Triteleia peyerimhoffi comb. n., a remarkably variable circum-Mediterranean scelionid (Hymenoptera, Platygastroidea)

Abstract Triteleia peyerimhoffi comb. n. (Kieffer, 1906) is redescribed taking into account its great variability and is considered the senior synonym of Triteleia dubia (Kieffer, 1908), Calliscelio lugens (Kieffer, 1910) and Triteleia striolata Kononova & Petrov, 2000, syn. n. Neotypes are designated for Triteleia dubia and Triteleia peyerimhoffi. Triteleia peyerimhoffi is a new record for Greece, France and Croatia and was reared for the first time from eggs of Orthoptera laid in the dead wood of Quercus sp. and Tilia sp. in Romania.


Introduction
Jean-Jacques Kieffer (b. 1857Kieffer (b. -d. 1925) was an Abbé, a clergyman, and taught natural history and religion at the Collége Saint-Austin at Bitche in Lorraine (Nominé 1926). His taxonomic work was published in a large number of scientific papers and in some comprehensive monographs, e.g. Das Tierreich and André's Hyménoptères d'Europe et d'Algérie. During his life, he described 49 genera and 465 species belonging to the Platygastroidea (Johnson 1992). Kieffer had a private collection, the remnants of which are at the Muséum national d'Histoire naturelle in Paris (Notton 2004). He also published on much material belonging to other collectors and museums, and this was returned to them, so Kieffer's types are scattered in collections around the world and many types have yet to be found. Hence, many species described by Kieffer have an uncertain status today.
Triteleia peyerimhoffi was described by Kieffer (1906) under the name Caloteleia peyerimhoffi (Kieffer interpreted Calotelea following Ashmead rather than Walker). The type material, all female, was obtained from eggs of Ephippiger confusus (now Uromenus brevicollis (Fischer) according to Sahnoun et al. (2010) from Algeria by Dr. Paul Peyerimhoff. Peyerimhoff (1908) published a paper about the biology of Ephippiger confusus and mentioned Caloteleia peyerimhoffi, including some notes about its oviposition behaviour.
In 1908, Kieffer, described Ceratoteleia, including species with a characteristic long metasoma with a horn on the first metasomal tergite, and with the length of the marginal vein varying from punctiform to the same length as the stigmal vein. Two years later (Kieffer 1910a), transferred Caloteleia peyerimhoffi to Ceratoteleia. In this genus Kieffer (1910b) described two more new species similar to C. peyerimhoffi: Ceratoteleia lugens from France and Ceratoteleia mediterranea from Italy. Ceratoteleia Kieffer is in fact the same as Caloteleia sensu Ashmead (Ashmead's unjustified emendation of Calotelea Westwood).
Kieffer obviously did not study Ashmead's type of Caloteleia grenadensis (in BMNH) and therefore his concept of Ceratoteleia is very heterogeneous, containing species of several genera (e.g. Triteleia, Calliscelio, Holoteleia, Probaryconus, etc.). Masner (1976) synonymized Ceratoteleia under Calliscelio Ashmead. Silvestri (1939) described the biology of Uromenus and gave some information about its parasitoids. He showed a picture of some parasitoids ovipositing into eggs of Uromenus, identified as Ceratoleia (a misspelling of Ceratoteleia). Petit et al. (2007) have some excellent pictures of Uromenus brevicollis and scelionids ovipositing into its eggs. They identified these as Catoteleia peyerimhoffi (a misspelling for Caloteleia peyerimhoffi).
Triteleia dubia was described by Kieffer (1908) in Apegus, subgenus Parapegus based on one male from Hungary. According to Kieffer, the subgenus Parapegus is distinct from the rest of Apegus because of the long marginal vein (almost equal with the stigmal in Parapegus and punctiform in other Apegus) and because of the sculpture of the head (foveolate in Parapegus and with striae in other Apegus). Kieffer (1908) divided all genera known to him without a dorso-ventrally flattened body and with the scutellum unarmed into two groups: firstly genera with three longitudinal grooves on the mesoscutum and secondly those with two or no longitudinal grooves on the mesoscutum. He included Triteleia and Apegus in the second group.
