Review Reports

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The manuscript presents an interesting application of multi-modal FTIR sensing (transmission, macro-ATR, and micro-ATR imaging) to investigate the chemical heterogeneity of Edible Bird's Nest (EBN). The paper is well-written, and the use of chemometrics to explore spectral variability is commendable. However, there are significant concerns regarding the study design, specifically the absence of adulterated samples, which undermines the central claim of the paper regarding "authentication." Furthermore, the presentation of the dataset size and sampling methodology requires clarification.

Major Comments:

1.    Mismatch between Title/Hypothesis and Experimental Design (Missing Adulterants):
The title and the abstract emphasize "Authentication" and "Counterfeit detection". However, the study exclusively analyzes authentic EBN samples to demonstrate natural heterogeneity. The authors argue that standard single-spectrum methods are insufficient for authentication because of this heterogeneity. This conclusion is speculative because no adulterated samples were analyzed.
Without a "negative" class (adulterated samples), it is impossible to determine the decision boundary. It is highly probable that common adulterants possess a chemical signature distinct enough to be detected by standard macro-ATR despite the natural heterogeneity of EBN. The authors acknowledge this gap in the Discussion, but it is a fundamental flaw for a paper claiming to establish an authentication framework. The authors should either include data on adulterated samples to validate their hypothesis or reframe the paper (and title) to focus solely on the "Chemical Characterization and Heterogeneity" of EBN.

2.    Justification for Multi-Modal/Micro-ATR Analysis:
This point is closely related to the comment above. The authors advocate for high-resolution micro-ATR imaging. While this technique successfully reveals micro-heterogeneity (e.g., lipid domains), the manuscript does not justify its necessity for routine quality control.
Micro-ATR imaging is time-consuming, expensive, and generates large datasets, making it impractical for industrial screening compared to standard macro-ATR. If the spectral differences between authentic EBN and common adulterants are macroscopically significant, the high sensitivity of micro-ATR might be unnecessary. Since no adulterants were tested, the authors have not proven that the multi-modal approach is actually required to detect fraud. A multi-point macro-ATR approach might be sufficient and far more practical. The discussion should critically address the cost-benefit ratio of the proposed methodology.

3.    Confusion regarding Sample Size and Spectral Count:
The description of the dataset is confusing. The Materials section mentions "six distinct raw EBN samples". However, the Results section presents a much larger volume of data (e.g., 246 transmission spectra, 25 mini-clusters).
It is unclear how the 6 physical samples translate into the final number of spectra and clusters. Are the 25 clusters in Figure 7 derived from distinct biological samples or are they statistical artifacts (sub-samples) from the original 6 nests? This raises concerns about pseudo-replication. A data flow diagram or a table clarifying the hierarchy (Sample --> Fragments --> Measurement Points --> Spectra) is necessary to assess the statistical validity.

Minor Comments:

4.    Band Assignment (Line 387):
The authors attribute the 3000–2800 cm-1 region solely to lipid content (CH2 stretching). In a protein-rich matrix like EBN, alkyl groups from amino acid side chains also contribute significantly to absorbance in this region. Assigning these bands exclusively to lipids without supporting data (e.g., lipid extraction or specific lipid markers other than the carbonyl band found only in micro-ATR) is an oversimplification.

5.    Figure Clarity (Figs 2, 3, and 4):
    a)    Figures 2a & 3a: The authors state these plots show mean spectra and standard deviations (SD). However, due to the overlay of multiple samples, the SD "ribbons" look like additional spectral lines, making the plot difficult to interpret.
    b)    Figure 2: Please explicitly label panel (a) as "Transmission Mode" for clarity.
    c)    Figure 4: The legend states: "mean spectrum refers to a single spectrum represented by one FPA measurement". This phrasing is contradictory (mean vs. single). Additionally, the plot shows two distinct lines for each cave: are these two different measurement spots? The SD is not visible. Please clarify the legend and visual representation.

