Weathering pro®les, mass-balance analysis, and rates of solute loss: Linkages between weathering and erosion in a small, steep catchment

Suzanne Prestrud Anderson* Center for Study of Imaging and Dynamics of the Earth, University of California, Santa Cruz, California 95064-1077, USA William E. Dietrich George H Brimhall Jr. Department of Earth and Planetary Sciences, University of California, Berkeley, California 94720-4767, USA

ABSTRACT where uplift ratesÐand, hence, physical- and transport-limited landscapes (Carson and denudation ratesÐare great enough to lead Kirkby, 1972). In the former, the rate of for- In a headwater catchment in the Oregon to a bedrock-dominated landscape. Chem- mation of erodible debris by physical and Coast Range, we ®nd that solid-phase mass ical denudation rates will increase with chemical processes controls the rate of land- losses due to chemical weathering are physical-denudation rates, but only as long scape lowering. Landscapes dominated by equivalent in the bedrock and the soil. as the landscape remains mantled by soil. bare bedrock are universally recognized as However, the long-term rate of mass loss weathering-limited. In contrast, landscapes per unit volume of parent rock is greater Keywords: chemical erosion, denudation, with deep regolith mantles are transport- in the soil than in the rock. We attribute physical weathering, soil dynamics, uplift, limited. The rate of landscape lowering is lim- this ®nding to the effects of biotic processes weathering. ited by the ef®ciency of transport processes, in the soil and to hydrologic conditions that and the weathered pro®le thickens through maximize contact time and water ¯ux INTRODUCTION time. Stallard (1985) demonstrated that these through the mineral matrix in the soil. This concepts could be used to explain the evolu- result stems both from earlier work in The weathered pro®le, consisting of layers tion of chemical loads of large rivers. which we demonstrated that rock and soil of weathered rock topped by soil, develops in Recent empirical advances (Heimsath et al., contribute equally to the solute ¯ux and response to chemical, physical, and biological 1997, 1999, 2000, 2001a) support the concept from arguments presented here that the ba- processes at the Earth's surface. The strong that regolith-production rates decline under in- sin is in dynamic equilibrium with respect interplay among these processes makes it dif- creasing soil cover. In an alternative model, to erosion and uplift. The silica ¯ux of 10.7 ®cult to disentangle their interactions in the regolith production reaches a maximum under t´km؊2´yr؊1 from the basin is several weathered pro®le. Physical weathering pro- a particular soil depth and is reduced under 7.1 ؎ times larger than the ¯ux from older soils cesses expose fresh rock and mineral surfaces both shallower and deeper soil cover (Ahnert, elsewhere, but comparable to the ¯ux from to chemical weathering, whereas chemical 1967; Carson and Kirkby, 1972; Stallard, sites with similar physical erosion rates. weathering reduces the strength of rock, mak- 1985; Anderson and Humphrey, 1989; Rosen- This result argues that physical denudation ing it more susceptible to physical breakdown. bloom and Anderson, 1994; Small et al., or uplift rates play an important role in set- The interactions between these processes are 1999). Mechanistic justi®cation for either type ting the chemical denudation rate. Physical widely recognized as critical in understanding of production rule is usually couched in terms processes appear to in¯uence chemical- the effects of changes in climate and of tec- of the effects of increasing soil depth on weathering rates in several ways. First, they tonic uplift rates on erosion (Raymo et al., chemical and/or thermomechanical (Ander- limit chemical evolution by removing ma- 1988), landscape evolution (Anderson and son, 1999) processes. terial, thus setting the residence time within Humphrey, 1989), and geochemical cycling Understanding the interplay between phys- the weathered rock and the soil. Second, (Gaillardet et al., 1999a). ical and chemical weathering processes and bioturbation mixes rock fragments into the The balance between removal of debris by their relationship to erosion rates has become more reactive soil and maintains high soil transport processes and the breakdown of rock increasingly important in understanding geo- porosity, allowing free circulation of water. into movable material by weathering exerts a chemical cycles (Stallard, 1992, 1995b). The Because the weathering in the soil is more strong control on landscape evolution. G.K. potential feedbacks between erosion and intense than in the rock, we argue that the Gilbert (1877) described the linkage between chemical weathering were brought to promi- chemical denudation rate will diminish regolith production and erosion rates, now nence by the erosion-driven climate change codi®ed by geomorphologists who differenti- hypothesis (Raymo et al., 1988; Raymo and *E-mail: [email protected]. ate conceptually between weathering-limited Ruddiman, 1992). Given that global silicate-

GSA Bulletin; September 2002; v. 114; no. 9; p. 1143±1158; 9 ®gures; 3 tables.

For permission to copy, contact [email protected] ᭧ 2002 Geological Society of America 1143 ANDERSON et al.

Coast Range drives suf®cient erosion to keep the soils relatively thin, but is not so high as to produce large tracts without appreciable soil development. The rainfall rate is high enough to produce measurable weathering in the rock and soil during their relatively brief transit through the weathered zone. Our exploration of weathered-pro®le devel- opment takes several tacks. First, we describe the weathered pro®le and its spatial variation. A chemical mass-balance technique (Brimhall and Dietrich, 1987) allows quanti®cation of the mass gains and losses of individual chem- ical elements relative to fresh parent rock. Second, we compare the alteration of the rock with measured solute ¯uxes from the catch- ment, permitting comparison of present-day weathering processes with time-integrated chemical losses from the bedrock. We can then quantify the rate of mass loss due to weathering in the bedrock and in the soil. To- gether, these analyses show that physical pro- cesses open the bedrock up to hydrologic pro- cesses and drive the production of soil. Chemical processes attack the rock vigorously once it is in contact with air, rainwater, and the biosphere, but this activity must be viewed Figure 1. Photograph of the CB1 and CB2 catchments and surrounding hillslopes. as a consequence of the environment produced View is to the southwest, oblique to the north-facing CB1 catchment. CB1 catchment by physical processes. is 50 m long and ϳ20 m wide near the ridge crest. CB2 is 95 m long and ϳ45 m wide. Soil pits were located in CB2 to avoid disturbance to CB1. Star shows location of deep Setting core (Fig. 2). The study site is the 860 m2 CB1 catch- ment, located in the Oregon Coast Range near weathering rates exert an important control on et al., 1997a; Anderson and Dietrich, 2001) Coos Bay, Oregon (Fig. 1). Topography of the the long-term C dioxide content of the atmo- and catchment hydrology (Montgomery et al., Oregon Coast Range is marked by steep hill- sphere (Holland, 1978; Berner, 1990), if sili- 1997; Anderson et al., 1997b; Torres et al., slopes bounded by bedrock and gravel chan- cate chemical weathering is linked to physical 1998). Previously, we have shown that half of nels. Alluvial ®lls are con®ned to narrow val- denudation and tectonic uplift (Ruddiman and solutes exported from the catchment are pro- leys along the coast inundated by late Prell, 1997), then weathering and erosion may duced in the soil and half in the bedrock. Holocene sea-level rise (Personius, 1993; Per- drive climate change. A number of workers Here, we quantify solid-phase mass gains and sonius et al., 1993). Uplift of rock beneath the have stressed the importance of physical ero- losses in the weathered pro®le, and we deter- Coast Range (Adams, 1984; Kelsey et al., sion in enhancing chemical-weathering rates mine the time for weathered-pro®le develop- 1994) continually generates steep topographic (Stallard, 1992, 1995a; Gaillardet et al., ment. This approach allows us to compare the gradients to drive incision of streams and ero- 1999a, 1999b; Riebe et al., 2001), whereas long-term rate of mass loss in the bedrock and sion of hillslopes. The record preserved in ma- White and Blum (1995) found little correla- soil. rine terraces yields rock-uplift rates of 0.03± tion between chemical weathering and physi- The ®eld site does not represent extremes 0.23 mm´yrϪ1 over the past 80±125 ka along cal erosion rates. The relationship between of either physical or chemical weathering con- the southern Oregon coast (Kelsey et al., physical processes and chemical weathering ditions. Instead, one expects both relatively 1994, 1996), probably ϳ0.1 mm´yrϪ1 in the has not been elucidated at a mechanistic level, vigorous physical processes (owing to the vicinity of Coos Bay (H. Kelsey, 1994, per- but this mission is most likely to be accom- site's steep slopes and active tectonics) and sonal commun.). plished in a small catchment, where there is active chemical weathering (driven by the Soils are thin in the Coast Range, but soil more control on rock type, vegetation, and lo- moderately heavy rainfall). Biological in¯u- depth varies considerably. Erosion rates over cal climate. ences on weathering processes are both phys- the southern Oregon Coast Range have been We present here a detailed study of the ical and chemical in nature and are well ex- 0.05±0.08 mm´yrϪ1 for the past 4±15 ka (Re- weathered pro®le in a headwater catchment in pressed in the catchment. Climate variation neau and Dietrich, 1991; Reneau et al., 1989), the Oregon Coast Range (Fig. 1) and of the since the Pleistocene has been relatively mod- roughly equivalent to the long-term uplift processes that have shaped it. We use chemi- est (Worona and Whitlock, 1995); hence, cur- rates. Reneau and Dietrich (1991) found that cal analyses of soil and bedrock as well as rent processes are similar to those that shaped bedrock-lowering rates (also called ``exhu- prior work on soil water chemistry (Anderson the landscape. The uplift rate in the Oregon mation rates'' by England and Molnar, 1990),

