Archive for April, 2014

Tracking The Advance Of Modernity By The Hybrid Nature Of IBD Segments – The Depletion Of Our Hybrid Heritage Revisited

April 28, 2014 6 comments

The implications of short DNA segment sharing between populations in Africa, Europe and Asia are compelling. Povysil and Hochreiter (2014), Austrian researchers on the Institute of Bioinformatics in Linz, published a preprint on new results that focus on population structure, based on the genetic heritage of humans, Neandertals and Denisovans. Applying HapFABIA, a new method for the identification of very short identity by descent (IBD) segments tagged by rare SNVs (single nucleotide variants) or tagSNVs, for now just based on chromosome 1 data from the 1000 Genomes Project (Phase 1, actually consisting of 1,092 individuals: 246 Africans, 181 Admixed Americans, 286 East Asians, and 379 Europeans), 105,652 valid IBD segments were detected:

The vast majority (101,209) of the detected IBD segments (105,652) is shared by Africans (at least one African possesses the segment), of which 64,894 are exclusively found in Africans. Only 9,003 and 4,793 IBD segments are shared by Europeans and Asians, respectively. 558 IBD segments are exclusively found in Europeans and 1,452 exclusively in Asians. (Povysil et al., 2014)

Mapping these results against the Denisovan genome at 31X coverage (Meyer et al., 2012), and the Altai Neandertal genome at 52X coverage (Prüfer et al., 2014), resulted in a global division into segments that are Neanderthal-like, Denisovan-like, or “Archaic”, ie. matching both the Denisova and Neandertal genome. Intercontinental sharing of segments should thus provide some indication of origin and post-admixture gene flow between the continents.


Researchers have genomes from three types of humans who coexisted during the last Ice Age: Neandertals (orange); Denisovans (blue); and Homo sapiens (yellow). Interbreeding was attested for the Denisovan and the modern human samples.

