Ancient Mitogenomes Reveal the Maternal Genetic History of East Asian Gray Wolves (Canis lupus)

ABSTRACT

The gray wolf (Canis lupus) is the only wild ancestor of dogs (Canis lupus familiaris) and serves a crucial role in understanding the highly controversial issue of dog origins. Recently, ancient DNA studies on gray wolves from different regions of the Eurasian continent have achieved significant breakthroughs, providing important clues about the dog origins. As one of the potential origin areas for dogs, East Asia has seen some research on ancient dogs; however, reports related to gray wolves remain limited. In this study, we sequenced seven new mitogenomes of ancient gray wolves from Northern China, integrating them with 497 ancient and modern canid mitogenomes from published data. Our results reveal the following: (1) East Asian gray wolves have maintained high genetic diversity from ancient times to the present; (2) multiple haplogroup A gray wolves from Northern China support the hypothesis that Northeastern Eurasia is a core region for dog origins; (3) a deep gray wolf lineage in East Asia has been identified in this study; (4) different mitogenomes concentrated at the Jinchankou site indicate that admixture may have frequently occurred in the northeastern edge of the Tibetan Plateau. These findings enhance our understanding of the maternal genetic history of gray wolves in East Asia.

Recent ancient DNA studies have achieved significant breakthroughs, providing clues that indicate dogs may have originated from East Eurasian gray wolves. Seven ancient gray wolf mitogenomes from Northern China revealed (1) East Asian gray wolves exhibit high genetic diversity; (2) Northeastern Eurasia served as a core region for dog domestication; (3) a new named deep ancestral wolf lineage in East Asia; and (4) genetic admixture likely occurred frequently along the northeastern margin of the Tibetan Plateau.

1. Introduction

Gray wolves (Canis lupus) have existed since the Pleistocene, with a wide geographical distribution covering most of the Northern Hemisphere (Koblmüller et al. 2016; Puzachenko and Markova 2016; Diedrich 2022). Although fossil records have been continuous since the Middle Pleistocene (Jiangzuo 2021; Iurino et al. 2022; Diedrich 2022), and deep local ancestries did not come close to extinction (Bergström et al. 2022), the existing populations may have originated from expansion events during the Last Glacial Maximum (LGM) (Freedman et al. 2014; Loog et al. 2020; Bergström et al. 2022).

As the direct and only ancestor of dogs (Canis lupus familiaris), gray wolves are increasingly a focus for the research of dog origins. Based on the analysis of ancient gray wolf genomes spanning 100,000 years, it is proposed that dogs may have two distinct wolf ancestors, which were related to Eastern Eurasian and Western Eurasian wolves (Bergström et al. 2022). Wolves from Eastern Eurasia likely contributed to the genetic heritage of early dogs across Siberia, the Americas, East Asia, and Northeastern Europe (Bergström et al. 2022). Zhang et al. (2024) conducted mitogenome studies on ancient dogs in East Asia, supporting a Northeastern Eurasian origin of East Asian dogs. Modern genetic studies have revealed several gray wolf populations in China (Wang et al. 2019, 2020a), highlighting the importance of Eastern Eurasian gray wolves. Southern China is also regarded as a possible place of origin of dogs, as modern dogs from this region have the highest genetic diversity worldwide (Pang et al. 2009; Wang et al. 2016).

Ancient DNA (aDNA) research has played significant roles in uncovering human evolutionary history (Fu et al. 2014; Yang et al. 2020; Liu et al. 2022) and the origins of domestic animals (Bergström et al. 2020; Bergström et al. 2022; Zhang et al. 2022; Librado et al. 2024). Genetic studies on ancient gray wolves in East Asia, particularly in China, are crucial for understanding the origins of dogs. However, although only a limited amount of mitochondrial data has been published from China (Zhang et al. 2020, 2024; Wang et al. 2024), it confirms the need for more ancient data to improve our understanding of the genetic history of both gray wolves and dogs. In this study, we successfully obtained seven full mitogenomes of gray wolves from five archaeological sites in Northern China (Table 1). By combining these with published mitogenomes of ancient and modern gray wolves and dogs (Table S1), we aim to provide new evidence to enhance our understanding of the maternal genetic history of gray wolves.

