Correlation between landscape fragmentation and sandy desertification: a case study in Horqin Sandy Land, China.

The exact roles of landscape fragmentation on sandy desertification are still not fully understood, especially with the impact of different land use types in spatial dimension. Taking patch size and shape into consideration, this paper selected the Ratio of Patch Size and the Fractal Dimension Index to establish a model that reveals the association between the area of bare sand land and the fragmentation of different land use types adjacent to bare sand land. Results indicated that (1) grass land and arable land contributed the most to landscape fragmentation processes in the regions adjacent to bare sand land during the period 1980 to 2010. Grass land occupied 54 % of the region adjacent to bare sand land in 1980. The Ratio of Patch Size of grass land decreased from 1980 to 2000 and increased after 2000. The Fractal Dimension Index of grass increased during the period 1980 to 1990 and decreased after 1990. Arable land expanded significantly during this period. The Ratio of Patch Size of arable land increased from 1980 to 1990 and decreased since 1990. The Fractal Dimension Index of arable land increased from 1990 to 2000 and decreased after 2000. (2) The Ratio of Patch Size and the Fractal Dimension Index were significantly related to the area of bare sand land. The role of landscape fragmentation was not linear to sandy desertification. There were both positive and negative effects of landscape fragmentation on sandy desertification. In 1980, the Ratio of Patch Size and the Fractal Dimension Index were negatively related to the area of bare sand land, showing that the landscape fragmentation and regularity of patches contributed to the expansion of sandy desertification. In 1990, 2000, and 2010, the Ratio of Patch Size and the Fractal Dimension Index were mostly positively related to the area of bare sand land, showing the landscape fragmentation and regularity of patches contributed to the reversion of sandy desertification in this phase. The absolute values of the coefficients were the highest for grass land in the regression models, so that grass land had the most important influence on sandy desertification.