This focus on the current distribution of purported Neanderthal, Denisovan and “Archaic” IBD segments is a valuable extension to previous previous results described in Hochreiter’s “Rare Haplotypes in the Korean Population” (2012) and “HapFABIA: Identification of very short segments of identity by descent characterized by rare variants in large sequencing data” (2013) that still may not have received due attention or assessment yet. The inflated portion of European-African shared DNA snippets in the African-European-Asian triangle already caught my attention in Expanding Hybrids And The Rise Of Our Genetic Common Denominator. It goes without saying that Asians should have more archaic DNA than Europeans, but despite the total Neanderthal contribution being apparently higher in the Asian genome, I found the accumulated European contribution as compared to the Asian contribution to be consistently higher for rare DNA segments. Due to different filtering and standards the quantitative differences were even more extreme for the 2014 preprint (8,6% European versus 4,6% Asian) than in mentioned previous publications (2012: 8,4% versus 6,1%; 2013: 11,8% versus 6,8%). Remarkably, this combines with an even more massive attestation of European-African shared segments, while Asian segments remain hardly shared by Africans. Reluctance to subsequently consider this shared European-African portion as essentially European seems to be the logical consequence of embracing Out of Africa. However, the African tendency for smaller sized IBD segments being noted in the preprint, doesn’t equally involve the shared European-African portion at all. Hochreiter et al. (2013)’s Figure 8 already revealed that the distribution curves of “Neanderthal” segment sizes don’t differ so much between Asian and European samples, as compared with on average shorter African samples. All the contrary, compared with the Asian reference the European curve in figure 29A of the preprint (2014) reveals a slight left-shift for the private Neanderthal-like IBD segments towards smaller sequences than the overall European “Neanderthal” distribution in figure 28B, suggesting the European IBD segments of the African shared part are on average distinguishably longer than typical for private African IBD segments. According to figure 23 also the overall Asian left-shift for African shared IBD segments is considerably higher than for Europe, though apparently this feature doesn’t involve the Neanderthal-like part. Actually, instead the right-shift of private Asian “Neanderthal” segments is closely linked with the Asian excess of Neanderthal DNA, apparently without this having resulted in a higher count of Asian IBD segments. Thus, the African tendency for smaller IBD segments as being noted in the preprint, seems to be separate from the shared European-African “Neanderthal” distribution! To the effect that a considerable part of African variation can actually be derived from the European distribution.
Actually, Europeans have more variation than Asians as they share more different IBD segments. Indeed, on average (Castellano et al., 2014) European heterozygosity (0.387 per thousand coding bases) is also higher than Asian heterozygosity (0.358), what is far less than African heterozygosity (0.507) and much higher than Neanderthal heterozygosity (0.128). Since ‘about 33% of the derived alleles seen in the Neandertals are shared with the present-day humans and about 19%, 24%, and 24% of the derived alleles seen in the Africans, Europeans, and Asians, respectively, are shared with the Neandertals’ (Castellano et al., 2014), it may be confirmed that Neanderthal-derived variability in Europe should be slightly higher for Europeans than Asians, and considerably higher for Africans.
After having filtered out the IBD segments being already present in the ancestral genome of humans and other primates, the preprint presents an interesting interbreeding analysis. On average more Asians share the few IBD segment, indicating less variation and genetic unity. Indeed, this may confirm East Asians as a rather recent population expansion. Or a bottleneck, if this would also suggest a previous reduction of population size involving some hypothesized similar hybrid population. At least, the East Asian element was not visible yet in Mal’ta Boy (24,000 BP), found pretty deep inside current Mongoloid territory. A late East Asian expansion may even be corroborated by the genetic results of Tianyuan if, like I already suggested in Restoring the Gap, Part 1 – The Delayed Non-African Expansions of Modernity, Tianyuan was actually less East Asian than could be expected from its location. Bottleneck rather than population expansion may possibly be evidenced in Native Americans. Some pre-bottleneck variation of the kind may have been preserved in South East Asia. Unfortunately, these regions were not yet included in the Phase 1 samples of the 1000 Genomes Project. So far I tend to consider the East Asian component in Native Americans and (Island) Southeast Asia rather part of some more recent East Asian expansion.
For reading the IBD length distribution graphs it should be considered that overlap of different distribution curves may have produced a somewhat convolute results, where the peaks that emerge may really derive from superposed distribution curves having density peaks at quite different segment lengths. Like in wave theory distribution curves may combine through superposition to create an interference pattern that feature distinct metrics, ie. the density peaks we see might be deceiving and actually derive from a combination of underpinning “hidden” density peaks. Retrieving actual distribution curves as components of the visible part may be complex and reminds me to the geophysical maps I once had to interpret as a student using deconvolution.
Remember, each distribution curve may be modeled as the statistical result of random segment length reduction into smaller segments as a function of time, and overlap between distribution curves may cause interference. Each distribution curve may decompose into its constituent elements that reflect another admixture event or regions of low recombination rate. Real-life dependencies such as geography, genetic cohesion and natural selection are likely to have influenced, or “distorted”, the statistical distribution curves.
For instance, Asians and Europeans have very similar IBD lengths distributions though both have some minor peaks at longer segments where the distribution curves deviate. Here the chromosome cross-over process may have been retained due to a combination of natural selection, sweep and possibly a higher proportion of homozygous cross-over events. One way to find out about the underpinning reasons for the resulting convolute distribution curve is find a procedure to decompose the overall curves into their constituent components by using proper deconvolution tools and proper assumptions.

The authors of the preprint did some decomposing themselves in the panels of chapter 4 “Analyses of Lengths of IBD Segments”, using a coarse geographic grid. This already yielded an interesting array of distinct distribution curves on a geographical basis that either appear independent or sharing some dependencies. Without any doubt a denser grid of regional human genome samples would help identify more constituent distribution curves, or admixture events, though probably not for each and every minor peak if these were due to uneven effects of natural selection. The best way to assess the nature of all spikes and anomalies of all the distribution curves would be to run the analyses on all human chromosomes separately, what the authors indeed intend to do. While the preprint only presents preliminary results on chromosome 1 of the 1000 Genomes Project, this will soon be completed with results from the other chromosomes for the final version.

figure 23

Figure 23 (Povysil et al., 2014): For Asian there is a clear bias to shorter IBD segments if they are shared with Africans. For Europeans this effect is less clear.

Let’s take a closer look at the various distribution curves for IBD segment lengths of figures 23-31, on the intuitive though – as demonstrated below – precarious assumption of an inverse relation between mean segment size and age. Figure 23 shows a noticeable signal of smaller sequences in Asia for the portion of IBD segments being shared with Africa, while for Europe the signal for an African shared ancestry background is only significant for the smallest segments below 10,000 base pairs (bp). Figure 24 shows this ancestry background signal is stronger for Europe than Asia if European/Asian/African shared segments are included, implying Europe may have played some role in transmitting small segments to Asia. Instead, Figure 25 shows that on average shared Eurasian segments are longer than private European or Asian segments, implying longer segments distributed more easily eg. by the process of natural selection than smaller segments. Where did these longer segments come from, and what pattern followed the early distribution of smaller segments outside Africa?
The next figures distinguish three types of IBD segments, ie. Neanderthal-like, Denisovan-like and “Archaic”, that are neither Neanderthal or Denisovan. For each type the segment length distribution is related to its global geographical location, for either “shared” segments that occur in various continents, or for “private” segments, that occur in a single continent only. Thus, figures 26 and 27 display geographically shared and private Denisovan segments respectively; figures 28 and 29 respectively display shared and private Neanderthal-like segments; and figures 30 and 31 display respectively shared and private “Archaic” segments.
figure 24

