lable at ScienceDirect

Soil Biology & Biochemistry 78 (2014) 90e96

Contents lists avai

Soil Biology & Biochemistry

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

Reduction of air- and liquid water-filled soil pore space with freezingexplains high temperature sensitivity of soil respiration below 0 �C

Colin Tucker*

University of Wyoming, Botany Department and Program in Ecology, 1000 E. University Avenue, Laramie, WY, 82071, United States

a r t i c l e i n f o

Article history:Received 13 February 2014Received in revised form16 June 2014Accepted 25 June 2014Available online 15 July 2014

Keywords:Heterotrophic soil respirationWinter ecologyDual Arrhenius MichaeliseMenten modelSoil organic matterSnowpack insulation

* Present address: University of Alaska, Fairbanks, IN. Koyukuk Drive, P.O. Box 757000, Fairbanks, AK, 99

E-mail addresses: cltucker@alaska.edu, colinleetuc

http://dx.doi.org/10.1016/j.soilbio.2014.06.0180038-0717/© 2014 Elsevier Ltd. All rights reserved.

a b s t r a c t

At temperatures just below 0 �C, the temperature sensitivity of heterotrophic soil respiration (RH) isorders of magnitude higher than above 0 �C. Two primary mechanisms have been proposed for thishigh sub-zero temperature sensitivity: changes in soil microbial community composition and physiology,or the physical effects of the transition of water between liquid and ice phases. In this study, the effect ofsoil freezing on RH was modeled using a simple modification of the Dual Arrhenius MichaeliseMentenmodel, to account for both the reduced liquid water content of the soil pore space and the reduced air-filled pore space as water expands during freezing. Using parameters derived from previous studies,RH was modeled at a range of sites throughout Southeast Wyoming, ranging from prairie to highelevation forest. Across the study region, RH at sub-zero temperatures was low in the prairie(0.002 mg C m�2 h�1 at �1 �C and optimal water content) and sagebrush (5$10�7 to 0.012 mg C m�2 h�1

at �1 �C and optimal water content) sites, with lower organic matter and higher sand content, and muchhigher in the high sub-alpine forest (0.71 mg C m�2 h�1 at �1 �C and optimal water content) andmeadow (3.5 mg C m�2 h�1 at �1 �C and optimal water content) sites with high soil organic mattercontent. The modeled Q10 (the multiplicative response of RH to a 10 �C increase in temperature) abovefreezing was ~3.2, while below freezing the median value ranged from 15 to 255, and the maximum was1.6$1024. These values capture the range of Q10's described in the literature, suggesting that the modelbased on changing liquid water contented presented here can explain much of these observed apparenttemperature responses. Hence, this model may prove valuable for predicting soil C fluxes in environ-ments that undergo seasonal freezing.

© 2014 Elsevier Ltd. All rights reserved.

1. Introduction

In regions at high elevation and latitude that experience long,cold winters, heterotrophic respiration (RH) from snow-coveredsoils may be a quantitatively important part of the annual carbon(C) cycle (Sommerfeld et al., 1993). While RH is generally low at coldtemperatures (Lloyd and Taylor, 1994), over the duration of thesnow-covered period the cumulative winter efflux of CO2 may be asmuch as 50% of the annual flux (Grogan and Chapin, 1999). Thisefflux is, however, highly sensitive to soil temperature. A number ofstudies have found that just below 0 �C, the temperature sensitivityof soil respiration increases sharply (Mikan et al., 2002; Monsonet al., 2006; Tilston et al. 2010; Schmidt et al., 2009). The Q10

(multiplicative increase in reaction rate to a 10 �C temperatureincrease) of soil respiration above 0 �C generally ranges from 1.5 to

nstitute of Arctic Biology, 902775, United States.ker@gmail.com.

3.4 (Raich and Schlesinger, 1992), while just below 0 �C, Q10 valuescommonly range from 60 to 200 (e.g., Mikan et al., 2002), and maybe high as 6.65$105 (Monson et al., 2006). Snow-covered soils incold regions may remain just above freezing for a significant frac-tion of the winter (Brooks et al., 2005; Buckeridge and Grogan,2008; Groffman et al., 2001; Miller et al., 2007), depending onthe timing and depth of the snow cover. Thus, changes in snowcover such as those occurring over the last half century in theIntermountain West (Groisman et al., 2004; Pierce et al., 2008) andprojected to continue over the coming century (Weare and Blossier,2012), may result in significant shifts in annual C loss from soils.

