Content uploaded by Cindy Gilda Santander
Author content
All content in this area was uploaded by Cindy Gilda Santander on Jun 26, 2019
Content may be subject to copyright.
Full Terms & Conditions of access and use can be found at
https://www.tandfonline.com/action/journalInformation?journalCode=iahb20
Annals of Human Biology
ISSN: 0301-4460 (Print) 1464-5033 (Online) Journal homepage: https://www.tandfonline.com/loi/iahb20
Searching for archaic contribution in Africa
Cindy Santander, Francesco Montinaro & Cristian Capelli
To cite this article: Cindy Santander, Francesco Montinaro & Cristian Capelli (2019): Searching for
archaic contribution in Africa, Annals of Human Biology, DOI: 10.1080/03014460.2019.1624823
To link to this article: https://doi.org/10.1080/03014460.2019.1624823
Accepted author version posted online: 04
Jun 2019.
Published online: 26 Jun 2019.
Submit your article to this journal
Article views: 67
View Crossmark data
REVIEW
Searching for archaic contribution in Africa
Cindy Santander
a
, Francesco Montinaro
a,b
and Cristian Capelli
a
a
Department of Zoology, University of Oxford, Oxford, UK;
b
Estonian Biocentre, University of Tartu, Tartu, Estonia
ABSTRACT
Context: Africa’s role in the narrative of human evolution is indisputably emphasised in the emer-
gence of Homo sapiens. However, once humans dispersed beyond Africa, the history of those who
stayed remains vastly under-studied, lacking the proper attention the birthplace of both modern and
archaic humans deserves. The sequencing of Neanderthal and Denisovan genomes has elucidated evi-
dence of admixture between archaic and modern humans outside of Africa, but has not aided efforts
in answering whether archaic admixture happened within Africa.
Objectives: This article reviews the state of research for archaic introgression in African populations
and discusses recent insights into this topic.
Methods: Gathering published sources and recently released preprints, this review reports on the dif-
ferent methods developed for detecting archaic introgression. Particularly it discusses how relevant
these are when implemented on African populations and what findings these studies have shown
so far.
Results: Methods for detecting archaic introgression have been predominantly developed and imple-
mented on non-African populations. Recent preprints present new methods considering African popu-
lations. While a number of studies using these methods suggest archaic introgression in Africa,
without an African archaic genome to validate these results, such findings remain as putative archaic
introgression.
Conclusion: In light of the caveats with implementing current archaic introgression detection methods
in Africa, we recommend future studies to concentrate on unravelling the complicated demographic
history of Africa through means of ancient DNA where possible and through more focused efforts to
sequence modern DNA from more representative populations across the African continent.
ARTICLE HISTORY
Received 28 January 2019
Revised 13 May 2019
Accepted 17 May 2019
KEYWORDS
Africa; archaic introgression;
ancient admixture;
human evolution
Current state-of-the-art for archaic introgression
The past decade has seen the industrial-scale sequencing
and studying of ancient and modern genomes alike (Slatkin
and Racimo 2016; Nielsen et al. 2017). Prior to this, studying
the relationship between archaic hominins and anatomically
modern humans was predominantly left to paleoanthropolo-
gists whom have proposed several evolutionary models since
the first half of the 20th century. The most popular hypothe-
ses, which have been in competition up until very recently,
were the multiregional evolution and the recent out of Africa
model (RAO). Multiregional evolution proposed that humans
evolved across the Old World from local ancestors through
an interbreeding network since the last 1.8 million years
(Wolpoff et al. 1984), while RAO argued for an origin derived
from Africa which then dispersed across the globe (Stringer
and Andrews 1988). The turning point in this debate is often
considered to be in the late 1980s, when Cann and col-
leagues (1987) conducted a study of mitochondrial DNA of
present-day individuals where the results suggested a recent
African origin for modern humans. This was followed by sev-
eral other Y-chromosome and autosomal DNA studies sug-
gesting similar results and, therefore, supporting the RAO
model for many years (Jobling and Tyler-Smith 2003).
Since the sequencing of Neanderthal (Green et al. 2010)
and Denisova genomes (Reich et al. 2010) we have been
able to characterise the evolutionary relationship between
archaic and modern humans using genetic data as well as
explore the hypothesis of gene flow among them (Wall et al.
2013; Vernot and Akey 2014; Vernot et al. 2016;Pr
€
ufer et al.
2017; Browning et al. 2018; Hajdinjak et al. 2018) (note here
that paleoanthropologists hold different positions on the sig-
nificance of the degree of variation present within the genus
Homo, and the classification for these different groups ranges
from species to populations. The terms modern and archaic
are used here for simplicity and are intended to refer to
modern human populations usually defined as H. sapiens
and broadly to all the other Homo groups that co-existed, as
for example Neanderthals and H. floresiensis, as well as
poorly characterised forms such as the Denisovans, and
others still unknown). It is now understood that neither
multi-regionalism nor RAO can strictly explain the complex
interactions between the hominin taxa (Galway-Witham and
Stringer 2018). A number of intermediate models have been
suggested where their main differences lie in how human
genetic dispersal has taken place while mixing with other
hominin groups outside of Africa. Intermediate models have
CONTACT Cindy Santander cindy.santander@zoo.ox.ac.uk Department of Zoology, University of Oxford, Zoology Research and Administration Building,
11a Mansfield Road, Oxford, OX1 3SZ, UK
ß2019 Informa UK Limited, trading as Taylor & Francis Group
ANNALS OF HUMAN BIOLOGY
https://doi.org/10.1080/03014460.2019.1624823
included the ‘Leaky Replacement’model, which is fundamen-
tally RAO with limited hybridisation between modern and
archaic populations. However, some of these models con-
tinue to assume an evolutionary history of humans diverging
off a closely related core African population without account-
ing for the presence of multiple populations within Africa
and their absorption through admixture before proximate
extinctions of other Homo taxa prior to the exodus from
Africa. Others have made attempts to integrate genomic,
archaeological, fossil and paleoenvironmental data to form
models inclusive of African population structure (Lahr and
Foley 1998; Harding and McVean 2004; Stringer 2016; Henn
et al. 2018; Scerri et al. 2018).
As research continued and progressed in this field, the
idea of admixture between archaic and modern humans has
become an indispensable element to consider in any human
evolution model. Evaluating admixture between non-Africans
and archaic hominins has been comparatively more straight-
forward than in Africans, given the availability of both
Neanderthal and Denisova genomes (Green et al. 2010; Reich
et al. 2010). The sequencing of non-African archaic hominins
has led to irrefutable interest in wanting to understand what
relationship they had with modern humans and moreover
what possibly connects us to them. This yearning has mani-
fested in a number of questions about our origin as a spe-
cies, but it has also given rise to whether this phenomenon
has occurred not just in Eurasia, but also in Africa (Garrigan
et al. 2005; Wall et al. 2009; Hammer et al. 2011; Lachance
et al. 2012; Hsieh et al. 2016; Skoglund et al. 2017).
In order to answer these queries, a number of tools have
been developed to identify and characterise signatures of
archaic admixture, predominantly depending on either vari-
ant distribution, linkage disequilibrium or both. Patterson’sD
statistic, for example, measures excess sharing of derived
alleles between two sister populations (ingroup) and an out-
group. If none of the ingroup populations received genetic
material from the outgroup, they should share about the
same number of derived alleles with the outgroup. While
this method can detect whether there is an asymmetrical
variant sharing between archaic and modern humans, it is
unable to pinpoint segments in the genome that are of
introgressed archaic origin (Patterson et al. 2012). Moreover,
demographic scenarios such as ancient population structure
can generate similar results (Eriksson and Manica 2012;
Theunert and Slatkin 2017).
Mutation and recombination impact inherited DNA and,
therefore, inevitably shape the segments of a putatively
archaic origin. Consequently, these processes may be har-
nessed when developing algorithms to detect tracts of
archaic introgression. Pedigree studies in humans would sug-
gest that about 80 new mutations occur each generation
leading to a mutation rate of 0.5–110
9
per base pair
per year (Scally and Durbin 2012; Besenbacher et al. 2015).
