Revision of the West Palaearctic Euura bergmanni and oligospila groups (Hymenoptera, Tenthredinidae)

Eight Western Palaearctic Euura species are here assigned to the bergmanni group (bergmanni, brevivalvis, dispar, glutinosae, leptocephalus, respondens, sylvestris, and viridis) and two species to the oligospila group (frenalis and oligospila). Euura pallens (Konow, 1903) (bergmanni group) is removed from the list of West Palaearctic taxa. Euura pyramidalis (Hellén, 1948) is treated as incertae sedis within the bergmanni group. Definitions of the bergmanni and oligospila groups are primarily based on genetic sequence data (mitochondrial COI and nuclear NaK and POL2). We report likely occurrence of heteroplasmy and amplification of NUMTs among some of the treated species, complicating the use of DNA barcoding in species discrimination. Based on morphological and genetic evidence, we establish that the correct name for the invasive willow sawfly in the southern hemisphere (South America, southern Africa, Australia, New Zealand), known there only in the female sex, is Euura respondens (Förster, 1854). The species is probably native to the Palaearctic (or even Holarctic) where males are common: possibly as common as females (examined from Europe and Central Asia). The name Euura oligospila (Förster, 1854) has been incorrectly used for the species in the southern hemisphere. The examination of type material and reliable association of males and females based on genetics revealed that females of E. oligospila are morphologically extremely similar to E. respondens (and to some other E. bergmanni group JHR 84: 187–269 (2021) doi: 10.3897/jhr.84.68637 https://jhr.pensoft.net Copyright Marko Prous et al. This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. RESEARCH ARTICLE Marko Prous et al. / Journal of Hymenoptera Research 84: 187–269 (2021) 188 species), but male penis valves and genetics enable reliable separation of these species. Morphological separation of females of E. oligospila and E. respondens is possible, but challenging. Identification keys for males and females of the bergmanni and oligospila groups are provided. The following 15 new synonymies are proposed: Nematus validicornis Förster, 1854, syn. nov. with Euura bergmanni (Dahlbom, 1835); Pteronidea woollatti Lindqvist, 1971, syn. nov. and Nematus turgaiensis Safjanov, 1977, syn. nov. with Euura brevivalvis (Thomson, 1871); Pteronidea pseudodispar Lindqvist, 1969, syn. nov. with Euura dispar (Zaddach, 1876); Nematus (Pteronidea) fastosus var. ponojense Hellén, 1948, syn. nov. and N. (P.) fastosus var. punctiscuta Hellén, 1948, syn. nov. with Euura frenalis (Thomson, 1888); Nematus declaratus Muche, 1974, syn. nov. and N. desantisi D.R. Smith, 1983, syn. nov. with Euura respondens (Förster, 1854); Pteronidea straminea Lindqvist, 1958, syn. nov., P. angustiserra Lindqvist, 1969, syn. nov., and P. disparoides Lindqvist, 1969, syn. nov. with Euura sylvestris (Cameron, 1884); Pteronidea breviseta Lindqvist, 1946, syn. nov., P. breviseta Lindqvist, 1949, syn. nov., P. abscondita Lindqvist, 1949, syn. nov., and P. lauroi Lindqvist, 1960, syn. nov. with Euura viridis (Stephens, 1835). Lectotypes are designated for 18 nominal taxa: Amauronematus longicornis Konow, 1897; A. spurcus Konow, 1904; Nematus bergmanni Dahlbom, 1835; N. brevivalvis Thomson, 1871; N. curtispina Thomson, 1871; N. (Pteronidea) fastosus var. ponojense Hellén, 1948; N. (P.) fastosus var. punctiscuta Hellén, 1948; N. glutinosae Cameron, 1882; N. microcercus Thomson, 1871; N. polyspilus Förster, 1854; N. prasinus Hartig, 1837; N. respondens Förster, 1854; N. salicivorus Cameron, 1882; N. validicornis Förster, 1854; N. virescens Hartig, 1837; Pteronidea curtispina var. luctuosa Enslin, 1916; Pteronus fastosus Konow, 1904; and P. pallens Konow, 1903.


Introduction
The genus Euura Newman, 1837, native to the Holarctic and Oriental regions, is the largest in the Nematinae (Tenthredinidae), containing about 650 species, or half of the subfamily (Taeger et al. 2010(Taeger et al. , 2018. The generic concept was significantly expanded in 2014 as a result of phylogenetic analyses of genetic data (Prous et al. 2014). The genus has remained taxonomically problematic due to the large number of species and high degree of morphological similarity. During 2017-2020 revisionary work on Euura was funded by the Swedish Taxonomy Initiative. Some of the results are presented here, focussing on a revision of the Euura bergmanni and oligospila groups in the West Palaearctic. The definition of these groups is largely based on an analysis of genetic sequence data. The decision to publish on these two groups before the main revision of Euura, is mainly due to the need to correct the taxonomic confusion surrounding the willow sawfly that has become invasive in the southern hemisphere (Caron et al. 2013(Caron et al. , 2014Malagón-Aldana et al. 2017). We provide evidence to distinguish it from closely related or distantly related but morphologically very similar species within the bergmanni and oligospila groups of Euura. Preparation of a revision of other European Euura species is in progress.

