(211)

J. Env. Bio-Sci., 2019: Vol. 33 (2):211-219 ISSN 0973-6913 (Print), ISSN 0976-3384 (On Line)

GEOSPATIAL MODELLING OF POTENTIAL HABITAT OF TIGER (PANTHERA

TIGRIS TIGRIS) IN CORBETT - KOSI RIVER CORRIDOR OF UTTRAKHAND, INDIA
H. S. Gehlot* and Piyush Joshi
Department of Zoology, Faculty of Science, J N Vyas University, Jodhpur, Rajasthan
*Naturalist, Corbett Tiger Reserve, Ramnagar, Nainital, Uttrakhand
Email*:gehloths@gmail.com

Received: 08-05-2019 Revised: 26-11-2019 Accepted: 10-12-2019

The present study attempts to assess the potential habitat of the tiger (Panthera tigris tigris) in Corbett - Kosi corridor of Uttrakhand
State of India using geospatial modelling from April 2008 to September 2010. MaxEnt modelling was used to generate the habitat
suitability maps and the jackknife test of this model was applied to develop the potential habitat map for chital, wild boar, sambar,
and tiger. As per this model, the AUC for training data were found to be 0.809 for chital, 0.0818 for wild boar, 0.746 for sambar and
0.800 for tiger. According to response curves, tiger preferred low slope but elevation landscape. Tiger favoured frequenting in sal
mix and teak forest type where the presence of sambarwas highest (48.7 %). Satellite remote sensing technique was used for
mapping different forest types and density classes in this region.
Key words: Geospatial Modelling, Panthera tigris tigris, Maxent modelling, Jack knife test.

India's national animal 'Tiger' despite being accorded the
highest protection under Schedule I of the Wildlife (Protection)
Act of India, 1972, has been forced to lose ground in recent
decades (John singh et al., 2010; Jhala et al., 2011).In Terai
Arc Landscape, conservationists are working to restore,
reconnect, and manage wildlife corridors to link 12 important
wildlife reserves and national parks that harbour wild tigers
across the 49,000 km² area (Wikramanayake et. al., 2004;
Dinerstein et. al., 2007). The goal is to manage tigers as a
meta-population in which dispersal between core refuges can
help to maintain genetic, demographic, and ecological
integrity.The Corbett National Park has the highest density of
tigers (17.8 ±1.4/100 km²) in the world and serves as a source
from where tigers are likely to disperse both westward as well
as eastward to maintain the populations in the Rajaji National
Park and Pilibhit Forest Division. However, the population
outside the Tiger Reserve (TR) is also of great significance
with the Ramnagar FD having a density of approximately 15
tigers per 100 km² with evidence of breeding individuals (Jhala
et. al., 2011). The tiger population of Corbett and Ramnagar
currently forms a single unit with important connectivity
provided by the Kosi River which flows between Corbett Tiger
Reserve (CTR) and Ramnagar Forest Division (RMR FD) and
certain stretches of forests along the river. Tiger crosses the
river for prey and maintains their territory. Habitat is a sum
total of the environmental condition of a specific place occupied
by wildlife species or a population of such species. All species
have specific habitat requirements, which can be described
by habitat factors. These factors were connected to the critical
characteristics of the habitat, such as vegetation, soil and the
spatial structure of landscape elements. Urbanization,
industrialization, infrastructure development projects,
agriculture, grazing, deforestation, wildlife trade, and poaching
continue to create tremendous stress on pristine natural habitat
and wildlife. Remote sensing and Geographic Information
System (RS and GIS) can be used as a tool for getting
information about the habitat preference of the wildlife species.
RS and GIS also help in monitoring areas of land for their
suitability to endangered species, through the integration of
various habitat variables of both spatial and non-spatial nature
(Davis and Goetz, 1990). The outputs of such models are
usually simple, easily understandable and can be used for the
assessment of environmental impacts or prioritization of
conservation efforts in a timely and cost-effective manner
(Kushwaha et al., 2004 and Zarri et. al., 2008). As a result, the
information acquired from RS data when supported by field
evidence of the presence of animal and biotic disturbances
has been used for wildlife habitat suitability analysis (Roy,
1993; Roy and Ravan, 1994). Consequently, a habitat suitability
model is developed primarily based on the distribution of
animals compared to the availability of habitat parameters (Buk,
2004).The present study was carried out using the RS and
GIS to predict and model the habitat suitability of tiger (Panthera
tigris tigris) and its prey sambar (Rusa unicolor), chital (Axis
axis) and wildboar (Sus scrofa) species in the Kosi - Corbett
river Corridor for better conservation.