TABLE 1.

Information of ancient samples in this study.

Library ID	Province	Site	Origin label	Tissue	Powder (mg)	Lib style	Haplogroup	Coverage	Contamination (%)		Age (BP)
									Mean	95% CI
L3278	Shaanxi	Quanhucun	H165:2	Calcaneus	78.6	DS.no UDG	M	51.33	0.04	0.01–0.9	5121 &
V0725	Heilongjiang	Harbin	V23966	Jaw	68.22	SS.full UDG	A	38.32	0.06	0.01–1.12	1–6768 *
V0729	Heilongjiang	Harbin	V26296	Jaw	74.47	SS.full UDG	X2	48.59	0.20	0.1–0.3	70 961–129 510 *
V8373	Qinghai	Jinchankou	A1368	Bone	29.6	SS.full UDG	X1	52.58	0.00	0.0–0.0	4000 #
V9533V9537	Qinghai	Jinchankou	A2043	Bone	36.5	SS.full UDG	E	124.83	0.00	0.0–0.0	4000 #
V9584	Gansu	Mazongshan	2017SMYT2605F100	Jaw	57.4	SS.full UDG	N	53.65	0.00	0.0–0.0	2200 #
V9597	Hebei	Xinglong	2018KXIIT1105F1②	Calcaneus	70	SS.full UDG	A	7.53	0.09	0.01–2.04	8000 #

^& The date based on the radiocarbon age results from Zhang et al. (2020).

^*The date based on the age estimated by tip dating method of BEAST v1.10.4 (Suchard et al. 2018) (Table S5; Figure S6).

^# The date based on the age estimated by the archaeological age information of the site (Dong et al. 2020; Lin et al. 2023; Zhang et al. 2024).

2. Materials and Methods

2.1. Sample Collection and Age Definition

Samples were obtained from five archaeological sites (Gansu: Mazongshan, n = 1; Hebei: Xinglong, n = 1; Heilongjiang: Harbin, n = 2; Qinghai: Jinchankou, n = 2; Shaanxi: Quanhucun, n = 1) (Table 1; Figure 1). All together, these samples are concentrated in Northern China, filling a gap of regions and time periods in wolves within East Asia. For these five sites, only the Harbin samples lack clear archaeological layer records, as they were collected from the bottom of Songhua River near Harbin City (approximately N45.92°N, 127.4°E) during underwater sand mining activities (Yang et al. 2019; Ni et al. 2020; Lu et al. 2021). Annual sand mining activities are estimated to uncover hundreds to thousands of bone and petrified wood fragments (Lu et al. 2021). Due to the non‐systematic way of sample collection, their precise stratigraphic layers remain undetermined (Yang et al. 2019; Lu et al. 2021). We performed Bayesian phylogenetic reconstruction with BEAST v1.10.4 (Suchard et al. 2018) to estimate the ages of the two Harbin samples (Table S5) (Zhang et al. 2024). For the remaining five samples, calibrated accelerator mass spectrometry (AMS) radiocarbon dates or archaeological age estimates were utilized (Table 1) (Dong et al. 2020; Zhang et al. 2020, 2024; Lin et al. 2023). The samples in this study were collected as part of a previous ancient dog project (Zhang et al. 2020, 2024). Due to the similar morphology between wolves and dogs, accurately identifying their highly damaged remains is challenging. We cannot definitively confirm that all samples derive exclusively from gray wolves due to occasional hybridization. In this study, specimens were classified based on maternal lineages.

FIGURE 1.

Geographic locations of the different archaeological sites in this study. The dots depict all five sites from East Asia, where the number of specimens is listed after the location. The samples are marked with colors ranging from red to green, indicating the range of their ages from older to more recent. Higher elevations are shown as darker colors. Base map was downloaded from Natural Earth (www.naturalearthdata.com).

