Taxonomic position of Hormaphis similibetulae Qiao & Zhang, 2004 (Hemiptera, Aphididae): molecular and biological evidences

Abstract The taxonomic position of Hormaphis similibetulae Qiao & Zhang, 2004 has been reexamined. The phylogenetic position of Hormaphis similibetulae was inferred by maximum parsimony, maximum likelihood and Bayesian analyses on the basis of partial nuclear elongation factor-1α and mitochondrial tRNA leucine/cytochrome oxidase II sequences. The results showed that this species fell into the clade of Hamamelistes species, occupying a basal position, and was clearly distinct from other Hormaphis species. A closer relationship between Hormaphis similibetulae and Hamamelistes species was also revealed by life cycle analysis. Therefore, we conclude that Hormaphis similibetulae should be transferred to the genus Hamamelistes as Hamamelistes similibetulae (Qiao & Zhang), comb. n.


Introduction
The aphid tribe Hormaphidini in subfamily Hormaphidinae (Hemiptera: Aphididae) consists of three genera, Hamamelistes, Hormaphis and Protohormaphis (Remaudière and Remaudière 1997). Hamamelistes and Hormaphis are disjunctively distributed in Eurasia and North America (Guo and Qiao 2005), where they are primarily associated with Hamamelis and secondarily associated with Betula. The taxonomy of these two genera was once in a mess at both the genus and species levels. They are easily confused with one another, and species of the same genus are difficult to distinguish morphologically. This confusion in the taxonomy was due partly to a limitation of diagnostic characteristics and partly to the fact that no combination had been established between different aphid forms on primary and secondary hosts. Distinction of Hamamelistes and Hormaphis is based mainly upon alatae, galls and life cycles. However, it is not easy to collect all morphs, and the observation of life cycles takes a long time. Molecular studies have shed light on these issues. Based on the mitochondrial cytochrome oxidase II (COII) gene, Aoki et al. (2001) clarified the Japanese Hamamelistes species, established the combination between generations on primary and secondary hosts, and elucidated their life cycles. von Dohlen et al. (2002) estimated the phylogeny of Hormaphidini using partial nuclear elongation factor-1α (EF-1α) and mitochondrial tRNA leucine/cytochrome oxidase II (COII) sequences, the monophyly of both Hamamelistes and Hormaphis was retrieved with strong support. Qiao and Zhang (2004) described Hormaphis similibetulae based on specimens of apterous viviparous females collected from small conical galls on leaves of Betula albosinensis in China (Tibet); the specimens were closely related to Hormaphis betulae (Mordvilko) but differed from the latter in body color and living habits. In this study, the taxonomic position of H. similibetulae was reassessed on the basis of nuclear EF-1α and mitochondrial tRNA/COII sequences. A discussion of life cycles was also included.

Materials and methods
The samples used in this study and the corresponding collection information are listed in Table 1. Eight species of Hormaphidini, covering all the species of Hamamelistes and Hormaphis were used as ingroups. Three species of Nipponaphidini were chosen as outgroups because Nipponaphidini is considered the sister group of Hormaphidini based on biological and phylogenetic data (Ghosh 1985, von Dohlen and Moran 2000, Ortiz-Rivas and Martínez-Torres 2010. Voucher specimens were preserved in 75% ethanol and deposited in the National Zoological Museum of China, Institute of Zoology, Chinese Academy of Sciences, Beijing, China. Total genomic DNA was extracted from single aphids preserved in 95% or 100% ethanol using a CTAB protocol modified from Doyle and Doyle (1987). Partial leucine tRNA and the cytochrome oxidase II (COII) gene was amplified with primers 2993+ (Stern 1994) and A3772 (Normark 1996). Sequencing reactions were performed using the corresponding PCR primers from both directions with BigDye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems, Foster City, CA, USA) and run on an ABI 3730 automated sequencer (Applied Biosystems). Sequences were assembled by Seqman II (DNAStar, Inc., Madison, WI, USA) and verified for protein coding frame- shifts to avoid pseudogenes (Zhang and Hewitt 1996) using Editseq (DNAStar, Inc.). All sequences were deposited in GenBank under the accession numbers JF730745-JF730749. All EF-1α sequences used in this study were downloaded directly from GenBank (for accession numbers see Table 1), and only exons were used for further analysis. Multiple alignments were done with ClustalX 1.83 (Thompson et al. 1997) and then verified manually. Nucleotide composition and pairwise distances based on Kimura's two-parameter model (K2P) (Kimura 1980) of the aligned sequences were calculated using MEGA 4.0 (Tamura et al. 2007).
Phylogenetic reconstructions were conducted by maximum parsimony (MP), maximum likelihood (ML) and Bayesian analyses for each single gene and a combined dataset. The partition homogeneity test (Farris et al. 1994) based on 100 replicates of a heuristic search algorithm was performed with PAUP*4.0b10 (Swofford 2002) to examine the incongruence between EF-1α and mtDNA. Unweighted MP and ML analyses were carried out using PAUP*. For ML analysis, the best-fit model of nucleotide substitution was selected for each dataset according to the Akaike information criterion (AIC) in Modeltest 3.7 (Posada and Crandall 1998). Heuristic searches were performed with 1000 (MP) or 100 (ML) random-addition sequences and treebisection-reconnection (TBR) branch swapping. Bootstrap (BS) analyses were used to assess the relative robustness of branches of the MP (1000 replicates) and the ML (100 replicates) trees (Felsenstein 1985). Bayesian analysis was conducted using MrBayes 3.1.2 (Ronquist and Huelsenbeck 2003) based on the model selected by Modeltest 3.7. In the combined analysis, the mitochondrial and nuclear data were partitioned, and a heterogeneous model was used for each gene partition. The parameters of the model were treated as unknown variables with uniform prior probabilities and were estimated during the analysis. Four Markov chains (three heated and one cold) were run, starting from a random tree and proceeding for one million Markov chain Monte Carlo generations, sampling the chains every 100 generations. Two concurrent runs were conducted to verify the results. The first 2500 trees were discarded as burn-in samples, the remaining trees were used to compute a majority-rule consensus tree with posterior probabilities (PP).

