the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 09 Apr 2026
Identifying sensitive areas of soil landscape evolution by digital soil mapping and complexity analysis in an agro-pastoral transitional zone
Abstract. Soils exhibit various development trends due to differences in initial conditions and external driving forces, leading to spatial and temporal variation in their properties. This study integrates digital soil mapping (DSM) with a modified evolutionary pathway approach (Δcomplexity) to quantify horizontal and vertical soil heterogeneity across an agro-pastoral transitional zone (APTZ) in northern China. We focused on particle size distribution (PSD) and magnetic susceptibility (MS), and quantified Δcomplexity between different soil depths to provide valuable insights into soil-landscape evolution. This study measured the PSD and MS of 1317 samples and determined Δcomplexity for 148 catchments in the study area. The R2 for the prediction accuracy of DSM ranged between 0.481 and 0.729. The Δcomplexity of PSD spans from -0.089 to 0.042 % cm-1 for both deep and shallow layers, whereas that of MS ranges from -0.6 to 0.96 10-8 m3 kg-1 cm-1. Vertical heterogeneity, which is determined by the positive and negative shifts in Δcomplexity in every 5 cm soil, is primarily associated with soil forming factors such as parent material, topography and vegetation, and it is further regulated by tree throw and human activities. For horizontal heterogeneity, which derived from pedogenesis and also soil redistribution induced by both water and wind erosion, is determined by the positive and negative variations in both shallow and deep complexity. Our approach enables the identification of highly heterogeneous areas that may be particularly sensitive to soil change and degradation. It offers a transferable framework for assessing soil development and guiding sustainable land use. The soil Δcomplexity metric provides a promising tool for analyzing and visualizing the state and trends of soil-landscape evolution, and provides a scientific basis for informing land management decisions.
Competing interests: At least one of the (co-)authors serves as topic editor for the special issue to which this paper belongs.
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 paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.- Preprint
(1464 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2026-1837', Bradley Miller, 21 Jun 2026
-
RC2: 'Comment on egusphere-2026-1837', Anonymous Referee #2, 23 Jun 2026
This manuscript investigates the sensitivity of soil landscape evolution in a typical transitional area that is vulnerable to environmental change. It is an interesting work which is based on robust digital soil mapping (DSM) at multiple depths, and then the evolutionary pathway analysis. The latter part is especially meaningful and brings new insights into how we can make DSM more useful, and how to rationally understand and interpret the soil maps. Such perspective is rarely considered by the large amount of DSM studies, and the attempt to move beyond static DSM to dynamic landscape evolution analysis using the Δcomplexity metric is a valuable direction. However, the manuscript still needs to be improved in several aspects. Following includes specific comments.
Major comments
The core for determining evolutional pathway is the changes of soil complexity among various depths. But this issue is not sufficiently described in the Introduction, which makes readers easy to get lost when encounter this in the method part. Specifically, complexity has not been mentioned in the Introduction, although it is so important for the manuscript and it is highlighted in the title. Such core term and theory should be better defined and introduced. For example, what is the link between complexity differences and the pedogenetic processes?
The study uses spatial heterogeneity (Δcomplexity) to infer temporal evolution ("evolutionary pathways") without any chronological control (e.g., radiocarbon dating). The link between spatial variance and temporal trajectory is asserted but not proven. You cannot robustly infer temporal evolution trajectories solely from spatial variance without independent age control. High heterogeneity might indicate active erosion, but it could also simply reflect mixed parent material. You must significantly tone down the claims of "evolutionary pathways." Reframe the study as an analysis of spatial heterogeneity patterns that suggest potential instability or mixing, rather than definitive evolutionary trajectories.
Why authors modify the approach of Van der Meij (2022)? The reasons should be well addressed in 2.5. The 148 catchments should be mentioned in this part because authors calculate the complexity based on such spatial units. How these 148 catchments were determined should also be introduced.
What does the sensitive and highly sensitive areas refer to? What are the differences between them? This should be more clearly described or defined in 2.6 or other parts of the manuscript. You need to validate these "sensitive areas" with ground truth. Do these areas actually show signs of erosion (rills, gullies) or degradation in the field? Without field validation, "sensitive area" is just a theoretical construct. Add a section validating your map against actual erosion features or land degradation signs.
In 2.3, it would be much better to first give a whole idea about what properties authors want to measure and tell readers why. Then, continue with the specifics. Such information might also be mentioned in 2.2, then it is easier to understand why authors implement soil sampling in that way.
