Island mysteries in the spotlight: Barbitistes kaltenbachi and Rhacocleis buchichii, the only bush-cricket species endemic to Croatia (Orthoptera, Tettigoniidae)

Abstract Hvar Saw Bush-cricket Barbitistes kaltenbachi Harz, 1965 (Phaneropterinae: Barbitistini) and Lesina Bush-cricket Rhacocleis buchichii Brunner von Wattenwyl in Herman 1874 (Tettigoniinae: Platycleidini) are flightless orthopterans restricted to a narrow area in the Mediterranean part of Croatia, both originally described from Hvar Island. In this study, all available information on these two interesting species is presented: data on morphology, bioacoustics, distribution, habitat, and a key to identification of the species belonging to genera Barbitistes and Rhacocleis in Croatia. The songs of both B. kaltenbachi and R. buchichii are described here for the first time, with the former one being the second known example of a synchronising and presumably duetting species. Both species were reassessed according to the IUCN Red List criteria, where B. kaltenbachi should be considered an endangered species, while R. buchichii is suggested to be downgraded to a less threatened category. Biogeography and evolution of the species are briefly discussed.


Introduction
Despite two centuries of research on Orthoptera in Croatia (Skejo et al. 2018), the only two bush-crickets endemic to the country, Hvar Saw Bush-cricket Barbitistes kaltenbachi Harz, 1965 and Lesina Bush-cricket Rhacocleis buchichii Brunner von Wattenwyl in Herman 1874, have escaped the eyes and ears of many researchers for quite some time. The knowledge of their distribution and biology is based merely on scattered documentation, with temporally and spatially well separated findings (Herman 1874;Harz 1965Harz , 1969Gausz 1970;Warchałowska-Śliwa et al. 2013;Wagner 2015;Puskás et al. 2018;Skejo et al. 2018).
Both species were originally described from the Island of Hvar (Lesina in Italian), one of Croatia's Adriatic islands, belonging to the Central Dalmatian archipelago. Until now Barbitistes kaltenbachi has been regarded as a Hvar-endemic, while Rhacocleis buchichii has been found on other Dalmatian islands ) as well as in mainland Croatia (Gausz 1970, Wagner 2015, Skejo et al. 2018). The same is true for most other Orthoptera species that were originally regarded as Hvar-endemics, such as Paramogoplistes novaki (Krauss, 1888), Chorthippus mollis lesinensis (Krauss, 1888) and Pholidoptera dalmatica maritima Zeuner, 1931(Willemse et al. 2009Rebrina and Brigić 2017;Puskás et al. 2018;Skejo et al. 2018).
Since so little has been published about Barbitistes kaltenbachi and Rhacocleis buchichii, the authors of the current paper felt the need to present a comprehensive overview of the information available. This paper is written around RF's finding of both species on the Island of Vis, well outside their formerly known distribution areas (briefly mentioned in Skejo et al. 2018). KGH's records of B. kaltenbachi from two sites on Hvar Is. (briefly mentioned in Warchałowska-Śliwa et al. 2013;Rebrina 2014;Skejo et al. 2018) are presented in more detail, together with a comprehensive analysis of the bioacoustics of the species. Unpublished records of R. buchichii are presented, together with an overview of the published data. Distribution, habitat, ecology, and possible evolutionary patterns of B. kaltenbachi and R. buchichii are discussed. To facilitate recognition in the field, an identification key to the Croatian species of Barbitistes and Rhacocleis is provided.

Study area
The Central Dalmatian islands (Fig. 1) form a tectonic unit composed of the geomorphologically similar islands of Vis, Hvar, and Brač (Borović et al. 1977). Unlike the typical Dalmatian orientation (direction NW to SE) exhibited by the majority of islands in the Adriatic Sea, these islands show Hvar orientation (direction W to E), witnessing their shared geological history. The Central Dalmatian islands harbour diverse habitats, e.g., rocky hills, small patches of forest, meadows (karst-poljes), ponds, and dry valleys. The backbone of the islands is formed by Mesozoic ridges of limestone and/or dolomite rock. The islands were connected to the Croatian and Italian mainland, by the ancient Po or Adriatic valley until approximately 11000-12000 years ago, when the sea level rose and the islands became separated (Lozić et al. 2012, Maselli et al. 2014. The Dalmatian coast is, just like the islands, warm and dry, with annual precipitation of only 700-900 mm. Isolated mountains (Svilaja Mt., Mosor Mt., and Biokovo Mt.) in the mainland have a mixed upper belt, rich in Mediterranean biota, with rare continental and oro-Mediterranean elements (Skejo et al. 2018).

