Mammal Study 41: 155–161 (2016)
© The Mammal Society of Japan Short communication
Food habits of Asian elephants Elephas maximus in a rainforest
of northern Peninsular Malaysia
Shiori Yamamoto-Ebina1, Salman Saaban2, Ahimsa Campos-Arceiz3,* and Seiki Takatsuki1
1 School of Veterinary Medicine, Azabu University, Japan
2 Biodiversity Conservation Division, Department of Wildlife and National Parks Peninsular Malaysia, Malaysia
3 School of Geography, University of Nottingham Malaysia Campus, Malaysia
Abstract. Little is known about the food habits of Asian elephants (Elephas maximus) in tropical
rainforests of Southeast Asia. In Peninsular Malaysia, elephant habitat has been extensively modied
by human intervention in the past few decades. Most of the primary forest has been logged or given
way to plantations, infrastructure, and human inhabitation. Here we compare the food habits of wild
elephants in three habitats of Belum-Temengor Forest Complex (BTFC): (1) primary forest, (2)
selectively-logged forest, and (3) by the side of a road that bisects the forest complex. We used
microhistological fecal analysis to describe elephants’ diet. Elephant dung in the primary forest was
mainly composed of non-grass monocotyledonous leaves (22%), woody debris (32%), and woody ber
(20%). Those in the logged forest were similar; non-grass monocotyledonous leaves accounted for
33%, woody debris for 24%, and ber for 26%. At the roadside, elephant dung was dominated by
grasses (47%). We conclude that by the road elephants shift their diet into grasses, suggesting that the
road acts like a large forest gap, promoting the availability of grasses and other early succession plants.
Elephant feeding by the road poses potential conservation conicts by means of road accidents and
increased contact with people.
Key words: diet, forest fragmentation, infrastructure, megafauna, selectively-logged forest.
Asian elephants (Elephas maximus) are endangered
(Choudhury et al. 2008) with a wild population estimated
in 39,000–47,000 individuals (Fernando and Pastorini
2011) in a range of ca 450,000 km2 in South and South-
east Asia (Sukumar 1989). Within the range, protected
areas cover only 72,000 km2 (~16% of the total) and hence
further population declines are expected (Leimgruber et
As the largest terrestrial animals, elephants have huge
demands for food, consuming as much as 150 kg of fod-
der per day (Vancuylenberg 1977), and large home ranges
(e.g., Alfred et al. 2012). The need to nd large amounts
of food determines elephants ranging patterns and other
aspects of their behavior. Understanding Asian elephant
feeding ecology and food habits is thus essential for the
conservation of the species, especially in the increasingly
human-modied landscapes they inhabit.
Elephants in Southeast Asia feed on a wide range of
plant types, often showing preference for grasses and
other monocotyledonous plants (Olivier 1978; English
et al. 2014a, 2014b); studies in southern China and
Myanmar showed that elephants feed on nutritious foods
other than grasses (Chen et al. 2006; Campos-Arceiz et al.
2008). Asian elephants are also known to consume a
range of eshy fruits, hence playing an important role as
agents of seed dispersal (Campos-Arceiz and Blake
2011). Nevertheless, information of elephant food habits
in Southeast Asian rainforests is still quite limited, and
further quantitative analyses are required.
Peninsular Malaysia, with a population of ca 1,500 wild
elephants (Saaban et al. 2011) is one of the strongholds of
elephant conservation in Southeast Asia and has experi-
enced a very rapid transformation that led to a decrease in
forest cover from nearly 90% in 1950 (FDTCP 2009) to
less than 40% in 2010 (Miettinen et al. 2011). Elephant
habitats in Malaysia have been replaced by forest planta-
*To whom correspondence should be addressed. E-mail: firstname.lastname@example.org
156 Mammal Study 41 (2016)
tions, mostly of rubber (Hevea brasiliensis) and oil palm
(Elaeis guineensis), dams, urban areas, and human infra-
structure (Daim 1995). Elephants now often enter into
plantations and raid crops, resulting in the so-called
human-elephant conict—currently the main threat for
Asian elephants in Malaysia (Saaban et al. 2011; Campos-
Tropical forests such as those in Peninsular Malaysia
have gaps formed after felling of large trees, resulting in
the invasion of bamboos and ginger shrubs. It is thus
expected that forest-dwelling Malaysian elephants can
feed on such gap-specialist plants as found in riverine
habitats of the Kinabatangan river, in Borneo (English et
al. 2014a, 2014b). If primary forests are logged, or a road
or a plantation is created contiguous to the forest, this will
result in open patches and forest edges. These changes
would facilitate the growth of grasses and forbs by
increasing light availability.