The arguments in 4.1 are interesting, but it is difficult to claim the advantages of the complexity method to DSM. These are the two phases of the approach, and the complexity analysis relies on the results of DSM. With the complexity analysis, the DSM results can be better visualized and interpreted. So, the title of 4.1 is suggested to be adjusted.
In Section 4.2, the authors attribute heterogeneity to "tree throw" and "human activities."This is speculative. How do you know it is tree throw and not tillage? Or bioturbation by small mammals? Broaden the discussion of mechanisms. If you cannot prove it is tree throw, refer to it generally as "bioturbation" or "pedoturbation." Compare your findings with other studies in APTZ zones that have identified specific erosion drivers.
Regarding the topic of landscape evolution, the approach of this study seems to be effective for landscape scale issues. Can author give any prospect about the significance of this approach to larger scale soil mapping?
Determination of soil complexity requires a certain area or a group of mapping grids. Catchment was chosen for the unit of calculating the complexity, is there any other possibility of selecting spatial unit for the complexity? For example, how to determine the complexity in flat landscapes? Such issue can be considered in the Discussion.
In both the Abstract and Conclusion, authors present the values of the changes of the complexity, but it is not clear what do these numbers mean and what information readers can get.
The information of the soil types is missing in 2.1, which is quite a pity especially when the soil is the topic of the study. More introduction about the soil characteristics would help readers understand the whole results.
The language of the manuscript is basically good, but it can be refined to make the manuscript easier to be read. Some expression should be improved. For example, In 2.3, “Clay is an important property …” is not so accurate, it should be the clay content. There are still some mistakes in the manuscript, for example, in Line 125, the reference is missing.
Minor comments:
Line 18: % cm-1->"% per cm"
Line 30: "Soil is a four-dimensional natural body" -> Consider a more direct opening
Line 57: "evolutionary pathways of Van der Meij (2022)" -> Ensure you clearly distinguish between Van der Meij's temporal model and your spatial adaptation.
Line 145: define window size for the equation 5
Table 2: Ensure units for RMSE are clear
Line 270: "DSM... lacks of quantification" -> "Lacks quantification" of what?
Citation: https://doi.org/10.5194/egusphere-2026-1837-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 194 | 91 | 19 | 304 | 13 | 16 |
- HTML: 194
- PDF: 91
- XML: 19
- Total: 304
- BibTeX: 13
- EndNote: 16
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
This manuscript presents a novel approach for quantifying soil complexity using digital soil mapping products. It demonstrates the utility for mapping a Δcomplexity metric for soil properties across depths. I particularly appreciate the integration of digital soil mapping with soil-landscape analysis, as this approach has the potential to reveal connections between spatial patterns in soil properties and the processes that shape them. The methodology is innovative and has the potential to advance quantitative assessments of soil heterogeneity.
Despite these strengths, I found the manuscript difficult to follow conceptually. Much of the confusion stems from ambiguity in the use of key terms and confounding factors that cloud the interpretation of connections between the proposed metric and soil evolutionary concepts. In my view, the manuscript is methodologically innovative but theoretically underdeveloped. The work would benefit from a clearer conceptual framework that distinguishes soil heterogeneity, complexity, development, convergence, divergence, and sensitivity to change.
If I understand the authors correctly, the manuscript implies that increasing vertical complexity reflects greater evolutionary sensitivity. However, I do not think this interpretation is fully supported by the string of literature that has developed this framework. I suspect some aspects of these concepts have been slowly lost as they have been applied in different ways across a sequence of papers. The present manuscript would benefit from stronger engagement with the work of Johnson and Watson-Stegner (1987) and Johnson et al. (1987), as well as revisiting the approaches in Phillips (1999, 2000, 2017).
The emphasis on vertical heterogeneity strongly echoes Johnson and Watson-Stegner’s evolutionary model of pedogenesis. They argued that soils simultaneously experience progressive differentiation, regressive simplification, proanisotropic (horizon-forming), and proisotropic (mixing and homogenizing) processes. The present manuscript does a good job of treating both positive and negative changes in complexity as meaningful. However, I think the present manuscript sometimes conflates heterogeneity with more advanced pedogenesis. Profile complexity can be increased or decreased through pedogenic processes, erosion, deposition, faunal activity, tree throw, and anthropogenic disturbance. For example, the authors mention the interception of wind-blown sediments that occurs to different degrees within the study area. A slow rate of this process would result in a progressive pathway of developmental upbuilding as pedogenesis assimilates the influx of sediment. However, if the rate of the same process overwhelms pedogenesis, then the soil shifts to a regressive pathway of retardant upbuilding. While many of these mechanisms are acknowledged, the interpretation often treats Δcomplexity as direct evidence of soil evolution rather than simply evidence of changing spatial organization.
