The centipede genus Eupolybothrus Verhoeff, 1907 (Chilopoda: Lithobiomorpha: Lithobiidae) in North Africa, a cybertaxonomic revision, with a key to all species in the genus and the first use of DNA barcoding for the group

Abstract The centipede genus Eupolybothrus Verhoeff, 1907 in North Africa is revised. A new cavernicolous species, Eupolybothrus kahfi Stoev & Akkari, sp. n., is described from a cave in Jebel Zaghouan, northeast Tunisia. Morphologically, it is most closely related to Eupolybothrus nudicornis (Gervais, 1837) from North Africa and Southwest Europe but can be readily distinguished by the long antennae and leg-pair 15, a conical dorso-median protuberance emerging from the posterior part of prefemur 15, and the shape of the male first genital sternite. Molecular sequence data from the cytochrome c oxidase I gene (mtDNA–5’ COI-barcoding fragment) exhibit 19.19% divergence between Eupolybothrus kahfi and Eupolybothrus nudicornis, an interspecific value comparable to those observed among four other species of Eupolybothrus which, combined with a low intraspecific divergence (0.3–1.14%), supports the morphological diagnosis of Eupolybothrus kahfi as a separate species. This is the first troglomorphic myriapod to be found in Tunisia, and the second troglomorph lithobiomorph centipede known from North Africa. Eupolybothrus nudicornis is redescribed based on abundant material from Tunisia and its post-embryonic development, distribution and habitat preferences recorded. Eupolybothrus cloudsley-thompsoni Turk, 1955, a nominal species based on Tunisian type material, is placed in synonymy with Eupolybothrus nudicornis. To comply with the latest technological developments in publishing of biological information, the paper implements new approaches in cybertaxonomy, such as fine granularity XML tagging validated against the NLM DTD TaxPub for PubMedCentral and dissemination in XML to various aggregators (GBIF, EOL, Wikipedia), vizualisation of all taxa mentioned in the text via the dynamically created Pensoft Taxon Profile (PTP) page, data publishing, georeferencing of all localities via Google Earth, and ZooBank, GenBank and MorphBank registration of datasets. An interactive key to all valid species of Eupolybothrus is made with DELTA software.


Introduction
Th e lithobiid subfamily Ethopolyinae is represented in Europe and Africa by a single genus, Eupolybothrus Verhoeff , 1907, which currently comprises around 20 valid species as well as a few poorly known species and subspecies, collectively arranged in seven subgenera (Zapparoli 2003, Minelli 2006. Th e genus ranges from Central and South Europe to the Middle East and Maghreb with highest species diversity in the Appenine and Balkan peninsulas (Zapparoli 2003). In North Africa, Eupolybothrus is known only from a single, quite ubiquitous species, E. nudicornis (Gervais, 1837), the range of which extends from northern Morocco to northwestern Libya. It also occurs in France, mainland Spain and Italy, as well as in several West Mediterranean islands (Manfredi 1939, Minelli 2006. Th e identity of E. nudicornis has been a subject of controversy for more than a century. Th e polymorphic external anatomy shown by the species throughout its broad geographic range led to the description of several morphologically similar taxa that were sometimes based only on a single type specimen (Minelli 2006). Currently the list of named taxa allied to E. nudicornis comprises six species and eight subspecies/varieties described from diff erent Mediterranean countries: E. koenigi (Verhoeff , 1891), E. lebruni (Dobroruka, 1968), E. monilicornis (Newport, 1849), E. elongatus alpinus (Brolemann, 1930), E. elongatus aprutianus (Manfredi, 1950), E. elongatus calabrus (Manfredi, 1933), E. elongatus imperanus (Verhoeff , 1937), E. elongatus levis (Verhoeff , 1943), E. elongatus oraniensis (Verhoeff , 1901), E. elongatus sardus (Manfredi, 1956), E. impressus corsicus (Brölemann, 1903), E. cloudsleythompsoni Turk, 1955, E. osellai Matic, Floca & Hurezeanu, 1992, and E. ruff oi Matic, Floca & Hurezeanu, 1992. Th e list has been reduced through the course of taxonomic revisions, with most of the names eventually being considered synonyms of E. nudicornis. Currently the taxonomic position of three of them, E. cloudsleythompsoni from Tunisia, E. osellai and E. ruff oi, both from Italy, is uncertain and needs reappraisal (Minelli 2006). A more comprehensive study of the whole group has never been undertaken and it is likely that some synonyms will turn out to be valid species after contemporary morphological and molecular methods are applied.
Th e aim of present paper is to put on record all North African material of Eupolybothrus amassed during recent years and also found in old collections of diff erent European museums. We redescribe E. nudicornis and describe a new species discovered in a cave in Tunisia. Th e new species is distinguished from the nearest congener morphologically as well as using the cytochrome c oxidase I gene (mtDNA-5' COI-barcoding fragment). We also discuss the morphological variability and postembryonic development of E. nudicornis and provide an overview of its habitat preferences and distribution in Africa. We outline some of the existing taxonomic problems in the genus Eupolybothrus and provide a key to all currently valid species of the genus.
