).
We are indeed confident that we reached the bottom of the bog (see response to comment #L261f). However, we cannot at the moment expand on the bottom sediment found at TLM and TAC, as we did not analyse the nature of the sediment other than its mineral content, and we do not have dates specific to those sediment layers (which would necessitate OLS dating due to the very low organic content). According to Rémillard et al. (2013), these deposits probably pertain to the Drift des Demoiselles (see response to comment L430). This unit represents the main surficial deposit on the Havre-Aubert Island and was interpreted to be from glacial origin near TAC and from glaciomarine origin near TLM (Rémillard et al., 2013). In all cases, Rémillard et al. (2017) showed that RSL passed below current sea level at Havre-Aubert Island by 10.7 ka cal BP, which helps constrain our chronology. We can add this information in section 5.1.1 (peat development).
Reviewer comment (section 3.1: Core sampling):
L150: What is the “correction factor”? A percentage? This is a bit unclear.
Authors response:
The correction factor, as described in the manuscript, is the factor by which we must multiply depth measurements in the peat monolith to obtain the original, uncompressed depth. It can be interpreted as a percentage: we must add 79% and 61% of the current (compressed) monolith length to obtain the original uncompressed length, or multiple each depth by 1.79 or 1.61, as we explained in the manuscript. We can add this to our manuscript to make this concept clearer. This only applies to the monoliths (upper 50 cm), not to the cores.
Reviewer comment (section 3.2: Chronological controls):
L161ff: I am not sure to understand. This means that there is no valid Pb-based chronology for the TLM monolith, right?
Authors response:
Thank you for pointing this out. We agree that this aspect of our age-depth model was not described in sufficient detail in the original text, which may have made it difficult to follow. However, the absence of a clearly defined supported 210Pb level in the TLM core does not mean that a valid 210Pb-based chronology is unattainable.
To clarify: 210Pb dating is based on measurements of total 210Pb, which consists of two components (Aquino-López et al., 2018; Blaauw et al., 2018; Cwanek et al., 2025):
- Supported 210Pb, produced in situ through the decay of 226Ra in the sediment, and
- Excess (unsupported) 210Pb, delivered from atmospheric deposition and decaying over time.
Supported 210Pb is assumed to be constant with depth, while excess 210Pb decays exponentially with time and, consequently, with depth in undisturbed sediments such as peat cores. A 210Pb-based chronology is developed by modeling the decay of excess 210Pb until it becomes indistinguishable from supported levels. This depth is often referred to as the equilibrium depth, which typically corresponds to approximately 100–150 years of sediment accumulation.
In traditional models such as the Constant Rate of Supply (CRS) model, it is necessary to measure total 210Pb down to the equilibrium depth to calculate the total inventory of excess 210Pb and determine the age-depth relationship accurately. However, our chronologies for TLM and TAC were developed using the Bayesian age-depth model Plum (Aquino-López et al., 2018), which is less reliant on reaching the supported level directly.
Plum incorporates prior distributions for parameters like sediment accumulation rates and excess 210Pb flux and uses Markov Chain Monte Carlo (MCMC) simulations to generate posterior distributions for these parameters, providing both age estimates and associated uncertainties.
This approach allows the model to infer supported 210Pb levels even when they are not directly observed in the data, inferring from prior information and from the behavior of the decay profile.
In the case of TLM, our sampling did not extend deep enough to observe the supported 210Pb level directly. However, because we did measure supported 210Pb in the nearby TAC core, collected from a site with similar environmental conditions compared to TLM, we were able to inform the TLM model using prior knowledge derived from TAC. Plum integrates this information within its Bayesian framework, allowing for a more robust estimate of supported 210Pb and resulting in a reliable chronology for TLM despite the data limitation.
We will revise the manuscript to include a clearer explanation of this modeling approach and how supported 210Pb was estimated for TLM.
Reviewer comment (section 3.4: LOI and ASI):
L173ff: I would suggest to stick with a simpler nomenclature for the LOI processing stages and to use dry weight (DW or DW105) and LOI550 for combusted samples. Also, I feel that drying peat samples at 105°C overnight appears a little short.
Authors response:
We can of course use the simpler, and more standard, nomenclature proposed by the reviewer.
The peat samples were dried overnight, usually from between 5-6 PM until 9-10 AM the next day, or about 15 to 17 hours. Heiri et al. (2001) recommend between 12-24 hours. We can change “overnight” by the approximate number of hours.
Reviewer comment (section 3.4: LOI and ASI):
L179: Unlike LOI, the ASI approach is not a very well established as a method and does require a more detailed description or more thorough and precise referencing. Did you consider the whole grain-size distribution of the samples or was there a significant amount of larger minerogenic particles present? It would be worthwhile to reference a bit more here and argue for the approach. Also, it reads here as if the fraction <125 µm is used as a proxy, but in the discussion, it is clearly stated that >125 µm is used. What did you use? Also, I would suggest to separate LOI and ASI, or change the title to “Variations in mineral content and ASI”. LOI is often used to identify the content of organic material in a sample, and I was a little confused that you went straight to the residual here.
