Forest elephant distribution and habitat use in the Bossematié Forest Reserve, Ivory Coast

Abstract

We studied the influence of human presence and habitat structure on the distribution and habitat use of forest elephants (Loxodonta africana cyclotis Matschie, 1900) in the heavily exploited Bossematié Forest Reserve in south-eastern Ivory Coast. From August 1993 to April 1994, we estimated the distribution of elephants by dung counts on transects and documented their habitat use by trail mapping. Elephant dung density increased with the distance to the forest border and to the nearest village. During the study, the elephants used only 60% of the forest regularly. They avoided the rest because of forestry operations and human presence. Forest roads, in light but regular use, had no influence on the elephants' spatial distribution. Elephants avoided coffee and cocoa plantations in all seasons. During the dry season, elephants walked more often through forest parts with greater tree canopy cover than during the wet season, when they avoided valley bottoms. The main influence on the spatial distribution of elephants in the study area was human presence, followed by habitat structure.

Full text

Pachyderm No 30 January–June 2001 37
Forest elephant distribution and habitat use in the Bossematié
Forest Reserve, Ivory Coast
Jörn Theuerkauf,1Hermann Ellenberg,2 Wolf Ekkehard Waitkuwait3
and Michael Mühlenberg4
1Department of Biology, Philipps University, 35032 Marburg, Germany
2Institute for World Forestry, Federal Research Centre for Forestry and Forest Products,
21031 Hamburg, Germany
3Projet forestier SODEFOR-GTZ, B.P. 878, Abengourou, Ivory Coast
4Centre for Nature Conservation, University of Göttingen, 37075 Göttingen, Germany
Present address of corresponding author:
Jörn Theuerkauf, Am Schäperkamp 3, 27711 Osterholz-Scharmbeck, Germany
e-mail: Theuerkauf.Joern@t-online.de
Additional key words: habitat structure, human influence, Loxodonta africana cyclotis
Abstract
We studied the influence of human presence and habitat structure on the distribution and habitat use of forest
elephants (Loxodonta africana cyclotis Matschie, 1900) in the heavily exploited Bossematié Forest Reserve
in south-eastern Ivory Coast. From August 1993 to April 1994, we estimated the distribution of elephants by
dung counts on transects and documented their habitat use by trail mapping. Elephant dung density increased
with the distance to the forest border and to the nearest village. During the study, the elephants used only 60%
of the forest regularly. They avoided the rest because of forestry operations and human presence. Forest
roads, in light but regular use, had no influence on the elephants’ spatial distribution. Elephants avoided
coffee and cocoa plantations in all seasons. During the dry season, elephants walked more often through
forest parts with greater tree canopy cover than during the wet season, when they avoided valley bottoms. The
main influence on the spatial distribution of elephants in the study area was human presence, followed by
habitat structure.
Résumé
Nous avons étudié l’influence de la présence humaine et de la structure d’habitat sur la répartition et l’utilisation
d’habitat des éléphants de forêt (Loxodonta africana cyclotis Matschie, 1900) dans la Forêt Classée de
Bossematié (sud-est de la Côte d’Ivoire) qui fut fortement exploitée. De août 1993 à avril 1995, nous avons
estimé la répartition des éléphants par des comptages de crottes sur transects et décrit leur utilisation d’habitat
par la cartographie de leurs traces. La densité des crottes d’éléphants a augmenté avec la distance de la lisière
de la forêt et du plus proche village. Au cours de l’étude les éléphants ont utilisé régulièrement seulement
60% de la forêt. Les éléphants ont évité le reste de la forêt à cause d’opérations forestières et de la présence
d’hommes. Les routes forestières en usage léger mais constant n’avaient aucune influence sur la distribution
spatiale des éléphants. Durant toutes les saisons, les éléphants ont évité les plantations de café et de cacao.
Pendant la saison sèche, les éléphants de forêt ont marché plus souvent dans les parties de forêt avec une
canopée plus fermée que pendant la saison des pluies où ils ont aussi évité les vallées. Dans le site de recher-
che, l’influence principale sur la répartition spatiale des éléphants était la présence humaine suivie par la
structure d’habitat.

