Estimating Chlorophyll Content, Production, and Quality of Sugar Beet under Various Nitrogen Levels Using Machine Learning Models and Novel Spectral Indices

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Comments to the Author:

General comments: This study investigated the response of various sugar beet parameters to different N levels, as well as examined the ability of various published and newly developed SRIs combined with GBR models to timely assess these parameters. This study provided a reliable method for tracking various attributes of sugar beet grown under different N levels. Although the study obtained some valuable results, it needs some revisions and English expression needs further improvement.

Introduction:

The introduction is too long, and some sentences have problems such as repeated expression, so it is suggested to simplify and modify, for example，

1.      Lines 59-60 and lines 60-61: Please add references.

2.      Lines 61-62: Please put references [2-4] in the previous statement.

3.      Lines 71-72: Please add references.

4.      The citation format of the references in the manuscript is inconsistent, such as, “Koch et al. [22]” “by [69]”.

5.      Line 151-153: “a close relationship was found between Chl content and NDVI.”and “as it is closely related to variations in Chl content and crop growth status” repeated expression. There are many similar cases in manuscripts.

6.      Lines 188-190: “Specifically, there is limited information available regarding the performance of GBR models based on common and 3D-SRIs in predicting the characteristics of sugarbeet under arid and semiarid conditions.” The sentence is not relevant to the purpose of this study and is recommended to be deleted.

Materials and Methods:

1.      Line 200: It is suggested to replace “Experimental Site, Agronomic Practices, and Nitrogen Treatments” with “Experimental Site and Experimental Design”.

2.      The year has a significant influence on the parameters, so it is recommended to add two years of meteorological data in Materials and Methods.

3.      Line 233: Why was chlorophyll measured at growth stage BBCH35? Please state the reasons in the manuscript.

Results and Discussion:

1.      It is suggested that the Results and discussion be divided into two parts, i.e. 3. Results and 4. Discussion.

2.      Lines 402-404 and Lines 388 repeat. There are many similar cases in manuscripts.

3.      Line 404: compared with 0 kg N ha^-1, SC was reduced in 30 kg N ha^-1. So “The increase in SC in the treatment using the low nitrogen level (30 kg N ha^-1)” description is not accurate.

4.      Figure 4 showed that there was a significant negative correlation between SC and nitrogen levels, however, positive correlation was observed between root yield and nitrogen levels. Thus, how to achieve a balance between maximizing RY and promoting high SC?   

Comments for author File: Comments.pdf

Comments on the Quality of English Language

There are some problems in English expression, such as semantic repetition, which need further improvement. In particular, the Introduction and Results and Discussion sections need to be improved.

Author Response

Reviewer 1

 

Comments and Suggestions for Authors

Comments to the Author:

General comments: This study investigated the response of various sugar beet parameters to different N levels, as well as examined the ability of various published and newly developed SRIs combined with GBR models to timely assess these parameters. This study provided a reliable method for tracking various attributes of sugar beet grown under different N levels. Although the study obtained some valuable results, it needs some revisions and English expression needs further improvement.

We greatly appreciate your critical observations as well as your constructive and helpful comments. We hope that we could address your questions/comments by the explanations and revisions made in the manuscript. We believe that the manuscript is substantially improved after making the suggested revisions.

Introduction:

The introduction is too long, and some sentences have problems such as repeated expression, so it is suggested to simplify and modify, for example，

1. Lines 59-60 and lines 60-61: Please add references.

Response: The authors are extremely thankful to the reviewer for this thoughtful point. The citations were added according to the comment of the reviewer.

2. Lines 61-62: Please put references [2-4] in the previous statement.

Response: Thanks for this comment. It was done in the previous statement.

3. Lines 71-72: Please add references.

Response: Thanks for this comment. The citations were added according to the comment of the reviewer.

4. The citation format of the references in the manuscript is inconsistent, such as, “Koch et al. [22]” “by [69]”.

Response: Thanks for this comment. The citation [69] was modified as citation Koch et al. [22] and all citations were checked.

5. Line 151-153: “a close relationship was found between Chl content and NDVI.”and “as it is closely related to variations in Chl content and crop growth status” repeated expression. There are many similar cases in manuscripts.

Response: Thanks for this comment. One of the sentences was removed and we checked all sentence in the manuscript.

6. Lines 188-190: “Specifically, there is limited information available regarding the performance of GBR models based on common and 3D-SRIs in predicting the characteristics of sugarbeet under arid and semiarid conditions.” The sentence is not relevant to the purpose of this study and is recommended to be deleted.

