the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 07 Jul 2026
Rain, wind or fire: which led the PAH fallout in Northern Madagascar across the Holocene?
Abstract. The interactions between wildfire, human and climate variability are not well understood. To have better predictions, past studies are useful for observing their behaviours on a long-time scale. To infer past fire activity, studies used charcoal and molecular biomarkers derived from biomass burning, such as Polycyclic Aromatic Hydrocarbons (PAHs). In the north of Madagascar, charcoals are of local origin and have shown increased abundance since 1000 cal a BP. In contrast, PAH are present throughout the Holocene. Therefore, PAHs likely provide a more regional information. Due to the seasonal wind regime, they are thought to originate from East Africa. However, the PAH signal pattern differs from the fire activity record of East Africa. Instead, it shows a similar pattern to precipitation variability in northern Madagascar. Higher PAH concentrations are observed during wetter period, whereas lower concentrations occur during drier periods. This suggest that PAH preservation is enhanced, which may facilitate leaching processes prior to their deposition in sediments. These results challenge the direct interpretation of PAH as straightforward indicators of fire events and dry periods.
- Preprint
(1470 KB) - Metadata XML
-
Supplement
(217 KB) - BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2026-3726', Anonymous Referee #1, 05 Aug 2026
-
RC2: 'Comment on egusphere-2026-3726', Anonymous Referee #2, 24 Aug 2026
Review of manuscript: egusphere-2026-3726 for Climate of the Past
Bellet et al present a new record of polycyclic aromatic hydrocarbon (PAH) concentrations––a widely used biomarker for past fire activity––from a sediment core from Lake Maudit, Northern Madagascar which spans the Holocene. By comparing their PAH record with previous estimates of fire activity in Eastern Africa and local precipitation, they conclude that PAHs are sourced from fires in Eastern Africa but their presence in Lake Maudit is not sensitive to fire amount in the source region, but instead to local rainfall conditions and their impact on wet scavenging of these compounds out of the atmosphere.
This work is well written, logically organized, contains well-made figures, and is an important new study that will provide meaningful information to understanding the past fire activity in Macassar and Eastern Africa. However, I have major issues with the arguments presented in this paper and do not think the conclusions follow from the data. However, I think with major revisions, namely a re-interpretation of the main conclusion, this record will be an important contribution to the field and worthy of prompt publication in Climate of the Past. Therefore, I recommend major revisions.
My major concern revolves around the authors conclusion that wet deposition rather than fire amount in the source region is the main driver of PAHs in their lake sediment core. Outlined below are my overarching comments followed by in-line comments and suggestions.
Major concerns:First a methodological concern. Concentrations of paleo-proxies in a sediment core are inherently confounded by changes in sediment accumulation that can bias the signal (for example see Marlon et al. 2013, QSR). The gold standard is to instead report the influx of a chemical or physical constituent of the sediment using the mass accumulation rate determined by the age model of the sediment core, which mitigates the impact of accumulation rate change effects on concentration. At a minimum, the authors should determine the accumulation rate of the PAHs instead of the concentration as this will provide a more appropriate comparison to regional fire activity.
Second, the authors appear to flip-flop on when PAH concentration in their core is driven by wet deposition vs increased fire activity. They claim that high PAHs early in the Holocene are the result of higher wet deposition, but when PAHs also increase in the last few thousand years, despite no accompanying rise in rainfall, they attribute that to higher fire activity. They can’t have it both ways. Moreover, they gloss over a growing body of evidence that changes in fire activity in grassy ecosystems is dominantly controlled by rainfall via controls on fuel abundance {refs}. Their claim that wet deposition is driving the signal needs to supersede the existing literature showing that rainfall increases lead to higher fires in grassy regions (e.g., Eastern Africa which the authors (I believe) rightfully identify as the source area of the PAHs). I feel like the authors set up a strong argument for PAHs in this Malagasy lake being sensitive to fire activity in Eastern Africa. A very interesting and useful result but instead pivot toward the end of the paper into a poorly supported alternative hypothesis, the reason for which is unclear to me.
