Land cover change and landscape fragmentation--comparing the utility of continuous and discrete analyses for a western Honduras region

Current studies of land cover change and landscape fragmentation rely predominantly on land cover classifications derived from remotely sensed images. However, limitations of traditional land cover classifications are numerous and well known. This research compares classification-based techniques (discrete data) to the use of vegetation indices (continuous data) for land cover modeling and analyses of landscape fragmentation for a study area in western Honduras. The study area typifies many regions of tropical developing countries, where a complex interaction of social and environmental factors have given rise to a dynamic mosaic of patches of reforestation and deforestation. Understanding the complex human dimensions of land use and land cover change in these parts of the world continues to present a challenge for researchers. The land cover modeling analysis compares two models using different formulations for the dependent variable: (i) a continuous analysis using a tobit model regressing the normalized difference vegetation index (NDVI), with non-forest values truncated at 0, on the variables elevation, slope, distance from roads and distance from the nearest market; and (ii) a discrete analysis using a probit model with threshold NDVI coverages (representing forest and non-forest). To examine the patterns of landscape fragmentation, a discrete analysis of a forest/non-forest classification using the software FRAGSTATS is compared to a continuous NDVI-based analysis using the local indicator of spatial association (LISA) statistic. Estimated marginal effects and overall predictive power are compared across the tobit and probit formulations. Because the tobit formulation included variation in the dependent variable for forested areas, greater information was retained regarding the subtle relationships among the independent variables and land cover. The LISA statistic, using the NDVI coverages as input, were helpful in the interpretation of the data and its spatial arrangement in the landscape, indicating that the LISA was a good complement to the FRAGSTATS, classification-based analysis. Given the LISA statistic is parametric and hence subject to outliers, whereas landscape metrics tend to be non-parametric, we found that the use of both FRAGSTATS and LISA together was more beneficial than either analysis by itself. While there is increasing awareness of the need for using continuous data for land cover modeling and fragmentation, this area remains little explored. The research presented provides insights for additional future applications utilizing continuous data analyses.