Content uploaded by Gill Braulik
Author content
All content in this area was uploaded by Gill Braulik on Nov 19, 2014
Content may be subject to copyright.
ENDANGERED SPECIES RESEARCH
Endang Species Res
Vol. 17: 201–215, 2012
doi: 10.3354/esr00425 Published online June 15
INTRODUCTION
The world’s freshwaters are biodiversity hotspots,
inhabited by almost 6% of described species yet cov-
ering only 0.8% of the earth’s surface. However, they
are also hotspots of endangerment and are experi-
encing declines in biodiversity far greater than those
in terrestrial ecosystems (Dudgeon et al. 2006,
Strayer & Dudgeon 2010). Freshwater dolphins and
porpoises are among the world’s most threatened
© Inter-Research 2012 · www.int-res.com*Email: gillbraulik@downstream.vg
†Deceased
Robust abundance estimate for endangered river
dolphin subspecies in South Asia
Gillian T. Braulik1,2,*, Zahid I. Bhatti3, Tahir Ehsan2, Babar Hussain4,
Abdul R. Khan5, Ashfaq Khan6, Uzma Khan7, Khalil U. Kundi8, Rafiq Rajput9,†, Albert
P. Reichert2,10, Simon P. Northridge1, Hussain B. Bhagat9, Richard Garstang2
1Sea Mammal Research Unit, Scottish Oceans Institute, University of St. Andrews, St. Andrews, Fife KY16 8LB, UK
2Pakistan Wetlands Programme, Islamabad, Pakistan
3Lahore Zoological Gardens, Shahrah-e-Quaid-e-Azam, Lahore, Pakistan
4World Wide Fund for Nature (WWF)-Pakistan, Fortune Center, Pakistan Employees Cooperative Housing Society (PECHS),
Shahrae Faisal, Karachi, Pakistan
5Halcrow Pakistan, PECHS, Karachi, Pakistan
6Adventure Tourism Section, Tourism Development Pakistan, Lahore, Pakistan
7WWF-Pakistan, Lahore, Pakistan
8Khyber Pakhtunkhwa (KPK) Wildlife Department, Peshawar, Pakistan
9Sindh Wildlife Department, Karachi, Pakistan
10Downstream Research Group, Macon, Georgia 31201, USA
ABSTRACT: Robust estimates of absolute abundance are vital for management of threatened spe-
cies, but these have rarely been generated for endangered South Asian river dolphins due to
methodological challenges. An estimate of abundance for the Indus River dolphin in 2006 was gen-
erated by conducting tandem vessel-based direct counts; conditional likelihood capture-recapture
models were then used to correct for missed animals. Group size and sighting conditions were in-
cluded as covariates, and abundances of the 3 largest subpopulations were estimated as 101 (coef-
ficient of variation, CV = 44.1%) between Chashma and Taunsa barrages, 52 (CV = 14.9%) be-
tween Taunsa barrage and Ghazi Ghat, and 1289 (CV = 33.4%) between Guddu and Sukkur
barrages. A total of 75.3% of groups were seen by both independent survey teams, and single ani-
mals were almost 5 times more likely to be missed than groups of 3 or more. Providing groups can
be matched with minimal error, this survey method shows good potential for abundance estimation
of dolphins in confined habitat and the shallow rivers of South Asia. Dolphin encounter rates within
the Guddu-Sukkur subpopulation (10.35 km−1) are the highest reported for any river dolphin.
Direct counts conducted over a 35 yr period, suggest that this subpopulation may have been in-
creasing in abundance, probably due to the cessation of hunting and possible immigration from
other subpopulations. The future of South Asian river dolphins is intimately tied to water security
in the region, and escalating and competing demands for freshwater mean that the long-term
future of South Asia’s river dolphins is uncertain.
KEY WORDS: River dolphins · Platanista · Abundance · Trends · South Asia · Capture-recapture ·
Pakistan
Resale or republication not permitted without written consent of the publisher
O
PEN
PEN
A
CCESS
CCESS
Endang Species Res 17: 201–215, 2012
mammals. Nine river systems or brackish lagoons in
Asia have resident cetacean populations, and all are
listed on the IUCN Red List as Endangered or Criti-
cally Endangered (with the exception of two that are
yet to be evaluated) (IUCN 2010). After decades of
concern about its status, the baiji, or Yangtze River
dolphin Lipotes vexillifer, the sole representative of
an entire mammalian family, is believed to be extinct
(Turvey et al. 2007). Asian river dolphins share their
lowland river habitat with hundreds of millions of
people, resulting in high mortality rates from hunting
or entanglement in fishing gear, populations frag-
mented by dams and barrages, and habitat severely
depleted by water extraction and degraded by pollu-
tion and altered flow regimes. The baiji’s extinction
clearly demonstrates that without appropriate and
timely action, the future of the remaining freshwater
cetaceans is precarious.
Status of the Indus dolphin
The South Asian river dolphin Platanista ganget-
ica occurs in a monotypic genus that includes the
Indus River dolphin Platanista gangetica minor, re -
sident in the Indus River system in Pakistan and
India, and the Ganges River dolphin Platanista
gangetica gangetica found in the Ganges-Brahma-
putra and Karnaphuli-Sangu River systems in In -
dia, Bangladesh, and Nepal. The species and both
subspecies are listed as Endangered in the IUCN
Red List (Braulik et al. 2004, Smith et al. 2004).
The Indus River rises in Tibet, flows through the
Him alayas and Karakoram mountains and then
flows south for approximately 2000 km across arid,
intensively cultivated and densely populated plains
of Pakistan. The lower Indus River system has been
transformed into an intensively managed irrigation
scheme, including several high dams, multiple
inter-river link canals, and numerous barrages
(gated-diversion dams) that divert the majority of
the river’s flow into canals to irrigate agricultural
land. The Indus dolphin persists within this system,
but its range has declined by 80%, its population is
fragmented by barrages, and its habitat is degraded
due to water diversion. In 2001, dolphins were
recorded within around 1000 km of the Indus main-
stem in 5 subpopulations separated by barrages,
and the metapopulation was estimated as approxi-
mately 1200 (Braulik 2006). A sixth very small dol-
phin subpopulation, located more than 600 km
from all the others, was recently discovered in the
Beas River, India (Behera et al. 2008).
Methodological challenges to surveying
Platanista spp.
The ability to estimate abundance with relative
accuracy and precision is imperative for assessing
the status of endangered species and monitoring the
effectiveness of conservation measures. The Sub-
committee on Small Cetaceans of the International
Whaling Commission (IWC) noted in 2000 that few
reliable abundance estimates were available for any
species of freshwater cetacean and that the habitat
and behaviour of these animals posed particular
problems for abundance estimation (IWC 2001).
Identifying an appropriate survey methodology for
South Asian river dolphins is especially challenging
as the 2 principal methods used to estimate abun-
dance, i.e. distance sampling and photo-identifica-
tion, are difficult or impossible to apply (Smith &
Reeves 2000, Dawson et al. 2008). In the absence of a
more robust method, direct counts in discrete river
sections have generally been conducted. The counts
have not included measures of precision and are
underestimates of the real population size, as no cor-
rection was made for animals that were missed when
they were underwater (availability bias) or that sur-
faced in view but were not recognised (perception
bias) (Marsh & Sinclair 1989, Smith et al. 2006). The
IWC sub-committee strongly recommended that
appropriate methods be developed so that surveys of
river cetaceans result in statistically robust estimates
of abundance (IWC 2001).
Capture-recapture analysis of photo-identified ani-
mals is commonly used to estimate abundance of
cetaceans (Hammond 2009). This method relies on
capturing images of uniquely marked animals; the
proportion of identified individuals recaptured dur-
ing subsequent sampling events is then used to esti-
mate population abundance (Borchers et al. 2002).
This method has very limited possibilities for Pla-
tanista spp. because (1) they are extremely difficult to
photograph as they surface alone, unpredictably, for
about one second and they do not approach boats,
and (2) they lack a prominent dorsal fin and rarely
possess any readily identifying features. Not a single
individual could be identified from 1200 photographs
of Ganges River dolphins (Smith & Reeves 2000).
