ArticlePDF Available

The genomic history and global expansion of domestic donkeys

Authors:

Abstract and Figures

Donkeys transformed human history as essential beasts of burden for long-distance movement, especially across semi-arid and upland environments. They remain insufficiently studied despite globally expanding and providing key support to low- to middle-income communities. To elucidate their domestication history, we constructed a comprehensive genome panel of 207 modern and 31 ancient donkeys, as well as 15 wild equids. We found a strong phylogeographic structure in modern donkeys that supports a single domestication in Africa ~5000 BCE, followed by further expansions in this continent and Eurasia and ultimately returning to Africa. We uncover a previously unknown genetic lineage in the Levant ~200 BCE, which contributed increasing ancestry toward Asia. Donkey management involved inbreeding and the production of giant bloodlines at a time when mules were essential to the Roman economy and military. science.org/doi/10.1126/science.abo3503
Modern donkey dataset and population evolutionary history. (A) Number and geographical distribution of modern donkey samples (n = 207). Pie charts show the ADMIXTURE proportion of domestic ancestry (gray), African wild ass ancestry (white), and kiang ancestry (black) averaged across all individuals from each country (56). For visualization, the total surface of each pie chart is scaled to 2%. (B) Smartpca (57) of modern donkeys, with the imputed ancient samples in black. PC, principal component. (C) Treemix phylogeny of modern domesticates (excluding individuals with high wild introgression, n = 201) (27). Node supports are estimated from 100 bootstrap pseudo-replicates (confidence <90% in red). The percentage values indicate admixture proportions inferred from Treemix (27). (D) SMC++ demographic trajectories (colored) and split time estimates (black) for pairs of main geographic regions (28), repeating the analysis on two datasets of three individuals per population (the second dataset is shown in semitransparency). Modern donkeys are colored and shaped according to geographical location and continents in all panels. Country Codes: Algeria (ALG), Brazil (BRA), China (CHI), Canary Islands (CYK), Denmark (Eas), Egypt (EGY), Spain (ESP), Ethiopia (ETH), Ghana (GHA), Iran (IRA), Ireland (IRE), Kazakhstan (KAZ), Kenya (KEN), Kyrgyzstan (KYR), Mauritania (MAU), Mongolia (MON), Nigeria (NIG), Oman (OMA), Portugal (PTG), Saudi Arabia (SAU), Senegal (SEN), Somalia (SOM), Sudan (SUD), Syria (SYR), Tibet (TIB), Turkmenistan (TKM), Turkey (TUK), Tunisia (TUN), Yemen (YEM), Croatia (YUC), Macedonia (YUM).
… 
Content may be subject to copyright.
  


 rf-   
Jt
 

   
 

   
  

=
  
    
    
  f   
     
;t/1· tf  
   
   
 
 
  




h
 





 I



 

·

I








t


 
 
I[


  

