Integrative taxonomy and analysis of species richness patterns of nocturnal Darwin wasps of the genus Enicospilus Stephens (Hymenoptera, Ichneumonidae, Ophioninae) in Japan

Abstract The predominantly tropical ophionine genus Enicospilus Stephens, 1835 is one of the largest genera of Darwin wasps (Hymenoptera, Ichneumonidae), with more than 700 extant species worldwide that are usually crepuscular or nocturnal and are parasitoids of Lepidoptera larvae. In the present study, the Japanese species of Enicospilus are revised using an integrative approach (combined morphology and DNA barcoding). On the basis of 3,110 specimens, 47 Enicospilus species are recognised in Japan, eight of which are new species (E. acutus Shimizu, sp. nov., E. kunigamiensis Shimizu, sp. nov., E. limnophilus Shimizu, sp. nov., E. matsumurai Shimizu, sp. nov., E. pseudopuncticulatus Shimizu, sp. nov., E. sharkeyi Shimizu, sp. nov., E. takakuwai Shimizu, sp. nov., and E. unctus Shimizu, sp. nov.), seven are new records from Japan (E. jilinensis Tang, 1990, E. laqueatus (Enderlein, 1921), E. multidens Chiu, 1954, stat. rev., E. puncticulatus Tang, 1990, E. stenophleps Cushman, 1937, E. vestigator (Smith, 1858), and E. zeugos Chiu, 1954, stat. rev.), 32 had already been recorded in Japan; three (E. biharensis Townes, Townes & Gupta, 1961, E. flavicaput (Morley, 1912), and E. merdarius (Gravenhorst, 1829)) have been erroneously recorded from Japan based on misidentifications, and four names that were previously on the Japanese list are deleted through synonymy. The following taxonomic changes are proposed: E. vacuus Gauld & Mitchell, 1981, syn. nov. (= E. formosensis (Uchida, 1928)); E. multidensstat. rev.; E. striatus Cameron, 1899, syn. nov. = E. lineolatus (Roman, 1913), syn. nov. = E. uniformis Chiu, 1954, syn. nov. = E. flatus Chiu, 1954, syn. nov. = E. gussakovskii Viktorov, 1957, syn. nov. = E. striolatus Townes, Townes & Gupta, 1961, syn. nov. = E. unicornis Rao & Nikam, 1969, syn. nov. = E. unicornis Rao & Nikam, 1970, syn. nov. (= E. pungens (Smith, 1874)); E. iracundus Chiu, 1954, syn. nov. (= E. sakaguchii (Matsumura & Uchida, 1926)); E. sigmatoides Chiu, 1954, syn. nov. (= E. shikokuensis (Uchida, 1928)); E. yamanakai (Uchida, 1930), syn. nov. (= E. shinkanus (Uchida, 1928)); E. ranunculus Chiu, 1954, syn. nov. (= E. yezoensis (Uchida, 1928)); and E. zeugosstat. rev. = E. henrytownesi Chao & Tang, 1991, syn. nov. In addition, the following new regional and country records are also provided: E. flavocephalus (Kirby, 1900), E. puncticulatus, and E. vestigator from the Eastern Palaearctic region, E. laqueatus from the Eastern Palaearctic and Oceanic regions, and E. maruyamanus (Uchida, 1928) from the Oriental region; E. abdominalis (Szépligeti, 1906) from Nepal, E. flavocephalus from Laos, E. formosensis from Laos and Malaysia, E. insinuator (Smith, 1860) from Taiwan, E. maruyamanus from India and Philippines, E. nigronotatus Cameron, 1903, E. riukiuensis (Matsumura & Uchida, 1926), and E. sakaguchii from Indonesia, E. pungens from 14 countries (Australia, Bhutan, Brunei, Indonesia, Laos, Malaysia, Nepal, New Caledonia, Papua New Guinea, Philippines, Solomon Islands, Sri Lanka, Tajikistan, and Taiwan), and E. yezoensis from South Korea. An identification key to all Japanese species of Enicospilus is proposed. Although 47 species are recognised in the present study, approximately 55 species could potentially be found in Japan based on ACE and Chao 1 estimators. The latitudinal diversity gradient of Enicospilus species richness is also tested in the Japanese archipelago based on the constructed robust taxonomic framework and extensive samples. Enicospilus species richness significantly increases towards the south, contrary to the ‘anomalous’ pattern of some other ichneumonid subfamilies.


Darwin wasps and Ophioninae
Darwin wasps, the family Ichneumonidae, are one of the most species-rich branches of the tree of life (Klopfstein et al. 2019), consisting of 1,601 genera and more than 25,000 valid species worldwide (Yu et al. 2016), with an estimated 60,000-100,000 species (Townes 1969;Gauld et al. 2002), but our knowledge currently lags far behind their true enormous diversity (Klopfstein et al. 2019). Darwin wasps are parasitoids of other holometabolous insects or spiders (very occasionally other arthropods or phytophagous) so they play an important role in terrestrial ecosystems as regulators of host insect populations (e.g., Townes 1969;Gauld 1991;Wahl 1993;Quicke 2015;Broad et al. 2018).

The genus Enicospilus
The cosmopolitan genus Enicospilus Stephens is the largest genus within the Ophioninae, with more than 700 valid species (e.g., Broad and Shaw 2016;Yu et al. 2016;Shimizu 2017;Gadallah et al. 2017;Johansson 2018;Shimizu 2020) and an estimate of more than 1,000 species (Townes 1971), suggesting that this is the most speciesrich genus in not only the subfamily but also Ichneumonidae as a whole. Enicospilus is a predominantly tropical genus, with more than 75% of the species occurring in the tropics (Gauld and Mitchell 1981).
As well as almost all other genera of Ophioninae, tropical species of Enicospilus are taxonomically relatively well known thanks to Ian Gauld's comprehensive and groundbreaking regional revisions (e.g., Mitchell 1978, 1981;Gauld 1988). However, fewer papers have been published for the temperate species, most of them focusing on the Western Palaearctic species (e.g., Viktorov 1957;Broad and Shaw 2016;Johansson 2018) and a few on the Eastern Palaearctic species (e.g., Uchida 1928;Chiu 1954;Tang 1990). Therefore, the true species diversity of the temperate fauna is largely unknown. Although a total of 39 species of Enicospilus have been recognised in Japan (Shimizu 2017; Table 1), the Enicospilus fauna of Japan and the far east of Asia has been particularly poorly known, with many taxonomic problems persisting. No identification keys to the Japanese species of Enicospilus have been published after Uchida (1928), while the keys provided by Chiu (1954) and Gauld and Mitchell (1981) include few Japanese species. Hence, a comprehensive study of the Japanese fauna is much needed to understand their true species diversity in the temperate region.

Integrative taxonomy
Species delimitation and taxonomic revisions of poorly known, hyperdiverse groups, such as Darwin wasps, based on traditional morphology-based taxonomy is challenging, but has recently been rapidly improved by advancing integrative approaches that combine multiple perspectives (population genetics, morphometrics, behaviour, host, chemical composition, etc.). A combined morphological and DNA barcoding (partial sequencing of a mitochondrial protein-coding gene, cytochrome c oxidase 1, CO1) approach is the most straightforward method and has been used by many authors for various taxa (e.g., Gibbs 2009;Fernandez-Triana 2010;Schwarzfeld and Sperling 2014;Pentinsaari et al. 2019). For the latter approach, the appropriate sequence divergence distance to delineate species is still open to debate and will differ among taxa and authors, but 2-5% (especially 2%) have been frequently used (e.g., Hebert et al. 2004;Lin et Table 1. Summary of taxonomic histories of the Japanese species of Enicospilus. Valid species names are in bold. Total species numbers were calculated as follow: (a) previous 'total species number ', minus (b) numbers of 'deleted species or names', plus (c) number of added species (i.e., 'new species or names' plus 'new records'). *Enicospilus combustus and E. ramidulus have been sometimes treated as a single species (e.g., Viktorov 1957;Townes et al. 1965;Gauld and Mitchell 1981) but we don't agree with this and follow the recent papers (e.g., Broad and Shaw 2016).

