Nitrogen use efficiency thresholds for rice in Bangladesh: Evidence from large-scale farmer-managed field data

Background Nitrogen (N) over-application in smallholder rice systems reduces fertiliser-use efficiency and may increase reactive N losses and greenhouse gas emissions. However, field-scale nitrogen-use efficiency benchmarks derived from farmer-managed systems in South Asia remain limited, particularly for Bangladesh’s contrasting Aman /Kharif wet-season and Boro /Rabi dry-season rice systems. Objectives This study aimed to: (i) characterise N application rates, grain yield, and partial factor productivity of nitrogen (PFPn) across farmer-managed rice fields in Bangladesh; (ii) identify practical N-efficiency thresholds under real-world management conditions; and (iii) assess how these thresholds differ across seasons and variety types. Methods We analysed 13,144 farmer-managed rice-field observations collected across Bangladesh from 2016 to 2021, covering two growing seasons, three variety types, and 13 represented agroecological zones (AEZs). Nitrogen application rate, grain yield, and PFPn, calculated as grain yield divided by applied N, were evaluated using descriptive statistics, seasonal and varietal comparisons, locally weighted regression, response-surface analysis, multiple regression analysis of PFPn predictors, analysis of variance across N-rate bands, and AEZ-stratified response modelling; adjusted response curves were presented for six selected AEZs. Results Mean N application rate was 120.3 ± 41.7 kg N ha⁻¹ and mean grain yield was 5806 ± 1417 kg ha⁻¹ . Response-surface analysis identified a practical efficiency envelope at 100–140 kg N ha⁻¹ , where mean grain yield reached 5988 kg ha⁻¹ and partial factor productivity of nitrogen (PFPn) was 49.7 kg grain kg⁻¹ N. Beyond 140 kg N ha⁻¹ , yield increments were marginal (255 kg ha⁻¹ for an additional 38 kg N ha⁻¹) while PFPn declined by 46.3%. Approximately 26.2% of surveyed fields applied N at or above 140 kg N ha⁻¹.Boro dry-season irrigated rice produced 32.9% higher mean yield than Aman wet-season rice despite receiving only 21.3% more N; mean PFPn did not differ significantly between seasons. Hybrid varieties achieved the highest mean yield but showed no PFPn advantage over improved inbred varieties. Conclusion Large-scale farmer-managed field evidence indicates that N application above 140 kg N ha⁻¹ provides limited yield benefit while substantially reducing nitrogen-use efficiency. The 100–140 kg N ha⁻¹ range represents a practical efficiency envelope for balancing rice yield and fertiliser productivity in Bangladesh. Implications These findings provide an empirical basis for refining season- and variety-specific N management recommendations in Bangladesh. Targeting over-fertilised fields through improved N placement, including urea deep placement and mechanised applicator support, could improve nitrogen-use efficiency without compromising grain production. Direct measurements of N losses and greenhouse gas emissions are needed to quantify the environmental benefits of these management changes.