Who Were The Late Archaic Hominins Of China?

Were the late archaic hominins of China "Southern Denisovans?"

This author answer the question in the negative, instead arguing for great regional diversification and evolution of distinct H. erectus populations in Asia, arguing that Homo floresiensis is a basal dwarf branch of H. erectus, and that the influences of Denisovans on Late Archaic hominins in China are limited to admixture with various late H. erectus descended populations rather than complete replacement.

I don't find this argument terribly compelling, but it isn't inconsistent with the spotty available evidence either. The article is also worthwhile as a particularly broad and comprehensive review of the relevant evidence over a broader geographic scope than is usually considered in a single paper.

Our traditional scheme during the twentieth century was that Homo erectus had thrived on the vast terrain of eastern Asia since the Early Pleistocene, followed by the appearance of a more advanced but still primitive form of Homo in China during the mid-Middle Pleistocene. Recent discoveries or (re-)recognitions of other archaic hominins, such as Homo floresiensis from an Indonesian island, Neanderthals and the “Denisovans” from southern Siberia, as well as an extremely robust mandible from Taiwan, now open up a different view. By incorporating these latest discoveries, this paper intends to offer a phylogenetic model of diverse archaic Asian hominins distributed from southern Siberia to Southeast Asia and India. On the basis of this new model, I discuss how paleoanthropological data inform the taxonomic identity of the Denisovans and the admixture event with modern humans.

Yousuke Kaifu, "Archaic Hominin Populations in Asia before the Arrival of Modern Humans: Their Phylogeny and Implications for the “Southern Denisovans”," 58 (S17) Current Anthropology S418 (December 2017). https://doi.org/10.1086/694318 (open access).

There is more discussion of this paper before the fold (the author thinks Javanese erectus may have been the Southern Denisovans, although I am skeptical).

But, it is also worth noting that this is post 150 of the year at this blog and is post 1501 since the blog's inception.


The abstract is coy about the author's conclusion, but a full model is laid out in the body text:

The evidence reviewed and reported above offers the following working hypothesis about evolutionary relationships of archaic Asian hominins (see also fig. 2). 

(1) In Java, a regional group of Homo erectus had experienced gradual, slow, continuous evolution from the Early through Late Pleistocene. A series of time-successive demes can be defined within this population lineage, but here I refer to them collectively as “Javanese H. erectus p-deme.” This view is reasonable given the geographic setting. Throughout the Pleistocene, Java experienced episodic range contraction and fragmentation as glacial cycles recurred, and it has been relatively isolated from the Asian mainland (Antón 2002). During the episodes of glacial low sea levels, Java was the southernmost constituent of Sundaland, a huge peninsula extending from the present-day Malay Peninsula to Borneo (Kalimantan) and Bali, whereas it was an island during the warm interglacial epochs, as is today (fig. 1). The past geographic range of the Javanese H. erectus p-deme on the Sundaland or Asian mainland is unclear at present because of the lack of fossil specimens outside Java. 

(2) The lineage of Homo floresiensis goes back to ∼1.0 Ma on the Flores Island. The species originated from a population with early Javanese H. erectus affinities and evolved in total isolation from the Middle Late Pleistocene H. erectus populations in Java, in the isolated island setting. 

(3) The situation in mainland Southeast Asia (the area excluding southern China) is unknown due to the poor fossil evidence from this region. 

(4) In northern China, a H. erectus paleodeme represented by the fossil collection from Zhoukoudian Locality 1 was present during the earlier Middle Pleistocene (Zhoukoudian p-deme). Its geographic range extended southward at least around the region of the Yellow River. 

(5) In southern China, there is currently no evidence to support morphocline or interaction between the northern Chinese and Javanese demes of H. erectus. Instead, the fossils from Hexian and Penghu strongly suggest the presence of another deme that had its own evolutionary history (Hexian/Penghu pdeme). 

(6) The divergence time of the lineages leading to the Javanese H. erectus p-deme and the Zhoukoudian p-deme is unclear, but it can be hypothesized as sometime between Dmanisi Homo, the most primitive Eurasian hominins (1.77 Ma; Lordkipanidze et al. 2013) and the minimum age for the earliest JavaneseH. erectus(1.2Ma). The evolutionary origin of the Hexian/Penghu p-deme is unclear, but it also has deep roots back to the Early Pleistocene if their robust dentognathic morphology was not a product of evolutionary reversal. 

(7) Late archaic Homo in northern China (Dali, Jinniushan, Xuchang, and Xujiayao) may or may not belong to the same p-deme. Their genealogical relationship with the Zhoukoudian p-deme and influence from Afro-European late archaic Homo are debated. At least the extreme hypothesis of complete replacement is unlikely. 

