The steel and the star

I recently recorded a podcast with Anders Bergstrom discussing his paper from a few years back, A Neolithic expansion, but strong genetic structure, in the independent history of New Guinea. This got me to thinking a bit about the patterns over the last ~5,000 years within the island, and more broadly. The island of New Guinea is about the size of Texas. That means it’s a bit larger than France. Much of the population is concentrated regionally in the highlands, where a productive system of gardening agriculture dates back 7,000 years.

One of the main results from Anders’ paper is that though New Guinea seems to undergone demographic expansion with the rise of agriculture, there is no evidence of star-phylogenies on the Y chromosome that you see elsewhere in the world, and genetic distances between populations seem to be rather high at a local scale. You’ll have to listen to the podcast (I think it will probably go live in August, so just subscribe) to get the precise way Anders said this, but one thing I got from the conversation is that the cultural and genetic diversity of the highlands is a function of evolution after a Neolithic expansion of a more homogeneous population. That is, I had assumed that the “Papuan” language family was an artificial construct where a bunch of different unrelated dialects was thrown together, but it seems perhaps they have a common genetic origin in an ancestral population that took up taro farming.

This has huge implications for the rate of linguistic evolution of human societies. Like genetic diversity linguistic diversity emerges in the context of cultural parameters. For example, without literacy and widespread trade, one can imagine oral dialects diverging rapidly. Similarly, without gene flow between neighboring populations, they can rapidly differentiate with small effective populations.

One thing I wonder about is how similar this was the spread of swidden agriculture in Europe. Where the Cardial and LBK cultures originally homogeneous, but eventually fractured into small paramountcies? And why and how did the steppe-derived populations roll over these populations so quickly, and give rise to the ‘star-phylogeny’ Y chromosomes we see today?

Bergstrom makes some general allusion to the emergence of metal. But at this point, geneticists usually pass the buck to prehistorians, archaeologists, and economists. What about the rise of metals resulted in the explosion of paternal lineages, and cranked up gene flow between neighboring populations?

The easy way to explain this is that spears and swords of metal impose the rule of the few upon the many. But I think we need to consider the economic consequence of widespread metal (especially iron) in agriculture, where clearing virgin and the second-growth forest became much easier for peasants, and the social and manufacturing systems needed to produce metal weapons and tools at scale. Combined with the mobility of the horse, the shift into the Bronze and Iron Age across Eurasia resulted in the rise of an almost totalitarian and globalist social order in comparison to the localized and decentralized village societies of the Neolithic.

10 thoughts on “The steel and the star

  1. Razib:

    In regard to steppe populations male mediated domination in much of Europe and South Asia, some, such as Rasmussen et al. 2015 and Roscovan et al. 2018, think this was due to yersinia pestis or a similar contagious disease that steppe nomads carried immunity to, but agriculturalist did not.

    Do you have another theory Razib?

  2. i think starvation and marginalization could do the trick too. if these early ag systems were less robust then the agro-pastoralist shock could lead to rapid replacement of native Y lineages.

  3. How about the New World? The Aztec and Inca managed empires without metal weapons, right? Copper is better for tools than weapons, right?

    Might Old World centralization have predated bronze, too?

  4. DK, but was there demographic replacement and star-phylogenies?

    seems to me that these were loose federations of sorts, like most pre-modern societies. but they didn’t seem to impose a cultural revolution…from what i have read the spread of quechua language actually occurred AFTER the spanish conquest (the inca nobility actually retained influence in highlands, though they mixed with spanish creoles).

  5. Very interesting stuff touching on cultural evolution and so I have a few quite (too?) long comments:

    Razib: That is, I had assumed that the “Papuan” language family was an artificial construct where a bunch of different unrelated dialects was thrown together, but it seems perhaps they have a common genetic origin in an ancestral population that took up taro farming.

    It kind of still might be; linguistically, there’s no evidence that they aren’t unrelated dialects. (Unrelated in the sense of “Potentially related at some level, but quite deeply in history, enough that the comparative method cannot reconstruct their relatedness”).

    Actually to push back a little on Anders’ paper I’d question:

    1) How can we actually be confident that there isn’t some composite of agricultural expansion and some much more longstanding population structure (probably not standing since OoA but older than the Holocene) that is a clade to world variation?
    2) Isn’t 10kya also pretty old for an agricultural expansion in New Guinea?
    3) What does the ydna and mtdna structure tell us here? How deep is the structure in the PNG highlands compared to the timing of agriculture? As I recall there is retention of relatively large y-dna population sizes in highland PNG, pre-holocene? In the y, lots of deep rooting divergences splitting immediately under D, C, and K, I think.

    Razib: Where the Cardial and LBK cultures originally homogeneous, but eventually fractured into small paramountcies?