Latitudinal diversity gradient in species richness
There has been much research into patterns of Darwin wasp species richness across latitudinal gradients. This has been summarised fairly recently by Santos and Quicke (2011), Jones et al. (2012) and Veijalainen et al. (2012Veijalainen et al. ( , 2013. Observations that there were apparently small numbers of individuals and species of Ichneumonidae in the tropics led to the idea that this represented one of a few insect examples of an 'anomalous' species richness gradient (e.g., Owen and Owen 1974;Janzen 1981). Various potential mechanisms have been proposed to explain the relative lack of ichneumonid species in the tropics, e.g., the resource fragmentation hypothesis by Janzen and Pond (1975), the predation hypothesis by Rathcke and Price (1976), and the "nasty" host hypothesis by Gauld et al. (1992). However, the increase in data from more recent large scale collecting has shown that the pattern is more complicated, with some ichneumonid subfamilies being potentially more species-rich at lower latitudes, some less species-rich, but that robust data are still lacking to accurately describe patterns of species richness, let alone propose mechanisms to explain the patterns (Santos and Quicke 2011;Veijalainen et al. 2012). Nevertheless, there are robust findings, such as that parasitoids of insect groups that are more diverse at higher latitudes are similarly more species-rich at higher latitudes, such as sawfly parasitoids of the subfamilies Ctenopelmatinae and Tryphoninae, and Diplazontinae, parasitoids of aphidophagous Syrphidae (Diptera) (Quicke 2015).
Most ophionines are typical nocturnal koinobiont parasitoids with their centre of species diversity in the (sub-)tropics (e.g., Gauld 1985Gauld , 1988Gauld and Mitchell 1981), with a few exceptions, e.g., the nocturnal genus Ophion is most abundant in cooler temperate regions (e.g., Schwarzfeld and Sperling 2014;Schwarzfeld et al. 2016), as is a Southern equivalent of Ophion, Alophophion Cushman (Alvarado 2014). Gauld (1987) suggested that their nocturnal habit is one factor that has adapted ophionines to tropical rainforest, where they would be exposed to high predation pressure in the daytime, based on the predation hypothesis. Moreover, Quicke (2015) pointed out that daytime temperatures are too hot for much active host searching in the lowland tropics so that nocturnal habits are more suitable than diurnal there. However, this research field is still under discussion (e.g., Gauld and Mitchell 1981;Gauld 1995;Jones et al. 2012) and needs additional data.

Aims of the present study
The Japanese archipelago is located in a long line between ca. 20-45°N, approximately 3,000 km from south to north, ranging from the southern subtropical to northern and high elevational subarctic zones, containing a high diversity of ecological habitats. Biogeographically, it also includes the Oceanic, Oriental, and Palaearctic regions and is a melting pot of species originating from these regions, so in some ways one of the most interesting biodiversity hotspots (e.g., Mittermeier et al. 2004). However, taxo-nomic knowledge of the speciose genus Enicospilus in Japan has been complicated due to the lack of revisionary studies based on comprehensive sampling and the difficulties of the traditional morphology-based taxonomy. For these reasons, to reveal the species diversity of Enicospilus in Japan, we revise the Japanese fauna using integrative approaches (i.e., combined morphological and DNA barcoding approach) and estimate their species richness based on the large number of specimens examined. In addition, as a genus that seems to be more species-rich at lower latitudes on a global scale, it is of interest to know whether Enicospilus diversity follows this trend at a local scale too, within the Japanese archipelago. Hence, we test a latitudinal diversity pattern in this group of nocturnal koinobiont parasitoids in Japan based on the constructed robust taxonomic framework.

Specimens examined
The specimens examined were studied in or borrowed from insect collections, or newly collected for the present study, mainly using High Intensity Discharge (HID) light traps (Fig. 1) by the first author. A total of 3,110 specimens of Enicospilus, of which 1,863 are from Japan and 1,247 from other countries, were examined. Type specimens examined are listed in the main text, but the data for non-types are listed in Suppl. material 1: Table S1. The first author also examined, however, many more specimens (more than 20,000 specimens) from all over the world to develop an improved perspective on species criteria and range of variation within Enicospilus.

Terms, indices, and abbreviations
The morphological terms mainly follow Broad et al. (2018). Terms for surface microsculpture follow Eady (1968) and Gauld and Mitchell (1981). The terms, abbreviations, and indices for head, wings and metasoma follow mainly Shimizu and Lima (2018), Shimizu (2020) and partly Broad et al. (2018). The indices, terms, and abbreviations     Table 3. Abbreviations for repositories consulted (not all are referred to in the main text and some are only in Suppl. material 1: Table S1).
The abbreviations for specimen repositories used in the present paper (some only in Suppl. material 1: Table S1) are listed in Table 3. In addition, the following abbreviations for collections are used: JMC J. Minamikawa collection at NIAES KUSIG K. Kusigemati

Literature records
There are many published distribution and host records that cannot be verified as we cannot access all voucher specimens or host remains underpinning these literature records; Enicospilus species have frequently been misidentified, as have their hosts, and there are various reasons why potential hosts and parasitoids are mis-associated (see Shaw 1994). There is no point repeating dubious distribution and host records, so we have just broadly summarised distributions and host ranges. However, Japanese and some reliable recent extralimital host records (e.g., Broad and Shaw 2016) are emphasised.

Order of prefectures
We used the following order of Japanese Prefectures in the distribution of Japan:

Taxon sampling
To test our assessment of taxonomy based on morphology, we employed a DNA barcoding approach. A total of 168 sequences of CO1 from 41 of 47 Japanese species of Enicospilus (including the species described in the present paper) and other Enicospilus species from the Eastern Palaearctic, Neotropical, and Oriental regions were sampled: 125 of those were newly sequenced and deposited in DNA Data Bank of Japan (DDBJ), and 43 were obtained from the Barcode of Life database (BOLD) and Gen-Bank. We selected seven ophionine genera: Afrophion Gauld, Dicamptus Szépligeti, Hellwigiella, Leptophion Cameron, Ophion, Rhynchophion Enderlein, and Thyreodon Brullé as outgroups. The species, identifiers for the specimens, collection localities, sample codes, and accession numbers for all terminal taxa used in the analyses are listed in Suppl. material 2: Table S2.

DNA extraction, amplification, and sequencing
Most specimens for DNA analysis were dried specimens borrowed from collections. Some specimens were newly collected for the present study; these were stored in 80.0-99.9% ethanol and, after DNA extraction, mounted as dried specimens, currently deposited in the respective insect collections. DNA was extracted from a single right mid leg or both right mid and hind legs using the DNeasy Blood and Tissue Kit (Qiagen, Düsseldorf, Germany). Partial sequences of CO1 were amplified using primers designed by Folmer et al. (1994): LCO1490 (5' -GGT CAA CAA ATC ATA AAG ATA TTG G -3') and  HCO2198 (5' -TAA ACT TCA GGG TGA CCA AAA AAT CA -3'). Polymerase chain reactions (PCR) were conducted using the KOD FX NEO kit (Toyobo, Ōsaka, Japan). PCR conditions were 2 min at 94 °C as an initial denaturation, and 35 cycles of 10 s at 98 °C of denaturation, 30 s at 48 °C of annealing, and 30 s at 68 °C of extension, then a final extension at 72 °C for 10 min. PCR products were purified using Illustra GFX kit (GE Healthcare Life Sciences, Marlborough, USA). The purified PCR products were amplified with the same primers using the BigDyeTM Terminator v.3.1 Cycle Sequencing kit (Applied Biosystems, Waltham, USA). In order to save cost, cycle sequencing reactions were carried out in 10.0 μl total volume consisting of 0.5 μl Readdy Reaction Mix, 2.0 μl 5× Sequencing Buffer, 1.2 μl each primer (10.0 pmol), 5.0 μl PCR products (10.0 ng / 100 bp), and 1.3 μl Deionized water. Cycling conditions were 25 cycles of 10 s at 96 °C, 5 s at 50 °C, and 4 min at 60 °C. Products were purified using the 3.0 M sodium acetate, 95% ethanol, 70% ethanol, and Hi-Di formamide. Cycle sequencing products were run on an ABI Prism 3100 Genetic Analyzer (Applied Biosystems, Waltham, USA), and the forward and reverse sequences were assembled using the DNA Dynamo Sequence Analyse Software (Blue Tractor Software, North Wales, UK). Some sequences were incomplete; however, they were included in analyses with the gaps coded as missing data.

Multiple sequence alignments
We conducted multiple sequence alignments in MAFFT v.7.409 (Katoh and Standley 2013), using default parameters: the final dataset is 626 bp in length, without indels.

Analyses
Analyses were performed with Bayesian Inference (BI) and maximum likelihood (ML) approaches. Each codon position within the CO1 fragment was treated as a different data block, and the best-fit substitution model was determined using PartitionFinder v.2.1.1 (Lanfear et al. 2017) with the greedy search algorithm under the corrected Akaike information criterion (AICc): the selected model was the GTR+I+Γ model for all positions. The BI analyses were conducted using MrBayes v.3.2.2 (Ronquist et al. 2012). We ran two independent runs of a Bayesian Markov chain Monte Carlo (MCMC) analysis of eight chains each, heating 0.1, random starting trees, and trees sampled every 1,000th generation for 10,000,000 generations. We considered that the two MCMC runs had converged if the average standard deviation of split frequencies was below 0.01 (Ronquist and Huelsenbeck 2003). Moreover, we checked for chain stationarity in Tracer v.1.6 (Rambaut and Drummond 2007). Then, we discarded half of the generations as a conservative burn-in, obtained estimates for the harmonic means of the likelihood scores from the remaining half of the generations using the sump command, and conducted a final check of the convergence of the runs by the value of a potential scale reduction factor (PSRF): if the runs were convergent enough, PSRF was less than 5% divergent from 1.0. Finally, a majority-rule consensus tree with the Bayesian inference posterior probabilities was obtained using the sumt command in MrBayes. The ML analysis was conducted in RAxML v.8.2.10 (Stamatakis 2014) with 1,000 bootstrap replications. We mapped the bootstrap percentages to each node of the reconstructed best-fit tree by the pgsumtree command in Phylogears v.2.2.0 (Tanabe 2008). The trees were checked and edited in FigTree v.1.4.3 (Rambaut 2006 and Adobe Illustrator. The p-distances were calculated using MEGA v.10.0.5 (Kumar et al. 2018).

