Six years of fruit fly surveys in Bangladesh: a new species, 33 new country records and discovery of the highly invasive Bactrocera carambolae (Diptera, Tephritidae)

Abstract We engaged in six years of snap-shot surveys for fruit flies in rural environments and ten protected forest areas of Bangladesh, using traps baited with male lures (cue-lure, methyl eugenol, zingerone). Our work has increased the recorded number of species of Tephritidae in the country from seven to 37. We summarize these surveys and report eight new country occurrence records, and a new species (Zeugodacus madhupuri Leblanc & Doorenweerd, sp. nov.) is described. The highlight among the new records is the discovery, and significant westward range extension, of Bactrocera carambolae Drew & Hancock, a major fruit pest detected in the Chattogram and Sylhet Divisions. We rectify the previously published erroneous record of Bactrocera bogorensis (Hardy), which was based on a misidentification of Zeugodacus diaphorus (Hendel). We also report the occurrence in Bangladesh of nine other Tephritidae, the rearing of three primary fruit fly parasitoids from Zeugodacus, and records of non-target attraction to fruit fly lures.


Introduction
The Dacini is a very diverse group of fruit flies, with 939 described species, including 83 pests of cultivated fruit and cucurbits (e.g., Doorenweerd et al. 2018). Of these, 118 are known to occur on the Indian subcontinent (David and Ramani 2011;Drew and Romig 2013;David et al. 2016David et al. , 2017Leblanc et al. 2018b). Fruit fly surveys in rural environments of Bangladesh, initiated in 2013, increased the published number of known species from seven to 27 (Leblanc et al. 2013(Leblanc et al. , 2014Hossain and Khan 2013;Khan et al. 2015Khan et al. , 2017. While Drew and Romig (2013) could not confirm the presence of Bactrocera dorsalis (Hendel) on the Indian subcontinent, variation in color pattern and preliminary molecular data from Bangladesh and African populations suggested that B. dorsalis is widespread on the subcontinent and that the species described as B. invadens Drew, Tsuruta & White is conspecific with B. dorsalis (Leblanc et al. 2013). That same year, B. philippinensis Drew & Hancock was declared a synonym of B. papayae Drew & Hancock (Drew and Romig 2013). Soon after B. papayae, along with B. invadens, were declared conspecific with B. dorsalis, with formal designation of synonyms (Schutze et al. 2015a(Schutze et al. , 2015b, leaving B. carambolae Drew & Hancock as a distinct species, based on genetic differences, morphological differences in aedeagus, wing shape and color pattern, non-random assortative mating with B. dorsalis, and significant differences in pheromone composition (Wee and Tan 2007;Schutze et al. 2012Schutze et al. , 2013Schutze et al. , 2015bTan et al. 2013). With this revised status, B. dorsalis is now widespread across tropical Asia, and introduced to most of Africa and several islands in the Pacific, while B. carambolae has been restricted to a smaller range in South-East Asia (Fig. 1) and introduced to South America. To generate a complete inventory of the economic species and assess the diversity of fruit flies in the protected forest areas of Bangladesh, we surveyed for fruit flies during 2013-2018 with a focus on rural areas and report here cumulative results from these surveys, focusing on previously unpublished new records. Using a morphological and molecular approach, we discovered numerous new country records, including the highly invasive B. carambolae, and a new species of Zeugodacus Hendel is described here.

