evaluating the effectiveness of human–orangutan conflict mitigation strategies in sumatra
TRANSCRIPT
Evaluating the effectiveness of human–orangutan
conflict mitigation strategies in Sumatra
Gail Campbell-Smith1*, Rabin Sembiring2 and Matthew Linkie1,3
1Durrell Institute of Conservation and Ecology, University of Kent, Canterbury, Kent CT2 7NR, UK;2Human-Orangutan Conflict and Mitigation Programme, Orangutan Information Centre, Medan, North Sumatra,
Indonesia; and 3Fauna & Flora International, Jupiter House, Station Road, Cambridge CB1 2JD, UK
Summary
1. Crop raiding by great apes is an emerging conservation issue across their range. It is important
because it involves highly threatened species that can cause significant economic damage and be
killed in retribution. Yet, to date, no quantitative study has sought to test possible solutions for pre-
venting this form of human–wildlife conflict.
2. From February 2007 to August 2009, we monitored crop-raiding patterns across a Sumatran
agroforest landscape to determine background levels of human–orangutan conflict. We also inten-
sively monitored a subset of 50 farms to assess changes in farmer attitudes towards orangutan man-
agement; differences between farmer reported and independently enumerated monetary loss from
crop raiding; and the effectiveness of mitigation techniques in reducing orangutan crop raiding on
35 treatment farms (25 trialling noise deterrents and 10 trialling tree nets) in comparison with 15
control farms over a pre-trial (12 month) and a trial (18 month) phase. Five months after the trials
had ended, the ongoing use or uptake of the techniques were assessed.
3. Across the wider landscape, background levels of mean daily orangutan crop-raiding incidents
per month (±SE) farms did not significantly differ between the pre-trial (9Æ1 ± 3Æ7) and trial
(7Æ1 ± 4Æ3) phases, whilst on the 35 treatment farms it reduced significantly. Furthermore, crop
yield increased (+60Æ8%) on the netted trial trees, but reduced ()27Æ4%) on the control farm trees.
Despite this, there was no subsequent use of this technique, unlike those farmers (40%) who contin-
ued using the less-effective noise deterrents.
4. Farmer participation in the project yielded unexpected and positive attitude changes, from pre-
ferring orangutan removal (pre-trial) to in situ management with crop protection measures (post-
trial). However, project participation may have increased farmer expectations of receiving compen-
sation because the treatment farmers consistently overestimated their crop losses, unlike the control
farmers who did not.
5. Synthesis and applications. Whilst human–orangutan conflicts caused substantial losses to local
livelihoods, the identification of an effective mitigation method (nets) neither guaranteed its contin-
ued use nor uptake. Developing easy to install nets for valuable tree crops is therefore recom-
mended. Nevertheless, the project intervention efforts did create benign farmer attitudes towards
orangutan management, an essential prerequisite for managing large-bodied mammals in conflict
with people.
Key-words: crop protection, crop raiding, great ape, human habitat alteration, human–wild-
life conflict, large mammal conservation, Pongo abelii
Introduction
To avert the current biodiversity extinction crisis, conservation
practitioners require a clear understanding of the prevailing
threats and the strategies to mitigate them. Whilst the threats
are usually known and primarily relate to the competition
between humans and biodiversity for limited space and
resources (Sodhi et al. 2009), the recent calls for evidence-
based conservation indicates that the appropriate solutions
are not yet fully known (Sutherland et al. 2004; Ferraro &
Pattanayak 2006).*Correspondence author. E-mail: [email protected]
Journal of Applied Ecology 2012, 49, 367–375 doi: 10.1111/j.1365-2664.2012.02109.x
� 2012 The Authors. Journal of Applied Ecology � 2012 British Ecological Society
For large-bodied mammals living in the biodiversity-rich
tropics, mitigating human–wildlife conflicts is a conservation
priority. These conflicts have been widely documented from
each tropical continent and involve problem animals such as
elephants Loxodonta africana that crop raid in Kenya (Sitati
et al. 2003; Chiyo et al. 2011), leopards Panthera pardus that
kill livestock in Pakistan (Dar et al. 2009) and jaguarsPanther-
a onca that attack humans in Brazil (Zimmermann,Walpole &
Leader-Williams 2005). Given the magnitude of this human–
wildlife conflict, it is surprising that only a few studies have
sought to identify solutions (Sitati, Walpole & Leader-
Williams 2005). An important first step is to determine the
spatio-temporal conflict patterns and assess the effectiveness of
current mitigation strategies (Naughton-Treves et al. 1998;
Linkie et al. 2007). From this, appropriate and, where
necessary, alternative conservation intervention measures
should be developed, trialled and quantitatively assessed,
whilst controlling for the influence of confounding variables,
such as distance to human settlements (Sitati et al. 2003).
An emerging form of human–wildlife conflict in the tropics
that is beginning to receive greater attention is that involving
non-human great ape species. Reports of crop raiding exist for
chimpanzees Pan troglodytes in Uganda and Guinea Republic
(Reynolds 2005; Hockings, Anderson & Matsuzawa 2009),
gorillas Gorilla beringei in Rwanda (Hockings & Humle 2009)
and orangutans (Pongo abelii andPongo pygmaeus) in Indone-
sia and Malaysia (Meijaard et al. 2010; Campbell-Smith et al.
2010, 2011a,b).
Resolving human–great ape conflict is a conservation imper-
ative because these species are amongst themost threatened on
earth; their large body size means that they can cause substan-
tial economic loss to farmers through crop raiding; their high
visibility in farms may distort the perceived damage caused;
these attributes cause people to be fearful of them; they are
killed by farmers in retribution for crop raiding (Reynolds
2006); their varied diet makes it difficult for farmers to protect
crops with a single strategy and their advanced intelligence
means that they can quickly learn how to circumvent mitiga-
tion strategies. Comprehensive guidelines for mitigating
human–great ape conflict have been proposed but remain
untested in the field, making it difficult to apply the appropri-
ate conflict mitigation techniques (Yuwono et al. 2007;
Hockings &Humle 2009).
In this study, we perform the first quantitative assessment of
farm-based mitigation techniques that aimed to prevent great
ape crop raiding, in this case an isolated population of critically
endangered Sumatran orangutans living in a mixed natural
forest–agricultural landscape. First, we determine levels of
orangutan crop raiding across the entire landscape over
30 months. Secondly, we determine the various spatial
parameters that best explain crop-raiding patterns in farms.
