Diversification in a fluctuating island setting: rapid radiation of Ohomopterus ground beetles in the Japanese Islands

Abstract

The Japanese Islands have been largely isolated from the East Asian mainland since the Early Pleistocene, allowing the diversification of endemic lineages. Here, we explore speciation rates and historical biogeography of the ground beetles of the subgenus Ohomopterus (genus Carabus) based on nuclear and mitochondrial gene sequences. Ohomopterus diverged into 15 species during the Pleistocene. The speciation rate was 1.92 Ma(-1) and was particularly fast (2.37 Ma(-1)) in a group with highly divergent genitalia. Speciation occurred almost solely within Honshu, the largest island with complex geography. Species diversity is highest in central Honshu, where closely related species occur parapatrically and different-sized species co-occur. Range expansion of some species in the past has resulted in such species assemblages. Introgressive hybridization, at least for mitochondrial DNA, has occurred repeatedly between species in contact, but has not greatly disturbed species distinctness. Small-island populations of some species were separated from main-island populations only after the last glacial (or the last interglacial) period, indicating that island isolation had little role in speciation. Thus, the speciation and formation of the Ohomopterus assemblage occurred despite frequent opportunities for secondary contact and hybridization and the lack of persistent isolation. This radiation was achieved without substantial ecological differentiation, but with marked differentiation in mechanical agents of reproductive isolation (body size and genital morphology).

Figure 1

(a) The Japanese Islands and the distribution of Ohomopterus ground beetles. The coastline 20 000 years ago (the last glacial period) is based on Ohshima (1990) and Yonekura et al. (2001). (b) Number of species in each prefecture or island. The green line encloses localities where two species (large and small) can co-occur, and the red line where three (occasionally four or five) species can co-occur.

Figure 2

Species relationships in Ohomopterus with the ranges of the three species groups, male copulatory piece and gross morphology of male (left) and female (right) beetles. The tree is a maximum-likelihood (ML) tree resulting from simultaneous analysis of six nuclear DNA sequences (4164 bp). The ML analysis with a genetic algorithm was performed in Garli v. 0.951 (Zwickl 2006) with the GTR+I+G substitution model. Node support was obtained by 1000 bootstrap pseudoreplications (bootstrap percentages shown when greater than 50%). The Bayesian posterior probability of a node is shown following the bootstrap value, based on a partitioned analysis with MrBayes v. 3.12 (Huelsenbeck & Ronquist 2001). Node ages were determined using PATHd8 (Britton et al. 2007) with 2.14 Ma to the divergence between Isiocarabus and Ohomopterus.

Figure 3

Phylogeny of mitochondrial ND5 and distribution of species among lineages. A neighbour-joining tree was constructed using PAUP^* v. 4.10b (Swofford 2002) with a substitution model (GTR+I+G) and parameters selected by Modeltest v. 3.07 (Posada & Crandall 1998) for distance correction. Node supports resulting from 1000 bootstrap analyses are shown only for major nodes (when greater than 50%). Nh, number of haplotypes.

Figure 4

Haplotype clade composition of each species based on the number of ND5 haplotypes (large circles) and the pattern of sharing of clades and haplotypes between sympatric/parapatric species (lines connecting species: dashed lines, no clade shared; thin solid line, clade shared; thick solid line, haplotype shared). For each species, the number of sample beetles for ND5 sequences (n), the total number of haplotypes (Nh) and nucleotide diversity (d; %±s.d.) are indicated with the number of haplotypes within each clade. Clades with single asterisks are those with introgressed haplotypes, as evidenced by a haplotype sharing a pattern with another species. Clades with double asterisks are those putatively originating from introgression. Putative original haplotype clades (those not affected by introgression) are indicated by small circles.

Figure 5

Distribution of ND5 lineages by prefecture and island. Coloured areas indicate distribution ranges and small open circles indicate sampling localities. The pie graph for each prefecture/island shows the haplotype clade composition among individuals sampled, showing the geographical pattern of occurrence of different haplotype clades. Arrows on the distribution maps indicate the direction of dispersal for some haplotype clades.

Figure 6

Statistical parsimony networks of ND5 haplotypes showing the geographical relationships between internal and tip clades. We inferred directions of dispersal as from the geographical ranges of interior clades to those of tip clades. (a) japonicus, (b) dehaanii, (c) albrechti and (d) insulicola.