Decomposition of distinct organic matter pools is regulated bymoisture status in structured wetland soils

Chelsea Arnold a, Teamrat A. Ghezzehei a, b, *, Asmeret Asefaw Berhe a, b

a Environmental Systems Graduate Group, University of California, Merced, USAb Life and Environmental Sciences Group, University of California, Merced, USA

a r t i c l e i n f o

Article history:Received 21 July 2014Received in revised form27 October 2014Accepted 29 October 2014Available online 13 November 2014

Keywords:CO2 fluxSoil respirationHydrologyMeadowsClimate extremesSoil structure

a b s t r a c t

Peatlands are garnering much attention for their greenhouse gas feedback potential in a warmingclimate. As of yet, the coupled biogeochemical and hydrological processes that control the amount andtiming of soil organic matter (SOM) mineralization and, ultimately, whether peatlands will be sinks orsources of atmospheric CO2 are not fully understood. Soil structure is a key feature of soils that mediatesthe coupling between biogeochemical and hydrological processes. However, we know very little abouthow soil structure responds when soils are exposed to wettingedrying cycles outside their normal range.In order to better understand how high elevation peatlands will respond to increasingly dry years, weincubated soils from high elevation meadows in the Sierra Nevada at 5 different water potentials andmeasured the CO2 flux for over one year. We found that the cumulative carbon mineralization had a U-shaped pattern, with the greatest mineralization at the wettest (!0.1 bar) and driest (!4 bar) waterpotentials, across all hydrologic regions of the meadow. We propose a conceptual model that reproducesa similar pattern by incorporating the concept of dual porosity medium, with two distinct pore-sizepopulations representing inter- and intra-aggregate porosity. Availability of water and oxygen to thetwo pore-size populations depends on the soil's equilibrium water potential. The model and the datasuggest that the decomposition rates of intra-aggregate SOM may increase due to prolonged droughtevents that lead to accelerated release of C from previously untapped pool.

© 2014 Elsevier Ltd. All rights reserved.

1. Introduction

The accumulation of carbon in wetland soils, including mostpeatlands and high-elevation meadows in the Sierra Nevada andother mountain ranges, is primarily driven by low rates ofdecomposition as the soils stay near saturation for the bulk of thegrowing season (Clymo, 1965; Malmer, 1986). The low rate ofdecomposition coupled to a high degree of above and belowgroundproductivity results in an ecosystem sink of carbon dioxide fromthe atmosphere (Arnold et al., 2014).

The key soil environmental conditions that regulate the rate ofsoil OM decomposition include (a) soil water content and potential,(b) air-filled porosity, and (c) temperature (Linn and Doran, 1984;Baldock and Skjemstad, 2000; Schmidt et al., 2011). Soil watercontent indirectly regulates decomposition (Berhe, 2012; Berhe

et al., 2012) by controlling the diffusion of microbial substrate onthe dry end as well as by limiting oxygen (through air-filledporosity) to microorganisms on the wet end (Kaiser et al., 2015;Orchard and Cook, 1983). The degree of cell hydration and fluidexchange between microbial cells and soil is governed by waterpotential gradients (Stark and Firestone, 1995). Moreover, there isalso an interdependency between temperature and water effectson OMdecomposition, wheremaximum rates of soil respiration aredependent upon temperature (Wildung et al., 1975).

Each of the above parameters that determine the physicalenvironment of the soil are inherently functions of the architectureof the soil matrix (Kay, 1998; Baldock and Skjemstad, 2000; Rawlset al., 2003). This refers to the distribution and arrangement of soilpores in addition to soil particles and aggregates. In many soils, thepore-sizes exhibit unimodal probabilistic distribution with adistinct modal pore-size. While others, including macroporous andaggregated soils as well as most organic soils, exhibit bimodal ormultimodal distributions (populations) of soil pore sizes. While themechanisms that form multimodality are not fully understood, ithas been observed that this property is prevalent in soils with high

* Corresponding author. School of Natural Sciences, University of California,Merced, 5200 N Lake Rd, Merced CA 95343, USA.

E-mail address: TAGhezzehei@UCMerced.edu (T.A. Ghezzehei).

Contents lists available at ScienceDirect

Soil Biology & Biochemistry

journal homepage: www.elsevier .com/locate/soi lbio

http://dx.doi.org/10.1016/j.soilbio.2014.10.0290038-0717/© 2014 Elsevier Ltd. All rights reserved.

Soil Biology & Biochemistry 81 (2015) 28e37

SOM and high degree of aggregation (Ghezzehei, 2012) and getsaccentuated by wetting and drying cycles (Horn and Smucker,2005). The large and inter-connected macropores (inter-aggregatepores) in these soils belong to a different pore-size population fromthe finer intra-aggregate pores. The change in soil structure due tocompaction and shrinkage (Arnold And Ghezzehei, in revision)adds another layer of complication to the pore size distributions ofmultimodal soils. However, there is no general prediction thatcould be made for how compaction impacts pore size distributionacross soil taxa as the different pore-size populations responddifferently to the deforming stresses (Kutílek et al., 2006).

In multimodal soils, water and air content tend to vary spatiallyeven under equilibrium water potential conditions. Thus, differentportions of a given soil could experience markedly different rates ofmineralization. Although from the surface the relative contribu-tions of the different components are not discernible. As a result,the response of such soils to changing hydrologic and/or temper-ature regimes may appear contradictory and difficult to interpret.For example, Baldock and Skjemstad (2000) found that soils with ahigh degree of larger pore sizes exhibited increased rates of carbonmineralization as compared to soils with a smaller range of pores(at equivalent values of air-filled porosity). In another study, Torbertand Wood, (1992) found that the degree of water-filled pore spacein addition to the size and structure of pore space heavily influencemicrobial activity in the soil. Studies in the laboratory have alsoshown that a lowering of the water table can stimulate carbon di-oxide flux from peat microcosms (Moore and Dalva, 1993; Funket al., 1994; Blodau et al., 2004). However field studies haveshown mixed results with some reporting no significant change insoil respiration (Freeman et al., 1996), and yet others reportingimportance of microtopographical controls (Strack andWaddington, 2007), and even significant carbon losses (Oechelet al., 1998). In one study, field based measurements of ecosystemrespiration showed little to no response to drying whereas labo-ratory incubated peat cores showed a decrease in respiration withdrying at the surface (Lafleur et al., 2005). Drying in an Alaskan feninduced plant stress, which turned out to be more responsible forreducing the carbon sink in their ecosystem through a reduction ofgross primary productivity rather than an increase in ecosystemrespiration (Chivers et al., 2009). In one study, loss of porosity wascorrelated with reduction of OM mineralization (Franzluebbers,1999).

