Review of Earth System Dynamics manuscript: egusphere-2025-1424

Summary: Akabane et al. present a new compilation of pollen, charcoal, and human occupation paleorecords from the Neotropics (Central and South America) and compare them with previously published paleoclimate records to assess the regional drivers of vegetation and fire regime changes since the last glacial maximum (LGM) in several subregions of the Neotropics. They ground their study with informative context from data on climate/vegetation/fire patterns in modern ecosystems to aid their paleo interpretations. Overall, these authors find significant spatial heterogeneity in the role that climate variability (temperature, rainfall, and atmospheric CO₂ concentrations) and human activity have on fire regimes and vegetation structure across the wide range of environments in the Neotropics. They find that during the LGM low temperatures and low CO₂ concentrations primarily restricted vegetation productivity in the dry subtropics leading to low tree cover and low fire due to fuel limitations, while in wetter tropical regions rainfall had a stronger impact on tree growth and fire activity indicative of moisture limitation. Through the deglaciation, millennial scale changes in precipitation had a stronger influence on both tree cover and fire activity in addition to the impact of increasing temperatures. During the more thermally stable Holocene, changing precipitation vis-à-vis changes in monsoon strength dominantly controlled tree cover and fire activity with lesser and more spatially variable impacts of human activity.

This new compilation is impressive, and the analysis is thorough and detailed. I really appreciate the authors efforts to parse the records by climate/ecology to undergo a nuanced dissection of the competing influences of many factors on vegetation structure and fire regime and how these differ across geographies. The manuscript is very well written. They frame and motivate the problem well, the arguments are logical, the figures are clear and impactful, and it is overall quite pleasant to read. This study warrants speedy publication in Earth System Dynamics following minor revisions. I break down my review into major overarching comments followed by in-line comments and recommendations.  

Overarching comments:

Throughout the manuscript, the term “biomass” appears to be used interchangeably with “tree cover”. As a means of estimating vegetation structural change, the authors use the fraction of arboreal pollen in sediment cores. This method estimates the fraction of vegetation in a region which is composed of trees, however this is not a measure of biomass per se. All else equal, more trees on a landscape would equate to more biomass, but a ratio alone does not tell you this. Grasses can make up significant portions of the total biomass of ecosystems, particularly in tropical savannas (e.g., Cerrado). The authors should take a careful look at the instances where they make claims about changes in biomass when they are actually measuring tree pollen fraction to infer changes in tree cover. Grass biomass is an important fuel source especially in tropical savannas like the Cerrado, so I urge caution to the authors on broadly equating increased tree cover with biomass in the context of fuel availability. I flag such instances in the in-line comments.  

The description of the charcoal records is insufficient. The authors spend a decent portion of their methods section describing the dominant pollen types in the records that they compiled for each region but only list the number of records for charcoal without further description. Charcoal comes in many forms which record different aspects of fire regimes across multiple spatial scales. For instance, are all of the charcoal in the synthesis microcharcoal? Or are macrocharcoal particle records also included? One must read between the lines and look at the column title in the supplement table to infer this. A more complete description of the charcoal records including at a minimum the size fraction of the records used in this synthesis is needed.

The use of charcoal/pollen ratio records in this context surprises me. The authors say they multiply such records by the sedimentation rates of the cores to get an influx like unit, but this cannot be done. For such data to be ecologically meaningful, either (1) the pollen accumulation rate would have to be linearly correlated with the sedimentation rate and not a product of vegetation coverage within the watershed of the lake or (nearby river in the case of marine cores), or (2) the pollen accumulation rate would have to have such low variance that the change in charcoal accumulation rate drives the observed signal. Without convincing evidence in support of either of those scenarios, one cannot expect the charcoal numbers in these ratios to reflect changes in burning on the landscape. I suggest the authors remove such records from the compilation.

Figure 1 would benefit from a panel plotting either the fire radiative power or burned area data used in the modern analysis. The amount of burning across these biomes is highly variable. A map of where fires occur today would help the reader in interpreting the paleorecord compilations presented here. I would encourage the authors to emphasize that the majority of the sites in this study, except those in region 5 (Central Eastern Brazil) in the Cerrado, are currently situated in more forested regions that do not burn as much as tropical savannas. This would tie in within the results of the modern analysis (Figure 3) to show that one might expect more fire in the past if grasses where a more dominant fraction of the local vegetation.

