Camera traps provide insight into factors influencing trap success of the swamp wallaby, Wallabia bicolor

Additional keywords: activity pattern, capture, macropod, moonlight, moon phase, non-target.

Trapping programs for mammals typically report low (<10%) capture success (e.g. Woinarski et al. 2001; Vernes and Pope 2006; Morrant et al. 2010; Albanese et al. 2011). Environmental factors may influence trap success. For example, Caley (1994) reported that captures of feral pigs, Sus scrofa, were affected by season and forest type in tropical habitats of the Northern Territory, and van Hensbergen and Martin (1993) showed that trap success of small mammals in South Africa was influenced by rainfall, wind speed, temperature, relative humidity and moonlight. Aspects of the trapping technique itself are also influential. Intrinsic factors such as the type of trap, arrangement of traps, trap position within a trapping array, free-feeding before trapping, type of bait and practical experience of setting traps can all impact on trap success (Greenberg et al. 1994; Ruette et al. 2003; Cunningham et al. 2005). While these environmental and intrinsic factors are known to influence trap success, there is little information on the ways in which target species interact with traps (but see Jury et al. 2001).

Although most macropod species cannot be trapped using conventional trapping techniques (Coulson 1996), some will enter traps. The red-bellied pademelon, Thylogale billiardierii, and Bennett’s wallaby, Macropus rufogriseus, have often been trapped in Tasmania (Le Mar and McArthur 2005; Wiggins et al. 2010), and rock-wallabies, Petrogale spp., have been trapped in many areas on the Australian mainland (Bluff et al. 2011; Willers et al. 2011). Another macropod species that can be trapped is the swamp wallaby, Wallabia bicolor. We conducted a trapping program for this target species, and simultaneously set cameras to monitor visits to the traps. We investigated (1) the influence of a range of external variables on the rate of visits by the target species, and (2) the behavioural responses of the wallabies to aspects of the traps themselves.

We conducted this study in the Upper Yarra catchment, one of the three main water catchments located in the Yarra Ranges National Park, situated ~100 km north-east of Melbourne. The forest at this location was burnt by wildfire in 1983 (Ough 2001), and has more recently been subjected to heavy browsing pressure by a high-density population of sambar, Cervus unicolor (Bennett 2008). The study site is bordered by the Upper Yarra Reservoir, located in a valley surrounded by a range of forest types including the Ecological Vegetation Classes (EVCs) Wet Forest, Damp Forest, Riparian Forest, Heathy Dry Forest and Shrubby Foothill Forest (Department of Sustainability and Environment 2008). The aspect in general is north- and west-facing, and the topography is steep in places and dissected by ephemeral streams. The elevation rises from the reservoir at ~380 m to 640 m above sea level on the adjacent ridge. Annual rainfall averages over 1000 mm (Bureau of Meterology 2013), with rainfall and temperature over the study period being similar to the long-term means.

The swamp wallaby is a generally solitary, medium-sized macropod found in a variety of habitat types along the east coast of Australia (Merchant 2008). Swamp wallabies are predominantly active during nocturnal hours, and rest during the day in dense cover (Swan et al. 2008). While commonly described as generalist browsers (Hollis et al. 1986), swamp wallabies feed selectively, preferring shrubs and forbs over other plant types (Davis et al. 2008; Di Stefano and Newell 2008). The main predators of the swamp wallaby are wild dogs, Canis familiaris (Triggs et al. 1984; Glen et al. 2006), and dingoes, Canis lupus dingo (Robertshaw and Harden 1986). While several studies have found swamp wallaby in scats of the red fox, Vulpes vulpes, Lunney et al. (1990) considered that consumption of carrion was the most likely source. Wild dog and fox trapping was regularly conducted at the study site, but had ceased for approximately a year before this study (last trapping program conducted 5 March to 26 April 2010). Following the course of this study, six tagged wallabies were killed by wild dogs between October 2011 and November 2012.