Historical account. Th e earliest record of the genus Eupolybothrus in North Africa was made by Carl Ludwig Koch (1841), who described Lithobius impressus from Alger and Oran in Algeria. A few years later, Lucas (1849) recorded the same species from other localities in the country: Lac Tonga and Houbeira, La Calle, Constantine, Bône, and Philippeville. Newport (in Lucas 1849) described two further species from Algeria, Lithobius monilicornis from Boudjaréa near Alger and L. elongatus from Lac Tonga, Houbeira and La Calle. Th e former was tentatively synonymised with L. impressus by Meinert (1872), who redescribed the species based on new material from Algeria. Eason (1972a) confi rmed Meinert's synonymy and suggested elongatus as a possibly good subspecies of E. impressus. Verhoeff (1891) described Lithobius koenigi from Tunis and ten years later Lithobius elongatus oraniensis from "Rio Salado, Chabael Ham, Djebel el Tessala, Saida and Hammam Bou Hadjar" (Verhoeff 1901), all in Algeria. Subsequently, L. koenigi was synonymised with Lithobius elongatus by Silvestri (1896) who recorded it also from Tunis, Carthage, Souk el Arba (now Jendouba), Aïn Draham, Babouch and Tabarka. Eason (1972a) proposed the synonymy of L. elongatus oraniensis with Eupolybothrus impressus.
In 1892 Pocock recorded Lithobius impressus from Kherrarta, Alger, Constantine and Hammam Ri'irha in Algeria, and from Tunis in Tunisia (Pocock 1892). Attems (1908) recorded Polybothrus koenigi from Aïn Draham in Kroumirie, Tunisia. In the checklist of North African myriapods Brolemann (1921) mentioned Bothropolys impressus for Tunisia, Algeria and Morocco, referring to it most of the previous records from North Africa (those of e.g., C.L. Koch 1841, Lucas 1849, Verhoeff 1891, Pocock 1892, Silvestri 1896, Attems 1908, and also questioned the occurrence of B. monilicornis and Lithobius elongatus in Algeria, and that of Lithobius nudicornis in Tunisia and Algeria. Subsequently, having material from Tipasa and Djebel Bou Zegza in Algeria, he revived koenigi as a variety of B. elongatus distinguishing it by the spinulation of leg-pairs 14 and 15 and the more slender prefemora of leg-pair 15 (Brolemann 1925(Brolemann , 1931a. Only these two taxa were included in his identifi cation key to the North African centipedes (Brolemann 1932). Silvestri (1897) was fi rst to draw attention to Lithobius nudicornis from Sicily being conspecifi c with E. impressus. Although this fact has been commented upon by several authors (e.g., Jeekel 1967, Eason 1972b it was Minelli (1983a) who quite recently validated the name. Minelli (1983a) wrote " Eason (1980) refers to this Sardinian Eupolybothrus under the name Eu. impressus (C.L. Koch); however, it is probable that a single species should be recognized in the complex nudicornisimpressuselongatus, with nudicornis as senior synonym. In any case, I agree with Eason in so far I recognize a single taxon of this group in Sardinia, despite some habitus diff erences between diff erent populations". All subsequent taxonomic publications where E. nudicornis is mentioned refer to it as a full species and do not recognize the existence of subspecies (e.g., Minelli 1983b;Foddai et al. 1995;Zapparoli 1994Zapparoli , 1995aZapparoli , 2006Zapparoli , 2009Zapparoli et al. 2004;Minelli 2006;Iorio 2008a,b).
Th e fi rst and hitherto only record of the genus in Libya comes from Manfredi (1939) who recorded Bothropolys elongatus königi from Bu Gheilan (Tripolitania). Turk (1955) described a new species, E. cloudsley-thompsoni, collected near a Roman aqueduct south of Tunis. Dobroruka (1968) described another species, Schizopolybothrus lebruni, from Djebel Mansour (Pont du Fahs) also in Tunisia. Zapparoli (1985) synonymised Dobroruka's species with E. impressus elongatus and suggested that E. cloudsley-thompsoni could also be its synonym but type specimens needed to be examined. Having on disposal specimens from Morocco (Rif ), Algeria (Djudjura Mts.) and Tunisia (Th ala, Nabeul), he confi rmed the earlier observation of Eason (1972a) that in North Africa the species is represented by two subspecies -E. i. impressus and E. i. elongatus, which can be readily distinguished by the tarsal spinulation of the penultimate and ultimate pairs of legs and by the shape of tergite 9. More recently, Zapparoli (1995a) recorded E. nudicornis from the Italian islands Lampedusa and Pantelleria, which are situated close to the Tunisian coast. Verhoeff (1907) split the genus Polybothrus into three subgenera, namely Allopolybothrus, Propolybothrus and Eupolybothrus. Crabill (1955) and Jeekel (1963) showed that they were validly proposed, opposing the opinion of Chamberlin (1925), as Jeekel (1963) designated Lithobius koenigi as the type species of subgenus Allopolybothrus. Subsequently, Jeekel (1967) reviewed the genus Eupolybothrus and resolved several nomenclatorial problems, also providing a list of all taxa assigned to the genus to that time. CHILOBASE (Minelli 2006) is a contemporary web-based database of centipede names and lists all currently valid species in the genus, nominal subgenera (see Jeekel 1963), and global species distributions.