These disagreements clearly warrant an explanation based on asystematic study that evaluates how different moisture regimesinfluence decomposition in soils with multimodal pore-size dis-tributions. The objective of this study is to provide a mechanisticunderstanding of how SOM decomposition varies across a widerange of soil water content in highly organic soils with bi-modalpore-size distribution. We conducted a long-term soil incubationexperiment utilizing soil samples from high elevation meadows inthe Sierra Nevada mountain range. Gas fluxes from samples thatwere kept at five different water potentials (0.1e4.5 bar) werecollected and analyzed for over one year. In order to aid in syn-thesizing the various effects across the range, we introduce asimple modeling framework that incorporates water and air con-tent distributions in a bimodal soil (Section 2). Projections of thismodel adequately explain the trends observed in our experimentsas well as reported elsewhere.

2. Theory

2.1. Environmental controls on mineralization

Here we develop a simplified conceptual/mathematical modelthat integrates the interactions and feedbacks between soil

structure, structural dynamics, and hydrologic conditions in con-trolling soil organic matter decomposition. The model does notattempt to resolve temporal dynamics of organic matter minerali-zation or spatial distribution of concentrations and rates ofdecomposition. Rather, the overall goal is to mathematicallyrepresent how complexity of soil structure, through its influence onphysical environmental conditions, controls the relative degree ofmineralization over a wide range of moisture regimes.

Consider a macroscopically homogeneous soil volume that is atconstant and homogeneous water potential. The instantaneous rateof mineralization of soil carbon in this volume can be describedusing a “one-pool” model (Jenny, 1980)

dCdt

¼ kC (1)

where C (C-mass/soil-mass) is the quantity of mineralizable C and k(1/time) is the rate constant. Assuming that the soil remains underfairly constant environmental conditions (i.e., the rate constant andthe pool of mineralizable carbon remain unchanged), Eq. (1) can besolved to provide an exponential decay of the soil C content

C ¼ C0e!kt (2)

where C0 is the initial mineralizable C pool. Alternatively, thequantity of the mineralized C over the course of a given period ofmineralization under stable conditions can be described as

Ct ¼ C0!1! e!kt

"(3)

Note that C0 represents only the portion of the total SOM that isaccessible for mineralization under the equilibrium physicalconditions.

We postulate that the initial mineralizable pool (C0) and rateconstant (k) of any given soil that is incubated under equilibriumconditions are dependent on physical conditions. This assumptionis justified by the numerous studies that documented the depen-dence of fitted C0 and k on physical conditions such as temperatureand water-filled porosity (Linn and Doran, 1984; Kechavarzi et al.,2010a). Under isothermal conditions (which covers most labora-tory experiment) the total mineralized C over a period of t ¼ T canbe related to water-filled porosity and air-filled porosity as

Ct!T"¼ C0

!1! e!kT

"¼ S fW ðqÞ fAðaÞ (4)

where q is volumetric water content, a ¼ f ! q is volumetric aircontent, and f is porosity. The functions fW and fA describe the roleof water content and air contents as necessary variables that con-trol access to nutrients and oxygen. Specifically, the second term ofEq. (4) implies that mineralization is dependent on sufficient sub-strate diffusion (which is positively correlated with water content)as well as efficient gas-exchange (which increases with air-filledporosity). The proportionality constant S is a factor that accountsfor total SOM, SOM chemistry, and other soil biogeochemicalcharacteristics that are not directly dependent on soil water con-tent status. The factor S, varies depending on the quality andquantity of the SOM. Without quantitative understanding of thevalue of S, we cannot use the model for quantitative prediction.Nevertheless, it gives mechanistic explanation of how water-content and air-content influence mineralization.

For the purpose of illustrating these concepts, we define thefunctions fW and fA as linearly dependent on water-filled and air-filled fractions of the total porosity, respectively, as

fW ðqÞ ¼q

fand fAðqÞ ¼

af

(5)

C. Arnold et al. / Soil Biology & Biochemistry 81 (2015) 28e37 29

Then the relative quantity of C mineralized during the incuba-tion period is given by substituting Eq. (5) in Eq. (4),

CtðTÞS

¼ q$af2 ¼ q

f! q2

f2 (6)

Water content distribution in soils is primarily governed by theequilibrium water potential of the soil. At equilibrium, the waterpotential of typical soil sample used in incubation experiments isuniform throughout the volume. Therefore, it is more appropriateto express both water content and air content in terms of equilib-rium water potential J. The water retention characteristic, whichthe water potential J to water content q is commonly expressedusing the Van Genuchten (1980) equation as

Se ¼#1þ ðajÞn

$!m (7)

The empirical parameters a, n, and m ¼ 1 ! (1/n) determine theshape of the water-retention curve. This model essentially assumesa unimodal pore-size distribution. The effective saturationSe ¼ ðq! qrÞ=ðqs ! qrÞ depends on the saturated water content isassumed to be equal to the porosity (qs ¼ f) and qr the residual(irreducible) water content, which will be assumed negligible forthe mathematical expediency in the subsequent illustrative ex-amples. Thus, Eq. (7) can be re-written to express water filledporosity as

q

f¼

#1þ ð!ajÞn

$!m (7)

Eq. (7) exhibits an S-shaped increase in water-filled porositywith increase inwater potential (from large negative value to zero).The most significant change occurs in the vicinity of j ¼ !1/a,which represents the mode of the pore-size distribution. Thispattern is characteristic feature of unimodal pore size distribution.

