Lichen communities as a multiscale correlative indicator of elevational and land use-land cover gradients in the Himalayas

Elevation and land use/ land cover (LULC) plays an important role in the diversity of lichens in the Himalayas. The elevation gradients and LULC can be remotely assessed using remote sensing (RS) and geographical information systems (GIS). The current study was done in the Chopta-Tungnath landscape in the Kedarnath wildlife sanctuary, western Himalaya, India. Digital elevation modelling of the study area was done using shuttle radar topography mission data (SRTM-DEM) processed in Esri ArcGIS® ArcMAP TM 10.5, to assess the elevation gradient of the study area and selection of four lichen sampling sites. The LULC maps of the study area were prepared using Landsat 8 and Google Earth Pro 7.3.2.5776 imagery processed using Leica TM ERDAS IMAGINE® 9.2. An elevation gradient of 2750 m to 3703m was recorded by SRTM-DEM. The LULC analysis resulted in ve LULC classes of which the four sampling sites fall in the 3 LULC classes. The principal component analysis (PCA), used to analyse the lichen communities along the RS-GIS recognized LULC classes. The study found lichen communities to be a proxy to the LULC classes in the Himalayas with clear gradients of growth forms and habitat subsets along the increasing elevation gradient. al. 2019). The Google Earth imagery, an open-source data freely available has been instrumental for visual supervised classication of Landsat data and have been found ecient in the overall remote sensing classication of LULC (Tilahun and Teferie 2015; Uddin et al. 2015; Debnath et al. 2017; Sharma et al. 2018; Mondal et al. 2019). In the present study, we have attempted digital elevational modelling of shuttle radar topography mission data in site selection and dening the elevation gradient of the Chopta-Tungnath landscape. The LULC estimation was done using Landsat-8 and Google earth data and their comparative eciency was assessed. The above-mentioned RS-GIS analysis was correlated with the change in lichen diversity along with elevation and LULC changes to examine the capability of lichen communities as indicators of RS-GIS dened LULC in the mountainous terrain of Chopta-Tungnath landscape situated in the southern extreme of the Kedarnath wildlife sanctuary, western Himalaya. The study hereby elucidates the inuence of LULC on the lichen communities along the elevation gradient of the Chopta-Tungnath landscape. The eciency of GEPr-LULC mapping over Landsat 8-FCC indicates their superior remote sensing LULC analytic applications. The clustering of lichen communities to specic LULC with dened combinations of growth forms and habitat subsets, concludes their ability and probable applications as indicators of different vegetational covers and land use in the Himalaya. The ndings can be used for developing forest management policies and can be of considerable help for biodiversity assessment in the Himalayas.


Introduction
Studies on Lichens for the last few decades have been found them to be good indicators of land use in both managed as well as natural habitats re ecting the effects of both anthropogenic (i.e., habitat perturbations and pollution) and natural factors (i.e., invasive species, competition, and land-use intensity) ( In the present study, we have attempted digital elevational modelling of shuttle radar topography mission data in site selection and de ning the elevation gradient of the Chopta-Tungnath landscape. The LULC estimation was done using Landsat-8 and Google earth data and their comparative e ciency was assessed. The above-mentioned RS-GIS analysis was correlated with the change in lichen diversity along with elevation and LULC changes to examine the capability of lichen communities as indicators of RS-GIS de ned LULC in the mountainous terrain of Chopta-Tungnath landscape situated in the southern extreme of the Kedarnath wildlife sanctuary, western Himalaya.

