A new species of Mesopolobus (Hymenoptera, Pteromalidae) from black locust crops

A new species of the genus Mesopolobus Westwood, 1833, Mesopolobus robiniae Lakatos & László sp. nov., is described and illustrated from east-central Europe (Romania and Hungary). The species was reared from black locust (Robinia pseudoacacia) seedpod samples, where it most likely parasitizes the black locust’s seed predator Bruchophagus robiniae Zerova, 1970. Here we present the new species and report on its ecological relationships within the European seed predator community of black locust. We also give details regarding type material and type locality, a detailed description with images, a differential diagnosis of the new species, and a modifi cation to the identifi cation key published by Graham (1969), that distinguishes this new species from closely related species. In addition, we provide information on the distribution, biology and results of barcoding analysis. We also provide the DNA sequence data to complement the morphological taxonomy.


Introduction
In the last century, the black locust (Robinia pseudoacacia L.) became a characteristic component feature of the Central and Eastern European landscape ). Its positive economic, but negative environmental impacts led to confl icts between nature conservationists, forestry workers, urban planning experts, beekeepers and the public (Benesperi et al. 2012;Dickie et al. 2014;Sádlo et al. When replacing native vegetation, the black locust reduces local biodiversity (Hanzelka & Reif 2015). Endangered light-demanding plants and invertebrates are threatened by its appearance through reducing light to plants growing beneath the canopy and above the forest fl oor, and changing the microclimate and soil quality (Lazzaro et al. 2018). These impacts can have effects throughout the food chain, by depriving birds of their insect prey, which depend on the plants that have been wiped out by the black locust (Hanzelka & Reif 2015). One of the central problems regarding black locust colonization is its capacity to rapidly increase soil nutrient concentration and to alter soil chemical properties which conditions then facilitate invasion by other non-native nitrophilous plant species (Enescu & Dănescu 2013).
The identifi cation of Mesopolobus species emerged from black locust crops was based on the most detailed identifi cation key up to date provided by Graham (1969). Using Graham's keys a number of characters (fore wing marginal vein length ratio to stigmal vein length, number of anelli and funicular segments, position of toruli to anterior margin of clypeus and to median ocellus, pilosity of the basal cell of fore wing, position of hypopygium tip along the metasoma) led us to key couplet 16 (Graham 1969: 643), where based on two character combinations, namely metasoma length ratio to head plus mesosoma length and metasoma breadth, which did not match our specimens. This suggested the specimens reared from the black locust pods were not represented in the keys, and were likely undescribed. We thus studied several Mesopolobus species represented in the keys and compared them morphometrically to the Mesopolobus females emerged from black locust seed pods. This approach provides robust insight into Mesopolobus morphology, which may play a major role in resolving the species delimitations in biocontrol studies. Complementing the morphometric study, we also analyzed mtCOI sequences of the emerged Mesopolobus females from black locust pods, and compared them to the available mtCOI sequences from the BOLD System and NCBI databases.
Our objectives were the following: i) to identify those morphometric characters that give the best discrimination of the females emerged from black locust seedpods from other Mesopolobus species. ii) to calculate the genetic distance values between the mtCOI sequence of the females emerged from black locust and the other Mesopolobus species. iii) to describe the females and the males of the species that emerged from black locust seedpods.

