BMNH British Museum (Natural History), London;

CCH Collection of Claudia Hemp, Bayreuth;

NHMW Naturhistorisches Museum, Vienna.

Sound files of all species studied acoustically are given in the Suppl. materials and will be deposited in OSF (Cigliano et al. 2023).

The specimens were housed either in plastic containers or gauze cages and were provided with a daily diet of Taraxacum officinale or Fragaria sp. leaves, as well as leaves from other shrubs and trees which were regularly replaced.

Specimens studied

For this study sounds from 9 species and 50 specimens (37 ♂♂, 13 ♀♀) were analyzed (about 39 GB sound data corresponding to 84 h recording at 2 × 64 kHz sampling rate), all collected in Tanzania by Claudia Hemp.

D. flavoviridis sp. nov.: Nguru Mountains, 2♂♂ + offspring, 4♂♂, 4♀♀, (CH8855-8862), December 2020; 1 ♂, March 2021.

D. jagoi Ragge, 1980: East Usambara Mountains, Amani/Zigi, 1♂ (CH8619), November 2018; 1♂, February 2021.

D. magombera sp. nov.: Udzungwa Mountains, 1♂, near Mangula Gate (paratype), 2016; 1♂ 2021.

D. sanje sp. nov.: Udzungwa Mountains, 1♂ + offspring – 3♂♂, 2♀♀ (CH8167-70), December 2015; 1♂, 2♀♀ (CH8863-65), December 2020.

D. ornata B. v. Wattenwyl, 1878: 1♂, West Usambara Mountains, Lutindi, January 2015; 1♂, East Usambara Mountains, Nilo Forest Reserve, April 2016; 2♂♂, 1♀, East Usambara Mountains, Amani/Zigi, (CH8093-5), November 2015; 1♂, January 2021; 3♂♂, 2♀♀ (CH8951-5), December 2021; 3♂♂, 1♀, Nguru Mountains, February 2021 (CH8150-2); 1♂, Coastal forest: near Kisarawe, Kazimzumbwi Forest Reserve, July 2015; 3♂♂, Zanzibar, near Jozani Nationalpark, June 2017 (CH8450), October 2018 (CH8620), and February 2022.

D. ngurumontana sp. nov.: 3♂♂, 1♀, Nguru Mountains above Dibago (CH8986-9), November 2021.

D. tanneri Ragge, 1980: 1♂, West Usambara Mountains, Mazumbai Forest Reserve, March 2021.

D. versicolor sp. nov.: 1♂, Uluguru Mountains, February 2016.

D. zernyi Ragge, 1980: 2♂♂, Udzungwa Mountains (CH8034), July 2015, and February 2016.

We could not obtain acoustic data on the following species: D. bulla Ragge, 1980, D. nitens Ragge, 1980, D. scutellata Hemp, 2017, D. takanoi Massa, 2021, and D. ugandana sp. nov. From D. grandis Ragge, 1980 the tick song (see below) is figured by Naskrecki and Guta (2019).

Recording

The male calling songs were recorded in the laboratory using a digital bat detector (Pettersson D1000X) mostly with a sampling rate of 100 kHz (rarely 192 or 300 kHz), often from field-collected animals, but in some species also from animals reared in the laboratory. The singers were caged in plastic tubes or gauze cages with microphone fixed or handheld at distances between 5 and 60 cm. The female response behavior was studied in the laboratory using virgin females, reared from field-collected nymphs (rarely) or from eggs. Duets were recorded in stereo using a Sony ECM-121 microphone (frequency response relatively flat up to 30 kHz according to own tests) and an Uher M645 audio microphone connected to a personal computer through an external soundcard (Transit USB, “M-Audio”; 64 kHz sampling rate). Here, male and female were placed separately into two plastic tubes (Drosophila tube 28.5 × 95 mm, Biosigma, Cona, Italy) mostly standing side by side, with one microphone placed inside or on top of each vial. Both microphones typically picked up male and female sounds, but with different amplitudes. Most recordings were made at temperatures between 20 and 25 °C, only rarely also between 15 and 28 °C.

The stridulatory movements in D. sanje sp. nov. were registered using an opto-electronic device (Helversen and Elsner 1977, modified as in Heller 1988) as well as the equipment described in Montealegre-Z and Mason (2005; output used for Fig. 27). In summary, the equipment used the optical system proposed by Hedwig (2000), and a sampling rate of 100 k-samples/s for both sound and motion traces. Sound signals were recorded using a wide-bandwidth response 1/8-inch microphone (Brüel and Kjaer, 4138-A-015, with pre-amplifier model 2670, Brüel and Kjaer, Nærum, Denmark), and the Polytec software (PSV 9.2, Waldbronn, Germany) for multi-channel data acquisition.

For the recording of wing vibrations the method of micro-scanning Doppler vibrometry was used, described in previous studies (e.g. Chivers et al. 2017). Specimens were immobilized by exposure to a triethylamine-based mix (FlyNap, Carolina Biological Supply Company, Burlington, NC, USA) for 3–5 min. The specimens were placed on a block of BluTack (Bostik, La Défense, Paris, France) and their legs were gently clamped to the block with small staple clamps. The wings were extended and maintained in position by putting a small drop of a mix of beeswax (Fisher Scientific Loughborough, UK) and Colophonium (Sigma-Aldrich, Dorset. UK) (50:50). The BluTack block was affixed to a brass plate which itself was attached to an articulated aluminium rod allowing the specimen to be maneuvered into the required position.