We trapped swamp wallabies to allow the fitting of radio-collars to investigate the habitat use of this species for another study. We used soft-walled double-layered traps (70 × 70 × 100 cm) specifically designed for swamp wallabies (Di Stefano et al. 2005). For ease of transportation, we modified the rigid frame used by Di Stefano et al. (2005) into a folding design. To reduce the potential for injury to trapped animals, we added dense water-repellent ‘pool noodle’ foam to pad the underside of the top cross-bar. To prevent traps rolling, potentially causing injury or allowing wallabies to escape, we secured the frame to the ground with tent pegs. We dug a shallow pit under the trap, which extended almost the full width and a third to half the length of the floor, to allow the trigger mechanism to function, and used a low-fragrance spray lubricant (Inox-mx3) on the trap hinges to ensure the door could swing freely (see Di Stefano et al. 2005 for further details on the trap mechanism). For practical purposes of trap transportation, baiting and checking, we placed all traps within 40 m of access tracks, which form a network across the catchment. Pollock and Montague (1991) recommended traps be distributed a minimum of 15 m apart to reduce distress and disturbance to wallabies caught in adjacent traps. Similarly, J. Di Stefano (University of Melbourne 2011, pers. comm.) recommended the placement of multiple traps within a small (e.g. 50 × 50 m) area so that if one trap were closed, another would be available in close proximity, and to increase the efficiency of checking traps. We therefore distributed traps in groups of two to four, 15–30 m apart, with each group >500 m apart in a variety of forest types (Ecological Vegetation Classes) including Riparian Forest, Damp Forest and Shrubby Foothill Forest. We deployed six folding traps on 9 April but, owing to the apparent low density of the target species, deployed a further eight standard, rigid traps on 1 June and another five on 24 July for a total of 19 traps. To further increase the speed of trap checks and minimise disturbance at traps when checks were conducted, we attached reflective tape to the door and door handle of all traps (S. Garnick, University of Melbourne 2011, pers. comm.) so that it was apparent by torch light from a distance at night whether the door was closed.

We commenced weekly free-feeding with roughly chopped carrots in and around each trap six weeks before trapping. We locked the trap door open to habituate wallabies and encourage them to enter the traps. Initially, we baited traps with only carrots, but after a month tried a mix of carrot, pear and apple, with the occasional addition of peanut butter, as used by Di Stefano and Newell (2008), because the carrots did not appear to be attractive to swamp wallabies. However, these additional baits generated a frenzy of bobuck, Trichosurus cunninghami, activity but apparently little swamp wallaby activity, so we increased the quantity of carrots at each trap, and did not offer other baits. Instead, we trialled bait boxes, robust plastic containers with holes drilled in the lid, containing apple and pear. We placed these boxes in the traps in addition to the carrots in the hope of providing an attractive scent at the trap, but again this did not appear to increase swamp wallaby activity so we removed the bait boxes after a month, and used only carrots thereafter. Following the free-feeding phase and bait trials, we placed most carrots inside each trap, as well as a few outside the trap (to entice swamp wallabies to approach), every day while trapping and at least every three days between trapping bouts. We trapped over 45 nights between 4 May and 8 September 2011, for a total of 620 trap-nights. Initially, we set traps 2–3 h before dusk and checked them once a day in the morning at first light, and then locked the door open. However, the few swamp wallabies that had been tagged in the first two months became regular visitors at dusk, so from June we additionally checked traps 2–3 h after sunset to ensure that these animals did not unnecessarily remain in traps overnight, and to maximise the number of available traps to untagged wallabies.

We set infrared-triggered wildlife cameras (Reconyx HC600 Hyperfire) at 10 traps on 7 July 2011 to better understand the apparent low visitation and capture rate. We installed a further three cameras on 15 July and 11 August for a total of 16 cameras operating until 8 September 2011, when trapping was completed. The number of operational cameras fluctuated according to the number of traps installed at the time and with occasional camera malfunction. We located cameras 1–3 m in front, or to the side, of traps depending on the availability of a suitable mounting point, and aimed the cameras at the door of traps. We set cameras without time delays (24 h) so that, once triggered, three consecutive images were recorded 1 s apart, and were generally sufficient to identify tagged swamp wallabies. Each image file recorded the date, time and temperature. The date and time stamp allowed us to classify visits to traps as daylight, twilight or night, and whether there was moonlight present. Visits falling between times of moonrise and moonset were classified as moonlight present, and all other visits classified as moonlight absent. We did not record cloud cover, so variation of light levels within these two categories were not considered in analysis. The date stamp also allowed us to determine the moon phase at the time of visits, which we classified as either gibbous (first quarter), new, crescent (third quarter) or full, encompassing three days before and after the phase. Infrared images are monochrome, so we used reflective tape in unique patterns on the collar and barrel of the swamp wallaby radio-collars to identify individuals. We recorded the duration (minutes within camera view) of each swamp wallaby visit to traps, whether they entered the trap, and finally whether they were trapped. To allow for the variation in the number of operational cameras, we used the visit rate of swamp wallabies (total visits/cameras) for analyses from 7 July until 8 September 2011. Due to the small sample size, we did not differentiate visits by tagged wallabies from untagged wallabies for analyses. We used camera data to examine the relationship of the swamp wallaby visit rate to time of day, time after sunset, moon phase, moonlight and temperature, which previous studies have suggested may influence swamp wallaby activity (Osawa 1989; Di Stefano 2007; Swan et al. 2008), and to additionally examine the events that led to the successful trapping of a swamp wallaby. Following examination of residual plots, we normalised the data using a square-root transformation, and analysed using analysis of variance in Genstat 14, with Fisher’s least significant difference post hoc tests for paired comparisons of means.

