Article

Genomic Analyses Reveal the Influence of

Geographic Origin, Migration, and Hybridization on
Modern Dog Breed Development
Graphical Abstract Authors
Heidi G. Parker, Dayna L. Dreger, Maud
Rimbault, Brian W. Davis, Alexandra B.
Mullen, Gretchen Carpintero-Ramirez,
Elaine A. Ostrander

Correspondence
[email protected]

In Brief
The domestic dog is divided into
hundreds of island-like populations
called breeds. Parker et al. examine 161
breeds and show that they were
developed through division and
admixture. The analyses define clades,
estimate admixture dates, distinguish
geographically diverse populations, and
help determine the source of shared
mutations among diverse populations.

Highlights Accession Numbers

d Neighbor joining cladogram of 161 breeds establishes 23 GSE90441
supported clades GSE83160
GSE70454
d Crossing between diverse clades was done recently to add GSE96736
new traits

d Migration of a breed to a new region alters both immigrant

and indigenous breeds

d Tracking recent crosses can identify the source of mutations

in multiple breeds

Parker et al., 2017, Cell Reports 19, 697708

April 25, 2017 2017 The Author(s).
http://dx.doi.org/10.1016/j.celrep.2017.03.079
Cell Reports

Article

Genomic Analyses Reveal the Influence

of Geographic Origin, Migration, and
Hybridization on Modern Dog Breed Development
Heidi G. Parker,1 Dayna L. Dreger,1 Maud Rimbault,1 Brian W. Davis,1 Alexandra B. Mullen,1
Gretchen Carpintero-Ramirez,1 and Elaine A. Ostrander1,2,*
1Cancer Genetics and Comparative Genomics Branch, National Human Genome Research Institute, National Institutes of Health,

Bethesda, MD 20892, USA

2Lead Contact

*Correspondence: [email protected]
http://dx.doi.org/10.1016/j.celrep.2017.03.079

SUMMARY Previous studies have addressed the genomic makeup of a

limited number of breeds, demonstrating that dogs from the
There are nearly 400 modern domestic dog breeds same breed share common alleles and can be grouped using
with a unique histories and genetic profiles. To track measures of population structure (Irion et al., 2003; Koskinen,
the genetic signatures of breed development, we 2003; Parker et al., 2004), and breeds that possess similar
have assembled the most diverse dataset of dog form and function often share similar allelic patterns (Parker
breeds, reflecting their extensive phenotypic varia- et al., 2004, 2007; Vonholdt et al., 2010). However, none of these
studies have effectively accounted for the variety of mechanisms
tion and heritage. Combining genetic distance,
through which modern breeds may have developed, such as
migration, and genome-wide haplotype sharing ana-
geographic separation and immigration; the role of hybridization
lyses, we uncover geographic patterns of develop- in the history of the breeds; and the timeline of the formation of
ment and independent origins of common traits. breeds. In this study, we overcome these barriers by presenting
Our analyses reveal the hybrid history of breeds an expansive dataset, including pure breeds sampled from mul-
and elucidate the effects of immigration, revealing tiple sections of the globe and genotyped on a dense scale. By
for the first time a suggestion of New World dog applying both phylogenetic methods and a genome-wide anal-
within some modern breeds. Finally, we used cladis- ysis of recent haplotype sharing, we have unraveled common
tics and haplotype sharing to show that some com- population confounders for many breeds, leading us to propose
mon traits have arisen more than once in the history a two-step process of breed creation beginning with ancient
of the dog. These analyses characterize the com- separation by functional employment followed by recent selec-
tion for physical attributes. These data and analyses provide a
plexities of breed development, resolving longstand-
basis for understanding which and why numerous, sometimes
ing questions regarding individual breed origination,
deleterious mutations are shared across seemingly unrelated
the effect of migration on geographically distinct breeds.
breeds, and, by inference, transfer of trait and dis-
ease alleles among dog breeds.
RESULTS

INTRODUCTION We examined genomic data from the largest and most diverse
group of breeds studied to date, amassing a dataset of 1,346
The dog, Canis familiaris, is the first domesticate earning a place dogs representing 161 breeds. Included are populations with
within nearly every society across the globe for thousands of vastly different breed histories, originating from all continents
years (Druzhkova et al., 2013; Thalmann et al., 2013; Vila` et al., except Antarctica, and sampled from North America, Europe,
1997, 1999). Over the millennia, dogs have assisted humans Africa, and Asia. We have specifically included breeds that
with hunting and livestock management, guarding house and represent the full range of phenotypic variation present among
field, and played crucial roles in major wars (Moody et al., modern dogs, as well as three breeds sampled from both the
2006). Providing a range of services from companionship to pro- United States and their country of origin. Samples from 938
duction of fur and meat (Wilcox and Walkowicz, 1995), the diver- dogs representing 127 breeds and nine wild canids were geno-
sity of talents and phenotypes combined with an unequalled typed using the Illumina CanineHD bead array following standard
emotional connection between dogs and humans has led to protocols. Data were combined with publically available informa-
the creation of more than 350 distinct breeds, each of which is tion from 405 dogs genotyped using the same chip (Hayward
a closed breeding population that reflects a collage of defining et al., 2016; Vaysse et al., 2011). For three dogs from one breed,
traits (http://www.akc.org). genotypes were retrieved from publically available sequence

Cell Reports 19, 697708, April 25, 2017 2017 The Author(s). 697
This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
 Ameri

ter Dog
ca

Toy and Miniature

Standard Poodle
n Hair
Chine

Portuguese Wa
Toy Fox rrier

Pumi du Tulear

riffon
Bichon Frise
Havanese
Rat Terri
less Te d
se Cre

Pap ssels G
Chih

Maltese

Poodle

og
Terrier

illon
Pu g D
Coto
uahu
ste

Puli
er

Bru
a
^

Gre
*
a
Pha t Pyre *
Cirn raoh Honees u
eco und a In ute
Ibiza delEtna Shib n Malam
nH ka D og
Komoound Alas ledge
ndor re e n land S
Mastino Kuvasz G H us ky
n
Abru zzese Siberia )
Levriero Me Sloug astiff (Am
hi #
ridionale Tibetan M
Azawakh
Anatolian Shephe Tibetan Mastiff (Ch)
rd @ Xigou
Afghan Hound Basenji
@
Saluki Boxer
Dachshund Bulldog
een Dogue de Bo
Griffon Vend rdeaux
Petit Basset und Boston Te
Basset Hoound French
rrier
C oonh Bulldo
Redbone Be a g
le Bull T g
errier
ound Minia
Fox H und ture
rH o Bull
Otte +
Terr
ier
er
er

