Phylogenetic Relationships (Biomolecules)
Biomolecules, in particular DNA, assist us in generating and testing hypotheses about human evolutionary history. Molecular analyses testing for and then utilizing a local molecular clock can inform us as to the timing of the split between different lineages or populations. When applied to the split between hominins and chimpanzees, for instance, the molecular clock estimates of their divergence date place constraints on interpretations of the growing fossil record from the Late Miocene and Early Pliocene. The pattern and distribution of modern human variation can be used to extrapolate back in time to infer when and where the modern human gene pool arose. Mitochondrial DNA and Y chromosome sequences and markers have been extensively surveyed in populations from around the world. Numerous nuclear loci and other markers, such as microsatellites and Alu insertions, have similarly been sampled and analyzed. More recently, high-throughput massively parallel sequencing technologies have allowed for the characterization of hundreds of human and nonhuman primate complete genomes. The majority of such analyses point toward a relatively recent origin for modern human diversity from a small population in Africa within the last 200 Ka, with a subsequent dispersal into Eurasia less than 100 Ka though there is some debate as to the timing of these events. While analyses of ancient mitochondrial sequences from archaic hominins strongly suggest that archaic females did not contribute to the modern human mitochondrial gene pool, whole-genome sequences of two archaic populations suggest limited interbreeding with modern humans in Eurasia but not Africa. Analyses of modern African genomes suggest that some populations also interbred with an as yet unknown archaic population or populations. Thus, while a complete replacement of archaic populations by African-derived modern humans is no longer fully tenable, only a limited amount interbreeding between anatomically modern human populations and archaic forebears is likely to have taken place.
KeywordsModern Human Much Recent Common Ancestor Nest Clade Analysis Human Evolutionary History Modern Human Population
Over 100 years ago, George Nuttall began his book, Blood Immunity and Blood Relationship, with a discussion of the classification of the order Primates stating, “The persistence of the chemical blood-relationship between the various groups of animals serves to carry us back into geological times, and I believe that we have but begun the work along these lines, and that it will lead to valuable results in the study of various problems of evolution” (Nuttal 1904, p. 4).
Nuttall’s research, carried out shortly after the discovery of blood groups in 1901, was based on qualitative and quantitative measures of the immunological reactions of various proteins in the blood. Immunological approaches were improved and systematically applied to questions about primate evolutionary history extensively in the 1960s through the works of Goodman (1961, 1963) and Sarich and Wilson (1966, 1967). It was also during this period that the concept of the molecular clock was first proposed (Zuckerkandl and Pauling 1962). By the 1970s, research increased directly at the DNA level, though only using approximate methods such as DNA–DNA hybridization and restriction mapping to measure the differences between species, populations, and individuals. This time period also saw the development of chromosomal banding techniques for evolutionary analysis (Chiarelli 1966; Dutrillaux 1979; Yunis and Prakash 1982). These techniques have been further developed and have helped us understand the rearrangements that are both shared and differ between humans and other primates using fluorescence in situ hybridization (FISH) and reciprocal chromosomal painting (Weinberg and Stanyon 1998). Chromosomal techniques are generally only used clinically within modern humans as our level of chromosomal variation is extremely low.
Techniques that directly measure differences at the DNA sequence level have advanced greatly in the last three decades. Earlier studies of species and population differences utilized restriction endonucleases, enzymes that cut a strand of DNA at a particular short sequence pattern, to estimate either genetic distances or to provide phylogenetically informative characters between individuals in the sample. Until 2005, the majority of molecular information was derived using a variety of manual to semiautomated technologies that allowed DNA sequences or microsatellite allele sizes to be relatively rapidly determined. The use of the variants of the polymerase chain reaction (PCR) allowed for minute samples from a variety of biomaterials including blood, saliva, hair, feces, bones, teeth, and other biological materials to be amplified and/or sequenced from only a few molecules of DNA. Amplification was generally followed by gel or capillary electrophoresis to determine sequences or allele sizes.
Practical constraints required the use of either relatively short sequences (hundreds to tens of thousands of bases) or variable markers such as retroelements, microsatellites, and SNPs. One popular class of molecular markers consists of retrotransposable elements, including short interspersed elements (SINEs) and long interspersed elements (LINEs). SINEs, particularly the Alu family, which exists in over 500,000 copies in human genomes, can vary in number and location between individuals and populations (Batzer et al. 1996). Because the absence of an Alu element at a particular location in the genome is the ancestral condition, the shared presence of an element is most likely indicative of common descent. Similarly, the longer LINE elements, which make up over 15 % of human genomes, can be used as markers of common evolutionary descent (Sheen et al. 2000; Boissinot and Furano 2005). Extremely variable short tandem repeats (STRs), also known as microsatellites, have also proven useful in individual identification and parentage assessment and to infer population relationships based on the analysis of the frequencies of different allele sizes (Bowcock et al. 1994). SNPs continue to be used to infer population relationships and evolutionary history (Yu et al. 2002).
Another class of genomic DNA elements is the endogenous retroviruses, which make up a surprisingly large portion of the human genome. Their type, copy number, and positions within the genome vary between populations, so they can provide useful evolutionary markers in the same way as the retrotransposable elements mentioned above (Turner et al. 2001). Extragenomic molecular data from pathogenic and commensal organisms can also be useful in inferring human evolutionary history. Tapeworm, lice, and stomach bacteria sequences have all been used to generate and test hypotheses about human population relationships and migrations (Hoberg et al. 2001; Disotell 2003; Leo and Barker 2005).
Single-nucleotide polymorphisms (SNPs) are typically characterized using DNA microarrays in which DNA probes for variants that are to be identified are usually attached to a solid surface. A DNA sample is then passed over the microarray to allow for hybridization of the source DNA to the probes which are then detected by fluorescence or chemiluminescence indicating a match. Current DNA microarrays can detect up to nearly two million SNPs, copy number variants, or other markers at once.
Beginning in 2005, several new sequencing platforms were developed that generate several orders of magnitude more data than previous methods, at dramatically reduced cost (Mardis 2013). Often referred to as Next Generation (NextGen) technologies, Second Generation (2ndGen) is probably a better term as there will always be a next generation. Though multiple platforms and methods are used, they all basically share several commonalities. First, rather than cloning or amplifying the DNA in advance, different synthetic adapters are added to the source DNA depending upon the platform used, and the fragments are amplified on a solid surface, either glass or a tiny bead, to which the adapters bind. Then the bound amplified fragments are sequenced by adding nucleotides that are detected one at a time as they are incorporated into the amplified clusters. The sequencing and detection step is carried out in a massively parallel manner so that hundreds of thousands to millions of DNA fragments are sequenced simultaneously. One downside to these techniques is that they generate relatively short sequences (from 50 to 60 to a few hundred bases long), with a relatively high error rate. These then need to be assembled into a genome or portion of a genome usually be comparing them to a closely related reference genome.