Given that Neanderthals have diverged from the common
ancestor with Homo sapiens about 520,000–630,000 years
ago (Green et al. 2010;Pr
€
ufer et al. 2017; Hajdinjak et al.
2018), we can expect that the DNA of any two humans will
be on average closer to each other when compared to a
Neanderthal’s DNA sequence.
Given that recombination is not evenly distributed across
the human genome, several projects have built maps detail-
ing recombination crossover rates for populations of
European ancestry (Kong et al. 2002;2010; Matise et al. 2007;
The HapMap Consortium et al. 2010) and of West African
ancestry (The HapMap Consortium et al. 2010; Hinch et al.
2011; Wegmann et al. 2011). From a study led by Hinch
et al. (2011), an African enrichment (AE) map, a map of hot-
spots unique to African ancestry, was deduced by comparing
the Icelandic deCODE and African American (AA) recombin-
ation patterns (Kong et al. 2010; Hinch et al. 2011).
Availability of resources, such as high-resolution genetic
maps which are relevant for the populations of interest, is
crucial for current methods that detect archaic introgression
in modern genomes.
Here we delineate some of the most up-to-date methods
used in detecting archaic introgression in both Africans and
non-Africans. We consider the pitfalls of these methods bear-
ing in mind the absence of a sequenced African archaic gen-
ome, the complex demographic histories yet to be
disentangled in the African population, and the overall scar-
city of modern African genomes, which are comprehensively
representative of the diverse ethnic-groups found on
the continent.
Methods to infer archaic introgression
Linkage disequilibrium-based methods
We can consider recombination and mutations as the basis
of understanding an expected length of an introgressed tract
in relation to the time since the admixture event.
Introgression is distinct from incomplete lineage sorting (ILS)
in that it should leave behind longer tracts, as the former
being more recent (Liang and Nielsen 2014). The difference
in the size of the tracts should provide a way to tell whether
a shared tract with an archaic hominin is in fact introgression
or the genomic relic from an earlier common ances-
tral population.
Prior to the sequencing of the full Neanderthal and
Denisova genomes, Plagnol and Wall (2006) sought to take
advantage of the logic that putatively introgressed tracts
would have had a limited number of generations (e.g.
2000) to be broken down by recombination. Those tracts
could be identified by linkage disequilibrium (LD) where var-
iants in an archaic segment should be strongly associated
with other archaic variants in the genome, in other words
introgressed archaic variants should be found in high LD.
The authors came up with the Sstatistic (Figure 1), a sum-
mary statistic of LD, which extracts this particular information
through a scoring scheme that searches for derived muta-
tions that are in high LD. This method has become widely
used in several studies which seek signals of archaic intro-
gression in both non-African and African populations alike as
it was originally designed to identify introgression without
knowledge of the donor population (Plagnol and Wall 2006;
Wall et al. 2009). However, the availability of archaic hominin
2 C. SANTANDER ET AL.
sequences outside of Africa have provided a means of fur-
ther corroborating haplotypes detected by Sin non-African
populations (Wall et al. 2013; Vernot and Akey 2014; Vernot
et al. 2016; Browning et al. 2018). Such studies have been
able to find more Neanderthal contribution in East Asians
than in Europeans potentially because of two admixture
waves—one in Eurasian ancestors and another in Asians
(Wall et al. 2013; Vernot and Akey 2014). However, these sig-
natures can also be influenced by the different effective
population sizes for Europeans and Asians as well as a dilu-
tion of Neanderthal signatures in Europeans due to gene
flow from or into Africa, which can confound the true contri-
bution from Archaics into non-Africans (Sankararaman et al.
2014; Petr et al. 2019).
A subsequent study was able to detect introgressed
sequences which are uniquely Neanderthal or Denisovan or
shared amongst populations that have received contribution
from both archaic humans, such as in the case of the
Melanesians and Asians (Vernot et al. 2016). A more recent
study introduced an S-like method, Sprime, which uses a
similar scoring scheme as its predecessor with the exception
that it now performs detection on whole chromosomes
instead of sliding windows and takes into account local
mutation and recombination rates (Browning et al. 2018).
The authors of this method report that Sprime helped them
to discern that Asians today carried Denisovan introgression
from two waves of Denisovan admixture, one from a popula-
tion closely related to the Altai Denisovan into East Asians
and another more distantly related to the Altai Denisovan in
Papuans and South Asians. A more recent study has elabo-
rated on the complexity of archaic contact between
Denisovans and modern humans (Jacobs et al. 2019).
Probabilistic machine learning methods
Another option to using LD-based methods in detecting
introgressed sequences is to incorporate parametric assump-
tions into a probabilistic framework. Two methods that are
utilised for this are hidden Markov models (HMM) (Baum and
Eagon 1967) and conditional random fields (CRF) (Lafferty
et al. 2001). They are implemented with the concept that
each single nucleotide polymorphism (SNP) across the gen-
ome is a hidden random variable with two states: either
human or archaic (Figure 2). These methods integrate what
we know about the biological processes such as human
recombination and mutation and what we understand about
the demographic events since the human exodus from
Africa. In the case of looking for introgression in non-
Africans, these methods are implemented with the assump-
tion that we would not expect to find Neanderthal or
Denisovan contributions in West Africans from archaic admix-
ture. With the resulting parametric information one can cal-
culate the probability that a given sequence is of archaic
origin and if it is present in modern humans via admixture.
Two different studies made use of HMM methods, but the
main difference between the two was that, respectively, one
provided a priori chosen parameters, whereas the other gath-
ered parameters from a reference dataset (Pr€
ufer et al. 2014;
Seguin-Orlando et al. 2014). Pr€
ufer et al. (2014) estimated a
70–100% archaic enrichment in East Asians compared to
Europeans. Seguin-Orlando et al. (2014), by looking at mod-
ern human aDNA, found the approximate time of admixture
between modern humans and Neanderthals occurred
16,600 years earlier than the ancient sample tested, in-line
with what other studies approximate (37,000–86,000 years
ago) (Green et al. 2010; Reich et al. 2010; Sankararaman et al.
2012; Fu et al. 2014).
Similar to HMM are CRF models which can incorporate
other forms of information related to the data, for example,
LD, haplotype structure and allele configurations from mul-
tiple samples. The parameters are then calibrated by training
them using simulations with particular demographic assump-
tions ranging from divergence dates to effective popula-
tion size.
Sankararaman et al. (2014) used a CRF model in their
study where the first emission function provided a high
probability of being archaically introgressed if a variant was
found in Non-Africans and in the Neanderthal reference gen-
ome but absent in Africa; akin to what the aforementioned
studies had done with their HMM methods. Using this CRF
model the authors detected in the 1000 Genomes Project
15% of introgressed sequence with 99% precision. As they
detected more archaic sequence (38%) their precision
decreased to 98%. This study found 1.17–1.20% of the
Figure 1. Schematic for Sscoring. Here we see phased chromosomes and a
test population that is represented in red and the outgroup in grey. S
attempts to search for the most optimal sums of scores for an overall sub-set of
SNPs at a given locus. It rewards fully linked pairs of sites, in other words where
two successive SNP positions do not differ, and the increase in that reward is
proportional to the distance between the positions. Consequently, Svalues
increase with increasing LD within a window. The SNP positions that provide
the optimal score are taken note of and can then be used to provide the delim-
ited region that gives that optimal score. The lower bottom of the figure
depicts an example of how Sgoes about calculating the most optimal sums
of scores. Seven positions within a window of 50,000 bp and only SNPs not pre-
sent in the outgroup are considered. Sites within the haplotype are labelled as
a white star (i.e. ancestral) or a grey star (i.e. derived), those highlighted in red
(target) are calculated to give the most optimal solution and that no other set
of SNPs gives a higher score. The first 10,000 bp displays higher linkage than
the following 10,000 bp where there is a change in genotype leading to a pen-
alty of 10,000. We can see that in the calculation that follows the positions
from about 30,000 bp onwards are rewarded for being linked and outweigh
any penalty from before. The scoring function works in rewarding linked sites
with a score of 5,000 plus the distance between sites, penalties are given
10,000 for up to five mismatches in the above example. Regions with more
than five mismatches or no linked sites are 1, or essentially no score. The total
Sscore for this example is 30,000.