Bea tralian C eepdog

Aus
des Fla ard
Sett

Old E
ndres

Greyhou
is
Sett

Whippet
ren and

Scottish Deerho
Borzoi
Sheepdog
Italian Greyhound

Irish Wolfhound

= 90-100
Malino
ri

rded
B
lish
don

nglis
Eng

Coll ttle Dog

= 70-89
Go r

Belgian Tervu

nd
Belgian

h Sh
a
ie

= 50-69
Bouvier

und

Figure 1. Cladogram of 161 Domestic Dog Breeds

Breeds that form unique clades supported by 100% of bootstraps are combined into triangles. For all other branches, a gold star indicates 90% or better, black
star 70% 89%, and silver star 50% 69% bootstrap support. Breeds are listed on the perimeter of the circle. A small number of dogs do not cluster with the rest
of their breed, indicated as follows: *cane paratore, +Peruvian hairless dog, #sloughi, @country-of-origin salukis, and ^miniature xoloitzcuintle.

files, and all were merged into a single dataset (Table S1). After and to enhance these analyses with unbiased haplotype sharing
pruning for low quality or genotyping rate, 150,067 informative for a robust assessment of canine population structure.
SNPs were retained. A bootstrapped cladogram was obtained using an identity-by-
Ascertainment bias has been shown to skew population ge- state distance matrix and a neighbor-joining tree algorithm (Sup-
netic calculations that require estimation of allele frequencies plemental Experimental Procedures). After 100 bootstraps, 91%
and diversity measures (Lachance and Tishkoff, 2013). It has of breeds (146/161) formed single, breed-specific nodes with
also been shown that ascertainment based on a single individual 100% bootstrap support (Figure 1). Of the 15 breeds that did
provides less bias than a mixed group (Patterson et al., 2012). not meet these criteria, seven (Belgian tervuren, Belgian
The SNPs used in this study were identified primarily within the sheepdog, cane corso, bull terrier, miniature bull terrier, rat ter-
boxer or from boxer compared to another genome (Vaysse rier, and American hairless terrier) were part of two- or three-
et al., 2011), which has exaggerated the boxer minor allele fre- breed clades that were supported at 98% or greater, and two
quency (MAF; 0.351 in boxer compared to 0.260 overall) but breeds (Lhasa apso and saluki) formed single-breed clades
has little affect the other breeds (MAF range, 0.2470.284). To that were supported at 50% and 78%, respectively. Four breeds
minimize the effect this might have, we have chosen to use dis- (redbone coonhound, sloughi, cane paratore, and Jack Russell
tance measures based on allele sharing rather than frequency terrier) were split within single multi-breed clades, and the last

698 Cell Reports 19, 697708, April 25, 2017

 Figure 2. Representatives from Each of the
23 Clades of Breeds
Breeds and clades are listed for each picture from
left to right, top to bottom.
A B C D E F (A) Akita/Asian spitz.
(B) Shih tzu/Asian toy (by Mary Bloom).
(C) Icelandic sheepdog/Nordic spitz (by Veronica
Druk).
(D) Miniature schnauzer/schnauzer.
G H I J K L (E) Pomeranian/small spitz.
(F) Brussels griffon/toy spitz (by Mary Bloom).
(G) Puli/Hungarian.
(H) Standard poodle/poodle.
(I) Chihuahua/American toy.
(J) Rat terrier/American terrier (by Stacy Zimmer-
M N O P Q man).
(K) Miniature pinscher/pinscher.
(L) Irish terrier/terrier.
(M) German shepherd dog/New World (by Mary
Bloom).
(N) Saluki/Mediterranean (by Mary Bloom).
R S T U V W (O) Basset hound/scent hound (by Mary Bloom).
(P) American cocker spaniel/spaniel (by Mary
Bloom).
(Q) Golden retriever/retriever (by Mary Bloom).
(R) German shorthaired pointer/pointer setter (by Mary Bloom).
(S) Briard/continental herder (by Mary Bloom).
(T) Shetland sheepdog/UK rural.
(U) Rottweiler/drover
(V) Saint Bernard/alpine.
(W) English mastiff/European mastiff (by Mary Bloom).

