Scaling Up Sap Flow Measurements from the Stem Scale to the Individual Scale for Multibranched Caragana Korshinskii on the Chinese Loess Plateau

Round 1

Reviewer 1 Report

The authors present a study investigating the dependence of sap flow of a shrub species on different environmental variables (even though soil moisture, which should be interesting in a semi-arid environment, is missing) and exploring methods to scale up sap flow from single stems to the whole shrub. The number of shrubs investigated is quite low, but given the high number of stems, this is understandable. The basic results are presented nicely, however the statistics and multivariate regression models are very basic (some basic concepts of regression such as multicolinearity should be checked). The discussion is very short and should be improved considerably (References are very much focused on Chinese authors so far, a general topic auch as upscaling shrub transpiration from single stems to individuums and canopies should consider broader contributions).

For specific comments please check the pdf-file attached.

Comments for author File: Comments.pdf

Author Response

Reviewer #1’s comments

The authors present a study investigating the dependence of sap flow of a shrub species on different environmental variables (even though soil moisture, which should be interesting in a semi-arid environment, is missing) and exploring methods to scale up sap flow from single stems to the whole shrub. The number of shrubs investigated is quite low, but given the high number of stems, this is understandable. The basic results are presented nicely, however the statistics and multivariate regression models are very basic (some basic concepts of regression such as multicolinearity should be checked). The discussion is very short and should be improved considerably (References are very much focused on Chinese authors so far, a general topic auch as upscaling shrub transpiration from single stems to individuums and canopies should consider broader contributions).

The response: We really appreciate the reviewer’s positive comments and made some corrections which we hope meet with approval.

Line 28: Not sure what you mean with "variable transpiration".

The response: Both vapour pressure deficit (VPD) and solar radiation (Rs) generally have notable effects on sap flow. Variable of transpiration is an integrated index which can be calculated from VPD and Rs (Line 133-136), and is often used to investigate their composite influences on sap flow (Reference no.9, 30, and 31).

Line 40: Be careful here not to mix up different methods. It would probably be better to talk first about heat-pulse and thermal diffusion and their use in trees and then focus on heat balance then to jump back and forth.

The response: done (Line 39-46).

Line 103: Please explain the abbreviations used in the table (PBN, SL, SBD, ESB), and also what you mean with canopy size (projected crown area of the whole bush?) in the table caption.

The response: The abbreviations used in the table (PBN, SL, SBD, ESB) were explained below the Table 1. Canopy size indicates the projected area of the entire canopy (Line 148-151).

Line 111: Personally, I don't like the term "variable of transpiration" and I don't think it's a fixed term. If you check e.g. Du et al (2011) they describe it as " an integrated index named variable of transpiration (VT) was used in addition to Rs and VPD", something like that would be better (and I did not find "variable of transpiration" at all in Zha et al (2017) which you also cite here)

The response: Reference citation errors have been corrected (Line 136).

Line 141: I guess you mean "average rainfall per event" that should be clearly stated (it could also be average rainfall per day, week or month)

The response: I have made some revisions (Line 91).

Figure 1. using different colors for P and T would increase clearness of the figure

The response: done (Figure 1).

Line 171: not sure what you mean with this sentence: when conditions for transpiration were favourable, sap flow should increase further. For me an explanation could be that soil water availability was limited, thus the plants where forced to close their stomata and limit transpiration and sap flow around midday

The response: We have revised the expressions.

Line 172: not sure what you mean with this sentence, especially the "black locust overlaps"

The response: What we were trying to say is that the peaks of SF and the meteorological factors are not synchronized to avoid strong transpiration of plants. I have made some modifications (Lines 374-376 ).

Figure 3. Are these averages for the whole measuring period or is it one single day? Should be clearly stated.

The response: I have clearly stated it in Figure 3.

Line 190: this is clear as transpiration and sap flow are mainly driven by this environmental factors..., given the semi-arid nature of the measuring site, data on soil moisture (soil water content or even better soil water potential) would be really interesting to see if there is a limiting effect of soil moisture on sap flow.

The response: We have explored the influences of soil moisture on sap flow, and the results are shown in the attached list (line 381-385).

