Collapse of Complex Societies: Did Drought Kill off the Mayans?

When I wrote my blog about the effect of climate on the rise and demise of complex societies, I had no inkling that there was another paper on this subject in the works. But the latest issue of Science (9.XI.2012) has an article by Douglas Kennett et al., Development and Disintegration of Maya Political Systems in Response to Climate Change.

This article is actually better than the 2011 article I critiqued in my previous blog. Not only the authors present the climate data, but they also have quantitative data on social complexity and warfare. They use two proxies for social complexity: the total number of dated monuments and the  number of urban centers with dated monuments. These data are plotted at 25-year intervals and nicely show the trajectory of social complexity in the Mayan lowlands:

The graph also shows the number of war-related events per 25 years, reflected in the dated inscriptions, and the Inter-Polity Warfare Index (the proportion of war-related events among all recorded events).

So far so good. But after presenting all these neat quantitative data, the article inexplicably stops, and there is not a hint of statistical analysis that I could find, either in the article, or in the Supplementary Online  Material. Is this the case ‘as usual’ in Science, when all ‘hard science’ is lavished on constructing climate data, and none on determining whether there is a statistically defensible effect on social dynamics?! Yet there is nothing particularly difficult in performing such an analysis – if the authors send me the data, I’ll be happy to do it for them. Such an analysis would need to be done thoughtfully (you need to take into account temporal autocorrelations, and there are a couple of alternative model specifications I can think of), but it is routine for any competent statistician. Yet it was not done.

In the absence of statistical analysis I am reduced to eyeballing the data, and it doesn’t look good for the climate hypothesis. In particular, the period of growth in social complexity during the Late Classic era (600-800 CE) coincided with worsening climate. I could go on, but this is precisely the reason we need statistics – to reduce the subjective element in evaluating the hypotheses.

There are some other puzzling aspects of the argument in the paper. I know very little Mesoamerican history, but wasn’t there another cycle associated with Chichen Itza during the 11th and 12th centuries? I was puzzled why there was not another  peak in the data, until I read the supplementary materials, where I  discovered that the inscriptions data were quantified until only 1000 CE. I’d like to see where the heyday of northern Mayan lowlands (Chichen Itza and several other sites) would fall in relation to the eleventh century’s drought, the worst in Mayan history.

I am also curious about the great Mexican drought of 1535. I am not sure how relevant it is that it caused 1 million of deaths in Mexico – most of Mexico is in the much drier belt than Yucatan. Was there a drought in Guatemala in 1535?

The news piece accompanying the article cites archaeologist Andrew Scherer, who notes that the chief crop, maize, requires only 400-600 mm of rain, whereas the Mayan heartland gets 2000-3000 mm annually. Even a severe drought that reduces this amount of rainfall by 40% should  not have catastrophic consequences for Mayan agriculture. Additionally, the shift of Mayan complex societies to northern Yucatan (Chichen Itza etc), which gets less rain than the heartland around Tikal, does not make sense in light of the climate hypothesis.

Furthermore, the Mayan collapse in southern lowlands (around Tikal) resulted in depopulation of over an order of magnitude – 10-fold, perhaps 100-fold.

You cannot achieve such a decline by reducing the carrying capacity by 40%.

It is clear to me that social factors played a much more important role in the Mayan collapse, with climate, at best, a triggering mechanism and, at worst, irrelevant to civilizational dynamics in this part of the world.

So what is the alternative explanation? The truth is that I mostly know Old World history, and very little about Mesoamerica. But this is a blog, so here I go.

First, there was not just one Mayan collapse, but probably a whole series of them. Southern lowlands collapsed in the 9th century, northern lowlands in the 13th century. Before that there was El Mirador collapse around 150 CE. I would not be surprised if, when our data on Mayan cliodynamics gets better, we will find there were societal collapses every 2-3 centuries. In other words, I would expect that the Mayan cultural sphere, just as any other complex society, experienced secular cycles.

As I discuss in my book War and Peace and War not every secular decline leads to civilizational collapse. Such declines interact with longer dynamics involving the waxing and waning of asabiya – social solidarity underlying the ability of the society for concerted collective action. Let’s use as an example the case of Roman Empire, for which we have much more data, and which I wrote about extensively. During the early and middle Republic (especially fourth and third centuries BCE) the Romans were a nation characterized by high asabiya (for reasons explained in War and Peace and War). The Roman state went through several secular disintegrative periods – the crisis of the first century BCE, the crisis of the third century CE, and the final dissolution during the fifth century CE. By the fifth century CE Italy became an ‘asabiya black hole’, an area practically devoid of people capable of cooperating with each other to construct a functioning society.

The collapse of the Roman Empire was as bad in magnitude as the Mayan collapse, especially when considered in relative terms. At its peak in the 2nd century CE Rome (the city, not empire) had over 1 million people. By the seventh century, Rome’s population declined 100-fold, perhaps even 1000-fold.

Ancient Rome truly died. We tend to forget it, but when it got going again, it even started in a different location. The center of new Rome shifted to areas that were previously sparsely built – to Campus Martius (Field of Mars) and Vatican.

Returning to southern Mayan lowlands, the degree of societal collapse there is not particularly remarkable by historical standards (in addition to the Roman example I can cite dozens of others). According to the climate hypothesis, once rainfall amounts increased, we should have seen revival of civilization in southern Mayan lowlands. But it did not happen. Was it because the area around Tikal turned into an asabiya black hole? Asabiya black holes are highly persistent. Southern Italy is still a low cooperation area, fifteen hundred years after the Roman collapse.

In my opinion, when looking for causes and mechanisms of the rise and collapse of complex large-scale societies we should first pay attention to social factors, most importantly the capacity of the society for cooperation. Clearly the  resource base of the society is important and climate fluctuations affect it. But it’s a secondary factor – a modifier, rather than a prime mover.