Thirdly, we intensively monitor 50 farms during a pre-trial
phase and a trial phase, during which the effectiveness of local
and project-introduced mitigation strategies are compared
between 35 treatment farms and 15 control farms. Fourthly,
we test farmer reported monthly monetary loss caused by
orangutan crop raiding with that recorded by independent
enumerators and whether their participation in the project
elicited concurrent reporting with the enumerators. Finally, we
measure changes in farmer attitudes towards orangutan
management and the uptake of the mitigation trial techniques
by farmers, 5 months after the trial phase ended.
Materials and methods
STUDY AREA
Fieldwork was conducted from February 2007 to August 2009 in a
3234 ha closed agroforest system in Batang Serangan region
(3�43¢58Æ99¢¢N, 98�11¢41Æ99¢¢E), North Sumatra, Indonesia (Fig. 1).
The study area is enclosed by commercial oil palm Elaeis guineensis
Jacq. plantations, human settlements (>6000 people) and a large
river, all impassable to orangutans. Inside, there is a small-scale oil
palm plantation (450 ha) and pockets of degraded natural forest that
are intermixed with c. 1350 smallholder farms (2784 ha) containing
cash crop species of oil palm and rubberHevea brasiliensis, as well as
other cash and subsistence crop species (hereafter cultivated fruits)
such as jackfruit Artocarpus integer, durian Durio zibethinus, petai
Parkia speciosa and an Asian legume called jengkol Archidendron
pauciflorum.
Previous research from this study area identified an isolated orang-
utan population consisting of 16 individuals (eight adults, five adoles-
cents and three newborn infants; Campbell-Smith et al. 2011b).
Orangutans, which are primarily frugivorous in natural forest habi-
tats, were found to crop raid agricultural fruits and eat tree bark, as a
putative fallback food resource, in this area (Campbell-Smith et al.
2011a,b). Within this area, orangutans are primarily arboreal, as they
are in natural forests, but will descend to travel on the ground where
the canopy continuity of large-sized trees is disrupted.
The nearest neighbouring orangutan population is located in the
primary forests of the 2Æ75 million ha Leuser Ecosystem, c. 25 km
away (Wich et al. 2008). From Batang Serangan, the orangutan pop-
ulation comes into daily contact with the local farmers, who have
recently called upon the Department of Forestry to remove these
orangtuans. Based on this, a project (implemented by the University
of Kent, in partnership with the Indonesian Department of Forestry
and University of North Sumatra) decided to intervene by trying to
find alternative solutions for managing the orangutans in situ whilst
reducing farmer losses from their crop-raiding forays.
DATA COLLECTION
Data were collected over two phases: pre-trial (February 2007–Janu-
ary 2008) and trial (March 2008–August 2009). For this, nine enumer-
ators, from local communities, were trained over 4 weeks in the use of
orangutan crop damage and conflict mitigation datasheets, which
were modified from those produced by the IUCN ⁄ SSC African Ele-
phant Specialist Group (Hoare 1999). The enumerators also received
botanical training in wild and cultivated species identification. The
enumerators then visited the study area of 1350 farms on a daily basis
to measure crop raiding across the entire landscape, to determine
background levels for the 30-month project duration. To achieve this,
the landscape was systematically surveyed, whereby it was divided
into two sections and farmswithin each section were randomly visited
by one of two teams.
Data on the quantity and frequency of crop raiding were collected
through firstly speaking with the farmer to record perceived damage
and then independently assessing the farm for all records of damage.
368 G. Campbell-Smith, R. Sembiring & M. Linkie
� 2012 The Authors. Journal of Applied Ecology � 2012 British Ecological Society, Journal of Applied Ecology, 49, 367–375
When a crop-raiding incident was encountered, information was
recorded on its location (using a global positioning system unit), the
tree species raided, the tree part eaten (fruit or bark) and the amount
of damage (number of individual fruits or stripped bark). For the trial
phase, the study then focussed on the main crop species that had been
raided during the pre-trial phase.
MIT IGATION TRIALS
Based on the crop-raiding results from the 12-month pre-trial phase
and farmer willingness to participate, 50 focal farms were randomly
selected from the wider landscape (to cover the varying levels of crop
raiding recorded). These farms were then assigned as a ‘treatment’ or
a ‘control’ farm. The project offered the farmers no incentives to par-
ticipate because they had already willingly agreed to work with the
project. Next, for 35 treatment farms, one of two types of mitigation
techniques was introduced by the project, with no intervention being
made on the remaining 15 control farms (Fig. 1). These techniques
were based on those with proven success for large-bodied mammals
(Sitati &Walpole 2006) that seemed appropriate for great apes, based
on their behaviour, and that could be easily replicated within this
study and used by the farmers thereafter. The techniques were the fol-
lowing:
• Noise deterrents – Two types of non-projectile firing noise
makers were employed on 25 treatment farms. Hand-held
firecracker cannons that were locally manufactured from
bamboo and tin and used calcium carbide to produce a loud
noise. These cannons were only fired if orangutans were
found crop raiding within the treatment farms (excluding the
net trial farms). Locally manufactured hand-held bamboo
drums were used by the enumerators if the firecracker can-
nons proved unsuccessful, that is, did not cause the orangu-
tan tomove off the tree being raided.
• Tree nets – To prevent orangutans from raiding fruiting
trees, barrier nets of 5 · 5 cm square mesh stitching nylon
rope were placed in the canopy of 14 separate jengkol trees on
10 treatment farms. These tree barrier nets closed off the
arboreal travel pathways of the orangutans. Sometimes a net
would enclose the entire tree canopy and at other times just
part of the tree canopy depending on the position of the focal
tree with respect to other trees nearby. For example, if a focal
fruit tree had trees surrounding it thereby facilitating orangu-
tan access from all areas, the entire canopy was enclosed. If
the focal fruit tree only had a few trees to one side of it, then
only that side of the canopy was closed off. Jengkol fruit was
selected because it is the most important cash crop according
to the farmers, after jackfruit that was found to experience
low fruiting during the trial phase.
Over an 18-month trial phase, the 50 focal farms were monitored
intensively (visited twice a week). Each time a mitigation event
Fig. 1. Location of the orangutan crop-raidingmitigation in 35 treatment and 15 control farmlands in a 3234 ha closed lowland (28–165 m a.s.l.)
agroforest system in Batang SeranganDistrict, North Sumatra, Indonesia.