RESEARCH ARTICLE
DOMESTICATION
The genomic history and global expansion
of domestic donkeys
Evelyn T. Todd
1
, Laure Tonasso-Calvière
1
, Loreleï Chauvey
1
, Stéphanie Schiavinato
1
,
Antoine Fages
1
, Andaine Seguin-Orlando
1
, Pierre Clavel
1
, Naveed Khan
1,2
, Lucía Pérez Pardal
3,4
,
Laura Patterson Rosa
5
, Pablo Librado
1
, Harald Ringbauer
6
, Marta Verdugo
7
, John Southon
8
,
Jean-Marc Aury
9
, Aude Perdereau
9
, Emmanuelle Vila
10
, Matilde Marzullo
11
, Ornella Prato
11
,
Umberto Tecchiati
11
, Giovanna Bagnasco Gianni
11
, Antonio Tagliacozzo
12
, Vincenzo Tinè
13
,
Francesca Alhaique
12
, João Luís Cardoso
14,15
, Maria João Valente
16
, Miguel Telles Antunes
17
,
Laurent Frantz
18,19
, Beth Shapiro
20,21
, Daniel G. Bradley
7
, Nicolas Boulbes
22
, Armelle Gardeisen
23
,
Liora Kolska Horwitz
24
, Aliye Öztan
25
, Benjamin S. Arbuckle
26
, Vedat Onar
27
, Benoît Clavel
28
,
Sébastien Lepetz
28
, Ali Akbar Vahdati
29
, Hossein Davoudi
30
, Azadeh Mohaseb
28,30
,
Marjan Mashkour
28,30,31
, Olivier Bouchez
32
, Cécile Donnadieu
32
, Patrick Wincker
9
,
Samantha A. Brooks
33
, Albano Beja-Pereira
3,4,34,35
, Dong-Dong Wu
36,37
, Ludovic Orlando
1
*
Donkeys transformed human history as essential beasts of burden for long-distance movement, especially
across semi-arid and upland environments. They remain insufficiently studied despite globally expanding
and providing key support to low- to middle-income communities. To elucidate their domestication history,
we constructed a comprehensive genome panel of 207 modern and 31 ancient donkeys, as well as
15 wild equids. We found a strong phylogeographic structure in modern donkeys that supports a single
domestication in Africa ~5000 BCE, followed by further expansions in this continent and Eurasia and
ultimately returning to Africa. We uncover a previously unknown genetic lineage in the Levant ~200 BCE,
which contributed increasing ancestry toward Asia. Donkey management involved inbreeding and the
production of giant bloodlines at a time when mules were essential to the Roman economy and military.
Domestic donkeys (Equus asinus)have
facilitated the movement of goods and
people for millennia, enabling trade and
transport across a broad spectrum of
landscapes (1). Despite their importan ce
to ancient pastoral societies, little is known
about the deep history of donkeys and the im-
pact of human management on their genomes.
This is most likely due to their undervalued
status and loss of utility in modern industrial-
ized societies. Donkeys are, however, extraordi-
nary working animals and remain essential for
developing communities, especially in semi-
arid environments (2). Understanding their
genetic makeup is not only key to assessing
their contribution to human history but also
to improving their local management in the
future.
The current archaeological record of early
donkeys is limited (1,3), which makes their
domestic origins and spread through the
world contentious. The reduced body size of
zooarchaeological ass remains in Egypt at El
Omari (4800 to 4500 BCE) and Maadi (4000 to
3500 BCE) has been interpreted as early evi-
dence of domestication (47). Carvings on the
LibyanpalettethatwerefoundinAbydos,
Egypt (3200 to 3000 BCE), depict lines of walk-
ing asses, cattle, and sheep, which also suggests
a domestication context (8,9). Together with
contemporary remains from the same region
that show morphological evidence for load ca r-
rying (10), these findings suggest that donkeys
could have been first domesticated within a
range extending from the Northeastern Sahara,
theNileValley,theAtbaraRiver,andtheRed
Sea Hills to Eritrea. In this model, donkeys were
domesticated by pastoralists to assist with mo-
bility around 5500 to 4500 BCE b eca use of th e
large-scale aridification of the Sahara (1). Inde-
pendent evidence based on patterns of mito-
chondrial (11,12) and nuclear sequence variation
(13) also points to the African origins of the don-
key, owing to their closer proximity to African
wild asses (Equus africanus spp.) than to Asian
wild asses (Equus hemionus spp.).
However, candidate regions outside of Africa
are also proposed as alternative domestication
centers. In Ash Shuman (Yemen), for example,
ass remains of disputed domestic status predate
those from Egypt by 2000 years (~6500 BCE)
(14). Likewise, textual, iconographic, and zoo-
archaeological material indicates a possible
additional center in Mesopotamia during the
fourth and third millennia BCE (1519), a
context in which first-generation hybrids of
donkeys and Syrian onagers have been iden-
tified genetically (20). The segregation of
mitochondrial variation in two main clades
may also support a dual domestication process,
for which the Nubian wild ass (Equus africanus
africanus) is securely identified as the progen-
itor of Clade I (11,12). As for the ancestor of
Clade II, it could either be the extinct Atlas
wild ass (Equus africanus atlanticus), which
was endemic to Northern Africa, or another un-
described subspecies that potentially ranged
outside of Africa. Whether a single, maternally
RESEARCH
Todd et al., Science 377, 11721180 (2022) 9 September 2022 1of9
1
Centre dAnthropobiologie et de Génomique de Toulouse (CAGT), CNRS UMR 5288, Université Paul Sabatier, Toulouse 31000, France.
2
Department of Biotechnology, Abdul Wali Khan University,
Mardan 23200, Pakistan.
3
CIBIO, Centro de Investigação em Biodiversidade e Recursos Genéticos, InBIO Laboratório Associado, Campus de Vairão, Universidade do Porto, Vairão 4485-661,
Portugal.
4
BIOPOLIS Program in Genomics, Biodiversity and Land Planning, Campus de Vairão, Universidade do Porto, Vairão 4485-661, Portugal.
5
Department of Animal Science, Sul Ross State
University, Alpine, TX 79830, USA.
6
Department of Archaeogenetics, Max Planck Institute for Evolutionary Anthropology, Leipzig 04103, Germany.
7
Smurfit Institute of Genetics, Trinity College
Dublin, Dublin D02 PN40, Ireland.
8
Earth System Science Department, University of California, Irvine, CA 92697, USA.
9
Genoscope, Institut de biologie François Jacob, CEA, Université dEvry,
Université Paris-Saclay, Evry 91042, France.
10
Laboratoire Archéorient, Université Lyon 2, Lyon 69007, France.
11
Dipartimento di Beni Culturali e Ambientali, Università degli Studi di Milano, Milan
20122, Italy.
12
Bioarchaeology Service, Museo delle Civiltà, Rome 00144, Italy.
13
Soprintendenza archeologia belle arti e paesaggio per le province di Verona, Rovigo e Vicenza, Verona 37121, Italy.
14
ICArEHB, Campus de Gambelas, University of Algarve, Faro 8005-139, Portugal.
15
Universidade Aberta, Lisbon 1269-001, Portugal.
16
Faculdade de Ciências Humanas e Sociais, Centro de
Estudos de Arqueologia, Artes e Ciências do Património, Universidade do Algarve, Faro 8000-117, Portugal.
17
Centre for Research on Science and Geological Engineering, Universidade NOVA de
Lisboa, Lisbon 1099-085, Portugal.
18
Palaeogenomics Group, Department of Veterinary Sciences, Ludwig Maximilian University, Munich 80539, Germany.
19
School of Biological and Behavioural
Sciences, Queen Mary University of London, London E1 4DQ, United Kingdom.
20
Department of Ecology and Evolutionary Biology, University of California, Santa Cruz, CA 95064, USA.
21
Howard
Hughes Medical Institute, University of California, Santa Cruz, CA 95064, USA.
22
Institut de Paléontologie Humaine, Fondation Albert Ier, Paris / UMR 7194 HNHP, MNHN-CNRS-UPVD / EPCC
Centre Européen de Recherche Préhistorique, Tautavel 66720, France.
23
Archéologie des Sociétés Méditéranéennes, Université Paul Valéry - Site Saint-Charles 2, Montpellier 34090, France.
24
National Natural History Collections, Edmond J. Safra Campus, Givat Ram, The Hebrew University, Jerusalem 9190401, Israel.
25
Archaeology Department, Ankara University, Ankara 06100,
Turkey.
26
Department of Anthropology, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599, USA.
27
Osteoarchaeology Practice and Research Center and Department of Anatomy,
Faculty of Veterinary Medicine, Istanbul University-Cerrahpaşa, Istanbul 34320, Turkey.
28
Archéozoologie, Archéobotanique, Sociétés, Pratiques et Environnements, Muséum National d'Histoire
Naturelle, Paris 75005, France.
29
Provincial Office of the Iranian Center for Cultural Heritage, Handicrafts and Tourism Organisation, North Khorassan, Bojnord 9416745775, Iran.
30
Archaezoology
section, Bioarchaeology Laboratory of the Central Laboratory, University of Tehran, Tehran CP1417634934, Iran.
31
Department of Osteology, National Museum of Iran, Tehran 1136918111, Iran.
32
GeT-PlaGe - Génome et Transcriptome - Plateforme Génomique, GET - Plateforme Génome & Transcriptome, Institut National de Recherche pour lAgriculture, lAlimentation et lEnvironnement,
Castaneet-Tolosan Cedex 31326, France.
33
Department of Animal Science, UF Genetics Institute, University of Florida, Gainesville, FL 32610, USA.
34
DGAOT, Faculty of Sciences, Universidade do
Porto, Porto 4169-007, Portugal.
35
Sustainable Agrifood Production Research Centre (GreenUPorto), Universidade do Porto, Vairão 4485-646, Portugal.
36
State Key Laboratory of Genetic
Resources and Evolution, Kunming Institute of Zoology, Chinese Academy of Sciences, Kunming, Yunnan 650201, China.
37
Kunming Natural History Museum of Zoology, Kunming Institute of
Zoology, Chinese Academy of Sciences, Kunming, Yunnan 650223, China.
*Corresponding author. Email: ludovic.orlando@univ-tlse3.fr
Downloaded from https://www.science.org at bibCNRS INSB on September 10, 2022
inherited marker captures the whole complex-
ity of underlying ancestries can, however, be
questioned after recent results from other ani-
mals [e.g., horses (21)]. Furthermore, previous
analyses of nuclear genetic variation in African
and non-African donkeys have failed to dis-
entangle their origins (13,22). Overall, this
lack of consensus between genetic and archae-
ological data means that the geographic and
temporal origin of donkeys and whether they
were do mesticated more than once remain
uncertain. The global spread of donkeys is also
unclear, as their worldwide patterns of genomic
diversity lack extensive characterization.
Modern donkeys originated in Africa and
spread into Eurasia
To address these issues, we sequenced 49 mod-
ern donkey genomes from underrepresented
