In-Orchard Sizing of Mango Fruit: 2. Forward Estimation of Size at Harvest

Round 1

Reviewer 1 Report

See attached file.

Comments for author File: Comments.pdf

Author Response

The draft entitled "Forward estimation of mango (Mangifera indica) fruit size at harvest" developed a technique that can forecast tree fruit yield. I have carefully reviewed this draft and found it interesting.

>>Thank you

My minor queries regarding the draft are:

1) Page 12, Table 3, I did not completely understand this table. Does this table demonstrate that the harvest prediction has a large error (basically for those fruits with mass less than 405 g)? I realized that you mentioned it in page 13, lines 338-339. If it is the case, then the title of the paper has to be changed.  

We agree – the legend was poorly written. Text has been revised and Table converted to figures for ease of visualisation. There is indeed error particularly for small fruit but we contend the technique still has practical value.

2) Page 13, lines 344-345: "Despite higher error in prediction of fruit mass using Eqn. 2, the prediction of tray size distribution was not significant impacted (Pearson’s chi-squared test at P < 0.05)." I did not completely understand this sentence.

This sentence was removed after reviewing the text.

3) Page 13, Table 4: I did not understand what you wanted to demonstrate in this table.

Explanatory text has been added, with this section thoroughly revised.

“ The frequency distribution of fruit tray size classes based on estimation of fruit mass using L and the average of W and T (Eqn. 2, Table 3) was not significantly different (X2 test, P < 0.05) compared to use of LWT (Eqn. 1, Figure 6).”

4) Page 14, lines 356-357: "The forward estimates did not perform with an expected accuracy of 100% for tray size distribution." Please improve this sentence. You have to specifically mention why predicting tray size distribution is a difficult task.

This section thoroughly revised.

“The percentage distribution of fruit by tray sizes was estimated at 4 and 3 weeks before harvest for example populations of three cultivars (Figure 6). A cultivar specific average growth rate as calculated from change across these measurement dates (Table 2) was used to project fruit sizes and tray size distribution at harvest, allowing comparison to actual harvest mass distribution (Figure 6). Pearson’s chi-squared test indicated a significant (P < 0.05) difference between actual and predicted harvest size frequency for the three cultivars, however the distribution is sufficiently accurate to inform harvesting and marketing plans. e.g., a delay in harvest date for a low size distribution population.”

5) Page 14, lines 357-366: "The closest estimation of fruit size at harvest was better predicted from measurements taken from 4 to 3 weeks before harvest for the cultivars in this study. The small number of samples collected could have affected the overall results. As recommended by Walsh et al., [18] the Eqn. 3 could be used to improve the sampling size. It is recommended the use of the in-field machine vision methodology for estimation of fruit size on tree orchard as proposed by Wang et al. [7] and Neupane et al. [8]. The on-vehicle imaging rig used for counting fruit from inter-rows as described in Anderson et al. [2] could be used for improving the sampling size, due to the possibility of doing large sample collections e.g., sample an entire row or an entire block. These methodologies could be coupled to fruit growth models as described in this experiment." Please rewrite this entire paragraph.

This section thoroughly revised and incorporated into Conclusions.

“Measurements made at four and three weeks before harvest can predict harvest mean fruit mass to within 8% of actual mean harvest mass and achieve 99.4% accuracy on estimated tray size distribution. For practical implementation, it is recommended that a first measurement of fruit size be made at a GDD associated with stone hardening stage, i.e., at the harvest target GDD minus around 400 to 450 units for Calypso, Honey Gold and Keitt (Amaral, [17]). A second measure should then be made one week later. As development rates vary with temperature, this recommendation will be associated with a shorter calendar period between stone-hardening and harvest in warmer compared to cooler growing areas.

However, manual measurement of fruit size using a statistically relevant sample size and strategy [18] is  labour intensive, and this represents a barrier to farm adoption. The results of the current study are encouraging for use of in-orchard machine vision derived estimates of fruit lineal dimensions, as proposed by Wang et al., [4]. RMSE on fruit mass increased only slightly, from 23.9 to 25.0 g, for regressions based on kLWT and kL((W+T))/2)2, respectively, and the frequency distribution of fruit tray size distributions was not significantly affected. Use of fruit area as a machine vision input, as suggested by Utai et al., [6], is also warranted.

