Terrestrial mammalian wildlife responses to Unmanned Aerial Systems approaches

Wildlife conservation using drones and artificial intelligence in Africa

The use of drones and artificial intelligence may offer more reliable methods of counting populations and monitoring wildlife.

Drone Observation for the Quantitative Study of Complex Multilevel Societies

Unmanned aerial vehicles (drones) have recently been used in various behavioral ecology studies. However, their application has been limited to single groups, and most studies have not implemented individual identification. A multilevel society refers to a social structure in which small stable "core units" gather and make a larger, multiple-unit group. Here, we introduce recent applications of drone technology and individual identification to complex social structures involving multiple groups, such as multilevel societies. Drones made it possible to obtain the identification, accurate positioning, or movement of more than a hundred individuals in a multilevel social group. In addition, in multilevel social groups, drones facilitate the observation of heterogeneous spatial positioning patterns and mechanisms of behavioral propagation, which are different from those in a single-level group. Such findings may contribute to the quantitative definition and assessment of multilevel societies and enhance our understanding of mechanisms of multiple group aggregation. The application of drones to various species may resolve various questions related to multilevel societies.

Towards the fully automated monitoring of ecological communities

High-resolution monitoring is fundamental to understand ecosystems dynamics in an era of global change and biodiversity declines. While real-time and automated monitoring of abiotic components has been possible for some time, monitoring biotic components-for example, individual behaviours and traits, and species abundance and distribution-is far more challenging. Recent technological advancements offer potential solutions to achieve this through: (i) increasingly affordable high-throughput recording hardware, which can collect rich multidimensional data, and (ii) increasingly accessible artificial intelligence approaches, which can extract ecological knowledge from large datasets. However, automating the monitoring of facets of ecological communities via such technologies has primarily been achieved at low spatiotemporal resolutions within limited steps of the monitoring workflow. Here, we review existing technologies for data recording and processing that enable automated monitoring of ecological communities. We then present novel frameworks that combine such technologies, forming fully automated pipelines to detect, track, classify and count multiple species, and record behavioural and morphological traits, at resolutions which have previously been impossible to achieve. Based on these rapidly developing technologies, we illustrate a solution to one of the greatest challenges in ecology: the ability to rapidly generate high-resolution, multidimensional and standardised data across complex ecologies.

Robust Algorithms for Drone-Assisted Monitoring of Big Animals in Harsh Conditions of Siberian Winter Forests: Recovery of European elk (Alces alces) in Salair Mountains

Simple Summary

Forest animals can be used as a sensitive indicator of the real state of biodiversity. The research objective was to study the potential of drone planes equipped with thermal infrared imaging cameras for large animal monitoring in the conditions of Siberian winter forests with snow background at temperatures of −5 °C to −30 °C. The surveyed territory included the Salair State Nature Reserve in the Kemerovo Region, Russia. Drone planes were effective in covering large areas, while thermal infrared cameras provided accurate information in the harsh winter conditions of Siberia. The research featured the population of the European elk (Alces alces), which is gradually deteriorating due to poaching and deforestation. The designed technical methods and analytic algorithms are cost-efficient and they can be applied for monitoring large areas of Siberian, Canadian and Alaskan winter forests.

Abstract

There are two main reasons for monitoring the population of forest animals. First, regular surveys reveal the real state of biodiversity. Second, they guarantee a prompt response to any negative environmental factor that affects the animal population and make it possible to eliminate the threat before any permanent damage is done. The research objective was to study the potential of drone planes equipped with thermal infrared imaging cameras for large animal monitoring in the conditions of Siberian winter forests with snow background at temperatures −5 °C to −30 °C. The surveyed territory included the Salair State Nature Reserve in the Kemerovo Region, Russia. Drone planes were effective in covering large areas, while thermal infrared cameras provided accurate statistics in the harsh winter conditions of Siberia. The research featured the population of the European elk (Alces alces), which is gradually deteriorating due to poaching and deforestation. The authors developed an effective methodology for processing the data obtained from drone-mounted thermal infrared cameras. The research provided reliable results concerning the changes in the elk population on the territory in question. The use of drone planes proved an effective means of ungulate animal surveying in snow-covered winter forests. The designed technical methods and analytic algorithms are cost-efficient and they can be applied for monitoring large areas of Siberian and Canadian winter forests.