In this paper, it is posited that the high sensitivity of soilrespiration to small changes in temperature just below 0 �C isprimarily a response to the transition of water between the liquidand solid phases (e.g., Tilston et al. 2010). The dynamics of theliquideice transition of water in soils may be complex and relatedto the soil texture, organic matter content and concentration ofdissolved solutes (Farouki, 1981; Lovell, 1957; Romanovsky andOsterkamp, 2000; Tilston et al., 2010) (Fig. 1). In general, soils

C. Tucker / Soil Biology & Biochemistry 78 (2014) 90e96 91

with a higher clay fraction, or higher soil organic matter, tend tofreeze more slowly, while sandy soils tend to freeze abruptly justbelow the 0 �C boundary. As soils freeze, a number of limitationsmay be imposed on soil microbes. Liquid water availability isreduced to micro-films, thus inhibiting the diffusion of substratesto microbial cells, and leading to a higher reliance on recycling ofmicrobial biomass and metabolic by-products as the primarysource of respired C (Ostroumov and Siegert, 1996; Rivkina et al.,2000; Schimel and Mikan, 2005). Additionally, the reduction inliquid water is accompanied by a decrease in the amount of air-filled pore space, because as water freezes, it expands. This reduc-tion of air-filled pore space, which may be accompanied by theformation of ice lenses (Talamucci, 2003), will result in reduceddiffusion of oxygen through the soil and increased limitation ofaerobic respiration by anoxic conditions.

In this study, I developed a modification of the recent DualArrhenius MichaeliseMenten (DAMM) model (Davidson et al.,2012) of soil respiration to account for changes in soil structureand function driven by freezing water. The DAMMmodel is a semi-mechanistic model where soil respiration responds to temperature,substrate availability and the amount of liquid water and air-filledpore space in the soil volume. The mean values of soil propertiesfrom several field sites were used as the input variables to predictand analyze the temperature and moisture responses of soil het-erotrophic respiration (RH) across the 0 �C boundary. These sitesinclude mixed grass prairie, sagebrush steppe, and high altitudeconiferous forest and meadow systems, spanning the range ofecological variability in southeast Wyoming. I hypothesized thatthe RH would be more sensitive to changing temperature below0 �C than above 0 �C, and this response would be different amongsites depending on soil density, texture and C content.

This study is largely a thought-experiment, wherein field dataon soil physical properties and organic C content are used, alongwith a modification of the DAMM model, to simulate the effect ofchanging liquid to frozen water ratios across the phase changetemperature range on the apparent temperature sensitivity of

Fig. 1. The fitted proportion unfrozen water (as unfrozen water/total water) plottedagainst soil temperature from 3 studies: Tilston et al. (2010) (3,5), Farouki (1981)(2,4,6,8,9), Romanovsky and Osterkamp (2000) (1) across a range of soil types. Thevalues of A and B in the legend refer to the parameters of the power function %uf ¼ A$abs(T)B.

heterotrophic soil respiration. Data are incorporated from severalfield sites that illustrate some of the potential range in the physicalcontrols on sub-freezing soil respiration, especially related to soiltexture and C content. This study should provide a good basis forfuture field and labwork to explore thesemechanisms in detail, andsuggestions on appropriate experiments to do so are provided inthe discussion.

2. Methods

2.1. The original DAMM model

The Dual Arrhenius MichaeliseMenten model (Davidson et al.,2012) combines simple functions for the effects of temperature,soil moisture and substrate carbon availability on heterotrophic soilrespiration (Eqn. (1)). Unlike other, more empirical, soil respirationmodels commonly used (e.g., Q10 model, Lloyd and Taylor (1994)model), the DAMM model explicitly considers the role of sub-strate and O2 transport through the liquid and gas phases of thesoil. Because a critical feature of freezing in soils is a sharp change inthe proportion of liquid, water and air-filled pore space in soils, thestructure of the DAMM model facilitates the current analysis.

In the DAMM model heterotrophic respiration (RH) is modeledas:

RH ¼ Vmax$½Sc�

kMs þ ½Sc�$½O2�

kMO2þ ½O2�

(1)

where Vmax is the maximum reaction velocity when neither oxygenconcentration ([O2]) nor substrate C concentration ([SC]) arelimiting, and kMO2

and kMs are the respective MichaeliseMentenconstants. Vmax is modeled as a function of temperature T (�C) viaan Arrhenius exponential temperature response function:

Vmax ¼ as$exp��Eas

RT

�(2)

where as is the pre-exponential factor (set to5.38$1010 mg C cm�3 soil), Eas is the activation energy(72.26 kJ mol�1), and R is the universal gas constant(8.314$10�3 kJ K�1 mol�1). Values for as and Eas were taken directlyfrom Davidson et al. (2012) for the model analysis conducted here.Similarly, representative values for kMO2

