Multi-Agent simulation of climate change Adaptation

The Tropical Andes continue to suffer the most radical climatic changes in South America. These changes generate alterations in its ecosystems, and therefore affect local populations, whose livelihoods are dependent on its diversity and functioning. This is particularly true for rural populations who rely on agriculture as their primary source of food and income. Although the biophysical pathways through which climate change can affect these populations have received extensive scientific attention, it is urgent to study the socioeconomic pathways, at scales that allow the development of vulnerability reduction strategies at the local level. The present study is part of the INCA project (International Network on Climate Change), which is a research network that analyses the local strategies of farmers under a changing climate in the Tropical Andes (Lindner et al. 2017). To contribute to this goal this study investigates climate-related vulnerability and climate change adaptation at local scales. First, the current vulnerability of farm household systems (FHSs) to climate-related hazards is assessed. This is done by looking at determinants that are internal (adaptive capital) and external (climate-related hazards) to the FHSs. Based on the recurrence of internal factors, FHSs are categorized into different groups. These groups are validated by observing the effects of climatic events that are specific to each group. The result of the analysis are different typologies or archetypes of climate-related vulnerability. The analysis adopts an archetype approach and develops methods based on multivariate analysis techniques. Second, the study analyzes the impacts of climate change, expressed as an increase in temperature conditions, at local levels. For this purpose, a multi-agent systems model of land-use/cover change is used, specifically the software package MPMAS. The model is the first attempt at a detailed representation of agents-environment interactions in the framework of climate change in the Tropical Andes. The simulation outcomes report on the adaptation of different farm household groups and the effects of climate change on the agricultural landscape. The research was conducted in selected communities in the Central Andes of Peru. The active integration of empirical data with secondary literature in the application of the research methods provided a suitable way to analyze the vulnerability and adaptive capacity of FHSs in the Tropical Andes in a comprehensive manner. Moreover, the use of participatory assessment techniques to obtain empirical data provided an additional perspective for the analysis and improved the understanding of the problem, contributing to deriving analytical generalizations that could hardly be obtained using only quantitative methods. The research results for the study area identify five archetypes of farm household’s vulnerability to climate-related hazards. For each archetype, distinct vulnerability-creating mechanisms are observed. For example, most vulnerable farm households have a very limited amount of adaptive capital: low levels of off-farm employment, few farm animals, small agricultural area, mostly rainfed, and low use of agro-ecological zones. In addition, they occupy predominantly the higher, and therefore less-productive, agro-ecological zones of the watershed. The analysis also makes it possible to derive spatial and thematic priorities for vulnerability reduction that are specific to each archetype. The modeling approach applied proved to be suitable for simulating the impacts of climate change at the local level. In particular, regarding the explicit simulation of FHSs, the productive landscape, and the way in which they interrelate and change in response to an increase in temperature conditions. The incorporation of heterogeneity and dynamics in the modeled population, the use of optimization techniques to simulate decision making, and the multi-periodicity of the model produce non-linearity, uncertainty and trajectory dependence. In addition, the use of vulnerability archetypes is a novel and robust way of creating a heterogeneous population for the initialization of the model. Simulation results show dynamic changes in the agricultural landscape as temperature increases. The area allocated to corn and olluco expands, while potato and oat areas diminish. Investment in tree plantations is largely unaffected. The effects of rising temperatures on farm households’ welfare show a general persistence of poverty in the study area. However, the effect on FHSs income is predominantly positive, allowing some to improve their food poverty position. The FHSs that manage to benefit from an increase in temperature have, on average, larger agricultural and forest areas, a greater amount of savings in the form of animals, hire more salaried labor and practice more mechanized agriculture than the FHSs whose situation did not improve. The results show that, in addition to the effects of climate change on crop productivity, there are other factors influencing land use decisions that deserve more attention in the analysis of vulnerability and climate change impacts. A better understanding of heterogeneity in climate vulnerability and climate impacts is an important step in meeting this demand