Figure 24 (Povysil et al., 2014): If European/Asian/African shared segments are included the African ancestry background signal is stronger for Europe than Asia, implying Europe may have played some role in transmitting small segments to Asia.

figure 25

Figure 25 (Povysil et al., 2014): On average the shared Eurasian segments are longer than private European or Asian segments, implying longer segments distributed more easily, eg. by the process of natural selection, than smaller segments.

The next figures distinguish three types of IBD segments, ie. Neanderthal-like, Denisovan-like and “Archaic”, that are neither Neanderthal or Denisovan. For each type the segment length distribution is related to its global geographical location, for either “shared” segments that occur in various continents, or for “private” segments, that occur in a single continent only. Thus, figures 26 and 27 display geographically shared and private Denisovan segments respectively; figures 28 and 29 respectively display shared and private Neanderthal-like segments; and figures 30 and 31 display respectively shared and private “Archaic” segments.
figure 26

Figure 26 (Povysil et al., 2014): The convoluted overall European distribution curve that match the Denisovan genome (fig.26B) has density peaks at 28,000 and 37,500 bp that may both be shared Eurasian.

figure 27 (Povysil et al., 2014): IBD segments that match the Denisovan genome have a straightforward private European component (fig.27A, density peaks at 12,000 and 46,000 bp).

Figure 27 (Povysil et al., 2014): IBD segments that match the Denisovan genome have a straightforward private European component (fig.27A, density peaks at 12,000 and 46,000 bp).

The convolute state for the European length distribution of IBD segments that match the Denisovan genome (fig.26B) has a straightforward private European component (fig.27A, density peaks at 12,000 and 46,000 bp). This means much of the European Denisovan-like segments are shared by other continents. Instead, the Asian distribution of the private component (fig.27A) is only slightly more spiked (peaks at 27,000, 40,000, 54,000 bp) than the overall Asian distribution (fig.26B), betraying few non-Asian influences here. More specifically, the convoluted overall European distribution curve (fig.26B) has additional peaks at 28,000 and 37,500 bp that may both be shared Eurasian.
Figure 29A shows a convolute state for the distribution of length for private European IBD segments that match the Neanderthal genome (peaks at 15,500 and 26,500 bp), approaching only the Asian distribution (not the African!) when shared segments are included (fig.28B). Apparently, Asians have their wide distribution curve and main density peak at 26,000 bp (nearly) in common with Europe.
figure 28 (Povysil et al., 2014): The distribution curves of IBD segments that match the Neanderthal genome and include geographically shared segments, are similar between Europe and Asia, but not for Africa (fig.28B).

Figure 28 (Povysil et al., 2014): The distribution curves of IBD segments that match the Neanderthal genome and include geographically shared segments, are similar between Europe and Asia, but not for Africa (fig.28B).

figure 29

Figure 29 (Povysil et al., 2014): A private European density peak at 15,500 bp is accompanied by a left shift towards smaller IBD segments, indicating an admixture event that is almost exclusively reminiscent in Europe.

The wide European (Neanderthal) and Asian (Neanderthal and Denisovan) distribution curves having peaks between 26,000 – 28,000 bp have all appearance to be due to contemporaneous main admixture events that involved Neanderthal in Europe and Neanderthal-Denisova hybrids in Asia. Actually, for this range I fail to notice a Denisovan admixture event in Asia that is separate from Neanderthal admixture. At this point, Europe’s density peak at 15,500 for Neanderthal segments (figure 29A) could be due to selective LD latency at work, confined to Europe, since without such possibly selection-driven spikes the distribution patterns of both Europe and Asia would converge to a similar bell-shaped curve. This possibility would corroborate a similar age for Asian and European admixture events. However, since this private European density peak at 15,500 bp is accompanied by a left shift towards an overall private tendency of smaller IBD segments, this feature rather seems to indicate a preceding admixture event that is almost exclusively reminiscent in Europe.
figure 30

Figure 30 (Povysil et al., 2014): The Archaic distribution curve for Europe that peaks at 10,000 bp (fig.30A) looks wide enough to be actually a superposition of constituent elements.

figure 31

Figure 31 (Povysil et al., 2014): The lack of private “Archaic” segments of European signature near the Asian peak at 26,000 bp (fig.31A) is telling about the Asian origin of European shared “Archaic” segments having this particular distribution.

