Nutrient Balance of Citrus Across the Mandarin Belts of India

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The article holds significant scientific and practical value in the study of mineral nutrition in Indian citrus. However, there are areas that need improvement, including research design, data analysis, depth of discussion, and clarity of writing. The following suggestions are proposed to enhance the logic, scientific rigor, and readability of the manuscript:

 

The introduction mentions that one reason for the lower-than-potential yield of Indian citrus is imbalanced plant nutrition, but it fails to specify the manifestations of this imbalance (e.g., which nutrients are deficient or in excess). It is suggested to add a detailed description of the existing nutritional issues, such as listing the primary symptoms of nutrient deficiencies or excesses.

 

 

In the Materials and Methods section, although the sample size and geographical locations are provided, there is no explanation of the sampling criteria (e.g., whether random sampling was used) or how the representativeness of the samples was ensured. A detailed description of the sampling methods is recommended.

 

 

In the Data Transformation section, while the calculations of clr and wlr are described in detail, there is no explanation of why these methods were chosen over others. A discussion on the theoretical basis and the pros and cons of these methods is suggested.

 

 

In the Results section, Tables 1 and 2 list numerous nutrients and gain ratios but lack summaries or interpretations of these data. It is recommended to add brief summaries below the tables to highlight which nutrients have the greatest impact on yield. Additionally, while the accuracy of the Catboost model is mentioned, it is not compared with other models (such as traditional linear regression models). A comparative analysis with other models is suggested to highlight the advantages of the model used.

 

 

The Discussion section references some literature on nutrient diagnosis but does not provide a detailed comparison of the study’s results with existing research. A detailed comparison with existing nutrient standards, particularly those from other Indian studies, is recommended.

 

 

From a modeling perspective, gain ratios are an important concept in machine learning, but their biological significance is not explained in the article. A discussion on how gain ratios reflect nutrient interactions is suggested. Additionally, while the accuracy of the model is mentioned, external validation (e.g., testing the model’s applicability on datasets from other regions) is lacking. It is recommended to add an external validation section to verify the model’s universality.

Author Response

The article holds significant scientific and practical value in the study of mineral nutrition in Indian citrus. However, there are areas that need improvement, including research design, data analysis, depth of discussion, and clarity of writing. The following suggestions are proposed to enhance the logic, scientific rigor, and readability of the manuscript:

Reviewers comments : The introduction mentions that one reason for the lower-than-potential yield of Indian citrus is imbalanced plant nutrition, but it fails to specify the manifestations of this imbalance (e.g., which nutrients are deficient or in excess). It is suggested to add a detailed description of the existing nutritional issues, such as listing the primary symptoms of nutrient deficiencies or excesses.

Authors response : Thank you for your suggestion . As suggested .we have now provided the manifestations nutrient imbalance , mostly the  multiple nutrient deficiencies (line 47-48  ) . We have also listed the primary symptoms of different nutrient deficiencies and excesses (line 48-64). Deficiency symptoms were not included in the database because they are reported on a weak nominal scale. This is why analytical data provided the strong ratio scale to quantify the nutrient status of mandarin tissues.

Reviewers comments : In the Materials and Methods section, although the sample size and geographical locations are provided, there is no explanation of the sampling criteria (e.g., whether random sampling was used) or how the representativeness of the samples was ensured. A detailed description of the sampling methods is recommended.

Authors response : Thank you so much for excellent suggestion . Agreeing to your suggestion , we have provided an additional information about sampling criteria of orchards  and how the representative samples were obtained (line143-144 ) .

Reviewers comments : In the Data Transformation section, while the calculations of clr and wlr are described in detail, there is no explanation of why these methods were chosen over others. A discussion on the theoretical basis and the pros and cons of these methods is suggested.

Authors response :  Thank you for your suggestion . We have added a whole section (2.3) on clr and wlr.Sicerely hope, it is considered favorably (line 190-214).

Reviewers comments  : In the Results section, Tables 1 and 2 list numerous nutrients and gain ratios but lack summaries or interpretations of these data. It is recommended to add brief summaries below the tables to highlight which nutrients have the greatest impact on yield. Additionally, while the accuracy of the Catboost model is mentioned, it is not compared with other models (such as traditional linear regression models). A comparative analysis with other models is suggested to highlight the advantages of the model used.

Authors response : Thank you for nice suggestion . We haveadded brief summaries reporting on Tables 1 and 2. We tested several ML models. Gradient Boosting (Catboost is a GB model) proved to be the most accurate.

 

Reviewers comments : The Discussion section references some literature on nutrient diagnosis but does not provide a detailed comparison of the study’s results with existing research. A detailed comparison with existing nutrient standards, particularly those from other Indian studies, is recommended.

Authors response: Thank you so very much for a pertinent suggestion . Indeed so appreciative of such suggestion . We have now made a comparison of nutrient standards concentrating on Indian studies. We have also made a comparison of nutrient standards developed by authors for different Indian citrus cultivar to provide more meaningful discussion ( Table 5 ) . Incidently , nutrient standards have been observed vary a great deal depending upon the interpretation tool , despite using the same  data- set . This is the reason , any nutrient standards developed for any cultivar needs to be cross-validated through field-response studies. We have made these issues figure in discussion covering your major concern . However, there were no experimental data (fertilizer experiments) in the database.

Reviewers comments : From a modeling perspective, gain ratios are an important concept in machine learning, but their biological significance is not explained in the article. A discussion on how gain ratios reflect nutrient interactions is suggested. Additionally, while the accuracy of the model is mentioned, external validation (e.g., testing the model’s applicability on datasets from other regions) is lacking. It is recommended to add an external validation section to verify the model’s universality.

