Roosting Site Usage, Gregarious Roosting and Behavioral Interactions During Roost-assembly of Two Lycaenidae Butterflies

© 2020 Academia Sinica, Taiwan Open Access Roosting Site Usage, Gregarious Roosting and Behavioral Interactions During Roost-assembly of Two Lycaenidae Butterflies Yuan-Mou Chang1,*, Kent A. Hatch2, Mei-Yi Ho1, Stephen H. Roxburgh3, Yi-Ting Wu1, Yi-Kuang Wang1, Shuang-Ru Wang1, and Zi-Xuan You1 1Department of Ecology and Environmental Resources, National University of Tainan, 33 Su-Lin Street, Section 2, Tainan 700, Taiwan. *Correspondence: E-mail: [email protected] (Chang) E-mail: [email protected] (Ho); [email protected] (Wu); [email protected] (YK Wang); v.w[email protected] (SR Wang); [email protected] (You) 2Department of Biology, Long Island University – Post, Brookville, NY, USA. E-mail: [email protected] (Hatch) 3CSIRO Land & Water, GPO Box 1700, Canberra ACT 2601 / Clunies Ross Street Acton ACT 2601. E-mail: [email protected] (Roxburgh) Received 15 July 2019 / Accepted 9 March 2020 / Published 10 April 2020 Communicated by Shen-Horn Yen Lycaenidae is one of the larger of the world’s butterfly families, based on number and diversity of species, but knowledge of roosting in this group is sparse. Zizina otis riukuensis and Zizeeria maha okinawana are two small lycaenids that are commonly found in urban settings and widely distributed across much of Asia. We conducted experiments on a university campus to determine the plant species and plant structures commonly used by these two blues when roosting. We also tested the hypothesis that gregarious roosting exists in these two blues by demonstrating the non-random distribution of roosting blues and the tight mapping of their roosts to the spatial distribution of specific plant species and/or specific plant structures, as well as by demonstrating behavioral interactions among individuals during roosting-assembly. We found that both Z. otis and Z. maha roosted primarily on flowers and fruits of Tridax procumbens and Vernonia cinerea. We also found that these blues formed conspicuous roosting aggregations with significant positive associations between the flowers and fruits of both T. procumbens and V. cinerea and the blues. Moreover, our behavioral observations showed that these blues expressed various levels of interaction during roosting gatherings. Based on these findings, we conclude that gregarious roosting exists in both Z. otis and Z. maha. To our knowledge, this paper represents one of the first demonstration of nocturnal gregarious roosting in lycaenids. This study also highlights the importance of institutional estates in providing roosting resources for butterflies in urban ecosystems. Key words: Nocturnal roosting, Zizina otis riukuensis, Zizeeria maha, Roosting aggregation, Lycaenidae, Institutional estates, Conservation biology. Citation: Chang YM, Hatch KA, Ho MY, Roxburgh SH, Wu YT, Wang YK, Wang SR, You ZX. 2020. Roosting site usage, gregarious roosting and behavioral interactions during roost-assembly of two Lycaenidae butterflies. Zool Stud 59:10. doi:10.6620/ZS.2020.59-10. BACKGROUND Over the past few decades, the conservation of biodiversity has become an important issue at both local and global scales, with losses in biodiversity continuing unabated despite international efforts, such as the “Aichi Targets” of the Convention on Biological Diversity, to reduce declines in biodiversity (Butchart et al. 2010; Tittensor et al. 2014). Loss of biodiversity in tropical ecosystems is particularly concerning, since species diversity and abundance are particularly high in these ecosystems, as are the social and economic pressures impacting species abundance and diversity (Collen et al. 2008). For example, butterfly species of South East Asia are under particular threat due to extensive deforestation and the rapid rate at which it is proceeding (Koh Zoological Studies 59:10 (2020) doi:10.6620/ZS.2020.59-10 1 © 2020 Academia Sinica, Taiwan 2007). Habitat degradation or loss through agricultural conversion of land and urbanization are the main factors that drive species to extinction. In Taiwan, the location of this study, much of the original broad-leaf forest below 500 m elevation has been converted into either urban or agricultural land (Li et al. 2013). However, urban or agricultural habitats are not devoid of biodiversity, and are affected by the same factors affecting ecosystems elsewhere such as climate, substrate, resident organisms and topography (Pickett et al. 2011). While there has been much focus