Egg parasitoids of the tea green leafhopper Empoascaonukii (Hemiptera, Cicadellidae) in Japan, with description of a new species of Anagrus (Hymenoptera, Mymaridae)

Abstract Fairyfly (Hymenoptera, Mymaridae) egg parasitoids of the tea green leafhopper Empoasca (Matsumurasca) onukii Matsuda (Hemiptera, Cicadellidae), an economically important pest in Asia of the tea plant, Camelliasinensis, were identified from specimens reared in Japan. Using a combination of genetic and morphological evidence, Anagrus (Anagrus) rugmanjonesi Triapitsyn & Adachi-Hagimori, sp. n., is described and illustrated. It is shown to be different from the most similar A.turpanicus Triapitsyn & Hu, an egg parasitoid of a leafhopper pest of cultivated grapes which is known from Xinjiang Uyghur Autonomous Region in China. Mitochondrial and nuclear ribosomal DNA sequence data provide clear evidence for the separation of A.rugmanjonesi from A.turpanicus and other members of the Anagrusincarnatus Haliday species complex. A key to females of the Japanese species of Anagrus Haliday is given. Two other species of Mymaridae, Aresconenocki (Subba Rao & Kaur) and Stethyniumempoascae Subba Rao, are also identified, albeit the latter one only tentatively. Both latter taxa are newly recorded from Japan, and E.onukii represents their new host association.


Introduction
The tea green leafhopper, Empoasca (Matsumurasca) onukii Matsuda (Hemiptera, Cicadellidae) (Fig. 1) is one of the major pests of tea plants in Japan and also in mainland China and Taiwan where it has been commonly misidentified as Empoasca vitis (Göthe) and Jacobiasca formosana (Paoli) (or as Empoasca formosana Paoli), respectively (Qin et al. 2015). Adults and nymphs of E. onukii cause leaf vein reddening, leaf margin yellowing, leaf curling, stunted shoot growth, and leaf drop, which results in economic losses of up to 33% (Xu et al. 2005). Eggs of E. onukii are laid singly, embedded in the soft tissues of tea bushes, such as veins of leaves and tender stems (Takagi 1978).
In Japan, tea green leafhoppers have developed resistance to the insecticides (Ozawa et al. 2009) used intensively against this pest. Thus, development of alternative control methods is desirable. Egg parasitoids offer one potential alternative for regulating tea green leafhopper populations. Takagi (1978) found mymarid wasps (Hymenoptera, Mymaridae), known as fairyflies in English, in tea fields in Japan, and later, Ojima et al. (2010) provided data on the population dynamics of three species parasitizing eggs of E. onukii in tea plantations in Kochi Prefecture. However, biological control-based integrated pest management (IPM) using these egg parasitoids has not yet been established.
Unfortunately, voucher specimens of the study by Takagi (1978) could not be located, and those of Ojima et al. (2010) were lost (I. Ojima personal communication). Thus, as the first step towards establishment of biological control-based IPM using egg parasitoids, we collected fairyflies in organic tea fields and identified them both morphologically and genetically.

Specimen collection
Tea shoots were collected from three organic tea fields in Takaoka (Takaoka, fields 4, 5, 6), Miyazaki City on October 10, 17, and 25, and in one organic tea field in Kitakata (Kita, field 1), Nobeoka City, Miyazaki Prefecture, on October 20, 2017. All tea plants belonged to variety 'Yabukita'. In each field, 75-95 new shoots (15-20 cm length) were collected, put into plastic bags, kept in a cooler box containing ice, and transported to the Laboratory of Applied Entomology, University of Miyazaki, Miyazaki. The shoots were then transferred to two different container sets for observing either eclosion of the nymphs of tea green leafhoppers or emergence of egg parasitoid adults. The first container type consisted of plastic bottles covered with black opaque plastic film. The bottom of the bottle was cut off and replaced with the lid of a candy bottle. The latter was filled with a wetted garden sponge into which 20 tea shoots with one leaf per shoot were inserted. A transparent plastic test tube was screwed on the top of the plastic bottle. This system allowed for observation and collection of the emerged wasps from tea shoots through the test tube, as they are attracted to light. The second set of containers consisted of test tubes. A tea shoot without leaves was inserted into a small cut of wet garden sponge. The shoot inserted into wet garden sponge was put into a test tube and sealed with parafilm. The emerged wasps were collected every 24 hours and were provided with honey solution until they died naturally. The dead wasps were collected, labeled, placed in 99.5% ethanol and stored at -20 °C until they were shipped to the first author. These specimens were used both for molecular analyses and taxonomic studies (as type material of the new species described below).

