Multi-model projections of trade-offs between irrigated and rainfed maize yields under changing climate and future emission scenarios

Abstract

Eighteen global circulation models (GCMs) were evaluated to determine the potential impacts of future climate change on irrigated and rainfed maize yields using the FAO AquaCrop model on an inter-annual and decadal basis (2020 s until 2090 s). Prior to deemed fit for future simulations, AquaCrop model was subject to comprehensive calibration and validation using extensive field-measured long-term datasets. We observed declines in (decadal) rainfed maize yields, ranging from 2.2% (0.2 t/ha) to 17% (1.4 t/ha) and from 8.1% (0.6 t/ha) to 21.5% (1.7 t/ha) under Representative Concentration Pathways (RCPs) RCP 4.5 and RCP 8.5, respectively. The range of declines was lower for irrigated yields [3.7% (0.5 t/ha) to 6.7% (1.0 t/ha) and 4.3% (0.6 t/ha) to 15.6% (2.2 t/ha) under RCP 4.5 and RCP 8.5, respectively]. Near maximal yield declines were distributed uniformly across the century and almost all decades exhibited > 10% yield declines under at least one emission scenario. Both economic (grain yield) advantage associated with irrigation (difference in irrigated and rainfed yields) and yield stabilizing benefit of irrigation (difference in rainfed and irrigated yield variability) are projected to decrease significantly (p < 0.05) under RCP 8.5. Rainfed maize yield variability was 533% and 200% greater than irrigated yield variability under RCP 4.5 and RCP 8.5, respectively. For RCP 4.5, the long-term mean inter-GCM (2020–2099) standard deviation in rainfed yields (4.6 t/ha) was 460% greater than that in irrigated yields (0.8 t/ha), while for RCP 8.5, this difference was 271% (4.6 t/ha vs. 1.2 t/ha). Tmax and Tmin were able to explain more variability in irrigated than rainfed maize yields, the difference being 229% and 126%, respectively. Precipitation change explained 46% and 50% of the variability in rainfed yield change under RCP 4.5 and RCP 8.5, respectively, and was 100% and 733% greater than what was explained for irrigated yield variability. The research findings hold significance for water allocation considering how dynamics of grain yields vs. availability of irrigation may manifest in the future.