Why Baltic herring and sprat are weak conduits for astaxanthin from zooplankton to piscivorous fish

Why Baltic herring and sprat are weak conduits for astaxanthin from zooplankton to

piscivorous fish

Xiang-Ping Nie,a,b Jenny Zie,a Norbert Haubner,c Bo Tallmark,c and Pauline Snoeijsa,*

a Department of Systems Ecology, Stockholm University, Stockholm, SwedenbDepartment of Ecology, Jian University, Guangzhou, Chinac Department of Ecology and Evolution, Uppsala University, Uppsala, Sweden

Abstract

Atlantic salmon living in the brackish Baltic Sea have lower muscle pigmentation than populations elsewhere.The pigment in question is the antioxidant and vitamin A precursor astaxanthin, which is synthesized bycrustaceans from algal carotenoids. Baltic salmon feed nearly exclusively on the clupeids sprat and herring. Toevaluate astaxanthin availability to salmon we assessed astaxanthin levels and isomeric composition in their preyfish. We also analyzed astaxanthin dynamics in the dominant piscivorous fish in the Baltic Sea, the Atlantic cod.The geometrical E- (trans-) and Z- (cis-) isomers were distributed selectively in fish tissues, with highest E : Zratios in salmon gonads (82 : 18) and lowest in herring gonads (24 : 76). Sprat and herring are not ideal prey withrespect to their high whole-body concentrations of Z-isomers, which have low bioavailability for salmon and cod.These Z-isomers predominantly accumulate in the clupeid gonads. A crucial mechanism for the transport ofastaxanthin from clupeids to piscivores is the direct transfer of crustacean astaxanthin (mainly all-E) from theclupeid stomachs. Low stomach astaxanthin content in clupeids decreases total astaxanthin transfer to highertrophic levels. In autumn, herring stomachs (including contents) had 12.5 times lower astaxanthin concentrationsthan sprat stomachs, and herring had 2.8 times less whole-body all-E-astaxanthin (by weight) than sprat. Theseresults confirm recent reports of starvation in the Baltic herring, which may further decrease astaxanthin levels inthe Baltic salmon. Cod did not have lower astaxanthin levels than their Atlantic counterpart, which may beattributed to their lower need for astaxanthin and higher food diversity.

The carotenoid astaxanthin is abundant in many aquaticanimals (Matsuno 2001). This fat-soluble pigment is astrong antioxidant (Miki 1991) and in fish it also acts as avitamin A precursor (Schiedt et al. 1985). Fish cannotproduce astaxanthin themselves and need to acquire itthrough their diet. The ecological significance of astax-anthin in aquatic food webs has been little studied, whereasastaxanthin as a commercial food additive has receivedmuch attention. For example, artificial astaxanthin is themost widespread carotenoid used for the red pigmentationof salmonids in aquaculture (Bjerkeng 2008). The antiox-idant properties of astaxanthin and other carotenoids aredue to the long chromophore of conjugated double bonds,which makes the molecule able to absorb light and toquench harmful excited oxygen and free radicals (Fig. 1;Higuera-Ciapara et al. 2006). A few algal species canproduce astaxanthin, but in aquatic systems it is mainlysynthesized by crustaceans from carotenoid precursors,especially b-carotene and zeaxanthin, provided by algae(Matsuno 2001). Consecutively, astaxanthin originatingfrom crustaceans is the primary astaxanthin source for fish.

The effects of environmental, nutritional, and toxicstress on organisms are often mediated through the cellularprocess of oxidative stress (Lesser 2006). The oxidativereagents, reactive oxygen species (ROS), are continuouslyproduced in all oxygenic living cells and are essential in thecontrol of cell functions in biological systems. If notregulated properly, the excess ROS can damage majormacromolecules such as membrane lipids, proteins, and

deoxyribonucleic acid (DNA) and inhibit normal cellfunctioning. Oxidative stress can result from a lack ofantioxidant capacity caused by a disturbance in theproduction, cellular distribution, or uptake of antioxidants.It can also result from an overproduction of ROS as aresponse to external stressors. Thus, in most situations lackof antioxidants cannot be expressed as a specific concen-tration of antioxidants and even less as a specificconcentration of a single antioxidant such as astaxanthin.Lack of astaxanthin and critical astaxanthin concentrationsdepend on synergistic interactions with other antioxidantssuch as a-tocopherol (vitamin E) and ambient antioxidativerequirements, e.g., the level of polyunsaturated fatty acids(Pickova et al. 1999). However, in organisms thataccumulate astaxanthin as a major antioxidant, such ascrustaceans and fish, lower concentrations than normal canbe used as a signal for reduced animal health. Lowastaxanthin levels, visible as paler muscle or egg color thannormal, indicate a decreased intake of astaxanthin or anincreased depletion of astaxanthin because of oxidativestress (Miki 1991).

Fish take up astaxanthin from their food in theintestines, after which it is transported in the blood plasmaand accumulated mainly in liver, muscle, gonads, and skin.Most salmonid fish accumulate high levels of astaxanthinin the muscle, but low levels of astaxanthin also occur in thelow-fat white muscle of other fish species, e.g., cod(Ytrestøyl and Bjerkeng 2007a). When female fish becomesexually mature, astaxanthin accumulates in the gonads toprotect fish roe and fry against oxidative stress (Izquierdoet al. 2001). In Atlantic salmon post-smolts, the dietary lack* Corresponding author: [email protected]

Limnol. Oceanogr., 56(3), 2011, 1155–1167

E 2011, by the American Society of Limnology and Oceanography, Inc.doi:10.4319/lo.2011.56.3.1155

1155

of astaxanthin activates oxidative stress indicators such asthe recovery of desaturated and elongated products of fattyacids, production of malondialdehyde (lipid peroxidation),and nonenzymatic production of 8-isoprostane in bloodplasma, which confirms the general antioxidant function ofastaxanthin (Bell et al. 2000). In the brackish Baltic Sea, themuscle of the Atlantic salmon is extremely pale. Petterssonand Lignell (1998) reported twelve times lower astaxanthinlevels in Baltic than in Atlantic salmon populations. Lowastaxanthin concentrations are correlated with the abun-dance of the M74 syndrome in Baltic salmon populations(Pettersson and Lignell 1999). Roe and yolk sac fry, i.e.,life-stages that still rely on the yolk for nutrition, fromfemales low in astaxanthin and thiamine (vitamin B1) havean increased mortality rate as a result of oxidative stress(Amcoff et al. 1999; Pettersson and Lignell 1999).

The aim of the present study was to explain whyastaxanthin concentrations are so low in the Baltic salmoncompared to Atlantic populations through detailed analysisof astaxanthin dynamics in the Baltic salmon’s two majorprey species, herring and sprat, including seasonal andspatial variation. A second aim was to examine whethercod, the major piscivorous fish in the Baltic Sea, also haslower astaxanthin levels than its Atlantic counterpart. Notonly the availability of astaxanthin in fish prey per se, butalso the relative proportions of its geometrical isomers, candetermine how much astaxanthin is transferred from feedto fish. The thermodynamically more stable all-E- (trans-)isomers are the most common geometrical isomers ofcarotenoids in nature (Liaanen-Jensen and Lutnaes 2008).Z-(cis-) isomerization requires energy in the form of light,heat, or reactive surfaces. In all-E-astaxanthin the moleculeis straight, and in 9Z- and 13Z-astaxanthin it is twisted atthe 9th and 13th C atom, respectively (Fig. 1). Thesedifferences in spatial arrangement of the astaxanthinmolecule among the three isomers cause differences inbiochemical traits such as a higher in-vitro antioxidant

activity of Z-isomers (Liu and Osawa 2007) and higherbioavailability of the all-E-isomer for salmon and cod(Bjerkeng et al. 1997; Østerlie et al. 1999; Ytrestøyl andBjerkeng 2007b). Therefore we assessed the concentrationsof the E- and Z-isomers of astaxanthin separately, inaddition to total astaxanthin concentrations. The resultsare discussed with respect to the special conditions forfood-web dynamics and threshold-like ecological shifts inthe species-poor brackish Baltic Sea.

