Impact of anthropogenic disturbance and climate on bamboo distribution in shifting cultivation landscapes of Northeast India

Bamboo, a multipurpose plant species found in tropical and subtropical regions, covers about 1% of the earth's landmass and provides numerous ecosystem services. This study focuses on mapping the distribution of bamboo in the Dima Hasao district of Assam, India, and examines the influence of anthropogenic disturbance and climate on bamboo occurrence in shifting cultivation landscapes. Bamboo distribution was mapped using spectral and textural variables from Sentinel-2 imagery (March and November 2022) and topographic data from the Shuttle Radar Topography Mission digital elevation model. Three machine learning classifiers, random forest (RF), support vector machine, and artificial neural network, were evaluated for bamboo classification. Among these, the RF classifier achieved the highest performance, with an overall accuracy of 87.54%, a producer's accuracy of 86.86%, and a user's accuracy of 83.35% when using the combination of March and November median imagery. The short-wave infra red (SWIR) bands were found to be important variables for land use land cover classification, while the normalized difference vegetation index based on the vegetation red edge 2 band (NDVIre2) emerged as the most significant variable for bamboo mapping. A disturbance map for bamboo growing areas was also generated using LandTrendr in Google Earth Engine based on the normalized burn ratio (NBR) from time-series Landsat data and validated using TimeSync. The results indicated that approximately 78.9% of bamboo-growing areas in the district had undergone high disturbance, largely attributed to frequent practice of shifting cultivation. The influence of climatic drivers on bamboo distribution was analyzed using the RF algorithm, and vapour pressure deficit was identified as the most influential factor. This first-of-its-kind study in Northeast India offers key insights into bamboo ecology and demonstrates the value of advanced classifiers in improving distribution accuracy. The study has important implications for forest policy and landscape management in shifting cultivation regions, providing a foundation for conservation planning, climate adaptation, and contributions to climate resilience and relevant UN Sustainable Development Goals.