Modelling bush encroachment dynamics using Intensity Analysis and the Cellular Automata model

Abstract

Bush encroachment is a globally recognized phenomenon linked to adverse effects, including the degradation of grasslands and loss in biodiversity, thereby challenging the conservation of keystone and flagship species, the recreational value of landscapes and local livelihoods. Therefore, a comprehensive analysis of bush encroachment is essential to gain insights into its past, present and future encroachment, as well as the severity of transitions. Using RapidEye and PlanetScope satellite imagery, this study adopted Intensity Analysis to examine past and current bush encroachment trends for the periods 2009–2014, 2014–2019 and 2019–2023, while the Cellular Automata (CA) model was used to project future encroachment trends for 2028 and 2033 within a protected area. The results indicated a continuous increase in bush encroachment within the study area. Analysis of land cover intensities shows an intensive change in the research area’s land cover in the first period (2009–2014) compared to subsequent periods. In the first two periods (i.e. 2009–2014 and 2014–2019), woody vegetation gains were more pronounced at the expense of grasslands. However, during the 2019–2023 period, woody vegetation gains were less intensive to grasslands. Moreover, throughout the study period, most grassland gains occurred in bare areas, whilst the primary cause of grassland losses was bush encroachment. The projection of future encroachment trends indicates a continued increase in woody vegetation over the next decade. The results also indicate that bush encroachment is projected to expand by 5.50 and 6.67% in 2028 and 2033, respectively. These findings highlight the urgent need to assess and enhance management schemes within the study area. Gaining critical insights into bush encroachment progression trends and transition intensities can help prioritise landscape management efforts and support decision-making for the restoration of grasslands.