NAAS Rating (2019)-4.43

 GEHLOT AND JOSHI
MATERIALS AND METHODS
Study area:The Kosi corridor includes part of Almora Forest
Division, Sarpduli Range of Corbett Tiger Reserve (CTR) and
Kosi Range of the Ramnagar FD. Broadly, this corridor
connects CTR with the forests of Ramnagar FD and
administratively comes under Nainital and Almora districts of
Uttarakhand (Table 1,Plate 1)
Satellite data: The digital image of Landsat 8 Operational
Land Imager (OLI) and Thermal Infrared Sensor (TIRS) were
used. The imaging consisted of nine spectral bands with a
spatial resolution of 30 meters for Bands 1 to 7 and 9. New
band 1 (ultra-blue) is useful for coastal and aerosol studies.
New band 9 is useful for cirrus cloud detection. The resolution
for Band 8 (panchromatic) is 15 meters. Thermal bands 10
and 11 are useful in providing more accurate surface
temperatures and are collected at 100 meters. Approximate
scene size is 170 km north-south by 183 km east-west.
SRTM (DEM):The Shuttle Radar Topographic Mission (SRTM)
Digital Terrain Model (DTM) with 30 m resolution, downloaded
from the www.srtm.usgs.govt. website, was used to generate
the slope, aspect and elevation data layers.
Software and equipment: ERDAS Imagine 14 and ArcGIS
10.3 were used for the processing of satellite data and GIS
work. Equipment like binoculars (8 x 40), camera (Canon 700
D with lens 55-300 mm), and Global Positioning System
(GARMIN e_trex vista) were used for the field data collection.
GPS was used for recording locations of tiger and its prey
species (sambar, chital and wild boar). MaxEnt 3.3 was used
for assessment of potential habitat of tiger and its prey species
in the study area. Wildlife fauna was surveyed using random
sampling techniques on the trail of the corridor while recording
the locations with the help of GPS. Maxent modelling was
used to generate the habitat suitability maps and calculate
the predictable habitat of the tiger encountered taking into
account the closeness to water availability, habitat type, slope,
aspect and prey species, etc.
Study Area Extraction: The boundary of the study area was
delineated taking Kosi river as a center and make an area of
interest (AOI) for image analysis.
Field Work and Ground Data Collection: An extensive field
survey was undertaken and the ground point collection such
(212)
as direct sighting, pugmark, claw mark on bark of tree, kill of
prey and scat were undertaken to obtaining information about
the accessibility of the area, the patterns, and distribution of
vegetation type, recording of location of tiger presence and its
prey sambar, chital, wildboar sighting.
Image Interpretation: The image feathers on the satellite
data were interpreted for vegetation type using various image
elements viz. tone, texture, patterns, shape, size, shadow,
location and association. Various forest type categories were
classified, correlated and interpreted using on-screen visual
interpretation. Maps were prepared using Universal Transverse
Mercator (UTM) projection and WGS-84.
ImageVerification: After interpretation of satellite image,
image verification was carried out through personal knowledge
of the field. The vegetation type of the study area was marked
on the satellite image during ground truth collection and geo -
coordinates were taken by using GPS.
Database Creation: Habitat suitability analysis requires the
generation of the accurate database on various life support
system as well as potential distribution factors in the habitat.
Five maps were required for habitat analysis using GIS. Digital
database of the study area, vegetation type, slope, elevation,
and aspect were prepared in a GIS environment. All the input
map layers were co-registered with sub-pixel accuracy. A
specific GIS model will be followed for the assessment of habitat
suitability of tiger in the study area (Fig. 1).
Analysis: The field database was geostatistically analyzed to
understand the habitat used by the animal and its prey species.
This was carried out using the Maximum Entropy Model
(MaxEnt).
Maximum Entropy Method (MaxEnt): MaxEnt is a recently
introduced modelling technique, achieving high predictive
accuracy and enjoying several additional attractive properties.
The performance of MaxEnt is influenced by a moderate number
of parameter.The data available for this problem typically
consists of a list of geo-referenced occurrence localities, i.e.,
a set of geographic coordinates where the species has been
observed. In addition, there are data on a number of
environmental variables, such as elevation, slope, aspect,etc.
They have been measured or estimated across a geographic
region of interest. The maximum-entropy principle is explained
(213) GEOSPATIAL MODELLING OF POTENTIAL HABITAT OF TIGER
here, the unknown probability distribution, which we denote
π , is over a finite set X (which we will later interpret as the set
of pixels in the study area).We refer to the individual elements
