Phylogeny and biogeography of the enigmatic ghost lineage Cylindrotomidae (Diptera, Nematocera)

Ghost lineages have always challenged the understanding of organism evolution. They participate in misinterpretations in phylogenetic, clade dating, biogeographic, and paleoecologic studies. They directly result from fossilization biases and organism biology. The Cylindrotomidae are a perfect example of an unexplained ghost lineage during the Mesozoic, as its sister family Tipulidae is already well diversified during the Cretaceous, while the oldest Cylindrotomidae are Paleogene representatives of the extant genus Cylindrotoma and of the enigmatic fossil genus Cyttaromyia. Here we clarify the phylogenetic position of Cyttaromyia in the stem group of the whole family, suggesting that the crown group of the Cylindrotomidae began to diversify during the Cenozoic, unlike their sister group Tipulidae. We make a comparative analysis of all species in Cyttaromyia, together with the descriptions of the two new species, C. gelhausi sp. nov. and C. freiwaldi sp. nov., and the revision of C. obdurescens. The cylindrotomid biogeography seems to be incongruent with the phylogenetic analysis, the apparently most derived subfamily Stibadocerinae having apparently a ‘Gondwanan’ distribution, with some genera only known from Australia or Chile, while the most inclusive Cylindrotominae are Holarctic.

-Wings pale brownish without conspicuous markings with end of marginal cell apically somewhat clouded (Cockerell, 1924 Diagnosis. Wing without color spots; Sc short, terminating in C before level of fork of Rs, far before level of crossvein r-m; vein r-r (R 2 ) terminating far before r′-m′ level and at m-m level, just beyond level of basal part of M 3 ; R 1 atrophied; R 2+3+4 longer than half length of Rs; d′-cell shorter than d-cell, narrowed at its base; crossvein m-cu positioned beyond fork of Mb on M 1+2 and M 3+4 ; M 3 shorter than d-cell.    According to our parsimony analysis (consensus tree), Cyttaromyia falls as sister group of all the extant genera of Cylindrotomidae, and thus belongs to the stem group of the family, and could correspond to a different subfamily. Also, the extant Cylindrotominae appear paraphyletic in respect to the Stibadocerinae because the two genera Cylindrotoma and Diogma fall in an unresolved polytomy with this subfamily plus a clade that contains the other cylindrotomine genera. Nevertheless, these results are preliminary and would need to be completed by the addition of characters, in particular molecular.
The Cylindrotominae (and also Cyttaromyia) have a Holarctic distribution, while the Stibadocerinae have a more disjunctive distribution in Indo-Malaysia, Australo-Papua and Southern Neotropics (Taiwan, China, Indonesia, Malaysia, India, Papua New Guinea, Philippines, Australia, and Chile). Such distribution resembles that of an ancient Gondwana group, with 'relic' taxa in Australia and Chile; but the present phylogenetic analysis would contradict this hypothesis, as the only known stem representative of the family is also Holarctic. Further analyses together with discoveries of fossil Cylindrotomidae in the Southern Hemisphere shall be necessary to clarify this complex, strange situation.
From a taxonomic point of view, Architipula is characterized by the occurrence of vein Sc tip beyond fork of Rs level, subequal to or a little shorter than veins R 2+3 and R 3 combined, distinctly inclined crossvein m-m between M 1+2 and M 3 , usually short and straight vein A 2 51 Cyttaromyia is characterized by the occurrence of two discal cells (d-cell and d′-cell), supernumerary crossvein r′-m′ connecting vein R 5 with M 1 near its origin, to produce two discal cells. Some similarities are present in the wing venations of Cyttaromyia and Cylindrotoma: separate M 1 and M 2 and relatively long vein M 1 . In Cylindrotoma the crossvein r′-m′ is atrophied, but the base of vein M 1 is strongly arched and only one discal cell (d-cell) is present. In all other Cylindrotominae, like Diogma (recorded from the Middle Eocene) 32 , or other genera with a younger fossil record, the crossvein r′-m′ is reduced and only one discal cell (d-cell) is present (Fig. 6B,C; Supplementary Figs. S3, S4).

Conclusion
The revision of Cyttaromyia obdurescens and the description of two new species Cyttaromyia gelhausi sp. nov. and Cyttaromyia freiwaldi sp. nov., allowed us to propose a key to the species of this genus. We have also made the first morphological phylogenetic analysis of the Cylindrotomidae, with in the rather surprising result of the putative paraphyly of the Cylindrotominae and a position of Cyttaromyia in the stem group of this family.

Material and methods
The study was based on material from the collection of the Muséum national d'Histoire naturelle (MNHN), Paris (five specimens) and American Museum National History (AMNH) (one specimen). The imprints from sediments of Green River Formation USA (age 50.3-46.2 Ma 42 ) were studied using a Nikon SMZ 1500 stereomicroscope equipped with a Nikon DS-Fi1 camera in University of Rzeszów. The microphotographs and measurements were taken with NIS-Elements D 3.0 software. Drawings were completed by tracing the photographs, nomenclature of wing venation was used 31    and M4 2x as long as distance between tip of M2 and M3; distances between tip of M2 and M4 and between tips of M4 and Cu almost equal in lengths. A2 elongate, slightly sinusoidal. Tip of A1 positioned behind fork of Rs. Haltera (Fig. 1D) elongate, shorter than 0.25x of fore wing. Abdomen (Fig. 1G) approximately 5x as long as thorax, ovipositor not very elongate.

Supplementary Data S3
Characters scored for phylogenetic analysis.
1. Proportion between length of antenna and length of body: antennae one-third body length or shorter (0); antennae comparatively long, longer than one-third body length (1).