Evolutionary Relationship between Two Firefly Species, Curtos costipennis and C. okinawanus (Coleoptera, Lampyridae), in the Ryukyu Islands of Japan Revealed by the Mitochondrial and Nuclear DNA Sequences

The phylogenetic relationship, biogeography, and evolutionary history of closely related two firefly species, Curtos costipennis and C. okinawanus, distributed in the Ryukyu Islands of Japan were examined based on nucleotide sequences of mitochondrial (2.2 kb long) and nuclear (1.1-1.2 kb long) DNAs. In these analyses, individuals were divided among three genetically distinct local groups, C. costipennis in the Amami region, C. okinawanus in the Okinawa region, and C. costipennis in the Sakishima region. Their mtDNA sequences suggested that ancestral C. costipennis population was first separated between the Central and Southern Ryukyu areas, and the northern half was then subdivided between C. costipennis in the Amami and C. okinawanus in the Okinawa. The application of the molecular evolutionary clocks of coleopteran insects indicated that their vicariance occurred 1.0–1.4 million years ago, suggesting the influence of submergence and subdivision of a paleopeninsula extending between the Ryukyu Islands and continental China through Taiwan in the early Pleistocene.


Introduction
The Ryukyu Islands form a chain of more than 200 islands extending for about 1,200 km between the Japanese mainland and Taiwan. The faunae of this area are divided among three portions, the Northern, Central, and Southern Ryukyu areas, and this pattern is considered to have been strongly influenced by changes of the land configuration of this area during the Neogene and Quaternary Periods that occurred due to the interaction between tectonic changes of the Ryukyu ridge and changes of the sea level [1][2][3]. Paleolands or land bridges connecting this area with Taiwan and continental China, super islands connecting neighboring islands, and geologically long-standing channels that emerged during these periods are considered to be major factors, as they acted as corridors and barriers to biological dispersal [3,4].
The genus Curtos Motschulsky is a group of fireflies including 16 species mainly distributed in southeastern Asia [5,6]. In Japan, only two species, Curtos costipennis (Gorham) and C. okinawanus Matsumura, are known to exist in the Ryukyu Islands. The former species, also extant on the Chinese continent and in Taiwan, is distributed widely in the Northern, Central, and Southern Ryukyu Islands excluding the Okinawa region in the southern half of the Central Ryukyu Islands, while the range of the latter species is restricted to the Okinawa region, interrupting the range of the former species. Based on morphological and behavioral observations of the two species, Ohba and Goto [7] suggested that C. okinawanus might have derived from C. costipennis. However, it is not clear where and how these species have evolved or how C. okinawanus came to occupy the Okinawa region in the middle of the range of C. costipennis. Otherwise, C. costipennis might have entered the Ryukyu Islands from  Taiwan or the Chinese continent, before and after the occurrence of C. okinawanus in the Okinawa region, while it is not clear whether or not many paleolands emerged during the evolutionary history of the two species. To answer these questions, we performed molecular phylogenetic analyses of the two species.
To do this, we collected specimens of the two species from 51 localities in the Ryukyu Islands, sequenced three sections of the mitochondrial and nuclear DNA of individual insects, and then constructed phylogenetic trees based on the sequences. Our objective was to examine (1) the phylogenetic relationship among local populations of the two species, (2) the geographic distribution patterns of genetically separated groups, and (3) the evolutionary history of the two firefly species. The results strongly suggested that C. costipennis is paraphyletic and that C. okinawanus have evolved from populations of C. costipennis that have been isolated in the Central Ryukyu Islands.