Authors response: Thank you. But this would be the next step after developing diagnostic tools based on crop surveys. Fertilizer trials must be established to verify models’ capacity to generalize.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The paper evaluates the potential to develop interactive nutrient models to diagnose the nutrient status of mandarin groves in India to transfer to a machine able to classify and orient nutrition models useful to increase the yield. A very interesting field for future application that needs a more adequate presentation.

-The lack of references on the state of art on this kind of application to citrus (if any) lets to conclude that this area of research is at the beginning. Therefore, before the approval for publication some agronomic guidelines need to be discussed. 

- Citrus management needs good fertilization program based on the species and varieties, the rootstocks, the age of trees, the soil type and pH, the environmental conditions, the pests and diseases, etc, affecting the groves. 

-Some citrus pathogens (well known in India) are very relevant in this area, since they affect specifically the root and/or the vascular system of the tree, limiting the water and nutrient uptake. Therefore, the evaluation of these factors is preliminary before the design an interactive application for citrus management. 

- The data analyzed refer to extremely different environmental conditions, age and rootstocks of the trees (some were grafted on rough lemon and other were just seedlings), and the soil type. No information is provided for diseases and climate. 

- Very appreciable, despite the results do not evidence much difference between clr and wlr, the authors correctly underline the need of a verification in universality tests in the field and do not exclude the potential role of pests and diseases. 

-The complex limiting factors suggest that the publication of the paper is not well supported in this stage, while it would be useful to include some additional factors that are easily relievable (at least visually) to provide useful discriminants for a classification.

 

Comments for author File: Comments.pdf

Author Response

The paper evaluates the potential to develop interactive nutrient models to diagnose the nutrient status of mandarin groves in India to transfer to a machine able to classify and orient nutrition models useful to increase the yield. A very interesting field for future application that needs a more adequate presentation.

Reviewers comments : The lack of references on the state of art on this kind of application to citrus (if any) lets to conclude that this area of research is at the beginning. Therefore, before the approval for publication some agronomic guidelines need to be discussed. 

Authors response : Thanks for your concern . Very true , such kind of application to citrus is still in the beginning only (Yamane et al., 2022). Nevertheless, we have already provided cultivarwise agronomic guidelines under materials and methods section.   

Reviewers response : Citrus management needs good fertilization program based on the species and varieties, the rootstocks, the age of trees, the soil type and pH, the environmental conditions, the pests and diseases, etc, affecting the groves. 

Authors response: Very true , I agree with you . In our studies, we  aimed that orchards selected do were not infested with any major pests or diseases, with their fruit yield data recorded for at least two seasons. And, we presumed that yield data is the manifestation of any such variation in varieties, the rootstocks, the age of trees, the soil type and pH ,and  the environmental conditions, including  the pests and diseases ( if any ). These facts, we can easily see in cultivarwise yield disparity, can be coined regional yield discrepancy.  

 Reviewers comments: Some citrus pathogens (well known in India) are very relevant in this area, since they affect specifically the root and/or the vascular system of the tree, limiting the water and nutrient uptake. Therefore, the evaluation of these factors is preliminary before the design an interactive application for citrus management. 

Authors response: Very true, thanks for raising this important issue. Phytophthora-induced foot -and root-rot, gummosis diseases and Huanglongbing (citrus greening disease) are the two major diseases in citrus orchards of India , with varying implications on yield and orchard life expectancy . Role of plant nutrition in triggering these two disease are still considered inconsequential, by and large. On the other hand, better plant nutrition provides a second line of plant defense against these diseases, but their exact priming effect is still not studied. Your point of contention is well taken, unfortunately, our primary aim was different. Thank you

Reviewers comments : The data analyzed refer to extremely different environmental conditions, age and rootstocks of the trees (some were grafted on rough lemon and other were just seedlings), and the soil type. No information is provided for diseases and climate. 

Authors response: Thanks for pointing out. The yield outcome and related leaf nutrient composition are the two attributes of implication of variations in rootstock and soil type. Regarding orchard age, we selected the orchards age coinciding with their peak productivity, so that the expressions about fruit yield and  leaf nutrients concentration realistically obtained for nutrient modelling .  

Reviewers comments : Very appreciable, despite the results do not evidence much difference between clr and wlr, the authors correctly underline the need of a verification in universality tests in the field and do not exclude the potential role of pests and diseases. 

Reviewers comments : The complex limiting factors suggest that the publication of the paper is not well supported in this stage, while it would be useful to include some additional factors that are easily relievable (at least visually) to provide useful discriminants for a classification.

Authors response : Tissue testing is thought to integrate all factors that affect crop performance (Munson and Nelson, 1990). Visual and web-based deficiency symptoms were not documented during crop surveys because they are reported on a weak nominal scale, are mixtures of several nutrient problems simultaneously and could inject error in predictive models (Pl see new section 4.2). We extended nutrient norms to include all nutritionally balanced specimens (whatever yield cutoff selected) in elaborating nutrient standards by more than doubling the number of specimens of each cultivar (Table 4). The analytical data provided the strong ratio scale to quantify the multi-nutrient status of mandarin tissues and provide multi-nutrient diagnosis ( Pl. see Figure 2).

We sincerely hope , our responses and the revisions exercised in the original manuscript would be considered sufficient by our learned reviewers. Thanks for your time and efforts

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Accept.

Author Response

Thanks so much for time and concerns. Actually, the entire study was carried out in orchards free of two major diseases viz., Phytophthora-induced foot, root rot and gummosis diseases and Huanglongbing (HLB , two most common diseases of citrus orchards in India. The same has now been reflected under the Materials and Methods section and taken as an issue in the conclusion section as a part of the future study ( revisions made in green color). Accordingly, the revised manuscript is submitted for onward consideration at your end. Thanks once again for your time and so sincere efforts