Introduction

The East Asian region—and in particular, the Japanese Archipelago—is considered to have high insect biodiversity [1], [2]. The high degree of Japanese insect biodiversity is a result of several mechanisms—in particular, the complex geological history. The Japanese Archipelago originated in the middle of the Miocene [3] as an independent formation of eastern and western Japanese landmasses. Extensive geographical changes and large-scale climatic changes throughout the islands facilitated the subsequent connection and disconnection of Japanese landmasses from the Eurasian continent, and the formation of tectonic lines (as the median tectonic line, MTL; and the Itoigawa-Shizuoka tectonic line, ISLT) [3], [4], [5]. These geological events—facilitating for the colonisation of insects from the continent and their subsequent diversification as endemic lineages (i.e. new species)—substantially contributed to the high diversity of insects in Japan [2].

The process of species diversification has been intensively explored through phylogeographical approaches [6], [7]. These approaches have allowed for the observation of the historical process responsible for the current geographical distribution of individuals [6]. Molecular approaches to phylogeographic studies, using specific genes—such as mitochondrial DNA (mtDNA) or nuclear DNA (nDNA)—allow for a better understanding of species diversity by resolving complex taxonomic groups of species (for instance, cryptic species and species groups) [7]. Molecular phylogeography has provided valuable insights into the historical process of Japanese Archipelago formation underlying insect diversification. Previous studies identified genetic differentiation within species between the Japanese landmasses and the Eurasian continent (for instance, the mayflies Isonychia japonica (Ulmer, 1919) [8]; caddisflies Palaeagapetus spp. (Ulmer, 1912) [9] and beetles Ohomopterus spp. (Nakane, 1953) [10] and the Carabina subtribe [11]). Dispersal events via land bridges (islands between continents) from the Eurasian continent to the Japanese Archipelago (of, for instance, the orthopteran Locusta migratoria Linnaeus, 1758, [12]; mayflies Ephron spp. (Williamson, 1802), [13]) or, in reverse, from the Japanese Archipelago to the Eurasian continent (of, for instance, water bugs Appasus spp. Amyot and Serville, 1843, [14]) were additionally identified before, during and after the formation of the Japan Archipelago.

Aquatic insects have advantages in the studies of phylogeography, as their specialised ecological requirements and habitat range make aquatic insect species susceptible to geological changes. Among the Plecoptera order [15], the family Nemouridae is one of the largest and most dominant aquatic insect groups. The family comprises 20 genera and more than 400 species distributed throughout the Northern Hemisphere and across the equator in the Sunda Archipelago [16]. Several genera of the Nemouridae family have distinct disjunctions in their distribution [15]. For example, Ostrocerca Ricker, 1952, Prostoia Ricker, 1952 and Soyedina Ricker, 1952 were found in both the extreme western and the extreme eastern regions of North America, but they were absent in the central area [17], [18]. Similar disjunctive distributions were also observed among Protonemura, Indonemoura, Sphaeronemoura Shimizu & Sivec, 2001, and Illiesonemoura Baumann, 1975 in the Palaearctic region [19], the western and eastern Himalayan ranges [20] and North and South India [15]. Podmosta Ricker, 1952 and Zapada Ricker, 1952 (Nemourinae Billberg, 1820) are two interesting cases distributed across the Nearctic region and East Asia [21], [22]. Previous studies have suggested that their current habitat distribution could be associated with mountain formation and land bridges. In Japan, the Nemouridae family is widely distributed with four genera [23]—Indonemoura; Protonemura; Amphinemura; and Nemoura. To date, 30 Nemoura species, 17 Amphinemura species, 12 Protonemura species and 1 Indonemoura species have been reported in Japan [16]. However, their evolutionary history in the Japanese Archipelago remains unknown.

We studied the molecular phylogeny of the aquatic insect Nemouridae (Plecoptera) in the Japanese Archipelago with comprehensive genera-level sampling using mitochondrial cytochrome c oxidase 1 (cox1) and nuclear histone 3 (H3) markers. We hypothesised that the Nemouridae family diversification could be linked to the geological formation of the Japanese Archipelago. Therefore, we estimated the phylogenetic relationships among Nemouridae species and genera with reference to their historical biogeography. We focused on geographic events of Japanese Archipelago formation and their influence on the divergence time among the genera and species using a combination of fossil records and the Archipelago formation history. Furthermore, to estimate the historical process of the phylogeography of Nemouridae in Japan, we compared the phylogenetic relationships among the specimens from South Korea, China and North America, that are assumed to be the potential sources of Japanese Nemouridae because of the geological formation history of the Japanese Archipelago.

