Depuration of geosmin- and 2-methylisoborneol-induced off-flavors in recirculating aquaculture system (RAS) farmed European whitefish Coregonus lavaretus

This is a sel -a chi ed e sion o an o iginal a icle. This e sion may di e om he o iginal in pagina ion and ypog aphic de ails. Au ho (s): Ti le: Yea : Ve sion: Copy igh : Righ s: Righ s u l: Please ci e he o iginal e sion: CC BY 4.0 h ps://c ea i ecommons.o g/licenses/by/4.0/ Depu a ion o geosmin- and 2-me hylisobo neol-induced o - la o s in eci cula ing aquacul u e sys em (RAS) a med Eu opean whi e ish Co egonus la a e us © The Au ho (s) 2019 Published e sion Lindholm-Leh o, P.C.; Vielma, J.; Pakkanen, H.; Alén, R. Lindholm-Leh o, P.C., Vielma, J., Pakkanen, H., & Alén, R. (2019). Depu a ion o geosmin- and 2- me hylisobo neol-induced o - la o s in eci cula ing aquacul u e sys em (RAS) a med Eu opean whi e ish Co egonus la a e us. Jou nal o Food Science and Technology, 56(10), 4585- 4594. h ps://doi.o g/10.1007/s13197-019-03910-7 2019 ORIGINAL ARTICLE Depu a ion o geosmin- and 2-me hylisobo neol-induced o - la o s in eci cula ing aquacul u e sys em (RAS) a med Eu opean whi e ish Co egonus la a e us P. C. Lindholm-Leh o 1 •J. Vielma 1 •H. Pakkanen 2 •R. Ale ´n 2 Re ised: 18 June 2019 / Accep ed: 1 July 2019 ÓThe Au ho (s) 2019 Abs ac Eu opean whi e ish Co egonus la a e us has inc easingly become an impo an species o aqua ic ood p oduc ion, especially in he No dic coun ies. Whi e ish is p oduced in adi ional cage and pond ope a ions, and in eci cula ing aquacul u e sys em (RAS) in which, un o - una ely, o - la o s and odo s, mos ly caused by geosmin (GSM) and 2-me hylisobo neol (MIB), can accumula e in ish lesh om he ci cula ing wa e . GSM and MIB ha e e y low human senso y de ec ion limi s and, he e o e, o en disliked by consume s e en a low concen a ions. In his s udy, concen a ions o GSM and MIB in RAS a med Eu opean whi e ish we e s udied by an analy ical me hod based on headspace solid phase mic oex ac ion and gas ch oma og aphy–mass spec ome y. Concen a ions we e de e mined in di e en pa s o ish: ille , neck, belly, and ail du ing a depu a ion pe iod and in depu a ion wa e . The highes ini ial concen a ions we e on a e age 32 ng g -1 (GSM) and 24 ng g -1 (MIB) in Eu opean whi e ish ille and 128 ng L -1 (GSM) and 94 ng L -1 (MIB) in wa e , espec i ely. A e a depu a ion pe iod o 16 days, concen a ions dec eased o below he de ec ion limi s, indica ing he impo ance o he depu a ion pe iod. Keywo ds Depu a ion Eu opean whi e ish Geosmin 2- Me hylisobo neol RAS a ming In oduc ion Aquacul u e has been a apidly g owing ood p oducing sec o and nowadays accoun s o mo e han hal o he ish used o human consump ion (FAO 2016). O - la o -in- duced ain ing o ish has adi ionally occu ed in ou doo ponds bu has o en been obse ed also in eci cula ing sys ems (Houle e al. 2011; Sch ade and Dennis 2005). In eci cula ing aquacul u e sys ems (RAS), unwan ed o - la o s wi h unpleasan odo s can accumula e in ci cula ing wa e and in ish lesh (Ha hu usingha and Da ey 2014; Houle e al. 2011; Howga e 2004). O - la o s in ish lesh ha e been widely documen ed o , e.g. a c ic cha Sal elinus alpinus (Houle e al. 2011), A lan ic salmon Salmo sala (Bu e al. 2012), ba amundi La es calca i e (Pe ci al e al. 2008), la gemou h bass Mic op e us sal- moides, Nile ilapia O eoch omis nilo icus (Yamp ayoon and Noomho m 2003), and ainbow ou Onco hynchus mykiss (Robe son e al. 2005). The p esence o o - la o in ish lesh is pe cei ed as a quali y de ec and o en disliked by consume s, which