BIN
barcode index number; HT holotype; LT lectotype; ST syntype or syntypes.
Specimens studied are listed in the Suppl. material 1 and in the Appendix 1. Additional specimens mentioned, but not examined morphologically, can be found in BOLD (http://www.boldsystems.org/).

Morphological methods
To photograph penis valves and lancets (valvula 1 or ventral part of the saw), genital capsules and ovipositors were separated from the specimen and macerated in KOH (10-15%) for 6-10 hours at room temperature, or treated with proteinase during DNA extraction. Temporary slide preparations of dissected lancets and penis valves in glycerine were made, and after photography, the parts were glued on a piece of cardboard, which was pinned with the corresponding specimen. In addition, relevant permanent slide preparations borrowed from institutional or personal (Veli Vikberg, Turenki) collections were photographed. Photos were taken with a digital camera attached to a microscope. Composite images with an extended depth of field were created from stacks of images using the software CombineZP (Alan Hadley; http://www.hadleyweb.pwp. blueyonder.co.uk/, although apparently no longer available) or Helicon Focus 7.6.4. Most of the lancets were photographed in two overlapping parts and a single image was created using the program Image Composite Editor (Microsoft) or with the plugin MosaicJ (Thévenaz and Unser 2007) implemented in ImageJ (Wayne Rasband; http:// imagej.nih.gov/ij/). Morphological terminology follows Viitasaari (2002).

Molecular methods
For species delimitation and to associate males and females, mitochondrial and nuclear DNA was sequenced from representatives of Euura belonging to the bergmanni and oligospila groups as well as other selected species. DNA was sequenced using Sanger (Prous et al. 2019) or Oxford Nanopore technologies. For Nanopore sequencing, amplicons belonging to different species or species groups were pooled and sequenced with the MinION R9.4.1 or R10.3 flow cells using a Ligation Sequencing Kit (SQK-LSK109). Most of the amplicons sequenced with MinION were amplified using different combinations of tailed forward and reverse primers (variable 4-12 bp added to the 5'-end) to confirm the identity of the final consensus sequences. The raw sequencing signal from MinION was basecalled (translated into a DNA sequence) with Guppy v4.0.11 or 4.2.3 in high accuracy mode. Using available sequences as query, corresponding single molecule Nanopore reads were identified with BLAST 2.9.0+ (https://www.ncbi.nlm.nih.gov/books/NBK279690/). A maximum of 3000 single reads were aligned with MAFFT v7.427 (Katoh and Standley 2013) and the maximum likelihood trees were built with FastTree 2.1.11 (Price et al. 2010). Based on the resulting trees, separate clusters of reads were identified and subsequently used to create consensus sequences. Based on 15-200 reads of each amplicon, MAFFT v7.427 together with EMBOSS cons v6.6.0.0 (http://emboss.open-bio.org/rel/dev/apps/cons. html) and abPOA 1.0.4 (https://github.com/yangao07/abPOA) were used to create initial consensus sequences that were further polished with Medaka 1.0.1 (https:// github.com/nanoporetech/medaka). Medaka variant calling was used to separate haplotypes of nuclear genes. A more detailed protocol and data analysis workflow will be published separately. For most specimens, one mitochondrial and two nuclear genes were sequenced. The mitochondrial gene used is a part (1078-1087 bp) of cytochrome c oxidase subunit I (COI). For a small number of specimens, a longer mitochondrial fragment (including partial tRNA-Cys and complete tRNA-Tyr upstream of COI) was amplified with the primers TW-J1301 (Simon et al. 2006) and A2590 (Normark et al. 1999), resulting in COI length of 1119 bp. For other COI and nuclear primers, see Prous et al. (2019). The two nuclear markers are fragments of sodium/potassium-transporting ATPase subunit alpha (NaK, 1654 bp) and DNA dependent RNA polymerase II subunit RPB1 (POL2, 2527-2552 bp or 2700-2709 bp). The NaK fragment does not include any introns, but POL2 has one short intron (58-84 bp) that was excluded from phylogenetic analyses. For some specimens, shorter sequences of each gene (in most cases due to lower quality of DNA) were obtained. After excluding the intron in POL2, alignment of all genes was straightforward because of the lack of insertions or deletions in the studied specimens (length differences were only due to the extent the gene regions were amplified and sequenced). Some of the analysed sequences have been published previously (GenBank accessions in the Suppl. material 1 dataset of the studied specimens). Additionally, some COI sequences were obtained from BOLD (http://www.boldsystems.org/). The newly obtained DNA sequences have been submitted to NCBI GenBank (accessions MW939671-MW939746, MW939748-MW939850, MW939852-MW939885, and MZ479384-MZ479675). To concatenate separate gene alignments, we used R (R Core Team 2019) package apex (Jombart et al. 2017). Phylogenetic analyses using maximum likelihood (ML) were done with IQ-TREE 1.6.1 or 1.6.12 (http://www.iqtree.org/) (Nguyen et al. 2015). By default, IQ-TREE runs ModelFinder (Kalyaanamoorthy et al. 2017) to find the best-fit substitution model and then reconstructs the tree using the model selected according to Bayesian information criterion (BIC). We complemented this default option with a SH-like approximate likelihood ratio (SH-aLRT) test (Guindon et al. 2010) and ultrafast bootstrap (Hoang et al. 2018) with 1000 replicates to estimate robustness of reconstructed splits. COI (658 bp barcoding region, minimum length 600 bp) and nuclear (combined NaK and POL2, minimum length 1529 bp) p-distances (proportion of nucleotide differences) were calculated in R with the package ape (Paradis and Schliep 2018). In addition, we used ape to calculate the proportion of ambiguous positions for nuclear genes (i.e. heterozygosities) to get the p-distances between the haplotypes of every female and heterozygous larva. Note that p-distances between haplotypes can be larger than maximum within-species distances calculated from the alignment. The reason is that ambiguous positions are treated as missing data in the ape function dist. dna. Alignments used for phylogenetic analyses are available as Suppl. material 2.