(0.121 cm3 O2 cm�3 air)and kMs (9.95$10�7 g C cm�3 soil) were derived from that study.More accurate predictions of soil respiration will require derivingthese parameters from field data in future studies. The concentra-tion of soluble substrate C (Sc) was derived from total soil C (STc) asfollows. STc values (see Table 1) were derived from literature esti-mates as well as field measurements described in Tucker et al.(2013) and Tucker et al. (in review) and values for the ChimneyPark sites were provided by B. Borkhuu and N. Brown. The amountof soluble C in the soil ([SSc]) was calculated as [SSc]¼ p$[STc], wherep (¼ 4.14$10�4 as per Davidson et al. 2012) is a parameter describingthe proportion of the total C pool [STc] that is soluble. Then, theconcentration of C at the reaction site ([Sc]) was calculated as[Sc] ¼ [Ssc]$Dliq$q where q is the volumetric water content of thesoil, and Dliq (¼3.17) is a unitless diffusion coefficient. The O2 con-centration at the reaction site [O2] is calculated as[O2] ¼ Dgas$0.209$a4/3 where Dgas (¼1.67) is a diffusion coefficientof O2 in air, 0.209 is the fractional abundance of O2 in air, and a is thevolume of air-filled pore space in the soil.

2.2. Modifications to the DAMM model

Here, the DAMM function is modified to account for changes in q

and a as soil water converts from liquid to ice. First, an empirical

Table 1Properties of the eight study sites relevant to the modeling approach. The % C, BD, and soil texture data were measured at the field sites. A and B values (for the unfrozenwatercontent calculation) were determined by selecting the values from the site (see Fig.1) with themost similar vegetation and soil properties. Subscripts of parameter A denote thestudy (as numbered in Fig. 1) fromwhich the parameters were derived. The far right column lists either the reference for the soil properties, or the last name of the individualsresponsible for data collection.

Site Vegetation %C BD % Sand % Silt % Clay A B Ref

PHACE Mixed-grass prairie 2.5 1.2 62 23 15 0.04494 �0.123 Carrillo et al. (2011)Sagebrush: Jelm Sagebrush steppe 1.4 1.38 88 8 4 0.00419 �0.181 TuckerSagebrush: Pole Mountain Sagebrush steppe 2.3 1.24 73 21 6 0.04494 �0.123 TuckerSagebrush: Centennial Sagebrush steppe 2.9 1.3 52 37 11 0.08726 �0.14 TuckerGLEES forest Sub-alpine forest 9.7 0.65 na na na 0.1693 �0.378 TuckerGLEES meadow Sub-alpine moist meadow 14 0.52 na na na 0.2601 �0.38 TuckerLodgepole Clear cut Montane forest clear cut 10 0.98 75 23.4 1.6 0.115 �0.234 Borkhuu, BrownLodgepole Bark beetle Montane forest beetle kill 4 1.26 52.6 43.2 4.2 0.115 �0.234 Borkhuu, Brown

C. Tucker / Soil Biology & Biochemistry 78 (2014) 90e9692

power function (Lovell, 1957) was used for the conversion of liquidwater to ice:

qliq ¼�q if T � 0 �Cq$A$jTjB if T <0 �C (3)

where A and B are empirical parameters derived from a literaturereview, and vary among the different sites. Parameter values werechosen from ecosystems and soil types similar to the ecosystems inthis study (Table 1, Fig. 1), however it is possible that actual valuesof these parameters are much different in our study sitee this is animportant caveat that will require further research. In general, theliterature describes a clear trend toward more rapid freezing withlower content of clay and organic matter (Fig. 1). Air-filled porespace is then calculated as

a ¼ 1� BD=PD� qliq � ice (4)

where BD is bulk density (g cm�3) of the soil at each site (Table 1),PD is the density of the mineral component of the soil(¼2.65 g cm�3), and the ratio BD/PD represents the non-ice solidvolume of the soil. The amount of ice is calculated as:

ice ¼�q� qliq

�$ð1000=917Þ (5)

such that the volume of water that transitions from the liquid to iceexpands proportional to the ratio of the densities of liquid waterand ice. As liquid becomes ice, the volume of air-filled pore spacedecreases due to the expansion of this water. Similarly, the volumeof water-filled pore space available for substrate diffusion decreasesas ice forms.

