Data for: Landscape change in Southern Ecuador: An indicator-based and multi-temporal evaluation of land use and land cover in a mixed-use protected area
Description
The data contains the geotif images of the maps derived from Landsat classifications in addition to the landscape metrics associated with the maps. We introduced seven LULC classes with the SVM algorithm following Baquero et al. (2004) and Homeier et al. (2008) land cover classification systems for the southern Ecuadorian Andes: 1) recent burned areas (BUR); 2) mixed development including residential, roads, and other infrastructure (MXD); 3) bare or exposed soils (BARE); 4) active pastures and agricultural areas (AGRP); 5) succession vegetation after human-induced disturbances (SUV) which may include areas with old pastures, native grasses, and/or bracken fern; 6) evergreen upper montane forest and humid shrub (UMV); and 7) herbaceous/shrub and dwarf bamboo páramo vegetation (PAR). SIDI depicts the probability that any two pixels selected at random would be different patch types. To quantify evenness, we used Shannon’s evenness index (SHEI). SHEI is a modification of Shannon’s diversity index and provides information on area composition and richness relative to other LULC classes. It accounts for the number of different land cover types observed along a straight line and their relative abundances. We assessed LULC pattern using two elements of landscape composition: 1) percent of landscape occupancy (PLAND), and 2) largest class patch (LCP) (i.e., percentage of total landscape area comprised by the largest class patch); and one component of configuration: proportion of like adjacencies (PLADJ) (i.e., degree of aggregation of class patches). Both PLAND and LCP help characterize landscape change and class dominance trajectories based on the analysis of multi-temporal data. We measured LCP using the largest patch index (LPI) as an indicator of class dominance (Herold et al. 2002). PLADJ helps characterize habitat continuity or fragmentation (i.e., contagion/interspersion characteristics). Gillanders et al. (2008), for example, used interspersion metrics such as PLADJ to analyze forest/insect interactions in temperate ecosystems. A landscape containing greater aggregation of patch types (i.e., a smaller number of larger patches, thus a more continuous landscape) will contain a higher proportion of like adjacencies than a landscape containing disaggregated patch types (i.e., higher number of smaller patches, thus a more fragmented landscape) (McGarigal et al. 2012). We also evaluated LULC change trajectories by quantifying percent class change (PCC) (i.e., proportion of each class that transitioned from class i to class j and obtained directly from transition matrixes). PCC, specifically from forested land cover types to anthropogenic land uses, provides critical insights about the rate of habitat loss caused by human-induced disturbances, and to a certain extent, the state of habitat conservation in a given area (Cohen et al. 2016).