Author Response

Please see attached file

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript presents a technically sophisticated application of multi-modal FTIR spectroscopy—including transmission, macro-ATR, and high-resolution micro-ATR imaging—to investigate the chemical heterogeneity of edible bird’s nest (EBN). The use of synchrotron-based IR microscopy and a well-documented chemometric pipeline is commendable, and the observation of micro-scale lipid-rich domains represents a potentially valuable contribution to understanding EBN’s intrinsic variability.

 

However, despite these methodological strengths, the study suffers from a fundamental and fatal flaw: an extremely limited sample size that severely undermines the validity, generalizability, and scientific impact of all conclusions. This limitation is not merely a minor weakness but a critical design defect that cannot be remedied by textual revision alone.

 

Major Concern: Inadequate Sample Size and Lack of Biological Replication

 

The entire study is based on only six independent EBN samples, all sourced from four provinces within a single country (Vietnam), with three of the six originating from the same province (Khanh Hoa). While the authors correctly criticize prior studies for relying on insufficient spectral replicates or simplistic single-spectrum matching, they themselves commit a more serious error: confusing technical replication with biological replication.

 

The reported “246 transmission spectra,” “103 macro-ATR spectra,” and “15 micro-imaging maps” are all derived from these same six physical nests.

No matter how many times a single nest is measured, it remains one biological unit. The observed spectral heterogeneity—however real at the micro-scale—cannot be extrapolated to the broader population of authentic EBNs across Southeast Asia, let alone globally.

Crucially, the authors claim that “spectral heterogeneity is independent of geographic origin.” This conclusion is statistically unsupported given that (i) only four Vietnamese provinces are represented, (ii) no samples from major EBN-producing countries (e.g., Indonesia, Malaysia, Thailand) are included, and (iii) the sample per location is grossly inadequate (n = 1–3).

This issue directly contradicts the authors’ own cited standard: they state (lines 84–86) that “20–50 spectra per group” are needed for robust chemometric validation of complex biological materials—yet they provide only one biological replicate per “group” (if provinces are treated as groups).

 

As a result, any proposed authentication framework built on this dataset risks overfitting to idiosyncrasies of six specific nests and may fail catastrophically when applied to real-world samples exhibiting natural inter-individual, inter-species (Aerodramus fuciphagus vs. A. maximus), or inter-regional variation.

 

Additional Supporting Concerns

 

While the sample size issue alone justifies rejection, several other weaknesses reinforce this recommendation:

 

No adulterated samples were tested. The paper proposes a new authentication strategy but never validates it against known counterfeits (e.g., gelatin, tremella, karaya gum)—the very threat the method is meant to address.

Overstated conclusions. Phrases such as “fundamentally flawed and scientifically unsound” (line 373) dismiss prior work without sufficient evidence, while claims about global applicability ignore the narrow geographic scope.

Recommendation

 

 

Given the irremediable limitation in biological sampling, I recommend rejection of this manuscript in its current form.

Author Response

Please see the attached file

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

This study applies multi-modal FTIR sensing—combining transmission, macro-ATR, and high-resolution micro-ATR imaging-to Edible Bird's Nest. This research is highly meaningful. However, there are some notable deficiencies in the statistics and writing of this paper.
(1) line 15. reveals intrinsic chemical heterogeneity. This claim can not be supported by data in this paper. The study merely identifies samples from various sources.
(2) 1. Introduction needs to be segmented for better readability. The review of different chemometric methods should be included.
(3) 2.1. Sample Collection and Preparation. What is the sample size? How to ensure the representativeness of the samples. 
(4) What is the appearance of samples from different sources? Are there any differences? Are there any photos?
(5) Adopt multi-point acquisition and train PCA-DA/SIMCA models on diverse datasets to distinguish genuine variability from adulteration. SIMCA14.1 software can be used to process the data. PLS-DA and OPLS-DA results should be added. Significant changes in components can be identified with the methods. This can help identify the differences.
(6) The dataset should be separated into calibration and validation sets.
(7) The author may consider adopting supervised pattern recognition methods, such as PLS-DA, LDA.
(8) The discrimination rate needs to be added. Add the corresponding confusion matrix. 

