www.sciencemag.org SCIENCE VOL 343 7 FEBRUARY 2014 617

PERSPECTIVES

Limits of Soil Production?

GEOCHEMISTRY

Arjun M. Heimsath

Steep mountain regions can weather faster

and produce soil more quickly than previously

thought.

Rocky mountain ranges may appear

static but are constantly in motion.

Tectonic forces push the mountains

up, while physical and chemical processes

break rocks down to sediment that is trans-

ported to river plains and ultimately to the sea.

This cycle is thought to regulate global cli-

mate over million-year time scales ( 1) while

also responding to climate forcing itself ( 2).

It remains unclear whether mountain uplift

drives climate change, or whether climatic

cooling drives uplift by causing faster ero-

sion ( 3). On page 637 of this issue, Larsen et

al. ( 4) provide data that help to quantify these

controls on mountain building, reporting

faster sediment production rates and higher

chemical weathering rates than previously

measured. Their results also provide key

insights into soil sustainability over shorter

time scales ( 5).

Mountain building can only drive global

climate trajectories if weathering of silicate

rocks removes enough CO2 from the atmo-

sphere over geologic time scales to lower

atmospheric concentrations of this critical

greenhouse gas. Proponents for mountain

controls on climate point to extensive bed-

rock exposure, ready for weathering, in young

mountain ranges and to the temporal correla-

tion between periods of active mountain build-

ing and global cooling. However, it remains

unclear whether mountainous regions are big

enough and weather fast enough to account

for the CO2 drawdown needed to change

climate ( 6) and whether the few measured

weathering rates can be extrapolated across

mountain ranges. It is also unknown whether

there are limits to the rate of soil production,

which helps to govern the presence of soil.

Soil can only persist at a given location

if erosion is not removing it faster than it is

being produced. On steep slopes there are

typically extensive areas of bare rock, as well

as areas where soil cover allows forests, tun-

dra, and other forms of life to exist (see the

fi gure). These steep slopes are thought, there-

fore, to exist at the threshold of soil produc-

tion and provide the opportunity to examine

the complex response of a hillslope to a wide

range of erosion rates ( 7). If there is an upper

limit to soil production rates ( 8), it is unclear

how soil cover can be present in regions

thought to be eroding well beyond the pur-

ported upper soil production limit.

Many challenges remain before these

debates can be fully resolved. First and fore-

most, rates of soil production, erosion, and

chemical weathering must be quantified

across different landscapes. Larsen et al.

now report exactly these data from the west-

ern Southern Alps of New Zealand. They

also document pervasive soil and vegetation

cover on slopes that erode faster than 1 mm

per year. Their fi ndings are based on concen-

trations of rare isotopes (10Be) produced in

the very grains of silicate minerals that react

with CO2 during chemical weathering. This

isotope is produced by cosmic-ray bombard-

ment of Earth’s surface and is widely used to

determine average erosion rates and point-

specifi c soil production rates. Larsen et al.

also measure concentrations of a nonreactive

element (Zr) in the same samples to quan-

tify the degree of chemical depletion in the

weathered rocks producing the sediments.

They use these depletion fractions in concert

with the erosion and soil production rates to

infer chemical weathering rates.

These data are not easy to come by. Larsen

and colleagues traversed some of Earth’s

most rugged topography to collect their sam-

ples. The data reveal an exponential decline

of soil production with increasing soil thick-

ness, defi ning a higher soil production func-

tion, at higher erosion rates, than previously

predicted. The data thus support the view that

feedback between erosion and soil produc-

tion enables rapidly eroding landscapes to

retain a cloak of soil ( 7). Larsen et al. also

show that chemical weathering rates are

higher than a previously suggested kinetically

controlled limit ( 9), providing key evidence

for the important role that mountains play in

controlling climate.

Resolving the couplings between silicate

weathering and global climate requires simi-

lar data from both mountains and lowlands.

Sample collection is the fi rst challenge. The

logistics are demanding even in locations not

ravaged by war or severely affected by human

development. Extracting and measuring 10Be

concentrations is expensive, involves spe-

cialized laboratoriess, and is time-consum-

ing. 10Be concentrations yield rates averaged

over hundreds to hundreds of thousands of

years, making it diffi cult to use the method

to quantify rates that change over time. Cal-

culated chemical weathering rates depend

on the assumption that Zr concentrations are

homogenous in unweathered rock and that

Competing processes. The steep slopes of Mount Lukens rise abruptly from the suburban sprawl of Los Ange-les, California. Landslides and debris fl ows remove sediment produced from the weathered rock, while active tectonic forces push the mountains higher. The extent of rock outcrop is shown in red on the overlain shaded relief map ( 12). Larsen et al. report that soil-mantled landscapes in much wetter New Zealand can persist in rapidly uplifting mountain ranges because of high rates of soil production and chemical weathering.

PH

OT

O C

RE

DIT

: R

OM

AN

A. D

IBIA

SE

/CA

LIF

OR

NIA

IN

ST

ITU

TE

OF

TE

CH

NO

LO

GY

School of Earth and Space Exploration, Arizona State Uni-versity, Tempe, AZ 85287, USA. E-mail: arjun.heimsath@asu.edu

Published by AAAS

7 FEBRUARY 2014 VOL 343 SCIENCE www.sciencemag.org 618

PERSPECTIVES

Next Season’s Hurricanes

ATMOSPHERE

Gabriel A. Vecchi 1 and Gabriele Villarini 2

Seasonal predictions of hurricane activity

remain challenging, especially at a regional

scale.

Tropical cyclones (TCs) are a haz-

ard to life and property ( 1, 2), as was

tragically apparent following Super

Typhoon Haiyan’s landfall in the Philippines

in 2013 and Hurricane/extratropical system

Sandy’s landfall in the New York tri-state

area in 2012. Yet TCs also provide vital water,

sometimes relieving drought ( 3). Predictions

of the path and intensity of individual TCs

are usually suffi ciently good sev-

eral days in advance that action can

be taken. In contrast, predictions

of seasonal TC activity months in

advance must still be made more

regionally relevant to produce

information that can be acted on,

for example, to improve storm pre-

paredness.

Seasonal TC predictions focus

on the probability of a range of

outcomes integrated over broad

regions, rather than the individual

storms and narrower geographic

foci used in 3- to 5-day weather forecasts.

Whereas weather-scale TC predictions may

lead to targeted actions such as evacuations,

seasonal predictions are currently used to

develop and price instruments to distribute

risk (such as insurance). Improved skill and

regional specifi city of seasonal TC prediction

could be useful to water resource, emergency,

and energy management efforts. Furthermore,

a better ability to forecast seasonal hurricanes

can help build a more robust understanding of

the ways in which climate controls hurricane

activity, perhaps leading to increased confi -

dence in multidecadal hurricane projections.

Basin-Wide Success

In recent years, several approaches have

been developed to predict seasonal TC activ-

ity averaged over an entire basin, such as the

North Atlantic or Northwest Pacifi c, sev-

eral months before the season in question.

These approaches include statistical ( 4) and

dynamical general circulation models ( 5–

7), as well as hybrid statistical-dynamical

methods ( 8– 10). They are used

in operational seasonal TC out-

looks made by meteorological

agencies. Evaluated over multi-

ple years and decades, these pre-

dictions are skillful at predicting

the year-to-year changes in the

total number of hurricanes, when

compared to forecasts based on

knowing only the long-term aver-

age or activity over the years pre-

ceding a season. The predictive

skill of basin-wide activity can

be seen in individual years. For

example, for months prior to the 2010 sea-

son, Atlantic hurricane frequency was con-

sistently predicted to be large (see the fi rst

figure), and 2010 was indeed the second

most active hurricane season since 1970.

Learning from Failure

Even though predictions are skillful in pre-

dicting year-to-year changes in TC activ-

ity over many years, they are not perfect. A

glaring example is the recent 2013 Atlantic

hurricane season (see the fi rst fi gure), for

which nature failed to follow the almost

unanimous prediction that the North Atlan-

tic should have a normal to slightly enhanced

number of hurricanes (~6 to 9). Instead, it

was one of the most anemic hurricane sea-

sons ever recorded.

A season like 2013 is humbling. Yet only

by understanding and learning from past

failed predictions will the prediction com-

munity be able to successfully move for-

ward. In disentangling the causes of the

low hurricane activity of 2013, we must ask

ourselves whether our prediction systems

neglected something foreseeable, and then

account for this in future predictions. But

1Geophysical Fluid Dynamics Laboratory, National Oceanic and Atmospheric Administration, Princeton, NJ 08542, USA. 2IIHR-Hydroscience and Engineering, University of Iowa, Iowa City, IA 52246, USA. E-mail: gabriel.a.vecchi@noaa.gov

Pre

dic

ted a

nd o

bserv

ed a

nnual num

ber

of

Nort

h A

tlanti

c h

urr

icanes

20

16

12

8

4

0

Obs.

1981 to 2012

average

From (8) MarchFrom (9) AprilFrom (6) AprilFrom (10) MayFrom (7) JuneFrom (4) JuneFrom (5) June

Obs.

2010 2013

p(N

> 1

1)

= 5

5%

p(N

< 3

) =

3%

Seasonal North Atlantic hurricane prediction. The very active 2010 season was successfully pre-dicted by a range of methodologies ( 4– 10), but these prediction systems generally failed for the very inactive 2013 season. Central estimates are cir-cles; vertical bars show ranges [70% range for ( 8, 10); ±1σ for ( 4– 7, 9)]. The legend gives the month when each prediction was issued. For ( 8), the pre-dicted exceedance probabilities for the observed hurricane counts are given to the left of the vertical bar. For data, see supplementary materials. C

RE

DIT

: (L

EF

T) V. A

LTO

UN

IAN

/SC

IEN

CE

the Zr is immobile in solution. Given these

challenges, field-based data such as those

reported by Larsen et al. are indispensable

and provide crucial tests for models ( 10).

Obtaining similar data for agricultural

soils presents challenges not faced by studies

such as that of Larsen et al., yet this is where

the greatest societal concerns lie ( 11). As

food demand increases, so too will the need

to conserve Earth’s soil resources. The extent

of soil conservation measures will depend

on which side of the soil production−erosion

balance agricultural soils fall.

References 1. M. E. Raymo, W. F. Ruddiman, Nature 359, 117 (1992).

2. K. X. Whipple, Nat. Geosci. 2, 97 (2009).

3. P. Molnar, P. England, Nature 346, 29 (1990).

4. I. J. Larsen et al., Science 343, 637 (2014); 10.1126/sci-

ence.1244908.

5. D. R. Montgomery, Proc. Natl. Acad. Sci. U.S.A. 104,

13268 (2007).

6. J. K. Willenbring, A. T. Codilean, B. McElroy, Geology 41,

343 (2013).

7. A. M. Heimsath, R. A. DiBiase, K. X. Whipple, Nat. Geosci.

5, 210 (2012).

8. J. L. Dixon, F. von Blanckenburg, C. R. Geosci. 344, 597

(2012).

9. K. L. Ferrier, J. W. Kirchner, Earth Planet. Sci. Lett. 272,

591 (2008).

10. A. J. West, Geology 40, 811 (2012).

11. D. R. Montgomery, Dirt: The Erosion of Civilizations

(Univ. of California Press, Berkeley, CA, 2007).

12. R. A. DiBiase, A. M. Heimsath, K. X. Whipple, Earth Surf.

Process. Landf. 37, 855 (2012).

10.1126/science.1250173

Published by AAAS

particularly relevant to nanoscale substrates, wherethe curvature is appreciable on molecular lengthscales. For example, solid domains consisting ofnarrow stripes radiating outward from a circularcore have been observed in diverse systems, suchas phases on lipid vesicles (17) and metal coatingson nanoparticles (14). Our analysis suggests thatthe width of the stripes and the core size shouldbe determined by the interplay among curvature,elasticity, and bulk energy. A similar interplaymay affect the assembly pathways of viral capsids.Recent in vitro experiments (26) show that capsidscan assemble following a two-step mechanismanalogous to our crystallization process: The capsidproteins first attach to a substrate (an RNAmolecule)and then bind together into an ordered shell. Theintermediate states of this process—long a sub-ject of speculation (11)—might contain voids that,like those in our curved crystals, help the capsidsavoid excess elastic stress (27, 28).

Similar rules may govern other crystallizationand packing problems where global geometry isincompatible with local lattice packing. For exam-ple, crystallization of tetrahedra is frustrated in flat(Euclidean) 3D space (6).However, logs and helicesof tetrahedra are prevalent in first-order phasetransitions of tetrahedra from disordered to densequasicrystalline phases (7). The existence of thesestructures, which were first described by Bernal(29), may reflect a similar physical principle, inwhich growth along one dimension allows a crystalto escape the restrictions of geometrical frustration.

References and Notes1. N. Steno, The Prodromus of Nicolaus Steno’s Dissertation

Concerning a Solid Body Enclosed by Process of NatureWithin a Solid; an English Version with an Introduction

and Explanatory Notes by John Garrett Winter (Macmillan,London, 1916).