two breeds (xoloitzcuintli and Peruvian hairless dog) were split To assess hybridization across the clades, identical-by-
between divergent clades. Nine of the breeds that were not descent (IBD) haplotype sharing was calculated between all
monophyletic were either newly recognized by the American pairs of dogs from the 161 breeds. Haplotypes were phased
Kennel Club (AKC) or not recognized at the time of sample using the program Beagle (Browning and Browning, 2013) in
collection and likely represent a breed under development. 100-SNP windows, resulting in a minimum haplotype size of
Two other non-monophyletic breeds are composed of dogs 232 kb, well above the shared background level established in
collected in two countries; the cane corsos collected in Italy previous studies (Lindblad-Toh et al., 2005; Sutter et al., 2004).
form a fully supported, single clade, as do the salukis collected The large haplotypes specifically target admixture resulting
in the United States. However, the cane corsos collected in the from breed formation rather than domestication, which previous
United States form a paraphyletic clade near the Neapolitan studies have not addressed. The total length of the shared hap-
mastiffs, and the salukis collected in the Middle East form lotypes was summed for each pair of dogs. Individuals from
multiple paraphyletic groups within a clade that includes the within the same breed clade share nearly four times more of their
US salukis and Afghan hounds. genome within large IBD haplotype blocks than dogs in different
Not including those that are breed specific, this study defined breed clades (median shared haplotype lengths of 9,742,000 bp
105 phylogenetic nodes supported by R90% of bootstrap and 2,184,000 bp, respectively; p [Kolmogorov-Smirnov (K-S)
replicates, 133 by R70%, and 150 supported by R50% of and Wilcox] < 2.2e 16; Figure 3A). Only 5% of the across-breed
replicates. We identify 29 multi-breed clades that are supported pairings have a median greater than 9,744,974 bp. These excep-
at R90%. Each of these clades includes 216 breeds and tions argue for recent admixture events between breeds, as evi-
together account for 78% of breeds in the dataset. 150 breeds, denced by the example of the Eurasier breed, created in the
or 93% of the dataset, can be divided into 23 clades of 218 1970s by mixing chow chow with other spitz-type breeds (Fogle,
breeds each, supported at >50%. These multi-breed clades 2000) (Figure 3B). The data reveal not only the components of the
reflect common behaviors, physical appearance, and/or related breed but also the explanation for its placement on the clado-
geographic origin (Figure 2). gram. The Eurasier (unclustered) shows significant haplotype
Eleven breeds did not group with significance to any other sharing with the samoyed (unclustered), keeshond (Nordic spitz),
breeds. Five breeds form independent clades and six others and chow chow (Asian spitz) (Figure 3B). Because all three
are paraphyletic to established clades with <50% bootstrap sup- breeds are located in different clades, unrelated to each other,
port (Table S2). The lack of grouping may indicate that we have the Eurasier falls between the component breeds and forms its
not sampled the closest relatives of these breeds or that these own single-breed clade. Haplotype-sharing bar graphs for
breeds comprise outcrossings that are not shared by similar each of 161 breeds, including 152 AKC breeds, are available in
breeds. Data S1. This provides a long-term resource for identifying

Cell Reports 19, 697708, April 25, 2017 699

 A B

Figure 3. Gross Haplotype Sharing across Breeds

(A) Boxplot of total haplotype sharing between all pairs of dogs from breeds within the same clade, across different clades, and within the same breed. The
difference between the distributions is highly significant (p < 2e-16).
(B) Example of haplotype sharing between three breeds (samoyed, chow chow, and keeshond) and a fourth (Eurasier) that was created as a composite of the
other three. Combined haplotype length is displayed on the y axis, and 169 breeds and populations are listed on the x axis in the order they appear on the
cladogram, starting with the jackal and continuing counterclockwise. Haplotype sharing of zero is set at 250,000 for graphing, a value just below what is detected
in this analysis. Breeds are colored by clade. 95% significance level is indicated by the horizontal line. Breed abbreviations are listed under the graph in the order
they appear and colored by clade. Definitions of the breed abbreviations can be found in Table S1.

populations that likely share rare and common traits that will be groups, suggesting recent creation of these breeds from multiple
invaluable for mapping the origins of deleterious and beneficial others or that they provide a popular modern breed component.
mutations. The overall low level of sharing across diverse breeds suggests
Strong evidence of admixture across the clades was found in that interclade crosses are done thoughtfully and for specific
117 breeds (Figure 4). A small number of these were identified in reasons, such as the introduction of a new trait or the immigra-
previous studies using migration analysis (Pickrell and Pritchard, tion of a breed to a new geographic region.
2012; Shannon et al., 2015) 30% of these breeds share with only As importation and establishment in a new country has been
one breed outside their clade. Therefore, more than half (54%) of shown to have a measurable effect on breed structure (Quignon
the breeds that make up the 23 established clades share large et al., 2007), we chose three breeds, the Tibetan mastiff, saluki,
haplotypes with one or zero breeds outside their clade, indi- and cane corso, for inclusion in the study, with each collected in
cating breed creation by selection based on the initial founder the country of origin as well as from established populations in
population rather than recent admixture. Only 6 of the 161 the United States. In each case, there is division of the breed
breeds share extensive haplotypes with many (>8) different based on collection location. The split between the US and

700 Cell Reports 19, 697708, April 25, 2017

Figure 4. Haplotype Sharing between Breeds from Different Phylogenetic Clades
The circos plot is ordered and colored to match the tree in Figure 1. Ribbons connecting breeds indicate a median haplotype sharing between all dogs of each
breed in excess of 95% of all haplotype sharing across clades. Definitions of the breed abbreviations can be found in Table S1.

Chinese Tibetan mastiffs is likely due to independent lineage for- not evident in the Italian cane corsos, as well as increased shared
mation stemming from an importation bottleneck, as is evident haplotypes with the other mastiffs. Cane corsos have been in the
from estimations of inbreeding coefficients (Chinese Tibetan United States for less than 30 years (American Kennel Club,
mastiffs average F = 0.07, and US Tibetan mastiffs average 1998).
F = 0.15). Similarly, the average inbreeding coefficient of salukis Our analyses were designed to detect recent admixture;
collected in the United States is twice as high as those sampled therefore, we were able to identify hybridization events that are
from the countries of origin (F = 0.21 and 0.10, respectively). described in written breed histories and stud-book records.
Since the US salukis form a more strongly bootstrapped clade Using the most reliably dated crosses that produced modern
than the country-of-origin dogs, we suggest that there is a less breeds, we established a linear relationship between the total
diverse gene pool in the United States. In comparison, the length of haplotype sharing and the age of an admixture event,
cane corsos from Italy form a single clade, while the cane corsos occurring between 35 and 160 years before present (ybp) (Fig-
from the United States cluster with the Neapolitan mastiffs, also ure 5A). Applying this equation to the total shared haplotypes
collected in the United States. Significant shared haplotypes are calculated from the genotyping data, we have validated this rela-
observed between the US cane corsos and the rottweiler that are tionship on a second set of recently created breeds arriving at