Third, the authors need to improve the description of the modern conditions (e.g., add fire and compare the wind fields and potential transport from source to sink of PAHs to show that the timing of fire and transport align in the potential sources they discuss) and bolster the data they compare their results to, e.g., including all of the fire records they discuss in the text and improve the labeling of figures. A description of the current fire regime of Madagascar and its relation to moisture and fuel as well as connecting the fire regimes of the potential source areas to Lake Maudit via wind direction comparison would greatly improve the background the reader needs to understand the paper.
Fourth, there are multiple mislabelings of hydrophobic vs hydrophilic PAHs that make some of the arguments difficult to parse. In revision, the authors should much more carefully make sure they are using the correct terms in each instance otherwise, their argument requires undue inspection by the reader.
In-line comments:
18: “past studies”, I think you mean “paleo studies”, as written implies studies done in the past
32: This is a large oversimplification; we know that fires are differentially changing particularly across ecosystem type (in opposite directions). See Andela et al., 2017, Science
48-50: PAHs used to study human impacts on fires, not as biomarkers of human impacts generally
69-70: I don’t think this is really the central question of the paper. The wet deposition impact should be considered as a caveat, it could have a role in determining PAH accumulation in lakes, but existing literature also shows that they correspond to historical fires (e.g., Vachula et al. 2022, Paleo3). Your evidence that the PAHs correspond to Eastern African fire activity is stronger than the evidence for rainfall control, so I disagree with this framing
74-76: Do these studies actually do a frequency analysis (e.g., Higuera et al., 2007, QSR)? A rise in grass combustion does not imply more frequent fires, just more burning (which could be higher burned area). You need to be very careful when discussing fire activity vs amount vs frequency as all have related but distinctly different meanings.
84-95: This paragraph and the previous to some extent read as discussion-like. The introduction should lead up to your central question, but you appear to do that mid-way through leading to (as previously discussed) a question that I do not think should be the central point of the paper. This warrants reframing and reorganizing.
101-115: The potential fire sources are a large component of this paper, yet in the region description you only discuss the winds, temperature and rainfall. Why not the modern distribution/seasonality of fire? They claim later that the wind regimes exclude the southern portion of Madagascar as a PAH source for lake Maudit, so it would be good to set this up. Moreover, how does the Eastern African fire seasonality correspond with the timing of northwesterly winds? Setting up these potential sources at the beginning here would make your job in the discussion a lot easier. Same goes for the Equatorial Asia and Australia argument. I think that you may be able to just skip that because it is so unlikely to be the dominant source due to its distance but showing that the winds and fire timing align is the baseline to set up if that source even makes sense to consider.
133: Figure 1D??
136: “East Africa”: It is my understanding that "Eastern" is preferred due to the colonial association with "East Africa". This is not a science comment, just a cultural one, so make of that what you will.
144: “supports” should be “supported”
145-147: What is the evidence of the fire location?
157: RStudio is an R editor, you need to credit the R language not RStudio here. And it appears you used CLAM not BACON, so it’s a little concerning you are citing the wrong software here.
178: correct French spelling of “series”
205-206: comprised is not the correct word here, and the sentence just stops in the middle??
208-209: More information here? What did the blanks look like? What was removed? This is really vague.
211-212: I don’t know what this means, do you mean the extraction efficiency? How does that compensate for evaporative loss, which happens post-extraction?
223-224: Where are the exclusion criteria discussed? Why do you choose the dates that you kept? Are their physical indicators of erosion, and if so is it safe to assume the ages can be linearly interpolated across this disturbed interval?
239: Linearly interpolated age depth model
271: Figure 3 requires panel labels and reference to where the reader can find the PAH abbreviation key
283: What regions?
286: “Perylen” should be “Perylene”
307-308: What size class of charcoal is this? And what do you mean by "and Poaceae?" in the fig caption. Is this a true charcoal count or char:pollen ratio? If this is a macro charcoal count, it is not surprising that it will not correspond with PAHs becasue larger charcoal has a more local source. If this is a char:pollen ratio however, this does not provide direct fire information but instead a mixture of burning relative to grassy fuel availably.
