The focus of the project was Cryptops from Germany, which encompass 85% of the here analysed specimens of the genus (Fig. 1). The remaining 15% (11) successfully sequenced specimens of Cryptops were collected in adjacent countries. Our sample includes six specimens from Austria, two from Italy, and one each from Croatia, Wales, and Slovenia. One of the Italian specimens is of special importance as it came from the type locality of the subspecies Cryptops parisi sebini Verhoeff, 1934. All specimens are stored as vouchers in 95% undenatured ethanol, either at the Museum Koenig, Bonn, Germany (ZFMK), the Senckenberg Museum für Naturkunde, Görlitz (SMNG) or the Bavarian State Collection of Zoology, Munich, ZSM (see Table 1, full specimen information in Suppl. material 1).

Figure 1. 

Distribution map of all successfully sequenced Central European specimens of Cryptops. Numbers refer to each specimen (see Table 1). Symbols and colours denote species. Blue rectangle = C. parisi; red circle = C. anomalans; green triangle = C. hortensis; brown diamond = C. croaticus; orange cross = C. umbricus; light blue, orange, and yellow symbols mark undetermined Cryptops species.

The specimens were collected by hand and transferred to vials containing 95% undenatured ethanol within days of collection. The vials contain an individual GBOL number with which the specimens can be connected to the accompanying data. After conservation the specimens were either sent to the GBOL facility at the ZFMK or to the corresponding laboratory at the ZSM. Upon arrival, all specimens were photographed (images are or will be uploaded to BOLD, http://www.boldsystems.org/), and a tissue sample was removed for DNA extraction. For this specific GBOL subproject, DNA extraction was attempted for 77 specimens of Cryptops as well as one each of Scolopendra cingulata and Theatops erythrocephalus as outgroups (See Table 1).

Maps were created with ArcGIS 10.

At the ZFMK, DNA was extracted from the tissue samples using the BioSprint96 magnetic bead extractor by Qiagen (Germany). After the extraction, samples were outsourced for PCR and sequencing (BGI China). For PCR and sequencing, the degenerated primer pair HCOJJ/LCOJJ (Astrin and Stüben 2008) was used, resulting in a success rate of >75% (38 of 49 extracted specimens).

At the ZSM, a single leg was removed from each specimen and sent in 96 well lysis plates to the Canadian Centre for DNA Barcoding (CCDB, Guelph, Canada) for standardized, high-throughput DNA extraction, PCR amplification and bidirectional Sanger sequencing (http://www.ccdb.ca/resources.php). For PCR and sequencing, a primer cocktail (Hebert et al. 2004, see Table 2) was used, resulting in a success rate of >90% (23 from 25 extracted specimens). All voucher information and the DNA barcode sequences, primer pairs and trace files were uploaded to BOLD (http://www.boldsystems.org).

Sequences were obtained for 61 Cryptops as well as the two outgroup specimens. The three available sequences of Central European Cryptops were added from a previously published dataset (Spelda et al. 2011). Sequence identities were confirmed with BLAST searches (Altschul et al. 1997). All 63 new sequences were deposited in GenBank (see Table 1 for accession numbers). In order to rule-out the accidental amplification of nuclear copies of the mitochondrial COI gene, the whole dataset was translated into amino acids (see Supplemental Material) following the ‘invertebrate’ code in MEGA6 (Tamura et al. 2013); internal stop codons were absent in our dataset. There were a total of 657 positions in the final dataset, gaps were absent.

