Associations of butterflies across different forest types in Uttarakhand, western Himalaya, India: implications for conservation planning

different forest sub-types. Lastly, I would also like to thank the Chief Wildlife Warden, Uttarakhand Forest Department and various Divisional Forest Officers of the state and their field staff for providing logistical support during field surveys, from time to time. Abstrac t : Champion & Seth classified Indian forests into different ‘forest types and sub-types’, based on similarity of dominant vegetation and structural arrangement of species in each. However, it is not known if the species composition and community structure of butterflies is also different in each forest sub-type. If this is the case then each forest sub-type harbouring unique species can be taken as units of conservation on a sub-regional scale. The present study assesses for the first time the species composition and community structure of butterflies across 20 different and prominent ‘forest sub-types’ found across the state of Uttarakhand, western Himalaya. Data collected over eight years (2006–2009; June 2012; 2017–2020) using random seasonal sampling covering 307 transects revealed 370 butterfly taxa. Hierarchical clustering of butterfly abundances revealed seven different butterfly communities spread over 19 forest subtypes. Of these four forest sub-types (3C/C2a moist Shiwalik sal forest; 12/C2c moist temperate deciduous forest; 12/C1a ban oak forest; & 3C/C2c moist Terai sal forest) were identified as most important as they hold most of the butterfly diversity of the state including 58 rare taxa identified according to ‘rarity’ out of the total. GIS based mapping of these 58 priority species over laid on the protected area network and forest cover distribution in the state revealed many forested sites outside the PA network supporting these rare taxa. These sites along a physio- geographical gradient with important forest sub-types and rare taxa can be recommended and listed as new sites for conservation in the state.


INTRODUCTION
Butterflies, amongst invertebrates, are suitable indicators for ecological studies (Lomov et al. 2006), as the taxonomy, geographical distribution and status of many species are relatively well known (Pollard 1977;Thomas 1983;Thomas & Mallorie 1985;Murphy & Wilcox 1986). They are phytophagous, primary herbivores, good pollinators and surrogates plant diversity living close by their food plants (Ehrlich & Raven 1964;Gilbert & Smiley 1978;Pyle 1980). The precise and restricted environmental requirements of particular butterflies make them of considerable value as a group of indicator taxa that indicate the broader effects of environmental changes or reflects a particular suite of ecological conditions or habitat heterogeneity (Pyle 1980;Gilbert 1980Gilbert , 1984Brown 1982;Rosenberg et al. 1986;Murphy et al. 1990;New 1991;Kermen 1992;Pearman et al. 1995). Strong association with vegetation structure and composition makes Lepidoptera a particularly useful bioindicator for monitoring eco-restoration programs (Kremen et al. 1993;New et al. 1995).
Habitat is an important requisite for the proliferation and conservation of a butterfly species (Gilbert & Singer 1975), as species prefer particular habitats, closely related to their life history, breeding, larval and adult food resources and destruction of forest severely affects species habitats (Wells et al. 1983) and many species which were once common become rare. Thus, identification and conservation of priority landscapes, is very important. Champion & Seth (1968) classified Indian forests into different 'forest types' their sub units as 'forest sub-types', based on the similarity of dominant vegetation and structural arrangement of species within each of them, i.e., 'IV montane temperate forest' is one of VI major 'forest types" found across India (other 5 categories being "I. moist tropical forests, II. dry tropical forests; III montane subtropical forests; V sub-alpine forests, and VI alpine forests" classified by Champion & Seth (1968)), while its lowest unit in the hierarchy is a 'sub-type', e.g., "12C 1 /1a Ban oak forest" (Here, '12' signifies "12 Himalayan moist temperate forest" in a group of three [the other two being 11 Montane wet temperate forests & 13 Himalayan dry temperate forests). Then further sub-division of this sub-group "12" into three groups: C 1 -C 3 , where "C 1 " signifies " C 1 lower western Himalayan temperate forest " (other 2 being "C 2 upper west Himalayan temperate forest" and "C 3 east Himalayan temperate forest") and lastly its last subdivision which is depicted as "1a", i.e., "1a Ban oak forest (Q. incana)" (Quercus incana = Q. leucotrichophora) amongst the set of two (the other being "1b Moru oak forest (Q. dialata)" (Quercus dilatata = Q. floribunda) (Champion & Seth 1968)]. In this way, different 'forest subtypes' have been classified and labelled in India.
However, it is not known if the species composition and community structure of lower groups of animals such as butterflies are also different within each 'forest-subtype' or each have a unique community of butterflies. If this is the case then each forest sub-type harbouring unique and rare species can be taken as a unit of conservation on a sub-regional scale (western Himalaya) or state level (Uttarakhand). In this study we tried to evaluate and examine potential 'forest sub-types' or 'a group of forest sub-types' that have unique butterfly diversity which can be taken up as units of conservation of biodiversity at the state level. Besides, this can also be helpful in identification of new conservation areas with forest habitats outside the PA network and thus fill gaps in their connectivity, in the state. The rationale behind this is that many butterfly species are restricted to forested habitats in the state, have geographical distribution spread across the Himalayan region, i.e., western, central, and eastern Himalaya along a wide altitudinal gradient, e.g., Pale Green Sailer Neptis zaida zaida Doubleday, [1848] or Broad-banded Sailer, N. sankara sankara (Kollar, [1844]) (Nymphalidae) both occur in the state between 800-2,500 m, as observed in the present study. Fragmentation of their forested habitats on a larger spatial and temporal scale, may lead to isolated populations, local extinctions that can significantly affect their distribution, as they do not migrate. Thus, gaps and connectivity of the protected areas needs to be maintained for long term conservation.

