Carbon dioxide fluxes increase from day to night across European streams

ARTICLE Ca bon dioxide fluxes inc ease om day o nigh ac oss Eu opean s eams Ka in A e meye 1,2,3✉, Joan Pe e Casas-Ruiz 4,5, Thomas Fuss6, Ada Pas o 4,5,34, Sophie Cau y-F aunié7, Danny Shea h8,35, Anna C. Nydahl1, Albe o Do e o9,10, Ana Paula Po ela 11,12, B ian C. Doyle13, Nikolay Simo 14, Ca he ine Gu mann Robe s8, Geo g H. Nied is 15, Xisca Timone 4,5, Vesela E imo a 16, Lau a Ba al-F aga5, Tea Bašić8,36, Joachim Aude 17,37, Anne Deininge 18,38, Geo gina Buss 8, S e ano Fenoglio 10,19, Nú ia Ca alán 4,5,39,40, El i a de Ey o 20, F ancesca Pilo o 18,41, Jo di-René Mo 4,21, Juliana Mon ei o22, Da id Fle che 8, Ch is ian Noss23, Mi iam Colls4,5, Magdalena Nagle 24, Liu Liu 23,25, Cla a Rome o González-Quijano26, Fe an Rome o4,5, Nina Pansch25, José L. J. Ledesma 17,27,28, Josephine Pegg 8,29, Ma cus Klaus18,42, Anna F eixa 4,5, Sonia He e o O ega25, Cla a Mendoza-Le a 7,23, Adam Bednařík30,43, Jé émy A. Fon ielle25, Pe e J. Gilbe 31, Lyubomi A. Kende o 32, Ma in Rulík30 & Pascal Bodme 23,33,44 Globally, inland wa e s emi o e 2 Pg o ca bon pe yea as ca bon dioxide, o which he majo i y o igina es om s eams and i e s. Despi e he global significance o flu ial ca bon dioxide emissions, li le is known abou hei diel dynamics. He e we p esen a la ge-scale assessmen o day- and nigh - ime ca bon dioxide fluxes a he wa e -ai in e ace ac oss 34 Eu opean s eams. We di ec ly measu ed fluxes ou imes be ween Oc obe 2016 and July 2017 using d i ing chambe s. Median fluxes a e 1.4 and 2.1 mmol m−2h−1a midday and midnigh , espec i ely, wi h nigh fluxes exceeding hose du ing he day by 39%. We a i- bu e diel ca bon dioxide flux a iabili y mainly o changes in he wa e pa ial p essu e o ca bon dioxide. Howe e , no consis en d i e s could be iden ified ac oss si es. Ou findings highligh widesp ead day-nigh changes in flu ial ca bon dioxide fluxes and sugges ha he ime o day g ea ly influences measu ed ca bon dioxide fluxes ac oss Eu opean s eams. h ps://doi.o g/10.1038/s43247-021-00192-w OPEN A lis o au ho a filia ions appea s a he end o he pape . COMMUNICATIONS EARTH & ENVIRONMENT | (2021) 2:118 | h ps://doi.o g/10.1038/s43247-021-00192-w | www.na u e.com/commsen 1 1234567890():,; Inland wa e s a e impo an sou ces o a mosphe ic ca bon dioxide (CO 2 ) pa ially o se ing he e es ial ca bon sink1,2. S eams and i e s he ein ep esen majo CO 2 emi e s3. Flu ial CO 2 fluxes a e p ima ily con olled by he gas exchange eloci y a he wa e -ai in e ace (k) and he g adien be ween he wa e and a mosphe ic pa ial p essu es o CO 2 (pCO 2 )4. Bo h pa ame e s a e highly a iable in space and ime5,6, causing unce ain y in he magni ude o egional and global flu ial CO 2 emissions2. The high spa io empo al a iabili y o kand wa e pCO 2 can be a ibu ed o a complex in e play o unde lying con ols. While k in s eams is mos ly d i en by wa e u bulence c ea ed by a - ia ions in flow and s eam mo phology7, he wa e pCO 2 is influenced by he deg ee o hyd ological connec i i y be ween he s eam and he adjacen ipa ian soils8as well as by in-s eam p ocesses (e.g., s eam me abolism). The supply o CO 2 om ex e nal sou ces, such as soil wa e o g oundwa e , in o s eams, a ies wi h each and season5,9. Fu he mo e, seasonal and diel changes in s eam pCO 2 a e a ibu ed o s eam me abolism d i en by empe a u e and sola adia ion10–13. Ecosys em espi a ion, a sou ce o CO 2 in he s eam, akes place h oughou he whole day, and g oss p ima y p oduc ion, a sink o CO 2 , occu s only du ing dayligh . Tempe a u e and sola adia ion also di ec ly influence wa e pCO 2 , he o me by changing he solu- bili y o he gas and he la e due o pho omine aliza ion14. Howe e , ques ions emain ega ding he magni ude and ela i e d i e s o seasonal and diel fluc ua ions o CO 2 fluxes in s eams. P esen ly, mos flu ial CO 2 emission alues a e de i ed om k es ima es based on wa e eloci y and s eam channel slope and on wa e pCO 2 alues indi ec ly calcula ed om alkalini y, pH, and empe a u e3. This app oach ails o cap u e he high spa- io empo al a iabili y obse ed o kand pCO 2 and he e o e can p o ide imp ecise es ima es o CO 2 fluxes15,16. Di ec field obse a ions p o ide he means o imp o e es ima es and unde s anding o he d i e s behind spa io empo al a iabili y, and hus he dynamics o CO 2 ou gassing om unning wa e s. Howe e , besides mos ly local s udies ha indi ec ly in e CO 2 fluxes om pCO 2 and k11,12,17,18, no di ec measu emen s exis ha compa e day- ime and nigh - ime CO 2 fluxes om s eams on a la ge spa ial scale. The aim o his s udy was o assess he magni ude and d i e s o s eam CO 2 flux a ia ions be ween day and nigh ac oss Eu opean s eams. We hypo hesized ha CO 2 fluxes would di e be ween day and nigh due o diel a ia ions in e es ial ino - ganic ca bon inpu s, in si u me abolism, and empe a u e. As highe empe a u es and sola adia ion may d i e di e ences in pCO 2 , we expec ed a highe di e ence be ween day- ime and nigh - ime fluxes wi h wa me empe a u es and a lowe la i- udes. Hence, we measu ed day- ime and nigh - ime fluxes o CO 2 a ou di e en pe iods h oughou one yea om 34 s eams (S ahle s eam o de s om 1 o 6) in 11 coun ies ac oss Eu ope ollowing a s anda dized p ocedu e. CO 2 fluxes we e measu ed s a ing a midday (11 a.m. G eenwich Mean Time (GMT)) and midnigh (11 p.m. GMT) wi h d i ing flux chambe s equipped wi h CO 2 senso s as desc ibed in Bas iken e al.19. In he majo i y o he Eu opean s eams, we ound inc eased CO 2 fluxes a he wa e –ai in e ace in he nigh compa ed o he day wi h a median inc ease o 0.5 mmol m−2 h−1. Mos o he obse ed CO 2 flux a iabili y was explained by changes in pCO 2 om day o nigh wi h mo e p onounced changes a lowe la i udes. Resul s and discussion Magni ude o CO 2 flux a ia ion om day o nigh . Midday CO 2 fluxes a he wa e –ai in e ace anged om −2.7 (up ake) o 19.9 mmol m−2h−1(emission) (1.4 [0.5, 3.1]; median [in e - qua ile ange (IQR)]; n=107) and midnigh fluxes anged om −0.3 o 25.6 mmol m−2h−1(2.1 [0.9, 3.7]; n=107) (Fig. 1a; Supplemen a y Table S3). Ou measu ed fluxes a e compa able o o he s udies conduc ed in empe a e and bo eal s eams ha used chambe s20,21 o empi ical models12,22,23, al hough hey we e in he lowe ange o he numbe s modeled in a s udy in he USA23 (Supplemen a y Fig. S2). The lowe numbe s migh be due o he lack o ibu a y inflows, la