Drones and Wildlife Management: Are We Tapping Into Their Full Potential in An Ethical Way?

Symposium

SESSION NUMBER: 82

Symposia will be available on-demand on their scheduled date, then again at the conclusion of the conference.

 
Small unmanned aerial systems (sUAS) and unmanned aerial vehicles (UAV; hereafter referred to as ?drones?) are increasingly being tested and used in wildlife management. Applications include incidental monitoring, systematic surveys, wildlife health assessments, habitat and landscape mapping, tool/bait delivery, and dispersal of wildlife. Although many of the current uses are preliminary or ?proof-of-concept?, drone technology has the ability to significantly reduce sampling and management costs/efforts, produce high quality data, and provide a unique vantage point to study complex systems. However, uncertainty over drone regulations, platform use restrictions/purchases, and nascent of post-processing data techniques present challenges to their wide-spread use and integration into wildlife management programs. Furthermore, a growing number of studies indicate potential ethical concerns over the exposure of wildlife to drones, yet few studies or management programs explicitly address this issue. To that end, we organized this symposium as a gathering of stakeholders to discuss potential ethical and logistic barriers to drone use in wildlife management. The symposium is broken up into six sections: 1) FAA and USFWS current regulations on drone operation, 2) potential impacts of drone operations on wildlife, 3) unique applications, 4) standardization for systematic drone surveys and reporting of drone operations 5) consideration and techniques for data post-processing, and 6) future directions and formation of a TWS Drone working group.

Overview of Current FAA Regulations and SOPs for Operating sUAS in General and Overview of Requirements and SOPs for Obtaining a Waiver to Conduct sUAS Operations on Military Lands
Rick Spaulding
The presentation will cover the general FAA rules, regulations, and requirements for operating small unmanned aerial system (sUAS) (or drone) in the National Airspace System. Topics to be covered include, but are not limited to, Part 107 rules, registering a drone, FAA airspace restrictions including altitude requirements, Part 107 Remote Pilot certification, and the B4UFLY mobile app. In addition, the presentation will cover the general requirements covering the operation of sUAS on U.S. Department of Defense lands and within military airspace, including obtaining a waiver to conduct sUAS operations on military lands. Examples will be provided from sUAS operations within Navy and USMC installations and ranges, with some additional coverage of any unique Army and Air Force requirements.
Regulations and Permits for Drone Use in Wildlife Management and Research
Stephen Earsom
Unmanned aircraft systems (UAS, drones) are rapidly gaining traction as useful tools in conservation. Policies regulating UAS, however, are often misunderstood. A review will be provided of certain federal statutes and regulations, and how drones can and cannot currently be employed for research and management of natural resources. The focus will be on US Fish and Wildlife Service statutes and regulations including the Airborne Hunting Act, Migratory Bird Treaty Act, Bald and Golden Eagle Protection Act, Endangered Species Act, National Wildlife Refuge System Improvement Act and other regulatory authorities within Title 50 of the Code of Federal Regulations.
The Faa Drone Integration Pilot Program (IPP): Implicationsfor Wildlife Management and Research
James L. Grimsley; Stephen L. Webb; Michael D. Proctor; Dillon B. Payne
The Choctaw Nation of Oklahoma (CNO) is one of nine lead participants in the FAA’s Unmanned Aircraft Systems (UAS) Integration Pilot Program (IPP). The UAS IPP is a national effort to accelerate the safe integration of UAS into the national airspace. The CNO teamed with the Noble Research Institute (NRI) to explore the viability of using UAS (“drones”) to increase efficiencies in rural and agricultural environments for greater productivity. Most of the CNO and NRI work has focused on livestock and herd management; however, these applications have many direct parallels and application to wildlife monitoring, research and management. Applications such as automated animal inventory and counting, health assessment, and location and tracking of large animals are similar between the domains. UAS are also proving useful for situations where sensitive habitats could be damaged or harmed by the presence of human observers or researchers on the ground. However, many of these applications, and further mission types, are not yet possible for broad application due to delays in the regulatory rulemaking process. For example, many wildlife applications may require operations that are beyond visual line of sight (BVLOS) for the drone pilots, and regular BVLOS operations are not yet allowed by the FAA. As part of the UAS IPP, the CNO and NRI have looked at these infrastructure gaps to develop a realistic vision for how the technology will progress and be available for broader application in agricultural and wildlife scenarios. We will present lessons learned from the UAS IPP such as the impact of noise, the ground crew requirements for typical operations, and a proposed research path forward for UAS technology for wildlife applications that includes BVLOS and night operations (e.g., thermal imaging) for surveying wildlife, payload capacity requirements for tracking animals with sensors, and delivery of baits or repellents.
