The katydid country: bioacoustics and ecology of tettigoniid communities from the Indian subcontinent

ABSTRACT The study provides the first description of acoustic communities from a subtropical rainforest in Northeast India and a previously unexplored site from the Western Ghats. We describe call structures of 15 katydid species from the Indian subcontinent. The habitat and seasonal variations of the communities were investigated at both sites. Most species observed produced ultrasonic broadband calls (bandwidth: 11.07–42.5 kHz); only two Pseudophyllinae members produced pure tone calls. The study includes calls of two previously undescribed species of the genus Ducetia (subfamily: Phaneropterinae) from the subcontinent. We observed diverse acoustic communities at both sites represented by five major subfamilies: Conocephalinae, Hexacentriane, Mecopodinae, Phaneropterinae and Pseudophyllinae. The acoustic communities at each site exhibited seasonal variations and habitat preferences. The post-monsoon season had a richer community composition than the dry season. We compared differences in the community compositions between habitats using a presence–absence matrix based on 526 individuals from the two sites. Species composition was found to be different among habitats (stress = .06, dimension: 2), with ANOSIM indicating separation of species among understorey, shrubland and grassland habitats.


Introduction
Katydids or bush crickets (sub-order Ensifera) represent one of the most ancient insect groups on planet earth (Song et al. 2015). Katydids can be found in various habitats, ranging from cloud forests to deserts (Bailey 1990;Korsunovskaya 2008). Katydids use acoustic signalling for long-range communication (Korsunovskaya 2008). The primary role of these signals is sexual, i.e. mate-attraction and courtship (Bailey 1990;Korsunovskaya 2008). The significance of acoustic signalling for inter-or intra-specific interactions and predator evasion are also widely studied (Kowalski et al. 2014;Morris et al. 2018).
Stridulation in katydids involves coordinated movements of specialised body parts. The forewings (tegmina) possess a series of microscopic cuticular pegs (file) on the left tegmen and a scelerized vein (plectrum) on the right tegmen (Stumpner et al. 2013). The sound signal is produced by the motion of flight muscles, such that the plectrum strikes against the file during the opening and closing movement of the wings. A membranous mirror present behind the plectrum amplifies and radiates the vibrations generated due to this movement (Montealegre et al. 2017). Due to its mechanical origin, the signal properties are highly uniform and show minimal inter-individual variation enabling conspecific recognition and inter-specific isolation (Walker 1964;Korsunovskaya 2008).
Descriptive studies based on acoustic signals have proven effective in delineating morphologically cryptic species (Walker et al. 2003;Hemp et al. 2015a;Iorgu et al. 2017Iorgu et al. , 2018Tan et al. 2020). The species-specific signals combined with katydids' sensitivity to the changes in their ecosystem make katydids an excellent indicator of ecosystem quality (Thompson 2017;Gasc et al. 2018;Fumy et al. 2020;Ter Hofstede et al. 2020) that can be studied in a non-invasive manner (Riede 1998;Diwakar et al. 2007). Despite this, there are only a few community-level studies of katydids available in the literature. Most katydid studies are either (a) taxonomic revision or description of taxa with notes of signals and occasional ecological observations (Walker et al. 2003; Montealegre-Z et al. 2011;Sarria-S et al. 2015;Hemp et al. 2015aHemp et al. , 2015bIorgu et al. 2017Iorgu et al. , 2018Tan et al. 2020) or (b) acoustic surveys of a community with observations on behaviour and occasional notes for a new taxon (Morris and Beier 1982;Belwood 1990; Morris et al. 1994;Heller 1995;Kowalski and Lakes-Harlan 2013;Stumpner et al. 2013;Heller et al. 2014Gasc et al. 2018;Ter Hofstede et al. 2020).
Studies outside the neotropics have found that katydid communities are fairly dynamic over the seasons (Ayal et al. 1999;Lehmann et al. 2014;Thompson 2017). Studies from subtropics and temperate regions also report the close association between different species and vegetation structures that could play a role in habitat segregation between sympatric congeners (Ayal et al. 1999;Grant and Samways 2016).
From the Indian subcontinent, records of acoustics and ecology of the ensiferan community are limited, considering the region has four biodiversity hotspots (Joppa et al. 2011;Pimm et al. 2014). Diwakar and Balakrishnan (2007a) documented the acoustic diversity of Kudremukh National Park in the Western Ghats. The study reported calls of nine katydid species along with an investigation of diel partitioning among sympatric callers. A study about the ensiferan community investigated the possibility of habitat partitioning in terms of microhabitat height (Diwakar and Balakrishnan 2007b) along with signal structure and sound intensity (Jain et al. 2014). Tiwari and Diwakar (2019) described the acoustic signals for six conehead katydid species from the two biodiversity hotspots.
Despite being recognised as one of the biodiversity hotspots due to a high degree of endemism in both flora and fauna, the biodiversity of the Northeast region remains unexplored to the same extent as that of the Western Ghats (Chitale et al. 2015). Thus, the aims of the present study were 1) to describe the acoustic diversity in the North-East Himalayan ecosystem and a previously unexplored site in the Western Ghats and 2) to investigate acoustic diversity of katydids in dry and post-monsoon seasons at the two sites. The study also investigated the differences in the acoustic community based on habitat.

