Integrative taxonomy of root aphid parasitoids from the genus Paralipsis (Hymenoptera, Braconidae, Aphidiinae) with description of new species

Abstract Species from the genus Paralipsis are obligatory endoparasitoids of root aphids in the Palaearctic. It is known that these species are broadly distributed, parasitizing various aphid hosts and showing great biological and ecological diversity. On the other hand, this group of endoparasitoids is understudied and was thought to be represented by a single species in Europe, viz., Paralipsisenervis (Nees). However, recent description of two new species indicated the possibility of cryptic speciation and recognition of additional Paralipsis species in Europe. In this research, Paralipsis specimens collected during the last 60 years from eight European countries, as well as one sample from Morocco, were subjected to molecular and morphological characterization. Newly designed genus-specific degenerative primers successfully targeted short overlapping fragments of COI of the mitochondrial DNA. Molecular analyses showed clear separation of four independent lineages, two of which are the known species P.enervis and P.tibiator, while two new species are described here, viz., P.brachycaudi Tomanović & Starý, sp. n. and P.rugosa Tomanović & Starý, sp. n. No clear specialization of the taxa to a strict root aphid host has been determined. The recognized mitochondrial lineages were distinct one from another, but with a substantial within-lineage divergence rate, clearly indicating the complexity of this group of parasitoids, on which further research is required in order to clarify the factors triggering their genetic differentiation. We reviewed literature data and new records of Paralipsisenervis aphid host associations and distributions. A key for the identification of all known Paralipsis species is provided and illustrated.


Introduction
Parasitoid wasps from the subfamily Aphidiinae (Hymenoptera, Braconidae) attack various aphid phylogenetic lineages, exhibiting several specialized associations with their hosts (Gagić et al. 2016). Few parasitoid wasps are specialized to parasitize only root aphid species (Starý 1961). However, there is no substantial biological knowledge about these obligatory parasitoid species of root aphids, probably as a consequence of them being of little economic importance and difficult to access for sampling. Moreover, it is well known that the parasitoids of root aphids developed obligatory relationships with ants (Starý 1966, Takada and Shiga 1974, Völkl 1992, Völkl et al. 1996. Although, there are many examples of relationships between ants that collected honeydew from aphids and protected the aphid colony and parasitoid wasps, it seems that chemical mimicry plays a more important role in some parasitoids [e.g., Lysiphlebus cardui (Marshall 1896)] than behavioural mimicry (Liepert and Dettner 1993). However, parasitoid wasps from the genus Paralipsis Foerster, 1863 have developed species-specific relationships with ants attending root aphids (Starý 1966, Takada and Shiga 1974, van Achterberg and Ortiz de Zugasti Carrón 2016. The genus Paralipsis is а good example as case of specific obligatory parasitoids of root aphids in Europe and the Palaearctic (Figure 1). Until recently, Paralipsis enervis (Nees) was considered to be the only European species, while P. eikoae (Yasumatsu) was known as a Far Eastern species (Starý and Schlinger 1967). However, after examining two samples of Paralipsis from Spain and the Netherlands, van Achterberg & Ortiz de Zugasti Carrón (2016) described two new species on the basis of morphological characters, viz., P. planus van Achterberg and P. tibiator van Achterberg and Ortiz de Zugasti Carrón. It is known that the genus Paralipsis shows great biological and ecological complexity and diversity in view of their acceptance of various aphid hosts and also having a broad geographical distribution.
Bearing in mind it was thought that the genus Paralipsis was represented by a single species in Europe, until additional two species were newly described recently, we started this research to address the possibility of cryptic speciation and recognition of additional Paralipsis species in Europe. Since rarely encountered, there is almost a complete lack of knowledge about morphology and reliable characters for diagnostics of these root aphid parasitoids.
