Does the size of the butterfly enhance detection? Factors influencing butterfly detection in species inventory surveys

Butterfly species’ abundance and factors influencing butterfly detection in Chimmony Wildlife Sanctuary, Kerala was studied from April to June 2018. The survey was carried out on 15 tracks of 2-km lengths surveyed two times resulting in the sampling effort of 60km. A total of 141 species of butterflies belonging to two orders, six families and 103 genera were observed during the study, of which 15 species were recorded as endemic. The majority of butterfly species belonged to the families Nymphalidae and Lycanidae. The size of butterflies varies significantly among families with the largest butterflies recorded in Papilionidae and Nymphalidae and the smallest butterflies from Hesperidae and Lycanidae. The factors that determine butterfly detection during the count was determined using multiple regression. The number of detections had a linear relation with abundance, size, and activities of the butterflies. The model was highly significant and explained 86.9% of the variation in the detection of butterflies (F=407.8; df=3; p<0.000). Abundance had a primary influence on detection followed by the size and activities of the butterflies. Further studies on relative detectability of different species of butterflies in the diversity and abundance estimation would help in refining methods of assessment of butterflies.


INTRODUCTION
Butterflies are universally popular among all fauna. They are very beautiful and come in various sizes, shapes, and colours. Different patterns on their body enhance their aesthetic value (Gupta & Majumdar 2012). The Western Ghats can be classified into three biogeographical parts based on the status and distribution of butterflies. They are the southern Western Ghats, central Western Ghats and the northern Western Ghats (Gaonkar 1996). Because of high levels of species endemism, the Western Ghats is listed under 34 global biodiversity hotspots. The region is prominent among all other biodiversity hotspots (Myers et al. 2000). The butterfly fauna of the Western Ghats consists of 346 species of butterflies under six families (Bhakre & Ogle 2018).
Most of the inventory surveys were carried out by sampling through forest paths and trails without any information on the sample area (Sudheendrakumar et al. 2000;Sreekumar & Balakrishnan 2001;Aneesh et al. 2013), hence it was not possible to estimate population density. The systematic surveys using fixed width transect or using pollard walk (Isaac et al. 2011) helps to estimate the population density of butterflies with the same sampling effort by recording additional information on length and width of the area sampled. It is essential to determine the different factors that determine the detection probability. Species-wise differences in the detection probability of butterflies were reported in the studies carried out in the United Kingdom (Isaac et al. 2011).
The family Nymphalidae is the most dominant family with a high number of species. A detailed diversity study of butterflies in Chimmony Wildlife Sanctuary (CWS) has not been done yet. Previous studies reported 24 species of butterflies in the study area (George 2012). We have investigated butterfly species size and abundance influence on the detection of butterflies in inventory surveys at CWS.

Study area
The study was conducted in Chimmony Wildlife Sanctuary, which spreads geographically within 76. 417N and 10.402E and 76.560N and 10.483E in Thrissur District of Kerala State (George 2012). The sanctuary was established in the year 1984. The sanctuary consists of parts of Kodassery Reserve with an extent of 85.07km 2 .
It is bounded by Nelliampathy Reserve Forest on the east, Peechi-Vazhani Wildlife Sanctuary on the northwest, and Sholayar Reserve Forest on the south (Fig. 1). The mean annual rainfall is 3,130mm. The sanctuary has a tropical humid climate, with three distinct seasons, dry season (December-March) followed by the south-west monsoon (April-July), and north-east monsoon (August-November). Temperature varies from 38.5°C to 15.6°C during different seasons. The minimum temperature falls below 15.6°C during December. The area is also vulnerable to forest fires during the dry season. The sanctuary has more than 250 streams and six man-made waterholes. Diverse vegetation and favourable climatic conditions in the sanctuary could support many species of butterflies.