Identification and taxonomy
Specimens were identified using Harz (1969), the original species descriptions (Herman 1874; Harz 1965) and photographs of the type material on the Orthoptera Species File (Cigliano et al. 2019). Taxonomy follows the Orthoptera Species File (Cigliano et al. 2019).
as a track of a stereo recording. After digitising the songs on a computer (sampling rate 44.1 kHz), oscillograms (after high pass filtering, typically around 1 kHz) and sound analyses were made using the following software: Turbolab (TL 4.0,Stemmer,Puchheim,Germany), Amadeus (Amadeus II, Martin Hairer, http://www.hairersoft. com), Audacity (Audacity 2.1.0; http://audacity.sourceforge.net) and Canary (Canary 1.2.4; Cornell Laboratory of Ornithology) on Apple. Due to the recorders' restricted frequency response, data on frequency were not evaluated. Each data point for the time pattern is based on not less than 20 independent measurements (except series data), given as mean ± standard deviation (SD) and coefficient of variance (CV) (see Suppl. material 3).

Analyses of Rhacocleis bioacoustics
Two sound recordings of two different individuals of this species have been made by Roy Kleukers (see Material examined under Rhacocleis buchichii), using a DCC-recorder (Philips DCC175) with a Shure Prologic condenser microphone. Sound recordings have been made in simple studio conditions at night, at 26-27 °C. Although the recordings are digital, some frequencies inaudible to humans were automatically removed before digitisation. Also, frequencies above ca. 20kHz are missing completely. The sound recordings are therefore not very useful for the analysis of frequencies. Yet, they are useful for temporal analysis of the song.
Temporal characters have been measured for both sound recordings using Wavelab 10 software (www.steinberg.net). A high pass filtering at 500Hz using phonetic software Praat 6.0.39 (www.praat.org) preceded the preparation of oscillograms for the Pitve specimen only.
Bioacoustic data of other Rhacocleis species has been derived from Reynolds (1998), Heller (1988), and Massa et al. (2012) (analysis of the song of R. japygia by Paolo Fontana).

Bioacoustics terminology
Syllable. sound produced during one cycle of wing movements. In Barbitistes only pulse-like closing hemi-syllables are known (Heller 1988) (Fig. 5A). In Rhacocleis both opening and closing hemi-syllables can be recognised; trigger syllable: pronounced syllable (sensu Stumpner and Meyer 2001); syllable period: time period measured from the beginning of a syllable to the beginning of the next; echeme (sensu Ragge and Reynolds 1998) or chirp (sensu Stumpner and Meyer 2001): a first-order assemblage of syllables; verse: a combination of echeme and trigger syllable; verse period: time period measured from the beginning of a trigger syllable to the beginning of the next (reciprocal value verse repetition rate). Details on songs of other Barbitistes and Rhacocleis species used for comparison can be found in Heller (1988), Ragge and Reynolds (1998) and Massa et al. (2012). For details on when and where Barbitistes-recordings were made see Suppl. material 3.

IUCN Red List Assessment
Data from the previous IUCN assessments for B. kaltenbachi  and R. buchichii (Skejo 2014;Hochkirch 2016a) were used, together with the new data presented in this study to calculate new values of AOO (area of occupancy) and EOO (extent of occurrence) using GeoCAT Editor (Bachman et al. 2011) (available at http://geocat.kew.org). IUCN criteria were applied to suggest new Red List status qualifications for both species (IUCN 2001).  Harz, 1965: Harz (1965: 443 (description of the species from Brunner von Wattenwyl's collection, 12 ♂♂ syntypes, and 16 ♀♀ syntypes, depicted in details, measured, based on males and females from Hvar Is., and a female from Trieste) (see Fig. 2); Barbitistes kaltenbachi Harz, 1965: Harz (1969: 75 (included in the key to species of the genus in Europe, depicted in details in figs 224, 237, 242-243, described and measured, cited from Hvar Is., and mentioned from a few mainland localities: Lukovo, Rijeka, Trieste); Barbitistes kaltenbachi Harz, 1965: Galvagni andFontana (1993): 204 (comparison with newly described B. vicetinus); Barbitistes kaltenbachi Harz, 1965: Heller et al. (1998: 25 (listed in the checklist of European Orthoptera); Barbitistes kaltenbachi Harz, 1965: Massa et al. (2012: 521 (presence in Italy, in the region of Trieste, unlikely); Barbitistes kaltenbachi Harz, 1965: Warchałowska-Śliwa et al. (2013 (reported from Hvar Is., Bogomolje, number and shape of chromosomes determined, as well as sex determination system X0); Barbitistes kaltenbachi Harz, 1965: Rebrina (2014 (Wagner 2015) 14 Zaostrog (Gausz 1970) "?" presence on Korčula Is. has never been confirmed (Harz 1969).  Harz (1965Harz ( , 1969) (E-J). The male specimen is labelled as a holotype, but since Harz (1965) did not designate a holotype, it should be labelled as syntype A pronotum and elytra of the syntype male in dorsal view B habitus of the male C labels of the male syntype D tip of abdomen with cerci and subgenital plate of the male E dorsal view of pronotum and elytra of a male F dorsal view of pronotum and elytra of a female G pronotum and elytrae of a male in lateral view H subgenital plate of a male seen from rear right I cercus of a male J base of ovipositor K ovipositor. Scale bars: 5 mm. Drawings are not to scale. Harz, 1965: Skejo et al. (2018: 19, 21 (endemic to Croatia, distribution in the country presented, as well as Rob Felix's photograph from Hum, Vis Is.).