ANNALS OF HUMAN BIOLOGY 3
autosomes in Europeans have Neanderthal ancestry and that
between 1.37–1.40% of East Asian autosomes have
Neanderthal ancestry. Two years after developing this CRF-
based method, Sankararaman et al. (2016) applied it to the
Simons Genome Diversity Project (Mallick et al. 2016), which
included 257 high coverage individuals from 142 worldwide
populations. Here both the Neanderthal and Denisova
genomes were used in the analysis and they identified 1.06%
of autosomal DNA in Europeans of Neanderthal origin, while
in East Asians the proportion was slightly higher, at 1.40%.
Furthermore, they found a small although still significant
amount of Denisova contribution in East Asians, estimated to
be 0.06%.
While they are inarguably powerful methods, so far HMM
and CRF have predominantly relied on two things: a priori
chosen parameters which are based on well-established
demographic assumptions and a reference dataset. Recently
an HMM method was released by Skov et al. (2018), which
does not depend on an archaic reference but does require a
specific phylogenetic arrangement that allows a large frac-
tion of variation to be removed using an outgroup
population to detect introgression. This model bases its logic
on the observation that an archaic tract introgressed into a
population should display a high density of variants not
found in populations which have not experienced introgres-
sion. Like the previous methods reviewed here, Skov et al.
(2018) focus on a scenario where admixture has occurred
between a deeply divergent archaic population (e.g.
Denisova) and a modern human population (ingroup;
Papuans) but where admixture has only occurred with the
ingroup and not with the outgroup (e.g. Yoruba). By remov-
ing variants that are shared with the outgroup, they inspect
the density of the remaining variants in the ingroup which
are, essentially, private SNPs. In their analysis they use Sub-
Saharan Africans from the 1000 Genomes Project as an out-
group. The authors demonstrate, with 89 Papuan genomes,
that they are able to recover more Denisovan introgressed
segments (77 Mb out of 164.23 Mb of archaic sequence in
Papuans) than previous methods (Vernot et al. 2016;
Browning et al. 2018), possibly because this approach does
not rely upon validation with the Altai Denisovan genome
reference, ultimately demonstrating the advantage of a
Figure 2. Overview of Hidden Markov model and conditional random field framework. Here is a depiction of how probabilistic models, such as HMM and CRF, can
estimate the ancestry (x
i
) of a SNP in a genome sequence (i). The possible hidden states for x
i
are either introgressed (archaic) or not introgressed (not archaic). In
the example, y
i
is a matrix composed of individuals from a target population, an archaic population, and two outgroup populations that act as references for mod-
ern human variation. These are considered at three SNP positions (below model). Sites which are consistent with introgression (x
1
) are determined by the derived
allele of the target population also present in the archaic population, but not in the reference. In the case of x
2
, the derived allele is only in one of the reference
populations suggesting that the derived allele is modern human variation and, therefore, inconsistent with introgression. In x
3
, both target and reference popula-
tions share the derived allele, but absent in the archaic population which is uninformative. Hidden states (ancestry) are connected to the observed data (y
i
) through
emission probabilities (red) for HMM or emission functions (vertical red and diagonal yellow) for CRF. These emission functions in a CRF model can be used to
evaluate whether a site is consistent or inconsistent with introgression, like in HMM. Additionally, they can also score whether a haplotype overall is closer to the
archaic sequence than to the reference haplotypes. Horizontal connections between x
i
and x
iþ1
denote transition probabilities (HMM) or transition functions (CRF)
which model linkage between ancestral states along the genome. These parameters depend on the recombination, admixture proportion, and time of admixture.
4 C. SANTANDER ET AL.
reference-free detection method. In addition to detecting
introgressed fragments, this method can also infer admixture
proportions as well as divergence time of human and archaic
populations. Although this method does not require an
archaic reference and does not require phased data, without
using a suitable outgroup, detecting introgression in the
African population with just genome samples remains
a challenge.
Detection of putative archaic introgression
in Africans
So far, we have reviewed studies that have implemented the
above methods outside Africa, but which have also relied on
an archaic reference to filter out false positive candidates as
well as recover false negatives. Without using an archaic ref-
erence, the authors of those studies have collectively shown
that methods like Sand Sprime can identify anywhere from
30–60% of all true positive introgressed sequences at low
false discovery rates, and in the case of Sprime, with an
accuracy of 93%, as was shown on simulated data for differ-
ent demographic histories (Vernot and Akey 2014; Vernot
et al. 2016; Browning et al. 2018). Skov et al. (2018) have
shown a low rate of false detection on both simulated and
real data using an HMM method without the need of an
archaic reference.
Earlier studies initially implemented Son African popula-
tions using genotype data for both coding and non-coding
regions (Plagnol and Wall 2006; Wall et al. 2009; Hammer
et al. 2011). In particular, a study looking at 61 non-coding
autosomal regions was able to infer 2% of genetic material
introgressed into contemporary Mandenka, Biaka, and San
around 35 kya, from an archaic population that split from
the ancestors of modern humans at approximately the same
time as Neanderthals and Denisovans, 700 kya (Hammer
et al. 2011). The authors use an inferential approach to test
whether the data they have for two hunter-gatherer popula-
tions (Biaka and San) and an agricultural population
(Mandenka) fit a scenario of no admixture or low levels of
admixture under two possible models—a two-population or
a three-population model. Under a two-population model
they test for archaic introgression by comparing Svalues
from their data to values estimated using parameters under
a no admixture scenario. In a complementary fashion, the
authors also introduce three summary statistics for an
approximate-likelihood method, which consists of first identi-
fying the two most divergent sequences for a locus and
establishing two groups where the rest of the sequences will
cluster to correspondingly. The protocol then calls for esti-
mating the distribution of the fraction of shared polymor-
phisms between the two groups (D
1
), the ratio of the
number of differences between them (D
2
), and finally the
size of the smaller of the two groups (D
3
). Each one is meant
to represent the time of introgression, the time of the
archaic-split, and finally the proportion of admixture, respect-
ively. This distribution of summary statistics is calculated by
simulating several ancestral recombination graphs (ARGs)
(Griffiths and Marjoram 1997). In using both Sand these
three summary statistics (D
1
,D
2
,D
3
), the authors rejected the
null hypothesis that an ancestral population with no admix-
ture gave rise to anatomically modern humans. Instead both
their inferential methods identify three exceptionally long
haplotypes (Table 1) at low-frequency amongst African
hunter-gatherers (central African rainforest hunter-gatherers
and San), signalling these regions as putatively introgressed
from an archaic population. Specifically, the authors suggest
that central Africa may have been the place of origin for an
extinct archaic hominin that admixed with modern humans
in light of finding all three haplotypes amongst the Mbuti of
Democratic Republic of Congo.
With the advances in sequencing technology whole-gen-
ome sequencing becomes accessible enough to sequence
multiple individuals of several populations. Lachance et al.
(2012) were the first to implement Son whole-genomes
from African Hunter-Gatherer populations: rainforest hunter-
gatherers from Cameroon, click-speaking Hadza, and
Sandawe from Tanzania. Consistent with previous studies
(Wall et al. 2009; Hammer et al. 2011), they find evidence of
archaic admixture in all three populations with candidate loci
corresponding to a time to recent common ancestor compar-
able to those observed in Europeans from Neanderthal intro-
gression. The authors, through coalescent simulations,
concluded that Swas robust enough to detect admixture
and differences in amounts of introgression.
An ensuing study by Hsieh et al. (2016a) corroborated this
evidence for archaic admixture in Africa by addressing the
confounding effects of the demographic history of the popu-
lation in question. Sp-value distributions were calculated
for simulations based on two models inferred from a previ-
ous study (Hsieh et al. 2016b) that incorporated both isola-
tion and gene flow with neighbouring farming populations.
These distributions were compared with Sp-values distribu-
tions from the observed data. This approach is an improve-
ment to what Lachance et al. (2012) performed on their data
by accounting for sequences that may have extreme Sval-
ues, but not being statistically significant if one considers the
effects of demography and genomic processes such as muta-
tion and recombination rates. Only the significant top 1% in
the distribution of these p-values were chosen as candidate
introgressed loci, resulting in a total of 265 candidate loci,
spanning 20 Mb in length. They estimated a false discovery
rate (FDR) between 19% and 68% in these top candidates.
