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

Abstract 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.


| INTRODUC TI ON
Wildlife ecology and conservation has benefitted over the past decade from the emergence of drones, or unmanned aerial vehicles (UAVs), as a useful and innovative field research tool (Corcoran et al., 2021;Jiménez López & Mulero-Pázmány, 2019;Linchant et al., 2015). Their relatively low cost and ease of use has seen them used for wildlife behavior, density and abundance monitoring (Chabot et al., 2015;Hodgson et al., 2013;Vermeulen et al., 2013), animal tracking (Cliff et al., 2018;Muller et al., 2019), antipoaching monitoring (Mulero-Pázmány et al., 2014), recording of songbirds (Wilson et al., 2017), and the mitigation of humanwildlife conflict (Hahn et al., 2017). The potential of drones for data collection is only just beginning to be realized, but has already increased efficiency of processing and automation of data collection when compared to traditional ecological methods (e.g., ground truthing surveys on foot) (Hodgson et al., 2016;Martin et al., 2012).
Recent studies demonstrate that animal behavioral modification occurs as a result of drone surveying (Arona et al., 2018;Barnas et al., 2018;Bennitt et al., 2019;Brunton et al., 2019;Ditmer et al., 2015Ditmer et al., , 2019Mulero-Pázmány et al., 2017;Pomeroy et al., 2015). Missions that demand clear imagery, such as performing species counts and abundance mapping, require flying close to the animal without causing major movement or undesirable behaviors (e.g., birds mobbing the drone). The flight parameters (e.g., altitude, speed) that cause these disturbance behaviors are largely unknown, and therefore, the rules and guidelines designed to protect animals regarding drone flight are underdeveloped.
Detecting and accurately observing nocturnal species is a challenge that many researchers face and a key reason as to why nocturnal species are understudied compared to diurnal species (Vine et al., 2009). The cryptic nature and secretive movements of nocturnal animals, coupled with their adaptations to low light, make close observation extremely difficult (Balme et al., 2009;Jayasekara et al., 2007). Attempts have been made with drones to detect mammals during the night using thermal cameras (Chrétien et al., 2016;Seymour et al., 2017;Spaan et al., 2019;Zhang et al., 2020), with the primary objective to detect and count the target species. Evidence also exists that radio tracking of wildlife can be facilitated using drone systems (Cliff et al., 2015;Muller et al., 2019), reducing the labor and time costs of searching for animals in rugged terrain or thick vegetation and streamlining the process compared to traditional ground-based tracking.
Currently, there is no information available as to whether there is a difference in the behavioral response exhibited by nocturnal animals to night-time drone flight compared to daytime drone flight. It is to be expected that nocturnal animals display a different reaction as compared to diurnal animals to drone flight based upon the ability to identify the drone by the target species, coupled with the environmental conditions that influence the nature of the reaction (Bevan et al., 2018). The difference between sound propagation during the day and night is significant due to changes in temperature, humidity, wind speed, and atmospheric pressure within the atmospheric boundary layer (Embleton & Daigle, 1991).
Increased turbulence during daylight hours and the formation of stable and reflecting layering of the lower atmosphere contribute to differences in sound propagation (Cosgrove, 1997). We hypothesize that these changes in environmental conditions have the potential to modify the behavior of species to drone flight from night-time to daytime due to the difference in sound propagation properties.
The southern hairy-nosed (SHN) wombat (Lasiorhinus latifrons), a nocturnal, burrowing marsupial herbivore (Taggart & Temple-Smith, 2008), is a highly suitable model species to assess the behavioral response of nocturnal animals to drone flight. The large size of the wombat combined with its easily distinguishable shape against the harsh and barren landscape facilitates nighttime observations in their natural environment. Spotlighting is a common technique employed at night to observe their behavior from great distances, for which disturbance is minimal (Taggart et al., 2003). The home range of the SHN wombat is relatively small (1.3-4.8 ha) (Finlayson et al., 2005), and activity is undertaken close to their warrens which allows for straightforward location at night.
The aim of this study was to determine how drone flight activity influences the behavior of nocturnal animals. Here, we investigate the behavioral response of SHN wombats to drone flight at different altitudes during the day and night. Testing of this method is required to assess the validity of utilizing drones for automated collection of wombat behavioral information and for the remote tracking of SHN wombats. Bare-nosed wombat warrens have previously been mapped successfully using a drone (Old et al., 2019), and therefore, it is important that the behavioral response of wombats to drone flight is explored to assess best practice flight methods. It is hypothesized that the animals will react more strongly to night-time compared to daytime drone flight. Potential reasons are twofold; nocturnal animals have a more developed sense of hearing and night-time meteorological conditions are more favorable for the propagation of sound.

