Changes in phytoplankton production and upwelling intensity off A Coruña (NW Spain) for the last 28 yr.

1 23 Ocean Dynamics Theo e ical, Compu a ional and Obse a ional Oceanog aphy ISSN 1616-7341 Volume 69 Numbe 7 Ocean Dynamics (2019) 69:861-873 DOI 10.1007/s10236-019-01278-y Changes in phy oplank on p oduc ion and upwelling in ensi y o A Co uña (NW Spain) o he las 28yea s An onio Bode, Ma a Ál a ez, Manuel Ruíz-Villa eal & Ma a M.Va ela 1 23 You a icle is published unde he C ea i e Commons A ibu ion license which allows use s o ead, copy, dis ibu e and make de i a i e wo ks, as long as he au ho o he o iginal wo k is ci ed. You may sel - a chi e his a icle on you own websi e, an ins i u ional eposi o y o unde ’s eposi o y and make i publicly a ailable immedia ely. Changes in phy oplank on p oduc ion and upwelling in ensi y o A Co uña (NW Spain) o he las 28 yea s An onio Bode 1 &Ma a Ál a ez 1 &Manuel Ruíz-Villa eal 1 &Ma a M. Va ela 1 Recei ed: 25 Sep embe 2018 /Re ised: 15 May 2019 /Accep ed: 24 May 2019 /Published online: 10 June 2019 #The Au ho (s) 2019 Abs ac The analysis o a 28-yea -long (1989–2016) se ies o mon hly measu emen s o chlo ophyll concen a ions and p ima y p o- duc ion a es a a shel s a ion o A Co uña (NW Spain) p o ided e idence o changes a se e al ime scales ha we e only pa ly ela ed o upwelling in ensi y. Chlo ophyll de e mina ions we e made in ace onic ex ac s and p ima y p oduc ion a es by he measu emen o 14 C-up ake by na u al phy oplank on popula ions in simula ed in si u condi ions. Wa ele analysis e ealed mul iple modes o a ia ion in bo h se ies, pa icula ly a high equencies, bu some we e only signi ican o pa o he se ies. Fo ins ance, he seasonal cycle was no uni o m h ough he se ies despi e he annual epe i ion o maxima and minima. A mul iannual ime scales, bo h se ies we e di ided in h ee quasi-decadal pe iods cha ac e ized by signi ican inc eases in mean alues. Fluc ua ions in chlo ophyll and p ima y p oduc ion co a ied wi h changes in upwelling in ensi y a annual scales, bu annual means showed low co ela ion. Changes in dissol ed nu ien concen a ions om con inen al sou ces we e he likely d i e s o he obse ed changes in p oduc i i y a la ge ime scales. Inc eases in he decadal mean a es o p oduc ion and concen a ions o chlo ophyll we e d i en by inc eased in ensi y o sp ing blooms associa ed o inc eased nu ien s and low salini y wa e in he su ace. In con as , blooms caused by upwelling nu ien s emained unchanged along he se ies. This s udy illus a es he complexi y o in e ac ions in coas al upwelling a eas a la ge ime scales, whe e changes in con inen al nu ien inpu s may a ec phy oplank on p oduc ion mo e han a ia ions in upwelling in ensi y. Keywo ds P ima y p oduc ion .Upwelling .Nu ien s .Quasi-decadal a iabili y .NW Spain 1 In oduc ion The p ima y p oduc ion o he global ocean is expec ed o change in u u e clima e scena ios. Mos models p edic o he end o his cen u y a gene al dec ease o ca. 30% o he alues obse ed o he pe iod be ween 1980 and 1990, main- ly due o he expansion o oligo ophic a eas in he sub opical and equa o ial biomes (Beh en eld e al. 2006; Polo ina e al. 2008; Cab é e al. 2015). Such a educ ion in p oduc ion will ansla e in dec easing ishe ies and ma icul u e p oduc ion and will a ec he global use o ma ine esou ces e en beyond he di ec demands o sea ood (Chasso e al. 2010; Wa son e al. 2015). The wa ming o he su ace ocean in hese egions enhances s a i ica ion and educes he inpu o nu ien s om subsu ace laye s, hus causing a educ ion in p ima y p oduc- ion. E idence om sa elli e-de i ed measu emen s indica ed an inc ease o he low-chlo ophyll a eas in he sub opical gy es o he main oceans (Beh en eld