The relationships within the Chaitophorinae and Drepanosiphinae (Hemiptera, Aphididae) inferred from molecular-based phylogeny and comprehensive morphological data

The Chaitophorinae is a bionomically diverse Holarctic subfamily of Aphididae. The current classification includes two tribes: the Chaitophorini associated with deciduous trees and shrubs, and Siphini that feed on monocotyledonous plants. We present the first phylogenetic hypothesis for the subfamily, based on molecular and morphological datasets. Molecular analyses were based on the mitochondrial gene cytochrome oxidase subunit I (COI) and the nuclear gene elongation factor-1α (EF-1α). Phylogenetic inferences were obtained individually on each of genes and joined alignments using Bayesian inference (BI) and Maximum likelihood (ML). In phylogenetic trees reconstructed on the basis of nuclear and mitochondrial genes as well as a morphological dataset, the monophyly of Siphini and the genus Chaitophorus was supported. Periphyllus forms independent lineages from Chaitophorus and Siphini. Within this genus two clades comprising European and Asiatic species, respectively, were indicated. Concerning relationships within the subfamily, EF-1α and joined COI and EF-1α genes analysis strongly supports the hypothesis that Chaitophorini do not form a monophyletic clade. Periphyllus is a sister group to a clade containing Chaitophorus and Siphini. The Asiatic unit of Periphyllus also includes Trichaitophorus koyaensis. The analysis of morphological dataset under equally weighted parsimony also supports the view that Chaitophorini is an artificial taxon, as Lambersaphis pruinosae and Pseudopterocomma hughi, both traditionally included in the Chaitophorini, formed independent lineages. COI analyses support consistent groups within the subfamily, but relationships between groups are poorly resolved. These analyses were extended to include the species of closely related and phylogenetically unstudied subfamily Drepanosiphinae, which produced congruent results. Genera Drepanosiphum and Depanaphis are monophyletic and sister. The position of Yamatocallis tokyoensis differs in the molecular and morphological analyses, i.e. it is either an independent lineage (EF-1α, COI, joined COI and EF-1α genes) or is nested inside this unit (morphology). Our data also support separation of Chaitophorinae from Drepanosiphinae.


Introduction
Remaudière and Remaudière [1] and this division is now generally accepted. The last genus to be incorporated into the Chaitophorini was Chaitogenophorus [2].
The question is whether this classification reflects the evolution of this group of aphids. A high level of polymorphism, morphometric differences in the spring and autumnal viviparous generations of the same species, connected with the presence in the life cycle of aestivating morphs (dimorphs) (e.g. Periphyllus, Trichaitophorus) or cryptic mode of life (e.g. Siphini) make the Chaitophorinae a difficult subject for study and the main reason for the taxonomic confusion in this subfamily [8,23,24]. Altogether Chaitophorinae have been previously studied morphologically and local faunas reviewed [25, 10,26,11,27], only the Palaearctic species in the genus Chaitophorus have been revised [28] and a monograph on the tribe Siphini published [8]. Data on the relationships within Chaitophorinae are rare [29][30][31], including cladistics analyses [8]. No phylogenetic studies on the Chaitophorinae as a whole have been published. The phylogenetic trees based on nuclear and mitochondrial genes [32][33][34][35] or the DNA of the obligate symbiont Buchnera aphidicola [36] included limited sampling of Chaitophorinae (five of the 170 described species). As these studies focused on higher relationships within Aphididae and only one species of the genera Periphyllus and Chaitophorus or Chaitophorus and Sipha (never combined representatives of both Chaitophorini and Siphini) were studied, the tribal status is untested. Wieczorek and Kajtoch [37] on the other hand, explored the phylogeny of Siphini using molecular data together with morphological and biological characters, and included the genera Periphyllus and Chaitophorus as outgrups. In this paper the monophyly of Siphini was confirmed, however the Chaitophorini did not form a monophyletic lineage. The unexpected result of these studies is that Chaitophorus may be more closely related to the monocotyledonous feeding Siphini than the Acer feeding genus Periphyllus. Thus, it is now important to test the monophyly and major relationships of Chaitophorinae using a broad taxonomic sample and analyzing mitochondrial and nuclear genes, morphological and biological dataset.
