Thank you for carefully reading our manuscript. We will take the helpful comments into account in the revised manuscript.

Citation: https://doi.org/10.5194/egusphere-2024-4036-RC1

RC1: This article is investigating the role of snow on treeline response to climatic 

warming by including snow (in addition to other abiotic and biotic variables) module in the 

vegetation model LAVESI. I like the idea with the model LAVESI. Results showed site 

specific response of treeline shifts and highlighted the role of snow in treeline dynamics. 

The approach including additional variable in vegetation model is interesting. The data 

are enough and relevant to test the authors’ hypothesis. I recommend a minor revision.

AC: We sincerely appreciate your valuable feedback and constructive comments 

during the first review of our preprint. Your comments will significantly enhance 

our manuscript. We will carefully address all referees' suggestions to further 

improve the quality of our study and ensure it meets the high standards of 

Biogeosciences.

RC1: I like your introduction, but it is a bit too long, it would very nice if you are able to 

make it more concise by reducing it by 10-20 % pages. 

AC: We will shorten the introduction while preserving all essential content.

RC1: Recently several studies reported role of biotic interactions on treeline dynamics. 

Please see Bader et al. 2021, Ecography, Liang et al. 2016, PNAS, Sigdel et al. 2024, 

Nature Plants.

AC: Thank you for the additional literature suggestions. We will consider to 

include them where fitting in the text.

RC1: Overall, the discussion is general and comprehensive for me and mostly authors 

focused on those influences and processes that have been investigated. It has been 

neglected role of snow duration, i.e., the period when snow covers the treeline, 

determines processes of tree regeneration and establishment. 

AC: The snow module accounts for snow duration, i.e., the period during which 

snow covers the treeline area: 

Building on the approach of Sato et al. (2007), the processes of snow 

accumulation and snowmelt were incorporated into the LAVESI model (see lines

231-232). As a result, the duration of snow cover is influenced by snow depth, 

which is affected by both snow accumulation and snowmelt. All subsequent 

influences on tree growth and treeline migration, which are dependent on snow 

depth, are consequently impacted throughout the duration of the snow cover. The 

effects of varying snow depths were implemented in the model in the following

abstracted manner:

If the height of the tree is higher than the current snow depth, the mortality rate of 

the tree increases by 50%. This penalty decreases linearly with increasing tree 

height and is eliminated once the tree reaches a height of 5 m. 

When seeds are covered by snow, their germination is enhanced, exhibiting a 

linear increase in germination rates with greater snow depth. Specifically, this 

relationship begins at zero and reaches a 30% increase in germination at a snow 

depth of 1 m. 

If seedlings and saplings are covered by snow, their growth is enhanced by 30%. 

The germination process is influenced linearly, either positively or negatively, 

based on whether the snow melts before or after the 110th day of the year. 

The actual day of snowmelt is divided by 110, and the deviation from 1.0 is added 

to the germination probability (see lines 237-243).

In general, the growing season length and its influence on seedling germination 

and tree growth are already incorporated into the vegetation model (see lines 187-

189).

We will provide a clearer explanation of these connections especially of the snow 

period length in the updated version of the manuscript.

RC1: Related to this, snow fungi might also explain part of the spatio-temporal pattern. 

AC: The influence of snow fungi was indirectly incorporated into the snow module, 

as prolonged high snow load or a shortened growing season negatively affects 

tree growth. In the discussion, we will highlight that snow fungi may also 

contribute to the observed spatio-temporal pattern.

RC1: Also snow movements and avalanches may affect the observed species differently. 

AC: The snow module includes snow load and avalanche occurrence. However, 

we did not vary these effects by tree species, as the simulation areas in our study

– as well as the North American and Siberian treeline ecotones in general – are 

predominantly dominated by spruce in Northwest Canada and Alaska, and by 

larch in Siberia.

We will include species-dependent differences in treeline migration in the 

discussion to provide further insights for studies focusing on species-specific 

responses.

RC1: Additionally, even though, different plant traits and interactions are included in the 

model, their influence on treeline sensitivity and migration is missing. 