Methods

Field sampling of fish—Specimens of Baltic sprat(Sprattus sprattus balticus Schneider), Baltic herring (Clu-pea harengus membras L.), and Atlantic cod (Gadus morhuaL.) were caught by trawling in the open Baltic Sea proper,between the Hano Bight in the south and the northernGotland area in the north (Fig. 2). The fish were taken at17 sampling stations during two offshore cruises with R/VArgos during 28 February–10 March and 21 November–01December 2005. All stations were included in the Baltic Seacod monitoring program coordinated by the InternationalCouncil of Exploration of the Sea (ICES) and located inICES Areas 25, 27, and 28 (Fig. 2). All trawls were carriedout between 08:00 and 10:00 h in the morning, trawlingtime was c. 45 min, and average depth was 58 m (minimum37 m, maximum 77 m). For comparison with a previousextensive study on astaxanthin concentrations in salmon(Pettersson and Lignell 1998), six specimens of wildAtlantic salmon (Salmo salar L.), caught on 28 November2005 at Sta. 99, were purchased from local fishermen atKarehamn, Oland. All fish were put on ice immediatelyafter sampling. Tissue samples from gonads, liver, muscle,and stomach (including stomach contents) of female fish

Fig. 1. Molecular structures of all-E-astaxanthin, 9Z-astax-anthin, and 13Z-astaxanthin. Fig. 2. Map of the Baltic Sea proper, showing the 18

sampling stations in ICES Areas 25, 27, and 28. Sta. 51–59 weresampled 28 February–10 March 2005, and Sta. 90–99 weresampled 21 November–01 December 2005.

1156 Nie et al.

were taken within 2 h from sampling and then packed andfrozen at 220uC until analysis. The length and weight ofeach specimen was noted before dissecting the fish.

Field sampling of zooplankton—Each morning beforetrawling and before sunrise (between 5:00 and 7:00 h), sixseawater samples of 1000–2000 liters were pumped up froma depth of 5–7 m; and the mesozooplankton . 200 mm werecollected on a plankton net, rinsed, and transferred to600 mL of filtered seawater. Quantitative subsamples werecollected on WhatmanTM GF/F filters for carotenoidanalysis, and one subsample was preserved with Lugolsacid iodine for the analysis of the biovolume andtaxonomic composition of the mesozooplankton commu-nity. All filters were packed in aluminum foil, quick-frozenin liquid nitrogen onboard, and transferred to 280uCstorage upon return to land.

Carotenoid extraction of fish samples—Altogether, 577tissue samples of female sprat, herring, and cod muscle,gonads, liver, and stomachs from the expeditions wereanalyzed for carotenoids (Table 1). One sample of eachorgan was analyzed from each individual fish. Not allspecies occurred at all sampling stations; sprat and herringgonads and stomachs and cod gonads from 17, 15, and 14stations, respectively, were included. Muscle and liversamples of these three species, as well as cod stomachs,were analyzed from six stations (Table 1). For six salmon

females from Sta. 99, triplicate samples of muscle, gonad,and liver were analyzed. The total organ weight of gonads,liver, and stomach (including contents) from each individ-ual fish was measured before c. 0.5 g of tissue was taken forcarotenoid extraction. For sprat, these organs were alwaysbelow 0.5 g, and sample weight was 0.1–0.3 g. Theextraction protocol followed Pettersson and Lignell (1998,1999) with some minor modifications. The samples werehomogenized in 2 mL of ice-cold acetone, using an Ultra-TurraxH T25 for 20 s at 25,000 rounds per minute (rpm)and transferred to a FalconTM tube. The homogenizer wasrinsed twice with 1 mL of acetone, and these fractions werealso added to the tube, adding up to 4 mL of acetone. Thesamples were refrigerated at 4uC to reach a total extractiontime of 30 min, including homogenization. The tubes werecentrifuged at 6000 3 g at 4uC for 2 min. Three milliliters ofthe sample was transferred to a new FalconTM tube, and3 mL of cyclohexane was added, together with 2 mL ofdouble-distilled water to obtain phase separation. Eachsample was shaken vigorously by hand for 3 min andcentrifuged at 6000 3 g at 4uC for 2 min; next, 1.5 mL ofthe carotenoid-containing cyclohexane phase was trans-ferred to an EppendorfTM tube, and the sample wasevaporated to dryness with nitrogen gas. The samples werethen resolved in 250 mL of mobile phase A (80% methanol+ 20% 0.5 mol L21 ammonium acetate buffer of pH 7.2).Due to the pigment’s light sensitivity, exposure to light wasminimized throughout all laboratory work.

Table 1. Results of 15 two-way factorial ANOVA analyses with log-transformed fish data as responsevariables and area of the Baltic Sea proper (ICES Areas 25, 27, and 28) and season (Mar and Nov) aspredictor variables. n 5 number of samples analyzed (from different fish specimens), S 5 number of samplingstations included, Asta 5 total astaxanthin.

Tissue andspecies n S

Area Season Area3season

F p F p F p

Body weight (g)

Sprat 84 15 13.68 ,0.001 0.26 0.611 4.91 0.010Herring 91 17 9.93 ,0.001 6.99 0.010 0.65 0.527Cod 80 14 14.36 ,0.001 0.04 0.851 14.45 ,0.001

Asta in gonads (mg g21)

Sprat 84 15 0.34 0.714 10.48 0.002 0.45 0.642Herring 91 17 0.63 0.535 0.68 0.412 0.33 0.719Cod 80 14 0.34 0.715 3.00 0.087 2.93 0.060

Asta in liver (mg g21)

Sprat* 17 5 0.00 0.993 0.77 0.396 —{ —Herring* 21 6 0.00 0.962 6.71 0.019 1.18 0.292Cod* 24 6 0.07 0.789 0.75 0.396 0.09 0.767

Asta in muscle (mg g21)

Sprat* 22 6 0.67 0.424 0.00 0.984 3.54 0.076Herring* 21 6 0.24 0.631 1.08 0.314 2.51 0.131Cod* 22 6 0.11 0.739 3.27 0.087 0.01 0.910

Asta in stomach (mg g21)

Sprat 81 15 0.23 0.795 8.63 0.004 1.14 0.325Herring 91 17 0.63 0.535 26.30 ,0.001 0.14 0.867Cod* 23 6 0.20 0.663 0.76 0.393 2.08 0.165

* Data from only two areas (ICES Areas 27 and 28).{ Dashes indicate main effects ANOVA instead of factorial ANOVA because of lack of data.

Astaxanthin dynamics in fish 1157

Carotenoid extraction of zooplankton samples—Caroten-oids were extracted from three zooplankton filters perstation (one filter per sample) according to Van Nieuwer-burgh et al. (2005). The filters were sonicated for 1 min incold 100% high-performance liquid chromatography(HPLC) grade methanol on ice (Vibra CellTM, amplitude92, pulse 0.9 s), and centrifuged at 6000 3 g at 4uC for4 min. Before analysis, 20% 0.5 mol L21 ammoniumacetate buffer was added to 80% of the methanol extract toequal mobile phase A.

Analysis of carotenoids—Syringes with 0.45-mm polytet-rafluoroethylene-polypropylene (PTFE-PP) filters (Ti-tanTM) were used to transfer the samples into 2-mL dark-glass vials. HPLC was carried out with an AgilentTM 1100System with a scanning ultraviolet (UV) diode arraydetector or with a ShimadzuTM LC-10ADvp System witha UV detector at 472 nm. A subsample of 75 or 100 mL wasinjected into a Gemini C18 (PhenomenexTM) column (3-mmparticle size, 30 3 3.00 mm, 110A) at a flow rate of0.3 mL min21 with gradients of mobile phase A (seeabove), B (90 : 10 acetonitrile : H2O), and C (ethyl acetate).The solvent gradient (time in min; A%, B%, C%) for theAgilent system was (0; 100, 0, 0), (2; 0, 100, 0), (19; 0, 20,80), (21.5; 0, 100, 0), (24; 100, 0, 0), (30; 100, 0, 0), and forthe Shimadzu system it was (0; 100, 0, 0), (2; 0, 100, 0), (17;0, 20, 80), (21; 0, 100, 0), (23; 100, 0, 0), (36; 100, 0, 0).Chromatograms were integrated using the softwareChromProcessorTM (Advanced Chemistry Development)or ClarityTM (DataApex). Identification of astaxanthinisomers was confirmed by absorption spectra. A commer-cial all-E-astaxanthin standard (synonym: all-trans-astax-anthin) was purchased from Sigma-AldrichTM, and the twoisomers 9Z- and 13Z-astaxanthin (synonyms: 9-cis- and 13-cis-astaxanthin) were produced from the standard powderby incubation in chloroform at 35uC for 110 h (Yuan andChen 1999). Five-point calibrations were carried out forboth HPLC equipments using the all-E-astaxanthin stan-dard, and astaxanthin concentrations were calculated withthese calibrations. Astaxanthin esters and isomers werequantified from peak areas compared to the peak of the all-E-isomer. The quantities of the 9Z- and 13Z-isomers weremultiplied by 1.20 and 1.56, respectively, to account fortheir lower specific absorption at 472 nm (Bjerkeng et al.1997; Maltby et al. 2003). Extracts of Haematococcuspluvialis aplanospores were used to confirm the elutiontimes of the astaxanthin esters. Throughout this article,carotenoid concentrations are expressed as mg carotenoidper g wet weight (wet wt) of fish tissue or mesozooplanktonbiomass.