ge woody deb is, and s ong hyd aulic jumps in he selec ed s eam sec ions (Supplemen a y Sampling manual). To assess s eam CO 2 flux a ia ions be ween day and nigh , we compu ed he di e ence o nigh - ime minus day- ime fluxes o each s eam and sampling pe iod, whe e posi i e numbe s indica e an inc ease om day o nigh and ice e sa (Fig. 1b). Di e ences in CO 2 fluxes amoun ed o 0.5 mmol m−2h−1[0.1, 1.4] (n=107) ac oss all si es and sampling pe iods, which is equi alen o a ela i e inc ease o 39% [4%, 100%] (n=101; n educed due o exclusion o ela i e compa isons o ze o flux a day- ime) (Fig. 2). Al oge he , hese esul s poin owa ds a high ele ance o nigh - ime CO 2 fluxes as epo ed ea lie o single p e-alpine s eams12, s eam ne wo ks13,17 o i e s18, and in a ecen compila ion o diel CO 2 da a om 66 s eams wo ldwide24. A ough annual ex apola ion o fluxes om ou s udy si es (Supplemen a y Me hods) shows ha he inclusion o nigh - ime fluxes inc eases annual es ima es o si e-specific s eam CO 2 emissions by 16% [6%; 25%] (Supplemen a y Table S4). Hence, ou measu emen s and he simplified ex apola ion o ou da a emphasize he need o collec and in eg a e nigh - ime CO 2 flux da a in o sampling p o ocols as well as egional upscaling e o s. Looking in o he indi idual compa isons, we ound 83 inc eases in median CO 2 fluxes om day o nigh wi h se en compa isons whe e he s eam e en swi ched om a sink o a sou ce o CO 2 o he a mosphe e (Supplemen a y Table S3). Howe e , we also ound ou compa isons whe e median CO 2 fluxes a day and nigh we e he same and 20 dec eases in he nigh (Supplemen a y Table S3). These esul s and also o he s udies13,25,26 sugges ha he di ec ion and s eng h o diel pCO 2 pa e n can be la gely a iable ac oss space and ime. Diel CO 2 flux di e ences a y as a unc ion o la i ude and wa e empe a u e. The diel di e ences in CO 2 fluxes we e sig- nifican ly nega i ely ela ed o la i ude (Table 1A), wi h sub- s an ial diel a ia ion mo e likely a lowe la i udes. Likewise, he in e ac ion wi h la i ude and he wa e empe a u e was sig- nifican (Table 1A), which migh be explained by highe em- pe a u es a lowe la i udes du ing he sampling pe iods and highe sola adia ion boos ing in-s eam p ima y p oduc ion27. This da ase is de i ed om only 34 s eams dis ibu ed ac oss di e en clima e zones in Eu ope. Howe e , o ou knowledge, i is cu en ly he la ges s udy o i s kind, using flux chambe s o measu e CO 2 fluxes, and compa e hose fluxes a day- ime and nigh - ime on such a spa ial scale. We ound no significan di e ences in he magni ude o diel di e ences in CO 2 fluxes ela ed o wa e empe a u e (Table 1A) using a linea mixed-e ec model (LME). Howe e , compa ing he CO 2 fluxes a midday o midnigh a he di e en sampling pe iods, we de ec ed significan diel changes in CO 2 fluxes in Oc obe , Janua y, and Ap il (Fig. 1a). Con a y o ou expec a ion ha highe di e ences can be expec ed a highe empe a u es, we did no de ec significan changes om day o nigh in July (Fig. 1a), du ing which pe iod he lowes changes in absolu e numbe s we e eco ded (0.3 mmol m−2h−1;Fig.1b). The highes di e ences o CO 2 fluxes om day o nigh we e measu ed du ing Ap il (1.1 mmol m−2h−1), ollowed by Janua y (0.5 mmol m−2h−1) ARTICLE COMMUNICATIONS EARTH & ENVIRONMENT | h ps://doi.o g/10.1038/s43247-021-00192-w 2COMMUNICATIONS EARTH & ENVIRONMENT | (2021) 2:118 | h ps://doi.o g/10.1038/s43247-021-00192-w | www.na u e.com/commsen and Oc obe (0.5 mmol m−2h−1). Lowe day-nigh changes in July could be explained by inc eased ipa ian shading educing pho osyn hesis28,29. Fo example, educed in-s eam pho osyn hesis in summe compa ed o sp ing has been shown o a subalpine s eam ne wo k29 o a empe a e o es ed headwa e s eam28. Howe e , compa ing he canopy co e o he s eams and he di e ences in CO 2 fluxes om day o nigh (Supplemen a y Fig. S3h) e ealed no clea pa e n. A p obable al e na e explana ion is ha CO 2 p oduc ion ia pho omine aliza ion du ing he day coun e ac ed a dec ease ia CO 2 fixa ion by pho osyn hesis30 and diminished diel pCO 2 and ul ima ely CO 2 flux changes. This highligh s he complex in e play be ween di e en ligh -dependen p ocesses in s eams influencing pCO 2 on a diel scale. The impo ance o yea - ound measu emen s is highligh ed by he Janua y da a se con aining he second-highes diel CO 2 flux changes. Eu opean ice- ee s eams may be pe cei ed “do man ” du ing hese pe iods and ep esen a i e CO 2 flux es ima es a e hus o en missing3. Ou Janua y da a showed a magni ude o flux compa ed o he es o he yea ac oss he Eu opean s eams as well as high diel a iabili y in CO 2 fluxes (Fig. 1). This may be a ibu ed in pa o he la i udinal co e age o ou s udy as we included s eams om he bo eal o he Medi e anean. Fo example, he wa e empe a u es o he Spanish s eams we e s ill ela i ely high in win e wi h a ound 2.8–9.5 °C du ing he day whe eas Swedish s eams showed hese empe a u es in Oc obe and Ap il. A s udy in he co e minous US looking in o s eam pCO 2 a iabili y also epo s a ying s eng hs o diel pCO 2 a iabili y, dependen on he in es iga ed s eam and ime25. Hence, diel pCO 2 and CO 2 flux a iabili y can be la ge in s eams o he no he n hemisphe e, s essing he need o un a el he si e- specific d i e s o and mechanisms behind hese diel changes. Fig. 1 Day- o-nigh changes o CO 2 fluxes a he wa e –ai in e ace o he sampled Eu opean s eams. S eam CO 2 fluxes (in mmol CO 2 m−2h−1)a day- ime (yellow) and nigh - ime (blue) (a) and he calcula ed changes om nigh minus day (ΔCO 2 flux) (b) o all da a and sepa a ely o each sampling pe iod. In he sampling pe iods compa isons in (a), CO 2 fluxes o indi idual s eam si es a e indica ed by ed (day) and ligh blue (nigh ) do s. The boxplo s isualize he median o all s eam si es (line), he fi s and hi d qua iles (hinges), he 1.5*in e -qua ile anges (whiske s), and he ou lie s ou side he ange o 1.5*in e -qua ile anges (black do s). On op o (a) a e p alues e ie ed om pai ed compa isons o median CO 2 fluxes es ed by Wilcoxon signed- ank es s and he sample size (n). Significan p alues wi h p< 0.05 a e in bold wi h an as e isk. The di e ences in he CO 2 fluxes (b)in mmol CO 2 m−2h−1 om day o nigh a e o Oc obe : 0.5 [0.1, 1.2]; Janua y: 0.5 [0.3, 0.9]; Ap il: 1.1 [0.1, 2.3]; July: 0.3 [−0.2, 1.1] (median [IQR]). Fig. 2 Rela i e changes in CO 2 fluxes om day o nigh o all da a oge he and o each sampling pe iod. A