Abstract
Natural forest regrowth on former agricultural land represents a large potential for climate change mitigation and nature conservation by rebuilding biogenic carbon stocks. Expansion and abandonment of agricultural land is typically influenced by trends in diets and agricultural intensification, which are two key parameters in the Shared Socioeconomic Pathways (SSPs). Recently, published datasets that explicitly map future land dynamics under different SSPs and climate change mitigation targets have been made available. These maps stem from different scenario assumptions, land classifications and modelling frameworks. This study aims to determine the role that these three factors play in shaping the estimates of the evolution of cropland and pastureland in a range of future scenarios. The analysis quantifies similarities and differences, and explore how different land-use data for the same SSP risk to achieve diverging estimates of carbon fluxes for the predicted land-cover transitions. We find that the datasets largely agree with the representation of cropland at the initial conditions (2015), although the identification of pastureland is ambiguous and shows large discrepancies due to the lack of a unique land-use category. Differences occur with the future projections until 2050, including scenarios for the same SSP and climate target. The accounting of CO2 sequestration from revegetation of abandoned agricultural land and CO2 emissions from forest clearance due to agricultural expansion shows a net reduction in vegetation carbon stock for most SSPs, except SSP1. However, the estimated cumulative increase in carbon stock until 2050 is 3.3 GtC for one dataset, and more than twice as much for another. Different datasets can thus bring to large differences in estimates of climate change mitigation potentials associated with future land transitions. The use of a common classification system with improved detection of alternative land uses, especially for pastureland, and the definition of common spatial drivers for land use changes can harmonize projections and reduce the variability of alternative background land-cover dynamics in environmental studies.