Ecology and E olu ion. 2017;1–13.
|
1
www.ecole ol.o g
Recei ed:28June2016
|
Re ised:18Janua y2017
|
Accep ed:7Feb ua y2017
DOI: 10.1002/ece3.2841
ORIGINAL RESEARCH
Oxygen and ca bon isoscapes o he Bal ic Sea: Tes ing hei
applicabili y in ish mig a ion s udies
Jy ki To niainen1 | Anssi Lensu2 | Pekka J. Vuo inen3 | Eloni Sonninen4 |
Ma ja Keinänen3 | Roge I. Jones2 | William P. Pa e son5 | Mikko Kiljunen2
Thisisanopenaccessa icleunde he e mso heC ea i eCommonsA ibu ionLicense,whichpe mi suse,dis ibu ionand ep oduc ioninanymedium,
p o ided heo iginalwo kisp ope lyci ed.
©2017TheAu ho s.Ecology and E olu ionpublishedbyJohnWiley&SonsL d.
1Na u alHis o yMuseum,Uni e si yo
Jy askyla,Jy askyla,Finland
2Depa men o BiologicalandEn i onmen al
Science,Uni e si yo Jy askyla,Jy askyla,
Finland
3Na u alResou cesIns i u eFinland,Helsinki,
Finland
4Labo a o yo Ch onology,Uni e si yo
Helsinki,Helsinki,Finland
5Saska chewanIso ope
Labo a o y,Depa men o Geological
Sciences,Uni e si yo Saska chewan,
Saska oon,SK,Canada
Co espondence
Jy kiTo niainen,Na u alHis o yMuseum,
Uni e si yo Jy askyla,Jy askyla,Finland.
Email:[email p o ec ed]
Funding in o ma ion
MajandTo NesslingFounda ion,G an /
Awa dNumbe :2010150,2011105,2012506
and2013040;Academyo Finland,G an /
Awa dNumbe :34139
Abs ac
Con en ional agsapplied oindi idualsha ebeenused oin es iga eanimalmo e-
men ,bu heseme hods equi e aggedindi idualsbe ecap u ed.Mapso egional
iso opic a iabili yknownas“isoscapes”o e po en ial o a iousapplica ionsinmi-
g a ion esea chwi hou aggingwhe einiso ope alueso issuesa ecompa ed o
en i onmen aliso ope alues.In hiss udy,wep esen hespa ial a iabili yinoxygen
(
δ18
OH
2O
)anddissol edino ganicca bon(δ13CDIC)iso ope alueso Bal icSeawa e .
Wealsop o ideanexampleo how heseisoscapescan e ealloca ionso indi idual
animal iaspa ialp obabili ysu acemaps,using hehigh- esolu ionsalmono oli h
iso opeda a omsalmondu ing hei sea- eedingphasein heBal icSea.Aclea la i-
udinaland e icalg adien was ound o bo h
δ18
OH
2O
andδ13CDIC alues.Thedi -
e encebe weensumme andwin e in heBal icSea
δ18
OH
2O
alueswasonlysligh ,
whe easδ13CDIC aluesexhibi edsubs an ialseasonal a iabili y ela ed oalgalp o-
duc i i y.Salmono oli hδ18Oo oandδ13Co o aluesshowedclea di e encesbe ween
eedinga easandseasons.Ou exampledemons a es ha dualiso opeapp oacho -
e sg ea po en ial o es ima ingp obable ishhabi a sonceissuesinmodelpa am-
e e iza ionha ebeen esol ed.
KEYWORDS
iso opiclandscape,mic omilling,modele alua ion,Salmo sala ,spa ialassignmen ,
spa ialin e pola ion
1 | INTRODUCTION
Se e al ma king app oaches ha e been employed o add ess ques-
ionsinmig a ionecology.Un il ecen ly,con en ionalex insicma k-
e s(i.e., ags)applied oindi idualsha ebeenused oin es iga e hei
mo emen (Lucas&Ba as,2000),bu heseme hods equi e agged
indi idualsbe ecap u ed oacqui espa ialin o ma ion.Du ing ecen
yea ssomein es iga ionsha ebeenconduc ed os udylong- e m
mo emen so indi idualadul A lan icsalmon(Salmo sala L.;Figu e1)
in heseausing ags ha eco den i onmen alcha ac e is icsalong
hemig a ion ou es(e.g.,Chi enden,Ådlands ik,Pede sen,Righ on,
&Rika dsen,2013).Howe e ,due o hep esen sizeo he ags, he
s udied ishha e obela geand, he e o e,dono necessa ily ep e-
sen hemajo i yo hepopula ion.
In insicbiochemicalma ke ssuchass ableiso opescanp o ide
anal e na i eapp oach o ackindi idualmo emen so e la gegeo-
g aphical dis ances such as be ween con inen s (Hobson & No is,
2008).All animals a e iso opically ma ked by he en i onmen hey
li einandby hei die .Theassignmen o anindi idual oace ain
a eawo ksbyes ima ingp obabili ieso occu ence o animalindi-
idualsbycompa ing aluesob ained om issuesamples oiso opic
landscapes(i.e.,isoscapes;Wunde ,2010).Ag ea ad an ageinusing
2
|
TORNIAINEN ET Al.
biochemical ma ke s is ha hey can be linked o hose indi iduals
ha ac ually su i ed he mig a ion o hei b eeding habi a s, and
he e o e,be e ep esen hepopula ion.As isho oli hsa ealmos
comple ely mine alized om he ca bona e o he en i onmen al
wa e (Kim,O′Neil,Hillai e-Ma cel,&Mucci,2007;Pa e son,Smi h,
&Lohmann,1993;Solomone al.,2006),analysisandcompa isono
o oli handwa e s ableiso opescan e eal heloca ionswhe e he
o oli ho anindi idual ishis o med.Howe e ,i hechosen issues/
ma e ials a e dissimila , spa ial assignmen (ma ching o issue and
sou ceiso ope alues)needs ac iona ionequa ionsbe ween hecho-
sen issueandsou ceiso ope aluesdue o hedi e en ac iona ion
o heelemen iso opes iaen i onmen alandphysiological ac o s.
