Seasonal and diurnal variation in CO fluxes from an agricultural bioenergy crop

Biogeosciences, 13, 5471–5485, 2016 www.biogeosciences.ne /13/5471/2016/ doi:10.5194/bg-13-5471-2016 © Au ho (s) 2016. CC A ibu ion 3.0 License. Seasonal and diu nal a ia ion in CO luxes om an ag icul u al bioene gy c op Ma i Pihla ie1,2, Ülla Rannik1, Sami Haapanala1, Olli Pel ola1, Na asinha Shu pali3, Pe i J. Ma ikainen3, Saa a Lind3, Niina Hy önen3, Pe u Vi kajä i4, Ma k Zahnise 5, and I an Mamma ella1 1Depa men o Physics, Uni e si y o Helsinki, P.O. Box 48, 00014 Uni e si y o Helsinki, Finland 2Depa men o Food and En i onmen al Sciences, P.O. Box 56, 00014 Uni e si y o Helsinki, Finland 3Biogeochemis y esea ch g oup, Depa men o En i onmen al and Biological Sciences, Uni e si y o Eas e n Finland, Yliopis o an a 1D-E, PO Box 1627, Kuopio campus, 70211, Finland 4Na u al Resou ces Ins i u e Finland, G een echnology, Halolan ie 31 A, 71750 Maaninka, Finland 5Ae odyne Resea ch, Inc. 45 Manning Road Bille ica, MA 01821-3976, USA Co espondence o: Ma i Pihla ie ([email p o ec ed]) Recei ed: 2 Decembe 2015 – Published in Biogeosciences Discuss.: 26 Janua y 2016 Re ised: 29 Augus 2016 – Accep ed: 31 Augus 2016 – Published: 4 Oc obe 2016 Abs ac . Ca bon monoxide (CO) is an impo an eac i e ace gas in he a mosphe e, while i s sou ces and sinks in he biosphe e a e poo ly unde s ood. Soils a e gene ally con- side ed as a sink o CO due o mic obial oxida ion p ocesses, while emissions o CO ha e been epo ed om a wide ange o soil–plan sys ems. We measu ed CO luxes using he mi- c ome eo ological eddy co a iance me hod om a bioene gy c op ( eed cana y g ass) in eas e n Finland om Ap il o No embe 2011. Con inuous lux measu emen s allowed us o assess he seasonal and diu nal a iabili y and o compa e he CO luxes o simul aneously measu ed ne ecosys em ex- change o CO2, N2O and hea luxes as well as o ele an me eo ological, soil and plan a iables in o de o in es i- ga e ac o s d i ing he CO exchange. The eed cana y g ass (RCG) c op was a ne sou ce o CO om mid-Ap il o mid-June and a ne sink h oughou he es o he measu emen pe iod om mid-June o No em- be 2011, excluding a measu emen b eak in July. CO luxes had a dis inc diu nal pa e n wi h a ne CO up ake in he nigh and a ne CO emission du ing he day ime wi h a maxi- mum emission a noon. This pa e n was mos p onounced in sp ing and ea ly summe . Du ing his pe iod he mos signi - ican ela ionships we e ound be ween CO luxes and global adia ion, ne adia ion, sensible hea lux, soil hea lux, el- a i e humidi y, N2O lux and ne ecosys em exchange. The s ong posi i e co ela ion be ween CO luxes and adia ion sugges s abio ic CO p oduc ion p ocesses, whe eas he ela- ionship be ween CO luxes and ne ecosys em exchange o CO2, and nigh - ime CO luxes and N2O emissions indica e bio ic CO o ma ion and mic obial CO up ake espec i ely. The s udy shows a clea need o de ailed p ocess s udies accompanied by con inuous lux measu emen s o CO ex- change o imp o e he unde s anding o he p ocesses asso- cia ed wi h CO exchange. 1 In oduc ion Ca bon monoxide (CO) is an impo an eac i e ace gas in he a mosphe e, whe e i pa icipa es in he chemical eac- ions wi h hyd oxyl adicals (OH), po en ially leading o he p oduc ion o he s ong g eenhouse gas ozone (O3). The e- ac ions o CO and OH dec ease he a mosphe ic capaci y o oxidize a mosphe ic me hane (CH4), hence indi ec ly a ec - ing he li e ime o his impo an g eenhouse gas. Al hough CO i sel abso bs only a li le in a ed adia ion om he Ea h, he cumula i e indi ec adia i e o cing o CO may be e en la ge han ha o he hi d powe ul g eenhouse gas, ni ous oxide (N2O; Myh e e al., 2013). An h opogenic ac i i ies ela ed o he bu ning o ossil uel and biomass (e.g. o es i es) as well as pho ochemical oxida ion o CH4 and non-me hane hyd oca bons a e he main sou ces o CO (Duncan e al., 2007), while he eac ion wi h OH is he ma- jo sink o CO in he a mosphe e (Duncan and Logan, 2008). Published by Cope nicus Publica ions on behal o he Eu opean Geosciences Union. 5472 M. Pihla ie e al.: Seasonal and diu nal a ia ion in CO luxes Soils a e globally conside ed as a sink o CO due o mi- c obial oxida ion p ocesses in he soil (Con ad and Seile , 1982; Po e e al., 1996; Whalen and Reebu gh, 2001; King and Webe , 2007). Acco ding o Con ad and Seile (1980) he soil consump ion o CO is a mic obial p ocess, i ol- lows i s -o de kine ics and can ake place in bo h ae obic and anae obic condi ions. A di e se g oup o soil mic obes a e capable o oxidizing CO. They include ca boxydo ophs, me hano ophs and ni i ie s (Fe enci e al., 1975; Jones and Mo i a, 1983; Bende and Con ad, 1994; King and Webe , 2007), hence hey po en ially link CO luxes o he exchange o CH4and N2O. In addi ion o CO consump ion, p oduc ion o CO has been ound in a wide ange o soils (Moxley and Smi h, 1998; Gödde e al., 2000; King, 2000; Va ella e al., 2004; Galbally e al., 2010; B uhn e al., 2013; an Aspe en e al., 2015), plan oo s (King and C osby, 2002; King and Hung ia, 2002), li ing and deg ading plan ma e ial (Ta e al., 1995; Schade e al., 1999; De endo p e al., 2011; Lee e al., 2012) and deg ading o ganic ma e (Wilks, 1959; Con- ad and Seile , 1985b). Al hough mic obial CO o ma ion may occu in anae obic condi ions (Funk e al., 1994; Rich and King, 1999), mos o en he CO p oduc ion has been ela ed o abio ic p ocesses such as he mal, UV- o isible ligh -induced deg ada ion o o ganic ma e o plan ma e ial (Con ad and Seile , 1985b; Ta e al., 1995; Schade e al., 1999; De endo p e al., 2011; Lee e al., 2012; an Aspe en e al., 2015; F ase e al., 2015). Pho odeg ada ion in ol es di ec and indi ec pho odeg ada ion o e.g. li e o o ganic ma e ial (King e al., 2012). In he di ec pho odeg ada ion, a molecule (e.g. lignin) has