Feral Horses and Bison at Theodore Roosevelt National Park (North Dakota, United States) Exhibit Shifts in Behaviors during Drone Flights

Drone use has been rapidly increasing in protected areas in North America, and potential impacts on terrestrial megafauna have been largely unstudied. We evaluated behavioral responses to drones on two terrestrial charismatic species, feral horse (Equus caballus) and bison (Bison bison), at Theodore Roosevelt National Park (North Dakota, United States) in 2018. Using a Trimble UX5 fixed-wing drone, we performed two flights at 120 m above ground level (AGL), one for each species, and recorded video footage of their behaviors prior to, during, and after the flight. Video footage was analyzed in periods of 10 s intervals, and the occurrence of a behavior was modeled in relation to the phase of the flights (prior, during, and after). Both species displayed behavioral responses to the presence of the fixed-wing drone. Horses increased feeding (p-value < 0.05), traveling (p-value < 0.05), and vigilance (p-value < 0.05) behaviors, and decreased resting (p-value < 0.05) and grooming (p-value < 0.05). Bison increased feeding (p-value < 0.05) and traveling (p-value < 0.05) and decreased resting (p-value < 0.05) and grooming (p-value < 0.05). Neither species displayed escape behaviors. Flying at 120 m AGL, the drone might have been perceived as low risk, which could possibly explain the absence of escape behaviors in both species. While we did not test physiological responses, our behavioral observations suggest that drone flights at the altitude we tested did not elicit escape responses, which have been observed in ground surveys or traditional low-level aerial surveys. Our results provide new insights for guidelines about drone use in conservation areas, such as the potential of drones for surveys of feral horses and bison with low levels of disturbance, and we further recommend the development of in situ guidelines in protected areas centered on place-based knowledge, besides existing standardized guidelines.

The emergence of marine recreational drone fishing: Regional trends and emerging concerns

Online evidence suggests that there has been an increase in interest of using unmanned aerial vehicles or drones during land-based marine recreational fishing. In the absence of reliable monitoring programs, this study used unconventional publicly available online monitoring methodologies to estimate the growing interest, global extent, catch composition and governance of this practice. Results indicated a 357% spike in interest during 2016 primarily in New Zealand, South Africa and Australia. From an ecological perspective, many species targeted by drone fishers are vulnerable to overexploitation, while released fishes may experience heightened stress and mortality. From a social perspective, the ethics of drone fishing are being increasingly questioned by many recreational anglers and we forecast the potential for increased conflict with other beach users. In terms of governance, no resource use legislation specifically directed at recreational drone fishing was found. These findings suggest that drone fishing warrants prioritised research and management consideration.

Perspectives in machine learning for wildlife conservation

Inexpensive and accessible sensors are accelerating data acquisition in animal ecology. These technologies hold great potential for large-scale ecological understanding, but are limited by current processing approaches which inefficiently distill data into relevant information. We argue that animal ecologists can capitalize on large datasets generated by modern sensors by combining machine learning approaches with domain knowledge. Incorporating machine learning into ecological workflows could improve inputs for ecological models and lead to integrated hybrid modeling tools. This approach will require close interdisciplinary collaboration to ensure the quality of novel approaches and train a new generation of data scientists in ecology and conservation.

Responses of turkey vultures to unmanned aircraft systems vary by platform

A challenge that conservation practitioners face is manipulating behavior of nuisance species. The turkey vulture (Cathartes aura) can cause substantial damage to aircraft if struck. The goal of this study was to assess vulture responses to unmanned aircraft systems (UAS) for use as a possible dispersal tool. Our treatments included three platforms (fixed-wing, multirotor, and a predator-like ornithopter [powered by flapping flight]) and two approach types (30 m overhead or targeted towards a vulture) in an operational context. We evaluated perceived risk as probability of reaction, reaction time, flight-initiation distance (FID), vulture remaining index, and latency to return. Vultures escaped sooner in response to the fixed-wing; however, fewer remained after multirotor treatments. Targeted approaches were perceived as riskier than overhead. Vulture perceived risk was enhanced by flying the multirotor in a targeted approach. We found no effect of our treatments on FID or latency to return. Latency was negatively correlated with UAS speed, perhaps because slower UAS spent more time over the area. Greatest visual saliency followed as: ornithopter, fixed-wing, and multirotor. Despite its appearance, the ornithopter was not effective at dispersing vultures. Because effectiveness varied, multirotor/fixed-wing UAS use should be informed by management goals (immediate dispersal versus latency).