Two years later Kieffer (1910a) kept Apegus divided into two subgenera: Apegus and Parapegus on the basis of the same characters, but later he considered Parapegus as a distinct genus (Kieffer 1914. Masner (1956) revised Parapegus and described a female allotype of P. dubius from one specimen from Moravia caught on 14 September 1936 by F. Gregor in grassland (specimen catalogue no. 3 109 -NMPC). Later Masner (1976) synonymized Parapegus with Macroteleia and transferred Parapegus dubius to Triteleia. This was correct because Masner recognised that the median longitudinal mesoscutal furrow in Triteleia is only a specific, not a generic character. Parapegus dubius was transferred to Triteleia because of the shape of T6 in the female, which is depressed dorsoventrally to form a flat triangle, and because T7 in the male is armed posterolaterally with two sharp spikes, or at least tiny points. Kozlov (1978), probably without seeing Masner's (1976) paper, described the genus Parapegus again, adapting Masner's (1956) description and reusing drawings from the same paper. This was the only mention of this species in the Russian literature. Kozlov andKononova (1985, 1990), Kononova and Petrov (2000) and Kononova and Kozlov (2008) do not mention Parapegus dubius or Triteleia dubia. Conversely Johnson (1992) mentioned T. dubia (Kieffer, 1908) as a valid Palaearctic species, Bin et al. (1995) recorded it from Italy, and Popovici (2005) recorded this species from Romania based on one female from the Bârnova forest (N.-E. Romania).
Our goal in this paper is to provide a modern description of this species, document its unusual variability and provide new data about its biology. The contributions of the authors are as follows: O.A. Popovici (character definition, species concept development, imaging, collection of new material, manuscript preparation); F. Bin (had the idea for the paper, provided most of the Italian material, and contributed to the section on biology); L. Masner (character definition, species concept development, provided new material from Italy, Hungary and France and elaborated the plan for this paper); M. Popovici (geometric morphometric analysis); David Notton (provided specimens from the BMNH and corrected the English of the paper). Any nomenclatural acts in this paper are to be attributed to O. A. Popovici and L. Masner.

Material and methods
Most of specimens seen were caught with a Malaise trap in various parts of Europe, especially the south and west including: France, Greece, Hungary, Italy and Romania. One specimen each from Croatia and Romania were swept. The remaining specimens were reared from the dead wood of Tilia species and Quercus species. Specimens were glued to triangular card points. For better examination, the maxillo-labial complex, one antenna, legs and wings of some specimens were removed and mounted on microscope slides. Specimens were examined using a Kruss MSZ54 stereomicroscope. Microscope slides were analyzed with a Euromex GE 3045 microscope, and drawings were made using a Reichart drawing tube.
The following characters were measured: body length; LH; WH; POL; LOL; DPO; HE; LE; distance between compound eyes (measured at level of anterior ocellus); Lt; distance between toruli; Lck; Hfd; surface of frontal depression covered with transversal striae; distance between compound eye and frontal depression; length of cheek; length and width for all antennal segments (A1….A12); length of mesosoma; width of mesosoma; length of mesoscutum; Lscut; Wscut; length of metascutellum; width of metascutellum; distance between lateral propodeal carina; width of lateral propodeal area; Lfw; Wfw; length of marginal vein; length of stigmal vein; length of postmarginal vein; Lhw; Whw; length of marginal fringe of hind wing (at level of hamuli); length of metasoma; length of T1; minimum width of T1; maximum width of T1; length of T2; maximum width of T2; length of T3; maximum width of T3; length of T4; minimum width of T4; length of T5; minimum width of T5; length of T6; minimum width of T6.
For each ratio in the description of species we used minimum -maximum (mean ± standard deviation).
