Endemism-based butterfly conservation: insights from a study in Southern Western Ghats, India

The Western Ghats, a biodiversity hotspot in India harbours a high percentage of endemic species due to its unique and diverse habitats. These species which cannot survive elsewhere due to their specialised habitat requirements are at high risk from climatic and anthropogenic disturbances. The butterfly fauna of the region although well documented has not been investigated intensively at local scales. In this study, we present information on species presence within 10 km × 10 km grid cells (n= 30; area=3000 km2) of 94 butterfly species in the Western Ghats region. The data on the species distribution within these grids which included three wildlife sanctuaries and four forest divisions was mapped. Indicator analysis was performed in R using multipatt function in indispecies package to determine species associated with sites/site combinations. The corrected weighted endemism indices of the study grids were estimated. The data collected over a 4-year period comprised of 393 records of 60 endemic species belonging to five families observed along 102 transects. Troides minos was the most widespread species occuring in 19 grids. Seventeen species indicative of sites and site combinations were obtained, of which Cirrochroa thais, Papilio paris tamilana, Papilio helenus daksha, Parthenos sylvia virens and Mycalesis patnia were significant. The highest corrected weighted endemism index was observed in grid 25 (14.44) followed by grids 24 (12.06) and 19 (11.86). Areas harbouring unique and range restricted species were Parambikulam WLS/TR: Kuthirakolpathy, Pupara, Kalyanathi, Top slip and Muthalakuzhy; Peechi-Vazhani WLS: Ayyapankadu; Thrissur FD: Chakkapara and Vellakarithadam; Nenmara FD: Karikutty, Pothumala and Nelliampathy estate; Vazhachal FD: Poringalkuthu dam, Meenchal and Vazhachal. The study area which covers 2.14% of the Western Ghats hotspot harbours almost 63.82% of the region’s endemic butterfly species making this particular region crucial for butterfly conservation and management. Studying the phylogenetic endemism of the butterflies, identification of microrefugia and testing the mountain geobiodiversity hypothesis with respect to butterflies are the suggested approaches to be adopted for fine-tuning research and conservation of butterflies in this fragile hotspot.