The completion of the first human genomes in 2001 by Lander et al. (2001) and Venter et al. (2001) was followed in 2005 by a complete draft of the chimpanzee and, in 2007, the macaque genomes, using conventional sequencing approaches allowing for even more sophisticated comparative analyses (Chimpanzee Sequencing and Analysis Consortium 2005; Rhesus Macaque Genome Sequencing and Analysis Consortium 2007). Second-Generation technologies were used to sequence the orangutan (Locke et al. 2011), gorilla (Scally et al. 2012), and bonobo (Prüfer et al. 2012) genomes. These genomes have not only allowed for far more in-depth analyses of the similarities and differences among the various ape lineages but provide information useful in inferring the polarities of molecular characters that vary among humans.
Finally, molecular data are being used to investigate the differences between humans and our primate relatives through studies of copy number variation, gene expression, epigenetics, and other underlying molecular and developmental processes; but these issues are beyond the scope of this review (e.g., Enard et al. 2002; Cheng et al. 2005; Eckhardt et al. 2006; Sudmant et al. 2013; Pääbo 2014).
Molecular studies have been used to draw inferences about the possible Eurasian origin of the African hominids including the ancestor of hominins, though not without controversy (Miyamoto et al. 1998; Stewart and Disotell 1998; Moyà-Solà et al. 1999; Heizmann and Begun 2001; Begun et al. 2012). However, until the discoveries of the Late Miocene hominids and/or hominins including Ardipithecus, Orrorin, and Sahelanthropus (Haile-Selassie 2001; Senut et al. 2001; Brunet et al. 2002), little could be said about origin of the hominin lineage itself. In fact, only its date, approximately 6 Ma, inferred from molecular clock estimates was available (Chen and Li 2001; Wildman et al. 2003). This date estimate is also not without controversy, though significantly older dates put forth by Arnason et al. (1996, 1998) and supported by Tavare et al. (2002) do not appear to be supported by more detailed molecular analyses (Raaum et al. 2005).
Recently a new debate has broken out over estimating divergence dates with molecular data. Typically, comparisons between two lineages for which good fossil evidence provides the age of at least one of them are converted into rates of change per year. This is of course dependent not only on proper placement of a fossil but also its appearance in the record near the point of divergence. Nevertheless, multiple estimates within the catarrhines have relatively consistently suggested a rate of approximately 1.0 × 10−9 bp−1 year−1 (Takahata and Satta 1997; Green et al. 2010). With whole-genome analyses now common and relatively inexpensive, a direct estimation of the human mutation rate can be estimated by examining parent–offspring triads across the whole genome. With a known mutation rate and an estimate of the generation time for the taxa being investigated, divergence dates can be estimated. Coupled with new estimates of generation times for human, chimpanzees, and gorillas (Langergraber et al. 2012), an overall rate of approximately 0.5 × 10−9 bp−1 year−1 or about half the previous rate has been estimated (Roach et al. 2010; 1000 Genomes Project Consortium 2010; Scally and Durbin 2012). However O’Roak et al. (2012) do not find such a slow rate in their whole-genome family triad study. Ho et al. (2011) provide an excellent discussion of the discrepancy in rates of evolution determined at different timescales, at both the morphological and molecular levels.
Fu et al. (2013) provide a third way of estimating molecular rates of evolution, at least for the mitochondrial genome. By sequencing mtDNA in fossils dated to within the last 40,000 years (the reliable range of radiocarbon dating), they can measure the effect of “branch shortening” in which fossil lineages are missing the substitutions that would have occurred had they survived to the present. The number of missing substitutions coupled with the dates of the samples provided remarkably consistent rates. These rates when applied to mitochondrial trees provide divergence estimates that fall within the range of the classically accepted dates discussed below. When high enough quality whole-genome data become available, this technique will be applied to nuclear DNA as well (Green and Shapiro 2013).
It should not come as a surprise that, with whole genomes of the apes available, the interpretation of when speciation occurred has become more complex. For instance, 15 % of the sequences in the western lowland gorilla genome are most similar to those in humans, and 15 % are most similar to those in chimpanzees despite the overall closer relationship of chimpanzees to humans (Scally et al. 2012). Furthermore, 3 % of the human genome is more closely related to chimpanzees or bonobos than they are to each other (Prüfer et al. 2013). These differences in gene lineages from the overall pattern of speciation may be due to several phenomena. In incomplete lineage sorting, ancestral polymorphism or variation in the common ancestral population may be partitioned into descendent populations and ultimately descendent species such that individual genes do not match the species phylogeny. If there is gene flow between the ancestral populations or the incipient species, the gene phylogeny may similarly differ from the species phylogeny. Thus, care must be taken in interpreting molecular divergence dates, especially those derived from concatenated datasets and whole genomes.
With the discovery of these Late Miocene fossils, numerous phylogenetic hypotheses were put forth suggesting their hominin status (Haile-Selassie 2001; Brunet et al. 2002) or that only one was an early hominin and the others were either fossil chimpanzees or gorillas or broadly ancestral to both the human and chimpanzee lineages (Senut et al. 2001). Given the vast amount of molecular data collected to assess the relationships among African apes including humans, “… genetic data can also give us trees that are well enough proportioned to be useful to us as paleontologists and that can provide constraints on our ‘flights of fancy,’ when calibrated by plausible paleontological or other historical data” (Pilbeam 1995). An approximately 6 Ma split between humans and chimpanzees (see Fig. 2), for instance, makes untenable the phylogenetic proposal put forth by Senut et al. (2001), in which Ardipithecus falls along the chimpanzee lineage and Orrorin falls well within the hominin lineage more than 2.5 Ma after a human–chimpanzee divergence (assumed by Senut et al. to have occurred around 8.5 Ma). On the other hand, if the purported much slower rate of evolution is applied, the human–chimpanzee split (and all others) becomes much older. Hawks (2012) points out that if older divergence dates are accepted, then fossils such as Chororapithecus at 10.5 million years old could indeed fall along the gorilla lineage as claimed by Suwa et al. (2007). The gorilla affinities of Chororapithecus have, however, been disputed by others (Gibbons 2007). Overall, given that the phylogenetic approach to estimating divergence dates and the method using missing substitutions from fossils concur and the whole-genome family triad methods give suspiciously ancient divergence estimates for many primate lineages, it is probably best to continue to use the faster rate estimates when inferring dates.