Cell Reports 19, 697708, April 25, 2017 701

A Figure 5. Total Haplotype Sharing Is Inversely
Correlated with the Time of Hybridization
between Breeds that Have Developed within
the Last 200 Years
(A) The time of hybridization in years before present
is graphed on the x axis and the median total
haplotype sharing on the y axis for six breeds of dog
with reliable recent histories of admixture in breed
formation or recovery. The trendline shows a linear
correlation with r2 = 1.
(B) The slope and intercept of the trendline from A
was applied to the median haplotype sharing values
from the data for four additional breeds with reliable
breed creation dates to establish accuracy of esti-
mated hybridization dates.

likelihood score over a zero migration tree

(Figure 6A). Only 2 of the 18 clades, New
World and Asian toy (Figures 6B and 6C),
showed evidence of extensive hybridiza-
tion between the breeds. Thus, the mod-
B ern breeds were likely created through se-
Breed 1 Breed 2 Total sharing Esmated Yrs Ago Predicted Year Historical year
lection for unique traits within an ancient
breed type with possible admixture from
unrelated breeds to enhance the trait.
Our hybridization analysis reveals evi-
dence for disease sharing across the
clades. For instance, collie eye anomaly
(CEA) is a disease that affects the develop-
ment of the choroid in several herding
breeds, including the collie, Border collie,
Shetland sheepdog, and Australian shep-
herd, all members of the UK Rural clade
(Lowe et al., 2003). The mutation and
haplotype pattern displayed IBD across
all affected breeds, and we speculated
that all share a common obviously
affected ancestor (Parker et al., 2007).
historically accurate time estimations (Figure 5B). Using the rela- We were unable to explain, however, the presence of the disease
tionship equation y = 1,613,084.67x + 262,137,843.89, where y in the Nova Scotia duck tolling retriever, a sporting dog devel-
is the total shared haplotype length and x is the number of years, oped in Canada from an unknown mixture of local breeds, which
we can estimate the time at which undocumented crosses or di- also shares the same haplotype. This perplexing observation can
visions from older breeds took place. For example, based on a now be explained, as this analysis shows that collie and/or Shet-
median haplotype sharing value of 66,993,738, the golden land sheepdog were strong, undocumented, contributors to the
retriever was separated from the flat-coated retriever in 1895, formation of the Nova Scotia duck tolling retriever and, therefore,
and the written history of the golden retriever dates to crosses the likely source of the CEA mutation within that breed
between multiple breeds taking place between 1868 and 1890 (Figure 7A).
(Figure 5B), a near-perfect match. Similarly, a mutation in the MDR1 gene (multi-drug resistance 1),
To determine if the multi-breed clades are formed through which causes life-threatening reactions to multiple drugs in many
recent admixture rather than through common ancestral sour- of the UK Rural breeds, has been reported in 10% of German
ces, we examined migration in 18 groups of four or more breeds. shepherd dogs (Mealey and Meurs, 2008). These data display a
These include 16 of the clades established on the tree, including link between the German shepherd dog and UK Rural breeds
nearby unclustered breeds, and two groups of small clades through the Australian shepherd, highlighting the unexpected
(American terrier/American toy and small spitz/toy spitz/schnau- role the Australian shepherd or its predecessor played in the devel-
zer) that are monophyletic, but not well supported. Using the pro- opment of the modern German shepherd dog (Figure 6B). Earlier
gram Treemix (Pickrell and Pritchard, 2012), and allowing 012 this year, the MDR1 mutation was identified in the chinook at a fre-
predicted migration events, we determined the effect of admix- quency of 15% (Donner et al., 2016). Our analysis reveals recent
ture on clade formation by calculating the increase in maximum admixture between this breed and the German shepherd dog as

702 Cell Reports 19, 697708, April 25, 2017

Figure 6. Assessment of Migration between Breeds within Clades
Admixture was measured in Treemix for 18 groups of breeds representing clades or combinations of small clades.
(A) Improvement to the maximum likelihood tree of each group as the result of admixture. The y axis shows fold improvement over the zero admixture tree.
(B) Cladogram of the New World breeds with European herders allowing four migration events. Arrows show estimated migration between breeds colored by
weight (yellow to red = 00.5).
(C) Cladogram showing migration within the Asian toy clade, including a neighboring breed, the Tibetan terrier. Pictures by Terri Gueck (TIBT), Yuri Hooker (INCA),
Mary Bloom (GSD and SHIH), Maurizio Marziali (CPAT), Mary Malkiel (COOK), and John and Debbie Caponetto (large and small XOLO/MXOL).

Cell Reports 19, 697708, April 25, 2017 703

 Figure 7. Haplotypes Shared with Breeds that Carry
Known Deleterious Mutations
Breeds are connected if the median shared haplotype size
exceeds the 95% threshold for interclade sharing. Sharing
between breeds that are known to carry the mutation is
colored black, and sharing with other breeds is colored
according to the breed that carries the mutation.
(A) Collie eye anomaly is found in a number of herding
breeds developed in the United Kingdom and some
sporting breeds developed in the United States.
(B) Multi-drug resistance 1 mutation is carried by many UK
herding breeds as well as the German shepherd.