regions and combined these with 158 public-
ly available genomes to capture worldwide di-
versity (Fig. 1A and table S1) (13,2325). We
constructed a fine-scale recombination map
from genomes that encompassed all phyloge-
netic groups, which we used to phase 13,013,551
variants (fig. S1 and tables S3 and S4). Prin-
cipal component analysis (26)revealedstrong
geographical substructuring, with donkeys
from Af rica , Europe, and Asia forming distinct
genetic clusters (Fig. 1B and figs. S2 and S3).
Todd et al., Science 377, 11721180 (2022) 9 September 2022 2of9
Fig. 1. Modern donkey dataset and population evolutionary history.
(A) Number and geographical distribution of modern donkey samples (n= 207).
Pie charts show the ADMIXTURE proportion of domestic ancestry (gray),
African wild ass ancestry (white), and kiang ancestry (black) averaged across all
individuals from each country (56). For visualization, the total surface of each
pie chart is scaled to 2%. (B) Smartpca (57) of modern donkeys, with the imputed
ancient samples in black. PC, principal component. (C) Treemix phylogeny of
modern domesticates (excluding individuals with high wild introgression, n= 201)
(27). Node supports are estimated from 100 bootstrap pseudo-replicates
(confidence <90% in red). The percentage values indicate admixture proportions
inferred from Treemix (27). (D) SMC++ demographic trajectories (colored)
and split time estimates (black) for pairs of main geographic regions (28),
repeating the analysis on two datasets of three individuals per population
(the second dataset is shown in semitransparency). Modern donkeys are colored
and shaped according to geographical location and continents in all panels.
Country Codes: Algeria (ALG), Brazil (BRA), China (CHI), Canary Islands (CYK),
Denmark (Eas), Egypt (EGY), Spain (ESP), Ethiopia (ETH), Ghana (GHA), Iran
(IRA), Ireland (IRE), Kazakhstan (KAZ), Kenya (KEN), Kyrgyzstan (KYR),
Mauritania (MAU), Mongolia (MON), Nigeria (NIG), Oman (OMA), Portugal
(PTG), Saudi Arabia (SAU), Senegal (SEN), Somalia (SOM), Sudan (SUD), Syria
(SYR), Tibet (TIB), Turkmenistan (TKM), Turkey (TUK), Tunisia (TUN), Yemen
(YEM), Croatia (YUC), Macedonia (YUM).
RESEARCH |RESEARCH ARTICLE
Downloaded from https://www.science.org at bibCNRS INSB on September 10, 2022
A Treemix phylogenetic reconstruction that
grouped modern donkeys according to sampl-
ing locations (27) confirms the earliest split
between African (Clade A) and mostly non-
African donkeys (Clade B) (Fig. 1C). Further
structure within Clade A separates donkeys
from the Horn of Africa (Ethiopia and Somalia)
plus Kenya from those from Western Africa
(Ghana, Mauritania, Nigeria, and Senegal).
Within Clade B, we find another major diver-
gence between European and Asian donkeys,
with east-to-west affinities in Europe from
the Balkans (Croatia and Macedonia) to Iberia,
Denmark, and Ireland. Conversely, Asian sub-
populations show west-to-east substructur-
ing from Iran and Central Asia to China and
Mongolia. Combined, these findings suggest
expansions from a central source into both
continents.
In Clade B, some of the most basal donkeys
are from the Southern Arabian Peninsula
(Om an and Yemen), whereas the single don-
key from Saudi Arabia analyzed here shows
European affinities, which is indicative of a
secondary translocation. Similarly, the Pega
donkey from Brazil is nested within Iberia,
which mirrors the colonization history of the
Americas. Clade B also includes donkeys
from Nubia (Egypt and Sudan) that show
affinities to the Levant (Syria) and Anatolia
(Turkey), as well as donkeys from Maghreb
(Tunisia), with closer genetic proximity to
European subpopulations. This suggests gene
flow into Africa from donkeys that were native
to Anatolia and the Levant, but not to the
Arabian Peninsula. Overall, this phy logenetic
reconstruction is compatible with both mod-
els of donkey domestication: a unique origin
in Africa followed by dispersals out and back,
or dual origins in Africa and the Southern
Arabian Peninsula.
The unique origin model posits a demo-
graphic expansion in Africa first and subse-
quent waves into Europe and Asia. By contrast,
dual origins would result in an earlier split of
demographic trajectories between African and
Eurasian subpopulations, given their deep phy-
logenetic divergence. To test these, we first per-
formed demographic modeling using SMC++
(28), which revealed the first expansion around
5200 BCE (7186 ± 742 years ago), in line with
archaeological evidence of domestication oc-
curring at this time (Fig. 1D and fig. S6). Addi-
tionally, when modeled from a possible African
source, SMC++ trajectories indicated more re-
cent and nearly coincidental expansions into
Asia around 2600 BCE (4573 ± 577 years ago)
and Europe around 2800 BCE (4806 ± 671 years
ago) (Fig. 1D). This is in line with the unique
origin model and the earliest archaeological
evidence of donkeys in Asia (Iranian Plateau
and the Indus Valley) and Europe (Portugal,
Greece, and Cyprus) in the mid-to-late third
millennia BCE (2934).
Furthermore, the Yemen and Oman sub-
populations do not branch basal to Clade B
according to fineSTRUCTURE (35), in contrast
to the expectations of the dual origins model,
but within Asian subpopulations (Fig. 2, A and
B). Lastly, the pairwise genetic distances be-
tween Ethiopia and non-African subpopula-
tions were greater than those from Yemen (fig.
S5). They both increased linearly with geo-
graphic distances and supported identical
dispersal rates (fig. S5; p= 0.775), which is in
line with a single wave of expansion at a con-
stant pace. Therefore, our analyses support
an early domestication in Africa that spread
at an even rate into the Arabian Peninsula
and Eurasia and flowed back into Nubia and
Maghreb. Modern subpopulations from the
Horn of Africa and Kenya so far best represent
the descendants of earliest donkeys.
Ancient donkey genomes reveal early and
rapid dispersal into Asia and secondary
contacts between Europe and Western Africa
The patterns of genetic variation within mod-
ern subpopulations may reflect recent breed-
ing history rather than early domestication
(36). Additionally, they could underrepresent
the contribution of lineages that were once
important but have since declined (37). Dating
population splits also assumes constant yet
unknown generation intervals. To address
these caveats and validate the domestica-
tion hi story reconstructed above, we gener-
ated a genomic time series spanning the past
~4000 years that included 31 ancient don-
keys from 11 different sites, which ranged
from the Atlantic shores (Portugal) to Central
Asia (Eastern Iran/Turkmenistan) (Fig. 3A and
table S2).
Ancient genomes sequenced to 0.77- to
5.05-fold coverage (table S6) were analyzed by
using two complementary methods: pseudo-
haploidization following (21) that resulted in
4,833,570 nucleotide transversion sites and
genotype imputation following (38), at 7,161,029
polymorphic sites present at >5% frequency in
modern donkeys (38,39). Imputation accuracy
was confirmed by the high consistency rates
between imputed and observed genotypes
after down-sampling of high-coverage modern
genomes and downstream analyses that were
largely consistent with those that were based
on pseudo-haploidized data (figs. S7 to S15
and table S6).
The three oldest samples from our dataset
consist of donkeys from Anatolia (Acemhöyük,
Turkey), which were radiocarbon dated to
2564 to 2039 BCE. Their age and phylogenetic
placement within Clade B confirm an early
expansion out of Africa by ~2500 BCE, which
is in agreement with SMC++ time estimates
(Fig. 2A and fig. S10). These samples, as well
as a donkey from Eastern Iran/Turkmenistan
that is affiliated to the BactriaMargiana Ar-
chaeological Complex (BMAC) (~2050 BCE;
sample Chalow3), branch before the forma-
tion of modern subpopulations from Central
Asia (Kazakhstan, Kyrgyzstan, Turkmenistan)
and Eastern Asia (China, Mongolia, Tibet)
(Fig. 2C). These subpopulations thus di-
verged after 2050 BCE, but potentially be-
fore the radiocarbon age of the donkey from
Doshan Tepe (1049 to 928 BCE), which appears
closer to modern subpopulations from Cen-
tral Asia in one Treemix analysis (Fig. 2D and
fig. S10).
Ancient samples from Iran (Shahr-i-Qumis,
800 BCE to 800 CE), including one Sassanid
(sample AM805), are not more closely related
to Central than to Eastern Asian modern sub-
populations, although their exact phylogenetic
placement remains unclear (Fig. 2, A and B,
and fig. S10). Their fineSTRUCTURE affinities
to modern Iran, Anatolia (Turkey), the Levant
(Syria), and Maghreb (Tunisia) support dif-
ferent genetic ancestry profiles from those
inferred at the nearby site of Doshan Tepe.
This indicates a population turnover in Iran
after ~1000 BCE but before ~500 CE, corre-
sponding to the radiocarbon time interval
of Doshan Tepe and a single specimen from
Shahr-i-Qumis.
Notably, all our ancient specimens from
Europe cluster within modern European do-
mesticates, supporting differentiation within
this continent before the oldest European
samples analyzed (Tarquinia, 803 to 412 BCE,
~2500 years ago; Fig. 3C). However, a donkey
from a Roman context in Marseille, a major
seaport trading center in Southern France
(Centre Bourse Marseille, 0 to 500 CE), dis-
played strong genetic affinities with modern
individuals from Western Africa (Fig. 3, B and
D). Additionally, SNP and haplotype shared-
ness with modern Western Africa were also
found in European donkeys from the Islamic
era in Portugal (Albufeira, 1228 to 1280 CE) and
Roman times in Northern France (Boinville-
en-Woëvre, 200 to 500 CE) (Fig. 3, B and E).
This reveals multiple contacts between Europe
and Western Africa from the Antiquity to
Middle Ages. Despite ancient European don-
keys showing Western African ancestry,
these contacts have affected Western Africa
more than Europe, in line wit h Tre emix
inferring gene flow predominantly in this
direction rather than the reverse (Fig. 1C).
All modern Irish donkeys and the two
Etruscan samples from Tarquinia are devoid
of Western African ancestry. This suggests the
preservation of old European genetic lineages,
at least in some modern subpopulations of
this continent.
Donkey management involved inbreeding and
introgression from divergent lineages