abso bed adia ion and unde - goes di ec changes such as agmen a ion, in amolecula ea angemen o elec on ans e om o o he molecule (King e al., 2012). In he indi ec pho odeg ada ion, ce ain pho osensi ize s abso b he incoming adia ion and ans e he ene gy o o he molecules such as iple oxygen, o m- ing eac i e in e media es such as single oxygen, hyd oxyl adical o hyd ogen pe oxide, which can u he change he chemis y o ano he non-ligh -abso bing molecule (e.g. cel- lulose) o pa o he same molecule whe e he pho osensi- ize esided (King e al., 2012). Indi ec pho odeg ada ion may also e e o adia ion-induced s imula ion o mic o- bial deg ada ion h ough b eaking down o ganic compounds making hem easily a ailable o mic obial deg ada ion (see King e al., 2012). The mal deg ada ion is iden i ied as he empe a u e-dependen deg ada ion o ca bon in he absence o adia ion and possibly oxygen (De endo p e al., 2011; Lee e al., 2012; an Aspe en e al., 2015). The sepa a ion be- ween CO o ma ion h ough he mal deg ada ion and pho- odeg ada ion is e y challenging because hey can bo h ake place simul aneously and he indi ec pho odeg ada ion may occu e en in he absence o sola adia ion i adequa e he - mal ene gy is p esen (Lee e al., 2012). Unde s anding o he biological p ocesses leading o CO elease and he impo ance o hese sou ces in e es ial ecosys ems a e poo ly unde s ood (Moxley and Smi h, 1998; King and C osby, 2002; V eman e al., 2011; He and He, 2014). Fo ma ion o CO om li ing g een plan s unde il- lumina ion and he p esence o oxygen was al eady ound in he la e 1950s by Wilks (1959) and Siegel e al. (1962). Mo e ecen ly, CO has been ound o be o med e.g. in plan oo s (King and C osby, 2002), in s essed plan s (He and He, 2014), du ing heme oxida ion (Engel e al., 1972; V e- man e al., 2011), in a oma ic amino acid deg ada ion p o- cesses (Hino and Tauchi, 1987) and in lipid pe oxida ion e- ac ions (Wol and Bidlack, 1976). Howe e , he impo ance o hese biological CO o ming p ocesses in he ne CO ex- change and, in gene al, o he global CO budge s ill emain la gely unknown (King and C osby, 2002). Mos o he epo ed CO lux measu emen s a e ei he sho - e m ield expe imen s (e.g. Con ad and Seile , 1985a; Funk e al., 1994; Zepp e al., 1997; Kuhlbusch e al., 1998; Moxley and Smi h, 1998; Schade e al., 1999; Va ella e al., 2004; B uhn e al., 2013; an Aspe en e al., 2015) o labo a- o y incuba ions wi h speci ic ea men s o he soil o plan ma e ial (Ta e al., 1995; King and C osby, 2002; Lee e al., 2012). Bo h CO up ake and emissions a e epo ed om soil– plan sys ems in di e en clima ic egions, and mos ly he CO luxes ange be ween −2 and 2 nmol m−2s−1(Con ad e al., 1988; Funk e al., 1994; Zepp e al., 1997; Moxley and Smi h, 1998; Schade e al., 1999; King, 2000; King and Hun- g ia, 2002; Va ella e al., 2004; Galbally e al., 2010). Based on he a ailable li e a u e, he e is a endency o sou h- o- no h g adien wi h highe CO emissions om opical and Medi e anean en i onmen s compa ed o bo eal and em- pe a e ecosys ems (e.g. Zepp e al., 1997; Kuhlbusch e al., 1998; King, 2000; Va ella e al., 2004; Galbally e al., 2010; Cons an e al., 2008; B uhn e al., 2013; an Aspe en e al., 2015). Howe e , he high a ia ion be ween CO up ake and emission a es does no ye allow us o classi y he ecosys em ypes o clima ic egions. Tall owe (And eae e al., 2015) and ai bo ne measu emen s ha e indica ed sou ce a eas o CO bo h in he Amazon basin (Ha iss e al., 1990) and in he No h Ame ican und a (Ri e e al., 1992, 1994) sugges ing a connec ion be ween high plan biomass and biological CO o ming p ocesses. To ou unde s anding his is he i s s udy o epo long- e m and con inuous ield measu emen s o CO luxes (FCO)using he mic ome eo ological eddy co a iance (EC) me hod. We measu ed FCO abo e a bo eal pe ennial g ass- land ecosys em, eed cana y g ass (RCG), o e a 7-mon h snow- ee pe iod in 2011 using wo pa allel lase abso p ion spec ome e s. We compa ed he FCO wi h simul aneously measu ed luxes o ca bon dioxide (CO2), ne ecosys em ex- change o CO2(NEE), ni ous oxide (N2O), hea and ene gy as well as wi h ele an soil, plan and me eo ological a i- ables. Based on p e ious s udies, we expec ha he diu nal and seasonal a ia ions in FCO a e s ongly dependen on a- dia ion and empe a u e. On he o he hand, we do no ex- pec s ong ela ionships be ween FCO and NEE o be ween FCO and N2O luxes due o he limi ed in o ma ion a ailable Biogeosciences, 13, 5471–5485, 2016 www.biogeosciences.ne /13/5471/2016/ M. Pihla ie e al.: Seasonal and diu nal a ia ion in CO luxes 5473 on he in ol emen o biological p ocesses in FCO and chal- lenges in sepa a ing pa allel abio ic and bio ic d i e s o FCO. We hypo hesize ha a nega i e co ela ion be ween FCO and NEE can indica e an in ol emen o a biological componen in CO p oduc ion, and ha a posi i e co ela ion be ween nigh - ime FCO and N2O lux may indica e an in ol emen o ni i ie s in CO consump ion. 2 Ma e ials and me hods 2.1 Measu emen si e The measu emen s we e conduc ed on a mine al ag i- cul u al ield loca ed in eas e n Finland (63◦9048.6900 N, 27◦1403.2900 E), cul i a ed wi h a pe ennial eed cana y g ass (Phala is a undinaceae, L. c . Pala on). The measu emen s co e ed a pe iod om snowmel o he new snow all, om Ap il o No embe 2011. Long- e m ( e e ence pe iod 1981– 2010) annual mean ai empe a u e in he egion is 3.2◦C and annual p ecipi a ion is 612 mm (Pi inen e al., 2012). The c op was cul i a ed a he beginning o June 2009. In 2011 a he beginning o he g owing season (23 May, day 143), he c op was e ilized wi h an NPKS e ilize con- aining 76 kg N ha−1(NO3-N : NH4-N =47 :53). The c op om he p e ious season was kep a he si e o e he win- e (Bu all, 1997) and was ha es ed on 28 Ap il (day 118; Lind e al., 