e al. 2006; Polo ina e al. 2008). E en when some o hese changes can be a ib- u ed o pho oacclima ion (Beh en eld e al. 2016), he This a icle is pa o he Topical Collec ion on he 50 h In e na ional Liège Colloquium on Ocean Dynamics, Liège, Belgium, 28 May o 1 June 2018 Responsible Edi o : Aida Al e a-Azcá a e Elec onic supplemen a y ma e ial The online e sion o his a icle (h ps://doi.o g/10.1007/s10236-019-01278-y) con ains supplemen a y ma e ial, which is a ailable o au ho ized use s. *An onio Bode an [email protected] Ma a Ál a ez ma a.al a [email protected] Manuel Ruíz-Villa eal manuel. [email protected] Ma a M. Va ela ma a. a [email protected] 1 Ins i u o Español de Oceanog a ía, Cen o Oceanog á ico de A Co uña, Apdo, 130, 15080 A Co uña, Spain Ocean Dynamics (2019) 69:861–873 h ps://doi.o g/10.1007/s10236-019-01278-y dec easing nu ien inpu s would s ill limi p ima y p oduc ion in hese egions (Jang e al. 2011; Chus e al. 2014). Howe e , o he egions a e expe iencing inc eases in p i- ma y p oduc ion (Malone e al. 2016). Fo ins ance, he loss o sea ice and he inc ease in he g owing season ha e caused an inc ease o 20% in annual p ima y p oduc ion in he A c ic om 1998 o 2010 (B own and A igo 2012) bu his end may be e e sed i nu ien inpu s a e insu icien (T emblay and Gagnon 2009). In con as , a lowe ex en and du a ion o win e sea ice caused a dec ease in p oduc ion in An a c ic wa e s (Ducklow e al. 2013). Mos o he s udies we e conce ned wi h p oduc ion based on new nu ien s (i.e., hose p o ided om ex e nal sou ces o he pho ic zone), bu a signi ican ac ion o o al p ima y p oduc ion is based on nu ien s egene a ed in si u (Capone e al. 2008). Because plank on espi a ion inc eases as e wi h empe a u e han phy oplank on p ima y p oduc ion (Lopez- U u ia e al. 2006), he wa ming o su ace wa e s would ha e a posi i e eedback on he e o ophy, hus inc easing egene - a ed nu ien s and p oduc i i y. Models based on he p edic ed wa ming du ing his cen u y en isage an o e all inc ease in p oduc i i y, and less expo and cap u e o CO 2 ,causedby he enhanced p oduc i i y by egene a ed nu ien s (K ale e al. 2015). One o he main con ibu o s o global p ima y p oduc ion is he eas e n bounda y upwelling ecosys ems (EBUS). Co e ing less han 3% o he wo ld’s ocean su ace, hey accoun o be ween 10 and 20% o oceanic p oduc ion (Messié e al. 2009;Malonee al.2016), ha e a signi ican ole in he clima e sys em (La ge and Danabasoglu 2006), and con ibu e o 40% o he epo ed global ish ca ch (Pauly and Ch is ensen 1995; Capone and Hu chins 2013). An inc ease in p ima y p oduc ion in EBUS has been pu o h as a conse- quence o he in ensi ica ion o he alongsho e winds caused by he di e en ial wa ming o he land and he ocean su ace (Bakun 1990). The inc ease in wind speed would enhance he inpu o nu ien s om upwelled wa e s and hus p ima y p o- duc ion. E idence o hese inc eases has been p o ided by models (Rykaczewski e al. 2015) and obse a ions (Dema cq 2009), bu hese also e ealed ha he la ge he e ogenei ies in he wind changes a egional scales we e mainly due o he polewa d displacemen o he high a mosphe ic p essu e cells a he han an e ec o inc easing land–sea empe a u e di e - ences (Rykaczewski e al. 2015). In addi ion, egene a ed nu- ien s ampli y he ini ial p oduc ion igge ed by he upwell- ing, hus inc easing o al p oduc ion (e.g., Al a ez-Salgado e al. 1997). Fu he mo e, palaeocological s udies sugges he pa icipa ion o addi ional ac o s, as he inpu s o con i- nen al wa e s, a ec ing phy oplank on composi ion and p o- duc ion in upwelling ecosys ems (Ab an es e al. 2011). In he case o he EBUS including he egion om NW A ica o he NW Ibe ian Peninsula, p ima y p oduc ion is la gely d i