can lead o high inancial losses in p oduce s’ ea nings. O - la o s a e ypically emo ed by depu a ing he ish in clean wa e . Un o u- na ely, his o en akes om days o weeks, and can lead e en o signi ican economic losses due o delays o ha es and high consump ion o clean wa e (Bu e al. 2012). O - la o is o en caused by wo sa u a ed bicyclic e penoids, geosmin (GSM, ans-1,10-dime hyl- ans-9- decalol) and 2-me hylisobo neol (MIB (1-R-exo)-1,2,7,7- e ame hyl-bicyclo[2.2.1]hep an-2-ol) (Ge be 1968, 1969), p oducing ea hy and mus y odo and as e. GSM Elec onic supplemen a y ma e ial The online e sion o his a icle (h ps://doi.o g/10.1007/s13197-019-03910-7) con ains sup- plemen a y ma e ial, which is a ailable o au ho ized use s. &P. C. Lindholm-Leh o [email p o ec ed] 1 Aqua ic P oduc ion Sys ems, Na u al Resou ces Ins i u e Finland (Luke), Su on ie 9A, 40500 Jy a ¨skyla ¨, Finland 2 Depa men o Chemis y, Uni e si y o Jy a ¨skyla ¨, Box 35, 40014 Jy a ¨skyla ¨, Finland 123 J Food Sci Technol h ps://doi.o g/10.1007/s13197-019-03910-7 and MIB a e p oduced as seconda y me aboli es by a a ie y o bac e ia, including ac inomyce es, cyanobac e ia, p o eobac e ia and ungi (Dickscha e al. 2005). Typically, s ep omyces, myxobac e ia, and ac inomyce es a e he o ganisms esponsible o GSM and MIB p oduc ion in indoo eci cula ing sys ems (Dickscha e al. 2005). Wa e phase and he bio il e s ha e been sugges ed as he main sou ce o GSM-p oduce s in RAS, bu so a , he e is no consensus on which RAS compa men is dominan in GSM and MIB p oduc ion (Lukassen e al. 2017). GSM and MIB a e semi- ola ile compounds wi h oc a- nol-wa e pa i ion coe icien s 3.57 o GSM and 3.31 o MIB (Howga e 2004). E en low concen a ions in wa e can be abso bed by ish and accumula ed in lipid- ich is- sues (Houle e al. 2011; Howga e 2004), gi ing an unpleasan as e and odo o wa e (Smi h e al. 2002) and ish (Howga e 2004). Addi ionally, bo h compounds a e pe cei ed a e y low concen a ions ( 1ngg -1 )by human senses (Da idson e al. 2014; Robe son e al. 2005). The a con en o ish has an e ec on he la o sen- sa ion and he senso y h eshold o lipophilic GSM and MIB (Howga e 2004; Zimba and G imm 2015). Senso y h esholds o GSM and MIB (muddy and ea hy la o ) inc ease wi h a con en s o ish (D ake e al. 2010; Howga e 2004), lea ing a highe concen a ion unde ec ed as he a con en inc eases. Th eshold alues a e used in senso y analyses and gene ally de ined as he p obabili y o de ec ion (Ridgway e al. 2010). Acco ding o he de ini- ion, 50% o popula ion will de ec he ain a ha le el. Howe e , indi iduals ha e di e en abili ies in de ec ing as es and odo s and some indi iduals can be mo e sensi- i e o ce ain compounds. GSM and MIB a e abso bed in o he ish ia gills, skin and gas oin es inal ac by lipid- ich issues, gills being he main pa h o up ake (Howga e 2004). GSM and MIB a e concen a ed in o he ish lesh un il an equilib ium s a e is eached be ween he wa e phase and he ish. Thei concen a ions in wa e and he ime o exposu e a e he main ac o s a ec ing hose in ish lesh (Howga e 2004), bu also e.g. ish species, wa e empe a u e, size, and age o ish a e o impo ance (Pe ci al e al. 2008). The exchange o chemicals is assumed o p oceed h ough a passi e p ocess, a ec ed by he lipophilic na u e o he compounds and he concen a ion in he aqueous pa o ish (Howga e 2004). Se e al me hods ha e been s udied and es ed o emo e o dec ease he o ma ion o GSM and MIB in RAS, including ozona ion (Powell and Scolding 2018), o he ad anced oxida ion p ocesses (AOPs) (Ru angwa and Ve degem 