Abstract:

The cartography of the vegetation increases the understanding of the extension and spatial distribution of different communities and has become an important tool for planning strategies for both conservation and use. Remnant grasslands in Buenos Aires province have been well-studied, though the vegetation of Sierra de la Ventana has yet to be mapped at a level of detail appropriate to its spatial heterogeneity. One reason for this, is that conventional photo interpretation or image classification techniques are less suitable for making good vegetation maps of areas with complex topography. In this study we used techniques of predictive vegetation mapping at a regional scale and applied it to the natural vegetation of Sierra de la Ventana, a hill system located in southwestern Buenos Aires with a large part devoted to agriculture and livestock grazing. The technique presented in this paper combines remote sensed data (Landsat TM7) and a digital elevation model at a regional scale, with a previous study that related the most conspicuous plant communities in the region with topography at a more detailed scale. We constructed an algorithm to assign each pixel to the corresponding Vegetation Unit (VU) that uses the geo-edaphic environment and topographic variables as input variables. The sixteen VU identified using the digital data (spatial resolution = 90 m) showed a high correlation with the VU identified in the field (r = 0.88, n = 100 ground control points). The technique developed in this study represents an advance on vegetation mapping and could be extrapolated to other hilly areas that already have a basic vegetation description. Our results provide new knowledge on the spatial heterogeneity of this area. The map of natural vegetation is a useful resource for improving grazing management. Some of the differences between field data from the foothill sites and the derived map may arise from the impacts of grazing. Our results could help to generate hypotheses on the causes of differences between the actual and the original vegetation. We found that all the VU were included inside the limits of the only protected area in the region, however the most extensive VU in the region are the less represented inside the protected area.