Content uploaded by Batbayar Galtbalt
Author content
All content in this area was uploaded by Batbayar Galtbalt on May 17, 2020
Content may be subject to copyright.
5
The Wildlife Picture Index: monitoring
Mongolian biodiversity
with camera trapping
Susan E. Townsend, Batbayar Galtbalt, Munkhjargal Myagmar, Jonathan Baillie
and Timothy G. O’Brien
Abstract
The Wildlife Picture Index (WPI) was developed to
measure trends in biodiversity at the landscape
level using camera trap sampling. As a case study
from 2009 to 2011, we tested the use of the WPI to
assess how well protected areas were functioning
to conserve wildlife (biodiversity) in Mongolia.
Mongolia supports a rich ungulate and carnivore
fauna, has low human population density and has
established protected areas, presenting ideal condi-
tions to test this new conservation tool for assessing
trends in biodiversity. Our effort resulted in 26 816
trap-nights producing 416 300 photographs from
five study areas. In this paper, we present species-
specific occupancy estimates for one study area and
the WPI for the study area for which we had three
seasons of data. The WPI effectively measured bio-
diversity and the status of individual species at a
landscape level within the Protected Areas. We
detected a total of 22 (possibly 24) mammalian spe-
cies ranging from six to 13 species at each site. In the
least protected area of a study site, the WPI rose
14.5% after the first year and declined 38% in the
third year, and in the more protected area, it rose
12.6% after the first year and then declined back to
baseline in the third year. This approach could
prove to be very useful and cost effective for long-
term monitoring and adaptive management.
Introduction
With recent advances in camera trapping meth-
ods, it is now possible to monitor trends in the
diversity, abundance and distribution of a broad
range of terrestrial mammals and birds in a vari-
ety of habitats ranging from savannah to deserts
to tropical ecosystems. Camera trapping is a par-
ticularly attractive approach for monitoring
because it is non-obtrusive, has low observer
error, is comparable across sites, data can be
aggregated for various indices, and photographs
allow for verifiability. Setting and maintaining
camera traps does not require highly skilled staff.
It is lower in cost when compared to other
approaches of equal rigor (O’Brien and Kinnaird
2013; Silveira et al. 2003).
Copyright 2014. CSIRO PUBLISHING.
All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted under U.S. or applicable copyright law.
EBSCO Publishing : eBook Collection (EBSCOhost) - printed on 5/17/2020 5:18 PM via DEAKIN UNIVERSITY LIBRARY
AN: 921133 ; Meek, Paul, Fleming, Peter, Ballard, Guy, Banks, Peter, Claridge, Andrew W., Sanderson, James, Swann, Don.; Camera Trapping : Wildlife Management
and Research
Account: deakin.main.eds
PArt 1 – cAmErA trAPPIng for AnImAL monItorIng: cAsE studIEs
46
Landscape level wildlife monitoring with
camera traps has generally been implemented in
forested tropical ecosystems (O’Brien et al. 2010;
Ahumada et al. 2011 [see Tropical Ecosystem
Assessment & Monitoring Network; <http://www.
teamnetwork.org>]), often targeting a single spe-
cies (tiger: Karanth et al. 2011; pygmy hippopoto-
mus: Collen et al. 2011) and certain trophic levels
(such as carnivores, see Pettorelli et al. 2010; Burton
et al. 2011).
The Wildlife Picture Index (hereafter, ‘WPI’) is a
metric for biodiversity using camera trapping
(O’Brien 2010; O’Brien et al. 2010). The WPI uses
species-specific occupancy statistics scaled to the
first year of camera trapping data; these data are
then used to produce the index using the geometric
mean for all species in a given year as per Buckland
et al. (2005). Because we were interested in species
occurrence and distribution, camera trapping is a
logical choice for documenting occupancy over
large areas. Because a species may go undetected
in a sample unit even if that species is actually pre-
sent, this ‘false absence’ leads to underestimation
of true occupancy. Unless the probability of detec-
tion is determined, measures of occupancy over
time are invalid (MacKenzie et al. 2005). Occupancy
modelling allows for estimation of true occupancy
and detection probabilities, and provides the basis
for the Wildlife Picture Index (O’Brien et al. 2010).
