Content uploaded by Simon Hedges

Author content

All content in this area was uploaded by Simon Hedges

Content may be subject to copyright.

31

Gajah 28 (2008) 31-40

Modelling Impacts of Poaching on the Sumatran Elephant Population

in Way Kambas National Park, Sumatra, Indonesia

Arnold F. Sitompul1,2, John P. Carroll3, James Peterson4 and Simon Hedges5

1Conservation Science Initiative, Medan, Indonesia

2Department of Natural Resources Conservation, University of Massachusetts, Amherst, MA, USA

3Warnell School of Forestry and Natural Resources, University of Georgia, Athens, GA, USA

4U.S.G.S. Georgia Cooperative Fish and Wildlife Research Unit, Warnell School of Forestry

and Natural Resources, University of Georgia, Athens, GA, USA

5Wildlife Conservation Society, International Programs, Bronx, New York, NY, USA

Introduction

Poaching has been known to have a large impact

on elephant populations in both Africa (e.g.

Douglas-Hamilton 1987; Poole & Thomsen

1989) and Asia (Sukumar 1989; Sukumar et al.

1998). There are fears that poaching of Asian

elephants has increased since CITES approved

an experimental one-off sale of ivory from

Botswana, Namibia, and Zimbabwe to Japan

in July 1999, following compliance with a

number of agreed conditions. Another one-off

sale from South Africa, Namibia, and Botswana

was approved in 2002 but that sale has not yet

taken place (CITES 2000; Milliken 2004). In

Sumatra, during the 1980s and 1990s, poaching

was not considered a major threat to elephants

(Blouch & Haryanto 1984; Blouch & Simbolon

1985; Santiapillai & Jackson 1990); however it is

feared that poaching activity has increased since

year 2000 (Sitompul et al. 2002; Hedges et al.

2005). While poaching activity is predicted to

continue increasing, accurate data on poaching is

very difﬁ cult to obtain. Furthermore, there have

been no ﬁ eld studies in Sumatra identifying the

impact of poaching on elephant abundance and

population trends.

Population modelling has been widely used in

wildlife ecology studies for many terrestrial large

mammals (e.g. Belovsky 1987; Berger 1990;

Rothley et al. 2005). Incorporating modelling

approaches as part of adaptive management

strategies, allows managers to develop more

effective conservation strategies (Cromsigt

et al. 2002) while reducing the uncertainty

about how the system responds to management

actions (Williams et al. 2002). Furthermore,

modelling allows managers to make an empirical

assessment of the species of interest and identify

and implement the management strategies that

are most likely to increase the probability of a

species persisting over a given time period.

However, developing detailed and accurate

population models for many species requires

extensive historical baseline data (i.e., population

size, age structure, sex-ratio, fecundity rate, and

natural survival and mortality rates). In Sumatra,

reliable baseline data for Sumatran elephant is

uncommon; however the results of a couple of

studies (Riley 2002; Hedges et al. 2005) provide

reliable data for the elephant population in Way

Kambas National Park. We believe that modelling

of elephant populations and poaching threats will

help managers identify key parameters to monitor,

and strategies to adopt, in order to minimize

extinction threats for Sumatran elephants.

In this paper, we estimate the potential impact

of poaching on the elephant population in Way

Kambas National Park (WKNP) using a stochastic

population model. We projected the population

trend under three different poaching scenarios:

no poaching, low poaching, and high poaching.

For each model, we predicted the population’s

age distribution, growth rate, and trends in

abundance estimates over 50 years. Finally, we

calculated the extinction probability for each

scenario and conducted sensitivity analyses to

identify the parameter that had the largest effect

on the model’s estimates.