Material and methods
Collections. Unless stated otherwise, the material treated herein has been collected from Tunisia in March 2008 andMarch 2009 by N. Akkari, P. Stoev andH. Enghoff , andalso in the course of individual excursions of N. Akkari to diff erent regions of the country in the period [2003][2004][2005][2006][2007][2008]. Th e material is preserved in 70% or 96% ethanol and is shared between the Field Museum of Natural History, Chicago (FMNH), National Museum of Natural History, Sofi a (NMNHS), Natural History Museum of Denmark, Copenhagen (ZMUC) and Biodiversity Institute of Ontario, Guelph (BIO). Additional type and non-type specimens of Eupolybothrus from North Africa housed in the Hungarian Natural History Museum (HNHM), the Natural History Museum, London (NHM), ZMUC and the private collection of Marzio Zapparoli (CMZ) were also incorporated in the present study. Photos were taken mainly with a Leica DFC 420 digital camera mounted on a Leica MZ16A stereomicroscope, and were processed using the program Automontage Pro software (Syncroscopy, Cambridge, UK) for image-stacking 3D focus expansion. Terminology for external anatomy follows Bonato et al. (in prep.). Molecular methods. Eleven specimens from 5 species were used for genetic examination of the divergence among species of the genus. Ten specimens that sample 4 species were barcoded in the context of a global campaign on Myriapoda initiated as a part of the 'Barcode of Life' project (iBOL WorkGroup 1.9 'Terrestrial surveillance') (Appendix C doi: 10.3897/zookeys.50.504-app.C). To this dataset we added a sequence from GenBank for a fi fth species, Eupolybothrus fasciatus (AY214420) (Edgecombe and Giribet 2004). Sequences are publicly available on BOLD (Ratnasingham and Hebert 2007; http://www.barcodinglife.org) within the project PSEKA and in GenBank (accession numbers in Table 6).
Lysis of the tissues was carried out in 50 μl volume of lysis buff er and proteinase K incubated at 56°C overnight. DNA extraction followed a standard automated protocol on 96-well glass fi bre plates (Ivanova et al. 2006). Th e 5' region of COI used as a standard DNA barcode was amplifi ed using M13 tailed primers LCO1490 and HCO2198 (Folmer et al. 1994). A standard PCR reaction protocol was used for PCR amplifi cations and products were checked on a 2% E-gel 96 Agarose (Invitrogen). Unpurifi ed PCR amplicons were sequenced in both directions using M13 tails as primers. Th e sequencing reactions followed standard protocols of the Canadian Center for DNA Barcoding (Hajibabaei et al. 2005), with products subsequently purifi ed using Agencourt CleanSEQ protocol (Agencourt) and processed using BigDye version 3.1 on an ABI 3730 DNA Analyzer (Applied Biosystems). Sequences were assembled with Sequencer 4.5 (GeneCode Corporation, Ann Arbor, MI, USA) and aligned by eye using BIOEDIT version 7.0.5.3 (Hall 1999). We observed no indels in this coding region of the mito-chondrial genome and therefore all base positions were aligned with confi dence in positional homology. Distance analyses were conducted with MEGA4 (Tamura et al. 2007) using a neighbor-joining (Saitou and Nei 1987) algorithm and distances corrected with the Kimura-2 parameter (Kimura 1980). Th e robustness of nodes was evaluated through bootstrap analysis of 1000 pseudoreplicates.
Cybertaxonomy. Th e present paper demonstrates several innovative methods of semantic tagging and semantic enhancements, text and data processing, publishing and dissemination in taxonomy, described in more detail in a forum paper published in the same issue (Penev et al. 2010). Among the most important semantic enhancements shown in the paper are: fi ne granularity XML (eXtensible Markup Language) tagging based on the US National Library of Medicine's DTD (Document Type Definitions) TaxPub extension (the tagging of the present paper was based on TaxPub Version 123, http://sourceforge.net/projects/taxpub); fi nal XML output of the paper validated against the NLM DTD TaxPub for archival in PubMedCentral and dissemination in XML to various aggregators, e.g., new descriptions to Encyclopedia of Life (http://www.eol.org) and all taxon treatments to Plazi (http://www.plazi.org); vizualisation of main tag elements within the text (e.g., taxon names, taxon treatments, DNA sequences, localities, type materials, etc.); mapping of localities listed in the whole paper or within separate taxon treatments; a dynamically created Pensoft Taxon Profi le (PTP) page for each taxon name mentioned in the paper; Genbank accession numbers autotagged and linked to the National Center for Biotechnology Information (NCBI, http://www.ncbi.nlm.nih.gov) and Barcode of Life (BOLD, http://www.boldsystems. org); external links of references to PubMed, Google Scholar, Biodiversity Heritage Library and/or other sources.
