The Drone Revolution of Shark Science: A Review

Round 1

Reviewer 1 Report

This manuscript provides a literature review of the state of drone use in shark research from around the world. They provide details on the state of research starting with how one should conduct shark research using a drone, current areas of research, how advancements in technology may pave the future for drone research and provide overall conclusions on the field of drones in shark research. Although the authors claim to provide an exhausting description of the application, advantages and limitations of using drones specifically for shark research, the paper is disjointed and belabored in places and has some omissions. Interestingly, they fail to cite two detailed book chapters focusing specifically on aerial and underwater drone use to study sharks published in 2019 that cover much of the same information (Kizka & Heithaus, Lowe et al.), which is an odd omission considering they cite another chapter from the same book (Shark Research – emerging technologies and applications for the field and laboratory). Since this field is rapidly expanding, they should consider reducing redundancies and focus on advances subsequent to the publication of these two reviews of aerial and UW drones.

Section 3.1 has a little bit of hand-waving in it since specifics on how drones can mitigate shark bites are missing, and the authors focus solely on shark identification. A little more detail and specifics with regards to how this technology has reduced shark bites on humans is necessary to demonstrate the efficacy of this technology for this purpose. If that has not be empirically demonstrated then they should indicate that clearly. I really don’t understand the point behind Section 3.5 – why are hammerheads the only threatened species that could benefit from this technology? It seems that using drones to survey any schooling species that use surface waters should benefit from this technology. Only two of the papers cited in this section studied hammerheads and therefore this seems more like a “future directions” section than an active arm of shark research using drones.

Section 4.3 is missing some important citations and focuses mostly on the tracking ability of the AUVs instead of the combined tracking and environmental variable quantification. Which brings about another aspect of the ms that is confusing – the authors frequently compare and contrast the use of drone vs telemetry; however, they fail to indicate that these two really are used to answer different questions about behavior.  Telemetry is often used to quantify behavior (e.g., movements, habitat use) of individuals over much longer periods of time than drone surveys.  Often the objectives of telemetry tracking studies are to tag and track enough individuals to characterize behavioral patterns within a population.  Many of the drone applications covered in this review really focus on short-term, snapshots of a population.  These are very different applications and yet often complementary; however, that isn’t really discussed in this review. Lastly, most of the future directions stated here in the paper are, “wait for technology to get better and cheaper”. As drones are just another tool, it would be great to have a (small) section about how drones can be used simultaneously with other methods or specific future directions. Despite these major concerns with substantial revisions, and some key specifications I believe this could be a valuable contribution.

Specific points are listed below.

Title: Succinct and accurate

Overview:

line 59-60: This statement is major overgeneralization.  There are good examples of shark populations that are increasing. Not sure a shark would consider a “shark interaction” with an animal (potential prey) as unfortunate, they would probably consider that just a meal. That might be perceived as unfortunate to a human. Also, an “unfortunate shark interaction” is a very awkward way to saying unprovoked shark bite on a human. Not all shark bites may be “devastating” depending on the species and location of the wound. In some areas the most common bites are from feeding during eco-tourism, which is very different from a bite from a 4 m white shark. Much of this verbiage tends to over sensationalize shark interactions.

The citation here is from a survey-based paper on human perception of the danger of sharks. I’m not sure how this backs up the claim in the statement.

line 96: Please expand on the distances that can be covered. Fixed-wing drones can often cover an order of magnitude greater distance than a multi-rotor drone. As it is the main draw for using fixed-wing drones, I think there is a benefit to putting some relative distances or area coverage per flight. This also becomes the most important comparison with human-flown aircraft. It would be beneficial to provide the benefits and detriments of each tech/application.

Line 117-124:  There are other restrictions in regards to study area besides environmental conditions. In many places, additional drone restrictions/prohibitions are in place to reduce wildlife, particularly bird disturbance. For instance, drone operations are prohibited within US National Parks and National Marine Sanctuaries, although research exemptions are occasionally allowed, altitude limits often reduce the benefits of many commercially available UAVs.

