Characterizing human-modelled landscapes at a stationary state: a case study of Minorca, Spain

Abstract

The island of Minorca, Spain, has become a patchy mosaic landscape as a result of centuries of human-induced fragmentation. The dynamics of the landscape elements and the spatial pattern of Minorca were analysed to test whether this human-modelled landscape was at a stationary state, and, if it was the case, to see whether the system could be characterized by a particular spatial pattern. Landsat TM satellite images were processed to derive land cover classification and vegetation index maps corresponding to the years 1984 and 1992, the best compromise between null cloudiness and maximum time span. The classification was used to test the stationary state by estimating the transition matrix of land cover. The Normalized Difference Vegetation Index (NDVI) was used to analyse the spatial pattern of change at the patch level. The spatial analysis of NDVI was based on four indices: fractal dimension, number and size distribution of patches, and spatial Kappa index. Those pattern descriptors were compared with values obtained from stochastic landscape simulations. Temporal analysis showed that land cover proportions remained approximately constant over the 8-year period, although interchange amongst patches existed. This suggested that the landscape of Minorca was close to a stationary state. The study of NDVI changes revealed that the spatial structure was different from that of neutral models and presented scale invariance properties. Results from land cover transitions and from NDVI patterning suggest that the landscape of Minorca is a critical system, in the framework of the self-organized criticality hypothesis, i.e. the mosaic of patches would self-organize at the frontier of a dynamic equilibrium constituted by the balance between disturbances and successional processes.