Phylogenetic analysis of the Niphargus orcinus species–aggregate (Crustacea: Amphipoda: Niphargidae) with description of new taxa

The Niphargus orcinus aggregate, also “subgenus Orniphargus” of the genus Niphargus, is a widely distributed assemblage of large‐bodied amphipods. Although it comprises the most prominent European groundwater animals, its phylogeny has not yet been scrutinized and its monophyly has never been tested. In this study, 30 putative “Orniphargus” taxa and 14 other Niphargus and outgroup species were examined cladistically using a matrix of 71 morphological characters. The strict consensus tree constructed of 34 equally parsimonious trees demonstrated that “Orniphargus” does not constitute a monophylum. Most of the species were divided into two large clades. Both clades contain species from the western Balkans (mainly the Dinaric Karst) with partly overlapping ranges. Niphargus croaticus pachytelson and N. orcinus redenseki are elevated to species rank. Four new species are described: N. dolichopus sp. n., N. polymorphus sp. n., N. dabarensis sp. n. (all Bosnia and Herzegovina) and N. lourensis sp. n. (Greece).


Introduction
With over 300 described species and subspecies, the genus Niphargus Schiö dte, 1947 (Amphipoda: Niphargidae) is the largest genus of European freshwater amphipods (Pinkster 1978;G. Karaman and Ruffo 1986;Fišer et al. 2002). Most of the species inhabit subterranean waters and constitute a substantial part of the European groundwater biodiversity (Sket 1999a(Sket , 1999b. The genus is distributed across most of Europe, mainlybut not exclusively -south of the Pleistocene ice sheet boundary (Ruffo 1953;G. Karaman and Ruffo 1986;Proudlove et al. 2003). Few species are known from the Arabian Peninsula, Turkey and Iran (G. Karaman 1986Karaman , 1998Karaman , 2003Bat et al. 2001). By contrast, over most of the Iberian Peninsula, only its presumed relative Haploginglymus Mateus and Mateus, 1958 can be found (G. Karaman and Ruffo, 1986).
The analysis includes all the features that vary between the taxa and that can be distinctly split into character states. The shape of some body parts has provided a substantial number of characters used in this study. The main character coding challenges were (1) describing the shapes of certain variable body parts in the form of indices, and (2) finding gaps that defined character states. Gap searching was constrained by a limited number of individuals available (these species rarely appear in high numbers), thus methods based on variability (e.g. Thorpe 1984;Archie 1985) failed. Gaps were searched for as follows. Index values of the appendage article under consideration were treated as intervals for every species. Partially overlapping intervals were joined into larger intervals until non-overlapping intervals, considered as character states, were found. Some problems occurred when only one or two specimens of a species available showed a boundary value. In such cases, characters were coded as polymorphic (Wiens 1995). The highest possible number of character states was chosen as suggested by Pogue and Mickevitch (1990). A commented list of coded characters and the character-taxon matrix can be found in Appendices A and B. The illustrations of the character states (in preparation) shall be published elsewhere.

Character analysis
Hypotheses of character evolution were evaluated to (1) examine sources of disagreement between our results and previous classifications and (2) review character support of the clades. Character distributions were studied in WinClada under ''fast'' (ACCTRAN) optimization (for rationale see Discussion). The only diagnosis designated (S. Karaman 1950c) was used as a reference for comparison of our results with previous classifications. In particular, the following parameters were evaluated for selected clades: (1) number of characters transformed at the basal node, (2) number of consistent characters, (3) number of reversals, and (4) number of parallelisms. For a character to be consistent it had to remain untransformed within a clade. We regarded reversals as transformations of the type aRbRa that occurred at least once within a clade. Character states attained independently within a clade at least twice were treated as parallelism. The character support for all major clades and subclades was compared.

General notes and abbreviations
We call ''spines'' strong, rigid structures and ''setae'' thin, flexible structures. ''Spines'' and ''setae'' often cannot be clearly distinguished. Sometimes the same group of spines/setae can contain setae, spines and various intermediate structures. We also observed that strong ''spines'' Figure 1. A strict consensus tree of 34 most parsimonious trees (length5472; CI50. 26,RI50.60). Values of Bremer support index (decay index) are indicated below branches. Taxa traditionally assigned to the ''Orniphargus'' species aggregate, as well as clades for which character analysis was performed, are encircled in boxes. Note 1: Despite of its position on the cladogram N. pectinicauda has never been considered as an ''Orniphargus'' taxon. Note 2: Taxa are named according to their lowest rank. For full names, see Tables I and II. of young specimens become thinner and more flexible (''setae'') in adults. For these reasons, when dealing with number of structures in a group, we simply count the number of structures in a group as ''number of spines/setae'' and do not distinguish between spines and setae.