on how deforestation causes loss of species, much less is known about how species persist in transformed landscapes (Jain et al. 2017). A better understanding of their behavior and natural history is basic to understanding how certain species persist in agricultural and especially urban landscapes (also see Wang and Hung 2019). When considering critical butterfly habitats, much attention is paid to larval host plants and habitats that provide food and other resources needed for breeding (Gilbert and Singer 1975). However, other critical habitats are also important to the survival of healthy adult butterfly populations, such as those needed for roosting and mating (Dennis 2004), or providing protein-rich food (i.e., pollen) for adults (Gilbert 1972; Mallet 1986). Adult butterflies are typically active during daylight hours, when ambient conditions are suitable for flight, but are inactive and roost during evening hours. Nocturnal roosting in butterflies varies widely in characteristic roosting locations and behaviors (Table 1). Individuals of most species roost alone, solitarily remaining and sleeping at night in the location where they find themselves in the late afternoon. However, some butterfly species form roosting aggregations in which individuals gather at a specific location to pass the night (Davis et al. 2012; Mallet 1986; Young and Thomason 1975). Roosting in groups is predominantly found in unpalatable species of the subfamilies Acraeinae, Danainae, Heliconiinae and Ithomiinae (Benson and Emmel 1973; Finkbeiner 2014; Finkbeiner et al. 2012; Howard and Davis 2009; Mallet 1986; Mallet and Gilbert 1995; Owen and Chanter 1969; Salcedo 2011 2010a b; Turner 1975; Urquhart and Urquhart 1979) and some palatable species of the Nymphalini (Barrett and Burns 1951) and Hesperiidae (DeVries et al. 1987) butterflies. Although there are many examples of roosting in groups from a variety of butterfly groups, reports of roosting in groups from the diverse family Lycaenidae (the “blues” or “hairstreaks”) are mainly descriptive (e.g., Heath and Emmet 1985; Thomas 1983) and no detailed case study on roosting in groups has been done in any of these species as far as we are aware. Moreover, roosting substrates may vary from site to site and species to species. Many species roost under leaves, some on tree trunks and some in low, dense vegetation (e.g., grasses, rushes, bushes). A few tropical species roost in caves or under cliff overhangs (Benson and Emmel 1973; Davis et al. 2012; Dennis 2004 1986; DeVries et al. 1987; Hoskins 2018; James 2006; Opler and Malikul 1992; Rawlins and Lederhouse 1978; Young and Thomason 1975). Lycaenidae is one of the larger families of the world’s butterfly species, based on the numbers and diversity (> 6000 species, 30% of all butterfly species) (Pierce et al. 2002; Zhang et al. 2019). Considering the great diversity of lycaenids and the wide range of habitats in which they occur, knowledge of roosting in this group is sparse (Table 1). This study focuses on investigating the roosting behavior and habitats of the lesser grass blue, Zizina otis riukuensis (Matsumura, 1929), and the Japaneses pale grass blue, Zizeeria maha okinawana (Matsumura, 1929), by observating wild populations on a university campus. These two small butterflies belong to one of the largest subfamilies in Lycaenidae, the Polyommatinae (Eliot 1973). They are commonly found in urban settings and are widely distributed across much of Asia, preferring cultivated areas, abandoned lots, urban parks and gardens (Chen 2015; Chowdhury et al. 2017; Harinath et al. 2015; Lu and Chen 2014; Nidup 2016; Otaki et al. 2010; Sing et al. 2016; Tsang and Bonebrake 2017; Venkata Ramana et al. 2014; Yago et al. 2008). Given their urban setting, the main threats to these two blues is the spraying of pesticides and the physical clearance of weeds and grasses in their preferred habitats (Sing et al. 2016). The adults of these two species are colonyforming, and swarm in great abundance in mostly open grassy habitats such as forest clearings, weedy lots, riverbanks, roadsides, parks and gardens. They are extremely active and flutter very close to the ground. They feed avidly on the nectar of daisies and other low-growing flowers, including Boerhavia, Medicago, Tridax, Trifolium and Vernonia (Hoskins 2018; Li 2007; Yuan Mou Chang, personal observation). In diffuse sunlight they will bask on the ground with their wings half open (Hoskins 2018). Around dusk, both blues settle into roosts in the vegetation, solitarily or in groups (Yuan Mou Chang, personal observation). Despite the fact that the adults of these two blues are common and easy to find, their roosting locations remain largely unreported, the one exception being a report showing that adult Z. otis roosts overnight on low vegetation or on bushes (Hoskins 2018). Our study had two goals. The first was to determine the roosting sites of these two blues, including which plant species they use and where on these plants (flower, fruit, leaf, receptacle and stem/ page 2 of 15 Zoological Studies 59:10 (2020) © 2020 Academia Sinica, Taiwan peduncle) these butterflies roost. We predicted that these two blues roost primarily on specific plant species and plant structures (prediction 1). The second was to test the hypothesis that gregarious roosting exists in these two blues. Two additional predictions resulted from our hypothesis that Z. otis and Z. maha roost gregariously: that the roosting individuals are distributed nonrandomly and match the spatial distribution of specific plant species and/or specific plant structures within roosting aggregations (prediction 2), and since evidence of behavioral interactions among individuals in close proximity is evidence of gregarious roosting (Finkbeiner 2014 2019; Mallet 1986; Salcedo 2011), that behavioral interactions occur among individuals of these species during roost-assembly (prediction 3). MATERIALS AND METHODS Study site and a brief description of the experimental design The study was conducted in a garden of the Rongyu campus of the National University of Tainan (about 2 hectares; 22°58'N, 120°13'E. About 24 m a.s.l., Tainan, Taiwan) (Fig. 1). The garden was surrounded Table 1. Typical nocturnal roosting sites of different butterfly species Species Roosting substrate Sources Hesperiidae Celaenorrhinus fritzgaertneri in a small cave (DeVries et al. 1987) Pyrgus malvae at the top of dead flower-heads (Hoskins 2018) Erynnis tages at the top of dead flower-heads (Hoskins 2018) Nymphalidae Clossiana euphrosyne on bracken fronds or on the flowers of rushes (Hoskins 2018) Coenonympha pamphilus in a head-downwards posture at the top of grass heads (Hoskins 2018) Danaus plexippus gregariously on various types of trees or shrubs (with a general preference for maples and conifers, pecans and oaks) (Davis et al. 2012) Heliconius charitonia gregariously on leafless twigs of Anguria trees (Waller and Gilbert 1982) Heliconius erato gregariously on leafless fine twigs or tendrils of dead vines (Finkbeiner et al. 2012; Mallet 1986) Heliconius ethilla gregariously on leafless twigs (Turner 1975) Heliconius sara shaded areas with plenty of thin dry vines and branches under relatively dense vegetation mats (Salcedo 2010b) Lasiommata megera under leaves, under the lower boughs of trees or crevices in banks and walls, on fences (Dennis 1986) Manataria maculata in shaded embankments, tree holes, and other dark hiding places (Hanson 2000) Maniola jurtina in a head-downwards posture at the top of grass heads (Hoskins 2018) Marpesia berania on the underside of the leaves of rubiaceous trees or other small trees (Benson and Emmel 1973) Melanargia galathea in a head-downwards posture at the top of grass heads (Hoskins 2018) Smyrna karwinskii gregariously in cavities of lava walls, tree trunks and on the underside of concrete slabs roofing alleys between sheds (Muyshondt and Muyshondt 1974) Papilionidae Papilio polyxenes asterius primarily on inflorescences, or apices, but also on stems, scapes or culms of daisies, other herbs and grasses, roost singly (Rawlins and Lederhouse 1978) Pieridae Colias eurytherne singly or gregariously in dense grass, in dense crown vetch, on the leaves of emergent forbs (Clench 1970) Phoobis sennae eubule gregariously on the yellow-green leaves of vines (Clench 1970) Pieris rapae gregariously on leaves or stems of the upper branches of Pittosporurn undulatum Vent. Var. val’iegatum (McFarland 1971) Lycaenidae Lycaenidae icarus on the flower-heads and stems of grasses and other plants (Frohawk 1914) Lysandra bellargus in small groups on tall vegetation (Thomas 1983) Plebejus argus on shrubs (bramble, gorse) and tall herbs (rank bunched grasses such as Dactylis glomerata, bracken and flowering herbs) (Dennis 2004) Pseudophilotes sinaicus on the tips of dead stalks/dry flower heads of Jasonia montana (James 2006) Zizina otis on low vegetation or on bushes (Hoskins 2018) page 3 of 15Zoological Studies 59:10 (2020) © 2020 Academia Sinica, Taiwan by the concrete buildings and included lawn, groves, perennial flowering