ANIMAL BEHAVIOUR, 2001, 61, 447–457

doi:10.1006/anbe.2000.1610, available online at http://www.idealibrary.com on

A quantitative assessment of the efficacy of an environmental

enrichment programme for giant pandas

RONALD R. SWAISGOOD*, ANGELA M. WHITE*, XIAOPING ZHOU†, HEMIN ZHANG†, GUIQUAN ZHANG†,
RONGPING WEI†, VALERIE J. HARE*, ERIN M. TEPPER* & DONALD G. LINDBURG*
*Center for Reproduction of Endangered Species, Zoological Society of San Diego
†China Research and Conservation Center for the Giant Panda, Wolong Nature Reserve, Sichuan

(Received 7 February 2000; initial acceptance 10 May 2000;

final acceptance 14 August 2000; MS. number: A8709R)

With the goal of enhancing psychological well-being, we developed an environmental enrichment

programme for eight subadult and six adult giant pandas, Ailuropoda melanoleuca, at the Wolong Breeding
Center in Sichuan, China. We used five different enrichment items, each selected because of the different
behavioural opportunities they promoted: (1) manipulable plastic objects; (2) a burlap sack full of straw;
(3) spruce branches; (4) a fruitcicle (apples frozen in a large ice block); and (5) a puzzle feeder. Each subject
received each of the five items on three occasions, for a total of 15 enrichment sessions over the course
of the 2.5-month study period. Each enrichment session was paired with a control session in which no
enrichment was present. Pandas spent significantly more time active, and displayed a greater variety of
object- and nonobject-directed behaviours when enrichment was present. We also found a statistically
significant reduction in the rate of and percentage of time engaged in the performance of stereotypic
behaviour and behaviours indicative of feeding anticipation. This effect was maintained even when
pandas were not interacting directly with an enrichment item, suggesting that enrichment had a
motivational carry-over effect in the aftermath of interactions with enrichment items. These results are
consistent with the ethological needs model of motivation in that opportunities to perform more natural
behaviour appeared to improve motivational indices of well-being. The pandas remained responsive to
enrichment across the 15 trials, providing no evidence for habituation or cross-habituation. Age proved
to be an important determinant of responsiveness; for example, adults, but not subadults, displayed a
significant preference for feeding enrichment. Future studies will assess whether these promising initial
results are maintained throughout the lives of captive giant pandas.
 2001 The Association for the Study of Animal Behaviour

The past decade has witnessed a remarkable rise in the
importance and application of the concept of environ-
mental enrichment, both in the zoo community and
among those striving to improve welfare in inten-
sively housed species used for human consumption and
biomedical research (Shepherdson et al. 1998). Environ-
mental enrichment is aptly defined as ‘an animal
husbandry principle that seeks to enhance the quality of
captive animal care by identifying and providing
environmental stimuli necessary for optimal psychologi-
cal and physiological well-being’ (Shepherdson 1998,
page 1). Enriched environments can promote a greater
diversity of the species-typical behavioural repertoire
and a concomitant reduction in stereotypic and other
Correspondence: R. R. Swaisgood, Center for Reproduction of
Endangered Species, Zoological Society of San Diego, P.O. Box 551,
San Diego, CA 92112, U.S.A. (email: rswaisgood@sandiegozoo.org).
X. Zhou, H. Zhang, G. Zhang and R. Wei are at the Wolong Nature
Reserve, Wenchuan, Sichuan, People’s Republic of China, 623006.
0003–3472/01/020447+11 $35.00/0
abnormal behaviours (Shepherdson et al. 1998). In
addition, several studies have demonstrated a variety of
positive effects of enrichment, including enhanced brain
function and learning ability, and reduced emotional
reactivity and indices of stress (Carlstead & Shepherdson
1994). Although many of these questions cannot be
addressed in research with endangered species (e.g.
enrichment effects on neurological function), the
weight of evidence from these and other studies suggests
a clear role for enrichment in maintaining endangered
species in captivity, promising to increase successful
mating and rearing of offspring, and promote the devel-
opment of more behaviourally competent candidates for
reintroduction to the wild.
Captive environments only rarely approximate the
variety of stimuli and environmental contingencies
present in nature. Captive environments in compari-
son with the wild are unchanging (lacking novelty),
spatially limited, stimulus-poor (lacking complexity), and
447  2001 The Association for the Study of Animal Behaviour
448 ANIMAL BEHAVIOUR, 61, 2
generally provide the inhabitant with little control over
its environment (Carlstead 1996; Poole 1998). The latter
lack of behavioural contingency is especially evident in
feeding practices wherein the animal is delivered its food
on a predictable schedule, and needs not perform any
behaviour to garner access to this resource. Captive diets
are also usually highly concentrated, processed foods that
require little effort to consume (e.g. Lindburg 1998). The
captive animal, therefore, is often left with a great deal of
time, but few behavioural opportunities to occupy that
time. One undesirable consequence of an unknown sub-
set of these factors is the development of stereotypic
behaviours, which seem to take the place of more
‘natural’ behaviours seen among free-living animals
(Hughes & Duncan 1988; Mason 1991b; Rushen et al.
1993). Stereotypies are repetitive behaviours, invariant in
