Look closely and listen carefully: unexpected cicada diversity in northern Sardinia, with the description of a new species (Cicadidae: Tibicina)

Abstract

Integrative taxonomy combines different data sources as a way to detect separately evolving metapopulation lineages or species. This method is applied to cicada populations found in northern Sardinia, based on acoustic, morphological and ecological data. Thereby, the genus Tibicina turned out to be more diverse than previously expected. Besides the anticipated Tibicina corsica corsica and T. nigronervosa, both scientifically described from the neighbouring island of Corsica, two poorly or previously unknown species have been discovered. Tibicina longisyllaba sp. nov. is here described as new. Tibicina corsica s.l. forms a small species complex where syllable periods of the song – the movement cycle of the acoustic organs – are specific without overlaps. Some coloration and hair cover characters, as well as the shape of the song apparatus, are not diagnostic on their own, but highly significant between species. Species occur sympatrically, parapatrically or allopatrically and prefer specific habitats from grassland to closed forests. All four Tibicina taxa are endemic to parts of Sardinia or to the Corso-Sardinian archipelago. The new findings are important for biodiversity conservation and public awareness. The loud, strident calling songs dominate the summer sounds of the islands. The group could form a showcase with respect to biogeography, phylogeny, speciation and sound production.

adaptive radiation, bioacoustics, climatic fluctuations, colonization, cryptic species, endemism, habitat specialization, Italy, specific mate recognition system

INTRODUCTION

The conservation of biodiversity can only succeed when the organisms of our planet are well known. In the last decades, new methods and equipment have enabled novel approaches to the study of biodiversity, which have sometimes resulted in the discovery of unknown cryptic or sibling species. Multiple sources of data in combination are essential for species delimitation and classification of difficult taxonomic groups (e.g. Schlick-Steiner et al., 2010). Here, I investigate hidden cicada taxa using an integrative approach incorporating acoustics, morphology, habitat requirements and distribution. Cicadas (Cicadidae) are among the most popular and best-liked insects. This is especially true in the Mediterranean Basin where some charismatic, loud-singing species can be found. Surprisingly, the group is scientifically poorly investigated. Almost 20 species among 70 have been described for Europe during the 21^st century alone (Gogala, 2019). Many of these new species are relatively small and belong to complexes of closely related species. The most frequently discussed example is Cicadetta montana (Scopoli, 1772) s.l., which consists now of more than a dozen cryptic species (e.g. Gogala & Trilar, 2004; Sueur & Puissant, 2007; Hertach et al., 2015, 2016). While morphology of these species is similar, song patterns are diverse and some of these songs have evolved to high complexity (Gogala et al., 2008; Hertach, 2011). Songs are species-specific in many acoustically active animals like birds, amphibians, grasshoppers and cicadas. They form the first, long-range step in pair formation (Sueur & Aubin, 2003). In cicadas, songs must be inherited and cannot be learnt since the adults are temporally and spatially separated from the long larval stages (Marshall et al., 2011). These insects are excellent short-distance flyers but have quickly tiring flight muscles, and the limited duration of the adult stage, of maximally a few weeks, hinders long-range dispersal, resulting in high phylogeographic structure (e.g. Buckley et al., 2001).

Fieldwork leading to this study revealed hidden diversity related to Tibicina corsica (Rambur, 1840) in the core of the Mediterranean, on the Italian island of Sardinia. Sardinia is the second largest island in the Mediterranean Sea, neighboured by Corsica to the north and includes a high number of smaller islands. The archipelago is known for its old age of 20 to 29 Mya (Alvarez, 1972; Cherchi & Montadert, 1982; Carmignani et al., 1995), when a microplate started to separate from the Iberian continental landmass. However, some authors argue that a narrow land-bridge remained until 7 to 8 Mya (Meulenkamp & Sissingh, 2003). In any case, the Messinian salinity crisis again offered migration paths for thermophilous organisms, when the Mediterranean Sea almost desiccated (5.6 to 5.3 Mya; Krijgsman et al., 1999; Ketmaier & Caccone, 2013). During the Quartenary ice ages, Sardinia and Corsica were sometimes connected to each other and also to the Tuscan archipelago due to sea-level fluctuations. Long separation, large size and a complex geology and orography led to a high degree of 6.7% exclusively endemic animal species in Sardinia (Minelli et al., 2006). Médail & Quézel (1999) list Sardinia and Corsica as one of the ten ‘Mediterranean hotspots of plant diversity and endemism’ where narrow endemism surpasses 10%. Phylogenetic studies used the Sardinian history to calibrate molecular clock rates (e.g. Caccone et al., 1994; Ketmaier & Caccone, 2013).

The seven European species of the genus Tibicina are of medium to large size with wingspreads from 50 to 90 mm. Described taxa differ in coloration of the wing venation and of the body with characteristic light spots on a dark thorax, as well as for the shape of the genitalia and body size. However, the songs – produced by a pair of so-called timbals – are simple and almost identical to human ears (Sueur & Aubin, 2003): monotonous buzzes lasting for several minutes that only differ in timbre. Some species sing more softly (T. nigronervosa Fieber, 1876), others more stridently [T. quadrisignata (Hagen, 1855)] and some own an easily audible fibrillation [T. steveni (Krynicki, 1837)]. An exception is T. haematodes (Scopoli, 1763), which starts with introductory chirps and already ends after 10 to 20 seconds. The low degree of acoustic divergence makes the genus an attractive model for testing the role of ecological and ethological characters in speciation (Sueur et al., 2007). A similar but much more species-rich position is taken up by the North American genus Okanagana (Chatfield-Taylor & Cole, 2019), which is closely related to the Palaearctic Tibicina (Sueur et al., 2007). Sueur & Aubin (2003) have demonstrated a simple method, on the basis of digital recordings, to distinguish the songs according to frequency bands and the number of syllables produced per second. Syllables are the complete cycle of both timbals, which are mandatorily moved one after the other with a short, temporal overlap in Tibicina. Syllable periods are astonishingly constant within species boundaries and hold up, for example, in T. steveni and T. quadrisignata over vast distances of nowadays dissected populations.

Tibicina corsica is a species known from the Mediterranean island of Corsica and from a limited area in southern France (Narbonne to Agde; e.g. Puissant, 2006), and it was reported for Sardinia (e.g. Brizio, 2015). French mainland populations are regarded at a subspecific level as Tibicina corsica fairmairei Boulard, 1984, but song and coloration are not evidently different from the insular populations of subspecies corsica (Boulard, 2000; Sueur & Aubin, 2003). It sings fairly loudly (up to 84.5 dB in 0.5 m distance; Sueur & Sanborn, 2003) and is the largest European grassland species besides Tibicina tomentosa (Olivier, 1790). In comparison with many other European cicadas, it can be easily observed due to its favourite habitats, its gregariousness and no obvious shyness. The next phylogenetically tested relative of T. corsica is T. nigronervosa, another endemic species described from Corsica (Sueur et al., 2007).

Costa (1883, 1884) already recognized the high coloration spectrum of Tibicina specimens in Sardinia and deduced up to three species in his detailed travel reports: Tibicina cisticola (Hagen, 1855), T. tomentosa and T. luctuosa (Costa, 1883), the latter he described as new. He obviously did not know about Tibicina corsica, which for a long time was mistaken for a smaller Cicadetta species (e.g. Puton, 1886; Metcalf, 1963). Different authors commented on this work on the basis of morphology (Boulard, 1977; Boulard & Mondon, 1995) or listed Sardinian taxa (Servadei, 1967; Nast, 1972; Guglielmino et al., 2000). While several studies have been published about Corsican taxa in modern times (Boulard, 1976; Bonfils & Della Giustina, 1978; Puissant & Sueur, 2001), Sardinia was poorly investigated in cicadas, and acoustic information on the songs has only been recorded recently and locally (Brizio, 2015). Without knowledge of Costa’s studies, I detected morphological trends coupled with clear differences in syllable periods, which seemed stable between populations from different regions. These populations, forming a small, unknown species complex, are described and discussed, and first taxonomic conclusions based on an integrative approach are drawn.