Barbitistes kaltenbachi
Type material. Type specimens of B. kaltenbachi are deposited in NMW, Vienna. Harz's (1965) description is based on 31 syntypes: 12 males (10 from Lesina (Italian for Hvar), 2 from an unknown locality, probably "Dalmatia"), 16 females (12 from Hvar, 1 from Trieste (Italy), 3 with unknown locality) and 3 nymphs from Lesina. A male syntype is erroneously labelled as holotype (Fig. 3) . Hum is the highest mountain on the island, with a peak reaching 587 m a.s.l. The observed specimens represent the first record of B. kaltenbachi on the Island of Vis, and outside of Hvar Is. Four individuals were observed, of which three were collected (see Material examined). Individuals were observed basking in the morning sun on outer leaves of Holm Oak (Quercus ilex), at a height of ca. 1.5 meters. Barbitistes was accompanied by Long-tailed Speckled Bush-cricket, Leptophyes laticauda (Frivaldszky, 1868) and Schmidt's Marbled Bush-cricket, Eupholidoptera schmidti (Fieber, 1861) in the same habitat. The collecting site was inspected for only an hour, and a more comprehensive survey would probably have yielded more sightings.
Additional information on morphology. The right tegmen of the female has several rows of stridulatory pegs, in resting position covered by the upper left tegmen. To produce sound, the pegs are probably contacted by the slightly enhanced inner edge or a sclerotised vein on the lower side of the left tegmen.
Distribution. Harz (1965) originally described B. kaltenbachi from Hvar Is., Trieste (mainland Italy), and an unspecified locality (see Type material). Later, Harz mentions the species from Rijeka and Lukovo in mainland Croatia (Harz 1969). Massa et al. (2012) point out that the occurrence of B. kaltenbachi in mainland Italy (Trieste) is based solely on a female specimen. Since female specimens of Barbitistes are very hard to identify to species level, because of a high degree of overlap in interspecific characteristics (Nadig 1987), the record from Trieste is considered unreliable. Furthermore, Harz's records from mainland Croatia are assumed to be based on females and are therefore regarded as doubtful. Therefore, we consider B. kaltenbachi as being absent from mainland Croatia and endemic to the Dalmatian Islands.
All known records of B. kaltenbachi are shown on the map in Fig. 1 and presented in Table 1. The first records of B. kaltenbachi after its description by Harz (1965,  1969) were the ones by KGH in 2006, on Likova Glava, near Bogomolje, and near Sućuraj, both on Hvar Is. (Fig. 1, 4B). In Warchałowska-Śliwa et al. (2013) and Rebrina (2014) coordinates are provided but after mapping, those appear to be incorrect, and are therefore omitted from Table 1. The distribution on both Hvar Is. and Vis Is. is probably much wider and the presence of the species on nearby islands such as Brač Is., Korčula Is., as well as neighbouring islets, is also possible.
Habitat. No published data on the habitat of B. kaltenbachi are available. On Vis Is. the species has been found on bushes and small trees of Holm Oak, in maquis intercepted with scattered patches of grassy vegetation and bare limestone rock (Fig. 4A). The habitat on Hvar Is. is similar to that on Vis Is. but lacks rocky outcrops (Fig. 4B). Barbitistes is an arbusticolous genus, as its members are usually living in bushes, shrubs and on low trees (Galvagni and Fontana 1993;Pavićević et al. 2014). Thus, Hvar Saw Bush-cricket inhabits typical Barbitistes -habitat.
