The Ecology of Agglomeration and
the Rise of Chaco Great Houses

Vernon, McCool, Brewer, Codding, Ortman

2023-03-28

kbvernon/saa_2023-chaco_greathouses

The word ‘agglomeration’ is pedantic, a gerund that draws hoity-toity intellectuals to it like moths to a flame. But, it’s also used a lot in urban and economic geography where it refers to “spatial clustering”. Of course, in urban planning, it’s just called ‘density’, so I guess, just call it what you want.

Why should we care about this process? From the perspective of an archaeologist, the answer is surely that density is caught up in virtually every social process you can think of - from inequality to intensification to state formation. From the perspective of a citizen of the modern world, of course, the answer is that we can’t let NIMBYs win. Building for density in cities is one of the most powerful tools we have for promoting sustainable development while also combating social and economic injustice, so understanding it, what causes it and what consequences it might have, is critically important from virtually every angle.

What is a Chaco great house?

What is a great house?

Monumental architecture, typically with one or more kivas surrounded by multi-story room blocks. They are also often found on prominent land forms and they are typically surrounded by dispersed pueblos or farms.

Outlines adapted from Van Dyke (2003), Figure 1.

Why are they built in these areas?

Why this place?

Map displays centroids of HUC10 watersheds with at least one great house.

Why are they built at these times?

Why this time?

Two major building periods in the 11th century, though initial building began much earlier and continued even after Chaco canyon itself faded.

Familiar hypotheses

1. Natural advantages
   Environment offers 🌳 resources, soil, 🌤️ climate️,💧 water!

2. Agglomeration effects
   Density offers 🚚 reduced transport costs, 👷 available labor️, and 🧪 knowledge spill-overs
   

3. But cities are costly, too!
   More people means 🤼‍competition for jobs, 📈 higher rents, 🍷 expensive food, and 😬 more people
   

Agglomeration effects = scaling effects that encourage density.

Clustering is costly!

IFD tells us that habitat suitability is negative density dependent, so how do we get agglomeration effects?

Allee effects are unstable

Positive effects at low density are eventually overwhelmed by competition.

But, unequal effects may be self-reinforcing

Given circumscription, the productivity of an unequal system will increase relative to alternative arrangements.

The key here is circumscription. It guarantees that subordinates will always be worse off by choosing not to participate. At the same time, dominants will benefit from subordination. So the relative advantages of the whole system will tend to increase.

In the context of a sedentary, food production economy, this process would be amplified as the opportunity costs of moving would already be really high.

See Wilson and Codding (2020) The marginal utility of inequality.

Co-regionalization

Can explore this process by looking for times and places where great houses and pueblos co-occur.

Values less than 1 mean observed counts are lower than expected. Values greater than 1 mean observed counts are greater than expected.

Relative rate is ratio of observed to expected counts, estimated using a multivariate space-time model fit with INLA in R.
Method described in Gómez-Rubio et al (2019) Bayesian joint spatio-temporal analysis of multiple diseases.

Unit of analysis

Results

⚠️ These results are preliminary!

1. Still experimenting with hyper-parameters.
2. Just an intercept-only model for now.

Spatial random effects

• Shared effect represents pattern irrespective of type.
  
• Specific effects represent places that deviate from the shared effect.
  
• So, there’s just less of everything in the NE, and even less of great houses there.

Sum of shared and specific is total effect for each.

Temporal random effects

• Shared effect represents pattern irrespective of type.
  
• Specific effects represent times that deviate from the shared effect.
  
• So, there’s just less of everything before 900 and after 1300,
• Slightly more of everything between those times,
• and even more great houses then.

Sum of shared and specific is total effect for each.

Covariance of spatio-temporal weights

Positive covariance means times and places with more of everything also have more of each. Negative covariances means more of one and less of the other.

Weights are contribution of shared effect to relative rates of each.

Conclusion

Conclusion

1. Results are suggestive 🤔 but inconclusive.
   
2. Potential regional trends 🌎 need more consideration.
   
3. Need to incorporate environmental covariates 🌳.
   
4. And so many other things…

Acknowledgments

• Joan Brenner-Coltrain
• Link O’Brennan
• Josh Watts
• R. Kyle Bocinsky