Modern Human Origins
Most studies of blood group allele frequencies and protein polymorphisms carried out in the 1960s and early 1970s that presented their findings in the form of a phylogenetic tree posited a basal split between Asians and an Afro-European cluster. In 1974, Nei and Roychoudury (1974) analyzed 21 blood group systems and 35 polymorphic proteins from which they inferred an initial African versus European–Asian split. In this rather prescient chapter, they extrapolated from estimated amino acid replacement rates and inferred that the basal split between Africans and Eurasians occurred approximately 120 Ka and that Europeans and Asians split around 55 Ka. Few additional studies attempting to infer modern human origins were carried out until the late 1980s.
Two seminal papers published in the late 1980s by Cann et al. (1987) and Vigilant et al. (1989), both working in Allan Wilson’s laboratory at the University of California at Berkeley, inferred a less than 200 Ka African origin for all human mitochondrial DNA (mtDNA) and, by extrapolation, perhaps for all modern populations. Known by various names, the “Mitochondrial Eve” or “Out-of-Africa” hypothesis, will hereafter be referred to as the Recent African Origin (RAO) model. This model stands in contrast to the regional continuity or multiregional (MRE) model in which local populations are thought to derive from the original groups that migrated into the various regions of the Old World over 1 Ma from Africa, with various amounts of gene flow between the different regions ever since (Wolpoff et al. 2000). Cann et al.’s (1987) study was based on phylogenetic inferences drawn from parsimony analysis of high-resolution restriction mapping of the whole mtDNA genome. To counter criticisms of the precision of restriction mapping, the geographical sampling, and the lack of an outgroup in Cann et al.’s original analysis, Vigilant et al. (1989) employed one of the first uses of PCR utilizing hair samples to generate nucleotide sequences in a phylogenetic analysis, followed by a sequence-based analysis with a much larger sample size (Vigilant et al. 1991). Through sequencing the D-loop or control region of the mtDNA genome, they were able to align human sequences with those of a chimpanzee in order to carry out a parsimony analysis rooted by an outgroup. The results were remarkably congruent with those of Cann et al. (1987), in inferring a similar timing and location for the origin of all contemporary human mtDNA: approximately 200 Ka in Africa.
The initial papers of Cann et al. (1987) and Vigilant et al. (1989, 1991) came under criticism for an important analytical flaw. Their parsimony trees suggesting a recent African ancestry for all modern mtDNA were derived from heuristic search strategies that did not find the most parsimonious trees for their respective data sets. Other researchers were able to infer trees without African roots that were more parsimonious (Maddison et al. 1992; Templeton et al. 1992). Since no search strategy is available to guarantee the most parsimonious tree is found for such large data sets, alternative strategies were utilized to infer the root of the modern human mtDNA tree. However, Stoneking et al. (1992) and Sherry et al. (1994) demonstrated that the much greater amount of mtDNA diversity found within Africa compared to outside of it was best explained by a longer period of time for it to accumulate within Africa. Additional smaller data sets chosen to represent the most diverse human sequences possible were also analyzed, and an African origin for modern mtDNA types was inferred (Kocher and Wilson 1991). Additional molecular dating inferences also supported the approximately 200 Kyr time frame inferred to explain human mtDNA diversity (Ruvolo et al. 1994).
A huge number of human complete mitochondrial genome sequences has been collected and subjected to phylogenetic and population genetic analyses. Due to the rapid rate of evolution of the mtDNA genome, short sequences such as those found in the D-loop or control region are not always useful over long timescales and may show spurious clustering due to homoplasy or multiple substitutions at the same site, including saturation of substitutions at a site. One solution when available is to characterize both the fast-evolving control region and several more slowly evolving region of the mtDNA genome to define haplogroups or related lineages of mtDNA haplotypes. In fact, sequencing the complete 16.5 kb mtDNA genome has become commonplace (Ingman et al. 2000; Herrnstadt et al. 2002). By the end of 2013, more than 20,000 complete human mitochondrial genomes had been deposited in GenBank. Such analyses show more geographic partitioning of mtDNA sequences than previous studies based on much shorter sequences revealed.
Various criticisms of using mtDNA sequence data that have been put forth include the possibility of nonmaternal inheritance, selection skewing inferences of geographic structure and rates of evolution, and the presence of recombination. To date, no firm evidence of paternal inheritance has been demonstrated in humans (Bandelt et al. 2005). Furthermore, a mechanism that destroys sperm mtDNA has been discovered, making paternal inheritance even more unlikely (Nishimura et al. 2006). Claims for selection acting strongly upon some human mtDNA lineages, especially related to humans’ entry into colder climates, have been put forth (Mishmar et al. 2003; Ruiz-Pesini et al. 2004). Others interpret the evidence for selection as mainly for purifying selection with only a restricted amount of positive selection in a small portion of the mtDNA genome (Elson et al. 2004). Eyre-Walker and Smith (1999) suggested that mtDNA genomes undergo recombination making inferences about their evolutionary history much less straightforward. The suggestion that mtDNA undergoes recombination has been amply countered by further analyses (Macaulay et al. 1999).
All in all, mtDNA analyses provide a very powerful tool for inferring the evolutionary history of humans and provide a remarkably consistent story as additional data and techniques are brought to bear. Mitochondria, however, only yield a maternal history of the organisms under study. To better understand the overall evolutionary history of any group, both male-specific and biparentally inherited loci are also needed.
The Y chromosome fulfills an analogous paternal role to maternally inherited mtDNA, as the majority of it does not recombine with regions of the X chromosome. This nonrecombining region (NRY) is also referred to as the male-specific portion (MSY) of the Y chromosome. While it was initially thought that little variation was present on the human Y chromosome, increasingly sophisticated molecular analytical techniques have allowed for the discovery of a wealth of variation and potential phylogenetically informative markers. Fortunately, unlike mtDNA, the naming of major lineages or haplogroups was regularized to unambiguously label the clades based upon their phylogenetic structure (The Y Chromosome Consortium 2002).