All 70 images included in this publication have been deposited in MorphBank (Appendix D doi:10.3897/zookeys.50.504-app.D). All the revised species were registered in ZooBank and Life Science Identifi ers (LSID) were assigned to them. Accession numbers were obtained from BOLD (see Appendix C doi:10.3897/zookeys.50.504app.C for complete metadata) and GenBank for all COI gene sequences. Datasets in spreadsheet format for specimen localities have been shared with the Global Biodiversity Information Facility (GBIF) via Appendix A doi:10.3897/zookeys.50.504-app.A. To illustrate all records of the species in North Africa interactively in Google Earth, KML fi les were generated and are available for download as Appendix E doi:10.3897/ zookeys.50.504-app.E. Th e interactive key for identifi cation of all currently valid species of Eupolybothrus was made with DELTA software http://delta-intkey.com (Appendix F doi:10.3897/zookeys.50.504-app.F).
Remarks. Several taxa assigned to Eupolybothrus remain species inquirendae. Here we briefl y review the current status of these taxa. Eupolybothrus stygis was described from Iljina pećina (cave) near Trebinje in Bosnia and Herzegovina (Folkmanova 1940), and Stoev (2001b) suggested that it could be identical with E. leostygis, a troglobitic species known from the same area (Eason 1983). In the key below it keys out together with E. acherontis, another poorly known species from Bosnia and Herzegovina. Stoev (2001a,b) noted that E. spiniger, E. acherontis and E. acherontis wardaranus could be identical with E. caesar. Being the oldest available name, in case of synonymy E. spiniger would have priority over E. caesar. Th anks to Verena Stagl, curator of myriapods at NHMW, we were able to obtain a photograph of the prefemur of the ultimate leg pair of male E. spiniger which shows no diff erences with that of E. caesar. However, until we personally examine the types we prefer to treat E. spiniger and E. caesar as separate species. Eupolybothrus valkanovi was based on a single female with unusually short gonopodial spurs found near Asenovgrad, Rhodope Mts in Bulgaria (Kaczmarek 1973). According to Stoev (2002) it is most likely conspecifi c with the morphologically similar E. transsylvanicus which is also known to occur in the area. Eupolybothrus sketi was described from male and female specimens found in the Jakupica Mts, in the Former Yugoslav Republic of Macedonia (Matic 1979). It is listed under the possible synonyms of E. transsylvanicus by Minelli (2006) but its status is yet to be clarifi ed. Morphologically, it is most closely related to E. zeus from Greece. Eupolybothrus tabularum was synonymised under E. excellens by Minelli and Zapparoli (1985) but was recently found to be a good species. A paper on this subject is currently in preparation by M. Zapparoli and will be published elsewhere. Although both species share some traits in common (like 15VCm spine), the long median protuberance on the prefemur of leg 15 in males convincingly distinguishes E. excellens from E. tabularum (see key below). Eupolybothrus macedonicus is hitherto known only from its type locality, Temna cave near Loutraki, North Greece (Zapparoli 2002). Likewise, E. verrucosus is presently known only from its original description based on a single female specimen from Moldova (Minelli 2006 Turk collection 1984.10.1.77 (NHM).
Antennae moderately long, approx. 15 mm, reaching midline of T6 when folded backwards; approx. 37% length of body, composed of 43-44 articles; fi rst three articles enlarged, with second being the largest (Fig. 1a); antenna gradually tapering towards the end; articles 5-20 broader than long, ultimate article of antenna about same length as penultimate or slightly longer (Fig. 1c). Basal two articles less setose than the others, which are densely covered with trichoid setae. Clypeus (Fig. 1d) with a cluster of six medium-sized setae situated asymmetrically at the left half of its apex; central clypeal area smooth, without setae, basal part of clypeus with a single row of setae, lateral clypeal margins with very few dispersed setae. Forcipule (Fig. 1f ): coxosternite subhexagonal, lateral margins feebly convex; anterior margin set off as a rim by furrow that is impressed behind all teeth; coxosternal teeth 5+5, almost equal-sized; median diastema shallow, U-shaped; intradental distance varying, generally increasing towards lateral teeth; porodont arising from a small node at lateral coxosternal margin, situated below the dental rim, and well laterad from lateralmost tooth; base of porodont as thick as adjacent tooth; coxosternite smooth, with one or two rows of setae in close proximity to dental rim; dorsal side of coxosternite with sparse minute setae, the apices of which are not visible from the ventral side. Distal part of tarsungulum about 3-3.5 times longer than proximal part, devoid of setae. Forcipular trochanteroprefemur, femur and tibia fringed with few minute setae (Fig. 1f ).
Coxal pores: small, circular, more concentrated on the outer part of pore-fi eld, forming 3-4 irregular rows; only 2-3 pores from the internal row on each coxa larger;  around 25-30 on legs 12-14 and about 17 pores on leg 15; pores of inner rows often separated by more than twice their own diameter, those of outermost row usually separated by less than their own diameter (Fig. 2b).