Species richness pattern analysis
The latitudinal diversity gradient (LDG) of species richness in the Japanese archipelago was analysed based on the constructed robust taxonomic framework and extensive samples. We divided the Japanese archipelago into six latitudinal zones of equal intervals (Table 4) and analysed the LDG based on the species richness in each zone to separate the Oriental and Palaearctic regions and reduce the effect of sampling biases along the Japanese archipelago. Species richness was usually counted in the prefectural capitals (Table 4), only using data from the specimens examined in the present study (Suppl. material 1: Table S1), because literature data are sometimes unreliable. Incomplete label data were not included in the analysis. To exclude regional biases in sample number, saturation species richness in each zone was estimated by extrapolation methods based on Chao1 richness estimator in EstimateS v.9.1.0 (Colwell 2013). We used Spearman's rank correlation to test correlations between latitudinal zones and species richness in each zone using R v.3.6.3 (R Core Development Team 2020).
To understand the regional pattern of sampling biases, numbers of four categories (specimens, collection events, collector, and species) were counted for each area. Each pattern is shown in the heat maps.
To infer the total species richness of the Japanese Enicospilus, the individual-based rarefaction curves were estimated based on ACE and Chao 1 richness estimators using EstimateS, with 100 runs of randomizations and the classic formula for Chao 1.

Integrative taxonomy
A total of 47 morphospecies were recognised in Japan: 32 of which were previously known from the Japanese fauna, eight were new to science, seven were new to Japan, and seven were excluded from the Japanese fauna (Table 5). All Japanese species and nomenclatural changes are summarised in Suppl. material 3.
For most Japanese Enicospilus species, the morphological and DNA barcoding results were complementary and consistent, and we could easily separate species (Fig. 6).
However, the results from each approach were inconsistent for a few species: three Japanese species, E. xanthocephalus Cameron, 1905, E. stenophleps Cushman, 1937, and E. puncticulatus Tang, 1990, and the non-Japanese E. flavicaput (Morley, 1912). In our DNA barcoding analysis, maximum p-distances within species were less than 2% in virtually all species, but 6% in E. stenophleps and 7% in E. flavicaput and E. xanthocephalus. These species were morphologically stable, hence, we were following the  traditional morphology-based taxonomy and not splitting them in the present paper. On the other hand, although E. puncticulatus exhibits a wide range of variation in morphology, especially of the shape of fore wing sclerites, maximum p-distances within this species were less than 1%. In the present paper, we treated them as a single species, Figure 7. Intraspecific variation of the fore wing sclerite development in E. shikokuensis (Uchida, 1928) A the proximal and distal sclerites separated and the central sclerite weak (SEN42) B the proximal and distal sclerites confluent and the central sclerite strong (SEN41). because the fore wing variation is likely to be continuous (see the species account for E. puncticulatus for more details).
Generic diagnosis. Enicospilus species are moderately to very large insects, fore wing length usually 10.0-30.0 mm, with ophionoid facies. Easily distinguishable from other Ophioninae by the following characters: fore wing discosubmarginal cell with extensive glabrous area (fenestra), often with one or more sclerites (e.g., Fig. 4); mandibles narrow, slightly to strongly twisted (e.g., Fig. 2); inner surface of fore tibial spur lacking membranous flange. Enicospilus species can be confused with the genus Dicamptus but easily distinguished by the weakly to strongly tapered and twisted mandible (mandible very weakly tapered and never strongly twisted in Dicamptus). A key to the Japanese genera of Ophioninae has also been provided by Shimizu and Watanabe (2017).
Head. Clypeus flat to strongly convex in profile, ventral margin acute, blunt, or impressed. Mandible weakly to strongly tapered and twisted, usually moderately long, outer surface with or without diagonal setose groove or line of punctures, and bidentate apically. Frons, vertex and gena shiny and smooth. Ocelli usually very large and posterior ocellus often close to or touching eye. Occipital carina usually complete, ventrally reaching oral carina or not. Antennae usually longer than fore wing, with usually more than 50 flagellomeres.
Mesosoma entirely weakly to moderately shiny with setae. Pronotum finely punctate or diagonally wrinkled and not specialised. Mesoscutum shiny and punctate to smooth with setae, evenly rounded in profile, and notauli usually absent. Scutellum moderately convex and usually with lateral longitudinal carinae. Epicnemium usually densely punctate with setae. Epicnemial carina present, straight to curved, inclined to curved to anterior margin of mesopleuron. Posterior transverse carina of mesosternum usually complete. Propodeum evenly rounded or declivous in profile; anterior transverse carina usually complete; anterior area longitudinally striate; spiracular area usually smooth; posterior area reticulate, wrinkled, striate, or rugose; and posterior transverse carina usually absent.
Wings. Fore wing pterostigma fairly slender; vein 1m-cu&M evenly curved, angulate or sinuate, usually without a ramulus; vein 2r&RS usually more or less widened and sinuate; discosubmarginal cell usually with bare fenestra, often with one or more sclerotised sclerites. Hind wing vein RS usually straight and rarely weakly curved; vein RA usually with 4-12 uniform hamuli.
Legs. Inner mesal surface of fore tibial spur without membranous flange. Outer distal margin of mid and hind trochantelli usually simple without decurved tooth. Hind tarsal claw moderately to strongly curved and usually simply pectinate.
Metasoma very slender. Spiracle of T1 far behind middle. Thyridium well developed. Ovipositor straight and almost always not longer than posterior depth of metasoma.
Colour. General body colour usually entirely testaceous, with posterior metasomal segments sometimes darker, but body sometimes entirely dark brown to black or pale. Wings usually entirely hyaline or weakly infuscate, but wings with strong infumate area in a few species; fenestra always hyaline; sclerites weakly to strongly pigmented amber.
Species criteria. We summarise the especially important diagnostic characters to identify Enicospilus species below.
Head (Fig. 2). The head provides many good characters to define species, as many previous authors have indicated (e.g., Mitchell 1978, 1981;Gauld 1988;Schwarzfeld and Sperling 2014;Johansson and Cederberg 2019). Among them, the width of lower face as well as of clypeus, colour of interocellar area (or stemmaticum), shape of clypeus, and mandibular characters are especially useful and easy to use.
Width of the lower face is usually stable within a species group and/or species, even if the species is widespread, and sometimes provides enough gaps between species, although a few species, such as E. capensis, exhibit considerable variation.
Although body colour can be very variable within species, the colour of the interocellar area is usually stable at the species level and a good diagnostic character.
The shape of the clypeus is also very useful. For instance, the nasute clypeus is one of the most critical diagnostic characters of E. riukiuensis and related species (Fig. 40D), and the flat and projecting clypeus of E. sakaguchii is distinctive (Fig.  41D). The shape of the ventral margin, i.e., acute, blunt, or impressed, is also a very useful diagnostic character.
Features of mandibles are some of the most important diagnostic characters of Enicospilus species. First, the outer mandibular surface sculpture, especially presence or absence of a diagonal setose groove or line of punctures between the dorsoproximal corner and base of the apical teeth, is important. For example, the outer mandible surface of E. ramidulus has a diagonal setose groove (Fig. 39B, D), but of E. pungens is smooth (Fig. 38B, D). Second, the torsion of the mandible is a useful character, although it is rather difficult to measure. For instance, the strongly twisted mandible of E. acutus sp. nov. is one of the most important diagnostic characters for this species (Fig. 11B, D). Finally, length and shape also provide good characters. For instance, the mandible of E. shikokuensis is very long, slender, strongly tapered proximally, and subparallel-sided distally (Fig. 44B, D), but of E. sakaguchii is very short, stout, and evenly tapered (Fig. 41B, D).
Some mandibular diagnostic characters at the species level, such as degree of torsion and length of teeth are possibly adaptive characters and have been considered to be related to modes of emergence from host insects; hence, these characters are usually easily modified and not phylogenetically restricted, so it is indeed useful for species level taxonomy. Mesosoma (Fig. 3). Mesosomal characters are also very informative. Surface microsculuptures of meso-and metapleuron are rather stable within species and also show large gaps between species and/or species groups. For example, the mesopleuron is coarsely longitudinally striate and metapleuron coarsely rugose in E. nigristigma (Fig.  31E), but the meso-and metapleuron are evenly moderately punctate and strongly shiny in E. unctus sp. nov. (Fig. 50E). Propodeum characters are useful as well. The posterior transverse carina of the propodeum of E. signativentris is unique in the Japanese species (Fig. 46E), and globally few Enicospilus species possess this carina. In other cases, the posterior area can be entirely densely punctate to finely reticulate in E. limnophilus sp. nov. (Fig. 25E), but coarsely concentrically striate in E. insinuator (Fig. 19E), providing an easy means to separate them. In this way, the propodeum is useful for definition of the species or species group when combined with other characters. Characters of the scutellum are also sometimes good for species recognition. For example, the quadrate scutellum of E. formosensis contributes to its identification, with the scutellum more or less trapezoidal or triangular in most Enicospilus species. Additionally, the length of the lateral longitudinal carinae of the scutellum is almost always stable within species and useful, with very rare exceptions, such as in E. limnophilus sp. nov., that exhibits a very wide range of variation (length of the carinae varies from 0.1-1.0× scutellum length).
Wings (Fig. 4). Wing characters have probably been regarded as the most important diagnostic characters in nocturnal Ophioninae by many previous authors (e.g., Gauld 1977;Gauld and Mitchell 1981;Shimizu 2020). These are much easier to use than other characters and can easily be measured.
First, the number, shape, and position of sclerites of the fore wing fenestra is usually very useful in Enicospilus. The number of sclerites varies in Enicospilus species from zero to four, but is nearly always stable within a species. The shape and position of sclerites are also very diverse, and some previous research has suggested that some species exhibit a wide range of intraspecific variation (e.g., Gauld and Mitchell 1981). However, it is stable in many cases and, if a species shows a wide range of intraspecific variation, it is likely that cryptic species are involved. Second, the shape and setosity of the fore wing fenestra is sometimes a very useful character. For example, among Japanese species, the fenestra of E. nigribasalis and E. stenophleps is long and its anterodistal corner interstitial to fore wing vein RS (Figs 30F, 47F), but in other Japanese species the fenestra is shorter and its anterodistal corner clearly antefurcal to RS (e.g.,Figs 25F,36F).
Finally, the length and shape of wing veins also offer very good diagnostic characters. For example, the shape of fore wing veins 1m-cu&M and 2r&RS, the position of fore wing vein 1cu-a, and values of indices (e.g., AI, CI, ICI) are useful to distinguish species.
Legs. Legs do not seem to provide many useful characters, but some characters, such as the density of spines on the outer surface of the fore tibia, and pectination of the hind tarsal claw, are useful in species definition. For example, E. maruyamanus and E. pudibundae are very difficult to distinguish from each other, but the hind tarsal claw of the former is entirely uniformly pectinate, and the latter is not pectinate proximally.
Metasoma (Fig. 5). The shape of the first metasomal segment (e.g., sinuous or straight in profile, slender or stout) is useful (Broad and Shaw 2016;Johansson and Cederberg 2019).
Body size. Measuring body length is rather difficult due to the wide range of contraction or expansion of metasomal segments after death, hence fore wing length is a more useful character. However, body size shows a very wide range of variation in many species, such as E. pseudoconspersae, although it is stable in a few species (e.g., E. concentralis and E. nigronotatus). Therefore, this character is occasionally useful for species definition.
Colour. As mentioned above, this character can easily change intraspecifically. For instance, E. nigropectus and E. signativentris show a considerable range of colour variation (body entirely blackish to entirely testaceous). However, this character is stable within some species (e.g., E. acutus sp. nov. and E. nigribasalis,as in Figs 11,30). Hence, this is also a useful critical character, but we need to be careful when relying on colour pattern.
Key to the Japanese species of Enicospilus 1 Fore wing fenestra lacking both sclerites and quadra (Fig. 16F) Mandible very strongly twisted by ca. 85°, therefore outer margin forming acute median longitudinal ridge between centroproximal part of mandible and base of apical teeth (this ridge is the ventral margin of the mandible) (Fig. 11B, D).

