mineral weathering rates in glacial drift soils (sw...

Available online at www.sciencedirect.com

(2008) 129–154www.elsevier.com/locate/chemgeo

Chemical Geology 249

Mineral weathering rates in glacial drift soils (SW Michigan, USA):New constraints from seasonal sampling of waters and

gases at soil monoliths

Lixin Jin a,⁎, Stephen K. Hamilton b, Lynn M. Walter a

a Department of Geological Sciences, University of Michigan, 2534 CC Little Building, 1100 N. University Ave.,Ann Arbor, MI 48109-1005, United States

b Kellogg Biological Station and Department of Zoology, Michigan State University, 3700 E. Gull Lake Drive,Hickory Corners, MI 49060-9516, United States

Received 2 July 2007; received in revised form 4 December 2007; accepted 5 December 2007

Edito
r: J. Fein
Abstract

Soil solutions and gases were sampled along 200 cm deep soil profiles from four instrumented soil monoliths in southwestMichigan,established on coarse-grained glacial drift deposits. Seasonal sampling enabled evaluation of thermodynamic versus kinetic controls oncarbonate- and silicate-mineral weathering rates, allowing better integration with past field hydrogeochemical studies of Michigan soiland surface water systems. Silicate-weathering products dominate water chemistry in the upper soil zones. Carbonate minerals,comprised of subequal amounts of calcite and dolomite, are only present at depths below 150 cm.When present, carbonate dissolution israpid and soil water Ca2+ and Mg2+ concentrations increase dramatically as observed in other natural soil study sites in southernMichigan. Soil water saturation states are near equilibriumwith respect to calcite and slightly less saturated with respect to dolomite. Thedivalent cations of soil waters and soil CO2 both show a seasonal trend, with concentration maxima occurring in September and minimain April, suggesting that soil water Ca2+ and Mg2+ concentrations are under equilibrium control with carbonate solubility limited bytemperature-dependent pCO2 rather than by direct effects of temperatures. Importantly, monolith soil water Mg2+/Ca2+ and calcite anddolomite saturation states are lower than those of streams in the samewatershed and also lower than those of soilwaters in otherMichiganwatersheds. Because carbonate weight percentages and chemical compositions in these sites are similar, this difference likely reflects theshort exposure path (thus short residence time) of soil waters to carbonate-rich horizons in the monoliths.

The dissolution reactions of primary aluminosilicate minerals are incongruent with respect to Al and Si due to kaolinite formation.However, major cations (Ca2+, Mg2+, K+ and Na+) are stoichiometrically released from silicate dissolution. Na⁎ (soil water Na+ aftercorrection for atmospheric input and derived primarily from plagioclase weathering) exhibits much less seasonality than divalentcations, with only slight elevations observed in the summer months. Soil water H4SiO4

0 concentrations show seasonal variationssimilar to the divalent cations, but are determined by the balance between production (silicate-mineral dissolution) and consumption(kaolinite precipitation). Plagioclase and amphibole are below saturation, and these dissolution reactions must be kineticallycontrolled. Through a conservative tracer study, about 15% to 45% of applied Br passed out of the monolith profiles in 40–160 daysand this long mineral–water contact time is especially important for slow reactions such as silicate dissolution.

⁎ Corresponding author. Current address: Center for Environmental Kinetics Analysis, Earth and Environmental Systems Institute, Penn StateUniversity, State College, PA 16802, United States.

E-mail address: [email protected] (L. Jin).

0009-2541/$ - see front matter © 2007 Elsevier B.V. All rights reserved.doi:10.1016/j.chemgeo.2007.12.002

mailto:[email protected]

http://dx.doi.org/10.1016/j.chemgeo.2007.12.002

130 L. Jin et al. / Chemical Geology 249 (2008) 129–154

Based on water chemistry and discharge, bulk reaction rates of calcite, dolomite (Ca0.5Mg0.5CO3), K-feldspar and plagioclaseare calculated to be at 3400, 3100, 220, and 320 mol ha−1 yr−1, respectively. Based on mass balance of soil composition, long-termplagioclase-weathering rates (over the past 12,500 years) are calculated at about 2400 mol ha−1 yr−1, much higher than the currentrates. This agrees with previous conclusions that weathering rates decrease with time, due to loss of reactive mineral surfaces.Furthermore, both long-term and short-term plagioclase dissolution rates in Michigan are relatively high compared to those in otherwatersheds with similar age, possibly due to fresh surfaces produced by glaciation, in combination with the high discharge and highplagioclase abundances.© 2007 Elsevier B.V. All rights reserved.

Keywords: Carbonates; Aluminosilicates; Soil water; Reaction kinetics; Carbon dioxide; Hydrological tracers

1. Introduction

Anthropogenic CO2 emissions (~8 Gt C/yr) haveraised the atmospheric CO2 partial pressure significantly(e.g., Schlesinger, 1997; Burgermeister, 2007). The draw-downof this additional atmospheric CO2 on decadal scalesdepends on the response of surficial inorganic and organiccarbon reservoirs, including ocean–atmosphere exchange,biotic assimilation on continents and in oceans, and short-term uptake by carbonate-mineral weathering (Sundquist,1993; Schlesinger and Lichter, 2001). On geologic time-scales, the paleo-atmospheric CO2 record suggests thatinput and output fluxes of Earth-surface carbon have beenin close balance (Berner, 1991, 1994; Berner and Caldeira,1997). The stabilization of atmospheric CO2 requires anegative feedback, and the most likely mechanism is CO2

consumption during silicate weathering (Walker et al.,1981; Bickle, 1996; Edmond and Huh, 1997; Boeglin andProbst, 1998; Gaillardet et al., 1999; Quade et al., 2003;Ridgwell and Zeebe, 2005). Thus, it is geologically signif-icant to understand how silicate- and carbonate-mineralweathering responds to changes in atmospheric CO2 levelsas well as global climate, especially changes in air temper-ature and precipitation.

Riverine chemistry has proven powerful in calculatingchemical weathering rates and understanding their depen-dence on mineralogical and climate variables. White andBlum (1995) compiled water chemistry data for riversdraining granite/granitoid watersheds and observed onlyweak correlations between field-derived weathering ratesand mean annual temperature and precipitation of thewatersheds. Our group has studied chemical weatheringin the Great Lakes region, a recently glaciated area withmixed mineralogy (carbonate and silicate minerals), pro-viding the opportunity to investigate the correlation be-tween climate change and chemical weathering processesin mid-continent settings (Fig. 1A; Ku, 2001; Williamset al., 2007b; Szramek et al., 2007). Ground waters andriver waterswere characterized in severalMichiganwater-sheds (e.g., the Huron and Kalamazoo watersheds shown

in Fig. 1A) to evaluate the weathering fluxes and carbondioxide consumption rates (Williams et al., 2007b;Szramek et al., 2007). The Great Lakes region wasshown to be a significant CO2 sink due to intensivecarbonate dissolution, which positively correlated withhigh runoff and carbonate-mineral solubility. Specifically,dolomite is more soluble than calcite in temperate-climateregions and, contrary to past estimates, contributes morethan half of cations and dissolved inorganic carbon (DIC)on a global scale (Berner and Berner, 1995; Williamset al., 2007b). Contribution from silicate dissolution isdifficult to evaluate from elemental relations in surfacewaters, as carbonate dissolution dominates the waterchemistry if carbonate minerals are present. Furthermore,elemental budgets can be confounded by contaminantssuch as deicing salts (CaCl2, NaCl), especially in themoredensely populated areas of southern Michigan (Williamset al., 2007b). Soil water cation chemistry, althoughaffected by precipitation composition, biological activity,and cation exchange processes, is predominantly con-trolled by the type and rate of chemical weatheringreactions. In near-surface soils, carbonate minerals havebeen preferentially removed, permitting silicate reactionsto be quantified directly. A recent study investigated soilwaters in Michigan soil profiles, focusing on cation massbalance and the silicate versus carbonate weatheringcontributions (Jin et al., 2008). Consistent with surfacewater studies, dolomite dissolution contributes morethan 90% of soil water Mg2+ and more than 40% ofsoil water Ca2+. Furthermore, silicate weathering wasshown to be dominantly controlled bymineral abundances(surface area) while carbonate weathering was controlledby soil zone CO2.

Despite the constraints these studies have placed, theactual determination of chemical weathering rates andtheir dependence on temperature and precipitation washard to quantify by collection and analyses of soil watersfrom natural soil columns, because the frequency ofsampling is limited by dry or frozen field conditions. Inthis study, mineral weathering reactions were investigated

http://dx.doi.org/10.1016/j.gca.2007.12.007

Fig. 1. Location map of the study area: the Kalamazoo watershed in the Great Lakes region (A); Kellogg Biological Station (KBS) Monoliths in theKalamazoo watershed (B); Schematic of the monolith with arrows indicating water-flow directions (C). The Kalamazoo and the Huron watersheds havesimilar climate, soil age and mineralogy. Chemical weathering in the natural soils of the Huron watershed has been previously studied (Jin et al., 2008).

131L. Jin et al. / Chemical Geology 249 (2008) 129–154

in four replicate soil monoliths at a Long-Term EcologicalResearch (LTER) site ofKelloggBiological Station (KBS,Michigan State University), situated in the Kalamazoowatershed of southwestern Michigan. The soil monolithsare intact soil profiles, enclosed by walls, allowingsampling and quantification of gravity infiltration exitingthe bottom of the profile. The soil mineralogy at KBSwascharacterized and the soil solute and gas chemistry wasfollowed over seasonal cycles. A LiBr tracer experimentwas conducted to understand the hydrologic flow in thesoils and to calculate the residence time of water andsolutes in the soil profiles. The goals of this study were:(1) to elucidate the relative importance of thermodynamicversus kinetic controls on carbonate- and silicate-mineralweathering; (2) to investigate seasonal variation of chemi-cal composition of soil pore waters and soil gas chemistryin relation to temperature and soil moisture; (3) to studythe hydrologic controls on silicate weathering by measur-

ing residence time of a conservative tracer in the soilprofiles; (4) to calculate the infiltration fluxes of majorcations and the present reaction rates of major minerals(calcite, dolomite, feldspars) in the soil profiles; (5) tocalculate long-term plagioclase weathering rates by soilelemental mass balance since soil formation; and (6) tocompare short-term and long-term plagioclase weatheringrates.

2. Field site description and methodology

2.1. Soils and geology

Southwestern Michigan soils are developed on highlyheterogeneous glacial deposits left upon retreat of theWisconsin ice sheet (ca. 12,500 years BP) (Dorr andEschman, 1970). The average thickness of glacial depo-sits in this region is about 100 m atop Mississippian age



bedrock units as the Coldwater Shale and sulfide-richMarshall Sandstone (Catacosinos andDaniels, 1991). Theglacial drift originated from long-distance transport oferosion products from both the crystalline CanadianShield and Paleozoic/Mesozoic sedimentary units of theMichigan Basin. The KBS field site is located on anoutwash plain in the Kalamazoo watershed (Fig. 1B).Soils here are sandy loams of the Kalamazoo series (fine-loamy, mixed, mesic Typic Hapludalfs) (Rosek, 1992;Hamilton et al., 2007). These well-drained soils lie onnearly level slopes and are moderately permeable in theupper part of the profile and very permeable in the lowerpart. The typical sequence of horizons is Ap, E, Bt1, 2Bt2and 2E/Bt. Textures of the Ap and E horizons are loamor sandy loam. The Bt1 is usually clay loam or sandyclay loam,whereas the 2Bt2 has a sandy loam texture. TheE/Bt consists of lamellae of loamy sand (Bt) and sand (E).

2.2. Climate and precipitation chemistry

At KBS, mean annual temperature is 9.7 °C, and meanannual precipitation is 890 mm/yr over the last 30 years,with about half falling as snow. KBS maintains dailyrecords of air temperatures, precipitation (as rain andsnow), and snow depth on ground. Daily precipitationwas calculated by adding the rain and snowmelt togetherassuming that 1-inch of snow is equivalent to 0.02-inch ofrainfall. This assumption is an oversimplification sincethe snow/rain conversion factor varies with history andaging of the snow, but from a weekly to monthly waterbudget perspective, error introduced by this assumption isrelatively small. Total precipitation was calculated to be1145 mm in the calendar year of 2004, which is about20% above the annual average. Monthly Thornthwaitepotential evapotranspiration (PET) was computed basedon weather data collected at KBS from 2004; this indexhas proven to be fairly reliable in the Midwestern USA(Thornthwaite, 1948; Palmer and Havens, 1958; Crumet al., 1990). KBS also has PET record extending over thelast 30 years.

Major ion chemistry of wet atmospheric depositionwas monitored weekly by the National AtmosphericDeposition Program/National Trends Network at KBSwithin 1 km of the study sites (NADP/NTN, 2005). Theannual average composition of wet precipitation here is:NO3

− (26 μM), SO42− (18 μM), NH4

+ (27 μM) and Ca2+

(5 μM), and pH averages about 4.5.

2.3. Monoliths and tracer experiments

The four replicate soil monoliths (identified here as #2,#6, #9 and #13) are 20 m apart in an area with negligible

topographic slope. Each was installed in 1989 byexcavating around a 1.22 m×2.29 m×2 m pedon andthen enclosing it within a stainless steel chamber. The areaaround the chamber was then backfilled, so that installa-tion was flush with the surrounding soil surface (Fig. 1C).Gravity drainage waters are collected continuously in areceiving bucket at the base of eachmonolith, at a depth of200 cm. A chamber on one side provides access tosamplers. The monoliths are sealed by walls but open tonatural precipitation at the top so that water discharge atthe base equals to the difference between precipitation andevapotranspiration. In addition to the gravity drainage atthe bottom, water was collected from 4 to 6 Prenart®tension soil-solution samplers (lysimeters) at shallow soildepths (between 20 and 100 cm). Inmonoliths #9 and #13,three replicate lysimeters were installed at a depth of180 cm, just 20 cm above the monolith base. Tension(~0.5 atm) was applied to the lysimeters, and the resultingsoil water was collected after 24 h. Since installation,the surfaces of the monoliths have been harvested andcultivated annually (rotations of corn/soybean/wheat) butwithout application of liming or fertilization.

