PS 73-126
Geographical modeling of spatial interaction between human activity and forest connectivity in an urban landscape of southeast China

Thursday, August 13, 2015
Exhibit Hall, Baltimore Convention Center
Yin Ren, Institute of Urban Environment, Chinese Academy of Sciences, Xiamen, China

Geographical detector models provide a quantitative approach for evaluating spatial correlations among ecological factors, population density and landscape connectivity. Here, we used a geographical model to assess the influence of different gradients of urbanization, human activities and various environmental factors on the connectivity of urban forest landscapes in Xiamen, China from 1996 to 2006. Our overarching hypothesis is that human activity has modified certain ecological factors in a way that has affected the connectivity of urban forest landscapes. Therefore, spatiotemporal distributions of landscape connectivity should be similar to those of ecological factors and can be represented quantitatively. Integral indices of connectivity and population density were employed to represent urban forest landscape connectivity and human activity, respectively. We then simulated the spatial relationship between forest patches and population density with Conefor 2.6 software. A geographical detector model was used to identify the dominant factors that affect urban forest landscape connectivity. 


The results showed that a distance of 600 m was the threshold of node importance. Mean annual temperature, mean annual precipitation, elevation, patch area, population density and dominant species had significant effects on the node importance. Mean annual temperature was more significant than population density in controlling the spatial pattern of the delta of the integral index of connectivity (dIIC). The spatial interaction between population density and various ecological factors as well as their linearly enhanced or nonlinearity enhanced urban forest landscape connectivity. In conclusion, a combination of graph theory and geographical detector models is effective for quantitatively evaluating interactive relationships among ecological factors, population density and landscape connectivity.