Line 197: I don't get why you use "although" here, there is no contradiction between the first and second part of the sentence.

The response: I have made revisions (Line 401).

Line 198: That is not true. VPD is the partial pressure difference between actual and saturated condition in the atmosphere (There is also a VPD if there are no plants). However, if the air within the leaf is saturated, the partial pressure difference between the leaf and the outside atmosphere is equal to VPD.

The response: I have made revisions (Line 388-389).

Line 206: Which results?

The response: I have clearly stated it (Line 236).

Figure 6. Interesting that you do not see saturation in the relationship between SF and VPD or PAR (and consequently VT and ET₀). For VPD this could be due to our semi-arid conditions, for PAR I am not sure (I would expect it in shaded, radiation-limited conditions which apparently not the case in your study).

The response: This may be due to the fact that soil moisture was not a limiting factor on sap flow (Line 381-385).

In general a multivariate (regression-) model of SF and climate variables (i.e. the ones which are not correlated) would be statistically more correct than to plot/calculate correlations between SF and each climate variable alone.

The response: I have made revisions (Table 2)

Figure 5. While I am a fan of showing time courses of variables, I think this figure is better suited for the supplementary material

The response: We have moved Figure 5 as a supplementary material (Line 703).

Line 223: I guess that PBN, SL, SBD, and W (ESB)are strongly correlated to each other (which could/should be shown) so it is quite clear that a high correlation of SF and one of them also leads to a high correlation between SF and the other. It would be more interesting to discuss which correlation was highest (in your case the one between SF and ESB) and why?

The response: We observed from the R² values of the regression models that the correlation between SF and W was highest. This may be due to the fact that a larger W will contribute to a larger numbers of leaves and also a larger leaf area that is positively correlated with plant transpiration.

Line 225: To me it is not clear how the "noon depression" phenomenon is connencted with SBD. In my opinion it should be discussed above in the SF vs climate variables section. (You could than add here that there is no noon depression in any stem size, as it could be that smaller and larger stems react differently)

The response: We have moved the related discussion to the SF vs climate variables section.

Again, I would move one or two of figures 7, 8, and 9 to the supplementary material.

The response: We have moved the figures 7 and 8 to the SF vs climate variables section.

Figure 7 Is this an example of one specific day (if so was it a sun or cloudy day) or and average daily course for the whole measuring period? This should be clearly stated in the figure caption.

The response: I have clearly expressed it (Lines 718-719).

Line 259: It is a pity that you do not have any soil moisture data to back up this claim. Can you think of any other possibility to estimate soil water availability and to check if it is affecting sap flow?

The response: the influence of soil moisture on sap flow has been provided in lines381-385.

Line 270: In Table 2, R² for the SF-ESB model is (slightly) better than for the SF-SBD model, so it should be discussed here why the SF-SBD model is used.

The response: It is true that the R² for the SF-ESB model is higher than that for the SF-SBD model, but the P-value of constant for the SF-ESB model is not statistically significant.

Table 2: To calculate statistically correct regression models multicollinearity of explanatory variables should be checked first (e.g. by calculating VIF), I assume that at least some PBN, SL, SBD, and ESB are correlated to each other. In this case only one of the correlated variables can be used in a regression model. In addition, adjusted R² should be used instead of R² to show the quality of the model, as R² will also improve when more explanatory variables are added. An additional model quality criteria such as AIC would improve the selection of the best model (I assume that the model using ESB as the only explanatory variable might actually be the best one)-To me it is no clear what the meaning of P_con, P_PBN, P_SL, P_SBD and P_W (P_ESB) is, this should be explained better.

The response: We have added VIF results in Table 2 to show the degrees of the multicollinearity of explanatory variables. The P_con, P_PBN, P_SL, P_SBD and P_ESB are referring to the P-values of the regression coefficients of PBN, SL, SBD, and W (ESB).

Table 3. I guess the SD values in table 3 are swapped

The response: I have made modifications.

Line 288: According to table 2 the SF- W(ESB) model was better than the SF-SBD model.

The response: The R² for the SF-W(ESB) model is slightly better than for the SF-SBD model, but the P-value of constant for the SF-W model is not statistically significant.