704 Cell Reports 19, 697708, April 25, 2017

well as previously unknown addition of Collie, both carriers of Based on these estimates, the excess haplotype sharing that
the MDR1 mutation. Haplotype sharing with the same affected we have identified represents the creation of breeds since the
breeds is found in the xoloitzcuintli, which allows us to predict Victorian era breed explosion. Most breeds within each clade
that this rare breed may also carry the deleterious allele but has share haplotypes at this level (<200 ybp); however, the lack of
yet to be tested. sharing across the clades, outside of very specific crosses, sug-
gests the clades were developed much earlier than the breed
DISCUSSION registries. Dividing the data by clade, the median haplotype
sharing is lowest in the Asian spitz (median = 0) and the Mediter-
Phylogenetic analyses have often been applied to determine the ranean clades (median = 516,900) (median range across all
relationships between dog breeds with the understanding that a clades = 03,459,000), indicating that these clades are most
tree structure cannot fully explain the development of breeds. divergent and possibly older than the rest. This fits well with pre-
Prior studies have shown that single mutations produce recog- vious studies that suggest the earliest dogs came from Central
nizable traits that are shared across breeds from diverse clades, and East Asia (Pang et al., 2009; Shannon et al., 2015). Interest-
suggesting that admixture across clades is a notable source of ingly, the mean haplotype sharing is slightly higher in the Asian
morphologic diversity (Cadieu et al., 2009; Parker et al., 2009; spitz clade than it is in the Mediterranean clade (mean =
Sutter et al., 2007). Studies of linkage disequilibrium and haplo- 1,596,000 and 1,317,000, respectively) (Figure S1), implying
type sharing suggest further that within regions of 1015 kb, that the Asian spitz breeds have been used in recent crosses
there exist a small number of haplotypes that are shared by while the Mediterranean breeds are currently more segregated.
the majority of breeds, while breed specificity is revealed only These data describe a staggered pattern of dog breed creation
in large haplotypes (Lindblad-Toh et al., 2005; Sutter et al., 2004). starting with separation by type based on required function
In this study, we observe that the majority of dog breeds either and the form necessary to carry out that function. This would
do not share large haplotypes outside their clade or share with have taken place as the need arose during early human progres-
only one remote breed. The small number of breeds that share sion from hunter-gather to pastoral, agricultural, and finally ur-
excessively outside their assigned clade could be recently ban lifestyles. During the last 200 years, these breed types
created from multiple diverse breeds or may have been popular were refined into very specific breeds by dividing the original
contributors to other breeds. For example, the pug dog groups functional dog into morphotypes based on small changes in
closely with the European toy breed, Brussels griffon (Figure 2F), appearance and with occasional outcrosses to enhance appear-
in the toy spitz clade but also shares extensive haplotypes with ance or alter behavior (e.g., reduce aggression, increase
the Asian toy breeds (Figure 2B) as well as many small dog biddability).
breeds from multiple other clades. This likely indicates the Though most breeds within a clade appear to be the result of
pugs early exportation from Asia and subsequent contribution descent from a common ancestor, the New World dogs and the
to many small breeds (Watson, 1906). Consider also the exten- Asian toys showed nearly 200% improvement in the maximum
sive cross-clade haplotype sharing in the chinook, a recently likelihood score by allowing for admixture between the breeds
created breed with multiple ancestors from different breeds. within the clade. Based on this analysis, the Asian toy dogs
Our data both recapitulates and enhances the written history were likely not considered separate breeds when first exported
of this breed (http://www.chinook.org/history.html) (Data S1). from their country of origin resulting in multiple admixture events
Extreme examples such as these underscore the complications (Figure 6C). Unexpectedly, the New World clade admixture
implicit in relying on phylogeny alone to describe breed relation- events center exclusively on the German shepherd dog, which
ships. Overall, our data show that admixture has played an informs both the development of this breed as well as immigra-
important role in the development of many breeds and, as new tion of dog breeds to the New World (Figure 6B). The inclusion of
hybrids are added to phylogenetic analyses, the topology of German shepherd dog with cane paratore, an Italian working
the cladogram will likely rearrange to accommodate. farm dog, likely indicates a recent common ancestor among
The ability to determine a time of hybridization for recent these breeds, as the German shepherd dog was derived from
admixture events can refine sparse historical accounts of breed a herding dog of unknown ancestry in the late 1800s (http://
formation. For example, when dog fighting was a popular form of gsdca.org). However, the hybridization of the German shepherd
entertainment, many combinations of terriers and mastiff or dog with the Peruvian hairless dog and the xoloitzcuintli, also a
bully-type breeds were crossed to create dogs that would excel hairless breed, is unexpected and could be the result of recent
in that sport. In this analysis, all of the bull and terrier crosses admixture to enhance the larger varieties of these breeds or
map to the terriers of Ireland and date to 18601870. This coin- could indicate admixture of generic herding dogs from Southern
cides perfectly with the historical descriptions that, though they Europe into South America during the Columbian Exchange.
do not clearly identify all breeds involved, report the popularity of Dogs have been in the Americas for more than 10,000 years,
dog contests in Ireland and the lack of stud book veracity, hence likely traveling from East Asia with the first humans (Wang
undocumented crosses, during this era of breed creation (Lee, et al., 2016). However, studies of mitochondrial DNA suggest
1894). that the original New World dogs were almost entirely replaced
The dates estimated from these data are approximations, as through European contact (Castroviejo-Fisher et al., 2011;
selection for or against traits that accompanied each cross, as Wayne and Ostrander, 1999; Witt et al., 2014) and additional
well as the size of the population at the time of the cross, would Asian migrations (Brown et al., 2015). As colonists came to the
have affected retention of the haplotypes within the genome. Americas from the 16th to the 19th centuries, they brought Old