form, and have no apparent functional consequences
(Mason 1991b). The expression of stereotypies in captive
animals is generally believed to reflect poor well-
being, although it still remains unclear exactly what
stereotypy performance tells us about the animal’s
subjective experience (Mason 1991a, b).
Researchers at zoological institutions have been
addressing the issue of environmental enrichment with
an increasingly powerful conceptual arsenal (e.g.
Hutchins et al. 1984; Carlstead & Shepherdson 1994;
Carlstead 1998; Forthman 1998; Poole 1998), but,
unfortunately, the sophistication of these ideas has out-
paced improvement in the application of empirical tests
of the efficacy of enrichment programmes and their
accompanying guiding conceptual framework. While
major inroads have been made with studies of intensively
housed species, zoo animal studies have been severely
compromised by small sample size. For example, we
searched the journal Zoo Biology for articles published
during the 1990s pertaining to environmental enrich-
ment at zoological institutions, and found that the
median sample size was four animals, with a maximum
of 11.
Other deficiencies in studies of enrichment are not
necessarily limited to zoo species. For example, a rela-
tively unexplored aspect of enrichment is the assessment
of differential responsiveness to specific items as a func-
tion of the individual’s age, and subsequent efforts to
tailor enrichment programmes to specific age groups
(Mench 1998). Perhaps most important is the need to test
whether enrichment effects behavioural change solely by
occupying the animal’s time, leaving less time available
to engage in abnormal or stereotypic behaviour, or
whether it has a long-lasting motivational effect that
carries over into the aftermath of an interaction with
enrichment. This idea is most effectively stated in the
‘ethological needs’ model of motivation advanced by
Hughes & Duncan (1988) and by its sister hypothesis, the
information primacy model (Inglis et al. 1997). Briefly,
while the types of behaviours addressed and evolutionary
explanations differ, both of these hypotheses advance the
notion that animals have certain basic needs associated
with the appetitive phase of goal-directed activity.
Thus, animals are reinforced not just by the functional
consequences of their behaviour (e.g. food and mate
acquisition), but also by the behaviours that are
employed to attain these endpoints (e.g. search, explore,
dig, extract, and so forth) or the information gathered
through exploration. These models adequately explain
findings such as pigs wasting time and energy ‘rooting’
tokens prior to caching them in for a primary reinforcer
(i.e. food; Breland & Breland 1961) and a preference by
starlings to search for food hidden in holes under plastic
flaps rather than simply consume food made freely avail-
able in a dish (Inglis & Ferguson 1986). Hughes & Duncan
go on to suggest that even if animals are well fed and
comfortably housed, they may suffer from a lack of
opportunity to express these appetitive behaviours.
Because it is the performance of these behaviours itself
that is reinforcing, the inability to engage in these behav-
iours may compromise the animal’s well-being, resulting
in the performance of stereotypic behaviour in their
stead.
With these ideas in mind, we developed an enrichment
programme for giant pandas, Ailuropoda melanoeuca,
designed to (1) promote natural species-typical behav-
iours and reduce abnormal or stereotypic behaviours and
(2) allow us to test some of the hypotheses pertaining to
psychological well-being and enrichment. The giant
panda is a dietary specialist, feeding almost exclusively
on bamboo (Schaller et al. 1985). Because the panda is
not particularly efficient at digesting this food source, it
must consume approximately 14 kg/day to meet its daily
energetic requirements. As a consequence, the panda
spends on average 55% of its time consuming bamboo.
Almost all the remaining time is spent resting, with other
activities occupying only 4% of its time. Because captive
pandas’ diet of bamboo is supplemented with more con-
centrated food, such as bread and milk, pandas spend less
time feeding (Mainka & Zhang 1994). As a dietary special-
ist with a relatively simple ecology (solitary, no complex
predator–prey interactions), many would predict that its
ethological needs are relatively simple and therefore are
less likely to develop stereotypic behaviours in captivity
(Kreger et al. 1998; Mench 1998). Yet many pandas
in captivity display a variety of stereotypies (R. R.
Swaisgood, unpublished data), suggesting that their needs
are not being met in their present circumstances. Thus,
an enrichment programme for pandas needs to provide
alternative ways for pandas to spend the time that they
would normally spend eating bamboo in the wild. A
common-sense solution might be to simply feed the
pandas only bamboo, but this is not a practical solution
because we cannot be certain that the bamboo we select
will meet all their nutritional needs.
In the experiment described here, we tested the effects
of five different enrichment items, each presented to the
animal on three separate occasions. The items provided
included three objects (a burlap sack filled with straw,
moveable plastic objects and spruce branches) and two
feeding enrichment items (fruit frozen in an ice block
and a puzzle feeder). All of these objects, both feeding and
nonfeeding, were selected for their varying properties
and consequent abilities to promote different sorts of
behaviours. This enrichment programme is supplemental
to other ongoing efforts to enrich the lives of the pandas
 SWAISGOOD ET AL.: ENRICHMENT FOR PANDAS 449