Surprisingly, from these putatively introgressed regions,
Hsieh et al. (2016a), using a variant of D
3
from Hammer et al.
(2011), inferred at least one admixture event with low
amounts of introgression around 9000 years ago, although a
fine-scale understanding of the nature of these recurrent
events in Africa still remains unresolved.
In a similar vein, a recent study that analysed 21 high
coverage African genomes within an Approximate Bayesian
computation (ABC) with Deep Learning framework (Mondal
et al. 2019) also estimated that interbreeding occurred
between modern humans in Africa (i.e. Khoe-San, Mbuti and
West Africans) and an archaic ghost population that diverged
from the basal human lineage around the same temporal
ANNALS OF HUMAN BIOLOGY 5
scale seen between Neanderthal and Denisovans (Lorente-
Galdos et al. 2019).
Finally, a genomic analysis comprising genomic material
from ancient individuals from Southern and Eastern Africa
and, therefore, lacking confounding genomic fragments
derived by recent demographic events, revealed that a
model depicting Southern Africa Khoe-San populations as
basal of all the African populations is not fully supported,
since the former show different relatedness to East and
Western Africa (Skoglund et al. 2017). Furthermore, different
West African populations show different relatedness to
ancient Khoe-San, which is inconsistent with being derived
from a homogeneous ancestral population which diverged
from ancient southern Africans. This scenario would be com-
patible with either ‘archaic’admixture in Western Africans,
affecting different populations heterogeneously, or with
long-term admixture which affected Western African groups
with different magnitudes.
Although the results of recently released preprints are
subject to change, we also report here some of the interest-
ing and strongly relevant findings recently submitted to pre-
print archives. Durvasula and Sankararaman (2019a)
developed a machine learning method, ArchIE, which makes
use of training datasets to calculate a set of features that are
potentially informative of introgression. A prediction about
archaic local ancestry could then be made for any given win-
dow by using a binary logistic regression model with the set
of computed parameters. Results are then summarised for
what is indicative of archaic admixture for each haplotype.
The authors found that ArchIE weighs first the number of
private SNPs followed by the skew of the distance vector in
the underlying logistic regression model. After training their
logistic regression predictor using the parameters from a
dataset with confirmed Neanderthal introgression in non-
African populations, the authors applied this method on
Yoruban individuals from the 1000 Genomes Project
(Durvasula and Sankararaman 2018). They implemented this
under the assumption that the predictor is expected to be
sensitive to introgression events from populations that
shared ancestral population structure with Yorubans. Their
results suggest that the archaic ancestry in Yoruba is best
explained by admixture with an archaic ghost population
more than the possibility of Neanderthal ancestry from back-
migration or from admixture with an extant modern human
population. In total, about 258 Mb of introgressed sequences
in the Yoruba were recovered using this method in several
protein coding regions at high frequency in the population
(Table 1). An update to this study earlier this year claimed a
further recovery of 482 and 502 Mb of archaic ancestry in
Yoruba and Mende populations, respectively, and that sub-
Saharan populations derive 2–19% of their genetic ancestry
from an archaic population that diverged before the split
between Neanderthals and modern humans (Durvasula and
Sankararaman 2019b).
In another recent preprint, Ragsdale and Gravel (2018)
explore classic statistics, as well as less familiar tests, to
measure introgression. These were used to infer a demo-
graphic model with archaic introgression within a likelihood
framework in the absence of an archaic reference genome.
By implementing this approach of joint statistics on inter-
genic data from the 1000 Genomes Project, the authors
found that the Luhya in Webuye, Kenya, and the Yoruba of
Ibadan, Nigeria, exhibited approximately 6–8% archaic
admixture, respectively. Moreover, this study shows that the
commonly used model of human demographic history,
derived from single-site allele frequency spectrum (AFS) and
corroborated by LD decay curves, tends to fit the real data
well, but significantly under-estimates the levels of LD
among rare alleles. They show that, by modelling archaic
introgression worldwide, including African admixture with an
archaic population that split off around 460–540 kya, this dis-
crepancy in the levels of LD among rare alleles is resolved.
Speidel et al. (2019) in a most recent preprint have
released a method, Relate, capable of inferring genome-wide
genealogies for thousands of samples which they also imple-
mented on African populations from the 1000 Genomes
Table 1. Archaic introgressed candidate loci in African populations.
Putatively introgressed regions Type Population Study
RP11-286M16 (chr 1) lincRNA Yoruba (Durvasula and Sankararaman 2018)
RN7SKP160 (chr 1) Pseudogene Yoruba (Durvasula and Sankararaman 2018)
4qMB179 Non-coding Biaka (Hammer et al. 2011)
KCNIP4 (chr 4) Protein coding Yoruba (Durvasula and Sankararaman 2018)
XRCC4 (chr 5) Protein coding Yoruba (Plagnol and Wall 2006; Wall et al. 2009)
MTFR2 (chr 6) Protein coding Yoruba (Durvasula and Sankararaman 2018)
TRPS1 (chr 8) Protein coding Yoruba (Durvasula and Sankararaman 2018)
13qMB107 Non-coding San (Hammer et al. 2011)
TJP1 (chr 15) Protein coding Yoruba (Plagnol and Wall 2006; Wall et al. 2009)
DUT (chr 15) Protein coding Yoruba (Plagnol and Wall 2006; Wall et al. 2009)
HSD17B2 (chr 16) Protein coding Yoruba (Durvasula and Sankararaman 2018)
KRT18P61 (chr 17) Pseudogene Yoruba (Durvasula and Sankararaman 2018)
NF1 (chr 17) Protein coding Yoruba (Durvasula and Sankararaman 2018)
RP1115E18 (chr 17) Pseudogene Yoruba (Durvasula and Sankararaman 2018)
18qMB60 Non-coding Biaka (Hammer et al. 2011)
MIR125B2 (chr 21) miRNA Yoruba (Durvasula and Sankararaman 2018)
Xp21.1 Non-coding Mbuti (Garrigan et al. 2005)
A00 (chr Y) Sex chromosome African-American & Mbo (Mendez et al. 2013)
Top 350 candidates
†
(across chr 1–22) Unknown (genic depleted) Sandawe, Western Pygmy, Hadza (Lachance et al. 2012)
Distinct 265 candidates(across chr 1–22) Genic & non-genic Biaka and Baka (Hsieh et al. 2016a)
†
Specific coordinates were not made available.
Coordinates available by requesting from author(s) of respective study.
6 C. SANTANDER ET AL.
Project. Their results support separate ancient events unique
to African populations, in particular an introgression event in
the Yoruba with a hominin not closely related to
Neanderthals (also diverging before the split between
Neanderthals and modern humans). This method might be
useful to implement on other African populations, such as
the San and the Mbuti, as more whole genomes for these
populations become available.
Caveats with the state-of-the-art
Methods that do not require an archaic reference can be
used to exploit modern whole-genome data, but still have
their caveats such as high false positive rates. There have
been but a few studies focusing on Africa in attempts to
detect archaic sequences and potentially introgressed
regions (Garrigan et al. 2005; Wall et al. 2009; Hammer et al.
2011; Lachance et al. 2012; Mendez et al. 2013; Hsieh et al.
2016a; Durvasula and Sankararaman 2018; Ragsdale and
Gravel 2018). Previous studies focusing on archaic admixture
outside of Africa have consistently used Africa as a proxy
population that did not experience ‘archaic introgression’,
which has essentially meant ‘did not experience contact with
Eurasian Archaics’such as Neanderthals and Denisovans
(Green et al. 2010; Reich et al. 2010; Sankararaman et al.
2014; Sankararaman et al. 2016;Pr
€
ufer et al. 2017; Durvasula
and Sankararaman 2018). Those that have taken up the ques-
tion of archaic introgression in Africa have had to do so in
the absence of an African archaic reference. Some of those
studies have used LD-based methods such as Sto search
for signals and have equated the identified regions, which
display a fairly old time to recent common ancestor (TMRCA)
and long haplotype, as evidence of archaic introgression
(Lachance et al. 2012). Although TMRCA does not always
equate population divergence when considering selection
and complex demographic models (Henn et al. 2018). Others
have, similarly, run Son African hunter-gatherer populations
and then chosen top candidates of archaic introgression by
comparing against simulations with a set of demographic
assumptions (Hsieh et al. 2016a) to account for confounding
effects on TMRCA due to selection. Nevertheless, both of
these studies emphasise the importance to better character-
ise the nature of admixture in Africa and, more generally, the
demographic history of early African populations.