1144 Geological Society of America Bulletin, September 2002 LINKAGES BETWEEN WEATHERING AND EROSION IN A SMALL, STEEP CATCHMENT determined from colluvium in-®lling rates of catchment divide, and the sandstone beds dip (38 mm´dϪ1) was applied over the entire catch- unchanneled valleys in the Oregon Coast no more than 8Њ±15ЊS into the slope. This core ment for one week. The intensity was selected Range, were equal to sediment yield in is therefore representative of the bedrock un- to typify storms in the region; it is equivalent streams. The equivalence of denudation rates derlying most of CB1 and CB2. Eleven shallow to a 24 h event with Ͻ1 yr recurrence interval across spatial scales from 1 to 1500 km2 sug- bedrock borings within the catchment were (Montgomery et al., 1997). Analysis of daily gests that there are no changes in sediment made with a portable diamond-bit corer (Mac- rainfall for two years shows that daily rainfall storage in the system. Recent catchment- Donald, 1988); these cores generally had poor totals Ն38 mm occur in ϳ10% of annual rain averaged regolith-production rates of 0.117 recovery, but pieces up to 0.5 m long helped days (National Climate Data Center [http:// mm´yrϪ1 (Heimsath et al., 2001b) also match characterize bedrock at points within the catch- lwf.ncdc.noaa.gov/oa/ncdc.html] records for uplift and denudation rates. Heimsath et al. ment. About 100 holes were hand-augered in North Bend Airport). In 1983 (annual rainfall found considerable variability in regolith- the catchment to delineate the soil and saprolite 150% of normal), days with Ն38 mm rain ac- production rates over a small spatial scale thickness. Examination of exposures of the soil/ counted for 26% of the annual total, while in (meters to tens of meters), however, which bedrock interface in landslide scars also con- 1992 (annual rainfall 92% of normal), days they attributed to both the stochastic nature of tributed to our understanding of the spatial dif- with Ն38 mm rain accounted for 16% of the the biogenic processes that form the regolith ferences in weathered-pro®le development. annual total. Experiment 3 was long enough and the effects of local competition between Twenty samples were taken from the 35 m that quasi-steady ¯ow conditions were reached drainage networks and episodic erosional core to characterize the unweathered parent in the soil and outlet channel (Anderson et al., events. At scales greater than these variations, material and the weathered rock. Fourteen soil 1997b; Torres et al., 1998). however, the landscape appears to be in dy- samples were collected with a piston corer in Soil water samples were ®eld ®ltered (0.45 namic equilibrium, and uplift rates are equal two soil pits in the CB2 catchment (Fig. 1). ␮m) and analyzed for cations and silica by to denudation rates. CB1 and CB2 soils are comparable, as they inductively coupled plasma±mass spectrome- are derived from the same bedrock and are at try and for anions by ion chromatography (see Catchment Description a similar stage in the hollow-®lling cycle Anderson and Dietrich, 2001, for details). Al- (Dietrich et al., 1986). kalinity was measured by titration with 0.1N The CB1 catchment is an unchanneled val- Rock bulk densities were determined by HCl by using the Gran method. ``Organic an- ley, a basin with convergent bedrock topog- water displacement of wax-coated samples ions'' were estimated from the difference be- raphy that forms the contributing area to a (average size, 72 cm3). Soil bulk densities tween cation and anion concentrations (Dahl- ®rst-order stream (Fig. 1). The outlet of the were determined for oven-dried piston core gren and Ugolini, 1989). CB1 catchment is the head of an ephemeral samples. Grain-density determinations were ®rst-order stream. The steep (45Њ) slopes of made with a pycnometer on milled rock or RESULTS the catchment are thinly mantled with soil, disaggregated soil. Rock and soil samples which supported a Douglas ®r (Pseudotsuga were prepared for chemical analysis by pro- The Weathered Pro®le menzeisii) forest until the area was logged in ducing pulps with a disc mill. Determinations 1987. It was replanted in 1988. The catchment of Al, Ti, Mg, Ca, Na, K, and Fe were done We divided the weathered pro®le into four is underlain by the Eocene Flournoy Forma- on La2O3 solutions of HF-digested samples layers based on differences in physical prop- tion, a rhythmically bedded graywacke sand- with a Perkin-Elmer 3030 ¯ame atomic ad- erties, fractures, and degree of weathering stone containing quartz, feldspar, and lithic sorption spectrometer. Determinations of Si (Fig. 2). There is great contrast, chie¯y in bulk (volcanic) fragments in a matrix of clay min- and Zr were done with XRF (X-ray ¯uores- density (Fig. 3) and organic C content, be- erals (Baldwin, 1974, 1975). cence) spectrometry (Spectrace 440) on fused tween soil and bedrock. Saprolite is present The CB1 catchment and the adjacent, and pressed pellets, respectively. C and N spottily in the catchment. We identi®ed two slightly larger CB2 catchment were monitored were analyzed with a Perkin-Elmer Model layers within the weathered bedrock below the hydrologically from December 1989 to Feb- PE2400 CHNS/O analyzer. saprolite. The predominant lithology in the ruary 1992 (CB2) or November 1996 (CB1). Thin sections of six samples from the core unweathered rock is ®ne-grained graywacke. Monitoring ended in both cases because of and two samples from the soil pits were point- Minor calcite-cemented graywacke was asso- landslides. Sprinkling experiments were counted to document the mineral assemblage. ciated with faults and fractures. mounted in CB1 to study the hydrology and Quartz was distinguished from feldspar on the The soil in the CB1 catchment is classi®ed water chemistry in the runoff (Anderson et al., basis of twinning, which may have resulted in as Haplumbrept by the Soil Conservation Sur- 1997a, 1997b; Montgomery et al., 1997; Tor- an undercounting of feldspar grains (our vey (Haagen, 1989), re¯ecting its limited ped- res et al., 1998). quartz counts are higher than those that Lovell ogenic development. The soil is an extremely (1969) and Dott (1966) obtained in formations porous (average bulk density of 740 Ϯ 190 METHODS similar to the Flournoy). kg´mϪ3) sandy loam. It is also organic rich; C Soil waters were sampled with 34 suction content averages 2.5 wt%, and reaches as high Spatial variations in the weathered pro®le lysimeters and two zero-tension lysimeters as 10 wt% in the surface litter (Anderson and were documented by coring, hand augering, during a February 1992 winter storm with a Dietrich, 2001). The soil surface is hummocky and digging soil pits. A 35-m-long, 65-mm- three-day total rainfall of 144 mm, and during owing to the presence of tree-throw mounds diameter core was obtained at the top of the a catchment-scale sprinkling experiment (ex- and the spoils and tunnels of burrowing ani- CB1 catchment with a truck-mounted drill rig. periment 3) in May±June 1992. Peak runoff mals. The boundary between soil and bedrock Core recovery averaged 93% below the upper during the storm was 10 times greater than is abrupt. A thin layer of saprolite underlies 3 m, which were air blasted without recovery. during the sprinkling experiment. For experi- the soil in some areas in the catchment. The The channel head is located ϳ42 m below the ment 3, a nearly steady ``rain'' of 1.6 mm´hϪ1 saprolite is orange to tan in color and lacks

Geological Society of America Bulletin, September 2002 1145 ANDERSON et al. organic material. The fact that it is penetrable with a hand auger or shovel implies that weathering has substantially weakened it. The bulk densities of the two samples we collected were 1850 and 1150 kg´mϪ3. The weathered bedrock below the saprolite can be divided into two layers. Bulk density is not a discriminator; it varied by only 60 kg´mϪ3 around an average of 2270 kg´mϪ3 throughout the core (Fig. 3). We use oxidized color and fracture density to subdivide the weathered rock into pervasively oxidized and fractured, partially oxidized rock layers. Open fractures, often stained with Fe or Mn oxides, increase in frequency toward the surface, where horizontal fractures occurred in addi- tion to high-angle fractures (Fig. 2). Veins de- crease in frequency toward the surface and were equally likely to have calcite or silica ®llings or linings. Most of the veins and frac- tures were steeply inclined (45Њ±80Њ from hor- izontal) and are probably tectonic in origin. Horizontal fractures were observed in the top 3 m of the core. The pervasively oxidized rock layer extends to a depth of 4.5 m below the surface, which is approximately the upper extent of pyrite preservation. Pyrite, the most reactive mineral in the parent material, is a good indicator of oxidation processes and related acid genera- tion. Core recovery was ϳ70%, and the rock was highly broken up over this interval. All of the rock recovered was tan to orange in color, and the surfaces of fractures, many of which were horizontal, were coated with brown to orange oxidized material. The tran- sition to the fractured and partially oxidized rock layer was rather abrupt at 4.5 m depth, and this layer extended to ϳ9 m. Core recov- ery was nearly 100%. Open fractures in this layer were stained and surrounded with oxi- dation bands up to 25 mm thick. The rock was competent and unoxidized between fractures and outside the oxidation halos. Below 9 m depth, the rock was essentially unweathered. Oxidation stains were seen on some fracture faces, but oxidation halos in the surrounding sandstone were absent. Few hor- izontal fractures were seen. Two fractures were notable below 9 m depth. An open frac- ture with Fe and Mn oxide stains on its sur- faces at 22 m depth was associated with shale

Figure 2. Log of the 35 m core, showing variations in lithology, fractures and veins, and weathered appearance. Rock sample locations shown with ®lled circles. The top 3 m were air blasted without recovery.

1146 Geological Society of America Bulletin, September 2002 LINKAGES BETWEEN WEATHERING AND EROSION IN A SMALL, STEEP CATCHMENT

beds. More oxidation was evident along a fault at 24 m depth. Below these features, all the rock was competent and gray in color, and no oxidation stains were seen.