Study site
The sampling was carried out in the protected areas of Hollongapar Wildlife Sanctuary in Assam and Bhagwan Mahavir Wildlife Sanctuary in Goa. At both locations, the sampling was conducted in the protected areas and the periphery. These peripheral areas consisted of grasslands and tea gardens in the case of the former and of grasslands in the latter. The two selected sites have a subtropical climate with pronounced wet post-monsoon and dry seasons.
The fieldwork was conducted from September 2015 to September 2020.

Acoustic diversity
Singing individuals were located acoustically, along transects in each habitat (Diwakar et al. 2007 Representative specimens were collected from outside the protected areas and preserved in 90% ethanol for taxonomic identification. All the calls and specimens are available in the Department of Environmental Studies, University of Delhi, for future reference.

Acoustic analysis
The best recordings in terms of S:N ratio per call-type/species were selected. The sound files selected for the analysis ranged from 30 to 240 s in duration. We performed spectral analysis of the recordings in SpectraPLUS-SC (2015, Version 5.1..33, Pioneer Hill Software, USA). We analysed calls from the ultrasound detector, and recordings from the low-frequency TASCAM recorder were only used to check for any low-frequency component. We analysed three spectral parameters: fundamental frequency, which is defined as the first peak in the spectrum (Baker and Chesmore 2020); peak frequency, which is the frequency with the highest amplitude (Baker and Chesmore 2020) and bandwidth (BW) was calculated as the frequency range 20 dB below the peak frequency (Diwakar and Balakrishnan 2007a;Kowalski and Lakes-Harlan 2013).
We used Raven Pro (2013, Version 1.5, Cornell Lab of Ornithology, Ithaca, NY) for temporal analysis. We followed the terminology in Baker and Chesmore (2020) for analyses and descriptions of katydid calls in our study. An echeme is the highest unit of call structure. An echeme may contain multiple syllables produced by incomplete tegmen movement. A repeated sequence of syllables or echeme constitutes a verse ( Figure 1).
We examined two parameters (duration and period) for both echeme and syllable. We defined 'duration' as the difference between the ending and beginning of a single syllable or echeme and 'period' as the difference between the beginnings of two consecutive song syllables or echemes. A syllable or echeme's 'repetition rate' was calculated as the reciprocal of its period. Thus, a total of six temporal parameters were measured viz., syllable duration (SyD), syllable period (SyP), syllable repetition rate (SyRR), echeme duration (ED), echeme period (EP), and echeme repetition rate (PRR). In addition to the six temporal parameters, we analysed an additional parameter verse duration (VD). In call-type/species that contained two or more different kinds of syllables, parameters for each syllables were analysed and reported separately.

