Characters of landscape patterns and correlation in Jinghe watershed

Landscape ecology describes spatial patterns of land uses and ecosystems.In order to manage landscapes sustainably,methods are needed to express spatial patterns in a suitable way.In the present paper six indices that describe spatial patterns are applied,namely patch number,patch area,patch density,dominance index,contagion index and the fractal dimension.These indices have been used for the analysis of the landscape patterns of all 31 counties located in the Jinghe watershed,Northwest China.Applying these indices,the different landscapes of the watershed have been classified into three types of spatial landscape patterns.The counties showed distinct variations in their indices.Agriculture,forest and grassland are the three dominant land use types in the watershed.High values of dominance and contagion indices were found in landscapes with relatively simple land use patterns,both for agricultural and grass land;these areas are mostly located in the lower reaches of the watershed,such as in Xianyang,Jingyang and Qianxian counties.Low values of dominance and contagion indices are typically associated with more diversified land use and with an almost equal share between agriculture,forest and grass land;These areas are mostly found in the loess uplands and in ravines,as well as on the loess plateau such as in Jingyuan,Pingliang,Longxian,Ningxian,and Xunyi counties.A combination of high dominance and contagion indices with a low fractal dimension was observed in Xianyang city,which is located in the relatively large plain loess scalariform areas in the lower reaches of the watershed.The significant positive correlation found between forestland area,dominance and contagion indices,and between grassland and fractal dimension indicate that forest and grassland are normally scattered in the watershed,signifying that the encroachment of agricultural land use into grassland leads to a high degree of fragmentation of land use.This clearly demonstrates the human manipulation of the landscape.The temporal changes of landscape indices observed in the watershed over the last decades have not been significant.The paper concludes that continued research on developing and applying landscape indices are critical for large-scale environmental impact assessment and landscape management.In particular,correlations should be established between index values and the actual ecological processes occurring on the ground.Basic questions to be addressed in further research should be the depiction of changes of the indices over time and how well these changes correspond to the actual ecological changes taking place on the ground.