Definition of bergmanni and oligospila groups
The bergmanni and oligospila groups of Euura are here primarily defined based on phylogenetic analyses of DNA sequence data ( Fig. 1), but are supported by morphological evidence. The names of the groups are based on the oldest valid species name in that group. Living females of both groups are usually mostly green (except E. leptocephalus) (Fig. 4), but males vary from nearly completely black (E. leptocephalus) to largely pale yellowish (e.g. Figs 17C,D,19E,F). There are other species within Euura that are green in life and could be confused with the bergmanni and oligospila groups, like Euura poecilonota (Zaddach, 1876) and Euura hypoxantha (Förster, 1854). Euura hypoxantha has a less pointed valvula 3 in lateral view (Fig. 5E) and the radix Figure 1. Maximum likelihood tree of Euura based on three genes (mitochondrial COI and nuclear NaK and POL2). Numbers at branches show SH-aLRT support (%) / ultrafast bootstrap support (%) values. Support values for weakly supported branches (<90) are not shown. Letters "f " and "m" stand for "female" and "male" if known. Numbers at the end of the tip labels refer to the length of the sequence and the number of ambiguous positions (e.g., heterozygosities). The tree was arbitrarily rooted between Euura and other Nematini. The scale bar shows the number of estimated substitutions per nucleotide position. and lamnium of the saw (Fig. 6D) are of similar length while in the bergmanni and oligospila groups (Figs 5F-H, 6A, B, 8-9) valvula 3 is more pointed in lateral view and the lamnium is distinctly longer than the radix. Euura poecilonota has a different saw (broader, with more serrulae, and different structure of basal serrulae, Fig. 6C). Based on penis valves (Figs 11A-J, 12A, B), males of the bergmanni and oligospila groups are easier to distinguish from each other and from other species of Euura than females. Males of E. poecilonota are not known for certain (likely candidate on Fig. 12C), because all previous associations have been almost certainly in error. Penis valves illustrated by Benson (1958) for E. poecilonota (under the name Nematus viridescens) belong to E. hypoxantha (Fig. 12D) and those by Macek et al. (2020) to E. dispar. The main difference between bergmanni ( Fig. 11A-J) and oligospila (Fig. 12A, B) group penis valves is in the shape of the valviceps (paravalva + pseudoceps), which is basally about as broad as apically and has a weak or distinct constriction in the middle in oligospila group (usually broader apically and without constriction in the middle in bergmanni group). Euura bergmanni group penis valves usually also have a distinctly deeper invagination between the valvispina and paravalva than in the oligospila group. Identification characters of species of the bergmanni and oligospila groups are summarised in the keys given below. Within the bergmanni group it is worth defining the viridis subgroup (brevivalvis, dispar, viridis, and glutinosae) that is genetically well supported (Fig. 2) and where mitochondrial COI sequences do not allow reliable separation of species. Females of this subgroup can be recognised also morphologically by saws and tendency to have a relatively long malar space compared to most other species in the bergmanni and oligospila groups.