2.3. Study sites

Eight sites were chosen located in southeast Wyoming at whichto predict RH from the model based on several criteria. First, thesites are the location of recent or ongoing experiments makingmeasurements of soil respiration during the winter. While soilrespiration data is not explicitly incorporated in this analysis, a nextstep will be to “confront the model with data” from each of thesesites to estimate site-specific parameters. The second criterion isthat these sites span a wide range of winter conditions andecological characteristics that capture much of the ecological vari-ability of the IntermountainWest. Table 1 lists the study sites, alongwith important ecological characteristics and site-specific modelparameter values. The Prairie Heating and Carbon Dioxide Enrich-ment (PHACE) experiment site is located in native prairie at~1800 m elevation. The PHACE site is the lowest elevation site andcorrespondingly has the shortest winters and warmest summerconditions. The three sagebrush (Artemisia tridentata) steppe sites

e Jelm, Pole Mountain, and Centennial e are the location ofexperimental snow manipulations, and are at 2452, 2652 and2543 m elevation, respectively. The Chimney Park site is the loca-tion of an ongoing experiment tracking the impact of pine beetleoutbreaks in a Lodgepole pine (Pinus contorta) forest. This sitecontains two sub-sitese a clear cut site, and a site that was attackedby pine beetles beginning six years ago. The highest elevation sites,with the longest winters and coolest summers are located at theGlacier Lakes Ecosystem Experiment Site (GLEES), where a numberof ecological and meteorological studies are conducted. The GLEESforest site is co-dominated by Picea engelmanii (Engelmann spruce)and Abies lasiocarpa (sub-alpine fir), with an understory of Vacci-nium scoparium (grouse whortleberry). The GLEES meadow site is amixed community of shrubs, grasses, sedges and forbs, with noobvious dominant species.

2.4. Model implementation

The modified DAMM model described in Eqns. (1)e(5) wasused to calculate predicted heterotrophic respiration (RH) rates ateach site from �5 �C to þ2 �C to capture the critical range whereliquid water transitions to ice. The full range of soil water contentpossible at each site was used to determine the effects of freezingat different water contents. Since the maximum water content ofthe soil is equal to 1 � BD/PD, this value was set as the maximumand RH was estimated over the range q ¼ (0.01, 1 � BD/PD). Whensoils freeze under saturated conditions, the total soil volume isthus greater than 1, which is ecologically and physically accurate.No effect of swelling on RH was implemented in this model,although it is likely that the resultant damage to roots and soilmicrobes, as well as soil structure, would have an importantimpact on soil processes. RH was calculated every 0.1 �C from �5to þ2 �C and in 0.01 unit steps from q ¼ 0.01 to q ¼ qmax whereqmax ¼ 1 � (BD/PD). The output of the modified DAMM model ismg CO2eC cm�3 h�1, which is converted to units of mgCO2eC m�2 h�1 by multiplying that value over the upper 10 cm ofthe soil profile. The model and all subsequent analyses were donein using ‘R’ statistical software. The ‘R’ code is provided inSupplement 5.1.

3. Results

3.1. Soil freezing in relation to soil type

The results of several studies (Farouki, 1981; Romanovsky andOsterkamp, 2000; Tilston et al., 2010) measuring the liquid watercontent of soils at sub-zero temperatures are presented in Fig. 1.Parameter values (A and B, Eqn. (3)) were selected from the studysite in Fig. 1 having soil and vegetation most similar to (i.e., similartexture and organic C content, similar dominant plant functional

C. Tucker / Soil Biology & Biochemistry 78 (2014) 90e96 93

types) the site of interest (Fig. 1, Table 1), along with the bulkdensity of the soils at that site, to predict soil liquid water content(qliq) above and below freezing. The maximum predicted qliq (basedon the assumption that the soil was saturated prior to freezing) ispresented in the apex of the upper limit of the red curve (along theq axis) Fig. 2. The high altitude meadow (with peaty soils and verylow BD) had the highest qliq at all temperatures and qliq was leastsensitive to freezing at this site. qliq at sub-zero temperatures in thehigh-altitude coniferous forest sites was higher than the shortgrassprairie and sagebrush sites, and the sagebrush site (Jelm) with verysandy (88% sand) soil experienced a 99% drop in qliq between 0 �Cand �0.1 �C.

3.2. Soil respiration in the different ecosystem types

Predicted RH above and below 0 �C was very different amongsites (Fig. 2, Table 2). Consistent with the very high C content andlow bulk density, RH at all temperatures was highest in the peatysoils of the GLEES meadow site. Conversely, consistent with the lowC content and high bulk density, RH was lowest at the Jelm siteacross all temperatures. At �1 �C, RH was effectively 0 at the Jelmsite. RH at �1 �C at the GLEES Meadow site was ~2$103 times higherthan at the PHACE site, while RH at þ1 �C it was only 2.5 timeshigher (Table 2).

