Aphis pomi and Aphis spiraecola ( Hemiptera : Sternorrhyncha : Aphididae ) in Europe – new information on their distribution , molecular and morphological peculiarities

Aphid species Aphis pomi (de Geer, 1773) is oligophagous on pomoideous host plants, whilst Aphis spiraecola Patch, 1914 is a polyphagous species alternating between Spiraea spp., its primary host, and a wide variety of secondary hosts, also including pomoideous species. Despite the biological distinction, these species are difficult to separate using their morphological characters. Partial sequences of mitochondrial COI and nuclear EF-1α genes were analyzed for samples from Central and Eastern Europe, Germany, Bulgaria, Italy, Turkey, China together with available data from GenBank. Interspecific pairwise sample divergences of the COI fragment ranged from 3.1 to 4.3%. One COI haplotype of A. pomi was predominant (n = 24), with a pan European distribution. The most abundant COI haplotype of A. spiraecola (n = 16) occurred in Lithuania, Latvia, Poland, Italy, Turkey and China. Interspecific pairwise sample divergences of the EF-1α fragment ranged from 0.6 to 1.2%. Analyzed partial sequences of EF-1α were identical in A. pomi. The most abundant EF-1α haplotype of A. spiraecola (n = 14) occurred in Lithuania, Poland, Italy, Turkey and China. The length of ultimate rostral segment appeared to be the most reliable morphological character for discrimination between apple and spirea aphid species. It allowed a 100% correct identification of A. pomi (n = 143) and 91.5% of A. spiraecola (n = 94) specimens in the European samples used for the molecular analysis. The existence of A. spiraecola in the Eastern Baltic region of Europe is documented for the first time.

In Europe, apple aphid is a common species, whilst spirea aphid is currently reported mostly from southern Europe, reaching British Isles, Germany and Ukraine in the north (Holman, 2009;Nieto Nafria et al., 2010).Yet morphology-based identification make some records uncertain (Jaskiewicz & Kot, 2007;Caglayan et al., 2013;Yovkova et al., 2013).Spirea aphid is reported to be the principle pest on citrus, occasionally also on Prunoidea (stone fruits), but not apple or other pomoideae in Europe (Barbagallo et al., 1997).
Recently, the spirea aphid was reported from a more northernly part of Europe, in Poland, on Kalanchoe blossfeldiana, Polyscias fabiana, Schefflera arboricola in a greenhouse (Labanowski, 2008) and in Belarus, on Spi-was used.ML analysis was performed using Tamura 3-parameter model with invariable sites (T92 + I) for COI and Tamura 3-parameter model (T92) for EF-1α, which were selected by MEGA 5 model selection option (Tamura et al., 2011).Bootstrap values for NJ, MP and ML trees were generated from 1000 replicates.Bayesian analysis was conducted in MrBayes 3.2.1 (Ronquist & Huelsenbeck, 2003) using Hasegawa-Kishino-Yano model with Invariable sites (HKY + I) for COI and Felsenstein model with Invariable sites and Gamma distribution (F81 + I + G) for EF-1α, which were selected by jModeltest (Posada, 2008).One run for 1,000,000 generations with tree sampling every 1,000 generations was performed using the uniform model of the molecular clock.
Statistical parsimony networks with 95% implemented connection limit were constructed using TCS v 1.21 (Clement et al., 2000).For analysis of partial COI sequences gaps were treated as missing data, while EF-1α fragment gaps were treated as a 5th state.

