The determination of critical nitrogen (N) dilution curves (CNDCs) has been the subject of intensive study for the last decades due to its relevance to diagnosing crop N status. However, to date, minimum steps, data requirements and robust science-based guidelines to estimate CNDCs based on experimental data have not been formalized in the literature to ensure the reliability of an established nutrient dilution curve. In this study, we conducted a systematic review of the literature on CNDCs, described the main characteristics of the datasets used to establish CNDCs in the past, and finally identified a set of criteria specifying the minimum characteristics that a dataset must satisfy to be used to establish an accurate CNDC. Published CNDC studies showed large heterogeneity in the number of experiments (from 1 to 35), fertilizer N rates (from 2 to 7), and sampling times (from 2 to 16) used to fit the CNDCs. Given that, we quantified the sensitivity of the CNDC parameters to five dataset characteristics (number of experiments, N rates, and sampling times, attainment of W_max plateau, and precision of the studies) using a Bayesian statistical model on a case-study estimation for maize. We found that the number of experiments is the main factor affecting the uncertainty of the fitted CNDC. Bootstrapping analyses showed that accurate CNDCs can be fitted with experiments having at least three N rates. Increasing the number of sampling times was more effective than using more than three N rates for reducing the uncertainty in the estimation of the parameters for the CNDCs. We also showed that more reliable CNDCs can be obtained when weighting each data by its precision. We concluded our analysis with formal recommendations of the minimum steps and data requirements to fit CNDC. The criteria established here should serve as a guide for the establishment of reliable dilution curves for N and other nutrients in the agricultural and biological sciences.

临界氮 (N) 稀释曲线 (CNDC) 的测定由于与诊断作物 N 状态有关，因此在过去几十年中一直是深入研究的主题。然而，迄今为止，基于实验数据估计 CNDCs 的最小步骤、数据要求和稳健的科学指南尚未在文献中正式确定，以确保已建立的营养稀释曲线的可靠性。在这项研究中，我们对有关 CNDCs 的文献进行了系统回顾，描述了过去用于建立 CNDCs 的数据集的主要特征，并最终确定了一组标准，指定数据集必须满足的最低特征才能用于建立准确的CNDC。已发表的 CNDC 研究表明，实验数量（从 1 到 35）存在很大差异，用于拟合 CNDCs 的肥料 N 率（从 2 到 7）和采样时间（从 2 到 16）。鉴于此，我们量化了 CNDC 参数对五个数据集特征（实验次数、N 速率和采样时间、W_最大限度高原和研究的精确度）使用贝叶斯统计模型对玉米进行个案研究估计。我们发现实验次数是影响拟合CNDC不确定性的主要因素。自举分析表明，准确的 CNDC 可以与至少三个 N 速率的实验相匹配。增加采样次数比使用超过三个 N 采样率更有效地减少 CNDC 参数估计的不确定性。我们还表明，当通过精度对每个数据进行加权时，可以获得更可靠的 CNDC。我们以适合 CNDC 的最低步骤和数据要求的正式建议结束了我们的分析。