Systematics
We identified the collected specimens using a stereo-zoom microscope (Olympus SZX7). The stridulatory apparatus of each specimen was photographed under the microscope (Olympus penlite EPL3). Body measurements were taken with digital Vernier callipers (Aerospace Digimatic, 0-150 mm; accuracy: ±.2 mm). As stridulatory files in katydids are curved, the teeth in the file become smaller and less discernible towards the edges. We followed the methodology of Anichini et al. (2017) for calculating teeth density for each individual by counting the teeth in the middle part of the file and dividing it by the length of the middle file region (Refer to Supplementary online material). Specimen identification followed the keys provided by Walker (1870), Brunner von Wattenwyl (1895), Ingrisch and Shishodia (2000), Barman (2003) Shishodia and Barman (2004), Ingrisch (2005), Nagar et al. (2015) and Heller et al. (2021).
We followed the Tettigoniidae classification described in Orthoptera Species File Online (Cigliano et al. 2022) to assign identified taxa to their respective subfamilies. All specimens were identified to genus level. Specimens that could not be assigned to a species were identified using temporary onomatopoeic names, such as 'chimer' and 'triller' (Tettigoniidae: Mecopodinae). The term 'call-type/species' is used synonymously with species throughout the paper as calls in Ensifera are reliable indicators of species (Kowalski and Lakes-Harlan 2013;.

Statistical analyses
As the recordings were made across a temperature range of 17.8-31.4ºC, we checked for the effect of temperature on the call parameters. Nine of the 15 call-type/species did not show any effect of temperature on the call parameters. In two species Hexacentrus unicolor and Orthelimaea insignis, there was a significant effect of temperature on the fundamental frequency. In two species Tegra novahollandiae viridinotata and T. viridivitta, there was a significant effect of temperature on echeme duration. In Ducetia sp. 2, we observed a significant effect of temperature on syllable duration. In Euconocephalus sp. 1, there was significant effect of temperature on syllable period. For the above species, the parameters were regressed to 26.3ºC (the temperature at which most calls were recorded) for comparison between species. To investigate the differences in katydid acoustic community between different habitats, we classified our sampling nights (n = 85) as sampling units. A total of 526 individuals were used for generating a presence-absence matrix across 126 habitat transects (60 understorey, 43 grassland and 23 shrubland). We used a nonmetric multidimensional scaling (NMDS) using Bray-Curtis similarity distances running 10,000 permutations to examine similarity between the habitats (Szinwelski et al. 2013;Gasc et al. 2018). We used analysis of similarity (ANOSIM), a non-parametric equivalent of ANOVA based on ranked dissimilarities, to test the null hypothesis, that there were no differences in the multispecies community between the sampled habitats. We used similarity percentage analysis (SIMPER) to investigate which species were most crucial in delineating the groups. All the multivariate analyses were performed on PAST (Version 4.02, Hammer et al. 2001).

Seasonal pattern in calling and habitat association
A total of 47 and 38 days of sampling were conducted at HGWS and BMWS, respectively. Each habitat was sampled every night between 1700 and 2200 hours in HGWS and between 1800 and 2300 hours in BMWS due to difference in time of sunset at the locations. Thus five sampling hours per day (235 sampling hours at HGWS and 190 sampling hours at BMWS) were obtained.
In HGWS, 30 and 17 nights of sampling were conducted in post-monsoon and dry seasons, respectively, and in BMWS, 26 and 12 nights. Individual insects were located in the field through visual scanning/listening each night. These insects were grouped based on the habitat they were found in. A total of 526 individuals were located from the two sites across two seasons.
A brief description of micro-habitat (habitat location, type of substrate, height, and degree of cover) and lunar phase was noted for every located individual.