Spatial Variation in Weathered-Pro®le Development

All of the weathered layers in the bedrock become thinner downslope (Fig. 4). The soil thickens downslope, reaching a maximum ϳ15 m above the channel head. Cross sections at right angles to the longitudinal pro®le show the soil accumulations in the hollow. Average soil depth is 0.70 Ϯ 0.41 m (n ϭ 104). Sap- rolite is present as a nearly continuous layer in the top third of the catchment. It is thin or missing under the deepest soil in the hollow axis in the lower third of the catchment, prob- ably owing to removal in periodic landsliding events. The saprolite varies tremendously in thickness where it is found (0.23 Ϯ 0.26 m, n ϭ 36, excluding borings with no saprolite). The variations in saprolite thickness found on the east nose are suggestive of blocky core stone-style weathering and are consistent with observations of Heimsath et al. (2001b) and with patterns seen in roadcuts. Figure 3. Bulk density, grain density, and porosity as a function of depth in soil and Mineral Assemblage Variations bedrock. Note the vertical-scale break below the saprolite layer; samples in the upper box are from soil pits 1 and 2 in CB2; samples below this are from the core in CB1 (Fig. 1). The mineral assemblage of the graywacke parent material is dominated by quartz, feld-

Figure 4. Longitudinal pro®le and cross sections of the catchment, without vertical exaggeration. The boundaries of weathered-rock lay- ers are extrapolated between tie points of the 35 m core at the top of the pro®le and bedrock borings indicated with vertical lines. Ac- cess rows across the site are num- bered consecutively from bottom to top; cross sections at a few rep- resentative rows are shown.

Geological Society of America Bulletin, September 2002 1147 ANDERSON et al. spar, and lithic (volcanic) fragments in a ®ne- and the mass of the element added or lost in or as percent mass changes relative to the grained groundmass (Table 1). A decline in the weathering process, mj,¯ux: mass of the element in parent rock with the plagioclase and an increase in reddish ground- dimensionless element-mass-transfer coef®- mass are the primary differences between the 1 cient, ␶ : (V ␳ C ) j,w parent rock and the pervasively oxidized rock. 100 ww j,w Variation between the parent-rock samples is mCC 1 ␶ ϵ j,flux 100 ϭϪj,w i,p 1. (4) as great as any variation within the weathered j,w ϭ (Vpp␳ Cj,p) ϩ mj,flux, (1) Vpp␳ CCCj,p j,p i,w pro®le for mica, lithic, hornblende, or alkali 100 feldspar contents. Pitted and weathered quartz where ␳ is the bulk density, C is the concen- Equations 3 and 4 show explicitly the de- and plagioclase grains are evident throughout j tration in weight percent of element j, and the pendence of these calculated mass gains and the core, but are present in greater abundance subscripts w and p refer to weathered material losses on elemental concentrations and parent- in oxidized rock samples. Fractures were lined and parent rock, respectively. In order to rock density. Errors in measurements of soil with reddish-colored groundmass in the per- quantify the mass gains or losses associated bulk density propagate only into the calcula- vasively oxidized rock layer (thin sections with weathering, the volume changes that ac- tion of strain (equation 2) and not into the CB33 and CB35), and the mineral grains edg- company weathering must be known. For an calculation of mass gains and losses (equa- ing the fractures were more likely to be pitted element that is immobile during weathering, tions 3 and 4). than those in unfractured parts of the thin sec- and that does not have an external source, the Addition of material to the soil surface tion. In these weathered-rock samples, a high- m term of equation 1 is zero. In this case, through eolian deposition is a possible source birefringence groundmass appeared to coat j,¯ux equation 1 can be solved for the volumetric of error in the mass-balance analysis. We think many mineral grains, especially those lining strain, ␧ , of the weathered material relative that eolian inputs are not likely to be signi®- fractures. i,w to the parent rock: cant in the CB1 soil given the long distance In the soil samples, quartz, feldspar, biotite, from Asian dust sources, the relatively wet cli- and volcanic-glass abundances were lower Vwp␳ Ci,p mate (minimizing local sources), and the than in the rock, whereas reddish groundmass ␧ϭi,w Ϫ1 ϭϪ1. (2) V ␳ C young age of the soils. and void spaces were more common. The ratio pwi,w of quartz to plagioclase increased by a factor Here the subscript i replaces j to emphasize Mass-Balance Model Results of three to ®ve. The groundmass was uniform- that equation 2 applies only for an immobile ly reddish in color and, in places, was nearly element. We used three of the deepest samples from opaque because of either organic material or The strain from equation 2 can be used to the core to de®ne parent material. The three oxides. All mineral grains within rock frag- calculate mass gains or losses due to weath- samples (CB306, CB318, and CB350; see Fig. ments in the soil were pitted and embayed, ering from equation 1. These are expressed as 2 and Table 1) are from Ͼ5 m below the per- and overgrowths were common. Plagioclase absolute mass changes per unit volume of par- manent water table at 25 m and are composed was found in the soil, but it was much less ent rock, ␦j,w, with units of mass per unit vol- of competent graywacke lacking visible oxi- abundant than in the fresh parent rock. ume: dation. The average chemical composition of

these samples de®ned the Cj,p and Ci,p terms WHOLE-ROCK CHEMISTRY: mj,flux in all calculations. Strains within the soil cal- ␦j,w ϵ QUANTIFICATION OF WEATHERING Vp culated by assuming Ti and Zr immobility de- ALTERATION scribe a nearly perfect 1:1 relationship; in the 1 C ϭ C ␳ i,p Ϫ ␳ C (3) calculations that follow we use Zr as the im- 100 j,wpC p j,p Mass-Balance Model ΂΃i,w mobile element.

TABLE 1. MINERALOGY OF SELECTED SAMPLES We quanti®ed elemental losses and gains that accompany weathering in the soil and Parent rock Pervasively oxidized rock Soil rock through the use of a chemical mass- Sample: CB350 CB318 CB306 CB43 CB35 CB33 Pit1 Pit2 balance model (Brimhall et al., 1985, 1991, Sample depth (m) 35.0 31.8 30.6 4.3 3.5 3.3 0.3 0.6 1992; Brimhall and Dietrich, 1987). The mod- Points counted 200 100 90 100 100 100 100 100 el is predicated on identi®cation and charac- Quartz 40 52 40 44 39 40 32 35 terization of the parent rock. Physical defor- Plagioclase 14 15 13 8 3 6 2 2 mation, or strain, accompanying weathering Alkali feldspar 3 4 0 2 1 1 0 0 Lithic 7 8 14 13 5 9 4 6 can be calculated by conserving the mass of Biotite 6 1 6 3 5 4 2 2 an element that is unaffected by chemical Hornblende 3 3 1 2 4 3 0 4 Muscovite 0 3 1 0 2 2 2 0 weathering. Thus, both the chemical transfor- Glass (volcanic) 6 1 2 2 15 0 0 0 mations and physical deformation of weath- Opaque 1 2 0 0 0 2 0 2 ering can be quanti®ed with this simple Groundmass 20 11 17 14 15 24 21 43 Reddish groundmass* 0 0 6 11 11 6 36 26 model. Void² 1001065354 The mass of a chemical element, j, in a vol- Quartz/plagioclase 2.9 3.5 3.1 5.5 13.0 6.3 15.0 16.0 ume of weathered material, Vw, is equal to the Note: All mineral abundances reported as percent of nonvoid points. mass of the element in the volume of fresh *Organic matter or oxides. ²Percentage of total points that were voids. parent rock, Vp, that weathered to produce Vw

1148 Geological Society of America Bulletin, September 2002 LINKAGES BETWEEN WEATHERING AND EROSION IN A SMALL, STEEP CATCHMENT

in absolute terms, mass losses rank Si Ͼ Ca Ͼ Na Ն Mg, whereas C dominates the mass gains. Both soil pits yield a total mass loss within the soil column of ϳ100 kg´mϪ2. The two saprolite samples stand out from both the soil and the weathered rock (Fig. 6). Both samples have Ca and Na losses compa- rable to those of the overlying soil, but lack the slight silica losses and signi®cant C ad- ditions of the soil. There are accumulations of Al and K in the saprolite not seen elsewhere. We tentatively attribute these to translocation of these elements from the soil because of podzolization. In contrast, depth trends in the mass transfers are not apparent below the sap- rolite. One sample at 3.5 m depth shows sig- ni®cant loss of Al, Si, and other cations, whereas samples at 3.3 and 3.6 m do not show these losses. All of these samples are above the highest occurrence of macroscopic pyrite grain accumulations in the core. The deple- tions measured in the sample at 3.5 m may re¯ect the local effects of acid hydrolysis driv- en by pyrite oxidation (Brimhall et al., 1985).

Figure 5. Pro®le of volumetric strain, ␧, in the soil and weathered-rock layers. Strain is Additional scatter in ␶j,w below the saprolite is calculated from equation 2 by assuming Zr immobility; error bars show the variations comparable to the variability in the parent ma- due to the range of parent-material composition. Note the depth-scale break below the terial. The saprolite appears therefore to be saprolite. truly transitional between the organic-rich, weathered soil and the barely altered weathered- rock layers. The bulk density of the saprolite Soil and bedrock are dramatically different Si in the soil do not fall outside the range of is lower than the weathered rock below it, also in their physical properties (Fig. 3). Soil bulk variability in the parent rock, except for the suggesting that it has undergone signi®cant densities are less than half those of bedrock, shallowest sample. Nonetheless, all ␶Si,w val- mass losses due to weathering. and porosities are correspondingly large. The ues within the soil are negative, rather than decline in bulk density arises from formation distributed about zero, suggesting that silica Comparison with Other Observations of porosity. Consequently, the measured strain has been lost from the soil, but the depletions is large and positive in the soil, but essentially are at the limits of detection with this method. This quantitative analysis of mass losses in zero within the rock (Fig. 5). The saprolite If the silica data are taken at face value, they the rock and soils is in accord with the mineral samples show positive strains that are inter- amount to ϳ10% loss in the soil relative to variations in the pro®le (Table 1). We con- mediate between deep soil and rock. Most of the parent rock. C (and NÐnot shown, but clude that the weathered appearance arises the soil has undergone a positive strain of ϳ2: well correlated with C) both show dramatic from oxidation of ma®c minerals, which i.e., a parcel of rock expands by 100% when gains in the soil, re¯ecting organic matter ac- changes the color of the rock without altering converted into soil. cumulation. K, which is biologically cycled, its composition or mass. A pair of samples at Variability in the parent-rock composition shows a slight accumulation in the shallowest 6.3 m depth from the inside and outside of an leads to some uncertainty in the mass-balance soil sample that probably re¯ects transport up oxidation halo around a fracture supports this calculations. In Figure 6, shaded swaths show into plants and redeposition in leaf litter on view. The pair of samples did not display sig- the effect of compositional differences be- the soil surface (note the association of posi- ni®cant differences in mass transfers (␶j,w) for tween the three parent-rock samples on ␶j,w. tive ␶K,w with an extremely high C addition). any element (Fig. 6), other than a slight ac- Only values that fall outside of this swath can K retention in the soil follows from the pres- cumulation of Fe in the oxidized sample. In be considered different from the parent rock. ervation of muscovite in at least one soil pit summary, measurable weathering is limited to This uncertainty is greatest for Al and Si, be- (Table 1). Al and Fe show slight mass gains the soils and saprolite, where Ca, Na, and Mg cause of the large variations in their concen- in the soil pro®le, although as with Si, the have been signi®cantly depleted, and silica trations in the parent rock and their small per- mass transfers for these elements do not rise has probably been lost, in agreement with the centage changes through the weathered above the parent-rock variability. Our parent- losses in quartz, feldspar, and volcanic glass pro®le. rock samples appear anomalously low in Al. seen in the thin sections. The mass transfers of individual elements Nearly all samples in the pro®le, including all Moreover, the data are consistent with the are greatest within the soil (Fig. 6). Up to 50% but one of the unweathered-rock samples, observation that physical disruption, such as of the Ca and 25% of the Mg and Na origi- show Al additions relative to the parent ma- bioturbation, is the primary mechanism of nally present in the rock has been removed terial. The absolute mass changes in the soil regolith production. The positive strain of 2 or from the soil. The calculated mass transfers of are summarized in Figure 7, which shows that more in the soil is roughly an order of mag-