Song description. The male calling song consists of short (up to ca. 10 ms), pulse-like syllables without clearly recognisable impulses (tooth impacts). Syllables are arranged in a stereotyped pattern. One loud syllable, the trigger syllable, assumed to be the marker for female response (Stumpner and Meyer 2001), is followed by a group of two or three (rarely one or four) 'softer' syllables (echeme) (Fig. 5A). The trigger syllables are 6 to 16 dB louder than syllables within the echeme (Suppl. material 3: Table S3.1). After a short interval, the same pattern (verse) is repeated again and again (Fig. 5B). Periods range from 280-400 ms for the verse (verse repetition rate ca. 3 Hz), 120-160 ms for the post-and pre-trigger periods, to 40-50 ms for the syllables in the echeme (Suppl. material 3: Table S3.2). These sequences of verses are variable in duration but may last more than 30 s (Suppl. material 3: Table  S3.3), at least while two males are in acoustic contact. During interactions males seem to prefer singing during the other male's song pauses, but they are not inhibited. Songs overlap quite often, even for many seconds. In situations of overlap, males typically synchronise the pattern: they produce trigger syllables nearly at the same time (Fig. 5B, C). The delay between the trigger syllables of two synchronising males is typically below 20 ms (Fig. 5E) and the roles of leader and follower Table 1. Known records of Barbitistes kaltenbachi with data on the sites and collection events. The #map correspond with the numbers in Fig. 1. Type specimens collected by various persons in different events (Harz, 1965)     proximal half of a cercus is in an abrupt but obtuse angle with the distal half, giving the cercus an angular sinuosity, like that of an open elbow. Barbitistes constrictus has nearly identically shaped cerci (Fig. 7B). All the other species of the genus have cerci of a different shape, in the absence of a thickened mid-part and in being more gradually curved, showing a smoother sinuosity in the apical part (Figs 7C-H), except for B. vicetinus, which has uniquely shaped cerci (Fig. 7F).
The song of B. kaltenbachi differs distinctly from the songs of all other Barbitistesspecies in the temporal pattern of the syllables (Fig. 8). Definite diagnostic differences between females cannot be given. Females of different species of Barbitistes show many similarities, some exhibiting intermediate characters (see e.g., Nadig 1987). In this paper, the photos of females are of specimens present together with males in the same bush and hence are identified as belonging to B. kaltenbachi. See Suppl. material 1:Table S1 for measurements of specimens in collections and literature data.
Key to the species of the genus Barbitistes in Croatia (males only) In Croatia, four species of Barbitistes occur, namely Barbitistes kaltenbachi, B. ocskayi, B. serricauda, and B. yersini (Rebrina 2014;Skejo et al. 2018). Males are easily identified to species level, but females are not. The key to males presented below is based on the morphology of cerci (angle and thickness) and subgenital plate (presence or absence of nose-like projection), and is adapted after Harz (1969). The key to the identification of females published by Harz (1969) has proven to be unreliable in practice, so it is not presented here. Cercus thickened in the middle, proximal half in an abrupt but obtuse angle with the distal part, giving the cercus an elbow-like appearance. Distal half angularly sinuous (Fig. 8)  Extent of occurrence (EOO): calculated from the known data 90 km 2 , maximal estimation ca. 780 km 2 (if all the hilly habitats in Hvar Is. and Vis Is. are included).
Newly proposed status. The species is reassessed here as Endangered. Up to now, the species has been known from only two Adriatic islands (Hvar and Vis). Inhabiting a restricted number of sites, it has an AOO of 12 km 2 to maximally 80 km 2 , and a known EOO of 90 km 2 (maximally estimated less than 800 km 2 ). This makes it qualify for EN, following the criteria B1ab(iii)+B2ab(iii). Since there are only three sites at which this species occurs with certainty, only a few threat events may be enough to wipe out or threaten proportional parts of the population. Forest fires, for example, occur quite often on the Adriatic islands and can damage important parts of the species' habitat, as can touristic and recreational developments and clear cutting of maquis and scrubland.  Karny (1907b): 131 (misspelled; mentions that the species has been known only from Hvar Is.); Rhacocleis buchichii Herman: Caudell (1908): 5 (listed in the catalogue of Decticinae species, mentioned as endemic to Hvar Is.); Rhacocleis bucchici Herm.: Werner (1919): 217 (misspelled; mentions species to occur on Brač Is., "south of Neresi, near the water tanks in the bushes, not common". Also mentions "species until then only known from Lesina"); Rhacocleis buchichii Herm.: Salfi (1924): 43, 45 (key to species); Rhacocleis bucchichi Br.: Ramme (1951): 118 (reported to have found the species in hilly terrain under Pistacia terebinthus L. and noted that the species was timorous and difficult to catch); Rhacocleis bucchichi Br.: La Greca (1959)  Historical misspellings. Historically, the species name of Rhacocleis buchichii was misspelled a lot, for obvious reasons. It was even misspelled more often than cited correctly (see above). One specific misspelling, bucchichi, can be explained by the fact that the person after whom the species was named is Grgur/Gregorio Bučić/Bucchich (1829-1911), a Croatian naturalist with a surname that contains two consonants (voiceless postalveolar affricate consonant č, pronounced /tʂ/; and voiceless alveolopalatal affricate ć, pronounced /tɕ/). These consonants were written in Croatian language in many different ways (ch, cch, ci, cci, chi) in the past, which has probably caused the confusion. Type material. Syntypes of Rhacocleis buchichii, one male and one female, are deposited in MfN, Berlin (Fig. 10). In the collection of NMW, Vienna, there are three specimens from the collection of Brunner von Wattenwyl (pers. comm. S Randolf ). These three specimens are not labelled as types, but since they are part of the typecollection, they should be considered syntypes too.  (Figs 1, 2A, 11). Several photographs of the specimen were taken in its natural habitat  . 11C), and it was subsequently collected (Fig. 11). As in the case of Hvar Saw Bush-cricket, this is the very first record of Lesina Bush-cricket for the island of Vis. It was recorded in the direct vicinity of the collecting site of Barbitistes kaltenbachi.