Enicospilus abdominalis (Szépligeti, 1906)
Newly recorded from Nepal. JAPAN: [Ryûkyûs] Okinawa (Shimizu and Maeto 2016;present study). This species is abundant in Taiwan and in other mountainous areas of the Oriental region, but only one Japanese specimen has been collected from Okinawa-hontô of the Ryûkyûs. This single Japanese individual could have been a wanderer from Taiwan or other southern areas.
Bionomics. Although this is one of the most common species in the Oriental region, there are no host records.
Differential diagnosis. This species is morphologically relatively close to E. laqueatus and E. yonezawanus. However, E. aciculatus is rather easily distinguished from all other species of Japanese Enicospilus by the following combination of character states: mandible with diagonal setose groove, and upper mandibular tooth more than 2.0× as long as lower (Fig. 10B, D); central sclerite of fore wing fenestra moderately sized, ill-defined, very weakly pigmented, and positioned in anterodistal corner of fenestra (Fig. 10F); and proximal angle of proximal sclerite of fore wing fenestra ca. 45° (Fig. 10F). This species is morphologically stable but sometimes exhibits colour variation (i.e., usually the metasoma is entirely orange brown, but rarely the posterior segments are black). Enicospilus acutus Shimizu, sp. nov. http://zoobank.org/F6D9DDEC-2D45-48D3-8E4D-E00C67CAA1A7 Figure 11 Etymology. The specific name is derived from the characteristic longitudinal acute ridge of the mandibular outer margin.  that outer margin forming acute median longitudinal ridge between near centro-proximal part of mandible and base of mandibular apical teeth (Fig. 11B, D) (mandible more or less weakly twisted less than 35° so that outer mandibular surface without any ridge or groove in E. maai); fore wing with AI more than 1.0 ( Fig. 11F) (AI less than 0.7 in E. maai); fore wing fenestra with strongly pigmented central sclerite (Fig. 11F) (central sclerite absent in E. maai); and mesopleuron closely striate to punctostriate (Fig. 11E) (mesopleuron coarsely striate in E. maai). This species is uniquely distinctive and easily distinguishable within the Japanese Enicospilus on account of the mandibular morphology and colour pattern.
Colour (Fig. 11). Entirely light to rather dark brown except for apex of mandible, posterior part of mesoscutum, and central part of frons infuscate. Wings moderately infuscate; sclerites pigmented and amber; veins dark reddish brown.

Enicospilus capensis (Thunberg, 1824)
Enicospilus fossatus Chiu, 1954: 63; HT ♀ from Malaysia, TARI, examined; synonymised by Gauld and Mitchell (1981: 385). Enicospilus indica Rao and Grover, 1960: 280; HT ♀ from India, MUC, destroyed (cf. Gauld and Mitchell (1981: 385)), not examined; synonymised by Gauld and Mitchell (1981: 385).  (Sonan 1940;present study). This species has a very wide distribution from South East Asia to South Africa. According to Gauld and Mitchell (1981), this distribution pattern is hardly surprising when considering many of their host moths are also widely distributed throughout the Old World tropics. Enicospilus capensis is frequently encountered as a parasitoid of economically important noctuid moths; however, only a single specimen has been collected in Japan.
Differential diagnosis. This species is usually very easily distinguished from all other Palaearctic Enicospilus species by the black mesosoma, thyridium, and posterior segments of metasoma, as in Fig. 13A. Enicospilus combustus has sometimes been confused with E. multidens stat. rev., E. ramidulus, and E. shikokuensis; moreover, some authors have treated E. combustus and E. ramidulus as a single species (e.g., Viktorov 1957;Townes et al. 1965;Gauld and Mitchell 1981). However, E. combustus is easily separated from E. shikokuensis by the separated proximal and distal sclerites of fore wing fenestra, as in Fig. 13F (proximal and distal sclerites usually obviously confluent in E. shikokuensis, as in Fig.44F), from E. multidens stat. rev. and E. ramidulus by the entirely more or less blackish mesosoma, as in Fig. 13A, E (mesosoma entirely orange-brown in E. multidens stat. rev. and E. ramidulus, as in Figs 29A, E and 39A, E respectively). Moreover, this species is similar to E. sharkeyi sp. nov. in colour pattern (Figs 13, 43), however, E. combustus can be readily distinguished from it by many characters, such as separated proximal and distal sclerites of fore wing fenestra, as in Fig. 13F (proximal and distal sclerites confluent in E. sharkeyi sp. nov., as in Fig. 43F), larger central sclerite of fore wing fenestra, as in Fig. 13F (central sclerite smaller in E. sharkeyi sp. nov., as in Fig. 43F), wider lower face, as in Fig. 13B (narrower in E. sharkeyi sp. nov., as in Fig. 43B), etc. Figure 14 Enicospilus concentralis Cushman, 1937: 305; HT ♀ from Taiwan, DEI, not examined.

