the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 01 Sep 2025
Mapping and modelling a boreal forest soil organic carbon predictor in the glacial till of Newfoundland, Canada
Abstract. Boreal forest soils store 30 % of global forest soil carbon, making them a crucial component of the carbon cycle. However, climate change at high latitudes is resulting in heightened temperatures and increasingly unpredictable precipitation patterns. Soil organic carbon (SOC) formation and stabilization is tied to precipitation patterns, thus climate change will inevitably influence the stability and longevity of boreal forest SOC. The current size and distribution of this reservoir is poorly understood, creating uncertainty under current and future climate scenarios. Previous research demonstrates mineral soil properties may be used to model boreal forest SOC accurately. The surface slope, depth of carbon enriched horizon, and climate characteristics are important parameters for modelling SOC and can generally be obtained or estimated via remote sensing. However, information about aluminum availability – the weatherable aluminum capable of interacting with organic matter to form stable carbon rich organometal complexes in mineral soils – is not widely available but is controlled by soil parent material. To bridge this gap, the Newfoundland and Labrador till geochemistry dataset was used here to map and model aluminum availability in glacial till across the island of Newfoundland as a function of geology and climate. The Random Forest Algorithm was employed to develop two models: one relying solely on the strength of geological and climatic variables, and the other drawing additionally on the spatial context of sample points. The first model performed well (R2=0.60), however, adding a spatial component increased the performance of the second model (R2=0.71). Bedrock type and proximity of the samples to certain units were indicated to be the strongest controls on aluminum availability, while environmental factors were less influential. Additionally, model uncertainty was calibrated empirically and mapped spatially, providing reliable and actionable information about the confidence of predictions over the study area. This project demonstrates the value of predictive geospatial modelling for till geochemistry mapping and delivers key aluminum availability predictions for deriving SOC reservoir estimates across Newfoundland.
- Preprint
(1699 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2025-3475', Anonymous Referee #1, 08 Oct 2025
-
AC1: 'Reply on RC1', Scout Quinn, 15 Jan 2026
Review for “Mapping and modelling a boreal forest soil organic carbon predictor in the glacial till of Newfoundland, Canada”
I read the manuscript "Mapping and modelling a boreal forest soil organic carbon predictor in the glacial till of Newfoundland, Canada" by Quinn et al. with great interest, as the goal of the paper is to create a high-resolution geospatial model predicting the distribution of Al rich weatherable minerals in glacial till in Newfoundland. The manuscript is overall well structured and easy to follow. However, I see a few major flaws that need to be addressed before the manuscript can be considered for publication.
We thank the reviewer kindly for their insightful and thorough comments throughout our manuscript. They will certainly help improve the manuscript and clarify several points.
General comments:
Available Al. The authors describe that Al availability (%Al) is calculated via the chemical index of alteration (CIA). This is not correct. The CIA measures the proportion of Al2O3 relative to mobile cations (Ca, K, Na). A high CIA means that the sample is highly weathered, Al-rich, and mobile cations are depleted. Thus, the CIA measures relative proportions during weathering, not which Al is actually bioavailable or extractable. In equation 2, the authors calculate the fraction of mobile cations remaining (100 – CIA)/100. This is then multiplied by Al203. To me, this is conceptually backwards, because a high CIA means more Al enrichment (more weathering), but their formula gives lower %Al when CIA is high. In essence, the authors calculate Al203 x (mobile cation fraction), to me, it is unclear what this represents physically. Either the authors need to describe better what they are using as input variables and why or they have to re-consider their calculations entirely.
We appreciate the need for clarification on this topic. The above comment possibly stems from a misunderstanding of the Al availability concept. The CIA does indeed measure the proportion of stable Al to more mobile cations, but more importantly it functions as an indicator of the degree of weathering a mineral in the parent material has undergone. Previous research indicates that organometal complexes form not with highly weathered Al in parent material, but rather fresh, poorly crystalline Al hydroxides, which deplete with extensive weathering (Slessarev et al., 2022). Therefore, while there may be proportionally more stable Al in a more highly weathered sample, it is less likely to contribute to the formation of soil Al oxyhydroxides conducive to organometal complexation and SOC storage. In our study the degree of weathering is determined using the CIA, then the proportion of mobile cations is found. This is used to determine how much of the initial Al content of the glacial till can be considered to be “available” to bind to SOC, representing the Al oxyhydroxides.