Early African demography: a complicated history
Our knowledge of what may have occurred since the diver-
gence of modern humans and other archaic forms within
Africa is limited both in the fossil record and in availability of
modern genomes. Few genetic studies have addressed early
human history in Africa, despite the amount of fossil and
archaeological findings elucidating on the emergence of ana-
tomically modern humans (AMH) (Campana et al. 2013;
Schlebusch and Jakobsson 2018).
In particular, East Africa is the most extensively excavated
area in the continent, while Central and Western Africa are
the least explored (Schlebusch and Jakobsson 2018). Some
of the oldest fossils of the genus Homo can be found in
southern Africa, with a fossil record that begins from 2 mil-
lion years ago, with pronounced transitional forms ranging
from 200–600 kya (Dusseldorp et al. 2013). Forms displaying
transitional AMH features have appeared in the record of
southern Africa around 100–300 kya and fully AMH emerge
here 120 kya. Strikingly, while the fossil record for North
Africa has been limited, it is where the oldest fossils meeting
the criteria of AMH can be found—particularly the ancient
remains from Jebel Irhoud, dated to 300 kya (Hublin et al.
2017). Interestingly in a younger time period of 100–60 kya,
North African remains still denote archaic morphological
traits, despite anatomical modernity (Rightmire 2009).
Fossils dating 160–180 kya in East Africa, Omo Kibish and
Herto have been observed to be fully AMH and have often
been used to support East Africa as the birthplace of modern
humans (White et al. 2003; McDougall et al. 2005). However,
an East African calvaria from Lukenya Hill in Kenya, while
regarded Homo sapiens, demonstrates morphological features
better represented in archaic hominins yet dated as recently
as 23 kya (Tryon et al. 2015).
Of the notably few and recent assemblages from western
and central Africa, some remains display features that are
more in line with archaic humans, but which are fairly young,
such as the Ishango site in the eastern part of the
Democratic Republic of Congo dated to 20–25 kya and Iwo
Eleru skeletal remains from Nigeria dating back to 13 kya
(Harvati et al. 2011). Yet the human fossils from Shum Laka
in Cameroon, which date 3–7 kya, are fully AMH (Lavachery
2001). Some have suggested, based on both morphology
and the mosaic-like fashion, that technological replacement
took place from central to western Africa, and that this could
support a model where archaic admixture took place
between surviving archaic populations and modern humans
(Scerri 2017; Schlebusch and Jakobsson 2018). As consistent
as some of these fossils may be with supporting archaic
introgression in Africa, understanding how admixture
impacts morphology to better understand the mosaic-like
distribution for early Homo sapiens across different regions
has been strongly overshadowed by the efforts to prove that
modern humans originated from Africa (Henn et al. 2018).
Despite the several assemblages of fossils found across
Africa, DNA preservation from these samples is often com-
promised—a great deal owed to climatic obstruction
(Campana et al. 2013). Alternatively, recovering demographic
history from modern genomes can be challenging with the
lack of large genomic datasets representing the whole of
Africa, but especially in the light of the continent’s complex
deep past. The movements, and subsequent gene flow, from
the agricultural expansion and spread of pastoralism have
cloaked the ancient variation pertaining to early modern
humans (Tishkoff et al. 2009; Montinaro et al. 2017).
Consequently, this has complicated demographic inferences
that can be recovered regarding ancient African population
structure. Although recent ancient samples have provided
insight into the relatedness between populations prior to
monumental demographic events (Skoglund et al. 2017),
ANNALS OF HUMAN BIOLOGY 7
these sort of samples with endogenous aDNA remain rare
and will most likely accumulate at a slower rate.
Simulating various demographic scenarios has corrobo-
rated our understanding of the demographic history of pop-
ulations (Gravel 2012). In the case of African demographic
history, modern genomes, albeit challenging, can still provide
a window into the past. As such, there have been genetic
studies that simulate important events in early migrations
such as the Bantu expansion and pastoralist movements
(Li et al. 2014; Gonz
alez-Santos et al. 2015; Marks et al. 2015).
This can assist our understanding of African pre-history in
order to make more appropriate demographic assumptions
about the populations on the continent. Henn et al. (2018)
lay out distinct models of early modern human origins and
corresponding support for them in the fields of morphology,
archaeology and genetics. These models can provide guid-
ance in model-testing and exploration of early demographic
events. Nevertheless, without more archaeological data and
more genomes, both ancient and modern, those explorations
might overall be of limited scope and self-fulfilling.
Back to the basics: the landscape of recombination and
genetic variation in African populations
In the absence of aDNA from an archaic hominin or ghost
population in Africa, we must rely on modern genomes to
uncover whether archaic introgression took place.
Demographic history alone is not sufficient when using mod-
ern genomes as windows into the past. We must also con-
sider biological processes in order to develop tools that aid
us in answering questions about our genetic history.
Consequently, it is crucial to be aware of the differences in
the genomic landscape of different populations. For example,
recombination maps have been built for populations with
European-ancestry (deCODE, HapMapCEU) (Kong et al. 2010;
The HapMap Consortium et al. 2010), West African-ancestry
(HapMapYRI) (The HapMap Consortium et al. 2010) and
mixed-ancestry (AAmap, AfAdm map) (Hinch et al. 2011;
Wegmann et al. 2011). These studies have led to the conclu-
sion that, comparatively, West Africans have more recombin-
ation hotspots across the genome than Europeans and,
therefore, crossovers are more evenly distributed, leading to
shorter LD distances in West Africans (The HapMap
Consortium et al. 2010; Hinch et al. 2011). Moreover, these
studies have been able to show that recombination events
appear to be concentrated at hotspots which correlate with
a particular ancestry (The HapMap Consortium et al. 2010;
Wegmann et al. 2011). Considering that African populations
show the highest levels of genetic diversity in both between-
and within-population, it is essential to note that the recom-
bination landscape across Africa may differ and, therefore,
LD-based methods developed to search for introgression in
non-African populations must be adjusted for these differen-
ces. For example, if we wanted to use Sto explore archaic
introgression in West Africans, we must account for the
shorter extent of LD and, therefore, expect shorter intro-
gressed haplotypes than what is seen in Eurasians, even if
admixture putatively took place at a similar date. Mutation
rate, too, has been shown to differ between populations and
still remains without a final consensus (Narasimhan et al.
2017; Ragsdale et al. 2018).
In a similar vein, it is vital to point out that the grand
majority of tools that explore human history in high-through-
put sequencing use SNPs to answer questions about evolu-
tion and variation. While SNPs are undoubtedly useful
genetic markers, structural variants (SVs) such as insertions,
deletions, duplications, and inversions make up most of the
variation between individuals (Sudmant et al. 2015). Those
with African-ancestry have been shown to exhibit significant
differences in SV profiles from other populations, which is
consistent with what has been observed with SNPs (Simons
et al. 2014; Sudmant et al. 2015; Sherman et al. 2019).
However, the complexity of SVs present on both the individ-
ual and population level can often go unappreciated, first
due to reference-bias (Sudmant et al. 2015; Sherman et al.
2019) but also because SVs and repetitive elements are not
commonly considered when developing tools that measure
genetic variation for non-clinical purposes (Santander et al.
2017). This reference bias can make it challenging to observe
what other regions of the genome are shared between cer-
tain populations and archaics, either because of ancestral
structure or introgression (Gardner et al. 2017;G
€
unther and
Nettelblad 2018).
When searching for archaic introgression in African popu-
lations, both Lachance et al. (2012) and Hsieh et al. (2016a)
found pronounced depletion of archaic sequence in genic
regions suggesting that introgressed loci in hunter-gatherer
populations are neutrally evolving remnants. However, a sub-
sequent study (Xu et al. 2017) has found that archaic intro-
gression in Africa contributes to the variation in the human
salivary gene MUC7, consequently affecting the composition
of the oral microbiome in modern Africans today. This study
also suggested that copy number variation in MUC7 has rap-
idly evolved under adaptive forces potentially shaped by
pathogenic pressures among primates (Xu et al. 2016).