Distribution. All known records of R. buchichii are depicted on the map in Fig.  1 and presented in Table 2. After its description from Hvar Is., the species has been found at scattered localities on the island and in a wide area outside the island: Zaostrog (Gausz 1970), Biokovo Mt. (present paper;Wagner 2015), Brač Is. (present paper;Puskás et al. 2018) and Troglav Mt. within the Dinara Massif (Skejo et al. 2018). Harz (1969) mentions the species' occurrence on Korčula Is., but its presence on the island has never been confirmed.
Habitat. The first information on the habitat of R. buchichii was given by the name bearer himself; Bucchich (1886) found the species in July in a hilly terrain in bushes, especially Erica. He mentions that the species does not occur in fields (see Bibliography). Novak (1888) also mentions hilly areas as the prime habitat, adding that the species can be found from June to October under Cistus monspeliensis. Some individuals entered the house, probably as stowaways in collected firewood. In gardens, it was found under roses. Ramme (1951) reported the species to live under Pistacia terebinthus and noted that it was timorous and difficult to catch. Gausz (1970) found a specimen in the littoral zone near Zaostrog, in the undergrowth (karst-steppe vegetation of 3-10 cm high) in a stand of Ficus carica and Olea europaea. Kleukers (pers. comm.) found R. buchichii in 1996 on a rocky slope with low bushes around the villages of Pitve and Vrisnik, Hvar Is. Wagner (2015) found R. buchichii on Biokovo Mt., under Black Pine, Pinus nigra, while JS found it on Troglav Mt. under Pistacia sp. On 25 Jul. 1982, KGH found many specimens at night, walking and jumping on the road between Jelsa and Gdinj.
Song description. We regard this song description as preliminary, as we were able to analyse the sound recordings of only two males. The calling song of R. buchichii consists of echemes that are repeated in a series of 4-13. However, in one of the recordings only the series of 2-5 occur. Echemes are repeated at a rate of ca. one or two echemes/s. In the available recordings, no continuous repetition of echemes is found. Echemes seem to have a more or less fixed structure, last ca. 160-220 ms and contain 7-9 syllables. Syllables are repeated at a rate of 40-50/s (26-27 °C).
Species diagnosis. Within Tettigoniidae, the members of the Platycleidini tribe have either an unarmed prosternum or the prosternum bears two spines (Massa et al. 2012). The Rhacocleis genus belongs to the latter group, together with Pterolepis Rambur, 1838, Antaxius Brunner von Wattenwyl, 1882, Anterastes Brunner von Wattenwyl and Yersinella Ramme, 1933, among others. Based on the shape of the cerci of the male and the subgenital plate of the female, La Greca (1959) placed R. buchichii into the Rhacocleis neglecta-species group (La Greca 1959), composed of R. neglecta (Costa, 1863), R. japygia La Greca, 1959 (both from central and southern Italy) and R. buchichii (Croatia). These species are characterised by 1) slender cerci with a decurved inner tooth close to the base (♂♂) and 2) a quadrate subgenital plate with a central keel and lateral depressions (♀♀). Rhacocleis buchichii is characterised by very small, neglecta-type titillators (Fig. 10J, K), but with a blunt and rounded apex (La Greca 1959). Cerci of the male are coni-cal, delicate, very long and have a medial tooth protruding ca. 1/5 from their base (Figs 10, 11). La Greca (1959) provides a detailed description of the subgenital plate of the female: it presents a median carina and two lateral carinulas arranged obliquely and converging to the centre, towards the apex of the median carina. Transversal grooves on both sides of the median carina are situated more towards the apex of the subgenital plate than towards its base, and are limited posteriorly by the two lateral carinulas (Fig. 10L).
Rhacocleis buchichii is easily distinguished from its only congener in Croatia, R. germanica (Herrich-Schäffer, 1840). The cerci of R. buchichii males are very slender, while the males of R. germanica have more robust cerci, with a less elongated distal part. Each cercus of male R. germanica has a long and straight inner tooth. Females of R. buchichii have a rectangular subgenital plate with a median keel, which is armed at the tip (Fig. 10L), while R. germanica females have a prolonged, oval subgenital plate with an apical incision. The two Croatian species also differ in colouration. Rhacocleis germanica is more reddish and brownish tinted, without a white band on the paranota of the pronotum. Rhacocleis buchichii has darker and greyer tints and has a clear pale paranotal band (Figs 10, 11). See Suppl. material 2: Table S2 for the measurements of specimens in collections and literature data.