Materials and Methods
Adults of C. costipennis and C. okinawanus were collected from 51 localities (Table 1) in the Ryukyu Islands ( Figure 1). Although several isles in the Tokara Islands of the Northern Ryukyu Islands are also known to be habitats of the former species [5], we could not obtain samples from these islets. For an outgroup taxon in phylogenetic analyses, the firefly Luciola kuroiwae Matsumura collected in Okinawa-jima Island was used. Specimens were stored in 99.9% ethanol. Template DNA was extracted from the insect body excluding the head and pronotum using a Wizard Genomic DNA Purification Kit (Promega Co., Madison, WI, USA) and dissolved in 100 µL sterilized distilled water.
For template DNAs extracted from individual insects, three DNA fragments, the mtDNA fragment containing the tRNA leu gene and portions of the cytochrome oxidase subunit I and II genes (CO), the fragment containing the 16S ribosomal RNA gene (rDNA), and the fragment containing internal transcribed spacer 2 (ITS2) of the nuclear rDNA, were amplified using the primer sets FFMT2210F/ FFMT3578R, AAMT12948F/FFMT13911R, and ITS2F/ ITS2R, respectively ( Table 2). The primers FFMT2210F, FFMT3578R, and FFMT13911R were designed based on sequences obtained in the preliminary experiments using several firefly species. Numerals in the name of primers for COI-COII and 16S rDNA indicate nucleotide position in the total mtDNA sequence of Drosophila yakuba [8] corresponding to the 5 end of the primer. The primers ITS2F The Scientific World Journal   FFMT13911R  GTA GTT TTG TAC CTT GTG TAT CAG GGT   ITS2   ITS2F  TGT GAA CTG CAG GAC ACA TG  ITS2R  CCT GTT CGC TCG CAG CTA CT  FFITS2F  GGT GAG CTC GTC CCC GCA TCG  FFITS2R  GTG TAA TAT CAT TTG ATA TCG (1) Numerals in the name of primers for COI-COII and 16S rDNA indicate nucleotide position in the total mtDNA sequence of Drosophila yakuba [8] corresponding to the 5 end of the primer. (2) AAMT12948F was designed in our previous study [9]. and ITS2R were designed based on the aligned homologous sequences of several insects downloaded from the DDBJ nucleotide sequence database. Amplified nuclear ITS2 was cloned using a TOPO TA cloning Kit (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions. One or two colonies were picked from individual insect and directly used for PCR and nucleotide sequencing. In addition to the primers used for PCR, FFMT3140R, FFITS2F, and FFITS2R designed based on sequences obtained in this study were used for sequencing. PCR, labeling, and sequencing were performed as described in Muraji et al. [9]. Sequences of representative individuals were submitted to the DDBJ/EMBL/GenBank nucleotide sequence databases (accession numbers: AB671246-AB671262, AB672623-AB672630).
Sequences were aligned as described in Muraji et al. [9] and used to compute basic statistical data and to generate phylogenetic trees based on the neighbor-joining method using MEGA ver. 4.1 software [10]. Maximum parsimony analyses were performed with PAUP * ver. 4.0b10 [11], using a heuristic search procedure with TBR swapping and 100 max. tree options.   was higher in CO (polymorphic sites: 25.8%, parsimony informative sites: 23.7%) than in rDNA (polymorphic sites: 15.7%, parsimony informative sites: 14.4%).

Results
In the phylogenetic analyses using the neighbor-joining method, the CO, rDNA, and combined (CO + rDNA) data sets produced the same topology in terms of the relationships among the haplotypic groups irrespective of the method used to calculate the genetic distances. The same topology was also recognized in all of the equally parsimonious trees generated using CO (length: 512; CI: 0.801; RI: 0.972; RC: 0.778), rDNA (length: 184; CI: 0.886; RI: 0.982; RC: 0.870), and combined data sets (length: 697; CI: 0.882; RI: 0.974; RC: 0.801). At the basal node of these trees, individuals were divided into two major lineages, group A + B and group C (Figure 2), and the former group was then subdivided into groups A and B.
Groups A, B, and C were specific to the Amami (islands of Amami-oshima and Tokunoshima), Okinawa (Okinoerabujima and Okinawa-jima), and Sakishima regions (Miyakojima, Ishigaki-jima, and Iriomote-jima), respectively. The ranges of groups A and C coincided with that of C. costipennis and that of B coincided with that of C. okinawanus. At the terminal nodes of the phylogenetic trees, 2, 3, and 3 minor haplotypic groups were also detected in groups A, B, and C, respectively. All of these groups were strongly supported by the bootstrap analyses (93-100%).
The nucleotide sequence of ITS2 was determined for 50 clones obtained from 29 individuals. The length and sequence of the fragments were variable among and within the populations of the Amami (1,150-1,171 bp   (1,089-1160 bp long), and, due to insertion/ deletion mutations, the sequences could not be aligned in several sections. Thus, a 856 bp long data set generated excluding 14 (315-329 bp in total length), 15 (322-369 bp), and 13 (247-314 bp) sections from the Amami, Okinawa, and Sakishima populations, respectively, was used for phylogenetic analyses. In these analyses, the topology of the neighbor-joining trees was highly consistent among the different methods used to calculate the genetic distances ( Figure 3). The topology also agreed with that of the equally parsimonious 100 trees (length: 269; CI: 0.903; RI: 0.990; RC: 0.894) obtained by the maximum parsimony method. In these trees, three groups, corresponding to groups A, B, and C in Figure 2, diverged simultaneously at the basal node. Subgroups at the lower level were recognized only in the populations that originated in the Sakishima region. All these groupings were supported strongly by the bootstrap analyses (99-100%).