en by he seasonal change in he dominan alongsho e winds a o ing he upwelling o subsu ace wa e s (A ís egui e al. 2006). Eas e n No h A lan ic Cen al Wa e s (ENACW) a e ad ec ed o he coas in he pe iod be ween Ma ch and Oc obe , when he winds a e p edominan ly equa- o wa d, in oducing la ge amoun s o new nu ien s in o he pho ic laye and enhancing p ima y p oduc ion (Al a ez- Salgado e al. 1997; Al a ez-Salgado e al. 2002). Changes in p oduc i i y we e desc ibed a geological ime scales (e.g., Ab an es e al. 2011), bu he e is con adic o y e idence as o he di ec ion o changes in ecen imes and he p ojec ions o he u u e. Recen in ensi ica ion o upwelling was desc ibed o he A ican coas (McG ego e al. 2007), bu o he analy- sis o wind da a showed di e ences in ends a sub egional scale (Al a ez e al. 2008; Pé ez e al. 2010; San os e al. 2011), a gene al weakening in he egion (Pa do e al. 2011) o no clea e idence o upwelling dec ease (Ba on e al. 2013; Benazzouz e al. 2015). Indeed, es ima ions based on egional clima e models p edic an inc ease in upwelling- a o able winds in he egion (Casabella e al. 2014). Sa elli e-de i ed chlo ophyll se ies did no e eal a consis en end h ough he egion (Bode e al. 2011; Al a ez e al. 2012: Benazzouz e al. 2015) bu in si u de e mina ions nea he coas sugges la ge yea o yea a ia ions (Bode e al. 2011)o e eninc easesa decadal ime scales (Pé ez e al. 2010). The con inuous se ies o in si u p ima y p oduc ion a e de e mina ions a ailable in he egion showed a sus ained inc ease o he pe iod 1989– 2010 a an upwelling si e (A Co uña), while a gene al dec ease in p ima y p oduc ion was obse edinCudille o,on he no he n Ibe ian coas , whe e upwelling has lowe in luence (Bode e al. 2011). The objec i e o his s udy is o analyze pe iodic and long- e m changes in he se ies o phy oplank on chlo ophyll and p ima y p oduc ion a es measu ed o A Co uña (Galicia, NW Spain) be ween 1989 and 2016 in ela ion o changes in upwelling in ensi y and nu ien inpu s. The ocus is on he changes obse ed in he whole wa e column, as he phy o- plank on p oduc ion occu s in he pho ic laye bu is a ec ed by he nu ien s p o ided om deep laye s du ing upwelling e en s. 2Me hods 2.1 Sampling The s udy si e was s a ion E2CO (43° 25.32′N, 8° 26.22′W, 80 m deep) om he RADIALES p og am (h p://www. se ies empo ales-ieo.com/) o he Ins i u o Español de Oceanog a ía (IEO, Spain). The s a ion was isi ed a app oxima ely mon hly in e als since 1989 wi h collec ion o wa e samples a 7 dep h le els using Niskin bo les a e CTD cas s (Seabi d SBE25). Since 2002 a Seabi d Compac Ca ousel (SBE32C) was employed o acili a e he sampling 862 Ocean Dynamics (2019) 69:861–873 o wa e du ing he CTD p o iles. The CTD was equipped wi h a LiCo unde wa e adiome e o he de e mina ion o pho osyn he ically ac i e adia ion (PAR) p o iles. Wa e sam- ples we e collec ed o de e mine phy oplank on chlo ophyll and p oduc ion, ino ganic nu ien s, dissol ed oxygen and ses on ca bon and ni ogen. Mo e de ails o sampling and desc ip ion o seasonal a ia ions a his s a ion can be ound elsewhe e (Casas e al. 1997;Bodee al.2015). In his s udy, we analyzed he ull se ies o obse a ions be ween Janua y 1989 and Decembe 2016. 