Table 1. Description of enrichment items used in the experiment

Enrichment item Description


Boomer Ball A hard plastic ball 25 cm in diameter.
Boomer Bobbin A hard plastic object 30 cm in height and shaped like a bobbin from a sewing
machine: wide and flat on the ends, tapering to a more narrow cylinder in the middle.
Burlap sack A large burlap sack (0.9×0.5 m) stuffed with rice straw.
Branches A pile of eight fresh-cut spruce branches, each approximately 1 m in length.
Fruitcicle A large block of ice approximately 20×30×10 cm with pieces of apple frozen on the
surface.
Puzzle feeder A large, hard segment of bamboo approximately 12 cm in diameter and 30–40 cm in
length. Both ends are naturally sealed, and a 5-cm2 hole is cut midway between the
endpoints and pieces of apple are placed inside.

at the Wolong Breeding Center. This experiment was
conducted in the spirit of a pilot study to assist in the
development of a long-term enrichment programme as a
standard husbandry practice at the this facility.
METHODS
Subjects and Study Site
Our study population consisted of 14 giant pandas:
three adult males, three adult females, three subadult
females and five subadult males. Subadults ranged from
1.5 to 4.5 years of age, with reproductive maturity occur-
ring at 5.5 years of age (Schaller et al. 1985). During the
course of this study, all pandas were housed in enclosures
consisting of an outdoor portion measuring approxi-
mately 98 m, and a 36 m indoor bedroom. Pandas
are able to interact through cage bars with neighbouring
individuals in adjoining enclosures, but opportunities for
physical contact is minimal. For details of housing and
diet, refer to Swaisgood et al. (1999). None of the
subjects had had any previous exposure to an enrichment
programme.
Experimental Design and Procedures
Table 1 describes each of the six enrichment objects,
selected for the diverse array of sensory properties and
behavioural opportunities that they afford (e.g. move-
ability, manipulability, malleability, consumability, etc.).
The Boomer Ball and Bobbins were presented simul-
taneously and analysed as a single enrichment treatment.
The remaining four items were provided separately, for a
total of five different enrichment treatments over the
2.5-month study period. All objects were novel to the
subjects at the onset of the study. Each subject was given
each of the five treatments on three different occasions
for a total of 15 enrichment trials. We chose this design so
that we could assess the relative efficacy of several differ-
ent enrichment items, ensure robust results, and provide
some opportunity for habituation to specific items, as
well as cross-habituation to multiple exposures to an
array of items. By ‘habituation’ we mean any evidence of
decreased responsiveness to enrichment items resulting
from repeated presentations.
Observations commenced after the item was placed
in the enclosure and continued for 45 min. Five time
periods corresponding to peak activity levels (R. R.
Swaisgood, unpublished data) were selected for the study:
0700–0745, 0815–0900, 0900–0945, 1400–1445 and
1445–1500 hours. We employed a balanced design such
that an equal number of sessions was obtained for each
enrichment item during each time period. All three trials
for an individual for a particular enrichment item were
conducted during the same 45-min time period to mini-
mize variability associated with a particular time of day
and maximize our ability to detect any habituation effects
across trials. To give the pandas more opportunities to
interact with the enrichment items and encourage any
possible habituation, enrichment items were left in the
pandas’ enclosures until approximately 1700 hours.
Approximately 24–48 h prior to the enrichment presen-
tation, we observed the panda for a 45-min control
session with no enrichment present. Each control session
occurred at the same time of day as the enrichment trial
with which it was paired. Thus, each enrichment trial was
paired with its own control trial 1–2 days prior to the
enrichment trial to control for any temporal changes in
motivation across the study period, and at the same
time of day as the enrichment trial to control for any
time-of-day effect.
Dependent Variables
We set a timer to emit an audible signal at 1-min
intervals, and used instantaneous point sampling to esti-
mate the percentage of time spent in various activities.
We recorded one of the following mutually exclusive
and exhaustive behaviours at the point sample. (1)
Stereotypic: any of the invariant, highly repetitive
behaviours observed previously in giant pandas (R. R.
Swaisgood, unpublished data). These include head toss,
pirouette, cage climb, tongue flick, pace, self-bite, paw
suck and body sway. (2) Feeding Anticipation: operation-
ally defined as being alert, within one body length of the
area where food is commonly delivered, and oriented in
the direction from which food is brought. (3) Object-
Directed: in contact with and actively interacting with an
object. Further divided into three subcategories: (a)
Object Play/Manipulate: panda engages in any of the
450 ANIMAL BEHAVIOUR, 61, 2
one–zero activities listed under this category below; (b)
Feed: panda consumes any food item derived from an
enrichment item; and (c) Object Investigation: panda
sniffs or licks object. (4) Several other behaviours (e.g.
feed, locomote, rest) not relevant to analyses performed
for this paper. (5) Active: panda engaged in some active
behaviour, that is, not sleeping, standing, or other-
wise resting (not mutually exclusive of the behaviours
identified above). We also recorded all-occurrences of
Stereotypic behaviour (except pacing) and Feeding
Anticipation, as defined above. Separate bouts of
behaviour were separated by more than 5 s.
Because of the vigour and complexity of interactions
with enrichment items, we recorded the occurrence of
specific behaviours directed to the enrichment item dur-
ing 1-min intervals using one–zero sampling. Those
behaviours subsumed under the general category Object
Play/Manipulation include the following. (1) Paw: panda
pushes, swats or rolls item on the ground using its
forepaws. (2) Manipulate: panda picks up, holds and
rotates object in forepaws and/or hindpaws. (3) Bite:
panda bites, mouths or chews object. (4) Roll: panda rolls
side-to-side or somersaults on top of or while holding
object in mouth or paws. (5) Head-Push: panda pushes
object with head, causing it to move. (6) Pursue: panda
causes object to move and follows it for at least 2 m. (7)
Shake: panda holds object in mouth and shakes head
vigorously from side to side. (8) Pounce: panda rears up
and jumps forcefully on top of object. (9) Carry: panda
holds object in mouth or paws and carries or drags object
at least 2 m. We also recorded two behaviours under the
general category of Object Investigation using one–zero
sampling. (10) Investigate: pandas places nose within
5 cm of object and sniffs or licks the object. (11) Flehmen:
while Investigating an object, panda raises upper lip,
baring the teeth, and inhales deeply. A final category of
object interaction not included in the above categories is
(12) Scent Anoint: panda picks up object and rubs it onto
its pelage, often accompanied by movements reminiscent
of a cat washing its face.
We used the above one–zero behaviours, together with
several other behaviours from our standard ethogram, to
calculate a Behavioural Diversity index. Each time a
particular behaviour was observed at least once during an
observation session, a score of ‘1’ was recorded for that
behaviour. In addition to the 12 one–zero behaviours
defined above, we recorded the following 20 behaviours:
five elements of locomotor play (playful run, jump,
somersault, headstand and roll without an object), loco-
mote, feed on bamboo, climb, water splash, water roll,
social play, nonplay social interaction, scent-mark, body
rub, and chemosensory investigation of six different sub-
strates (concrete floor, dirt/grass, wall, cage bars, wood
and large stone/vertical concrete structure). Thus, an
individual could receive a maximum score of ‘32’ for
an observation session. Other behaviours, such as
Stereotypic behaviours and Feeding Anticipation, were
omitted from the Behavioural Diversity index because
they are generally believed to indicate poor well-being
(Mason 1991a, b; Shepherdson et al. 1998). Behaviours
included in the Behavioural Diversity index were chosen
to reflect an array of common ‘normal’ behaviours that
could be construed as an index of psychological well-
being. Although such an index is not a proven manifes-
tation of well-being (indeed, there are no undisputed
indices), environmental exploration, play and the expres-
sion of species-typical behaviours are at least correlated
with other measures of well-being (Chamove & Anderson
1989; Broom & Johnson 1993; Poole 1998); therefore, the
question of causation aside, this measure may bear some
relationship with well-being. In addition, if animals have
the need to express some specific behaviour patterns,
then the performance of a greater variety of behaviours
will increase the probability that one or more important
behaviours are expressed (see Discussion).
Data Analysis
Residual plots were inspected for deviations from
normality and the homogeneity assumption was tested
with Hartley’s Fmax test (Sokal & Rohlf 1998). Where
necessary, we arcsine-transformed data based on percent-
ages and log-transformed those based on rates. If, follow-
ing transformation, the data still did not meet the
assumptions of parametric tests, we employed a non-
parametric counterpart. Parametric tests were comprised
of a repeated measures factorial analysis of variance
(ANOVA) with two within-subjects (experimental con-
dition, trial) and two between-subjects (sex, age class)
factors. We excluded control trials from the analysis and
employed the same statistical model (with enrichment
item replacing experimental condition) to examine the
effects of different enrichment items. Type III ANOVA
was used because it provides a conservative adjustment
for unequal sample sizes.
RESULTS
Description of Responses to Enrichment
The pandas almost invariably approached the enrich-
ment item as soon as they became aware of its presence. A
typical interaction began with an investigatory phase in
which the subject, sometimes displaying signs of caution,
approached and sniffed and gently touched the object
with the forepaw. A manipulatory phase generally fol-
lowed the initial investigation, in which the subject
picked up the object and rotated the object repeatedly in
its forepaws, examining it closely both olfactorily and
tactilely. Almost invariably the panda rolled on to its back
and manipulated the object with all four paws simul-
taneously. This slow methodical examination typically
gave way to more vigorous play, particularly among
subadults, including behaviours such as Bite, Roll, Shake,
Pursue, and Pounce. Although different enrichment items
promoted different kinds of behaviours, pandas did not
show any overt preferences for one enrichment item over
any other, as evidenced by our failure to obtain any
significant effects of enrichment item on the percentage
of time spent interacting with enrichment (data not
shown; but see adult/subadult comparisons below).