Authors response:
We agree with the reviewer that this section would benefit from greater clarity and a more thorough justification of the ASI approach. Indeed, this comment echoes suggestions made by Reviewer 1 in that ASI and LOI should be treated separately in the results (see our response to Reviewer 1). We suggest:
- Having a dedicated Methods section for ASI, which would include a complete definition of ASI and additional references to support its use for paleo-storm records. These would include work from Orme et al. (2016), Kylander et al. (2020), Vandel et al. (2019), and the recent work from Vaasma et al. (2025). We would move some of the material from section 5.1.3 (Aeolian grain-size and associated wind speed) in this new section in order to justify our use of ASI earlier in the paper, and clearly state that the mineral fraction above 125 or 63 µm is used for the ASI calculations, not the fraction below these grain sizes.
Regarding whole grain-size distribution: we did not conduct a full grain-size distribution analysis on the peat minerals, because the very low mineral content in our samples made laser diffraction difficult, not only due to detection limits, but also because the limited amount of mineral material meant that the entire core would have been consumed in the process, preventing us from using it to calculate ASI. On average, the proportion of aeolian sand in the total mineral residue was about 5% at both TLM and TAC (excluding the basal sediment portion), while most of the remaining residue was dust-sized material (< 125 or 63 µm). We choose these grain-size fraction based on previous peat-based paleo-storm studies (e.g., Goslin et al. 2019; Vandel et al. 2019).
Reviewer comment (section 3.6.1: Data processing)
L192: Often (but not always) LOI correlates well with measured values of TOC based on element analysis, so arguably LOI is an absolute measure (with a bit of uncertainty and the occasional outlier). Would the results look much different would you skip the data processing as described?
Authors response:
We agree that LOI often correlates well with TOC from elemental analysis and can serve as an approximate absolute measure of organic content. In our case, the overall patterns and conclusions would not change substantially if we used untransformed LOI data.
However, we applied a log transformation to LOI primarily to align with the approach used by Bertrand et al. (2023), and to account for potential compositional (closed-sum) effects, as its sum-constrained nature can still introduce spurious correlations when used in multivariate analyses. The log transformation helps mitigate this issue and ensures that our statistical analyses (e.g., PCA, correlation, clustering) are more robust.
Reviewer comment (section 3.7: Paleo-storm identification):
L228: You mention the identification of (non-)aeolian processes here. Did you by any chance measure full grain-size distributions on the samples from the core or on a few of the samples from the bottom of the cores at least? This would be very useful to pinpoint the differences in sources and processes.
Authors response:
We did not measure full grain-size distributions on the core samples. Sample size was too limited for systematic grain-size analysis, particularly in the ombrotrophic sections where the mineral content is low (see our response to the previous comment).
We fully agree that full grain-size distributions would provide additional insights into sediment sources and processes, particularly in the basal, minerotrophic sections of the core. However, for the purposes of this paper, our focus is specifically on reconstructing storms from the ombrotrophic portion of the core, where aeolian input is dominant. This distinction is supported by our geochemical data, which allows us to identify where ombrotrophic conditions prevail. While grain-size distributions could help further characterize the processes and sediment sources, especially in the minerotrophic sections, and would indeed be valuable for extending the storm record deeper into the core, we believe this type of analysis lies beyond the scope of the current study.
Reviewer comment (section 4.1: Chronostratigraphy)
L261f: The sand layers at the bottom of the core are relatively thin. How can you be certain that they are a basal sand layer that marks the bottom of the peat sequence (and not just an interbed, e.g. from overwash)?
Authors response:
While the sand layers at the bottom of the cores are only a few centimeters thick, this is consistent with what is typically retrieved using a Russian corer, as the dense nature of mineral-rich sediments at the base often limits further penetration. Although we cannot entirely rule out the possibility of an interbedded deposit, several lines of evidence support our interpretation of these sands as basal layers marking the onset of peat accumulation.
First, these sand layers are positioned directly beneath peat strata that, in both cores, indicate the earliest phases of peatland development (possible lacustrine sediments at TLM and a wood-dominated fen at TAC). This stratigraphic context aligns with what we would expect at the transition from mineral to organic accumulation (for example, see Magnan (2014). Second, ecological succession patterns inferred from the macrofossil (peat type) record, combined with geochemical signatures, are consistent with a natural progression from a mineral-dominated environment to peatland establishment. Finally, the likelihood of these sands being overwash deposits is low, given the distance from any significant water body and the elevated position of the sites, which makes storm-induced overwash or marine incursions unlikely. Altogether, the sedimentological, ecological, and geomorphological evidence supports our interpretation of these layers as basal sands marking the initiation of the peat sequence.
