Biophysical and Socio-Economic Opportunities and Constraints for Bridging Wheat Yield Gaps in Central Asia

CONTEXT: Wheat is central to food security in Central Asia, but the magnitude, composition, and drivers of wheat yield gaps remain poorly quantified, particularly across contrasting production systems in Uzbekistan and Tajikistan. Decomposing whether yield reductions arise from technology, resource/input-use, or management constraints is essential for designing targeted sustainable intensification strategies.OBJECTIVE: This study quantified wheat yield gaps across major wheat-producing regions of Uzbekistan and Tajikistan, decomposed these gaps into technology, resource/input-use, and management components, and identified the main biophysical, agronomic, and socio-economic factors associated with observed yield variation.METHODS: Farm-level survey data were collected from 2,216 wheat-producing households, including 1,526 households across eight provinces in Uzbekistan and 690 households across three provinces of Tajikistan. Yield gaps were estimated using APSIM Next Generation benchmark yields, stochastic frontier analysis, and empirical upper-tail farmer yields. Random forest regression and partial dependence plots were used to identify important yield predictors and response thresholds across irrigated, rainfed, and supplementally irrigated wheat systems.RESULTS AND CONCLUSIONS: Wheat yield-gap patterns differed clearly among production systems. Irrigated wheat in Uzbekistan achieved 67% of the water-limited benchmark yield of 7.8 t ha⁻¹, with technology, management, and resource/input-use gaps accounting for 20%, 10%, and 3% of the benchmark, respectively. Rainfed wheat in Uzbekistan achieved only 29% of the water-limited benchmark yield of 3.5 t ha⁻¹, with technology-related constraints accounting for 57%, indicating large scope for improved dryland wheat technologies and extension support. In Tajikistan, wheat cultivated with 1–3 supplemental irrigations achieved 48% of the water-limited benchmark yield of 4.2 t ha⁻¹, with resource/input-use constraints accounting for 48% of the benchmark. Random forest analysis showed high predictive performance in Uzbekistan and identified potassium application rate, irrigation status, nitrogen rate, seed rate, and variety release year as important yield predictors. Partial dependence analysis indicated response thresholds around 50 kg K ha⁻¹, 100 kg N ha⁻¹, and 150 kg seed ha⁻¹ in Uzbekistan. In Tajikistan, province was the dominant predictor, indicating that location-specific production conditions explained more yield variation than input-rate variables alone.SIGNIFICANCE: Wheat yield gaps in Central Asia are system- and site-specific and require integrated interventions. Uzbekistan requires integrated customized and bundled solutions including variety dissemination, and precision nutrient management, whereas Tajikistan requires improved access to inputs, credit, and location-specific advisory services. The strong association between potassium application and yield in Uzbekistan also suggests that fertilizer recommendations should move beyond nitrogen- and phosphorus-centred approaches toward more balanced nutrient management.