Author Response File: Author Response.docx

Reviewer 2 Report

The authors have made a big stride in upscaling morpohological level traits up to tree level transpirational flows and it is very commendable. These kinds of efforts are helping a lot the science community in understanding better the vegetative water transport mechanisms. It is well-written, clear and simple to understand. Below are my comments to help improve the paper.

 

Figures 3 and 4 should appear in order of appearance in the text (interchanged?).

 

Line 186 should be put in Line 187. So drag the Figure 3 up to move the figure description below Figure 3.

Line 113. What is Rs?

 

Line 188 – 208. Your result indicated that VPD has the strongest correlation with SF. However, VPD and PAR are two very closely related parameters. So to remove the effect of PAR on VPD component, maybe you can try exploring at normalizing SF with PAR, then plot VPD vs SF residuals and see if your result still holds true. You can also can rank all your meteorological variables according to their importance in SF using any multivariate analysis or General Additive Modelling analysis using the akaike weights to get the best model. In this way, you can rank these climate variables to determine the best predictor for SF. This can strengthen your paper although this is optional.

Relative to these two suggestions, a wealth of information about light-normalized evapotranspiration and GAM analysis can be seen in a paper of Aguilos et al., 2018 on Interannual and Seasonal Variations in Ecosystem Transpiration and Water Use Efficiency in a Tropical Rainforest.  https://www.mdpi.com/1999-4907/10/1/14 where climatic factors affecting evapotranspiration were closely discussed. Another good paper can also be found on this paper:  Kume, T.; Takizawa, H.; Yoshifuji, N.; Tanaka, K.; Tantasirin, C.; Tanaka, N.; Suzuki, M. Impact of soil drought on sap flow and water status of evergreen trees in a tropical monsoon forest in northern Thailand. For. Ecol. Manag. 2007, 238, 220–230.

Author Response

Reviewer #2’s comments

The authors have made a big stride in upscaling morphological level traits up to tree level transpirational flows and it is very commendable. These kinds of efforts are helping a lot the science community in understanding better the vegetative water transport mechanisms. It is well-written, clear and simple to understand. Below are my comments to help improve the paper.

The response: We really appreciate the reviewer’s positive comments.

Figures 3. and 4. should appear in order of appearance in the text (interchanged?).

The response: done.

Line 186: should be put in Line 187. So drag the Figure 3 up to move the figure description below Figure 3.

The response: done (Line 353-355).

Line 113: What is Rs?

The response: Rs represents solar radiation. I have clearly stated it.

Line 188 – 208: Your result indicated that VPD has the strongest correlation with SF. However, VPD and PAR are two very closely related parameters. So to remove the effect of PAR on VPD component, maybe you can try exploring at normalizing SF with PAR, then plot VPD vs SF residuals and see if your result still holds true. You can also rank all your meteorological variables according to their importance in SF using any multivariate analysis or General Additive Modeling analysis using the akaike weights to get the best model. In this way, you can rank these climate variables to determine the best predictor for SF. This can strengthen your paper although this is optional.

The response: The explored regression models between SF and meteorological parameters in our study is not used to predict SF. Actually, the regression models between SF and shrub morphological traits are used to predict SF.

Relative to these two suggestions, a wealth of information about light-normalized evapotranspiration and GAM analysis can be seen in a paper of Aguilos et al., 2018 on Interannual and Seasonal Variations in Ecosystem Transpiration and Water Use Efficiency in a Tropical Rainforest. https://www.mdpi.com/1999-4907/10/1/14 where climatic factors affecting evapotranspiration were closely discussed. Another good paper can also be found on this paper: Kume, T.; Takizawa, H.; Yoshifuji, N.; Tanaka, K.; Tantasirin, C.; Tanaka, N.; Suzuki, M. Impact of soil drought on sap flow and water status of evergreen trees in a tropical monsoon forest in northern Thailand. For. Ecol. Manag. 2007, 238, 220–230.

The response: We have studied the two papers carefully and cited them in this paper.

***************************************************************

Reviewer 3 Report

Please, see the attached pdf. 

