The eastern margin of the Tibetan Plateau represents one of the morphologically most active regions on Earth, where the interplay of recent crustal deformation and subsequent fluvial landscape adjustment has affected the course of continental-scale rivers by river piracy events. Based solely on field observations, such an event has been hypothesised for two of the largest tributaries of the Yangtze River: the Jialing and Hanjiang Rivers. To test this hypothesis, we employ a novel combination of independent methods including a provenance study based on age distributions of detrital zircons from both modern riverbeds and river terraces and a morphometric analysis of river channels and drainage divides. We supported the morphometric analysis with a time-dependent numerical model describing the evolution of river channel long profiles and drainage divides in a succession of river capture events. Analysed zircon ages show clearly distinguishable distributions for the modern Jialing and Hanjiang Rivers, but similar distributions for the recent Hanjiang River up to its topmost terraces. This suggests that the capture of the Hanjiang headwaters by the Jialing River is unlikely to have taken place during the last 1.2 million years. However, several knickpoints in the main stem and the tributaries of the Jialing River cluster at an elevation of about 900 m and separate steeper (downstream) from less steep channel segments (upstream), which is consistent with the morphological expression of a major capture event. χ mapping indicates drainage divide asymmetry at catchment scale with on average steeper rivers on the Jialing side, whereas Gilbert metrics show a symmetric divide at hillslope scale. This numerical model explains this apparent contradiction by the travel time of capture-related knickpoints from the capture point towards the watershed, where χ mapping indicates divide asymmetry immediately after the river capture, while Gilbert metrics are only affected as soon as the knickpoints reach the channel heads and the divide effectively starts moving. Based on knickpoint travel distances and constraints on regional incision / uplift rates, we estimate the possible date of river capture to be the Pliocene. This is earlier than the formation of the terraces investigated in the provenance study but recent enough that most of the drainage divides are still unaffected and currently almost stable. Only the wind gap located in the almost dry valley connecting the two competing drainage systems is likely to have shifted towards the Hanjiang side. We suggest that this resulted in the capture of another important tributary of the Hanjiang River (the Heishui River) by the Jialing drainage system. Our results illustrate the complex evolution of drainage networks along the eastern margin of the Tibetan Plateau, and highlight the importance for combining provenance and morphometric analyses in regions of active landscape rejuvenation where river captures are likely to occur.

青藏高原东缘是地球上形态最活跃的地区之一，最近的地壳变形和随后的河流景观调整的相互作用通过河流海盗事件影响了大陆尺度河流的流向。仅根据实地观察，有人假设长江的两条最大支流嘉陵江和汉江发生了这样的事件。为了验证这一假设，我们采用了一种新的独立方法组合，包括基于现代河床和河流阶地碎屑锆石年龄分布的物源研究以及河道和排水分水岭的形态测量分析。我们使用瞬态数值模型支持形态测量分析，该模型描述了在一系列河流捕获事件中河道长剖面和排水沟的演变。分析的锆石年龄显示现代嘉陵江和汉江的分布清晰可辨，但最近的汉江直到其最顶层阶地的分布相似。这表明在过去的 120 万年里，嘉陵江对汉江源头的捕获不太可能发生。然而，嘉陵江干流和支流中的几个关键点聚集在海拔约 900 m 处，将陡峭的河段（下游）与陡峭的河道段（上游）分开，这与一次重大捕获事件的形态表达是一致的。 . 但最近的汉江到其最顶层阶地的分布相似。这表明在过去的 120 万年里，嘉陵江对汉江源头的捕获不太可能发生。然而，嘉陵江干流和支流中的几个关键点聚集在海拔约 900 m 处，将陡峭的河段（下游）与陡峭的河道段（上游）分开，这与一次重大捕获事件的形态表达是一致的。 . 但最近的汉江到其最顶层阶地的分布相似。这表明在过去的 120 万年里，嘉陵江对汉江源头的捕获不太可能发生。然而，嘉陵江干流和支流中的几个关键点聚集在海拔约 900 m 处，将陡峭的河段（下游）与陡峭的河道段（上游）分开，这与一次重大捕获事件的形态表达是一致的。 .χ映射表明集水区尺度上的排水分水线不对称，嘉陵一侧的河流平均较陡峭，而吉尔伯特指标显示了山坡尺度上的对称分水岭。这个数值模型通过从捕获点到分水岭的捕获相关的拐点的旅行时间来解释这个明显的矛盾，其中χ映射表明在河流捕获后立即划分不对称，而吉尔伯特指标仅在节点到达渠道头并且划分有效开始移动时才会受到影响。根据拐点行进距离和区域切口/抬升率的限制，我们估计河流捕获的可能日期是上新世。这早于来源研究中调查的阶地的形成，但最近足够了，大多数排水分水岭仍然不受影响，目前几乎稳定。只有位于连接两个相互竞争的排水系统的几乎干涸的山谷中的风隙可能已经向韩江一侧移动。我们认为这导致嘉陵排水系统夺取了汉江的另一个重要支流（黑水河）。