Evaluating aerial net gunning and chemical immobilisation for capture of invasive sambar deer (Rusa unicolor) and red deer (Cervus elaphus) in alpine Australia

Study site

Aerial net gunning operations took place in the southern section of Kosciuszko National Park (36.54°S, 148.35°E), spanning 186 km² in southeastern New South Wales, Australia (Fig. 1). Operations took place over 11 days, when weather conditions were favourable (wind <15 km h⁻¹ and no rain), within two 2-week periods in April (autumn) and November (spring) 2021. The capture area is within Australia’s Snowy Mountains region, and elevation ranges from 1150 to 1917 m above sea level (NASA/METI/AIST/Japan Spacesystems and US/Japan ASTER Science Team 2019). The region is mountainous, and 32.48% of the area has a slope of 18° or greater (NSW Department of Planning and Environment 2021). Vegetation in the capture area is dominated by eucalypt woodlands with tussock grass, fern and shrubby understoreys, and wet open tussock grasslands (Australian Government Department of Climate Change, Energy, the Environment and Water 2021). The capture area has extensive creek and river systems with multiple tributaries (Fig. 1). Temperature from the nearest weather station (~3 km from northern boundary of capture area), measured at 0900 and 1500 hours on operational days, ranged from −2.4 to 10.6°C in operation one and from 4.6 to 12.4°C in operation two (Bureau of Meteorology 2020). Sambar deer, fallow deer and red deer were present within the capture area, and overall deer population density was estimated within 3 weeks of both aerial net gunning operations using distance sampling, via aerial thermal imaging surveys that took place across a larger area (279 km²) of the southern section of Kosciuszko National Park, including the capture area. The first thermal survey took place during 1−3 April 2021, when deer density was estimated at 15.23 ± 4.00 deer km⁻² (18 days prior to commencement of operations), and the second thermal survey took place during 16−17 November 2021, when deer density was estimated at 11.08 ± 2.99 deer km⁻² (5 days after completion of operations; E. O’Dwyer-Hall, unpubl. data).

Fig. 1.

Map of capture area, targeting sambar deer (Rusa unicolor) and red deer (Cervus elaphus), in Kosciuszko National Park, Australia. Yellow bounding box indicates capture area, yellow points indicate locations of successfully captured deer and red points indicate locations of deer mortalities during capture. Watercourses are shown on map in blue (major) and green (minor). The location of the capture area relative to Australia is indicated by the frame in the top left corner. Image sources: Esri, Maxar, Earthstar Geographics.

Field trials

To ensure safe and effective operations, the net gun was deployed in a series of ground and aerial trials to ensure even spread of the net at operation heights 5–10 m above ground level. Running shots were also trialled on a static target to assist pilot and net gun operator to identify operational requirements prior to attempting capture of live targets.

Equipment and procedure

A Eurocopter AS350 B3 Squirrel helicopter, crewed with pilot, navigator and net gun operator (NSW National Parks and Wildlife Service Feral Animal Aerial Shoot Team trained shooter) was used to search the area for deer. Once a deer was sighted, the pursuit phase began, categorised into two parts. Part one involved directing the deer toward open and reasonably level ground, with an emphasis on maintaining a low level of stress and physical exertion on the deer. Part two involved a more rapid pursuit to approach within firing range and net deployment. Part one of the pursuit was limited to 10 min; however, for deer moving uphill, this was reduced to 5 min to reduce the likelihood of fatigue. Once the deer was in a sparsely vegetated flat area suitable for net gunning, part two commenced, which was a maximum of 5-min high intensity chase. During this time, the helicopter flew at a low level (2–5 m above ground level) and speed (≤45 km h⁻¹). Times for each phase of the pursuit were recorded (approximate times were recorded in operation one) (Supplementary Table S1) and ceased if the deer was observed to be panting or running unsteadily or the maximum pursuit times reached. A .308 breakaway net gun (ACE Capture, Invercargill, New Zealand) loaded with .308 blanks (22 grain, Australian Munitions ADI AP70N gunpowder) launched four 186-g weights through four barrels angled to spread a 170 × 170 mm (April) or 280 × 280 mm (November) aperture net when fired downward at heights of <5 m (Fig. S1). Following an initial shot, additional nets were fired if deemed necessary by the shooter, based on animal size, to further restrain the animal (Table S1).

