Content uploaded by Vincent Raphael Nyirenda
Author content
All content in this area was uploaded by Vincent Raphael Nyirenda on Jan 30, 2015
Content may be subject to copyright.
Journal of Ecology and the Natural Environment Vol. 4(7), pp. 186-193, April 2012
Available online at http://www.academicjournals.org/JENE
DOI: 10.5897/JENE11.142
ISSN 2006 - 9847©2012 Academic Journals
Full Length Research Paper
Population status of the African elephant in Zambia
Chansa Chomba1*, Chuma Simukonda2, Vincent Nyirenda1 and Francesca Chisangano3
1Zambia Wildlife Authority, Directorate of Research, Planning, Information and Veterinary service P/B1,
Chilanga Zambia.
2Zambia Wildlife Authority, Department of Research P/B 1 Chilanga, Zambia.
3Zambia Wildlife Authority, CITES office, P/B 1 Chilanga, Zambia.
Accepted 15 February, 2012
Prior to the 1960s Zambia had an estimated elephant population of over 250, 000, but by 1989 it had
fallen to about 18 000 individuals. After about 10 years of improved law enforcement operations and the
involvement of local communities, populations stabilized and even started to increase by 1996. In order
to update and compare population estimates, aerial surveys using similar methods used in the past
were conducted between October 17 and December 5, 2008. Methods used were aerial sample counts,
total counts and block counts in seven ecosystems; Nsumbu-Mweru wa Ntipa – Lusenga, Bangweulu,
Luangwa Valley, Lower Zambezi, Upper Zambezi, Kafue and Chete-Sikula Islands. The national
population estimate was 27, 529 an increase of 18% over the 1996 estimate. Of this population estimate
20, 200 (73%) were in the Luangwa, 3348 (12%) in Kafue, 2464 (9%) in Upper Zambezi and 1299 (5%) in
Lower Zambezi ecosystems. The Luangwa Valley was thus the most important elephant habitat in
Zambia. The current survey omitted West Lunga ecosystem which should be surveyed in future.
Ground census should be introduced in the Nsumbu ecosystem because the itigi thickets impede
visibility from the air and thus aerial counts may yield under estimates. It is also being recommended
that some populations be translocated from well populated areas to ecosystems where populations are
still too low so as to speed up population recovery from poaching depression.
Key words: Aerial surveys, estimate, trend, distribution, Luangwa, Kafue.
INTRODUCTION
Zambia had one of the largest elephant (Loxodonta
africana) populations south of the Sahara (Anon, 1961).
Prior to the late 1960s it was estimated that the elephant
population was over 250, 000 (Anon, 1961) and wide
spread (Ansell, 1960). Survey results supported by
anecdotal observations by field staff indicated a preci-
pitous decline of the population between 1976 and 1986
mainly due to excessive poaching (Chansa and
Kampamba, 2005). By 1989 for instance, the numbers
had fallen from about 250,000 to 18, 000 individuals
(Anon, 1989). The species was later upgraded to
Appendix I of the Convention on International Trade in
Endangered Species of Wild Fauna and Flora (CITES) in
*Corresponding author. E-mail: Chansa.chomba@zawa.org.zm
or ritachansa@yahoo.com.
1989 to save it from local extermination. Under CITES
transferring a species to Appendix I restricts international
trade and it is assumed that since poaching of elephant is
driven by international trade in ivory, placing elephant in
Appendix I of CITES would minimise poaching pressure.
From the 1990s, the government of Zambia with the help
of international partners increased investments in law
enforcement activities. In the Luangwa Valley for
instance, the Norwegian Agency for International
Development (NORAD) invested substantial amounts of
financial resources from the mid 1980s to date. With this
consistent financial support from NORAD, the elephant
population stabilized and even showed trends of
increase.
National surveys conducted by NPWS staff in 1992,
1994 and 1996 (Anon, 2010) estimated 22,467 in 1992,
21,000 in 1994 and 22,518 individuals in 1996 (Anon,
2002). Field reports submitted to headquarters by NPWS
Chomba et al. 187
Figure 1. Location of the elephant survey areas in 2008, Zambia.
staff from Luangwa and Kafue ecosystems indicated a
preponderance of juveniles and sub adults which was
indicative of an increasing population. In 2008
government provided financial support to conduct a
nation wide aerial survey to update the national elephant
population estimates as a prerequisite to the preparation
of a proposal presented at the Conference of the Parties
(COP) 16 of CITES. At COP 16 Zambia presented a
proposal for approval by parties a proposal to down list
the elephant population from Appendix I which restricts
international commercial trade to Appendix II which
permits commercial trade since data from the surveys
had shown that the population was increasing and was
no longer endangered.
