A revision of the subgenus Dudaica Strand of the genus Drosophila Fallén, with descriptions of six new species (Diptera, Drosophilidae)

Abstract The subgenus Dudaica Strand of the genus Drosophila Fallén has been known to comprise only two species: Drosophila (Dudaica) senilis Duda, 1926 (recorded from Indonesia, Philippines, Vietnam, Bhutan, and India) and D.malayana (Takada, 1976) (recorded from Malaysia). In the present study, this subgenus is revised, with D.malayana redescribed and six new species discovered and described from China, Malaysia, and Indonesia: gracilipalpis Katoh & Gao, sp. n., puberula Katoh & Gao, sp. n., albipalpis Katoh, Toda & Gao, sp. n., qiongzhouensis Katoh & Gao, sp. n., orthophallata Katoh, Toda & Gao, sp. n., and dissimilis Katoh & Gao, sp. n. Both morphological and molecular data (DNA barcodes) are used to distinguish the above species. A key to species of this subgenus is provided.


Introduction
established the monotypic subgenus Macropalpus Duda (type species: Drosophila senilis Duda, 1926 from Sumatra) in the genus Drosophila Fallén, and defined it by the following diagnostic characters: (1) palpus distinctly large, long and broad, lacking prominent setae, (2) scutellum large, apically broadly rounded, and (3) costal break turned inwards onto thickened end of R 1 . Later, Strand (1943) proposed Dudaica as a replacement name of the subgenus Macropalpus, since the name Macropalpus had been preoccupied. In a revision of the genus Zygothrica Wiedemann, Grimaldi (1990a) transferred Z. malayana Takada, 1976 from Zygothrica into Drosophila (Dudaica), referring to Takada's (1976) original description, and mentioned that an undetermined species of Dudaica, but distinct from malayana, was present in New Guinea. According to previous records, D. senilis is widely distributed in the Oriental region, from not only Sumatra (the type locality) but also the Philippines and Java (Wheeler 1981), India (Gupta and Sundaran 1990), Bhutan (De and Gupta 1996), and Vietnam (Sidorenko 1996).
The phylogenetic position of the subgenus Dudaica remains unresolved. Grimaldi (1990b) proposed a revised phylogenetic classification of the family Drosophilidae, based on a cladistic analysis for a set of 120 species (including D. senilis) representing genera and subgenera of the family. Grimaldi's final consensus cladogram placed D. senilis most close to D. (Drosophila) monochaeta Sturtevant, 1927, and both formed a cluster with Idiomyia s. lat. However, he was "not confident of the homologies for the two features suggesting this relationship [reduction in number of interfrontal setulae (ap. 67) and a reduced, simple spermatheca (ap. 217)]". Yassin (2013) revised the subgeneric classification of Drosophila in light of molecular and morphological data, and proposed diagnoses for the subgenera, including Dudaica. However, his proposal that Dudaica is closely related to the genera Hirtodrosophila Duda, Paraliodrosophila Duda and Zygothrica was elicited solely by a single morphological trait, i.e., the shape of the "gonopods", but not by molecular data.
In this paper, we revise the subgenus Dudaica, and add six new species discovered from China, Malaysia, and Indonesia to this subgenus, by identifying them with the aid of DNA sequences of the 658-bp barcoding region of the mitochondrial COI (cytochrome c oxidase subunit I) gene. We also redescribe the known species D. malayana, based on specimens newly collected from Malaysia and Indonesia. Finally, a key to all the eight species of Dudaica is given.