2016). The sp ing and ea ly summe (days 118– 160) was cha ac e ized by as -g owing c op wi h he c op heigh inc easing om abou 10 cm in mid-May o 1.7 m in la e June (day 180), eaching he maximum heigh o 1.9 m in ea ly July. The ield was 6.3ha in size and om he sam- pling loca ion o he EC measu emen sys em he oo p in was homogenous in all di ec ions, ex ending 162, 137, 135 and 178 m o N, E, S and W espec i ely. The e is a sligh sou h- o-no h slope in he ield and he we es a ea lies in he no he n co ne o he oo p in , which o en has s anding wa e du ing he pe iod o snowmel (Ap il). The soil a he si e is classi ied as a Haplic Cam- bisol/Regosol (Hype eu ic, Sil ic; IUSS Wo king G oup WRB, 2007) and he ex u e o he opsoil (0–28 cm) a - ied om clay loam o loam based on he US Depa men o Ag icul u e (USDA) ex u al classi ica ion sys em. Wi hin he ploughing laye om he su ace o abou 30cm, soil pH a ies om 5.4 o 6.1, and soil o ganic ma e con en a ied be ween 3 and 11 % espec i ely. The a e age C /N a io in he ploughing laye was 14.9 ( anging om 14.1 o 15.7). We pe o med oo p in analysis in o de o iden i y he sou ce a ea o he lux measu emen s. Two limi ing cases we e analysed: i s , a low c op ep esen ing he beginning o he campaign, and second, a canopy wi h 1.9m in heigh ep esen ing he RCG canopy a e midsumme . The mea- su emen heigh s 2.2 and 2.4 m we e used in he analysis. In he i s case, we ep esen ed he low canopy as he su - ace wi h ae odynamic oughness 0.04 m (de e mined om measu emen s), in he second case, a canopy wi h lea a ea dis ibu ion cha ac e is ic o RCG c ops was ep esen ed by a be a dis ibu ion. In bo h cases he sou ces we e assumed a he soil su ace. Such an assump ion was made due o lim- i ed in o ma ion on sou ce–sink beha iou (see Sec . 3 be- low) and also in o de o ob ain mo e conse a i e oo p in es ima es. Th ee s abili y classes ep esen ing uns able ( he Obukho leng h L= −10m), nea -neu al (L= −100 m) and s able (L= +10 m) condi ions we e conside ed. The oo p in e alua ion was pe o med by using he Lag angian s ochas ic ajec o y simula ions (e.g. Rannik e al., 2003). The upwind dis ances con ibu ing 80 % o he lux we e iden i ied o low/high canopy as ollows: 53/23 m, 83/34 m and 166/60 m o uns able, nea -neu al and s able s a i ica- ions espec i ely. The conduc ed oo p in analysis e eals ha he p esence o a canopy signi ican ly educes he oo - p in ex en . No e ha he conse a i e oo p in scena io wi h no canopy is applicable only o a sho pe iod o ime due o as canopy g ow h a he beginning o he campaign (see Fig. 1d). Conside ing ha p e ailing wind di ec ion du - ing he measu emen pe iod was om SE and SSW di ec- ions, and he wind di ec ion in e al 110–315◦con ibu ed 90 % o he hal -hou pe iods used in he analysis, he oo - p in analysis con i ms ha he oo p in was su icien and he measu emen s well ep esen he RCG canopy. 2.2 CO lux measu emen s The EC measu emen s we e made as a pa o he ICOS (In- eg a ed Ca bon Obse a ion Sys em) Finland p og amme du ing Ap il o No embe 2011. He e we epo he e- sul s o FCO calcula ed om he concen a ion measu emen s using wo con inuous-wa e quan um cascade lase s: AR- CW-QCL (model CW-TILDAS-CS Ae odyne Resea ch Inc., see e.g. Zahnise e al., 2009) and LGR-CW-QCL (model N2O/CO-23d, Los Ga os Resea ch Inc., see e.g. P o encal e al., 2005). The measu emen s by AR-CW-QCL ex ended he whole measu emen pe iod om Ap il o No embe 2011 (days 110–325), whe eas o LGR-CQ-QCL da a a e a ail- able om la e summe o he end o he measu emen pe- iod (days 206–330). Fluxes om he wo analyse s a e com- pa ed; howe e , due o he longe da a co e age, he diu nal and seasonal a ia ion in FCO is assessed using da a om AR-CW-QCL only. The AR-CW-QCL and LGR-CQ-QCL we e he same as used in he s udy by Rannik e al. (2015) whe ein ou lase -based as - esponse gas analyse s used o measu e ni ous oxide (N2O) luxes we e compa ed. The measu emen heigh was 2.2m un il 30 June 2011 (day 181) when he heigh was aised o 2.4 m due o he g ow h o RCG. The gas inle s o he closed-pa h analy- se s we e loca ed 10 cm below a sonic anemome e (USA-1, Me ek Ge many GMBH) used o measu ing u bulen wind componen s. In addi ion, CO2and H2O luxes we e mea- su ed a he si e wi h an in a ed gas analyse (LI7000 – Li- Co Inc., Lincoln, NE, USA) connec ed o a sonic anemome- www.biogeosciences.ne /13/5471/2016/ Biogeosciences, 13, 5471–5485, 2016 5474 M. Pihla ie e al.: Seasonal and diu nal a ia ion in CO luxes e (R3-50, Gill Solen L d., UK). The closed-pa h gas anal- yse s we e loca ed in an ai condi ioned cabin a abou 15 m eas om he ai inle and he anemome e s. This wind di ec- ion (50–110◦sec o ) was he e o e disca ded om u he analysis due o possible dis u bances o lux measu emen s. Sample lines (PTFE) we e shielded and hea ed sligh ly abo e ambien ai empe a u e. Sample lines we e 16 m in leng h, hei inne diame e s we e 4 and 8 mm, he sample ai low a es we e 13.2 and 11.6 LPM (Rannik e al., 2015). Based on ma e ial es ing wi h LGR-CW-QCL, he PTFE ubing was ound ine wi h espec o CO in a cons an - low se -up and low a e o 2.5 LPM (unpublished da a). The EC measu e- men s we e sampled a 10 Hz equency. Fu he de ails on he EC se -up, ins umen speci ica ions and da a acquisi ion, can be ound in Rannik e al. (2015) and Lind e al. (2016). 