Comments for author File: Comments.pdf

Author Response

Reviewer #3’s comments

The main idea of the manuscript of scaling-up the sap flow measurements from the stem to the individual is very good and also very useful for shrubs with several stems. However, establishing a strong model to obtain the individual sap flow from stem measurements must have several samples, and 3 individuals are not enough to be confident with the results.

The response: This is true that only 3 individuals were studied in this paper, due to the expensive costs of the devices used. Altogether 21 stems of different morphological characteristic were used to measure sap flow. We think, maybe this is understandable.

If possible I recommend to get more information from other individuals, and also during other periods, due to the data are only from the rainy period and maybe the drought could affect that relationship.

If it is not possible, I suggest to re-structure the manuscript focusing in other objectives and do more complex statistical analysis to describe better the response of sap flow to meteorological data and morphological traits (combination of both).

The response: We have modified the main objectives of this study and improved the statistical methods.

The highlights are too general and do not summarize the main results or implications of manuscript.

The response: We have modified the highlights.

INTRODUCTION
-The introduction needs to be re-structured to avoid repetitions of the ideas (e.g. lines 44-45and 47-48. And include more appropriate references

The response: done (Line 48-49).

- Linen38. I suggest introducing the idea that sap flow can be measured with several methods, so then this sentence will be connected with the following ideas about the different methods and their advantages and disadvantages.

The response: The response: done (Line 39-46).

-Lines 45 and 46: at the end of each of the sentences is necessary to include references.

The response: done (Line40 and42)

-Line 50: the idea of stem leaf area is not clear (what it means).

The response: The stem leaf area indicates the measured area of all the leaves on a stem.

-Line 52: the point after Zhang must be removed.

The response: done (Line 57).

-Last paragraph: it is not well justified why is important to connect the sap flow with the morphological traits, and the meteorological conditions.

The response: We have made some revisions (66-67).

-There is not information about the success of the program to control the degradation and soil erosion, and the relevance of the study species in this control.

The response: We have made some revisions (66-67).

MATERIALS AND METHODS
Study area

-It is not clear if the 3 individuals are in the forest or in the grassland area…and this is relevant for the model, and the possible effect of plant competition for the resources (specially water).

The response: done (Line 79).

-Line 82: “and others” should be specified which species are in the area due to the possible effect of competence.

The response: done (Line 89).

Measurement of sap flow

-Line 87: who authors know the age of the individuals? Specify how the age is measured or include an appropriate reference.

The response: done (Line 127-128).

-Line 92: It is indicated (abstract) that the sap flow sensors were installed in 21 stems, but here the authors indicate that there were only 16 selected stems. Who is possible?

The response: A total of 16 stems were selected at the beginning of the measurement, and 5 stems were replaced with new branches due to abnormal data during the period of measurement.

-Line 98: regarding to the 5 branches that were replaced with new branches, do authors used all the data? In the case they didn’t used the abnormal data they had a period without information for the 21 stems, so the replicates are less than exposed at the beginning. It must be clarified.

The response: The abnormal data has been eliminated, and the normal data came from 21 stems (including 5 new stems) was used in the paper. The unit of sap flow is gram per hour, and each value represents the average of each hour in a day (Figure 5, 6, 8,9, and 10).(Line 144-146)

-Line 99: the authors indicate to check Table 1 when they talk about the five branches that were removed due to abnormal data, but the Table 1 did not include that information. Maybe it must be indicated in the Table, including also which % of the data was considered as abnormal.

The response: done (Line 151-153).

-Line 101: maybe the authors should use W and not ESB for the stem biomass due to it has not sense to change it.

The response: done

Table 1: I suggest including the meaning of the abbreviations in the caption and not at the end of the Table.

The response: done (Line 151-152).

-Do authors consider the number of bifurcations of each stem? Could it influence the sap flow?

The response: We did not consider the number of bifurcations of each stem. Maybe it could affect the sap flow, which will be further explored in subsequent experiments.

Measurements of meteorological variables

-Line 110: Is the meteorological station in an open area? How far from the individuals is the meteorological station?

The response: We have clearly provided that information (Line 175-176).
-It is not justified why the authors calculate the evapotranspiration (ET₀) and also the “transpiration” (VT), it seems that they are correlated.