The song (Table 3) is different from the song of R. germanica in the number of syllables per echeme, being ca. ten in R. germanica, and 7-9 in R. buchichii. In R. japygia echemes are repeated in short series (2-8 echemes) or more or less continuously, with a repetition rate of ca. 1-2.5 echemes/s. Echemes consist of 5-7 syllables which are repeated at a rate of ca. 30/s. In R. neglecta, echemes are produced in long series or continuously, at a rate of 1-3 echemes/s. Echemes consist of 3-5 syllables. Syllables are repeated at a rate of ca. 25-40/s. Echeme repetition rate and the number of syllables per echeme are considered to be the main distinguishing features of different species.  Wagner (2015) Key to the species of the genus Rhacocleis in Croatia
Newly proposed status. Lesina Bush-cricket inhabits the Adriatic islands of Hvar, Brač and Vis, as well as certain mountains in mainland Dalmatia. Since the species has recently been found on Vis Is. and its presence on Troglav Mt. has now been confirmed, its known range has extended significantly compared to the previous assessment. Based on the above calculations, we propose the species to be downgraded to a less threatened category.
The species occurs in Natura 2000 protected areas (Vis Is., Hvar Is., Brač Is.) and in a protected natural park (Biokovo Natural Park), but it is expected to occur outside the protected areas as well, where the main threats to the species' survival persist. Based on numerous references, the habitat type of the species, to be considered in a future reassessment, has to be extended with scrubland (see above, under Habitat). Table 3. Bioacoustic data of R. buchichii, R. germanica, R. japygia, and R. neglecta. Presented are echeme repetition rates per second, number of syllables per echeme, and syllable repetition rate per second. Data of R. germanica and R. neglecta obtained from Ragge and Reynolds (1998) and Heller (1988), data of R. japygia obtained from Massa et al. (2012).

Dalmatian endemics from a biogeographical perspective
The Adriatic islands are of a rather young evolutionary age (Malvić 2016) compared to large Mediterranean or Greek islands (Robertson andDixon 1984, Willemse et al. 2018). Therefore, the rate of endemism on the Adriatic islands is relatively low (Kenyeres et al. 2009). During most of the time from Oligocene to Pleistocene, the Adriatic islands seem to have been a part of the mainland Balkans (Giuli et al. 1987). From the lower Pliocene, when the Adriatic sea was large and extended far into the northwest, through the Würm ice age (ca. 100,000-11,700 years ago), when the sea level was low, as well as in the early Holocene, when the sea level increased, the islands were connected to the mainland (Velić & Malvić 2011, Malvić 2016. They may have been isolated during some interglacials, but these periods were probably short. Nevertheless, a few endemics known from the central Dalmatian islands, mainly documented in plants (Bogdanović et al. 2014), may justify the status of this island group as a distinct biogeographic unit. Among animals, however, very few examples are known, most of which refer to troglobitic taxa. Of particular note is the troglobitic beetle genus Spelaeobates Müller, 1901, endemic to the central Dalmatian islands and adjacent Croatian mainland. S. kraussi Müller, 1903and S. pharensis Müller, 1901are Brač and Hvar endemics, respectively (Pretner 1968, Gottstein Matočec et al. 2002) while S. novaki Müller, 1901 is found on Dugi otok Is. and adjacent Croatian mainland, at the foot of Velebit Mt. (Jalžić 1983). Allegrucci et al. (2017) recently mentioned a new, yet undescribed species of Troglophilus (T. sp. 1) from the caves on Mljet Is. However, these troglobitic and troglophilous forms live in special habitat 'islands', not necessarily related to true islands.
The terrestrial fauna of most Croatian islands still remains profoundly understudied and the Orthoptera of the Central Dalmatian islands, in particular, have never been studied systematically (Skejo et al. 2018). At the present state of knowledge, however, the occurrence of endemic or subendemic Orthoptera species is surprising and note- worthy. How and when could these forms speciate? Surprisingly, the available data indicate rather different histories for the two species studied in this paper.
To start with the less complex case, the Lesina Bush-cricket Rhacocleis buchichii is endemic to Dalmatian islands (Brač, Hvar, Vis) and mountains (e.g., Troglav and Biokovo), and occurs sympatrically with R. germanica, the latter being very common in the Mediterranean region of Croatia. The two species show clear morphological differences and do not seem to be closely related. On the other side of the Adriatic Sea, in Italy, two species presumably closely related to R. buchichii and each other occur, namely R. neglecta and R. japygia (La Greca 1959). Rhacocleis neglecta is widespread in the Italian mainland, while R. japygia occurs in a small area between Basilica and Puglia (Massa et al. 2012). All three species of the so-called neglecta-group (neglecta, buchichii, japygia;La Greca 1959) have similar cerci in males and quadrate subgenital plates in females. Thus, it is highly probable that these species share the most recent common ancestor. La Greca (1959) considers R. buchichii as originating from the Apennines, Italy. The ancestor of R. buchichii may have inhabited the area currently covered by the Adriatic Sea during one of the periods when the sea level was low (e.g., during Würm; Malvić 2016). After the sea level rose and the Balkan Peninsula became separated from the Italian Peninsula, speciation became possible.