Study area:
The study was conducted in the temperate-alpine habitats of Chopta-Tungnath (between 30°28'39"-30°29'51" N latitude and 79°12'9" to 79°13'21" E longitude), a pasture and a trekking-pilgrimage area situated on the south-western fringe of Kedarnath wildlife sanctuary in the Garhwal Himalayas of Rudraprayag district, Uttarakhand ( Figure 1A). The climate of the landscape is characterized by severe frost, diurnal to seasonal blizzards, hailstorms, and daily orographic precipitation at higher altitudes, throughout the year (Khare et al. 2010).
The precipitation ensues in the form of snow, sleet-hail, rains, and showers throughout the year (Rai et al. 2012). The snowfall occurs from November to April. Snow melting in April is the major source of soil water before the monsoon. Maximum rainfall is recorded in July-August ( Figure 1B). The mean monthly atmospheric temperature ranges from a maximum of 19°-37°C (May to October) to a minimum, as low as -15°C (December to February) ( Figure 1B).
The area is known for the shrine of Tungnath, situated at the Tungnath ridge a major relief structure dividing the drainage of the region.
The shrine is associated with alpine grassland (the Tungnath Bugyal). The Tungnath shrine lies 2 km below the Chandrashila Peak, the highest point of the landscape. The landscape is characterized by rocky outcrops having moderate to steep slopes. The topography of the area is dominated by ridges formed by exposed rocks and patches of at temperate and alpine grasslands. The soil in the area is thin-layered, coarse-textured/ sandy loam at lower altitudes and sandy at higher altitudes, with proper drainage and acidic pH (pH 3.0-5.5) 3.2. Remote sensing (RS) and geographical information systems (GIS) analysis: 3.2.1. Topographic studies using shuttle radar topography mission-digital elevation model (SRTM-DEM) data: The topographic study of the Chopta-Tungnath landscape was done on shuttle radar topography mission-digital elevation model (SRTM-DEM) data. The SRTM was an international effort consisting of an 11-day mission of space shuttle Endeavour in February 2000, which mapped almost the entire earth from 56°S to 60°N with STRM payload employing interferometric synthetic-aperture radar technique and obtained a high-resolution digital topographical data of earth (Nikolakopoulos et al. 2006). SRTM-DEM data covers compiled by Consultative Group for International Agriculture Research Consortium for Spatial Information (CGIAR-CSI) covers about 80% of the globe.
The SRTM 90m DEMs are with a resolution of 90m and are available for free download as 5×5-degree tiles at http://srtm.csi.cgiar.org/.
The 5×5-degree tile, SRTM, version 4, 90 m data of the study area was downloaded as Geo TIFF le from http://srtm.csi.cgiar.org/srtmdata/ (Fig.2). The data le was pre-processed for noise reduction, identi cation, and elimination of manmade terrain features (if any), and for estimating and eliminating the forest canopy data (Köthe et al. 2009). The SRTM Geo TIFF le was processed in Esri ArcGIS ® ArcMAP TM 10.5. The study area was clipped from the regional downloaded SRTM Geo TIFF le, using the shape le created from Google Earth Pro 7.3.2.5776 (GEPr). The processed SRTM data along with prior knowledge of the authors was used for the selection of probable sample areas. The nal georeferenced and geotagged elevation map of the study area with sampling sites and landmarks was prepared using cumulative input data from SRTM-DEM, GEPr, and previous eld studies (Fig.2).