Vibration-compliant areas of forewings, and associated frequency characteristics, were measured using a micro-scanning laser Doppler vibrometer (Polytec PSV-500;Waldbronn, Germany) fitted with a close-up attachment. The mounted specimens were positioned so that the extended wings were perpendicular to the lens of the laser unit. A loudspeaker was positioned above the laser unit and facing the animal to broadcast the sound stimulus. The acoustic stimulus used was periodic chirps, generated by the Polytec software (PSV 9.2), passed to an amplifier (A-400, Pioneer, Kawasaki, Japan), and sent to the loudspeaker (Ultrasonic Dynamic Speaker Vifa, Avisoft Bioacoustics, Glienicke, Germany). The periodic chirps spanned frequencies between 2 and 100 kHz, and the stimulus was flattened so all frequencies were represented at 60 ± 1.5 dB (SPL re. 20 μPa) at the position of the wings. A Brüel and Kjaer 1/8-inch condenser microphone was placed at the position of the wings to monitor and record the acoustic stimulus at the position of the wings as a reference. The laser system was used in scan mode.

Analysis

Song measurements were obtained using Amadeus II and Amadeus Pro (Martin Hairer; http://www.hairersoft.com). Oscillograms of the songs were prepared using Turbolab (Bressner Technology, Germany). Data are presented as mean ± standard deviation. For spectral analysis, short song parts (echemes or syllables) were analyzed using the mean of 512 points wide Hanning windows, overlapping by 100 points.

Morphology

The stridulatory files were photographed using a Canon EOS 550 D fixed on a Leica MZ 125 stereomicroscope or a Sony Cyber-shot DSC-P120 on an Olympus SZ Binocular Stereo Zoom Microscope. Programs used for processing the photos were EOS Utility, Helicon Focus, and Adobe Photoshop.

To get an impression of the three-dimensional geometry, the auditory tracheae of D. sanje sp. nov. were scanned using a SkyScan 1172 X-ray μ-CT scanner (Bruker Corporation, Billerica, MA, USA) with a resolution of 12.9 μm (50 kV source voltage, 200 μA source current, 200 ms exposure and 0.2° rotation steps). For obtaining a series of orthogonal slices, the μ-CT projection images were reconstructed with NRecon (v. 1.6.9.18, Bruker Corporation, Billerica, MA, USA). The three-dimensional segmentation of the auditory tracheae were performed with the software Amira-Aviso 6.7 (Thermo Fisher Scientific, Waltham, MA, USA) (Celiker et al. 2022).

Acoustical terminology

Tettigonioids produce their songs by opening and closing movements of their tegmina. The sound resulting during one cycle of movements is called a syllable, often separable in opening and closing hemisyllable (Ragge and Reynolds 1998). Syllable duration: time period measured from the beginning of the first impulse to the last; syllable period: time period measured from the beginning of a syllable to the beginning of the next (reciprocal value: syllable repetition rate, SRR). In many species and also in Dioncomena, these syllables can be grouped into echemes: first-order assemblage of syllables (duration, period and rate as for syllable). However, the male songs are typically long and complex, contain different elements and cannot be described by syllable and echeme only. The uninterrupted sequence of sounds lasting typically many seconds and which is answered by the female is called main song. Its final part - where typically the female responds – is called coda (a term from music, Italian, meaning tail). Isolated syllables (or short echemes) which are produced in intervals of a few to many seconds are onomatopoeically called ticks (correspondingly ticking song). Each syllable consists of impulses: a simple, undivided, transient train of sound waves (here: the damped sound impulse arising as the effect of one tooth of the stridulatory file).

The Dioncomena species under investigation can be classified into three groups based on their color pattern, habitat, and ecology. The first group is the jagoi-group, characterized by a green and black color pattern with turquoise colors in males (D. jagoi, D. magombera sp. nov., D. flavoviridis sp. nov., D. sanje sp. nov., and probably D. ugandana sp. nov., Figs 1, 10, 11). Initially, individuals resembling D. jagoi were collected from forest areas in Tanzania, but further examination of their morphology and songs revealed that D. jagoi is only found in the East Usambara Mountains and Kimboza Forest Reserve. Similar species collected from the Nguru and Udzungwa Mountains and also Kimboza Forest Reserve were identified as distinct species based on differences in their songs and morphology, particularly in their stridulatory files and color patterns. Therefore, we describe three new species in the D. jagoi-group: D. sanje sp. nov., D. magombera sp. nov., and D. flavoviridis sp. nov. These species are common in the Udzungwa Mountains and along the Eastern Arc chain, with D. flavoviridis sp. nov. found in lowland wet and submontane forests of the Nguru Mountains. These species are found in low densities scattered in closed forest in Uganda and lowland to submontane forests along the Eastern Arc chain in Tanzania (see map Fig. 36).

The ornata-group comprises three species: D. ornata found in southern Kenya and northern to central Tanzania, D. zernyi inhabiting lowland to submontane forested areas in southern Tanzania, and D. grandis distributed in the border region of Mozambique with Zambia. None of the three species were found to occur together (refer to map Fig. 36). Male D. scutellata’s color pattern indicates its possible relation to D. zernyi, though its live color pattern is unknown as one of the two known specimens has lost its color due to preservation in alcohol (stored in Copenhagen), and the other is preserved in the collection of London. All four species have a three-parted stridulatory file.

