Assessing high altitude andean wetlands using plant community structure: A multivariate analysis and remote sensing approach

Abstract

High altitude wetlands of the Central Andes Cordillera in South America are unique ecosystems with valuable ecosystem functions and one of the environments most threatened by climate change. They play a significant role in sustaining endemic biota, in providing the grasslands for herd of alpacas, llamas and vicuñas and by storing water and releasing it during the year to one of the driest regions on the earth, the Atacama Desert. This ecosystem is dependent on groundwater sources, and vegetation regulates the amount of water available during the dry periods. In Chile, the increasing demand for water requires more technical knowledge and research in order to prevent further degradation. The objective of this research is the description of Tarapacá and Atacama regions’ wetlands plant communities, the abiotic factors and human impacts that are more strongly associated with them by multivariate analysis and a remote sensing approach. Chapter 1 is a review of high altitude Andean wetlands and their importance. In Chapter 2, I identified differences in plant communities’ structure. Each region was distinguished by 5 different plant communities according to the vegetation wetland types. Abiotic factors and physical attributes that were more strongly associated with plant communities were the number and width of principal streams found on the wetland and amount of rocks, bare land and percent of organic matter along the vegetation transects. Using field work and remote sensing, in Chapter 3, I performed a spectral discrimination among plant communities using IKONOS-2 and Geoeye1 high resolution satellites images. They were used to identify which bands and vegetation indices were the most effective for discriminating vegetation classes. Vegetation classes did express different spectral behaviors. The classes with more reflectance variation were mixed grasses with Oxychloe andina, mixed grasses with salt patches and mixed grasses with Zameioscirpus atacamensis, while classes dominated by O. andina, Z. atacamensis and Festuca chrysophylla expressed less variation on the spectral range. General Discriminant Analysis showed that the most important spectral bands and vegetation indices for distinguishing differences between vegetation classes were Band 1-blue, band 4-NIR and the Wide Dynamic Range Vegetation Index.