The main objective of the survey conducted in 2008
therefore, was to establish the status of elephants in
Zambia by determining population size, distribution and
carcass ratio as a way of estimating the impact of
poaching and documenting cross border elephant
movements. This was because Zambia had planned to
resubmit the proposal to down list the elephant
population from Appendix 1 to Appendix II which was
rejected at COP 12 for reconsideration by the parties to
CITES at COP 15.
MATERIALS AND METHODS
Study area
The survey covered seven ecosystems; Luangwa, Kafue, Upper
Zambezi, Lower Zambezi, Bangweulu, Nsumbu-Mweru wa Ntipa-
Lusenga Plains (Tanganyika) and Chete-Sikula Islands (Lake
Kariba Islands) (Figure 1). The total area covered was 166,713 km2
(22%) of the country’s land surface area and comprised 17 National
Parks, 25 Game Management Areas (GMAs) and 3 open areas
which was 69% of the country’s protected area system.
The aerial survey method used in each area was selected based
on size and terrain of survey area. Aerial sample counts were used
in large areas (≥ 100 km2) and this involved transect counts in more
or less flat terrain and block counts in mountainous terrain. For
small areas (< 100 km2) total counts were made.
Transect sample counts
Streamers were marked using two nylon ropes with cones attached
188 J. Ecol. Nat. Environ.
at one end for stability in the air, were tied to each of the struts of a
plane. The streamers were placed parallel to each other as the
aircraft flew at about 91 m a.g.l. The distance between the
streamers on each strut was arranged so that this distance
represented a strip of about 150 m wide on the ground (Norton –
Griffiths, 1978), so that a total strip width of 300 m is covered on the
ground. The survey area was divided into strata based on density
and distribution obtained from previous survey results.
Transects were laid out at right angles to a road or river (principal
physical features) and were parallel to each other. Sampling
intensity ranged from 10 to 15% so that sampling effort could be
allocated between strata based on animal density. Strata with high
variance received more sampling effort than strata with low
variance. Two fixed high wing six seater Cessna 206 and Cessna
207 aircrafts fitted with Garmin 296 GPS and Radar altimeter were
used as described by Bell and Grimsdell (1973), Howard, Jeffery
and Grimsdell (1984) and modified by Chansa and Kampamba
(2009) by using GPS and applying it to plot spatial distribution of
animals. Transects were flown at a nominal (theoretical) height of
91 m which gave a width of 300 m on the ground and cruising at an
average speed of 160 km/hour which were also used in the
previous surveys and made the data comparable.
The flying height of 91 m above ground was directly proportional
to the calculated width of 300 m on the ground. In defining the
sampling strip width, a pair of parallel nylon ropes stabilized by
plastic funnels at the end of each nylon rope was attached to the
wing strut on each side of the aircraft. The strip width was
determined by the distance between nylon ropes and the flying
height. Wider parallel nylon ropes and higher flying height would
give wider sampling strips and vice versa. Having decided on the
flying height and strip width, the nylon ropes were placed in their
correct spacing. The area observed at 91m was calculated based
on the formula developed by Norton-Griffiths (1978) as follows:
W = W.h/H
where;
h - is the height of the observers’ eye from the ground,
W - is the required strip width, and
H - is the required flying height.
The aircraft was propped into flying position. The observer then sat
in a position that was comfortable for counting and an eye position
was chosen. The position of the inner nylon rope was selected. The
chosen position was marked in such a manner that the observer’s
line of sight would be clear of the wheel of the aircraft but close to
the aircraft’s body. The lower window marker code named ‘a’ was
then fixed such that it was in line with the eye position and inner
nylon rope.
A mark code named ‘A’ was placed on the ground in line with
marker ‘a’ and the inner nylon rope. The height of the observer’s
eye code named ‘h’ was measured from the ground. A second
marker code named ‘B’ was placed on the ground and the distance
between ‘A’ and ‘B’ was denoted by ‘w’ and calculated based on the
formular indicated above.