Material and methods

Study sites and sample collection

Our sampling sites in Japan comprised 32 sampling sites on Hokkaido Island, 83 on Honshu Island and 27 on Shikoku Island. The field sampling was conducted without the necessity of any special permit. None of the Nemouridae species was found on Hokkaido Island during sampling. All species reported from Hokkaido are known to occur on either Honshu or Shikoku Islands. Herein, we only reported on the sampling sites wherein specimens were found. We collected 20, 7, 8 and 1 species of the genera Nemoura, Amphinemura, Protonemura and Indonemoura, respectively, on 110 sampling sites in Japan (Fig 1, S1 Table). Additionally, 14 species collected from 8 sampling sites of Mainland China and 2 of South Korea (S1 Table, Fig 2) and 100 specimens of the three species Zapada columbiana (Claassen, 1923), Z. cinctipes (Banks, 1897), and Podmosta delicatula (Claassen, 1923) of subfamily Nemourinae collected from 15 sampling sites of North America (western United States of America and Alaska) were included in our analysis. We added these samples from outside of Japan because of their geographical proximity to the Japanese Archipelago and their geological formation histories. The maps were prepared in QGIS software v 2.18 under the GNU free Documentation License.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. The Japanese islands and distribution of 110 sampling sites from where Nemouridae samples were collected (open circles).

The map was prepared using QGIS v 2.18 under the GNU free documentation License with political boundaries from the Global Database of Administrative Areas (https://gadm.org/).

https://doi.org/10.1371/journal.pone.0210269.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Concatenated Bayesian phylogeny (cox1 + H3) of the East Asian Nemouridae family.

The phylogenetic tree nodes were calibrated using 180 Ma based on fossil records + 15 to 30 Ma based on the Japanese Archipelago formation. Calibration and geological time are shown at the bottom of the tree. A 95% HPD is indicated as a horizontal grey bar and posterior probabilities are shown for each node. Circle symbol (ᴏ) in the nodes indicates intra-species diversification based on the eastern and western Japanese boundaries of the Fossa Magna region. Inserted upper map shows sample site locations for Japan (black), China (red) and South Korea (blue) as dots. Colour branches indicate sample location distribution as shown in the map. The map was prepared using QGIS v 2.18 under the GNU free documentation License with political boundaries from the Global Database of Administrative Areas (https://gadm.org/).

https://doi.org/10.1371/journal.pone.0210269.g002

We collected adult insects using hand nets around riversides. We stored samples in 80% ethanol in the field, and replaced the ethanol with fresh 99.5% ethanol after morphological identification. We identified individuals according to the taxonomical keys of [21], [23], [24], [25], [26], [27], [28], [29], [30], [31] and [32]. Undescribed species in our study were morphologically sorted based on our taxonomic expertise and inconclusive taxonomic keys.

Results

Genetic diversity and DNA phylogeny

For studying the phylogeny of Nemouridae in the Japanese Archipelago, we analysed two molecular markers. Cox1 sequences were of 658 bp length, with 247 polymorphic sites, 237 parsimony-informative sites, 10 singletons and a mean nucleotide substitution rate of 0.151. H3 sequences were of 328 bp length, with 67 polymorphic sites, 54 parsimony-informative sites, 13 singletons and a mean nucleotide substitution rate of 0.051. No gaps were detected for either cox1 or H3 sequences. In total, for cox1 and H3, we identified 128 and 68 haplotypes, respectively.

