Aromatic Profile of Hydroponically and Conventionally Grown Tomatoes
, Nikola Vietorisová 1, Patrícia Martišová 2, Jana Štefániková 2, Anna Mravcová 3 and Vladimír Vietoris 1,*Round 1
Reviewer 1 Report
As a general comment, the paper explains clearly the methodology followed to take and test the samples, both with the enose and with the tasters. However, I miss much more detail in the statistical analysis section. In particular, it is not clear to me what are the features obtained from the enose that are subsequently fed to the enose algorithm. There are also no figures showing the response of the different sensors of the enose.
The authors claim that enose is suitable to discriminate between hydroponically and conventionally grown crops. However, data provided does not support this claim. The authors do not provide the results of classification in a training and test phase. Furthermore, they don't provide a classifier, they just select a region of space and assign it to a class. Also, the number of samples studied is very low for such a strong claim.
There are some minor issues with the use of English. For instance, in line 37, I think that there is a typo in the first sentence: "aroma is the combined sensation of taste and...". Also, it is not clear to me what "extractive" means in lines 212 and 232.
In line 237, it is not clear to me what are the five markers that the authors use.
Regarding the sensory analysis carried out by the evaluators, some more details, such as the number of samples tasted per session, the rest periods between samples, etc, would also be welcome.
In general, I think that the paper is potentially interesting, however, some more data is required to back the claims presented, and much greater detail in the data analysis section is also neccesary.
Author Response
* Research design must be improved - changed. * Methods must be adequately described (must be improved) - changed * Results must be clearly presented (must be improved) - added * Conclusions must be improved - done. * More details in statistical analysis section:a.) What features obtained from e-nose are subsequently fed to the e-nose algorithm - it is software feature to select discriminative power attributes. Sentence added.
b.) There are no figures showing the response of the different sensors of the enose - The loading of the PCA are most discriminative markers by Kovacs retention indexes.
* The authors claim that enose is suitable to discriminate between hydroponically and conventionally grown crops, however, data provided does not support that claim - changed.
* The authors do not provide the results of classification in a training and test phase - in the procedure of the enose measurements, there is no training or test phase, just calibration phase, done by calibration kit by manufacturer.
* They don't provide a classifier, they just select a region of space and assign it to a class. - partialy changed
* The number of samples studied is very low for such a strong claim - added note for future research.
* There are some minor issues with the use of English. In line 37, there is a typo in the first sentence - corrected.
* It is not clear to me what "extractive" means in lines 212 and 232 - tending or serving to extract; of, involving, or capable of extraction.
* In line 237, it is not clear to me what are the five markers that the authors use - described, it was feature of the software.
* sensory analysis by the evaluators some more details would be welcome: number of samples tasted per session, the rest periods between samples, etc. - added
Reviewer 2 Report
The manuscript is well prepared but has serious flaws. Most importantly, the aim of the study was to distinguish between hydroponically-grown and conventionally –grown tomatoes by means of their flavor profile. To test this, the hydroponically grown fruits came directly from Slovakia while the conventionally grown tomato fruits were from abroad. Thus the latter had completely different conditions during cultivation, harvest and transport, which makes it really difficult to compare them. The outcome of the presented study is therefore rather questionable.
Moreover, the results section is mixed with discussion elements. Please change this or combine the results and discussion section.
The discussion, in contrast, is rather a review of literature with very little relation to the results obtained in the presented study. This has to be changed.
Some minor remarks:
Materials and methods: Please add the supplier of the software that you used for statistical analyses.
The labelling of the axes is missing in all figures.
Figures 5-8: What is shown? Averages with standard deviation? What is the number of replicates?
Lines 414-415: This conclusion is not surprising as they have different genetic resources.
Author Response
* Conclusions must be improved - done. * Research design must be improved - partially changed. * The results section is mixed with discussion elements. Please change this or combine the results and discussion section - combined.* Discussion is a review of literature with very little relation to the results obtained in the presented study - changed.
* Material and methods: add the supplier of the software that you used for statistical analysis - added RStudio version 3.6.3 * The labeling of the axes is missing in all figures - it is output generated by conventional software, we cannot input some additional axes.
* Figures 5-8: What is shown? Averages with standard deviation? What is the number of replicates? - added in text.
* Lines 414-415: this conclusion is not surprising as they have different genetic resources - deleted.
Reviewer 3 Report
After a careful reading, this article is interesting, and the bibliography included is appropriate. Generally, it is a good written article. Most of references are recent and appropriate. All these points are advantages of this work
But there are some comments
- Figures 8-10 should be more readable (numbers are not sharp)
- Adjust the numbers in the tables to show the same number of decimal places
Author Response
* Figures 8-10 should be more readable - figures are enlarged.
* Adjust the numbers in the tables to show the same number of decimal places - adjusted.
Round 2
Reviewer 1 Report
Although the manuscript has improved, my greatest concern, regarding the lack of proper machine learning methodology still remains.
The authors still do not provide a figure showing the response of the sensors of the enose, nor do they provide indications as how they extract the features from those curves. They do not follow a train-validation scheme, nor do they provide a proper classification model.
These are, in my opinion, fundamental aspects that need to be included in a paper claiming that an enose can be used to discriminate between hydroponically and conventionally grown tomatoes.
Author Response
We have added a discriminatory power table, which is evaluated by the native AlphaSOFT electronic nose software. Machine learning on several types of algorithms is subsequently performed on this data, and the results are also described. For the strongest traits associated with hydroponic cultivation, 2 odorants were evaluated, we report their retention times and names.
We threw out things that did not relate to the aroma results and lightened the data visualization a bit. Added data canonical correlation analysis between sensory and instrumental data and report it.
Reviewer 2 Report
The manuscript has improved a lot albeit the research design with tomatoes from completely different provenances is questionable. However, the conclusions are adapted to this design, focussing on the tomatoes that are available on the market. Nevertheless, it is still doubtful if e-nose can be used to distinguish betweeen hydroponically- and conventionally-grown tomatoes deriving from different conditions and varieties...
Author Response
We have added a discriminatory power table, which is evaluated by the native AlphaSOFT electronic nose software. Machine learning on several types of algorithms is subsequently performed on this data, and the results are also described. For the strongest traits associated with hydroponic cultivation, 2 odorants were evaluated, we report their retention times and names.
We threw out things that did not relate to the aroma results and lightened the data visualization a bit. Added data canonical correlation analysis between sensory and instrumental data and report it.