Background
The forests in the Kerala region of the Western Ghats, a biodiversity hotbed, have recently been the target of unpredictable monsoons (Mishra et al., 2018). Studies show an increasing trend of extreme rainfall (Roxy et al., 2017) as well as increases in minimum temperature in the Western Ghats and peninsular India (Dash, Nair, Kulkarni, & Mohanty, 2011;Mondal, Khare, & Kundu, 2015). Butterfly migrations of several species of Papilionidae, Nymphalidae and Pieridae have been reported to coincide with the monsoonal system in peninsular India (Bhaumik & Kunte, 2018;Kunte, 2005). Apart from decades of anthropogenic disturbances leading to forest fragmentation (Jha, Dutt, & Bawa, 2000;Menon & Bawa, 1998;Nair, 1991), recent studies have highlighted other detrimental activities like mining, road construction and irrigation projects (Bharucha, 2006).
In the backdrop of increasing habitat degradation, formulation of priorities for conservation in the Western Ghats is challenging. Many approaches may be adopted but the simplest and most effective would be focusing our attention and action towards reducing the loss of biodiversity based on a framework of vulnerability and irreplaceability (Margules & Pressey, 2000). Subsequently, areas with exceptionally high concentrations of endemic species were prioritised for conservation and the global 'biodiversity hotspot' concept was developed to address this crucial issue (Mittermeier, Myers, Robles-Gil, & Mittermeier, 1999;Myers, 1988Myers, , 1990Myers, , 2003Myers, Mittermeier, Mittermeier, da Fonseca, & Kent, 2000). Although hotspots are designated as areas for priority conservation action, the fauna and flora of many hotspots are poorly studied and relevant data are insufficient for effective conservation planning (Mittermeier et al., 2004;Mittermeier, Turner, Larsen, Brooks, & Gascon, 2011). Moreover, it is also likely that true hotspots may go unrecognised due to lack of organised data, biogeographical biases and regional misconceptions (Noss et al., 2015).
In the hotspot analysis whereby 25 areas were identified, the Western Ghats was among the top eight critical regions in terms of endemism and extent of original primary vegetation (Myers et al., 2000) and later designated as 'hyperhot' for conservation prioritisation (Brooks et al., 2002). The number of global hotspots was later expanded to 35 (Mittermeier et al., 2011). The Western Ghats is a 1600 km mountain chain running almost parallel to India's western coast and spread over six states-Gujarat, Maharastra, Goa, Karnataka, Tamil Nadu and Kerala. It includes two biosphere reserves, the Nilgiris Biosphere Reserve (11,040 km 2 ) and the Agasthyamalai Biosphere Reserve (3500 km 2 ). Kerala lies between 8°1 8′ and 12°48′ N latitude and 74°54′ and 77°12′ E longitude in the south-west region of the Indian peninsula between the Arabian sea and the Western Ghats (Fig. 1). Topographic heterogeneity (from sea level to 2695 m at its highest point) and a strong precipitation gradient (annual rainfall of < 50 cm in eastern valleys to > 700 cm along western slopes) has given rise to remarkable diversity in flora and fauna. The forests in the state are classified into wet evergreen, semi-evergreen, moist deciduous, dry deciduous and thorn forest types (Champion & Seth, 1968). Studies by Reddy, Jha, & Dadhwal (2016) on the extent, distribution and changes in forests of the Western Ghats reveal a net loss of 35.3% of forest   (Gaonkar, 1996). Endemism and species richness are widely used indicators of conservation value and an index combining both has been calculated and mapped at regional, continental and global scales (Crisp, Laffan, Linder, & Monro, 2001;Kier & Barthlott, 2001;Kier et al., 2009;Venevsky & Venevskaia, 2005). However, such studies at local scales (less than 5000 km 2 ) are still scarcely seen. As a result of detailed analyses of Australian flora, Crisp et al. (2001) concluded that the corrected endemism index is a useful method to detect centres of endemism using species-in-grid-cell data. Studies in the last decade indicate that microrefugia may be formed due to topographic variations at scales of metres (Dobrowski, 2011) as well as local influences (De Frenne, Rodríguez-Sánchez, Coomes, et al., 2013) and that these effects are not reflected in GIS models of climate change (De Frenne et al., 2013;Keppel et al., 2017). However, most of these studies are based on flora (Bátori et al., 2017;Keppel et al., 2017;Noss, 2013) and mammals (Camacho-Sanchez et al., 2018). In a study of bush frogs in the Western Ghats, Vijayakumar, Menezes, Jayarajan, & Shanker (2016) highlighted the evolutionary significance of massifs which harbour unique refugia due to steep topographical and environmental shifts. Thus, protection of refugia resulting from mountain topography and climatic stability which buffers lineages against extinction is the current trend as it assures future protection (Klein et al., 2009;Mosblech, Bush, & van Woesik, 2011;Stewart, Lister, Barnes, & Dalén, 2010). Gaonkar (1996) details the state-wise distribution of the butterflies of the Western Ghats and Kunte (2008) delineated their distribution within the four zones whilst assigning conservation values to species.
In this context of impending challenges both biotic and abiotic to the forest ecosystems in the Western Ghats, we venture to ask the key question-Can endemic butterflies be used as indicators for conservation management at local scales in the Western Ghats hotspot? In this study, we have mapped the distribution of endemic butterfly species in an area of 3000 km 2 in the central region of the Kerala part of the Western Ghats and calculated the endemic richness index of various grids of the study area. We expect that the sampling of endemic butterfly species at this microscale level will help detect microrefugia and unique habitats in this fragile hotspot. The indicator species occurring in the study area were identified using the R software.

Methods
The study was carried out in the central region of the Kerala part of the Southern Western Ghats which included three wildlife sanctuaries: Peechi-Vazhani wildlife sanctuary (Peechi-Vazhani WLS), Chinmony wildlife sanctuary (Chinmony WLS), Parambikulam wildlife sanctuary/tiger reserve (Parambikulam WLS/TR) and four forest divisions: Nenmara forest division (Nenmara FD), Vazhachal forest division (Vazhachal FD), Thrissur forest division (Thrissur FD) and Chalakudy forest division (Chalakudy FD). The study area was divided into 10 km × 10 km grids and a total of 30 grids were obtained (Fig. 2). In each grid cell, ten transects were covered over the study period and the length of transects in the different grids ranged between 800 m and 1000 m.   Butterflies in the study grids were sampled along the transects using the Pollard line transect method (Pollard & Yates, 1993) with slight modifications. Butterflies sighted within 5 m on either side and in front of the observer walking at a constant pace of 1 km/h were recorded. The individuals that could not be identified by sight were either caught with an insect net for close examination or photographed and released. The butterflies were identified using suitable keys (Evans, 1932;Kehimkar, 2008;Wynter-Blyth, 1957). The sanction obtained from the Kerala Forest and Wildlife Department (No. WL 10-36790/ 15) for sampling in protected areas in the Western Ghats prohibited collection of endemic species. Hence, identification of smaller species belonging to families Lycaenidae and Hesperidae by sight was difficult. The sampling was done over a 4-year period from May 2015 to April 2019 and the transect data collected along 102 transects (Table 1) was used to map the distribution of endemic species/subspecies onto the grids of the study area. The species occurrence matrix was prepared by scoring the presence/absence (1/0) of endemic species within the sampling grids. The geographic coordinates of transects were marked using Global Positioning System (GPS; GPSMAP 76Cx) which has good receptivity in forest areas. The GPS readings are plotted over geo-referenced Survey of India (SOI) by using open source Geographic Information System (GIS) software. The base layers such as water bodies, forest and boundaries were digitised from SOI topo sheets and updating of layers from latest satellite imageries done using GIS and remote sensing software. The final distribution map of endemic species whereby each point represents the occurrence of a single individual within the study area was prepared using the GIS software (Environmental Systems Research Institute (ESRI), 2011).
Using the sampling data for 4 years, the relationship between the observed species occurrence in the surveyed sites and site groups was analysed and the indicator species determined (De Cáceres & Legendre, 2009;De Cáceres, Legendre, Wiser, & Brotons, 2012;Dufrene & Legendre, 1997). Analyses were implemented in RStudio ver. 3.6.2 (RStudio Team, 2015) and indicative species were identified using multipatt function in package indispecies ver.1.7.9. The total count of species within each grid cell was measured as the species richness. Weighted endemism (WE) is a function of species richness and range size rarity (Crisp et al., 2001;Kier & Barthlott, 2001):