The use of the in-field machine vision methodology for estimation of fruit size on tree orchard was proposed by Wang et al. [7] and Neupane et al. [8] for implementation on the farm vehicle mounted imaging platform driven at approx. 5 km/h described by Anderson et al. [2]. Fruit size estimates made using these methodologies could be coupled to fruit growth models as described in this experiment, although it remains to be documented whether the accuracy of on-the-go measurements is a limitation on practical application. Decreased accuracy is expected, associated with the need to remove partly occluded fruit from consideration.”

6) Page 14, lines 369-371: "Measurements made at four and three weeks before harvest can predict harvest mean fruit mass to within 8% of actual 370 mean harvest mass and achieve 99.4% accuracy on estimated tray size distribution." Based on my understanding, your method could not predict tray size distribution accurately. I did not understand how your method achieved 99.4% accuracy for this task.

Again, this section has been thoroughly revised.

” Mango fruit size at harvest estimated from lineal measurements of L, W and T made up to four weeks before harvest and a week later were between -15 and 18% (mean 8.6%) of actual mean fruit mass at harvest (Table 2), and can be used in combination with estimates of the number of fruit in the orchard, made by manual or machine vision methods as described by Anderson et al [2], for prediction of orchard yield in tonnes/ha.”

Reviewer 2 Report

Please see comments in the file!

Comments for author File: Comments.pdf

Author Response

The reviewer has provided comments on a pdf of the manuscript:

1. format error (blank line) in the Abstract

Response: corrected

2. "Introduction is hard to read. No logic"

Response: The Introduction has been thoroughly revised

3. "viz. What is is?"

Response: viz is a term that means 'namely'. Word has been replaced for reader convenience.

4. li 180 "what kind of tray? Are there any figure?"

Response: Cardboard trays are ubitiquous in the Australian domestic and export markets.  A figure seems over-kill, however some further words of explanation have been added.

“The impact of error on fruit mass assessment was considered in context of packing into the 7 kg fruit trays typically used by the Australian fruit industry. These reinforced cardboard trays are marketed with either 10, 12, 14, 16, 18, 20 or 22 fruit per tray, with corresponding fruit mass averages of 720, 600, 514, 450, 400, 360 and 327 g.”

5. line 333 "Sampling. This section should be in Methods"

Response: We contend that this is part of a discussion of the implementation of the method, and so is appropriate in the current section.

6. line 367: Section should be numbered 4 not 3

Response: corrected

Reviewer 3 Report

The submitted manuscript by Amaral and Walsh entitled “Forward estimation of mango fruit (Mangifera indica) size at harvest” is well-written, and generally, the author's data and statical analysis are clear. The aim and scope of the Special Issue "Plant-Based, proximal and Remote Sensing in orchards and Vineyards- State of the Art, Challenges, Data Fusion and Integration" and  Horticulturae journal are in line with the current manuscript.

Author Response

We thank the reviewer for their positive appraisal. 

Reviewer 4 Report

The authors presented work on using computer vision in the field to measure fruit size to support the prediction of mango fruit size at harvest.

The manuscript is all right as far as I am concerned. The design of the research exercise is straightforward, with an adequate description of the methods. The results support the conclusions, where the authors recommend a first fruit measurement at a GDD associated with the kernel hardening stage and a second one week later for practical implementation.

Some minor flaws should be considered before proceeding with the publication of the manuscript. My comments are below:

 - in my opinion, acronyms and formulas (line 14) should be avoided as much as possible in the Abstract section

 - the word "lineal" (lines 19, 37, 88, 155, etc.) referring to measurements is uncommon in the metric system. It would be better to replace it with "linear", in my opinion

- the symbols used in formulas 2 and 3, lines 172 and 173, although evident, should be made explicit

- lastly, the number of references is somewhat limited. Often authors refer to the same author several times. Therefore, the References section should include a broader sector overview.

Author Response

(reviewer comment in followed by <<our response)

The authors presented work on using computer vision in the field to measure fruit size to support the prediction of mango fruit size at harvest.The manuscript is all right as far as I am concerned. The design of the research exercise is straightforward, with an adequate description of the methods. The results support the conclusions, where the authors recommend a first fruit measurement at a GDD associated with the kernel hardening stage and a second one week later for practical implementation.