A set of wasps collected during the last 60 years from eight European countries and Morocco were initially subjected to morphological characterization. In addition, DNA was extracted from available Paralipsis specimens to perform the amplification and sequencing of the mitochondrial DNA barcoding region of cytochrome c oxidase subunit I (COI). We developed DNA amplification protocol and designed new internal genus-specific degenerative primers in order to retrieve short overlapping COI fragments for molecular characterization of the wasps. Subsequently, we used an integrative approach analyzing the morphological and molecular results to recognize phylogenetic lineages and cryptic species within the analyzed Paralipsis specimens. Two new species in Europe were described. In addition, we reviewed the host aphids and distribution of associations for Paralipsis enervis. A new determination key including all previously known and two newly described species is provided and illustrated.

Insect material
We were provided with Paralipsis specimens collected during the last 60 years from eight European countries (Czech Republic, France, Germany, Lithuania, Moldova, Serbia, Slovakia, and Spain), in addition to one non-European sample from Morocco ( Figure 2). Material was obtained by rearing from 17 different plant/aphid trophic associations, which included specimens emerging from 13 different aphid hosts (Table  1). Additionally, the paratype of P. planus was provided by the Naturalis Biodiversity Center, Leiden, the Netherlands.  Our examination of Paralipsis specimens took into account reliable morphological characters used in aphidiinae taxonomy (number of flagellomeres, shape of flagellomere 1 and 2, number of labial and maxillary palpomeres, size and shape of fore tarsus, shape of hind tibia and femur, wing venation pattern, pterostigma shape, ratio between the pterostigma and radial vein 1, petiole shape, propodeal areolation, and ovipositor shape) (Kavallieratos et al. 2005, van Achterberg and Ortiz de Zugasti Carrón 2016. The morphological terminology used in this article for diagnostic characters of aphidiines is based on Sharkey and Wharton (1997).

DNA extraction, PCR, and sequencing
The barcoding region of the mitochondrial cytochrome oxidase c subunit I gene (COI) was chosen for phylogenetic study as a proven informative marker in species delineation for numerous aphidiines (Derocles et al. 2012  DNA extraction was conducted using a commercial DNeasy Blood and Tissue Kit (Qiagen Inc., Valencia, California, USA) following the manufacturer's instructions. Initially, we attempted to amplify the barcoding region of the COI gene from dry material using the standard primer pair LCO1490/HCO2198 (Folmer et al. 1994). Each reaction was carried out in a volume of 20 μl, according to the following protocol: i) initial denaturation 95 °C/5 min; ii) 35 cycles including three steps, viz., 1 min/94 °C, 1 min/54 °C, and 30 sec/72 °C; and iii) final extension at 72 °C for 7 min.
Since the standard primer pair failed to successfully amplify the barcoding region in more than three specimens, the next step was to test the suitability of the internal degenerative primers designed by Mitrović and Tomanović (2018) for dry museum specimens of other Aphidiinae genera. Partial success was achieved in such trials amplifying random fragments, but predominantly in the first 200-350 base pairs (bp) of the barcoding region, imposing the need to design new Paralipsis-specific primers with the aim to target the middle and last portions of mitochondrial DNA fragments. In the absence of reference COI sequences of Paralipsis parasitoids in the available public databases, we used our own sequences to design internal primers for dry material. These primers were positioned to amplify the missing fragments of COI, which could later be concatenated to longer barcoding sequences ( Figure 3). Trials of retrieving the COI barcodes included PCR reactions combining the standard primers LCO1490 and HCO2198, ones designed by Mitrović and Tomanović (2018), and newly designed Paralipsis-specific primers, targeting overlapping fragments of different lengths and positions (Table 2, Figure 3).