For comparative analysis, a comprehensive mitogenome dataset was built, including ancient and modern canids worldwide. Full mitogenomes for some samples essential for our analysis were not available. To address this, we reconstructed the mitogenomes from the whole‐genome data, following the established protocol described by Hennelly et al. (2021).

2.2. DNA Extraction, Amplification, Sequencing, and Authenticity Criteria for aDNA

All procedures for aDNA experiments were carried out in the clean room at the Institute of Vertebrate Paleontology and Paleoanthropology, Chinese Academy of Sciences. Bone powder from each specimen, ranging from 29.6 to 78.6 mg, was prepared using a Dremel tool and single‐use drill bits. The aDNA laboratory work followed the same protocol described in Zhang et al. (2020). Double‐stranded or single‐stranded libraries with no uracil–DNA–glycosylase (UDG) or full UDG treatment were prepared from collected samples (Kircher et al. 2012; Meyer et al. 2012) (Table 1). For the in‐solution capture step, probes for 242 mammalian mtDNAs were utilized (Slon et al. 2016), resulting in the enrichment of endogenous aDNA fragments for sequencing. Data processing, reference genome alignment, and contamination evaluation followed the methods and parameters described in Zhang et al. (2024), and the aDNA damage patterns were detected using mapDamage v.2.0 (Jónsson et al. 2013) (Figures S1 and S2).

2.3. Haplogroup Classification

Haplogroup classification was initially performed using mitotoolpy‐seq.py (http://dometree.kiz.ac.cn/) as described by Peng et al. (2015), which defined most of the haplogroups. Novel haplogroups from newly sequenced individuals were defined in this study according to the phylogenetic trees (Figure 2; Tables S1 and S2; Figures S3–S5).

FIGURE 2.

A simplified Bayesian phylogenetic tree (see Figure S5). Posterior values are indicated by dots of different colors, ranging from green to red, representing posterior support values from 0 to 1, respectively. Red squares represent ancient gray wolves from China, with dotted outlines indicating samples collected from Wang et al. (2024) and solid outlines representing samples generated in this study. Circles of different colors represent various regional populations of domestic dogs and gray wolves. The clade of haplogroup A is enlarged for better viewing.

2.4. |Network Construction and Phylogenetic Trees

To investigate the genetic population structure of gray wolves, seven sequences generated from our study together with 186 published sequences were used to construct the median‐joining network using PopART v.1.7.1 (Leigh and Bryant 2015) (Figure 3A).

FIGURE 3.

(A) Median‐joining network of 193 samples, including gray wolves and dogs. Haplotypes are depicted as circles, with sizes proportional to the number of individuals. Several important lineages are highlighted with dotted lines. (B) Bar and line chart of the genetic diversity index before and after removing highland wolves. The positions of Hd1 and Pi1 are before removing highland wolves, Hd2 and Pi2 are after removing highland wolves.

To investigate phylogenetic relationships, a maximum likelihood tree that included all ancient and modern sequences (n = 504) was constructed using RAxML v8.2.12 (Stamatakis 2014), utilizing the GTRGAMMA + I model and a bootstrap value of 1000, with two coyotes (GenBank assembly accession: MZ042347 and MZ042357) as the outgroup (Figure S3). To reduce the computational burden, we selected 195 representative sequences (including two coyotes as outgroup) for the construction of both maximum likelihood and Bayesian phylogenetic trees (Figure 2; Figures S4 and S5). The GTR + G + I nucleotide substitution model was determined by jModeltest2 (Darriba et al. 2012), with clock rate set from 2 × 10⁻⁹ to 2 × 10⁻⁷ in uniform distribution. The Bayesian phylogenetic tree was constructed using BEAST v1.10.4 (Suchard et al. 2018), employing a strict molecular clock and a Bayesian skyline model with a piecewise‐linear tree structure, as outlined by Zhang et al. (2020). A total of 100 million iterations were performed, sampling every 10 000 iterations. All effective sample size (ESS) values were confirmed to exceed 300 using Tracer v1.6.0 (http://tree.bio.ed.ac.uk/software/tracer). After discarding the first 20% of the generated trees as burn‐in, the Bayesian phylogenetic tree was visualized using FigTree v1.4.4 (Figure 2; Figure S5).