32

Methods

Study area

Field data used in the model were collected in

Way Kambas National Park (WKNP), Sumatra,

Indonesia. WKNP is located in eastern part of

Lampung Province in south-eastern Sumatra

(4o62’–5o26’ S and 105o54’–105o90’ E), and

is 1235 km2 in area The entire park is < 50 m

above sea level and annual rainfall is 2000–

3000 mm. Vegetation types are typical tropical

lowland and swamp forest. Most of the park was

logged in the 1960s and 1970s, so most of the

forested area in the park is relatively degraded.

Nonetheless, the park has still been categorized

as the second highest priority for Sumatran

elephant conservation (Santiapillai & Jackson

1990). The park boundary is approximately 227

km long and 65% (148 km) of it is bordered by

34 villages. The elephant population in the park

was estimated to be 180 (95% CI = [144, 225])

in 2002 (Hedges et al. 2005). The government

of Indonesia established an Elephant Training

Centre (ETC) in the south-eastern area of the

park in the early 1980s; the purpose of this ETC

was to house “problem elephants” captured as

a result of human–elephant conﬂ ict and habitat

conversion in WKNP and other parts of Lampung

Province (Hedges et al. 2005). The “problem

elephants” were then tamed and trained at the

ETC for tourism purposes. The ETC in WKNP

is the largest such centre in Sumatra and during

2000–2002 was known to contain about 100

elephants (authors’ pers. obs.).

Methods

We developed a stage-based stochastic population

model to determine the impact of poaching in

the park based on known rates of illegal killing

of elephants in WKNP (Sitompul et al. 2002).

Population trajectories and maximum population

size under different scenarios were predicted

for elephants in WKNP using a Leslie matrix

projection model (Leslie 1945, 1948). The model

consisted of four different life-history stages:

calf, juvenile, subadult, and adult and operated

on an annual time step basis (Fig. 1). The calf

stage included any elephant <1 year old, juveniles

included ages 1–5 years, subadult elephants

included individuals >5–15 years old, and adults

included individuals >15 years old (Sukumar

1989). Each simulation began by assigning

individuals to one of the four life history stages:

calves were 8.04% of the population, juveniles

were 28.57%, subadults 50%, and adults 13.39%,

based on the demographic conﬁ guration of the

elephant population in WKNP in Reilly (2002).

The number of calves produced each time step

was a function of the number of adults and sub-

adults and fecundity. Stage-speciﬁ c maximum

annual fecundity rate was assumed to be constant

over time and estimated to be 0.225 for both

subadult and adult elephants, and was based on

long-term studies of Asian elephants in other

regions (Sukumar 1989). Stage-speciﬁ c natural

survival rate was assumed to be similar to

Asian elephants in India and averaged 0.85 for

the calf, 0.96 for the juvenile, 0.98 for the sub-

adult, and 0.85 for the adult life history stages.

We incorporated stochasticity into the model by

randomly generating annual survival rates from

a beta distribution with the mean speciﬁ ed above

and a standard deviation that was 10% of the

mean.

Figure 1. Model ﬂ ow for population estimation

and demographics as a function of recruitment,

survival and poaching for elephants projected for

50 years in Way Kambas National Park.

33

For each simulation scenario, we ran 1000

replicate simulations for a 50 year time period,

and observed the ﬁ nal population structure at

year 50. Mean and 95% conﬁ dence interval (95%

CI) of population size, population structure, and

population growth rate (λ) were calculated. In

addition, a quasi-extinction coefﬁ cient (EC) was

estimated as the proportion of the 1000 replicate

simulations that resulted in extinction before 50

years.

We evaluated the effect of poaching on elephant

populations using three different scenarios.

The ﬁ rst scenario, which we called the control,

assumed that the elephant population in the park

was fully protected, resulting in no anthropogenic

removal of elephants (no poaching and elephant

capture due to conﬂ ict with human). The second

scenario assumed poaching occurred at a low rate

deﬁ ned as the mean number of elephants known

to have been removed from the population per

year due to poaching over the years 2000–2004.

