Introduction
The millipede fauna of the Crimea has recently been reviewed (Golovatch 2008), with only 14 species from 11 genera, seven families and six orders being involved. Of these, only two or three species are possibly Crimean endemics, whereas most show more or less widely (Euro-)Mediterranean distributions.
The more so important is the present discovery of still another new millipede in a cave in the Crimea. Unlike all other faunal elements, this new species might prove to represent the first truly relict, palaeoendemic in the diplopod list of that peninsula.
Material serving as the basis for the present contribution was captured with pitfall traps, later transferred into 75% alcohol and is currently deposited in the collection of the Zoological Museum, State University of Moscow, Russia. Specimens were studied and illustrated using standard stereomicroscopic, photographic and drawing equipment.
Systematic position of Caucasodesmus
Originally, Caucasodesmus was treated as a genus of the small Holarctic family Macrosternodesmidae (Golovatch 1984/85). However, Shear and Shelley (2007), in their recent reassessment of the Macrosternodesmidae, have ejected Caucasodesmus from that family, leaving it unclassified. These authors have also advanced a new terminology of the various parts of a polydesmidan gonopod. This has been further refined even more recently (Shear et al. 2009), in particular in accepting such denominations as exo- and endomere.
The superfamily Trichopolydesmoidea can be defined by its gonopod prefemoral (= setose) part orientated mostly transversely to the body’s main axis, extending mesally across the entire width of the coxae (Hoffman 1982; Simonsen 1990). Within this superfamily, where Caucasodesmus undoubtedly belongs, there are several, mainly small families in addition to Macrosternodesmidae: Trichopolydesmidae, Neoarctodesmidae, Furhmannodesmidae and Mastigonodesmidae. To find a new, more suitable place for Caucasodesmus, their diagnoses must briefly be reiterated, especially as regards their gonopod conformation.
Macrosternodesmidae (Shear and Shelley 2007; Shear et al. 2009): In this Holarctic family, the gonopod aperture is large, transversely oval. The coxae completely fill the respective halves of the aperture, excavated mesad to accommodate the telopodites; the prefemora are horizontal or angling ventromesad, giving rise to the acropodite and often, but not always, to an additional projection; the acropodite part distal to the origin of a solenomere (distal zone) variably configured, sometimes folded, flattened, and not recognizable as such; the solenomere is long and narrow, arising subterminally, with neither a hairpad nor an accessory seminal chamber (= ampulla); the seminal groove opens terminally.
Nearctodesmidae (Shelley 1994; Shear et al. 2009): This small Nearctic group shares basically the same gonopod conformation with Macrosternodesmidae. No wonder it has sometimes been treated as only a subfamily or even a possible synonym of the latter family.
Trichopolydesmidae (Ceuca 1958; Mauriès 1983): This small, mainly Mediterranean family shows basically the same gonopod structure as the previous two groups, except in the prefemoral part sometimes being shortened (e.g. Galliocookia Ribaut, 1955, Occitanocookia Mauriès, 1980 and a few others), i.e. rather strongly resembling the condition observed in the family Polydesmidae (see Hoffman 1980; Simonsen 1990), while the telopodite is bi- or uniramous, far less elaborate, often with a long flagelliform solenomere. Yet I am inclined to follow Mauriès (1983) in treating such somewhat deviating genera as representing rather peculiar Trichopolydesmoidea.
Mastigonodesmidae (Mauriès 1980, 1982): This very small, purely western Mediterranean group of polydesmidean millipedes is sometimes regarded as only one of the numerous genera of Polydesmidae (Hoffman 1980; Simonsen 1990), apparently because the gonopod prefemoral part in Mastigonodesmus Silvestri, 1898, is also shortened, but, due to its globose gonocoxae and a peculiar, parabasal, long and coiled solenomere, it seems more similar to trichopolydesmoids. So I am again inclined to follow (Mauriès (1980, 1982) in regarding this group as representing rather peculiar Trichopolydesmoidea as well.
Fuhrmannodesmidae (Golovatch 1994): This profoundly diverse, pantropical group of small polydesmideans shows a wide range of situations transitional in gonopod conformation between the typical Polydesmoidea and the typical Trichopolydesmoidea. At least in the Neotropical fauna, the gonopod coxae can be small and devoid of a gonocoel, with (sub)erect telopodites, yet more often the coxae are enlarged and deeply excavate for the accommodation of more stout, usually elaborate telopodites that can have a shortened to medially stretched prefemoral part supporting either crossing or parallel acropodites, the latter with or without a distinct solenomere. This highly heterogeneous assemblage certainly merits splitting into several natural families, but such a task is by necessity to be deferred because of the numerous genera and species involved, many of which still require revision.
Based on the above diagnoses and distributions, it appears to be quite difficult to unequivocally reallocate Caucasodesmus. The correlated absence of both a cannula and a seminal groove is probably a sufficiently strong apomorphy to erect still another family of Trichopolydesmoidea for the accommodation of solely this genus, but I refrain here from doing so pending more information becomes available. New taxa are still being regularly described, new synonymies established, and old types revised. Instead I reassign Caucasodesmus to Trichopolydesmidae as a family not only representing the oldest taxon in the superfamily, but also one which shows the same basic traits of gonopod structure and a coherent distribution pattern.