2.5. Principal Component Analysis

A two‐dimensional (2D) plot from a principal component analysis (PCA) can visualize the relationships among multiple individuals more clearly than a phylogenetic tree with hundreds of sequences. Following the methodology described in Zhang et al. (2022), we converted the sequences into numeric values by replacing the DNA nucleotides of each sequence with a Boolean vector represented by 0s and 1s (Konishi et al. 2019) and conducted the PCA using this Boolean vector data. Finally, we utilized the ggplot2 package in R (v4.3.1) to create a PCA scatter plot based on the first principal component (PC1) and the second principal component (PC2) values (Figure 4; Figure S7) (Zhang et al. 2022).

FIGURE 4.

(A) Principal component analysis (PCA) based on 504 canids, including dogs and gray wolves. Lowland and highland wolves form distinct clusters, while the V0729_Harbin (assigned into haplogroup X2 in Figure 2) differs from both clusters. (B) PCA based on 488 canids, including gray wolves and dogs, after removing the highland wolves. The solid shapes represent ancient samples, while the hollow shapes represent modern samples, with solid red squares indicating ancient gray wolves from China.

2.6. Genetic Distance and Genetic Diversity Analysis

To investigate genetic diversity and population differentiation, populations were grouped geographically, and duplicate sequences within the same populations were removed. The genetic differentiation index (F _st) and corresponding p‐values between populations were calculated based on mitogenome sequences using Arlequin v3.5.2.2 (Excoffier and Lischer 2010), with 10,000 permutations performed to ensure the stability and reliability of the analysis results. Specific genetic distances and p‐values are presented in Table S3 (values less than 0 were modified to 0). Visualization of the results was performed using the heatmap.2 function from the gplots package in R v4.3.1 (Figure S8).

For the calculation of genetic diversity indices, all sites containing ambiguous or unknown bases were removed, and the indices were calculated using DnaSP v6.12.03 (Rozas et al. 2017). To further assess the impact of highland wolves on genetic diversity, we calculated the genetic diversity indices for the sequences after removing the highland wolves. The results were visualized as a stacked line graph using the ggplot2 package in R v4.3.1 (Figure 3B; Table S4).

3. Results

In this study, we successfully obtained six new and one re‐sequenced (Zhang et al. 2020, 2024) complete mitogenomes (depths of 7.5‐ to 124.8‐fold coverage, mean 53.8‐fold) (Figures S1 and S2). Five samples with radiocarbon dating age or archaeological age from approximately 8000 to 2200 years before present (BP) (Table 1; Figure 1). Bayesian analysis using BEAST v1.10.4 (Suchard et al. 2018) estimated that the V0725_Harbin sample dates to approximately 3 ka BP (95% HPD interval: 1–6768 BP), while V0729_Harbin likely represents a much older individual, dating to around 100 ka BP (95% HPD interval: 70 961–129 510 BP) (Table S5; Figures S5 and S6). Phylogenetic analysis indicates that these samples belong to gray wolves (Figure 2; Figures S3–S5). The estimated contamination rate for all samples was less than 5% (0%–0.2%, Table 1). We performed an extensive survey for comparative data and gathered a dataset of 497 ancient and modern canid mitogenomes, including gray wolves (123 ancient and 314 modern samples), and dogs (29 ancient and 31 modern samples) across Eurasia and the Americas (Table S1).

3.1. Ancient Chinese Wolves Were Assigned to Seven Haplogroups

We reconstructed phylogenetic trees using maximum likelihood and Bayesian methods. All phylogenetic trees are generally consistent and support the division of all gray wolves into highland and lowland wolves (constituting the majority of gray wolves). Ten ancient gray wolves from China (seven from this study and three from Wang et al. 2024) were assigned to seven haplogroups: A (Dongmingga, n = 1; Harbin, n = 1; Honghe, n = 1; and Xinglong, n = 1), D (Shuangta, n = 1), E (Jinchankou, n = 1), M (Quanhucun, n = 1), N (Mazongshan, n = 1), X1 (Jinchankou, n = 1), and X2 (Harbin, n = 1) (Figure 2; Tables S1 and S2). Multiple lineages within the same population have been found in other terrestrial ancient and modern wolves (Figure 2; Figures S3–S5). The median‐joining network also showed that ancient wolves from Northern China are distributed across seven lineages (Figure 3A).