Calculation of whole-body concentrations of astax-anthin—We measured total body weight while sampling.Gonad, liver, and stomach weights were measured beforecarotenoid analyses. To be able to estimate whole-bodyconcentrations of astaxanthin for the two clupeid fishspecies, 24 herring specimens were separated into eightfractions and weighed separately: (1) gonads, (2) liver, (3)stomach, (4) other viscera and blood, (5) muscle and skin,(6) head and gills, (7) skeleton, (8) fins, to obtain the

percentage of these body parts in relation to the whole fishbody. The relative proportions of the body parts excludinggonads, liver, and stomach of the closely related sprat wereassumed to be equal to those of herring. In the calculationsit is assumed that body astaxanthin occurs in gonad, liver,other viscera, blood, stomach, and muscle (including fatand skin) and that astaxanthin concentrations in otherparts of the body (head, gills, skeleton, and fins) arenegligible.

Zooplankton identification and biomass conversions—Sixmesozooplankton subsamples were identified to the genuslevel and counted. In each subsample, the body lengths ofthe 25 first-encountered individuals of each genus weremeasured and converted to biovolume using shape-adaptedequations. These estimates appeared robust as theycompared well with published biovolume data for the samegenera sampled from the Baltic Sea at the same time of year(Hernroth 1985). Wet wt was calculated from biovolume,using a specific gravity of 1.025 as determined for thesouthern Baltic Sea by Chojnacki (1983). For comparisonswith literature data, we used the conversions dry weight(dry wt) 5 21.3% wet wt and C content 5 44.7% dry wt(Mauchline 1998).

Data analysis—ANOVA analyses and calculations of thePearson product moment correlation coefficient (RP) wereperformed with the statistical package StatisticaTM version9. A principal components analysis (PCA) was performedto summarize how similar the astaxanthin isomer compo-sitions of the different fish tissues were to each other and tothat in zooplankton, and how variation in carotenoidcomposition related to area, season, and tissue type. ThePCA was carried out with the program CanonicalCommunity Ordination (CANOCO; ter Braak and Smi-lauer 2002) on the relative abundances of the all-E-, 9Z-,and 13Z-isomers (% of the total astaxanthin concentra-tion), and these data were log-transformed prior toanalysis. Dummy variables (scores ‘‘1’’ or ‘‘0’’) wereassigned to the nominal (categorical) variables samplingarea, season, and tissue type and were tested by multipleregression analysis on the results of the PCA. Thus, thesevariables were not used to construct the ordination, buttheir influence on the ordination results was testedpassively. Throughout this paper, statistical significance isaccepted at p , 0.05.

Results

Astaxanthin detection—The HPLC analyses detectedseven carotenoid peaks in the fish samples (Fig. 3). Thesame carotenoids, as well as canthaxanthin and astaxanthinesters, were detected in the zooplankton samples. Astax-anthin was by far the major carotenoid in all samples(. 95% of total carotenoid concentration). The elutiontimes of the three dominant geometrical isomers, all-E-,9Z-, and 13Z-astaxanthin, in the fish samples agreed withthose in the chloroform-dissolved standard solution(Fig. 3). Three smaller peaks (Peaks 1, 3, and 5) alsoshowed typical astaxanthin spectra with maxima at 470–

1158 Nie et al.

480 nm. Peak 1 is situated before all-E-astaxanthin (Fig. 3)and may be di-Z-astaxanthin based on its elution time. Aseventh peak (Peak 7) rarely occurred after 13Z-astax-anthin and may be 15Z-astaxanthin based on its elutiontime. However, the spectrum of this latter peak could notconfirm astaxanthin because of the small peak size. Peaks 3and 5 were situated immediately before 9Z-astaxanthin and13Z-astaxanthin, respectively, and may be related to theseastaxanthin forms. When we refer to ‘‘total astaxanthin’’this represents the sum of the three major peaks, whichwere positively identified as all-E-, 9Z-, and 13Z-astax-anthin. For the zooplankton, ‘‘total astaxanthin’’ alsoincludes the astaxanthin esters.

Astaxanthin isomeric composition—Carotenoid composi-tion varied more among fish tissues than among areas andseasons. This is shown in the PCA ordination plot (Fig. 4)by the positions of the centroids for area and season, whichare situated close to each other near the center of theordination plot, while the centroids for the tissues arescattered all around but not in the center. The two firstordination axes had eigenvalues 0.61 and 0.33, showingthat axes 1 and 2 together explained most of the variationin carotenoid composition. The astaxanthin isomericcomposition was dominated (. 50%) by all-E-astaxanthin

for all tissues, except for sprat and herring gonads andsalmon liver, which were dominated by astaxanthin Z-isomers (Fig. 4). The highest proportions of all-E-astax-anthin (74–82%) were found in salmon gonads and muscleand in cod gonads, as well as in the zooplankton (75–85%).Sprat and herring gonads are situated in the upper-leftquadrant of the ordination and had the highest proportionsof 13Z-astaxanthin (74 and 76%, respectively). Salmonliver had the highest level (37%) of 9Z-astaxanthin.Generally, the proportion of 9Z-astaxanthin was slightlyhigher in all salmon tissues compared with the other threespecies. Sprat muscle, liver, and stomachs had 62–64% all-E-astaxanthin, herring muscle, liver, and stomachs 51–56%, cod liver 51%, cod muscle 63%, and cod stomachs71%. The isomeric composition of cod stomachs, includingcontents, was closest to that of sprat muscle and stomachand zooplankton (Fig. 4).

Astaxanthin concentrations in different tissues—The totalastaxanthin concentrations were extremely variable amongindividual fish. Despite the large number of samplesanalyzed, the 95% confidence limits were generally high(Fig. 5). Astaxanthin concentrations in gonads varied withseason for sprat, with 3.5 mg g21 in March and 8.5 mg g21 inNovember, but not with area of the Baltic Sea (Table 1;

Fig. 3. Examples of chromatograms showing the seven peaks detected during retention timeof 12 to 26 min. 1, Peak 1; All-E, all-E-astaxanthin; 3, Peak 3; 9Z, 9Z-astaxanthin; 5, Peak 5; 13Z,13Z-astaxanthin; and 7, Peak 7.

Astaxanthin dynamics in fish 1159

Fig. 5). For the other three species, gonad concentrationsdid not vary with season or area, and average values were1.9, 3.9, and 0.7 mg astaxanthin g21 for herring, cod, andsalmon, respectively.

In liver and muscle tissue, the total astaxanthinconcentrations were much lower than in the gonads, andthey did not vary significantly with area of the Baltic Sea orwith season, except for herring liver, which had eight timeslower astaxanthin concentrations in November (0.016 mgg21) than in March (0.129 mg g21) (Table 1; Fig. 5). Insprat, cod, and salmon, average liver concentrations were0.320, 0.022, and 0.044 mg g21, respectively. Averagemuscle concentrations were 0.055, 0.020, 0.018, and0.033 mg g21 in sprat, herring, cod, and salmon, respec-tively. Muscle concentrations in sprat and salmon weresignificantly higher than those in herring and cod (one-wayANOVA, Tukey’s HSD test), and there was no significantdifference between sprat and salmon.

Astaxanthin concentrations in sprat and herring stom-achs varied with season, displaying low values in Novem-ber, but not with area of the Baltic Sea (Table 1; Fig. 5).Average stomach concentrations in March and Novemberin sprat were 3.7 and 1.8 mg g21 and in herring 1.04 and0.15 mg g21, respectively. Maximum stomach concentra-tions (including stomach contents), indicating the astax-anthin level at maximum stomach fullness in this study,were 19.2 and 12.0 mg g21 in sprat and herring, respectively.

In cod the stomach values did not vary significantly withseason or area and were on average 0.15 mg g21.

