Taxonomic status and behavioural documentation of the troglobiont Lithobiusmatulici (Myriapoda, Chilopoda) from the Dinaric Alps: Are there semiaquatic centipedes in caves?

Abstract Lithobiusmatulici Verhoeff, 1899 is redescribed based on type material and newly collected specimens. Strandiolusjugoslavicus Hoffer, 1937, described from another cave in the same region in Bosnia and Hercegovina, is presented as a junior subjective synonym of L.matulici (syn. nov.). L.matulici is shown to be most closely related to Lithobiusremyi Jawłowski, 1933, type species of the subgenus Thracolithobius Matic, 1962. The completeness of the chitin-lines on the forcipular coxosternite is discussed as a promising character for interspecific differentiation within Lithobiomorpha. Documentation of hitherto unknown semiaquatic behaviour in L.matulici and other cave-dwelling centipede species from Herzegovinian-, Montenegrin- and Pyrenean caves is presented.


Introduction
Many species of lithobiomorph centipedes have been described from European caves during the 19 th and 20 th centuries (e.g. Verhoeff 1899; Matic and Dărăbanţu 1968), as well as more recently (e.g. Negrea and Minelli 1994;Iorio 2009Iorio , 2015Stoev et al. 2013;Akkari et al. 2017). The degree of cave adaptation in the morphology of these species is rather variable: while some of them still have ocelli and rather short appendages similar to those in epigeic species, other taxa present highly troglomorphic characters, such as being completely blind and having strongly elongated legs and antennae (Folkmanová 1940;Lewis 1981). Regarding the Dinaric Mountains on the Balkan Peninsula and considering only the species with functionally articulated tarsi, five species variously placed in six genera/subgenera have been described as belonging to the latter, troglomorphic group: Lithobius (Oligobothrus) matulicii [sic] Verhoeff, 1899;Strandiolus jugoslavicus Hoffer, 1937;Mesobothrus troglomontanus Folkmanová, 1940; Lithobius (Troglolithobius) sketi Matic & Dărăbanţu, 1968;and Lithobius (Thracolithobius) remyi Jawłowski, 1933. In addition to their troglomorphic features, all of these taxa might be considered as troglobionts according to the definition of Sket (2008), as they have only been found in caves and never in surface (epigean) habitats. Most of these species are known only from their original description and only from their one or two type locality cave(s) in South Herzegovina, Montenegro, and North Albania (Fig. 1). When revising the taxonomy of the above mentioned (sub)genera, Stoev (1997) concluded that probably none of these are natural taxa and synonymised Strandiolus Hoffer, 1937, Hemibothrus Folkmanová, 1946(replacement name for Mesobothrus Folkmanová, 1940 due to homonymy) and Troglolithobius Matic, 1967under Lithobius Leach, 1814. Regarding L. matulici, S. jugoslavicus, M. troglomontanus, and L. sketi he stated that: "It will be no great surprise if the four Balkan "species" are in fact highly variable cave populations of one or two species. Only additional collecting and/or type revision can settle this problem." (Stoev 1997: 90).
Just as suggested more than 20 years ago, freshly collected specimens from that area combined with the study of type material allowed us to revise one of these species, L. matulici, and to show that one of the others, S. jugoslavicus, is its junior subjective synonym.
Some morphological and behavioural characters not highlighted in earlier descriptions are discussed here in detail: 1. The posteriorly rounded form of the 14 th tergite might indicate a close relation of L. matulici to members of the subgenus Thracolithobius Matic, 1962 (Zapparoli andEdgecombe 2011); 2. The completeness of the chitin-line on the forcipular coxosternite is an important specific character within several genera in Geophilomorpha (Bonato et al. 2011), but until now, it has not been used in Lithobiomorpha. Our unpublished preliminary studies show that this character is also probably useful for interspecific differentiation in this group, as it seems to have different character states (i.e. incomplete, or complete -as in L. matulici) which are stable within species; 3. An amphibious lifestyle in freshwater has not been reported for lithobiomorph centipedes yet, and there is only one species with such behaviour within Chilopoda as a whole. Documentation of underwater activity in cave-dwelling species is presented here, from which at least one is ascertained to be L. matulici; another observation made in a Pyrenean cave indicates that this behaviour might be actually rather widespread among cave-dwelling centipedes, similarly as in troglobiont millipedes, where a few amphibious species are already known (Enghoff 1985).