The implication of this kind of system on the relative quantity ofC mineralized at different levels of soil water content is illustrated

in Fig. 1. The effect of water-filled and air-filled porosities (fWand fA,respectively) are depicted in Fig. 1b. In real soils, these functions areprobably different from simple straight lines, but they capture thekey role of water and air content. Specifically, fW highlights the factthat increase in water content promotes increased transport ofmicrobial communities as well as nutrient diffusion. It also impliesthat increase in water-filled porosity does bear intrinsic adverseaffect towards mineralization. Conversely, the function fA empha-sizes the need for efficient gas-exchange in respiration andmineralization. Because of the competitive interaction betweenthese two factors, the optimal condition for mineralization istypically in the mid-range water content. The model given in Eq.(6), which mathematically captures this competition, describes therelationship between water-filled porosity and relative minerali-zation as a down-facing parabola (Fig. 1b). This pattern is consistentwith several reported experiments (Linn and Doran, 1984) andmechanistic models (Grant and Rochette, 1994). For qualitativecomparison, we plotted experimental results of Linn and Doran(1984) on the secondary y-axis of Fig. 1b. This model satisfactorilycaptures the limiting effects of both air-content and water contentat extreme ends and indicates that mid-range water content asoptimal for mineralization.

2.2. Role of soil structure

Note that an underlying assumption behind the abovemodel (aswell as the single pool model) is that water, air and mineralizable Care uniformly distributed in the soil. However, this assumption failswhen dealing with structured soils with distinctly different pop-ulations of micro- and macro-pores. Soils with internal structuralheterogeneity often exhibit heterogeneous water and air content,even under equilibrium water potential conditions. Such soils canbe conceptualized as composites of multiple overlapping soils(continua) each with distinct water retention characteristics. Thisapproach is often employed, for example, when the intra- andinter-aggregate pores form distinct populations that drain atmarkedly different levels of water potential. The water retentioncharacteristics of these overlapping continua can be describedseparately using separate Van Genuchten (1980) equations withtheir sum representing the bulk composite soil (Durner, 1994).

q

f¼

X2

i¼1fi#1þ ðaijÞ

ni$!mi (7)

where i ¼ 1 and i ¼ 2 represent the macro- and micro-pores and f1and f2 are their corresponding porosities (f1 þ f2 ¼ f). Notethat !1/a1 > !1/a2 because the large inter-aggregate pores requireonly a slight drop in water potential in order to drain. Eq. (7) ex-hibits monotonous increase in water-filled porosity with twohumps near j ¼ !1/a1 and j ¼ !1/a2, which represent the modesof the inter- and intra-aggregate pore-size distributions.

2.3. Effect of consolidation

In addition to dual-porosity, certain structured soils especiallythose are formed in humid environments and/or with expansiveclays shrink as they dry. Most shrinkage occurs when the soil isfairly moist and malleable. Therefore, such shrinkage preferentiallyreduces the volume of the inter-aggregate pores, which drain firstwith very small reduction in water potential. Therefore, we canrepresent desiccation-induced loss of inter-aggregate porosity us-ing an exponential decay function.

Fig. 1. Illustrative model projections of effect of water-filled porosity on fW and fA (a)and relative quantity of C mineralization (b). For qualitative comparison experimentaldata of relative C mineralization from a study by Linn and Doran (1984) are plotted onsecondary y-axis in part (b).

C. Arnold et al. / Soil Biology & Biochemistry 81 (2015) 28e3730

f1ðjÞ ¼ fi1 þ fc

1e!rj (8)

where fi1 is the irreducible intra-aggregate porosity, fc

1 is thecompressible intra-aggregate porosity, and r is the exponentialdecay constant.

In Fig. 2, we illustrate the combined effect of structure andcompressibility on the relative amounts of C mineralized atdifferent water potential values. The parameters used for thisillustration are given in Table 1. The porosities (dotted lines) andwater retention characteristics (solid lines) of the inter- and intra-aggregate pore spaces are shown in Fig. 2a (blue and red lines,respectively). The total porosity and water retention characteristicof the composite are shown in black lines. Note that the inter-aggregate porosity is reduced by shrinkage at high water poten-tial values. Likewise, most of the inter-aggregate pores arecompletely drained at water potential of >!1 bar. In contrast, thefine intra-aggregate pores remain completely saturated until thewater potential is reduced below!1 bar. The corresponding fW andfA function for both the component pore domains are shown inFig. 2b. As previous, both are linearly related to the water-filled and

air-filled porosities. The optimal mixture of air and water in theinter-aggregate pores occurs at approximately !0.05 bar. Thisrepresents fairly wet condition at the bulk soil leveldexcept for theinter-aggregate pores, which are moderately drained. In contrast,the intra-aggregate pore spaces achieve optimal mixture of air andwater content at approximately !5 bar. The resulting relativequantities of C mineralization at different levels of water potentialare shown in Fig. 2c. These plots indicate that peak Cmineralizationin the inter- and intra-aggregate pore space occurs distinctlydifferent levels of water potential. The combined curve, whichdenotes the response of the bulk soil, is characterized by double-peaks. Note the parameter S in Eq. (4), which denotes quality/quantity of SOM, microbial communities etc, is likely to be differentin the micro and macropores. Therefore, the relative contributionsof the two pools is further dependent on this differential distribu-tion of SOM as well as differences in SOM quality and microbialcommunities. Nevertheless, the relative proportions of mineraliz-able C follow the double peaks shown in Fig. 2c. To experimentallytest this pattern, one needs to conduct Cmineralization experimentof structured soils over wide range of water potentials and withsufficient increments over the entire range. We are aware of onlyone such study, which investigated CO2 efflux from agriculturalpeat soils (Kechavarzi et al., 2010b). Within the range of waterpotential covered in that study (!1 to !0.01 bar), the majority ofpeat soils depict one distinct peak near the wet end and moderaterise at lower water content. These observations appear to be inqualitative agreement with the projections of the above conceptualmodel.