An Overview of the Use of Drone Technology for the Study and Management of Birds
Dominique Chabot; David M. Bird
In an era of rapid global change, ornithologists are relying on an ever-increasing suite of novel tools to answer questions and solve problems related to avian research, monitoring, and management. Over the past 10 years, the use of unmanned aerial vehicles, or drones, has exploded in popularity in ecological studies in general, and ornithology in particular. Drones have many advantages over traditional survey and research techniques, as they may minimize human mortality due to plane crashes, reduce cost and disturbance, increase accuracy, and allow the collection of high-resolution data over large and/or otherwise inaccessible areas. Some of the major areas of application of drones that have emerged in avian research and conservation include: (1) photographic population surveys, including breeding colonies and non-breeding aggregations and the use of different types of camera sensors (e.g. visible, thermal-infrared); (2) individual nest inspections, most often involving raptors; (3) radio-tracking surveys, involving the use of radio-telemetry sensors; (4) acoustic surveys, involving the use of song-recording sensors; (5) avian habitat research and monitoring, involving high-resolution 2D and 3D mapping; and (6) bird dispersal, either for nuisance birds or to deter birds from hazards. This review provides an update on the progress, including success and failures, for ornithologists using, or planning to use, this technology.
Animal Responses to Vehicle Approach: Applications to Use of sUAS in Wildlife Research
Bradley F. Blackwell
Controlled research examining animal responses to vehicle approach is relatively limited, and efforts that have accurately quantified behavioral and physiological metrics relative to approach by aircraft, including small Unmanned Aircraft Systems (<25 kgs; UAS), are few. However, with recent technological advances, sUAS and associated technologies are now at the forefront of ecological research in terms of field tools. Importantly, foundational theory that informs models of animal response to vehicle approach (including terrestrial vehicles and aircraft) is likely inadequate with regard to object novelty, visually salient attributes, vehicle size and speed, as well as approach altitude, direction, and angle, among other metrics. In this presentation, I will 1) briefly summarize current antipredator theory relative to animal escape decision-making; 2) outline necessary metrics to understanding animal (primarily bird) perceived risk from vehicle (including sUAS) approach; 3) discuss how these metrics can be obtained in the field; and 4) consider different operational scenarios in wildlife management by which the likelihood of UAS detection by target animals can be reduced or enhanced for the specific task. sUAS hold promise as tools in animal surveys and hazing applications, but efficacy depends upon a sound understanding of responses that can potentially introduce bias or negatively affect performance.
How Do Birds Perceive Drones? Implications for Using UAS for Wildlife Surveillance Systems or Wildlife Deterrents
Esteban Fernandez-Juricic
Small unmanned aircraft system (sUAS) technology has been applied for wildlife management in survey and hazing/dispersal contexts. However, animal (particularly avian) response to sUAS approach hinges in large part on detection of the aircraft and the potential perception of risk by its approach. However, avian visual perception is radically different from that of humans. Additionally, different bird species see their worlds in different ways. Consequently, our ability to use UAS (unmanned aircraft system) to survey wildlife populations by decreasing UAS detectability or to deter wildlife by enhancing UAS detectability will be a function of how the target species perceives them visually. We will discuss how variation in avian color vision, visual acuity, and motion perception in different bird species could modify the ability to visually resolve the presence of UAS at different distances. We will also establish which UAS colors could increase or decrease their visual conspicuousness for different bird species. We will derive some recommendations to make UAS more or less visible in different ecological conditions.