WE¼Σ 1=C
where C is the number of grids in which each species occurs.
The total WE index of each grid was obtained by summing the WE indices of all species recorded in that particular grid. Finally, the corrected weighted endemism index (CWEI) was calculated for each cell by dividing the weighted endemism index by the total count of species in that particular cell (Linder, 2000). Since the proportion of endemics in a grid cell is measured, this index corrects the species richness effect.

CWEI¼WE=K
where C is the number of grid cell in which each endemic species occurs, and K is the total number of species in a grid cell.

Results
The data set for sampling comprised of 94 endemic species (Tables 2 and 3) which include 60 species recorded as endemic to Western Ghats and Sri Lanka (Gaonkar, 1996), 1 species endemic to peninsular India and 33 subspecies reported as endemic to the area (Kunte, Nitin, & Basu, 2018). The number of species occurrence varied from one to thirty-eight and consisted of only presence points. Overall, 393 sightings of endemic species and subspecies were recorded within the 30 grids over the 4 years. The distribution of 60 endemic species/subspecies recorded during the study was mapped onto grids of the study (Figs. 3, 4, 5, 6 and 7).
Indicator analysis identified seventeen indicator species of which five, namely Cirrochroa thais, Papilio paris tamilana, Papilio helenus daksha, Parthenos sylvia virens and Mycalesis patnia were significant at p≤0.001 and the remaining twelve species were significant at p≤0.05 (Table 4). In the former group, two were endemic to Western Ghats and Sri Lanka whilst three were endemic to the Western Ghats at subspecies level. An interesting and unexpected trend noticed is that out of the seventeen indicator species, five species were endemic to the Western Ghats, four were endemic to the Nenmara FD and Vazhachal FD. The number of transects in the above locations which had sightings of these species were as follows: Cirrochroa thais (15), Papilio paris tamilana (8), Papilio helenus daksha (19), Parthenos sylvia virens (17) and Mycalesis patnia (15). The endemic species indicative of Chalakudy FD and Thrissur FD were Troides minos and Pachliopta hector which were common and widespread species.