posi i e alue indica es an inc ease in CO 2 fluxes du ing he nigh and ice e sa (exp essed as a %-change o he day ime alues). Ou lie s (>1.5*IQR) we e excluded o illus a ion pu poses as he la ge ela i e a ia ion in hese fluxes was due o mino absolu e a ia ion in fluxes close o ze o. The median ela i e changes we e posi i e h oughou all sampling pe iods, anging om 32% [0.6%, 95%] in Oc obe , 38% [16%, 50%] in Janua y, 60% [7%, 177%] in Ap il, o 24% [−16%, 69%] in July (median [IQR]; n=26, 21, 28, and 26, espec i ely). COMMUNICATIONS EARTH & ENVIRONMENT | h ps://doi.o g/10.1038/s43247-021-00192-w ARTICLE COMMUNICATIONS EARTH & ENVIRONMENT | (2021) 2:118 | h ps://doi.o g/10.1038/s43247-021-00192-w | www.na u e.com/commsen 3 Diel CO 2 flux a iabili y d i en by changes in wa e pCO 2 .To unde s and he mechanisms behind he obse ed changes in CO 2 fluxes om day o nigh , we fi s selec ed he wo p ima y con- ols o CO 2 fluxes a he wa e –ai in e ace, i.e., he gas exchange eloci y and wa e pCO 2 and explo ed he influence o hese pa ame e s on absolu e CO 2 flux changes using an LME. The diel CO 2 flux a iabili y in Eu opean s eams could be mos ly a ibu ed o changes in wa e pCO 2 (Table 1B), whe eas changes in he gas exchange eloci y kappea ed less impo an . In ac , we did no measu e significan a ia ions in k om day o nigh in ou s eams (Fig. 3; Supplemen a y Fig. S4h). Al hough diel a iabili ies o gas exchange eloci ies ha e been epo ed o CO 2 and o he gases31,32, he majo i y o he in es iga ed s eams in his s udy did no show hose changes. The pCO 2 as a majo d i e o diel CO 2 flux a iabili y was also iden ified by a global compila ion o high- equency CO 2 measu emen s24. Conse- quen ly, i no majo changes in physical d i e s o gas exchanges occu ha s ongly a ec he u bulence, such as hea y ain e en s, i is su ficien o ocus on pCO 2 o assessing diel flux changes a he wa e –ai in e ace. In a second s ep, we es ed he influence o biogeochemical pa ame e s ha a y on a diel scale on wa e pCO 2 day- o-nigh di e ences (Table 1C). This LME iden ified a link be ween he day- o-nigh changes in wa e pCO 2 and wa e dissol ed O 2 , wi h pCO 2 gene ally inc easing and O 2 dec easing om day o nigh (Supplemen a y Fig. S4b, c). This po en ially eflec s a diel cycle o CO 2 con olled by aqua ic p ima y p oduc ion and espi a ion (in-s eam me abolism). Hence, e en hough in si u me abolism may play a mino ole in de e mining he baseline pCO 2 and flux in smalle s eams (mos ly con olled by e es ial inpu s23), ou esul s sugges ha me abolism can be an impo an d i e o he diel fluc ua ions in CO 2 fluxes. Indeed, inc eased wa e pCO 2 du ing he nigh has been a ibu ed o a dec ease in CO 2 fixa ion by p ima y p oduce s13,18,24, al hough a ecen s udy sugges s ha he adjacen g oundwa e can also show measu able bu less p onounced diel pCO 2 a ia ions33. P e ious esea ch sugges s ha in si u mine aliza ion o CO 2 should play a la ge ole in CO 2 dynamics in la ge s eams because hey a e less influenced by ex e nal CO 2 sou ces23. Ne e heless, we did no find any end in CO 2 flux day- o-nigh di e ences wi h s eam wid h o discha ge as a p oxy o size (Supplemen a y Fig. S3c, ) o wi h s eam o de (Supplemen a y Fig. S5) al hough o he s udies sugges change o e a size g adien 23,34. Fu he mo e, he LME es ing hyd omo phological and ca chmen a iables on pCO 2 day- o- nigh di e ences (Table 1D) did no e eal significan ela ion- ships wi h ei he o hese d i e s. This could ei he be due o he ac ha we missed he bes p oxy ha de e mines day- o-nigh di e ences in pCO 2 in Eu opean s eams o ha he e a e no common d i e s among he in es iga ed s eams. La ge diel a iabili y o CO 2 pa e ns wi hin one Swedish s eam26 o among US headwa e s eams25 ha e been desc ibed, which complica es he iden ifica ion o gene al d i e s. Hence, u he esea ch is needed o deciphe he diel a iabili y o he sou ces and dynamics o pCO 2 in s eams and o unde s and he en i onmen al, hyd omo phological, and ca chmen d i e s be o e hei impo ance on a egional o global scale can be assessed. In-s eam me abolism wi h pho osyn he ic CO 2 fixa ion diminishing pCO 2 du ing he day may explain he inc ease in CO 2 fluxes om day o nigh , bu canno explain why in some ins ances we measu ed a lowe CO 2 flux a nigh . Po en ial explana ions o a lowe nigh flux migh include: (i) highe a mosphe ic CO 2 concen a ions due o he absence o e es ial CO 2 fixa ion du ing nigh and he e o e a lowe wa e –a mosphe e pCO 2 g adien , (ii) pho omine aliza ion o Table 1 Resul s o he linea mixed-e ec models (LME). Response a iable Fixed e ec χ2(1) pSign (A) Tes ing spa ial and empo al hypo heses CO 2 flux di e ence om day o nigh La i ude 7.4207 0.006 − Wa e empe a u e (day) 0.0168 0.897 Wa e empe a u e (day)*la i ude 4.9594 0.026 + (B) Tes ing physical and biogeochemical d i e s o CO 2 flux changes CO 2 flux di e ence om day o nigh ΔWa e pCO 2 4.9497 0.026 + ΔGas ans e eloci y k 0.5613 0.454 (C) Tes ing biogeochemical d i e s o pCO 2 changes pCO 2 di e ence om day o nigh ΔWa e O 2 concen a ion 7.9879 0.005 − ΔpH 0.0345 0.853 ΔConduc i i y 0.0293 0.864 ΔT w −T a a(p oxy o hea flux) 1.6720 0.196 ΔWa e empe a u e 0.8731 0.350 (D) Tes ing ca chmen and hyd omo phological d i e s o pCO 2 changes pCO 2 di e ence om day o nigh Day leng h 1.7244 0.189 S eam we ed wid h 0.3748 0.540 Discha ge 3.4458 0.063 % o es 0.0950 0.758 Ca chmen a ea 2.3656 0.124 aHea flux calcula ed as wa e empe a u e (T w ) minus ai empe a u e (T a ). (A) Ma ginal R2=0.12, condi ional R2=0.18, sample size =107. (B) Ma ginal R2=0.08, condi ional R2=0.10, sample size =77. (C) Ma ginal R2=0.13, condi ional R2=0.33, sample size =78. (D) Ma ginal R2=0.11, condi ional R2=0.13, sample size =68. The e ec s o la i ude and wa e empe a u e du ing he day (A) and he e ec o day- o-nigh di e ences o pCO 2 and he gas ans e eloci y (Δ=nigh minus day alues) (B) on he day- o-nigh di e ence o CO 2 fluxes we e es ed. Fu he mo e, he e ec o day- o-nigh di e ences o physical and biogeochemical pa ame e s (C) and he e ec o ca chmen and hyd omo phological ela ed pa ame e s (D) on he day- o-nigh di e ences o pCO 2 we e e alua ed. S eam ID was included as a andom e ec on he in e cep . Significances o fixed e ec s we e assessed wi h likelihood a io es s wi h deg ees o eedom =1. The slope di ec ion (sign) o he e ec is indica ed wi h –o +when significan . Significan p