Recent studies looking at African hunter-gatherer and
farmer populations have explored how deleterious genetic
variation amongst human populations is affected by changes
in population size and gene flow (Lopez et al. 2018).
Specifically, hunter-gatherers are efficient in purifying selec-
tion, despite having a recent population collapse (Simons
et al. 2014; Lopez et al. 2018). Long-term selection against
archaic introgression in Africa has not been looked at,
although recent studies have shown that non-African popula-
tions demonstrate selection against introgression in regula-
tory regions more than in protein-coding regions (Petr et al.
2019). A study exploring the genetic relationship between
sequenced archaic hominins and Africans showed substantial
IBD sharing between Africans (East and West) and
Denisovans best explained by interbreeding between the
ancestor of humans and other archaic hominins (Povysil and
Hochreiter 2016). Whether these short IBD fragments are
skewed in their distribution across genic and non-genic
regions remains to be confirmed as well if they are putatively
introgressed haplotypes under any form of non-neutral selec-
tion. However, Durvasula and Sankararaman (2018) have
8 C. SANTANDER ET AL.
detected a number of regions with archaic ancestry in
Yoruba which are in genic regions and that may be under
positive selection (Table 1). Nevertheless, it is important to
point out that without ancient African samples to confirm
archaic introgression in present-day African populations,
signs of adaptive introgression will remain putative.
Overcoming the over-simplification of Africa:
future directions
Detection of archaic introgression with current state-of-the-
art methods relies on at least one of three things: (1) an
archaic reference, (2) known demographic history and (3)
large sample size. While acquiring an African archaic refer-
ence will take time considering the limitations of the current
technology and the integrity of available samples, this is a
possibility that shows signs of becoming more feasible
(Skoglund et al. 2017). In the meantime, efforts should be
strengthened to enrich the geographic coverage for genetic
data across Africa. More modern African genomes will eluci-
date the history of different populations across the continent
and how they relate to each other in terms of ancient gene
flow and structure in light of more recent migration.
Simulations without more insight into demographic history,
which are used to calibrate whether archaic introgression is
present in absence of an archaic reference, can only be as
accurate as the demographic assumptions they are modelled
upon (Henn et al. 2018). Moreover, caution should be taken
when implementing methods where certain biological
assumptions are better in line with non-African populations
and where extrapolating what has been done in non-African
populations to African populations can potentially lead to
both false positives and false negatives alike.
In the absence of both an archaic reference and clear
demographic inferences for Africa, new (Speidel et al. 2019)
or less explored (Ragsdale and Gravel 2018) methods might
be better suited in answering questions surrounding the
nature of archaic introgression in Africa once more modern
African genomes become available. Several questions still
remain such as: did admixture take place between archaic
hominins and modern humans or with the ancestors of mod-
ern humans? Or did ancient admixture take place between
existing populations in Africa today and ancient ghost popu-
lations? How often and for how long did admixture poten-
tially take place? Lastly, more needs to be investigated
regarding putative adaptive introgression in Africans (Xu
et al. 2017; Durvasula and Sankararaman 2018) and, in gen-
eral, evidence of selection for or against archaic ancestry in
modern African genomes (Durvasula and Sankararaman
2018). New ancient and modern genetic data will allow us to
explore and infer better demographic models so that we
may move on to reliably answer these outstanding questions
which are crucial to understanding our modern
human origins.
Acknowledgements
The authors would like to thank St. Hugh’s College, the Leverhulme
Trust and Comisi
on Nacional de Investigaci
on Cient
ıfica y Tecnol
ogica,
Gobierno de Chile. We would also like to thank all the students,
researchers and collaborators who have contributed to the studies that
elucidate on the genomic diversity across African populations and to the
people who have donated their DNA to make those studies possible.
Disclosure statement
The authors declare no competing interests.
ORCID
Cindy Santander http://orcid.org/0000-0003-3021-6809
References
Baum LB, Eagon JA. 1967. An inequality with applications to statistical
estimation for probabilistic functions of markov processes and to a
model for ecology. Bull Am Math Soc. 73(3):360–364.
Besenbacher S, Liu S, Izarzugaza JMG, Grove J, Belling K, Bork-Jensen J,
Huang S, et al. 2015. Novel variation and de novo mutation rates in
population-wide de novo assembled Danish trios. Nat Commun. 6(1):
5969.
Browning SR, Browning BL, Zhou Y, Tucci S, Akey JM. 2018. Analysis of
human sequence data reveals two pulses of archaic denisovan admix-
ture. Cell 173(1):53.e9–61.e9
Cann RL, Stoneking M, Wilson AC. 1987. Mitochondrial DNA and human
evolution. Nature. 325(6099):31.
Campana MG, Bower MA, Crabtree PJ. 2013. Ancient DNA for the archae-
ologist: The future of African research. [place unknown]; [cited 2018
Dec 7].
Durvasula A, Sankararaman S. 2018. Recovering signals of ghost archaic
admixture in the genomes of present-day Africans. bioRxiv. [Internet]
[cited 2018 Oct 1]:285734.
Durvasula A, Sankararaman S. 2019a. A statistical model for reference-
free inference of archaic local ancestry. PLoS Genet. 15(5):e1008175.
Durvasula A, Sankararaman S. 2019b. Recovering signals of ghost archaic
introgression in African populations. bioRxiv. [Internet]. [cited 2019
Apr 30]:285734.
Dusseldorp G, Lombard M, Wurz S. 2013. Pleistocene homo and the
updated stone age sequence of South Africa. S Afr J Sci. 109:1.
Eriksson A, Manica A. 2012. Effect of ancient population structure on the
degree of polymorphism shared between modern human populations
and ancient hominins. Proc Natl Acad Sci. 109:13956–13960.
Fu Q, Li H, Moorjani P, Jay F, Slepchenko SM, Bondarev AA, Johnson
PLF, et al. 2014. Genome sequence of a 45,000-year-old modern
human from western Siberia. Nature. 514(7523):445–449.
Galway-Witham J, Stringer C. 2018. How did Homo sapiens evolve?
Science. 360:1296–1298.
Gardner EJ, Lam VK, Harris DN, Chuang NT, Scott EC, Stephen Pittard W,
Mills RE, Devine SE. 2017. The mobile element locator tool (MELT):
population-scale mobile element discovery and biology. Genome Res.
27(11):1916–1929.
Garrigan D, Mobasher Z, Kingan SB, Wilder JA, Hammer MF. 2005. Deep
haplotype divergence and long-range linkage disequilibrium at
Xp21.1 provide evidence that humans descend from a structured
ancestral population. Genetics. [Internet] [cited 2018 Dec 7];170:
1849–1856.
Gonz
alez-Santos MG, Montinaro F, Oosthuizen O, Oosthuizen E, Busby
GBJ, Anagnostou P, Destro-Bisol G, et al. 2015. Genome-Wide snp
analysis of southern african populations provides new insights into
the dispersal of bantu-Speaking groups. Genome Biol Evol. 7(9):
2560–2568.
Gravel S. 2012. Population genetics models of local ancestry. Genetics.
191(2):607–619.
ANNALS OF HUMAN BIOLOGY 9
Green RE, Krause J, Briggs AW, Maricic T, Stenzel U, Kircher M, Patterson
N, et al. 2010. A draft sequence of the neandertal genome. Science.
328(5979):710–722.
Griffiths RC, Marjoram P. 1997. An ancestral recombination graph. Prog
Popul Genet Hum Evol Minneapolis, MN, 1994. New York (NY):
Springer.
G€
unther T, Nettelblad C. 2018. The presence and impact of reference
bias on population genomic studies of prehistoric human popula-
tions. bioRxiv. [Internet]. [cited 2018 Dec 10]:487983.
Hajdinjak M, Fu Q, H€
ubner A, Petr M, Mafessoni F, Grote S, Skoglund P,
et al. 2018. Reconstructing the genetic history of late Neanderthals.
Nature 555:652–656.