Four individuals from Heilongjiang (Dongmingga, Harbin, and Honghe) and Hebei (Xinglong) are classified within haplogroup A. These individuals are within the basal lineages that cluster with the rest of those (mainly dogs) belonging to haplogroup A (Figure 2; Figures S3–S5). The individual from Jilin (Shuangta) is also at the basal part of the haplogroup D. One individual (V9533V9537_Jinchankou) from the Jinchankou site, Qinghai, belongs to haplogroup E, which is commonly found in extant individuals from Europe and Russia. Haplogroups M and N are commonly found in ancient/modern Eurasian wolves (Figures 2 and 3A; Figures S3–S5). Haplogroups X1 and X2 are newly named based on the phylogenetic trees established in this study (Figure 2). Haplogroup X1 consists of individuals (including V8373_Jinchankou, approximately 4000 BP) primarily from a high‐altitude area. Haplogroup X2 represents a previously unidentified deep maternal lineage (possibly extinct before LGM) located at the basal part of all other lowland wolves. This lineage only includes one ancient individual (V0729_Harbin), which diverged from other lowland wolves around 138 824 BP (95% HPD interval, 125 058–152 863 BP) (Figure 2; Figure S5).

3.2. Ancient Chinese Wolves With High Genetic Diversity

Individuals with relatively clear identification as wolves are divided into 11 groups based on their geographical origin: ancient Northern China (ANC, n = 8), ancient Siberia (ASIB, n = 36), ancient Central Eurasia (ACE, n = 5), ancient Europe (AEU, n = 48), ancient North America (ANA, n = 17), Japan (n = 15), modern East Asia (MEA, n = 50), modern Siberia (MSIB, n = 15), modern Central Eurasia (MCE, n = 28), modern Europe (MEU, n = 107), and modern North America (MNA, n = 88).

The heatmap results based on the genetic differentiation index (F _st) indicate that the gray wolves from ANC, ACE, and MEA clustered together, with MEA being the closest to ANC. Japanese wolves stand apart from other groups, indicating the genetic distinctiveness of the population (Table S3; Figure S8).

ANC and MEA have much higher nucleotide diversity among all investigated populations (Figure 3B; Table S4). However, this may be attributed to the presence of highland wolves in both ANC and MEA. After removing all highland wolves, we found that the Hd (haplotype diversity) and Pi/π (nucleotide diversity) values for these two populations significantly decreased; however, ANC remained the highest diverse among all groups (Figure 3B; Table S4). The PCA analysis of the lowland wolves showed that eight ancient wolves from China (five from this study and three from Wang et al. 2024) exhibited a significantly scattered distribution, also indicating a high genetic diversity of ancient wolves from the region (Figure 4; Figure S7).

3.3. Highland and Lowland Wolves Clearly Divided

The PCA analysis using all 504 mitogenomes visually illustrates the relationship between the newly sequenced individuals from Northern China and other wolves and dogs worldwide. The results showed high‐altitude wolves (highland wolves) to be distinctly separated from other low‐altitude wolves (lowland wolves). One individual (V0729_Harbin) did not cluster with either the highland or lowland wolves (Figure 4A), consistent with the newly designated haplogroup X2 in the phylogenetic trees (Figure 2; Figures S3–S5). As mentioned, the highland wolves are mainly grouped into haplogroup X1, this lineage split from the other lineages around 171 012 BP (95% HPD interval, 15 3379–18 7367 BP), and the most recent common ancestor (TMRCA) of haplogroup X1 is estimated to date back to 17 985 BP (95% HPD interval, 14 225–22 389 BP) (Figure 2; Figure S5).