D. ornata populations on the mainland exhibit distinct coloration, with black wings and turquoise areas on the head, face, and dorsal abdomen in males. In contrast, the Zanzibar population lacks turquoise coloration, but males have a vivid yellow, yellow-red to red pronotum, similar to females (see Hemp 2021 for images of live specimens). Females of mainland populations of D. ornata are less brightly colored, lacking the conspicuous turquoise color typical of males, and their tegmina are often more brown or light black, but not as deep black as males. The body of D. ornata is uniformly colored green or yellow-green, without lateral fasciae, as is typical for the jagoi-group. A dorsal black or light black fascia is usually present. D. zernyi males and females are less conspicuously colored, with brown tegmina and a green body with white and/or yellow fasciae, and both sexes show no differences in coloration (Fig. 16). D. grandis has a color pattern very similar to mainland D. ornata. Like D. ornata, this species produces tick-songs (depicted in Naskrecki and Guta 2019), but its full song is not known. Naskrecki and Guta (2019) state that D. grandis is relatively abundant on low herbaceous vegetation along forest edges and thus is classified in the ornata-group, occupying a similar habitat. Species of the ornata-group are found in forest edges and clearings, often aggregated with several males and females sitting on bushes and tall herb vegetation together (this behavior is known for D. ornata and D. zernyi only).

Species of the tanneri-group, consisting of D. tanneri, D. versicolor sp. nov., and D. ngurumontana sp. nov., are found in forests of the montane zones along the Eastern Arc Mountains. All three species are highly colorful, with patterns of turquoise, yellow, red to red-brown, and black, especially in males (Figs 12, 15). Like members of the ornata-group, individuals of all three species are found along forest edges or on forest clearings, and are often aggregated.

All studied species of Dioncomena produced long lasting calling songs, often with durations between one to several minutes (see Fig. 19). These songs differ from the well-known continuous and long lasting tettigoniid songs because they consist of several different elements following each other in a more or less fixed sequence and they end in a coda during or after which a female ready to mate typically responds.

Figure 19. 

Oscillograms of male calling songs of Dioncomena species. Overview (150-s-sections).

Here at first the songs of the species of the D. jagoi-group are described (in alphabetical order: flavoviridis sp. nov., jagoi, magombera sp. nov., sanje sp. nov.), then those of ornata-group (ornata, zernyi) and finally those of the tanneri-group (tanneri, versicolor sp. nov., and ngurumontana sp. nov.).

D. flavoviridis sp. nov.

The main part of the calling song (Fig. 19; Suppl. material 1) starts with a 25–30 s long series of relatively uniform syllables. At the beginning, the duration of these bi-parted syllables is around 40 ms (SRR 12 Hz), but increases towards the end to 55 ms with a simultaneous reduction of the syllable intervals (SRR 17 Hz). Now the amplitude modulation changes and single long (140 m) syllables are introduced in a rhythm of about 1 Hz (Fig. 20A). Shortly afterwards short (200–400 ms) silent intervals appear, always before every second long syllable. At the beginning the long syllables are sometimes produced at a lower rate and before each gap (Fig. 21). The (3–5) syllables before the gap have a much higher amplitude (typically > 6 dB) than the song before (Fig. 25). This pattern is produced for another 30 s nearly until the end of the main part. Here often a few soft syllable groups are observed.

Figure 20. 

Oscillograms of male calling song of Dioncomena flavoviridis sp. nov. (A, B) and D. jagoi (C). A. Change of amplitude pattern in first half of the main part; B. Transition from ticking song to ‚slow‘ song; C. ‚Slow‘ song.

Figure 21. 

Oscillograms of male calling songs of Dioncomena species. Central portion of main part of song; 20-s-sections.

As long as the male does not hear a female response it produces this main part at intervals of many minutes. In the mean time long sequences of other song parts can be heard, called here ‚ticking song‘ and ‚slow song‘. Some time after the end of a main part, typically very short, isolated sound elements can be heard, following each other at intervals of a few seconds – the ticking song (Fig. 22A). These elements consist mainly of an isolated impulse, sometimes accompanied by a second softer and a series of very much softer impulses. After some time they are combined with longer syllables, resulting in the ‚slow song‘. Its basic units are echemes repeated at a rate of about 0.5 Hz (Fig. 20B). Each echeme is made up of 10–15 short syllables, the last being an isolated impulse, and is separated from the next by 0.3–0.5 s. The short syllables (25–30 ms; ca. 20 impulses) follow each other with 5 Hz. This ‚slow song‘ can change to the next main part. During long lasting tick sequences also isolated impulses combined with an impulse series are observed (Fig. 22B). Two males singing together can alternate (Fig. 22C).

Figure 22. 

Oscillograms of male tick songs of Dioncomena species. A–C. D. flavoviridis; A, B. ticks; C. alternating males; D. D. jagoi; E. D. grandis (from Naskrecki and Guta 2019); F, G. D. magombera sp. nov.; H, I. D. sanje sp. nov.; J–L. D. ornata; M. D. zernyi; N. D. tanneri; O. D. versicolor sp. nov.; P. D. ngurumontana sp. nov.

A female ready to mate responds one to many times during the gaps which follow the high-amplitude syllables of the main part of the male song (Figs 23, 24; Suppl. material 2). Its response consists of one to several impulses, often one louder than the others. The delay of the response is about 200 ms, measured from the end of the last male syllable (Fig. 25).

Figure 23. 

Oscillograms of male-female-duets in Dioncomena species. Overview (150-s-sections).