Morphometrics
Samples representing different clades in the molecular tree and haplotype network were used to verify the characters commonly used in the morphology-based keys discriminating both species (Halbert & Voegtlin, 1992;Blackman & Eastop, 2000;Foottit et al., 2009).The following characters were selected: URS -ultimate rostral segment length; SIPHON -siphunculus length; CAUDA -length of cauda (apical part); MT2-4(5) -numbers of marginal tubercles on abdominal tergites II-IV(V); HCAU-DA -numbers of caudal hairs; SIPHON/CAUDA -ratio of siphuncular length to caudal (apical part) length.Measurements of slide-mounted apterous viviparous females were made using the interactive measurement system Micro-Image (Olympus Optical Co. GmbH).
After the construction of networks based on statistical parsimony (Fig. 1) 31 partial COI sequences of A. pomi and 30 sequences of A. spiraecola were collapsed into seven haplotypes each.The number of COI haplotypes, sequence length and sample or sequence numbers are given in Table 3; details for each sample (country, host plant and collection date) are given in Tables 1-2.
raea alba outdoors (Rakauskas & Buga, 2010).A. spiraecola-like aphids were collected from Spiraea sp.outdoors also in Latvia (2008) and Lithuania (2005, 2012-2014) Rakauskas, unpubl.).The aim of this study is to identify the available European samples of the A. pomi-spiraecola species complex using partial sequences of mitochondrial COI and nuclear EF-1α genes and test the reliability of the morphological characters used to discriminate between these two species (Blackman & Eastop, 2000;Foottit et al., 2009).

Samples
Aphid material collected in 2004-2013 included forty nine samples from ten European countries, Turkey and China (Table 1).Microscope slides in Canada balsam were prepared according to Blackman & Eastop (2000).Ethanol-preserved and mounted specimens are stored at the Department of Zoology, Vilnius University.

DNA extraction, fragment amplification and sequencing
For molecular analysis, a single aphid from one plant was considered as a unique sample.Total genomic DNA was extracted from each aphid using the DNeasy Blood & Tissue kit (Qiagen), which involved at least a 2 h digestion of tissue with proteinase K.For the amplification of mitochondrial COI and nuclear EF-1α gene fragments previously published primers, Aphis-L-465 / Aphis-H-1068 and Eloaphis-F / Eloaphis-R (Turči na vičienė et al., 2006), were used.PCR amplification was carried out in a thermal cycler (Eppendorf) in 50 µl volumes containing 2 µl genomic DNA, 5 µl of each primer (10 µM), 5 µl of PCRreaction buffer, 5 µl of dNTP mix (2mM each), 4-8 µl of 25mM MgCl 2 and 1.25 U of AmpliTaq Gold 360 polymerase (5U/µl) and ddH 2 O to 50 µl.The cycling parameters were as follows: denaturizing at 95°C for 10 min, denaturizing at 95°C for 30", annealing at 49°C (for COI) or 57°C (for EF-1α) for 30" and extension at 72°C for 30" (32-37 cycles in total), and a final extension for 5 min.
PCR products were purified and sequenced at the Institute of Biotechnology, Vilnius University (Vilnius, Lithuania).The amplification primers were also used as sequencing primers.DNA sequences for each specimen were confirmed with both sense and anti-sense strands and aligned in the BioEdit Sequence Alignment Editor (Hall, 1999).Partial COI sequences were tested for stop codons and none were found.The sequence data have been submitted to GenBank, Accession numbers are given in Table 1.