“Archaic” IBD segments, that couldn’t be identified exclusively as Neanderthal nor Denisovan, show private sister peaks for Europeans (at 34,000 bp, fig.31A) and Africans (at 33,000 bp, fig.31B). This apparently reveals the shared African-European (Mediterranean?) origin of a comparatively late admixture event, without any clear Asian affiliation.
If so, “Mediterranean” archaic segments are puzzling. Iberians feature some unexpected correlations, ie. a lower Neanderthal contribution. I am not sure this could be explained by higher rates of African gene flow: AMR populations already show that AFR admixture rather causes a boost of “ancient African” IBD segments than a reduction of verifiable archaic contribution. Unless this same pattern could be isolated from the data in a forthcoming study, it could be postulated the Denisovan and Neanderthal “deficit” in the Iberian population (figures 1 and 2) was actually caused by moving segments into the Archaic category, rather than this deficit being explained by African admixture. Indeed, the latter should have implied a massive introgression of African short IBD segments not unlike the tendency observed in the Admixed American (AMR) sample.
However, the 1000 Genome Project Phase 1 only contains Sub-Saharan African samples as a reference. “Mediterranean” could have a North African affiliation instead, while possible collateral Sub-Saharan influences as attested by the archaic segments in Europe that peak at 5000 bp in Africa can’t be excluded. I wonder if higher Neanderthal contributions attested in Ötzi and some Northern African populations may indeed be due to this apparently late admixture event, and if this wouldn’t exclude Northern Africa as an early agent of modern human genes.
Another European Archaic distribution curve that peaks at 10,000 bp (fig.30A) looks wide enough to be actually a superposition of segments associated with the fully African segments that peak at 5,000 bp (fig.31B) and fully European segments that correspond with the private density peaks for “Neanderthal” segments at 15,500 bp (fig.29A) and “Denisovan” segments at 12,000 (fig.27A, possibly a superposition all by itself). We’ve already seen the slight left-shift of the European distribution curve towards lower lengths compared with the Asian curve. The resolution of these measurement is just not high enough to be conclusive, though it is interesting these European short IBD segments may be connected and reminiscent of what may have been some of the earliest admixture events outside Africa.
Finally, the lack of private “Archaic” segments of European signature near the Asian peak at 26,000 bp (fig.31A) is telling about the Asian origin of European shared “Archaic” segments having this particular distribution.
I would conclude that the private Asian “archaic” density peak at 26,000 bp (fig.31A) fits in nicely with the other distribution curves having density peaks between 26,000 – 27,000 bp (fig.27A, private Asian “Denisovan; fig.29A, private Asian and European “Neanderthal”) that indicate a wide range of contemporaneous Eurasian admixture events involving Neanderthal in Europe and Neanderthal-Denisova hybrids in Asia. Most likely the agent of this admixture were Neanderthal populations that ‘sometimes’ already inherited such hybrid genes all by themselves.
Moreover, a large tail of density peaks could be distinguished that may or may not be relevant for assessing the admixture history of modern humans. It has to be reiterated that the dataset of 1000 Genome Project Phase 1 is limited to a very course grid of East Asians, Europeans, Sub-Saharan Africans plus some mixed American populations, all of whom were so far only analyzed for IBD segments on chromosome 1. For instance, the large chunks of Denisovan DNA discovered in Oceania were not examined. This having said, the largest IBD segments may rather be an indication of long time isolation of already admixed populations, since higher levels of homozygosity imply lower rates of DNA segment size decay by cross-over.
In summary, on top of a Sub-Saharan multitude of small sized IBD segments, largely confined to the African continent, a coarse grid of admixed IBD segment subsets in different grades of size decay may be perceived, or admixture layers on a world-wide scale, that for now could be distinguished as follows:

    1. An early “European” layer of Neanderthal-like, Denisovan-like and “Archaic” IBD segments having a (projected) density peak at about 15,000 bp
    2. A “Eurasian” layer of basically Neanderthal-like IBD segments, supplemented in East Asia by Denisovan-like IBD segments, having a density peak at about 26,000-27,000 bp
    3. A “Mediterranean” subset of “Archaic” IBD segments having a density peak at about 34,000 bp
    4. A “Eurasian” layer of basically Neanderthal-like IBD segments, supplemented in East Asia by Denisovan-like IBD segments, having a density peak at about 40,000-42,000 bp
    5. A European (and possibly shared Asian) tail of “Denisovan” IBD segments having a density peak at about 45,000 bp
    6. An Asian tail of “Denisovan” IBD segments having a density peak at about 54,000 bp
    7. A European tail of “Neanderthal” IBD segments having a density peak at about 56,000 bp

