Next Article in Journal
AMFEF-DETR: An End-to-End Adaptive Multi-Scale Feature Extraction and Fusion Object Detection Network Based on UAV Aerial Images
Next Article in Special Issue
Advances in Surveying Topographically Complex Ecosystems with UAVs: Manta Ray Foraging Algorithms
Previous Article in Journal
Enhancing Unmanned Aerial Vehicle Path Planning in Multi-Agent Reinforcement Learning through Adaptive Dimensionality Reduction
Previous Article in Special Issue
Assessment of Ground and Drone Surveys of Large Waterbird Breeding Rookeries: A Comparative Study
 
 
Article
Peer-Review Record

Vegetation Type Preferences in Red Deer (Cervus elaphus) Determined by Object Detection Models

Drones 2024, 8(10), 522; https://doi.org/10.3390/drones8100522
by Annika Fugl 1, Lasse Lange Jensen 1, Andreas Hein Korsgaard 1, Cino Pertoldi 1,2,* and Sussie Pagh 1
Reviewer 1: Anonymous
Reviewer 2: Anonymous
Reviewer 3: Anonymous
Drones 2024, 8(10), 522; https://doi.org/10.3390/drones8100522
Submission received: 12 August 2024 / Revised: 16 September 2024 / Accepted: 20 September 2024 / Published: 26 September 2024
(This article belongs to the Special Issue Drone Advances in Wildlife Research: 2nd Edition)

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The article describes the determination of dietary and habitat preferences of large herbivores by drones. The scientific value of the data on distribution, habitat use and vegetation preference is not very high and the article does not bring anything new. A comparison of the method used with other methods could be of greater value. But that was not the aim of the study. Methodological procedures correspond to the objectives and data evaluation is at a standard level.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

This paper uses a drone equipped with a thermal camera to monitor the spatial distribution of red deer in a moor in Denmark at night, and identifies its body state and uses the YOLO model to determine its behavior pattern. The behavior pattern is combined with the local vegetation classification to infer its preference for vegetation type. In addition, the spatial distribution map of red deer is drawn by combining the RTK positioning module on the drone and the vegetation classification map. The paper is well-written and the work done in this paper is quite detailed under the conditions described in the paper. However, compared with similar works, especially in combination with the existing research results in the field of drones, the paper has the following shortcomings:

1) In the literature review, there is basically no mention of other research on animal population distribution and habits using drones; in fact, the paper is not innovative in the use of drones to monitor large animals; of course, there is some innovation in using thermal cameras to conduct related research at night;

2) Since the drone is completely controlled manually to collect data, the author does not give the flight route of the drone. The flight of the drone has nothing to do with the movement route of the red deer. It seems that the drone only takes pictures when it finds them. In the absence of complete monitoring of its nighttime activity route, it does not seem to be very convincing to deduce its foraging preference. In fact, the conclusion of the paper is not essentially different from the conclusion of general zoological researchers. Therefore, it is difficult to say that the new behavioral habits of red deer have been discovered through the methods of this paper;

3) Judging from the pictures provided in the paper, the drone flew very low when identifying the species of deer. In this case, the drone is likely to interfere with the deer's eating or lying behavior, thus affecting the statistical and analytical results of the paper;

4) The data collection work of the paper was carried out in March in Denmark. Whether the seasonal factor is the key factor in determining the foraging preference of red deer, the paper should give a detailed description of the growth status of the entire vegetation in the current season. If necessary, relevant data in other seasons, especially summer, should be collected to provide more sufficient information support for this paper. As it stands, the paper lacks sufficient data and is not convincing.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

This study investigated the use of aerial drones (UAV) equipped with thermal cameras and object detection models to monitor the distribution and behaviour of red deer in different vegetation types. Applied research using these methods are a valuable contribution to the literature as they broaden our understanding of the different approaches and challenges associated with drone wildlife monitoring. Furthermore, this research presents clear and useable management implications that could be applied from their results. They summarise the limitations of their study and approach well making it possible for other researchers to assess the strength of the results.  

 

General comments 

In general, the research is well thought out and the authors have provided detailed explanations and justifications for their approach. However, the introduction could do with a rework to focus more on the core story or message of the paper. The ability to capture behavioural data on the deer using drones is extremely valuable for better understanding what crops are favoured by the deer, discriminating between species and better understanding how they use the landscape. However, I think the manuscript would be greatly improved by focusing the introduction more on explaining and highlighting these benefits and less on a detailed discussion of the biology and ecology of the species. Much of these could be covered more briefly with reference other literature. Their approach could also be applied to many other species or scenarios (such as feral pig management) where identifying behavioural states from drone data is still not common practice. Therefore, an acknowledgement of this, or even a more general framing of the introduction and paper could be help it reach a broader audience. This could also include a change to the title of the paper to include drones – e.g., “Vegetation type preferences in red deer (Cervus elaphus) determined by object detection models using aerial drones” or UAV data.  