The objective of this study is to provide a quantitative
description of the food habits of elephants in a Malaysian
rainforest environment and compare the diet composition
of elephants living in primary forest, selectively logged
forest, and by the side of a road bisecting the forest.
Materials and methods
This study was conducted in Belum-Temengor Forest
Complex (BTFC), Perak, northern Peninsular Malaysia
(Fig. 1). BTFC is one of the largest blocks of continuous
forests in Peninsular Malaysia; it is contiguous with
Halabala Wildlife Sanctuary and Bang Lang National
Park, in southern Thailand, and forest areas in the Malay-
sian states of Kedah and Kelantan. BTFC is a hilly area
with an altitudinal range of 130–1,500 m above sea level
and vegetation that includes lowland dipterocarp, hill
dipterocarp, and montane forests. Much (152 km2) of
BTFC landscape is submerged under Tasik Temengor, a
large man-made lake dammed in the late 1970s. BTFC is
rich in biodiversity, including over 3,000 species of
owering plants, 185 bird species, and a wide range of
From a management point of view, BTFC is divided in
two blocks—the Royal Belum State Park in the north
and the Temengor Forest Reserve in the south—bisected
by the East-West highway (Federal Route 4; Fig. 1).
Royal Belum was gazetted in 2007 and with an area of
Fig. 1. Map of Peninsular Malaysia with a close-up of Belum-Temengor Forest Complex, where this study took place. White dots indicate roughly
the sampling areas.
Yamamoto-Ebina et al., Diet of Malaysian elephants 157
1,175 km2. Most of Royal Belum has never been logged.
Temengor on the other hand is a production forest where
logging is allowed and currently ongoing. The East-West
highway is a two-lane road that connects the ~125 km
between the towns of Gerik and Jeli. This road was com-
pleted in 1975 and a band of 3 km to the north and south
of the road was logged in 1970–1995 (MNS 2009). Ele-
phants are frequently seen crossing the road and feeding
in the grasslands on the side of the East-West highway.
Sample collection and analysis
Elephant dung samples were collected in three kinds
of environments within BTFC: (1) in primary forests of
Royal Belum, (2) in selectively logged forests of
Temengor, and (3) by the road side (Fig. 1). We col-
lected a total of 30 independent samples, nine from the
primary forest, 10 from the logged-forest, and 11 from
the roadside. Each sample was collected from a separate
dung pile, from which we collected a spoonful amount
of dung from the internal part of one dung bolus. The sam-
ples were collected in June-July 2013 and preserved in
60% ethanol for subsequent analysis.
We used microhistological fecal analysis (Stewart
1967) and the point-frame method (Chamrad and Box
1964) to produce a quantitative description of the dung
composition. This approach is widely used for ruminants
(e.g., Campos-Arceiz et al. 2004). The principle is consis-
tently usable for elephant dung too. We used a slide glass
gridded with a 1.0 mm aperture and sizing 40 mm by 80
mm. On the slide glass, a frame of 20 × 50 mm was set,
and we put fecal contents into the frame. Because this
method can only be used to analyze small particles and
elephants often defecate large indigested food fragments,
we cut dung contents into smaller pieces with scissors.
They were washed over a sieve of 0.5 mm aperture and
the retained fragments were spread over the slide glass,
and analyzed under a binocular microscope at a magni-
tude of ×40 and ×100. Each fragment covering the cross-
ing point of the grid was identied and counted. We
counted up to 200 points and this process was replicated
three times with each sample and the results averaged
(Takatsuki and Tatewaki 2012).
Initially we categorized plant fragments into 14 groups
but since some of these categories were very rare, we
grouped some of them together, reducing the nal list to
nine categories: (1) grass leaves, (2) monocotyledonous
leaves other than grasses, (3) the culms of grass, bamboo,
and ginger, (4) banana culms, (5) dicotyledonous leaves,
(6) woody debris (small broken debris of wood), (7)
woody ber (i.e. brous material of woody plants other
than leaf veins), (8) others (including fruit parts, seeds,
bark, and root), and (9) unidentied materials.
