Landscape Features and Climatic Forces Shape the Genetic Structure and Evolutionary History of an Oak Species (Quercus chenii) in East China

Abstract

Major topographic features facilitate intraspecific divergence through geographic isolation. This process may be enhanced by environmental isolation along climatic gradients, but also may be reduced by range shifts under rapid climatic changes. In this study, we examined how topography and climate have interacted over time and space to influence the genetic structure and evolutionary history of Quercus chenii, a deciduous oak species representative of the East China flora. Based on the nuclear microsatellite variation at 14 loci, we identified multiple genetic boundaries that were well associated with persistent landscape barriers of East China. Redundancy analysis indicated that both geography and climate explained similar amounts of intraspecific variation. Ecological differences along altitudinal gradients may have driven the divergence between highlands and lowlands. However, range expansions during the Last Interglacial as inferred from approximate Bayesian computation (ABC) may have increased the genetic diversity and eliminated the differentiation of lowland populations via admixture. Chloroplast (cp) DNA analysis of four intergenic spacers (2,866 bp in length) identified a total of 18 haplotypes, 15 of which were private to a single population, probably a result of long-term isolation among multiple montane habitats. A time-calibrated phylogeny suggested that palaeoclimatic changes of the Miocene underlay the lineage divergence of three major clades. In combination with ecological niche modeling (ENM), we concluded that mountainous areas with higher climatic stability are more likely to be glacial refugia that preserved higher phylogenetic diversity, while plains and basins may have acted as dispersal corridors for the post-glacial south-to-north migration. Our findings provide compelling evidence that both topography and climate have shaped the pattern of genetic variation of Q. chenii. Mountains as barriers facilitated differentiation through both geographic and environmental isolation, whereas lowlands as corridors increased the population connectivity especially when the species experienced range expansions.

Keywords: East China; Quercus chenii; chloroplast haplotypes; elevation; environmental isolation; geographic isolation; landscape features; microsatellites.

Figure 1

(A) Sampling sites of Quercus chenii in China, and geographical distribution of three ancestral groups corresponding to the genetic cluster I (red), II (blue), and III (yellow) as inferred by Bayesian cluster analysis based on the genetic variation at 14 nuclear microsatellite loci. Circles and triangles in the center of each pie chart indicate populations located at the high and low elevation regions, respectively. Pie chart sizes are proportional to the sample size. (B) Major genetic boundaries identified by BARRIER 2.2 (Manni et al., 2004). The blue lines represent the Voronoi tessellation. The red, orange, and green lines indicate geographic barriers with bootstrap support of >90%, 70%–90%, and 50%–69.9%, respectively. (C) and (D) show natural habitats of populations WN (lowlands) and HS (highlands), respectively. Population codes are shown in Table 1 and Supplementary Table S1 .

Figure 2

Seven demographic scenarios compared in approximate Bayesian computation (ABC) for both highland and lowland populations of Quercus chenii. (1) constant effective population size at both t ₁ and t ₂; (2) a recent bottleneck at t ₁; (3) an old bottleneck at t ₂; (4) a recent expansion at t ₁; (5) an old expansion at t ₂; (6) an old expansion at t ₂ followed by a recent bottleneck at t ₁; and (7) an old bottleneck at t ₂ followed by a recent expansion at t ₁. Parameter abbreviations include effective population sizes (N ₁–N ₅) and generation-scaled times (t ₁ and t₂).

Figure 3

Heatmap of pairwise F _ST values among the 18 populations of Quercus chenii. Blue and red bars indicate highland and lowland populations, respectively. Population codes are shown in Table 1 and Supplementary Table S1 .

Figure 4

Linear correlations between allelic richness (A _R), genetic diversity within populations (H _S), genetic admixture index (A _D) and elevation of each population of Quercus chenii. Pie charts show proportions of the genetic cluster I (red), II (blue), and III (yellow) as inferred by Bayesian cluster analysis for each population. Red and black circles indicate lowland and highland populations, respectively.

Figure 5

Biplot from redundancy analysis (RDA) showing the association of genetic variation at the 14 nuclear microsatellite (nSSR) loci with geographic and climatic variables. Small red crosses at the center represent the 177 allelic variables. Eight eigenvectors corresponding to positive eigenvalues of the principal coordinates of neighbor matrix (PCNM) were used as geographic variables. Eight climatic variables include elevation, annual mean temperature (bio1), mean diurnal temperature range (bio2), temperature seasonality (bio4), mean temperature of driest quarter (bio9), annual precipitation (bio12), precipitation seasonality (bio15), and precipitation of warmest quarter (bio18). Orange and gray dots indicate individuals from highlands and lowlands, respectively. The proportion of total genetic variation explained by each axis is shown in parentheses.

Figure 6

Geographical distribution (A) and median-joining network (B) of 18 chloroplast DNA haplotypes of Quercus chenii. Circles and triangles in the center of each pie chart in (A) indicate populations located at the high and low elevation regions, respectively. Population codes are shown in Table 1 and Supplementary Table S1 . Circle sizes in (B) are proportional to the frequency of a haplotype across all populations. The small black dots indicate inferred intermediate haplotypes not detected in this investigation. Dash lines indicate two mutations between haplotypes. When branches represent more than two mutations, numbers of mutations are labeled in brackets. F and A represent outgroups, Q. fabri and Q. aliena, respectively. **, shared by seven populations; *, shared by two populations.

Figure 7

BEAST-derived chronograms for 18 chloroplast DNA haplotypes of Quercus chenii, with species from Quercus section Quercus and genus Castanea as outgroups. Blue bars indicate the 95% highest posterior density (HPD) credibility intervals for node ages (million years ago, Ma). Posterior probabilities (>0.9) are labeled above nodes. Geological time abbreviation: Pli, Pliocene; Q, Quaternary. **, shared by seven populations; *, shared by two populations.

Figure 8

(A–C) Putative areas with moderate (yellow) or high (pink) ecological stability for Quercus chenii based on the average (A), minimum (B), and 1-SD (C) of occurrence probabilities across the Last Interglacial (LIG), the Last Glacial Maximum (LGM), the Mid Holocene, and the present estimated by MAXENT 3.4.1 (Phillips et al., 2018), using 80% and 90% (for average), and 50% and 80% (for minimum and 1-SD) of the maximum values as thresholds. (D–F) Potential dispersal corridors during the LGM (D), the Mid Holocene (E), and at the present (F) for Q. chenii estimated by SDMTOOLBOX (Brown, 2014). Black dots indicate the 18 sampling sites of this study.