Figures 12 and 13. I really like the time snapshot analysis presented in these figures. However, it is a little bit difficult to have to read across the panels and compare the colors between points to see if the z-scores increased or decreased through time by comparing the color to the previous time slice. When I first looked at these maps, I expected that they were displaying the trends rather than the mean z-scores for the time slice. One minor change you could make to these figures would be to change the markers depending on whether the mean z-score increased or decreased compared to the last time slice. E.g., upward triangle for increase above some threshold between 21-19 to 19-14.8, downward triangle for decrease below some threshold, and circle for no change outside of some threshold. This would really help as the reader is going through your later portion of the discussion when you are providing a broad overview of the trends through time. Additionally, the time slice labels are backward, they should be in chronological order, e.g., “(c) 21 – 19 ka”. You have it right in the figure caption, just reversed in the panel labels.

Regarding data availability, I suggest that the authors make the smoothed z-score time series of AP and charcoal influx available in the supplement or permanently stored in a public archive. These new synthesis curves are valuable information for paleoclimatologists, paleoecologists, anthropologists, climate modelers, etc., who might wish to compare them with their data. As I am sure the authors are aware from the webplotdigitizing they did for this study, it is always a relief when the key data for a paper are easily accessible for future analysis and plots don’t need to be unnecessarily and painstakingly recreated.

In-line comments:

19-20: I suggest you remove the “in the one hand” and “on the other hand” they are just not necessary to the point of the sentence which is already clear and concise.

24-25: Perhaps add “with additional impacts from human activity”  to the end of the sentence which starts with “Temperature …”

26: “Biomass growth” this should be “tree growth”

46: “process” should be “processes”

50: Glacial/interglacial cycles were occurring (although much more muted) in the Neogene. I suggest you change this to something like “onset of pronounced glacial/interglacial cycles in the Quaternary”

51: Change “were responsible for” to “played a significant role in”

52: Please clarify here if you mean in setting the modern ecosystem distributions or modulating changing ecosystem distributions through time.

56-57: “Weaker fire regime” is not very clear to a general reader. I am okay with the use of weaker and stronger fire regimes throughout the paper, but take a sentence here to explain what characteristics constitute and “weak” versus “strong” fire regime. Additionally, you could maybe also clarify also here what “fire activity” means because you use this general term in the text as well.

81 and throughout: You interchange “mm.yr-1” and “mm yr^-1" I recommend you adopt the second in all cases, the added period is unnecessary.

86-87: “…marked by weak seasonality with mean monthly temperatures ranging between 25 and 27 °C and mean annual precipitation of…” would be better as a comma-separated list: ““…marked by weak seasonality, mean monthly temperatures ranging between 25 and 27 °C, and mean annual precipitation of…”

93-94: It is not clear to me which positive feedback loop you are referring to. E.g. fire impacts on vegetation structure and knock-on effects on temperature and future fire likelihoods, fires emitting greenhouse gases leading to overall warming and thus more fires, etc. Please clarify.

99: “area” should be “areas”

103-104: “moist-laden” should be “moisture-laden”

199: I suggest you change “…open grasslands to closed shrublands and woodlands…” to “…open grasslands and savannas to closed shrublands and woodlands…”

121: “…the occurrence South Atlantic…” should be “…the occurrence of the South Atlantic…”

125-145: This is largely a style choice, so up to you, but in all preceding paragraphs you list temperature ranges from low to high (which is convention) but here you list high to low. I see that you might be doing this intentionally as you describe the regions from north to south, but it is a little weird to read temperature ranges from 16 to 3 °C.

140: “…southward-displaced…” should be “…southwardly displaced…”

165: For example, here is the only place in the text you mention microcharcoal. See overarching comments for method recommendations.

201-203: “For charcoal composites, …” You need to either provide strong justification that this is a viable method or remove such records from the compilation. See overarching comment.

295: “Despite represented by…” should be “Despite being represented by…”

303: Is this a mistaken paragraph break at the end of this line?

315: Figure 4 should have y-axis labels. A single label common for all subplots would be fine. The two columns appear unnecessarily squished together horizontally; I think you can add a little separation between the two which should give ample room for the y-axis labels.