Community composition
Of the 21 call-type/species recorded in this study from the two sites, 14 were from HGWS and 7 from BMWS. The katydid community at each site was represented by five major subfamilies: Conocephalinae, Hexacentrinae, Mecopodinae, Phaneropterinae and Pseudophyllinae ( Figure 2(a,b)). Of the 14 call-type/species from HGWS, four call-type /species belonged to Conocephalinae and Mecopodinae and two call-type/species each to Hexacentrinae, Phaneropterinae and Pseudophyllinae ( Figure 2). In BMWS, two calltype/species belonged to Conocephalinae and Pseudophyllinae each and one each to Hexacentrinae, Mecopodinae and Phaneropterinae.
Two call-type/species of Mecopodinae, Mecopoda elongata 'helicopter' and M. elongata 'train' from HGWS and BMWS, respectively, were similar to acoustic descriptions provided in Nityananda and Balakrishnan (2006). Acoustic and morphometric description for four Conocephalinae call-type/species, Conocephalus sp. X, C. melaenus, Euconocephalus pallidus and R. indica (E. indicus), have been described previously in Tiwari and Diwakar (2019). Hence, the acoustic description of the 15 call-type/species is presented below.

Acoustic description
The following subsection describes the acoustic diversity of katydids from the two sites. Acoustic descriptions of 15 call-type/species arranged by subfamilies is as follows: (i) Subfamily: Conocephalinae: Conocephalus maculatus Le Guillou (1841) produces a chirping call with an echeme of 7-9 syllables (SyD 17 ± 3 ms). The call is produced continuously at a high repetition rate (49 ± 2 syllables/s). Compared to the sympatric E. pallidus, C. maculatus has a spectrum with bandwidth of 28.11 ± 9.74 kHz and a peak frequency at 31.8 kHz (Tables 1 and 2, Figure 3(i), S2i).
Euconocephalus sp. 1 produces a prolonged buzzing call. The call consists of short syllables (SyD 27 ± 1 ms). The call-type/species has a repetition rate of 33 ± 2 syllables/s and a spectrum with bandwidth of 16.37 ± 3.3 kHz with peak frequency at 12.2 kHz. Euconocephalus sp. 1 had the lowest peak frequency among the four Conocephalinae from HGWS (Tables 1 and 2, Figure 3 (ii), S2ii).   (2) Euconocephalus pallidus* (n = 10) (ii) Subfamily: Hexacentrinae Hexacentrus sp. 1 produces a buzz-like conehead katydids with cycles of amplitude modulation. Individual syllables are very short and poorly differentiated with a high duty cycle (SyD 2 ± 0 ms, 330 ± 11 syllables/s). The frequency spectrum has a bandwidth of 15.92 ± 3.08 kHz, with a peak at 11 kHz (Tables 1 and 2, Figure 4(i), S3i).  The call of H. major Redetenbacher 1891 consists of a long continuous buzz of poorly defined syllables (SyD 1 ms). In contrast to the amplitude modulation in Hexacentrus sp. 1, H. major call alternates a long section of high amplitude syllables with a short burst of low amplitude syllables 427 ± 33 syllables/s). The call-type/species has a broadband frequency spectrum (22.85 ± 3.25 kHz), with an ultrasonic peak at 30.3 kHz. (Tables 1and 2, Figure 4 (ii), S3ii).
M. elongata 'triller' also produces a continuous trill with a discrete syllable structure. However, unlike the 'chimer', 'triller' exhibits amplitude modulation with louder syllables followed by a group of faint syllables and ended with the louder syllables again. The syllables differ only in amplitude and not in duration (SyD 11 ± 2 ms) and had a high repetition rate (65 ± 11 syllables/s). The broadband call (BW: 29.06 ± 11.83 kHz) has a peak frequency of 26.83 kHz (Tables 1 and 2, Figure 5 (ii), S4ii).