In order to ensure that the strip width calculated above is correct,
aerial calibration was carried out. Markers were laid out on flat
ground or an airstrip at intervals of 20 m. The aircraft then flew at
the required height passing at right angles to the markers. The
observers then counted and photographed the number of markers
between the nylon rope streamers.
The strip width was then calculated as follows:
W= w.H/h
where;
h=average height of aircraft,
w=average strip width,
H=selected census flying height, and
W=nominal strip width at H.
In this case, the selected flying height was 91 m. Flying higher than
91 m could have made transects wider than the calculated width of
300 m leading to an over estimate of numbers and vice versa.
While flying along the transects, observers kept their heads in a
constant position with reference to a masking tape marking on the
window to avoid parallax error. The Front Seat Observer (FSO)
recorded flight height from the Radar altimeter every 30 seconds.
The right seat observer sat just behind the pilot and the left seat
observer behind the recorder, and communicated through an
intercom system.
The FSO called out to the rear seat observers the start and end
of each transect. Each rear seat observer recorded elephant
observations within the strip width on each side of the plane. At
each sighting, the observer called out to the recorder for the
location way point, which the observer entered into the dictum
recorder and sightings made were also recorded. The Recorder
also entered the way-point into the hand-held GPS. Groups of
animals greater than 20 oblique or near-vertical photographs were
taken using a hand held camera with high resolution camera and
the number of frames of pictures taken was recorded on the dictum
as described by Grimsdell (1978). Recorded observations were
transcribed by the observers onto data sheets immediately after
each flight. On the ground, all GPS data from the front seat
observer and photographs from the digital cameras were down
loaded into an aerial survey database.
Aerial block sample counts
Aerial block sample counts conducted were based on the method
described by Norton-Griffiths (1978) as aerial quadrat sampling.
The survey area was mapped and divided into quadrat squares of
approximately 3 km2 placed systematically to cover 10% sampling
intensity of the survey area. A super cab aircraft which can
manoeuvre at slower speed than a Cessna was used. The pilot
searched each block sequentially while the observers counted
animals observed within blocks. The average search time was three
minutes per block. More time was spent when there were more
than 20 animals in a group. Animals sighted close to the block
boundary were also counted as being within the block. In block
sample counts, streamers and radar altimeter were not necessary
because total counts were conducted, but the observer sitting
arrangements were maintained and the recorder only recorded
GPS recordings of sightings. The pilot circled each block and ended
with a diagonal line to head for the next block. Because of the small
size of the blocks only one observer counted at a time, and so the
pilot flew the other way to give chance to the other observer. Total
counts were conducted in Inangu (13.2 km2); Chete (25.5 km2),
Sekula (7107 km2) and Mosi-oa-Tunya (67.38 km2).
Population estimates from the 1992, 1994 and 1996 surveys
were also collected from Zambia Wildlife Authority archives. These
surveys used similar survey methods as in the present study
making data comparable with the 2008 estimates.
Elephant carcasses
Elephant carcasses were recorded and classified based on the
criteria developed by Douglas-Hamilton and Hillman (1981) as
shown in Table 1. Calculated elephant carcass ratios are good
indicators of population increase. For instance, carcass ratios of <
8% indicate population increase. To distinguish elephant carcasses
killed by poachers from other mortality factors was not easy. In this
study however, it was generally assumed that poached elephants
Chomba et al. 189
Table 1. Classification of elephant carcasses based on the seven ecosystem surveyed in 2008, Zambia.
Class of elephant carcass
Description of carcass class
Carcass 1
Fresh; still has flesh giving the body a rounded appearance. Vultures still present and ground still moist from body fluids.
Vegetation trampled. Carcass; probably less than 1 month old.
Carcass 2
Rot patch and skin still present. Skeleton not scattered. No Vulture droppings, no trampled vegetation. Less than 1 year
old. Probably 3 to 8 months since death.
Carcass 3
Clean bones, skin usually absent, and vegetation re-grown in rot patch. More than 1 year old.
Carcass 4
Very old, bones scattered and turning grey. Very old; carcass could be up to 10 years old.
had tusks removed while those dying from natural mortality
had tusks/ivory.