Sequences were aligned by hand in Bioedit (Hall 1999). The final dataset included 66 nucleotide sequences with 657 positions (63 newly sequenced). Phylogenetic analyses were conducted in MEGA6 (Tamura et al. 2013). A Modeltest, as implemented in MEGA6 (Tamura et al. 2013), was performed to find the best fitting maximum likelihood substitution model. Models with the lowest BIC scores (Bayesian Information Criterion) are considered to describe the best substitution pattern. Included codon positions were 1st+2nd+3rd+Noncoding. Modeltest selected the General Time Reversible model (Nei and Kumar 2000) with gamma distribution and invariant sites as best fitting model (lnL -4725.286624, Invariant 0.505, Gamma 1.65919, R 3.11, Freq A: 0.2844, T: 0.3433, C: 0.2113, G: 0.1606). The tree with the highest log likelihood (-4725.2866) is used here to infer the genetic distances and evolutionary history of the analyzed specimens. Initial tree(s) for the heuristic search were obtained automatically by applying Neighbor-Join and BioNJ algorithms to a matrix of pairwise distances estimated using the Maximum Composite Likelihood (MCL) approach, and then selecting the topology with superior log likelihood value. A discrete Gamma distribution was used to model evolutionary rate differences among sites (5 categories (+G, parameter = 1.6591)). The rate variation model allowed for some sites to be evolutionarily invariable ((+I), 50.5% sites). The bootstrap consensus tree inferred from 1000 replicates (Felsenstein 1985) is taken to represent the evolutionary history of the analyzed taxa. The tree is drawn to scale, with branch lengths measured in the number of substitutions per site.

The number of base differences per site between sequences is shown in figures and tables (Fig. 3; Suppl. material 2). The analysis involved 66 nucleotide sequences. Codon positions included were 1st+2nd+3rd+Noncoding. All ambiguous positions were removed for each sequence pair. There were a total of 657 positions in the final dataset. Evolutionary distance analyses were conducted in MEGA6 (Tamura et al. 2013). Two frequency distribution diagrams of all pair-wise intra- and inter-specific distances were produced to further evaluate species divergence in Cryptops. All samples of each species were grouped in the first analysis, while Cryptops parisi was split into the three separate clades C. parisi sensu stricto, C. parisi sebini and C. parisi lineage3 in the second analysis.

The monophyly of the genus Cryptops is strongly supported (97%) in our tree (Fig. 2). One undetermined Cryptops sp. collected from the tropical rainforest greenhouse at the Leipzig Zoo in eastern Germany (Fig. 1: 55) is in a basal position juxtaposed to all other Cryptops specimens (Fig. 2). The remaining Central European Cryptops are split into two clades, of which only the C. parisi/C. croaticus clade receives high statistical support (96%). The unsupported clade unites C. umbricus and C. anomalans, three specimens of uncertain identity, and C. hortensis (Fig. 2). C. anomalans is in a basal position regarding this second (unsupported) clade with a single haplotype spread all over Germany, forming a monophylum with C. umbricus from Solnhofen, Germany, representing the first record from this country (Fig. 1: 46). The uncertain Cryptops sp. from Slovenia is a sister group to a weakly supported clade (76% bootstrap support) uniting two unidentified Cryptops sp. specimens with C. hortensis (Fig. 2). The latter two unidentified Cryptops sp. specimens from eastern Austria and Croatia are grouped together, but this grouping is not statistically supported.

Figure 2. 

Maximum likelihood tree under the GTR+G+I model, 1000 bootstrap replicates. Colours and symbols correspond to Maps (Figs 1, 4). Country of origin given after specimen number: AT = Austria; DE = Germany; GB = Wales; HR = Croatia; IT = Italy; SL = Slovenia. Photograph shows a specimen of Cryptops parisi s.s. from Breckerfeld (photo A. Steiner), western Germany. For full data on all specimens, see Table 1.

Figure 3. 

Frequency distribution of pairwise intraspecific (blue) and interspecific (red) distances. All lineages of C. parisi treated as one species, C. parisi sensu lato. Basic table see Suppl. material 1.

The monophyly of the 18 specimens of C. hortensis is strongly supported (100%). Of the shallow clades inside C. hortensis (Fig. 2), only one, a clade uniting five different haplotypes from Italy, eastern and western Germany (Fig. 4), receives some statistical support (78%). Interestingly, a second specimen from Friedeburg, Saxony-Anhalt, the same locality as one of the five haplotypes mentioned above (see Table 1), groups within a separate clade (Fig. 2).