2.2 Measu emen s Phy oplank on chlo ophyll (Chl) was measu ed on plank on collec ed on Millipo e cellulose ace a e 0.8 μmpo esize(un il 1992) o Wha man GF/F il e s, ex ac ed in 90% ace one o e nigh and measu ed luo ome ically (Pa sons e al. 1984). F om he yea 2000 onwa ds, he ex ac s we e ana- lyzed using he spec o luo ome ic me hod (Ne eux and Panouse 1987) and in e calib a ion adjus men s be ween me hods we e made o ensu e he compa ibili y o he se ies o he en i e sampling pe iod. As a esul o he in e calib a- ion, luo ome ic measu emen s we e mul iplied by 0.65 o es ima e spec o luo ome ic Chl. P ima y p oduc ion a es (PP) we e es ima ed by measu ing he up ake o 14 C-labeled bica bona e in samples incuba ed in simula ed empe a u e and i adiance condi ions equi alen o hose o he o iginal sampling dep h using unning su ace wa e and neu al den- si y sc eens, espec i ely (Bode and Va ela 1998;Bodee al. 2011). Samples o p ima y p oduc ion de e mina ions co esponded o dep hs ecei ing 100%, 50%, 25%, 10%, and 1% o su ace PAR, as measu ed wi h he CTD PAR senso . Two 100 mL eplica es plus one da k bo le we e illed on boa d wi h wa e o each dep h, anspo ed he labo a o y (wi hin 2 h o collec ion) and inocula ed wi h 148 (± 57) kBq o NaH 14 CO 3 . Incuba ions we e made unde di ec sunligh a he labo a o y (A Co uña) o 2–3 h a noon and we e e mi- na ed by il a ion on he same ype o il e s used o chlo o- phyll de e mina ions. The excess 14 C was emo ed om he il e s by adding 10% HCl o e nigh and he adioac i i y de e mined in a liquid scin illa ion coun e (LKB Wallac 1409 un il 2006 and Packa d T iCARB 2900TR he ea e ) using au oma ic quenching co ec ion wi h in e nal s anda ds. Scin illa ion cock ails we e Ins a-Gel (Pe kin Elme ) un il 1996 and Ul ima-Gold XL (Pe kin Elme ) he ea e . Ra es o PP we e co ec ed o ino ganic ca bon inco po a ion in he da k bo le and exp essed as mg C m −3 h −1 o each dep h le el. Ino ganic nu ien concen a ions (ni a e+ni i e, phos- pha e, and silica e) we e de e mined in samples p ese ed o- zen in polye hylene ials (−20 °C) by segmen ed low analy- sis using a Technicon AA-II (1989–2006), a B an-Luebbe AA3 (2007–2012), o a Seal Analy ics QuAA o 39 (2013– 2016) analyze . The colo ime ic me hods employed we e basedin hosedesc ibedbyG assho e al.(1983). Dissol ed oxygen was de e mined in duplica e 125-mL bo - les using he Winkle i a ion p ocedu e (Pa sons e al. 1984) and pa icula e o ganic ca bon and ni ogen (POC and PON, espec i ely) in ses on collec ed on Wha man GF/C (1989– 1992) o GF/F il e s analyzed in a Pe kin Elme CHN 2400 (1989–2009) o a Ca lo E ba CHNSO 1108 elemen al analyz- e (since 2010). Nu ien anomalies ela i e o he expec ed concen a ions in ENACW ( he main inpu om upwelling) we e de e mined as he di e ence be ween he obse ed con- cen a ions in bo om wa e (70 m dep h) o sampling da es du ing upwelling e en s and ENACW nu ien concen a- ions, he la e es ima ed om in si u empe a u e using he equa ions p o ided by Ál a ez-Salgado e al. (Al a ez- Salgado e al. 2002). Such equa ions we e de e mined wi h da a om 14 c uises in he NW Ibe ian upwelling sys em be ween 1977 and 1998, excluding da a om s a ions wi h wa e dep h < 1000 m o a oid he nu ien en ichmen in shel bo om wa e s due o in ense mine aliza ion. De ailed analysis o nu ien in luence on he Chl and PP se ies was made by conside ing wo main sou ces o nu ien s o p ima y p oduc- ion in he s udy a ea: he bo om wa e s du ing he upwelling pe iod (Ma ch o Oc obe ) and he con inen al wa e s ( i e discha ge and uno ) a ec ing he su ace mainly du ing he win e –sp ing pe iod (Feb ua y o May). Upwelling in ensi y (UI) was ob ained om he Ekman anspo de i ed om su ace winds (h p://www. indicedea lo amien o.ieo.es/) in a cell o 1° × 1° cen e ed a 44° N, 9° W, using da a om a mosphe ic p essu e a sea le el de i ed om he WXMAP model (González-Nue o e al. 2014). Posi i e alues o his index (exp essed in m 3 s − 1 km −1 ) indica e ne upwelling pe iods when su ace wa e is anspo ed o sho e while nega i e alues indica e an accu- mula ion o su ace wa e agains he coas (downwelling). Fo he pu pose o he p esen s udy, 6-hou ly UI alues we e a e aged o e he 15 days p eceding each sampling e en o E2CO s a ion, as p e ious s udies indica ed ha Chl was co - ela ed wi h upwelling e en s occu ing se e al days be o e sampling (e.g., Noguei a e al. 1997). In luence o uno du ing sp ing was es ima ed om ain- all obse a ions in A Co uña (Agencia Es a al de Me eo ologia, AEMET, Spain). In his s udy, ain all da a we e accumula ed o he 15 days p eceding each sampling e en a sea. Raw da a o wa e and plank on obse a ions used in his s udy can be accessed h ough he PANGAEA eposi o y (h ps://doi.pangaea.de/10.1594/PANGAEA.885413). 