Land use land cover (LULC) classi cation:
The collative use of google earth imagery and Landsat data was done to assess and prepare the land use/ land cover of the study area ( Fig. 4). The Google earth land use/land cover map was prepared by visual interpretation of data based on size, shape, tone, texture, association, and relationship to other objects (Fig. 4).
To study the Landsat-based land use/ land cover (LULC) of the study area, the Landsat 8 satellite data (path 145, row-39) of 18 th February 2014 and 10 th June 2014 was downloaded from the Earth Explorer website (https://earthexplorer.usgs.gov/ ). The satellite data was imported, stacked and the subset of area of interest (AOI) was created using Leica TM ERDAS IMAGINE ® 9.2. The image was processed for preparing the unsupervised false-colour composite (FCC) map of the AOI using three bands (5, 4 and, 3), which was used in the eld excursions for ground-truthing. For Google earth (GE) LULC studies the GE imagery of the Chopta-Tungnath landscape was downloaded using Google Earth Pro 7.3.2.5776 (GEPr). The images were imported, stacked, noise reduction, image enhancement, and georeferencing were done as pre-processing (Fig. 3). The GEPr imagery was processed for the preparation of the LULC map of AOI using 8 classi cation classes in Leica TM ERDAS IMAGINE ® 9.2. The preliminary LULC maps i.e., FCC image by Landsat 8 and GEPr LULC class maps were nally interpreted by supervised classi cation based on ground-truthing (Fig. 3).
Reconnaissance eld visits/surveys were carried out in different months during 2014-2018 to establish the relationship between land use/ land cover and their tonal variation on the satellite data ( Fig. 4). Ground truthing of the Landsat 8 FCC maps/ GEPr LULC class maps was done using handheld GPS (Garmin GPSMAP ® 76S).
The nal land use/land cover (LULC) was interpreted using digital and visual analysis of Landsat 8 satellite/ GEPr-LULC data. Supervised classi cation was performed with the training sites of known targets and then the spectral signatures of these sites were extrapolated to other unknown classes (Fig. 3). For this, the Gaussian maximum likelihood classi cation (GMLC) algorithm was used. The classi er used the training statistics to compute a probability value of whether it belongs to a particular class, which allows for the within-class spectral variance. In this image, the analyst used prior knowledge to weigh the probability function. GMLC provided the highest classi cation accuracies (Lillesand et al. 2015). For visual analysis elements of visual interpretation like tone, texture, shadow, was used to classify the land cover of the study area using Google Earth imagery and ground-truthing observations during eld visits (Fig. 3). The LULC class area of each land use was calculated in km 2 and percent. The comparative error matrix and accuracy of visual (Google Earth Pro) and digital (Landsat 8, 2014) interpretation for the LULC classes were assessed using kappa index statistics and area measurements.  The SRTM-DEM obtained with a pixel size of 90 m covered a total area of 10.25 km 2 with elevational variation ranging from 2750 m to 3703m (Fig. 5). The collative LULC studies using Landsat 8, FCC, and Google Earth imagery identi ed ve LULC classes ( Table 2, Fig 6).
Among the identi ed LULC classes the mixed conifer forest dominated followed by temperate grassland, Rhododendron sub-alpine forests, alpine grassland, and snow ( Table 2). The accuracy assessment found the LULC classes derived by the visual interpretation using Google Earth Pro imagery to be more e cient than the LULC classes derived by the digital classi cation using Landsat 8, 2014 data ( Table 3). The reconnaissance eld visits/ surveys carried out during the study period (i.e., 2014-2018) further observed strati cation of vegetation along increasing elevational gradients in the landscape. The LULC classes identi ed through RS-GIS studies were de ned by these vegetational strati cations. The mixed conifer forests were dominated by strands of Quercus semecarpifolia and Rhododendron arboreum with few patches of Abies pindrow and Taxus baccata trees (Fig. 7). The temperate grasslands developed in the open canopy area in the coniferous forests (Fig. 7). The Rhododendron sub-alpine forests entirely consisted of coppices of Rhododendron campanulatum (Fig. 7). The alpine grassland was dominated by the vegetation of herb species of Anemone, Potentilla, Aster, Geranium, Meconopsis, Primula, and Polemonium, with scattered patches of shrubs of Rhododendron anthopogon and Juniperus species (Fig. 7). 4.2. Average lichen community structure, patterns.

Lichen communities and RS-GIS de ned LULC classes:
The PCA analysis required 3 components (axis) to account for a 100% variation in the data set. The rst two axes of PCA explained 87.2% of the variance, and each axis explained 69.6 and 16.6 % of the variance, respectively (Fig.9). Sites 1 and 2 mapped separately whereas sites 3 and 4 mapped coherently due to their inherent similarity and differences in the diversity of constituent lichen species at each site (Fig. 9). The PCA biplot further concluded that the lichen community at site 1 was indicative of the RS-GIS recognized LULC class mixed conifer forest, whereas site 2 of Rhododendron sub-alpine forest and sites 3 and 4 of alpine grassland (Fig. 9).

Figure 3
An overview of RS-GIS based land use land cover (LULC) analysis.

Figure 4
The detailed map of study area depicting all the sampling sites and landmarks in Chopta-Tungnath landscape.

Figure 5
The SRTM-DEM visualized map with sampling sites and major landmarks tagged.