Mongolia has seen rapid social, economic and
environmental change over the past two decades
due to a transition to the market economy. This
transition resulted in major shifts in the human
population and, to some extent, a breakdown of
regulatory mechanisms. This, coupled with high
unemployment, led to a flourishing illegal and
wildlife trade, now estimated to be worth more
than US$100 million annually (Wingard and
Zahler 2006). It has been difficult to define the
impact that this has had on Mongolian wildlife, as
few scientifically robust studies on population
trends have been carried out over the past two dec-
ades. The Mongolian Red List for Mammals, initi-
ated by the World Bank, indicates that massive
declines have taken place; 79% of large herbivores
and 12% of carnivores are listed as threatened with
extinction (Clark et al. 2006). While the Red List is
an excellent first step, the certainty of many of the
status assessments is low due to poor data quality
and the reliance on anecdotal information and
expert opinion.
As a case study from 2009 to 2011, we tested the
WPI as a tool to assess how well protected areas
were functioning to conserve wildlife (biodiver-
sity) in Mongolia with the intent of having the park
administration staff implement long-term monitor-
ing. Mongolia supports a rich ungulate and carni-
vore fauna, has low human population density and
has established protected areas, presenting ideal
conditions to test this new conservation tool for
assessing trends in biodiversity. Mongolia supports
both temperate grassland and forested ecosystems
and is part of the Holarctic ecozone.
We hypothesised that biodiversity and species-
specific occupancy estimates would be higher in
the more restricted management zones in the pro-
tected areas than in the less restricted management
zones. We expected there would be less anthropo-
genic influence in the more restrictive zone. In
order to test this, we set up camera-trapping grids
in two different zones in each study area.
In this paper, we present representative find-
ings from one of our study areas to illustrate base-
line information that can be generated from one
season and present the WPI from another site,
Myangan Ugalzat (MU), from which we have three
seasons of data. Establishing baseline occupancy
estimates and diversity and being able to detect
trends over time are two useful outcomes of this
approach towards effective management.
Methods
Five study areas representing four ecoregions that
were sampled over 3 years (Plate 7). These study
areas included: (1) Myangan Ugalzat (MU) located
in Tsetseg Soum, Hovd Aimag in western Mongo-
lia in the south-western portion of the Altai Sayan
region; (2) Khonin Nuga (KN) located in a valley in
the western Khentii region of Northern Mongolia;
EBSCOhost - printed on 5/17/2020 5:18 PM via DEAKIN UNIVERSITY LIBRARY. All use subject to https://www.ebsco.com/terms-of-use
47
5 – tHE WILdLIfE PIcturE IndEx: monItorIng mongoLIAn bIodIvErsIty WItH cAmErA trAPPIng
(3) Onon Balj (OB) located in the Eastern Khentii in
the Amur River Basin ecoregion; (4) Mongol
Daguur (MD) located in north-eastern Dornod
Aimag in the Eastern Steppe of Mongolia; and (5)
Numrug (NU) located in the farthest east of Mon-
golia in Dornod Aimag. In this paper, we present
representative findings from KN and MU.
Our field work was conducted during the
summer season between June and August in 2009,
2010 and 2011. In MU during the summer season of
2009, we deployed 40 SG550 camera traps in
2 × 20 km2 grids with 1 km spacing; one in a zone
that allowed grazing (Buffer Zone, Grid I) and one
that did not (Core Area, Grid II). In 2010 and 2011,
we deployed 100 SG550 camera traps at stations at
MU, KN and OB and at MU, MD and NU, respec-
tively, in 2 × 50 km2 grids (1 km spacing) in two
different management zones at each study area; a
less restricted zone (Grid I) and a more restricted
zone (Grid II). The first camera station location for
each grid was randomly selected. We predicted our
occupancy estimates, terrestrial mammalian diver-
sity and the presence of expected red-listed species
would be higher in Grid II due to less human influ-
ence and more protection for wildlife than Grid I,
characterised by more human influence and less
protection for wildlife.
Data analysis
Protocols were standardised and followed TEAM
published guidelines (<http://www.teamnetwork.
org/data/manage>). All photographs were cata-
logued by attaching location coordinates to EXIF
data and extracting EXIF data into excel spread-
sheets using Picture Information Extractor (PIE)
software. All photographs were viewed for animal
identification and then entered into the excel
spreadsheet. Occupancy estimates were deter-
mined for all species that were reliably detected.
We used single-season occupancy models to esti-
mate occupancy (ψ) and detection probabilities (ρ)
for each species (MacKenzie et al. 2003) using the
program PRESENCE (v3.2, Hines 2006). Occupancy
models account for imperfect detection and pro-
vide unbiased estimates of occupancy. To apply
these models, detection histories were compiled for
each species at each site (Grids I and II) for each
camera station in a series of ones (detection) and
zeros (non-detection). Each day the camera was set
and functioning was considered a replicate sample.
