Data-Driven Agronomic Solutions to Close Wheat Yield Gaps and Achieve Self-Sufficiency in Uzbekistan

Abstract

CONTEXT Agriculture is a cornerstone of Uzbekistan's economy, accounting for 25 % to the national gross domestic product and employing 26 % of the workforce. Since independence, wheat intensification has been a national priority, with cultivated land expanding from 0.63 million hectares (Mha) to 1.24 Mha and productivity increasing from 1.66 t ha−1 in 1991 to 4.55 t ha−1 in 2023. However, on-farm yields remain below attainable yield, leading to a reliance on wheat imports to meet domestic demand. Closing this yield gap is critical for achieving national wheat self-sufficiency.

OBJECTIVES This study aims to identify key yield-limiting factors and develop evidence-based, agroecologically optimized bundled solutions to enhance wheat productivity in Uzbekistan. By integrating multiple analytical approaches, the research seeks to provide targeted agronomic recommendations for improving sustainability and self-sufficiency.

METHODS A combination of systematic reviews, crop modeling, and machine learning was used to analyze wheat yield gaps and optimize agronomic practices. Agricultural Production Systems sIMulator (APSIM) -Wheat model was calibrated, validated and used to simulate wheat yields over 36-years across four agro-ecological zones (AEZs): Khorezm (arid saline lowland), Kashkadarya (semi-arid highland), Samarkand (semi-arid mid-altitude), and Jizzakh (arid high-altitude). The simulations optimized seeding dates, nitrogen fertilizer rates, cultivar selection, and water management practices. Additionally, a meta-analysis of 90 studies and machine learning were employed to identify key determinants of wheat yield variation.

RESULTS AND CONCLUSIONS To achieve self-sufficiency, Uzbekistan requires an average wheat yield of 6.62 t ha−1, necessitating a 45 % (2.07 t ha−1) increase from current levels (4.55 t ha−1), while the yield gap of 3.25 t ha−1 exists. The study identified nitrogen fertilization, irrigation, rainfall, cultivar selection, and seeding dates as the primary determinants of yield. Wheat yield declined significantly when plant-available water content dropped below 50 %, establishing a critical threshold for sustainable productivity. Precision nutrient management included applying 150–180 kg N ha−1, up to 120 kg P₂O₅ ha−1, and 75 kg K₂O ha−1. Conservation agriculture showed a 26 % increase in yields compared to conventional tillage. High-yielding, stress-tolerant wheat varieties released after 2010 increased wheat productivity by up to 22 %. Seeding between September 15 and October 15 maximized yields, while delayed sowing reduced yield by up to 57 kg ha−1 day−1. Seed rates of 160–180 kg ha−1 improved plant density and yields, preventing excessive competition or underutilization.

SIGNIFICANCE This study offers a science-based framework for improving wheat productivity in Uzbekistan through AEZ-specific, resource-efficient bundled solutions. By integrating crop modeling, machine learning, and systematic reviews, this study provides scalable solutions to enhance input use efficiency, resilience to climate variability, and sustainable intensification. Beyond Uzbekistan, these findings hold relevance for wheat production in other arid and semi-arid regions facing similar food security challenges.