The primary challenge to the application of line or
strip transect methods in the Indus, Ganges and
Brahmaputra Rivers is that rivers are very shallow
and survey vessels are restricted to travelling down
the thalweg (the line that follows the deepest part of
the river) along a single curving transect that periodi-
cally approaches alternate banks as the river mean-
202
Braulik et al.: Abundance of Indus River dolphins
ders. Indus dolphin distribution is biased towards the
deeper water found along the survey transect (G.
Braulik unpubl. data). To avoid the unrealistic as-
sumption that animals in the population are randomly
distributed when using distance sampling, the ran-
dom or systematic placement of an adequate number
of transects is required. This allows for reliable ex-
trapolation (1) from observations made in the sampled
area to the entire region of interest, and (2) of the ob-
served perpendicular sighting distances to estimate
the proportion of animals counted. A thalweg transect
survey unavoidably samples unrepresentative habitat
as it passes through areas with higher densities, and
in addition the animals are unlikely to be uniformly
distributed in the surveyed strips. Other less signifi-
cant challenges to distance sampling in this environ-
ment include measuring perpendicular sighting dis-
tances when surveying moving objects from a sharply
curving path (Hiby & Krishna 2001), frequent con-
strictions in the river channel that cut off the potential
detection width causing a narrowing or unusual
shoulder in the detection function (Dawson et al.
2008), and the presence of a downstream population
density gradient (Braulik 2006) that complicates ex-
trapolation of data from one area to another.
Transects running from bank to bank, perpendicu-
lar to the flow, are used for line transect surveys of
cetaceans in the Amazon River (Vidal et al. 1997,
Martin & da Silva 2004), but in the comparatively
shallow, sand-bedded, South Asian rivers naviga-
tional constraints preclude this approach. A single
transect parallel to, and a standard distance from, the
river banks has also been used for strip transect sur-
veys in the Amazon (Vidal et al. 1997, Martin & da
Silva 2004) and for adapted line transect surveys on
the Yangtze River (Zhao et al. 2008), but this is not
possible on South Asian rivers as channel width
changes rapidly and vessels cannot maintain a stan-
dard distance from the banks due to shallows. Aerial
surveys have not been attempted for South Asian
river dolphins, but high water turbidity would pre-
vent animals being detected below the surface, and
the extremely brief surfacing time would make
detection from above unlikely.
Aerial surveys of terrestrial and marine mammals
frequently obtain simultaneous counts using inde-
pendent observer teams, so that mark-recapture can
be used to correct abundance estimates for missed
animals (Samuel & Pollock 1981, Graham & Bell
1989, Marsh & Sinclair 1989, Crete et al. 1991, Car-
retta et al. 1998, Hiby & Lovell 1998). A similar
method using independent teams on a single vessel
was used by Smith et al. (2006) to estimate abun-
dance of Ganges River dolphins and Irrawaddy dol-
phins Orcaella brevirostris in the Bangladesh Sun-
darbans. We have adapted these methods, and pre-
sent here an appropriate and robust method for
estimating abundance of dolphins in confined habi-
tats such as rivers in South Asia. Abundance of Indus
River dolphins was estimated by conducting direct
counts using independent observation teams on ves-
sels travelling in tandem along a thalweg transect,
and conditional-likelihood capture-recapture models
were used to correct for missed animals.
MATERIALS AND METHODS
Field surveys
The survey was conducted from 23 March to 24
April 2006 when Indus discharge was at its annual
low. The expedition covered habitat in each of the 5
extant Indus dolphin subpopulations, including river
channels between Jinnah barrage and Ghazi Ghat
bridge (65 km downstream of Taunsa barrage),
Guddu and Sukkur barrages and Sehwan Sharif and
Kotri barrage (Fig. 1). Field methods were the same
as those employed during a direct count survey in
2001 (see Braulik 2006). Two oar-powered boats trav-
elled downstream, in tandem and separated by 1 to
3 km (5 to 35 min). Significant secondary channels
were identified using satellite images and were sur-
veyed by the rear boat, while the forward vessel con-
tinued along the main channel in non-tandem survey
effort. All observers received training prior to the sur-
vey and each vessel had a minimum of 2 observers
with prior dolphin survey experience. Survey effort
was recorded and the effect of wind on the river sur-
face evaluated according to the following ‘river state’
scale: 0 = water surface glassy; 1 = ripples without
crests; 2 = small wavelets with crests but no white-
caps; 3 = large wavelets with scattered white-caps;
4 = small waves with fairly frequent white-caps.
When dolphins were sighted, their distance was esti-
mated, and nearby objects, such as the river banks,
were measured with laser range-finders to improve
the accuracy of distance estimates. Generally, dol-
phins were sighted downstream of the survey vessels
and remained relatively stationary, so that the ves-
sels approached and passed through groups while
proceeding downstream. We recorded a ‘detection
location’ with a GPS when a dolphin was first sighted
and an ‘exact location’ when the centre of the group
was perpendicular to the vessel. Animals judged to
be less than 1 m in length were designated as calves.
203
Endang Species Res 17: 201–215, 2012
Indus dolphins typically occur as loose aggregations
of individuals, so following previous authors, a group
was defined as animals occurring within 500 m in
similar fluvial habitat such as meander bends or
channel constrictions (Smith et al. 2006). In the lower
half of the Guddu-Sukkur river section, encounter
rates were very high; dolphins occurred continuously
with no obvious gaps between groups, and we used
river features, such as constrictions, or mid-channel
bars, to delineate groups and facilitate counting.
Identification of matched sightings
All sighted dolphin groups were considered to be
‘captured’ and were then classified as duplicates or
‘matches’ if they were seen by both survey vessels, or
unique if they were ‘missed’ by 1 boat. Sightings were
classified as matches based on the distance between
their ‘exact’ geographic positions com-
bined with any group movement di-
rection noted in the field. Using Arc -
View 3.2, the distance between
dolphin groups was measured along
the centre of the river channel, and a
frequency distribution of the distance
between the exact geographic posi-
tions of potentially matched dolphin
groups generated. The obvious clump-
ing of distances was used to guide se-
lection of an appropriate distance
threshold with which to classify groups
as matched. Use of a small threshold
distance would result in recognition of
more missed sightings and a larger
abundance estimate; conversely a
wide threshold distance would gener-
ate fewer missed sightings and a
smaller estimate of abundance. Field
observations indicate that Indus dol-
phins often remain in circumscribed
habitat for periods of up to an hour (G.
Braulik pers. obs.) and that a relatively
small threshold distance could be ap-
plied. We selected a threshold distance
that allowed for some dolphin move-
ment between detections but did not
specifically model movement because
no quantitative data have been pub-
lished on the movements or swimming
speeds of Indus or Ganges dolphins
that could reliably contribute to a
model. We assumed that matches
were made without error; however, to test how robust
our results were to the selection of threshold distance,
we conducted a sensitivity analysis to compare the in-
fluence of 6 candidate thresholds (300 to 800 m) on the
number of sightings identified as matched.
Abundance estimation
Estimating number of groups
Abundance was estimated separately for each sub -
population. Sightings made during tandem survey
effort were analysed using mark-recapture for closed
populations in a Huggins conditional-likelihood
model implemented by the program MARK. Capture
probabilities are modelled as a function of sighting
covariates according to the following formula
(Huggins 1989):
204
Fig. 1. Map of the Indus River system illustrating the barrages that form the
boundaries between the 5 dolphin subpopulations. Each subpopulation is de-
noted by the following acronyms that include the sequential number of the
subpopulation and the barrages it is bounded by: 1J-C (Jinnah to Chashma),
2C-T (Chashma to Taunsa), 3T-G (Taunsa to Guddu), 4G-S (Guddu to
Sukkur), 5S-K (Sukkur to Kotri). MD: moderate density subsection of the
Guddu to Sukkur subpopulation; HD: high density subsection of the Guddu
to Sukkur subpopulation
Braulik et al.: Abundance of Indus River dolphins
(1)
where logit = link function, p = the probability of
capture, i= forward or rear survey boat, k= dolphin
group, j= covariate, β0= intercept and βj= slope for
covariate value xj.