Following capture, the catch team helicopter landed >100 m from the deer and proceeded quietly towards it. The deer’s head was covered with a blanket and a blindfold was fitted if safe to do so. In operation two, an additional net was draped over entangled deer to provide an additional level of containment. Following this, a second helicopter with a pilot, veterinarian, veterinary nurse and collar technician landed nearby to the catch team helicopter. Morphometric measurements (body length) were taken from recumbent sambar deer to allow estimation of body mass. Length−weight relationships were developed from adult male and female sambar deer shot and weighed in Victoria in 2012 (Table S1; Bartareau 2019; Watter et al. 2020; Hampton et al. 2021). We did not have access to equivalent data for red deer, so body mass estimates were based on published estimates for adult male and female red deer (Amos 2015). Final estimated mass and drug doses were adapted upon visual inspection of each deer by the veterinarian (Tables 1 and S1). The veterinarian immobilised animals via intramuscular hand injection into the hindquarters. Sambar deer and red deer received medetomidine (40 mg/mL; prepared by a commercial compounding pharmacy; Bova Compounding, Caringbah, NSW, Australia) and tiletamine-zolazepam (Zoletil®; Virbac, Milperra, NSW, Australia). Medetomidine was given to all deer (~0.1 mg kg⁻¹ in operations one and two), and sambar deer also received tiletamine-zolazepam (~1.0–5.0 mg kg⁻¹; Tables 1 and S1). With the exception of one animal, red deer did not receive tiletamine−zolazepam. Two considerations informed the rationale for the different approaches taken to immobilisation of sambar deer versus red deer. Studies have shown that net-gunned red deer can be safely handled without any chemical restraint (Latham et al. 2020), and that red deer captured via other methods can be safely handled with relatively low immobilisation doses (Wolkers et al. 1994). For sambar deer, however, there is little known about effective capture methods (Hampton et al. 2019a). In addition, owing to the much larger size of sambar deer, the risk of injury to staff was deemed to be much higher for incompletely immobilised animals (Hampton et al. 2019a). The tiletamine−zolazepam dose used was reduced between operations one (~2.5 mg kg⁻¹) and two (~1.0 mg kg⁻¹) (Table 1), in order to reduce recovery times in operation two. Following drug administration, each deer was observed until completely immobilised (recumbency); in some cases, re-administration of medetomidine and/or tiletamine−zolazepam was required if deer were incompletely immobilised (see Table S1). The induction time, the duration between the administration of the drugs and immobilisation being achieved, was recorded for each deer (Table 2; Hampton et al. 2021).

Once drugs were administered and the deer was immobilised, the blanket was removed and a blindfold was attached, allowing the capture and veterinary team to untangle and remove the nets from the animal. Hobbles were applied to front and back legs, and personnel were protected from antlers by applying antler covers (sections of ‘pool noodles’) to antler tips and/or shielding antlers using an antler protection board (see Fig. S1). The deer was repositioned in sternal recumbency with the pharynx higher than the rumen, nose downhill and tongue extended, to reduce the chances of regurgitation of ruminal fluid and aspiration of fluid to the lungs. The animal’s rectal temperature, heart rate, respiratory rate and oxygen saturation (Mindray PM-60 Vet Pulse Oximeter) were recorded every 5 min until medetomidine was antagonised via administration of the reversal agent atipamezole (0.1 mg kg⁻¹) (see Table S1). A sterile eye lubricant was applied, and any injuries were recorded. If oxygen saturation fell below 90%, intranasal oxygen was administered. This was delivered via a size C medical oxygen bottle with an Allied Healthcare Products B&F oxygen pressure regulator in operation one, and an INHALO® medical oxygen cylinder (CD size) with an integrated oxygen regulator in operation two. Oxygen was administered at a flow rate of 1–2 L min⁻¹ and was increased to 5–10 L min⁻¹ if oxygen saturation fell below 85%. If oxygen saturation fell below 80%, animals were immediately given a third of a dose of the reversal agent (see below) by intramuscular injection. If the rectal temperature of a deer fell below 38.0°C (i.e. hypothermia), the deer would be warmed using blankets, or above 40°C (i.e. hyperthermia), the deer would be cooled using water spray and fanning (Hampton et al. 2022). If respiration rate fell to <20 breaths min⁻¹, a third of a dose of atipamezole was administered. Once inspection had determined that the deer was in good physical condition with no significant injuries (i.e. those that would inhibit normal movement), several morphometric measurements were recorded: length from nose to the base of tail (cm); neck circumference (cm at mid neck); and full girth (cm). The veterinarian provided a qualitative immobilisation quality score for each deer (Table S2; Harms et al. 2018).