Geological Society of America Bulletin, September 2002 1149 ANDERSON et al.

1150 Geological Society of America Bulletin, September 2002 LINKAGES BETWEEN WEATHERING AND EROSION IN A SMALL, STEEP CATCHMENT

Figure 6. Calculated mass gains and losses of major cations, silica, and Al in the weathered rock and soil relative to un- weathered parent rock. On each plot, the upper scale expresses changes as a percent- age relative to the element mass in the par- ent rock (␶j,w), whereas the lower scale shows the absolute mass changes per unit

volume of parent rock (␦j,w). Error bars represent variations due to the range of parent-material composition on each cal- culation. Vertical shaded swath shows the effect of this variability on the three parent- rock samples; samples that plot outside of this zone clearly differ in composition from parent material. Circled pair of samples at 6.3 m depth (within the weathered rock) are from the oxidation halo around a frac- ture and the gray, fresh-looking rock out- side the halo. Two calcite-cemented samples are off scale to the right on the Ca plot. N nitude greater than can be explained by addi- tion of organic matter, given a calculated ␦j,w of ϳ100 kg´mϪ3 for C in the soil and any plau- sible bulk density for organic matter. Disso- lution of silicate minerals produced mass loss- es in the soil relative to the parent rock of 100±150 kg´mϪ3 (Fig. 7), which corresponds to a porosity increase of 0.04±0.06 for a min- Ϫ3 eral density of 2600 kg´m . At most, then, Figure 7. Summary of absolute mass gains and mass losses (␦j,w) in two soil pits in the only ϳ5% of the soil porosity can be attri- CB2 catchment. Si and the cations Ca, Na, and Mg are depleted throughout the soils buted to dissolution of minerals. Most of the depth; C, N, and Al show mass additions in the soil (shaded side of plot). The mass losses porosity and dramatic positive strains in the can be summarized as Si Ͼ Ca Ͼ Na Ն Mg; mass gains are dominated by C. The deepest soil can be credited neither to dissolution of sample in each pro®le is saprolite collected from the bottom of the soil pits. The total minerals nor to incorporation of organic mat- mass loss in a column of soil for each pit is reported. ter. This ®nding leads to the conclusion that the strains are derived from purely physical increases in void space, such as would be accord with our conclusion from the soil water trom et al., 1998; White et al., 2002). We caused by stress-releasing exfoliation fractur- chemistry (Anderson and Dietrich, 2001). The therefore compare solute pro®les with the solid- ing and by bioturbation- and creep-driven solute-concentration pro®les of the soil water phase mass transfers. mixing of rock and organic matter. should lead to the development of Spodosols Under the steady ¯ow conditions of our It is easy to draw the conclusion that phys- within the likely residence times of the soils sprinkling experiment, soil water solute con- ical disruption of the soils is important in their on these slopes. Spodosols are found on all centrations are proportional to solute ¯uxes. genesis from observations of tree throw, ani- uplifted marine terraces in the vicinity (Bock- How representative are the soil water solute mal burrowing (particularly by mountain bea- heim et al., 1996). The lack of spodic horizons concentrations during steady ¯ow in our sprin- vers, Aplodontia rufa), the rapid degradation in the CB1 soil we attribute to more frequent kling experiments of typical conditions? of landslide scars in the ®eld area, and the disruption of the soil and litter by physical Clear-cut logging in 1987 is thought to have very low bulk density of the soil. Exfoliation processes than occurs on the low-gradient had minimal effect on soil water and runoff and raveling on bare bedrock surfaces, and terraces. compositions measured in 1992. Catchment scour during landslides and debris ¯ows, may water yield and stream chemistry generally re- also be important in the transformation of rock Comparison with Solute Fluxes turn to baseline within ®ve years (Jewett et al., into soil, although these processes are likely 1995; Swank et al., 2001). In Oregon Coast to be of secondary importance because they To the extent that current weathering pro- Range soils, nutrient concentrations do not operate over limited areas. Without measuring cesses are re¯ected in the soil pro®le, there vary between 5 yr stands and old growth (En- the rates of these physical processes, however, should be a correspondence between the pat- try and Emmingham, 1995). Also, runoff it is dif®cult to verify their in¯uence on soil tern of solute-mass ¯uxes and the mass-transfer chemistry varies Ͻ20% per tenfold variation production. The argument just outlined is in coef®cients calculated from the soil (Stones- in discharge over a range of four orders of

Geological Society of America Bulletin, September 2002 1151 ANDERSON et al. magnitude. Runoff composition variations are we ®nd that the greatest solid-phase mass tant in the genesis and subsequent evolution attributed to changes in ¯ow paths and resi- losses in the soil are from silica and these are of the soils in the CB1 catchment. dence time of water in bedrock under different relatively uniformly distributed through the hydrologic conditions (Anderson et al., soil pro®le (Fig. 7). The humped solute pro- RATE OF MASS LOSS IN SOIL 1997a). Soil water solute concentrations were ®les of Al, Fe, and K should result in trans- VERSUS ROCK quite stable throughout experiment 3 (Ander- location and accumulation of these species son and Dietrich, 2001). In Figure 8, we show within the soil; instead, all show small mass From the analysis of mass losses in the sol- that soil water during experiment 3 is com- gains (accumulation) in the solid phase id phase, it is easy to get the impression that parable with soil water from the peak of a throughout the soil pits, with no evidence of all weathering takes place within the soil. This winter storm (February 20, 1992). The storm a zone of removal. Ca and Na display rela- interpretation is at odds, however, with a con- had a peak rainfall intensity of 15 mm´hϪ1,an tively uniform solute concentrations with clusion from experiment 3 that solute ¯uxes order of magnitude greater than experiment 3, depth, after a steep rise in the top 0.2 m of from the site were derived roughly equally and produced instantaneous discharge 10 the soil. This circumstance should yield mass from the soil and from the bedrock (Anderson times greater than experiment 3 (Anderson et losses of these elements concentrated near the and Dietrich, 2001). The total solute ¯ux al., 1997b). Because of the remarkable consis- soil surface, yet both show uniform mass loss- through the soil, calculated from the soil water tency in soil water during these different hy- es throughout the soil solid phase. Mg solute TDS (total dissolved solids) concentrations drologic conditions and different times in the concentrations are lowest of all the cations, and vadose-zone water ¯ux, equaled half of growing season, we feel that the steady ¯ow which is re¯ected in Mg mass losses in the the total solute ¯ux out of the catchment dur- condition is representative of typical solute solid phase being lower than Na and Ca. In ing experiment 3. Soil water travels through ¯ux conditions. general, the solute-concentration pro®les sug- the bedrock before emerging at the channel The mass balance of any constituent in a gest that mass losses should be concentrated head, and therefore the additional solute in the vertical column of soil during steady solute near the soil surface, rather than be uniform runoff must be derived from bedrock. Because ¯ux conditions is with depth. the soil thickness is less than the weathered- Some of the discrepancies between the sol- rock thickness, this observation suggests that M ute behavior and the mass changes in the solid weathering in the soil is more intense than inץ Qd´ϭϪ١ .t phase may arise from the fact that solutes and the bedrockץ To explore this proposition, we computed soils were not sampled within the same pro- Q the rate of mass loss per unit volume of parentץ QץQxyzץ ϭϪ Ϫ Ϫ , (5) ®le. A more signi®cant difference between the -z rock measured over the time scale for develץ yץ xץ two records, however, is that the solute pro- opment of the weathered pro®le. We ®rst es- ®les show one moment in time, whereas the where M is the mass and Qd is the dissolved tablished the residence time for parcels of rock solid-phase pro®les show the accumulated ef- mass ¯ux per unit area of that constituent. in the weathered rock and of soil layers in the fects of weathering over signi®cant periods of Noting that the solute ¯uxes may be written weathered pro®le. We used several methods to time (several thousand years). Mechanical Qi ϭ UiC, where Ui is the water velocity in address the determination of residence time mixing will tend to obliterate vertical patterns the i direction and C is the solute concentra- for the soil and a single methodÐbased on of mass change through homogenization. tion, and given the speci®c conditions of uplift ratesÐfor the weathered rock. Resi- Changes in solute concentrations of K, Al, Ca, steady vertical ¯ow achieved in experiment 3, dence time in both layers must be set by the equation 5 simpli®es to and Na are greatest in the top 0.5±0.6 m of uplift rate and layer thickness, assuming that the lysimeter pro®les (Fig. 8), and therefore, these layers have constant thickness through -C gradients in the solid-phase mass gains and time. This condition is equivalent to an asץ Mץ ϭϪU . (6) -z losses of these elements should be found with- sumption of a landscape in dynamic equilibץ t zץ in this depth range in the soil pits. The relative rium (Hack, 1960; Pavich, 1986); that is, Because the vadose-zone water during exper- uniformity of the mass losses of Ca and Na maintenance of landforms and the weathering iment 3 moved vertically at a constant veloc- through the soil pro®les and the absence of a pro®les on them, while erosion rates exactly zone of removal of Al and K in the solid phase ity, Uz, within the soil (Torres et al., 1998), match uplift rates. interpretation of mass change in the soil re- (Fig. 6) suggest that, over time, bioturbation sulting from the solute ¯ux is straightforward. homogenizes the soil to at least these depths. Residence Time from Uplift Rates Regions of increasing concentration with Si concentrations in the soil solution increase depth undergo net mass removal, whereas re- steadily downward through the lysimeter pro- Because the weathered-rock pro®le ranges gions of negative solute concentration gradi- ®les, which should result in uniform mass from 3 to 8 m thick in the CB1 catchment, 5 ents accumulate the solute in question. losses of silica in the soil pro®les, a pattern Ϯ 1 m is a reasonable average thickness. The The increasing Si concentrations in soil wa- that cannot be disrupted by bioturbation. Ex- average soil depth in the CB1 catchment is ters with depth (Fig. 8) show that silica is lost cept for the litter sample at the top of soil pit 0.70 m. However, because the bulk density of in solution throughout the soil pro®le owing 2, and the saprolite samples at the bottoms of the soil is signi®cantly lower than that of rock, to chemical weathering. Our analysis of the both soil pits, mass losses of silica are rela- the equivalent thickness of rock represented solid phase suggest silica losses throughout tively uniform in the soil pits, in agreement by the soil, or the ``effective thickness'' of the the soil pro®le (Fig. 6). If we take the soil with the prediction from solute pro®les. Thus, soil, is just 0.23 m. For local uplift rates, we silica data at face value, a stance in accord the soil water chemistry, the soil development, used 0.1 Ϯ 0.07 mm´yrϪ1, which is centered with both the thin-section analysis and the fact and the elemental chemistry of the soils all on the most likely uplift rate for the region that all soil samples have negative ␦Si,w values, point to physical processes being very impor- (H. Kelsey, 1994, personal commun.), but also