AC: Species-dependent plant traits are incorporated into the vegetation model. 

Since the simulation areas are dominated by spruce or larch, we consider the 

influence of other tree species on the sensitivity analysis as negligible.

 

RC1: It is not clear why site-specific specifics sensitivities are observed? What are the 

mechanisms behind such discrepancies? It needs to be justified based on site specific 

evidences. Authors discussed different factors (environmental and traits) but fail to link 

site specific variability. Thus, it needs to synthesize the results rather than presenting 

results directly. 

AC: We acknowledge the need for a more integrated discussion of site-specific 

sensitivities. Rather than treating the study sites separately, we will synthesize the 

results to highlight overarching patterns and mechanisms driving the observed 

differences. By linking environmental factors and species traits more explicitly, we 

aim to provide a clearer, more comprehensive understanding of the site-specific 

variability in treeline dynamics.

RC1: Generally, authors just compare their results with other similar studies. To make it 

convincing, it needs deeper discussion with key scientific evidence and how it helps to 

advance our understanding on treeline ecology under changing climate.

AC: In the discussion, we will highlight the implications of our results for the 

future of North America's treelines in the context of global warming.

RC1: Also, ecological implications of the modified model should be highlighted.

AC: In the discussion, we will also further elaborate on the ecological implications 

for North American forests, including biodiversity, the albedo effect, and their role 

as a carbon sink.

RC1: L13-14: As site wise results are presented, it is better to mention the areas/sites 

considered in analysis.

AC: We will mention the study sites at this point in the abstract.

RC1: L64-66: Please see recent treeline studies used individual-based model to simulate 

plot-based data including both biotic and abiotic variables (Zheng et al. 2024, Eco Lett).

AC: Thank you for the literature suggestion. We will carefully review the methods 

and findings of this study.

RC1: L184: Please mention the sites considered in this study before this sentence.

AC: We will introduce the study sites earlier in the manuscript.

RC1: L307-308: Even though the trees traits data were adopted from previous 

publications, better to elaborate how these data were retrieved and sources.

AC: We will compile a complete table of species traits and model variable values in 

LAVESI and include it in the manuscript appendix.

RC1: Table 3 caption: impact factors could be replaced by predicting variables.

AC: We will change this term according to your suggestion.

RC1: No any parameter is significance from Road to central Alaska. What is mean? Is 

this model not well able to predict the sensitivity of Alaskan treelines?

AC: At the Road to Central, Alaska site, no single factor was found to be 

statistically significant. This may be due to stronger interactions between multiple 

factors, which could obscure the effects of individual parameters. Another 

possible explanation is that this site is more influenced by stochastic processes.

RC1: L392-395: It seems to methodology rather than results.

AC: We will relocate this section to the Methods section as suggested.

RC1: L464-469: Just speculations, no evidence supports the findings.

AC: As mentioned above, we will provide a more detailed explanation of the site-specific differences in the sensitivity analysis results, thereby clarifying or reassessing the significance of this section.

RC1: L532: either data or reference should be presented.

AC: We will insert a reference to ‘Table 1: Characteristics of the study sites.’ at this 

point.

RC1: L550: How did species-specific stand structures varies across the sites. It should 

be site and species specific. How population structure varied across the study sites? and 

so on.

AC: Since spruce or larch clearly dominates at all three study sites, we have not 

yet conducted a detailed comparison of species composition. However, we will 

include a general comparison of the stand structure of the study sites in the 

manuscript.

Citation: https://doi.org/10.5194/egusphere-2024-4036-RC2

RC2: The manuscript entitled "The significant role of snow in shaping alpine treeline 

responses in modelled boreal forests” aims to investigate the effect of snow on the 

treeline shift. However, in my opinion, reading the article looks like this is not the main

focus of the manuscript. The structure of the article is center on the sensitivity analysis of 

the forest dynamic model LAVESI. In the methodology section is reported the description 

of a new way to incorporate the effect of the snow cover in the LAVESI model, but the 

results section is just partially focused on the outcomes of this process. Some issues in 

the incorporation of this process are detected and reported in the specific comments 

below. Moreover, the study areas are not adequately described and neither pictures or 

orthophoto are reported. For me it is complicate to understand the real characteristics of 

the three investigated sites.