Each day the camera station was ‘down’ or not
functioning was treated as a missing value.
We modelled each of our grids separately. We
ran two predefined models and used the model
with lowest delta AIC (Akaike’s Information Crite-
rion) to estimate probability of detection and occu-
pancy as calculated by the software PRESENCE
(Hines 2006). The first model estimated the same
occupancy probability for all camera station loca-
tions and that detection probability (ρ) was con-
stant across both camera station location and
survey occasions (i.e. two parameters). The second
model assumes that all camera station locations
have the same probability of occupancy (ψ), but
that ρ varies between the surveys although at each
survey occasion, ρ is the same at each camera sta-
tion location. Several species were detected too
infrequently to generate occupancy estimates and,
in those cases, we report the observed occupancy.
The Wildlife Picture Index (WPI)
Methods to calculate the WPI including the R script
are discussed in detail in O’Brien et al. (2010) and
O’Brien (2010). The most essential details are pre-
sented below. To develop the WPI, species-specific
occupancy estimates were generated from the
camera trap grid data for each season. Each survey
season had a series of repeated surveys represented
by each camera trap day. The index used each spe-
cies-specific occupancy estimate for species i at site
j in year k. In this case, 3 years of seasonal data pro-
vided three periods to generate occupancy esti-
mates and an index. Occupancy in year k was
divided by the estimated occupancy at the initial
season, Oijk = ψijk/ψij1. Each species-specific index
measured change in occupancy from initial condi-
tions. The estimate for k = 1 is always 1. The WPI
for year k and site j and n species in geometric mean
of scaled occupancy statistics for n species was cal-
culated as follows:
EBSCOhost - printed on 5/17/2020 5:18 PM via DEAKIN UNIVERSITY LIBRARY. All use subject to https://www.ebsco.com/terms-of-use
PART 1 – CAMERA TRAPPING FOR ANIMAL MONITORING: CASE STUDIES
48
IO
jk ijk
i
n
n
1
=
=
%
The WPI was calculated for MU, the only site for
which we had 3 years of occupancy estimates.
Results
Camera trapping effort resulted in 26 816 camera
trap-nights for all sites for all years. This effort
resulted in 416 294 images and 133 341 events (Table
5.1). We detected a total of 22 (possibly 24) mam-
malian species ranging from 6–13 species at each
site. Red Listed species were detected at all of our
study sites (Table 5.2).
For all sites, we established a list of detected
species and baseline occupancy estimates for each
management zone in year 1. An example of base-
line results from year 1 at KN showed a higher
diversity of carnivores (five versus two species)
Table 5.1. Camera trapping effort for Wildlife Picture Index (WPI) study sites in Mongolia protected areas, 2009–11.
Protected area Year Trap- nights Images Events
Area
(km2)
Myangan Ugalzat 2009 561 14 410 4 803 40
Myangan Ugalzat 2010 5 889 35 141 9 752 100 *
Myangan Ugalzat 2 011 3 065 18 033 6 011 100 †
Onon Balj 2010 3 649 114 790 34 271 100
Khonin Nuga 2010 7 250 105 218 35 073 100
Mongol Daguur 2011 3 16 4 67 145 22 912 100
Numrug 2011 3 238 61 556 20 519 10 0
Total 26 816 416 293 133 341 500
* includes area used in 2009
† same area as 2010
Table 5.2. IUCN Red List of Threatened Species-listed mammalian species detected at WPI study sites in Mongolia (from
Clark et al. 2006) ranked in descending threatened status order.
KN = Khonin Nuga, OB = Onon Balj, MU = Myangan Ugalzat, MD = Mongol Daguur, NU = Numrug. D = detected.
Species Common Name IUCN Status* KN OB MU MD NU
Cervus elaphus Red deer Critically Endangered D D D
Marmota sibirica Siberian marmot Endangered DDD
Moschus moschiferus Musk deer Endangered D
Ovis ammon Argali Endangered D
Alces alces Elk (moose) Endangered D D
Procapra gutturosa Mongolian gazelle Endangered D
Martes zibellina Sable Vulnerable D
Lynx lynx Eurasian lynx Near Threatened D D
Otocolobus manul Pallas’ cat Near Threatened D
Canis lupus lupus Grey wolf Near Threatened DDDDD
Vulpes corsac Corsac Near Threatened D D
Vulpes vulpes Red fox Near Threatened DDDD
Sus scrofa Wild boar Near Threatened D D
Capra sibirica Siberian ibex Near Threatened D
* IUCN Red List online <http://www.iucnredlist.org>, accessed September 2013.