Reviewer comment (section 4.1: Chronostratigraphy):
L265: Arguably, the marked drop in PAR could be interpreted as a hiatus rather than a stage of very slow accumulation (which slightly overstretches the ability of the 14C chronology). Also, there is a lot happening in the core at about the same time (or slightly thereafter). The peat looks much brighter in the core images and the ASI picks up up-core from here. Wouldn’t it be advisable to add an uncertainty to the rates and to subdivide a little more, especially in the case of TAC?
Authors response:
We agree that the marked drop in PAR could reflect a hiatus rather than a stage of slow accumulation. In its current form, our text privileges the slow accumulation interpretation without fully acknowledging the uncertainty around this interval. We will revise the wording to explicitly recognize the potential for a hiatus and add a clearer uncertainty statement.
While our current subdivision is based on a CONISS analysis and works well within the context of establishing broad periods in our peat cores, we can describe the changes happening in Zone 3 of TAC in more details in section 4.1. For instance: the lower boundary of the potential hiatus begins shortly after ~970 BCE, where the previously steady age-depth relationship begins to shift, and PAR values start to decline. The PAR continues to drop until the next radiocarbon date at 470 BCE (147.5 cm), which coincides with the base of a thick minerogenic layer. The following date at 134.5 cm is from just above this layer and is calibrated to 970 CE—suggesting a ~1500-year interval of minimal or no accumulation. This sequence suggests a hiatus in peat accumulation between ~1000 BCE and ~1000 CE, with a particularly low accumulation rate (~0.003 cm/yr) from 970–470 BCE, dropping further (~0.001 cm/yr) between ~400 BCE and 970 CE. Peat accumulation resumes abruptly around 1000 CE at a rate comparable to pre-hiatus levels.
Reviewer comment (section 4.1: Chronostratigraphy):
L272: Figure 2. I do not find the top panels above the age-depth models particularly informative (also, the last bit of text to the right is cut off) and you should consider to remove them. It would be helpful for the readers understanding to have a table of the radiocarbon ages with uncalibrated and calibrated ages (ideally as ranges), the 2-sigma probabilities, and info on the material dated and possibly the lithology of the unit dated. (NB: Sorry! I just found the table in the appendix – it could be useful to have this integrated into the main text). While the age model for TLM looks fairly straight forward, there is a tricky hiatus (or continued deposition followed by erosion?) at 150 cm. Would there be a good way (visually) to scale the y-axes the same way? The fact that the oldest peat deposits have a surprisingly similar age is easily overlooked!
Authors response:
We can move the full age-depth model figure with the top panels to the Appendix, and move table C1 from the Appendix to section 4.1. We can also scale the y-axis of TAC, so that it fits the scale of TLM (from 2000 CE to 6000 BCE), to make visual comparisons easier.
Reviewer comment (section 4.2: Downcore variations in lithology):
L280ff: I feel it could be sensible to exclude the ASI (as a separate proxy) from the description of the lithology, which is (mostly) descriptive otherwise.
Authors response:
This was pointed out by Reviewer 1. We suggest treating ASI in a separate section. See our response to Reviewer 1 under “Interpretation and reliance on Ti”.
Reviewer comment (section 4.2: Downcore variations in lithology):
L300: Figure 3. I find it difficult to compare the sites visually. Would it be an option to display the data scaled for age and displayed with a linear axis (even distance of the tick-marks between e.g. centuries)? In a similar way, the values for ASI and mineral content should be scaled the same way and you could consider to use a log scale to make the small variations more visible. Some of the labels are not entirely visible in the current version (e.g., Si, ASI).
Authors response:
We appreciate the reviewer’s thoughtful suggestion regarding the visualization of Figure 3. While we understand the value of age-scaled plots for temporal comparison, we have chosen to present this particular figure on a depth scale for the following reasons:
Our goal with Figure 3 is to provide a descriptive overview of the core stratigraphy and compositional changes in relation to depth, which reflects the physical structure of the sediment archive. This is especially important at this stage of the manuscript, where we introduce the characteristics of each core. Plotting by depth allows for a direct visual connection to the sedimentary column itself, which is how the cores were collected, logged, and sampled in the field.
We acknowledge that trends over time are also important, and we address these more explicitly in subsequent figures 5, 7, and 8, where data are plotted by modeled age with evenly scaled time axes to facilitate inter-core comparison and trend analysis.
Regarding the comparison of ASI and mineral content values across cores, we agree that consistent scaling will improve readability, and we will harmonize the axis scales accordingly. A log scale is already used in the current figure (highlighted in grey) to emphasize subtle variations; we will make this more explicit in the figure caption and will add a second x-axis to indicate the log-transformed values clearly.
Reviewer comment (section 4.4: Paleo-storm reconstructions):
L347: I am a little surprised, that zone 3 of the fore from TAC was not excluded. The hiatus in the record (and the lack of information to fully explain it) is a bit puzzling, but if your storm proxy is the input of allochthonous material into the peat bog, a hiatus can not be ignored, as the entire zone 3 is potentially the time-integrated results of almost 2kyr of all kinds of processes (limited plant growth, a higher ground water table, flooding with marine water, etc.) potentially paired with a near constant influx of sand.