The number of elephants poached in the park

was estimated from the total number of carcasses

with signs of poaching activity found in the

park in the 2000–2002 period (n=8 elephants)

plus 8 elephants that had been found killed by

poachers in the 2003–2004 period (Sitompul et

al. 2002; Hedges et al. 2005; WCS unpub. data).

We assumed only sub-adult and adult elephants

were poached. The third scenario assumed that

high poaching would occur in the park based on

continued human population growth and land use

trends in Lampung Province. High poaching was

deﬁ ned as a 2x increase on the previously deﬁ ned

low poaching rate described above. Because the

relationship between poaching and population

size is unknown, we modelled poaching rates as a

function of population size using four alternative

functions: (1) poaching was constant over time;

(2) poaching was a negative linear function of

population size; (3) poaching was an exponential

decay function of population size; and (4)

poaching was a logistic function of population

size. For the high poaching rate scenario,

poaching functions were kept the same as in the

low poaching rate scenario. For each poaching

function, the number of sub-adult and adult

elephants poached from the park was randomly

assigned using a Poisson distribution and the

scenario-speciﬁ c rate. Thus, the rate of poaching

per year, in the model, was assumed to be additive

to the stage-speciﬁ c natural mortality. We did not

include sex-speciﬁ c differences in poaching rate

because there was no information on such sex-

speciﬁ c differences for WKNP. There is evidence

that adult female elephants are also poached in

Sumatra and their toenails, genitalia, and other

body parts are collected for use in traditional

medicines (Sitompul et al. 2002).

Several other assumptions were required in

constructing the models. Natural mortality

rates used were derived from data on Indian

elephants, which might be different than

Sumatran elephants. However, it is unlikely that

they would be substantially different because

elephants in India and Sumatra have similar

life histories. Furthermore, we did not include a

carrying capacity function because the carrying

capacity of the study area is not well studied

(but is thought to be much higher than the

present population size) and because our primary

concern was preventing declining populations

and local extinction, the effect of density-

dependent factors as the population approached

carrying capacity was considered unimportant.

However, model scenarios projecting increases

in population will need reﬁ nement and some

measure of carrying capacity should be included

as those data become available. Finally, potential

genetic problems associated with small isolated

elephant populations (e.g. inbreeding depression)

were not included in our model.

Sensitivity analyses

The purpose of the sensitivity analyses was

to determine the relative inﬂ uence of each

parameter and alternative poaching model on

model estimates (Williams et al. 2002). Relative

sensitivity of model estimates can be evaluated by

varying model input parameters over a speciﬁ ed

range and examining the change in model

outputs. For this study, we evaluated the relative

sensitivity of the year 50 model estimates to each

parameter by calculating a Sensitivity Index (SI)

using regression analysis to calculate the slope

and uncertainty of each poaching function and

then multiplying the slope and uncertainty of

34

the parameter to calculate the SI following the

methods of Wiegand et al. (1998). We evaluated

the sensitivity of reproductive parameters of

sub-adult and adult elephants by varying the

reproductive rates from 0.19 to 0.25, with 0.01

increments. We also evaluated model sensitivity

to the survival rate parameter for the calf to

juvenile transition and the sub-adult to adult

transition by varying the survival parameter for

each life history stage from 0.75 to 0.90, with

0.05 increments. To understand the sensitivity of

the population model to the alternative poaching

functions, we varied poaching rate from the low

poaching scenario’s 50% to 200% of the estimate

values in 10% increments. The results of these

sensitivity analyses for the high poaching rate

scenario will be identical to the low poaching

rate scenario since the difference between

the low and high poaching rate scenarios is

simply the magnitude of the poaching rate. All

simulation modelling and sensitivity analyses

were conducted using SAS (SAS version 8.2).

Results

Projection of the WKNP elephant population

over a 50-year period showed the population

increasing from 180 elephants to 594 elephants

(95% CI = [570, 618]) if we assumed that

poaching stopped. The extinction coefﬁ cient for

the control population was 0.0 and population

growth rate (λ) was 1.02 (0.0001 SE). Under the

low poaching rate scenarios we also showed that

the elephant population would increase (Fig. 2).

