We determined the nucleotide sequences of a part of the mitochondrial cytochrome c oxidase I gene (1,000 bp) for twelve species of Asian phytophagous ladybird beetles belonging to the genus Epilachna, and constructed molecular phylogenetic trees for ten “Henosepilachna” species, using two “Epilachna” species as outgroups. Based on the suggested phylogenetic trees, we discussed taxonomic issues and the direction of host shift in these epilachnines.
Ladybird beetles of the subfamily Epilachninae are phytophagous, and most species are distributed in tropical and subtropical zones of the world (Gordon, 1975). Most species are host specific, and their host plants cover diverse taxonomic groups of angiosperms (Schaefer, 1983; Pang and Mao, 1979; Katakura et al., 1992). Some species are notorious for causing serious damage to important crops such as legumes, solanums and cucurbits (Dieke, 1947; Schaefer, 1983).
The phylogenetic position of Epilachninae in the family Coccinellidae seems well established. The subfamily is considered to represent a lineage of advanced coccinellids that has a close relationship with the higher predaceous coccinellid group, the subfamily Coccinellinae (Sasaji, 1968). However, the supraspecific classification of Epilachninae, especially of Old World species, is still controversial (Iablokoff-Khnzorian, 1980; Richards, 1983; Fürsch, 1991). There are two morphologically distinct groups in Asian and African species: one is characterized by toothless tarsal claws and the lack of a spilt in the sixth visible abdominal sternite of females; whereas the other is characterized by toothed tarsal claws and a divided sixth sternite. The two groups are further divided into species groups (Dieke, 1947). Some authors treat the two groups as two distinct genera, Epilachna Dejean and Henosepilachna Li (Li and Cook, 1961; Fürsch, 1990, 1991), respectively. On the other hand, others assign them to the single genus Epilachna (Iablokoff-Khnzorian, 1980; Richards, 1983), considering that the separation by these characters was unreliable. Although the present study is mainly concerned with Henosepilachna, all the species treated here are tentatively placed in Epilachna to avoid nomenclatural problems by yielding new combinations. When necessary, however, the species belonging to Henosepilachna and Epilachna (sensu Li and Cook, 1961) are indicated as “Henosepilachna” and “Epilachna”, respectively (see Katakura et al., 1994).
In order to establish a satisfactory classification system, it is necessary to clarify phylogenetic relationships between various species and species groups (Katakura et al., 1994). Reconstruction of phylogenetic relationships among extant taxa of epilachnines is also indispensable to understand the evolutionary changes in relationships between ladybird beetles and their host plants.
Katakura et al. (1994) analyzed the phylogenetic relationships of several species groups of Asian epilachnines based on the female internal reproductive organs and modes of sperm transfer. They suggested that “Henosepilachna” and “Epilachna” are sister groups and that “Epilachna” is further divided into two sister groups. Unfortunately, however, their analysis provided little information on the phylogenetic relationships between the members of “Henosepilachna”.
In the present paper, we analyzed the phylogenetic relationships of ten species of Asian epilachnines belonging to “Henosepilachna” using mitochondorial DNA sequences, assuming two species of “Epilachna” (sensu Li and Cook, 1961) as outgroups. Since the rate of nucleotide substitution in animal mitochondrial DNA is generally higher than that in nuclear DNA (Brown et al., 1979), it is suitable for the analysis of phylogenetic relationships among closely related taxa such as those used in this study. Based on the suggested phylogeny, we discuss taxonomic issues and infer the host plants of the ancestral species of these epilachnines.
MATERIALS AND METHODS
We used ten species of “Henosepilachna” and two species of “Epilachna”. The species, species groups, provenance and host plants of these specimens are given in Table 1. The name E. pusillanima is used for the species referred to as E. dodecastigma in Katakura et al. (1988, 1994), because the identity of E. dodecastigma (sensu Wiedemann, 1823) is still ambiguous (Booth and Pope, 1989). Two Indonesian species, whose taxonomic status are not yet determined, are referred to using the species-specific code number or code letter (E. sp. 3 and E. sp. H) (cf. Katakura et al., 1994). All the ladybird beetles used for the phylogenetic study were collected from 1994– 95. Of the ten “Henosepilachna” species, nine belong to the vigintioctopunctata group and one to the enneasticta group according to Dieke (1947). Four species, i.e., E. vigintioctomaculata, E. niponica, E. pustulosa and E. yasutomii are very closely related and are assigned to the so-called Epilachna vigintioctomaculata species complex, which is further classified into two groups (group A: E. vigintioctomaculata; group B: other three species) based on their morphological characters (Katakura, 1981). Both of the “Epilachna” species belong to the admirabilis group (Dieke, 1947).