Taxonomy and morphometric description
The description for the katydid call-type/species is provided in Table 3. The descriptions of habitus and external genitalia of the 15 call-type/species can be found in Supplementary online material; Figure S8-S22. The identifying characters for each species have been added in the Supplementary online material.
The katydids exhibited a wide range in their body size. Both the smallest and largest individuals were from the Conocephalinae, C. maculatus, and Euconocephalus sp. 1 ( Table 3). The longest stridulatory files were found in the species of Mecopodinae, the longest file length in M. fallax (6.3 ± 0.4 mm). Despite their relatively small size, the two Ducetia call-type/species exhibited a high teeth density, 80 ± 7/mm for Ducetia sp. 1 and 97 ± 8/mm for Ducetia sp.2, second only to the Pseudophyllinae Phyllozelus siccus, 131/mm.
In BMWS, the katydids could be divided similarly into two. The post-monsoon community, including calls of C. maculatus, H. major, M. elongata 'train' and T. viridivitta; the dry season community consisted of calls of E. pallidus, Ducetia sp. 2 and Phyllozelus siccus (Figure 2(a)).  one Pseudophyllinae (T. n.viridinotata) were recorded from shrubs and bushes (Figure 2(b)).

Community distribution
The multivariate analysis showed the community composition to be significantly different among the habitats (stress = .06, dimension: 2, Figure 8, Table S1-S2). The 10 forest-canopy call-type/species clustered on the positive side of the X-axis. The grassland call-type/species clustered on the negative side of the X-axis. The shrubland call-type/species clustered separate from the grassland and understorey communities. Here again, we observed overlap between call-type/species from two locations.
The ANOSIM indicated complete separation among communities from the three habitats (p(same) <.001, Bonferroni p-values <.01; Table S3). The low R-value of the ANOSIM statistic suggests significant differences within the habitat. The shrubland species Ducetia sp.1 and understorey dwelling M. elongata 'train' were the most important species for delineating the habitat groups in the SIMPER results with 21% cumulative contribution (Table 4; average dissimilarity: 98.37%).