Trap success of swamp wallabies was low (9.4%), we captured just 23 individual swamp wallabies and had 38 subsequent recaptures over 620 trap-nights. The camera traps detected 1317 swamp wallaby visits at a mean (±s.e.) rate of 1.39 ± 0.14 per day (24 h). The temperature during visits ranged from –2 to 19°C (mean ± s.e. = 6.2 ± 0.1°C), but was not related to the visit rate (F_1,543 = 0.10, P = 0.753). The majority (84%) of visits occurred at night (F_2,122 = 60.28, P < 0.001) (Fig. 1), but the frequency of visits decreased throughout the night (F_1,766 = 107.58, P < 0.001) (Fig. 2). Nocturnal visit rate was also influenced by the phase of the moon (F_3,60 = 10.69, P < 0.001), being highest around crescent and new moon, but moon phase did not affect the duration of these visits (F_3,58 = 0.39, P = 0.759). In addition, visits were also more frequent when the moon was set (F_1,60 = 82.67, P < 0.001), and moon phase and moonlight interacted (F_3,60 = 5.92, P = 0.001), such that almost all visits under moonlight occurred during crescent and new moons (Fig. 3).

Fig. 1.  Mean number (±standard error) of swamp wallaby visits per trap in 24 h classified by sun position, Yarra Ranges National Park, July–September 2011. Daylight = sunrise to sunset; Twilight = civil sunrise (first light) to sunrise, and sunset to civil sunset (last light); Night = civil sunset to civil sunrise. Solar times obtained from Geoscience Australia (2011).

Fig. 2.  Mean number of swamp wallaby visits per trap in each hour after sunset (y = –0.013x + 0.174, P < 0.001, adjusted R² = 0.122), Yarra Ranges National Park, July–September 2011. Sunset data obtained from Geoscience Australia (2011).

Fig. 3.  Mean number (±standard error) of nocturnal swamp wallaby visits per trap classified by moon phase (±3 days) and presence of moonlight, Yarra Ranges National Park, July–September 2011. Lunar data obtained from Geoscience Australia (2011). Fisher’s least significant difference post hoc test was used to calculate overall significant differences between presence of moonlight and for the interaction of moonlight and moon phase where a < b.

The outcome of swamp wallaby visits to traps that were set (n = 340) is summarised in Fig. 4. Wallabies did not enter traps on 59% of visits. We could not discern why wallabies did not enter traps for 28% of visits, but some reasons were evident: wallabies consumed bait outside the trap then left the area on 11% of visits, and the door was already closed on 17% of visits, due to other animals, including swamp wallabies (7% of visits), bobucks (6%) and wombats (<1%), as well as some unknown causes (3%). Bobucks were the most frequent non-target species to visit the traps, but were generally able to escape through a small gap in the wool bags that form the trap walls (Fig. 5a), so we had to release only two individuals. We did not quantify the activity of bobucks, except when they coincided with swamp wallaby visits to traps, but bobucks were generally early visitors to traps each evening and preferentially consumed bait outside traps before moving into the traps. When bobucks entered traps, they commonly took one piece of carrot, often without triggering the trap, then left the trap to consume it before returning for another. Common wombats occasionally entered, but only two triggered the traps, and then escaped by digging at the floor until they broke a cable tie holding the outer bag in place. Interference accounted for the remaining 3% of visits when the wallaby did not enter the trap: bait was either removed by other animals during the night, or another swamp wallaby or bobuck inside the trap prevented access to it by physical obstruction or aggression towards the new visitor (Fig. 5b). Other non-target species visited the traps and consumed bait but usually did not trigger the traps: bush rats, Rattus fuscipes, European rabbits, Oryctolagus cuniculus, and satin bower birds, Ptilonorhynchus violaceus. Nonetheless, we largely prevented bush rats removing bait from traps by using bigger pieces of carrot so they could not be easily carried, and because satin bower birds are active only during the day we closed the door of traps that attracted particularly large numbers of these birds during the day.

Fig. 4.  Flow-chart showing outcomes of swamp wallaby visits to set traps (n = 340), Yarra Ranges National Park, July–September 2011.