Phillips' framework distinguishes complexity from convergence and divergence. For Phillips, divergence refers to similar initial conditions producing increasingly different outcomes through time, whereas convergence refers to different initial conditions producing increasingly similar outcomes. The present study measures contemporary spatial structure and infers evolutionary tendencies from that structure. Without temporal constraints (observations made over time or via soil dating for space-time substitution) to support independent reconstruction of developmental pathways, the manuscript does not directly demonstrate convergent or divergent evolution. The authors hint at this in the section on potential refinements for future work, but I think without those controls, the interpretation of trajectories from Δcomplexity becomes overextended. According to Phillips, complexity cannot be equated to divergence or convergence in the absence of evidence of trajectories.
The manuscript appears to assume that highly heterogeneous landscapes are more sensitive. I’m not sure I am thinking about sensitivity the same way the authors are, but I don't think heterogeneity always indicates sensitivity. Highly heterogeneous systems may represent stable long-term equilibria, while relatively homogeneous systems may be close to critical thresholds and highly responsive to external forcing. Consequently, the conceptual leap from high Δcomplexity to high sensitivity doesn’t seem fully justified.
Much of the justification for the proposed approach and the assumptions underlying its interpretation appear only in the discussion. I recommend moving more of this rationale into the introduction and methods, particularly regarding why Δcomplexity should be interpreted as an indicator of sensitivity and whether greater heterogeneity necessarily implies greater sensitivity to future change. I also think it would help to establish how the proposed method will be validated, such as through a falsifiable hypothesis or a predefined test that allows for more objective evaluation.
Overall, the methodology is innovative and potentially valuable for soil-landscape analysis. The Δcomplexity metric appears capable of revealing meaningful patterns in the spatial organization of soil properties. However, the manuscript would benefit from a stronger theoretical foundation and a clearer distinction among complexity, development, convergence-divergence, and sensitivity, with additional theoretical development to address multiple developmental pathways. In interpreting the Δcomplexity metric, there needs to be a better explanation, preferably with data, that constrains the potential for the current state of complexity to have been produced by different pathways.
General Comments
Specific Comments
L39 – Would be useful to define the type of scale being referred to here. Are the differences in these studies the extent or the unit of analysis?
L40 – Is a chronosequence study not a study of the landscape?
L48 – Is this implying space-time substitution or something else?
L51 – Please elaborate on how understanding soil heterogeneity is helpful in regions with significant soil variation.
L67 – The study area figure is helpful, but this text could be more descriptive to help us readers better understand the context. For example, pointing out that the QD are limited to some upland flats in the southeast portion of the study area.
L69 – Does this create a potential for developmental or even retardant upbuilding of the soil with influx of new sediment?
L86 – Please add a note about why not all samples could be taken. This could be especially important if some locations had bedrock at depths less than this study’s soil core depth of 40 cm.
L95 – Suggest replacing “laser particle sizer” with “laser diffraction” to be more explicit about the method.
L96 – Please describe the pretreatment method for removing the organic matter.
L98 – Was this three measurements of the same subsample, or three subsamples pulled separately from the whole sample?
L99-100 – Do the authors have any concern about compatibility with pipette method results? Many studies have shifted the clay-silt break from 0.002 to 0.006 mm (American system) to match the results of those methods better.
L122 – The elevation data probably wasn't derived from the DEM, because the DEM is elevation data. In any case, please cite the source of the DEM.
L114-131 – Results of the validation for the digital soil mapping are presented in the results, but the methods of that validation should be provided in the methods section. Please provide the conditions of the validation data set, including how the validation points were selected and how many there were.
L134 – In connecting the digital soil maps with analysis of variability, is there any danger of the digital soil mapping product having more or less variability due to modeling constraints? For example, could there be a tendency towards producing more homogeneous maps at greater depths due to fewer samples and weaker relationships with covariates measured on the surface?
L142-143 – I'm not sure "divergent" and "convergent" are appropriate terms for the vertical heterogeneity. For lateral heterogeneity, we could assume some homogeneity with the original parent material, with spatially variable soil-forming factors driving soil properties to diverge. Thinking about changes in soil-forming factors, I would guess that stronger pedogenesis drives greater heterogeneity at the surface. At the same time, the deposition of new sediment dampens pedogenesis and makes the shallower depths more homogeneous. The authors should explain their rationale for using these interpretive terms, rather than simply extending the use of terms from previous papers to describe different things.