Phylogeny of the bergmanni and oligospila groups
Based on three genes (COI, NaK, POL2), the branching order at the base of Euura phylogeny is poorly resolved, but several strongly supported groups can be identified, such as the bergmanni and oligospila groups (Fig. 1). Both groups were strongly supported as monophyletic also by all three genes analysed separately (not shown). Females of the bergmanni and oligospila groups can be very difficult to distinguish despite being distantly related (as suggested by estimated branch lengths: Fig. 1) within Euura. Because the phylogenetic relationships among these and other Euura species groups remain unresolved based on current data, it cannot be said whether high morphological similarity between the bergmanni and oligospila groups is because of convergence, parallelism, retention of ancestral characters, or some other reason.
Because of clear conflict between mitochondrial and nuclear data within the bergmanni and oligospila groups themselves, we analysed these gene sets separately for each group. Within the bergmanni group, the viridis subgroup is the only relationship above species level that is strongly supported both by COI and nuclear genes. The other relationships are poorly supported or even strongly contradictory (Fig. 2). Species outside the viridis subgroup are all monophyletic based on all genes, although identification of specimens of sylvestris, leptocephalus, and bergmanni from North America based on barcodes (no nuclear data available) is uncertain, as we have not examined these specimens (available photos in BOLD are at least consistent with the identities suggested by barcodes). COI barcodes do not allow species identification within the viridis subgroup (Fig. 2), where most specimens (containing all species) fall within a tight cluster of similar sequences (divergence less than 1.8%). A smaller number of COI sequences within the viridis subgroup ( Fig. 2) are more divergent (up to 3.4%), although interestingly there are specimens that appear to contain two or more COI variants (differing by 0.5-2.3%) falling within the main cluster as well as outside of it (see further discussion under the section "Possible heteroplasmy and NUMTs in Euura"). All viridis subgroup species are monophyletic based on combined analysis of both nuclear genes and based on POL2 only (Fig. 2). NaK, however, does not clearly separate the viridis subgroup species, as these tend to be in multiple clusters Figure 2. Maximum likelihood trees of Euura bergmanni group (COI left, nuclear NaK and POL2 upper right) and viridis subgroup (NaK lower left, POL2 lower right). Numbers at branches show SH-aLRT support (%) / ultrafast bootstrap support (%) values. Support values for weakly supported branches (<90) are not shown. Letters "f " and "m" stand for "female" and "male" if known. Numbers at the end of the tip labels refer to the length of the sequence and the number of ambiguous positions (e.g., heterozygosities). Note the COI heteroplasmic variants for brevivalvis ZMUO.030869 and ZMUO.030870 (in bold).
( Fig. 2). In the oligospila group, there are two main clusters based on COI (excluding the possible NUMT) and nuclear genes, but specimen composition differs between the marker sets (Fig. 3). According to COI, both main clusters contain both species (frenalis and oligospila), but according to nuclear genes (combined analyses or separate) the species are monophyletic (Fig. 3). Nevertheless, unlike in the viridis subgroup, COI may still enable identification of oligospila group species, at least in most cases. At least in Europe, frenalis appears to be restricted to two COI clusters, one of which is quite distinct (minimum distance 3.65% to a cluster containing apparently only oligospila), while the other is hardly different (0.3%) from some oligospila specimens (Fig. 3). Sequences of eight specimens of oligospila in BOLD are apparently NUMTs because they include a stop codon and in some cases also indels (insertions or deletions). For another specimen (ZMUO.030844), two COI variants were co-amplified and sequenced, one of which belonged to the NUMT cluster and the other to one of the oligospila clusters (Fig. 3).

Delimitation of species
Unfortunately, mitochondrial COI data is unreliable for species delimitation as it often conflicts with morphology and nuclear data, which is a common pattern in sawflies (Linnen and Farrell 2007;Prous et al. 2017Prous et al. , 2020. Due to the high degree of similarity among females of the bergmanni and oligospila groups, and poor resolution of DNA barcoding, we have primarily relied on male penis valves and nuclear sequence data to delimit species. There are usable morphological characters to distinguish also females, but these tend to be less reliable than penis valve characters in males. Close congruence between morphology and two independent nuclear markers (and often also mitochondrial COI) enabled us to delimit species rather reliably (see individual species treatments for details). Nuclear sequence data is informative Figure 3. Maximum likelihood trees of Euura oligospila group based on COI (left) and nuclear genes (NaK and POL2, right). Numbers at branches show SH-aLRT support (%) / ultrafast bootstrap support (%) values. Support values for weakly supported branches (<90) are not shown. Letters "f" and "m" stand for "female" and "male" if known. Numbers at the end of the tip labels refer to the length of the sequence and the number of ambiguous positions (e.g., heterozygosities). Note the COI NUMT for oligospila ZMUO.030844 (in bold). about within-species variation even without prior knowledge about species boundaries, because the number of variable sequence positions (due to heterozygosity) of heterozygous females (males are haploid) can be counted. When considering mean or maximum distances within species, either based on all specimens or counting heterozygous positions in females individually, the values are remarkably similar. In the oligospila group, all the females are heterozygous with mean and maximum distances between haplotypes of 0.28 and 0.53%. The same within-species distance values for all specimens of the oligospila group are 0.22 and 0.68%. In the bergmanni group, 94% of the females are heterozygous with mean and maximum distances between haplotypes of 0.29 and 0.83%. The same within-species distance values for all specimens of the bergmanni group are 0.11 and 1.09%. In other words, haplotypes within a single female are, on average, about as distant as two different specimens of the same species.