In the present paper, we also extend our analysis by including the Drepanosiphinae (with one tribe the Drepanosiphini), which is another poorly studied subfamily of aphids, closely related to Chaitophorinae. Drepanosiphinae comprises five genera and about 40 species, all associated with Acer spp. The genera Drepanaphis (17 species) and Shenahweum (one species) are Nearctic, genera Drepanosiphoniella (four species) and Drepanosiphum (nine species) are west Palaearctic, whereas the genus Yamatocallis (eight species) is distributed in eastern Asia. The aphids live on leaves, usually not in dense colonies and are not attended by ants. All known species are monoecious and holocyclic [4][5][6]. Among Drepanosiphinae the interspecific relationships within Drepanaphis are known [38]. Recently, the taxonomic revisions of Shenahweum [39], Drepanosiphoniella [40] and Drepanosiphum [41] have been published. Although Drepanosiphum platanoidis (Schrank, 1801) is a model species in numerous studies on the ecology of aphids [42][43][44][45][46][47], relationships within this subfamily are unstudied. The collective evidence from aphid parasites [48], morphology of extant taxa [49][50] and fossils [51] indicate that Drepanosiphinae and Chaitophorinae are sister groups. Molecular data, however poorly sampled as only two species of Drepanosiphinae were studied, supports this view [32][33][34], or even indicate that the Drepanosiphinae and Chaitophorinae could be combined in a single unit [52].
Chaitophorinae and Drepanosiphinae are one of the major lineages within the Aphididae. It is important to determine the relationships within as well as between particular genera in these subfamilies. This analysis presents the first phylogenetic hypothesis including both subfamilies. Compared with our previous study [37], an expanded dataset was used to test: (1) whether Siphini and Chaitophorini are mutually monophyletic within the subfamily Chaitophorinae and Drepanosiphini within Drepanosiphinae; (2) the taxonomic positions of some genera and their redefinition; (3) the hypothesis that Drepanosiphinae and Chaitophorinae could be combined into a single unit. We used sequences obtained from the mitochondrial gene cytochrome oxidase subunit I (COI) and the nuclear gene elongation factor-1α (EF-1α), supplemented by 91 morphological and biological characters, to reconstruct the relationships between these aphids.

Molecular data
Taxon sampling. Sequenced taxa. We obtained molecular data for a total of 36 species. The specimens used for molecular studies were preserved in 99.8% ethanol. Specimens from the same clones were also preserved in 70% ethanol for producing slides of voucher specimens and identification. The specimens were mounted in Berlese liquid on slides. Voucher specimens for each sample were identified by K. Wieczorek based on morphological diagnostic features using standard literature-based keys and a comparison with previously identified specimens kept in the University of Silesia, Department of Zoology, Katowice, Poland (UŚ). All samples and voucher slides were also deposited in the collection of UŚ. Most sequences were directly obtained from the collected specimens. Details of the sequenced taxa, voucher information and numbers of GenBank sequences for all the species studied (both downloaded and newly submitted) are presented in S1 Table. Ingroup. 25 species belonging to eight genera of Chaitophorinae were included in this study. All genera of Siphini (five) and three (of seven total genera) of the larger genera of Chaitophorini were sampled. There are only one or two rare species in each of the genera Lambersaphis, Yamatochaitophorus, Chaitogenophorus and Pseudopterocomma and they occur only at a few locations and therefore they were not included in the molecular study.
Genes. Molecular analyses were based on the mitochondrial gene cytochrome oxidase subunit I (COI) and the nuclear gene elongation factor-1α (EF-1α). Based on previous molecular studies on aphids, a mitochondrial gene was selected to provide resolution at lower taxonomic levels (generic and specific) [53][54][55][56], whereas a nuclear gene was used to provide a deeper resolution within the subfamily [57][58][59][60].