Inbreeding is a common reproductive strategy
for breeding animals with desirable traits (40).
Todd et al., Science 377, 11721180 (2022) 9 September 2022 3of9
RESEARCH |RESEARCH ARTICLE
Downloaded from https://www.science.org at bibCNRS INSB on September 10, 2022
To assess whether ancient donkey breeders
made use of inbreeding, we measured the pro-
portion of autosomal runs of homozygosity
(ROH) using three independent techniques,
all of which provided consistent results (figs.
S13 and S14). We detected inbreeding, but no
significant changes in levels between modern
and ancient donkeys (Wilcoxon rank sum test,
p=0.395)(Fig.4,A,B,andC).Conversely,
modern horses show higher inbreeding lev-
els than their ancient counterparts (Wilcoxon
rank sum test, p< 0.001), which mirrors pre-
vious reports of reduced heterozygosity and
increased deleterious mutation load in re-
cent times (Fig. 4, D, E, and F, table S11)
(21,41). Longer ROH tracts are more common
in modern horses and donkeys than in the
past, which is consistent with inbreeding
from closer generations in their genealogies
(Fig. 4, C and F). Overall, our analyses sup-
port recent major changes in reproductive
management inflating inbreeding in horses,
but not in donkeys.
Admixture modeling suggests ongoing intro-
gression from African wild asses into modern
donkeys from Africa and the Southern Arabian
Peninsula (with between 0.24 to 6.99% of ad-
mixture) (Fig. 1A, fig. S4, and table S5). This
is in line with free-ranging local management
practices that allow for continued interbreed-
ing with wild and feral subpopulations (4,42).
The limited but substantial amount of wild
genetic material from kiangs in one modern
donkey from China also supports admixture
between taxa generally regarded as separate
species. This confirms previous reports of
mitochondrial introgression (43)andgenomic
admixture despite different karyotypes (24).
Notably, all but one ancient donkey (sample
Tur168) carried remnants of outgroup mate-
rial (0.21 to 4.15%; Fig. 3B), which poten-
tially results from recent range contractions
of wild subpopulations and ancient manage-
ment practices providing more opportunities
for wild introgression.
The genome of MV242, a donkey from Israel
dating to the Hellenistic period (350 to 58 BCE),
displayed the largest fraction of divergent
genetic material (Fig. 3B, 4.15% ± 0.019). In
Treemix, this sample showed a deeper place-
ment than all donkeys present in our panel,
except the Somali wild ass [Equus africanus
somaliensis (E.a.som)] (Fig. 2E). Significantly
positive f4 (E.a.som, MV242; Horn+Ken, x)
statistics revealed MV242-related genetic an-
cestry in some modern subpopulations (x), espe-
cially toward Central and Eastern Asia (Fig. 5E).
This ancestry was already present in the BMAC
sample from Iran (sample Chalow3, Fig. 5F),
which indicates contact ~2050 BCE at the
Todd et al., Science 377, 11721180 (2022) 9 September 2022 4of9
Fig. 2. Haplotype sharedness and phylogenetic placement of ancient
European donkeys. (A) Haplotype sharedness clustering of modern (n= 168) and
ancient donkeys (n= 31) reconstructed using fineSTRUCTURE (35). Modern
domesticates are colored following Fig. 1, and ancient individuals are numbered
according to Fig. 3A. Cluster supports are shown in as probabilities on each node
if >0.8. MV242 placement is incongruent with Treemix (Fig. 2E) because of the
limited representation of divergent ancestries in the modern reference panel
used for imputation. (B) Co-ancestry matrix based on haplotype sharedness.
Co-ancestry values are averaged for co-clustered individuals. (Cto E) Treemix
phylogenies of three ancient specimens shown in black (C: Chalow3, D: Doshan
Tepe, and E: MV242) placed within the subpopulations defined in Fig. 1C (27).
Branches that are not scaled are shown as dashed lines. s.e., standard error.
RESEARCH |RESEARCH ARTICLE
Downloaded from https://www.science.org at bibCNRS INSB on September 10, 2022
latest. It was, however, absent in Acemhöyük
at that time, which suggests that the MV242
divergent lineage ranged into Eastern Iran/
Turkmenistan, but not Turkey. This lineage
also left genetic material in modern Anatolia,
the Levant, Nubia, andMaghreb,butnotin
Western Africa, which is consistent with don-
keys carrying MV242-related ancestry flowing
back into some African regions. In addition,
this ancestry was also present in Southwest-
ern European subpopulations (Canary Islands,
Portugal, and Spain), but neither in the mod-
ern Balkans and Ireland nor in any ancient
European sample analyzed here (Fig. 5, E and
F). Combined, our results suggest a range for
the MV242-related lineage from the Levant
into Asia rather than Europe and Africa.
Despite its divergent genetic makeup, MV242
carries a mitochondrial haplotype character-
istic of Clade II (Fig. 5A). Our tip-calibrated
coalescent analyses revealed that the time to
the most common recent ancestor of that
Clade was 32,226 BCE and not 332,580 to
142,980 BCE (Fig. 5B), as previously reported
(12,44). Because the same holds true for
Clade I, both clades could have coexisted in
sympatry 25,000 years later when donkeys
were first domesticated (Fig. 5B). Additionally,
no phylogeographic structure is apparent in
patterns of mitochondrial variation, both in
modern and ancient subpopulations, as ancient
specimens from Asia and Europe, sometimes
from the same archaeological sites, were placed
across both clades (Fig. 5A). Y-chromosomal
variation was also associated with little, if any,
population structure (Fig. 5, C and D). Com-
bined, our results dismiss mitochondrial DNA
and the Y-chromosome as reliable markers of
domestication history in donkeys.
Todd et al., Science 377, 11721180 (2022) 9 September 2022 5of9
Fig. 3. Ancient donkey dataset, genetic affinities to outgroups, and
modern donkeys. (A) Geographical distribution, estimated age, and sample
names of ancient donkeys (n= 31). Pie charts represent the proportion
of individuals with dun coat color (white), heterozygotes (gray), and derived
coat color (red) at each site. Genotype probabilities 0.99 are denoted
with ** and 0.9 with *. (B) Heatmap displaying outgroup f3-statistics in the
form of (modern, ancient; kiang) (58). Bar charts represent the proportion
of wild ancestry (kiang, onager, zebra, E.a.som) in each ancient individual
with standard errors estimated from ADMIXTURE with 100 bootstrap pseudo-
replicates (56). (Cto E) Treemix phylogenies of ancient specimens from
three archaeological sites shown in black [C: Tarquinia (Tarquinia214,
Tarquinia501), D: Bourse (BourseB, BourseC), E: Albufeira 1x1] placed within
the subpopulations defined in Fig. 1C (27). Branches that are not scaled are
shown as dashed lines. s.e., standard error.
RESEARCH |RESEARCH ARTICLE
Downloaded from https://www.science.org at bibCNRS INSB on September 10, 2022
Romans bred improved donkeys for producing
mules essential for their military power
and economy
Beyond documenting domestication history
at the global scale, our genomic dataset also
included three jennies (females) and six jacks
(males) from the same archaeological site
(Boinville-en-Woëvre) (Fig. 3A). These were
found in a dedicated farming area of a Roman
villa, providing insights into local management
practices in Roman Northern France (200 to
500 CE). One jack (sample GVA349) appeared
particularly inbred, with long ROH indicative
of recent consanguinity (Fig. 4A), and was ge-
netically related to four jacks and one jenny
(family group GVA1, including samples GVA125,
GVA347, GVA348, GVA349, GVA353, and
GVA354; table S10). Additionally, two jennies
showed genetic relatedness coefficients equiv-
alent to full siblings (family group GVA2, in-
cluding samples GVA355 and GVA358; table
S10). This indicates breeding management
within close kin, potentially aimed at select-
ing for desirable traits. Genotype imputation
at TBX3 (13) revealed the presence of dun and
derived colored coats, but no evidence for the
dominant alleles associated with white spots
or long hair was found in the sequence align-
ments at KIT (45)andFGF5 (fig. S11 and tables
S7 to S9) (46). The latter two phenotypes are,
however, relatively common in modern breeds
from France, which suggests post-Roman se-
lection for these traits.
The abundance of donkeys at Boinville-en-
Woëvre stands as an exception in Roman
France, as mules dominated all other assemb-
lagesfromthistime(47). Contemporaneous
Roman sites report mules of a large and uni-
form size, indicating selective breeding in
the parental species for expensive animals of
exceptional stature [Varro (2, 6)] (48). Mor-
phometric m easurements previously revealed
five donkeys from family group GVA1 as giant
(148 to 156 cm at the withers) (47). We found
that sample GVA359 had a similarly large size
Todd et al., Science 377, 11721180 (2022) 9 September 2022 6of9
Fig. 4. Inbreeding in domestic donkeys and horses. (A) Distribution of total ROH length in modern versus ancient donkeys. (B) Total length of ROH in donkey
genomes through time. (Cand D) Same as (A) and (B), but for 79 modern and 75 ancient horses. ROH tracts were identified using the program ngsF-HMM (59).
RESEARCH |RESEARCH ARTICLE
Downloaded from https://www.science.org at bibCNRS INSB on September 10, 2022
(144 cm) and genetic affinities to Western
Africa. This may indicate restocking to enhance
body size from distant bloodlines carrying
divergent ancestry or from wild populations.
Outgroup admixture was significantly higher
at Boinville-en-Woëvre than in other ancient
donkeys except the divergent MV242 specimen
(p= 0.045). Significantly negative f4(kiang,
MV242; Fiumarella1, Boinville-en-Woëvre)
statistics support restocking into family group
GVA1 only, from a lineage more divergent
than MV242 (Fig. 5G). Additionally, f4(kiang,
E.a.som; Fiumarella1, GVA1) statistics reject
unbalanced allele sharedness between E.a.som
andGVA1,rulingoutrestockingfromE.a.som
or more divergent populations (Fig. 5H). Com-
bined, these findings uncover a lineage, phylo-
genetically intermediate between MV242 and
E.a.som, that contributed to the genetic makeup
of some Roman donkeys at Boinville-en-Woëvre.
Together with the evidence of genetic related-
ness and inbreeding, this suggests Boinville-en-
Woëvre as a likely mule production center that