Bionomics.
Recorded from Erebidae by Sharma (1985). Differential diagnosis. This species especially resembles E. grandis (Cameron, 1905) and E. plicatus (Brullé, 1846), but is distinguishable by the smaller size, shorter antennae, and more matt and uniformly punctate meso-and metapleuron (Fig. 16A, E). Other than this species, all Japanese Enicospilus species have at least one fore wing sclerite; hence, it is fortunately very easily identifiable.
Remarks. Gauld and Mitchell (1981) had separated E. formosensis and E. vacuus based on differences of value of CI. However, the CI of these 'species' are continuous and no other morphological differences could be recognised. Tang (1990) also suggested that these names represented the same species. Hence, E. vacuus is newly synonymised under E. formosensis in the present paper. Figure 19 Ophion insinuator Smith, 1860: 141; HT ♀ from Moluccas, OUMNH, not examined. Enicospilus zyzzus Chiu, 1954: 23; HT ♀ from China, TARI, examined; synonymised by Gauld and Mitchell (1981: 353).   Chiu (1954) has recorded this species from Japan, but lacking collection locality data. Most Japanese specimens were collected in Yakushima by Malaise traps.

Enicospilus javanus (Szépligeti, 1910)
Remarks. Enicospilus javanus exhibits a very wide range of variation in the shape of fore wing sclerites, body size, and colour, as mentioned by Gauld and Mitchell (1981: 262). We examined 24 female specimens from the same locality and date collected by I. D. Gauld in Brunei, and these specimens indeed show an extremely wide range of morphological variation. However, the variation is more or less continuous. This variation suggests that this species includes some cryptic species and integrative approaches are needed to reveal species boundaries. Regarding the Japanese population, it is very stable, suggesting that there are no cryptic species included.   Newly recorded from Japan. JAPAN: [Kantô-Kôshin] Tôkyô and Chiba. This species is collected only from a big city in Japan.
Bionomics. Unknown. Differential diagnosis. All Japanese specimens had been misidentified as E. yonezawanus, because they lack the central sclerite of fore wing fenestra and have a strongly pigmented triangular proximal sclerite, as in Figs 21F, 54F. However, E. jilinensis can be distinguished from E. yonezawanus by many characters, such as the smooth outer mandibular surface, smaller fore wing fenestra, entirely punctate mesopleuron, and larger value of ICI.
Remarks. Some morphological features (e.g., flattened clypeus, moderately long and slender mandible, and absence of central sclerite of fore wing fenestra) suggest a relation to E. shinkanus, which should be tested when fresh specimens are available.  Distribution. Eastern Palaearctic region (Shimizu 2017;present study Bionomics. Unknown. Differential diagnosis. As in Fig. 22, the characteristic colour pattern easily distinguishes this species from other Enicospilus species. Morphologically, E. kikuchii resembles E. melanocarpus, but is distinguishable by the following combination of character states: mesosoma, T1, T2, and T5-8 black (Fig. 22A, E) (usually most of body yellow-brown with black T5-8 in E. melanocarpus, as in Fig. 28A, E); and metapleuron roughly diagonally punctostrigose (Fig. 22E) (uniformly punctate or finely diagonally punctostriate in E. melanocarpus, as in Fig. 28E).
Legs. Hind leg with coxa in profile 1.8× as long as deep; basitarsus 1.9× as long as second tarsomere; fourth tarsomere 3.1× as long as wide; tarsal claw simply pectinate.
Male. Unknown. Figure 24 Henicospilus laqueatus Enderlein, 1921 Distribution. Afrotropical and Oriental regions (Yu et al. 2016); new to the Eastern Palaearctic (Hachijô-jima, Tôkyô, Kantô-Kôshin, Japan) and Oceanic (Nishi-jima, Tôkyô, Ogasawara, Japan) regions (cf. Suppl. material 1: Table S1); this is a predominantly (sub-) tropical species. This species is widely distributed from the Afrotropical to Oriental regions and Gauld (1982) suggested that it has possibly been introduced from Asia to Africa, although there is no reliable evidence to support or refute his hypotheses.
Bionomics. Unknown. Differential diagnosis. This species resembles E. pseudantennatus, E. vestigator, and E. tripartitus on the shape of fore wing fenestra, sclerites, and venation. However, E. laqueatus is easily distinguishable from them by the outer mandibular surface morphology (i.e., outer mandibular surface with diagonal setose groove between dorsoproximal corner and base of apical teeth in E. laqueatus, but smooth or just densely punctate with stout setae in the other three species, as summarised in Table 6). In addition, this species morphologically resembles E. aciculatus and E. yonezawanus but is distinguished from them by its strongly pigmented central sclerite of fore wing fenestra (Fig. 24F) (central sclerite very weakly pigmented or vestigial in E. aciculatus, as in Fig. 10F, and completely lacking in E. yonezawanus, as in Fig. 54F).

Enicospilus limnophilus Shimizu, sp. nov.
http://zoobank.org/2383AAC6-A428-4933-A7DC-50400B672084 Figure 25 Etymology. This species probably prefers lakes. Hence, the specific name is derived from the Greek limne + philos meaning lake and lover respectively.  Bionomics. All specimens have been collected from marshes or lakes of rather cooler regions, suggesting that it is restricted to hosts that inhabit open, aquatic conditions. However, some factors, such as a progression of plant succession, isolation of habitats, and increasingly dry conditions, have led many wetland insects to become endangered in Japan (e.g., Yoshida et al. 2019).
Differential diagnosis. The distally setose fore wing fenestra is unique to this species within the Asian Enicospilus fauna (Fig. 25F), hence E. limnophilus sp. nov. is morphologically very easily recognisable.
Remarks. This is a fairly morphologically uniform species, although GOI (= 1.9-2.9) and length of lateral longitudinal carinae of scutellum (along anterior 0.1-0.9 of scutellum) exhibit a very wide range of variation within the same population. Figure 26 Henicospilus maruyamanus Uchida, 1928: 220; LCT ♀ from Japan, designated by Townes et al. (1965: 329), SEHU, examined. acter states: uniformly pectinate hind tarsal claw; weakly to moderately sinuate fore wing vein 1m-cu&M (Fig. 26F); and punctostriate meso-and metapleuron (Fig. 26E) (cf. Table 7).  Bionomics. Unknown. Differential diagnosis. This species is similar to E. flavicaput and some of the type series were misidentified as that species, but E. matsumurai sp. nov. is distinguishable from E. flavicaput by the shape and position of the central sclerite of the fore wing fenestra (Fig. 27F), sculpture of the mesosoma (Fig. 27E), and wing venation (Fig. 27F). This species also resembles E. kunigamiensis sp. nov. but is distinguishable by several characters (cf. 'Differential diagnosis' under E. kunigamiensis sp. nov.). In addition, the large size of this species is also a useful diagnostic character.
Bionomics. No host records from Japan. Gauld and Mitchell (1981) and Nikam (1990) report rearings from a disparate range of hosts in the families Erebidae, Lasiocampidae and Noctuidae, which clearly warrants investigation. Differential diagnosis. This species is morphologically most similar to E. sauteri, but is distinguished from it by the uniformly setose marginal cell of fore wing (Fig. 28F) (marginal cell proximally widely glabrous in E. sauteri, as in Fig. 42F), and oval central sclerite of fore wing fenestra (Fig. 28F) (central sclerite linear in E. sauteri, as in Fig.  42F). Enicospilus melanocarpus is also sometimes confused with E. ramidulus but is distinguished from it by the sculpture of the mesosoma (i.e., meso-and metapleuron punctate to punctostriate in E. melanocarpus, as in Fig. 28E, but entirely punctate in E. ramidulus, as in Fig. 39E), shape of the sclerites (i.e., proximal and distal sclerites confluent in E. melanocarpus, as in Fig. 28F, but not confluent in E. ramidulus, as in Fig. 39F), etc.
In Japanese collections, they are sometimes confused with E. ramidulus and E. yezoensis, as both species have a similar colour pattern (i.e., entirely testaceous body with posterior metasomal segments strongly infuscate, as in Figs 28A, 39A, and 53A). However, in Japan E. melanocarpus is restricted to Ryûkyûs and Ogasawara (i.e., the Oceanic and Oriental regions of Japan), with E. ramidulus and E. yezoensis in the Palaearctic area of Japan. We summarise the diagnostic characters in Table 8 for E. melanocarpus, E. ramidulus, E. sauteri and E. yezoensis, all of which have testaceous bodies with the metasoma black posteriorly.  Figure 29 Enicospilus multidens Chiu, 1954: 75; HT ♀ from Japan, TARI, examined; stat. rev.