1152 Geological Society of America Bulletin, September 2002 LINKAGES BETWEEN WEATHERING AND EROSION IN A SMALL, STEEP CATCHMENT 0, 1992, during Figure 8. Soil waterthe concentration largest pro®les storm on of May the 31, winter 1992, (dotted during lines). the Lysimeter steady nests ¯ow numbered conditions according of to sprinkling the experiment access 3 row (solid on lines) which and they on are February located 2 (see Fig. 4).

Geological Society of America Bulletin, September 2002 1153 ANDERSON et al.

TABLE 2. RESIDENCE TIME ESTIMATES at CB1 are roughly identical (Anderson and Layer Thickness Effective Residence times (k.y.) Dietrich, 2001), the change in density is the (m) thickness 14 same within the layers. This circumstance (m) From From ␦j,w C date Reneau uplift and and Dietrich should be manifested by similar integrated rates solute ¯ux (1991) mass losses in the soil layer and the weathered- Soil 0.70 0.23 2.3 Ϯ 1.8 3.5 Ϯ 1.4 4.07 Ϯ 0.09 5±6 rock layer. Weathered rock, 5550Ϯ 36 including saprolite From the measured annual water ¯ux of 1.6±1.8 m/yr, average bedrock-derived cations plus silica concentration in the runoff of 9.4 Ϯ 3.0 ppm (Anderson and Dietrich, 2001), encompasses most of the likely range. For this Other Measures and uplift rate of 0.10 Ϯ 0.07 mm/yr, we ob- uplift rate, the residence time in the weathered- tain an average change in density within the weathered bedrock layer of 160 Ϯ 123 kg/m3. rock layers is 50 Ϯ 36 k.y., and the residence We have two additional independent age es- A change in density of this magnitude is cer- time in the soil is 2.3 Ϯ 1.8 k.y. (Table 2). timates for the soil. A 14C age of 4070 Ϯ 90 tainly measurable; therefore, the lack of ob- Much of the error comes from the wide range yr B.P. (Center for Accelerator Mass Spec- servable density variation within the bedrock of possible uplift rates, which affect both res- trometry, Lawrence Livermore National Lab- core samples does not stem from too small a idence times the same way. A parcel of rock oratory) on charcoal from the base of soil pit signature. The average bulk density of all the spends ϳ20 times longer traversing the weath- 1 in the CB2 catchment is in good agreement samples from the core is 2270 Ϯ 60 kg/m3, ered-rock layers than passing through the soil. with the values in Table 2. Reneau and Die- whereas bulk densities of two saprolite sam- trich (1991) calculated residence times of 5±6 ples from the soil pits are 1150 and 1850 kg/ Residence Time from Solute Fluxes k.y. for colluvium in the southern Oregon m3 (Fig. 3). These saprolite samples have un- Coast Range, based on 14C age dates of col- dergone density losses of 1120 and 420 kg/ luvial ®lls. An alternative method is to use solute ¯ux- m3, respectively. That the saprolite samples es and our measurements of mass loss in the display mass losses much greater than the av- Depth of Signi®cant Mass Loss in the solid phase of the soil. Several others have erage density change we have just computed Weathered Rock found a close correspondence between modern for the weathered-rock layer is reasonable for solute ¯uxes and long-term weathering rates samples from the top of a pro®le in which the determined from calculated soil mass losses The similarities between the residence times degree of weathering declines with depth. (Stonestrom et al., 1998; White et al., 1998). computed for the soil from the approaches just Our sampling was not suf®ciently dense at The residence time for material in the weath- outlined supports the idea that the local land- the top of the weathered rock to allow us to scape is in dynamic equilibrium. This state- ered pro®le, tw, is calculated from the sum of identify the thickness of the weathered-rock ment means that the soil does not change in mass losses in the soil ⌺(␦j,w) integrated over layer that has undergone signi®cant mass loss- thickness with time and that regolith produc- the soil depth, the mean annual runoff, Qannual, es or to de®ne how these mass losses vary and the average sum of dissolved concentra- tion and erosion both occur at the same rate. with depth. To put limits on the thickness of Å It is likely that this dynamic equilibrium ex- tions of the elements considered, ⌺jCj: this signi®cantly weathered rock layer, we ap- tends to the weathered-rock layers as well. A peal to our observation that integrated mass consequence of the combination of dynamic Hs losses in the weathered rock must equal those (␦ ) dz equilibrium and equal partitioning of solute ͸ j,w of the soil. The integrated mass losses in the ͵ j 0 ¯uxes from the soil and the bedrock is that we t ϭ , (7) bedrock are equal to the average density w Q CÅ should expect to see equivalent solid-phase annual͸ j change multiplied by an appropriate depth j mass losses in the soil and bedrock despite the (⌬␳Å)H. This can be equated to the integrated disparity in residence times within these lay- mass loss in the soil: ers. We make this prediction because in steady where HS is the soil depth. The numerator is state, the residence time, tr, and thickness of the total mass loss from the soil solid phase, Hs and the denominator is the rate of mass loss layers, H, are related through the uplift rate, u (⌬␳Å )H ϭ (␦ ) dz, (9) RR ͸ j,w H/t . Material in the thinner soil layer thus ͵ j in solution. ϭ r 0 The numerator of equation 7, computed as has a shorter residence time than in the thicker weathered rock in proportion to the uplift rate. the product of the average ␦j,w values for Si, where the left side shows the mass losses in Ca, Na, and Mg from the soil pits (ϭ 94.5 Ϯ We can calculate the average change in den- the rock and the right side shows the mass 27.7 kg´mϪ3) and the average soil thickness, sity, ⌬␳Å, in a layer due to mass loss in solution losses in the soil. We have already found is 66.2 Ϯ 19.4 kg´mϪ2. For the denominator, as that the right side of equation 9 is 66.2 Ϯ we used the mean annual runoff from the CB1 19.4 kg´mϪ2 and that ⌬␳Å is 160 Ϯ 123 Ϫ1 Q (C Ϫ C ) Ϫ3 catchment of 1.6±1.8 m´yr (Anderson, ⌬␳Å ϭϪ w0, (8) kg´m . Therefore, the representative thick- 1995), and the average dissolved cations plus u ness, HR, for the weathered-rock layer is silica concentration in the soil during steady 0.41 Ϯ 0.34 m. runoff in experiment 3 of 11.0 Ϯ 3.0 ppm where Qw is the water ¯ux (in m/yr), and (CÐ Constraints on the distribution of mass loss-

(Anderson and Dietrich, 2001). The soil resi- C0) is the change in solute concentration as es in the weathered rock are depicted in Figure dence time calculated from equation 7 is 3.5 water ¯ows through the layer. Because the net 9. The total mass loss in the weathered-rock Ϯ 1.4 k.y. solute ¯uxes from the soil and bedrock layers layer is equal to the area of the box in Figure