Isoscapeso e po en ial o a iousapplica ionsinen i onmen al
andecological esea ch(Bowen,Wassenaa ,&Hobson,2005;Dawson
&Siegwol ,2007;Hobson&Wassenaa ,2008),whe eby heiso ope
alueso selec ed issuesa e ela ed oen i onmen aliso ope alues,
andno jus oneelemen ,bu se e alelemen s(i.e.,mul i-iso opeiso-
scape;e.g.,Hobsone al.,2012;Ga cía-Pé ez&Hobson,2014).The
mos ambi iousapp oachesha ebeen hecons uc ionso isoscapes
onaglobalscale(e.g.,Amundsone al.,2003;Bowen&Re enaugh,
2003;LeG ande&Schmid ,2006),whe ese e als udiesha eled o
con incing esul s ega dinganimalmig a ionamongmany axain e -
es ialandaqua icen i onmen s(Bes &Schell,1996;Chambe lain,
Bensch, Åkesson, & Ande sson, 2000; Hanson, Wu s e , EIMF, &
Todd,2010,2013;Hobson&Wassenaa ,1996;Wassenaa &Hobson,
1998).Howe e ,al houghin o ma ionabou animalmo emen sand
spa ialusageo hei habi a sa hein e con inen alscaleisimpo an
inecological esea chandconse a ion,manyc uciale en soccu also
a smalle scaleswi hin hedis ibu iono smalle dis ancemig an s.
Un o una ely, hea ailabili yo isoscapeda a o s udiesa amo e
localspa ialscaleappea s obespa seo hedis anceo hesu ey
s a ionso globalisoscapeda amaybe oola ge o adequa elocal
iso opicdisc imina ion(seeBowen&Re enaugh, 2003).The e o e,
addi ionaliso opicda aa eneeded oinc ease he esolu iono global
isoscapes oenablemo ep ecise econs uc iono animalloca ions
andmo emen .
Theaimso hiss udywe e(1) op o ideho izon aland e ical
iso opicg iddedda ase s(i.e.,isoscapes)o oxygen(
δ18O
H
2O
)anddis-
sol edino ganicca bon(δ13CDIC) o hewa e o heBal icSea.(2)
Asanexamplewedemons a e hepo en ialo heseisoscapesusing
woA lan icsalmon indi iduals om he Ri e Simojoki.Combined
wi h heiso opeda a om hesalmonandspa ialp obabili ysu ace
maps,weshowp obableloca ionso indi idual ishin a ious ime
poin sdu ing hei sea- eedingmig a ionphasein heBal icSea.We
alsodemons a ehowpa ame e iza iono hemodelsin luenceson
loca iones ima es.
2 | MATERIAL AND METHODS
2.1 | Sampling and iso ope analyses o wa e
Bal icSeawa e sampleswe ecollec eddu ing h eedi e en c uises
by heR/VA andao heFinnishEn i onmen Ins i u e(SYKE).To
e alua eapossibleseasonalimpac onseawa e iso ope alues,we
collec ed wosumme se sandonewin e se o samples.Theda es
o hec uiseswe e(1)24May o11June2010,(2)9Augus o27
Augus 2010,and(3)17Janua y o3Feb ua y2011.The i s c uise
co e ed hewholeBal icSea om heBo hnianBay o heSou he n
Bal icP ope ,excluding heGul o Finland(Figu e2),whichwas he
onlya easampleddu ing hesecondc uise.The hi dc uise,inwin-
e ,co e ed heBal icSeaexcep a eassou ho Go land.Seawa e
wassampleda adep ho 10m ome e ysamplings a ion(al o-
ge he 316samples om134s a ion isi s,blackdo sinFigu es3
and4).F om25s a ions,wa e wasalsosampled e icallya 5–50m
FIGURE1 A lan icsalmon(Salmo sala )
FIGURE2 Mapo heBal icSeawi h heloca iono heRi e
Simojoki(uppe mos igh co ne ).Thema kedpa hwaysshow
ansec sused o c ea ing e icalin e pola ionsinFigu e4
|
3
TORNIAINEN ET Al.
in e alsdependingonwa e dep ha hesi e(Figu e4).Maximum
dis ancesbe weensamplings a ionswe eless han100km.Wa e
samples we e aken using a CTD/Rose e sample (Rose e 1015,
Seabi d,SBE911/Gene alOceanics,SIS:Plus500).
Rose esample eco dsconduc i i y, omwhichsalini ywasau-
oma icallycalcula ed.F ome e ysamplingoccasiono wa e o iso-
opeanalysispu pose,salini ywas eco ded.Samplewa e collec ed
o δ13CDICanalysis,we einjec edin o12-mlbo osilica eExe aine
ials(ca .no438B;LabcoL d.,HighWycombe,UK)p epa edin he
labo a o y,whe e0.2mlo 85%o hophospho icacid(H3PO4)was
addedin oeach ial,whichwas hensealedwi hacap(con aininga
ubbe sep um)and lushedand illedwi haheliuma mosphe e.In
he ield2–4mlo seawa e omeachs a ionandsampledep hwas
injec ed h ough he ubbe sep umin o he ial.Samplewa e o
δ18O
H
2O
analysiswe ecollec edin20-mlglassscin illa ion ialsand
illed ullensu ingnoai bubbles.Allsampleswe es o edinanunli
FIGURE3 In e pola edmaps o he
Bal icSeao
δ18
OH
2O
alues(a,b)and
δ13CDIC alues(c,d) om10me e sin
summe (a,c)andwin e (b,d).Blackdo s
ep esen samplingloca ions
(a)
(c)
(b)
(d)
Longi ude(E)
La i ude(N)
–7
–8
–9
–10
2
1
0
–1
–2
510 15 20 25 30 35 510 15 20 25 30 35
66
64
62
60
58
56
54
66
64
62
60
58
56
54
Summe Win e
δ18O(‰)δ
13C(‰)
FIGURE4 In e pola edc oss-sec ional
su aceso e ical
δ18
OH
2O
andδ13CDIC
alues.