Sociability strongly affects the behavioural responses of wild guanacos to drones

Drones are being increasingly used in research and recreation but without an adequate assessment of their potential impacts on wildlife. Particularly, the effect of sociability on behavioural responses to drone-associated disturbance remains largely unknown. Using an ungulate with complex social behaviour, we (1) assessed how social aggregation and offspring presence, along with flight plan characteristics, influence the probability of behavioural reaction and the flight distance of wild guanacos (Lama guanicoe) to the drone's approach, and (2) estimated reaction thresholds and flight heights that minimise disturbance. Sociability significantly affected behavioural responses. Large groups showed higher reaction probability and greater flight distances than smaller groups and solitary individuals, regardless of the presence of offspring. This suggests greater detection abilities in large groups, but we cannot rule out the influence of other features inherent to each social unit (e.g., territoriality) that might be working simultaneously. Low flight heights increased the probability of reaction, although the effect of drone speed was less clear. Reaction thresholds ranged from 154 m (solitary individuals) to 344 m (mixed groups), revealing that the responsiveness of this guanaco population to the drone is the most dramatic reported so far for a wild species.

Comparing unmanned aerial systems with conventional methodology for surveying a wild white-tailed deer population

Abstract Context Ungulate populations are subject to fluctuations caused by extrinsic factors and require efficient and frequent surveying to monitor population sizes and demographics. Unmanned aerial systems (UAS) have become increasingly popular for ungulate research; however, little is understood about how this novel technology compares with conventional methodologies for surveying wild populations. Aims We examined the feasibility of using a fixed-wing UAS equipped with a thermal infrared sensor for estimating the population density of wild white-tailed deer (Odocoileus virginianus) at the Cedar Creek Ecosystem Science Reserve (CCESR), Minnesota, USA. We compared UAS density estimates with those derived from faecal pellet-group counts. Methods We conducted UAS thermal survey flights from March to April of 2018 and January to March of 2019. Faecal pellet-group counts were conducted from April to May in 2018 and 2019. We modelled deer counts and detection probabilities and used these results to calculate point estimates and bootstrapped prediction intervals for deer density from UAS and pellet-group count data. We compared results of each survey approach to evaluate the relative efficacy of these two methodologies. Key results Our best-fitting model of certain deer detections derived from our UAS-collected thermal imagery produced deer density estimates (WR20204_IE1.gif, 95% prediction interval = 4.32–17.84 deer km−2) that overlapped with the pellet-group count model when using our mean pellet deposition rate assumption (WR20204_IE2.gif, 95% prediction interval = 4.14–11.29 deer km−2). Estimates from our top UAS model using both certain and potential deer detections resulted in a mean density of 13.77 deer km−2 (95% prediction interval = 6.64–24.35 deer km−2), which was similar to our pellet-group count model that used a lower rate of pellet deposition (WR20204_IE3.gif, 95% prediction interval = 6.46–17.65 deer km−2). The mean point estimates from our top UAS model predicted a range of 136.68–273.81 deer, and abundance point estimates using our pellet-group data ranged from 112.79 to 239.67 deer throughout the CCESR. Conclusions Overall, UAS yielded results similar to pellet-group counts for estimating population densities of wild ungulates; however, UAS surveys were more efficient and could be conducted at multiple times throughout the winter. Implications We demonstrated how UAS could be applied for regularly monitoring changes in population density. We encourage researchers and managers to consider the merits of UAS and how they could be used to enhance the efficiency of wildlife surveys.