Despite the fact that the similarity between the three species was mentioned as early as 60 years ago (La Greca 1959), a comprehensive comparative study has never been conducted. In terms of bioacoustics, the songs of the three allopatric Rhacocleis species are relatively similar, but all four mentioned species of this genus can be identified by their songs. However, very few sound recordings of these species exist and a more thorough study of their bioacoustics, including female behavioural response experiments, is required to test this hypothesis.
The situation with Hvar Saw Bush-cricket Barbitistes kaltenbachi is quite different and much more complex. This flightless bush-cricket is endemic to two Dalmatian islands (Hvar Is. and Vis Is.). The song of B. kaltenbachi is unique within the genus and does not resemble that of any other known species (Figs 5,6). On Hvar Is., B. kaltenbachi occurs sympatrically with B. yersini, a common and widespread species in the Western Balkans (Hochkirch et al. 2016a, Skejo et al. 2018, with an isolated, trans-Adriatic occurrence in Central Italy (Massa et al. 2012). The true origin of this isolated population would be an interesting subject for a bioacoustic/genetic study.
According to the cercal morphology, B. kaltenbachi could be closely related to B. constrictus, their cerci being nearly identical (Fig. 8). On the other hand, B. constrictus has a completely different distribution and habitat. It inhabits central, northern and eastern Europe (Harz 1969) and is typically found in coniferous forests (Hochkirch et al. 2016b), whereas B. kaltenbachi is a species of dry and hot maquis.
Looking at the distribution ranges of other Barbitistes species, one can recognise an intriguing gap between the northern and southern distributional areas of B. ocskayi (see Hochkirch et al. 2016b, Skejo et al. 2018, in which the distribution of B. kaltenbachi seems to fit perfectly. If B. kaltenbachi were an offshoot of B. ocskayi, it would be relatively young (perhaps isolated during one of the interglacials), but in this case it means that its unique song and distinctive cercal morphology evolved very fast. Since both characters are known to be under sexual selection (Lehmann 1998), a rapid evolution seems possible. Under these premises, it would be particularly interesting to get a molecularly based estimate of the age of the species involved. If the distribution of kaltenbachi is restricted to Hvar Is. and Vis Is. -during a quick search we only found B. yersini on Brač Is. and the mainland opposite to Hvar Is. -this could be ascribed to its occurrence (or origin?) in only one of the two separate local Würm glacial refugia, that were situated south of the Neretva River (see Podnar et al. 2004).
Based on the shape of the subgenital plate, B. kaltenbachi is similar to ocskayi. Based on cercal morphology, however, B. ocskayi and B. yersini seem to resemble each other more. Barbitistes vicetinus is a species restricted to Northern Italy, which exhibits an isolated occurrence comparable to that of B. kaltenbachi. Despite a superficial resemblance of the male cerci of B. kaltenbachi and B. vicetinus (the latter also has a somewhat thickened mid-part; Fig. 7F) the authors consider the cerci of the two species very different from each other. Barbitistes vicetinus' cerci show a right angle between the proximal and distal parts, and almost lack any sinuousity in the distal part (compare B. kaltenbachi in Key to the species). Furthermore, their songs are different (Fig. 8).

Acoustic behaviour of Barbitistes kaltenbachi
Barbitistes kaltenbachi belongs to Phaneropterinae, a bush-cricket subfamily in which females of most species do not respond to male songs only by phonotactic approach, but also react with their own acoustic signals. These sounds are used by males for locating females (see Heller et al. 2015 for a review). Female responses are often short, occur very fast after releasing male the song element (in less than 100 ms; see Heller et al. 2018) and are scarcely detectable by predators, but can be exploited by conspecific males (Villarreal and Gilbert 2014). Therefore, it is not surprising that a variety of song modifications which can be interpreted as forms of defense against eaves-dropping rivals, are known (Heller and. The rivals, on the other hand, can attempt to mask female response (Bailey et al. 2006).
