Assessing water-energy-food-ecosystem nexus policy trajectories under uncertainty in the Inkomati-Usuthu water management area, South Africa

• Water-energy-food-ecosystems management is extremely complex • A WEFE nexus system dynamics model is developed • The model is applied to the Inkomati-Usuthu basin, South Africa • Results show complex policy interactions leading to synergy and trade-offs • Results can help inform integrated policy formulation The water-energy-food-ecosystems (WEFE) nexus promotes holistic management of natural resources. WEFE sectors are linked through socio-economic connections, for example the food sector depends on water availability, and through policies largely developed in silos. South Africa has c. 80% of households reporting inadequate access to food and water resources, but has mineral wealth supported by high-value agriculture and tourism. The primary energy source is coal, with aging infrastructure leading to intermittent energy supply. This paper presents the development of a system dynamics WEFE nexus model in the Inkomati-Usuthu Water Management Area, capturing interactions between sectors to 2050 under climate and socio-economic pathways. The model integrates policies to assess their impact across sectors. Implemented one-at-a-time, policy impacts tend to be confined to the sector to which they are developed. The land sector is a key nexus impact driver, and land-based policies have wide impacts across sectors. Food production is shown to drop up to 52% compared to the reference, with nitrogen leaching dropping by up to 37%. With all policies implemented simultaneously, impacts across sectors are greater and most sectors benefit for example crop production is enhanced by up to three times, nitrogen leaching drops by up to 48%, and greenhouse gas emissions are reduced by up to 19%, reflecting policy design. By integrating multiple uncertainties, in terms of modelling biases and strong radiative forcing, variables such as crop yield and biomass growth under some RCP8.5 simulations can expand with predicted values well above feasible levels. Results trends were validated by local stakeholders and against observation. Results suggest that interactions between policies are very complex, an important message for policy makers dealing with natural resources management. Results show unintended consequences (trade-offs) of siloed policy development and implementation, with some policies countering the effects of others. For example, land and ecosystems preservation policies tend to reduce local food production, an issue for food security concerns. There are hundreds of millions of combinations, opening opportunities for machine learning to search vast spaces and suggest ‘optimal’ policy strategies. This work contributes to: i) improved understanding of multi-sectoral response to policy implementation; ii) providing updates to policy makers regarding resource response to policy actions; and iii) promoting holistic, integrated thinking about natural resources management and policy development in the region.