AC: Thank you for your insightful feedback and constructive critique during the 

first review of our preprint. Your comments provide important guidance for 

improving our manuscript. We will carefully consider all referees' suggestions to 

enhance the clarity and quality of our study, ensuring it aligns with the high 

standards of Biogeosciences.

RC2: The approach adopted in this manuscript is the progressive variation of the model 

inputs and so it is basically a sensitivity analysis of the LAVESI model. 

AC: We did not conduct a progressive variation of the model inputs but, as 

correctly noted, performed a sensitivity analysis. In the context of pattern-oriented 

modeling, this is a standard approach in ecological individual-based modeling 

(see Grimm & Railsback, 2005, Individual-based Modeling and Ecology, Princeton 

University Press; Grimm et al., 2005, Science, 310: 987-991).

RC2: Furthermore, there is no comparison between the model outputs and the actual 

forest characteristics making impossible to evaluate the reliability of the model outcomes 

and consequently the future scenarios. 

AC: In pattern-oriented modeling, a simple model is iteratively refined to more 

accurately reproduce observed patterns. As part of the validation process, model 

outputs are compared with actual forest characteristics. While we have fitted the 

treelines, this was not done in a spatially explicit manner. The original LAVESI 

model (Kruse et al., 2016) has been systematically extended to incorporate 

dispersal processes influenced by wind speed and direction (LAVESI-WIND 1.0; 

Kruse et al., 2018), landscape topography and additional boreal forest species 

(LAVESI-CryoGrid v1.0; Kruse et al., 2022), and forest fire dynamics (LAVESI-FIRE; 

Glückler et al., 2024). Both the pattern-oriented modeling approach and the 

sensitivity analysis have been consistently applied across these studies.

RC2: The manuscript need to be deeply restructured and also the methodological 

approach deeply revised. In my opinion the final suggestion for the manuscript is a 

rejection encouraging the authors to work on it for a resubmission.

AC: As also pointed out in Referee Comment 1, we will revise and clarify the 

description while maintaining the methodology, as it aligns with established 

ecological modeling standards.

RC2: Below I report the specific comments.

Introduction: the section is too long and I suggest the authors to reduce it of about 50 

lines. Moreover a general comment to this section is to restructure the description of the 

drivers and factors influencing the treeline shift. As it is, in my opinion it results a little bit 

complete to follow.

AC: We will shorten the introduction while preserving all essential content.

Currently, the influencing factors are categorized into abiotic climatic factors,

species-dependent factors, abiotic non-climatic factors, and other biotic factors. 

We will further emphasize this classification and illustrate it with a graphical 

overview.

RC2: Line 187: for run, the authors intended years? So, the model time resolution is 

equal to 1 year if I understood correctly. Please specify it.

AC: The model description outlines that within one simulation step, which is one 

year, the relevant processes are incorporated consecutively as submodules (see 

lines 175-176). We will reiterate this in line 187 for clarity.

RC2: Section 2.1: how do the species-specific parameters included in the model were 

derived? Please refer also to the species you are going to simulate reported in table 1.

AC: We will compile a complete table of species traits and model variable values in 

LAVESI and include it in the manuscript appendix. We will provide a clearer 

explanation of the site-specific results by directly linking environmental factors 

and species traits to site-specific variability, thereby enhancing the understanding 

of the observed differences across sites.

RC2: Line 194: How does the tree mortality rate is included in the model? Please specify 

what tree and environmental related parameters where used to model this process.

AC: The fundamental methodology for determining seed and tree mortality was 

originally developed by Kruse et al. (2016) and has been updated since. We will 

provide a clear explanation of how the tree mortality rate was calculated, detailing 

the environmental parameters that influence this process and the mechanisms 

behind their effects.