2.3 Suppo ing measu emen s A wea he s a ion loca ed a he si e moni o ed con inuously se e al me eo ological and soil pa ame e s such as ai em- pe a u e (Tai )and ela i e humidi y (RH; model: HMP45C, Vaisala Inc.), p ecipi a ion (P ; model: 52203, R.M. Young Company), global (Rglob)and ne adia ion (Rne ; model: CNR1, Kipp & Zonen B.V.), pho osyn he ically ac i e adia- ion (PAR, model: SKP215, Skye ins umen s L d.), soil hea lux a 7.5 cm dep h (G; model: HPF01SC, Hukse lux), soil empe a u es a 2.5, 5, 10, 20 and 30 cm dep hs (Tsoil; model: 107, Campbell Scien i ic Inc.) and soil wa e con en a 2.5, 5, 10 and 30 cm dep hs (SWC) (model: CS616, Campbell Scien i ic Inc.). All me eo ological da a we e eco ded as 30 min mean alues and s o ed using a da a logge (model: CR 3000, Campbell Scien i ic Inc.). Lea a ea index (LAI) was measu ed a app oxima ely weekly in e als du ing he main c op g ow h pe iod using a plan canopy analyse (model: LAI-2000, LiCo ). G een a ea index (GAI) was es ima ed on a weekly basis om plo s adjacen o he LAI measu emen s acco ding o Wilson e al. (2007) and Lind e al. (2016). The GAI measu emen s we e conduc ed om h ee loca ions (1 ×1 m2)and wi hin each om h ee spo s (8 ×8 cm2)by coun ing a numbe o g een s ems (Sn)and g een lea es (Ln)pe uni a ea and measu ing he g een a ea o lea es (La)and s ems (Sa). The GAI was calcula ed as GAI =(SnSa)+(LnLa). 2.4 Da a p ocessing and analysis The EC da a p ocessing was pe o med wi h pos -p ocessing so wa e EddyUH (Mamma ella e al., 2016). Fil e ing o elimina e spikes (Vicke s and Mah , 1997) was pe o med acco ding o an app oach, whe e he high- equency EC da a we e despiked by compa ing wo adjacen measu emen s. I he di e ence be ween wo adjacen concen a ion measu e- men s o CO was g ea e han 20ppb, he ollowing poin was eplaced wi h he same alue as he p e ious poin . The spec oscopic co ec ion due o wa e apou impac on he abso p ion line shape was accoun ed o along wi h he dilu ion co ec ion. LGR-CW-QCL au oma ically co ec ed he wa e apou e ec using a buil -in module in he LGR da a acquisi ion so wa e. The same spec oscopic co ec- ion was applied o AR-CW-QCL a e a so wa e upda e in July 2011. P io o his so wa e upda e, he espec i e di- lu ion and spec oscopic co ec ions o AR-CW-QCL high- equency CO mole ac ion da a we e pe o med du ing he pos -p ocessing phase acco ding o Rannik e al. (2015) wi h he ins umen -speci ic CO spec oscopic coe icien (b=0.28) de e mined in he ield. P io o calcula ing he u bulen luxes, a 2-D o a ion (mean la e al and e ical wind equal o ze o) o sonic anemome e wind componen s was calcula ed acco ding o Kaimal and Finnigan (1994) and all a iables we e linea ly de ended. The EC luxes we e calcula ed as 30min co a i- ances be ween he scala s and e ical wind eloci y ol- lowing commonly accep ed p ocedu es (e.g. Aubine e al., 2000). Time lag be ween he concen a ion and e ical wind speed measu emen s induced by he sampling lines was de- e mined by maximizing he co a iance. Due o he la ge in- ne diame e (8 mm) o he sampling line in LGR-CW-QCL, he esul ing lag ime was 4.2 s compa ed o ha o 0.91 s o AR-CW-QCL wi h he sampling line inne diame e o 4 mm. The inal p ocessing was, howe e , done by ixing he ime lag o a oid unphysical a ia ion o lag occu ing due o andom lux e o s. Spec al co ec ions we e applied o accoun o he low and high- equency a enua ion o he co a iance. The i s -o de esponse imes o he EC sys- ems we e de e mined o be 0.07 and 0.26 s o he AR-CW- QCL and LGR-CW-QCL sys ems espec i ely, ollowing he me hod by Mamma ella e al. (2009). This esul ed in di e - en lux co ec ion ac o s mainly due o ube damping: o AR-CW-QCL he 5 and 95 pe cen ile alues o lux unde - es ima ion we e 2.1 and 12.2 % and o LGR-CW-QCL hey we e 5.7 and 21.4 %. Da a quali y sc eening was pe o med acco ding o Vicke s and Mah (1997) o ensu e exclusion o he sys em mal unc ioning as well as unphysical and/o unusual occasions in measu emen s. We chose o pe o m es s on single ime se ies o ensu e quali y o measu emen s used in he analysis and did no use he lux s a iona i y es (Foken and Wichu a, 1996) because he CO luxes a e e- quen ly small and ha e la ge ela i e andom e o s. In such cases he es s based on ela i e e o s a e no expec ed o pe o m well (e.g. Rannik e al., 2003). A e quali y sc een- ing, 66.0 % o he FCO da a (AR-CW-QCL) we e a ailable, wi h da a co e age o 59.2 % du ing he day ime and 75.9 % du ing he nigh - ime. Fo de ails o he da a p ocessing and quali y sc eening, see Rannik e al. (2015). To e alua e in de ail he seasonal changes in FCO and ac- o s a ec ing he luxes, he da a we e di ided in o six pe i- ods (days 110–145 (20 Ap il–25 May) a e sp ing (S), days 146–160 (25 May–9 June) a e ea ly summe (ES), days 161– 181 (10–30 June) a e midsumme (MS), days 205–240 (24 Biogeosciences, 13, 5471–5485, 2016 www.biogeosciences.ne /13/5471/2016/ M. Pihla ie e al.: Seasonal and diu nal a ia ion in CO luxes 5475 Table 1. Mean, median and 25–75 h pe cen iles o he CO luxes (FCO, nmol m−2s−1)measu ed in a ead cana y g ass (RCG) c op a Maaninka. The luxes a e sepa a ely calcula ed o day ime (FCO_day, sun ele a ion, hsun > 0) and nigh - ime (FCO_nigh ,hsun < 0) and as a ne lux o e all FCO da a (ne FCO) o he six measu emen pe iods (S is sp ing, ES is ea ly summe , MS is midsumme , LS is la e summe , A is au umn, LA is la e au umn) and o e he ull measu emen pe iod (All) om Ap il o No embe 2011. FCO_day FCO_nigh ne FCO Pe iod, days mean median 25–75 h mean median 25–75 h mean median 25–75 h pe cen ile pe cen ile pe cen ile S, 110–145 0.97 0.68 −0.15 2.00 −0.64 −0.56 −0.97 −0.20 0.41 0.09 −0.57 1.28 ES, 146–160 0.24 0.08 −0.29 0.57 −0.67 −0.49 −0.72 −0.33 0.03 −0.10 −0.45 0.43 MS, 161–181 −0.07 −0.08 −0.40 0.24 −0.67 −0.52 −0.86 −0.22 −0.22 −0.18 −0.55 0.16 LS, 205–240 0.36 0.30 −0.07 0.87 −0.76 −0.49 −0.96 −0.19 −0.09 −0.04 −0.53 0.49 A, 241–295 −0.12 −0.18 −0.48 0.13 −0.66 −0.61 −0.90 −0.32 −0.44 −0.44 −0.77 −0.10 LA, 296–325 −0.62 −0.59 −0.94 −0.26 −1.05 −1.01 −1.37 −0.65 −0.92 −0.89 −1.25 −0.49 All, 110–325 0.21 0.01 −0.41 0.55 −0.77 −0.66 −1.06 −0.33 −0.25 −0.34 −0.79 0.17 July–28 Augus ) a e la e summe (LS), days 241–295 (29 Augus –23 Oc obe ) a e au umn (A) and days 296–325 (24 Oc