Studying shifting cultivation landscapes is fundamentally impeded by the difficulty in mapping and distinguishing shifting cultivation, settled farms and forests. There are foundational challenges in defining shifting cultivation and its constituent land-covers, distinguishing them from other land-uses during satellite image analysis, conceptualizing a suitable mapping framework and identifying consequent methodological specifications. Questions have already been raised about the veracity of official statistics of existing land-uses in the Northeast Indian region. Forest estimates are in doubt too. The objective in the paper is therefore two-fold: 1) to present a rigorous methodological framework and mapping protocol combined with extensive fieldwork and use them to undertake a two-season satellite image analysis to map the forest-agriculture frontier of West Garo Hills district, Meghalaya, Northeast India, and 2)to examine if results of the mapping correspond with official estimates of shifting cultivation and forests.
Empirically, shifting cultivation is the most extensive land-use in West Garo Hills district, followed by tree plantations and old-growth forest confined to only a few locations (mapping accuracy of ~80%). The extent of commercial tree-plantation crops is positively correlated with shortened fallow periods (high intensity of land-use), i.e., plantation expansion on hills occurs at the cost of land allocated for shifting cultivation leading to intensification (and possibly lowered production). The findings of the paper are in sharp contrast to other estimates, including those from Forest Survey of India (FSI), The Wastelands Atlas of India, and state government statistics, all of which showcase the West Garo Hills landscape as primarily forested (79% forest cover according to FSI) with only small fractions under shifting cultivation. We demonstrate that this is a consequence of the poorly defined land categories and poorer understanding of what is to be constituted as shifting cultivation and forest.
The main implications of the results are three-fold: 1) Conceptually, ecologists/physical geographers/social scientists/policy-experts involved in land-use mapping need to be attentive to the socially constructed nature of land categories. The importance of picking socially relevant land categories is crucial during mapping and not those ones for which there is global implicit preferences or others that are easily separable using image analysis. 2) Methodologically, lack of field-level understanding of shifting cultivation, its sub-components (fallows, different aged cultivation fields) and farm management, and other land-uses such as tree plantations and forests hamper clarity in defining and partitioning the landscape, undertaking mapping and establishing accuracy of estimates that are relevant for both environmental decision making and ensuring people's livelihoods. 3) From a policy perspective for land management, the paper calls for an attentive revision of India’s official land-use mapping protocols and has wider significance for remote sensing-based mapping of other shifting cultivation landscapes.