0·00
1·00
2·00
3·00
4·00
5·00
6·00
7·00
8·00
9·00
10·00
Febr
aury
Febr
aury
Mar
chA
pril
May
June
July
Aug
ust
Sep
tem
ber
Oct
ober
Oct
ober
Nov
embe
r
Nov
embe
r
Dec
embe
r
Dec
embe
r
Janu
ary
Janu
ary
Sep
tem
ber
Mar
chA
pril
May
June
July
Aug
ust
Mea
n m
onth
ly e
cono
mic
loss
(£)
Control farmerEnumeratorTreatment farmerEnumerator
Pre-trial phase Trial phase
Fig. 2. Comparison of control and treatment farmer reported mone-
tary loss against enumerated loss during a pre-mitigation trial phase
andmitigation trial phase.
Mitigating human–orangutan conflict 369
� 2012 The Authors. Journal of Applied Ecology � 2012 British Ecological Society, Journal of Applied Ecology, 49, 367–375
occurred (either firing a cannon or an orangutan attempting to forage
from a netted tree), the enumerators collected data on length of time
to carry out the mitigation (firing duration or foraging time), type of
technique used, distance travelled by an orangutan in response, and
whether successful or not.
Data on climate, food availability (cultivated and wild), farm pro-
file, crop protection strategies (local and project-introduced), mone-
tary losses from crop raiding and mitigation use after the trials were
collected from the 50 farms and their farmers.
CLIMATE AND FOOD AVAILABIL ITY
Rainfall data were collected twice a day (06Æ00 and 18Æ00 h) using a
standard precipitation tube rain gauge at a central location
(3�43¢577¢¢N, 98�11¢455¢¢E) in the study site. On a daily basis, the
smallholder farms were systematically monitored for fruit availabil-
ity on edible tree species. Unripe and ripe cultivated and wild fruits
were recorded as ‘present’ on a farm whether fruit availability was
above 50% of the canopy covered (Campbell-Smith et al. 2011b).
These data were compiled into fruit calendars that provided infor-
mation on daily and monthly availability of wild and cultivated
fruits on farms.
FARM PROFILE
For each focal farm, data were collected on its area (ha), wild and cul-
tivated fruit abundance, distance from the nearest village, and guard-
ing techniques in use. The traditional guarding techniques used on a
crop-raiding orangutan (throwing sticks or stones and using a bark-
ing dog or shouting) were present on all farms but over the duration
of the study were inconsistently and ineffectively applied. To deter-
mine absolute abundance and density of cultivated and wild tree spe-
cies, the total number of individual adult trees were counted for the
entire farm and divided by its size (ha). Tree size [diameter at breast
height (d.b.h.)], as a proxy for arboreal travel pathways, was mea-
sured along a 50 · 2 m transect, which was randomly placed within
the farm. We then focused only on jengkol as this was the main crop
being protected by the static mitigation trials. For the 10 treatment
farmers participating in the net trials, the farmers first estimated their
yearly yield (in kg). Similarly, the 15 control farmers were also asked
to estimate their yearly jengkol yields.
MONETARY LOSS
To compare perceived (farmer reported) against observed (enumera-
tor recorded)monetary loss fromorangutan crop raiding, a 24-month
study covering the pre-trial and trial phases (February 2007–January
2008 and September 2008–August 2009, respectively) was conducted
on the damage of the five main cultivated species (jackfruit, durian,
jengkol, petai and rubber) that the farmers identified as being most
important to them. To quantify this loss, the 50 focal farmers
reported their daily monetary loss from orangutans to the enumera-
tors. This was then related to the same damage that was indepen-
dently recorded by the enumerators and converted to a monetary (£)
value based on the then market prices and currency exchange rate.
We then focused only on jengkol fruit damage as this was the main
cultivated fruit being protected by the barriermitigation trials. The 10
participating farmers first estimated the volume of damage (in kg)
that was then converted to a monetary (£) value as above. The project
enumerators then estimated the volume of damage independently for
each crop-raiding incident, which was converted into a monetary
value using the same technique.
POST-PROJECT SURVEY
Changes in attitudes of the 50 focal farmers towards orangutan man-
agement were measured before and after the mitigation trials by ask-
ing (i) are orangutans dangerous and, if so, why? and (ii) what is your
preferred solution for reducing orangutan crop raiding on your farm?
To investigate opinions towards the mitigation techniques and to
assess whether farmers would continue to use them (including uptake
by control farmers) and why, the 50 focal farmers were interviewed
5 months after the formal trials had ended (Graham & Ochieng
2008).
DATA ANALYSIS
Spatial data were mapped using ArcGIS software v.9.2 (ESRI Inc.,
Redlands CA,USA), tabulated, where required, and then exported to
spss v.16 (SPSS, Chicago, IL, USA). The continuous variables were
normalized to reduce the disproportionate influence of outliers.
To examine orangutan crop-raiding patterns, an independent
crop-raiding incident was taken as the basic unit of measurement,
defined as a single crop-raiding event by an orangutan on the same
farm on the same day, irrespective of whether it raided more than
once (Naughton-Treves 1998). To compare crop raiding in the non-
focal farms over two project phases (12 months during pre-trial and
10 of the 18 trial months), which occurred over an unequal number of
months and also betweenmonths with a different number of days, the
mean number of daily crop-raiding incidents per month and per farm
was calculated (referred to hereafter as crop-raiding frequency).
Linear regression models were used to determine which combina-
tion of spatial factors, during the pre-trial and trial phases, best
explained overall crop-raiding frequency on each of the 50 focal
farms. Prior to this, a Pearson’s rank correlation coefficient (r) test
was performed to identify non-independence between factors. From
this, the independent factors included within the statistical models, in
various combinations, were farm size, mean distance to nearest
village, abundance of wild and cultivated tree species, wild and
cultivated tree size with d.b.h. ‡ 25 cm and use (presence ⁄ absence) oftraditional guarding. This method was repeated for the trial phase.
Linear regression models were used to test the additional relationship
between project-introduced mitigation methods (measured by their
presence or absence) and crop-raiding frequency on the 50 focal
farms.
The performance of the regression models was evaluated based on
their delta Akaike Information Criterion (DAIC) values, Akaike
weights (wi) and adjustedR2 values. In the spatial analysis, it was nec-
essary to test for non-independence caused by spatial autocorrelation
as farms close to each other may have shared similar characteristics.
The presence of spatial autocorrelation within the final models was
tested by calculating Moran’s I statistic adjusted for small distances
within Crime-Stat Software v.3.2 (N. Levine and Associates,
Annadale, VA,USA).
