A strong 'filter' effect of the East China Sea land bridge for East Asia's temperate plant species: inferences from molecular phylogeography and ecological niche modelling of Platycrater arguta (Hydrangeaceae)

Abstract

Background: In East Asia, an increasing number of studies on temperate forest tree species find evidence for migration and gene exchange across the East China Sea (ECS) land bridge up until the last glacial maximum (LGM). However, it is less clear when and how lineages diverged in this region, whether in full isolation or in the face of post-divergence gene flow. Here, we investigate the effects of Quaternary changes in climate and sea level on the evolutionary and demographic history of Platycrater arguta, a rare temperate understorey shrub with disjunct distributions in East China (var. sinensis) and South Japan (var. arguta). Molecular data were obtained from 14 P. arguta populations to infer current patterns of molecular structure and diversity in relation to past (Last Interglacial and Last Glacial Maximum) and present distributions based on ecological niche modelling (ENM). A coalescent-based isolation-with-migration (IM) model was used to estimate lineage divergence times and population demographic parameters.

Results: Combining information from nuclear/chloroplast sequence data with nuclear microsatellites, our IM analyses identify the two varieties as genetically distinct units that evolved in strict allopatry since the mid-Pleistocene, c. 0.89 (0.51-1.2) Ma. Together with Bayesian Skyeline Plots, our data further suggest that both lineages experienced post-divergence demographic growth, followed by refugial isolation, divergence, and in the case of var. arguta post-glacial admixture. However, past species distribution modelling indicates that the species' overall distribution has not greatly changed over the last glacial cycles.

Conclusions: Our findings highlight the important influence of ancient sea-level changes on the diversification of East Asia's temperate flora. Implicitly, they challenge the notion of general temperate forest expansion across the ECS land bridge, demonstrating instead its 'filter' effect owing to an unsuitable environment for certain species and their biological (e.g., recruitment) properties.

Figure 1

Distribution of ITS ribotypes and the 95% plausible network of ribotypes in Platycrater arguta. (a) Distribution of ITS ribotypes. All ribotypes, except for H24 denoted by orange colour, are population-specific, and represented with different colours corresponding to the major phylogroups. Population codes are identified in Additional file 1: Table S1. (b) TCS-derived network of genealogical relationships between the 33 ribotypes. The small, solid black circles represent missing ribotypes. The sizes of circles are approximately proportional to sample size (n), with the smallest circles representing n = 1 and the largest representing n = 10.

Figure 2

Distribution of Tpi haplotypes and the 95% plausible network of Tpi haplotypes in Platycrater arguta. (a) Distribution of Tpi haplotypes. Population codes are identified in Additional file 1: Table S1. The distributions of shared haplotypes are denoted by colour, while private haplotypes are white. (b) TCS-derived network of genealogical relationships between the 31 haplotypes. The small, solid black circles represent missing haplotypes. The sizes of circles are approximately proportional to sample size (n), with the smallest circles representing n = 1 and the largest representing n = 21.

Figure 3

Phylogenetic relationships of ITS ribotypes (H1–H33) of Platycrater arguta. Individuals of Hydrangea anomala, H. chinensis and Schizophragma hydrangeoides were used as outgroup taxa. Phylogenetic analyses using maximum parsimony (MP) and maximum likelihood (ML) produced trees with the same topology regarding major lineages. Only the MP strict consensus tree is presented. Numbers above and below the branches indicate, respectively, MP and ML bootstrap values (> 50%).

Figure 4

Phylogenetic relationships of Tpi haplotypes (T1–T31) of Platycrater arguta. Hydrangea chinensis was used as outgroup taxon. Phylogenetic analyses using maximum parsimony (MP) and maximum likelihood (ML) produced trees with the same topology regarding major lineages. Only the MP strict consensus tree is presented. Numbers above and below the branches indicate, respectively, MP and ML bootstrap values (> 50%).

Figure 5

The proportion of genetic clusters detected by STRUCTURE analysis for the model with peaks at K = 2 and K = 7. The smallest vertical bar represents one individual. The assignment proportion of each individual into population clusters is shown along the y-axis. Note that STRUCTURE provided strongest support for K = 7, both when considering the probability of the data LnP(D) and ΔK (see text and Additional file 7: Figure S1).

Figure 6

Modelled climatically suitable areas of Platycrater arguta in East Asia at different times. (a) the present; (b) the Last Interglacial (LIG/Eemian: c. 130,000–114,000 yr BP); and (c) the Last Glacial Maximum (LGM: c. 21,000–18,000 yr BP). The current ecological niche model was established with six bioclimatic data layers on the basis of 59 sites of presence records of the species (black dots) using MAXENT 3.2.1 [76] and then projected onto a set of climatic variables simulated by MIROC 3.2 [80] to infer the extent of suitable habitats during the LGM and the LIG (see text). The map in (c) reflects changes in coastline and shelf exposition during the LGM due to lowered sea level (−110 m than at present; e.g., [7]). Yellow River and Yangtze Rive plaeo-channels in the exposed East China Sea basin are modified after [86]. The logistic value of habitat suitability is shown according to the grey-scale bars.