Hammer MF, Woerner AE, Mendez FL, Watkins JC, Wall JD. 2011. Genetic
evidence for archaic admixture in Africa. Proc Natl Acad Sci. 108(37):
15123–15128.
Harding RM, McVean G. 2004. A structured ancestral population for the
evolution of modern humans. Curr Opin Genet Dev. 14(6):667–674.
Harvati K, Stringer C, Gr€
un R, Aubert M, Allsworth-Jones P, Folorunso CA.
2011. The later stone age calvaria from iwo eleru, Nigeria: morph-
ology and chronology. Relethford JH, editor. PLoS One. [Internet]
[cited 2019 Jan 18];6:e24024.
Henn BM, Steele TE, Weaver TD. 2018. Clarifying distinct models of mod-
ern human origins in Africa. Curr Opin Genet Dev. [Internet] [cited
2018 Nov 13];53:148–156.
Hinch AG, Tandon A, Patterson N, Song Y, Rohland N, Palmer CD, Chen
GK, et al. 2011. The landscape of recombination in African Americans.
Nature. [Internet] [cited 2018 Dec 10];476:170–175.
Hsieh PH, Woerner AE, Wall JD, Lachance J, Tishkoff SA, Gutenkunst RN,
Hammer MF. 2016a. Model-based analyses of whole-genome data
reveal a complex evolutionary history involving archaic introgression
in Central African Pygmies. Genome Res. 26:291–300.
Hsieh PH, Veeramah KR, Lachance J, Tishkoff SA, Wall JD, Hammer MF,
Gutenkunst RN. 2016b. Whole-genome sequence analyses of Western
Central African Pygmy hunter-gatherers reveal a complex demo-
graphic history and identify candidate genes under positive natural
selection. Genome Res. 26(3):279–290.
Hublin J-J, Ben-Ncer A, Bailey SE, Freidline SE, Neubauer S, Skinner MM,
Bergmann I, et al. 2017. New fossils from Jebel Irhoud, Morocco and
the pan-African origin of Homo sapiens. Nature. 546(7657):289–292.
Jacobs GS, Hudjashov G, Saag L, Kusuma P, Darusallam CC, Lawson DJ,
Mondal M, et al. 2019. Multiple deeply divergent denisovan ancestries
in Papuans. Cell. [Internet] [cited 2019 May 8];177:1010–1021.e32.
Jobling MA, Tyler-Smith C. 2003. The human Y chromosome: An evolu-
tionary marker comes of age. Nat Rev Genet. 4(8):598–612.
Kong A, Gudbjartsson DF, Sainz J, Jonsdottir GM, Gudjonsson SA,
Richardsson B, Sigurdardottir S, et al. 2002. A high-resolution recom-
bination map of the human genome. Nat Genet. 31:241–247.
Kong A, Thorleifsson G, Gudbjartsson DF, Masson G, Sigurdsson A,
Jonasdottir AA, Walters GB, et al. 2010. Fine-scale recombination rate
differences between sexes, populations and individuals. Nature. 467:
1099–1103.
Lachance J, Vernot B, Elbers CC, Ferwerda B, Froment A, Bodo JM, Lema
G, et al. 2012. Evolutionary history and adaptation from high-cover-
age whole-genome sequences of diverse African hunter-gatherers.
Cell 150:457–469.
Lafferty JD, McCallum A, Pereira FCN. 2001. Conditional random fields:
probabilistic models for segmenting and labeling sequence data. In:
Proc Eighteenth Int Conf Mach Learn; Jun 28–Jul 01; Williams College,
Williamstown, MA.
Lahr MM, Foley RA. 1998. Towards a theory of modern human origins:
geography, demography, and diversity in recent human evolution.
Am J Phys Anthropol. 41:137–176.
Lavachery P. 2001. The holocene archaeological sequence of Shum Laka
rock shelter (Grassfields, western Cameroon). African Archaeol Rev.
18(4):213–247.
Li S, Schlebusch C, Jakobsson M. 2014. Genetic variation reveals large-
scale population expansion and migration during the expansion of
Bantu-speaking peoples. Proc R Soc B Biol Sci. 281(1793):20141448.
Liang M, Nielsen R. 2014. The lengths of admixture tracts. Genetics.
197(3):953–967.
Lopez M, Kousathanas A, Quach H, Harmant C, Mouguiama-Daouda P,
Hombert JM, Froment A, et al. 2018. The demographic history and
mutational load of African hunter-gatherers and farmers. Nat Ecol
Evol. 2(4):721–730.
Lorente-Galdos B, Lao O, Serra-Vidal G, Santpere G, Kuderna LFK, Arauna
LR, Fadhlaoui-Zid K, et al. 2019. Whole-genome sequence analysis of
a Pan African set of samples reveals archaic gene flow from an extinct
basal population of modern humans into sub-Saharan populations.
Genome Biol. 20:77.
Mallick S, Li H, Lipson M, Mathieson I, Gymrek M, Racimo F, Zhao M, et
al. 2016. The Simons genome diversity project: 300 genomes from
142 diverse populations. Nature. 538(7624):201.
Marks SJ, Montinaro F, Levy H, Brisighelli F, Ferri G, Bertoncini S, Batini C,
et al. 2015. Static and moving frontiers: the genetic landscape of
Southern African Bantu-speaking populations. Mol Biol Evol. 32(1):
29–43.
Matise TC, Chen F, Chen W, De La Vega FM, Hansen M, He C, Hyland
FCL, et al. 2007. A second-generation combined linkage-physical map
of the human genome. Genome Res. 17(12):1783–1786.
McDougall I, Brown FH, Fleagle JG. 2005. Stratigraphic placement and
age of modern humans from Kibish, Ethiopia. Nature. 433(7027):
733–736.
Mendez FL, Krahn T, Schrack B, Krahn AM, Veeramah KR, Woerner AE,
Fomine FLM, et al. 2013. An African American paternal lineage adds
an extremely ancient root to the human y chromosome phylogenetic
tree. Am J Hum Genet. 92(3):454–459.
Mondal M, Bertranpetit J, Lao O. 2019. Approximate Bayesian computa-
tion with deep learning supports a third archaic introgression in Asia
and Oceania. Nat Commun. 10:246.
Montinaro F, Busby GBJ, Gonzalez-Santos M, Oosthuitzen O, Oosthuitzen
E, Anagnostou P, Destro-Bisol G, et al. 2017. Complex ancient genetic
structure and cultural transitions in Southern African Populations.
Genetics. [Internet] [cited 2019 May 8];205:303–316.
Narasimhan VM, Rahbari R, Scally A, Wuster A, Mason D, Xue Y, Wright J,
et al. 2017. Estimating the human mutation rate from autozygous
segments reveals population differences in human mutational proc-
esses. Nat Commun. 8:303.
Nielsen R, Akey JM, Jakobsson M, Pritchard JK, Tishkoff S, Willerslev E.
2017. Tracing the peopling of the world through genomics. Nature.
541(7637):302–310.
Patterson N, Moorjani P, Luo Y, Mallick S, Rohland N, Zhan Y,
Genschoreck T, et al. 2012. Ancient admixture in human history.
Genetics. 192(3):1065–1093.
Petr M, P€
a€
abo S, Kelso J, Vernot B. 2019. Limits of long-term selection
against Neandertal introgression. Proc Natl Acad Sci. [Internet] [cited
2019 Jan 16];116:1639–1644.
Plagnol V, Wall JD. 2006. Possible ancestral structure in human popula-
tions. PLoS Genet. [Internet] [cited 2017 Oct 9];2:e105.
Povysil G, Hochreiter S. 2016. IBD sharing between Africans, Neandertals,
and Denisovans. Genome Biol Evol. 8(12):3406–3416.
Pr€
ufer K, Filippo C. d, Grote S, Mafessoni F, Korlevi
c P, Hajdinjak M,
Vernot B, et al. 2017. A high-coverage Neandertal genome from
Vindija Cave in Croatia. Science. [Internet] [cited 2017 Oct 5]:
eaao1887.
Pr€
ufer K, Racimo F, Patterson N, Jay F, Sankararaman S, Sawyer S, Heinze
A, et al. 2014. The complete genome sequence of a Neanderthal from
the Altai Mountains. Nature. 505:43–49.