maintained the bloodlines of giant donkeys
Todd et al., Science 377, 11721180 (2022) 9 September 2022 7of9
Fig. 5. Uniparental marker phylogenies and introgression of divergent
lineages. (A) Mitochondrial phylogeny constructed using IQ-TREE (60)with100
bootstrap pseudo-replicates marked with a black triangle if >90%. (B) Posterior
distributions of the time to the most recent common ancestors of all mitochondrial
haplotypes, Clade I and Clade II labeled with their modes. (Cand D)Sameas(A)
and (B) for the Y-chromosome. (Eto H)f4-statistics(58)exploringthegenetic
contribution of divergent lineages into modern and ancient donkeys. Z scores were
corrected for multiple testing, and red bars with asterisks show p<0.05.
RESEARCH |RESEARCH ARTICLE
Downloaded from https://www.science.org at bibCNRS INSB on September 10, 2022
that were selected through familial breeding
and restocking. This center may illustrate how
Romans sustained the eno rmous demand for
mules, which is documented in the nearby
Rhine frontier(49), and fueled transportation
networks throughout the Empire (47).
Discussion
Our study solves long-standing debates about
donkey domestication. We support domesti-
cation starting from a unique African source
~5000 BCE. Donkeys subsequently spread
into Eurasia from ~2500 BCE, and Central and
Eastern Asian subpopulations differentiated
~2000 to 1000 BCE. Genetic affinities char-
acteristic of modern Western Europe were
already formed by 500 BCE. After early do-
mestication, African donkeys further differ-
entiated in the West and the Horn of Africa
plus Kenya but also received streams of ge-
netic ancestry from Western Europe as well as
a region encompassing the Levant, Anatolia,
and Mesopotamia. Donkey domestication in-
volved limited but detectable wild introgres-
sion. It did not entail inflated inbreeding in
recent times, in contrast to horses. The pro-
cesses of donkey and horse domestication
differed substantially, as horses were domes-
ticated twice (50) and rapidly spread across
Eurasia from the lower Don-Volga region
~2000 BCE (21). Their regional differentia-
tion remained relatively limited owing to
strong connectivity at continental distances
early on and until oriental bloodlines were
propagated throughout the world during the
past 1000 years (41,51). The extent to which
the different domestication trajectories of
donkeys and horses were only driven by their
respective roles in human societies or also re-
flected management practices adapted to their
respective mating and social behavior (52)re-
mains to be explored.
This work clarifies global patterns of donkey
domestication and movements but also high-
lights many directions for future research. For
example, it remains unknown whether domes-
tic donkeys only dispersed out of Africa by
land through the Sinai Peninsula or across
the Red Sea from Ethiopia to Yemen. Addi-
tionally, modern subpopulations from the Horn
of Africa plusKenya were found to be the first
expanding. This may suggest early domesti-
cation there or that donkeys domesticated
elsewhere in Africa entering the region more
recently. Further research is needed to clarify
the timing of pastoral spread into the Red Sea
Sudanese region and the Horn of Africa. Cur-
rent dates range from ~2500 BCE in Ethiopia
and Eritrea (53)to~3000BCEinNorthern
Kenya (54). Donkeys are not present in the
archaeological record of Western Africa before
the beginning of the common era, either (55),
which postdates by 3000 years the time when
donkey populations from the Horn of Africa
plus Kenya and Western Africa are inferred
to have split genetically. This may indicate an
early yet undocumented arrival in the region
or a slow migration westward, only reaching
the modern range later. Improving the current
African archaeological record thus appears
paramount to refining the exact context un-
derlying early donkey domestication and sub-
sequent population movements.
Further genomic studies in other regions
would also largely benefit the understanding
of donkey diversity and history. Resolving the
genetic structure of equine remains from the
third millennium BCE of Southwest Asia will
be challenging because of postmortem DNA
decay but essential to mapping the geographic
range of the divergent lineage identified here
(sample MV242), as well as to understanding
dispersal mechanisms in greater detail. The
same holds true for Chalcolithic and Bronze
Age Europe, which remain genetically undocu-
mented in our dataset, and onwards. Develop-
ing genetic knowledge of ancient European
donkeys will further clarify patterns of exchange
across the Mediterranean region, including
during and after Roman times, as revealed in
this study. It will also provide insights into
the dispersal mechanisms underpinning the
genetically supported presence of donkey
remains in Portugal ~2200 BCE (33). Genetic
characterization of local archaeological sites at
the population scale may uncover additional
mule breeding centers other than the one re-
ported here. This will shed light on the diver-
sity of breeding management strategies that
were developed by Romans to supply their
continental-wide economy and military with
adequate animal resources (49). For now, both
the absence of mules and the rarity of horse
mares at Boinville-en-Woëvre (47)suggestthat
mares were brought in for mating before re-
turning pregnant to their owners. Alternatively,
donkey breeders may have visited other farms
with their jacks to cover mares.
Efforts should continue to characterize the
modern donkey diversity around the world,
especially in Saudi Arabia, which is currently
characterized by a single individual, as well
as in Africa, for which no populations located
south of the Equator have been sampled. Such
efforts may not only refine the historical leg-
acy of past populations into the modern world
but also uncover the genetic basis of desert
adaptations, which could prove invaluable for
future donkey breeding in the face of global
warming.
REFERENCES AND NOTES
1. P. Mitchell, The Donkey in Human History: An Archaeological
Perspective (Oxford Univ. Press, 2018).
2. S. L. Norris, H. A. Little, J. Ryding, Z. Raw, PLOS ONE 16,
e0247830 (2021).
3. B. Kimura, F. Marshall, A. Beja-Pereira, C. Mulligan,
Afr. Archaeol. Rev. 30,8395 (2013).
4. F. Marshall, Rethinking Agriculture: Archeological and
Ethnoarcheological Perspectives (Left Coast Press, 2007).
5. S. Bokonyi, in Studi di paletnologia in onore di Salvatore M.
Puglisi, M. Liverani, A. Palmieri, R. Peroni, Eds. (Università di
Roma La Sapienza, 1985), pp. 495499.
6. J. Boesneck, A. Von den Driesch, in El Omari: A Neolithic
Settlement and Other Sites in the Vicinity of Wadi Hof, Helwan,
F. Debono, B. Mortensen, Eds. (Philipp von Zabern, 1990),
pp. 99107.
7. J. Boesneck, A. von den Driesch, R. Ziegler, in Maadi III: The
Non-lithic Small Finds and the Structural Remains of
the Predynastic Settlement, I. Rizkana, J. Seeber, Eds.
(Philipp von Zabern, 1989), pp. 87128.
8. K. M. Cialowicz, Les palettes Egyptiennes aux motifs
zoomorphes et sans decoration. Etudes de l'art predynastique
(Jagiellonian University, 1991).
9. D. J. Brewer, D. B. Redford, S. Redford, Domestic Plants and
Animals: The Egyptian Origins (Aries and Phillips, 1994).
10. S. Rossel et al., Proc. Natl. Acad. Sci. U.S.A. 105, 37153720
(2008).
11. A. Beja-Pereira et al., Science 304, 1781 (2004).
12. B. Kimura et al., Proc. Biol. Sci. 278,5057 (2011).
13. C. Wang et al., Nat. Commun. 11, 6014 (2020).
14. M. Cattani, S. Bokonyi, in Essays on the Late Prehistory of the
Arabian Peninsula. Serie Orientale Romana XCIII, S. Cleuziou,
T. Maurizio, Z. Juris, Eds. (Istituto Italiano per l'Africa e l'Oriente,
2002).
15. R. H. Meadow, H. P. Uerpmann, Equids in the Ancient World,
vol. 2 (Dr Ludwig Reichert Verlag, 1991).
16. J. Clutton-Brock, Mammal Rev. 22,7985 (1992).
17. J. Zarins, R. Hauser, The domestication of equidae in
third-millennium BCE Mesopotamia. Cornell University Studies
in Assyriology and Sumerology 24, XI + 432 (2014).
18. E. Vila, in Equids in Time and Space: Papers in Honour of Véra
Eisenmann, M. Mashkour, Ed. (Oxbow, 2006), pp. 101123.
19. J. Boessneck, A. von den Driesch, U. Steger, Baghdader
Mitteilingen 15,149189 (1984).
20. E. A. Bennett et al., Sci. Adv. 8, eabm0218 (2022).
21. P. Librado et al., Nature 598, 634640 (2021).
22. S. Rosenbom et al., Mol. Phylogenet. Evol. 85,8896 (2015).
23. G. Renaud et al., Sci. Adv. 4, eaaq0392 (2018).
24. H. Jónsson et al., Proc. Natl. Acad. Sci. U.S.A. 111,
1865518660 (2014).
25. L. Zeng et al., Zool. Res. 42, 450460 (2021).
26. A. L. Price et al., Nat. Genet. 38, 904909 (2006).
27. J. K. Pickrell, J. K. Pritchard, PLOS Genet. 8, e1002967 (2012).
28. J. Terhorst, J. A. Kamm, Y. S. Song, Nat. Genet. 49, 303309
(2017).
29. D. Reese, Mediterr. Archaeol. Archaeom. Int. J. 13, 289320
(2013).
30. B. A. Knapp, Biblic. Archaeol. 55,5272 (1992).
31. S. Ratnagar, Trading Encounters: from the Euphrates to the
Indus in the Bronze Age (Oxford Univ. Press, 2004).
32. M. A. Zeder, in Equids in the Ancient World, R. H. Meadow,
H. P. Uerpmann, Eds. (Reichert, 1986), pp. 164193.
33. J. L. Cardoso, J. T. Vilstrup, V. Eisenmann, L. Orlando,
J. Archaeol. Sci. 40, 44834490 (2013).
34. S. Amiri, M. Mashkour, F. Mohaseb, R. Naseri, in Archaeozoology
of Southwest Asia and Adjacent Areas XIII: Proceedings of the
Thirteenth International Symposium, University of Cyprus,
Nicosia, Cyprus, June 7-10, 2017, J. Daujat, A. Hadjikoumis,
R. Berthon, J. Chahoud, V. Kassianidou, J. D. Vigne, Eds. (2021),
pp. 199219.
35. D. J. Lawson, G. Hellenthal, S. Myers, D. Falush, PLOS Genet. 8,
e1002453 (2012).
36. L. Girdland Flink et al., Proc. Natl. Acad. Sci. U.S.A. 111,
61846189 (2014).
37. L. A. F. Frantz, D. G. Bradley, G. Larson, L. Orlando,
Nat. Rev. Genet. 21, 449460 (2020).
38. R. Hui, E. DAtanasio, L. M. Cassidy, C. L. Scheib, T. Kivisild,
Sci. Rep. 10, 18542 (2020).
39. B. L. Browning, S. R. Browning, Am. J. Hum. Genet. 98,
116126 (2016).
40. T. N. Kristensen, A. C. Sørensen, Anim. Sci. 80, 121133 (2007).
41. A. Fages et al., Cell 177, 14191435.e31 (2019).
42. F. Marshall, L. Weissbrod, in Tracking Down the Past.
Ethnohistory Meets Archaeozoology, G. Grupe, J. Peters,
G. McGlynn, Eds. (M. Leidor, 2009),
pp. 5979.
