Forest ecosystem respiration estimated from eddy

Forest ecosystem respiration estimated from eddy covariance and chamber measurements under high turbulence and substantial tree mortality from bark beetles

H E A T H E R N.S P E C K M A N 1,2,J O H N M.F R A N K 3,2,J O H N B.B R A D F O R D 4,B R I A N N A L.M I L E S 5,W I L L I A M J.M A S S M A N 3,W I L L I A M J.P A R T O N 1and MIC HAEL G.RYAN 1,31

Natural Resource Ecology Laboratory and Graduate Degree Program in Ecology,Colorado State University,Fort Collins,CO 80523,USA,2Department of Botany and Program in Ecology,University of Wyoming,Laramie,WY 82071,USA,3Rocky Mountain Research Station,U.S.Forest Service,Fort Collins,CO 80526,USA,4U.S.Geological Survey,Southwest Biological Science Center,Flagstaff,AZ 86001,USA,5Department of Horticulture and Landscape Architecture,Colorado State University,Fort Collins,CO 80523,USA

Abstract

Eddy covariance nighttime ?uxes are uncertain due to potential measurement biases.Many studies report eddy covariance nighttime ?ux lower than ?ux from extrapolated chamber measurements,despite corrections for low tur-bulence.We compared eddy covariance and chamber estimates of ecosystem respiration at the GLEES Ameri?ux site over seven growing seasons under high turbulence [summer night mean friction velocity (u *)=0.7m s à1],during which bark beetles killed or infested 85%of the aboveground respiring biomass.Chamber-based estimates of ecosys-tem respiration during the growth season,developed from foliage,wood,and soil CO 2ef?ux measurements,declined 35%after 85%of the forest basal area had been killed or impaired by bark beetles (from 7.1?0.22l mol m à2s à1in 2005to 4.6?0.16l mol m à2s à1in 2011).Soil ef?ux remained at ~3.3l mol m à2s à1throughout the mortality,while the loss of live wood and foliage and their respiration drove the decline of the cham-ber estimate.Eddy covariance estimates of ?uxes at night remained constant over the same period,~3.0l mol m à2s à1for both 2005(intact forest)and 2011(85%basal area killed or impaired).Eddy covariance ?uxes were lower than chamber estimates of ecosystem respiration (60%lower in 2005,and 32%in 2011),but the mean night estimates from the two techniques were correlated within a year (r 2from 0.18to 0.60).The difference between the two techniques was not the result of inadequate turbulence,because the results were robust to a u *?lter of >0.7m s à1.The decline in the average seasonal difference between the two techniques was strongly correlated with overstory leaf area (r 2=0.92).The discrepancy between methods of respiration estimation should be resolved to have con?dence in ecosystem carbon ?ux estimates.

Keywords:bark beetles,chambers,disturbance,EC,ecosystem respiration,respiration modeling,soil ef?ux,turbulence,u*?ltering

Received 28May 2014and accepted 11July 2014

Introduction

Each year,terrestrial ecosystems sequester 2.3GT of carbon,roughly 26%of annual anthropogenic global carbon emissions (Le Quere et al.,2009).The balance between photosynthesis and respiration determines carbon storage,but respiration appears to vary with the environment more than photosynthesis,and to largely control ecosystem carbon loss or gain (Valentini et al.,2000).Despite its importance,respiration is less studied than photosynthesis and there are numerous uncertain-ties in its measurements (Valentini et al.,2000).

Most ecosystem respiration (R e )measurements are derived from eddy covariance (EC),and currently there are >500EC towers established across the globe,pro-viding nearly continuous ?uxes from a wide variety of ecosystems (Baldocchi et al.,2001,www.?https://www.wendangku.net/doc/8b9890754.html,).EC measurements have greatly improved our understanding of the response of ecosystem carbon and water ?uxes to the environment and to disturbance (Wofsy et al.,1993;Goulden et al.,1998;Valentini et al.,2000;Law et al.,2002;Baldocchi,2003;Amiro et al.,2010).However,uncertainties in EC ?uxes measured at night (Goulden et al.,1996;Lavigne et al.,1997;Baldoc-chi,2003;Thomas et al.,2013)make inferences dif?cult for R e ,photosynthesis or gross primary productivity,

Correspondence:Heather N.Speckman,tel.+13039462281,fax +13077662851,e-mail:hspeckma@https://www.wendangku.net/doc/8b9890754.html,

708

?2014John Wiley &Sons Ltd

Global Change Biology (2015)21,708–721,doi:

10.1111/gcb.12731

and daily,seasonal,and annual sums of net ecosystem exchange.

The largest uncertainty in EC measurements of R e is that the atmospheric mixing required for the technique may be lower,absent,or different at night(Goulden et al.,1996;Lavigne et al.,1997;Baldocchi,2003;Tho-mas et al.,2013).During the daytime,convective heat-ing mixes the atmosphere and?ux can be recorded by the tower’s EC instrumentation(Massman&Lee,2002). At night,there is no convective heating from the sur-face and in the absence of mechanical mixing,which may not penetrate the canopy from above(Aubinet, 2008;Wharton et al.,2009;van Gorsel et al.,2011;Tho-mas et al.,2013),CO2produced near the ground may be advected downhill.Advective?uxes are not recorded by the EC tower,causing a systematic under-estimation of the true nighttime ecosystem carbon?ux. If unaddressed this‘night problem’makes ecosystems appear to be unrealistically large sinks of carbon,when they could be a carbon source(Goulden et al.,1996; Aubinet,2008).

The traditional method for dealing with a lack of tur-bulence in EC is a procedure known as u*?ltering,in which all measurements below a certain friction veloc-ity(u*)threshold are removed and then replaced via gap?lling(Goulden et al.,1996).There are many limita-tions of u*?ltering(Ruppert et al.,2006;Van Gorsel et al.,2007;Aubinet,2008),such as the selection of the u*threshold is subjective(Gu et al.,2005),and a small difference in the u*threshold can change the?ux from a carbon sink to a carbon source(Miller et al.,2004; Ruppert et al.,2006).Through u*?ltering,many sites lose~50%of their nighttime EC values,causing further uncertainty of the true nighttime?ux(Feigenwinter et al.,2004;Misson et al.,2007).

Another large and perhaps related uncertainty for EC estimates of R e(R EC)is the nearly universal and sys-tematic bias between R EC and R e estimated using cham-ber measurements of components and extrapolation models(R T;see Table1).R EC and R T should generate similar numbers,but studies in a variety of ecosystems reported that u*?ltered R EC that were signi?cantly lower than R T.For example,EC estimates of respiration were27%lower than chamber measurements and poorly correlated(r2=0.06–0.27)in Canadian boreal forest(Lavigne et al.,1997).In a deciduous forest in northern USA,EC respiration estimates were50% lower than chamber estimates,despite a good correla-tion between them(r2=0.62,Bolstad et al.,2004).Simi-lar results were described in Chinese temperate forests (Wang et al.,2010),a eucalyptus forest in the Australian highlands(Van Gorsel et al.,2007),managed meadows in the European Alps(Wohlfahrt et al.,2005),North American semiarid grasslands(Myklebust et al.,2008), and Brazil’s Amazon rainforest(Chambers et al.,2004). Many other studies have shown EC estimates of R e to be lower than chamber estimates(Table2).

We compared EC and chamber estimates of R e at the windiest EC site in North America,where bark beetle mortality also killed most of the aboveground biomass during the study(Fig.1).If a lack of turbulence and advection cause the discrepancy between R EC and R T, then the two measurement types should be roughly equal in this highly turbulent environment.We set out to determine:(i)if the EC and chamber methods for esti-mating nightly mean R e differed;(ii)if any difference between the methods decreased as turbulence increased;(iii)if the methods differed in estimating the impact of85%tree mortality on R e;and(iv)if any differ-ence between the methods decreased as the tree canopy died and changed the coupling between the subcanopy (including the forest?oor)and the atmosphere. Materials and methods

Study area

Glacier Lake Ecosystem Experimental Site(GLEES)is a subal-pine forest located in Wyoming’s Snowy Range,approxi-mately55km west of Laramie(41°21.9920N,106°14.3970W). This high elevation site(3190m),maintained by the US Forest Service Rocky Mountain Research Station(Musselman et al., 1994),has a mean annual temperature ofà2°C and a mean annual precipitation of1200mm,mostly as snow.The forest is dominated by old growth Engelmann spruce(Picea engel-mannii Parry ex Englem)and subalpine?r[Abies lasiocarpa (Hook.)Nutt]with an average canopy height of18m.The age distribution of the forest at GLEES suggests either a stand-replacing disturbance>400years ago with a very slow recov-ery,or a series of smaller disturbances over the last400years (Bradford et al.,2008).Mean annual u*is0.94m sà1,higher than any other tower(www.?https://www.wendangku.net/doc/8b9890754.html,).