Response: Thanks for this comment. It was removed.

Materials and Methods:

1. Line 200: It is suggested to replace “Experimental Site, Agronomic Practices, and Nitrogen Treatments” with “Experimental Site and Experimental Design”.

Response: Thanks for this comment. It was replaced.

2. The year has a significant influence on the parameters, so it is recommended to add two years of meteorological data in Materials and Methods.

Response: Thanks for this comment. Two years of meteorological data was added.

3. Line 233: Why was chlorophyll measured at growth stage BBCH35? Please state the reasons in the manuscript.

Response: Thanks for this comment. We utilized BBCH35 to measure the Chl content as it represents a crucial stage in the plant's growth and development. At this stage, there is a significant rise in both root mass and sugar content. And the information was added from line 253 to line 255.

Results and Discussion:

1. It is suggested that the Results and discussion be divided into two parts, i.e. 3. Results and 4. Discussion.

Response: Thank you for your suggestion. We have decided to combine the results and discussion into one section to effectively explain and discuss the outcomes of the three main goals addressed in this study. These objectives are: (i) to examine the effects of different nitrogen levels on several characteristics of sugar beet, such as Chlt, Chla, Chlb, RY, SY, and SC; (ii) to create multiple 2D and 3D-SRIs from the spectral reflectance of a sugar beet canopy and evaluate their performance in assessing the characteristics of sugar beet; (iii) to evaluate the performance of the GBR model based on common and 3D-SRIs, as well as the data fusion of both common and 3D-SRIs, to forecast the characteristics of sugar beet under different nitrogen levels.

Importantly, the interrelationships among the three objectives have also been considered.

 

2. Lines 402-404 and Lines 388 repeat. There are many similar cases in manuscripts.

       Response: Thanks for this comment. Lines 402-404 were removed

3. Line 404: compared with 0 kg N ha^-1, SC was reduced in 30 kg N ha^-1. So “The increase in SC in the treatment using the low nitrogen level (30 kg N ha^-1)” description is not accurate.

Response: the sentence “The increase in SC in the treatment using the low nitrogen level (30 kg N ha^-1) or without N fertilizers may be attributed to the fact that a lack of nitrogen supply leads to a decrease in root size and moisture content. As a result, the SC in the root increases” has been changed into “The increase in SC in the treatment without N fertilizers, compared to those fertilized with N, can be attributed to the fact that a deficiency in nitrogen supply leads to a reduction in root size and moisture content. As a result, the SC in the roots increases.

4. Figure 4 showed that there was a significant negative correlation between SC and nitrogen levels; however, positive correlation was observed between root yield and nitrogen levels. Thus, how to achieve a balance between maximizing RY and promoting high SC?

Response: Thank you for this comment. As we mentioned in lines 448-460: an optimal supply of N fertilizer is crucial for achieving maximum RY and optimum SC. However, excessive amounts of nitrogen fertilizer can lead to an increase in RY but a decrease in SC in the root. Conversely, when nitrogen fertilizer amounts are too limited, the opposite is true. Therefore, it is important to apply N fertilizer to sugar beets in an amount that strikes a balance between maximizing RY and promoting high SC. Therefore, in our study we found that linear functions were effective in fitting the relationship between N levels and SC. However, quadratic functions were found to be effective in fitting the relationship between N levels and the yields of root and sugar (Figure 4). Based on the quadratic functions, the optimal amount of N fertilizer required for maximizing RY was 155.9 kg N ha^-1. This indicates that if the N application rate increases than 156 kg N ha^-1, the RY and SC would be decrease.

Comments on the Quality of English Language

There are some problems in English expression, such as semantic repetition, which need further improvement. In particular, the Introduction and Results and Discussion sections need to be improved.

Response: many thanks for your comment. Many instances of semantic repetition have been altered and the English language has been enhanced in numerous areas within MS.

 

Reviewer 2 Report

Comments and Suggestions for Authors

The paper is well-written and interesting.  However, I am struggling to get the novelty of the investigation. These relationships are already known and published. The authors rightfully provide those references where the work was done [25, 26, 27 28 &29]. The question is then why this investigation?

Furthermore, the motivation or justification for developing new SRIs is very weak. It is not clear from the manuscript what their research problem or research question is, or why the existing SRIs are inadequate for these purposes. This information needs to be provided.