Figure 24. 

Oscillograms of male-female-duets in Dioncomena species. Detail (20-s-sections).

Figure 25. 

Oscillograms of male-female-duets in Diocomena species. Detail (3-s-sections).

D. jagoi²

The song of D. jagoi has some similarities to that of D. flavoviridis sp. nov. The main part of the calling song (Fig. 19; Suppl. material 3) starts with a ca. 60 s long series of relatively uniform syllables which show, however, some variability in amplitude modulation and rhythm resulting in an irregular echeme-like pattern. The duration of these bi-parted syllables varies between 50 and 60 ms (SRR 12–17 Hz; Figs 20C, 26A), but does not change with time, while the amplitude increases slowly and continuously. The series ends abruptly (or with a transition phase of some seconds) and 0.5 s long echemes appear, separated by intervals of about 1 s. They consist of 10–12 syllables (SRR ca. 25 Hz) with increasing amplitude (Fig. 26A). At the beginning of this final part (duration 30–60 s) some short echemes are interspersed (or they form the transition part exclusively).

Directly before the main part often a ‚slow song’ is observed (see D. flavoviridis sp. nov.). Its basic units are echemes repeated at a rate of about 0.5 Hz (Fig. 20). Each echeme is made up of 10–15 short syllables, the last being an isolated impulse, and is separated from the next by 0.5–0.8 s. The short syllables (40–50 ms) follow each other with 8 Hz. D. jagoi produces also ‚ticking songs‘ (Fig. 22D) as D. flavoviridis sp. nov. However, the ticks are not isolated impulses but nearly always short echemes (duration < 100 ms).

D. magombera sp. nov.

The song of D. magombera sp. nov. is even more complicated than that of D. flavoviridis sp. nov. and D. jagoi. Typically, it consists of a sequence of three large parts (Fig. 19; Suppl. material 14) from which the last two may correspond to the main part in D. flavoviridis sp. nov. In the following, they are described using this name. Female sound recordings are not available. This so-called main part starts with a long (30 s), homogeneous series of short echemes (ca. 100 ms; ERR 4 Hz) containing 4–5 syllables each (SRR 30 Hz) (Fig. 26B). Each syllable is made of one large impulse and a series of softer ones its number increasing during the echeme. At the beginning of the second half of the main part these short echemes are interspersed by longer (1.5 s) and more complex combination of syllables (Fig. 26C). Each combination starts with syllables similar to that in the short echemes but then the number of soft impulses becomes larger, the single loud impulse disappears and the rhythm of the now bi-parted syllables becomes faster (50 Hz). With time, this core syllable combination becomes shorter but at the beginning and end other syllables are added (Fig. 26D) until the main part ends with a few short echemes. The first part of the song – before the so-called main part – contains long syllable combinations quite similar to that immediately after the homogenous echeme series. Sometimes – in contact with other males? – some three-part sequences follow immediately one after another. Occasionally the males produce, ticking songs’ with ticks similar (Fig. 22G) or slightly different (Fig. 22F) as in D. jagoi.

Figure 26. 

Oscillograms of male calling songs of Dioncomena species. Detail (3-s-sections). A. D. jagoi; B–D. D. magombera sp. nov.; E, F. D. zernyi; G, H. D. tanneri; I, J. D. versicolor sp. nov.

D. sanje sp. nov.³

The males of D. sanje sp. nov. produce very long uninterrupted sequences. During two hours of recording, for example, an isolated male made two sequences each lasting more than 20 min. These long songs consisted of two parts (section in Fig. 19; Suppl. material 8) each taking several minutes and each consisting of syllable groups followed by one or several impulses. At the beginning one type of combination (ca. 16 syllables with a duration of 30 ms each, SRR 15 Hz) is repeated in a rhythm of 0.7 Hz, but at the end it becomes shorter (1 Hz) while the syllable duration (38 ms) and SRR (19 Hz) slightly increase. Each syllable is separated from the next by a distinct gap of more than 10 ms (and a soft hemisyllable; Fig. 27A). Therefore these sound combinations are called ‚gappy’. Each series is followed by a loud, short, tick-like syllable. After a somewhat irregular transition period of ca 10 s another syllable combination is heard, produced a little bit faster (1.25 Hz) (Fig. 21). Here both hemisyllables are similar in amplitude leaving no large gaps in the syllable group (Fig. 27B). This combination is called ‚dense’. It is followed by a single syllable of a different type which varies in amplitude between softer and louder than the preceding group (compare Fig. 21 and Fig. 27B) and which ends with a few loud isolated impulses. The syllable is called high frequency (HF) syllable, because the carrier frequencies of the first part are much higher than in all other song parts (Fig. 29; only very weak in Fig. 25 due to the restricted frequency range of this recording; Suppl. material 9). Outside the long uninterrupted sequences the males produce long (1–3 min) series of regularly spaced elements of a ticking song (Fig. 22H). Occasionally also short sequences of gappy sound combinations are observed (Fig. 22I).

Figure 27. 

Oscillograms of stridulatory movement and song [synchronous registration of left tegmen movement (red; upward deflection represents opening, downward closing) and sound (black)] in D. sanje sp. nov. A. Gappy sound combination; B. Dense sound combination with HF-syllable (*).

The females respond preferentially and regularly to the dense sound combinations (Figs 23, 24; Suppl. material 10). Their sounds follow about 50–70 ms after the end of the male impulse group (Fig. 25). If the male switches ‚back’ to the gappy pattern, the females stop responding. However, they react acoustically sometimes also to the gappy combinations.