Response of Turkey Vultures in a Landfill Context to Approach by Drones
Morgan B. Pfeiffer; Bradley F. Blackwell; Thomas W. Seamans; Bruce N. Buckingham; Josh L. Hoblet; Esteban Fernández-Juricic; Steven L. Lima; Travis DeVault
Small unmanned aircraft system (sUAS) technology has been used in wildlife management to disperse animals from select areas. Some new sUAS have been developed to look and fly like a predatory bird (ornithopters), but little is known about their efficacy relative to other sUAS (e.g. fixed wing). We conducted a study recording turkey vulture (Cathartes aura) responses to an ornithopter, fixed-wing and multirotor sUAS performing targeted and overhead flights. Turkey vultures are considered a hazard to aviation safety, which warrants a need for deterring them around airports. Prior to field experiments, we calculated the minimum number of sUAS flights (n = 93) needed to obtain 95% power for the interaction of sUAS platform and approach on the proportion of vultures that remained after a flight. Consequently, we conducted 100 sUAS flights over 38 days. We launched an ‘eye in the sky’ multirotor sUAS (180 m away) and ascended to 119 m before positioning over the focal vulture in order to record bird reactions. Flight-initiation distance (FID) for reacting vultures (n = 200) was calculated using geo-referenced images from the ‘eye in the sky’ sUAS recordings. Escape reaction variables were evaluated for temporal autocorrelation and modeled accordingly. Here we present our results for proportion of vultures that remained, reaction time, FID and latency to return. Our results will help understand perceived risk of vultures to sUAS operations in order to enhance bird dispersal.
Bison and Wild Horses: Large Mammal Responses to Drones
Chris Felege; Robert Newman; Blake McCann; Susan Ellis-Felege
In recent years, drones have become widely adopted as a tool for wildlife ecologists to conduct surveys. Drones provide a variety of benefits from safety, to repeatability, to high resolution imagery that can be used for population estimation and habitat assessments. One assumption and often touted benefit of drones is that they are a non-invasive method of studying animals. However, there are very few studies that address behavioral impacts of drones on terrestrial mammals and those that exist suggest species-specific responses. The objective of our work was to evaluate behavioral responses of charismatic megafauna, bison (Bison bison) and wild horses (Equus caballus), to drones which we observed during habitat assessments at Theodore Roosevelt in July 2018. We flew a fixed-wing Trimble UX5 at 400 ft AGL over horse and bison herds in a lawn-mower survey pattern and recorded behavioral responses on a Samsung Galaxy 7 phone camera. We reviewed the video to categorize behaviors of vigilance, resting, feeding, aggression, traveling, and grooming at least 15 minutes before a flight, during the flight (15-23 minutes), and 15 minutes after the flight. We found horses (n = 21) increased vigilance during the flight and this increased vigilance persisted during the 15-minute post-flight observation period. Bison (n = 20) spent 25% less time, on average, resting and increased traveling and feeding behaviors during the flight and these behaviors persisted during the post-flight observations. We found our drone surveys altered behaviors; however, future work should evaluate potential long-term fitness implications and physiology of such behavioral shifts. Given all flights occurred at the highest altitude permitted under FAA Part 107 regulations, and in a manner not targeting these animals but rather surveying the landscape they were on, wildlife managers should consider if species-specific regulations are necessary when drones are used.
USDA Wildlife Services’ Integration of Unmanned Aerial Vehicles for Wildlife Damage Management
Steven Smith
Wildlife Services (WS) first applied Unmanned Aerial Vehicles (UAV) outside of research in 2015 when the first two Certificate of Authorizations (COA) were granted by the Federal Aviation Administration (FAA) to personnel in Georgia and Idaho. In the five years since WS’ use of UAVs has expanded to include nearly 100 certified pilots and an internal Working Group for providing oversight, training, and support. Wildlife Services pilots have logged over 500 hours of flight time on multiple platforms for applications including damage assessment and surveys, location scouting, remote trap checks, harassment of injurious species, and locating target animals. In 2019 WS was issued an agency Certificate of Waiver (COW) to conduct night operations. As of May 2020 ten WS pilots have been trained and certified to carry out night operations under this COW for locating and surveying white-tailed deer, feral swine and other nocturnal species.
