Figure 1.
Geographical distribution of the sampled 35 Chinese wolves in this study.
The average altitudes are also shown on the map.
Figure 2.
Depth of coverage of the nine Chinese wolf genomes.
The proportion of the genome exceeding a given minimum depth of coverage is shown. The numbers in legend are the mean genome wide coverage.
Figure 3.
Population structure from genome data (excluding highly linked SNPs with r2>0.2).
A: Structure assignments based on complete genome data from nine Chinese wolves. B: Principle component analysis of complete genome data from nine Chinese wolves. Xinjiang, XJ: XJ24, XJ30; Inner Mongolia, IM: IM06, IM07; Tibet, TI: TI09, TI32; Qinghai, QH: QH11, QH16. See table S1 and the text for abbreviations and localities.
Figure 4.
Pairwise sequential Markovian coalescent (PSMC) analysis of nine Chinese wolf genomes reflecting the genomic distribution of heterozygous sites.
Time scale on the x-axis is calculated assuming a mutation rate of 1×10−8 per generation and generation time equal to 3.
Figure 5.
Genomic regions with strong selective sweep signals in Tibet wolves.
The three hypoxia-related genes (EPAS1, RYR2, and ANGPT1) which each include at least one fixed non-synonymous SNP in highland wolves are highlighted. (A) Distribution of θπ ratios (θhighπ/θlowπ) and FST values calculated in 100-kb sliding windows in 20-kb steps. Data points in red (corresponding to the 5% empirical θπ ratio distribution, where θπ ratio is 5.311, and the 5% empirical FST distribution, where FST is 0.259) are regions under selection in highland wolves. (B) Genome-wide distribution of FST and Δπ along autosomes 1–38 (chromosomes are separated by color). Each dot represents 100 Kb genome regions. A dashed horizontal line indicates the top 5% level (FST>0.259; Pi>5.311) used for extracting outliers, where another dashed horizontal line FST>0.489 shows the top 1% level. (C) The genotypes observed in 9 full-genome data around EPAS1, RYR2, and ANGPT1. Every plot includes a gene region (top) and genotypes around the region (below). The y-axis denotes individuals: 1—RKWL [21], 2-IM06, 3-IM07, 4-XJ24, 5-XJ30, 6-QH11, 7-QH16, 8-TI09, and 9-TI32. See table S1 and the text for abbreviations and localities. The x-axis denotes the locations on genome.
Figure 6.
The number of nonsynonymous SNPs that might affect protein function based on SIFT [24], MAPP [25], and PolyPhen2 [26].
These SNPs genotypes differ at the 5% level between highland and lowland wolves.
Figure 7.
Top: Partial alignments of ANGPT1 (left) and RYR2 (right) amino acid sequences.
The sequences with more than 15% gaps or less than 60% identity with lowland ortholog sequences were filtered out. Dots (.) represent residues identical to lowland wolves' sequence. The mutation S214T in RYR2 and M201V in ANGPT1 in highland wolves were denoted with arrows. Where highland showed the wolves from Tibet and Qinghai and lowland showed the wolves from Xinjiang and Inner Mongolia. Bottom: Genotypes plots of three hypoxic genes (top) and LD patterns (triangle plot, bottom). These genotypes derive from Sanger sequencing of 35 wolves and were encoded as homozygous reference (ref), heterozygote (het), and homozygous alternative (alt). Red regions represent a high degree of LD. I: intron; N: non-synonymous; S: synonymous.
Figure 8.
Selection candidates involved in the HIF pathway that were found to show evidence of positive selection in highland wolves.
Solid lines indicate a direct relationship between enzymes and metabolites. Dashed lines indicate that more than one step is involved in the process. The genes outlined in black boxes were under selection and those indicated in gray boxes were provided for reference. See discussion of interactions in the text.