It remains to be seen how many of these categories will persist once the dataset has been extended to include all chromosomes. Once the effects of natural selection and regions of low recombination rate are filtered out, and the subsets of restricted geographic currency tightly assigned to conservative regions where isolated populations could retain their homozygosity and larger IBD segments much longer since admixture – as possibly the case in Northern Africa (subset 3) – the true admixture story could be revealed.
Let’s first try to map these different density peaks on the current insights that concern the expansion of modernity. Purported ‘evidence for a single dispersal […] during a rapid worldwide expansion out of Africa in the Late Pleistocene’ dictated the prevailing view on modern human origins for a long time. Strange enough, most often a coastal route through south Asia was sought for the initial, single dispersal event out of Africa, nowadays dated ‘as early as ~130 ka ago‘ – while ‘in comparative craniometric studies the Levantine series and other early modern humans from Africa have consistently closer affinity to recent Australians than to other modern human populations’. Apparently, something else caused modern African population to be quite different, and other non-Australian populations as well: ‘However, recent genetic studies, as well as accumulating archaeological and paleoanthropological evidence, challenge this parsimonious model. They suggest instead a “southern route” dispersal into Asia as early as the late Middle Pleistocene, followed by a separate dispersal into northern Eurasia’ (Reyes-Centeno et al., 2014). Indeed, ‘a second dispersal through the Levant at ~50 ka and into northern Eurasia’ (Reyes-Centeno et al., 2014)!
Unfortunately, this new view of dual human expansion rather reads as an apologia in defense of the coastal South Asian dispersal hypothesis than that it supplies a clarification on the newly proposed second route through northern Eurasia. Indeed, Indian Indo-European and Dravidian samples ‘exhibit closer genetic and phenotypic affinity to the hypothetical second dispersal descendants (the Japanese, Aeta/Agta, and Central Asian populations)’, though it is remarkable such a detail apparently deterred the authors, declared Recent Out of Africa (ROA) adherents, only now from scoffing at the evidence.
I dare to suggest that as for now unpublished inside information on the 45 ka old Ust-Ishim genome – at 42X coverage, recently announced by Svante Pääbo – from northern Eurasia (near Omsk, southwestern Siberia) may have been decisive for this sudden change of heart. Indeed, if Mal’ta Boy (MA-1, a ~24 ka old non-Mongoloid Siberian whose genome was published by Raghavan et al., 2013) had already been identified as a direct descendent of Ust-Ishim, both early modern hominines may be allocated the same Central Asian affinity, and should thus be recognized as members of the same lineage being held responsible for the replacements of the Australoid elements in India – that should have been there instead according to the “classic” Out of Africa hypothesis.
Now, the present preprint of Povysil et al. (2014) may only give a hint on this first “Australoid” dispersal phase of modern humans. The 1000 Genomes Project Phase 1 they used, simply doesn’t include the South Asian nor Oceanian genomes necessary for an analysis. However, the overall private Asian distribution curves (figs.23-25) all reveal themselves “wider” than the corresponding European distribution curves, thus besides indicating higher densities of the longest IBD segments the Asian curves also demonstrate higher densities of the smallest IBD segments. The same applies for African shared. Fig.23 shows length Asian vs. Asian/African and European vs. European/African. For Asian there is a clear bias to shorter IBD segments if they are shared with Africans and here is a peak at shorter segments. For Europeans this effect is less clear, though there still is an enrichment of short segments around 5,000 bp for African sharing.