To compare control and treatment farm characteristics, manova
and t-tests were performed on the following variables: farm size, dis-
tance to the nearest village, use of traditional guarding methods,
abundance of large trees (d.b.h. ‡ 25-cm abundance) and key tree
species (jackfruit, jengkol, rubber, durian and petai), cultivated fruits
and crop-raiding levels during the pre-trial phase.
To test the effectiveness of the tree barrier nets, the crop yields (kg)
for treatment (n = 10 farms, 14 focal jengkol trees) and control
(n = 15 farms, 12 jengkol trees) during the pre-trial and trial phases
were measured and the effect of the nets measured (t-test). Next,
Wilcoxon signed-rank test was used to examine whether there was a
370 G. Campbell-Smith, R. Sembiring & M. Linkie
� 2012 The Authors. Journal of Applied Ecology � 2012 British Ecological Society, Journal of Applied Ecology, 49, 367–375
change in crop-raiding frequency over each of the two phases on both
treatment and control farms and the effectiveness of the mitigation
techniques and to examine perceived farmer crop loss with enumer-
ated crop loss for the 15 control farms and 35 treatment farms over
each of the two phases. To investigate how often each farm was re-
raided, crop-raiding intervals (number of days) per monthwere calcu-
lated for both the 15 control and the 35 treatment farms over each of
the two phases.
Next, to investigate the relationship between farmer perceptions
of orangutans and their conservation management, chi-squared
tests were performed both before and after mitigation trials.
Finally, the ongoing use and uptake of farm-based deterrents and
perceptions of effectiveness were analysed as the percentage of
treatment farmers that were still using project-introduced mitiga-
tion techniques 5 months after the study trial period had ended
and the percentage of control farmers that were now independently
using these techniques.
Results
BACKGROUND RATES OF CROP RAIDING
Across the entire study area, a total of 801 independent crop-
raiding incidents (frequency ± SE: 26Æ4 ± 5Æ2) were recordedby the enumerators on 179 farms during the pre-trial phase.
From the 12 cultivated fruit species recorded as raided
(Table 1), 4569 individual fruits and tree bark (mean daily fruit
volume per month ± SE: 124Æ2 ± 72Æ7 and bark:
26Æ5 ± 10Æ5) were damaged. Most (73%) incidents involved
consumption of five species that the farmers also reported as
their most important cultivated species: jackfruit (38%), jeng-
kol (15%), rubber (13%), durian (4%) and petai (3%). The
trial phase focused on monitoring these five crop species and
enumerators recorded 667 independent crop-raiding incidents
(frequency ± SE = 21Æ9 ± 4Æ1) that damaged 3604 individ-
ual crops and tree bark (mean daily volume per month ± SE;
fruits: 92Æ8 ± 32Æ3 and bark: 25Æ5 ± 6Æ8), on 143 farms (448
independent crop-raiding incidents in the 50 focal farms and
219 independent crop-raiding incidents in the 93 non-focal
farms). Excluding the 50 focal farms from both the 179 farms
in the pre-trial phase (n = 129 non-focal ⁄non-treatment
farms) and the 143 farms in the trial phase (n = 93 non-focal ⁄ -non-treatment farms), levels of mean daily orangutan crop-
raiding incidents per month (±SE) for the five main cultivated
fruits in 222 non-focal ⁄non-treatment farms did not signifi-
cantly differ between the pre-trial (9Æ1 ± 3Æ7) and trial
(7Æ1 ± 4Æ3) phases (n = 22 months, U = 44Æ0, Z = )1Æ056,P = 0Æ291).
COMPARISON OF CONTROL AND TREATMENT FARM
CHARACTERISTICS
Both the 15 control and 35 treatment farms shared similar
characteristics with no differences (manova) in traditional
guardingmethods (F1,49 = 0Æ092,P = 0Æ763), tree abundance(F1,49 = 0Æ435, P = 0Æ513), cultivated fruits, (F1,49 = 0Æ102,P = 0Æ751), distance to the nearest village (F1,49 = 0Æ209,P = 0Æ649) and tree d.b.h. ‡ 25 cm abundance
(F1,49 = 0Æ055, P = 0Æ815), as well as the abundances of jack-fruit (F1,49 = 3Æ335, P = 0Æ074), jengkol (F1,49 = 0Æ0240,P = 0Æ762) and rubber (F1,49 = 0Æ335, P = 0Æ565). Differ-
ences were found in the size of the farms (F1,49 = 6Æ013,P < 0Æ01), abundance of durian (F1,49 = 14Æ049, P < 0Æ001)and petai (F1,49 = 6Æ632, P < 0Æ01), and crop-raiding levels
during the pre-trial (F1,49 = 25Æ827, P < 0Æ001). Treatment
farms tended to be larger (t-test, d.f. = 48, t = 2Æ452,P < 0Æ01), had a higher abundance of durian (t-test,
d.f. = 48, t = 3Æ748, P < 0Æ001) and petai trees (t-test,
d.f. = 48, t = 2Æ575, P < 0Æ01), and during the pre-trial
phase received more mean crop raiding per month (t-test,
d.f. = 48, t = 5Æ082,P < 0Æ001) than the control farms.
SPATIAL PATTERNS OF CROP-RAID ING FREQUENCY
During the pre-trial phase, crop-raiding frequency in the 50
focal farms was primarily related to two key factors; farm
size and abundance of trees with d.b.h. ‡ 25 cm (Models 2Æ1and 2Æ2, Table 2) with no effect from spatial autocorrelation
(Moran’s I = )0Æ02, P > 0Æ1), distance of the farms to vil-
lages, abundance of wild and cultivated trees, abundance of
the five main raided crop species or traditional guarding
techniques employed. Orangutans tended to raid farms that
were small (<1–3 ha) and had a greater abundance of large
trees, whether cultivated or wild. In the trial phase, crop-
raiding frequency was primary related to project-introduced
mitigation techniques (Models 2Æ3, Table 2), with no effect
from spatial autocorrelation (Moran’s I = )0Æ02, P > 0Æ1),from the type of deterrent mitigation techniques employed
(i.e. noise deterrents or barriers), farm size, traditional guard-
ing techniques employed, distance of the farms to villages,
abundance of wild and cultivated trees or abundance of the
five main raided crop species. Orangutans tended to raid
farms where no mitigation techniques were employed even
though these farms were found to have higher levels of crop
raiding in the pre-trial phase.