Collecting and curation
Starting in 2013, we periodically maintained a series of traps (described in Leblanc et al. 2015a) separately baited with male lures plus a 10×10 mm piece of dichlorvos (DVVP) insecticide strip to kill trapped flies. Cue-lure and methyl eugenol were included as commercially available plugs (Scentry Biologicals, Billings, Montana) whereas zingerone lure, also used in the surveys since 2016, was prepared by dipping dental cotton wicks in zingerone (= vanillylacetone) (Sigma-Aldrich) melted over a hot plate and allowed to solidify in the wicks. We deployed traps at 383 sites through- out the country for periods ranging from one to 14 days, either as individual sites scattered over rural areas or as series of 11-26 sites, about 50 m apart, concentrated in selected rural areas and in 10 different protected forest areas (Nishorgo Support Project 2007) (Fig. 2, Table 1). Sampled flies were stored in 95% ethanol in a -20 °C freezer, to preserve DNA for analysis. All flies were identified by the first three authors, using available keys Romig 2013, 2016). Before drying flies for doublemounting, we pinned them through the scutum with a minuten pin and soaked them in ethyl-ether for 3-12 hours to fix and preserve their natural coloration. We photographed specimens using a Nikon D7100 camera attached to an Olympus SZX10 microscope and used Helicon Focus Pro ver. 6.7.1 to merge pictures taken at a range of focal planes. To measure specimens, we used an ocular grid mounted on an Olympus SZ30 dissecting microscope.    We also reared parasitoids and hyperparasitoids from readily available, heavily fly-infested snake gourd (Trichosanthes cucumerina) collected at the AERE campus (Dhaka). Infested gourds were weighed and placed on a cloth-covered small bowl (to collect excess juice from decay), over moist sawdust (as a pupation media) in a fine nylon netted cage. Pupae were separated from the sawdust using fine-meshed sieve and placed in a petri dish inside a very fine-netted plastic cage to collect emerged fruit flies and parasitoids.

Morphological terms and taxonomic assignment
Morphological terminology used in the descriptions follows White (1999) and assignment of species to genera follows Doorenweerd et al. (2018). The genus Zeugodacus, of which a new species is described in this paper, is treated as separate from Bactrocera Macquart and Dacus Fabricius, based on recent molecular-based phylogenetic assessments (Krosch et al. 2012;Virgilio et al. 2015;Dupuis et al. 2017;San Jose et al. 2018a). Despite recent efforts to reassign species to subgenera (e.g., Hancock and Drew 2018 a, b), the understanding of higher relationships of species within Dacini is still in state of flux, and a number of traditionally recognized subgenera and species complexes (Drew and Romig 2013) are demonstrated to be polyphyletic groups of convenience defined on the basis of highly homoplastic morphological characters and male lure relations (e.g., Leblanc et al. 2015b;San Jose et al. 2018a;Catullo et al. 2019). For this reason, we have not attempted to include subgenera in the country's species list.

DNA extraction, PCR and sequencing
Methods for DNA extraction, PCR primers and conditions, and Sanger sequencing follow those of San Jose et al. (2018a). We attempted to amplify and sequence regions of the Cytochrome C Oxidase I (COI) and Elongation Factor 1-alpha (EF1-alpha) genes. It has previously been shown that COI cannot be used to differentiate Bactrocera dorsalis from B. carambolae ). However, we found that there are five diagnostic single nucleotide polymorphisms (SNP's) that separate B. dorsalis from B. carambolae in the 762 base-pair (bp) fragment of EF1-alpha that we used for multimarker phylogenetic studies (San Jose et al. 2018a). We therefore sequenced this segment to confirm or refute the identity of B. carambolae. For Zeugodacus madhupuri, we attempted to amplify a large section of 1540 bp of COI as well as EF1-alpha, but we only successfully amplified the COI-3P region. Amplified regions of COI-5P proved to be nuclear pseudogenes after sequencing (data not shown), and EF1-alpha did not yield any PCR product, possibly due to degradation of the template DNA. We aligned newly generated sequences with the published data of EF1-alpha or COI, respectively, from San Jose et al. (2018a) and performed maximum likelihood analyses using IQTree (Nguyen et al. 2015). We allowed IQTree to determine the substitution model via its integrated modeltest and ran a standard maximum likelihood analyses with 1000 ultrafast bootstraps and 1000 Sh-aLRT bootstraps. We consider branches with support values >95% for ultrafast bootstraps and >80% for Sh-aLRT bootstraps as well supported. Resulting trees were optimized for publication using FigTree 1.4.3 and Adobe Illustrator. Data from this study are available from the BOLDSYSTEMS Digital Repository: https://doi.org/10.5883/DS-BANG01.

