A novel GIS-based ensemble technique for rangeland downward trend mapping as an ecological indicator change

Abstract

Rangelands provide important ecosystem services worldwide. The present study was aimed to map rangeland degradation in a critical mountainous rangeland ecosystem of Iran. The study was carried out based on seven years intensive fieldwork and recording 1147 locations with downward trends in the quality of the rangelands. Twelve conditional factors and two important ensemble algorithms including Probability density-Index of entropy (PD-IOE) and Frequency ratio-Index of entropy (FR-IOE), were used to produce rangeland downward trend (RDT) susceptibility maps. The results of validation showed that PD-IOE hybrid model with area under curve (AUC = 0.901) and standard error (SE = 0.011) is more accurate than FR-IOE hybrid model (AUC = 0.881 and SE = 0.012). In addition, our results indicate that altitude, distance to river, and distance to road are the most important factors for rangeland degradation. In addition, the places with higher altitude and less distance to roads and rivers endured more degradation and these places have downward trends. Based on the achieved results, 2% and 10% of study area fall into the very high and high classes of downward trends, respectively. Overgrazing and early grazing are two main drivers for rangeland degradation in the study area, and the rangeland managers and decision makers should define and develop strategies to reduce pressure on rangelands and promote strategies to restore these important ecosystems.