>>Thank you

Some minor flaws should be considered before proceeding with the publication of the manuscript. My comments are below:- in my opinion, acronyms and formulas (line 14) should be avoided as much as possible in the Abstract section

>> We attempted this but it proved very awkward to be spelling out ‘growing degree days’ at each occurrence, and to be explaining the two formulae in words.  We have thus returned to use of the abbreviations and formulae, albeit with removal on one abbreviation.

 - the word "lineal" (lines 19, 37, 88, 155, etc.) referring to measurements is uncommon in the metric system. It would be better to replace it with "linear", in my opinion

>>We don’t believe 'linear' is a metric term.  Linear/lineal are English words, synonyms (lineal is ‘relating to or consisting of lines; linear’ Oxford Dictionary). As we have used this term in predecessor papers, we propose to retain its use.

- the symbols used in formulas 2 and 3, lines 172 and 173, although evident, should be made explicit

>>Done

- lastly, the number of references is somewhat limited. Often authors refer to the same author several times. Therefore, the References section should include a broader sector overview.

>>There are now 22 unique references. The companion paper [22] extends the reference list with focus to the use of machine vision in fruit sizing, while the references of the current paper focus to mango fruit allometry, fruit growth models and traditional measurement techniques. We submit that the reference coverage across the two papers is reasonable (while noting we have a review paper on fruit sizing in preparation).

Reviewer 5 Report

This manuscript presented a meaningful method for the mango yield prediction, which would benefit the mango industry and the research of orchard fruit yield prediction. With the powerful introduction and clear materials, the manuscript would be interesting for most readers. However, the analysis of the results was not that strong enough. If more discussion of the application could be addressed, it would be a greater work.

1.        The introduction gave enough references to support the meaning of this work. However, as a method involved manually measurement, how to apply it in modern orchard should be emphasized.  

2.        “This result supports the proposed use of machine vision estimates of fruit lineal for estimation of fruit mass”. It’s better to give an experiment support for this conclusion. If couldn’t, more discussion should be addressed.

3.        Readers would be confused by how and whether the model match the real value. The manuscript did a great work while the writing was a little bit week.

Author Response

(Reviewer text is followed by <<our response)

This manuscript presented a meaningful method for the mango yield prediction, which would benefit the mango industry and the research of orchard fruit yield prediction. With the powerful introduction and clear materials, the manuscript would be interesting for most readers. However, the analysis of the results was not that strong enough. If more discussion of the application could be addressed, it would be a greater work.

1. The introduction gave enough references to support the meaning of this work. However, as a method involved manually measurement, how to apply it in modern orchard should be emphasized.  

>> Text added at line 34, 51 and 93

2. “This result supports the proposed use of machine vision estimates of fruit lineal for estimation of fruit mass”. It’s better to give an experiment support for this conclusion. If couldn’t, more discussion should be addressed.

>>There is a companion paper on the use of machine vision in fruit sizing submitted concurrently to the current manuscript. As the companion is now published, we reference it in the revised manuscript (li 376).

3. Readers would be confused by how and whether the model match the real value. The manuscript did a great work while the writing was a little bit week.

>> We feel Table 2 is clear on the presentation of the comparison of predicted and actual mean mass (for 15 estimates), while Figure 6 and Table 3 report on the predicted and actual frequency distribution for fruit mass.

>>The text statement on estimates of mean fruit mass is: “For measurements taken on weeks 4 and 3 before harvest, estimates were between -6% and 8% (mean and SD of absolute errors of 4.5 ± 2.4%) of actual mass (Table 2), with an RMSE and bias of predicted fruit mass at time of harvest compared to actual fruit mass of 47.5 and 25.9 g, 61.0 and 6.0 g, and 43.8 and -34.6 g, for the ‘Calypso’, ‘Honey Gold’ and ‘Keitt’ populations, respectively.”

>>The text statement for frequency distribution is “Pearson’s chi-squared test indicated a significant (P < 0.05) difference between actual and predicted harvest size frequency for the three cultivars (data not shown), however the distribution is sufficiently accurate to inform harvesting and marketing plans. e.g., a delay in harvest date for a low size distribution population.”, with reference to Figure 6 and Table 3.

>>we have attempted to reword Discussion sections for clarity on this point.

Round 2

Reviewer 2 Report

More better!

Author Response

A new version was done based on editor and reviewer feedback. No specific comments were done by reviewer 2

Author Response File: Author Response.docx