Group abundance and variance during tandem
effort was estimated within MARK using a Horvitz-
Thompson like estimator (Horvitz & Thompson 1952):
(2)
where gˆ tis the estimated number of groups present
during tandem effort and pˆ kis the estimated proba-
bility that group kwas detected by either platform.
Covariates included in the model were selected
based on similar studies, combined with knowledge of
the Indus River environment and of Indus dolphin be-
haviour. Perception bias of Ganges dolphins was in-
fluenced only by group size, not by sighting conditions
or channel width (Smith et al. 2006). Detection of har-
bour porpoise Phocoena phocoena was influenced by
group size and sea state (Hammond et al. 2002), and
the vaquita P. sinus by sea state (Gerrodette et al.
2011). We included group size, river surface state and
sighting vessel as covariates in models for each sub-
population. Group sizes used were those recorded in
the field. Although possible error in group size estima-
tion is not explicitly accounted for, the effect on the
abundance estimate and variance is expected to be
small. When group size estimates for matched groups
differed, the estimate from the forward vessel was al-
ways used, as this was considered more reliable. Mod-
els were selected using Akaike’s information criterion
for small sample sizes (AICc), according to recom-
mended guidelines (Burnham & Anderson 2002). To
account for uncertainty in model selection, if the best
fitting models were separated by less than 2 AICc
points they were averaged on the basis of their nor-
malized AICc weights (Stanley & Burnham 1998).
Estimating mean group size
Mean group size was estimated ignoring errors in
recorded size but attempting to correct for smaller
groups being less detectable. To produce an estimate
of mean group size (s¯)in each subpopulation, the
detected number of groups of each group size (nj)
were corrected by the average detection probability
of a group of particular size (pˆ j) output by MARK, and
this was used to estimate a group size distribution,
from which the mean was determined:
(3)
The variance in estimated mean group size was
generated from the sample variance of the estimated
recorded group sizes, after adjusting for variability in
detection probability.
Non-tandem effort
The capture-recapture method described above was
reliant on tandem survey effort; however, portions of
non-tandem survey effort were conducted in all river
sections. In addition, the 4G-S subpopulation (Fig. 1)
was subdivided into 2 strata: the upper 106 km with
moderate dolphin density (MD) (3.63 dolphins linear
km−1), and a lower 98 km with high density (HD) (9.06
dolphins linear km−1). In the HD subsection there were
no obvious gaps between aggregations, which made it
impossible to determine whether sightings from the 2
vessels matched. Capture-recapture using tandem
survey data was therefore not conducted on the HD
subsection, and these data were treated as non-tan-
dem. To account for groups missed in each subpopula-
tion during non-tandem survey periods we applied a
correction factor (ƒm) determined from the tandem-
effort survey: ƒm=gˆ t/gft where gft is the number of
groups seen by the forward vessel during tandem ef-
fort. The group size from tandem effort was applied to
sightings made during nearby non-tandem effort.
However, in the 4 G-S HD subsection, which was sur-
veyed entirely in non-tandem effort, group size was
substantially larger than in all other areas, and cor-
rected group size was therefore calculated using the
method described above, applying the group size de-
tection probabilities determined from tandem survey
effort in the 4 G-S MD subsection. Sighting conditions
in side channels were very different from the main
channel, and it was considered inappropriate to apply
the main channel group correction factor to these ar-
eas, so individuals seen in side channels (ns) were
added to main channel sightings without correction.
Estimating abundance
A binomial generalised linear model (GLM) was used
to test the relationship between the time lag between
tandem vessels and the probability that a dolphin
group was missed. Based on this, where the time lag
between survey vessels was long, tandem survey data
logit ln()pp
px
ik
ik
ik
jjk
j
=−
⎛
⎝⎞
⎠=+
∑
10
ββ
ˆˆvar( ˆ)ˆ(ˆ)gpgpp
k
k
g
k
k
g
itt
tt
==−
==
∑∑
11
1
11
s
nj
p
n
p
j
j
j
s
j
j
j
s
=
⋅
=
=
∑
∑
ˆ
ˆ
max
max
1
1
205
Endang Species Res 17: 201–215, 2012
were reclassified as non-tandem, so that the probability
that a sighting was missed was in dependent of time
lag. Abundance of dolphins in each subpopulation seen
during tandem survey effort N
ˆtwas estimated as gˆ ts
–t
and the coefficient of variation, CV:
(4)
Abundance of dolphins in each subpopulation seen
during non-tandem survey effort N
ˆnt was estimated
as gˆ nt s
–nt + nsand the CV:
(5)
The CV of the correction factor is derived from:
(6)
Total subpopulation abundance was generated by
summing the tandem and non-tandem sightings, and
total metapopulation abundance by totalling the
abundance in each subpopulation. CVs were com-
bined using the delta method, and if there were
shared factors between strata these were factored
out to account for covariance (Buckland et al. 2001,
Gerrodette et al. 2011).
Log-normal confidence intervals (CI), where the
lower limit cannot be smaller than the number of
unique individuals sighted (Mt+1), were calculated
according to the following (Williams et al. 2002):
(7)
where:
(8)
The confidence intervals generated for the current
study are particularly skewed because the lower
bounds are constrained to lie between the abun-
dance estimates and Mt+1, and as sighting probabili-
ties were high these values are similar. By contrast,
the upper bound of the confidence interval is uncon-
strained and is influenced by the precision of the
abundance estimates.
Availability bias
The contribution of dolphin availability to total
detection bias (corrected for by the tandem surveys)
was investigated using radial dolphin sighting dis-
tances, vessel speed, and the dive and surface
behaviour of groups of different sizes. In the dry sea-
son of 2008, between Chashma and Taunsa barrages,
dive times of groups or individuals were recorded
with a stop watch from a vantage point on the river
bank. A dive time was the interval between surfac-
ings that lasted longer than 2 s. Group size was
recorded when dive time monitoring began, and if
this subsequently changed, the time and new group
size were recorded. Water depth at group locations
was not specifically measured but included a range
of those used by the subspecies. Surfacing of this
species is so rapid and unpredictable that it is not
possible to accurately measure surface interval in the
field; therefore, surfacings were recorded with digi-
tal video and surface time measured by sequentially
viewing each frame. Footage was recorded at 25
frames s−1, resulting in surface intervals accurate to
0.08 s. The probability that a group was available to
be seen by observers was determined according to
the following (Barlow et al. 1988): a = st + w / st + d,
where a = the number of surfacings when a survey
vessel was present, st = mean surface time, d= mean
dive time, and w= the time window that individuals
or groups were within range of observers, deter-
mined using sighting distances and vessel speed.
RESULTS
The survey covered 808 km of the Indus River main
channel and 126 km of adjacent secondary channels.