Ragsdale AP, Gravel S. 2018. Models of archaic admixture and recent his-
tory from two-locus statistics. bioRxiv. [Internet]. [cited 2018 Dec 10]:
489401.
Ragsdale AP, Moreau C, Gravel S. 2018. Genomic inference using diffu-
sion models and the allele frequency spectrum. Curr Opin Genet Dev.
[Internet]. [cited 2019 Jan 22];53:140–147.
Reich D, Green RE, Kircher M, Krause J, Patterson N, Durand EY, Viola B,
et al. 2010. Genetic history of an archaic hominin group from
Denisova cave in Siberia. Nature. [Internet] [cited 2017 Oct 23];468:
1053–1060.
Rightmire GP. 2009. Out of Africa: modern human origins special feature:
middle and later Pleistocene hominins in Africa and Southwest Asia.
Proc Natl Acad Sci U S A. 106(38):16046–16050.
10 C. SANTANDER ET AL.
Sankararaman S, Mallick S, Dannemann M, Prufer K, Kelso J, Paabo S,
Patterson N, Reich D. 2014. The genomic landscape of Neanderthal
ancestry in present-day humans. Nature. [Internet] 507:354–357.
Sankararaman S, Mallick S, Patterson N, Reich D. 2016. The combined
landscape of denisovan and neanderthal ancestry in present-day
humans. Curr Biol. [Internet] 26:1241–1247.
Sankararaman S, Patterson N, Li H, P€
a€
abo S, Reich D. 2012. The date of
interbreeding between neandertals and modern humans. Akey JM,
editor. PLoS Genet. [Internet] [cited 2018 Oct 19];8:e1002947.
Santander CG, Gambron P, Marchi E, Karamitros T, Katzourakis A,
Magiorkinis G. 2017. STEAK: A specific tool for transposable elements
and retrovirus detection in high-throughput sequencing data. Virus
Evol. [Internet] 3:1–12.
Scally A, Durbin R. 2012. Revising the human mutation rate: implications
for understanding human evolution. Nat Rev Genet. 13:745–753.
Scerri E. 2017. The stone age archaeology of West Africa. Oxford Res
Encycl African Hist. [Internet] [cited 2019 Jan 18];1.
Scerri EML, Thomas MG, Manica A, Gunz P, Stock JT, Stringer C, Grove M,
et al. 2018. Did our species evolve in subdivided populations across
Africa, and why does it matter? Trends Ecol Evol. [Internet] [cited
2018 Jul 18];38(8):582–594.
Schlebusch CM, Jakobsson M. 2018. Tales of human migration, admix-
ture, and selection in Africa. Annu Rev Genomics Hum Genet.
[Internet] 19:405–428.
Seguin-Orlando A, Korneliussen TS, Sikora M, Malaspinas AS, Manica A,
Moltke I, Albrechtsen A, et al. 2014. Genomic structure in Europeans
dating back at least 36,200 years. Science. 346(6213):1113–1118.
Sherman RM, Forman J, Antonescu V, Puiu D, Daya M, Rafaels N,
Boorgula MP, et al. 2019. Assembly of a pan-genome from deep
sequencing of 910 humans of African descent. Nat Genet. 51:30–35.
Simons YB, Turchin MC, Pritchard JK, Sella G. 2014. The deleterious
mutation load is insensitive to recent population history. Nat Genet.
46:220–224.
SkoglundP, Thompson JC, Prendergast ME, Mittnik A, Sirak K, Hajdinjak
M, Salie T, Rohland N, Mallick S, Peltzer A, et al. 2017. Reconstructing
prehistoric African population structure. Cell. [Internet]. [cited 2017
Sep 21];171:59.e21–71.e21.
Skov L, Hui R, Shchur V, Hobolth A, Scally A, Schierup MH, Durbin R.
2018. Detecting archaic introgression using an unadmixed
outgroup.Racimo F, editor. PLOS Genet. [Internet] [cited 2018 Oct 1];
14:e1007641.
Slatkin M, Racimo F. 2016. Ancient DNA and human history. Proc Natl
Acad Sci.
Speidel L, Forest M, Shi S, Myers S. 2019. A method for genome-wide
genealogy estimation for thousands of samples. bioRxiv. [Internet]
[cited 2019 Apr 23]:550558.
Stringer C. 2016. The origin and evolution of homo sapiens. Philos Trans
R Soc B Biol Sci. 371(1698):20150237.
Stringer CB, Andrews P. 1988. Genetic and fossil evidence for the origin
of modern humans. Science. 239(4845):1263–1268.
Sudmant PH, Rausch T, Gardner EJ, Handsaker RE, Abyzov A, Huddleston
J, Zhang Y, et al. 2015. An integrated map of structural variation in
2,504 human genomes. Nature. [Internet] 526.
The HapMap Consortium, Altshuler D, Gibbs R, a Peltonen L, Dermitzakis
E, Schaffner S, Yu F, Bonnen PE, et al. 2010. Integrating common
and rare genetic variation in diverse human populations. Nature.
467(7311):52–58.
Theunert C, Slatkin M. 2017. Distinguishing recent admixture from ances-
tral population structure. Genome Biol Evol. [Internet]. [cited 2018
Sep 28];9:427–437.
Tishkoff SA, Reed FA, Friedlaender FR, Ehret C, Ranciaro A, Froment A,
Hirbo JB, et al. 2009. The genetic structure and history of Africans
and African Americans. Science. [Internet] [cited 2017 Sep 18];324:
1035–1044.
Tryon CA, Crevecoeur I, Faith JT, Ekshtain R, Nivens J, Patterson D, Mbua
EN, Spoor F. 2015. Late Pleistocene age and archaeological context
for the hominin calvaria from GvJm-22 (Lukenya Hill, Kenya). Proc
Natl Acad Sci. [Internet] [cited 2019 Jan 18];112:2682–2687.
Vernot B, Akey JM. 2014. Resurrecting surviving Neandertal lineages
from modern human genomes. Science. [Internet] [cited 2017 Oct 23];
343:1017–1021.
Vernot B, Tucci S, Kelso J, Schraiber JG, Wolf AB, Gittelman RM,
Dannemann M, et al. 2016. Excavating Neandertal and Denisovan
DNA from the genomes of Melanesian individuals. Science. 352(6282):
235–239.
Wall JD, Lohmueller KE, Plagnol V. 2009. Detecting ancient admixture
and estimating demographic parameters in multiple human popula-
tions. Mol Biol Evol. 26(8):1823–1827.
Wall JD, Yang MA, Jay F, Kim SK, Durand EY, Stevison LS, Gignoux C,
et al. 2013. Higher levels of Neanderthal ancestry in east Asians than
in Europeans. Genetics. 194(1):199–209.
Wegmann D, Kessner DE, Veeramah KR, Mathias RA, Nicolae DL, Yanek
LR, Sun YV, et al. 2011. Recombination rates in admixed individuals
identified by ancestry-based inference. Nat Genet. [Internet] [cited
2018 Dec 10];43:847–853.
White TD, Asfaw B, DeGusta D, Gilbert H, Richards GD, Suwa G, Howell
FC. 2003. Pleistocene Homo sapiens from Middle Awash, Ethiopia.
Nature. 423(6941):742–747.
Wolpoff MH, Wu XZ, Thorne AG. 1984. Modern Homo sapiens origins: a
general theory of hominid evolution involving the fossil evidence
from East Asia. In: Smith FH, Spencer F, editors. Orig Mod Humans A
World Surv Foss Evid. New York (NY): Alan R. Liss.
Xu D, Pavlidis P, Taskent RO, Alachiotis N, Flanagan C, Degiorgio M,
Blekhman R, et al. 2017. Archaic hominin introgression in africa con-
tributes to functional salivary MUC7 genetic variation. Mol Biol Evol.
[Internet] [cited 2018 Dec 30];34:2704–2715.
Xu D, Pavlidis P, Thamadilok S, Redwood E, Fox S, Blekhman R, Ruhl S,
Gokcumen O. 2016. Recent evolution of the salivary mucin MUC7. Sci
Rep. 6:31791.
ANNALS OF HUMAN BIOLOGY 11