Soil water transport through the profile was investi-gated by applying a tracer to snow on the soil surface.LiBr was chosen because of the low background concen-trations in soils and low toxicity to plants (Jabro et al.,1991; Thies et al., 2002). The chemical behavior of Br− isconservative, as soil pH is around 5.5 to 6, which limitsany significant Br− sorption to clays. To establish baselineconcentrations for Li+ andBr− tracers, the drainagewaterswere sampled before the application of the tracer. About60 g of laboratory-grade LiBr powder was dissolved in20 L of melted snow, and then 5 L of this tracer solutionwas evenly spread over each monolith with a gardensprayer. LiBr solution was applied to coincide as closelyas possible with the onset of a major snowmelt event(27 Feb 2004). Gravity drainage samples were collectedapproximately weekly for the first 3 months after tracerinjection, followed by less frequent sampling of thelysimeters at all depths and the gravity drainage at the baseof the monoliths. Two types of gravity drainage waterswere collected at the base (200 cm): the water that wasactually dripping from the monolith drainage pan on eachsampling date and the water that had accumulated overtime in the receiving bucket. We will hereafter refer to theinstantaneous sample as the “drip” sample and theintegrated sample of drainage between sampling eventsas the “tub” sample. The volumes of drainage in thebuckets were measured, the instantaneous dripping ratesrecorded, and then the buckets were emptied for the nextsampling interval. Average water-flow rates were calcu-lated from 1 Mar to 10 Jun 2004.


2.4. Collection and analyses of soil, soil gas and soilwater samples

Five gas sampling tubes were installed at depthsranging from 20 to 100 cm in each monolith and sampleswere collected multiple times over the year to investigateseasonal variations. Soil gas was withdrawn by syringefrom Teflon capillary tubing (0.5 mm ID) extending atleast 50 cm into the monolith from the side wall. Partialpressures of soil gas CO2 (pCO2) were measured by gaschromatography (GC) with thermal conductivity detec-tion, calibrated with 4 standards ranging from 500 to10,000 ppmv CO2.

Soil cores up to 200 cm deep were collected with apneumatic coring device about 5 m away from eachmonolith, and the cores were sectioned every 10 cm. Soilsamples were dried at 45 °C and ground to pass a sieve ofmesh 200 (75 µm diameter). Mineralogy of these sampleswas determined by Scintag powder X-ray diffraction(XRD) in the ElectronMicrobeamAnalysis Lab (EMAL)at the University of Michigan. Thin sections of parentsediments were examined by scanning electron micro-scope (SEM) at EMAL to identify the trace minerals andto characterize the features of mineral surfaces. Sedimentsless than 2 μM in diameter (clay fraction) were separatedusing sedimentation method and analyzed for mineralogyby XRD. Representative bulk soils were leached by aquaregia (3HCl:1HNO3) at room temperature for 3 h toselectively dissolve carbonates and Al/Fe oxyhydroxides(Sparks, 1996). The acid leachate provides the bestestimate of the carbonate content (dolomite and calcite) asit has been found to remove negligible amounts of cationsfrom silicate minerals in these soils (Jin, 2007). Anothersoil core was collected in a mature deciduous forest nearthe monolith site. Elemental composition of representa-tive bulk soils from this profile was measured by X-rayfluorescence (XRF, Rigaku SMAX) at theMichigan StateUniversity, following the method of Hannah et al. (2002).

Water temperature was measured in the field with atemperature meter. The pH values of the soil waters weremeasured in the field immediately after sample collectionwith a Corning 315 pH meter with an Orion Ross glasscombined electrode. The electrode was calibrated withtwo low-ionic-strength standard buffer solutions (pH 4.10and 6.97), and the pH measurement precision is 0.05 pHunits. Water for chemical analyses was filtered through0.45-μmWhatman glass-fiber filters and stored in HDPEbottles at 4 °C. For cation samples, about 30-mL filteredsolution was collected and acidified to pH 4 with ultra-pure concentrated HNO3. Major cations and anions wereanalyzed by inductively coupled plasma-optical emissionspectrometry (ICP-OES) and ion chromatography (IC),

respectively, in the Environmental and Analytical Geo-chemical Lab (EAGL) at the University of Michigan. Theprecision of IC and ICP analyses was better than 3% formajor elements and 10% for minor elements. Totalalkalinity was determined by weak acid titration within aday of sampling over a pH range of 2 to 3 using the Granmethod (Edmond, 1970; Gieskes and Rogers, 1973;Stumm andMorgan, 1996). The uncertainty for alkalinitytitrations is less than 2% for most of our samples, exceptfor those with very low alkalinity values where uncer-tainty is 0.05 meq/L.

Mineral saturation state calculations were performedfor soil water samples where complete geochemical ana-lyseswere available (pH, temperature, alkalinity andmajorion concentrations) using the USGS program Solmi-nEQ.88 (Kharaka et al., 1988). Saturation indexes (SI) ofthe solution with respect to calcite are defined as SI=Log([Ca2+]⁎ [CO3

2−]/Kcalcite), where [Ca2+] and [CO32−] are

activities of Ca2+ and CO32−, and Kcalcite is the solubility

constant of calcite. The normalized saturation state is usedto compare the ionic activity product (IAP) of dolomiteCa0.5Mg0.5 (CO3) on the samemolar basis as that of calcite(CaCO3). Thiswas accomplished by taking the square rootof IAPdolomite from SolminEQ.88 and the thermodynamicdata for dolomite stability from Hyeong and Capuano(2001) were used to evaluate the saturation index ofintermediate-order dolomite. Given the uncertainties in pHmeasurement, the degree of saturation is accurate within0.3 SI units.

3. Results

3.1. Soil mineralogy

The major minerals identified by XRD in parentmaterials of monoliths were quartz, K-feldspar, calcite,dolomite, plagioclase, and amphibole in the order ofdecreasing abundance, with trace amounts of mica,magnetite, ilmenite, and pyrite. Clay fractions of parentsediments were dominated by quartz, with little illite andkaolinite. Plagioclase composition of the glacial depositshad been previously reported as Na0.8Ca0.2Al1.2Si2.8O8

throughout Michigan watersheds (Ab of 0.80±0.10;Williams et al., 2007a). Elemental compositions of theacid leachable fractions and bulk soils are reported inTable 1. Completely leached from the shallow soils,calcite and dolomite co-occurred in the soil zones ofmonoliths and deciduous forest beginning at around150 cm depth. Precipitation of secondary Al/Fe oxyhy-droxides produced a visible red-brown layer in theB-horizon, generally just above the top of the carbonate-bearing layer. The carbonate fraction ranges from 5 to

Table 1Elemental composition of the acid leachable fractions and bulk soils at KBS

Acid leachable

Deciduous forested site: Monolith soil core:

Depth (cm) Mg (mmol/kg) Ca (mmol/kg) Calcite wt.% Dolomite wt.% Depth (cm) Mg (mmol/kg) Ca (mmol/kg) Calcite wt.% Dolomite wt.%

10 21.4 31.1 ML #6 135 31.4 87.320 22.4 24.5 145 737.5 1991.7 12.5 13.645 28.2 23.9 155 810.1 1886.4 10.8 14.975 28.9 10.8 165 622.4 1804.1 11.8 11.5100 11.8 4.8 175 743.6 1766.3 10.2 13.7150 37.7 9.2 183 577.1 1213.5 6.4 10.6175 194.3 413.8 2.2 3.6 190 371.2 1126.2 7.6 6.8190 845.0 2685.0 18.4 15.5 ML #9 175 498.4 1307.9 8.1 9.2

184 612.1 1558.6 9.5 11.3193 288.5 977.0 6.9 5.3

Elemental composition by XRF at deciduous forested site

Depth(cm)

Density a

(g/cm3)Zr(ppm)

SiO2

wt.%TiO2

wt.%Al2O3

wt.%MgOwt.%

CaOwt.%

Na2Owt.%

K2Owt.%

Zr/TiO2⁎1000 Zw/Zp

b

10 1.6 358 80.35 0.53 7.67 0.62 0.79 0.86 1.89 0.68 0.220 1.6 365 81.56 0.54 7.45 0.43 0.47 0.86 1.90 0.68 0.245 1.7 348 80.95 0.56 8.13 0.50 0.46 0.83 1.98 0.62 0.275 1.8 112 84.90 0.20 7.21 0.52 0.50 1.00 1.53 0.56 0.5100 2.0 103 91.66 0.12 4.21 0.22 0.34 0.76 1.10 0.86 0.5150 2.0 109 84.81 0.23 6.76 0.58 0.65 1.13 1.38 0.47 0.5175 2.3 63 84.29 0.16 5.69 1.15 2.35 1.19 1.12 0.39 0.7190 c 2.3 47 61.76 0.12 4.39 3.28 13.79 0.97 0.96 0.39 1.0a Soil density data are from LTER data depository at KBS.b This ratio reflects volume change during soil development, and if less than 1, indicating soil volume loss.c The soil at 190 cm is used as representative parent material in long-term weathering rate calculation.


30 wt.% in the soils, averaging at 20 wt.% (Table 1).Dolomite and calcite occur in approximately equal massin the carbonate soil zones, similar to previously studiedsoil profiles in Michigan (Jin, 2007). Also, soil textureswere similar between the soil monolith and deciduousforest sites, being loamy in the shallow soils, sandybelow ~100 cm depth, and much coarser in the C-horizon (below 150 cm). Weight percentages of MgOand Na2O in the bulk soils are linearly correlated with theabundances of amphibole and plagioclase where carbo-nate is absent. The KBS site has similar soil mineralogy

Table 2Basic climate information at KBS in the year of 2004

Unit Jan Feb Mar Apr M

Precipitation (P) mm 62 38 118 11 25Mean monthly temperature °C −6 −2 6 12 1PET1 mm 0 0 5 41 8PET2 mm 0 0 19 55 9Discharge (P-PET1) mm 62 38 113 −30 16

PET1 is potential evapotranspiration averaged from 30-year record.PET2 is potential evapotranspiration calculated from Thornthwaite equationMonthly precipitation and temperature data are from a National Weather Se

to the soils in the Huron watershed, but with different soiltextures (Jin et al., 2008). The organic matter contentdecreased with depth in each monolith but reached up to2 wt.% of dry weight at the uppermost horizons of thesoil (Jin, 2007).

3.2. Water budget

Data on mean air temperature, total precipitation,Thornthwaite PET, long-term mean PET at KBS(30 years), and water infiltration (difference between

ay Jun Jul Aug Sep Oct Nov Dec Total

2 100 80 169 44 85 112 74 11457 20 23 21 19 12 6 −13 140 180 142 80 40 10 0 7215 119 140 114 95 45 16 0 6989 −40 −100 27 −36 45 102 74 424

(Thornthwaite, 1948).rvice station at KBS.


Fig. 2. Variation of the mean monthly precipitation (P), air temperature(T) and potential evapotranspiration (PET) at KBS in year 2004. PET1is monthly PET averaged from 30-year record at KBS, while PET2stands for monthly PET calculated following Thornthwaite (1948).


precipitation and PET) are reported in Table 2. MonthlyPET values show a seasonal cycle typical of temperateregions (Fig. 2), with PET exceeding precipitationthroughout the warmer months while precipitationexceeds PET during the cooler months. Thus, rechargewould normally be significant only in early spring andlate fall, although it could also occur during a mid-winterthaw or after a particularly heavy rainfall in the summer.Negative infiltration values in Table 2 indicate a net

Fig. 3. Br− concentrations of soil solutions collected from soils less than 2B: monolith #6, C: monolith #9, and D: monolith #13). Breakthrough peaks alater in deeper soils as shown in monolith 6. Three replicate tension samplersrelease, suggesting that flow paths through the soils are heterogeneous.

loss of soil water to evapotranspiration, assuming thatthe plant canopy was fully developed so that evapo-transpiration would reach its potential maximum. Thebulk density of the Kalamazoo soils is around 1.7 g/cm3

(LTER site at KBS, unpublished data), which wouldrequire an approximate porosity of 30% given themineral assemblages. Thus, the monolith is a large waterreservoir (around 1600 L of water if saturated) and has abuffering capacity for the water budget. After a lowmoisture period, inputs by precipitation would haveto re-wet the soils before any significant infiltrationwould commence. In 2004, an estimated 424 mm out of1145 mm of precipitation (~37%) infiltrated the soils,and reached the water table to recharge the aquifer,ultimately discharging to streams. Water budget studieson a nearby sub-catchment within the Kalamazoo water-shed showed that 580 mm of the annual precipitation(890 mm) was evapotranspired back to the atmosphere,with the rest (~35%) discharged by streams (Allen et al.,1972; Rheaume, 1990). Calculations of upland soil waterbalance from these two perspectives thus yield consis-tent annual evapotranspiration to precipitation ratios of~65%.

00 cm as a function of time after tracer application (A: monolith #2,rrived soon after the application of Br− in the surface soils and arrivedat 180 cm deep at monoliths # 9 and #13 have different patterns of Br


3.3. Conservative solute tracer experiment

The downward transport of bromide through the fourmonoliths was followed for several months after itsapplication. Li+ was not detectable in any of these soilpore waters, in contrast to Br−. Both clay minerals andsolid organic materials in the soils have negative surfacecharges, which permit the Li+ adsorption, while anionssuch as Br− would not be affected. Compared to theconcentrations resulting from the applied tracer solution,the baseline Br− concentrations can be considerednegligible. Patterns of Br− concentrations over timefor each monolith are shown in shallow soils (less than200 cm; Fig. 3) and in drainage at the bottom (at 200 cm;Fig. 4). Even though the number of shallow soil watersamples was limited by dry conditions, a general trendcould be discerned. The tracer moved through theshallow soil layers (~25 cm–30 cm) very quickly, asevidenced by early sharp peaks in Br− concentrations(Fig. 3), consistent with the very permeable nature of thesandy shallow soils. In monolith #6, Br− concentration

Fig. 4. Br− concentrations of drip (solid) and tub (open) samples collected at t(A: monolith #2, B: monolith #6, C: monolith #9, and D: monolith #13). DripBr− release curves differ among the four monoliths. Insets show how the Br− c100 days after tracer application. In monolith #2, Br− decreased with time reincrease in Br− concentrations was observed after rainfall events.

breakthrough peaks were observed from 4 depths insequential order from shallow to deep (Fig. 3B).

