Content uploaded by Parthankar Choudhury
Author content
All content in this area was uploaded by Parthankar Choudhury on Jul 20, 2021
Content may be subject to copyright.
44
Original Research Article
SpatialDistributionandSuitableHabitatsofFeralHorsesinDibruSaikhowa
National Park cum Bi osphe re Re serve: A Studyusing G eospatial
Technologies
Shamikhu Changmai1, Parthankar Choudhury1* and K. K. Sarma2
1Wildlife Conservation Research Laboratory, Department of Ecology and Environmental Science, Assam University, Silchar,
788011, India
2North Eastern Space Applications Centre, Umiam, Meghalaya, 793103, India
*Corresponding author: parthankar@rediffmail.com
Received: October 9, 2020; revised: March 13, 2021; accepted: March 30, 2021
https://doi.org/10.17605/OSF.IO/J8UXF
Abstract: Identifying spatial distribution status and assessing suitable habitats of animal species are important prerequisites
for its conservation and habitat management. Dibru Saikhowa National Park cum Biosphere Reserve is the one of few known
places in India where feral horses are found. A study was carried out to investigate their spatial distribution and thereafter to
find their suitable habitats using geospatial technologies. Unsupervised classification was executed over Landsat 8 OLI image to
understand the land use land cover of the study area. Ground Control Points collected using GPS device were attributed with
biological data, thus formed the point layer, that has been interpolated to assess the spatial distribution of feral horse herds.
Suitable habitats were identified using Weighted Overlay Method. Field observation and secondary data has also been considered
to set the final weightage. The study area is dominated by semi evergreen patches, scrub and open forests, salix, swamps and
grasslands. Feral horses (around 150-160 individuals) were estimated, and found to be distributed in partially connected small
river islands called
chaporis
, mostly dominated by grasses mixed with
Tamarix dioca
and other non-grass species. The area is
threatened by severe riverine pressure, human interference and livestock grazing. Geospatial tools are important for modelling
of wildlife habitat with less human effort, cost and time. The output of the study is a scientific documentation having scope of
further extension, and it will be useful for conservation of Indian feral horses in particular habitats.
Key words: Chapori, Dibru saikhowa, Feral horse, Habitat, Spatial distribution, Weighted overlay.
Introduction
Feral horses (
Equus ferus
) are large bodied mobile grazers
adapted to a range of flexible physiological, behavioral and
morphological attributes that have enabled them to thrive
under a wide range of ecological conditions (Beever, 2003).
Feral horses usually live and move in groups, known as herds.
Each herd usually consists of 3-5 individuals, but sometimes
12-15 individuals may also occur (Berman, 2008). Feral horses
mostly prefer grassy river flats, forest and woodland habitats
(Lenehan, 2010). They use open areas where predators can
be seen from a distance. Studies have suggested that these
animals occur commonly within 5-6 km range of water.
Availability of feral horses has been reported from many
countries across the world. In India, however, the only known
place where feral horses are found is Dibru Saikhowa National
Park cum Biosphere Reserve (hereafter referred as DSNP-
BR). These horses are the offspring of the army horses left
Journal of Bioresources 8(1): 44-51 (2021)
Online: ISSN 2582-2276
45
after the World War II (North East News, 2020). Feral horses
are locally known as
Janghali Ghura
or
Bonoria Ghura
. After
the massive earthquake of 1950, these animals remain trapped
in the floodplain island of Dibru Saikhowa (Bhuyan, 2011).
However, there is no scientific documentation available about
the Indian feral horses.
Geospatial technology plays an important role for
monitoring and conservation of biodiversity as well as in habitat
management. Biological database along with ground data can
be mapped and incorporated geographically to decide
conservation priorities for a particular species (Manel
et al.
,
2001; Salem, 2003). Use of GPS devices has greatly enhanced
wildlife researches and conservational activities with better
accuracy (Cagnacci
et al.
, 2010). With gradual development in
geospatial technologies, the collection and analysis of field
data and its prediction has become much easier and effective
(Sonti, 2015). Information acquired from remote sensing data
supported by field evidences of the presence of animal species
have been proved versatile for wildlife habitat suitability
analysis (Roy, 1993; Roy and Ravan, 1994). LULC of a particular
area is prepared using satellite data and then the known
habitat preference and environmental conditions of the species
based on ground observation are evaluated. The qualitative
ratings in each information layers can be modeled to develop
habitat suitability maps (Ejigu, 2016).