The clade uniting C. parisi sensu lato and C. croaticus receives high statistical support (96%). While both specimens of C. croaticus show the same haplotype, the 32 specimens of C. parisi s. l. are separated into three statistically well-supported (99–100%) clades. The basalmost clade (Fig. 2) includes seven specimens and represents three different haplotypes from the eastern alpine region (Fig. 4: green). The remaining two clades of C. parisi are clearly related (92% support); one represents a western clade (Fig. 4: yellow) and the other is found slightly more to the east (Fig. 4: blue) and also includes the topotypoid of the subspecies C. parisi sebini.

Figure 4. 

Distribution map of all successfully sequenced Central European specimens of Cryptops parisi. Different colours mark the three different clades. Yellow = C. parisi sensu stricto; blue = C. parisi sebini; green = C. parisi lineage 3 (potentially C. cf. hortensis sensu Pichler 1987).

Cryptops specimens differ from the outgroups Scolopendra and Theatops by 19.8–25.7% (Supplementary Material 2). Interspecific and intraspecific distances of the different nominal Cryptops species show no overlap (Fig. 3). Interspecific distances lie between 13.4–21.1% (Fig. 3), with the lowest observed between C. croaticus and C. parisi s. l. (13.4–14.8%) as well as between C. anomalans and C. umbricus (13.9%). Otherwise, interspecific distances are always >16%, with the highest value of >20% observed between C. anomalans and C. hortensis, as well as between C. parisi sebini and C. hortensis. Intraspecific distances are between 0–11.3%. However, intraspecific distances are low, 0–3.3%, if we treat the three distinct lineages of C. parisi as distinct species (Fig. 5).

Figure 5. 

Frequency distribution of pairwise intraspecific (blue) and interspecific (red) distances. The three lineages of C. parisi treated as different species. Basic table see Suppl. material 1.

Clear intraspecific distances in German or even Central European Cryptops are low. The specimens filling the majority of our barcoding gap between 3 and 11.3% are the different lineages of C. parisi, which differ by 8.4–11.3% from one another (Fig. 3) and might represent distinguishable taxonomic units (see below). Two specimens directly at the edge between inter- and intraspecific distances (Fig. 5), the two Cryptops “sp. 2” specimens from Austria and Croatia (15.9%), require a careful re-study (see below).

Cryptops parisi and C. hortensis belong to the South European and Central Asiatic European chorotypes respectively (Zapparoli 2006). In central Europe C. hortensis and C. parisi s. l. seem to exclude each other either geographically or ecologically. In the lowland areas of north-western Germany and in the Upper Rhine valley it is usually C. hortensis that occurs, while in the lower mountain ranges usually C. parisi is present. Nevertheless, C. parisi mainly avoids higher altitudes. In the eastern part of Germany C. parisi dominates.

Cryptops parisi is generally classified as a mesophilous woodland species (Spelda 1999, Minelli and Iovane 1987, Voigtländer et al. 1997), but may also occur outside of forests, especially in northern Germany where more anthropogenic influenced places are inhabited.

The two clearly differentiated genetical lineages in C. parisi s. s. in Germany (see below) are reflected in distinct ecological differences in the preferred habitats between the western and eastern parts of Germany. In the more Atlantic areas in the West, the species prefers woodland like in its main distribution area. In the more continental influenced East, C. parisi inhabits open-dry habitats such as dry meadows, mesoxeric meadows and their successional shrub-stages, as well as dwarf-shrub heaths (Voigtländer 2003a, 2003b, 2005).