DNA extraction, amplification and sequencing. Genomic DNA was extracted from one to three individuals from the same colony of each species using the NucleoSpin Tissue Kit (Macherey-Nagel, Germany). Amplification of the partial mitochondrial COI and the nuclear EF-1α was done using the following pairs of primers, respectively: LepF and LepR [61] and Shirley and Prowler [62]. PCR was done in 30-μL reaction volumes with 3.0 μL of 10 × PCR buffer, 3.0 μL of 25 mm MgCl2, 0.6 μL of deoxynucleotide triphosphate (dNTP) mixture, each in a 10 mm concentration, 0.6 μL of each 15 mm forward and reverse primers, 1.0-3.0 μL of 100 ng of genomic DNA, 0.2 μL of Taq DNA polymerase and sterile and de-ionized water (up to 30.0 μL). The cycling profile for the PCR was as follows: 95˚C for 4 min, 35 cycles of 95˚C for 30 s, 50˚C (for COI) 55˚C (for EF-1α) for 1 min, 72˚C for 2 min and a final extension period at 72˚C for 10 min. After purification (NucleoSpin Extract II (Macherey-Nagel)), the PCR fragments were sequenced using a BigDye Terminator v.3.1. Cycle Sequencing Kit (Applied Biosystems, Foster City, CA, USA) and ran on an ABI 3100 Automated Capillary DNA Sequencer.
Sequence edition and alignment. Sequences were checked and aligned using BIOEDIT v.7.0.5.2 [63] and CLUSTALX [64]. No stop codons or indels that would indicate the presence of nuclear pseudogenes were found in the mitochondrial protein-coding genes. Introns in EF-1α sequences contained a large number of variably sized indels, which were removed prior to further analysis. All sequences were deposited in GenBank, and accession numbers are given in S1 Table. The Akaike Information Criterion in MrModeltest 2.3 [65] in conjunction with PAUP Ã [66] was used to determine the best-fitting nucleotide substitution model. The GTR+I +G model was chosen for COI (proportion of invariable sites I = 0.5890, gamma distribution shape parameter G = 1.1591), the GTR+I+G model for EF-1α (proportion of invariable sites I = 0.5035, gamma distribution shape parameter G = 0.7386) and GTR+I+G for joined COI and for EF-1α alignments (proportion of invariable sites I = 0.5049, gamma distribution shape parameter G = 0.7605). The final dataset used for phylogenetic analyses contained 634 bp of COI and 458 bp of EF-1α. As the sequences generated from individuals from a single colony were identical (for all species we recorded only single haplotypes for COI and EF-1α), for further analyses only a single sequence of each gene was used. We managed to obtain COI sequences for 27 species and EF-1α sequences for 31 taxa. The coverage of species using both markers was only partial due to difficulties in obtaining PCR products for some species, mainly when using COI primers. Attempts to use standard barcode primers (LCO1490 and HCO2198; [67]) also failed to generate amplicons for these species.

Morphological and biological data
Taxon sampling. Ingroup. 29 species belonging to all genera of Chaitophorinae (with exception of Chaitogenophorus for which there were no samples) were included in this study (Fig 1A-1J).