STUDY AREA
The study was carried out in Uttarakhand state of India which covers an area of 53,483 km 2 , which is 1.63% of the geographical area of the country, and lies between 28. 716-31.466 N latitude & 77.566-81.05 E longitude. This predominantly mountainous state, shares its borders with Himachal Pradesh to the west and Uttar Pradesh to the south. It also shares international borders with Nepal in the east and China (Tibet) to the north. The state is mainly representative of the western Himalaya, the climate and vegetation vary greatly with altitude, from glaciers at the highest elevations, and temperate to subtropical at the lower elevations. Nanda Devi peak is the highest point at 7,816 m in the state while the lowest areas at ~100m lie in the Terai grasslands.

J TT
The average annual rainfall is 1,500 mm and the annual temperature varies from below 0 o C to 43 o C. Major rivers, Ganga, Yamuna, Ramganga, & Sharda, drain the state along with their tributaries. The Himalayan range in Uttarakhand is divided into the distinct non-montane and montane physiographic zones. The lower zone comprises the 'Bhabhar' region in non-montane lowland woodlands having Gangetic moist deciduous forests and the Terai region (below 500 m) running parallel to it, which comprises mainly the marshes and grasslands (Botanical Survey of India 2021). The montane region is divided into sub-Himalaya, which consists of the Shiwalik ranges, the lower Himalayan ranges, and the Doon (flat long valleys) lying north of the Shiwaliks (~ 500-1,000 m). Above this region are the lesser Himalaya (~ 1,000-3,000 m) followed mid Himalaya (~ 3,000-4,000 m) and then greater Himalaya (~ 4,000-6,000 m) (Khanduri et al. 2013) and the trans-Himalaya (above 5,000 m), also known as the Tethys Himalayas and the Indo-Tibet plateau, the region is in the rain shadow area that transforms into the cold desert.
Forests cover an area of 24,303.04 km 2 in the state, which constitutes 45.44% of the state's geographical area (FSI 2019). The state is represented by biogeographic zone 2B western Himalaya and 7B Shiwaliks of India (Rodgers & Pawar 1988 (Nayar & Sastry 1987, 1988, 1990; https://indiabiodiversity.org/). The PA network cover 12 percent of the total geographical area of the state, which includes six national parks, seven wildlife sanctuaries, four conservation reserves, and one biosphere reserve (Appendix 1).

Previous studies on butterflies in Uttarakhand
Studies on natural history and checklists of different areas in Uttarakhand state have been carried out as early as 1886 (Doherty 1886;Mackinnon & de Nicéville 1899;Hannyngton 1910-11;Ollenbach 1930;Shull 1958Shull , 1962Baindur 1993;Smetacek 2002Smetacek , 2004Bhardwaj et al. 2012;Bhardwaj & Uniyal 2013; Bhandari 2003Bhandari , 2006Singh & Sondhi 2016;Verma & Arya 2018;Sondhi & Kunte 2018; and the total number of butterfly species recorded in the state so far is ~ 500 species, based on these records. However, none of these studies give an account on the association of butterfly species with different forest sub-types as classified by Champion & Seth (1968), found across the state of Uttarakhand. The author had earlier studied butterfly-forest type associations in 11 major "forest sub-types" in the state of Arunachal Pradesh (eastern Himalaya), India (Singh 2017) and identified four forest sub-types: 2B/1S1 sub-Himalayan light alluvial plains semi-evergreen forests; 2B/C1a Assam alluvial plains semi-evergreen forests; 2B/2S2 eastern alluvial secondary semi-evergreen forests, and 3/1S2 b Terminalia-Duabanga as major forest sub-types supporting 415 butterfly taxa along with many rare and endemic species in the northeastern region and eastern Himalaya, but the forest sub-types occurring in these two Himalayan states are totally different from each other.

