Changes in the partial pressure of carbon dioxide in the Mauritanian-Cap Vert upwelling region between 2005 and 2012

Biogeosciences, 14, 3859–3871, 2017 h ps://doi.o g/10.5194/bg-14-3859-2017 © Au ho (s) 2017. This wo k is dis ibu ed unde he C ea i e Commons A ibu ion 3.0 License. Changes in he pa ial p essu e o ca bon dioxide in he Mau i anian–Cap Ve upwelling egion be ween 2005 and 2012 Melcho González-Dá ila1, J. Magdalena San ana Casiano1, and F ancisco Machín1,2 1Ins i u o de Oceanog a ía y Cambio Global, G upo QUIMA, Uni e sidad de Las Palmas de G an Cana ia, 35017, Las Palmas de G an Cana ia, Spain 2Depa amen o de Física, Uni e sidad de Las Palmas de G an Cana ia, 35017, Las Palmas de G an Cana ia, Spain Co espondence o: Melcho González-Dá ila (melcho [email p o ec ed]) Recei ed: 9 Ma ch 2017 – Discussion s a ed: 29 Ma ch 2017 Re ised: 26 July 2017 – Accep ed: 28 July 2017 – Published: 31 Augus 2017 Abs ac . Coas al upwellings along he eas e n ma gins o majo ocean basins ep esen egions o la ge ecological and economic impo ance due o he high biological p o- duc i i y. The ole o hese egions o he global ca bon cycle makes hem essen ial in add essing clima e change. The physical o cing o upwelling p ocesses ha a o p o- duc ion in hese a eas a e al eady being a ec ed by global wa ming, which will modi y he in ensi y o upwelling and, consequen ly, he ca bon dioxide cycle. He e, we p esen mon hly high- esolu ion su ace expe imen al da a o em- pe a u e and pa ial p essu e o ca bon dioxide in one o he ou mos impo an upwelling egions o he plane , he Mau i anian–Cap Ve upwelling egion, om 2005 o 2012. This da a se p o ides di ec e idence o seasonal and in e - annual changes in he physical and biochemical p ocesses. Speci ically, we show an upwelling in ensi ica ion and an inc ease o 0.6 Tg y −1in CO2ou gassing due o inc eased wind speed, despi e inc eased p ima y p oduc i i y. This in- c ease in CO2ou gassing oge he wi h he obse ed de- c ease in sea su ace empe a u e a he loca ion o he Mau- i anian Cap Blanc, 21◦N, p oduced a pH a e dec ease o −0.003 ±0.001 y −1. 1 In oduc ion The excess o CO2in he a mosphe e, la gely esponsible o global clima e change, has p omp ed esea ch on he ole o he oceans in he ca bon cycle. The aim in ecen decades has been o assess how he oceans ac as sou ces o sinks wi hin he ca bon cycle. To achie e his goal, highly esol ed spa ial and empo al obse a ions ep esen a i e o he dis ibu ion o CO2 luxes be ween he ocean and a mosphe e a e nec- essa y. Au oma ed ins umen s on olun ee obse ing ships (VOSs) se e o p o ide as many obse a ions h oughou he global ocean as possible. This is in addi ion o da a collec ed on scien i ic c uises and a long- e m moo ings (i.e., As o e al., 2005: Lüge e al., 2004, 2006; González-Dá ila e al., 2005, 2009; Schus e e al., 2009; Ullman e al., 2009; Wa - son e al., 2009; Padin e al., 2010; G ube e al., 2002; Do e e al., 2003; San ana-Casiano e al., 2007; Ba es e al., 2014). Wi h he amoun o da a al eady ga he ed (h p://www. soca .in o/; P eil e al., 2013), clima ologies ha p esen a - e age CO2 luxes be ween he a mosphe e and he ocean ha e been de eloped, iden i ying a eas ac ing as a sou ce o sink (Key e al., 2004; Takahashi e al., 2009). Howe e , he low spa ial esolu ion o hese da abases limi s he applicabili y, especially in coas al a eas. Upwelling egions a e pa icu- la ly unde - ep esen ed in such la ge da abases. Upwelling p esen s a dynamic p ocess ha aises nu ien and CO2- ich wa e om ela i ely deep a eas o he su ace. The nu i- en s eaching he pho ic zone p omo e p ima y p oduc ion, which consumes CO2. This p ocess gene a es a CO2 lux in o he ocean. On he o he hand, upwelling also b ings up CO2 om deep seawa e , which gene a es unce ain y abou he ac ual ole o upwelling a eas as a sou ce o sink o CO2 (Michaels e al., 2001). Indeed, upwelling a eas may ac as a sou ce o sink o CO2depending on hei loca ion (Cai e al., 2006; Chen e al., 2013), whe e upwelling egions a low la i- udes mainly ac as a sou ce o CO2(Feely e al., 2002; As o e al., 2005; F iede ich e al., 2008; San ana-Casiano e al., 2009; González-Dá ila e al., 2009) and hose a midla i udes Published by Cope nicus Publica ions on behal o he Eu opean Geosciences Union. 