38 Pachyderm No 30 January–June 2001
Introduction
Following poaching and habitat loss, the number of
forest elephants in West Africa has declined dramati-
cally during the last decades (Barnes 1999). In 1988/
1989, Merz and Hoppe-Dominik (1991) estimated
that 200 elephants lived in south-eastern Ivory Coast.
In 1993, only about 55 remained (W.E. Waitkuwait,
Y. Guiro, J. Theuerkauf, H. Ellenberg, unpubl. data,
1993): roughly 20 in the Songan-Tamin-Mabi-Yaja
Forest Reserve complex (1700 km²), 5 in the Béki
Forest Reserve (150 km²), and 30 in the Bossematié
Forest Reserve (220 km²). For the last of the forest
elephants in south-eastern Ivory Coast to survive, it
was necessary to implement measures to prevent fur-
ther poaching and to protect large forest tracts that
meet the habitat needs of the animals. To contribute
towards establishing conservation priorities for ele-
phants in south-eastern Ivory Coast, we investigated
the influence of human presence and forestry opera-
tions on the distribution and habitat use of the ele-
phants in the Bossematié Forest Reserve. The study
was part of a long-term biomonitoring programme
included in a rehabilitation project for the reserve.
The project was created to develop methods for the
restoration and sustainable use of all formerly over-
exploited forest reserves in south-eastern Ivory Coast.
Barnes et al. (1991) and Hall et al. (1997) showed
that in central Africa elephant densities are directly
related to distance from human settlements or roads.
We therefore investigated if human presence affected
the distribution of the elephants in and around our
study area. Elephants that live in large forest com-
plexes prefer habitats with an open canopy and dense
herbaceous tangle (see Barnes et al. 1991, Prins and
Reitsma 1989, Struhsaker et al. 1996). However, tim-
ber extraction had been heavy in our study area (Wöll
1992), and the canopy was much more open than that
in the forest complexes where the above-mentioned
studies had been conducted. The study area had a low
density of fruit trees, an important food source for
forest elephants in West Africa (Alexandre 1978,
Short 1981, White et al. 1993, Theuerkauf et al.
2000b). The objectives of this study were to deter-
mine 1) the influence of roads, villages, the forest
border and forestry operations on the distribution of
elephants and 2) the habitat types that elephants se-
lect. The results were necessary to design manage-
ment measures aimed at improving the survival
chances of the elephants.
Study area
The Bossematié Forest Reserve (Forêt Classée de
Bossematié) is a moist, semi-deciduous rain forest in
the lowland zone of south-eastern Ivory Coast (6°22' to
6°33' N, 3°24' to 3°35' W). The annual distribution of
rainfall is bimodal with peaks in September–October
and April–July; August and December–February are
dry. Mean annual rainfall for 1961–1990 was 1330 mm
but varied significantly from year to year (Schroth
unpubl.). The 220-km² forest block is surrounded by
fallow land as well as cocoa, banana and coffee planta-
tions. Human density outside the forest is about 25 in-
habitants per km2 in the north and west of the study
area and about 50 inhabitants per km2in the south and
east (Kientz unpubl.). There are no settlements in the
forest itself. Cars, bicycles and pedestrians pass regu-
larly but infrequently on the forest roads, which con-
nect villages situated around the forest.
During the 30 years up to 1988, the forest had been
heavily exploited for timber (Wöll 1992), resulting
in a mosaic of degraded forest with few old fruit trees
and many small clearings. Only 16% of the forest
has a canopy cover over 30%, whereas about 8% of
its surface consists of mostly illegal cocoa and coffee
plantations, clustered mainly along the southern mar-
gin of the forest (Wöll 1992). During the time of this
study, forestry operations (cleaning around target
trees, replanting, inventorying) were concentrated in
the northern part of the forest.
Elephant poaching was common in the 1980s un-
til 1990 when the rehabilitation project for the
Bossematié Forest Reserve started and elephant
poachers were arrested. During this study, some 27
to 33 elephants lived in the reserve (Theuerkauf et al.
2000a) and no elephant was observed to migrate into
or out of the study area.
Methods
We counted elephant droppings on permanent 2-km
transects to document the distribution of elephants
within the study area. Because the elephants some-
times walked along the permanent transects, we used
the presence of elephant droppings in 250-m sectors
to represent the relative density of elephants, which
ensured that agglomerations of droppings produced
by an elephant group that had walked along a transect
did not distort the data. We then calculated the pro-
portion of sectors with elephant dung on each transect.