Materials and methods
All specimens employed in the present study were collected from China, Malaysia, and Indonesia (Table 1). They were mostly captured from herb layer in forest by net sweeping, and preserved immediately in either 70% or 100% ethanol for morphological observation and DNA sequencing, respectively. Specimens were first identified as of the subgenus Dudaica, based on their overall resemblance to the two known species of this subgenus, D. senilis and D. malayana, especially in body color pattern, shape of palpus, and structures of male/female terminalia. The holotype specimen of D. malayana was examined for the reference. As for D. senilis, we referred to Duda's (1926) original description, Gupta and Sundaran's (1990) redescription of terminalia, and Grimaldi's (1990b) character states in his cladistic analysis. All specimens were then sorted into known or putatively new species in light of morphology. For this, external morphology was examined, numbers of morphometric characters were measured, and detailed structures in male/ female terminalia, head and mouth parts were observed by the same methods as in Li et al. (2014).
The specimens were then subjected to DNA barcoding analysis (Hebert et al. 2003), with total DNA extracted from a right hind-or mid-leg, or small piece(s) of abdominal tissue picked from the abdominal dissection cut, using the TIANamp ® Genomic DNA Kit. DNA sequences of the 658-bp barcoding region of the mitochondrial COI gene were then amplified with the Folmer et al.'s (1994) primer pair, following the procedures as in Li et al. (2014). The PCR products were purified and sequenced with ABI3730 sequencer. The obtained DNA sequences were edited and aligned in the Seq-Man module of the DNAStar package (DNAStar Inc. 1996) and MEGA7 (Kumar et al. 2016), respectively. A molecular phylogenetic tree was constructed by using Bayesian Inference (BI) method in MrBayes v3.2.6 (Ronquist et al. 2012), with the sequence data partitioned into two subsets by codon position, i.e., 1st+2nd codon positions and 3rd codon position. In BI, two independent runs of MCMC with four chains each (three heated and one cold) were conducted simultaneously for 5,000,000 generations, and trees were sampled every 100 generations. The analysis was stopped after verifying convergence statistics using Tracer v1.6 , and the first 20% of the tree samples were discarded as burn-in. Nucleotide substitution model was determined for each data set using jModelTest 2.1.10 (Guindon andGascuel 2003, Darriba et al. 2012) using the Bayesian Information Criterion (BIC; Schwarz 1978). In addition, we employed the Automatic Barcode Gap Discovery (ABGD; Puillandre et al. 2012) and the General Mixed Yule-coalescent (GMYC; Pons et al. 2006) analyses for the molecular species delimitation. The ABGD analysis was run on the web-interface (http://wwwabi.snv.jussieu.fr/public/abgd/abgdweb.html) with the default settings [P min = 0.001, P max = 0.1, Steps = 10, X (a proxy for minimum gap width) = 1.5, Nb bins (for distance distribution) = 20]. All three distances applicable in the web-interface, JC69 (Jukes and Cantor 1969), K2P (Kimura 1980), and simple distances (i.e., p-distances) were used for the analyses. The GMYC was performed using the package "splits" (http://rforge.r-project.org/projects/splits) in R, with the single-threshold strategy and default scaling parameters. An ultrametric tree for the GMYC was generated by BEAST v2.4.5 (Bouckaert et al. 2014) using the Yule prior and the HKY (Hasegawa et al. 1985) with a proportion of invariable sites (+I) model, with 5,000,000 MCMC generations. In addition, the intra-and inter specific p-distances for the species in Dudaica were calculated with MEGA7 and summarized. For species illustration, the external morphology and detailed structures of male and female terminalia, and head and mouth parts were microphotographed using a Dino-Lite ® Microscope Eyepiece Camera (ANMO Electronics Corporation). We followed McAlpine (1981) for morphological terminology, and Zhang and Toda (1992) for definitions of measurements and indices. The examined specimens are deposited in the following institutions:

Results
A total of 34 COI sequences of 658-bp were determined in this study (Table 1). We failed to determine the COI sequence for the male specimen #00177 (of D. orthophallata sp. n. to be described here), probably due to poor quality of the total DNA extracted from this specimen, which was collected in 1999. Also, one female specimen of D. malayana, which was the oldest one examined here, was not used for DNA analysis. The HKY+I model was selected for both of the '1st+2nd codon positions' and '3rd codon position' partitions as the best nucleotide substitution model for BI analysis. Figure 1 illustrates the unrooted BI tree built with the 34 COI sequences and the results for the molecular species delimitation. The ABGD and GMYC analyses resulted in the same hypothesis: the studied sequences were sorted into seven hypothetical species, except for at P (prior intraspecific divergence) = 0.001 in ABGD using JC69 and K2P distances where eight species including a paraphyletic one were proposed. These seven Molecular Operational Taxonomic Units (hypothetical species) were supported by morphological data as well. In addition, the highest intraspecific (i.e., within-lineage) p-distance was 0.0163, while the lowest interspecific (i.e., amonglineage) p-distance was 0.0796, indicating a broad barcoding gap (Table 2). Thus, in consequence of integrative species delineation based on molecular and morphological data, we recognized seven (one known and six new) species within our studied samples of the subgenus Dudaica: D. malayana (Takada, 1976), D. gracilipalpis sp. n., D. puberula sp. n., D. albipalpis sp. n., D. qiongzhouensis sp. n., D. orthophallata sp. n., and D. dissimilis sp. n.
Among them, D. qiongzhouensis sp. n. and D. albipalpis sp. n. are very similar in morphology to each other. The diagnoses for these species are supplemented with "pure" molecular diagnostic characters, which are defined as sites with fixed status in the COI sequence alignments within the focal species but differing from the other species (Sarkar et al. 2002, DeSalle et al. 2005 (Table 3).
Thorax (Figure 2A): Postpronotal lobe milky white in upper half, dark brown in lower half, with 2-3 prominent but no short setae. Right and left dorsocentral setae nearly parallel. Notopleuron milky white. Thoracic pleura with three, sometimes rather indistinct stripes.
Legs (Figures 2A,10A): Foreleg first tarsomere shorter than total length of four succeeding tarsomeres. Mid-and hind-leg first tarsomeres shorter than or as long as total length of four succeeding tarsomeres.
Thorax ( Figure 2B): Postpronotal lobe with two prominent and rarely 1-2 short setae. Stripes on thoracic pleura mostly confluent with each other.
Legs ( Figure 2C): First tarsomeres of all legs shorter than total length of four succeeding tarsomeres; mid-leg first tarsomere with one subproximal short, black spine.
Legs ( Figures 2D): First tarsomeres of all legs slightly shorter than total length of four succeeding tarsomeres.
Measurements ( Diagnosis. Palpus slender, not so flat, somewhat rod-shaped in lateral view, with a few moderate setae on outer lateral surface ( Figure 9E). Paramere apically rounded in lateral view ( Figure 14K). Nucleotide status in COI sequence = C and C at sites 92 and 226, respectively (Table 3).
Thorax ( Figure 3A): Postpronotal lobe with 1-2 prominent and 1-3 short setae. Legs ( Figure 3A): Foreleg first tarsomere shorter than total length of four succeeding tarsomeres. Mid-and hind-leg first tarsomeres slightly shorter than or nearly as long as total length of four succeeding tarsomeres.
Legs ( Figure 3B): Foreleg first tarsomere shorter than or as long as total length of four succeeding tarsomeres. Mid-and hind-leg first tarsomeres as long as total length of four succeeding tarsomeres.

Distribution. Malaysia (Sabah).
Etymology. Referring to the straight aedeagus in the new species. Remarks. The paratype female specimens #03905 and #03906 were identified as conspecific with the holotype male specimen #00177 (DNA sequence data unavailable), based on close morphological affinity between them. This species can also be distinguished from the other Dudaica species by oviscapt valve with 17-19 marginal ovisensilla ( Figure 15I, J) in addition to the diagnosis. Diagnosis. Palpus short, club-shaped, medio-to baso-laterally dark grayish brown, with one prominent seta apically and several long setae ventrally ( Figure 9G). Cercus without caudoventral process ( Figure 16E). Paramere apically somewhat quadrate in lateral view ( Figure 16J), ventroapically with three sensilla (Figure 16I, J). Aedeagus distally dilated laterally, somewhat lunate in lateral view ( Figure 16F-H).
Description (characters in common with D. orthophallata sp. n. not repeated). Adult ♂. Head ( Figures 3C, 4G, 5G, 7C, 8G, 9G): Longest axis of eye nearly rectangular to body axis. Frontal vitta grayish white. Fronto-orbital plate slightly grayish; anterior reclinate orbital seta situated between proclinate and posterior reclinate orbital setae. Occiput and postgena dark brown, marginally milky white. Arista with six dorsal and three ventral branches. Supracervical setae 16-19 per side; postocular setae 17-19 per side. Cibarium slightly thickened on anterior margin; medial sensilla ca. nine per side and posterior sensilla ca. nine per side; first and second medial sensilla weaker than and anteriorly apart from others. Clypeus not thickened at median portion, laterally dark brown.
Thorax ( Figure 3C): Postpronotal lobe pale brownish white in upper half, dark brown in lower half; setae broken. Dorsocentral and scutellar setae broken. Notopleuron pale brownish white. Thoracic pleura nearly entirely dark brown, without stripes.
Wing ( Figure 3C) slightly wrinkled at basal portion of R 4+5 ; longitudinal veins pale brown except for basal section of M 1 (brown) and CuA 1 (brown).
Legs ( Figure 3C) pale grayish yellow to pale yellow. Foreleg first tarsomere shorter than total length of four succeeding tarsomeres. Mid-and hind-leg first tarsomeres slightly shorter than total length of four succeeding tarsomeres, without subproximal spine.
Distribution. China (Yunnan). Etymology. Referring to the morphological difference from the other species in the subgenus Dudaica.
Remarks. This species is the most different in morphology from the other Dudaica species, such as the shape of palpus and parameres, apical prominent seta on palpus, and number of sensilla on parameres. Those characters are also seen in many other species than Dudaica, suggesting the plesiomorphic states of these characters.