3. Methods

3.1. Site description

Soils for the incubation were collected from a meadow at3200 m elevation in the Harvey Monroe Hall Research Natural Area(Hall RNA). The Hall RNA is located at the crest of the Sierra NevadaMountains, on the eastern side of the Sierras adjacent to YosemiteNational Park. It is characterized by a cool climate with cold wetwinters and cool dry summers. Mean daily temperatures rangefrom !4.9 &C to 12.9 &C and average precipitation is 642 mm/year(Taylor, 1984). The soils in the Hall RNA are characterized asInceptisols with suborders Andic Cryumbrepts and LithicCryumbrepts.

Meadows in the high elevation ecosystems in the Sierras arecharacterized by hydrologic gradients that are defined by thegeomorphology of the landscape. In the fall of 2011, bulk meadowsoils were collected from across the hydrologic gradient in thestudy meadow. The hydrologic gradient in the meadows wasdelineated by vegetation (Allen-Diaz, 1991), with dry meadowsassociated with Carex filifolia, intermediate meadows with Ptila-grostis kingii and wet meadows associated with Carex vesicaria/

Fig. 2. Illustrative model projections of effect of wide-range of water potential onrelative C mineralization in structured and compressible soils: (a) water retentioncharacteristics and porosity, (b) fW and fA, and (c) relative quantity of C mineralization.Blue and red lines denote inter- and intra-aggregate porosities and black lines are thecurves for the composite (bulk) soil. The parameters used are reported in Table 1.

Table 1Soil properties for the Hall RNA soils are depicted in the table. Standard error isreported for pH and bulk density measurements. Particle size distribution was notcompleted on the wet meadow 0e10 cm layer due to peat layer.

Parameter Unit Value

a1 bar!1 50a2 bar!1 0.29n1 e 2n2 e 1.8fi1 e 0.2

fr2 e 0.1

f2 e 0.6r e !20

C. Arnold et al. / Soil Biology & Biochemistry 81 (2015) 28e37 31

utriculata and Carex scopulorum. At ten random locations in each ofthe wet, intermediate and dry regions of the meadow, bulk soilsamples were collected from the following three depths: 0e10 cm,10e20 cm, and 20e40 cm (Fig. 3). The wet meadow soils areclassified as a fibrous peat in the surface layer, but the intermediateand dry locations had mineral soils with high organic matter con-tent. At the dry meadow locations, the depth of the lower soilsample only extended to 30 cm. These soils were packed on ice andtransported to the laboratory at the University of California,Merced, where they were composited by location (dry, intermedi-ate, wet) and depth, sieved to 8 mm to remove large roots andgravels and repacked to native bulk density in 6.35 cmtall' 5.08 cmwide plastic tubes. The repacked soil cores were thenallowed to saturate from below overnight and weighed.

3.2. CO2 measurements

Three replicate cores of each layer (n ¼ 135) were drainedrespectively to one of the following water potentials(J ¼ !0.1, !0.3, !1.0, !2.5 or !4 bars) on a pressure plate appa-ratus following methods of Klute, (1986). These water potentialswere chosen to represent a range of potential dryness levels thatcould be expected in these soils. The dry meadow regions mayexperience dryness of !4 bar during a hot summer, but this levelwould not be experienced in the intermediate and wet meadowregions in a normal year. It would take severe desiccation for!4 barto be reached in those regions of the meadow. The repacked coreswere then placed in one-liter mason jars, covered with plastic wrapand incubated at 20 &C in the dark for over one year. The jars wereweighed weekly and water was added to maintain the initialdrained weight. Gas samples were pulled at the following intervals

(day 1, 3, 7, 14, 28, 42, 63, 99, 140, 248, 392) by capping the masonjars for 24 h and extracting 15 ml of gas through a septa inserted inthe jar lid. These samples were then analyzed on a gas chromato-graph, (Shimadzu GC2014) fitted with a thermal conductivity de-tector for determination of carbon dioxide concentrations.

3.3. Soil properties

Soil texture was analyzed at the University of CaliforniaAnalytical Laboratory by the hydrometer method (Sheldrick andWang, 1993). Soil pH was analyzed on three separate replicates ina 1:1 water dilution (Thomas et al., 1996). Percent carbon and ni-trogen were analyzed in duplicate on ground bulk soil samples

Fig. 3. Bulk soils and intact cores were collected from three main hydrologic regions inthe meadow (dry, intermediate and wet) at three different depths. The sample iden-tifications referred to in the remainder of the manuscript are illustrated in the figure,with D, I, or W, referencing the dry, intermediate or wet site and the T, M, or Breferencing the (0e10 cm), (10e20 cm) or (20e30 cm edry only and 20e40 cm)depths.

Table 2Soil properties for the Hall RNA soils are depicted in the table. Standard error is reported for pH and bulk density measurements. Particle size distributionwas not completed onthe wet meadow 0e10 cm layer due to peat layer.

Depth % Sand % Silt % Clay Total % C Total % N C:N pH (H20) Bulk density

Dry 0e10 67 28 5 5.7 ± 0.08 0.39 ± 0.01 14.8 ± 0.02 3.8 ± 0.42 0.89 ± 0.1710e20 67 27 6 3.3 ± 0.04 0.24 ± 0.01 13.9 ± 0.15 4.2 ± 0.35 1.26 ± 0.1120e30 65 26 9 2.5 ± 0.04 0.19 ± 0.0 13.4 ± 0.18 4.5 ± 0.32 1.56 ± 0.14

Intermediate 0e10 71 23 6 10.4 ± 0.12 0.7 ± 0.01 14.8 ± 0.04 4.5 ± 0.06 0.51 ± 0.0610e20 64 31 5 3.2 ± 0.02 0.24 ± 0.01 13.7 ± 0.35 5.2 ± 0.03 1.04 ± 0.0620e40 61 32 7 2.3 ± 0.01 0.17 ± 0.0 13.6 ± 0.03 5.4 ± 0.07 1.25 ± 0.03

Wet 0e10 a a a 33.5 ± 0.07 2.2 ± 0.01 15.3 ± 0.06 5.3 ± 0.13 0.17 ± 0.0110e20 64 32 4 12.6 ± 0.05 0.89 ± 0.01 14.1 ± 0.12 5.7 ± 0.03 0.46 ± 0.0720e40 73 25 5 10.3 ± 0.08 0.66 ± 0.0 15.7 ± 0.11 4.8 ± 0.62 0.54 ± 0.16

a Not tested.