Pachyderm No 30 January–June 2001 39
We noted elephant droppings that we could see from
the transect line and removed all droppings after count-
ing to ensure an independent sample for the next count.
From November 1993 to February 1994, we counted
on 8 transects monthly. Before the November count,
we had removed droppings in October. In March 1994,
we cleared and included in the April counts a further 8
transects. To assess human influence on the distribu-
tion of the forest elephants, we compared by regression
analyses the percentage of transect sectors with elephant
dung with the distances between the centres of the 16
transects and the nearest road, the nearest forest border
and the nearest village.
To determine how the forest elephants used the habi-
tat, we searched for fresh elephant trails that crossed
forest roads. We then followed and mapped the trail by
compass and topofil (a distance-measuring instrument)
with the assistance of an experienced elephant tracker.
We followed the trails of 16 different elephant groups
for distances between 0.8 and 15.7 km (total 82.2 km)
and juxtaposed the trails we had followed to a tree
canopy stratification map and a soil map.
The tree canopy stratification map of the Bossematié
Forest Reserve (Sabenin and Kra unpubl.), based on
aerial pictures taken in 1992 and 1993, distinguished
between strata of big trees (30–40 m in height) and small
trees (up to 20 m). We pooled the strata into five classes:
canopy cover < 10%, 10–20%, 20–30%, > 30%, and
plantations. The soil map (Gerold and Hetzel unpubl.),
which was mainly based on the topography of the study
area, identified Ferralsol and Plinthosol on high ground
and the upper parts of slopes, Cambisol on the middle
and lower parts, and Arenosol and Gleysol in valley
bottoms. We distinguished between two classes: high
ground (Ferralsol, Plinthosol) and valley (Cambisol,
Arenosol, Gleysol).
We measured the length of each trail through the
different habitat types to calculate the percentage of the
habitats on the trail. To compensate for the different
trail lengths, we used a weighting factor when calculat-
ing mean percentages and confidence limits. We con-
sidered the average trail length (5.1 km) as the standard
sample unit. The weighting factor (sample unit) of a
trail was therefore not 1 but the length of the given trail
divided by the mean length of the 16 trails. We consid-
ered that elephants selected or avoided a habitat if the
actual (expected) percentage of the habitat in the ele-
phant range was below or above the confidence inter-
val of the mean percentage that elephants used the habi-
tat. To define the elephant range, we drew a minimum
convex polygon around the elephant trails and deter-
mined the actual percentage of each habitat type in the
area of the minimum convex polygon.
Results
The percentage of sectors with elephant dung on
transects varied little between months (SD = 2%) but
Table 1. Percentages of 250-m sectors with forest elephant dung on 16 transects, each 2 km long, in the
Bossematié Forest Reserve, November 1993 to April 1994
Transect November December January February April Mean
1––––00
20000255
3 – – – – 13 13
4 – – – – 13 13
5000000
61300003
7 0 0 38 0 13 10
8002525010
9000000
10 – – – – 13 13
11 – – – – 13 13
12 38 25 0 13 13 18
13 – – – – 13 13
14 0 0 0 25 0 5
15––––00
16––––00
all 6 3 8 8 7 7

40 Pachyderm No 30 January–June 2001
more between transects (SD = 6%), which reflected
the heterogeneous distribution of elephants in the for-
est (table 1). The percentage of sectors with elephant
dung increased with the distance from the middle of
a transect to the nearest border of the forest (fig. 1a).
The linear relationship was, however, not significant
(P = 0.067), because there were no elephant drop-
pings on transects in the northern part of the study
area in which forestry work was concentrated at the
time of the study. When we excluded the northern
transects (1, 5 and 9) from the regression analysis,
the linear relationship between the distance to the
forest border and the percentage of sectors with el-
ephant dung became significant (P = 0.008). In 1992,
when no forestry work was under way in the north-
ern region, the probability of observing elephant
tracks at waterholes in the northern region was de-
pendent (P < 0.001) on the distance to the forest bor-
der (fig. 1b).