D. ornata

The main part of the song of D. ornata contains two long parts (Fig. 19; Suppl. material 6). It begins with a long homogeneous series of hemisyllables (SRR 14–25 Hz (-30 Hz at 26 °C), duration 15–150 s; Fig. 28A). Only one male was found to produce bi-parted syllables (Fig. 28B). The directly following part consists of a characteristic short combination of a few sound elements (Fig. 28C) which is repeated for many seconds (repetition rate 2.3–2.6 Hz; durations up to 110 s observed). How these sound elements are produced is unclear and not all elements are found in every recording. Possibly each group results from one stridulatory movement cycle (syllable) using all parts of the complicated file (Fig. 2). Sometimes a series of syllables resembling that from the first half of the main part are found isolated without the second half of the main part. Between the main parts often single ticks or occasionally even long series of ticks can be heard, rarely immediately before a main part. These short sounds often consist of pairs of syllables (Fig. 22K), but more complicated combinations are also observed (Fig. 22J, L).

Figure 28. 

Oscillograms of male calling songs of Dioncomena species. Detail (1-s-sections). A–C. D. ornata: A. Section of first part; B. Section of first part in atypical male; C. Section of second part; D. D. tanneri; E, F. D. ngurumontana sp. nov.: E. Change from fast to slow rhythm; F. ‚coda‘.

The females respond only during the second half of the main song (Figs 23, 24; Suppl. material 7). Typically, the responses are registered 50–150 ms after the end of the longest part of the short combination (Fig. 25).

D. zernyi

From D. zernyi only few recordings are available. According to the data, the main part of the song (uninterrupted part) is relatively short (Figs 19, 21; Suppl. material 13). It starts with a series of hemisyllables with an irregular but slowly increasing SRR (e.g., 2 Hz to 15 Hz in 10 s). During the next few seconds the syllables become bi-parted until the pattern changes into a sequence of short isolated syllables (or echemes?) (SRR 5 Hz; Fig. 26E). This part lasts about 5 s. Then the singing continues with a kind of ticking song with elements separated by intervals of about 3–6 s (Fig. 22). This ticking song is sometimes heard for hours without the main part. Before the main part often the male produces short echeme sequences (Fig. 26F).

D. tanneri

Also from D. tanneri only few recordings are available. In structure, the song has some similarities to that of D. versicolor sp. nov. In the longest recording of the main part, the song starts with several combinations consisting of two different types of syllables (Figs 19, 21; Suppl. material 11). A series of about 30 syllables (SRR 14 Hz) each with ca. 12 impulses is followed by a single short syllable (1–2 impulses). After a few of these combinations one distinctly longer one of the same type is heard. Then a long series of syllables (18 s) is observed, which is relatively homogeneously in rhythm but shows changes in amplitude (Fig. 28F). After its end a short (2 s) series of bi-parted syllables (SRR 25 Hz) is produced followed by some short isolated syllables (or echemes?) (SRR 6 Hz) (Fig. 26G). These combination of two different types of echemes is repeated several times with varying durations. At the end the isolated syllables become longer (Fig. 26H) and are followed at first by one or two impulses which finally disappear.

D. ngurumontana sp. nov.

An isolated male of D. ngurumontana sp. nov. produced the main part of its song typically at intervals of many minutes. It often lasted about one minute and consisted of two parts (Fig. 19; Suppl. material 4). The song started with a long (ca. 50 s), homogeneous series of syllables (SRR 7 Hz) which was – after some irregular echemes – followed by a much shorter final part (Fig. 21). This ‚coda‘ began with a series of bi-parted syllables (SRR 28 Hz) and closed with a short (<= 1 s) sequence of hemisyllables slightly decreasing in amplitude and SRR (25 to 10 Hz; Fig. 28E). In between the main parts single tick elements were observed. If a male is or has been in acoustical contact with a female it often produced much longer (up to 5.5 min documented) songs. While the coda remained similar in structure, the first part became much longer and the SRR changed often. Fast series with 15–20 Hz alternated with 6 Hz parts (Fig. 28F), both lasting several to many seconds (5–60 s).

A female ready to mate responds with a series of impulse-like syllables after the male coda (Fig. 23; Suppl. material 5). Sometimes the responses overlap with the latest syllables of the male but there was no short-time relation to any male song parts as in the other Dioncomena species (Figs 24, 25). The response series sometimes lasted nearly one second.

D. versicolor sp. nov.

Also from D. versicolor sp. nov. only few recordings are available. According to the data, the main part of the song (uninterrupted part) is relatively short (Figs 19, 21; Suppl. material 12). Before this part, the song consists of 1–2 s long series of bi-parted syllables (SRR 20–28 Hz) crescendoing slightly in amplitude (Fig. 26J). These echemes are produced continuously and with silent intervals (ERR ca. 0.4 Hz) for about a minute. Then the male starts to fill the intervals with a series of isolated syllables (at first syllable period 65 ms). During the last two ‚intervals‘, these syllables (or echemes?) become longer, louder and more widely spaced (SRR 7 Hz; (Fig. 26I) 1), until after a very long (ca. 4 s) echeme the song ends. Outside this main part the male produces also a ‚ticking song’. It can be heard for quite a long time and consists often of two syllables (Fig. 22O).