2. I. Sunagawa, Crystals: Growth, Morphology, & Perfection(Cambridge Univ. Press, Cambridge, 2005).

3. A. T. Skjeltorp, Phys. Rev. Lett. 58, 1444–1447 (1987).4. J. S. Langer, Rev. Mod. Phys. 52, 1–28 (1980).5. K. G. Libbrecht, Rep. Prog. Phys. 68, 855–895 (2005).6. D. Nelson, F. Spaepen, Solid State Phys. 42, 1 (1989).7. A. Haji-Akbari et al., Nature 462, 773–777 (2009).8. A. R. Bausch et al., Science 299, 1716–1718 (2003).9. P. Lipowsky, M. J. Bowick, J. H. Meinke, D. R. Nelson,

A. R. Bausch, Nat. Mater. 4, 407–411 (2005).10. W. T. M. Irvine, V. Vitelli, P. M. Chaikin, Nature 468,

947–951 (2010).

11. W. H. Roos, R. Bruinsma, G. J. L. Wuite, Nat. Phys. 6,733–743 (2010).

12. I. R. Bruss, G. M. Grason, Proc. Natl. Acad. Sci. U.S.A.109, 10781–10786 (2012).

13. G. A. DeVries et al., Science 315, 358–361 (2007).14. H. Bao, W. Peukert, R. Klupp Taylor, Adv. Mater. 23,

2644–2649 (2011).15. R. Lipowsky, Nature 349, 475–481 (1991).16. J. Korlach, P. Schwille, W. W. Webb, G. W. Feigenson,

Proc. Natl. Acad. Sci. U.S.A. 96, 8461–8466 (1999).17. A. Bandekar, S. Sofou, Langmuir 28, 4113–4122

(2012).18. See supplementary materials on Science Online.19. W. W. Mullins, R. F. Sekerka, J. Appl. Phys. 34,

323 (1963).20. W. W. Mullins, R. F. Sekerka, J. Appl. Phys. 35, 444 (1964).21. S. Schneider, G. Gompper, Europhys. Lett. 70, 136–142

(2005).22. A. Y. Morozov, R. F. Bruinsma, Phys. Rev. E 81, 041925 (2010).23. Y. Chushak, A. Travesset, Europhys. Lett. 72, 767–773

(2005).24. D. R. Nelson, Defects and Geometry in Condensed Matter

Physics (Cambridge Univ. Press, Cambridge, 2002).25. C. Majidi, R. S. Fearing, Proc. R. Soc. London Ser. A 464,

1309 (2008).26. R. F. Garmann, M. Comas-Garcia, A. Gopal, C. M. Knobler,

W. M. Gelbart, J. Mol. Biol. 10.1016/j.jmb.2013.10.017(2013).

27. W. S. Klug, W. H. Roos, G. J. Wuite, Phys. Rev. Lett. 109,168104 (2012).

28. A. Luque, D. Reguera, A. Morozov, J. Rudnick, R. Bruinsma,J. Chem. Phys. 136, 184507 (2012).

29. J. D. Bernal, Proc. R. Soc. London Ser. A 280, 299–322 (1964).30. S. Asakura, F. Oosawa, J. Chem. Phys. 22, 1255 (1954).31. R. Roth, B. Götzelmann, S. Dietrich, Phys. Rev. Lett. 83,

448–451 (1999).

Acknowledgments: We thank F. Spaepen for helpful discussions.Supported by the Harvard Materials Research Science andEngineering Center through NSF grant DMR-0820484.

Supplementary Materialswww.sciencemag.org/content/343/6171/634/suppl/DC1Materials and MethodsFigs. S1 to S4References (32–58)Movies S1 to S4

19 August 2013; accepted 7 January 201410.1126/science.1244827

Rapid Soil Production and Weatheringin the Southern Alps, New ZealandIsaac J. Larsen,1*† Peter C. Almond,2 Andre Eger,2‡ John O. Stone,1David R. Montgomery,1 Brendon Malcolm2

Evaluating conflicting theories about the influence of mountains on carbon dioxide cycling andclimate requires understanding weathering fluxes from tectonically uplifting landscapes. Thelack of soil production and weathering rate measurements in Earth’s most rapidly upliftingmountains has made it difficult to determine whether weathering rates increase or decline inresponse to rapid erosion. Beryllium-10 concentrations in soils from the western Southern Alps,New Zealand, demonstrate that soil is produced from bedrock more rapidly than previouslyrecognized, at rates up to 2.5 millimeters per year. Weathering intensity data further indicatethat soil chemical denudation rates increase proportionally with erosion rates. These highweathering rates support the view that mountains play a key role in global-scale chemicalweathering and thus have potentially important implications for the global carbon cycle.

Plate tectonics has long been thought to in-fluence climate through links among rockuplift, relief generation, erosion, silicate

weathering, and CO2 cycling (1, 2). Determin-

ing the form of the functional relationship be-tween hillslope erosion and weathering rates iscritical for understanding how or if mountainsinfluence global weathering budgets and climate

Fig. 4. (A) Measurements of domain length l (black) and width w (gray) as a function of time for a dropletwith R = 18 mm (18). Images of domain morphologies at various times are shown at top (see also movieS1). The transition from isotropic to anisotropic growth occurs at about 100 min. (B) Distribution of l andw, both normalized to the radius of curvature R, for curved crystals. The dashed line marks the isotropicregime where w = l. w/R is roughly constant, in agreement with the model.

www.sciencemag.org SCIENCE VOL 343 7 FEBRUARY 2014 637

REPORTS

(3–6). Current models predict that soil weather-ing is supply-limited and dependent on the sup-ply of fresh minerals when erosion rates are low,but that weathering becomes increasingly kineti-cally limited as erosion rates increase, due to re-duced mineral residence times in the soil (3–5).However, there are few observational constraintson soil production and soil weathering rates inEarth’s most rapidly uplifting mountains, and itis unclear whether chemical weathering rates con-tinue to increase or decline as erosion rates in-crease beyond about 100 metric tons km−2 year−1

(~0.04 mm year−1) (6).We determined soil production and catchment-

scale denudation rates by measuring concentra-tions of in situ–produced 10Be in soil and sediment,respectively, and inferred chemical denudationrates using Zr mass balance (7) in the westernSouthern Alps of New Zealand (Fig. 1). The studysites are located east of the Alpine Fault, whererock uplift, exhumation, and erosion rates reach10 mm year−1 (8–11) and watershed-scale weath-ering fluxes are extremely high (12). Mean an-nual precipitation exceeds 10m year−1, supportingdense temperate montane rainforest and subalpineshrub ecosystems that grow on <1-m-thick soilsformed from highly fractured schist (13–15).

Soil production rates on ridgetops range from0.1 to 2.5 mm year−1 and decline exponentiallywith increasing soil thickness at two of the threesites with a sufficient number of samples (n ≥ 5)to define a regression relationship (Fig. 2A). Theform of these exponentially declining soil produc-tion functions is consistent with those determinedfor other landscapes (16). However, comparisonwith data compiled from sitesworldwide indicatesthat soil production rates in the western SouthernAlps reach levels more than an order of magni-tude greater than measured elsewhere (Fig. 2B),demonstrating that soil production can play a far

greater role in mountain denudation than pre-viously documented.

Chemical depletion fractions increase withsoil thickness (Fig. 2C) and decline with increas-ing soil production rates (Fig. 2D), which is con-sistent with declining soil residence times thatmodels predict at high denudation rates (3–5) andmeasurements from the San Gabriel Mountains(17). However, the high P values preclude con-cluding that the regression slopes are significant-ly different from zero. Thus, weathering may besupply-limited at the very high denudation rateswe observe. Regardless of whether the chemicaldenudation data conform to supply- or kinetical-ly limited weathering, the chemical denudationrates inferred from the Zr and 10Be data increaselinearly with physical erosion rates (Fig. 2E). Com-parison of our soil chemical denudation ratesagainst a compilation of worldwide data indi-cates that soil chemical denudation rates in thewestern Southern Alps are the highest valuesyet determined, and the power-law scaling ex-ponent of ~1.0 indicates that chemical denudationrates increase linearly as erosion rates increaseby nearly three orders of magnitude (Fig. 3A).The relationship between soil chemical denuda-tion and physical denudation rates observed inthe western Southern Alps is consistent with datafrom more slowly eroding landscapes (18) andextends the range of soil production rates overwhich weathering and denudation rates increasein proportion to one another to much highervalues than previously recognized. If the rela-tionship becomes nonlinear or reverses to an in-verse correlation, as predicted by several models(3–5), it does so at rates higher than we observein the western Southern Alps. Hence, our find-ings do not support the view that mountains areinefficient sites for weathering because of “speedlimits” to soil production and chemical denuda-tion (19); values from the western Southern Alpsbreak the proposed speed limits for both soilproduction and soil weathering.

10Be concentrations in sediment indicate thattotal denudation rates for watersheds in the west-ern Southern Alps range from 1 to 9 mm year−1

(Fig. 1). Small catchments that drain areas within1 to 2 km of the Alpine Fault have the lowestdenudation rates of 1 to 2 mm year−1. Largercatchments and subcatchments that drain areas

farther to the east of the Alpine Fault denude atrates of 4 to 12 mm year−1. The spatial patternof decreasing watershed-scale denudation towardthe Alpine Fault is consistent with crustal ve-locity fields predicted by simple shear deforma-tion (20) and vertical deformation measured byglobal positioning systems (21).

The maximum soil production rate that wemeasured in the western Southern Alps equals orexceeds denudation rates for the more slowlyeroding watersheds. However, for a given catch-ment, soil production rates are a fraction of thecatchment-averaged denudation rates. Hillslopesbelow the tree line in the western Southern Alpshave very little bedrock exposure, even thoughsoil production rates are lower than catchment-averaged denudation rates. The landscape main-tains a soil mantle because the return interval forlandslides, which must account for the balanceof catchment-averaged denudation not attributa-ble to soil production (10), is long enough at anypoint on the landscape that soils develop betweenfailures (22). The soil-mantled hillslopes and highsoil production rates in the western SouthernAlps indicate that, contrary to previous sugges-tions (23), hillslopes are not necessarily strippedof soil at high uplift rates. Although strong evi-dence for a transition from soil-mantled to bed-rock hillslopes exists for the SanGabrielMountains(22, 24), mean annual precipitation in the SanGabrielMountains is an order of magnitude lowerthan in the western Southern Alps. We suspectthat the emergence of bare bedrock hillslopes intectonically active landscapes is modulated inlarge part by climate, with soil production able toproceed at higher rates in wetter, and thus morevegetated, regions, such as our study area.

The high mean annual precipitation promotesrapid leaching (14) and supports high annualvegetation productivity (25), which we suggest,along with pervasively fractured bedrock, con-tribute to the high soil production rates that weobserve. The primarily biotic driver of soil pro-duction is vegetation, because burrowing mam-mals, which contribute to soil production in otherlandscapes (16), are not endemic to New Zealand.Coarse and fine roots readily penetrate bedrockby exploiting foliation planes and fractures gen-erated by tectonics (15) and erosional unloading(fig. S1). Root expansion in fractures, especially

1Department of Earth and Space Sciences and Quater-nary Research Center, University of Washington, Seattle,WA 98195–1310, USA. 2Department of Soil and Physi-cal Sciences, Lincoln University, Christchurch 7647 NewZealand.

*Corresponding author. E-mail: larseni@caltech.edu†Present address: Division of Geological and Planetary Sci-ences, California Institute of Technology, Pasadena, USA.‡Present address: Institute of Earth and EnvironmentalScience, University of Potsdam, Germany.

Fig. 1. Study sites. Location of the field area on New Zealand’s South Island (A), sample locations, and watershed scale denudation rates (T1 SE) (B).

7 FEBRUARY 2014 VOL 343 SCIENCE www.sciencemag.org638

REPORTS

foliation planes within a few cm of the soil-bedrock interface, likely plays a key role in con-verting rock to soil, both by physically breakingrock and by enhancing chemical weatheringthrough organic acid production and increasedsubsurface CO2 concentrations.

The high soil production rates in the westernSouthern Alps are consistent with estimated soilresidence times of only a few centuries (14) andrapid weathering and leaching (26) documentedat other sites on the South Island’s west coast.Soil production rates also increasewithwatershed-scale denudation rates, as observed in more slow-ly eroding landscapes (22). Soil production rateshave not been measured in other mountains un-dergoing uplift rates comparable to those in thewestern Southern Alps. However, in the Taiwanorogen, hillslopes are soil-mantled even thoughlandslides—which primarily mobilize soil—erodethe landscape at rates >2mmyear−1 (27). Similarly,hillslopes in the Tsangpo Gorge of the easternHimalaya are soil mantled, despite erosion rates>4 mm year−1 (28), suggesting that soil produc-

tion rates in other landscapes are also sufficientlyrapid that hillslopes are not stripped to bedrock,despite high rates of rock uplift and erosion.

Global river solute and sediment yield data(29) demonstrate that catchment-averaged chem-ical denudation rates span a similar range ofvalues as global soil chemical denudation rates,although the power-law scaling exponent less thanunity differs from the soils data (Fig. 3B). Riversolute data indicate that catchment-scale chemicaldenudation rates in the western Southern Alps areof the samemagnitude as soil chemical denudationrates. However, chemical denudation accounts for<5% of catchment-scale denudation (12, 30).Similarly, the mean soil chemical denudation ratesthat we determined are 1 to 7% of the 10Be-basedtotal denudation rates determined for each water-shed, whereas mean chemical denudation is 16 to32% of the soil production rate at each of theridgetop sites. The differing ratio of chemical tototal denudation rates observed for soils versuscatchments is likely due to a high proportion ofunweathered rock in the debris delivered to rivers

by landslides. Amore detailed geochemical budgetis required to assess weathering fluxes from deepbedrock (6) and landslide-affected terrain (31),but if soil dynamics at the sites we sampled areat least broadly representative of those occurringthroughout the landscape, our results suggest thatsoil weathering may be an important contributorto the high weathering flux in the western South-ern Alps (12, 30, 32).