plants, herbs, shrubs and large trees. We chose the campus garden as the study site because the campus garden provides a typical habitat (grassy lawn) that these lycaenids like to use. The study site experiences a humid tropical savanna climate (Köppen climate classification) with hot summers (Temperature range: 26–33°C) and dry cool winters (Temperature range: 14–24°C) (1981–2010 records, Tainan city government, http://www.tainan.gov.tw/tainanE/). The study was done in two stages. The first stage gathered data to determine the roosting sites of these two blues and address prediction 1, focusing on investigating the plant species and plant structures for Fig. 1. (a) Map of Rongyu Campus, National University of Tainan. White squares indicate the plots for the night roosting surveys of butterflies from November 2015 to March 2016 (n = 10). Red squares (plots a–f) are subsites used for the random patterns test, confirming the non-random distribution of the roosting blues within roosting aggregations, as well as for observation of the social and behavioral interactions among individuals during roost-assembly. (b) A closer look of plots a, b, and c. (c) A closer look of plots d, e, and f. N page 4 of 15Zoological Studies 59:10 (2020) © 2020 Academia Sinica, Taiwan roosting and took place in November 2015–March 2016. The second stage tested the hypothesis that gregarious roosting occurs in these two blues, and this stage was carried out in November–December 2018. Following prediction 2, we sought to demonstrate gregarious roosting by showing a correlation between the spatial distribution of specific plant species and that of roosting blues (i.e., a non-random distribution) in roosting aggregations. Following prediction 3, we focused on the observation of behavioral interactions among individuals during roosting-assembly. Stage 1 Determination of plant species and structures of plants where roosting occurs (prediction 1) Based on our preliminary observations of the behavior and locations of these butterflies, we set up ten plots (each 3 × 3 m2 in area, delineated with red plastic strings) to record the plant species and plant structures on which roosting occurs. This part of the study was conducted from late fall to early spring (November 24th, 2015 to March 31st, 2016). During this period, the nearest weather station (Tainan, 2.3 km north-west of Rongyu campus) reported a mean daily air temperature of 20°C, with mean monthly temperatures ranging from 17.3 (January 2016) to 25.1°C (November 2015) (Central Weather Bureau, http://www.cwb.gov.tw/). The overall rainfall during the study period was 556.7 mm. We surveyed the above plots for four to five days each week during the study period. Each plot was surveyed systematically and carefully using a head lamp, beginning approximately 1 h after sunset time of 17:30–18:00, and lasting for 2–4 hours. Surveys were not conducted on days with heavy rain. Once a butterfly was located, we identified it to the species level, recorded the time, the species of plant it was on, the structure of the plant upon which the butterfly roosted, and the height above ground level of the head of the butterfly. Plant structure was recorded according to the following five categories: flower (i.e., the head inflorescence of the daisies Tridax procumbens, Vernonia cinerea, Emilia sonchifolia, Youngia japonica, Ixeris chinensis, Ageratum conyzoides; the globose heads of Mimosa pudica; the spike inflorescence of Kyllinga brevifolia and Kyllinga nemoralis; the raceme inflorescence of Axonopus compressus and Digitaria; the panicle inflorescence of Sporobolus indicus and Eragrostis amabilis; the umbel inflorescence of Hedyotis corymbosa and Fimbristylis dichotoma), fruit (the cypselas of the daisies, including ripe fruits with winged achenes ready for wind-dispersal and residual fruits after achene dispersal), leaf, receptacle (the part after the residual fruit has left), and stem/ peduncle. The observations were made with the aid of a magnifying glass with LED lights, and photos were taken using a digital camera when the individuals were difficult to identify. The identities and scientific names of the butterflies are in accordance with Lu and Chen (2014). Plants were photographed and/or collected and identified in consultation with Prof. Tsung-Hsin Hsieh, Department of Ecology and Environmental Resources, National University of Tainan. The scientific names of the plants follow the Flora of Taiwan (Huang 2000). Toward the end of the experiment, we took aerial photographs and recorded