Enicospilus nigribasalis (Uchida, 1928)
Bionomics. No host records from Japan. Recorded as a parasitoid of Ericeia inangulata (Guenée) (Erebidae) by Chiu et al. (1984) and Chen et al. (2009). Differential diagnosis. This species is very easily distinguishable from any other Enicospilus species on account of the characteristic colour pattern, especially of the metasoma and wings, as in Fig. 30A. Gauld and Mitchell (1981) suggested that E. nigribasalis is closely related to E. ashbyi and E. pallidus; however, this species can easily be distinguished from them by colour pattern, shape of wing veins, and shape of the fore wing fenestra and sclerites. Figure 31 Enicospilus nigristigma Cushman, 1937: 309; HT ♀ from Taiwan, DEI, not examined.
Bionomics. Unknown. Differential diagnosis. This species is one of the largest Japanese ophionines, along with E. nigronotatus and Dicamptus nigropictus (Matsumura, 1912). The habitus and sculpture of this species are very similar to E. stimulator (Smith, 1865), but  (Uchida, 1928) ♀ from Japan A habitus B head, frontal view C head, dorsal view D head, lateral view E mesosoma, lateral view F central part of fore wing.

Enicospilus nigronotatus Cameron, 1903
Distribution. Oriental region (Yu et al. 2016). Newly recorded from Indonesia. JAPAN: [Ryûkyûs] Okinawa (Shimizu and Maeto 2016;present study). Bionomics. Unknown. Differential diagnosis. This species is an extremely large insect, as is E. nigristigma. Gauld and Mitchell (1981) suggest that this species is related to E. trilobus and we agree with this. Although E. nigronotatus is probably related to E. trilobus, E. nigronotatus can be distinguished from it by characters such as the shape of the fore wing fenestra, i.e., fenestra very long, proximally extensively glabrous and proximal end of fenestra widely touching the anterior margin of the discosubmarginal cell, as in Fig. 32F, but fenestra short and the proximal end of the fenestra widely separated from the anterior margin in E. trilobus.
Bionomics. Unknown. Differential diagnosis. This species is very easily distinguished from all other Japanese Enicospilus by the black interocellar area (Fig. 33B, C), characteristic bulletshaped proximal sclerite (Fig. 33F), and the absence of the central sclerite (Fig. 33F).
Morphological characters indicate that E. nigropectus is very closely related to E. montaguei (Turner, 1919), but can be distinguished by the longer fore wing fenestra and irregularly rugose scutellum of E. nigropectus. Moreover, this species has been confused with E. abdominalis by many authors but is very easily separated from it (cf. Differential diagnosis of E. abdominalis).
Differential diagnosis. This species can be very easily distinguished from all other Enicospilus species by its characteristic sclerites of the fore wing fenestra (i.e., proximal sclerite entirely weakly pigmented and half-moon-shaped, and margin of the proximal sclerite distinctly separated from the margin of the fenestra, as in Fig. 34F).
This species exhibits a wide range of morphological variation in size and colour pattern. The proximal sclerite is usually weakly pigmented but is strongly pigmented in the holotype of E. tenuinubeculus.
Enicospilus pseudopuncticulatus Shimizu, sp. nov. http://zoobank.org/24488D4C-E242-4199-B87E-98B67E27BC4A Figure 35 Etymology. This species is very close to E. puncticulatus, hence the specific name based on their similarity. Bionomics. Unknown. Differential diagnosis. Although this species is very close to E. puncticulatus based on both morphology and DNA barcoding (Fig. 6), it is rather readily distinguishable by the linear central sclerite and straight fore wing vein 1cu-a, as in Fig. 35F.
Differential diagnosis. This species resembles E. biharensis, E. maruyamanus and E. transversus, all of which are rather difficult to distinguish from each other. However, E. pudibundae can be distinguished by the evenly curved fore wing vein 1m-cu&M (Fig. 36F), lack of proximal pectinae of the hind tarsal claw, and entirely closely punctate meso-and metapleuron (Fig. 36E). Gauld and Mitchell (1981) separated E. biharensis, E. pudibundae and E. transversus by the value of CI, but with the caveat, "Whether this character will prove to be completely reliable we doubt". However, we consider these species to be certainly distinct when using a combination of characters (cf. Table 7). Figure 37 Enicospilus puncticulatus Tang Bionomics. Unknown. Differential diagnosis. This species is most similar to E. pseudopuncticulatus sp. nov., but easily distinguished by the rounded central sclerite and more or less curved fore wing vein 1cu-a, as in Fig. 37F. Furthermore, E. puncticulatus resembles E. melanocarpus, but can be distinguished by the usually separated proximal and distal sclerites (Fig. 37F) (proximal and distal sclerites strongly confluent in E. melanocarpus, as in Fig. 28F) and entirely testaceous metasoma (Fig. 37A) (posterior metasomal segments almost always black in E. melanocarpus, as in Fig. 28A).

Enicospilus puncticulatus Tang, 1990
Remarks. Specimens with various shapes of sclerites of the fore wing fenestra but which are very similar in sculpture are identified as this species by the key provided by Tang (1990). It has proved impossible to segregate morphospecies based on discrete differences in sclerites, and our DNA barcodes differ by less than 1%; if E. puncticulatus represents a species complex, it seems that the CO1 gene is not useful for delimiting species in this complex. Figure 38 Ophion pungens Smith, 1874: 396; HT ♂ from Japan, NHMUK, examined, photographs provided by Shimizu and Broad (2020: fig. 24 Rao and Nikam, 1969: 343; LCT ♂ from India, designated by Gauld and Mitchell (1981: 304), NHMUK, examined; syn. nov. Enicospilus unicornis Rao and Nikam, 1970: 103; HT ♀ from India, MUC, not examined; junior primary homonym of Enicospilus unicornis Rao & Nikam, 1969; syn. nov. Figure 39 Ichneumon ramidulus Linnaeus, 1758: 566; HT, sex and locality unknown, not examined. Distribution. Afrotropical, Oriental, and trans-Palaearctic regions (Yu et al. 2016); this is a predominantly Palaearctic species and may be restricted to there, i.e., all reliable distribution records have been only from the Palaearctic region. Enicospilus ramidulus is one of the most frequently encountered Enicospilus species throughout the Palaearctic.
Bionomics. Recorded from a wide variety of hosts, but some records are undoubtedly the result of misidentifications of the ichneumonid. Reliable rearings are from species of Noctuidae, particularly the subfamily Hadeninae (Broad and Shaw 2016).

31.VII.1931, T. Shiraki leg. (TARI); PT of Enicospilus vontalis
Distribution. Afrotropical, Australasian, Eastern Palaearctic, Oceanic and Oriental regions (Yu et al. 2016). This is a widely distributed species in tropical to temperate regions of the Old World. Gauld and Mitchell (1981) thought that the presence of E. riukiuensis in the Afrotropical region (Madagascar) resulted from human trans-Indian Ocean trade, although there is no reliable evidence to support or refute this hypothesis.
Bionomics. Unknown. Differential diagnosis. This species is one of the most easily identified Enicospilus species based on the following character states: clypeus nasute (Fig. 40B, D); mandible evenly tapered, rather short and stout, with upper tooth as long as lower (Fig. 40B, D); proximal and central sclerites strongly sclerotised (Fig. 40F).
Remarks. The central sclerite of the Madagascan specimens is smaller than others. The colour of the interocellar area is usually a useful character for identification of Enicospilus, however, the interocellar area varies considerably in colour in E. riukiuensis.
Differential diagnosis. This species is very easily distinguished from all other species of Enicospilus by the shape of the clypeus (flat and projecting, with a distinct gap between clypeus and mandibles in profile, as in Fig. 41B, D), mandible (short and stout, as in Fig. 41B, D), and fore wing sclerites (Fig. 41F).
Remarks. We could find no morphological differences between E. sakaguchii and E. iracundus syn. nov., except the faint presence or absence of the central sclerite.
Distribution. Eastern Palaearctic and Oriental regions (Yu et al. 2016); this is a predominantly Oriental species.
Bionomics. Unknown. Differential diagnosis. This species resembles E. melanocarpus but can be distinguished by the presence of a glabrous area in the proximal part of the fore wing marginal cell (Fig. 42F) (marginal cell uniformly setose in E. melanocarpus, as in Fig. 28F) and linear central sclerite (Fig. 42F) (central sclerite usually circular to oval in E. melanocarpus, as in Fig. 28F) (also see Table 8).  (Enderlein, 1921) ♂ from Taiwan A habitus B head, frontal view C head, dorsal view D head, lateral view E mesosoma, lateral view F central part of fore wing.
Remarks. This species is morphologically very stable except that the mesosoma varies from entirely dark to reddish. Figure 44 Henicospilus combustus var. shikokuensis Uchida, 1928: 224; LCT ♀ from Japan, designated by Townes et al. (1965: 334), SEHU, examined.  (Uchida, 1928) ♀ from Japan A habitus B head, frontal view C head, dorsal view D head, lateral view E mesosoma, lateral view F central part of fore wing.
Bionomics. No host records from Japan. Enicospilus shikokuensis is one of the most frequently collected ichneumonids in spring in Japan, and it seems to be univoltine.
Differential diagnosis. As mentioned in the diagnosis of E. multidens stat. rev., this species is sometimes confused with E. multidens stat rev., but can be distinguished by the characters listed in the diagnosis section of E. multidens stat. rev. Gauld and Mitchell (1981) compared E. shikokuensis to E. ramidulus, but E. shikokuensis is easily distinguishable by the much wider lower face (Fig. 44B), longer and slenderer mandible (Fig. 44B, D), larger size, etc.
Differential diagnosis. According to Gauld and Mitchell (1981), this species is very similar to E. rufus (Brullé, 1846), but distinguished by its longer fore wing fenestra (Fig. 45F). It is also sometimes confused with E. sakaguchii due to their similar clypeus shape (i.e., flat, projecting apically above mandibles in profile, as in Figs 41D and 45D) and absence of the fore wing central sclerite (Figs 41F, 45F), but easily distinguished by many mandibular characters, such as the mandible rather long in E. shinkanus (Fig. 45B, D) but very short and stout in E. sakaguchii (Fig. 41B, D); and outer mandibular surface smooth in E. shinkanus (Fig. 45D) but with a diagonal setose groove in E. sakaguchii (Fig. 41B, D).
Differential diagnosis. This species is morphologically close to E. abdominalis but can easily be distinguished from it, and also from all other Japanese species, by the strong posterior transverse carina of the propodeum (Fig. 46E) and characteristic colour pattern (T4 is usually conspicuously brighter than adjacent segments) (Fig. 46A).
Remarks. Enicospilus signativentris is more or less morphologically stable, although it exhibits a very wide range of colour variation (i.e., from entirely orange to entirely dark brown). DNA barcoding analysis supports the conclusion that variable body colour represents intraspecific variation. There was no difference of p-distance between the entirely testaceous (SEN97 from Ōsaka) and the entirely dark brown individuals (SEN98 from Wakayama). Figure 47 Enicospilus stenophleps Cushman, 1937: 309; HT ♀ from Taiwan, DEI, not examined.
Type series. A holotype male only. Bionomics. Unknown. Differential diagnosis. This species is more or less similar to E. laqueatus, but easily distinguished by the position and shape of the central sclerite: central sclerite positioned in the anterodistal part of the fenestra, smaller and moderately sclerotised in E. takakuwai sp. nov., as in Fig. 48F, but positioned in the centrodistal part of the fenestra, larger and strongly sclerotised in E. laqueatus, as in Fig. 24F.
Legs. Hind leg with coxa in profile 1.7× as long as deep; basitarsus 1.9× as long as second tarsomere; fourth tarsomere 3.3× as long as wide; tarsal claw simply pectinate.