δ18
OH
2O
alues omsou h ono h
in(a)summe and(b)win e ,and omwes
oeas in(c)summe and(d)win e .δ13CDIC
alues, omsou h ono hin(e)summe
and( )win e ,and omwes oeas in
(g)summe and(h)win e .Blackdo s
ep esen samplingloca ionsanddep hs
(a) (b)
(c) (d)
(e)
(g)
( )
(h)
δ18O (‰)δ
13C (‰)
Dep h(m)
0
100
200
300
0
50
100
150
–5
–6
–7
–8
–9
–10
–11
–5
–6
–7
–8
–9
–10
–11
20 22 24 26 28 20 22 24 26 28
56 58 60 62 64 56 58 60 62 64
Longi ude(E)
2
1
0
–1
–2
–3
–4
–5
2
1
0
–1
–2
–3
–4
–5
0
100
200
300
0
50
100
150
20 22 24 26 28 20 22 24 26 28
56 58 60 62 64 56 58 60 62 64
La i ude(N)
Longi ude(E)
La i ude(N)
Summe Win e
4
|
TORNIAINEN ET Al.
e ige a o (+4°C)inada k oompendinglabo a o yanalysis.In he
labo a o y,simila ialsasused o δ13CDICsampleswe e illedwi h
0.5mlseawa e andequilib a edwi hCO2 o a leas 24h a 25°C.
Analyseso sampless a edwi hinaweeka e a i al o helabo a-
o y, i s heδ13CDICsamplesimmedia elya e a i al.Theδ13CDIC
aluesa eexp essed ela i e oVPDB(ViennaPeeDeeBelemni e),
measu ed and calib a ed/no malized agains in e na ional IAEA
(In e na ionalA omicEne gyAgency[IAEA])s anda dsNBS19(o TS-
Limes one;Calciumca bona e,δ13C=+1.95‰)andLSVEC(Li hium
ca bona e,δ13C=−46.6‰);±0.14‰.Valuesa eexp essed ela i e o
VSMOWandcalib a ed/no malized o heVSMOW(ViennaS anda d
Mean Ocean Wa e [VSMOW], δ18O=0‰)—SLAP (S anda d Ligh
An a c icP ecipi a ion,δ18O=−55.5‰)scale;±0.1‰.Allwa e sam-
pleswe eanalyzed o bo hδ13CDICand
δ18O
H
2O
a heLabo a o y
o Ch onology, Finnish Museum o Na u al His o y, Uni e si y o
Helsinki, using a GasBench II and Del a Plus XL (The mo Fishe
Scien i ic,B emen,Ge many).
S a is ical es ingo seasonaldi e encesin
δ18O
H
2O
andδ13CDIC
alues(pai edsamples es )waspe o medusingPASWS a is ics18
o Windows(SPSSInc.,Chicago,IL,USA).In e pola edisoscapes o
δ13CDICand
δ18O
H
2O
we econs uc edusingOceanDa aView(ODV)
so wa e e sion4.5.5(Schli ze ,2002,2011).
2.2 | O oli h sampling, mic omilling and s able
iso ope analysis
In o de o es applicabili y o Bal ic Sea isoscapes in ish mig a-
ions udies,weanalyzedo oli hs om woexample emaleA lan ic
salmon(he ea e Bal icsalmono salmon)o iginallycaugh ,as hey
we e e u ning o heRi e Simojoki(Figu e2) ospawnin2008,by
heNa u alResou cesIns i u eFinland(LUKE)aspa o hep og am
omoni o yolk-sac ymo ali y(M74synd ome:e.g.,Keinänene al.,
2012; o iginal LUKE code numbe s SS5447 and SS5461, he ea e
FISH1and2, espec i ely).Bo hsalmonhadspen woyea s eed-
ingin he sea.Theo igin (wild o ha che y- ea ed) andage o he
salmonwasde e mined om hescalenucleusandscaleg ow hpa -
e n(Hiili i a,Ikonen,&Lappalainen,1998).FISH1waso wildo igin
(5,300g,80cm)andFISH2waso ha che y- ea edo igin(6,300g,
81cm). Bo h sagi al o oli hs we e emo ed om he head o he
salmon,cleanedindeionizedwa e o emo eany emainingo ganic
issue,andd iedo e nigh a 60°C.
Bo ho oli hswe esampled o δ13Candδ18Oanalysisusing he
cus om-buil h ee-dimensional compu e -con olled mic omilling
sys emin heSaska chewanIso opeLabo a o ya heUni e si yo
Saska chewan ollowing he p ocedu e o Wu s e , Pa e son, and
Chea ham(1999).Thissys emallowed31–32samplingpa hs obe
ollowed conco dan wi h g ow h banding in bo h o oli hs. Iso ope
a ios o samples we e de e mined using a Finnigan MAT 253 di-
ec lycoupled oaKiel-IVau oma edca bona ep epa a ionde ice
(The mo–Fishe Scien i ic).Accu acy and p ecision we e moni o ed
by ou ineanalysiso NBS-19s anda d,yieldingas anda dde ia ion
o eplica es anda ds ha wasconsis en lyless han0.09 o bo h
δ13Oandδ18O alues.Allo oli hiso opemeasu emen sa e epo ed
in hes anda ddel ano a ion(pe mil) ela i e o heVPDBs anda d
asδ18Oo oandδ13Co o.F om heo oli hda ao eachsalmon, h ee
samplemillingpa hswe eselec ed o close inspec ion,namely1s
seawin e (1SW), he ollowingsumme (2SS),and he2ndseawin e
(2SW) oassign hesalmon oBal icSeaa eas o eachpe iod.Du ing
heseamig a ionphase,Bal icsalmonexpe iencehighseasonal luc-
ua ions in ambien empe a u e.These pe iods a e clea ly seen in
heo oli hs,aso oli hca bona ewi h he highes δ18Oo o aluesis
acc e eddu ingcoldwin e pe iod,while helowes δ18Oo o alues
ep esen highes empe a u esin hesumme (Figu e6;Wu s e &
Pa e son,2003).
2.3 | C ea ing iso opic (isoscapes) and empe a u e
maps o he Bal ic Sea
Isoscapes o δ13CDICand
δ18O
H
2O
aswellas empe a u emapswe e
cons uc edusingODVso wa e.Usedmapda a o ho izon aliso-
opemapswe e ecei ed om heda abaseo heLeibnizIns i u e
o Bal ic Sea Resea ch Wa nemünde (IOW; Sei e , Taube , &
Kayse ,2001)and o bo hho izon almapsand e icalp o iles om
he Gene al Ba hyme ic Cha o he Oceans (GEBCO_08 G id).