The high denudation rates in the westernSouthern Alps are consistent with global fluxesdetermined using river load data, which demon-strate that young, wet mountains make up only14% of the ocean-draining land area but accountfor 62% of sediment, 38% of total dissolved solid,and 60% of dissolved silica delivered to Earth’soceans (29). Hillslopes in rapidly uplifting moun-tains appear to bypass the hypothesized weath-ering speed limit because high rates of orographicprecipitation support ecosystems that physicallybreak up rock and promote weathering andleaching, allowing rapid soil production to contrib-ute to the high orogen-scale weathering fluxes.

Fig. 2. Soil production rates, chemical depletion data, and chemicalversus physical denudation rates. (A) Soil production rate versus soil thick-ness for the western Southern Alps. Soil production rates decline exponentiallywith increasing soil thickness at the Gunn Ridge and Rapid Creek sites, and therespective exponential regression fits are y = 1.71(+0.71/–0.50)e–0.056(T0.012)x,R2 = 0.82, P = 0.005 and y = 3.18(+1.33/–0.94)e–0.054(T0.015)x, R2 = 0.83, P = 0.03.(B) Western Southern Alps soil production rates and soil production functions plottedwith a worldwide compilation of soil production functions; see supplementary ma-terials for data sources. (C) Chemical depletion fraction (CDF) versus soil thickness forthe western Southern Alps. CDF values increase with soil thickness as y =

0.0073(T0.0040)x – 0.052(T0.12), R2 = 0.39, P = 0.13 for Gunn Ridge and y =0.0077(T0.0023)x + 0.18(T0.057), R2 = 0.79, P = 0.04 for Rapid Creek. (D) CDFversus soil production rates for the western Southern Alps. CDF values decline assoil production rates increase, as y = –0.11(T0.070)ln(x)+0.036(T0.081), R2 =0.35, P = 0.16 for Gunn Ridge and y = –0.10(T0.062)ln(x)+0.35(T0.039), R2 =0.48, P = 0.20 for Rapid Creek. (E) Chemical denudation rate versus physicaldenudation rate for the western Southern Alps. Chemical denudation rates increaselinearly as physical denudation rates increase. The linear reducedmajor axis (RMA)regression fit shownby the line is y=0.49(T0.036)x–0.052(T0.017),R2=0.89,P<0.001. Error bars and parentheses indicate T1 SE; P values based on t statistics.

www.sciencemag.org SCIENCE VOL 343 7 FEBRUARY 2014 639

REPORTS

References and Notes1. T. C. Chamberlin, J. Geol. 7, 545–584 (1899).2. M. E. Raymo, W. F. Ruddiman,Nature 359, 117–122 (1992).3. K. L. Ferrier, J. W. Kirchner, Earth Planet. Sci. Lett. 272,

591–599 (2008).4. E. J. Gabet, S. M. Mudd, Geology 37, 151–154 (2009).5. G. E. Hilley, C. P. Chamberlain, S. Moon, S. Porder,

S. D. Willett, Earth Planet. Sci. Lett. 293, 191–199 (2010).6. A. J. West, Geology 40, 811–814 (2012).7. Materials and methods are available as supplementary

materials on Science Online.8. R. J. Norris, A. F. Cooper, J. Struct. Geol. 23, 507–520 (2001).9. J. M. Tippett, P. J. J. Kamp, J. Geophys. Res. 98 (B9),

16119–16148 (1993).10. N. Hovius, C. P. Stark, P. A. Allen,Geology 25, 231–234 (1997).11. D. M. Hicks et al., J. Hydrol. N.Z. 50, 81–142 (2011).12. A. D. Jacobson, J. D. Blum, Geology 31, 865–868 (2003).13. G. A. Griffiths, M. J. McSaveney, N. Z. J. Sci. 26, 197–209

(1983).14. L. R. Basher, P. J. Tonkin, M. J. McSaveney, Z. Geomorphol.

69, 117–131 (1988).15. B. A. Clarke, D. W. Burbank, J. Geophys. Res. 116 (F4),

F04009 (2011).16. A. M. Heimsath, W. E. Dietrich, K. Nishiizumi,

R. C. Finkel, Nature 388, 358–361 (1997).17. J. L. Dixon, A. S. Hartshorn, A. M. Heimsath, R. A. DiBiase,

K. X. Whipple, Earth Planet. Sci. Lett. 323–324, 40–49 (2012).18. C. S. Riebe, J. W. Kirchner, R. C. Finkel, Earth Planet. Sci.

Lett. 224, 547–562 (2004).19. J. L. Dixon, F. von Blanckenburg, C. R. Geosci. 344,

597–609 (2012).20. J. Braun, F. Herman, G. Batt, Earth Planet. Sci. Lett. 300,

197–204 (2010).21. J. Beavan et al., Geophys. Res. Lett. 37, L16305 (2010).22. A. M. Heimsath, R. A. DiBiase, K. X. Whipple, Nat. Geosci.

5, 210–214 (2012).23. K. X. Whipple, E. Kirby, S. H. Brocklehurst, Nature 401,

39–43 (1999).24. R. A. DiBiase, A. M. Heimsath, K. X. Whipple, Earth Surf.

Process. Landf. 37, 855–865 (2012).25. R. G. Hilton, P. Meunier, N. Hovius, P. J. Bellingham,

A. Galy, Earth Surf. Process. Landf. 36, 1670–1679 (2011).26. A. Eger, P. C. Almond, L. M. Condron, Geoderma 163,

185–196 (2011).27. Z.-X. Tsai, G. J.-Y. You, H.-Y. Lee, Y.-J. Chiu, Earth Surf.

Process. Landf. 38, 661–674 (2013).28. I. J. Larsen, D. R. Montgomery,Nat. Geosci.5, 468–473 (2012).29. J. D. Milliman, K. L. Farnsworth, River Discharge to the

Coastal Ocean: A Global Synthesis (Cambridge Univ.Press, New York, 2011).

30. W. B. Lyons, A. E. Carey, D. M. Hicks, C. A. Nezat,J. Geophys. Res. 110 (F1), F01008 (2005).

31. E. Gabet, Earth Planet. Sci. Lett. 264, 259–265 (2007).32. J. Moore, A. D. Jacobson, C. Holmden, D. Craw, Chem.

Geol. 341, 110–127 (2013).

Acknowledgments: We thank the NSF East Asia and Pacific SummerInstitutes program (OISE-1015454 to I.J.L.), Royal Society ofNew Zealand, NASA Earth and Space Science fellowship program,Geological Society of America, and University of WashingtonDepartment of Earth and Space Sciences for support; A. Heimsath,J. Roering, F. von Blanckenburg, and N. Hovius for stimulatingconversations; H. Greenberg for geographic information systemsupport; K. Larsen for assistance collecting sediment samples;R. Sletten for assistance importing samples; A. Heimsath andA. J. West for thorough reviews; and the New Zealand Department ofConservation for granting access to study sites. Data are availablein the supplementary materials. I.J.L. and P.C.A. designed and plannedthe field component of the study; I.J.L., A.E., and B.M. conductedthe field sampling; A.E. described soils; and I.J.L. and J.O.S. separated10Be. I.J.L. wrote the manuscript, with input from all coauthors.

Supplementary Materialswww.sciencemag.org/content/343/6171/637/suppl/DC1Materials and MethodsSupplementary TextFigs. S1 to S7Tables S1 to S7References (33–92)19 August 2013; accepted 30 December 2013Published online 16 January 2014;10.1126/science.1244908

Fig. 3. Chemical versus physical denudation rate for worldwide soil and river data. (A) TheRMA power-law fit to the soil mass flux data is y = 0.37(+0.12/–0.09)x1.025(T0.059), R2 = 0.38, P <0.001, indicating that chemical denudation increases linearly as physical denudation rates in-crease by nearly three orders of magnitude. (B) The RMA power-law fit to the river mass flux data(29) is y = 2.66(+0.39/–0.35)x0.63(T0.028), R2 = 0.41, P < 0.001. The hypothesized globalweathering speed limit (19) (horizontal dashed line) is shown for reference. Also shown are riverflux–based denudation data from the western Southern Alps (11, 12, 30); note that river and soilchemical denudation rates in the western Southern Alps are of similar magnitude. The upper andright axes with length units are related to the mass flux axes by a density of 2.65 g cm−3 andthus are approximate and for reference only, because rock density varies among the samples. Seesupplementary materials for data sources.

7 FEBRUARY 2014 VOL 343 SCIENCE www.sciencemag.org640

REPORTS

www.sciencemag.org/cgi/content/full/science.1244908/DC1

Supplementary Materials for

Rapid Soil Production and Weathering in the Western Alps, New Zealand

Isaac J. Larsen,* Peter C. Almond, Andre Eger, John O. Stone, David R. Montgomery, Brendon Malcolm

*To whom correspondence should be addressed. E-mail: larseni@caltech.edu

Published 16 January 2014 on Science Express

DOI: 10.1126/science.1244908

This PDF file includes:

Materials and Methods Supplementary Text Figs. S1 to S7 Tables S1 to S7 References

2

Materials and Methods Soil, bedrock, and sediment sampling We sampled soil from five ridges in the western Southern Alps (Fig. S2, S3). To avoid sampling material that experienced geochemical alteration during downslope transit (33, 34), soils were collected from the main ridge or from local, meter-scale convexities on smaller divides emanating from the main ridge, rather than downslope locations. The morphology and horizon development of each soil was described prior to sampling. We discarded organic-rich O-horizon materials and collected material generally from the entire soil column; from the top of the A-horizon down to the parent material. The depth to bedrock (mineral soil thickness) was measured at each site. Soil bulk density was estimated using the compliant cavity method (35). Local slope at each site was measured with a clinometer. Bedrock was sampled from the lowermost depth of each soil pit, which reached into fractured rock, as well as from the nearest outcrop to each pit. At one soil pit (Alex Knob Pit 4) we collected soil in 10 cm thick increments, which correspond closely to the soil horizons at the site (Fig. S4). We collected one sample from a bedrock outcrop in the Karangarua catchment. We also collected sand-sized river sediment from sandbars and channel margin deposits. Sample processing and analysis All samples were wet sieved to isolate the 250–850 μm grain-size fraction. The 250-850 μm grain-size fraction was treated with warm HCl for 24 h, followed by treatment with a combination of warm NaOH and H2O2 for 24 h; both treatments were then repeated. Selective dissolution in 2% HF was used to isolate quartz. Refractory heavy minerals were removed using lithium heteropolytungstate (LST) heavy liquid. A surfactant was used to aid removal of muscovite and feldspar. Samples were boiled in NaOH prior to the final HF etch. 9Be carrier was added to quartz aliquots prior to dissolution and Be separation at the University of Washington Cosmogenic Isotope Laboratory (36, 37). BeO was packed into cathodes with Nb powder and 10Be/9Be ratios were measured via accelerator mass spectrometry (AMS) at Lawrence Livermore National Laboratory. Splits of each bulk soil sample (generally ~1 kg) were crushed and pulverized. Zirconium (Zr) concentrations were measured on sub-splits of the pulverized material. Loss-on-ignition (LOI) was also measured for sub-splits and soil Zr concentrations were corrected for LOI. Surfaces of bedrock samples were cut or ground off to remove potentially weathered joint or foliation surfaces prior to Zr measurement. Bedrock density was measured following the removal of joint surfaces by weighing samples in air and in water. All Zr measurements were made via x-ray fluorescence on pressed powder samples at ALS Minerals, Vancouver, Canada. We generally calculated the chemical depletion fraction (CDF) (18, 38) using the bedrock collected from the base of each pit, rather than bedrock collected from outcrops, because preliminary measurements of the outcrop samples suggested Zr concentrations were spatially variable, which has been observed in other studies (39). However, the Gunn Pit 4 and Gunn Pit 6 rock samples from the base of the soil pits had higher Zr concentrations than the soil, so we used Zr

