By: Hollis Miller, NHRE Intern, with Meghan Mulkerin, Collections Specialist and Research Scientist Contractor.
When I was accepted to the Smithsonian’s Natural History Research Experience (NHRE) Internship, I was so excited. I had always loved the Smithsonian and now I was getting the chance to work there! I first heard about the program through Liz Cottrell, co-director of NHRE, who gave a lecture at my school the year before. Along with my acceptance letter came a short description of the research project that I would be working on, entitled “Computational Modeling of Climate Change and Human Adaptation.” It sounded cool, especially since I am very interested in climate change, but I was a little worried because I knew nothing about computer models.
Thankfully, upon my arrival in the office of Dr. Rogers, I learned that I simply had to learn how to use and apply the models – no computer programming necessary on my part. Even so, it still took me several weeks of reading, playing with the computer model and instruction from Dr. Rogers, before I was ready to officially design and execute my research project.
The computer model I am using is called HouseholdsWorld (designed and tested under NSF grant BCS-0527471), which simulates a nomadic herding society in Mongolia. It is an agent-based model, which is a complex type of computer simulation that allows individual people or households (i.e. the agents) to make their own decisions each day. If you have ever played the Sims, you know that each Sim has its own personality and operates independently, although its decisions can be influenced by other people and social norms. This is basically the same concept as in agent-based modeling, albeit HouseholdsWorld has far less exciting graphics.
Each agent in HouseholdsWorld has its own kinship network and its own life history of birth, marriage, childrearing, herding animals, and death. But even death is not the end to the agent’s participation in the model, because the agent’s living relatives will continue to remember them for several generations, much like you recall who your grandparents or great-grandparents are. This complexity is necessary because it reflects how societies in the real world operate, thus making the model more useful and applicable.
Here's another picture of the goats and sheep, which are typical of the kinds of animals a simulated household in our computer model would own. Who needs a graphical user interface anyway? Dr. Rogers has tons of these pictures from his travels in Mongolia. Photo Credit: Dr. J. Daniel Rogers.
In HouseholdsWorld, the agents are households that move across a 10,000 km2 landscape consisting of dense forests, open grasslands and rivers. Each household is a member of a camp, which is a small living group that generally contains 1 to 4 households, and is also a member of a clan, which is basically an extended lineage of families. The main goal of each household is to find the most abundant grass in order to support its herd of animals. In turn, this supports the household as well, because wealth is measured by the number of animals a household or clan owns. They also use the animals as a source of meat, milk and hides.
Households move in the model according to five social rules:
1) relying on the memory of the camp to return to locations that had good grass in previous years
2) searching for nearby abundant grass
3) staying close to other camp members
4) staying close to other clan members
5) avoiding members of foreign clans
How strictly the households follow these rules can be changed by the researcher.
Bad weather is the main source of trouble for the households. There are two types of bad weather in the model, droughts and zuds, which are severe winter storms. When weather events occur in HouseholdsWorld, they decrease the amount of food available for the animals to eat, which causes both the animal and human populations to suffer. Generally, the weather events are based on historical climatological data from Mongolia, but the model can be programmed to increase the number of weather events, which could be used to approximate modern climate change.
Zuds are very dangerous for herd animals and the people who rely on them. Millions of animals can die from prolonged zuds. Image Credit: FAO: Agriculture and Consumer Protection Department, Chapter II - Cold, semi-arid Asia.
This summer I explored how the social connections people make with one another affect the ability of the society to bounce back after severe weather events. That ability is called resilience. Resilience is similar to flexibility. For example, think about a rubber band. You can stretch a rubber band and when you release the pressure, it returns to its original shape, however, if the rubber band is stretched too far, it will break. Similarly, the more times you stretch a rubber band, the less well it rebounds back to its original shape. Societies can act in similar ways, such as how the Roman Empire expanded throughout many centuries, but eventually reached a point of collapse because the government could no longer control unrest in the far-reaching corners of the empire. Equally so, a society that finds its adaptive capacity stretched too frequently, or several times in quick succession—such as back to back snowstorms—might be less able to cope with the events.