GEBCO_08G idhad obecon e edin oNe CDF o ma compa -
iblewi hODVwi hRS a is icsso wa e 3.0.1(RCo eTeam,2013)
using package RNe CDF (Michna, 2012). In e pola ed maps we e
p oducedusingDa aIn e pola ingVa ia ionalAnalysis(DIVA)g id-
dingso wa e(T oupine al.,2012)includedinODV(DIVApa am-
e e s:scaleleng hschosenau oma ically;signal- o-noise a io=40;
quali ylimi =3.0;excludingou lie s).Ve icalp o ileswe ealsoc e-
a edusingDIVAg idding(scaleleng hschosendependingonused
da a;signal- o-noise a io=40;quali ylimi 3.0;excludingou lie s).
Thecoas linesin hemapsa ebasedon heGlobalSel -consis en
Hie a chical High- esolu ion Sho elines da abase 2.1 (Wessel &
Smi h,1996).
As hespa ialco e ageo collec ed
δ18O
H
2O
was ela i elyspa se
o assignmen models,andsomela ge a easlackedmeasu emen s,
wec ea edp edic i emodels o es ima e
δ18O
H
2O
alues om he
Bal icSeawa e salini y(S)da awhichisanexcellen p edic o o
δ18O
H
2O
andiscommonlyused oes ima eoceanseawa e iso ope
alues(e.g.,LeG ande&Schmid ,2006).The
δ18O
H
2O
–S ela ionship
is equen lylinea ly ela ed(LeG ande&Schmid ,2006),bu in he
Bal ic eshwa e in low omla ge i e sin oGul o Bo hniaandGul
o Finlandha eane ec on he ela ionships.The e o e,wec ea ed
sepa a emodels o heGul o Bo hnia, heGul o Finlandand o
he es o he emaininga eao heBal icSea.Shapeo hep edic i e
model o eacha eawasselec edbasedongoodnesso i andmodel
esidualswe ee alua edusingquan ile–quan ileplo s.Modelswe e
u he applied oseawa e salini ysamplingsu eysob ained om
Bal icMa ineEn i onmen P o ec ionCommission(HELCOM)da a-
base(Ande sson,2014) oc ea eseasonal
δ18O
H
2O
maps o heBal ic
Sea.Da aonly o summe (July–Augus )andwin e (Feb ua y–Ma ch)
o hepe iod hes udysalmonhadspen in hesea(2007–2008)
we eselec ed.Aswe ounds a is icallysigni ican seasonaldi e ence
in measu ed
δ18O
H
2O
alues, h ee sepa a e
δ18O
H
2O
maps [win e
|
5
TORNIAINEN ET Al.
2007(1SW),summe 2007(2SS)andwin e 2008(2SW)]we ec e-
a ed obeusedinassignmen models.Modeled alueswe ecombined
wi hmeasu ed alues om hesameseason ogainbe e spa ialco -
e age. Ma ked wi hin-season di e ences we e no ound be ween
HELCOMandou ownmeasu emen sbasedon isual alida iono sa-
lini yda a,ob ained omdep hs5–15m,indica ing ha also
δ18O
H
2O
aluesbeha e hesameway.
Salmon summe empe a u es we e ixed a 11.5°C, calcula ed
omp e e ed ue10ms ayingdep hso salmoninBal icsalmon
da a s o age ag s udy by Wes e be g, S u laugsson, Ikonen, and
Ka lsson(1999).Fo win e empe a u es(2007–2008),weusedall
a ailable empe a u e p o iles collec ed om HELCOM da abase
(Ande sson,2014)co e ing hewholeBal icSea(mean empe a u es
omdep hs5–15m)andc ea edanin e pola edsu ace o hep e-
e ed10ms ayingdep h.In e pola edme abolicallyde i edbica -
bona eδ13Cdie alueswe ede i ed omδ13C alueso salmondie a y
speciesa ound heBal icSea(AppendixS3).
2.4 | C ea ion o o oli h- ela ed isoscapes om
δ18OH2O, δ13CDIC, δ13Cdie and wa e empe a u e
Unde s anding he dependence o ambien wa e empe a u e and
δ18O
H
2O
alues on δ18Oo o is necessa y o ob ain compa able al-
ues o δ18Oo o and
δ18O
H
2O
o spa ial assignmen o salmon (e.g.,
Pa e sone al.,1993).Mo eo e , heδ13Co o alueisamixo bica -
bona eδ13CDIC omambien wa e andme abolicallyde i edbica -
bona e δ13Cdie (Solomon e al., 2006; Wu s e & Pa e son, 2003).
We he e o eco ec ed
δ18O
H
2O
andδ13CDIC aluesusingli e a u e-
de i ed ac iona ionequa ions,a e which heco ec ed aluesa e
conside edas“o oli hisoscape”o bo helemen s(AppendixS1).This
pe mi sadi ec compa isono o oli handBal icSeawa e iso opes,
he ebyp o idingap obabilis icspa ialassignmen o salmondu ing
hei sea- eedingphase.
Ini ially,
δ18O
H
2O
alues a e calcula ed ela i e o he VSMOW
scaleandδ18Oo o alues ela i e o heVPDBscale om heIAEA.