The patterns of transport of Br− into deeper soilhorizons (below 100 cm) differed markedly among themonoliths. Even within a given monolith, soil waterssampled at the same depth (A, B and C at 180 cm) showeddifferent patterns of Br transport over time (Fig. 3D).These differences were likely driven by heterogeneoussoil flow paths. Arrival of Br− peaks at the depth of200 cm in the four monoliths exhibits complex temporaland spatial variation (see Fig. 4). The most rapid Br−

transport was observed in monolith #2. Concentrations ofBr− in monolith #13 showed an early peak immediatelyfollowing tracer addition, and then decreased sharply.Following that, Br− concentrations slowly increased againand showed another peak around 100 days after tracerapplication. In contrast, Br− concentrations increasedslowly in monoliths #6 and #9 and did not show peaksuntil 100 and 150 days, respectively, after the additionof the tracer. Daily precipitation data are presented in theinsets of Fig. 4. Except formonolith #2,Br− concentrations

he bottom of each monolith as a function of time after tracer applicationand tub samples show the same trend in each monolith even though theoncentrations of drip and tub respond to daily precipitation over the firstgardless of precipitation intensity. However in monolith #6 and #9, an

Fig. 5. Alkalinity versus Ca2++Mg2+ in soil solutions from KBS (A).Carbonate weathering dominates soil water composition in the deep soilcolumn, producing 2:1 ratio (dashed line) between Ca2++Mg2+ concen-trations and alkalinity. Shaded area is the range of soil waters and groundwaters in the Huron watershed (Jin et al., 2008). Soil water Mg2+

concentrations versus Ca2+ concentrations at KBS (B). The broken line inB has Ca2+/Mg2+ ratio of 3, indicating 1:1 bulk dissolution rates betweendolomite and calcite. Soil water Na⁎ concentrations versus silica concen-trations at KBS (C). The Na⁎/silica ratios of most soil waters are greaterthan 1/3, indicating of silica loss to secondary mineral precipitation.Shallow (solid) and deep (open) samples are defined by soil waterswithout and with the influence of carbonate dissolution.

Fig. 6. Ca2+ concentrations of drip (solid) and tub (open) samplescollected at the bottom of each monolith, (A: monolith #2, B: monolith#6, C: monolith #9, and D: monolith #13) show clear seasonalvariations with maxima in the summer and early fall and minima in thewinter and early spring. Drip and tub samples have very similar Ca2+

concentrations except in the summer months, suggesting that Ca2+,largely a product of carbonate dissolution, is not diluted by varyinginfiltration because carbonate-mineral equilibrium is rapidly attained.Therefore, the difference of Ca2+ concentrations between drip and tubsamples in the summer likely results from CO2 degassing and possibleprecipitation of calcite after the water exits the soil monoliths.


in the monoliths were sensitive to precipitation intensity.For example, between day 30 and day 60, there was norain and Br− concentrations remained unchanged. Afterday 60, heavy rain fell at KBS, and thereafter soil water

Br− concentrations increased in the drainage water (seeinsets of Fig. 4).

3.4. Soil water chemistry and soil zone pCO2

Chemical composition of soil waters from each mono-lith is reported in Suppl. Table 1, together with Na⁎ con-centrations and saturation indices with respect to calcite.The soil water data can be evaluated by ionic stoichio-metry because soil water samples dominantly influencedby carbonate dissolution should lie on 2:1 relationship



between concentrations of major divalent cation (Ca2++Mg2+) and alkalinity in equivalents (Fig. 5A). The soilwaters and ground waters from the neighboring Huronwatershed in southeast Michigan (Jin et al., 2008) are alsoplotted in Fig. 5A for comparison. Inter-laboratory dataquality comparison was made of analyses of replicatesamples at the EAGL lab at the University of Michiganand Dr. S. Hamilton's lab at KBS; over 90% of the majorelement concentrations agreed within 3%. The shallowsoil waters (b100 cm) were very dilute and alkalinity wasless than 200 µeq/L. The dominant cations of these soilwaters are Mg2+, Ca2+, K+, and Na+ with soil water Ca2+

concentrations were less than 600 μM andMg2+ less than300 μM (Fig. 5B). The pH values of soil solutions weregenerally lower than 7 in the shallow layers and increasedto 7–8 below 150 cm depth. Soil waters at 180 cm

Fig. 7. The soil gas pCO2 (A), soil water (drip water, 200 cm) Mg2+ (B), Mg2+/time for each monolith.

collected by Prenart lysimeters and drip and tub samplesat 200 cm were Mg2+–Ca2+–HCO3

− dominated water(Fig. 5B; up to 700 μM Mg2+ and 2500 μM Ca2+). TheCa2+ concentrations in drainage water from each of themonoliths show a similar seasonal trend, with maxima inSeptember and minima in April (Fig. 6). Drip and tubsamples collected on the same day had very similar Ca2+

concentrations during most of the year, except in thesummermonthswhen the soilmoisturewas low andwaterdischarge was minimal. The Mg2+ concentrations in deepsoil solutions showed the same trend as Ca2+ (Fig. 7B). Inearly spring, soil water Mg2+ and Ca2+ concentrationswere about 300 μM and 1000 μM, respectively, while insummer and early fall months, Mg2+ and Ca2+ concentra-tions almost doubled. In contrast, Mg2+/Ca2+ ratios wereinvariant throughout the year at around 0.3 (Fig. 7C). In

Ca2+ molar ratios (C), H4SiO40 (D), Na⁎ (E), and Cl− (F) as a function of



these soils, the dominant sources of Mg2+ and Ca2+ werechemical weathering of carbonate minerals; Mg2+/Ca2+

ratios in the soil solutions reflected the relative contribu-tion of calcite and dolomite dissolution to these twodivalent cations. Mg2+/Ca2+ ratios of 0.3 would requireroughly equal bulk molar contributions of calcite anddolomite dissolution.

Na⁎ is defined as Na+ concentrations in soil solutionsafter correcting for the atmospheric inputs according tothe following equation:

Na⁎ ¼ Naþ−½Cl−�ðNaþ=Cl−Þprecipitation� ð1Þ

In contrast to divalent cations, the shallow versusdeeper soil waters do not significantly differ in Na⁎ andsilica (H4SiO4

0) concentrations (Fig. 5C) with Na⁎/silicamole ratios less than 3. Seasonal variations of H4SiO4

0,Na⁎, and Cl− concentrations in drainage soil solutionsare plotted in Fig. 7D, E and F. Silica exhibits a patternsimilar to divalent cations (Fig. 7D), with concentrationsranging from 60 to 140 μM. The Cl− concentrationswere elevated somewhat in February and March. Afterthat, Cl− concentrations decreased markedly andremained constant at around 20 μM for the rest of the

Fig. 8. NO3− (A) and SO4

2− (B) concentrations as a function of time indrainage water (drip samples) from each monolith, following thecalendar year rather than month as in Fig. 7. Shaded area is the range ofsoil water NO3

− concentrations from the Huron watershed, where studysites are in natural forests (Jin et al., 2008).

year (Fig. 7F). Focusing on sodium data obtained afterCl− concentrations had stabilized, Na⁎ concentrationsincrease gradually from April and reach a peak value of140 μM in July, and decline to around 60 μM (Fig. 7E).

In addition to carbonate alkalinity, NO3− and SO4

2−

can also contribute significantly to the anion budget insoil solutions, with NO3

− concentrations ranging from50 to 500 μM and SO4

2− from 100 to 200 μM. The NO3−

and SO42− concentrations in drip and tub samples are

plotted with calendar year (Fig. 8A and B) because ofthe association of these anions with annual agriculturalrotation at the sites. The NO3

− and SO42− concentrations

show different temporal trends; with SO42− concentra-

tions remaining almost constant over the year whileNO3

− concentrations steadily decrease with time over theduration of this study.

Soil pCO2 at KBS was higher than the atmosphericCO2 partial pressure at all depths and on all collection dates(Table 3). Soil pCO2 ranged from 1000 to 18,000 ppmvand fluctuated by about an order of magnitude over time atany particular sampling point. The soil pCO2 generallyincreased with depth then remained constant at each soilprofile, clearly indicating the effect of gaseous diffusion toatmosphere (Reardon et al., 1979; Bacon and Keller,1998). Soil pCO2 values at about 100 cm depth (thedeepest gas samples available) are plotted in Fig. 7A,showing a strong correlation of pCO2 with season. In thedeeper soil layers of KBS, pCO2 was about 50 times ashigh as atmospheric CO2 in the summer. Even in thewinter, soil pCO2was still 5 times higher than atmosphericCO2 partial pressures, indicating that frozen ground andsnow cover may impede gas exchange and maintain highpCO2 values even in the absence of active respiration.

4. Discussion

4.1. Seasonal variation in solute concentrations

The complex nature of temperature and soil moisturecontrols on mineral weathering has been increasinglyrecognized (Velbel, 1993; Richards and Kump, 2003;Anderson, 2005). Temperature regulates dissolution ratesaccording to the Arrhenius equation (a kinetic factor),but also modifies the solubility constants of minerals(a thermodynamic equilibrium factor). Temperature alsoexerts indirect controls on chemical weathering by modi-fying the biological activity, therefore the soil zone pCO2

and biological sources ofmineral acidity (e.g., respiration,nitrification). Precipitation dilutes the soil solutes andflushes water out of soil zones, thereby controlling dis-solution rates through chemical affinity (saturation state)effects. Closer inspection of seasonal variations in soil


Table 3Soil gas CO2 concentrations at four monoliths of KBS

Samples pCO2

(ppmv)Log(pCO2)

Samples pCO2

(ppmv)Log(pCO2)

Samples pCO2

(ppmv)Log(pCO2)

Samples pCO2

(ppmv)Log(pCO2)

Monolith #2 Monolith #6 Monolith #9 Monolith #13

21a2: 25 cm 61Ap4: 25 cm 91Ap1: 25 cm 131Ap1: 30 cm11/25/03 1971 −2.71 11/25/03 3328 −2.48 11/25/03 2688 −2.57 11/25/03 904 −3.043/31/04 4337 −2.36 3/31/04 2728 −2.56 3/31/04 2292 −2.64 3/31/04 1071 −2.976/10/04 588 −3.23 6/10/04 11799 −1.93 6/10/04 8717 −2.06 6/10/04 10121 −1.997/27/04 10024 −2.00 7/27/04 10739 −1.97 7/27/04 4482 −2.35 7/27/04 4566 −2.3410/21/04 5198 −2.28 10/21/04 5997 −2.22 10/21/04 6370 −2.20 10/21/04 3990 −2.4012/22/04 2070 −2.68 12/22/04 1651 −2.78 12/22/04 1667 −2.78 12/22/04 555 −3.264/15/05 2068 −2.68 4/15/05 2599 −2.59 4/15/05 2665 −2.57 4/15/05 807 −3.095/5/05 2447 −2.61 5/5/05 2112 −2.68 5/5/05 1647 −2.78 5/5/05 1894 −2.728/26/05 4968 −2.30 8/26/05 6288 −2.20 8/26/05 5838 −2.23 8/26/05 4364 −2.36

22b2: 51 cm 62Btt1: 40 cm 92: 40 cm 132b2: 60 cm11/25/03 1906 −2.72 3/31/04 3197 −2.50 7/27/04 6608 −2.18 11/25/03 1001 −3.003/31/04 3997 −2.40 6/10/04 13233 −1.88 10/21/04 1903 −2.72 3/31/04 1750 −2.766/10/04 22053 −1.66 7/27/04 11326 −1.95 12/22/04 637 −3.20 6/10/04 17629 −1.757/27/04 11959 −1.92 10/21/04 6181 −2.21 4/15/05 825 −3.08 7/27/04 14316 −1.8410/21/04 7872 −2.10 12/22/04 2296 −2.64 5/5/05 1135 −2.94 10/21/04 6540 −2.1812/22/04 2591 −2.59 4/15/05 2820 −2.55 8/26/05 3229 −2.49 12/22/04 1455 −2.844/15/05 2938 −2.53 5/5/05 2074 −2.68 4/15/05 2296 −2.645/5/05 3223 −2.49 8/26/05 6177 −2.21 5/5/05 2160 −2.678/26/05 9259 −2.03 8/26/05 8933 −2.05

23c1: 63 cm 63c2: 55 cm 93BtB1: 53 cm 133BtB1: 68 cm3/31/04 4198 −2.38 11/25/03 4345 −2.36 11/25/03 3226 −2.49 11/25/03 2069 −2.686/10/04 21837 −1.66 3/31/04 3599 −2.44 3/31/04 3057 −2.51 6/10/04 14752 −1.837/27/04 12605 −1.90 6/10/04 5799 −2.24 6/10/04 15284 −1.82 7/27/04 10541 −1.9810/21/04 8548 −2.07 7/27/04 15595 −1.81 7/27/04 14472 −1.84 10/21/04 5777 −2.2412/22/04 2868 −2.54 10/21/04 7773 −2.11 10/21/04 8360 −2.08 12/22/04 903 −3.044/15/05 3489 −2.46 12/22/04 1988 −2.70 12/22/04 2383 −2.62 4/15/05 2404 −2.625/5/05 3392 −2.47 4/15/05 2136 −2.67 4/15/05 3193 −2.50 5/5/05 2858 −2.548/26/05 9077 −2.04 5/5/05 2785 −2.56 5/5/05 3168 −2.50 8/26/05 10766 −1.97

