Revision of the World Species of Megaphragma Timberlake (Hymenoptera: Trichogrammatidae)

Simple Summary Parasitoid wasps of the genus Megaphragma are some of the smallest known insects, being as small as some unicellular protozoans. Their life history is not known in great detail, but all species with known biology are parasitoids of thrips eggs (Thysanoptera) and as such, they are potential biological control agents of these pests. At the current state of knowledge of the genus, it is impossible to identify with confidence most of the Megaphragma species (original descriptions lack essential details or illustrations; molecular markers are available for very few species; many species are still undescribed while others were described multiple times). We provide the first revision of the genus that includes the formal descriptions and naming of 22 species and a key to all 32 valid species. Abstract Megaphragma species are important models for basic organismal research, and many are potential biological control agents. We present the first extensive revision of species of the genus Megaphragma based on morphological and molecular data. Our revision includes all previously described species, 6 of which are synonymized, and 22 of which are described here as new. We also provide the first key to all species of the genus and reconstruct their phylogeny based on 28S and CO1 molecular markers. The following species are synonymized with M. longiciliatum Subba Rao: M. aligarhensis Yousuf and Shafee syn. nov.; M. amalphitanum Viggiani syn. nov.; M. decochaetum Lin syn. nov.; M. magniclava Yousuf and Shafee syn. nov.; M. shimalianum Hayat syn. nov. M. anomalifuniculi Yuan and Lou syn. nov. is synonymized with M. polychaetum Lin. The following species are described as new: M. antecessor Polaszek and Fusu sp. nov.; M. breviclavum Polaszek and Fusu sp. nov.; M. chienleei Polaszek and Fusu sp. nov.; M. cockerilli Polaszek and Fusu sp. nov.; M. digitatum Polaszek and Fusu sp. nov.; M. fanenitrakely Polaszek and Fusu sp. nov.; M. funiculatum Fusu, Polaszek, and Viggiani sp. nov.; M. giraulti Viggiani, Fusu, and Polaszek sp. nov.; M. hansoni Polaszek, Fusu, and Viggiani sp. nov.; M. kinuthiae Polaszek, Fusu, and Viggiani sp. nov.; M. liui Polaszek and Fusu sp. nov.; M. momookherjeeae Polaszek and Fusu sp. nov.; M. nowickii Polaszek, Fusu, and Viggiani sp. nov.; M. noyesi Polaszek and Fusu sp. nov.; M. pintoi Viggiani sp. nov.; M. polilovi Polaszek, Fusu, and Viggiani sp. nov.; M. rivelloi Viggiani sp. nov.; M. tamoi Polaszek, Fusu, and Viggiani sp. nov.; M. tridens Fusu, and Polaszek sp. nov.; M. uniclavum Polaszek and Fusu sp. nov.; M. vanlentereni Polaszek and Fusu sp. nov.; M. viggianii Fusu, Polaszek, and Polilov sp. nov.

de Costa Rica (MZUCR). Natural History Museum of Oman (NHMO); Natural History Museum, London, UK (NHMUK: Natalie Dale-Skey); Musée Royal de l'Afrique Centrale, Tervuren, Belgium (MRAC: Eliane de Coninck); Plant and Food Research New Zealand (formerly DSIR: Jocelyn Berry); University of California, Riverside, USA (UCRC: Serguei Triapitsyn). Additional material was received for identification from several institutions, and a number of recent collections by the authors and Dr John Noyes (NHMUK) contributed substantial material to this revision.

Morphology
All material was examined on microscope slides for morphological characters using an Olympus BX63 microscope with Nomarski differential interference contrast (DIC) with 40× and 100× objectives. Since the lysis buffer used for DNA extraction (see below) contains proteinase K, there is no need to clear the body with KOH as usually performed before slide mounting. Instead, specimens were extracted from the lysis buffer with an adjustable volume pipette (0.5 to 10 µL) set at 1-2 µL to reduce liquid loss and transferred to distilled water to remove unwanted reagents. Afterward, they were dehydrated using a series of ethanol solutions of increasing concentration and cleared in clove oil as described by Noyes [34]. Afterward, some specimens were mounted laterally in Canada balsam while others were dissected and wings, antennae, head, and body were mounted separately under different coverslips following [34]. However, mounting the abdomen dorsal side up has the disadvantage of making the setae on the sides of the tergites very difficult to see. Where possible, the setae on the disc of the fore wing were counted on the upper and lower surfaces. Body colour was observed on both card-mounted specimens and on slide-mounted specimens in which the generally unremarkable body pigmentation remains preserved.
Selected specimens were dried using a critical point drier and examined with an electron microscope as described in Polilov [26].
In species with a single discal fore wing seta, its length is important: "short" = shorter than or equal to the distance between the 2 proximal wing fringe setae (i.e., those closest to the seta); long = longer than the distance between the 3 proximal wing fringe setae (see Figure 1c).
A peculiar type of metafemoral spine with a unique shape, structure, and position is present in all species of the ghesquierei-group. Probably non-homologous metafemoral spines are present in other species groups.

Molecular Methods
DNA was extracted using a Qiagen DNeasy Blood and Tissue Kit with modifications as described in Cruaud et al. [35]; specimens were lysed whole for 6-8 h without grinding, then frozen at −80 • C overnight and thawed at room temperature before addition of buffer AL. After about the first 100 extractions the freezing stage was omitted as it appeared not to increase DNA yield significantly.
The primer pair D23F (5 -GAGAGTTCAAGAGTACGTG-3 ) [36] and 28Sb also known as D3B (5 -TCGGAAGGAACCAGCTACTA-3 ) [37,38] was used to amplify an approximately 850 bp fragment from the 5 end of the nuclear ribosomal 28S gene spanning the D2-D3 region. In the instances where there was no detectable PCR product, we performed a second PCR using 1 µL of the primary PCR product and the semi-nested primer pair D23F combined with the newly designed reverse primer 28Sbsn (5 -GATGGTTCGATTAGTCTTTCG-3 ), which amplified an approximately 800 bp fragment of the 28S rDNA.
The PCR conditions were as described in Fusu and Polaszek [42] except the standard barcode region was amplified at 42 • C.
All PCR products were checked by gel electrophoresis in 1% agarose gels, cleaned using AxyPrep PCR clean-up beads as per manufacturer's instructions, then sequenced bidirectionally using BigDye terminator reaction mix v.3.1 in a 3730xl DNA analyser (Applied Biosystems) at the NHMUK sequencing facility.
The forward and reverse sequences were assembled and corrected using the Staden Package v.1.7.0 [44]. The resulting sequences were aligned in Mega v.7.0.14 [45] with the Clustal W program [46] for the CO1 gene; the 28S gene was aligned with the MAFFT web server [47] using the E-INS-i algorithm, a gap opening penalty of 2, leave gappy regions option activated, and UPGMA as a tree-building method for the guide tree. The CO1 sequences were also translated to amino acids to detect eventual stop codons that indicate NUMTs. The two alignments were first used in single-gene phylogenetic analyses in RAxML-NG v.1.0.0 [48] to detect eventual long branches and misplaced sequences (that might indicate pseudogenes or contaminants) that are to be checked/removed prior to the concatenation of the two datasets. A phylogenetic analysis of the concatenated but unpartitioned dataset using a simple substitution model (K2P) was also conducted in RAxML-NG since over-parameterization of the substitution and partition models might be a problem in a maximum likelihood framework [49], especially when using a comparatively small alignment. For the partitioned analyses, data blocks were delimited in Mesquite v.3.10 [50], CO1 being divided by codon position and 28S was treated as one block. The best partitioning scheme and substitution models were selected using PartitionFinder2 v.2.1.1 [51], with branch lengths proportionally linked and the search option set to all.
Partitioned analyses were run in RAxML-NG [48], which is maximum likelihood (ML) based, and MrBayes v.3.2.7 [52], which is based on Bayesian inference (BI) with the following substitution models as indicated by PartitionFinder2: GTR+G for 28S, HKY+I+G, TIM+I+G, and TIM+G for the 1st, 2nd, and 3rd codon positions of CO1, respectively. For MrBayes, we substituted TIM with GTR, since the former model is not available in this program. In MrBayes two parallel analyses, each with four chains, were run for 10 7 MCMC generations, with trees and lnLs sampled every 100 generations; all estimated parameters were unlinked among partitions except for branch lengths; convergence of all parameters and estimated sample size values (ESS) above 200 were assessed by examining the trace files in Tracer v1.7.1 [53]. Support for the maximum likelihood analysis was estimated with rapid bootstrapping (number of replicates determined by the autoMRE criterion [54]).
Bootstrap percentages (BP) over 85% were considered as strong support and BP smaller than 65% as weak. Posterior probabilities (PP) over 0.95 were considered as strong support and those below 0.90 as weak. The trees were imported and modified in FigTree v1.4.4 [55] and Adobe Illustrator.
All sequences were uploaded to GenBank (accession numbers ON555486-ON555643 for 28S and ON557406-ON557518 for CO1). Since a part of the DNA extractions did not yield PCR products, the presence of a DNA code after the label data of a specimen does not necessarily mean that it has an associated DNA sequence. A complete list of specimens with associated DNA sequences and their repository is provided in Appendix A.