Results and discussion
The final alignments of EF-1α (excluding three introns) and tRNA/COII sequences consisted of 826 and 761 sites, with 131 and 165 parsimony-informative sites, respectively. A single 1-to 2-base-long indel was found in the tRNA. The genetic distance between two distinct samples of H. similibetulae was 0 for EF-1α and 0.001 for tRNA/COII. The distances of both genes between H. similibetulae and Hamamelistes species were much smaller than those between H. similibetulae and the other Hormaphis species (EF-1α: average of 0.040 and range of 0.038-0.042 to Hamamelistes, average of 0.082 and range of 0.078-0.092 to Hormaphis; tRNA/COII: average of 0.080 and range of 0.071-0.085 to Hamamelistes, average of 0.106 and range of 0.102-0.112 to Hormaphis).
For phylogenetic analyses, the partition homogeneity test found no significant conflict between EF-1α and mtDNA (P=0.05), indicating that information from both genes could be combined. Combined analysis resulted in similar topology to that obtained in single gene analyses and with higher support for most nodes, so only the combined dataset results were presented. MP analysis yielded eight most parsimonious trees with a length of 611 steps (CI=0.705401, RI=0.845626). ML analysis produced one ML tree based on the optimal model GTR+G selected by AIC in Modeltest 3.7. The 50% majority-rule consensus tree inferred from Bayesian analysis is shown in Fig. 1 and resulted in a topology essentially identical to that obtained in ML analysis, but was different from the strict consensus of MP trees in the position of H. similibetulae. All ingroup taxa constituted a monophyletic group with respect to these outgroups and formed two clades. Clade I (100% MP BS, 100% ML BS, 1.00 PP) was comprised of H. betulae, H. cornu, and H. hamamelidis. Clade II (99% MP BS, 99% ML BS, 1.00 PP) consisted of all the Hamamelistes species and H. similibetulae. Within clade II, two distinct samples of H. similibetulae clustered together (100% MP BS, 100% ML BS, 1.00 PP) and were placed as the outermost branch in ML and Bayesian analyses, just as the results based on EF-1α. However, MP analysis revealed the same topology as the mitochondrial analysis: H. similibetulae and Hamamelistes spinosus were sister groups, although the support value was low (53% BS), and together formed the basal lineage within clade II.
The results of genetic distances and phylogenetic analyses strongly suggested that H. similibetulae was more closely related to Hamamelistes than to Hormaphis. H. similibetulae was distinguished by its unique biology, forming galls on leaves of Betula. Because of the high morphological similarity with H. betulae (Mordvilko), it was placed under the genus Hormaphis (Qiao and Zhang 2004). However, the distinction of apterae of Hamamelistes and Hormaphis from the secondary host Betula is very difficult: both of them are aleyrodiform, dorsoventrally compressed, have body segments fused, short antennae with only 2-4 segments, fore and middle legs without tarsi, and hind legs with rudimentary unsegmented tarsi and lack claws. These reductions appear to be related to the organisms' sedentary habits on Betula and represent the adaptive convergences selected by their temperate habitat. Although species of both genera migrate between Hamamelis and Betula, their life cycles are quite different and have proven extremely valuable in distinction (Pergande 1901, von Dohlen and Gill 1989, Aoki and Kurosu 1991, von Dohlen and Stoetzel 1991, Aoki et al. 2001. Firstly, Hamamelistes have two-year life cycles due to a long gall phase, while Hormaphis complete their life cycles within one year. Secondly, on Hamamelis, Hamamelistes induce spiny or corallike galls on leaf or flower buds, whereas Hormaphis cause conical galls on the leaves. Lastly, Hamamelistes inhabit cockscomb-like or blister-like galls on leaves of Betula, but Hormaphis live freely on the leaves, not causing any deformation. In China, there is only one species of Hamamelis, H. mollis, distributed in Sichuan, Hubei, Anhui, Zhejiang, Jiangxi, Hunan and Guangxi Provinces (Zhang and Lu 1995). According to the absence of primary host at high elevations in the Tibetan Plateau, Qiao and Zhang (2004) inferred that H. similibetulae was autoecious on Betula albosinensis. We agree with their inference, as Ha. betulinus and H. betulae were also observed living all year round parthenogenetically on Betula in Europe due to lack of primary host (Heie 1980). Although the life cycle of H. similibetulae requires further research, it appears to be more similar to that of Hamamelistes than to that of Hormaphis.

Conclusion
The phylogenetic position of Hormaphis similibetulae was inferred by MP, ML and Bayesian analyses on the basis of nuclear EF-1α and mitochondrial tRNA/COII sequences. In all phylogenetic analyses, H. similibetulae clustered firmly with Hamamelistes and was placed as a basal lineage, clearly differed from other Hormaphis species. Life cycle similarities also indicated that H. similibetulae was more closely related to Hamamelistes species. We therefore conclude that H. similibetulae should be transferred to the genus Hamamelistes as Hamamelistes similibetulae (Qiao & Zhang), comb. n.