The key was created in Xper3 version 1.4.0. Pictures could be uploaded as JPEGs and are hosted on a free Dropbox account. The original TIF-files are deposited on zenodo.org. Descriptors were ranked with the built-in ranking system of Xper3 (rank 1 = little important, 5 = highly important) according to their distinct nature and relative simplicity to study. Measurements, which were taken for all species, were read in as quantitative data with fixed ranges. Completeness of the database was checked using the inbuilt tools for analysis of the data.

As the official user manual of Xper3 (Vignes-Lebbe et al. 2017) is somewhat rudimentary, we hence include here a short account of the main features and possibilities offered by Xper3 and how we used them.

The user of Xper3 has several tabs to edit the descriptive data in the software. In the “Items” section all taxa are listed and can be provided with additional detailed information (e.g. distribution, first author, habitat, indicator values etc.) and with illustrating material like pictures, movies, sound files or any other file (e.g. maps, text files). In addition, new taxa can be added and edited in the same way as already existing taxa. The “Descriptive model” section lists all descriptors and can also be edited as described for the “Items” section. States for each descriptor can be edited or new ones can be added.

Xper3 has the possibility to weight the descriptors. This is a great way to simplify and speed up the identification process, allowing quick and reliable results. For instance, problematic (e.g. shape of propodeal foramen) or rather difficult (e.g. degree of paraspiracular inclination) characters can be weighted less, which reduces the chances for users to deal with them already at the beginning of the identification process. However, there is no guarantee that such characters will never appear in the top part of the list.

The section “Description” offers the possibility to assign each descriptor the respective states for each item.

Xper3 comes with some diagnostic tools to check the completeness of the database or to analyse the data. A view of the complete description matrix with the possibility to search for undescribed items and the revision of these is a great help to make sure that each taxon is completely described.

Furthermore, it is possible to select various items or assign them to groups for comparison. The output is a matrix, which also shows all descriptor states of every included item. In addition, the states for each descriptor, where the items differ or are congruent, are displayed separately.

The tool “Checkbase” is an automated search for various errors in the database, for instance items with identical descriptions, undescribed or inapplicable descriptors and descriptors described as unknown. A direct link to these errors provides a handy means for quick and easy revision. Finally, it is possible to generate different kinds of keys. Besides the interactive version, flat or tree-shaped dichotomous versions can be generated in various output formats (HTML, PDF, SDD, Zip archive etc.).

To work with the key, one selects any character which seems conspicuous or unique to the specimen and decides which state is appropriate. Of course, it is recommended to start with the top characters, as they have been weighted higher by the author for good reason. Once a selection is made, the software will exclude every taxon which does not fit this set of states and give a list of the remaining possible names. If more than one name remains, more characters have to be studied. If only one name remains, this should normally be the name one is looking for. This can be checked by comparing the specimen with all the pictures linked to this taxon and the complete diagnosis. The section “History” provides a helpful means to look up the identification process and correct potential mismatches without resetting the complete selection.

The taxa to be included in the key were based on a list of valid European species for the genus Pteromalus from the Universal Chalcidoidea Database (Noyes 2016). A first list yielded 374 species names, from which we excluded 301 names a priori. These names did not belong to species of our target species group, are of dubious status or belong to species of which the type specimens are lost.

Of the remaining 73 names, we included 28 species belonging to the P. albipennis species group (sensu Graham 1969). Pteromalus annae, P. arnicae and P. sonchi were excluded, since we regard these as possible synonyms of P. caudiger, P. albipennis and P. intermedius. Pteromalus leucanthemi might be considered a synonym of P. albipennis (Lutz 2014). P. cingulipes, currently a synonym of P. albipennis, has been demonstrated to be a valid species (Maletti et al. in press) and is therefore treated separately. P. ametrus is morphologically extremely close to P. temporalis but since our paper is not a revision of the taxon, we consider it as a valid species.

Of the remaining 47 names, we included 9 species from 8 species groups (sensu Graham 1969) as representatives. Additionally, 9 names that either belong to species regarded as species sola (Graham 1969) or to recently described species were selected. Finally, we included one new species, Pteromalus capito Baur sp. n. In total, 47 species were included (Table 1).