Cell Reports 19, 697708, April 25, 2017 705

World livestock, and therefore the dogs required to manage and depictions show sighthound-type hunting dogs that date back
tend the livestock, to the New World (Crosby, 1972). Many of the 4,000 ybp (Alderton, 2002; Fogle, 2000), and one of the earliest
newly introduced animals outcompeted the native animals, known writings regarding segregation of dogs based by type
which may explain the surprising and very strong herding dog clearly delineates hunting dogs from working dogs (Columella,
signature in the native hairless breeds of South and Central 1954). The new cladogram presented herein suggests that the
America that were not developed to herd. In this analysis, we switch from hunting to agricultural pursuits may have initiated
observe that the ancient hairless breeds show extensive hybrid- early breed formation and that this occurred in multiple regions.
ization with herding dogs from Europe and, to a lesser extent, These data show that geographical region can define a founda-
with each other. We also identify two additional clades of New tional canid population within which selection for universally rele-
World breeds, the American terriers and the American toys (Fig- vant behaviors occurred independently, separating the regional
ures 2I and 2J), two monophyletic clades of small-sized breeds groups also by function long ago.
from North/Central America, which include a set of related ter- The lack of admixture across clades that appear to share a
riers, and the Chihuahua and Chinese crested. Written records common trait suggests that these traits may have arisen inde-
state that the terriers trace their ancestry to the feists, a pendently, multiple times. For example, these data show no
North American landrace dog bred for hunting (http://www. recent haplotype sharing between the giant flock guards of the
americantreeingfeist.com,akc.org). The Chihuahua and Chinese Mediterranean and the European mastiffs (Figure S2D). These
crested are both believed to have originated in Central America breed types required large size for guarding; however, each
(American Kennel Club, 1998; Parker et al., 2017), despite the used that size in a different way, a fact that was recognized at
nomenclature of the latter, which implies Asian ancestry. In least 2,000 years ago (Columella, 1954). The flock guards use
contrast, most new breeds developed in the Americas were their size to defeat animal predators, while the mastiffs use their
created from crosses of European breeds and cluster accord- size to keep human predators at bay, often through fierce coun-
ingly (i.e., Boston terrier [European mastiff], Nova Scotia duck tenance rather than action. The phylogenetic placement of these
tolling retriever [retriever], and Australian shepherd [UK Rural]). breeds and lack of recent admixture suggests that giant size
The separation of the older American breeds on the cladogram, developed independently in the different clades and that it may
despite recent European admixture, suggests that both clades have been one of the earliest traits by which breeds were segre-
may retain the aboriginal New World dog genomic signatures in- gated thousands of years ago.
termixed with the European breed haplotypes, similar to the The cladogram of 161 breeds presented here represents the
admixture among European, African, and Native American ge- most diverse dataset of domestic dog breeds analyzed to
nomes that can be found in modern South American human pop- date, displaying 23 well-supported clades of breeds represent-
ulations (Mathias et al., 2016; Ruiz-Linares et al., 2014). This is ing breed types that existed before the advent of breed clubs
the first indication that the New World dog signature may not and registries. While the addition of more rare or niche breeds
be entirely extinct in modern dog breeds, as has been previously will produce a denser tree, the results here address many unan-
suggested (Leonard et al., 2002). swered questions regarding the origins of breeds. We show that
In addition to the effects on the native population, our analysis many traits such as herding, coursing, and intimidating size,
of geographically distinct subsets of the same breeds shows that which are associated with specific canine occupations, have
some degree of admixture also occurs in the imported breeds likely been developed more than once in different geographical
when first introduced into a new country. These data suggest locales during the history of modern dog. These data also
two outcomes of breed immigration that mirror human immigra- show that extensive haplotype sharing across clades is a likely
tion into a new region: the immigrant population is less diverse indicator of recent admixture that took place in the time since
than the founding population, and there is often admixture with the advent of breed registries, thus leading to the creation of
the native population in early generations (Baharian et al., most of the modern breeds. However, the primary breed types
2016; Zhai et al., 2016). were developed well before this time, indicating selection and
We observe further evidence of the role geography plays in the segregation of dog populations in the absence of formal breed
distribution of breeds within the clades. For instance, both the recognition. Breed prototypes have been forming through selec-
UK Rural and the Mediterranean clades include both sighthound tive pressures since ancient times depending on the job they
and working dog breeds, two highly divergent groups in terms of were most required to perform. A second round of hybridization
physical and behavioral phenotype. Sighthounds are lithe and and selection has been applied within the last 200 years to create
leggy hunters, built to run fast, and have a strong prey drive. the many unique combinations of traits that modern breeds
Working dogs include both the tall and heavy flock guards that display. By combining genetic distance relationships with pat-
are bred to live among herds without human interaction, prevent- terns of haplotype sharing, we can now elucidate the complex
ing predator attacks, and mid-sized herders (Figure 2T), which makeup of modern dogs breeds and guide the search for genetic
are agile and bred to work closely with humans to control the variants important to canine breed development, morphology,
movement of the flock without harming them. Despite the behavior, and disease.
opposing phenotypes under selection, both breed types form
single clades stemming from distinct geographical regions.
EXPERIMENTAL PROCEDURES
Haplotype analysis shows no recent admixture between the
geographically distinct clades, suggesting that these groups Further details and an outline of resources used in this work can be found in the
arose independently (Figures S2A and S2B). Archeological Supplemental Experimental Procedures.

706 Cell Reports 19, 697708, April 25, 2017

ACCESSION NUMBERS Druzhkova, A.S., Thalmann, O., Trifonov, V.A., Leonard, J.A., Vorobieva, N.V.,
Ovodov, N.D., Graphodatsky, A.S., and Wayne, R.K. (2013). Ancient DNA anal-
The accession numbers for the raw data files for the SNP genotype arrays ysis affirms the canid from Altai as a primitive dog. PLoS ONE 8, e57754.
reported in this paper are GEO: GSE90441, GSE83160, GSE70454, and Fogle, B. (2000). The New Encyclopedia of the Dog, Second Edition (Dorling
GSE96736. Kindersley Publishing, Inc.).
Hayward, J.J., Castelhano, M.G., Oliveira, K.C., Corey, E., Balkman, C.,
SUPPLEMENTAL INFORMATION Baxter, T.L., Casal, M.L., Center, S.A., Fang, M., Garrison, S.J., et al. (2016).
Complex disease and phenotype mapping in the domestic dog. Nat. Commun.
Supplemental Information includes Supplemental Experimental Procedures, 7, 10460.
two figures, two tables, and two data files and can be found with this article Irion, D.N., Schaffer, A.L., Famula, T.R., Eggleston, M.L., Hughes, S.S., and
online at http://dx.doi.org/10.1016/j.celrep.2017.03.079. Pedersen, N.C. (2003). Analysis of genetic variation in 28 dog breed popula-
tions with 100 microsatellite markers. J. Hered. 94, 8187.
AUTHOR CONTRIBUTIONS Koskinen, M.T. (2003). Individual assignment using microsatellite DNA reveals
unambiguous breed identification in the domestic dog. Anim. Genet. 34,
H.G.P. conceived of project, performed analyses, created figures, and pre- 297301.
pared the manuscript. D.L.D. created figures and assisted in manuscript prep-
Lachance, J., and Tishkoff, S.A. (2013). SNP ascertainment bias in population
aration. M.R. ran SNP chips and worked on early analysis. B.W.D. and A.B.
genetic analyses: why it is important, and how to correct it. BioEssays 35,
performed experiments. G.C.-R. performed sample collection and DNA isola-
780786.
tion. E.A.O. organized and directed the study and contributed to manuscript
preparation Lee, R.B. (1894). A History and Description of the Modern Dogs of Great Britain
and Ireland (Horace Cox).