Taxonomic studies
Morphological identifications of the Anagrus sp., made by the first author, were based mainly on females because males of many species of Anagrus Haliday are often similar.
Results of the genetic analysis were key in determining the separation of the new species of Anagrus from A. turpanicus Triapitsyn & Hu from Xinjiang Uyghur Autonomous Region in China; this species is an egg parasitoid of a leafhopper pest of culti-vated grapes, Arboridia kakogawana (Matsumura) (Hu and Triapitsyn 2016), which is the most similar based on morphology of both sexes. The genetic analysis was also useful to separate the new species from other members of the Anagrus incarnatus Haliday species complex (Triapitsyn et al. , 2019. For the taxonomic description of the new species, the morphological terms of Gibson (1997) and Triapitsyn (2015) were used. All measurements (as length or length: width for the wings) are given in micrometres (µm). Abbreviations used in the description and key are:

F
funicle segment of the female antenna or flagellomere of the male antenna; mps multiporous plate sensillum or sensilla on the antennal flagellar segments (= longitudinal sensillum or sensilla, or sensory ridge(s)).
Specimens from ethanol were dried using a critical point drier, then point-mounted and labeled. Selected specimens were dissected and slide-mounted in Canada balsam. Slide mounts were examined under a Zeiss Axioskop 2 plus compound microscope (Carl Zeiss Microscopy, LLC, Thornwood, New York, USA) and photographed using the Auto-Montage system (Syncroscopy, Princeton, New Jersey, USA). Photographs were retouched where necessary using Adobe Photoshop (Adobe Systems, Inc., San Jose, California, USA).
Specimens examined are deposited in the collections with the following acronyms: BLKU Biosystematics Laboratory, Faculty of Social and Cultural Studies, Kyushu University, Fukuoka, Japan; UCRC Entomology Research Museum, Department of Entomology, University of California, Riverside, California, USA.

DNA extraction, amplification, and sequencing
DNA was extracted from two individual female wasps using the "HotSHOT" method of Truett et al. (2000), in a total volume of 80 µL. This non-destructive method allowed for the recovery and slide-mounting of each specimen following extraction; each slide was then labeled with the assigned P. F. Rugman-Jones' primary molecular voucher PR number and UCRC database UCRC ENT number. For reasons described below, DNA was also extracted from one male (PR18-486) using a Chelex 100 method described by Stouthamer et al. (1999). This specimen was destroyed by grinding and is listed below under "Other (non-type) material examined". Two "preliminary" specimens of the same species of Anagrus were also subject to a destructive extraction protocol in Japan. The polymerase chain reaction (PCR) was employed in an attempt to amplify the "barcoding" region of the mitochondrial cytochrome c oxidase subunit I gene (COI) using LCO1490 (5'-GGTCAACAAATCATAAAGATATTGG-3') and HCO2198 (5'-TAAACTTCAGGGTGACCAAAAAATCA-3'; Folmer et al. 1994), as described in Rugman-Jones et al. (2012). This primer combination has previously proved good for use with HotSHOT-extracted specimens of Anagrus (e.g. Triapitsyn et al. 2018Triapitsyn et al. , 2019. However, in this instance, amplification of COI from the HotSHOT-extracted specimens failed for these and several alternative primer combinations (data not shown). In contrast, amplification of COI from the Chelex 100 -extracted specimen, and the two "preliminary" specimens extracted in Japan, worked fine. Reactions were performed in 25 µL volumes on a Mastercycler ep gradient S thermocycler (Eppendorf North America Inc., New York, New York, USA) and amplification was confirmed by gel electrophoresis.
All PCR products were cleaned using a DNA Clean & Concentrator™-5 kit (Zymo Research Corporation, Irvine, California, USA) and direct sequenced in both directions at the Institute for Integrative Genome Biology, University of California at Riverside. The parity of forward and reverse reads was checked using SEQUENCHER 4.9 (Gene Codes Corporation, Ann Arbor, Michigan, USA) and priming regions were removed manually in BioEdit version 7.0.5.3 (Hall 1999). The online tool, EMBOSS Transeq (Rice et al. 2000) was used to translate the protein coding COI sequence into its amino acid chain, confirming the absence of indels and pseudogenes. All sequences were deposited in GenBank (Benson et al. 2008).