Percentage of time interacting with enrichment varied
from about 28% with the Ball/Bobbin to 36% with the
Fruitcicle. These analyses also revealed that all five
enrichment items yielded similar levels of ‘improvement’
among all of our behavioural measures (e.g. stereotypy
performance).
Effects of Enrichment on Behaviour: Enrichment
versus Control Trials
Our findings demonstrate that enrichment altered sev-
eral aspects of the pandas’ behaviour in that the presence
of enrichment produced significant main effects for all of
the behaviours measured. Pandas spent a significantly
greater proportion of their time Active during enrichment
than control trials (F1,395 =25.7, P<0.0001; Fig. 1a).
Pandas also displayed greater Behavioural Diversity when
enrichment was present (F1,395 =406.3, P<0.0001; Fig. 1b).
However, Object-Directed behaviour was only rarely
observed during control trials (XSE=0.30.07 behav-
iours for subadults and zero for adults), probably because
there were few objects that were moveable in the
enclosures. Pandas were occasionally seen to play with
bamboo, bread, porridge pans and loose cement blocks.
Since object play behaviours were difficult to perform
during control trials, we eliminated object play from our
diversity index and reanalysed the occurrence of the
remaining behavioural variables to determine whether
the presence of enrichment affected the expression of
behaviours not immediately directed towards enrich-
ment items. The resulting analysis still yielded signifi-
cant results, with pandas displaying a mean of 5.5
0.2 nonobject-directed behaviours during control trials
and 7.80.3 during enrichment trials (F1,395 =59.5,
P<0.0001).
In addition to enhancing the occurrence of ‘desirable’
behaviours, the enrichment programme was also associ-
ated with a significant reduction in ‘undesirable’ behav-
iours. For example, Feeding Anticipation occurred
significantly less frequently in the presence of enrich-
ment than when enrichment was absent (F1,395 =21.7,
P<0.0001; Fig. 1c). This same effect was also evident for
the percentage of time engaged in Feeding Anticipation
(F1,395 =31.6, P<0.0001; data not shown). In addition,
pandas displayed Stereotypic behaviour less frequently
(F1,395 =7.6, P=0.006; Fig. 1e) and spent a smaller pro-
portion of time engaged in Stereotypic behaviour
(F1,395 =17.9, P<0.0001; Fig. 1g) when enrichment was
present. Stereotypic behaviours that contributed to rate
estimates were predominantly composed of head toss,
pirouette, cage climb and tongue flick, whereas those
contributing to the percentage of time estimates primar-
ily consisted of pace, self-bite, paw suck and cage climb.
To test whether enrichment produced a motivational
carry-over effect in the aftermath of enrichment inter-
actions during an enrichment trial, we calculated esti-
mates of Stereotypic behaviour when the panda was not
interacting with the enrichment item. Analysis of these
data revealed that even when pandas were not
enrichment-directed, they displayed a significant reduc-
tion in both the rate (F1,395 =3.9, P=0.04; Fig. 1f) and
SWAISGOOD ET AL.: ENRICHMENT FOR PANDAS 451
percentage of time spent in Stereotypic behaviour
(F1,395 =12.8, P=0.0004; Fig. 1h). Similarly, Feeding
Anticipation was reduced even when pandas were not
interacting directly with the enrichment item (F1,395 =8.2,
P=0.004; Fig. 1d).
Age and Sex Differences in Responsiveness to
Enrichment
Although there were pronounced and highly signifi-
cant main effects of age on most of the behavioural
measures reported here (see Fig. 1a–h), these results are
beyond the scope of this paper. Sex effects were minimal,
but likewise are unrelated to the topic at hand. Of interest
for this paper are how sex and age interact with exper-
imental condition (enrichment versus control). Sex did
not interact significantly with experimental condition
(F1,395 <2.5, NS), strongly suggesting that males and
females do not differ with regard to responsiveness to
enrichment. In contrast, age of the subject interacted
significantly with experimental condition for two
measures. The presence of enrichment appeared to bring
about a more dramatic reduction in Feeding Anticipation
among subadults than adults (experimental con-
dition*age class statistical interaction: F1,395 =4.5, P=0.04;
Fig. 1c), and was also more effective at promoting higher
activity levels among subadults (F1,395 =14.2, P=0.0002;
Fig. 1a). We also examined the percentage of time spent
interacting with enrichment (including Feeding) with
regard to feeding versus nonfeeding enrichment. The
results of this test revealed a significant interaction
between age class and enrichment type (F1,55 =5.1,
P=0.03). Separate analyses for each age class showed that
adults (22.24.4% time spent with feeding versus 9.7
1.5% for nonfeeding enrichment; F1,26 =6.3, P=0.02), but
not subadults (40.74.0% versus 44.24.2%; F1,38 =0.1,
P=0.79), displayed a significant preference for feeding
enrichment. However, these differences did not arise
from a tendency of adults to spend more time Feeding per
se: adults spent a mean of 9.7% of their time Feeding, in
comparison with 12.9% for subadults.
Evidence for Habituation across Trials
We were unable to detect any evidence that the pandas
habituated to these enrichment items across the 15 trials,
despite the fact that all items were left in the enclosures
until the end of the day. Although trial produced a
significant main effect for two behaviours, the pattern of
change was inconsistent with habituation. Indeed, the
percentage of time spent in Stereotypic behaviour actu-
ally decreased significantly across trials (F1,154 =3.7,
P=0.03: 4.51.4, 2.10.9 and 1.80.7% for trials 1, 2
and 3, respectively). Conversely, the percentage of time
spent in Object-Directed behaviour with enrichment
items was highest in trial 3 (F1,154 =3.7, P=0.03:
29.92.9, 26.52.9 and 35.93.4%). Behavioural
Diversity also increased by the end of the study, albeit not
quite significantly (F1,154 =2.9, P=0.06: 11.70.6, 11.2
0.6, and 12.90.7). Trial effects did not approach signifi-
cance for any other behavioural variable (F1,154 <2.3, NS).
452 ANIMAL BEHAVIOUR, 61, 2

100 18
90 (a) 16 (b)

Behavioural diversity
80 14
70
%Time active

12
60
10
50
8
40
30 6
20 4
10 2
0 0
Adult Subadult Adult Subadult
7 7
Feeding anticipation bouts/h

Feeding anticipation bouts/h

(c) (d)
6 6

5 5

4 4

3 3

2 2

1 1

0 0
Adult Subadult Adult Subadult
30 30
(e) (f)
25 25
Stereotypic bouts/h

Stereotypic bouts/h

20 20

15 15

10 10

5 5

0 0
Adult Subadult Adult Subadult
16 16
(g) (h)
14 14
%Time stereotypic

%Time stereotypic

12 12
10 10
8 8
6 6
4 4
2 2
0 0
Adult Subadult Adult Subadult
Figure 1. Behavioural effects of enrichment as a function of panda age class. (a) Percentage of time Active; (b) Behavioural Diversity; (c) rate
of Feeding Anticipation (bouts/h); (d) rate of Feeding Anticipation when not interacting with enrichment item; (e) rate of Stereotypic
behaviour; (f) rate of Stereotypic behaviour when not interacting with enrichment item; (g) percentage of time performing Stereotypic
behaviour; (h) percentage of time performing Stereotypic behaviour when not interacting with enrichment. Values are X+SE. : Control;
: enrichment.

