Contrasted patterns of local adaptation to climate change across the range of an evergreen oak, Quercus aquifolioides

Abstract

Long-lived tree species are genetically differentiated and locally adapted with respect to fitness-related traits, but the genetic basis of local adaptation remains largely unresolved. Recent advances in population genetics and landscape genomic analyses enable identification of putative adaptive loci and specific selective pressures acting on local adaptation. Here, we sampled 60 evergreen oak (Quercus aquifolioides) populations throughout the species' range and pool-sequenced 587 individuals at drought-stress candidate genes. We analyzed patterns of genetic diversity and differentiation for 381 single nucleotide polymorphisms (SNPs) from 65 candidate genes and eight microsatellites. Outlier loci were identified by genetic differentiation analysis and genome-environment associations. The response pattern of genetic variation to environmental gradient was assessed by linear isolation-by-distance/environment tests, redundancy analysis, and nonlinear methods. SNPs and microsatellites revealed two genetic lineages: Tibet and Hengduan Mountains-Western Sichuan Plateau (HDM-WSP), with reduced genetic diversity in Tibet lineage. More outlier loci were detected in HDM-WSP lineage than Tibet lineage. Among these, three SNPs in two genes responded to dry season precipitation in the HDM-WSP lineage but not in Tibet. By contrast, genetic variation in the Tibet lineage was related to geographic distance instead of the environment. Furthermore, risk of nonadaptedness (RONA) analyses suggested HDM-WSP lineage will have a better capacity to adapt in the predicted future climate compared with the Tibet lineage. We detected genetic imprints consistent with natural selection and molecular adaptation to drought on the Qinghai-Tibet Plateau (QTP) over a range of long-lived and widely distributed oak species in a changing environment. Our results suggest that different within-species adaptation processes occur in species occurring in heterogeneous environments.

Keywords: Fagaceae; Hengduan Mountains; Quercus; Tibet; adaptation; genomic divergence; landscape genomics.

FIGURE 1

Geographical distribution, location of study sites, and mapping of the Bayesian genetic clusters in Quercus aquifolioides. (a) Species range (gray shadow) and sampling locations (black dots). (b) Two main clusters indicated by red (Tibet lineage) and green (HDM‐WSP lineage) based on Bayesian cluster analysis of eight nuclear microsatellite markers from 587 individuals in 60 populations

FIGURE 2

Principal component analysis based on (a) all SNPs, (b) non‐F _ST outlier SNPs, and (c) SSRs. Cool color dots (blue/green) illustrate Tibet lineage, and warmer color dots (red/orange) represent HDM‐WSP lineage

FIGURE 3

Venn diagrams showing genes containing outlier SNPs detected in the Tibet and HDM‐WSP lineages by (a) BAYESCAN and FDIST2 (i.e., F _ST outlier SNPs) and (b) BayEnv and LFMM (i.e., GEA outlier SNPs). The number of genes is given in each area, and the total number of genes for each lineage using each method is given in parentheses. The common and lineage‐specific outlier genes are in bold

FIGURE 4

Mantel tests of genetic distance (F _ST/(1 − F _ST)) against geographic and environmental distances in each lineage and for all populations combined

FIGURE 5

Ranked importance of environmental variables based on analysis of a generalized dissimilarity model (GDM) for each lineage and for all populations combined. Exact values are shown in Table S14

FIGURE 6

Risk of nonadaptedness plot (RONA) for three environmental factors under RCP26 (top panels) and RCP 85 (lower panels) prediction models in 2070. Bars represent weighted means (by R ² value), and lines represent standard error for each population in the Tibet and HDM‐WSP lineages. The exact R ² values are shown in Table S16