alues < 0.05 a e in bold. Fig. 3 Diel changes in CO 2 fluxes (FCO 2 ) and o he physical and chemical pa ame e s o Oc obe /Janua y/Ap il and July, espec i ely. The physical and chemical pa ame e s comp ise a mosphe ic CO 2 (Ai CO 2 ), he di e ences o CO 2 concen a ions in he wa e minus he ai (CO 2 g adien ), he wa e –ai gas ans e eloci y (k), he di e ences o empe a u es in he wa e minus he ai (T w −T a ), he wa e empe a u e (WT), he oxygen concen a ion in he wa e (O 2 ), pH in he wa e , he pa ial p essu e o CO 2 in he wa e (pCO 2 ), and conduc i i y (Cond). The a ows indica e significan inc eases (↑) o significan dec eases (↓) om day o nigh and he line indica es no significan change (―) es ed by a Wilcoxon signed- ank es (see Supplemen a y Fig. S4 o mo e in o ma ion). The di e ences be ween he sampling pe iods Oc obe / Janua y/Ap il and July, espec i ely, de ec ed in his Eu opean s udy a e highligh ed in ed. ARTICLE COMMUNICATIONS EARTH & ENVIRONMENT | h ps://doi.o g/10.1038/s43247-021-00192-w 4COMMUNICATIONS EARTH & ENVIRONMENT | (2021) 2:118 | h ps://doi.o g/10.1038/s43247-021-00192-w | www.na u e.com/commsen o ganic ma e o CO 2 coun e ac ing he CO 2 fixa ion by p ima y p oduce s du ing day- ime, and (iii) lowe u bulence due o a dec ease in s eam discha ge in he nigh . We ound significan inc eases in a mosphe ic CO 2 close o he in es iga ed s eams a nigh . Howe e , his was usually accompanied by concomi an inc eases in wa e pCO 2 and he e o e did no ansla e in o smalle CO 2 g adien s be ween he wa e –ai in e ace (Fig. 3; Supplemen a y Fig. S4b, e, i). P oduc ion o CO 2 due o pho omine aliza ion o dissol ed o ganic ca bon (DOC) could play a ole in diel CO 2 dynamics in s eams wi h high amoun s o colo ed e es ial o ganic ma e 35. In he highly colo ed s eams, diel CO 2 pa e ns can addi ionally be influenced by DOC shading diminishing ben hic p ima y p oduc ion36. In Oc obe , we measu ed DOC concen a ions in a subse o he in es iga ed s eams o ano he s udy37 whe e an ag icul u al s eam in Sweden and pea land-domina ed s eams in G ea B i ain had high DOC concen a ions (>10 mg L−1) whe eas he median DOC was much lowe wi h 2.6 mg L−137. Due o he limi ed da a, we could no es he e ec o DOC on pCO 2 changes and we can nei he confi m no exclude ha pho omine aliza ion migh play a ole o diel pCO 2 and consequen ly CO 2 flux a iabili y in he s udied s eams. We did find, none heless, ha he majo i y o he s eams whe e CO 2 fluxes we e lowe du ing he nigh also had a lowe gas ans e eloci y (k 600 ), likely due o a sligh dec ease in s eam discha ge and he e o e u bulence. Thus, while he e was a gene al endency o inc eased pCO 2 om day o nigh (only 4 ou o 20 dec eases in CO 2 fluxes om day o nigh showed a concomi an dec ease in wa e pCO 2 ), indi idual s eams a single ime poin s seemed o expe ience diel fluc ua ions in discha ge as desc ibed elsewhe e38. This can simul aneously educe he gas exchange eloci y o he s eam and he e o e cause lowe nigh - ime CO 2 fluxes. In his s udy, we only measu ed s eam discha ge du ing he day, and he e o e he impo ance o his mechanism emains o be confi med. Maximum CO 2 flux di e ences migh be e en highe —lim- i a ions o he s udy design. Fo o ganiza ional easons, he sampling scheme o his collabo a i e s udy was s anda dized o fixed imes o measu emen s o he day and he nigh . All eams ac oss Eu ope s a ed hei measu emen s a 11:00 (midday) and 23:00 GMT (midnigh ) du ing each sampling pe iod, which has consequences o he magni ude o he obse ed diel a iabili y o he CO 2 fluxes. The la ges diel di e ences in s eam pCO 2 ha e gene ally been de ec ed a he end o he day compa ed o he end o he nigh 12,18,39. In an ag icul u al Swedish s eam, diel max- imum and minimum CO 2 concen a ions we e eached a 04:00 and 16:00 (GMT), espec i ely, du ing sp ing and ea ly summe pe iods (la e Ap il o ea ly July) whe e diel dynamics we e mos p onounced26. In hese scena ios, sampling midday and mid- nigh , as conduc ed in his s udy, would be close o hose maxima and minima as hey can be eached al eady ea lie du ing he day (see Supplemen a y Fig. S6 in May). Howe e , he maxima and minima o diel CO 2 dynamics in s eams can a y la gely (see Supplemen a y Fig. S6 in Oc obe , Ap il, July). In ano he example o Ge man s eams39, he imes o minima and maxima di e be ween s eams and imes, and he fixed ime poin s chosen in his s udy would miss he maximum di e ences ha can be obse ed (see Supplemen a y Fig. S7 in Augus ). Hence, ou es ima es could be conse a i e as we compa ed fixed ime poin s a midday and midnigh . In gene al, CO 2 flux measu e- men s in s eams a e highly sensi i e owa ds he ime o he day because diel minimum and maximum o pCO 2 can a y la gely om mon h o mon h bu also om day o day. As we ound ha he diel a iabili y o pCO 2 was he majo d i e o diel CO 2 fluxes, we ecommend u u e s udies ha plan o measu e CO 2 fluxes di ec ly wi h he chambe me hod, o addi ionally moni o he diel a iabili y o pCO 2 wi h logge s a a high empo al eso- lu ion. This app oach will p o ide he oppo uni y o es ima e i he measu emen s a e done du ing peak imes o no . While ou esul s p o ide a fi s insigh in o he d i e s o day- nigh di e ences in CO 2 fluxes, he high unce ain y in he models as well as he some imes opposing pa e ns—inc eases and dec eases om day o nigh in di e en s eams and sampling pe iods—poin owa ds di e en d i e s a ying on a empo al and spa ial scale. We ecommend ha u u e s udy designs inco po a e high- equency CO 2 da a oge he wi h biogeochemical a iables om he s eam (e.g., O 2 ) and he a mosphe e (e.g., CO 2 o empe a u e)40. Addi ionally, we ecommend including adioac i e o s able ca bon iso ope signa u es o ack po en ial sou ces o CO 2 and hei changes in s eams41,42 o be e assess e es ial–aqua ic linkages. Linking empo al pa e ns o flu ial CO 2 fluxes wi h hei d i e s ac oss la ge spa ial scales is a pa h owa ds a mo e accu a e unde s anding o hei ole in egional and global ca bon cycles. Ou esul s demons a e ha , in many s eams ac oss Eu ope, nigh - ime CO 2 fluxes exceed day- ime, esul ing in a po en ial unde es ima ion o global CO 2 emissions om inland wa e s i no conside ed. I is hus c i ical o accoun o he diel a iabili y o flu ial CO 2 fluxes o accu a e daily and annual es ima es o CO 2 emissions om inland wa e s. Me hods Sampling scheme. The p ojec included 16 eams dis