of X as points. The distribution  assigns a
non-negative probability π (x) to each point x, and these
probabilities sum to 1. Our approximation of π is also
probability distribution, and we denote it ˆ π . The entropy of π
is defined aswhere is the natural logarithm,the entropy is non-
negative and is at most the natural log of the number of elements
in X. (S.J. Phillips et. al., 2004) and The MaxEnt is a recently
introduced modelling technique, achieving high predictive
accuracy and enjoying several additional attractive properties.
The performance of Max Ent is influenced by a moderate
number of parameters (Phillips, 2006).
RESULT AND DISCUSSION
First, the false color composite map of the study area was
prepared (Plate 2). It was also used to identify environmental
variable associated with the tiger occurrence in the study area
using MaxEnt.The Area under the Receiving Operator Curve
(AUC) for training data for the study was analyzed for tiger.
The predictors that affected the tiger occurrence in the model
are the presence of prey species especially sambar(48.7) and
sal mix forest. The occurrence of the tiger was more with low
aspect, elevation, and slope. According to jack knife training
gain, the contribution of aspect and elevation were not preferred
by a tiger while other variables like forest type and riverine
area were favorite. The study area was divided into 9 different
habitats (Table 2) and the forest type map was developed (Plate
3A). The SRTM was used to generate the map of aspect (Plate
3B), elevation (Plate 3C) and slope (Plate 3D). Various
environmental variables and animal presence points were used
as points into the MaxEnt model to generate the habitat
suitability maps for tiger and its prey species. Receiver
Operating Characteristic (ROC) curve for wild species in this
corridor was prepared.The occurrence of the tiger was more
with low aspect, elevation, and slope.The AUC for training
data for the study were 0.809 for chital, 0.0818 for wild boar,
0.746 for sambar and 0.800 for tiger (Fig. 2). As per the
response curves, the chital, wild boar and sambar preferred
elevation areas in lower slopes but prey species did not prefer
high dense forest (low NDVI) thus the habitat suitability map
produced in the present study is very similar to the map
produced by Ghosh (2013) for habitat suitability of tigers in
Indo-Bhutan Manas Tiger Conservation Landscape (IBMTCL).
As per the result of this study, chital, wild boar, and tiger
favoured river bed forest type. Kushwa et.al.,(2008) and Ghosh
(2013) reported that tiger prefers vicinity of water bodies with
an increase in distance from settlements the probability of
occurrence of tiger also increase. The tiger preferred low slope,
elevation area, and aspect according to response curves (Fig.
3). Kushwaha et al.,(2008) noticed that the tiger prefers
relatively open area and not very dense forests if abundant
prey is available. While the sambar preferred (48.7 %) sal mix
forest type which is the most preferred food of tiger thus tiger
favoured sal mix and teak forest type (Table 3). The map
generated has shown that the Habitat Suitability of tiger in the
study area is related mainly to forest type, water availability,
slope, aspect and topography of the area. Water availability is
a basic requirement for the survival of the tiger and other wild
animals in the study area. In the corridor, there is a Kosi river
which is an only water source. The validity of the model
developed was tested against the GPS locations of tiger
presence through direct or indirect sighting in this corridor;
the overall accuracy of the model (AUC) is 0.800 for tiger. The
tiger prefers the habitat where as sambar number is good as
per jackknife of AUC for tiger (Fig. 4). The main objectives of
Habitat Modelling are predicting the distribution of wildlife
species in a geographical area with high species diversity
(Butterfield, 1994), locating of species of concern (Sperduto,
1996), predicting the area of suitable habitat that may not be
currently used by species (Lawton, 1991). Most important
factors determining the habitat use by a carnivore are the
availability of suitable prey and hunting success (Blame et al.,
2007) and Landscape attributes and prey abundance appear
to be the proximate factors driving resource selection in
carnivores (Stephens and Krebs, 1986).Our study suggests
that tigers show a very strong pattern of habitat selection in
this unique landscape; they prefer sal mix forest of this corridor
with the presence of their main prey species, but avoid areas
with dense human habitations and anthropogenic pressure
(Plate 4). Similar to observations have been reported by Smith
et al., (1998), Johnsingh et al. (2004). In this study, satellite
RS has been used for mapping different forest types and density
classes in this region same technique have been used as
 GEHLOT AND JOSHI (214)