2.3 S a is ics Fo he pu pose o his s udy, alues o all a iables we e summa ized o each sampling da e. Chlo ophyll, p ima y Ocean Dynamics (2019) 69:861–873 863 p oduc ion a es, POC and PON we e in eg a ed, and nu ien and oxygen da a we e a e aged in he eupho ic zone (1% o su ace PAR). The dep h o he eupho ic zone a ied be ween 14 and 53 m and showed no empo al ends (Online Resou ce 1). The o iginal PP hou ly a es we e no scaled o daily a es o p ese e he o iginal se ies and o a oid assump ions on he app op ia e scaling ac o s. O he s udies on subse s o he same se ies p o ide a ious es ima es o daily p oduc ion a es a he same s a ion (e.g., Bode and Va ela 1998;Bodee al. 2011). Tempe a u e and salini y measu emen s we e ep e- sen ed by su ace alues (SST and SSS, espec i ely) and nu ien s p o ided by he upwelling by hei concen a ions in he bo om laye (70 m). Time se ies o all a iables we e explo ed o he p esence o ou lie s, no mali y, homoscedas- ici y, and colinea i y among a iables (e.g., Zuu e al. 2010) using he PAST 3.17 package (Hamme e al. 2001). Mul iannual pe iods o change in Chl, PP, and UI we e i s iden i ied using he cumula i e sum o he mean dis- ances me hod (Ibañez e al. 1993) by emo ing om each empo al obse a ion he mean o he ime se ies and plo - ing he cumula i e sum o esiduals. Changes in he slope o he cumula i e sum cu e om posi i e o nega i e o ice e sa indica e phases o which he mean o he da a is highe o lowe han he mean o he o iginal ime se ies. Fu he analysis o pe iodici y a small (mon hs) and la ge ime scales (yea s) in he ime se ies was made by compu - ing he con inuous wa ele ans o ma ion on each ime se ies a e no maliza ion (log 10 - ans o ma ion) and egu- la iza ion o pe iods o 30 days using he quad a ic in e - pola ion me hod in PAST. The wa ele s enable he ex ac- ion o ime-dependen ampli ude cycles ela ed o equen- cies (To ence and Compo 1998; Cazelles e al. 2008). The plo o he squa ed co ela ion s eng h wi h he scaled mo he Mo le wa ele allowed o inspec ing he se ies a small, in e media e, and la ge equencies simul aneous- ly. The signi icance o he iden i ied pe iodici ies was assessed wi h a χ 2 es and he cone o in luence compu ed o disca d bounda y e ec s (To ence and Compo 1998). Wa ele analysis and ela ed es s we e made wi h PAST. The in luence o en i onmen al a iables on Chl and PP se ies was i s analyzed using he annual means o each se ies. En i onmen al a iables we e i s no malized, cen- e ed, and s anda dized. The signi icance o au o- and c oss-co ela ion be ween se ies o a iables was de e - mined a e Da is (1986). Co ela ion analysis was used o emo e highly co ela ed a iables (Online Resou ce 2) and SST, SSS, mean oxygen, POC and bo om wa e nu ien concen a ions, along wi h UI we e e ained o a p incipal componen analysis (PCA) e lec ing he main modes o a iabili y in hese se ies. The se ies o Chl and PP we e hen p ojec ed on he space o he i s compo- nen s o iden i y he main associa ions. In a second s ep, changes in he en i onmen al a iables be ween he phases o Chl and PP se ies iden i ied wi h he cumula i e sums me hod we e analyzed a in a-seasonal scale. The signi i- cance o a iabili y in