2015), algicides (Ha hu usingha and Da ey 2014), adso p ion, o example, wi h ac i a ed ca bon (Bu e al. 2012), and zeoli es (Ghasemi e al. 2018). Mo e ecen ly, me hods based on pho oca alysis, such as modi- ied TiO 2 wi h sun ligh (Fo iou e al. 2015) and palladium modi ied ungs en ioxide pho oca alys (Xue e al. 2016) o deg ade o - la o compounds seem p omising in he u u e. Howe e , depu a ing wi h clean wa e is so a he only e icien me hod a ailable o emo e he o - la o s. Depu a ion can ake om days o weeks, depending on, o example, he ini ial concen a ions o GSM and MIB, he species, and he size o ish, bu equi es la ge olumes o wa e (Bu e al. 2012). Addi ionally, ish a e ypically no ed du ing he depu a ion o ensu e as good wa e quali y and e icien depu a ion as possible, which can lead o hei weigh loss. Eu opean whi e ish Co egonus la a e us is a whi e- le- shed eshwa e species which belongs o he amily Sal- monidae and a genus Co egonus spp. The g oup includes ma ine, anad omous, and eshwa e species widely dis- ibu ed in No he n Eu ope, No h Ame ica, and Asia (Boi eanu e al. 2016). In ecen yea s, Eu opean whi e ish has inc easingly become an eme ging species o aqua- cul u al ood p oduc ion in he No dic coun ies. Typically, whi e ish is consumed as aw pickled, smoked, ille o o he p oduc s, including ca ia (Se a ¨la ¨2008). So a , concen a ions in o GSM and MIB in ish lesh o di e en species ha e been a subjec o limi ed numbe o s udies. Addi ionally, only ew s udies ha e been con- duc ed o de ec he loca ion o o - la o accumula ion in ish (Zimba and G imm 2015). Acco ding o ou knowl- edge, his is he i s s udy ega ding GSM and MIB con- cen a ions in Eu opean whi e ish du ing a depu a ion pe iod. Aim o his esea ch was o ex ensi ely s udy he loca ion o o - la o accumula ion in ish lesh and he ime equi ed o su icien depu a ion. Ma e ials and me hods Chemicals and ma e ials S anda d solu ion (T aceCERT Ò , 100 lgmL -1 in MeOH) o (?/-) GSM and MIB we e pu chased om Me ck. High-pe o mance liquid ch oma og aphy (HPLC) g ade me hanol and hexane we e ob ained om J.T Bake and solid NaCl (pu i y 98%) om Me ck. Ul a-high quali y (UHQ) wa e om Millipo e (Bed o d, MA, USA) was used in he analyses. Addi ionally, 10 mL headspace (HS) glass ials and poly e a luo oe hylene (PTFE) sep um caps we e pu chased om Me ck. Manual solid phase mic oex ac ion (SPME) assembly wi h an ex ac ion ibe coa ed wi h S ableFlex di inyl- benzene/ca boxene/polydime hyl siloxane (DVB/CAR/ PDMS), 1 cm, 50/30 lm (pa no. 57328-U) in a manual holde we e pu chased om Supelco (Me ck). J Food Sci Technol 123 Sampling and p e ea men P ocess condi ions in RAS Eu opean whi e ish was aised in a comme cial indoo RAS a m. The ish we e ed un il he beginning o he depu- a ion pe iod wi h comme cially p oduced d ied pelle s, con aining 0.8–1.7% phospho ous and 5.9–8.5% ni ogen wi h highe concen a ions in ju enile eeds. In gene al, he main ing edien s o eeds a e ish meal, ish oil, soya, whea , and apeseed oil, supplemen ed wi h i amins and ace elemen s. A he a m, a e age biomass o Eu opean whi e ish was 30 , ed 200 kg eed pe day. In he RAS p ocess, ammonium was emo ed by ni i ica ion in he bio- il e , while dissol ed ca bon dioxide was emo ed om he wa e by packed ae a ion owe . In he ci cula ion sys em, i e wa e o wa e om a d ill well was used as he sou ce o he enewal wa e . The ci cula ion wa e was disin ec ed wi h ozone and UV ligh , pH adjus ed wi h sodium hyd oxide o sodium bica bona e, and ae a ed be o e ci cula ing back o anks. Solid o ganic ma e ial was