Trans boundary movements
Questionnaire surveys and personal interviews were
conducted with Zambia Wildlife Authority in the field,
community resource board members and tour operators in
elephant areas bordering neighbouring countries. Hunters
were among the most resourceful interviewees as they
have historical data on elephant movements particularly in
areas where elephants were some times killed on
retribution after destroying crops or causing human
fatalities. Key questions were formulated to know whether
the interviewee knew the international boundary between
Zambia and the
neighbouring country. Those with knowledge of the
international boundary were asked other questions relating
to whether they had seen and how long ago they saw
elephants cross the border into or out of Zambia. In areas
where protected areas existed on both sides of the
international boundary, consultation was made with staff of
the wildlife agency of that country regarding records
relating to elephant movements. In areas where Zambia
has implemented Trans frontier Conservation Areas, data
collection was easy. In areas such as the Kavango -
Zambezi (KAZA) Transfrontier Conservation Area, where
elephants are fitted with GPS collars for monitoring
purposes, information in form of maps made after down
loading the data were collected and used to document
common elephant movement routes.
Data analysis
Transect sample counts
Transect sample counts data was analyzed using an Excel
based application (Jolly, 1969) method 2 for unequal sized
sample units. The estimate was calculated from the
product of the overall density of animals counted per unit
area and the census zone of stratum based on the formula:
Ŷ = Z. Ŕ
where;
Z= Area of the census zone, and
Ŕ= is the ratio of all the animals counted to the whole area
searched, thus
an estimate of the overall density of animals per unit area.
Block counts
Block counts’ data was analyzed using Jolly’s method 1 for
equal sized sampling units. This method was used
because blocks were almost of the same size and shape
except for those on the edges of strata. The population
estimate was calculated from the average number of
animals counted in each sampling unit based on the
formula by Jolly (1969):
Ŷ = Population estimate = N. Ỹ
where;
N = the total ‘possible’ number of block units in the
population
Ỹ = the sample mean =∑Y ∕ n
Y= population size of surveyed area
n= sample
Elephant carcass ratio
The elephant carcass ratios were calculated for each
ecosystem based on the method described by Douglas-
Hamilton and Hillman (1981), where the estimated
carcasses were divided by the sum of all carcasses and
estimated population.
RESULTS
The total area surveyed was 166,712.5 km2 which
was 22% of Zambia’s land surface and 69% of the
country’s protected area system. The 69%
protected area surveyed contained 17 National
Parks, 25 Game Management Areas and 3 were
Open areas, (Lundazi, Sikongo and Kazungula).
190 J. Ecol. Nat. Environ.
Table 2. Elephant population estimates and carcass ratio for the seven ecosystem surveyed in 2008, Zambia.
Ecosystem
Population estimate
Carcass ratio details
95 % CL
Lower limit
Upper limit
Carcass ratio
(%)
Significance Level**
Nsumbu-Mweru wa Ntipa-
Lusenga (Tanganyika)
27
47
0
74
0
NA
Bangweulu
136
137
0
NA
Luangwa Valley
20,200
3,592
17,129
24, 313
1.57
Low
Lower Zambezi
1299
860
439
2159
14.4
High
Upper Zambezi
2464
2239
225
4703
3.2
Low
Kafue
3348
933
2415
4281
0.9
Low
Chete-Sikula (Kariba)
55*
0
NA
*Total count; ** Low if below recommended 8% and high if above it.
West Lunga ecosystem was not surveyed because of
insufficient financial resources.
The national population estimate was 27,529 ± 4405
elephants distributed in four major elephant ecosystems;
Luangwa, Kafue, Upper (including Mosi oa Tunya-
Kazungula-Sioma areas) and Lower Zambezi Out of the
total estimate 20,200 (73%) were in the Luangwa
ecosystem, 3348 (12%) in the Kafue ecosystem, 2464
(9%) in Upper Zambezi and 1299 (5%) in the Lower
Zambezi area (Table 2). The Tanganyika ecosystem had
27 elephants and was the lowest. In the Bangweulu
ecosystem, there were 136 elephants all sighted in
Kansaka National Park and 54 were at Chete and Sikula
islands on Lake Kariba. The largest number of elephants
were in luangwa and Kafue which were also the largest
ecosystem (Figure 2).
In comparison with previous surveys, the 2008 estimate
showed an increase of 18% over the previous estimate in
1996 Regression analysis showed a positive increase (y
= 1670 4 x + 19203; R2 = 0. 5705) (Figure 4).