The present study was carried out with aim to
investigate the spatial distribution status of feral horses in
DSNP-BR and to identify their suitable habitats using various
tools of Remote Sensing and GIS. The outcome of the study
is a baseline documentation on Indian feral horses.
Materials and methods
St udy Area
The present study has been carried out in DSNP-BR situated
in flood plain areas of the river Brahmaputra, towards eastern
corner of the state of Assam, India. The park extends between
27°30¹3ºN – 27°47¹30º N latitude and 95°10¹16ºE – 95°45¹10º
E longitude covering an area of 340 sq. km. The area is bounded
by mighty Brahmaputra and Siang in the north, Lohit, Dibang
and Noa Dihing in the east and Dibru in the south (Figure 1).
The elevation of the area ranges from 110m to 126m with an
average of 118m. The area experiences tropical monsoon with
long, hot summer and short, cool winter. Annual average
rainfall is 2300 to 3800 mm (Choudhury 1998). The area is
mostly dominated by semi-evergreen forests, grasslands, salix-
swamps and scrub forests disturbed by human activities.
Land Use Land Cover c lassification and hab itat
mapping
It is necessary to understand the landscape with proper
identification of its features for wildlife habitat studies and its
management. Several information about the earth surface can
be extracted from remotely sensed satellite images. In the
present study, Landsat 8 OLI satellite image with 30m spatial
resolution and 134/41 (Path/Row) procured from USGS Glovis
was used to classify the study area. Unsupervised classification
technique was executed on the subset area of interest (AOI)
to extract meaningful classes of interest using ISODATA
classification algorithm in ERDAS Imaging. Unsupervised
classification provides clusters of pixels with similar spectral
characteristics that have to be interpreted into meaningful
classes of interest (Jensen, 1996). In a very heterogeneous
area with complex spectral variations, unsupervised
Fig. 1. Location map of DibruSaikhowa National Park. This figure shows
where the study area (i.e. DibruSaikhowa National Park) is located and
what are the major rivers surrounded the park.
Shamikhu Changmai et al., 2021 Feral horsesin Dibru Saikhowa National Park
46
Fig. 2. Interpretation keys for LULC classification. This figure shows the
sample pixels taken from the original subset satellite image to classify each
LULC category. Based on the pixel characteristics and ground knowledge,
the final LULC map of the study area has been prepared.
Note: Images shown against each LULC category in Figure 2 represents
sample of pixels taken from the original subset satellite image of the study
area. Based on pixel characteristics and ground knowledge, the final LULC of
the area has been prepared.
classification provides accurate result (Rozenstein and Karnieli,
2011). The classified image was filtered using a neighbourhood
majority function with 3×3 matrix. Recoding was performed
for misclassification of features based on ground knowledge
and very high resolution image from Google Earth. Landsat
image was interpreted using some important interpretation
keys (Fig. 2).
Estimation of spatial dist ribution of feral horses
Interactions with local people and forest staffs; and extensive
field survey using GPS device was conducted to record the
Fig. 3. Flow chart of Kriging Interpolation method. This figure shows the
overall procedure adopted to interpolate a raster surface from collected ground
points. Interpolation process allow to estimate and predict unknown value from
a known geographic point data within a specified distance. This method was
adopted to estimate the spatial distribution of feral horses in the study area.
Fig. 4. Flow chart of Weighted Overlay Analysis. This figure shows the
overall procedure adopted to identify suitable habitats of feral horses. The
LULC classes were converted to raster layers and then assigned each class
with some predetermined weights. Various parameters were considered to
evaluate suitable habitats. All reclassified raster files were added in weighted
overlay tool with correct scale value to identify suitable habitats.
Shamikhu Changmai et al, 2021 Feral horsesin Dibru Saikhowa National Park
47
locations of feral horses. During the field survey, the number
of herds encountered, number of individuals found in each
herd, their movement, daily activities, the type of habitat where
they are found, etc. was periodically observed and recorded
accordingly. GPS locations were collected wherever feral horses
were encountered. Every ground control point (GPS location)
was attributed with the biological data collected through
physical observations. The ground locations collected using
GPS thus formed the point layer that was interpolated using
the Kringing Ordinary with spherical semivariogram model
(Fig. 3) to generate a continuous surface of spatial distribution
of feral horses within the study area.