Authors response:
We recognize the uncertainties associated with Zone 3 of the TAC core, including the apparent hiatus and the challenges it presents for interpretation. Nonetheless, we chose to retain this section in our reconstruction for several reasons.
First, the Ti and ASI patterns in TAC Zone 3 mirror remarkably well the storm-event frequency reconstructed from TLM, suggesting that these patterns may still reflect climatically driven aeolian inputs, even if the peat accumulation in TAC was discontinuous or altered by decomposition. The strong correspondence between sites supports the idea that Zone 3 captures a meaningful aeolian signal, potentially time-integrated, but informative.
While we cannot definitively confirm ombrotrophic conditions during this interval, it remains plausible that the apparent hiatus reflects a period of reduced or altered peat accumulation, perhaps due to decomposition rather than a complete break in deposition. In this scenario, aeolian sand may still have continued to accumulate at the surface and reflect storm-related inputs. We recognize the limitations of interpreting such intervals but feel that omitting this section could downplay potentially meaningful regional patterns observed across both cores.
In line with our response to Reviewer 1 (“Exclusion of part of the TAC record”), we believe it is helpful to retain this section while clearly indicating its uncertainties. By shading the relevant interval (as already done in Figure 5 and now also added to Figure 8), we aim to support transparent interpretation. Rather than excluding this portion of the record entirely, our intention is to provide the available data with appropriate context so that readers can consider all the evidence.
Reviewer comment (section 5.1: Key similarities and differences between TLM and TAC):
L383: There is a very nice match between the two reconstructed event densities for the 880-550 BCE window, that unfortunately falls into the hiatus at TAC (and hence in a time window, where the age model may be highly unreliable. It is very interesting regardless, but I really feel that you should discuss this more. For the 600-800 CE window, I don’t see a way the age model can be trusted for TAC. The phase is nicely visible in the Ti record for TLM, but bare visible in the ASI record.
Authors response:
We agree that the 880–550 BCE interval in TAC falls near the onset of the hiatus and may carry some uncertainty. However, this section is bounded by two radiocarbon dates (970 [1080–840] BCE and 475 [630–395] BCE), which we think provides a reasonable level of chronological control. While we acknowledge that the 600–800 CE interval in TAC is less well constrained and relies primarily on a single upper date (945 CE [775–1040 CE]), the increased uncertainty in this section is reflected in our figures, where age ranges for each potential event are indicated. We will expand our discussion in this section to better clarify the limitations of the age model, particularly around the hiatus, and emphasize the importance of interpreting those intervals with caution.
Reviewer comment (section 5.1.2: Aeolian sediment sources):
L430: Not sure, I can follow the argument. Earlier you argued that the XRF values are relative and (unless I missed it) no absolute values were measured. In the data displayed in Figure 3, I do not see an elevated Fe content in the TAC core. The highest relative values are in fact found in the bottom section of core TLM.
Authors responses:
Thank you for pointing this out. As currently presented, the relative relationships between elements in Figure 3 are indeed difficult to interpret due to the smoothing effect of the CLR transformation. To clarify, we plan to display and interpret the geochemical data in Figure 3 as elemental ratios over the sum of lithogenic elements (as outlined in our response to Reviewer 1 under “XRF Data Processing and Normalization”). This approach is intended to better highlight relative changes in individual elements while accounting for shifts in overall sediment composition. While the CLR-transformed data remain essential for statistical analysis (the PCA and correlations), displaying elemental ratios should make trends — such as elevated Fe near the top of TAC — more clearly visible. We will make sure this change is clearly explained in both the main text and the figure caption.
Reviewer comment (section 5.1.2: Aeolian sediment sources):
L430: A depletion in minerals other than quartz is not a necessity and highly dependent on e.g., local properties (lithology of the source rock), the sediment budget of a site (and hence the time of exposure in the system prior to deposition), and transport distance. In order to make this understandable for all readers, the local properties need to be described more.
Authors response:
Additional details will be incorporated into Sections 2.1 and 5.1.2 to clarify the local factors influencing sediment composition, such as sediment pathways, sources, and physical characteristics, to better contextualize our interpretation of the higher abundance of hematite-coated quartz at TAC compared to TLM.
At Beach du Cap (the beach NE of TLM), sediments likely originate from the Butte de La Croix, a nearby (2-3 km SE of the beach) rocky cliff undergoing rapid erosion (up to 1 m / year according to Bernatchez et al. 2012), with a main longshore drift going from the Butte towards Beach du Cap. Similarly, at TAC, the sandstone cliffs - also part of an actively receding rocky shoreline - supply sediments to nearby accreting beaches such as du Bassin and Anse à la Cabane (Bernatchez, 2012). These areas lie within the same geomorphological unit composed primarily of shale, limestone, and sandstone (Owens & McCann, 1980).