emo ed wi h a d um il e and phospho ous ma e ial was p ecipi a ed by using polyaluminum chlo ide and polyme , lea ing 2.5% o o al phospho ous. The loccula ed ma e- ial was led o a band il e o emo e he emaining wa e . In he beginning o he depu a ion, he ank o 127 m 3 wi h 5500 kg o ish illed wi h ci cula ion wa e was pu on low h ough. Ri e wa e was led o he depu a ion ank (3 L s -1 ), displacing he ci cula ion wa e o e ime. Tempe a u e o he incoming wa e emained unadjus ed, being nea 0 °C in he win e . The ishes we e no ed du ing he depu a ion. Fish we e sampled om one depu- a ion ank du ing one p ocedu e. Sampling o ish and ci cula ion wa e Sampling o RAS a med Eu opean whi e ish was pe - o med a e 1, 3, 7, 10, 13 and 16 days o depu a ion. Fi e ishes we e aken o each sampling. The ish we e ins an ly ozen a e sampling and s o ed a -24 °C un il mel ed and used o he analyses. On a e age, each ish weigh ed 600 g wi h a slaugh e yield o 90.1%. De- ozen ishes we e ille ed and pieces (15 g ±1 g) o neck, ail, s omach, and ille as desc ibed by Ha hu usingha and Da ey (2016) we e collec ed o analysis, consis ing o 5 ish wi h ou sampling poin s pe ish. Fou pa allel analyses we e pe - o med o each sample piece (n = 5 94). All analyzed pa s o ish consis ed o lesh, while isible pieces o non- edible pa s we e disca ded. Wa e om he depu a ion ank was aken o analysis a e 1, 2, 5 and 8 days. Wa e samples we e collec ed in 1 L high-densi y polye hylene (HDPE) bo les and s o ed ozen a -24 °C be o e he analyses. Pa allel analyses (n = 4) we e pe o med o each sample. Lipid con en s Lipid con en s o each sample piece (neck, ail, belly, and ille ) o ish a e 1, 3, 7, 10, 13 and 16 days o depu a ion we e de e mined by Synlab (Synlab, Analy ics and Se - ices Finland Oy). The acc edi ed me hod (Synlab in e nal me hod 076) included lyophiliza ion combined wi h acid hyd olysis o de e mine he lipid con en o he ish lesh. P e ea men GSM and MIB we e ex ac ed om he sample ma ix by headspace solid phase mic oex ac ion (HS-SPME), mod- i ied om he me hod epo ed by Peng e al. (2014). Fi s , 1 g o ish lesh o 1 mL o wa e was placed in a 10 mL HS ial. A sa u a ed NaCl (aq) solu ion was p epa ed and 750 lL added o he ial. The me hod o s anda d addi ion wi h i e addi ions was used o cons uc calib a ion cu es o quan i ica ion o GSM and MIB in bo h ish and wa e samples. Fou eplica es o each sample we e p epa ed and analyzed o de e mine he s anda d de ia ions. Sealed sample ials we e placed in a wa e ba h a 60 °C. The sep um o he sample ial was pie ced wi h a needle and he DVB/CAR/PDMS ibe was exposed in he headspace o ex ac ion. The ex ac ion ook 30 min be o e in oducing he ibe di ec ly in o he GC–MS injec ion po o deso p ion. Gas ch oma og aphy–mass spec ome y (GC–MS) An Agilen 6890 se ies/5973N GC/MSD (Palo Al o, CA, USA) sys em wi h a Phenomenex Zeb on ZB-5MSi (To - ance, CA, USA) capilla y column (30 m 90.25 mm 9 0.25 lm) was used o sepa a e and de ec he analy es. The empe a u e o he injec o was adjus ed o 270 °C in he spli less mode. The ca ie gas was helium a he low a e o 0.7 mL min -1 . The empe a u e o he o en s a ed a 45 °C o 3 min and inc eased 30 °C min -1 o achie e 300 °C ( o al ime 14.5 min). The elec on impac (EI)-MS condi ions we e selec ed 230 °C o ion sou ce, 5 min delay ime and ionizing ol age o 70 eV. Selec ed ion moni o ing (SIM) mode was used o he de ec ion o GSM and MIB wi h m/z 112, 126, 182 (GSM) and m/z 95, 135, 168 (MIB). Base peak a eas o m/z 95 and 112 we e used o he quan i ica ion o GSM and MIB. Me hod alida