the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 11 Jul 2025
LUCATOOv1 – A new land use change allocation tool and its application to the planetary boundary for land system change with the LPJmL model
Abstract. Anthropogenic alterations to terrestrial ecosystems resulting from land use transformation and agricultural intensification represent a significant driving force of global environmental change. The planetary boundary for land system change is one approach to determine an upper tolerable limit to such modifications, measured in terms of the remaining extent of major forest biomes in temperate, tropical, and boreal climatic zones on the different continents. Here, we introduce a land use change reallocation tool (LUCATOO) that can accurately represent the spatial distribution of agricultural land use for different statuses and transgression levels of the planetary boundary for land system change. By representing such configurations of global land cover and land use patterns, the tool facilitates a systematic assessment of the impacts of afforestation and deforestation scenarios on the status of this and other interconnected planetary boundaries. LUCATOO has been developed in the programming language R, is openly accessible, and can be readily adapted for land use change scenarios in applications beyond the planetary boundaries framework.
- Preprint
(1025 KB) - Metadata XML
-
Supplement
(366 KB) - BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2025-2202', Anonymous Referee #1, 12 Aug 2025
-
RC2: 'Comment on egusphere-2025-2202', Anonymous Referee #2, 21 Nov 2025
In their manuscript, entitled „LUCATOOv1 – A new land use change allocation tool and its application to the planetary boundary for land system change with the LPJmL model“, the authors describe the LUCATOO model, a land use change allocation model which can directly be applied on the planetary boundary approach.
This is an important research topic, since drivers of land-use change are often not well represented in existing approaches and their impacts and interlinkages on various environmental systems are complex and understudied. The direct link of this approach to the established framework of the planetary boundaries (PBs) potentially allows to better understand combined land use change impacts on PBs.
The paper is well written and well structured; it also suits well to the journal “Geoscientific Model Development”. Nevertheless, there are several major methodological aspects that in my opinion should be clarified.
First, it should be described in more detail, where the BP-LSC values (85% tropical, 50% temperate, 85% boreal) for the scenario I (at the planetary boundary) exactly come from and how they were estimated or if this is just an arbitrary assumption. I also tried to find out in Richardson et al. 2023 and found the numbers there, but not an explanation how they were quantified. To me, these assumptions largely determine the results and therefore require a more detailed explanation.
In the same way, it is not clear to me, how the values for scenario II (60% tropical, 30% temperate, 60% boral) and scenario III (40% tropical, 20% temperate, 40% boreal) were set. Are there any reasons, assumptions or other studies applying Earth System models underlying these values? Further, given the high uncertainty to these numbers, the authors should apply a sensitivity analysis to demonstrate the sensitivity of the model for different assumptions.
Another question in this context: Why do temperate forests have a much lower value for the ‘safe operating space’? Is this because they have already been deforested by a large extent?
Another major issue affects the definition of the intensification scenario, that in my opinion is also a specific form of expansion, but within regions that already have cultivated areas. Thus, the spatial scale of the approach determines if agricultural expansion is interpreted as intensification or expansion, which can be largely misleading.
The main methodological issue is that the authors uniformly apply intensification or afforestation factors across biomes. This is very unrealistic. Also compared to other land use models, I don’t see any theory behind this allocation algorithm. Other land use models (e.g. Globiom) consider a large range of different factors that spatially vary widely, such as capital, labour, or land productivity (in Globiom, this is also provided by a mechanistic crop model), and costs (e.g. fertilizer costs for intensification that have very different costs in different regions). Most land use models are coupled with CGE or PE models to consider dynamically changing patterns of supply and demand for different regions but also global trade between these regions. All this is not considered in the LUCATOO but would largely impact on the spatial patterns and the degree of both, intensification and expansion.
Therefore, I would be careful applying the model for impact assessments of LULCC on PBs. The identified spatial patterns matter a lot. Therefore, approaches that investigate possible impacts on the environment or on PBs or analyse other trade-offs (e.g. with biodiversity, food security, carbon sequestration, GHG emissions, etc.) should be able to consider main drivers of land use change.That said, the major weakness of this study is a lack of the validation of the approach that demonstrate e.g. if simulated land-use change patterns for historical periods are realistic and can be reproduced.