Epilachna beetles used in the present study, and their sampling locations and host plants
For E. vigintioctomaculata, E. niponica, E. pustulosa, E. yasutomii, E. vigintioctopunctata and E. admirabilis, mitochondrial DNA (mtDNA) was extracted from a living adult specimen following the method of Tamura and Aotsuka (1988). For other species, total DNA was extracted from a specimen stored in alcohol by using Steller's (1990) method. The mtDNA samples from E. vigintioctomaculata were digested by appropriate restriction enzymes to fractionate cytochrome c oxdase subunit I (COI) gene and were then cloned into plasmid vector pUC118 with E. coli K12.MV1184 as a host. The mtDNA recombinants were subcloned by using exonuclease III and mungbean nuclease (Henikoff, 1984) and/or appropriate restriction enzymes. Single strand templates for sequencing were obtained from these subclones by taking advantage of the pUC118/119-M13KO7 system (Vieria and Messing, 1987). The nucleotide sequences were determined by using an ABI autosequencer according to the protcol supplied by the manufacturer.
Using the E. vigintioctomaculata sequence obtained as a reference, we designed a set of PCR primers to amplify a region containing the whole COI gene. The primer sequences were: 5′-TTTACCGCCTAATTCAGCCA-3′ and 5′-AGAATTCATGGGGTTTAAATCCAGTGC-3′, in the latter of which the sequence of seven bases from the 5′terminal was an adapter to facilitate subsequent cloning procedures. Depending on the species mentioned above, either mtDNA or a total DNA sample was used for PCR templates. The reaction mixture (100 μl) for PCR contained: Tris-HCl pH 8.3, 50 mM KCl, 2 mM MgCl2, 0.001% gelatin, 200 μM each dNTP, 200 nM each primer, approximately 50 ng template DNA, 2.5 units Taq polymerase. Since an addition of Pfu polymerase improves the fidelity of PCR (Barnes, 1994), we used a mixture of 2.5 units Taq and 0.0125 units Pfu polymerases. Amplifications were performed for 25 cycles in a DNA thermalcycler using the following parameters: 94°C for 30 sec; 60°C for 1 min; 72°C for 1 min, except for the last cycle where 72°C for 8 min. The DNA fragments amplified were then cloned and sequenced by essentially the same method as that for the sequence of E. vigintioctomaculata.
Since we determined nucleotide sequences from only a single cloned DNA, they are subject to PCR error. Nevertheless, as shown later, PCR error is negligible in the phylogenetic analyses, because the differences in sequence among the species used are much greater than those expected by PCR error.
Phylogenetic analysis for sequence data
We determined the nucleotide sequences of a part of mitochondrial cytochrome c oxidase subunit I gene (COI) for 12 individuals (Table 1). The number of nucleotide sites determined and used for phylogenetic analyses was 1,000 bp from the initial codon on the sense strand. Jukes and Cantor's (1969) method was used for estimating the number of substitutions per site for all possible pair of sequences. Using the distance matrix, we constructed a phylogenetic tree by the minimum-evolution (ME) method (Rzhetsky and Nei, 1992). To estimate the confidence probability for each interior branch, the bootstrap method (Felsenstein, 1985) was performed with 1,000 replications.
The nucleotide sequence of COI region (1,000 bp) was determined for each “Henosepilachna” and “Epilachna” individual. The sequences have been deposited in databases (DDBJ, EMBL and GenBank) under the following accesion numbers: Epilachna enneasticta, AB002173; E. sp. 3, AB002174; E. boisduvali, AB002175; E. septima, AB002176; E. pusillanima, AB002177; E. admirabilis, AB002178; E. sp. H, AB002179; E. vigintioctopunctata, AB002180; E. pustulosa, AB002183; E. yasutomii, AB002184; E. niponica, AB002185.
A total of 343 nucleotide substitution sites and 38 amino acid replacement sites were detected in the 12 sequences. Variations such as insertion and deletion were not found. Within “Henosepilachna,” the number of nucleotide substitutions was fewest between the members of the group B of the E. vigintioctomaculata complex (one site: E. yasutomii vs. E. niponica), and most numerous between E. boisduvali and E. septima (180 sites).
Figure 1a is a phylogenetic tree for the ten “Henosepilachna” sequences constructed using the ME method. Two “Epilachna” sequences were used as outgroups. The bootstrap probabilities were given in the upper or lower side of each branch. To reconstruct the condensed tree (Nei, 1996) showing only the reliable topology of the specimen, branches whose bootstrap values were lower than 95% in Fig. 1a were multifurcated (Fig. 1b). These trees convey the following information:
“Henosepilachna” diverged into four groups: (a) E. pusillanima and E. sp. 3, (b) E. septima, (c) E. boisduvali and (d) other six species (the E. vigintioctomaculata complex, E. vigintioctopunctata and E. enneasticta) (Fig. 1b).
In group (d), the E. vigintioctomaculata complex (which includes four species), E. vigintioctopunctata and E. enneasticta were trifurcated and their phylogenetic relationships were not resolved.