EBSCOhost - printed on 5/17/2020 5:18 PM via DEAKIN UNIVERSITY LIBRARY. All use subject to https://www.ebsco.com/terms-of-use
49
5 – THE WILDLIFE PICTURE INDEX: MONITORING MONGOLIAN BIODIVERSITY WITH CAMERA TRAPPING
and higher occupancy for the brown bear (Ursus
arctos) in Grid I (Fig. 5.1). In contrast for ungulates
(and a small herbivore, the mountain hare, Lepus
timidus), diversity and occupancy were similar for
both grids (Figs 5.1 [for L. timidus] and 5.2). In the
case of KN, we had anticipated that Grid I would
be subject to more human influence than Grid II.
However, as was the case for most of our study
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Lepus timidusMeles
leucurus
Martes
zibellina
Lynx lynx Canis lupusUrsus arctos
Occupancy
Grid 1
Grid 2
Fig. 5.1. Occupancy estimates for mountain hare (Lepus timidus) and carnivores (Meles leucurus = Asian badger, Martes
zibellina = sable, Lynx lynx = Eurasian lynx, Canis lupus = grey wolf and Ursus arctos = brown bear) at Khonin Nuga
Protected Area for Grids I and II, 2010 (error whiskers = ± standard error; columns without error bars are naïve estimates).
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Sus scrofaAlces alcesCapreolus
pygargus
Cervus elaphusMoschus
moschiferus
Occupancy
Grid 1
Grid 2
Fig. 5.2. Occupancy estimates for the ungulates (Sus scrofa = wild boar, Alces alces = moose, Capreolus pygargus = roe
de er, Cervus elaphus = red deer and Moschus moschiferus = musk deer) for Grid I and II at Khonin Nuga 2010 (error
whiskers = ± standard error; columns without error bars are naïve estimates).
EBSCOhost - printed on 5/17/2020 5:18 PM via DEAKIN UNIVERSITY LIBRARY. All use subject to https://www.ebsco.com/terms-of-use
PART 1 – CAMERA TRAPPING FOR ANIMAL MONITORING: CASE STUDIES
50
areas except for NU, Grid II had more human influ-
ence than what was expected. For KN, unantici-
pated tree extraction was taking place, which was
documented on our camera traps.
The WPI was calculated for MU using occu-
pancy estimates for those species (Table 5.3) for
which we could obtain occupancy estimates for
the three seasons. The Siberian ibex and argali
were detected in the Grid I but not Grid II; signifi-
cant in so far as these species were expected in the
Core Area (Grid II) of the park where there was
better habitat. Over the 3 years, while not allowed
for ‘core areas’ of the National Park, there were
consistently high levels of livestock and nomadic
herder families. In Grid I, the WPI rose 14.5% after
the first year and declined by 38% after the second
year (Fig. 5.3). In Grid II, the WPI rose 12.6% after
the first year and then declined back to the 2009
level (Fig. 5.4).
Discussion
Over 3 years, the WPI was implemented at five
study sites in Mongolia resulting in a baseline for
four protected areas and the WPI for one protected
area (MU). This paper presents an example of year
1 occupancy estimates from one site and the WPI
from another. The method developed for the WPI
also produces baseline occupancy estimates and
species-specific information on the presence and
status of mammalian species (> 1 kg) in and around
Mongolian protected areas.
For all five study areas, we saw no consistent
relationship between number of species (diversity),
occupancy and its level of protection. We hypothe-
Table 5.3. Mammalian species detected in camera trap grids for the Buffer Zone (Grid I) and Core Area (Grid II) in Myangan
Ugalzat National Park, Mongolia.