The percentage of sectors with elephant dung also
increased with increasing distance to the nearest vil-
lage when we excluded the northern region (P=
0.017) but not when we included all transects in the
analysis (P= 0.246). There was no linear relation-
ship between the percentage of sectors with elephant
dung and the distance to the nearest forest road (P=
0.988).
We grouped transects into three zones to draw a
map of the forest presenting an estimate of elephant
distribution (fig. 2). It appeared that in 1993 and 1994,
humans reduced the area available to the elephants to
130 km+ (60% of the forest) in the core of the forest
(fig. 2). We found six carcasses of elephants poached
before 1990, which were all in the centre of the for-
est in a mean distance of 600 m from a main forest
road (range 500–800 m).
In the dry season, elephants walked more than ex-
pected in areas where the canopy cover was higher
than 20% and less than expected in more open forest
parts (fig. 3). Forest elephants tended to use areas with
a high canopy cover more often in the dry season than
in the wet season. In the wet season, elephants used
open forest (< 10% cover) more than expected. In
both dry and wet seasons, elephants strongly avoided
plantations (outside 99.8 and 98.6% confidence in-
terval, respectively).
High ground and valleys accounted for 78% and
22% respectively of the minimum convex polygon
surface. During the wet season, only 13% of the ele-
phant trails were in valleys, which was less than ex-
pected, whereas they walked more on high ground
(87%) than expected. The actual percentages of high
ground and valleys were both outside the 98.3% con-
fidence interval of the mean percentages of elephant
use. During the dry season, elephants used valleys
more (35%) and high ground less (66%) than expected
but only at the 91.1% confidence interval level. At
the peak of the dry season, elephants were observed
to leave the forest on two occasions to eat banana
plants.
Figure 1. a) linear regression between the distances
from the middle of a transect to the nearest forest
border and the percentages of 250-m sectors with
elephant dung on 16 transects in 1993/1994; b)
linear regression between the distances from
waterholes to the nearest forest border and the
observation probability of elephant tracks at
waterholes in 1992 (data from Mühlenberg et al.
unpubl.). Dotted lines: all transects or all water-
holes; continuous line: excludes transects of the
northern region (1, 5 and 9). Transects and
waterholes of the northern region are shown as
open circles, those of other regions as closed
circles.
0246
0246
20
10
0
20
10
0
ay = 0.028x + 0.02
y = 0.022x + 0.02
y = 0.039x – 0.003
b
Distance to forest border (km)

Pachyderm No 30 January–June 2001 41
ies found very low elephant densities in the first 10–
20 km from settlements. In south-eastern Ivory Coast,
conditions are more difficult for forest elephants be-
cause of widespread deforestation and ubiquitous land
exploitation, which have resulted in small and frag-
mented forests. In our study, elephants avoided a zone
within 4 km of villages and reached their maximum
density in the centre of the forest, 7–11 km from
settlements. The distance to the forest border was a
better measure of human influence than the distance
to the nearest village because it reflected the human
activity around the forest, which was not limited to
villages. The forestry operations during the time of
the study also influenced elephant distribution, prob-
ably because of the long-time human presence,
whereas forest roads did not affect it. Roads, how-
ever, seemed to have facilitated poaching, as all ele-
phants poached before this study were killed near
roads.
Barnes et al. (1991) found a direct relationship
between the density of elephant dung and the per-
centage of secondary forest, which implies that ele-
phants prefer the parts of the forest with the lowest
canopy cover. Elephant preference for forest parts
with higher canopy cover in the Bossematié Forest
Reserve contradicts these findings and suggests that
Figure 3. Canopy cover classes crossed by forest
elephants compared with the actual percentages
present in the used area (minimum convex polygon
in fig. 2). The 95% confidence intervals of elephant
use indicate avoidance or preference when they are
below or above actual percentages.
Figure 2. Estimated distribution of forest elephants in 1993/1994 in the Bossematié Forest Reserve and
location of human settlements in its surroundings.