Line 141-149: Authors fail to mention the influence of certifications, training and federal regulations (e.g., differences between recreational/hobby and commercial use). In some countries, drone operations for the purposes of science and research are considered commercial operations and thus require federal certification and licensing. This is an important distinction, particularly since there is growing interest in citizen science and contributions of drone video and flight data via public apps such as iNaturalist and others. Since many recreationists use the same drone technology, what is the future roll of citizen science in using drones for shark research?

line 155-156: This seems extremely glossed over. For instance, a tethered ROV in a nearshore, coral reef environment study vs. an open-water AUV study require much different considerations. As ROV/AUV are being included in this review I think this needs to be fleshed out much more.

3.1 Drones as a tool for shark bite mitigation:

Figure 2 – I like this figure. 

Line 168: I believe this section ultimately suffers quite a bit by the authors not being specific about how drone surveillance leads to shark bite mitigation. Are they a stand-in for shark spotters? If they spot a shark, will lifeguards/officials pull people out of the water? Are they being used to study shark and human nearshore behaviors to educate the public? The authors do a great job of providing limitations for drones as a tool, but don’t really connect the dots for how they will be specifically used.

Line 177: what about fog or marine layer?  It doesn’t have to be raining for a drone to be grounded.  US FAA does not allow flight if visibility is impaired by fog or low altitude cloud cover.

Line 196-197: It may be worth it to provide some information on the economic costs of cull-based methods as well as their obvious ecological ones. Obviously, culling vs drone surveillance flights serve different purposes, but they also have very different cost/benefits.  Some cost/benefit comparisons might be useful here.

Line 199-202: I think the point here is that it shouldn't be the only monitoring tool as it is only partially effective.  Indicating the need for simultaneous multiple tools. Also, how does this utility further reduce shark bite risk? Do you mean that using drone to spot specifically for potentially dangerous species and using that info to pull people out of the water can “potentially” reduce risk.  There is no indication that this actually reduces risk, it primarily informs marine safety personnel and water users.

Line 211-212: I think the authors need to be careful with these statements since as stated in this ms, there is no hard, quantifiable evidence that the presence of shark-spotting drones reduce shark bite frequency.

Line 214-216: The conservation aspects of this paragraph seem weak and out of place. Either be specific (“provide opportunistically collected data to conservation agencies” etc) or take them out. 

3.3. Drone Studies of Shark Behavior and Social Interactions:

Line 305-306: So, what you’re saying is that behaviors and social interactions will be limited to calm, good-weather sample days.

3.4 Shark Behavior Around Whale Carcasses:

I’m not really sure why this is in its own section?  Isn’t this really related to predation and group feeding?  By separating it out makes it seem like this makes some unique contribution to our understanding of group feeding or scavenging. This section is rather belabored and long considering the overall value.

Line 324-326: This seems a little oddly worded. I would flip the structure: “Intricate shark behaviors and interactions…..”

Line 335-336: I'm not sure you can say this view gives researchers the "most" complete view, but certainly a different perspective of these behaviors.

 

3.5 Drone research of hammerhead sharks:

I think this section is inconsequential and certainly over emphasized.  Yes, they have a unique shape making them easier to identify from other sharks, but I think the point they are trying to make is that because they school and can be close to the surface they are observable.  Why isn’t this important for ANY schooling species, regardless of their unique head shape or conservation status?  Why couldn’t this technology be use for silky sharks that are also listed and protected under CITES?

Line 390-392: The authors continue to compare utility of drones vs telemetry; however, these are two totally different tools for answering very different questions.  These are unfair comparisons.

Line 429-433: In my opinion, this entire section should be expanded to include any schooling species or reduced to one paragraph.

3.6 Drone studies of reef sharks

The organization of these sections seems to be related to either group behavior or habitat; however, the authors have commented very little on habitat relationships.  Unlike using telemetry to follow individuals to ascertain behaviors, drones can help provide context to behaviors because they allow for simultaneous observations of the habitat and presence of other marine life for focal individuals.  This is something that should be focused on in these sections, since in many ways, despite the short duration of observation provide important context to observed behaviors.  I think these sections would be much more valuable if all were reframed within that sort of approach. 