EC data collection and processing

The GLEES Ameri?ux EC tower was established in its current location in2004;the tower is23m tall,with the above-canopy

Table1Terms used for the paper

R e The true nighttime ecosystem respiration.

R EC Eddy covariance estimates of R e using u*?ltered

nighttime data.

R T Chamber estimate of R e.Calculated from Eqn(4).

R W Respiration from woody tissues as estimated by

chambers.

R F Respiration from foliage as estimated by chambers.

R S Soil respiration as estimated by chambers.

R light Ecosystem respiration estimated from daytime EC

light response curves.

?2014John Wiley&Sons Ltd,Global Change Biology,21,708–721

RESPIRATION FROM EDDY COVARIANCE AND CHAMBERS709

sensors installed between22.6and25.8m in height.Air tem-perature(T a)was measured by a RTD-810resistance ther-mometer with an OM5-1P4-N100-C signal-conditioning module(Omega Engineering,Inc.,Stamford,CT,USA)and a Met–One radiation shield(076B-4radiation shield,Met One Instruments,Inc.,Grants Pass,OR,USA).Soil temperature was measured at0.05,0.10,0.20,0.50,and1.02m depths using a Hydra probe(Vitel,Inc.,Chantilly,VA,USA)(Frank et al., 2014).

At GLEES,net ecosystem exchange of carbon(NEE),water, and energy are calculated from the sum of vertical?ux(eddy covariance)and changes in carbon canopy storage(Lee et al., 2004).CO2concentration for the EC was measured using a LI-Cor7500(Li-Cor Biosciences,Lincoln,NE,USA),collected at a frequency of20Hz and compiled into30min statistics.Can-opy storage of CO2was estimated from a vertical pro?le of CO2concentration,measured once a minute at eight different heights(LI-Cor6262,until August2008,then a LI-Cor7000). Wind speed and direction were measured using a sonic ane-mometer(model SATI/3Vx,Applied Technologies,Inc.,Long-mont,CO,USA).

Our comparison of R EC and R T estimates of R e used nightly (PAR<2l mol mà2sà1)averages of NEE collected with EC during the snow-free summer nights(July1st–October1st)from2004to2011when mean u*was>0.2m sà1(Goulden et al.,1996;Gu et al.,2005)for every half hour of the night. The nights used for the respiration comparison had a mean u* of0.74m sà1,lower than the annual mean of0.94m sà1.

EC footprint and forest mortality

Fluxes observed by the EC tower originate from a‘footprint’upwind of the tower(Massman&Lee,2002).We used wind direction to determine the GLEES EC tower footprint,and used forest survey plots within the footprint for measure-ments of leaf area and sapwood volume needed to extrapolate foliar respiration and wood CO2ef?ux,and annual changes with tree mortality(Bradford et al.,2008;Frank et al.,2014).In 2004,36circular survey plots(each201m2),arranged into nine clusters,were established to estimate carbon pools in live vegetation,dead wood,and soil to a depth of30cm,and the ?uxes of annual litter fall and wood net primary production (Bradford et al.,2008).Twenty four of the36plots were within the footprint,and we used measurements of tree species, diameter,and height from these plots to compute tree leaf area,live and tree biomass,standing dead tree biomass, sapwood volume,and growth increment using allometric

Table2List of studies documenting EC estimates of R e being lower than chamber estimates

Reference Ecosystem type Site location Comparison

Barr et al.,2002Boreal Forest,aspen Saskatchewan,Canada EC

Bolstad et al.,2004Deciduous hardwoods Wisconsin,USA EC50%

Cook et al.,2008Deciduous hardwoods Wisconsin,USA EC

Dore et al.,2003Scrub-oak peatland Florida,USA EC

Flanagan&Johnson,2005Mixed grassland Alberta,Canada EC

Grunwald&

Bernhofer,2007

Subalpine spruce forest Tharandt,Germany EC

Hermle et al.,2010Boreal Forest,black spruce Quebec,Canada EC

Kutsch et al.,2008Deciduous hardwoods Thuringia,Germany EC

Lavigne et al.,1997Boreal Forest,black spruce Quebec,Canada EC27%

Nagy et al.,2011Sandy grassland Bugac,Hungary EC

Ohkubo et al.,2007Cypress evergreen forest Shiga Prefecture,Japan EC

Reth et al.,2005Meadow and brown?eld Lindenberg,Germany EC

Riveros-Iregui&

McGlynn,2009

Mountain pine forest Montana,USA EC

Schrier-Uijl et al.,2010Peatland dairy farm Oukoop,Netherlands EC16%

Tang et al.,2008Deciduous hardwoods Michigan,USA EC

Thomas et al.,2013Douglas-?r forest Oregon,USA EC

Van Gorsel et al.,2007Highland eucalyptus forest New South Wales,Australia EC

to u*?lter

Wang et al.,2010Mixed temperate forest Changbai Mountain,China EC

Wohlfahrt et al.,2005Mountain meadow Neustift,Austria EC26%

*Study only measured soil respiration,which roughly equaled eddy covariance data.It is assumed that aboveground?uxes are>0, resulting in total chamber?ux being greater than eddy covariance numbers.

?2014John Wiley&Sons Ltd,Global Change Biology,21,708–721 710H.N.SP ECK MAN et al.

equations (Kaufmann &Troendle,1981;Kaufmann et al.,1982;Ryan,1989).

Bark beetles are endemic to the study site,but starting in 2007–2008,their population rose and the rate of tree mortality dramatically increased (Figs 1and 2).Mortality from the bark beetle epidemic was assessed by annual surveys of all plots from 2009to 2011.Trees with a DBH >10cm were classi?ed as ‘infested’if they displayed any evidence of bark beetles such as pitch tubes,beetle entrance holes,or boring dust.Trees were classi?ed as ‘dead’once they lacked any green needles.In 2011,it was estimated that ~85%of the forest basal area was infested or killed by bark beetles.Because the forest survey was not conducted in 2006–2008,forest mortality for these years were modeled using a logistic regression,a shape sug-gested by dendrochronology data and MODIS estimates of leaf area (Frank et al.,2014).

Overview of chamber measurements

All chamber measurements were taken using a closed-system approach (Field et al.,1991),except where noted.For each measurement,a chamber was attached to a biological sub-strate (such as a leaf,wood,or soil);tubing connected the chamber to a portable infrared gas analyzer (IRGA)that mea-sured the increase in CO 2over time.For woody and foliage measurements,air?ow into and from the chamber was mea-sured to check for leaks.For soil respiration measurements,the collar was inserted into the mineral soil.Air inside the chamber was then ?ushed with outside air to lower the CO 2concentration to ambient prior to starting the measurements,and during ?ushing and measurements a fan within the cham-ber ensured that the air was mixed for the foliage and wood samples (the soil chamber used tubing with many small holes to mix the air within the chamber).Fluxes were calculated from the linear or exponential change in CO 2concentration over time (~60s)in a known volume of air (Field et al.,1991)and ?ts with r 2<0.98were excluded.The equipment used for each measurement type and period is listed in Table 3.

Chamber measurements of wood CO 2ef?ux

CO 2ef?ux from wood was measured on 14Engelmann spruce and 11subalpine ?r live stems (across the range of age,diame-ter,and canopy position),and on four recently dead trees (two ?r and two spruce).All trees were located within 100m of the EC tower.Measurements were made three times during the summer of 2010and ?ve times in the summer of 2011.Mea-surements of CO 2ef?ux from wood were made using a 250ml clear polycarbonate chamber,temporarily strapped to the neoprene gasket on a 7910cm aluminum plate attached with putty to the smoothed outer bark at ~1.3m height.These plates remained attached to the tree from summer

2010

Fig.1Repeat photography from the GLEES EC tower contrast-ing the forest in 2003with 65m 2ha à1healthy tree basal area,and the aftermath of massive bark beetle tree mortality in 2012with <10m 2ha à1healthy tree basal area.