To clearly address this issue we plan to make the following edits within the manuscript:
L121: “SOC is stored in Al OMCs formed from fresh, primary weathering products (Slessarev et al., 2022) supported by poorly weathered rather than extensively weathered forms of Al in the parent material. To assess the degree of mineral weathering, the chemical index of alteration (CIA) (Nesbitt & Young, 1984) was calculated following Eq. (1):
And L123: “The Al%, or the Al presumed to be still “available” to support the formation of Al OMCs following weathering processes, was then derived using Eq 2:”
Motivation/Framing. I find the title and motivation laid out in the abstract and introduction very misleading. As it is described right now, the motivation is to improve SOC predictions in boreal forests using Newfoundland as a case study. However, the paper itself is about predicting %Al which is thought by the authors to be important for predicting SOC. However, the authors do not show that their %Al maps actually improve SOC predictions. Either the authors need to quantitatively show that or reframe the motivation and title of the paper.
We appreciate this comment. The reviewer is correct, and we propose to revise the title to better reflect the aim of the paper; for example Spatial modelling of reactive aluminum in glacial till: informing a boreal forest soil carbon predictor.” The motivation for this work emerges from an existing body of research demonstrating the value of reactive Al in SOC retention in boreal forest soils (eg. Patrick et al. 2022; Spinola et al., 2026). Indications that Al is one of the strongest controls on SOC storage in these soils makes mapping Al for use in future SOC models relevant.
Geospatial modeling. The authors need to better describe and explain what the difference between the two random forest models is. From the manuscript, it is not clear to me what exactly it means that one model is “drawing additionally on the spatial context of sample points”. What are the “layers” that represent the distance to each bedrock unit? Is that the distance each sample has to all bedrock units present in the region? Isn’t each sample assigned to one bedrock unit? I guess I miss something here. Additionally, it is not clear why the presented set of covariates was chosen. For example, what is the underlying process for including “Distance from Ocean”?
We agree that further clarification of the Distance to Bedrock layers will improve the manuscript. The reviewer is correct: these predictors are Euclidean distance layers from the nearest of each bedrock unit class. Motivation for testing these layers in one of the models is that glacial till is derived not only from the rock unit upon which the sample is taken, but also from ground up rock from nearby units, with distant units affecting the composition of the till less than near units. It introduces “spatial context” to the model by incorporating information about the distance between sample points and proximal bedrock units. Distance to the ocean was included because following glacial retreat, marine inundation affected areas of the coastline, resulting in reworking and weathering of surficial deposits near the coastline. This was thought to potentially impact Al% near the coastline. We propose to add this information to clarify the manuscript.
Specific comments:
L262f: I don’t think that based on the presented results that the authors can conclude the map addresses the need to incorporate %Al into SOC modeling. The maps do not show any relationship with SOC.
We appreciate this comment and believe that revision to the title will better reflect the aim of paper, which is not to demonstrate a relationship between SOC and Al, but to develop a methodology for mapping and modelling reactive Al with publicly available till geochemistry data for use in future SOC modelling efforts.
L295: I don’t see any PDP plots in the manuscript.
Partial dependence plots (PDP) were not initially included given the large number of predictor variables (51). It is true that exploration of these plots could support interpretation of the model though. If the reviewers and editor think it necessary, a suitable compromise may be to provide the top n contributing variables for each model (e.g., top 5 or 10).
Nesbitt, H. W., & Young, G. M.: Prediction of some weathering trends of plutonic and volcanic rocks based on thermodynamic and kinetic considerations, Geochimica et Cosmochimica Acta, 48, 1523–1534, https://doi.org/10.1016/0016-7037(84)90408-3, 1984.