Table 1: Details of Kosi - Corbett Corridor of Uttrakhand

Connectivity Village / Frequency of use
Name of Anthropogenic
between Forest type khatta fall in by key wildlife Major threat
Linkage pressure
forest division the corridor species
Mohan - Corbett - Tropical Kumeria and Ramnagar- Elephant herd 2-10, IMPCL factory
Kumeria Ramnagar dry Khunkhet Ranikhet Road Tiger 2-3, Mohan, Grazing
Forest deciduous NH 121 many Ungulate herd pressure by cattle
Division
Sunderkh Corbett - Sal forest Chukam, Ramnagar- Elephant herd 2 -10, Encroachment,
al - Mohan Ramnagar and teak Mohan, Ranikhet Road, Tiger 2-3 Lopping, Grazing
Forest plantation Amarpur, Garjiya temple Many prey species pressure by cattle,
Division Devichaur&Su Fuel wood collection
nderkhal
Bijrani- Corbett - Teak Forest deport, Elephant herd 2-10 pressure by cattle
Dhikulli Ramnagar plantation Aamdanda, Mazar at Ringora Tiger 2-3, Uncontrolled
Forest Ringora and Heavy Traffic many Ungulate herd development in the
Division on NH 121 corridor area &
Increasing number
of resorts in the
vicinity

Table 2: Forest type of study area Table 3: Estimates of relative contributions of the
environmental variables for Habitat Suitability of tiger
S No Forest Type Area (in sq km)
1 Sal forest 74.00 Percentage Permutation
Variable
2 Sal mix forest 125.06 contribution importance
3 Teak forest 11.34 Sambar_avg 48.7 4.2
4 Riverine grasses 04.64 Aspect 23.4 19.5
5 Agriculture 2.168 Forest Type 19.3 19.1
6 River bed 12.87 Elevation 4.3 44.5
7 River 2.05 Forest Density 2 3.4
8 Settlement 2.07 Chital_avg 1.7 7
9 Scrub land 5.71
Slope 0.6 2.2
Total 239.94
Wild_boar_avg 0 0

digital techniques by several other earlier workers( Roy et al.,

(1986), Kushwaha and Madhavan Unni, (1986), Parihar et.al.,
1986), Singh,(1986). Every species were the product of their
habitats (Riclefs, 1973). But During this study, we found that
some dominant component of the habitat of physical variables
can be used to quantify the habitat.