The fecal composition diversity at each habitat type
was expressed using the Shannon-Wiener’s diversity
H’ = –
Pi ln Pi
where Pi is the proportion of i in the fecal composition.
Fecal composition similarity across habitats was esti-
mated using Whittaker’s percentage of similarity (PS):
PS = 1
min (Pai, Pbi)
where Pai and Pbi were the proportion of food i at habitat
a, and habitat b, respectively.
Statistical analyses we conducted using R statistical
environment (R Core Team 2016). Differences in diver-
sity (H’) and percentage values of the major food catego-
ries were compared among the three habitats using the
Kruskal-Wallis test. When differences were signicant,
we conducted multiple comparison tests and assessed
signicance using the Dwass, Steel, Critchlow, Flinger W
statistic (function pSDCFlig in R’s NSM3 package with
method = “Asymptotic”; Schneider et al. 2015).
The fecal composition diversity index (H’) ranged from
1.38 to 1.48 and was not different across habitat types
(c2 = 4.6, df = 2, P = 0.098). Fecal composition was very
similar between primary and logged forest (PS = 73.5%)
and quite different between the logged forest and by the
roadside (PS = 36.6%). PS between the primary forest
and the roadside was 51.6%.
Overall, the diet of elephants in BTFC was dominated
by four food categories that accounted for roughly 80% of
the dung composition: grass leaves, the leaves of other
monocots (ginger, palms, and others), woody debris, and
woody ber. Other plant types and parts contributed
small amounts (Fig. 2). The distribution of the different
food categories was, however, very distinct across habitat
types (Fig. 2). Grass leaves, for example, contributed dif-
ferently to elephant diet across habitats, being the domi-
nant food category by the roadside (mean ± SD = 46.6 ±
13.3%) but almost absent in the logged forest (1.1 ± 2.1%;
158 Mammal Study 41 (2016)
Fig. 2). The contribution of grasses in the three habitats
differed signicantly among each other (Table 1). The
contribution of other monocot leaves also differed across
habitats showing the opposite trend to grass leaves—they
were most abundant in the logged forest (33.4 ± 12.9%)
and least by the roadside (3.5 ± 2.7%; Fig. 2). The dif-
ferences were also signicant (Table 1), although the
pair-wise comparisons suggest that only the difference
between logged forest and the roadside are signicant
(Table 1). The contribution of woody debris was also dif-
ferent: they accounted for 32.2 ± 10.7% and 23.5 ± 6.3%
at the primary and logged forest, respectively, but 10.7 ±
6.8% at the roadside (Fig. 2). Woody debris contribution
in the primary forest was greater than that of the logged
forest and the roadside, and in the logged forest greater
than in the roadside (Table 1). This was similar to woody
ber, which accounted for 20.3 ± 10.5% and 26.1 ± 12.6%
in the primary forest and the logged forest, and 13.1 ±
9.6% at the roadside. The value of woody ber by the
roadside was smaller than in the logged forest but there
were no differences between primary and logged forest,
nor between primary forest and the roadside (Table 1).
This is one of the few quantitative analyses of wild
elephant food habits in tropical rainforest environments
of Southeast Asia (see also Olivier 1978; English et al.
2014a, 2014b) and to our knowledge the rst to use dung
microhistological analysis to identify diet contents.
Forest elephants have the most diverse diets of any wild
herbivores (Blake 2002). Since direct observation of wild
elephant feeding is difcult in tropical rainforest, previ-
ous studies have generally relied on indirect observa-
tion of feeding signs (e.g., English et al. 2014a, 2014b).
Our approach—using microhistological analyses—has
allowed us to identify the relative importance of differ-
ent kinds of plant types as elephant food in the three
different habitats within BTFC. We found that although
food diversity—at the resolution we studied it—was not
different across habitats, the relative contribution of the
different food blocks were considerably different.
The main constituents of the elephant feces in the pri-
mary forest were grass leaves, monocot leaves, woody
debris, and woody ber. Dicot leaves unexpectedly ac-
Fig. 2. Elephant diet composition in three habitat types in Belum-Temengor Forest Complex based on microhistological analyses of elephant
dung. GBG culm = grass, bamboo, and ginger culm; others = fruit parts, seeds, bark, and root. Error bars represent standard errors.
