Quantitative analysis of habitat fragmentation using computer-generated landscapes and satellite images
Abstract
Steady advances in our understanding of the effects of specific landscape spatial attributes have led to the formulation of models of population and community dynamics in fragmented habitats. These models have not been readily tested or applied in real landscapes, due to difficulty in relating theoretical spatial models to real disturbed environments. One reason for this difficulty is a lack of quantitative indices of spatial pattern that are easily generated, and can be used to characterize habitat fragments and broad-scale spatial configuration. A FORTRAN program (HISA - Habitat Island Spatial Analysis) was written to calculate indices of spatial pattern from digital landcover data. A proximity index (PX) was developed that reflects the proximity of each patch to a suite of neighbors (landscape context), providing an indication of the density of forest habitat in the vicinity of each patch. Simulated landscapes of forest and non-forest landcover were generated to investigate the relationship between the proportion of forest and indices of patch spatial pattern. The simulated landscapes were also compared with real agricultural landscapes. The results support the use of percolation models as neutral models in landscape ecology, and the performance of the indices studied with neutral models can be used to interpret indices calculated for real landscapes. HISA was used to analyze the spatial quality of landscapes in Indiana as wild turkey habitat. Mean values of spatial pattern parameters were compared statistically between various quality levels of wild turkey habitat. Landscapes of optimal and sub-optimal spatial quality levels were more different from each other than sub-optimal and poor landscapes. Spatial parameters were used to mathematically classify landscapes. A Landsat Thematic Mapper image was used to model potential land acquisition for conservation reserves on a highly fragmented landscape in northwestern Indiana. Simulations of various reserve configurations that bridged areas of low PX values showed that linkage corridors and a "string-of-pearls" configuration were the most efficient way to increase PX, but they were predominantly edge habitat. A system of a few, larger reserves did not increase PX as much as a single, large reserve, unless the total area of added reserves was small.
Degree
Ph.D.
Advisors
Parker, Purdue University.
Subject Area
Ecology|Forestry|Remote sensing
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.