ibu ed ac oss 11 Eu opean coun ies. E e y eam sampled one o h ee s eams (Supplemen a y Table S1) e e y 3 mon hs (Oc obe 2016/Janua y 2017/Ap il 2017/July 2017) wi hin a ime ame o 2 weeks h oughou a whole yea . These sampling pe iods oughly co e he seasons au umn/win e /sp ing/summe al hough, due o he la ge la i udinal co e age o he sampling si es, he seasons and hei cha ac e is ics a y la gely. In o al, 34 s eam si es (Supplemen a y Fig. S1) we e isi ed each sampling pe iod du ing he specified 2 weeks’ ime ame excep o 11 s eams in Janua y ha we e ozen du ing he sampling weeks (Supplemen a y Table S3). CO 2 fluxes we e measu ed once e e y sampling pe iod wi h d i ing flux chambe s equipped wi h CO 2 senso s. This me hod has p o en o be a eliable and leas biased di ec measu emen o CO 2 fluxes a he wa e –ai in e ace in s eams19,43.CO 2 concen a ions in he chambe headspace we e logged e e y 30 s o e a pe iod o 5–10 min du ing each un, and CO 2 fluxes we e calcula ed based on he a e o change o e ime in pCO 2 in he chambe headspace. A each s eam, we measu ed CO 2 fluxes wi h he flux chambe (fi e imes), pCO 2 in he a mosphe e and wa e wi h he CO 2 senso s in he flux chambe (de ails desc ibed in Supplemen a y Me hods), pH, empe a u e, conduc i i y, and oxygen in he wa e wi h a mul ip obe (Supplemen a y Table S2). These measu emen s we e s a ed a 11:00 and 23:00 (GMT) and las ed app oxima ely wo hou s and a e e e ed o as midday and midnigh h oughou his a icle. S eam wid h, dep h, canopy co e , and discha ge we e de e mined du ing he day (see Supplemen a y Sampling manual o de ails). In addi ion, he ollowing in o ma ion was collec ed o each s eam once du ing he s udy: s eam o de , clima e zone, ca chmen a ea un il he endpoin o he in es iga ed s eam si e and he pe cen age o co e age o di e en land use classes in his ca chmen a ea, and p edominan geology (Supplemen a y Table S1). Calcula ions o CO 2 fluxes and gas ans e eloci y. Flux a es we e ob ained om he linea slopes o he pCO 2 in he chambe headspace o e ime and flux was accep ed i he coe ficien o de e mina ion (R2) o he slope was a leas 0.6544. An excep ion was made in cases whe e he slope was close o ze o and he pCO 2 in he a mosphe e and wa e (measu ed a he same ime) we e a equilib ium. These fluxes we e se o ze o. Final flux a es F(mmol CO 2 m−2h−1) we e calcula ed acco ding o Eq. (1)45: F¼S103PV RTA 60 60;ð1Þ whe e Sis he slope (ppm s−1), Pis he pCO 2 in he a mosphe e (a m), Vis he olume (mL) o he d i ing chambe , Ris he gas cons an (82.0562 mL a m K−1 mol−1), Tis he chambe ai empe a u e (K), Ais he bo om a ea o he chambe (m2), and he las e m is he con e sion om seconds o hou s. In his s udy, we ollowed he sign con en ion whe eby posi i e alues indica e a CO 2 flux om he s eam o he a mosphe e (sou ce) and nega i e alues indica e a flux om he a mosphe e o he s eam (sink). The magni udes o a ia ions be ween day- ime and nigh - ime measu emen s a e addi ionally s a ed as pe cen inc eases, which COMMUNICATIONS EARTH & ENVIRONMENT | h ps://doi.o g/10.1038/s43247-021-00192-w ARTICLE COMMUNICATIONS EARTH & ENVIRONMENT | (2021) 2:118 | h ps://doi.o g/10.1038/s43247-021-00192-w | www.na u e.com/commsen 5 we e compu ed by di iding he di e ence be ween he alues a nigh minus day by he alue a day and exp essing he esul as a pe cen change om day o nigh . We used F(Eq.(1)) o calcula e he gas ans e eloci y (kin cm h−1)by in e ing he equa ion o Fick’s law o gas di usion, acco ding o Eq. (2): k¼F kHðCO2wa e CO2ai Þ100;ð2Þ whe e kH is Hen y’s cons an (in mol L−1a m−1) adjus ed o empe a u e46. Fo compa ison o ans e eloci ies be ween si es and sampling pe iods and wi h he li e a u e, k(Eq. ( 2)) was s anda dized o k 600 (Eq. (3)): k600 ¼k600 Sc  0:5 ð3Þ whe e kis he ans e eloci y a in si u empe a u e (T), Sc is he Schmid numbe o in si u empe a u e T, he Schmid numbe o 20 °C in eshwa e is 600, and ep esen ing a hyd odynamic ough wa e su ace ypical in s eams he exponen o −0.5 was chosen47. S a is ical analyses. All s a is ical analyses we e pe o med wi h median alues o h ee o fi e floa ing chambe uns pe day and nigh , espec i ely, using he s a is ical p og amming language R48 ( e sion 3.5.1). Samplings ha gene a ed less han h ee alues o ei he day o nigh due o an R2o he slope <0.6544 we e excluded om u he analysis educing he numbe om 136 o 107 day–nigh compa isons. Fo ou s a is ical es s, he alpha le el was se o α=0.05. Significan di e ences be ween day- ime and nigh - ime measu emen s o each sampling pe iod ac oss all s eams we e es ed wi h Wilcoxon signed- ank es s49 whe e median day- ime and nigh - ime alues o each s eam si e we e pai ed (Fig. 1a). The same es s we e conduc ed o he o he biogeochemical a iables measu ed a midday and midnigh (see Fig. 3; Supplemen a y Fig. S4). Wi h a fi s linea mixed-e ec model (LME) we es ed he la i udinal and wa e empe a u e e ec on CO 2 flux di e ences om day o nigh . A second LME was buil o e alua e he wo majo d i e s o CO 2 flux di e ences om day o nigh : pCO 2 and gas exchange eloci y (k). A hi d LME was subsequen ly used o de e mine he biochemical ac o s po en ially influencing he di e ences o he nigh - ime minus day- ime pCO 2 , which was iden ified as he only significan d i e in he second LME. Finally, a ou h LME was buil o e alua e he e ec o ca chmen and hyd omo phological pa ame e s on he day- o-nigh di e ences o pCO 2 . Fo hese es s, we used he “lme ” unc ion o he R-package “lme4”50 wi h maximum-likelihood es ima ion. Fixed e ec s o he LME wi h biogeochemical pa ame e s o pCO 2 di e ences om day o nigh included absolu e di e ences om day o nigh o oxygen concen a ion in he wa e , pH, conduc i i y, empe a u e g adien o a mosphe e and wa e , and wa e empe a u e. Fixed e ec s o he LME wi h ca chmen and hyd omo phological pa ame e s included day leng h (i.e., sun hou s om sun ise o sunse ), s eam we ed wid h, discha ge, % o es o he ca chmen , and ca chmen a ea. These a iables a e mos ly emo ely a ailable o s eams. Fo he LMEs we included s eam ID as a andom e ec allowing di e en in e cep s o each s eam o accoun o pseudo eplica ion (one da a poin pe sampling pe iod pe s eam) and z-scaled all fixed e ec s wi h he “scale” unc ion be o e unning he models. S a is ical significances o fixed e ec s we e assessed wi h likelihood a io es s using he unc ion “d op1”51. The espec i e LMEs we e ollowed by a model alida ion, checking he esiduals o no mal dis ibu ion and homogenei y o a iances52. A sepa a ion o he da ase o check i d i e s be ween inc eases om day o nigh and dec eases om day o nigh di e did no e eal accep able models in e ms o model alida ion (i.e., esiduals we e no no mally dis ibu ed). Al hough ou da ase p o ided a la ge spa ial co e age on day–nigh di e ences in CO 2 fluxes in Eu opean s eams, i did no ha e he s a is ical powe o es o significan d i e s sepa a ely o inc eases and dec eases. Da a a ailabili y The da a ha suppo he findings o his s udy a e openly a ailable in figsha e a h ps:// doi.o g/10.6084/m9.figsha e.12717188. Code a ailabili y This manusc ip includes no code. Recei ed: 13 No embe 2020; Accep ed: 21 May 2021; Re e ences 1. Bu man, D. E. e al. Aqua ic ca bon cycling in he con e minous Uni ed S a es and implica ions o e es ial ca bon accoun ing. P oc. Na l. Acad. Sci. USA 113,58–63 (2016). 