Male fi rst genital sternite with emarginated posterior margin (Fig. 2b), posterior angles broadly rounded, sternal surface densely covered with numerous long brownish setae; gonopod hidden behind the edge of fi rst genital sternite; small, basal part larger covered with six setae, apical part with 2 setae.
Female gonopods with 2+2 moderately long, apically pointed spurs and a simple, falcate claw (Fig. 2c). First article with approx. 14 setae concentrated on small protuberance at its posterior part; posterior half of second article with approx. 20 dorsal and dorso-lateral setae of various sizes; gonopodial claw with 5 moderately long lateral setae.
Post-embryonic development. Meinert (1872) described immature stadia of E. nudicornis (sub Lithobius impressus), based on his study of specimens from Algeria, Granada (Spain) and Ischia (Italy). He recognized four classes: -"Pullus" with 10 pairs of legs + 2 incompletely developed pairs (= LIII) -"Pullus" with 12 pairs of legs + 3 incompletely developed pairs (= LIV) -"Juvenis" (smaller PL) -"Junior" (larger PL) His information is summarized in Table 2. No information on the number of specimens in each group is available. Meinert's data agree well with our own observations, except for the higher number of forcipular coxosternal teeth in the larval stadia.  Daas et al. (1996) studied post-embryonic development of E. nudicornis (sub E. elongatus Newport) in Algeria. Th e data from their table 2 are given here as Table 3. Th e fi rst larval stadia were reared from eggs, whereas the older stadia were obtained from fi eld-collected animals. By comparison to developmental series of other lithobiids, the numbers of legs given for LIII and LIV are anomalous (cf. Murakami 1958, Andersson 1981. Further data on the post-embryonic development based on Tunisian specimens of E. nudicornis are provided by Silvestri (1896: 149, sub L. elongatus) but they mostly refer to grown individuals and say nothing about the larval and earlier post-larval stadia (see Table 4).
In Table 4 we provide the character states for the diff erent larval and postlarval stadia obtained from part of the studied material. Th e defi nition of postlarval stadia follows Daas et al. (1996) and is based on the length of respective specimen.  Table 3. Character states in larval and postlarval stadia in E. nudicornis according to Daas et al. (1996, sub E. elongatus). Each entry is based on at least four observations.   Emberger (1966). Th e species occurs in the northwestern mountains of Kroumirie and Mogods (Aïn Draham, Béni Mtir, Hamman Bourguiba, Béja) where it interconnects with the populations in northern Algeria (La Calle, Constantine, Annaba, Skikda, Alger and Djurdjura Mts.). In the North it is known also from the coast (Bizerte, Ghar El Melh), from the plain of Mateur (Ichkeul National Park) and along the Gulf of Tunis (Nahli Park, Sidi Th abet, Tunis, Carthage). In the Cap Bon Peninsula the species is quite common along the coast (Nabeul, Oued el Abid, Sidi Erraiès, Kélibia, Dar Chichou, El Haouaria) but is also found inland, in Jebel Abderrahman. In Central Tunisia it occurs from the High Tell in the West (Le Kef, Tajerouine, Dugga, Nebeur), virtually from the whole Dorsale Ridge which stretches from Chambi and Kesra in the West to Jebel Bargou, Jebel el Fahs and Jebel Zaghouan in the East and further South from the plain of Kairouan (Sbikha, Th uburbo Majus) to the eastern coast in the so called Sahel (Sousse, Hergla, Békalta, Mahdia). In the South, the species was recorded from the mountains of Gafsa (Jebel Bou Ramli), the western Saharian platform (Midès) and from the Dahar Mts further east. It has been found as far south as Matmata and Tataouine.
Diagnosis. A species of Eupolybothrus with long antennae, approx. 90% length of body, composed of 68 articles; eyes composed of 18 ocelli; colour uniformly yellowwhitish; anterior margin of forcipular coxosternite with 7+7 teeth; TT 9, 11, 13 with posterior triangular projections; leg 15 56-57% length of body, with a single claw on pretarsus; prefemur of leg 15 with a long, conical dorso-median protuberance emerging from its posterior part and pointing posterior-dorsad; coxal pores generally large, round to ovoid; around 15-20 on legs 12 and 13 and about 20-24 on legs 14 and 15; posterior margin of male fi rst genital sternite straight.
