Impact of an enclosure rotation on the activity budgets of two captive giant pandas (Ailuropoda melanoleuca): An observational case study

Initially, the aim of this study was to establish a behavioural baseline for the giant pandas to assist zookeepers with the identification of maladaptive behaviours and oestrous-related behaviours. However, partway through the study, zoo staff implemented husbandry changes to rotate the giant pandas’ enclosures as a strategy to interrupt and reduce the female giant panda’s (Funi) performance of stereotypic somersaults and pacing. As this was such a significant shift, the aim of the project changed to evaluate the impact of the rotation on the giant pandas’ behaviour and enclosure use. Aside from the enclosure rotation, the animals’ routines and enclosures remained unchanged for the duration of the study.

The giant pandas in this study were Funi (female) and Wang Wang (male), born at the Wolong Giant Panda Research Centre in Sichuan Province, China. Funi was born on 23/08/2006 (dam: Long Xin; sire: Lu Lu), and Wang Wang on 31/08/2005 (dam: Mao Mao; sire: Lin Lin), and both were mother-reared. In 2008, because of the Wenchuan earthquake, Wang Wang and Funi were moved to the Baixiongpin Yaan Research Centre in China. From there they were re-located to the Adelaide Zoo in November 2009, where they had been on exhibit to zoo visitors since December 2009.

During the study, Funi and Wang Wang were on-exhibit to zoo visitors daily for six, 45-minute sessions with 15-minute scheduled breaks from visitors between sessions, and a 75-minute break daily at 2.30p.m. This viewing schedule was implemented by zoo management to allow the giant pandas regular, quiet time away from large visitor crowds, which often reached capacity due to the novelty of the exhibition (visitors pre-booked tickets for each session). During breaks, the entire ‘Bamboo Forest’ area (including giant panda & red panda enclosures) was closed to zoo visitors.

Funi and Wang Wang were housed separately, each with an indoor and outdoor enclosure. The glass-fronted indoor ‘day rooms’ (see Figure 1) were temperature-controlled (20^° - 22^°C) and were connected to an indoor lock-away area that was off exhibit and out of view to zoo visitors. The 600sqm outdoor areas featured a variety of terrains (e.g. hills, waterfalls, rocks, trees and grassy knolls) designed to encourage physical activity (see Figure 2). When on- exhibit to zoo visitors, Funi and Wang Wang had access to their outdoor areas and their day rooms. During short break times, they sometimes had access to their indoor lock-away areas, and they always had access to this area during the long break at 2.30p.m., for resting, feeding or training. On several occasions, the animals remained on exhibit for periods outside these hours for private tours or functions.

Figure 1.

Figure 2.

From the time of their arrival in November 2009 until the 7th of September in 2010, Funi occupied enclosure one and Wang-Wang occupied enclosure two (see Figure 2). On the 7th of September, zookeepers implemented an enclosure rotation, with Funi moved to Wang Wang’s enclosure, and vice versa. Zookeepers implemented the enclosure rotation in response to observations of stereotypic pacing and somersaults exhibited by Funi. A detailed description of these behaviours can be found in the results section.

When on-exhibit, zookeepers provided Funi and Wang Wang with six randomly scheduled scatter-feeds daily. Each giant panda had access to up to 20 kilograms of bamboo leaves and culm daily, of several varieties including Arrow, Black, Bambusa, Golden Painted, Golden Goddess, Oldhamii, Buddha and Moso. In addition, their daily diet was supplemented with panda cake (steamed crushed bamboo, flour, oil and eggs): approximately 1.6 kilograms for Funi and 2 kilograms for Wang Wang. Both animals were also fed up to 2 kilograms of fruit (apple, pear) and vegetables (sweet potato & for Wang Wang only - carrots). Part of their daily diet was offered to them throughout the day and part as a final night feed at approximately 4.30pm, when they were locked indoors overnight.