Carcasses
The total number of carcasses estimated was 75 with 3%
as mean carcass ratio at national level. Of the 75, only an
estimated number of 20 (25%) were in category 1. North
Luangwa National Park in the Luangwa Valley had the
lowest carcass ratio of 0.1%, while the South Luangwa
National Park had 0.6 % which indicated low elephant
mortality in both National Parks. Musalangu GMA had
13.1% which was the highest carcass ratio of all PAs in
the Luangwa Valley ecosystem. According to Douglas-
Hamilton and Hillman (1981), the figure of 13.1% was
higher than the recommended upper limit of 8%. Of all
carcasses recorded in the Luangwa Valley, 76% were
found in West Musalangu GMA suggesting that poaching
was high in this area. The mean carcass ratio for the
Luangwa Valley ecosystem however, was only 1.57and
was so poaching was also assumed to be low. This
assumption is based on the principle that high level
poaching is when the carcass ration is ≥ 8% (Douglas-
Hamilton and Hillman, 1981).
The highest incidences of elephant carcasses were in
Lower Zambezi system which had 14.37% while upper
Zambezi had 3.2%. West Petauke GMA which was
considered to be part of the Lower Zambezi ecosystem
had 7% while Kafue had a low carcass ration of 0.9%.
Bangweulu, Chete and Sikula islands on Lake Kariba and
the Tanganyika-Nsumbu-Lusenga Plains ecosystem
recorded no carcass incidence (Table 2). Of all the
carcasses recorded during the survey, 62% were in
Lower Zambezi National Park, implying that this was the
most affected ecosystem by poaching. Generally,
poaching was low across all the ecosystems, except for
the Lower Zambezi ecosystem where the carcass ratio
was higher than 8%.
Tran boundary movements
There were seven routes where elephants were recorded
to cross international boundaries into neighbouring
countries. These were; West Lunga – Congo DR route,
Kasanka/Bangweulu – Congo DR route, Tanganyika –
Congo DR route, Nyika (Zambia) – Nyika (Malawi) route,
Lukusuzi (Zambia) – Vwaza (Malawi) route, Lower
Zambezi (Zambia) – Mana Pools (Zimbabwe) route, and
Sioma – Namibia/Botswana route (Figure 3).
DISCUSSION
The elephant population in Zambia has recovered from
the poaching pressure of the 1970s to late 1980s and is
increasing (Figure 4). The elephant population estimate
of 27,529 is perhaps lower than the true population as
some areas such as the West Lunga ecosystem was not
surveyed, and the Nsumbu area had an under estimate.
The itigi thicket in Nsumbu-Mweru was Ntipa complex for
Chomba et al. 191
Figure 2. Major elephant ecosystems and elephant distribution in 2008,
Zambia.
instance, makes aerial surveys inappropriate for the area
due to impaired visibility. Ground counts and use of dung
count would yield a higher estimate in this ecosystem
(Jachmann, 2001).
The Luangwa Valley was the most important elephant
range in Zambia, a pattern which was also recorded in
the 1960s (Chansa and Kampamba, 2005).A comparison
of population estimates between National Parks and
Game Management areas showed a highernumber in
National Parks than Game Management Areas. This
may be attributed to higher security and absence of
human settlements in National Parks. In Game
Management Areas, human encroachment heightened
human – elephant conflicts and many communities did
not tolerate elephants in their vicinity. Perhaps this is one
of the reasons why Musalangu GMA recorded the highest
elephant carcass ratio (13.1%) in the entire Luangwa
Valley.
The low carcass ratio in the Upper Zambezi ecosystem
could be a result of improved law enforcement
surveillance after the establishment of the Kavango-
Zambezi (KAZA) Transfrontier Conservation Area.
Regarding elephant carcass ratios at national level,
Douglas-Hamilton and Hillman, (1981) indicated that
stable or expanding populations show elephant-carcass
ratio ranging from 2 to 8% while decreasing populations
have ratios greater than 8%. This implies that the
elephant population in Zambia is generally on the
increase except for the Lower Zambezi area where the
carcass ratio was higher than 8%.
The transfrontier movement routes recorded during this
survey were attributed to the country’s central location
where it shares borders with eight countries. Since most
ecosystems are large (>10 000 km2) in extent, it becomes
inevitable for elephant range to coalesce and sprawl with
those of the neighbouring countries. In fact the concept of
192 J. Ecol. Nat. Environ.
Figure 3. Trans-boundary elephant routes in 2008, Zambia.