ion The analy ical me hod was alida ed in e ms o linea i y, le el o de ec ion (LOD), le el o quan i ica ion (LOQ), J Food Sci Technol 123 sensi i i y, p ecision, and epea abili y. Me hod alida ion was de e mined in e ms o ci cula ion wa e and ish samples. Linea i y Linea i y was s udied by i e s anda d addi ions o GSM and MIB solu ions o ci cula ion wa e , anging 10–50 ng L -1 and 20–60 ng g -1 o ish lesh. Bo h sample ypes showed adequa e co ela ion coe icien s and linea - i y: R 2 0.9877 (GSM) and R 2 0.9878 (MIB) o ci cula ion wa e and R 2 0.9862 (GSM) and R 2 0.9883 (MIB) o ish lesh. Sensi i i y, le els o de ec ion, and quan i ica ion Fo he ch oma og aphic analysis o GSM and MIB, LOD and LOQ we e de e mined based on signal- o-noise a io (S/N) = 3 o LOD and S/N = 10 o LOQ in he SIM mode. The esul s o wa e and ish samples a e lis ed in Table 1. Addi ionally, sensi i i ies we e calcula ed and also lis ed in Table 1. P ecision and epea abili y Yields o s anda d addi ion o GSM and MIB s anda ds a e lis ed in Table 2. Yields ha e been calcula ed bo h o ci cula ion wa e and whi e ish lesh. Addi ions ha e been conduc ed wi h i e eplica i e measu emen s (n = 5) and based on hem, calcula ed he s anda d de ia ions (SDs). Depu a ion model Modelling he depu a ion o Eu opean whi e ish in gene al and he dec ease o GSM and MIB in RAS sys em was pe o med as epo ed by Sch am e al. (2017). The model was p epa ed o show he decline o GSM and MIB in ela ion o e.g. ish biomass, wa e low a e, ank olume, and le els in he wa e . Based on his, he elimina ion a e cons an s we e de e mined o Eu opean whi e ish, which a e gene ic, bu s ill speci ic o he size, lipid con en and empe a u e. Equa ion 1desc ibes he dec ease o o - la- o concen a ions in ish in clean wa e , assuming negli- gible accumula ion o chemicals in he wa e . CBð Þ¼CBð ¼0Þeðk2 Þð1Þ C B( = 0) is he ini ial concen a ion o GSM o MIB in ish in he beginning o he depu a ion pe iod, he ime o depu a ion in hou s, k 2 elimina ion a e cons an o GSM o MIB o he wa e , and C B( ) he concen a ion o GSM o MIB in ish as a unc ion o ime. Ra e cons an k 2 is calcula ed o each body pa ( ille , neck, belly, and ail) by he non-linea eg ession analysis based on he dec ease o concen a ions and he depu a ion ime by Sys a 13.1 (SYSTAT So wa e, Inc. 2009). By inse ing he k 2 alues in Eq. 1and ecalcula ing pe d y lipid con en s, ln- ans o med concen a ions we e ob ained o each body pa . Resul s and discussion GSM and MIB in whi e ish Concen a ions o GSM and MIB in Eu opean whi e ish ille anged om 0 o 32 ng g -1 , dec easing du ing he 16 days o depu a ion (Fig. 1). Be o e he depu a ion pe - iod, concen a ions we e ai ly high compa ed o 0.2–0.9 ng g -1 o MIB and 0.10–0.115 ng g -1 o GSM in A lan ic salmon (S. sala ) epo ed by Da idson e al. (2014). Ha hu usingha and Da ey (2016) s udied ba a- mundi (L. calca i e ) in a ci cula ion sys em wi h 6% (w/ w) body a and ound 0.01–7 ng g -1 GSM and 0.01–7 ng g -1 MIB. On he o he hand, highe concen a- ions o GSM and MIB ha e been epo ed by Zimba e al. (2012). They ound 4.8–19.7 ng g -1 MIB and 0.27–0.59 ng g -1 GSM in RAS a med ainbow ou (O. mykiss). Sch am e al. (2017) s udied depu a ion o Eu o- pean eel (Anguilla anguilla) wi h 33.7% (w/w) a con en and ound high concen a ions o GSM (22 ng g -1 ) be o e he depu a ion. The esul s we e in he same ange wi h hose o his s udy. The o e all a con en o Eu opean whi e ish wi h a e age size o 600 g was ela i ely high, bu dec eased o e he ime o depu a ion (Table 3). Howe e , i is in he same ange wi h