Description. Holotype: Length (from anterior margin of cephalic plate to posterior margin of telson) approx. 30 mm; cephalic plate slightly broader than long (Fig.  3a); head 2.7 mm long, 3 mm wide; leg 15 aprox. 17 mm long, or 56-57% length of body. Colour generally uniformly yellow-whitish; only forcipular coxosternal teeth, posterior half of forcipular tarsungulum brown; anterior 1/3 of cephalic plate slightly darker yellowish; interrupted black line stretches along the midline of body and can be traced on all but last tergite. Cephalic plate smooth, wider than T1 (Fig. 3a); a median notch contributing to biconvex anterior margin; marginal ridge with a distinct median thickening occupying almost 50% breadth of plate; posterior margin straight or slightly convex; central part of cephalic plate concave; transverse suture situated at about 1/3 of anterior edge; posterior limbs of transverse suture visible, connecting basal antennal article with anterior part of ocellar area; setae on cephalic plate very few, dispersed, without regular arrangement. Ocelli: 1+3,4,5,5; seriate ocelli greyish, oval to elliptical, in 4 rows: fi rst seriate ocellus of the exteriormost row largest, ocelli of the middle two rows mediumsized, those of inferior row smallest; posterior ocellus as large as the fi rst seriate ocellus. Tömösváry's organ small, circular, situated on subtriangular sclerotisation immediately below the inferiormost row of seriate ocelli (Fig. 3b).
Left antenna long, approx. 27 mm, reaching or slightly surpassing posterior margin of T12 when folded backwards; 90% length of body, composed of 68 articles; right antenna damaged, composed of at least 34 articles; basal two articles enlarged (Fig. 3a), most articles longer than broad; last 12 articles more elongated than others; ultimate article about same length as penultimate (Fig. 3c). Basal two articles less setose than others, which are densely covered with trichoid setae.
Clypeus with a cluster of about 30-33 long to medium-sized setae situated at apex and near the lateral margin (Fig. 3d).
Forcipule (Fig. 3e): coxosternite subhexagonal, lateral margins feebly convex; anterior margin set off as a rim by furrow that is impressed behind all teeth; coxosternal teeth 7+7, inner tooth smaller than others, its apex well posterior to outer tooth; median diastema small, strongly narrowed by the inner teeth; intradental distance varying, generally increasing towards lateral teeth; porodont arising from a small node below the dental rim, situated posteriad to teeth and well laterad to lateralmost tooth; base of porodont as thick as adjacent tooth or slightly thinner; coxosternite densely setose anteriorly; setae generally long, in approximately 7-8 irregular rows; another row of long setae visible behind anterior margin. Forcipular trochanteroprefemur medially concave with a small subtriangular outgrowth emerging at its posterior part; distal part of tarsungulum about six times longer than proximal part, devoid of setae; forcipular prefemur, femur and tibia fringed with a row of setae (sparse and irregular on the posterior half of prefemoral part).
Origin of name. derives from the Arabic word kahf ‫)ﮐﻬﻒ(‬ meaning 'cave', and kahfi denotes 'living in cave'. Habitat. E. kahfi occurs in a chasm of approximately 30 m depth which after descending continues as a narrow horizontal gallery ending in a small hall. Th e total length of the cave is approximately 50 m. Th ere are just a few humid spots on the fl oor, with almost no organic substance. Th e juvenile specimen was collected creeping on the wall at the end hall, while the adult was found under a lump of clay, approximately one meter below the place of descent. In the cave E. kahfi co-occurs with troglomorphic isopods, spiders of the genus Meta C.L. Koch, 1836, pseudoscorpions of the genus Roncus L. Koch, 1873, harvestmen, troglobitic diplurans, trichopterans and gastropods. With very few exceptions, Tunisian specimens of E. nudicornis fi t the morphological diagnosis of E. n. elongatus well. Eason (1972a) reported on specimens with intermediate characters from Constantine, Algeria, and wrote "…it seems that the characters separating elongatus and impressus are unstable", and "..…in spite of the intermediate examples from Constantine, it seems advisable to retain, for the time being, the distinction between elongatus and impressus…but they should be regarded as only subspecifi cally distinct." Most of the specimens we studied lack triangular projections on T9, though sometimes they were angulated or only slightly projecting behind the rear margin (specimens from Ichkeul National Park, ZMUC). All specimens except for one adult female from Jebel Chambi and one adult female from near Oued El Abid village lacked spines on tarsi of legs 15. Only two specimens out of hundreds possessed tarsal spines. Likewise, the shape of tergite 9 seems to be also infrapopulationally variable. Th e specimens from Spain (Granada) in Meinert's collection (ZMUC), which were studied by E.H. Eason in 1980, all lack triangular projections and tarsal spines and should therefore be attributed to E. n. elongatus even if geographically this area is situated within the range of the nominate form. Th e general colour of the body also varies considerably among the populations, from uniformly dark brown in e.g., the Ichkeul specimens, to uniformly castaneous and dark yellowish-brownish in most of the other examined specimens. Some specimens (e.g., those from near Zahret Médine and Bulla Regia, NMNHS) possess a dark middorsal band.