To make descriptions clearer for the reader, some shapes of some characters expressed as ratios in phylogenetic analysis were replaced here by descriptive means. The numbers can be obtained from the corresponding author, see also Appendices A and B. See also comments of characters and note in Appendix A.

Phylogenetic analysis
The heuristic search found 12 equally parsimonious trees 462 steps long, with a consistency index (CI) of 0.27 and a retention index (RI) of 0.61. The ratchet search found another 22 trees of the same length and similar topology. Following Nixon and Carpenter (1996), the strict consensus method was used to summarize the topological agreement among cladograms. The cladograms differed at six nodes. Collapsing them led to a tree 472 steps long (CI50.26, RI50.59) ( Figure 1). Most of the nodes had low Bremer support values, ranging from one to seven.
Rooting Niphargus on Gammarus fossarum split the Niphargus taxa in two lineages ( Figure 1). One was represented by the tiny interstitial species [N. skopljensis and N. longidactylus], whereas the rest of the analyzed species represented a series of paraphyletic terminal taxa and three distinct clades. All ''Orniphargus'' species were nested within other Niphargus species, none of them showing affinities to Gammarus. The ''Orniphargus'' group itself emerged as a paraphyletic group. Most of its members were subdivided into two large, non-sister clades, referred to as clades A and B ( Figure 1). Species considered transitional by morphology (N. stefanellii, N. virei and N. polymorphus) occupied basal positions in all cladograms, except for N. miljeticus, which was positioned among so-called ''stygius sensu lato'' species.

Character analysis
The notorious morphological variability of niphargids was reflected through the analysis of character transformation. It can be summarized in three general aspects. (1) Shape and    1,CI50.28,RI50.37;right). Note that the body shape is plesiomorphic in most of the ''Orniphargus'' taxa, and large body size is a highly convergent (possibly troglomorphic) character. reversals and parallelisms than clade A. Conversely, clade A is characterized by lower Bremer support values and lower character consistency. The character transformation analysis identified some fundamental problems with previous diagnosis of the group (S. Karaman 1950c). Most of the proposed diagnostic characters are either plesiomorphic or homoplastic . Furthermore, some characters (e.g. length of antenna, length of uropod I-II rami) may have more states or need to be subdivided into several characters (e.g. depth of coxal plates, shape and length of uropod III). Illustrative is the case of coxal plates I-IV, which are supposed to be ''more deep than wide'' (S. Karaman 1950c). If all coxal plates in all species exhibited congruent patterns of shapes, they should be regarded as serial homologues and coded as a single character. In fact, however, they transform independently from each other. Coding them as a single character with character state ''more deep than wide'' represents an oversimplification and loss of phylogenetic information.  19,CI50.18,RI50.53;left) and ''gnathopod II article 6 size'' (character 39,CI50.33,RI50.7;right). Long antennae are considered as troglomorphic, whereas a large-sized gnathopod II is supposed to be a synapomorphy of ''Orniphargus'' (S. Karaman (1950c)).

Changes of taxonomic rank
The analysis revealed two nomenclature errors, which originated from an earlier work (S. Karaman 1950c). Niphargus croaticus and N. arbiter occur syntopically. In the course of redescription, S. Karaman misinterpreted the identity of Jurinac's N. croaticus, and in his redescription of what he believed to be N. croaticus, criticized the original description as inaccurate. Gordan , who described S. Karaman's ''N. croaticus'' as a new species, N. arbiter, and redescribed the true N. croaticus, later corrected the mismatch. Sket ( , 1960 erroneously named N. orcinus redenseki and N. croaticus pachytelson based on S. Karaman's nomenclature and drawings. Niphargus o. redenseki resembles N. croaticus, the description of which was at that time rejected, whereas N. c. pachytelson resembles N. arbiter that was at that time designated as N. croaticus. Niphargus o. redenseki is in fact very similar to the true N. croaticus, but has a shorter (i.e. less troglomorphic, see Discussion) antenna I and pereopod VII, and partially reduced pleonal spinosity (which might point to more advanced troglomorphy). Different patterns  30,CI50.5,RI50.86;left) and ''coxa III shape'' (character 40,CI50.33,RI50.78;right). The difference in the distribution of the two characters contradicts the notion that they might be serial homologues.
of troglomorphic features of two sympatric taxa suggest on independent history, thus N. redenseki should be raised to species level.