Fig. 4. Water retention curves for the intact soils cores fitted with Durner's multi-modal retention function (Eq. (7)).

C. Arnold et al. / Soil Biology & Biochemistry 81 (2015) 28e3732

from each location and depth in the meadow on a CostechElemental Analyzer (Costech Analytical Technologies, Inc.).

At the same time that the bulk soils were collected, we extractedtwo intact and paired soil cores from the same three depths in thelocations directly adjacent to where the bulk samples werecollected. One sample core of each pair was saturated overnightand drained to !0.3 bar on the pressure plate apparatus. The finaldrainedweight wasmeasured and the coreswere incubated in one-liter mason jars for 392 days with the rest of the compositedsamples. Water was added each week to maintain the constantdrained weight. Gas samples were drawn at intervals for over oneyear at the same time as the composited cores. The second set ofcores were saturated overnight in the laboratory and drained on apressure plate apparatus over a month long period to determinethe water retention curve for the samples. The resulting waterretention data was fitted with Durnerevan Genuchten waterretention model (Van Genuchten, 1980; Durner, 1994), which isappropriate for structured soils that exhibit bi-modal pore sizedistribution (Eq. (7)). The water retention curves were generatedusing only the intact cores to gain an accurate depiction of the fieldstate of the soils. Bulk density values were estimated from thewater retention curve data.

3.4. Data analysis

In order to determine the initial pool of labile carbon availablefor decomposition, the cumulative carbon mineralization data wasfitted with an exponential rise to max curve (Eq. (3)). This one poolmodel carries the assumption that all of the carbon in the sample

decays at the same rate. A two-pool model was tested on the data,but since the majority of data could be fitted with the one poolmodel, we chose to utilize the one pool model for all the samples.

A one-way ANOVA, was performed to compare the differencebetween means rate of carbon dioxide mineralization across waterpotentials. If the model was significant (p < 0.1), a Tukey HSD posthoc test was performed in order to determine which treatmentswere significant. All statistical analysis was done on the R statisticalsoftware (rproject.org).

4. Results

4.1. Soil properties

The soils in this regionwere classified as a sandy loam, with highpercentage of organic matter. The percent carbon in top 0e10 cmdepth ranged from approximately 6%e34% across the hydrologicgradient (dry to wet) in the meadow. While the percent carbonbelow 10 cm rapidly dropped to 2e3% carbon in the dry and in-termediate meadow regions, the wet meadow region maintainedhigh (10e12%) carbon values at similar depths (Table 2). Bulkdensity ranged from 0.89 to 0.17 g/cm3 moving from dry to wetacross the meadow in the top 10 cm of the soil. The bulk density ofthe wet meadow soil increased to a maximum of 0.54 g/cm3 at adepth of 40 cm, compared to approximately 1.56 g/cm3 and 1.25 g/cm3 in the dry and intermediate regions respectively. The soils arehighly to moderately acidic with pH values ranging from 3.8 to 5.7.

An examination of the water retention curves from the series ofintact cores showed that there is an initial large drop in water

Fig. 5. Mean cumulative CO2eC evolution data in mg C/g soil for the composited cores with the error bars representing standard error of three replicates. The one pool model (Eq.(3)) was fitted to the average of the three replicates. The columns represent soils drained to various water potentials (0.1, 0.3, 1.0, 2.5, and 4.0 bar) while the rows represent the depthin the soil profile (0e10 cm, 10e20 cm, and 20e40 cm).

C. Arnold et al. / Soil Biology & Biochemistry 81 (2015) 28e37 33

content as the soils progress from saturation to !0.05 bar (Fig. 4).After the large pores are rapidly drained, the subsequent release ofwater by these sandy organic soils became gradual as the dryingprogressed. There was a secondary pulse of water released as thesoils experience further drying at !4.0 bar. This is characteristic ofsoils that exhibit a bi-modal pore size distribution. This bimodalpattern was consistently observed in all the soils (across depth andmoisture classes).

Note that the water retention curve of the composited samples,which were not separately determined, could be different from thatof the intact cores because some stones (20e30 cm layer of drymeadow soil only) and larger roots were removed during repack-ing. Nevertheless, because the samples were handled very gentlyand packed close to their field bulk density we assume the bimo-dality of the water retention curves was preserved. Therefore, theabove patterns of the water retention curves can be safely used to

make a qualitative comparison with C mineralization pattern pro-jected by the conceptual model introduced earlier in this paper(Fig. 2).

4.2. Carbon mineralization

Overall, we found the largest carbon loss in the top 0e10 cm ofthe wet region of the meadow (between 30 and 40 mg C/g soil) atall water potentials (Fig. 5). The intermediate and dry meadowregions also showed the highest carbon loss in the top 0e10 cm.Below 10 cm, there was a reduction in the carbon lost in all regionsof the meadow and a convergence of values across the dry throughwet regions. The intact series of cores showed similar trends, butslightly higher carbon loss in the wet meadow 0e10 cm depth thanwas found in the composited cores (Fig. 6).

Averaged across the different water potentials, there was asignificant loss of carbon in all of the dry cores, the intermediate0e10 cm depth, and the wet (0e10 cm and 20e40 cm) depths(Table 3). There was a decreasing trend in the total amount ofcarbon lost in the wet meadow (WT) up to !2.5 bar and then asignificant increase at !4.0 bar (Fig. 7). Although the trend isevident at depth in the wet meadow, only the 20e40 cm depthshowed significant differences in carbon mineralization among thedifferent water potentials (p < 0.05). In this case, the lowest carbonmineralization values were found at the !0.3 and !1.0 bar waterpotentials, with a subsequent increase at !2.5 and !4.0 bar. Thisgeneral trend of the lowest carbon mineralization in the !2.5 barfollowed by a significant increase was also found in the dry andintermediate regions of the meadow. The intermediate region hadthe highest values of carbonmineralization at the 0e10 cm depth inthe !4.0 bar treatment. The effect was attenuated at depth and nosignificant differences among water potentials was found. In thedry region of the meadow, there were significant differencesapparent at all depths (0e10 cm,10e20 cm and 20e30 cm). The dry0e10 cm and 10e20 cm depths displayed the highest flux inthe !0.1 bar treatment followed by decreasing carbon minerali-zation until !2.5 bar. There was a significant increase from !2.5to!4 bar above 20 cm, but no significant difference below 20 cm. Acomparison of the total carbon loss from the composited and intactcores indicates a close alignment of values, with the intact coresshowing slightly higher carbon loss than the composited cores(Fig. 8).

