Morphological and molecular identification of arrhenotokous strain of Diglyphuswani (Hymenoptera, Eulophidae) found in China as a control agent against agromyzid leafminers

Abstract Diglyphus species are ecologically and economically important on agromyzid leafminers. In 2018, a thelytokous species, Diglyphuswani Liu, Zhu & Yefremova, was firstly reported and described. Subsequently, the arrhenotokous D.wani were discovered in Yunnan and Guizhou Provinces of China. We compared the morphological characteristics of thelytokous and arrhenotokous strains. However, the females of two strains had a strongly similar morphology and showed subtle differences in fore- and hind-wings. The difference was that forewing of arrhenotokous female was with denser setae overall, showing that costal cell with 2 ~ 4 rows of setae on dorsal surface and the setae of basal cell with 15 ~ 21 hairs and forewing of thelytokous female was with two rows of setae on dorsal surface and basal cell with 10 ~ 15 hairs generally. The setation beneath the marginal vein of the hind-wing of arrhenotokous female is denser than the same area of thelytokous female. To explore the genetic divergence between thelytokous and arrhenotokous strains of D.wani, the mitochondrial and nuclear gene were applied and sequenced. The polygenic analyses revealed that two strains can be distinguished by COI, ITS1 and ITS2. The mean sequence divergence between the two strains was 0.052, 0.010 and 0.007, respectively. Nevertheless, the 28S gene was unfeasible due to its containing a sharing haplotype between different strains. The two strains of D.wani are dominant parasitoids against agromyzid leafminers and such effective discernible foundation provides future in-depth studies on biological characteristics, along with insight into field application of two strains of D.wani.


Introduction
Agromyzidae belongs to Diptera and is a family consisting of about 2750 species (Tschirnhaus et al. 2000) and approximately 110 species of them are known to be the main pests of cultivated crops world-wide (Dempewolf 2020). In China, over 130 Agromyzidae species have been reported. Of these, at least six species, including indigenous Chromatomyia horticola, Liriomyza chinensis and invasive L. sativae, L. huidobrensis, L. trifolii and L. bryoniae, are major agricultural leaf-mining pests, especially on vegetables Liu et al. 2013). For decades, the main prevention for agromyzid leafminers has been chemical control with pesticides ). With the frequent use and abuse of chemical pesticides, agromyzid leafminers have gradually developed resistance to insecticides (Parrella and Keil 1984;Tokumaru and Yamashita 2004) and natural enemies have decreased (Trumble and Toscano 1983;Hernández et al. 2011). Therefore, it requires sustainable, effective and biocontrol strategies to regulate the damage of agromyzid leafminers. Notably, applying Hymenoptera parasitoids are considered to be primary strategies, because these species are the most effective natural enemies against agromyzid leafminers (Parrella 1987;Liu et al. 2009;Mujica and Kroschel 2011;Ridland et al. 2020).
In Hymenoptera parasitoids, some species have two reproduction modes: (1) arrhenotoky, where haploid males arise from unfertilised eggs and diploid females from fertilised eggs and (2) thelytoky, which is obligate parthenogenesis and produces only female progenies or occasional males (Heimpel and de Boer 2008). Amongst Diglyphus species, a thelytokous parasitoid named D. wani was firstly reported and displayed favourable biocontrol potential showing three types of host-killing behaviour (hostfeeding, parasitism and host-stinging) (Ye et al. 2018).
In arthropods with haplodiploid sex determination mechanism, thelytokous strains may exist with their corresponding arrhenotokous strains (van der Kooi et al. 2017). In Eulophidae, several species with two strains (reproduction modes) have been reported, such as Neochrysocharis formosa (Adachi-Hagimori et al. 2011;Yang et al. 2017) and Pnigalio soemius (Gebiola et al. 2012). For D. wani, whether there is also an arrhenotokous strain is not clear. In the field investigations, we firstly discovered arrhenotokous D. wani in Yunnan Province of China, which was a dominant parasitoid on agromyzid leafminers and established a stable colony in the laboratory. We preliminarily attempted to make a morphological distinction, but two strains of D. wani were likely to be so similar that it would be difficult to discriminate each other accurately. However, accurate identification was essential for potential application of D. wani. Thus, in addition to traditional morphological classification, molecular methods were also adopted, because multiple gene markers, such as the cytochrome c oxidase subunit I gene (COI) and nuclear internal transcribed spacers (ITS1 and ITS2), have been also applied widely for species identification (Campbell et al. 1993;Chen et al. 2004;Sha et al. 2006;Munro et al. 2011;Om et al. 2017;Ye et al. 2018).
In this paper, the combination of morphological and molecular tools (COI, ITS1, ITS2 and 28S) was applied to characterise and compare differences between arrhenotokous and thelytokous strains of D. wani. The results will promote the future biocontrol application of two strains of D. wani.