Material and methods
To gather information about black locust seedpod insect inhabitants we collected seedpod samples in black locust plantations and patches for four years, in the early spring of 2009 in Romania and between 2013-2015 in Romania and Hungary (Table 1). Samples were placed in plastic cups, containing 20-100 seedpods and covered with punched plastic wrap. Samples were kept in a covered balcony with a temperature and humidity close to outdoors at Babeș-Bolyai University (Cluj-Napoca, Romania) and at University of Debrecen (Debrecen, Hungary). Emerged individuals were monitored and collected monthly from seedpod samples for a year, and stored in 70% ethanol. The dominant emerging species were the seed predator of black locust seeds, Bruchophagus robiniae, and its parasitoid, the undescribed Mesopolobus species (Lakatos et al. 2018).
Identifi cation and description of the emerged Mesopolobus species have been made under an Olympus SZ51 binocular microscope, with an 80× magnifi cation and LED lighting. Images were produced by a Canon EOS 600D and a Canon EF 100 mm f/2.8 USM Macro Lens. Morphological nomenclature follows Graham (1969). The provided identifi cation key is modifi ed from the keys to genus Mesopolobus of Graham (1969). Type material is deposited in the Museum of Zoology, Babeş-Bolyai University, Cluj-Napoca (MZBBU). Specimen identifi cation codes: holotype-MZBBU HYM000011; 14 paratypes-MZBBU HYM000012 to HYM000025, measured female specimens: HYM000026 to HYM000039.
One female specimen of M. longicollis Graham, 1969 was measured using ImageJ from photographs of the type provided by Oxford University Museum of Natural History.
We measured 19 morphometric variables, corresponding to those used in the taxonomy of Pteromalidae for calculating typically used ratios (e.g., Graham, 1969) (Table 2), on a total of 55 dry-mounted Mesopolobus females belonging to the new species (Supp. fi le 1). Measurements were made with an Olympus SZ51 stereo microscope (objective: 110AL2X; eyepiece: WHSZ10X) under 60× and 80× magnifi cation using a calibrated eye-piece micrometer (2.5 mm subdivided into 100 units). For all measurements we ensured that the points of reference were equidistant from the objective of the microscope.
Body ratios of Mesopolobus female specimens were analysed using the Multivariate Ratio Analysis (MRA) tool (Baur & Leuenberger 2011). Variation structure of Mesopolobus specimens was analysed by PCA in shape space to identify the principal components accounting for the variation. For the visualization of each character's contribution we used PCA ratio spectrum. Body ratios with best discriminant power were determined using the LDA ratio extractor (Baur & Leuenberger 2011). Analyses were made with R statistical software ver. 3.6.3 (R Core Team 2020).
Genomic DNA was extracted from three individuals using DNeasy Blood and Tissue kits (Qiagen Inc., Valencia, CA), following the protocol provided by the manufacturer. Mitochondrial cytochrome c oxidase subunit I (COI) sequences were amplifi ed using the standard LCO1490 and HCO2198 primer pair (Folmer et al. 1994) in a 50 μl reaction volume at a 45°C annealing temperature. PCR products were purifi ed with the Wizard SV Gel and PCR Clean-Up System (Promega, USA) and sent for sequencing to Macrogen Inc. (Korea).
Sequences were downloaded and verifi ed with the Basic Local Alignment Search Tool (BLAST) (Johnson et al. 2008). Further, sequences for all available Mesopolobus species were also downloaded from the NCBI database and the BOLD System (for reference numbers see Fig. 2). The sequences were aligned using a Clustal W algorithm (Thompson et al. 1994) in BioEdit (Hall 1999). A phylogenetic tree was inferred in MrBayes (Ronquist et al. 2012), assuming a GTR+G+I model. Interspecifi c p-distances were calculated in MEGA X (Kumar et al. 2018).  (Graham 1969), distance between anterior and posterior margin of the head, measured laterally 60 × ltg.b Seventh gastral tergite breadth Greatest breadth of the seventh gastral tergite, greatest distance between the outermost lateral edges of the seventh gastral tergite

Multivariate Ratio Analysis of variation in body size and shape
We fi rst performed a series of shape PCAs on all specimens based on 19 morphometric characters. We identifi ed principal components contributing to morphometric variation of all Mesopolobus females without prior species-determination by applying the PCA in isometry free shape space function to all specimens as a single group. Then we applied the PCA in isometry free shape space only to the group of females which were closest to those emerged from black locust seedpods. When we included all females in shape space, PC1 and PC2 accounted for 58% of the variation of the entire sampled population. When analysing only those species pairs which were closest to our target group in shape space, PC1 and PC2 accounted for 61% and 74% of the variation respectively. The fi rst principal components are congruent with the separation of species, although a clear cut between the clusters could not be established (Fig. 1).
On the fi rst scatterplot (Fig. 1a)  The PCA ratio spectrum for the species pair M. amaenus and M. robiniae sp. nov. (Fig. 2a) identifi ed ltg.l at the extreme high end, and stv.l at the extreme low end of the spectrum. These characters were also found to contribute to species discrimination. The allometry ratio spectrum for the fi rst species pair was dominated almost by the same ratio, stv.l and ltg.l (Fig. 2b), which is also the most important ratio concerning the fi rst shape PC which shows to be the most allometric one. The PCA ratio spectrum for the species pair M. fasciiventris and M. robiniae sp. nov. (Fig. 2c) identifi ed pcl.l at the extreme high end, while stv.l at the extreme low end of the spectrum. These characters, except for stv.l, were found to contribute to species discrimination. The allometry ratio spectrum for the second species pair was dominated almost by the same ratio, pcl.l and ltg.b (Fig. 2d), that is not the most important ratio concerning the fi rst shape PC.
For the species pair M. amaenus and M. robiniae sp. nov. the LDA ratio extractor identifi ed stv.l/lgt.l and hea.l/stv.l as the fi rst two best discriminating ratios. These two combined ratios successfully separated the two species (Fig. 3a) hea.l/stv.l (0.11-1.12, 0.93, 0.13) vs (1.11-1.53, 1.26, 0.11). This suggests that the two species can be separated when judgement is based on a series of individuals. Further, the calculated D.shape is much higher than D.size in all of the two best discriminative ratios, indicating that species are mostly separated by differences in shape of characters (Table 3).
For the species pair M. fasciiventris and M. robiniae sp. nov. the LDA ratio extractor identifi ed pcl.l/mav.l and clv.l/hea.l as the fi rst two best discriminating ratios. These two combined ratios successfully separated the two species (Fig. 3b). The calculated ratios for the LDA-suggested characters (M. fasciiventris vs M. robiniae sp. nov., range, mean, sd) are: pcl.l3/mav.l (0.07-0.10, 0.08, 0.02) vs (0.11-0.20, 0.15, 0.05); clv.l/hea.l (0.36-0.66, 0.50, 0.14) vs (0.52-0.68, 0.63, 0.04). This suggests that the two species can be separated when judgement is based on a series of individuals. Further, the calculated D.shape is much higher than D.size in all of the two best discriminative ratios, indicating that species are mostly separated by differences in shape of characters (

