Effects Of Vegetation Restoration On Landscape Pattern In Hongya Country

The great forestry ecological engineering projects, i.e. Conversion of Farmland to Forest (CFF) and Natural Forest Conservation (NFC), have played great positive roles in promoting the ecological environment constructions of western China. The reconstruction, recovery, and protection of vegetation in these projects, changed not only the composition but also the configuration of landscape pattern. In this study, the landscape pattern of Hongya Country in 1994 and 2009 was compared to examine the effects of vegetation restoration on landscape pattern at the mountain-hilly region of the edge of Sichuan Basin, China. Firstly, based on the calculation of six landscape indices, the landscape pattern derived from Landsat TM images were quantified to obtain the major changes of landscape pattern. The changes of index values indicate different effects of vegetation restoration on landscape pattern at the gradient of elevation for both CFF and NFC. But the direct compare of raw index values cannot distinguish the compositional and configurational changes of landscape pattern, indices are more uninterpretable. Secondly, a grid sampling based landscape pattern analysis approach was conducted to separate the different processes of landscape change. The statistical changes of sampled index values exhibited various trends for different increasing and decreasing processes of compositional change for both the types of forest and farmland. As a consequence, the relationship curves between composition and landscape index values were fitted to illustrate the various trends of different change processes at different elevation gradient zones. As a result of the graphical illustration, the changes of index values are almost interpretable, and the effects of vegetation restoration on compositional and configurational changes could be distinguished by the change trajectories of landscape indices for the multiple change processes in Hongya. At last, the advantage, procedure, theoretical dependence, and underlying problems of the grid sampling based approach for landscape pattern analysis were summarized and discussed, and introduced a "patch mosaic paradigm" and "landscape gradient" integrated conceptual model, named "mosaic-gradient". Overall, the results could be arranged as follows:(1) From 1994 to 2009, the proportion of forest in Hongya increased by 5.72% from 60.58% to 66.30%, and the farmland and grassland decreased by 5.12% and 1.51% respectively. The increase of forest and decrease of farmland and grassland are the major effects of vegetation restoration on the compositional change of landscape pattern.(2) The landscape pattern change of Hongya exhibited a little defragmentation trend from 1994 to 2009. At the elevation gradient, the spatial distribution of forest, which covers most of the land, was more expanded continuously to lower and gentle areas throughout the whole landscape; while, as a contrast, the farmland was more shrank to lower and gentle areas and it’s pattern was more fragmented. In low area (below 1000m), the changes of landscape were mainly induced by transformations of landscape classes, especially for the heavy amounts of farmland-forest conversions. And in high area (above 1500m), there were little transformations between classes, but more significant changes of landscape configuration. The change processes between 1000-1500m were relatively complex, partly for the overlapping influences of CFF and NFP that coupled with other human induced disturbances. At the slope gradient, forest≤ 5° increased clearly for the intense impact of human activities; moreover, forest at the slope near 25° also exhibits a little clear increasing trend, which indicating more forest-farmland conversions in the implement of CFF due to the larger amount of slope-farmland in this region.(3) The effects of CFF on landscape pattern were mostly represented as the compositional change at the low area below the elevation of 1000m, especially at the lower area below 500m, where the scattering filled forest patches converted from farmland induced more patch edges and made the pattern more fragmented, representing a more significant change of composition. The configurational effects of CFF on landscape pattern were mostly exhibited at the elevation zone between 500-1000m, where the increased amount of forest transformed from farmland indicating compositional change and the aggregation of forest patch that indicating configurational change both have the equal likely magnitude of influence on the pattern of this region. The middle elevation zone between 1000-1500m exhibited no significant change for both composition and configuration, this was likely due to the compositive complex influences of CFF, NFC, and other human induced disturbances. The effects of NFC that impact the configuration of landscape pattern were mainly exhibited at the region higher than 1500m, and particularly at 1500-2500m. As the matrix of this region, forest was more gap-filled and connected, and the dominance, contagion that established indicating a well repaired and enhanced integrity of forest landscape.(4) The various change processes underlying are also the causes of why the changes are difficult to interpret when using indices to quantifying landscape pattern. The grid based sampling approach could relate the change processes to multiple landscape indices and make the results more interpretable. The procedure of this approach could mainly content four steps. First, utilize spatial grids to segment the categorical map into samples. Second, discriminate different change processes and their driver forces for all the samples statistically. In this thesis, the discrimination is based on the spatial continuous gradient concept of geographical variable, elevation. Third, use fitted carves to illustrate the common relationships between landscape indices and class proportions. Forth, draw the change trajectories on each carve for all the discriminated processes graphically, and analyze the trend, intensity, and stage of each trajectory to distinguish the compositional and configurational changes for each process. This approach integrate the "patch mosaic paradigm" and "landscape gradient model", and could be well applied to landscape pattern analysis with more clear interpretation of raw index value change, so call it "mosaic-gradient" model.