Transfrontier Conservation Areas in the southern African
sub region was based on this observation.
Conclusion
The most elephant populated range in Zambia was the
Luangwa Valley, currently holding more than 75 % of the
national population. This population is largely restricted to
Zambia except for the limited movements that may occur
between Zambia and Kasungu National Park in Malawi.
The National Parks were the most secure habitats for
elephants in Zambia and the long term survival of
elephants in Zambia will depend to a large extent on the
protection of elephant populations inside National Parks
and transfrontier areas. The establishment of a
Transfrontier Conservation area between Lower Zambezi
National Park in Zambia and Mana Pools National Park in
Zimbabwe is recommended as this will enable the two
countries to work together in curbing poaching incursions
which are currently high as reflected in the high carcass
ratio of 14.1%.
Regarding law enforcement, Zambia Wildlife Authority
needs to upscale operations and where possible consider
population supplementation in the Nsumbu-Mweru wa
Ntipa – Lusenga Plains ecosystem. Such effort would
speed up the recovery from the poaching depression of
the 1970s-1980s of the remnant elephant. The
Bangweulu ecosystem also requires similar conservation
measures as the population is also still low at 136
individuals. In terms of population estimates, the
Nsumbu-Mweru wa Ntipa ecosystem which has itigi
thicket that impairs visibility from the air, ground counts
and dung counts based on the methods described by
(Kangwana, 1996) will be required to compare estimates
obtained from aerial techniques.
ACKNOWLEDGEMENTS
This survey could not have taken place without financial
support of the Government of the Republic Zambia. We
Chomba et al. 193
Figure 4. Elephant population trends since 1992, Zambia.
are also indebted to Dr. Colin Craig for his technical
guidance during the planning and data analysis phases.
The survey costs for the upper Zambezi ecosystem were
funded by the African Wildlife Foundation – Kazungula
Heartland. The Flying Mission Zambia Ltd and ZAMFARI
Zambia Ltd. provided aircrafts and committed staff at
short notice. Several members of staff of Zambia Wildlife
Authority provided input at various levels. These were;
Twakundine Simpamba, Edward Chilufya, Chaka
Kaumba, Lisa Mwiinga, Shantanda Shabwela, Jassiel
Msoka, Jarton Shawa, Rhoda Kachali, Jeremy Sikazwe,
Priscilla Mkanga Sichone, Daniel Mwizabi, Benson
Kabungo, Bryson Thole and Davison Mumba.
REFERENCES
Anon (2010). Proposal to down list the African Elephant population of
Zambia from Appendix I to Appendix II of CITES. Zambia Wildlife
Authority, Chilanga.
Anon (1961). A Wildlife Policy for Northern Rhodesia. Ministry of Native
Affairs, Lusaka.
Anon (1989). Operations of the Save the Rhino Trust. Lusaka.
Ansell WFH (1960). Mammals of Northern Rhodesia, Lusaka.
Bell RHV, Grimsdell JJR (1973). The persecuted black lechwe of
Zambia. Orxy., 12: 77-92.
Chansa W, Kampamba G (2005). The National elephant conservation
plan. Delux Printers, Lusaka. pp. 1-6.
Chansa W, Kampamba G (2009). The population status of the Kafue
lechwe in the Kafue Flats, Zambia. Afr. J. of Ecol., 48: 837-840
Douglas-Hamilton I, Hillman A (1981). Using Elephant Carcasses and
Skeletons as indicators of Population Trends in Low-Level Aerial
Survey techniques. ILCA Monograph.
Howard GW, Jeffery RHV, Grimsdell JR (1984). Census and population
trends of balck lechwe in Zambia. Afr. J. Ecol., 22: 175-179.
Jackmann H (2001). Estimating abundance of African Wildlife. Kluwer
Academic Publishers, London.
Jolly GM (1969). Sampling Methods for Census of Wildlife Populations.
E. Afr. Agric. For. J., 34: 46-49.
Kangwana K (1996). Studying elephants. African Wildlife Foundation
Technical Hand Book Series 7. Nairobi. pp. 20-37.
Norton-Griffiths M (1978). Counting Animals. African Wildlife Leadership
Foundation, Nairobi.