Chl, PP, and en i onmen al a iables was de e mined using pe mu a ional ANOVA es s (PERMANOVA) compu ed wi h PAST. 3 Resul s 3.1 Seasonali y s. long- e m pe iodici y All se ies eco ded du ing he las 28-yea pe iod showed ecu en maxima and minima each yea bu also mul iannual pe iods o ela i ely mino change (Online Resou ce 3). As desc ibed in p e ious s udies, minimum wa e empe a u e alues (and maximum nu i- en concen a ions) we e obse ed in la e win e in su ace wa e s, whe e he con e se si ua ion was mo e likely in summe (e.g., Casas e al. 1997). The equen incidence o upwelling e en s du ing mos o he yea , bu mainly be ween Ma ch and Oc obe , caused he cooling and nu i- en en ichmen o he su ace and subsu ace laye s a alues nea hose o he la e win e minimum. Such chang- es we e e lec ed by he high a iabili y in he dis ibu ion o Chl and PP, wi h peaks a ec ing he whole pho ic zone du ing he en i e upwelling season each yea (Fig. 1). Howe e , he magni ude o he annual changes a ied a low equencies, as shown by he accumula ed mon hly anomalies. The en i e se ies o Chl and PP could be di id- ed in o h ee pe iods o 9–10 yea s, when he e we e ab up changes in he slope o he accumula ed anomalies o he se ies (mainly o Chl bu also o PP). The i s pe iod (1989–1997) was cha ac e ized by a dec easing end in he anomalies, wi h Chl and PP alues below he o al mean o he se ies. This dec ease was s abilized in he second pe iod (1998–2006), bu he alues we e s ill lowe han he mean. Finally, a hi d pe iod (2007–2016) was cha ac e ized by an inc easing end in he anomalies and Chl and PP alues inally eaching alues abo e he mean. No wi hs anding he gene al synch ony obse ed be ween Chl and PP, some di e ences we e appa en , mainly ela ed wi h he slow bu pe sis en inc ease o he PP anomalies du ing he cen al pe iod (1998–2006). Annual epe i ion o maximum and minimum alues was also p esen by he upwelling index se ies, bu in his case, he mul iannual pe iods eme ging om he accumu- la ed anomalies we e clea ly di e en om hose o Chl and PP (Fig. 1). Two main pe iods (1989–2003 and 2004– 2016) showed a nea ly symme ical bell pa e n o UI: inc easing he posi i e anomalies o a maximum nea he cen e o he pe iod and hen dec easing again o low alues. I mus be no ed ha posi i e anomalies domina - ed in bo h pe iods, indica ing ha he alues we e mainly 864 Ocean Dynamics (2019) 69:861–873 abo e he mean o he whole se ies, wi h nega i e anom- alies concen a ed in he ansi ion be ween pe iods a he middle o he se ies. When compa ed wi h he phases desc ibed o Chl o PP, UI anomalies did no show a consis en pa e n o change, as alues inc eased and de- c eased du ing each o he pe iods. In addi ion, he mean alues o UI o he h ee pe iods iden i ied o Chl and PP we e no signi ican ly di e en (PERMANOVA, F 2,326 = 0.945, p= 0.381). A u he compa ison o mon hly, annual, and long- e m a iabili y was p o ided by he wa ele analysis, showing a signi ican annual pe iodici y (i.e., maxima nea a 12-mon h pe iod) bu also a o he equencies in all a iables (Fig. 2). Fo ins ance, he seasonali y in Chl was no always signi ican bu only in he ini ial yea s o he pe iods 1989–1997 and 1998–2006, while since 2007, he e we e e en wo annual peaks (i.e., signi ican maxi- ma nea 6- and 12-mon h pe iods). In addi ion, he e was signi ican pe iodici y in Chl a equencies be ween 3 and 5 yea s. Simila pe iodici ies we e obse ed o he PP se ies, bu in his case, he signi icance o he annual seasonali y was es ic ed o he phases a e 1998 and he bimodal peaks o only some yea s o he las pe iod (2007–2016). In bo h cases, and because o he o al leng h o he se ies, he quasi-decadal phases iden i ied wi h he accumula ed anomalies in Fig. 1appea