Each replicate was fitted with the one pool model and a modelutilizing the average of fitted parameters is plotted along with thedata in Fig. 6. The means and standard-errors of the fitted initiallabile carbon pool (C0) and rate constant (k) are reported in Fig. 9.Note that the fitted C0 was largest at !0.1 and !4.0 bar in the0e10 cm and 10e20 cm depths of the dry and wet meadow soils(Fig. 9). The intermediate meadow region showed a less pro-nounced trend in the top layer, but the largest amount of C wasmineralized in the !2.5 to !4.0 bar treatments. Below 10 cm inthe intermediate meadow, the labile carbon pool was largest inthe !2.5 and !4.0 bar treatments. The fraction of the labile carbonpool to the total carbon pool in the meadow was similar above20 cm in the dry meadow, with labile carbon making up to15e20% of the total amount of carbon in the !0.1 and !4.0 bartreatments. The wet meadow showed a similar trend although the10e20 cm depth showed the largest proportion of labile tototal carbon. The fraction of labile carbon in the intermediateregion of the meadow was highest below the 10 cm depth inthe soil in the !2.5 and !4.0 bar treatments. The fitted decom-position rate constant (k) showed the highest rate of decompo-sition at the !0.3 bar treatment across all moisture regions anddepths.

Fig. 6. Mean cumulative CO2eC evolution data in mg C/g soil for the intact cores (driedto 0.3 bar prior to incubation) with the error bars representing standard error. The onepool model (Eq. (3)) was fitted to the average of the three replicates.

Table 3One-way ANOVA results of final C loss. The degrees of freedom of the treatments(water potential) and the replicates are 4 and 10, respectively.

Moisture Regime

Dry Intermediate Wet

Depth Top F ¼ 17.3, P < 0.0002 F ¼ 6.1, P < 0.01 F ¼ 2.8, P < 0.09Middle F ¼ 8.5, P < 0.003 F ¼ 0.8, P ¼ 0.533 F ¼ 2.5, P ¼ 0.11Bottom F ¼ 2.7, P < 0.091 F ¼ 1.4, P ¼ 0.295 F ¼ 3.5, P < 0.05

C. Arnold et al. / Soil Biology & Biochemistry 81 (2015) 28e3734

5. Discussion

The results clearly show that in most regions of themeadow, thelargest carbon mineralization occurs when the meadow is very wet

(0.1 bar). This contradicts the typical inverted-U trend of mineral-ization with changing water-filled porosity (Linn and Doran, 1984;Grant and Rochette, 1994; Baldock and Skjemstad, 2000; VanHuissteden et al., 2006) and illustrated in Fig. 1b for unimodalsoil. In fact, what we observed is consistent with the projections ofthe bimodal model, which exhibits two peaks of mineralization athigh and low bulk soil moisture status. The soil water retentioncharacteristics shown in Fig. 4 suggest that the macropores of allthe soils were significantly drained at >!0.1 bar water potential.Therefore, the peak mineralization of the C held in the macroporeslikely occurs at !0.1 bar or higher because this range permitsmotility of microorganisms (Carson et al., 2010) and diffusion ofnutrients in the available water as well as efficient gas-exchangethrough the drained macropores. Once the soil was drainedbelow !1 bar, the moisture that remains in the macropores is mostlikely to be patchy and low in quantity as suggested by the waterretention curves. Therefore, the treatments drained below!2.5 barare expected to exhibit decreasing pool of labile SOM (Skopp et al.,1990) held in the macropores.

In contrast, the micropores remained fully saturated until thesoils dried below !0.5 bar. In fact, the most change in the micro-pore water content was observed at <!2 bar for most of the soils.Therefore, SOM mineralization in the micropores was not likely to

Fig. 7. The final carbon loss (mg C/g soil) from the composited cores with error bars representing standard error of three replicates. Lower case letters represent significancedetermined via a Tukey post hoc test (p < 0.1).

Fig. 8. The comparison of total carbon loss from composited versus intact soil cores,with the black line indicates the 1:1 line between the samples. The vertical error barsrepresent the standard error (n ¼ 3) of the intact soil cores while the horizontal errorbars represent the standard error (n ¼ 3) of the composited cores.

C. Arnold et al. / Soil Biology & Biochemistry 81 (2015) 28e37 35

be significant contributor in the !0.1, !0.3, and !1 bar treatments.But as air invades the micropores and oxygen becomes readilyaccessible to the decomposers, the pool of labile SOM in the mi-cropores increases and can explain the high C mineralization weobserved the driest water potential (!4.0 bar). This phenomenonwas also observed in a study by Kechavarzi et al. (2010a) thatshowed increased carbon mineralization with drying. The cumu-lative carbon mineralization at 4.0 bar was comparable to that ofthe 0.1 bar treatment in most cases. To verify that this was not anartifact of the reconstruction of the cores, we examined the cor-relation between the intact and composited cores. As shown inFig. 8, the intact cores were consistent with the composited sam-ples, with only a slight but consistent shift above the 1:1 line.

The slightly higher C loss from the intact cores can be explainedby the fact that the intact cores contained coarser fractions(including larger root fragments) and undisturbed structure.However, the pattern of loss was similar to the reconstructed cores,in that we saw large C loss at the wettest and the driest waterpotentials. This observation partly supports the assumption that

the water retention curves of the composited samples wereconsistent with that of the intact cores.