Fish body weight and total astaxanthin concentrations—For sprat, herring, and cod the total astaxanthin concen-trations in the four different tissue types were independentof fish body weight (RP; p . 0.05), although body weight ofthe sampled fish was highly variable. The sprat specimensweighed 9–17 g, herring 19–267 g, cod 71–1698 g, andsalmon 3125–4700 g. Generally, herring weight was below100 g, but at Sta. 55 and 94 only large female specimens(117–267 g) were available. Cod was usually well above200 g, but at Sta. 51 only small female specimens (, 100 g)were available. Body weight varied with area and, forherring, also with season (Table 1). Average herring bodyweight was 66 g in March and 80 g in November. Salmonwas only sampled at one station in November (n 5 6specimens), and the total astaxanthin concentrations ingonads, liver, and muscle were independent of fish bodyweight (RP; p . 0.05).

The June herring used to estimate body dimensions hadhigher gonad weight, 7.7% of the total body weight, thanthe March and November herring used for astaxanthinanalyses (Table 2). With the help of the June relative tissueweights of herring, excluding gonads, liver, and stomach,the clupeid fish analyzed for astaxanthin were estimated toconsist of 64–67% muscle, 5.4–5.7% blood and visceraother than liver and stomach, and 24–25% head, gills,skeleton, and fins (Table 3). The percent gonad weight ofthe total body weight was significantly higher in herringthan in sprat in both March and November (Table 3; t-tests) and significantly higher in March than in Novemberin both species (Table 3; t-tests). The percent stomachweight (including contents) of total body weight wasaround 1% for both herring and sprat and significantlyhigher in March than in November for both species(Table 3; t-tests).

Model of total astaxanthin concentrations in sprat andherring—The independence of total astaxanthin concentra-tions in the different tissues from body weight and samplingarea makes it possible to construct a simple model forcalculating total body astaxanthin in sprat and herring inthe Baltic Sea proper in spring and autumn (Table 4). Thismodel is based on body proportions and tissue concentra-tions for March and November (Table 3). Whole-bodyconcentrations of total astaxanthin calculated with thismodel are, for sprat, 0.140 mg g21 in March and 0.112 mg g21

in November and, for herring, 0.106 mg g21 in March and0.084 mg g21 in November. With respect to astaxanthinquantity irrespective of isomer composition, sprat is thus,in both seasons, a c. 25% better prey (by weight) forpiscivorous fish than herring.

Bioavailable astaxanthin concentrations in sprat andherring—The relative gonad weight in herring was higherthan that in sprat (Table 3). In herring most of the totalbody astaxanthin was located in the gonads, as much as90% in November, but in sprat this was around 45% inboth seasons (Table 4). Thus, piscivorous fish that con-

Fig. 4. Results of principal component analysis (PCA) onlog-transformed relative carotenoid data, showing carotenoidscores as arrows. The length and direction of the arrows areproportional to the influence of the respective forms ofastaxanthin in the ordination plot. The effects of the dummyvariables for sampling area, season, and tissue type were testedpassively by multiple regression analysis on the results of the PCA,and their scores in the ordination plots are shown as centroids.

1160 Nie et al.

sume females of these clupeid fish will acquire relativelymore astaxanthin from herring gonads than from spratgonads. In contrast, relatively more astaxanthin is trans-ferred from sprat stomachs to the predator than fromherring stomachs, twice as much in March and eight timesas much in November (Table 4). This also has consequenc-es for the composition of the astaxanthin isomerstransferred to piscivorous fish because these accumulateselectively in the different clupeid tissue types with

minimum bioavailable E-isomer levels in herring and spratgonads (Fig. 4). The difference between the two clupeid fishwith respect to all-E-astaxanthin concentrations is large.For sprat in November, 50% of the astaxanthin consumedby the piscivorous fish consists of all-E-astaxanthin, but fora herring this is only 24% (Table 5). In March thedifference is less: 46% in sprat and 32% in herring. Basedon this, estimates of whole body concentrations of all-E-astaxanthin are, for sprat, 0.064 mg g21 in March and

Fig. 5. Seasonal and spatial variation of total astaxanthin concentrations in mg per g fresh weight of sprat, herring, and cod gonad,liver, muscle, and stomach (including stomach contents). 0 5 No samples were analyzed from this area. Error bars are 95%confidence limits.

Astaxanthin dynamics in fish 1161

0.056 mg g21 in November and, for herring, 0.034 mg g21 inMarch and 0.020 mg g21 in November. The concentrationsof all-E-astaxanthin in sprat are 1.9 (March) to 2.8(November) times those in herring. With respect toastaxanthin quality (bioavailable to cod and salmon), spratis thus a two to three times better prey (by weight) thanherring.

Zooplankton—The mesozooplankton community almostexclusively consisted of four genera of calanoid copepods(Table 6). Only 0.7% (March) and 2.0% (November) of thetotal community biovolume was made up by otherzooplankton taxa (copepods and cladocerans). The relativeabundances of Acartia spp. and Centropages hamatus(Liljeborg) were highest in March, and the relativeabundance of Temora longicornis (O. F. Muller) was

highest in November. The relative abundance of Pseudo-calanus acuspes (Giesbrecht) was low in both seasons (3.8–6.0%). Total community biomass and total astaxanthinavailability per liter of seawater were 3.5 and 4.6 timeshigher in November than in March, but the average size ofthe zooplankton organisms was 23% lower in November(Table 6). Also, the food quality of the zooplankton withrespect to astaxanthin was significantly higher in Novem-ber than in March. Total astaxanthin concentrations were25% higher in November, and the proportion of all-E-astaxanthin was 85% in November compared with 75% inMarch (Table 6).

Discussion

Astaxanthin levels in zooplankton—The total astaxanthinconcentrations in the copepod-dominated mesozooplank-ton (c. 46 mg g wet wt21) were in the same order ofmagnitude as published data for different marine areas innorthern Europe. Adult Acartia bifilosa (Giesbrecht) andPseudocalanus acuspes from the southern Baltic Sea

Table 2. Weight of different body tissues expressed as percentof total body weight of 24 female herring individuals caught byfishing rod on 29 Jun 2010 in the Stockholm archipelago at GlobalPositioning System 59u17.599N, 18u39.909E. CI 5 confidenceinterval.

Weight Mean695% CI

Body (g) 27.863.5Gonads (%) 7.762.1Liver (%) 1.760.2Stomach (%) 1.860.2Other viscera and blood (%) 5.260.6Muscle, including skin (%) 61.161.9Head and gills (%) 19.661.1Skeleton (%) 1.260.0Fins (%) 1.860.1

Table 3. Mean values 6 95% confidence interval for weights and astaxanthin concentrations for sprat and herring. Asta 5 totalastaxanthin, wet wt 5 wet weight.

Sprat Herring

Mar Nov Mar Nov

Weight

Total body (g wet wt) 12.260.5 (n544) 12.560.7 (n540) 65.6616.0 (n548) 79.6611.8 (n543)Gonad (% of body wet wt) 1.6760.29 (n544) 0.6560.10 (n540) 4.7961.02 (n548) 3.1060.61 (n543)Liver (% of body wet wt) 1.0860.27 (n512) 1.1060.09 (n55) 1.2460.23 (n512) 1.2660.36 (n59)Stomach, including contents (% of body

wet wt) 1.0460.07 (n543) 0.8360.10 (n538) 1.3260.07 (n548) 0.9360.07 (n543)Muscle (% of body wet wt)* 66.12 66.96 63.68 65.10Other viscera and blood (% of body wet

wt)* 5.63 5.70 5.42 5.54Head, gills, skeleton, and fins (% of body

wet wt)* 24.46 24.77 23.55 24.08

Asta

Content in fish body (mg){ 1.7160.07 (n544) 1.4060.08 (n540) 6.9661.69 (n548) 6.6960.99 (n543)Concentration in gonad (mg g wet wt21) 3.47961.508 (n544) 8.47663.155 (n540) 1.49660.353 (n548) 2.43860.767 (n543)Concentration in liver (mg g wet wt21) 0.33760.175 (n512) 0.28160.269 (n55) 0.12960.096 (n512) 0.01660.017 (n59)Concentration in stomach (mg g wet

wt21) 3.66161.157 (n543) 1.84461.164 (n538) 1.04360.584 (n548) 0.14760.070 (n543)Concentration in muscle (mg g wet wt21) 0.05760.037 (n512) 0.05460.043 (n510) 0.02760.018 (n512) 0.01060.012 (n59)

* Extrapolated from Table 2.{ Calculated using measured total body weights, the seasonal average percent weight of the different body parts, and the seasonal average astaxanthin

concentrations for these body parts.

Table 4. Model for calculating total astaxanthin (Asta) indifferent body parts of sprat and herring in different seasons,based on the data presented in Table 3.