MATERIAL AND METHODS

Fieldwork

Fieldwork on Sardinia was conducted in late June and early July 2008, in the middle of July 2013, during the first half of July 2017 and the second half of July 2018. For comparison purposes, Corsica was visited in late June and early July 2017. Cicadas were detected by their striking calling songs. In the years 2013, 2017 and 2018, specimens were consequently recorded first, then caught with a net and finally photographed. Some individuals were released after this procedure, others were collected for further study.

Song recordings were made with a Marantz PMD 660 and a Tascam DR-44WL (sampling frequency at 48 kHz) and different microphones (especially Telinga Pro 7). Temperature of the perch surface (grass, branches of shrubs or trees), where the singing male was presumably sitting, was measured with a TFA ScanTemp 410 infrared thermometer in order to relate findings with abiotic factors, since temporal song parameters can be dependent on the body temperature (Fonseca & Revez, 2002; Sanborn, 2006; Hertach et al., 2015). The measured temperature is not identical with the body temperature, but at least it is a practical approach for efficient fieldwork.

Habitat structure was recorded following the eight classes of Puissant & Sueur (2001) and Puissant (2006) for each local population. This rough assessment for Mediterranean habitats measures the dominance of ligneous plants and the medium height of dominant plants (Table 1). In this study, a location could belong either entirely to one or partly to a maximum of two vegetation classes.

Table 1.

Vegetation classes after Puissant & Sueur (2001)

Class	Dominance of ligneous plants	Medium height of dominant plants
1	< 20%	< 0.5 m
2	20–40%	< 0.5 m
3	20–40%	0.5–2.0 m
4	40–60%	< 0.5 m
5	40–60%	0.5–2.0 m
6	> 60%	< 0.5 m
7	> 60%	0.5–2.0 m
8	> 40%	> 2.0 m

Data from 46 locations was gathered from the most northern peninsula, Santa Teresa Gallura, southward to a line linking roughly Alghero, Macomer, Fonni and Cala Gonone (Fig. 1). This dataset was compared with local populations from Corsica (11) and southern and central Sardinia (5). Investigations were to a greater extent conducted in putative contact zones of taxa to search for indications of recent hybridization, like, for example, intermediate song patterns (e.g. Marshall et al., 2011; Hertach et al., 2015).

Figure 1.

Map with investigated populations in northern Sardinia and southern Corsica. The size of the circles is relative to the number of analysed specimens (N_max = 15, N_min = 1).

Song analyses

Songs were analysed using Raven Pro 1.4 (Cornell Lab of Ornithology). Illustrations of oscillograms were generated with the package SEEWAVE (Sueur et al., 2008) and graphs with ggplot2 (Wickham, 2009) on the R software platform (R Core Team, 2014). Terminology of the variables used is adapted from Sueur & Aubin (2003). Their term ‘group of pulses’ was replaced by ‘syllable’. I simplified the frequency parameters for this taxonomic purpose to the Raven Pro functions. The main variable is ‘Centre Frequency’, where 50% of the energy is below and 50% above the measured value. Methods of measured variables are summarized in Table 2 and partly visualized in Figure 2.

Table 2.

Methods of acoustic analysis

	Abbreviation	Term	Measuring mode per individual
Durations	SYD	Syllable duration	4 evenly selected groups per 5 syllables = 20 values
	ISYD	Intersyllable duration (pause)	Calculated from SYD ends and next starts = 16 values
	SYP	Syllable period	Calculated by SYD + ISYD = 16 values
	PP	Pulse period	Taken at the most flawless group of syllables from the beginning of the third to the beginning of the forth pulses = 5 values
Counts	NP	Number of pulses per syllable	Estimation from the most flawless group of syllables = 5 values
Frequencies	Centre Freq	Centre Frequency	5 evenly selected periods of 2 s = 5 values
	Q1 Freq	1. Quartile Frequency	5 evenly selected periods of 2 s = 5 values
	Q3 Freq	3. Quartile Frequency	5 evenly selected periods of 2 s = 5 values
	Max Freq	Maximal Frequency	5 evenly selected periods of 2 s = 5 values

Figure 2.

Calling song examples of the Tibicina OTUs from northern Sardinia. A, sonograms of LONG and CIST (10-s phrases). B, oscillograms (time vs. amplitude) of four different OTUs (0.5-s phrases, left and 0.1-s phrases, right) with terms marked.

Significances among the operational taxonomic units (OTU) were verified with Wilcoxon–Mann–Whitney rank sum tests (U-test). Duration characters were also correlated with perch surface temperature (linear regressions).

Morphological analyses

Only song-identified males and females in the close neighbourhood of singing males were used in the morphological analyses. Specimens were, in particular, compared for coloration characters, hair cover, the timbals and the genitalia. Some principal body measurements were taken with vernier callipers or a Wild M3Z stereomicroscope from vouchers. Specimens for figures were photographed in high resolution with a Leica DVM6. The morphological descriptions follow the terminology of Moulds (2005) and Puissant & Sueur (2002). Comprehensive morphometric studies are planned for a later date.

Because dried specimens tend to loose coloration characters and hairs, individuals were photographed alive in the field from the dorsal, ventral and lateral side. Coloration tints of the forewing venation, posterior margin of the pronotum and spots on the scutum were compared by measuring the RGB-values from these photographs with programs like Adobe Photoshop. Outliers of the first round were controlled again. Hair cover and dark portions of sectors on mesonotum and abdomen were estimated by sorting the photographs (labelled with a random number only) over all OTUs in one alignment from extreme to converse extreme phenotypes. Specimens then received a rank for their position in the alignment. Sorting was once repeated and mean ranks were used for statistics. Wilcoxon–Mann–Whitney rank sum tests (U-test) were performed among OTUs for these ranks and for other measurements. Normally the sun was shining intensely when outdoor photos were taken and insects were exposed optimally not to cause shadows. However, I am aware that these conditions were not perfectly standardized but were at least a simple approach to make small differences more objective.

Distribution pattern and ecology

Maps were produced with ArcGIS (map source: http://www.worldclim.org and http://www.diva-gis.org). The location of all populations in general, and specimens analysed in this study in particular, are given in the Supporting Information (Appendix S1). Counts of inhabited vegetation classes (after Puissant & Sueur, 2001) were tested among OTUs with chi-square contingency tests.

RESULTS

Four distinct groups of Tibicina in northern Sardinia arise from the acoustic analysis. Therefore, the following operational taxonomic units (OTU) are defined, which are colour-coded in the illustrations: CORS (= orange), NIGR (= violet), CIST (= brown) and LONG (= red). These OTUs were tested against morphological, ecological and chorological traits.

Song patterns

Calling songs of all Tibicina OTUs are simple, long-lasting buzzes. The most indicative variable is the syllable period (SYP). CORS is in the range of T. c. corsica and NIGR in the range of T. nigronervosa from Corsica. While the SYP is clearly shorter (69%) in CIST in comparison with CORS and T. c. corsica, it is longer (119%) in LONG with no overlaps (Figs 2, 3A; Table 3). NIGR (and T. nigronervosa) SYPs are shortest. SYP is composed of variable PP, NP and ISYD (Fig. 3B). These basic acoustic elements show large overlaps; at least some of them differ significantly between OTUs, others not. Only their combination is relevant.

Table 3.