Analysis of suitable habitats
Spatial decisions involve a large set of datasets, parameters
and different alternatives. Accordingly, many spatial problems
give rise to GIS based Multicriteria Decision Analysis (MCDA)
or Multicriteria Evaluation (MCE) (Malczewski, 2006). MCE
was performed using a raster based map algebra expression.
Important factors like available food resources, distance from
water bodies, habitat type, anthropogenic disturbances,
highlands, etc. were considered to evaluate suitable habitats
(Talukdar
et al.,
2020). All vector files were converted to raster
and then reclassified using reclassify tools. LULC classes were
assigned with certain predetermined weights (1=
Not Suitable
,
2=
Least Suitable
, 3=
Moderately Suitable
, 4=
Suitable
and 5=
Highly Suitable
). The reclassified raster files were added as
input in weighted overlay tool with equal and correct scale
values. Thus, the suitable habitat sites were identified after
performing Weighted Overlay analysis in ArcGIS (Fig. 4).
Res ults
Land Use Land Cover of the study area
The study area is dominated by semi evergreen forests which
consist of evergreen forests along with some patches of moist
deciduous forests. Scrub forests are found at the peripheries
of dense forests and located closer to horse habitations. These
areas are open type with biotic as well as abiotic interferences.
There are grasslands mixed with
Tamarix dioca
and other non-
grass species. Wetlands are found in the form of small ponds or
beels. Rivers and streams are also found. Sandy areas are found
along river beds. Rural builtup consisting of two forest villages
viz. Laika and Dadhia along with agriculture including all types
of farm lands are found in the study area (Fig. 5).
Spatial distribution of feral horses
Feral horses were recorded towards northern and north
eastern portion of the reserve. The horses were found to stay
Fig. 5. Land Use Land Cover of DibruSaikhowa. In this figure, the major
land use land cover types found in and near DibruSaikhowa has been
represented. This help to understand the overall landscape of the study area
and to know the type of habitat present with visual information.
Fig. 6. GPS locations of feral horses encountered during the study. In this
figure all those locations are show where feral horses were encountered
during the filed survey. Every location was collected using a GPS device.
Total of 144 times feral horses were encountered at different locations.
Shamikhu Changmai et al., 2021 Feral horsesin Dibru Saikhowa National Park
48
Fig. 7. Preferred habitats where feral horses were found. This figure represents
the number of times feral horses are encountered in a particular type of
habitat during the study. The figure clearly reflects the preferred habitat
types. Horses were maximum found in grassland and riverine habitat.
Fig. 8. Spatial distribution of feral horses during winters. This figure
shows where feral horses are distributed in winter months during the study.
The spatial distribution was developed from Kriging Interpolation. The blue
portions show places with less distribution, and towards red portions their
distribution gradually increases.
in groups of 3 to 9 individuals. However, on few occasions,
single individual was also recorded. 144 times feral horses
were encountered at different locations (Fig. 6). Some herds
were found to share common area and they follow similar
movement. During the study no feral horses were found in
dense forest areas. Ten times horse herds were found near
cattle farms (locally called as
khuties
). Thirteen times these
animals were found in dry river beds completely dominated
by sand. Their availability gradually increases near water bodies
and grasslands. Twenty-two times found in wetlands dominated
by different grass species; and twenty-eight times found in the
areas mostly dominated by
Tamarix dioca
and short grasses.
However, maximum horse herds were found in riverine
grasslands mixed with
Tamarix dioca
and other shrubs (thirty-
five times), and thirty-six times in open highland areas (Fig.
7). The horse herds were mostly found distributed in open
areas dominated by riverine grasslands mixed with
Tamarix
diocea
and other non-grass and shrub species, specially
between the rivers Brahmaputra, Siang and Dibang.
Interestingly, feral horses show different distribution
pattern in different seasons. As shown in the Figure 7 and 8, red
portions are the areas where maximum horses were found and
they were distributed closely. On the other hand, their numbers
and distribution gradually decreases towards blue portions.
During the study it was found that horse herds were distributed
widely during winters (Fig. 8). There is scarcity of grasses and
other plants in winters. Moreover, they can easily move across
their habitat for grazing and other activities. Another important
reason is that horses frequently visit near
khuties
in search of
food and salts. On the other hand, during summers, the horses
were found to be distributed mostly along the bank of Dibang
(Fig. 9). There is easy availability of food plants. Moreover,
horses cannot move to long distance due to flood.