Asta location

Sprat (%) Herring (%)

Mar Nov Mar Nov

Gonads 41.3 48.8 67.6 89.8Stomach 27.2 13.6 13.0 1.6Liver 2.6 2.8 1.5 0.2Rest of fish body 28.9 34.8 17.8 8.3

1162 Nie et al.

contained on average 100 and 115 mg g wet wt21,respectively (Łotocka et al. 2004); and, in Calanushelgolandicus (Claus) from the North Sea, astaxanthinconcentrations varied between 11 and 43 mg g wet wt21

(Sommer et al. 2006). Reported concentrations in copepod-dominated zooplankton from the Norwegian Sea varybetween 51 and 303 mg g wet wt21 (Van Nieuwerburgh etal. 2005; Van der Meeren et al. 2008). Larger crustaceanzooplankton organisms also have similar total astaxanthinlevels, e.g., 46–93 mg g wet wt21 in the krill speciesMeganyctiphanes norvegica (M. Sars) and Thysanoessainermis (Krøyer) (Storebakken 1988). In our fish samplesall astaxanthin was in the free form, but in the zooplanktonit was partly present as esters of fatty acids (33% in Marchand 40% in November). This is typical for crustaceans;reported ester proportions are usually higher in copepodsthan in our samples (. 45%; Van Nieuwerburgh et al.2005), and in krill up to 96% can be esterified (Foss et al.

1987). In the gastrointestinal tract of fish, the esters arehydrolyzed before being absorbed. No astaxanthin esterswere detected in our fish stomach samples, a result whichindicates that this is a fast process.

Astaxanthin transfer from zooplankton to clupeids—Thezooplankton astaxanthin concentrations (c. 46 mg g wetwt21) were much higher than the total body concentrationsof the clupeid fish (0.08–0.14 mg g wet wt21). However,individual clupeid stomachs (including contents) couldcontain up to 19 mg astaxanthin g wet wt21, indicatingthe consumption of crustaceans. The feeding of both spratand herring is related to prey relative abundance and size(Viitasalo et al. 2001; Casini et al. 2004). Sprat and smallerherring (up to 14–20 cm in length) are strictly zooplank-tivorous, selecting principally the copepods Temora long-icornis and Pseudocalanus acuspes and the cladoceranBosmina maritima (P.E. Muller). Larger herring areessentially nektobenthos feeders, predating on mysids,amphipods, and polychaetes (Casini et al. 2004). The lowastaxanthin concentrations in the clupeid stomachs inNovember, especially for herring, cannot be explained bylow astaxanthin availability in the zooplankton becausethis was 4.6 times higher in November than in March. Apossible explanation for the low stomach astaxanthincontent in herring in November is that zooplankton preysize was too small in relation to fish size (Viitasalo et al.2001). The average biovolume of the zooplankton organ-isms was 23% lower and the herring body weight was 21%higher in November compared to March. Furthermore, in

Table 5. Distribution of astaxanthin geometrical isomers insprat and herring (whole body), in different seasons, based on thedata presented in Table 4 and seasonal average isomer concen-trations in the different body parts.

Sprat (%) Herring (%)

Mar Nov Mar Nov

All-E-astaxanthin 45.7 50.1 31.7 23.99Z-astaxanthin 3.5 4.5 4.8 5.513Z-astaxanthin 50.8 45.4 63.5 70.6

Table 6. Mean values 6 95% confidence interval for community composition, body size,biomass, total astaxanthin availability, and astaxanthin concentrations in zooplanktoncommunities (in the Baltic Sea proper in Mar and Nov 2005). wet wt 5 wet weight. p-valuesshow the results from comparisons of means for Mar and Nov (unpaired t-tests).

Mar(n59 stations)

Nov(n58 stations) p-value

Zooplankton genera (% of total biovolume )

Acartia 69.768.1 45.066.8 ,0.001Centropages 12.961.4 5.562.5 ,0.001Pseudocalanus 6.065.2 3.862.3 0.489Temora 10.664.7 43.667.0 ,0.001Other zooplankton 0.760.7 2.060.8 0.027

Body volume (mm3 individual21) 0.05660.006 0.04360.002 0.001Biomass (mg wet wt L21) 0.06760.023 0.23560.058 ,0.001Total astaxanthin availability (mg L21) 2.660.8 12.063.3 ,0.001

Astaxanthin concentrations (mg g wet wt21)

Free all-E-astaxanthin 20.362.9 26.063.5 0.027Free 9Z-astaxanthin 1.960.6 1.660.3 0.440Free 13Z-astaxanthin 5.160.9 3.060.6 0.002Astaxanthin monoesters 8.360.8 10.261.7 0.061Astaxanthin diesters 5.161.0 10.063.7 0.020Total astaxanthin 40.764.3 50.763.7 0.004

% of free astaxanthin

All-E-astaxanthin 74.563.4 84.662.9 ,0.0019Z-astaxanthin 6.861.7 5.461.5 0.24413Z-astaxanthin 18.761.9 10.062.0 ,0.001

Astaxanthin dynamics in fish 1163

winter, herring from 14 to 15 cm in length already feed onnektobenthic organisms (Casini et al. 2004), which may bescarce and/or low in astaxanthin.

Astaxanthin levels in sprat and herring—No previousdata on astaxanthin levels in sprat, herring, or other clupeidfish, to which our data could be compared, were found inthe literature. Sprat contained higher astaxanthin concen-trations than herring and cod, which may indicate that thesprat does not face a lack of astaxanthin in the Baltic Seaproper. Herring showed several signs of a possible lack ofastaxanthin, especially in November. For example, herringliver concentrations in November were eight times lowerthan in March, whereas they were stable and high in spratin both seasons. Herring stomachs contained almost noastaxanthin in November in all three ICES areas, whereassprat stomachs contained less astaxanthin in Novemberonly in Area 25. The differences in astaxanthin concentra-tions between the two clupeids are mainly due to relativegonad size (larger in herring) and stomach astaxanthincontent (higher in sprat). In other body parts, astaxanthinconcentrations are much lower than in gonads andstomach. Thus, whole-body astaxanthin concentrations ofthe clupeids are strongly influenced by their food intake.Low stomach astaxanthin content in clupeids decreasestotal astaxanthin transfer to higher trophic levels.

Astaxanthin levels in salmon—Our data from Sta. 99confirm the results of previous studies that wild salmon inthe Baltic Sea proper has low astaxanthin concentrationscompared to its Atlantic counterpart. Average muscleconcentration was as low as 0.033 mg g21. In a previouscomprehensive study, Pettersson and Lignell (1998) report-ed an average of 0.2 mg g21 for wild spawning females fromthe Baltic Sea. This can be compared to 0.5 mg g21 forAtlantic salmon living in freshwater Lake Vanern (Sweden)and 2.4 mg g21 for the North Atlantic (Pettersson andLignell 1998). Schiedt et al. (1981) measured concentrationsof 3–8 mg g21 for wild salmons caught at different places inEurope and Canada. Cultivated Atlantic salmon fed withastaxanthin in the diet usually contains 3–11 mg g21

(Czeczuga et al. 2005). Our salmon gonad concentrationsof between 0.4 and 1.5 mg g21 correspond with the lowerhalf of the range 0.5–3.1 mg g21 for mature eggs of wildAtlantic salmon in the Baltic Sea reported by Petterssonand Lignell (1999). They defined 0.74 mg g21 as a thresholdconcentration level below which healthy egg batches (notdisplaying the M74 syndrome) do not occur. Theseconcentrations can be compared with 2.6–4.4 and 5.6–7.9 mg g21 for mature eggs of wild Atlantic salmon fromLake Vanern and the North Atlantic, respectively (Petters-son and Lignell 1999). Our low astaxanthin values are notrelated to the use of different analytic methodologiesbecause we used the same method for astaxanthinextraction as Pettersson and Lignell (1998, 1999), but thegonads in our study may have been less mature.

Astaxanthin levels in cod—Cod in the Baltic Sea properdo not seem to experience a lack of astaxanthin comparedto their Atlantic counterparts. The cod is a fish with low

natural astaxanthin concentrations in its muscle tissue.Muscle concentrations in the Baltic Sea were on average0.02 mg g21, the same as those reported by Ytrestøyl andBjerkeng (2007a) from the Atlantic Ocean. Ytrestøyl andBjerkeng (2007a) also showed, by intraperitoneal injectionswith a high astaxanthin dose, that the low-fat muscle of codcan hold a maximum of 1.2 mg astaxanthin g21. This is 25times lower than the high-fat muscle tissue of salmon,which can hold a maximum of 30 mg g21. In cod wemeasured average gonad concentrations of 3.9 mg astax-anthin g21, which is higher than previously reported.Grung et al. (1993) measured 0.5 mg g21 of astaxanthin inmature eggs of wild cod and 0.2–0.4 mg g21 in mature eggsof farmed cod. Salze et al. (2005) reported even lowerconcentrations (wild 0.14 mg g21, farmed 0.05 mg g21).However, the latter study used a detection method in whichastaxanthin may be lost by first extracting the total lipidfraction and then analyzing astaxanthin only in thisfraction. The 40- to 50-fold variation in our tissue samplesof cod muscle and gonads reflects a large variation betweenindividuals, which in the case of the gonads is mostprobably related to the degree of gonad maturation.