As for the second modern human dispersal phase, by now proposed to run through northern Eurasia, the Phase 1 data analysed doesn’t include the proper samples either. East Asia, nor even Japan, isn’t a proper approximation for ancestral Northern Eurasia: so much became clear from the genetic composition of Mal’ta Boy. However, Europe probably is, since this Siberian sample shared considerable drift and ancestry with western Eurasians. Apparent early modern influences in Europe may indeed be attested by eg. the “modern” remains in Kent’s Cavern (KC4), dated by Higham et al. at 44,2- 41,5 kyr cal BP., or the pre-Aurignacian remains in Italy’s Grotta del Cavallo that at Uluzzian culture level had an estimated age of 45-43 ka cal BP (Benazzi, 2012) – both being thus of an age similar to Ust-Ishim! As mentioned before, the complete DNA sequence of the latter is forthcoming, and though ‘Neandertal DNA in the femur suggest that the Ust-Ishim man lived soon after the interbreeding, which Pääbo estimated at 50,000 to 60,000 years ago’ (Gibbons, 2014) there is no mention of Denisovan admixture – what may potentially contradict an eastern identity of the “Denisovan” peak at 12,000 bp in Europe (fig.27A) – in case we should expect any related eastern origin of this kind. At least, the low segment-length distribution of the Denisovan-like IBD segments (whose density peaks at 12,000 bp) suggests an earlier incorporation in the modern gene pool than the main body of Eurasian Neanderthal-Denisovan segments whose densities peak between 26,000 – 27,000 bp. It would be most valuable if any of those “early short” European segments could eventually be recognized in the genome of Ust-Ishim!
In other words, I would identify the first category of IBD segments – having a deconvoluted density peak at about 15,000 bp for the early “European” layer comprising Neanderthal-like, Denisovan-like and “Archaic” IBD segments – as indicative for what currently passes as “the second modern human dispersal through northern Eurasia”. However, I am not sure the private European density peaks at 12,000 for “Denisovan-like” IBD segments (fig.27A) and near 10,000 for “Archaic” IBD segments (fig.31A) should indeed be of this same category. Why, if these components have a completely different distribution in Europe and Asia. The “Archaic” portion is more significant than the Denisovan portion in both Asia and Europe (fig.10, not shown here), what might reduce European “Denisovan” to actually a mere subset of the European “Archaic” category. Such a European connotation of ancient IBD segments could be due as well to a local strain of Neanderthal that possibly inherited from another, more ancestral member of the Denisovan clade, such as the ~400,000-year-old hominine of Sima de los Huesos, Spain, whose mitochondrial DNA indicated an early split (Meyer et al., 2013). It can’t be excluded any of these events included hominines that were actually less related to the Neanderthal and Denisovan reference samples.

Could Neanderthal DNA confirm such kind of ancient local variation? According to Castellano et al. (2014) ‘genetic diversity of Neandertals was lower than that of present-day humans and that the pattern of coding variation suggests that Neandertal populations were small and isolated from one another.’ However, despite the time-gap with the Altai Neanderthal, ‘[t]he fact that the three Neandertals carry longer tracts of homozygosity and differ more from one another than presentday humans within continents further suggests that Neandertals may have been subdivided in small local populations’. Apparently, a smaller long-term effective population for Neanderthal than present-day humans didn’t prevent them from diverging locally. Actually, Neanderthal heterozygosity may vary. It’s considerably higher in the west (0.143 for El Sidrón, Spain, ~49 ka old – or up to 0.163 at 20X coverage) than in the east (0.113, Altai, ~50 ka old), Vindija at ~44 ka old being intermediate (Table S12, Castellano et al., 2014). Apart from possible dating issues, El Sidrón heterozygosity is comparable to (or higher than) Denisovan heterozygosity (0.145 at a minimum age of 48–30 kyr ago – or down to 0.131 at 20X coverage), the latter being attributed both Neanderthal and Archaic admixtures: it was ‘estimated that 17% of the Denisovan DNA was from the local Neandertals. […] Four percent of the Denisovan genome comes from yet another, more ancient, human’ (Pennisi, 2013). Likewise for Europe, inherited archaic admixture of a divergent local source through a “hybrid” variety of Neanderthal is certainly a feasible possibility. Unfortunately, there is no way to predict the physical differences of such a local variety: ‘dental remains from the Sima de los Huesos of Atapuerca, for which ages between 350,000 and 600,000 years have been proposed, already carry Neanderthal-like morphological features that are not seen in the Denisova molar’ (Reich et al., 2010), while the archaic features in the Denisovan tooth specimen may rather derive from archaic introgression (Veeramah & Hammer, 2014). However, if Neanderthals were the “carrier” hominines from where those short non-Neanderthal IBD segments introgressed, then it could be postulated these segments should be found embedded within larger trunks of Neanderthal-like segments. Were they? Again, this can’t be retrieved from the data. Still, it’s crucial for understanding the admixture process and the hominines involved to find some positional correlation between the different types of IBD segments. If no correlation exists whatsoever, the null hypothesis of modern humans being the one and only agent for hybridization between hominine (sub)species can’t be rejected, and still this result may essentially contradict paleogenomic evidence of archaic hybridization that already happened before the appearance of modern humans. Especially the smaller IBD segments that inevitably float around independently within the modern human genomes would appear to evidence a tight association with the modern human lineage as an “agent” for direct, ancient admixture. That the smallest IBD segments being shared across continental populations date back to a time before humans moved out of Africa may be valid for the African distribution, though not necessarily (or much less so) for Asian and European distributions. This has been explained above: much of the smaller shared African segments clearly follow non-African distribution patterns.