Butterfly abundance estimation
Butterfly species abundance was estimated using fixed-width transect method in CWS from April 2018 to August 2018. Totally, 15 strip transects of 2km were selected along paths with 2-m width on either side of the transect and sampled twice that resulted in the sampling effort of 60km. The surveys were conducted between 09.30h and 13.30h when the butterflies were most active. The butterflies observed in the field were photographed for further clarification and identification. Butterflies were identified using field guides (Kunte 2006;Palot 2015;Kehimkar 2016;Bhakre & Ogale 2018) and specialists were consulted in case of uncertainty in the identification of species. The butterflies were photographed using a Nikon 3100 DSLR camera with 18-50mm and 70-300 mm lens. The butterfly survey routes were marked with GPS ( Fig.1).
Statistical analysis was performed by using Windowsbased statistical package Microsoft Excel, PAST (Hammer et al. 2001) and SPSS. The diversity indices such as Simpson and Shannon-Wiener index of butterfly species from each habitat were analysed with the help of software PAST. Butterfly size difference among different families was tested using one-way analysis of variance (one-way ANOVA). The factors that determine the detection of butterflies, such as abundance, activities (0-resting; 1-foraging, flying, mud puddling, etc), size of butterflies were tested using multiple regression. Both response and independent variables were log-transformed due to positive skewness of data. Linearity was examined by plotting the relationship between the response variable (number of detections) and each predictor variable (abundance and size) using Lowess plot. To investigate multicollinearity between the environmental covariates, a correlation analysis was conducted before using J TT multiple regressions to assess the relationships between the response variable and predictor variables, thereby providing valid parameter estimates and p values. The data were analyzed using SPSS Statistics 21 (IBM SPSS Inc., Chicago, Illinois, USA).

RESULTS
Totally, 141 butterfly species were documented in CWS from April to June 2020. Butterfly species composition varied among different families, with Nymphalidae and Lycanidae constituting 62%. Families such as Hesperidae, Papilionidae, and Pieridae were constituted 16.3%, 12.8%, and 8.5%, respectively. Only one species (Double-banded Judy) was recorded in the family of Riodinidae. Thus there is significant variation in the number of species recorded among different families (X 2 =67.3; df=5; p<0.01). The majority of butterfly species belong to Nymphalidae and Lycanidae in Chimmony Wildlife Sanctuary.
In total, 15 species are found to be endemic to the Western Ghats region (Table 1). Butterfly species such as Indian Ace, Shiva Sunbeam, Blue Oakleaf, Danaid Eggfly, Gladeye Bushbrown, Malabar Tree Nymph, Tailed Palmfly, Tamil Catseye, and Southern Birdwing are endemic species (Images 1-45). There are four species of butterflies such as Orchid Tit, Malabar Banded Swallowtail, Crimson Rose, and Danaid Eggfly listed in the Schedule I of the Indian Wildlife Protection Act (1972). In total there are 20 species of butterflies that are catalogued in the Schedules of IWPA and provide protection to the butterflies. Common Lineblue is the most abundant butterfly followed by Common Crow and Common Emigrant in CWS. There were more than 100 individuals of all these butterflies that were recorded in the study area. There were 42 species that were recorded only once during the time of the survey.

Factors that determine detection of butterflies
The size of butterflies varies among families with the largest sized butterflies recorded from Papilionidae and Nymphalidae (102.8±23mm and 70.1±20.1mm). Hesperidae (37.5mm) and Lycanidae (30.6mm) are the smallest-sized butterflies. Pieridae and Riodinidae are the medium-sized butterflies (57.7mm and 45mm, respectively). There is a significant difference in the size of butterflies among different families (F= 118.20; df= 5; p< 0.001).
The relationship between the number of detection, abundance, and size of butterflies were tested using multiple regression. The number of detection had linear relation with abundance, size, and activities of the butterflies. The model was highly significant and explained 86.9% variation in the detection of butterflies (F= 407.76; df= 3; p< 0.00; Table 2). All the three predictors had positive abundance and size positively influenced number of detections. From the standardized   SEM-Standard error of mean | SPRC-Standardized Partial Regression Coefficient partial regression, it was inferred that abundance (b 1 = 0.74) had the primary influence on the detections, followed by size (b 2 = 0.19), and activity of the butterflies (b 3 = 0.02; Fig. 2).