Molecular species delimitation
Based on molecular analysis of the available samples, Mesopolobus robiniae sp. nov. is placed closest to Mesopolobus verditer (Norton 1868) (Fig. 4). The three individuals represented only one haplotype (653 bp) that was deposited in GenBank with the MF098549 accession number. The alignment of the downloaded sequences was 468 bp long and consisted of 3 Pteromalus species (used as outgroup) and 14 Mesopolobus species, including the one described in this paper.
The phylogenetic relationship between the species is unresolved based on the available COI sequence data, but the tree shows a well-supported differentiation (PP = 1) of the new species, with M. verditer  as the closest species (Fig. 4). The differentiation is also supported by the p-distance values with a minimum of 12.5% between M. robiniae sp. nov. and M. verditer, and a maximum of 16.52% between M. robiniae sp. nov. and M. tibialis (Table 4). Diagnosis Mesopolobus robiniae sp. nov. is characterized by having the following morphological characters: head with uniform reticulation, antennae inserted at or above ventral edge of compound eyes, with three anelli, head broader than mesoscutum, pronotal collar moderately long, marginal vein of fore wing about twice as long as the stigma vein, metasoma longer than mesosoma, with only slightly projecting ovipositor sheaths, body green, antennae proximally testaceous with funiculus and clava infuscate, femora and tibiae testaceous, with hyaline wings. Mesopolobus robiniae sp. nov. differs from closely resembling species by the ratios of metasoma to head plus mesosoma length, of temples to eye length, of marginal to stigma vein, of pronotal collar to mesoscutum length and of antennal clava to head length. Molecular results: the three sequenced individuals represented one haplotype (GenBank accession number: MF098549); based on the BI tree the new species, shows a well-supported differentiation from M. verditer as the closest species and from other congenerics with the maximal differentiation from M. tibialis.

Etymology
The new Mesopolobus species is named after the host plant of its seed predator host, the black locust (Robinia pseudoacacia). COLORATION. Body green, sometimes with golden refl ections; metasoma bronze-black distally, some of the tergites occasionally with blue or violet fl ecks. Coloration of antennae: scape, pedicellus and anelli testaceous, sometimes last anellus infuscate, all funicular segments and clava infuscate, occasionally brown. Coxae concolorous with the mesosoma, femora and tibiae testaceous, the tips of the fi fth tarsi fuscous to black. Tegulae hyaline, usually slightly yellow posteriorly. Wings hyaline; venation pale yellow.