Intervals were measured by an automated stopwatch which sounded an alarm every 90 seconds. Behavioural categories were operationally defined according to an abbreviated ethogram adapted from Swaisgood (17). The ethogram consisted of 14 behavioural categories, containing 58 behaviours, each of which was assigned a unique code. During data analysis, similar behavioural categories were collapsed together, and behaviours that were not observed at all were excluded from analysis. In all, 15 behaviours were included in this study.

Data were collected using instantaneous time sampling (18), at 180-second intervals during three-hour blocks (10am – 1pm and 1.30pm – 4.30pm); as well as all-occurrences frequency counts (for two behaviours only, stereotypic somersaults and scent marking). Data collection occurred for a total of 180 hours over 30 randomly scheduled days during the period of July – September 2010 (total of 180 hours), which included 23 weekdays and seven weekend or school holiday days. Baseline data was collected for 22 days prior to enclosure rotation that occurred on September 7th. Following this, post intervention data was collected for eight days.

At each time sample, the focal giant panda’s behaviour and location in the enclosure was recorded on the same data sheet, with areas coded according to the enclosure map (see Figure 2). Each giant panda’s behaviour and location was recorded on alternating, 90-second intervals. Additionally, two behaviours; stereotypic somersaults (Funi only) and scent marking were recorded as all-occurrences frequency counts, meaning that they were recorded each time they were observed regardless if this coincided with a time sample or not. These behaviours were recorded (and reported) as frequency counts because they occurred very infrequently, yet are considered to be behaviours of significance for husbandry purposes (e.g. as possible indicators of stress – somersaults; or oestrous/mating – scent marking).

When off-exhibit, the giant pandas were observed on a large-screen monitor via security cameras, which could be controlled from a panel located at a desk area in the public-viewing area in front of the indoor day rooms (with access only given to zoo staff members or researchers).

Inter-rater reliability tests were conducted to assess the extent to which the observation protocol was susceptible to observer bias (e.g. 19, 20). Two independent observers completed a one-hour training session with the primary researcher (J.R.), before collecting data for a 90-minute observation session (60 scans each). The primary researcher and both independent observers agreed on behaviour classification and location in enclosure for 95.5% of scans.

The giant pandas’ activity budgets are presented in Table 2, with the exception of stereotypical somersaults and scent marketing, as these were recorded as frequency counts. Both giant pandas engaged in a variety of different behaviours, and spent different proportions of time engaged in these behaviours. Funi spent just over 45% of her time resting, and this remained largely unchanged after the enclosure rotation. Wang Wang also spent a large proportion of time resting (about 33 – 34% of his time), and he spent an almost equivalent amount of time feeding, which remained largely unchanged after the enclosure rotation. Funi spent about 20% of her time feeding, which decreased slightly (by about 3%) after the enclosure rotation. She spent about 13% less time feeding than Wang Wang.

Following the enclosure rotation, Wang Wang exhibited an increase in performance of scent-marking, when this behaviour was observed four times on the (first day) and three times on the (second day). The increase in scent marking on those two days coincided with the first observations of ‘mock charging’, which is most similar to a charge or lunge according to the ethogram (17). This behaviour was observed on two occasions, once directed at a zookeeper on the other side of a glass wall, and once directed at an enrichment item – a large plastic ball hanging from a tree by a chain. Wang Wang had previously ripped the ball enrichment item off the tree, thereby dislodging a piece of rubber tube from around the chain, which he initially played with before lunging at it. Coinciding with the enclosure rotation was the turning on of the outdoor waterfalls by keepers, which resulted in Wang Wang playing in the water, ‘running’ laps of the outdoor enclosure, and then mock charging. He appeared to be very excitable because of the enclosure rotation.