New Species Left Undescribed
We have identified several species that are clearly new based either on their DNA sequences or morphology (or both), but are not described herein for one or more of the following reasons: • Species known from males only. Within (e.g.) the ghesquierei-group, several new species have been identified (at least 7 or 8-see Figure 2), which are known only from males. Since in most cases females are essential for species recognition (e.g., antennal structure, ovipositor length), we have refrained from describing these species here. • Incomplete specimens. In several instances, new species are indicated by both morphology and DNA sequences, but a crucial morphological character is missing, most often the antennae. These specimens and their sequences have been curated pending the discovery of fresh, complete specimens. • Poorly-mounted specimens. In a few cases, slide-mounted specimens not represented by DNA sequences appear to be very likely new species. In many cases, the material is simply not in good enough condition for the designation of a holotype to represent the species.
Clearly, there is overlap and gradation between the above categories, and we have used our discretion when deciding whether or not to describe specimens. In all cases, information as to our opinion of species status is included on the specimens.

A Note on Figures Supplementing the Descriptions
While all new species are fully described, in many cases there are aspects of the morphology that differ so little between species that images of these structures would be superfluous. In these cases, "cf Figure" is used, where the reader is referred to a figure that to all intents and purposes can serve to illustrate the species while actually depicting a different one. This is especially true for many species of the ghesquierei-group, where several species are morphologically indistinguishable, and to a lesser extent for M. mymaripenne, M. noyesi sp. nov., and M. polilovi sp. nov. in the mymaripenne-group. In every case of extreme morphological similarity, robust molecular data are available to support separate species status. In addition, where possible, illustrations were made from holotypes. Under each photograph, we mention whether it is that of a holotype, neotype, or paratype.

Phylogenetic Analyses
We obtained DNA sequences for a total of 170 Megaphragma specimens (158 sequences for 28S and 113 sequences for CO1) (Appendix A). The 28S alignment was 1068 bp in length, while the CO1 alignment was 652 bp, though only shorter sequences (DNA mini-barcodes) were obtained for some species/specimens.  The phylogenetic trees from the single-locus analyses are in general agreement though, for example, the position of M. antecessor sp. nov. differs drastically between the two, while some species are present in one data matrix but not the other (Supplementary Figures S2  and S3). They are also in general agreement regarding the clustering of specimens into putative species: all species that are distinct based on CO1 are also distinct based on 28S. An exception is two specimens of M. digitatum sp. nov. that are very divergent on CO1, but almost identical on 28S. Both the ML and BI trees from the combined and partitioned dataset show similar topologies, with minor differences; hence, in Figure 2b on the BI tree, both posterior probabilities and bootstrap support were plotted at the nodes. The species that were included in the analysis are split into two major groups: one consisting of mostly species of the mymaripenne-group (species groups are discussed below), the other of species of the ghesquierei-group. Both have high posterior probabilities (0.99 and 1, respectively) but low bootstrap support, indicating strong support based on a low number of characters. Two other small groups are formed by two species each in the antecessorand polychaetum-groups. The antecessor-group is sister to M. liui sp. nov. in both of these trees, while two unnamed species of the polychaetum-group (SAM1 and SAM2) are basal to the mymaripenne-group instead of clustering with M. giraulti sp. nov. and M. cockerilli sp. nov.
(the other two species of the group). The two species of the longiciliatum-group, though forming a monophyletic group with a posterior probability of 1, and a medium bootstrap support of 69%, render the mymaripenne-group paraphyletic.
The two trees based on the combined analysis of both genes have a major difference from the unpartitioned analysis that used a simple substitution model and not the best fit model ( Figure 2a); in this latter tree, M. liui is recovered in a basal position in the ghesquiereigroup, where it belongs based on its morphology. Another major difference between the partitioned analyses and the unpartitioned one is the position of the antecessor-group that is not sister to the ghesquierei-group in the first analyses ( Figure 2b) as would be expected by its morphology, while it is retrieved as basal to all other Megaphragma in the second tree ( Figure 2a).
Even very short DNA sequences are sufficient to place a specimen, though in some cases this is also the explanation for the unstable and likely erroneous position in the phylogenies for some species. For example, HUM9 is correctly placed in the cluster with other M. noyesi based on a 296 bp CO1 sequence and the same is true for M. momookherjeeae sp. nov. retrieved as sister to M. antecessor based on a 394 bp CO1 sequence (the morphology of both species places them in the antecessor-group). Megaphragma antecessor, M. liui, and M. momookherjeeae that have their positions on the trees drastically altered depending on the analysis (partitioned versus unpartitioned) are represented by short sequences: 519 bp for 28S and 366 for CO1, 344 bp for 28S and 370 for CO1, and 394 bp for CO1, respectively. Diagnosis. Female (Figure 1a). Body rather compact, extremely small, length 0.16-0.3 mm. Antenna (Figure 1b) inserted at mid level of the internal orbital line, with short radicle, scape usually elongate, pedicel, anellus, single funicle segment present or absent, clava one -, two -, or three-segmented. Antennal formula: 1 (scape), 1 (pedicel), (1) (anellus), 1 (funicle), 2 (clava); or 1,1,(1),1,1 or 1,1,(1),0,3. The antenna is counted as four-or fivesegmented, since the anellus is not counted among the antennomeres. Claval segment 1 without multiporous placoid sensilla. Mandible with two small teeth. Maxillary palp very small and labial palp vestigial. Eye black unless otherwise stated. Mesosoma rather high, usually shorter than metasoma. Pronotum very short; mid lobe of mesoscutum not much longer than wide, either smooth or with polygonal or striate sculpture; one pair of adnotaular setae. Scutellum shorter than mid lobe of mesoscutum, with a pair of setae. Metanotum short; propodeum slightly longer than metanotum, or, in the middle, even longer, with a well-developed central area (disc) that may bear crenulae. Propodeal spiracle placed in an oval groove, and near the internal margin with two very small setae. Fore wing (Figure 1c) extremely narrow compared with other Trichogrammatidae genera, 5.3-10× as long as maximum discal width, with short submarginal vein; costal cell and parastigma not distinct; marginal vein very long, with one short seta at the base and with one or two setae centrally, which when paired may be of similar or very different lengths; stigmal vein very short with one or two short setae on the stigma; disc with one or a few setae in one or two rows or glabrous (when there is one seta it is located on the dorsal surface of the wing, when discal setae are more numerous they are located on both dorsal and ventral surfaces of the wing, cf Figure 22d,e). Hind wing without discal fringe on front margin. Legs robust, often with striate sculpture on coxae, also on femora and tibiae. T7 and T8, respectively, without spiracle and cercus.

