Please see the entire comment in the attached pdf file

The manuscript claims to adapt the REVEALS model for global climate-vegetation reconstructions. However, several critical issues compromise the validity of its findings, please see below for further details.

While quantitative palynologists globally aspire to achieve global-scale reconstructions of vegetation, the reality is that this goal is probably decades away. The manuscript presents a methodologically flawed approach that fails to meet the scientific standards required for reliable global vegetation reconstruction using the REVEALS model. The assumption that RPPEs (relative pollen productivities) are constant continentally, hemispherically and temporally undermines the meticulous and dedicated empirical work conducted by the scientific community across recent decades in creating RPPEs. Particularly, this applies to regions across the Southern Hemisphere and tropical/sub-tropical, where RPPEs are being developed for the first time.

We recommend rejection (or extensive revisions) to be undertaken to address the issues outlined below, ensuring that the model is applied appropriately and effectively within its intended ecological, temporal and geographical constraints. If a revised version is produced, this should only be geographically constrained (probably just the Northern Hemisphere might be feasible in fact).

Major issues:

Inadequate regional calibrations: The generalization of RPPEs across broad geographical scales (hemispheres) ignores crucial ecological and bioclimatic regional variations. This approach most likely leads to significant inaccuracies in the vegetation reconstructions, in spite of what the presumed ‘validation’ approach suggests (see below).

Questionable data assumptions and methodological gaps: The use of northern hemisphere RPPE values for taxa not natively present in the southern hemisphere, such as Alnus in Australia, introduces substantial and confusing biases. Presumably, the authors have not consulted the relevant scholars who worked within this field and the geographical areas mentioned. Similarly, defaulting RPP to 1 for taxa without specific data oversimplifies pollen-vegetation relationships. The paper does not adequately address the absence of data for the Southern Hemisphere, leading to a misleading portrayal of global vegetation.

It is suggested >50% of RPPEs are missing for Australia and Oceanic pollen records. So, in this work a decision was made to run these records using the Northern Hemispheric RPPEs, despite very different bioclimatic and ecological contexts. This extrapolation of Northern Hemisphere RPPEs to southern locations missing PPEs without considering ecological or bioclimatic differences is particularly problematic. RPPEs empirically produced using ground truthing work (field surveys and surface pollen collection) were ignored, especially across the Southern Hemisphere (see some references below).

Duffin, K. I., & Bunting, M. J. (2008). Relative pollen productivity and fall speed estimates for southern African savanna taxa. Vegetation History and Archaeobotany, 17, 507-525.

Mariani, M., Connor, S. E., Theuerkauf, M., Kuneš, P., & Fletcher, M. S. (2016). Testing quantitative pollen dispersal models in animal-pollinated vegetation mosaics: An example from temperate Tasmania, Australia. Quaternary Science Reviews, 154, 214-225.

Mariani, M., Connor, S. E., Fletcher, M. S., Theuerkauf, M., Kuneš, P., Jacobsen, G., ... & Zawadzki, A. (2017). How old is the Tasmanian cultural landscape? A test of landscape openness using quantitative land‐cover reconstructions. Journal of Biogeography, 44(10), 2410-2420.

Mariani, M., Connor, S. E., Theuerkauf, M., Herbert, A., Kuneš, P., Bowman, D., ... & Briles, C. (2022). Disruption of cultural burning promotes shrub encroachment and unprecedented wildfires. Frontiers in Ecology and the Environment, 20(5), 292-300.

Oversimplified and incorrect spatial and temporal settings: The inclusion of incorrect basin types in the model without appropriate adjustments is very concerning. Why are marine records included for a model explicitly designed to work for large lakes of closed basins with wind dispersal as the only mechanism for pollen deposition?

The manuscripts states that ‘all sites that were not classified as lakes were run with peatland settings’ = can we consider the ocean a peatland? REVEALS cannot work with marine records and it definitely does not make sense to apply the ‘peatland’ settings for marine records with some random arbitrary basin radius (100m?). Further, using a deep temporal scope (50ka) without any consideration for massive climatic shifts (likely larger than the effect of regional RPPEs values vs regional bioclimate variations) are concerning oversights, making any pre-Holocene glacial REVEALS reconstructions unrealistic with current interglacial PPEs.