3860 M. González-Dá ila e al.: Changes in he pa ial p essu e o ca bon dioxide mainly ac as a sink o CO2(F ankignoulle and Bo ges, 2001; Hales e al., 2005; Bo ges and F ankignoulle, 2002; Bo ges e al., 2005; San ana-Casiano e al., 2009; González-Dá ila e al., 2009). Se e al an h opogenic in e ac i e e ec s s ongly in luence eas e n bounda y upwelling sys ems (EBUSs), in- cluding uppe ocean wa ming, ocean acidi ica ion, and ocean deoxygena ion (G ube , 2011; Feely e al., 2008; Keeling e al., 2010). Mo eo e , e idence o inc easing wind speed ha would a o upwelling (Bakun, 1990; Dema cq, 2009; Oe de e al., 2015) suppo s he possibili y o a change in he dynamics o hese highly p oduc i e a eas. Recen ly, eddy- esol ing egional ocean models ha e shown how upwelling in ensi ica ion can cause a majo impac on he sys em’s bi- ological p oduc i i y and CO2ou gassing (Lachka and G u- be , 2013; Oe de e al., 2015). Wind obse a ions and e- analysis p oduc s a e con o e sial ega ding he Bakun in- ensi ica ion hypo hesis (Bakun, 1990). Using di e en wind da abases o he Cana y egion, Ba on e al. (2013) con- cluded ha he e was no e idence o a gene al inc ease in he upwelling in ensi y o no hwes A ica. Ma cello e al. (2011) ound an in ensi ica ion o he upwelling sys em in he same a ea du ing a 20-yea pe iod, while he alongsho e wind s ess emained almos s able. C oppe e al. (2014) ound ha coas al summe wind speed inc eased, esul ing in an inc ease in upwelling- a o able wind speeds no h o 20◦N and an inc ease in downwelling- a o able winds sou h o 20◦N. San os e al. (2005, 2012) showed ha sea su ace empe a u e (SST) was no homogeneous ei he along la i- ude o longi ude and depended on he upwelling index (UI) in ensi y. Va ela e al. (2015) demons a ed opposi e esul s wo ldwide depending on he leng h o da a, season e alua ed, and selec ed a ea wi hin he same wind da a se o be ween da a se s. Fo he Mau i anian egion, when wind s ess da a we e used (Va ela e al., 2015), a mo e pe sis en inc easing end in upwelling- a o able winds no h o 21◦N and a de- c easing end sou h o 19◦N was de e mined. S a ing in June 2005, he QUIMA-VOS line isi ed he Mau i anian–Cap Ve upwelling egion no hwes o A ica on a mon hly basis (Fig. 1 and Table S1 in he Supplemen ) p oducing o he i s ime a high- esolu ion da abase o SST and pa ial p essu e o CO2exp essed as ugaci y CO2. This da abase shows he a ia ions in he CO2sys em unde changes in he upwelling condi ions in he Cana y ecosys em om 27 o 10◦N o he pe iod 2005 o 2012. Mo e da a o he egion om o he su eys exis (h p://www.soca .in o/; P eil e al., 2013) bu hey we e no conside ed in his s udy as hey do no ollow he same ack as he QUIMA-VOS line. Those da a a e s ongly in luenced by he dis ance o he upwelling cells wi h he co esponding physical e ec s in he pa ial p essu e o CO2. Figu e 1. Ship ack (black line) in he a ea om 28◦N (G an Ca- na ia, he Cana y Islands) o 10◦N. The loca ions o Cap Blanc and Cap Ve a e indica ed. Mon hly OceanColo Web (h ps:// oceancolo .gs c.nasa.go /) da a o a e age chlo ophyll aconcen- a ion (mgm−3)we e included in a MATLAB ou ine and annu- ally a e aged. The map has been gene a ed using MATLAB 7.12 R2011a. 2 Expe imen al 2.1 S udy egion The VOS line c osses he eas A lan ic Ocean om he no h o Eu ope (English Channel) o Sou h A ica, calling a G an Cana ia, he Cana y Islands, wi h a pe iodici y o 2 mon hs, which p o ides mon hly da a (sou hwa d o no hwa d sec- ions). In his wo k, he a ea be ween G an Cana ia a 27 and 10◦N has been selec ed in o de o s udy he Mau i anian– Cap Ve upwelling egion. On i s ou e sou h (Fig. 1), he ship lea es G an Cana ia and goes s aigh o 100 km o Cap Blanc a 21◦N, 17◦450W. I hen ollows his longi ude, pass- ing a 100km o Cap Ve un il 12◦N, whe e i changes di- ec ion o Cape Town, eaching 10◦N, 17◦W a 330km o he coas o Guinea. Be ween 22 and 20◦N, he ship eaches he 500m isoba h. Sou h o 15◦N, he ship mo es be ween he 1000 and 500 m isoba hs. On i s ou e no h, he ship ol- lows he e e se ack. 2.2 Expe imen al da a Expe imen al da a we e ob ained unde he EU p ojec s CARBOOCEAN and CARBOCHANGE (www.Ca boOcean.o g and h ps://ca bochange.b.uib.no/) and now also a ailable a h p://www.soca .in o/ (P eil e al., Biogeosciences, 14, 3859–3871, 2017 www.biogeosciences.ne /14/3859/2017/ M. González-Dá ila e al.: Changes in he pa ial p essu e o ca bon dioxide 3861 2013). An au onomous ins umen o he de e mina ion o he pa ial p essu e o CO2de eloped by C aig Neill ollow- ing NOAA ecommenda ions was ins alled on a VOS line. This was ope a ed by he Medi e anean Shipping Company S.A. om 2005 o 2008 and Mae sk om 2010 o 2012. This VOS line (QUIMA-VOS) an be ween he UK and Cape Town om July 2005 o Janua y 2013 (Table S1 in he Supplemen ). Tempe a u e was