2. D ake, T. W., Raymond, P. A. & Spence , R. G. M. Te es ial ca bon inpu s o inland wa e s: a cu en syn hesis o es ima es and unce ain y. Limnol. Oceanog . Le 3, 132–142 (2018). 3. Raymond, P. A. e al. Global ca bon dioxide emissions om inland wa e s. Na u e 503, 355–359 (2013). 4. MacIn y e,S.,Wanninkho ,R.&Chan on,J.P.T acegasexchangein eshwa e and coas al ma ine sys ems: flux ac oss he ai wa e in e ace. In Me hods in Ecology: Biogenic T ace Gases: Measu ing Emissions om Soil and Wa e (eds Ma son, P. & Ha iss, R.) 52–97 (Blackwell Publishing, 1995). 5. Du e , C., Bu man, D. E., Ma x, A., Ribolzi, O. & Hu ley, L. B. CO 2 e asion along s eams d i en by g oundwa e inpu s and geomo phic con ols. Na . Geosci. 11, 813–818 (2018). 6. Roche ‐Ros, G., Sponselle , R. A., Lidbe g, W., Mö h, C. & Giesle , R. Landscape p ocess domains d i e pa e ns o CO 2 e asion om i e ne wo ks. Limnol. Oceanog . Le . 4,87–95 (2019). 7. Hall, R. O. & Ulse h, A. J. Gas exchange in s eams and i e s. WIREs Wa e e1391 (2019). h ps://doi.o g/10.1002/wa 2.1391 8. Hope, D., Palme , S. M., Bille , M. F. & Dawson, J. J. C. Va ia ions in dissol ed CO 2 and CH 4 in a fi s -o de s eam and ca chmen : an in es iga ion o soil–s eam linkages. Hyd ol. P ocess. 18, 3255–3275 (2004). 9. Ho gby, Å., Gómez-Gene , L., Esco fie , N. & Ba in, T. J. Dynamics and po en ial d i e s o CO 2 concen a ion and e asion ac oss empo al scales in high-alpine s eams. En i on. Res. Le . 14, 124082 (2019). 10. Guasch, H., A mengol, J., Ma í, E. & Saba e , S. Diu nal a ia ion in dissol ed oxygen and ca bon dioxide in wo low-o de s eams. Wa e Res. 32, 1067–1074 (1998). 11. Lynch, J. K., Bea y, C. M., Seidel, M. P., Jungs , L. J. & DeG andp e, M. D. Con ols o i e ine CO 2 o e an annual cycle de e mined using di ec , high empo al esolu ion pCO 2 measu emen s. J. Geophys. Res. 115, G03016 (2010). 12. Pe e , H. e al. Scales and d i e s o empo al pCO 2 dynamics in an Alpine s eam. J. Geophys. Res. Biogeosci. 119, 1078–1091 (2014). 13. Roche -Ros, G., Sponselle , R. A., Be gs öm, A.-K., My s ene , M. & Giesle , R. S eam me abolism con ols diel pa e ns and e asion o CO 2 in A c ic s eams. Glob. Chang. Biol 00,1–14 (2019). 14. Koehle , B., Landelius, T., Weyhenmeye , G. A., Machida, N. & T an ik, L. J. Sunligh -induced ca bon dioxide emissions om inland wa e s. Glob. Biogeochem. Cycles 28, 696–711 (2014). 15. Golub, M., Desai, A. R., McKinley, G. A., Remucal, C. K. & S anley, E. H. La ge unce ain y in es ima ing pCO 2 om ca bona e equilib ia in lakes. J. Geophys. Res. Biogeosci. 122, 2909–2924 (2017). 16. Raymond, P. A. e al. Scaling he gas ans e eloci y and hyd aulic geome y in s eams and small i e s. Limnol. Oceanog . Fluids En i on. 2,41–53 (2012). 17. Schelke , J., Singe , G. A., Ulse h, A. J., Hengsbe ge , S. & Ba in, T. J. CO 2 e asion om a s eep, high g adien s eam ne wo k: impo ance o seasonal and diu nal a ia ion in aqua ic pCO 2 and gas ans e . Limnol. Oceanog . 61, 1826–1838 (2016). 18. Reiman, J. H. & Xu, Y. J. Diel a iabili y o pCO 2 and CO 2 ou gassing om he lowe Mississippi Ri e : implica ions o i e ine CO 2 ou gassing es ima ion. Wa e 11, 43 (2019). 19. Bas iken, D., Sundg en, I., Na chimu hu, S., Reyie , H. & Gål alk, M. Technical No e: cos -e ficien app oaches o measu e ca bon dioxide (CO 2 ) fluxes and concen a ions in e es ial and aqua ic en i onmen s using mini logge s. Biogeosciences 12, 3849–3859 (2015). 20. Looman, A., Mahe , D. T., Pendall, E., Bass, A. & San os, I. R. The ca bon dioxide e asion cycle o an in e mi en fi s -o de s eam: con as ing wa e –ai and soil–ai exchange. Biogeochemis y 132,87–102 (2017). 21. C aw o d, J. T. e al. CO 2 and CH 4 emission om s eams: pa e ns, con ols, and egional significance. Glob. Biogeochem. Cycles 28, 197–210 (2014). 22. Teodo u, C. R., Del Gio gio, P. A., P ai ie, Y. T. & Cami e, M. Pa e ns in pCO 2 in bo eal s eams and i e s o no he n Quebec, Canada. Glob. Biogeochem. Cycles 23, GB2012 (2009). 23. Ho chkiss, E. R. e al. Sou ces o and p ocesses con olling CO 2 emissions change wi h he size o s eams and i e s. Na . Geosci. 8, 696–699 (2015). 24. Gómez-Gene , L. e al. Global ca bon dioxide e flux om i e s enhanced by high noc u nal emissions. Na . Geosci.1–6 (2021). h ps://doi.o g/10.1038/ s41561-021-00722-3 25. C aw o d, J. T., S anley, E. H., Do nblase , M. M. & S iegl, R. G. CO 2 ime se ies pa e ns in con as ing headwa e s eams o No h Ame ica. Aqua . Sci. 79, 473–486 (2016). 26. Wallin, M. B., Aude , J., Peacock, M., Sahlée, E. & Win e dahl, M. Ca bon dioxide dynamics in an ag icul u al headwa e s eam d i en by hyd ology and p ima y p oduc ion. Biogeosciences 17, 2487–2498 (2020). 27. Dema s, B. O. L. e al. Impac o wa ming on CO 2 emissions om s eams coun e ed by aqua ic pho osyn hesis. Na . Geosci. 9, 758–761 (2016). ARTICLE COMMUNICATIONS EARTH & ENVIRONMENT | h ps://doi.o g/10.1038/s43247-021-00192-w 6COMMUNICATIONS EARTH & ENVIRONMENT | (2021) 2:118 | h ps://doi.o g/10.1038/s43247-021-00192-w | www.na u e.com/commsen 28. Robe s, B. J., Mulholland, P. J. & Hill, W. R. Mul iple scales o empo al a iabili y in ecosys em me abolism a es: esul s om 2 yea s o con inuous moni o ing in a o es ed headwa e s eam. Ecosys ems 10, 588–606 (2007). 29. Ulse h, A. J., Be uzzo, E., Singe , G. A., Schelke , J. & Ba in, T. J. Clima e- induced changes in sp ing snowmel impac ecosys em me abolism and ca bon fluxes in an Alpine S eam Ne wo k. Ecosys ems 21, 373–390 (2018). 