Material examined
HEAD. 1.1 (range 1.02-1.18) times as broad as mesoscutum; in dorsal view 2.25 (2.07-2.52) times as broad as long, with temples rounded off and between one third and one fourth as long as eyes; the distance between posterior ocelli (POL) 2.11 (1.75-2.80) times oculo-ocellar distance (OOL). Head in front view suboval with the genae moderately buccate. Eyes separated about 1.59 (1.18-1.74) times their length. Malar space more than half (0.68 (0.55-0.76)) the length of an eye. Breadth of oral fossa 1.93 (1.69-2.36) times malar space. Clypeus strigose, its anterior margin moderately emarginate. Head uniformly and moderately reticulate. Antennae inserted low on head, lower edge of toruli at or hardly above level of ventral edge of eyes; distance between clypeal margin and toruli 0.69 (0.54-0.8) times the distance between median ocellus and toruli. Scape length 1.23 (1.09-1.4) times eye length, scape almost reaching lower edge of median ocellus; combined length of pedicellus and fl agellum 0.87 (0.76-0.96) times breadth of head; pedicellus (profi le) 2.06 (0.75-2.5) times as long as broad, about as long as anelli plus fi rst funicular segment; fl agellum rather weakly clavate, proximally as stout as or slightly stouter than pedicellus; fi rst and second anelli short, twice or slightly more than twice as broad as long, third anellus longer than previous anellus and about 1.5 times as broad as long; funicular segments subquadrate, proximal ones sometimes slightly longer than broad, distal ones occasionally very slightly transverse; clava 1.9 (1.5-2.29) times as long as broad, 0.83 (0.66-1.15) as long as three preceding funicular segments together; sensilla in one row on each segment, sparse on funicle, more numerous on the clava.
MESOSOMA. 1.52 (1.38-1.74) times as long as broad. Pronotal collar moderately long medially, 0.21 (0.16-0.26) times (one sixth to one fi fth) as long as mesoscutum, and much longer at sides, strongly and coarsely reticulate, clearly margined. Mesoscutum 1.58 (1.28-1.82) times as broad as long, rather coarsely reticulate discally, more fi nely laterally, without piliferous punctures. Scutellum 0.9 (0.82-0.94) times as broad as long, moderately convex, fi nely reticulate, frenum slightly more coarsely reticulate. Axillae fi nely reticulate. Dorsellum a narrow, alutaceous transverse crest separated from scutellum by simple suture. Propodeum medially slightly less than half (0.41 (0.36-0.48)) as long as scutellum; median area 2.39 (2-3) times as broad as long, well-defi ned laterally, plicae distinct throughout and sharp over at least their distal half; median carina distinct, straight; panels of median area fi nely, slightly irregularly reticulate; nucha transversely aciculate, separated from median area by impressed line; posterior foveae, at sides of nucha, moderately deep; spiracles oval, longer than broad, separated by nearly half their length from metanotum. Postspiracular sclerite broad, shiny, weakly and irregularly sculptured. Mesepisternum moderately fi nely reticulate, its upper triangular area smooth; mesepimeron rather more coarsely reticulate than mesepisternum, metapleuron smooth. Legs rather short; femora rather stout; mid tibiae fairly slender, 7.44 (4.88-9) times as long as their maximum breadth. Fore wing rather broad; costal cell fairly broad, its upper surface bare, lower surface with a complete row of hairs and some additional hairs scattered over distal third to half; basal cell bare, open below; basal vein bare or with one to three hairs; speculum open below, on upper surface of wing extending below proximal end of the marginal vein; surface beyond speculum thickly pilose; marginal vein 2.19 (2-2.47) times as long as stigmal vein; postmarginal vein shorter than marginal, 0.73 (0.63-0.81) times as long as marginal.

Morphological comparison
Females of Mesopolobus robiniae sp. nov. were not identifi able based on Graham's keys (Graham 1969), but several morphologically and morphometrically related species were found for which the differing characters will be enumerated in the order that the species appear in Graham's key. The species M. robiniae sp. nov. has a shorter metasoma compared to head plus mesosoma in M. maculicornis. The species M. jucundus (Walker, 1834) has a curved stigmal vein compared to M. robiniae sp. nov. Mesopolobus robiniae sp. nov. differs from M. fasciiventris by its males having 3 anelli and 5 funicular segments while in the latter there are 2 anelli and 6 segments. Females of M. robiniae sp. nov. differ from those of M. fasciiventris in the ratios of pcl.l3/mav.l and clv.l/ hea.l (for abbreviations see Table 2, for differences Table 3). The head of M. apicalis in dorsal view has temples nearly three quarters as long as the eyes, while M. robiniae sp. nov. has its head in dorsal view with temples appearing one quarter to one third as long as the eyes. The metasoma of M. amaenus is less than twice as long as broad and almost as long as the mesosoma, while in the case of M. robiniae sp. nov. the metasoma is not less than twice as long as broad, but it is as long as the mesosoma. The species M. longicollis has the pronotal collar 1/7 to 1/6 as long as the mesoscutum and its metasoma is less than twice as long as broad compared to M. robiniae sp. nov. The species M. diffi nis and M. meditteraneus differ from M. robiniae sp. nov. because the latter has longer marginal vein (1.4 to 1.6) than length of the stigmal vein.
The species M. verditer is not present in the keys of Graham (1969) because it has a North American distribution. It differs from M. robiniae sp. nov. in the following: antennal funicle segments shorter than their breadth, while in M. robiniae sp. nov. they are at least as long as their breadth. The ratio of the stigmal vein to the last gastral tergite length is 1.91-2.50 in M. verditer, while in M. robiniae sp. nov. is between 0.08-1.15. Mesopolobus sericeus differs from M. robiniae sp. nov. fi rst by having 2 anelli and 6 funicular segments, but also in having the ratio of the stigmal vein to the last gastral tergite length 1.41, while in the other species this ratio is smaller (0.8-1.15). In M. typographi the median area of the propodeum is 1.75-2 times as broad as long (Graham 1969) while in M. robiniae sp. nov. is 0.82-0.94 times as broad as long (N = 15).
Based on von Rosen's key (von Rosen 1958), the morphological identifi cation of specimens led us to M. mediterraneus as the closest species, from which females of M. robiniae sp. nov. differs, apart from the previously mentioned longer marginal than stigmal vein, in also having a longer pronotal collar and a shorter metasoma than the combined length of head and mesosoma. Gahan (1932: 739) (Noyes 2020). Mesopolobus verditer is distributed in the Nearctic and Germany (Thompson 1958). Moreover, M. verditer differs from M. robiniae sp. nov. in having a reticulated middle area of propodeum and oblique wrinkles, as does also from M. amaenus and M. longicollis (von Rosen 1958).
We propose the following update to the key of Mesopolobus species of Graham (1969)

