The paper which surveys the relationship of the 40,000 year old Tianyuan sample is finally out in Current Biology, 40,000-Year-Old Individual from Asia Provides Insight into Early Population Structure in Eurasia. There isn’t anything too surprising here. Here is the part of the abstract that presents new finding:
…we generated genome-wide data from a 40,000-year-old individual from Tianyuan Cave, China…We find that he is more related to present-day and ancient Asians than he is to Europeans, but he shares more alleles with a 35,000-year-old European individual than he shares with other ancient Europeans, indicating that the separation between early Europeans and early Asians was not a single population split. We also find that the Tianyuan individual shares more alleles with some Native American groups in South America than with Native Americans elsewhere, providing further support for population substructure in Asia  and suggesting that this persisted from 40,000 years ago until the colonization of the Americas. Our study of the Tianyuan individual highlights the complex migration and subdivision of early human populations in Eurasia.
The Tianyuan sample lived about ~40,000 years ago in China, and it does not seem to have been the direct ancestor of modern East Eurasians. It also seems to have had some relationship to the Australo-Melanesian affiliated population which contributed ancestry to the indigenous peoples of South America. Additionally, it also shares ancestry above what you’d expect with a 35,000 year old Paleolithic European, the GoyetQ116-1 sample, which is found in an Aurignacian context.
There are some direct conclusions that one can infer from this paper. First, as known beforehand the divergence between East Eurasians and West Eurasians has to predate 40,000 years before the present since this sample already shares drift with East Eurasians far more than West Eurasians. In the paper, the authors give an interval of 40,000 to 80,000 years before the present, which seems advised. Remember that “Basal Eurasians” separated before the divergence of East and West Eurasians.
Second, “ghost” populations were common. There are at minimum two ancient Eurasian populations, represented by the Oase1 sample in Romania from 40,000 years ago, and the 45,000 year old Ust’-Ishim from Siberia, who were not closely related to any populations which left descendants today.
Third, the human “family tree” looks more like a human “family bramble.” One of the interesting points in this paper is that Tianyuan shares drift with Goyet, but does not share drift with El-Miron, which seems to be descended in large from a population like Goyet. The key here is to note that Goyet is the closest proxy to some of the ancestors of El-Miron, but it may not be the ancestor at all. So if Goyet-like populations were heterogeneous in relation to East Eurasian, then El-Miron may descend from a group which never mixed with East Eurasians.
This is clear when you read many of these ancient DNA papers closely. The Mal’ta boy was representative of a population which contributed to both Northern Europeans (via Eastern Hunter-Gatherers) and Amerindians, but the deeper results also indicated that the common contributor to these populations was not the Mal’ta population, but related to them. That is, there is no expectation that the sparse sampling of ancient DNA in many regions and epochs will find the ancestral populations, as opposed to groups related to the ancestral populations.
This is a looking-through-the-glass-darkly situation. The true pattern of population relationships of the past needed to be inferred from a finite set of individuals randomly drawn from those populations. If most of those populations left no descendants due to common and repeated local extinction events, then it may be that most of the time we’re going to have to triangulate to the “true” ancestral groups, who left descendants simply due to luck.
Finally, this should really put the nail in the coffin of the idea that we can think of ancient populations are algebraic recombinations of modern populations. Modern groups almost certainly sample only a small part of the distribution of ancient populations.