The linear poaching function produced an elephant

population in year 50 of 422 (95% CI = [403,

441]). The extinction coefﬁ cient using the linear

function was also 0.0 and λ was 1.02 (0.0002 SE).

If poaching in the park behaves as an exponential

extinction function, the elephant population in

year 50 was estimated to be 325 (95% CI = [308,

342]). The extinction coefﬁ cient for this function

was 0.009 and λ was 1.01 (0.0002 SE). The

constant and logistic poaching functions in the

model produced estimates of elephant population

size of 253 (95% CI = [235, 271]) and 263 (95%

CI = [245, 281]), respectively. The extinction

coefﬁ cient with constant poaching was 0.099,

and logistic poaching resulted in an estimate of

0.086. The population growth rate with constant

poaching was 1.0 (0.0005 SE) and λ with logistic

poaching was 1.0 (0.0005 SE; Table 1). The age

distribution after 50 years for the control and low

poaching rate scenarios changed slightly from

one dominated by sub-adults towards one more

dominated by adults (Fig. 3).

Population models with high poaching rate

scenarios showed a different trend to the low

poaching rate scenarios over the 50-year period.

In the high poaching rate scenarios, only linear

and exponential decay poaching patterns showed

that the elephant population in WKNP would

increase over the 50 years (Fig. 4). Population

size in year 50 for the linear and exponential

decay poaching functions was estimated to

be 274 (95% CI = [263, 285]) and 217 (95%

CI = [211, 226]), respectively. The extinction

coefﬁ cient for the linear and exponential

poaching functions was 0.0 and λ was 1.0

(0.0002 SE). For the exponential decay poaching

function, the extinction coefﬁ cient was 0.01 and

λ was 1.0 (0.0003, SE). In contrast, the constant

poaching and logistic poaching functions in the

high poaching scenarios showed that elephant

population in WKNP would decline dramatically

(Fig. 4). Final population size in year 50 for the

constant and logistic poaching functions was

41 (95% CI = [33, 49]) and 37 (95% CI = [30,

44]), respectively. The extinction coefﬁ cient

for constant poaching was 0.75 and for logistic

poaching it was 0.76. The population growth rate

was 0.97 (0.008 SE) for constant poaching and

0.97 (0.009 SE) for logistic poaching (Table 1).

The age distribution in the high poaching rate

scenarios showed similar patterns to the low

poaching scenarios, with more adult individuals

found at the end of each simulation (Fig. 5).

Sensitivity analyses

Sensitivity analyses for each natural parameter

revealed high levels of variation in the model. The

result of the sensitivity analyses for the sub-adult

and adult reproductive parameters showed that

small changes in the adult reproductive parameter

caused large changes in the ﬁ nal population size.

For example, an increase of 6% in the adult

reproduction rate could cause a 76.01% change in

ﬁ nal population size. In contrast, a 6% change in

35

0

100

200

300

400

500

600

700

0 5 10 15 20 25 30 35 40 45 50 55

Time

Control

Constant

Exponential

Linear

Logistic

Figure 2. Simulated population trends of Asian elephants for 50 years

under control and low poaching scenarios in Way Kambas National

Park,. Density dependent effects using low poaching level scenarios were

developed (constant, exponential, linear and logistic).

0

50

100

150

200

250

300

350

400

Calf

Juvenile

Subadult

Adult

Figure 3. Projection of the age structure of the elephant population in

Way Kambas National Park after 50 years of simulation, presented in the

current population (start) and in control and low poaching scenarios.

0

100

200

300

400

500

600

700

0 5 10 15 20 25 30 35 40 45 50 55

Time

Control

Constant

Exponential

Linear

Logistic

Figure 4. Simulated population trends over 50 years period under control

and high poaching scenarios in Way Kambas National Park. Density de-

pendent effects using high poaching level scenarios were developed (con-

stant, exponential, linear and logistic).