ACKNOWLEDGMENTS

The authors are thankful Dr. S Nandy and Dr. SarnamSingh,

Department of FED and Dr. S.P.S. Kushwaha, Dean, Indian
Institute of Remote Sensing, Dehradun and Dr. Akhilesh
Scientist of Gwalior for encouragement during the preparation
of this paper during my NNRMS training period. I am deeply
indebted to Dr. S P Goyal, Senior Scientist of Wildlife Institute
of India, Dehradun, and Dr. H. C. Bohra, Sr. Scientist, CAZRI,
Jodhpur for his advice and guidance from time to time. I am
obliged to Dr. Parag M Dhakhete IFS, Mr.Lalit Mohan of
Uttrakhand Forest Department, Mr. Ashish Thakur, Rimpi,
(215) GEOSPATIAL MODELLING OF POTENTIAL HABITAT OF TIGER

GudduTiwari Nature Guide of Corbett Tiger Reserve for providing

essential facilities and technical support during fieldwork.

REFERENCES

1. Blame, G. A., L. T. B. Hunter and R. Slotow.: Feeding habitat selection

by hunting leopards Pantherapardus in a woodland savanna: prey
catchability versus abundance. Animal Behaviour 74: 580- 598
(2007)
2. Buk, K. G.: Diet Selection and Habitat Suitability for Black Rhino in
Augrabies Falls National Park, South Africa. MSc. Thesis, University
of Copenhagen, Copenhagen (2004)
3. Butterfield BRB, Csuti JMS: Modeling vertebrate distribution for Gap
analysis. In: R.I.Miller, e., editor, mapping the diversity of nature.
London: Chapman and Hall, 53-68 (1994)
4. Coulombe H N: Classification, inventory and analysis of Fish and
Wildlife habitats. The proceedings of the National Symposium -
Washington D.C., pp. 604 (1977)
5. Davis FW, Goetz S: Modeling vegetation pattern using digital terrain
data. Landscape Ecology, 4:69-80 (1990)
6. Dinersteinet al.: A statistical explanation of Maxent for ecologists.
Divers Distributers 17, 43-57(2011)
7. Ghosh, S: Habitat suitability analysis for tiger (Pantheratigris) in the
Indo - Bhutan Manas Tiger Conservation Landscape (Ph.D. Thesis)
Department of geography and earth sciences, Aberystwyth
University, U.K. (2013)
8. Guisan, A. and Zimmerman N. E.: Predicting habitat distribution models
in ecology. Ecol. Model. 135, 147-186 (2000)
9. Jaynes, E.T.: Information theory and statistical mechanics. Phys.
Rev. 106, 620-630(1957)
10. Jhalaet al.: Status of the Tigers, Co-predators, and Prey in India,
2010. National Tiger Conservation Authority, Govt. of India, New
Delhi, and Wildlife Institute of India, Dehradun TR 2011/003 pp. 302
(2011)
11. Johnsinghet al.: Conservation status of tiger and associated
species in the Terai Arc Landscape, India. _ RR-04/001, Wildlife
Inst. of India, Dehradun (2004)
12. Johnsinghet al.: Status and conservation of tigers in the Indian
subcontinent. Tigers of the world 2nd edition (2010)
13. Johnson L.B.: Analyzing spatial and temporal phenomena using
geographical information systems. Landscape Ecology, 4:31-43
(1990)
14. Kushwaha S P S and MadhavanUnni, N V: Application of Remote
Sensing techniques in forest cover monitoring and habitat evaluation.
A case study in Kaziranga National Park, Assam Proc. Seminar -
cum - Workshop on Wildli/e Habitat Evaluation using Remote Sensing
technique Dehradun, India 22-23 October, 1986 (Dehradun. India:
[IRS and WII), pp. 146-156 (1986)
GEHLOT AND JOSHI (216)
(217) GEOSPATIAL MODELLING OF POTENTIAL HABITAT OF TIGER
GEHLOT AND JOSHI (218)
(219) GEOSPATIAL MODELLING OF POTENTIAL HABITAT OF TIGER
15. KushwahaSPS, Khan A, Habib B, Quadri A, Singh A.: Evaluation of
sambar and muntjac habitats using geo-statistical modelling. Current
Science, 86 (10): 390-1400 (2004)
16. KushwahaSPS, Quadri A, Singh A and Khan A.: Habitat Preference