The relationships between specimens were analyzed using Principal Component Analysis (PCA). This was performed using log-transformed data on a variance-covariance matrix (Klingenberg 1996). The Jolliffe cut-off value was used to indicate the number of significant principal components and standard errors of these were also determined with a bootstrap procedure (Boot N = 1000). We used the Kolmogorov-Smirnov test to show the distribution of metric data within all populations (all populations were normally distributed, p>0.05). The Levene test was used to test the homogeneity of the variance, a one-way ANOVA was conducted on the whole data set to test significant differences between variables and then a post hoc test defined pairwise differences in variables of populations (Sokal and Rohlf 1995). Depending on the result of the Levene test we used the Tukey or Games-Howell test.
The metasoma was analyzed using geometric morphometric methods based on the Kendall theory of shape (Kendall 1977). The shape is a configuration of Cartesian coordinates of landmarks which are discrete anatomical homologues (Zeldith et al. 2004). Generalized Procrustes Analysis (GPA) was performed to superimpose landmark configuration; it removes variation due to differences in translation, orientation, size and superimposes the objects in a common coordinate system. We generated thinplate spline deformation grids to visualize metasomal shape differences. Size and shape components of this configuration were separately analyzed. The size of the metasoma was measured as a centroid size (CS). The CS is a geometric scale which is mathematically defined by the square root of the sum of squared distances between all landmarks and their centroid (Zelditch et al. 2004).We collected 21 homologous landmarks on the metasoma using tpsDig2 (Rohlf 2006). The overall metasomal size variation has been presented as a box plot, while differences between populations have been tested with a Pairwise Analysis of Variance (ANOVA) using post-hoc Tukey Least Square Distance test.
Relationships between metasoma shapes were investigated using Principal Component Analysis (PCA). PCA was performed on all shape variables in order to define the greatest axes of metasoma shape variation in the dataset. The visualization of the shape differences was made with thin-plate spline deformation grids. Overall differences between metasoma shapes have been tested by a Multivariate Analysis of Variance (MANOVA) and the permutation tests (1000 permutations) for Mahalanobis distances among populations were performed to confirm the significant differences. The effect of size on shape was investigated by multiple regression.

Acronyms of collections
Colour: body black; antenna brown with reddish tint on some parts: radicle yellow with reddish tint; A1-5 with reddish tint on the ventral side; wing veins brown; legs light brown, sometime yellowish; middle of femora with dark tint.
Structure of ovipositor in T. dubia, shows this species was misplaced in Apegus by Kieffer, because in Apegus, the ovipositor has a completely different structure, being Ceratobaeus-type (Fig. 5d).
The aedeagus (Fig. 5a) has two parts: the basal ring and aedeago-volsellar shaft. The basal ring is well developed, and represents 0.4 of copulatory organ length and 0.7 of aedeago-volsellar shaft. The aedeago-volsellar shaft has two aedeagal apodemes and two digiti volsellares. Each digitus has a row of five pits, each with a short tooth. The digiti, teeth and aedeagal apodemes are darker, more sclerotized than the rest of the copulatory organ.
Biology. Triteleia peyerimhoffi is the third member in a tritrophic system, the other two being the plant-hosts and the orthopteran-host. We examined specimens of T. peyerimhoffi obtained from the following plants: Asphodelus sp. (Asphodelaceae); Ferula sp., Magydaris tomentosa (both Apiaceae), Tilia sp. (Malvaceae) and Quercus sp. (Fagaceae). In most cases, T. peyerimhoffi was obtained from the tettigoniids Uromenus brevicollis insularis or Platycleis albopunctata (Fig. 6). The relationship between Asphodelus ramosum -Uromenus brevicollis insularis -Triteleia peyerimhoffi (under the name T. dubia) was previously noted by Cavalcaselle (1968) and the relationship between Uromenus brevicollis -Triteleia peyerimhoffi, was noted by Silvestri (1939) and Petit et al. (2007). Triteleia peyerimhoffi was collected from the end of June until the first part of October, with peak numbers in August.