8/26/05 11069 −1.96 8/26/05 12675 −1.90

24d1: 76 cm 64: 66 cm 94B31: 61 cm 134BtB3: 81 cm11/25/03 1342 −2.87 3/31/04 3167 −2.50 11/25/03 3456 −2.46 11/25/03 1827 −2.743/31/04 3993 −2.40 6/10/04 13774 −1.86 3/31/04 2237 −2.65 3/31/04 595 −3.236/10/04 22666 −1.64 7/27/04 13720 −1.86 6/10/04 14124 −1.85 6/10/04 17586 −1.757/27/04 20474 −1.69 10/21/04 7530 −2.12 7/27/04 15587 −1.81 7/27/04 14133 −1.8510/21/04 8993 −2.05 12/22/04 1929 −2.71 10/21/04 8050 −2.09 10/21/04 6579 −2.1812/22/04 2830 −2.55 4/15/05 3133 −2.50 12/22/04 1848 −2.73 12/22/04 1528 −2.824/15/05 3595 −2.44 5/5/05 2516 −2.60 4/15/05 2514 −2.60 4/15/05 2376 −2.625/5/05 3489 −2.46 8/26/05 10273 −1.99 5/5/05 3036 −2.52 5/5/05 3107 −2.518/26/05 11281 −1.95 8/26/05 13246 −1.88 8/26/05 6198 −2.21

25e1: 82 cm 65e4: 74 cm 96: 89 cm11/25/03 2569 −2.59 11/25/03 2520 −2.60 11/25/03 3938 −2.403/31/04 3308 −2.48 3/31/04 2157 −2.67 3/31/04 2597 −2.596/10/04 20316 −1.69 6/10/04 11601 −1.94 6/10/04 13633 −1.877/27/04 17479 −1.76 7/27/04 15866 −1.80 7/27/04 15843 −1.8010/21/04 7433 −2.13 10/21/04 7941 −2.10 10/21/04 9052 −2.0412/22/04 1992 −2.70 12/22/04 1569 −2.80 12/22/04 2338 −2.634/15/05 3530 −2.45 4/15/05 2977 −2.53 4/15/05 3597 −2.445/5/05 3264 −2.49 5/5/05 2598 −2.59 5/5/05 3478 −2.468/26/05 9244 −2.03 8/26/05 14806 −1.83 8/26/05 17052 −1.77


Table 3 (continued)

Samples pCO2

(ppmv)Log(pCO2)

Samples pCO2

(ppmv)Log(pCO2)

Samples pCO2

(ppmv)Log(pCO2)

Samples pCO2

(ppmv)Log(pCO2)

Monolith #2 Monolith #6 Monolith #9 Monolith #13

67 g2: 95 cm11/25/03 2481 −2.613/31/04 1944 −2.716/10/04 10814 −1.977/27/04 16190 −1.7910/21/04 7832 −2.1112/22/04 1464 −2.834/15/05 2821 −2.555/5/05 2682 −2.578/26/05 10646 −1.97


water and gas chemistry allows identification of key reac-tion controlling factors.

To quantify the ion fluxes under different environ-mental conditions, one must consider whether naturalwaters have reached equilibrium with the primary orsecondary minerals. If chemical equilibrium is attained,the water composition would depend on the phases andthermodynamic properties of secondary minerals, andweathering fluxes should be a simple function of waterdischarge and elemental solubility. If, on the other hand,the system is far from chemical equilibrium, the evolu-tion of water composition will be rate-limited and varywith kinetic-related factors such as time, mineral surfacearea, mineral reactivity, pH regime, and chemicalaffinity. The responses of solute concentrations tovarying soil moisture conditions provide an indicationof kinetic versus equilibrium control on weatheringrates.

Some studies of stream water chemistry report powerlaw relationships between ion concentrations (such asCa2+ and Mg2+) and discharge, showing variations overtime as streams carry variable mixtures of precipitation,surface runoff, and shallow and deep ground water (e.g.,Hooper et al., 1990; Kirchner, 2003). Others have ob-served that concentrations of divalent cations vary littlewith discharge for carbonate mineral-dominated water-sheds (Williams et al., 2007b; Szramek et al., 2007). On alarger scale, differences in stream water chemistry couldalso reflect changes in relative contributions from dif-ferent areas with different lithology (e.g., Tipper et al.,2006). Solution chemistry in a soil depth profile, in con-trast, depends only on the direction and rate of chemicalreactions along a defined flow path influenced by arelatively stable mineralogical assemblage (Wolt, 1994).With frequent sampling of soil waters and dischargemeasurement from the soil monoliths, we aimed to un-derstand the chemical weathering mechanisms and tocalculate reaction rates.

About 37% of the 1145 mm of precipitation infiltratedthe soil monoliths andwas thus available for groundwaterrecharge (see Fig. 2 and Table 2). Given the dimensions ofthe monoliths, this would equal a discharge of 1185 L forthe year 2004. Field water-flow rates are listed for all fourmonoliths for the first 3 months of the tracer experiments(Table 4). Discharge peaks followed rainfall intensityclosely, with a lag time of less than 15 days. As expected,instantaneous and average flow rates can differ signifi-cantly, indicating variable discharge on even a weeklytimescale. Drip and tub soil water samples had very simi-lar Ca2+ and Mg2+ concentrations, except for the summermonths (Figs. 6 and 7B and Suppl. Table 1). The Ca2+ andMg2+ concentrations did not change with drainage dis-charge, suggesting that carbonate-mineral weatheringwasnot controlled by soil moisture or infiltration rates butby mineral solubility, consistent with results from riv-erine flux studies in carbonate-bearing watersheds (e.g.,Szramek et al., 2007). Concentrations of silica and othersilicate-weathering derived cations (such asNa⁎) (Fig. 7Dand E) decreased moderately during the major snowmeltevent in the middle February of 2004 but generally variedlittle with discharge. This suggests that instead of thepower law relationship reported for riverine silicate-weath-ering fluxes, chemical weathering in these soil profilesappears to be the first-order control on water chemistry,consistent with results from Clow and Drever (1996).

4.1.1. Na+ andCl− variation: atmospheric contributionsand evapoconcentration

Because Cl− is not a product of chemical weatheringof native soil minerals nor is it influenced by biologicalactivity, it can serve as a conservative tracer for hydro-logical processes (e.g., Quade et al., 2003; West et al.,2005). After the initial snowmelt input, concentrationsof Cl− in the four monoliths showed little seasonalvariability, demonstrating that evapotranspiration is notresponsible for the increases in concentrations of major

Table 4Water flow rates recorded in the first 3 months of the tracer experiments at KBS (unit: mL/min)

Date Monolith #2 Monolith #6 Monolith #9 Monolith #13

Instantaneous a Average b Instantaneous Average Instantaneous Average Instantaneous Average

3/1/04 60 7.1 0.5 0.4 1.2 0.4 120.0 1.33/2/04 – 31.9 – 0.4 34.0 33.6 c NA 18.63/4/04 16 12.2 c 10.4 7.7 17.2 12.2 c 12.8 12.2 c

3/5/04 10 11.7 9.6 9.8 14.0 11.7 12.0 10.23/9/04 7.2 6.8 c 8.8 6.8 c 8.5 6.8 c 7.0 6.8 c

3/11/04 5.4 5.5 5.2 6.0 5.4 5.6 5.4 5.83/15/04 8 3.8 5.6 3.8 4.0 3.7 4.4 4.03/16/04 1.2 3.9 3.6 3.9 3.2 3.1 3.6 3.13/22/04 b1 2.5 2.4 2.5 3.0 2.6 2.4 2.63/23/04 0.8 2.0 2.0 2.6 1.6 2.0 2.0 2.03/31/04 6.4 3.3 c 5.2 2.6 5.2 2.9 5.2 2.64/13/04 1 2.0 c 2.0 2.0 c 1.0 2.0 c 1.2 2.0 c

4/22/04 0.5 1.1 0.8 1.2 0.5 1.0 0.7 1.15/7/04 0.2 0.7 0.3 0.7 0.2 0.6 0.2 0.65/24/04 34 1.5 c 32.0 1.5 c 26.0 1.5 c 28.5 1.5 c

5/25/04 21 25.5 21 25.5 20.0 20.7 21.0 24.7

a Instantaneous flow rates are observed on sampling events and correspond to drip samples.b Average flow rates are calculated from volume of water in the bucket over the duration between this and previous sampling events. They

correspond to tub samples.c Minimum flow rate due to overflow of buckets.


cations and silica observed during the warmer months(Fig. 7F). Similar conclusions were also reached byWhite et al. (2005b) in silicate-rich Merced soils ofCentral California. Thus, the fluctuations in major solutechemistry of the soil solutions may be attributed directlyto seasonal variation in chemical weathering processes.

The elevated soil water Cl− and Na+ concentrationsobserved in February andMarch likely result from elutionof snowpack solutes, presumably deposited via wet or dryatmospheric deposition. NADP data also show that Na+

and Cl− concentrations in wet precipitation tend to behighest in the winter/spring and lowest in the summer.Comparison of the rainfall chemistry among differentstations in Michigan along a N–S gradient (theTahquamenon watershed in Upper Peninsula, the Muske-gon watershed at northern Lower Peninsula and theKalamazoo watershed of this study in southern LowerPeninsula) showed that Na+ and Cl− concentrationsdecreased in the watersheds from the north to south in thewinter/spring, but they were very similar in the summer(Suppl. Table 2). Thus, the different Na+ and Cl−

concentrations observed in the rainfall during winter/spring among these watersheds are probably explained byatmospheric inputs of deicing salts from local sources. Inthe Upper Peninsula of Michigan, population density islow and deicing salts are not applied as frequently andover as large an area compared to the Lower Peninsula.But in the summer, the atmospheric source of dissolvedsolutes is similar for all of Michigan; therefore, there is

no difference in rainwater Na+ and Cl− concentrationsamong these three watersheds.

4.1.2. Soil water Ca2+ and Mg2+, carbonate-mineraldissolution, and soil zone pCO2

When present in soils, carbonate minerals are thedominant contributors of Ca2+ and Mg2+ (Caroll, 1970;Horton et al., 1999; Jacobson et al., 2002, 2003; Whiteet al., 2005a; Jin et al., 2008). Saturation indices of soilsolutions with respect to calcite and dolomite are plottedin Fig. 9. All the soil solutions collected by Prenartlysimeters from the 180 cm depth were near equilibriumwith respect to calcite (SI=0.0±0.5) and undersaturatedwith respect to dolomite (SI=−1.0±0.5), and neither SIvaried much with season (Fig. 9A and B). In contrast, thesaturation indices of drip soil solutions from200 cm depthshowed seasonal variation, tending to be oversaturatedwith respect to calcite in the summer and fall months(Fig. 9C and D). This difference between Prenart and dripsamples may be attributed to more rapid degassing of theCO2 from the drip samples thereby increasing calcitesaturation state. Oversaturation reached maximum valuesin the summer months, when soil CO2 greatly exceededatmospheric CO2. In contrast, Prenart samplers are iso-lated from the atmosphere before sampling and mostlikely provide the best measure of in-situ saturationindices for soil waters. The tub samples had even longerexposure time to air before collection, which could lead tohigher oversaturation and calcite precipitation, consistent

Fig. 9. Saturation indices of calcite (A) and dolomite (B) of the Prenart samples (180 cm, lysimeters) and saturation indices of calcite (C) and dolomite(D) in drainagewater (200 cm, drip samples) as a function of time. A SI value of 0±0.5 suggests that soil water is at equilibrium (between dashed lines).


with the lower Mg2+ and Ca2+ concentrations observed inthe tub relative to drip samples in the summermonths. Theconstant saturation indices of carbonate minerals through-out the year also suggest that soil waters became saturatedwith respect to carbonate minerals rapidly tracking largeseasonal variations in pCO2.

Considering that all of the Prenart soil water sampleshad similar saturation indexes, the changes in Ca2+ andMg2+ concentrations reflect the seasonal variation ofcalcite and dolomite solubility, which decreases withincreasing temperature and increases with increasingpCO2. Thus, higher Ca

2+ andMg2+ concentrations of thedrainage water in the summer must be related to theseasonal increase in soil pCO2, which in turn is related toseasonal variation in soil temperature (Fig. 7A). Micro-bial and root respiration are the major processes thatregulate the daily and long-term soil CO2 and they aretemperature-dependent and, in the case of roots, linked tothe seasonal growth cycle of the plants.

Equations have been proposed by previous studies topredict soil pCO2 levels based on air temperature (e.g.,Brook et al., 1983; Drake, 1983; Wood et al., 1993;Andrews and Schlesinger, 2001; Hui and Luo, 2004).Anderson et al. (2003) studied stream waters in Alaskaand found a positive correlation between calculatedpCO2 and the amount of calcite dissolved per liter even

though pyrite dissolution contributed more than 10% ofprotons. The most common and consequential effects onsoil-solution pH arose from the aqueous phase distribu-tion of inorganic carbon species and the gas-phase/liquid-phase partitioning of CO2. The pH values of thePrenart soil solutions ranged from 7.5 to 8.2 and showedan opposite seasonal trend compared to the soil pCO2.This agreed with the general trend of modeled pH valuesassuming soil solutions are saturated with measured soilpCO2 and calcite, where a ten-fold increase in pCO2

results in a full unit decrease in soil-solution pH. Thesolubility of carbonate minerals decreases with increas-ing temperature, and this direct effect of temperature istherefore secondary to the temperature-dependent pCO2

in affecting carbonate dissolution and soil-solution pH.

4.1.3. Sources and reactivity of NO3− and SO4

2−

The NO3− concentrations decreased in drainage water

over the course of this study (Fig. 8A). The decrease inNO3

− since 2003 corresponds to the corn (2002)–soybean(2003)–wheat (2004) crop rotation cycle, because soy-beans tend to increase soil nitrogen contents by biologicalN2 fixation. Nitrification produces soil nitrate and deni-trification removes it. Thus, NO3

− leaching depends on thebalance of plant demands and these two microbiallymediated processes. In previously studied forested sites of

Fig. 10. Evolution paths of soil water chemistry (Ca2+ and Mg2+) bycarbonate dissolution. Ranges of the soil waters and the ground watersfrom the Huron watershed and stream waters from the Kalamazoowatershed are shown for comparison (data from Jin et al., 2008;Szramek et al., 2007). See text for discussion of the two-stage evolutionof carbonate weathering reactions.


the Huron watershed, soil-solution NO3− concentrations

were much lower than in the four monoliths (shaded areain Fig. 8A; data from Jin, 2007). The high SO4

2− concen-trations were relatively constant throughout the year(Fig. 8B). One study in this area showed that dry and wetdeposition together only contributed about 18% of theSO4

2− in the streams (Rheaume, 1990). Pyrite dissolutionmust be a significant SO4

2− source, as Marshall Sandstoneis enriched in sulfide minerals such as pyrite. The streamsin the Kalamazoo watershed had been shown to havemuch higher SO4

2− concentrations than those in otherMichigan watersheds, because of the existence of shallowsubcropping of the Marshall sandstone beneath glacialdrift (Szramek et al., 2007).