Patrick, M. E., Young, C. T., Zimmerman, A. R., & Ziegler, S. E.; Mineralogic controls are harbingers of hydrological controls on soil organic matter content in warmer boreal forests, Geoderma, 425, Article 116059, https://doi.org/10.1016/j.geoderma.2022.116059, 2022.
Slessarev, E.W., Chadwick, O.A., Sokol, N.W., Nuccio, E.E., Pett-Ridge, J.: Rock weathering controls the potential for soil carbon storage at a continental scale, Biogeochemistry 157, 1-13, https://doi.org/10.1007/s10533-021-00859-8, 2022.
Spinola, D., Jones, D., Portes, R., Stewart, A., Sanborn, P., & D’Amore, D. (2026). A continental-scale study of Spodosols across North America and implications for soil organic carbon dynamics. Catena (Giessen), 263, Article 109743. https://doi.org/10.1016/j.catena.2025.109743
Citation: https://doi.org/10.5194/egusphere-2025-3475-AC1
-
AC1: 'Reply on RC1', Scout Quinn, 15 Jan 2026
-
RC2: 'Comment on egusphere-2025-3475', Anonymous Referee #3, 22 Dec 2025
The manuscript by Quinn et al., titled ‘Mapping and modelling a boreal forest soil organic carbon predictor in the glacial till of Newfoundland, Canada’ tackles an important topic and aims improving soil organic carbon predictions for a special region in Canada.
In the introduction the authors focus on using soil available aluminium as a predictor to improve mineral SOC stabilization, in addition to remote sensing obtained environmental characteristics, such as surface topography, C enriched horizon depth and climate characteristics. While the authors provide an interesting new approach, the manuscript will benefit from some clarifications or changing the title reflect the focus of the study.
The authors use the CIA as foundation for their model comparison to get Al% from soil geological databses. The CIA indeed describes an enrichment of Al relative to other cations (Ca, Na and K), and also describes feldspar breakdown into clay minerals. So one major question arises for me, could one also use Ca, Na and K loss and/or clay content/fraction changes as indicator for weathering as alternative confirmation to check model performances?
The aim of the paper (line 81: a high-resolution geospatial model predicting the distribution of Al rich weatherable minerals in glacial till in Newfoundland) is not really reflected in the title. Maybe reconsider a more concise title? So far there is nearly not really a focus on SOC modeling, but rather the manuscript focusses on Al distribution modeling.
I am also missing alternative approaches to map Al and subsequent SOC to see if Al representation really would improve SOC modelling.
Moreover, another cross validation would be to compare different weathering indices, and see if there are potentially different results available, and a better description of the random forest models would be needed.
Citation: https://doi.org/10.5194/egusphere-2025-3475-RC2 -
AC2: 'Reply on RC2', Scout Quinn, 15 Jan 2026
The manuscript by Quinn et al., titled ‘Mapping and modelling a boreal forest soil organic carbon predictor in the glacial till of Newfoundland, Canada’ tackles an important topic and aims at improving soil organic carbon predictions for a special region in Canada.
In the introduction the authors focus on using soil available aluminium as a predictor to improve mineral SOC stabilization, in addition to remote sensing obtained environmental characteristics, such as surface topography, C enriched horizon depth and climate characteristics. While the authors provide an interesting new approach, the manuscript will benefit from some clarifications or changing the title to reflect the focus of the study.
The authors use the CIA as foundation for their model comparison to get Al% from soil geological databases. The CIA indeed describes an enrichment of Al relative to other cations (Ca, Na and K), and also describes feldspar breakdown into clay minerals. So one major question arises for me, could one also use Ca, Na and K loss and/or clay content/fraction changes as indicator for weathering as alternative confirmation to check model performances?