DNA sequence data analysis
In addition, available partial sequences of mitochondrial COI (1 of A. pomi and 11 of A. spiraecola) and nuclear EF-1α (7 of A. spiraecola) were downloaded from GenBank (Table 2).To avoid any discrepancies when analyzing data, sequences of both fragments were aligned and those matching partial sequences obtained from samples collected during this study were selected for further procedures.For sequences from GenBank geographic origin of samples and their host plants were obtained from publications (Table 2).
Phylogenetic analyses with a sequence of Nasonovia ribisnigri (Mosley, 1841) (tribe Macrosiphini, family Aphididae) as outgroup species, included Neighbour joining (NJ), Maximum parsimony (MP), Maximum likelihood (ML) and Bayesian inference in phylogeny (BI).NJ, MP and ML analyses were performed using MEGA 5 (Tamura et al., 2011).Out of 31 partial COI sequences of A. pomi, host plant information was available for 30 of them (Tables 1-2).The most abundant COI haplotype No. 1 (n = 24) occurred in samples collected from Malus, Pyrus, Cotoneaster, Crataegus, Spiraea, Sorbus, Cydonia and Amelanchier, and is clearly not host specific.Unique COI haplotypes were also not host specific as they were collected from the same hosts (Malus, Spiraea and Crataegus) as the most common  haplotype.This is also the case for the rare haplotype No. 5, which was recorded in samples from Aronia and Pyrus.
Of the 30 partial COI sequences of spirea aphid there is host plant information for 24.As in case of the apple aphid, the most common haplotype of spirea aphid appeared to be polyphagous, as it was collected from a broad spectrum of hosts, including those of the apple aphid: Spiraea (n = 4), Chaenomeles (n = 3), Prunus (n = 3), Malus (n = 3), Cotoneaster (n = 1), Pittosporum (n = 1) and Armeniaca (n = 1).Remaining haplotypes of spirea aphid shared the same hosts with the commonest haplotype, except three samples of haplotype No. 6, collected from Viburnum in Italy (Sicily).
The maximum parsimony (MP) analysis of partial COI sequences resulted in 618 equally parsimonious trees (length = 105, CI = 0.79, RI = 0.98).ML tree (T92 + I model) had a similar topology to the NJ (K2P distances) and BI (HKY + I model) analyses.NJ, MP and ML bootstrap values over 40% together with BI posterior probabilities over 0.50 are given at the respective nodes of the same tree in Fig. 2. The two Aphis species form distinct strongly supported clusters.The apple aphid clade is highly homogenous.Only four specimens from Lithuania, Latvia, Ukraine and China do not group with the remaining samples (Fig. 2).The clade of the spirea aphid appears more complex, comprising four moderately supported branches, one of them being represented only by Chinese samples (n = 5).Noticeably, GenBank sequence No. FJ965690 from China deposited as A. spiraecola, grouped outside the spirea aphid clade, both in the haplotype network (Fig. 1) and phylogenetic tree (Fig. 2).It appeared closer to the outgroup sequence of Nasonovia ribis-nigri, belonging to the tribe Macrosiphini of the aphid subfamily Aphidinae (Aphididae).This indicates an incorrect identification of the sequenced aphid specimen, because the genus Aphis belongs to the tribe Aphidini.
The analyzed region of EF-1α consisted of two parts of three exons and two introns, which were not removed before further analysis.The alignment of this fragment contained 510 sites, 6 of which were variable and parsimony informative.The average nucleotide composition was: T -30.3%, C -18.2%,A -31.4% and G -20.1%.The overall transition/transversion bias was R = 1.335.Interspecific pairwise sample divergences between spirea and apple aphid species ranged from 0.6 to 1.2% (average 0.9%).The range of the intraspecific pairwise sample divergences (K2P model) for the spirea aphid was 0-0.6% (average 0.2%), whilst all sequences of the apple aphid appeared identical.Noticeably, EF-1α sequences of the apple aphid differed from the closest haplotype of the spirea aphid in terms of only three base changes (Fig. 3), which is usually the characteristic of closely related haplotypes, attributable to the same species in the haplotype network.
All partial EF-1α sequences for the apple aphid (n = 30) from 9 genera of pomoideous hosts were identical, thus, no correlation between haplotypes and host plant could be detected.Out of 26 partial EF-1α sequences for the spirea aphid, host plant information was available for 20 and the commonest haplotype was associated with host plants of 6 genera: Prunus (n = 4), Malus (n = 3), Chaenomeles (n = 3), Cotoneaster (n = 2), Spiraea (n = 1) and Pittosporum (n = 1).Haplotype No. 2 was collected predominantly from Spiraea (n = 4), with just one sample from Prunus.Samples from Spiraea (n = 3) and Viburnum (n = 2) were of haplotype No. 3. Two remaining unique haplotypes were not unique in terms of their host plant associations, being collected from Spiraea (haplotype No. 4) and Malus (No. 5).As for COI haplotypes, our data do not indicate any host based background in the haplotype distribution of the spirea aphid partial EF-1α sequences.
The maximum parsimony (MP) analysis of partial EF-1α sequences resulted in 1010 equally parsimonious trees (length = 56, CI = 0.80, RI = 0.98).ML tree (T92 model) had a similar topology to the NJ (K2P distances) and BI (F81 + I + G) analyses.NJ, MP and ML bootstrap values over 40% together with BI posterior probabilities over 0.50 are given at the respective nodes of the same tree in Fig. 4. The apple and spirea aphids form distinct clusters.The apple aphid clade is homogenous because the sequences are all identical.The clade of the spirea aphid appears more complex and includes three moderately supported branches, none of them with geographic or host plant specificity.Noticeably, spirea aphids from Italy and USA collected from Viburnum and Spiraea respectively, were grouped together both by their COI and EF-1α partial sequences (COI haplotype No. 6 and EF-1α haplotype No. 3, respectively, Tables 2-4, Figs 1-4).Of the 14 specimens of spirea aphid with identical sequence of EF-1α (haplotype No. 1, Table 4) 12 also had identical COI sequences (haplotype No. 1, Table 3).Such congruence might indicate evolutionary specificity of certain lineages of spirea aphid.
Halbert & Voegtlin (1992), followed by Blackman & Eastop (2000), suggest three morphological characters can be used to discriminate between apterous viviparous females of apple and spirea aphids: numbers of marginal tubercles on abdominal tergites II-IV (present in apple aphid, absent in spirea aphid); numbers of caudal hairs (10-19 hairs Fig. 3. Haplotype network for EF-1α fragment (508 positions in final set, 95% connection limit, gaps treated as 5 th state) haplotypes of A. pomi and A. spiraecola.The haplotype with the highest outgroup probability is displayed as a square, while others are displayed as ovals.For sample information, see Tables 1-3.BG -Bulgaria, BY -Belarus, CN -China, CZ -Czech Republic, DE -Germany, EE -Estonia, IT -Italy, KR -Korea, LV -Latvia, LT -Lithuania, PL -Poland, TR -Turkey, UA -Ukraine, US -United States of America.
table 4. EF-1α haplotypes of Aphis pomi and Aphis spiraecola revealed by the haplotype network analysis.Sample numbers are the same as in Tables 1-2.