3

concentrations in bedrock from outcrops to calculate the CDF for those sites. Bedrock weathering beneath the soil would lead to the interpretation that the CDF are minimum values, but bedrock density values suggest relatively little mass loss due to incipient saprolite formation (Table S1). Soil oxide percentages were measured via x-ray fluorescence on fused samples by ALS Minerals, Vancouver, Canada. We used oxide data to calculate the Chemical Index of Alteration (CIA), as CIA = [Al2O3 / (Al2O3 + CaO + K2O + Na2O)] x 100; without corrections for apatite and carbonate content (40). We acknowledge that Zr is an imperfect tracer of chemical weathering, due in part to eolian deposition (17, 41, 42) and potential mobility (43, 44), given the high mean annual precipitation in the western Southern Alps (Fig. S5), but argue that it provides first-order constraints on chemical depletion; which is supported by the overall trend of increasing CDF values with increasing CIA values (Fig. S6). Interpretation of 10Be concentrations The 10Be concentration in quartz grains within a soil provides an estimate of the soil production rate (i.e., the total denudation rate, or the sum of physical and chemical denudation) at that site, provided the soil is vertically well-mixed (45, 46). We tested the assumption of well-mixed soil with the Alex Knob Pit 4 samples. The soil 10Be concentrations for different depth intervals indicate the soil is well-mixed (Fig. S4). Few studies have evaluated whether soils are vertically-well mixed, but our results are consistent with those expected from bioturbation, as observed in other landscapes (47-49). Interpretation of 10Be concentrations in soil and sediment as denudation rates requires the assumption of isotopic steady-state, such that the in-going and out-going 10Be flux from a soil profile or watershed is constant over time (45, 49-51). Landslides, which are common in the western Southern Alps (10), have the potential to upset the 10Be balance at both the soil profile and watershed scales. For example, sampling a surface immediately after a landslide removed 1 m of rock would cause erosion rates to be over-estimated by about a factor of three (52). Given the rapid denudation rates in the western Southern Alps, it is not feasible to use a second nuclide, such as 26Al, to test the isotopic steady-state assumption (51), so we carefully selected sampling sites in order to minimize the likelihood of violating the steady-state assumption. At the soil profile scale, landslides can expose bedrock shielded from cosmic rays, resulting in low 10Be concentrations that are out of equilibrium with long term soil production rates (51, 52). Measuring 10Be concentrations in soils formed on recent landslide scars would result in over-estimation of soil production rates. The time (t) required to return to isotopic equilibrium declines with increasing denudation rate (51) following:

t =Λ

ρ ⋅ D (eq. S1)

4

where Λ is the attenuation length for 10Be production below the surface (160 g cm-2), ρ is rock density, (2.65 g cm-3) and D is the denudation rate (cm yr-1). For example, for a soil production rate of 2 mm yr-1, isotopic equilibrium will be re-established 300 yr following a landslide. We carefully selected sampling sites to avoid areas with evidence of landsliding. We limited our sampling to convex ridgetops (Fig. S2, S3) where chronic, biogenic disturbance-driven geomorphic processes dominate (53) and avoided planar, threshold hillslopes where episodic landslides are the dominant erosion mechanism (28, 54). Previous work on spatial patterns of vegetation and soils in the western Southern Alps indicates ridges are the most stable portions of the landscape with respect to disturbance by landsliding (55, 56). While in the field we did not observe scars of any recent landslides that breached topographic divides and lowered ridgetop elevations, although we did observe recent landslide scars downslope from ridges. The landslides we did observe resulted in distinct topographic depressions, as even “shallow” landslides erode into bedrock, given the dm-scale soil depths in the western Southern Alps (57). We observed no similar topographic depressions on the ridgetops we sampled, indicating that the landslide return interval for ridgetops is likely to be much longer than those reported for the landscape as a whole. The mean time between landslides (return interval) for a given point on the landscape in the western Southern Alps has been estimated to be both 2,100–15,000 yr (25) and ~300 yr (15). Note that the different landslide return intervals are based on the same landslide mapping data (10), but the authors make different assumptions about the continuity (or dis-continuity) of landslide area-frequency distributions; the 2,100–15,000 yr estimate is based directly on the mapped landslide distribution. It is worth noting that both landslide return interval estimates (which likely overestimate landslide frequency on ridgetops) are sufficiently long that 10Be equilibrium can be re-established between landslides for the high soil production rates we measure. In the case where earthquakes on the Alpine Fault caused co-seismic shaking and ridgetop landsliding, as observed following the Chi-Chi earthquake in Taiwan (58), a soil production rate of 2 mm yr-1 would be sufficiently rapid to establish a new 10Be equilibrium in the time interval since the most recent (Mw>7.6; 1717 A.D.) earthquake (59). Moreover, none of our 10Be concentrations are consistent with coseismic landsliding during the most recent Alpine Fault rupture—the apparent exposure ages are too old (Table S2). If the soil 10Be data were alternatively interpreted as soil exposure ages that yield information on the time since the last landslide, a soil production rate can be determined by dividing the soil depth by the soil age. For this method to yield true soil production rates, it must be assumed that no erosion of soils formed in-situ on the landslide scar has occurred. Such an approach would yield lower soil production rates than those calculated by interpreting the 10Be data as steady-state soil production rates. However, given the steep slope gradients at most of our sampling sites (Table S3), the assumption of no erosion is implausible, hence interpreting soil 10Be concentrations as exposure ages would underestimate soil production rates. Tree throw is an episodic driver of soil production (60-62) that could also cause 10Be concentrations to be out of steady-state. We explored the potential influence of tree throw on soil production rates with a model similar to those developed to assess potential

5

errors in denudation rates caused by sampling bedrock surfaces following spallation of a slab of rock (52, 63). We do not attempt to simulate biotic-geomorphic interactions at our field sites, as has been done at other sites with more complete data on vegetation characteristics (62), but use the model results to assess whether the measured 10Be concentrations can reasonably be interpreted as denudation rates. Here we assume that tree throw removes rock from the soil-bedrock interface in a manner analogous to episodic erosion of an outcrop and track the 10Be concentration at the soil-bedrock interface. The model does not simulate mixing of detached material with overlying soil and we assume our results reflect maximum estimates of potential errors, as incorporation of rock transported by tree-throw into a well-mixed regolith would provide additional buffering that minimizes temporal variation of 10Be concentrations in soil. The model also assumes that all denudation is due to physical erosion, which also requires interpreting the error estimates as maximum values. The model uses a finite-difference approach with an annual time-step; the initial condition is a steady state erosion rate, after which erosion becomes unsteady, with periods of no erosion punctuated by periodic removal of a rock slab with a specified thickness. We modeled soil production rate scenarios of 1 mm yr-1 and a 2.5 mm yr-1, as we were most concerned with potential over-estimation of higher soil production rates. There are two 1 mm yr-1 cases, one in which 100 mm of rock is removed once per century and one in which 300 mm is removed once every 300 yr. There are also two 2.5 mm yr-1 cases, one in which 250 mm is eroded once per century and one in which 750 mm is removed once every three centuries. The modeling results show that, after reaching steady-state, the range of erosion rates interpreted from 10Be in the 1 mm yr-1 scenario is 0.93–1.09 mm yr-1 for the case in which all the erosion occurs once per century and 0.79–1.29 mm yr-1 for the case in which all the erosion occurs once every three centuries (Fig. S7). One of our measured soil production rates is 1.00±0.08 mm yr-1 and the true uncertainty is likely greater due to uncertainty in 10Be production rate scaling (64, 65). Hence uncertainty in soil production rates predicted by our 1 mm yr-1 un-steady erosion rate models is of comparable magnitude to analytical and production rate uncertainty. The range of erosion rates interpreted from 10Be in the 2.5 mm yr-1 scenarios are 2.05–3.11 mm yr-1 for the case in which all the erosion occurs once per century, which is very similar to the uncertainty in one of our measured values of the same magnitude (2.47±0.43 mm yr-1). The range of inferred erosion rates for the 2.5 mm yr-1 scenario in which all erosion occurs every 300 years is 1.43 to 4.94 mm yr-1 (Fig. S7). The magnitude of the erosion (750 mm) is unrealistically high for a single tree fall event, especially given that shrubs are the dominant vegetation at the ridgetops we sampled. Hence, the potential error modeled by this scenario is more representative of error caused by shallow landsliding, which we minimized with our sampling scheme, as described above. Additionally, whereas the soil production rates we measure in the western Southern Alps are substantially higher than those determined elsewhere using in situ-produced 10Be, they are consistent with values estimated using other methods. For example, meteoric 10Be has been used to infer a soil production function in with a y-intercept of 2.1 mm yr-1

6

for very weak shale bedrock (66). Similarly, soil on anthropogenically-produced bedrock exposures has been shown to form at rates of 5–10 mm yr-1 (given the volume change associated with rock-soil conversion, the equivalent bedrock lowering rates would be about half these values) on sandstone and shale bedrock (67). These two examples highlight the role weak lithology plays in driving rapid soil production rates, and we expect climate and biotic activity to play similar roles in driving high soil production rates in the western Southern Alps. At the catchment scale, landslides stochastically deliver sediment to channels. The 10Be concentration in landslide-derived sediment varies with landslide depth. Unless sediment is sufficiently well-mixed such that the 10Be concentration in sediment transported past the sediment sampling point is steady, the 10Be concentrations will not accurately represent the true denudation rate (68, 69). The catchment scale at which 10Be in sediment can be used reliably to infer the denudation rates of the upstream watershed has been modeled to be on the order of 70–100 km2 (68, 69). The large catchments we sampled that drain from the Main Divide to the Alpine Fault have areas of 340–450 km2, which are sufficiently large to expect steady 10Be flux. 10Be disequilibrium is expected to be more likely at small catchment areas, typically resulting in overestimation of denudation rates (68). Denudation rates for the two small (1.6 km2; 12.6 km2) catchments nested within larger catchments are within 1σ error or lower than denudation rates for the larger catchments, and hence do not exhibit evidence of isotopic disequilibrium. The lack of disequilibrium is likely due to the high rates of landsliding in the western Southern Alps, as rivers are always transporting sediment derived from a range of landslide depths. Additionally, though it is not evidence of erosional steady state, we do note that the catchment-scale denudation rates are generally consistent with erosion and exhumation rates averaged over a range of timescales (9-11, 70). CRONUS calculator inputs and denudation rate calculations We used the CRONUS calculator (71) to calculate denudation rates (Table S1, S2) from our 10Be concentrations. The theoretical framework that the CRONUS calculator uses to calculate denudation rates is identical whether the sample is bedrock collected from an outcrop, stream sediment, or vertically-mixed soil (45). To maintain consistency with other studies, we use the 10Be production rate calibration data encoded in the CRONUS calculator, which are from a wide-range of global sites, in determining denudation rates. More recent calibration efforts from the eastern Southern Alps of New Zealand (65) has resulted in a 10Be production rate estimate that is ~14% lower than those in the version of the CRONUS calculator we used (Wrapper script v.2.2; Main calculator v.2.1; Objective function v.2; Constants v.2.2.1; Muons v.1.1). Adopting the more proximal calibration data would result in roughly a 14% reduction in the denudation rates we report. Because 10Be production rate calibration schemes will continue to be improved upon, we report our results and the data required to reproduce our denudation rates in Tables S1-S7. The 1-standard error uncertainties in the denudation rates calculated using the CRONUS calculator include errors in the number of 9Be atoms added to each sample, errors in the 10Be concentration of procedural and carrier blanks, and errors in the AMS isotope ratio measurements, added in quadrature.

7

We calculated catchment-averaged 10Be production rate scaling factors by determining production rates for each grid cell in a DEM, using the elevation and latitude of each grid cell, as described by Greg Balco (http://depts.washington.edu/cosmolab/P_by_GIS.html). We used production rate scaling factors to calculate the elevation input for the CRONUS calculator, using the mean catchment latitude and Lal’s polynomial 10Be production rate scaling scheme (51). The catchment-averaged production rates we use assume zero 10Be production for the portions of each catchment with permanent snow and ice cover (11%, 13%, and 4% of the Karangarua, Whataroa, and Hokitika catchments, respectively), using 1:50,000 scale data from Land Information New Zealand (http://data.linz.govt.nz/layer/352-nz-mainland-snow-polygons-topo-150k/). The thickness of all soil and sediment samples was set to an arbitrarily low value of 0.1 cm, as the 10Be concentration at the surface is the relevant value for determining denudation rates in both vertically-mixed soils and sediment (45). The density values are mean values from multiple bedrock samples from each soil pit; catchment bedrock density values are the mean value of all soil pit samples within the watershed.

Topographic shielding of soil pits was calculated based on the hillslope angle of the sample site. The thick vegetation in the western Southern Alps generally blocked our view of the horizon, so we did not collect data for determining a shielding correction due to distant topography. Given our samples were from ridgetops, distant topography is expected to cause little shielding. We were able to climb on top of the single bedrock outcrop we sampled, which did afford a view of the horizon. The shielding factor determined for distant topography at this site is extremely small (0.999), indicating that by neglecting to account for shielding from distant topography, we are introducing very small errors that are orders of magnitude lower than the AMS measurement uncertainty. We calculated catchment-scale topographic shielding factors by first assuming that cosmic rays are conserved, such that all cosmic rays that enter a catchment produce 10Be within the catchment. We then assumed that all cosmic rays entering a catchment pass through a plane that projects from the catchment outlet to the maximum catchment elevation. The angle of the catchment surface plane was then used to determine catchment scale shielding factors in a manner analogous to determining the shielding factor for a sample on a sloping surface (72). The denudation rates we report in Tables S2 and S4 are based on time-dependent 10Be production and the Lal-Stone (51, 73) latitude and altitude scaling scheme.