RC2: Line 221: the threshold of 5°C for distinguish between rainfall and snowfall needs 

to be accurately motivate. The value of 5 sounds quite high in my opinion. How do you 

motivate it? Temperature daily fluctuation in case of precipitation is really low so I would 

expect that a mean daily temperature slightly less than 5°C will lead to rainfall 

precipitation. This is a crucial point, and it needs to be clarify and accurately motivated 

since it can majorly affect the simulation results.

AC: Our approach is based on the empirical function from Sato et al. (2007), which 

in turn refers to the precipitation partitioning approach from Ichii et al. (2003). Ichii 

et al. (2003) used a gradual transition, where snowfall accounts for 50% of total 

precipitation at 2°C. In our model development, we initially incorporated this 

approach but subsequently adjusted the threshold as part of a model tuning 

process. Specifically, we compared simulated and observed snow accumulation 

patterns and iteratively refined the temperature threshold to improve model 

accuracy. However, we acknowledge that this value may vary depending on 

regional climate conditions, and we will clarify this tuning process more explicitly 

in the revised manuscript.

RC2: Line 245-250: about avalanche induced mortality, the authors never mentioned the 

role of the slope as predisposing factor for snow avalanche release. It is generally 

defined that a slope between 27° and 55° is the range where avalanches can occur.

AC: We implemented snow movements downslope in extreme years which happen 

with a chance increasingly when exceeding the 99.9th percentile of snow height in 

the full climate forcing series. This movement is simply implemented as 

downslope independent from the slope angle. Further, in this simplification we 

named this process avalanche. In the snow module description, we then described 

the process and stated that mortality is added where a modelled avalanche 

impacts the leading edge of tree stands, with its destructive force diminishing 

downslope due to protecting tree growth (see lines 250-251). We will provide a 

more detailed explanation that snow accumulation and snow movement were 

modeled for each grid cell, considering the slope direction.

RC2: Section 2.3: For me this section results slightly complicate to follow because it is a 

pure description of the model and the relative processes. I would encourage the authors 

to use formulas and/or diagrams or flowcharts in order to have also a clear view of the 

developed model.

AC: We will include additional formulas where needed and in-text formulas are not 

sufficient. We will create a flowchart to clarify the interrelated processes. 

Additionally, a graphical overview of all influencing factors will be provided to 

facilitate a better understanding of the overall framework.

RC2: Table 1: are the locations reported in a local or a geographic coordinate system? It 

looks to me east north but the xmin etc. is confusing me.

AC: We will update the names to 'Longitude' and 'Latitude' for clarity.

RC2: I would suggest to add also the total area modelled for each study site.

AC: We will also specify the total simulation area for each study site at this point.

RC2: A small map would be beneficial to clearly identify the location of the investigated 

sites.

AC: As requested, we will include a map to visually illustrate the locations of the 

study sites.

RC2: Lines 304-308: Why do the authors compute the sensitivity analysis of the 

parameters of the model? For me it is unclear the motivation of this operation.

AC: As previously mentioned, this approach is a standard method in ecological 

modeling (Grimm & Railsback, 2005; Grimm et al., 2005). The methods section 

further explains that the sensitivity analysis is conducted to evaluate and compare 

the influence of various factors on the migration rate of the alpine treeline and to 

identify key drivers and assess how variations in these factors influence treeline 

dynamics (see lines 299-300).

RC2: Section 5: for really understand the morphology of the sites and the condition it 

would be beneficial to add an image with the orthophoto of the investigated areas with 

contour lines or something similar.

AC: Satellite images of the study sites are already included in the preprint. 

However, we have noticed that Figures A1 and B1 mistakenly display the same 

image, which we will correct. We initially decided against including these images 

in the main section to avoid excessive length but are happy to move them if 

desired. Additionally, we can provide photos and full panorama images of the field 

plots online.

RC2: Section 5: It would be of great value if the authors can compare the actual 

condition of the study site with the results of the spin up simulation. In this way it is 

possible to assess the reliability of the model and therefore the validity of the future 

scenarios.