Material and methods
For light microscopy, specimens from Bravenik Cave (Bosnia and Herzegovina, Grab (near Trebinje), 42°35.97'N, 18°25.29'E) were cleared in a mixture of lactic acid and glycerol (3:1) on temporary slides. Two specimens were later cleared also in potassium-hydroxide and mounted in Euparal on permanent slides (all deposited in the Myriapoda Collection of the Hungarian Natural History Museum, Budapest, Hungary: inventory numbers HNHM chilopr-377-378; HNHM chilo-6330). Slides were examined under a Leica DM 1000 microscope equipped with a drawing tube for preparing line drawings. The map for Figure 1 was generated with QGIS version 3.2.2. (QGIS Development Team 2018).
The following abbreviations are used in the text and tables: a-anterior, C-coxa, D-dorsal, F-femur, m-median, p-posterior, P-prefemur, T-tibia, t-trochanter, V-ventral.  Remark on the origin of name. The species was dedicated to Lucijan von Matulić (teacher at a high school in Trebinje and founder of the first Speleological Society in Bosnia and Herzegovina in Trebinje in 1911), thus it was emended to "matulici" by Folkmanová (1946).
Material examined. Type material: female holotype on two slides (Slide No. 266 and 267) housed by the Museum für Naturkunde, Berlin. The slides were mounted in Canada balsam, but in an inappropriate way since they are partially dried out (Figs 2, 3). Such drying may probably happen because of the mixing of the Canada balsam with a diluting-agent, like glow-oil or xylene, at a too high of a level.
Slide No. 267: posterior part of body from 12 th segment, legs missing except right 14 th leg and the 15 th pair detached. Right ultimate leg was probably not macerated in any clearing agents before slide mounting, since the muscles are well visible inside (Fig.  3). All the other parts cleared, probably via potassium hydroxide, because their muscles were dissolved.
Redescription based on material examined and on literature. Where differences between specimens from different caves occur, they are highlighted at the given characters.

Leg pairs Ventral
Absent in S. jugoslavicus according to Hoffman (1937), but present in specimens from the same cave according to Matic and Dărăbanţu (1968: (1937), but absent in specimens from the same cave according to Matic and Dărăbanţu (1968: table 3). § Present in only one specimen from Bravenik Cave (see Table 2). | The presence of spines on femora instead of prefemora in Matic and Dărăbanţu (1968 : table 3) is most probably a typing or printing error, i.e. marking the spines in the wrong column of the table. ¶ The ventral plectrotaxy given for leg 14 by Matic and Dărăbanţu (1968: tab. 3), -,m,m,mp,m, i.e. more spines on femur than on prefemur is very unusual in Lithobius, thus a printing error in the table might be suspected. # Only spine "p" present in specimens from the Vjetrenica Cave according to Matic and Dărăbanţu (1968: table 3). Only one spine in S. jugoslavicus from the same cave according to Hoffman's (1937) plectrotaxy table, which is spine "m" according to the illustration in the same work (Hoffman 1937: fig. 1). Both spines "p" and "m" present in the holotype of L. matulici and in the specimens from Bravenik Cave. of accessory spines highest on legs 11-12: about 0.5 of claw's length for the anterior and 0.3 for the posterior spine (from Vjetrenica Cave Hoffer (1937: fig. 6) illustrated for leg 13 ratios of about 0.4 in both spines, while Matic and Dărăbanţu (1968: fig. 4g) illustrated for leg 10 ratios of 0.8 and 0.2). 3,4,4,3-5,5,5,5 coxal pores arranged in one line. In the original description Verhoeff (1899) mentioned 2(+1),3,4,3 as number for coxal pores in the holotype, but in fact it is 4,4,4,3 on legs 12-15 respectively; in S. jugoslavicus only legs 14-15 were documented with 5 and 4 coxal pores respectively (Hoffer 1937: fig. 10; in the text erroneously mentioned 4 and 5 respectively, which would be an unusual pattern in Lithobiomorpha). For the specimens from the same Vjetrenica Cave Matic and Dărăbanţu (1968) mentioned 5,5,5,5 coxal pores, while in the specimens from Bravenik Cave we found 3,4,4(5),3(4).