measu ed a h ee posi ions along he sampling ci cui : in he in ake (Sea-Bi d SBE38L), in he equilib a o (Sea-Bi d he mosalinog aph SBE21 and in e nal PT100 he mome e ), and in he oxygen senso (Op ode 3835, Aande aa™). A e he seawa e pump, he in ake is di ided in o wo lines, one eeding he CO2sys em and he o he eeding he oxygen senso , he luo ome e , and he Sea-Bi d he mosalinome e . Di e ences be ween equilib a o and in ake empe a u es we e cons an in ime due o he high seawa e low bu a ied among ships due o he di e en loca ions o he equipmen . Values a ied be ween 0.06 ◦C when he equipmen was placed close o he in ake and 0.35◦C when he equipmen was one loo abo e and inside he engine oom. The SST was also ob ained om he NOAA_OI_SST-V2 da a p o ided by he NOAA/OAR/ESRL PSD om Boulde , Colo ado, USA (h p://www.es l.noaa.go /psd). These da a had a spa ial es- olu ion o 1◦la i ude and 1◦longi ude and mon hly a e ages we e used. The co ela ion be ween ou expe imen al SST da a and sa elli e da a was be e han ±1◦C, and imp o ed o ±0.4 ◦C a e emo ing he mos a ec ed upwelling egions (19–22 and 14–16◦N), which ela ed o he high a iabili y imposed by he upwelling. The CO2mola ac ion, xCO2, was ob ained e e y 150s in seawa e , while a mosphe ic xCO2da a we e ob ained e - e y 180min. The seawa e in ake was loca ed a a 10m dep h. The sys em was calib a ed e e y 3 hou s by measu ing ou di e en s anda d gases wi h mixing a ios in he anges o 0.0, 250–290, 380–410, and 490–530 ppm o CO2in he ai , p o ided by NOAA and aceable o he Wo ld Me eo ologi- cal O ganiza ion scale. The p ecision o he sys em is g ea e han 0.5µa m and he accu acy es ima ed wi h espec o he s anda d gases is o 1µa m inside he s anda ds’ ange. Fo xCO2 alues highe han he highes s anda d (532.04ppm), he accu acy will be educed, e en when linea i y was ob- se ed in all cases inside he s anda ds ange. The ugac- i y o CO2( CO2, µa m) was calcula ed om xCO2a e co ec ing o empe a u e di e ences be ween in ake and equilib a o , acco ding o he exp essions o seawa e gi en by DOE (1994). No malized CO2(N CO2) de i ed om he mean SST o he a ea (Tmean)was compu ed ollowing Takahashi e al. (1993) as (N CO2)= CO2·exp[0.0423(Tmean −SST)].(1) In o de o compu e a second ca bona e sys em a iable, he su ace o al alkalini y (AT) was compu ed om sea su ace salini y (SSS) and SST (Lee e al., 2006). pHTa he in si u empe a u e was compu ed om CO2and ATand wi h a - e age annual su ace ocean o al phospha e and o al silica e concen a ions o 0.5 and 4.8 µmolkg−1, espec i ely, om he Wo ld Ocean A las 2009, using he ca bonic acid acidi y cons an s by Meh bach e al. (1973) e i ed by Dickson and Mille o (1987). Ai –sea CO2 luxes (FCO2, mmol m−2d−1) we e e alu- a ed as FCO2=0.24 ·k·s·( COsw 2− COa m 2), (2) whe e 0.24 is he scale ac o , kis he gas ans e eloc- i y, sis he CO2solubili y, COsw 2is he seawa e ugaci y o CO2, and COa m 2is he a mosphe ic ugaci y o CO2. In o de o e alua e ( COsw 2− COa m 2), COa m 2da a we e lin- ea ly in e pola ed o he COsw 2 ime ec o . A posi i e alue o FCO2co esponds o CO2ou gassing om he ocean. k (cm h−1)was e alua ed wi h he ollowing pa ame e iza ion (Nigh ingale e al., 2000): k=(0.222 ·W2+0.333 ·w) ·(Sc/660)−1/2,(3) whe e Wis he wind speed a 10m abo e he sea su ace (m s−1)and Sc is he Schmid numbe . The a iables in ol ed in es ima ing FCO2da a (i.e., COsw 2, COa m 2, SST, and SSS) we e i ed o sinusoidal ex- p essions (Lüge e al., 2004) o a gi en la i ude as ollows: X(la )∗=a0+a1( −2005)+a2sin(2π )+a3cos(2π ) +a4sin(4π )+a5cos(4π ),(4) whe e aia e he i ing coe icien s, is he sampling ime exp essed as yea ac ion, and X(la )∗ ep esen s any o he ou i ed a iables. This p ocedu e allowed us o e- cons uc he se ies o expe imen al da a o pe iods wi h- ou mon hly da a. The a iables we e decomposed in o an in e annual e m X(la )∗ =a0+a1( −2005)plus a pe iodi- cal e m X(la )∗ p=a2sin(2π )+a3cos(2π )+a4sin(4π )+ a5cos(4π ), ha is, X(la )∗=X(la )∗ +X(la )∗ p. The pe iod- ical e m accoun s o he high- equency seasonal a iabil- i y, while he in e annual e m ma ks he yea - o-yea end. Fi s , obse a ions we e g ouped in a na u al yea o a gi en la i ude, as i hey had been aken in a single yea (no co - ec ion was done o in e annual a iabili y). The mean sea- sonal clima ology da a associa ed wi h he pe iodic coe i- cien s (i.e., a2,a3,a4, and a5) h oughou he sampling pe- iod we e de e mined. Nex , he in e annual coe icien a1 was calcula ed by i ing he esiduals esul ing om sub- ac ing he pe iodical componen , X(la )∗ p, om he o iginal a iable X(la ). By ixing hese i e coe icien s (a1–a5), new dis ibu ions o COsw∗ 2, COa m∗ 2, SST∗, and SSS∗we e cons uc ed wi h a daily esolu ion based on he cu e i s gi en o each a iable as in Eq. (4), p o iding he coe i- cien a0. The accu acy o his i ing p ocedu e was checked by bo h compu ing he co ela ion be ween expe imen al and econs uc ed alues and by de e mining he mean esid- uals. The Pea son coe icien s we e always o e 0.87 o www.biogeosciences.ne /14/3859/2017/ Biogeosciences, 14, 3859–3871, 2017 3862 M. González-Dá ila e al.: Changes in he pa ial p essu e o ca bon dioxide SST (a e age 0.94 ±0.03), o e 0.69 o bo h COsw 2and COa m 2(a e age o 0.79 ±0.07 and 0.82 ±0.04, espec- i ely), and o e 0.67 o SSS (a e age 0.79 ±0.07). The mean esidual on he de e mina ion o hose ou a i- ables we e ±3.7 µa m, ±1.5 µa m, ±0.22 ◦C, and ±0.05 o COsw∗ 2, COa m∗ 2, SST∗, and SSS∗, espec i ely. When he mon hly sa elli e SST alues we e conside ed, he new SST∗ unc ion a e aged o each mon h p oduced alues wi hin ±0.47 ◦C, con i ming ha his p ocedu e was able o i non- sampled pe iods. I was assumed ha he same p ocedu e was alid o non-sampled CO2.Finally, daily FCO∗ 2 ime se- ies be ween 10 and 27◦N wi h a la i udinal esolu ion o 0.5◦we e calcula ed wi h a s anda d e o o es ima ion o 0.5 mmol m−2d−1(15 % o e o ) ha p oduced mean esid- uals (expe imen al FCO2–FCO∗ 2)o 0.4 mmol m−2d−1and Pea son co ela ion coe icien s be ween expe imen al and compu ed FCO∗ 2o > 0.6, p < 0.01. Chlo ophyll awas calcula ed om measu emen s made by he Mode a e- esolu ion Imaging Spec o adiome e (MODIS) aboa d NASA’s Aqua sa elli e. We used mon hly a e ages wi h a spa ial esolu ion o 9 km supplied by Ocean- Colo Web (h ps://oceancolo .gs c.nasa.go ). Wind da a we e downloaded om he NCEP CFSR da abase a h p:// da.uca .edu/pub/c s .h ml de eloped by NOAA and e ie ed om he NOAA Na ional Ope a ional Model A chi e and Dis ibu ion Sys em and main ained by he NOAA Na ional Clima ic Da a Cen e . The spa ial eso- lu ion is app oxima ely 0.3×0.3◦and he empo al esolu ion is 6 h. The e e ence heigh o he wind da a is 10m. Rain all da a we e collec ed by he p ecipi a ion ada in- s alled on he T opical Rain all Measu ing Mission (TRMM) sa elli e (h p://p ecip.gs c.nasa.go ). Mon hly a e ages wi h a spa ial esolu ion o 0.5 ×0.5◦(p oduc 3A12, e sion 07) we e used (Fig. S1 in he Supplemen ) in o de o explain changes in seasonal su ace salini y dis ibu ions. 3 Resul s and discussion 3.1 Physical p ope ies The a iabili y o he Mau i anian–Cap Ve upwelling was analyzed in e ms o he upwelling index (Nykjae and Van Camp, 1994) (Fig. 2) using sa elli e wind da a. Nega i e UI alues co espond o upwelling- a o able condi ions and posi i e alues o downwelling- a o able condi ions. The lowes nega i e alues o he index co espond o mo e in- ense upwelling. Resul s clea ly dis inguish wo main suba - eas in he upwelling sys em: (1) no h o 20◦N, he upwelling condi ions we e a o able h oughou he yea , al hough he highes upwellings we e obse ed om Ma ch o Sep em- be wi h a no hwa d shi om 20 o 22◦N. (2) Sou h o 20◦N, a ma ked seasonali y was obse ed wi h a o able up- welling condi ions du ing au umn and win e , wi h he maxi- mum in ensi y obse ed du ing Janua y and Feb ua y. In his Figu e 2. Time se ies o upwelling index (UI, ×10−3m2s−1)in he Mau i anian–Cap Ve upwelling egion along he ship ack compu ed ollowing Nykjae and Van Camp (1994). Blue colo s a e ela ed o upwelling e en s and ed colo s o downwelling e en s. egion, a downwelling egime is p esen be ween May and No embe when he summe ade winds a e eplaced by he monsoonal winds ad ec ing wa m wa e (Fig. 3a) no hwa d along he sho e (Nykjae and Van Camp, 1994). Ou esul s (Fig. 2) a e qui e consis en wi h p e ious esea ch (Nykjae and Van Camp, 1994; Ma cello e al., 2011; San os e al., 2005, 2012; C oppe e al., 2014) bu include he yea s 2010 o 2012, when he UI a a ound 20–21◦N p esen ed a shi o he upwelling in ensi y om high (−2000 m2s−1) o s ong (−2800 m2s−1). The analysis o upwelling ends along his ou e has been con o e sial since i is highly dependen on he selec ed egion (San os e al., 2012). The in e annual e o- lu ion o he UI o e he pe iod 2005 o 2012 (Fig. 4, g een line) o each deg ee in la i ude indica es an inc ease in he UI (mean con idence in e al o 9m2s−1)as showed by San- os e al. (2012). No h o 15◦N, he upwelling index con i med he s onge upwelling obse ed since 1995–1996 in his egion a e mo e han a 10-yea ( om a leas 1982 o 