According to the results of the cladistic analysis ( Figure 1), N. croaticus pachytelson belongs to a different lineage than N. croaticus. Its distinct position in the cladogram, as well as marked morphological differences (presence of dorsal telsonic spines, shallower telsonic cleft, several setae on ventro-posterior margin of pereomera VII, lack of pleonal spines, single aesthetasc per article of flagellum I, single distal row of setae on gnathopod I/3, innerside positioned denticulated spines on propodus II, setae on dactyla I-II in groups), allow us to elevate the taxon to species rank.

Description
Head and trunk (Figures 6,11). Body length up to 17.5 mm. Head length 10-11% of body length; rostrum absent. Pereonites I-V without setae; pereonite VII with 1-2 posteroventral setae and up to 12 strong spines dorso-posteriorly. Dorsal spines may be present also on pereonite VI.
Mouth parts ( Figure 8). Inner lobes of labium longer than half of the outer lobes.
Left mandible: incisor with 5 teeth, lacinia mobilis with 4 teeth; between lacinia and molar row of serrated spines, long seta at the base of molar. Right mandible: incisor process with 4 teeth, lacinia mobilis with several small denticles, between lacinia and molar a row of serrated spines; molar process with a long basal seta. Ratio of mandibular palp article 2:article 3 (distal) is 1:1.1-1.2. Proximal palp article without setae; the second article with 9-10 setae groups and single setae; distal article with 1 A group of 7 setae; 3-4 B groups; 37-38 D single setae; 6-7 E single setae.
Maxilla II inner lobe slightly smaller than outer lobe; both of them apically setose. Maxilliped palp article 2 with 8-10 rows of setae along inner margin; distal article with a dorsal seta, a group of small setae at the base of the nail. Maxilliped outer lobe with 13 flattened spines and 4 serrated setae; inner lobe with 3-4 flattened apical spines and 8-9 serrated setae.
Gnathopod I (Figure 9). Ischium with postero-distal row of setae. Carpus 0.6-0.65 of basis length and 0.8-0.85 of propodus length. Anterior margin of carpus with the distal and 1 additional group of setae; carpus posteriorly with transverse rows of setae proximally, a row of lateral setae and rows of submarginal setae; postero-proximal bulge large (1/3 of carpus length), positioned medially. Propodus rectangular, palm short, convex and slightly inclined. Along posterior margin 7 rows of denticulated setae. Anterior margin with 34-37 setae in 7 groups, antero-distal group with 11-12 setae. Group of 3 facial setae proximally of palmar spine; small groups of surface setae present. Palmar corner with strong palmar spine, single supporting spine on inner surface and 3 denticulated spines on outer side. Nail Figure 9. N. dolichopus sp. n., holotype. Gnathopod I (above) and gnathopod II (below). length 0.2-0.25 of total dactylus length; along anterior margin 10 single setae; short setae along inner margin present.
Gnathopod II (Figure 9). Basis width:length is 1:0.3. Ischium with postero-distal row of setae. Carpus 0.6 of basis length and 1-1.05 of propodus length. Anterior margin of carpus with distal and 2 additional groups of setae; carpus posteriorly with transverse rows of setae proximally, a row of lateral setae and submarginal setae groups; posteroproximal bulge large (1/3 of carpus length), positioned medially. Propodus small (compared to the body) and equal to propodus of gnathopod I. Propodus rectangular, palm short, convex and more inclined than palm of gnathopod I. Posterior margin with 9-10 rows of denticulated setae. Anterior margin with 18-20 setae in 5 groups; antero-distal group with 9-10 setae. Group of 3 facial setae proximally of palmar spine; small groups of surface setae present. Palmar corner with strong palmar spine, single supporting spine on inner surface and 2-4 denticulated spines on outer side. Nail length 0.2-0.25 of total dactylus length. Along anterior margin 10-11 single setae; short setae along inner margin present.  or setae. Dactylus VII length 0.2 of propodus VII length. Dactyli V-VII with dorsal plumose seta; at the base of nail a spine and a seta.
Variability. The studied specimens do not differ in spination and shape; minor differences in number of setae and spines may be a consequence of different body lengths.