Author Response

Please see the attached file

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

I would like to thank the authors for their comprehensive revision and for constructively addressing the major concern regarding the lack of adulterated samples. The decision to reframe the manuscript’s scope from an authentication framework to a foundational assessment of chemical heterogeneity is excellent and significantly enhances the scientific validity of the work. Additionally, the technical clarifications regarding spectral band assignments and figure presentation are fully satisfactory. I am pleased to recommend the revised manuscript for publication.

Author Response

We are very grateful for accepting our paper.

Reviewer 2 Report

Comments and Suggestions for Authors

Major Concerns

 

Fundamental flaw in experimental design

All spectral data (246 transmission spectra, 103 macro-ATR spectra, and 15 micro-ATR images comprising 61,440 pixel spectra) originate from only six biological samples—all collected from Vietnam and belonging to a single species (A. fuciphagus). The study critically confuses technical replicates with biological replicates: no matter how many times a single nest is measured (even at 4,096 spatial pixels), it remains one biological unit and cannot represent population-level variation of edible bird's nests.

 

Heterogeneity is not a discriminative feature for authentication

Microscopic chemical heterogeneity is a universal physical phenomenon. Both authentic EBN and artificial adulterants (e.g., gelatin, tremella, or other mixtures) inevitably exhibit spatial inhomogeneity at microscopic scales due to phase separation during drying or mixing. Lipid aggregation, for instance, occurs naturally in authentic samples (from dietary residues) and can be easily mimicked in adulterated products by adding common fats or oils. Crucially, the study tested no adulterated samples whatsoever, making it impossible to determine whether heterogeneity patterns of authentic EBN are statistically separable from those of counterfeits. Equating "presence of heterogeneity" with "authentication capability" confuses a descriptive observation with a discriminative criterion.

 

Overgeneralization of conclusions

Despite revision, the manuscript still implies that observed heterogeneity represents a defining or universal feature of authentic EBN. Six samples from a single geographic region and species cannot establish population-level prevalence or stability of any spectral feature. The study demonstrates only phenomenon existence within these limited samples, not population regularity.

 

Characteristics of Heterogeneity

Heterogeneity observed in this study exhibits four key attributes. First, it is ubiquitous: any multi-component natural or artificial material displays chemical inhomogeneity at microscopic scales. Second, it is non-specific: lipid domains appear in both authentic samples (from biological secretion variability) and adulterants (from artificial phase separation). Third, it is scale-dependent: transmission spectroscopy masks heterogeneity through sample homogenization, whereas micro-ATR imaging reveals it through spatially resolved measurement. Fourth and most critically, it is non-discriminative by itself: the mere presence of heterogeneity cannot authenticate samples. Only differential heterogeneity patterns—statistical differences in domain size distribution, spatial co-localization with other biomolecules, or chemical composition between authentic and adulterated samples—could potentially serve as authentication criteria. Such differential patterns were not investigated in this work.

 

Recommendations

Increase biological replicates substantially: minimum three batches (representing distinct origins or species) with at least ten independent nests per batch, covering major producing regions and species.

Essential addition of adulterant controls: systematically prepare and analyze common counterfeit materials (gelatin-based mixtures, tremella, karaya gum, etc.) using identical micro-ATR protocols to determine whether statistically separable heterogeneity patterns exist between authentic and adulterated samples.

Strictly limit all conclusions to the six Vietnamese samples examined. Remove any phrasing implying population-level generalizability (e.g., change "authentic EBN exhibits heterogeneity" to "the six examined Vietnamese samples exhibited intra-sample heterogeneity").

Revise the Discussion to explicitly distinguish between (i) documenting heterogeneity existence (the actual contribution of this study) and (ii) establishing heterogeneity as an authentication basis (which requires authentic versus adulterant comparison and was not performed).

 

Decision

Reject. The insufficient biological replication (n = 6 from a single geographic region and species) represents an experimental design flaw that cannot be remedied by textual revision alone. Furthermore, the absence of any adulterant controls makes authentication claims scientifically unsupported. The authors are encouraged to redesign the study with adequate biological replicates and systematic adulterant testing, focusing on identifying differential heterogeneity patterns rather than merely documenting heterogeneity existence, before resubmission.

Author Response

Please see attached file

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

The manuscript has been revised as required.

Author Response

We are very grateful for accepting our paper.