341: “…high level…” and “…fire regime…” should be “…high levels…” and “…fire regimes…”

349: “p-values” can just be “p”

351: “condition” should be “conditions”

354: drop the “and” before “likely”

359: “…coeval to…” should be “…coeval with…”

357-358: I do not really see a slope break? The decline appears to be part of the larger trend of declining AP % since the EH. So, I would avoid calling it abrupt.

378: “…featured a reduced…” should be “…featured reduced…”

380-382: I thought Amazonia is generally moisture limited, so isn’t it surprising that low rainfall leads to low fire activity?

385-389: This intra-region spatial variability is really interesting. Are there enough records to split Amazonia up further and potentially make curves of E vs W or N vs S? If so, that could make for a nice supplementary figure. But if not, okay!

 436: “…the influence different…” should be “…the influence of different…”

438: “moist-laden” should be “moisture-laden”

441: “…speleothem cores suggest primarily reflect…” remove either “suggest” or “primarily reflect”

444-446: This is an interesting point. I wonder if you could comment further here about the potential impacts of llama grazing on grassy fuel loads?

501-505: It seems to me that the high rainfall observed during HS1 and during the YD  EH match peaks in the charcoal influx and BA matched a trough in the charcoal influx, suggesting fuel limited conditions which would make sense for the Cerrado savanna vegetation.

Also, what do you think happened ca. 19-17 ka? Why are the z-scores for charcoal influx so low? It’s hard to tell from the figure, but does the number of records go to zero at points in this interval? It would be good to comment on this. And perhaps obscure portions of the curve which might have very low confidence due to lack of sufficient data.

517: “Biomass growth” should be “tree growth”

517-518: “moist-laden” should be “moisture-laden”

535-536: I would recommend “biomass” in both cases should be “tree cover”

545: “biomass” should be “tree cover”

566: “This was likely consequence from significantly…” should be “This was likely the consequence of significantly…”

568: “tree cover increases along warming…” should be “tree cover increases along with warming…”

574: “biomass” should be “tree cover”

589-590: The human population was already quite high by 7 ka, so why is the influence on fires delayed until 3.5 ka?

604: “biomass” should be “tree biomass”

606: I think it is worth pointing out to the readers that Haas et al., (2023) is a modeling study and we do not yet fully understand the impacts of low atmospheric CO₂ on global fire regimes.

Overall, I really enjoyed this well-written and interesting paper, and I am excited to see the manuscript published following these revisions. Nice work!

Nicholas O’Mara

Review of the manuscript entitled "Vegetation and fire regimes in the Neotropics over the last 21,000 years"

The manuscript presents an excellent and timely synthesis, combining analyses of modern data with an extensive compilation of fossil records and archaeological evidence. The work is well written, clearly organized, and supported by high-quality figures that effectively convey the results.

I would like to offer a few comments and questions that I hope may help to further strengthen the manuscript:

1-Definition of arboreal pollen

The authors calculated arboreal pollen (AP) percentages as the sum of woody taxa (trees and palms), excluding mangrove and aquatic taxa, fern spores, and unidentified types. This approach is widely used in paleoecological studies and serves as a valuable proxy for reconstructing past vegetation structure.

However, given the floristic complexity of tropical ecosystems, I would like to kindly suggest that the authors provide further clarification regarding the taxonomic criteria used to define AP. Specifically, many plant families in these ecosystems include both arboreal and non-arboreal life forms, which may significantly influence AP percentages depending on how these groups were categorized. It would be helpful to know how the authors distinguished between arboreal and non-arboreal taxa within such families, and whether any standardized criteria were applied in this process.

For instance, I wonder whether the palm Mauritia flexuosa was considered part of the AP in the Cerrado records. Given that Mauritia palms are often highly abundant in local swamp environments (veredas), their inclusion could potentially inflate AP percentages without necessarily indicating a broader regional forest expansion. Clarifying this point would be particularly valuable for interpreting AP trends in relation to regional woody cover dynamics. Providing these additional details could enhance the transparency and reproducibility of the study, and also refine the paleoecological interpretations drawn from the AP trends.

2-Charcoal data scaling

The use of z-score transformation for scaling charcoal data, as applied by the authors, is a widely accepted and established method in paleo-fire research, allowing for effective comparison of variability within and between records.

