Methods

The study was carried out during March – October 2010. Knowledge on the focal landscape was gained through the Survey of India topographic maps (48 J 15, 1978), Google Earth data (http://googleearth.com) and earlier studies on the sacred groves and general landscape (Chandran and Gadgil 1998; Nagendra and Gadgil 1999; Nagendra 2001).

Study of landuse dynamics-
Land use dynamics of the focal region has been studied using data acquired at regular interval through space borne sensors (Landsat Thematic Mapper and IRS-LISS IV (multispectral)). Landsat data of 1989, 1999 and 2010 were downloaded from the public domain (http://www.landsat.org) and IRS data (2010) were procured from the NRSC, Hyderabad (http://nrsc.gov.in). These data were geometrically corrected for the UTM coordinate system of zone 43 using GCPs (ground control points). Land use analysis involved i) generation of False Colour Composite (FCC) of remote sensing data (bands – green, red and NIR). This helped in locating heterogeneous patches in the landscape ii) selection of training polygons (these correspond to heterogeneous patches in FCC) covering 15% of the study area and uniformly distributed over the entire study area, iii) loading these training polygons co-ordinates into pre-calibrated GPS, iv) collection of the corresponding attribute data (land use types) for these polygons from the field. GPS helped in locating respective training polygons in the field, v) supplementing this information with Google Earth (http://googleearth.com) and Bhuvan (http://bhuvan.nrsc.gov.in), and vi) 60% of the training data has been used for classification, while the balance is used for validation or accuracy assessment (Ramachandra et al. 2012).

Land use analysis was carried out using supervised pattern classifier - Gaussian Maximum Likelihood Classifier (GMLC) algorithm. Remote sensing data was classified using signatures from training sites that include predominant land use types. Mean and covariance matrix were computed using estimate of maximum likelihood estimator.  This technique is proved to be a superior classifier as it uses various classification decisions using probability and cost functions (Ramachandra et al. 2012). Spectral classification inaccuracies were measured by a set of reference pixels. Based on the reference pixels, confusion matrix, kappa (κ) statistics and producer's and user's accuracies were computed.

Mapping of current landscape elements-
The position of sacred groves and areas covered by each were ascertained using Global Positioning System (GPS) (Garmin eTrex Vista, USA) and by transferring the data to MapInfo (version 11.0). Other elements of landscape viz, human settlement, forest plantation, forest and agricultural areas were demarcated with the help of the Survey of India topographic maps (1:50000 scale), Google Earth, village maps and forest maps pertaining to the region.  

Status of tree species in sacred groves –
Census of tree populations was done grove wise as groves were relatively small in size, the largest one measuring just 18,000 m2 (i.e.1.8 ha) and rest were mostly in fractions of a hectare. Species wise abundance was recorded and measurements like girth at breast height (GBH, at 137 cm), height was taken for tree individuals (≥ 30 cm GBH).  

To know the future of endemic tree species within the sacred groves, their saplings (> 1m in height but < 30 cm GBH) and seedlings (< 1 m in height) were counted in randomly distributed sample plots (5 m X 5m each) covering 50% of total area of the larger groves (> 5000 m2). For smaller groves i.e. < 5000 m2, the entire forest floor was searched for juvenile members.

Tree species study outside the sacred groves-
A small forest patch was present in the south-west corner of the studied landscape. The plant diversity was measured using transect cum quadrat method. The transect had five tree quadrats of 20 m X 20 m size laid alternatively to the right and left, at intervals of 20 m between quadrats, with nested shrub and herb quadrats i.e. 5 m X 5 m and 1 m X 1 m respectively within these tree quadrats. Our sampling covered a total of 2000 m² area.

Endemic tree species hardly occurred in other elements of landscape such as roadside, household gardens and farms. Nevertheless their ightings if any were recorded.  

Recording of disturbances -
Due to small size, all except the largest three (> 5000 m²) were fully covered for disturbance measurement. For three largest groves (i.e. Ara39, Kal49 and Mat25), fifty percent of the grove area was covered by 10m X 10m plots. The plots were in alternate order along the line transects and were 10 m apart. Seven distinct indicators of disturbance i.e. presence of invasive species, exotic plantation, root exposure (which is an indicator of soil erosion), presence of cattle dung (indicator of grazing), protection status, distance from main road, and grove area were recorded from each studied grove.    

For invasive species (e.g. Lantana camara and Eupatorium sp.), and forest department plantation (Acacia auriculiformis) the degree of presence was categorized on a scale of 1-3: 1= no presence; 2= < 50% in edge (for invasive), and presence in edge (for plantation); 3= >50% in edge / interior region (for invasive) and interior region (for plantation). In order to measure root exposure (an indicator of soil erosion) the following was adopted: 1 = ≤ 10% exposure; 2=25%; 3=≥ 50% . For cattle dung it was 1=no / ≤ 10%; 2=>10%. Both the measures were taken plot-wise, and then averaged for three largest groves (i.e. Ara39, Kal49 and Mat25). The distance of the groves from main road was measured in MapInfo (version 11.0) and categorized on a scale of 1-3: 1=0.1-0.4 km; 2=0.05-0.09 km; 3=0-0.04 km. Grove area was calculated, and categorized based on the size classes (5000 m² , 2501-5000 m2 and ≤ 2500 m2) for further analysis. Protection status was measured by coding: 1= presence of fencing; 2=no fencing.

Data analysis –
Sacred grove distribution – Distances between the groves and the only forest patch was calculated through Euclidean distance by using open source GIS software GRASS (http://ces.iisc.ernet.in/grass). The area-perimeter ratio of the polygons (corresponding to groves spatial extent) was calculated through MapInfo (Version 11.0).

Endemic population status and disturbances- The population data collected on endemic species was categorised into GBH and height classes to understand their distribution pattern.  For forest tree species, non-parametric richness indicators and diversity indices were computed through EstimateS version 8.2 (Colwell 2009). The distribution of juvenile members (saplings and seedlings) in the study area was categorised as per land use types based on their abundance data.  The severity of disturbance has been measured through scoring method with all disturbance parameters given equal weight. The value was expressed in terms of relative disturbance ((scored value/maximum disturbance value)* 100) (Devar 2008). Due to categorical nature of the disturbance parameters a categorical principle component analysis (CATPCA) was conducted to identify the major disturbance factors and their association with groves (SPSS trial version 17).