Nitrogen (N) is a critical element in many ecological and biogeochemical processes in forest ecosystems. Cycling of N is sensitive to changes in climate, atmospheric carbon dioxide (CO₂) concentrations, and air pollution. Streamwater nitrate draining a forested ecosystem can indicate how an ecosystem is responding to these changes. We observed a pulse in streamwater nitrate concentration and export at a long-term forest research site in eastern North America that resulted in a 10-fold increase in nitrate export compared to observations over the prior decade. The pulse in streamwater nitrate occurred in a reference catchment in the 2013 water year, but was not associated with a distinct disturbance event. We analyzed a suite of environmental variables to explore possible causes. The correlation between each environmental variable and streamwater nitrate concentration was consistently higher when we accounted for the antecedent conditions of the variable prior to a given streamwater observation. In most cases, the optimal antecedent period exceeded two years. We assessed the most important variables for predicting streamwater nitrate concentration by training a machine learning model to predict streamwater nitrate concentration in the years preceding and during the streamwater nitrate pulse. The results of the correlation and machine learning analyses suggest that the pulsed increase in streamwater nitrate resulted from both (1) decreased plant uptake due to lower terrestrial gross primary production, possibly due to increased soil frost or reduced solar radiation or both; and (2) increased net N mineralization and nitrification due to warm temperatures from 2010 to 2013. Additionally, variables associated with hydrological transport of nitrate, such as maximum stream discharge, emerged as important, suggesting that hydrology played a role in the pulse. Overall, our analyses indicate that the streamwater nitrate pulse was caused by a combination of factors that occurred in the years prior to the pulse, not a single disturbance event.

氮 (N) 是森林生态系统许多生态和生物地球化学过程中的关键元素。氮循环对气候、大气二氧化碳 (CO ₂ ) 浓度和空气污染的变化很敏感。排出森林生态系统的溪水硝酸盐可以表明生态系统如何应对这些变化。我们在北美东部的一个长期森林研究地点观察到溪流硝酸盐浓度和出口的脉冲，导致硝酸盐出口比前十年的观测值增加了 10 倍。溪流硝酸盐的脉冲发生在 2013 年水年的参考流域中，但与明显的扰动事件无关。我们分析了一系列环境变量来探索可能的原因。当我们在给定的河水观测之前考虑变量的先行条件时，每个环境变量与河水硝酸盐浓度之间的相关性始终较高。在大多数情况下，最佳先期超过两年。我们通过训练机器学习模型来预测溪流硝酸盐脉冲之前和期间的溪流硝酸盐浓度，评估了预测溪流硝酸盐浓度的最重要变量。相关性和机器学习分析的结果表明，溪流硝酸盐的脉冲式增加是由于以下两个原因造成的：（1）由于陆地总初级生产力降低而导致植物吸收减少，这可能是由于土壤霜冻增加或太阳辐射减少或两者兼而有之；(2) 由于 2010 年至 2013 年气温升高，净氮矿化和硝化作用增加。此外，与硝酸盐水文输送相关的变量（例如最大河流流量）也变得很重要，表明水文在脉冲中发挥了作用。总的来说，我们的分析表明，溪流硝酸盐脉冲是由脉冲发生前几年发生的多种因素引起的，而不是单一的扰动事件。