DISCUSSION met, at least during the admittedly brief 2.5-month

period during which this study was conducted. When
This investigation of our first enrichment efforts with the enrichment was present, pandas spent a greater pro-
giant pandas at the Wolong Breeding Center achieved a portion of their time Active, and displayed a greater
measure of success in that all of our identified goals were variety of Object-Directed and other behaviours than

during control observations in which enrichment was
absent. This addressed our concerns regarding excessive
inactivity and probable depauperate behavioural diversity
among captive pandas. Enrichment also promoted a con-
comitant reduction in the rate of and proportion of time
spent engaged in Stereotypic behaviour and Feeding
Anticipation. Alleviating Feeding Anticipation is of
special importance because this activity is closely associ-
ated with the development of stereotypies (Falk 1977;
Hughes & Duncan 1988; Mason 1991b). It is during
this time that appetitive behaviour is activated, wherein
the animal is motivated to find and acquire food.
Depauperate environments apparently do not provide
appropriate opportunities for animals to channel this
motivation, and subsequently many animals develop
rigid, entrenched stereotypic routines (e.g. adjunctive
behaviour) while waiting for food delivery. Reduction
of Feeding Anticipation may, therefore, remove the
motivational context in which many stereotypies arise.
Finally, age played a prominent role in governing respon-
siveness to enrichment, an effect which centred primarily
around subadults’ penchant for play and adults’ interest
in feeding. Subadults showed no overall preference
between the feeding (fruitcicle, puzzle feeder) and non-
feeding enrichment items (burlap sack, ball/bobbin,
branches), whereas adults significantly preferred feeding
enrichment.
An interesting, but ultimately unanswerable, question
arising from this research is how these behavioural effects
of enrichment relate to the subjective experience of the
animal. Do these measures of reduced abnormal behav-
iour, increased behavioural diversity and so forth indicate
improved ‘psychological well-being’ of the animal?
Although the driving force behind our enrichment pro-
gramme was guided by the nebulous concept of well-
being, this thorny issue raises a host of problems
regarding its measurability and interpretation (Mason
1991a, b; Dawkins 1990). In light of these limitations, it is
perhaps best for our present purposes to limit our evalu-
ation to that which is quantifiable, and maintain the
hope that these effects are manifestations of improved
well-being and reflect meaningful enhancement of living
conditions for the pandas.
Motivational Effects of Enrichment: Support for
the Ethological Needs Model?
The ethological needs model of motivation posits that
animals may suffer if they are unable to carry out normal
behaviour patterns (Hughes & Duncan 1988). When the
environment affords few opportunities to perform certain
natural behaviours, particularly those associated with
feeding, animals often display these activities even when
they yield no functional consequence. For example,
chickens conditioned to stand on a platform for 15 s will
ground-scratch while waiting, even though this behav-
iour is not part of the conditioned response (Breland &
Breland 1961). If behavioural opportunities are severely
constrained (i.e. in an impoverished environment), many
animals develop stereotypies that replace these natural
behaviours. Inglis and his colleagues offer a different
SWAISGOOD ET AL.: ENRICHMENT FOR PANDAS 453
interpretation of these apparent behavioural needs, argu-
ing that it is not the performance of these appetitive
behaviours that is important, but rather the information
gathered through the exploratory process associated with
locating and extracting hidden or dispersed food
resources (Inglis & Ferguson 1986; Inglis et al. 1997).
These researchers do not relate their model (the infor-
mation primacy model) to suffering or stereotypy, yet this
bridge is easily built, as implied in Mench’s (1998) linkage
of information primacy with enrichment needs (see also
Shepherdson et al. 1993). Following Toates (1983),
Mench also reduces the emphasis of these models on
feeding behaviour, expanding them to include the con-
struction of a cognitive map of other relevant environ-
mental dimensions. It is in this more general spirit with
which we approached our ‘test’ of the ethological needs
model.
If enrichment is to have an effect on stereotypic behav-
iour via the reduction of ethological needs, then it must
reduce the occurrence of stereotypies even when the
animal is not actively interacting with the enrichment
item. Thus, overall rates or time allocated to stereotypy in
the presence and absence of enrichment is insufficient
to test the ethological needs model. To adequately test
enrichment efficacy, therefore, it is necessary to factor out
the time spent interacting with enrichment and analyse
how the animal spends the remainder in the aftermath of
an enrichment interaction. Taking this idea to its logical
conclusion, we need furthermore to demonstrate that
enrichment carried out throughout an animal’s life can
meet its ethological needs such that the development of
stereotypies is prevented in the first place.
The latter ambitious goal is beyond the scope of this
paper, but our results clearly show that enrichment
affected the pandas’ motivation to engage in Stereotypic
behaviour and Feeding Anticipation in the aftermath of
enrichment interactions. After factoring out the time
spent with enrichment, pandas still performed stereo-
typies at a lower rate and spent less time engaged in
stereotypic behaviour. Feeding Anticipation was con-
comitantly reduced, suggesting that feeding motivation
may play a role in stereotypy performance (cf. Falk 1977).
Also consistent with the ethological needs model is the
observation that enrichment promoted a greater diversity
of nonobject-directed activities, again suggesting a carry-
over effect that not only reduced abnormal behaviour,
but also encouraged a more natural behavioural reper-
toire. (Note, however, that while this analysis may test
some predictions of the ethological needs model, it does
not address the more practical question of long-term
effects on motivation.)
Qualitative observations also support these conclu-
sions. On several occasions we observed that episodic
interactions with enrichment items appeared to break the
‘momentum’ toward stereotypic behaviour. For example,
behaviours such as head toss, pirouette, cage climb and
pace seemed to ‘build up’ after 2–4 min of fast-paced,
apparently ‘anxious’ locomotion, often associated with