Discussion
In Central Africa, where elephants range over large
forests, their density is directly related to the distance
from human settlements or major roads, according to
Barnes et al. (1991) and Hall et al. (1997). Both stud-
minimum convex polygon around the
elephant trails
roads and border of the forest
transects
waterholes controlled in 1992
human settlements
low elephant density
area that elephants used before forest
operations began
mean elephant density
high elephant density
1
> 30 20-30 10-20 < 10 plantations
Canopy cover (%)
40
20
0
dry-season
crossing
wet-season
crossing 95% confidence
intervals
actual percentages
of cover

42 Pachyderm No 30 January–June 2001
the forest has been so heavily exploited that it is now
too degraded to offer optimal conditions for the ele-
phants. In Taï National Park of Ivory Coast, Merz
(unpubl.) observed higher density of elephants in sec-
ondary forest than in primary forest but also inten-
sive use of primary forest parts because of the abun-
dance of fruit trees in the latter. In Bia National Park
of Ghana, forest elephants mainly select open areas
but eat the bark and fruits of trees in areas with high
canopy cover (Short 1981). We suggest that our ele-
phants selected forest parts that had a higher canopy
cover to obtain fruits from the remaining mature trees.
Furthermore, elephants selected shade-tolerant spe-
cies (Theuerkauf et al. 2000b), which were rare after
the heavy exploitation.
Vanleeuwe and Gautier-Hion (1998) found that the
foraging paths of elephants were mostly in forest parts
with a canopy cover of 80%, as this provided the ele-
phants with herbaceous plants and fruits. As in our
study, those elephants avoided dense forest parts with
an open canopy. The elephants they studied, however,
fed regularly in clearings. During our fieldwork, we
noticed no preference for clearings, which in the study
area are usually covered by the introduced neotropical
Asteraceae Chromolaena odorata, an invasive plant
without importance in the elephants’ diet. Elephants
in our study area also avoided plantations, which were
mainly cocoa and coffee and provided them with little
food.
It is likely that the selection of open or closed parts
of a forest is also influenced by weather. In the dry
season, the preference was for closed canopy, per-
haps because exposure to the sun would increase water
loss. The elephants’ avoidance of valleys in the wet
season may be a result of the poorer soils (Gerold
and Hetzel unpubl.), which could bear vegetation of
lower nutritional value than that found on high ground.
In the dry season when water is scarce on high ground
and soils in valleys remained moister than on high
ground, elephants used valleys more frequently. Al-
though the habitat use of elephants changed with the
seasons, this did not affect the distribution of elephants
in the forest. Waterholes are numerous in the study
area and elephants use them often in the dry season
(Theuerkauf and Ellenberg 2000). Fruits of different
trees (for example, Desplatsia chrysochlamys) as well
as crop plants (for example, banana) supplement the
water requirements of elephants during the driest
months (January to March).
We conclude that the heavy timber exploitation and
plantation of coffee and cocoa have lowered the habi-
tat quality of the Bossematié Forest Reserve for the
elephants. We have suggested that the optimal habitat
structure for forest elephants may be found in forests
with a canopy cover denser than in our study area.
We therefore provided a list of 25 food plants of par-
ticular importance to the elephants. This list was in-
cluded in the management plan for the Bossematié
Forest Reserve and the species now stand under pro-
tection during forestry operations. Large-scale forestry
operations also negatively affected the distribution of
elephants. Forest exploitation should therefore be lim-
ited to small areas and short periods to prevent el-
ephants from avoiding the areas altogether. However,
forestry activity deterred poachers and thus also had
a positive effect, which led to the high recruitment of
the elephant population observed every year since the
protection of the Bossematié Forest Reserve (W.E.
Waitkuwait, unpubl. data, 1998).
Acknowledgements
This study was part of the pilot project Réhabilitation
de la Forêt Classée de Bossematié under the
programme Réhabilitation des forêts à l’est de la Côte
d’Ivoire, a collaboration between the Société de
Développement des Plantations Forestières
(SODEFOR), the Deutsche Gesellschaft für
Technische Zusammenarbeit (GTZ) and the
Kreditanstalt für Wiederaufbau (KfW). We thank H.-
J. Wöll, H. Fickinger, N. Seabé and the SODEFOR
team, who took care of financial and logistical aspects;
J. Slowik, H. Plachter and the late H. Remmert for
cooperation; and S. Rouys for useful comments on
earlier drafts of this article. We particularly thank Y.
Guiro for sharing his elephant tracking experience.
References
Alexandre, D.Y. (1978) Le rôle disséminateur des éléphants
en forêt de Taï, Côte d’Ivoire. Terre et Vie 32, 47–72.