A GPS tracking collar with inbuilt mortality sensor, programmed to record animal locations hourly, was then fitted (G52D Iridium, Advanced Telemetry Systems, Isanti, MN, USA). Each collar was fitted with an accelerometer, which estimated the proportion of time each collar or deer was moving between hourly fixes. Chloromide® antiseptic spray (Troy Animal Healthcare, Glendenning, NSW, Australia) was applied to any wounds. The processing time, the time between administration of the immobilising dose and the first reversal dose, was recorded for each deer (Table S1). Once all procedures were finalised, medetomidine was reversed with atipamezole ~0.3 mg kg⁻¹ by intramuscular injection. Some deer were given further doses of atipamezole if they did not show signs of recovering within ~15 min of the administration of a first dose (Table S1). Each of these cases were designated as a ‘prolonged recovery’.

Immediately after administration of atipamezole, hobbles were removed, the blindfold was replaced with a small cloth, and catch and veterinarian teams (excluding the veterinarian) moved to >20 m from the deer to allow it a clear path of escape upon recovery. The veterinarian remained to support and monitor the deer in sternal recumbency. When the veterinarian had assessed that the deer had recovered sufficiently, the blindfold cloth and antler protection were removed, and the veterinarian moved slowly away from the recovering deer. A firearm was available to enable immediate euthanasia to be performed should a severe adverse animal welfare event (e.g. a broken leg) result from an unsteady recovery (Hampton and Arnemo 2022). The recovery time was also recorded for each deer – the time from the first reversal dose to the time the deer was first standing (Table S1). This study was approved by The University of Sydney Animal Ethics Committee (Project number: 2020/1844).

Post-release monitoring

To measure movement and mortality for 45 days post-capture, we used generalised additive models (GAMs). Previous studies evaluating post-release welfare and movement of captured deer have monitored deer movement for 30 days post-capture for mule (Odocoileus hemionus) and fallow deer (Northrup et al. 2014; Bengsen et al. 2021); 50 days for roe deer (Capreolus capreolus) (Morellet et al. 2009) and 45 days for red deer (Becciolini et al. 2019).

Data were remotely downloaded for all collars deployed on sambar deer (14) and red deer (five) after 45 days. Mean hourly movement and activity were calculated for each deer for 1–45 days following collaring. Day 1 consisted of hourly data from the first 24 h following collaring for each deer. Days 2–45 commenced at the same time each day for each deer, depending on their time of collaring on day 1.

For each deer, accelerometry data was used to calculate mean proportion of time spent active per day for days 1–45 following collaring. Daily mean Euclidean distance between hourly location fixes from 1 to 45 days following collaring was also calculated. Then, GAMs were used to assess changes in movement and activity over the monitoring period, because the collared deer were predicted to have non-linear responses over time (Becciolini et al. 2019; Bengsen et al. 2021).

To examine trends in deer activity in the days following collaring, four GAMs were fitted, modelling changes in activity separately for males and females of the two species. The models were fitted with mean daily activity (proportion of time spent active) as the response variable, days since capture as the smooth term and operation as an explanatory variable and multiplier in the smoothed term. Separate smooth functions were generated to model activity 1–45 days since capture for operations one and two. Models were fitted with a binomial distribution in mgcv; however, underdispersion was detected in the binomially distributed models, which was corrected using a quasi-GAM model (Zuur et al. 2009).

To examine changes in deer movement in the days following collaring, four GAMs were fitted, as above, with mean Euclidean distance between hourly location fixes, averaged daily, as the response variable, days since capture as the smooth term and operation as an explanatory variable and multiplier in the smoothed term. Again, separate smooth functions were generated to model mean distance between hourly location fixes, averaged daily, for the two operation periods. The response variable underwent logarithmic transformation to improve normality of the residuals. GAMs were fitted with Gaussian distributions, restricted maximum likelihood and thin plate regression spline smoothers. All data wrangling and analyses were performed in R ver. 4.1.2, interfaced through RStudio 2021.09.1. Data visualisations were generated using ggplot2 and visreg packages.