Discussion
We have described call patterns of 15 katydid species from the Indian subcontinent in this study. The acoustic records from the subcontinent are available primarily for calltype/species from the Western Ghats (Diwakar and Balakrishnan 2007a). From North-East India, katydid bioacoustics has received limited attention (Tiwari and Diwakar 2019). Besides HGWS, we also described the acoustic community from a previously unknown site in the Western Ghats.
Our calls for H. unicolor matched with the calls of H. unicolor recorded from Malaysia (Heller 1981from Cigliano et al. 2022. Calls of C. maculatus described in the study matched with calls recorded from Sumatra (Ingrisch 1993from Cigliano et al. 2022. Our survey showed a rich acoustic community at both locations. HGWS had a diverse community with pronounced seasonal dynamics changing across seasons. The ambient sound recordings from both sites across the seasons also indicate that our results represent the active acoustic community in different seasons (Diwakar et al. Unpublished date).
We found a diverse community of species with members from five major subfamilies. We found some subfamilies more frequently, while others were rarer, e.g. in HGWS, Conocephalinae and Mecopodinae had greater diversity than the other four families. In BMWS, the subfamilies Conocephalinae and Pseudophyllinae were represented by more call-type/species than others.
The katydid calls covered a wide range of spectral and temporal variations. The calltype/species followed their subfamilies' stereotypical patterns; the Conocephalinae exhibited high-duty song with both ultrasonic peak and a broad bandwidth like species described elsewhere (Korsunovskaya 2008;Hemp et al. 2015b). Euconocephalus sp. 1 did not resemble description of previously described taxa from the subcontinent (Farooqi and Usmani 2018a). The Mecopoda call-type/species showed spectral patterns like call-type/species described from Western Ghats (Nityananda and Balakrishnan 2006;Diwakar and Balakrishnan 2007a;Cigliano et al. 2022). The study reports the first description of M. fallax from the Indian subcontinent. Identification of the call-type/species was based on the acoustic description from South Asia (Heller et al. 2021;Cigliano et al. 2022).
The Pseudophyllinae showed the greatest variation in their songs. Two of the four calltype/species studied (Phyllomimus sp. and Phyllozelus siccus) had chirping calls and a narrow bandwidth with a low peak frequency like other previously known palaeotropical Pseudophyllinae (Heller 1995;Diwakar and Balakrishnan 2007a). The two species could not be identified to species level. Species key for both the taxa are based on original records (Brunner von Wattenwyl 1895) and need revision. The other two species of the Tegra genus showed a relatively broader bandwidth as compared to the other Pseudophyllinae species described from neotropics (Stumpner et al. 2013;Ter Hofstede et al. 2020). Tegra sp. calls were recorded from the shrubs outside the forest in HGWS and from the understorey in BMWS. There were no size differences between the puretone and broadband call-type/species except for the greater teeth number in the stridulatory file in pure-tone singers in this study (Heller 1995;Stumpner et al. 2013). The Phyllomimus sp. described in this study had significantly higher teeth density than previous descriptions of the genus, P. inversus (Heller 1995).
Two of the three Phaneropterinae call-type/species in the study exhibited low-duty temporal patterns with an ultrasonic frequency component and broadband spectrum typical of phaneropterine (Korsunovskaya 2008). Our study is the first acoustic description for O. insignis. The other phaneropterine, Ducetia sp. 1 and 2, could not be assigned to any known call-type/species of Ducetia (Personal communication with KG Heller).  documented acoustic signals for all the song types classified originally as D. japonica and revised the status for four species. They described five additional species based on the acoustic signal. The call patterns of the two Ducetia sp. from our study did not correspond to any species described in . Ducetia sp. 2 call resembled Ducetia malayana ; however, the call of Ducetia sp. 2 lacked isolated syllables of D. malayana described in their study. Of the three Ducetia species described from the subcontinent, D. japonica (Thunberg 1815) had been revised into five species, and none of those matched with calls described in this study. The description of D. dichotoma (Ingrisch and Shishodia 1998) is based on a female specimen that could not be used to identify male specimens. Further, neither of the call-type/species matched the description of the recently described D. serratus by Nagar et al. (2015). Our results with Ducetia call-type/species reiterate the importance of bioacoustics in the taxonomy of Tettigoniidae and the presence of more cryptic species in the field. This study is the first acoustic description for H. major. The third call-type /species i.e. Hexacentrus sp. 1 did not resemble any Hexacentrus species described from the subcontinent (Shishodia et al. 2010;Farooqi and Usmani 2018b). The three call-type/species showed broadband spectral pattern and significant temporal variation in their signals.
The acoustic community differed distinctly between the three habitats. The ordination neatly resolved the community differences between the habitats from two locations. Calltype/species from the same habitats of two locations clustered in multidimensional space. The overlap in communities in our results indicate the influence of environmental conditions on the acoustically active community (Gasc et al. 2018). Furthermore, 12 of the 21 call-type/species in the study were recorded from particular habitats only. This could be due to specific microclimatic and microhabitat requirements of the individual call-type/species (Jain et al. 2014;Gasc et al. 2018). The role of the physical structure of the individual's environment on efficiency of signal transmission and reception has been studied in Orthoptera (Römer 1993;Jain and Balakrishnan 2012). It is also evident in communities at both of our locations. At both sites (HGWS and BMWS), broadband calltype/species with uniform temporal patterns occupied open grasslands, maximising the transmission distances for such signals (Römer 1993). The shrubland and forest community call-type/species had a more mixed signal structure, like other tropical communities (Heller 1995;Diwakar and Balakrishnan 2007a). We can attribute this variation to the heterogeneous nature of the vegetation in these habitats that is generally sufficient to provide acoustic space for all individual call-type/species (Jain and Balakrishnan 2012;Jain et al. 2014). Our results from this study suggest that variation in spectro-temporal parameters among signal patterns is usually enough to secure sympatric signals a niche in their respective soundscapes (Jain et al. 2014).
This study also suggests possible convergence to common signal traits between allopatric signallers to optimise communication in their respective ecosystems. This study has contributed to the bioacoustics of katydid species from Northeast India. Future studies will aim to understand the effect of the allopatric callers with similar signal structures on the calling behaviour of an individual in their natural environment.