Confusion about the use of names
Due to the high degree of similarity of females in the bergmanni and oligospila groups, many of the names have confusing histories, and misinterpretations are also common in the recent literature. Here, only the more prominent or recent examples will be discussed. Particularly, there has been confusion about the identity of the invasive willow sawfly in the southern hemisphere (South America, southern Africa, Australia, New Zealand), which we here identify as Euura respondens (Förster, 1854). The name Nematus oligospilus [= E. oligospila (Förster, 1854)] has been incorrectly used for this sawfly in the southern hemisphere due to the high degree of similarity between females of E. oligospila and E. respondens. These two species prove to be genetically distant within Euura (Fig. 1) and can also be clearly separated based on penis valves (males of E. respondens are not known in the introduced parthenogenetic populations in the southern hemisphere). Accordingly, the nominal species Nematus desantisi Smith, 1983 described from Argentina, is here treated as a synonym of E. respondens and not of E. oligospila. There are further name mix-ups involving other bergmanni group species. Macek et al. (2020) used the name Nematus viridis for a species here called E. dispar, and the name N. breviseta for E. viridis. In using the name N. breviseta, Macek et al. (2020) seem to have relied on Zhelochovtsev (1988), where the name N. viridis is not even used. Benson (1958) did use the name N. viridis, but did not recognise E. dispar. Confusingly, Benson's (1958) drawing of the penis valve of N. viridis actually represents E. bergmanni (copied by Lacourt 2020). Benson apparently had not seen or recognised males of E. viridis and E. dispar.

Possible heteroplasmy and NUMTs in Euura
Remarkably, multiple mitochondrial COI variants are frequently observed within individuals of Euura (unpublished), including in the bergmanni and to a lesser extent in the oligospila groups. The viridis subgroup of the bergmanni group is the most problematic in this regard, where many individuals have secondary peaks in Sanger chromatograms, indicating the presence of multiple variants. Usually, the secondary peaks are rather weak, enabling the determination of the dominant variant, but sometimes different variants are represented at similar frequencies, making it necessary to code variable sequence positions with IUPAC ambiguity symbols. Some of the specimens referred to here were re-amplified and sequenced with Nanopore technology, enabling us to determine the exact variants and place them in the tree (Figs 2, 3). Nanopore sequencing even revealed more than two variants per individual in one case (ZMUO.030870), which Sanger sequencing cannot indicate reliably. The intra-individual COI variants in the viridis subgroup could indicate genuine heteroplasmy rather than nuclear-encoded mitochondrial pseudogenes (NUMTs), because the variants have neither stop codons, nor frame shifting indels, and are not in any other way unusual in the viridis subgroup context, i.e. do not have unusual nucleotide composition or display long branches in the tree. Interestingly, clear unreliability of mitochondrial COI barcodes in indicating species identity is also restricted to the viridis subgroup. If the mis-match between mitochondrial and nuclear sequences with respect to species boundaries in the viridis subgroup is mainly due to occasional hybridisations between species (a pattern expected in theory for haplodiploid species and at least partly supported by empirical studies; Linnen and Farrell 2007;Patten et al. 2015;Prous et al. 2020), then this could also explain the presence of extensive heteroplasmy, that has been suggested to be more likely when heterospecific hybridization has occurred (Ladoukakis and Zouros 2017;Mastrantonio et al. 2019). In the oligospila group there appears to be a cluster of divergent NUMTs, because all of them contain the same in-frame stop codon as well as in some cases the same frame shifting indels (see the specimens listed under E. oligospila). These possible NUMTs can co-amplify with the genuine COI sequence (ZMUO.030844; Fig. 3) or amplify instead of the mitochondrial variant (eight specimens in BOLD).

Females
Abdomen posteriorly with modified segments 8-9 forming the ovipositor and its sheath (Fig. 5A). Length 4.5-9.0 mm. With the exception of E. leptocephalus the species are extensively green in life (only exceptionally yellow) and yellowish when dried. Clypeus emarginate ventrally; flagellum 2.6-3.4 times as long as head width; malar space 1.1-2.2 times as long as diameter of front ocellus; claws bifid (West Palaearctic taxa); valvula 3 dorsally roughly 1.5-2.0 times as broad as a cercus, weakly tapering, and without invagination posteriorly; lancet with 16-21(22) serrulae, and long and narrow in most species.