Taxonomic comments.
Kieffer did not appreciate the variability of this species since he described the female of this species in 1906 in Caloteleia; males two years later in Apegus; and the male and female again (1910b) as Ceratoteleia lugens. According to Kieffer (1910a) Ceratoteleia and Apegus are very close, differing in details of the female clava and first metasomal tergite. We did not find the type specimens of these species. It seems that the types of Ceratoteleia peyerimhoffi and Parapegus dubius are lost. We looked for them in the collections of BMNH (Popovici and Notton), MNHN (collections of Jean-Jacques Kieffer and Paul de Peyerimhoff de Fontenelle) (Dr. Masner, Dr. Fusu & Dr. Claire Villemant) and HNHM (Dr. Sandor Csősz) but without success. It is possible that the type specimens of Ceratoteleia lugens are in the Naturhistorisches Museum, Vienna or in the Museo Civico di Storia Naturale di Trieste, but we have not visited these collections (Dr. Dominique Zimmermann and Dr. Fusu looked for this species in the Naturhistorisches Museum, but without success). When describing the male of Ceratoteleia lugens, Kieffer mentioned: 'chez le mâle, tous les tergites sont transversaux, le 6 e porte de chaque côté de son bord postérieur un petit appendice'. The bidentate or bispinose last tergite is a character state confirming that this species belongs to Triteleia. This species was obtained from Foeniculum sp. (Apiaceae); the host plant of the tettigoniids which Triteleia peyerimhoffi is known to attack. From Kieffer's original descriptions it is impossible to find reliable characters to separate Ceratoteleia peyerimhoffi from Ceratoteleia lugens and from Parapegus dubius.
To clarify the taxonomic status of these species (ICZN, article 75.3.1), we here designate neotypes for Parapegus dubius and Ceratoteleia peyerimhoffi. We consider that the types of these species have been lost or destroyed: since we were unable to locate them in BMNH, MNHN, HNHM or in the Naturhistorisches Museum, Vienna. For Parapegus dubius we designate as neotype one female labeled: Hungary, Veröce 47°49.58'N, 19°1,30'E, 122m, 2-18.ix.2005, leg. Z. Nyiro (Malaise trap, CNCI). For Ceratoteleia peyerimhoffi we only have one male from Algeria. Because we have many specimens from Italy from the same host as the type specimens and because the male from Algeria is very similar to males from Italy, we decided to designate as a neotype a female of Ceratoteleia peyerimhoffi labeled: Italy, Guspini, 3.VIII.1933 (reared from Asphodelus). This neotype will be deposited in BMNH. Kononova and Petrov (2000) described a new Palaearctic species of Triteleia from southeast Bulgaria, T. striolata, based on two females, adding Israel to the distribution in 2008. Based on the description of its sculpture, ratios between sclerites, emergence dates, distribution and examination of pictures of the habitus, antenna and forewings of the holotype, we conclude that T. striolata is a junior synonym of T. peyerimhoffi.
Among the Palaearctic species described by Kieffer in Ceratoteleia there is one further species that has an uncertain status: Ceratoteleia mediterranea. Currently it is placed in Calliscelio (Johnson 1992;Kononova and Kozlov 2008), but we are convinced it is a Triteleia, and possibly another junior synonym of T. peyerimhoffi. We have not found the type specimens, although the senior author and Dr. Masner saw two females in MNHN identified by Maneval as C. mediterranea and the senior author saw a similar specimen in FBIN also identified as C. mediterranea. The main difference between these specimens and T. peyerimhoffi is the overall size and the ratio between the length and maximum width of the metasoma. It is possible these specimens are extreme examples of T. peyerimhoffi, but until we see more specimens we prefer to not include these specimens in T. peyerimhoffi.