Outcrops in these areas are described in detail in Remillard et al. (2013). At TAC, the cliff stratigraphy includes a basal unit of white, well-rounded sand, overlain by a 30 cm organic-rich layer dated to 47–50 ka, then by colluvium, and finally by a 1–4 m thick reddish (hematite coating) silt-clay/sand diamict known as the Drift des Demoiselles (Dredge et al., 1992). The cliffs near TLM are composed solely of this unit, reaching up to 6 m in thickness (Remillard et al. 2013). This diamict contains ~80% sandstone-sized particles (60-200 microns) (with hematite coating) predominantly made up of siliciclastic minerals (quartz and chert), as well as siltstone and mudstone.
At both locations, the sand-size sediment source is similar, originating either from relic coastal sand or the Drift des Demoiselles diamict. At TAC, this highly erodible sediment layer lies closer in both elevation and distance to the bog than the nearby beaches, making it a likely primary sediment source for the bog. In contrast, the cliffs southeast of TLM are more distant and less extensive; here the closer and more extensive beaches likely act as the primary sediment source.
While, to our knowledge, no formal sediment budget has been established for the Magdalen Island beaches, the dominant sediment source - reddish sandstones – supports the interpretation that wave action during nearshore transport erodes iron oxide coatings before deposition at the beach and eventually into the bog. These local conditions help explain the dominance of hematite-coated quartz in the upper portion of TAC (assumed through high Fe content), and the absence of a strong Fe signal in the upper portion of TLM.
Reviewer comment (section 5.1.2: Aeolian sediment source):
L433: It would be useful to show the correlation between the two different (!) ASI values and elements in the manuscript or the supplement.
Authors response:
The correlations between the two ASI values and geochemical elements are included in Appendix E. To improve clarity, we will expand the discussion of these correlations in the results section, particularly focusing on the relationship between ASI and geochemistry in the ombrotrophic section of the cores, as addressed in our response to reviewer 1 under “Interpretation and reliance on Ti.”
Reviewer comment (section 5.1.3: Aeolian grain-size and associated wind speed):
L441: "relatively unconsolidated" is difficult to understand as a description. If the sources allow for transport of sediment to a distal site during storms, it would be interesting to know what happens at a more proximal location. Did you observe the formation of primary dunes on the cliff top or the beaches and what are the grain-size properties here? In other words, what happens in the source areas during other conditions than storms?
Authors response:
We agree that the phrase “relatively unconsolidated” is vague. We will revise the text to more clearly refer to “loose beach sand” and “easily erodible sandstone cliffs” as the main sediment sources.
Regarding the formation of primary dunes: the coastal dune system of the Magdalen Islands is bordered throughout by an active foredune system, supported by abundant beach and intertidal sand (Morin, 2001). These dunes vary in morphology depending on sediment supply. In areas of high accumulation, complex dune ridge systems have developed, with two notable examples on Havre-Aubert Island in the south (near Plage du Bassin, close to TAC) and in the northeast, near TLM, as previously noted by the Reviewer. These systems are described in more details in our response above to comment L131ff.
Most beaches are bordered by a single foredune, which remains active due to a more limited sand supply compared to ridge systems. Storm waves frequently erode the foredune, forming dune scarps that expose sand to wind. This sand is often carried inland and deposited in back-dune environments (Morin, 2001). In these cases, common on the southern and eastern coasts of Havre-Aubert Island, the dunes tend to migrate landward.
Morin (2001) also documented a decrease in mean grain-size from the intertidal zone to the foredune crest. At the top of the foredune, mean grain size reached 35 µm, reflecting the lower transport capacity of wind relative to water and highlighting the aeolian character of foredune formation.
Storm-driven overwash and dune breaching are also significant processes. Satellite imagery and previous reports show clear evidence of overwash lobes, particularly in the back-barrier saltmarshes between the main islands. A government report (Bernatchez & Remillard, 2019) documented the impacts of four major storms in November 2018. At Plage du Bassin, a likely sediment source for both TAC and TLM, dunes retreated by an average of 2.82 meters, with multiple breaches and overwash deposits observed. The frontal dune experienced severe erosion and vegetation loss, indicating a highly dynamic storm response. In contrast, dune breaching is rare in prograding areas like the ridge system near TLM or in zones of high sediment supply, such as Sandy Hook.
In summary, dune systems near our sites are very active and usually have positive or constant sediment budget; where it is very high, it creates prograding systems (ridges); where it is more limited, dunes tend to migrate landward, with evidence of aeolian transport and morphology landforms in these areas. We can include a discussion regarding dune dynamics in the Site description and where relevant in the Discussion.
Reviewer comment (section 5.1.3: Aeolian grain-size and associated wind speed):
L449ff: It would be useful to know more about the typical wind directions during these different storm events. This does have a huge a huge impact on how the sources and your storminess archives connect.