To enable di ec compa ison o δ18Oo o(VPDB) wi h δ18Owa e (VSMOW),
δ18Owa e (VSMOW) alues we e con e ed o VPDB using ollowing
equa ion( omCla k&F i z,1997):
Oxygen iso ope alues o o oli hs e lec hose o he ambien
wa e (Campana,1999;Fa ell&Campana,1996;Tho old,Jones,&
Campana, 1997), wi h a empe a u e-dependen ac iona ion (e.g.,
Pa e sone al.,1993).Followingcommonp ac ice,weused helinea
empe a u e-dependen ac iona ion(e.g.,Pa e sone al.,1993).
whe e Tis empe a u e(103/K),whe eKisambien wa e empe a-
u einKel in,andpa ame e α is he ac iona ion ac o be ween
o oli h and ambien wa e [α=(δ18Oo o(VPDB)+1,000)/(
δ18O
H
2O
(VPDB)+1,000)]. Model ac iona ion cons an s a and b in ou p e-
ious s udies co e ing salmonid ishes a e a iable (a=−41.14,
b=20.43:Godiksene al.,2010;a=−33.43,b=17.88:Hansone al.,
2013; a=−33.49, b=18.56:Pa e sone al., 1993and a=−41.69,
b=20.69:S o m-Suke,Dempson,Reis ,&Powe ,2007).Toe alua e
hee ec and a ia iono di e en pa ame e s osalmonassignmen ,
henume icallymos dis an pa ame e s omeacho he (i.e.,Hanson
e al.,2013;Pa e sone al.,1993)we eselec ed o salmonassign-
men (i.e., o p obabili ysu aces),he ea e Model3andModel1,
espec i ely.Alsoana e ageo all ou models(a = −37.44,b = 19.39)
wascalcula edandusedacco dingly,he ea e Model2(AppendixS2).
The ollowingequa iongi es heo oli h- ela edδ18O(VPDB)isos-
cape aluesin ela ion owa e empe a u e(T)and
δ18O
H
2O
(VPDB) al-
ues, ha canbecompa ed oδ18Oo o aluesenabling heassignmen
o salmon oseaa easbasedon hei o oli hiso ope alues:
Ano oli h- ela edδ13Co oisoscapewascalcula edas ollows(e.g.,
Solomone al.,2006;Wu s e &Pa e son,2003) oallowcompa ison
be weenδ13CDICandδ13Co o:
whe e δ13Cdie isameanδ13C alueo salmonp ima yp eyspeciesin
heBal icSea(sp a [Sp a us sp a us],Bal iche ing[Clupea ha engus
memb as]and h ee-spineds ickleback[Gas e os eus aculea us])ina
pa icula loca ion(AppendixS3)andMis hep opo iono me abolic
ca bonin hesalmono oli h(She wood&Rose,2003):
whe e Kcaud(A lan icsalmoncaudal in a io;Minns,King,&Po ,
1993)is2.4.
2.5 | Salmon assignmen using o oli h and wa e
iso ope alues
Toes ima e heloca ionso salmonindi idualsin hei 1SW,2SS
and second 2SW om δ18Oo o and δ13Co o alues, we calcula ed
p obabili ydensi ysu aces o eachsalmonbyusingade e min-
is icg idco e ing heBal icSea ollowing heapp oachp esen ed
inWunde (2010)wi hRS a is icsso wa e 3.0.1(RCo eTeam
2013).We i s had o ein e pola eallDIVAin e pola ion esul s
(wa e and die -basediso opes, empe a u es,e c.which we ein
di e en kindso nonde e minis icg ids) omODVin oade e -
minis icg id oenableus ocalcula e he ac iona ionequa ions
o allg idloca ions.This ein e pola ionwaspe o medwi hlocal
(Nmax =4) in e se-dis ance weigh ing in e pola ion a ailable in
heRpackagegs a (Pebesma,2004).Thep obabili ydensi ysu -
ace calcula ion app oach assumes wa e - empe a u e-dependen
δ18O(VPDB) aluesandDIC-die -basedδ13C alues obeno mallydis-
ibu eda eachg idpoin and hencalcula es hep obabili ieso
ob aining hemeasu ed alueso heo oli h omno maldis ibu-
ionswi hmeanpa ame e sacco ding o heVPDB-co ec edin e -
pola ediso ope alues.S anda dde ia ionso δ18Oandδ13C alues
we ees ima edbasedonobse ed a ia ionso
δ18
O
H2O
(ɛ=0.019)
andδ13CDIC(ɛ=0.216) aluesob ained om hesameloca ionand
dep h,and he a ia iono obse edδ18Oo o aluesin heo oli hs
δ18
O
H2O(VPDB)
=0.97002×δ
18
O
H2O(VSMOW)
−
29.98.
1,000×ln α=a+b∕T,
δ18
O
(VPDB)
=e
(
a+b∕T
)
∕
1,000
×(δ
18
O
H2O(VPDB)
+1,000)−
1,000.
δ13
C
o o =
M
×δ13
C
die +
(1
−
M)
×δ13
C
DIC,
M=0.025+0.066×Kcaud,
6
|
TORNIAINEN ET Al.
o se e al ishindi iduals(ɛ=0.207es ima ed omGodiksene al.,
2010).
The esul ing p obabili y su aces do no ep esen wo-
dimensionalp obabili ysu aces( o alp obabili yineachmapsisno
scaled obe1)and heminimumandmaximump obabili y aluesa e
no hesameinallmapsasiscommonin hiskindo app oach(see
e.g.,Wunde ,2010).Ins ead, hesesu aces isualizewhe e hep ob-
abili yo ob aining hemeasu ed alueo heo oli h(o hep obabili y
o p esenceo pa icula indi iduala ace ain ime)is heg ea es ,
andwhe ei is(much)lowe .In hosemapswhe ebo hoxygenand
ca bon iso opes ha e been aken in o accoun , p obabili ies we e
calcula ed by mul iplying iso ope-wise p obabili ies, assuming inde-
pendence,whichmayno bes ic ly ue o hiskindo phenome-
non.Howe e ,ou da adidno allow o ulles ima iono co a iance
be ween ueda a aluesdue opa iallydi e ingmeasu emen o
obse a ion loca ions. Ne e heless, he co a iance o in e pola ed
su aceswasabou 10- oldsmalle han he a ianceso indi idual
iso opes.The e o e,weexpec ha anypossiblee o in he esul s,
due odependenceiniso ope alues,issmall.
Tos udy,howsensi i eou app oachis omeasu emen e o s
ela ed o hes ableiso opes,weconduc edasimplesensi i i yanal-
ysisbyaddingand/o sub ac ing heobse eds anda dde ia ionso
wa e samples(basedonli e a u e)o δ13Candδ18O o/ om heac-
ualmeasu ed alues,andpe o med hecalcula iono hep obabili y
o p esence o FISH2on he i s win e (1SW)using hemodelby
Hansone al.(2013) o ac iona ion.Theamoun ±SDo wa e iso-
opemeasu emen a ia ion(basedonli e a u e)waschosenbecause
i also e lec sabou ±2SDo ou ownes ima eo measu emen ac-
cu acy(0.09)o o oli hiso ope alues.So, heselimi sa ealso a he
close obeingabou 95%con idencein e alo hemeasu ediso ope
alues.