C. anomalans is viewed as a species introduced to Germany and England (Eason 1964; Voigtländer 1988). Specimen records are rare, e.g. the species has only recently been recorded from Germany, where it only occurs in localized areas, usually in parks or gardens (Lindner 2010, Decker and Hannig 2011). Our findings show that a single haplotype (Fig. 2) is present in western, eastern and southern Germany (Fig. 1), while all other Cryptops (see Fig. 3), as well as Geophilomorpha species (Wesener et al. 2015) show different haplotypes across a large geographical area. An identical haplotype from different localities might be interpreted as recent human introductions from a homogenous source population or a rapid spread of C. anomalans in Germany.

Our analyses first showed one outlier C. anomalans specimen from Solnhofen, Bavaria (Fig. 1), which strongly differs by 13.9% from the common German haplotype. This was the only specimen of C. anomalans in a previous analysis involving German centipedes (Spelda et al. 2011). A morphological check against similar species showed that it was indeed not C. anomalans but represents C. umbricus, a first record for Germany. This finding shows the usefulness of the barcoding method in detecting previously unrecorded species.

Cryptops sp. 1 is only represented in our dataset by a single specimen from Slovenia, which is unfortunately missing the pre-ultimate legs and can therefore not easily be determined morphologically.

Cryptops sp. 2 is represented by two specimens that are separated by a wide genetic distance of 15.9%. This distance usually falls right into the lower limit observed between different Cryptops species (Fig. 3). The two specimens are from the eastern lowlands of Austria (Burgenland) and Croatia (Brestova). Unfortunately, the Austrian specimen is heavily damaged with missing posterior segments, which prevents any determination. As both specimens of Cryptops sp. 2 are related, but potentially not conspecific, they are discussed here together.

These two specimens are similar to C. hortensis, but are missing the ventral furrow on the prefemora of the ultimate leg pair. An available name for one of these lines might be C. rucneri Matic, 1966. This species was synonymised with C. hortensis by Koren (1986), followed by Spelda (1999), but treated as a valid species later (Stoev (2002). The presently discovered genetic diversity brings this name into consideration again. One argument for the identity of one of our lines with C. rucneri is the configuration of the prefemur of the ultimate legpair, where Matic (1966) did not mention a ventral furrow. Although Matic (1966, 1972) did not describe and depict the poison gland in great detail, his figures clearly show that in both C. hortensis sensu Matic (1972a) and C. rucneri, the calyx of the poison glands lie mainly in the femur and tibia of the forcipule. Matic also records C. rucneri from Italy (Matic 1967), Austria: Carynthia (Matic 1972b), and Slovenia (Matic 1979).

Maybe this specimen is the same species to which Pichler (1987) refers to as Cryptops cf. hortensis from North Tyrol. The shape of the poison gland was not illustrated for C. cf. hortensis. The poison gland allows a clear separation from C. parisi even in very early stages. Without checking the poison gland, juvenile specimens of C. parisi, which lack the characteristics of adult specimens (a central depression on the forcipular tergite and the pair of occipital sutures), can be easily mistaken for C. hortensis. Pichler (1987) records an unidentate labrum for C. cf. hortensis, as does Matic (1966) for C. rucneri. Pichler’s (1987) fig. 18 of the 21st pleurocoxa corresponds to fig. 4 of Matic (1966) for C. rucneri.

Of the two specimens of Cryptops sp. 2, the one from Brestova is the most probable to represent C. rucneri. This specimen was collected only 30 kilometres distant from the type locality of C. rucneri and shows the characteristic elongated 20th leg pair, which is unfortunately missing in the other specimen (as well as in our Cryptops sp. 1). Nevertheless, while having only three sequences of these eastern C. hortensis-relatives and without being able to provide a revision of the hortensis/rucneri-complex we prefer at the moment to keep these specimens under the name Cryptops sp.