    My impression, as a non-archaeologist and non-expert, is that archaeologically there is more of a tendency to see more geographically widespread and stratified societies in Late Neolithic-Chalcolithic societies of Europe (although this is not totally ubiquitous), and not to see this as a break associated with the time period when steppe ancestry also emerges in much of Europe (the third millennium BCE). Independently of any connection to metallurgy (or at least, very, very weakly associated).

    For example – Neolithic kings to the Staffordshire hoard. Hoards and aristocratic graves in the European Neolithic: the birth of a ‘Barbarian’ Europe?

    p1, Introduction: This paper will discuss the significance of these parallels between earlier and later periods and debate the following hypothesis: that the pairing of elite grave/hoard of precious goods, and the dialectical relationship between them (as it has been described for the Bronze and the Iron Ages), reflects a social configuration that originates in the Neolithic and endures quite unmodified through the Chalcolithic and the Bronze and Iron Ages, as well as into the pre-literary and pre-Christian European societies of the first millennium AD

    p8, DIscussion: The example of western Europe shows that this dialectical relationship, as well as G1-type graves, can arise within a non-metalworking context, among ‘small Neolithic village societies’. Its genesis is thus completely independent of metallurgy and of the major technical innovations of the fourth millennium (namely animal traction and wheeled vehicles), which are also often said to have opened the way to significant social and historical changes.

    Even at the NW periphery, work on megalithic tombs and the non-local scale of coordination around Stonehenge (pre-Beaker period), larger than village scale coordination, seem to indicate fairly wide-ranging political connections (some of the genetics that has come or where abstracts have been given suggest hierarchical elites over time associated with some of these monuments).

    Highland PNG is a very broken territory, by topology, and late Neolithic-Chalcolithic Europe mostly was not.

    Razib: And why and how did the steppe-derived populations roll over these populations so quickly, and give rise to the ‘star-phylogeny’ Y chromosomes we see today?

    My guess would be that the advantage in Northern Europe and Spain and the West and Central Steppe is primarily ecological (to do with a certain sort of set of pastoral practices) that had an explosive advantage (and was that ecological advantage then is immediately responsible for “star shaped” y expansions).

    The spread of IE to the Anatolia and Iran, assuming the Yamnaya/CW hypothesis is correct, probably I would guess, reflects more of a sort of elite recruitment process (lots of J2 lineages it seems and not much post-chalcolithic R1 for the most part). I would guess in these parts, empires and state forming expansions unrelated to IE speaking people significantly leveled linguistic diversity over time, to the scale where elite recruitment by IE outsiders could then swallow them? (Or this may just be the case for Iranian).

  6. I’m not sure how to quantify how star-like a phylogeny is, but some G2a expansions seem relatively star-like? Here is G-P303 from Rootsi et al (2012):

    Either way these G2a guys came out of nowhere to dominate the Y-DNA landscape by the early Neolithic. So perhaps it’s not metals per se, but just some advantageous technological / cultural innovation in general?

  7. @pax, to extend your argument, if disease mattered in the Americas, why do we see there a sex bias in Latin American admixture? Disease can’t have “killed only the men”, ergo, it must have been guns and steel only wot won it, not “Guns, Germs and Steel”!

    It’s pretty obvious what seems a bit wrong with that as an argument; disease isn’t being proposed here as an alternative to conquest, but an alternative to ideas that some superiority in military force (through “Kurgan male warrior ideology”, horse based pastoralist mobility and so on) allowed mobile male steppe ancestry groups to move in and then admix with local females in sex biased and marginalize pre-existing groups.

    (Just as in the case of Iberians moving into the Americas, despite sex biased admixture being obvious, that did not happen primarily by way of a more militarist ideology, metallurgy, advantages of having horses, etc.)

    Now disease being a significant factor may or may not actually be correct (and there are reasons to think it may not have been, and it’s currently not well supported by positive evidence), but it seems kind of wrongheaded to say disease could not be a factor due to sex biased admixture pulses, when exactly the very case we know of most prominently in reality where disease was a factor (and the overwhelming factor at that) shows sex biased admixture pulses.

  8. @PF, agree about not sure about how to quantify star-like exactly to check for within G2a.

    On the topic of other star-like expansions, there was a good paper on reconstructing y-chromosome phylogeny within Sardinians in 2013 – (Dienekes blog post about it) / (sci-hub-able paper).

    Strong star-like expansion in I2a1a1 (nomenclature used in the paper) within Sardinia, modal in the populations, look like could be postdating the Western European star-like expansion within R1b-M269 (adna suggests that’s about 3000 BCE). G2a doesn’t look to show the same pattern, but most of the Sardinian G2a does fall under about three or so branches that look to have experienced bursts at the same time (not a “slow and steady” constant growth rate since earliest Neolithic founding of the island?).

    Ancient dna on Sardinia this year supports R1b-M269 as post Nuragic period (1800BCE-200AD), as doesn’t show up in their set 3000-2000 BCE – So I would guess probably mainland North or Central Italian, where the star expansion of R1b-M269 happened within a North Italian post-Bell Beaker population, then some of this population migrated as a whole to Sardinia.