Morphological and biological characters. A total of 91 characters were scored for 43 species, including 83 morphological and eight biological characters. Morphological characters were evaluated for viviparous (apterous and alate females) and sexual (oviparous females and males) generations of the species studied (from five to 15 individuals). The characters used for the morphological analysis include previously published characters [8,37] as well as a number of newly developed ones. Characters of specimens viewed under a Nikon Ni-U light microscope and photographed using a Nikon DS-Fi2 camera or a scanning electron microscope were scored. For the SEM photographs, individuals collected in the field were preserved in 70% ethanol for several days and prepared following a mofidied Kanturski et al. [68] method as follows. The samples were transferred into 6% phosphotungstic acid (PTA) solution in 70% ethanol for 24 hours. Dehydration was achieved by using a graded ethanol/water series of 80%, 90% and 96% with 20 minutes at each concentration and 30 minutes in two changes of absolute ethanol. Dehydrated specimens were subsequently dried in 1:3, 1:2 and 2:3 ratio solutions of hexamethyldisilazane (HMDS) in absolute alcohol for 30 minutes and two changes in undiluted HMDS. Samples were mounted on aluminum stubs using double-sided adhesive carbon tape and sputter-coated in a Pelco SC-6 sputter coater (Ted Pella Inc., Redding, CA, USA) to  Table. The complete matrix is presented in S2 Table and S3 Table. Descriptions of character are as follows: Apterous viviaparous females.   Phylogenetic analyses. Molecular dataset. Phylogenetic inferences were obtained individually for each of the genes and for joined alignments using Bayesian inference (BI) and Maximum likelihood (ML). BI was run using MrBayes 3.1 [69][70] with 1 cold and 3 heated Markov chains for 10 000 000 generations, and trees were sampled every 1000th generation. Each simulation was run twice. Convergence of Bayesian analyses was estimated using Tracer v. 1.5.0 [71], all trees sampled before the likelihood values stabilized were discarded as burn-in, and the remainder used to reconstruct a 50% majority rule consensus tree. ML analyses were implemented in RAxML 7.2.6 [72][73] with the GTR+I+G model and the same model parameters as in the Bayesian analyses. Branch support for ML analyses was assessed by bootstrapping with 1000 replicates. All trees were visualized using TreeView 1.6.6 [74].
Morphological dataset. Morphological analyses were rooted with H. betulinus. Datasets were analyzed with MP under equal weights using TNT v1.1 [75]. New technology searches were applied consisting of 10 000 random addition sequence replicates and TBR. (TBR) branch swapping. Clade support was assessed with 1000 replicates of the bootstrap [76]. Bayesian analyses were also performed in the same way as described above.

Molecular data analyses
The topologies of both, Bayesian and Maximum Likelihood trees generated for COI gene, EF-1α gene and for joined sequences were congruent for particular markers, therefore only BI topologies are presented on figures (Figs 9-11), with added values of bootstraps from analogous nodes on ML trees.
Analysis of EF-1α gene resolved five clades with most nodes highly supported (Fig 9). The genus Periphyllus was recovered as paraphyletic. Two clades included the following species of Periphyllus: i) European species P. coracinus (Koch, 1854), P. lyropictus Kessler, 1886, P. hirticornis Walker, 1848 and ii) Asiatic species P. koelreuteriae (Takahashi, 1919), P. californiensis (Shinji, 1917), P. acerihabitans Zhang, 1982, P. testudinaceus (Fernie, 1852). The latter includes also Trichaitophorus koyaensis Takahashi, 1961  ) was found to be an independent phylogenetic lineage of the remaining Drepanosiphinae and the clade constituted by Chaitophorinae.   COI was sequenced for relatively fewer taxa than EF-1α. The COI tree (Fig 10) confirmed paraphyly of Periphyllus with the same species clustering as in the EF-1α tree (data for T. koyaensis not included). The monophyly of the genus Chaitophorus was also supported by the COI tree, however this genus was clustered with Asiatic Periphyllus, but with rather weak support. Siphini aphids, based on mtDNA, seemed to be sister to the Chaitophorini clade but with very weak support (0.37 PP). Subfamily Drepanosiphinae was again found to be sister to the Chaitophorini-Siphini clade and divided into two subclades, much as in EF-1α tree. Yamatocallis tokyoensis was most distant in COI tree and formed a third lineage next to the Drepanosiphinae, Siphini and Chaitophorini lineages.
Phylogenetic trees constructed on joined datasets (COI and EF-1α genes) (Fig 11) showed generally similar topologies like abovementioned trees. Monophyly of Drepanosiphinae could not be confirmed on phylogenetic trees based on joined sequences. On the other hand Chaitophorini-Siphini was very well supported (1.00 PP). Moreover, within this clade, three phylogenetic lineages were clearly confirmed (all with 1.00 PP): i) European species of Periphyllus, ii) Asiatic species of Periphyllus and iii) Chaitophorus with Siphini. The third clade constituted with also two well supported lineages (both 1.00 PP): members of Chaitophorus and sister to them-Siphini species.