Male calling songs of all species belonging to Barbitistini sensu stricto (genera Barbitistes, Metaplastes, and Ancistrura) are characterised by short, isolated syllables showing a species-specific pattern (Heller 1988). They mostly contain a short sequence of syllables (up to ten) followed by a larger interval, after which an isolated syllable is produced. This syllable is called trigger syllable, because a female ready to mate responds directly to it. Barbitistes males (e.g., B. serricauda and B. kaltenbachi; Fig. 5) sometimes produce so-called "extra syllables" (Stumpner and Meyer 2001) at the time of an expected female response, possibly to hinder the rivals' ability to locate answering females. However, there are almost no data describing the calling behaviour of two or several males singing together. A rival male could e.g., disturb the silent interval (before the next trigger syllable of another male) by sounds reducing the probability of female response. In some species, trigger periods are certainly long enough for such attempts (e.g., Ancistrura nigrovittata, B. constrictus, B. serricauda), but in others, sequences of verses are probably too short (e.g., B. alpinus, B. yersini), making the prediction of the trigger periods difficult. Barbitistes kaltenbachi songs are relatively long, so the rivals' disruption attempts should not be difficult. However, a rival would arguably not gain much from such behaviour, because its own song could be disturbed by the song of the 'attacked' male in a similar way. Thus, it could be more worthwhile to synchronise the song as closely as possible with the song of a neighbour.
During the duets of B. kaltenbachi, trigger syllables/verses of both males started within only 20 ms (Fig. 5). Some duetting frogs have a mean difference of 79 ms (Legett et al. 2019). The shortest onset difference tested by Snedden and Greenfield (1998) in a synchronising tettigoniid was 26 ms. When synchrony is near-perfect (sensu Legett et al. 2019; delays of 5 ms or less), as observed here in many instances, males increase the amplitude of their signals by overlapping. In any case, both males are still able to hear female responses, provided the female is in the hearing range of both. This system works even if males use auditory time windows, as reported from other Barbitistini (Heller and von Helversen 1986). To our knowledge, synchronising in duetting bush-cricket species has been observed only in Amblycorypha parvipennis (Shaw et al. 1990). This species has syllables (called phonatomes in Shaw et al. 1990) with a duration of 100 ms. Female responses follow ca. 120 ms after the beginning of a male syllable. In Barbitistini, synchronising would have been considered unlikely due to the fast female response and narrow auditory time windows.
Unfortunately, as nothing is known about female acoustic behaviour or the potential male reaction, the hypotheses regarding B. kaltenbachi acoustic behaviour can only be based on what is known of other species. Barbitistes females respond to the trigger syllable after a delay of ca. 40 ms (Stumpner and Meyer 2001), probably caused by an audiomotoric reflex (named acousto-motorical reflex in Ancistrura nigrovittata by Dobler et al. 1994). In this case, a preference for the leading call would be obvious. Females would (and can only) respond to the first sound. This is comparable to a precedence effect discussed in the context of synchrony and alternation in chorusing animals (Greenfield 2002). However, males are not necessarily under pressure to call first. This depends a lot on the specific borders of the auditory time window. The trigger syllable of the leading male may start the auditory time window not only in the female, but also in the male follower. Ideally, it should not be reset by the follower's own trigger syllable. On the other hand, singing at a different time from the leading male may result in covering the female response. Thus, exact synchronisation might be the optimal strategy. Of course, much more data, especially from females, are necessary to prove this hypothesis, but it seems a fascinating prospect for acoustic co-operation -even if it is considered to 'make the best of a bad situation'.

Concluding remarks
With this paper, we attempted to enhance the information on the distribution of two Croatian endemic bush-crickets, as well as the knowledge of their morphology and bioacoustics. The songs of Barbitistes kaltenbachi and Rhacocleis buchichii are described here for the first time. The IUCN Red List status has been reassessed here for both species; we suggest R. buchichii to be downgraded to a less threatened category, while B. kaltenbachi should be upgraded to 'Endangered'.
The knowledge of the biology of both species is still scarce. Barbitistes kaltenbachi is suspected to be active early in the season, while R. buchichii is probably active late in the season, based on many records of this species from October (see Tables 1, 2). JS visited Hvar Is. two times, in late July and late August 2017, focusing on finding both species, but without success. Both dates were apparently too late in the season for B. kaltenbachi, taking into account that KGH collected specimens of this species on Hvar Is. at the end of May, without observing nymphs at that time.
The fact that the type series of B. kaltenbachi consists of a fairly large number of specimens (22) from the same locality (Hollier and Bruckner 2015) could suggest that the species exhibits gradations in certain years. Gradations are a well-known aspect of the biology of other Barbitistes species (Galvagni andFontana 1993, Stumpner et al. 2015), some of which were, and still are, sometimes characterised as pest species (e.g., Bei-Bienko 1954, Cavaletto et al. 2019. However, outbreaks of B. kaltenbachi have never been reported. To enrich the knowledge of the Orthoptera fauna of Vis Is. and other Adriatic islands, the authors would like to suggest visiting entomologists to pay attention to these and other species, and enter their sightings, accompanied by photos, on websites such as Observation (www.observation.org) or iNaturalist (www.iNaturalist.org).