Feeding Anticipation. Sometimes pandas appeared to
break away from this somewhat predictable trajectory to
investigate, consume, or play with enrichment items.
454 ANIMAL BEHAVIOUR, 61, 2
Following these interactions, the panda’s demeanor often
appeared to be calmer for several minutes, as evidenced
by a slow-paced gait and reduced Feeding Anticipation.
This disruptive effect may explain why pandas displayed
less stereotypy even when they were not interacting with
the enrichment item.
How specific are these results to the ethological needs
model of motivation? A test of the ethological needs
model requires that any motivational effects be achieved
solely via the performance of specific behaviours, not
their functional consequences. While it might be argued
that the two feeding enrichment items achieved their
effects via the functional consequence of food acqui-
sition, the same cannot be said for the three nonfeeding
enrichment items. These items provided no secondary
reinforcer, only the reward associated with the oppor-
tunity to explore, manipulate and play. That is, all they
provided was the opportunity to perform certain behav-
iours, strongly implicating behavioural performance as
the causal variable in reducing motivation to engage in
stereotypic behaviour. It might be argued, however, that
sensory feedback and the opportunity to investigate these
objects played a causal role (i.e. information primacy
model), rather than the performance of the behaviour
itself. Without highly invasive experimental interven-
tion, such as deafferentation, these alternatives remain
empirically indistinguishable. Regardless, whether these
effects arise from performance of these behaviours or the
information obtained from them is of little practical
consequence as long as stereotypic motivation is miti-
gated and well-being promoted. Moreover, appealing to
some version of the ethological needs/information pri-
macy model seems to offer a plausible motivational
explanation for our results.
It should be noted that our experimental design did not
allow us to determine which behaviours were important
to perform to reduce these putative ethological needs. We
agree with Veasey et al. (1996, page 17), skeptics of the
ethological needs model, that it is not necessary for an
animal to perform all of the behaviours occurring among
wild-living individuals, but also concur that the ‘expres-
sion of some behaviours is essential for adequate welfare’.
Indeed, it may not be the expression of any particular
behaviour that is important, and any number of alterna-
tive behaviours may provide sufficient substitution for
those performed in the wild (see below). The fact that
all five enrichment items effected similar degrees of
enhancement in plausible measures of well-being, yet
each produced a distinct set of behaviours, implies that
not one, but some unknown combination of several
behaviours promoted by these enrichment items consti-
tute important ethological needs. Of course, our research
would be strengthened if we could determine precisely
which behaviours are important, but the fact that enrich-
ment did lead to a substantive increase in behavioural
diversity at least maximizes the probability that one or
more behavioural needs were satisfied. On a more practi-
cal level, even in species more amenable to experimental
manipulation, it is extremely difficult to deprive individ-
uals of opportunities to perform specific behaviours, and
measure the effects of such deprivation. In our research,
and indeed any study aimed at quickly arresting negative
consequences of suboptimal captive conditions, it is
neither practical nor ethical to engage in a long-term
series of investigations aimed at understanding the effect
of selective deprivation of specific behaviour patterns.
Mimicking Behavioural Opportunities Present in
the Wild
The limited space available in the confines of captive
environments makes it difficult to recreate the animal’s
natural environment in miniature. Wild-living animals
encounter a temporally and spatially variable environ-
ment, such that they are exposed to nearly constant
change at some level: trees, bushes, rocks, terrain, bodies
of water, conspecifics, other forms of wildlife. All of
these aspects of the natural environment have varying
properties that afford infinite behavioural opportunities.
Certainly, a great deal of the ‘enrichment’ in a wild
animal’s life stems from the sheer variety to which it is
exposed. Thus, practitioners of enrichment must selec-
tively choose items that afford opportunities to express
those behaviours that occur naturally in the wild. This
notion is based in part on the hypothesis that animals
have ethological needs wherein they are motivated to
perform certain behaviours even when their most basic
needs of food and comfort are met (Hughes & Duncan
1988). In the case of the giant panda, it is difficult to
determine just what its ethological needs might be
because so little is known about its behavioural repertoire
in the wild. As solitary animals that only encounter
conspecifics occasionally (Schaller et al. 1985), it seems
likely that most needs for social interaction are met at a
captive facility such as Wolong. Each panda has access
through cage bars to other pandas, and occasional full
contact when placed together during mating intro-
ductions. Opportunities for chemical communication,
pandas’ primary means of communication throughout
most of the year, might also constitute a suitable etho-
logical need, but the management strategy at Wolong
(Swaisgood et al. 2000) probably meets this need. The
clear majority of a wild panda’s activity budget is filled
with feeding on bamboo (55%, Schaller et al. 1985), while
at the Wolong Breeding Center pandas spend only 25% of
their time feeding (Mainka & Zhang 1994). In addition,
pandas spend a greater proportion of their time inactive
in captivity (61% versus 41%).
Our enrichment programme addressed some of these
perceived problems regarding activity budget and prob-
able depauperate behavioural diversity. Because increas-
ing time allocated to bamboo consumption was not a
practical solution (they are fed bamboo ad libitum), we
sought to occupy their time in alternative activities. The
two feeding enrichment items, in particular, were pro-
vided in the hope of stimulating feeding-related activi-
ties. Although the behavioural tasks required to extract
food from these items deviates from those associated with
bamboo consumption, the opportunity to employ more
appetitive feeding behaviours may partially fulfill this
putative ethological need. The pandas’ foraging strategy
requires both a great deal of processing (e.g. mastication)