Megaphragma
Male: As female, but often with postanellar antennomeres shaped differently. Genitalia tubular, very simple and usually small.
Relationship. The closest relatives of Megaphragma appear to be Prestwichia Lubbock and Sinepalpigramma Viggiani and Pinto [56]. Unfortunately, sequences for neither of these genera were available for comparison. We have used an Epoligosita Girault, two Oligosita Walker, and a Probrachista Viggiani species as outgroups. These Oligositinae genera are close phylogenetically to Megaphragma according to a previous molecular study [33]. Speciesgroup relationships are discussed below.
Distribution: Cosmopolitan. Hosts and biology. The known species of Megaphragma are all egg parasitoids of Thysanoptera (Supplementary Figure S4) [57][58][59]. Biological data are available only for a few species, e.g., M. mymaripenne, M. longiciliatum (as M. amalphitanum) [12][13][14], and are given below where available. It is interesting that at the same locality there may be more than one species of Megaphragma, even in Europe. Megaphragma viggianii and M. polilovi were found in Italy at the same locality and on the same host, while in a single sample from near Barkás Lake in Hungary, there are three species (M. longiciliatum, M. noyesi, and the undescribed species represented by the specimen HUM4, close to M. longiciliatum but distinct genetically).

Species groups in Megaphragma
On the basis of present knowledge, the following species group are proposed in Megaphragma: M. mymaripenne-group: antenna with a single funicle segment that is longer than wide (this feature also shared by polychaetumand longiciliatumgroups); T1 with longitudinal and/or transverse cells, with some denticles laterally within the cells (Figures 17f, 18f and 20b); T2-T4 each with a pair of short setae. M. longiciliatum-group: same as mymaripenne-group, but without cells on T1. According to the phylogenetic analysis, the group appears to be derived from within the mymaripenne-group having lost the denticulate cells on T1. M. polychaetum-group: antenna with a long, cylindrical, funicle segment; spatulate sensilla at the end of each clava segment, and a robust terminal sensillum on C2; fore wing disc with more than seven setae, often arranged in two rows. Male antenna is particularly distinctive, with an elongate C1, short C2 usually with very long sensilla.
Included species: M. cockerilli Polaszek and Fusu sp. nov., M. giraulti sp. nov., M. polychaetum Lin, M. kinuthiae Polaszek, Fusu, and Viggiani sp. nov. Our molecular analysis also includes two males of this group, representing two undescribed species (vouchers SAM1 and SAM2, NHMUK). They have the antennal structure characteristic for males of the group, but our analyses recover them basal to the mymaripenneand longiciliatumgroups instead of clustering them with the other two species of the polychaetum-group.
M. ghesquierei-group: antenna without funicle segment and with clava three-segmented, because the funicle is fused with the clava along an oblique suture. Fore wing disc with one seta on the dorsal surface or no setae. Propodeum characteristically produced centrally, almost always with a row of crenulae. Metafemur with a robust spine close to the connection with the tibia. Because of the intergradation in the structure of the antenna between the ghesquierei and other groups, we do not currently consider Paramegaphragma as a valid genus for the species in the ghesquiereiplus stenopterumgroups. It is possible that future analyses, especially including multigene or reduced genome representation data, may lead to the reinstatement of Paramegaphragma Lin as a valid genus. The two species formerly assigned to Paramegaphragma by Lin [9], M. stenopterum and M. macrostigmum, are not closely related and clearly belong to different species-groups (stenopterum-group and macrostigmum-group, respectively), though on morphological grounds stenopterum-group is clearly related to ghesquierei-group or even integral part of it. This is another reason for not recognizing Paramegaphragma. M. stenopterum-group: same as M. ghesquierei but with clava two-segmented. The antennal structure is very suggestive of the ghesquierei-group, given the similarity between the apparent C1 of the stenopterum-group and that of the ghesquierei-group; i.e., it is actually a funicle completely fused to the clava. In the antecessor-group, the funicle is distinct albeit transverse and anneliform. Pending further evidence, we consider the stenopterum-group as possibly nested within the ghesquierei-group. Megaphragma macrostigmum and M. caribea (macrostigmum-group) were considered by previous authors to belong in a group with M. stenopterum [60], and M. macrostigmum with M. stenopterum were both originally included by Lin [9] in his genus Paramegaphragma. However, the former two species lack any of the obvious apomorphies of the ghesquierei-group except for the apparently lost funicle. Members of the macrostigmum-group are otherwise similar in the structure of the fore wing and sculpture of the mesoscutum to the species in the longiciliatum-, mymaripenne-, and polychaetumgroups and are probably not related to the ghesquiereiand stenopterumgroups. Our molecular analysis did not include M. stenopterum, the only member of this species group.
Included species: M. stenopterum (Lin). M. antecessor-group: antenna with a transverse funicle segment not much larger than the anellus, and clava one-or two-segmented. Metatibia with a characteristic row of setae (Figures 11d and 21j). The structure of the antenna seems intermediate between that characteristic of the longiciliatumand mymaripennegroups and that of the ghesquierei-group.
In the latter species group, the antenna is apparently without a funicle, as the funicle is completely fused with the clava by an oblique suture and, hence, the clava appears threesegmented. Our phylogenetic analysis shows that M. antecessor and M. momookherjeeae, while resembling the ghesquierei species group in many features (including fore wing structure and the robust spine on metatibia), appear outside it, and basal to all remaining M. macrostigmum-group: as explained above, M. macrostigmum and M. caribea, while undoubtedly very closely related to each other, appear to have no connection with the ghesquierei-group (our molecular analysis did not include either of these two species). The antenna has the clava two-segmented and no funicle as in the stenopterum-group; the fore wing structure, however, is not similar to the ghesquierei-group but suggestive of the longiciliatum-group, especially M. priesneri; metafemur without the robust spine characteristic for the antecessor-, ghesquierei-, and stenopterumgroups.