1995) pe iod o weake upwelling (San os e al., 2012). Local zonal di e - ences be ween ocean and coas al SST ends de e mined wi h sa elli e da a con i med he in ensi ica ion o he upwelling egime along he A ican coas o he pe iod 1982 o 2000 (San os e al., 2005) and ex ended by San os e al. (2012) un- il 2010 and u he ex ended in his s udy un il 2012 (da a no shown). This has been desc ibed as a decadal-scale shi o he upwelling egime in ensi y (Ma cello e al., 2011; San- os e al., 2012). Sou h o 15◦N, he annual UI alues and ends (Figs. 2 and 4) bo h o he upwelling ( alues close o −2800 m2s−1 in Janua y) and downwelling ( alues eaching 1850 m2s−1 in July) pe iods a e becoming s onge . A 11–12◦N, whe e downwelling is becoming s onge , his esul s in nega i e annual empe a u e a es ha app oach ze o. The UI se es as an indica ion o decadal a iabili y o he summe mon- soon winds and associa ed no hwa d ad ec ion o wa m wa- e along he coas (San os e al., 2012). The highes upwelling in ensi y along he VOS line was loca ed a he capes, Cap Blanc and Cap Ve . F om sa el- li e chlo ophyll ada a, especially o Cap Blanc, gian ila- Biogeosciences, 14, 3859–3871, 2017 www.biogeosciences.ne /14/3859/2017/ M. González-Dá ila e al.: Changes in he pa ial p essu e o ca bon dioxide 3863 10 15 20 25 18 20 22 24 26 28 30 La i ude Tempe a u e ( C) (a) W 10 15 20 25 32 33 34 35 36 37 La i ude Salini y W (b) 10 15 20 25 18 20 22 24 26 28 30 La i ude Tempe a u e ( C) Sp 10 15 20 25 32 33 34 35 36 37 La i ude Salini y Sp 10 15 20 25 18 20 22 24 26 28 30 La i ude Tempe a u e ( C) Sm 10 15 20 25 32 33 34 35 36 37 La i ude Salini y Sm 10 15 20 25 18 20 22 24 26 28 30 La i ude Tempe a u e ( C) Au 10 15 20 25 32 33 34 35 36 37 La i ude Salini y Au Figu e 3. In si u da a o column (a) SST and column (b) SSS in he Mau i anian–Cap Ve coas al egion g ouped by seasons: win e (W; Decembe , Janua y, and Feb ua y), sp ing (Sp; Ma ch, Ap il, and May), summe (Sm; June, July, and Augus ), and au umn (Au; Sep embe , Oc obe , and No embe ). The a e aged alues o all c uises in Table S1 a e shown in black o each season including he 95% con idence limi s. The colo code o each c uise is indica ed in Table S1. men s wi h chlo ophyll concen a ions abo e 1mg m−3pe - sis yea - ound, sp eading om he coas o se e al hun- d ed kilome e s o sho e (Fig. 1). No h o Cap Blanc he upwelled wa e o igina es om he No h A lan ic Cen- al Wa e , and mixes wi h Sou h A lan ic Cen al Wa e (SACW) owa ds he sou h (Mi els aed , 1983). Sou h o Cap Blanc, he upwelling o nu ien - ich SACW (Mi el- s aed , 1983) p omo es phy oplank on g ow h be ween Cap Blanc and Cap Ve . Towa ds 12◦N, upwelling is also ed by he No h Equa o ial Unde cu en (Hagen and Schemainda, 1984). Mo eo e , he en i e no hwes A ican coas is also in luenced by he A ican dese dus anspo by he mid- oposphe ic Ha ma an winds o igina ing om he cen al Saha a, which supplemen s he le els o mic onu ien s (such as i on) o he adjacen ma ine ecosys em (Mi els aed , 1983; Neue e al., 2004). The s udy a ea is also a ec ed by he mig a ion o he In- e opical Con e gence Zone (ITCZ), ela ed o maximum p ecipi a ion a es (Has en a h, 1995). To ha e a signi ican sa elli e p ecipi a ion eco d in ou egion o in e es , p e- cipi a ion da a we e in eg a ed longi udinally be ween 25.25 and 9.75◦W. Time se ies o he la i udinal dis ibu ion o in- eg a ed p ecipi a ion (Fig. S1 in he Supplemen ) iden i ied he a e age posi ion o he ITCZ ela ed o maximum p ecip- i a ion a es. The ITCZ was loca ed a i s sou he nmos posi- ion (2◦N) du ing win e , eaching i s no he nmos posi ion www.biogeosciences.ne /14/3859/2017/ Biogeosciences, 14, 3859–3871, 2017 3864 M. González-Dá ila e al.: Changes in he pa ial p essu e o ca bon dioxide −0.2 −0.1 0 0.1 0.2 0.3 10 12 14 16 18 20 22 24 26 28 Change pe yea La i ude (°N) (a) T ends in T T ends in S T ends in UI 0 1 2 3 4 5 10 12 14 16 18 20 22 24 26 28 Change pe yea (b) T ends in CO2 sw T ends in CO2 a m Figu e 4. La i udinal dis ibu ion o he in e annual ends o he upwelling index (UI) and o he ou expe imen al a iables along he QUIMA-VOS line in eg a ed o e e e y deg ee be ween 2005 and 2012. Panel (a) p esen s he ends o upwelling index (UI, ×10−3m2s−1, mean con idence in e al o 9m2s−1), SST (◦C y −1, con idence in e al 0.13◦C), and SSS (y −1, con idence in e al 0.06) and (b) he ends o COsw 2and COa m 2(con i- dence in e als 4.23 and 0.44µa m). (14–16◦N) a ound summe . The ITCZ eached ou a ea o in e es (>10◦N) om la e sp ing o la e summe . The la i