Euura bergmanni group
The group is mostly defined based on phylogenetic analyses of sequence data (Fig. 1 least in females there tend to be distinct differences between the generations also in the length of malar space and perhaps postocellar area. In overwintering generations, the malar space tends to be distinctly shorter (Fig. 5M) than in later generations (Fig. 5N). Genetic data. COI. Based on 13 specimens, maximum within species distance is 3.65% and the nearest neighbour, diverging by a minimum of 7.1%, is the viridis subgroup. Only one BIN: BOLD:AAG3539.
Nuclear. Based on 5 specimens, maximum within species distance is 0.19% (0.23% based on haplotypes of individual females). The nearest neighbour, diverging by a minimum of 3.5%, is viridis. Host plants and behaviour. Hosts: a wide variety of Salix species, including alba, fragilis (Weiffenbach 1985), aurita, viminalis (Boevé 1990), caprea, pentandra, phylicifolia (Kangas 1985), and purpurea (Benander 1966). Lindqvist (1956) recorded up to four generations per year in Finland. Lindqvist (1941) stated that adults from the overwintering generation were very much darker than the next generation, and that adults of the 3 rd and 4 th generations were paler still. The characteristic continuous double dorsal line of the larva makes their identification usually straightforward. This double line is usually white in early generation larvae, but pink or even red in later generations.
Type material. Nematus bergmanni Dahlbom, 1835. Lectotype, here designated, ♀, MZLU2017334, MZLU. Dahlbom cited a publication by Bergman (1763), in which adults were mentioned, which Dahlbom considered to belong to this species. There is no trace of Torbern Bergman material in the UUZM collection (Hans Mejlon, personal communication: March 11, 2019). Following this citation, Dahlbom described a larva, evidently from his original observations: "Larva prasina linea dorsali lata livida vel purpurascente et utrinque fusco-marginata" [Larva leek green with broad blue or purplish band and dark-bordered at both sides], with the additional information [translated from Latin] "Frequently observed on willows around Lund in Scania from 26 August to 2 October". Although a label on the lectotype bears the date "14 Aug.", this might refer to the date of emergence of an adult reared from a larva, and therefore does not necessarily contradict Dahlbom's statement.
Nematus curtispina Thomson, 1871. Lectotype, here designated, ♀, MZLU2017334, MZLU [the same specimen as the LT of bergmanni Dahlbom]. Koch (2000) mentioned this same specimen as LT, with details of its labelling. This was not, however, a valid taxonomic act, because he omitted an explicit statement that he was designating this specimen (see ICZN 2003).
Nuclear. Based on 10 (only NaK) or 8 (NaK and POL2) specimens, maximum within species distance is 1.09% (only NaK) or 0.92% (NaK and POL2) and 0.12% based on haplotypes of individual females. The nearest neighbour, diverging by a minimum of 0% (only NaK) or 0.41% (NaK and POL2), is viridis. The 0% distance between viridis and brevivalvis for NaK is because one of the haplotypes of one female of viridis (ZMUO.030835) isidentical to several brevivalvis specimens.
Amauronematus spurcus Konow, 1904. Lectotype, here designated, ♀, GBIF-GISHym3848, SDEI. Koch (2000) mentioned this same specimen as LT, with details of its labelling, together with a female paralectotype. This was not, however, a valid taxonomic act, because he omitted an explicit statement that he was designating this specimen (see ICZN 2003).
Nuclear. Based on 6 specimens, maximum within species distance is 0.25% (0.15% based on haplotypes of individual females). The nearest neighbours, diverging by a minimum of 0.63%, are brevivalvis and glutinosae.
Host plants and behaviour. Hosts: Betula pendula (Kangas 1985) and B. pubescens (Kontuniemi 1960). Probably two generations per year, of which larvae identified as pseudodispar belong to the second generation (Lindqvist 1969).
Nuclear. Based on 3 specimens, maximum within species distance is 0.07% (0.05% based on haplotypes of individual females). The nearest neighbour, diverging by a minimum of 2.96%, is viridis.
Distribution. West Palaearctic and Nearctic (Sundukov 2017). Mainly in subarctic and arctic areas. Specimens studied are from Finland, Norway, and Sweden.
Nuclear. Based on 3 specimens, maximum within species distance is 0.18% (0.27% based on haplotypes of individual females). The nearest neighbour, diverging by a minimum of 3.82%, is viridis.
Host plants and behaviour. Hosts: a large number of Salix species, as well as sometimes Populus species (Dapoto and Giganti 1994;Koch and Smith 2000). Host records from the countries where E. respondens has become invasive are considered to be reliable, because no similar sawflies occur there, but records from Europe should be treated with caution, because the larvae of E. respondens and E. oligospila are apparently very similar. Up to six generations per year have been recorded, in Argentina (Alderete et al. 2002).
Type material. Nematus respondens Förster, 1854b. Lectotype, here designated, ♂, GBIF-GISHym3404, ZSM. The specimen is completely destroyed and only one badly damaged penisvalve remains. However, the shape of the penis valve of this taxon is highly characteristic.