Selective uptake of geometrical Z-isomers in sprat andherring—The proportions of Z-isomers in whole-bodysprat (50–54%) and herring (68–76%) were considerablyhigher than those in the mesozooplankton (15–25%). Themuscle and liver of the clupeids directly reflected the E : Zratio in the stomach (including contents), 36–38% in spratand 44–49% in herring. The highest proportions of Z-isomers were found in the gonads, c. 75% in both species.This is the opposite to salmon and cod, which had lowgonadal proportions of Z-isomers (18% and 26%, respec-tively). These observations suggest selective accumulation,and perhaps also isomerization, of astaxanthin Z-isomersin clupeids. It is unknown if accumulation of Z-isomers is ageneral feature for clupeid fish. In our study, herringgenerally had higher proportions of Z-isomers than sprat,which may depend on differences in prey selectivity or fishphysiology. In turn, the E : Z ratios in crustaceans maydepend on their microbial diet. Copepods experimentallyfed with a diet containing higher amounts of b-caroteneand zeaxanthin produced high levels of astaxanthin andhigher E : Z ratios (Rhodes 2007).

Selective uptake of geometrical E-isomers in salmon andcod—In salmon we recorded 75% Z-astaxanthin in the livercompared with 25% in muscle and 18% in gonads.Salmonid muscle has reportedly low proportions of Z-isomers, e.g., 3% in Atlantic salmon (Bjerkeng and Berge2000) and 5–6% in rainbow trout (Bjerkeng et al. 1997).Compared to these published data, our salmon musclesamples contained a high proportion of Z-isomers. It isprobable that wild (Baltic) salmon has higher proportionsof Z-isomers than the cultivated salmon used by Bjerkenget al. (1997) and others because of nutritional constraints.Baltic Sea salmon nearly exclusively feeds on sprat andherring, whereas cultivated salmon receives all-E-astax-anthin as a food additive. A high proportion of Z-isomersin the liver agrees with previous feeding experiments with

1164 Nie et al.

salmonids, which suggest that a selective mechanismaccumulates Z-isomers in the liver. A possible explanationwould be isomerization in the liver as initially suggested byBjerkeng et al. (1997), but high proportions of Z-isomers inthe liver rather support the observed accumulation of all-E-astaxanthin in the blood plasma and muscle by filtering outZ-isomers in the liver (Bjerkeng and Berge 2000). Anothermechanism explaining the accumulation of all-E-astax-anthin in blood and muscle of salmonids fish is that theintestinal absorption of the Z-isomers is less efficient thanthat of all-E-astaxanthin (Bjerkeng et al. 1997; Østerlie et al.1999; Bjerkeng and Berge 2000). In experiments with oraladministration of different astaxanthin E and Z isomericcompositions in the fish feed, the fecal percentage of Z-isomers was significantly higher than that in the feed (Østerlieet al. 1999). It can be argued that Z-isomerization of all-E-astaxanthin takes place during gastrointestinal residence.However, as discussed by Østerlie et al. (1999), higher fecallevels, lower intestinal tissue levels, and lower blood plasmalevels of total astaxanthin in trout fed with E : Z 5 64 : 36compared to trout fed with E : Z 5 97 : 3 strongly supportselective intestinal absorption of the all-E isomer withsubsequent incorporation into blood transport proteins anddeliverance to muscle cells. A third mechanism, selectiveuptake of all-E-astaxanthin by muscle cells from plasma, wasdiscovered when astaxanthin (E : Z 5 75 : 25) was adminis-tered by intraperitoneal injection to Atlantic salmon andrainbow trout (Ytrestøyl and Bjerkeng 2007a,b). In cod werecorded 50% of Z-astaxanthin of the total astaxanthin in theliver compared with 37% in muscle and 26% in gonads. Likesalmon, selective uptake of all-E-astaxanthin in Atlantic codhas previously been reported by Ytrestøyl and Bjerkeng(2007a), which is here confirmed by our data.

Astaxanthin transfer to salmon and cod—Herring andsprat in the Baltic Sea form huge schools constituting c.97% of the pelagic biomass (Orlowski 2001). Atlanticsalmon feeds opportunistically on available prey, which inthe Baltic Sea almost exclusively consists of herring andsprat (Hansson et al. 2001). Adult salmon does not prey oncrustaceans in the Baltic Sea, because the largest availablecrustaceans in the pelagic zone are mysids of 1.5 to 2 cm inlength, which are less abundant and smaller in size than krillin marine waters (Conti et al. 2005). In the North AtlanticOcean, salmon consumes a highly varied diet, and up to 80%of the prey can consist of crustaceans (Jacobsen and Hansen1996). Thus, in contrast to its Atlantic counterpart, theBaltic salmon acquires practically all its astaxanthin fromthe low body concentrations of herring and sprat, which inaddition are dominated by Z-isomers. This most probablyexplains the extremely low astaxanthin concentrations in theBaltic salmon populations. If clupeid fish consume a highnumber of zooplankton crustaceans, their stomachs willcontain high concentrations of astaxanthin with a highproportion of the all-E-isomer, which has the highestbioavailability for piscivorous fish. Thus, a considerablepart of the astaxanthin available to salmon and codoriginates directly from crustaceans in the clupeid stomach.

Cod did not have lower astaxanthin levels than itsAtlantic counterpart, which may be attributed to its lower

need for astaxanthin and higher food diversity. Cod in theBaltic Sea feeds on a much more varied diet than doessalmon, with roughly 25–50% herring and sprat, 20–50%Saduria entomon L., and 20–50% mysids, polychaetes, andother invertebrates (Uzars 1994). While mysids do containastaxanthin (Allahpichay et al. 1984), the isopod S.entomon has been reported not to contain astaxanthin(Czeczuga 1996). Thus, in the Baltic Sea mysids mayprovide an astaxanthin source for cod but not for salmon.

Food-web changes and astaxanthin dynamics—Declininghealth of Baltic fish has emerged in concert with threshold-like ecological shifts in the pelagic system of the Baltic Seaproper during the past century (Casini et al. 2009;Mollmann et al. 2009). There are several reports that theherring has decreased muscle fat, growth rate, anddecreased condition in the Baltic Sea proper (Cardinaleand Arrhenius 2000; Casini et al. 2010), which is in linewith low astaxanthin concentrations as a biomarker forgeneral health. Although the fat content of the Balticherring possibly has increased somewhat since 1995 (A.Bignert pers. comm.), it is still leaner than it is in theBothnian Sea north of the Baltic Sea proper (Bignert et al.2007). These changes in the herring are thought to becaused by increased feeding competition among clupeids inthe Baltic Sea proper (Casini et al. 2010) and/or docu-mented shifts in zooplankton abundance and composition(Mollmann et al. 2005). Baltic sprat biomass has increasedbecause the populations of the sprat’s main predator, theAtlantic cod, have decreased through fishing mortality andrecruitment problems (Cardinale and Arrhenius 2000;Casini et al. 2006, 2009). The low astaxanthin concentra-tions in herring liver and muscle are probably directlycaused by low astaxanthin intake, as seen from thedifferences between March and November in stomach,liver, and muscle, especially in ICES Area 28. Loss of bodyfat may also directly affect body astaxanthin contentbecause the pigment is hydrophobic and uptake andtransport are fat-dependent (Torrissen et al. 1989).

The M74 syndrome in the Baltic salmon is correlated tooxidative stress and low astaxanthin and thiamine levels(Pettersson and Lignell 1998, 1999; Amcoff et al. 1999). Asimilar reproductive disturbance related to thiaminedeficiency, early mortality syndrome (EMS), is found insalmonid fish in the Great Lakes in North America. This isexplained by the introduction of a clupeid fish with highthiaminase activity, the alewife (Honeyfield et al. 2005). Nolarge changes in the feeding ecology of the Baltic salmonhave occurred. Hansson et al. (2001) compared salmonstomach contents before (1959–1962) and after (1994–1997)the outbreak of M74. They found that the consumption ofsprat decreased from 88% to 77% in January–April, whileconsumption of herring increased from 9% to 22% (byweight). In September–December consumption of spratdecreased from 93% to 58%, and consumption of herringincreased from 6% to 37% (data extrapolated from fig. 8ain Hansson et al. 2001). However, it cannot be excludedthat increased consumption of herring lowers thiamine andastaxanthin levels in salmon because herring has a 10-foldhigher thiaminase activity than sprat (Wistbacka and

Astaxanthin dynamics in fish 1165

Bylund 2008), lower astaxanthin concentrations, and ahigher proportion of Z-isomers than sprat (this study).