Cryptops sp. 3, previously determined as C. cf. doriae Pocock, 1891 is only known from the Leipzig Zoo in eastern Germany, where it was collected in a large tropical greenhouse (Decker et al. 2014). It was provisionally identified as C. doriae, a member of the doriae-group, which is characterized by having teeth on femur, tibia and tarsus of the ultimate legs (Lewis 2011). C. doriae was already reported from a tropical biome in England (Lewis 2007) and is so far the only introduced tropical Cryptops species with records in Europe (Stoev et al. 2010). A BLAST search of our specimen against the sequences of C. doriae already deposited on GenBank (11.2015) reveals a large genetic distance between our specimen and the ones from the Pacific, which is the reason we refer to our specimen as Cryptops sp. 3.

Cryptops croaticus was originally described from Bakar (formerly Buccari) in Croatia (Verhoeff 1931) and subsequently recorded from other localities in Croatia, Slovenia and Bosnia-Herzegovina (Matic 1966, 1979, Kos 1992), Greece (Matic 1976), Bulgaria (Stoev 1997a, 2002), and Italy (Matic 1960, 1968, Matic and Darabantu 1971, Minelli 1985, 1992). Currently, C. croaticus seems to be absent or not yet found in Hungary (Dányi 2008). One subspecies (C. croaticus burzenlandicus) was described from Romania (Verhoeff 1931) and was subsequently synonymised with the nominal subspecies (Matic 1972a), another subspecies, C. croaticus albanicus, has been described from Albania (Verhoeff 1934) and was later synonymized under C. anomalans (Stoev 1997b). Several subspecies have been described from Italy, namely C. croaticus bergomatius (Verhoeff 1934), C. croaticus longobardius and C. croaticus baldensis (Manfredi 1948), subsequently cited by Conci (1951) and Boldori (1969). Based on this wide distribution, the occurrence of C. croaticus in Austria is not unexpected. In Austria, it is currently only known from a southern exposed slope, which is home to numerous relic species adapted to a warmer climate. C. croaticus shares its habitat with the recently rediscovered population of Scolopendra cingulata in Austria (Oeyen et al. 2014), as well as the thermophilic beetle Carabus hungaricus and other thermophilic animals (Böhme et al. 2014). However, the determination of our specimens as C. croaticus is only based on the characters given in the original description (Verhoeff 1931) as no better description exists. Numerous important characters, such as the last leg pairs, are unfortunately missing in our specimens. A revision of C. croaticus is urgently needed (Matic 1966) as it may be that some of the nominal subspecies represent independent species. One way to clarify this is to collect and sequence topotypic material. Once C. croaticus has been properly revised, a re-evaluation of the Austrian specimens should be undertaken.

The three lineages of specimens placed in C. parisi by morphological characters differ 8.4–11.3% from one another, while their intra-lineage genetic distance is much lower at 0–1.1%. A large barcoding gap becomes clearly visible in our dataset when we treat the three different lineages of C. parisi as separate species (Figs 4, 5). Endosymbionts like Wolbachia (Hurst and Jiggins 2005) are an unlikely explanation for the different lineages, as such endosymbionts have never been recorded in the Myriapoda (Witzel et al. 2003).

One lineage clearly represents the C. parisi sensu stricto (Fig. 2: yellow). This group shows a western distribution in Germany, with a single specimen from southern Germany (Fig. 4). The type locality of C. parisi is, as the species epithet implies, Paris, France. Our only sample from Great Britain (Wales) also falls into this group. Intra-lineage variation is low with 0–1.7%. Inner structure of the lineage is limited due to the small genetic distances inside the group, but one group containing only few haplotypes differing in a single or two basepairs from one another is well-supported. This group contains specimens from western Germany, as well as a single specimen each from southwestern (ZFMK-TIS 2520349) and southeastern Germany (ZFMK-DNA-112780049), but these two were collected in a park and a garden.