L152 – How were catchments determined? What were their resulting sizes?
L156 –What are these underlying surface conditions?
L159 – Should “changes” be “change”?
L159-160 – I don’t understand this sentence.
L160 – Please remove all extra periods, here and throughout the manuscript.
L170-175 – Is there a potential connection between the uncertainty index and variability?
L185 – No need to change terminology so often. In several instances in this paper, I have been uncertain whether different terms were describing the same thing or whether there was an important difference. In this case, I am more confident that there is no difference between environmental covariates and environmental predictors, so why use different terms?
L190 – Please define abbreviations that appear in Figure 3.
L209 – How does this relate to our understanding of soil geomorphology in floodplains? As a starting point, let's consider that the parent material in floodplains tends to have many layers. Also, evidence for changes in erosion rates in catchments is often found in the morphology of floodplain soils.
L211 – I think it would be useful to define what is meant by “transform.” I think this is the first time the term is used in the results section of this manuscript.
L229 – How the analysis of different environmental conditions was done should be described in the methods.
L247-248 – I think a stronger argument for how analysis of current vertical heterogeneity indicates a timeline of different directions of soil processes. This isn’t the same as a chronosequence.
L272 – Is it the visualization that is missing? Don't we want quantitative measures?
L288-289 – Were spatial connections among the digital soil map’s cells actually revealed? Does variability substitute for connectivity? Wouldn't some sort of topology be needed?
L294 – Because pedogenic processes for vertical complexity do not have a 1:1 relationship with redistribution processes, I'm not convinced by this statement. Mainly, I can think of both depositional and erosional processes that result in a more homogeneous soil profile (at least between depths of 0 to 40 cm). I encourage the authors to either acknowledge their assumption or clearly argue why the vertical complexity metric does not suffer from confounding factors.
L296 – I think the presumption of trajectory based on one point in time needs some additional support. I'm concerned that the assumption is being made that greater complexity equates to a trajectory of greater development. If that is not the case, I think the text could be improved to clarify the basis for the interpretations.
L342 – Perhaps specify the listed soil redistribution processes as primary in this study area.
L369 – Okay, yes, but usually there is some measure of age to determine where the location is on the timeline relative to other locations studied.
L388-390 – These are well-established processes that influence the vertical heterogeneity of soil. However, I'm not sure this study conclusively connects the results of the complexity metric with them. These things coincided, but the case for their connection was largely made in the discussion section. Perhaps different wording could soften these conclusions to align with the data presented in this paper.
L396-397 – This sentence illustrates why I am questioning the definition of sensitive areas. I keep thinking a sensitive area is one that responds more to environmental forces. If a soil is highly susceptible to erosion and loses its surface 40 cm, couldn't the result be a vertically homogeneous profile?
Technical Corrections
L35 – Should this sentence be simplified? The current text seems somewhat circular with “information of soil heterogeneity for revealing soil variation”, unless the authors want to differentiate heterogeneity from variation, or maybe better, variability.
L43 – This citation appears to be missing from the reference list. I realize this is a companion paper to the 2022 paper, but I think it should still be included in the reference list.
L64 – The number of decimal places for the area number seems excessive.
L117 – Fix typo. “b” to “be”
L125 – Clean-up authors' note to themselves to add a needed reference.
L170 – Typo. Change “cay” to “clay”
L210 – Should “catchment” be “catchments”?
L315 – Change “human” to “humans”
L388 – This sentence is awkwardly written. Please reword.
References mentioned in this review that are not already in the manuscript’s reference list
Johnson, D.L., and Watson-Stegner, D.: Evolution model of pedogenesis, Soil Science, 143(5), 349–366, https://doi.org/10.1097/00010694-198705000-00005, 1987.
Johnson, D.L., Watson-Stegner, D., Johnson, D.N., and Schaetzl, R.J.: Proisotropic and proanisotropic processes of pedoturbation, Soil Science, 143(4), 278–292, https://doi.org/10.1097/00010694-198704000-00005, 1987.
Phillips, J.D.: Divergence, convergence, and self-organization in landscapes, Annals of the Association of American Geographers, 89(3), 466–488, https://doi.org/10.1111/0004-5608.00158, 1999.
Phillips, J.D.: Signatures of divergence and self-organization in soils and weathering profiles, The Journal of Geology, 108(1), 91–102, https://doi.org/10.1086/314386, 2000.