Estimating biodiversity
We used EstimateS software (Colwell 2013) to generate species accumulation curves. We estimated species diversity with the incidence-based Chao 2 algorithm, which does not include abundance in its extrapolation, thereby avoiding abundance bias in our data related to how strongly each species is attracted the lures and controlling for the predominance of a few agricultural pests in the samples. Diversity estimations were done comparing forest and rural sites, and the individual protected forest areas separately, with 100 randomizations without replacement for confidence intervals. It is understood that diversity estimates are underestimations, because they are based solely on species attracted to the male lures used in our sampling.

Results
Between April 2013 and September 2018, we collected a total of 23,939 specimens of Dacine fruit flies, representing 29 species (Table 1), among 1012 samples (372 cuelure, 357 methyl eugenol, 271 zingerone; and a few others hand-collected or bred from fruit) across 383 sites (Fig. 2). We report a number of new country occurrence records, including a major pest species in need of management attention (Bactrocera carambolae) and describe a new species, increasing the number of species of Tephritidae in Bangladesh from 27 (Leblanc et al. 2013(Leblanc et al. , 2014Hossain and Khan 2013;Khan et al. 2015Khan et al. , 2017 to 37 (29 Dacini and eight from other tribes).

Biodiversity and species accumulation curves
Rural sites were dominated by three pest species: Bactrocera dorsalis (61.3% of specimens captured), Zeugodacus cucurbitae (Coquillett) (17.5%), and Z. tau (Walker) (12.6%). Forest sites also yielded large numbers of B. dorsalis (39.1%), as well as the non-economically important B. rubigina (Wang & Zhao) (41.4%), whereas cucurbit pests were less common (2.6% Z. cucurbitae and 9.2% Z. tau). The Chao 2 algorithm estimated overall number of species is 30.0 in forest sites and 25.7 in rural sites (Fig. 3A). Among the surveyed protected forests (Fig. 3B), the highest diver- Estimates of species numbers based on the Chao 2 estimator, with the 95% confidence interval ranges). Data used to generate these curves and estimates include two species not yet definitely identified and not included on Table 1. sity was collected in three locations in the Chattogram District: Sita-Kunda Eco-Park, (15 species), Fashiakali Wildlife Sanctuary (12 species) and Chunati Wildlife Sanctuary (12 species), with the estimated Chao 2 number of species ranging from 15 to 18. Most other sites had a moderate diversity of 6-10 species, with estimated numbers of 8-12 species. For unclear reasons, only three species were collected in Lawachora National Park, possibly due to trapping done during the tail end of the rainy season (October 2016) and/or its relatively small size and geographic isolation from other forested areas within a densely populated environment dominated by agriculture. Paradoxically, the smallest sampled protected area, Sita-Kunda Eco-Park (808 ha) yielded the highest species diversity. The estimated total number of species in Bangladesh, based on the Chao 2 algorithm from a species accumulation curve including all sites, is 37.5 species, relatively few compared to the 118 species known to occur in the Indian subcontinent.
A higher estimate might have been attained had species not attracted to lures been more actively collected and had access to the Chattogram Hills tracts forests not been severely restricted due to security concerns.