However, recent studies (e.g., McMichael et al., 2021; Gosling et al., 2021) have highlighted some potential limitations of z-score scaling, particularly its tendency to distort the structure of charcoal peaks by inflating small-scale peaks and minimizing the influence of large peaks. Moreover, as noted by McMichael et al. (2021), this method "does not retain a consistent value that represents the absence of fire across sites", which can be especially relevant in tropical ecosystems where documenting both the presence and absence of fire is critical for understanding fire regime variability.

As an alternative, proportional relative scaling has been suggested (Gosling et al., 2021), which transforms charcoal data to a 0–100 scale while retaining the zero value to consistently represent fire absence. This approach also avoids upweighting rare charcoal finds in otherwise charcoal-poor sequences, potentially providing a more ecologically meaningful representation of fire activity.

Given these recent discussions, I wonder whether the authors considered alternative scaling methods, and if so, what motivated the decision to apply the z-score transformation. I believe that elaborating on this methodological choice could be valuable for readers and for future studies in similar tropical contexts.

3-Relationship between tree cover, biomass, and fire

The interpretation proposed by the authors—linking positive correlations between tree cover and fire frequency to fuel-limited regimes, and negative correlations to moisture-limited regimes (e.g. lines 349-350; 421-422; 535-537; 554-555;574-575)—is broadly consistent with established ecological theory on fire-vegetation-climate interactions. This framework provides a useful lens for interpreting paleoecological data, particularly in ecosystems where fuel availability is a key constraint on fire activity.

Indeed, in ecosystems where fire regimes are typically fuel-limited, such as deserts, xeric shrublands, and dry savannas and grasslands, fire occurrence is constrained by low biomass production and the discontinuity of fuel, despite often dry climatic conditions (e.g., Krawchuk et al., 2009). In these cases, it is true that an increase in biomass is necessary to reach the threshold at which fire can propagate across the landscape (e.g., Pausas & Ribeiro, 2013).

However, it is important to recognize that this relationship is not linear. Once a certain level of biomass is reached—sufficient to produce continuous fuel loads—further increases in biomass, particularly through increased tree cover, do not necessarily lead to higher fire frequency. In fact, dense woody cover can reduce fire frequency (which is acknowledged by the authors in lines 516-517) by suppressing the herbaceous layer, increasing shading and moisture retention, and creating microclimatic conditions less favorable to combustion (e.g., Staver et al., 2011).

Moreover, the assumption that increasing tree cover directly facilitates increased fire activity under fuel-limited conditions may not always hold. In some cases, both variables—tree cover and fire frequency—could increase independently as parallel responses to external climatic drivers. For instance, a scenario in which warmer temperatures promote tree expansion, while drier conditions simultaneously enhance fire frequency, could result in a positive correlation that does not necessarily reflect a causal, fuel-mediated link. Such a situation might be relevant to the patterns observed in some of the analyzed regions.

Conversely, in cases of negative correlations, it is important to consider that a reduction in fire frequency could also be a precondition for tree cover expansion, rather than its consequence (e.g., Staver et al., 2011). These alternative causal pathways underscore the importance of considering the directionality of the relationships and the potential influence of external climatic factors.

Additionally, it is crucial to emphasize that in many fire-prone ecosystems, especially savannas, herbaceous fuels—notably C4 grasses—are the primary drivers of fire regimes, while woody biomass plays a secondary role (e.g., Bond & Keeley, 2005). Therefore, correlations between tree cover and fire activity may not fully capture the dynamics of fuel availability and fire propagation (the authors seem to consider throughout the ms that arboreal cover (interpreted from AP) equals biomass growth and fuel availability).

Overall, the interpretations made by the authors are reasonable within their theoretical framework, but incorporating these additional ecological nuances could further strengthen the discussion and provide a more comprehensive understanding of the complex interplay between vegetation dynamics and fire regimes.

4-Potential role of megafauna extinction:

One additional factor that may have influenced vegetation structure and fire regimes in the Neotropics during the late Quaternary is the extinction of megafauna at the Pleistocene-Holocene transition (or later - e.g. Faria et al., 2025). Large herbivores are known to play a critical role in shaping vegetation through grazing, browsing, and trampling, thereby modulating fuel loads and fire regimes (Gill et al., 2009; Doughty et al., 2016). The disappearance of these animals in (some parts of) South America may have contributed to changes in woody cover and fuel accumulation, potentially influencing fire dynamics independently or synergistically with climatic and anthropogenic factors. While I understand that this topic may be beyond the primary scope of the manuscript, I wonder whether the authors considered this as a possible additional driver in some regions, or whether any of the available paleoecological records reflect such transitions.