Remarks and affinities. In general appearance, N. dolichopus with small gnathopods, extremely long pereopods V-VII without distal lobes on narrow bases and extremely long antenna I, as well as with well-developed spination of pleonites and pereopods, resembles both N. croaticus and N. balcanicus. However, it represents a unique combination of characters that can be found in both species. Short flagellum of antenna II, approximately equal size of propods of gnathopods I and II, shallow telson cleft and spines on urosomite III are characters shared by N. croaticus, from which it can be distinguished by the presence of strong dorsal spines on pereonites VI-VII. These spines are also characteristic of N. balcanicus. However, N. balcanicus has dorsal spination much better developed, also on pereonites I-V; it has a much longer flagellum of antenna II, posteriorly produced coxae V-VI, proxi-posterior lobe on pereopod VII, strong spines on dactyls III-VII, and well developed plumose setae on uropods I-II. On the other hand N. balcanicus lacks spines on urosomite III and additional anterior setae on the carpus of gnathopods I-II. Etymology. The epithet dabarensis is derived from the type locality.

Description
Head and trunk (Figures 6, 14). Body length up to 25.5 mm. Head length 9% of body length; rostrum absent. Pereonites I-VI without setae; pereonite VII with 3 postero-ventral setae. Mouth parts (Figure 12). Inner lobes of labium longer than half of the outer lobes.
Left mandible: incisor with 5 teeth, lacinia mobilis with 4 teeth; between lacinia and molar row of serrated spines, long seta at the base of molar. Right mandible: incisor process with 4 teeth, lacinia mobilis with several small denticles, a row of serrated setae and molar process with a long basal seta. Mandibular palp articles 2:article 3 (distal) is 1:1.25. Proximal palp article without setae; the second article with 9 setae groups and single setae; distal article with 1 A group of 14 setae; 5 B groups; 37 D single setae; 5 E single setae. Maxilla I distal palp article with 8 apical setae. Outer lobe of maxilla I with 7 uni-, bi-or pluri-toothed spines; inner lobe with 10 setae.
Maxilla II inner lobe slightly smaller than outer lobe; both of them with numerous apical setae.
Maxilliped palp article 2 with 11 rows of setae along inner margin; distal article with a dorsal seta and a group of small setae at the base of the nail. Outer lobe with 15 flattened spines and 10 serrated setae; inner lobe with 4 flattened apical spines and 6 serrated setae.
Gnathopod I (Figure 13). Ischium with postero-distal row of setae. Carpus 0.5 of basis length and 0.7 of propodus length. Anterior margin of carpus only with distal group of setae; carpus posteriorly with transverse rows of setae proximally, a row of lateral setae and rows of submarginal setae; postero-proximal bulge large (1/3 of carpus length), positioned proximally. Propodus large ''hoof'' shaped, palm long, convex and oblique. Posterior margin with 14 rows of denticulated setae. Anterior margin with 32 setae in 4 groups; antero-distal group with 11 setae. Group of 4 facial setae proximally of palmar spine; small groups of surface setae present. Palmar corner with strong palmar spine, single supporting spine on inner surface and 3 denticulated spines on outer side. Nail length 0.2 of total dactylus length; along anterior margin 11 single setae; short setae along inner margin present.
Gnathopod II (Figure 13). Basis width:length is 1:0.3. Ischium with postero-distal row of setae. Carpus 0.5 of basis length and 0.75 of propodus length. Anterior margin of carpus with distal group of setae; carpus posteriorly with transverse rows of setae proximally, a row of lateral setae and rows of submarginal setae; postero-proximal bulge large (1/3 of carpus length), positioned proximally. Propodus large (compared to the body), much larger than propodus of gnathopod I. Propodus large ''hoof''-shaped, palm long, convex and oblique. Posterior margin with 17 rows of denticulated setae. Anterior margin with 13 setae in 4 groups; antero-distal group with 12 setae. Group of 4 facial setae proximally of palmar spine; small groups of surface setae present. Palmar corner with strong palmar spine, single supporting spine and 2 denticulated spines on inner surface. Nail length 0.2 of total dactylus length. Along anterior margin 7 single and a pair of setae (totally 9 setae); short setae along inner margin present.
Uropod I protopodite with 7 dorso-lateral spines and 5 dorso-medial spines. Exopodite:endopodite length is 1:1.05; rami straight. Uropod I rami with groups of spines and/or setae along lateral and medial side. Endopodite with 8 setae and/or spines in 4 groups of spines and/or setae; apically 5 spines. Exopodite with 21 setae and/or spines in 6 groups of spines and/or setae; apically 5 spines.
Uropod II exopodite:endopodite length is 1:1.05. Uropod III length 0.2 of body length. Protopodite with 2 lateral spines and 8 apical spines and setae. Endopodite 0.5 of protopodite length; apically with 1 spine and 1 plumose seta; laterally 2 setae. Exopodite of uropod III slightly flattened; distal article 0.12 of proximal article length. Proximal article with 6 groups consisting of setae, plumose setae and spines along inner margin and 3 groups of spines and setae along outer margin. Distal article of exopodite without lateral setae; apically 3 spines and 3 setae.