4.2 Artificial Intelligence for Shark Monitoring, Detection, and Alerting

It is unclear what the authors see as the main value of this technology?  So, does AI provide a non-expert the ability to identify sharks to species from real-time video?  Is it meant to provide a fully automated system from drone missions to provide on-the-fly image recognition and alerts?  Or do they see the value in using AI engines for scrubbing through 100s of hours of geo-referenced drone footage for automated compilation of survey data?

Line 655-657: It should also be mentioned that training the A.I. can be extremely time intensive depending on labelling/training software available, which makes tuning the A.I. even harder.

Line 673: look into Ho et al. 2017, “Predicting Coordinated Group Movements of Sharks With Limited Observations Using AUVs”. They have shown they can track multiple animals, get trajectories, and predict movement within frame using A.I.

4.3 The Potential of Underwater Drones

This section is particularly under-represented and lacking of sufficient review of the literature.

Line 698: Author’s name is incorrectly spelled “Clark”.  What about White et al 2016?

Line 698-713:  Again, previous reviews of this technology for studying shark behavior has covered all this and the main value of using this technology isn’t really alluded in this review.  What is the point of having this major piece of autonomous oceanographic equipment follow a tagged shark?  This equipment can characterize environmental conditions surround the focal individual, providing important context as to the behaviors observed.  None of this is discussed.

Line 748-750: This seems out of place or too broad a statement. It has been argued in multiple places in this review that drones do not affect the natural behavior of sharks. What about White et al. 2016?

Line 749: Autonomous “glider” vehicles do not have propellors – see Lowe et al. 2019 for specifics of this technology

Outlook and Conclusions

Another important aspect of using drones is the importance of universal storage and archiving data records (e.g, geo-referenced video/images). Too bad the authors don’t discuss the value of data repositories for drone footage and associate flight data that can be archived and shared for use by other researchers to address different questions and make historic comparisons in the future.  Not all technologies provide such robust datasets and I know of very few data repositories for these types of data.

Line 764-765:  Again, telemetry and biologger studies use these tools to address different questions.  So, while tagging may cause some influences in behavior, the duration of these tracking events compared to drone observation are quite different.

Author Response

Drones

 

07 January 2020

 

Dear Ms Leah Li (Assistant Editor) and reviewer 1

 

I have attached an electronic copy of the revised manuscript (drones-1035097) ‘The Drone Revolution of Shark Science: a Review’.  We have made most of the reviewer’s revisions plus additional changes and believe it has considerably improved the manuscript.  I have listed our responses to the various comments  in the attached letter and highlighted these in the main document.

 

We thank the reviewers for their comments. We have also included additional information throughout the manuscript and believe that the revised version is now acceptable for publication.  Please contact me if you require further changes.

 

 

Yours sincerely

Dr Paul Butcher

Principal Research Scientist

New South Wales Fisheries, NSW Department of Primary Industries

Author Response File: Author Response.docx

Reviewer 2 Report

I really enjoyed reading the paper on the use of drones for shark research by Butcher et al. While I am not a shark biologist or a drone specialist, the paper was easy to read and understand. They discussed the history of drones as they pertain to sharks and the future implementation of drones in shark research. I really don’t have any changes to the draft. There was a double comma on line 504, and that’s about it for technical edits.

 

I did have two questions for the authors concerning drones: 1) can aerial drones be used to rapidly deploy and two hydroacoustic sensors, and 2) can drones be used to tag sharks? I would think that you could dart a shark with a tag fired from a drone. Or at least it would be fun to develop. Lastly, in the marine mammal world, drones are used to collect DNA. Can drones be used similarly for sharks?

Author Response

Drones

 

07 January 2020

 

Dear Ms Leah Li (Assistant Editor) and reviewer 2

 

I have attached an electronic copy of the revised manuscript (drones-1035097) ‘The Drone Revolution of Shark Science: a Review’.  We have made most of the reviewer’s revisions plus additional changes and believe it has considerably improved the manuscript.  I have listed our responses to the various comments  in the attached letter and highlighted these in the main document.