Discussion
Out of the 94 endemic species and subspecies reported from the Western Ghats, 60 species were recorded and mapped in this study. The study area which covers 2.14% of the Western Ghats hotspot harbours almost 63.82% of the region's endemic butterfly species making this particular region as important and crucial for conservation and management. Assessment of selected sites with respect to butterflies indicate that locations like Vazhachal Reserve Forest, Nelliampathy Reserve Forest, Parambikulam Wildlife Sanctuary/Tiger Reserve, and Peechi-Vazhani Wildlife Sanctuary harbour endemic species and should be prioritised in biodiversity conservation plans. A simple monitoring protocol using endemic butterflies was developed and the GIS mapping provided information on the distribution of endemic species within the study area. These monitoring studies clearly emphasise the well documented fact that reliable field data along with robust analytic tools will help guide conservation of these fragile endemics in this biodiversity hotspot. Kessler and Kluge (2008) postulated that distribution patterns of endemic species along tropical elevational  gradients usually reach a maximum richness between 500 and 2000 m. In this study, the grids with the highest peaks, Padagiri (1585 m; grid 19) and Karimala (1438 m; grid 25) also showed high endemism indices of 11.89 and 14.14 respectively. Mangattu Kumban (grid 9) at mid elevation of 635 m had an endemism index of 7.78 (Fig. 9). These findings thus support the elevational gradient-species richness relationship proposed above. The main advantage of the CWEI is its noncorrelation to species richness and ability to distinguish range-restricted species at a very subtle level. We could identify unique pockets where such species occur within the study area: Parambikulam WLS/TR: Kuthirakolpathy, Pupara, Kalyanathi, Top slip and Muthalakuzhy; Peechi-Vazhani WLS: Ayyapankadu; Thrissur FD: Chakkapara and Vellakarithadam; Nenmara FD: Karikatty, Pothumala and Nelliampathy estate; Vazhachal FD: Poringalkuthu dam, Meenchal and Vazhachal (Fig. 10). The biotic and abiotic factors in these areas should be rigorously studied to determine if they are microrefugial habitats of these rare species.
It is a well-established fact that current spatial distribution and diversity patterns are a reflection of a long evolutionary and biogeographical history. In order to elucidate these complex mechanisms we suggest further studies in this vulnerable hotspot by adopting the following three approaches: (1) studying the phylogenetic endemism (Rosauer, Laffan, Crisp, Donnellan, & Cook, 2009) would help uncover the events that have shaped the rich diversity of this region having Gondwanan, Sundaland and recent biogeographical elements with respect to butterflies. The butterfly fauna of this region with over 300 well documented species would an ideal template for such investigations (2) identification of microrefugia which Harrison and Noss (2017) caution will assume greater relevance against the backdrop of climate change would be another area for butterfly research in this hotspot and (3) finally, the Western Ghats with its steep gradients and undulating terrain would be the perfect arena to test the 'Mountain Geobiodiversity Hypothesis' (Mosbrugger, Favre, Muellner-Riehl, Päckert, & Mulch, 2018) with respect to butterflies. Endemic species are useful indicators of habitat quality and can also act as umbrella species for conservation planning and management. Given the high diversity and endemism among the butterfly communities of the   Kerala part of the Western Ghats, implementation of effective conservation actions would require an integrated approach involving: (1) management of vulnerable and unique habitats at microscale level as landscape level management may sometimes fail to recognise truly 'hot' microhabitats (2) conservation-driven research with emphasis on phylogenetic endemism and microrefugia of species (3) continuous monitoring of habitat and populations based on community forest management through stakeholder participation (4) raising conservation awareness in local communities living in close proximity to and highly dependent on forest resources.

Conclusions
The distribution of 60 endemic species/subspecies was recorded and mapped within the 30 study grids over the 4-year study period. Overall, 393 sightings of endemic species and subspecies were recorded and the family Papilionidae had the highest number of sightings (194) followed by Nymphalidae (116), Pieridae (54), Hesperidae (17) and Lycenidae (12). Out of the total of 37 species endemic to the Western Ghats, 21 species (56%); 18 species of the total of 24 (75%) species endemic to Western Ghats, Sri Lanka and peninsular India and 21 species of a total of 33 (63%) endemic subspecies were recorded. The highest number of Western Ghats endemics was recorded in family Nymphalidae (7) followed by Papilionidae (5), Hesperidae (4), Lycaenidae (3) and Pieridae (2). Indicator analysis identified seventeen indicator species of which five, namely Cirrochroa thais, Papilio paris tamilana, Papilio helenus daksha, Parthenos sylvia virens, and Mycalesis patnia were significant and were indicative of habitats in Chinmony WLS, Peechi Vazhani WLS, Parambikulam WLS/TR, Nenmara FD and Vazhachal FD. The endemic species indicative of Chalakudy FD and Thrissur FD were Troides minos and Pachliopta hector.
The highest corrected weighted endemism index was observed in grid 25 followed by grids 24 and 19. Sixteen grids showed values ranging from 4.07 to 7.75 and seven grids had values between 2.07 and 3.89. The grids with the highest peaks of the study area, Padagiri, Karimala and Mangattu Kumban also showed high endemism indices. Kuthirakolpathy, Pupara, Kalyanathi, Top slip, Muthalakuzhy, Ayyapankadu, Chakkapara Vellakarithadam, Karikatty, Pothumala, Nelliampathy estate, Poringalkuthu dam, Meenchal, and Vazhachal were areas harbouring unique and range restricted species.
This study has shown interesting geographic patterns of the spatial structure of endemism richness in a highly critical hotspot area. Conservation management in the Indian context is expected to benefit if biodiversity can be characterised to more local levels (Bossuyt et al., 2004). This study shows that even within hotspots, endemicity is not uniform and our efforts should be to focus on small areas that represent unique species associations. Even though the addition of more taxa will be useful for a more complete overview, we believe that these are primary areas in the central region of the Kerala part of the Western Ghats that harbour species of conservation value. Moreover, these are also species having a complex evolutionary history and should therefore be monitored and studied in further depth, especially when designing conservation strategies. Thus as we advance into a future wrought with climatic instabilities and increased human impacts, research should be finetuned and the delineation of phylogenetic endemism patterns and identification of microrefugia would definitely be a step forward in the right direction for butterfly conservation in this fragile hotspot.