30. Co y, R. M., Wa d, C. P., C ump, B. C. & Kling, G. W. Sunligh con ols wa e column p ocessing o ca bon in a c ic esh wa e s. Science (80-.) 345, 925–928 (2014). 31. Tobias, C. R., Böhlke, J. K., Ha ey, J. W. & Busenbe g, E. A simple echnique o con inuous measu emen o ime‐ a iable gas ans e in su ace wa e s. Limnol. Oceanog . Me hods 7, 185–195 (2009). 32. Be g, P. & Pace, M. L. Con inuous measu emen o ai –wa e gas exchange by unde wa e eddy co a iance. Biogeosciences 14, 5595–5606 (2017). 33. Riml, J., Campeau, A., Bishop, K. & Wallin, M. B. Spec al decomposi ion e eals new pe spec i es on CO 2 concen a ion pa e ns and soil-s eam linkages. J. Geophys. Res. Biogeosci. 124, 3039–3056 (2019). 34. Liu, S. & Raymond, P. A. Hyd ologic con ols on pCO 2 and CO 2 e flux in US s eams and i e s. Limnol. Oceanog . Le 3, 428–435 (2018). 35. Lindell, M. J., G anéli, H. W. & Be ilsson, S. Seasonal pho o eac i i y o dissol ed o ganic ma e om lakes wi h con as ing humic con en . Can. J. Fish. Aqua . Sci. 57, 875–885 (2000). 36. Ask, J., Ka lsson, J., Pe sson, L. & Ask, P. Te es ial o ganic ma e and ligh pene a ion: e ec s on bac e ial and p ima y p oduc ion in lakes. Limnol. Oceanog . 54, 2034–2040 (2009). 37. B a o, A. G. e al. The in e play be ween o al me cu y, me hylme cu y and dissol ed o ganic ma e in flu ial sys ems: a la i udinal s udy ac oss Eu ope. Wa e Res.144, 172–182 (2018). 38. Schwab, M., Klaus, J., Pfis e , L. & Weile , M. Diel discha ge cycles explained h ough iscosi y fluc ua ions in ipa ian inflow. Wa e Resou . Res. 52, 8744–8755 (2016). 39. Bodme , P., Heinz, M., Pusch, M., Singe , G. & P emke, K. Ca bon dynamics and hei link o dissol ed o ganic ma e quali y ac oss con as ing s eam ecosys ems. Sci. To al En i on. 553, 574–586 (2016). 40. Vachon, D. e al. Pai ed O2–CO 2 measu emen s p o ide eme gen insigh s in o aqua ic ecosys em unc ion. Limnol. Oceanog . Le .5, 287–294 (2019). 41. Campeau, A. e al. S able ca bon iso opes e eal soil-s eam DIC linkages in con as ing headwa e ca chmen s. J. Geophys. Res. Biogeosci. 123, 149–167 (2018). 42. Campeau, A. e al. Cu en o es ca bon fixa ion uels s eam CO 2 emissions. Na . Commun. 10,1–9 (2019). 43. Lo ke, A. e al. Technical no e: d i ing e sus ancho ed flux chambe s o measu ing g eenhouse gas emissions om unning wa e s. Biogeosciences 12, 7013–7024 (2015). 44. T emblay, A., Va al y, L., Ga neau, M. & Roehm, C. G eenhouse Gas Emissions-Fluxes and P ocesses: Hyd oelec ic Rese oi s and Na u al En i onmen s (Sp inge Science & Business Media, 2005). 45. Duc, N. T. e al. Au oma ed flux chambe o in es iga ing gas flux a wa e –ai in e aces. En i on. Sci. Technol. 47, 968–975 (2013). 46. Golden um, J. A. GHG Measu emen Guidelines o F eshwa e Rese oi s: De i ed F om: he UNESCO/IHA G eenhouse Gas Emissions om F eshwa e Rese oi s Resea ch P ojec (In e na ional Hyd opowe Associa ion (IHA), 2010). 47. Jähne, B. e al. On he pa ame e s influencing ai –wa e gas exchange. J. Geophys. Res. Ocean. 92, 1937–1949 (1987). 48. R Co e Team. R: A Language and En i onmen o S a is ical Compu ing (R Founda ion o S a is ical Compu ing, 2018). 49. Wilcoxon, F. Indi idual compa isons by anking me hods. Biome . Bull. 1, 80–83 (1945). 50. Ba es, D., Maechle , M., Bolke , B. & Walke , S. Fi ing linea mixed-e ec s models using lme4. J. S a . So w. 67,1–48 (2015). 51. Zuu , A., Ieno, E. N., Walke , N., Sa elie , A. A. & Smi h, G. M. Mixed E ec s Models and Ex ensions in Ecology wi h R (Sp inge Science & Business Media, 2009). 52. Zuu , A. F. & Ieno, E. N. A p o ocol o conduc ing and p esen ing esul s o eg ession‐ ype analyses. Me hods Ecol. E ol. 7, 636–645 (2016). Acknowledgemen s We hank he ini ia o s o he fi s Collabo a i e Eu opean F eshwa e Science P ojec o Young Resea che s, he Eu opean Fede a ion o F eshwa e Sciences (EFFS) boa d, he Eu opean F esh and Young Resea che s (EFYR) and he ep esen a i es o he F esh Blood o F esh Wa e (FBFW) mee ings. We also hank he se en na ional eshwa e socie ies financing his p ojec , namely he Ibe ian Associa ion o Limnology (AIL; Spain and Po ugal), Deu sche Gesellscha ü Limnologie e.V. (DGL; Ge many), Swiss Socie y o Hyd ology and Limnology (SGHL; Swi ze land), I alian Associa ion o Oceanog aphy and Limnology (I aly), F eshwa e Biological Associa ion (FBA; Uni ed Kingdom), F ench Limnological Associa ion (AFL; F ance), Aus ian Limnological Socie y (SIL- Aus ia), as well as he Leibniz-Ins i u e o F eshwa e Ecology and Inland Fishe ies o addi ional unds. Addi ional unding was awa ded o J.P.C.-R. by a Juan de la Cie a pos doc o al g an om he Spanish Go e nmen (FJC2018-037791-I), o A.P.P. by a Ph. D. g an om he Fundação pa a a Ciência e Tecnologia (SFRH/BD/115030/2016), o B.C.D. by he Ma ine Ins i u e’s Cullen Ph.D. ellowship (G an No. CF/15/05), o N.C. by he Eu opean Union’s Ho izon 2020 esea ch and inno a ion p og am unde he Ma ie Sklodowska-Cu ie g an ag eemen (No. 839709), o J.M. by FCT (Po uguese Science Founda ion) h ough a Ph.D. g an (SFRH/BD/131924/2017), o J.P. by he DSI/ NRF Resea ch Chai in Inland Fishe ies and F eshwa e Ecology, o A.F. by he Juan de la Cie a pos doc o al g an om he Spanish Go e nmen (FJCI-2017–33171), and o C.M.-L. by he F ench Na ional Agency o Wa e and Aqua ic En i onmen s (ONEMA, Ac ion 13, “Colma age, échanges nappe- i iè e e p ocessus biogéochimiques”). We acknowledge Luigi Naselli-Flo es and An onio Camacho o hei encou agemen and suppo du ing he p ojec . We also hank Da id Bas iken, Ing id Sundg en, and Thanh Duc Nguyen o he in oduc ion o he logge and chambe cons uc ion and ad ice o measu emen s o CO 2 fluxes wi h he chambe , Vincen Fugè e o his help in se ing up he linea mixed-e ec models, and Vik o Rosenbe g o c ea ing he map. Fu he mo e, we a e e y hank ul o Ra ael Ma cé and Paul del Gio gio o hei hough ul commen s on he manusc ip and we hank h ee anonymous e iewe s o cons uc i e inpu s ha imp o ed he manusc ip . Open access unding p o ided by Uni e si y o Vienna. Au ho con ibu ions K.A. and P.B. concei ed he s udy design, coo dina ed he p ojec and con ibu ed equally o his wo k; all au ho s collec ed and analyzed he field da a and K.A. and P.B. ga he ed and pe o med he quali y check o all da a; K.A., P.B., and J.P.C.-R. co-w o e he pape wi h he help o M.K., G.H.N., and N.C. All au ho s (K.A., J.P.C.-R., T.F., A.P., S.C.-F., D.S., A.C.N., Al.D., A.P.P., B.C.D., N.S., C.G.R., G.H.N., X.T., V.E., L.B.