If the key leads to species names representing whole species groups (e.g. altus, sequester etc.), the resulting identification has to be treated with caution. In this case, we refer to (Graham 1969) for a sound identification.

Nomenclature and classification of Pteromalidae follow Noyes (2016). Terminol­ogy of body parts follows Gibson et al. (1997) and for terms concerning sculpture of the integu­ment and for some particular expressions used in the description, we refer to Graham (1969). The separation of the plica of the propodeum into an anterior and a poste­rior part is according to Graham (1969). The above mentioned progress in chalcid taxonomy applies merely to females. Due to their short life cycle, males can exhibit much more variation in important traits without suffering too many restrictions in fitness and are therefore much harder to identify reliably (Graham 1969, Bouček and Rasplus 1991). In addition, males are not available for many species. Therefore, only females were examined in this work. For the identification of males, we refer to Graham (1969).

All specimens relevant to this work are listed in appendix A. Examined specimens are deposited in the following collections (abbreviations according to Noyes (2016)): Biological Museum (Entomology), Lund University, Lund, Sweden (LUZM); Natural History Mu­seum Bern, Bern, Switzerland (NMBE); Oxford University Museum of Natural History, University of Oxford, Oxford, UK (OUM), Swedish Museum of Natural History, Stockholm, Sweden (NHRS).

Most characters were studied on dried and card mounted specimens (Noyes 1982). Concerning character selection and definitions of character states, we often followed Graham (1969). The descriptive data of P. aartseni were based on the original description by Gijswijt (1972). For better examination of the back of the head, slides were prepared using the method described by Noyes (1982). Microtomy was applied to access the paraspiracular inclination at optimal conditions for digital microphotography. This is a new character, which describes the shape of the paraspiracular sulcus. It is best seen in lateral or dorsolateral view. However, hind legs, wings or the propodeal callus may mask the view, which makes it very difficult to spot the differences. Processed specimens are listed in Suppl. material 5. Specimens were all preserved in 90% ethanol and stored in a refrigerator before processing. LR White (soft grade) (London Resin Company Ltd., Reading, Berkshire), a resin commonly used in immunocytochemistry (e.g. Wynford-Thomas et al. 1986), served as the tissue embedding medium. The specimens were first immersed in 100% ethanol for one hour to completely dehydrate. Then head, legs, wings and gaster were removed and preserved again in 90% ethanol. The remaining body was transferred to LR White for three hours. The LR White bath was exchanged three times. After the third replacement, 10 ml LR White and one drop of LR White accelerator were mixed thoroughly and the specimen was placed within the resin until it became viscous and the object kept its position. The medium was allowed to polymerise hermetically sealed in a water bath overnight. Polymerised resin blocks were grinded using a Leica RM 2145 microtome. The specimen was orientated in a way that the glass blade grinded it laterally. The material was grinded until the spiracle and the paraspiracular sulcus were truncated (Fig. 1). Resin blocks were afterwards immersed in 100% ethanol overnight to dissolve from the body. Then all parts of the respective specimen were dried using the AXA-method (Alcohol/Xylene-Amyl acetate method), a procedure which minimises shrinking and deformation (Achterberg van 2009). Dried body parts were glued on a card point with fish glue (Syndetikon) (Noyes 1982) and stored in the collection of the NMBE. The drying and mounting procedure was applied to all ethanol specimens which are accounted for in this work.

Figure 1.  

Pteromalinae sp., schematic dorsal view of the mesosoma. The line indicates the orientation of grinding. After Gibson et al. (1997).

a

b

c

Figure 2.

Pteromalus spp., metasternum in caudal view. Scale bar = 1 mm.

a: Pteromalus caudiger.  
b: Pteromalus inclytus.  
c: Pteromalus intermedius.  

a

b

Figure 3.

Pteromalinae sp., schematic lateral view of the mesosoma after grinding. Arrows indicate the position of the paraspiracular inclination. After Gibson et al. (1997).

a: Inclination sharp.  
b: Inclination blunt.  

a

b

c

d

Figure 4.