 
Specific comments 

Section 1.2 in the intro includes a lot of discussion points and unnecessary detail. I would recommend moving a lot of this to the discussion and/or trimming back the details. Use references to justify rather than explaining all of the biology. 

Section 1.3 could also be shortened down significantly. There is a lot of summarising of other literature rather than focusing on its relevance to this research. 

Reword study aims to be in the same style – e.g., Number two should read “The types of behaviour of red deer using…” 

Section 2.1 – would be improved with a table summarising the flight information and environmental conditions. It was also not clear what was meant by this line: To enhance the reliability of observations and identify potential moving patterns, specific takeoff points were used multiple times 

The drone flight method should go after introducing the drone/in the drone data collection section. “The scouring drone method developed by Povlsen et al. (2023) was used [18]. This method entails...” 

The following line should be referenced: Data on vegetation cover was achieved from the Danish National environmental portal 

Section 2.2 - Whether video or photo capture was used should be mentioned here not later in the methods section 

Section 2.4 - The authors mentioned how many frames were used for training but not how many were included in the data collection phase 

Section 2.5 Model performance this section seems more like results than methods. Possibly a condensed version of this could be mentioned in the methods where needed to justify the approach and the rest in the results. 

Section 2.6 – Should be included in data collection section instead. 

2.7.1 “each population was assigned coordinates is repeated from the previous section. Recommend removing this line. 

Section 2.7.2. This section could be reworded to improve the flow. It is a bit repetitive/disjointed at present. The descriptions of the different vegetation types might be better expressed in a table so that the reader can refer to this when reading the results and discussion. 

Figure 3 and 4. These maps should include some basic map elements such as a scale bar, north arrow, reference map for Denmark and a box indicating where figure 4 fits into figure 3). The solid shapes by population size make it difficult to determine which habitat the populations are in. It may be better to make these either hollow shapes or reduce the amount of information that is included in the figure. For example, I would recommend that time of day only be included in the magnified view. Since population size across habitat types is covered in more detail in later figures it may not be necessary to include in these intital figures. 

 

Figure 5. Grass fields appears to have less points than the sample size (14 vs. 7).  

 

Section 3.4 - there are a lot of numbers in this section that would be easier to follow if summarised in a table 

 

Appendix 

Appendix A and B are unnecessary, Appendix B should be incorporated into Figure 3 as inset map. Appendix A – When this is referenced in section 1.2 the authors should reference the original source material instead. However, I was unable to access the original reference material based on the reference provided as it was incomplete.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The article is prepared at a standard level. Its main weakness is the low level of innovation and interest. I recommend further limiting its length.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The revised paper does cite more literature, but only adds a note in lines 48-49 in the main text, and does not compare these research works with the paper work to illustrate the necessity and highlight the paper work. The author did not make any changes to the rest of the paper.

I agree with the author's point in the cover letter, but the problem is that the research work described in this paper and the conclusion are only in the early stages of the entire research work. The  innovation mentioned by the author, "We combined it with an object detection model that can analyze behavior, which seems novel to us.", is actually not novel compared with peer research.

It is recommended that the author add more seasonal data, including vegetation growth conditions, to enhance the persuasiveness of the paper.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 3

Reviewer 2 Report

Comments and Suggestions for Authors

I agree with part of the author's response in the coverletter, but suggest that the author provide further explanation in the discussion section on the reasons for the "We are not able to add more seasonal data and/or vegetation growth conditions" , so as to eliminate the possibility that some readers may think that the data in this study are insufficient.

 

After all, I am not an expert in ecology, and this paper was submitted to Section Drones in Ecology. Moreover, the author insisted that "it is one rare study of automatic behavioural detection... ", maybe from view of ecology, author is right, so after two revisions, I agreed to publish it.  

Author Response

We have added following to the discussion section to address the seasonal limitation: "Winter food shortage would be expected to affect the behaviour of red deer around the time when this study was conducted. Due to logistical constraints, it was not possible to collect data across multiple seasons or account for varying vegetation growth conditions. As a result, certain seasonal behaviors and changes in vegetation were not captured in this study and should be implemented in future studies."

Back to TopTop