Products of PCR were obtained in 40 μl following the protocol described by Mitrović and Tomanović (2018). All barcoding products were sequenced with forward and reverse primers for each part of the barcoding region using automated sequencing equipment (Macrogen Inc, Seoul, South Korea). Short barcode fragments were manually edited in FinchTV ver. 1.4.0 (www.geospiza.com), aligned and concatenated using the ClustalW program integrated in MEGA5 (Tamura et al. 2011). Sequenced mitochondrial barcodes were subjected to maximum likelihood best fit model analysis using the MEGA5 program. According to the obtained Akaike information criterion scores, the best fit model to calculate evolutionary distances was the Tamura-Nei model (Tamura and Nei 1993).

Results
Barcoding fragments of COI were successfully recovered from 18 specimens. The material subjected to molecular analyses was of different ages in terms of the time passing between sampling until DNA extraction; several of the oldest had been preserved in collections for nearly 60 years. This probably caused DNA disintegration, which resulted in failed attempts to recover the barcoding region with the LCO1490/ HCO2198 standard primer pair. The newly designed Paralipsis-specific primers made it possible through diverse combinations to retrieve short subsequences of different length and position from disintegrated DNA of archival specimens. Prior to molecular analyses, all the barcoding sequences were aligned and trimmed to the same length of 568 bp. Comparison of COI barcodes identified 14 haplotypes (PH1-PH14) distinguished by a total of 83 variable sites, of which 51 were parsimony-informative (Table  3). The phylogenetic relationship was inferred using the MP and ML methods, which resulted in trees sharing identical topology with no substantial differences in bootstrap support ( Figure 4).
Phylogenetic analysis showed molecular differentiation on the basis of COI barcoding fragments, with recognition of four distinct lineages. The first group includes seven haplotypes: PH1, PH2, PH4, PH5, PH9, PH11, and PH13, which morphologically correspond to the first known species in this genus and in Europe, P. enervis. The specimens were sampled from different aphid hosts (Forda, Aphis, Anuraphis, Dysaphis) in association with different plants originating from Serbia, Germany, France, Lithuania, and the Czech Republic. The average overall divergence rate between the haplotypes within this group was 1%, with distances ranging from 0.4 to 2.5% (Table 4).
The second lineage, a "Mediterranean" clade, includes haplotypes PH12 and PH14 from Spain, and haplotype PH10 from Morocco. The overall divergence rate within this group was 2.8%. Genetic distances show that the haplotype PH12 associated with Forda formicaria is intermediary, diverging from the haplotype PH10 from Aphis rumicis (2.4%) and from the haplotype PH14 associated with Forda formicaria (2%), while the genetic distance between the other two was 4% (Table 4). Haplotype PH14 belongs to the paratype specimen of the newly described species P. tibiator. On the basis of morphological examination, it can be concluded that the haplotypes PH12 and PH10 belong to P. tibiator, although with evident high intraspecific genetic diversity. These three specimens clearly differ from the other congeners in having an elongated flagellomere 1 (F 1 ) (the ratio between F 1 and F 2 is 1.3-1.4) and a large number of longitudinal placodes on F 1 and F 2 in males (5-6 in P. tibiator versus 0-2 in other Paralipsis).
The third distinct lineage on the phylogenetic tree consists solely of the haplotype PH7, with unknown host data. The single specimen available from Moldova is characterized by having a very rugose and irregularly carinated propodeum. It is described as the new species P. rugosa sp. n., clearly separated genetically, with average distance from the first, second, and fourth lineage of 7.3, 7.7, and 9.6%, respectively.
PA26 MH475332 Figure 4. Phylogenetic relationship inferred using the maximum parsimony (MP) method. The consistency index is (0.533333), the retention index is (0.681818), and the composite index is 0.476540 (0.363636) for all sites and parsimony-informative sites. The MP tree was obtained using the subtreepruning-regrafting (SPR) algorithm with search level 1, in which the initial trees were obtained by the random addition of sequences (10 replicates). The percentage of replicate trees in which >50% of the associated taxa clustered together in the bootstrap test (500 replicates) is shown next to the branches (in red color). Since the topology is identical, the bootstrap support of branches obtained by the maximum likelihood method is presented in black color as well. Barcoding haplotypes of the analyzed archival Paralipsis specimens are designated with codes from PH1 to PH14, species name, host aphid and host plant.  Three haplotypes (PH3, PH6, and PH8) originating from Brachycaudus sp. and Tetraneura ulmi aphid hosts from Central Europe (Czech Republic and Germany) are grouped within the fourth distinct lineage. The barcoding haplotypes differ in the range of 1.8 to 3.6%, with an average overall interlineage divergence rate of 2.8% (Table 4). Specimens of Paralipsis within this lineage are characterized by having a more elongated petiole and ovipositor sheath in comparison with other congeners, and are described as the new species P. brachycaudi sp. n.