Enicospilus tripartitus Chiu, 1954
Bionomics. Unknown. Differential diagnosis. This species resembles E. laqueatus, E. pseudantennatus, and E. vestigator in the shapes of the sclerites, but can easily be distinguished by the dense and stout setae and punctures of the outer mandibular surface (Fig. 49B, D), deep basal concavity of the outer mandibular surface (Fig. 49B, D), etc., as summarised in Table 6.
Legs. Hind leg with coxa in profile 1.9× as long as deep; basitarsus 2.1× as long as second tarsomere; fourth tarsomere 3.7× as long as wide; tarsal claw simply pectinate.
Colour (Fig. 50). Entirely red-brown except for head yellow-brown and apex of mandible black. Wings hyaline. Fore wing sclerites pigmented and amber. Wing veins red-brown to amber.

Enicospilus xanthocephalus Cameron, 1905
Bionomics. No Japanese rearings. A range of hosts have been recorded in the literature, with some looking more reliable than others.
Differential diagnosis. This species is sometimes confused with E. flavocephalus because their body size, general colour, body shape, etc., are very similar, as in Figs 17 and 52. However, E. xanthocephalus is easily distinguished by the black interocellar area (Fig. 52B, C), shape of fore wing veins and sclerites (Fig. 52F), etc. (cf. Differential diagnosis of E. flavocephalus for details). The significantly large value of AI (more than 2.0) is also characteristic of E. xanthocephalus and helps identification. Figure 53 Henicospilus yezoensis Uchida, 1928: 227; LCT ♀ from Japan, SEHU, examined. Enicospilus ranunculus Chiu, 1954: (Fukuda and Kusigemati 1986;present study). *New records.

Enicospilus yezoensis (Uchida, 1928)
Bionomics. Unknown. Differential diagnosis. This species is similar to E. melanocarpus and E. ramidulus. However, E. yezoensis is easily distinguished from all other species of Enicospilus by the following combination of character states: proximal and distal sclerites separated (Fig. 53F); central sclerite comma-shaped (Fig. 53F); face wide and subquadrate (Fig. 53B); gena wide and not constricted behind eye in dorsal view (Fig. 53C, D); and diagonal groove of outer mandibular surface with dense and long setae (Fig. 53B, D) (also see Table 8).

Bionomics.
Unknown. Differential diagnosis. Enicospilus yonezawanus is one of the most common Enicospilus species in Japan and easily distinguished from all other Enicospilus species by the following combination of character states: ventral margin of clypeus impressed (Fig.  54B, D); fore wing fenestra with triangular proximal sclerite and without central sclerite (Fig. 54F); and meso-and metapleuron closely punctostriate (Fig. 54E). Enicospilus yonezawanus is also sometimes confused with E. jilinensis but can easily be separated (cf. Differential diagnosis of E. jilinensis).
Remarks. There is some variation in the shape of the proximal sclerite, but in Japanese specimens it is usually very stable. Chiu, 1954, stat. rev. Figure 55 Enicospilus zeugos Chiu, 1954: 64; HT ♀ from Taiwan, TARI, examined; stat. rev. Enicospilus henrytownesi Chao and Tang, 1991:  Distribution. Oriental region (Yu et al. 2016). Newly recorded from Japan. JAPAN: [Ryûkyûs] Okinawa. Bionomics. Unknown. Differential diagnosis. This species can very easily be distinguished from all other Enicospilus by the unique shape of the fore wing sclerites (Fig. 55F).

Enicospilus zeugos
Remarks. Both E. zeugos stat. rev. and E. henrytownesi syn. nov. had been synonymised under E. grammospilus (Enderlein, 1921) by Gauld and Mitchell (1981: 316) and Shimizu (2018: 93) respectively. However, the shape of fore wing veins and sclerites and the results of DNA barcoding analysis indicate that E. zeugos stat. rev. (with E. henrytownesi syn. nov. as a junior synonym) is easily separated from E. grammospilus, hence we propose a revised status for E. zeugos stat. rev. here.

Enicospilus species erroneously recorded from Japan
The following species have been recorded from Japan in error so were not included in the present study. Townes, Townes & Gupta, 1961. Gupta (1987) recorded this species from Japan, but this record was probably based on a misidentification. Figure 55. Enicospilus zeugos Chiu, 1954, stat. rev. ♀ from Japan A habitus B head, frontal view C head, dorsal view D head, lateral view E mesosoma, lateral view F central part of fore wing.
Enicospilus flavicaput (Morley, 1912). This species had been recorded from Japan by Matsumura and Uchida (1926), and some additional Japanese specimens were also identified as E. flavicaput. However, all Japanese specimens identified as this species were based on misidentifications of other Enicospilus species, including a new species (E. matsumurai sp. nov.). Enicospilus merdarius (Gravenhorst, 1829). This species had been recorded from Japan by Matsumura and Uchida (1926) and Uchida (1928). However, all Japanese specimens identified as this species have proved to belong to other Enicospilus species. Enicospilus merdarius has been recorded from all over the world except the Afrotropical, Antarctic and Australasian regions (Yu et al. 2016). As Gauld and Mitchell (1981) mentioned, however, any individuals of Enicospilus with two fore wing sclerites and an entirely orange-brown metasoma have been identified as this species, and the name 'E. merdarius' had frequently been used as a blanket term. Broad and Shaw (2016)

Species richness analysis
A total of 12, 8, 19, 31, 25, and 33 species were observed from the latitudinal zones A to F respectively, and saturation species richness was estimated at 13. 98, 8.65, 20.00, 36.99, 32.11 and 56.46 in each zone (Figs 56,57; values rounded to the nearest whole number for biological meaningfulness). Enicospilus species richness in Japan significantly decreases towards the north (Spearman's rank correlation coefficient = -0.89, p-value = 0.03; Fig. 57). Regional patterns of the four categories (i.e., number of specimens, collection events, collectors, and species) are visualised as heat maps in Fig. 58. All of these indicate apparent regional biases. The number of specimens, collection events, and collectors were remarkably high in Ryûkyûs (285,112,and 69 in Kagoshima,and 242,181,and 125 in Okinawa respectively, as in Table 9 and Fig. 58), and there were many collection events in Shizuoka (= 94) and Hiroshima (= 101) too. In contrast, three of six prefectures in Tôhoku (Akita, Iwate, and Miyagi), three of four in Hokuriku (Toyama, Ishikawa, and Fukui), four of nine in Kantô-Kôshin (Ibaraki, Gunama, Yamanashi, and Chiba), two of four in Tôkai (Gifu and Aichi), four of six in Kinki (Kyôto, Shiga, Ōsaka, and Nara), three of five in Chûgoku (Tottori, Okayama, and Yamaguchi), one of four in Shikoku (Kagawa), three of seven in Kyûshû (Saga, Ōita, and Miyazaki), and Ogasawara were sparsely represented in all categories (Table 9; Fig. 58). In particular, the number of specimens, collection events, and collectors as well as of species were significantly scarce in the following regions: zero in one prefecture of Chûgoku (Tottori) and one of Shikoku (Kagawa); and one or two in three of Tôhoku (Akita, Iwate, and Miyagi), one of Kantô-Kôshin (Chiba), one of Tôkai (Gifu), one of Kinki (Shiga), and one of Chûgoku (Okayama) ( Table 9; Fig. 58).  Enicospilus species richness across Japan significantly decreases towards the north (Spearman's rank correlation coefficient = -0.89, p-value = 0.03). Table 9. Regional patterns of the number of specimens, collection events, collectors, and species. Bold indicates especially small numbers (fewer than 5). Individual-based observed and rarefaction numbers of Enicospilus species in Japan are shown in Table 10 and as a species accumulation curve in Figure 59, based on ACE and Chao 1 estimators. Estimated species number of Enicospilus in Japan differed between the two estimators and was lower with ACE than Chao 1 estimator: 54.79 and 55.02 species at 1,850 individuals in ACE and Chao 1 estimators respectively.