Yamamoto-Ebina et al., Diet of Malaysian elephants 159
counted for a small proportion. It is unclear whether the
elephants did not feed on dicot leaves or digestibility of
them is higher than other food plants. As we expected,
grasses were less important in the forest than by the road
(Fig. 2). However, against our expectations, grasses were
more important in the primary than in the logged forest.
This is probably a sampling effect since the samples in
the primary forest were collected in areas near Temengor
Lake, where grasses are relatively common. It is interest-
ing that in both primary and logged forest habitats, non-
grass monocotyledonous plants including gingers and
palms were important for elephants. These plants grow in
gaps and other disturbed habitats, which again reinforces
the idea of elephants as edge (or gap) specialists (Campos-
Arceiz 2013). Gaps are relatively common in Southeast
Asian dipterocarp forests, where large trees tend to have
small root systems and often fall down producing forest
gaps. Elephants are likely to take advantage of these gaps
and other disturbed habitats such as streams and lake beds.
In the logged forest, monocot leaves and ber repre-
sented a higher proportion of the diet and grass leaves
were almost absent. This seems to reect well the habitat
vegetation where bamboos, gingers, and palms grow
abundantly while grasses are uncommon. The great
similarity between the primary and the logged forest is
attributed to the dominance of monocot leaves, woody
debris, and ber.
By the roadside, elephants were largely reliant on
grasses like Imperata spp., Panicum spp. and other mono-
cot plants that grow abundantly on the sides of the road.
The greater contribution of banana (Musa spp.) was also
unique to this habitat. These are probably wild bananas
that grow abundantly in disturbed habitats. Southeast
Asian rainforests are often described as ‘food deserts’,
where food for large herbivorous and frugivorous mam-
mals is very limited (e.g., Corlett 2007). This must be
especially the case for elephants. In this context, roads
like the East-West highway may be ecologically similar
to a very large gap, where grasses and other early succes-
sion plants grow abundantly. For elephants in BTFC, the
road probably represents a trade-off between abundance
of high quality food (mostly grasses) and high levels of
risk such as trafc and contact with people. Judging from
our casual observations (Campos-Arceiz personal obser-
vation) and data from GPS-telemetry (Campos-Arceiz et
al. unpublished data), some elephants in BTFC spend a
substantial amount of time near the East-West highway,
which brings potential conservation conicts and might
result in changes in the behavior and ecological function
Although our sample size was relatively small and our
sampling did not cover different seasons (e.g., fruiting
periods), this study proves the usefulness of the point-
frame method to study elephant food habits, and the
Table 1. Kruskal-Wallis and Dwass-Steel-Crichlow-Fligner multiple comparisons statistical test outputs for relative contribution of major food
categories to the diet of Asian elephants in three habitats of Belum-Temengor Forest Complex
Kruskal-Wallis test Dwass, Steel, Critchlow, Fligner
Food category chi-sq df P-value Pair W statistic P-value
Grass leaves 41.08 2 1.20E-09 PF-LF –4.63 0.003
PF-Rd 6.39 0
LF-Rd 7.89 0
Other monocot leaves 14.84 2 0.0006 PF-LF –0.021 0.99
PF-Rd –4.73 0.002
LF-Rd –4.95 0.001
Dicot leaves 10.66 20.005 PF-LF 4.54 0.004
PF-Rd 2.2 0.27
LF-Rd –2.64 0.15
Woody debris 34.06 24.02E-08 PF-LF –4.03 0.012
PF-Rd –6.88 < 0.001
LF-Rd –6.36 < 0.001
Woody ber 12.33 2 2.00E-03 PF-LF 1.76 0.43
PF-Rd –2.87 0.11
LF-Rd –4.94 0.001
PF = primary forest; LG = logged forest; and Rd = roadside.
160 Mammal Study 41 (2016)
results showed clear patterns. In our samples there was no
presence of fruit but a few months earlier than our sam-
pling, there was a fruiting episode of Mangifera sp. and
Irvingia malayana and, at that time, many elephant dung
piles contained fruit remains (Campos-Arceiz personal
observations). Longer-term analyses covering fruiting
episodes and wider range analyses covering different
habitat types are needed.
