Potential Distribution and Suitable Habitat for Chestnut (Castanea sativa)

Round 1

Reviewer 1 Report

minor flaw only mainly in how scientific names are written. if it is in accordance with the journal requirement, please ignore my comments. Comments highlighted/marked in yellow.

Comments for author File: Comments.pdf

Author Response

Please see the attachment

Author Response File: Author Response.docx

Reviewer 2 Report

The work presented concerns the potential range of an important species such as the chestnut C. sativa. Data from the Caucasus were used to estimate the potential range in both the Caucasus and Europe. The result of the comparison of two databases (WorldClim and Chelsa) is an interesting output. I will certainly cite this article in the future when it will be published. However, I have a few comments:

The text is a little messy; some sentences are unnaturally split into two; many latin names are not italic, there is a bunch of unnecessary spaces. Sometimes “chestnut” is written in the middle of the sentence by capital letters and sometimes not. I think the text should be carefully check by the authors.

L57-59 – this fragment about precipitation is unclear; 500-3500 mm is truly optimal range or just suitable range? In Europe minimal values are 600 and 800 mm (why two values?) so 500 mm shouldn’t be in the optimal range

Figures 5 and 6 should be improved, because it is difficult to compare it as they have completely different colour set. I suggest to add the legend with colours, to know the values of suitability and to change the maps to use one colour set.

I am not fully convinced by the explanation that snow cover is that important for the species. Most of its range covers areas with a Mediterranean climate where snowfall is rare, and the impact of precipitation in the cold quarter can easily be explained by the fact that most precipitation in this region is winter precipitation. Can the authors discuss this thread more or add any references/observer references?

Authors collected part of spots from the GBIF database. In this case, it is necessary to add the link to the created by GBIF list of spots. It should be added in the Materials and Methods section or directly in the References section.

 

Minor technical remarks:

L20 – unnecesary space before „Historic”

L35 “relict {..} from the last glacial maximum” sounds strange, because it suggests that chestnut Is a glacial relict. It should be rephrased.

L620 – unnecesary space before „However”

L 64-65 1,000 m is with colon and 1450 not – it should be one style

L68 – Lack of dot after “soils”

L67 – Phytophthora cinnamomi and Cryphonectria parasitica should be italic, also it should be authors abbreviation there

L76 – Lack of space after “[18]”

L89 – latin name in the bracket is unnecessary

L99  - word “In” should be not capitalized

L162 - unnecesary space before „The biggest”

L184 - unnecesary space before „4 coordinates”

L201 - unnecesary space before „The occurences”

L246 - unnecesary space before „The models”

L335, L339 – should be “Hyrcanian” not “Hyrcani”

L382 – should be “Lesser Caucasus”, not “Laser Caucasus”

L383 – Black Sea should be capitalized

L527 – why here is new paragraph? the last sentence of previous one start the topic

 

The text is generally comprehensible, although there are minor errors. Some sentences could be combined to maintain fluency.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 3 Report

In the article titled "Potential Distribution and Suitable Habitat for Chestnut (Castanea sativa)", the authors modeled the habitat requirements of chestnut, using presence-only data collected in Georgia, Turkey and Azerbaijan. The authors used two sources of climate data, CHELSA and WorldClim. Part of the study is the comparison of the two sources of climate layers.  The paper is susceptible to some improvements.

There are two aspects that need to be clarified in the document. It is probably just a matter of wording.

The first is concerning sampling.

a) In the abstract, the authors say that they used 620 occurrence locations, which coincides with the number described in L.173-191. But what was done with the 37954 points obtained from GBIF (L. 205) were they only used to validate the results? If so, emphasize to the audience from the beginning of the paragraph.

b) the type of sampling used does not seem to be the same in all cases. For example, the authors write that stratified random sampling was used for the 166 plots in Turkey and Georgia. But, the rest of the sampling points. What type of sampling was used?  

c) On the other hand, the 330 sample plots provided by the Ministry of Environmental Protection and Agriculture of Georgia are not duplicated with the information that the authors established (L.173-174).

The second point is related to the historical anthropogenic expansion of the Chestnut's range mentioned in the Abstract. However, this is no longer addressed in the discussion, nor elsewhere in the paper. The authors should discuss it further in the light of their results.

Other aspects that could be improved:

L. 53, 57 and others... Avoid excessive use of "However" throughout the document. It is repeated 21 times throughout the document.

Fig.1.- Avoid overlapping the world map with the area studied.

L. 175. About "old forest inventory maps", how old are the maps?? Mention the year or period, please.

L.180 -190. Avoid starting a sentence or paragraph with numbers. For example, instead of writing, "330 presence sample plots were provided by [...]", perhaps start the paragraph as follows: Georgia's Ministry of Environmental Protection and Agriculture provided 330 sample plots,....

L. 173-194. Improve the wording. It is still confusing as is. There is a lack of fluency.

Citation N. 48. Check if the GBIF citation is correct.

 

L. 205. How did you ensure that the 37954 presence records obtained from GBIF were not the same as those obtained from other sources described in L.180-194?

 

L. 220, on "pairwise correlations". Which coefficient did you use, Pearson's linear correlation coefficient? Clarify.

 

L. 223. Rephrase the following paragraph: "119 uncorrelated variable combinations...".

L.225. "we selected 4 additional groups of the factors". Mention which ones they are.

Improve the resolution of Fig. 4 AND perhaps letter each panel.

 

L419-420. Disagree. The use of various climate sources is not rare. It is very common. Here are some works.  I do not put these papers to be cited. I put them to show the authors that there are already several papers.