Environmental conditions were excellent: 92% of sur-
vey effort was conducted in a river surface state of 2 or
less, and 46% was in glassy conditions. There was no
significant difference between the daily counts re -
corded by each boat (paired t-test, p = 0.704). Mean
dolphin radial sighting distance was 401 m (SD =
279.1 m), consistently greater than half the mean river
width (200 to 300 m), and sightings often occurred at
distances up to 1 km (Fig. 2). Although these are radial
distances, not perpendicular distances, this still illus-
trates that the majority of surfacings within the river
channel could be detected. Dolphin encounter rate
and mean group size increased from the northern ex-
treme of the range downstream to Sukkur barrage. As
mean group size increased, the distance between
groups decreased. Direct counts derived from the sum
of group size estimates of the forward survey vessel
plus animals sighted in secondary channels totalled: 1
in section 1J-C (Jinnah to Chashma barrages); 82 in
section 2C-T (Chashma to Taunsa barrages); 44 be-
tween Taunsa barrage and Ghazi Ghat bridge (3T-
GG); 1275 in section 4G-S (Guddu to Sukkur bar-
rages); and 4 in section 5S-K (Sukkur to Kotri
barrages). Calves accounted for approximately 14%
CV CV CV
ttt
(ˆ)(
ˆ)()[]Ngs≈+
22
CV CV CV
nt m nt
(ˆ)(
ˆ
ƒ) ( )[]Ns≈+
22
SE mt
ft
(ˆ
ƒ) var( ˆ)
=g
g
Lower CI = Upper CI =
t+1 t
MCMC+⎛
⎝
⎜⎞
⎠
⎟+
+
ˆ
ƒ(ˆ
ƒ
010
))
ˆ
ƒˆ,
exp . var( ˆ)
ˆ
ƒ
0
02
196 1
=−
=+
⎛
⎝
NM
CN
t+1 and
ln⎜⎜ ⎞
⎠
⎟
⎡
⎣
⎢⎤
⎦
⎥
⎧
⎨
⎪
⎩
⎪
⎫
⎬
⎪
⎭
⎪
12
206
Braulik et al.: Abundance of Indus River dolphins
of total individuals in 2C-T, 7% in 3T-GG, and 11% in
the 4G-S section. Dolphins were sighted in secondary
channels only in the 2C-T (4 ind.) and 4G-S (5 ind.)
sections, and encounter rates in these channels were
very low, 0.08 km−1 and 0.3 km−1, respectively, com-
pared to adjacent uncorrected main channel en-
counter rates of 0.27 km−1 and 6.23 km−1, respectively.
Identification of matched sightings
Matching of sightings in most areas was unam-
biguous because there were long distances between
detections, encounter rate was low (<1 dolphin
km−1), groups can only move in 2 directions, up- or
downstream, and most potentially matched groups
were very close to one another. Of potentially
matched sightings (those within 2 km) the geo-
graphic locations of 88% occurred within 600 m, and
78% were within 400 m (Fig. 3). Six hundred metres
was selected as the appropriate threshold distance
for determining matched sightings because it encom-
passed the majority of probable matches, allowed for
some group movement between surveys, and was
greater than the 500 m distance used to define a
group. In the matching sensitivity analysis, contrast-
ing the 300, 400, 500, 700 and 800 m distance thresh-
olds to the 600 m threshold resulted in different
missed/matched classifications for a maximum of 5
groups (26 versus 31 matches) and less than 5% dif-
ference in the corrected number of groups estimated
by the Huggins mark-recapture model (Table 1). The
largest differences corresponded to the narrowest
300 m threshold. The sensitivity analysis clearly
demonstrates that changing the threshold distance
used to define matched groups does not exert a great
influence on the resulting estimates of abundance.
Estimation of abundance
Based on the GLM, 4 sightings that occurred when
there was more than 35 min separating the vessels
207
Distance nfnrnfr gˆ tSE(gˆ t)
threshold (m)
300 41 41 26 56.8 3.6
400 41 41 29 55.1 2.8
500 41 41 31 54.2 2.4
600 41 41 31 54.2 2.4
700 41 41 32 53.8 2.2
800 41 41 32 53.8 2.2
Table 1. Comparison of different distance thresholds used to
identify matched and missed Indus River dolphin Platanista
gangetica minor sightings with the 600 m distance (in bold)
finally selected. Number of groups seen by the forward ves-
sel (nf), rear vessel (nr) and by both vessels (nfr) during tan-
dem survey effort between Chashma and Ghazi Ghat; gˆ t:
correct number of groups calculated using the Huggins
model; SE(gˆ t): standard error of the correct number of
groups
Fig. 2. Frequency of radial sighting distances of the Indus
River dolphin Platanista gangetica minor. The available
survey strip, represented by half mean river width, varies
between 200 and 300 m depending on the section of river
Fig. 3. Distance between the exact geographic positions of
potentially matched groups of Platanista gangetica minor.
Vertical line indicates the 600 m distance threshold selected
to classify sightings as matched in the tandem surveys
Endang Species Res 17: 201–215, 2012
were reclassified as occurring during non-tandem
effort, and in the remaining dataset the probability
that a dolphin group was missed was independent of
survey time lag (GLM, p = 0.12). In 50% of tandem
survey effort, the vessels were separated by less than
10 min, and in 75%, the boats were less than 20 min
apart. Sighting probability was high; 75.3% of
groups were seen by both independent survey
teams, and single animals were almost 5 times more
likely to be missed than groups of 3 or more. Mark-
recapture analysis was conducted on the tandem sur-
vey data from each of the 3 largest dolphin subpopu-
lations but was not conducted on the sightings that
occurred between Jinnah and Chashma barrages (1
animal) and Sukkur and Kotri barrages (4 animals)
due to the small sample size.
Chashma to Taunsa
In the 2C-T section, during tandem survey effort
the forward boat recorded 27 groups, the rear boat
recorded 26 groups, and 18 sightings were matches,
with 17 unique. Nine groups were sighted during
non-tandem main channel effort, and 4 individuals
were recorded in side channels. Missed groups were
significantly smaller than matched groups (Mann-
Whitney U-test, W= 258.5, p < 0.0001) (Fig. 4). Of the
35 sightings, 48.6% were missed by 1 team, includ-
ing 77.8% of single animals, and 37.5% of groups of
2. All groups of 3 or more were sighted by both ves-
sels (Fig. 5). There was no significant effect of river
state on the proportion of sightings that were missed
(Mann-Whitney U-test, W= 170.5, p = 0.76). The best
supported model was separated by more than 2 AICc
points from all others and included a single capture
probability influenced only by the covariate group
size. The mean group size observed in the field, 1.98,
was corrected for size bias to 1.50 (CV = 8%) based
on sighting probabilities generated by the model.
The final abundance estimate for this subpopulation
was 101 (95% CI = 74−317, CV = 44.1%) (Table 2).
Taunsa to Ghazi Ghat
Between Taunsa barrage and Ghazi Ghat, 14
groups were seen by the forward vessel and 15 by
208
Fig. 4. Frequency of missed and matched sightings of Platanista gangetica minor (A) between Chashma and Taunsa barrages
by group size and (B) by river surface state (see ‘Materials and methods; Field surveys’ for details), and (C) between Guddu
and Sukkur barrages by group size and (D) by river surface state. Number of groups is given at the top of each column
Braulik et al.: Abundance of Indus River dolphins
the rear vessel. Thirteen were classified as matched,
and only 3 were missed. There were 3 non-tandem
sightings and no groups recorded in side-channels in
this section. The top 3 candidate models that were
averaged included the influence of river surface state
on sighting probability. As the final models did not
include the covariate group size, the mean group size
recorded in the field 2.63 (CV = 12.7%) was assumed
to be unbiased. Abundance for this portion of the 3T-
G subpopulation was estimated to be 52 (95% CI =
50−102, CV = 14.9%) (Table 2).
Guddu to Sukkur
In the MD subsection of the 4G-S subpopulation,
43 groups were seen by the forward vessel, 35 by
the rear vessel, and 33 were classified as matched.