ed a he bounda ies o he cone o in luence (i.e., nea 9 yea s) bu only we e signi ican o PP. In con as , he a iabil- i y in UI was signi ican a mul iple low and high equen- cies bu maximum alues concen a ed in all cases a he annual equencies, pa icula ly in ecen yea s. The sym- me y o he whole UI se ies was indica ed by he ela i e maximum in he local powe spec um a equencies nea 8 yea s (i.e., 6.6 in he log 2 scale). As indica ed wi h he wa ele analysis, he mon hly dis i- bu ions o Chl and PP se ies a ied signi ican ly be ween phases (Fig. 3). While he dis ibu ion o mon hly means o bo h a iables was qui e homogeneous o he i s wo phases, he e was a ma ked peak du ing he i s hal o he yea o he 2007–2016 phase (Wilcoxon signed ank es , p< 0.05). Consequen ly, means o he win e –sp ing season in 2007–2016 we e highe han hose in p e ious pe iods (PERMANOVA and Bon e oni es s, p< 0.05, Online Resou ce 4). When conside ing whole phases, he mean alues o Chl and PP inc eased signi ican ly h ough he s udy pe iod (PERMANOVA and Bon e oni es s, p< 0.05, Fig. 4). 3.2 En i onmen al a iabili y F om he PCA on he selec ed annually a e aged a i- ables, h ee main componen s accoun ing o 35.4, 17.9, -12.0 -8.0 -4.0 0.0 4.0 0.0 0.5 1.0 1.5 2.0 2.5 1989 1992 1995 1998 2001 2004 2007 2010 2013 2016 Accum. anomal y m C g( PP -2 h-1) -1.0 -0.8 -0.6 -0.4 -0.2 0.0 0 100 200 300 1989 1992 1995 1998 2001 2004 2007 2010 2013 2016 Accum. anomaly -10.0 -5.0 0.0 5.0 10.0 15.0 20.0 -6.0 -3.0 0.0 3.0 1989 1992 1995 1998 2001 2004 2007 2010 2013 2016 Accum. anomaly xednI gnillewpU (a) (c) (b) Chl (mg m-2) Fig. 1 Time se ies ( hin lines) o a pho ic-zone in eg a ed chlo ophyll (Chl, mg m −2 ), b p ima y p oduc ion (PP, gCm −2 h −1 ), and cEkman anspo (upwelling index, m 3 s −1 km −1 ) and he co esponding accumula ed anomalies om he mean alue o each se ies ( hick lines). The e ical dashed lines indica e h ee phases o change in Chl and PP Ocean Dynamics (2019) 69:861–873 865 and 16.7% o o al a iance we e selec ed (Fig. 5, Online Resou ce 5). The i s componen was posi i ely co ela ed mainly wi h ni a e and phospha e concen a- ions and salini y, and nega i ely wi h POC and su ace empe a u e. In u n, he highes posi i e co ela ions o he second componen we e obse ed o oxygen concen- a ion, while he main nega i e co ela ions we e hose o silica e. The hi d componen was posi i ely co ela ed wi h all a iables excep wi h salini y, which showed a ma ked nega i e co ela ion. The e o e, he i s compo- nen sepa a ed yea s o ela i ely high nu ien concen a- ion, low wa e empe a u e, and dominance o saline wa- e s om hose o high concen a ion o ses on, highe empe a u e, and lowe salini y and nu ien concen a- ions. The second componen highligh ed he opposi e e- la ionship be ween oxygen and silica e concen a ions, and he hi d componen appea ed ela ed wi h yea s wi h ela i ely high inpu s o nu ien - ich con inen al wa e s by uno (i.e., low salini y wa e s in Online Resou ce 2). The mean upwelling in ensi y had a ela i ely weak co ela ion wi h all h ee PCA componen s and he same occu ed wi h Chl and PP when p ojec ed on he PCA 1990 1996 2002 2008 2014 (a) (c) 2.8 8.0 22.6 64.0 2.8 8.0 22.6 64.0 2.8 8.0 22.6 64.0 1990 1996 2002 2008 2014 500 100 150 200 250 300 pe iod (mon hs) 500 100 150 200 250 300 500 100 150 200 250 300 (b) scale (mon hs)scale (mon hs)scale (mon hs) Fig. 2 Local wa ele powe spec um o pho ic-zone in eg a ed a chlo ophyll, bp ima y p oduc ion, and cupwelling index ime se ies. The X-axis (in e al, mon hs) indica es he leng h o he se ies and he Y-axis he log 2 ime scale, wi h he signal obse ed a a scale o only wo consecu i e da a poin s a he op, and a a scale o one ou h o he whole sequence a