Our results provide interesting hints on one of the
causes of the widespread human-elephant conict that is
threatening the species throughout its range. Since ele-
phants are edge specialists (Campos-Arceiz 2013), they
readily use human-disturbed environments, including
food crops and newly planted rubber and oil palm planta-
tions. Attracted both by the crop and other early succes-
sion plants, elephants come into conict with farmers as
part of their natural optimal foraging strategy. Mitigating
human-elephant conict into human-elephant coexistence
is the most important challenge for Asian elephant con-
servation and will require a sound understanding of the
behavioral and ecological drivers of the conict (Campos-
In summary, elephants in the two forest types away
from the road fed mainly on monocot leaves other than
grasses and hard dicot material while those at the road-
side fed mainly on grasses. This supports the statements
that Asian elephants can elastically utilize forage plants
depending on habitats (Sukumar 1990; Mohapatra et al.
2013). The presence of roads like the East-West highway,
bisecting otherwise continuous patches of forest, modi-
es elephant food habits and creates conservation con-
icts. In this case, mechanisms must be put in place to
reduce the risk of road accidents and human disturbance
of elephants roaming by the roadside.
Acknowledgments: This study is part of the Manage-
ment & Ecology of Malaysian Elephants (MEME), a joint
research project between the Department of Wildlife and
National Parks (DWNP) Peninsular Malaysia and the
University of Nottingham Malaysia Campus. We are very
grateful to DWNP, and especially to its Director General
Dato’ Abdul Rasid Samsudin, for the permits to conduct
this research and for the continuous support in the eld.
SY is grateful to Malaysia’s Economic Planning Unit
(EPU) for granting her the necessary permit (UPE:
40/200/19/3003) to conduct eld research in Malaysia.
Field activities were generously nanced by grants from
Yayasan Sime Darby (grant M0005.54.04), Marinescape
Eco Aquariums (grant M0004.54.04), and Azabu Univer-
sity. Mr. Sampath K. K. Ekanayaka advised and supported
us for analysis; MEME members supported eld sam-
pling and logistics; Ms Kayal Vizi A/P Karupannan
helped SY to obtain research permits, and Y. Yamada
encouraged us during the study—we are very grateful to
all of them for their kind support.
Alfred, R., Ahmad, A. H., Payne, J., Williams, C., Ambu, L. N., How,
P. M. and Goossens, B. 2012. Home range and ranging behaviour
of Bornean elephant (Elephas maximus borneensis) females.
PLOS ONE 7: e31400. DOI: 10.1371/journal.pone.0031400.t007.
Blake, S. 2002. The Ecology of Forest Elephant Distribution and Its
Implications for Conservation. ICAPB, Edinburgh, University of
Edinburgh, 303 pp.
Campos-Arceiz, A. 2013. The dangerous myth of the Noble Beast.
Gajah 38: 5.
Campos-Arceiz, A. and Blake, S. 2011. Megagardeners of the forest—
the role of elephants in seed dispersal. Acta Oecologica 37: 542–
Campos-Arceiz, A., Lin, T. Z., Htun, W., Takatsuki, S. and Leimgruber,
P. 2008. Working with mahouts to explore the diet of work ele-
phants in Myanmar (Burma). Ecological Research 23: 1057–1064.
Campos-Arceiz, A., Takatsuki, S. and Lhagvasuren, B. 2004. Food
overlap between Mongolian gazelles and livestock in Omnogobi,
southern Mongolia. Ecological Research 19: 455–460.
Chamrad, A. D. and Box, W. T. 1964. A point frame for sampling
rumen contents. The Journal of Wildlife Management 28: 473–
Chen, J., Deng, X., Zhang, L. and Bai, Z. 2006. Diet composition and
foraging ecology of Asian elephants in Shangyong, Xishuangbanna,
China. Acta Ecologica Sinica 26: 309–316.
Choudhury, A., Lahiri Choudhury, D. K., Desai, A., Duckworth, J.
W., Easa, P. S., Johnsingh, A. J. T., Fernando, P., Hedges, S.,
Gunawardena, M., Kurt, F., et al. (IUCN SSC Asian Elephant
Specialist Group). 2008. Elephas maximus. In: IUCN (2014).