Another important point that should be considered by the authors and added to the discussion is that land use requirements are only defined by conservation goals in this study. Other sustainability goals, such as food security (SDG2) or renewable energy production are not captured by the PBs? The results of this analysis therefore could lead to trade-off with different SDGs that not considered within the PB approach.
Specific comments:
Ln 16: To state that the model ‘can accurately represent the spatial distribution of agricultural land use for different statuses …’, a model validation would be required that shows that the model is really able to reproduce historical land use transitions. This is however not done by the authors. Due to the lack of a validation approach, I also wonder why the authors use the term ‘accurately’ in this context.
Ln 20: instead of saying ‘is openly accessible’, I recommend providing the exact license here (e.g. CC BY 4.0).
Ln 62f: To assess the impact on the stability of the Earth system, the land-use model would need to be coupled with a global Earth System model. This could also help to quantify the unknown extent to which forest cover must remain intact to sustain safe planetary conditions (as mentioned in line 50). As I understand, this is then again referred to in line 83-86. Maybe, this could be kept together?
Ln 71: There are approaches that investigate impacts of agricultural intensification various key agricultural externalities, e.g. Folberth et al. 2020 (doi: https://doi.org/10.1038/s41893-020-0505-x), and approaches that assess trade-offs for agricultural expansion, e.g. Schneider et al. 2025 (doi: https://doi.org/10.1038/s41893-024-01410-x), and approaches that look at trade-offs between agricultural expansion, intensification, biodiversity and GHG emissions, e.g. Zabel et al. 2019 (doi: https://doi.org/10.1038/s41467-019-10775-z).
Ln 79: Could you cite an article here that demonstrates that LUCATOO has the ‘potential to be used for …’.
Ln 93: ‘match’ here means that it does not exceed the PB-LSC?
Ln 90-102: Where do these numbers come from. Are these realistic scenarios? Until when – there is no time period determined. Particularly for scenario III, I wonder if there is any socio-economic basis for this assumption?
Figure 2: It should also be added to the figure caption of Figure 2, that the values refer to the global biome areas. As such, ‘safe operating space’ also refers to impacts on the global Earth System and does not mean that there could be severe regional risks. Maybe this should be added to the discussion for clarification.
Ln 125f: Several questions on the LPJmL setup: What spatial dataset do you use for different managements, such as irrigation, fertilizer application, sowing dates for the agricultural crops? Is management kept constant over time or does it change over time? How is this handled in scenarios (that are not simulated in this study). Do land-use changes depend on management? This is not getting clear in the entire manuscript.
Ln 128: Why is historical climate model data (GFDL-ESM4) used here and not reanalysis data, such as ‘GSWP3-W5E5’? Which time period is simulated?
Figure 3: According to Fig. 3, LPJmL must run not only historical periods, but for the scenarios also future periods? This is not getting clear from the previous description of LPJmL simulations, in which the authors say that historical climate model data is used for the simulations. This is a bit confusing, because no future scenarios are simulated in this study. Nevertheless, the scenarios refer to the future -> compare Fig. 3.
Ln 145-151: Not clear: How are crops allocated to cells (which mechanism is behind) and what management is assumed here?
Ln 153: Why the first 30 years of the simulation period (1850 – 1879) and not using PI control forcings until 2100 as a reference, which would be more consistent?
Ln 158: This information comes late and fits better to section 2.1.
Ln 164-213: How does the model decide on which pixels intensification or expansion occurs? What mechanisms and theories are behind it? This is not explained.
Ln 171: Is afforestation the right wording here (usually afforestation is done actively by humans) or is it reforestation, or forest restauration? Further, it should be added to the discussion for the afforestation case, that afforested areas require time to take on functions of primary ecosystems, which should be relevant for PB assessments. Are there any elasticity functions in LUCATOO that constrain the transformation between different land uses and covers over time?
Ln 172 and Ln 196: ‘Deforestation can result from intensifying LULCC in cells where CFTs are already present, or by expanding LULCC into currently pristine cells (where CFTs are absent). ‘ and ‘Here, LULCC intensification replaces PFTs with CFTs’.