Female first genital sternite longer than wide, with 22-40 evenly scattered setae (40 in holotype; not known in specimens from Vjetrenica Cave); posterior border almost straight (Fig. 14) (not known in specimens from Vjetrenica Cave). Female gonopods with thin setae and 2+2 elongated spurs on first article (holotype in Fig. 14; unequal spurs in younger adults as in Figs 15, 16; 1+1 in a subadult specimen in Fig.  17). Lateral side of female gonopods with 7-12 moderate to long setae on first article, 5-8 setae on second and 1 or 2 setae on third article, arranged as in Figures 14-16 (only 4 setae on first article in a specimen from Vjetrenica Cave according to Matic and Dărăbanţu (1968: fig. 4d) but their drawing is probably inaccurate in this detail); dorsal side of gonopod with about 4 weak spines on second article and 1-3 minute spines on third article (Figs 14-16), medial side of female gonopods without setae (not known in specimens from Vjetrenica Cave). Gonopodal claw bipartite (on left gonopod of holotype (Fig. 14) misinterpreted by Verhoeff (1899: fig. V) as tripartite); medial tip smaller than lateral (Fig. 16).
Remarks on synonymy. Strandiolus jugoslavicus was described by Hoffer (1937) on a single male specimen from the Vjetrenica Cave ("grotte sur le mont 'Brencovac' près de Zavala en Popovo polje", 42°50.752'N 17°59.028'E) without comparison with Lithobius matulici Verhoeff, 1899, known from another cave only about 32 km away. The depository of the type is unknown, and it was not found at the National Museum in Prague (Dolejš 2015) where that part of Hoffer's material is housed that we know to exist. However, the original description is very detailed, supplemented with illustrations, and fits in every important character with Verhoeff's original description, but also with the holotype of matulici, as well as the fresh material studied by us. It also fits the five topotypic specimens described by Matic and Dărăbanţu (1968). The fact that Matic and Dărăbanţu (1968) identified these topotypic specimens as L. matulici (without any notes on S. jugoslavicus) also supports our conclusion that S. jugoslavicus is a subjective junior synonym of L. matulici (syn. nov.). Because also neighbouring caves might be completely isolated from each other, high-level genetic separation of cave populations might occur even in cases where no morphological differences of the specimens are obvious. Future molecular studies might easily support our decision based on morphology.
Taxonomic remarks. The posteriorly semicircular form of the 14 th tergite has not been highlighted for this species by the earlier authors, although it was illustrated by Hoffer's (1937: fig. 1) drawing on the habitus of the holotype of S. jugoslavicus and Verhoeff (1899) mentioned that the posterior corners of the tergites 3, 5, 8, 10, 12, and 14 are exceptionally strongly rounded. It is present in the holotype of matulici (Fig. 7) and in our fresh specimens as well. Matic and Dărăbanţu (1968) seem to have overlooked this character, as they only mentioned that the tergites are without posterior triangular projections. Hoffer (1937) characterised the tergites as of oval in shape, but for more details he referred to his drawing with the holotype which depicts tergite 14 with rounded posterior margin.
The shape of the 14 th tergite seems to indicate a close relation of L. matulici to the members of the subgenus Thracolithobius Matic, 1962 (Zapparoli andEdgecombe 2011), especially to its type species, Lithobius remyi, described from the Gradje Cave (Montenegro), which is only 95-150 km from the known occurrences of L. matulici, and also reported from the North Albanian Merkurth Cave (Stoev 1996). As the posteriorly semicircular form of the 14 th tergite is the key character defining Thracolithobius, we could consider L. matulici as member of this subgenus, but we refrain to do for reasons of nomenclatural stability. Including L. matulici in Thracolithobius would result in a situation in which the generic name Strandiolus Hoffer, 1937 would became a subjective senior synonym of Thracolithobius Matic, 1962 according to the principle of priority (ICZN 1999: Art. 23), because its type species, Strandiolus jugoslavicus Hoffer, 1937, is synonymised in the present paper under L. matulici (see above). Strandiolus was synonymised under Lithobius (s.s.) by Stoev (1997) (also proposed earlier informally and without explanation by Folkmanová (1946) in a key) because its differential characters are either actually common in Lithobius (s.s.) -three 'claws' on legs 1-13, reduced leg plectrotaxy, notched lateral edges of head, absence of tergal projections, form of maxillae II -or adaptations to the cave environment -absence of ocelli, elongation of legs and narrow anterior sternites, depigmentation, high number of antennal articlesand as such of no taxonomical importance. Meanwhile, Thracolithobius Matic, 1962 is considered as a valid subgenus (Stoev 1997;Shelley 2006;Ćurčić et al. 2008;Zapparoli and Edgecombe 2011) with three species -L. dacicus Matic, 1959, L. inexpectatus Matic, 1962, L. remyi Jawlowski, 1933 -but the monopyhly of this group might be questioned. The only common character defining this subgenus is the shape of the 14 th tergite, a character that however has already been proven to vary at the inter(sub)specific level in Lithobius (Andersson 1979) and in another lithobiomorph genus, Eupolybothrus (Stoev et al. 2013;Akkari et al. 2017). Apart from this character, the members of the subgenus seem to be rather different in several other features (e.g. presence/absence of ocelli and a wart-like structure on forcipular tarsungulum) and L. matulici differs actually from the members of Thracolithobius even in an aspect of the 14 th tergite: the rounded shape is present in matulici also in females, while it is known only from males in the other species. Although at least L. remyi and L. matulici seem to be similar also in some other features (lack of ocelli, strong porodonts, coxosternal dentation) this may be also due to convergent adaptation to a similar lifestyle in cave environments.