Morphological data analyses
The morphological analysis includes representatives of all genera of Chaitophorinae (with exception of Chaitogenophorus) and Drepanosiphinae. The morphological analysis conducted in TNT (consensus on six trees Fig 12) corroborate monophyly of Siphini and paraphyly of Periphyllus, with the same subdivision of the studied taxa studied as in the earlier analysis. However, the species belonging to the genus Chaitophorus, on the basis of one synapomorhy (the shape of siphunculi), were clustered with the genus Periphyllus. Both were sister to the species from the tribe Siphini. L. pruinosae and P. hughi (not included in molecular analysis) formed independent lineages with T. koyaensis as a sister to the latter lineage, whereas Y. albus (all Chaitophorini) was nested inside Siphini. Drepanosiphinae, on the basis on two synapomorphies, formed a clade independent of the remaining taxa with Y. tokyoensis nested inside this unit. Bayesian tree resulting from the Bayesian analysis of morphological dataset with weak of supports makes its interpretation too speculative (S1 Fig).

Comparison of the tribal and generic relationships with historical classification of Chaitophorinae
The major point of discordance between our molecular data and the classical taxonomy of Chaitophorinae is the tribal division of this subfamily. Analysis of the EF-1α gene and joined datasets of COI and EF-1α genes, highly supported closer relationships between Chaitophorus and the genera included in Siphini. The COI trees are much less resolved as it constitute of several parallel phylogenetic lineages which group independently: Siphini, Chaitophorus, European Periphyllus and Asiatic Periphyllus (Fig 10). However, in all molecular-based trees Chaitophorini do not form a monophyletic clade. Another result that contrasts with historic views is the position of the genus Periphyllus.
Taxonomists have found this highly polymorphic genus consisting of 42-44 species [4] confusing for many years. In the life cycle of most species of this genus about 15 types of morphs (including aestivating highly specialized first instar nymphs) among the normal generations of individuals have been identified [25]. Next to these unique biological features, recognition characters for this genus includes abdominal tergum membranous in wingless viviparous females, dorsal cuticle smooth, presence of spinules on distal part of tibiae and conical siphunculi with distinct reticulation. Our analyses of both molecular and morphological data  show that this taxon is paraphyletic. The European species analyzed (P. coracinus, P. lyropictus, P. hirticornis) were clustered together as were the Asiatic species (P. koelreuteriae, P. californiensis, P. acerihabitans, P. testudinaceus). In the latter clade (and within the genus as a whole) P. californiensis and P. testudinaceus are the most widely distributed in the Northern Hemisphere and recorded from the most species of maples [5]. Interestingly, both species are abundant on the Pacific Coast and rarely collected along the Atlantic Coast and elsewhere in eastern North America [25]. The diameter of primary rhinarium on penultimate antennal segment, length of basal seta of antennal segment VI and number of ventral setae are, among others, morphological characters clearly distinguishing two clades of Periphyllus. Our study points to a potentially interesting feature of Periphyllus but firm conclusions about the evolutionary patterns of species in this genus cannot yet be made because our analysis did not include members of the genus native to Central Asia nor the Nearctic.
Among all the taxa of Chaitophorinae studied Trichaitophorus koyaensis seems to be the most closely related to Periphyllus (Fig 9). Trichaitophorus seems to be even more polymorphic than Periphyllus, with complicated life cycles and numerous intermediate morphs characterized by variation in setal length and shape, as well as winged females strongly resembling Periphyllus [23,[96][97]. On the other hand wingless viviparous females are characterized by unique characters like 5-or 4-segmented antennae, fused head and pronotum, very short and sparse dorsal setae in exception of lanceolate marginal ones, as well as slightly elevated siphunculi. Further field studies, in conjunction with laboratory analysis of additional species, morphs and genes, may ultimately show that the two genera are not justified as currently structured.