Trapping effort is shown in Table 5.1. D = detected
Order Species Common Name Buffer Zone Core Area
Lagomorpha Ochotona sp. Pika D
Lepus tolai Tolai h ar e D D
Rodentia Marmota sibirica Siberian marmot D D
Spermophilus undulatus Souslik D
Carnivora Canis lupus lupus Grey wolf D
Gulo gulo Wolverine D
Mustela eversmanni Siberian ferret D
Otocolobus manul Manul D
Vulpes vulpes Red fox D D
Vulpes corsac Corsac D
Artiodact yla Capra sibirica Siberian ibex D
Ovis ammon Argali D
0
0.5
1
1.5
2
2.5
2008 2009 2010 2011 2012
WPI
Year
mean
LCL
UCL
Fig. 5.3. The WPI plotted from 2009–11 for Grid I
(Buffer Zone), Myangan Ugalzat National Park, Mongolia
(mean = geometric mean, LCL = lower confidence limit,
UCL = upper confidence limit)
EBSCOhost - printed on 5/17/2020 5:18 PM via DEAKIN UNIVERSITY LIBRARY. All use subject to https://www.ebsco.com/terms-of-use
51
5 – THE WILDLIFE PICTURE INDEX: MONITORING MONGOLIAN BIODIVERSITY WITH CAMERA TRAPPING
sised that diversity and occupancy would be lower
in management zones that were subject to greater
human influence. However, we did not anticipate
that the zones subject to the greatest amount of
human influence were in fact the more restricted
zones (contrary to their management objectives);
we found that the areas that were subject to less
human influence did support higher diversity and
occupancy, supporting the hypothesis that human
activity had a negative influence on biodiversity.
Evaluating the effectiveness of each management
zone was confounded by the fact that humans were
more prevalent in areas that had the most restric-
tive zoning. For example, humans and livestock
were far more abundant in the Core Area (Grid II)
of MU than in the Buffer Zone (Grid I) even though
no families or their livestock were allowed to use
the Core Area.
Additionally, for KN, tree extraction was found
to be prevalent in an area thought to be lacking vir-
tually any human influence due to its remoteness
and difficulty to access. In this case, and it is
unclear why, the carnivores were more prevalent
and diverse in the area without tree extraction
despite the presence of a variety of ungulates
(prey). These baseline occupancy estimates indi-
cated that trophic levels (or guilds) may be dispro-
portionately affected by differences in land use,
anthropogenic influence and other variables (see
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2008 2009201020112012
WPI
Year
mean
LCL
UCL
Fig. 5.4. The WPI plotted for 2009–2011 for Grid II
(Core Area), Myangan Ugalzat National Park, Mongolia
(mean = geometric mean, LCL = lower confidence limit,
UCL = upper confidence limit).
also Ahumada et al. 2011). For protected area man-
agement, these baseline estimates can be used to
set quantitative goals and measure the influence of
land use or management changes over time. Addi-
tionally, these baseline estimates and documenta-
tion of presence or absence can provide essential
information on how well certain areas are meeting
conservation objectives or goals.
The WPI was calculated for the MU study area.
The WPI (relative to the first year) showed a down-
ward trend in Grid I and relatively steady trend
(little change) in Grid II over the 3 years. Both
grids had a slight increase in biodiversity from
2009 to 2010. However, in 2011 Grid I had a 38%
decrease relative to 2009. By contrast, Grid II had
no change from 2009 to 2011 indicating that the
park was conserving biodiversity at baseline rates.
However, the numerous livestock and humans
present within the Core Area (Grid II) are likely
adversely affecting biodiversity. The WPI could
prove to be a very practical, quantitative tool to
evaluate the functioning of protected areas in
Mongolia and towards setting and obtaining con-
servation and management goals.
Mongolia represents a relatively untouched
ecosystem despite having been inhabited by
nomadic herders for thousands of years at low
densities. The landscape has little development,
fencing or paved roads. This relatively untouched
status is changing rapidly due to unprecedented
levels of mining and commensurate infrastruc-
ture development in recent years. The Mongolian
ecosystems harbour most or all of its constituent
members that were historically present. Further-
more, many IUCN Red Listed species are still pre-
sent in higher relative numbers than in
surrounding countries where these species are
currently extant or have recently been extirpated
(e.g. ibex, argali and snow leopards). Finally, Mon-
golia may represent a continental bellwether for
climatic change due to its isolation from moderat-
ing maritime influences. For all of these reasons,
using the WPI as a monitoring tool to measure
trends in biodiversity, and also for important spe-
cies about which we know little in terms of preva-
lence and distribution, shows real promise for
EBSCOhost - printed on 5/17/2020 5:18 PM via DEAKIN UNIVERSITY LIBRARY. All use subject to https://www.ebsco.com/terms-of-use
PArt 1 – cAmErA trAPPIng for AnImAL monItorIng: cAsE studIEs
52
temperate regions where biodiversity and ecosys-
tem health may rely on much larger geographic
areas.