The E. vigintioctomaculata complex diverged into two subclusters corresponding to two morphologically defined groups. One subcluster consists of the E. vigintioctomaculata sequence, which is classified in group A, and the other subcluster consists of the members of group B, namely E. pustulosa, E. niponica and E. yasutomii. In the latter subcluster, only one or two nucleotide substitutions were detected by pairwise comparisons between the three sequences.
Because the number of species analyzed in the study was very small and limited to paticular groups, we could not refer to the phylogenetic relationship between “Henosepilachna” and “Epilachna”. However, the present study clarified some noteworthy aspects of the “Henosepilachna”.
One was the phylogenetic position of E. enneasticta. Among “Henosepilachna”, E. enneasticta and related species (the E. enneasticta group, sensu Dieke, 1947) are distinct in genitalia morphology in both sexes and the associated abdominal segments of females (Dieke, 1947). The last visible sternite of the female is of particular taxonomic interest. In E. enneasticta, the last visible sternite of the female is split but fused on the suture. This characteristic might be an intermediate condition between “Epilachna”, which has a non-split last sternite, and typical “Henosepilachna”, which has split sternite (Kapur, 1967). The occurrence of such an equivocal condition is one of the reasons why some authors reject “Henosepilachna” as a valid genus (Richards, 1983). Furthermore, E. enneasticta showed a somewhat different condition in the morphology of the female internal reproductive organ (Katakura et al., 1994).
The present study showed that Epilachna enneasticta is not an intermediate form that links members of “Epilachna” with those of “Henosepilachna”. It is rather an advanced form of “Henosepilachna” derived from a solanum-feeding lineage (Fig. 1b). Morphological features characteristic to E. enneasticta (modified genitalia in both sexes and the abdominal segments in females) can thus be regarded as autoapomorphies.
Another point clarified by the present study is the phylogenetic relationships among the members of the E. vigintioctomaculata complex. The E. vigintioctomaculata complex is composed of a number of closely related but morphologically and/or biologically different forms (Katakura, 1981). It has been paid much attention from students of evolutionary biology, in particular those who have specific interest in speciation (Katakura, 1997). Based on morphological evidence and information on the host plants, the E. vigintioctomaculata complex has been classified into two groups, A and B. In group B, three species have been recognized by differences in host plants and the geographic allopatric-sympatric relationships (Katakura, 1981). The present study supports the dichotomy of the species complex (Fig. 1b). As the genetic divergence between the three species of the group B was very small, our data could not provide positive evidence for a trichotomy. This means that the differences in morphology and/or the host plants (Katakura, 1981) must have developed during a short geological time. Detailed analyses of the phylogenetic relationships between various members of the E. vigintioctomaculata complex will be reported elsewhere (Kobayashi et al., in preparation).
Finally, we discuss the direction of host shifts in the studied groups of epilachnines based on the suggested phylogeny (Fig. 1a, b). Such an attempt is indispensable to understand the evolutionary relationships between ladybird beetles and their host plants. Indeed, reconstruction of phylogenetic relationships among species of insects has often succeeded in detecting the direction of their adaptation to particular host species in the course of phyletic evolution (Futuyma and McCafferty, 1990; Futuyma et al., 1995).
It has been known that the majority of “Henosepilachna” species feed on either cucubitaceous or solanaceous plants (Shaefer, 1983; Katakura et al., 1992). The most parsimonious interpretation of Fig. 1a and b with respect to the direction of host shifts is that the host plant of the ancestral species of “Henosepilachna” was cucurbits. Later, two types of host shifts occurred in the two lineages. One was a shift toward the species of Compositae in the E. sp. 3 lineage, and the other was a shift toward solanaceous plants in the common ancestor of the cluster comprizing E. enneasticta, E. vigintioctopunctata and the E. vigintioctomaculata complex. In the latter group, further shifts to other plants followed. The ability of E. vigintictomaculata in the E. vigintioctomaculata complex to feed on cucurbits (Katakura, 1981) can be interpreted in two different ways. It may be a plesiomorphic condition inherited from a cucurbits-feeding ancestor. Alternatively, it may have been acquired in a recurrent evolution after once shifting to solanaceous plants occurred.
The validity of the scenario of the host shift presented above, however, must be examined by further extensive phylogenetic analyses of various “Henosepilachna” species.
We would like to express our gratitude to Prof. S. Nakano, Hiroshima Shudo University, Prof. K. Nakamura, Kanazawa University, Mr. S. Kahono, Bogor Zoological Museum, and Mr. Y. Shirai, National Institute of Agro-Environmental Siences, for collecting beetles. We are also grateful to Messrs. T. Moriya, T. Kato and G. Toba, Tokyo Metropolitan University, for valuable advice, and two anonymous referees for useful comments.
This study was partly supported by a Grant-in-Aid for Scientific Reserach from the Ministry of Education, Science, Sports and Culture of Japan (No. 09440258) to H.K.