Molecular data
In order to confi rm the delineation of the new species E. kahfi , we used DNA barcoding to bring genetic support to the morphological observations. Th e COI barcodes examined from 11 specimens (Table 6) among 5 species of Eupolybothrus exhibited a 20.8% mean value for interspecifi c divergences (Table 7). Th e lowest value was 16.61% between E. transsylvanicus and E. litoralis, and the highest was 23.98% between E. transsylvanicus and E. nudicornis. By contrast, for the two species for which we were able to measure it, we observed a low infraspecifi c value, 1.14% for E. nudicornis (sampled from three diff erent populations in Tunisia; see Table 6) and 0.3% for E. transsylvanicus. Th e neighbor joining tree built from this dataset shows the clear separations between the diff erent barcode clusters corresponding to the diff erent species (Fig. 5). Italy, Fogliano Mt, near Viterbo, Lazio AY214420 --------------------------   In addition to the molecular support, E. kahfi is very well characterized morphologically, having a long, conical dorso-median protuberance at the prefemur of leg 15, a unique trait within the genus. Th is is of importance for discussions of cryptic diversity because it will permit pointing to the right genetic entity/COI cluster as the bearer of the species name, and then to assign a new name to the other sibling species. Although it may be possible to get the sequence from museum material for this purpose (Hausmann et al. 2009), it is easier and cheaper to barcode the fresh holotype at the same time it is described. Jeekel (1967) assigned E. nudicornis and allied taxa to the subgenus Allopolybothrus Verhoeff , 1907, characterised by the following set of morphological characters: absence of VCm spine and presence of VCa spine on leg 15, single pretarsus of leg 15, posterior triangular projections on TT 9, 11, 13 (often reduced on T 9), spinulation of leg 15: 0, 1, 3-4, 1-2, 0-1, (1), male gonopods short, single-segmented. In the same publication he wrote "… the practical value of these subgenera is doubtful" and indeed all those characters are found also in species from other subgenera and thus are of very little value for establishing the phylogenetic relationships among the species. Most of the subgenera of Eupolybothrus comprise only a limited number of species, some of which are poorly described and known from a single specimen only. It is beyond the scope of this publication to revise the whole genus, nevertheless we would like to point out that the currently accepted subgeneric division of Eupolybothrus is outdated and will most likely be altered once several poorly known taxa are revised and contemporary phylogenetic methods are applied.
No taxonomically signifi cant diff erences were found between the syntype specimens of E. cloudsley-thompsoni and E. nudicornis, which confi rms Zapparoli's (1985) suspicion that both might be identical. Th e examined specimens lack posterior triangular projections on T9 and tarsal spines which characterize the Tunisian populations. Instead of trying to distinguish the new species from the other North African congeners known at that time, Turk (1955) compared the new species with the morphologically and geographically quite distant E. segregans Chamberlin, 1952 andE. praecursor (Attems, 1902) from Turkey and Lebanon, respectively, both currently considered synonyms of Eupolybothrus litoralis (cf. Zapparoli 1991Zapparoli , 1995b. He also wrongly at-tributed one juvenile Lithobius castaneus to the syntype series of E. cloudsley-thompsoni and failed to illustrate the porodonts. Turk's species was improperly justifi ed, and we regard E. cloudsley-thompsoni to be conspecifi c with E. nudicornis. Matic et al. (1992) described two new Italian species of Eupolybothrus, E. osellai and E. ruff oi from the Cozian Alps and Apuan Alps, respectively. Both species were very vaguely diagnosed and described, as no comparison with other congeners was made. Th ey are morphologically similar to E. nudicornis, and except for some minor diff erences in the spinulation there are no sound traits that allow separation from the latter. Th e possible synonymy with E. nudicornis has aleady been suspected (Minelli 2006). Th e question whether E. nudicornis represents a single polymorphic species or a species-complex comprising cryptic (sub-)species is also beyond the scope of this paper and requires examination of additional material from Europe and extension of molecular sampling. A fact of interest is the absence of E. nudicornis from the Balearic Islands (see e.g., Sammler et al. 2006), and its extreme rarity in Spain, which can hardly be explained as an artifact of collecting activities in these regions.
Post-embryonic development. Information on the post-embryonic development of species of Eupolybothrus is generally poor, as more comprehensive studies have been published only for E. nudicornis (Meinert 1872, Daas et al. 1996), E. grossipes and E. litoralis (Eason 1970), E. dolops (Zapparoli 1998) and E. transsylvanicus (Mitić and Tomić 2008). Th e number of post-larval stadia was found to be species-specifi c but could also vary intraspecifi cally in the diff erent parts of the species' range (Andersson 1981). Th us, Eason (1970) distinguished and described six post-larval stadia in E. grossipes, which corresponds to the number of stadia found also in E. transsylvanicus (Mitić and Tomić 2008). Daas et al. (1996) also reported six post-larval stadia for the Algerian populations of E. nudicornis (sub elongatus). Murakami (1958) reported eight post-larval stadia (including matures) in Bothropolys rugosus (Meinert, 1872) (sub B. asperatus). Our data on postembryonic development (Table 4) agree with those given by Meinert (1872) ( Table 2), except for the higher number of forcipular coxosternal teeth in the larval stadia. Compared with the data of Daas et al. (1996) (Table 3), there are some diff erences; for example, our data show higher number of antennomeres in larval stadia III and IV and less in PLI. Th is could be due to geographical variation.