WorldClim v. 2 (www.worldclim.org/version2), MERRAclim (Vega et al., 2016, 2017), CHEL SA (w w w w.chelsa-climate.org), and ocean databases MARSPEC (www.marspec.org) and Bio-ORACLE v. 2.0 (www.bio-oracle.org): DOI: 10.1111/2041-210X.13124

Worldclim and Chelsa datasets: DOI: 10.2495/SDP-V14-N2-105-117

 Worldclim and Chelsa: doi: 10.3897/neobiota.59.36299:

Etcetera.

L. 462.Regarding the claim the authors make about "WorldClim model is less reliable" I think it is worth discussing a bit more and contrasting these results with those of other authors. Also, maybe the authors should look for more work with other approaches and methods of analysis.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 4 Report

This paper focuses on the potential habitat distribution of Chestnut (Castanea sativa) in the Caucasus using the Maxent model, comparing the CHELSA climate dataset with the WorldClim climate dataset, and validating the model using the habitat electivity index (HEI).

1. Why did you choose CHELSA version 2.1 data instead of using another database (GLDAS, and MERRA-2)?

2. Why is there such a difference when CHELSA and Wordclim have different variables (Table 2, ID 3 and ID 6) after conducting the screening? There are nine variables in common, and the nine variables they have in common should be used as the final variables. The model is closely related to his choice of parameters, the more or less variables will lead to an impact on accuracy, usually building models using r ≤ 0.8, further screening is recommended.

3. Line 223, why is there a combination of 119 unrelated variables.

4. The distribution point data used differed significantly in time from the bioclimatic data and whether this difference could cause some problems.

5. Insufficient literature citations at the habitat electivity index (HEI).

6. The AUC values in the model comparisons all exceeded 0.95, so why would a Habitat Selectivity Index (HEI) < 0 be present (Table 3, ID 1 and ID 3).

7. The contribution of the five variables in Figure 3 should add up to one, and some of the individual variables are already contributing more than one.

8. The introduction states that Chestnut requires an average temperature above -3.5 °C in the coldest month, which corresponds to the response curve of BIO11 in Wordclim (Figure S6), but not to that of BIO11 in CHELSA (Figure 4). The response curves for the Wordclim variables (BIO11) are more biologically appropriate, the CHELSA variables do not match the biological growth characteristics, and the Wordclim data are more appropriate than the CHELSA data. the AUC values as well as the HEI values also illustrate this point (Table 3, ID 2 and ID 5).

9. The images are not clear enough, for example, the global map of Figure 2 can be shown without the region name, the y-axis of Figure 4 is not clear enough, and the region name of Figure 6 is not clear.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 4 Report

There are still some questions that have not been responded to replies, and some of answers are hardly going to convince me.

Point 1: After screening, CHELSA and Wordclim had different variables (Table 2, ID 3 and ID 6), of which 9 were common. However, since the comparison is between two databases, it is more appropriate to use the 9 variables they have in common as the final variables. This comparison is only meaningful and more credible.

Point 2: How were the values of the variables in Figure 3 calculated? It is not possible to accurately see the extent to which each variable contributes, and it is suggested that they be replaced with contribution rates.

Point 3: We did not find agreement in the statements of Dolukhanov (2010). Chestnut (Castanea sativa) is a warm-temperate deciduous species, that likes a mean yearly temperature ranging between 8 °C and 15 °C, and monthly mean temperatures over 10 °C during 6 months, which is in agreement with the statements of Conedera et al. (2016). In the response curves of BIO10 and BIO11 of CHELSA (Fig. 4), when the hottest seasonal mean temperature and the coldest seasonal mean temperature reach the suitable annual mean temperature, it is instead unsuitable for Chestnut. In the response curve of BIO10 of CHELSA, the probability of the presence of Chestnut is less than 0.1 when BIO10 is located between 8 °C and 15°C, and the probability of the presence of Chestnut is 0.5 when BIO11 is located between 8 °C and 15°C, which is the lowest point of the curve. In Afif-Khouri 's et al. (2011) article, the average summer temperatures in north-west Spain are between 15.70 °C and 17.50°C, and north-west Spain is a very suitable area for Chestnut. This agrees with the response curve for BIO10 in Wordclim (Fig. S6), but not with the response curve for BIO10 in CHELSA. In the introduction, it is stated that Chestnut requires a mean temperature above -3.5 °C in the coldest month, and that the coldest seasonal mean temperature is warmer than the coldest monthly mean temperature, however the highest point of the response curve for BIO11 in CHELSA is at -3 °C, and that Chestnut can withstand short periods of frost between -18 °C and -25 °C. However, BIO11 represents the average temperature of the coldest season, and not a short period of time. The response curves of BIO10 and BIO11 in Wordclim show that the hottest season mean temperature between 0 and 15°C is the optimal temperature for Chestnut, and the coldest season mean temperature between 0 and 10°C is the optimal temperature for Chestnut. The response curves for the Wordclim variables were more biologically consistent, while the CHELSA variables were not consistent with biological growth characteristics, and the Wordclim data were more appropriate than the CHELSA data.

Point 4: The AUC and HEI values (Table 3, ID 2 and ID 5) also indicate that the Wordclim data are more appropriate than the CHELSA data.

·      Conedera M, Tinner W, Krebs P, et al. Castanea sativa in Europe: distribution, habitat, usage and threats. In book: European Atlas of Forest Tree Species, Publication Office of the European Union, Luxembourg, 2016, pp. 78-79.

Afif-Khouri E, Álvarez-Álvarez P, Fernández-López M J, et al. Influence of climate, edaphic factors and tree nutrition on site index of chestnut coppice stands in north-west Spain. Forestry, 2011, 84(4): 385-396.

Author Response

Please see the attachment

Author Response File: Author Response.docx