Assessment of the tiger and its prey ungulates in the Corbett Tiger
Reserve. Fauna of Corbett Tiger Reserve, Uttrakhand 33-35 (2008)
17. Lawton JH, Woodroffe,G.L.: Habitat and the distribution of water
volves: why are there gap in a species range. Journal of Animal
Ecology, 60:79-91 (1991)
18. Menon, S. and Bawa K. S.: Applications of geographic information
systems, remote sensing, and landscape ecology approach to
biodiversity conservation in the Western Ghats. Current Science
73 (2): 134-144 (1997)
19. Nandy et. al.: (2007). Monitoring the Chilla-Motichur wildlife corridor
using geospatial tools, Journal for Nature conservator 15 (2007)
237-244
20. Panth et al.: Habitat analysis for Sambarin Corbett National Park,
U.P. Photonirvachak24(1): 32-52 (1999)
21. Parihar et al.: Study of wildlife habitat using high resolution space
photographs. A case study in Kanha National Park, ISRO Special
Publication, ISRO-SP-17-86, Results from the joint Indo-Soviet Remote
Sensing Experiment. TERRA on board Salyut- 7 (1986)
22. Phillips, S.J., Miroslav, D., Schapire, R.E.: Maxent Software for
Species Distribution Modelling, http://cs.princeton.edu/?schapire/
Maxent (2004)
23. Phillips et all., Maximum entropy modeling of species geographic
distributions Ecol. Model. 190: 231259 (2006)
24. Phillips, S.J., Dudik, M.: Modelling of species distributions with
MaxEnt: new extensions and a comprehensive evaluation.
Ecography31, 161-175 (2008)
25. Riclefs R C: Ecology. Chiron Press, Neston, Massachussetts, pp.
861 (1973)
26. Roy, P. S.: Remote sensing for forest ecosystem analysis and
management In: Environmental Problems and Prospects in India, pp.
335-363 Oxford & IBH Publishing Co. Pvt Ltd, New Delhi (1993)
27. Roy, P. S. and Ravan, S. A.: Habitat management for biodiversity
maintenance using aerospace remote sensing. In: Tropical
Ecosystems: A synthesis of Tropical Ecology and Conservation,
pp. 308-345. (Balakrishnan, M., Borgstrom, R. and Bie, S. eds).
Oxford & IBH Publishing Co. Pvt Ltd, New Delhi (1994)
28. Roy et al., Analysis of vegetation types using Remote Sensing
techniques for wildlife habitat evaluation Kanha National Park, Proc.
Seminar-cum-Workshop on WildhJi, Habitat Evahtation using Remote
Sensing technique, Dehradun, India (Jointly organised by IIRS and
WII), pp 83-116 (1986)
29. Schamberger M, Krohn WB: Status of the habitat evaluation
procedures. Trans North American Wildlife Natural Resource
Conference, 47:154-164 (1982)
30. Singh Ashbindu: Change detection in the tropical forest environment
of North-East India using Landsat In: Remote, Sensing and Tropical
land management Eds M., J Eden and J T Parry (1986)
31. Smith, J. L. D. et al.: Landscape analysis of tiger distribution and
habitat quality in Nepal. Conservation Biol. 12: 1338_1346 (1998)
32. Sperduto MB: Congleton Predicting rare orchid (small whorled
pogonia) habitat using GIS, Photogrammetric Engineering and
Remote Sensing, 62: 1269-1279 (1996)
33. Stephens, D. W. and J. R. Krebs:Foraging Theory. Princeton, New
Jersey: Princeton University Press (1986)
34. Thomas J W. Maser C and Roderick: Piparian zones In: Wildlife
habitats in managed forest - the Blue Mountains of Oregon and
Washington (ed.), lack Ward Thomas. U.S. Dept. Agriculture
Handbook, pp. 553 (1979)
35. Wikramanayake et al.., Designing a conservation landscape for
tigers in human-dominated environments. Cons. Biol. 18 (3): 839-
844 (2004)
36. Zarriet al.: Habitat suitability assessment for the endangered Nilgiri
Laughing thrush: a multiple logistic regression approach. Current
Science, 94: 1487-1494 (2008)