IBD-segment size is not necessarily related directly to age, though the overall distribution of each admixture event – or selective sweep – could be assumed to be it is, or should be. The African diversity of small length IBD segments, however, is more difficult to correlate to a certain time-related profile. Hence, private African density peaks at 5,000 bp for Neanderthal-like, Denisovan-like and ‘Archaic’ alike have all appearance to form a modern human evolutionary “baseline” of great time depth. However, if – as suggested by current paleogenetic evidence – homozygous family groups of low effective population sizes were the norm for archaic hominines, and to a lesser extend even for pre-Neolithic modern humans, small length IBD segments could never have formed or thrived in the variety they do nowadays. Hence, one central assumption that emerge concerning the origin of IBD segment variety is hybridization between concurrent hominine lineages rather than balanced selection of concurrent polymorphisms as a result of within group diversification. This, despite of the fact that the opposite, diversification as a function of time, was always one of the cornerstones of the Recent Out of Africa paradigm. However, the latter can’t explain local accumulations of rare tagSNVs on the chromosome. It is quite unlikely such mutations arise independently on various lineages by chance, so the number of tagSNVs on each IBD segment determines its distance from the common genotype, as well as the IBD segment’s detectability. Hochreiter at Dienekes Anthropology Blog: ‘Short segments have often many tagSNVs: e.g. 3kbp segments have 50 SNVs. We found that short segments are tagged as good as long segments, therefore the false discovery rate should not increase for short segments.’
The mere size of small IBD segments marked by such an array of clearly distinguished tagSNVs suggests an already advanced stage of DNA fragmentation by the process of chromosome cross-over throughout generations. Moreover, it suggests heterozygous interaction with DNA of a lineage that essentially lack those tagSNVs, thus in the initial stage of having an appreciable genetic distance. Still, there must be a limit in breaking up segments by cross-over before the IBD simply disappear below detection level. Also, in the same process of fragmentation small pieces of the segment might be chopped off that only contain a few tagSNV markers that thus lost its tight association with the rest of the segment. Probably this will never occur for coding base pairs, though in all other cases this process may be inevitable and found to continuously reduce the IBD segment size, or eventually even to drain the number of detectable IBD segments in store. What happens next, would these chipped off fragments that lost genetic linkage, and residuals of disintegrated segments, including tagSNVs gone astray, simply disappear in a sink hole, or instead be randomly incorporated into common segment variability? Most probably, such fragmented, unrecognizable chomps of “grinded” DNA just recombine with an alternative segment, typically the most common one. Like already asserted in my previous blog articles, this would render a big deal of eg. Neanderthal derived introgressed DNA unaccounted for by the present detection methods that all depart only from much larger units, ie. candidate introgressed sequences.
On the long term this ongoing process of fragmenting the originally quite homozygous strains of introgressed archaic DNA by cross-over thus should have had a dual effect: the reduction of introgressed segments to lower density peak values, having few fixed differences compared to the archaic reference genome; and the integration of fragmented introgressed DNA, chipped off below sequence detection level, into the main body of genome variety. Indeed, whatever their own explanation, Vernot & Akey (2014) detected 6.1X more fixed differences (FD) for the sequences they labeled as “non-introgressed” (17.3 vs. 2.8 FD/Mb) in comparison with the fixed differences between modern humans and Neanderthal for introgressed sequences. This value was claimed as uniform, while larger non-introgressed sequences having more fixed differences were considered significantly depleted of introgressed (Neanderthal) sequences. The latter could thus be instead rather the end result of an advanced state of disintegration of the introgressed sequences, and the subsequent full integration by recombination of smaller introgressed base pairs units into common genome variability.
More radical long term results might be expected with hybrid recombination of coding DNA, either increasing the effects of purifying selection against hybrid heterozygosity, or positive selection of mutated genes that possibly incorporated a recombined mixture of ancestral and derived alleles – to the effect this DNA lost its recoverable archaic correlation. Recombination shuffles the allele content between homologous chromosomes, but needs a certain degree of heterozygosity for producing new combinations of DNA. Since these processes thus are the very result of the disintegration of introgressed DNA, Sankararaman’s (2014) ‘unexpected finding’ that ‘regions with reduced Neanderthal ancestry are enriched in genes’ doesn’t necessarily imply ‘selection to remove genetic material derived from Neanderthals.’. His observation that ‘functionally important elements’ (low B) are ‘significantly correlated to low Neanderthal ancestry’ explicitly relates to the availability of low variability chunks of Neanderthal DNA exempt of any type of recombined code. Indeed, if hybrid recombination spawned new coding variants, successful or deleterious, reduction of introgressed DNA that remained in its original state, ie. without recombination, should be expected.