Distribution. E. nudicornis is distributed throughout the whole of Maghreb, although from Morocco and Libya it is so far known only from single localities -near Bab Berred (Tetouan) (Zapparoli 1985) and in Bu Gheilan (Manfredi 1939), respectively (Map 2). Th e majority of records come from North Algeria (Map 3) and Tunisia (Map 1). Th e species distribution in North Africa covers an area of approx. 894 000 sq. km, or a distance of 1,720 km East-West and 520 km North-South. Th e species occurs also on Malta and Gozo (Zapparoli et al. 2004). In Europe it is known from France (Basses Alpes, Alpes Maritimes, Corsica) and Italy (Sardinia and circum-Sardinian islands, Ponziane Isl. [Santo Stefano Is.], Ischia Is., Sicily, Eolie [Filicudi, Lipari, Salina, Vulcano], Egadi [Favignana, Levanzo], Ustica, Lampedusa and Pantelleria Islands. In Spain it is hitherto known only from Granada (Meinert 1872) and Linares (Attems 1952).
E. kahfi is known only from its type locality, the cave Sidi Bou Gabrine (Fig. 6a). Th e cave is situated in the limestone massif Jebel Zaghouan (Fig. 6b) at a distance of approximately 500 m from the marabout Sidi Bou Gabrine (Map 4). Th e southwestern part of the mountain is composed of Jurassic limestone strata of mostly Sinemurian to Tithonian age (Schlüter 2006). Th ere are at least 30 caves in Jebel Zaghouan and around 20 in the neighbouring mountains (Mohammed Tiouiri pers. comm.) and it is very likely that E. kahfi will be found elsewhere once more profound biospeleological investigations are carried out.

Habitats.
In Tunisia E. nudicornis is recorded from coniferous and broad-leaf woods of diff erent composition and dominant structure: 1) oak forests dominated by Quercus suber and Erica arborea; 2) coniferous forests dominated by Pinus halepensis; 3) mixed forests with P. halepensis, Quercus ilex and Stipa tenacissima; mixed forests with Olea europaea and Pistacia lentiscus; mixed woods with Eucalyptus and Th uja; 4) Olea europaea orchards. E. nudicornis has been found also in open habitats such as meadows with scattered vegetation, coastal slopes with planted vegetation, rocky terrains overgrown with shrubs not far from the sea (approx. 10-50 m from the water line), coastal sandy habitats with very scattered halophilous vegetation, maquis, arid rocky slopes with shrubs and stones, deserted rocky plains with Opuntia and sparse palm trees, suburban and urban habitats, and ruins. Minelli and Iovane (1987) consider it a "fairly euryecious" species in Italy, where it often inhabits woods (Aquifolium-Fagetum, Quercus cerris, Q. ilex, Castanea, Ostrya), but also open habitats (Plantago cupanii, Calycotome, Genisto-Potentilletum, Cynosuro-Leontodontetum), occasionally found also on dunes, gardens, Olea stands, but seldom in caves. According to Zapparoli (2006), in the Central Apennines the species is most common in pastures, grasslands and open or shrub montane habitats above 900-1000 m. It occurs also in the Fagus-shrub ecotone, in garigues and calanques, seldom in Quercus cerris or Ostrya woods, olive groves, Pinus spp. reforestations, urban and suburban gardens and parks. On Malta and Gozo, E. nudicornis is known from a range of habitats including widien (valleys carrying water only during the wet season), leaf litter under Acacia and Ceratonia siliqua trees, in garique, coastal vegetation, gardens and urbanised areas (Zapparoli et al. 2004 land it also inhabits woods of Quercus ilex (Zapparoli 1995a). In Sardinia it is known from sea level up to 1800 m, in oakwoods (Quercus ilex), pine and Eucalyptus plantations, as well as in garrigue and agricultural habitats (walnut orchards); also recorded from caves and in endogeous habitat (Zapparoli 2009). Unlike E. nudicornis, E. kahfi is known only from a cave showing traits of adaptation for life underground (e.g., long legs and antennae, pale coloration). It is worth mentioning that still very little is known about the cavernicolous lithobiomorphs in North Africa. Cave-dwelling lithobiomorphs are hitherto unknown from Libya and Egypt. Only three species have hitherto been recorded from caves in Algeria and Morocco, these all being members of Lithobius Leach, 1814 (cf. Boutin et al. 2001. Only Lithobius chikerensis Verhoeff , 1936 shows troglomorphic traits (long antennae, large Tömösváry's organ, reduced ocelli) and was categorised as a troglophile (Zapparoli 1984). It is known from the Ben Add cave in Oran, Algeria and from the caves Daya Chiker, Friouat and Ras el Ma in Taza province, Morocco (Brolemann 1931b, Verhoeff 1936, Manfredi 1956, Matic 1967, Zapparoli 1984. Th e other two species, Lithobius crassipes L. Koch, 1862 andLithobius dieuzeidei Brolemann, 1931 are occasional cave-dwellers and represent trogloxenes at most (Zapparoli 1984). 2 (5) Ocelli absent (Fig. k-1), posterior triangular projections at least on TT 9, 11, 13 ( Fig. k-4)