Quartz is resistant to dissolution; hence the mean residence time for quartz in soils is longer than the mean residence time of all minerals (74). The enrichment of quartz biases denudation rate estimates (75). We corrected our soil denudation rate measurements for quartz enrichment by assuming Zr is similarly enriched in our samples (74, 75). The chemical erosion factors (CEF) are generally small, with an average of 1.06 (Table S3). For all corrections we assume a soil density of 1.0 g cm-1 (approximately the mean of the measured values (76)) and an attenuation length of 160 g cm-2. Chemical weathering

Soil and bedrock Zr concentrations were used to determine chemical depletion fractions (CDF), where:

8

CDF = 1−Zr[ ]rock

Zr[ ]soil

(eq. S2)

which is equivalent to the ratio of chemical denudation to total denudation (18, 38). Hence chemical denudation (W) can be determined:

W = D ⋅ 1−Zr[ ]rock

Zr[ ]soil

(eq. S3)

where D is the soil production or total denudation rate. Since:

D = W + E (eq. S4)

and E is the physical denudation rate, E can be determined from D and W. Errors in D were propagated in calculating W and E. Bedrock Zr was measured in a composite sample consisting of one or more pieces of bedrock from the base of each pit. Similarly, soil Zr was measured in a split of a large, homogenized sample that included rock fragments; hence we assume these data represent mean Zr concentrations. Because measurement errors in the Zr data are small (±2 ppm) relative to the concentrations we measured (Table S3), and are not expected to vary among samples, we did not propagate Zr measurement error in our calculations of W. World-wide soil production function, physical denudation, and chemical denudation data

The soil production functions in Fig. 2b are compiled from refs. 16, 22, 60, 77-82. The soil production functions are as originally published; for example, no corrections have been made for updates of 10Be production rate scaling. The soil physical and chemical denudation rate data in Fig. 3 are compiled from refs. 17, 18, 39, 83-86. The physical and chemical denudation data are as originally published, except some of those from ref. 18, which have been re-calculated using chemical erosion factors reported in ref. 87. The catchment-based physical and chemical denudation rate data are for 299 ocean-draining rivers from ref. 29. Pre-dam denudation data were used when these data were reported and data from five rivers in the database (Colorado, Haihe, Rhine, Patuxent, and Severn) were excluded from the statistical analyses due to high anthropogenic influence on either chemical or physical denudation; other rivers in the dataset have, of course, been impacted by humans and a full description of the data is available in ref. 29. Removal of five rivers resulted in an increase in the R2 value of the chemical-physical denudation regression relationship, but had very little influence on the power-law scaling exponent. The chemical denudation data from the western Southern Alps are from refs. 12 and 30, whereas the physical denudation data are updated suspended sediment discharge values from ref. 11. Data are shown only for rivers with measurements; unlike the original publications, no extrapolated values are shown.

9

Supplementary Text Study site descriptions We limited our sampling to sites near (or directly on) the ridgecrest because the ridge-tops lacked morphologic evidence for recent landsliding, and are hence more likely to be in steady-state with respect to 10Be and soil thickness, as described above. Topographic data are not available for our sites at a resolution suitable for calculating quantitative metrics of hillslope morphology, but the ridges are convex landforms with local-scale topographic heterogeneity. Qualitatively, the overall ridge curvature varied among the sites; the ridge at Rapid Creek site was sharply convex and the ridge-top was only a few m wide in some areas, in contrast to the broader tens of meters wide ridge-tops at the other sites, which may explain the higher soil production rates at the Rapid Creek site (88, 89). The ridges are not smooth over length scales greater than ~10 m, which differs from many sites where soil production rates have been measured previously and morphology is more consistent with the assumption of steady-state soil thickness (16). Some of the local-scale topographic variability may arise from tree-throw, though the shrubs that were the dominant vegetation at many of the sample locations are not likely to generate well-developed pit-and-mound topography; instead the local scale topographic variability is likely a function of the structural heterogeneity of the pervasively jointed and fractured bedrock. Additionally, ridge-top topography throughout the Southern Alps is influenced by gravitational collapse of bedrock, generating ridge rents or sackung (90), which we suggest also contributes to local-scale topographic variability. At the Karangarua site, we observed offset in bedrock consistent with gravitational collapse. Areas adjacent to the Karangarua and Gunn Ridge sites have been identified as areas that contain or are suspected to contain deep-seated gravitational failures (91) and the gravitational collapse may extend spatially beyond the mapped failures to the sites we sampled and contribute to the non-smooth character of the ridges. We did not observe evidence of gravitational failure at the Fox, Alex Knob, and Rapid Creek sites. Soil production rates have been previously measured at sites with locally variable topography and the potential for stochastic erosion process via careful site selection (22, 60). Because slopes were not smooth, we did not sample in straight, down-slope transects, but carefully selected sites to avoid areas with evidence for past disturbance or disequilibrium. The soil morphology in each pit was examined prior to sampling. Soils exhibiting evidence of truncation or burial (e.g., buried horizons, multiple hardpans) were not sampled and the soils we did collect had well-developed organic horizons and exhibited well-developed root systems. Our primary objective in sampling was to select soils that spanned a range of soil thicknesses and did not exhibit evidence of non-steady soil thickness. We observed few areas where soil thickness was <10 cm, save scattered bedrock outcrops standing high above the soil surface. We note that we measured 10Be concentration in quartz separated from soil, rather than underlying bedrock, which is a different approach than most previous soil production rate studies (16, 22, 60, 77-82), as this method does not require a cosmic ray shielding correction based on an assumption of steady-state soil thickness. Although the assumption of steady-state soil thickness is required to determine soil production functions (the regression relationship between soil production rate and soil thickness), the denudation rates we measure do not depend on

10

this assumption and are robust, providing that soils are vertically-mixed (45, 46). Whereas it is difficult, if not impossible to strictly test the assumption of steady soil thickness, the assumption of vertically-mixed soils can be tested with samples collected from multiple depth intervals, and data from Alex Knob Pit 4 indicate the soils we sampled are vertically mixed. Only the Karangarua, Gunn Ridge, and Rapid Creek sites have enough samples (n ≥ 5) to define soil production functions. The Gunn Ridge and Rapid Creek sites exhibit well-defined soil production functions, consistent with the interpretation that the soils we sampled are in steady-state with respect to soil thickness. However, the Karangarua site does not exhibit an exponential decline in soil production rate with increasing soil thickness, suggesting that the soil depths and 10Be concentrations rates for some of the soils we sampled may not be in equilibrium. The principle outlier from the Karangarua site (Pit 3) is a thin soil (15 cm) with an extremely low CDF of 0.01. In a landscape with episodically driven erosion, it might be expected that the thin, unweathered soil formed on bedrock exposed by a recent mass failure and would yield an unreasonably high denudation rate. However, the denudation rate is quite low and similar to the bedrock outcrop we sampled at this site, suggesting this area, which was the highest elevation sampled at the Karangarua site, may have been exposed bedrock or grassy, tussock vegetation that has more recently been colonized by shrub vegetation, given the lower bioturbation erosion efficiency of grassland vegetation (92).

Landsliding results in a landscape with surfaces of varying age and geochemical development. If the pattern of landsliding is heterogeneous in space, then a watershed may maintain a geochemical quasi-steady state, despite the episodic nature of landslide triggering. In the western Southern Alps, landslides are distributed relatively evenly between valley bottoms and ridges (58), suggesting landsliding is spatially heterogeneous, at least with respect to elevation.

11

Fig. S1. Photos of roots penetrating bedrock exposed by a recent tree fall. (A) Coarse (cm-scale) roots growing in a foliation plane have opened an 8−10 cm fracture and have broken off a ~10 cm thick slab of schist. (B) Fine (mm-scale) roots growing within planes of weakness within the schist contribute to chemical weathering and can physically spall mm-thick pieces of bedrock.

12

Fig. S2. Sample locations. (A) Hillshade map of the study area. The red boxes outline the bounds of panels C–G. (B) Location of the study area on New Zealand’s South Island. Locations of soil samples on ridgetops at the Karangarua (C), Fox (D), Alex Knob (E), Gunn Ridge (F), and Rapid Creek (G) study sites. Note that the map in panel A is the same as Figure 1, which contains denudation rate results.

13

Fig. S3. Photos of the ridges sampled in the (A) Karangarua, (B) Whataroa, and (C) Hokitika catchments. The arrows show the approximate location of the sample sites. Note that the dense vegetation is rooted in a near-continuous soil mantle.

14

Fig. S4. Alex Knob Pit 4 soil depth profile. (A) 10Be concentrations for samples from three depths. The upper, middle, and lower samples span the 0–10 cm, 10–20 cm, and 20–30 cm depths, respectively. The black curve is the predicted 10Be concentration for an unmixed soil, using a bulk density of 1.48 g cm-3, the mean for the mineral Bw and BC horizons. The data do not follow the trend predicted by no mixing, indicating the soil has been mixed vertically by bioturbation. Error bars indicate 1 standard error. (B) Soil production rates inferred from the 10Be concentrations. (C) Photo of the soil with designated soil horizons. The soil horizons correspond closely (within 1 cm) to the sample depth increments. Munsell color of horizons: A: 10YR 2/2; Bw: 10YR 3/2.5; BC: 2.5Y 4/2-3.

15

Fig. S5. Mean and standard deviation of monthly rainfall at the Cropp at Waterfall rain gage. The rain gage is in the Cropp River catchment, which is adjacent to the Rapid Creek catchment we sampled, at an elevation of 975 m. The mean annual precipitation at this site is 11.52 m. Data are from July 1982 to October 2012, courtesy of the New Zealand National Institute of Water and Atmospheric Research. The mean annual temperature is approximately 5.5 °C (ref. 14).

16

Fig. S6. Chemical Index of Alteration (CIA) versus CDF values for soils from the Western Southern Alps. The two, independent weathering indices generally increase with one another (although one of the Rapid Creek sites is an outlier), suggesting that Zr-based CDF values provide a first order view of the degree of chemical weathering at our study sites.

17

Fig. S7. Variability in inferred soil production rates due to unsteady erosion. The upper and lower sets of curves are for 2.5 mm yr-1 and 1 mm yr-1 soil production rate scenarios, respectively. For both scenarios, the thin, horizontal black line shows steady soil production, the thick black line shows soil production rates inferred from 10Be when all erosion occurs once per century, and the gray line shows soil production rates inferred from 10Be when all erosion occurs once every three centuries. The variability in inferred soil production rates caused by unsteady erosion is of the same (or lower) magnitude than analytical and 10Be production rate scaling uncertainty, except for the 2.5 mm yr-1 scenario where all erosion occurs once ever 300 yr. In this case, the 750 mm of erosion is unrealistically high for a single tree fall event, but comparable to what might be expected due to shallow landsliding. As explained in the text, we designed our sampling strategy to avoid sampling areas subject to recent landsliding.

18

Table S1. CRONUS calculator inputs. All samples assume a standard atmosphere and are normalized to the 07KNSTD 10Be standard. The input sample thickness is 0.1 cm for all samples, with the exception of the Karangarua bedrock outcrop sample, which had a thickness of 1.6 cm. Sample location coordinates are reported in Table S7. Sample name Lat. Lon. Elev.

(m) Density (g cm-3)

Shielding factor

10Be conc. (atom g-1)

10Be conc. uncertainty

(atom g-1) Karangarua-Pit-1 -43.6485 169.8466 1030 2.73 0.9889 28100 1300 Karangarua-Pit-2 -43.6493 169.8499 1082 2.60 0.9819 57000 2300 Karangarua-Pit-3 -43.6486 169.8521 1112 2.74 0.9564 73800 1600 Karangarua-Pit-4 -43.6480 169.8458 959 2.73 0.9797 20840 590 Karangarua-Pit-5 -43.6479 169.8458 961 2.72 0.9260 24800 1000 Karangarua-bedrock -43.6493 169.8500 1085 2.74 0.9988 66000 2500 Karangarua-sed -43.6500 169.9260 1004 2.70 0.9998 2220 180

Fox-Pit-1 -43.4943 169.9984 932 2.66 0.9874 35200 900 Fox-Pit-2 -43.4944 169.9986 942 2.65 0.9839 70600 1300 Fox_sed -43.4906 170.0040 731 2.66 0.9874 4800 430

Alex-Knob-Pit-2 -43.4168 170.1574 846 2.68 0.9600 48700 1100 Alex-Knob-Pit-3 -43.4199 170.1534 947 2.69 0.9633 43000 1500 Alex_Knob_Pit_4_0-10_cm -43.4169 170.1577 836 2.70 0.9874 49800 1300

Alex_Knob_Pit_4_10-20_cm -43.4169 170.1577 836 2.70 0.9874 55700 2300

Alex_Knob_Pit_4_20-30_cm -43.4169 170.1577 836 2.70 0.9874 57500 1100

Docherty_Creek_sed -43.4102 170.1380 562 2.69 0.9985 2840 180

Gunn-Pit-1 -43.4040 170.4046 866 2.73 0.9874 14440 680 Gunn_Pit_2 -43.4047 170.4050 832 2.66 0.9723 20080 830 Gunn-Pit-3 -43.4044 170.4048 856 2.57 0.9446 19530 770 Gunn_Pit_4 -43.4027 170.4027 953 2.70 0.9857 43600 1900 Gunn-Pit-5 -43.4034 170.4037 910 2.65 0.9874 30400 1500 Gunn-Pit-6 -43.4046 170.4048 838 2.65 0.9750 20300 800 Gunn-Pit-7 -43.4050 170.4102 555 2.67 0.9750 23900 900 Gunn-Ridge-sed -43.3975 170.4010 944 2.68 0.9819 870 200 Whataroa-sed -43.3722 170.4890 1017 2.68 0.9998 1260 330

Rapid-Creek-Pit-1 -43.0294 171.0175 966 2.97 0.9633 24900 1100 Rapid-Creek-Pit-2 -43.0282 171.0173 897 2.61 0.9260 11100 560 Rapid-Creek-Pit-3 -43.0267 171.0170 856 2.59 1.0000 8020 330 Rapid-Creek-Pit-4 -43.0289 171.0175 946 2.59 0.8916 3160 500 Rapid-Creek-Pit-5 -43.0273 171.0173 832 2.65 0.9819 6830 340 Rapid-Creek-sed -43.0308 170.9860 1070 2.68 0.9948 1640 150 Hokitika_sed -43.0828 171.0424 1124 2.68 0.9997 760 120