When considering both pools together, themodels as well as thedata indicate that the mid-range water content (!1 barand !2.5 bar) is suboptimal for mineralization because themineralization is limited by lack of adequate water or oxygen in themacropores and micropores, respectively. This mechanistic expla-nation also appears to be sufficiently robust to explain the seem-ingly paradoxical observations of several studies that weresummarized at the outset of this paper.

The notion of having two pools (protected and unprotected) oflabile carbon is not new and has been previously applied to explainmineralization in aggregated grassland and/or cultivated soils(Beare et al., 1994; Mikha and Rice, 2004). What this study adds isthat soil moisture regime plays a significant role as a switchingmechanism that activates and deactivates the different pools atseparatemoisture levels. Specifically, themodel and data presentedin this work indicate the presence of a threshold soil moisture statethat separates the two pools.

Fig. 9. Best fit model parameter (Co, Co:Ca, and k), where Co represents the pool of labile carbon, Co:Ca represents the ratio of the initial labile pool to the total carbon pool (Ca), and krepresents the decomposition constant (day!1). Error bars represent the standard error of 3 replicates.

C. Arnold et al. / Soil Biology & Biochemistry 81 (2015) 28e3736

For the particular high elevation meadow soils we studied, thedegree of drying required to exploit the labile pool of carbon of themicropores in the intermediate to wet soils may not occur in anormal year. However, if these systems were severely desiccated(for example during extreme droughts), the secondary pool of labilecarbon would become accessible and result in a new pulse of CO2.

6. Conclusion

This study highlights the important controls that soil structureimposes on the coupled biogeochemical and hydrological in-teractions of organic soils. We showed that the presence of hydro-logical thresholds in these soilswith bi-modal pore-size distributioncan enhance carbon mineralization in response to drying. This hasenormous implicationsnotonly for thesehighelevationecosystems,but forhigh latitude ecosystemsandother soils that experiencebi-tomulti-modal porosity as well. In these soils, we may see an initialincrease in carbon mineralizationwith desiccation, but that may bereduced until a critical threshold of drying is reached. With an in-crease in drought-like extreme conditions in the Western UnitedStates, there is potential for this second pool of labile carbon to bemineralized in the high elevation wetlands. This would lead towidespread meadow degradation and a loss of ecosystem servicesthat these ecosystemsprovide in the context ofmountainhydrology.Furthermore, this result warrants further look into the coupled hy-drology and biogeochemistry of wet, organic-rich soils in high alti-tude and high latitude ecosystems that are expected to experiencedrying due to anticipated changes in climate.

Acknowledgment

The authors thank Jesseca Burkhart for her valuable contribu-tions to the laboratory and field work. Funding for this work wasprovided from startup funds to AAB and TAG, and GraduateResearch Council grant from UC Merced. The authors have noconflict of interest to declare.

References

Allen-Diaz, B.H., 1991. Water table and plant species relationships in Sierra Nevadameadows. American Midland Naturalist 126, 30e43.

Arnold, C., Ghezzehei, T.A., Berhe, A.A., 2014. Early spring, severe frost events, anddrought induce rapid carbon loss in high elevation meadows. Plos One 9,e106058.

Arnold, C.L., Ghezzehei, T.A., 2014. A method for characterizing desiccation inducedconsolidation and permeability loss of soft soils. Water Resources Research.http://dx.doi.org/10.1002/2014WR015745 (in press).

Baldock, J.A., Skjemstad, J., 2000. Role of the soil matrix and minerals in protectingnatural organic materials against biological attack. Organic Geochemistry 31,697e710.

Beare, M., Hendrix, P., Cabrera, M., Coleman, D., 1994. Aggregate-protected andunprotected organic matter pools in conventional-and no-tillage soils. SoilScience Society of America Journal 58, 787e795.

Berhe, A.A., 2012. Decomposition of organic substrates at eroding vs. depositionallandform positions. Plant and Soil 1e20. http://dx.doi.org/10.1007/s11104-011-0902-z.

Berhe, A.A., Suttle, K.B., Burton, S.D., Banfield, J.F., 2012. Contingency in the directionand mechanics of soil organic matter responses to increased rainfall. Plant andSoil 358, 371e383. DOI 310.1007/s11104-11012-11156-11100.

Blodau, C., Basiliko, N., Moore, T.R., 2004. Carbon turnover in peatland mesocosmsexposed to different water table levels. Biogeochemistry 67, 331e351.

Carson, J.K., Gonzalez-Qui~nones, V., Murphy, D.V., Hinz, C., Shaw, J.A., Gleeson, D.B.,2010. Low pore connectivity increases bacterial diversity in soil. Applied andEnvironmental Microbiology 76, 3936e3942.

Chivers, M., Turetsky, M., Waddington, J., Harden, J., McGuire, A., 2009. Effects ofexperimental water table and temperature manipulations on ecosystem CO2fluxes in an Alaskan rich fen. Ecosystems 12, 1329e1342.

Clymo, R., 1965. Experiments on breakdown of Sphagnum in two bogs. The Journalof Ecology 53, 747e758.

Durner, W., 1994. Hydraulic conductivity estimation for soils with heterogeneouspore structure. Water Resources Research 30, 211e223.

Franzluebbers, A., 1999. Microbial activity in response to water-filled pore space ofvariably eroded southern Piedmont soils. Applied Soil Ecology 11, 91e101.

Freeman, C., Liska, G., Ostle, N., Lock, M., Reynolds, B., Hudson, J., 1996. Microbialactivity and enzymic decomposition processes following peatland water tabledrawdown. Plant and Soil 180, 121e127.

Funk, D.W., Pullman, E.R., Peterson, K.M., Crill, P.M., Billings, W., 1994. Influence ofwater table on carbon dioxide, carbon monoxide, and methane fluxes fromtaiga bog microcosms. Global Biogeochemical Cycles 8, 271e278.

Ghezzehei, T., 2012. Soil structure. In: Huang, P.M., Li, Y., Sumner, M.E. (Eds.),Handbook of Soil Science, Properties and Processes, vol. 1. CRC Press, BocaRaton, Fla, pp. 1e17.

