This study generated 90 sequences belonging to 13 European cicada species and reports for the first time the genetic relatedness among all nine cicada species described to date within genus Tettigettalna. Primers LepF and LepR (Hajibabaei et al. 2006) were efficient to obtain COI fragment sequences for DNA barcoding in all tested cicada species, except for Tympanistalna gastrica. The coamplification of nuclear copies of mitochondrial genes has been commonly reported and might be favoured when universal primers are used (Song et al. 2008; Moulton et al. 2010). Therefore, the redesign of more efficient primers is probably necessary to attain the reliable amplification of mitochondrial COI copies in T. gastrica.
The total GC content in COI sequences varied among Tettigettalna species (30.9–34.0%). These values are similar or slightly larger than the ones observed in other Hemiptera (Philaenus spumarius, GC = 29.7%, Seabra et al. 2010; Ceroplastes spp., GC = 20.4%, Deng et al. 2012) but are within the range reported for several other insect orders (27.7–39.5%, Hebert et al. 2003a). Although the sampling size is modest for most species here investigated, some showed high levels of haplotype diversity (T. argentata and T. h. galantei), while others show very little sequence variation (T. aneabi).
Genetic distance and DNA barcoding
The genetic distance based on COI sequence variation between cicada genera investigated here (Tettigettalna, Tettigettacula, Tympanistalna and Cicada) is clear (> 9.0%) and enables the use of this marker as an efficient DNA barcoding tool for genus-level assignment. However, the power of COI to clearly diagnose closely related species within the genus Tettigettalna is compromised by some overlap of intraspecific and congeneric genetic distances. Initial DNA barcoding tests with COI in animal groups suggested that correct species diagnosis would be assured when a clear gap exists between mean intra- and interspecific divergence and the later should exceed the former by at least an order of magnitude (Hebert et al. 2004a). This premise indeed holds true for most animal groups and lineages, but recent studies showed that within- and among-species divergence may overlap more often that initially expected. The absence of such a DNA barcoding gap is usually attributed to recent divergence, incomplete lineage sorting, introgression, genetic geographical structure and insufficient sampling of geographical variation (Meyer & Paulay 2005; Wiemers & Fiedler 2007; Linares et al. 2009; Bergsten et al. 2012).
According to the present study, most Tettigettalna species appear well differentiated but show limited genetic divergence, and their origin probably results from relatively recent speciation events. Similarly, some studies with other cicada genera suggested episodes of species radiation to explain diversification patterns (Buckley & Simon 2007; Sueur et al. 2007; Marshall et al. 2008). For this study, the barcoding gap was considered as the difference between maximum intraspecific distance and the smallest distance to the closest congener, because the use of mean interspecific distances can produce artificially inflated barcoding gaps (Meier et al. 2008). T. josei and T. estrellae are the only species in the genus that fully pass DNA barcoding tests: they are both monophyletic and there is a clear gap between intraspecific variation and genetic distance to every other Tettigettalna spp. that enables their correct identification. The species T. boulardi apparently passes most DNA barcoding tests too, but only two specimens were analysed and intraspecific variation might have been underestimated.
Three Tettigettalna species show geographical genetic structure: T. argentata, T. helianthemi and T. defauti. Population genetic structure in T. argentata could be expected given its wide distribution range (Fig. 2). Although geographical variation in male calling songs has not been documented yet in this species, northern populations are genetically distinct from the southern ones. For T. helianthemi and T. defauti, population genetic structure is higher than expected from their restricted distribution range, and in both cases, cryptic divergence was detected in deep association with the Sierra Nevada mountain range (Fig. 2). These mountains are part of the Betic Cordillera in southern Spain, which is considered as a hotspot for Mediterranean biodiversity, harbouring many endemic species (Médail & Quézel 1999; Hewitt 2011). Changes in vegetation and climate are steep along the altitudinal cline, contrasting with the semi-arid lowlands. The two T. helianthemi subspecies recognized by Puissant and Sueur (2010) are morphologically distinct, have different calling songs and are geographically segregated. While T. h. helianthemi subspecies is associated with dry scrubland plains in the southeast of Sierra Nevada, T. h. galantei seems to be less thermophilous and can be found up the mountains till the top (circa 2000 m of altitude). The distinctiveness of these subspecies is clearly supported here by genetic divergence, but we also found that T. h. galantei is polyphyletic based on COI sequences. Three specimens collected in the southwestern slope of Sierra Nevada (near Lanjarón, Table S1, Supporting information) are divergent and were recovered as more related to T. boulardi, a species apparently limited to the eastern provinces of Murcia and Valencia. These results indicate that T. h. galantei from that area must be flagged for further taxonomic scrutiny, as these individuals might represent a cryptic species not yet described. The relationship between T. defauti and T. armandi also remains unclear as evidenced by the unresolved polytomy in the Bayesian analysis, although T. armandi is clearly undersampled and further investigation is needed to clarify this relationship.
Finally, T. argentata, T. mariae and T. aneabi show little COI differentiation and they seem to form a species complex. The cluster composed by these three species corresponds to the most derived haplotypes within Tettigettalna. Because these species have distinct male calling songs, acoustic behaviour probably plays an important role in reproductive isolation within this complex, but it remains unclear whether haplotype sharing between T. mariae and T. argentata is due to introgression or incomplete lineage sorting. The distribution of T. mariae is restricted to a small coastal region in the south of Portugal, and this species is often found in sympatry or close parapatry with T. argentata (VLN and RM, personal observation). Thus, acoustic behaviour isolation must be critical to prevent hybridization among this sibling species pair, and further studies involving mate preference tests are needed to address this question.
Despite the success of COI in DNA barcoding for several animal groups, the use of a single DNA fragment is often not enough to diagnose closely related species, as it happens in several examples among insects (e.g. Kaila & Ståhls 2006; Meier et al. 2006; Wiemers & Fiedler 2007; Langhoff et al. 2009; Žurovcová et al. 2010). Some authors advocate the use of a multigene approach for animals, as commonly implemented in plant DNA barcoding (Elias et al. 2007; Dupuis et al. 2012). The use of nuclear genes could help to control for misleading COI patterns caused by pseudogenes or species hybridization. However, finding informative nuclear genes for recently diverged species (as we might suspect for T. mariae) is challenging because nuclear genes usually evolve much slower than mitochondrial ones. Dasmahapatra et al. (2010) demonstrated the utility of AFLPs to determine species limits in butterflies, but a multilocus approach with anonymous loci will probably defeat the purpose of DNA barcoding as a universal, easy to implement, fast and cost-effective tool for species diagnosis.