Similar species.
Females are most similar to oligospila group, E. respondens, and E. bergmanni, from which it differs usually by having a longer malar space. Lancet is usually narrower compared to E. bergmanni. Clypeus is usually less deeply emarginate compared to oligospila group. Ventral part of 2 nd to 4 th suture of lancet is weakly or distinclty curved basally in E. sylvestris, but oblique and more or less straight or weakly curved apically in E. respondens. Males distinguishable from other species by their relatively distinct penis valves. Genetic data. COI. Based on 22 specimens, maximum within species distance is 3.68% (5.68% when including also Nearctic-only BINs) and the nearest neighbour, diverging by a minimum of 5.45%, is viridis subgroup. BINs: BOLD:AAG3515 (Holarctic), BOLD:AEH2646 (ZMUO.038944, Finland), and possibly also Nearctic BOLD:AAU8841, BOLD:ACJ5634, BOLD:ACI4984, BOLD:AAG3521, BOLD:ACN0565.
Nuclear. Based on 14 specimens, maximum within species distance is 0.97% (0.83% based on haplotypes of individual females). The nearest neighbour, diverging by a minimum of 2.72%, is viridis.
Host plants and behaviour. Hosts: Salix spp. and at least occasionally Populus tremula. Cameron (1884) included a description of the larva of N. sylvestris in the species' description, and gave the host as Salix caprea. Other Salix species recorded as hosts are: pentandra, phylicifolia (Kangas 1985), and myrsinifolia (Kontuniemi 1971). We have collected or reared larvae from S. caprea, S. pentandra, S. hegetschweileri, S. myrsinifolia, and Populus tremula. Apparently there can be more than one generation per year (Kontuniemi 1971).
Host plants and behaviour. Hosts: most host records under the name viridis and all those under breviseta refer to Betula, e.g. B. pendula (Kontuniemi 1960), B. pubescens (Kontuniemi 1960;Tenow 1963;Hanhimaki et al. 1995), and Betula utilis (Schedl 2010). However, other sources mention several additional hosts, all of which require checking, because they may involve misidentifications of the sawfly species: see also above, under E. glutinosae.
Distribution. Palaearctic (Sundukov 2017; current data). Specimens studied are from Finland, Germany, Sweden, and United Kingdom. Nematus prasinus Hartig, 1837. Lectotype, here designated, ♀, GBIF-GISHym3388, ZSM. Koch (2000) mentioned this same specimen as LT, with details of its labelling. This was not, however, a valid taxonomic act, because he omitted an explicit statement that he was designating this specimen (see ICZN 2003).
Nematus polyspilus Förster, 1854. Lectotype, here designated, ♀, GBIF-GISHym3386, ZSM. Koch (2000) mentioned this same specimen as LT, with details of its labelling. This was not, however, a valid taxonomic act, because he omitted an explicit statement that he was designating this specimen (see ICZN 2003).
Pteronidea breviseta Lindqvist, 1946. No identifiable syntypes found: the HT of P. breviseta Lindqvist, 1949Lindqvist, was collected in 1948, and so cannot be a syntype of the taxon described in 1946.

Similar species.
Most similar species are sylvestris and respondens, from which it differs by having more prominent serrulae (cf. Figs 7B, 9C, D). It is possible that pallens is a synonym of respondens (shape of the basal sutures of the lancet seem to be most similar to this species), even though the serrulae seem to be more prominent than in other specimens of bergmanni group examined so far. Lindqvist (1972) synonymised straminea, which we treat as a synonym of sylvestris, with pallens. Male unknown. Distribution. East Palaearctic. Removed from the list of West Palaearctic taxa. Specimens studied are from Russia (Irkutsk Oblast).
Notes. Seems to belong to the bergmanni rather than the oligospila group, because of its long malar space according to the original description (Hellén 1948). Overall colouration, small size (5.5 mm), and high northern locality (69.42°N, 86.25°E) suggests that it could be conspecific with E. sylvestris.

Euura oligospila group
The group is mostly defined based on phylogenetic analyses of sequence data (Fig. 1).
No clear female morphological characters distinguish it from the highly similar but distantly related bergmanni group. Penis valves, however, enable rather easy separation of the oligospila group from the bergmanni group and the other Euura. The valviceps in the oligospila group is basally about as broad as apically, has a weak or distinct constriction in the middle, and the invagination between valvispina and paravalva is indistinct (Fig. 12A, B). Usually multiple generations per year, except probably at high latitudes and altitudes. Salix species are the hosts of the two European species. Larvae are cryptically coloured, solitary, and at least the later instars feed mostly from the edges of leaves. The group has a natural distribution in the Holarctic.

Similar species.
See the key couplets 10 (females) and 4 (males). Genetic data. COI. Based on 11 specimens, maximum within species distance is 6.38% and the nearest neighbour, diverging by a minimum of 0.3%, is oligospila. BINs: BOLD:AEA7654 and BOLD:ABZ2416, but possibly also Nearctic BOLD:ACA8095 and BOLD:AAV4677.
Nuclear. Based on 10 specimens, maximum within species distance is 0.67% (0.53% based on haplotypes of individual females). The nearest neighbour, diverging by a minimum of 0.72%, is oligospila.
Two generations in southern Finland according to Lindqvist (1961). Distribution. Palaearctic (Sundukov 2017, current data), but possibly also Nearctic. Mainly in the North. Specimens studied are from Finland, Norway, Russia (Murmansk Oblast, Nenets Autonomous Okrug), and Sweden.
Pteronus fastosus Konow, 1904. Lectotype, here designated, ♀, GBIF-GISHym3851, SDEI. Koch (2000) mentioned this same specimen as LT, with details of its labelling. This was not, however, a valid taxonomic act, because he omitted an explicit statement that he was designating this specimen (see ICZN 2003).