Some changes in the giant pandas’ activity budgets were observed following the enclosure rotation (see table 2 for full descriptions). Funi demonstrated a notable increase in climbing following the rotation, from < 2% to more than 5%. The reactionary behaviour category (i.e. door-directed and keeper-directed behaviours) was higher for Wang Wang than Funi (< 4%), with about 7% of his time spent engaged in this behaviour before the rotation, and a slight decrease (about 2%) observed after the rotation. Hypothesis 1 was supported, since active behaviours increased for both giant pandas after the enclosure rotation. The increase was most noticeable for Funi (from 11 – 21%) than for Wang Wang (16 – 19%).

The Spread of Participation Index (SPI) was calculated to indicate the giant pandas’ relative use of the different areas of their enclosure (21). SPI scores range from 0.0 to 1.0, with values lower than 0.40 indicating that the animal uses the different areas of their enclosure equally, whilst values between 0.40 and 0.70 indicate that the animal uses the different areas of their enclosure unequally, and values greater than 0.70 indicate a limited use of available enclosure areas. Prior to the enclosure rotation, Funi’s baseline SPI score was 0.57, and this decreased slightly after the enclosure rotation to 0.52. These SPI values indicate that Funi did not utilise all areas of her enclosure equally both at baseline and after the enclosure rotation. Wang Wang’s baseline SPI score was 0.45 and this decreased to 0.31 after the enclosure rotation when Wang Wang used areas of his enclosure more equally. Hypothesis 2 was supported, since the SPI scores decreased following enclosure rotation. However, Funi continued to use some areas of the enclosure more frequently. The enclosure rotation resulted in an SPI score for Wang Wang that was ideal in terms of making use of all areas of his enclosure.

Note that in this section, stereotypic pacing is differentiated from quasi-stereotypic pacing, which refers to pacing with some variation and is considered to be less intense. As shown in Figure 3, Funi demonstrated a marked decrease in stereotypic pacing following the enclosure rotation. Wang Wang’s performance of this behaviour remained relatively stable across the study period, comprising around 5 – 7% of his activity budget. Thus, Hypothesis 3 was supported for Funi but not for Wang Wang.

Figure 3.

This hypothesis was only relevant for Funi, since no idiosyncratic aberrant behaviours were observed for Wang Wang. On day 17 of data collection (6 days prior to the enclosure rotation), Funi was observed to engage in stereotypic somersaults. Figure 4 shows the sequence of this behaviour (or topography of the response). Funi incorporated the somersaults into a stereotypic pacing routine in the indoor enclosure. On several occasions, this behavioural repertoire of combined pacing and somersaulting behaviour became more intense. The pacing became faster and Funi scrambled up the single faux tree in the enclosure before falling out of the tree, and then running towards the metal gate next to the door to lock-away area and slamming her weight against the door as she somersaulted, which created a loud noise. Thus, it was a louder, more intense (in terms of speed) version of the behaviour, and more odd since it included climbing a tree and falling out of it. The occurrence of this behaviour, as well as Funi’s performance of pacing, formed part of the justification for zookeepers to execute the enclosure rotation. The frequency of somersaults was much higher prior to the enclosure rotation (average of 46 per day) compared to after the enclosure rotation (average of 11 per day) (see figure 5). Note that no somersaults occurred prior to day 17, nor after day 29. Therefore, Hypothesis 4 was supported.

Figure 4.

Figure 5.

Both Funi and Wang Wang engaged in stereotypic pacing throughout this study, (between 4 and 11% of activity budget) at levels comparable to those for other giant pandas in previous research (3 – 6). Funi also performed an unusual form of highly rigid, and physically rough stereotypic somersaults. Whilst the aetiology of this behaviour is unknown, a possible explanation was that it was a conditioned behaviour in response to attention from keepers. The evidence for this was that Funi directed this behaviour at a Perspex slide that was adjacent to the keeper entrance to the giant panda building. Keepers occasionally interacted (hand feeding, playful scratching, calling name) with Funi at this location, which may have served as a potential reinforcer for this behaviour in the initial stages of its development. That is, it may be akin to the ‘superstitious’ behaviour in laboratory animals, where a response has been accidentally paired with food (22, 23), but in this instance may have been paired with food and/or attention and praise.