Bringing this idea back to HouseholdsWorld, I wanted to know what happened to the society when I weakened the connections that people have with each other, so I set up three experiments:
1) No Bailout – in this experiment I weakened the economic connectivity that people had with one another by turning off the bailout function in the model. The bailout function is when households give some of their animals to other family members in times of need. However, in this experiment I turned off that feature, so households did not share animals with one another.
2) Alien Distance – in this second experiment I weakened the strength of kinship ties by lessening the number of people that each household recognized as family.
3) Social Rules – in the final experiment I changed the order of importance of the social rules. If you look back to the list of five social rules above, you will see that the first two involve finding the best grass, either by remembering where good grass was in previous years or simply by going to the best grass that is nearby. I switched these two rules in this experiment so that households are more likely to settle for nearby grass than seek out better grass that may be further away. In terms of connections between people, this experiment represents decreased communication among camp members.
In order to have a control group to compare the experiment results to, I also did a baseline run of the model, using real historical and ethnographic data for how nomadic herding societies in Mongolia operate. After comparing each of my experiments to the baseline, I came up with some interesting results that suggest that the type connections people make with one another matters in terms of the population and function of the society.
If you would like to read more about my experiment, you can download my final poster presentation here: Download Miller_NHRE2014 (smaller)
1) No Bailout
a. Population decreased because poorer households could not survive
b. Society lost the social safety net when households no longer shared animals with one another
2) Alien Distance
a. Population increased
b. Fewer family members meant that each household had fewer people to worry about and potentially take care of, so they were able to act more independently – like a modern urban setting where people may be more concerned about their immediate family than their extended family
c. This would also mean that there were many more available marriage partners, making it easier for the society to reproduce
d. Families were on average less wealthy than in either the baseline or the No Bailout experiment
3) Social Rules
a. Population dropped significantly over the generations as birth rates decreased
b. This is probably due to a reduced number of marriage partners because in a smaller group, people are more likely to be related to one another and therefore not allowed to marry based on the customs
c. Households are much wealthier and there is a middle class larger than in any of the other experiments because of less competition for grass
This graph shows the total population for each of the experiments plotted throughout the 330 year time frame. Blue represents the Alien Distance experiment, red is the baseline run, green is the No Bailout experiment, and purple is the Social Rules experiment. The first 250 years of the run are an initialization phase when no extreme weather events occur, hence the fairly stable population during that period. The later section with high variability in the population is the phase when extreme weather events are introduced. These data suggest that connectivity is important to demographic success, as evidenced by the sometimes extreme deviance from the baseline run, especially in the Social Rules experiment.
The data revealed that lessened social connectivity does affect demographic success (i.e. population) and thus resiliency, but the direction of that effect depends on the type of connectivity in question. Humans make choices based on observations, inherited knowledge and the input of other humans. In short, they make the system more complicated and more difficult to predict as even subtle changes can have massive cascading effects. Understanding the dynamics of how societies and their natural environments interact with one another is especially important now, because the Earth system is currently being tested by humans through environmental degradation and climate change. Agent-based models, such as HouseholdsWorld, could be instrumental in learning how best to proceed regarding climate change and how our connections with one another around the globe are going to help and/or hurt us and the societies we have built.
Cioffi-Revilla, Claudio, J. Daniel Rogers, and Maciek Latek. 2010. The MASON HouseholdsWorld Model of Pastoral Nomad Societies. In Simulating Interacting Agents and Social Phenomena: The Second World Congress on Social Simulation. Keiki Takadama, Claudio Cioffi-Revilla, and Guillaume Deffaunt, eds. Pp. 193–204. Agent-Based Social Systems. Tokyo: Springer.
Rogers, J. Daniel, Teresa Nichols, Theresa Emmerich, Maciej Latek, and Claudio Cioffi-Revilla. 2012. Modeling Scale and Variability in Human-Environmental Interactions in Inner Asia. Ecological Modelling 241: 5–14.
Walker, B. H, and David Salt. 2006. Resilience Thinking : Sustaining Ecosystems and People in a Changing World. Washington, DC: Island Press.
This blog post was made possible by support from the National Science Foundation. HouseholdsWorld was created under NSF Grant: BCS-0527471. Hollis Miller's internship through the Natural History Research Experience program was supported by NSF's REU Site, EAR-1062692.