Similar species.
See the key couplets 10 (females) and 4 (males). Can be small, about 4.5 mm (ZMUO.030844). Genetic data. COI. Based on 12 specimens, maximum within species distance is 5.93% and the nearest neighbour, diverging by a minimum of 0.3%, is frenalis. ZMUO.035689, ZMUO.031369, ZMUO.035712. Specimen ZMUO.030844 has two COI variants, one belonging to BOLD:AEA6205 and the other one in the NUMT cluster (it has a stop codon in the barcoding region and a 1 bp insertion outside the barcoding region).
Nuclear. Based on 10 specimens, maximum within species distance is 0.68% (0.52% based on haplotypes of individual females). The nearest neighbour, diverging by a minimum of 0.72%, is frenalis.
Host plants and behaviour. Hosts: Salix species (Cameron 1882 [types of N. salicivorus]; Macek et al. 2020). A large number of Salix species, and a few species of Populus, are named as hosts of oligospila (or oligospilus) in the literature, but such records from Australasia, southern Africa and South America all refer to E. respondens (see above). Because of widespread mixing up in Europe of E. respondens and E. oligospila, it is often not clear as to which species the published records refer. Lorenz and Kraus (1957) listed Ulmus as a host of oligospila, but this is probably based on Conde (1938), whose meaning is not clear, and probably only indicates that an adult had been collected from Ulmus. Probably has two generations per year in southern Sweden (Benander 1966).    On underside of card to which the specimen is gummed: "Bred 11.5.78; the larva X on Salix viminalis: Wor'sh."

Discussion
The genus Euura has been taxonomically challenging and probably will remain so for years to come, but fortunately progress is also continuously being made. Here we revised the taxonomy of the bergmanni and oligospila groups, containing the bulk of the "green" Euura. These "green" species have been notoriously difficult to identify. In addition to very similar morphology, taxonomic oversplitting of species has further complicated identification of species in both groups. The most recent attempt to revise the "green" species was by Koch (2000), who considered only females and included fewer species than here. Unfortunately, his key (as well as all previous keys) cannot be relied upon to identify species. We hope that the keys for females and males and accompanying photographs provided here are an improvement, although identification remains difficult, especially of females, where very few usable characters have been found and the differences are often minute, and sometimes not even definitely constant. Thanks usually to relatively clear differences in penis valves, males are easier to identify. Unfortunately, even the distinction of the males of some species is difficult (viridis and glutinosae), and potentially useful characters require confirmation by rearing or further sequencing of nuclear genes. The taxonomic decisions made here were greatly facilitated by genetic data that enabled reliable association of males and females. Although extensive rearings would have also enabled association of females and males, it can be argued that without the genetic data, the true identity of the invasive willow sawfly (E. respondens) might have remained undiscovered for a long time, because males are not known in the southern hemisphere populations. One other notable result is that E. leptocephalus, which is never green in life, belongs to the bergmanni group according to genetic data. This is also consistent with the structure of the saw and penis valve. Previously, only Lindqvist (1960b) had argued that E. leptocephalus (under the name Pteronidea leptocephala) is closely related to E. bergmanni. As in many other sawfly groups (Linnen and Farrell 2007;Prous et al. 2017Prous et al. , 2020Schmidt et al. 2017), identification of species based on mitochondrial barcodes is often not reliable in the bergmanni and oligospila groups (Figs 2, 3). The most problematic in this regard is the viridis subgroup (brevivalvis, dispar, viridis, glutinosae) of bergmanni group, where probably all species can have identical COI sequences (even if >1000 bp long) while at the same time divergence within species can be around 2-3% (Fig. 2). The other species in the bergmanni group (bergmanni, leptocephalus, respondens, sylvestris) are all clearly distinguishable based on barcodes: as far as currently known, each species has only one or rarely two BIN clusters in Europe. In the oligospila group, both species are split among multiple BIN clusters, but in most cases each cluster appears to contain either E. frenalis or E. oligospila, except BOLD:ABZ2416, which contains both species (Fig. 3).
One issue that still needs attention is taxonomic revision of Nearctic taxa of the oligospila and bergmanni groups, which could affect the use of names for Palaearctic species.
Although morphological differences between adults of different generations have already been observed in a few species of Tenthredinoidea, such as Pristiphora leucopus (Hellén, 1948) (Grearson and Liston 2012), a difference in the appearance of larvae of different generations of the same species is so far only recorded for Euura bergmanni and E. dispar.
Vårdal (NHRS), Matti Viitasaari (Helsinki), and Veli Vikberg (Turenki). We also thank the reviewers Leonardo Malagón-Aldana, Yuri Sundukov, and Andreas Taeger for helping to improve the manuscript. This paper is part of the special issue "Hymenoptera studies through space and time: A collection of papers dedicated to the 75 th anniversary of Arkady S. Lelej". MP is grateful to Arkady Lelej, as well as to Maxim Proshchalykin and Valery Loktionov, for the opportunity of collecting in the Russian Far East. We thank also Maxim Proshchalykin for his editorial work.

Appendix 1
List of specimens studied, available also as Suppl. material 1 excel