Chaitophorus is the largest genus within the Chaitophorinae, recognized by dorsum sclerotized with a distinct reticulation and, with some exceptions (e.g. Ch. populicola Thomas, 1878), a knobbed cauda. Our analysis supports the monophyly of this genus, however the hypothesis that willow-feeding species are monophyletic within Chaitophorus and separate from poplar-feeding species is not supported. Moreover, the species with a certain feeding position i.e. leaf-feeding species versus petiole-feeding species do not constitute clear clades as well as ant-attended versus not ant-attended species. Shingleton and Stern [98] constructed a molecular phylogeny of 15 species of Chaitophorus based on mtDNA sequences and obtained similar results. Our research based on a different set of species and molecular markers is congruent with the more general theory that during evolution Chaitophorus has several times switched host plants (from poplars to willows), feeding position and ant tending [98]. The high plasticity of this genus is also reflected in variation in the shape and thickness of the dorsal setae, which is correlated with the seasonal development and distribution of particular species [26][27]99]. Chaitophorus, with its ability to switch host plants and feeding position in the course of evolution, is considered to be the ancestral form for the Chaitophorinae [8,[29][30]. However, lack of fossil evidence makes discussion of the origin of the subfamily somewhat hypothetical.
As our research are based on a different set of species in molecular (eight of twelve total genera) and morphological (eleven of twelve total genera) analysis is difficult to directly compare the obtained results. In particular, the position of Lambersaphis and Pseudopterocomma, traditionally included in the Chaitophorini, is not well-justified. The forewing veins of the species in both genera characteristically have a pigmented border like the Nearctic species of Chaitophorus and similar affinities to host plants, however they are placed in the most isolated positions in the cladogram (Fig 12). Moreover, Pseudopterocomma is characterized by unique set of features like processus terminalis covered in numerous, fine, hair-like setae, presence of secondary rhinaria on antennal segment III and IV both in winged and wingless viviparous females, as well as porous, not reticulated siphunculi. Similarly, Lambersaphis is characterized by very short processus terminalis, slightly elevated siphunculi without reticulation and short, needle-like dorsal setae. The position of Yamatochaitopho rus albus (Takahashi, 1961) also remains unclear, as this species was nested inside Siphini clade (Fig 12). Traditionally this genus is placed close to Trichaitophorus, as these two genera share similar morphological characters and differ by pattern of dorsal chaetotaxy.

Relationships within Drepanosiphinae
Our analyses of mitochondrial and nuclear genes resulted in a stable phylogenetic reconstruction with well-supported clades (Figs 9-11). In the genus Drepanosiphum, D. oregonensis and D. platanoidis were clustered together and sister to D. aceris. Drepanaphis species (D. parva and D. acerifoliae) also were clustered together and sister to Drepanosiphum. Although morphological analysis comprises representatives of all genera of Drepanosiphinae, the combination of synapomorphies: enlarged fore (Fig 8C and 8D) or mid femora and presence of rastral spines on hind tibiae (Fig 5A-5C), also supports this division (Fig 12). The only exception is the position of Yamatocallis tokyoensis. In the molecular analysis this species was placed in an independent lineage far from the remaining Drepanosiphinae. In our morphological analysis, on the other hand, Y. tokyoensis is nested within Drepanosiphinae, with a sister relation to species of Drepanaphis (Fig 12), which is congruent with traditional taxonomy. Its position is supported by one synapomorhy: pigmented forewings (Fig 2D). Originally, members of Yamatocallis were placed in the Nearctic Drepanaphis [100][101], as in general appearance species of these genera are similar. However, a combination of characters like accessory rhinaria located close to the major rhinarium, abdomen without dorsal tubercles and elongated siphunculi with reticulated apex (Fig 7N), clearly distinguish Yamatocallis from this genus and other taxa of Drepanosiphinae. In addition, Fukatsu [102] reports the secondary intracellular symbiotic bacterium YSMS in Y. tokyoensis (and Y. hirayamae Matsumura, 1917), which is treated as conserved throughout the evolution of the genus. Our molecular analyses show that Yamatocallis is farther from other species of this subfamily than previously thought [4,22]. The presence of this unique secondary mycetocyte symbiont, whose time of acquisition was estimated as the Miocene, may also indicate the separation of this genus. In this epoch dramatic geological and climatic changes took place. Isolation of Eastern Asia by the uplift of the Himalayas fits with the hypothesis that Yamatocallis was isolated from the other Palaearctic and Nearctic species of Drepanosiphinae.

