A model suggests that active deformation in mountains causes river networks to constantly reorganize, providing an explanation for the paradoxical formation of almost flat surfaces high in craggy mountain ranges. See Letter p.526
For more than a century, Earth scientists' curiosity has been piqued by the existence of areas with low topographic relief, some nearly flat, perched high in rugged mountain ranges. One common explanation1 posits that these surfaces are relicts of large peneplains — low-relief features formed as the ultimate result of fluvial erosion — that were once close to sea level before being uplifted by mantle convection or plate tectonics and then dissected by rivers or streams. On page 526 of this issue, Yang et al.2 propose a very different mechanism, whereby low-relief surfaces in mountain landscapes form transiently as a result of the dynamic reorganization of river networks.
One of Earth science's greatest challenges is to document vertical movements of the crust at geological timescales. Step-like landforms called terraces are commonly used as passive markers of small-scale deformations that occurred during the past million years or less, but it is usually impossible to use such markers for large uplifted regions and events that occurred over longer time periods. Geologists are therefore forced to use other palaeoaltimetric methods3 that generally have large uncertainties and are difficult to implement over large regions.
The lack of reliable tools is particularly problematic for studies of regional uplift caused by deep-seated geodynamic processes, or when investigating the upheaval of wide orogenic plateaus (which form as a result of colliding tectonic plates). In these cases, it is tempting to find other passive markers, such as the dissected and uplifted remnants of peneplains that are assumed to have once been nearly horizontal and close to sea level. By interpreting low-relief surfaces perched across the southeast Tibetan plateau in this way, a broad south-dipping slope of the plateau surface has been inferred4, which was proposed as evidence of differential thickening of the underlying crust. Aspects of this interpretation have been questioned5,6, but a two-stage mechanism has always been assumed, in which the formation of low-relief surfaces precedes river dissection.
Yang et al. challenge this idea. They first propose a new description of the fluvial network that drains southeastern Tibet, by using a modified metric7 of the stream power equation, a widely used model in which fluvial incision into bedrock primarily depends on river slope and discharge. This model predicts that when regional uplift increases (as expected in the dissected-peneplain scenario), the profiles of all the major rivers and tributaries will display a change of slope (a 'knickpoint') at similar elevations8 if the geometry of the fluvial network is fixed. By examining several regions in southeastern Tibet, the researchers show that the main tributaries do not display such a regular pattern. Instead, knickpoint elevations are widely scattered.
The authors then show that the long profiles of tributaries that drain low-relief surfaces are systematically shallower than expected, whereas those of tributaries that drain the slopes surrounding the low-relief surfaces are steeper. The authors interpret these peculiar features9 as evidence of recent drainage captures — the diversion of headwater regions of main rivers to nearby tributaries — and dynamic reorganization of the river network (Fig. 1). The low-relief surfaces therefore cannot be relict landscapes, and must instead correspond to areas that formerly had normal relief, but where the main stream has lost its power to incise bedrock because its headwater has been captured. If subjected to sustained uplift, such areas would gain elevation, and their local relief would be smoothed down because of erosion of the surrounding hill slope. The process is a kind of topographic inversion, because formerly incised valley bottoms and rugged topographies end up as flatter surfaces at high elevations.
To simulate this mechanism, the authors built a computational model in which a section of crust is squeezed between two rigid plates while undergoing constant thickening — a situation that may have occurred in southeastern Tibet10. This model and the associated video (see the paper's Supplementary Information2) wonderfully illustrate how crustal shrinkage reduces the overall drainage area that feeds the rivers, causing continuous reorganization in which 'victim' rivers lose their upstream area to 'pirate' river networks. Crucially, they demonstrate how stream piracy in this deforming setting is a self-sustained or cascading mechanism. Once a network has lost part of its drainage area, its ability to incise the crust decreases, its elevation above surrounding major rivers increases, and it becomes easier for pirate networks to capture even more of its drainage area.
Yang and colleagues' decision to reject the classical explanation of the low-relief surfaces in southeastern Tibet merits some discussion. First, part of the observed scattering of knickpoint elevations might result from river-incision behaviour that is not encapsulated in the simplified stream power model used by the authors, from local variations of tectonics, or from the initial topography of the raised low-relief surface. Second, the authors' model does not easily apply to the northern part of the studied region (north of 30° N), where there have been low rates of erosion during the past 50 million years11 and where reduced river incision has probably limited the reorganization of fluvial networks; low-relief surfaces in this region represent more than half of the landscape, whereas the model predicts a much smaller fraction.
Future modelling should investigate the roles of horizontal strain and vertical uplift in the dynamics of river capture. Would the model have led to such a dynamic stream reorganization and production of low-relief areas if the authors had considered much lower finite erosion11 and the present north–south flow deformation12 in southeastern Tibet, which is in sharp contrast to the east–west shortening simulated in the model?
In any case, the process proposed by Yang et al. changes our thinking about the genesis of low-relief surfaces in mountainous areas, and will reignite debate about their origin and use as uplift markers in other orogenic settings, such as the Pyrenees13 or the Eastern Andes cordillera14. The multicapture scenario will also alter the way we look at the evolution of river networks in many landscapes, and, as a corollary, affect interpretations of sedimentary archives that have recorded past erosion of regions undergoing deformation. Finally, if stream reorganization is highly sensitive to horizontal and vertical tectonics, it highlights both the richness and the complexities provided by river-network geometry for unravelling the tectonic history of orogenic features10.Footnote 1
Davis, W. M. Ann. Assoc. Am. Geogr. 1, 21–83 (1911).
Yang, R., Willett, S. D. & Goren, L. Nature 520, 526–529 (2015).
Kohn, M. J. (ed.) Paleoaltimetry: Geochemical and Thermodynamic Approaches (Mineral. Soc. Am., 2007).
Clark, M. K. et al. J. Geophys. Res. 111, F03002 (2006).
Hoke, G. D., Liu-Zeng, J., Hren, M. T., Wissink, G. K. & Garzione, C. N. Earth Planet. Sci. Lett. 394, 270–278 (2014).
Liu-Zeng, J., Tapponnier, P., Gaudemer, Y. & Ding, L. J. Geophys. Res. 113, F04018 (2008).
Perron, J. T. & Royden, L. Earth Surf. Process. Landforms 38, 570–576 (2013).
Niemann, J. D., Gasparini, N. M., Tucker, G. E. & Bras, R. L. Earth Surf. Process. Landforms 26, 1317–1332 (2001).
Willett, S. D., McCoy, S. W., Perron, J. T., Goren, L. & Chen, C.-Y. Science 343, 1248765 (2014).
Hallet, B. & Molnar, P. J. Geophys. Res. 106, 13697–13709 (2001).
Clark, M. K. et al. Geology 33, 525–528 (2005).
Shen, Z.-K., Lü, J., Wang, M. & Bürgmann, R. J. Geophys. Res. 110, B11409 (2005).
Gunnell, Y. et al. Earth Planet. Sci. Lett. 278, 208–218 (2009).
Gubbels, T. L., Isacks, B. L. & Farrar, E. Geology 21, 695–698 (1993).