For capture outcomes and animal welfare metrics, we report descriptive statistics (means ± s.d. and 95% confidence intervals [CIs]) to avoid performing multiple post hoc statistical tests with low power (Hampton et al. 2019b; Hampton et al. 2021). Ninety-five percent CIs were calculated using the Clopper–Pearson exact method to account for many cases in which sample sizes were small (n < 10) and data did not follow a normal distribution.

Mortalities and injuries

Three mortalities occurred in total: two in operation one and one in operation two. The two mortalities that occurred in operation one (one sambar deer male, one red male) arose from an unknown acute cause while the animals were being pursued, with no definitive diagnosis made. As a result of these mortalities, and the presumed role that stress imposed by pursuit played in them, the maximum allowed time for phase one was reduced from 10 min to 5 min for operation two. In operation two, one sambar deer died at the time of capture when it became entangled in the capture net, fell into a small water body (a pond) and drowned. The yearling male was dragged from the pond and resuscitation was attempted by the navigator for 60 s with no response. Following this incident, aerial net gunning was not attempted on any deer <100 m from a water body. Five male deer (four sambar deer, one red deer) sustained an antler injury during capture, and five deer (four sambar deer, one red deer) bit their tongue during capture and sustained a superficial tongue laceration (Table S1). No other injuries were observed (Table 2).

Post-release movement and activity

For all deer groups, the mean daily activity (proportion of time spent active) tended to be lowest immediately following collaring. For red deer males, mean activity increased rapidly between days 1 and 10, before declining slightly, and then continuing to increase (Fig. 2). For red deer, mean daily activity varied from 0.10 (day 1) to 0.57 (day 38) for males and 0.22 (day 2) and 0.43 (day 35) for females. For sambar deer, mean daily activity varied from 0.23 (day 3) to 0.44 (day 27) for males and 0.21 (day 1) and 0.46 (day 33) for females. GAM results showed that mean daily activity for both deer species varied significantly over time for all models (red males: F = 20.55, P < 0.001, red females: F = 14.00, P < 0.001, sambar deer males: F = 15.90, P < 0.001, sambar deer females: F = 7.84, P < 0.001; Supplementary Material S2). Mean activity in the 45 days following collaring varied significantly between operations one and two for male and female sambar deer and red female deer, with these groups showing higher levels of activity in operation two compared with operation one. Red male deer activity did not vary significantly between the periods (red males: t = −0.44, P = 0.662, red females: t = 15.06, P < 0.001, sambar deer males: t = 18.98, P < 0.001, sambar deer females: t = 5.29, P < 0.001; Supplementary Material: S2).

Fig. 2.

GAM plots showing daily mean proportion of time active for 14 sambar deer (Rusa unicolor; nine males, five females) and five red deer (Cervus elaphus; two males, three females) for 1–45 days following collaring. For both species, male and females are modelled separately. Red lines represent modelled activity responses for deer collared in operation one (April), and blue lines represent responses from operation two (November). Shaded error bands show 95% confidence intervals.

Modelled movement responses showed that for red deer males, mean movement (hourly distance travelled) increased sporadically throughout the monitoring period, but for red deer females, mean movement declined for ~15 days following collaring, before increasing slightly for the remainder of the monitoring period (Fig. 3). For female sambar deer, and male sambar deer collared in operation one, modelled mean movement peaked immediately following collaring, then decreased in the days immediately following collaring, before plateauing around 10 days after collaring, and then remained relatively stable, with smaller positive and negative fluctuations for the remainder of the monitoring period (Fig. 3). This trend was less pronounced for male sambar deer collared in operation two (Fig. 3). These peaks in modelled movement immediately following collaring were due to some sambar deer dispersing large distances in the days following collaring, then remaining within a small area for the remainder of the monitoring period (Figs 4 and S2). Deer mean movement for red deer varied from 19 m per hour (day 4) to 400 m per hour (day 43) for males and 12 m per hour (day 2) to 187 m per hour (day 1) for females. For sambar deer, mean movement varied from 21 m per hour (day 3) to 143 m per hour (day 1) for males and 22 m (day 4) to 118 m per hour (day 1) for females. Mean movement varied significantly over time for female sambar deer, but not for male and female red and male sambar deer (red males: F = 1.40, P = 0.140, red females: F = 1.94, P = 0.109, sambar deer males: F = 1.45, P = 0.240, sambar deer females: F = 3.90, P < 0.001; Supplementary Material: S2). Mean movement in the 45 days following collaring varied significantly between the two collaring periods in all models except the model explaining female sambar deer movement (red males: t = −3.56, P < 0.001, red females: t = 4.42, P < 0.001, sambar deer males: t = 9.39, P < 0.001, sambar deer females: t = 1.59, P < 0.001; Supplementary Material: S2).