1154 Geological Society of America Bulletin, September 2002 LINKAGES BETWEEN WEATHERING AND EROSION IN A SMALL, STEEP CATCHMENT

librium leads to the squared dependence on the layer thickness, as the residence time with- in a layer depends on thickness. The ratio of rate of mass loss in the soil to the weathered rock in the CB1 catchment therefore depends on the thickness of the layer that has undergone signi®cant mass loss in the weathered rock. If the calculated 0.4 m aver- age depth of mass loss is appropriate for the rock, and the soil is 0.23 m in depth (rock equivalent), then this ratio is 3. For the pro- ®les shown in Figure 9, the ratio ranges from 0.21 to 11. The smaller ratio, which implies much greater rate of mass loss in the rock than in the soil, is for the case that yields the thin- nest zone of weathering in the rock, whereas the larger ratio is for a weathered-rock layer of 1 m thickness. Using the average saprolite thickness of 0.23 Ϯ 0.26 m yields a ratio of 1, meaning that the rate of mass loss within Figure 9. Plausible models of vertical distribution of mass losses in the weathered rock. the bedrock is the same as that in the soil. Although our data on rate of mass loss in Box shows integral constraint on mass losses determined from (⌬␳Å R)HR. Average measured saprolite thickness (dashed line) falls within this layer of signi®cant mass loss. Points at 0 the soil versus the rock is equivocal, we think depth show measured density differences in saprolite samples collected at the base of soil that a greater rate of mass loss in the soil is pits (the top of the weathered rock); point at 2.3 m depth shows the uppermost core the most likely scenario. The degree of weath- sample. Pairs of curves (one pair light, one pair bold) show linear and exponential dis- ering at the top of the weathered-rock pro®le tributions of ⌬␳ that ®t the integral constraint for each of the two saprolite measurements. clearly varies signi®cantly throughout the In all cases, the depth of signi®cant mass losses due to weathering is con®ned to a layer catchment. The samples of rock collected at Ͻ1 m thick (generally Ͻ0.4 m thick). the base of our two soil pits differ in ⌬␳ by a factor of two. Saprolite is absent in parts of the catchment and up to 1 m thick elsewhere. Ϫ3 9 whose sides are ⌬ϭ␳Å 160 kg´m and HR by stochastic events of relatively small length Given this variability, our computed average ϭ 0.4 m. Measurements of ⌬␳ in the two sap- scale, in agreement with the assessment of depth of signi®cant mass loss of 0.4 m in the rolite samples and of ⌬␳ ϭ 0 for the upper- Heimsath et al. (2001b). bedrock provides the best measure of the typ- most sample of the bedrock core are shown as We can now address the rate of mass loss ical weathered-rock pro®le. This thickness im- points. By using the saprolite ⌬␳ values and in the soil versus the bedrock. The rate of plies that rate of mass loss in the soil is ap-

the integral constraint on total mass loss, we mass loss per unit volume of rock, or Iw,is proximately three times greater than in the can construct the ⌬␳ pro®les for cases of lin- ⌬M/(V⌬t), where ⌬M is the integrated mass rock. ear and exponential decline in mass loss with loss, V is volume, and ⌬t is the time over There are a number of reasons that weath- depth. In all four cases, signi®cant mass losses which the mass loss occurs. If we consider ering may be more intense within the soil than are con®ned to the top1moftheweathered mass losses in a column of rock or soil, and the rock. First, the effects of vegetation on rock, and most likely to the top 0.4 m, well therefore replace volume with thickness (i.e., weathering will be strongest in the soil. In above our uppermost core sample at 2.3 m volume per unit area), and note that residence general, vegetation affects chemical-weathering into the weathered rock. These pro®les com- time is the thickness divided by the uplift rate, rates by producing weathering agents, cycling pare well with an average saprolite thickness the rate of mass loss in a layer of thickness H cations, and altering soil hydrology (Kelly et in our borings (in the 36 cores where we en- can be expressed as al., 1998). Organic acids, which mediate sili- countered it) of 0.23 Ϯ 0.26 m. Saprolite, cate weathering through chelation of cations however, was not found throughout the catch- ⌬M and metals, peak in concentration in the upper I ϭ . (10) ment. This analysis yields average mass loss w H(H/u) 0.5 m of the CB1 soil (Fig. 8). Second, by- in the weathered rock throughout the catch- passing ¯ow routes (root holes, soil fractures) ment. The good agreement between saprolite Because ⌬M for the soil equals ⌬M in the are not signi®cant in the CB1 soils, whereas thickness and average depth of signi®cant weathered rock in CB1, the ratio of rate of fracture ¯ow is important in the bedrock. mass loss in the catchment implies that these mass loss in the soil to rate of mass loss in Thus, more water passes through the soil ma- mass losses are probably not con®ned to just the weathered rock reduces to trix than through the bedrock matrix, giving the saprolite layer itself, but may also be greater contact with mineral surfaces in the found in the uppermost weathered rock where IH2 soil. Chemical-weathering rates depend on SRϭ , (11) 2 saprolite is missing. That the degree of weath- IHRS both water velocity (which sets contact time) ering in the top of the bedrock, as represented and water ¯ux. At CB1, the maximum in con- by the bulk density loss, is so variable sug- where subscripts S and R refer to soil and tact time and water ¯ux appears to be met in gests that conversion of rock to soil is driven weathered rock, respectively. Dynamic equi- the soil. We found that water-¯ow rates

Geological Society of America Bulletin, September 2002 1155 ANDERSON et al.

TABLE 3. LONG-TERM SILICA FLUXES

Location Reference Mean annual Mean annual Soil age Si ¯ux precipitation temperature (ka) (t km±2 yr±1) (m) (ЊC) CB1 This study 1.6±1.8 9* 3.5 Ϯ 2.5 10.7 Ϯ 7.1 Rio Icacos, Puerto Rico White et al. (1998); Stonestrom et al. (1998) 4.2 22 200 14.9 Panola, Georgia Stonestrom et al. (1998) 1.2 16 350 4.8 Riverbank soil, California Stonestrom et al. (1998) 0.3 16 250 2.2 Strawberry Rock soil, California Chadwick et al. (1990) 1±1.5 12 240 2.1 *Data from National Climate Data Center, for North Bend Municipal Airport, assuming a 6 ЊC/km lapse rate.

Ϫ3 through the soil vadose zone were controlled organic acids, highPCO2 , and oxygen all con- Ϯ 28.3 kg´m ), the average soil thickness, by rainfall rates (Anderson et al., 1997b; Tor- tribute to enhancing chemical-weathering and the average residence time for material in res et al., 1998), commonly ϳ1 mm´hϪ1,or rates. Chemical weathering is most intense the soil of 3.5 Ϯ 1.4 k.y. (Table 2)Ðis 10.7 ϳ10Ϫ6 m´sϪ1. These rates are comparable to within the soil layer, but it is also signi®cant Ϯ 7.1 t´kmϪ2´yrϪ1. This value falls above the the few measurements of saturated hydraulic within the rock. Water ¯ow is a key to the long-term silica ¯uxes calculated in several conductivity in near-surface weathered-bedrock rates of chemical processes within both rock other catchments on the basis of similar anal- matrix of 10Ϫ5 to 10Ϫ7 m´sϪ1 (Montgomery et and soil, and the reason that half of the solute yses of soil and runoff chemistry (Table 3). al., 1997). A signi®cant fraction of the water from the CB1 catchment derives from the rock Among these, only the example from the trop- ¯ow through the bedrock, however, occurs is that nearly all water ¯ows through the bed- ical Rio Icacos soil yields a greater silica ¯ux through fractures, where high velocities rock in its passage through the catchment. than does the CB1 soil. Extremely high rain- (ϳ10Ϫ3 m´sϪ1) and little contact with mineral Throughout the weathered pro®le, chemical fall and temperature may account for the high surfaces both work to limit solute acquisition. weathering processes operate in tandem with weathering rates from the Rio Icacos soil. Al- Thus, the water ¯ux through the rock matrix physical processes except during two physi- though we cannot rule out weathering of vol- is less than in the soil because some ¯ow by- cally driven transformations. These are the canic glass as the source of high silica ¯uxes passes the rock in fractures or along the soil/ formation of regolith (soil) from the underly- from CB1, the most striking difference be- rock interface. We found no evidence for by- ing rock and the transport of soil out of the tween the remaining sites and the CB1 catch- pass ¯ow in the soil. catchment. These two key events check fur- ment is the age of the soil. This difference ther chemical evolution of the rock (in the ®rst suggests that the ongoing rejuvenation of CB1 DISCUSSION case) or of the soil (in the second). Because soils by physical regolith-production process- these events are physically driven, physical es sustains higher chemical-weathering rates Weathering of the rock starts with opening processes set the time scaleÐand hence the than would be reached in a more stable envi- of fractures. As rock moves up through the degree of development of the weathered pro- ronment. Riebe et al. (2001) found in granitic pro®le, the number of fractures increases (Fig. ®leÐand chemical weathering plays a second- Sierra Nevada catchments that silica ¯uxes 2). These provide avenues for water ¯ow ary role. were greatest in the areas undergoing the most through the rock, but the passage of that water Although physical processes control rapid physical denudation. The CB1 silica ¯ux is relatively rapid. With time, the number of weathered-pro®le development and regolith of 10.7 Ϯ 7.1 t´kmϪ2´yrϪ1 and erosion rate of fractures increases and oxidation halos grow, formation, the degree of chemical weathering 110±180 t´kmϪ2´yrϪ1 (Reneau and Dietrich, eventually coalescing to form a pervasively that takes place within this framework is re- 1991) are comparable to the 2±8 t´kmϪ2´yrϪ1 oxidized rock layer. In the upper 0.4 m of this markable. Within the 1±10 k.y. residence time range that Riebe et al. (2001) reported for Si- layer (on average), weathering produces sig- in the saprolite and soil, 25%±50% of cations erran catchments with total erosion rates of ni®cant mass losses; consequently, changes in and perhaps 10% of the silica in the rock are ϳ100 t´kmϪ2´yrϪ1. bulk density are found. Where physically un- removed. By comparison, Chadwick et al. The connection between uplift and silicate disturbed, chemical degradation continues to (1990) measured Si losses of 29% and Na chemical weathering appears to be as follows. the point of loss of mechanical strength, and losses of 57% in a 240 ka soil in northern At very high uplift or physical-denudation saprolite forms. Although we have not tried to California. If the residence time of material in rates, soil may be very thin or absent. This measure it, it seems likely that matrix hydrau- the CB1 soil were increased, it seems likely circumstance will yield low, but nonnegligible lic conductivity must increase upward through that aluminous clay phases would accumulate chemical denudation rates associated with the pervasively oxidized rock and saprolite and change the soil's hydraulic properties. For weathering within the bedrock. Fracture ¯ow, layers, as weathering produces porosity. In- example, the soil studied by Chadwick et al. the likely dominant hydrologic pathway in creasing water ¯ow through the rock matrix (1990) contained 30% secondary clay. The ef- bedrock, limits both the surface area of min- as it weathers provides a positive feedback on fect of clay formation on soil water movement erals in contact with water and the contact the chemical-weathering rate within the rock. is one source of the difference in rate of mass time. At lower uplift and physical-denudation Finally, material is transformed from rock into loss between these sites. Another difference is rates, the landscape may be thinly soil man- soil by physical or physical and biological the lack of topographic gradient driving ero- tled, and chemical weathering will take place processes that tear up and transport rock frag- sion and refreshing material in the soils that in both the soil and the rock. This situation ments. These regolith-forming processes feed Chadwick et al. (1990) studied. exists at CB1, where half of all weathering material into the ®nal stage of weathering The long-term silica ¯ux from the CB1 takes place within the rock, and silica ¯uxes within the pro®le. In the soil, water ¯ows catchment soilsÐcomputed from the mea- are quite high. At low or negligible uplift and readily through the matrix. The presence of sured average Si loss in the soil (␦Si,w ϭ 54.1 physical-denudation rates, soils will thicken,