Paralipsis enervis -a review of host aphids and distribution of associations
The presented review includes evidence obtained for the most part from consulted published references about the species. The material was often re-visited, which was possible due to its preservation in available collections (IECR and FBS). The review also includes some new supplementary records (*).

Anoeciinae
Anoecia corni (Fabricius): Germany (Völkl et al. 1996). Anoecia sp.: England (Pontin 1960), Czech Republic (Starý 1961), France (Noury 1962 (Starý 1961(Starý , 2006. Brachycaudus cardui (L.): Czech Republic (Starý 1961(Starý , 2006 This integrated review contains broad information and also allows a cross-comparison of all the known host aphid-parasitoid locations of P. enervis in the Western Palaearctic. The true distribution range of P. enervis is somewhat more extensive than that derivable from the above review, since in most of the countries the parasitoid wasp was determined from individually sampled specimens with no data on the associated host aphids. Similarly, the distribution data reflect strength of the respective field research efforts. It seems that the northern distribution limits are the Scandinavian countries. The vertical distribution also manifests some peculiarities. Paralipsis enervis was also reared from the root aphid Dysaphis reaumuri sampled in the Alps (France) at approximately 2200 meters (see the review).

Descriptions of new species in Europe
On the basis of morphological examination of our available material from across Europe and the Mediterranean and using the COI mitochondrial barcoding marker, we confirmed the existence of the recently described Paralipsis species P. tibiator. In addition, two new Paralipsis species are described below. Diagnosis. The new species morphologically resembles P. enervis in petiole shape, absence of longitudinal placodes from flagellomeres 1 (F 1 ) and 2 (F 2 ), and fore wing venation pattern. Paralipsis brachycaudi sp. n. differs from P. enervis in having a longer petiole (Figure 13), the ratio between petiole length and width at the spiracle level is 1.50-1.60 in P. brachycaudi sp. n., while in P. enervis it is 1.30-1.40; somewhat shorter F 1 and F 2 ( Figure  7) (the ratio between length and width of F 1 and F 2 is approximately 2.00 in P. brachycaudi sp. n., as opposed to 2.20-2.30 in P. enervis); and a propodeum that is smooth with just a few rugosities at the side (Figure 9), while the propodeum in P. enervis sometimes possess rugosities in the central parts which indicate for the presence of a central areola. Additionally, F 1 and F 2 are light-brown to yellow in P. brachycaudi sp. n., while in P. enervis only half of flagellomere 1 is yellow and the remaining parts of the flagellomeres are brown. The ovipositor sheath in P. brachycaudi sp. n. (Figure 14) is more elongated than in P. enervis.
Mesosoma: Mesoscutum smooth, and only moderately sculptured within small central area, usually with four rows of setae along its dorsolateral part. Mesoscutum 1.4 times as long as wide. Scutellum (Figure 8) smooth elongated, bearing 20-30 long setae in the central part. Propodeum (Figure 9) smooth, sometimes with rugosities laterally. Upper and lower parts of propodeum with 3-5 and 15-20 long setae on each side (Figure 9). Fore wing ( Figure 12) densely pubescent, with long marginal setae, longer than those on fore wing surface. Vein 2-1A sclerotized (Figure 12). Pterostigma triangular, 1.7-1.9 times as long as its width (Figure 12). Second-fourth segments of fore tarsus in dorsal view ( Figure 10) almost as long as wide (1.1-1.2 times as long as wide) and medium sized of apical bristles. Hind tibia medially and femur subbasally parallel-sided (Figure 11).
