| The Giant Panda,(Ailuropoda melanoleuca)is the flagship species of wildlife conservation in China and the world.The Qinling Mountains are the most eastern and most northern branches of giant pandas in China,and are of great significance to the population and habitat protection of giant pandas in China.This paper based on land use/cover of Qinling Mountains and wild distribution of Qinling giant pandas in 2000 and 2010 data.Supported by software platforms such as FRAGSTATS4.2.1,ArcGlS10.2,GWR4.0 and R3.3.2 quantified landscape spatial pattern characteristics and changes of the Qinling Mountains in 2000 and 2010,and analyzed spatial utilization of the Qinling giant panda in 2000 and 2010,and introducing geographical weighted regression(GWR)model to couples the landscape spatial pattern of the Qinling mountains with the giant panda space utilization dynamics and compare it with the ordinary least squares(OLS)model.The main findings are as follows:(?)From the perspective of landscape composition,the composition of various landscape elements in the Qinling Mountains from 2000 to 2010 has not changed much.In 2000,the forest area accounted for 37.03%of the total area,and the grassland accounted for 34.79%of the total area.In 2010,forest land accounted for 37.22%of the total area,and grassland accounted for 34.15%of the total area.Forest and grassland are the most dominant landscape elements,which constitute the matrix of the Qinling mountains landscape pattern.The spatial distribution characteristics of each landscape elements are obviously different,the forests(including forest,shrub and woodland)are mainly distributed in the relatively high altitude area in the middle of the study area.The grassland is surrounded by forest,the cropland and resident are scattered in the low-altitude gully area in the northwest,south,southeast,and the low-altitude plain in the northeast.There is no obvious distribution pattern in the waters and other land.(?)From the perspective of landscape shape characteristics,the shape changes of various landscape elements in the Qinling Mountains from 2000 to 2010 are not large.The order of the patch shape index of the landscape elements in 2000 and 2010 is from large to small is cropland>grassland>forest>shrub>woodland>urban>water>other land.The shape index of cropland and grassland is large,indicating that the two landscape types are complex and easy to promote interaction between the patch and the external environment.The fractal dimension of each landscape element is not much different.The fractal dimension of the forest(including forest,shrub and woodland)is close to 1.5,indicating that the self-similarity of these patch is low.Habitats are complex and human disturbance is small.(?)From the perspective of landscape fragmentation,there is a significant difference in the average patch area and fragmentation index of the landscape elements in the Qinling Mountains.There is little difference between the patch density,the dumpiness index and the cohesion index.From 2000 to 2010,the degree of fragmentation of various landscape elements does not change much.In the study area of the forests(including forest,shrub and woodland),the patch dumpiness index was greater than 93%,and the cohesion index approached 99%,indicating that the patch type was spatially clustered and connected.In 2000,the landscape contagion index was 59.7563,and in 2010 it was 59.4054,which experienced a downward trend.In 2000,the Shannon's diversity index was 1.4147,and in 2010 it was 1.4273,which experienced an upward trend,indicating that the degree of fragmentation of the Qinling mountain landscape has been strengthened from 2000 to 2010.The main reason is the fragmentation of the overall landscape pattern in the study area due to the presence of artificial patch components such as cropland,resident and other land use.(iv)The ordary least square(OLS)and geographically weighted regression(GWR)models were used to couple the spatial distribution of giant pandas and the landscape spatial pattern in 2000 and 2010.It is found that based on the 2000 giant panda distribution data,the R~2 of the OLS model is 0.226,the R~2 of the GWR model is 0.365,and the difference between the GWR model and the AICc of the OLS model is much larger than 3(AAICc=23.289),and the residual square decreased by 36.953 of the GWR model compared to the OLS model.Based on the 2010 giant panda distribution data,the R~2 of the OLS model is 0.379,the R~2 of the GWR model is 0.506,and the difference between the GWR model and the AICc of the OLS model is much larger than 3(?AICc=23.441),and residual square decreased by 24.85 of the GWR model compared to the OLS model.The results show that the AIC,R~2 and adjust R~2 of the GWR model are significantly better than the OLS model,and the fitting effect of the model is significantly higher than that of the OLS model.The local regression coefficient of the GWR model can more effectively describe the spatial heterogeneity between the distribution of wild animals and the environment,and reveal the complex spatial relationship between the spatial distribution of giant pandas and environmental variables.In summary,this study systematically analyzes the spatial pattern of the Qinling mountains in 2000 and 2010 and its impact on the spatial distribution of giant pandas.The results can provide a scientific and effective theoretical basis for protected area managers to further protect giant pandas and other wild animals.In this study,the GWR method was used to explore the response mechanism of giant pandas to landscape spatial pattern changes in the Qinling mountains.This method can provide reference for further accurate quantitative analysis of the selection and utilization of giant pandas and other wild animals. |
PDF Full Text Request