The problem of imperfect detection is addressed by carrying out the survey and the data analysis in an occupancy framework described by MacKenzie et al (2002). The authors suggested a new method based on “detection-non-detection” of a species. The occupancy method allows the use of proportion of area occupied (POD) as a low-cost surrogate for species abundance. The occupancy model is based on the premise that changes in the proportion of area occupied by a species may be corresponding with changes in its population size. Presence/absence surveys can be conducted at a number of sites across a broad landscape, with the history of its presence/absence being maintained. The model allows building detection probability built over capture history of the species and also incorporates habitat covariates such as habitat types, forest type, vegetation composition and biotic influences to account for variation in detectability and occupancy. This also takes in to account variations in occupancy based on habitat characteristics. For a large-scale species survey, proportion of area occupied is a reasonable state variable to be used as suggested by Mackenzie et al. (2002, 2003, 2004) and Linkie et al (2007).
The first objective of our survey was to evaluate the distribution status of Malabar giant squirrel in the Western Ghats of Maharashtra. This was achieved by using the occupancy modeling framework (Mackenzie et al. 2002).
The area of the survey was from Northern boundary of Harishchandragad-Kalsubai WLS of Ahemadnagar District in the north to Chandgad Taluka of Kolhapur District in the south. The survey was carried out in a grid-wise manner. The study area was divided into grids of 2.5 minutes (latitude) x 2.5 minutes (longitude). This is equivalent to a rectangle of 4.332 km (latitude) x 4.613 km (longitude) at the northernmost extreme and 4.465 km (latitude) x 4.610 km (longitude) at the southernmost extreme of the surveyed area. Degrees were used as a unit for demarcating the grids instead of kilometers because it is a more intuitive unit to use in the field with help of a GPS and does not make a substantial difference to the grid size. The average grid size was 20.28 km2 A total of 184 grids were sampled in the study area. Each grid was divided into four sub-grids of 1.25 minutes (latitude) x 1.25 minutes (longitude) that is 5.07 km2. Each sub-grid represented one spatial replicate. So each grid had a minimum of one and a maximum of four spatial replicates. If a giant squirrel was detected in a spatial-replicate it was given a ‘1’ and if it was undetected it was given a ‘0’. In this way, we created capture histories for each of the 184 grids.
The second objective was estimating the density of the giant squirrel in the study area. Distance sampling along line transects is a popular technique for population estimation of animals. Jathana et al (2008) used distance sampling along pre-marked straight line transects to estimate giant squirrel populations. Marking line transects for measuring abundance is not a feasible option given the extensive study area and limited study period of 9 months. The survey was therefore carried out along natural trails and paths in the forests as an alternate options suggested by Krishna and Hiby (2001).There is a possibility of bias in this approach because trails may influence the distribution of species by either attracting a species or repelling it for various reasons such as change in habitat and resource availability. A second cause of bias can be change of detectability along trails due to openings in the canopy due to trails. In the present survey a vast majority of the sampling trails passed through closed canopy forests with the forest canopy covering the trails so that there was no break in the canopy. Since the canopy above the trails was generally continuous squirrel distribution is unlikely to be affected by the presence of the trails. Because of closed canopy detectability was also unlikely to be affected by the presence of the trails. Hence we contend that distance sampling carried out during our survey was substantially unbiased.
We used number of squirrels per total effort (encounter rates =n/l) to compare its relative abundance across the habitat types within the study area.
3.4 Threat Assessment
The Third Objective was to assess the threats to the giant squirrel population in the study area. Some types of threats such as firewood collection, lopping, tree cutting and cattle grazing are localized and can be assessed by direct observations on the trail. A few other types of threats are understood better by evaluating the larger area as they may not be localized but may have long-term, permanent and even continued impact. Examples of such types of threats are poaching, hunting, habitat loss due to fragmentation and deforestation, impact due to old and new development projects, and disturbance due to human presence.
Livestock presence: Under this category, presence of livestock, and signs of cattle dung were included.
Firewood Collection: Observations of people carrying head-loads or signs of cut firewood were recorded.
Tree cutting and lopping: Signs of old and new stumps of trees, cut branches, illegal felling of trees, stacked logs, and timber harvesting were recorded.
Poaching and hunting: Direct information on squirrel hunting could not be collected because it is usually done on the spot by slicing of nests, catching the pups of with guns. On the other hand, use of snares, traps and hides were easy to detect as they are installed in the forests. Presence of snare etc does not directly indicate hunting of squirrel but it does indicate overall hunting pressure in the area.
Human Presence: Presence of people, tourists, and villagers were recorded.
Presence of Domestic Dogs: Domestic dogs are used often for hunting of squirrels and it also indicates human presence in the area.
Development Projects: Upcoming projects for highways, canals, dams, mining projects, windmills, railways were recorded by discussing with local officers and people.
Plantations: Plantations of teak, eucalyptus or any other species observed on the trails were recorded.
Information was collated from findings based on occupancy analysis, density estimates and threat assessment for individual sites for identifying vulnerable populations of giant squirrels.