319-320: Presence of charcoal does not imply fire frequency, only occurrence. You cannot separate frequency from burned area with charcoal amount. To get frequency you have to look at a peak vs background analysis (see Higuera et al., 2007, QSR).
337-338: “We suppose… variation”. This similarity runs counter to your deposition control argument. If wet deposition rather than production via fires is driving the signal, you would expect a stronger rainfall/PAH connection between the hydrophilic vs hydrophobic PAHs.
337-339: You are mixing up the hydrophobic/hydrophilic, hydrophobic = heavy, particle bound. This is mislabeled in figure 4.
343-345: “Secondly… Holocene.” What is this sentence saying? You should just state what each record shows and how it relates to yours. Also, what is the “cf” in the first line?
346-347: “In the southeastern… Holocene.” Why isn’t this plotted in figure 4 as well? This is important to your argument.
350-351: See my major comment, this needs to be set up earlier in the manuscript.
3543-355: “distinctive wind pattern” You can really drive this home with a supp figure showing the wind field during peak Eastern African fire season.
365: “Lake Tanganika and Lake Victoria” You are comparing your record to these records of fire in the source region you identify for the PAHs, why not include these in the comparison plot (Fig 4), being able to visually compare these records is critical to your readers to make the comparison and evaluate your claim that they are "markedly distinct" from your record.
367-368: Delete sentence and just start next sentence with "In contrast..."
368: Replace “huge” with “high”
369: “until 11,000 cal a BP” This mischaracterizes the data, they show elevated fire activity in Turkana from ~15-6ka.
370: Karp et al. show that fire declined with local declines in rainfall showing that declining fuel availability drove reductions in fire.
374-375: Is is different, but overall consistent that fires where high during 11-6 ka and had a secondary increase in the last few thousand years. But agreed that this supports your assertion that the lake PAHs are sourced generally from Eastern Africa.
375-376: What is the evidence for this?? They broadly agree with the Eastern Africa fire history.
378: Figure 4: You should more clearly label the time series to show which are from the same lake and which are not. Especially because some of these records are from the mainland vs. Madagascar. “Hydrophilic” should also be replaced with “Hydrophobic”.
379-380: Panel B: What is plotted here? Caption says charcoal and pollen, but y-axis says charcoal?
381: Two panel D’s, which is which?388: “This period…” Yes, and Karp 2023 argues that the precipitation is the main control on the fuel abundance and thus fire activity. If you disagree with this conclusion, you need to provide a strong counter argument.
392-393: This is context dependent. It matters if you are in a forest or grassland because precipitation controls on fire can have opposite effects (fuel vs moisture limited fire regimes).
401-403: Then why do your hydrophilic and hydrophobic PAHs follow the same patter? If PAH accumulation is all deposition driven, there should be deviation between the philic/phobic PAH time series, which there does not appear to be.
414-415: This can just as easily (and has been previously) interpreted as a fuel driven change in fire occurrence. In a few sentences you report that PAHs increased with modern human activity driving more fires without an associated increase in rainfall. So why is it that when rainfall is high, it is deposition that drives PAHs and not fire amount, but when rainfall is low it doesn’t matter and PAHs are driven by fire amount. How do you have this both ways?
428-429: I do not agree with this, if anything this speleothem record shows disagreements in the timing of PAH accumulation and local rainfall change: e.g., between 10-6 ka PAHs have a broad decline between 9-7 ka but while precipitation declines at 10.5-9 ka, it is just as high from 8.5-6 ka. If deposition was the main control one would expect a much tighter relationship with the local precipitation that displayed.
445: Huh? It's elevated. When did you discuss it being reduced during this time?
448-449: “However,… precipitation.” I don’t think this argument is supported by the data.