Variability. Only one female was studied. Another specimen found is preserved in 96% ethanol for molecular analyses.
Remarks and affinities. Size and shape of propodus of gnathopods I-II as well as position of palmar spines on gnathopod II are characteristic for species N. arbiter, N. salonitanus, N. rejici and N. lourensis. From these taxa, N. dabarensis can be distinguished by features that are shared with N. vjetrenicensis, N. bilecanus, N. hercegovinensis and N. trullipes: large bulge on carpus I-II, disto-posteriorly produced coxal plates V-VI, by the presence of proximoposterior lobe of basis VII, dactyls V-VII with 2-3 dorsal plumose setae as well as uropods I-II with plumose setae. Finally, narrow bases without distal lobes of pereopods V-VII and such a long antenna I (similar as in N. croaticus and N. balcanicus) are not present in any of the above listed species.

Diagnosis
Pleonites I-III with setae along dorso-posterior margin. Telson with apical spines only; these are very long (0.61 telson length). Carpus I-II with postero-proximal bulge large (1/3 of carpus length); gnathopod II propodus large ([length+width+diagonal] 0.29 body length); much larger than propodus of gnathopod I (1:0.75). Denticulated spines in palmar corner of gnathopod II hidden behind the palmar spine. Coxae V-VII with single posterior setae. Pereopods V-VII with numerous strong spines. Uropod III very spiny and very short (0.19 of body length); exopodite flattened, distal article with setae along one margin only.
Telson length:width is 1:0.85; cleft 0.65 of length. Only 3-4 apical telson spines per lobe, they are very long, 0.6 of telson length. Plumose setae inserted laterally in the middle part of telson.
Mouth parts (Figure 16). Inner lobes of labium longer than half of the outer lobes.
Left mandible: incisor with 5 teeth, lacinia mobilis with 4 teeth; a row of serrated spines between lacinia and molar and a long seta at the base of molar. Right mandible: incisor process with 4 teeth, lacinia mobilis with several small denticles, a row of serrated setae and molar process with a long basal seta. Mandibular palp article 2:article 3 (distal) is 1:1.4. Proximal palp article without setae; the second article with 8 setae groups and single setae; distal article with 2 A groups of 1+7 setae; 4 B groups; 41 D single setae; 6 E single setae.
Maxilla II inner lobe slightly smaller than outer lobe; both of them with numerous apical setae.
Maxilliped palp article 2 with 9 rows of setae along inner margin; distal article with a dorsal seta and a group of small setae at the base of the nail. Outer lobe with 10 flattened spines and 7 serrated setae. Inner lobe with 4 flattened spines and 10 serrated setae apically.
Gnathopod I (Figure 17). Ischium with single postero-distal row of setae. Carpus 0.5 of basis length and 0.65 of propodus length. Anterior margin of carpus with distal group of setae only; carpus posteriorly with transverse rows of setae proximally and a row of lateral setae; postero-proximal bulge large (1/3 of carpus length), positioned proximally. Propodus ''hoof''-shaped, palm long, oblique and convex. Posterior margin with 10 rows of denticulated setae. Anterior margin with 25 setae in 6 groups; antero-distal group with 10 setae. Group of 3 facial setae proximally of palmar spine; small groups of surface setae present. Palmar corner with strong palmar spine, single supporting spine on inner surface and 4 denticulated spines on outer side. Nail length 0.3 of total dactylus length; along anterior margin 3 single setae and a group of 2 setae; short setae along inner margin present.
Gnathopod II (Figure 17). Basis width:length is 1:0.3. Ischium with single postero-distal row of setae. Carpus 0.5 of basis length and 0.6 of propodus length. Anterior margin of carpus with distal group of setae; carpus posteriorly with transverse rows of setae proximally, a row of lateral setae, and a short row of setae at the base medially; posteroproximal bulge large (1/3 of carpus length), positioned proximally. Propodus much larger than propodus I, ''hoof''-shaped, palm long, oblique and convex. Posterior margin with 12 rows of denticulated setae. Anterior margin with 10 setae in 5 groups; antero-distal group with 12 setae. Group of 3 facial setae proximally of palmar spine; small groups of surface setae present. Palmar corner with strong palmar spine, single supporting spine and 2 denticulated spines on inner surface. Nail length 0.25 of total dactylus length. Along anterior margin 6 single setae; short setae along inner margin present.