Previously described species
Megaphragma caribea Delvare (Figure 3a Body yellow, with the following slightly darkened: occiput, meso-and meta-coxae, apices of meso-and metafemora. Metasoma with pale brown transverse bands. Male: Antenna with C1 longer than in female ( Figure 19b). Distribution: Colombia, Guadeloupe. Host: Selenothrips rubrocinctus (Giard). DNA data: no DNA sequences. Comments: The species was described in detail by the author. Megaphragma caribea is clearly close to M. macrostigmum (Lin). At present, their discrimination is based on the absence of long UST on the basal clava (C1) of the antenna of the latter species (Figure 6a). Since the original description did not indicate whether the species-group name caribea is a noun or an adjective, following Art. 31.2.2. of ICZN, we treat it as a noun and do not make a gender agreement. (Figure 3d (Figure 3f) 9× as long as maximum width; longest fringe seta 5× as long as maximum discal width. Fore wing disc without setae. Marginal vein with two long subequal setae centrally. Stigmal vein not enlarged, with two sensilla apically. Middle tibia with one large spine basally; metafemur with spine. T1 with elongate cells laterally, 2-3× as long as wide; T2-T4 without setae laterally. Ovipositor 1.7× as long as mesotibia.
Male: As female but C3 with fewer MPS and with ASC apically. Mid lobe of mesoscutum anteriorly with reticulate sculpture, remainder with longitudinal striation continuing onto scutellum (cf Figure 3g). Propodeum with a large subtriangular central area. Fore wing 7× as long as maximum width ( Figure 4a); the disc pointed distally, without setae. Metasoma with tergites with some short transverse striation centrally, and each with a pair of lateral setae (Figure 4b).
Body dark brown, with the following paler: frons and occiput, scutellum and propodeum, tarsi. Metasoma with tergites and sternites appearing as dark bands (in the slide-mounted types). Fore wing basally strongly infuscate with a dark marginal vein.
Male: Similar to female in all aspects of morphology except genitalia characters. Comments: The species is rather easily recognizable by the combination of features of the antenna, mid lobe of mesoscutum, propodeum, fore wing, and metasomal tergites.
The species was intended to be described by Nowicki, but was published by Ghesquière [61] (p. 36) because Nowicki's manuscript on several African Trichogrammatidae never reached the journal Revue de zoologie et de botanique Africaines in Tervuren where Ghesquière was working. Ghesquière [61] (p.37) gives the date of collection as "XII.1937", but, as given above, the holotype is labeled: "I.1938".  Mid lobe of mesoscutum anteriorly with reticulate sculpture. Propodeum with a very short central area. Fore wing 8× as long as wide ( Figure 5g). Metasoma (Figure 5h) without subpolygonal sculpture on tergites, but with some ridges, T2-T4 each with a pair of long setae. Ovipositor 1.1× as long as mesotibia.
Body brown to dark brown, with the following paler: antenna, legs. Metasoma with tergites and sternites appearing as dark bands (in the slide-mounted types). Fore wing completely hyaline.
Male: Similar to female in most characters except genitalia; antennal funicle slightly more elongate than in female, and clava darker than remainder of antenna. C2 without long UST; C3 shorter than in female (Figure 5f). Comments: Megaphragma longiciliatum is the most widely distributed Megaphragma species; hence the large number of synonyms. We have examined 150 specimens from 14 countries and have DNA sequences for 18 specimens from 8 very widely distributed countries. We have carefully assessed morphological variation within the specimens examined, and consider that it encompasses the morphological characteristics of the type material of the species synonymized above [7,9,10,63,64].
The holotype of M. longiciliatum is in extremely poor condition. The mountant, presumably gum chloral, has turned black. It is to be hoped that in a few years' time, the holotype will be destroyed completely, and one of the paratypes, all of which are still in excellent condition, can be designated a neotype. Unfortunately, there is no current provision under the Code to legitimately replace a holotype specimen that has deteriorated irremediably. Body uniformly pale brown, fore wing slightly to moderately infuscate below marginal vein.
Male: Similar to female in most characters except genitalia; antennal funicle slightly shorter than in female, C1 longer, C2 with long UST (Figure 6b  Mid lobe of mesoscutum anteriorly with subpolygonal sculpture, but often appearing smooth in slide-mounts. Propodeum with a very short central area. Fore wing (Figure 6e) 9-10× as long as wide, marginal vein with two long setae in the middle, setae on disc more or less regularly in a row of 10-15 setae, and longest fringe seta 5-6× as long as maximum disc width. T1 with sculpture represented by a combination of transverse and longitudinal cells, lateral ones twice as long as wide; sides of some cells with denticles present. The subsequent tergites show rather variable sculpture, differing from the pattern on the first tergum. T2-T4 each with a pair of very short setae.
Body uniformly pale brown, scutellum paler than mesoscutum. Legs pale, wings hyaline. Clava slightly darker than the remainder of the antenna.
Male (hitherto undescribed): same as female but antenna slender, with funicle twice as long as wide and C1 about 1.7× as long as C2. T1 with sculpture not as complete as in the female. Genitalia simple, tubular, 4.5× as long as wide (cf Figure 7f). Species-group placement: M. mymaripenne-group. Distribution: Argentina, Brazil, Chile, Costa Rica, Dominican Republic, Ecuador, Guadeloupe, Israel, Mexico, USA, and Venezuela.
Hosts: Megaphragma mymaripenne is a solitary egg endoparasitoid of several species of Panchaetothripinae (Thripidae). The most common host is the widespread Heliothrips haemorrhoidalis. The populations recorded in the USA [12] are represented mainly by females. The population reared in Argentina from maize and identified as M. mymaripenne [11] differs from the known populations of the species: the reared specimens from maize appear to be normally bisexual.
Comments: This species was described in detail by Timberlake [65], and additional features were given by Viggiani [6]. Megaphragma mymaripenne is extremely difficult to distinguish morphologically from the closely related species M. polilovi, and even from the more distantly related species M. noyesi, with which it has been previously confused. They differ, however, in the length and shape of C1, length of the scape and colour of the radicle, and length of the ovipositor, respectively, as outlined in the key. Without the molecular data, these subtle differences would have been overlooked or treated as intraspecific variability. The correlation between the molecular clades and morphological characters indicates, however, that there are three species involved.
Records from Israel are the only Old-World records for this species; previous records of M. mymaripenne, e.g., from Italy [13,14,66], turned out to be misidentifications of the new species M. polilovi. Comments: The type material of M. anomalifuniculi was not available to the authors. According to the illustration given by Yuan and Lou [67], M. anomalifuniculi appears to be similar, if not identical, to M. polychaetum Lin. The features concerning the funicular segment appear to derive from a preparation artifact.   Diagnosis. Female: Antenna (Figure 8d) with pedicel slightly shorter than scape, funicle as long as half pedicel. Clava two-segmented with two long UST on C1 of female; C2 with one MPS and two MT.