the vegetation composition of each plot. In each plot, plant species were identified and the percentage of ground they covered estimated. The lawn was not mowed during the study period to allow the grasses to grow and flower. Stage 2 1. Non-random spatial distribution of roosting blues and spatial distribution of associated plant species (prediction 2) To understand whether the spatial distributions of roosting blues are significantly associated with the spatial distribution of the flowers and fruits of T. procumbens and V. cinerea, we examined the spatial relationship between the roosting blues and the flowers and fruits of T. procumbens and V. cinerea on six plots (plots a–f; Fig. 1, Fig. 3) from November 14th to 16th, 2018, using a Canon EOS 5D Mark III with an EF 24–105 mm f/4L IS II USM Lens (Canon Inc. Tokyo, Japan) (Fig. 1). We started to search for roosting aggregations on the evening of November 3, 2018 and found aggregations in the garden. Based on the numbers of roosting blues in these aggregations, we selected the six aggregations with the most individuals (Fig. 1, plots a– f) to test the hypothesis that gregarious roosting exists in these two blues. Of these six aggregations, three (plots a–c) contained many flowers and fruits of T. procumbens, and blues were observed to roost on them. Plots d to f had many flowers and fruits of V. cinerea, and blues were observed roosting on them as well. We continued to observe roosting aggregations on these plots until November 13 to confirm that aggregations occurred each night at these plots during this period. We sampled the spatial patterns of T. procumbens and blues on the evening (about 9–10 pm) of November 14th, 2018 for plots a, b and c (Fig. 3a, 3b, 3c). Because these three plots were relatively large (about 1.5 m × 1.5 m per plot) compared to the size of the blues themselves (< 1 cm), it was difficult to identify each blue when page 5 of 15Zoological Studies 59:10 (2020) © 2020 Academia Sinica, Taiwan viewing a digital photo of the entire plot on the computer, due to lack of resolution. We thus developed a “thread marking method” to make roosting sites stand out in digital photos. By marking the roosting locations of each blue within a plot with colored threads, we were able to analyze the spatial patterns of T. procumbens and blues. This approach proved to be suitable and can be used to clearly mark the roosting locations. To accurately mark roosting locations used by blues, we first took a series of photos of each plot. These photos were taken from a squatting position, but from a variety of different angles to determine all of the locations the blues were roosting. This step was important because the act of attaching threads to plants where the butterflies were roosting typically caused the blues to fly away. Next, based on this series of photos, we tied a piece of thread to each structure on which blues had roosted. Because it was difficult to tie a thread on a flower and a fruit, we tied a thread on the “peduncle” right under a flower or a fruit where roosting occurred. We used red cotton threads for T. procumbens and yellow plastic threads for other plants where blues had roosted. On the morning (about 7–8 am) of November 15th, 2018, we took a photo of the roosting locations as indicated by the threads from the top of a ladder above each of the three plots (Fig. 3a, 3b and 3c) and used the locations of the two different color threads in a random patterns test (Roxburgh and Chesson 1998). For V. cinerea and blues, we did not use the “thread marking method,” but directly took photos from the side of plants and blues to sample the spatial patterns of V. cinerea and blues in the evening of November 16th, 2018. Veronia cinerea is an erect herb, and blues typically roost at different heights on the flowers/fruits that are situated on the top of the slender, grooved and ribbed stems (Figs. 2b, 3d, 3e and 3f). These plots were smaller (about 0.5 m × 0.5 m per plot) than those mentioned above, so that the entire plot fit within the photo and the photo could be taken at a distance that made it is easy to identify each blue’s location when viewing a digital photo of the entire plot on the computer. The spatial patterns of plants and blues in photos were subdivided into grid cells (Fig. 3) using Pages (Pages Version 7.2, Apple Inc, CA) for the random pattern test (see below) (Roxburgh and Chesson 1998). 