Under the method section:

1) There is a need to clarify why they only investigated one cultivar DSA007 (why this one?). 

2) Provide more details or equipment used to apply the fertilizer. Otherwise, provide a reference to the process.

3) Provide a reference for the agronomic procedure; "From planting until harvesting, all recommended agronomic and plant protection practices were followed". Alternatively, describe the procedure here.

A much more important weakness in the manuscript is around the need to develop 20 additional SRIs. This sounds like fishing for the right signal (data dredging). Provide strong motivation for these 20 additional SRIs. 

 

 

Author Response

Reviewer# 2

 

Comments and Suggestions for Authors

We greatly appreciate your critical observations as well as your constructive and helpful comments. We hope that we could address your questions/comments by the explanations and revisions made in the manuscript. We believe that the manuscript is substantially improved after making the suggested revisions.

The paper is well-written and interesting.  However, I am struggling to get the novelty of the investigation. These relationships are already known and published. The authors rightfully provide those references where the work was done [25, 26, 27 28 &29]. The question is then why this investigation? Furthermore, the motivation or justification for developing new SRIs is very weak. It is not clear from the manuscript what their research problem or research question is, or why the existing SRIs are inadequate for these purposes. This information needs to be provided.

Response: Thanks for this comment. We highlighted that from line 161to line 172. Although many SRIs may be calculated easily and have great potential to analyze and predict plant traits, these indices are often generated based on only two to three wavelengths. As a result, SRIs are less effective in estimating plant attributes under diverse growth conditions and are more sensitive to the different factors influencing canopy spectral signatures such agronomic treatments, phenological growth stages, cultivars, seasons, and climatic conditions. Additionally, while there are hundreds of SRIs available, just a small number of SRIs are used in published work. Furthermore, the regression analyses used for assessing the relationship between spectral reflectance and plant attributes are based on only one SRI. Therefore, there are strong arguments in favour of combining multiple different SRIs into a single index when estimating plant features in order to enhance the prediction analysis and modeling of plant attributes using machine learning.

 

Under the method section:

• There is a need to clarify why they only investigated one cultivar DSA007 (why this one?). 

Response: Thank you for your comment. This particular cultivar is commonly used in the area of study and has a similar phenotype structure to other cultivars. Additionally, this is the first time we are testing spectral indices and GBR modeling with Sugar beet under Egypt's conditions. However, in future work, it is important to use different genotypes to validate this data.

 

• Provide more details or equipment used to apply the fertilizer. Otherwise, provide a reference to the process.

Response: Thanks for this comment. It was add in the manuscript at line 233.

• Provide a reference for the agronomic procedure; "From planting until harvesting, all recommended agronomic and plant protection practices were followed". Alternatively, describe the procedure here.

Response: All recommended agronomic and plant protection practices were diligently followed, from the planting stage to harvesting, in accordance with Leilah and Khan [7]. This information was added from line 224 to line 225

A much more important weakness in the manuscript is around the need to develop 20 additional SRIs. This sounds like fishing for the right signal (data dredging). Provide strong motivation for these 20 additional SRIs. 

Response: Thanks for this comment. The created contour maps in Figure1 can successfully be used to derive as many SRI as required so it is easy to develop numerous numbers of SRIs. We mainly focused on the most accurate ones for detecting various investigated sugar beet attributes. From contour maps based on three bands and colour scale can be detect the best R² for each parameters of sugar beet.

Reviewer 3 Report

Comments and Suggestions for Authors

The authors estimated chlorophyll content, yield, and quality of sugar beet under different N rates using machine learning models and spectral indices. The topic is valuable and the study is significant. Before it can be published, I have the following comments for this paper.

 

1. Line 210, experiment design needs to be described more specifically, Randomized Complete Block Design? 5 replicates?

2. Line 232-246, The concentrations of Chlt, Chla, and Chlb were based on fresh weight, which varied depending on leaf moisture. Can these concentrations be calculated using dry matter weight?

3. Line 316, The experiment was conducted for two seasons (two plantings), the data was pooled into one dataset? How many data were used for validation?

4. Line 355, Table 2. How were the ANOVAs conducted? Using a design of split plot? Why are there two Errors? (Error-1 and Error-2).  

5. Line 526, Table 3, please indicate the number of samples within each N levels.

6. Line 565, Table 4, I suggest the data can be showed in heatmap.

7. Line 608, Table 5, concentrations of Chlt, Chla, and Chlb should be the contents in dry weights.

8. Line 608, Table 5, Explain why the R2s are high for training datasets, but are very low for testing dataset? It is unacceptable if the R2 is low for the training dataset. In that case, the models are needed to be re-constructed to increase prediction accuracy.