udinal dis ibu ions o measu ed SST and SSS along he essel ack a e shown in Fig. 3, g ouped by sea- sons (labeled W, Sp, Sm, and Au). The empe a u e gene - ally dec eased om 10 ◦N o abou 20–21◦N, whe e he ship mee s he Mau i anian upwelling. F om he e o he no h, he empe a u e ises as he ship lea es he upwelling a ea on i s way o he Cana y Islands. In si u empe a u e a 27◦N shows empe a u es in he ange o 18 o 24 ◦C wi h he minimum in win e and maximum in la e summe o ea ly au umn. The annual empe a u e ange was somewha highe a 20◦N, wi h a summe maximum o a ound 26 ◦C and minimum in sp ing o abou 17 ◦C. A 10◦N, empe a u es we e he high- es h oughou he yea (>25 ◦C), wi h minimum alues in win e and maximum in la e sp ing and la e au umn. The low alues obse ed du ing he end o summe a e ela ed o he a i al o he ITCZ (Fig. S1 in he Supplemen ) a hose la i- udes. The he mal dis ibu ion shows a empe a u e inc ease as we mo e o he Equa o and a no able cooling a he up- welled wa e s o Mau i ania. The upwelling o cold wa e om he Cap Ve a ea was only de ec ed du ing win e ime and he beginning o sp ing. Salini y minimum alues we e no mally loca ed a 10◦N, inc easing o maximum alues a he Cana ies’ la i ude. The minimum alues o salini y we e excep ionally low du ing au umn om 10 o 16◦N by bo h he eshwa e inpu om i e s ha inc ease hei ou low du ing his season (Nicholson, 1981) and by he no hwa d shi o he ITCZ du ing his ime o he yea . Anomaly ields o empe a u e and salini y (da a no shown) we e calcula ed as he di e ence be ween he obse - a ions and he mean alues a each season o indi idual la - i udes. Fo empe a u e, he la ges anomalies in win e and sp ing we e loca ed sou h o 18◦N, wi h alues o ±2◦C, ela ed o he seasonal cycle o he Cap Ve upwelling. Du - ing summe he pa e n changed and he la ges anomalies we e de ec ed in he upwelling a ea a 18–22◦N, wi h alues o ±5◦C when he upwelling index o he Mau i anian a ea was highes (Fig. 2). In au umn he empe a u e anomalies we e shi ed sligh ly o he no h, 20–24◦N, wi h alues o ±3◦C ela ed o he obse ed pulses in upwelling- a o able winds ha a ec ed he su ace seawa e p ope ies. On he o he hand, salini y anomalies showed a e y homogeneous pa e n in all la i udes o win e , sp ing, and summe , wi h alues gene ally wi hin ±0.5. Howe e , du ing au umn im- po an anomalies sou h o 18◦N we e obse ed, wi h alues in he ange o ±1.5. In his egion, he upwelling de elop- men , he i e discha ge, and he ainy season con olled he obse ed dis ibu ion (Yoo and Ca on, 1990). To conclude, he da a show a pe manen annual upwelling egime obse ed no h o 20◦N and a seasonal egime ac oss 10–19◦N, in acco dance wi h he clima ology o p e ious s udies. The da a also con i m an inc ease in upwelling con- di ions no h o 20◦N and an inc ease in downwelling con- di ions sou h o 20◦N. 3.2 Ca bon dioxide a iabili y The la i udinal dis ibu ion o he seasonal COsw 2da a (Fig. 5a) showed he highes alues be ween 18 and 23◦N o all seasons due o he a iabili y imposed by he up- welling o Mau i ania. COsw 2was consis en ly g ea e han he COa m 2. Du ing win e , when he Cap Ve upwelling de- elops (Fig. 2), he 12–15◦N egion also p esen ed highe COsw 2 alues han hose in he a mosphe e. COsw 2da a showed a la i udinal shi be ween he seasons ollowing he shi obse ed in he upwelling index: in win e , he la ges alues we e loca ed be ween 19 and 24◦N; in sp ing, hey we e loca ed be ween 16 and 22◦N; and du ing summe and au umn, he la ges COsw 2 alues we e eco ded in he ange 20 o 23◦N. The di e ence be ween COsw 2no malized o he mean SST o 22◦C o he egion (N COsw 2)and COsw 2 (1 CO2=N COsw 2− COsw 2, Fig. 5b) ein o ced he a i- abili y a 20–23◦N all yea a ound and a 12–17◦N du ing win e and sp ing, indica ing ha upwelling is he majo ac- o con ibu ing o he CO2 a iabili y. Acco ding o Takahashi e al. (1993), COsw 2inc eases wi h empe a u e a a a e o 4.3% µa m ◦C−1(be ween 15 and 26 µa m ◦C−1in his a ea) in a he modynamically con- olled sys em. A 27◦N, as SST inc eases, he a e was only 7.45 µa m ◦C−1due mainly o biological up ake and also o CO2ou lux. A 20◦N he a e became nega i e wi h a alue Biogeosciences, 14, 3859–3871, 2017 www.biogeosciences.ne /14/3859/2017/ M. González-Dá ila e al.: Changes in he pa ial p essu e o ca bon dioxide 3865 10 15 20 25 200 300 400 500 600 700 800 La i ude CO2 ( a m) (a) W 10 15 20 25 −150 −100 −50 0 50 100 150 La i ude CO2 ( a m) W (b) 10 15 20 25 200 300 400 500 600 700 800 La i ude CO2 ( a m) Sp 10 15 20 25 −150 −100 −50 0 50 100 150 