Morphological Identification
The collected parasitoid samples were transferred to plastic tubes filled with 99.7% ethanol and then stored at -20°C for subsequent classification. These samples were examined with a stereomicroscope (Olympus Corporation, SZX-16, Tokyo, Japan). Terminology and measurement methods referred to Gibson (2003). The abbreviations used are: F1-F2, first to second flagellomeres; SMV, MV, PMV and STV, which are submarginal, marginal, post-marginal and stigmal veins; OOL, the minimum distance between an eye margin and the adjacent posterior ocellus; and POL, the minimum distance between the posterior ocelli. Measurements of body, gaster and ovipositor lengths were taken using an optical microscope (Keyence Corporation, VHX-2000, Tokyo, Japan). Relative measurements were used for the other parts. The ratio of gaster to ovipositor was calculated in Microsoft Excel 2016 using Mean ± SD (standard deviation). Photographs of arrhenotokous and thelytokous D. wani were taken by an Olympus CX31 microscope and an Olympus BX43 microscope with a Helicon Focus system, respectively. Of Diglyphus parasitoids that we surveyed, D. crassinervis was close to D. wani relatively in terms of morphology. Additionally, D. isaea was a common parasitoid on agromyzid leafminers. We selected the two species to discover further phylogenetic relationships between them and D. wani.

Parasitoid DNA extraction
Using the QIAGEN blood or tissue genome kit (Germany) we followed the steps according to the manufacturer's standard protocol of kit to extract DNA of a single parasitoid. The DNA was stored at -20°C for molecular research.
After the PCR reaction, taking 4 μl of the PCR product, mixing it with 0.3 μl of 10× Loading buffer, then electrophoresing products in 1% agarose solution containing Gold View II (Solarbio, Beijing, China), setting voltage 100 V, current 400 mA and 30 minutes. After the electrophoresis, we observed the results in the gel imaging system and saved the photos. The PCR unpurified products containing the target bands were sent to Tsing Ke Biological Technology, Beijing of China, for Bi-directional sequencing.
When the gene sequence peak map showed double peaks in Bi-direction, the sequences needed to be cloned. After the PCR products were purified, the target fragments were ligated into the pEASY-T3 cloning vector (Transgen Biotech, Beijing, China) and transferred into E. coli competent cells Trans-T1 (Transgen Biotech, Beijing, China) according to the manufacturer's instructions. Finally, using the universal M13 vector primer to detect whether the target fragments were successfully connected, each sample tested five positive clones to evaluate the difference between clones. In this study, the sequence divergence of clones of every sample was small about 0 ~ 0.003, usually about 0.001. Thus, we randomly selected a sequence for phylogenetic analysis.

Sequence analysis
All sequences were analysed by BLAST (Basic Local Alignment Search Tool) in the NCBI database to determine whether the amplified sequences belonged to mitochondria and nuclear genes. The sequences were aligned by using the CLUSTAL W tool of MEGA 7.0 (Kumar et al. 2016) and using the default options. Pairwise and mean sequence divergence, variation sites and parsimony informative sites were estimated, based on the Kimura-2 parameter (K2-P) (Kimura 1980). For COI, the sequences were translated into the amino acid sequence, based on the invertebrate mitochondrial genetic code so as to examine no stop codes. Then, version 5 of the DNASP (Librado and Rozas 2009)was used to calculate gene haplotypes.

Phylogenetic analysis
The phylogenic tree was constructed with UPGMA (the unweighted pair group method, based on arithmetic averages) methods, based on the K2-P model and were performed with MEGA 7.0 (Kumar et al. 2016). Bootstrap values were obtained after conducting 1000 replications for sequence divergence and phylogenetic relationships. Bootstrap support > 70% and taxonomically relevant splits, were indicated above branches of the phylogenic tree.