(8) It is possible that Maba and Narmada formed their own p-deme of southern Asian late archaic Homo with some relationship with the genetic pool in western Eurasia. 

(9) Genetic evidence shows eastward dispersal of Neanderthals into the Russian Altai (Okladnikov and other sites) in the late Pleistocene. 

It should be noted that, unlike genetic studies, paleoanthropology has no power to detect subtle introgression between extinct hominin groups. I propose the above model acknowledging this limitation, but with the hope that it serves as a useful guide to infer genetic relationships among the extinct hominin populations in Asia. 

This model is radically different from the classic model that Asian H. erectus was a single evolutionary entity that exhibited clinal geographic variation across the continent from Indonesia to northern China (Antón, 2002; Kidder and Durband, 2004; Wolpoff, 1999; see the phylogenetic scheme illustrated in figure 9.8 of Rightmire 2001) but is more or less similar to the recent scheme illustrated by Stringer and Barnes (2015), where each population lineage is treated as relatively independent from each other.

This is then applied the the vexing data regarding Denisovan admixture in modern humans. The author argues that Siberian Denisovans may simply be Neanderthal hybrids who are admixed with evolved H. erectus, rather than a pure type, and that the pure time individuals in Asia who admixed with modern humans may have been quite different than the Siberian Denisovans. He argues that the hybrid Siberian Denisovans may have a deceptively young TMRCA date as a result, and that instead, an evolved Javan clade of H. Erectus is the source of Denisovan admixture in modern humans.

Personally, I think this hypothesis gives too little weight to how solid the genetic inferences are, but the author's discussion and analysis of this question is as follows:

Among the many unanswered questions about the Denisovans, I here discuss the issue of admixture with modern humans. There is evidence of gene flow from the Denisovans to modern humans, but the observed proportion of the Denisovan genetic component in each extant population shows an unexpected geographic pattern. It was substantially higher in Oceania (particularly New Guinea and Australia, or Sahul) and the eastern Indonesian islands (up to ∼5% of the genomes), but it was minimal in Sunda, mainland Asia, and the Americas (∼0.2%) and was absent in Europe and Africa (Prüfer et al. 2014; Reich et al. 2011). One explanation for this finding is that archaic Denisovans lived over “an extraordinarily broad geographic and ecological range, from Siberia to tropical Asia” and that the Denisovan gene flow occurred not in southern Siberia but in Southeast Asia (Reich et al. 2011). Cooper and Stringer (2013) even suggested that the Denisovans crossed the ocean and later admixed with early modern humans on the islands between the Sunda and Sahul lands (fig. 1).

Although there is ongoing controversy about the timing and routes taken by early modern human populations who dispersed from Africa deep into Asia, there is little doubt that the initial colonizers of Sahul were descendants of populations who migrated using the routes south of the Himalaya Mountains in the Late Pleistocene (Kaifu, Izuho, and Goebel 2015). Therefore, Denisovan admixture detected from the extant Australo-Melanesian populations must have occurred somewhere along this route in South or Southeast Asia. Two fundamental questions emerge: (1) Did the “southern Denisovans” belong to the same p-deme as the original Siberian Denisovans? (2) Who was present in South and Southeast Asia when modern humans first entered this region during the Late Pleistocene?

As for the first question, there is currently no empirical fossil evidence to suggest the presence of a single morphotype that was distributed across the extensive region between Siberia and Southeast Asia (I call this the “Greater Denisovan hypothesis”). More specifically, the two existing Denisovan molars, particularly the more recently reported Denisova 8, are morphologically unique among the known fossil collections from Asia (Sawyer et al. 2015). In my view, the phylogenetic model illustrated in figure 2 points to a rather patchy distribution of several distinct hominin groups (e.g., the Zhoukoudian and Hexian/Penghu p-demes) in each region of Asia. For these reasons, we may need to think of a scenario different from the Greater Denisovan hypothesis. 

The alternative scenario we should consider is that the donor of Denisovan DNA to the Australo-Melanesians belonged to its own p-deme but shared a substantial portion of DNA with the Siberian Denisovans because of their shared ancestry. Another hypothesis that is compatible with this is that the “Siberian Denisovan” had no taxonomic reality, but it was a Neanderthal population admixed with another local primitive hominin. Namely, the reported occurrence of both the Denisovan and Neanderthal DNAs in the same Denisova cave (and possibly even in the same stratigraphic unit in it) does not mean alternating occupations by different p-demes (Sawyer et al. 2015) but is ascribed to individual genetic variation within the same population: some of the members inherited unmixed Neanderthal nuclear DNA but others did not, and the more ancient mtDNA identified for the three Denisovan individuals derived from the other archaic population. The unidentified occupants from the Lower Paleolithic open-air site of Karama near the Denisova cave (Derevianko and Shunkov 2005) may or may not represent such a non-Neanderthal local population in the Altai. 