Final remarks

Overall, I believe this manuscript makes an important and valuable contribution to our understanding of long-term vegetation and fire dynamics in the Neotropics. It integrates multiple lines of evidence in a thoughtful and rigorous way, and I am confident it will be a useful reference for researchers working in this field. I appreciate the opportunity to revise this study and hope my comments are helpful to further refine this already excellent manuscript.

Raquel Cassino

Minor comments:

Lines 237 - 241: replace “use” by “used”

Line 436: “which suggests the influence “of” different mechanisms”

Line 441: “Furthermore, while δ18O in ice and speleothem cores suggest primarily reflect rainy season precipitation” - remove “suggest”.

References:

Bond, W.J., & Keeley, J.E. 2005. Fire as a global ‘herbivore’: the ecology and evolution of flammable ecosystems. Trends in Ecology & Evolution, 20(7), 387–394.

Doughty, C. E., Wolf, A., & Malhi, Y. 2016. The legacy of the Pleistocene megafauna extinctions on nutrient availability in Amazonia. Nature Geoscience, 9, 800–803.

Faria, F. H. C., Carvalho, I. S., Araújo-Júnior, H. I., Ximenes, C. L., & Facincani, E. M. 2025. 3,500 years BP: The last survival of the mammal megafauna in the Americas. Journal of South American Earth Sciences, 153, 105367.

Gill, J. L., Williams, J. W., Jackson, S. T., Lininger, K. B., & Robinson, G. S. 2009. Pleistocene megafaunal collapse, novel plant communities, and enhanced fire regimes in North America. Science, 326(5956), 1100-1103.

Gosling, W. D., Maezumi, S. Y., Heijink, B. M., Nascimento, M. N., Raczka, M. F., van der Sande, M. T., Bush, M. B., & McMichael, C. N. H. 2021. Scarce fire activity in north and north-western Amazonian forests during the last 10,000 years. Plant Ecology & Diversity, 14(1), 89–99.

Krawchuk, M.A., et al. 2009. Global pyrogeography: the current and future distribution of wildfire. PLoS ONE, 4(4): e5102.

McMichael, C. N. H., Heijink, B. M., Bush, M. B., & Gosling, W. D. 2021. On the scaling and standardization of charcoal data in paleofire reconstructions. Frontiers in Biogeography, 13(1), e49431

Pausas, J.G., & Ribeiro, E. 2013. The global fire–productivity relationship. Global Ecology and Biogeography, 22(6), 728–736.

Staver, A.C., Archibald, S., & Levin, S.A. 2011. The global extent and determinants of savanna and forest as alternative biome states. Science, 334(6053), 230–232.

The manuscript presented by Akabane et al., presents a synthesis of how vegetation and fire regimes in the Neotropics have responded to climatic changes over the last 21,000 years in seven subregions of the Neotropics. The authors used a modern analysis of vegetation distribution and fire activity in relation to current climatic conditions, complemented by a compilation of 243 vegetation and 127 fire records to assess changes over the past 21 ka BP.

The manuscript is well-structured and comprehensible, offering a significant contribution to the understanding of long-term ecosystem dynamics in the regions studied. Although there is a scarcity of paleo records in some areas, this research is a good example of the importance of databases and open data in enhancing our understanding of Neotropical ecosystems, and it calls for additional research to address the existing knowledge gaps.

GENERAL COMMENTS

The title proposed by the authors “Vegetation and fire regimes in the Neotropics over the last 21,000 years” suggest an extensive examination of the entire neotropical region. However, this study is restricted to seven specific subregions, omitting the northern Andes, parts of Bolivia and Chile, and large portions of Brazil and Argentina. This selection was partly due to data availability, yet it resulted in a focus on these seven subregions rather than a comprehensive analysis of the Neotropics.

