Current and potential distribution of selected savanna tree species in the South African Lowveld by applying agent-based modeling
In this dissertation, the impacts on trees by elephants in the Kruger National Park and the use of trees by firewood collectors outside the protected area were compared for asimulation period from 2010 to 2050. A first research focus aimed at a deeper understanding of the ecological adaptation strategies of different tree species. For this purpose, partial order methods were applied to an existing data set. Three very different strategies were identified and each assigned to a representative tree species. Acacia nigrescens invests in the root system for nitrogen production, while Combretum apiculatum stores carbon in different parts of the plant as a defense strategy against browsers. Whereas, Sclerocarya birrea with its large average leaf area of easily digestible leaves, on the other hand, leads to a high photosynthesis rate and nutrient input due to the longer residence time of large browsers. At the same time, all these tree species are characterized by their ability to withstand drought, fire, and recurrent tree cutting. To develop an agent-based model, the agents and the environment had to be conceptualized first. The three agent types, "Tree", "Elephant" and "Firewood collector" were defined. A so-called "Goal-Oriented Action Planning" (GOAP) approach for collecting firewood was adapted from game development to model human adaptive behavior. Agent-based modeling made it possible to observe individual trees over time and to compare different tree species and age groups. Additionally, the two areas considered have been simulated with two different climate scenarios. Particularly noteworthy here is the comparison between the pure consideration of aboveground biomass and the detailed information on age class and species level. Even before changes in the biomass values became apparent, effects such as species shift, loss of adult trees and scrub encroachment were visible in the simulation results. This is a clear difference to macroscopic modeling. Even if trees react very slowly to changes, trends can be derived from the model results that resemble an early warning system. Thus, simulation results can be processed directly into the practice of decision-makers on-site in Africa. Besides, the combination of computer science and geography has contributed to a reusable methodical workflow and adaptable model elements for the scientific consideration of current and potential distributions of certain tree species in different regions.
Use and reproduction:
No CC License (german copyright law applies)