In soil monoliths, nitric and sulfuric acids canbecome a major driver of chemical weathering reactionsand an important component in anion charge balance(Fig. 5A). These strong acids could react with carbonateminerals to produce either CO2 or HCO3

−, which is animportant process for atmospheric CO2 cycling (Spenceand Telmer, 2005; Hamilton et al., 2007).

4.2. Controls on chemical weathering of carbonatesand silicates

4.2.1. Soil exchangeable cation pool and its equilibriumwith soil solutions

Seasonal variation in concentrations of major solutes(Ca2+, Mg2+ and silica) was observed at the study sites,while Mg2+/Ca2+ molar ratios remained constant through-out the year (Figs. 6 and 7). Similar seasonal trends inconcentrations but with constant ratios between divalentcations and between univalent cations have been observedin other soil water and stream water studies (Berner et al.,1998; Grasby et al., 1999; White et al., 2005b). At theHubbard Brook Experimental Forest (New Hampshire,USA), positive correlationswere found between soilwaterCa2+ andMg2+ concentrations aswell as betweenNa+ andK+ concentrations, and this was interpreted as control byequilibria between the very dilute soil solutions and anexchangeable pool associatedwith clay and organicmatter(Berner et al., 1998). A clear sinusoidal trendwas apparentfor silica, Ca2+ and K+ over an annual cycle, which wasinterpreted as being related to the acid production in thesoil as a function of temperature. Another possible expla-nation, the effect of varying temperature on the rate ofrelease of cations and silica from primary mineral disso-lution, was refuted because no similar seasonal variationof HCO3

− was observed. White et al. (2005b) also em-phasized cation exchange capacity in the control of soil-solution Mg2+/Ca2+ ratios, although that study examinedrelatively clay-rich soils in central California.

At our study sites, no linear correlation was observedbetween K+ and Na+, or between Na+ and Ca2+ in ourmore concentrated soil solutions. Furthermore, the clayfraction was dominated by quartz and only significant inthe first 90 cm, with soils becoming sandy below thisdepth. Another component of the soil cation exchangecomplex, organic matter, was only abundant at the top30 cm of these soils. The sharp increase in divalent cationconcentrations occurred right below the carbonate layer(150 cm), below the zone of potential control by thecation exchange pool. Taken together, these indicate avery minor role for soil cation exchange processes in thesoil water solute budgets for the soil monolith at KBS.

4.2.2. Dolomite versus calcite dissolutionTheMg2+/Ca2+ molar ratios of soil waters at KBS are

about 0.3, lower than those of soil waters and groundwaters in the Huron watershed and also lower than thoseof streams in the Kalamazoo watershed (Fig. 10;Szramek et al., 2007; Hamilton et al., 2007; Jin et al.,2008). This cannot be explained by cation exchangeprocesses as previously discussed, nor by lower Mg2+

contribution from silicate dissolution, as amphibolecontents and soil water Mg2+ concentrations in shallowcarbonate-free soil zones were similar between KBS andthe Huron watershed (Table 1; Jin et al., 2008). Thus,lower Mg2+/Ca2+ ratios reflect a proportionately smallercontribution from dolomite dissolution relative to calcite,even though similar amounts of dolomite and calcite arepresent in all southern Michigan soils. This is consistentwith the fact that Prenart soil water samples were still




Table 5Recovery and residence time of Br− in one-year tracer experiment

Recovery (%) Residence time (days) a

Monolith #2 45.0 43.4Monolith #6 15.8 162.7Monolith #9 44.3 156.8Monolith #13 40.9 109.3a Residence time of Br is calculated only for recovered Br; The

actual residence time of Br should be much longer than this.


undersaturated with respect to dolomite but at equili-brium with calcite.

Theoretical models have been presented previously toreconstruct evolution paths of soil water chemistry underthe influence of carbonate dissolution and to evaluatedolomite versus calcite dissolution kinetics in systemsopen to CO2 (Jin et al., 2008). As shown in Fig. 10, thesemodels include two stages: first, calcite and dolomitedissolve at a certain rate and soil water reachesequilibrium with calcite but remains undersaturated withdolomite; second, dolomite continues to dissolve to equi-librium, driving calcite to supersaturation. Model resultsindicated Mg2+/Ca2+ ratios of the soil waters dependdominantly on the relative dissolution rates of calcite anddolomite and the progress of hydrogeochemical evolutionof the system. Soil waters in the Huron watershed havereached the second stage in their evolution, and similardolomite and calcite dissolution rates were required (Jinet al., 2008). Soil waters at KBS are at equilibrium withcalcite, suggesting that carbonate dissolution at KBS hasonly evolved to the first stage, and Mg2+/Ca2+ ratios of~0.3 require similar dolomite and calcite dissolution rates,as in the Huron watershed. Model results at the end ofstage one would produce a saturation index of dolomitearound −0.4, consistent with those calculated for KBSsoil waters (Fig. 9D).

Therefore, both theoretical models and water chem-istry suggest that at KBS dolomite dissolution has notreached equilibrium within the soil zones and willcontinue to evolve chemically in deeper soil zones and/orbelow the water table. This is consistent with the higherMg2+/Ca2+ ratios observed in surface waters of theKalamazoo watershed (Fig. 10; Szramek et al., 2007).We attribute different carbonate dissolution scenariosbetween KBS and the Huron watershed to the shortercontact time of dolomite and soil waters at KBS, due toboth the coarser soil textures and the shorter carbonate-bearing soil column path length exposed to dolomitedissolution before water exits the monoliths.

4.2.3. Residence time of water in the soil columnSolute transport in soil media can be conceptualized

as consisting of variable proportions of preferential flowand piston flow. In a preferential flow-dominated profile,well-connected macropores form flow channels. Solutesin flow channels are rapidly transported downward prefer-entially while those in the surrounding matrix (micro-pores) are transported only after encountering a flowchannel (Harrington and Bales, 1998; Feng et al., 2001).Profiles of Br− in monolith #2 clearly exhibit preferentialflow (see Figs. 3 and 4)where Br− peaks in drainagewaterarrive immediately after tracer application and thereafter,

Br− concentrations decreased regardless of infiltrationrate. The CO2 profile is governed by production rates,diffusion upward to atmosphere and advective transportdownward by fluids. Soil CO2 and soil water are moreeasily moved in a well-connected channel than in theisolated matrix, and this may be the reason that pCO2

concentrations and soil water Ca2+ concentrations are thelowest in monolith #2. In piston flow, water progressivelymigrates through the medium, carrying solutes down thehydraulic head gradient (Hibberd, 1984) with channel andlateral flow relatively insignificant. This pattern of trans-port is well illustrated by monoliths #6 and #9 (see Fig. 4)where increases in Br− concentrations in drainage waterscoincide with rain events.

The Br− recovery after 1 year was calculated basedon the following equation:

Recovery % ¼ ∑ð½Br�t⁎VH2OÞ=½Br�total ð2Þ

While [Br] t is the concentration of Br− in the tub at time

t, VH2O is the volume of the water recorded from t−1 to t,t is the number of days since tracer injection, and [Br]totalis the total amount of Br− applied. The buckets over-flowed during high rainfall events and the volume of thebucket was used for the calculation, which should under-estimate the recovery rate. Approximately 40% of Br−

was recovered from monoliths #2, #9 and #13 and only16% of Br− was recovered from monolith #6. Using theamount of Br− recovered, and Br− concentration/timedata, the residence time of Br− was calculated based onthe following equation:

Residence time ¼ ∑ð½Br�t⁎VH2O⁎ tÞ=∑ð½Br�t⁎VH2OÞ

ð3ÞThe residence time of recovered Br− in these mono-

liths ranged from 43 to 160 days (Table 5). Because morethan half of the Br− was retained in the soil profile after1 year, it is reasonable to assume that residence time forall the Br is two or more years. Soil texture affects water-flow patterns as well as residence time of solutes in a soil





profile. Residence time affects the proportion of soilwater that is lost to evapotranspiration versus infiltrateddownward into the ground water reservoir.

Residence time is also a very important determinant ofwater/rock interaction, controlling whether mineral equi-librium can be attained in the soil zone. This is particularlyimportant for the silicate minerals that react much moreslowly than do carbonate minerals (Gilbert, 1877; Carsonand Kirkby, 1972; Heimsath et al., 2001; Gabet et al.,2006). Previous studies have shown that soil waters inmicropores and macropores have very different majorsolute compositions (Richards and Kump, 2003). How-ever, the residence time is defined as the average time thebromide tracer stayed over the whole 200 cm soil profile.KBS soils tend to be heterogeneous; soils fromO-, A- andC-horizons are sandy loams and loams, while B-horizonshave a much higher clay content. Thus, it is reasonable toassume that water resides in the first 150 cm for most ofthe time, flowing rapidly out of the more permeabledeeper C-horizon. This textural difference could explainwhy dolomite which occurs only in the C-horizon canremain undersaturated despite the long average waterresidence time.

4.2.4. Silica, Na⁎ and silicate weatheringThe Na⁎ value is a proxy for plagioclase dissolution

because plagioclase is the only significant mineralsource of Na in these soils. Amphibole can containsmall amounts of Na (less than 1 wt. %), averaged fromsemi-quantitative EDX-SEM analysis. Given that am-phibole is less abundant but has similar reactivity toplagioclase (Blum and Stillings, 1995), the contributionof Na from amphibole dissolution is relatively minor. Incalculation of soil water Na⁎, Na+/Cl− ratios of the wetprecipitation were averaged over 1 year, and this sea-sonal variation is the major source of error according toEq. (1) (about 17 μM). As the study area is more than300 m away from the road, the direct effect of road saltis minimum, and the elevated Cl− is due to wet/dryprecipitation as discussed in previous Section 4.1.1.Silica in the soil solutions must have been producedmainly by the dissolution of the silicate mineralsplagioclase, mica and amphibole because quartz andK-feldspar react much more slowly. The molar ratio ofsoil-solution silica to Na⁎ concentration is less than 3,requiring stoichiometric loss of silica (Fig. 5C). Silicatesdissolve incongruently, with gibbsite stable at low silicaconcentrations (b20 μM) and kaolinite and smectiteat progressively higher silica concentrations (Brickerand Garrels, 1965; Helgeson et al., 1969; Colman andDethier, 1986; White and Blum, 1995; Berner andBerner, 1996). At KBS, silica concentrations range from

60 to 140 μM, placing soil waters within the stabilityfield of kaolinite, similar to previously studied Michiganwatersheds (Jin et al., 2008). Soil waters at KBS weregenerally undersaturated with respect to plagioclase andamphibole, but oversaturated with kaolinite. Thus, theseasonal variation of silica observed in the KBS soilwaters is not solely controlled by silicate dissolutionbut also by precipitation of secondary minerals. Mineralsaturation state results show that kaolinite was less over-saturated in the summer months even though silicaconcentrations were higher and kaolinite solubility waslower at this time. This is probably due to the lower pHcaused by higher pCO2 values in the summer. Since thekinetics of kaolinite precipitation are complicated andpoorly studied in soils, the variation of silica concentra-tions with season cannot be ascribed to a single reactionmechanism.

Silica concentrations and fluxes have been used asproxies of silicate-weathering intensity in riverine sys-tems (Bluth and Kump, 1994; White and Blum, 1995;White et al., 1999). However, silica fluxes are compli-cated by potential silica consumption (e.g., in-riverprocesses such as uptake by diatoms and kaoliniteprecipitation) (Meybeck, 1987; Velbel, 1989; Berneret al., 1998; Huh et al., 1998; Miretzky et al., 2001;Pawellek et al., 2002). Also, silica is involved inbiogenic cycles of terrestrial environments, which is asizable pool of Si (Conley, 2002). Dissolved silica alsoforms complexes with dissolved organic molecules,increasing its solubility (Gérard et al., 2003). Takentogether, stream water and soil water silica concentra-tions in Michigan can only provide a minimum estimatefor silicate-mineral weathering rates.

Concentrations of Na⁎ decreased only slightly inresponse to the increasing mean seasonal drainage rate(Fig. 7E), indicating that geochemical processes in the soiltend to mitigate the dilution effect attributable toprecipitation inputs. The higher Na⁎ concentrations ob-served in the summer months could result from elevatedplagioclase dissolution rates associated with high pCO2

and/or high temperature. Decreases in soil pH caused byincreased pCO2 may not in themselves influence silicatedissolution rates because experiments have shown thatreaction rates are invariant between pH 5 and 8 (e.g.,Oelkers et al., 1994; Oelkers and Gislason, 2001). Theeffect of pCO2 on silicate weathering, therefore, may beindirect, from the CO2 fertilization of organic activity andchelating ofAl by carbonic ions (Brady andCarroll, 1994;Berg and Banwart, 2000).

As previously stated, residence time of solutions inthe soils might be sufficient for solute concentrations tobuild up enough to decrease the dissolution rates via a



chemical affinity effect (Clow and Drever, 1996). Soilwater Na⁎ at KBS ranged from 40 to 120 μM, slightlylower than those in the Huron watershed (50–150 μM;Jin et al., 2008). Plagioclase in the KBS soils is alsoslightly less abundant than in soils of the Huronwatershed, based on Na2O wt.% in bulk soil. This agreeswith our previous observation that silicate weathering isprimarily controlled by mineral contents, and specificallyby weathering surface areas.

