Assessing Lantana camara spatial distribution in response to rainfall and temperature variations in Inkomati catchment in Mpumalanga

Abstract

Lantana camara is an invasive plant affecting global natural ecosystems, including South African ecosystems. Preceding studies have focused on accurately identifying and mapping its spread, without understanding its adaptability to different environments. The aim of this study was to map and distinguish L. camara from different Land Use Land Cover (LULC) types, using Google Earth Engine (GEE) platform, Sentinel 2 satellite data and Random Forest algorithm in Inkomati Catchment, Mpumalanga. The study also modelled the influence of rainfall and temperature on the L. camara, and predicted its future distribution, using the Maximum Entropy (Maxent) Species Distribution Modelling (SDM) approach. Overall, the accuracy of the identification of L. camara was 90.27%, with a user's and producer's accuracy of 80% and 89%, respectively. L. camara was detected mostly within the eastern, south-eastern and south-western areas of the Komati sub-catchment and in the north-eastern parts, of the Nwanedzi River sub-catchment in the Kruger National Park. In comparison to other LULC classes, the prevalence of L. camara was found to be sparse and heterogeneous across the catchment. L. camara spatial distribution was significantly influenced by the mean temperature of the coldest and driest quarters and the rainfall of the wettest quarter. The current and future climatic scenario models built using the selected bioclimatic variables, yielded high species predictive accuracies of 0.937 and 0.932, respectively. These findings are useful in assessing landscapes vulnerable to L. camara invasions, the potential of remote sensing, and species distribution modelling towards the management of species invasion.