The response: done (Line 178-180).

-Equations 2-4: it is necessary to indicate which is the meaning of each of the parameters considered in those equations.

The response: done (Line 184-189).

-Lines 121-122: Who the authors know that the daily averaged soil heat flux density is small and can be ignored? A reference is required.

The response: done (Line 189).

Model establishment and validation

-I consider that the model should include meteorological data, due to the known relationship, also demonstrated here, between sap flow and meteorological data.

-Sap flow data from 14 stems to build a strong model is not real, moreover, if only 3 individuals are considered, due to the data are not representative of the individuals and much less of the population.

The response: this is a limitation of our study.

Data analysis

-It is not described how they analyse the meteorological data.

The response: done (Line 177-189).

-It is not described how they calculate the sap flow data represented in the figures. Are they means for all the stems (21?) for just a part of the stems (14?) or maybe is a sum of the stems from an individuals and later is a mean for the three individuals?

The response: done (Line 230-249).

RESULTS AND DISCUSSION
The discussion is very scarce and not well referenced. There are not references for the study species, and the reasoning is superficial. Meteorological conditions during the study period. This information could be included in the methodology, when the study site is described.

The response: done (Line 90-125).

Diurnal variations in sap flow

-Line 151: Reference the figure 3.

The response: done (Line 244-245).
-The figure 3 should be placed before the figure 4. And it is not known what information includes de sap flow variable (mean of stems, mean of individuals….). I suppose that is the sap flow measured and not the estimated. Moreover, it could be convenient to reference the sap flow data to a unit of area (maybe projected area), due to the individuals seem that they do not have the same size, and this could influence the absolute value of the sap flow.

The response: done (Line 355-361).
-The figure 4 do not includes the ET₀, why?

The response: The ET₀ calculated by function 5 is the crop evaporation per day, and it is not the instantaneous value. So the hysteresis between the sap flow and ET₀ cannot be analyzed.
- Lines 164-177: there is no information about the tolerance of the species to PAR or temperature. It is a relevant information to interpret the responses of sap flow to meteorological variables as PAR or temperature.

The response: Tie et al. (2017) found the hysteresis loops of sap flux density were observably smaller for PAR, VPD, and T under the low radiation condition (PAR < 231.48 μmol∙m^-2∙s^-1) compared to the high radiation condition (PAR>231.48 μmol∙m^-2∙s^-1), lag time from 0.5-1.0 h, but the rotation directions of the hysteresis loops did not change (Reference no. 5). The PAR is above 231.48 μmol∙m^-2∙s^-1 occurring seven days during the whole measurement period, which was much sporadic. So we only considered the overall effect of PAR population.

-Line 173: there is not a references justifying that it is a self-protection mechanism to avoid black locust. Moreover, there is not information (in the introduction) that the study species has competence with this other species.

The response: It is our fault to have not clearly expressed it. We have revised it (Line 377).

-Lines 176-177: there is not any reference justifying that the study species has a conservative water strategy, and the results showed here do not justify this strategy due to the study was developed during the rainy season.

The response: There is a reference justifying that the study species has a conservative water strategy (Reference no.37). The experiment was conducted during the rainy season, but rainfall distribution was uneven from 0.2 mm to 46.0 mm.

Effects of meteorological factors on SF

-Lines 196-198: is this sentence supported by statistical analysis? Include that information.

The response: done (Line 400-402).

-Did authors explore statistical models checking the effect of meteorological data on sap flow variations?

The response: done.

-Lines 203-204: more references for the study species are necessary, references from related species are not enough. In the case there is not information about the study species it must be indicated in the text to avoid confusion.
-Figure 6 and figure 9: why the p –value is cero? This is not statistically correct. In the case it is very low and near 0 the authors should indicate that the p-value is <0.0001

The response: done (Figure 6 and figure 9).

-The regressions done in the figure 6 consider several points (I suppose that from several stems) so, the effect of the individual should be considered due to some points are more similar to others due to be from the same individual. This effect should be included in the model. (the same for the regressions of figure 9).