Pteromalus capito Baur sp. n., a, b: holotype female; c, d: paratype female. Scale bar = 0.5 mm.

a: Head, lateral view.  
b: Fore wing, dorsal view.  
c: Pronotal collar, lateral view.  
d: Propodeum, dorsal view.  

Characters were selected and measured according to the procedure described in Baur (2015) (Table 2). The distances were measured using ImageJ, version 1.51j8 (Schneider et al. 2012). To avoid variation due to fluctuating asymmetry (e.g. Bechshøft et al. 2008), measurements of paired characters were taken on the left hand side, where available and possible. Especially for type material, this was not always possible and in order to provide at least some measurement, we also measured photographs taken on the right hand side (noted in Suppl. material 2). Prior to taking photographs and measuring, a Bland-Altman testing procedure (Bland and Altman 1986, Maletti et al. in press) was performed to ensure comparability with former gaugers, from whom data was used for morphometric comparison with the new species (Baur 2015, Lutz 2014, Maletti et al. in press, Suppl. material 8). Ranges of measurements and ratios used in the key were calculated with R 3.0.1 (R Core Team 2013).

We also used the Keyence microscope for making stack-images of qualitative character states. A 3-digit individual code including the notion “Palbi” (e.g. “Palbi 536”) was provided for specimens that were measured, photographed or used as reference specimens for comparison with the newly described species. Specimens already possessing a unique label from former projects (like e.g. “alb”, “Baur” etc.) or that can otherwise be recognised without a doubt (such as holotypes), were not given an additional label.

To further test the hypothesis that P. capito Baur sp. n. is indeed a separate species, a shape principal component analysis (shape PCA) and a PCA ratio spectrum were applied incorporating 28 of the measurements listed above. The analysis of ratios of body measurements is an important part of morphometric taxonomy in Chalcidoidea, since many cryptic species often differ significantly in body proportions but not in qualitative characters (Baur and Leuenberger 2011). The method has since been known as Multivariate Ratio Analysis (MRA, e.g. László et al. 2013, Gebiola et al. 2017). For further details, we refer to Baur et al. (2014). Due to their higher measurement error, eye length, head length and temple length have been excluded. Analyses of the morphometric data and all graphics were undertaken using R 3.0.1 (R Core Team (2013)) and the package “ggplot2” (Wickham 2016). The best ratios for the separation of P. capito Baur sp. n. from P. albipennis and P. cingulipes were found using the LDA ratio extractor (Baur and Leuenberger 2011, Baur et al. 2014). All measurements and the full R code used for the shape PCA are listed in Suppl. materials 10, 11.

During field trips in summer 2016 in Switzerland and Italy, additional wasp specimens were collected using sweep nets (Noyes 1982) and by collecting infested flower heads of Asteraceae. The latter were transferred to cardboard boxes (length x width x height: 25 x 12 x 4 cm) with a single hole at the front end, where a transparent plastic tube was fixed (following Noyes 1982). The boxes were kept in a climate chamber at 20°C for 9 months. Emerging flies and wasps were attracted to the light and subsequently ended up in the tubes, where they could be easily collected. All animals were then killed with ethyl acetate, preserved in 90% ethanol and stored in a refrigerator (Suppl. materials 6, 7).

No study so far has examined the area of the metasternum in Pteromalidae for distinctive characters although it seems quite promising. In Fig. 2a, b, c, we can see the propodeal foramen flanked by the foramen of the metacoxa. In between, one pair of fine wrinkles delimits the inner area of the metasternum from the paraspiracular inclination. These wrinkles may be complemented ventrally by additional wrinkles in some species, providing a first new distinctive character. Below the propodeal foramen, a fissure is present in some species, dividing the core area of the metasternum vertically into two parts. An alternative state is present, when some sort of broad impression bulges the core area inwards. However, it is also possible for species to lack both of these states and simply show a flat and smooth surface. As an additional character, the shape of the propodeal foramen can be defined as being either oval or round. Ventrally of the metacoxal foramen, a horizontal set of wrinkles concludes the metasternum. For better visibility, the mesocoxa were removed as well.