The log-likelihood of the GMYC model at the optimal threshold (111.02) was significantly better than the null model of a single coalescent (logL = 56.99) in the likelihood ratio test (p < 0.001). Most clades have GMYC-support values higher than 0.9, implying that the probability of the clades being delimited as separate GMYC-species among the alternative models of delimitation (within a 95% confidence set) is higher than 0.9. The single-threshold model delimited 61 putative species (S1 Table) (confidence interval: 58–65) composed of 39 clusters (confidence interval: 38–42), indicating that some of the inferred putative species were singletons (i.e. only one sequence). These results agreed with our 34 morphologically identified and 15 undescribed (five species of Protonemura, seven of Nemoura, one of Indonemoura and two of Amphinemura) species. Eight species (I. nohirae (Okamoto, 1922), A. decemseta Okamoto, 1922, A. zonata Okamoto, 1922, A. longispina Okamoto, 1922, A. megaloba (Kawai, 1960), N. uenoi Kawai, 1954, N. chinonis Okamoto, 1922, and N. cf. cercispinosa Kawai, 1960) showed two putative DNA-species. While A. decemseta showed multiple putative DNA-species (three putative DNA-species), N. sanbena Shimizu (1993) and P. kohnoae Shimizu, 1998, showed two putative DNA-species in the same sampling site suggesting the presence of cryptic species. The congruence of H3 phylogenetic groups provided confirmation of DNA-based groups detected by GMYC.

We observed the genetic diversity of the species per island (Table 1). Honshu had the highest number of species (26 species), haplotype richness (63) and mean nucleotide substitution rate (average 0.027). Five species were found throughout the three Japanese islands (Honshu, Shikoku and Kyushu), i.e. A. decemseta, A. zonata, N. cf. cercispinosa, N. chinonis and I. nohirae, with a mean nucleotide substitution rate ranging from 0.011 to 0.126 and a total of 23 haplotypes. N. sanbena haplotypes were observed in two different branches in the phylogenetic tree, both within N. cf. cercispinosa and as an isolated branch.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Regional distribution of sample size (n), haplotype richness (h) and mean nucleotide substitution rate of Nemouridae species among the three main islands in Japan, based on mitochondrial DNA (cox1) sequences.

Total species richness was 26, 23 and 6 for Honshu, Shikoku and Kyushu, respectively.

https://doi.org/10.1371/journal.pone.0210269.t001

Phylogeographic pattern between Nemoura from Japan and North America

DNA sequences in the Japanese clade of Indonemoura (single species, I. nohirae) showed a high homology with those in the Alaskan species of Z. columbiana (COI: KM874174; >93% sequence similarity) and Z. cinctipes (H3: EF622600; >98% sequence similarity) based on blastn results.

Based on the Blast results and the previous reported geographical distribution of these species in Asia and North America, we decided to observe the phylogenetic relationships of these Alaskan specimens (Z. columbiana, Z. cinctipes and P. delicatula) with stoneflies taxa from Japan. The ML phylogenetic trees for both cox1 and H3 (Fig 3) showed that the Indonemoura Japanese clade clustered with three North American species (Z. columbiana, Z. cinctipes and P. delicatula) and the Indonemoura Mainland China clade clustered with the East Asian Nemouridae genera (Nemoura, Protonemura and Amphinemura). The pairwise nucleotide substitution rate based on cox1 between the Indonemoura Japanese clade and Zapada spp. or P. delicatula from North America ranged from 0.13 to 0.15, whereas a higher pairwise nucleotide substitution rate based on cox1 of 0.26 was observed between the Indonemoura Japanese and Mainland China clades (Table 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Maximum likelihood trees based on both cox1 and H3 markers for comparison between the East Asia Nemouridae family and three North American Nemourinae species: Zapada cinctipes, Z. columbiana and Podmosta delicatula.

Inserted upper map shows sampling site locations in North America (western USA and Alaska) as black dots. The map was prepared using QGIS v 2.18 under the GNU free documentation License with political boundaries from the Global Database of Administrative Areas (https://gadm.org/).

https://doi.org/10.1371/journal.pone.0210269.g003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Pairwise nucleotide substitution rate based on cox1 between the East Asian Nemouridae and North American (western USA and Alaskan) species.

https://doi.org/10.1371/journal.pone.0210269.t002

Discussion

We studied mitochondrial cox1 and nuclear H3 gene sequences to determine the patterns of diversification and phylogenetic relationships of species belonging to four genera of stoneflies of the Nemouridae family in the Japanese Archipelago. We estimated the divergence among Nemoura, Amphinemura, Indonemoura and Protonemura to have occurred in the early and mid-Cretaceous (around 100 Ma), which is compatible with previous studies based on fossil records [56], [57]. Our results suggested that these four genera might have dispersed and colonised different areas of the Eurasian continent—including the Japanese landmasses—when they were still connected to the Eurasian continent. Among the four genera, the diversification of Indonemoura occurred earlier (120 Ma) than that of the other genera (<100 Ma), suggesting that it is an ancient genus. The geological isolation of colonised areas [58], the long evolutionary time [59] and poor dispersal ability of Indonemoura [23], [60] might have accounted for their ancient diversification.