Biology
Based on our rearing, M. robiniae sp. nov. seems to be an early fl ying parasitoid species. Individuals of the species emerged during spring consequently in all study years. Our black locust seedpod samples were collected mostly in March, and the emergence peak of M. robiniae sp. nov. was in April, with a decrease in May. After May, we rarely encountered any individuals of this parasitoid species.
The host of M. robiniae sp. nov. may be Bruchophagus robiniae but there is no information regarding the host plant of B. robiniae before the introduction of black locust. Another possibility is that M. robiniae sp. nov. initially had another host, but switched from it to B. robiniae. Either possibility is plausible; before 1970 (Zerova 1970) the species B. robiniae was not known, and M. robiniae sp. nov. was not described until now. The parasitoid community of black locust is understudied, and the available literature makes no mention of parasitoids in this community (Farkas & Terpó-Pomogyi 1974;Perju 1998), with the exception of our ecological study concerning the seed-predator community of black locust in Eastern Europe (Lakatos et al. 2016).

Discussion
The multivariate ratio analysis (MRA) and the mtDNA sequence analysis resulted in the successful separation of the Mesopolobus species emerging from black locust seedpods from the other congeneric relatives. The morphometry-based shape PCA helped us identify which species fall closer to the specimens emerged from black locust crops. This delimitation was important since the available specifi c keys (Graham 1969) did not lead us to a closest relative based on the combination of morphology and morphometry.
In a PCA ratio spectrum, only ratios calculated with variables lying at the opposite ends of the spectrum are relevant for a particular shape PC and the most allometric ratios are also found at the opposite ends of the allometry ratio spectrum (Baur et al. 2014) we apply multivariate ratio analysis (MRA). The PCA ratio spectrum and the allometry ratio spectrum plots revealed a large (M. amaenus and M. robiniae sp. nov.) and moderate (M. fasciiventris and M. robiniae sp. nov.) amount of allometric variation in the identifi ed discriminating morphometric character pairs (Fig. 2). However, this is not of concern in our case, because on the one hand the species we found to be closely related based on morphometry were clearly separated based on the molecular results, and the combinations of the usually used ratios do not overlap with species in the keys of Graham (1969), since there is no possibility to progress beyond key couplet 16. On the other hand, the ratios found with the LDA ratio extractor tool have small D.size values compared to D.shape values ( Table 3) which means that separation was mainly due to shape rather than size. The LDA ratio extractor tool found that the species pairs could be separated without overlapping based on the fi rst ratio pairs. These ratios in combination with morphologic characters gave a confi dent separation of the closely related species.
The origin of M. robiniae sp. nov. is as yet unknown, since it has to be a host shifting species. Black locust was introduced to Europe 300 years ago, and in its native area it has no Bruchophagus seed predator, nor the associated parasitoids (Stone 2009). So, the new Mesopolobus species may not be monophagous on B. robiniae, which is similarly a host-shifting seed predator. Nonetheless, it is befi tting of the name robiniae, since parasitoids are also affected by the host plant of their herbivorous host. As part of their host fi nding strategy, parasitoids may search for a specifi c plant or plant part (as seedpods) housing any potential herbivorous host species (Cronin & Abrahamson 2001).