Fig. 5.  Images recorded by camera traps showing interactions between swamp wallabies, Wallabia bicolor, bobucks, Trichosurus cunninghami, and traps. (a) A bobuck emerges from a triggered trap with a piece of carrot bait via the small gap in the wool bags that form the trap walls, preventing entry and potential capture of a swamp wallaby. (b) A bobuck inside a trap shows aggression towards a swamp wallaby, preventing entry and potential capture. (c) A swamp wallaby stands just inside the trap to reach carrot bait, without triggering the trap. (d) The rump and tail of a swamp wallaby blocks the door from closing properly, enabling escape. Yarra Ranges National Park, July–September 2011.

Swamp wallabies entered traps on 41% of visits, but were trapped in only 14% of visits. The unsuccessful captures following entry (27% of trap visits) could be attributed to three causes. Failure to trigger the trap on 21% of visits encompassed occasions when the wallaby stood just inside the trap and reached in to obtain bait without disturbing the trigger mechanism (14% of visits) (Fig. 5c), and times when the trigger did not release despite wallaby activity inside the trap (7%). Door failures (3% of visits) occurred when the hinges had rusted and did not swing freely to close the door, or when a wallaby triggered the trap when reaching inside but its tail or rump blocked the door from closing, so was able to escape before the lock bar could fall into place (Fig. 5d).

Swamp wallaby trap success in the Yarra Ranges National Park during 2011 was low (9.4%), although this is probably not unusual; Di Stefano (2007) reported a similar trap success of 11.4%. Swamp wallabies at our site visited traps most frequently during darkness, declining steadily by ~13% per hour after sunset. We checked traps 2–3 h after sunset; however, future programs could schedule trap checks and closures more closely aligned with the declining visit rate found in this study. Visits to traps were strongly influenced by lunar phase and presence of moonlight at this site, with the wallabies favouring the low-light conditions of new and crescent moon or when the moon had set. The influence of moonlight on foraging behaviour of small mammals is well documented (e.g. Daly et al. 1992; Kotler et al. 2010), but there has been limited study of the effects of moonlight on larger mammals. Kie et al. (1991) showed that mule deer, Odocoileus hemionus, altered feeding patterns, and Carter and Goldizen (2003) found increased vigilance of brush-tailed rock-wallabies, Petrogale penicillata, on bright moonlit nights. These behavioural responses to bright moonlight have been attributed to increased susceptibility to predation because the light enhanced the ability of predators to detect prey. In swamp wallabies, Osawa (1989) reported a trend towards fewer road-kills, suggesting reduced feeding activity on road verges, on full and gibbous moon nights. Our data are consistent with these studies and suggest that a fear of predators at this site reduced visitation to traps by swamp wallabies under bright moonlit conditions.

When swamp wallabies visited the traps, they still did not enter them on more than half of these visits. In most cases we were unable to determine why this occurred; however, in some cases the wallabies simply consumed the bait outside the trap, then left, or the door had already been closed, primarily by other swamp wallabies or bobucks. The crepuscular to nocturnal habits of bobucks were similar to those of our target species, so we were unable to deter bobucks from the traps without impacting on potential swamp wallaby captures. Had we not placed a few carrots outside the trap we suspect more entries and subsequent door closures by this non-target species would have occurred. By also ensuring that there were large quantities of bait inside the trap, there was greater likelihood that bait would still be present should a swamp wallaby visit the trap.

When swamp wallabies entered traps but did not trigger the door mechanism, the wallaby usually stood just inside the door and reached in to obtain bait, despite our placing bait in the rear corners of the trap. To combat this problem, three changes could be made to traps: (1) increase the length of the pit beneath the trap, (2) fasten bait to the rear of the trap, perhaps in a mesh pocket to avoid entanglement, or (3) increase the length of traps. The first two changes are easily implemented and should be thoroughly tested before considering the third option, which would make traps more difficult to construct and more cumbersome to operate.

The use of camera traps in this study allowed us to explore potential relationships between a set of environmental variables and the rate of visits to traps by our target species, the swamp wallaby. Once wallabies had visited the traps, the camera traps then allowed us to identify aspects of trap design, in addition to behavioural responses of both target and non-target species, which further influenced trap success. Although our primary goal was to identify the most influential environmental and intrinsic factors in order to improve the efficiency of our trapping program, our findings also provide insights for other researchers working with swamp wallabies. Future trapping programs should be scheduled with consideration for time of day, moon phase, and moon rise and set times. Some simple steps can also be taken to reduce the impact of non-target species, and modifications to the trap itself could be made to enhance capture rates. This study highlights the potential of using cameras on traps to better understand the environmental and intrinsic factors that can influence trap success for a wide range of species.