Authors response:
Prevailing winds in the Gulf of St. Lawrence and around the Magdalen Islands are predominantly westerlies, especially from the northwest. These contribute to more pronounced coastal erosion and narrower beaches on the islands’ western shores (narrower beaches), while the eastern coasts remain relatively sediment-rich (Forbes et al., 2004; Morin, 2001). In contrast, post-tropical cyclones often bring strong northeasterly winds to the region. Overall, wind directions during storm events are highly variable and strongly dependent on storm tracks, influencing the transport and deposition of aeolian sediments from multiple potential source areas.
Reviewer comment (section 5.2 Modern storm attribution):
L448ff: This is a very interesting section and you are really working at the limits of the ASI approach here. You are considering a time window of ca. 170 years with an averaged accumulation rate on the order of 3 mm year (much less for TLM based on the age model shown in Fig. C1). It would be really interesting to hear your opinion here: Is it at all possible to use the mineral content of peat to get to a storminess reconstruction (ideally with the attribution of single events) for these time scales? Have you considered to lump the data (e.g., to decades)?
Authors response:
We thank the reviewer for this insightful comment. As noted, we are indeed working at the limits of the ASI approach. Most peat-based reconstructions avoid modern storm attribution, particularly in Europe, where a long history of peat harvesting can compromise recent records, such as in Kylander et al. (2023). However, our study area has not been affected by peat extraction, which provided a unique opportunity to explore both the potential and limitations of storm attribution in recent peat layers. This analysis is intended as a preliminary step toward a broader investigation of this approach in an upcoming paper.
We believe that mineral content in peat can be used to reconstruct storminess, and potentially even to attribute individual events, on an annual scale, particularly in ombrotrophic coastal settings. This assumption is supported by Vandel et al. (2019), who applied ASI to four well-dated short coastal peat cores in Estonia and found positive correlations between ASI and annual instrumental storminess data. Similarly, Allen et al. (2021) demonstrated a strong correspondence between plastic concentrations in peat and historical trends in European plastic production and consumption, further validating the use of peat for high-resolution environmental reconstructions.
While neither study attempted single-event attribution, we believe this remains possible in principle. Similar event-scale attributions have been achieved in other continuously accumulating environments, such as blue holes (e.g., Winkler et al. (2022)). Although single-event attribution is not the central focus of our paper, our study lays the groundwork for a future investigation specifically targeting this question.
That said, achieving annual or sub-annual resolution requires high-resolution proxies. ASI at one cm intervals is likely too coarse to resolve individual events reliably, especially deeper in the core, where annual layers are thinner than 1 cm. In these cases, data aggregation (e.g., decadal averaging) may be more appropriate, as the reviewer suggested. However, our use of Ti data at one mm resolution opens the door to investigating annual or even finer-scale variations, particularly in the upper sections of the core.
Reviewer comment (section 5.2: Modern storm attribution):
L462: This is news! My understanding up to this point was, that the beaches NE of TLM would be the major source of sediment to the site or those immediately to the south.
Authors response:
For this specific comment, we are not entirely certain which part is being referred to as news by the reviewer. If the reviewer is referring to our mention of the gravel roads as a potential recent sediment source, we agree that this point could be better introduced earlier in the manuscript, particularly in the site setting section. If the reviewer was referring to a different part of the text, we would welcome further clarification to address the comment more directly.
Reviewer comment (section 5.3-5.4: Western North Atlantic Basin comparison):
L507-608: This section is very important, but I struggle a bit with three things here: (1) The storminess reconstruction stands a bit on shaky grounds for a couple of reasons (understandable reasons, not necessarily flaws!), but the identified storm periods are here taken as fact and I miss the careful arguing a bit, (2) Your results are placed in general climate context here, but while reading through this I repeatedly wondered, if the purpose is to argue for the validity of ASI (and your age model for TLM) or to show the added value of your records, and (3) Several different scales are naturally mixed here and your hypothesis is stuck between the coast of West Africa and distance relationships between source and sink in sediment transport. To summarize, I like the section overall, but I am not sure of its purpose.
Authors response:
We agree that the storminess reconstruction, while promising, rests on several assumptions and methodological limitations that require clearer discussion. As noted throughout our responses to both reviewers, we will revise earlier sections of the manuscript to more explicitly acknowledge these uncertainties, particularly regarding the ASI proxy, Ti-based interpretations, and the age models. This should provide stronger foundations for Sections 5.3 and 5.4 and help them stand more independently.
Regarding the purpose of Section 5.3, our main goal is to demonstrate the added value of our records by comparing them to other storm reconstructions from the western North Atlantic, specifically from the northern Bahamas, northeastern USA, and eastern Canada. Despite the challenges and limitations of our approach, we find consistent storminess patterns across these regions, which supports the broader relevance of our peat-based archive.
Section 5.4 builds on this comparison by situating the shared storminess trends within a broader climatic context. Our intention is not to re-validate the ASI proxy or the TLM chronology here, but rather to engage with ongoing discussions about large-scale drivers of storm activity and highlight how our record contributes to that dialogue. We will revise the text to sharpen the focus of our arguments. In doing so, we aim to clarify the purpose of each section and streamline the narrative to better reflect the unique contribution of our study.