Another study of 9 mega-base pairs (Mb) from over 1,200 males which using a rate of 0.53 × 10−9 bp−1 year−1 (which is close to the estimates from triad de novo rates) found a MRCA of ~200 Ka (Francalacci et al. 2013). Mendez et al. (2013) discovered a new Y chromosome lineage in an African American individual that is also found in extremely low frequency in some Central Africans that they name A00. They inferred a MRCA date of 338 Ka and suggested either fundamental reassessment of the models for Y chromosome origins or the possibility of archaic hominin introgression. Elhaik et al. (2014) strongly critiqued Mendez et al.’s (2013) analyses on several grounds and recalculate the MRCA to 208 Ka. An analysis of 69 complete Y chromosome sequences estimates the MRCA between 120 and 156 Ka, in line with mtDNA estimates (Poznik et al. 2013). All these Y chromosome studies concur in the African origin of modern diversity, with the larger analyses placing the timing of this origin within the range of the estimated MRCA for mtDNA depending upon which evolutionary rates and fossil calibration points were used.
Pre-Second-Generation sequencing era studies were remarkably consistent. A study of over 10,000 base pairs on a region of the X chromosome with low levels of recombination also is compatible with the mtDNA and Y chromosome results. Kaessmann et al. (1999) found an approximately 535 Ka most recent common ancestor for the alleles of this region. This is broadly consistent with the mtDNA and Y chromosome dates, given that the effective population size of the X chromosome is three times that of the other two loci, so coalescent estimates will be approximately three times as old as well. A 3,000 bp region of the β-globin locus on chromosome 11 yielded an estimate of 750 Ka, which is again broadly consistent with being four times older then mtDNA and Y chromosome dates (Harding et al. 1997).
Nuclear loci such as the compound haplotype composed of an STR locus and an Alu deletion polymorphism on chromosome 12 at the CD4 locus demonstrate a similar pattern to the mtDNA and Y chromosome patterns of variation (Tishkoff et al. 1996). An African origin for the variation at this locus is estimated in the same time frame as that inferred from mtDNA and Y chromosome data with dramatically reduced variation found outside of Africa. Phylogenetic trees derived from numerous microsatellite (STR) loci similarly find their roots within Africa with reduced variation outside of it, though divergence date estimates cannot be easily calculated from such data (Bowcock et al. 1994). A similar pattern is found with SNPs (Yu et al. 2002) and polymorphic Alu insertions (Batzer et al. 1996).
Altogether, the majority of analyses of relatively short molecular sequences and markers suggest a recent African origin for the diversity of modern human genomes (Jorde et al. 2000; Takahata et al. 2001; Excoffier 2002; Satta and Takahata 2002). However, interpretations that contradict a scenario of a recent African origin have been put forth (Harris and Hey 1999; Hawks and Wolpoff 2001; Templeton 2005). Harris and Hey (1999), for instance, interpret PDHA1 (an X chromosome locus) sequence diversity as yielding a 1.86 Ma common ancestor, which would fall outside of the range of estimates derived from the above loci. The PDHA1 analysis has been called into question, due to the probability that the locus is under selection which makes inferences as to coalescence dates difficult (Disotell 1999). Nested clade analyses (Templeton 2002, 2005) suggest more than one major exodus from Africa, an early one, approximately 1.9 Ma, one around 600–700 Ka, and a final one around 100 Ka with evidence for range expansion, long-distance dispersal, and isolation by distance complicating the picture. These analyses have generated a healthy skepticism, especially over the efficacy and accuracy of nested clade analysis (Cann 2002; Knowles and Maddison 2002; Satta and Takahata 2002; Panchal and Beaumont 2010).
Another approach to understanding our evolutionary history comes from examining the particular patterns of molecular variation found throughout the world. Several studies using microsatellite or short tandem repeat (STR) markers have found patterns that are best explained by a series of serial founder effects emanating from Africa outward to other regions of the world. Both Prugnolle et al. (2005) and Ramachandran et al. (2005) find the amount of genetic variation measured in a variety of ways linearly decreases the further populations are from Africa.
Analyses of the several thousandfold increase in amount of human molecular data collected with the advent of microarray SNP typing and Second-Generation (2ndGen) sequencing have corroborated most of the above findings. The Recent African Origin (RAO) model with important caveats discussed in detail below best explains the patterns of human molecular diversity observed today.
Neanderthals, Denisovans, and Other Archaic Hominins
Another opportunity for biomolecules to shed light on hominin phylogeny involves the direct characterization of DNA from fossils. While the earliest analyses of ancient hominin DNA focused on mtDNA, 2ndGen technologies now allow entire genomes to be sequenced. The presence of hundreds to thousands of copies of the mtDNA genome in most cells makes it an ideal candidate for extraction from poor or degraded sources of tissue, such as teeth and bone, including fossils. Ancient DNA (aDNA) analyses are however fraught with difficulties (Cooper and Poinar 2000; Mulligan 2005). Ancient DNA, when present, even under ideal preservation conditions is likely to be damaged and fragmented. More importantly, it is almost certainly contaminated with modern DNA from the environment, excavators, curators, scientists who have handled the material, and molecular laboratory personnel. Extraordinary precautions and techniques need to be carried out to lower the probability of mistakenly accepting such modern contaminants as the sequences from the ancient material (Cooper and Poinar 2000; Mulligan 2005). Despite these difficulties, aDNA provides a unique and important window in the evolutionary history and processes.
By the end of 2007, partial mtDNA sequences from around 18 Neanderthal individuals have been gathered (Krings et al. 1997; Ovchinnikov et al. 2000; Schmitz et al. 2002; Serre et al. 2004; Beauval et al. 2005; Lalueza-Fox et al. 2005; Caramelli et al. 2006; Lalueza-Fox et al. 2006; Orlando et al. 2006; Krause et al. 2007). These sequences form a reciprocally monophyletic clade with the thousands of modern human mtDNA sequences analyzed to date and are estimated to have diverged from modern humans somewhere between 365 and 853 Ka, with an average between 550 and 600 Ka. Even with this preliminary sampling of multiple individuals from different time periods and geographic locations, it is unlikely that a Neanderthal sequence that falls within the modern mtDNA gene pool will be discovered (Krings et al. 2000). Wolpoff (1998) suggested that because the original Feldhofer Neanderthal sequence is more similar to some modern human sequences than some other modern sequences are to other moderns, their mtDNA gene pools overlapped. This however was a misleading analysis as a cladistic analysis of the same data clearly demonstrates a complete separation of Neanderthals and moderns into reciprocally monophyletic clades (Disotell 1999). This observation has been further strengthened by all additional Neanderthal sequences and molecular analyses.