Suitable habitats
Horizontal and vertical study of natural habitat across the
landscape for any wildlife species is very crucial in wildlife
ecology. It is essential to know about the requirements for
food, water, reproduction, pray-predator relation, competitors,
etc. The suitability of habitats was categorised into five levels
of suitability (Fig. 10). The study reveals that, open highland
grasslands are highly suitable habitats (represented as green
portion). Riverine grasslands mixed with
Tamarix dioca
and
other shrubs are suitable habitats (represented as lemon green
portion). Riverine beds, near water bodies and also those
areas near to
khuties
are moderately suitable habitats
(represented as yellow portion). Areas near human
settlements, agricultural activities and mixed cultivation are
Shamikhu Changmai et al, 2021 Feral horsesin Dibru Saikhowa National Park
49
Fig. 9. Spatial distribution of feral horses during summers. This figure
shows where feral horses are distributed in summer months during the
study. The spatial distribution was developed from Kriging Interpolation.
The blue portions show places with less distribution, and towards red portions
their distribution gradually increases.
Fig. 10. Suitable habitats of feral horses in DibruSaikhowa. This figure
shows the suitable habitats of feral horses in the study area. The area is
categorized into five levels of suitability. Red portions are not suitable habitats,
and towards green portions suitability gradually increases.
Fi g. 11. Feral horses observed in different habitat conditions in
DibruSaikhowa. This figure shows various types of habitat condition where
feral horses are found in the study area.
(i) a horse herd in sandy area in ArnaTapu
(ii) a male horse in river bank near KobuChapori
(iii) a horse herd grazing in grasslands at Churketapu
(iv) a horse herd in wetland near Meili Camp
least suitable habitats (represented as orange portion). While
dense forest areas are found as not suitable habitats
(represented as red portion).
Discussion
There is no readily available scientific database on Indian
feral horse. Therefore, the present study has tried to establish
a baseline documentation on population status, spatial
distribution and suitable habitats of feral horses in DSNP-BR
which is considered as the only known habitat of this creature in
India. Feral horses stay in groups of 3 to 7 or sometimes more
individuals. Around 150-160 individuals have been estimated
during the study. The herds mostly prefer to stay and move in
open areas where predators can be seen easily. However,
movement and distribution of horse herds depend on their
daily activities. In early part of the day herds are mostly found
in open riverine grassland areas. During noon time they are
mostly found under bushy shrubs, and towards afternoon and
evening period they usually visit near waterbodies and swampy
areas dominated by different grass species. Much of the time
they spent in grazing. Open highlands and riverine grasslands
are most preferred habitats. Horses are equally found near
grazing lands and local cattle farms called
khuties
. Suitable habitats
are found mostly in and near
chapori
or
tapu
areas like Churke
Tapu, Arna Tapu, Kathalkuthi near Mieli Camp, Kobu Chapori,
and other nearby areas where habitat is mostly dominated by
riverine grasslands mixed with some shrubs species (Fig. 11).
However, areas with regular human interventions and dense
Shamikhu Changmai et al., 2021 Feral horsesin Dibru Saikhowa National Park
50
vegetation are least preferred by feral horses. Brahmaputra,
Siang and Dibang flow as a natural barrier between the identified
home range and the rest of the world. Annual flood, severe
erosion along with livestock grazing has been identified as major
threats to their habitat. In addition, various human activities
and cattle farms cause destruction of habitat into smaller
fragments. Cattles are found to be major competitors resulting
in excessive grazing in grassland habitats.
Geospatial technology in wildlife sciences is an effective
tool for managing, analyzing and mapping of wildlife data that
offer solutions for planning, problem analysis, monitoring and
conservation of wildlife. The output of the present study has
been represented in the form of maps, which will help in easy
understanding with better visual information about the landscape
pattern of the study area, major forest types, seasonal distribution
of feral horses and their suitable habitats, etc. Such readily
available information is helpful for the park management.
Moreover, satellite images of different time periods along with
periodic ground monitoring can reveal about the dynamic
changes occurring in the preferred habitats of feral horses in
the study area. Since, feral horses are found only in DSNP-BR,
therefore, it is of prior necessary to carry out some scientific
studies, create database, and conserve this beautiful animal. The
study has scope for further extension of scientific analyses and
predictions in this area and beyond.