| Study site
This study was conducted at Kooloola Station, near Swan Reach  (Taggart et al., 2007. Limestone (calcrete) forms the foundation for the soil, with zones of alluvial clay present (Walker et al., 2007). The area experiences a semiarid climate, characterized by hot, dry summers, (maximum 48°C) and cool dry winters (approximately 270 mm annual rainfall). The area is also periodically subjected to droughts (lowest 10% of rainfall records) (Taggart et al., 2003.
Kooloola Station has been commonly used for SHN wombat research across the past 25 years (Finlayson et al., 2005;Taggart et al., 1998Taggart et al., , 2007, with regular monitoring activities including spotlighting, camera trapping, sample collection, and tagging/collaring.
As a consequence of the high frequency of spotlighting undertaken at Kooloola, wombats have become habituated to the presence of human activity on nearby tracks and roads.

| Drone flights
No regular drone flights have been conducted at Kooloola Station prior to this study; hence, the responses of the wombats to the drone were nonbiased due to the limited exposure.

| Test drone flights
Preliminary flights (n = 16 flights, n = 30 total transects) were flown over wombats using the same transect methodology and at the same flight altitudes used in the final data collection. The preliminary flights were conducted to refine the methodology and ensure reliable capture of data. The final data collection occurred during 3 consecutive days in October with similar weather conditions to reduce variability of responses. Locating wombats for daytime drone flights (5:30 p.m.-7:30 p.m.) Searching for wombats for daytime drone flight experiments involved driving a 4WD vehicle slowly along tracks on the property searching for wombats on, or near, their warrens. Once a wombat was spotted, the vehicle was positioned behind the closest bush and stopped to ensure the wombat was not disturbed prior to launching the drone. The drone was then deployed at a safe distance from the vehicle and out of sight of the wombat. The observers stood F I G U R E 1 Locations of drone deployments on Kooloola Station, Swan Reach, South Australia. Drone deployments were undertaken on existing vehicle tracks on the property and located close to wombat warrens motionless and semi-obscured from the wombat to avoid any disturbance and monitored the animal's behavior through binoculars (8 × 40 mm).

| Drone flights and wombat behavior
Locating wombats for night-time flights (9:30 p.m.-2 a.m.) Searching for wombats at night involved two observers walking along the vehicular tracks at the station using a spotlight (Ledlenser™ H14R.2 headlamp; Low 60 lumens, Power 450 lumens and Boost 1,000 lumens) to scan for wombats on or near their warrens. The drone pilots followed behind at a distance of approximately 100 m.
If a wombat was identified, a red light was used to signal to the drone pilots to stop and set up ready to launch the drone. While this was happening, the spotlight was shone on the wombat for ~2 min to ensure that any change in behavior observed following deployment of the drone was caused by the presence of the drone and the sound it created and not associated with the spotlight. Previous experience spotlighting wombats over multiple surveys across decades has indicated that wombats are highly tolerant of spotlights, of which poor eyesight can be attributed to. Typically no change is observed in behavior following detection by spotlight unless the animal is within a close proximity (~25 m). Wombat behavior was classified as follows:

