Identifying the extent of the spatial expression of landscape fragmentation based on scale effect analysis in Southwest China

Abstract

As the primary method used to describe landscape patterns, the landscape index method is widely applied throughout landscape ecology research, and the scale effect (grain size or extent) of the landscape index has continually been one of the important factors in landscape pattern research. Compared to research regarding the diversity of grain size, research regarding the scale effect resulting from the extent change has been given less attention; it should be given more. This paper applies patch density (PD) and splitting index (SPLIT), two main indicators of landscape fragmentation, to the 30-meter-grain land cover in Southwest China, and proposes an optimization algorithm to calculate the landscape index to explore the suitable extent and spatial expression of landscape fragmentation research. The results show that (1) the optimization algorithm of Fragstats, the landscape index calculation software, combined with R language and ArcGIS improves the software memory limitation, increases the calculation speed, expands the calculation range and simplifies the calculation process; (2) to avoid a calculation error resulting from the cutting effect and information loss caused by the homogenization effect, the most suitable extent of landscape pattern research in Southwest China is 32 km × 32 km；(3) clustering can characterize the spatial distribution of landscape indices, and spatial autocorrelation can characterize the differences within landscape indices, especially the aggregation relationship with adjacent landscape indices; combining these two can better show the regional fragmentation characteristics and internal correlations; and (4) based on the optimum extent of 32 km × 32 km, the landscape fragmentation in Southwest China could be studied. The results showed that the severely fragmented areas were concentrated in Guizhou Province, the border area between Yunnan Province and Guizhou Province, and the western part of Chongqing. SPLIT in Guizhou Province was mainly affected by population density; PD in Guangxi Zhuang Autonomous Region was mainly affected by annual precipitation; and PD in Sichuan Province was mainly affected by annual temperature and elevation. This study provides a more convenient and feasible method for comprehensively understanding regional landscape fragmentation and promoting the coupling analysis between landscape indices and other indices.