dIScuSSIon
Molecular markers are widely used to reveal cryptic insect species, including aphids (Rakauskas et al., 2011).Partial COI sequences used for DNA barcoding were analyzed for A. pomi (n = 76) and A. spiraecola (n = 56) by Foottit et al. (2009).The values of interspecific pairwise sample divergences were higher (mean 5.0%, range 4.8-5.1%)than those obtained for the COI fragments used in this study (mean 3.6%, range 3.1-4.3%).The majority of the spirea aphids (50 out of 56) from North America, Australia, Guam, Palau and Marshall Islands, also have identical COI barcode sequences (Foottit et al., 2009).The remaining individuals from New Zealand (n = 1), New York (n = 2) and British Columbia (n = 3) differed from the most abundant COI haplotype by one to three base changes, giving a maximum pairwise within-species divergence of 0.6%.In our study, maximum pairwise within-species divergence of the spirea aphid was 0.5% (average 0.2%, range 0.0-0.5%).This is in accordance with the conclusion of Foottit et al. (2009) that the variation in biological characteristics among populations of spirea aphid was greater than in those of apple aphid.
Analysis of COI barcode fragments indicate that sequences for the apple aphid collected in North America are identical (Foottit et al., 2009).In our study we recorded a greater diversity in COI fragments from European specimens of the apple aphid.Six haplotypes were detected in 30 samples.This might be because of the presumed Palaearctic origin of the apple aphid, which is reported to be a non-native species in the Nearctic, where it was first noted in North America in 1844 (Foottit et al., 2006).This fact could account for the homogeneity of apple aphid COI sequences from North American populations (Foottit et al., 2009).Unlike the partial COI sequences, our study indicates that partial sequences of the nuclear EF-1α from European samples of the apple aphid are very homogeneous (1 haplotype, Table 4).Such homogeneity might be attributed to the isolated mode of reproduction of this species: gynoparae and males are apterous resulting in a high incidence of intraclonal inbreeding.Genetic consequences of such a reproductive system appear similar to those of anholocyclic populations resulting in a few predominant table 5. Summary statistics for the key morphological characters of apterae of Aphis pomi and A. spiraecola.For comparison the same data from Foottit et al. (2009) are given.URS -ultimate rostral segment length (all lengths in µm), SIPHON -siphunculus length, CAUDA -length of cauda (apical part), MT2-4(5) -numbers of marginal tubercles on abdominal tergites II-IV(V), HCAUDA -numbers of caudal hairs, Siph/cauda -ratio of siphuncular length to caudal (apical part) length.

