The temporal dynamics of dissolved organic matter (DOM) are inherently linked with the functioning of aquatic ecosystems. Because DOM represents a complex mixture of millions of different compounds, the statistical analysis of DOM dynamics poses a huge challenge. Here, we present a statistical approach based on hierarchical clustering of time series that groups DOM compounds with synchronous dynamics. We applied this approach to time series of Fourier-transform ion cyclotron resonance mass spectrometry data of DOM sampled over a period of 26 months near Helgoland, an island in the Southern North Sea. We identified three DOM clusters, which represented a total of 1392 different molecular formulae and showed distinct chemical properties and noticeably compound matches within the PubChem database. Correlations of the three DOM clusters with abundance data of prokaryote and phytoplankton species and with environmental parameters provided consistent indications on the potential origin of the clustered compounds. The first cluster integrated terrestrial DOM originating from riverine discharge reaching Helgoland waters. The second cluster was attributed to DOM related to phytoplankton and microbial activity, whereas the third cluster was interpreted as representing the marine refractory DOM background. Accordingly, while further partitioning divided each of the first two clusters into five sub-clusters with distinct temporal dynamics and molecular characteristics, the third cluster persisted as a stable feature. Applying a purely mathematical approach, we thus confirmed the differential dynamics of individual DOM compounds and compound groups and showed that temporal dynamics of dissolved molecules are linked to their origin and transformation history.

中文翻译：

---

具有同步动力学的溶解有机化合物共享化学性质和来源

溶解有机物 (DOM) 的时间动态与水生生态系统的功能有着内在的联系。由于 DOM 代表了数百万种不同化合物的复杂混合物，因此对 DOM 动力学的统计分析提出了巨大的挑战。在这里，我们提出了一种基于时间序列层次聚类的统计方法，该方法将具有同步动态的 DOM 化合物分组。我们将这种方法应用于在北海南部黑尔戈兰岛附近 26 个月内采样的 DOM 的傅立叶变换离子回旋共振质谱数据的时间序列。我们确定了三个 DOM 簇，它们总共代表 1392 个不同的分子式，并在 PubChem 数据库中显示出不同的化学性质和明显的化合物匹配。三个 DOM 聚类与原核生物和浮游植物物种的丰度数据以及环境参数的相关性为聚类化合物的潜在来源提供了一致的指示。第一个集群整合了源自到达黑尔戈兰水域的河流排放的陆地 DOM。第二个集群归因于与浮游植物和微生物活动相关的 DOM，而第三个集群被解释为代表海洋耐火 DOM 背景。因此，当进一步划分将前两个簇中的每一个分为具有不同时间动态和分子特征的五个子簇时，第三个簇作为一个稳定的特征持续存在。应用纯数学方法，