Somersaults have been observed previously in captive giant pandas, but it is the context in which it occurs that is likely to determine whether it can be considered a behavioural indicator of stress or not. For example, Liu et al. (4) observed rolling and somersaulting as an integral part of play with objects. The behaviour observed in Funi is most similar to that reported for many of the adult giant pandas at the Chengdu Giant Panda Breeding Centre in China, where somersaulting occurs within pacing bouts, particularly when confined to indoor areas (6). Although, not physically confined to the indoor area, Funi, preferred to ‘confine’ herself in this area, perhaps because of visitor noise or other aversive stimuli present outdoors.

Zookeepers implemented the enclosure rotation with the primary aim of reducing Funi’s performance of stereotypic somersaults and pacing. It was a significant event for Funi and Wang Wang, as it was their first opportunity to experience not only the different enclosure, but also each other’s scents up-close for the first time. For Funi, the enclosure rotation resulted in a marked decrease in the performance of stereotypic pacing (Hypothesis 3 supported), cessation of performance of stereotypic somersaults (Hypothesis 4 supported), and a noticeable increase in active behaviours, particularly climbing (Hypothesis 1 supported). For Wang Wang, the enclosure rotation resulted in a marked increase in sexual communication behaviours, scent marking and mock charging. For Wang Wang, there was a small increase in active behaviours (Hypothesis 1 supported), but no noticeable change in stereotypic pacing/ quasi-stereotypic pacing (Hypothesis 3 not supported). The increased activity and opportunities for exploration highlight the potential value of enclosure rotations as a form of enrichment, which has previously been described in relation to other species (e.g. gorillas (16)). The observed increase in scent marking after the enclosure rotation for Wang Wang is likely to represent territorial behaviour, as male giant pandas use scent marking to mark out new territory and in response to the scent of another giant panda (24).

Previous studies indicate that enrichment practices such as increasing the difficulty of accessing food or the inclusion of toys in the enclosure may not be sufficient in reducing stereotypy in giant pandas if the behaviour has developed over a long period of time (7). Indeed, traditional enrichment practices (e.g. the provision of sawdust) proved ineffective in disrupting Funi’s performance of stereotypic somersaults. The enclosure rotation was very effective in reducing Funi’s performance of stereotypic pacing and idiosyncratic aberrant somersaults. Additionally, there was an increase in Wang Wang’s behavioural diversity following the enclosure rotation (e.g. novel performance of mock charging). However, in terms of the SPI or use of all areas of the enclosure, the enclosure rotation resulted in Wang Wang using all parts of the enclosure (Hypothesis 2 supported), whereas, Funi continued to spend her time less equally in all parts of her enclosure (Hypothesis 2 not supported). The enclosure rotation represented a significant change to the giant pandas’ routine and environment, which had positive effects on both Funi and Wang Wang’s behaviour. It is not known whether these effects would persist through subsequent enclosure rotations, although this would be a valuable investigation for future studies to explore.

This case study evaluated the impact of an enclosure rotation as a form of enrichment on the behaviour of two giant pandas at the Adelaide Zoo in South Australia. Results indicate that the enclosure rotation was successful as it reduced Funi’s performance of stereotypy (pacing and somersaults), and increased Wang Wang’s behavioural diversity and activity levels. Stereotypy is a behavioural indicator of stress, and if chronic, may be indicative of reduced welfare. Rather than being deterred by small sample sizes available for conducting empirical observational studies to test efficacy of enrichment practices in zoos, researchers should be encouraged to publish case studies, and disseminate their findings widely. This may help expand the body of evidence available concerning how different species and individual animals respond to different types of environmental enrichment. The availability of this type of information is a key determinant of zoos’ ability to rely on evidence-based husbandry techniques.