The preparation of translucent slide material of the back of the head yielded no distinguishing characters. Fig. 6 shows the heads of P. achillei and P. cingulipes in posterior view. When comparing P. capito Baur sp. n. to closely related species, such as P. albipennis, it was obvious that the paraspiracular inclination differed. By grinding the specimens laterally to the level of the spiracle, the tegula, lateral panel of axilla, prepectus, acropleuron, upper mesepimeron and metapleuron of the respective side were removed almost completely. Peripheral structures like the lateral lobe of mesoscutum, axilla and lower mesepimeron were at least truncated. For better visibility of the area of interest, the metacoxa, wings and gaster were removed before grinding. Treating the specimens with a microtome clarified the differences in the paraspiracular inclination. The edge of the propodeum describes either a sharp bend towards the metasternum or else is gently rounded (Fig. 3).

The development of software to analyse large amounts of descriptive data allowed the creation of new taxonomic keys, which led to a facilitated identification process. As an example, the identification of P. albipennis needs 16 steps in Graham (1969), whereas with our interactive key only 6 steps are necessary in the best case. The number of steps strongly depends on the selection of the characters, which in our key are ranked according to their relevance for the P. albipennis species group. This means that this identification path is optimised and that species from other groups may take longer than in Graham (1969). Although Graham’s key comes with a lot of detailed drawings, only a fraction of the characters could be illustrated. This makes it often difficult to understand the differences between two states. Our key is completely illustrated and comes with 585 pictures for all taxa, characters and states. These pictures are directly displayed with every object and therefore easily accessible during the entire identification process. In addition, every taxon is portrayed with its complete diagnosis and a set of pictures illustrating most of the characters of the key. Keys created in Xper3 offer the possibility to share the database with co-workers, who can bring in their latest results. Depending on the rights the owner grants them, they are able to only read or edit the database. This means that they can add new taxa and descriptors, edit existing taxa/descriptors or upload additional illustrations. Constant actualisation is therefore assured. Any changes in the database will not affect the permanent URL, which means that everybody having access to the link will always be working with the latest version. Xper3 comes with the ability to export the database in SDD or CSV files, which can also be processed by other software. Together with the independence from platforms and operating systems, this software allows transfer and editing of data also via alternative ways.The treatment of quantitative characters like measurements and ratios by Xper3 in a non-discrete way is not completely satisfying. It is impossible for an identification key to cover the complete range of variation of a character and therefore users will always have to deal with specimens exceeding the range. Especially if only a few specimens are available for measurement, indicated ranges are not very indicative (see Suppl. material 1). However, for practical reasons and to give it a try, we mostly relinquished making discrete groups, except for some measurements which separate a few closely related species (e.g. see Table 3). The discrete nature of most taxa is nonetheless ensured with the qualitative characters alone.

It must be pointed out that the long term storage of the descriptive data and the pictures as well as the future access to the original database is a major concern. We simply do not have the experience of storing digital data over longer time periods (>20 years) (Baker et al. 2006Janzon 1984). In order to enable future generations to access our key and the data underlying it, we deposited all raw files and the original TIF-pictures on several data repository sites. Since Dropbox, where the pictures linked to the key are hosted, is a commercial provider and is not designed and obliged to ascertain long-term access to scientific data, this way seems practical to us with the current state of knowledge. To enable access to the original database with the right to edit, it would be best to designate a successor, who carries on our work. However, it is not clear if such a person may be found in time. Therefore, we arranged with the team of Xper3 that, if somebody wishes to get writing permission to take care of our database after the corresponding author or his successor has retired, he or she can contact the Xper3 team, which will grant the access. However, it must be clear that the corresponding author must be contacted first if still possible.