?2014John Wiley &Sons Ltd,Global Change Biology ,21,708–721

RESPIRATION FROM EDDY COVARIANCE AND CHAMBERS 711

through autumn2011.System volume was calculated as the sum of volume of the chamber,tubing,IRGA,and gasket-to-tree bark space(measured for each tree).Wood CO2ef?ux measurements were expressed per unit sapwood volume (l mol mà3of sapwood sà1),and the sapwood volume under-neath each gasket was calculated using an allometric equation and geometric formulas for a cylinder and wedge(Ryan, 1990).

Chamber measurements of foliar respiration

Foliar respiration was measured in situ on?ve spruce and?ve ?r branches at night once during the summer of2010(from the EC tower and from the ground),and on eight spruce and eight?r branches three times during the summer of2011(at a height of2–3m on trees within100m of the EC tower).Sam-pled branches represented the range of tree sizes and light positions(even for trees sampled near the ground),included a range of foliage ages including new growth,were~30cm long,~1cm diameter at the proximal end,and had about ~350cm2of projected leaf area.Foliage chambers were clear polycarbonate and split length-wise,with a neoprene gasket sealing the chamber around the branch.In2010,5l chambers were used(each half3091597.5cm),and in2011these chambers were replaced by smaller3l chambers(each half 3091592cm).Leaf temperatures were measured with an infrared thermometer.Fluxes were scaled by the effective pro-jected of leaf area(l mol mà2of effective projected LAI sà1) (Kaufmann&Troendle,1981;Scurlock et al.,2001).

At the end of each summer,branches were harvested and leaf area measured using a volume displacement method (Chen et al.,1997).After the?nal2011measurements, branches were harvested,immediately recut underwater,and their stems were kept submerged during transport to the lab. Within18h of cutting,foliar respiration was again measured in the laboratory at22°C.Temperature response curves were tested on a subset of?ve branches(three spruce and two?r), measuring respiration at5,10,15,and20°C using a tempera-ture controlled cuvette(Hubbard et al.,1995).Foliage was allowed to acclimate to the new temperature for10min prior to each measurement.

Chamber measurements of soil respiration

Soil respiration was measured with survey chambers through-out the EC footprint,from2004to2011,before and during the extensive tree mortality.For survey measurements,108collars were permanently installed in the36plots located in a km2 around the EC tower(described above),and84of these col-lars were within the probable EC footprint(Frank et al.,2014) and used for constructing the model for soil respiration.Col-lars were circular(731cm2area),made of PVC pipe,and installed~5cm depth in the mineral soil,leaving~5cm of collar above the soil.A6l PVC chamber was placed on the collar for measurement.Soil respiration at each collar was measured~3times per summer in2004–2006and2009–2011, but not measured from collars containing standing water(a few collars in the?rst measurement after snow melt).Soil temperature was measured at10cm depth using a Penetra-tion Probe(Omega Engineering,Stamford,CT,USA)and soil moisture was measured at10cm depth in three different spots near the collar(HydroSense,Campbell Scienti?c,Logan, UT,USA).Fluxes were expressed as l mol mà2of ground area sà1.

Modeling observed chamber?uxes

To compare with nightly EC means,we developed models for continuous estimates of respiration?uxes for each ecosystem component(woody tissues,foliage,and soils).Models used substrate temperature,moisture,phenology,and tree species; model quality was evaluated using AIC and r2.

CO2ef?ux from woody tissues displayed strong seasonal variability,and was modeled with the log-linear model:

R w?S v expew0tw1Dtw2D2tw3STe1Twhere R W is observed woody respiration rates(l mol mà3of sapwood volume sà1),D is day of year,S is a species identi-?er,and w0–w3are model coef?cients.To convert to units of ?ux per ground area(R W,l mol mà2sà1),respiration is multi-plied by the average sapwood volume per ground area per species(S v,cm3mà2)from the plot sampling(Ryan,1990; Sprugel,1990;Lavigne et al.,1997),adjusted for mortality each year.Bark beetles infect sapwood with blue-stain fungus,and we assumed that wood CO2ef?ux from beetle-infested trees was50%of uninfected trees(likely an overestimate;sensitivity analysis in discussion).

Foliar respiration was modeled with the log-linear equation:

R F?LAI expef0tf1T ATe2Twhere R F is foliar respiration per unit ground area (l mol mà2sà1),T A is air temperature(°C)as observed by

Table3Summary chamber measurements

Type Measurements taken Time frame IRGA used Manufacturer

Woody Tissues146on25live boles2010LCA-4(open path)ADC,Hoddeston,England 39on4dead boles2010LI-820(closed path)Li-Cor Biosciences,Lincoln,NE,USA

2011Ciras-2(closed path)PP Systems,Amesbury,MA,USA Foliage85on26branches2010LI-820(closed path)Li-Cor Biosciences,Lincoln,NE,USA

2011Ciras-2(closed path)PP Systems,Amesbury,MA,USA Soil1282on84collars2004–2006,

2009–2011

LI-820(closed path)Li-Cor Biosciences,Lincoln,NE,USA

?2014John Wiley&Sons Ltd,Global Change Biology,21,708–721 712H.N.SP ECK MAN et al.

the EC tower,and f0–f1are model coef?cients.Equation(2) is a mathematically equivalent to the commonly used Q10 equation(Lloyd&Taylor,1994).To model foliar respira-tion to units of?ux per ground area,we multiplied exp (f A+f B T A)by the average effective projected leaf area (LAI,m2mà2)of the EC footprint,estimated from the annual forest inventory survey and adjusted for mortality each year.Continuous measurements of T A were provided by the EC tower.Bark beetle-infested trees retain needles for~2years after infection,but with greatly impaired phys-iology(Frank et al.,2014).We assumed that foliar respira-tion for beetle-infested trees had50%of the rate of uninfested trees.

Soil respiration was modeled using the linear model:

R s?s0ts1T sts2he3Twhere R S is soil respiration per ground area(l mol mà2sà1), T S is soil temperature at10cm(°C)for collars,h is percent vol-umetric water content for collars,and s0–s2are model coef?-cients.Equation(3)was?t using mean values of R S,T S,and h observed during each?eld session from the84soil collars in the EC footprint,and used with continuous measurements of T S and h from probes buried at10cm depth near the EC tower to generate continuous estimates.

Continuous estimates of?uxes from woody tissues,foliage, and soils from the models in Eqns(1–3)were averaged for the same time period as used for the night EC measurements to estimate total ecosystem respiration from chambers[R T, Eqn(4)]:

R T?R WtR FtR Se4TValues of R T were compared to u*-?ltered EC(R EC),using lin-ear regression and paired t-tests.

Results

Tree mortality from bark beetles

Healthy tree basal area declined from65m2haà1in 2005to10m2haà1in2011(85%decrease,Figs1and 2).In2011,15m2haà1of basal area still retained nee-dles,but was infested by bark beetles.Trees attacked by bark beetles have severely impaired physiology, likely respire little,and will die completely in1–2years physiology(Frank et al.,2014).Trees which survived the bark beetle epidemic are smaller than their prede-cessors(mean stand live DBH in2005was24.5cm,vs.

18.2cm in2011).Healthy sapwood volume decreased from330m3haà2in2005to21m3haà2in2011(6%of the original).Healthy effective projected leaf area simi-larly decreased from6.1m2mà2(?1.7,95%con?dence interval)in2005to0.9?0.3m2mà2in2011(7%of the 2005values).These allometeric estimates of LAI are slightly higher than MODIS estimates for the same site (Frank et al.,2014),a trend commonly observed in coni-fer forests(Wang et al.,2004).Chamber respiration measurements

CO2ef?ux from woody tissues(R W)increased until the end of July and decreased afterward(Fig.3a),a pattern attributed to seasonal changes in wood growth and photosynthetic activity(Ryan,1990).This trend was similar in2010and2011,and modeled using Eqn(1) (R W=exp(à9.57+1.31Dà0.00032D2+1.00S,

r2=0.67,n=146,Fig.3a).Firs respired more per unit

sapwood volume than Engelmann spruce,but had less sapwood volume per unit of tree basal area.We did not measure diurnal variation in CO2?ux with sapwood temperatures(Ryan et al.,1995),however,seasonal var-iation in sapwood temperature was not a signi?cant predictor of R W after accounting for seasonal trends. Respiration from dead tree boles was zero(39measure-ments on4trees).

