Distinct late Pleistocene subtropical-tropical divergence revealed by fifteen low-copy nuclear genes in a dominant species in South-East China

Abstract

In East Asia, genetic divergence is usually considered to be correlated to different floristic regions, however, subtropical-tropical divergence is largely ignored, compared to widely explored temperate-subtropical divergence. Lindera aggregata (Lauraceae), a dominant species in South-East China was selected to address this issue. Fifteen low-copy nuclear genes (LCGs) and four chloroplast DNA (cpDNA) fragments were used to detect its evolutionary history. In LCGs, STRUCTURE and dated Bayesian phylogeny analyses detect distinct subtropical-tropical divergence since late Pleistocene. Approximate Bayesian calculation (ABC) further supports the distinct subtropical-tropical divergence, and close related Taiwan and South China populations are diverged at the last interglacial. Isolation by distance, isolation by environment and isolation by resistance analyses suggest the current climatic difference rather than geographical distance contributes to the genetic differentiation. Principle component analysis shows populations of tropical cluster occur in warmer area with higher precipitation. Ancestral area reconstruction based on Bayesian phylogeny indicates that ancestral L. aggregata populations are distributed in tropical region. In cpDNA, although unique haplotypes are found in tropical region, distinct subtropical-tropical divergence is absent. In conclusion, distinct late Pleistocene subtropical-tropical divergence of L. aggregata is triggered by climate. It is likely that L. aggregata is originated in Southwest-South China and experienced hierarchical dispersal from south to north. The South China Sea land bridge has dual role in connecting or isolating Taiwan and mainland populations since the last glaciation.

Figure 1

(a) Colour-coded grouping of the 29 Lindera aggregata populations according to STRUCTURE with the most likely group number K = 5 (ArcGis 10.2, ESRI, www.esri.com), the dashed black line represents the boundary of subtropical (Sino-Japanese Floristic Region) and tropical (Paleotropic Floristic Region) regions. The shaded area indicates the species’ distribution range. (b) BEAST-derived phylogeny for the 29 populations. The colour of populations indicates their genetic cluster as inferred by STRUCTURE. Statistically significant posterior probabilities (PP > 0.95) are labeled in black dots.

Figure 2

The three divergence scenarios with posteriors probability (PP) among southern, northern and Taiwan clusters of Lindera aggregata (a–c). The effective population size of the three clusters is labeled as N_south, N_north and N_tw. t1/t2, divergence times for the depicted event.

Figure 3

(a) The four major floristic divisions (A–D) in southeast China according to Wu et al., (ArcGis 10.2, ESRI, www.esri.com), and (b) ancestral area reconstructions based on the Bayesian binary Markov chain Monte Carlo (BBM) method implemented in RASP using the BEAST-derived phylogeny of 29 Lindera aggregata populations (see Fig. 1b). Pie charts of each node illustrate the marginal probabilities for each alternative ancestral area derived from BBM, black colour indicate unknown ancestral area.

Figure 4

Principal component analysis (PCA) plots with 19 climatic variables (see Supplementary Table S3) of 29 Lindera aggregata populations. Different colours correspond to the five genetic clusters inferred by STRUCTURE analysis.