4.3. Mineral weathering rates from solute fluxes andfrom soil mass balance

4.3.1. Short-term weathering rates from solute fluxesSzramek andWalter (2004) showed that most streams

draining glaciated watersheds in southern Michigan aretoo contaminated by road deicing salts to evaluatesilicate-weathering contributions to the overall cationfluxes. A similar study was made of riverine Na⁎ fluxesin two northern Michigan watersheds, where pollutionwas at minimal values due to extremely low populationdensities (Williams, 2005). Here, silicate-weatheringfluxes were in the upper range of those reported byWhiteand Blum (1995) in the streams draining granitoidwatersheds, even thoughmean annual temperatures wererelatively low in northern Michigan (Williams, 2005).The results presented here from the soil monolith studiespermit evaluation of silicate-weathering fluxes from theKalamazoo watershed, typical of southern Michiganlandscapes, because of the availability of soil waterbudgets. To compare the Kalamazoo watershed silicate-weathering rates with prior studies, elemental fluxes via

Table 6ACation fluxes and short-term mineral weathering rates

Species ofinterest

Fluxes(mol ha−1 yr−1)

Minerals ofinterest

Chemicalweathering rates(mol ha−1 yr−1)

Ca2+ 5500 Calcite (CaCO3) 3400Mg2+ 1700 dolomite

(Mg0.5Ca0.5CO3)3100

K+ 210 K-feldspar(KAlSi3O8)

210

Na+ 250 Plagioclase (Ab0.8) 320H4SiO4

0 340

Water discharge=1185 L yr−1.

Table 6BLong-term plagioclase-weathering rates (soil age=12.5 ka)

By Zr mass balance: 2400 mol ha−1 yr−1

By Ti mass balance: 1300 mol ha−1 yr−1

Long-term weathering rates are calculated based on data in Table 1; seetext for details.

infiltration in the soil monoliths are calculated based onion concentrations and drainage water discharge of theyear 2004 (Table 6A). Weathering rates of plagioclasewere calculated at 320 mol ha−1 yr−1 from Na⁎ fluxes.

Glacial drift deposits typically host unconfinedaquifers with close hydrogeochemical links betweensoil waters and ground water/surface water systems (e.g.,Szramek and Walter, 2004; Williams et al., 2007b) andsoil zone processes can exert strong controls on surfacewater chemistry. Previous studies have shown that soilNa⁎ concentrations in several Michigan watershedsincrease sharply in shallow soil horizons and then remainrelatively constant with depth, suggesting that silicateweathering occurs early in the soil water column withcarbonate weathering proceeding at greater depths(Williams et al., 2007a; Jin et al., 2008). High DOCconcentrations and low pH values of the shallow soilwaters also favor silicate hydrolysis reactions (e.g.,Oelkers et al., 1994; Williams et al., 2007a; Jin, 2007).The Br tracer results presented earlier estimate soluteresidence time in the monoliths to be long, allowing slowkinetic silicate dissolution reactions to proceed towardsequilibrium. Once the carbonate-mineral dissolutionstarts in the deeper soil profile, soil water pH increasesand DOC decreases rapidly, both preventing furthersilicate-mineral dissolutions. Thus, short-term weath-ering rates derived from the soil monoliths studies aregood estimates of silicate-weathering fluxes and they areespecially valuable in watersheds like the Kalamazoowhere streamwater chemistry cannot be used to comparedirectly to riverine watershed-scale cation flux data(e.g., White and Blum, 1995).

The Na⁎ fluxes in this study are plotted versus annualrunoff in Fig. 11A, as well as data from granitoidwatersheds in eastern North America compiled by Whiteand Blum (1995) and from two northern Michigan water-sheds by Williams (2005). The contribution of runoff(x axis) and the volume-weighted Na⁎ concentrations(slope) to the overall cation fluxes can be evaluated fromthese relations. Regardless of glacial history, plagioclase-weathering rates are much lower in the watersheds withextremely low mean annual temperature (e.g., RawsonLake, Exp. Lake, and LochValewatersheds) and higher inthe warmer watersheds (e.g., Coweeta and Panola water-sheds). Mineral surface areas from modern glacial envi-ronments are shown to be enhanced by glacial abrasion,but relative silicate-weathering rates still are low due tothe extremely low temperature in the vicinity of theglaciers (Anderson, 2005).

Williams et al. (2007a) compared soil-solution chem-istry in natural and experimental soils profiles developedon Pleistocene glacial drift in the same Michigan


Fig. 11. Na⁎ fluxes from surface water geochemical studies as a functionof annual runoff (A). Slopes (straight lines) indicate average Na⁎

concentrations of natural waters in μM. Data fromMichigan Cheboygan(Cheb) and Tahquamenon (Tahq) watersheds are from Williams (2005);other data in eastern North American watersheds are compiled by Whiteand Blum (1995). Their data sources include Schlinder et al. (1976),Allen et al. (1993), Siegel and Pfannkuch (1984), April et al. (1986),Mastet al. (1990), Yuretich andBatchelder (1988), Likens et al. (1977), Nortonet al. (1994), Swank andWaide (1988), Peters (personal communication,1994), Buell and Peters (1988), Cleaves et al. (1970). Long-term Na lossby soil mass balances as a function of soil age (B). Data from previouswork are also presented for comparison (April et al., 1986;Kirkwood andNesbitt, 1991; Taylor and Blum, 1995; Nezat et al., 2004). Mean annualtemperature and mean annual precipitation of these watersheds are listed.


watershed. Here, plagioclase dissolution rates wereaccelerated in disturbed experimental soils. It was furthersuggested that the disturbed soils allowed more reactivemineral surface to be exposed, analogous to soils inglaciated regions (Williams et al., 2007a). Jin et al. (2008)compared soil water Na⁎ concentrations in three Michigan

watersheds along climate and lithology gradients, andshowed that plagioclase weathering is most intensive in theHuron watershed which also has the highest abundance ofplagioclase in the soils. In the soil monoliths of theKalamazoowatershed, glacially derivedmixedmineralogysediments are weathered at relatively high mean annualtemperature, both of which favor rapid silicate weathering.Thus, as shown in Fig. 11A, the high weathering ratescombined with high mean annual precipitation make theNa+ fluxes from the mid-continent region among thelargest short-term fluxes in the eastern USA region.

Although up to 10% of the soil water Ca2+ and Mg2+

can be provided from silicate dissolution in glacial driftsoil profiles in Michigan (Jin et al., 2008), silicateweathering introduces little uncertainty in calculatingcarbonate dissolution rates from divalent cation fluxes.Following these estimates, 90% of Ca2+ and Mg2+

fluxes are used herein to calculate calcite and dolomiteweathering rates (calcite: 3400 mol ha−1 yr−1; dolomite:3100 mol ha−1 yr−1). Carbonate weathering rates areprimarily controlled by pCO2-dependent solubility andannual discharge in soil monoliths, in agreement withsurface water fluxes calculated in Michigan streams(Szramek and Walter, 2004; Williams et al., 2007b;Szramek et al, 2007).

The monoliths studied are manipulated agriculturalsites, and surface soils have been cultivated since 1987.Compared to the natural forests, soil solutions here arehigh in nitrate concentrations, which lower the soilwater pH and potentially accelerate the leaching ofmajor cations. Thus, the short-term weathering ratescalculated from Na-fluxes may have been slightly over-estimated relative to those in natural environments.

4.3.2. Long-term plagioclase-weathering rates fromsoil elemental mass balances

Elemental inventories in soil profiles reflect loss ofcations over the thousands of years required for soil devel-opment and thus can estimate long-term weathering ratesby both physical and/or chemical erosion (e.g., Taylor andBlum, 1995; White et al., 1996; Bain et al., 2001; White,2002; Nezat et al., 2004; Price et al., 2005). This methodassumes the following mass balance relationships (Brim-hall and Dietrich, 1987; Chadwick et al., 1990):

mWCj;W ¼ mPCj;P þ mj;flux ð4Þ

where m is the mass of soil, Cj is the concentration ofelement j in the soil, and subscripts P and W indicateparent material and weathered soil. Because there is littletopographic relief in the vicinity of the soil monoliths it isreasonable to assume that elements leave the profile only




through chemical weathering. In Eq. (4), mj,flux is theamount of element j that is lost (or gained) in soil profilesand is defined as zero for immobile elements such as Ti andZr. Thus, Eq. (4) can be simplified and the variation of soilmass be evaluated (Zr as an example):

mP ¼ mWCZr;W=CZr;P ð5Þ

If mP in Eq. (4) is replaced by Eq. (5), the amount ofelement that is lost from soils can be calculated as:

mj;flux ¼ mWCj;W � mPCj;P

¼ mW Cj;W � Cj;PCZr;W=CZr;P

� � ð6Þ

where mass of the soils can be calculated by ZW and ρW,thickness and density of the soils:

mW ¼ ZWqW ð7ÞSome soil development has been shown to be iso-volumetric and thus the thickness does not change duringconversion from parent materials to soil profiles (Zp=Zw;White et al., 2005b). Because up to 30 wt.% of carbonateminerals originally present in soils have been removedby dissolution (Table 2), iso-volumetric development ofthe soil profile could not be assumed in Michigan. Forexample, soils developed in karst regions of south Chinahave been shown to collapse by almost 100% (Ji et al.,2004). The amount of volume loss relative to the parentmaterial can be evaluated by the following equation:

ZpqpCZr;p ¼ ZwqwCZr;w ð8Þ

Rearranging gives:

Zw=Zp ¼ qpCZr;p=qwCZr;w ð9Þ

The volume change factor (Zw/Zp) is much less than 1 inthe KBS soil profiles (Table 1), indicating that significantvolume lossmust be considered to arrive at valid long-termweathering fluxes from these carbonate-rich soils.

Because Cj,w varies with soil depth, the weathered soilzone is divided into discrete layers and the total elementalloss is integrated over the entire soil column. Ti and Zr arecommonly used as examples of immobile elements andthey mainly occur in soils as stable accessory mineralsilmenite and zircon. Using the calibrated Zp values fromEq. (9) yields long-term plagioclase-weathering ratesbased on Ti and Zr mass balances of 1300 and 2400 molha−1 yr−1 respectively (see Table 6B). Zr/Ti (as TiO2)ratios are constant at the two deepest soil samples(including the parent material at 190 cm), but are muchhigher in the shallow weathered soils, indicating that Ti

has been preferentially removed via chemical weatheringrelative to Zr (Table 1). Because ilmenite is less stable insurface environments, plagioclase-weathering ratesderived by Zr concentrations may be more accurate,than those derived from Ti mass balance.

The long-term plagioclase-weathering rates calculatedfrom soil mass balance are about 8 times the presentweathering rates calculated from solute chemistry andfluxes (24000 vs. 320 mol ha−1 yr−1). This is in agree-ment with the general trend that weathering rates decreaseas a soil profile develops (Taylor and Blum, 1995).Silicate-weathering rates diminish over time due to theloss of reactiveminerals and fresh mineral surfaces, partlydue to development of Al- or Si-rich leached layers orfurther deposition of secondary Al and Fe oxyhydroxidesat the mineral surfaces (e.g., Bain et al., 1993; Taylor andBlum, 1995; White and Brantley, 2003).

In Fig. 11B, long-term Na loss (proportional toplagioclase-weathering rates according to the An contentof the plagioclase) is plotted versus soil ages. Alsopresented are data from Wind River Mountain (Wyom-ing), Panther Lake (New York), Plastic Lake (Ontario)and Hubbard Brook (NewHampshire) (April et al., 1986;Kirkwood andNesbitt, 1991; Bain et al., 1993; Taylor andBlum, 1995; Nezat et al., 2004). A power law relationbetween plagioclase-weathering rates and soil age isobserved for these sites collectively (Nezat et al., 2004).The plagioclase dissolution rate in Kalamazoo soil issignificantly higher than those in other watersheds of thesame age.

There are differences among thewatersheds in terms ofclimate and parent materials. Relative to the localitiesshown in Fig. 11B, the Kalamazoo watershed is in theupper range of mean annual temperatures, and is mid-range in terms of precipitation. The sedimentary rocksources for Michigan glacial drift soils likely increasethe weathering rates due both to particle surface areas andhigh plagioclase abundances. This would be consistentwith the observations from current silicate-weatheringrates as well. Additional study of silicate-weathering ratesin mixed mineralogy, sedimentary sourced profiles wouldbe required to quantify how important each of thesevariables is in controlling short- and long-term rates.

5. Conclusions

Soils, soil waters and gases were characterized in soilmonoliths at the KBS LTER agricultural site (SWMichigan) to study chemical weathering processesduring the initial contact of infiltrating waters withsoils and underlying glacial drift materials. Soil solutionsattained a strong signature of carbonate-mineral


weathering by a depth of 200 cm, and Ca2+ and Mg2+

concentrations reflected roughly equal contributionsfrom calcite and dolomite dissolution with little additionby silicate-mineral weathering. Frequent sampling hasilluminated temporal variations of soil gas pCO2 and soilwater Ca2+ and Mg2+, with concentration maxima in thesummer and minima in the winter. Carbonate weatheringreactions are sufficiently rapid that soil waters haveequilibrated with calcite but remained slightly under-saturated with respect to dolomite by 50 cm below theupper extent of carbonates.

The saturation states of calcite and dolomite changedlittle throughout the year and did not vary with precipi-tation intensity, implying first-order control of Ca2+ andMg2+ concentrations by carbonate-mineral solubility.Thus, higher carbonate solubility and resultant higherCa2+ andMg2+ concentrations in the soil solutions duringthe warm season were related to temperature-dependentsoil pCO2. Soil water Mg2+/Ca2+ molar ratios at KBS areslightly lower than those in the Huron watershed as aresult of less contribution from dolomite dissolution.This is consistent with the fact that soil waters atKBS were still undersaturated with respect to dolomite.Previously developed theoretical models reproduced thefield-observed chemistry data and suggested that theextent of dolomite dissolution was reduced due to shortercontact time with soil waters.