Discussion
Triteleia peyerimhoffi is a widely distributed species in southern Europe and because of this wide distribution it appears to form local races which appear different. Therefore, it is easy for researchers working with a limited series of specimens from a limited geographic region to misinterpret this intraspecific variability as representing additional species. Indeed, specimens from the extremes of the species' distribution may look like different species; however, studying a large number of specimens shows an almost continuous morphological gradation between these extremes. In other parasitic Hymenoptera PCA was successfully used to separate closely related species (Popovici and Buhl 2010;Fusu 2010;Polaszek 2004;Bernardo et al. 2008). We used PCA here to see if, and how morphology of this species varied with distribution.
The first six principal components of PCA explain 80.34% of the total variance (one indicated by Jolliffe cut-off value = 0.002; eigenvalues of the first six PCs > 0.002). 95% bootstrapped confidence intervals are given for the eigenvalues (Table 1). The first three principal components of this analysis are plotted in Figures 7a and 7b; they explain 67.35% (PC1), 6.99% (PC2) and 6% (PC3) respectively of the total variance. Both graphs of PC1 on PC2 and PC1 on PC3 displayed a trend of separation between specimens (Figs 7a, 7b). The contributions of variables in this dispersion are showed in the Table 2.
Because in both the graph of PC1 on PC2 and PC1 on PC3, but especially in PC1 on PC2 we obtained a relatively clear separation between the specimens from France and specimens from Italy, we wanted to see if there is a significant difference between these two populations.
One-way ANOVA of all variables indicated significant differences between populations for all variables excepting one (LT6 min: F= 1.865, p= 0.13). The analysis proceeded by pair-wise comparisons (post-hoc) using the Games Howel test after the Levene test revealed unequal variances (p <0.05). The Games Howel test indicated small but significant differences between populations for most variables (p<0.05) ( Table 3). The main differences appear between specimens from Italy and specimens from France, in this case almost all characters showed significant differences. Also there were a few significant differences between specimens from France and specimens from Greece (for six characters), between specimens from France and specimens from Hungary (for one character) and between specimens from France and specimens from Romania (for one character). There were no significant differences between specimens from France and the specimen from Croatia. Interestingly there weren't any significant differences between specimens from Italy and specimens from Greece, Croatia, Romania and Hungary. Also, there were no significant differences between specimens from Greece, Romania, Croatia and Hungary. These results demonstrate that all these specimens from different populations belong to the same species, specimens from France and specimens from Italy being at different extremes with populations from Greece, Croatia, Romania and Hungary, lying between these two populations. Because characters of the metasoma were very important in the PCA analysis we decided to analyze the size and shape of the metasoma in T. peyerimhoffi to reveal more detail of the variability within this species.
One way ANOVA found significant differences in centroid size between populations (F= 7.17; P = 0.001). The distribution of the log-transformed centroid size is shown in Fig. 8. Levene's test revealed a normal distribution of all populations (p>0.05) and Tukey's pairwise mean comparisons showed a significant difference between three of them (Table 4). Again, in the analysis of the metasoma, there were some significant differences between France and Italy, Greece and Romania and between France and Greece. There were no differences however between Hungary and Romania or between Hungary and Greece, no difference between Italy and Romania or between Italy and Greece, no difference between France and Romania, or between specimens from France and Hungary or Croatia (Table 4). We conclude, as before, that although there are some significant differences between some populations, overall the specimens belong to the same species.