Citation: https://doi.org/10.5194/egusphere-2026-3726-RC2
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 102 | 41 | 18 | 161 | 23 | 16 | 14 |
- HTML: 102
- PDF: 41
- XML: 18
- Total: 161
- Supplement: 23
- BibTeX: 16
- EndNote: 14
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
General comments
This manuscript presents the results of a scientific research into the use of Polycyclic Aromatic Hydrocarbons (PAHs) as biomarkers of biomass burning, measured in a mountain lake in northern Madagascar throughout the Holocene, and their relationship with regional climatic drivers such as wind and rainfall patterns. This work is relevant to the research field, presenting an interesting approach based on the different chemical characteristics of PAH compounds to improve the estimation of their origin and location, and to determine whether they can be used as a proxy for wildfire signals preserved in environmental archives. Text editing and additional information, where demanded, would be beneficial.
Specific Comments
This manuscript addresses relevant scientific questions within the scope of CP by presenting novel data that supports the use of the concept of physicochemical properties to trace the geographic origin of chemical compounds and the main mechanisms by which they are transported over time.
The scientific hypothesis is clearly outlined, as is the methodology used to test it. However, the analytical methodology for determining PAHs requires further clarification to ensure it is fully understandable and reproducible without consulting the references.
While some sentences and figures require clarification, the results of this study are clearly presented and sufficient to test the stated hypothesis, providing adequate support for the interpretations, particularly given the approach based on the physicochemical properties of PAHs. However, the originality of this work should be better highlighted. Some insights have been presented and a comprehensive understanding based on this data has been achieved. However, the conclusions section needs improvement as it is too condensed.
The references used are adequate for the discussion of this topic. The title is OK. Although it contains all the mandatory elements, the abstract and the rest of the text should be improved in terms of language editing to make them clearer and more fluent. Some terms/words are highlighted in the technical corrections section.
Technical Corrections
I would prefer the terms 'Cal BP' or 'Cal yr BP', as these are more common, but this is just a suggestion.
Lines 57-59: It is missing a reference for this information.
Line 60: Why? Neither the above paragraph nor this text makes it clear why charcoal is limited as a regional wildfire proxy. Please could you improve this explanation?
Line 63: “current observations” - is this related to the 2017 reference? Is there a more recent study on the seasonality of PAH concentrations in the environment?
Line 81: “no CHARCOAL was found”.
Line 120 (Figure 1 Legend): Please complete the legend of Figure A with the winter, summer and annual mean precipitation information shown.
Line 133: Correct to “Fig. 1D”.
Line 134: I suggest the following text correction: “…we USED the 11-metre-LONG sediment cores LM1A and LM1B, taken in 2017 using a Russian PEAT corer (Teixeira et al., 2021).”
Please also include any other information from the sampling methodology described by Teixeira et al. (2021) that is relevant to PAH analysis and interferences, such as sample preservation. Also, please specify which analytical controls were employed to address potential contamination issues arising from the sampling procedures.
Lines 145-147: “Simultaneously, charcoal counts exhibit that fires started to occur, probably not in Montagne d’Ambre, but more likely occurring in the low elevation areas surrounding the mountains.”
Could you please rewrite this sentence to make it clearer?
Line 162: Were the samples spiked with internal standards prior to extraction? If so, which internal standards were added?
Lines 165-170: Please standardise the chemical notation for solutions according to the IUPAC recommendations. For example: 1:1 (v/v); heptane/dichlorometane (98:2, v/v)…
Do you really mean 98/2 v/v? Mostly heptane (4.9 ml), with just a few drops of DCM (100 µl)? Or was a different proportion used for this fraction's eluent? Which fraction is this, F2? Which proportion was used for the 2mL HEP/DCM solution for F2?
It is informed that “If pigment remained in fraction 1, a second separation by SiO2 column chromatography was performed.” How exactly? Is it just for F1, and are the same eluents used?
Please review the methodology information regarding the eluents and their proportions used to recover each fraction of compounds and use standard terms throughout the text.
Lines 171-172: Could you please review this sentence, as it is not clear enough? Could you also provide more information on whether fractions 1 and 2 were evaporated together or separately after being combined? Or were the two extracts analysed for PAHs separately? Was the DMSO removed from the extract prior to injection?
More detailed information on these steps is needed to make them reproducible.