Fig. 3.

GAM plots showing daily mean distance moved per hour for 14 sambar deer (Rusa unicolor; nine males, five females) and five red deer (Cervus elaphus; two males, three females) for 1–45 days following collaring. Sambar deer and red males and females are modelled separately. Red lines represent modelled movement responses for deer collared in operation one (April), and blue lines represent responses from operation two (November). Shaded error bands show 95% confidence intervals.

Fig. 4.

Post-capture movements of sambar deer (Rusa unicolor) and red deer (Cervus elaphus) in alpine Australia. Movement tracks are coloured according to speed (m h⁻¹), measured as Euclidean distance between successive locations recorded hourly. Black points on maps indicate the starting point of each movement path. All movement paths are oriented north.

Animal welfare outcomes

The overall probability of mortality during capture was 17% for red deer and 12% for sambar deer (Table 2); however, the small sample size meant that 95% CIs were broad (Table 2; Hampton et al. 2019b). Between operations one and two, however, the capture method was refined and the frequency of injury and mortality during capture was generally reduced (Table 2). In operation one, the mortalities of one sambar deer and one red deer while being chased by the helicopter led us to reduce the maximum time allowed for phase two from 10 to 5 min for operation two, in line with the maximum chase time reported for helicopter net gunning of red deer in a montane environment in New Zealand (Latham et al. 2020). Subsequently, no captured deer died during chases in operation two, although other factors may have been involved in the two deaths (e.g. individual susceptibility). Reductions in maximum allowed chase times have been associated with fewer animal mortalities for helicopter darting of chital deer (Axis axis) in a tropical environment in Australia (Hampton et al. 2021; Amos et al. 2023).

The mortality of a male sambar deer due to drowning in a small water body while entangled in a net resulted in a change to the capture method whereby animals were only pursued when tracking directly away from water. The capture method described here could potentially be adjusted or refined to suit capture attempts in environments different to the one in which we worked. Our study was in a relatively cold (compared with much of Australia) alpine environment. Should the method be deployed to capture sambar deer within their native Asian range, which is predominately tropical and sub-tropical (Timmins et al. 2015), we suspect that sambar deer would be more susceptible to heat stress.

Immobilisation methods

Chemical restraint of wild Cervidae always incurs animal welfare risks at the time of capture and during the post-capture period (Dechen Quinn et al. 2012; Caulkett and Arnemo 2014; Ortega et al. 2020; Thompson et al. 2020). Therefore, it is important to monitor the post-release responses of animals to new or modified capture methods. Chemical immobilisation was used to mitigate the risk to human handlers, but helicopter net gunning operations in northern New South Wales showed that fallow deer can be captured and processed safely and efficiently without the need for chemical restraint (Bengsen et al. 2021). However, physically restraining smaller fallow deer poses greater human and animal welfare risks than restraining red deer (~50–80 kg heavier than fallow) or sambar deer (~140–200 kg heavier than fallow) (Geist and Bayer 1988). Red deer have been captured via net gunning without chemical immobilisation (Latham et al. 2020), and future studies could investigate applying that method in sambar deer.

Induction and recovery times were reduced for both deer species between operations one and two (Table 2), due to improved body mass estimation procedures and refined drug regimes. The prolonged recovery times observed in operation one were reduced in the second operation by reducing the dose of tiletamine–zolazepam (Table 1). The non-reversible nature of tiletamine–zolazepam has practical disadvantages related to prolonged recoveries, increased risk of associated physiological states such as hypothermia, and increased requirements for monitoring (Mayberry et al. 2014). Once tiletamine–zolazepam doses were reduced in November, mean recovery times were considerably reduced (10 min in November compared with 47 min in April), and rapid recoveries were observed with the completely reversible immobilisation regime of medetomidine alone that was given to red deer (Tables 1 and S1). To allow comparable induction times with greater ease of administration, intranasal drug administration could be considered. Intranasal drug administration is generally easier for non-veterinarians in the field, has fewer and less severe complications and requires less training and experience than intravenous or intramuscular drug administration (Shury et al. 2010).