19

Table S2. Denudation rate and apparent exposure ages determined using the CRONUS calculator. Note that we do not use the soil denudation rate data presented here in our analyses, but the CEF-corrected values reported in Table S4. Note: 1.0 Mg (megagram) = 1.0 t (tonne) = 1000 kg. Sample name Denudation

rate (mm yr-1)

Denudation rate

uncertainty (mm yr-1)

Denudation rate

(Mg km-2 yr-1)

Denudation rate

uncertainty (Mg km-2 yr-1)

Apparent exposure

age (yr)

Apparent exposure

age uncertainty

(yr) Karangarua-Pit-1 0.29 0.02 790 60 2650 260 Karangarua-Pit-2 0.15 0.01 400 30 5150 480 Karangarua-Pit-3 0.11 0.01 300 20 6600 580 Karangarua-Pit-4 0.37 0.03 1010 70 2100 190 Karangarua-Pit-5 0.30 0.02 820 60 2630 250 Karangarua-bedrock 0.13 0.01 350 30 5890 550 Karangarua-sed 3.70 0.39 9990 1040 210 30

Fox-Pit-1 0.22 0.02 590 40 3600 320 Fox-Pit-2 0.11 0.01 290 20 7040 610 Fox_sed 1.44 0.16 3840 420 580 70

Alex-Knob-Pit-2 0.15 0.15 390 30 5440 480 Alex-Knob-Pit-3 0.18 0.18 480 40 4420 400 Alex_Knob_Pit_4_0-10_cm 0.14 0.14 390 30 5460 480 Alex_Knob_Pit_4_10-20_cm 0.13 0.13 350 30 6070 570 Alex_Knob_Pit_4_20-30_cm 0.12 0.12 340 20 6260 540 Docherty_Creek_sed 0.15 0.15 5900 520 390 40

Gunn-Pit-1 2.19 0.19 1380 110 1570 150 Gunn_Pit_2 0.51 0.04 960 70 2270 210 Gunn-Pit-3 0.36 0.03 980 70 2220 210 Gunn_Pit_4 0.38 0.03 480 40 4370 420 Gunn-Pit-5 0.18 0.01 670 60 3170 310 Gunn-Pit-6 0.25 0.02 960 70 2270 210 Gunn-Pit-7 0.36 0.03 680 50 3370 310 Gunn-Ridge-sed 0.25 0.02 24200 6100 90 20 Whataroa-sed 9.02 2.26 17700 5100 120 30

Rapid-Creek-Pit-1 0.28 0.02 830 70 2560 240 Rapid-Creek-Pit-2 0.67 0.05 1740 140 1260 120 Rapid-Creek-Pit-3 0.96 0.07 2490 190 870 80 Rapid-Creek-Pit-4 2.38 0.42 6170 1080 360 60 Rapid-Creek-Pit-5 1.07 0.09 2840 230 770 80 Rapid-Creek-sed 5.19 0.59 13900 1600 150 20 Hokitika_sed 11.65 2.04 31200 5500 70 10

20

Table S3. Soil loss-on-ignition (LOI), soil and bedrock Zr data, chemical depletion fractions (CDF), and chemical erosion factor (CEF) correction for quartz enrichment in soil (74, 75, 89), soil depth, and soil sampling site slope data. Note that the CEF used here is equivalent to the quartz dissolution factor (Cd) of ref. 75, where Zr enrichment in soil is used as a proxy for quartz enrichment. Saprolite weathering is assumed to be minimal in the CEF calculations (89). Sample name Soil

LOI (%)

Zrsoil (ppm)

LOIcorrected Zrsoil

(ppm)

Zrrock (ppm)

CDF CEF Soil thickness

(cm)

Local slope

(degrees) Karangarua-Pit-1 3.9 237 246 199 0.19 1.05 40 24 Karangarua-Pit-2 15.0 288 331 213 0.36 1.07 21 28 Karangarua-Pit-3 2.6 259 266 262 0.01 1.00 15 37 Karangarua-Pit-4 5.8 209 221 193 0.13 1.02 21 29 Karangarua-Pit-5 10.9 330 366 266 0.27 1.02 10 44

Fox-Pit-1 13.4 255 289 236 0.18 1.04 32 25 Fox-Pit-2 12.2 275 309 226 0.27 1.04 20 27

Alex-Knob-Pit-2 8.8 257 280 201 0.28 1.09 41 36 Alex-Knob-Pit-3 5.7 289 305 214 0.30 1.04 15 35 Alex_Knob_Pit_4_0-10_cm

8.5 224 243 181 0.26 1.06 31 29

Alex_Knob_Pit_4_10-20_cm

6.9 260 278 181 0.35 1.09 31 29

Alex_Knob_Pit_4_20-30_cm

5.5 215 227 181 0.20 1.04 31 29

Gunn-Pit-1 12.7 220 248 219 0.12 1.02 24 25 Gunn_Pit_2 7.6 238 256 223 0.13 1.02 25 32 Gunn-Pit-3 6.0 256 271 239 0.12 1.02 29 40 Gunn_Pit_4 8.2 262 284 206 0.27 1.08 39 26 Gunn-Pit-5 9.9 226 248 219 0.12 1.02 30 25 Gunn-Pit-6 6.0 243 258 199 0.23 1.05 27 31 Gunn-Pit-7 10.8 208 230 193 0.16 1.04 34 31

Rapid-Creek-Pit-1 5.1 208 219 107 0.51 1.23 40 35 Rapid-Creek-Pit-2 10.1 253 279 169 0.39 1.11 30 44 Rapid-Creek-Pit-3 7.1 233 249 170 0.32 1.04 16 0 Rapid-Creek-Pit-4 6.8 221 236 157 0.33 1.04 12 50 Rapid-Creek-Pit-5 25.4 213 267 207 0.23 1.03 15 28

21

Table S4. CEF-corrected soil denudation rate data. T = total denudation (soil production) rate, W = chemical denudation rate, and E = physical denudation rate. Sample name T

(mm yr-1) ±T

(mm yr-1) T

(Mg km-2 yr-1) ± T

(Mg km-2 yr-1) W

(mm yr-1) ± W

(mm yr-1) W

(Mg km-2 yr-1) ± W

(Mg km-2 yr-1) E

(mm yr-1) ± E

(mm yr-1) E

(Mg km-2 yr-1) ± E

(Mg km-2 yr-1)

Karangarua-Pit-1 0.30 0.02 830 70 0.056 0.005 150 10 0.25 0.02 680 70 Karangarua-Pit-2 0.16 0.01 430 30 0.054 0.005 140 10 0.11 0.01 280 40 Karangarua-Pit-3 0.11 0.01 310 20 0.0016 0.0001 4.4 0.3 0.11 0.01 300 20 Karangarua-Pit-4 0.38 0.03 1030 80 0.047 0.004 130 10 0.33 0.03 900 80 Karangarua-Pit-5 0.31 0.02 830 60 0.082 0.006 220 20 0.22 0.02 610 70 Fox-Pit-1 0.23 0.02 610 40 0.041 0.003 110 10 0.19 0.02 510 40 Fox-Pit-2 0.12 0.01 310 20 0.030 0.002 80 10 0.09 0.01 230 20 Alex-Knob-Pit-2 0.16 0.01 430 30 0.041 0.003 110 10 0.12 0.01 320 30 Alex-Knob-Pit-3 0.18 0.01 500 40 0.053 0.004 140 10 0.13 0.01 350 40 Alex_Knob_Pit_4_0-10_cm 0.15 0.01 410 30 0.037 0.003 100 10 0.12 0.01 310 30 Alex_Knob_Pit_4_10-20_cm 0.14 0.01 380 30 0.045 0.004 120 10 0.10 0.01 260 30 Alex_Knob_Pit_4_20-30_cm 0.13 0.01 350 20 0.025 0.002 70 10 0.11 0.01 280 30 Gunn-Pit-1 0.51 0.04 1410 110 0.059 0.005 160 10 0.46 0.04 1250 110 Gunn_Pit_2 0.37 0.03 980 80 0.047 0.004 120 10 0.32 0.03 860 80 Gunn-Pit-3 0.39 0.03 1000 80 0.046 0.004 120 10 0.34 0.03 890 80 Gunn_Pit_4 0.19 0.02 520 40 0.049 0.004 130 10 0.14 0.02 390 40 Gunn-Pit-5 0.26 0.02 690 60 0.030 0.003 80 10 0.23 0.02 610 60 Gunn-Pit-6 0.38 0.03 1000 80 0.082 0.006 220 20 0.29 0.03 780 80 Gunn-Pit-7 0.26 0.02 700 50 0.041 0.003 110 10 0.22 0.02 590 50 Rapid-Creek-Pit-1 0.34 0.03 1020 80 0.14 0.01 420 40 0.20 0.03 600 90 Rapid-Creek-Pit-2 0.74 0.06 1930 160 0.26 0.02 690 60 0.48 0.06 1250 170 Rapid-Creek-Pit-3 1.00 0.08 2600 200 0.31 0.02 790 60 0.70 0.08 1810 210 Rapid-Creek-Pit-4 2.47 0.43 6390 1110 0.80 0.14 2060 370 1.67 0.45 4330 1180 Rapid-Creek-Pit-5 1.10 0.09 2910 240 0.24 0.02 640 50 0.86 0.09 2270 240

22

Table S5. Blank information, sample size, Be carrier weights, isotope ratio, and quartz yield data. 9Be carrier concentration error=0.8%. na=not applicable. 10Be/9Be ratios for samples are corrected for blanks. Sample name Blank name Sample

mass (g quartz)

Be carrier

(μg)

10Be/9Be ratio

10Be/9Be ratio

uncertainty

Quartz yield from bulk

soil (%)

Quartz yield from 250-850 μm fraction (%)

Karangarua-Pit-1 Blank_ijl_15feb2012 31.5248 244.4 5.417E-14 2.43E-15 2.3 35.2 Karangarua-Pit-2 Blank-ijl31aug2012 20.3623 244.9 7.093E-14 2.85E-15 2.0 32.9 Karangarua-Pit-3 Blank_ijl_15feb2012 31.5192 244.9 1.421E-13 3.1E-15 2.5 56.1 Karangarua-Pit-4 Blank_ijl_15june2012 33.2208 244.8 4.232E-14 1.20E-15 3.2 38.2 Karangarua-Pit-5 Blank_ijl_15feb2012 32.3650 244.5 4.913E-14 1.97E-15 2.4 28.9 Karangarua-bedrock Blank_ijl_3aug2012 24.2542 245.8 9.749E-14 3.70E-15 na 37.1 Karangarua-sed Blank_ijl_15feb2012 37.4834 244.0 5.102E-15 4.13E-16 na 30.1

Fox-Pit-1 Blank-ijl31aug2012 25.9308 245.8 5.551E-14 1.38E-15 3.4 17.6 Fox-Pit-2 Blank-ijl31aug2012 26.8048 246.0 1.152E-13 2.2E-15 5.0 52.1 Fox_sed Blank_ijl_3aug2012 38.6035 246.2 1.128E-14 1.00E-15 na 20.7

Alex-Knob-Pit-2 Blank-ijl31aug2012 24.9813 245.2 7.423E-14 1.66E-15 2.4 39.7 Alex-Knob-Pit-3 Blank-ijl31aug2012 24.7428 246.0 6.471E-14 2.31E-15 2.6 32.3 Alex_Knob_Pit_4_0-10_cm

Blank_ijl_3aug2012 27.2564 246.2 8.254E-14 2.21E-15

1.4 23.4

Alex_Knob_Pit_4_10-20_cm

Blank_ijl_3aug2012 27.3397 246.1 9.254E-14 3.81E-15

1.6 22.8

Alex_Knob_Pit_4_20-30_cm

Blank_ijl_3aug2012 27.8965 245.9 9.765E-14 1.89E-15

1.3 33.0

Docherty_Creek_sed Blank_ijl_3aug2012 37.6864 245.9 6.514E-15 4.17E-16 na 20.5

23

Table S6. Blank information, sample size, Be carrier weights, isotope ratio, and quartz yield data. 9Be carrier concentration error=0.8%. na=not applicable. 10Be/9Be ratios for samples are corrected for blanks. Sample name Blank name Sample

mass (g quartz)

Be carrier

(μg)

10Be/9Be ratio

10Be/9Be ratio uncertainty

Quartz yield from bulk soil

(%)

Quartz yield from 250-850 μm fraction (%)

Gunn-Pit-1 Blank_ijl_15june2012 25.5671 244.4 2.260E-14 1.06E-15 1.4 25.4 Gunn_Pit_2 Blank_ijl_3aug2012 28.2016 245.5 3.452E-14 1.42E-15 1.2 32.7 Gunn-Pit-3 Blank_ijl_15june2012 25.8279 244.6 3.086E-14 1.21E-15 1.7 45.2 Gunn_Pit_4 Blank_ijl_15june2012 25.0069 243.8 6.689E-14 2.86E-15 1.6 34.5 Gunn-Pit-5 Blank-205-IL-JS 24.8587 243.6 4.640E-14 2.26E-15 1.8 26.5 Gunn-Pit-6 Blank-ijl31aug2012 24.7996 246.2 3.064E-14 1.14E-15 1.7 46.4 Gunn-Pit-7 Blank-ijl31aug2012 25.7644 245.2 3.757E-14 1.44E-15 1.2 23.5 Gunn-Ridge-sed Blank-205-IL-JS 33.5699 242.6 1.794E-15 4.04E-16 na 21.7 Whataroa-sed Blank_ijl_15feb2012 38.2650 242.8 2.966E-15 7.84E-16 na 25.4