Based on the phylogenetic relationships of both molecular markers (cox1 and H3), we observed that the three genera Nemoura, Amphinemura and Protonemura were monophyletic and clustered as three independent groups, as previously observed by morphological systematics [16]. However, Indonemoura was paraphyletic. This genus was divided into two clades corresponding to the Mainland China and the Japanese clades. Surprisingly, the Japanese clade of Indonemoura (single species, I. nohirae) clustered together with North American species (Z. columbiana, Z. cinctipes and P. delicatula), with a low pairwise nucleotide substitution rate (<0.15). The distribution range of these two North America genera covers North America and Eastern Asia. Previous studies suggested that their distribution could be related to the land connection (i.e. the islands) between Alaska and Eastern Asia [15]. Dispersal by island connectivity between Alaska, the Aleutian Islands, the Kamchatka peninsula and the Kuril Islands has been observed in other stonefly families (for instance, Arcynopteryx dichroa (McLachlan, 1872), Capnia nearctica Banks, 1919, Mesocapnia variabilis (Klapálek, 1920) and Nemoura arctica Esben-Petersen, 1910) [61]. However, the distribution of Indonemoura on these islands is unknown.

The complex history of the geological formation of the Japanese Archipelago may provide a possible alternative explanation. The ancestral Japanese landmasses were located on the borders of four major tectonic plates, of which two are continental plates—the Eurasian plate and the North American plate [4] (S2 Fig). The eastern Japanese landmass was located on the North American plate, whereas the western Japanese landmass was located on the Eurasian plate [5]. The dispersal and colonisation of Indonemoura might have occurred from the North American plate to the Eurasian continent or vice versa (from the Eurasian continent to the North American plate) before their geographic separation in an ancient time (around 70 to 80 Ma) [62]. Dispersal events between Eurasian and Japanese landmasses are commonly reported for aquatic insects [10], [63]. Particularly, a dispersal event between North America and the Japanese Archipelago was detected by the phylogenetic relationship of the monophyletic group of caddisflies, Palaeagapetus spp. [9]. However, no prior studies have observed speciation events of aquatic insects associated with geological events that occurred in ancient times (>12 Ma). Our result suggests an ancient divergence time and a distribution pattern of Indonemoura, consistent with a hypothesis of an ancient colonisation influenced by the connection of the Japanese landmass with the North American plate in the Eurasian continent.

Nemouridae species diversification, as has been observed in other species of aquatic insects, such as beetles [10], caddisflies [9], water bugs [14] and mayflies [8], [13], was also observed to be affected by the geological formation of the Japanese Archipelago. The diversification of the Nemouridae species occurred during the Paleogene period (<50 Ma). This geological period is consistent with the movement of landmasses (S2 Fig) about 70 Ma ago [4] and the active geological formation of the Japanese Archipelago around 20 Ma ago [5], which could be the cause of the Nemouridae diversification, as previously reported for the mayfly Dipteromimus flavipterus Tojo and Matsukawa, 2003 (35 Ma) [2].

Indonemoura nohirae is the single species of Indonemoura on the Japanese Archipelago [25], [26]. The morphology of their terminalia resembles that of Protonemura rather than of Indonemoura, but the characteristic gill formula justifies their taxonomical classification in Indonemoura [25], [26]. To date, there are 24 Indonemoura species from China [16], [24] and 30 species belonging to the Himalayan and Oriental regions in East Asia [15], [20]. These species are morphologically different from I. nohirae in Japan [15], [16], [20], [24], [25], [26]. We hypothesise that the Indonemoura species of East Asia could be forming separate phylogenetic clades clustered by geographical regions. For the hypothesis testing, further collection of molecular data on Indonemoura from wider areas such as Northeast China, Southeast China, Mongolia, Russia and other countries in Asia is needed in future studies.