The behavioral responses of a nocturnal burrowing marsupial (Lasiorhinus latifrons) to drone flight

The use of drones in wildlife research and management is increasing. Recent evidence has demonstrated the impact of drones on animal behavior, but the response of nocturnal animals to drone flight remains unknown. Utilizing a lightweight commercial drone, the behavioral response of southern hairy-nosed wombats (Lasiorhinus latifrons) to drone flights was observed at Kooloola Station, Swan Reach, South Australia. All wombats flown over during both day and night flights responded behaviorally to the presence of drones. The response differed based on time of day. The most common night-time behavior elicited by drone flight was retreat, compared to stationary alertness behavior observed for daytime drone flights. The behavioral response of the wombats increased as flight altitude decreased. The marked difference of behavior between day and night indicates that this has implications for studies using drones. The behavior observed during flights was altered due to the presence of the drone, and therefore, shrewd study design is important (i.e., acclimation period to drone flight). Considering the sensory adaptations of the target species and how this may impact its behavioral response when flying at night is essential.

Improving Animal Monitoring Using Small Unmanned Aircraft Systems (sUAS) and Deep Learning Networks

In recent years, small unmanned aircraft systems (sUAS) have been used widely to monitor animals because of their customizability, ease of operating, ability to access difficult to navigate places, and potential to minimize disturbance to animals. Automatic identification and classification of animals through images acquired using a sUAS may solve critical problems such as monitoring large areas with high vehicle traffic for animals to prevent collisions, such as animal-aircraft collisions on airports. In this research we demonstrate automated identification of four animal species using deep learning animal classification models trained on sUAS collected images. We used a sUAS mounted with visible spectrum cameras to capture 1288 images of four different animal species: cattle (Bos taurus), horses (Equus caballus), Canada Geese (Branta canadensis), and white-tailed deer (Odocoileus virginianus). We chose these animals because they were readily accessible and white-tailed deer and Canada Geese are considered aviation hazards, as well as being easily identifiable within aerial imagery. A four-class classification problem involving these species was developed from the acquired data using deep learning neural networks. We studied the performance of two deep neural network models, convolutional neural networks (CNN) and deep residual networks (ResNet). Results indicate that the ResNet model with 18 layers, ResNet 18, may be an effective algorithm at classifying between animals while using a relatively small number of training samples. The best ResNet architecture produced a 99.18% overall accuracy (OA) in animal identification and a Kappa statistic of 0.98. The highest OA and Kappa produced by CNN were 84.55% and 0.79 respectively. These findings suggest that ResNet is effective at distinguishing among the four species tested and shows promise for classifying larger datasets of more diverse animals.

Networked Unmanned Aerial Vehicles for Surveillance and Monitoring: A Survey

As a typical cyber-physical system, networked unmanned aerial vehicles (UAVs) have received much attention in recent years. Emerging communication technologies and high-performance control methods enable networked UAVs to operate as aerial sensor networks to collect more complete and consistent information with significantly improved mobility and flexibility than traditional sensing platforms. One of the main applications of networked UAVs is surveillance and monitoring, which constitute essential components of a well-functioning public safety system and many industrial applications. Although the existing literature on surveillance and monitoring UAVs is extensive, a comprehensive survey on this topic is lacking. This article classifies publications on networked UAVs for surveillance and monitoring using the targets of interest and analyzes several typical problems on this topic, including the control, navigation, and deployment optimization of UAVs. The related research gaps and future directions are also presented.

Evaluation of the use of drones to monitor a diverse crocodylian assemblage in West Africa