DISCUSSION
Composition of butterflies varied among different families. A total of 141 species of 1,986 individuals were observed from CWS. Though the study was carried out in a limited period, the number of species reported was higher than earlier reports of the study area (George 2012). The number of species recorded in the study area was more than other protected areas in Kerala; Sudheendrakumar et al. (2000) recorded 124 species at adjacent Parambikulam Tiger Reserve. A total of 71 species from Aralam WS (Sreekumar & Balakrishnan 2001) have been recorded. The results, however, are not directly comparable outside the protected areas. The number of species recorded in Kerala Agricultural University was 139 species of butterflies (Aneesh et al. 2013). The reason for comparison is the geographical proximity of KAU compass to the study area. The study area is part of the network of protected areas such as Peechi-Vazhani towards north, Sholayar Reserve Forest in the south and Parambikulam Tiger Reserve in the east. The major habitat of the study area is evergreen and moist deciduous forest. Earlier studies recorded higher species diversity and richness in the similar habitats (Sudheendrakumar et al. 2000). Thus, the contiguous forest and evergreen habitat supports higher species diversity and endemism in the study area.
Family Nymphalidae and Lycanidae represented 62% of the total. Families such as Hesperidae, Papilionidae, and Pieridae were comparatively less. They are, 16.3%, J TT 12.8%, and 8.5%, respectively. Out of two butterflies in the family Riodinidae of Kerala and Western Ghats, one species (Double-banded Judy) was recorded from the study area during the period of study. There is a significant variation in the species composition among different families. Family Nymphalidae dominated over other families. In almost all the studies conducted in butterflies of Western Ghats (Sudheendrakumar et al. 2000;Sreekumar & Balakrishnan 2001;Aneesh et al. 2013) Nymphalidae is the family showing the maximum number of species because this is the family representing more number of species in the Western Ghats. The study area harbours 40.7% of butterfly species of Western Ghats (Bhakre & Ogle 2018).
In total there are 20 species of butterflies that are listed in various schedules of Indian Wildlife Protection Act (1972) that provide protection to these butterflies. Only 14.2% of butterflies of recorded species are protected under IWPA. Hence it is important to include all the endemic species in the IWPA and butterflies which are more charismatic, and rapidly declining species need to be listed under the schedules. Common Lineblue is the most abundant butterfly followed by Common Crow and Common Emigrant in CWS. The other species such as Common Mormon, Chocolate Pansy, Narrowbanded Blue Bottle, Blue Mormon, Tailless Lineblue, Three-spot Grass Yellow, and Great Orange Tip were recorded. Similar species composition was recorded in Parambikulam TR (Sudheendrakumar et al. 2000) and Aralam WS (Sreekumar & Balakrishnan 2001).

Factors that determine detection of butterflies
The study highlights the differences in the species detection based on size and abundance and importance of differences in detection probability of butterfly species inventory surveys. Butterfly species such as Common Lineblue, Common Crow, Common Emigrant, Common Mormon, Three-spot Grass Yellow, Narrow-banded Bluebottle, and Blue Mormon were more frequently sighted. All these species are conspicuous, larger in size, active flyers, and some species show mud-puddling behaviour as well. This could have resulted in higher abundance and detectability. Studies on butterflies have shown that detection of same species tends to vary according to habitats (Pellet et al. 2012). Further, survey technique could also influence the abundance and density estimation. Thus our preliminary examination on butterfly detectability showed the influence of size, abundance, and activities. The number of detection had a direct relation with the abundance, size, and activities of the butterflies.

J TT
The model was highly significant and explained 86.9% variation in the detection of butterflies. Both abundance and size positively influenced the number of detections. From the standardized partial regression, abundance (b 1 = 0.74) had the primary influence on the detection of butterflies, followed by size (b 2 = 0.19) and activity (b 3 = 0.02). Similar species-wise differences in the detection of butterflies were reported in the studies carried out in the United Kingdom (Isaac et al. 2011;Pellet et al. 2012). Further investigation on the detectability of butterflies based on size, colouration, and habitats will help to estimate population size rather than species abundance.