Using Drones in the Management of Agriculture-Wildlife Conflict: Can Large Blackbird Flocks Be Deterred from Commercial Sunflower Fields?
Page Klug
A promising tool in wildlife damage management is the unmanned aircraft system (UAS). UAS are an ideal frightening device given they are a multi-functional tool for agriculture, are easy to fly, and can be modified to enhance the antipredator response of nuisance species. Thus, we have begun evaluating UAS as a hazing tool to minimize blackbird damage to sunflower. My first objective is to discuss the antipredator response of captive red-winged blackbirds (Agelaius phoeniceus) to three UAS platforms (i.e., multirotor, fixed-wing, and predator model) approaching at direct and overhead trajectories. My second objective is to discuss field evaluations of UAS efficacy as a hazing tool including the influence of altitude and horizontal distance on the escape response of large mixed-flocks of blackbirds (Icteridae). We did not observe an effect of trajectory on alert response in captive birds, however, blackbirds alerted to the predator model 8 seconds earlier than the fixed-wing and 13 seconds earlier than the multirotor. Additionally, blackbirds returned to foraging earlier and alarm-called and took flight less frequently in response to multirotor approaches compared to the predator model. Overhead approaches failed to elicit flight, suggesting UAS hazing may be most effective at low altitude, direct approaches. In direct approaches, only the multirotor failed to elicit an escape response. In the field blackbird flocks responded to all three platforms by taking flight. In a separate study, blackbird flocks did not respond to a fixed-wing flown at 52 m above ground level (AGL), but exhibited responses to a multirotor when flown within 30 m AGL. Flocks responded to a large spraying UAS (5 m AGL; 4 m/s) at an average of 38 ± 9.6 m. Future research will explore increasing the negative stimulus (e.g., multiple drones in coordination and deploying nonlethal chemical repellents) to enhance deterrence on large flocks.
Drones Extend the Reach of Wildlife Management Tools
Tim Shields
Aerial drones offer wildlife biologists and managers tools to extend the reach of established techniques as well as opportunities for novel approaches to previously insoluble problems. Population management of nuisance birds includes lethal removal of individuals and preventing clutches from hatching by coating eggs with oil, a substance which prevents oxygen flow to the bird embryo. Lethal removal of adult birds has drawbacks: poisoning is controversial and shooting expensive and logistically challenging. Egg oiling can be limited by the physical location of nests, however we highlight a drone-based solution. Through the use of a remote fluid application system (RFAS) mounted on commercially available UAVs we have been able to treat large numbers of Common Raven (Corvus corax) nests in the interest of desert tortoise (Gopherus agassizii) and greater sage grouse (Centrocercus urophasianus) conservation, both of which species suffer from heightened predation from this subsidized predator. Our work has included treating raven nests on electrical utility transmission towers, an important artificial nesting substrate, as well as a variety of natural substrate nests. Having refined the technology and methodology of drone-based remote egg oiling, the potential to address avian threats to other threatened and endangered species is clear.
Overview of sUAS Imagery for Wildlife Monitoring
Cary McCraine
Unmanned aerial vehicles (UAVs) have proven their usefulness in a multitude of fields including precision agriculture, surveying, and wireless networks. The ability to carry an array of different sensors makes them an extremely versatile tool for anyone trying to collect small scale remotely sensed data. There are numerous cases of UAVs being used in wildlife settings such as to deter nuisance animals and monitor wildlife activity. This talk will primarily concentrate on photogrammetry of wildlife using UAVs. It will emphasize the basics of sensor selection and how to optimally collect data based on the end goal. Additionally, it will cover a couple sample workflows and lessons learned from past wildlife/UAV projects.