Pereopods III-IV ( Figure 18). Lengths of pereopods III:IV is 1:0.95. Dactylus IV 0.3 of propodus IV length; nail length 0.5 of total dactylus length. Dactyli III-IV with dorsal plumose seta; at the base of nail a spine and a seta.
Uropod II exopodite:endopodite length is 1:1.1. Uropod III length is 0.2 of body length. Protopodite with 2 lateral spines and 8 apical spines and/or setae. Endopodite 0.4 of protopodite length; apically with 1 spine and 1 plumose seta; laterally 1 seta. Exopodite of uropod III flattened; distal article 0.15 of proximal article length. Proximal article with 6 groups of setae, plumose setae and very long and strong spines along inner margin; 5 groups of very long and strong spines and setae along outer margin. Distal article of exopodite with 3 lateral groups of setae; groups set along one margin; apically 3 setae. Variability. Only one female was studied.
Remarks and affinities. The most remarkable features of the species are extremely strong and long spines on telson and appendages as well as setae along the dorso-posterior margins of the pleonites. These characters can be found also in N. podgoricensis, which can be distinguished by the presence of spines on urosomite III and plumose setae on uropod I. Another important distinguishing feature between N. podgoricensis and N. lourensis are proportions of gnathopods and the position of spines on the palmar corner of gnathopod II, which in the new species resembles N. arbiter, N. salonitanus, N. rejici and N. dabarensis. However, none of these taxa has such long telson spines or setae along the dorso-posterior margins of the pleonites.
Niphargus polymorphus n. sp. (Figures 7, 20-23 Etymology. The name was given on the basis of the mosaic appearance of certain features characteristic of various other species (coxae, bases and number of hooks in retinacles) as well as for the polymorphism of some characters (dorsal setae on carpus of gnathopods I-II, additional spine on dactyl VII), which are usually stable in other species.

Diagnosis
Pleonites I-III with spines and some setae along dorso-posterior margin. Telson with 8-9 apical spines per lobe. Coxal plate of gnathopod I antero-ventrally produced; coxal plates II-IV narrow (width:depth is 1:0.5-0.75); coxal plate IV with well developed posterior lobe. Coxal plates V-VI with posterior lobe much larger than anterior lobe. Bases V-VI with disto-anterior lobe equally developed as disto-posterior lobe; basis VII with extremely large disto-posterior lobe. Pleopods I-III with 3 hooks in retinacles.
Pleonites I-III with up to 29 spines and setae along dorso-posterior margin. Mouth parts (Figure 20). Inner lobe of labium longer than half of the outer lobe.
Left mandible: incisor with 5 teeth, lacinia mobilis with 4 teeth; between lacinia and molar row of serrated spines, long seta at the base of molar. Right mandible: incisor process with 4 teeth, lacinia mobilis with several small denticles, a row of serrated setae and molar process with a long basal seta. Mandibular palp articles 2:3 (distal) is 1:1.15. Proximal palp article without setae; the second article with 12 setae groups and single setae; distal article with 1 A group of 16 setae; 6 B groups; 41 D single setae; 6 E single setae. Maxilla I distal palp article with 10-12 apical setae. Outer lobe of maxilla I with 7 uni-, bi-or pluri-toothed spines; inner lobe with 8 setae.
Maxilla II inner lobe slightly smaller than outer lobe; both of them with numerous apical setae.
Maxilliped palp article 2 with 11 rows of setae along inner margin; distal article with a dorsal seta and a group of setae at the base of nail. Outer lobe of maxilliped with 15-16 flattened spines and 6-8 serrated setae. Inner lobe with 5 flattened spines and 9-11 serrated setae apically.
Gnathopod II (Figure 21). Basis width:length is 1:0.3. Ischium with single postero-distal row of setae. Carpus 0.6-0.65 of basis length and 1.1-1.15 of propodus length. Anterior margin of carpus with distal group of setae, sometimes additional groups of setae are present; carpus posteriorly with transverse rows of setae proximally, a row of lateral setae and submarginal setae groups; postero-proximal bulge large (1/3 of carpus length), positioned proximally. Gnathopod II propodus small (compared to the body) and larger than propodus of gnathopod I. Propodus rectangular, palm short, convex and slightly inclined. Posterior margin with 17-18 rows of denticulated setae. Anterior margin with 15 setae in 4-5 groups; antero-distal group with 12-15 setae. Group of 4-5 facial setae proximally of palmar spine; small groups of surface setae present. Palmar corner with strong palmar spine, single supporting spine on inner surface and 3 denticulated spines on outer side. Nail length 0.2 of total dactylus length; along anterior margin 6-7 single setae and a distal group of 2 setae; short setae along inner margin present.