Megaphragma polychaetum
Mid lobe of mesoscutum ( Figure 8f) anteriorly with subpolygonal sculpture; propodeum with a very short central area. Fore wing (Figure 8g) 7× as long as wide, with maximum distal width less than 2× width measured at apex of marginal vein ( Figure 8g); maximum fringe seta length 4× maximum discal width; setae on ventral disc surface short, penultimate one not reaching to the base of the distal (Figure 22e). T1 without sculpture, but with a row of microspines; T2-T4 each with a pair of setae, shorter than their corresponding tergites. Ovipositor 1.1× mesotibia.
Entire head and body are very dark. Legs and antenna paler. Wings hyaline.
Male: Similar to female in most characters except genitalia. Antenna with funicle and C1 more elongate than in female, without UST on C1; C2 much shorter than in female ( Figure 8e) Comments: Following extensive inquiries over the decades since 1990 in Egypt and Denmark, the holotype (and indeed the remainder of the type series of four specimens) appears to be lost. A specimen with data almost identical to the holotype is in the NHMUK, but has aberrant antennae. Nevertheless, we here designate that specimen as neotype, given that the data are very similar to those of the original type [68] (only the collection date differs by less than a month). Furthermore, all of the remaining morphology accords perfectly with the original description. Unfortunately, extensive efforts to collect fresh specimens in both Egypt and Israel failed.
The neotype designation for M. priesneri (4) giving reasons (no references available heretofore) for believing that the original type material is lost (Article 75.3.4); (5) selecting a neotype specimen consistent with the original description of the species and that was collected not long (less than 1 month) after the original description (specimen in this case) and, as such, represents the type species (Article 75.3.5); (6) choosing a neotype from the originally cited type locality, Tanta, Egypt (Article 75.3.6); and (7) recording that the neotype is the property of a recognized scientific institution, NHMUK in London (Article 75.3.7).  Comments: This species, described in detail by the author, has the unique combination of a single seta on the fore wing and four-segmented antenna without any apparent funicle. A transverse, anelliform funicle is present in M. uniclavum, the only other species with a four-segmented antenna and a single seta on the fore wing.
There are differences between the Chinese text and the English text of the original description concerning the collecting dates of the type series. The examined paratype is mentioned in the English part but not in the Chinese part. Mid lobe of mesoscutum and scutellum longitudinally striate (Figure 10d). Propodeum with a pronounced subtriangular central area (Figure 21c). Fore wing 8× as long as wide, with one long central seta on the marginal vein, one discal seta, longest fringe seta 4-5× as long as maximum discal width. Metasoma with a row of crenulae on T2 (Figure 21c).
Head and metasoma very dark, mesoscutum brown, the remainder, including legs and antenna, paler. Fore wing infuscate basally.
Male: Similar to female in most characters except genitalia.
Mid lobe of mesoscutum (Figure 11d) with fine longitudinal striation; vertical/ventral anterior mid lobe of mesoscutum with coarse, reticulate sculpture ( Figure 11d); propodeum with subtriangular area centrally (Figure 11d) with 3-4 large crenulae; propodeum with hind margin arcuate. Fore wing (Figure 11b) 8.5× as long as maximum width, maximum distal width is 92× the maximum basal width; disc with a single short seta, and longest fringe seta 6.5× maximum discal width. Marginal vein with four setae, the second (from the wing base) robust and blunt; central setae equal in length. Campaniform sensillum present below second seta, a line joining the sensillum to the fourth seta. Stigmal vein with a row of three campaniform sensilla apically. Mesotibia with two large spines basally; metafemur with spine; metatibia with a row of fine, blunt setae extending almost the entire inner length, increasing abruptly in length at the distal tibia (exact length not visible in Figure 11d since setae positioned almost vertically; a similar row of setae is found in M. momookherjeeae and M. uniclavum, Figure 21j). T1 with smooth area centrally, flanked by two or three longitudinal grooves and a longitudinal cell laterally, extending for the length of the tergum; T1-T4 with very long setae laterally, each longer than its tergum; T2 with a curved row of 6-8 spicules on each side. Ovipositor 1.6× as long as mesotibia.
Mid lobe of mesoscutum (cf Figure 13b) smooth with some irregular longitudinal striate sculpture; propodeum (cf Figure 13a) elongate and curved centrally and posteriorly, crenulae present. T1 without elongate cells laterally; T2-T4 with short setae laterally. Ovipositor 2× as long as mesotibia. Mesotibia with one large spine basally; metafemur with spine. Fore wing (cf Figure 13c) 8× as long as maximum width, maximum distal width equal to maximum basal width; discal setae absent, longest fringe seta 4.7× as long as maximum discal width. Marginal vein with two long setae centrally, approximately equal in length. Stigmal vein moderately enlarged, with three sensilla apically.
Male: Characteristics as for female (Figure 13a-c) (except antenna and genitalia); although, metasoma darker than in female. Antenna as in Figure 12h, with a much shorter C3 compared to the female.  Description. Female: Antenna (Figure 13d) five-segmented (excluding anellus); pedicel as long as funicle; funicle 4× as long as wide; C2 longer than C1. C1 with two prominent dorsal UST, proximal UST almost as long as entire clava; C2 with two MT and an SB, which is only slightly shorter than C2. MPS apparently absent.
Mid lobe of mesoscutum ( Figure 13e) entirely with large, coarse reticulation; propodeum with a rhomboid, laterally arcuate central area, its hind margin truncate, with fine crenulae. T1-T4 largely smooth, with scattered denticles and no setae laterally. T5 and T6 with a pair of long setae centrally. Ovipositor 1.1× as long as mesotibia. Mesotibia without spines basally; metafemur without spine; metatibia with a row of five spines within the distal inner half; a robust spine towards the apex of the outer surface. Fore wing (Figure 13f Mid lobe of mesoscutum ( Figure 14b) with longitudinal striate sculpture extending to scutellum; propodeum with central area extending posteriorly, crenulae present; T1 with one elongate cell or groove laterally, 2-3× as long as wide; T2-T4 without setae laterally; T5 with long setae laterally. Ovipositor 1.5× as long as mesotibia. Mesotibia with one large spine basally; metafemur with spine. Fore wing (Figure 14c) 8.5× as long as maximum width, maximum distal width equal to maximum basal width; the disc with a single long seta;longest fringe seta 5× as long as maximum discal width. Marginal vein with two setae centrally; proximal seta 5-7× as long as distal seta, extending to the end of the marginal vein (in Figure 14c, the distal seta is barely visible in the space between the proximal one and the marginal vein). Stigmal vein moderately enlarged, with two sensilla apically.
Body largely brown, the following paler: legs except coxae and metafemora. Antenna with pedicel pale; scape and C1-C3 darker. Fore wing strongly infuscate basally; stigmal and marginal vein brown; marginal vein very dark centrally.
Male: Largely as in female. C1 and C2 with scattered SS; C2 with 2-3 MT apically; C3 with long apical and ventral UST.   (Figure 14g) five-segmented (excluding anellus), pedicel twice as long as funicle, the latter trapezoid, and slightly longer than wide; C1 slightly shorter than C2; C1 with 2 dorsal UST; C2 with ≥3 MPS 1 SB and a short SS. Mid lobe of mesoscutum smooth, with weakly impressed large subpolygonal sculpture (not visible in paratype which has been strongly macerated). Metanotum and propodeum relatively long centrally, each about half the length of scutellum. Propodeum with short central area, without crenulae. T1 (Figure 15a) sculpture with cells converging centrally, lateral cells 2-3× as long as wide, without denticles (some denticles present on innermost cells). T2-T4 without long setae laterally, each with similar sculpture comprising a central irregular oval cell and elongate lateral cells; those on T3 and T4 divided medially. Ovipositor 1.3× as long as mesotibia. Mesotibia with one large spine basally; metafemur without spine. Fore wing (Figure 15b) 10× as long as wide, maximum distal width 1.2× maximum basal width; the disc with a single irregular row of five setae; longest fringe seta 5.5× maximum discal width. Marginal vein with two long setae centrally.
Mid lobe of mesoscutum and scutellum without apparent sculpture; metanotum and propodeum medially short. Metasoma with a row of microspines on each segment. T1 without cells (Figure 4f). Ovipositor 1.1× as long as mesotibia. Mesotibia without spines basally. Metafemur without prominent spine. Fore wing (Figure 5a) 9× as long as wide, maximum distal width 1.5× maximum basal width; disc with 10 setae irregularly arranged in 1-2 rows; fringe with longest seta 6× maximum discal width. Marginal vein with two long setae centrally, of equal length. Stigmal vein moderately enlarged, with three sensilla apically.
Body brown/yellow. Mesosoma largely pale, but mid lobe of mesoscutum brown anteriorly. Scape and pedicel pale, C1-C3 brown. Fore wing slightly infuscate basally.  Head with toruli separated by about their own width. Mid lobe of mesoscutum smooth; propodeum with straight hind margin, without crenulae; T1 non-reticulate; T2-T4 with short setae laterally. Mesotibia with two large spines basally; metafemur with spine; metatibia with a row of fine, blunt setae extending almost their entire inner length, increasing abruptly in length distally. Fore wing (Figure 5d) about 9× as long as maximum width, maximum distal width 1.3× maximum basal width; discal setae arranged in two rows, each with 4-5 setae, longest fringe seta 10× as long as maximum discal width. Marginal vein with two long setae centrally. Stigmal vein with two sensilla apically. Ovipositor 1.1× as long as mesotibia.
Male: Unknown. Paratypes: same data as holotype (6♀, NHMUK). All specimens are on the same slide; the holotype is circled in red.
Species-group placement: polychaetum-group. The species appears closest to M. giraulti based on morphology.
Distribution: Kenya. Host: Not identified, but possibly Scirtothrips dorsalis (Hood), a species common on tea in Kenya.
DNA data: no DNA sequences. Etymology: Named for our colleague and friend Dr Wanja Kinuthia, National Museums of Kenya, Nairobi.