Bactrocera carambolae new to Bangladesh
We collected 167 specimens of B. carambolae among 55 methyl eugenol samples, mostly in protected forest sites in the Chattogram District (Table 1, Fig. 2). One specimen morphologically consistent with B. carambolae was collected further north, in the Rema-Kalenga Wildlife Sanctuary (Sylhet Division) (UHIM molecular voucher ms07278), but its identity could not be confirmed molecularly because the amplification of EF1-alpha failed repeatedly, possibly due to degradation. All specimens are morphologically consistent with the diagnostic features of B. carambolae: subapical spots on fore femora, costal band slightly overlapping and expanded beyond apex of R 2+3 , presence of narrow transverse black band across anterior margin of tergum III, widening to cover lateral margins (Drew and Hancock 1994; Drew and Romig 2013) (Fig. 4). Species identity of a selection of nine specimens was further confirmed through sequencing of a region of EF1-alpha with five diagnostic SNP's that differentiate B.    (David et al. 2017), but differs in that the fuscous medial band and lateral markings on the abdomen are pale and less extensive than in Z. brevipunctatus, dark marking on legs are fulvous rather than fuscous, and Z. madhupuri consistently has two pairs of equally well-developed scutellar setae. Zeugodacus brevipunctatus, along with most other species of subgenus Sinodacus Zia has only one pair of scutellar setae (Hancock and Drew 2018a).
Molecular diagnostics. We obtained COI-3P sequences for three specimens, aligned them with the available COI-3P sequences from San Jose et al. (2018a) and performed maximum likelihood analyses. The full tree is available in Supplementary material 2: Figure S2, and a subset of Z. madhupuri and its closest relatives is shown in Figure 5. Based on our reference dataset, the new species is most similar to Z. hengsawadae (Drew & Romig) and Z. heinrichi (Hering) at around -11% pairwise distance and can be diagnosed reliably using COI. Note however that Zeugodacus brevipunctatus was not represented in our COI dataset.
Description of adult. Head (Fig. 6A, B). Vertical length 1.65-1.95 mm. Frons, of even width, 1.11-1.23 times as long as broad; fulvous with anteromedial hump covered by short red-brown hairs; orbital setae large-sized and dark fuscous and strong: one pair of superior and three pairs of inferior fronto-orbital setae present, the most anterior pairs nearly contiguous; lunule yellow. Ocellar triangle black. Vertex fuscous. Face fulvous to yellow with a broad transverse black band at mid height; length 0.58-0.75 mm. Genae fulvous, with or without a faint fuscous subocular spot; red-brown seta present. Occiput fulvous and yellow along eye margins; with two pairs of large occipital dorsal setae and lateral occipital rows with 5-8 light to dark setae. Antennae with scape, pedicel and first flagellomere fulvous and arista black (fulvous basally); length of segments: 0.20-0.25 mm; 0.25-0.35 mm; 0.83-0.93 mm.
Wings (Fig. 7A). Length 6.78-7.78 mm; basal costal (bc) and costal (c) cells fuscous; microtrichia in outer corner of cell costal only; remainder of wings with a pale fulvous tint except fuscous subcostal cell, broad fuscous costal band overlapping con-  fluent with R 4+5 , a large dark fuscous apical spot from apex of R 2+3 , and englobing apical portions of veins R 4+5 and M (from interception with dm-cu), a broad fuscous anal streak ending at apex of A 1 + CuA 2 ; dense aggregation of microtrichia around A 1 + CuA 2 ; supernumerary lobe well developed.
Abdomen (Figs 6C, D, 7C). Elongate oval and petiolate; terga free; pecten present on tergum III; posterior lobe of surstylus long (Fig. 7C); abdominal sternum V with a shallow concavity on posterior margin. Tergum I as long as wide and sterna I and II longer than wide. Tergum I fulvous with apical margin narrowly yellow. Tergum II yellow with a median fulvous heart-shaped marking with an inner anteromedial fuscous marking. Terga III-V fulvous with fuscous along basal margin of tergum III, a narrow median longitudinal band reaching apex of tergum V, narrowly along lateral margins of terga III-IV, and as broad lateral markings on tergum V, anterior to ceromata, and with pale fulvous along base of tergum IV and on tergum V medially, anterior to ceromata (shining spots), which are also pale fulvous.
Etymology. The species name is an adjective that refers to the Madhupur National Park, where all specimens were collected.