ACKNOWLEDGMENTS Leonard, J.A., Wayne, R.K., Wheeler, J., Valadez, R., Guillen, S., and Vila`, C.
(2002). Ancient DNA evidence for Old World origin of New World dogs. Science
We gratefully acknowledge support from the Intramural Program of the Na- 298, 16131616.
tional Human Genome Research Institute. We thank Sir Terence Clark for col- Lindblad-Toh, K., Wade, C.M., Mikkelsen, T.S., Karlsson, E.K., Jaffe, D.B.,
lecting DNA samples from multiple breeds of sighthounds from their countries Kamal, M., Clamp, M., Chang, J.L., Kulbokas, E.J., 3rd, Zody, M.C., et al.
of origin in Africa and Asia; Mauricio Lima, Flavio Bruno, and Robert Gennari for (2005). Genome sequence, comparative analysis and haplotype structure of
collecting samples from native Italian breeds; and Lei Song for collecting sam- the domestic dog. Nature 438, 803819.
ples from native Tibetan mastiffs. Lowe, J.K., Kukekova, A.V., Kirkness, E.F., Langlois, M.C., Aguirre, G.D.,
Acland, G.M., and Ostrander, E.A. (2003). Linkage mapping of the primary dis-
Received: January 3, 2017 ease locus for collie eye anomaly. Genomics 82, 8695.
Revised: February 10, 2017
Mathias, R.A., Taub, M.A., Gignoux, C.R., Fu, W., Musharoff, S., OConnor,
Accepted: March 28, 2017
T.D., Vergara, C., Torgerson, D.G., Pino-Yanes, M., Shringarpure, S.S.,
Published: April 25, 2017
et al.; CAAPA (2016). A continuum of admixture in the Western Hemisphere re-
vealed by the African Diaspora genome. Nat. Commun. 7, 12522.
REFERENCES
Mealey, K.L., and Meurs, K.M. (2008). Breed distribution of the ABCB1-1Delta
(multidrug sensitivity) polymorphism among dogs undergoing ABCB1 geno-
Alderton, D. (2002). Dogs (Dorling Kindersley, Ltd.).
typing. J. Am. Vet. Med. Assoc. 233, 921924.
American Kennel Club (1998). The Complete Dog Book, 19th Edition Revised
Moody, J.A., Clark, L.A., and Murphy, K.E. (2006). Canine history and breed
(Howell Book House).
clubs. In The Dog and Its Genome, E.A. Ostrander, U. Giger, and K. Lind-
Baharian, S., Barakatt, M., Gignoux, C.R., Shringarpure, S., Errington, J., Blot, blad-Toh, eds. (Cold Spring Harbor Laboratory Press), pp. 118.
W.J., Bustamante, C.D., Kenny, E.E., Williams, S.M., Aldrich, M.C., and
Pang, J.F., Kluetsch, C., Zou, X.J., Zhang, A.B., Luo, L.Y., Angleby, H., Arda-
Gravel, S. (2016). The great migration and African-American genomic diversity.
lan, A., Ekstrom, C., Skollermo, A., Lundeberg, J., et al. (2009). mtDNA data
PLoS Genet. 12, e1006059.
indicate a single origin for dogs south of Yangtze River, less than 16,300 years
Brown, S.K., Darwent, C.M., Wictum, E.J., and Sacks, B.N. (2015). Using mul- ago, from numerous wolves. Mol. Biol. Evol. 26, 28492864.
tiple markers to elucidate the ancient, historical and modern relationships
Parker, H.G., Kim, L.V., Sutter, N.B., Carlson, S., Lorentzen, T.D., Malek, T.B.,
among North American Arctic dog breeds. Heredity (Edinb.) 115, 488495.
Johnson, G.S., DeFrance, H.B., Ostrander, E.A., and Kruglyak, L. (2004).
Browning, B.L., and Browning, S.R. (2013). Improving the accuracy and effi- Genetic structure of the purebred domestic dog. Science 304, 11601164.
ciency of identity-by-descent detection in population data. Genetics 194,
Parker, H.G., Kukekova, A.V., Akey, D.T., Goldstein, O., Kirkness, E.F.,
459471.
Baysac, K.C., Mosher, D.S., Aguirre, G.D., Acland, G.M., and Ostrander,
Cadieu, E., Neff, M.W., Quignon, P., Walsh, K., Chase, K., Parker, H.G., Von- E.A. (2007). Breed relationships facilitate fine-mapping studies: a 7.8-kb dele-
holdt, B.M., Rhue, A., Boyko, A., Byers, A., et al. (2009). Coat variation in the tion cosegregates with Collie eye anomaly across multiple dog breeds.
domestic dog is governed by variants in three genes. Science 326, 150153. Genome Res. 17, 15621571.
Castroviejo-Fisher, S., Skoglund, P., Valadez, R., Vila`, C., and Leonard, J.A. Parker, H.G., VonHoldt, B.M., Quignon, P., Margulies, E.H., Shao, S., Mosher,
(2011). Vanishing native American dog lineages. BMC Evol. Biol. 11, 73. D.S., Spady, T.C., Elkahloun, A., Cargill, M., Jones, P.G., et al. (2009). An ex-
Columella, L.J.M. (1954). On Agriculture (De Re Rustica), Vol. Books 59 (E.S. pressed fgf4 retrogene is associated with breed-defining chondrodysplasia in
Forster and E.H. Heffner, Trans.) (Harvard University Press). domestic dogs. Science 325, 995998.
Crosby, A.W., Jr. (1972). The Columbian Exchange (Greenwood Publishing Parker, H.G., Harris, A., Dreger, D.L., Davis, B.W., and Ostrander, E.A. (2017).
Company). The bald and the beautiful: hairlessness in domestic dog breeds. Philos. Trans.
Donner, J., Kaukonen, M., Anderson, H., Moller, F., Kyostila, K., Sankari, S., R. Soc. Lond. B Biol. Sci. 372, 372.
Hytonen, M., Giger, U., and Lohi, H. (2016). Genetic panel screening of nearly Patterson, N., Moorjani, P., Luo, Y., Mallick, S., Rohland, N., Zhan, Y., Gen-
100 mutations reveals new insights into the breed distribution of risk variants schoreck, T., Webster, T., and Reich, D. (2012). Ancient admixture in human
for canine hereditary disorders. PLoS ONE 11, e0161005. history. Genetics 192, 10651093.