In either case, the above discussion raises a possibility that the morphotype of the “southern Denisovan”is not necessarily similar to the original Siberian Denisovan, and the former may be represented in the already-known fossil samples from South or Southeast Asia.

With these predictions in mind, we turn to the second question, which is the original occupants on the southern dispersal route by early modern humans. The phylogenetic model illustrated in figure 2 suggests the presence of at least three distinct p-demes on this route in the Late Pleistocene: a late archaic Homo (the Maba/Hathnora p-deme) in the region of India and the mainland Southeast Asia, late H. erectus (Ngandong p-deme) in Java, and H. floresiensis (Liang Bua p-deme) on Flores. On the basis of this observation, I discuss the following different scenarios with regard to the possible donor of the Denisovan DNA, mainly from the perspective of the Asian hominin fossil record. 

A. The Maba/Hathnora p-deme. The divergence time of this hypothetical deme from the lineage leading to Homo sapiens is unknown, but it was probably sometime in the Middle Pleistocene if this deme was related to African or European Middle Pleistocene Homo. This date is consistent with the estimated divergence time of autosomal DNA between the Siberian Denisovans and modern humans (550–765 ka; Prüfer et al. 2014). However, the lack of Denisovan DNA in present-day South and mainland Southeast Asian populations contradicts the hypothesis that the Maba/Hathnora p-deme was the source of the Denisovan DNA seen in present-day Oceania (Cooper and Stringer 2013; Reich et al. 2011). 

B. Javanese Homo erectus. The past geographic range of this long-lasting lineage is unclear, but it is quite possible that it was widespread across the Sunda region (fig. 1). This is probably why Stringer and Barnes (2015) referred to this lineage as “Sunda H. erectus.” If so, the migrating ancestral population of Australo-Melanesians must have come across them, giving a chance of introgression. The divergence time between Javanese H. erectus and H. sapiens is probably much older than 1.2 Ma, but the inconsistency with the reported divergence time between the Siberian Denisovans and modern humans may not be a problem if the former was an admixed population between the local primitive hominins and Neanderthals as discussed above. If this H. erectus deme was the source of the Denisovan DNA found from Oceania, this event might explain some H. erectus–like cranial characters in fossil and extant aboriginal Australians, as noted before by proponents of the multiregional hypothesis of modern human origins (e.g., Hawks et al. 2000; Wolpoff and Lee 2015). The absence of Denisovan DNA in the present-day populations of Sunda can be explained by the later expansion of farming communities from the north and resultant genetic overprinting (Matsumura, Oxenham, and Lan Cuong 2015). 

C. Homo floresiensis. Flores, the home of H. floresiensis, is on the way from Sunda to Sahul. So early H. sapiens probably encountered this diminutive, small-brained, archaic hominin. However, I doubt whether they substantially interbred with each other, primarily because signatures of such admixture are not evident in the skeletal morphology of Australo-Melanesians.

The existing Asian fossil record is admittedly scant, but it is always productive to intensively review what we can and cannot say on the basis of the available materials. Contrary to the classic scheme, the accumulating fossil evidence suggests the presence of regionally different evolutionary lineages or p-demes of archaic Homo in Pleistocene Asia with no clear geographic morphoclines (fig. 2). The current paleoanthropological evidence is also useful to question some aspects of the existing hypotheses regarding Denisovan introgression to modern humans, which derived from the genetic data. In particular, I found that the “Javanese Homo erectus as the southern Denisovans hypothesis” is tenable and worth further investigation. Of course, it is possible that the fossils of the southern Denisovans are yet to be discovered or still remain unrecognized in existing fossil collections. However, closer scrutiny of other possible scenarios is needed before we reach this conclusion.

Two closing points.

First, the author's assessment that "there is little doubt that the initial colonizers of Sahul were descendants of populations who migrated using the routes south of the Himalaya Mountains in the Late Pleistocene" overstates the certainty of that evidence. It is certainly plausible, but it isn't certain, and admixture in North Asia could solve other puzzles in early human migration. There have been papers making a quite plausible genetic phylogeny case for a Northern route migration to Southeast Asia. 

Second, I don't think that you you need "the presence of a single morphotype that was distributed across the extensive region between Siberia and Southeast Asia" to get the results that we see. It is also plausible that Denisovan populations filtered into Southeast and East Asia and Oceania primarily in ecological niches where they had the greatest comparative advantage relative to H. Erectus, such as in highlands and on islands, while H. Erectus continues to be predominant in environments they found to be more favorable, perhaps in mainland lowlands. If H. Erectus were incompatible with modern humans reproductively, but Denisovans were compatible, at least somewhat, with modern humans, strategically settled Denisovans could still leave a genetic impact on the first wave of modern humans that H. Erectus did not.