Table 1. Cultivated species and parts eaten* by orangutans and the
plant parts that both field researchers and farmers stated as
constituting human–orangutan conflict
Crops grown Latin name Type
Areca nut Areca catechu BK*, BR*, YL
Asam Gelugur Garcinia atroviridis YL, FR
Bamboo Bambusa spp. BK, BR
Durian Durio zibethinus BK, FL*, FR*, YL
Jackfruit Arthocarpus integer BK, FR*
Jengkol Archidendron pauciflorum BK, FR*, YL
Jering Arichidendron sp. BK, FR
Mahogany Swietenia mahagoni BK, FR, BR
Mango Mangifera sp. FR*
Oil palm Elaeis guineensis FR*, YL*, BR*
Petai Parkia speciosa FL*, FR*
Rubber Hevea brasiliensis BK*, FL*, FR*,
LV, SD*, YL
BK, bark (inner cambium); BR, branch; FL, flower; FR, fruit;
LV, old leaf; SD, seed; YL, young leaf).
Mitigating human–orangutan conflict 371
� 2012 The Authors. Journal of Applied Ecology � 2012 British Ecological Society, Journal of Applied Ecology, 49, 367–375
MIT IGATION TRIAL EVALUATION
Comparisons between the two project phases revealed a signifi-
cant decrease in mean crop-raiding incidents per month on the
35 treatment farms (Wilcoxon signed rank, Z = 2Æ007,P < 0Æ05) and a non-significant increase in the control farms
(Wilcoxon signed rank, Z = 0Æ188, P = 0Æ851, Table 3). Of
the 359 crop-raiding incidents on the 35 treatment farms, the
enumerators encountered orangutan crop raiding on 49 occa-
sions and used firecracker cannons on 32 of these occasions
(65%), bamboo drums on two occasions (4%) and both on 15
occasions (30%), which eventually drove orangutans away
most (84%) of the time. On average, it took 69Æ4 ± 36Æ5 min
(±SD) until the orangutan moved away a mean distance of
191Æ3 ± 117Æ7 m. In comparison, average orangutan feeding
time on cultivated trees if undisturbed was 81Æ4 ± 54Æ1 min,
suggesting that the firecracker cannons had limited effect.
During the pre-trial phase, there was no significant differ-
ence in jengkol yield between farms (n = 10) that later utilized
tree barrier nets (during the subsequent trial phase) and control
farms (n = 12) that did not (paired t-test, d.f. = 9, t = 0Æ602,P = 0Æ562). However, after the introduction of the nets (trial
phase), the treatment farms had an average jengkol yield
increase (60Æ8%) on the focal trees (mean yield ± SE,
pre-trial: 69Æ0 ± 14Æ8 kg; trial: 176Æ0 ± 39Æ0 kg), whereas the
control farms had an average decrease (27Æ4%; pre-trial:
64Æ3 ± 18Æ8 kg; trial: 46Æ7 ± 11Æ1 kg; paired t-test, d.f. = 9,
t = 2Æ502, P < 0Æ01). Similarly, across the 10 treatment
farms, crop raiding of jengkol in the pre-trial phase
(frequency ± SE = 3Æ1 ± 1Æ3) was significantly higher than
during the trial phase (1Æ2 ± 0Æ7; paired t-test, d.f. = 9,
t = 2Æ237, P < 0Æ05). The average time (days per
month ± SE) between crop-raiding events on the 35 treat-
ment farms increased by 37Æ3% from the pre-trial (5Æ9 ± 0Æ5)to trial (8Æ1 ± 0Æ7) phases, whereas on the control farms it
increased by 125Æ6% during these phases (from 3Æ9 ± 1Æ5 to
8Æ8 ± 1Æ0).
MONETARY LOSS
Comparing the effect of project intervention (Fig. 2), there
was no difference between control farmers estimated loss and
enumerated loss during the pre-trial phase (perceived,
£15 ± 5; enumerated £10 ± 3; Wilcoxon signed rank,
Z = 1Æ136,P = 0Æ256) or the trial phase (perceived, £17 ± 3;
enumerated £13 ± 3; Wilcoxon signed rank, Z = 1Æ307,P = 0Æ191), whereas the treatment farmers overreported dur-
ing both phases (pre-trial: perceived £48 ± 7, enumerated
£12 ± 2; Wilcoxon signed rank, Z = 4Æ996, P < 0Æ001 and
trial: perceived £19 ± 3, enumerated £12 ± 3; Wilcoxon
signed rank,Z = 1Æ940,P < 0Æ05).
FARMER PERCEPTIONS ON HUMAN–ORANGUTAN
CONFLICT MANAGEMENT
The trials did not change focal farmers’ fear of the orangutans,
as 40% of respondents claimed that they were dangerous
before the trials and 46% after the trials (v2 = 0Æ320,
Table 2. Linear regression models describing the spatial patterns of mean monthly crop-raiding incidents by orangutans in 50 focal farmlands
over two project phase, formodels within 10DAIC of the top-rankedmodel
Models K DAIC wi r2
Incidents (pre-trial phase)
2Æ1. T-tree abundance + T-farm size 3 0Æ00 0Æ621 0Æ3892Æ2. T-tree abundance 2 0Æ99 0Æ379 0Æ365
Incidents (trial phase)
2Æ3. Mitigation trials 2 0Æ00 0Æ230 0Æ5302Æ4. T-tree abundance 2 0Æ91 0Æ146 0Æ3502Æ5. T-tree abundance + T-farm size 3 2Æ81 0Æ056 0Æ170
T-TA is the transformed (normalized) tree abundance of wild and cultivated tress with a d.b.h. ‡ 25 cm data, T-farm size is the trans-
formed (normalized) farm size data, K is the number of parameters in the model, DAIC is the difference between each model and the
top ranked model and wi is the AIC model weight.
Table 3. Mitigation technique assessment and change in mean daily crop-raiding incidents per month (±SE) over a 12 month pre-trial phase
and an 18-month trial phase
n
Pre-trial
Mean incidents ⁄ farmTrial
Mean incidents ⁄ farm Mean change (%)
Farm type
Treatment 35 0Æ64 ± 0Æ06 0Æ57 ± 0Æ09 )10Æ9*Control 15 0Æ25 ± 0Æ02 0Æ33 ± 0Æ10 +31Æ7
Mitigation technique
Noise deterrent 25 0Æ60 ± 0Æ07 0Æ62 ± 0Æ12 3
Barrier 10 0Æ76 ± 0Æ13 0Æ45 ± 0Æ12 )41
*P < 0Æ05.