1156 Geological Society of America Bulletin, September 2002 LINKAGES BETWEEN WEATHERING AND EROSION IN A SMALL, STEEP CATCHMENT

and we think that chemical-weathering rates rock at CB1 is lower than within the soil. The balance relations between chemical composition, vol- ume, density, porosity, and strain in metasomatic hy- will be reduced. At this end of the spectrum, implication is that areas undergoing the high- drochemical systems: Results on weathering and pe- low physical-process rates mean that soils can est uplift rates will not produce the highest dogenesis: Geochimica et Cosmochimica Acta, v. 51, develop into mature pro®les. Development of silicate-weathering rates. Instead, silicate- p. 567±587. Brimhall, G.H., Lewis, C.J., Ford, C., Bratt, J., Taylor, G., clay may impede water ¯ow through the soil, weathering rates will be maximized in actively and Warin, O., 1991, Quantitative geochemical ap- whereas thickening the soil may lead to chem- eroding, soil-mantled landscapes. proach to pedogenesis: Importance of parent material reduction, volumetric expansion, and eolian in¯ux in ical saturation control on chemical-weathering lateritization: Geoderma, v. 51, p. 51±91. rates (e.g., White et al., 2001). Together, these ACKNOWLEDGMENTS Brimhall, G.H., Chadwick, O.A., Lewis, C.J., Compston, effects suggest that silicate chemical-weathering W., Williams, I.S., Danti, K.J., Dietrich, W.E., Power, M.E., Hendricks, D., and Bratt, J., 1992, Deforma- rates will be greatest at some intermediate up- We thank Tim Teague, C. Lewis, and L. Abbott for assistance with the laboratory work and O. tional mass transport and invasive processes in soil lift or physical-denudation rate, where physi- Chadwick for discussions about the mass-balance evolution: Science, v. 255, p. 695±702. cal processes are rapid enough to continually Carson, M.A., and Kirkby, M.J., 1972, Hillslope form and model. We appreciated thoughtful reviews by S. process: Cambridge, UK, Cambridge University Press, add fresh rock into the weathering pro®le, yet Brantley, R. Stallard, and associate editor J. Hanor. 475 p. low enough to permit development of a soil. This work was supported by a National Aeronautics Chadwick, O.A., Brimhall, G.H., Jr., and Hendricks, D.M., and Space Administration global change graduate fel- 1990, From a black to a gray boxÐA mass balance lowship (NGT30083), the National Science Founda- interpretation of pedogenesis: Geomorphology, v. 3, CONCLUSIONS tion (EAR8417467), the U.S. Geological Survey Wa- p. 369±390. ter Resources Division (14080001G12111), and the Dahlgren, R.A., and Ugolini, F.C., 1989, Aluminum frac- Weyerhauser Company. CSIDE (Center for Study of tionation of soil solutions from unperturbed and teph- The weathered pro®le in this steep, actively ra-treated Spodosols, Cascade Range, Washington, eroding headwater basin consists of a loose, Imaging and Dynamics of the Earth) contribution USA: Soil Science Society of America Journal, v. 53, number 438. poorly developed 0.7 m deep soil overlying p. 559±566. ϳ Dietrich, W.E., Reneau, S.L., and Wilson, C.J., 1986, Hol- 3±8 m of weathered rock. The total mass lost lows, colluvium, and landslides in soil-mantled land- by chemical weathering is as great from the REFERENCES CITED scapes, in Abrahams, A.D., ed., Hillslope processes: Boston, Allen and Unwin, p. 361±388. rock as from the soil. Because the thickness Dott, R.H., Jr., 1966, Eocene deltaic sedimentation at Coos Adams, J., 1984, Active deformation of the Paci®c North- of the weathering layer is greater in the rock Bay, Oregon: Journal of Geology, v. 74, p. 373±420. west continental margin: Tectonics, v. 3, p. 449±472. England, P., and Molnar, P., 1990, Surface uplift, uplift of than in the soil (normalized to equivalent den- Ahnert, F., 1967, The role of the equilibrium concept in the rocks, and exhumation of rocks: Geology, v. 18, interpretation of landforms of ¯uvial erosion and de- sity), the rate of mass loss is greater in the soil p. 1173±1177. position, in Macar, P., ed., L'Evolution des Versants: Entry, J.A., and Emmingham, W.H., 1995, In¯uence of for- than in the rock. Soils are important sites of Les Congress et Colloques de l'Universite de Liege, est age on nutrient availability and storage in conif- chemical weathering, because the combination v. 40, p. 23±41. erous soils of the Oregon Coast Range: Canadian Anderson, R.S., 1999, Near-surface thermal pro®les in al- of hydrology and biology that they support Journal of Forest Research, v. 25, p. 114±120. pine bedrock: Implications for the frost weathering of Gaillardet, J., DupreÂ, B., Louvat, P., and AlleÁgre, C.J., generally enhances solute-production rates. rock: Arctic and Alpine Research, v. 30, p. 362±372. 1999a, Global silicate weathering and CO consump- Anderson, R.S., and Humphrey, N.F., 1989, Interaction of 2 Although weathering within bedrock is often tion rates deduced from the chemistry of large rivers: weathering and transport processes in the evolution of ignored (e.g., Riebe et al., 2001), this work Chemical Geology, v. 159, p. 3±30. arid landscapes, in Cross, T., Quantitative Dynamic Gaillardet, J., DupreÂ, B., and AlleÁgre, C.J., 1999b, Geo- demonstrates that bedrock can also be a sig- Stratigraphy: Englewood Cliffs, New Jersey, Prentice- chemistry of large river suspended sediments: Silicate Hall, p. 349±361. ni®cant site of chemical weathering. weathering or recycling tracer? Geochimica et Cos- Anderson, S.P., 1995, Flow paths, solute sources, weath- Ϫ2 Ϫ1 mochimica Acta, v. 63, p. 4037±4051. The silica ¯ux of 10.7 Ϯ 7.1 t´km ´yr ering, and denudation rates: The chemical geomor- from the CB1 catchment is higher than that phology of a small catchment [Ph.D. thesis]: Berkeley, Gilbert, G.K., 1877, Report on the geology of the Henry Mountains (Utah): Washington, D.C., U.S. Geograph- from other catchments with considerably older University of California, 380 p. Anderson, S.P., and Dietrich, W.E., 2001, Chemical weath- ical and Geological Survey, Publication of the Powell weathering pro®les (Table 3) and is compa- ering and runoff chemistry in a steep headwater catch- Survey, Government Printing Of®ce, 160 p. rable to silica ¯uxes from catchments with ment: Hydrological Processes, v. 15, p. 1791±1815. Haagen, J.T., 1989, Soil survey of Coos County, Oregon: U.S. Department of Agriculture, Soil Conservation similar physical-denudation rates (Riebe et al., Anderson, S.P., Dietrich, W.E., Torres, R., Montgomery, D.R., and Loague, K., 1997a, Concentration-discharge Service, 269 p. 2001). We infer that physical processes en- relationships in a steep, unchanneled catchment: Water Hack, J.T., 1960, Interpretation of erosional topography in hance chemical-weathering rates by rejuve- Resources Research, v. 33, p. 211±225. humid temperate regions: American Journal of Sci- Anderson, S.P., Dietrich, W.E., Montgomery, D.R., Torres, ence, v. 258-A, p. 80±97. nating and refreshing material within the soil R., Conrad, M.E., and Loague, K., 1997b, Subsurface Heimsath, A.M., Dietrich, W.E., Nishiizumi, K., and Finkel, ``reactor.'' The hydrology of a young, biotur- ¯ow paths in a steep unchanneled catchment: Water R.C., 1997, The soil production function and land- scape equilibrium: Nature, v. 388, p. 358±361. bated soil is also conducive to high solute Resources Research, v. 33, p. 2637±2653. Baldwin, E.M., 1974, Eocene stratigraphy of southwestern Heimsath, A.M., Dietrich, W.E., Nishiizumi, K., and Finkel, ¯uxes because of the lack of impeding hori- Oregon: Oregon Department of Geology and Mineral R.C., 1999, Cosmogenic nuclides, topography, and the zons. These observations suggest that chemi- Industries Bulletin, no. 83, 38 p. spatial variation of soil depth: Geomorphology, v. 27, p. 151±172. cal-weathering rates from catchments depend Baldwin, E.M., 1975, Revision of the Eocene stratigraphy of southwestern Oregon, in Weaver, D.W., Hornaday, Heimsath, A.M., Chappell, J., Dietrich, W.E., Nishiizumi, on physical-denudation rates and therefore on G.R., and Tipton, A., eds., Paleogene symposium and K., and Finkel, R.C., 2000, Soil production on a re- uplift rates. However, we infer that the rela- selected technical papersÐConference on future en- treating escarpment in southeastern Australia: Geolo- ergy horizons of the Paci®c Coast: Long Beach, Cal- gy, v. 28, p. 787±790. tionship between chemical weathering and up- ifornia, American Association of Petroleum Geolo- Heimsath, A.M., Chappell, J., Dietrich, W.E., Nishiizumi, lift rates is not likely to be monotonic. When gists, p. 49±64. K., and Finkel, R.C., 2001a, Late Quaternary erosion uplift rates are high enough that the landscape Berner, R.A., 1990, Atmospheric carbon dioxide levels over in southeastern Australia: A ®eld example using cos- Phanerozoic time: Science, v. 249, p. 1382±1386. mogenic nuclides: Quaternary International, v. 83±85, becomes bedrock dominated, hydrologic con- Bockheim, J.G., Marshall, J.G., and Kelsey, H.M., 1996, p. 169±185. tact with mineral surfaces diminishes (fracture Soil-forming processes and rates on uplifted marine Heimsath, A.M., Dietrich, W.E., Nishiizumi, K., and Finkel, ¯ow dominates), and enhancement of weath- terraces in southwestern Oregon, USA: Geoderma, R.C., 2001b, Stochastic processes of soil production v. 73, p. 39±62. and transport: Erosion rates, topographic variation and ering by biotic processes declines. Both of Brimhall, G.H., Alpers, C.N., and Cunningham, A.B., 1985, cosmogenic nuclides in the Oregon Coast Range: these effects will tend to lower the chemical Analysis of supergene ore-forming processes and Earth Surface Processes and Landforms, v. 26, ground water solute transport using mass balance prin- p. 531±552. denudation rate. This result parallels our ar- ciples: Economic Geology, v. 80, p. 1227±1256. Holland, H.D., 1978, The chemistry of the atmosphere and guments that rate of mass loss within the bed- Brimhall, G.H., and Dietrich, W.E., 1987, Constitutive mass oceans: New York, John Wiley and Sons, 351 p.