3 | RESULTS
3.1 | δ18OH2O, and δ13CDIC alues in he Bal ic Sea
Di e encein
δ18O
H
2O
aluesbe weenseasonsa 10mdep ha he
same loca ions was s a is ically signi ican (Pai ed samples T es :
N = 31,p < .001).Howe e , hemean(±SD) alueso
δ18O
H
2O
we e
e y simila (meansumme : −7.9‰±0.83, meanwin e : −7.6‰±0.74).
Inaddi ion,
δ18O
H
2O
aluesshowedaclea sou h ono hla i udinal
dec ease oma ound−6‰ oa ound−10‰(Figu e3a,b).Aweake
longi udinaldec ease oma ound−6‰ oa ound−7.5‰occu ed
om heeas e nendo heGul o Finland o hewes e nsideo he
No he nBal icP ope (Figu e3a,b).Dis inc di e encesin
δ18O
H
2O
aluesbe weenadjacen basinswe ealsoobse ed(Bo hnianBay—
Bo hnianSea—Bal icP ope —Gul o Finland),bu heBo hnianSea
and heGul o Finland,e en houghno di ec lyconnec ed,had e y
simila alues(Figu e3a,b)due o hein luxo eshwa e om i e s.
Mean (±SD) δ13CDIC alues we e signi ican ly di e en (Pai ed
samplesT es :N = 31,p < .001)be weensampling imes(meansum-
me :0.3‰±0.81,meanwin e :−0.9‰±0.54).δ13CDIC aluesdec eased
wi hla i udeinsumme oma ound2‰ oa ound−2‰(Figu e3c).
The di e encewas smalle in hewin e (Figu e3d).An inc easing
endwasobse edinδ13CDIC alues om heGul o Finland o he
Bal icP ope inbo hseasons, oma ound−1‰ oa ound2‰in
summe and oma ound−2‰ oa ound0‰inwin e (Figu e3c,d).
Ve icalin e pola iono he ansec sshowed ha
δ18O
H
2O
aluesin
heCen alBal icP ope we ehighe below~50m oma ound−6.5‰
oa ound−5.5‰a hebo omandashighas−4‰in heSou he n
Bal icP ope (Figu e4a,b),whe easaclea dec easing endinsumme
δ13CDIC alues oma ound−1‰ oa ound−5‰a hebo omwas
obse ed (Figu e4e, ). Ve ical in e pola ion also showed s abili y o
δ18O
H
2O
aluesbe weensumme andwin e ,whe easδ13CDIC alues
exhibi ed a ia ion be ween summe and win e in he uppe wa e
laye (up om ~50m; Figu e4e–h). Below ~50m bo h alues we e
a he s ableo e ime,excep in helesssalinea eas( heBo hnianSea,
heBo hnianBayand heGul o Finland)whe eδ13CDIC aluesshowed
mino dec easesdue oau umnwa e column u no e (Figu e4).
The ewasa s ong
δ18O
H
2O
–S ela ionshipand he coe icien s
o de e mina ion o all hemodelswe eclose o1.Rela ionshipap-
pea ed obelinea only o Sou he nBal icSea(R2=.98),whilein
Gul o Finland(R2=.97)andGul o Bo hnia ela ionshipswhe eo
loga i hmicshape(R2=.99;Figu e5).
3.2 | Ca bon and oxygen iso ope alues o
salmon o oli hs
S ableiso opeanalysis iamic omilling om heo oli hnucleus o he
o oli hedgeshowedclea a ia ionino oli hiso ope alues.Inaddi-
ion,bo hiso ope alueso he woo oli hsanalyzedshowedsimila
li e-his o y ends om henucleus o heo oli hedge.δ18O aluesin
heo oli hnucleus(FISH1:−10.9‰;FISH2:−10.6‰)we ema kedly
lowe han hehighes aluesobse edin heou e pa o heo oli h
adius(FISH1:−5.9‰;FISH2:−6.6‰),andwi h helowes aluesa
heedgeo heo oli h(bo ho oli hs:−11.6‰;Figu e6a,b).
FIGURE5 Rela ionshipsbe weenseawa e salini yand
δ18
OH
2O
measu edin(a)Gul o Bo hnia,(b)Gul o Finlandand(c)Bal ic
P ope .P esen edmodelswe eappliedwhenes ima ing
δ18
OH
2O
o
eacha ea
–12
–11
–10
–9
–8
–7
–6
–5
–4
0246810121
41
6
a
c
b
Salini y (‰)
δ18O(‰)
y = 0.263x - 8.684
R2= 0.98
y = 2.245ln(x) - 11.243
R2 = 0.97
y = 3.447ln(x) - 13.664
R2= 0.99
|
7
TORNIAINEN ET Al.
Fo bo ho oli hs,δ13C aluesalsoshowedclea a ia ionalong
he o oli h adius, om lowes δ13C alues in he o oli h nucleus
(FISH1:−16.1‰;FISH2:−14.4‰) ohighes in heou e pa o
heo oli h adius(FISH1:−3.4‰,FISH2:−3.1‰),wi hin e media e
aluesa heedgeo heo oli h(FISH1:−11.1‰;FISH2:−10.8‰;
Figu e6a,b).Bo ho oli hnuclei(i.e.,du ing heju enilephase)exhib-
i edlowδ13Co oandδ18Oo o alues.Theju enilephaseiso ope alues
o FISH1 ended obeless a iable han hoseo FISH2.Theo oli h
iso ope alueso appa en pos -smol mig a ion om he eshwa e
omo esalinea e hemos dis inc be ween hese wosalmon.FISH
1o oli h δ18O alues inc eased o hehighes aluesmo esha ply
han hoseo FISH2, o whichδ13C aluesinc easedinamo elinea
manne .O oli hδ18O alues o FISH1we ehighe du ing he i s
seawin e han hose o FISH2,while aluesdu ing hesecondsea
win e andappa en spawningmig a ion(s eepdeclineinbo hiso ope
aluesa e 2SW)we esimila o bo hsalmon(Figu e6a,b).
