Methods
Figure 2: Flow chart of the methodology outline
The methodology outline is shown in the fig. 2. A total of 56
occurrences points of two sympatric hornbills i.e Great hornbill (n=31)
and Rufous-necked hornbill (n=25) were collected from the field survey.
The field survey was conducted for four months (December 2018 - March
2019). During the field survey, wherever we have sighted hornbills, we
took the photo, communicating with local villagers and recorded GPS
coordinates (latitude and longitude) along with the elevation. After the
field survey, we used 21 environmental variables (BIO_1 to BIO_19,
Mean actual Evapotranspiration (AET), and Global Aridity Index (AI)
(www.worldclim.org) were downloaded
from WorldClim dataset at a resolution of 1 km2.
Topograhic variables namely-elevation, aspect, and slope were derived
from digital elevation model of Shuttle Radar Topographic Mission (SRTM)
having the resolution of 30m.
A test of auto-correlation was done using DIVA-GIS to remove correlated
sites. After the test, 51 presence location are retained and 5
auto-corrrelated points were discarded( 2 presence location for great
hornbill and 3 presence location for Rufous-necked hornbill. Likewise,
correlation test of 24 variable layers was conducted using the ENM Tools
version 1.3(Warren et al., 2010) to test multi co-linearity between
predictor variables. . The criteria adopted to predict multicollinearity
between the variables was Pearson’s correlation coefficients (r).
Pearson’s correlation coefficient (r) is a measure of the strength of
the association between the two variables. After running the correlation
test, variables with correlation coeffecient value higher than 0.75 were
discarded from the further analysis. Bio_1, Bio_2, Bio_4, Bio_5,
Bio_6, Bio_7, Bio_8, Bio_9, Bio_11, Bio_12, and Bio_19 were
removed and rest 13 variables were only used for generating the model.
To generate the habitat suitability map, all the remaining variables
were first converted to ASCII layer to feed into the MaxEnt 3.4.1. The
model was iterated 100 times with 20 replication using the bootstrap to
avoid any biases or extreme value that may arise and output was acheived
as the mean among them. All rest of the settings we kept as default in
the MaxEnt. The MaxEnt output of ASCII format are converted to raster
and later categorized in ArcGIS for further evaluation.
The suitability map obtained will have the value between 0-1 where 0
represents least suitable and 1 represents most suitable habitat for the
species. The maps were further classified into 4 categories: most
suitable (0.6-1), moderately suitable (0.4-0.6), marginally suitable
(0.2-0.4) and least suitable (0-0.2) (Lu et al., 2012).