Cereal crops are of critical importance to food security and rural livelihoods in Ethiopia, being cultivated on about 10 M ha annually and engaging nearly 16 M smallholders (CSA, 2022a). Moreover, two-thirds of the national cereal production is used for household consumption (CSA, 2022b). Tef, maize, sorghum, and wheat constitute 71% of the cereal area and 77% of the cereal production in country (CSA, 2022a) and average yields at national level more than doubled in the past two decades (Figure 1A; CSA, 2001; CSA, 2022a; Hodson et al., 2020). Yield gains occurred in tandem with area expansion (Figure 1A), translating into impressive increases in cereal production at national level (Figure 1B). Tef remains the most important staple crop in the country (Tadele & Hibistu, 2021), followed by maize (Abate et al., 2015) and wheat (Silva et al., 2021). Past yield trends were the result of increases in fertilizer use, improved genetics, and supportive extension system among other factors (Abate 2015; Assefa et al., 2020; Silva et al., 2023a). Yet, farmers’ yields still only reach about 20-40% of what is possible with agronomic best practices (van Ittersum et al, 2016; Assefa et al. 2020; Silva et al., 2021). Such low productivity levels cannot satisfy cereal demand in the country resulting in high import dependencies. This is especially true for wheat (Silva et al., 2023a; Senbeta et al., 2023) for which imports amounted to 1.4 M t in 2022, corresponding to 865 M US$ (FAOSTAT, 2024). Current yield constraints are related to poor agronomy, including sub-optimal crop management and low input levels (Assefa et al. 2020; Silva et al., 2021), and low adoption of soil fertility management practices, but field-specific constraints are yet to be identified. Overcoming such yield constraints towards yield levels of about 50% of the water-limited yield, equivalent to doubling current farm yields, could ensure cereal self-sufficiency at the national level whereas full yield gap closure would transform the country into a net cereal exporter (van Ittersum et al., 2016).