In the section on vegetation settings, the authors provide a detailed description of the selected subregions, emphasizing the primary climatic drivers, such as the ITCZ, SASM, and ENSO. However, the analysis and discussion neglect the direct climatic influences from the Pacific Ocean (and records from the west flank of the Andes), particularly ENSO. The authors note in section 195: “Datapoints outside the defined subregions (black dots in Fig. 2a) were excluded from subregional analyses because they were either isolated or located outside subregional definitions, e.g., high montane sites from NNeo >3000 m; low altitudes from CAn < 2200 m; or positioned on the Andean west flank, which has a distinct climate control relative to the east flank and Altiplano”. The climatic dynamics influenced by the Pacific Ocean cannot be overlooked, particularly given that their effects can be detected globally (e.g. ENSO). If the analyzed records did not exhibit a clear signal, this should be explicitly stated. Similarly, if the sampling resolution is insufficient for the western flank, as has been acknowledged for other regions with limited data coverage such as Northeastern Brazil (NEB), this limitation should also be addressed. In the same line, it is unclear why ecosystems from high mountains, such as paramos, are excluded from the analysis. Several of them have proven useful for understanding past climatic variability (e.g. Haggemans et al., 2022, Espinoza et al., 2022, Ledru et al., 2022). It is worth mentioning that the northern part of the Andes was completely excluded in this study.

In the methodology, it is not entirely clear how the authors have standardized the pollen data. It draws my attention to the part where they have “excluded mangrove and aquatic taxa, fern spores and unidentified types”. In the case of fern spores, the authors should evaluate or clarify if they have considered tree ferns in their AP composite. In some regions, such as the Atlantic and Andean Forests, tree ferns, like Cyathea, Lophosoria, or Dicksonia species thrive under specific conditions, with water availability being a common factor. In pollen records, they serve as key indicators of humid conditions and are included in the pollen sum (Kesler et al., 2011, Salazar et al., 2013, de Gasper et al., 2021).

Regarding fire dynamics, it is unclear how the authors determined high or low fire activity. The methodology from the compiled data is not evident. Additionally, the type of data used to reconstruct fire regimes, whether macrocharcoal, microcharcoal, or both is not specified. Similarly, the authors should pay careful attention to the interpretation of fire dynamics in ecosystems such as savannas, in which the fuel for fire primarily comes from grasses. The authors used AP to determine the available biomass to be burned; however, for these types of systems, grasses and herbs should also be considered (Alvarado et al., 2020).

I hope these suggestions are useful to complement this outstanding review, and I congratulate the authors on their efforts to contribute to the understanding of Neotropical ecosystems at different time scales.

REFERENCES

Steiger, N.J., Smerdon, J.E., Seager, R. et al. (2021). ENSO-driven coupled megadroughts in North and South America over the last millennium. Nat. Geosci. 14, 739–744

Hagemans et al., (2022). Intensification of ENSO frequency drives forest disturbance in the Andes during the Holocene. Q. Sci. Rev. 294, 107762

Espinoza, I. G., Franco-Gaviria, F., Castañeda, I., Robinson, C., Room, A., Berrío, J. C., et al. (2022). Holocene fires and ecological novelty in the high colombian cordillera oriental. Front. Ecol. Evol. 10:895152.

Ledru, M.-P., Aquino-Alfonso, O., Finsinger, W., Samaniego, P., & Hidalgo, S. (2022). Changes in the vegetation and water cycle of the Ecuadorian páramo during the last 5000 years. The Holocene, 32(9), 950-963.

Kessler, M., Kluge, J., Hemp, A., Ohlemüller, R. (2011). A global comparative analysis of elevational species richness patterns of ferns. Glob. Ecol. Biogeogr. 20, 868–880.

Salazar, L., Homeier, J., Kessler, M., Abrahamczyk, S., Lehnert, M., Krömer, T., & Kluge, J. (2013). Diversity patterns of ferns along elevational gradients in Andean tropical forests. Plant Ecology & Diversity, 8(1), 13–24

De Gasper, A. L., Grittz, G. S., Russi, C. H., Schwartz, C. E., Rodrigues, A. V. (2021). Expected impacts of climate change on tree ferns distribution and diversity patterns in subtropical Atlantic Forest. Perspect. Ecol. Conserv. 19, 369–378.

Alvarado, S. T., Andela, N., Silva, T. S., and Archibald, S.(2020) Thresholds of fire response to moisture and fuel load differ between tropical savannas and grasslands across continents, Global Ecol. Biogeogr., 29, 331–344