We thank the reviewer for their constructive and insightful comments. Ca, Na, and K are related to Al, but they are already used to calculate the parent material Al availability through the CIA. Thus, these elements are already used in the CIA calculation, and would not be appropriate as an independent evaluation of model performance. Doing so would risk “data leakage”, where information about the response variables from model training is available during the validation step. Clay content/fraction changes are not available in the till geochemistry database for these samples so this option was not explored. This is not a calculation typically included in till geochemistry datasets, illustrating the relevance in further application of this research. The use of Al availability directly addresses the mechanisms behind this form of SOC storage (organometal complexation), and reactive metals prove to be a more reliable indicator of SOC storage capacity in podzols than clay and silt (Spinola et al., 2026; Patrick et al. 2022; Rasmussen et al., 2018). In this case, cross validation was chosen as the most informative indicator of model performance, lacking an independent test dataset.
The aim of the paper (line 81: a high-resolution geospatial model predicting the distribution of Al rich weatherable minerals in glacial till in Newfoundland) is not really reflected in the title. Maybe reconsider a more concise title? So far there is nearly not really a focus on SOC modeling, but rather the manuscript focuses on Al distribution modeling.
We appreciate this comment, which has been raised by both reviewers. We plan to revise the title to better reflect the aim of the paper, for example “Spatial modelling of reactive aluminum in glacial till: informing a boreal forest soil carbon predictor.”
I am also missing alternative approaches to map Al and subsequent SOC to see if Al representation really would improve SOC modelling.
The role of minerals in soil organic carbon sequestration in podzols is well established (eg. Patrick et al., 2022; Patrick et al., 2025; Rasmussen et al., 2018; Slessarev, 2022; etc.). The spatial distribution of reactive metals including aluminum has been proven to determine that of SOC (eg. Ren et al., 2024), indicating that including it in future SOC models across our study area and other landscapes dominated by podzolic (spodsolic) soils would be valuable (section 1, L60-62, L67-82). Additional references on this topic will be added in this section for clarification.
Moreover, another cross validation would be to compare different weathering indices, and see if there are potentially different results available, and a better description of the random forest models would be needed.
While comparison of different indices would be interesting, it is outside of the scope of this study. The CIA was chosen based on findings from previous work in the field (e.g., Patrick, 2023).
We are unclear about the ways in which our current description of the Random Forest models are lacking. The algorithm is described generally in the second paragraph of section 2.4, including relevant references, and our specific implementations are described in the paragraph directly following. We would invite clarification about this comment.
Patrick, M. E., Young, C. T., Zimmerman, A. R., & Ziegler, S. E.Mineralogical controls are harbingers of hydrological controls on soil organic matter content in warmer boreal forests, Geoderma, 425, Article 116059, https://doi.org/10.1016/j.geoderma.2022.116059, 2022.
Patrick, M.E.: The roles of parent material, climate, and geomorphology in soil organic carbon response to short-term climate change in moist boreal forests, Doctor of Philosophy Dissertation, Department of Earth Sciences, Memorial University of Newfoundland, pp 265, https://doi.org/10.48336/SKF9-1132 2023.
Patrick, M. E., Myers-Pigg, A. N., Gates, Z., Gaviria Salazar, C., Morry, A. K., Prestegaard, K., & Ziegler, S. E. (). Hillslope hydrologic influences on soil dissolved organic carbon fate informs extreme precipitation impacts on boreal forest mineral soil stocks. Catena (Giessen), 259, Article 109361. https://doi.org/10.1016/j.catena.2025.109361, 2025.
Rasmussen, C., Heckman, K., Wieder, W. R., Keiluweit, M., Lawrence, C. R., Berhe, A. A., Blankinship, J. C., Crow, S. E., Druhan, J. L., Hicks Pries, C. E., Marin-Spiotta, E., Plante, A. F., Schädel, C., Schimel, J. P., Sierra, C. A., Thompson, A., & Wagai, R.: Beyond clay: Towards an improved set of variables for predicting soil organic matter content, Biogeochemistry, 137, 297–306. https://doi. org/10.1007/s10533-018-0424-3, 2018.
Ren, S., Wang, C., & Zhou, Z., Global Distributions of Reactive Iron and Aluminum Influence the Spatial Variation of Soil Organic Carbon, Global Change Biology, 30, e17576-n/a., https://doi.org/10.1111/gcb.17576, 2024.
