Taxonomy of the Colocasiomyia gigantea species group (Diptera, Drosophilidae), with descriptions of four new species from Yunnan, China

Abstract Species of the genus Colocasiomyia de Meijere feed/breed on inflorescences/infructescences of the plants from the families Araceae, Arecaceae and Magnoliaceae. Although most of them utilize plants from the subfamily Aroideae of Araceae, three species of the recently established C. gigantea species group make use of plants of the subfamily Monsteroideae. We describe four new species of the gigantea group found from Yunnan, China: Colocasiomyia longifilamentata Li & Gao, sp. n., C. longivalva Li & Gao, sp. n., C. hailini Li & Gao, sp. n., and C. yini Li & Gao, sp. n. The species delimitation is proved in virtue of not only morphology but also DNA barcodes, i.e., sequences of the partial mitochondrial COI (cytochrome c oxidase subunit I) gene. Some nucleotide sites with fixed status in the alignment of the COI sequences (658 sites in length) are used as “pure” molecular diagnostic characters to delineate species in the gigantea group.


Introduction
To date, as many as 90 species (of them only 25 species described) have been found in the genus Colocasiomyia de Meijere, 1914. All these species visit and breed on flowers of the families Araceae, Arecaceae or Magnoliaceae (Sultana et al. 2006. So far, six species groups, i.e., baechlii, cristata, toshiokai, crassipes, zeylanica and gigantea (Okada 1990, Sultana et al. 2002, have been established in this genus. The gigantea group was recently erected by Fartyal et al. (2013) for two species [C. gigantea (Okada, 1987) and C. scindapsae Fartyal & Toda, 2013] from Southeast Asia and one species (C. rhaphidophorae Gao & Toda, 2013) from China (Table 1). Host plants of these species belong to the Rhaphidophora clade of the subfamily Monsteroideae of True Araceae, according to the most recent phylogenetic studies of the Araceae (Cusimano et al. 2011, Nauheimer et al. 2012. On the other hand, the other species groups utilize either the subfamily Aroideae or the other families. Fartyal et al. (2013) investigated the reproductive ecology of the three species, and revealed interesting ecological and morphological adaptations of the flies to the host plants. In addition, Fartyal et al. (2013) conducted a cladistic analysis of 70 morphological characters of 34 Colocasiomyia species, covering all the six species groups of the genus. Their results lent essential support to the monophyly of the gigantea group, placing it as the sister clade to the cristata species group.  (Okada, 1987) Java, Indonesia; Solomon Is. Epipremnum pinnatum (L.) Engle Fartyal et al. (2013) Our recent field surveys in Yunnan Province, China brought new, insightful information on the evolution of flower-breeding habits in the gigantea group. We found four new species of this group visiting inflorescences of Rhaphidophora decursiva (Roxb.) Schott (Table 1); at least three of them were found breeding on inflorescences/infructescences of this plant. In Pu'er (central-southern part of Yunnan), C. rhaphidophorae cohabited with the above-mentioned three new species on inflorescences/infructescences of R. decursiva. Thus, the Chinese members of the gigantea group are mostly sympatric and overlapping in host plant selection with each other (cohabitation), in contrast to the allopatry and monopolization of host plant in the other members, C. gigantea and C. scindapsae.
The four new species of the gigantea species group are described here, based on species delimitation in virtue of morphological and molecular (DNA sequences of the mitochondrial cytochrome c oxidase subunit I gene, COI, as the DNA barcoding marker) characters. Table 2 shows the fly samples/specimens involved in the present study. Most of them were collected from the field in southwestern China, Malaysia and Indonesia. Some were reared from inflorescences/infructescences of host plants; after dissection of inflorescences/infructescences under a stereoscopic microscope in laboratory, the fly eggs were isolated and transferred into Petri dishes with decayed pistils as food and then reared at 25°C in an incubator until adults emerged.

Morphological observation
We followed the same method as in Fartyal et al. (2013) for the observation of external morphology, measurement of morphometric characters and preparation of dissected organs. The male and female terminalia and cibarium of the new species were microphotographed using a DinoLite Digital Eyepiece Camera and drawn according to these digital pictures using CORELDRAW® X4 (Corel Corporation). Fine structures of foreleg (tibia, 1st and 2nd tarsomeres) and oviscapt were microphotographed for all the seven species of the gigantea group using a HITACHI TM3000 tabletop scanning electron microscope (SEM). We followed McAlpine (1981) for the morphological terminology, and Zhang and Toda (1992) for the definitions of measurements and indices. New type specimens were deposited in Kunming Natural History Museum of Zoology, Kunming Institute of Zoology, Chinese Academy of Sciences (KIZ) and Hokkaido University Museum, Hokkaido University, Sapporo, Japan (SEHU).