METHODS
Random sampling surveys were carried out for eight years under two different projects (2006-2009 and 2017-2020, respectively) across 11 districts of Uttarakhand state covering all the six butterfly seasons (spring, summer, pre-monsoon, monsoon, postmonsoon, autumn, and winter;Smith 1989) of the year. Surveys were carried out using 'Pollard Walk' on the line transects (Pollard & Yates 1993). Sampling on each transect (ca. 1 km) was done and butterflies were observed up to 20 m on both the sides of the trail for 1 h in a stretch between 1000 h and 1600 h to collect data on individual butterfly species abundance. Each sampling survey was carried out by the author, while 1-2 helpers were also used for recording data, collection of insect and plant material from time to time. Coordinates of all the locations for 307 samplings carried out were recorded using a GPS (Etrex Garmin Vista) (Figure 1) covering 20 major forest sub-types (FSI 2011; Figure 2 & Appendix ii) existing across the state of Uttarakhand.
Dominant vegetation (mainly trees & shrubs) in the respective forest sub-types were also identified and confirmed by ground truthing by laying down 10 x 10 m quadrates, collected plant material and preparing herbariums. Photographs and herbarium specimens were identified in the field and many were identified and confirmed from plant taxonomists based at Systematic Botany Branch, Botany Division, FRI, Dehradun and literature (Brandis 1906;Rai et al. 2017; http://www. gbif.org).

Evaluating species of conservation priority: rarity analysis of butterflies
The degree of "rarity' characterizing a species is usually an indicator of extinction risk (Rabnowitz et al. 1986;Pimm et al. 1988;Arita et al. 1990;Primarck 1993;Gaston 1994;Brown 1995;Gaston & Blackburn 1995) and provides a basis to identify threatened species (Rabinowitz 1981;Arita et al. 1990;Daniels et al. 1991;Berg & Tjernberg 1996). In general, species characterized by small geographic range, habitat specialization, and low abundance, are at higher risk of extinction than a widely distributed, habitat generalist and with high abundance. Rabinowitz et al. (1986) have examined types of rarity, and in what important ways rare species differ from one another. They first distinguish three traits, characteristic of all taxa recorded: (i) Geographical range -whether a species occurs over a broad area or whether it is endemic to a particular area; (ii) Habitat specificity -the degree to which a species occurs in a variety of biotopes' or 'habitats' is restricted to one or a few specialized sites versus generalists; and (iii) Local population size -whether a species occurs in large populations somewhere within range or has small populations whenever it is found.
In the present study, Rabinowitz et al. (1986) classification of rarity based on the three above traits was used. Only those species were filtered out the total as rare which had: (i) narrow geographical range, i.e., those species which had narrow distribution restricted only to western and central Himalaya as against those with wide distribution, i.e., Himalaya, northeastern India, & Peninsular India; (ii) restricted to two or less forest sub-types as against more than two forest sub-Hierarchical clustering of different forest sub-types based on butterfly species distribution and relative abundance.
The data of relative abundance of all the species of butterflies sampled against 20 different forest subtypes was pooled and averaged to relative abundance per sampling in each of the forest sub-type to remove varied sampling bias and was done using statistical software "NCSS Data Analysis 2021, v21.0.2", to know the dissimilarly of forest sub-types in terms of butterfly species composition.
The relative abundance of species ranged 1-1,596 individuals. These species were then ranked into four abundance classes based on their quartile division, i.e.,  (Table 4 and an "Appendix iii" with an account of 370 taxa). Sixty-seven species sampled are listed under various schedules of the Indian Wildlife Protection Act, 1972 (appendix: Schedule I-8 species; Schedule II-51 & Schedule IV-8). The seasonality of butterflies suggests that most of the species are in flight during 'post-monsoon' and 'pre-monsoon' seasons followed by 'monsoon' season, respectively when more than 270 species are in flight (Figure 3) in the state. The pattern of seasonality in Uttarakhand is very similar to the trend found in western and central Himalaya (Wynter-Blyth 1957) where two peaks are known to occur in a year, the bigger one during the 'post-monsoon' season and a slightly smaller one during the 'pre-monsoon' season.