Rapid-Creek-Pit-1 Blank_ijl_15june2012 32.3075 244.6 4.917E-14 2.13E-15 1.9 35.0 Rapid-Creek-Pit-2 Blank_ijl_15june2012 29.0639 244.8 1.972E-14 9.90E-16 2.7 14.0 Rapid-Creek-Pit-3 Blank_ijl_15feb2012 30.5438 245.1 1.496E-14 6.2E-16 2.2 23.5 Rapid-Creek-Pit-4 Blank_ijl_15june2012 31.0972 244.8 6.008E-15 9.40E-16 1.6 18.2 Rapid-Creek-Pit-5 Blank-ijl31aug2012 30.1100 245.4 1.254E-14 6.2E-16 2.1 18.4 Rapid-Creek-sed Blank_ijl_15feb2012 37.2632 244.7 3.739E-15 3.51E-16 na 20.4 Hokitika_sed Blank_ijl_3aug2012 37.8328 245.8 1.741E-15 2.78E-16 na 20.0

Blank_ijl_15feb2012 Blank_ijl_15feb2012 na 244.5 4.782E-16 1.183E-16 na na Blank_ijl_15june2012 Blank_ijl_15june2012 na 244.8 1.030E-15 6.22E-16 na na Blank_ijl_3aug2012 Blank_ijl_3aug2012 na 246.1 5.361E-16 1.169E-16 na na Blank-205-IL-JS Blank-205-IL-JS na 242.1 2.947E-16 2.214E-16 na na Blank-ijl31aug2012 Blank-ijl31aug2012 na 245.4 4.489E-16 2.964E-16 na na

24

Table S7. Sample locations. UTM zone 59S, WGS84 datum. Sample name Northing (m) Easting (m)

Karangarua-Pit-1 5166518 406980 Karangarua-Pit-2 5166429 407250 Karangarua-Pit-3 5166510 407429 Karangarua-Pit-4 5166579 406920 Karangarua-Pit-5 5166581 406921 Karangarua-bedrock 5166440 407258 Karangarua-sed 5174514 403756

Fox-Pit-1 5183804 419019 Fox-Pit-2 5183797 419036 Fox_sed 5185321 418733

Alex-Knob-Pit-2 5192552 431785 Alex-Knob-Pit-3 5192209 431464 Alex_Knob_Pit_4 5192543 431814 Docherty_Creek_sed 5196260 429580

Gunn-Pit-1 5194145 451786 Gunn_Pit_2 5194074 451823 Gunn-Pit-3 5194102 451809 Gunn_Pit_4 5194294 451636 Gunn-Pit-5 5194214 451719 Gunn-Pit-6 5194079 451809 Gunn-Pit-7 5194043 452241 Gunn-Ridge-sed 5194083 452561 Whataroa-sed 5206455 452430

Rapid-Creek-Pit-1 5235916 501425 Rapid-Creek-Pit-2 5236056 501407 Rapid-Creek-Pit-3 5236216 501384 Rapid-Creek-Pit-4 5235974 501423 Rapid-Creek-Pit-5 5236151 501407 Rapid-Creek-sed 5237500 500974 Hokitika_sed 5240528 499687

25

References 1. T. C. Chamberlin, An attempt to frame a working hypothesis of the cause of glacial periods on

an atmospheric basis. J. Geol. 7, 545–584 (1899). doi:10.1086/608449

2. M. E. Raymo, W. F. Ruddiman, Tectonic forcing of late Cenozoic climate. Nature 359, 117–122 (1992). doi:10.1038/359117a0

3. K. L. Ferrier, J. W. Kirchner, Effects of physical erosion on chemical denudation rates: A numerical modeling study of soil mantled hillslopes. Earth Planet. Sci. Lett. 272, 591–599 (2008). doi:10.1016/j.epsl.2008.05.024

4. E. J. Gabet, S. M. Mudd, A theoretical model coupling chemical weathering rates with denudation rates. Geology 37, 151–154 (2009). doi:10.1130/G25270A.1

5. G. E. Hilley, C. P. Chamberlain, S. Moon, S. Porder, S. D. Willett, Competition between erosion and reaction kinetics in controling silicate-weathering rates. Earth Planet. Sci. Lett. 293, 191–199 (2010). doi:10.1016/j.epsl.2010.01.008

6. A. J. West, Thickness of the chemical weathering zone and implications for erosional and climatic drivers of weathering and for carbon cycle feedbacks. Geology 40, 811–814 (2012). doi:10.1130/G33041.1

7. Materials and methods are available as supplementary materials on Science Online.

8. R. J. Norris, A. F. Cooper, Late Quaternary slip rates and slip partitioning on the Alpine Fault, New Zealand. J. Struct. Geol. 23, 507–520 (2001). doi:10.1016/S0191-8141(00)00122-X

9. J. M. Tippett, P. J. J. Kamp, Fission track analysis of the Late Cenozoic vertical kinematics of continental Pacific crust, South Island, New Zealand. J. Geophys. Res. 98, (B9), 16119–16148 (1993). doi:10.1029/92JB02115

10. N. Hovius, C. P. Stark, P. A. Allen, Sediment flux from a mountain belt derived by landslide mapping. Geology 25, 231–234 (1997). doi:10.1130/0091-7613(1997)025<0231:SFFAMB>2.3.CO;2

11. D. M. Hicks, U. Shankar, A. I. McKerchar, L. Basher, M. Jessen, I. Lynn, M. Page, Suspended sediment yields from New Zealand rivers. J. Hydrol. N.Z. 50, 81–142 (2011).

12. A. D. Jacobson, J. D. Blum, Relationship between mechanical erosion and atmospheric CO2 consumption in the New Zealand Southern Alps. Geology 31, 865–868 (2003). doi:10.1130/G19662.1

13. G. A. Griffiths, M. J. McSaveney, Distribution of mean annual precipitation across some steepland regions of New Zealand. N. Z. J. Sci. 26, 197–209 (1983).

14. L. R. Basher, P. J. Tonkin, M. J. McSaveney, Geomorphic history of a rapidly uplifting area in a compressional plate boundary: Cropp River, New Zealand. Z. Geomorphol. 69, 117–131 (1988).

15. B. A. Clarke, D. W. Burbank, Quantifying bedrock-fracture patterns within the shallow subsurface: Implications for rock mass strength, bedrock landslides, and erodibility. J. Geophys. Res. 116, (F4), F04009 (2011). doi:10.1029/2011JF001987

26

16. A. M. Heimsath, W. E. Dietrich, K. Nishiizumi, R. C. Finkel, The soil production function and landscape equilibrium. Nature 388, 358–361 (1997). doi:10.1038/41056

17. J. L. Dixon, A. S. Hartshorn, A. M. Heimsath, R. A. DiBiase, K. X. Whipple, Chemical weathering response to tectonic forcing: A soils perspective from the San Gabriel Mountains, California. Earth Planet. Sci. Lett. 323-324, 40–49 (2012). doi:10.1016/j.epsl.2012.01.010

18. C. S. Riebe, J. W. Kirchner, R. C. Finkel, Erosional and climatic effects on long-term chemical weathering rates in granitic landscapes spanning diverse climate regimes. Earth Planet. Sci. Lett. 224, 547–562 (2004). doi:10.1016/j.epsl.2004.05.019

19. J. L. Dixon, F. von Blanckenburg, Soils as pacemakers and limiters of global silicate weathering. C. R. Geosci. 344, 597–609 (2012). doi:10.1016/j.crte.2012.10.012

20. J. Braun, F. Herman, G. Batt, Kinematic strain localization. Earth Planet. Sci. Lett. 300, 197–204 (2010). doi:10.1016/j.epsl.2010.08.018

21. J. Beavan, P. Denys, M. Denham, B. Hager, T. Herring, P. Molnar, Distribution of present-day vertical deformation across the Southern Alps, New Zealand, from 10 years of GPS data. Geophys. Res. Lett. 37, L16305 (2010). doi:10.1029/2010GL044165

22. A. M. Heimsath, R. A. DiBiase, K. X. Whipple, Soil production limits and the transition to bedrock-dominated landscapes. Nat. Geosci. 5, 210–214 (2012). doi:10.1038/ngeo1380

23. K. X. Whipple, E. Kirby, S. H. Brocklehurst, Geomorphic limits to climate-induced increases in topographic relief. Nature 401, 39–43 (1999). doi:10.1038/43375

24. R. A. DiBiase, A. M. Heimsath, K. X. Whipple, Hillslope response to tectonic forcing in threshold landscapes. Earth Surf. Process. Landf. 37, 855–865 (2012). doi:10.1002/esp.3205

25. R. G. Hilton, P. Meunier, N. Hovius, P. J. Bellingham, A. Galy, Landslide impact on organic carbon cycling in a temperate montane forest. Earth Surf. Process. Landf. 36, 1670–1679 (2011). doi:10.1002/esp.2191

26. A. Eger, P. C. Almond, L. M. Condron, Pedogenesis, soil mass balance, phosphorus dynamics and vegetation communities across a Holocene soil chronosequence in a super-humid climate, South Westland, New Zealand. Geoderma 163, 185–196 (2011). doi:10.1016/j.geoderma.2011.04.007

27. Z.-X. Tsai, G. J.-Y. You, H.-Y. Lee, Y.-J. Chiu, Modeling the sediment yield from landslides in the Shihmen Reservoir watershed, Taiwan. Earth Surf. Process. Landf. 38, 661–674 (2013). doi:10.1002/esp.3309

28. I. J. Larsen, D. R. Montgomery, Landslide erosion coupled to tectonics and river incision. Nat. Geosci. 5, 468–473 (2012). doi:10.1038/ngeo1479

29. J. D. Milliman, K. L. Farnsworth, River Discharge to the Coastal Ocean: A Global Synthesis (Cambridge Univ. Press, New York, 2011).

30. W. B. Lyons, A. E. Carey, D. M. Hicks, C. A. Nezat, Chemical weathering in high-sediment-yielding watersheds, New Zealand. J. Geophys. Res. 110, (F1), F01008 (2005). doi:10.1029/2003JF000088

27

31. E. Gabet, A theoretical model coupling chemical weathering and physical erosion in landslide-dominated landscapes. Earth Planet. Sci. Lett. 264, 259–265 (2007). doi:10.1016/j.epsl.2007.09.028

32. J. Moore, A. D. Jacobson, C. Holmden, D. Craw, Tracking the relationship between mountain uplift, silicate weathering, and long-term CO2 consumption with Ca isotopes: Southern Alps, New Zealand. Chem. Geol. 341, 110–127 (2013). doi:10.1016/j.chemgeo.2013.01.005

33. E. G. Green, W. E. Dietrich, J. F. Banfield, Quantification of chemical weathering rates across an actively eroding hillslope. Earth Planet. Sci. Lett. 242, 155–169 (2006). doi:10.1016/j.epsl.2005.11.039

34. K. Yoo, R. Amundson, A. M. Heimsath, W. E. Dietrich, G. H. Brimhall, Integration of geochemical mass balance with sediment transport to calculate rates of soil chemical weathering and transport on hillslopes. J. Geophys. Res. 112, (F2), F02013 (2007). doi:10.1029/2005JF000402

35. R. Burt, Ed., Soil Survey Investigations Report No. 51, Version 1.0, Natural Resources Conservation Service, U.S. Department of Agriculture, 407 pp. (2009).

36. J. O. Stone, Extraction of Al and Be from quartz for isotopic analysis, <http://depts.washington.edu/cosmolab/chem/Al-26_Be-10.pdf> (2001).

37. R. Ditchburn, N. Whitehead, The separation of 10Be from silicates, in 3rd workshop of the South Pacific Environmental Radioactivity Association, pp. 4–7 (1994).

38. C. S. Riebe, J. W. Kirchner, D. E. Granger, R. C. Finkel, Strong tectonic and weak climatic control of long-term chemical weathering rates. Geology 29, 511–514 (2001). doi:10.1130/0091-7613(2001)029<0511:STAWCC>2.0.CO;2

39. K. L. Ferrier, J. W. Kirchner, R. C. Finkel, Weak influences of climate and mineral supply rates on chemical erosion rates: Measurements along two altitudinal transects in the Idaho Batholith. J. Geophys. Res. 117, (F2), F02026 (2012). doi:10.1029/2011JF002231

40. H. W. Nesbitt, G. M. Young, Early Proterozoic climates and plate motions inferred from major element chemistry of lutites. Nature 299, 715–717 (1982). doi:10.1038/299715a0

41. G. H. Brimhall, C. J. Lewis, J. J. Ague, W. E. Dietrich, J. Hampel, T. Teague, P. Rix, Metal enrichment in bauxites by deposition of chemically mature eolian dust. Nature 333, 819–824 (1988). doi:10.1038/333819a0

42. K. L. Ferrier, J. W. Kirchner, R. C. Finkel, Estimating millennial-scale rates of dust incorporation into eroding hillslope regolith using cosmogenic nuclides and immobile weathering tracers. J. Geophys. Res. 116, (F3), F03022 (2011). doi:10.1029/2011JF001991

43. A. C. Kurtz, L. A. Derry, O. A. Chadwick, M. J. Alfano, Refractory element mobility in volcanic soils. Geology 28, 683–686 (2000). doi:10.1130/0091-7613(2000)28<683:REMIVS>2.0.CO;2

44. M. E. Hodson, Experimental evidence for mobility of Zr and other trace elements in soils. Geochim. Cosmochim. Acta 66, 819–828 (2002). doi:10.1016/S0016-7037(01)00803-1

28

45. D. E. Granger, J. W. Kirchner, R. Finkel, Spatially averaged long-term erosion rates measured from in situ-produced cosmogenic nuclides in alluvial sediment. J. Geol. 104, 249–257 (1996). doi:10.1086/629823

46. D. E. Granger, C. S. Riebe, Cosmogenic nuclides in weathering and erosion, in Treatise on Geochemistry, vol. 5, J. Drever, Ed. (Elsevier, London, 2007).