Semiaquatic behaviour. One lithobiomorph specimen was found in July 2014 while one of the authors, G. Balázs, was diving in Vjetrenica Cave. The specimen was in a water-filled part of the cave (Donje Vjetrenica), freely and consciously walking on the underwater bottom at a depth of 3 metres, at a distance of about 30 metres from any terrestrial microhabitats (i.e. chambers with air). This specimen was without any signs of distress (no spasms, no enfeeblement). There was no flood in the cave at that time, the water was still (not flowing), and thus a simple flushing away of the specimen from the water's edge might be ruled out. This individual spent another 2 hours in the water, while kept captured by the diver and escaped later during photographic documentation. In the photograph (Fig. 18), the 14 th tergite of the specimen seems clearly rounded posteriorly, and thus it can be considered as L. matulici with confidence. Similar cases of lithobiomorph specimens on the bottom of water ( Bonato et al. (2010). However, a weak suture is also present along the stripe in geophilomorphs, and weak sclerotisation is present along the suture also in lithobiomorphs (orig. obs.). Thus, the homology of the two structures seems probable, and we prefer to also use this established term (Bonato et al. 2010) in Lithobiomorpha, just as it has already been used by Latzel (1880). While the chitin-line is an incomplete suture (i.e. not reaching the posterior margin of the coxosternite) in several lithobiomorph species, it is complete in L. matulici. Our preliminary unpublished studies reveal that a complete chitin-line is probably not rare at all (e.g. in Lithobius forficatus (Linnaeus, 1758), Lithobius microps Meinert, 1868, andLithobius burzenlandicus Verhoeff, 1931). The states of this character seem to be stable within species, as well as in specimens of different age which promises that it might be useful for some cases of interspecific differentiation.
Semiaquatic behaviour. Semiaquatic behaviour in terms of actively and regularly moving into the water has never before been reported for lithobiomorphs, but even for Myriapoda as a whole there have been few examples. In the following paragraphs a short overview is given (for Chilopoda as well as for millipedes), starting from observation of animals actively seeking water to species enduring inundation out of necessity in flood-prone areas.
Only two publications mention active semiaquatic behaviour in Chilopoda. One is the only report of centipedes entering freshwater on their own free will (Armitage 1982). This short paper reports on several specimens of the geophilomorph Strigamia maritima (Leach, 1817) found on two occasions in a small stream in England, where they possibly entered the water to hunt for caddisfly larvae (Armitage 1982). The other case is of a scolopendromorph specimen which was possibly hunting underwater (Moraes and Chagas-Júnior 2009). The centipede was found dead in a sea anemone which had probably caught it under water.
A semiaquatic lifestyle is more frequently noted for millipedes. Some species have been reported from under stones in streams in France (Causard 1903) and Australia (Burrows et al. 1994), and one species in South America is known to be able to live submerged for several months in subadult stadia (Adis 1986). Three additional, possibly highly water-adapted species have been reported from Guyana and from widely dispersed Atlantic and Pacific islands (Golovatch and Kime 2009). From caves there are several millipede species described as semiaquatic, for example some julids and polydesmids in the Italo-Balkan region of Europe Enghoff et al. 1997;Antić et al. 2017). These cave millipedes enter water on purpose, spend a long time submersed, and have modified mouthparts, which are probably adapted to filtering and screening suspended organic particles from the water . Similar mouthparts are also known from some other cave-dwelling millipede species from the Caucasus and Papua New Guinea (Enghoff 1985), suggesting that semiaquatic behaviour might be more common in diplopods than generally acknowledged. Some observations show centipedes to choose swimming as a way of escape when attacked or disturbed. Zulka (1991) published the first observations of this for Lithobius curtipes (C.L. Koch, 1847) and Lamyctes emarginatus (Newport, 1844), which entered water from objects standing out of surrounding water when he tried to catch them. Even when there are terrestrial pathways for escape, some species or at least specimens chose water: Siriwut et al. (2016) mentioned an individual of Scolopendra cataracta Siriwut, Edgecombe & Panha, 2016 that entered a stream to escape from the collector, and the same behaviour was observed by one of the authors (I.H. Tuf pers. obs.) in Lithobius forficatus (Linnaeus, 1758) and Lithobius mutabilis L. Koch, 1862.