Genetic analysis
Representative COI sequences previously obtained by Triapitsyn et al. (2018Triapitsyn et al. ( , 2019 for members of the incarnatus species complex (and associated out-group taxa) were combined with the current COI data. The sequence data was subsequently aligned using MAFFT version 7.050 (Katoh and Standley 2013) and the Q-INS-i algorithm with default settings. The aligned COI dataset contained 23 terminal taxa (including outgroups), 587 nucleotide positions, and no gaps. Genetic variation among our sequences was estimated by calculating uncorrected p-distances between all possible sequence pairs, using MEGA version 6 (Tamura et al. 2013). All ambiguous positions were removed for each sequence pair. A neighbor-joining (NJ) tree based on those p-distances was subsequently constructed, again using MEGA. Branch support was estimated using a bootstrap procedure with 1000 replicates.
As phylogenetic inference from ITS2 is typically problematic due to large interspecific differences that make alignment of this region difficult and somewhat ambiguous, ITS2 sequences were examined "by eye" to corroborate the status of our specimens as a single species, and to compare them with other Anagrus species by using a BLAST search of the NCBI database.  Diagnosis. The new species is a member of the incarnatus species group of the Anagrus (Anagrus) as defined by Chiappini et al. (1996), and its A. incarnatus species complex, studied by Triapitsyn et al. (2018). Female antenna (Fig. 3b) with F2 not the longest funicular segment (usually F4 is or, sometimes, F6); mps on F4 (1), F5 (1 or 2), F6 (2), and clava (5); midlobe of mesoscutum without adnotaular setae (Figs 4d, 5b); fore wing disc sometimes with a distinct, but small subapical bare area (Fig. 4b) but often this bare area is either somewhat indistinct (Fig. 4a) or absent (Fig. 4c); ovipositor 2.3-2.5× length of protibia.