0

50

100

150

200

250

300

350

400

Start Control Constant Linear Exponential Logistic

Calf

Juvenile

Subadult

Adult

Figure 5. Projection of the age structure of the elephant population in Way

Kambas after 50 years of simulation, presented in the current population

(start) and in control and high poaching scenarios.

36

sub-adult reproduction rate only caused a 26.84%

change in ﬁ nal population size (Fig. 6). For the

survival parameter, sensitivity analyses showed

that juvenile survival and young survival rates had

relatively similar impact on the ﬁ nal population

size. An increase of 5% in survival of young and

juvenile elephants independently caused a change

of 29.25% and 29.87% in ﬁ nal population size,

respectively (Fig. 7). However, the adult survival

parameter had a far more sensitive effect on the

ﬁ nal population size compared to the sub-adult

survival parameter. Changing the adult survival

parameter 5% could cause an 86.54% change in

ﬁ nal population size. In contrast, a 5% change

in the sub-adult parameter only caused a 37.46%

change in ﬁ nal population size (Fig. 8).

Sensitivity analysis for the four poaching function

parameters showed clear differences in model

sensitivity (Fig. 9, Table 2). The logistic poaching

function appeared to have the greatest inﬂ uence,

which is shown by it having the lowest index

(SI = -2.626) followed by the constant poaching

function (SI = - 0.013). The linear, constant, and

exponential poaching functions appeared to have

relatively similar sensitivity in the model (Fig. 9).

The level of uncertainty of poaching parameter in

the model showed that the exponential parameter

had the lowest uncertainty compared to the other

three poaching parameters (Table 2).

Discussion

Our model clearly demonstrates that in the control

(no poaching) scenarios the elephant population

in the park will increase over time. Furthermore,

the low poaching rate scenarios also show the

elephant population increasing. These results

imply that the low poaching rates observed in the

past did not have a serious negative impact on the

elephant population in the park. The population

growth rate in the low poaching rate scenarios

remained about 1.0 or above and extinction

encounter rate after 1000 simulations was less

than 0.1. However, if we doubled the poaching rate

from the minimum known rate observed in 2000–

2002, as in the high poaching scenarios, we found

that the population could decline dramatically

for the logistic poaching and constant poaching

functions, with the extinction coefﬁ cients for

both functions increasing signiﬁ cantly up to

about 75%. For both the constant and logistic

poaching functions, the magnitude of poaching

pushed the population into negative growth rates.

In contrast, the linear and exponential poaching

functions did not differ much from the lower

poaching scenarios. In this situation, poaching

(linear and exponential functions) seemed to

have little effect on the population even though

the magnitude of the poaching increased two fold

from the low poaching scenarios. It is clear from

these results that further study of the WKNP

population, and other Asian elephant populations,

is necessary in order to decide which poaching

function best describes reality and therefore

allow us to better model population trajectories

under different scenarios.

The age distribution in the model showed that

the proportional representation of the different

age stages in the population shifted towards the

adult age stage for the low and high poaching rate

scenarios. The overall pattern of age distribution

for both poaching scenarios was the same, with

Table 1. Summary of model result representing ﬁ nal population size; population growth rate and

extinction encounter using all possible scenarios in the model. ƒ = poaching function of population

size. N50= population at year 50; λ = population growth ate; EC= Extinction Coefﬁ cient.

Scenarios ƒ N50 95%CL λ95%CL EC

Control 594 23.59 1.02 0.0002 0

Low-poaching constant 253 17.87 1.01 0.001 0.099

linear 422 19.03 1.02 0.0004 0

exponential 325 16.63 1.01 0.0006 0.009

logistic 263 17.80 1.00 0.0009 0.086

High-poaching constant 41 7.86 0.97 0.016 0.75

linear 274 11.08 1.00 0.0005 0

exponential 217 9.40 1.00 0.0007 0.01

logistic 37 7.09 0.97 0.018 0.76

37

Figure 6. Response on predicted elephant population size in 50 years

simulation for various combinations of adult reproduction rate (y-axis) and

sub adult reproduction rate (x-axis). Line in different color represents el-

ephant population size for speciﬁ c adult and sub adult reproduction rate.