Measurements of calling songs in Sardinian Tibicina OTUs and known Corsican species. Mean values among individuals ± SD. N_ind = number of individuals (for definition of song variables see Fig. 2 and Table 2)

OTU or taxon	Nind	SYP [ms]	ISYD [ms]	PP [ms]	NP	Centre Freq [kHz]	Q1 Freq [kHz]	Q3 Freq [kHz]	Max Freq [kHz]
LONG	30	19.6 ± 0.6	2.5 ± 0.8	1.41 ± 0.16	13.8 ± 0.6	9.50 ± 0.33	8.96 ± 0.30	10.10 ± 0.32	9.53 ± 0.53
CIST	28	11.4 ± 0.6	1.6 ± 0.5	0.93 ± 0.13	11.1 ± 1.0	9.94 ± 0.28	9.39 ± 0.27	10.51 ± 0.28	10.10 ± 0.35
CORS (Sardinia)	15	16.5 ± 0.4	1.4 ± 0.4	1.28 ± 0.15	13.2 ± 0.9	9.98 ± 0.28	9.45 ± 0.27	10.50 ± 0.27	10.14 ± 0.37
T. c. corsica (Corsica)	12	16.5 ± 0.5	1.6 ± 0.5	1.19 ± 0.10	13.6 ± 0.8	9.69 ± 0.20	9.15 ± 0.15	10.26 ± 0.23	9.92 ± 0.33
NIGR (Sardinia)	10	9.0 ± 0.2	1.6 ± 0.4	0.79 ± 0.09	9.4 ± 0.8	9.76 ± 0.20	9.17 ± 0.15	10.38 ± 0.22	9.86 ± 0.24
T. nigronervosa (Corsica)	9	9.0 ± 0.1	1.3 ± 0.4	0.87 ± 0.10	9.6 ± 0.9	9.34 ± 0.38	8.64 ± 0.35	10.03 ± 0.36	9.36 ± 0.59

Figure 3.

Calling song characters of the Tibicina OTUs from Sardinia and the known species from Corsica. Mean values per individual. A, syllable periods (SYP) vs. centre frequencies (N_ind = 103). B, contribution of PP, NP and ISYD to the species-specific SYP (N_ind = 103). C, two converse examples of temperature dependency of SYP (N_ind = 19 resp. 23).

Centre frequency means of all OTUs vary from 9.3 ± 0.4 to 10.0 ± 0.3 kHz. They are not diagnostic owing to a large overlap (Fig. 3; Table 3). Indeed, differences are significant between lower LONG and higher CORS and CIST in Sardinia (for statistics see section Diagnosis in species description chapter further on). Tibicina c. corsica from Corsica also sing significantly lower than CORS from Sardinia (Wilcoxon–Mann–Whitney rank sum test P = 0.009). The same is true for T. nigronervosa (Corsica) and NIGR (Sardinia; P = 0.016). Audio examples of the four OTUs are presented at the web pages Songs of European singing cicadas (http://www.cicadasong.eu).

Perch surface temperature is not responsible for these OTU-specific song differences. However, there seems to exist a temperature dependence for the SYP in CIST (linear regressions, R² = 0.27, T = 26° to 39 °C, N_ind = 24) and CORS (R² = 0.43, T = 28° to 39 °C, N_ind = 23), but not in NIGR (R² = 0.10, T = 27° to 35 °C, N_ind = 12) and LONG (R² = 0.00, T = 30° to 38 °C, N_ind = 19; Fig. 3C). The good correlations are surprisingly positive, meaning that kinetics of the timbals become slower with higher temperature.

Morphology

Hair cover and the colour tints in the three morphologically closely related OTUs, CIST, LONG and CORS, are not clearly distinct in dried specimens, which leads to difficulties in determining material from collections. However, live individuals show evident differences. CIST is normally with light ochre basal forewing venations (Fig. 4; Table 4), dense silvery to golden hairs on the mesonotum and the ventral side of the abdomen (Fig. 5; Supporting Information, Appendices S2, S3). Crescent-shaped spots on the mesonotum are sometimes large and merge into variable patterns. LONG has more vividly yellow basal venations, less hair on the mesonotum and on the ventral side of the abdomen with more of the black surface shining through and the crescent-shaped spots are of more constant size and never fuse. CORS is generally close to CIST (e.g. hair cover), but less diverse in size of mesonotal spots and more variable in coloration tints of forewing venation. There are no obvious differences to T. c. corsica from the island of Corsica. The RGB-values measured at the posterior margin of the pronotum and at the spots on the scutum did not yield important differences among the OTUs LONG, CIST and CORS, with the exception of the R-values at the pronotum being slightly lower in LONG. NIGR has a divergent coloration: it has black forewing venations with the exception of the costa up to the node and four small orange spots on the mesonotum, as is typical for Corsican T. nigronervosa.

Table 4.

Statistics of morphological characters in LONG, CIST and CORS (for definition of characters see Figs 4, 5, 7E). N_ind = number of individuals. Characters not mentioned were not significant. CORS and T. c. corsica are merged for easier illustration

		Mean ± standard deviation			Wilcoxon-Mann-Whitney p-value
Character	Nind	LONG	CIST	CORS	LONG vs. CIST	LONG vs. CORS	CIST vs. CORS
Hair cover of submedian sigilla	87	61.1 ± 22.7 (rank)	35.2 ± 23.7 (rank)	36.3 ± 20.9 (rank)	0.0003	0.0001	0.7930
Black on sternites IV to VI	78	18.9 ± 11.7 (rank)	54.7 ± 18.1 (rank)	48.2 ± 18.5 (rank)	0.0000	0.0000	0.2760
Colour tint venation 1 (RGB)	95	85.4 ± 27.0 (B-value)	126.9 ± 21.4 (B-value)	107.0 ± 29.0 (B-value)	0.0000	0.0048	0.0111
Colour tint venation 2 (RGB)	95	81.2 ± 29.2 (B-value)	117.0 ± 18.2 (B-value)	106.9 ± 34.6 (B-value)	0.0000	0.0015	0.3238
Colour tint of posterior margin of pronotum (RGB)	95	212.2 ± 19.5 (R-value)	198.0 ± 21.0 (R-value)	203.1 ± 17.0 (R-value)	0.0120	0.0411	0.3303
Short timbal ribs	59	10.0 ± 0.4 (number)	8.9 ± 0.6 (number)	9.5 ± 0.6 (number)	0.0000	0.0037	0.0028
Body length to timbal width	59	11.1 ± 0.4 (ratio)	11.9 ± 0.6 (ratio)	11.9 ± 0.5 (ratio)	0.0001	0.0000	0.5205

Figure 4.

Morphological trends in the Corso-Sardinian Tibicina OTUs CIST, CORS and LONG (part I). Colour tints of basal and central forewing venation. RGB-values taken at two positions (see numbered ovals in example CIST) from photographs of live individuals. R-values constant for boxes, G-values at vertical axes, B-values at horizontal axes. Size of circles is relative to the measured occurrences (N_ind = 95, N_max = 12, N_min = 1). Two typical venation phenotypes of LONG and CIST (right, below). CORS and T. c. corsica are merged for easier illustration.

Figure 5.

Morphological trends in the Corso-Sardinian Tibicina OTUs CIST, CORS and LONG (part II). Individual ranks of overall alignments of hair cover of submedian sigilla on mesonotum vs. black portion shining through hairs on sternites IV to VI (see green framed body sectors; N_ind = 70). Photographs visualize extreme phenotypes. Individuals were labelled with a random number and sorted twice for each character from photographs of live individuals and alignments differed only by a mean of 4.7 (sigilla) respectively 6.5 rank positions (sternites). Significances were similar and mean ranks of individuals of repeated alignments are shown here. CORS and T. c. corsica are merged for easier illustration.

Timbals have alternating long and short ribs, which show OTU-typical ranges in numbers from NIGR and T. nigronervosa (7.0 ± 0.5 short ribs) to LONG (10.0 ± 0.4 short ribs; Table 4). Similarly, the ratios of body length to timbal width vary from 12.8 ± 0.5 (NIGR and T. nigronervosa) to 11.1 ± 0.4 (LONG; Table 4).

Distribution and habitat requirements

The three OTUs, LONG, CIST and CORS, are not randomly distributed in northern Sardinia, but each forms a set of local populations with clear spatial limits. CIST is the most widespread Tibicina but is absent from the eastern coastland and becomes rare toward the western coast and the Nurra plain. Its core areas are the inland plains. Vegetation structure is usually open with low dominance of woody plants, including many typical grassland localities (Fig. 6). LONG is limited to the more wooded east and north-east, and often near the coast. Males normally sing in bushes or trees of closed Mediterranean shrubland or forests.

Figure 6.