A second distinct group (Fig. 2: Blue) contains the topotypic specimen of the subspecies C. parisi sebini Verhoeff, 1934. C. parisi sebini was recently synonymised under C. parisi because no morphological differences could be detected (Lewis 2011). However, the distinctiveness of the subspecies C. parisi sebini should be re-evaluated, as our genetic data supports this monophyletic subspecies (100% bootstrap support) with a high genetic distance to C. parisi s. s. (8.4–9.4%) in combination with low intra-lineage variation (0–0.6%) despite the large geographical distances between the analyzed specimens from Italy and eastern Germany. This C. parisi group 2 shows a distribution to the east of C. parisi s. s., with localities in eastern northern Italy and the eastern half as well as the south of Germany (Fig. 4). Another name potentially available for this clade is C. parisi rhenanus Verhoeff, 1931, which is characterized by its extremely elongated calyx of the poison gland (Verhoeff 1931). If both names turn out to represent the same species, this taxon would have priority over C. parisi sebini, with which it is compared in the original description (Verhoeff 1934). Unfortunately, Verhoeff (1931) never designated a type for C. parisi rhenanus. The specimens represented in the Bavarian State Collection of Zoology originate from a large number of localities.

The specimens of C. parisi s. l. belonging to a third group (Fig. 2: green), referred here as C. parisi lineage 3, are morphologically and genetically distinct and may also be identical to the specimens of C. cf. hortensis in the literature (Pichler 1987, Lewis 2011). Our specimens of C. parisi lineage 3 come mainly from alpine habitats in Austria and Germany. In the most recent revision of the species group (Lewis 2011), these specimens were listed in the key under C. parisi, but with remarks concerning its unique morphology. Coxal pores are too numerous (~50) for C. hortensis and more closely resemble the lower end of C. parisi. Other morphological characters prompted Lewis (2011) to place these specimens in his key under C. parisi, an affinity confirmed here by our genetic analysis.

However, the large genetic distance of 10–11.3% between C. parisi lineage 3 and C. parisi s. s. as well as to the lineage containing C. parisi sebini, combined with a low intraspecific distance (0–1.1%) are clear indications that these specimens might represent a species of its own.

To find names for our two eastern lines of C. parisi one has to go back to C. L. Koch, who described three Cryptops species from around Regensburg, Germany: C. ochraceus C. L. Koch, 1844 from the Keilstein (a calcareous mountain east of Regensburg), C. sylvaticus C. L. Koch, 1844 from the Naab-valley (north of Regensburg) and C. pallens C. L. Koch, 1847 from the moat of Regensburg. More information on these species, such as the precise type localities and more detailed descriptions, are provided in Koch (1863), which has often resulted in these species erroneously being assigned to the date of this second publication.

Attems (1930) indicated that it would be impossible to assign these species to either C. hortensis or C. parisi, while Matic (1972a) simply synonymized them with C. hortensis. Both did not take note of the central depression, often darker than the adjacent parts of the tergite, as a character separating C. parisi from C. hortensis, at least for adult specimens from southern Germany (own observation, JS). This depression is also described by Attems (1930) as existing in some C. parisi specimens, but is not otherwise mentioned in the available keys separating the two species (Attems 1930, Brölemann 1930, Verhoeff 1931, Eason 1964, Matic 1972a, Koren 1982, 1986, Iorio and Geoffroy 2008). Verhoeff (1934) also described this character in C. parisi sebini. Koch (1863) clearly states and depicts the depression for his species C. sylvaticus and C. ochraceus. It seems only to be missing in C. pallens, which represents a juvenile specimen. Another argument against a synonymy of these species with C. hortensis is the absence of the latter species in our extensive collections from eastern Bavaria. Topotypoids of C. ochraceus have already been collected and might clarify this species in the near future.

It should be noted that Matic (1972a) depicts a C. parisi with a short poison gland. This specimen surely represents a different species.

Future prospects should include the parallel sequencing of nuclear genes to confirm the relationships drawn from the mitochondrial barcoding fragment. To clarify the taxonomic relationships within Cryptops parisi, it would be important to collect further samples to enable an extensive morphological evaluation.