AC: The treeline was adjusted to match the current observed position through a 

fitting process, which, although conducted, has not yet been explained in detail 

within the manuscript. We will address this more thoroughly in the revised 

version. However, it is important to note that achieving an exact match in a 

stochastic model is inherently challenging. The model generates mean positions 

based on multiple runs, with some simulations advancing further while others lag 

behind. This variability is a result of the stochastic nature of the model, and it 

allows us to explore the associated uncertainties. For further clarification, please 

refer to the section on stochasticity in Grimm & Railsback (2005).

RC2: Line 330: What is the definition of treeline? It is necessary to define how the 

authors have identified it. Is it based on a certain density of trees or something similar?

AC: The definition of the treeline in this study is described in lines 328-329: 

Treelines were determined using threshold-based criteria of one hundred trees per 

hectare.

RC2: General comment for the methodology section: Authors investigated which factors 

are the most important/influent for the migration rate. I would say that the common 

approach for simulating future scenarios is to have a solid and reliable model either 

tested with field data or from literature in similar conditions. Consequently, models are 

used to predict future change by just modifying the environmental variables or the initial 

conditions, obtaining in this case the probable shift of the treeline. In this manuscript the 

procedure is not classical, and the authors look like they want to investigate which 

variable of the model can affect the treeline shift and how. To me this approach is more 

similar to a simple exercise respect to really understand the possible dynamic of the 

treeline shift for the three study sites. This feeling is corroborated by the absence of 

comparison between the current condition of the forests and simulation spin up for the 

three study areas. In other word the factors influencing the tree line shift are the 

environmental one and not the parameters of the model.

AC: The aim of our study and sensitivity analysis was to evaluate and compare the 

influence of various factors on the migration rate of the alpine treeline, identify key 

drivers, and assess how variations in these factors influence treeline dynamics. 

As outlined by Grimm & Railsback (2005) and Grimm et al. (2005), this approach is 

a standard method for determining the influence of input factors on model output. 

This methodology has also been applied in several other LAVESI studies, such as 

Kruse et al. (2016, 2018, 2022) and Glückler et al. (2024). Similar approaches have 

been used in other studies, including Zurell et al. (2011), who linked species 

distribution models with an individual-based model to assess climate-induced 

range shifts in black grouse, and Malchow et al. (2024), who used a spatially 

explicit individual-based model to analyze red kite population dynamics. 

Additionally, Zurell et al. (2018) applied species distribution models to evaluate the 

risks faced by migratory birds due to climate and land cover changes. These 

studies demonstrate that integrating process-based models with sensitivity 

analyses is a widely used method for understanding ecological dynamics under 

environmental change.

RC2: Line 359: please add a couple of sentences to introduce the results instead of 

directly showing Table 3.

AC: In accordance with the suggestion, we will add an introductory sentence at 

this point.

RC2: Table 5: The table should be moved to the results section.

AC: As suggested, we will move Table 5 to the results section.

RC2: Lines 458-462: I would say this is a quite obvious result. The study areas are 

different, and I would expect exactly this result.

AC: We will provide a more detailed explanation of the site-specific differences in 

the sensitivity analysis results, thereby clarifying or reassessing the significance 

of this section.

RC2: General comment to the results and discussion section: The authors report the 

outcomes of the modelled scenarios discussing and comparing the relative results, but 

the real problem is that a reliable scenario has not been identified. In the manuscript 

there is no identification of a modelled scenario that can be representative of 

current/actual forest conditions and of the future one. Without this definition is not 

possible to quantify the alpine treeline shift.

AC: We appreciate the reviewer’s concern regarding the identification of a 

representative scenario for current and future forest conditions. However, our 

study is not designed to provide deterministic predictions of future treeline 

positions. Instead, it is a sensitivity study aimed at understanding the relative 

influence of different factors on treeline migration dynamics. Our model was tuned 

to represent observed patterns, but the primary objective is to assess the key 

drivers and interactions shaping treeline shifts rather than to predict an exact

future distribution. By systematically varying input factors, we can identify the 

processes most relevant for treeline migration, which can help refine future 

projections and improve ecological understanding rather than provide a single 

"correct" scenario. We will clarify this point more explicitly in the manuscript.