Variability. The paratype is without additional spine on dactyls VII as well as without additional setae groups along anterior margin of gnathopods I-II. Spine distally of retinacles occurs irregularly. Males not known.
Remarks and affinities. Niphargus polymorphus resembles N. carcerarius in most characters, from which it can be distinguished by larger body size, spines along the dorso-posterior margin of pleonites and three hooks in the retinacles of the pleopods. Otherwise N. polymorphus shares some characters with N. longiflagellum (wide bases and structure of distal lobes of pereopods V-VII, large disto-posterior lobe on coxa IV, shape of gnathopods) as well as N. steueri (shape of coxal plate I, much larger posterior lobe on coxae V-VI) and N. pectinicauda (narrow coxal plates of pereopods I-IV, large disto-posterior lobe on coxa IV, much larger posterior lobe on coxae V-VI). However, N. polymorphus can be easily distinguished from all these species by the presence of the three hooks in the retinacle of pleopods and a higher number of apical telson spines.

Phylogeny of ''Orniphargus'' taxa
The results of the present study do not confirm the findings of an earlier allozyme analysis (Bullini and Sbordoni 1980) that put ''Orniphargus'' outside Niphargus, as a sister group to Gammarus. However, we cannot compare the results of both studies directly, since the only ''Orniphargus'' species used in the allozyme study was N. patrizii (Sbordoni pers. commun.) -a species not included in our analysis.
Descriptions of groups or subgenera of Niphargus were often accompanied by comments such as ''some of the characters may be different'' (S. Karaman 1950c) or ''…distinctively mosaic type evolution […] makes exact delimitation of any group extremely difficult…'' (Sket and Notenboom 1993). A comparison of our results with S. Karaman's diagnosis summarized in Table II illustrates the pitfalls of previous classifications. The most prominent feature, a large and massive body, is a combination of a phylogenetically noninformative plesiomorphy (stout, gammariform shape) and convergently acquired large size typical of many subterranean crustaceans including other niphargids as well as unrelated amphipods, e.g. Megagidiella (Bogidiellidae) or Trogloleleupia (Ingolfiellidae) (Holsinger 1994;Koenemann and Holsinger 1999). Similar conclusions can be drawn also for less obvious characters.
In our case, an almost completely resolved cladistic hierarchy has been obtained. However, only some smaller groups of taxa can be considered as supported (e.g. clades a-2.1, b-1). Larger clades (e.g. clades A, a-2) suffer from week node support, a high number of homoplasies (parallelisms, reversals) and, consequently, low character consistency. In addition, multistate characters and numerous polymorphic characters might indicate rapid character evolution or parallel evolution.
Homoplasies may be explained, at least in part, as a consequence of homogeneous selection forces acting in the subterranean environment. Nevertheless, characters derived on account of ''subterranean selection'', so-called troglomorphisms, should not necessarily be a priori regarded as homoplasies and may provide substantial information on the phylogeny of subterranean animals. Marques and Gnaspini (2001) addressed that question and proposed a method for coding troglomorphisms to minimize both errors due to homoplasies as well as loss of phylogenetically relevant information. The method was flawed (Desutter-Grandcolas et al. 2003;Harris et al. 2003) and, to date, no other method has been proposed to solve the problem successfully.
However, troglomorphosis is not a general phenomenon in subterranean animals, and sometimes different, closely related species or even allopatric populations of the same species reveal different patterns of troglomorphic characters (Turk et al. 1996;Aden 2005) as well as different degrees of changes of ancestral states (Prevorčnik et al. 2004;Christiansen 2005). We may assume that parallel evolution in subterranean environments results in similar, but not the same morphological differentiation (Prevorčnik et al. 2004). Our case shows that defining character states as exactly as possible can considerably refine hypotheses of primary homology. The same character state of a multistate troglomorphic character found in two taxa may either be a synapomorphy or a homoplasy due to parallel evolution. In our study, extremely elongated antennae I and pereopods VII are synapomorphies of the clades B and b-1, respectively, while moderately elongated appendages emerge in several different lineages convergently.