IUCN Red List of Threatened Species. Version 2014.1. Available
at www.iucnredlist.org (Downloaded on 25 October 2014).
Corlett, R. T. 2007. What’s so special about Asian tropical forests?
Current Science 10: 14–20.
Daim, M. S. 1995. Elephant translocation: The Malaysian approach.
Gajah 14: 43–48.
English, M., Ancrenaz, M., Gillespie, G., Goossens, B., Nathan, S. and
Linklater, W. 2014a. Foraging site recursion by forest elephants
Elephas maximus borneensis. Current Zoology 60: 551–559.
English, M., Gillespie, G., Ancrenaz, M., Ismail, S., Goossens, B.,
Nathan, S. and Linklater, W. 2014b. Plant selection and avoidance
by the Bornean elephant (Elephas maximus borneensis) in tropical
forest: does plant recovery rate after herbivory inuence food
choices? Journal of Tropical Ecology 30: 371–379.
FDTCP. 2009. CFS: Master Plan for Ecological Linkages. Final Report,
Federal Department of Town and Country Planning, Peninsular
Malaysia. Unpublished report. Kuala Lumpur, Malaysia.
Fernando, P. and Pastorini, J. 2011. Range-wide status of Asian
elephants. Gajah 35: 15–20.
Leimgruber, P., Gagnon, B., Wemmer, C., Kelly, D. S., Songer, M. A.
and Selig, E. R. 2003. Fragmentation of Asia’s remaining wild
lands: implications for Asian elephant conservation. Animal
Conservation 6: 347–359.
Miettinen, J., Shi, C. and Liew, S. C. 2011. Deforestation rates in
Yamamoto-Ebina et al., Diet of Malaysian elephants 161
insular Southeast Asia between 2000 and 2010. Global Change
Biology 17: 2261–2270.
MNS. 2009. A Study on the Effect of Conversion of Natural Forest to
Plantations along the East-West Highway in the Belum-Temengor
Forest Complex. Kuala Lumpur.
Mohapatra, K. K., Patra, A. K. and Paramanik, D. S. 2013. Food and
feeding behavior of Asiatic elephant (Elephas maximus Linn.) in
Kuldiha Wildlife Sanctuary, Odisha, India. Journal of Environ-
mental Biology 34: 87–92.
Olivier, R. C. D. 1978. On the Ecology of the Asian elephant. Unpub-
lished PhD Thesis, Universal of Cambridge.
R Core Team. 2016. R: A Language and Environment for Statistical
Computing. R Foundation for Statistical Computing, Vienna,
Austria. Available at https://www.R-project.org/ (Accessed 1 June
Saaban, S., Othman, N., Yasak, M. N. B., Nor, B. M., Zar, A. and
Campos-Arceiz, A. 2011. Current status of Asian elephants in
Peninsular Malaysia. Gajah 35: 67–75.
Schneider, G., Chicken, E. and Becvarik, R. 2015. NSM3: Functions
and Datasets to Accompany Hollander, Wolfe, and Chicken—
Nonparametric Statistical Methods, Third Edition. R package ver-
sion 1.5. Available at https://CRAN.R-project.org/package=NSM3
(Accessed 1 June 2016).
Stewart, D. R. M. 1967. Analysis of plant epidermis in feces: a tech-
nique for studying the food preferences of grazing herbivores.
Journal of Applied Ecology 4: 83–111.
Sukumar, R. 1989. The Asian Elephant: Ecology and Management.
Cambridge University Press.
Sukumar, R. 1990. Ecology of the Asian elephant in southern India. II.
Feeding habits and crop raiding patterns. Journal of Tropical
Ecology 6: 33–53.
Takatsuki, S. and Tatewaki, T. 2012. Applicability of the point-frame
method as a food habit analysis method for omnivorous mam-
mals: a case study on medium-sized carnivores. Honyuruikagaku
[Mammalian Science] 52: 167–177 (in Japanese with English
Vancuylenberg, B. W. B. 1977. Feeding behaviour of the Asiatic
elephant in Southeast Sri Lanka in relation to conservation.
Biological Conservation 12: 33–54.
Received 22 December 2015. Accepted 4 June 2016.
Editor was Masaharu Motokawa.