The field foundation of the study rested on experiments conducted from 2023 to 2025 at all three sites, arranged in a randomized complete block design with 24 treatment combinations. These spanned three sowing windows (early, mid, and late), four nitrogen rates (0, 46, 69, and 92 kg N per hectare), and two bread wheat cultivars, Shaki and Boru, both widely grown for their rust resistance and broad adaptability. Seeding rate, row spacing, and pest control were standardized so that the effects of sowing date, nitrogen, and variety could be isolated. The soils differed markedly: Bore’s profiles were moderately acidic with high clay content but sharply declining organic carbon at depth, Kulumsa’s held high cation exchange capacity above 43 cmol/kg with low organic carbon, and Sinana’s carried the highest surface organic carbon at 2.76 percent, supporting its productivity.

Calibration was performed at Kulumsa using all 24 treatments across the three seasons, with seven genetic coefficients estimated through the GLUE tool within DSSAT version 4.8.5. The model reproduced anthesis and maturity dates within roughly three days of observations, with normalized root mean square errors of 3.7 percent or less for phenology. Grain yield and biomass were simulated with moderate errors of roughly 5 to 11 percent, and harvest index deviations stayed at or below 0.06. Shaki, the longer-duration variety, consistently out-yielded Boru in both observations and simulations, reaching observed grain yields of up to 4,610 kg per hectare in 2023.

Independent validation at Bore and Sinana, whose data played no role in parameter estimation, tested whether the calibrated coefficients transferred across contrasting agroecologies. Performance was strong at both locations: pooled analysis at Bore yielded an R-squared of 0.942, an RMSE of 188 kg per hectare, a normalized RMSE of 4.8 percent, and a d-index of 0.983, while Sinana performed even better with an R-squared of 0.951, RMSE of 143 kg per hectare, and d-index of 0.987. Phenology was simulated within plus or minus three days, canopy attributes such as leaf area index and tiller number within 4 to 6 percent error, and grain yield within roughly 3.5 to 4.7 percent normalized error. These figures sit comfortably within the thresholds commonly cited for successful crop model evaluation and gave the team confidence to run future climate scenarios.

Climate projections came from four regional climate model combinations in the CORDEX-Africa ensemble at 0.44-degree resolution, bias-corrected by quantile mapping for rainfall, maximum and minimum temperature, and solar radiation, and including carbon dioxide fertilization effects. Two Shared Socioeconomic Pathways were examined, the moderate SSP2-4.5 and the high-emission SSP5-8.5, over a near-future window of 2037 to 2068 and a far-future window of 2069 to 2100, benchmarked against a 1991 to 2022 baseline. Growing-season minimum temperatures were projected to rise by 1.4 to 2.3 degrees Celsius in the near future and 1.8 to 2.7 degrees in the far future under SSP2-4.5, and by up to 4.2 degrees by late century under SSP5-8.5. Elevated night-time temperatures accelerate respiration and shorten grain filling, a key mechanism behind the simulated yield declines.

The projected losses were substantial and intensified over time. Under SSP2-4.5, all three sites showed yield reductions of roughly 11 percent in the near future, deepening to between 16 and 22 percent later in the century. Under SSP5-8.5, reductions reached 25 to 30 percent in the near future and 28 to 32 percent in the far future. Strikingly, the three locations responded almost identically, suggesting that regional climate drivers outweigh local site conditions in shaping highland wheat vulnerability, in contrast with earlier studies that reported more differentiated site patterns. The projected declines fall within the range of recent multi-region DSSAT studies, which anticipated losses of 18 to 20 percent by 2040 and 28 to 30 percent by century’s end under comparable pathways.