3.4. Dogs Within Haplogroup A Associated With Northeastern Eurasian Wolves

Haplogroup A is the predominant haplogroup in both ancient and modern dogs (Zhang et al. 2020, 2024). In this study, we found four gray wolves (approximately 8000–4250 BP, one sample from Harbin unknown age) from Northern China belonging to haplogroup A (Figure 2; Table S2; Figures S3–S5). Network analysis shows that dogs appear in multiple branches, with ancient and modern wolves from Northeastern Eurasia positioned together with dogs within haplogroup A (Figure 3A). This indicates a possibly derived maternal relationship between Northeastern Eurasian wolves and dogs of haplogroup A, which was also supported by the phylogenetic trees (Figure 2; Figures S3–S5).

Phylogenetic trees indicate that several ancient Northeastern Eurasian wolves are closely related to ancient and modern dogs within haplogroup A (Figure 2; Figures S3–S5). Dogs of haplogroup A clustered with at least five relatively deep wolf lineages. All five lineages consist of individuals from Northeastern Eurasia. The basal lineage of haplogroup A (lineage ①, represented by three individuals from the Russian Far East, approximately 31 500 to 33 500 BP) separated from others approximately 38 098 BP (95% HPD interval, 36 287–40 062 BP). Lineage ② is represented by one individual from Northern China, approximately 5750 BP. Lineage ③ is represented by ancient individuals from Northern China and several modern wolves. Lineage ④ is represented by two ancient individuals found in Europe (Switzerland and Germany, approximately 14 500 to 15 500 BP) and with one individual from the Russian Far East (Belaya Gora, approximately 18000 BP). Lineage ⑤ is the closest lineage to dogs, which is represented by two wolves from the Russian Far East from 2400 BP and present (from Primorsky and Chukotka), and separated from others by approximately 23 655 BP (95% HPD interval, 20 595–26 882 BP) (Figure 2).

4. Discussion

The origin of dogs from gray wolves is widely accepted; however, their geographical origin remains controversial. Genetic research on gray wolves can provide supplementary evidence that is crucial for understanding their population history and can offer another perspective on the domestication origins of dogs. These mitogenomes of ancient gray wolves across Northern China contribute to the understanding of the genetic history of gray wolves in East Asia.

4.1. Ancient Wolves From Northern China Exhibit High Genetic Diversity

Gray wolves display a complex maternal population genetic history (Koblmüller et al. 2016; Loog et al. 2020; Bergström et al. 2022). Our study successfully obtained seven mitogenomes and combined them with three published mitogenomes, resulting in a total of 10 from Northern China (Wang et al. 2024). We found that they are within seven lineages, which showed very high maternal genetic diversity (Figures 2 and 3A; Tables S1 and S2). This is similar to ancient gray wolves found in Europe and Siberia (AEU and ASIB), which exist widely in many lineages (Thalmann et al. 2013; Loog et al. 2020) (Figure 2; Figures S3–S5).

Although most gray wolves across Eurasia exhibit a high level of maternal genetic diversity, ancient gray wolves from Northern China have the highest (Figure 3B; Table S4), revealing broad maternal origins. One individual in this study is from the highland wolf lineage (V8373_Jinchankou_4000), and one is from a previously undefined deep lineage (V0729_Harbin). The presence of these two special individuals is the main reason for the higher maternal genetic diversity of ancient Chinese gray wolves (n = 10) compared to other populations (Figures 2 and 3A; Figures S3–S5).

We also found that individuals of diverse lineages (such as A, N, and E) appear across various regions (Figures 2 and 3A; Figures S3–S5), which may be related to the ability of gray wolves to migrate extensively, possibly facilitated by back‐crossing with dogs (Aspi et al. 2009; Thalmann et al. 2013; Jarausch et al. 2023). This could lead to frequent gene flow between different regions (Pilot et al. 2014). Modern Chinese wolves also exhibit high genetic diversity (Zhang et al. 2015); however, this is mainly due to the existence of highland wolves. After removing highland wolves, the genetic diversity of modern Chinese wolves is significantly lower than that of ancient Chinese wolves (Figure 3B; Table S4).