43. E. A. Bennett et al., PLOS ONE 12, e0174216 (2017).
44. L. Wang et al., Front. Genet. 12, 759831 (2021).
45. B. Haase, S. Rieder, T. Leeb, Anim. Genet. 46, 321324 (2015).
46. R. Legrand, L. Tiret, M. Abitbol, Genet. Sel. Evol. 46,65
(2014).
47. S. Lepetz et al., J. Archaeol. Sci. Rep. 40, 103250 (2021).
Todd et al., Science 377, 11721180 (2022) 9 September 2022 8of9
RESEARCH |RESEARCH ARTICLE
Downloaded from https://www.science.org at bibCNRS INSB on September 10, 2022
48. G. K. Kunst, Anthropozoologica 31, 109118 (2000).
49. C. Johnstone, in Feeding the Roman Army: The Archaeology of
Production and Supply in NW Europe, S. Stalibrass, R. Thomas,
Eds. (Oxbow Books, 2008), pp. 128145.
50. C. Gaunitz et al., Science 360, 111114 (2018).
51. S. Felkel et al., Sci. Rep. 9, 6095 (2019).
52. F. B. Marshall, K. Dobney, T. Denham, J. M. Capriles, Proc. Natl.
Acad. Sci. U.S.A. 111, 61536158 (2014).
53. J. Lesur, E. A. Hildebrand, G. Abawa, X. Gutherz, Quat. Int. 343,
148158 (2014).
54. E. A. Hildebrand et al., Proc. Natl. Acad. Sci. U.S.A. 115,
89428947 (2018).
55. K. MacDonald, R. Hutton MacDonald, in The Origins and
Development of African livestock: Archaeology, Genetics,
Linguistics and Ethnography, R. M. Blench, K. MacDonald,
Eds. (University College London Press, 2000), pp. 89.
56. D. H. Alexander, K. Lange, BMC Bioinformatics 12, 246
(2011).
57. N. Patterson, A. L. Price, D. Reich, PLOS Genet. 2, e190 (2006).
58. N. Patterson et al., Genetics 192, 10651093 (2012).
59. F. G. Vieira, A. Albrechtsen, R. Nielsen, Bioinformatics 32,
20962102 (2016).
60. B. Q. Minh et al., Mol. Biol. Evol. 37, 15301534 (2020).
ACKNOWL EDGME NTS
Funding: L.K.H. thanks the director of the Tel es-Safi/Gath
excavation, A. Maeir, for facilitating sampling. The DonkeyBank
collection of modern donkey DNA samples is supported by the
European Unions Horizon 2020 Research and Innovation
Programme (grant agreement no. 857251). L.P.P. is funded by
national funds from FCT Fundação para a Ciência e a Tecnologia,
I.P. Genoscope and GeT-PlaGe sequencing platforms are partly
funded by France Génomique National infrastructure, funded
as part of Investissement davenirprogram managed by
Agence Nationale pour la Recherche (contract ANR-10-INBS-09),
and their Grands Projetsfunding instrument (BUCEPHALE). This
project has received funding from the CNRS, University
Paul Sabatier (AnimalFarm IRP), and the European Research Council
(ERC) under the European Unions Horizon 2020 research and
innovation program (grant agreements 885729-AncestralWeave,
295729-CodeX, 853272-PALAEOFARM, and 681605-PEGASUS).
Author contributions: Conceptualization: L.O.; Materials and
Reagents: J.-M.A., A.P., P.W., O.B., C.D., and L.O.; Archaeological
samples and contextual information: M.V., E.V., M.Marzullo, O.P.,
U.T., G.B.G., A.T., V.T., F.A., J.L.C., M.J.V., M.T.A., N.B., A.G., L.K.H.,
A.Ö., B.S.A., V.O., B.C., S.L., A.A.V., H.D., A.M., M.Mashkour, and
L.O.; Sampled modern donkeys: A.B.-P., D.-D.W., L.P.R., and S.A.B.;
Radiocarbon dating: J.S.; DNA sequencing and preparation: L.T.C.,
L.C., S.S., A.F., A.S.-O., P.C., N.K., L.P.P., J.M.A., A.P., O.B., C.D.,
P.W., A.B.-P., and L.O.; Data analysis: E.T.T. and L.O. with input from
P.L. and H.R.; WritingSupplementary Information: E.T.T., with
input from L.O.; WritingMain Article: E.T.T. and L.O. with input
from A.F., L.P.R., P.L., E.V., F.A., L.A.F., B.S., L.K.H., B.S.A.,
M.Mashkour, S.A.B., A.B.P., and all coauthors. Competing interests:
The authors declare that they have no competing interests. Data
and materials availability: The sequence data generated in this
study are availablefor download at the European Nucleotide Archive
(accession no. PRJEB52849). The variant data for this study
have been deposited in the European Variation Archive (EVA) at
EMBL-EBI under accession number PRJEB55549. The recombination
map can be downloaded from https://osf.io/k7x8b/. The accession
numbers for each individual sample and all other data used
in this study are included in tables S1, S2, and S11 of the
supplementary materials. License information: Copyright © 2022
the authors, some rights reserved; exclusive licensee American
Association for the Advancement of Science. No claim to original
US government works. https://www.science.org/about/science-
licenses-journal-article-reuse
SUPPLEMENTARY MATERIALS
science.org/doi/10.1126/science.abo3503
Materials and Methods
Figs. S1 to S15
Tables S1 to S11
References (61136)
MDAR Reproducibility Checklist
View/request a protocol for this paper from Bio-protocol.
Submitted 28 January 2022; accepted 15 August 2022
10.1126/science.abo3503
Todd et al., Science 377, 11721180 (2022) 9 September 2022 9of9
RESEARCH |RESEARCH ARTICLE
Downloaded from https://www.science.org at bibCNRS INSB on September 10, 2022
Use of this article is subject to the Terms of service
Science (ISSN ) is published by the American Association for the Advancement of Science. 1200 New York Avenue NW, Washington, DC
20005. The title Science is a registered trademark of AAAS.
Copyright © 2022 The Authors, some rights reserved; exclusive licensee American Association for the Advancement of Science. No claim
to original U.S. Government Works
The genomic history and global expansion of domestic donkeys
Evelyn T. ToddLaure Tonasso-CalvièreLoreleï ChauveyStéphanie SchiavinatoAntoine FagesAndaine Seguin-
OrlandoPierre ClavelNaveed KhanLucía Pérez PardalLaura Patterson RosaPablo LibradoHarald RingbauerMarta
VerdugoJohn SouthonJean-Marc AuryAude PerdereauEmmanuelle VilaMatilde MarzulloOrnella PratoUmberto
TecchiatiGiovanna Bagnasco GianniAntonio TagliacozzoVincenzo TinèFrancesca AlhaiqueJoão Luís CardosoMaria
João ValenteMiguel Telles AntunesLaurent FrantzBeth ShapiroDaniel G. BradleyNicolas BoulbesArmelle GardeisenLiora
Kolska HorwitzAliye ÖztanBenjamin S. ArbuckleVedat OnarBenoît ClavelSébastien LepetzAli Akbar VahdatiHossein
DavoudiAzadeh MohasebMarjan MashkourOlivier BouchezCécile DonnadieuPatrick WinckerSamantha A. BrooksAlbano
Beja-PereiraDong-Dong WuLudovic Orlando
Science, 377 (6611), • DOI: 10.1126/science.abo3503
Donkeys’ African origins
Donkeys have been important to humans for thousands of years, being the primary source of work and transport for
many cultures. Unlike horses, little was known about the origin and domestication of donkeys. Todd et al. sequenced
the genomes of modern and ancient donkeys and found evidence of an eastern African origin over 7000 years
ago, with subsequent isolation and divergence of lineages in Africa and Eurasia. They also reveal the imprint of
desertification on divergence among groups and specifics about donkey breeding and husbandry, including selection
for large size and the practice of inbreeding. —SNV
View the article online
https://www.science.org/doi/10.1126/science.abo3503
Permissions
https://www.science.org/help/reprints-and-permissions
Downloaded from https://www.science.org at bibCNRS INSB on September 10, 2022
... Together, these advancements have enabled ancient genomes to be generated from a range of domestic animals, including dogs (Bergström et al., 2020), cats , pigs (Frantz et al., 2019), horses (Librado et al., 2021), donkeys (Todd et al., 2022), cattle (Verdugo et al., 2019), sheep (Rossi et al., 2021), and goats , and their wild progenitors as far back as half a million years (e.g., Orlando et al., 2013). These high resolution datasets, alongside traditional analyses of the archaeological record (especially material culture and zooarchaeological context), and extensive modern genomic databases (e.g., Dog10K; Meadows et al., 2023), have helped track the geographic and temporal origin of domestic lineages (e.g., horses; Librado et al., 2021), revealing the inextricably linked evolutionary histories of human cultures and their domestic animals, especially in dogs (Bergström et al., 2020). ...
... Mules, the hybrid offspring of a male donkey and female horse, were essential to the Roman economy and military, given their stature and endurance. As a result, bloodlines of giant donkeys were maintained for the purpose of mule production during this period (Todd et al., 2022). In Britain, interbreeding between local wild aurochs and ancestral landrace cattle was permitted by early herders (Park et al., 2015). ...
Chapter
The increased availability of ancient nuclear genomes has helped to illuminate the complexity of genetic evolution and overcome the difficulties inherent in inferring the past from the present. This “nuclear revolution” has challenged and overturned long-held assumptions about the domestication process by demonstrating that many key domestic traits were not under selection in the early stages of management, and that reproductive isolation is not necessary to maintain domestic phenotypes.
... The donkey (Equus asinus), is a remarkable member of the Equidae family that has played a pivotal role in shaping human civilization since its domestication approximately 5000 years ago (Todd et al. 2022). As a reliable beast of burden, donkeys have facilitated long-distance commerce and movement across a broad spectrum of terrains, including semi-arid and upland regions with limited food and water availability. ...
Article
Full-text available
Donkeys (Equus asinus) have played a vital role in agriculture, transportation, and companionship, particularly in developing regions where they are indispensable working animals. The domestication of donkeys marked a significant turning point in human history, as they became essential for transportation, agriculture, and trade, especially in arid and semi-arid areas where their resilience and endurance were highly valued. In modern society, donkeys are indispensable due to their diversified applications, including meat, dairy, medicine, and functional bioproducts, supporting economic, cultural, and medical industries. Despite their critical importance, research on donkeys has historically been overshadowed with studies on horses. However, recent advancements in high-throughput sequencing and bioinformatics have significantly deepened our understanding of the molecular landscape of donkey genome, uncovering their unique adaptations, genetic diversity, and potential therapeutic applications. Microsatellite and mitochondrial DNA (mtDNA) markers have proven effective in assessing the genetic diversity of donkeys across various regions of the world. Additionally, significant strides have been made in characterizing differentially abundant genes, proteins, and metabolic profiles in donkey milk, meat, and skin, and in identifying specific genes/proteins/metabolites associated with sperm quality, motility, and reproduction. Advanced genomic technologies, such as genome-wide association studies and the identification of selection signatures, have also been instrumental in delineating genomic regions associated with phenotypic and adaptive traits. This review integrates data from diverse studies, including those on genetic diversity, transcriptomics, whole genome sequencing, protein analysis, and metabolic profiling, to provide a comprehensive overview of donkey biology. It underscores the unique characteristics of donkeys and emphasizes the importance of continued research to improve their genetic management, conservation, and agricultural use, ensuring their ongoing contribution to human societies.