 

We thank the reviewers for their comments. We have also included additional information throughout the manuscript and believe that the revised version is now acceptable for publication.  Please contact me if you require further changes.

 

Yours sincerely

Dr Paul Butcher

Principal Research Scientist

New South Wales Fisheries, NSW Department of Primary Industries

Author Response File: Author Response.docx

Reviewer 3 Report

That authors present a review style paper documenting the current use, utility, and future use of aerial and aquatic drones in shark research. Overall, I would say that the paper is relatively well written, and relatively comprehensive concerning the current use of drones. My two major comments concerning the paper are 1) I would have liked to have had the authors document/suggest “best practices” in the use of drones. How and when can they best be used and suggestions for researchers on how to best utilize them, it is also important to suggest when they will have limited utility. and 2) I was slightly confused by the organization of section 4, the authors use these as case examples, however each section is grouped by a different factor some being species focused, environment focused, and scientific question focused. All of these sections seem to have significant overlap with each other, with some sections being very specific, while others are quite broad. For instance, there is “Behavior around whale carcasses”, which overlaps with “shark predation events”, but also overlaps with the general case of “Shark Behavior and Social Interactions”. Otherwise, I think the paper add a contribution to the field and is acceptable for publication.

 

 

Line 49: The authors state that their benefits are “obvious”, I think this statement can be removed and have the authors just state their benefits

Line 51: The authors mention “breeding grounds” where sharks can be seen. This should have a citation if they are referring to a specific location or species. Shark breeding grounds are generally not well known.

Line 110-112:” Regardless of imaging equipment, sensor type and drone choice are influenced by the "data link", i.e., resolution of video telemetry, data bandwidth and meta-data captured”. I am unsure what this sentence means. Does sensor type refer to the type of imaging sensor? If the data is being recorded onboard for post -processing, telemetry does not seem particularly important. From a transect/autonomous perspective this is not true.

Line 120: By “network” are the authors referring to a network like a cellular/GSM network? I thought most drones directly communicated with the base station, or are the authors referring to this “telemetry P2P communication as a network?

Line 133: instead of “distinguishable from the air”, I would just say “visible”. I was initially confused by this statement.

Line 134: Delete “necessarily”

Line 168. While it is implied, the authors never actually state how drone would be used for shark bite mitigation. I assume it would involve regular drone patrol/transects by lifeguards to find sharks. Or is it to follow up on reported shark sightings? While maybe outside the scope of this section there is also the problem of how to best use the information of a confirmed shark sighting….

Line 206-207: “beyond line-of-sight approvals would currently unlikely fulfil safety requirements…” confusing end to the sentence, please rephrase

Like 207 “However, in the coming years this will li…” delete “the”

Lines 237: “In addition, the presence of the observer can alert prey or impact the behaviour of the predator, even resulting in an aborted predation attempt” There should probably be a citation after this.

Line 243: “However under some circumstances …” There should probably be a citation after this sentence, unless the authors are referring to drones being about to remotely observe.

Line 248: I appreciate the use of “Snells window”, however, I did not know what it was and had to google it. Either further describe the phenomenon or just delete “through Snells window”

Line 255: Is there a citation for swimming velocity? It seems in frames where there are no static objects (unable to see the sea floor), calculating accurate fine scale swimming speed would be suspect, as even minor variations in the movement/orientation/elevation of the drone/camera would lead to errors.

Line 364-266: “Advances…” This sentence just kind of hangs here and there is an entire section on sensors and advancements, I would suggest deleting it.

Line 444: Researchers also use “RUVs” they are the same setup, just without the bait, removing the “unnatural” circumstances mentioned here. But are of course limited in their visibility and probability of observation

Line 458 – I am not sure about the citation “93”. I do not believe there is a discussion of how the accuracy of active acoustic tracking decreases over time.