Management, however, changed the picture considerably. In the field data, omitting nitrogen cut yields by 12.4 to 30.3 percent relative to the standard 46 kg N per hectare baseline, while 69 kg N lifted yields by 12.8 to 35.4 percent and 92 kg N by 25.6 to 50 percent, especially with early planting. Early sowing combined with high nitrogen delivered the largest gains, up to 50 percent for Shaki, whereas late planting produced yield reductions exceeding 40 percent even at high nitrogen rates, demonstrating that fertilizer cannot compensate for a delayed start. In the future simulations, early planting consistently showed the smallest losses: under SSP5-8.5 in the far future at Sinana, early-planted Boru with 92 kg N lost 32 percent of baseline yield, compared with 44 percent for mid planting and 58 percent for late planting.

Nitrogen’s protective effect weakened as climate stress intensified and showed diminishing returns. Moving from zero to 92 kg N improved outcomes by roughly 16 percentage points under moderate near-future conditions at Bore and Kulumsa, but only about 11 points at the more heat-exposed Sinana, and under the harshest SSP5-8.5 far-future conditions Bore showed no net benefit at all. The incremental gain from raising nitrogen from 69 to 92 kg was small, often just a few percentage points, echoing wider evidence that blanket fertilizer recommendations are suboptimal under changing climates. Variety choice mattered throughout: Shaki, with its longer grain-filling duration and apparently greater nitrogen assimilation efficiency, held a consistent advantage, even achieving a 5.6 percent yield increase above baseline under SSP2-4.5 near-future conditions with early planting and 46 kg N, while Boru under identical management lost 12.1 percent.

The economic analysis, which held input and output prices constant at a wheat price of 65 ETB per kilogram, framed these agronomic findings in financial terms. Early-planted Shaki with 92 kg N produced the highest calculated net return, 258,475 ETB per hectare under SSP2-4.5 in the near future and 213,625 ETB under SSP5-8.5 in the far future, while late planting without nitrogen generated negative returns as severe as minus 20,500 ETB under the harshest scenario. Because the marginal gain from the step from 69 to 92 kg N was modest, around 32,000 to 34,000 ETB, the authors note that capital-constrained farmers might favor the lower rate. The team is careful to stress the limits of the work: ensemble spread, crop model assumptions, calibration confined to Kulumsa, a defined set of management options, and fixed prices all constrain interpretation, and the economic figures are comparative estimates rather than forecasts of future farm income.

The overall message is sobering but actionable. No single practice fully offsets warming: even the best combination of early planting, adequate nitrogen, and the Shaki variety limited far-future losses under SSP5-8.5 to roughly 15 to 20 percent rather than eliminating them. Yet early planting alone reduced losses by 15 to 25 percentage points compared with normal sowing and by 25 to 40 points compared with late sowing under the most severe conditions, making it the single most powerful lever available to farmers today. The researchers argue that extension services should shift from static planting calendars to climate-informed sowing decisions, that nitrogen recommendations must be tailored site by site with decision-support tools such as DSSAT, and that Shaki’s resilience traits deserve priority in breeding and seed distribution. Sustaining Ethiopian highland wheat beyond mid-century, they conclude, will also require transformative measures, including heat-tolerant varieties and integrated soil water management, alongside global emissions cuts.

Subject of Research: Model-based assessment of climate change impacts and adaptation strategies for wheat production in the southeastern Oromia highlands, Ethiopia

Article Title: Model-based assessment of climate change impacts on wheat yield under different crop management practices in the southeastern Oromia highlands, Ethiopia

Article References: Tesfaye, Y., Dechassa, N., Alemayehu, Y., & Birhan, D. A. (2026). Model-based assessment of climate change impacts on wheat yield under different crop management practices in the southeastern Oromia highlands, Ethiopia. Environmental Challenges, 25, Article 101701. https://doi.org/10.1016/j.envc.2026.101701

Image Credits: AI Generated

DOI: 10.1016/j.envc.2026.101701

Keywords: wheat, climate change, Ethiopia, DSSAT, crop modeling, sowing date, nitrogen management, SSP scenarios, highlands, food security, adaptation, CORDEX-Africa