Because the DAMM model considers both O2 availability in theair-filled pore space, and availability of substrate C in the soil so-lution, the model predicts optimal qliq for soil respiration at inter-mediate soil water content (Fig. 2). As soil freezes, this optimal qliqdrops substantially, because the solid content of the soil increases.In all soils, the optimal qliq is between 40 and 70% of saturated liquidwater content.

Fig. 2. Predicted heterotrophic respiration (RH, mg CO2eC m2 h�1) at the 8 different field siteeach site. The gray surface represents the response to total soil water content (q), while theRespiration is predicted from the upper 10 cm of the soil profile. (For interpretation of the rarticle.)

3.3. Apparent temperature sensitivity of soil respiration

The apparent temperature sensitivity of soil respiration wascalculated as the Q10 of respiration at optimum water content:

Q10 ¼�R2R1

�10=T2�T1(6)

Q10 was calculated below and above 0 �C (Table 2). This estimate isreferred to as the apparent Q10 because below freezing it capturesboth temperature sensitivity and response to changing soil mois-ture since they are closely coupled. The maximum Q10 occurs justbelow 0 �C, however, the calculated Q10 values within less than~0.25 �C below 0 �C may be unrealistically high, driven by the po-wer function (Eqn. (3)) for frozen water content,because lim

T/0�ðA$jT jBÞ ¼ ∞, and a small change in T in this range is

associated with huge change in qliq. While the aforementionedequation for the proportions of liquid water in frozen soils has along history (starting with Lovell (1957)), it may in fact be moreappropriate to use an exponential function where the proportionequals 1 when T ¼ 0 �C. The median Q10 below 0 �C is much higherthan above 0 �C, because this value reflects both the exponentialtemperature sensitivity of the Arrhenius function as well as thepower function relating liquid water content to temperature. TheQ10 at optimumwater content was the same for all sites above 0 �C,reflecting shared exponential temperature sensitivity parameters.

4. Discussion

A number of studies of soil respiration in winter dominatedsystems have documented very high temperature sensitivity (i.e.,

s between �3 and þ1 �C across the full range of volumetric water content (q) possible atred surface represents the response to unfrozen, liquid water content (qliq) below 0 �C.eferences to color in this figure legend, the reader is referred to the web version of this

Table 2Simulated Q10 and respiration rates above and below freezing. The presented Q10 values are themedian for soil respiration above (0e2 �C) and below (�5e0 �C) freezing, alongwith the minimum and maximum values in each temperature range. Q10 values were calculated from Eqn. (6). The values above zero are invariant across sites (so only onevalue is presented) because the temperature sensitivity is determined entirely by the proscribed Eas which is held constant across sites. Values of RH,max are the maximumheterotrophic respiration (mg C m�2 h�1) at �1 �C and þ1 �C, at each site.

Site Q10 T < 0 �C Q10 T > 0 �C RH,max

Median Minimum Maximum Median Minimum Maximum (�1 �C) (1 �C)

PHACE 14.4 7.11 3.80Eþ11 3.22 3.19 3.24 0.002 10.7Jelm 29.2 10.2 7.33Eþ16 5.00E-07 6.1Pole Mtn. 14.4 7.11 3.80Eþ11 0.003 13.3Centennial 18 7.95 1.71Eþ13 0.012 10.03GLEES Forest 194 26.8 1.64Eþ24 0.71 24GLEES Meadow 225 31.2 1.34Eþ17 3.5 26.7Lodgepole Clear cut 55.7 14 2.60Eþ22 0.16 20.5Lodgepole Bark beetle 55.7 14 2.60Eþ22 0.03 11.2

C. Tucker / Soil Biology & Biochemistry 78 (2014) 90e9694

values of Q10 ranging from 20 to 6.65$105) of soil respiration below0 �C (Mikan et al., 2002; Monson et al., 2006; Tilston et al., 2010). Inthis study, I provide a theoretical mechanism that explains this hightemperature sensitivity as a function of the conversion of liquidwater to ice. Ice formation reduces heterotrophic soil respiration(RH) via twomechanisms: 1) reduction of liquid water for substratediffusion and 2) reduction air-filled pore space for O2 diffusion tosoil microbes. These mechanisms may attain different levels ofimportance depending on the initial saturation of the soil e a drysoil undergoing freezing may experience a complete cessation ofsubstrate diffusion, while a wet soil undergoing freezing mayexclude all remaining air filled pore space. Because the decline inliquid water content with decreasing temperatures is a function ofsoil physical properties, respiration rates below 0 �C are stronglydetermined by those same properties. RH effectively disappears insandy soils with low C content below 0 �C, while in peaty soils, thedecline is much less drastic. Nonetheless, predicted RH in the peaty,GLEES meadow soil at optimal water content was 7 times higher at1 �C than at �1 �C (Table 1).

