Mapping the spatial distribution of tree species selected by elephants (Loxodonta africana) in Venetia-Limpopo Nature Reserve using Sentinel-2 imagery

Trees that elephants select are generally at risk to local extirpation. Elephants are the primary drivers of change to large savannah trees, pushing trees over, ring barking or damaging them. These impacts have an adverse effect on the diversity and structure of woody vegetation. Many tree species have slow growth and recruitment rates. This makes it difficult for ecosystems to recover after damage, especially in areas with high densities of elephants. In this study, we mapped the spatial distribution of six endemic and priority tree species that are selected by elephants using Sentinel-2 imagery. Data on the GPS locations of the selected trees were collected and a machine learning algorithm, Random Forest, was used to discriminate between species and to map their spatial distribution in the Venetia-Limpopo Nature Reserve. The algorithm successfully discriminated the different tree species with an overall accuracy of 79.6%, ranking the vegetation red edge (Band-7) and water vapour (Band-9) as the most important bands in the classification. Colophospermum mopane is a dominant tree species in terms of its distribution on the reserve, and according to the classification results, canopies of this species cover approximately 17% of the study area, while Commiphora spp. cover 19%. These findings indicate the effectiveness of Sentinel-2 imagery for mapping savannah tree species that are valuable for conservation and selected by elephants.