Author Response

Reviewer 3

Comments and Suggestions for Authors

The authors estimated chlorophyll content, yield, and quality of sugar beet under different N rates using machine learning models and spectral indices. The topic is valuable and the study is significant. Before it can be published, I have the following comments for this paper.

 We greatly appreciate your critical observations as well as your constructive and helpful comments. We hope that we could address your questions/comments by the explanations and revisions made in the manuscript. We believe that the manuscript is substantially improved after making the suggested revisions.

 

1. Line 210, experiment design needs to be described more specifically, Randomized Complete Block Design? 5 replicates?

Response: In our study, we focused on evaluating a single factor with five treatments. Therefore, the most suitable design to assess this factor is a Randomized Complete Block Design (RCBD). However, during the ANOVA analysis, the data were analyzed in a combined manner.

2. Line 232-246, the concentrations of Chlt, Chla, and Chlb were based on fresh weight, which varied depending on leaf moisture. Can these concentrations be calculated using dry matter weight?

Response: In this study, we simultaneously measured the plant chlorophyll content and spectral reflectance of the canopy. As a result, we found it more suitable to correlate the chlorophyll concentration with the fresh weight rather than the dry weight. Additionally, we ensured that all plants in the nitrogen treatments received a consistent amount of irrigation, ensuring that the moisture content of the plants in all N treatments remained relatively stable.

3. Line 316, the experiment was conducted for two seasons (two plantings), the data was pooled into one dataset? How many data were used for validation?

Response: many thanks for these comments. A total of approximately 50 sugar beet samples were used for training and validation. Out of these samples, 80% (40 samples) were used to train and validate the regression model, while the remaining 20% (10 samples) were used to compare predicted values with computed values. It was highlighted from line 339 to line 342.

3. Line 355, Table 2. How were the ANOVAs conducted? Using a design of split plot? Why are there two Errors? (Error-1 and Error-2).

Response: Thank you for your comment. The experiment was conducted using a Randomized Complete Block Design (RCBD), and the data were analyzed in a combined manner. As a result, there are two errors identified, one for the year and another for the treatment, as indicated in Table 2. 

5. Line 526, Table 3, please indicate the number of samples within each N levels.

Response: Thank you for your comment. It was added at the title of table 3 from line 569 to 570.

6. Line 565, Table 4, I suggest the data can be showed in heatmap.

Response: Many thanks for these comments. Table 4 was replaced by heatmap as figure 5

Response: Many thanks for these comments. Table 4 was replaced by heatmap.

7. Line 608, Table 5, concentrations of Chlt, Chla, and Chlb should be the contents in dry weights.

Response: In this study, we simultaneously measured the plant chlorophyll content and spectral reflectance of the canopy. As a result, we found it more suitable to correlate the chlorophyll concentration with the fresh weight rather than the dry weight. Additionally, we ensured that all plants in the nitrogen treatments received a consistent amount of irrigation, ensuring that the moisture content of the plants in all N treatments remained relatively stable.

8. Line 608, Table 5, Explain why the R² are high for training datasets, but are very low for testing dataset? It is unacceptable if the R² is low for the training dataset. In that case, the models are needed to be re-constructed to increase prediction accuracy.

Response: Many thanks for these comments. In the present study, we undertook a comprehensive comparison of different vegetation indices, including two-dimensional Spectral Reflectance Indices (2D-SRIs), three-dimensional Spectral Reflectance Indices (3D-SRIs), and a combination of both, termed as Advanced Spectral Reflectance Indices (ASRIs). Our objective was to identify the most effective indices that yield high predictive accuracy when validated. Utilizing the R-squared yield (RY) as a metric, we observed values of 0.815, 0.743, and 0.725 for 2D-SRIs, 3D-SRIs, and ASRIs, respectively. Based on these results, 2D-SRIs emerged as the superior predictive feature, especially when implemented in conjunction with Gradient Boosting Regression (GBR). Alternative models were developed to provide a comparative analysis and facilitate the selection of the most efficacious model. While the predictive accuracy for some variables may not be exceptionally high, the outcomes are satisfactory. Moreover, the parameters we have introduced in this research are novel and hold promise for future studies that may employ new methodological approaches for enhanced prediction.