Silicate-weathering studies based on analysis of Naconcentrations in surface waters of Michigan haveproven problematic due to contamination by deicingsalts and fertilizers. Soil water studies, on the otherhand, can have better controls on sources of cations andmineral compositions. At KBS, Na⁎, an indicator ofplagioclase dissolution, ranged from 40 to 160 μM withslightly higher concentrations during warmer months. Incontrast to carbonate-mineral weathering, the soilsolution was always undersaturated with respect toplagioclase, suggesting the plagioclase dissolution rateswere subject to kinetic regulation by factors such astemperature and availability of reactive surfaces. There-fore, the higher Na⁎ during warmer months could be dueto higher reaction rates as influenced by higher soiltemperature and higher pCO2, and longer residence timeof infiltrating water in the soil profiles. The LiBr tracerexperiments suggest that water residence times in200 cm deep soil profiles varied among the monoliths,but were likely on the order of a couple of years. Silicaconcentrations were elevated in the summer, similar tothe divalent cations, but not likely controlled solely bysilicate dissolution rates as soil waters fall into thekaolinite field in stability diagrams and silicate weath-ering is thus incongruent.

Seasonal variability of stream water chemistry hasbeen attributed to mixing of waters from different parts ofwatershed with different lithology or mixing of dilutesurface runoff with relatively solute-rich ground waters(Douglas, 2006; Tipper et al., 2006). This study suggeststhat at least in glacial drift watersheds, seasonal variationin stream water composition could also be due to theresidence time of water in subsurface flow paths (silicateweathering) and temperature- and pCO2-controlledsolubility (carbonate weathering). Infiltration for year2004 was estimated by the difference between pre-cipitation and evapotranspiration and cation fluxes werecalculated. Cations were assigned to minerals and short-term weathering rates of calcite, dolomite and plagioclasedissolution rates were calculated at 3400, 3100 and320mol ha−1 yr−1, respectively based on soil water solutebudgets. Plagioclase dissolution rates at KBS are high,because of fresh mineral surfaces and high waterinfiltration rates at this glaciated mid-continent region.Furthermore, soil profile mass balances suggest that thelong-term plagioclase-weathering rates are about2400 mol ha−1 yr−1, much higher than the short-termaverage, and this is consistent with the general trend thatchemical weathering rates of silicate minerals decreasewith age due to loss of fresh mineral surfaces. It is alsonotable that the long-term plagioclase-weathering ratesobserved in KBS soils are much higher than those fromstudies of soil sequences of similar age, as a result of highmineral surface areas by glacial abrasion, high meanannual temperature and plagioclase abundances.

Acknowledgements

Special thanks to David Weed and Corey Lambert fortheir assistance in the field and lab. The tracer experimentwas partially funded by a Scott Turner Award from theDepartment of Geological Sciences at the University ofMichigan and a student grant from Geological Societyof America to L. Jin. The KBS LTER site including themonoliths were supported by grants 9810220 and 0423627from the US National Science Foundation. Support fromUSNational Science Foundation grants EAR-0208182 and-0518965 to LMW is also acknowledged. We thank editorJeremy Fein for handling the manuscript and SteveDworkin and another anonymous reviewer for theircomments, which greatly improved this paper.

Appendix A. Supplementary data

Supplementary data associated with this article can befound, in the online version, at doi:10.1016/j.chemgeo.2007.12.002.

http://dx.doi.org/doi:10.1016/j.chemgeo.2007.12.002

http://dx.doi.org/doi:10.1016/j.chemgeo.2007.12.002


References

Allen, W.B., Miller, J.B., Wood, W.W., 1972. Availability of Water inKalamazoo County, Michigan. U.S. Geological SurveyWater-Supply.paper 1973, 129 pp.

Allen, C.J., Roulet, N.T., Hill, A.R., 1993. The biogeochemistry of pristineheadwater Precambrian shield watersheds: an analysis of materialtransport within a heterogeneous landscape. Biogeochemistry 22,37–79.

Anderson, S.P., 2005. Glaciers show direct linkage between erosion rateand chemical weathering fluxes. Geomorphology 67, 147–157.

Anderson, S.P., Longacre, S.A., Kraal, E.R., 2003. Patterns of waterchemistry and discharge in the glacier-fed Kennicott River, Alaska:evidence for subglacial water storage cycles. Chemical Geology202, 297–312.

Andrews, J.A., Schlesinger,W.H., 2001. Soil CO2 dynamics, acidification,and chemical weathering in a temperate forest with experimental CO2

enrichment. Global Biogeochemical Cycles 15, 149–162.April, R.A., Newton, R.M., Coles, L.T., 1986. Chemical weathering in

two Adirondack watersheds: past and present-day rates. Geologi-cal Society of American Bulletin 97, 1232–1238.

Bacon, D.H., Keller, C.K., 1998. Carbon dioxide respiration in thedeep vadose zone: implications for groundwater age dating. WaterResources Research 34, 3069–3077.

Bain, D.C., Mellor, A., Robertson-Rintoul, M.S.E., Buckland, S.T.,1993. Variations in weathering processes and rates with time inchronosequence of soils from Glen-Feshie, Scotland. Geoderma57, 275–293.

Bain, D.C., Roe, M.J., Duthie, D.M.L., Thomson, C.M., 2001. Theinfluence ofmineralogy onweathering rates and processes in an acid-sensitive granitic catchment. Applied Geochemistry 16, 931–937.

Berg, A., Banwart, S.A., 2000. Carbon dioxide mediated dissolution ofCa-feldspar: implications for silicate weathering. Chemical Geology163, 25–42.

Berner, E.K., Berner, R.A., 1996. Global Environment: Water, Air, andGeochemical Cycles. Prentice Hall.

Berner, R.A., 1991. Atmospheric carbon dioxide levels overPhanerozoic time. Science 249, 1382–1386.

Berner, R.A., 1994. GEOCARB II: a revised model of atmospheric CO2

over Phanerozoic time. American Journal of Science 294, 56–91.Berner, R.A., Caldeira, K., 1997. The need for mass balance and

feedback in the geochemical carbon cycle. Geology 25, 955–956.Berner, R.A., Rao, J., Chang, S., O’Brien, R., Keller, C.K., 1998.

Seasonal variability of adsorption and exchange equilibria in soilwaters. Aquatic Geochemistry 4, 273–290.

Bickle, M.J., 1996. Metamorphic decarbonation, silicate weatheringand the long-term carbon cycle. Terra Nova 8, 270–276.

Blum,A.E., Stillings, L.L., 1995. Feldspar dissolution kinetics. In:White,A.F., Brantley, S.L. (Eds.), Chemical Weathering Rates of SilicateMinerals, vol. 31. Mineralogical Society of America, pp. 291–352.

Bluth, G.J.S., Kump, L.R., 1994. Lithologic and climatologic controls ofriver chemistry. Geochimica et Cosmochimica Acta 58, 2341–2359.

Boeglin, J., Probst, J., 1998. Physical and chemical weathering ratesand CO2 consumption in a tropical lateritic environment: the upperNiger basin. Chemical Geology, 148, 137–156.

Brady, P.V., Carroll, S.A., 1994. Direct effect of CO2 and temperatureon silicate weathering: possible implications for climate control.Geochimica et Cosmochimica Acta 58, 1853–1856.

Bricker, O.P., Garrels, R.M., 1965. Mineralogic factors in natural waterequilibria. In: Faust, S., Hunter, J.V. (Eds.), Principles and applicationsof water chemistry. John Wiley, New York, pp. 449–469.

Brimhall, G.H., Dietrich, W.E., 1987. Constitutive mass balancerelations between chemical-composition, volume, density, poros-ity, and strain in metasomatic hydrochemical systems-results onweathering and pedogenesis. Geochimica et Cosmochimica Acta51, 567–587.

Brook, G.A., Folkoff, M.E., Box, E., 1983. A world model of soilcarbon dioxide. Earth Surface Processes and Landforms 8, 79–88.

Buell, G.R., Peters, N.E., 1988. Atmospheric deposition effects on thechemistry of a stream in northwest Georgia. Water Air Soil Pollution25, 125–135.

Burgermeister, J., 2007. Missing carbon mystery: case solved? NatureReports of Climate Change. doi:10.1038/climate.2007.35.

Carson, M.A., Kirkby, M.J., 1972. Hillslope Form and Process.Cambridge University Press, New York. 475pp.

Carroll, D., 1970. Rock Weathering. Plenum Press, New York.Catacosinos, P.A., Daniels, P.A.J., 1991. Stratigraphy of Middle

Proterozoic to Middle Ordovician formations of the MichiganBasin. Special Paper-Geological Society of America 256, 53–71.

Chadwick, O.A., Brimhall, G.H., Hendricks, D.M., 1990. From a blackto a grey box — a mass balance interpretation of pedogenesis.Geomorphology 3, 369–390.

Cleaves, A.T., Godfrey, A.E., Bricker, O.P., 1970. Geochemicalbalance in a small watershed. Geological Society of AmericanBulletin 81, 3013–3032.

Clow, D.W., Drever, J.I., 1996. Weathering rates as a function of flowthrough an alpine soil. Chemical Geology 132, 131–134.

Colman, S.M., Dethier, D.P., 1986. Rates of Chemical Weathering ofRocks and Minerals. Academic Press, Orlando, Fla.

Conley, D.J., 2002. Terrestrial ecosystems and the global biogeochem-ical silica cycle. Global Biogeochemical Cycles 16. doi:10.1029/2002GB001894.

Crum, J.R., Robertson, G.P., Nurnberger, F., 1990. Long-term climatetrends and agricultural productivity in southwest Michigan. 1988LTER Climate Variability Workshop, University of Colorado.

Dorr Jr., J.A., Eschman, D.F., 1970. Geology of Michigan. Ann Arbor,MI, University of Michigan Press. 476pp.

Douglas, T.A., 2006. Seasonality of bedrock weathering chemistry andCO2 consumption in a small watershed, the White River, Vermont.Chemical Geology 231, 236–251.

Drake, J.J., 1983. The effects of geomorphology and seasonality on thechemistry of carbonate groundwater. Journal of Hydrology 61,223–236.

Edmond, J.M., 1970. High precision determination of titration alkalinityand total carbon dioxide content of sea water. Deep-Sea Research 17,737–750.

Edmond, J.M., Huh, Y., 1997. Chemical weathering yields from base-ment and orogenic terrains in hot and cold climates. In: Ruddiman,W.F. (Ed.), Tectonic Uplift and Climate Change. Plenum Press,pp. 329–351.

Feng, X., Kirchner, J.W., Renshaw, C.E., Osterhuber, R.S., Klaue, B.,Taylor, S., 2001. A study of solute transport mechanisms using rareearth element tracers and artificial rainstorms on snow. WaterResources Research 37, 1425–1435.

Gabet, E.J., Edelman, R., Langner, H., 2006. Hydrological controls onchemical weathering rates at the soil–bedrock interface. Geology34, 1065–1068.

Gaillardet, J., Dupré, B., Louvat, P., Allègre, C.J., 1999. Global silicateweathering and CO2 consumption rates deduced from thechemistry of large rivers. Chemical Geology 159, 3–30.

Gérard, F., Ranger, J., Ménétrier, C., Bonnaud, P., 2003. Silicateweathering mechanisms determined using soil solutions held athigh matric potential. Chemical Geology 202, 443–460.

http://dx.doi.org/10.1038/climate.2007.35

http://dx.doi.org/10.1029/2002GB001894

http://dx.doi.org/10.1029/2002GB001894


Gieskes, J.M., Rogers, W.C., 1973. Alkalinity determinations ininterstitial waters of marine sediments. Journal of SedimentaryPetrology 43, 272–277.

Gilbert, G.K., 1877. Report on the Geology of the HenryMountains, inU.S. Geographical and Geological Survey of the Rocky MountainRegion. Department of the Interior, Washington D.C. 170pp.

Grasby, S.E., Hutcheon, I., McFarland, L., 1999. Surface-water–ground-water interaction and the influence of ion exchange reactions on riverchemistry. Geology 27, 223–226.

Hamilton, S.K., Kurzman, A.L., Arango, C., Jin, L., Robertson, G.P.,2007. Evidence for carbon sequestration by agricultural liming.GlobalBiogeochemical Cycles 21 (2) GB202110.1029/2006GB002738.

Hannah, R.S., Vogel, T.A., Patino, L.C., Alvarado, G.E., Perez, W.,Smith, D.R., 2002. Origin of silicic volcanic rocks in Central CostaRica: a study of a chemically variable ash-flow sheet in the TiribiTuff. Bulletin Volcanology 64, 117–133.

Harrington, R.F., Bales, R.C., 1998. Modeling ionic solute transport inmelting snow. Water Resources Research 34, 1727–1736.

Heimsath, A.M., Dietrich, W.E., Nishiizumi, K., Finkel, R.C., 2001.Stochastic processes of soil production and transport: erosionrates, topographic variation and cosmogenic nuclides in theOregon Coast Range. Earth Surface Processes and Landforms26, 531–552.

Helgeson, H.C., Garrels, R.M., Mackenzie, F.T., 1969. Evaluation ofirreversible reactions in geochemical processes involving mineralsand aqueous solutions II: applications. Geochimica et Cosmochi-mica Acta 33, 455–482.

Hibberd, S., 1984. A model for pollutant concentrations during snow-melt. Journal of Glaciology 30, 58–65.

Hooper, R.P., Christophersen, N., Peters, N.E., 1990. Modelingstreamwater chemistry as a mixture of soilwater end-members —an application to the Panola Mountain catchment, Georgia, U.S.A.Journal of Hydrology 116, 321–343.

Horton, T.W., Chamberlain, C.P., Fantle,M., Blum, J.D., 1999. Chemicalweathering and lithological controls of water chemistry in a high-elevation river systems: Clark's Fork of the Yellowstone River,Wyoming and Montana. Water Resources Research 35, 1643–1655.

Huh, Y., Panteleyev, G., Babich, D., Zaitsev, A., Edmond, J.M., 1998. Thefluvial geochemistry of the rivers of Eastern Siberia: II. Tributaries ofthe Lena, Omoloy, Yana, Indigirka, Kolyma, and Anadyr draining thecollisional/accretionary zone of the Verkhoyansk and Cherskiy ranges.Geochimica et Cosmochimica Acta 62, 2053–2075.