Twenty seven groups were seen during non-tandem
main-channel effort and 5 individuals in a side
channel. All groups of 6 or more individuals were
seen by both vessels, but 50% of single animals
were missed. Matched groups were significantly
larger than those that were missed (Mann-Whitney
U-test, W= 88, p < 0.01) and there was no obvious
effect on sighting probability attributable to river
state (Mann-Whitney U-test, W= 254, p = 0.13)
(Fig. 4). The model with the lowest AICc included
the covariate group size and a different capture
probability for each vessel. Corrected group size
estimates were 4.73 (CV = 11.0%) in the MD sub-
209
River section nf nr nfr Mt+1 pˆ ƒm ƒmCV gˆ gˆ CV ns s¯ s¯CV N
ˆ N
ˆCV 95% CI er
Strata
2 C-T
Tandem 27 26 18 – 0.864 – – 48.5 21.1 – 1.50 8.0 73 22.5 – –
Non-tandem 9 – – – – 1.80 37.9 16.2 – 4 1.50 8.0 28 38.7 – –
Total – – – 70 – – – – – – – – 101 44.1 74 – 317 0.34
3 T-GG
Tandem 14 15 13 – 0.999 – – 16.4 5.0 – 2.63 12.7 43 13.7 – –
Non-tandem 3 – – – – 1.17 5.80 3.5 – 0 2.63 12.7 9 14.0 – –
Total – – – 50 – – – – – – – – 52 14.9 50 –102 0.80
4 G-S MD
Tandem 43 35 33 – 0.879 – – 47.0 18.0 – 4.73 11.0 223 21.1 – –
Non-tandem 27 – – – – 1.09 19.7 29.5 – 5 4.73 11.0 145 22.5 – –
4 G-S HD
Non-tandem 91 – – – – 1.09 19.7 99.5 – 0 9.26 9.1 922 21.7 – –
Total – – – 1189 – – – – – – – – 1289 33.4 1192 – 4120 6.30
Grand total – – – 1309 – – – – – – – – 1442 57.2 1312–7014
Table 2. Summary of Indus dolphin Platanista gangetica minor subpopulation abundance estimation. For river sections see
Fig. 1. MD: moderate density subsection of the Guddu to Sukkur subpopulation; HD: high density subsection of the Guddu to
Sukkur subpopulation. Strata refers to survey effort conducted with either tandem vessels or a single vessel, each of which
were analysed differently (see ‘Materials and methods’). CV: coefficient of variation; nf, nr, nfr: number of sightings seen by the
forward, rear and by both vessels, respectively, during tandem survey effort; Mt+1= number of unique individuals sighted dur-
ing the survey; pˆ : sighting probability; ƒm: group correction for non-tandem effort; gˆ: corrected number of tandem or non-tan-
dem effort sightings; ns: number of individuals recorded in side channels; s¯ : corrected mean group size; N
ˆ: abundance
estimate; er: encounter rate; (–): not determined/not applicable
Fig. 5. Probability that Indus dolphin Platanista gangetica
minor groups of different sizes were sighted by a survey ves-
sel. Dotted lines represent 95% confidence intervals. Data
from Chashma-Taunsa barrages only
Endang Species Res 17: 201–215, 2012
section and 9.26 (CV = 9.1%) in the HD subsection.
The group correction factor of 1.09 was applied to
the sightings in the HD sub-section to give a final
abundance estimate of 1289 (95% CI = 1192−4120,
CV = 33.4%) for this subpopulation (Table 2).
Metapopulation abundance
The sum of the above 3 abundance estimates and
the animals sighted between Jinnah and Chashma,
and Sukkur and Kotri, was 1447 (CV = 57.2%). A
total of 300 km of dolphin habitat between Ghazi
Ghat and Guddu barrage was not covered by the
present survey, and therefore to provide an approxi-
mate estimate of subspecies abundance, we include
data from previous surveys. In 2001, 200 dolphins
were recorded in the 300 km section that was missed
in 2006 (Braulik 2006). The direct counts recorded in
the surveyed portion of this subpopulation in 2001
and 2006 (45 versus 44, respectively) were very simi-
lar, indicating that no large changes have occurred
(Braulik 2006). However, conservatively allowing
abundance in the unsurveyed area to have changed
±50% in the intervening 5 yr means that there may
have been between 100 and 300 individuals in the
unsurveyed stretch in 2006, and we therefore sug-
gest that the subspecies numbered approximately
1550 to 1750 in 2006.
Evaluation of potential bias
A total of 1156 dive times were collected from 33
groups ranging in size from 1 to 5 individuals.
Group dive time steadily decreased as group size
increased because there is no synchronization in
surfacing behaviour in this species (Table 3). Dol-
phin surfacings lasted from 0.60 to 1.76 s, averaging
1.01 s (SD = 0.28; CV = 27.3%; n = 103) and 61% of
surfacings lasted for less than 1 s. Although not
specifically investigated, surface interval of individ-
uals appeared to be unaffected by group size, and
individuals seldom surfaced at the same time even
in large groups. Consequently, the proportion of
time spent at the surface increased with group size
and ranged from 1.3 to 4.7% (Table 3). A frequency
distribution of dolphin radial sighting distances
(Fig. 2) indicated that detection probability was
consistently high up to 400 m from the vessel and
then slowly declined. It would take 4.81 min, trav-
elling at the survey speed of 5 km h−1 to cover
400 m and this was used as the time window within
which animals could be detected. All of the dive-
surface cycles recorded were considerably shorter
than 4.81 min, and we conclude that the contribu-
tion of availability bias to total detection bias was
negligible. Because there are more frequent surfac-
ings in large groups, there are many more opportu-
nities for them to be detected, which explains the
lower detection probability of single animals; how-
ever, even single animals would be expected to
surface several times in view of observers (Table 3).
For matched sightings, as group size increased so
did the variability in the size estimates (GLM, p <
0.001). However, 30% of the group size estimates
were identical, and 75% of estimates were within 2
individuals despite the time delay between surveys.
DISCUSSION
Evaluation of survey methods
Closed population capture-recapture analysis
includes assumptions that, if violated, result in biased
estimates of abundance. The assumption of popula-
tion closure was met as each subpopulation is
bounded into a linear strip by the lateral river banks
and up- and downstream between irrigation bar-
rages with closed gates. The 2 surveys were sepa-
rated by less than 35 min, so demographic changes
would not have occurred. The assumption of equal
capture probability for all groups in all circumstances
was not met, but capture heterogene-
ity was accounted for in the model.
Groups moving, and therefore not
being recognized as matched (analo-
gous to capture loss), may have
occurred on occasion, but the high
proportion of matched sightings
(75.3%) and the fact that matched
sightings were on average only 200 m
from one another indicate that this
was not a significant source of bias.
210
Group Dive time (s) % Time at Surfacings within
size (n; 95%CI) surface sighting time window
1 78 (181; 60−97) 1.3 3.7
2 56 (282; 28−83) 1.8 5.1
3 51 (337; 29−72) 2.0 5.6
4 & 5 22 (356; 6−37) 4.7 12.8
Table 3. Sighting availability of Indus dolphin Platanista gangetica minor
groups calculated from the mean sighting distance of 401 m which would be
traversed in 4.8 min at the target survey speed of 5 km h−1. n: no. of dives
Braulik et al.: Abundance of Indus River dolphins
To reduce the potential for capture loss, group move-
ment direction was included in the matching process,
and the time lag between survey vessels was kept as
short as possible. However, a separation of at least 1
km was essential so that the forward vessel did not
interfere with, or influence, the ability of the rear
observers to locate dolphins.
The greatest uncertainty involved in this study is
the ability to correctly recognise captures. Individu-
als within Indus dolphin groups are often quite dis-
persed, and therefore the group locations used in
the matching process are inherently inexact. In
addition, the exact position was recorded when a
group was judged to be perpendicular to the vessel,
so does not necessarily represent the center of the
group, which may have contributed to errors in
recognising matched sightings. However, the fre-
quency distribution of distances between potentially
matched sightings demonstrates that there is little
ambiguity in identifying matches, the sensitivity
analysis showed little change in the number of
matches even when quite different distance thresh-
olds were used, and the great majority of sightings
could be readily determined as matched or missed.
A final capture-recapture assumption is that capture
does not affect the probability of recapture, and the
potential for this was also accounted for in the mod-
els. The best-fitting models for the 4G-S MD sub-
section included a lower capture probability for the
rear survey vessel, which might suggest that dol-
phins avoided the vessels in the high density area.