Acknowledgements
The authors are thankful to the forest officials and staffs of
Dibru Saikhowa National Park cum Biosphere Reserve for
allowing us to carry out the study and assist during the field
survey. The authors are thankful to Mr. Joynal Abedin, his
staffs and local people from nearby villages for providing us
necessary information. The authors also extend sincere thanks
to anonymous reviewers for their comments and suggestions.
Refe rences
Beever E. 2003. Management implications of the ecology
of free-roaming horses in semiarid ecosystems of the western
United States. Wildlife Society Bulletin. 31: 887-895.
Berman D. 2008. Horse
Equus
caballus
. In: The Mammals
of Australia, Third Edition. (eds. S Van Dyck and R Strahan),
Reed New Holland, Australia. Pp: 753-754.
Bhuyan A. 2011. Warhorses Gone Wild (2011, June 30).
Retrieved from https://openthemagazine.com/features/india/
warhorses-gone-wild/. Cagnacci, F., Boitani, L., Powell, R.A.
and Boyce, M.S. 2010. Animal ecology meets GPS- based
radiometry: A perfect storm of opportunities and challenges.
Biological Sciences. 365: 2157-2162.
Choudhury AU. 1998. Mammals, birds and reptiles of
Dibru-Saikhowa Sanctuary, Assam, India
.
Oryx. 32(3): 192-
200. doi:10.1046/j.1365-3008.1998.d01-36.x
Ejigu D. 2016. Application of remote sensing and GIS in
ecology. Review Paper. Bahir Dar University, Ethiopia. Pp: 1-48.
Jensen JR. 1996. Introductory Digital Image Processing: A
Remote Sensing Perspective (Second Edition), Prentice Hall,
New Jersey, USA.
Lenehan J. 2010. Ecological impacts of feral horses in grassy
woodland and open forest Gorge country in a temperate subtropical
woodland. In: Ph.D. Thesis, University of New England.
Malczewski J. 2006. Ordered weighted averaging with fuzzy
quantifiers: GIS based multicriteria evaluation for land-use
suitability analysis. International Journal of Applied Earth
Observation and Geoinformation. 8: 270-277.
Manel S, Williams HC and Ormerod , SJ. 2001.
Evaluating presence-absence models in ecology. The need to
account for prevalence. J. Appl. Ecol. 38: 921-931.
North East News. 2020. Why ex-Army personal ‘target’
feral horses of Dibru Saikhowa National Park? Accessed in
March 10, 2020. Retrieved from <https://nenow.in/north-east-
news/assam/why-ex-army-personnel-target-feral-horses-of-
dibru-saikhowa-national-park.html.
Roy PS. 1993. Remote sensing for forest ecosystem analysis
and management. In: Environmental Problems and Prospects
in India, (eds. M. Balakrishnan). Oxford and IBH Publishing
Co. Pvt Ltd, New Delhi. Pp: 335-363.
Roy PS and Ravan SA. 1994. Habitat management for
biodiversity maintenance using aerospace remote sensing. In:
Tropical Ecosystems: A synthesis of Tropical Ecology and
Conservation. (eds. M. Balakrishnan, R. Borgstrom and S. Bie).
Oxford and IBH Publishing Co. Pvt Ltd, New Delhi. Pp: 308-345.
Shamikhu Changmai et al, 2021 Feral horsesin Dibru Saikhowa National Park
51
Rozenstein O and Karnieli A. 2011. Comparison of
methods for land-use classification incorporating remote
sensing and GIS inputs. Applied Geography. 31: 533-544.
Salem BB. 2003. Application of GIS to biodiversity
monitoring. J. Arid Environ. 54: 91-114.
Sonti SH. 2015. Application of Geographic Information
System (GIS) in Forest Management. Journal of Geography
and Natural Disasters
.
5: 145.
Talukdar NR, Choudhury P, Ahmad A, Ahmed A,
Ahmad F and Al Razi H. 2020. Habitat suitability of the
Asiatic elephant in the trans-boundary Patharia Hills Reserve
Forest, northeast India. Modeling Earth Systems and
Environment. 6:1951-1961. https://doi.org/10.1007/s40808-020-
00805-x
Shamikhu Changmai et al., 2021 Feral horsesin Dibru Saikhowa National Park
