Next Article in Journal
Spatial–Temporal and Decoupling Effect of Agricultural Carbon Pollution Synergy in Ecologically Fragile Areas
Next Article in Special Issue
Evaluation of Deep Learning Models for Insects Detection at the Hive Entrance for a Bee Behavior Recognition System
Previous Article in Journal
Research on the Impact of Meteorological Disaster Shocks and Risk Perception on Farmers’ Cooperative Pest and Disease Control Behavior
 
 
Article
Peer-Review Record

Machine Learning-Based Acoustic Analysis of Stingless Bee (Heterotrigona itama) Alarm Signals During Intruder Events

Agriculture 2025, 15(6), 591; https://doi.org/10.3390/agriculture15060591
by Ashan Milinda Bandara Ratnayake 1,2,*, Hartini Mohd Yasin 3, Abdul Ghani Naim 4, Rahayu Sukmaria Sukri 5, Norhayati Ahmad 5, Nurul Hazlina Zaini 5, Soon Boon Yu 5, Mohammad Amiruddin Ruslan 5 and Pg Emeroylariffion Abas 1,*
Reviewer 1: Anonymous
Reviewer 2: Anonymous
Agriculture 2025, 15(6), 591; https://doi.org/10.3390/agriculture15060591
Submission received: 1 February 2025 / Revised: 6 March 2025 / Accepted: 8 March 2025 / Published: 11 March 2025

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Please see the attachment.

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript presents a well-structured and technically sound study on the use of machine learning techniques to classify acoustic signals produced by Heterotrigona itama guard bees during intrusion events. The methodology is clearly described, and the experimental design is robust. The findings demonstrate the feasibility of sound-based intrusion detection, which is an innovative approach in stingless bee research.

However, the manuscript is excessively long and would benefit from substantial reduction. Certain sections, particularly the introduction and discussion, contain redundant information and could be more concise. Additionally, the conclusion should be streamlined to focus only on the most critical findings and their implications.

Comments:

  1. Too much key words.
  2. Line 63: Missing space between reference and word.
  3. Line 67: Missing space.
  4. Line 68: Delete full stop.
  5. Line 109-184 – Appreciate for extensive description of dimensionality reduction methods, but in Introduction section you need to referrer on the topic of your paper and review of past studies which are already accomplished and have brought some specific insight on this topic. You can delete this section.
  6. Lines 219-249 – You need just to mention equipment that you have used, not to describe them in detail.
  7. Section “Data collection Procedure” – Clarify how 32 samples were collected given the setup: two hives, two replicates, two conditions, 10-minute recordings, and 2-minute breaks between artificially induced alarms.

The manuscript presents an important contribution to bioacoustic monitoring of stingless bees using machine learning. However, to improve its readability and effectiveness, significant reductions in length are necessary, particularly in the introduction, discussion, and conclusion sections. The study is methodologically sound. and with these revisions, it would be well-suited for publication.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

The manuscript is well written. The methods are presented with a high level of detail such that anyone wishing to use the methodology for other social insects should be able to. I have included suggestions in the manuscript to improve grammar in a few instances. I have two major concerns. The first concern is that there is no comparison of the authors results with other studies on predictive modeling of sound classes. This is a necessary part of any Discussion section. The authors should cite relevant studies that either support their results or suggest that their results are better than previous predictive models. 

In addition, the authors have developed a highly predictive model of the auditory defense alarms of a species of stingless bee. However, they have not back trasnformed their Mel-Frequency Cepstral Coefficients that were determined be the key features in their predictive model. The reader needs to know what frequencies (Hz) or tones or amplitudes that were important in separating the alarm sounds made by the bees from the non-alarm sounds. It is well and good to build a highly predictive model, but the authors need to tell the readers what the key alarm sounds are and how they are different from bee sounds in the hive when not under attack by invaders.

Because of these two points I have concerns about I feel that the paper needs to address these points before being considered for publication. 

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

After receiving the second-round peer review invitation from the editor of Agriculture, I promptly set aside my current tasks and logged into the MDPI system to review the authors' revised manuscripts. I noticed that the authors had made significant revisions to the manuscript, and my recommendation to the editor is that it is now ready for publication, and I will not be reviewing it further.

Reviewer 3 Report

Comments and Suggestions for Authors

The authors have done a thorough job of addressing my suggestions and concerns. They have cited key studies that provide more background and perspective on the performance of their machine learning model. They have also elucidated in more detail the features of bee defense sounds that allow prediction of aggressive defense behaviors in H. itama. I feel that this manuscript is much improved and should be published with the incorporated edits by the authors.

Back to TopTop