In all other areas, dolphins had a uniform capture
probability which indicates no vessel attraction or
avoidance behaviour. An additional source of poten-
tial downward bias in this survey is that animals
were missed because they were too distant from the
observers. To minimise this bias, geographic cover-
age of available habitat was maximised by survey-
ing the entire length of the Indus main channel and
deploying a separate boat to survey large side chan-
nels, behind islands and the far side of wide channel
habitat. Detection probability was maximised by
surveying only in excellent and good survey condi-
tions and at a relatively slow speed. Perpendicular
sighting distance could not be generated; however,
mean sighting distance was consistently greater
than half the mean river width, so the assumption
that the majority of dolphins within the river channel
could be detected is not unreasonable. In especially
wide sections of the river it is still probable that dol-
phins were missed. This can be addressed in future
surveys if channel width at sighting locations is col-
lected and included as a covariate in models.
Availability and perception bias
Indus dolphin groups of all sizes surfaced fre-
quently and were available to be detected by
observers; the failure of observers to detect or recog-
nise surfacings (perception bias) was therefore pri-
marily responsible for missed groups. Dawson et al.
(2008) suggested that perception bias tends to be
largest for species (such as Platanista gangetica) that
occur as single animals or in small groups and do not
show much of the body when surfacing. Perception
bias is often highest for inexperienced observers
(Barlow 1988), and it is likely that observer inexperi-
ence contributed to the higher proportion of missed
groups in the 2C-T subpopulation, which was the
first to be surveyed. This reinforces the importance of
training and observer experience in future surveys.
Bias in group size estimation
Estimation of group size for this species is challeng-
ing because groups are dispersed and individuals do
not surface in synchrony. Increased variability and
bias of group size estimates with increasing group
size has been documented in many other surveys
(Carretta et al. 1998, Gerrodette et al. 2002, Rugh et
al. 2008). One hypothesis to explain the decrease in
precision of Indus dolphin group size estimates as
group size increases is that the unsynchronised sur-
facing behaviour and lack of group cohesion means a
longer observation time is required to estimate the
size of large groups; during a downstream survey,
however, sufficient time is not always available.
Biased group size estimates are potentially a smaller
problem for Indus dolphins than for marine dolphins
that form schools of several hundreds and where
sighting conditions are often poor (Barlow 1988);
however, future surveys of this species will need to
consider methods to improve or calibrate group size
estimates, especially in high density areas.
Abundance and encounter rate
At least 10 Indus dolphin subpopulations have been
extirpated in the last century (Reeves et al. 1991). The
farthest upstream (1J-C) and downstream (5S-K) In-
dus subpopulations and the one in the Beas River are
each estimated at 10 or fewer individuals and are un-
likely to persist in the long-term, leaving only 3 that
are potentially viable. Conserving the 2C-T subpopu-
lation is of high priority, because it is the smallest of
211
Endang Species Res 17: 201–215, 2012
these 3 and its loss would mean Indus dolphins re-
main in only approximately 550 km of river, dramati-
cally increasing the vulnerability of the subspecies.
Encounter rate in the 4G-S subpopulation was 6.30
dolphins linear km−1, peaking at 10.35 dolphins lin-
ear km−1 in an 80 km section. These are the highest
encounter rates reported for any river cetacean. High
densities of Amazon River dolphins have been
recorded in specific favourable habitats (4.2 linear
km−1) (Martin & da Silva 2004), but the encounter
rates reported here are more than double those in the
Amazon and occur over a much wider area. Other
freshwater cetaceans normally occur at less than 2.00
dolphins linear km−1 (Smith et al. 2001, Kreb 2005,
Choudhary et al. 2006, Beasley 2007, Smith et al.
2007, Zhao et al. 2008, Wakid 2009). Given the
degree of disturbance to the natural flow and sedi-
ment transport regime of the Indus River system and
the fact that this subpopulation receives pollution
from upstream, it is hard to understand how the local
environment can support such unusually high densi-
ties of dolphins. However, the present survey was
conducted when animals were concentrated by dry-
season flow levels. The high observed density is pre-
sumably ephemeral because for much of the year
river discharge is higher, and density and competi-
tion for resources is reduced. At present, little quan-
titative data is available as a basis for comparing
habitat quality, prey availability, and dolphin mortal-
ity rates in the 4G-S subpopulation with other parts
of the Indus River to show why this area is so impor-
tant for Indus dolphins.
Trends in abundance of the Guddu to Sukkur
subpopulation
The direct counts generated by the present survey
and those from 2001 (Braulik, 2006) used identical
field methods and recorded very similar counts in
every subpopulation except for between Guddu and
Sukkur (Table 4) where the 2006 count was 64.5%
greater than 2001. Direct count surveys were also
conducted by Sindh Wildlife Department (SWD) dur-
ing the same time period. They reported 500 dol-
phins between Guddu and Sukkur barrages in 2001,
and 807 in 2006 (SWD unpubl. data); an increase of
61.4% over the same 5 yr period. The absolute counts
recorded by the 2 groups were different due to differ-
ent methods, but both recorded a similar increase.
Since 1974, at least 22 direct counts of the Guddu-
Sukkur dolphin subpopulation have been conducted
by SWD (Reeves & Chaudhry 1998, Bhaagat 1999,
Braulik 2006). Survey methods have not been com-
prehensively documented, and the reported dolphin
counts do not include measures of precision and pre-
sumably also underestimate the actual population
size, as no corrections were made for animals that
were missed. However, they included standard ele-
ments, with visual observers on oar-powered vessels
travelling downstream during the dry season, and
there was some consistency in observers over years.
If the methods, and hence the proportion of dolphins
missed, remained relatively stable over time, the sur-
veys can provide an indication of trends in abun-
dance in that subpopulation. The counts increase
from 138 in 1974 (Pilleri & Zbinden 1973−74) to 902 in
2008 (SWD unpubl. data) (Fig. 6) and show an expo-
nential, statistically significant, rate of increase (lin-
ear regression: F= 135, p < 0.001) equivalent to
approximately 5.65% per yr. The lack of confidence
intervals on all of these counts means that firm con-
clusions about population growth rates cannot be
made; however, all surveys indicate that there has
been some increase in the Guddu to Sukkur subpop-
ulation since the 1970s. A power analysis showed
that count CVs as large as 54% would have allowed
the observed trend to be detected at a 5% confidence
level (Gerrodette 1987, Gerrodette & Brandon 2000).
Until the 1970s, dolphins were hunted in Sindh
(Pilleri & Zbinden 1973−74). Hunting was banned by
the Sindh Wildlife Act (1972) and, in 1974, the Indus
Dolphin Reserve was established between Guddu
and Sukkur barrages. The probable increase in dol-
212
Subpopulation 2001 2006
(Braulik 2006) (present study)
1J-C 2 1
2C-T 84 82
3T-GG 45 44
4G-S 775a1275
5S-K 18 4b
aIn 2001, 602 dolphins were counted, and after extrapola-
tion of a conservative mean encounter rate (3.6 km−1) to
an unsurveyed 33.3 km segment, 725 were estimated
(Braulik 2006). As the unsurveyed segment was in a
very high density area, application of the encounter rate
from adjacent channels (5.0 km−1) is more realistic and
we therefore applied this to generate a revised estimate
of 775 animals in 2001 in this subpopulation
bThe whole 5S-K subpopulation was not surveyed in
2006, so figures cannot be directly compared between
years
Table 4. Comparison of direct counts of Indus dolphin
Platanista gangetica minor subpopulations recorded in 2001
and 2006 using identical survey methods
Braulik et al.: Abundance of Indus River dolphins
phin abundance between Guddu and Sukkur likely
signifies population recovery following the cessation
of dolphin hunting. It is unlikely that the increase is
due to improvements in habitat or prey availability,
as new dams and barrages continue to be con-
structed, the Indus flood cycle and sediment trans-
port processes have been greatly disrupted, dry sea-
son discharge has declined, and levels of pollution
have increased dramatically as Pakistan becomes
industrialized (World Bank 2005). The river corridor
in upper Sindh is a tribal area subject to banditry and
lawlessness, resulting in low levels of human activity
compared to other parts of the river that may con-
tribute to low dolphin mortality rates. Another factor
that could contribute to the increase is the role of
immigration from other subpopulations. It is possible
that dolphins traverse irrigation barrages and move
between subpopulations during the few weeks of the
year that gates are fully open (Reeves et al. 1991,
Braulik 2006). Many factors are likely to influence
whether animals move through a particular barrage,
including its design, river discharge, hydrology,
adjacent dolphin density, and most importantly, how
the barrage is operated and how frequently the gates
are opened. It is likely that dolphins never traverse
some barrages but frequently traverse others. How-
ever, during January 2009, in the only few weeks of
the year when the gates of Sukkur barrage are open,
a radio-tagged dolphin moved through the barrage 3
times, finally becoming trapped downstream (WWF-
Pakistan unpubl. data). If the predominant move-
ment of migrants is downstream, this ‘downstream
migratory attrition’ would result in the decline of all
upstream subpopulations and the increase of those
downstream (Reeves 1991, Reeves et al. 1991). As the
increase in counts from 2001 to 2006 between Guddu
and Sukkur exceeded the total number of animals
recorded in all other subpopulations, immigration
cannot be solely responsible. The large proportion of
calves and juveniles observed (11%) suggests that
this subpopulation is reproducing rapidly and if
immigration does occur, it is likely supplementing
the increase, rather than being solely responsible for
it. It is essential that there is continued monitoring of
the Indus subpopulations, using standard survey
methods to provide more robust data for determining
trends in abundance, and radio or satellite tracking
of dolphins in different locations and seasons will
help to shed light on dolphin movement between
subpopulations.