Eight species (I. nohirae, A. decemseta, A. zonata, A. longispina, A. megaloba, N. chinonis, N. uenoi and N. cercispinosa) showed interesting patterns of intra-species separation into two genetic groups corresponding to eastern and western areas of the Fossa Magna region of Honshu Island (S1 Fig). Honshu is the centre of insect biodiversity [10]; apart from its extensive territorial space, it is the main island with a geological history [3], [4], [5]. We found supporting evidence on the genetic diversity of these eight species. We found a larger mean nucleotide substitution rate and haplotype number in the Honshu region than in other islands (Table 1). The mean nucleotide substitution rate and haplotype diversity are indications of biodiversity [64], which could lead to evidence of speciation [65]. Out of eight species, the diversification of six species (A. decemseta, A. zonata, A. longispina, A. megaloba, N. uenoi and N. cercispinosa) occurred during the late Neogene period (20 to 22 Ma). This event corresponded with the double-door (i.e. the union of eastern and western Japan; S2 Fig) geological model and the formation of the Itoigawa-Shizuoka tectonic line (ISLT) at around 20 Ma [5], [66]. The speciation of aquatic insects was often observed to be influenced by these two geological events [2]. Additionally, species diversification—from eastern or western Japan of the Fossa Magna region—showed recent diversification events (3 to 5 Ma) corresponding with the formation of the small islands in northeastern or southwestern edge areas of Japan (Fig 1). The northeastern islands created land bridges between the Japanese Archipelago and China or Korea, whereas the southwestern islands connected Taiwan or the Philippines with the Japanese Archipelago [5], [66]. This connectivity promoted immigration events in Japan that might have contributed to the formation of the current genetic diversity, as previously observed in mayflies [13] and beetles [10].

The evolutionary divergence of the Nemouridae family was promoted by the complex geological formation of the Japanese Archipelago. Despite the different evolutionary rate of both molecular markers, Bayesian analysis found congruence between them; however, failed to find congruence with their morphological taxonomy. The main morphological character used for identification of adult stoneflies species is its genital morphology. The evolution of genital morphology is; however, governed by within-population sexual selection rather than environmental or geological history of the locations [67]. Conversely, the genetic variation of natural populations has been observed to be directly associated with environmental [68] and geological variations [2]. Therefore, the genetic variation could reflect an independent course in the evolutionary history of Indonemoura than do the morphological characters used for their taxonomy. However, we detected that N. sanbena shared haplotypes from different lineages, revealing a possible introgression or incomplete sorting of ancestral polymorphisms [10]. This is an often reported phenomenon in stoneflies [40], [69], [70], which remains as unresolved species. Resolving the problems between the process of evolution of morphological characters and the genetic variation within species will improve our future understanding of the origin of the species and the local species distribution.

Finally, our inference of divergence time was based on the coalescent simulation approach. Despite the frequent use of this approach, a biased sampling of lineages and extreme state-dependent molecular substitutions rate heterogeneity are known to potentially cause erroneous inference of divergence time [71]. One of the causes of biased sampling is under-sampling (i.e. the incomplete-coverage samples included in the phylogenetic tree). Previous studies demonstrated that under-sampling increases the estimates of the posterior probabilities (i.e. variance of age estimates become less precise) [72] and led to bias and low precision of the divergence time of shallow nodes (i.e. evolutionary recently divergent taxa) [73], [74]. However, under-sampling tends not to affect deep nodes (i.e. internal nodes close to the root) time scales estimations or the tree shape [75], [76], [77]. Therefore, in order to confirm the accuracy of the time scales estimations of the recently divergent taxa in our study, additional taxonomic samples would be required to be included in the molecular clock analysis. Conversely, the state-dependent molecular substitution rate heterogeneity could be address by combining node calibrations generated by more than one calibration analyses, as recommended by [71], [78]. A cautious method such as the combined uses of fossil records and biogeographic ages as employed in our analysis may minimize the risk of such erroneous inference.

Supporting information

Acknowledgments

We thank the following colleges for contributing Nemouridae specimens for our analysis: Akatsuki Yoshinari, Kobayashi Kyota, Weihai Li and Boris C. Kondratieff. We also acknowledge anonymous reviewers for constructive comments on an earlier version of the manuscript.