Figure 7. Response on predicted elephant population size in 50 years

simulation for various combinations of juvenile survival rate (y-axis) and

calf survival rate (x-axis). Line in different color represents elephant popu-

lation size for speciﬁ c juvenile and calf survival rate.

Figure 8. Response on predicted elephant population size in 50 years

simulation for various combinations of adult survival rate (y-axis) and sub

adult survival rate (x-axis). Line in different colour represents elephant

population size for speciﬁ c adult and subadult survival.

0

0.5

1

1.5

2

2.5

3

0 0.5 1 1.5 2 2.5

Rate of change

Constant

Exponential

Linear

Logistic

Figure 9. Response of population size in 50 years simulation to the rate of

change on the poaching function parameter performed in the model. Dif-

ferent colour line represents different poaching function in the model.

38

the highest proportion of the population formed by

the adult stage followed by the calf, juvenile, and

sub-adult stages. If we examine the relationship

between population growth and age structure after

simulation, we ﬁ nd that for the low poaching rate

scenarios the population is predicted to grow after

50 years. A similar pattern was also found for the

exponential and linear poaching functions in the

high poaching rate scenario. If the population is

growing, that means the population growth rate

is equal to or more than one. In this situation we

would expect the age distribution at the end of

simulation year to be dominated by the younger

age classes. However, our models did not predict

this, suggesting that improved survival of sub-

adult and adult elephants in the population over

a relatively short projection period (50 years)

relative to an elephant’s lifespan provided our

populations with much greater numbers of older

individuals. As a result, there was not enough new

recruitment to shift the age distribution towards

the younger age classes.

Sensitivity analyses

Our sensitivity analyses showed that variation

in reproduction parameters for adults had the

greatest impact on model variability. Relatively

small changes in adult reproduction rate could

cause a signiﬁ cant impact on ﬁ nal population

size. Therefore, reproduction rate of adult

elephants needs to be determined accurately if

models such as ours are to be useful management

tools and to allow the demographic condition

of populations of interest to be assessed. If we

assumed reproduction rate in the population to

be deterministic, and compared the sensitivity of

the survival rate, we found the model was more

sensitive to the adult survival parameter compared

to the subadult survival parameter. Sukumar

(1989) suggested that among adult elephants,

female survival rate had a more signiﬁ cant effect

on the population than did male survival rate. His

study suggested that if adult male elephants have

low survival, the population could still grow if

female survival rate was high. Similar results

have also been demonstrated for other long-lived

species such as grizzly bears in Yellowstone

National Park (Eberhardt et al. 1994).

Sensitivity analyses for the poaching parameter

revealed a clear sensitivity to poaching function in

the model and this was reﬂ ected in the sensitivity

index value for the parameter. Sensitivity analyses

showed the logistic poaching function was the

most sensitive poaching function. This is most

likely because the number of elephants poached

per year was maintained at the maximum level and

at the same time randomization was incorporated

into the function. Clear differences can be found

if we compare the sensitivity of the logistic to the

constant poaching function: the constant poaching

function tended to be less sensitive, even though

the number of elephants poached per year was

maintained at the maximum level, presumably

because no randomization was incorporated into

this poaching function.