| Field data collection and behavioral observations
No behavioral response exhibited (0) Alert but stationary (1) Alert with movement in any direction, but did not retreat to burrow (2) Alert and retreat into burrow (3) SHN wombat behavior was categorized and tabulated into an ethogram (Table 1). Behavioral observations using binoculars (8 × 40 mm) occurred approximately 40-120 m from the wombat/s during drone flights. Other anthropogenic sources of disturbance, such as cars driving on a road adjacent to the property, or planes flying overhead, were noted but were rarely present. In the event these disturbances occurred, flights were postponed until vehicles were clear of the area and the study environment returned to its natural state. A wombat was considered alert if it was observed to lift its head and prick up its ears as a result of the disturbance activity (e.g., drone noise, spotlight, vehicle noise). If there was no change in behavior pre-and post-drone flight, the wombat was considered to have been unaffected behaviorally by drone flight.

| Data analysis
All statistical analyses of wombat behavior taken from the field behavioral observations were undertaken in the R environment (R Core Team, 2020). A general linear mixed effects model (GLMM) was used to evaluate the significance of the covariates time of day, flight altitude, moon phase, and distance between launch site and the wombat. None of these covariates except flight altitude had a significant effect on wombat behavior.

| Survival analysis
In order to evaluate the response of wombats to drone flight altitude, we used the "survival" package (Therneau, 2020). In this instance, "survival" denotes a wombat remaining above ground and not retreating into its burrow. "Survival" probability was calculated as a response to drone altitude, but not for time. This analogy allows for the estimation of confidence intervals around the wombats staying above ground and to separate between night and day responses.

| Ethics
This study was conducted under the University of Adelaide Animal Ethics permit number S-2018-112a. All drone flights were undertaken with prior approval from the Civil Aviation Safety Authority (CASA), and the pilot was in the possession of a Remote Pilot Licence (RePL).

Type of behavior Description of behavior
No Response (0) No observable reaction is displayed. Animal maintained original behavior (e.g., resting and feeding) Alert, no movement (1) Focal animal is alert, head raised, and ears pricked up, searching for the source of the sound Alert with movement, but no retreat into burrow (2) Focal animal is alert, head raised, and ears pricked up, searching for the source of the sound; and displacement of animal from original position Alert and retreat into burrow (3) Focal animal is alert, head raised, and ears pricked up, searching for the source of the sound, full retreat into burrow TA B L E 1 Classification of SHN wombat behavior to night-time and daytime drone flight

| RE SULTS
Behavioral responses differed depending upon drone flight altitude and time of day ( Figure 2). We were able to locate 9 wombats during daylight hours and 59 animals during the night. All wombats exhibited a form of vigilance behavior (1, 2, and 3) during flights, and no wombats were observed to be undisturbed by the drone irrespective of time of day. Note that there is no behavior response type 0 at any altitude in Figure 2.