-F., T.B., J.A., An.D., G.B., S.F., N.C., E.d.E., F.P., J.-R.M., J.M., D.F., C.N., M.C., M.N., L.L., C.R. G.-Q., F.R., N.P., J.L.J.L., J.P., M.K., A.F., S.H.O., C.M.-L., A.B., J.A.F., P.J.G., L.A.K., M.R., P.B.) commen ed on he manusc ip . Compe ing in e es s The au ho s decla e no compe ing in e es s. Addi ional in o ma ion Supplemen a y in o ma ion The online e sion con ains supplemen a y ma e ial a ailable a h ps://doi.o g/10.1038/s43247-021-00192-w. Co espondence and eques s o ma e ials should be add essed o K.A. Pee e iew in o ma ion Communica ions Ea h & En i onmen hanks he anonymous e iewe s o hei con ibu ion o he pee e iew o his wo k. P ima y Handling Edi o s: Joshua Dean and Joe Aslin. Rep in s and pe mission in o ma ion is a ailable a h p://www.na u e.com/ ep in s Publishe ’s no e Sp inge Na u e emains neu al wi h ega d o ju isdic ional claims in published maps and ins i u ional a filia ions. Open Access This a icle is licensed unde a C ea i e Commons A ibu ion 4.0 In e na ional License, which pe mi s use, sha ing, adap a ion, dis ibu ion and ep oduc ion in any medium o o ma , as long as you gi e app op ia e c edi o he o iginal au ho (s) and he sou ce, p o ide a link o he C ea i e Commons license, and indica e i changes we e made. The images o o he hi d pa y ma e ial in his a icle a e included in he a icle’s C ea i e Commons license, unless indica ed o he wise in a c edi line o he ma e ial. I ma e ial is no included in he a icle’s C ea i e Commons license and you in ended use is no pe mi ed by s a u o y egula ion o exceeds he pe mi ed use, you will need o ob ain pe mission di ec ly om he copy igh holde . To iew a copy o his license, isi h p://c ea i ecommons.o g/ licenses/by/4.0/. © The Au ho (s) 2021 COMMUNICATIONS EARTH & ENVIRONMENT | h ps://doi.o g/10.1038/s43247-021-00192-w ARTICLE COMMUNICATIONS EARTH & ENVIRONMENT | (2021) 2:118 | h ps://doi.o g/10.1038/s43247-021-00192-w | www.na u e.com/commsen 7 1 Limnology/Depa men o Ecology and Gene ics, Uppsala Uni e si y, Uppsala, Sweden. 2 Wasse Clus e Lunz –Biologische S a ion, Lunz am See, Aus ia. 3 Depa men o Func ional and E olu iona y Ecology, Uni e si y o Vienna, Vienna, Aus ia. 4 Ca alan Ins i u e o Wa e Resea ch (ICRA), Gi ona, Spain. 5 Uni e si y o Gi ona (UdG), Gi ona, Spain. 6 Flu ial Ecosys em Ecology, Depa men o Ecology, Uni e si y o Innsb uck, Innsb uck, Aus ia. 7 INRAE, UR Ri e ly, Cen e de Lyon-Villeu banne, Villeu banne, Cedex, F ance. 8 Depa men o Li e and En i onmen al Sciences, Bou nemou h Uni e si y, Poole, UK. 9 Depa men o Sciences and Technological Inno a ion, Uni e si y o Piemon e O ien ale, Alessand ia, I aly. 10 ALPSTREAM –Alpine S eam Resea ch Cen e , Os ana, I aly. 11 Resea ch Cen e in Biodi e si y and Gene ic Resou ces (CIBIO-InBIO), Uni e si y o Po o, Vila do Conde, Po ugal. 12 Facul y o Sciences, Uni e si y o Po o, Po o, Po ugal. 13 Cen e o F eshwa e and En i onmen al S udies, Dundalk Ins i u e o Technology, Dundalk, Co Lou h, I eland. 14 Na ional Museum o Na u al His o y, Bulga ian Academy o Sciences, Sofia, Bulga ia. 15 Ri e and Conse a ion Resea ch, Depa men o Ecology, Uni e si y o Innsb uck, Innsb uck, Aus ia. 16 Depa men o Aqua ic Ecosys ems, Ins i u e o Biodi e si y and Ecosys em Resea ch, Bulga ian Academy o Sciences, Sofia, Bulga ia. 17 Depa men o Aqua ic Sciences and Assessmen , Swedish Uni e si y o Ag icul u al Sciences, Uppsala, Sweden. 18 Depa men o Ecology and En i onmen al Science, Umeå Uni e si y, Umeå, Sweden. 19 Depa men o Li e Sciences and Sys ems Biology, Uni e si y o Tu in, Tu in, I aly. 20 Ma ine Ins i u e, Fu nace, Newpo , Co Mayo, I eland. 21 Depa men o E olu iona y Biology, Ecology and En i onmen al Sciences, Facul y o Biology, Uni e si y o Ba celona (UB), Ba celona, Spain. 22 Cen e o Ecology, E olu ion and En i onmen al Changes (cE3c), Faculdade de Ciências, Uni e sidade de Lisboa, Lisboa, Po ugal. 23 Ins i u e o En i onmen al Sciences, Uni e si y o Koblenz-Landau, Landau, Ge many. 24 Ins i u e o Mic obiology, Uni e si y o Innsb uck, Innsb uck, Aus ia. 25 Expe imen al Limnology, Leibniz-Ins i u e o F eshwa e Ecology and Inland Fishe ies (IGB), S echlin, Ge many. 26 Ecohyd ology, Leibniz-Ins i u e o F eshwa e Ecology and Inland Fishe ies (IGB), Be lin, Ge many. 27 Cen e o Ad anced S udies o Blanes, Spanish Na ional Resea ch Council, Blanes, Spain. 28 Ins i u e o Geog aphy and Geoecology, Ka ls uhe Ins i u e o Technology, Ka ls uhe, Ge many. 29 Sou h A ican Ins i u e o Aqua ic Biodi e si y, Makhanda, Sou h A ica. 30 Depa men o Ecology and En i onmen al Sciences, Palacký Uni e si y Olomouc, Olomouc, Czech Republic. 31 En i onmen al Resea ch Ins i u e, Uni e si y o Highlands and Islands (UHI), Thu so, Sco land, UK. 32 Depa men o Gene al and Applied Hyd obiology, Sofia Uni e si y “S . Klimen Oh idski”,Sofia, Bulga ia. 33 Chemical Analy ics and Biogeochemis y, Leibniz-Ins i u e o F eshwa e Ecology and Inland Fishe ies, Be lin, Ge many. 34 P esen add ess: Depa men o Biology, Aa hus Uni e si y, Aa hus C, Denma k. 35 P esen add ess: Ins i u e o Global Heal h, Facul y o Medicine, Uni e si y o Gene a, Campus Bio ech, Gene a, Swi ze land. 36 P esen add ess: Cen e o En i onmen , Fishe ies and Aquacul u e Science (Ce as), Lowes o , Su olk, UK. 37 P esen add ess: Depa men o Bioscience, Aa hus Uni e si y, Silkebo g, Denma k. 38 P esen add ess: No wegian Ins i u e o Wa e Resea ch, Oslo, No way. 39 P esen add ess: Labo a oi e des Sciences du Clima e de l’En i onnemen (LSCE), CEA, CNRS, UVSQ, Gi -Su -Y e e, F ance. 40 P esen add ess: Uni ed S a es Geological Su ey, Boulde , CO, USA. 41 P esen add ess: En i onmen al A chaeology Lab, Depa men o His o ical, Philosophical and Religious s udies, Umeå Uni e si y, Umeå, Sweden. 42 P esen add ess: Depa men o Fo es Ecology and Managemen , Swedish Uni e si y o Ag icul u al Sciences, Umeå, Sweden. 43 P esen add ess: Global Change Resea ch Ins i u e o he Czech Academy o Sciences, B no, Czech Republic. 44 P esen add ess: G oupe de Reche che In e uni e si ai e en Limnologie, Dépa emen des Sciences Biologiques, Uni e si é du Québec à Mon éal, Mon éal, Canada. ✉email: [email p o ec ed] ARTICLE COMMUNICATIONS EARTH & ENVIRONMENT | h ps://doi.o g/10.1038/s43247-021-00192-w 8COMMUNICATIONS EARTH & ENVIRONMENT | (2021) 2:118 | h ps://doi.o g/10.1038/s43247-021-00192-w | www.na u e.com/commsen