DNA barcoding
A total of 54 individuals representing all the three known and four morphologically identified, putatively new species (Table 2) were subjected to DNA sequencing of the COI barcode fragments (Hebert et al. 2003). In addition, three more species, C. colocasiae (Duda, 1924), C. sulawesiana Okada & Yafuso, 1989 and C. xenalocasiae (Okada, 1980) from the sister cristata group were included, for comparison of sequences. DNA was extracted using small piece(s) of abdominal tissue or the right hindleg of a single adult using the TIANamp® Genomic DNA Kit. The primer pair used for the PCR and sequencing of the COI fragment was either that designed by Folmer et al. (1994: 5'-GGTCAA CAAAT CATAA AGATA TTGG -3', HCO2198: 5'-TAAAC TTCAG GGTGA CCAAA AAATC A -3') or that by Hebert et al. (2004, LepF1: 5'-ATTCA ACCAA TCATA AAGAT ATTGG -3', LepR1: 5'-TAAAC TTCTG GATGT CCAAA AAATC A -3'). The 20 µl PCR reaction volume contains 0.1 µl TaKaRa Ex Taq® (5 U/µl), 0.4 µl of each primer (10 µM), 2 µl 10× Ex Taq Buffer (Mg + Plus), 1.6 µl dNTP mixture (2.5 mM for each), 15 µl ddH 2 O and 0.5 A total of 57 COI sequences (54 of the gigantea group and three of the cristata group) were determined. The sequences were edited in the SEQMAN module of the DNASTAR package (DNASTAR Inc. 1996), and aligned in MEGA5 (Tamura et al. 2011). Then the inter-and intraspecific genetic distances were calculated for the species of the gigantea group using the Kimura 2-parameter (K2P) model in MEGA5. In addition, we also conducted a character-based species barcoding. Since there were end gaps (missing data) in some of the sequences of each species, the end sites with an overlapping of less than three (two for C. longivalva) sequences were excluded. The sites being fixed within a species but differing from the other species (including the three species of cristata group) were manually selected as diagnostic sites (i.e., "pure" diagnostics, Sarkar et al. 2002, DeSalle et al. 2005, for each species.