In the ''Orniphargus'' case, the problem of troglomorphisms is far from being trivial. Several species exhibit traits that can be regarded as progressive (e.g. elongated appendages) or reductive (setae and spine reductions) troglomorphisms (Langecker 2000;Christiansen 2005). However, a discussion on troglomorphisms should be restricted to phylogenetically well-supported comparisons of non-troglomorphic ancestor and troglobiotic descendant relationships (Dessuter-Grandcolas et al. 2003). At present, all niphargids display at least two prominent reductive characters, which might be regarded as troglomorphic: lack of eyes and lack of integumental pigmentation. Furthermore, no epigean and non-troglomorphic niphargid ancestor is known. The presumably plesiomorphic species N. valachicus, hypothesized to have lived already in surface waters of the ancient Paratethys area (Sket 1981), does not appear basally in the cladogram. Moreover, over 30-million-year-old fossils found in Baltic amber (Coleman and Myers 2000;Coleman and Ruffo 2002) closely resemble some recent species and display some features that can be regarded as troglomorphic.

West Balkan clades A and B
Most of the large plesio-troglomorphic ''Orniphargus'' species live in the western Balkans. Most of the taxa from that territory belong to clades A and B (Figure 1). Surprisingly, N. longiflagellum is excluded from either clade, although it was originally described as N. orcinus longiflagellum. The biogeographic patterns of clades A and B resemble each other remarkably. Both clades are distributed across the western Balkans. The basal positions of both clades are held by south-eastern Dinaric species (N. podgoricensis and N. lourensis [clade A], N. zavalanus [clade B]). Further, both clades contain holodinaric (a-2.2 vs. b-2) as well as southeastern Dinaric subclades (a-2.1 vs. b-1) (biogeography of the western Balkans sensu Sket [1994]). Interestingly, the ranges of both southeastern Dinaric subclades overlap only slightly, possibly pointing to two independent centers of speciation.
Taxonomy of ''Orniphargus'' and remarks on species with ambiguous morphology While the most parsimonious trees rejected the monophyly of ''Orniphargus'' or the ''orcinus group'' of species, the support for this rejection, measured as consistency and Bremer support indices, is weak. We can, therefore, maintain that that the taxonomic validity of ''Orniphargus'' remains doubtful. However, as a comprehensive revision of the whole genus is underway, new arguments for its rejection will probably emerge. The use of this and other large groups of niphargids remains problematic for further taxonomic practice. It can hardly be expected that these groups will prove monophyletic after phylogenetic scrutiny. The subdivision into smaller and well-supported monophyla that can readily be diagnosed morphologically seems to be more productive for the subgeneric taxonomy of Niphargus (e. g. clades a-1, a-2, b-1, b-2).
Niphargus virei Chevreux, 1896 from France, and N. miljeticus Straškraba, 1959 from Mljet Island (southern Adriatic Sea) show some similarities with ''Orniphargus'' (Straškraba 1959). In addition, N. hvarensis was considered as a species that shares morphological features both with ''Stygoniphargus'' and ''Orniphargus'' (S. Karaman 1952b. The French species N. virei was recently shown to be genetically rather heterogeneous, and speciation within this ''species'' cannot be excluded (Lefébure et al. 2006). However, early designations of N. virei as N. orcinus virei  or N. miljeticus as N. virei (Schellenberg 1937) were not correct. Both N. virei and (N. miljeticus and N. hvarensis) are distinct lineages evolved independently from the rest of the studied species and from each other. Our analysis clearly positions the clade (N. miljeticus and N. hvarensis) among the socalled ''stygius sensu lato'' species. Interestingly, our results suggest that the two populations identified as N. hvarensis do not constitute a monophylum. Furthermore, N. miljeticus may be synonymous with N. hvarensis. However, the high variability in these populations demands additional analyses with a wider sampling of different populations. Because we were not able to obtain the holotypes of N. miljeticus, decisive taxonomic statements are not possible, neither are they within the scope of this paper.
Affinities between Italian and west Balkan species can only be discussed with reference to N. stefanellii, the only Italian species studied. It occupies the basal-most position among the studied ''Orniphargus'' taxa, and does neither form a monophylum with N. hebereri nor with N. skopljensis, as stated by Straškraba (1972) or G. . Niphargus polymorphus sp. n. resembles both ''Orniphargus'' (spines along posterior margin of pleonites and solid body) and N. pectinicauda (very deep coxae II-IV, well-developed posterior lobe on coxae V-VI, widened and lobated bases on pereopods V-VII with threehooked retinacles). The latter is along with N. longidactylus a presumed member of the ''transitivus'' group of species (Sket and Notenboom 1993). Sket and Notenboom (1993) recognized paraphyly of the transitivus sensu lato group, whereas the present study suggests it is polyphyletic. Testing position of N. polymorphus