Assessing rates and causes of global forest fragmentation based on Globcover V2.2

Accurate and spatially explicit forest loss and fragmentation information is currently of primary concern for land use decision makers and conservation communities to address many pressing issues including habitat planning, biodiversity conservation, carbon budgets, and sustainable forest management. In this study, based on the latest and unprecedented high resolution global land cover dataset of Globcover v2.2, mathematical morphology coupled with the percolation theory was adopted to classify each forest pixel into one of the six fragmentation components (interior, perforated, patch, edge, transitional and undetermined) by using the sliding window analysis technique to establish spatially explicit forest fragmentation maps. Our analysis revealed that at the analytical scale of 9 by 9 pixels, Australia-Pacific had the severest fragmentation and South America the least. This was evidenced by 24.43% of Australia-Pacific's forest staying in “interior” condition while 57.84% of the forest in South America belonging to “interior”. On the other hand, derivations of geospatial indices suggested that the forest in both of Africa and South America were heavily disturbed by agricultural activities, whereas natural disturbances played a crucial role in forest fragmentation of North America. Additionally, forest fragmentation maps provided valuable insights and implications regarding the development of land use strategies, habitat planning for rare and endangered species, and sustainable forest management. By utilizing the fragmentation maps, such as the location of the interior and perforated components, and identifying areas with highly human-induced fragmentation, human can make effective risk assessments and crucial targets for protection and remediation.