Diglyphus wani Liu, Zhu & Yefremova, 2018
Type material. The type specimens of arrhenotokous D. wani were deposited in the Institute of Plant Protection, Chinese Academy of Agricultural Sciences, Beijing, China.
Arrhenotokous male (Figs 1A, B). Body length 1.0-1.9 mm, forewing length 0.9-1.2 mm. Body light green with a metallic tint; tegulae dark brown, antenna and mandibles brownish, labial and maxillar palpae pale yellow, compound eyes dark red. Legs with dark green and metallic coxae, brownish and metallic trochanters, anterior 3/4 to the middle of all femora dark brown and metallic, posterior pale yellow, all tibiae dark brown with metallic shine, except base and apical 1/5-2/5 part white or pale yellow, hind tibia with anterior surface dark to white-yellow and posterior surface dark, tarsi yellow, except last 4 th tarsomere (dark brown) and 3 rd tarsomere (brownish), wings hyaline.
Head (Figs 1C, F). Head wider than height. Toruli inserted a little above the level with the lower margin of eyes. Malar sulcus present, straight, mouth width 1.6× of malar space.
Metasoma (Figs. 1G and 1H). Petiole short. Gaster 1.8-1.9× as long as broad. Genitalia: digitus with two developed and two reduced spines.  Arrhenotokous female. (Fig. 2A). The arrhenotokous female was similar to the thelytokous female in morphological characteristics (Table 2). We only found a little difference on fore-and hind-wings between arrhenotokous and thelytokous D. wani (Figs. 2A, B). For the arrhenotokous and thelytokous females, the forewing with denser setae overall, the costal cell with 2 ~ 4 rows and 2 rows of setae on dorsal surface,  2D) is denser than the same area of the thelytokous female (Fig. 2F).

ITS1 gene
The ITS1 gene sequences of arrhenotokous and thelytokous strains were 617 bp and 636 ~ 680 bp, respectively. A total of eight variation sites were detected in the thely-  In comparison with the COI gene, the ITS1 gene showed lower haplotype diversity, showing six haplotypes (ITS1-1 ~ ITS1-6) when gaps were not considered. Of ITS1 gene haplotypes, only one haplotype (ITS1-1) was found in the arrhenotokous strain; however, the thelytokous strain had five haplotypes (ITS1-2 ~ ITS1-6). The haplotype sequence of D. wani, D. isaea and D. crassinervis were uploaded to GenBank (accession number: MW393894, MW393901). The mean sequence divergence was 0.010 between two strains and 0.241 ~ 0.265 between related Diglyphus species (Table 3). Similar to the COI analysis, D. wani species formed two major branches, which were thelytokous and arrhenotokous strains, respectively, separated from D. isaea and D. crassinervis (Fig. 4).

ITS2 gene
The ITS2 sequence length of arrhenotokous and thelytokous strains was 389 bp and 388 bp, respectively. Sequence analysis showed three variation sites and no parsimony informative sites when analysing sequences of two strains integrally. The identities of the ITS2 sequences of arrhenotokous species were 87% with D. begini (MH818358.1) and 77% with D. isaea (MH818359.1) in GenBank.
A total of five haplotypes (ITS2-1 ~ ITS2-5) was found when gaps were not considered. Amongst them, there were two haplotypes (ITS2-1 ~ ITS2-2) of the arrhenotokous strain and three haplotypes (ITS2-3 ~ ITS2-5) of the thelytokous strain. The haplotype sequence of D. wani, D. isaea and D. crassinervis were uploaded to GenBank (accession numbers: MW394012, MW394018). The mean sequence divergence was 0.007 between two strains and 0.064 ~ 0.107 between interspecies variation (Table 3). The phylogenetic relationship of the ITS2 region is shown in Fig. 5. The two strains of D. wani form two branches including arrhenotokous and thelytokous strains, respectively, which grouped with D. crassinervis.
Two haplotypes were found within two strains. The haplotypes sequences of D. wani, D. isaea and D. crassinervis were uploaded to GenBank (accession numbers: MW393685, MW393688). Nevertheless, two strains shared a common haplotype. Haplotype 28S-1 was across all arrhenotokous and partial thelytokous individuals and haplotype 28S-2 was included in the other thelytokous individuals. The phylogenetic analysis showed haplotype 28S-1 and D. crassinervis formed one branch due to the same sequences, then clustered with 28S-2 and D. isaea (Fig. 6).