obe –21 No embe ) a e la e au umn (LA)). The di ision in o hese pe iods was based on seasonal changes in c op g ow h and de elopmen o changes in FCO and empe a u e, while he leng hs o he pe iods we e kep as simila in leng h as possible. Also, FCO we e no measu ed du ing an ins u- men al b eak be ween days 181 and 204. To compa e diu nal changes in he FCO, he da a we e u he di ided in o day- ime (FCO_day)and nigh - ime (FCO_nigh )da a. We used sun ele a ion angle h<0 o nigh - ime and h> 0 o day ime. Pea son co ela ions be ween day ime and nigh - ime hal - hou a e age luxes and o he measu ed pa ame e s we e de- e mined. Da a p ocessing was pe o med wi h Ma lab e - sion R2014a (The Ma hWo ks, Inc., Uni ed S a es) and he s a is ical es ing wi h IBM SPSS s a is ics 23 (IBM Co po- a ion, Uni ed S a es). To e alua e he g oss CO emission du ing he day ime (g oss day ime CO emission), we calcula ed he g oss day- ime CO emission in wo ways (1) by assuming an equi a- len CO up ake o day ime and nigh - ime (cons an up ake) and (2) by aking in o accoun empe a u e dependency (Q10 o 1.8) in CO up ake acco ding o Whalen and Reebu gh (2001). Based on a cons an CO up ake, he g oss day ime CO emission was calcula ed by sub ac ing he nigh - ime FCO (FCO_nigh ) om he day ime FCO (FCO_day), p esen ed in Table 1. The up ake CO luxes e e s o he es ima ed CO up ake aking place du ing he day, based on measu ed CO up ake alues a nigh . The empe a u e-co ec ed day ime CO up ake (day ime CO up ake, Q10 1.8) is calcula ed by ex apola ing he measu ed nigh - ime CO luxes (FCO_nigh ; Table 1) using he di e ence be ween day and nigh soil em- pe a u es (2.5 cm dep h; 1 soil)and he Q10 alue o 1.8 (Whalen and Reebu gh, 2001). The empe a u e-dependen day ime CO up ake (R2) was sol ed om he equa ion Q10 =R2 R110 (T 2−T1), whe e Q10 is 1.8 (Whalen and Reebu gh, 2001), R1 is he nigh - ime FCO (ne FCO_nigh ; nmol m−2s−1), and T2−T1 is he empe a u e di e ence be ween day ime (T2) and nigh - ime (T1) soil empe a u e a 2.5 cm dep h (◦C). The empe a u e-co ec ed g oss day ime CO emissions (g oss day ime CO emission; Q10 1.8) was es ima ed by sub ac - ing he empe a u e-co ec ed day ime CO up ake (day ime CO up ake, Q10 1.8) om he day ime FCO (FCO_day). These g oss CO emission and up ake a es we e es ima ed o each o he six measu emen pe iods and a e p esen ed in Table 2. 3 Resul s 3.1 Seasonal a ia ion The RCG ield was a ne sou ce o CO om mid-Ap il in he sp ing o mid-June (days 110–160), a e which he si e u ned o a ne sink un il he end o he measu emen pe iod in No embe 2011 (days 161–325; Fig. 1 ). Cumula i e CO lux (cumula i e FCO)cu es, calcula ed by cumula ing he hal -hou ly luxes, show ha he si e was a ne sink o CO o e he 7-mon h measu emen pe iod (Fig. 1 ). Du ing day- ime, he ne CO luxes (FCO_day)we e posi i e in sp ing and ea ly summe (days 110–160) and again du ing la e summe (days 205–240). These day ime emissions we e highes in sp ing (Table 1). Nigh - ime CO luxes (FCO_nigh )we e neg- a i e (CO up ake) h oughou he whole measu emen pe iod wi h a end o inc easing CO consump ion owa ds la e au- umn (Table 1). The sp ing emission pe iod (days 110–145) co e ed a ime (days 110–118) wi h a s anding d y c op om he p e i- ous yea . The old c op was ha es ed on 28 o Ap il (day www.biogeosciences.ne /13/5471/2016/ Biogeosciences, 13, 5471–5485, 2016 5476 M. Pihla ie e al.: Seasonal and diu nal a ia ion in CO luxes Table 2. Mean, median and 25–75 h pe cen iles o he es ima ed g oss day ime CO emission (g oss day ime CO emission, nmol m−2s−1), empe a u e-co ec ed day ime CO up ake (day ime CO up ake; Q10 1.8) and empe a u e-co ec ed g oss day ime CO emission (g oss day ime CO emission; Q10 1.8) calcula ed o he ead cana y g ass (RCG) c op a Maaninka. The CO emission and up ake a es a e calcula ed o six measu emen pe iods (S is sp ing, ES is ea ly summe , MS is midsumme , LS is la e summe , A is au umn, LA is la e au umn) and o e he ull measu emen pe iod (all) om Ap il o No embe 2011. The es ima ed g oss day ime CO emission is calcula ed in wo ways: (1) assuming a cons an CO up ake and (2) assuming empe a u e-dependen CO up ake. G oss day ime CO emission based on a cons an CO up ake (way 1, Sec . 2.4) e e s o he di e ence be ween day ime luxes (FCO_day)and nigh - ime luxes (FCO_nigh ) p esen ed in Table 1. The empe a u e-co ec ed g oss day ime CO emission (g oss day ime CO emission; Q10, 1.8) e e s o he di e ence be ween day ime luxes (FCO_day; Table 1) and day ime CO up ake (Q10, 1.8). The day ime CO up ake (day ime CO up ake; Q10, 1.8) is calcula ed by ex apola ing he nigh - ime CO luxes (FCO_nigh ) o day ime using he di e ence be ween day and nigh soil empe a u es (2.5 cm dep h; 1 soil)and he Q10 alue o 1.8 (Whalen and Reebu gh, 2001), as desc ibed in Sec . 2.4. G oss day ime CO emission 1 soil Day ime CO up ake (Q10, 1.8) G oss day ime CO emission (Q10, 1.8) Pe iod, DOY mean median 25 h–75 h Tday-Tnigh mean median 25 h–75 h mean median 25 h–75 h pe cen ile pe cen ile pe cen ile S, 110–145 1.61 1.24 0.83 2.20 2.1 −1.24 −1.09 −1.89 −0.39 2.22 1.76 1.74 2.39 ES, 145–160 0.91 0.57 0.43 0.91 1.2 −1.27 −0.92 −1.36 −0.63 1.51 1.00 1.06 1.20 MS, 160–181 0.59 0.45 0.46 0.46 0.7 −1.23 −0.96 −1.58 −0.41 1.15 0.89 1.18 0.65 LS, 205–240 1.12 0.79 0.89 1.07 0.9 −1.42 −0.91 −1.78 −0.36 1.77 1.21 1.71 1.24 A, 240–295 0.54 0.42 0.41 0.45 1.0 −1.24 −1.13 −1.68 −0.59 1.11 0.95 1.19 0.72 LA, 295–325 0.42 0.42 0.43 0.39 0.3 −1.90 −1.84 −2.49 −1.18 1.28 1.25 1.56 0.92 ALL, 110–325 0.98 0.68 0.65 0.88 3.5 −1.58 −1.37 −2.19 −0.68 1.79 1.38 1.78 1.23 118), a e which he g ound consis ed mainly o sho dead plan ma e ial and li e and a slowly sp ou ing new RCG. The second emission pe iod in ea ly summe (days 146–160) was cha ac e ized