372 G. Campbell-Smith, R. Sembiring & M. Linkie
� 2012 The Authors. Journal of Applied Ecology � 2012 British Ecological Society, Journal of Applied Ecology, 49, 367–375
d.f. = 1, P = 0Æ572). However, the reasons related to fear did
change with some (40%) farmers citing large body size or
orangutan aggression (i.e. they would bite; pre-trial) and then
most (70%) citing crop damage after the trial (v2 = 10Æ571,d.f. = 4, P < 0Æ05). Pre-trial, most (58%) farmers thought
that crop raiding would be mitigated most effectively through
the removal of the orangutans, yet after the trials, most (58%)
thought that the mitigation techniques presented a better solu-
tion, with only aminority (28%) preferring orangutan removal
(v2 = 48Æ400, d.f. = 6,P < 0Æ01).After the trials, all mitigation strategies were considered to
have been effective by both treatment farmers (94% agreeing)
and control farmers (93%) with no difference in perceived
effectiveness of individual mitigation techniques (v2 = 7Æ049,d.f. = 4, P = 0Æ133). Whether a focal farmer experienced an
observed crop raiding decrease, or not, did not influence their
perception of technique effectiveness (v2 = 1Æ775, d.f. = 1,
P = 0Æ183). Most (90%) of the focal farmers asked to be con-
sidered for future mitigation trials, with a preferred combina-
tion of static and mobile methods (50%), tree nets only (35%),
or firecracker canons only (15%) and no preference for bam-
boo drums (0%). A minority (10%) of farmers did not wish to
be considered for future mitigation trials because they consid-
ered the techniques to be poorly suited for their planned plant-
ing of oil palm, which they stated as being unpalatable to
orangutans.
POST-PROJECT MIT IGATION TRIAL USE
From the 35 treatment farmers, some (40%) continued using
the mitigation techniques 5 months after the study, with all of
these choosing bamboo drums (even though previously cited
as being the least preferredmethod), because it was inexpensive
and the easiest to use. From those (60%) not continuing with
the mitigation techniques, most (72%) gave the reasons that
the techniques took too long to set up and were difficult to use,
or were too expensive to use (24%) or other (4%). From the 10
farmers trialling the tree nets, half preferred this method, but
only if the enumerators continued to manage the nets, because
they claimed not to understand the hangingmethod and, more
likely, that the process was too time-consuming.
Discussion
A common problem faced in most human–wildlife conflict sit-
uations, particularly those involving large-bodied mammals, is
that the successful identification of crop protection measures
enables further farmland expansion into the species’s range,
especially if it is not protected (Sitati & Walpole 2006). How-
ever, in the case of Batang Serangan, the problem ismore acute
because the isolated orangutan population co-inhabits a small
and finite space with the farmers. Managing these orangutans
therefore presents a dilemma.Whilst implementing an effective
crop protection strategy should, in turn, benefit local live-
lihoods, it is predicted to remove important food sources for
orangutans, with agricultural crops comprising 21% of
their diet (Campbell-Smith et al. 2011b). Prior to this study,
translocation of the orangutan population to a larger forest-
inhabiting populationwas being considered by theDepartment
of Forestry as it was the most realistic management option,
especially given the intensifying farmer pressure on the depart-
ment to do so.However, the positive change in recorded farmer
attitudes following project intervention presents a new oppor-
tunity to manage these orangutans in their current location.
The mitigation strategies that were trialled had mixed success;
hence, future work should focus on developing a simplified tree
net. It should be easy to install and locally accepted. It could be
used to protect the most commercially valuable trees, which
would exclude some, but not all of the cultivated fruits from the
orangutans’ diet. These trial results provide important insights
for conservationmanagers seeking to reduce conflicts with any
great ape species (i.e. tree nets) or other large-bodied mammal
species (i.e. firecracker cannons).
Numerous studies have found that crop raiding occurs more
frequently in farmlands located close to the forest edge
(Naughton-Treves 1997). In contrast, orangutans in Batang
Serangan lived within the agroforest system and spatial crop-
raiding patterns were, instead, explained by farmland size and
tree abundance. Small farms supporting a higher abundance
of large trees were more prone to crop raiding in the pre-trail
phase because they were more accessible to orangutans, con-
taining a denser network of arboreal travel routes (Thorpe,
Holder & Crompton 2007). The regression models also sug-
gested that additional explanatory factors might be missing.
For example, once orangutans were situated within the farm,
other factors such as human presence and calorific content of
the food that they were eating may have influenced how long
they stayed and howmuch they ate.
To successfully mitigate wildlife crop raiding, a combination
of preventative measures (i.e. early warning and direct crop
protection) and reactive measures (i.e. foraging disturbance)
based on the spatial patterns of crop raiding are recommended
(Sitati & Walpole 2006). No early warning systems, for exam-
ple, watch towers, were used or trialled in this study, because a
farmer, if on-site, would be able to detect an approaching
orangutan without difficulty. The tree nets proved effective in
preventing crop raiding on the focal trees, a view also shared
by most of the farmers. However, it is likely that these nets
caused orangutans to search for more profitable patches, that
is, the nearest fruiting trees without nets, thereby displacing the
conflict rather than reducing overall crop-raiding levels across
the landscape, as indicated by crop raiding decreasing on treat-
ment farms and increasing on control farms. Expanding the
net barrier technique to a greater number of trees would be pre-
dicted to result in a trade-off between the time an orangutan
spent searching for unprotected trees (that would require less
handling time) against trying to capture nutritious fruits from
the netted trees.
The socio-economic methodology applied in our study to
assess the long-term effectiveness of the different conflict miti-
gation techniques is transferable to other studies on large-bod-
ied mammal species. We found that most of the farmers
expressed their willingness to continue using the tree nets after
project intervention, but none actually did so. The initial cost
Mitigating human–orangutan conflict 373
� 2012 The Authors. Journal of Applied Ecology � 2012 British Ecological Society, Journal of Applied Ecology, 49, 367–375
of purchasing a net (£58) may have seemed exorbitant to the
farmers, especially as this cost far exceeded the average annual
monetary loss incurred on an unprotected jengkol tree (£8).
Whilst this cost makes the nets an unlikely purchase for farm-
ers, it would certainly not be too expensive for a conservation
donor or government agency. Besides the financial investment,
the labour investment required may have deterred farmers.