The results from this study show that knowledge of thecomposition of essential compounds as a measure of foodquality is important for fishery management. This is especiallyimportant in areas that, like the Baltic Sea, are low inbiodiversity so that food choice for aquatic animals is limited.

AcknowledgmentsWe are grateful to Ake Lignell, Anders Bignert, Clare

Bradshaw, Peter Sylvander, Olle Hjerne, and two anonymousreviewers for constructive comments to the manuscript, to theSwedish Board of Fisheries for making available offshore ship-time for our project, to the crew of R/V Argos for their hospitalityand help in many ways, and to Ann-Christin Rudolphi, expeditionleader of the two cruises, for organizational skills and flexibility.For practical assistance in collecting the fish tissue samples wethank Erik Karlstrom, Kalle Haikonen, Markus Forslund, JormaLeijonmalm, Kristina Leijonmalm, and Richard Wennerholm.The research presented in this paper was financed by a visitingscientist scholarship to X.-P.N. (China Scholarship Council) andresearch grants to P.S. (Formas 21.9/2003-1033, Formas 21.0/2004-0313, EU Strukturstod FiV Dnr 231-0692-04).

References

ALLAHPICHAY, I., C. SHIMIZU, AND M. KONO. 1984. Pigmentationof cultured red sea bream, Chrysophrys major, usingastaxanthin from Antarctic krill, Euphausia superba, and amysid, Neomysis sp. Aquaculture 38: 45–57, doi:10.1016/0044-8486(84)90136-4

AMCOFF, P., H. BORJESSON, P. LANDERGREN, L. VALLIN, AND L.NORRGREN. 1999. Thiamine (vitamin B1) concentrations insalmon (Salmo salar) brown trout (Salmo trutta) and cod(Gadus morhua) from the Baltic Sea. Ambio 28: 48–54.

BELL, J. G., J. MCEVOY, D. R. TOCHER, AND J. R. SARGENT. 2000.Depletion of a-tocopherol and astaxanthin in Atlantic salmon(Salmo salar) affects autoxidative defence and fatty acidmetabolism. J. Nutr. 130: 1800–1808.

BIGNERT, A., E. NYBERG, K. L. SUNDQVIST, AND K. WIBERG. 2007.Spatial variation in concentrations and patterns of the PCDD/Fand dioxin-like-PCB content in herring from the northern BalticSea. J. Environ. Monit. 9: 550–556, doi:10.1039/b700667e

BJERKENG, B. 2008. Carotenoids in aquaculture: Fish andcrustaceans, p. 237–254. In G. Britton, S. Liaanen-Jensen,and H. Pfander [eds.], Carotenoids, v. 4: Natural functions.Birkhauser Verlag.

———, AND G. M. BERGE. 2000. Apparent digestibility coeffi-cients and accumulation of astaxanthin E/Z isomers inAtlantic salmon (Salmo salar L.) and Atlantic halibut(Hippoglossus hippoglossus L.). Comp. Biochem. Physiol. B127: 423–432, doi:10.1016/S0305-0491(00)00278-9

———, M. FØLLING, S. LAGOCKI, T. STOREBAKKEN, J. J. OLLI, AND

N. ALSTED. 1997. Bioavailability of all-E-astaxanthin and Z-isomers of astaxanthin in rainbow trout (Oncorhynchus mykiss).Aquaculture 157: 63–82, doi:10.1016/S0044-8486(97)00146-4

CARDINALE, M., AND F. ARRHENIUS. 2000. Decreasing weight-at-age of Atlantic herring (Clupea harengus) from the Baltic Seabetween 1986 and 1996: A statistical analysis. ICES J. Mar.Sci. 57: 882–893, doi:10.1006/jmsc.2000.0575

CASINI, M., V. BARTOLINO, J. C. MOLINERO, AND G. KORNILOVS.2010. Linking fisheries, trophic interactions and climate:Threshold dynamics drive herring Clupea harengus growth inthe central Baltic Sea. Mar. Ecol. Prog. Ser. 413: 241–252,doi:10.3354/meps08592

———, M. CARDINALE, AND F. ARRHENIUS. 2004. Feedingpreferences of herring (Clupea harengus) and sprat (Sprattussprattus) in the southern Baltic Sea. ICES J. Mar. Sci. 61:1267–1277, doi:10.1016/j.icesjms.2003.12.011

———, ———, AND J. HJELM. 2006. Inter-annual variation inherring, Clupea harengus, and sprat, Sprattus sprattus,condition in the central Baltic Sea: What gives the tune?Oikos 112: 638–650, doi:10.1111/j.0030-1299.2006.13860.x

———, AND oTHERS. 2009. Trophic cascades promote threshold-like shifts in pelagic marine ecosystems. Proc. Natl. Acad. Sci.USA 106: 197–201, doi:10.1073/pnas.0806649105

CHOJNACKI, J. 1983. Standard weights of the Pomeranian Baycopepods. Int. Rev. Ges. Hydrobiol. 68: 435–441, doi:10.1002/iroh.19830680314

CONTI, S. G., D. A. DEMER, AND A. S. BRIERLEY. 2005. Broad-bandwidth, sound scattering, and absorption from krill(Meganyctiphanes norvegica), mysids (Praunus flexuosus andNeomysis integer), and shrimp (Crangon crangon). ICES J.Mar. Sci. 62: 956–965, doi:10.1016/j.icesjms.2005.01.024

CZECZUGA, B. 1996. Carotenoprotein complexes in Mesidoteaentomon (L.) (Crustacea: Isopoda) from Baltic Sea. Folia Biol.(Krakow) 44: 73–77.

———, E. CZECZUGA-SEMENIUK, AND S. V. TRYGGVADOTTIR. 2005.Carotenoid content in various body parts of Atlantic salmon(Salmo salar L.) and Arctic charr (Salvenus alpinus L.)specimens from an ocean ranching farm. Oceanol. Hydrobiol.Stud. 34: 31–42.

FOSS, P., B. RENSTRØM, AND S. LIAANEN-JENSEN. 1987. Naturaloccurrence of enantiomeric and meso astaxanthin. 7. Crusta-ceans including zooplankton. Comp. Biochem. Physiol. B 86:313–314, doi:10.1016/0305-0491(87)90298-7

GRUNG, M., Y. S. SVENDSEN, AND S. LIAANEN-JENSEN. 1993. Thecarotenoids of eggs of wild and farmed cod. Comp. Biochem.Physiol. B 106: 237–242, doi:10.1016/0305-0491(93)90294-F

HANSSON, S., AND oTHERS. 2001. Stomach analyses of Balticsalmon from 1959–1962 and 1994–1997: Possible relationsbetween diet and yolk-sac-fry mortality (M74). J. Fish Biol.58: 1730–1745.

HERNROTH, L. [ED.], Recommendations on methods for marinebiological studies in the Baltic Sea—mesozooplankton bio-mass assessment. Baltic Mar. Biol. 10: 1–32.

HIGUERA-CIAPARA, I., L. FELIX-VALENZUELA, AND F. M. GOYCOO-

LEA. 2006. Astaxanthin: A review of its chemistry andapplications. Crit. Rev. Food Sci. Nutr. 46: 185–196, doi:10.1080/10408690590957188

HONEYFIELD, D. C., J. P. HINTERKOPF, J. D. FITZSIMONS, D. E.TILLITT, J. L. ZAJICEK, AND S. B. BROWN. 2005. Development ofthiamine deficiencies and early mortality syndrome in lake troutby feeding experimental and feral fish diets containing thiami-nase. J. Aquat. Anim. Health 17: 4–12, doi:10.1577/H03-078.1

IZQUIERDO, M. S., H. FERNANDEZ-PALACIOS, AND A. G. J. TACON.2001. Effect of broodstock nutrition on reproductive perfor-mance of fish. Aquaculture 197: 25–42, doi:10.1016/S0044-8486(01)00581-6

JACOBSEN, J. A., AND L. P. HANSEN. 1996. The food of Atlanticsalmon, Salmo salar L., north of the Faroe Islands.Conference and Meeting Document M 10. Anadromous andCatadromous Fish Committee, ICES Annual Science Con-ference, 1996, Reykjavik, Iceland.