3.3 | P obabili y su aces o salmon loca ions in he
sea- eeding phase
Basedon heModel1(Hansone al.,2013)assignmen s, hemos
p obable i s seawin e loca iono FISH1wasin heBal icP ope
(Figu e7a).Du ing henex summe and hesecondwin e ,FISH1
appea ed ooccupywa e sclose o heGul o Riga,and hea easo
hesou hwes andno he nBal icP ope , espec i ely(Figu e7b,c).
The esul s o FISH2we e e ysimila ,excep ha in he i s win e
i appea ed obeloca edin heBo hnianSeao in heGul o Riga
(Figu e7d– ).Compa ed o heModel1,assignmen so heModel
3(Pa e sone al.,1993)we eloca edmo eno he nandalsocolde
a easo oa easwi hlowe salini y( heGul o Finland;Figu e7).
Model2( hea e agemodelo ou modelsusedin hes udy)salmon
assignmen swe ebe weenHansone al.(2013)andPa e sone al.
(1993)models(Figu e7).Someo heassignmen so hemodels2
and3we econ ined osuchasmalla ea ha esul so hemodels
we eno e y isiblein heassignmen maps(Figu e7).
To al o eigh di e en esul s (Figu e8) we e ob ained om
hesensi i i yanalysisinaddi ion o he esul ob ainedwi hac ual
measu ed alues.Inc easing heo oli hδ18O aluecauses hemos
p obablea eain heGul o Bo hnia oshi sligh ly owa dwes ,and
dec easingi causes hemos p obablea ea oshi sligh ly owa d
eas .Inc easingδ13C alueseems omake heassignmen mo econ-
cen a eda oneloca ionanddec easingi causes he esul obeless
de e mined.
4 | DISCUSSION
4.1 | Iso ope alues o he Bal ic Sea
Wep esen ed hedis ibu iono δ13CDICand
δ18O
H
2O
aluesin he
Bal icSeadu ingbo hsumme andwin e iain e pola edho izon al
FIGURE6 Iso ope alueso wo
exampleRi e Simojokisalmono oli hs
mic o-milledandanalyzed o oxygenand
ca bons ableiso opes:(a)FISH1(wild
o igin)and(b)FISH2(ha che y- ea ed).
Openci cles ep esen δ13C aluesand
illedci clesδ18O alues.Iso opepoin
aluesusedin hesalmonassignmen o
hep obableloca ionsin heBal icSea
du ing hei i s (1SW)andsecondsea
win e (2SW)andsecondseasumme
(2SS)a eindica ed
–20.0
–18.0
–16.0
–14.0
–12.0
–10.0
–8.0
–6.0
–4.0
–2.0
–12.0
–11.0
–10.0
–9.0
–8.0
–7.0
–6.0
δ
13
C(‰)δ
13
C(‰)
δ18O(‰)
–20.0
–18.0
–16.0
–14.0
–12.0
–10.0
–8.0
–6.0
–4.0
–2.0
–12.0
–11.0
–10.0
–9.0
–8.0
–7.0
–6.0
020406080100
Dis ance om nucleus (%)
δ18O(‰)
(a)
(b)
1SW 2SS 2SW
1SW 2SS 2SW
8
|
TORNIAINEN ET Al.
iso opicmaps omadep ho 10mandalso ia e icalc oss-sec ions.
Di e encesbe weenseaa easinbo hiso ope alueswe eobse ed.
Ho izon alδ13CDIC aluesclea lydi e edbe weensumme andwin-
e ,whe eas
δ18O
H
2O
alues emained a he simila .Ve ical
δ18O
H
2O
alues emaineds ablebe weensumme andwin e ,whe easδ13CDIC
showedma keddi e encesbe weensumme andwin e ,especially
nea hesu ace.
The dis inc i e hyd ologic cha ac e is ics o he Bal ic Sea a e
clea ly e lec edinou measu emen s.
δ18O
H
2O
aluesin heBal icSea
a emainlycon olledby hein lux om heDanishs ai so dense ,
mo e saline wa e wi h highe
δ18O
H
2O
(see e.g., Dickson, 1973;
Ma häus&Lass,1995)and he eshwa e wi hlowe
δ18O
H
2O
alues
unningo om heca chmen a ea.Theobse ed e icalg adien in
iso ope aluesisalsodue o hepene a iono mo esalinewa e wi h
highe
δ18O
H
2O
aluesunde helessdense eshwa e wi hlowe
δ18O
H
2O
alues.Thisdense wa e s agna esin hedeepe basinsiso-
la edby hesillsbe weendi e en Bal icSeaa eas(Ma häuse al.,
2008)andleads o hea ealseg ega iono
δ18O
H
2O
alues.Asimila
phenomenoncanbeseeninma ineen i onmen sa ound hewo ld
(Schmid , Bigg, & Rohling, 1999) and has been eco ded om he
Sou he nBal icSea(F ohlich,G abczak,&Rozanski,1988;Punning,
Vaikmae,&Maek i,1991).Tempe a u e- ela edau umno e u no
hewa e columnb eaksdown hesumme s a i ica ionandmixes
hewa e abo e hehalocline(down o~50m),bu hedeepe wa e
FIGURE7 P obabili ysu aceso
loca ions o woexampleRi e Simojoki
A lan icsalmons(FISH1and2)du ing hei
eedingphaseo i s seawin e (1SW;a
andd, espec i ely),secondseasumme
(2SS;bande, espec i ely)andsecond
seawin e (2SW;cand , espec i ely)in
heBal icSea.Calcula ionsa ebasedon
salmono oli handBal icSeawa e δ18O
andδ13C alues.Whi eandhigh-sa u a ion
colo sindica e ha all aluesu aces
usedin hecalcula iono p obabili ies
( empe a u e,
δ18
OH
2O
,δ13CDICandp ey
iso ope alues)a e eliableincon as o
lowcon idencep obabili iesindica edin
g ayandg ayishcolo s.P obabili ysu ace
esul sindica edin ed ep esen high
p obabili yo p esenceacco ding o he
usedmodelo Hansone al.(2013)(Model
1),g eencolo indica eshighp obabili y
o p esencewi h hea e agemodelo
all ou usedmodelsin hiss udy(Model
2;Godiksene al.,2010;Hansone al.,
2013;Pa e sone al.,1993;S o m-Suke
e al.,2007)andbluecolo ep esen s
highp obabili yo p esenceon hemodel
p obabili ieso Pa e sone al.(1993)
(Model3)
La i ude
(a)
Model 1 high p ob.
