Quaternary climate change drives allo-peripatric speciation and refugial divergence in the Dysosma versipellis-pleiantha complex from different forest types in China

Abstract

Subtropical China harbours the world's most diverse temperate flora, but little is known about the roles of geographical and eco-climatic factors underlying the region's exceptionally high levels of species diversity and endemism. Here we address this key question by investigating the spatio-temporal and ecological processes of divergence within the Dysosma versipellis-pleiantha species complex, endemic to subtropical China. Our cpDNA phylogeny showed that this monophyletic group of understory herbs is derived from a Late Pliocene ancestor of the Qinghai-Tibetan Plateau (QTP)/Southwest China. Genetic and ENM data in conjunction with niche differentiation analyses support that the early divergence of D. versipellis and D. pleiantha proceeded through allo-peripatric speciation, possibly triggered by Early Pleistocene climate change, while subsequent climate-induced cycles of range contractions/expansions enhanced the eco-geographical isolation of both taxa. Furthermore, modelling of population-genetic data indicated that major lineage divergences within D. versipellis likely resulted from long-term allopatric population isolation in multiple localized refugia over the last glacial/interglacial periods, and which in turn fostered endemic species formation (D. difformis, D. majoensis) from within D. versipellis in Southwest China. These findings point to an overriding role of Quaternary climate change in triggering essentially allopatric (incipient) speciation in this group of forest-restricted plant species in subtropical China.

Figure 1

(a) Geographic distribution of the 56 chloroplast (cp) DNA (trnL‒trnF, trnL‒ndhJ, trnS‒trnfM) haplotypes (H1–H56) detected in the Dysosma versipellis-pleiantha complex, created in illustrator v15.0 (http://www.adobe.com/products/illustrator.html) (see Supplementary Table S2 for population codes). Red dashed and solid lines denote the major western vs. central-east split within D. versipellis s. lat. (i.e. D. versipellis, D. difformis, D. majoensis) and the divergence between the latter and D. pleiantha further east, respectively, as identified by phylogenetic tree and network analyses. (b) Geographic ranges of D. versipellis, D. pleiantha, D. difformis, and D. majoensis, created in illustrator. The base map was drawn using ArcGis v.9.3 (ESRI, Redlands, CA, USA). (c) Ninety-five percent statistical parsimony network of the 56 cpDNA haplotypes identified in the species complex generated in tsc v1.21 (http://darwin.uvigo.es/software/tcs.html). The small open circles represent missing haplotypes. The size of circles corresponds to the haplotype frequency.

Figure 2. Bayesian inference (BI) phylogeny of the Dysosma versipellis-pleiantha complex based on cpDNA (trnL‒trnF, trnL‒ndhJ, trnS‒trnfM) sequences with the remaining Dysosma species treated as part of the ingroup and Podophyllum peltatum and Sinopodophyllum hexandrum as outgroups.

Posterior probabilities (PP > 0.50) and maximum likelihood (ML) bootstrap values (>50%) are sequentially indicated above the branches. Colored branches identify major haplotype lineages (west, central-east, east) within the species complex. Symbols following the haplotype numbers indicate the species bearing this haplotype. Nodes of interest are marked as A–G, while the corresponding beast-derived age estimates (including their 95% HPD intervals) are shown in Supplementary Table S3.

Figure 3

(a) Histogram of the assignment test for 40 populations (577 individuals) of the Dysosma versipellis-pleiantha complex based on genetic variation at 15 EST-SSR loci using structure v2.3.4 (http://pritchardlab.stanford.edu/structure.html). Each vertical bar represents one individual and its probability of membership for each of the K = 4 clusters. (b) Geographic distribution of the four structure clusters within and among populations of the species complex created in illustrator v15.0 (http://www.adobe.com/products/illustrator.html). The symbol next to each sampling locality identifies the respective Dysosma species, while the filled color represents the cluster assigned to that population (population codes are identified in Supplementary Table S2). The base map was drawn using ArcGis v.9.3 (ESRI, Redlands, CA, USA). (c) Principal coordinates analysis (PCoA) of the 577 individuals from the four species of the complex based on EST-SSR variation (15 loci) using arlequin v3.5 (http://cmpg.unibe.ch/software/arlequin35/).

Figure 4

Predicted distributions of D. versipellis s. lat. (including D. versipellis, D. difformis and D. majoensis), and D. pleiantha under (a,b) current climate conditions (1950–2000) and (c,d) the Last Glacial Maximum (LGM; 21 kya BP) based on ecological niche modelling using maxent v3.3.1 (http://www.cs.princeton.edu/~schapire/maxent/). Maps were generated using ArcGis v9.3 (ESRI, Redlands, CA, USA).