Probably the most frequent reasons for myriapods to come into contact with water are tides and floods. From tide-affected seashores there are numerous reports of more than 40 centipede taxa (see review by Barber 2009Barber , 2011. Almost all of these are geophilomorphs, many of which are considered as real halophiles with adaptations to submergence (Binyon and Lewis 1963;Barber 2011), while the recorded ubiquitous lithobiomorph species do not appear to be truly halophilic; the only exception might be Lithobius ellipticus Takakuwa, 1939(Barber 2009). Several millipede species are able to survive river floods by living actively under floodwaters for weeks (Golovatch and Kime 2009), while some centipedes have dormant submerged egg stage (e.g. Lamyctes adisi Zalesskaja, 1994(Zalesskaja 1994 and Lamyctes emarginatus (Zulka 1991;Zerm 1997)). The centipede Lithobius curtipes is able to survive under water for more than one week under experimental conditions (Tufová and Tuf 2005), while in an experiment Scolopendra subspinipes Leach, 1816 was found to swim on the water surface, probably as a strategy for escape during floods (Lewis 1980). Another scolopendromorph, Edentistoma octosulcatum Tömösváry, 1882, does not swim, but in an experiment by Lewis (1980) simply walked along the bottom when inundated. L. matulici and related species inhabit caves where flash floods are common and which have active streams of highly fluctuating water levels, depending on the precipitation and/or snow melting at the surface region above them (Spahić 2015). In addition to the necessity of tolerating submergence during floods, the ability to submerge voluntarily and to move under water might be also useful in moving between parts of the cave that are separated by water. Semiaquatic behaviour might be potentially highly adaptive in caves also for another reason: in subterranean habitats food sources are limited and an expansion of the prey spectrum with the inclusion of the aquatic biota can help a terrestrial predator increase its fitness. This might be highly significant, especially when aquatic biota (e.g. Niphargus amphipods) represents the main part of the available biomass of possible prey, like in the caves discussed here (Gergely Balázs pers. obs.; Márton Mede in litt.). Due to similar conditions and forces, such adaptation might be hypothesized to emerge in parallel multiple times in different caves, just like in some hydrophilous millipede taxa (Enghoff 1985).
Two other cave-dwelling Lithobius (s.s.) species from the Dinaric Mountains. Lithobius sketi Matic & Dărăbanţu, 1968 was described as belonging to the subgenus Troglolithobius Matic, 1967 (junior synonym of Lithobius according to Stoev 1997), which included also L. matulici at that time, and it was stated to be very similar to that species. Although no rounded edge of tergite 14 is known for L. sketi, additional studies are needed to verify this character in this species. Lithobius troglomontanus (Folkmanová, 1940) was described from Vodna Cave (Vodna Pećina), Montenegro, but it is missing from the list of Mitić et al. (2007). Although Kos (1992) considered L. troglomontanus closely related to L. matulici and L. sketi and also as belonging to the subgenus Troglolithobius Matic, 1967, it seems to be different indeed from L. matulici in the shape of tergite 14 (with cornered posterior edges). It shares several characters with L. sketi, i.e. the structure of female gonopods (unipartite gonopodal claws, 1+1 spurs) and the arrangement of the coxal pores (smaller pores forming a second row), but they do differ in several important characters. Some of these differences (small Tömösváry's organ, short ultimate legs and antennae) actually show troglomontanus to be morphologically not very cave-adapted, and thus, surface collecting around the type locality cave might prove it to be not a real troglobiont species. Based on the differences, we consider L. troglomontanus and L. sketi to be two valid species, but molecular phylogenetic studies are needed on each of these cave taxa to clarify their actual relation to each other and to the members of other subgenera within Lithobius.