4.2. Dogs Possibly Originating From Northeastern Eurasian Wolves

Currently, no specific gray wolf population worldwide has been found that is closer to dogs than others (Freedman et al. 2014; Freedman and Wayne 2017; Wang et al. 2019). Genomic studies have shown that modern dogs and Eurasian gray wolves are monophyletic (Fan et al. 2016). The maternal haplogroups of dogs are widely distributed in six major branches among wolves (haplogroups A–F, with dogs mainly in haplogroup A) (Figure 2; Figures S3–S5) (Thalmann et al. 2013; Peng et al. 2015; Doan et al. 2023; Zhang et al. 2020, 2024). The mixing of maternal haplogroups among wolves and dogs may result from gene flow between dogs and gray wolves after dog domestication. Recentlhe oldest dog from East Asia (Harbin, approximately 12 900 BP) was identified as belonging to haplogroup A, which is close to ancient dogs in America (Zhang et al. 2024). The common ancestor of all dogs within haplogroup A can be traced back to the end of the LGM, approximately 20 000 BP (95% HPD interval, 21 900–17 800 BP) and expanded significantly (Zhang et al. 2024). We consider that this population expansion event laid the foundation for haplogroup A to be the most prominent mitochondrial haplogroup in both ancient and modern dogs.

Dogs of haplogroup A originated from ancestral gray wolves within haplogroup A. Based on the results of phylogenetic trees, we can clearly see that at least five relatively deep wolf lineages, all including wolves from Northeastern Eurasia, are closely related to ancient and modern dogs within haplogroup A (Figure 2; Figures S3–S5). Our study found that several ancient wolves from Northern China (lineage ②: Dongmingga; lineage ③: Xinglong, Honghe, and Harbin) and the Russian Far East (lineage ④: Belaya Gora; lineage ⑤: Chukotka and Primorsky) are located at the basal part of haplogroup A (Figure 2; Figures S3–S5). The results of these phylogenetic trees suggest that the ancient wolves from Northern China may be associated with populations related to the progenitor of dogs. These results from ancient wolves support the hypothesis that the progenitors of dogs first lived in a vast area of Northeastern Eurasia, consistent with the findings of Bergström et al. (2022) and Zhang et al. (2024).

4.3. Highland and Lowland Wolves Lived Together in the Northeastern Edge of the Tibetan Plateau

The wolves on the Tibetan Plateau (including Tibetan wolves and Himalayan wolves) possess approximately 39% of their nuclear DNA from an unidentified ancient wolf‐like lineage that is significantly diverged from the genomes of contemporary Holarctic wolves and dogs (Wang et al. 2020a). This may be due to the admixture events between early migrations of gray wolves onto the Tibetan Plateau and preceding canids that had already adapted to the highland environment. The Tibetan wolf population acquired a significant genetic proportion of deep ancestry, including the endothelial PAS domain protein 1 gene (EPAS1) (Wang et al. 2020a), enabling them to rapidly adapt to high‐altitude environments.

Our study found that the mitogenome of modern wolves on the Tibetan Plateau forms a deep lineage (X1) with significant genetic differentiation from other gray wolf populations, similar to previous studies (Loog et al. 2020; Doan et al. 2023). The Jinchankou site, Qinghai (both plant and animal samples have been radiocarbon dated to 3.7–4.1 ka BP) is exclusively associated with the Qijia culture and is located at the northeastern edge of the Tibetan Plateau (approximately 2400 m above sea level) (Dong et al. 2020; Wang et al. 2020b; Zhang et al. 2020). This site exhibits high diversity in animal resources, predominantly wild animals (mainly deer), whereas domestic animals (sheep and dogs) also constitute a considerable proportion. In this study, we found both highland and lowland wolf haplogroups (Figures 2 and 4A; Figures S3–S5), indicating that two significantly different maternal haplogroups may have coexisted around this site, with ancient highland wolves being reported for the first time (Figures 2 and 4A; Figures S3–S5). The lowland wolf from this site belongs to haplogroup E, which has been found almost exclusively in European gray wolves and a small number of East Asian dogs (e.g., in Korea) (Figures 2 and 3A; Figures S3–S5), indicating a close relationship with European gray wolves. Three individuals belonging to haplogroup A1b3 (mainly found in ancient East Asian dogs) and two individuals belonging to haplogroup C (mainly found in Western Eurasian dogs) were also found at this site (Zhang et al. 2020, 2024). The above discovery of gray wolves and dogs suggests that the northeastern edge of the Tibetan Plateau may be a key area for the admixture of highland and lowland gray wolves and dogs. This is consistent with previous studies suggesting that the admixture may occur in the marginal areas of the Tibetan Plateau (Werhahn et al. 2020; Hennelly et al. 2021).