Taxonomy
Morphologically, A. rugmanjonesi is most similar to the Palaearctic species A. turpanicus, to which its female specimens with a more or less distinct bare area on the fore wing disc key in Li et al. (2018). Both taxa have F2 of the female antenna not the longest funicular segment whereas in the other members of the A. incarnatus species complex it is the longest one (Triapitsyn 2015;Hu and Triapitsyn 2016;Triapitsyn et al. 2018). In A. turpanicus, however, the mesosoma is mostly yellowish brown except anterior half or so of mesoscutum is brown and frenum is yellowish white (Hu and Triapitsyn 2016), whereas in A. rugmanjonesi the scutellum and mesosoma (laterally, except the pronotum) are contrastingly white (Fig. 2a). Also, in A. rugmangonesi the clava is at most as long as the combined length of F5 and F6 whereas it is always longer than that in A. turpanicus. The fact that the two species also substantially differ genetically (Fig. 7) provides a good justification for their differentiation as two separate entities. In Li et al. (2018), those specimens of A. rugmanjonesi that lack a more or less distinct bare area on the fore wing disc key to Anagrus nilaparvatae Pang & Wang, a well-known egg parasitoid of rice planthoppers (Hemiptera, Delphacidae) and leafhoppers in Asia. The latter taxon was recently synonymized under Anagrus incarnatus, and F2 of its female antenna is always the longest funicular segment . A key to females of the Japanese species of Anagrus is provided below, as the previous key by Sahad and Hirashima (1984) is outdated.
Description. Female (holotype and paratypes). Body length of dry-mounted, critical point-dried paratypes 400-460 µm, and of the slide-mounted paratypes 560-590 µm. Body (Figs 2a, 3a) mostly brown to dark brown except face, gena, and propodeum light brown and scutellum and mesosoma laterally (except pronotum) white; posterior half or so of mesoscutum and apex of gaster often light brown to offwhite; scape, pedicel and F1 pale to light brown, remaining funicular segments light brown, and clava brown; legs mostly pale to light brown, wings hyaline. Antenna (Fig. 3b) with scape 2.9-3.8× as long as wide, with cross-ridges, 1.9-2.2× length of pedicel; F1 a little longer than wide, about half of pedicel length; F2 at least slightly shorter than following funicular segments, F4 usually the longest funicular segment (except sometimes F6 the longest); mps on F4 (1); F5 (1 or 2), and F6 (2); clava with 5 mps, 2.8-3.3× as long as wide, either as long as combined length of F5 and   F6 or slightly shorter. Midlobe of mesoscutum without adnotaular setae (Fig. 4d). Fore wing (Fig. 4a-c) 7.0-8.0× as long as wide, longest marginal seta 2.6-2.9× maximum wing width; distal macrochaeta 1.6-1.7× length of proximal macrochaeta; disc with several rows of setae in addition to admarginal rows of setae (1 complete row originating behind apex of venation and 3 or 4 irregular rows in the broadest part of disc), sometimes leaving a distinct, but small subapical bare area at posterior margin ( Fig. 4b) but often this bare area either somewhat indistinct (Fig. 4a) or absent (Fig.  4c). Hind wing (Fig. 4a) 22-24× as long as wide, longest marginal seta 6.0-7.0× maximum wing width; disc mostly bare except for an incomplete row of setae along anterior margin and a compete row of setae along posterior margin. Ovipositor (Fig. 3c) extending anteriorly almost to mesophragma in slide-mounted specimens and exserted a little beyond apex of gaster posteriorly (by 0.12-0.15× total ovipositor length). Second valvifers (= external plates of ovipositor), e.g., Chiappini et al. (1996), each with 3 setae (Fig. 3c) Male (paratypes). Body length of the slide-mounted paratypes 560-585 mm. Body color mostly as in female except entire flagellum brown. Antenna (Fig. 5a) with scape 2.4-2.7× as long as wide, F1 at least a little shorter than following flagellomeres. Fore wing 7.2-7.6× as long as wide, with or without (Fig. 5d) a more or less bare area in the broadest part. Genitalia (Fig. 5c) length 124-127 µm.
Etymology. This new species is named by the first author in honor of his colleague and one of the co-authors of this communication, Paul F. Rugman-Jones, whose contributions towards determination of the identities of the nominal taxa within the Anagrus incarnatus species complex using molecular methods and genetic analyses have been invaluable.
Distribution. Palaearctic region: Japan. Host. Cicadellidae: Empoasca (Matsumurasca) onukii Matsuda. Biology. In eggs of E. onukii on tea plants, A. rugmanjonesi was observed to develop as a solitary endoparasitoid (Ojima et al. 2010). Takagi (1978) monitored population dynamics of A. rugmanjonesi (as Anagrus sp.) in a tea plantation by using sticky suction traps. The dynamic curve indicated that A. rugmanjonesi was a multivoltine species and was most abundant in September. The study site of Takagi (1978) was not mentioned but is known to be the former Kanaya Town, Shizuoka Prefecture, Japan (K. Takagi personal communication), which is now part of Shimada City.
Comments. The photographs of "Mymaridae sp. A" provided in Ojima et al. (2010) leave no doubt that their specimens from Kochi Prefecture, Shikoku Island belonged to both sexes of Anagrus rugmanjonesi n. sp. Frenum of scutellum with triangular paramedial plates widely separated from each other; metafemur short, less than 2× trochanter length, trochantellus incision almost halfway between coxa-trochanter and femur-tibia articulations