DNA barcoding
The alignment of the 57 COI sequences spanned 658 nucleotide sites in length, with 184 variable sites, among which 160 were parsimony informative. For the inter-and intraspecific K2P distances see Table 3. The largest intraspecific K2P distance in the gigantea group was found in C. scindapsae (= 0.0102), while the smallest interspecific one was found between C. rhaphidophorae and C. longifilamentata (= 0.0135). This implies that the "barcoding gap" (Meyer and Paulay 2005) is too narrow, only 0.0033, to validate the distance-based species delimitation in the gigantea group. Fig. 1 shows nucleotides at the sites where "pure" diagnostics for any species of the gigantea group are included. At least one diagnostic site was recognized for each species. For example, the site 226 is diagnostic for C. rhaphidophorae: this site has a fixed status of C (Cytosine) in this species, but T (Thymidine) in the other species. The sites 136 and 505 (both with fixed status of C) are diagnostic for C. longifilamentata.
Remarks. The characters described to be common among the three known species of the gigantea group by Fartyal et al. (2013) are shared also by the four new species to be described here, except for the distance between antennal sockets larger than the socket width: the former is as large as the latter in C. hailini sp. n. Description. Adult male. Head: Supracervical setae 12-15 per side. Dorsomedial arm of tentorial apodeme approximately 1/2 as long as dorsolateral arm. Eye red, somewhat roundish, with very sparse interfacetal setulae. Frontal vitta mat, black. First flagellomere not concave on inner margin. Facial carina trapeziform, medially wider than twice width of first flagellomere, as long as pedicel and first flagellomere combined. Palpus convex on ventrodistal portion. Cibarial posterior sensilla minute, 2 or 3 per side (Fig. 23). Labellum with 22 pseudotracheae per side.
Female. Head, thorax, wing and legs as in male. Terminalia: Tergite VII mid-dorsally not constricted; VIII entirely pubescent, with 5 setae in transverse (against body axis) row on discolored posteroventral portion. Oviscapt distally narrowing; distal narrow portion as long as proximal, broad portion, with approximately 15, 6 and 3 trichoid ovisensilla per side on ventral, dorsal and apical margins, respectively, and a tiny, peg-like ovisensillum near dorsosubapical margin (Fig. 28).  Egg. Filaments 2, approximately 1.8-2.4 times as long as length of egg body. Etymology. The specific name "longifilamentata" refers to the long filaments of egg. Distribution. China (Yunnan). Remarks. Although this species closely resembles C. rhaphidophorae in the external morphology and structures of male and female terminalia, it can be easily distinguished from the latter by the epandrium having several setae on the dorsal to lateral portion (Fig. 24)  Description. Adult male. Head: Supracervical setae 13 per side. Dorsomedial arm of tentorial apodeme 1/3 as long as dorsolateral arm. Eye red, somewhat roundish, with very sparse interfacetal setulae. Frontal vitta mat, black. First flagellomere not concave on inner margin. Facial carina trapeziform, medially wider than twice width of first flagellomere, as long as pedicel and first flagellomere combined. Palpus convex on ventrodistal portion. Cibarial posterior sensilla minute, 1 per side (Fig. 29). Labellum with 20 pseudotracheae per side.
Measurements ( Female. Head, thorax, wing and legs as in male. Terminalia: Tergite VII mid-dorsally not constricted; VIII pubescent nearly entirely, with 3−4 setae in a transverse row on unpubescent medio-posterior portion. Oviscapt distal narrow portion twice as long as proximal, broad portion, with approximately 18, 6 and 3 trichoid ovisensilla per side on ventral, dorsal and apical margins, respectively, and tiny, peg-like ovisensillum near mid-dorsal margin (Fig. 36).
Measurements ( Remarks. Adults of this species were very rarely captured from inflorescences of R. decursiva. So far we have never get any adult of this species by laboratory rearing from the host inflorescences/infructescences. This species is distinguished from the other members of the gigantea group: epandrium somewhat notched above insertion of ventral lobe; epandrial ventral lobe prolonged like rod, apically with grooved, finger-like peg (Figs 30-32); surstylus 1/3 as long as epandrial ventral lobe, distally nearly parallel with the latter (Figs 31, 33); paramere large, as long as hypandrium (Figs 34, 35); distal narrow portion of oviscapt twice as long as proximal, broad portion (Fig. 36). Description. Adult male. Head: Supracervical setae approximately 7 per side. Dorsomedial arm of tentorial apodeme 1/3 as long as dorsolateral arm. Eye red, somewhat roundish, with very sparse interfacetal setulae. Frontal vitta mat black. First flagellomere concave on inner margin. Facial carina trapeziform, medially wider than twice width of first flagellomere, as long as pedicel and first flagellomere combined. Palpus convex on ventrodistal portion. Cibarial posterior sensilla minute, 1 per side (Fig. 37). Labellum with 11 pseudotracheae per side.
Abdomen: Tergites glossy, entirely dark brown; II to VI+VII each bearing setulae and setae in approximately 3 transverse rows; setae of posteriormost row largest. Sternites pale brown to brown; VI posteriorly not bilobed.
Wing: Veins yellow. Halter dark brown except for grayish yellow stalk. Legs: Foreleg second tarsomere with 6 pegs (Figs 8, 15). Foreleg coxa large, with approximately 10 long setae on underside near attachment to trochanter. Small preapical dorsal setae present only on tibiae of hindlegs.
Abdomen: Tergites glossy, entirely dark brown; II to VI+VII each bearing setulae and setae in approximately 3 transverse rows; setae of posteriormost row largest. Sternites II-V pale brown; VI blackish brown, and bilobed posteriorly.
Measurements ( Remarks. Much fewer adults of this species were collected from inflorescences of R. decursiva in comparison to C. hailini sp. n. and C. longifilamentata sp. n. Breeding of this species on R. decursiva was confirmed by laboratory rearing of eggs laid on inflorescences of the host plant. See the Remarks for C. hailini sp. n. with respect to morphological differences from it.