| D ISCUSS I ON
This study is the first to examine the disturbance behavior of a noc- These results are consistent with observations undertaken on other animals. A study conducted on 7 African herbivore species, of which the drone was either lowered vertically toward the animal from >100 m or approached horizontally at either 10, 20, or 30 m above ground level from approximately 400 m away, showed that the type of response and the tolerance of the drone varied for each species (Bennitt et al., 2019). Some wombats displayed a higher tolerance level to the drone by remaining above ground after the drone had flown over at 30 m altitude, although this was an uncommon occurrence relative to retreat behavior (n = 17). The results suggest that flight altitude is a significant factor of instigating disturbance behaviors for not only daytime but night-time drone flight. Currently, little is known on the disturbance behaviors of any nocturnal animal to drone flight; however, it is likely that the different responses of the wombats to drone flight at night compared to daylight hours was influenced by its sensory adaptations, ability to detect the drone, the sound emitted by the drone, its state of fear and the environmental conditions at the time of drone flight (Bevan et al., 2018).
It is difficult to quantify the extent to which wombats were Night-time meteorological conditions at the field site during drone flights differed markedly to daytime conditions, with time of day appearing to influence the ability of the wombat to detect the drone. During night-time flights, wombats commonly appeared to be searching for the source of the sound, and this was characterized by the lifting of the wombat's head and the pricking up of its ears following drone deployment. These observations confirm that wombats could detect the drone auditorily. It is likely that the wombats were able to hear the drone well before they could see it due to their reputably poor eyesight (Triggs, 2009) associated with their nocturnal and burrowing lifestyle. This lifestyle would favor a heightened sense of smell and hearing, rather than keen eyesight (Taggart & Temple-Smith, 2008). In this study, the drone was flown at heights potentially too high for the wombats to detect the red and green lights on the drone. A large percentage of the wombats observed at night instantly pricked up their ears as soon as the rotors of the drone started spinning for take-off, and less so during the day, most likely due to the scattering loss of sound and masking of drone sound from wind and atmospheric turbulence (Attenborough, 2007;Barber et al., 2010). This observation is consistent with a study of African herbivores, for which the animals also appeared to search for the sound of the drone when it was well out of eyesight (Bennitt et al., 2019). Awareness of the sensory capabilities of the target species is important for the implementation of appropriate study design that allows for minimal disturbance.
Sound propagation varied significantly with time of day.
During this study, night-time conditions were mostly clear with low wind (low cloud cover, average wind speed <5 km/hr). These conditions give rise to ground-based temperature inversions (Attenborough, 2007;Wilson et al., 2015), causing strong downward refraction of sound. As sound levels decline at a slower rate during inversions, this allows sound to travel faster and more directly between two points in these conditions, and in this case, rapidly and more directly downward from the drone to the wombat. Daytime conditions at the field site were also relatively clear with low wind (low cloud cover, wind speed <5 km/hr). High solar radiation conditions result in higher atmospheric turbulence with unstable stratification and upward refraction of sound (Embleton & Daigle, 1991), rendering it more difficult for the wombat to hear the sound emitted from the drone flying overhead. It is likely that the change in conditions from daytime to night-time altered the sound propagation properties emitted from the drone and hence altered the sound characteristics and associated behavioral reaction of the wombats studied.
It is highly probable that the noise emitted from the drone is a novel anthropogenic sound for the wombats, given that the location of the study site is away from any significant urban centers.
Such sounds have the potential to cause disturbance behaviors dependent upon the perceived predation risk by the target species (Meillère et al., 2015;Quinn et al., 2006;Shannon et al., 2014).  (Ware et al., 2015). This was the response of some wombats to the drone and the noise it emitted. It is also common for some species to show no behavioral response to potentially fearful stimuli but elicit physiological symptoms. This has been demonstrated in a study on black bears and their response to drone flight (Ditmer et al., 2015) and is likely the case for the "stationary alert" wombats that did not retreat. Factors such as age, sex, reproductive status, and body condition score all influence an animals perception of predation risk (Gaynor et al., 2019), and it is likely that these factors influenced the vigilance behavior observed in some form.

| CON CLUS ION
The ability of drones to collect data efficiently and with high precision and accuracy is changing the way field researchers design ecological studies. Despite drone use becoming more common in the field, little is known about how animals react to drone flight, particularly at night. The marked difference in behavioral response between day and night flights observed in this study suggests that drone flight has a different influence on the behavior of nocturnal animals compared to previous studies on diurnal species (Bennitt et al., 2019;Bevan et al., 2018). Studies of nocturnal animals that involve drone flight should acknowledge this difference. It is likely that the strong behavioral differences, observed between day and night flights can be reduced by habituation. Test flights to observe and gauge the reaction of the target species prior to primary drone field tests may therefore be appropriate.

ACK N OWLED G M ENTS
We thank Dan Godden for assistance with field work; the station owners for permission to undertake research on their property; and Steven Delean for assistance with statistical analyses.

CO N FLI C T O F I NTE R E S T
None declared.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data used to undertake analyses are available at Dryad (https://doi. org/10.5061/dryad.5dv41 ns6c).