and handling effort (stripping leaves, peeling culms), but
does not require much searching. The puzzle feeder
requires significant handling time, whereas the fruitcicle
promotes the processing phase of feeding in that the
animal must chew the ice block vigorously to extract food
items. Although these foraging tasks differ in the exact
form of behaviours associated with feeding on bamboo, it
might be argued that they promoted analog behav-
iours and thereby satisfied some ethological need (sensu
Veasey et al. 1996). The remaining enrichment items did
not address any specific known behaviours that pandas
perform in the wild, but may none the less provide
behavioural opportunities present in the wild. In
addition, the handling of objects incorporated some of
the same behavioural elements observed when feeding on
bamboo. For example, the branches in particular pro-
moted behaviour resembling bamboo feeding, includ-
ing stripping of the needles with the forepaws and
the bark with the teeth. While wild pandas would
not encounter burlap sacks, balls or bobbins, they cer-
tainly encounter other manipulable objects that they
investigate, manipulate or play with occasionally.
Practitioners of enrichment programmes have long
debated the relative merits of ‘artificial’ versus ‘natural’
enrichment (e.g. Lindburg 1998; Poole 1998). Of course,
the ‘naturalness’ of enrichment items varies along a
continuum from only those biotic and abiotic commodi-
ties occurring in the natural environment of the species
to natural items that do not occur in the species’ range to
natural objects modified by humans to completely
human-made objects. Initially, our philosophical stance
fell in the ‘natural’ camp, but practicality of implemen-
tation, efficacy of enrichment items at producing the
desired effect, and constraints operating in certain captive
environments have swayed us towards a balanced
approach including both ‘artificial’ and ‘natural’ enrich-
ment as needs dictate (cf. Veasey et al. 1996). However, all
of the enrichment items provisioned during this study,
regardless of their location on the naturalness con-
tinuum, were equal in their overall efficacy with regard to
occupying time, promoting Activity and Behavioural
Diversity, and reducing Stereotypic behaviour and
Feeding Anticipation. If it is the performance of these
behaviours that is important, then the goal of enrich-
ment programmes should be to promote such behaviour
using whatever means, natural or artificial, that prove
most provocative of the desired outcome.
Implications for Husbandry and Captive Breeding
One limitation of the present study was its short-term
nature (2.5 months) and rather limited opportunity for
habituation to take place. However, although several
studies have found substantive habituation even after a
single trial (reviews in Shepherdson et al. 1998), we found
no evidence of it across three trials. We also found no
evidence of cross-habituation across the 15 trials; that is,
exposure to one type of enrichment did not reduce
responsiveness to others. This initial finding provides
hope that the continued application of this enrichment
programme (in expanded form) will not lead to the
SWAISGOOD ET AL.: ENRICHMENT FOR PANDAS 455
diminishment of its efficacy. Given that we maintain
sufficient variability in the objects we provision, we are
encouraged that the programme will maintain its practi-
cal utility (cf. Weld & Erwin 1990; Crockett 1998).
Indeed, we have continued and expanded this pro-
gramme since its inception; although monitored through
less labour-intensive and quantifiable means, we are
encouraged that the pandas have continued to interact
with these enrichment items for nearly a year. We will
follow up this programme with detailed quantitative
assessments at periodic intervals throughout the lives of
these pandas, thereby enabling the long-term effective-
ness and developmental consequences of the programme
to be monitored. It is our goal to develop an enrichment
programme/husbandry protocol that ultimately alters the
developmental trajectory such that abnormal behaviours
never develop.
Regardless of whether this lofty goal is attained, we
hope that enrichment will have a beneficial effect on the
breeding success at the facility. As discussed by Carlstead
& Shepherdson (1994), environmental enrichment can
promote captive propagation in a variety of ways. For
example, pairing for mating purposes or the birth of an
offspring can represent a significant challenge to an
animal kept in a sterile, unchanging environment. That
captive giant pandas may suffer from such deficits is
suggested in the fear-induced maternal abandonment
observed in some panda mothers (Zhang et al. 2000) and
the inappropriate behavioural responses that plague
many attempts at mating introduction (Lindburg et al.
1998; R. R. Swaisgood, unpublished data). Enrichment
can provide the environmental complexity and novelty
required to develop the ability to cope with such changes,
as suggested by the reduced emotional responsiveness
observed in animals kept in enriched environments
(Renner & Rosenzweig 1986, 1987). Enriched environ-
ments have also proved effective at alleviating stress
(Carlstead & Shepherdson 1994; Carlstead 1996), and
presumably may mitigate its negative consequences for
health and reproduction. Promotion of physical exercise,
as suggested by the vigour with which many pandas
interacted with the enrichment items, may also yield
beneficial effects on health. Reintroduction of captive
giant pandas to the wild is an important long-term goal of
China’s giant panda breeding programme (MacKinnon et
al. 1989), and thus the improved behavioural competence
and postrelease survival associated with enriched en-
vironments for some species (Sheperdson 1994; Miller et
al. 1998) would be an additional welcome consequence.
Finally, the enrichment programme described in the
present study is not occurring in a vacuum, but rather is
accompanied by other efforts, such as the creation of
large naturalistic outdoor enclosures, changes in diet and
feeding protocols, and other measures to enrich the
pandas’ lives in captivity.
Acknowledgments
We thank the keepers at the Wolong Breeding Center for
their assistance in carrying out this programme and
China’s State Forestry Administration for its support. This
456 ANIMAL BEHAVIOUR, 61, 2
research was funded by the Zoological Society of San
Diego. We also wish to thank Charles and Susan Lang for
their donation in support of this project. Ronald R.
Swaisgood’s position was generously funded by a do-
nation from J. Dallas Clark. The research presented here
was evaluated and approved by the Animal Behavior
Society’s Animal Care Committee on 6 September 2000.
References
Breland, K. & Breland, M. 1961. The misbehavior of organisms.
American Journal of Psychology, 16, 681–684.
Broom, D. M. & Johnson, K. G. 1993. Stress and Animal Welfare.
London: Chapman & Hall.
Carlstead, K. 1996. Effects of captivity on the behavior of wild
mammals. In: Wild Mammals in Captivity (Ed. by D. G. Kleiman,
M. E. Allen, K. V. Thompson & S. Lumpkin), pp. 317–333.