Post-release behaviour

We reported on the movement and activity of collared red and sambar deer for 45 days post-capture to allow for comparisons with similar studies (Northrup et al. 2014; Becciolini et al. 2019; Bengsen et al. 2021). Although no mortalities were reported during this period, accelerometer data showed that red and sambar deer of both sexes tended to exhibit the lowest activity rates immediately following collaring. Then, throughout the monitoring period, activity rates tended to increase until 30–40 days post-collaring for both species, when activity decreased for all modelled responses, except for female red deer collared in operation two, whose activity continued to increase (Fig. 2). Similarly, activity increased over the first 10 days following capture, before stabilising for fallow deer captured via an aerial net gun, processed without chemical immobilisation (Bengsen et al. 2021). Accelerometer data may be a more sensitive indicator of mild impairment, so for the present study, the increase in activity over the monitoring period, may indicate that deer are recovering from impairment as a result of the collaring procedure (Bengsen et al. 2021).

Free-ranging deer are known to exhibit changes to movement rates and site fidelity resulting from capture and chemical immobilisation procedures (e.g. Morellet et al. 2009; Northrup et al. 2014). For male and female sambar deer, movements peaked immediately following capture and steadily decreased for ~10 days following capture and collaring before plateauing or increasing (Fig. 3). Some deer appeared to exhibit a disturbance response immediately following capture, like sambar deer males 49 and 6, which travelled more than 12 and 8 km (respectively) from their capture locations in ~7 days, undertaking faster long-distance movements interspersed with short periods of slower speed movements, before remaining within a small area and moving slower for the remainder of the monitoring period (Figs 3, 4 and S2). Sambar deer experiencing acute stress are known to undertake long-distance movements interspersed with periods of rest (Semiadi et al. 1994), so capture may have contributed to increased movement rates for a short period following capture. However, this disturbance response was not displayed by all sambar deer; some sambar deer of both sexes remainined within a small distance of the capture area for the entire monitoring period, like sambar deer female 11 (Fig. 4). Short-term increases in movements for the period following capture have previously been reported for many Cervidae species, including moose (Alces alces) and male roe deer, which exhibit flight behaviour and show greater displacement from their home ranges following capture (Morellet et al. 2009; Neumann et al. 2011), as well as for an antelope species, nilgai (Boselaphus tragocamelus; Baumgardt et al. 2023). Like sambar deer, male roe deer and moose are largely solitary and tend not to exhibit strong social bonds (Geist 1963; Hewison et al. 1998), so may be more likely to move large distances following capture.

Comparing movement and activity rates pre- and post-capture would have allowed the delineation of movement changes due to capture (e.g. Morellet et al. 2009; Neumann et al. 2011; Northrup et al. 2014; Baumgardt et al. 2023) but was not possible in the present study. Elevated movement rates for the period immediately following collaring did not occur in red deer (Fig. 3), which did not move as far from their point of capture as sambar deer (Fig. S2), despite moving greater average maximum distances than sambar deer throughout the monitoring period (Fig. S2). Red deer are more social (Clutton-Brock et al. 1982) and may seek to re-establish social bonds following disturbance, remaining closer to their capture area, as shown for red deer female 15 and male 40 (Fig. 4). Conversely, post-capture movements of red deer hinds were not consistent with a flight response to move away from the capture area into forest cover for a period of ~8 days following capture (Becciolini et al. 2019). In the present study, low sample sizes preclude us from more definitely characterising red deer movement post-capture.

Conclusion

This study used an aerial net gun and chemical immobilisation to capture 14 free-ranging invasive sambar deer and five red deer in alpine Australia. The method was used across two periods, with refinement of procedures in the second period reducing mortality probability to ≦0.12, with further refinement and reduction of adverse event frequency likely to be feasible. The method presented here can be adapted and refined, to capture sambar deer and red deer within their native and invasive ranges.