Discussion
In many insect orders, both arrhenotokous and thelytokous strains can be commonly found, such as Hemiptera and Psocodea (Bøcher and Nachman 2011;Yang et al. 2015;van der Kooi et al. 2017). In Hymenopteran parasitoids, species with arrhenotoky and thelytoky are not rare (Schneider et al. 2002;Adahi-Hagimori et al. 2011;Gebiola et al. 2012). However, systematic taxonomical studies on different strains of the conspecific parasitoids are relatively few. Our results indicated that D. wani confirmed both arrhenotokous and thelytokous reproduction modes existed in this species. Besides, the current study is the first directly targeting the morphological and molecular identification of arrhenotokous and thelytokous strains of D. wani.
In general, arrhenotokous and thelytokous strains of Hymenopteran parasitoids are similar in morphology. They may differ in body colour, body length, eyes, wing size and shape, spermathecae and ovaries occasionally (Reineke et al. 2004;Reumer et al. 2013;Petrović et al. 2015;Gebiola et al. 2017). The important distinguishing features we found in the fore-and hind-wings provided an enormous convenience for quickly distinguishing two strains of D. wani. These features were mainly on the density of setae in the costal cell and basal cell. At the same time, based on COI gene, ITS1 gene and ITS2 gene, the sequences divergence between D. wani and related Diglyphus species was far greater than inter-strains divergence. Phylogenetic analysis results showed that the COI gene, ITS1 gene and ITS2 gene can distinguish two strains of D. wani according to the cluster of phylogenetic trees. The COI gene was the best maker to distinguish the two strains of D. wani due to a greater sequence divergence, followed by the ITS gene and the 28S gene cannot distinguish them, because the sequence conservation of the ITS gene and 28S gene was significantly higher than that of the COI gene. Thus, the COI gene can be used as a more effective marker to judge different strains of D. wani, as well as strains of Tetrastichus coeruleus (Hymenoptera: Eulophidae) (Reumer et al. 2013).
Although Hebert et al. (2003) analysed that COI-based sequences divergences amongst the 13320 species and argued 2% gene divergence possessing at least 400 bp of COI sequence was employed as a threshold for species diagnosis, it is controversial (Mallet and Willmot 2003), especially for different strains of a species. Besides, the length of the sequence will affect the delimitation of this threshold (Yang et al. 2017). Yang et al. (2017) reported the COI gene divergence of two strains of N. formosa were 2.3% and 3.9% when using a primer combination (COI1 and COI2) to amply the 520 bp region and another primer combination (LCO1490 and HCO2198) to amply the 710 bp region, respectively. The COI gene sequence divergence between two strains of T. coeruleus was 3.3% ~ 3.7% according to 991 bp of the sequence (Reumer et al. 2013). In this study, the gene divergence between two strains of D. wani was more than 2%, based on the 744 bp of COI sequence. Therefore, the threshold of 2% COI gene divergence is not available for species delimitation in some situations (Murata et al. 2009;Reumer et al. 2013;Yang et al. 2017;Fujie et al. 2019). Furthermore, some species obtaining two strains may have become a genetically-distinct complex or cryptic species on account of a high level of genetic divergence. Cryptic species are at least superficially morphologically indistinguishable, but have distinct genetic structures (Bickford et al. 2007). Based on the COI gene, the sequence divergence between two strains of N. formosa from China was 2.3%, amongst which the thelytokous strain had a closer genetic relationship with thelytokous N. formosa from Japan (Yang et al. 2017). However, the sequence divergence between thelytokous and arrhenotokous strains of N. formosa in Japan is 8.6% (Adachi-Hagimori et al. 2011). Molecular analyses suggested that N. formosa could be a complex of at least two cryptic species, the first one including the thelytokous strain from Japan and two strains of N. formosa from China, the second one from Japan which was arrhenotoky (Yang et al. 2017, unpublished data).
In general, a crossing experiment was carried out to verify whether there were reproductive barriers between the two strains of a parasitoids (Arakaki et al. 2000;Kraaijeveld et al. 2009;Reumer et al. 2013). Thelytokous Leptopilina clavipes (Hymenoptera: Figitidae) was infected with Wolbachia and males were produced by antibiotic treatments (Kraaijeveld et al. 2009). The discoveries were that arrhenotokous males and males derived from thelytokous strains can mate with thelytokous and arrhenotokous females (Kraaijeveld et al. 2009). In contrast, in the parasitoid T. coeruleus whose thelytoky is the result of infection with Wolbachia, although thelytokous females were attractive to arrhenotokous males, thelytokous females were unreceptive to males (Reumer et al. 2014). For thelytokous D. wani, we did not detect thelytoky-inducing endosymbionts reported previously; moreover, high temperature or antibiotic treatment for five generations did not reverse the thelytokous reproductive pattern to produce males (unpublished data). We also conducted laboratory crossing between strictly thelytokous females and arrhenotokous males of D. wani; however, no male progeny was produced (unpublished data).
Previous studies demonstrated thelytokous D. wani had high fecundity and three types of host-killing behaviour (Ye et al. 2018). The arrhenotokous strains of D. wani also exhibited strong biocontrol potential and the two strains of D. wani most notably attacked agromyzid leafminers, especially against C. horticola, L. sativae and L. huidobrensis in the field. In the follow-up studies, it is particularly important to compare and evaluate the biological characteristics of the two strains and to clarify control efficiency when releasing one strain alone, releasing two strains together or releasing them with other parasitoids jointly.