Conclusions
In medium and low density areas the survey
method was effective, but at very high densities,
greater than 6 dolphins linear km−1, it was not possi-
ble to apply. Such densities, however, are excep-
tional and do not occur elsewhere in Asia. Whether
using tandem vessels or independent observers on a
single vessel, the direct count capture-recapture sur-
vey method following a thalweg transect shows good
potential for abundance estimation of dolphins in
confined areas or shallow rivers, such as the Indus,
Ganges, Brahmaputra, and Ayeyarwady, where dol-
phin densities are generally low and traditional
methods for estimating abundance cannot be easily
applied. One distinct advantage of the method is that
despite the more complex analysis, the existing rela-
tively low-cost field survey methods do not need to
be greatly modified. However, in each new survey it
is essential that there be a careful evaluation of the
reliability of matches.
The future of the South Asian river dolphins is inti-
mately tied to the region’s water security. South Asia
has approximately 25% of the world’s human popu-
lation but only 4.5% of its renewable water resources
(Babel & Wahid 2008). The vast majority of water
needs are met by the region’s rivers, per capita water
availability is plummeting, and in Pakistan demands
213
Fig. 6. Natural logarithm of Indus River dolphin Platanista
gangetica minor direct counts recorded between Guddu and
Sukkur barrages between 1974 and 2008. The natural loga-
rithm was applied to the counts to transform the exponential
increase into a linear increase that could be examined using
linear regression. SWD: Counts conducted by Sindh Wildlife
Department; WWF: World Wildlife Fund (WWF)-Pakistan
and Ministry of Environment’s Pakistan Wetlands Pro-
gramme; Other: counts conducted by authors other than
SWD or WWF, for details see Braulik (2006) and Bhaagat
(1999). The WWF point labelled 1 represents data from
Braulik (2006), and the WWF point labeled 2 represents
the direct count recorded in the present study
Endang Species Res 17: 201–215, 2012
214
for fresh water are predicted to outstrip availability
before 2025 (Siddiqi & Tahir-Kheli 2004). The tempo-
rary increase in river discharges predicted due to cli-
mate change will be overwhelmed by increased
water demand (Palmer et al. 2008). There are numer-
ous plans to increase river withdrawals, to transfer
water between rivers or river basins, and increase
water storage capability by constructing high dams
(Ghosh et al. 2003, Bosshard & Lawrence 2006, de
Fraiture & Wichelns 2010, Dutta 2010), which will
further deplete river flows and disrupt the natural
flow regime. Losses of freshwater biodiversity are
inevitable and the prospects for the South Asian river
dolphins uncertain.
Acknowledgements. The Pakistan Provincial Wildlife De -
partments, Police Forces, Irrigation Departments, Ministry
of Environment, and Water and Power Development
Authority (WAPDA) facilitated the expedition. Logistical
support was undertaken by R. Ahmed, M. Azam, M.
Abduhu and A. Rana of the Adventure Foundation Pakistan.
Members of the scientific teams included: K. Javed, M.
Ahmad, M. Chaudhry, S. Khan, S. Hussain, S. Ahmad, M.
Hamid, I. Khaskheli, A. Bullo and Z. Ali. Invaluable support
was provided by A. Habib, M. Malik, I. Tajwer, A. Khan, G.
Akbar and M. Arshad. The manuscript was improved by
input from P. Hammond, R. Williams and R. Reeves. Valu-
able statistical input was provided by S. Hedley. Funding
came from WDCS, WWF-Pakistan, and the Pakistan Wet-
lands Programme which is funded by UNDP-GEF and the
Netherlands Embassy.
LITERATURE CITED
Babel MS, Wahid SW (2008) Freshwater under threat: South
Asia. United Nations Environment Programme, Nairobi
Barlow J (1988) Harbor porpoise, Phocoena phocoena,
abundance estimation for California, Oregon, and Wash-
ington: I. Ship surveys. Fish Bull 86: 417−432
Barlow J, Oliver CW, Jackson TD, Taylor BL (1988) Harbor
porpoise, Phocoena phocoena, abundance estimation for
California, Oregon, and Washington: II. Aerial surveys.
Fish Bull 86: 433−444
Beasley I (2007) Conservation of the Irrawaddy dolphin,
Orcaella brevirostris (Owen in Gray, 1866) in the
Mekong River: biological and social considerations influ-
encing management. PhD thesis, James Cook University,
Townsville
Behera SK, Nawab A, Rajkumar B (2008) Preliminary inves-
tigations confirming the occurrence of Indus River dol-
phin Platanista gangetica minor in River Beas, Punjab,
India. J Bombay Nat Hist Soc 105: 90−126
Bhaagat HB (1999) Introduction, distribution, conservation
and behavioral ecology of Indus blind dolphin (Platanista
indi) in River Indus (dolphin reserve), Sindh-Pakistan.
Tiger Paper 26: 11−16
Borchers DL, Buckland ST, Zucchini W (2002) Estimating
animal abundance: closed populations. Springer-Verlag,
London
Bosshard P, Lawrence S (2006) Pakistan’s rot has World Bank
roots. Far Eastern Economic Review May 2006: 39−43
Braulik GT (2006) Status assessment of the Indus River dol-
phin, Platanista gangetica minor, March−April 2001. Biol
Conserv 129: 579−590
Braulik GT, Smith BD, Chaudhry AA (2004) Platanista
gangetica ssp. minor. In: IUCN 2010 IUCN Red List of
Threatened Species, Ver 20102. www.iucnredlist.org/
apps/redlist/details/41757/0 (accessed 04 August 2010)
Buckland ST, Anderson DR, Burnham KP, Laake JL,
Borchers DL, Thomas L (2001) Introduction to distance
sampling: estimating abundance of biological popula-
tions. Oxford University Press, Oxford
Burnham KP, Anderson DR (2002) Model selection and
multi- model inference: a practical information-theoretic
approach, 2nd edn. Springer-Verlag, New York, NY
Carretta JV, Forney KA, Laake JL (1998) Abundance of
southern California coastal bottlenose dolphins esti-
mated from tandem aerial surveys. Mar Mamm Sci 14:
655−675
Choudhary SK, Smith BD, Dey S, Dey S, Prakash S (2006)
Conservation and biomonitoring in the Vikramshila
Gangetic dolphin sanctuary, Bihar, India. Oryx 40:
189−197
Crete M, Vandal D, Rivest LP, Potvin F (1991) Double counts
in aerial surveys to estimate polar bear numbers during
the ice-free period. Arctic 44: 275−278
Dawson S, Wade P, Slooten E, Barlow J (2008) Design and
field methods for sighting surveys of cetaceans in coastal
and riverine habitats. Mammal Rev 38: 19−49
de Fraiture C, Wichelns D (2010) Satisfying future water
demands for agriculture. Agric Water Manag 97: 502−511
Dudgeon D, Arthington AH, Gessner MO, Kawabata Z-I
and others (2006) Freshwater biodiversity: importance,
threats, status and conservation challenges. Biol Rev 81:
163−182
Dutta AP (2010) Reservoir of dams. Down to earth. May 1-15:
32-39 www.downtoearth.org/in/content/reservoir-dams
Gerrodette T (1987) A power analysis for detecting trends.