The areas containing the stridulatory organs differ clearly in structure between the left and the right tegmen. The part of the left tegmen which carries the stridulatory file at its lower side is non-translucent and shows the typical species-specific coloration like the other parts of the tegmen. In the corresponding part of the right tegmen in all species one large translucent mirror cell is found (Fig. 29A). Its frontal edge (right edge in resting animal) is distinctly elevated above the mirror plane.

The stridulatory files in Dioncomena species differ widely in tooth number, length and structure (Fig. 2, Table 1). The lowest tooth numbers are found in D. ngurumontana sp. nov. and D. nitens, followed by a broad range with many species similar in number and two species (D. ornata and D. grandis) with high numbers. Five species (flavoviridis sp. nov., jagoi, sanje sp. nov., grandis, scutellata) have longer files than the others. The variation in complexity does not seem to be correlated with any other file characteristic. Several species (grandis, scutellata, ornata, zernyi; see Fig. 2) have very unusual tri-parted files with three regions differing in file density, but even in these four species length and tooth number of the different parts vary distinctly.

The stridulatory organs of females are typical for Phaneropteridae (see e.g. Heller et al. 1997); several to many short rows of teeth on cross veins in the upper side of the right tegmen are present, while the scraper is formed by the amplified inner edge of the left tegmen.

As to be expected for songs consisting of damped impulses (non-resonant song), in most species studied the spectra are relatively broad-banded with the carrier frequencies showing the highest amplitudes situated between 20 and 30 kHz (Fig. 30). Checking sonograms for spectral differences, often in some components (hemisyllables etc.) the higher frequencies (partly harmonics) are stronger than in others parts of the song (for examples see Fig. 28; details see legend). Only in D. sanje sp. nov. one special type of hemisyllable was registered with the main part having its maximum at 60 kHz (Fig. 30; green), called HF-syllable (high frequency). The syllables before and the following impulses are much lower in frequency (Fig. 31). When studied by laser Doppler vibrometry, the single mirror cell shows a narrow peak around 40 kHz (Fig. 32; n = 2 specimens: 33 kHz, 46 kHz), demonstrating a distinct mismatch of natural frequency of the mirror and the peak frequencies of the either of the call types. The left wing is completely damped and probably not involved in sound radiation.

Figure 29. 

Dioncomena flavoviridis sp. nov. A. Base of male left tegmen; B. Base of male right tegmen; C. Base of female left tegmen; D. Base of female right tegmen. Scale bar: 5 mm.

Figure 30. 

Power spectra of the male calling songs of the Dioncomena species studied. Asteriks in the respective color mark the positions of the analysed sounds, for D. flavoviridis sp. nov. in Fig. 23, for D. jagoi in Fig. 20, for D. magombera sp. nov. in Fig. 25, for D. sanje sp. nov. see Figs 26, 29 (dense combination: blue and violett, high frequency syllable: green, gappy combination: red), for D. ornata in Fig. 27, for D. zernyi in Fig. 25E (song), Fig. 21M (ticks), for D. tanneri in Fig. 25, for D. versicolor sp. nov. in Fig. 25, for D. ngurumontana sp. nov. in Fig. 23.

Figure 31. 

Sonogram of second half of the dense sound combination of Dioncomena sanje sp. nov. with HF-syllable.

Figure 32. 

Forewing vibrational response in Dioncomena sanje sp. nov. A. Displacement of forewings in response to acoustic stimulus. The left wing is completely damped and probably not involved in sound radiation B. Deflection through phases of one oscillation C. Top panel: velocity gain of the mirror of the right wing as recorded through laser Doppler vibrometry. Middle panel: relative intensity of a section of the ‚normal‘ call. Bottom panel: relative intensity of the HF-syllable. Note the mismatch of natural frequency of the mirror and the peak frequencies of the either of the call types.

After the observation that in D. sanje sp. nov. one syllable type has a different spectral composition from the others, we began to analyze the stridulatory movements in this species (Fig. 27). Interestingly, in the gappy as well as in the dense sound combination the most intense song parts are produced during opening movements of the tegmina. In the gappy combination the opening hemisyllables are much louder than the closing ones, and also the tick-like syllable of this combination is produced during an opening movement (Fig. 27A). In the dense combination the differences in amplitude between opening and closing hemisyllable are much smaller. The HF-syllable is also produced during an opening movement. Here the amplitude of the movement is larger than in the other syllables of the combination and the speed of the movement slower, but always the same part of the file seems to be used. Even the few loud isolated impulses are produced during the opening movement of the tegmina but with the teeth situated farer away from the articulation than that used for the other parts of the combination.

The tympanic membranes near to the auditory sensillae in the fore tibia are exposed, naked, and show no obvious modifications. However, the major input to the hearing organs is via the auditory spiracles in the prothorax, especially for higher sound frequencies (reviewed e.g. by Strauß 2019; Römer 2020). The large acoustic spiracles are permanently open, and continue into the prothoracic auditory bullae, the first parts of the auditory (acoustic) trachea (Bailey 1990). The outer parts of auditory trachea run without obvious change in diameter from the bullae to the sensory organs in the tibiae (Fig. 33). In Dioncomena, these bullae are very large – as in many phaneropterids -, but even the space before the openings is enormously enlarged due to the inflated paranota (Figs 34, 35) in many species. This structure is called prebulla. These prebullae show an impressive size variation in Dioncomena, also depending on the sex (Fig. 35, Table 2). While this variation is easily recognizable with the naked eye, it is much more difficult to quantify. Calculating relative pronotal width (overall width / overall length) gives only a first clue (Table 2), the scatter probably resulting from differently shaped pronota. Combining visual impression and measurements, prebullae are most distinct in the males of the D. jagoi-group, D. grandis and D. nitens. In females, only D. flavoviridis sp. nov. approaches or even surpasses the size of the male prebulla (Fig. 35A). The males of the other species have smaller prebullae, while they are completely missing in the females of some species.

