Frost Risk Management in Chickpea Using a Modelling Approach
Round 1
Reviewer 1 Report
The submitted manuscript by Y. Chauhan and R. Merrill present an interesting study on the risk of frost after flowering for chickpea in the Northern Grain Region (NRG) of Australia. The probability of these events, which may cause a significant yield loss, have been investigated by means of a modified version of the Agricultural Production System Simulator (APSIM) forced with observed climatic data (from 1961 to 2017) over 46 checkpoints of NGR, which are representative of the main sites of chickpea production. The modified version of the APSIM model accounts for the effects of soil moisture, which, in a previous study, has been demonstrated to be significant for a more performant simulation of the flowering dates of the chickpea. Built on the comparison between the statistical features of the simulated flowering dates and of the last frost day of the year, the authors assessed the historical frost risk and its trends over the 46 checkpoints. Depending on these features, the 46 sites have been clustered in 6 different sub-regions: among them, the mid-latitude regions, i.e. the Southern Downs and the Darling Downs, appeared to be those with the highest frost risk as compared to the regions further north and south. Also, focusing on the Dalby site in the Darling Downs sub-region, the authors analysed a hierarchy of different strategies to minimise the risk of yield loss:
- They investigated the impact of different sowing dates for PBA Boundary chickpea variety, identifying the 15th of May (i.e. at 30% frost risk level) as the optimal sowing day for Dalby site.
- In addition to the sowing management, they investigated the effects of the use of different chickpea varieties, evidencing the possibility to further optimise the yield by selecting the most suitable cultivar for each specific region, e.g. late flowering varieties (PBA Boundary, Tyson) against early flowering varieties (Sonali, PBA Seamer).
Overall, I find the manuscript very interesting, the methodology robust, the text rather clear and the bibliography pertinent. This study may have a strong impact on decision-making for Australian agriculture, as it can support chickpea producers for an optimal management of their cultivations. Moreover, this study may be also relevant for future studies focusing on long-term adaptation strategies to climate change, as the same methodology here used for historical data may be easily applied to future climate projections. Therefore, I think that the manuscript deserves to be published on Agronomy.
I have just a few minor suggestions that, in my opinion, could improve the clarity and the linearity of the paper before the acceptance:
Lines 90-94: please specify in the text that TTm is the modified thermal time and TT is the thermal time in the previous version of the APSIM model.
Line 106: where the estimation of 5% loss in chickpea due to a frost event come from? Is it also in [11]?
Line 119: please specify the spatial resolution of SILO weather data.
Line 125: it should be -3 degree (minus is missing).
Section 2.3. I think that the different dates of sowing for Dalby, i.e. 10 June for 10% frost risk, 22 May for 20% frost risk and 15 May for 30% risks are results of the study (they come from Fig.5) and should not be put in this Section. I suggest a qualitative description of the four different management strategies here (e.g. just describing them with the attributes “super conservative” “conservative” “slightly conservative” and “highly risky” with the associated frost risk probabilities), and to evidence these specific dates after Fig. 5 (see also comments for lines 251-253).
Line 135: please specify that Dalby belongs to the Darling Downs sub-region, as the latter was already introduced in the Abstract.
Lines 136-137: I am wondering if logically it makes more sense to associate the 20% risk with the attribute “conservative” and the 30% risk with the attribute “slightly conservative” (see also comments on Line 265 below).
Line 165: In relation to Figure 1: where these profiles have been calculated? Is it one specific site or the ensemble mean of all the 46 locations? In the latter case, please also display the ensemble spread.
Figure 5 (left panels): What is the difference between the solid and the dotted curves? Is the former simply related to the results from the modified version of the APSIM model, and the latter to the previous version of the APSIM model? Please specify it in the caption.
Lines 251-253: I suggest to present here the dates of sowing corresponding to the four different strategies for Dalby site, and to evidence them with a simple table (see also comments on Section 2.3).