Ecology 68: 1364−1372
Gerrodette T, Brandon J (2000) Trends. Ver. 3.0 Windows.
Southwest Fisheries Science Center, La Jolla, CA
Gerrodette T, Perryman WL, Barlow J (2002) Calibrating
group size estimates of dolphins in the eastern tropical
pacific ocean. Southwest Fisheries Science Center,
National Marine Fisheries Service, NOAA, La Jolla, CA
Gerrodette T, Taylor BL, Swift R, Rankin S, Jaramillo-Legor-
reta AM, Rojas-Bracho L (2011) A combined visual and
acoustic estimate of 2008 abundance, and change in
abundance since 1997, for the vaquita, Phocoena sinus.
Mar Mamm Sci 27:
E79−100
Ghosh A, Sikdar PK, Ashoke KD (eds) (2003) Interlinking of
Indian Rivers. ACB Publications, Kolkata
Graham A, Bell R (1989) Investigating observer bias in aerial
survey by simultaneous double-counts. J Wildl Manag
53: 1009−1016
Hammond PS (2009) Mark-recapture. In: Perrin WF, Würsig
B, Thewissen JGM (eds) Encyclopedia of marine
mammals, 2nd edn. Academic Press, San Diego, CA,
p 705−709
Hammond PS, Berggren P, Benke H, Borchers DL and others
(2002) Abundance of harbour porpoise and other
cetaceans in the North Sea and adjacent waters. J Appl
Ecol 39: 361−376
Hiby L, Krishna MB (2001) Line transect sampling from a
curving path. Biometrics 57: 727−731
➤
➤
➤
➤
➤
➤➤
➤
➤
➤
➤
➤➤
Braulik et al.: Abundance of Indus River dolphins
Hiby AR, Lovell P (1998) Using aircraft in tandem formation
to estimate abundance of harbour porpoise. Biometrics
54: 1280−1289
Horvitz DG, Thompson DJ (1952) A generalization of sam-
pling without replacement from a finite universe. J Am
Stat Assoc 47: 663−685
Huggins RM (1989) On the statistical-analysis of capture
experiments. Biometrika 76: 133−140
IUCN (International Union for Conservation of Nature)
(2010) IUCN Red List of Threatened Species, ver. 2010.2.
www.iucnredlist.org (accessed 3 August 2010)
IWC (International Whaling Commission) (2001) Annex K
report of the standing sub-committee on small cetaceans.
J Cetacean Res Manag 3: S263−290
Kreb D (2005) Abundance of freshwater Irrawaddy dolphins
in the Mahakam River in East Kalimantan, Indonesia,
based on mark-recapture analysis of photo-identified
individuals. J Cetacean Res Manag 6: 269−277
Marsh H, Sinclair DF (1989) Correcting for visibility bias in
strip transect aerial surveys of aquatic fauna. J Wildl
Manag 53: 1017−1024
Martin AR, da Silva VMF (2004) River dolphins and flooded
forest: seasonal habitat use and sexual segregation of
botos Inia geoffrensis in an extreme cetacean environ-
ment. J Zool 263: 295−305
Palmer MA, Liermann CAR, Nilsson C, Florke M, Alcamo J,
Lake PS, Bond N (2008) Climate change and the world’s
river basins: anticipating management options. Front
Ecol Environ 6: 81−89
Pilleri G, Zbinden K (1973−74) Size and ecology of the dol-
phin population (Platanista indi) between Sukkur and
Guddu Barrages, Indus River. Investigations on Cetacea
5: 59−70
Reeves RR (1991) Conservation of the bhulan (blind river
dolphin) in the Punjab. Natura, WWF-Pakistan, Lahore
p 3−22
Reeves RR, Chaudhry AA (1998) Status of the Indus River
dolphin Platanista minor. Oryx 32: 35−44
Reeves RR, Chaudhry AA, Khalid U (1991) Competing for
water on the Indus Plain: Is there a future for Pakistan’s
river dolphins? Environ Conserv 18: 341−349
Rugh DJ, Muto MM, Hobbs RC, Lerczack JA (2008) An
assessment of shore-based counts of gray whales. Mar
Mamm Sci 24: 864−880
Samuel MD, Pollock KH (1981) Correction of visibility bias
in aerial surveys where animals occur in groups. J Wildl
Manag 45: 993−997
Siddiqi TA, Tahir-Kheli S (eds) (2004) Water needs in
South Asia: closing the demand-supply gap. Global
Environment and Energy in the 21st century, Hon-
olulu, HI
Smith BD, Reeves RR (2000) Survey methods for population
assessment of Asian river dolphins. In: Reeves RR, Smith
BD, Kasuya T (eds) Biology and conservation of freshwa-
ter cetaceans in Asia. IUCN, Gland, p 97−115
Smith BD, Ahmed B, Ali ME, Braulik GT (2001) Status of the
Ganges river dolphin or shushuk Platanista gangetica in
Kaptai Lake and the southern rivers of Bangladesh. Oryx
35: 61−72
Smith BD, Braulik G, Sinha RK (2004) Platanista gangetica
ssp. gangetica IUCN 2010 IUCN Red List of Threatened
Species, ver. 20102 www.iucnredlist.org/apps/redlist/
details/41756/0 (accessed 4 August 2010)
Smith BD, Braulik G, Strindberg S, Ahmed B, Mansur R
(2006) Abundance of Irrawaddy dolphins (Orcaella bre-
virostris) and Ganges River dolphin (Platanista gangetica
gangetica) estimated using concurrent counts made by
independent teams in waterways of the Sundarbans
mangrove forest in Bangladesh. Mar Mamm Sci 22:
527–547
Smith BD, Shore RG, Lopez A (eds) (2007) Status and
conservation of freshwater populations of Irrawaddy
dolphins. Wildlife Conservation Society Working Paper
No. 31
Stanley TR, Burnham KP (1998) Information-theoretic model
selection and model averaging for closed-population
capture-recapture studies. Biom J 40: 475−494
Strayer DL, Dudgeon D (2010) Freshwater biodiversity con-
servation: recent progress and future challenges. J N Am
Benthol Soc 29: 344−358
Turvey ST, Pitman RL, Taylor BL, Barlow J and others (2007)
First human-caused extinction of a cetacean species?
Biol Lett 3: 537−540
Vidal O, Barlow J, Hurtado LA, Torre J, Cendón P, Ojeda Z
(1997) Distribution and abundance of the Amazon River
dolphin (Inia geoffrensis) and the tucuxi (Sotalia fluvi-
atilis) in the upper Amazon River. Mar Mamm Sci 13:
427−445
Wakid A (2009) Status and distribution of the endangered
Gangetic dolphin (Platanista gangetica gangetica) in the
Brahmaputra River within India in 2005. Curr Sci 97:
1143−1151
Williams BK, Nichols JD, Conroy MJ (2002) Analysis and
management of animal populations. Academic Press,
London
World Bank (2005) Pakistan country water resources assis-
tance strategy. Agriculture and Rural Development Unit,
South Asia Region, World Bank
Zhao X, Barlow J, Taylor BL, Pitman RL and others (2008)
Abundance and conservation status of the Yangtze fin-
less porpoise in the Yangtze River, China. Biol Conserv
141: 3006−3018
215
Editorial responsibility: Brendan Godley,
University of Exeter, Cornwall Campus, UK
Submitted: August 25, 2010; Accepted: February 11, 2012
Proofs received from author(s):
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤➤
➤
➤