47. M. C. Jungers, P. R. Bierman, A. Matmon, K. Nichols, J. Larsen, R. Finkel, Tracing hillslope sediment production and transport with in situ and meteoric 10Be. J. Geophys. Res. 114, (F4), F04020 (2009). doi:10.1029/2008JF001086

48. E. T. Brown, F. Colin, D. L. Bourlès, Quantitative evaluation of soil processes using in situ-produced cosmogenic nuclides. C. R. Geosci. 335, 1161–1171 (2003). doi:10.1016/j.crte.2003.10.004

49. E. T. Brown, R. F. Stallard, M. C. Larsen, G. M. Raisbeck, F. Yiou, Denudation rates determined from the accumulation of in situ-produced 10Be in the Luquillo Experimental Forest, Puerto Rico. Earth Planet. Sci. Lett. 129, 193–202 (1995). doi:10.1016/0012-821X(94)00249-X

50. P. Bierman, E. J. Steig, Estimating rates of denudation using cosmogenic isotope abundances in sediment. Earth Surf. Process. Landf. 21, 125–139 (1996). doi:10.1002/(SICI)1096-9837(199602)21:2<125::AID-ESP511>3.0.CO;2-8

51. D. Lal, Cosmic ray labeling of erosion surfaces: In situ nuclide production rates and erosion models. Earth Planet. Sci. Lett. 104, 424–439 (1991). doi:10.1016/0012-821X(91)90220-C

52. E. E. Small, R. S. Anderson, J. L. Repka, R. Finkel, Erosion rates of alpine bedrock summit surfaces deduced from in situ 10Be and 26Al. Earth Planet. Sci. Lett. 150, 413–425 (1997). doi:10.1016/S0012-821X(97)00092-7

53. J. J. Roering, J. W. Kirchner, W. E. Dietrich, Hillslope evolution by nonlinear, slope-dependent transport: Steady state morphology and equilibrium adjustment timescales. J. Geophys. Res. Solid Earth 106, (B8), 16499–16513 (2001). doi:10.1029/2001JB000323

54. D. W. Burbank, J. Leland, E. Fielding, R. S. Anderson, N. Brozovic, M. R. Reid, C. Duncan, Bedrock incision, rock uplift and threshold hillslopes in the northwestern Himalayas. Nature 379, 505–510 (1996). doi:10.1038/379505a0

55. A. Reif, R. B. Allen, Plant communities of the steepland conifer-broadleaved hardwood forests of central Westland, South Island, New Zealand. Phytocoenologia 16, 145–224 (1988).

56. J. Harrison, Soil distribution and landscape dynamics, Camp Creek, Westland, in Proceedings of the soil dynamics and land use seminar. I. B. Campbell, Ed. New Zealand Society of Soil Science, Lower Hutt, and New Zealand Soil Conservators Association, 65–77 (1985).

57. I. J. Larsen, D. R. Montgomery, O. Korup, Landslide erosion controlled by hillslope material. Nat. Geosci. 3, 247–251 (2010). doi:10.1038/ngeo776

29

58. P. Meunier, N. Hovius, J. A. Haines, Topographic site effects and the location of earthquake induced landslides. Earth Planet. Sci. Lett. 275, 221–232 (2008). doi:10.1016/j.epsl.2008.07.020

59. R. Sutherland, D. Eberhart-Phillips, R. A. Harris, T. Stern, J. Beavan, S. Ellis, S. Henrys, S. Cox, R. J. Norris, K. R. Berryman, J. Townend, S. Bannister, J. Pettinga, B. Leitner, L. Wallace, T. A. Little, A. F. Cooper, M. Yetton, M. Stirling, Do great earthquakes occur on the Alpine Fault in central South Island, New Zealand? Geophysical Monograph Series 175, 235–251 (2007). doi:10.1029/175GM12

60. A. M. Heimsath, W. E. Dietrich, K. Nishiizumi, R. Finkel, Stochastic processes of soil production and transport: Erosion rates, topographic variation and cosmogenic nuclides in the Oregon Coast Range. Earth Surf. Process. Landf. 26, 531–552 (2001). doi:10.1002/esp.209

61. J. J. Roering, J. Marshall, A. M. Booth, M. Mort, Q. Jin, Evidence for biotic controls on topography and soil production. Earth Planet. Sci. Lett. 298, 183–190 (2010). doi:10.1016/j.epsl.2010.07.040

62. E. J. Gabet, S. M. Mudd, Bedrock erosion by root fracture and tree throw: A coupled biogeomorphic model to explore the humped soil production function and the persistence of hillslope soils. J. Geophys. Res. 115, (F4), F04005 (2010). doi:10.1029/2009JF001526

63. L. J. Reinhardt, T. B. Hoey, T. T. Barrows, T. J. Dempster, P. Bishop, L. K. Fifield, Interpreting erosion rates from cosmogenic radionuclide concentrations measured in rapidly eroding terrain. Earth Surf. Process. Landf. 32, 390–406 (2007). doi:10.1002/esp.1415

64. G. Balco, J. Briner, R. C. Finkel, J. A. Rayburn, J. C. Ridge, J. M. Schaefer, Regional beryllium-10 production rate calibration for late-glacial northeastern North America. Quat. Geochronol. 4, 93–107 (2009). doi:10.1016/j.quageo.2008.09.001

65. A. E. Putnam, J. M. Schaefer, D. J. A. Barrell, M. Vandergoes, G. H. Denton, M. R. Kaplan, R. C. Finkel, R. Schwartz, B. M. Goehring, S. E. Kelley, In situ cosmogenic 10Be production-rate calibration from the Southern Alps, New Zealand. Quat. Geochronol. 5, 392–409 (2010). doi:10.1016/j.quageo.2009.12.001

66. J. A. McKean, W. E. Dietrich, R. C. Finkel, J. R. Southon, M. W. Caffee, Quantification of soil production and downslope creep rates from cosmogenic 10Be accumulations on a hillslope profile. Geology 21, 343–346 (1993). doi:10.1130/0091-7613(1993)021<0343:QOSPAD>2.3.CO;2

67. J. D. Phillips, A. V. Turkington, D. A. Marion, Weathering and vegetation effects in early stages of soil formation. Catena 72, 21–28 (2008). doi:10.1016/j.catena.2007.03.020

68. N. A. Niemi, M. Oskin, D. W. Burbank, A. M. Heimsath, E. J. Gabet, Effects of bedrock landslides on cosmogenically determined erosion rates. Earth Planet. Sci. Lett. 237, 480–498 (2005). doi:10.1016/j.epsl.2005.07.009

69. B. J. Yanites, G. E. Tucker, R. S. Anderson, Numerical and analytical models of cosmogenic radionuclide dynamics in landslide-dominated drainage basins. J. Geophys. Res. 114, (F1), F01007 (2009). doi:10.1029/2008JF001088

30

70. F. Herman, E. J. Rhodes, J. Braun, L. Heiniger, Uniform erosion rates and relief amplitude during glacial cycles in the Southern Alps of New Zealand, as revealed from OSL-thermochronology. Earth Planet. Sci. Lett. 297, 183–189 (2010). doi:10.1016/j.epsl.2010.06.019

71. G. Balco, J. O. Stone, N. A. Lifton, T. J. Dunai, A complete and easily accessible means of calculating surface exposure ages or erosion rates from 10Be and 26Al measurements. Quat. Geochronol. 3, 174–195 (2008). doi:10.1016/j.quageo.2007.12.001

72. S. A. Binnie, W. M. Phillips, M. A. Summerfield, L. K. Fifield, Sediment mixing and basin-wide cosmogenic nuclide analysis in rapidly eroding mountainous environments. Quat. Geochronol. 1, 4–14 (2006). doi:10.1016/j.quageo.2006.06.013

73. J. O. Stone, Air pressure and cosmogenic isotope production. J. Geophys. Res. Solid Earth 105, (B10), 23753–23759 (2000). doi:10.1029/2000JB900181

74. E. E. Small, R. S. Anderson, G. S. Hancock, Estimates of the rate of regolith production using 10Be and 26Al from an alpine hillslope. Geomorphology 27, 131–150 (1999). doi:10.1016/S0169-555X(98)00094-4

75. C. S. Riebe, J. W. Kirchner, D. E. Granger, Quantifying quartz enrichment and its consequences for cosmogenic measurements of erosion rates from alluvial sediment and regolith. Geomorphology 40, 15–19 (2001). doi:10.1016/S0169-555X(01)00031-9

76. I. J. Larsen, Hillslope erosion and weathering rates in Earth’s most rapidly uplifting mountains. Ph.D. dissertation, University of Washington, 128 p. (2013).

77. J. J. Owen, R. Amundson, W. E. Dietrich, K. Nishiizumi, B. Sutter, G. Chong, The sensitivity of hillslope bedrock erosion to precipitation. Earth Surf. Process. Landf. 36, 117–135 (2011). doi:10.1002/esp.2083

78. J. L. Dixon, A. M. Heimsath, J. Kaste, R. Amundson, Climate-driven processes of hillslope weathering. Geology 37, 975–978 (2009). doi:10.1130/G30045A.1

79. A. M. Heimsath, J. Chappell, W. E. Dietrich, K. Nishiizumi, R. C. Finkel, Late Quaternary erosion in southeastern Australia: A field example using cosmogenic nuclides. Quat. Int. 83, 169–185 (2001). doi:10.1016/S1040-6182(01)00038-6

80. A. M. Heimsath, J. Chappell, W. E. Dietrich, K. Nishiizumi, R. C. Finkel, Soil production on a retreating escarpment in southeastern Australia. Geology 28, 787–790 (2000). doi:10.1130/0091-7613(2000)28<787:SPOARE>2.0.CO;2

81. A. M. Heimsath, D. J. Furbish, W. E. Dietrich, The illusion of diffusion: Field evidence for depth-dependent sediment transport. Geology 33, 949–952 (2005). doi:10.1130/G21868.1

82. A. Heimsath, D. Fink, G. R. Hancock, The ‘humped’ soil production function: Eroding Arnhem Land, Australia. Earth Surf. Process. Landf. 34, 1674–1684 (2009). doi:10.1002/esp.1859

83. B. C. Burke, A. M. Heimsath, A. F. White, Coupling chemical weathering with soil production across soil-mantled landscapes. Earth Surf. Process. Landf. 32, 853–873 (2007). doi:10.1002/esp.1443

31

84. B. C. Burke, A. M. Heimsath, J. L. Dixon, J. Chappell, K. Yoo, Weathering the escarpment: Chemical and physical rates and processes, south-eastern Australia. Earth Surf. Process. Landf. 34, 768–785 (2009). doi:10.1002/esp.1764

85. K. P. Norton, F. von Blanckenburg, Silicate weathering of soil-mantled slopes in an active Alpine landscape. Geochim. Cosmochim. Acta 74, 5243–5258 (2010). doi:10.1016/j.gca.2010.06.019

86. J. L. Dixon, A. M. Heimsath, R. Amundson, The critical role of climate and saprolite weathering in landscape evolution. Earth Surf. Process. Landf. 34, 1507–1521 (2009). doi:10.1002/esp.1836

87. C. S. Riebe, D. E. Granger, Quantifying effects of deep and near-surface chemical erosion on cosmogenic nuclides in soils, saprolite, and sediment. Earth Surf. Process. Landf. 38, 523–533 (2013). doi:10.1002/esp.3339

88. J. J. Roering, How well can hillslope evolution models ‘explain’ topography? Simulating soil transport and production with high-resolution topographic data. Geol. Soc. Am. Bull. 120, 1248–1262 (2008). doi:10.1130/B26283.1

89. M. D. Hurst, S. M. Mudd, R. C. Walcott, M. Attal, K. Yoo, Using hilltop curvature to derive the spatial distribution of erosion rates. J. Geophys. Res. 117, (F2), F02017 (2012). doi:10.1029/2011JF002057

90. A. C. Beck, Gravity faulting as a mechanism of topographic adjustment. N.Z. J. Geol. Geophys. 11, 191–199 (1968). doi:10.1080/00288306.1968.10423684

91. O. Korup, Large landslides and their effect on sediment flux in South Westland, New Zealand. Earth Surf. Process. Landf. 30, 305–323 (2005). doi:10.1002/esp.1143

92. M. W. Hughes, P. C. Almond, J. J. Roering, Increased sediment transport via bioturbation at the last glacial-interglacial transition. Geology 37, 919–922 (2009). doi:10.1130/G30159A.1