During the winter, RH is highly sensitive to changes in temper-ature near and below 0 �C. Calculated Q10 temperature sensitivitiesat the different sites above 0 �C were ~3.2, meaning that for a 10 �Cincrease in temperature, RH would approximately triple. Below0 �C, themedian temperature sensitivity from this study range from15 to 255, suggesting ten- to hundred-fold increases in RH inresponse to a 10 �C increase in temperature, while the maximumvalues range up to 1.6$1024. The modeled temperature sensitivitiesfrom this study below 0 �C capture the range of this parameterdemonstrated in studies in the field (e.g., 104e6.65$105, Monsonet al. (2006)) and lab (e.g., 1e23.4, Clein and Schimel (1995),63e237, Mikan et al. (2002), 8.6e950, Tilston et al. (2010),

Fig. 3. Seven-day moving average of soil temperatures, measured at 5 cm depth at the sitesthis data was available: the PHACE data are from 2008 to 2009, the sagebrush site data are frfrom 2009 to 2010. There are long strings of missing data from Chimney Park from mid-N

22.5e5.7$107, Schmidt et al. (2009)), and this study provides a solidtheoretical basis for those previous results. Biotic factors, such asshifting microbial community composition (Schmidt et al., 2009)are not precluded as factors affecting temperature sensitivity nearthe freezing point of water. However, this study demonstrates thata semi-mechanistic model based on the diffusion of oxygen in air-filled pore space and soluble substrates in unfrozen water cansimulate very high Q10 values commonly reported. This resultsuggests the physical effects of freezing water explain much of theobserved temperature response.

Temperature sensitivity in the DAMM model is primarily afunction of Vmax (Eqn. (2)) e where an activation energy of72.26 kJ mol�1 corresponds to a Q10 of ~3.2 between �5 and þ5 �C.Below freezing, an additional source of temperature sensitivity isadded as the liquidesolid phase change modifies the availability ofsubstrate C and O2 according to MichaeliseMenten kinetics (Eqn.(1)). As soils thaw from �1 to 0 �C, substrate diffusion increases asice becomes liquid, and soil air-filled pore space increases becausethe volume of liquid water is smaller than that of ice, resulting in alarge increase in the reaction rate. Thus, while holding activationenergy constant, the apparent temperature sensitivity is very highin this temperature range compared to just above 0 �C. Because thiseffect is due to the kinetics of water freezing in soil, the shape of thefreezing curve (which is highly variable among soil types) de-termines the rate and temperature sensitivity of soil respiration justbelow freezing. Since the rate of soil water freezing below 0 �C is afunction of soil properties such as texture, bulk density and organicmatter content that vary among sites, the calculated rate and Q10 ofRH at sub-zero temperatures also varied widely among sites.

To evaluate the importance of winter respiration on theecosystem carbon balance of across these sites, it is necessary to

presented in this study. The soil temperature time series are from different years whenom 2011 to 2012, the GLEES data are from 2008 to 2009, and the Chimney Park data areovember to early January, and for all of February, which are interpolated as a flat line.

C. Tucker / Soil Biology & Biochemistry 78 (2014) 90e96 95

take into account the intensity of winter freeze and the duration ofthewinter period. Fig. 3 presents soil temperatures at 5 cm depth ateach of the study sites over the period fromNov 1st to Sept. 1st. Theperiod during which soils are below or very near (within 0.2 �C) tofreezing is much longer at the higher elevation sites (~8 months)than the lowest sites (<3 months) (Fig. 3). Conversely, the twohighest sites, GLEES forest and meadow, experience relatively sta-ble, moderate soil temperatures throughout the winter period (�1to þ0.2 �C), because deep snows at these sites provide substantialinsulation. Thus, at the sites where sub-freezing RH is predicted tobe highest, winter lasts much longer and winter soil conditions arealso less severe, and winter RH is likely a muchmore important partof the annual C flux at the highest elevation sites than at the lowersites.