Foliar respiration(R F)varied with temperature,but temperature corrected foliage respiration did not vary across season[R F=exp(à1.96+0.10T A),r2=0.63,

n=85,Fig.3b].For every10°C increase in air temper-ature(Q10),R F increased by a factor of2.7?0.2.Foliar respiration per leaf area did not differ between?rs and spruces.

Soil respiration(R S)was in?uenced both by tempera-ture and soil moisture(soil temperature was the domi-nant in?uence,R S=à1.98+0.60T S+0.044h, r2=0.83,n=1282,Fig.3c).Model?t was substantially

better using a linear rather than an exponential temper-ature response.R S did not decline after the bark beetle epidemic,nor was there any signi?cant relationship between observed soil respiration rates and distance to live or dead trees(t>0.1both when comparing collars near vs.far from trees,and collars before vs.after the death of nearby).Mean R S was estimated to be 3.3?0.09l mol mà2sà1in2005,and 3.8?0.13l mol mà2sà1in2011(2011values slightly higher because of the heavy precipitation that year).

The total ecosystem mean summer nightly respira-tion estimated from chambers(R T)was estimated to have declined35%after85%of the tree basal area was killed or infested with bark beetles,from 7.1?0.22l mol mà2sà1in2005to 4.6?0.16l mol mà2sà1in2011(Fig.4a).This decrease was entirely from the loss of aboveground biomass:R W declined82%after the epidemic(1.4?0.06 l mol mà2sà1in2005to0.26?0.02l mol mà2sà1in 2011,Fig.4b),and R F declined75%,from 2.4?0.10l mol mà2sà1in to2005to0.60?0.03l mol mà2sà1in2011,Fig.4c).R S did not decrease in response to bark beetle tree mortality;however,R S in 2011was slightly higher than other years due to high soil moisture that year(3.3?0.09l mol mà2sà1in 2005to3.8?0.1l mol mà2sà1in2011,Fig.4d).

?2014John Wiley&Sons Ltd,Global Change Biology,21,708–721

RESPIRATION FROM EDDY COVARIANCE AND CHAMBERS713

In2005,19%(?0.5%standard error)of R T was esti-mated to originate from woody tissue ef?ux,34% (?0.6%)from foliage respiration,and47%(?0.6%)from soils.These proportions are similar to those found in other studies(Lavigne et al.,1997).In2011,after85%of the tree basal area had been killed of beetle infested, only5%(?0.2%)of R T came from woody tissues,13% (?0.4)from foliage,and the remaining82%(?0.5%) from soils.EC and comparison to chambers

Unlike chamber estimates,nighttime EC measurements of respiration(R EC)did not decline after85%of the for-est basal area had been infested or killed by bark bee-tles(F-test,P>0.1),with a mean nighttime NEE of 2.9l mol mà2sà1in2005(?0.2)and3.1in2011(?0.1; Fig.6a).

We also used daytime EC data to estimate R e from the intercept of light response curves(R light,Hutyra et al.,2008).These intercepts remained constant throughout the epidemic(R light of 2.4?0.3 l mol mà2sà1in2005and2011;Fig.6a).However, maximum CO2assimilation rates(A max)and quantum yield of photosynthesis(Φ)both declined50%due to the epidemic(Frank et al.,2014).Using Bayesian analy-sis,daytime EC data were able to correctly approximate the degree of bark beetle mortality independent of actual forest inventories(Frank et al.,2014).Other EC sites have shown an increase in estimates of R e follow-ing an insect epidemic(Clark et al.,2010;Mathys et al., 2013).

Daytime and night EC values consistently estimated R e values much lower than those estimated by cham-bers,but the difference between R EC and R T declined as tree mortality increased(Figs4a,5,and6a;accessed via paired t-test,t>0.1).In2005,before the bark beetle tree mortality,EC estimated R e to be on average60%lower than chamber estimates(?0.14%).In2011,EC estimates of R e were only32%lower than chambers(?0.02%). Despite the large difference in absolute values,the two estimates of R e were correlated(yearly r2ranging from 0.17to0.60)with the slopes for each year~0.9,which implies a~constant slope and an intercept that decreased with tree mortality.After85%of the above-ground biomass was killed or infected by bark beetles,

Fig.3The variability in chamber measurements of wood,foli-

age,and soil CO2ef?ux were?t to models driven by environ-

mental variability or phenology.(a)Observed and modeled

CO2ef?ux from woody tissues measured on live boles in2010

and2011was highly seasonal,varied with sapwood volume,

and was greater for?r.Data points are mean ef?ux and stan-

dard error for each species during each measurement session(n

~10).(b)Observed and modeled foliar respiration varied with

temperature.Data points represent foliage respiration rates

observed in the?eld(circles)and lab(triangles).(c)Observed

and modeled soil ef?ux rates varied with soil temperature and

moisture and did not differ between the mostly intact forest

measurements(2004–2006)and high tree mortality years(2009–

2011).Data points are mean and standard error soil CO2ef?ux

for each measurement period(n~84measurements,taken over

2–3days),standardized to20%volumetric soil moisture.

?2014John Wiley&Sons Ltd,Global Change Biology,21,708–721 714H.N.SP ECK MAN et al.

R T estimated from chamber measurements decreased by 35%,with no change in EC estimates of R e (Fig.4a and 5).The difference between R T and R EC was strongly related to live leaf area,which declined with tree mortality (Fig.6b,r 2=0.92).

Discussion

Ecological implications of chamber measurements

Chamber measurements enable discerning how indi-vidual ecosystem components react to environmental factors and disturbances.Over the bark beetle mortality period assessed,modeled mean CO 2ef?ux from woody tissues declined 72%and foliage respiration declined 74%from the loss of sapwood and live foliage.Stand-ing dead tree boles had no measurable CO 2ef?ux,and decomposition of standing aboveground dead wood will likely remain negligible until the trees fall (Harmon et al.,2011).

Unlike the wood and foliage,soil respiration remained constant during the large tree mortality from the bark beetles (F -test,P >0.1).This result was consistent with another study of soil respiration after

?2014John Wiley &Sons Ltd,Global Change Biology ,21,708–721

RESPIRATION FROM EDDY COVARIANCE AND CHAMBERS 715

bark beetle tree mortality(Morehouse et al.,2008),with tree girdling in a pi~n on-juniper woodland(Berryman et al.,2013)and in a eucalyptus plantation(Binkley et al.,2006),but it differed from girdling studies in which soil respiration rapidly declined after girdling (50%decline,e.g.H€o gberg et al.,2001).A lack of soil respiration response to bark beetle attack in this and other studies could be caused by(i)roots respiring stored carbohydrates for several years after beetle attack or(ii)the decline of autotrophic respiration is off-set by an increase in heterotrophic respiration from decomposition of newly fallen foliage and dead roots (Morehouse et al.,2008;Berryman et al.,2013).Resam-pling forest?oor in2011showed that the litter fall from dead trees increased forest?oor mass by40% between2005and2011and increased litter quality as the C:N ratio dropped from71to50(H.N.Speckman, M.G.Ryan,unpublished data).In a nearby lodgepole pine ecosystem with similar tree mortality from mountain pine beetle,nitrogen in the increased litter from dead trees did not appear in streams(Rhoades et al.,2013).In our study,at least some of the N from the dead foliage and needles remained within the for-est ecosystem,as N was68%more abundant in the forest?oor after the epidemic(H.N.Speckman,M.G. Ryan,unpublished data).In the lodgepole pine study, the increased nitrogen increased decomposition (Rhoades et al.,2013).

Exploration of uncertainty in R EC(u*?ltering,an alternative technique for estimating R e and energy balance)does not explain the discrepancy between R EC and R T

Eddy covariance requires turbulence to be above a cer-tain threshold to properly function(Goulden et al., 1996;Baldocchi,2003).At least a portion of the discrep-ancy between EC and chamber estimates of R e may result from insuf?cient turbulence(Van Gorsel et al., 2007;Aubinet et al.,2010),even though most of the studies in Table2?ltered R EC values to exclude those with low u*.We investigated the possibility that the threshold empirically calculated for this study (0.2m sà1)was insuf?cient for estimating R e by com-paring R EC under different u*?lters,and R EC vs.R T with R EC selected under u*?lters as high as0.7m sà1 with an additional requirement that storage‘?ux’be <0.4l mol mà2sà1.R EC proved robust to these?lters, maintaining roughly the same absolute difference and correlation between the two datasets(see Table4). While increased turbulence bought R EC and R T slightly closer,the two datasets still do not converge,providing strong evidence that insuf?cient turbulence above the canopy and selection of a u*?lter were not responsible for the discrepancy R EC and R T.