I understand that intensification often comes with a compaction of land and more monocultures. Nevertheless, the conversion of forest into cropland is an expansion by definition. Anything else is just an effect of the spatial scale of the applied approach (approx. 50 km2 in this study). To me, both cases are an expansion of cropland and do not refer to intensification.Ln 190-200: Intensification and afforestation are applied uniformly across all cells of the biome. Given the spatial heterogeneity of both soils and climate, but also the different socio-economies, the potential for afforestation and intensification varies strongly across regions within a biome. Why not using yields from LPJmL under different degrees of intensification levels (e.g using fertilizer application) or potential yields that can be used as a maximum threshold for intensification and to calculate yield gaps that can be closed by a certain amount for different intensification scenarios. This would result in much more realistic spatial patterns of intensification. Not clear if other feedbacks of intensification on PBs are considered, e.g. due to higher use of pesticides and fertilizers?
Ln 205: How is bioclimatic suitability of the CFT assessed? Please cite data/paper. Why is this done? Is this done at subscale within a half degree grid cell? Are just wetlands and waterways excluded, or also water bodies such as lakes and rivers? What about impervious surfaces, cities, etc.?
Ln 210: Usually land use transitions are not linear over time. Often, sigmoidal functions describe this better.
Ln 251: Suggest to refer to Fig. 6.
Figure 6: Very small and difficult to see details in the maps. Also the resolution in the compiled pdf is insufficient to zoom in.
Ln 265-267: Not clear, this study doesn’t investigate the impacts on other PBs. Also, I don’t agree that this study helps to corroborate the assumed thresholds.
Ln 291-294: ‘LUCATOO opens a new realm of research…’.
I think that's a bit far-fetched. Many other land use models exist, that include a representation of land use change drivers – which LUTATOO is obviously missing completely. Other existing land use models in principle can also be applied on PB impacts.Ln 295: ‘Our land use change reallocation tool is easily extendable’.
This has already been stated several times. To investigate other Earth System processes, I would not suggest using this model, due to the already mentioned weaknesses.Ln 300: Intact forest biomes are not necessarily hotspots of forest biodiversity. I would say ‘and thus can be hotspots of …’.
Ln 310: Similar assumptions are used by most of the existing land use models. Often, existing road networks are used to determine future agricultural expansion. Nevertheless, this assumption could quickly become obsolete, for example if a new road is built through the rainforest.
Ln 328: This reads a bit ignorant. Why can other approaches not be adjusted or modified to depict specific anthropogenic pressure levels? Actually, they already do.
Citation: https://doi.org/10.5194/egusphere-2025-2202-RC2
Model code and software
LUCATOO - A new land use change allocation tool and its application to the planetary boundary for land system change with the LPJmL5 model Arne Tobian et al. https://doi.org/10.5281/zenodo.14525230
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 1,918 | 59 | 25 | 2,002 | 47 | 29 | 41 |
- HTML: 1,918
- PDF: 59
- XML: 25
- Total: 2,002
- Supplement: 47
- BibTeX: 29
- EndNote: 41
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
General comments
This study introduces a new land allocation tool to produce spatially explicit datasets consistent with different states of the LSC planetary boundary. The conceptual framework is enticingly simple and forms a novel contribution. Using the PB framework, the authors construct unique scenarios which provide a much-needed contribution to the currently limited diversity of LULCC scenarios in the literature. However, the simplicity of the approach is not supported with sufficient discussion about its potential limitations. In particular, the manuscript would benefit from further discussion about the land use allocation algorithm employed which omits key factors known to influence LULCC, including land productivity, land use and conversion costs, and socioeconomic factors. Including these factors could result in significantly different patterns of LULCC than those simulated here and add further layers of uncertainty not explored.
Specific comments
(L47) While forest loss is a prominent example, it might be worth highlighting that the impacts of LULCC are not just limited to forests but can affect a diverse range of ecosystems. For example: peatland drainage resulting in carbon emissions, soil carbon loss due to overgrazing of semi-natural grasslands. Additionally, LULCC impacts aren’t necessarily binary (e.g. forest vs. no forest) but can exist on a continuum based on different levels of land use intensity. While this study focuses on absolute deforestation and afforestation, it’s important to note the complexity of forest degradation more generally.