REFERENCE LIST
Allen, D., Allen, S., Le Roux, G., Simonneau, A., Galop, D., & Phoenix, V. R. (2021, Nov 9). Temporal Archive of Atmospheric Microplastic Deposition Presented in Ombrotrophic Peat. Environ Sci Technol Lett, 8(11), 954-960. https://doi.org/10.1021/acs.estlett.1c00697
Aquino-López, M. A., Blaauw, M., Christen, J. A., & Sanderson, N. K. (2018). Bayesian Analysis of 210-Pb Dating. Journal of Agricultural, Biological and Environmental Statistics, 23(3), 317-333.
Barnett, R., Bernatchez, P., Garneau, M., & Juneau, M. N. (2017). Reconstructing late Holocene relative sea‐level changes at the Magdalen Islands (Gulf of St. Lawrence, Canada) using multi‐proxy analyses. Journal of Quaternary Science, 32(3), 380-395. https://doi.org/https://doi.org/10.1002/jqs.2931
Bernatchez, P., Drejza, S., & Dugas, S. (2012). Marges de sécurité en érosion côtière: évolution historique et future du littoral des îles de la Madeleine. Laboratoire de dynamique et de gestion intégrée des zones côtières, Université du Québec à Rimouski. Rapport remis au ministère de la Sécurité publique du Québec.
Bernatchez, P., & Remillard, A. M. (2019). Documentation des impacts de l’érosion côtière dans les côtes naturelles et anthropisée les plus touchées par les tempêtes de l’automne 2018 aux Îles-de-la-Madeleine. https://ldgizc.uqar.ca/Web/docs/default-source/default-document-library/idlm2018_rapport_temp%C3%AAte.pdf
Bertrand, S., Tjallingii, R., Kylander, M. E., Wilhelm, B., Roberts, S. J., Arnaud, F., Brown, E., & Bindler, R. (2023). Inorganic geochemistry of lake sediments: A review of analytical techniques and guidelines for data interpretation. Earth-Science Reviews, 104639. https://doi.org/https://doi.org/10.1016/j.earscirev.2023.104639
Billy, J., Robin, N., Certain, R., Hein, C., & Berné, S. (2013). Barrier shoreline evolution constrained by shoreface sediment reservoir and substrate control: The Miquelon-Langlade Barrier, NW Atlantic. Journal of Coastal Research(65 (10065)), 2089-2094. https://doi.org/10.2112/si65-353.1
Billy, J., Robin, N., Hein, C., FitzGerald, D., & Certain, R. (2018). Dominance of Inherited Geologic Framework on the Development of Coastal Barrier System. Journal of Coastal Research(85 (10085)), 406-410. https://doi.org/10.2112/si85-082.1
Billy, J., Robin, N., Hein, C. J., Certain, R., & FitzGerald, D. M. (2014, 2014/11/01/). Internal architecture of mixed sand-and-gravel beach ridges: Miquelon-Langlade Barrier, NW Atlantic. Marine Geology, 357, 53-71. https://doi.org/https://doi.org/10.1016/j.margeo.2014.07.011
Blaauw, M., Christen, J. A., Bennett, K. D., & Reimer, P. J. (2018, 2018/05/15/). Double the dates and go for Bayes — Impacts of model choice, dating density and quality on chronologies. Quaternary Science Reviews, 188, 58-66. https://doi.org/https://doi.org/10.1016/j.quascirev.2018.03.032
Cwanek, A., Aquino-Lopez, M. A., Kołaczek, P., Lamentowicz, M., Fiałkiewicz-Kozieł, B., Gałka, M., Słowiński, M., Łuców, D., Marcisz, K., Obremska, M., Czerwiński, S., & Łokas, E. (2025, 2025/02/01/). Strengthening potential of recent peat dating. Journal of Environmental Radioactivity, 282, 107594. https://doi.org/https://doi.org/10.1016/j.jenvrad.2024.107594
Dredge, L. A., Mott, R. J., & Grant, D. R. (1992). Quaternary stratigraphy, paleoecology, and glacial geology, Îles de la Madeleine, Quebec. Canadian Journal of Earth Sciences, 29(9), 1981-1996. https://doi.org/https://doi.org/10.1139/e92-154
Forbes, D. L., Parkes, G. S., Manson, G. K., & Ketch, L. A. (2004). Storms and shoreline retreat in the southern Gulf of St. Lawrence. Marine Geology, 210(1-4), 169-204. https://doi.org/https://doi.org/10.1016/j.margeo.2004.05.009
Heiri, O., Lotter, A. F., & Lemcke, G. (2001). Loss on ignition as a method for estimating organic and carbonate content in sediments: reproducibility and comparability of results. Journal of Paleolimnology, 25(1), 101-110. https://doi.org/https://doi.org/10.1023/A:1008119611481