Megaphragma liui
Polaszek and Fusu sp. nov. (Figure 16a Head with toruli very close together, separated by about one-third their own width. Mid lobe of mesoscutum (Figure 16b) with fine longitudinal striations, but also with distinct large reticulate cells; propodeum medially with strongly produced hind margin, with two crenulae (Figure 16b). T1 smooth centrally, but with 8-10 elongate cells laterally (Figure 16b); T2-T4 with short setae laterally, lateral cells present. Mesotibia without large spines basally, but a robust spine present at the apex of mesofemur; metafemur with spine; metatibia with a group of fine, sharp setae on inner surface apically. Metacoxa and metafemur (Figure 16c) with distinct longitudinal sculpture ventrally, contrasting with transverse sculpture dorsally. Fore wing (Figure 16d) 7× as long as maximum width, maximum distal width 1× maximum basal width; disc distally pointed, without setae (but one wing with a possible indication of a minute seta); longest fringe seta 4× as long as maximum discal width. Marginal vein with two setae centrally, the proximal one very robust, about 1.5× as long as distal. Stigmal vein with one elongate sensillum apically. Ovipositor 1.9× as long as mesotibia.
Body entirely brown, mesosoma pale posteriorly, T1 with pale areas laterally. C1 very dark, pedicel paler than the remainder of the antenna. Fore wing strongly infuscate basally. Legs dark, tarsi pale.
Male: Unknown.  Species-group placement: ghesquierei-group. In the concatenated and partitioned analysis, this species is not included in the group (Figure 2b), though in an unpartitioned analysis with a simple substitution model it is one of the most basal species of the ghesquierei-group (Figure 2a). It is also retrieved as part of the group in the tree based on the 28S sequences alone, where it is sister to M. rivelloi but on a very long branch (Supplementary Figure S3). Hence, morphology, and partly molecular analyses, indicate that our inclusion of the species is correct.