ha (11.1%) epo ed by Boi eanu e al. (2016) o a med C. ma aena, weigh ing 290–380 g. Kause e al. (2011) ound 12.2% lipid con en in Eu opean Table 1 Le el o de ec ion (LOD), le el o quan i ica ion (LOQ), and sensi i i y o Eu opean whi e ish (ng g -1 ), and ci cula ion wa e samples (ng L -1 ) Sample ype Compound LOD LOQ Sensi i i y Eu opean whi e ish Co egonus la a e us GSM, ng g -1 1.4 2.1 5.5 MIB, ng g -1 0.5 1.5 0.9 Ci cula ion wa e GSM, ng L -1 2.1 2.8 2.3 MIB, ng L -1 0.8 1.6 1.2 GSM—Geosmin, MIB—2-me hylisobo neol J Food Sci Technol 123 Table 2 Yields o s anda d addi ion o GSM and MIB o ci cula ion wa e (10–50 ng L -1 addi ion) and o Eu opean whi e ish lesh (20–60 ng g -1 addi ion) wi h ela i e s anda d de ia ions (±RSD) GSM (in wa e ) MIB (in wa e ) GSM (in whi e ish) MIB (in whi e ish) S anda d addi ion (ng L -1 ) Yield (%) S anda d addi ion (ng L -1 ) Yield (%) S anda d addi ion (ng g -1 ) Yield (%) S anda d addi ion (ng g -1 ) Yield (%) 10 94.5 ±2.8 10 96.0 ±2.6 20 101.1 ±3.5 20 101.7 ±3.6 20 94.1 ±2.7 20 105.7 ±2.1 30 97.6 ±2.9 30 103.2 ±3.0 30 99.4 ±2.9 30 103.1 ±1.3 40 96.3 ±3.1 40 100.1 ±3.2 40 96.9 ±3.5 40 104.8 ±2.8 50 103.8 ±3.4 50 105.6 ±2.9 50 98.6 ±2.0 50 95.6 ±2.2 60 103.7 ±3.0 60 103.2 ±3.3 GSM—Geosmin, MIB—2-me hylisobo neol Fig. 1 Concen a ions o GSM (a) and MIB (b) in ille , neck, belly, and ail o Eu opean whi e ish lesh du ing he depu a ion o 16 days (±SD) J Food Sci Technol 123 whi e ish, anging in size 540–710 g, while Ai aksinen and Riihima ¨ki (2014) epo ed lipid con en s o abou 9% o a med whi e ish C. la a e us wi h ha es weigh o 995 g. The lipid con en a e 10 days in depu a ion show sudden inc ease in all body pa s o ish, al hough he lipid con en in o al shows up o 50% dec ease du ing he ime o depu a ion. This can be due o a ia ion in ish size a di e en sampling poin s, suppo ed by he inc eased limi s o e o a ha poin . The di e ences in a con en s we e pa ly explained by he size o ish. Howe e , o he ac o s also ha e an e ec on he a con en . Suomela e al. (2017) showed ha he ype o eed has a s ong in luence on he o al lipid con en o Eu opean whi e ish and on he a y acid p o ile o ish. Addi ionally, cul u e condi ions a ec he a y acid com- posi ion as well as wa e empe a u e, becoming mo e unsa u a ed in colde wa e (Haa d 1992). The highes concen a ions o GSM we e de ec ed in ille and belly and o MIB in ille and neck. In all pa s o ish, concen a ions dec eased o e he depu a ion pe iod (Fig. 1). So a , he e a e only ew s udies a ailable ega ding o - la o s and a con en s in di e en pa s o ish. Ha hu usingha and Da ey (2016) ound he highes lipid con en in he belly and he lowes in he ail pa o ba amundi. Howe e , hey did no de e mine concen a- ions o o - la o s in di e en pa s o ish. Pe ci al e al. (2008) s a ed ha GSM and MIB we e no homogeneously dis ibu ed in he ish bu based on lipid concen a ions, while Pe e sen e al. (2011) ound no co ela ion be ween a con en and concen a ions o GSM and MIB, only a posi i e co ela ion wi h he ish size. Pe ci al e al. (2008) also s a ed ha highe le els de ec ed in la ge ish is mos likely caused by he highe lipid con en o la ge ish. In his s udy, he esul s indica ed lowe concen a ions o MIB in he belly, he pa wi h ypically highe lipid con- en s compa ed o ille and ail. In he case o GSM, he highes concen a ions we e de ec ed in ille and belly, and somewha lowe in he neck and ail. Howe e , he e we e no signi ican di e ences be ween he a e age alues o di e en body pa s. I is possible, ha GSM and MIB as lipophilic compounds we e mos ly