Model 2 high p ob.
Model 3 high p ob.
Low p obabili y
Low con idence
La i ude
(b)
Model 1 high p ob.
Model 2 high p ob.
Model 3 high p ob.
Low p obabili y
Low con idence
Longi ude
La i ude
(c)
Model 1 high p ob.
Model 2 high p ob.
Model 3 high p ob.
Low p obabili y
Low con idence
(d)
Model 1 high p ob.
Model 2 high p ob.
Model 3 high p ob.
Low p obabili y
Low con idence
(e)
Model 1 high p ob.
Model 2 high p ob.
Model 3 high p ob.
Low p obabili y
Low con idence
54 56 58 60 62 64 6654 56 58 60 62 64 66
10 15 20 25 30
54 56 58 60 62 64 66
10 15 20 25 30
Longi ude
( )
Model 1 high p ob.
Model 2 high p ob.
Model 3 high p ob.
Low p obabili y
Low con idence
FISH 1 FISH 2
|
9
TORNIAINEN ET Al.
column emains a he s ablemain ainingmo ema inecha ac e is ics
(Lass&Ma häus,2008).Howe e ,inou s udy,
δ18O
H
2O
alues e-
maineds ablebe weensumme andwin e ,andonlyoccasionals ong
sal wa e pulses(e.g.,Dickson,1973;Ma häus&Lass,1995)could
change his si ua ion. Lack o win e iso ope measu emen s sou h
om Go land migh ha e a ec ed he in e pola ed alues deno ed
in hemapswi hg ayishcolo s).TheFinnishEn i onmen Ins i u e
(SYKE;h p://www.i ame ipo aali. i/en_GB/:2.6.2014)andHELCOM
(Ande sson,2014)ha e epo eds ablehyd ologyandsalini yo he
Bal icdu ing hes udiedyea s.Basedon hosea ibu es,i isno e y
likely ha
δ18O
H
2O
aluesinsumme wouldbe e ydi e en om he
aluesobse edinwin e .
δ13CDIC aluesa econ olledbysimila mechanismsas hose ha
con ol he wa e column cha ac e is ics and
δ18O
H
2O
.The clea es
dis inc ion be ween he mechanisms based on empe a u e- ela ed
bioac i i y. P e e en ial inco po a ion o 12C du ing pho osyn hesis
FIGURE8 Sensi i i yanalysiso p obabilis icspa ialassignmen Model1 o FISH2in1s win e a hesea(1SW).O iginalo oli hδ18Oand
δ13C alues( igu ein hemiddle)we eadjus ed±1SD.Highp obabili ymeanswhe e hep obabili yo ob aining hemeasu ed(o de ia ed)
alueo heo oli h(o hep obabili yo p esenceo hisindi iduala 1SW) om hec ea edo oli hisoscapeis heg ea es by heused
model.Theha ched(low-sa u a ioncolo ed)a easindica e egionswhe e hedensi yo ealobse a ionso a leas oneisoscape(usedin he
calcula iono p obabili yo p esence)is oolow o eliablep edic ion
10 15 20 25 30
δ18Oa uppe , δ13Ca lowe limi
High p obabili y
Low p obabili y
10 15 20 25 30
δ18O a uppe limi
High p obabili y
Low p obabili y
10 15 20 25 30
Bo h a uppe limi
High p obabili y
Low p obabili y
10 15 20 25 30
δ13C a lowe limi δ13C a uppe limi
High p obabili y
Low p obabili y
10 15 20 25 30
Bo h iso opes as measu ed
High p obabili y
Low p obabili y
10 15 20 25 30
High p obabili y
Low p obabili y
10 15 20 25 30
54 56 58 60 62 64 6654 56 58 60 62 64 6654 56 58 60 62 64 66
Bo h a lowe limi
High p obabili y
Low p obabili y
10 15 20 25 30
δ18O a lowe limi
High p obabili y
Low p obabili y
10 15 20 25 30
54 56 58 60 62 64 66
54 56 58 60 62 64 66
54 56 58 60 62 64 66
54 56 58 60 62 64 66
54 56 58 60 62 64 66
54 56 58 60 62 64 66
δ18Oa lowe , δ13C a uppe limi
High p obabili y
Low p obabili y
Longi ude
La i ude
O oli hδ13C alue inc eases →
O oli h δ18O alue inc eases →
10
15
20
2
5
30
High
p o
b
a
bili
y
L
o
w p o
b
a
bil
i
y
10 1
5
20
25
30
High
p o
b
a
bili
y
L
o
w p o
b
a
bili
y
10
1
5
20
25
30
High
p o
b
a
bili
y
L
o
w p o
b
a
bili
y
10
15
20
25
30
δ
1
3
C
a
l
o
w
e
l
i
m
i
δ
1
3
C
a
u
ppe
li
m
i
Hi
g
h p obabili
y
L
o
w p obabili
y
10
15
20
25
30
Bo h iso o
p
es as measu ed
H
i
g
h p obabili
y
L
o
w p obabili
y
10
15
20
25
30
Hi
g
h p obabili
y
L
o
w p obabili
y
B
o
h
a
l
o
w
e
limi
Hi
g
h p obabili
y
L
o
w p obabili
y
δ
1
8
O
a
l
o
w
e
l
i
m
i
H
i
g
h p obabili
y
L
o
w p obabili
y
5
4
5
6
5
8
6
0
6
2
6
4
6
5
4
5
6
5
8
6
0
6
2
6
4
6
5
4
5
6
5
8
6
0
6
2
6
4
66
5
4
5
6
5
8
6
0
6
2
6
4
6
6
5
4
5
6
5
8
6
0
6
2
6
4
66
5
4
5
6
5
8
6
0
6
2
6
4
6
6
δ
1
8
O
a
l
o
w
e
,
δ
1
3
C
a
u
p
p
e
l
i
m
i
Hi
g
h p obabili
y
L
o
w p obabili
y