(L50-54) A little more background about how the 50% and 85% boundaries were chosen would be useful. It’s not clear how the third sentence of that paragraph leads on from the previous two. Where are we currently with respect to these boundaries? I think there needs to be more justification for setting a lower boundary for temperate forests or at least a discussion of the limitations of this assumption. The original source (Steffen et al. 2015) states “this is a provisional boundary only” – are there no recent updates? This lower boundary is justified in Steffen et al. by referencing Snyder et al 2004. However, I find this claim is not at all clear just from that paper - the climate impacts of removing temperate forests seems comparable to other biomes.
(L167-169) Why were continents chosen as the regional boundaries? That seems somewhat arbitrary. Wouldn’t ecologically relevant boundaries such as ecoregions be more appropriate? Or national boundaries given the importance of domestic policies.
(L197) Uniform intensification is rather unrealistic. LULCC is influenced by a range of factors including land productivity and costs, national and international demand for commodities, and proximity to existing managed lands. This leads to complex LULCC patterns which are rarely spatially uniform, particularly on continental scales. A discussion on the limitations of this assumption is needed.
(L215) Figure 5. The notation used in the figure could be improved. It’s not clear whether i, m, p etc. are parameters, variables or sets. For example, using the key in the top right corner, I would translate “i < LU_scn” as “subset of scenario dataset is less than scenario dataset” – while I can guess the intended meaning with the help of the caption, it’s perhaps a bit unconventional. Maybe it would be clearer with something like Si < SLU_scn where S is the LSC boundary variable. “m = i*fac.re[m]” and similar is particularly difficult to parse – is this representing the transformation of subset i into subset m? In which case, perhaps this could be written as “m = fre(i)” where fre() is the reduction function?
(L226-230) Some background information on previous assessments would fit well in the introduction (see comment for L50-54).
(L255) Figure 6. Higher resolution image needed. It’s interesting that each scenario shows either reforestation (planetary boundary) or deforestation (risk and strong transgression) in all biomes but not a mixture of both. Why is that? Given that different biomes in different continents are at or below the planetary boundaries (Table 1), shouldn’t result in a more heterogenous response? Also, the uniform application of the intensity factors within each biome is very apparent here. I think there needs to be discussion whether this is realistic, given that observed LULCC is spatially (and temporally) heterogeneous.
(L265-267) It’s not clear how this has been demonstrated. You have produced maps consistent with the PB-LSC boundary but there was no further analysis of how other PBs are affected under this scenario. Or is this referencing Richardson et al. 2023 (as it appears so further down)?
(L289) What did Drüke et al. 2024 find?
(L295-298 and L306) This is an important point of discussion that should be expanded on (also see previous comment)
(L310-321) – As previously commented, the reallocation of CFTs based purely on area is an important limitation here. A more detailed allocation tool would consider other factors such as potential yields, land suitability and production costs as well as trade-offs between agricultural expansion and intensification. Similarly, afforestation could be prioritised based on preservation of ecosystem services such as biodiversity and carbon storage. On a more fundamental level, it’s also not clear whether the scenarios presented here are internally consistent – for example, is the amount of deforestation in the strong transgression scenario even feasible given socioeconomic constraints? How much demand growth (food, timber etc.) would be required to cause this much deforestation?
(L329) “cannot be adjusted or modified to depict specific anthropogenic pressure levels” – to the contrary, many land system models work explicitly with “anthropogenic pressure levels”, although these can be expressed in different ways (e.g. demand for commodities, marginal utility of ecosystem services). Prominent examples include the major IAMs (IMAGE, REMIND-MAgPIE etc.) and other frameworks such as LandSyMM. While these haven’t extensively explored the PB framework, there’s no reason why PB-oriented scenarios couldn’t be constructed within these models.
Technical corrections
(L65) Replace “allocation models” with “land use models”
(L90) “The following _” section?
(L260-264) Too repetitive and non-specific, particularly “bridge the conceptual gap of an adjustable depiction”