concen a ed in he a y issue. Only he edible lesh pa s we e s udied and he a y pa s disca ded which migh ha e led o lowe de ec ed concen a ions. On he o he hand, his could be due o di e en species o ish and la ge size (Pe ci al e al. 2008). The concen a ions o GSM we e mos ly below he de ec ion limi s a e 13 days o depu a ion (Fig. 1), while in he s udy o Robe son e al. (2005), he concen a ions dec eased apidly a i s (24 h) and hen slowe du ing he es o he o al 168 h o depu a ion. In he case o GSM, howe e , he e is a mo e p onounced dec ease in concen- a ions a e he i s 7 days (Fig. 1). On he o he hand, Da idson e al. (2014) epo ed a slow dec ease in MIB concen a ion du ing he i s 6 days and an inc eased a e o dec ease owa ds 10 days o depu a ion. The depu a ion pe o mance and ime o depu a ion depend on se e al ac o s, such as ini ial o - la o concen a ion in ish, wa e empe a u e, ish densi y, wa e enewal a e, and he o - la o concen a ion in he depu a ion sys em supply wa e (D ake e al. 2010; Howga e 2004), leading o a ying a es o depu a ion. The lowes ini ial le els o MIB we e ound in he belly, e en hough i con ained he highes amoun o a . This may be explained by he ac ha he lipophilic GSM and MIB ind hei ways especially in o he a issue, while bo h lesh and a we e chosen o he chemical analysis. Howe e , he a e o emo al in he belly was simila o hose in ille , neck, and ail. The a e cons an k 2 o he dec ease o MIB was 0.007 h -1 o neck, 0.007 h -1 o belly, 0.005 h -1 o il- le , and 0.008 h -1 o ail. In he case o GSM, he a e cons an k 2 was 0.008 h -1 o neck, 0.008 h -1 o belly, 0.008 h -1 o ille , and 0.007 h -1 o ail. These we e somewha lowe han he a e cons an k 2 0.014 h -1 epo ed by Sch am e al. (2017), desc ibing mo e e icien depu a ion. This is consis en wi h he slow change o ci cula ing wa e o clean wa e in he beginning o he depu a ion. The dec ease o GSM and MIB concen a ions o e he ime o depu a ion o each body pa has been p esen ed Fig. 2(ln GSM ng g -1 lipid). Fi s , he a e cons an k 2 was calcula ed o ille , neck, belly, and ail o bo h GSM and MIB (Supplemen a y ile S1), which we e used in Table 3 Lipid con en s (w/w, ±limi o e o , %) o Eu opean whi e ish Co egonus la a e us a e 1, 3, 7, 10, 13 and 16 days o depu a ion Depu a ion ime, days Lipid con en pe d y lesh (%) Neck Fille Belly Tail 1 24.9 ±2.5 15.4 ±1.5 30.1 ±3.0 19.5 ±2.0 3 18.0 ±1.8 11.7 ±1.1 15.6 ±1.6 15.6 ±1.6 7 17.5 ±1.8 11.4 ±1.1 17.0 ±1.7 10.1 ±1.0 10 21.2 ±2.1 13.8 ±1.4 22.7 ±2.3 12.2 ±1.2 13 13.0 ±1.3 9.8 ±1.0 17.3 ±1.7 9.3 ±1.9 16 12.8 ±1.3 9.4 ±1.9 13.3 ±1.3 9.7 ±1.0 J Food Sci Technol 123 ecalcula ing he decease pe lipid con en . The a e o emo al in di e en body pa s p oceeds a simila a es. The ini ial concen a ions in ille we e highe han in he o he body pa s, despi e he ac ha he highes lipid con en s we e ini ially ound in he belly and neck (Table 3). The esul s also show ha e en a e 16 days o depu a ion, in some cases, low le els o o - la o com- pounds emained in he ish lesh. GSM and MIB in wa e Concen a ions in wa e dec eased du ing he depu a ion ime, as shown in Fig. 3. This is in ag eemen wi h he ac ha a his pa icula a m, he ci cula ion wa e was g adually changed o clean wa e du ing he ime o depu a ion. In he beginning, he concen a ions we e ai ly high, on a e age 128 ng L -1 o GSM and 94 ng L -1 o MIB. The esul s can be conside ed ypical o ci cula ion wa e o a RAS a m, e en hough he concen a ions a y widely among RAS a ms based on ope a ional choices, p ocess condi ions, and wa e quali y (Ha hu usingha and Da ey 