Indicating structural connectivity in Amazonian rainforests from 1986 to 2010 using morphological image processing analysis

Tropical forests in the Amazon are rapidly being cleared. Deforestation has produced marked and ecologically important fluctuations in the spatial patterns of forest areas. Morphological spatial pattern analysis (MSPA) has been introduced to quantitatively map forest areas into various spatial pattern classes, but the remaining landscape here non-forest' class is not evaluated well using this method as it is simply clumped into a single class. Spatial arrangements of non-forest areas, including fragmentations occurring within forest and along its exterior boundaries, can have different impacts on related landscape properties, especially structural connectivity, and must be extracted and evaluated within this same analysis. To emphasize the spatial distribution of non-forest areas, therefore, we further classified non-forest areas into two classes (hole' and outer background') and proposed them as additional MSPA tools. Long-term forest-cover maps (from 1986 to 2010) of a deforested landscape in the Amazon were used as an example to illustrate the application of MSPA, as well as to verify the necessity of the proposed pattern classes, and based on the results, a conceptual model was built to indicate structural connectivity in the study area. This article enhances MSPA application by adding two non-forest classes, deploying a new way to understand the deforestation process, and provides a MSPA database to support forest condition evaluation and long-term conservation programmes in the Amazon.