Dubious optimization and validation: Optimizing PPEs to match remote sensing data risks validating the model based on its own assumptions rather than providing an unbiased estimation of past vegetation, which REVEALS is designed to do. This circular reasoning undermines the scientific integrity of the model's outputs. While an interesting concept this needs to be validated separately on a much smaller spatially and higher resolution scale before such a widespread application. This cannot really be called a ‘validation’.

The reconstructed forest cover for the past 500 years was compared to modern remote sensed cover. Why not a smaller and more recent age bin was considered? In the past 500 years many areas of the world have been colonised by Europeans and have experienced major shifts in vegetation structure, as management transferred from Indigenous to colonial regimes (e.g. the Americas and Australia). This means that forest cover over the whole 500 years bin is not comparable to modern remote sensing data. This highlights a Eurocentric view of the global vegetation patterns.

An example of validation of RPPEs using modern vegetation data (with surveys) has been done in the following papers:

Mariani, M., Connor, S. E., Fletcher, M. S., Theuerkauf, M., Kuneš, P., Jacobsen, G., ... & Zawadzki, A. (2017). How old is the Tasmanian cultural landscape? A test of landscape openness using quantitative land‐cover reconstructions. Journal of Biogeography, 44(10), 2410-2420.

Mariani, M., Connor, S. E., Theuerkauf, M., Herbert, A., Kuneš, P., Bowman, D., ... & Briles, C. (2022). Disruption of cultural burning promotes shrub encroachment and unprecedented wildfires. Frontiers in Ecology and the Environment, 20(5), 292-300.

We’ve read with interest the manuscript by Schild et al. and are supportive of its goal to develop and publicly share an open set of global vegetation reconstructions for the late Quaternary.  Such reconstructions will be of interest to many in the Earth system modeling, paleoclimate, carbon cycle, and paleoecology communities.  

With this affirmation noted, we write this public comment to raise the concern that the current version of the manuscript does not adhere to best practices in open data citation and provenancing.  

The analyses presented by Schild et al. are based upon data from the Legacy Pollen Dataset V1.0 (hereafter LPD1.0) (Herzschuh et al., 2022), which in turn are mostly drawn from living data curated and made openly available by communities of paleoecologists working together to support the Neotoma Paleoecology Database (hereafter Neotoma) (Williams et al., 2018).  Specifically, of the original data compiled for LPD1.0, 3147 records were from Neotoma (90.6% of all records) and 324 (9.4%, mostly in China and Siberia) are from other sources.  Herzschuh et al. (2022) then added value to their compiled research dataset by 1) removing some records based on quality-control criteria and 2) building a harmonized set of taxonomic names. These actions follow recommended practices for large-scale paleoecological data syntheses (Flantua et al., 2023).

The Schild et al. ms. appropriately cites LPD1.0 but fails to cite Neotoma and/or include an acknowledgement of Neotoma, its Constituent Databases, and data contributors.  Given >90% of LPD1.0 data is sourced directly from Neotoma, this omission is a violation of Neotoma’s Data Use Policy (https://www.neotomadb.org/data/data-use-and-embargo-policy) and Creative Commons license (CC-BY 4.0, https://creativecommons.org/licenses/by/4.0/deed.en). The CC-BY license allows re-use and modification of Neotoma data, but requires that any downstream datasets that use Neotoma continue to provide attribution to Neotoma.  Flagship open-science journals like ESSD and synthesis papers like Schild et al. can further advance open science by setting a good example to other users by following best practices in data citation.

The lack of attribution to Neotoma’s Constituent Databases severs the link between the data and the Data Stewards who compiled these data.  These Stewards have regional expertise in the systematics and ecology of the taxa and the factors governing the dynamics, diversity, and distribution of species and ecosystems.  Including regional experts in global-scale syntheses is also a best practice, given the high diversity and heterogeneity of paleofloristic datasets (Flantua et al., 2023).  