Acknowledgements
Funding and support were provided by the Zoo-
logical Society of London, NEMO2,World Bank and
the United Nations Development Program (UNDP).
Additional invaluable in country support was pro-
vided by the Mongolia Program of the Wildlife
Conservation Society, the National University of
Mongolia, the Park Administration of Mongolia
and Border Defence Agency and, from the World
Bank, Tony Whitten.
References
Ahumada JA, Silva CEF, Gajapersad K, Hallam C,
Hurtado J, Martin E, McWilliam A, Mugerwa B,
O’Brien T, Rovero F, Sheil D, Spironello WR,
Winarni N, Andaelman SJ (2011) Community
structure and diversity of tropical forest mam-
mals: data from a global camera trap network.
Philosophical Transactions of the Royal Society of
London. Series B, Biological Sciences 366, 2703–2711.
Buckland ST, Margurran AE, Green RE, Fewster
RM (2005) Monitoring change in biodiversity
through composite indices. Philosophical Transac-
tions of the Royal Society of London. Series B, Bio-
logical Sciences 360, 243–254.
Burton AC, Sam MK, Kpelle DG, Balangtaa C,
Buedi EB, Brashares JS (2011) Evaluating persis-
tence and its predictors in a West African carni-
vore community. Biological Conservation 144,
2344–2353.
Clark EL, Munkhbat J, Dulamtseren S, Baillie JEM,
Batsaikhan N, Samiya R, Stubbe M (2006) Mon-
golian Red List of Mammals. Regional Red List
Series Vol. 1. (Zoological Society of London,
London).
Collen B, Howard R, Konie J, Daniel O, Rist J (2011)
Fields surveys for the endangered pygmy hip-
popotamus Choeropsis liberiensis in Sapo
National Park. Oryx 45, 35–37.
Hines JE (2006) PRESENCE v3.2 – Software to esti-
mate patch occupancy and related parameters.
USGS-Patuxent Wildlife Research Center.
Karanth KU, Nichols JD, Kumar NS (2011) Estimat-
ing tiger abundance from camera trap data: field
surveys and analytical issues. In AF O'Connell,
JD Nichols and KU Karanth (Eds) Camera Traps
in Animal Ecology: Methods and Analyses. Pp.
97–117. (Springer, London).
MacKenzie DI, Nichols JD, Hines JE, Knutson MG,
Franklin AD (2003) Estimating site occupancy,
colonization, and local extinction when a species
is detected imperfectly. Ecology 84, 2200–2207.
MacKenzie DI, Nichols JD, Sutton N, Kawanishi K,
Bailey LL (2005) Improving inference in popula-
tion studies of rare species that are detected
imper fectly. Ecology 86, 1101–1113.
O’Brien TG (2010) Wildlife Picture Index: Implementa-
tion Manual v. 1.0. Wildlife Conservation Society
Working Paper No. 39. (Wildlife Conservation
Society, New York).
O’Brien TG, Baillie JEM, Krueger L, Cuke M (2010)
The Wildlife Picture Index: monitoring top
trophic levels. Animal Conservation 13, 335–343.
O’Brien TG, Kinnaird MF (2013) The Wildlife Pic-
ture Index: a biodiversity indicator for top
trophic levels. In B Collen, N Pettorelli, S Durant,
J Baillie and L Krueger (Eds) Biodiversity Moni-
toring and Conservation: Bridging the Gaps between
Global Commitment and Local Action. Pp. 45–70.
(Wiley-Blackwell Publishing, Cambridge).
Pettorelli N, Lobora AL, Msuha MJ, Foley C, Durant
SM (2010) Carnivore biodiversity in Tanzania:
Revealing the distribution patterns of secretive
mammals using camera traps. Animal Conserva-
tion 13, 131–139.
Silveira L, Jacomo ATA, Diniz-Filho JAF (2003)
Camera trap, line transect census, and track sur-
veys: a comparative evaluation. Biological Con-
servation 114, 351–355.
Wingard JR, Zahler P (2006) Silent Steppe: The Illegal
Wildlife Trade Crisis in Mongolia. Mongolia Discus-
sion Papers, East Asia and Pacific Environment and
Social Development Department. (World Bank,
Washington).
EBSCOhost - printed on 5/17/2020 5:18 PM via DEAKIN UNIVERSITY LIBRARY. All use subject to https://www.ebsco.com/terms-of-use
