Linking Participatory, Bayesian, and Agent-Based Modeling Techniques to Simulate Coupled Natural-Human System: A Case Study with Ranchers in Sonora, Mexico

Abstract

Conflicts for water between human use and ecological function have increased in recent years for semi-arid rivers. In the semi-arid Rio Sonora watershed, precipitation varies widely both annually and seasonally. In a harsh semi-arid climate with limited water resources, ranching remains one of the few remaining options to support livelihoods. Water resources are an important component to ranching operations, but are limited and costly. The method a rancher adopts to manage his herd is one of many decisions heavily influenced by both water and money. In an uncertain political and environmental climate, concern rises about the future of the rancher industry, rancher families, and their way of life. As a case study for the Rio Sonora Watershed, a series of workshops were held in the city of Rayon, Sonora, Mexico. The purpose of the workshops was to identify important decisions made by ranchers in the study area, and attempt to predict the likelihood of their choices under various environmental conditions. We applied a new methodology for characterizing decision-making, Bayesian cognitive mapping, which creates a probability of the likelihoods decisions will be made. The construction of a Bayesian cognitive map requires stakeholder involvement in two steps: development of an acyclic graph structure and data collection. The Bayesian cognitive map of Sonoran ranchers is then be applied to an agent-based model to incorporate temporal dynamics. Developing Bayesian cognitive maps of human decision-making is a time-intensive endeavor; however, the approach is holistic, easy to use, encourages stakeholder participation, and values individual variation. This chapter provides an introduction into how Bayesian cognitive maps can be created via participatory approaches and translated into an agent-based model.