he bo om. The con inuous black lines deno e he 5% signi icance a eas and he cone o in luence (i.e., egion o he wa ele spec um in which edge e ec s become impo an ). The colo s indica e he squa ed co ela ion s eng h o each se ies wi h he scaled mo he wa ele . The e ical dashed lines indica e he h ee phases iden i ied in Fig. 1 0 20 40 60 80 J F M A M J J A S O N D mon h 1989-1997 1998-2006 2007-2016 0 100 200 300 400 500 JFMA MJJASOND m C gm( PP -2 h -1 ) mon h (a) (b) Chl (mg m-2) Fig. 3 Mean (± SE) mon hly alues o pho ic-zone in eg a ed a chlo ophyll (Chl, mg m −2 )andbp ima yp oduc ion(PP, mg C m −2 h −1 ) o he di e en phases iden i ied in Fig. 1 866 Ocean Dynamics (2019) 69:861–873 space (Fig. 5). Fo he la e a iables, he highes co e- la ions we e posi i e wi h he hi d componen , ollowed by hose wi h he i s (also posi i e) and he second (nega i e) componen s. All hese co ela ions indica e ha Chl and PP a annual ime scales we e a he in luenced by a combina ion o a iables, mainly ela ed wi h he nu ien inpu s, han by a simple esponse o upwelling in ensi y. Mean nu ien concen a ions in bo om wa e s, excep sil- ica e, we e simila o he h ee phases (Fig. 6). Fu he mo e, he e was no a clea pa e n o change in he dis ibu ion o anomalies om he e e ence ENACW nu ien concen a- ions be ween phases (Online Resou ce 6). In e annual a i- abili y was high and anomalies we e mos ly posi i e. In con- as , he e was a signi ican inc ease in mean concen a ions o all nu ien s in su ace wa e s om he i s o he las phase o he s udy (Fig. 6). Such inc eases we e concu en wi h a signi ican dec ease o mean salini y and an inc ease in ain- all, pa icula ly a e he phase be ween 1989 and 1997 (PERMANOVA and Bon e oni es s, p< 0.05, Fig. 7). 4 Discussion 4.1 High- s. low- equency oscilla ions The analysis o a se ies spanning 28 yea s showed signi ican quasi-decadal a iabili y in p ima y p oduc ion in a coas al si e a ec ed by upwelling. This esul con i ms p e ious anal- ysis o a sho e e sion o he se ies o he pe iod 1989–2006 (Bode e al. 2011) and aligns wi h simila quasi-decadal a - iabili y ound in o he EBUS (Heymans e al. 2004; La aniegos and Ohman 2007; Dema q Dema cq 2009). Changes in he egional clima e a ec ing wind egimes and cu en s we e o en in oked as he mos likely d i e s o he obse ed a iabili y a long- e m scales. Fo ins ance, Cha ez e al. (2011) no ed ha El Niño–Sou he n Oscilla ion (ENSO) anomalies in 1997–1998 s ongly dep essed p ima y -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 PC2 PC1 Chl PP O2 POC SST SiO4 SSS NO3 PO4 UI (a) -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 -1.0 -0.5 0.0 0.5 1.0 PC3 PC2 PP Chl SSS SiO4 SST O2 POC UI PO4 NO3 (b) Fig. 5 P ojec iono pho ic-zonein eg a ed chlo ophyll (Chl) and p ima y p oduc ion (PP) on he space o he 3 p incipal componen s (PC1, PC2, and PC3) ex ac ed om he se ies o en i onmen al a iables. Co ela ions be ween he en i onmen al a iables and he p incipal componen s a e indica ed by ec o s. SST sea su ace empe a u e, SSS sea su ace salini y, NO3, PO4, and SiO4 o al ni a e, phospha e, and silica e concen a ion nea he bo om, O2 mean oxygen concen a ion in he pho ic zone, POC mean pa icula e o ganic ca bon in he pho ic zone, UI upwelling index 0 10 20 30 40 50 60 70 1989-1997 1998-2006 2007-2016 a a,b b 0 50 100 150 200 250 300 350 400 1989-1997 1998-2006 2007-2016 m C gm( PP -2 h-1) a b b (a) (b) Chl (mg m-2) Fig. 4 Mean (± SE) annual alues o eupho ic-zone in eg a ed a chlo ophyll (Chl, mg m −2 )andbp ima y p oduc ion (PP, mg C m −2 h −1 ) o he di e en phases iden i ied in Fig. 1. The le e s abo e each ba iden i y signi ican means (PERMANOVA and Bon e oni es s, p<0.05) Ocean Dynamics (2019) 69:861–873 867