Supplemental information:
Figure S1 Phenotypes of the different populations of Aquilegia viridiflora complex in the common garden.
Figure S2 A. k-means cluster analysis of phenotype based on the first two principal components, B. Principal component analysis (PCA) plot for the 66 A. viridiflora complex individuals based on the first two principal components.
Figure S3 Phylogenetic relationship between Aquilegia viridiflora complex and other columbine species whose distribution was overlap with Aquilegia viridiflora complex. A. Neighbor-joining (NJ) tree, B. maximum likelihood (ML) tree.
Figure S4 Correlation analysis between phenotypes. * represents a significant relationship between different phenotypes, 0.01 < p <0.05. ** represents a significant relationship between different phenotypes, 0.001 < p <0.01. *** represents a significant relationship between different phenotypes, p <0.001.
FigureS5Cross-validation results corresponding to different K values in the ADMIXTURE analysis of A. viridiflora complex.
Figure S6 A. Median-joining haplotype network of 66 A. viridiflora complex based on chloroplast genome, B. NeighborNet diagram based on chloroplast genome.
Figure S7 Five gene flow events of Aquilegia viridiflora complex. Each branch represents a population, and arrows indicate migration events that occur between populations.
Figure S8 A. Four groups of possible models for inferring demographic parameters. The gray arrows represent the current gene flow, the red arrows represent the ancient gene flow, and the model in the dotted line is the best model selected according to the values of the likelihoods and AIC.
Figure S9 Comparison of observed data and expected data based on the best model for A. NE and EL, B. NE and CN, C. EL and CN, D. NE and NW, E. EL and NW, and F. CN and NW.
Figure S10 Phenotype of Aquilegia viridiflora complex. A. Corolla diameter, B petal length (not including spur), C. angle between petals and spurs, D. spur length, E. pistil length, F. Number of inflorescences. Except that the difference between the two groups marked in the figure is not significant, there is a significant difference between any other two groups.
Figure S11 RDA plot of A. viridiflora complex showing SNPs with are potentially associated with seven environmental factors.
Table S1 Sampling information and the value of seven environmental factors.
Table S2. Results of the HyDe.
Table S3. Polymorphisms in the chloroplast genome of A. viridiflora complex.
Table S4. Haplotype in the chloroplast genome of A. viridiflora complex.
Table S5. Results of the D-statistics.
Table S6. The AIC value of different models shown in Figure S7.
Table S7. Inferred parameters estimates with 95% confidence intervals for the best-fitting demographic scenario modelled in fastasimcoal2.
Table S8. The gene list under the selection.
Table S9. The gene list correlation the environment.