Spatiotemporal distribution of negative human-elephant interactions in Wayanad district, Kerala, India

Negative human-elephant interactions (HEI) present a significant threat to the long-term conservation of Asian elephant populations and negatively impact the psychological well-being of local communities. The Brahmagiri-Nilgiri Eastern Ghats complex in Kerala is a key landscape for Asian elephant conservation and supports the largest single breeding population across their range. However, negative encounters between people and elephants are increasing in frequency and have become a prominent cause for concern. Despite this, the use of predictive distribution models to map the spatio-temporal patterns of human-elephant interactions across the landscape remains poorly explored. Compiling 1942 individual conflict incidents from compensation records, we dynamically extracted 16 ecological and anthropogenic variables identified in the literature as important drivers of interactions between people and Asian elephants. Using an ensemble modelling framework incorporating 10 algorithms, we constructed predictive distribution models for the wet and dry seasons from 2011 to 2023 to map the spatiotemporal distribution of regions at consistent risk of negative human-elephant interactions. Final consensus models achieved a mean accuracy of 0.91 (AUC) and 0.73 (TSS) respectively and suggested that the top predictor in influencing interactions is human population density. Regions within and adjacent to the Wayanad Wildlife Sanctuaries (I and II) have the highest predicted likelihood of interactions, however risk-level differs according to season. During the dry season, areas including Kidanganad, Nulpuzha, and Pulpalli, within the Kurichiat, Muthanga, and Sulthan Battery Forest ranges, were identified as having the largest land area at risk. Conversely during the wet season, the Tirunelli and Trisshaleri areas in the Tholpetty forest range demonstrated the highest risk. Results provide valuable insights to inform effective mitigation strategies at the landscape-level.