Thus, to overcome the commonly cited problem that net instal-
lation was too time-consuming and difficult, training the farm-
ers in assembling the nets might help, but our field experience
found that it took 3–4 people c. 4 h to install a single net
around the entire tree canopy, and often at the cost of encoun-
tering aggressive wasps.
Motivating the farmers to use the tree nets was identified as
a main constraint to their use. This was certainly not the case
for the bamboo drums (40% used post-project by the treat-
ment farmers). These drums, although simple to use and inex-
pensive, provided negligible benefits to farmers, probably
because the orangutans were already used to varying levels of
human disturbance. Similarly, in Kenya, inexpensive chilli
smoke briquettes had the greatest uptake amongst farmers
(93%) trying to protect their crops from elephants, even
though the technique was not found to be effective (Graham&
Ochieng 2008). The firecracker canon deterrents also did not
guarantee a reduction in crop-raiding frequency. Only after
repeated firing (on average 60 min) by the enumerators, did a
crop-raiding orangutan leave the farm. Thus, an enhanced
scare tactic might test the effectiveness of iron canons, which
make a louder noise but are more expensive (£4) and slightly
heavier (1Æ5 kg) than the bamboo variety used (£0Æ65 and
1Æ0 kg). It is unknown whether this associated increase in dis-
turbance would precipitate an increase in orangutan stress, or
indeed, whether orangutans would quickly become habituated
to this false threat, as evidenced with crop-raiding elephants
and certain noise deterrents used in Namibia (O’Connell-
Rodwell et al. 2000).
Besides identifying mitigation technique effectiveness, or
otherwise, changes in farmer behaviour are recommended for
reducing crop loss. In Batang Serangan, farmers typically work
from 06Æ00 to 11Æ00 h and, most probably in response, orangu-
tans appear to wait until the afternoon to crop raid on the
unoccupied farms (Campbell-Smith et al. 2011b). This was
also found with crop-raiding yellow baboons (Papio cynoceph-
alus) in Kenya (Maples et al. 1976). For Batang Serangan, a
communal guarding system that works on a rotational basis
might be most effective, once a proven scare tactic technique
has been identified, but again this would greatly depend on
farmermotivation.
Several studies have used farmer estimates of crop losses, by
focusing on the volume of crops damaged (Hill 2000; Perez &
Pacheco 2006; Priston 2009) or the monetary value of the crop
loss (Wang, Curis & Lassoie 2005). Where these studies have
been accompanied by independent assessments of crop losses,
farmers have been found to overestimate their losses. In our
study, such overestimates were also recorded for the treatment
farmers. The enumerators and farmers used similar calculation
methods during the two project phases, so improved calcula-
tion accuracy is unlikely to explain why only treatment farmers
overestimated their crop loss during both phases. Whilst the
disparity between treatment farmers and enumerator reports
did lessen during the trial phase, it was still significant. A more
plausible explanation is that the farmers intentionally overesti-
mated in the hope that this would result in an equivalent form
of compensation, even though from the outset the enumerators
consistently explained that this would not happen. Surpris-
ingly, overreporting by control farmers was not found. These
farmers experienced lower levels of crop loss to orangutans
and this might have reduced their demands for compensation.
Conflicts between people and large-bodied mammal species,
for example, elephants (Sitati, Walpole & Leader-Williams
2005) and orangutans (this study), can pose a significant threat
to local livelihoods. Thus, the further identification of effective
crop protection techniques through treatment and control field
trials remains an important area for future research by practi-
tioners. Crop raiding can also reduce local tolerance towards
the conservation of the real or perceived crop pest species.
However, in our study, an initial (pre-trial) farmer view of the
orangutan as a dangerous species that should be removed to
mitigate conflict changed (post-trial) to it being considered as a
crop pest that should be managed through improved crop pro-
tection. This unexpected benefit resulting from farmer partici-
pation in the project suggests that future conservation efforts
should also explore ways to increase community participation
in a broader range of field activities and measure whether this
yields further benefits, or not.
Acknowledgements
We thank the Indonesian Institute of Sciences and the Director General of
Nature Conservation for the opportunity to conduct this research and the
Sumatran Orangutan Conservation Programme and the Orangutan
Information Centre for providing logistical support. We are grateful to
Professor Ir. Zulkifli Nasution and Hubert V.P. Simanjorang (University of
North Sumatra), our field assistants and the farmers in Batang Serangan for
invaluable assistance with our work. Funding was provided by the U.S. Fish &
Wildlife Service-Great Ape Conservation Fund (for the project ‘Bittersweet
knowledge: can people and orangutans live in harmony’, grant #31221208 and
#31221435 held by Nigel Leader-Williams), Born Free Foundation, Humane
Society International, SeaWorld and Bush Gardens Conservation Fund,
PanEco, People’s Trust for Endangered Species, Orangutan Republik
Education Initiative, Orangutan Foundation, Primate Society of Great Britain
and BorneoOrangutan Society-Canada.
References
Campbell-Smith, G., Simanjorang, H.V., Leader-Williams, N. & Linkie, M.
(2010) Local attitudes and perceptions towards crop-raiding by Sumatran
orangutans (Pongo abelii) and other non-human primates in Northern
Sumatra, Indonesia.American Journal of Primatology, 72, 866–876.
Campbell-Smith, G., Campbell-Smith, M., Singleton, I. & Linkie, M. (2011a)
Apes in Space: managing an imperilled orangutan population in Sumatra.
PLoSONE, 6, e17210.
Campbell-Smith, G., Campbell-Smith, M., Singleton, I. & Linkie, M. (2011b)
Raiders of the lost bark: Orangutan foraging strategies in a degraded land-
scape.PLoSONE, 6, e20962.
Chiyo, P.I., Moss, C.J., Archie, E.A., Hollister-Smith, J.A. & Alberts, S.C.
(2011)Usingmolecular and observational techniques to estimate the number
and raiding patterns of crop-raiding elephants. Journal of Applied Ecology,
48, 788–796.
Dar, N.I., Minhas, R.A., Zaman, Q. & Linkie, M. (2009) Predicting the
patterns, perceptions and causes of human-carnivore conflict in and
374 G. Campbell-Smith, R. Sembiring & M. Linkie
� 2012 The Authors. Journal of Applied Ecology � 2012 British Ecological Society, Journal of Applied Ecology, 49, 367–375
around Machiara National Park, Pakistan. Biological Conservation, 142,
2076–2082.
Ferraro, P.J. & Pattanayak, S.K. (2006) Money for nothing? A call for empiri-
cal evaluation of biodiversity conservation investments. PloS Biology, 44,
482–488.