We also explored the use of two alternatives to the u*?ltering technique for estimating R e,one that worked

?2014John Wiley&Sons Ltd,Global Change Biology,21,708–721 716H.N.SP ECK MAN et al.

well in a Eucalypt forest and elsewhere(Van Gorsel et al.,2007;van Gorsel et al.,2008,2009),and a light-curve method that matched R T in a wet tropical forest (Hutyra et al.,2008).The Van Gorsel et al.(2007)tech-nique assumes that immediately after sunset the atmo-sphere is stable and advection is small compared to storage and vertical turbulent?uxes,and develops a relationship between the maximum respiration mea-sured after sunset(R max)and soil temperature for a monthly window.This relationship is used with mea-sured soil temperature to derive a continuous estimate of R EC for the ecosystem.R max estimates of respiration were generally higher than u*-?ltered R EC values and much closer to chamber estimates of R e(van Gorsel et al.,2009).The R max technique failed at this study site because(i)variability in NEE at night frequently obscured selection of a R max(perhaps because the site was so turbulent at night);and(ii)when R max values could be estimated,R max had no relationship with soil temperature.The intercept of a light response curve (R Light)to estimate R e with EC?ux in the day was simi-lar to R EC,did not decline with tree mortality,and did not match R T in this study(P>0.1for all tests;Fig6a) (Frank et al.,2014).

Energy balance closure(comparison of measured net radiation with the sum of sensible+latent heat?ux plus heat storage change)is frequently used as an indicator of EC data quality(Foken,2008).Energy bal-ance closure at this study site averaged82%from2005 to2011(Frank et al.,2014),did not vary signi?cantly from year to year or with bark beetle mortality (t>0.1),and is similar to that reported in other stud-ies(Aubinet et al.,2000;Wilson et al.,2002).The com-parison during nighttime30minute periods was similar to periods in the day,but noisier(r2=0.38 night vs.0.72day).Exploration of uncertainty in R T(chamber placement, sampling bias,model bias,and literature values for component?uxes)does not explain the discrepancy between R EC and R T

The?rst potential error for chamber measurements is a bias caused by physical placement of the chamber, which can alter temperature,air pressure,and diffusion gradients(Baldocchi,2003),particularly for soil respira-tion(Bain et al.,2005).Because the foliage measure-ments were taken at night,the wood measurements were on the large thermal mass of the stem,and the chambers were scrubbed with air at ambient CO2con-centration prior to measurement,chamber bias is unli-kely for these.The soil pore space can hold about a day’s?ux in easily disrupted storage(Ryan&Law, 2005),and CO2can be pulled from storage by wind and measured by EC(Bowling&Massman,2011),and chambers might alter this wind-driven?ux(Bain et al., 2005).We have not directly tested this potential bias, but the chamber top included tubing to equilibrate pressure between the inside and outside of the cham-ber,making this bias for soil respiration less likely.

A second dif?culty with chamber measurements is the dif?culty in obtaining an unbiased sample for foliar respiration and wood CO2ef?ux for extrapolation on large trees.Foliar respiration varies with foliar age and throughout the canopy(Ryan et al.,1996;Cavaleri et al., 2008)and likely varied after the onset of mortality.We considered foliage age and light environment in our samples by measuring?ux for a large sample(the distal ~15cm of a branch)that usually included the full com-plement of foliage ages(older foliage receives less light as foliage develops distal to it,Schoettle,1990).Our samples included foliage from a few beetle-infested trees(possibly attacked between the2nd and3rd foli-age sample),but we did not analyze that factor sepa-rately.Wood CO2ef?ux can be greater at higher locations in the canopy and for smaller branches and trees(Ryan et al.,1996;Cavaleri et al.,2006).If so,our estimates of wood CO2ef?ux may be underestimates, particularly because we did not estimate ef?ux for trees <10cm DBH.Our84soil respiration collars were estab-lished with a systematic sampling design and their locations should be unbiased.

CO2ef?ux rates at the tissue level for foliage and wood were comparable with other studies in boreal and subalpine forests.For example,wood CO2ef?ux per unit bark surface area was0.4–4.2l mol mà2sà1 for spruce and0.5–2.1l mol mà2sà1for?r in this study,compared with0.4–1.4l mol mà2sà1for Engel-mann spruce about250km south of these measure-ments(Ryan,1990),0.4–1.0l mol mà2sà1for black spruce in boreal forests(Lavigne&Ryan,1997;Ryan

Table4Comparison between EC and chambers insensitive

to changes in u*?lter

2005

Filter u*>0.2m sà1

R T=0.96R EC+4.3

Filter u*>0.7m sà1

R T=0.97R EC+3.6

2011

Filter u*>0.2m sà1

R T=0.79R EC+2.2

Filter u*>0.7m sà1

R T=1.02R EC+0.9

Where:R T=Chamber estimate of R e.Calculated from

Eqn(4).

R EC=Eddy covariance estimates of R e using u*?ltered night-

time data.

?2014John Wiley&Sons Ltd,Global Change Biology,21,708–721

RESPIRATION FROM EDDY COVARIANCE AND CHAMBERS717

et al.,1997),and0.5l mol mà2sà1;for Scots pine in Finland(Zha et al.,2007).Foliar respiration per leaf area was0.3–0.5l mol mà2sà1for spruce and?r in this study,compared with0.3–0.7l mol mà2sà1for boreal black spruce(Ryan et al.,1997)and1.3l mol mà2sà1 for boreal Scots pine(Zha et al.,2007).Soil respiration was higher for this study(summer average of 3–3.8l mol mà2sà1)than in boreal spruce (~2.5l mol mà2sà1,Lavigne et al.,1997;Wang et al., 2003),but comparable for spruce-?r in Newfoundland, Canada(4l mol mà2sà1,Moroni et al.,2009).R EC aver-aged3.0l mol mà2sà1)for this study,higher than the 2.5l mol mà2sà1)found in boreal spruce(Lavigne et al.,1997),but was comparable to that for Scots pine in Finland(Zha et al.,2007).The spruce-?r forest at GLEES had three times more tree biomass (150Mg haà1)than the black spruce forests in the Lav-igne et al.(1997)study(Gower et al.,1997).The greater biomass would yield larger wood,foliage,and root res-piration at the GLEES site,perhaps explaining the higher R T in this study compared to the boreal sites.

A third source of error is that the models extrapolated between measurements that were infrequent(~monthly) relative to the continuous R EC record.To investigate if the difference between R T and R EC resulted from model-ing error,we compared CO2ef?ux from wood,foliage, and soil taken over a4day period(August7th–10th, 2011)to both R EC and the chamber models[Eqns(1–3)]. The mean u*value for the three studied nights(August 7th,9–10th)was0.61m sà1(August8had u* <0.2m sà1).R EC values during this time were40%lower than observed R T and33%lower than modeled R T(Fig-ure S1),suggesting that at least for this point in time, modeling error did not cause the difference R EC and R T. To estimate R T for foliage and wood for trees attacked by bark beetles but not yet killed,we assumed that foli-age and wood respiration rate was half that of healthy trees.We tested the impact of this assumption by com-paring the R T calculated with the50%rate assumption with R T estimated assuming infested trees have either zero foliage respiration or wood CO2ef?ux,or the same rate as healthy trees.R T varied<1.0l mol mà2sà1 between the0%and100%rates for a given year,and the regression coef?cients between R T and R EC values remained within one standard error of the original.We also note that soil respiration,the only chamber?ux measured throughout the entire study,and with a large sample size and unbiased sample design,was larger or equal to R EC throughout the study(Fig.6a).