The results of this study are based entirely on calculations fromthe modified DAMM model using literature derived parametervalues and mean values at each site of soil physical properties suchas texture, bulk density and C content. To demonstrate the empir-ical validity of these results, an experiment combining several setsof measurements will need to be conducted. First, the unfrozenwater content of frozen soils was determined by selecting param-eter values from sites with similar ecology or soil structure, yet it ispossible that the actual liquideice transition in soils from each siteis significantly different than the values presented here. Thus, soilliquid water contents need to be measured for the range of sitesevaluated in this study across the critical range from �5 to 0 �C.Unfrozen water content of frozen soils can be measured via either2H nuclear magnetic resonance spectroscopy (Tilston et al., 2010) ora modified TDR-probe method (Romanovsky and Osterkamp,2000). Second, the actual soil respiration rates across this samerange of temperatures will need to be carefully measured e ideally,these measurements would be conducted in the lab on soil in-cubations (on either intact cores or sieved and compressed soilsamples) in an anti-freeze bath, where temperature can becontrolled with a good deal of precision. Finally, several parametersfor the DAMM model and freezing power function should bedetermined by fitting the model to the measurements justdescribed.

5. Consequences for carbon-climate feedback

Declining snowpack across the western United States has beenattributed to ongoing climate change (Pierce et al., 2008), and thisdecline is projected to continue over the coming century (Weareand Blossier, 2012). As snowpack declines, the soil temperature ismore directly coupled to the air temperature. In areas where thesnowpack is already ephemeral and shallow (such as the mixedgrass prairie), this decline in snowpack is unlikely to result in asubstantial reduction in winter RH. Indeed, the warmer air tem-peratures are likely to stimulate RH, particularly during the shoul-der seasons where soil freezing is delayed (fall) or thawing isaccelerated (spring), as has been suggested by the carbon cycle-climate positive feedback hypothesis (Cox et al., 2000). In areaswhere the snowpack forms early, accumulates deeply, and thawsvery late in the year, a small shift in the timing of snowpack for-mation may have substantial consequences for the ecosystem Cbalance. The persistent occurrence of sub-zero air temperaturesbefore the snowpack accumulates deeply may result in the soilfreezing deeply before being buried. The insulated soil would thenwarm only very slowly or not at all throughout the winter. Giventhe dramatic sensitivity of RH to the formation of ice, winter soilrespiration could be significantly reduced under these conditions.Given that in high elevation systems winter soil respiration rep-resents ~25% of the annual flux, this reduction could be a sub-stantial fraction of the annual soil C efflux. Thus, the balance of

changes in fall air temperatures and changes in the timing andextent of early winter snow accumulation may have importantconsequences for the C balance of these systems.

Acknowledgments

I would like to thank Kiona Ogle and Elise Pendall for valuableinsight into the analysis presented here, as well as help withmanuscript revision. I would like to thank Thijs Kelleners forcomments on an early version of the manuscript. Bujidmaa Bor-khuu and Nick Brown supplied data for the Chimney Park field site.Further thanks to two anonymous reviewers for their very valuablecomments on this manuscript. Funding was provided by NASAEarth and Space Science Fellowship (NNX10AP26H).

Appendix A. Supplementary data

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.soilbio.2014.06.018.

References

Brooks, P.D., McKnight, D., Elder, K., 2005. Carbon limitation of soil respirationunder winter snowpacks: potential feedbacks between growing season andwinter carbon fluxes. Glob. Change Biol. 11, 231e238.

Buckeridge, K.M., Grogan, P., 2008. Deepened snow alters soil microbial nutrientlimitations in arctic birch hummock tundra. Appl. Soil Ecol. 39, 210e222.

Carrillo, Y., Pendall, E., Dijkstra, F.A., Morgan, J.A., Newcomb, J.M., 2011. Response ofsoil organic matter pools to elevated CO2 and warming in a semi-arid grassland.Plant Soil 347, 339e350.

Clein, J.S., Schimel, J.P., 1995. Microbial activity of tundra and taiga soils at sub-zerotemperatures. Soil Biol. Biochem. 9, 1231e1234.

Cox, P.M., Betts, R.A., Jones, C.D., Spall, S.A., Totterdell, I.J., 2000. Acceleration ofglobal warming due to carbon-cycle feedbacks in a coupled climate model.Nature 408, 184e187.

Davidson, E.A., Samanta, S., Caramori, S.S., Savage, K., 2012. The dual Arrhenius andMichaeliseMenten kinetics model for decomposition of soil organic matter athourly to seasonal time scales. Glob. Change Biol. 18, 371e384.