Line 265: Here the “slightly conservative” approach is associated with a 30% frost risk, while in lines 136-137 and in Figure 6 it is associated with a 20% frost risk. Please make it conform (see also comments on lines 136-137).
Figure 6 (righ panel): It is not exactly clear to me what “probability of exceedance” stands for. Also, make the attribute “conservative” and “slightly conservative” conform to line 265.
Otherwise, I find the manuscript already in a good shape for acceptance.
Author Response
Responses to Reviewer 1 Comments and suggestions
Lines 90-94: please specify in the text that TTm is the modified thermal time and TT is the thermal time in the previous version of the APSIM model.
Response: Done. The sentence now read as “…where TTm is the modified thermal time after accounting for the decreases associated with soil water, TT is the thermal time in the previous version of the APSIM model and FASW is fractional available soil water.”
Line 106: where the estimation of 5% loss in chickpea due to a frost event come from? Is it also in [11]?
Response: The source of 5 % loss for each frost risk has been elaborated. Adjusting potential simulated yield with this loss, the performance of model in predicting yield improved considerably (See Chauhan et al. 2019).
Line 119: please specify the spatial resolution of SILO weather data.
Response: The coordinates of 45 locations that we used were described by Chauhan et al. (2017) which is an open access article published in Field Crops Research. The coordinates of an additional locations have been given. The relative positions of the location have been shown in Fig. 3.
Line 125: it should be -3 degree (minus is missing).
Response: Yes. Thanks and it has been corrected.
Section 2.3. I think that the different dates of sowing for Dalby, i.e. 10 June for 10% frost risk, 22 May for 20% frost risk and 15 May for 30% risks are results of the study (they come from Fig.5) and should not be put in this Section. I suggest a qualitative description of the four different management strategies here (e.g. just describing them with the attributes “super conservative” “conservative” “slightly conservative” and “highly risky” with the associated frost risk probabilities), and to evidence these specific dates after Fig. 5 (see also comments for lines 251-253).
Response: The dates have been removed and as advised only attributes of the strategy have been given.
Line 135: please specify that Dalby belongs to the Darling Downs sub-region, as the latter was already introduced in the Abstract.
Response: Yes, that is true that Dalby is in the Darling Downs region of Queensland and this has been indicted now.
Lines 136-137: I am wondering if logically it makes more sense to associate the 20% risk with the attribute “conservative” and the 30% risk with the attribute “slightly conservative” (see also comments on Line 265 below).
Response: The 30% risk was conservative and 20% as slightly conservative. Since these have now been mentioned more clearly, there should be little confusion.
Line 165: In relation to Figure 1: where these profiles have been calculated? Is it one specific site or the ensemble mean of all the 46 locations? In the latter case, please also display the ensemble spread.
Response: This is the mean of all 46 locations. The soil water holding characteristics were described In Table 4 of Chauhan et al. (2017).
Figure 5 (left panels): What is the difference between the solid and the dotted curves? Is the former simply related to the results from the modified version of the APSIM model, and the latter to the previous version of the APSIM model? Please specify it in the caption.
Response: Yes, the solid lines and solid symbols in the right panels of charts are flowering time computed by adjusting for soil water and the dotted curve lines and open symbols of the previous version of the APSIM model. This has now been specified in the caption.
Lines 251-253: I suggest to present here the dates of sowing corresponding to the four different strategies for Dalby site, and to evidence them with a simple table (see also comments on Section 2.3).
Response: It has been done. Please note that it is the earliest date that planting was done. Planting could be done even later if there was no rain on this date until 15 July.
Line 265: Here the “slightly conservative” approach is associated with a 30% frost risk, while in lines 136-137 and in Figure 6 it is associated with a 20% frost risk. Please make it conform (see also comments on lines 136-137).
Response: Corrected. The 30% is indeed the ‘conservative approach’
Figure 6 (righ panel): It is not exactly clear to me what “probability of exceedance” stands for. Also, make the attribute “conservative” and “slightly conservative” conform to line 265.