These Neanderthal sequences do not cluster among modern European sequences, as might be expected if they gave rise to the Europeans or extensively interbred with the new migrants into Europe as would be predicted under the multiregional model. However, both Nordborg (1998) and Relethford (2001) point out that different amounts of crossbreeding between Neanderthals and early moderns could have still been possible with the Neanderthal mtDNA lineages having gone extinct due to normal stochastic processes over the last 30 Ka. The Neanderthal sequences do show geographic and temporal structure however. The oldest sequences and eastern-most Neanderthals cluster together to the exclusion of western European samples younger than 48 Ka. Fabre et al. (2009) and Dalén et al. (2012) suggest that the western population may have experienced a bottleneck and population replacement while the eastern populations were more stable through time.
To further test hypotheses of modern human origins, several researchers have attempted to recover and sequence early modern human aDNA. One of these attempts provides a good illustration of the numerous difficulties of aDNA analysis. Adcock et al. (2001) claimed to have recovered an mtDNA sequence from an early modern human fossil skeleton from Australia, known as Lake Mungo III, then thought to date to approximately 60 Ka [this specimen has since been redated to 40 Ka (Bowler et al. 2003)]. The sequence fell outside of the range of modern human mtDNA diversity and clustered with a sequence located on chromosome 11 of the modern human genome, a known mitochondrial pseudogene (numt). Their interpretation was that early modern humans reached Australia before the most recent African exodus that gave rise to the rest of the world’s mtDNA diversity less than 100 Ka. This analysis seems deeply flawed for several reasons. First, the standard protocols suggested to avoid contamination with modern DNA (Cooper and Poinar 2000; Mulligan 2005) were not rigidly followed (Cooper et al. 2001). The sequence is most likely in fact a contaminating numt or has been damaged to yield spurious nucleotide substitutions (Cooper et al. 2001). Smith et al. (2003) point out that it is extremely unlikely for aDNA to have survived at the Lake Mungo site due to the environmental conditions present. Finally, reanalysis with additional Australian and African sequences yields a tree very different from that originally put forth (Cooper et al. 2001). Caramelli et al. (2003) attempted to sequence several early modern specimens from Paglicci Cave in Southern Italy. Their sequences fully fall within the range of modern human sequences. These sequences are therefore either modern contaminants, or early modern mtDNA sequences indeed fall within the range of all modern mtDNA present today.
Serre et al. (2004) therefore took a different approach to investigating early modern human and Neanderthal mitochondrial diversity. They realized that demonstrating the presence of early modern mtDNA at that time was nearly impossible, so they tested five early modern fossil samples along with four Neanderthal samples for the presence of Neanderthal-specific mtDNA motifs. Included among the early human samples were fossils from Vindija, Croatia, and Mladeč (Czech Republic) that have been claimed to be transitional between Neanderthals and early moderns (Wolpoff 1999). Their reasoning was that, if interbreeding occurred between the two groups, the presence of Neanderthal mtDNA in early modern individuals would be more likely since it would not have had a great amount of time to go extinct as Nordborg (1998) and Relethford (2001) potentially proposed for the absence of Neanderthal mtDNA today. Serre et al. (2004) were able to amplify all four Neanderthal samples with “Neanderthal-specific” primers. None of the early modern human fossils yielded amplification products, though they did for more generalized “hominoid-specific” primers, suggesting DNA was present. Furthermore, faunal samples from the same sites all yielded DNA products, suggesting that the conditions at the sites were adequate for the preservation of aDNA.
Currat and Excoffier (2004) carried out a simulation study to model the conditions necessary to detect Neanderthal introgression with mtDNA. They extensively modeled different scenarios of modern human expansion into Europe with competition and admixture with Neanderthals. They found the mtDNA data at the time was only compatible with a less than 0.1 % interbreeding rate that would mean fewer than 120 matings over a 12,000-year period of overlap. One of the most important components of their model demonstrated that at the leading edge of an expanding population where interbreeding is most likely acts like a wave carrying new mutations and introgressed alleles to higher and higher frequencies. This iterative founder effect phenomenon is often referred to as “surfing the wave.”
With development of 2ndGen sequencing methods, the potential of retrieving archaic hominin nuclear DNA improved dramatically. The first such studies (Noonan et al. 2006; Green et al. 2006) sampled a 38 Ka specimen (Vi-80) from Vindija Cave, Croatia, whose mtDNA analysis suggested contained 98 % endogenous Neanderthal DNA and only 2 % modern human contamination. Noonan et al. (2006) directly cloned DNA from the specimen (without amplification) and generated 62 kb of Neanderthal DNA. The sequences had the particular patterns of damage usually found in ancient DNA and were thus of presumed Neanderthal origin. They estimated an average divergence time between their sequences and those of modern humans at 706 Ka with the population split at 370 Ka.
Divergence time estimates for different loci within the genome will almost always be older than the population split, because nearly all populations have some level of variation within them. One test for the populations’ divergence date is to look at variants within each population. For instance, if modern humans and Neanderthals separated a long time ago, Neanderthals would only rarely have the derived version of a modern human variant because if the variant appeared only in the modern lineage, and not in the common ancestor of modern humans and Neanderthals, the derived modern variant would not be found in Neanderthals. On the other hand, if Neanderthals and modern humans split recently or were significantly admixed, then derived modern human variants should be common in the Neanderthal genome. With only three derived modern human variants in their Neanderthal sample, Noonan et al. (2006) concluded that little to no interbreeding had occurred.
Green et al. (2006) using the same sample as Noonan et al. (2006) generated more than a million bases of sequence using a standard 2ndGen technique involving bead-based amplification. They estimated a divergence time of 516 Ka and found 30 % of the SNPs were identical to human-derived alleles. From this, they concluded significant admixture occurred. However, Wall and Kim (2007) demonstrated that much of Green et al.’s (2006) sequence was modern human contamination most likely introduced in the commercial facility utilized for the final sequencing. Thus, as of 2007, there was little evidence of any Neanderthal admixture with modern humans (Hodgson and Disotell 2008).
