The potential link between erosion rates at the Earth’s surface and changes in global climate has intrigued geoscientists for decades1,2 because such a coupling has implications for the influence of silicate weathering3,4 and organic-carbon burial5 on climate and for the role of Quaternary glaciations in landscape evolution1,6. A global increase in late-Cenozoic erosion rates in response to a cooling, more variable climate has been proposed on the basis of worldwide sedimentation rates7. Other studies have indicated, however, that global erosion rates may have remained steady, suggesting that the reported increases in sediment-accumulation rates are due to preservation biases, depositional hiatuses and varying measurement intervals8,9,10. More recently, a global compilation of thermochronology data has been used to infer a nearly twofold increase in the erosion rate in mountainous landscapes over late-Cenozoic times6. It has been contended that this result is free of the biases that affect sedimentary records11, although others have argued that it contains biases related to how thermochronological data are averaged12 and to erosion hiatuses in glaciated landscapes13. Here we investigate the 30 locations with reported accelerated erosion during the late Cenozoic6. Our analysis shows that in 23 of these locations, the reported increases are a result of a spatial correlation bias—that is, combining data with disparate exhumation histories, thereby converting spatial erosion-rate variations into temporal increases. In four locations, the increases can be explained by changes in tectonic boundary conditions. In three cases, climatically induced accelerations are recorded, driven by localized glacial valley incision. Our findings suggest that thermochronology data currently have insufficient resolution to assess whether late-Cenozoic climate change affected erosion rates on a global scale. We suggest that a synthesis of local findings that include location-specific information may help to further investigate drivers of global erosion rates.
Access optionsAccess options
Rent or Buy article
Get time limited or full article access on ReadCube.
All prices are NET prices.
Subscribe to Journal
Get full journal access for 1 year
only $3.90 per issue
All prices are NET prices.
VAT will be added later in the checkout.
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
We thank P. Valla for discussions on the topic. D. Jerolmack and A. Ault provided feedback on an early version of the manuscript. T.F.S. acknowledges support from the Emmy Noether Program of the Deutsche Forschungsgemeinschaft (DFG), grant number SCHI 1241/1-1, and the Helmholtz Association’s Initiative and Networking Fund. P.A.v.d.B acknowledges support from the Institut Universitaire de France (IUF). ISTerre is part of Labex OSUG@2020 (ANR10 LABX56). R.C.T. was funded by the DFG grant number TH 1371/5–1.Reviewer information
Nature thanks R. Flowers, E. Kirby and J. Spotila for their contribution to the peer review of this work.
Extended data figures and tables
Extended Data Fig. 1 Empirical semi-variogram of erosion rates derived from AFT data from the European Alps.
a, Gamma plot, showing the mean of the squared differences (variance) in erosion rates for each distance bin. The plot shows a negative exponential pattern, as described in ref. 18. The variance at a distance of zero, defined as the ‘nugget’, is non-negligible. The nugget is considered to represent the measurement error plus the variance at distances smaller than the smallest measurable distance40. The ‘sill’ is the relatively steady variance value reached at distances beyond the initial increase40. The nugget-to-sill ratio, which in this case is about 0.35, indicates that approximately 35% of the variance is unaccounted for with the spatial correlation function40. b, Box plot of the range of squared differences in erosion rates, with boxes outlining quartiles of data, and error bars corresponding to 1.5 times the inner quartile distance. Horizontal lines within boxes show medians of data within each distance bin.
Extended Data Fig. 2 Global coverage and inferred cause of resolved changes in late-Cenozoic erosion rates.
The data are based on the inversion results of ref. 6. Locations where erosion rates are resolved both from 2–0 Myr ago and from 6–4 Myr ago are shown with thin black circles, whereas locations where erosion rates are resolved in all four 2-Myr time bins between 8 Myr ago and the present are shown with thick black circles. Circles filled with multiple colours indicate contributions from multiple factors.
a, Input exhumation rates are 1 mm yr−1 in the external zone and 0.25 mm yr−1 in the internal zone; they are held constant for 20 Myr. b, Predicted thermochronological ages; different symbols for thermochronological systems and colour scale for ages are as in Fig. 3a. c–e, Predicted erosion rates for a correlation length (λ) of 30 km from 6–4 Myr ago (c), from 2–0 Myr ago (d), and normalized difference in erosion rates from the comparison of c to d (as defined in equation (1)), shown only where the resolution in each time bin is greater than 0.25 (e). Contours in c–e show the predicted resolution. f–h, Same as c–e, but for a correlation length of 10 km. Ma, million years ago.
Circles are locations where erosion rates are resolved (resolution >0.25) from both 2–0 Myr ago and from 6–4 Myr ago from the extended data of ref. 6. Black boxes outline the regions described in the case studies of Mount Cook (Fig. 4), Fiordland (Supplementary Fig. 8) and Marlborough fault system (Supplementary Fig. 16).
a, Input exhumation rates are predicted from a kinematic model of overthrusting along a fault with a ramp–flat geometry30; rates vary from 2.2 mm yr−1 adjacent to the fault to zero far from it, and they are held constant for 10 Myr. b, Predicted thermochronological ages; different symbols for thermochronological systems and the colour scale for ages are as in Fig. 4a. c–e, Predicted erosion rates for a correlation length of 30 km from 6–4 Myr ago (c), from 2–0 Myr ago (d), and normalized difference in erosion rates from the comparison of c to d (as defined in equation (1)), shown only where the resolution in each time bin is greater than 0.25 (e). Contours in c–e show the predicted resolution. f–h, Same as plots c–e, but for a correlation length of 10 km.
a, Input exhumation rates increase smoothly from zero at the eastern boundary to 1.2 mm yr−1 at the western boundary and are held constant for 12 Myr. b, Predicted thermochronological ages. Symbols for thermochronological systems and colour scale for ages are as in Supplementary Fig. 1a. c–e, Predicted erosion rates for a correlation length of 30 km from 6–4 Myr ago (c), from 2–0 Myr ago (d), and normalized difference in erosion rates from the comparison of c to d (as defined in equation (1)), shown only where the resolution in each time bin is >0.25 (e). Contours in c–e show the predicted resolution. f–h, Same as c–e, but for a correlation length of 10 km. i–k, Same as f–h, but for a prior erosion rate (ep) of 0.1 mm yr−1 instead of 0.35 mm yr−1.
This file contains the regional summaries and the Edot2age Matlab script. For each of the regions with erosion histories that are purportedly well resolved by the 1D inversion performed by ref. 6 (32 regions in total, 30 of which show accelerations), we present a literature review to compare the original data and interpretations to the erosion-rate histories reported in ref. 6. The Matlab script was used to determine what thermochronologic age corresponds to a given 1D steady exhumation rate with assumed crustal and thermal properties