... Ancient DNA (aDNA, the analysis of DNA molecules preserved in remains of ancient organisms) has extensively advanced insights into many evolutionary and ecological questions, including the demographic dynamic of human populations [1,2,3,4], the adaptation and domestication of animals [5,6,7,8], and the coevolutionary trajectories of pathogens [9,10]. However, aDNA research has thus far relied predominantly on the isolation of DNA from well-preserved fossilised specimens, typically bone, hair, and teeth. ...
Preprint
Full-text available
Ancient environmental DNA (aeDNA) is now commonly used in paleoecology and evolutionary ecology, yet due to difficulties in gaining sufficient genome coverage on individual species from metagenome data, its genetic perspectives remain largely uninvestigated. Hybridization capture has proven as an effective approach for enriching the DNA of target species, thus increasing the genome coverage of sequencing data and enabling population and evolutionary genetics analysis. However, to date there is no tool available for designing capture probe sets tailored for aeDNA based population genetics. Here we present eProbe, an efficient, flexible and easy-to-use program toolkit that provides a complete workflow for capture probe design, assessment and validation. By benchmarking a probe set for foxtail millet, an annual grass, made by the eProbe workflow, we demonstrate a remarkable increase of capturing efficiency, with the target taxa recovery rate improved by 577-fold, and the genome coverage achieved by soil capture-sequencing data even higher than data directly shotgun sequenced from the plant tissues. Probes that underwent our filtering panels show notably higher efficiency. The capture sequencing data enabled accurate population and evolutionary genetic analysis, by effectively inferring the fine-scale genetic structures and patterns, as well as the genotypes on functional genes.
... Feral burros (Equus asinus) are found in a variety of environments worldwide including the United States. They were first domesticated approximately 5000 BC in Northeast Africa from African wild ass (Equus africanus) (Todd et al., 2022), resulting in donkeys, which were subsequently introduced to North America by Spanish explorers in the 1530s (Protsch & Berger, 1973;Seegmiller & Ohmart, 1981). Donkeys, also called burros, were widely used as pack animals in the southwest United States. ...
Article
Full-text available
Animals adjust their habitat use patterns in response to changes in their physiological needs and environmental conditions. Understanding the mechanisms underlying resource selection and space use across time and space reveals effects of the environment on animals' decisions. We explored the effects of habitat availability and heterogeneity on the seasonal and annual space use and resource selection of two free‐roaming feral burro (Equus asinus) populations in the United States within distinct climate and habitat conditions: the Sonoran Desert and the Colorado Plateau. As an introduced yet protected species in the United States, understanding burros' interactions with habitat elements is important for their conservation and management, as well as the conservation of sympatric wildlife. We used GPS locations of female burros (72 animals across both study areas) to delineate annual and seasonal ranges and resource selection patterns. We evaluated effects of mean and CV of habitat covariates, including forage, distance to water, and topography, representing availability and heterogeneity of resources, on seasonal and annual range size of burros. Moreover, we explored how burro seasonal and annual resource selection patterns were affected by availability and heterogeneity of resources. In the Sonoran Desert study area, burros had smaller seasonal and annual ranges and constant resource selection patterns across a year, likely due to a freshwater lake in the area, making water a nonlimiting resource. Human presence was the greatest factor affecting range size and resource selection in the Sonoran Desert, where burros selected for areas near roads and human recreation. In the Colorado Plateau study area, where resources were more seasonal, we found larger range sizes and fluctuating resource selection patterns compared to the Sonoran Desert population. Spatial variation in forage, water, and topography significantly affected range size of burros inhabiting the Colorado Plateau study area. Productive habitats with available water support smaller ranges and a more consistent pattern of resource selection. Our results highlight the positive effect of habitat heterogeneity and the negative effect of habitat productivity on range size of animals. Our findings contribute to an improved understanding of habitat requirements for free‐roaming burros that currently live under various climate and habitat conditions globally.
... The animal has proved amazingly durable as adopted consort for all kinds of human effort to "make hay" of their environment. Deriving from pre-historic wild ancestors in North Africa, now numbering some 40 million world-wide, first domesticated around 3000 BCE, the burro has proved crucial as partner especially to the poor throughout history (Todd et al. 2022;Rossel et al. 2008;Corbett 2005, 311). ...
Chapter
And then another meander of thought in which we hunker down on Jesus’ own seminal choice for an ass-consort, when taking his posse straight into the central nation-state “lair” for a final showdown (Mk 11:1–11 and the other gospel treatments). Chapter 11 goes deep into this culminating episode in the Nazarene’s confrontation with the Powers, conjuring the context as a public battle of disparate parades—Jesus-on-a-burro coming from the east with Passover pilgrimage crowds faced off in the space of urban Jerusalem against Pilate-on-a-horse with equestrian legion coming from the west, strutting might and militancy. At stake is the meaning of the monarchical evocations entertained. Is the Galilean prophet in quest of provoking a Maccabean-like revolt under the aegis of a Davidic Redivivus? Or is his orchestration subtly pillorying not only Roman pretension, but also the historic Israelite choice for royal pomp and wealth betraying Israel’s original vocation (I Sam 8:4–22; 10:17–19; 12:1–19)—and doing so by way of a “ridden riddle”? Did Jesus’ choice for a common burro over a royal mule subtly signal Israel’s original MO as a “kingdom” (!) of “priests” recovering nomadic autonomy and localized savvy learned from Sinai sands and Canaanite hill country (Exod 19:6)? The tack here reads the messianic theatrics as parabolic polemics: riding down Temple-State ostentation by adopting an ancient convention of solid “positioning” on the back of the long-time “friend” whose particular abilities had long enabled human viability in that otherwise harsh ecology. And it begs reading in relationship to archeo-shards of memory of our species’ first self-depictions as far back as (17,000 BCE) Lascaux Cave paintings of magnificent animal compositions flanked by miniature and minimalist human depictions as well as later (13,000 BCE) Trois-Freires Sorcerer evocations of human-animal composites—an owl-eyed, horse-bear-human-deer-bison eco-being in true-to-form rendition, not yet self-isolating in the delusion and solipsism of civilization and its aftermath. What if the central thrust of this final epiphany is indeed a recovered monstrosity—a “God” becoming incarnate in history not merely in anthropocentric anatomy, but actually a burro-human hybridity generic and genuine to that particular local ecozone and significant of localized symbiosis in whatever unique partnerships it marshals in quite varied ecozones?
... Many assisted-migrant populations retain genetic diversity lost from their native conspecific populations (Bradshaw et al., 2006;Marchesini et al., 2021;Todd et al., 2022). Genetic diversity is itself an element of biodiversity, considered by many to have intrinsic value (Crozier, 1997); is important for future evolutionary dynamics; and may be a lifeline to the persistence of these species globally (Booy et al., 2000). ...
Article
Full-text available
International and national conservation policies almost exclusively focus on conserving species in their historic native ranges, thus excluding species that have been introduced by people and some of those that have extended their ranges on their own accord. Given that many of such migrants are threatened in their native ranges, conservation goals that explicitly exclude these populations may overlook opportunities to prevent extinctions and respond dynamically to rapidly changing environmental and climatic conditions. Focusing on terrestrial mammals, we quantified the number of threatened mammals that have established new populations through assisted migration (i.e., introduction). We devised 4 alternative scenarios for the inclusion of assisted‐migrant populations in mainstream conservation policy with the aim of preventing global species extinctions. We then used spatial prioritization algorithms to simulate how these scenarios could change global spatial conservation priorities. We found that 22% (70 species out of 265) of all identified assisted‐migrant mammals were threatened in their native ranges, mirroring the 25% of all mammals that are threatened. Reassessing global threat statuses by combining native and migrant ranges reduced the threat status of 23 species (∼33% of threatened assisted migrants). Thus, including migrant populations in threat assessments provides a more accurate assessment of actual global extinction risk among species. Spatial prioritization simulations showed that reimagining the role of assisted‐migrant populations in preventing species extinction could increase the importance of overlooked landscapes, particularly in central Australia, Europe, and the southwestern United States. Our results indicated that these various and nonexhaustive ways to consider assisted‐migrant populations, with due consideration of potential conservation conflicts with resident taxa, may provide unprecedented opportunities to prevent species extinctions.