Megaphragma momookherjeeae
Polaszek and Fusu sp. nov. (Figure 17a Mid lobe of mesoscutum smooth; propodeum without distinct central area. T1-T4 largely smooth, T2-T4 with long setae laterally. Ovipositor exserted, exceptionally long for the genus, more than 3× as long as mesotibia (Figure 17b). Mesotibia with a very robust spine basally, 0.4× tibial length; metafemur without spine; metatibia with a row of about 17 spines along almost the entire inner length, and 4 robust spines toward the apex of the outer surface. Fore wing (Figure 17c) 9× as long as maximum width;maximum distal width 1× maximum basal width; disc with a single, minute seta; longest fringe seta 6.5× as long as maximum discal width. Marginal vein apparently with three long setae centrally, of equal length. Stigmal vein with three sensilla apically. Species-group placement: M. antecessor-group. Resembling also the ghesquierei-group in some features (e.g., fore wing with one seta), but clearly not clustering with that group in any molecular analyses. It is recovered as sister to M. antecessor, and the two clustered together as basal to all species-groups except the ghesquierei-group (partitioned analysis), or as the most basal species group of Megaphragma (unpartitioned analysis).
Distribution: Costa Rica. DNA data: CO1: one sequence. Etymology: Named for Mo Mookherjee, a friend of the first author (AP).
Mid lobe of mesoscutum smooth, anteriorly with subpolygonal sculpture (cf Figure 7b). Metanotum and propodeum narrow centrally, the latter without an extension or crenulae. T1 sculpture (Figures 7a and 17f) with cells converging centrally, lateral cells 3× as long as wide, each with 3-5 inward-pointing denticles; T2-T4 without long setae laterally, all with coarse reticulate sculpture becoming lateral distally. Ovipositor 1.7× as long as mesotibia. Mesotibia with two large spines basally; metafemur with spine; metatibiae with a row of fine, blunt setae extending almost their entire inner length, increasing abruptly in length distally. Fore wing (Figure 7e) 9.5× as long as wide, maximum distal width 1.1× maximum basal width; disc with a single row of six setae; longest fringe seta 4.5× maximum discal width. Marginal vein with two long setae centrally.
Body entirely brown, metasoma slightly darker posteriorly. Pleural parts of mesosoma and hind legs except for tarsi lighter. Wings hyaline.
Male: Largely as in female. Antenna with C1 slightly longer than C2; T1 (Figure 7b) with an incomplete pattern of cells. Aedeagus as in Figure 7f.  (Figure 18a) five-segmented (excluding anellus); pedicel almost twice as long as funicle; funicle slightly longer than wide; C1 shorter than C2; C1 with two dorsal UST; C2 with three elongate MPS extending beyond clava apex.
Mid lobe of mesoscutum and scutellum smooth. Metanotum and propodeum narrow centrally, the latter without an extension or crenulae.
Mesotibia without large spines basally; metafemur without spine; metatibia with a row of robust spines extending along the inner surface of distal half.
Fore wing 9× as long as wide, maximum distal width 1.3× maximum basal width; disc with a single row of six setae, and longest fringe seta 5× maximum discal width. Marginal vein with two long setae centrally, subequal in length (cf Figure 18g). T1 sculpture with cells converging centrally, six lateral cells present, 2-3× as long as wide, each with 1-3 inward-pointing denticles; T2-T4 with lateral cells indicated, with lateral setae not detected (cf Figure 18f). Ovipositor as long as mesotibia. Body largely brown, mesosoma largely pale. T1 very dark brown in contrast to rest of body. Legs pale. Wings hyaline.
Male: Largely as in female. Antenna with C1 longer than C2; T1 with an incomplete pattern of cells. Genitalia tubular as in other species, 3× as long as wide.  Species-group placement: mymaripenne-group. Distribution: UK (England); Czech Republic, Hungary. DNA data: CO1: 21 sequences from 2 countries; 28S: 23 sequences from 2 countries: Hungary, UK.
Etymology: Named for Dr John Noyes of the Natural History Museum, London, for his outstanding contribution to our knowledge of Chalcidoidea. As a chalcid collector, John is unmatched so far, and a major proportion of the material in this study was collected by him.
Mid lobe of mesoscutum (Figure 18c) smooth, with 5-6 deep striae anteriorly; propodeum with straight hind margin, without crenulae. T1 non-reticulate; T2-T4 with short setae laterally. Ovipositor 1.1× as long as the mesotibia. Metafemur without spine but a prominent seta present. Fore wing (Figure 18d) 10× as long as maximum width, and maximum distal width 1.4× maximum basal width; discal setae arranged in 2-3 rows, each with 3-5 setae, longest fringe seta 5.5× as long as maximum discal width. Marginal vein with two long setae centrally. Stigmal vein with two sensilla apically. Species-group placement: polychaetum-group. Distribution: Colombia, New Zealand. DNA data: no DNA sequences. Etymology: Named for Emeritus Professor John Pinto, formerly of University of California, Riverside, in recognition of his monumental contribution to our understanding of Trichogrammatidae.
Comments: Known so far from only two specimens to date; this species has an apparently extraordinary distribution, being known from Colombia and New Zealand. It seems very likely that it will turn up elsewhere and is probably another cosmopolitan Megaphragma species.
Mid lobe of mesoscutum and scutellum smooth (Figures 18f and 19d). Propodeum with a distinct central area with lateral boundaries in line with those of mesoscutum and scutellum ( Figure 19d); two lateral lobes present behind propodeum central area ( Figure 19d); propodeum without crenulae. T1 sculpture (Figures 18f and 20b) with cells converging centrally, about six lateral cells, 2× as long as wide, mesal cells each with 2-3 inwardpointing denticles. T2-T4 with 2-3 lateral cells, with short setae laterally. Ovipositor 1.3× as long as mesotibia. Mesotibia with two large spines basally; metafemur without spine; metatibia with a row of fine, blunt setae extending almost its entire inner length, increasing abruptly in length distally. Fore wing (Figure 18g) 9.5× as long as wide, maximum distal width 1.4× maximum basal width; disc with a single row of six setae, longest seta of fringe 5.8× maximum disc width. Marginal vein with two long setae centrally.
Antenna (Figure 18e) with radicle brown, very dark compared to the remainder of the antenna; remainder of body largely brown, mesosoma largely pale, mid lobe of mesoscutum brown anteriorly, an indistinct brown spot on the scutellum. Legs pale. Wings hyaline.
Male: Largely as in female. Antenna with C1 slightly longer than C2; T1 with an incomplete pattern of cells. Genitalia tubular as in other species, 3× as long as wide.
DNA data: CO1: five sequences; 28S: five sequences; all from Italy. Etymology: Named for our colleague Alexey Polilov, co-author of this paper, for his outstanding contribution to our knowledge of the Trichogrammatidae and miniaturization in insects.
Comments: Found in Italy at the same locality and on the same host as M. viggianii.    Description. Female: Antenna (Figure 21d) five-segmented (excluding anellus); funicle absent; hence, clava three-segmented, with C1 and C2 almost fused; C1 without UST; C2 with one elongate UST, reaching more than half the length of C3; C3 with MPS, SB, and SS.

Megaphragma rivelloi
Mid lobe of mesoscutum with longitudinal striate sculpture extending to scutellum (cf Figure 14b); propodeum with central area extended posteriorly, crenulae present (cf Figure 15e). T2-T4 without setae laterally. Ovipositor 1.7× as long as mesotibia. Mesotibia with one large spine basally; metafemur with spine (cf Figure 16c, upper). Fore wing 6× as long as the maximum width, maximum distal width 1.2× maximum basal width; disc with a single long seta (Figure 21e), longest fringe seta 3× maximum discal width. Marginal vein with two setae centrally, equal in length, extending to the end of the marginal vein. Stigmal vein moderately enlarged, with two sensilla apically.
Male: Largely as in female. C1 and C2 with scattered SS; C2 with 2-3 MT apically; C3 with long apical and ventral UST. C3 is darker than preceding segments.  Mid lobe of mesoscutum with longitudinal striate sculpture (cf Figure 14b); propodeum with central area extended posteriorly, crenulae absent (cf Figure 21l). T1 with elongate cells laterally, 2-3× as long as wide (cf Figure 12b); T2-T4 with a short, robust seta near lateral margin; T5 centrally with subparallel striations and with long setae laterally (Figure 21h). Ovipositor 1.7× as long as the mesotibia. Mesotibia with one large spine basally; metafemur with spine. Fore wing 8.5× as long as the maximum width, maximum distal width equal to maximum basal width; disc with a single long seta (cf Figure 12c), longest fringe seta 3.5× as long as maximum discal width. Marginal vein with one long seta centrally, extending to the end of the marginal vein. Stigmal vein moderately enlarged, with four sensilla apically.
Male Head with few features discernible due to mounting position, and extensively obscured by eye pigment. Mid lobe of mesoscutum with fine longitudinal striation (cf Figure 14b); vertical/ventral anterior mid lobe of mesoscutum with coarse, reticulate sculpture (cf Figure 3g); propodeum elongate centrally, longitudinally striate with three large crenulae. T2-T4 with very long setae laterally, each longer than its tergum; T3 and T4 with large irregular cells laterally. Mesotibia with two large spines basally; metafemur with spine; metatibia with a row of fine, elongate setae extending almost their entire inner length (Figure 21j). Fore wing 5× as long as maximum width; longest fringe seta 6.7× as long as maximum discal width, maximum distal width versus maximum basal width is unclear; disc (cf Figure 11b Species-group placement: antecessor-group. Megaphragma uniclavum is so far unique for the genus having a single claval segment. Despite the very unusual fore-wing structure, it is clearly affiliated with M. antecessor (similar mesoscutal sculpture and row of setae on metatibia).
Distribution: Costa Rica. DNA data: no DNA sequences.