Management implications

Our model suggests that the elephant population

in WKNP will not decline over the next 50

years provided poaching rates remain at the low

level observed in 2000–2002. While this result

is encouraging, there is a possibility that the

2000–2002 poaching rate data used in this study

underestimated real poaching rates in the park at

that time because they were based on the number

of elephant remains found without dedicated

carcass searches. There is, therefore, a possibility

that the number of elephants killed because of

poaching was higher than our estimate, and our

models suggest this if this were so the increased

poaching could push the population toward

negative growth. Moreover, even if the 2000–

2002 data were representative of actual poaching

rates at that time an evidence-based adaptive

management approach to protecting the park’s

Table 2. Sensitivity analysis of the poaching

parameter. Poaching was speciﬁ ed as function

of population size. β0= parameter value;

α(β,β0)= slope; Δ(β)= approximate uncertainty

in the parameter; SI(β,β0)= sensitivity index of

parameter β within point β0.

Poaching β0α(β,β0)Δ(β) SI(β,β0)

Constant 2.848 -0.258 0.049 -0.013

Exponent. 2.630 -0.105 0.012 -0.001

Linear 2.780 -0.161 0.000 0.000

Logistic 4.050 -1.802 1.457 -2.626

39

elephants would require monitoring of poaching

rates to determine, for example, whether law

enforcement targets were being achieved.

Therefore a poaching monitoring program

(e.g. systematic carcasses searching) should be

established as a priority for management of the

park’s elephant population. This could perhaps

involve the use of detection dogs (sniffer dogs) to

improve carcass detection efﬁ ciency, as elephant

carcasses are surprisingly difﬁ cult to ﬁ nd in

forested environments. In addition to improving

detection rates, the limited number of arrests in

relation to elephant poaching and the existence

of local ivory markets clearly also need to be

addressed (Hedges et al. 2005). Interestingly,

reducing poaching could also reduce human–

elephant conﬂ ict around WKNP because

research in Africa has shown that poachers

hunting elephants in forests can drive them into

closer proximity to surrounding farmland thus

increasing crop depredation rates (e.g. Nchanji

2005).

Finally, this model did not incorporate habitat

degradation or destruction in and around the

park. However, illegal killing of elephants

and other wildlife is known to be correlated

with road building, agricultural encroachment,

and other forms of habitat degradation and

destruction that facilitate human access into

wildlife-inhabited areas (Duckworth & Hedges

1998), and so elephant population management

in WKNP and elsewhere on Sumatra should also

focus on reducing habitat destruction, especially

encroachments into elephant habitat.

Acknowledgements

The study was conducted as a collaboration

between the Wildlife Conservation Society and

the Indonesian Ministry of Forestry’s Directorate

General of Forest Protection and Nature

Conservation (PHKA). The project was funded

by the Wildlife Conservation Society and the US

Fish & Wildlife Service (through Asian Elephant

Conservation Fund grants 1448-98210-00-G496,

98210-1-G806, and 98210-2-G292), the National

Geographic Society, and WWF-US. Data analysis

was supported by the Warnell School of Forestry

and Natural Resources, the University of Georgia,

USA. We thank Clint Moore and Michael Conroy

for valuable advice on modelling. Finally we

thank Margaret Kinnaird, Tim O’Brien, Josh

Ginsberg, and Martin Tyson for support and

advice during the project.

References

Belovsky, G.E. (1987) Extinction model and

mammalian persistence. In: Viable Populations

for Conservation. Soule, M.E. (ed) Cambridge

University Press. New York. pp 35-57.

Berger, J. (1990) Persistence of different-sized

populations: An empirical assessment of rapid

extinctions in bighorn sheep. Conservation

Biology 4: 91-98.

Convention on International Trade in Endangered

Species of Wild Fauna and Flora (CITES) (2000)

Experimental trade in raw ivory of populations in

Appendix II: interpretation and implementation

of the convention. Conservation of and trade in

elephants [Doc 11.31.1]. In: Eleventh Meeting of

the Conference of the Parties. CITES Secretariat,

Geneva, Switzerland.

Blouch, R.A. & Haryanto (1984) Elephants in

Southern Sumatra. Unpublished report, IUCN/

WWF Project 3033, Bogor, Indonesia.