Metasoma: Petiole (Figure 13) smooth, with prominent spiracular tubercles, its length 1.50-1.60 times its width at spiracles and maximum width at level of spiracles 0.7 times distance between spiracle and apex of tergite 1; 10-15 setae positioned on posterior dorsolateral margin on each side. Ovipositor sheath (Figure 14) elongated, dorsally straight, narrowed toward tip, bearing 2-6 long setae on the ventral and dorsal surface. Length of ovipositor sheath 2.25-2.87 times its maximum width.
Male: unknown. Etymology. The name of the new species is derived from that of its aphid host. Distribution. Czech Republic, Germany.

Diagnosis.
The new species differs clearly from all known Paralipsis species in having a strongly rugose propodeum ( Figure 18) and scutellum ( Figure 17) that are irregularly deep carinated, while other Paralipsis species are characterized by a smooth propodeum, sometimes with moderately expressed rugosities. Also, P. rugosa sp. n. is with F 1 longer than F 2 (the ratio between F 1 and F 2 is approximately 1.15) ( Figure 16), while P. enervis, P. brachycaudi sp. n., and P. planus have F 1 equal or subequal to F 2 . An exception is P. tibiator, which has much longer F 1 than F 2 (the ratio of F 1 and F 2 is about 1.4). Further, F 1 and F 2 are very short (proportion of length and width of F 1 and F 2 are 1.76 and 1.50, respectively) ( Figure 16).
Description. Female: Head rounded, smooth, narrower than mesosoma at tegulae, bearing dense setae ( Figure 15). Head 1.1 times wider than long medially. Eyes oval, small, with scarse and long setae. Tentorial index 0.67. Clypeus with ten long setae ( Figure 15). Maxillary and labial palpi with one palpomere each. Ocular-ocelar line: diameter of posterior ocellus: Postocelar line =12:4:14. Malar space: height of eye =11:13. Antenna 15-segmented, slightly thickened at apex (Figure 16). Scapus subapically with subparallel side at lateral view. Pedicel subspherical. F 1 (Figure 16) longer than F 2 and about 1.76 times as long as its maximum width at the middle, and F 2 and F 3 about 1.50 times as long as its maximum width at the middle. F 1 and F 2 without longitudinal placodes. Penultimate flagellomera 1.6 times as long as wide. F 1, F 2 and F 3 without, and F 4 with one short longitudinal placode. Flagellomeres covered uniformly with short appressed and semi-erect setae (Figure 16). Mesosoma: Mesoscutum smooth, sculptured laterally, with very dense setae laterally. Mesoscutum 1.4 times as long as wide. Scutellum subspherical, strongly rugose with about 15 setae ( Figure 17). Propodeum (Figure 18) extremely rugose. Upper and lower parts of propodeum with 5-6 long setae on each side. Fore wing ( Figure 21) densely pubescent, with long lower marginal setae, longer than those on fore wing surface. Pterostigma triangular, 1.62 times as long as its width. Vein 2-1A sclerotized. Metacarpus absent. Second-fourth segments (Figure 19) of fore tarsus in dorsal view distinctly longer than wide (1.4-1.8 times as long as wide) and medium sized of apical bristles. Hind tibia medially and femur subbasally parallel-sided ( Figure 20).