Potential explanation for EC-chamber discrepancy

The difference between R EC and R T estimates of R e declined with the progression of tree mortality for bark beetles and R ECàR T was strongly correlated with tree leaf area(Fig.6b),suggesting that the origin of this dis-crepancy is linked to the amount of canopy.Thick for-est canopy can impede the mechanical mixing of air above the canopy and air within and below the canopy at night.If this occurs,?ux measurements on top of the tower become decoupled from ground and canopy sources of CO2,even with high u*values being recorded above the canopy(Amiro,1990;Loescher et al.,2003;Kutsch et al.,2008;van Gorsel et al.,2011; Sera?movich et al.,2011;Thomas et al.,2013).This phe-nomena may be prevalent in tall forest canopies,with high leaf area,where a recent study in a tall Douglas-?r forest showed88%of night measurements were decou-pled from the tower?ux measurement(Thomas et al., 2013).

Before the tree mortality from bark beetles,the site’s thick canopy(effective projected LAI6.1m2mà2,?1.7 95%con?dence interval)might have inhibited the mix-ing of air above and below the canopy,despite the site’s high winds(Fig.7).If low or intermittent mixing occurred,CO2from soil respiration(the largest source) and foliage and wood ef?ux would have moved off site without being observed by the tower-mounted EC sys-tem and pro?le.As the bark beetle mortality pro-gressed and the canopy thinned(to0.9?0.3m2mà2 in2011),more turbulence could have penetrated through the canopy,allowing the EC system to observe proportionally more of the CO2sources.The lack of change in R EC as tree mortality increased might be explained by the offsetting effects of a reduction in eco-system respiration(from tree mortality)and the increase in the EC tower’s ability to observe the true ecosystem respiration(because of better mixing in the thinning of canopy).Better mixing would also result in the observed convergence of chamber and EC estimates with a thinner canopy.

Turbulence during the day is also generated from convection,mostly derived from heating of the forest canopy,in addition to wind from above the canopy; with convection,air can be mixed even with a thick canopy(Fig.7).This could explain why daytime EC estimates of A max and quantum yield of photosynthesis (Φ)changed during the epidemic(Frank et al.,2014), but not nighttime R EC measurements.It is uncertain why the intercept of EC light response curves(R light) would not be affected by a change in forest LAI.These ideas could be explored through the installation of sec-ond EC system located below the canopy to identify coupling between subcanopy and above-canopy?ux. At another site,such a system greatly improved the quality of EC measurements,and generated R EC and R T estimates of R e within3%of each other(Thomas et al., 2013).

?2014John Wiley&Sons Ltd,Global Change Biology,21,708–721

718H.N.SP ECK MAN et al.

A bias in chamber measurements could explain a portion of the relationship in Fig.6b.For example,if the bias yielded an overestimate of foliar and wood res-piration,then R T>R EC and this inequality would widen with increasing forest LAI.

Need to quantify measurement uncertainty

Both chambers and eddy covariance are subject to many sources of error,ranging from instrument calibra-tion to uncertainty in calculation coef?cients.Error sources for chambers include IRGA calibration,cham-ber volume,estimation leaf and sapwood volume,and allometerics used for upscaling(Lavigne et al.,1997; Davidson et al.,2002;Loescher et al.,2006).In addition to advection and lack of nighttime turbulence,eddy covariance error sources include IRGA calibration, measurements of wind speed,turbulence sampling error,and footprint spatial variability(Hollinger& Richardson,2005;Oren et al.,2006;Aubinet et al.,2012). Formal analysis of all these errors is seldom performed and is dif?cult using traditional statistical techniques. New Bayesian statistical techniques have enabled the successful quanti?cation of these errors(Hollinger& Richardson,2005;Lasslop et al.,2010)and is recom-mended for EC and chamber estimates of R e. Implications

EC is a powerful and widely adopted technique for measuring ecosystem?uxes,and has generated a sub-stantial improvement in understanding of the response of carbon and waster?uxes to the environment and to disturbance.However,many?ndings have relied on the u*?ltering and gap-?lling techniques for estimates of R e used to estimate seasonal or annual sums of car-bon?ux,to make inferences about the environmental and vegetation controls over R e,and to derive estimates of ecosystem photosynthesis.The discrepancy between chamber and EC estimates of R e should be resolved before con?dence can be attained in the true measure-ment of ecosystem carbon?ux and its components. Knowledge of the true ecosystem?uxes will greatly advance scienti?c understanding of local carbon cycling,allow for more accurate carbon budgets,and improve the development of global ecological models. Acknowledgements

We thank Raechel Owens,Erik Skeie,Chase Jones,and Lance Asherin for data collection and Ben Bird,Kristen Scott,and Kurt Speckman for assistance with data analysis and prose.This research was funded by US Forest Service Rocky Mountain Research Station through the Forest Service National Climate Change Program.Plot data collection,including soil respiration prior to2010was funded by NASA grants CARBON/04-0225-0191.Further thanks to A.Scott Denning and Jay Ham MG Ryan acknowledges the support of CSIRO’s McMaster Fellowship during manuscript preparation.

References

Amiro BD(1990)Drag coef?cients and turbulence spectra within3boreal forest cano-pies.Boundary-Layer Meteorology,52,227–246.

Amiro BD,Barr AG,Barr JG et al.(2010)Ecosystem carbon dioxide?uxes after distur-bance in forests of North America.Journal of Geophysical Research:Biogeosciences, 115,1–13.

Aubinet M(2008)Eddy covariance CO2?ux measurements in nocturnal conditions: an analysis of the problem.Ecological Applications,18,1368–1378.

Aubinet M,Grelle A,Ibrom A et al.(2000)Estimates of the annual net carbon and water exchange of forests:the EUROFLUX methodology.Advances in Ecological Research,30,113–175.

Aubinet M,Feigenwinter C,Heinesch B et al.(2010)Direct advection measurements do not help to solve the night-time CO2closure problem:evidence from three dif-ferent forests.Agricultural and Forest Meteorology,150,655–664.

(a)(b)(c)

Fig.7Turbulence is generated by a different mechanism during the day vs.night.Blue arrows represent relatively colder carbon-rich air,and red arrows warmer carbon-poor air.During the day,(a)convective heating connects air below and above canopy.At night,tur-bulence is generated by above-canopy wind shear and requires mechanical mixing.A thick forest canopy(b)prevents this turbulence from mixing with the air below,decoupling?ows,and much of the respiration?ux can?ow away via advection.A thinner canopy(c) allows some above-canopy turbulence to penetrate,resulting in partial coupling and allowing proportionally more respiration?ux to be measured by the eddy covariance tower(see text for references).

?2014John Wiley&Sons Ltd,Global Change Biology,21,708–721

RESPIRATION FROM EDDY COVARIANCE AND CHAMBERS719

Aubinet M,Vesala T,Papale D(2012)Eddy Covariance:A Practical Guide to Measure-ment and Data Analysis.Springer Science+Business Media B.V.,London.173–209.

Print.

Bain WG,Hutyra L,Patterson DC,Bright AV,Daube BC,Munger JW,Wofsy SC (2005)Wind-induced error in the measurement of soil respiration using closed dynamic chambers.Agricultural and Forest Meteorology,131,225–232.

Baldocchi DD(2003)Assessing the eddy covariance technique for evaluating carbon dioxide exchange rates of ecosystems:past,present and future.Global Change Biol-ogy,9,479–492.

Baldocchi D,Falge E,Gu LH et al.(2001)FLUXNET:a new tool to study the temporal and spatial variability of ecosystem-scale carbon dioxide,water vapor,and energy ?ux densities.Bulletin of the American Meteorological Society,82,2415–2434.

Barr AG,Grif?s TJ,Black TA et al.(2002)Comparing the carbon budgets of boreal and temperate deciduous forest stands.Canadian Journal of Forest Research,32,813–822. Berryman E,Marshall JD,Rahn T,Litvak M,Butnor J(2013)Decreased carbon limita-tion of litter respiration in a mortality-affected pi~n on–juniper woodland.Biogeo-sciences,10,1625–1634.

Binkley D,Stape JL,Takahashi EN,Ryan MG(2006)Tree-girdling to separate root and heterotrophic respiration in two Eucalyptus stands in Brazil.Oecologia,148, 447–454.

Bolstad PV,Davis KJ,Martin J,Cook BD,Wang W(2004)Component and whole-sys-tem respiration?uxes in northern deciduous forests.Tree Physiology,24,493–504. Bowling DR,Massman WJ(2011)Persistent wind-induced enhancement of diffusive CO2transport in a mountain forest snowpack.Journal of Geophysical Research:Bio-geosciences,116,352–370.

