Greening drylands: The potential and limits of afforestation as a nature-based solution in the Aralkum Desert

Dryland afforestation is increasingly deployed as a Nature-based Solution (NbS) for climate mitigation and ecosystem restoration, yet its long-term effectiveness remains uncertain because carbon gains may be offset by biophysical trade-offs and water limitations. Existing assessments rarely integrate carbon sequestration, surface albedo, and climate vulnerability, limiting understanding of how these processes interact to determine restoration outcomes. Here we present the first integrated assessment of large-scale afforestation in the South Aral Sea – a region whose near-complete desiccation may have shifted the area from a carbon sink toward a source of emissions. We evaluate the current carbon pool, surface albedo effects, and climate vulnerability to assess both the potential and inherent limits of dryland afforestation as an NbS. By combining field surveys of saxaul (Haloxylon) shrublands, analysis of land cover change, and six decades of regional climate data, we estimate that the vegetated parts of the South Aral Sea (approximately 1.6 million hectares) store 20.4 million tonnes of carbon (74.7 million tonnes CO₂-equivalent) in soil and vegetation, with a mean carbon density of 12.7 Mg C ha⁻¹ . Albedo analysis reveals strongly context-dependent climate trade-offs: vegetation establishment increases surface reflectivity on former water bodies but decreases it on highly reflective salt pans. Over the past six decades, regional temperatures have increased by approximately 0.5°C per decade (more than twice the global average) while drought intensity has increased markedly since the 1990s. Temperature-adjusted drought indices (SPEI-12) indicate substantially stronger drying trends (R² = 0.55) than precipitation-only indices (SPI-12; R² = 0.03), demonstrating that rising temperatures amplify water stress beyond precipitation changes alone and may constrain the vegetation density that can be sustained over time.