by as -g owing RCG c op, high e ilize - induced N2O emissions (Shu pali e al., 2016), inc easing ai and soil empe a u es, g owing lea a ea and inc easing NEE (Fig. 1). A e he c op had eached i s maximum heigh o 1.9 m in mid-June (a ound day 160), he si e s a ed o ac as a ne sink o CO, ollowed by a pe iod o ne day ime emis- sions du ing la e summe in July-Augus (days 205–240). The au umn (A, LA) was cha ac e ized by dec easing day- ime FCO (FCO_day)and slowly d opping ai and soil empe - a u es, dec easing adia ion in ensi y and dec easing pho o- syn he ic ac i i y o he c op (less nega i e NEE; Fig. 1). Compa ison o he wo gas analyse s, AR-CW-QCL and LGR-CW-QCL, du ing he pe iod when bo h we e ope a- ional (days 205–325), shows ha he measu ed FCO ag ee easonably well (Fig. 1 ). A co ela ion sca e plo o he FCO om LGR-CW-QCL agains FCO o AR-CW-QCL e- sul s in a co ela ion coe icien o 0.95 and a slope o 0.96 (da a no shown). Acco ding o his compa ison, LGR-CW- QCL shows sligh ly (4 %) smalle luxes compa ed o AR- CW-QCL; howe e , he di e ence be ween he wo analy- se s is e y small, gi ing us con idence in he use o ei he analyse in u he analysis. 3.2 Diu nal a ia ion The FCO had a dis inc diu nal pa e n wi h an up ake in he nigh - ime and an emission du ing he day ime wi h max- imum emissions a noon (Fig. 2). This pa e n was mos p onounced du ing he sp ing, on days 110–145, when he maximum day ime CO emissions eached 2.7 nmol m−2s−1 (Fig. 2). The ne FCO was posi i e (emission) in sp ing and ea ly summe , a e which he nigh - ime up ake domina ed, making he si e a ne sink o CO (Fig. 2, Table 1.). Nigh - ime FCO show a nea cons an up ake o CO o e he whole mea- su emen pe iod wi h a mean o −0.77 nmol m−2s−1o e he whole measu emen pe iod (Fig. 2, Table 1.). The diu nal FCO o e he six measu emen pe iods closely ollowed he daily pa e n o Rglob wi h a maximum FCO (emission) a a ound noon and minimum FCO (highes up- ake) a midnigh (Figs. 2 and 3). The highes adia ion in- ensi y was eached du ing he ea ly summe (days 146–160), while he maximum FCO we e obse ed in sp ing (days 110– 145; Figs. 2 and 3). Diu nal a ia ion in soil empe a u e was highes in sp ing and ea ly summe and always peaked du - ing he a e noon (Fig. 3). Compa ed o he FCO, he diu nal a ia ion in CO2ex- change, exp essed he e as NEE, was e y small du ing sp ing (days 110–145; Fig. 4). A apid inc ease in LAI and GAI a a ound day 150 (Fig. 1d) led o an inc ease in CO2up ake du ing day ime, which is seen in a dis inc diu nal pa e n wi h high CO2up ake (nega i e NEE) du ing day ime and a small posi i e NEE du ing nigh - ime (Fig. 4). Maximum NEE alues we e eached du ing mid-June (days 161–181) a e which he NEE slowly dec eased and he CO2up ake disappea ed by mid-Oc obe (day 290; Figs. 1 and 4). Du ing ea ly summe , he luxes o N2O ollowed a simila daily pa e n as ha o FCO wi h highe day ime N2O emis- sions compa ed o nigh - ime luxes (Shu pali e al., 2016). This pe iod o high N2O emissions (days 143–158) was a di- ec esponse o he NPKS e ilize applica ion on 23 May, and i las ed o abou 15 days. A e his, an opposi e diu nal Biogeosciences, 13, 5471–5485, 2016 www.biogeosciences.ne /13/5471/2016/ M. Pihla ie e al.: Seasonal and diu nal a ia ion in CO luxes 5477 Table 3. Pea son co ela ion ma ix o hal -hou day ime CO luxes (FCO_day)du ing six pe iods (S is sp ing, ES is ea ly summe , MS is midsumme , LS is la e summe , A is au umn, LA is la e au umn) a he eed cana y g ass c op in Maaninka. MCO is CO mixing a io, NEE is ne ecosys em exchange, RESP is ecosys em espi a ion, FN2Ois N2O lux, His sensible hea lux, LE is la en hea lux, Tai is ai empe a u e, Rglob is global adia ion, Rne is ne adia ion, Gis soil hea lux, Tsoil is soil empe a u e a 2.5 cm, SWC is soil wa e con en a 2.5 cm. FCO_day FCO_day FCO_day FCO_day FCO_day FCO_day S, 110–145 nES, 146–160 nMS, 161–180 nLS, 205–240 nA, 241–295 nLA, 296–325 n MCO 0.080∗711 0.128∗∗ 510 −0.116∗436 −0.074 488 0.038 851 −0.284∗∗ 288 NEE −0.188∗∗ 711 −0.469∗∗ 510 −0.308∗∗ 436 −0.488∗∗ 488 −0.237∗∗ 850 −0.25∗∗ 288 RESP 0.015 711 0.274∗∗ 510 0.272∗∗ 436 0.257∗∗ 488 0.198∗∗ 850 0.077 288 FN2O−0.219∗∗ 669 0.000 453 −0.293∗∗ 426 −0.026 478 −0.085∗850 −0.172∗∗ 287 H0.729∗∗ 711 0.329∗∗ 510 0.234∗∗ 436 0.427∗∗ 488 0.132∗∗ 851 −0.076 288 LE 0.402∗∗ 418 0.398∗∗ 401 0.514∗∗ 224 0.625∗∗ 307 0.317∗∗ 573 0.289∗∗ 185 RH −0.537∗∗ 711 −0.176∗∗ 510 −0.303∗∗ 436 −0.434∗∗ 488 −0.081∗851 −0.179∗∗ 288 Tai 0.425∗∗ 711 0.344∗∗ 510 0.36∗∗ 436 0.433∗∗ 488 0.241∗∗ 851 0.073 288 Rglob 0.760∗∗ 711 0.498∗∗ 510 0.373∗∗ 436 0.549∗∗ 488 0.265∗∗ 851 0.256∗∗ 288 Rne 0.760∗∗ 711 0.515∗∗ 510 0.376∗∗ 436 0.558∗∗ 488 0.277∗∗ 851 0.218∗∗ 288 G0.575∗∗ 711 0.473∗∗ 510 0.406∗∗ 436 0.485∗∗ 488 0.247∗∗ 851 0.033 288 Tsoil 0.191∗∗ 711 0.282∗∗ 510 0.318∗∗ 436 0.358∗∗ 488 0.206∗∗ 851 0.071 288 SWC −0.099∗∗ 711 0.033 510 0.095∗436 0.086 488 −0.105∗∗ 851 0.095 288 ∗∗ Co ela ion is signi ican a he 0.01 le el (2- ailed). ∗Co ela ion is signi ican a he 0.05 le el (2- ailed). pa e n was obse ed du ing which he N2O emissions we e on a e age 50% highe du ing he nigh han du ing he day (Shu pali e al., 2016). The g oss day ime CO emissions we e es ima ed in wo ways: (1) assuming an equal CO up ake du ing day and nigh (cons an up ake) and (2) accoun ing o empe a u e-dependen CO up ake acco ding o Whalen and Reebu gh (2001). The g oss CO emissions calcula ed in ei- he way, show ha in he day ime he si e emi ed CO h oughou he whole measu emen pe iod wi h he highes emissions in sp ing and la e summe (Table 2). Du ing mid- summe and au umn he day ime emissions we e ma kedly smalle and less han hal o he emissions in sp ing. The smalles g oss CO emissions we e measu ed in la e au umn (Table 2). When he empe a u e dependency in he CO up ake was aken in o accoun , using a Q10 alue o 1.8 (Whalen and Reebu gh, 2001), bo h he day ime CO up ake (day ime CO up ake, Q10, 1.8), and he day ime emission (day ime CO emission, Q10, 1.8) we e almos wice as high as he a es wi hou he empe a u e co ec ion (Table 2). 