Figure 33. 

Outer parts of the auditory tracheae in D. sanje sp. nov. A–C. Female auditory tracheae in lateral (A), dorsal (B), and anterior (C) axes; D–F. Male auditory tracheae in lateral (D), dorsal (E), and anterior (F) axes; G. Male head and thorax (sagital-section) showing membrane which connects the two bullae; H. Male thorax (cross-section) at the point of the membrane in G.

Figure 34. 

Auditory tracheae in pronotum of male Dioncomena sanje sp. nov. from behind, abdomen removed.

Figure 35. 

Dorsal view of pronotum in Dioncomena and related genera, indicating different degrees of widening. A. D. flavoviridis sp. nov.; B. D. sanje sp. nov.; C. D. ornata, Amani; D. D. ornata, Jozani, Zanzibar; E. Kefalia laeta Hemp, 2017; F. Meruterrana elegans Sjöstedt, 1912; G. Atlasacris brevipennis Massa, 2015 (photo OSF); 1 male, 2 female.

However, not only the interspecific, but also the intraspecific variability may be quite large as can be seen in the data of D. ornata (Table 2, Fig. 35C, D), the only species with a relatively large number of populations studied.

Surprisingly, prebullae were discovered in a series of other East Africa phaneropterids, but always in males only. They are very pronounced in all species of the genus Kefalia which is morphologically similar to Dioncomena (Table 2, Fig. 35E). However, prebullae were found also in genera with uncertain relationship [Merumontana Sjöstedt, 1912 (Table 2, Fig. 35E) and in some species of Ivensia Bolivar, 1890] or even placed in other tribes [Atlasacris Rehn, 1914 (Table 2, Fig. 35G)]. Some other species were also checked and no prebullae were found [from OSF: Arostratum oblitum Massa, 2015 ♂, ♀, Pronomapyga grandis Rehn, 1914 ♂, ♀, Pro. graueri Rehn, 1914 ♂, ♀, Gabonella cothurnata (Bolívar, 1906) ♂, ♀, Poecilogramma annulifemur Karsch, 1887 ♂, ♀, Poe. cloetensi (Griffini, 1908) ♂, ♀), Poe. striatifemur Karsch, 1887 ♂, ♀, Pardalota asymmetrica Karsch, 1896 ♂, ♀, P. haasi Griffini, 1908 ♂, ♀, P. karschiana Enderlein, 1907 ♂, ♀, P. reimeri La Baume, 1911 ♂, ♀, P. superba Sjöstedt, 1913 ♂, ♀, P. versicolor Brunner von Wattenwyl, 1878 ♂, ♀; from specimens in Collectio Hemp: Odonturoides hanangensis Hemp, 2009 ♂, ♀, O. insolitus Ragge, 1980 ♂, ♀, Monticolaria meruensis Sjöstedt, 1910 ♂, ♀, M. manyara Sjöstedt, 1910 ♂, ♀, M. kilimandjarica Sjöstedt, 1910 ♂, ♀].

The two bullae of the Dioncomena are also coupled via a membrane or ‘septum’ in males but not in females (Fig. 33A–F). The surface area of this connection in one micro-CT scanned male of D. sanje sp. nov. is ~ 0.56 mm² (Fig. 33G, H).

Among bush-crickets, phaneropterines are well-known for their complicated songs. Their sounds may contain up to five syllable types and more than 1000 syllables per song unit (see Heller et al. 2015). However, species with these extremely complicated songs are not common. Species with more than 200 syllables per song unit, for example, are known only from six genera worldwide. Most Dioncomena species are above this value. They all have complex songs, strongly differing between the species (Fig. 18). Most of them have clearly structured songs with a starting, a middle and an end sequence, only D. sanje produces a continuous song lasting sometimes many minutes with two parts alternating. All species are diurnal, i.e. their songs are heard only during daytime (and sometimes at dusk). However, regular tick sounds – often for hours – were recorded in some species also at night (e.g. D. jagoi, D. zernyi, D. ornata).

Females respond very precisely at certain points of the male song (Figs 23–25) – an indication that these duets are very important for mate localization. However, what happens really before the sexes meet? Unfortunately, data from the behavior in the field do not exist. In the lab and without female contact males were observed to sit at one place for the whole day and sing. One could assume that the females approach a singing male silently – by day and night? – and start to respond (during daytime) when his song is loud enough and sufficient for her standards. In such a scenario the male ticking song could be used for a rough mate localization by the female. The last steps before the contact would have to be done by the male (strategy 3 in Spooner 1995).