Line 180: Instead of merely citing the reference, I strongly recommend that you provide a brief description of the applied method, highlighting any possible adaptations.
Line 196: Please correct “at at 25ºC”
Line 197: I would be careful using this concept, given that PAHs are not naturally hydrophilic. Instead, I would use the concept of increasing hydrophobicity as described by Garagnon et al. (2024).
Line 197: “solubility coeficient > 10-2 mg.L-1 (N, Ace, Flu, Phe, Ant, Fla and Pyr)”. Please provide the reference for this information, which was used to support this grouping.
Lines 205-206: This is an incomplete sentence. Please review it and provide the full information, including the equation, not just the reference.
Line 235 (Legend of Table 1): I suggest rewriting this sentence, e.g.: "Bold and italic lines represent inverse age, which were not taken into account in the model.”
Line 239 (Legend of Figure 2): I suggest rewriting these sentences, e.g.: "Linear age-depth model of Lake Maudit created using the CLAM package. Red points represent inverse age, which were not considered in the model.”
Line 258: “since 2800 cal a BP” is repeated. Please review this sentence.
Line 271 (Legend of Figure 3): I suggest using the lettering scale (A–D) for these four graphs and including this information in the legend.
Line 282: Please format this reference “(Ana Lúcia C. Lima et al., 2005)” along the text in accordance with the journal's rules.
Line 298: “(Pyr, Flu, Fla, Fig. SM 3)” - Figure 3 shows the distribution of N, Phe, Flu and Fla. It does not show Per or Pyr. Please adapt the figure to show the comparisons mentioned in the text.
Line 307: Please correct to correct to “… few CHARCOAL particles”
Lines 309-310: Where can it be seen/checked? In which figure? It cannot be seen clearly in Fig. 4.
“Particles >160um”? Which results show this? Do you mean the results of Teixeira et al. (2021)? If so, please clarify the size fractionation used.
Line 312: What do you mean by “presently occur”? When?
Lines 312-314: These two sentences are not easy to understand. Could you please rewrite this text to explain it further?
Lines 320-321: “Therefore, the northern region of Madagascar cannot be considered as a source of PAHs”
Which PAHs does this refer to? The pyrogenic ones? Could you please explain this relationship further, given that you pointed out a contrast between charcoal and PAHs above?
Line 322-323: “but this would reflect more remote sources supplying PAHs.”
Supplying all PAHs, or the "hydrophilic" ones?
Line 327: Please correct to “originated”
Line 329: “Where came from PAH during the Early and Mid-Holocene?”
Please review the grammar of this sentence.
Lines 341-342: “Therefore, this implies that although distant, the fire source is not excessively far.”
What does this mean in terms of distance? Is there any study of the atmospheric transportation distances of PAHs that can be cited here?
Line 367: Please correct to “In a different site” and clarify which one is it.
Line 371: Please correct to “Lake Masoko recorded high value of charcoal accumulation…”
Lines 375-376: “PAHs likely originate from regional biomass burning, but the variation was not due to fire activity variation.”
Please clarify this information.
Lines 378-383 (Legend of Figure 4):
Is there any data within the blue scale in Figure 4A? Could the scale be adapted to display the values more clearly, as was done in Fig. 4-C?
In the axis titles of Figures 4-D and 4-E, please indicate the locations and references of the charcoal data.
The legend is incoherent with Figure 4D. Please check it.
Line 443-452: I recommend rewriting the conclusion to make it clearer and address/clarify the following points:
“which are molecules produced by incomplete combustion” - that CAN be produced.” As mentioned in this manuscript, some PAHs can also have a biogenic origin.
“However, the signal of PAH differs from the fire activity signal” – fire activity of where?
“These results challenge the direct interpretation of PAHs in terms of fire events and dry periods”. Could you please clarify here the challenges you mention in terms of dry periods?
Lines 501-502: Please complete this reference
Lines 503-504: Please inform this reference’s DOI
Lines 596-599: Please inform this reference’s DOI
Lines 611-612: Please complete this reference
Line 643: DOI not found. Please check it.