5. Conclusion

Our study provides important data for understanding the maternal genetic history of gray wolves in East Asia and the origins of dogs. We found that East Asian gray wolves have maintained high maternal genetic diversity from ancient times to the present, primarily due to mitogenomes associated with the Tibetan Plateau. We identified several haplogroup A individuals in the gray wolves from Northern China that is shared with the majority of dogs worldwide, further supporting the hypothesis that Northeastern Eurasia is a core region for the origins of dogs (Bergström et al. 2022; Zhang et al. 2024). Additionally, the presence of different mitogenomes of gray wolves and dogs concentrated at the Jinchankou site suggests the northeastern edge of the Tibetan Plateau to be one of the key contact areas where admixture events between gray wolves and dogs may have occurred. Future research should focus on this region, with a more representative sample size of ancient gray wolves and dogs. Genome‐wide data will also be needed to provide detailed population and admixture history during the transition from gray wolves to dogs.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1 Fragmentation and misincorporation patterns of seven ancient wolves by mapDamage2.0.

Figure S2 Sequence length histograms of seven ancient wolves.

Figure S3 A maximum likelihood tree constructed using RAxML with 506 sequences. Samples from this study are highlighted in red, with statistical support values evaluated using 1000 bootstrap steps, indicated at each node.

Figure S4 A maximum likelihood tree constructed using RAxML with 195 sequences. Samples from this study are highlighted in red, with statistical support values evaluated using 1000 bootstrap steps, indicated at each node.

Figure S5 A Bayesian phylogenetic tree constructed using 195 sequences. Posterior values are indicated by dots of different colors, ranging from green to red, which represent posterior support values from 0 to 1, respectively. Samples marked in red were generated in this study.

Figure S6 Bayesian tipdating analysis for individuals from Harbin, Heilongjiang. (A) Age estimate value based on the mitochondrial Bayesian reconstruction. (B) Marginal densities for age estimates based on the mitochondrial Bayesian reconstruction.

Figure S7 PCA based on 488 canids, including dogs and gray wolves, after removing the highland wolves and dogs. The solid shapes represent ancient samples, and the hollow shapes represent modern samples, with solid red squares indicating ancient gray wolves from China.

Figure S8 F _ST heatmap constructed using 215 wolves from different regions.

Table S1 Selected individuals from published studies.

Table S2 Haplogroup classification of Chinese wolves used in this study.

Table S3 Genetic differentiation index (F _ST) and corresponding p‐value results.

Table S4 Genetic diversity of different populations.

Table S5 Bayesian tip dating values by BEAST v1.10.4 (Suchard et al. 2018).

Zhang, M. , Wang C., Zheng Y., et al. 2026. “Ancient Mitogenomes Reveal the Maternal Genetic History of East Asian Gray Wolves (Canis lupus).” Integrative Zoology 21, no. 5: 1367–1377. 10.1111/1749-4877.13005

Data Availability Statement

The data reported in this study were deposited in GenBase of the National Genomics Data Center, Beijing Institute of Genomics, Chinese Academy of Sciences/China National Center for Bioinformation under accession numbers C_AA106955.1–C_AA106961.1. All software are cited in the Materials and Methods are publicly available.