It is surprising that, even in one of the taxonomically best studied areas like Central Europe, new species can be found. Pteromalus capito Baur sp. n. clearly belongs to the P. albipennis species group showing all its distinctive characters (see above). In the recent past, some new species have been described for this species group (Janzon 1984) which later turned out to be just forms of existing species (Lutz 2014). However, the morphometric and morphological data both support the status of P. capito Baur sp. n. as a distinct new species, although cryptic in appearance. The shape PCA separates it completely from P. albipennis and P. cingulipes, its morphologically closest relatives. The ratios with the greatest distinctive power (antenna length to OOL and eye distance to POL) provide not only a sound but also easily measurable means for identification. The broad head (in lateral view) separates it morphologically from all other species of the P. albipennis species group. Some P. albipennis specimens may show a similarly pale colouration of their wing setae which can be misleading. However, the shape of the paraspiracular inclination, although not blunt exclusively for P. capito Baur sp. n. in the species group, in combination with the other characters, is a sound guide to a reliable identification. These qualitative characters have proven to be very consistent within the species, therefore providing good support. Only recently (Maletti et al. in press), P. capito Baur sp. n. was included in a study using DNA barcoding. It could be separated very clearly from all other species. These 5 specimens are also part of our morphologic analysis (see above). Our rearing conditions suggest that P. capito Baur sp. n. develops as a parasitoid of larvae of Tephritidae (Diptera). Its host range and therefore also its plant association do not significantly differ from other closely related species. From every sample with Pteromalus capito Baur sp. n., P. albipennis emerged as well. Currently, only Tephritis crepidis Hendel can be confirmed as host because it is the only host species with which P. capito Baur sp. n. emerged exclusively. However, it is very likely that the other species constitute hosts, too. In addition, we can confirm that Pteromalus brachygaster and P. scandiae, which were newly described by Graham (1969) and, according to his key, are morphologically very similar to Pteromalus tibiellus, are in fact original species. This holds true for P. conformis as well, which appears very similar to Pteromalus inclytus Förster. These taxa have not been taxonomically reviewed since their original description.

The paraspiracular inclination has been shown to be very consistent within species and has therefore proved to be a valuable means for distinguishing especially P. capito Baur sp.n. and P. albipennis but also other species. It is not absolutely necessary to grind the specimens with a microtome when studying this character. With some experience and a good light microscope, it is possible to spot the differences without preparation. Grinding specimens was merely used to take good pictures illustrating the differences, which was otherwise impossible with the available equipment. Thirteen species showed a sharp edge of the paraspiracular inclination, of which all belong to the P. albipennis species group, in comparison to 30 species with a blunt edge (not examined for two species). As far as we know, there have been no comments in the scientific literature mentioning this structure in Pteromalidae. Studying of the area of the metasternum astonishingly revealed a whole bunch of distinctive characters. The wrinkles connecting the propodeal with the metacoxal foramen may show some sort of stabilising function of the metasternum. We assume that the impression below the propodeal foramen is linked to gaster size or ovipositor length, since species with a large gaster tend to have such an impression more often. The presence or absence of a fissure vertically below the propodeal foramen may be linked to developmental processes. The shape of the propodeal foramen, although consistent in most species, may only be an artefact of a small sample size, since we did not have the possibility to dissect a large quantity of specimens for each species. Therefore, this character may need further investigation. The metasternum has also attracted the interest of other researchers in the field of Pteromalidae. Krogmann and Burks (2009) included a brief characterisation of the metasternum in their description of Doddifoenus wallacei sp. n. Gibson (2015) gave the relative length of the metasternum and compared the position of the base of the mesocoxa in relation to the base of the metacoxa in his revision of the Notanisus oulmesiensis species group. However, the most extensive work so far on this character has been done by Grissell (1995) in his World Catalog for Toryminae species (Chalcidoidea: Torymidae). He studied the metasternum using drawings and scanning electron micrographs and worked out different character states to separate the genera. Together with our work, this raises the question whether the metasternum also provides useful taxonomical characters for other groups of Chalcidoidea.The examination of the back of the head yielded no useful characters here. The area shows hardly any tangible structures besides the occipital foramen. Since Pteromalus sp. live on very similar food sources (pollen, nectar), their mouthparts also do not differ consistently. However, Lotfalizadeh et al. (2007) have demonstrated that the back of the head shows taxonomically important characters for Eurytominae (Chalcidoidea: Eurytomidae).