Bradford JB,Birdsey RA,Joyce LA,Ryan MG(2008)Tree age,disturbance history, and carbon stocks and?uxes in subalpine Rocky Mountain forests.Global Change Biology,14,2882–2897.

Cavaleri MA,Oberbauer SF,Ryan MG(2006)Wood CO2ef?ux in a primary tropical rain forest.Global Change Biology,12,2442–2458.

Cavaleri MA,Oberbauer SF,Ryan MG(2008)Foliar and ecosystem respiration in an old-growth tropical rain forest.Plant,Cell and Environment,31,473–483. Chambers JQ,Tribuzy ES,Toledo LC et al.(2004)Respiration from a tropical forest ecosystem:partitioning of sources and low carbon use ef?ciency.Ecological Appli-cations,14,S72–S88.

Chen JM,Rich PM,Gower ST,Norman JM,Plummer S(1997)Leaf area index of bor-eal forests:theory,techniques,and measurements.Journal of Geophysical Research: Atmospheres,102,29429–29443.

Clark KL,Skowronski N,Hom J(2010)Invasive insects impact forest carbon dynam-ics.Global Change Biology,16,88–101.

Cook BD,Bolstad PV,Martin JG et al.(2008)Using light-use and production ef?-ciency models to predict photosynthesis and net carbon exchange during forest canopy disturbance.Ecosystems,11,26–44.

Davidson EA,Savage K,Verchot LV,Navarro R(2002)Minimizing artifacts and biases in chamber-based measurements of soil respiration.Agricultural and Forest Meteorology,113,21–37.

Dore S,Hymus GJ,Johnson DP,Hinkle CR,Valentini R,Drake BG(2003)Cross vali-dation of open-top chamber and eddy covariance measurements of ecosystem CO2exchange in a Florida scrub-oak ecosystem.Global Change Biology,9,84–95. Feigenwinter C,Bernhofer C,Vogt R(2004)The in?uence of advection on the short term CO2-budget in and above the forest canopy.Boundry-Layer Meteorology,113, 201–224.

Field CB,Ball JT,Berry JA(1991)Photosynthesis:principles and?eld techniques.In: Plant Physiological Ecology:Field Methods and Instrumentation(eds Pearcy RW,Ehle-ringer J,Mooney HA,Rundel PW),pp.209–253.Chapman and Hall,London. Flanagan LB,Johnson BG(2005)Interacting effects of temperature,soil moisture and plant biomass production on ecosystem respiration in a northern temperate grass-land.Agricultural and Forest Meteorology,130,237–253.

Foken T(2008)The energy balance closure problem:an overview.Ecological Applica-tions,18,1351–1367.

Frank JM,Massman WJ,Ewers BE,Huckaby LS,Negron JF(2014)Ecosystem CO2/ H2O?uxes are explained by hydraulically limited gas exchange during tree mortality from spruce beetles.Journal of Geophysical Research:Biogeosciences,119, 1195–1215.

van Gorsel E,Leuning R,Cleugh HA,Keith H,Kirschbaum MUF,Suni T(2008) Application of an alternative method to derive reliable estimates of nighttime res-piration from eddy covariance measurements in moderately complex topography.

Agricultural and Forest Meteorology,148,1174–1180.

van Gorsel E,Delpierre N,Leuning R et al.(2009)Estimating nocturnal ecosystem respiration from the vertical turbulent?ux and change in storage of CO2.Agricul-tural and Forest Meteorology,149,1919–1930.van Gorsel E,Harman IN,Finnigan JJ,Leuning R(2011)Decoupling of air?ow above and in plant canopies and gravity waves affect micrometeorological estimates of net scalar exchange.Agricultural and Forest Meteorology,151,927–933.

Goulden ML,Munger JW,Fan SM,Daube BC,Wofsy SC(1996)Measurements of car-bon sequestration by long-term eddy covariance:methods and a critical evaluation of accuracy.Global Change Biology,2,169–182.

Goulden ML,Wofsy SC,Harden JW et al.(1998)Sensitivity of boreal forest carbon balance to soil thaw.Science,279,214–217.

Gower ST,Vogel JG,Norman JM,Kucharik CJ,Steele SJ,Stow TK(1997)Carbon dis-tribution and aboveground net primary production in aspen,jack pine,and black spruce stands in Saskatchewan and Manitoba,Canada.Journal of Geophysical Research:Atmospheres,102,29029–29041.

Grunwald T,Bernhofer C(2007)A decade of carbon,water and energy?ux measure-ments of an old spruce forest at the Anchor Station Tharandt.Tellus Series B-Chemi-cal and Physical Meteorology,59,387–396.

Gu LH,Falge EM,Boden T et al.(2005)Objective threshold determination for night-time eddy?ux?ltering.Agricultural and Forest Meteorology,128,179–197.

Harmon ME,Bond-Lamberty B,Tang JW,Vargas R(2011)Heterotrophic respiration in disturbed forests:a review with examples from North America.Journal of Geo-physical Research:Biogeosciences,116,498–512.

Hermle S,Lavigne MB,Bernier PY,Bergeron O,Pare D(2010)Component respira-tion,ecosystem respiration and net primary production of a mature black spruce forest in northern Quebec.Tree Physiology,30,527–540.

H€o gberg P,Nordgren A,Buchmann N et al.(2001)Large-scale forest girdling shows that current photosynthesis drives soil respiration.Nature,411,789–792. Hollinger DY,Richardson AD(2005)Uncertainity in eddy covariance measuremnts and its application to physiological models.Tree Physiology,25,873–885. Hubbard RM,Ryan MG,Lukens DL(1995)A simple,battery-operated,temperature-controlled cuvette for respiration measurements.Tree Physiology,15,175–179. Hutyra LR,Munger JW,Hammond-Pyle E et al.(2008)Resolving systematic errors in estimates of net ecosystem exchange of CO2and ecosystem respiration in a tropi-cal forest biome.Agricultural and Forest Meteorology,148,1266–1279.

Kaufmann MR,Troendle CA(1981)The relationship of leaf area and foliage biomass to sapwood conducting area in four subalpine forest tree species.Forest Science,27, 477–482.

Kaufmann MR,Edminster CB,Troendle CA(1982)Leaf Area Determinations for Subalpine Tree Species in the Central Rocky Mountains.UDSA Forest Service,Rocky Mountain Forest and Range Experiment Station,Research Paper RM-238,Fort Collins,CO. Kutsch WL,Kolle O,Rebmann C,Knohl A,Ziegler W,Schulze ED(2008)Advection and resulting CO2exchange uncertainty in a tall forest in central Germany.Ecologi-cal Applications,18,1391–1405.

Lasslop GM,Reichstein M,Detto M et al.(2010)Comment on Vickers et al.:Self-corre-lation between assimilation and respiration resulting from?ux partitioning of eddy-covariance CO2?uxes.Agricultural and Forest Meteorology,150,312–314. Lavigne MB,Ryan MG(1997)Growth and maintenance respiration rates of aspen, black spruce and jack pine stems at northern and southern BOREAS sites.Tree Physiology,17,543–551.

Lavigne MB,Ryan MG,Anderson DE et al.(1997)Comparing nocturnal eddy covari-ance measurements to estimates of ecosystem respiration made by scaling cham-ber measurements.Journal of Geophysical Research,102(D24),28977–28986.

Law BE,Falge E,Gu L et al.(2002)Environmental controls over carbon dioxide and water vapor exchange of terrestrial vegetation.Agricultural and Forest Meteorology, 113,97–120.

Le Quere C,Raupach MR,Canadell JG et al.(2009)Trends in the sources and sinks of carbon dioxide.Nature Geoscience,2,831–836.

Lee X,Massman WJ,Law BE(eds.)(2004)Handbook of Micrometeorology:A Guide for Surface Flux Measurement and Analysis.Kluwer,Dordrecht.

Lloyd J,Taylor JA(1994)On the temperature dependence of soil respiration.Func-tional Ecology,8,315–323.

Loescher HW,Law BE,Mahrt L et al.(2006)Uncertainties in,and interpretation of, carbon?ux estimates using the eddy covariance technique.Journal Geophysical Research,111,1–19.

Loescher HW,Oberbauer SF,Gholz HL,Clark DB(2003)Environmental controls on net ecosystem-level carbon exchange and productivity in a Central American trop-ical wet forest.Global Change Biology,9,396–412.