Discussion
We revised the Japanese species of Enicospilus using a combined morphological and DNA barcoding approach to delimit and describe species. Some studies have suggested that genetic introgression has rather frequently occurred in the Ichneumonidae by Wolbachia endosymbionts, leading to misleading DNA barcode signals (e.g., Klopfstein et al. 2016). The inconsistency of each approach for a few species (E. xanthocephalus, E. stenophleps, E. puncticulatus, and E. flavicaput (non-Japanese specimens)) suggests that Table 10. Rarefaction of Enicospilus species in Japan estimated using ACE and Chao 1 estimators.

Individuals
Observed number of species Estimated number of species only on the character states of fore wing sclerites as much as possible, and should use a combination of more than two morphological characters for accurate species identification. Moreover, sequence data on databases (e.g., GenBank, BOLD, and DDBJ) are sometimes based on misidentified specimens. For example, a sequence of E. ramidulus (accession number: AB917966) has been deposited in GenBank based on a misidentification of a Netelia species. Therefore, we always have to use not only either DNA barcoding data or morphological characters, but also some other characters to accurately delimit species, as many previous authors have suggested (e.g., Sperling 2014, 2015;Klopfstein et al. 2016;Johansson and Cederberg 2019). Padial et al. (2010) have also suggested that more than three different sources may be needed for reliable species delimitation. Species richness of Enicospilus significantly increases from north to south in Japan (Fig. 57); this latitudinal diversity pattern has not been demonstrated for many groups of ichneumonids (e.g., Townes 1969;Owen and Owen 1974;Gauld and Mitchell 1981;Gauld 1987;Quicke 2015), but is usual in ophionines (e.g., Mitchell 1978, 1981;Gauld 1985) and is probably true for some other subfamilies, although the data are not yet available to test this (Veijalainen et al. 2012). Observed species richness did not increase by uniform percentage from north to south, for example fewer species were found in southern zones B and E than in northern zones A and D, presumably because of the smaller amount of suitable habitat and smaller geographic area in some zones. However, the Japanese archipelago is fortunately located in a rather narrow longitudinal range and we have used six latitudinal zones to reduce the effect of Figure 59. Individual-based species accumulation curve, comparing the observed and estimated numbers of Enicospilus species in Japan, based on ACE and Chao 1 estimators. regional sampling biases in the analysis, therefore our results provide enough information to describe a latitudinal trend in species richness of Enicospilus species.
An overwhelmingly larger number of species in zone F (= Ryûkyûs) is worthy of special mention, especially as the diversity of habitats in Ryûkyûs is apparently narrower than the other zones; these suggest that Ryûkyûs is one of the biodiversity hotspots of Enicospilus species. In contrast, although Ogasawara is in the southern subtropical region, only 3 species (E. laqueatus, E. melanocarpus, and E. signativentris: 6% of Japanese Enicospilus species) were found there. This is very low species richness compared to 31 species (66% of Japanese species) in Okinawa; this is probably because Ogasawara is a group of small remote oceanic islands and the wasps have arrived there recently, indicating their surprisingly strong dispersal abilities. Moreover, E. signativentris is frequently reared from cocoons of agricultural pest moths on leaves of Brassicaceae plants (e.g., Cabbage), suggesting that human activities have facilitated their dispersal.
Nine of 47 (19%) Japanese Enicospilus species are endemic to Japan, and the rest (81%) are shared with other countries. Most of the non-endemic species are probably derived from the southern tropics, but some trans-Palaearctic species (E. combustus and E. ramidulus) probably dispersed from the continental temperate region via the Korean Peninsula or Sakhalin. Moreover, no endemic Enicospilus species are recognised in Ogasawara.
In the present study, a total of 47 species are recognised in Japan. However, some species, such as E. puncticulatus, probably consist of some potential cryptic species. Many Enicospilus species described in the present study and by previous authors have been described based on very few specimens (sometimes only the holotype). The estimated species number in Japan based on the ACE and Chao 1 estimators is ca. 55 species. Therefore, the taxonomy of Japanese Enicospilus is not complete.
Sampling of Enicospilus is biased to known biodiversity hotspots, as with many other organisms. Ryûkyûs are one of the most famous Japanese collecting sites as well as a biodiversity hotspot, receiving much attention from many collectors and scientists. Because many endemic species are found in Ryûkyûs, this is a good location to study phylogeography and biogeography. Hence, we have had access to more collection events, and many specimens of Enicospilus, compared to other regions (Table 9; Fig. 58). Although Enicospilus of the northern part of Japan were generally not well sampled, those of Hokkaidô have been rather well collected, because Hokkaidô is also a famous collection site with rich fauna and flora. However, there are not many Enicospilus specimens from Nagano Prefecture, although there is one of the most famous Japanese insect collection sites. This is probably because the aim of most insect collectors is to collect diurnal insects, especially longicorn beetles or butterflies, in high elevational Alps.
Regional sampling biases also seem to be related to the distribution of universities with traditional entomological laboratories or of active and large entomological societies. Enicospilus as well as all other insects of Hokkaidô, Ehime/Kôchi, and Fukuoka prefectures are well sampled, because of Hokkaidô University (SEHU), Ehime Univer-sity (EUM), and Kyûshû University (KUEC) respectively. Enicospilus of Niigata and Hyôgo prefectures are also well sampled, probably because of the Essa entomological society in Niigata and Teneral in Hyôgo.
In regions which are difficult to access and/or far from large cities (and which are not famous collection sites, nor near entomological laboratories and societies), Enicospilus are not well sampled, as is the case for all insects. There are regions of high potential biodiversity which are under-sampled, such as low elevational and coastal laurel forests on the Pacific side and Western Japan, grassland on karst in Chûgoku and Shikoku mountains, and alpine areas of the Chûbu region.
These regional sampling biases affect not only analyses of species richness of Enicospilus in the present study, but also our general understanding of biodiversity and how to conserve it. Biodiversity is threatened with decline or loss in many environments. For instance, grasslands and marshes are declining by a progression of plant succession, sometimes due to changes in human land use, and in Japanese forests, over-browsing by a population explosion of deer frequently results in a bare forest floor; thus, many insects found there, including parasitoid wasps, are declining and threatened with extinction (e.g., Takatsuki 2009;Sakata and Yamasaki 2015;Nakahama et al. 2018Nakahama et al. , 2020. To conserve such insects, we have to gather information on their diversity and distribution, as well as their ecology. Further comprehensive sampling efforts are strongly needed. More comprehensive sampling can be achieved through a combination of professional study and citizen science, amateur collecting. Either alone will be insufficient. In Japan, there are many people who enjoy entomology, especially field collecting (like hunting) and making private collections. This hobby, and feeling insects to be very special, is often called "Mushi-ya" (Takada 2014). Practitioners of Mushi-ya are very skilled in collecting insects and support faunistic and taxonomic studies as well as all other fields of natural history in Japan. Large numbers of specimens examined in the present study were actually provided by Mushi-ya. Hence, we have to continue to strengthen cooperation with citizen scientists, especially Mushi-ya.

Conclusions
The study of a large specimen base has been vital for resolving the taxonomic confusion that has surrounded Japanese Enicospilus. We can now identify them at a species level and therefore conduct more applied research, although some undescribed or unrecorded species are probably still present in Japan. Species richness of Enicospilus in Japan significantly increases from north to south, as in many other ophionines, although a southern group of small remote oceanic islands, Ogasawara, have only 6% of Japanese Enicospilus species. However, there are regional sampling biases, and the knowledge of the taxonomy and species richness patterns of Japanese Enicospilus is incomplete. Therefore, we need further sampling to fill the regional gaps and complete the taxonomy.