Discussion
We have demonstrated here a progress in methodology of DNA amplification by designing Paralipsis-specific degenerative primers to retrieve disintegrated DNA fragments from archived museum specimen collections of which can be considered as biobanks and used to discover new species (Yeates et al. 2016). Sequencing of the COI barcoding region of 18 specimens collected across the Western Palaearctic over a long period of time did not determine any clear specialization of taxa to a strict root aphid host. There is no geographical structuring of genetic variation between specimens associated with the same aphid host within the lineages P. enervis, P. tibiator, nor P. brachycaudi sp. n. However, it confirmed the existence of Paralipsis tibiator, a species recently described by van Achterberg and Ortiz de Zugasti Carrón (2016). Although a second recently described species, viz., P. planus (van Achterberg and Ortiz de Zugasti Carrón 2016), was not available for molecular analysis, the general morphological description (petiole shape, wing venation pattern, antennae) indicates that it is close to P. enervis, so we sup-pose it belongs to the P. enervis lineage. However, since P. planus was described on the basis of a single specimen, it is necessary to further explore the morphological and genetic variability of this species in order to finally confirm its taxonomic status. In addition, the present study revealed two new Paralipsis species, P. rugosa sp. n. and P. brachycaudi sp. n. Four separate phylogenetic lineages showed clear distinction, but with significant intralineage genetic variation between the haplotypes associated with different aphid/host associations. All phyletic lineages share aphid hosts from the subfamilies Aphidinae and Eriosomatinae. Many Eriosomatinae are specialized gall-producing aphids, but only on primary host plants, while this is not the case on secondary host plants (grasses), where they are parasitized by Paralipsis wasps and other specialized root aphid parasitoids. It is necessary to examine in detail all known records of P. enervis in the light of the diagnosis given for the new Paralipsis species described in the present paper. Probably, P. enervis represents a complex of cryptic species, which is a common case among aphid parasitoids (Mitrovski-Bogdanović et al. 2013, Derocles et al. 2016. Although our molecular analyses were restricted to only 18 COI barcoding sequences retrieved from dry Paralipsis specimens, we recognized four phyletic lineages on the phylogenetic tree with a sister position of P. brachycaudi sp. n. in relation to P. enervis + P. rugosa sp. n. + P. tibiator lineages. The strong rugosities of the propodeum and scutellum in P. rugosa sp. n. represent an autapomorphic character state, while its possession of very short flagellomeres is a plesiomorphic character state. We recognize the elongated ovipositor sheath and petiole in P. brachycaudi sp. n. as apomorphic characters. We did not find any strong support for the existence of Paralipsis host-specific lineages. Brachycaudus aphid hosts were found only in the P. brachycaudi sp. n. lineage, while other aphid hosts are mainly shared between P. enervis and P. tibiator. Although Forda root aphids are distributed throughout the whole of Europe, P. tibiator attacked them only in Mediterranean-type habitats. It is known that the distribution of parasitoids is usually narrower than that of their aphid hosts, due to the more specific habitat and microhabitat of parasitoids (Starý 1970). The records of P. enervis associated with Brachycaudus aphid hosts should be carefully examined, as they may be referable to P. brachycaudi sp. n. All findings of P. tibiator are from Mediterranean areas.
Although most of our samples originated from central and southern Europe, Paralipsis species are distributed in several European countries, including ones in the northern part of the continent (van Achterberg 2012, Staverløkk and Ødegaard 2016). However, no Paralipsis species have been recorded in more than half of the countries of Europe (van Achterberg 2012). In the present study, we have not explored the relationships between ants and Paralipsis wasps. However, future research should reveal existing relationships of the two newly described species with ants and result in new knowledge about the biology and ecology of these Paralipsis wasps.
sented research. We thank Frédérique Bakker (Collection Manager Hymenoptera, Naturalis Biodiversity Center, Leiden, The Netherlands) for loaning us the P. tibiator paratype. This study was supported by the Ministry of Education, Science, and Technological Development of the Republic of Serbia (III43001). P Starý was supported by the Institute of Entomology, Biology Centre of the Czech Academy of Sciences. The sampling of radicicolous aphids on the Iberian Peninsula was conducted in the context of Project CGL2015-68188-P, funded by "Ministerio de Economia, Industria y Competitividad" of Spain (MIMECO).