Identifying Wildlife from Aerial Imagery Using Cnns
Sathish Samiappan; Meilun Zhou
Aerial surveys and monitoring of wildlife using unmanned aerial systems (UAS) are cost-effective methods to assess changes in abundance, distribution, and species identification. The UAS can be fitted with a variety of image sensors to detect light in both visible, near-infrared, and thermal wavelengths. High-resolution imagery collected from the UAS enables new possibilities that are not available in the past. Manually analyzing UAS collected imagery can be laborious and unfeasible in most scenarios, so automated assessment by computer models is desired. The supervised pattern classification algorithms such as Convolutional Neural Networks (CNN) is shown to produce excellent classification of objects. Building a supervised learning framework for the recognition of wildlife involves a training step and a testing step. During training, features of the animals are learned from the training images or examples, which have already been labeled by an expert (a label is ground reference annotation). During the testing step, the trained model evaluates the incoming image and outputs a label or prediction. Deep learning approaches such as CNN for detecting and estimating animals from aerial imagery over large areas has gained popularity in the past five years. As object recognition models, deep learning algorithms such as CNN offer more accurate estimation compared to traditional statistical and other non-parametric approaches and is popular for wildlife survey. CNN’s not only training the classification machine learning model but also includes the salient feature extraction step tailored to the problem. The key aspect in animal detection from the UAS collected imagery is the representation of animal appearance heterogeneity due to different species, color, pose variations, motion blur due to animal and UAS movements, scene illumination changes. CNN is proved to extract features that are invariant to the abovementioned constraints. This talk will cover the fundamentals of CNN architecture, open-source tools, and challenges.
Standardized Protocol for Reporting Methods When Using Drones for Wildlife Research
Susan N. Ellis-Felege; Andrew F. Barnas; Dominque Chabot; Amanda Hodgson; David W. Johnston; David M. Bird
Drones are increasingly popular tools for wildlife research, but it is important that the use of these tools does not overshadow reporting of methodological details required for evaluation of study designs. The diversity in drone platforms, sensors, and applications necessitates the reporting of specific details for replication, but there is little guidance available on how to detail drone use in peer-reviewed articles. Here, we present a standardized protocol to assist researchers in reporting of their drone use in wildlife research. The protocol is delivered in six sections: Project Overview; Drone System and Operation Details; Payload, Sensor, and Data Collection; Field Operation Details; Data Post-Processing; and Permits, Regulations, Training, and Logistics. Each section outlines the details that should be included, along with justifications for their inclusion. To facilitate ease of use, we have provided two example protocols, retroactively produced for published drone-based studies by the authors of this protocol. Our hopes are that the current version of this protocol should assist with the communication, dissemination, and adoption of drone technology for wildlife research and management.
The Data Deluge: Integrating AI to Streamline Image Analysis for Wildlife Surveys
Sally E. Yannuzzi; Travis J. Desell; AbdElRahman ElSaid; Joshua Riedy; Barrett M. Sather; Jared Westrem; Susan N. Ellis-Felege
Wildlife surveys are vital to informing state and federal regulations, hunting seasons, management techniques, and population monitoring. However, their implementation, data quality, and prompt completion can be sacrificed as a result of time, money, accessibility, and surveyor fatigue and expertise. With technological advances such as drones, wildlife scientists have been able to improve data collection in many of these areas. However, reviewing imagery captured by drones can be a time intensive process in itself that is not void of detection error. While it has been shown that less experienced reviewers such as citizen scientists can effectively be used to aid in image processing, this still takes countless hours and can slow down data feeds. Integrating artificial intelligence through the use of a trained neural network can significantly speed up this data deluge. While initial training of the neural network can be time consuming, once completed, image reviewing can occur in real-life time. Through the application of this technique, wildlife data collection can be performed semi-autonomously and rapidly, producing higher quality data for less cost and increased efficiency.

 
Organizers: Morgan Pfeiffer, USDA WS National Wildlife Research Center, Sandusky, OH; Rick Spaulding, ManTech International Corp., Bainbridge Island, WA
 
Supported by: Drone Working Group, Military Lands Working Group

Symposium
Location: Virtual Date: October 1, 2020 Time: -