Chicago: University of Chicago Press.
Carlstead, K. 1998. Determining the causes of sterotypic behaviors
in zoo carnivores: toward appropriate enrichment strategies. In:
Second Nature: Environmental Enrichment for Captive Animals (Ed.
by D. J. Shepherdson, J. D. Mellen & M. Hutchins), pp. 172–183.
Washington, D.C.: Smithsonian Institution Press.
Carlstead, K. & Shepherdson, D. J. 1994. Effects of environmental
enrichment on reproduction. Zoo Biology, 13, 447–458.
Chamove, A. S. & Anderson, J. R. 1989. Examining environmental
enrichment. In: Psychological Well-being of Primates (Ed. by E.
Segal), pp. 183–202. Philadelphia: Noyes.
Crockett, C. M. 1998. Psychological well-being of captive non-
human primates: lessons from laboratory studies. In: Second
Nature: Environmental Enrichment for Captive Animals (Ed. by D. J.
Shepherdson, J. D. Mellen & M. Hutchins), pp. 129–152.
Washington, D.C.: Smithsonian Institution Press.
Dawkins, M. S. 1990. From animal’s point of view: motivation,
fitness and animal welfare. Behavioral and Brain Sciences, 13, 1–61.
Falk, J. L. 1977. The origin and functions of adjunctive behavior.
Animal Learning and Behavior, 4, 325–335.
Forthman, D. L. 1998. Toward optimal care for confined ungulates.
In: Second Nature: Environmental Enrichment for Captive Ani-
mals (Ed. by D. J. Shepherdson, J. D. Mellen & M. Hutchins),
pp. 236–261. Washington, D.C.: Smithsonian Institution Press.
Hughes, B. O. & Duncan, I. J. H. 1988. The notion of ethological
‘need’, models of motivation and animal welfare. Animal
Behaviour, 36, 1696–1707.
Hutchins, M., Hancocks, D. & Crockett, C. 1984. Naturalistic
solutions to behavioral problems of captive animals. Zoologische
Garten, 54, 28–42.
Inglis, I. R. & Ferguson, N. J. K. 1986. Starlings search for food
rather than eat freely-available, identical food. Animal Behaviour,
34, 614–616.
Inglis, I. R., Forkman, B. & Lazarus, J. 1997. Free or earned food? A
review and a fuzzy model of contrafreeloading. Animal Behaviour,
53, 1171–1191.
Kreger, M. D., Hutchins, M. & Fascione, N. 1998. Context, ethics,
and environmental enrichment for zoos and aquariums. In: Second
Nature: Environmental Enrichment for Captive Animals (Ed. by
D. J. Shepherdson, J. D. Mellen & M. Hutchins), pp. 59–82.
Washington, D.C.: Smithsonian Institution Press.
Lindburg, D. G. 1998. Enrichment of captive mammals through
provisioning. In: Second Nature: Environmental Enrichment for
Captive Animals (Ed. by D. J. Shepherdson, J. D. Mellen &
M. Hutchins), pp. 262–276. Washington, D.C.: Smithsonian
Institution Press.
Lindburg, D. G., Huang, X. M. & Huang, S. Q. 1998. Reproductive
performance of giant panda males in Chinese zoos. In: Proceedings
of the International Symposium on the Protection of the Giant
Panda, Ailuropoda melanoleuca (Ed. by A. Zhang & G. He),
pp. 67–71. Chengdu, China: Sichuan Publishing House of Science
and Technology.
MacKinnon, J., Bi, F., Qiu, M., Fan, C., Wang, H., Yuan, S., Tian, A.
& Li, J. 1989. National Conservation Management Plan for the Giant
Panda and Its Habitat. Hong Kong: China Alliance Press.
Mainka, S. A. & Zhang, H. 1994. Daily activity of captive giant
pandas (Ailuropoda melanoleuca) at the Wolong Reserve. Zoo
Biology, 113, 13–20.
Mason, G. J. 1991a. Stereotypies and suffering. Behavioral Processes,
25, 103–115.
Mason, G. J. 1991b. Stereotypies: a critical review. Animal Behaviour,
41, 1015–1037.
Mench, J. A. 1998. Environmental enrichment and the importance
of exploratory behavior. In: Second Nature: Environmental
Enrichment for Captive Animals (Ed. by D. J. Shepherdson, J. D.
Mellen & M. Hutchins), pp. 30–46. Washington, D.C.:
Smithsonian Institution Press.
Miller, B., Biggins, D., Vargas, A., Hutchins, M., Hanebury, L.,
Godbey, S., Anderson, S., Wemmer, C. & Oldemeier, J. 1998.
The captive environment and reintroduction: the black-footed
ferret as a case study with comments on other taxa. In: Second
Nature: Environmental Enrichment for Captive Animals (Ed. by
D. J. Shepherdson, J. D. Mellen & M. Hutchins), pp. 97–112.
Washington, D.C.: Smithsonian Institution Press.
Poole, T. B. 1998. Meeting a mammal’s psychological needs:
basic principles. In: Second Nature: Environmental Enrichment for
Captive Animals (Ed. by D. J. Shepherdson, J. D. Mellen & M.
Hutchins), pp. 83–94. Washington, D.C.: Smithsonian Institution
Press.
Renner, M. J. & Rosenzweig, M. R. 1986. Object interactions in
juvenile rats (Rattus norvegicus): effects of different experiential
histories. Journal of Comparative Psychology, 100, 229–236.
Renner, M. J. & Rosenzweig, M. R. 1987. Enriched and Impoverished
Environments: Effects on Brain and Behavior. New York: Springer-
Verlag.
Rushen, J., Lawrence, A. B. & Terlouw, E. M. C. 1993. The
motivational basis of stereotypies. In: Stereotypic Animal Behaviour
(Ed. by A. Lawrence & J. Rushen), pp. 41–64. Wallingford: CAB
International.
Schaller, G. B., Hu, J., Pan, W. & Zhu, J. 1985. The Giant Pandas of
Wolong. Chicago: University of Chicago Press.
Shepherdson, D. J. 1994. The role of environmental enrichment in
the captive breeding and reintroduction of endangered species.
In: Creative Conservation: Interactive Management of Wild and
Captive Animals (Ed. by G. Mace, P. J. S. Olney & A. Feistner),
pp. 167–177. London: Chapman & Hall.
Shepherdson, D. J. 1998. Tracing the path of environmental
enrichment in zoos. In: Second Nature: Environmental Enrichment
for Captive Animals (Ed. by D. Shepherdson, J. Mellen & M.
Hutchins), pp. 1–12. Washington, D.C.: Smithsonian Institution
Press.
Shepherdson, D. J., Carlstead, K., Mellen, J. D. & Seidensticker, J.
1993. The influence of food presentation on the behavior of small
cats in confined environments. Zoo Biology, 12, 203–216.
Shepherdson, D. J., Mellen, J. D. & Hutchins, M. (Eds) 1998.
Second Nature: Environmental Enrichment for Captive Animals.
Washington, D.C.: Smithsonian Institution Press.
Sokal, R. & Rohlf, F. 1998. Biometry. New York: W. H. Freeman.
Swaisgood, R. R., Lindburg, D. G. & Zhou, X. 1999. Giant pandas
discriminate individual differences in conspecific scent. Animal
Behaviour, 57, 1045–1053.
Swaisgood, R. R., Lindburg, D. G., Zhou, X. & Owen, M. A. 2000.
The effects of sex, reproductive condition and context on discrimi-
nation of conspecific odours by giant pandas. Animal Behaviour,
60, 227–237.

Toates, F. M. 1983. Exploration as a motivational and learning
system: a cognitive incentive view. In: Exploration in Animals and
Humans (Ed. by J. Archer & L. I. A. Birke), pp. 55–71. Berkshire:
Van Nostrand Reinhold.
Veasey, J. S., Waran, N. K. & Young, R. J. 1996. On comparing the
behaviour of zoo housed animals with wild conspecifics as a
welfare indicator. Animal Welfare, 5, 13–24.
SWAISGOOD ET AL.: ENRICHMENT FOR PANDAS 457
Weld, K. & Erwin, J. 1990. Provision of manipulable objects to
cynomolgus macaques promotes species-typical behavior.
American Journal of Primatology, 20, 243.
Zhang, G. Q., Swaisgood, R. R., Wei, R. P., Zhang, H. M., Han,
H. Y., Li, D. S., Wu, L. F., White, A. M. & Lindburg, D. G. 2000.
A method for encouraging maternal care in the giant panda. Zoo
Biology, 19, 53–63.