Barnes, R.F.W. (1999) Is there a future for elephants in West
Africa? Mammal Review 29, 175–199.
Barnes, R.F.W., Barnes, K.L., Alers, M.P.T., and Blom, A.
(1991) Man determines the distribution of elephants in
the rain forests of northeastern Gabon. African Journal
of Ecology 29, 54–63.
Gerold, G., and Hetzel, F. (unpubl.) Bodenkarte des Forêt
Classée de la Bossematié. 1993 map. University of
Göttingen.

Pachyderm No 30 January–June 2001 43
Hall, J.S., Inogwabini, B.I., Williamson, E.A., Omari, I.,
Sikubwabo, C., and White, L.J.T. (1997) A survey of
elephants (Loxodonta africana) in the Kahuzi-Biega Na-
tional Park lowland sector and adjacent forest in eastern
Zaire. African Journal of Ecology 35, 213–223.
Kientz, A. (unpubl.) Participation de la population riveraine
à la protection de la Forêt Classée de Bossematié.1993
report. Deutsche Gesellschaft für Technische Zusam-
menarbeit.
Merz, G. (unpubl.) Untersuchungen über Lebensraum und
Verhalten des afrikanischen Waldelefanten im Taï-
Nationalpark der Republik Elfenbeinküste unter dem
Einfluß der regionalen Entwicklung. PhD thesis, 1982.
University of Heidelberg.
Merz, G., and Hoppe-Dominik, B. (1991) Distribution and
status of the forest elephant in the Ivory Coast, West
Africa. Pachyderm 14, 22–24.
Mühlenberg, M., Slowik, J., and Waitkuwait, W.E.
(unpubl.) Waldrehabilitierung in der Ostregion der Côte
d’Ivoire. Bericht über die ökologischen Begleitmaß-
nahmen in der Projektphase 1991–1993. 1993 report.
Deutsche Gesellschaft für Technische Zusammenarbeit.
Prins, H.H.T., and Reitsma, J.M. (1989) Mammalian bio-
mass in an African equatorial forest. Journal of Animal
Ecology 58, 851–861.
Sabenin, Y., and Kra, Y. (unpubl.) Etude expérimentale de
la stratification dans la Forêt Bossematié. 1993 map.
IGCI/IGT.
Schroth, G. (unpubl.) Rapport de mission pour le projet
Stabilisation des systèmes de production agricole dans
la région d’Abengourou. 1992 report. Deutsche
Gesellschaft für Technische Zusammenarbeit.
Short, J. (1981) Diet and feeding behaviour of the forest
elephant. Mammalia 45, 177–185.
Struhsaker, T.T., Lwanga, J.S., and Kasenene, J.M. (1996)
Elephants, selective logging and forest regeneration in
the Kibale Forest, Uganda. Journal of Tropical Ecology
12, 45–64.
Theuerkauf, J., and Ellenberg, H. (2000) Movements and
defecation of forest elephants in the moist semi-decidu-
ous Bossematié Forest Reserve, Ivory Coast. African
Journal of Ecology 38, 258–261.
Theuerkauf, J., Ellenberg, H., and Guiro, Y. (2000a) Group
structure of forest elephants in the Bossematié Forest
Reserve, Ivory Coast. African Journal of Ecology 38,
262–264.
Theuerkauf, J., Waitkuwait, W.E., Guiro, Y., Ellenberg, H.,
and Porembski, S. (2000b) Diet of forest elephants and
their role in seed dispersal in the Bossematié Forest Re-
serve, Ivory Coast. Mammalia 64, 447–459.
Vanleeuwe, H., and Gautier-Hion, A. (1998) Forest elephant
paths and movements at the Odzala National Park,
Congo: the role of clearings and Marantaceae forests.
African Journal of Ecology 36, 174–182.
White, L.J.T., Tutin, C.E.G., and Fernandez, M. (1993)
Group composition and diet of forest elephants,
Loxodonta africana cyclotis Matschie 1900, in the Lopé
Reserve, Gabon. African Journal of Ecology 31, 181–
199.
Wöll, H.-J. (1992) Le projet Rehabilitation de la Forêt
Classée de Bossematié, Côte d’Ivoire, région d’Aben-
gourou. Tropenbos Series 1,163–179.