The response: The relationship between the average SF of one day for all branches and the mean value of the daily meteorological factors is shown in Figure 6, and the relationship between the daily average SF of all branches and their corresponding morphological traits is shown in Figure 9.
Effect of morphological traits on stem-scale SF
-Lines 228-232: it is not well justified why the authors talk about sapwood area but they do not calculate it (which it is not possible in small stems), so, this must be clarified and connected well with the need of considering the diameter of the stems.

The response: done (128-133).
-Figure 7: Why the authors consider only this diameters? How many stems are considered in each diameter? Only one? In that case why they do not represent all the diameters? This should be also described and justified in the methods.

The response: It is our fault to have not clearly expressed it. We have revised it (Lines 158-159). (Line 94-97).
-Figure 8: average from what? Stems? Individuals?

The response: The average SF values are the average SF for several stems in three groups of different morphological categories.

-Figure 8: why the authors consider those ranges for the different morphological traits?

The response: It is our fault to have not clearly expressed it. According to the existing morphological characteristics, different characteristics are divided into three groups: large, medium and small, so that the effect of different morphological traits on sap flow is more significant and easier to understand.

-Lines 255-262: why the authors talk about other species, why are relevant here? Are they similar to the study species? Are they coexisting species?

The response: We talked more about the other species due to the limited information on the studied species.

CONCLUSIONS-Authors should recognize the limitations of their study: for example, that they adjusted an estimation model only for the rainy period but maybe it is not a good model for the drought period, or in case that it is not relevant justify it.

The response: This is a limitation of our study.

-Lines 284 and specially lines 289-290: can be said emphatically due to the number of individuals is not enough, and it is not representative of the population.

The response: done.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

The authors adjusted the introduction and method's section according to the reviewers comments. In some of these corrections the English is rather poor (e.g. line 38-44) and should be improved. However, the concerns regarding a more profound statistical analysis and a more in-depth discussion were hardly addressed at all. I therefore have to stay with my recommondation of "major revision".

Some specific comments were added in the pdf attached.

Comments for author File: Comments.pdf

Author Response

Responses to the reviewers’ comments

Reviewer #1’s comments

The authors adjusted the introduction and method's section according to the reviewers comments. In some of these corrections the English is rather poor (e.g. line 38-44) and should be improved. However, the concerns regarding a more profound statistical analysis and a more in-depth discussion were hardly addressed at all. I therefore have to stay with my recommendation of "major revision".

The response: We have carefully checked and improved the English language in the paper. The statistical analysis used and the discussion sections have been greatly improved and clearly stated.

Line 38-44: The introduced part is really hard to read and needs rephrasing.

The response: done (Line38-53).

Line 93: The first sentence of this paragraph should be deleted as it is basically repeated in the second sentence.

The response: done (Line 93).

Line 147-149: How do you define "normal" and "abnormal" data?

The response: if SF data measured from one branch during the daytime was 0, and at the same time, the SF data measured from other branches has a value. Then we disassembled the gauge wrapped in the stem, and found that there was some water in the place where the gauge was in contact with the stem. This will influence the measurement of SF. So we determined that the SF data was abnormal. We believe abnormal SF data from the good working status of instrument, such as battery powered status, voltage and current status on each gauge.

Line 259-260: I cannot see this in figure 3d: the increase of SF in the morning is equal to VT, in the afternoon the decrease of SF lags behind the decrease of VT. In figure 3b the course of SF generally lags behind PAR.

The response: done (Line 256-258).

Table 2: According to your cover letter, VIF-values were added to Table 2, but i cannot see them here. I still not sure if you use R² or adjusted-R², you definitely should use the latter. The correct notation of very small p-values would be > 0.001 but not 0.000 as p-values are never 0 but can be very small...

The response: We use SPSS software to carry out univariate and multivariate linear regression analysis between sap flow and morphological characteristics (PBN, SL, SBD, and W) of all stems measured. Fifteen linear regression models were obtained. Then eleven models were screened out by using collinear diagnosis (if the VIF value > 10, then the model was omitted). Finally, the best model (function 10) was obtained according to the large R² and small P value of each variable (Table 2).

Author Response File: Author Response.docx