Slessarev, E.W., Chadwick, O.A., Sokol, N.W., Nuccio, E.E., Pett-Ridge, J.: Rock weathering controls the potential for soil carbon storage at a continental scale, Biogeochemistry 157, 1-13, https://doi.org/10.1007/s10533-021-00859-8, 2022.
Spinola, D., Jones, D., Portes, R., Stewart, A., Sanborn, P., & D’Amore, D. (2026). A continental-scale study of Spodosols across North America and implications for soil organic carbon dynamics. Catena (Giessen), 263, Article 109743. https://doi.org/10.1016/j.catena.2025.109743
Citation: https://doi.org/10.5194/egusphere-2025-3475-AC2
-
AC2: 'Reply on RC2', Scout Quinn, 15 Jan 2026
Data sets
Replication Data for: Mapping and modelling a boreal forest soil organic carbon predictor in the glacial till of Newfoundland, Canada Version 1.1 Scout M. Quinn et al. https://doi.org/10.5683/SP3/6AONRU
Model code and software
smquinn99/SMQuinn_Mapping_and_modelling_a_boreal_forest_soil_organic_carbon_predictor_in_glacial_tills_of_NL Scout M. Quinn https://doi.org/10.5281/zenodo.16109126
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 1,899 | 91 | 33 | 2,023 | 33 | 31 |
- HTML: 1,899
- PDF: 91
- XML: 33
- Total: 2,023
- BibTeX: 33
- EndNote: 31
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Review for “Mapping and modelling a boreal forest soil organic carbon predictor in the glacial till of Newfoundland, Canada”
I read the manuscript "Mapping and modelling a boreal forest soil organic carbon predictor in the glacial till of Newfoundland, Canada" by Quinn et al. with great interest, as the goal of the paper is to create a high-resolution geospatial model predicting the distribution of Al rich weatherable minerals in glacial till in Newfoundland. The manuscript is overall well structured and easy to follow. However, I see a few major flaws that need to be addressed before the manuscript can be considered for publication.
General comments:
Available Al. The authors describe that Al availability (%Al) is calculated via the chemical index of alteration (CIA). This is not correct. The CIA measures the proportion of Al2O3 relative to mobile cations (Ca, K, Na). A high CIA means that the sample is highly weathered, Al-rich, and mobile cations are depleted. Thus, the CIA measures relative proportions during weathering, not which Al is actually bioavailable or extractable. In equation 2, the authors calculate the fraction of mobile cations remaining (100 – CIA)/100. This is then multiplied by Al203. To me, this is conceptually backwards, because a high CIA means more Al enrichment (more weathering), but their formula gives lower %Al when CIA is high. In essence, the authors calculate Al203 x (mobile cation fraction), to me, it is unclear what this represents physically. Either the authors need to describe better what they are using as input variables and why or they have to re-consider their calculations entirely.
Motivation/Framing. I find the title and motivation laid out in the abstract and introduction very misleading. As it is described right now, the motivation is to improve SOC predictions in boreal forests using Newfoundland as a case study. However, the paper itself is about predicting %Al which is thought by the authors to be important for predicting SOC. However, the authors do not show that their %Al maps actually improve SOC predictions. Either the authors need to quantitatively show that or reframe the motivation and title of the paper.
Geospatial modeling. The authors need to better describe and explain what the difference between the two random forest model is. From the manuscript, it is not clear to me what exactly it means that one model is “drawing additionally on the spatial context of sample points”. What are the “layers” that represent the distance to each bedrock unit? Is that the distance each sample has to all bedrock units present in the region? Isn’t each sample assigned to one bedrock unit? I guess I miss something here. Additionally, it is not clear why the presented set of covariates was chosen. For example, what is the underlying process for including “Distance from Ocean”?
Specific comments:
L262f: I don’t think that based on the presented results that the authors can conclude the map addresses the need to incorporate %Al into SOC modeling. The maps do not show any relationship with SOC.
L295: I don’t see any PDP plots in the manuscript.