Other new records
Bactrocera abbreviata (Hardy) new country record: One specimen collected in zingerone-baited trap, in October 2016, at the Institute of Nuclear Medicine and Allied Sciences (Chattogram). Known from the Philippines, China and Thailand (Drew and Romig 2013). Doorenweerd et al. (2018) noted that it may be conspecific with and junior synonym to B. bipustulata Bezzi, known from Sri Lanka and southern India. It has been bred in Thailand from Chionanthus ramiflorus and Olea salicifolia (family Oleaceae) (Allwood et al. 1999). Bactrocera pendleburyi (Perkins) new country record: Eight specimens in six zingerone samples in forest sites of Chattogram District (Sita-Kunda, Chunati, Fashiakhali). Previously known from Peninsular Malaysia and Thailand, its presence in Bangladesh is a significant range extension. A non-pest species bred from Symplocos cochinchinensis, S. racemosa (Symplocaceae), and Gmelina arborea (Verbenaceae) (Allwood et al. 1999 (Drew and Romig 2013), it was recently recorded from India (David et al. 2017). The use of zingerone-baited traps collected large numbers in Sri Lanka , Bangladesh (Table 1), and as far east as Vietnam (Leblanc et al. 2018a) and south as Sarawak and Sulawesi (CD, unpublished), consistent with the widespread distribution of rose-apple.  (Leblanc et al. 2013(Leblanc et al. , 2014) as belonging to this species.
Other Tephritidae: Diarrhegma modestum (Fabricius) (Acanthonevrini) was originally collected in Dhaka by Hossain and Khan (2013), and more recently in rural areas of Rajshahi District. Khan et al. (2017)  Tephraciura basimacula (Bezzi) (Tephrellini) new record: one specimen was hand-collected at the AERE, in Dhaka. This species, also known from southern India and Sri Lanka, breeds in flowerheads or seedpods of Acanthaceae (Hancock 2010).

Non-target records
A total of 5626 specimens of Ankylopteryx anomala (Brauer) (Neuroptera, Chrysopidae) were collected in methyl eugenol traps, almost all in the forested areas. This species is widespread across tropical Asia, from Sri Lanka to Taiwan, and its attraction to methyl eugenol is well documented (Leblanc et al. 2015c). Two unidentified moths in the family Crambidae were caught in two separate methyl eugenol traps in Sita-Kunda Eco-Park. These may be real instances of attraction, because attraction of two species of flower-visiting crambids was demonstrated in Hawaii (Leblanc et al. 2009). Likewise, six specimens of one unidentified bee species in the genus Amegilla (subgenus Zonamegilla) (Hymenoptera, Apidae) ( Fig.  8A, B) were collected in zingerone-baited traps in Sita-Kunda (three specimens in three traps), Chimbuk Hills (3 specimens in one trap) and Chunati (one specimen in one trap), and one specimen of the same species entered a methyl eugenol trap in Chattogram. The zingerone attraction record is credible and worth further investigation. One specimen of Amegilla calceifera (Cockerell) was also caught in a zingerone trap in Nepal in 2017 (LL, unpublished). The single specimen in the methyl eugenol trap may be accidental, though honeybee attraction to that lure has been reported (Leblanc et al. 2009).

New records of fruit fly parasitoids
Psyttalia fletcheri (Silvestri) (Hymenoptera, Braconidae) new record: One kilogram of infested snake gourd yielded 427 fruit fly puparia, from which emerged 81 specimens of Zeugodacus cucurbitae, 286 specimens of Z. tau, and 43 specimens of P. fletcheri. A laboratory colony of this species was established at AERE in preparation for a pilot area-wide control program. Adult P. fletcheri are fed with a 10% sugar solution and honey, and oviposit in third instar larvae of Z. cucurbitae and Z. tau. Psyttalia fletcheri was rarely observed in commercially cultivated crop fields, likely due to the frequent applications of pesticides and low prevalence of alternate wild host fruits surrounding the fields. Spalangia sp. and Pachycrepoideus vindemmiae (Rondani) (Hymenoptera, Pteromalidae) new records: Two 5 kg bottle gourds (Lagenaria siceraria) each infested with larvae of Z. cucurbitae and Z. tau, yielded 383 puparia and seven Spalangia and 387 pupae and nine P. vindemmiae, respectively. Laboratory colonies of both parasitoids were established at AERE. Hosts determined to be suitable for both species are puparia of Z. cucurbitae, Z. tau, B. dorsalis, and B. zonata with a preference for Z. cucurbitae by Spalangia and for B. dorsalis by P. vindemmiae. , with partial support from USDA funding to the University of Idaho. We greatly appreciate help from Dan Nitta with molecular work. Terry Griswold (USDA-ARS Bee Laboratory, Logan, Utah, USA) has identified the bees collected in the zingerone traps and Dave Hancock has identified the specimen of Tephraciura basimacula.