Context West African crocodylian populations are declining and in need of conservation action. Surveys and other monitoring methods are critical components of crocodile conservation programs; however, surveys are often hindered by logistical, financial and detectability constraints. Increasingly used in wildlife monitoring programs, drones can enhance monitoring and conservation efficacy. Aims This study aimed to determine a standard drone crocodylian survey protocol and evaluate the drones as a tool to survey the diverse crocodylian assemblage of West Africa. Methods We surveyed crocodile populations in Benin, Côte d’Ivoire, and Niger in 2017 and 2018, by using the DJI Phantom 4 Pro drone and via traditional diurnal and nocturnal spotlight surveys. We used a series of test flights to first evaluate the impact of drones on crocodylian behaviour and determine standard flight parameters that optimise detectability. We then, consecutively, implemented the three survey methods at 23 sites to compare the efficacy of drones against traditional crocodylian survey methods. Key results Crocodylus suchus can be closely approached (>10 m altitude) and consumer-grade drones do not elicit flight responses in West African large mammals and birds at altitudes of >40–60 m. Altitude and other flight parameters did not affect detectability, because high-resolution photos allowed accurate counting. Observer experience, field conditions (e.g. wind, sun reflection), and site characteristics (e.g. vegetation, homogeneity) all significantly affected detectability. Drone-based crocodylian surveys should be implemented from 40 m altitude in the first third of the day. Comparing survey methods, drones performed better than did traditional diurnal surveys but worse than standard nocturnal spotlight counts. The latter not only detected more individuals, but also a greater size-class diversity. However, drone surveys provide advantages over traditional methods, including precise size estimation, less disturbance, and the ability to cover greater and more remote areas. Drone survey photos allow for repeatable and quantifiable habitat assessments, detection of encroachment and other illegal activities, and leave a permanent record. Conclusions Overall, drones offer a valuable and cost-effective alternative for surveying crocodylian populations with compelling secondary benefits, although they may not be suitable in all cases and for all species. Implications We propose a standardised and optimised protocol for drone-based crocodylian surveys that could be used for sustainable conservation programs of crocodylians in West Africa and globally.

Fewer bat passes are detected during small, commercial drone flights

Advances in technological capabilities, operational simplicity and cost efficiency have promoted the rapid integration of unmanned aerial vehicles (UAVs) into ecological research, providing access to study taxa that are otherwise difficult to survey, such as bats. Many bat species are currently at risk, but accurately surveying populations is challenging for species that do not roost in large aggregations. Acoustic recorders attached to UAVs provide an opportunity to survey bats in challenging habitats. However, UAVs may alter bat behaviour, leading to avoidance of the UAV, reduced detection rates and inaccurate surveys. We evaluated the number of bat passes detected with and without the presence of a small, commercial UAV in open habitats. Only 22% of bat passes were recorded in the presence of the UAV (0.23 ± 0.09 passes/min) compared to control periods without the UAV (1.03 ± 0.17 passes/min), but the effect was smaller on the big brown bat/silver-haired bat (Eptesicus fuscus/Lasionycteris noctivagans) acoustic complex. Noise interference from the UAV also reduced on-board bat detection rates. We conclude that acoustic records attached to UAVs may inaccurately survey bat populations due to low and variable detection rates by such recorders.

Behaviourally mediated biases in transect surveys: a predation risk sensitivity approach

Variation in the behaviour of individuals or species, particularly their propensity to avoid or approach human observers, their conveyances (e.g., cars), or their proxy devices (e.g., drones) has been recognized as a source of bias in transect counts. However, there has been little attempt to predict the likelihood or magnitude of such biases. Behavioural ecology provides a rich source of theory to develop a general framework for doing so. For example, if animals perceive observers as predators, then the extensive body of research on responses of prey to their predators may be applied to this issue. Here we survey the literature on flight initiation distance (the distance from a predator or disturbance stimulus at which prey flee) for a variety of taxa to suggest which characteristics of the animal, the observer, and the environment may create negatively biased counts. We also consider factors that might cause prey to approach observers, creating positive bias, and discuss when and why motivation for both approach and avoidance might occur simultaneously and how animals may resolve such trade-offs. Finally, we discuss the potential for predicting the extent of the behaviourally mediated biases that may be expected in transect counts and consider ways of dealing with them.

