2.5 Population abundance and density estimation
A free software program SPACECAP package version 3.0.2 (Gopalaswamy et
al., 2012) was run to estimate snow leopard density from genetic
samples; a user-friendly software package, implementing a Bayesian
Spatially Explicit Capture Recapture (SECR) analysis (Royle et al.,
2009). This software is recommended by Efford et al. (2009) for spatial
CR data originating from different types of non-invasive sources, such
as camera traps, hair snares or fecal DNA samples. SECR models in the
SPACECAP package directly estimates animal density by explicitly using
information on capture histories in combination with spatial locations
of captures under a Bayesian modeling framework.
Following Singh et al. (2010), three types of input files were prepared
to analyze data in the SPACECAP package: I. e. (i) Animal Capture
Details, (ii) Trap Deployment Details and (iii) State-space Details. The
trap deployment file consisted of trap location ID and spatial location
of trap IDs in X and Y- coordinates (Universal Transverse Mercator UTM
projection system) along with information on the occasions when each
individual location was operational during the sample collection
duration. Here the mid-center of grids was used as detectors (location)
and we also reduced the number of sampling occasions by pooling the data
together across three days due to a low encounter probability, as
suggested by Janecka et al., 2011. The trap deployment data were
organized in a two-dimensional matrix of individually identified
locations and sampling occasions in a binary format (1s and 0s),
indicating whether an individual location was or was not operational on
a particular sampling occasion. A mesh of points (potential home range
center) in each cell (0.25 km2) was generated using
the Arc Geographical Information System (GIS) in a sampling array and
1.6 km buffer. The potential home range center file consisted of X and Y
coordinates of all the potential activity centers in the UTM Projection
System, and a habitat suitability indicator (using 1s or 0s)
representing the potential activity centers within suitable or
unsuitable habitat. The analysis was run in SPACECAP v 3.1 under the
program R environment. For MCMC simulation, 60000 iterations were
selected, 1000 burn in values (number of initial values to discard
during the MCMC analysis) and 1 as thinning which was recommended by
Royle et al. (2009). Only iteration numbers defined by the thinning rate
are stored for the analysis (Singh et al., 2010). Since there was
uncertainty about the total number of animals, which was likely to be
larger than the minimum number identified during fecal DNA analysis, the
value needed to be “augmented”. Here fecal DNA analysis identified 19
individual snow leopards in this survey, thus data augmentation has been
set at 100 as a rule thumb this was 5 times greater (Royle et al.,
2009).