3.3 D i ing ac o s o CO luxes The mos p onounced ela ionships be ween FCO and o he measu ed scala s we e ound o he day ime da a (sun ele a- ion h> 0) du ing he wo emission pe iods in he sp ing and ea ly summe (Table 3, Fig. 5). Fu he mo e, he s onges co ela ions we e ound in sp ing be ween FCO_day and Rglob ( =0.760, p< 0.01), Rne ( =0.760, p< 0.01), H ( =0.729, p< 0.01) and G( =0.575, p< 0.01). These pos- i i e co ela ions emained signi ican bu became weake o- wa ds he end o he measu emen pe iod (Table 3, Fig. 5). S ong nega i e co ela ions we e ound in sp ing be ween FCO_day and RH ( = −0.537, p< 0.01), and du ing he ea ly summe wi h NEE ( = −0.469, p< 0.01), while he co - ela ion be ween day ime FCO and MCO,FN2Oo ecosys- em espi a ion (RESP) we e e y weak h oughou he 7- mon h measu emen pe iod (Table 3). Nigh - ime (h< 0) FCO (FCO_nigh )co ela ed weakly wi h FN2O( = −0.336, p< 0.01), H( =0.315, p< 0.01), and LE ( =-0.241, p< 0.05) in he sp ing and wi h SWC ( =0.308, p< 0.01) du ing ea ly summe (Table 4). A s ong nega i e co ela- ion was ound be ween FCO_nigh and FN2Odu ing midsum- me ( = −0.607, p< 0.01) and la e au umn ( = −0.514, p< 0.01) and a posi i e co ela ion was ound be ween FCO_nigh and LE ( =0.459, p< 0.05) du ing midsumme (Table 4). 4 Discussion Based on he 7-mon h EC lux measu emen s a he RCG c op, we demons a e ha he EC me hod is sui able o mea- su ing CO luxes (FCO) om a pe ennial ag icul u al c op. We show ha he soil–plan sys em ac ed as a ne sou ce o CO in sp ing and ea ly summe and a ne sink o CO o e he la e summe and au umn, and ha he FCO had a clea di- u nal pa e n, wi h ne CO emissions in he day ime and ne CO up ake a nigh . This sou ce–sink pa e n exis ed o e he whole measu emen pe iod wi h dec easing ne emissions o- wa ds he end o he au umn. To ou knowledge, simila long- e m and con inuous FCO da a se ies measu ed by he EC me hod o e any ecosys em ype does no exis , hence his s udy is unique in b inging new insigh o he unde s anding o sho - e m diu nal and long- e m seasonal FCO dynam- ics a ecosys em le el. Combining he con inuous FCO da a www.biogeosciences.ne /13/5471/2016/ Biogeosciences, 13, 5471–5485, 2016 5478 M. Pihla ie e al.: Seasonal and diu nal a ia ion in CO luxes Table 4. Pea son co ela ion ma ix o hal -hou nigh - ime CO luxes (FCO_nigh )du ing six pe iods (S is sp ing, ES is ea ly summe , MS is midsumme , LS is la e summe , A is au umn, LA is la e au umn) a he eed cana y g ass c op in Maaninka. MCO is CO mixing a io, NEE is ne ecosys em exchange, RESP is ecosys em espi a ion, FN2Ois N2O lux, His sensible hea lux, LE is la en hea lux, Tai is ai empe a u e, Rglob is global adia ion, Rne is ne adia ion, Gis soil hea lux, Tsoil is soil empe a u e a 2.5 cm, SWC is soil wa e con en a 2.5 cm. FCO_nigh FCO_nigh FCO_nigh FCO_nigh FCO_nigh FCO_nigh S, 110–145 nES, 146–160 nMS, 161–180 nLS, 205–240 nA, 241–295 nLA, 296–325 n MCO −0.045 380 −0.043 142 −0.279∗∗ 134 −0.165∗∗ 324 −0.110∗∗ 1149 −0.041 700 NEE 0.069 380 −0.167∗142 −0.118 134 −0.049 324 0.024∗∗ 1149 0.025 700 RESP 0.056 380 0.015 142 −0.006∗∗ 134 0.125∗∗ 324 0.062∗1149 0.072 700 FN2O−0.336∗∗ 350 0.034 120 −0.607∗∗ 126 −0.197∗∗ 307 0.009 1140 −0.514∗∗ 696 H0.315∗∗ 380 0.170∗142 0.002 134 0.051 324 −0.021∗∗ 1149 0.080∗700 LE −0.241∗74 0.099 72 0.459∗20 −0.078 62 0.135∗∗ 453 0.161∗∗ 279 RH 0.027 380 −0.016 142 −0.057 134 −0.12∗∗ 324 −0.033 1149 −0.041∗∗ 700 Tai 0.107∗380 −0.013 142 0.092 134 0.249∗∗ 324 0.138∗∗ 1149 0.098∗∗ 700 Rglob 0.077 380 0.118 142 −0.096 134 −0.02 324 −0.001 1149 −0.041∗∗ 700 Rne 0.011 380 0.111 142 0.026 134 0.087 324 0.043 1149 −0.053∗∗ 700 G0.050 380 0.029 142 0.121 134 0.207∗∗ 324 0.175∗∗ 1149 0.162∗∗ 700 Tsoil 0.075 380 −0.146 142 −0.035 134 0.167∗∗ 324 0.038 1149 0.117∗∗ 700 SWC 0.043 380 0.308∗∗ 142 0.212 134 0.138∗324 0.093∗∗ 1149 0.008 700 ∗∗ Co ela ion is signi ican a he 0.01 le el (2- ailed). ∗Co ela ion is signi ican a he 0.05 le el (2- ailed). Table 5. Repo ed CO luxes measu ed in di e en ecosys ems and clima ic egions, using chambe s ( anspa en o da k), mic ome eo olog- ical lux g adien o eddy co a iance me hods and he epo ed da a pe iod, measu emen equency and he momen o he measu emen s. Re e ence Ecosys em, clima e, coun y Measu emen me hod Da a pe iod, FCO measu emen equency, (nmol m−2s−1) momen o measu emen Zepp e al. (1997) Black sp uce o es , bo eal, Mani oba, Canada Chambe s, anspa en 3 mon hs, weekly, day ime −1.06 Zepp e al. (1997) Jack pine o es , bo eal, Mani oba, Canada Chambe s, anspa en 3 mon hs, weekly, day ime −0.58 King (2000) Pine o es , No h-eas , Walpole, Maine, USA Chambe s, da k 1.3 yea s, biweekly, day ime 1.12 King (2000) Mixed ha dwood-coni e ous o es , Walpole, Maine, USA Chambe s, da k 1.3 yea s, biweekly, day ime 0.62 King (2000) Pine o es , G i in, Geo gia, USA Chambe s, da k 1 yea , bimon hly, day ime −0.21 King (2000) Pine o es , Ti on, Geo gia, USA Chambe s, da k 1 yea , bimon hly, day ime −0.95 Kuhlbusch e al. (1998) Black sp uce, bo eal, Mani oba, Canada Chambe s, da k 1 yea , bimon hly, day ime −1.11 Galbally e al. (2010) Mallee, Eucalyp us sp. Ecosys em, opical, Aus alia Chambe s, anspa en 1 yea , bimon hly, day ime 0.61 Kisselle e al. (2002) Ce ado, campo sujo, opical, B azil Chambe s, anspa en 1 yea , mon hly, day ime 3.16 Kisselle e al. (2002) Ce ado, s ic o sensu, opical, B azil Chambe s, anspa en 1 yea , mon hly, day ime 2.66 Va ella e al. (2004) Na u al ce ado, opical, B azil Chambe s, anspa en 1.5 yea s, mon hly, day ime 1.91 Va ella e