The lack of a direct link to Neotoma datasets in this manuscript creates a break in provenance between the pollen datasets used by Schild et al. and the living datasets curated by Neotoma data stewards and Neotoma’s Constituent Databases.  Updates to Neotoma datasets are uncommon, but they happen as, e.g. age-depth models (chronologies) are updated or metadata errors are discovered and fixed by the community of Data Stewards associated with the various Constituent Databases.  Hence, the static datasets used by Schild et al. are now branched from the primary datasets curated by Neotoma and there is a risk that the static datasets in LPD1.0 will become stale relative to the living datasets in Neotoma.  

Fortunately, the solutions are simple.  First, the manuscript can add a citation and/or acknowledgment of Neotoma and the Constituent Databases that curated and provided these data:  the African Pollen Database, ALPADABA, European Pollen Database, Indo-Pacific Pollen Database, Latin American Pollen Database, and North American Pollen Database.  Citation endpoints and model text for acknowledgment are provided in the Neotoma Data Use Policy (https://www.neotomadb.org/data/data-use-and-embargo-policy) .

Second, the broken chain of dataset provenance and associated risk of data staleness can be removed by including a supplementary data table that has a column with a DOI link to each Neotoma dataset.  This solution also helps recognize data contributors by making their data more visible and cited.  (The LPD1.0 metadata table archived at Pangaea [https://doi.pangaea.de/10.1594/PANGAEA.929773] has DOIs to some original journal articles and dataset IDs for Neotoma datasets, but not Neotoma DOIs.)  The R-FossilPol package (Flantua et al., 2023) functionality includes a service that will generate a data table with DOIs for each Neotoma dataset (for example, see: https://github.com/HOPE-UIB-BIO/FOSSILPOL-example-Scandinavia/blob/main/Outputs/Meta_and_references/data_assembly_meta_2023-01-20.csv).  

To support this effort, we have developed and shared a GitHub Gist showing how to generate a table of DOIs and data citations using the `doi()` function from the neotoma2 R package (https://gist.github.com/SimonGoring/4a0b09b83373791a92c28b7191d07000). We have also opened an issue in the `neotoma2` GitHub repository () to improve the documentation for the `doi()` function, and to improve the package vignette to better integrate data citation best-practices.

John (Jack) W. Williams, Chair, Executive Working Group, Neotoma Leadership Council

Jessica Blois, University of California-Merced; Associate Chair, Executive Working Group, Neotoma Leadership Council

Thomas Giesecke, Utrecht University, Executive Working Group, Neotoma Leadership Council

Simon Goring, IT Team Lead, Neotoma Paleoecology Database

Sarah Ivory, Pennsylvania State University; Executive Working Group, Neotoma Leadership Council

References Cited

Flantua, S. G. A., Mottl, O., Felde, V. A., Bhatta, K. P., Birks, H. H., Grytnes, J.-A., Seddon, A. W. R., and Birks, H. J. B.: A guide to the processing and standardization of global palaeoecological data for large-scale syntheses using fossil pollen, Glob. Ecol. Biogeogr., n/a, https://doi.org/10.1111/geb.13693, 2023.

Herzschuh, U., Li, C., Böhmer, T., Postl, A. K., Heim, B., Andreev, A. A., Cao, X., Wieczorek, M., and Ni, J.: LegacyPollen 1.0: a taxonomically harmonized global late Quaternary pollen dataset of 2831 records with standardized chronologies, Earth Syst Sci Data, 14, 3213–3227, https://doi.org/10.5194/essd-14-3213-2022, 2022.

Williams, J. W., Grimm, E. G., Blois, J., Charles, D. F., Davis, E., Goring, S. J., Graham, R., Smith, A. J., Anderson, M., Arroyo-Cabrales, J., Ashworth, A. C., Betancourt, J. L., Bills, B. W., Booth, R. K., Buckland, P., Curry, B., Giesecke, T., Hausmann, S., Jackson, S. T., Latorre, C., Nichols, J., Purdum, T., Roth, R. E., Stryker, M., and Takahara, H.: The Neotoma Paleoecology Database: A multi-proxy, international community-curated data resource, Quat. Res., 89, 156–177, https://doi.org/10.1017/qua.2017.105, 2018.