LESSER, M. P. 2006. Oxidative stress in marine environments:Biochemistry and physiological ecology. Annu. Rev. Physiol.68: 253–278, doi:10.1146/annurev.physiol.68.040104.110001

LIAANEN-JENSEN, S., AND B. F. LUTNAES. 2008. E/Z isomers andisomerization, p. 17–36. In G. Britton, S. Liaanen-Jensen, andH. Pfander [eds.], Carotenoids, v. 4, Natural functions.Birkhauser Verlag.

1166 Nie et al.

LIU, X., AND T. OSAWA. 2007. Cis astaxanthin and especially 9-cisastaxanthin exhibits a higher antioxidant activity in vitrocompared to the all-trans isomer. Biochem. Biophys. Res.Commun. 357: 187–193, doi:10.1016/j.bbrc.2007.03.120

ŁOTOCKA, M., E. STYCZYNSKA-JUREWICZ, AND L. A. BŁEDZKI. 2004.Changes in carotenoid composition in different developmen-tal stages of copepods: Pseudocalanus acuspes Giesbrecht andAcartia spp. J. Plankton Res. 26: 159–166, doi:10.1093/plankt/fbh021

MALTBY, J. B., L. J. ALBRIGHT, C. J. KENNEDY, AND D. A. HIGGS.2003. Effect of route of administration and carrier onbioavailability and kinetics of astaxanthin in Atlantic salmonSalmo salar L. Aquacult. Res. 34: 829–838, doi:10.1046/j.1365-2109.2003.00888.x

MATSUNO, T. 2001. Aquatic animal carotenoids. Fish. Sci. 67:771–783, doi:10.1046/j.1444-2906.2001.00323.x

MAUCHLINE, J. 1998. The biology of calanoid copepods. AcademicPress.

MIKI, W. 1991. Biological functions and activities of animalcarotenoids. Pure Appl. Chem. 63: 141–146, doi:10.1351/pac199163010141

MOLLMANN, C., R. DIEKMANN, B. MULLER-KARULIS, G. KORNI-

LOVS, M. PLIKSHS, AND P. AXE. 2009. Reorganization of a largemarine ecosystem due to atmospheric and anthropogenicpressure: A discontinuous regime shift in the central BalticSea. Global Change Biol. 15: 1377–1393, doi:10.1111/j.1365-2486.2008.01814.x

———, G. KORNILOVS, M. FETTER, AND F. W. KOSTER. 2005.Climate, zooplankton and pelagic fish growth in the centralBaltic Sea. ICES J. Mar. Sci. 62: 1270–1280, doi:10.1016/j.icesjms.2005.04.021

ORLOWSKI, A. 2001. Behavioural and physical effect on acousticmeasurements of Baltic fish within a diel cycle. ICES J. Mar.Sci. 58: 1174–1183, doi:10.1006/jmsc.2001.1117

ØSTERLIE, M., B. BJERKENG, AND S. LIAANEN-JENSEN. 1999.Accumulation of astaxanthin all-E, 9Z and 13Z geometricalisomers and 3 and 39RS optical isomers in rainbow trout(Oncorhynchus mykiss) is selective. J. Nutr. 129: 391–398.

PETTERSSON, A., AND A. LIGNELL. 1998. Low astaxanthin levels inBaltic salmon exhibiting the M74 syndrome, p. 26–30. In G.McDonald, J. D. Fitzsimons, and D. C. Honeyfield [eds.],Early life stage mortality syndrome in fishes of the GreatLakes and Baltic Sea. Am. Fish. Soc. Symp. 21.

———, AND ———. 1999. Astaxanthin deficiency in eggs and fryof the Baltic salmon (Salmo salar) with the M74 syndrome.Ambio 28: 43–47.

PICKOVA, J., A. KIESSLING, A. PETTERSSON, AND P. DUTTA. 1999.Fatty acid and carotenoid composition of eggs from twononanadromous Atlantic salmon stocks of cultured and wildorigin. Fish. Physiol. Biochem. 21: 147–156, doi:10.1023/A:1007860908911

RHODES, A. C. E. 2007. Dietary effects on carotenoid compositionin the marine harpacticoid copepod Nitokra lacustris. J.Plankton Res. 29: i73–i83, doi:10.1093/plankt/fbl068

SALZE, G., D. R. TOCHER, W. J. ROY, AND D. A. ROBERTSON. 2005.Egg quality determinants in cod (Gadus morhua L.): Eggperformance and lipids in eggs from farmed and wildbroodstock. Aquacult. Res. 36: 1488–1499, doi:10.1111/j.1365-2109.2005.01367.x

SCHIEDT, K., F. J. LEUENBERGER, AND M. VECCHI. 1981. Naturaloccurrence of enantiomeric and meso-astaxanthin. 5. Ex Wildsalmon (Salmo salar and Oncorhynchus). Helv. Chim. Acta64: 449–457, doi:10.1002/hlca.19810640209

———, ———, ———, AND E. GLINZ. 1985. Absorption,retention and metabolic transformations of carotenoids inrainbow trout, salmon and chicken. Pure Appl. Chem. 57:685–692, doi:10.1351/pac198557050685

SOMMER, F., C. AGURTO, P. HENRIKSEN, AND T. KIØRBOE. 2006.Astaxanthin in the calanoid copepod Calanus helgolandicus:Dynamics of esterification and vertical distribution in theGerman Bight, North Sea. Mar. Ecol. Prog. Ser. 319: 167–173, doi:10.3354/meps319167

STOREBAKKEN, T. 1988. Krill as a potential feed source forsalmonids. Aquaculture 70: 193–205, doi:10.1016/0044-8486(88)90096-8

TER BRAAK, C. J., AND P. SMILAUER. 2002. CANOCO referencemanual and user’s guide to CANOCO for Windows: Softwarefor canonical community ordination (version 4.5). Micro-computer Power.

TORRISSEN, O. J., R. W. HARDY, AND K. D. SHEARER. 1989.Pigmentation of salmonids—carotenoid deposition and me-tabolism. CRC Crit. Rev. Aquat. Sci. 1: 209–225.

UZARS, D. 1994. Feeding of cod (Gadus morhua callarias L.) in thecentral Baltic in relation to environmental changes. ICESMar. Sci. Symp. 198: 612–623.

VAN DER MEEREN, T., R. E. OLSEN, K. HAMRE, AND H. J. FYHN.2008. Biochemical composition of copepods for evaluation offeed quality in production of juvenile marine fish. Aquacul-ture 274: 375–397, doi:10.1016/j.aquaculture.2007.11.041

VAN NIEUWERBURGH, L., I. WANSTRAND, J. LIU, AND P. SNOEIJS.2005. Astaxanthin production in marine pelagic copepodsgrazing on two different phytoplankton diets. J. Sea Res. 53:147–160, doi:10.1016/j.seares.2004.07.003

VIITASALO, M., J. FLINKMAN, AND M. VIHERLUOTO. 2001.Zooplanktivory in the Baltic Sea: A comparison of preyselectivity by Clupea harengus and Mysis mixta, with referenceto prey escape reactions. Mar. Ecol. Prog. Ser. 216: 191–200,doi:10.3354/meps216191

WISTBACKA, S., AND G. BYLUND. 2008. Thiaminase activity ofBaltic salmon prey species: A comparison of net- andpredator-caught samples. J. Fish Biol. 72: 787–802, doi:10.1111/j.1095-8649.2007.01722.x

YTRESTØYL, T., AND B. BJERKENG. 2007a. Dose response in uptakeand deposition of intraperitoneally administered astaxanthinin Atlantic salmon (Salmo salar L.) and Atlantic cod (Gadusmorhua L.). Aquaculture 263: 179–191, doi:10.1016/j.aquaculture.2006.10.021

———, AND ———. 2007b. Intraperitoneal and dietary adminis-tration of astaxanthin in rainbow trout (Oncorhynchusmykiss)—plasma uptake and tissue distribution of geometri-cal E/Z isomers. Comp. Biochem. Physiol. B 147: 250–259,doi:10.1016/j.cbpb.2007.01.010

YUAN, J. P., AND F. CHEN. 1999. Isomerization of trans-astaxanthin to cis-isomers in organic solvents. J. Agric. FoodChem. 47: 3656–3660, doi:10.1021/jf981319u

Associate editor: Robert R. Bidigare

Received: 21 August 2010Accepted: 06 January 2011

Amended: 15 December 2010

Astaxanthin dynamics in fish 1167