Blouch, R.A. & Simbolon, K. (1985) Elephants

in Northern Sumatra. Unpublished report, IUCN/

WWF Project 3033, Bogor, Indonesia.

Cromsigt, J.P.G.M., Hearne, J., Heitkönig,

I.M.A. & Prins, H.H.T. (2002) Using models

in the management of black rhino populations.

Ecological Modeling 149: 203-211.

Douglas-Hamilton, I. (1987) African elephants:

population trends and their causes. Oryx 21: 11-

24.

Duckworth, J.W. & Hedges, S. (1998) Tracking

Tigers: A Review of the Status of Tiger, Asian

Elephant, Gaur, and Banteng in Vietnam,

Lao, Cambodia, and Yunnan (China), with

Recommendations for Future Conservation

Action. WWF Indochina Programme, Hanoi,

40

Vietnam.

Eberhardt, L.L., Blanchard, B.M. & Knight,

R.R. (1994) Population trend of the Yellowstone

grizzly bear as estimated from reproductive and

survival rates. Canadian Journal of Zoology 72:

1-4.

Hedges, S., Tyson, M.J., Sitompul, A.F., Kinnaird,

M.F., Gunaryadi, D. & Aslan (2005) Distribution,

status and conservation needs of Asian elephant

(Elephas maximus) in Lampung Province,

Sumatra, Indonesia. Biological Conservation

124: 35-48.

Leslie, P.H. (1945) On the use matrices in

population mathematics. Biometrika 33: 183-

212.

Leslie, P.H. (1948) Some further notes on the

use of matrices in population mathematics.

Biometrika 35: 213-245.

Milliken, T. (2004) African Elephants and the

Thirteenth Meeting of the Conference of the

Parties to CITES, Bangkok, Thailand 2004. A

TRAFFIC Brieﬁ ng Document. TRAFFIC East/

Southern Africa, Harare, Zimbabwe.

Nchanji, A.C. (2005) Elephant poaching weapons

and new experiences from the Banyang-Mbo

wildlife sanctuary, Cameroon. Pachyderm 39:

33-42.

Poole, J.H. & Thomsen, J.B. (1989) Elephants

are not beetles: implications of the ivory trade

for the survival of the African elephant. Oryx 23:

188-198.

Reilly, J. (2002) Growth in the Sumatran

elephant (Elephas maximus sumatranus) and age

estimation based on dung diameter. Journal of

Zoology 258: 205–213.

Santiapillai, C. & Jackson, P. (1990) The Asian

Elephant: An Action Plan for its Conservation.

IUCN/SSC Asian Elephant Specialist Group,

Gland, Switzerland.

SAS Institute. Proc Regression, Version 8.02.

SAS Institute, Cary, North Carolina, USA.

Sitompul, A.F., Hedges, S. & Tyson, M.J. (2002)

Elephant Deaths in Sumatra as a Result of

Poaching or Elephant Capture Operations, 1

January 2000 to 1 November 2002. Unpublished

Report, Wildlife Conservation Society – Indonesia

Program, Bogor, Indonesia.

Sukumar, R. (1989) The Asian Elephant.

Cambridge University Press, Cambridge, UK.

Sukumar, R., Ramakrishnan, U. & Santosh,

J.A. (1998) Impact of poaching on an Asian

elephant population in Periyar, southern India: A

model of demography and tusk harvest. Animal

Conservation 1: 281-291.

Wiegand. T., Naves, J., Stephan, T. & Fernandez,

A. (1998) Assessing the risk of extinction for

the brown bear (Ursus arctos) in the Cordillera

Cantabrica, Spain. Ecological Application 68:

539-570.

Williams, B. K., Nichols, J.D. & Conroy, M.J.

(2002) Analysis and Management of Animal

Populations. Academic Press, California, USA.

Corresponding author’s e-mail:

asitompu@forwild.umass.edu