La i ude CO2 ( a m) Sp 10 15 20 25 200 300 400 500 600 700 800 La i ude CO2 ( a m) Sm 10 15 20 25 −150 −100 −50 0 50 100 150 La i ude CO2 ( a m) Sm 10 15 20 25 200 300 400 500 600 700 800 La i ude CO2 ( a m) Au 10 15 20 25 −150 −100 −50 0 50 100 150 La i ude CO2 ( a m) Au Figu e 5. Fugaci y o CO2da a in he Mau i anian–Cap Ve coas al egion g ouped by seasons: win e (W; Decembe , Janua y, and Feb u- a y), sp ing (Sp; Ma ch, Ap il, and May), summe (Sm; June, July, and Augus ), and au umn (Au; Sep embe , Oc obe , and No embe ). Column (a) COsw 2la i udinal dis ibu ion. Column (b), di e ence be ween measu ed and COsw 2 alues no malized o a cons an empe - a u e o 22 ◦C. The a e aged alues o all c uises in Table S1 a e shown in black o each season including he 95% con idence limi s. The colo code o each c uise is indica ed in Table S1. o −10.9 µa m ◦C−1, clea ly indica ing he impo an injec- ion o cool and CO2- ich seawa e a he upwelling a ea. The injec ion is no being compensa ed o by he solubili y no by he biological ca bon pumps. A 10◦N, he a e was s ill nega i e bu only −4.3 µa m ◦C−1as a esul o he seasonal upwelling. N COsw 2was ela ed wi h SST (da a no shown) in o de o accoun o e ec s no emo ed du ing no mal- iza ion. A la i udes 19 o 21◦N, in he upwelling icini y o Cap Blanc, an in e se ela ionship o 70–100µa m ◦C−1 was ound du ing win e and sp ing, while in summe and au umn he in e se ela ionship a e was educed o 12– 18 µa m ◦C−1. While he upwelling indexes a hose la i udes we e qui e cons an h oughou he yea , di e en a es ob- se ed should be ela ed o biological consump ion o he CO2excess. Howe e , du ing win e and sp ing he injec ion o CO2in he upwelling is no dec eased by he biological ac i i y in he a ea. Bu du ing he Chlo ophyll amaximum (la e sp ing and summe ), mos o he CO2was consumed and/o expo ed and, he e o e, he a e was s ongly educed. www.biogeosciences.ne /14/3859/2017/ Biogeosciences, 14, 3859–3871, 2017 3866 M. González-Dá ila e al.: Changes in he pa ial p essu e o ca bon dioxide Figu e 4 depic s he obse ed in e annual ends (a1coe i- cien in Eq. 4) o he ou expe imen ally eco ded de ended pa ame e s, oge he wi h he UI end. Con idence in e als o he compu ed mean annual alues o SST, SSS, COa m 2, and COsw 2we e 0.13 ◦C, 0.06, 0.44, and 4.23 µa m, e- spec i ely. The e was a clea SST end whe eby seawa e along he VOS line ack was ge ing coole wi h maximum cooling a es a he loca ion o Cap Blanc (21◦N) and Cap Ve upwellings (15◦N) wi h a es highe han −0.2 ◦C y −1. Da a om he i s 3 yea s (2005 o 2008) a 21◦N showed lowe empe a u es wi h highe cooling a es ha eached −0.7 ◦C y −1, al hough 3 yea s o da a a e no ep esen a i e. The a ea c ossed by he VOS line along 17◦450W om 22 o 10◦N is loca ed inside he 1000m isoba h ha is well inside he mean on al ac i i y in he Cana y egion, abou 200km wide (Wang e al., 2015). The di e en changes in empe - a u e in he coas al slope and o sho e wa e s a e ela ed o he di e en o igins o he wa e s upwelled om dep hs o abou 100m o he su ace (Mi els aed , 1983) ha sp ead o he coas al a ea. The o sho e wa e SST is less a iable owing o longe esidence ime in he ocean su ace. These e ec s and he ac ha he VOS line keeps a ack line ha c ossed he upwelling cells a a dis ance o he coas ha a ies among cells con ibu e o he obse ed spa ial a i- abili y. The e was no a emp o compa e la i udinal and lon- gi udinal e ec s on he obse ed alues. Ou expe imen al da a, howe e , do no show any posi i e SST a es in he up- welling a ec ed a ea, and only when he ship app oached he Cana y Islands did he ends become less nega i e, eaching a alue o +0.02 ◦C y −1a 27◦N, simila o hose ob ained o oceanic A lan ic wa e (Ba es e al., 2014). COa m 2 o he a ea showed he in e annual inc ease o abou 2 ±0.3 µa m y −1obse ed in a mosphe ic s a- ions, while COsw 2p esen ed a he e ogeneous dis ibu ion. Sou h o 18◦N, he a e o inc ease was always highe han ha in he a mosphe e eaching a maximum alue o 4.1 ±0.4 µa m y −1a 10◦N. A 27◦N, COsw 2inc eased a a a e o 1.7 ±0.2 µa m y −1simila o ha de e mined a he ESTOC ime se ies si e (González-Dá ila e al., 2010) lo- ca ed a 29◦100N, 15◦300W. In he Cap Blanc a ea, COsw 2 inc eased a an a e age a e o 2.5±0.4 µa m y −1wi h he highes alues in he pe iod 2005 o 2008 (a a e o 4.6 ±0.5 µa m y −1was compu ed wi h only hose yea s). A ound Cap Blanc, COsw 2always p esen ed lowe a es o inc ease han in he a mosphe e wi h alues well be- low 1 µa m y −1. The obse ed dec ease in SST and he ends in COsw 2can only