Response: The probability of exceedance has been clarified in the legend.
Reviewer 2 Report
A well written paper on an interesting topic.
The title Lines 2-4---The title should represent the article’s content and facilitate retrieval in indexes developed by secondary literature services. The terms in the title should be limited to those words that give significant information about the article’s content. It is best to start the title with key words—not with words such as "Enhancing". A good title briefly identifies the subject, indicates the purpose of the study, and introduces key terms or concepts. The recommended limit is 12 words. ---Frost risk management in chickpea through modeling etc...
I don't think using historical data as far as 1980 enhanced our understanding of the trends. Figure 4 to me clearly shows that trends changed about 2000 for the majority of clusters. Why not limit historical data from 2000 up to present. Besides the cultivars have changed over time making it difficult to make comparisons----I guess this is when G xE X M comes into play.
Lines 160-162. Could you be more specific in describing the Statistical methods---just reporting that you used R does not cut it for me. For instance, what libraries within R did you use. What kind of analysis did you do---probability analysis? non linear regressions? linear models? How were the graphs developed? in ggplot2? or what?
Figure 5-- Please explain why Day of last frost/ flowering day are between 200 and 300. Are you sing Julian Date? If so, this must be stated.
Lines 326-341---changing to new cultivars---how does it affect the risk to frost? Are newer cultivars bred for frost tolerance? What can a chickpea breeding program take home as a tool in their breeding program from your research. Give me the so what in more practical ways?
Author Response
Responses to Reviewer 2 Comments and suggestions
The title Lines 2-4---The title should represent the article’s content and facilitate retrieval in indexes developed by secondary literature services. The terms in the title should be limited to those words that give significant information about the article’s content. It is best to start the title with key words—not with words such as "Enhancing". A good title briefly identifies the subject, indicates the purpose of the study, and introduces key terms or concepts. The recommended limit is 12 words. ---Frost risk management in chickpea through modeling etc...
Response: The title has been revised to read
“Frost risk management in chickpea using a modelling approach”
I don't think using historical data as far as 1980 enhanced our understanding of the trends. Figure 4 to me clearly shows that trends changed about 2000 for the majority of clusters. Why not limit historical data from 2000 up to present. Besides the cultivars have changed over time making it difficult to make comparisons----I guess this is when G xE X M comes into play.
Response: Thanks for the suggestion. While we agree to some extent with the reviewer, we still we feel showing it from 1980 onward will make a reader miss the point that climate of our region began to change from 1980 onward in relation to the prior period. In some environments. such as Dawson Callide (Fig. 4 a), there were more frosts prior to 1980 than now. Hence we would like to retain the Figure 4 as such.
Lines 160-162. Could you be more specific in describing the Statistical methods---just reporting that you used R does not cut it for me. For instance, what libraries within R did you use. What kind of analysis did you do---probability analysis? non linear regressions? linear models? How were the graphs developed? in ggplot2? or what?
Response: The statistical methods have now been more elaborated to give an idea of R programs various resources including libraries being used.
Figure 5-- Please explain why Day of last frost/ flowering day are between 200 and 300. Are you sing Julian Date? If so, this must be stated.
Response: Yes it was Julian day which has been clarified in the legend.
Lines 326-341---changing to new cultivars---how does it affect the risk to frost? Are newer cultivars bred for frost tolerance? What can a chickpea breeding program take home as a tool in their breeding program from your research. Give me the so what in more practical ways?
Response: Cultivars were not bred for frost tolerance but differed in frost susceptibility due to their flowering time. This only suggests that we do not need to specifically breed for frost as there are other approaches to manage the problem. Breeders can focus on identifying maturity that will have list risk for a given environment and the approach outlined in the paper could be used for this purpose. This point has been elaborated.
Author Response File: 
Author Response.pdf