Using 2ndGen sequencing techniques, Green et al. (2008) generated a complete mtDNA genome from a Neanderthal from Croatia (Vindija 33.16). Briggs et al. (2009) sequenced five additional complete mtDNA Neanderthal genomes including Feldhofer 1 and 2, Vindija 33.25, El Sidrón in Spain, Mezmaiskaya 1, and Mezmaiskaya 2 (only a partial mtDNA genome) using a 2ndGen approach that targeted mtDNA. The younger individuals (38–70 Ka) had only about a third of the variation found in modern humans today, while the oldest sample was most divergent. The Mezmaiskaya 2 individual, despite only being 42 Ka, clustered with the younger western Neanderthals. A complete mtDNA genome from a 30 Ka modern human sample from Kostenki, Russia, was also sequenced using these techniques (Krause et al. 2010a). It clusters inside modern human haplogroup U2, which is common in North Africa, western Asia, and Europe.
With over two-dozen Neanderthal individuals sampled as of 2009, there was no evidence of mtDNA gene flow with modern humans (Currat and Excoffier 2004; Serre et al. 2004; Hodgson and Disotell 2008). With no sign of Neanderthal mtDNA in the tens of thousands of modern humans sampled to date, it was suggested that Neanderthal–human hybrids would have been rare while male hybrids might be sterile (Mason and Short 2011). According to Haldane’s rule, the heterogametic sex in interspecific hybrids will be absent, rare, or sterile (Short 1997).
Improvements to 2ndGen sequencing techniques and new ways of avoiding or at least identifying contamination allowed Green et al. (2010) to successful produce a complete draft Neanderthal genome to 1.3-fold coverage. They generated 5.3 gigabases (Gb) of sequence from three Croatian female Neanderthal samples, using two different techniques that reduced microbial background and enriched the endogenous DNA present in the bones. They were able to cover about 60 % of the Neanderthal genome with less that 1 % error (Green et al. 2010). Along with the complete genomes of five diverse humans and small amounts of sequence from El Sidón, Feldhofer Cave, and Mezmaiskaya Neanderthals, they estimate the population split between humans and Neanderthals at occurred 270–440 Ka. Since modern human mtDNA coalesces around 200 Ka, the Neanderthal–human population split falls within the range of coalescence for nuclear genes (four times that of mtDNA). Therefore, it is expected that many alleles should be shared between humans and Neanderthals.
Green et al. (2010) found an excess of shared alleles between Neanderthals and the genomes they sampled from a French, Han Chinese, and Papua New Guinean individual but not two Africans (San and Yoruba). This suggests that non-African populations share more ancestry than African populations with Neanderthals, indicating some level of admixture. Interestingly enough, the Chinese and Papuan individuals share as much ancestry with Neanderthals as the French individual. By examining the extended haplotypes (regions of the genome that are similar between two individuals), they noted longer haplotypes in Neanderthals and non-Africans than in Africans. This suggests that the admixture was recent, since these regions were not broken up by recombination.
Green et al. (2010) estimated between 1 % and 4 % of non-African modern human alleles introgressed from Neanderthals. Furthermore, this gene flow was within the last 100 Kyr. They proposed two alternate scenarios to explain this admixture. Since the Neanderthal–human split occurred within the time frame in which modern human nuclear DNA diversity developed, if there was ancient substructure, the African modern human population, some African populations could be more closely related to Neanderthals than to others. If such a population or populations also later gave rise to the modern humans that exited Africa, non-Africans and Neanderthals would share more alleles than Neanderthals and other Africans. Given that only two African individuals were sampled, this could not be ruled out. Eriksson and Manica (2012) note that any analyses of potential admixture need to take such substructure into account. Yang et al. (2012) carried out simulations and compared them to data from the Complete Genomics Diversity Panel (Drmanac et al. 2010) and concluded that ancient African substructure does not explain Green et al.’s (2010) finding. Eriksson and Manica (2014) however argue that Yang et al.’s (2012) simulations were inadequate and ancient population substructure cannot be ruled out.
Green et al.’s (2010) favored scenario is that admixture occurred shortly after the modern human exodus from Africa carrying Neanderthal alleles both into western Asian and Europe as well as eastern and southeast Asia. One estimate of the timing of this potential gene flow is between 37 and 80 Ka (Sankararaman et al. 2012). Hodgson et al. (2010) suggested an alternative hypothesis in which limited admixture occurred slightly earlier, when African fauna and early modern humans expanded into western Eurasia around 100 Ka before retreating back into Africa due to climatic shifts. Neanderthal alleles would then be present in low frequency in northeast Africa. Populations from there later migrated out of Africa, either through the Sinai, the Arabian Peninsula, or both, carrying these alleles with them into Eurasia several tens of thousands of years later. These alleles would have become more common due to the iterative founder effect, surfing the wave to higher and higher frequencies in Europe and Asia (Currat and Excoffier 2004, 2011).
Updating their admixture models based on mtDNA (Currat and Excoffier 2004) to whole genomes, Currat and Excoffier (2011) found that under a wide variety of demographic scenarios, very low levels of interbreeding would be necessary to yield 1–4 % admixture. They further speculate that there would have been some kind of avoidance of interspecific mating or lower fitness in hybrids. They estimate that during the entire time and range of overlap, as few a few hundred matings may have occurred. Depending upon when and where those events occurred, different populations and different individuals are likely to share different Neanderthal alleles (Wills 2011). Vernot and Akey (2014) and Sankararaman et al. (2014) infer that up to 20–30 % of the Neanderthal genome is spread out among modern humans, a few nonoverlapping percent at a time.
With the continuing improving methodologies to extract and manipulate ancient DNA and higher and higher throughput 2ndGen sequencing technologies, an entire mtDNA genome followed by a 1.9× coverage full genome was generated from a 50 Ka partial juvenile distal phalanx and a single molar from Denisova Cave in the Altai Mountains of southern Siberia (Krause et al. 2010b; Reich et al. 2010). Using new techniques and remaining fragments of the phalange and some of the original extracted material, Meyer et al. (2012) generated a much higher coverage (31×) Denisovan genome which covers 99 % of the “mappable” genome. Despite being only 100 km from known Neanderthal sites, the Denisovan mtDNA genome is equally distantly related to both Neanderthals and modern humans, diverging around 1 Ma (Krause et al. 2010b). This date is too late to belong to Homo erectus and too early for the common ancestor of modern humans and Neanderthals. The Denisovan nuclear DNA on the other hand clusters with Neanderthals with an average divergence around 640 Ka.
The discrepancy between the mtDNA and nuclear divergence dates between Denisovans and Neanderthals could have two possible explanations. The Denisovans may have hybridized with an as yet unknown archaic hominin that migrated out of Africa after Homo erectus but before the common ancestor of Neanderthals and modern humans. Or, if the population that gave rise to Neanderthals and humans was quite variable and included the Denisovan haplotype, that haplotype may have gone extinct in both the modern human and Neanderthal lineages. This is known as incomplete lineage sorting. To date, neither of these hypotheses can be ruled out.