Line 458-460, I am not sure I understand what the “strength” of drones is here. With unlimited trackers and fuel, active tracks could go indefinitely as well. More likely in the case of a drone, the shark would ultimately swim deeper and be visually lost.

Line 461- I am unsure what “more individual tracks” means here. Do the authors mean, since the Drone tracks are shorter, they need more of them to get to a comparable ‘time” threshold as active tracks? Additionally, I am cautious about the use of drones for “habitat use” as they are restricted in the habitats that can effectively find sharks, doesn’t that inherently bias their habitat use estimates?

Lines 496 – I would have liked to have a small discussion about size estimation in this section, via stero cameras and parallax or via parallel lasers. Additionally, some future directions other than just AI would have been nice. Maybe having drones cooperate to film the same individual from multiple angles, allowing for 3d reconstruction of the body, or something.

Lines 504 – there is an extra comma

Line 517 – RGB or other cameras have “additional cost and expertise”?

Like 519 – The authors state that we have not “fully realized” the benefits of these other sensors, but the next paragraphs state basically that other sensors seem to have limited utility in this case.

Line 529 – Anecdotally from who? Maybe include a personal communication statement here

Line 556 – I think the authors need to be explicit when they are talking about artificial intelligence, about what is onboard the drone, what is done in real time remotely, vs what is done in post processing

Line 674- I am biased as I am one of the authors, but I think the authors should cite Lowe et al 2018 “Use of Autonomous Vehicles for Tracking and Surveying of Acoustically Tagged Elasmobranchs” in this section, as it is a review of the utility of underwater drones in shark research. Other chapters in this book (Shark Research: emerging Technologies and Applications for the Field and Laboratory) would also be pertinent to cite throughout this manuscript, such as chapter 4, “Using Aerial Surveys to investigate the Distribution, Abundance and behavior of sharks and rays”.

Line 701 – There have been several more studies using this system, allowing for calculation of depth, greater accuracies, fusion with shark borne IMU information, as well as multi-drone collaborative tracking, which I think is neat!

Line 749 – Should probably not say glider and propeller, Gliders traditionally use change in buoyancy to generate movement not a propellor  

Line 786 – The need for higher resolution telemetry is only needed in some applications. As a general surveying tool with analysis done in post- processing, it is unnecessary.

Author Response

Drones

 

07 January 2020

 

Dear Ms Leah Li (Assistant Editor) and reviewer 3

 

I have attached an electronic copy of the revised manuscript (drones-1035097) ‘The Drone Revolution of Shark Science: a Review’.  We have made most of the reviewer’s revisions plus additional changes and believe it has considerably improved the manuscript.  I have listed our responses to the various comments  in the attached letter and highlighted these in the main document.

 

We thank the reviewers for their comments. We have also included additional information throughout the manuscript and believe that the revised version is now acceptable for publication.  Please contact me if you require further changes.

 

 

Yours sincerely

Dr Paul Butcher

Principal Research Scientist

New South Wales Fisheries, NSW Department of Primary Industries

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

Overall, the changes made by Butcher et al. are appreciated and directly respond to comments requested. The substantial addition to section 3.1 reduce the overly sensational statements and address the lack of evidence for shark bite mitigation. This strengthens their statement that this is one tool in the tool box for bite mitigation. Section 3.4, in my opinion, is still redundant with other sections. However, the preface for section 3 as a whole makes it clear what the intent of the section is. The restructuring of 3.5 makes this section stronger and clearer, and the paper overall stronger. The clarifications and additions address most of my major concerns. I believe this to be a worthwhile contribution to the field.

I would suggest that authors take care with statements such as "Finally, drone systems are already being deployed for spotting dangerous sharks at public beaches", which simply is not accurate.  A drone cannot tell if a particular shark is dangerous.  It can be used to spot "potentially" dangerous sharks. 

Author Response

Dear Reviewer 1

 

Thank you for your comments.

 

We agree with you in regards to using the term 'potentially' where we have already used this at line 60.

 

We have now changed this throughout the manuscript at line 653, 777 and 878.

 

We believe that with these changes the manuscript is now ready for publication.

 

Kind regards

 

Paul