|dc.description.abstracteng||Several studies suggested that forest fragmentation, which is an effect of deforestation, and edge effect have an impact on the biomas and carbon stock in tropical forest. For Amazon and Atlantic Forest biomes, most studies have shown using in situ measurements and remote sensing data that biomass and carbon stock reduce within the first 300 meter of a forest edge to its center. For the Cerrado biome, there is currently no consensus whether or not there is an edge effect on biomass and carbon stock. Therefore, this study aims to better analyze the forest fragmentation and edge effect on the vegetation, such as biomass-carbon stock and canopy greenness, in the Cerrado biome. The most common method used to assess the edge effects on vegetation in tropical forest is direct measurement, which is difficult to replicate, cost intensive and time consuming. Therefore, the use of satellite images may be an alternative to monitor vegetation cover within the context of edge effects. In order to monitor forest fragmentation and carbon storage in the Cerrado biome, different approaches were investigated with fragmented areas in the city of Nova Mutum- Mato Grosso: (1) mapping the different type of vegetation using optical and synthetic aperture radar (SAR) remote sensing images, (2) estimating the biomass and carbon stock from in situ measurements within the context of edge effect and (3) monitoring the edge effect over the long-term using time-series Landsat satellite images.
First, the use of optical and SAR images to map the different types of vegetation in the transitional area between the Cerrado and Amazon biomes was investigated. Using this approach, the diverse vegetation types of the transition areas were studied. The findings indicated that by applying a supervised random forest classification, the highest overall accuracy and kappa coefficient were obtained by using only Sentinel 2A images for the classification process. However, out of the three classifications, two (Sentinel 2A with TanDEM-X and Sentinel 2A with Sentinel 1A) that used radar and optical images recorded the highest overall accuracy and kappa values. Bands 5, 11, and 12 from Sentinel 2A satellite image, texture images from Sentinel 1A cross-polarization, and coherence from TanDEM-X images were the most important variables that separated each vegetation class similar to the variable importance from the random forest algorithm. After obtaining a better understanding of the diverse vegetation types in the study area, we assessed the impact of fragmentation and edge effect on biomass and carbon stock using in situ measurements that were collected in July and August 2017. Using this approach, we investigated the woody components of tree layer and shrub layer by recording key variables such as the diameter at breast height (DBH), total tree-shrub height, wood density, basal area and tree species. Here, the DBH and the total tree-shrub height were the explanatory variables of the allometric model in the Cerradão. For the Cerrado denso on the other hand DBH and the wood density were the explanatory variables of the allometric model. In contrast to our working hypothesis, the results showed no significant differences in the quantity and the distribution of AGB and carbon stocks between edge and center of the fragments of both vegetation types. Rather, the results showed a significant difference for the AGB and aboveground carbon stocks between the two investigated vegetation types. We thus suggest that the edge effect on biomass patterns found in the Amazon cannot be compared with those of the Cerrado biome. It is important to stress that our analyzes were performed with a single measurement, therefore, to have a better understanding of these impacts, a long-term approach is required. The last analysis of this thesis was to evaluate the edge effect in the long-term based on NDVI values of the transitional area between the Cerrado and Amazon biomes. The method described in this study corroborates studies that assessed edge effect on vegetation within the Cerrado and Amazon biomes. In this study, we applied a different approach to investigate possible edge effects using vegetation index from freely available satellite images. Our results showed a positive significant change (p-value < 0.00005) via the NDVI values in relation to distance from the nearest edge. The closer the vegetation was to the edge, the lower their respective NDVI value. Furthermore, our results showed that long-term edge effect patterns found in the Amazon biome cannot be extrapolated to Cerrado. This observation is mainly due to the stabilization of NDVI trends after two years of deforestation within the area. This suggest that more studies are needed to adequately understand the dynamics of edge effect in Arc of Deforestation, which directly affect biomass and carbon estimations.
In this thesis, different methods were used to assess edge effects on the vegetation, such as biomass-carbon stock and greenness, in the Amazon-Cerrado ecotone. In the small scale, using fieldwork data, we could not find any evidence that fragmentation affects the carbon stock, due to the fact that the natural resources of the Cerrado biome have been widely exploited within the past 50 years, and thus, has a general decreased in biomass and carbon stock in the edge and also the center of the fragment. However, we have significant results from remote sensing long-term data, in which the NDVI is affected by the forest fragmentation and edge effect on the vegetation (canopy greenness), the closer to the edge, the lower the NDVI value. This shows that the use of satellite images has allowed an analysis of a larger period compared to fieldwork. One explanation for these findings is that the natural resources of the Cerrado biome have been widely exploited within the past 50 years, and thus, has decreased an overall biomass and carbon stock in all areas, which was found in the fieldwork data due to the few samples that were measured. However, the use of satellite images has allowed an analysis of the fragmentation effect with a larger amount of samples compared to fieldwork. not only in the edges. These outcomes of this thesis provide a solid research direction for further studies on edge effect in the Amazon-Cerrado ecotone. Long-term analysis using both field data and remote sensing is required to fully understand the fragmentation and edge effects in the Cerrado.||de