Arescon enocki (Subba Rao & Kaur, 1959)
Neurotes enocki Subba Rao & Kaur, 1959: 233 (illustrations), 235-237, 238 (key Comments. This species was redescribed and illustrated by Subba Rao (1966) and Triapitsyn (2016) based on specimens from India, so it is quite easily recognizable; particularly, Triapitsyn (2016) provided habitus digital images of both sexes of this species. The photographs of "Mymaridae sp. C" provided in Ojima et al. (2010) leave no doubt that their specimens belonged to both sexes of A. enocki, which thus is newly recorded from the Palaearctic region, where it is definitely an Oriental fauna element in southern Japan where tea is grown. Unfortunately, as their material was lost, we are unable to further illustrate the Japanese specimens of this species.
Comments. The photographs of "Mymaridae sp. B" provided in Ojima et al. (2010) leave no doubt that their specimens belonged to both sexes of a Stethynium sp., which almost certainly were conspecific with ours from the same genus.
As discussed by Triapitsyn (2002), S. empoascae is extremely similar morphologically to usually lighter-colored specimens of Stethynium triclavatum Enock to the extent that it may be impossible to distinguish them in some countries (like China, Egypt, India, Japan, Nepal, Pakistan, etc.) where both species can potentially occur. Yet, females of S. empoascae from Australia and India, which could be a different, more subtropical and tropical species, seem to be slightly different from the majority of the European and North American specimens of S. triclavatum, which supposedly occurs in the countries with a more temperate climate (Triapitsyn 2002). Ultimately, molecular studies comparing freshly preserved specimens from Australia, Europe, India, Japan, and North America (now lacking) would need to be conducted to confirm separation of these two nominal species or, otherwise, provide genetic evidence of their possible conspecificity. At this point, however, we can only tentatively assign our specimens to S. empoascae based on some of the very minor morphological features mentioned in Triapitsyn (2002) as well as the fact that they were collected in Japan on the two islands with a subtropical climate. To facilitate recognition of this species, we provide illustrations of its female antenna (Fig. 6a), mesosoma and metasoma (Fig. 6b), and a pair of wings (Fig. 6c).

Molecular analyses
Sequences of the COI gene provided strong evidence that A. rugmanjonesi is distinct from A. turpanicus and other members of the A. incarnatus species complex. Three COI haplotypes were identified for A. rugmanjonesi, with maximum 1.9% divergence (based on uncorrected p-distances) among those haplotypes (GenBank accessions MK544853-MK544855; Fig. 7). All substitutions were synonymous. In turn, A. rugmanjonesi was at least 5.1% divergent from all other accepted species in the incarnatus species group, with A. turpanicus being the most similar (Fig. 7). The exact relationship between A. rugmanjonesi and A. turpanicus (and indeed other species) was largely unresolved, with weak branch support towards the base of the NJ tree (Fig. 7).
The ITS2 sequences of the three PRJ-extracted specimens were each 603 bp long and identical, confirming them as a single species (GenBank accessions MK564750-MK564752). A BLAST search revealed no match with anything currently in the GenBank database; the closest accessions again belonging to Anagrus turpanicus (MK024909-MK024911). A MAFFT alignment of our sequences with those of A. turpanicus resulted in a matrix with many substitutions and several sizable indels, resulting in a difference in length of approximately 40 bp.

Discussion
This study further confirms the effectiveness of simple molecular techniques for separating morphologically similar species, in this case A. rugmanjonesi, A. turpanicus, and other members of the A. incarnatus species complex (Fig. 7). Description of A. rugmanjonesi would be difficult based solely on morphology because of its close morphological similarity to A. turpanicus. The convincing molecular data confirms that two species are involved. Their separation also makes sense from the habitat point of view, the latter species being an egg parasitoid of a leafhopper pest of grapevines in a very hot and dry environment of a desert oasis in Xinjiang Uyghur Autonomous Region of China. Morphologically, however, they are clearly two sister species, and that is also corroborated by the genetic data presented herein. Lin (1981) reported one undescribed species of Megaphragma Timberlake (Hymenoptera, Trichogrammatidae) as an egg parasitoid of E. onukii (as Empoasca formosana Paoli) on tea plant in Taiwan, but that record was obviously erroneous, likely due to an inadequate rearing method, because members of this genus are known to be egg parasitoids of thrips (Thysanoptera).