2. Behavioral traits during roost-assembly Categorization of behavioral traits of blues during roost-assembly followed Mallet (1986) and Salcedo (2011). We identified four distinct behavioral traits that are consistently exhibited by flying individuals during roost-assembly, including (1) brief approach: direct flight to a roosted individual without physical contact, (2) hovering (or fanning): hovering above a roosted individual without physical contact, (3) stopping on the same perch: briefly stopping close to a roosting individual on the same perch (flower/fruit/ leave) without approaching or touching the individual, and (4) clutching: brief physical contacts between an approaching individual’s claws and the wings of a roosted individual. We recorded both approach and hovering behaviors only if they occurred within 15 cm of the roosted individual, and we recorded whether the approaching or hovering butterfly performed the behavior once or more than once. We also recorded the responses of the recipients involved: (1) no reaction, (2) movement or rotation, (3) fending off without dislodgement by vigorously flapping wings flapping without letting go of the perch, and (4) fending off and leaving: vigorously flapping the wings and then letting go of the perch. We also recorded the response of the approaching butterfly to the occurrence of fending off and taking off by the roosted individual: (1) both the approaching individual and the roosted individual leaving the perch at the same time, (2) the approaching individual usurping the perch but then flying off again to find another perch, and (3) the approaching individual usurping the perch without leaving. Behavioral interactions were observed and recorded with the aid of binoculars during the roostassembly in plots a-f in the afternoon hours (1510–1640) of November 28th [observation period (OP) = 34 min], 29th (OP = 49 min), and 30th (OP = 30 min), and Dec. 1st (OP = 39 min), 5th (OP = 71 min), and 7th (OP = 70 min), 2018. These blues began to congregate in the roost sites about one to two hours before sunset (YuanMou Chang, personal observation). During this period, the blues exhibited a set of roost-assembly behavioral interactions until they were all roosted gregariously (Salcedo 2011; Yuan Mou Chang, personal observation). We observed and recorded the interactions of butterflies from a distance of about 1.5 to 2 m from the roost sites. When an interaction was detected, the observers recorded the interaction according to the above categories. We were not able to record the individual butterflies to the species level because these two species are too small and similar to distinguish with naked eyes or binoculars from these distances. Statistics To test the prediction that these blues roost primarily on specific plant species and structures (prediction 1), we compiled the data from the 10 plots between November 2015 and March 2016 and used a chi-square test to compare the numbers of butterflies on page 6 of 15Zoological Studies 59:10 (2020) © 2020 Academia Sinica, Taiwan Fig. 2. (a) Zizina otis and Zizeeria maha roosting gregariously on Tridax procumbens. (b) Zizina otis and Z. maha roosting gregariously on Vernonia cinerea. (c) A Z. otis roosting on a flower of T. procumbens. (d) A Z. otis roosting on a flower of V. cinerea. (e) A Z. maha roosting on a flower of T. procumbens. (f) A Z. otis roosting on the fruits of T. procumbens. (g) A Z. otis roosting on a fruit of V. cinerea (left) and a Z. otis roosting on the receptacle of T. procumbens (right). (h) A Z. otis roosting on a leaf of V. cinerea. (i) Two Z. otis roosting on a leaf of Mimosa pudica. (j) Two Z. otis roosting on a leaf of Imperata cylindrica. (k) A Z. otis roosting on a peduncle of T. procumbens. The red dots in (a) and (b) are the roosting blues. (a) (b) (e) (h)(g)(f) (i) (j) (k) (c) (d) page 7 of 15Zoological Studies 59:10 (2020) © 2020 Academia Sinica, Taiwan each of the plant species and structures. Based on the results of standard contingency table chi-square test, which assumes randomness of blues from each plant species and structures, we inferred whether there is a preference for roosting site in terms of plant species and structures (Askew 1982; Ludwig and Reynolds 1988; Rouquette and Thompson 2007). The data used for the chi-square test may have the issue of pseudoreplication if the same butterfly individuals may have been counted multiple times at their roosting sites on various evenings. However, since these blues are too small to be marked, this so far is the best statistical analysis for understanding the answers of prediction 1. We also calculated the average roosting height according to the