The result of analysis of the shape of the metasoma is shown on the plot of principal components. The first two principal components are shown in Figure 9. PC1 ac-  counts for 35.21% of the shape variability and PC2 explains 31.38% of the variability. Thin plate deformation grids show the transformation of shape along the two axes. The MANOVA permutation test found insignificant overall difference between shapes (p>0.05) of metasoma in the six examined populations. The result of this test, confirm our view that all populations belong to the same species. So, in Fig. 9 it is impossible to separate specimens from their provenance, because there is a mixture of specimens belonging to the different populations. Hence we consider this variation in size and shape of the metasoma to be intraspecific variability. Furthermore, we found an almost continuous gradation between the shortest and the longest metasoma (Fig. 10). Multiple regression of shape on size was performed. The results revealed that only 10.88% of variability of shape is predicted by size (Permutation test against the null hypothesis of independence/ Number of randomization rounds: 10000, P-value: <0.001).  We therefore assert that the great variability of this species to geographical variation. The correlation of geography with these variables was shown by Spearman's rank coefficient. There was a significant correlation between longitude and PC1 (Spearman's correlation = -0.52; p<0.01) and latitude and PC2 (Spearman's correlation = -0.62; p<0.01). Therefore, in the case of this species, there is a correlation between longitude and LA3, LT5, width of metascutellum, LT4, LE and LT6 and also, between latitude and minimum width of T6, length of T6, Lt, length of T3, length of T4, length of T2 and length of T5. The relation between latitude and longitude on PC1 and PC2 can be seen in Fig. 11A & B. So, the gap between populations on Fig.  7A and 7B can be explained as a gap between the sites where the specimens were collected. The majority of specimens from Italy were collected from Sardinia and just three from continental Italy. Similarly for the other material: from Greece we analyzed specimens from northern Greece only; from Romania, only specimens from the North-East; from France, only specimens from a limited area. Hence there are many unrepresented areas between the sampled populations. Consequently we think this is the reason for the gap between the populations from France and Italy. Also, we have few specimens from northern Italy, and this population is separated from that in France by the Western Alps.
A large variation in size, shape and sculpture of the metasoma is not something that is unusual within scelionid species, e.g., Vecher (1980) notes the intraspecific variability of the metasoma in Telenomus angustatus reared from tabanid eggs. He analysed 5274 females and 1009 males and noted great variability in the shape and size of the metasoma and in the sculpture of T1 and especially T2. The most interesting thing is that variability was observed in specimens which were the progeny of thelytokous females. He confirmed his conclusion that all specimens belonged to the same species with allozyme electrophoresis.
Within the Chalcidoidea, Popescu (2004) found a great variability in the size and shape of antennomeres in females of Idiomacromerus pallistigmus Askew and Eridontomerus arrabonicus Erdös (both Torymidae). The specimens were reared as from Blascoa ephedrae Askew (Pteromalidae) from Ephedra distachya L. (Ephedraceae) and Tetramesa scheppigi (Schlechtendal) (Eurytomidae) from Stipa lessingiana (Poaceae) respectively. Kononova (2001) gave great importance to the influence of hosts in the morphology of parasitoids. She wrote: 'adaptation of scelionids to hosts from more advanced orders cause sharp modifications of not only the habitus (strongly shortened body), but also some other morphological characters (e.g. shortening of the abdomen through reduction of its apical segments)'.
The influence of hosts on the morphology of parasitoids was demonstrated by Johnson et al. (1987). They studied the intraspecific variability in Telenomus alsophilae reared from different hosts in the laboratory under controlled conditions and empha- sised the strong influence of hosts in the morphology of the antennae of their parasitoids. Another interesting fact shown by Johnson et al. (1987) was that specimens reared from field collected eggs showed substantially greater coefficients of variation than the laboratory-reared specimens most likely as a result of uncontrolled environmental variables. It is very probable that T. peyerimhoffi has not just a single host, but uses a number of similar tettigoniid hosts and differences between the size and shape of the eggs of different host species, and differences between the size and shape of eggs within the same host species under different environmental conditions are a source of intraspecific variability within and between parasitoid populations.
We conclude that T. peyerimhoffi is a species with a wide circum-Mediterranean distribution and is one of the most important eggs parasitoids of Uromenus brevicollis. As an alternative host it uses Platycleis albopunctata which explains its presence in areas (e.g. Romania) where there is no Uromenus, but where Platycleis albopunctata common (Iorgu I., pers. comm.).