Habitat specialization of the Tibicina OTUs from northern Sardinia. A, portions of eight vegetation classes in all four Tibicina OTUs from Sardinia. Data are shown per square root of observed individuals (N_ind = 586). Foreground drawing after Puissant & Sueur (2001). B, portions of eight vegetation classes in partly sympatrically occurring CIST and LONG. Data is visualized per location and the two class variables are displayed on the axes (dominance of ligneous plants vs. vegetation height; N_loc = 42). Bold numbers are vegetation class numbers and black dashes indicate the number of counts. Example: Vegetation class 3 fits for 8.5 locations in CIST, containing 2.5 locations in sympatry with LONG and 6 locations in allopatry to LONG. Typical Northern Sardinian habitats of: C, CIST (grassland, Pattada); D, CORS (low shrubland, Stintino); E, LONG (tall shrubland, Priatu); F, NIGR (woodland, Rena Majore).

The distribution area of CORS is surprising. It not only occurs in the south of Sardinia (Brizio, 2015; Supporting Information, Appendix S1), but also on the Stintino peninsula in north-western Sardinia, and is not present near Corsica. Habitats are intermediate to CIST and LONG. NIGR was rarely found near the coast, but is normally a species of higher elevations, occurring in closed forest. Its populations appear to be relictual.

Up to three OTUs occur in sympatry. CIST and CORS form a contact zone toward the Stintino peninsula, while LONG is allopatric to CORS (50 km distance or more). All sympatric OTU pairs were found syntopically sometimes. Whether local populations of CIST and LONG are allopatric or sympatric does not influence the habitat preferences substantially (Fig. 6B).

Status of operational taxonomical units

Combining acoustic differences, morphological and ecological traits, and the distribution patterns, I conclude that four different Tibicina species have been found in northern Sardinia. Integrative taxonomy reveals no conflicts: syllable periods (SYP) are without overlap between OTUs (Fig. 3A). Colour tints, hair cover and timbal characters are good indicators for different OTUs (Figs 4, 5; Table 4). Each OTU occupies its own ecological niche; especially illustrative is the segregation between partly sympatric LONG and CIST (Fig. 6). I have never found intermediate song patterns indicating recent hybridization, neither between sympatrically distributed taxa nor at the closest peripheries of parapatrically occurring taxa. All of these OTUs constitute ‘separately evolving metapopulation lineages’ in the sense of the species concept of de Queiroz (2007).

There are no important indications that CORS is different from Corsican Tibicina corsica corsica (Rambur, 1840). Songs match the patterns described by Sueur & Aubin (2003) and morphology does not conflict with the original descriptions by Rambur (1840) and the redescription by Webb (1979). Sardinian specimens are slightly smaller (e.g. 22.2 ± 1.2 to 22.7 ± 1.5 mm body length in males), which can at least partly explain the higher frequency domain (see: Bennet-Clark & Young, 1994). Thus, Tibicina corsica corsica can be verified as a Sardinian species (see also: Brizio, 2015). The same is true for NIGR, which I conclude is identical to Tibicina nigronervosa Fieber, 1876: body lengths appear again to be partly responsible for significant frequency differences (22.7 ± 0.6 in Sardinia to 23.9 ± 1.1 mm in Corsica). Therefore, I support the opinion of Boulard & Mondon (1995) that Tibicina luctuosa (Costa, 1883) is a junior synonym of T. nigronervosa Fieber, 1876. At one of Costa’s locations (Monte Limbara), T. nigronervosa was positively confirmed.

CIST is provisionally treated as Tibicina sp. indet. This species may have been previously described under the name Cicada cisticolaHagen, 1855, a taxon that is currently listed as a junior synonym of T. c. corsica (e.g. Boulard & Mondon, 1995). The true status of Cicada cisticola remains uncertain. Hagen described the taxon in detail as a variety of Cicada tomentosa (= Tibicina tomentosa) based on a series received from the Natural History Museum of Berlin. The one male and two females had been collected by Gené in ‘Sardinia’ without closer indication of loci typici. The syntype male still exists in the Berlin collection (Fig. 7C). It exhibits large, yellowish markings on the pronotum and mesonotum, light ochre basal forewing venations, a hairy, almost golden habitus and a low coverage of black at the sternites. Body length to timbal width ratio is high (ratio 13.1) and short timbal ribs are nine (Fig. 7F). All these characters in combination exclude LONG. However, the morphological distinction of CIST and T. c. corsica is not clear. Some characters mentioned above are often less pronounced in T. c. corsica than in CIST. We will try to positively assign the existing type specimen to a modern song-verified taxon with morphometrics or/and ancient DNA in an ongoing study.

Figure 7.

Morphology of Tibicina longisyllaba and relatives. A, holotype male of Tibicina longisyllaba in dorsal view and ventral view of the abdomen from Tempio Pausania (black bar = 1 cm). B, paratype female of Tibicina longisyllaba from Tempio Pausania. C, syntype male of Tibicina cisticola from the Natural History Museum of Berlin with original labels. D, genitalia of T. longisyllaba. Timbals with alternating short and long ribs of (E) T. longisyllaba holotype in comparison to (F) T. cisticola syntype (equal magnification, almost similar body size, 24.4 resp. 24.8 mm; white perpendicular line = timbal width definition). Morphological extremes in natural habitats: G, male of T. nigronervosa; H, male of T. sp. indet.

The fourth taxon LONG I describe as a new species below. It is not congruent with any other described Tibicina spp. in general and those listed for Sardinia in particular [e.g. Nast (1972); i.e. Tibicina picta (Fabricius, 1794) and T. baetica (Rambur, 1840)], which are both synonyms of Tibicina tomentosa (Boulard & Mondon, 1995; Sueur & Puissant, 2003) and have loci typici in France and Andalusia, respectively.

Tibicina longisyllaba sp. nov.

LSID: urn:lsid:zoobank.org:act:E06E9415-FC2C- 487F-B0BB-696599C53BE4

Type material

The type series consists of 17 males and eight females. It is kept in the Natural History Museum of Bern (NMBE) (holotype; Fig. 7A, D, E), the Natural History Museum of Basel (NHMB) and the private collection of the author.

Holotype male:

Verbatim label information: ‘Italy/SW Tempio Pausania, SARD/40.8689°/9.0227°, 132 m asl/29.7.2018, leg. Thomas Hertach’ (label rectangular, white, printed) and ‘Tibicina longisyllaba ♂/det. T. Hertach 2018/Collection Code no. 27.026’ (label rectangular, white, printed) and ‘HOLOTYPUS ♂/Tibicina longisyllaba sp. n. Hertach 2020’ (label rectangular, light red with dark red margin, printed).

Paratypes:

All paratypes with labels ‘PARATYPUS XX Y, Tibicina longisyllaba sp. n. Hertach 2020’ (label rectangular, white with red margin, printed) at which ‘XX’ is the number of the paratype and ‘Y’ the sex of the specimen. Paratype males: Monte Pinu, Sardinia, 40.9367°/9.3933°, 280 m a.s.l., 28.6.2008, leg. T. Hertach (paratype 1); N of Ala dei Sardi, Sardinia, 40.6889°/9.3663°, 560 m a.s.l., 26.7.2013, leg. T. Hertach (paratypes 2 and 3); La Licciola, Santa Teresa Gallura, Sardinia, 41.2149°/9.2608°, 60 m a.s.l., 10.7.2017, leg. T. Hertach (paratype 4); above Badesi, Sardinia, 40.9734°/8.9045°, 280 m a.s.l., 12.7.2017, leg. T. Hertach (paratypes 5 and 6); N of Padru, Sardinia, 40.7828°/9.5146°, 240 m a.s.l., 15.7.2017, leg. T. Hertach (paratype 7); Murta Maria, Sardinia, 40.8938°/9.5926°, 9 m a.s.l., 17.7.2018, leg. T. Hertach (paratype 8); Loculi – Lula, Sardinia, 40.3934°/9.5163°, 51 m a.s.l., 19.7.2018, leg. T. Hertach (paratypes 9, 10 and 11); Capo Comino, Sardinia, 40.5321°/9.8105°, 25 m a.s.l., 21.7.2018, leg. T. Hertach (paratypes 16, 17 and 18); Bidderosa, Sardinia, 40.4710°/9.7761°, 48 m a.s.l., 21.7.2018, leg. T. Hertach (paratype 20); Capo Comino, Sardinia, 40.5333°/9.8067°, 10 m a.s.l., 6.7.2008, leg. T. Hertach (paratype 24). Paratype females: Loculi – Lula, Sardinia, 40.3934°/9.5163°, 51 m a.s.l., 19.7.2018, leg. T. Hertach (paratypes 12 and 13); Cala Gonone, Sardinia, 40.2868°/9.6357°, 36 m a.s.l., 20.7.2018, leg. T. Hertach (paratype 14); above Dorgali, Sardinia, 40.3000°/9.6344°, 297 m a.s.l., 20.7.2018, leg. T. Hertach (paratype 15); Capo Comino, Sardinia, 40.5321°/9.8105°, 25 m a.s.l., 21.7.2018, leg. T. Hertach (paratype 19); Bidderosa, Sardinia, 40.4710°/9.7761°, 48 m a.s.l., 21.7.2018, leg. T. Hertach (paratypes 21 and 22); SW of Tempio Pausania, Sardinia, 40.8689°/9.0227°, 132 m a.s.l., 29.7.2018, leg. T. Hertach (paratype 23; Fig. 7B).

Diagnosis

Morphology:

There are good and highly significant indicators for T. longisyllaba (Table 4), although I am not able to distinguish all specimens with absolute certainty from T. c. corsica and T. sp. indet.: (1) basal forewing venation up to the node vividly yellow, instead of light ochre as in T. sp. indet. and many T. c. corsica (Fig. 4); (2) submedian sigilla on mesonotum less hairy than in T. sp. indet. and T. c. corsica (often below 20% hair cover; Fig. 5); (3) black portion of sternites IV to VI on abdomen shining through hairs normally larger than 20%, while often smaller in T. sp. indet. and T. c. corsica (Fig. 5); (4) males: short timbal rib number on average higher, normally 10; (5) males: body length to timbal width ratio in mean lower (11.1 ± 0.4).

Tibicina nigronervosa can be distinguished by the dark basal forewing venation (Fig. 7G), only 7.0 ± 0.5 (maximally 8) short ribs at the timbals and a high body length to timbal width ratio (12.8 ± 0.5). Furthermore, lateral light spots on the pronotum do not exist and rear cuneiform spots on the mesonotum are normally isolated from the lateral band. Habitus and coloration of T. longisyllaba are also reminiscent of Tibicina garricolaBoulard, 1983 occurring on the mainland in France and the Iberian Peninsula, but the more trapezoid pronotum with prominent collar carries light (orange-reddish-brownish) fasciae in T. garricola. Lateral yellow patterns at the mesonotum are reduced, ending in a total of normally eight scutal spots. The shape of the genitalia is different with: uncus in lateral view dorsally more rounded, and ventrally strongly emarginate toward the tip; theca with relatively long and dangling lateral lobes; vesica considerably broadened in upwards curve (Boulard, 1983; Quartau et al., 2001; Stéphane Puissant, pers. comm.).

Acoustics (Figs 2, 3; Table 3; Supporting Information, Appendix S4):

The loud, buzzing calling song of T. longisyllaba is strident and fibrillations are clearly audible, while in T. sp. indet. and T. nigronervosa no structure is audible in the song, and in T. c. corsica it is less distinct. Syllable periods are long (19.6 ± 0.6 ms) with no overlaps of measured extrema among Corso-Sardinian species. Centre frequency is significantly lower than in all other Tibicina taxa from Sardinian populations (Wilcoxon–Mann–Whitney rank sum test: P < 0.001 for T. sp. indet.; P < 0.001 for T. c. corsica and P = 0.018 for T. nigronervosa). However, significant frequency differences are not seen when Corsican specimens are compared to T. longisyllaba (Wilcoxon–Mann–Whitney rank sum test: P = 0.081 for T. c. corsica and P = 0.264 for T. nigronervosa).

Ecology:

Tibicina longisyllaba clearly prefers habitats with a dominance of ligneous plants higher than 40% in comparison to sympatrically occurring T. sp. indet. (Fig. 6B; chi-square contingency test for two classes < 40%, > 40%; P = 0.001, N_loc = 42). Vegetation is normally also taller (chi-square contingency test for the three classes < 0.5 m, 0.5–2.0 m, > 2.0 m; P = 0.028, N_loc = 42).

Morphological description

Measurements of holotype (for other males of the series and females see Table 5):

Body length: 24.4 mm, body width (abdomen, tergite II): 9.2 mm, forewing length: 28.8 mm, forewing width: 11.6 mm.

Male holotype with remarks on the variability in the paratype series (Fig. 7):

Head: About same width as mesonotum. Black with silvery hairs, small triangular yellow patch on epicranial suture and four marks along the frontal margins of vertex (the latter rarely reduced in paratypes). In dorsal view, vertex laterally next to the postclypeus almost horizontal and forming a conspicuous angle with the postclypeus. Compound eyes in live specimens grey-brown with darker spots. Ocelli orange to brownish, forming an obtuse triangle. Distance between lateral ocelli and next compound eye equal. Antennae blackish with exception of yellow margins of scapes. Postclypeus toward frons angled and ventrally with longitudinal groove, black with silvery hairs, lateral margins and upper central part yellow. Anteclypeus black with frontal margin yellow. Lorum also black with margins yellow, gena with prominent yellow margin. Rostrum scarcely reaching mid trochanter, labrum yellowish to brownish, mentum with longitudinal brown-yellow drawings, labium dark brown.

Thorax:

Pronotum generally relatively loosely haired and black except for a yellow median line, two yellow patches between the paramedian and the lateral fissures toward the pronotal collar, which are fused by a small connection (in paratypes only exceptionally fused, 6%, and sometimes completely missing, 25%). Pronotal collar large and yellow, its lateral angles pronounced and rounded with a considerable portion of dark colour, margin frontal to the lateral angles slightly S-shaped in dorsal view. Scutum generally black and with characteristic markings (Figs 5, 7; see also: Supporting Information, Appendix S2): (1) both submedian sigilla with a crescent-shaped yellow spot at their posterior bases and (2) starting frontal to the scutal depression a yellow band that forms a frontal corner and then laterally surrounds the lateral sigilla (in paratypes rarely interrupted). Parapsidal suture marked yellowish (in paratypes sometimes missing). From frontal margin with submedian sigilla to scutal depressions almost hairless, laterally and toward the rear more silvery hairs (for intraspecific variability see: Supporting Information, Appendix S2). Cruciform elevation predominantly yellow, its lateral depressions mainly dark. Metanotum centrally visible and yellowish, laterally with darker portions (in some paratypes also centrally with dark patches). On the ventral side, central segments generally darker than lateral segments, which are often light yellow. Opercula small and slender and distant from each other, mainly yellow (in paratypes basal parts often dark brown).

Abdomen:

Abdomen inverted U-shaped in cross-section. Tergite I centrally, with undulating posterior margin. Tergites III to VIII black with orange to yellow posterior margins; margins centrally small, laterally more dominant. Tergite II mainly orange to yellowish laterally. Sternites III to VII and epipleurites frontally black and toward the rear orange; maximal black portions at sternites IV to VI, hair cover comparatively thin to almost absent (for intraspecific variability see: Supporting Information, Appendix S3). Sternite VII relatively drawn-out but still broader than long. Sternite VIII barge-like, long and tapered, yellow and black (in paratypes also often completely dark, 48%, or sometimes completely light, 22%). Timbals lacking a cover, with regularly alternating long and short ribs. Ten short ribs present (Fig. 7E; in paratypes exceptionally 9, 6%, or 11, 6%). Timbals relatively large, ratio body length to timbal width 10.5 (in paratypes minimum 10.2, maximum 11.8).