be explained by a ein o ced up- welling. No h o 18◦N, he lowes a e o inc ease in COsw 2 compa ed o COa m 2, oge he wi h a dec ease in empe a- u e, indica ed ha upwelling is also a o ing an inc ease in he ne communi y p oduc ion a ound he Mau i anian up- welling, consuming and/o expo ing he CO2- ich upwelled wa e s a o ed by he la e al anspo o he Mau i anian cu - en (Lachka and G ube , 2013; Va ela e al., 2015). The Figu e 6. pH o su ace wa e s in o al p o on scale and a in si u SST compu ed om o al alkalini y (based on egional co ela ions wi h SST and SSS; Lee e al., 2006) and CO2a 21 ±0.25◦N. The e o ba s ep esen he s anda d de ia ion o he compu ed da a o each c uise o he selec ed la i ude. The black cu e shows he ha monic i ing o Eq. (4) o he da a and he co esponding linea end is also shown. upwelling in ensi ica ion e ec s obse ed in he ends o ou expe imen al da a suppo he ecen wind s ess ends (C oppe e al., 2014; Va ela e al., 2015; San os e al., 2012) o inc eased upwelling- a o able winds, a leas o he pe- iod 2005–2012 in he Cana y upwelling egion (Figs. 2 and 4). The in ensi ica ion o he upwelling esul s in a change in he measu ed upwelled wa e p ope ies due o ei he highe upwelling eloci ies o deepe sou ce upwelled wa e s. How- e e , wha emains unclea om hese eco ds is o wha ex- en hose changes e lec upwelling a ia ions due o clima e change o cing e sus na u al decadal a iabili y in he up- welling a eas occu ing o e in e annual imescales. Because he upwelling in ensi y is changing, o he a i- ables will also be a ec ed. pHT ,is a 21 ±0.25◦N was com- pu ed om CO2and alkalini y pai s o da a. Alkalin- i y was compu ed om egional co ela ions wi h SST and SSS (Lee e al., 2006), which could unde ep esen sea- sonal and in e annual a ia ions in upwelling a eas. How- e e , pH compu ed om CO2 alues a e ela i ely insen- si i e o e o s in AT, and CO2con ols he magni ude and a iabili y o pH (a 60µmol kg−1change in ATwill a ec a 0.1% in pH, ha is, abou 0.01 pH uni s). Fig- u e 6 depic s he compu ed pHT ,is(AT, CO2)da a and he ha monic i ing o Eq. (4) p o iding he seasonal a iabil- i y and in e annual end. Conside ing he small sys ema ic biases in in e annual dynamics, we de e mined a dec ease in pH a a a e o −0.003 ±0.001 y −1(Fig. 6). This de- c ease is one o he highes a e alues de e mined in se - e al ime se ies s a ions (Ba es e al., 2014), whe e oceanic SST has only sligh ly inc eased in he las decades. How- e e , a he Mau i anian upwelling a ea and a he loca- ion whe e ou VOS line app oached his egion, SST de- c eased a a a e o −0.22 ±0.06 ◦C y −1(Fig. 4). Solely Biogeosciences, 14, 3859–3871, 2017 www.biogeosciences.ne /14/3859/2017/ M. González-Dá ila e al.: Changes in he pa ial p essu e o ca bon dioxide 3867 Figu e 7. La i udinal dis ibu ions o seasonal and annual CO2 luxes (FCO2, mol m−2). Fluxes o CO2we e compu ed using Nigh ingale e al. (2000) pa ame e iza ion and sa elli e winds wi h a esolu ion o 6h. (a) In eg a ed yea o yea om 2005 o 2012 and (b) la i udinally in eg a ed o 2005 o 2012 oge he wi h annual alues o he No h A lan ic Oscilla ion (NAO) index. La i udinal dis ibu ions o FCO2 seasonally in eg a ed om 2005 o 2012 a e depic ed o (c) win e (Decembe , Janua y, and Feb ua y), (d) sp ing (Ma ch, Ap il, and May), and (e) summe (June, July, and Augus ). his dec ease in empe a u e would inc ease he pH by a a e o +0.004 y −1and he CO2would dec ease by 4 µa m y −1. The ne e ec o he inc ease in he amoun o ich CO2and lowe pH upwelled wa e s in he Mau i anian upwelling would be, he e o e, a dec ease in he pH a e o o e −0.007 ±0.002 uni s y −1and an inc ease in CO2o +6.5 ±0.7 µa m y −1(wi h pe iods whe e hose a es could each alues o −0.015 y −1in pH and +10.5 µa m y −1in CO2as eco ded du ing 2005–2008). Those alues a e g ea ly compensa ed o by he impo an dec ease in he SST esul ing in he de e mined a es o −0.003 ±0.001 pH uni s and +2.5 ±0.4 µa m o CO2pe yea . This new da a se o expe imen al alues con i med a de- c ease in SST and ends in COsw 2 ha can only be ex- plained by ein o ced upwelling condi ions ha a o an in- c ease in he ne communi y p oduc ion a ound he Mau i a- nian upwelling oge he wi h a mo e co osi e en i onmen wi h pH a es ha change by mo e han −0.007 ±0.002 y −1 a 21◦N. Howe e , he dec ease in SST in he upwelling cell bu e s his pH a e o alues a ound −0.003 ±0.001 y −1 and +2.5 ±0.4 µa m y −1in CO2, s ill among he highes obse ed in o he ime se ies. www.biogeosciences.ne /14/3859/2017/ Biogeosciences, 14, 3859–3871, 2017