Farouki, O.T., 1981. Thermal Properties of Soils. DTIC Document.Groffman, P., Driscoll, C., Fahey, T., Hardy, J., Fitzhugh, R., Tierney, G., 2001. Colder

soils in a warmer world: a snow manipulation study in a northern hardwoodforest ecosystem. Biogeochemistry 56, 135e150.

Grogan, P., Chapin III, F.S., 1999. Arctic soil respiration: effects of climate andvegetation depend on season. Ecosystems 2, 451e459.

Groisman, P.Y., Knight, R.W., Karl, T.R., Easterling, D.R., Sun, B.M., Lawrimore, J.H.,2004. Contemporary changes of the hydrological cycle over the contiguousUnited States: trends derived from in situ observations. J. Hydrometeorol. 5,64e85.

Lloyd, J., Taylor, J.A., 1994. On the temperature dependence of soil respiration. Funct.Ecol. 8, 315e323.

Lovell, C., 1957. Temperature effects on phase composition and strength of partially-frozen soil. Highw. Res. Board Bull. 168, 74e95.

Mikan, C.J., Schimel, J.P., Doyle, A.P., 2002. Temperature controls of microbialrespiration in arctic tundra soils above and below freezing. Soil Biol. Biochem.34, 1785e1795.

Miller, A.E., Schimel, J.P., Sickman, J.O., Meixner, T., Doyle, A.P., Melack, J.M., 2007.Mineralization responses at near-zero temperatures in three alpine soils.Biogeochemistry 84, 233e245.

Monson, R.K., Lipson, D.L., Burns, S.P., Turnipseed, A.A., Delany, A.C., Williams, M.W.,Schmidt, S.K., 2006. Winter forest soil respiration controlled by climate andmicrobial community composition. Nature 439, 711e714.

Ostroumov, V.E., Siegert, C., 1996. Exobiological aspects of mass transfer in micro-zones of permafrost deposits. In: Brack, A., Horneck, G., Friedmann, E.I.,Meyer, M.A., Reitz, G., Banin, A. (Eds.), Life Sciences: Space and Mars RecentResults. Pergamon Press Ltd, Oxford, pp. 79e86.

Pierce, D.W., Barnett, T.P., Hidalgo, H.G., Das, T., Bonfils, C.L., Santer, B.D., Bala, G.,Dettinger, M.D., Cayan, D.R., Mirin, A., Wood, A.W., Nozawa, T., 2008. Attributionof declining western U.S. snowpack to human effects. J. Clim. 21, 6425e6444.

Raich, J.W., Schlesinger, W.H., 1992. The global carbon dioxide flux in soil respirationand its relationship to vegetation and climate. Tellus Ser. B Chem. Phys. Mete-orol. 44, 81e99.

Rivkina, E.M., Friedmann, E.I., McKay, C.P., Gilichinsky, D.A., 2000. Metabolic activityof permafrost bacteria below the freezing point. Appl. Environ. Microbiol. 66,3230e3233.

Romanovsky, V.E., Osterkamp, T.E., 2000. Effects of unfrozen water on heat andmass transport processes in the active layer and permafrost. Permafr. Periglac.Process. 11, 219e239.

C. Tucker / Soil Biology & Biochemistry 78 (2014) 90e9696

Schimel, J.P., Mikan, C., 2005. Changing microbial substrate use in Arctic tundra soilsthrough a freezeethaw cycle. Soil Biol. Biochem. 37, 1411e1418.

Schmidt, S., Wilson, K., Monson, R., Lipson, D., 2009. Exponential growth of “snowmolds” at sub-zero temperatures: an explanation for high beneath-snowrespiration rates and Q10 values. Biogeochemistry 95, 13e21.

Sommerfeld, R.A., Mosier, A.R., Musselman, R.C., 1993. CO2, CH4 and N2O fluxthrough a Wyoming snowpack and implications for global budgets. Nature 361,140e142.

Talamucci, F., 2003. Freezing processes in porous media: formation of ice lenses,swelling of the soil. Math. Comput. Model. 37, 595e602.

Tilston, E.L., Sparrman, T., Oquist, M.G., 2010. Unfrozen water content moderatestemperature dependence of sub-zero microbial respiration. Soil Biol. Biochem.42, 1396e1407.

Tucker, C.L., Bell, J., Pendall, E., Ogle, K., 2013. Does declining carbon-use efficiencyexplain thermal acclimation of soil respiration with warming? Glob. ChangeBiol. 19, 252e263.

Weare, B.C., Blossier, B., 2012. Snow water changes with elevation over the WesternUnited States in IPCC AR4 models. Clim. Change 112, 1059e1069.