Spatial Modelling of the Temporal Patterns and Intensity of Wire Snare Poaching and Predicting Land Cover Change Dynamics in a Semi-Arid Protected Area

The spatial and temporal dynamics of poaching, along with continuous land cover alterations like deforestation and agricultural expansion, hinder efficient wildlife management. Changes in land cover could either generate new poaching opportunities or impede access to previously exploited areas. With the doubling of Africa's human population, protein resources will be strained, boosting the purchase and harvest of bushmeat for sustenance and income. In regions where meat poaching transpires, wire snaring is a prevalent technique due to its affordability, efficacy, and ease of acquisition, installation, and concealment. Due to their non-selective nature, snares can inflict severe by-catch mortality on a range of species. Yet, the necessity of projecting future values of a time series traverses across a range of fields. Powerful methods have been developed to capture these components by defining and estimating statistical models. Policymakers must plan several months or years ahead, since drawing up policies and actual policy implementation may take several months or years. The aims of this study were to (i) estimate the spatiotemporal patterns and intensity of wire snare poaching and (ii) predict future land cover dynamics using land change models and assess how these changes may influence poaching risk in the coming years. The Autoregressive Integrated Moving Average (ARIMA) was utilised for time series analysis and forecasting. Kernel density estimator (KDE) was used to smooth point data (in this case the locations of wire snares) to create a continuous surface that shows areas of high and low density. The analysis of land use and land cover takes into account the utilisation of Landsat satellite image products. Satellite images for the years 2020, 2022, and 2024 were utilised as inputs for forecasting future land cover scenarios using cellular automata artificial neural network (CA-ANN). The results from the ARIMA show an increase in the wire snares which would enhance the possibility for human–wildlife conflicts by the year 2028. Kernel density estimators pinpoint regions where wire snares are most concentrated; conservation teams can focus their patrols, thus helping to conserve species more efficiently. CA-ANN reveals marginal changes in land use and land cover which might enhance the likelihood for human–wildlife conflicts. Time series forecasting helps estimate when and where poaching activity is likely to spike. By identifying monthly trends, conservation teams can take preventative efforts rather than reacting after poaching has occurred.