Measuring disturbance at swift breeding colonies due to the visual aspects of a drone: a quasi-experiment study

There is a growing body of research indicating that drones can disturb animals. However, it is usually unclear whether the disturbance is due to visual or auditory cues. Here, we examined the effect of drone flights on the behavior of great dusky swifts Cypseloides senex and white-collared swifts Streptoprocne zonaris in 2 breeding sites where drone noise was obscured by environmental noise from waterfalls and any disturbance must be largely visual. We performed 12 experimental flights with a multirotor drone at different vertical, horizontal, and diagonal distances from the colonies. From all flights, 17% caused <1% of birds to temporarily abandon the breeding site, 50% caused half to abandon, and 33% caused more than half to abandon. We found that the diagonal distance explained 98.9% of the variability of the disturbance percentage and while at distances >50 m the disturbance percentage does not exceed 20%, at <40 m the disturbance percentage increase to > 60%. We recommend that flights with a multirotor drone during the breeding period should be conducted at a distance of >50 m and that recreational flights should be discouraged or conducted at larger distances (e.g. 100 m) in nesting birds areas such as waterfalls, canyons, and caves.

Counting Mixed Breeding Aggregations of Animal Species Using Drones: Lessons from Waterbirds on Semi-Automation

Using drones to count wildlife saves time and resources and allows access to difficult or dangerous areas. We collected drone imagery of breeding waterbirds at colonies in the Okavango Delta (Botswana) and Lowbidgee floodplain (Australia). We developed a semi-automated counting method, using machine learning, and compared effectiveness of freeware and payware in identifying and counting waterbird species (targets) in the Okavango Delta. We tested transferability to the Australian breeding colony. Our detection accuracy (targets), between the training and test data, was 91% for the Okavango Delta colony and 98% for the Lowbidgee floodplain colony. These estimates were within 1–5%, whether using freeware or payware for the different colonies. Our semi-automated method was 26% quicker, including development, and 500% quicker without development, than manual counting. Drone data of waterbird colonies can be collected quickly, allowing later counting with minimal disturbance. Our semi-automated methods efficiently provided accurate estimates of nesting species of waterbirds, even with complex backgrounds. This could be used to track breeding waterbird populations around the world, indicators of river and wetland health, with general applicability for monitoring other taxa.

An experimental approach to evaluate the potential of drones in terrestrial mammal research: a gregarious ungulate as a study model

Research on the use of unmanned aircraft systems (UAS) in wildlife has made remarkable progress recently. Few studies to date have experimentally evaluated the effect of UAS on animals and have usually focused primarily on aquatic fauna. In terrestrial open arid ecosystems, with relatively good visibility to detect animals but little environmental noise, there should be a trade-off between flying the UAS at high height above ground level (AGL) to limit the disturbance of animals and flying low enough to maintain count precision. In addition, body size or social aggregation of species can also affect the ability to detect animals from the air and their response to the UAS approach. To address this gap, we used a gregarious ungulate, the guanaco (Lama guanicoe), as a study model. Based on three types of experimental flights, we demonstrated that (i) the likelihood of miscounting guanacos in images increases with UAS height, but only for offspring and (ii) higher height AGL and lower UAS speed reduce disturbance, except for large groups, which always reacted. Our results call into question mostly indirect and observational previous evidence that terrestrial mammals are more tolerant to UAS than other species and highlight the need for experimental and species-specific studies before using UAS methods.

Fright or Flight? Behavioural Responses of Kangaroos to Drone-Based Monitoring

: Drones are often considered an unobtrusive method of monitoring terrestrial wildlife; however research into whether drones disturb wildlife is in its early stages. This research investigated the potential impacts of drone monitoring on a large terrestrial mammal, the eastern grey kangaroo (Macropus giganteus), in urban and peri-urban environments. We assessed the response of kangaroos to drone monitoring by analysing kangaroo behaviour prior to and during drone deployments using a linear modelling approach. We also explored factors that influenced kangaroo responses including drone altitude, site characteristics and kangaroo population dynamics and demographics. We showed that drones elicit a vigilance response, but that kangaroos rarely fled from the drone. However, kangaroos were most likely to flee from a drone flown at an altitude of 30 m. This study suggests that drone altitude is a key consideration for minimising disturbance of large terrestrial mammals and that drone flights at an altitude of 60–100 m above ground level will minimise behavioural impacts. It also highlights the need for more research to assess the level of intrusion and other impacts that drone surveys have on the behaviour of wildlife and the accuracy of the data produced.