As interesting as the discovery of potential admixture between Neanderthals and Eurasians is the finding that up to 4.8 % Denisovan alleles are found in Melanesians (Reich et al. 2010). Along with 2.6 % Neanderthal ancestry, Melanesians may have up to 7.4 % of their genome composed of alleles found in archaic hominins. Denisovan alleles are also found in aboriginal Australians, near Oceanic, Polynesian, Fijian, and east Indonesian, but not south Asian or east Asian populations (Reich et al. 2011). Denisovans were also not very diverse, with only about 20 % of modern African and ~30 % of the variation found in Eurasians. There is also a reduced amount of admixed X chromosome alleles potentially suggesting it was mostly male-mediated gene flow. Unfortunately, all archaic genomes generated to date come from females, so we do not know what archaic Y chromosomes look like.
A toe phalanx discovered in 2010 in Denisova Cave has yielded an extremely high-quality (52× coverage) genome of a female Neanderthal (Prüfer et al. 2013). With two high-quality archaic genomes now available, it was possible to estimate that the common ancestor of Denisovans and Neanderthals split from the modern human lineage between 553 and 589 Ka, while the two archaic lineages split approximately 381 Ka. The Altai Neanderthal was relatively inbred and probably derived from a population that went through a severe bottleneck. The higher-quality genomic data also reduce the amount of Neanderthal DNA thought to have introgressed into Eurasians to 1.5–2.1 %. Given that the branch length of the genome derived from the toe is shorter than the one derived from the finger, it is thought to be from slightly older sediments.
Further complicating the picture of admixture among the various hominins of the Middle Pleistocene is the observation that the mtDNA genome sequenced from a specimen from Sima de los Huesos in Spain is related to Denisovan mtDNA (Meyer et al. 2014). The Sima de los Huesos and Denisovan mtDNA genomes diverged around 700 Ka. The femur from which it was derived is classified as Homo heidelbergensis and is from sediments dated to over 300 Ka. Estimating the number of missing substitutions in the mtDNA genome, that is, those that would have occurred since the individual died, yields an expected age of 400 Kyr. Meyer et al. (2014) suggest that the most plausible evolutionary scenario is that the Sima de los Huesos hominins are broadly ancestral to both Denisovans and Neanderthals, which somehow maintained two deeply divergent mtDNA lineages.
Multiple scenarios are thus available to explain the complex patterns of relationships among the various Middle Pleistocene hominins and modern humans. The amounts, directions, and timings of introgression events are under healthy debate. There is still the possibility that what we are calling introgression may be the result of ancient population substructure (Eriksson and Manica 2014). Even the number of lineages involved is debatable. Does the Denisovan mtDNA haplotype represent another potential lineage? Hammer et al. (2011) infer approximately 2 % admixture from an unknown archaic population into some Africans population based on modern diversity in Africa. Similarly, based on whole-genome analyses, Lachance et al. (2012) infer introgression, from an unknown archaic population or populations, into Pygmy and click-speaking Hadza and Sandawe populations. Will east Asian fossils yield more surprises if or when molecular data is generated from them? Will the Flores Island specimens yield DNA with new and improved techniques, despite poor preservation?
Biomolecules have many advantages over morphological characters for phylogenetic analyses. The sheer volume of data potentially available is staggering. More importantly, nontrivial hypotheses regarding homology are generally more robust than those inferred for morphological characters and systems. The independence of characters and traits is more easily achieved at the molecular level, allowing multiple independent phylogenetic hypotheses to be generated and examined for concordance. On the other hand, all molecular phylogenies are necessarily gene trees, which can have different histories from the species or populations in which they reside. With whole-genome sequencing now available, including for a limited number of fossil taxa, the complexities of evolution are more readily apparent. Homoplasy and selection are more easily detectable at the molecular level. With high-quality ancient genomes, molecularly derived estimates of the ages of fossils are now possible. Fossils, on the other hand, can test hypotheses that have been put forth and suggest novel combinations of traits that we are not clever enough to have thought possible. A combination of approaches and techniques will provide us with the best insights into our evolutionary history.
- 1000 Genomes Project Consortium (2010) A map of human genome variation from population-scale sequencing. Nature 467:1061–1073Google Scholar
- Bandelt HJ, Kong QP, Parson W, Salas A (2005) More evidence of non-maternal inheritance of mitochondrial DNA? J Med Genet 42:957–960Google Scholar
- Batzer MA, Arcot SS, Phinney JW, Alegria-Hartman M, Kass DH, Millegan SM, Kimpton C, Gill P, Hochmeister M, Ioannou PA, Herrera RJ, Boudreau DA, Scheer WD, Keats BJ, Deininger PL, Stoneking M (1996) Genetic variation of recent Alu insertions in human populations. J Mol Evol 42:22–29PubMedGoogle Scholar
- Beauval C, Maureille B, Lacrampe-Cuyaubere F, Serre D, Peressinotto D, Bordes JG, Cochard D, Couchoud I, Dubrasquet D, Laroulandie V, Lenoble A, Mallye JB, Pasty S, Primault J, Rohland N, Pääbo S, Trinkaus E (2005) A late Neanderthal femur from Les Rochers-de-Villeneuve, France. Proc Natl Acad Sci U S A 102:7085–7090PubMedCentralPubMedGoogle Scholar