Preference for Forest Sub-types
The highest number of species were recorded in 12/ c1a Ban Oak Forest (292 species; Fig.4) followed by 3C/ C2a Moist Shiwalik Sal Forest (220) and 12/C2c Moist Temperate Deciduous Forest (165), respectively which suggests that these forest sub-types hold the major diversity of butterflies found in the state. The number of species sampled were the least in 13/C2b Dry Deodar Forest (14), 15/C1 Birch Rhododendron Scrub (6) and 15/E1 Dwarf Rhododendron Scrub (2), respectively ( Figure 4) suggesting them to be poor butterfly habitats, while the other 14 forest sub-types lay between them.
The percentage of butterfly species in each forest sub-type in relation to the proportional area covered by each in the state ( Figure 5), suggests that forest subtypes: 9/C1b Upper or Himalayan Chir Pine Forest; 12/ C2b West Himalayan Upper Oak/Fir Forest and 14/C1 B Western Himalayan Sub-alpine Birch/Fir Forest, support a relatively lower number of butterfly species per unit area as compared to the rest of the other forest subtypes (Figure 4). On the other hand forest sub-types: 3C/C2 Moist Shiwalik Sal Forest; 12/C1a Ban Oak Forest; 12/C2C Moist Temperate Deciduous Forest and 12/ C1d Western Mixed Coniferous Forest have a relatively higher density of butterfly species per unit area amongst all the forest sub-types covered ( Figure 5). The primary reason for this is that pure conifer forest stands support less diversity of butterflies as compared to the pure broad leaved or mixed conifer-broad leaved forests, as the diversity of nectar and larval food plants available are more diverse in the latter two than in the former.

Hierarchical clustering of forest sub-types
It was found that 7 forest-types butterfly clusters, 5 independent forest-subtypes and 2 clusters of 2 and 11 forest sub-types, respectively exist in the state (Fig.6) (Figure 6). Four forest sub-types that are most important in the state in terms of number of both butterfly species and with distinct dissimilarity of butterflies are 3C/C2a Moist Shiwalik Sal Forest; 12/C2c Moist Temperate Deciduous Forest; 12/C1a Ban Oak Forest and 3C/C2c Moist Terai Sal Forest.

Species preference of forest sub-types
Scatter plot (Figure 7) of individual butterfly species (n= 370) suggests that only one generalist species (Painted Lady Vanessa cardui) had preference for all 14 forest sub-types. While the number of species showing preference for more than five or more forest sub-types were fewer as compared to species showing preference for less than four forest sub-types (Figure 7 Horizontal bars) in the state. The maximum number of species showed preference for two forest sub-types (n= 90 species) followed by preference for only one forest subtype (n= 60 species). This suggests that a large number of habitat specialist species exist in the state.

Rarity in butterflies sampled in Uttarakhand: taxa of conservation priority
Out of the 370 taxa sampled in Uttarakhand, 58 were evaluated as rare species of conservation priority /concern based on rarity analysis (Rabinowitz 1981;Rabinowitz et al. 1986) (Appendix IV).
The 58 taxa of conservation concern evaluated based on rarity are scattered all across the state in at least 12 forest sub-types (Figure 8). It was also determined that most of the butterfly taxa of conservation priority occur in 12/C1a Ban Oak Forest followed by 12/C2c Moist Temperate Deciduous forest, 3C/C2 Moist Shiwalik Sal Forest and a few taxa in 12/C2b Western Himlayan Upper Oak/Fir Forest; 12/C1d Western Mixed Coniferous Forest, repectively (Figures 8-15).
The present study proved that individual 'forest sub-types'(Champion & Seth 1968) or a group of 'forest sub-types' having high species richness, unique and rare butterfly taxa can be taken up as units of conservation at the state level in the Himalayan region as representatives of lower groups of animals, i.e., butterflies. Three most important forest sub-types: 12/C1a Ban Oak Forest followed by 12/C2c Moist Temperate Deciduous Forest and 3C/C2 Moist Shiwalik Sal Forest, respectively, hold the maximum number of butterflies, including many rare and protected taxa, in the state amongst the 20 forest sub-types evaluated, thus they form priority over the rest.
The 58 butterfly taxa conservation priority in the state lies both within and outside the PA network, but mainly in forested areas ( Figure 16). Concentrations

J TT
of 58 species of conservation priority are marked in 17 circles ( Figure 16) and at least 12 of these occur outside the PA network based on the findings of the present study. Important forest sub-types identified falling in these clusters having species of conservation concern can thus be recommended for conservation or future PAs. Seventeen concentrations/clusters that are located in different physiographic zones represented in the state are, three in Trans Himalaya; three in Greater Himalaya; eight in Lesser Himalaya; one in Shiwalik/Dun; one in Bhabar; and one in Tarai area along an elevation gradient, rather than a few as currently represented in the PA network of the state (Figure 17 & Appendix V). Also, new conservation sites can be identified from these 17 clusters/concentrations of rare buttefly taxa especially in the 'Lesser Himalaya' where the number of PAs are almost negligible. This type of approach in identifying areas of conservation priority is more inclusive and suitable at a sub-regional or state level in restoring linkages and corridors in the PA network, rather than solely based on a broader geographic scale, i.e., zoogeographic zones. Many of these sites with high butterfly richenss that lie outside the PAs and close to the villages and towns with suitable logistical support for boarding, lodging and travel can be promoted for suitainable and inclusive butterfly ecotourism activities in the state.