Graham, M.D. & Ochieng, T. (2008) Uptake and performance of farm-based
measures for reducing crop raiding by elephants Loxodonta africana among
smallholder farms in LaikipiaDistrict, Kenya.Oryx, 421, 76–82.
Hill, C.M. (2000) Conflict of interest between people and baboons: crop-raiding
inUganda. International Journal of Primatology, 212, 299–315.
Hoare, R.E. (1999) A Standardised Data Collection and Analysis Protocol for
Human-Elephant Conflict Situations in Africa. IUCN ⁄ Species Survival Com-
missionAfrican Elephant Specialist Group, Nairobi, Kenya.
Hockings, K., Anderson, J.R. & Matsuzawa, T. (2009) Use of wild and culti-
vated foods by chimpanzees at Bossou, Republic of Guinea: feeding dynam-
ics in a human-influenced environment. American Journal of Primatology, 7,
1–11.
Hockings, K. & Humle, T. (2009) Best Practice Guidelines for the Prevention
and Mitigation of Conflict Between Humans and Great Apes, pp. 40.
IUCN ⁄ SSCPrimate Specialist Group,Gland, Switzerland.
Linkie, M., Dinata, Y., Nofrianto, A. & Leader-Williams, N. (2007) Patterns
and perceptions of wildlife crop-raiding in and around Kerinci Seblat
National Park, Sumatra.Animal Conservation, 101, 127–135.
Maples, W.R., Maples, M.K., Greenhood, W.F. & Walek, M.L. (1976) Adap-
tations of crop-raiding baboons in Kenya. American Journal of Physical
Anthropology, 45, 309–316.
Meijaard, E., Albar, G., Nardiyono, Rayadin, Y., Ancrenaz, M. & Spehar, S.
(2010) Unexpected ecological resilience in Bornean orangutans and implica-
tions for pulp and paper plantationmanagement.PLoSONE, 5, e12813.
Naughton-Treves, L. (1997) Farming the forest edge: vulnerable places and
people around Kebale National Park, Uganda. Geographical Review, 871,
27–46.
Naughton-Treves, L. (1998) Predicting patterns of crop damage by wildlife
aroundKibaleNational Park.Conservation Biology, 121, 156–168.
Naughton-Treves, L., Treves, A., Chapman, C. & Wrangham, R. (1998) Tem-
poral patterns of crop-raiding by primates: linking food availability in crop-
lands and adjacent forest. Journal of Applied Ecology, 35, 596–606.
O’Connell-Rodwell, C.E., Rodwell, T., Rice, M. & Hart, L.A. (2000) Living
with the modern conservation paradigm: can agricultural communities co-
exist with elephants? A five year case study in EastCaprivi,Namibia.Biologi-
cal Conservation, 93, 381–391.
Perez, E. & Pacheco, L.F. (2006) Damage by large mammals to subsistence
crops within a protected area in amontane forest of Bolivia.Crop Protection,
259, 933–939.
Priston, N.E.C. (2009) Exclosure plots as a mechanism for quantifying damage
to crops by primates. International Journal of Pest Management, 553, 243–
249.
Reynolds, V. (2005) The Chimpanzees of Budongo Forest: Ecology, Behaviour
and Conservation. OxfordUniversity Press, Oxford.
Reynolds, V. (2006) Budongo’s Chimpanzees and the Kinyara Sugar Works.
Budongo Forest Project – Volume 7, Number 1, pp. 4–6. e-newsletter: http://
culture.st-and.ac.uk/bcfs/documents/newsletter/BFPJune2006.pdf.
Sitati, N.W. & Walpole, M.J. (2006) Assessing farm-based measures for miti-
gating human-elephant conflict in Transmara District, Kenya. Oryx, 403,
279–286.
Sitati, N.W., Walpole, M.J. & Leader-Williams, N. (2005) Factors affecting
susceptibility of farms to crop raiding by African elephants: using a predic-
tivemodel tomitigate conflict. Journal of Applied Ecology, 42, 1175–1182.
Sitati, N.W., Walpole, M.J., Smith, R.J. & Leader-Williams, N. (2003) Predict-
ing spatial aspects of human-elephant conflict. Journal of Applied Ecology,
40, 667–677.
Sodhi, N.S., Lee, T.M., Koh, L.P. & Brook, B.W. (2009) Ameta-analysis of the
impact of anthropogenic forest disturbance on Southeast Asia’s biotas. Bio-
tropica, 41, 103–109.
Sutherland, W.J., Pullin, A.S., Dolman, P.M. & Knight, T.M. (2004) The need
for evidence-based conservation. Trends in Ecology and Evolution, 19, 305–
308.
Thorpe, S.K.S., Holder, R.L. & Crompton, R.H. (2007) Origin of human
bipedalism as an adaptation for locomotion on flexible branches. Science,
316, 1328–1331.
Wang, S.W., Curis, P.D. & Lassoie, J.P. (2005) Farmer perceptions of crop
damage by wildlife in Jigme Singye Wangchuck National Park, Bhutan.
Wildlife Society Bulletin, 342, 359–365.
Wich, S.A., Meijaard, E., Marshall, A.J., Husson, S., Ancrenaz, M., Lacy,
R.C., van Schaik, C., Sugardjito, J., Simorangkir, T., Traylor-Holzer, K.,
Doughty, M., Supriatna, J., Dennis, R., Gumal, M., Knott, C.D. & Single-
ton, I. (2008) Distribution and conservation status of the orangutan (Pongo
spp.) on Borneo and Sumatra: howmany remain?Oryx, 42, 329–339.
Yuwono, E.H., Susanto, P., Saleh, C., Andayani, N., Prasetyo, D. & Atmoko,
S.S.U. (2007) Guidelines for the Better Management Practices on Avoidance,
Mitigation andManagement of Human-Orangutan Conflict in and Around Oil
Palm Plantations, pp. 56.WWF, Indonesia, Jakarta, Indonesia.
Zimmermann, A., Walpole, M.J. & Leader-Williams, N. (2005) Cattle ranch-
ers’ attitudes to conflicts with jaguar Panthera onca in the Pantanal of Brazil.
Oryx, 394, 406–412.
Received 1 August 2011; accepted 10 January 2012
Handling Editor: Julia Jones
Mitigating human–orangutan conflict 375
� 2012 The Authors. Journal of Applied Ecology � 2012 British Ecological Society, Journal of Applied Ecology, 49, 367–375