al. (2004) Pas u e (B achia ia b izan ha), opical, B azil Chambe s, anspa en 1.5 yea s, mon hly, day ime 1.20 King (2000) C opland, co n, Walpole, Maine, USA Chambe s, da k 1.3 yea s, biweekly, day ime 2.19 King (2000) C opland, so ghum/whea , G i in, Geo gia, USA Chambe s, da k 1 yea , bimon hly, day ime 1.16 King (2000) C opland, co on/peanu s/win e whea , Ti on, Geo gia, USA Chambe s, da k 1 yea , bimon hly, day ime 1.03 Galbally e al. (2010) C opland, whea , opical, Aus alia Chambe s, anspa en 1 yea , bimon hly, day ime 0.98 Cons an e al. (2008) G assland, bo eal, Quebec, Canada Flux g adien 1 yea , diu nal cycle −2.11 B uhn e al. (2013) G assland, empe a e, Denma k Chambe s, da k 2 mon hs, mon hly, day ime −0.78 B uhn e al. (2013) G assland, empe a e, Denma k Chambe s, anspa en 2 mon hs, mon hly, day ime 0.36 an Aspe en e al. (2015) G assland, Medi e anean, I aly Chambe s, anspa en 5 weeks, summe , diu nal cycle 0.35 an Aspe en e al. (2015) G assland, Medi e anean, I aly Flux g adien 1 mon h, 30 min, diu nal cycle 1.74 his s udy G assland, eed cana y g ass, bo eal, Finland Eddy co a iance 7 mon hs, 30 min, diu nal cycle −0.25 wi h simul aneously measu ed CO2, N2O and ene gy luxes as well as me eo ological and soil a iables allowed us o dis inguish d i ing a iables o he FCO and demons a e he sui abili y o he EC me hod o analyse ecosys em-le el CO exchange dynamics. Due o he ac ha he EC me hod mea- su es ne luxes, we canno di ec ly sepa a e di e en p o- cesses such as CO p oduc ion and consump ion. Howe e , based on p ocess unde s anding and ou da a, we made an assump ion ha mos o he CO p oduc ion akes place du - ing day ime and ha he nigh - ime CO up ake is due o mi- c obial ac i i y. A e hese assump ions, we di ided he da a in o day ime and nigh - ime pe iods in o de o analyse sea- sonal changes in dependencies be ween CO emissions and up ake and hei d i ing a iables. Cumula i e FCO o e he whole 7-mon h measu emen pe- iod showed ha he RCG c op was a ne sink o CO. This cu- mula i e FCO es ima ion may be biased due o he ins umen- al b eak du ing July (days 181–205), du ing which we do no ha e an es ima e o he CO luxes. Also, due o he ac ha he da a p ocessing emo ed mo e day ime alues (40.8 % e- Biogeosciences, 13, 5471–5485, 2016 www.biogeosciences.ne /13/5471/2016/ M. Pihla ie e al.: Seasonal and diu nal a ia ion in CO luxes 5479 Figu e 1. (a) Daily mean ai and soil empe a u es, (b) global a- dia ion sum (Rglob),(c) daily p ecipi a ion sum (P )and soil wa e con en (SWC), (d) weekly lea a ea index (LAI; black) and g een a ea index (GAI; g ey), (e) ne ecosys em exchange o CO2(NEE), and ( ) cumula i e CO luxes calcula ed om hal -hou mean CO luxes (cumula i e FCO; black lines) and day ime mean CO luxes (FCO_day; g ey) o e he 7-mon h measu emen pe iod in a eed cana y g ass c op. Measu emen pe iods (S is sp ing, ES is ea ly summe , MS is midsumme , LS is la e summe , A is au umn, LA is la e au umn) a e sepa a ed by solid lines. mo ed) compa ed o nigh - ime da a (24.1 % emo ed), he nigh - ime CO up ake is weighing mo e in he cumula i e lux es ima ion, po en ially leading o smalle and mo e neg- a i e ne luxes han es ima ed based on an equal numbe o lux da a om day ime and nigh - ime. We es ed a simple s a is ical gap- illing me hod o ob ain a balanced numbe o day ime and nigh - ime da a. Howe e , as his gap- illing did no change he in e p e a ion o he esul s, and as we do no ha e an app op ia e p ocess model o accoun o up ake and emission p ocesses, we decided no o p esen hese esul s. Based on seasonal a ia ion, we can di ide he FCO in o a dis inc emission pe iod and an up ake pe iod. Du ing he emission pe iod (days 110–160), he soil–plan sys em was a s ong sou ce o CO du ing he day ime and a small sink du ing nigh - ime. Fu he mo e, he emission pe iod was di- Time (h) 0 6 12 18 24 FCO (nmol m-2 s-1) -2 0 2 4S, 110–145 Time (h) 0 6 12 18 24 FCO (nmol m-2 s-1) -2 0 2 4ES, 146–160 Time (h) 0 6 12 18 24 FCO (nmol m-2 s-1) -2 0 2 4MS, 161–181 Time (h) 0 6 12 18 24 FCO (nmol m-2 s-1) -2 0 2 4LS, 205–240 Time (h) 0 6 12 18 24 FCO (nmol m-2 s-1) -2 0 2 4A, 241–295 Time (h) 0 6 12 18 24 FCO (nmol m-2 s-1) -2 0 2 4LA, 296–325 Figu e 2. Diu nal cycle o hal -hou mean CO luxes (FCO, nmol m−2s−1) om he eed cana y g ass c op om six dis inc pe iods du ing he Ap il o No embe 2011. G ey a eas indica e he momen o sun ise and sunse , and he e ical ba s indica e ±1 SD o he luxes. ided in o a sp ing emission pe iod (days 110–145) and an ea ly summe emission pe iod (days 146–160), which di - e ed om each o he based on he day ime CO emission a es and ela ionships wi h o he measu ed a iables such as adia ion and NEE. The highes CO emissions we e ob- se ed soon a e he snowmel du ing sp ing om Ap il o ea ly May when he ai and soil empe a u es we e a he low, he c op was no ye ac i ely pho osyn hesizing (low LAI, low NEE) and adia ion in ensi y was al eady a he high. As sugges ed by King (2000), he ele a ed sp ing ime CO emissions p obably esul ed om he deg ada ion o he las yea ’s eadily a ailable c op and li e , which ha e been shown o be a signi ican sou ce o CO (King, 2000; King e al., 2012; Lee e al., 2012). Dec easing amoun s o his ead- ily deg adable li e also pa ly explains he dec easing end in CO emissions du ing sp ing and ea ly summe (King, 2000). In gene al, he FCO a es om he RCG c op in his s udy all in o he same ange as hose epo ed om di e en na - u al and managed ecosys ems ac oss he di e en clima ic egions (Table 5). The e is a endency o highe CO emis- sions om opical and Medi e anean ecosys ems compa ed o no he n and bo eal ecosys ems. The da a compa ison also indica es ne CO up ake om o es ecosys ems (Zepp e www.biogeosciences.ne /13/5471/2016/ Biogeosciences, 13, 5471–5485, 2016