Objectives Critical nitrogen dilution curves (CNDCs) have been constructed for nitrogen nutrition diagnosis in many crops. However, the acquisition of the main variable factor, biomass, requires destructive sampling, which significantly reduces the practical feasibility of these curves. We attempted to construct a critical nitrogen dilution curve for flue-cured tobacco using accumulated growing degree days (AGDD, referred to as the "AGDD curve") and compared the parameter values with those of the curve using shoot dry matter accumulation (referred to as the "dry matter curve")

Methods The field experiment was conducted in Linqu County, Shandong Province, in 2025, using the flue-cured tobacco cultivar Zhongchuan 208. The experiment employed a two-factor (planting date and nitrogen application rate) complete block design. The two transplanting dates were April 30 and May 10. The nitrogen application rates included five levels: N1 (0 kg/hm²), N2 (45 kg/hm²), N3 (90 kg/hm²), N4 (135 kg/hm²), and N5 (180 kg/hm²). Aboveground dry matter accumulation in tobacco plants was measured at key growth stages, including the rosette, vigorous growth, bud emergence, topping, and maturity stages. Using a hierarchical Bayesian framework model and the Markov Chain Monte Carlo (MCMC) algorithm, two critical nitrogen dilution curves for the aboveground dry matter accumulation of flue-cured tobacco were constructed. One curve was driven by accumulated temperature (T-curve), and the other by aboveground dry matter accumulation (M-curve). The ability of the two curves to differentiate between nitrogen-limited and non-nitrogen-limited conditions was compared. The critical nitrogen concentration and nitrogen nutrition index were calculated using measured data and fitted using the two curves to test the accuracy of the two curves.

Results Both planting date and N application rate significantly affected the aboveground dry matter accumulation of flue-cured tobacco. The critical nitrogen dilution curves constructed with aboveground dry matter accumulation (PDM) and accumulated growing degree days (AGDD) as driving variables, are respectively Nc = 3.00 PDM^−0.18 and Nc = 2.32 AGDD^−0.33. For the M-curve: the 95% posterior distribution ranges for parameters A1 and A2 were 2.71−3.40 and 0.10−0.29, with mean values of 3.00 and 0.18, respectively. The curve’s uncertainty level ranged from 0.14% to 1.76%, with relative uncertainties for A1 and A2 being 0.23% and 1.12%, respectively. The discrimination ability between the nitrogen surplus group and the nitrogen deficit group was 75%. The normalized root mean square error (n-RMSE) of the Nitrogen Nutrition Index was 23%. For the T-curve: the 95% posterior distribution ranges for parameters A1 and A2 were 2.23−2.41 and 0.23−0.44, with mean values of 2.32 and 0.33, respectively. The curve’s uncertainty level ranged from 0.12% to 1.12%, with relative uncertainties for A1 and A2 being 0.08% and 0.62%, respectively. The discrimination ability between the nitrogen surplus group and the nitrogen deficit group was 82%. The normalized root mean square error (n-RMSE) of the Nitrogen Nutrition Index was 13%.

Conclusions Compared to the dry matter-based curve, the accumulated temperature-based curve offers clearer differentiation between nitrogen deficit and nitrogen surplus conditions. The fitted nitrogen nutrition index demonstrates a stronger linear correlation with the actual nitrogen nutrition index, evidenced by a lower normalized root mean square error (n-RMSE) of 13%, indicating higher simulation accuracy. Given the practicality and accessibility of accumulated temperature data, the critical nitrogen dilution curve developed using accumulated temperature as the driving variable is well-suited for non-destructive diagnosis of nitrogen status in flue-cured tobacco. .