Discussion
In this study, we found that Japanese Curtos fireflies were divided among three genetically separated local populations, corresponding to C. costipennis in the Amami region, C. okinawanus in the Okinawa region, and C. costipennis in the Sakishima region (Figures 2 and 3). Although the nucleotide sequences of C. okinawanus formed a monophyletic group (group B), those of C. costipennis were separated between two distinct groups, A and C, and the monophyly of this species was not recognized in analyses using both mtDNA  and nuclear DNA sequences (Figures 2 and 3). In addition, the mtDNA haplotypes of C. okinawanus (group B) were positioned between those of C. costipennis (groups A and C) (Figure 2). These results indicate that the currently recognized C. costipennis species is paraphyletic. Otherwise, C. okinawanus must be a subspecies or a geographic strain included within the single species C. costipennis. Although the two species closely resemble each other in both morphology and behavior [7], they are apparently distinct in terms of elytral coloration (C. costipennis: dark yellow except for elytral apex, C. okinawanus: entirely brownish black), and this characteristic is recognized as diagnostically important [6]. However, the present study revealed that such a characteristic did not reflect evolutionary relationships correctly. On the other hand, several researchers recognized variations in body size, the shape of the male genital organ, and the number of punctures on the elytra among local populations of C. costipennis [5,6]. Reevaluation of these characteristics is needed to confirm the taxonomic status of the two species.
Among the three major groups detected in this study, a closer relationship was recognized between C. costipennis and C. okinawanus distributed in neighboring regions (the Amami and Okinawa regions in the Central Ryukyu Islands) than between the populations of C. costipennis separated between remote areas (the Amami and Sakishima regions) ( Figure 4). This phenomenon seems to be consistent with the view of Ohba and Goto [7] that C. okinawanus might have derived from C. costipennis in the Central Ryukyu Islands. As discussed below, their evolutionary history can be well understood by considering the influence of the Pleistocene paleogeography of this area.
Among the combined mtDNA sequences obtained in this study, mean substitution rates of 6.8% (6.1-7.4%) and 5.7% (5.4-6.0%) were estimated between groups A + B and C, and between A and B, respectively. From these values, divergence times of 1.2-1.4 and 1.0-1.1 million years were estimated by applying an mtDNA evolutionary clock, 5-5.7% per million years, calibrated for several coleopteran insect groups [12][13][14]. For these periods, the hypothesized paleogeography of the Ryukyu Islands [1,15] postulated a large paleopeninsula extending from the Chinese continent to the Northern Ryukyu Islands through Taiwan (1.2 and 1.7 million years ago). Recent studies of the paleomarine environment indicated that this peninsula began to break around 1.6 million years ago [16]. Subsequently, the peninsula submerged to form a chain of small islands, which has remained for more than one million years. Although the land 8 The Scientific World Journal configuration in the late Pleistocene is rather controversial, the seabed topography suggests the emergence of several large paleo-islands connecting neighboring islands [2,4].
If such a hypothesis is accepted, the common ancestor of major groups is considered to have dispersed through the Ryukyu Islands along the large paleopeninsula, and their separation is considered to have occurred during the course of the gradual subsidence and subdivision of the paleopeninsula. The divergence times estimated above suggest that the channel between the Central and Southern Ryukyu areas (Kerama Gap in Figure 4) and that between the Amami and Okinawa regions opened sequentially around 1.3 and 1.0 million years ago, respectively. After a long period of isolation among small islands, the local populations must have been connected again by superislands that emerged in the late Pleistocene. Because Miyako-jima Island is known to have submerged beneath the sea in the mid-Pleistocene [17], the occurrence of fireflies on this island suggests a land connection in the late Pleistocene. The integrity of the nucleotide sequences within the respective regions ( Figures  2 and 3) suggest that the super islands in this period were also separated among the Amami, Okinawa, and Sakishima regions. Minor haplotypes at the terminal nodes of the phylogenetic trees are considered to be due to the subdivision of the super islands caused by the sea level rise thereafter. The coexistence of two different minor haplotypes on the islands of Ishigaki-jima, Iriomote-jima, and Okinawa-jima ( Figure 4) suggests that super islands emerged several times in this period.
The fauna of the Ryukyu Islands is known to be separated among the Northern, Central, and Southern Ryukyu areas by geologically long-standing channels at the Tokara and Kerama Gaps (Figure 4). These channels are considered to have long acted as barriers to biological dispersal [3], and their role was reflected in the relictual distribution and endemism of many terrestrial animals including mammals, birds, reptiles, amphibians, and insects [3,18,19] in the Central Ryukyu Islands. Although the faunal borderline is ambiguous between the Amami and Okinawa regions within the Central Ryukyu area, differences in species, subspecies, and mtDNA sequences have also been seen in many terrestrial animals among the regions [3,[19][20][21][22]. As in the fireflies examined in this study, Neolucanus beetles [22] showed sequential population vicariances in which the mtDNA haplotype was first separated between the areas north and south of the Kerama Gap, and the northern half was then further subdivided between the Amami and Okinawa regions, suggesting that the channel between the Amami and Okinawa regions (>500 m depth) opened after the opening of the Kerama Gap (>1,000 m depth) and had slightly less importance in determining faunal differences.
We believe that the speciation of C. okinawanus must be understood in this context. Considering the paleogeography of this area, the repetitive entry of fireflies into the Central Ryukyu Islands from neighboring areas is unlikely. Our conclusion is that the species derived in the Okinawa region from a population of C. costipennis isolated in the Central Ryukyu area. The fact that the two species can engage in interspecific copulation under laboratory conditions [23] suggests that the speciation was based on the geographic segregation between the Amami and Okinawa regions.
In this study, we examined the evolutionary scenario of Curtos fireflies distributed in the Ryukyu Islands. However, because this study did not treat specimens that originated from the Northern Ryukyu Islands and Taiwan, we could not evaluate the function of two geologically long-standing channels at the Tokara Gap and the Yonaguni Depression ( Figure 4). To complete the evolutional scenario of this species group, further studies using these populations as well as populations from the Chinese continent are needed.