Genitalia (Fig. 7D):

Similar to Tibicina corsica. Pygofer dorsally black, ventrally yellow and almost straight without upper lobes, flattened in cross-section for upper part, at the apex strongly emarginate. Anal styles tongue-like to drop-shaped. In lateral view, predominantly black and dominant uncus dorsally straight, ventrally convex and slightly emarginate near the tip; in dorsal view concave at the tip. Claspers absent. Theca continuously broadened toward the apex, lateral lobes short. Thin and slightly flattened vesica curved upwards and more or less narrowing subsequently toward the gonopore.

Legs:

All legs with dark brown to black longitudinal fasciae on a yellow ground. Front femora with four spines: the first spine isolated, solid sharp and directed upwards, the second spine broad and straight, the third spine tiny and triangular and the forth tiny and thin (in paratypes forth spine often missing), the latter three forming a group. Distance between first and second spines much larger than length of these two main spines. Mid-legs spineless, hind legs with six solitary tibial spurs (paratypes 4 to 7) and tibial comb carrying a dozen spurs. Large meracantha tongue-like, reaching the posterior horizontal extensions of the opercula.

Wings:

Forewing hyaline except for smoky to black parts at the basal cell, at the base of the clavus and the pterostigma. Cubitus anterior vein originating approximately at the centre of the basal cell and, therefore, clearly separated from median vein. Eight apical cells. Veins vividly yellow (RGB-values of alive holotype 221/171/58 for basal median vein and 234/190/81 for distal parts of median vein near the node; for intraspecific variation see Fig. 4) with the following exceptions: costal vein black (in paratypes rarely also yellowish), including its functional distal prolongation (= subcostal vein), majority of veins forming the apical cells brown to black. Basal junction of anal veins yellow and basal membrane orange. Hindwing transparent, except for smoky base, margins of jugum and plaga additionally orange to whitish, base of costal cell light orange. Its veins predominantly yellow, but with a tendency to more dingy yellow or brownish parts, basally but especially apically. The ambient vein dark brown to black, as well as the anal veins 2 and 3. Posterior radial and medial veins not fused at their base.

Female paratypes (Fig. 7B):

On average, shorter body than in males (Table 5). Female body and wing coloration are close to the holotype and paratype males. On the head, two females have the central yellow band of postclypeus prolonged to dorsal part and a minority has yellow spots merged along the frontal margin of vertex. On the thorax, metanotum centrally often orange, rarely with darker portions. Opercula even smaller than in males, sometimes more angled, meracanthus surpasses the caudal extensions of the opercula. On the abdomen, tergite I with normally less undulating posterior margin. Lateral yellow to orange portions reduced at tergite II. Sternite VII with a V- to U-shaped notch to the middle and with a broad orange margin, frontally dark brown to black. Sternites IV to VI often with high portions of black (Fig. 5; Supporting Information, Appendix S3). Ventral parts of tergite IX orange to brownish, dorsal surface black with orange to brownish, progressively narrowing margin. Its dorsal beak black, pointed and long. Ovipositor brown, its sheath black. Ratio of body length to ovipositor length (including sheath) 4.18 ± 0.09. Wings and legs identical to males, the facultative forth spine at forelegs missing.

Table 5.

Body and wing sizes within the Corso-Sardinian Tibicina species. N_ind = number of specimens

Males		T. longisyllaba (Nind = 17)	T. sp. indet. (Nind = 20)	T. c. corsica (Nind = 22)	T. nigronervosa (Nind = 8)
Body length [mm]	Mean ± SD	23.5 ± 1.4	22.9 ± 0.9	22.3 ± 1.3	23.1 ± 1.0
	Maximum	25.7	24.9	24.6	25.0
	Minimum	20.1	21.3	19.9	21.7
Body width [mm]	Mean ± SD	8.9 ± 0.4	8.7 ± 0.3	8.3 ± 0.4	8.5 ± 0.5
(tergite II)	Maximum	9.6	9.4	8.8	9.2
	Minimum	7.8	8.1	7.5	7.7
Forewing length [mm]	Mean ± SD	27.2 ± 1.1	26.1 ± 0.9	25.1 ± 1.2	26.9 ± 0.8
	Maximum	28.8	27.3	28.1	28.2
	Minimum	24.3	24.6	22.3	25.8
Forewing width [mm]	Mean ± SD	10.9 ± 0.5	10.7 ± 0.4	10.1 ± 0.5	10.4 ± 0.5
	Maximum	11.6	11.9	10.8	10.9
	Minimum	9.4	10.0	9.3	9.4
Females		T. longisyllaba (Nind = 8)	T. sp. indet. (Nind = 2)	T. c. corsica (Nind = 9)	T. nigronervosa (Nind = 0)
Body length [mm]	Mean ± SD	21.3 ± 0.7	23.1 ± 0.2	21.3 ± 1.4
Body width [mm]	Mean ± SD	7.8 ± 0.4	8.4 ± 0.2	7.9 ± 0.4
Forewing length [mm]	Mean ± SD	26.2 ± 0.7	26.4 ± 0.4	25.3 ± 0.8
Forewing width [mm]	Mean ± SD	10.5 ± 0.5	10.7 ± 0.3	10.2 ± 0.4

Acoustic behaviour

Similar to all other European Tibicina, the fundamental element of the T. longisyllaba calling song is a monotonous buzz that normally lasts for several minutes without interruption. The song starts and ends abruptly, and is loud and audible up to a distance of 100 m (Fig. 2). The timbre of the buzz is strident and fibrillations are perceptible for the human auditory system caused by the slow micro rhythm: SYD (syllable duration; 17.1 ± 0.9 ms) are long and ISYD (pauses between the syllables; 2.5 ± 0.8 ms) extraordinarily pronounced, which ends in SYPs of 19.6 ± 0.6 ms (mean of individuals responsible for extrema: 18.5 ms and 20.9 ms). SYP is the best character of the species-specific song pattern (Fig. 3A). Long SYDs are a product of two variables: (1) the number of pulses (NP) is at least as high as in T. c. corsica and clearly higher than in T. sp. indet. and T. nigronervosa and (2) the period of pulses (PP) is longer than in all other Corso-Sardinian taxa. ISYD, NP and PP contribute from individual to individual in a variable degree to the specific SYP (Fig. 3B; Table 3).

Centre frequency means vary between 8.7 kHz and 10.1 kHz for individuals. Half of the energy was maximally measured in a frequency band from 8.4 to 10.7 kHz (Q1 and Q3 frequencies; Table 3). The frequency domain is probably positively correlated with body length (linear regressions; R² = 0.30) and body width (R² = 0.20), but the number of specimens with good acoustic recordings and body measurements is still low (N_ind = 7).

Tibicina longisyllaba has, besides the calling song, an additional repertoire of acoustic signals. Their function has not been completely clarified: courtship, rivalry or alarm. At least one male was recorded during a long sequence of alarm behaviour. In the process, sequences of a few seconds up to 12 seconds were emitted with important amplitude modulations at the end but also in the middle of the sequence. Low-intensity parts were produced by only one timbal, the second was inactivated. High-intensity parts were characterized by a raised abdomen. Hereby, both timbals or only one timbal were working. This behaviour is close to the courtship songs described for the genus by Sueur & Aubin (2004). Males can also produce audible wing-flicks.

Etymology

The species name ‘longisyllaba’ refers to the acoustic behaviour of the species, from Latin longus, long and syllaba, syllable. Syllable periods of T. longisyllaba break the European genus record so far held by T. steveni (17.4 ± 0.7 ms, N_ind = 14; pers. data; Supporting Information, Appendix S4). While general works mention 100 to almost 500 timbal movements per second typical for cicadas worldwide (Burton & Burton, 2002), T. longisyllaba sing with only 51.5 ± 1.5 movements per timbal and second. However, North American Okanagana canadensis (Provancher, 1889) slow down to 24.7 ± 0.9 movements by extremely prolonging the intersyllable durations (ISYP; Chatfield-Taylor & Cole, 2019).