2016). GSM and MIB a e assumed o be exchanged be ween wa e and ish by passi e di usion ia he gills. The depu a ion p ocess, d i en by he ugaci y and chemical po en ial, seeks equilib ium be ween he lipid ac ions o ish and wa e , eaching equilib ium as he ugaci ies in lipids and wa e a e equal (Howga e 2004). E en hough Fig. 2 Ln ans o med GSM (a) and MIB (b) concen a ions in ille , neck, belly, and ail o Eu opean whi e ish (ng g -1 lipid) as a unc ion o depu a ion ime, 1–16 days (h). Dashed line desc ibes he concen a ion o de ec ion limi s o GSM o MIB based on senso y and analy ical me hods J Food Sci Technol 123 GSM and MIB we e exc e ed om he ish du ing he depu a ion pe iod, he olume o wa e is e y la ge compa ed o inc ease in GSM and MIB le els and ypi- cally, no ne inc ease in hei concen a ions could be de ec ed (Sch am e al. 2017). Howe e , e y high ish densi ies and low wa e exchange a es can lead o inc eased le els o compounds exc e ed by he ish (Sch am e al. 2009). In his s udy, he ish load du ing he depu a ion was a abou 43 kg m -3 . The esul s showed a dec ease in concen a ions o GSM and MIB bo h in ish and wa e (Figs. 1and 3), sugges ing an adequa e olume o pu e wa e o he depu a ion p ocedu e. The concen- a ions de ec ed in he depu a ion wa e mos ly o igina ed om he ci cula ion wa e used o ill he depu a ion ank a he beginning o he depu a ion pe iod, and GSM and MIB eleased by he ish. Addi ionally, bio ans o ma ion o GSM and MIB o hei deg ada ion p oduc s migh occu (Sch am e al. 2018). Conclusion This s udy is he i s o epo GSM and MIB concen a- ions in Eu opean whi e ish lesh and hei dis ibu ion in di e en pa s o ish. The le els ound in Eu opean whi e ish we e likely caused by he high le els in ci cula- ion wa e and accumula ion in o he ish. In he u u e, s udies ega ding he accumula ion in di e en species unde simila condi ions can gi e mo e insigh . The gene al depu a ion model was buil o p edic he emo al o o - la o s om Eu opean whi e ish mo e gene ally, wi h he e ec s o p ocess condi ions o his pa icula plan in mino pa . The esul s show he g ea e ec o a con en in Eu opean whi e ish lesh. The dec ease p oceeds ai ly simila ly in all body pa s, e en hough he ini ial concen a ions we e di e en . Addi ion- ally, he a con en s dec eased up o 50% om hei o iginal alues du ing he depu a ion, showing he e ec o as ing. The esul s show ha depu a ion up o 16 days is equi ed o MIB, bu o GSM sho e ime seems su i- cien . In his s udy, ci cula ion wa e was g adually chan- ged o esh wa e du ing he depu a ion pe iod, inc easing he ime equi ed o ull emo al o o - la o s. Op i- miza ion o he depu a ion ime is c ucial in o de o educe he p oduc ion cos s bu s ill p oduce he high quali y ish p oduc s. The impo ance o depu a ion is emphasized pa icula ly, when p oducing mild- la o ed Eu opean whi e ish which mus no p o ide any o - la o sensa ion, no e en o he mos sensi i e consume . Acknowledgemen Open access unding p o ided by Na u al Resou ces Ins i u e Finland (LUKE). Financial suppo by he Eu o- pean Ma i ime and Fishe ies Fund is g a e ully acknowledged. Au ho con ibu ions The expe imen was planned by JV and PCL- L. HP was consul ed ega ding he me hod de elopmen and he analy ical wo k. PCL-L conduc ed he sample p epa a ion, analyses, and d a ed he manusc ip . RA, HP and JV c i ically examined and e ised he manusc ip . Con lic o in e es The au ho s decla e ha hey ha e no con lic s o in e es . Fig. 3 Concen a ions o GSM and MIB in depu a ion ank du ing 1, 2, 5 and 8 days, n = 4 (±SD) J Food Sci Technol 123