- Caramelli D, Lalueza-Fox C, Vernesi C, Lari M, Casoli A, Mallegni F, Chiarelli B, Dupanloup I, Bertranpetit J, Barbujani G, Bertorelle G (2003) Evidence for a genetic discontinuity between Neanderthals and 24,000-year-old anatomically modern Europeans. Proc Natl Acad Sci USA 100:6593–6597PubMedCentralPubMedGoogle Scholar
- Chimpanzee Sequencing and Analysis Consortium (2005) Initial sequence of the chimpanzee genome and comparison with the human genome. Nature 437:69–87Google Scholar
- Disotell TR (1999) Human evolution: origins of modern humans still look recent. Curr Biol 9:647–650Google Scholar
- Elhaik E, Tatarinova TV, Klyosov AA, Graur D (2014) The ‘extremely ancient’ chromosome that isn’t: a forensic bioinformatic investigation of Albert Perry’s X-degenerate portion of the Y chromosome. Eur J Hum Genet (Epub ahead of print)Google Scholar
- Eriksson A, Manica A (2014) The doubly conditioned frequency spectrum does not distinguish between ancient population structure and hybridization. Mol Bio Evol 31:1618–1621Google Scholar
- Eyre-Walker A, Smith NH (1999) How clonal are human mitochondria? Proc R Soc Lond B Biol Sci 266:477–483Google Scholar
- Herrnstadt C, Elson JL, Fahy E, Preston G, Turnbull DM, Anderson C, Ghosh SS, Olefsky JM, Beal MF, David RE, Howell N (2002) Reduced-median-network analysis of complete mitochondrial DNA coding-region sequences of the major African, Asian, and European Haplogroups. Am J Hum Genet 70:1152–1171PubMedCentralPubMedGoogle Scholar
- Kocher TD, Wilson AC (1991) Sequence evolution of mitochondrial DNA in humans and chimpanzees: control region and a protein-coding region. In: Osawa S, Honjo T (eds) Evolution of life: fossils, molecules and culture. Springer, Tokyo, pp 391–413Google Scholar
- Krause J, Serre D, Viola B, Prüfer K, Richards MP, Hublin JJ, Derevianko AP, Pääbo S (2007) Neanderthals in central Asia and Siberia. Nature 444:902–904Google Scholar
- Krause J, Briggs AW, Kircher M, Maricic T, Zwyns N, Derevianko A, Pääbo S (2010a) A complete mtDNA genome from an early modern human from Kostenki. Russia Curr Biol 20:231–236Google Scholar
- Langergraber KE, Prüfer K, Rowney C, Boesch C, Crockford C, Fawcett K, Inoue E, Inoue-Muruyama M, Mitani JC, Muller MN, Robbins MM et al (2012) Generation times in wild chimpanzees and gorillas suggest earlier divergence times in great ape and human evolution. Proc Natl Acad Sci USA 109:15716–15721PubMedCentralPubMedGoogle Scholar
- Leo NP, Barker SC (2005) Unravelling the evolution of the head and body lice of humans. Parisitol Res 98:44–47Google Scholar
- Macaulay V, Richards M, Sykes B (1999) Mitochondrial DNA recombination – no need to panic? Proc R Soc Lond B Biol Sci 266:2037–2039Google Scholar
- Maddison DR, Ruvolo M, Swofford DL (1992) Geographic origins of human mitochondrial DNA: phylogenetic evidence from control region sequences. Syst Biol 41:111–124Google Scholar
- Mardis ER (2013) Next-generation sequencing platforms. Annu Rev Anal Chem 6:287–303Google Scholar
- Mason PH, Short RV (2011) Neanderthal-human hybrids. Hypothesis 9:1–5Google Scholar
- Mendez FL, Krahn T, Schrack B, Krahn AM, Veeramah KR, Woerner AE, Fomine FL, Bradman N, Thomas MG, Karafet TM, Hammer MF (2013) An African American paternal lineage adds and extremely ancient root to the human Y chromosome phylogenetic tree. Am J Hum Genet 92:454–459PubMedCentralPubMedGoogle Scholar
- Miyamoto MM, Young TS, Ward C, Zihlman AL, Lowenstein JM, Groves C, Covert H, Stewart C-B, Disotell TR (1998) Primate evolution – in and out of Africa (correspondence). Curr Biol 8:R647–R650Google Scholar
- Nuttal GHF (1904) Blood immunity and blood relationships. Cambridge University Press, CambridgeGoogle Scholar
- Pilbeam D (1995) Genetic and morphological records of the Hominoidea and hominid origins: a synthesis. Mol Phylogenet Evol 5:155–168Google Scholar
- Rhesus Macaque Genome Sequencing and Analysis Consortium (2007) Evolutionary and biomedical insights from the rhesus macaque genome. Science 316:222–234Google Scholar
- Ruvolo M, Zehr S, von Dornum M, Pan D, Chang B, Lin J (1994) Mitochondrial COII sequences and modern human origins. Mol Biol Evol 10:1115–1135Google Scholar
- Senut B, Pickford M, Gommery D, Mein P, Cheboi K, Coppens Y (2001) First hominid from the Miocene (Lukeino Formation, Kenya). CR Acad Sci Paris Ser Iia 332:137–144Google Scholar
- Serre D, Langaney A, Chech M, Teschler-Nicola M, Paunovic M, Mennecier P, Hofreiter M, Possnert G, Pääbo S (2004) No evidence of Neanderthal mtDNA contribution to early modern humans. PLoS Biol 2:313–317Google Scholar
- Short RV (1997) An introduction to mammalian interspecific hybrids. J Heredity 88:355–357Google Scholar
- Templeton AR (2005) Haplotype trees and modern human origins. Yrbk Phys Anthropol 48:33–59Google Scholar
- Tishkoff SA, Dietzsch E, Speed W, Pakstis AJ, Kidd JR, Cheung K, Bonné-Tamir B, Santa- chiara-Benerecetti AS, Moral P, Krings M, Pääbo S, Watson E, Risch N, Jenkins T, Kidd KK (1996) Global patterns of linkage disequilibrium at the CD4 locus and modern human origins. Science 271:1380–1387PubMedGoogle Scholar
- Underhill PA, Shen P, Lin AA, Jin L, Passarino G, Yang WH, Kauffman E, Bonne-Tamir B, Bertranpetit J, Francalacci P, Ibrahim M, Jenkins T, Kidd JR, Mehdi SQ, Seielstad MT, Wells RS, Piazza A, Davis RW, Felmand MW, Cavalli-Sforza LL, Oefner PJ (2000) Y chromosome sequence variation and the history of human populations. Nat Genet 26:358–361PubMedGoogle Scholar
- Weinberg J, Stanyon R (1998) Comparative chromosome painting of primate genomes. ILAR J 39:77–91Google Scholar
- Wolpoff MH (1998) Concocting a divisive theory. Evol Anthropol 7:1–3Google Scholar
- Wolpoff MH (1999) Paleoanthropology. McGraw-Hill, BostonGoogle Scholar
- Y Chromosome Consortium (2002) A nomenclature system for the tree of human Y-chromosomal binary haplogroups. Genome Res 12:339–348Google Scholar
- Zuckerkandl E, Pauling L (1962) Molecular disease, evolution, and genetic heterogeneity. In: Kasha M, Pullman N (eds) Horizons in biochemistry. Academic, New YorkGoogle Scholar