Massman WJ,Lee X(2002)Eddy covariance?ux corrections and uncertainties in long-term studies of carbon and energy exchanges.Agricultural and Forest Meteorol-ogy,113,121–144.

Mathys A,Black TA,Nesic Z et al.(2013)Carbon balance of a partially harvested mixed conifer forest following mountain pine beetle attack and its comparison to a clear-cut.Biogeoscienes,10,5451–5463.

?2014John Wiley&Sons Ltd,Global Change Biology,21,708–721

720H.N.SP ECK MAN et al.

Miller SD,Goulden ML,Menton MC,da Rocha HR,de Freitas HC,Figueira A,de Sousa CAD (2004)Biometric and micrometeorological measurements of tropical forest carbon balance.Ecological Applications ,14,S114–S126.

Misson L,Baldocchi DD,Black TA et al.(2007)Partitioning forest carbon ?uxes with overstory and understory eddy-covariance measurements:a synthesis based on FLUXNET data.Agricultural and Forest Meteorology ,114,14–31.

Morehouse K,Johns T,Kaye J,Kaye A (2008)Carbon and nitrogen cycling immedi-ately following bark beetle outbreaks in southwestern ponderosa pine forests.Forest Ecology and Management ,255,2698–2708.Moroni MT,Carter PQ,Ryan DAJ (2009)Harvesting and slash piling affect soil respi-ration,soil temperature,and soil moisture regimes in Newfoundland boreal for-ests.Canadian Journal of Soil Science ,89,343–355.

Musselman RC,Connel BH,Conrad MA (1994)The Glacier Lake Ecosystem Experi-ments https://www.wendangku.net/doc/8b9890754.html,DA Fort Service General technical Report RM ,249,1–94.

Myklebust MC,Hipps LE,Ryel RJ (2008)Comparison of eddy covariance,chamber,and gradient methods of measuring soil CO 2ef?ux in an annual semi-arid grass,Bromus tectorum .Agricultural and Forest Meteorology ,148,1894–1907.

Nagy Z,Pinter K,Pavelka M,Darenova E,Balogh J (2011)Carbon ?uxes of surfaces vs.ecosystems:advantages of measuring eddy covariance and soil respiration simultaneously in dry grassland ecosystems.Biogeosciences ,8,2523–2534.

Ohkubo S,Kosugi Y,Takanashi S,Mitani T,Tani M (2007)Comparison of the eddy covariance and automated closed chamber methods for evaluating nocturnal CO 2exchange in a Japanese cypress forest.Agricultural and Forest Meteorology ,142,50–65.

Oren R,Hsieh C,Stoy P et al.(2006)Estimating the uncertainty in annual net ecosys-tem carbon exchange:spatail variation in turbulent ?uxes and sampling error in eddy-covarience measurements.Global Change Biology ,12,883–896.

Reth S,Gockede M,Falge E (2005)CO 2ef?ux from agricultural soils in Eastern Ger-many -comparison of a closed chamber system with eddy covariance measure-ments.Theoretical and Applied Climatology ,80,105–120.

Rhoades CC,McCutchan JH,Cooper LA et al.(2013)Biogeochemistry of beetle-killed forests:explaining a weak nitrate response.Proceedings of the National Academy of Sciences ,110,1756–1760.

Riveros-Iregui DA,McGlynn BL (2009)Landscape structure control on soil CO 2ef?ux variability in complex terrain:scaling from point observations to watershed scale ?uxes.Journal of Geophysical Research:Biogeosciences ,114,1–14.

Ruppert J,Mauder M,Thomas C,Luers J (2006)Innovative gap-?lling strategy for annual SUMS of CO 2net ecosystem exchange.Agricultural and Forest Meteorology ,138,5–18.

Ryan MG (1989)Sapwood volume for three subalpine conifers:predictive equations and ecological implications.Canadian Journal of Forest Research ,19,1397–1401.Ryan MG (1990)Growth and maintenance respiration in stems of Pinus contorta and Picea engelmannii .Canadian Journal of Forest Research ,20,48–57.

Ryan MG,Law BE (2005)Interpreting,measuring,and modeling soil respiration.Bio-geochemistry ,73,3–27.

Ryan MG,Gower ST,Hubbard RM,Waring RH,Gholz HL,Cropper WP,Running SW (1995)Woody tissue maintenance respiration of four conifers in contrasting climates.Oecologia ,101,133–140.Ryan MG,Hubbard RM,Pongracic S,Raison RJ,McMurtrie RE (1996)Foliage,?ne-root,woody-tissue and stand respiration in Pinus radiata in relation to nutrient sta-tus.Tree Physiology ,16,333–343.

Ryan MG,Lavigne MB,Gower ST (1997)Annual carbon cost of autotrophic respira-tion in boreal forest ecosystems in relation to species and climate.Journal of Geo-physical Research ,102(D24),28871–28884.Schoettle AW (1990)The interaction between leaf longevity and shoot growth and foliar biomass per shoot in Pinus contorta at two elevations.Tree Physiology ,7,209–214.

Schrier-Uijl AP,Kroon PS,Hensen A,Leffelaar PA,Berendse F,Veenendaal EM (2010)Comparison of chamber and eddy covariance-based CO 2and CH 4emission estimates in a heterogeneous grass ecosystem on peat.Agricultural and Forest Mete-orology ,150,825–831.Sera?movich A,Thomas C,Foken T (2011)Vertical and horizontal transport of energy and matter by coherent motions in a tall spruce canopy.Boundary-Layer Meteorology ,140,429–451.

Sprugel DG (1990)Components of woody-tissue respiration in young Abies amabilis trees.Trees ,4,88–98.Tang JW,Bolstad PV,Desai AR,Martin JG,Cook BD,Davis KJ,Carey EV (2008)Eco-system respiration and its components in an old-growth forest in the Great Lakes region of the United States.Agricultural and Forest Meteorology ,148,171–185.

Thomas CK,Martin JG,Law BE,Davis K (2013)Toward biologically meaningful net carbon exchange estimates for tall,dense canopies:multi-level eddy covariance observations and canopy coupling regimes in a mature Douglas-?r forest in Ore-gon.Agricultural and Forest Meteorology ,173,14–27.Valentini R,Matteucci G,Dolman AJ et al.(2000)Respiration as the main determinant of carbon balance in European forests.Nature ,404,861–865.

Van Gorsel E,Leuning R,Cleugh HA,Keith H,Suni T (2007)Nocturnal carbon ef?ux:reconciliation of eddy covariance and chamber measurements using an alternative to the u *-threshold ?ltering technique.Tellus Series B-Chemical and Physical Meteo-rology ,59,397–403.Wang CK,Bond-Lamberty B,Gower ST (2003)Soil surface CO 2?ux in a boreal black spruce ?re chronosequence.Journal of Geophysical Research ,108,1–8.

Wang Y,Woodcock CE,Buermann W et al.(2004)Evaluation of the MODIS LAI algorithm at a coniferous forest site in Finland.Remote Sensing of Environment ,91,114–127.

Wang M,Guan DX,Han SJ,Wu JL (2010)Comparison of eddy covariance and cham-ber-based methods for measuring CO 2?ux in a temperate mixed forest.Tree Physi-ology ,30,149–163.

Wharton S,Schroeder M,Paw KT,Falk M,Bible K (2009)Turbulence considerations for comparing ecosystem exchange over old-growth and clear-cut stands for lim-ited fetch and complex canopy ?ow conditions.Agricultural and Forest Meteorology ,149,1477–1490.

Wilson K,Goldstein A,Falge E et al.(2002)Energy balance closure at FLUXNET sites.Agricultural and Forest Meteorology ,113,223–243.

Wofsy SC,Goulden ML,Munger JW et al.(1993)Net exchange of CO 2in a mid-lati-tude forest.Science ,260,1314–1317.

Wohlfahrt G,Bahn M,Haslwanter A,Newesely C,Cernusca A (2005)Estimation of daytime ecosystem respiration to determine gross primary production of a moun-tain meadow.Agricultural and Forest Meteorology ,130,13–25.

Zha TS,Xing ZS,Wang KY,Kellomaki S,Barr AG (2007)Total and component carbon ?uxes of a Scots pine ecosystem from chamber measurements and eddy covari-ance.Annals of Botany ,99,345–353.

?2014John Wiley &Sons Ltd,Global Change Biology ,21,708–721

RESPIRATION FROM EDDY COVARIANCE AND CHAMBERS 721