Hétu, B., Rémillard, A. M., Bernatchez, P., & St-Onge, G. (2020). Landforms and Landscapes of the Magdalen Islands: The Role of Geology and Climate. In O. Slaymaker & N. Catto (Eds.), Landscapes and Landforms of Eastern Canada. World Geomorphological Landscapes (pp. 431-451). Springer, Cham. https://doi.org/https://doi.org/10.1007/978-3-030-35137-3_19
Kylander, M. E., Martínez-Cortizas, A., Sjöström, J. K., Gåling, J., Gyllencreutz, R., Bindler, R., Alexanderson, H., Schenk, F., Reinardy, B. T., & Chandler, B. M. (2023). Storm chasing: Tracking Holocene storminess in southern Sweden using mineral proxies from inland and coastal peat bogs. Quaternary Science Reviews, 299, 107854. https://doi.org/https://doi.org/10.1016/j.quascirev.2022.107854
Kylander, M. E., Söderlindh, J., Schenk, F., Gyllencreutz, R., Rydberg, J., Bindler, R., Martínez Cortizas, A., & Skelton, A. (2020). It's in your glass: a history of sea level and storminess from the Laphroaig bog, Islay (southwestern Scotland). Boreas, 49(1), 152-167. https://doi.org/https://doi.org/10.1111/bor.12409
Magnan, G. G., M. (2014). Evaluating long-term regional climate variability in the maritime region of the St. Lawrence North Shore (eastern Canada) using a multi-site comparison of peat-based paleohydrological records. Journal of Quaternary Science, 29(3), 209-220.
Morin, I. (2001). Geomorphologie et evolution du systeme dunaire des iles de la Madeleine, Quebec. Universite Laval.
Orme, L. C., Reinhardt, L., Jones, R. T., Charman, D. J., Barkwith, A., & Ellis, M. A. (2016). Aeolian sediment reconstructions from the Scottish Outer Hebrides: Late Holocene storminess and the role of the North Atlantic Oscillation. Quaternary Science Reviews, 132, 15-25.
Owens, E., & McCann, S. (1980). The coastal geomorphology of the Magdalen Islands, Québec. The Coastline of Canada. Geological Survey of Canada, Paper, 80-10.
Rémillard, A. M., Buylaert, J.-P., Murray, A., St-Onge, G., Bernatchez, P., & Hétu, B. (2015). Quartz OSL dating of late Holocene beach ridges from the Magdalen Islands (Quebec, Canada). Quaternary Geochronology, 30, 264-269. https://doi.org/https://doi.org/10.1016/j.quageo.2015.03.013
Rémillard, A. M., Hétu, B., Bernatchez, P., & Bertran, P. (2013). The Drift des Demoiselles on the Magdalen Islands (Québec, Canada): sedimentological and micromorphological evidence of a Late Wisconsinan glacial diamict. Canadian Journal of Earth Sciences, 50(5), 545-563. https://doi.org/10.1139/cjes-2011-0115
Rémillard, A. M., St-Onge, G., Bernatchez, P., Hétu, B., Buylaert, J.-P., Murray, A. S., & Lajeunesse, P. (2017). Relative sea-level changes and glacio-isostatic adjustment on the Magdalen Islands archipelago (Atlantic Canada) from MIS 5 to the late Holocene. Quaternary Science Reviews, 171, 216-233. https://doi.org/https://doi.org/10.1016/j.quascirev.2017.07.015
Vaasma, T., Vandel, E., Sugita, S., Tõnisson, H., Suursaar, Ü., Kont, A., & Vilumaa, K. (2025). Storminess reconstruction in the northeastern Baltic Sea region over the past 7600 years based on aeolian sand influx into coastal bogs. The Holocene, 35(1), 61-74. https://doi.org/10.1177/09596836241285783
Vandel, E., Vaasma, T., Sugita, S., Tõnisson, H., Jaagus, J., Vilumaa, K., Anderson, A., & Kont, A. (2019). Reconstruction of past storminess: Evaluation of an indicator approach using aeolian mineral grains buried in peat deposits, Estonia. Quaternary Science Reviews, 218, 215-227. https://doi.org/https://doi.org/10.1016/j.quascirev.2019.06.026
Winkler, T. S., van Hengstum, P. J., Donnelly, J. P., Wallace, E. J., D'Entremont, N., Hawkes, A. D., Maio, C. V., Sullivan, R. M., & Woodruff, J. D. (2022). Oceanic passage of hurricanes across Cay Sal Bank in The Bahamas over the last 530 years. Marine Geology, 443, 106653.
Citation: https://doi.org/10.5194/egusphere-2025-400-AC1
-
AC1: 'Reply on RC2', Antoine Lachance, 12 May 2025
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 285 | 262 | 20 | 567 | 18 | 28 |
- HTML: 285
- PDF: 262
- XML: 20
- Total: 567
- BibTeX: 18
- EndNote: 28
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1