Distribution and ecology

Tibicina longisyllaba is most probably endemic to north-eastern Sardinia, with 22 local populations known until now (Fig. 8). In the north and in the east, the distribution area is limited by the coast, and a distance of 30 km to the sea was never surpassed. Most western local populations are found from north to south in Badesi–Erula–Ala dei Sardi–Lula and Orgosolo. The southern limits are not clear. The known extension from north to south is about 110 km. The elevational range goes from sea level to 560 m a.s.l. (Ala dei Sardi) with almost half of the records below 60 m and only four above 300 m a.s.l.

Figure 8.

Distribution of the genus Tibicina in northern Sardinia and southern Corsica. All records are confirmed by song assessments. Large circle: ≥ 30 specimens detected. Medium circle: 6–29 specimens. Small circle: 1–5 specimens.

The distribution area contains many locations with tall shrubland (maquis; Fig. 6E). It is a rocky region on predominantly crystalline grounds. However, also limestone-influenced habitats are colonized in the Supramonte region. Of the local populations, 60% were found in the most extreme vegetation classes 7 and 8 (Fig. 6). Thus, favourite habitats are woodlands with more than 2-m tall ligneous plants or closed tall shrubland with more than 60% dominance and ligneous plants with a minimal height of 0.5 m. Shrubland is composed of Cistus spp., Pistacia lentiscus L., Euphorbia spp. and broom species. Populations have become established in some coastal forests with pines (Pinus spp.), which are normally planted. The species was occasionally found in gardens of settlements, but populations are normally rather stenoecious. Often T. longisyllaba is the only species in its habitats. In one place it was found syntopic with T. nigronervosa, which colonized the neighbouring forest. In another place it was found syntopic with T. sp. indet. with unclear habitat sharing due to low acoustic activity. In the costal pine forests, T. longisyllaba is associated as a minority with Cicada orni. Population densities of T. longisyllaba are often lower than in T. c. corsica or T. sp. indet.; males seem to occupy larger territories. However, this is not the case in coastal habitats between Siniscola and Orosei where the most important populations have been found.

DISCUSSION

Biodiversity and model group character

The last decade led to an unexpected augmentation of the known European cicada species number (e.g. Gogala & Trilar, 2004; Quartau & Simões, 2005; Puissant & Sueur, 2010; Hertach et al., 2015). These contributions toward a more complete knowledge of the biodiversity are an important basis for all efforts of conservation in an era of worldwide rapid species decline. Cicadas, with their loud songs and big body size, are among a favourite choice for flagship insects. With Tibicina I here present a genus that is more diverse than generally expected, with four instead of two Sardinian and nine instead of seven European species. The verification of this hidden Tibicina corsica species complex would have been challenging before modern recording and analysing devices made the investigation of the microtemporal structures of the calling songs possible. However, with the diversity detected, many T. longisyllaba individuals can be recognized by a trained observer in the field due to fibrillations in the buzz (similar to T. steveni).

Besides the genus Tibicina, only Cicada orni Linnaeus, 1758 was able to invade Sardinia and has survived there until now. This is the single taxon that is not endemic. Even though Sardinia and Corsica are separated nowadays by only 12 km and were connected during glacial maxima in the Pleistocene (e.g. 20 000 years ago; Grill et al., 2007) they share only 50% of the cicada fauna (compare Puissant & Sueur, 2001). These facts underline the low dispersal abilities of cicadas on a long-term scale (compare Arensburger et al., 2004). Indeed, different species communities for the two islands are also common in other insects (e.g. Guglielmino et al., 2000), even in butterflies (Dapporto, 2010).

Corso-Sardinian Tibicina could form a model group for different reasons. Their biogeography and phylogeny will be exemplary, including migrations between the islands and the mainland (southern French endemic T. corsica fairmairei). It also forms a showcase for speciation processes in two aspects and their interactions: first, physiological evolution that finally led to morphological peculiarities (colour, hair cover and song apparatus) and, second, ecological adaptations that resulted in different behaviour and habitat specialization. The driving factors of the Tibicina corsica group speciation (including T. nigronervosa) should be investigated more profoundly in a phylogenetic and biogeographic study.

Evolution and speciation

Tibicina songs did not evolve completely different patterns in contrast to many other cicada species complexes (e.g. Cicadetta or Tettigettalna). The songs are simple, only a few parameters can vary (Sueur & Aubin, 2003). This is also true for the North American Okanagana sister-genus (Chatfield-Taylor & Cole, 2019), possibly due to the early branching position of both genera within the Cicadidae (Moulds, 2005; Marshall et al., 2018) and a more primitive song apparatus.

Sardinian Tibicina species represent a good example of, presumably, recent evolution. The most important acoustic character is the syllable period (SYP). This is valid for the whole genus (Supporting Information, Appendix S4). SYPs have a low intraspecific variation consistent with a static trait (Gerhardt, 1991; Figs 2, 3A; Table 3). Interestingly, there are individually different evolutionary paths, toward a species-specific syllable range (Fig. 3B). Whether syllable periods function as important attributes within the specific mate recognition system (e.g. Paterson, 1985) is debatable. Sueur & Aubin (2002) tested Tibicina haematodes with conspecific, non-conspecific and simulated songs, and males reacted to playbacks of other Tibicina species and signals lacking syllable structures with an acoustic activity of almost similar intensity as to original control songs. Conditions were artificial in their experiments, without contact to conspecific males, but the authors think that a set of acoustic, morphological and ecological characters must be crucial to avoid interspecific contacts. In contrast, Chatfield-Taylor & Cole (2019) calculated a critical song distance on the basis of syllable rates and peak frequencies valuable for random sympatric congeners and natural communities of up to six Okanagana species. They conclude that songs represent a primary isolating barrier, despite lacking structural complexity. In this context, the positive temperature dependencies of time parameters (SYP) in T. c. corsica and T. sp. indet. are remarkable and contradict physiological expectations (Fig. 3C). Timbal muscles generally contract more rapidly and with greater force in cicadas when the temperature of the muscles increases. Syllables become shorter or are actively regulated toward constant values in order to maintain specific traits (Sanborn, 2006; Hertach et al., 2015). Here, it appears that with increasing heat these two species not only retard but, strangely, overcompensate timbal kinetics, which is difficult to interpret ethologically.

One could expect morphological adaptations at the timbals to produce different syllable periods in the examined group. In fact, Tibicina longisyllaba timbals are of relatively large size (Fig. 7D; Table 4), but a general good correlation between width and syllable period does not exist, since T. sp. indet. and T. c. corsica have almost equal timbal widths but different syllable patterns. The number of (short) ribs correlates better with the syllable periods but still shows largely overlapping values. Males singing at lower frequencies are expected to have a larger body size (Bennet-Clark & Young, 1994; Sueur & Aubin, 2003). This norm is by tendency achieved on intraspecific and interspecific levels (Fig. 3A; Tables 3, 5). In any case, the T. corsica group offers a simple system to better understand the sound production of cicadas in general.

Morphological characters overlap in three of four Sardinian Tibicina species, but there is a clear order from darker coloured and almost hairless to lighter and hairy species: T. nigronervosa → T. longisyllaba → T. c. corsica → T. sp. indet. (Figs 4, 5, 7). Exactly the same order can also be observed for the habitat requirements, from T. nigronervosa, which, as one extremity, prefers closed shrubland or forests to T. sp. indet. specialized for grassland (Fig. 6). Possibly, this congruence developed not by chance, but as a response and adaption to habitats: dark colours are an advantage for optimized camouflage on the dark bark of trees, while beige and ochre colour tints dominate in desiccated grassland habitats in summertime. Hair cover potentially supports thermoregulation, which is more crucial in a grassland habitat with important diurnal temperature fluctuations than in more balanced habitats with trees and bushes. In this way T. longisyllaba and T. nigronervosa may progressively have lost more and more hairs and/or the other two species have grown a denser cover. Some outliers can possibly be interpreted as recently hatched (more hairs) and old (less hairs) individuals.

