Modeling Inter-annual and Seasonal Distribution of Crop Depredation by Wild Asian Elephants in Eastern Thailand during 2009 to 2017

摘要

In Thailand, crop depredation by wild elephants have been intensified and negatively impacted local communities' quality of life as well as wild elephant's long-term conservation success. Despite increasing concern and urgent needs for solution, limited studies explore landscape-scale spatiotemporal pattern of this conflict. The goal of this study was, hence, to fill this gap and identify potential conflict distribution across season and year during 2009 to 2018. Specifically, we applied Maximum Entropy (Maxent) method and separately constructed models for resource-related scenario and direct human pressure scenario for wet and dry season (total of 4 models). Candidate predictors which characterize vegetation productivity (NDVI), meteorological drought condition (KBDI), landscape composition (standard deviation of NDVI), land cover (MOD1S land cover and Landsat permanent water pixel), topographic condition (elevation and TRI). and human pressure (Human Density, Roads, and Built-up) were used. Then, we applied our proposed two-dimensional conflict matrix based on thresholding approach to categorized predictive results into four groups. These include 1) high conflict area, 2) attractive area with uncertain pressure, 3) occasional refuge, and 4) unattractive which can be linked to different management action. With high temporal availability of satellite-derived dataset, we project our model on each year from 2009 to 2017. Multivariate Environmental Similarity Surface (MESS) was calculated to identify dissimilarity between each year to the training models, while regression slope was used to identify trend. Differences in habitat preference during wet and dry season can be identified. Variable response implied higher tolerant to direct human pressure in dry season. We estimated over 500 km~2 of frequent conflict occurring yearly in part of Chantaburi and Chachoengsao provinces due to high habitat suitability in that areas. Across 9 years, our model did not predict VIsible changes in direct human pressure, but large variation in resource-related suitability. KBDI was the key limiting factors causing drastic inter-annual response in distribution of potential conflict especially during extreme climatic condition (El Nino), while land cover changes suggested a subtle influence causing gradual changes in potential conflict over time. Management of crop depredation by elephants should consider not only variation in seasonality, but also resilience to extreme climatic events occur internally. Our method and potential conflict matrix can be applied for wildlife conflict study in other regions.