Grant, R., Rochette, P., 1994. Soil microbial respiration at different water potentialsand temperatures: theory and mathematical modeling. Soil Science Society ofAmerica Journal 58, 1681e1690.

Horn, R., Smucker, A., 2005. Structure formation and its consequences for gas andwater transport in unsaturated arable and forest soils. Soil and Tillage Research82, 5e14.

Jenny, H., 1980. The Soil Resource, Origin and Behaviour. Springer, New York.Kaiser, M., Kleber, M., Berhe, A.A., 2015. How air-drying and rewetting modify soil

organic matter characteristics: an assessment to improve data interpretationand inference. Soil Biology & Biochemistry 80, 324e340. http://dx.doi.org/10.1016/j.soilbio.2014.10.018.

Kay, B., 1998. Soil structure and organic carbon: a review. In: Soil Processes and theCarbon Cycle. CRC Press, Boca Raton, Fla.

Kechavarzi, C., Dawson, Q., Bartlett, M., Leeds-Harrison, P., 2010a. The role of soilmoisture, temperature and nutrient amendment on CO2 efflux from agriculturalpeat soil microcosms. Geoderma 154, 203e210.

Kechavarzi, C., Dawson, Q., Leeds-Harrison, P., 2010b. Physical properties of low-lying agricultural peat soils in England. Geoderma 154, 196e202.

Klute, A., 1986. Water Retention: Laboratory Methods, Methods of Soil Analysis, Part1. Physical and Mineralogical Methods, second ed. Soil Science Society ofAmerica, Madison, WI, pp. 635e662.

Kutílek, M., Jendele, L., Panayiotopoulos, K.P., 2006. The influence of uniaxialcompression upon pore size distribution in bi-modal soils. Soil and TillageResearch 86, 27e37.

Lafleur, P., Moore, T., Roulet, N., Frolking, S., 2005. Ecosystem respiration in a cooltemperate bog depends on peat temperature but not water table. Ecosystems 8,619e629.

Linn, D., Doran, J., 1984. Effect of water-filled pore space on carbon dioxide andnitrous oxide production in tilled and nontilled soils. Soil Science Society ofAmerica Journal 48, 1267e1272.

Malmer, N., 1986. Vegetational gradients in relation to environmental conditions innorthwestern European mires. Canadian Journal of Botany 64, 375e383.

Mikha, M.M., Rice, C.W., 2004. Tillage and manure effects on soil and aggregate-associatedcarbonandnitrogen. Soil ScienceSocietyofAmerica Journal68, 809e816.

Moore, T., Dalva, M., 1993. The influence of temperature and water table position oncarbon dioxide and methane emissions from laboratory columns of peatlandsoils. Journal of Soil Science 44, 651e664.

Oechel, W.C., Vourlitis, G.L., Hastings, S.J., Ault, R.P., Bryant, P., 1998. The effects ofwater table manipulation and elevated temperature on the net CO2 flux of wetsedge tundra ecosystems. Global Change Biology 4, 77e90.

Orchard, V.A., Cook, F., 1983. Relationship between soil respiration and soil mois-ture. Soil Biology and Biochemistry 15, 447e453.

Rawls, W., Pachepsky, Y.A., Ritchie, J., Sobecki, T., Bloodworth, H., 2003. Effect of soilorganic carbon on soil water retention. Geoderma 116, 61e76.

Schmidt, M.W.I., Torn, M.S., Abiven, S., Dittmar, T., Guggenberger, G., Janssens, I.,Kleber, M., Kogel-Knabner, I., Lehmann, J., Manning, D.A.C., Nannipieri, P.,Rasse, D.P., Weiner, S., Trumbore, S., 2011. Persistence of soil organic matter asan ecosystem property. Nature. http://dx.doi.org/10.1038/nature10386.

Sheldrick, B.H., Wang, C., 1993. Particle-size distribution. In: Carter, M.R. (Ed.), SoilSampling and Methods of Analysis. Canadian Society of Soil Science. LewisPublishers, Ann Arbor, MI., pp. 499e511.

Skopp, J., Jawson, M., Doran, J., 1990. Steady-state aerobic microbial activity as a func-tion of soil water content. Soil Science Society of America Journal 54, 1619e1625.

Stark, J.M., Firestone, M.K., 1995. Mechanisms for soil moisture effects on activity ofnitrifying bacteria. Applied and Environmental Microbiology 61, 218e221.

Strack, M., Waddington, J., 2007. Response of peatland carbon dioxide and methanefluxes to a water table drawdown experiment. Global Biogeochemical Cycles 21,GB1007. DOI 1010.1029/2006GB002715.

Taylor, D.W., 1984. Vegetation of the Harvey Monroe Hall Research Natural Area,Inyo National Forest, California. Unpub. report for the US Forest Service, PacificSouthwest Forest and Range Experiment Station.

Thomas, G., Sparks, D., Page, A., Helmke, P., Loeppert, R., Soltanpour, P.,Tabatabai, M., Johnston, C., Sumner, M., 1996. Soil pH and Soil Acidity. Methodsof Soil Analysis. Part 3-Chemical Methods, pp. 475e490.

Torbert, H., Wood, C., 1992. Effects of soil compaction and water-filled pore space onsoil microbial activity and N losses. Communications in Soil Science & PlantAnalysis 23, 1321e1331.

Van Genuchten, M.T., 1980. A closed-form equation for predicting the hydraulicconductivity of unsaturated soils. Soil Science Society of America Journal 44,892e898.

Van Huissteden, J., van den Bos, R., Alvarez, I.M., 2006. Modelling the effect ofwater-table management on CO2 and CH4 fluxes from peat soils. NetherlandsJournal of Geosciences 85, 3e18.

Wildung, R., Garland, T., Buschbom, R., 1975. The interdependent effects of soiltemperature and water content on soil respiration rate and plant root decom-position in arid grassland soils. Soil Biology and Biochemistry 7, 373e378.

C. Arnold et al. / Soil Biology & Biochemistry 81 (2015) 28e37 37