When ancestors of current Tibicina taxa settled in the archipelago, they found diverse habitats with a poor cicada fauna. Different ecological niches were available. The step by step adaptation to inhabit these niches provoked different traits and specialization, which conceivably led to separately evolving unities and species in the sense of de Queiroz (2007), a probable, smaller case of an adaptive radiation. It is striking that T. nigronervosa invariably occupies an extreme position in acoustic, morphological and ecological aspects, and T. c. corsica always an intermediate one. However, T. nigronervosa is closely related to the other three species and appears to be part of the group. This is conspicuous for the genitalia, for example, which show almost no qualitative differences.

Any indications of intermediate song patterns have not been found. For that reason, current gene flow between closely related species should be limited; large hybridization zones do not exist. The set of acoustic, morphologic and ecologic peculiarities, combined with some geographic disruptions, is enough to maintain stable characters. These traits allow T. sp. indet., T. longisyllaba and T. nigronervosa to occur sympatrically (and, in rare cases, syntopically; Fig. 8), which is generally accepted as a dependable property for valid species (e.g. Mallet, 2008). Intermediate T. c. corsica is distributed allopatrically to parapatrically to T. longisyllaba and T. sp. indet.

Sueur & Puissant (2002) demonstrated that ecological niches of sympatric Tibicina species differ either for the preferred habitat structure or for the adult seasonal pattern in southern France. In northern Sardinia, seasonal differences are not obvious. Habitat requirements of the partly sympatric T. sp. indet. and T. longisyllaba, for example, show a highly significant trend but are not perfectly distinct. Historical progressions of the habitats probably contribute to this partial overlap. Here are two fictitious exemplifications: a pasture that was typically colonized by the grassland specialist T. sp. indet. remains the habitat of the species after the location was abandoned and shrubland established, until the sister species T. longisyllaba replaces it; or, T. longisyllaba can possibly be observed in early grassland states of a post-fire-succession, when the habitat was previously closed shrubland. Since cicadas are bad dispersers (Karban, 1981; Simões & Quartau, 2007), crowding-out effects of better adapted species could take a long time.

Distribution and biogeographical aspects

A similar Corso-Sardinian faunistic distribution pattern (Fig. 8) has not been found in the literature, neither for the whole Tibicina group nor for T. longisyllaba. It also only rudimentarily fits the biogeographical plant sectors (Fenu et al., 2014). Tibicina longisyllaba is more or less a lowland species, often found near the coast. In contrast, the majority of endemic butterflies are mountain taxa (Grill et al., 2007). Other endemic animals are widespread on the whole island [e.g. the Sardinian tree frog, Hyla sarda (De Betta, 1853)] or they have extremely restricted distribution areas. The latter pattern is especially known from some rare cases of Sardinian species radiations, e.g. cave-dwelling organisms like Leiodidae beetles (e.g. Caccone & Sbordoni, 2001) or Hydromantes salamanders (e.g. Carranza et al., 2008) and freshwater organisms like Hydraenidae beetles and pseudoscorpions (Minelli et al., 2006). Thus, contrary to Tibicina, radiations have mainly been provoked by patchy or linear habitats rather than widely distributed ones.

The north-western limit of the T. longisyllaba distribution perfectly fits a change in geology from older (Mesozoic–Eocene) and predominantly crystalline cover to younger (Oligocene–Miocene) alkaline volcanics and sediments (Cherchi & Montadert, 1982). Possibly this change influences the habitat structure and land use, and as a consequence, the competition between T. longisyllaba and T. sp. indet.

The time of invasion of the Corso-Sardinian microplate by the Tibicina genus can currently not be answered. Other poorly dispersing organisms settled in the archipelago during the Messinian Salinity Crisis (5.6 to 5.3 Mya; e.g. Ketmaier et al., 2006; Grill et al., 2007;,Carranza et al., 2008). A clearly older, independent evolution of the group forced by the start of the detachment of the microplate from the continent (20 to 29 Mya) is less plausible, given the preliminary molecular data published on the genus (Sueur et al., 2007).

Current distribution patterns are influenced by Pleistocene climatic fluctuations. Corsica was much more affected by glaciation than Sardinia (Kuhlemann et al., 2008; Bisconti et al., 2011). It is even imaginable that Sardinia formed the glacial refuges for thermophilous Tibicina species, and colonization or the last recolonization of Corsica is young, as well as the settlement on the French continent. Modelling of the climatic conditions during the last glacial maximum demonstrates suitable habitats for the frog Hyla sarda, another lowland species, in vast coastal areas almost all around Sardinia, which are nowadays largely flooded by the sea (Bisconti et al., 2011). All these areas were likely convenient for Tibicina species. Climate-induced displacements could have shaped isolated populations and supported speciation. The T. c. corsica populations on the Stintino peninsula are unexpected. Additional investigations are necessary for a better understanding. However, a certain degree of isolation is given by the Nurra plain, similar to the Campidano plain in south-western Sardinia, which separates the Iglesiente–Sulcis region from the rest of the island and was flooded by the sea during extremely warm periods (e.g. Early Pliocene; see: Cherchi & Montadert, 1982).

Conservation

The described diversity of Tibicina in northern Sardinia has an important impact on regional conservation strategies. Tibicina c. corsica has a smaller and more dissected distribution area than generally assumed. Tibicina nigronervosa is a rare species in the research area with isolated local populations and probably suffered more than any other Cicadidae species due to deforestation by early settlers. It is also expected to be influenced most seriously by global warming because it is often a mountain species (see: Regato & Salman, 2008). Tibicina longisyllaba is known from an area of maximally 3500 km²; Tibicina sp. indet. of 5000 km² (Fig. 8). These two endemic species are generally prone to habitat loss. While T. sp. indet. is dependent on the maintenance of a traditional, extensive agriculture, negative effects of human pressure are nowadays highest on coastland due to industrial tourism, which affects T. longisyllaba habitats (see: Grill et al., 2007). Tibicina longisyllaba is considered a vulnerable and attractive species, which sings every year among thousands of tourists, with one of the most powerful local natural sounds, and probably belongs among the loudest European cicadas (compare: Boulard & Mondon, 1995; Sueur & Sanborn, 2003).

[Version of record, published online 19 June 2020; http://zoobank.org/urn:lsid:zoobank.org:pub:3D44BB78-AFDF-42AB-85F0-AA2380500AC0]

ACKNOWLEDGEMENTS

Stéphane Puissant (Natural History Museum of Dijon) and Kevin Gurcel (Annecy) accompanied the author during the stay in Corsica and contributed specimens and experience thereby. They also helped with information about Tibicina garricola. Cesare Brizio (CIBRA, Pavia) exchanged his Tibicina recordings from southern Sardinia. Jürgen Deckert from the Natural History Museum of Berlin provided the remaining Tibicina cisticola type specimen. Roberto A. Pantaleoni (University of Sassari) organized literature and information about Tibicina specimens in the local collection. I am grateful to Meinrad Küchler, Swiss Federal Institute for Forest, Snow and Landscape Research WSL, and Hannes Baur, Natural History Museum of Berne, for methodical support. Elsa Obrecht revised the language of former versions. Two anonymous reviewers provided suggestions that substantially improved the article. Special thanks go to Kurt Bollmann, Swiss Federal Institute for Forest, Snow and Landscape Research WSL, who supervised the Tibicina corsica project. The study was partly financed by a grant of the Basler Stiftung für biologische Forschung. The author reports no conflict of interest.

SUPPORTING INFORMATION

Additional Supporting Information may be found in the online version of this article at the publisher's web-site.

Appendix S1. Database of the records of T. c. corsica, T. longisyllaba, T. nigronervosa and Tibicina sp. indet.

Appendix S2. Variability of mesonotum (hair cover, coloration) in live specimens of Tibicina longisyllaba, T. sp. indet. and T. c. corsica (representative selection).

Appendix S3. Variability of ventral side of abdomen (hair cover, coloration) in live specimens of Tibicina longisyllaba, T. sp. indet. and T. c. corsica (representative selection).

Appendix S4. Syllable periods (SYP) of European Tibicina species. Adapted and completed from Sueur & Aubin (2003).