    The ancient dna paper finds the sole I2a1a1 from that star expansion that is now modal among Sardinians shows up at around 1800BCE, so that fits with a Bronze Age expansion (somewhere!). See – (from the paper’s supplement)

    There’s also that star-like expansion in R1b-V88 in the Sahel around the same time to think about –“Finally, the R-V88 lineages date back to 7.85 kya and its main internal branch (branch 233) forms a “star-like” topology (“Star-like” index = 0.55), suggestive of a demographic expansion. More specifically, 18 out of the 21 sequenced chromosomes belong to branch 233, which includes eight sister clades, five of which are represented by a single subject. The coalescence age of this sub-branch dates back to 5.73 kya, during the last Green Sahara period.”

  9. Jared Diamond notes in Guns, Germs and Steel how fractured and demographically structure Papuans area today with a very strong connection to very different ecological niches.

    We have historical attestation from observers who were neither Hattic nor Hittite (they were Akkadian traders keeping records mostly until the Hittites did so themselves) with no real dog in the fight about Hittite expansion and that story emphasizes the role of metallurgical superiority very heavily, a finding that is largely corroborated by the archaeology (culturally Hittite community show traces of more metal and more advanced metallurgy than anyone else in the region at the time).

    In Europe, horses, metal and more sustainable food production all seem to be key components. First Wave farming seemed to be followed by a slow moving wave of agricultural collapse that spanned to long a time frame to be easily attributed to a climate event. As for disease, in Europe, I think the chain of causation is more likely to be famine and/or malnutrition ==> compromised immune systems ==> devastating disease outbreaks, than an introduction to new disease alone. If disease was a huge factor the initial contact between steppe hunter-gatherers and first farmers in the western steppe ca. modern Moldova would have been an epicenter of massive disease and we don’t see that. Also disease travels more quickly than armies as the North American experience were disease emptied out huge swaths of land before Europeans arrived, and there doesn’t seem to be evidence for a sweep of disease that moved that fast in the late Neolithic or early Bronze Age.

    Linguistically, I don’t think that we have good enough data about Papuan languages to make good conclusions one way or the other. As a matter of linguistic scientific history, the Papuan language “family” was definitely an areal classification rather than a claim that there was a genetic connection. I think it is a fairly safe preliminary guess that Papuan is more diverse than North American Native American languages. And, we’ve had four or five hundred years of really solid scholarship of those languages and while there are now far fewer language isolates than there were earlier on, we are still at perhaps six or eight major North American Native American language families that can be broken down any further. We’ve also made progress in South American languages but that is more fractured now (in part just due to patchy data and lack of scholarship), but it still looks like we’ll have more language families in South American in the end than in North America because the landscape isn’t so open and connected and the language families that have been worked out reasonably well are geographically still quite compact. I’d be quite surprised if the Papuan languages could be broken down to even just two to four language families in a much shorter time period with much less scholarship and patchier data in light of all that. Yes, just like the Founding population of the Americas probably didn’t have more than two or three languages tops and might have had just one 14,000 years ago which has now produced all of the non-Na Dene, non-Inuit languages of the Americas, if you dig deep enough, surely there was some common ground as a starting point for Papuan languages but that starts three times earlier, more like 45,000 years ago.

    Re the agriculture 10kya point — I think that I recall that the data on this was pretty solid although I can’t find a citation right at hand. But, I do think it is worth recalling that, for example, there is lots of comparative anthropology agriculture to indicate that hoe agriculture leads to very different cultural and political outcomes than plough agriculture, and the garden agriculture of the Papuan highlands is more like the former than the latter, and that it was the latter that swept Europe so decisively.

    I do think that it is likely to be a scenario, a bit like the fertile crescent or Meso-America where there was a package of crops and domesticated animals of some sort with each developed in a geographic area and integrated over time into a package complete enough to allow a transition from terrestrial hunting-gathering, without nearly as much demographic mixing. Farmers in the Levant and Iran were very different genetically even though both were using Fertile Crescent agricultural packages that originated partially in Iran, partially in Anatolia and partially in the Levant and was integrated together via thin trade networks. Likewise, geographically corn, beans and rice were domesticated in very different places in Meso-America and then integrated although there wasn’t nearly as much genetic diversity since everyone derived from such a small founding population compounded by serial founder effects. I would think that the process in Papua was similar and that development in Papua may have stunted in part because the total package there may not have been quite as complete at the total packages in Meso-American, the Fertile Crescent, the Sahel and southern India, and East Asia. Maybe it needed more domesticated animal proteins, for example, in addition to not having an analog to horses and a lack of metallurgy. Isolation in the highlands may have also made metallurgy for war purposes with potentially superior outsiders less of an existential priority.


Comments are closed.