... Donkeys likely underwent domestication twice, as indicated by the two mitochondrial DNA haplogroups among modern donkeys. Ancient DNA analyses indicate that one haplogroup descends from the Nubian wild ass (Equus africanus africanus) (Todd et al., 2022). Donkeys were important in facilitating trade networks and contributed to the mobility of pastoral groups. ...
Article
Full-text available
Between 2016 and 2021, the fish-salting tank 10 from Workshop 23, located in the Roman Ruins of Troia, was subjected to archaeological excavation. The workshop, situated in an area severely impacted by coastal erosion due to tidal activities in the Sado River estuary, remains in a significantly deteriorated state. Among the remains, tank 10, entirely covered by a sand dune and positioned in the southern corner of the workshop, is notable for its large size, with an estimated production capacity of approximately 40 m³. Preliminary analysis of the ceramic remnants suggests that there was activity at the tank until at least the end of the 2nd century CE, aligning with findings from adjacent tanks 7, 8 and 9. The faunal remains unearthed include a layer of fish preparations found directly above the tank's flooring, dating back to its final operational phase—yet to be analysed—and a considerable quantity of mammal bones, notably from two donkeys (Equus asinus). These animals were integral to the Roman world, primarily for transporting diverse cargo. Body part representation and taphonomic analyses of the recovered remains do not indicate accumulation due to natural processes, or as a result of post-consumption discarding. The two donkeys, found above the roof- and wall-collapsed levels, show no significant marks on the bone surfaces examined. Preliminary studies involving ZooMS and isotopic analysis of these remains provide critical insights into the presence of donkeys in Troia, marking the first identification of such a species among an extensive zooarchaeological collection gathered from various fish-salting workshops excavated since 2007.
Preprint
Full-text available
International and national conservation policies almost exclusively focus on conserving species in their historic native ranges, thus excluding species that have dispersed on their own accord or have been introduced by people. Given that many of these "migrant" species are threatened in their native ranges, conservation goals that explicitly exclude these migrant populations may overlook opportunities to prevent extinctions and respond dynamically to rapidly changing environmental and climatic conditions. Focusing on terrestrial mammals, we quantified the extent to which migration, in this case via introductions, has provided new homes for threatened mammal species. We then devised alternative scenarios for the inclusion of migrant populations in mainstream conservation policy with the aim of preventing global species extinctions and used spatial prioritization algorithms to simulate how these scenarios could change global spatial conservation priorities. We found that 22% of all identified migrant mammals (70 species) are threatened in their native ranges, mirroring the 25% of all mammals that are threatened. Reassessing global threat statuses by combining native and migrant ranges reduced the threat status of 23 species (~33% of threatened migrants). Thus, including migrant populations in threat assessments provides a more accurate assessment of actual global extinction risk among species. Spatial prioritization simulations showed that reimagining the role of migrant populations to prevent global species extinction could increase the importance of overlooked landscapes, particularly in central Australia. Our results indicate that these various and non-exhaustive ways to consider migrant populations, with due consideration for potential conservation conflicts with resident taxa, may provide unprecedented opportunities to prevent species extinctions. We present these alternatives and spatial simulations to stimulate discussion on how conservation ought to respond, both pragmatically and ethically, to rapid environmental change in order to best prevent extinctions.
Article
Full-text available
It has been forty years since the first edition of this book, as an Oriental Institute doctoral dissertation, was completed. Now, in a fully revised and much expanded study, CUSAS 24 presents a comprehensive discussion of the philological, historical, and archaeological evidence for the range of equidae known now from much of Western
Article
Full-text available
Seven donkey breeds are recognized by the French studbook. Individuals from the Pyrenean, Provence, Berry Black, Normand, Cotentin and Bourbonnais breeds are characterized by a short coat, while those from the Poitou breed (Baudet du Poitou) are characterized by a long-hair phenotype. We hypothesized that loss-of-function mutations in the FGF5 (fibroblast growth factor 5) gene, which are associated with a long-hair phenotype in several mammalian species, may account for the special coat feature of Poitou donkeys. To the best of our knowledge, mutations in FGF5 have never been described in Equidae. We sequenced the FGF5 gene from 35 long-haired Poitou donkeys, as well as from a panel of 67 short-haired donkeys from the six other French breeds and 131 short-haired ponies and horses. We identified a recessive c.433_434delAT frameshift deletion in FGF5, present in Poitou and three other donkey breeds and a recessive nonsense c.245G > A substitution, present in Poitou and four other donkey breeds. The frameshift deletion was associated with the long-hair phenotype in Poitou donkeys when present in two copies (n = 31) or combined with the nonsense mutation (n = 4). The frameshift deletion led to a stop codon at position 159 whereas the nonsense mutation led to a stop codon at position 82 in the FGF5 protein. In silico, the two truncated FGF5 proteins were predicted to lack the critical β strands involved in the interaction between FGF5 and its receptor, a mandatory step to inhibit hair growth. Our results highlight the allelic heterogeneity of the long-hair phenotype in donkeys and enlarge the panel of recessive FGF5 loss-of-function alleles described in mammals. Thanks to the DNA test developed in this study, breeders of non-Poitou breeds will have the opportunity to identify long-hair carriers in their breeding stocks.
Article
Full-text available
The estimation of individual ancestry from genetic data has become essential to applied population genetics and genetic epidemiology. Software programs for calculating ancestry estimates have become essential tools in the geneticist's analytic arsenal. Here we describe four enhancements to ADMIXTURE, a high-performance tool for estimating individual ancestries and population allele frequencies from SNP (single nucleotide polymorphism) data. First, ADMIXTURE can be used to estimate the number of underlying populations through cross-validation. Second, individuals of known ancestry can be exploited in supervised learning to yield more precise ancestry estimates. Third, by penalizing small admixture coefficients for each individual, one can encourage model parsimony, often yielding more interpretable results for small datasets or datasets with large numbers of ancestral populations. Finally, by exploiting multiple processors, large datasets can be analyzed even more rapidly. The enhancements we have described make ADMIXTURE a more accurate, efficient, and versatile tool for ancestry estimation.
Article
Full-text available
Genetic data from extant donkeys (Equus asinus) have revealed two distinct mitochondrial DNA haplogroups, suggestive of two separate domestication events in northeast Africa about 5000 years ago. Without distinct phylogeographic structure in domestic donkey haplogroups and with little information on the genetic makeup of the ancestral African wild ass, however, it has been difficult to identify wild ancestors and geographical origins for the domestic mitochondrial clades. Our analysis of ancient archaeological and historic museum samples provides the first genetic information on the historic Nubian wild ass (Equus africanus africanus), Somali wild ass (Equus africanus somaliensis) and ancient donkey. The results demonstrate that the Nubian wild ass was an ancestor of the first donkey haplogroup. In contrast, the Somali wild ass has considerable mitochondrial divergence from the Nubian wild ass and domestic donkeys. These findings resolve the long-standing issue of the role of the Nubian wild ass in the domestication of the donkey, but raise new questions regarding the second ancestor for the donkey. Our results illustrate the complexity of animal domestication, and have conservation implications for critically endangered Nubian and Somali wild ass.
  • S L Norris
  • H A Little
  • J Ryding
  • Z Raw
S. L. Norris, H. A. Little, J. Ryding, Z. Raw, PLOS ONE 16, e0247830 (2021).
  • B Kimura
  • F Marshall
  • A Beja-Pereira
  • C Mulligan
B. Kimura, F. Marshall, A. Beja-Pereira, C. Mulligan, Afr. Archaeol. Rev. 30, 83-95 (2013).
Les palettes Egyptiennes aux motifs zoomorphes et sans decoration
  • K M Cialowicz
K. M. Cialowicz, Les palettes Egyptiennes aux motifs zoomorphes et sans decoration. Etudes de l'art predynastique (Jagiellonian University, 1991).