Geological Society of America Bulletin, September 2002 1157 ANDERSON et al.

Jewett, K., Daugharty, D., Krause, H.H., and Arp, P.A., ocean geochemical cycles: Geology, v. 16, using a steady-state perspective: Annual Review of 1995, Watershed responses to clear-cutting: Effects on p. 649±653. Earth and Planetary Sciences, v. 23, p. 11±39. soil solutions and stream water discharge in central Reneau, S.L., and Dietrich, W.E., 1991, Erosion rates in the Stonestrom, D.A., White, A.F., and Akstin, K.C., 1998, De- New Brunswick: Canadian Journal of Soil Science, southern Oregon Coast Range: Evidence for an equi- termining rates of chemical weathering in soilsÐSol- v. 75, p. 475±490. librium between hillslope erosion and sediment yield: ute transport versus pro®le evolution: Journal of Hy- Kelly, E.F., Chadwick, O.A., and Hilinski, T.E., 1998, The Earth Surface Processes and Landforms, v. 16, drology, v. 209, p. 331±345. effect of plants on mineral weathering: Biogeochem- p. 307±322. Swank, W.T., Vose, J.M., and Elliott, K.J., 2001, Long-term istry, v. 42, p. 21±53. Reneau, S.L., Dietrich, W.E., Rubin, M., Donahue, D.J., hydrologic and water quality responses following Kelsey, H.M., Engebretson, D.C., Mitchell, C.E., and Tick- and Jull, A.J.T., 1989, Analysis of hillslope erosion commercial clearcutting of mixed hardwoods on a nor, R.L., 1994, Topographic form of the Coast Rang- rates using dated colluvial deposits: Journal of Geol- southern Appalachian catchment: Forest Ecology and Management, v. 143, p. 163±178. es of the Cascadia margin in relation to coastal uplift ogy, v. 97, p. 45±63. Torres, R., Dietrich, W.E., Loague, K., Montgomery, D.R., rates and plate subduction: Journal of Geophysical Re- Riebe, C.S., Kirchner, J.W., Granger, D.E., and Finkel, search, v. 99, p. 12245±12255. and Anderson, S.P., 1998, Unsaturated zone processes R.C., 2001, Strong tectonic and weak climatic control and the hydrologic response of a steep, unchanneled Kelsey, H.M., Ticknor, R.L., Bockheim, J.G., Mitchell, of long-term chemical weathering rates: Geology, C.E., 1996, Quaternary upper plate deformation in catchment: Water Resources Research, v. 34, v. 29, p. 511±514. p. 1865±1879. coastal Oregon: Geology Society of America Bulletin, Rosenbloom, N.A., and Anderson, R.S., 1994, Hillslope v. 108, p. 843±860. White, A.F., Blum, A.E., Schulz, M.S., Huntington, T.G., and channel evolution in a marine terraced landscape, Peters, N.E., and Stonestrom, D.A., 2002, Chemical Lovell, J.P.B., 1969, Tyee Formation: Undeformed turbi- Santa-Cruz, California: Journal of Geophysical Re- dites and their lateral equivalents: Mineralogy and pa- weathering of the Panola granite: Solute and regolith search, v. 99B, p. 14013±14029. leogeography: Geological Society of America Bulle- elemental ¯uxes and the weathering rate of biotite, in Ruddiman, W.F., and Prell, W.L., 1997, Introduction to the Hellmann, R., and Wood, S.A., eds., Water rock in- tin, v. 80, p. 9±22. uplift-climate connection, in Ruddiman, W.F., ed., Tec- teraction, ore deposits, and environmental geochem- MacDonald, L.H., 1988, An inexpensive, portable system tonic uplift and climate change: New York, Plenum istry, a tribute to David A. Crerar: Geochemical So- for drilling into subsurface layers: Soil Science Soci- Press, p. 3±15. ciety Special Publication 7, p. 37±59. ety of America Journal, v. 52, p. 1817±1819. Small, E.E., Anderson, R.S., and Hancock, G.S., 1999, Es- White, A.F., and Blum, A.E., 1995, Effects of climate on Montgomery, D.R., Dietrich, W.E., Torres, R., Anderson, chemical weathering in watersheds: Geochimica et S.P., Heffner, J.T., and Loague, K., 1997, Hydrologic timates of the rate of regolith production using Be-10 and Al-26 from an alpine hillslope: Geomorphology, Cosmochimica Acta, v. 59, p. 1729±1747. response of a steep, unchanneled valley to natural and White, A.F., Blum, A.E., Schulz, M.S., Vivit, D.V., Stones- applied rainfall: Water Resources Research, v. 33, v. 27, p. 131±150. Stallard, R.F., 1985, River chemistry, geology, geomor- trom, D.A., Larson, M., Murphy, S.F., and Eberl, D., p. 91±109. 1998, Chemical weathering in a tropical watershed, Pavich, M.J., 1986, Processes and rates of saprolite pro- phology, and soils in the Amazon and Orinoco basins, in Drever, J.I., ed., The chemistry of weathering: Dor- Luquillo Mountains, Puerto Rico: I. Long-term versus duction and erosion on a foliated granitic rock of the short-term weathering ¯uxes: Geochimica et Cosmo- drecht, Netherlands, D. Reidel Publishing Company, Virginia Piedmont, in Colman, S.M., and Dethier, chimica Acta, v. 62, p. 209±226. p. 293±316. D.P., Rates of chemical weathering of rocks and min- White, A.F., Bullen, T.D., Schulz, M.S., Blum, A.E., erals: Orlando, Florida, Academic Press, p. 551±590. Stallard, R.F., 1992, Tectonic processes, continental free- Huntington, T.G., and Peters, N.E., 2001, Differential Personius, S.F., 1993, Age and origin of ¯uvial terraces in board, and the rate-controlling step for continental de- rates of feldspar weathering in granitic regoliths: Geo- the central Coast Range, western Oregon: U.S. Geo- nudation, in Butcher, S.S., Charlson, R.J., Orians, chimica et Cosmochimica Acta, v. 65, p. 847±869. logical Survey Bulletin Report B 2038, 56 p. G.H., and Wolfe, G.V., eds., Global biogeochemical Worona, M.A., and Whitlock, C., 1995, Late Quaternary Personius, S.F., Kelsey, H.M., and Grabau, P.C., 1993, Ev- cycles: London, UK, Academic Press, p. 93±121. vegetation and climate history near Little Lake, Cen- idence for regional stream aggradation in the central Stallard, R.F., 1995a, Relating chemical and physical ero- tral Coast Range, Oregon: Geological Society of Oregon Coast Range during the Pleistocene±Holocene sion, in White, A.F., and Brantley, S.L., eds., Chemical America Bulletin, v. 107, p. 867±876. transition: Quaternary Research, v. 40, p. 297±308. weathering rates of silicate minerals: Mineralogical Raymo, M.E., and Ruddiman, W.F., 1992, Tectonic forcing Society of America Reviews in Mineralogy, v. 31, MANUSCRIPT RECEIVED BY THE SOCIETY APRIL 20, 2001 REVISED MANUSCRIPT RECEIVED JANUARY 2, 2002 of late Cenozoic climate: Nature, v. 359, p. 117±122. p. 543±564. MANUSCRIPT ACCEPTED JANUARY 21, 2002 Raymo, M.E., Ruddiman, W.F., and Froelich, P.N., 1988, Stallard, R.F., 1995b, Tectonic, environmental, and human In¯uence of late Cenozoic mountain building on aspects of weathering and erosion: A global review Printed in the USA

1158 Geological Society of America Bulletin, September 2002