Characters
Aphis clones that appear to be considerably different from one another (Kanbe & Akimoto, 2009).
In general, species level identification of apple and spirea aphids by means of COI and EF-1α partial sequences coincided with that based on commonly used morphological characters (Halbert & Voegtlin, 1992;Blackman & Eastop, 2000;Foottit et al., 2009).However, none of the above mentioned morphological characters (Table 5) on their own can ensure a 100% correct discrimination between apple and spirea aphids.All four characters should be used to determine the identity of series of individual aphids in each sample.The situation might be even more complicated due to the presence of mixed colonies of both species.For example, apterous viviparous females (n = 5) collected from Cotoneaster in Lublin, Poland (sample 08-115) had clear morphological characters typical of the apple aphid: ultimate rostral segment length 148-156 µm; numbers of marginal tubercles on abdominal tergites 4-6; ratio of siphuncular length to caudal length 2.40-2.80.For DNA extraction one winged individual aphid was used and its sequence grouped together with those of the spirea aphid.When the morphology of the voucher specimen was checked, it resembled the spirea aphid.This case demonstrates that mixed colonies might also complicate identification by means of DNA sequences, because several individuals per colony should be subject to DNA analysis.

Fig. 1 .
Fig.1.Haplotype networks for COI fragment (621 positions in final set, 95% connection limit, gaps treated as missing data) haplotypes of A. pomi and A. spiraecola.The haplotype with the highest outgroup probability is displayed as a square, while others are displayed as ovals.For sample information, see Tables1-3.BG -Bulgaria, BY -Belarus, CN -China, CZ -Czech Republic, DE -Germany, EE -Estonia, IT -Italy, LV -Latvia, LT -Lithuania, PL -Poland, TR -Turkey, UA -Ukraine, US -United States of America.

Fig. 2 .
Fig. 2. Maximum Likelihood (ML) tree showing phylogenetic relationships among A. pomi and A. spiraecola based on partial sequences of mitochondrial COI (621 positions in final set).Numbers above branches indicate support of NJ (left, > 40%) and MP (right, > 40%) based on bootstrap test with 1000 replicates, and numbers below branches indicate support of ML (right, > 40%) bootstrap test with 1000 replicates and posterior probabilities of BI analysis (left, > 0.50).Sample numbers are the same as in Tables 1-2, together with the abbreviated symbol of the relevant country BG -Bulgaria, BY -Belarus, CN -China, CZ -Czech Republic, DE -Germany, EE -Estonia, IT -Italy, LV -Latvia, LT -Lithuania, PL -Poland, TR -Turkey, UA -Ukraine, US -United States of America.

Fig. 4 .
Fig. 4. Maximum Likelihood (ML) tree showing phylogenetic relationships among A. pomi and A. spiraecola based on partial sequences of nuclear EF-1α (510 positions in final set).Numbers above branches indicate support of NJ (left, > 40%) and MP (right, > 40%) based on bootstrap test with 1000 replicates, and numbers below branches indicate support of ML (right, > 40%) bootstrap test with 1000 replicates and posterior probabilities of BI analysis (left, > 0.50).Sample numbers are the same as in Tables 1-2, together with the abbreviated symbol of the relevant country BG -Bulgaria, BY -Belarus, CN -China, CZ -Czech Republic, DE -Germany, EE -Estonia, IT -Italy, KR -Korea, LV -Latvia, LT -Lithuania, PL -Poland, TR -Turkey, UA -Ukraine, US -United States of America.

table 2 .
Partial sequences of COI and EF-1α from GenBank that were used in the present study for comparison.