Evaluating the dynamics and eco-climatic predictors of forest conversion and restoration in Old Oyo National Park, Nigeria using geospatial and machine learning techniques

Abstract

Forest fragmentation and restoration in protected areas over a period provide insights into the success of any ecological intervention. This study aimed to determine the landscape pattern, dynamics, intensity, and eco-climatic drivers of forest fragmentation and restoration within the last 4 decades in Old Oyo National Park (OONP), Nigeria. The data collection and analysis used field observations, and geospatial and machine learning techniques. Three-time series (1986, 2003, and 2019) of Landsat satellite imageries downloaded and subjected to supervised image classification and landscape analysis (LA). The results revealed that the mixed open savanna areas covered the highest landmass of OONP. The dense woodland and forest outlier areas (DWFOA) reduced from 579.34 km2 (23.07%) in 1986 to 489.84 km2 (19.50%) in 2003 and increased to 653.46 km2 (26.02%) in 2019. LA revealed an increase in the forest fragmentation based on the number of patches and mean patch area. Twenty-six and twenty-eight out of forty-three eco-climatic variables are the most significant drivers to the changed and restored DWFOA between the year 1986 and 2019. Partial dependence plots showed a strong relationship of bush burn severity on the probability of the conversion and restoration of DWFOA. Other significant variables include distance to host communities, distance to roads, distance to rivers, elevation, average monthly rainfall (March–April, June–October), and slope. Despite the forest restoration, the evidence of forest fragmentation calls for measures to mitigate the climate change, indiscriminate bush burning and other anthropogenic pressure indicators.