Cell Reports 19, 697708, April 25, 2017 707

Pickrell, J.K., and Pritchard, J.K. (2012). Inference of population splits and mix- phenotypic variation between dog breeds using selection mapping. PLoS
tures from genome-wide allele frequency data. PLoS Genet. 8, e1002967. Genet. 7, e1002316.
Quignon, P., Herbin, L., Cadieu, E., Kirkness, E.F., Hedan, B., Mosher, D.S., Vila`, C., Savolainen, P., Maldonado, J.E., Amorim, I.R., Rice, J.E., Honeycutt,
Galibert, F., Andre, C., Ostrander, E.A., and Hitte, C. (2007). Canine population R.L., Crandall, K.A., Lundeberg, J., and Wayne, R.K. (1997). Multiple and
structure: assessment and impact of intra-breed stratification on SNP-based ancient origins of the domestic dog. Science 276, 16871689.
association studies. PLoS ONE 2, e1324. Vila`, C., Maldonado, J.E., and Wayne, R.K. (1999). Phylogenetic relationships,
Ruiz-Linares, A., Adhikari, K., Acuna-Alonzo, V., Quinto-Sanchez, M., Jara- evolution, and genetic diversity of the domestic dog. J. Hered. 90, 7177.
millo, C., Arias, W., Fuentes, M., Pizarro, M., Everardo, P., de Avila, F., et al.
Vonholdt, B.M., Pollinger, J.P., Lohmueller, K.E., Han, E., Parker, H.G.,
(2014). Admixture in Latin America: geographic structure, phenotypic diversity
Quignon, P., Degenhardt, J.D., Boyko, A.R., Earl, D.A., Auton, A., et al.
and self-perception of ancestry based on 7,342 individuals. PLoS Genet. 10,
(2010). Genome-wide SNP and haplotype analyses reveal a rich history under-
e1004572.
lying dog domestication. Nature 464, 898902.
Shannon, L.M., Boyko, R.H., Castelhano, M., Corey, E., Hayward, J.J.,
Wang, G.D., Zhai, W., Yang, H.C., Wang, L., Zhong, L., Liu, Y.H., Fan, R.X., Yin,
McLean, C., White, M.E., Abi Said, M., Anita, B.A., Bondjengo, N.I., et al.
T.T., Zhu, C.L., Poyarkov, A.D., et al. (2016). Out of southern East Asia: the nat-
(2015). Genetic structure in village dogs reveals a Central Asian domestication
ural history of domestic dogs across the world. Cell Res. 26, 2133.
origin. Proc. Natl. Acad. Sci. USA 112, 1363913644.
Watson, J. (1906). The Dog Book, Volume II (Doubleday, Page & Company).
Sutter, N.B., Eberle, M.A., Parker, H.G., Pullar, B.J., Kirkness, E.F., Kruglyak,
L., and Ostrander, E.A. (2004). Extensive and breed-specific linkage disequi- Wayne, R.K., and Ostrander, E.A. (1999). Origin, genetic diversity, and genome
librium in Canis familiaris. Genome Res. 14, 23882396. structure of the domestic dog. BioEssays 21, 247257.
Sutter, N.B., Bustamante, C.D., Chase, K., Gray, M.M., Zhao, K., Zhu, L., Pad- Wilcox, B., and Walkowicz, C. (1995). Atlas of Dog Breeds of the World, Fifth
hukasahasram, B., Karlins, E., Davis, S., Jones, P.G., et al. (2007). A single Edition (T.F.H. Publications).
IGF1 allele is a major determinant of small size in dogs. Science 316, 112115. Witt, K.E., Judd, K., Kitchen, A., Grier, C., Kohler, T.A., Ortman, S.G., Kemp,
Thalmann, O., Shapiro, B., Cui, P., Schuenemann, V.J., Sawyer, S.K., Green- B.M., and Malhi, R.S. (2014). DNA analysis of ancient dogs of the Americas:
field, D.L., Germonpre, M.B., Sablin, M.V., Lopez-Giraldez, F., Domingo- identifying possible founding haplotypes and reconstructing population his-
Roura, X., et al. (2013). Complete mitochondrial genomes of ancient canids tories. J. Hum. Evol. 79, 105118.
suggest a European origin of domestic dogs. Science 342, 871874. Zhai, G., Zhou, J., Woods, M.O., Green, J.S., Parfrey, P., Rahman, P., and
Vaysse, A., Ratnakumar, A., Derrien, T., Axelsson, E., Rosengren Pielberg, G., Green, R.C. (2016). Genetic structure of the Newfoundland and Labrador
Sigurdsson, S., Fall, T., Seppala, E.H., Hansen, M.S., Lawley, C.T., et al.; population: founder effects modulate variability. Eur. J. Hum. Genet. 24,
LUPA Consortium (2011). Identification of genomic regions associated with 10631070.

708 Cell Reports 19, 697708, April 25, 2017