During the production of the complicated songs the sounds produced during opening movements of the tegmina are louder than those during the closing movements. This was demonstrated by the recording of the stridulatory movement in D. sanje sp. nov., where the closing movement can sometimes even be completely mute. In the other species the sound production is not known in detail, but the structure of the file teeth suggest the same mechanism: the teeth are often curved with the tip towards the articulation (e.g., Fig. 2H) and/or the steep side of the teeth faces this direction (e.g., Fig. 2B). Sound production by opening wing movements (called ‚reverse stridulation‘ by Montealegre-Z 2012) is rare among Tettigonioidea and among phaneropterines documented only for Phaneroptera spp. Serville, 1831 and Monticolaria kilimandjarica (see Heller et al. 2015). The stridulatory files in Dioncomena are quite diverse. However, in no species lumps (or wing stoppers) as in Ischnomela Stål, 1873 (Montealegre-Z 2012) or a bend in the file like in Phaneroptera (Heller et al. 2021) are found, which were interpreted as a protection against „complete separation of the wings and subsequent erroneous wing overlapping during the ongoing cycle“ (Montealegre-Z 2012). However, another character of reverse stridulation, tooth spacing gradually increasing towards the anal margin of the wing (Montealegre-Z 2012), can be observed in several species, although it is not necessary for a pure-tone sound not found in Dioncomena. Within the genus, the number of stridulatory teeth varies between 30 and 100. More surprising, not only the number of teeth differs by a factor of three, but also their arrangement shows huge differences. In Dioncomena flavoviridis sp. nov. the file bears ca. 50 uniformly spaced teeth (Fig. 2B), in D. jagoi and D. magombera sp. nov. the basal end of the file differs distinctly from the rest (Fig. 2A, C), and in the three species D. ornata, D. grandis and D. zernyi the file is divided into three parts of very different tooth density (Fig. 2E–G). Phaneropterines are well known for complicated files and in some cases the function of the file modification is understood (e.g., genus Acrometopa Fieber, 1853; Hemp et al. 2018). In Dioncomena, however, the situation is more complicated. From the amplitude pattern alone the use of different file parts cannot be safely predicted. The stridulatory movement is known only in D. sanje sp. nov., a species with a relatively simple file. Nevertheless, using this file the males are able to produce syllable types with a widely different spectral composition (see Fig. 30).

Except D. sanje sp. nov., most sounds produced by investigated Dioncomena males were relatively similar in the spectral composition (Fig. 30). The broad main peak was nearly always situated around 20 kHz. Some song parts had ultrasonic components in addition, only in D. zernyi and D. ngurumontana sp. nov. very broad-banded sounds were observed with peaks around 45 kHz. D. sanje sp. nov. males, however, produced a special type of syllables at certain points of their song. These HF-syllables have a peak at 60 kHz and only weak components with lower frequencies (Fig. 30, green curve; Fig. 31). They are placed after the dense sound combination and before the loud, short ticking sound which may trigger the female response. Possibly the HF-syllable helps the female to estimate the distance to the singing male as hypothesized for the HF-syllables in Phaneroptera (Heller et al. 2021). Surprisingly, the HF-syllables seem to be produced with exactly the same part of the file as the dense sound combination (Fig. 27B) which is much lower in frequency. In the phaneropterine Gonatoxia helleri Hemp, 2016 the males sing with different frequencies very likely by using differently structured parts of the file (Heller and Hemp 2017). The different frequencies are also not easily explained by the tegmen morphology, since as in D. flavoviridis sp. nov. (Fig. 29) and all Dioncomena species studied so far, the right tegmen bears only one large mirror cell. The mismatch of the natural frequency of the mirror and the peak frequencies of the either of the call types (in D. sanje sp. nov.) suggests that the apparent sharp tuning of the mirror is not being exploited during stridulation (i.e. matched to the tooth strike rate): a case of non-resonant stridulation. The observed tuning peak in the mirror may be the natural response of a membrane which acts to radiate all and any input frequencies or a vestigial trait of a resonantly calling ancestor.

Sounds at 20–30 kHz are certainly easily perceived by the animals since they are in the typical frequency range of tettigoniid signals. At higher frequencies the ears often become less sensitive. However, there are exceptions where the hearing threshold falls continuously until at least 40 kHz (see Poecilimon elegans Brunner von Wattenwyl, 1878 in Strauß et al. 2014). These special adaptions cannot be recognized from morphology and have to be studied neurophysiologically.

In some Dioncomena species (and some other species; see Table 2) the ears show unusual modifications, but they are probably not related to improved frequency sensitivity. The inflation of the pronotal lobes, the prebullae, have most likely evolved to increase the general auditory sensitivity and/or the directionality of the ears. The physiological function of these ears having huge auditory tracheae with left and right one touching each other are not well understood. A first detailed study was done by Celiker et al. (2022), but they could not model the cross-talk between left and right ear. That this connectivity exists – similar to the situation in field crickets (Michelsen and Löhe 1995) – was already described in 2011 by Mohsin and Schul in a conference presentation. Improving sensitivity and/or the directionality would easily explain why mostly the males possess prebullae: to detect the short, rare and probably faint female signals.

But why are these structures not found in all Dioncomena species? Their presence in the closely related Kefalia (Grzywacz et al. in prep.) indicates that they evolved in the common ancestor of both, if not earlier. At least in Dioncomena the size of the lateral lobes is strongly correlated with the behavior and ecology of the species, i. e. the population density. The largest inflations of the pronotal lobes are found in species living scattered within closed forest while species living on clearings and along forests edges often found in clusters have less or not inflated pronotal lobes, especially in the females. Obviously the costs of such ears are not negligible (see Stumpner and Heller 1992; Lehmann 2012; Strauß et al. 2014) and the hearing sensitivity evolves towards an optimum but not a maximum. Unfortunately, hard data for this point are nearly completely missing (see Helversen et al. 2012) but from the variability seen in and between populations of D. ornata one could assume that here costs and benefits are nearly balanced.