Megaphragma vanlentereni
Mid lobe of mesoscutum ( Figure 22b) posteriorly smooth, anteriorly with large, coarse reticulation; propodeum with broad, truncate hind margin, with two widely separated lobes distally and laterally, without crenulae. T1 with a central "V" composed of minute denticles, a row of coarser denticles laterally (Figure 20c), and one elongate cell laterally, about 2× as long as wide. Metasoma dorsally with rows of denticles laterally on T2 and T3; T2-T4 with moderately long setae laterally. Ovipositor as long as the mesotibia.
Mesotibia without spines basally, a single seta present; metafemur without spine but with three robust setae; metatibia with a row of five or six fine setae extending along half its inner length. Fore wing (Figure 22c) 6× as long as maximum width; maximum distal width 1.5× maximum basal width and more than 2× width measured at the apex of the marginal vein; longest fringe seta 4× as long as the maximum discal width; disc with setae in one or two rows of 4-6 setae, and setae on ventral surface long, the penultimate one reaching to the base of the distal (Figure 22d). Marginal vein with two setae centrally, the proximal approximately 2× the length of the distal. Stigmal vein has four sensilla apically.
Head and metasoma dark brown; central/posterior mesoscutal mid lobe and lateral scutellum paler. Remainder of body, including legs, pale. Flagellum darker than remainder; wings hyaline.
Male: Antenna with pedicel almost as long as scape; funicle with two small setae; clava two-segmented; C1 >2× C2; C1 with three setae; C2 with three flagelliform setae; two multiporous plate sensilla and a short terminal process present. Metasoma dorsally with rows of denticles laterally on T1-T3. One individual with funicle apparently fused with C1.
Comments: Found in Italy at the same locality and on the same host as M. polilovi. According to the CO1 sequence of the mitochondrial genome deposited on GenBank by Nedoluzhko et al. [23], their species is not M. amalphitanum (=M. longiciliatum) but M. viggianii. Our molecular analysis retrieves the two as sister species. Both have the mid lobe of the mesoscutum reticulate anteriorly, and long lateral setae on T2-T4, though these are longer in the former. In M. longiciliatum T2-T4 have very long setae, at least as long as the tergite (Figure 5h), whereas in M. viggianii T2-T4 have shorter setae, much shorter than the tergite (cf Figure 8f).

Discussion
The present study is the culmination of more than 10 years of intensive collecting and examination of several thousand Megaphragma specimens from all over the world, including type material of all but one of the previously described species. Without the molecular dimension, our conclusions would have been very different. For example, the separation of M. noyesi and M. polilovi from M. mymaripenne would not have been possible, and these species have been confused in the past. Within the ghesquierei-group, morphological differences between species that are very distinct based on DNA, are completely undetectable in many cases. Some of this is no doubt due to the limitations of light microscopy, even when using techniques such as Nomarski differential interference contrast (DIC), coupled with focus stacking. Morphological evolution at species level is apparent for several structures, most notably the antenna, but also the setae and spines on the middle and hind legs, the fore wing, the structure of the propodeum, and that of metasomal terga showing variation in microsculpture and chaetotaxy (of diagnostic value). Species can be relatively easily grouped based on features of these morphological characters, and the ghesquierei-group, in particular, can be defined based on features of three characters: fusion of the funicle with the clava, development of the central propodeum, and metafemoral spine. The mymaripenne-group is less easily defined, with the loss of the otherwise characteristic sculpture of T1 having occurred in the longiciliatum subgroup. The shape of the fore wing, and its discal and marginal ciliation appear to be critical in reflecting species evolution in this genus. Future studies should carefully assess the setation of the upper and underside of the fore wing, something that is very difficult once the specimen has been slide-mounted. No doubt future studies, including more scanning electron microscope imaging, will reveal additional patterns of morphological variation in the genus.
Perhaps the most surprising discovery of this study is the extraordinary distribution of some species. Megaphragma longiciliatum, under our new and broader definition, is found from Southeast Asia to Northwest Europe, as well as in North America, the Congo, and the Middle East. In this respect, as well as in terms of their physical size, the Megaphragma species parallel some Protozoa. The phrase "everything is everywhere, but the environment selects" [69], originally applied to Protozoa, certainly seems to apply to several Megaphragma species. Previous theories attempting to explain ubiquitous distributions of particular species of organisms have attributed this to their large population sizes, rather than to any inherent properties of such groups [70]. This argument appears to be so entirely back to front (i.e., "some species are cosmopolitan because they have huge populations") that it can be easily dismissed. It is precisely the inherent property of minuteness, among other attributes discussed below, that is the main reason for these species having cosmopolitan distributions. Minuteness is directly related to dispersive ability, which can be largely passive for minute organisms; although, Megaphragma are known to be good at directional flight [28]. Our study, despite being very patchy in terms of the sample sizes of most species, suggests a mixed pattern of dispersal and distribution, with ubiquitous species as well as apparently endemic ones, as shown previously for about 200 Protozoa species [71]. Minuteness also directly affects the relationship between the species and its immediate microhabitat. We can assume that for an organism whose adults are around one-quarter of a millimeter, and whose developmental stages are entirely within a closed environment (the thrips egg), the macroecological factors of climate and temperature are less important, at least for some species. Hence, e.g., M. rivelloi appears to be as suited to the humid rainforests of SE Asia as it is to the much drier countryside of Southern England, and the same must be true for other species. Humidity, and especially avoidance of desiccation, are critical for the survival of minute terrestrial organisms, and Megaphragma species are known to have very thin cuticle [18] (see also Supplementary Figure S1). Even if air masses could transport minute hymenopterans quickly across the globe, desiccation would be a major impediment [72]. However, given that for most of their adult life Megaphragma species are likely to be in close proximity to living plant tissue, and hence access to moisture, it is probable that these immediate microclimatic conditions override the macroecological conditions already mentioned, and hence some of the remarkable latitudinal distributions of, e.g., M. longiciliatum, M. pintoi, and M. rivelloi. As well as both direct and passive colonization of new geographical areas, human movement of plant material containing thrips eggs, and Megaphragma life stages is inevitable, and has certainly contributed to a large extent to the cosmopolitan distribution of some species. Adaptation to, and dependence on, microhabitat conditions, especially humidity, have undoubtedly also contributed to their human-mediated distribution through the movement of plant material. Another explanation for species of egg parasitoids having extremely wide distributions has been their assumed defencelessness, and the relative uniformity of their hosts [73]. Whilst the former appears to be generally true, the latter clearly is not. Parasitoids that parasitize eggs are in some cases able to additionally parasitize Lepidoptera larvae, braconid cocoons, and even act as hyperparasitoids of other egg parasitoids. This extreme range of hosts has been reliably documented in Centrodora darwini (Girault) [74]. Other egg parasitoids appear to be extremely specific in which species they will either attack and/or develop successfully on, and this is particularly true of many species in the hyperdiverse genus Telenomus (e.g., [75]). Thus, a huge range of levels of host species specificity exists across egg parasitoids. In several taxa, eggs are entirely free of any parasitoids (aphids, whiteflies, and scale insects), while their (relatively) close relatives (leafhoppers and their relatives) are very heavily parasitized [76]. Clearly, the ability to evade or resist being parasitized at the egg stage is highly heterogeneous across the insects. In the case of Megaphragma, it is difficult to tell how host specific they are; in the few cases where there are rearing records from more than one host, they belong to different genera of the same family. Most likely as in other groups of parasitoid wasps, members of Megaphragma are a mix of generalists and specialist species [77].
As stated above, in several lineages Megaphragma has diversified to produce numerous cryptic species, many of which appear indistinguishable morphologically, at least using the techniques employed in this study. All future studies of these and similar organisms must rely to an extent on DNA data, and it may be that species will be described solely on differences in DNA where these differences can be demonstrated to be reasonably accurate proxies for biological species distinctions.

Informed Consent Statement: Not applicable.
Data Availability Statement: All sequences were deposited in GenBank. Voucher specimens are deposited in the collections listed under Specimens and depository abbreviations. All data and materials will be publicly available.