plant structure upon which the butterfly roosted. We used a random patterns test to test prediction 2, that roosting blues and the flowers and fruits of T. procumbens and V. cinerea will be spatially correlated, given the assumption that the butterflies exhibit patchiness in their distribution and using data collected in 2018 (Roxburgh and Chesson 1998). When organisms are patchily distributed (i.e., distributed nonrandomly in space), the existence of clumped patterns (the exhibition of positive spatial autocorrelation) violates the within-species spatial randomness as well as the independence between species assumptions of Fig. 3. The observed spatial patterns and the grid of the flowers and fruits of Tridax procumbens and blues at three subsites (a, b, c), and the spatial patterns and the grid processed of flowers and fruits of Vernonia cinerea and blues at three subsites (d, e, f). page 8 of 15Zoological Studies 59:10 (2020) © 2020 Academia Sinica, Taiwan traditional statistical tests for detecting interspecific associations, resulting in an elevated Type I error rate, i.e., an increase in the risk of concluding a test is statistically significant and therefore that species are associated, even when the species actually are not associated (Dale et al. 1991; Legendre 1993; Tavaré and Altham 1983). The random pattern test was developed to detect interspecific associations in which withinspecies patchiness is retained (Roxburgh and Chesson 1998). We compiled the number of observations of each behavioral trait collected between November 28 and December 7, 2018 to address the prediction that these blues exhibit behavioral interactions among individuals during roost-assembly (prediction 3). RESULTS Determination of plant species and structures of plants where roosting occurs (Prediction 1) A total of 1257 observations of Zizina otis and 266 observations of Z. maha were recorded. Z. otis and Z. maha roosted on 16 and 10 plant species, respectively, out of the 17 plant species present (Table 2). However, the main plant species on which these two blues roosted were T. procumbens and V. cinerea (Z. otis, χ2 = 13276.1, d.f. = 2, P < 0.001; Z. maha, χ2 = 2203.4, d.f. = 2, P < 0.001). The total percentage covered by these plants in all ten plots was only 5.9% and 1.7%, respectively. For Z. otis, 45% of individuals (n = 570) and 37% of individuals (n = 459) roosted on T. procumbens and V. cinerea, respectively. For Z. maha, 54% of individuals (n = 145) and 29% of individuals (n = 76) roosted on T. procumbens and V. cinerea, respectively (Table 2). The main plant structures used by both Z. otis and Z. maha for roosting were flowers and fruits (Fig. 2c, 2d, 2e, 2f), although receptacles (Fig. 2g), leaves (Fig. 2h, 2i, 2j), and peduncles (Fig. 2k) were also used (Z. otis, χ2 = 1916.6, d.f. = 4, P < 0.001; Z. maha, χ2 = 227.6, d.f. = 4, P < 0.001) (Table 3). The average roosting height above ground level for Z. otis and Z. maha was 20–28 cm, ranging from 8.9 cm to 60 cm (Table 3). Spatial distribution of roosting blues and plants within roosting aggregations (Prediction 2) We found that these blues formed conspicuous roosting aggregations, typically with one individual to a flower or fruit (Figs. 2a b, 3). The total numbers of roosting individuals we sampled was 35, 40, 30, 15, 11 and 13 in plots a, b, c, d, e and f, respectively (Fig. 3). The random pattern test demonstrated significant positive associations between the flowers and fruits of T. procumbens and blues, and between the flowers and fruits of V. cinerea and blues (Table 4). Table 2. The percentage cover of plant species and the number of Zizina otis and Zizeeria maha roosting on these species Percentage cover of plant species (%) No. of roosting Zizina otis (with percentage shown, %) No. of roosting Zizeeria maha (with percentage shown, %) Tridax procumbens 5.9 570 (45.3) 145 (54.4%) Vernonia cinerea 1.7 459 (36.5) 76 (28.6%) Other plant species 92.4 228 (18.1) 45 (16.9%) Emilia sonchifolia 3.5 59 (4.7) 25 (9.4) Mimosa pudica 6.4 61 (4.9) 2 (0.8) Fimbristylis dichotoma 0.3 29 (2.3) - Kyllinga brevifolia 14.4 22 (1.8) - Youngia japonica 4.7 21 (1.7) 5 (1.9) Axonopus compressus 27.2 8 (0.6) - Digitaria 7.7 7 (0.6) 1 (0.4) Ixeris chinensis 1.4 7 (0.6) 6 (2.3) Imperata cylindrica 6.8 6 (0.5) 1 (0.4) Eragrostis amabilis 11.6 3 (0.2) 3 (1.1) Sporobolus indicus 3.9 2 (0.2) - Kyllinga nemoralis 3.4 1 (0.1) 2 (0.8) Ageratum conyzoides 0.2 1 (0.1) - Hedyotis corymbosa 0.1 1 (0.1) - Conyza bonariensis 0.8 - - page 9 of 15Zoological Studies 59:10 (2020)