Drepanosiphinae versus Chaitophorinae
Despite the increased use of molecular methods in phylogenetic analyses, morphology continues to play a significant role in the understanding of the evolutionary biology and systematics of many groups of organisms [103]. According to Quednau's hypothesis [50], based on morphological and biological characters, Drepanosiphinae evolved as a sister group of the Chaitophorinae and probably have a common ancestral form in Taiwanaphis-or Monaphis-like aphids. Close relationships between these subfamilies are reflected in the similarities in their morphology (i.e. absence of sclerotisation of segment II of the rostrum, absence of wax glands), anatomy (i.e. gastrointestinal tract without a filter chamber [104]), similar internal male reproductive system [31] and male genitalia [105] or bionomy (associations with host plants, similar type of summer diapause). According to this hypothesis (also indicated in Fig 12), during their evolutionary scenario, representatives of Drepanosiphinae probably lost some apomorphic features and became Periphyllus-like (Chaitophorinae). The intermediate characters between species of Chaitophorinae and Drepanosiphinae occur in the representatives of the genus Drepanosiphoniella, i.e., presence of apterous morphs in the life cycle (Fig 2C), nude primary sensoria, (Fig 4D) or lack of leaping legs (Fig 8B), features common in most Chaitophorinae. As representatives of Drepanosiphoniella were not included in the molecular studies, the position of this genus can only be discussed based on the morphological and biological characters. Currently fossils of about eight genera and 20 species of Drepanosiphinae are described (Eocene, Middle Miocene), but only one fossil of Chaitophorinae (Chaitophorus salijaponicus niger Mordvilko, 1929) is known from the Late Pliocene-Early Pleistocene (Peary Land, Greenland) [106]. Its also supports the hypothesis that Drepanosiphinae are an independent lineage within drepanosiphine aphids (sensu Quednau, [50]), which is also congruent with the biological data. At least Drepanosiphum has several highly specialized parasitoids whose life cycles are closely synchronized with the life cycle of the aphid-host [107] and this relationship developed a long time ago in parallel during the evolution of both insects [48].

Conclusions
The generally accepted view of the classification of Chaitophorinae features the strict subdivision of two bionomic groups-monocotyledonous feeding Siphini and deciduous tree or shrub feeding Chaitophorini. Commonly accepted diagnosis define the Chaitophorini includes 6-segmented antennae and elevated siphunculi with reticulated apices. Due to this fact, Lambersaphis and Pseudopterocomma should be excluded from the Chaitophorini, as both have rather short, even pore-shaped siphunculi without reticulation and in Pseudopterocomma the antennae of the wingless viviparous females are 6-or 5-segmented, characters more closely fitting Siphini. Genera Trichaitophorus and Yamatochaitophorus, both included in Chaitophorini, are characterised by 6-or 5 (4)-segmented antennae and short siphunculi in wingless viviparous females whereas winged females have 6-segmented antennae and elevated and clearly reticulated siphunculi thereby strongly resembling Periphyllus. Therefore, the number of antennal segments and reticulation of siphunculi should not to be treated as good characters for tribal subdivision.
Our molecular analyses, supported by morphological and biological data, revealed at least four clades within Chaitophorinae: (1) Siphini closely related to (2) Chaitophorus, (3) paraphyletic Periphyllus with Trichaitophorus (and Yamatochaitophorus) and (4) the most distant Lambersaphis and Pseudopterocomma. All of these genera share the presence of four gonaphophyses, which is also the synapomorphy for Chaitophorinae as well as the entire anal plate.
The relationships within Drepanosiphinae are much clearer, with the exception of Yamatocallis, which seems to be an independent lineage. Supporting information S1