In conclusion, there is no reason to consider the smallest IBD segments as anything else but preserved hybrid heterozygosity, not unlike the longer segments. Rather than attesting ancient interaction that involves a modern human host population, already rife of the currently known common segments, recombination set the stage for genetic change. Increased heterozygosity by intense hybridization in between a larger quantity of ancient hominine lineages should have accelerated IBD segment size reduction, segment disintegration and recombination. Extensive African hybridization may even have preceded a possible African or non-African backflow of the known common segments that persist in modern human DNA, thus increasing variability even more by continued recombination.
The authors were right to drop earlier attempts to date the admixtures based on the length of IBD segments. Basic assumptions on neutral inheritance and genetic offset with regard to mean recombination patterns between recognizable ancestors without gene flow, can’t be met. Relative dating is equally challenging, now heterozygosity itself determines the speed of disintegration of introgressed sequences. Hybridization and hence increased heterozygosity, may be postulated as symptomatic for a certain level of modernity, though the few available ancient genomes already attest increased heterozygosity for Denisovans and some Neanderthal before the arrival of modernity, while their cultural level can’t be distinguished as inferior compared to contemporary modern humans. Could the mere profusion of hominine lineages in Africa have been the trigger for hybridization to start off, ie. early enough to explain their disproportional availability of the smallest IBD segments? For the various distribution curves, including the smallest IBD segments, to have any dating value at all, it would be most relevant to investigate geographic clines in more detail and to exclude the effects of natural selection. I reckon the distinction between statistic models and the quirks of natural selection will become clear once, as announced in the preprint, data from the other chromosomes will be included in the analysis – and hopefully at a denser grid of genome samples. Once private IBD segments on geographically shared distribution curves may be identified thus to reveal the geographic ranges of the ancestral hominines involved, this will be very helpful in recognizing the migrational patterns of the admixed populations that we all are.

Updated 1st of May, 2014: Map of sequenced hominines added and a reference to the article of Ann Gibbons on the announced genome of Ust-Ishim.


  • About the 1000 Genomes Project
  • Benazzi – The first modern Europeans, 2012, link
  • Castellano et al. – Patterns of coding variation in the complete exomes of three Neandertals, link
  • Dienekes – Sepp Hochreiter on IBD sharing between modern humans, Denisovans and Neandertals, Blog April 14, 2014, link
  • Gibbons – Oldest Homo sapiens Genome Pinpoints Neandertal Input, 2014, link
  • Higham et al. – The earliest evidence for anatomically modern humans in northwestern Europe, 2011, link
  • Hochreiter et al. – Rare Haplotypes in the Korean Population, at ASHG 2012, link
  • Hochreiter et al. – HapFABIA: Identification of very short segments of identity by descent characterized
    by rare variants in large sequencing data, 2013, link
  • Meyer – A High-Coverage Genome Sequence from an Archaic Denisovan Individual, 2012, link
  • Meyer et al. – A mitochondrial genome sequence of a hominin from Sima de los Huesos, 2013, link
  • Pennisi – More Genomes From Denisova Cave Show Mixing of Early Human Groups, 2013, link
  • Povysil et al. – Sharing of Very Short IBD Segments between Humans, Neandertals, and Denisovans, 2014, link
  • Prüfer et al. – The complete genome sequence of a Neanderthal from the Altai Mountains, 2013, link
  • Raghavan et al. – Upper Palaeolithic Siberian Genome Reveals Dual Ancestry of Native Americans, 2013, link
  • Reich et al. – Genetic history of an archaic hominin group from Denisova Cave in Siberia, 2010, link
  • Reyes-Centeno et al. – Genomic and cranial phenotype data support multiple modern human dispersals from Africa and a southern route into Asia, 2014, link
  • Sánchez-Quinto et al. – North African Populations Carry the Signature of Admixture with Neandertals, 2012, link
  • Sankararaman – The genomic landscape of Neanderthal ancestry in present-day humans, 2014, link
  • Veeramah & Hammer – The impact of whole-genome sequencing on the reconstruction of human population history, 2014, link
  • Vernot & Akey – Resurrecting Surviving Neandertal Lineages from Modern Human Genomes, 2014, link