Geoscience Reference
In-Depth Information
Another example relates to the modeling of the Pueblo population in Mesa
Verde, in Southwest Colorado, between 600 and 1300 [KOH 07, KOH 08]. In this
case, fine-grained information has also been reconstructed detailing the
environmental resources in terms of agricultural potentials (corn), water and hunting
(cervids); these resources vary ever year according to climate variations. The agents
make rational decisions, such as minimizing their caloric expenditure by
maximizing their proximity to the resources. It appears that the first simulations
underestimated rather systematically the evolution of the population: it appeared to
be limited, in the simulations, by the strong decrease of cervids, a consequence of
hunting to cover the protein needs of the population. This deviation between
archaeological estimates and simulation outputs led the authors to formulate a new
hypothesis that involved the conception of turkey domestication. This domestication
probably replaced the protein source that the cervids represented.
The authors have developed a family of models, gradually introducing more
complete interactions, for example exchanges between households when one of
them has suffered a setback when harvesting, following a principle of reciprocity.
The simulations have been able to reproduce a concentration process of the
settlement with formation of villages greater in size. Alternating between
the simulations' results and archaeological reflections has enabled to show that the
abandonment of the region was the result of a combination of climate deterioration
and conflicts. In this example, the MAS is a tool to help reflect on a thematic issue,
the understanding of the functioning of the Pueblo societies and of the factors that
cause them to abandon a territory. In this example, as in the previous one, the MAS
has enabled to show that the classical explanation based on a simple causality of
climate change was insufficient.
4.3.4.
Agent modeling on composite entities
As we emphasized earlier, agents can represent entities other than individuals
(persons, animals, etc.), that is to say aggregated or composite entities, particularly
groups (hunter-gatherer, for example) or settlements (hamlets or cities). We
illustrate such a choice with two examples: one concerns the fall of the Maya
civilization; the other, the dynamics of the European system of cities, to which we
will dedicate more attention. In both cases, the assumption is that the settlements are
indivisible entities and the driving interactions of the systems involved are defined at
the level of these settlements, as nodes in a network of exchanges, and not at the
individual or household level as in the previous examples.
4.3.4.1.
The MayaSim model: simulating the dynamics of the Maya settlement
MayaSim [HEC 13] models a socio-ecological system linking different
environmental dimensions (climate, soil and hydrology) following an approach