Hui, D., Luo, Y., 2004. Evaluation of soil CO2 production andtransport in Duke Forest using a process-based modeling approach.Global Biogeochemical Cycles 18 GB4029/2004GB002297.

Hyeong, K., Capuano, R.M., 2001. Ca/Mg of brines in Miocene/Oligocene clastic sediments of the Texas Gulf Coast: buffering bycalcite/disordered dolomite equilibria. Geochimica et Cosmochi-mica Acta 65, 3065–3080.

Jabro, J.D., Lotse, E.G., Simmons, K.E., Baker, D.E., 1991. A fieldstudy of macropore flow under saturated conditions using abromide tracer. Journal of Soil and Water Conservation 376–380.

Jacobson, A.D., Blum, J.D., Walter, L.M., 2002. Reconciling theelemental and Sr isotope composition of Himalayan weatheringfluxes: insights from the carbonate geochemistry of stream waters.Geochimica et Cosmochimica Acta 66, 3417–3429.

Jacobson, A.D., Blum, J.D., Chamberlain, C.P., Craw, D., Koons, P.O.,2003. Climatic and tectonic controls on chemical weathering in theNew Zealand Southern Alps. Geochimica et Cosmochimica Acta67, 29–46.

Ji, H., Wang, S., Ouyang, Z., Zhang, S., Sun, C., Liu, X., Zhou, D.,2004. Geochemistry of red residua underlying dolomites in karst

terrains of Yunnan-Guizhou Plateau II. The mobility of rare earthelements during weathering. Chemical Geology 203, 29–50.

Jin, L., 2007. Mg- and Ca-carbonate versus silicate dissolution rates inmid-latitude, glaciated soil profiles: implications for riverineweathering fluxes and global geochemical budgets. Ph.D.dissertation, University of Michigan, Ann Arbor, MI.

Jin, L., Williams, E.K., Szramek, K.J., Walter, L.M., Hamilton, S.K., 2008.Silicate and carbonate mineral weathering in soil profiles developed onPleistocene glacial drift (Michigan, USA): mass balances based on soilwater geochemistry. Geochimica et Cosmochimica Acta, 72 (4),1027–1042. doi:10.1016/j.gca.2007.12.007.

Kharaka, Y.K., Gunter, W.D., Aggarwal, P.W., Perkins, E.H., DeBraal,J.D., 1988. SOLMINEQ.88: a computer program for geochemicalmodeling of water-rock interactions. U.S. Geological Survey.

Kirchner, J.W., 2003. A double paradox in catchment hydrology andgeochemistry. Hydrological Processes 17, 871–874.

Kirkwood, D.E., Nesbitt, H.W., 1991. Formation and evolution of soilsfrom an acidified watershed: Plastic Lake, Ontario, Canada.Geochimica et Cosmochimica Acta 55, 1295–1308.

Ku, T.C.W., 2001. Organic carbon–mineral interactions in near surfaceenvironments: implications for the global carbon cycles. Ph.D.dissertation, University of Michigan, Ann Arbor, MI.

Likens, G.E., Bormann, F.H., Pierce, R.S., Eaton, J.S., Johnson, N.M., 1977. Biogeochemistry of a Forested Ecosystem. SpringerVerlag.

Mast, M.A., Drever, J.I., Baron, J., 1990. Chemical weathering in theLoch Vale Watershed, Rocky Mountain National Park, Colorado.Water Resources Research 26, 2971–2978.

Meybeck, M., 1987. Global chemical weathering of surficial rocksestimated from dissolved river loads. American Journal of Science287, 401–428.

Miretzky, P., Conzonno, V., Cirelli, A.F., 2001. Geochemical processescontrolling silica concentrations in groundwaters of the Salado Riverdrainage basin, Argentina. Journal of Geochemical Exploration 73,155–166.

National Atmospheric Deposition Program /National Trends Network(NADP/NTN) , 2004. Means of Annual Volume-Weighted RainWater Composition Data from Pellston (MI09), Ann Arbor (MI52)and Seney (MI48) downloaded in 2005. NADP/NTN CoordinationOffice, Illinois StateWater Survey, 2204 Griffith Dr., Champaign, IL61820.

Nezat, C.A., Blum, J.D., Klaue, A., Johnson, C.E., Siccama, T.G., 2004.Influence of landscape position and vegetation on long-termweathering rates at the Hubbard Brook Experimental Forest,New Hampshire, USA. Geochimica et Cosmochimica Acta 68,3065–3078.

Norton, S.A., Kahl, J.S., Fernandez, I.J., Rustad, L.E., Scofield, J.P.,Haines, T.A., 1994. Response of the West Bear Brook Watershed,Maine, USA, to the addition of (NH4)2SO4: 3-year results. ForestEcology and Management 68, 61–73.

Oelkers, E.H., Gislason, S.R., 2001. The mechanism, rates andconsequences of basaltic glass dissolution: I. An experimentalstudy of the dissolution rates of basaltic glass as a function ofaqueous Al, Si and oxalic acid concentration at 25 °C and pH=3and 11. Geochimica et Cosmochimica Acta 65, 3671–3681.

Oelkers, E.H., Schott, J., Devidal, J., 1994. The effect of aluminum,pH, and chemical affinity on the rates of aluminosilicatedissolution reactions. Geochimica et Cosmochimica Acta 38,2011–2024.

Palmer,W.C.,Havens, A.V., 1958. A graphical technique for determiningevapotranspiration by the Thornthwaite method. Monthly WeatherReview 86, 123–128.


Pawellek, F., Frauenstein, F., Veizer, J., 2002. Hydrogeochemistry andisotope geochemistry of the upper Danube River. Geochimica etCosmochimica Acta 66, 3839–3854.

Price, J.R., Velbel, M.A., Patino, L.C., 2005. Rates and time scales ofclay-mineral formation by weathering in saprolitic regoliths of thesouthern Appalachians from geochemical mass balance. Geologi-cal Society of America Bulletin 117, 783–794.

Quade, J., English, N., DeCelles, P.G., 2003. Silicate versus carbonateweathering in the Himalaya: a comparison of the Arun and SetiRiver watersheds. Chemical Geology 202, 275–296.

Reardon, E.J., Allison, G.B., Fritz, P., 1979. Seasonal chemical andisotopic variations of soil CO2 at Trout Creek Watershed,California. Global Biogeochemical Cycles 15, 383–391.

Rheaume, S.J., 1990. Geohydrology and water quality of KalamazooCounty, Michigan, 1986–1988. U.S. Geological Survey Water-Resources Investigation Report 90-4028. 102pp.

Richards, P.L., Kump, L.R., 2003. Soil pore-water distributions andthe temperature feedback of weathering in soils. Geochimica etCosmochimica Acta 67, 3803–3815.

Ridgwell, A.R., Zeebe, R.E., 2005. The role of the global carbonatecycle in the regulation and evolution of the Earth system. Earth andPlanetary Science Letters 234, 299–315.

Rosek, M.J., 1992. Soil Survey of Lux Arbor (unpublished). Dept. ofSoil Sciences. Michigan State University, E. Lansing.

Schlesinger, W.H., 1997. Biogeochemistry: An Analysis Of GlobalChange. Academic Press, San Diego. 588pp.

Schlesinger, W.H., Lichter, J., 2001. Limited carbon storage in soil andlitter of experimental forest plots under increased atmosphericCO2. Nature 411, 466–469.

Schlinder, D.W., Newbury, R.W., Beaty, K.G., Cambell, P., 1976.Natural water and chemical budgets for a small Precambrian LakeBasin in Central Canada. Journal of Fish Research Board Canada33, 2526–2543.

Siegel, D.I., Pfannkuch, H.O., 1984. Silicate dissolution influence onFilson Creek chemistry, northeastern Minnesota. Geological Societyof America Bulletin 95, 1446–1453.

Sparks, D.L., 1996. Methods of Soil Analysis: Part 3. ChemicalMethods and Processes. Soil Science Society of America, BookSeries, vol. 5. SSSA, Madison, WI.

Spence, J., Telmer, K., 2005. The role of sulfur in chemical weatheringand atmospheric CO2 fluxes: evidence from major ions, δ13CDIC,and δ34SSO4 in rivers of the Canadian Cordillera. Geochimica etCosmochimica Acta 69, 5441–5458.

Stumm, W., Morgan, J.J., 1996. Aquatic Chemistry (3rd ed.). Wiley-Interscience, New York.

Sundquist, E.T., 1993. The global carbon dioxide budget. Science 259,934–941.

Swank, W.D., Waide, J.B., 1988. Characterization of baselineprecipitation and stream chemistry and nutrient budgets forcontrol watersheds. In: Swank, W.T., Crossley Jr., D.A. (Eds.),Forest Hydrology and Ecology at Coweeta. Springer Verlag,pp. 57–79.

Szramek, K., Walter, L.M., 2004. Impact of carbonate precipitationon riverine inorganic carbon mass transport from a mid-continent, forested watershed. Aquatic Geochemistry 10,99–137.

Szramek, K., McIntosh, J.C., Williams, E.L., Kanduc, T., Ogrinc, N.,Walter, L.M., 2007. Relative weathering intensity of calcite versusdolomite in carbonate-bearing temperate zone watersheds: carbonategeochemistry and fluxes from catchments within the St. Lawrenceand Danube river basins. Geochemistry, Geophysics, Geosystems 8,Q04002. doi:10.1029/2006GC001337.

Taylor, A., Blum, J.D., 1995. Relation between soil age and silicateweathering rates determined from the chemical evolution of aglacial chronosequence. Geology 23, 979–982.

Thies, H., Nickus, U., Arnold, C., Psenner, R., 2002. A hydrologicaltracer experiment with LiCl in a high mountain lake. HydrologicalProcesses 16, 2329–2337.

Thornthwaite, C.W., 1948. An approach toward a rational classifica-tion of climate. Geographic Review 38, 55–94.

Tipper, E.T., Bickle, M.J., Galy, A., West, A.J., Pomiès, C., Chapman,H.J., 2006. The short term climatic sensitivity of carbonate andsilicate weathering fluxes: insight from seasonal variations in riverchemistry. Geochimica et Cosmochimica Acta 70, 2737–2754.

Velbel, M.A., 1989. Weathering of hornblende to ferruginous productsby a dissolution–reprecipitation mechanism: petrography andstoichiometry. Clays and Clay minerals 37, 515–524.

Velbel, M.A., 1993. Temperature dependence of silicate weathering innature: how strong a negative feedback on long-term accumulationof atmospheric CO2 and global greenhouse warming? Geology 21,1059–1062.

Walker, J.C.G., Hays, P.B., Kasting, J.F., 1981. A negative feedbackmechanism for the long-term stabilization of Earth's surfacetemperature. Journal of Geophysical Research 86, 9776–9782.

West, A.J., Galy, A., Bickle, M., 2005. Tectonic and climatic controlson silicate weathering. Earth and Planetary Science Letters 235,211–228.

White, A.F., 2002. Determining mineral weathering rates based onsolid and solute weathering gradients and velocities: applicationto biotite weathering in saprolites. Chemical Geology 190,69–89.

White, A.F., Blum, A.E., 1995. Effect of climate on chemical weatheringin watersheds. Geochimica et Cosmochimica Acta 59, 1729–1747.

White, A.F., Brantley, S.L., 2003. The effect of time on the weatheringof silicate minerals: why do weathering rates differ in thelaboratory and field? Chemical Geology 202, 479–506.

White, A.F., Blum,A.E.,Marjorie, S., Schulz,M.S., Bullen, T.D., Harden,J.W., Peterson, M.L., 1996. Chemical weathering rates of a soil chro-nosequence on granitic alluvium: I. Quantification of mineralogicaland surface area changes and calculation of primary silicate reactionrates. Geochimica et Cosmochimica Acta 60, 2533–2550.

White, A.F., Blum, A.E., Bullen, T.D., Vivit, D.V., Schulz, M.,Fitzpatrick, J., 1999. The effect of temperature on experimental andnatural chemical weathering rates of granitoid rocks. Geochimica etCosmochimica Acta 63, 3277–3291.

White, A.F., Schulz,M.S., Loowenstern, J.B., Vivit, D.V., Bullen, T.D.,2005a. The ubiquitous nature of accessory calcite in granitoidrocks: implications for weathering, solute evolution, and petrogen-esis. Geochimica et Cosmochimica Acta 69, 1455–1471.

White, A.F., Schulz, M.S., Vivit, D.V., Blum, A.E., Stonestrom, D.A.,Harden, J.W., 2005b. Chemical weathering rates of a soil chrono-sequence on granitic alluvium: III. Hydrochemical evolution andcontemporary solute fluxes and rates. Geochimica et CosmochimicaActa 69, 1975–1996.

Williams, E.L., 2005. Carbonate cycling and mineral weathering intemperate forested watersheds: an integrated study of solution andsoil chemistry. Ph.D. dissertation. University of Michigan, AnnArbor, MI.

Williams, E.L., Walter, L.M., Ku, T.C.W., Baptist, K.K., Budai, J.M.,Kling, G.W., 2007a. Silicate weathering in temperate forest soils:insights from a field experiment. Biogeochemistry 82, 111–126.

Williams, E.L., Szramek, K.J., Jin, L., Ku, T.C.W., Walter, L.M.,2007b. The carbonate system geochemistry of shallow ground-water–surface water systems in temperate glaciated watersheds

http://dx.doi.org/10.1029/2006GC001337


(Michigan, USA): significance of open-system dolomite weath-ering. Geological Society of American Bulletin 119, 515–528.

Wolt, J.D., 1994. Soil Solution Chemistry: Applications to Environ-mental Science and Agriculture. John Wiley and Sons, Inc.

Wood, B.D., Keller, C.K., Johnston, D.W., 1993. In site measurementof microbial activity and controls on microbial CO2 production inuntreated zone. Water Resource Research 29, 647–659.

Yuretich, R.F., Batchelder, G.L., 1988. Hydrogeochemical cycling andchemical denudation in the Fort River Watershed, Central Massa-chusetts: an appraisal of mass-balance approaches. Water ResourcesResearch 24, 105–114.

mineral weathering rates in glacial drift soils (sw...

Documents