STABLE ISOTOPES ISSUE

M. E. Perga Æ D. Gerdeaux

‘Are fish what they eat’ all year round?

Received: 16 September 2004 / Accepted: 18 February 2005 / Published online: 11 May 2005� Springer-Verlag 2005

Abstract Isotope turnover in muscle of ectotherms de-pends primarily on growth rather than on metabolicreplacement. Ectotherms, such as fish, have a discontin-uous pattern of growth over the year, so the isotopic sig-nature of muscle (d13C and d15N) may only reflect foodconsumed during periods of growth. In contrast, the liveris a regulatory tissue, with a continuous protein turnover.Therefore, the isotopic composition of liver should re-spond year round to changes in the isotopic signature offood sources. Therefore, we predicted that (1)Whitefish inLake Geneva would have larger seasonal variation in theisotopic variation of the liver compared to that of themuscle tissue, and (2) the isotope composition of fishmuscle would reflect a long-term image of the isotopecomposition of the food consumed only throughout thegrowth period. To test these expectations, we comparedthe isotope compositions of Whitefish muscle, liver andfood in a 20-month study. We found that the seasonalamplitude of isotope variation was two to three timeshigher in liver compared to muscle tissue. Duringthe autumn andwinter, when growthwas limited, only theisotopic signature of liver responded to changes in theisotope composition of the food sources. The d13C andd15N of muscle tissue only reflected the food consumedduring the spring and summer growth period.

Keywords Isotope turnover rate Æ Ectotherms ÆStable isotope analyses Æ Whitefish Æ Growth

Introduction

Over the past 20 years, stable isotope analysis (SIA) hasbeen extensively used to study food webs in variousecosystems. During this time, questions have emergedabout the underlying assumptions on which this tech-nique relies (Tieszen et al. 1983; Hobson and Clark 1992;Gannes et al. 1997), many of which concern isotopeturnover. Isotope turnover is defined as the change intissue isotope composition attributable to growth andmetabolic tissue replacement (Mac Avoy et al. 2001).The isotope turnover rate defines the delay necessary tothe isotope composition of the tissue of the consumer toreach equilibrium with that of the food source. Isotopeturnover has undergone numerous experimental studies,involving various species, mainly endotherms (Tieszenet al. 1983; Kelly 2000; Roth and Hobson 2000; Kurle2002; Kurle and Worthy 2002; Hobson and Bairlein2003), and to a lesser extent, ectotherms, such as krill(Frazer et al. 1997), insects (O’Brien et al. 2000) and fish(Hesslein et al. 1993; Pinnegar and Polunin 1999; Herzkaand Holt 2000). Isotope turnover rates have beendetermined for different tissues of captive organisms,following an isotopic switch in their diet. In endotherms,the isotope composition of organs with a fast turnover(such as liver) changed their isotopic composition morerapidly than slower growing tissues (such as muscle)(Tieszen et al. 1983; Hobson and Clark 1992). Little isknown about the isotope turnover process in ecto-therms, such as fish, even though most recent SIAstudies have been performed in aquatic ecosystems.Many questions concerning the isotope turnover rate infish therefore need to be addressed.

Stable isotope analysis of aquatic food webs is usuallyperformed on fish dorsal muscle, which is assumed toreflect the isotope signature of the food consumed. Incontrast to endotherms, for which SIA can be performedon various tissues to approximate feeding history ondifferent time scales, there is no clear evidence that liverand muscle can be used in tandem to obtain different

Communicated by Jim Ehleringer

M. E. Perga (&) Æ D. GerdeauxStation d’Hydrobiologie Lacustre, INRA,75, Av. de Corzent B.P. 511,74203 Thonon les Bains Cedex, FranceE-mail: mperga@uvic.caTel.: +1-250-4724739Fax: +1-250-7217120

Present address: M. E. PergaDepartment of Biology, University of Victoria,PO Box 3020, Stn CSC, Victoria, BC, CanadaV8W 3N5

Oecologia (2005) 144: 598–606DOI 10.1007/s00442-005-0069-5

time-scale for fish diets. The few studies involving fishhave reached contradictory conclusions about turnoverrates in muscle and liver (Hesslein et al. 1993; Doucettet al. 1999; Johnson et al. 2002). The isotope composi-tion of food sources may vary continuously during theyear, especially in aquatic ecosystems, where the isotopecomposition of plankton can vary by 20& (Zohary et al.1994). In light of seasonal variability of plankton d13Cand d15N, the isotope composition of the liver couldprovide a better short-term assessment of the isotopecomposition of the food in the field than muscle tissue.

Most of the studies of isotope turnover have so farrelied on laboratory experiments under controlled con-ditions with a constant food supply, resulting in con-stant fish growth (Hesslein et al. 1993; Pinnegar andPolunin 1999). However, in the field, animals, andespecially ectotherms, experience several different met-abolic phases during the year, including phases of so-matic growth, gonadic growth, and basal metabolism,during which no growth occurs. In fish, liver is a regu-latory tissue with a constant protein turnover, whereasmuscle is the most representative tissue of growth whenprotein synthesis and deposition are considered (McMillan and Houlihan 1989; Peragon et al. 1994; de laHiguera et al. 1999). Furthermore, muscle isotopeturnover depends mainly on growth (Hesslein et al.1993) and therefore muscle isotope turnover is likely tobe strongly affected by the consumer physiologicalphases. In particular, muscle tissue isotopic compositionmay not respond to changes in d13C and d15N of foodsources year round.

To tackle these issues, we compared the changes inthe d13C and d15N compositions of liver and muscletissue of Whitefish (Coregonus lavaretus) to changes inthe isotopic composition of their food on a monthlybasis. Whitefish was chosen because it is a zooplankti-vore fish (Mookerji et al. 1998), so the isotope compo-sition of its food varies year round (Zohary et al. 1994).Somatic growth of Whitefish in peri-alpine lakes haspreviously been measured by back-calculation fromscales and shown to occur from March to September(Caranhac 1999). We therefore expected that at any gi-ven time Whitefish muscle would reflect the isotopecomposition of the food eaten during the previousgrowth period, whereas that of the liver would reflect theisotope composition of food consumed throughout theyear, and respond more readily to wintertime dietchanges.

Material and methods

Study lake

Located on the Swiss-French border, Lake Geneva is thelargest lake in Western Europe (area=582 km2, maxdepth=309 m), with a catchment area of 7,400 km2 .Regarding phosphorus concentrations, Lake Geneva ismeso-eutrophic (http://www.cipel.org).

Fish sampling

There is little inter-individual variability in C. lavaretusstomach contents in large and deep lakes (Ponton 1986;Mookerji et al. 1998), and a sample of ten individualscan be taken to be representative of the C. lavaretuspopulation. From January 2002 to September 2003, tenadult Whitefish were bought every month from profes-sional fishermen in Lake Geneva. Fish could not besampled in November and December as fishing wasclosed to allow fish spawning. Fish were kept in a coolbox until they were processed in the laboratory withinan hour. The stomachs were removed, and the stomachcontents preserved in Ethanol 50%. The liver and partof the dorsal muscle were removed and frozen. We didnot remove lipids from our samples prior to analysis.The correction for the lipid content effect on muscled13C described by Mc Connaughey and Mc Roy (1979)has been shown to be applicable to Whitefish muscle(Dufour 1999). Moreover, as d15N tracks the origins ofprotein, and d13C tracks both proteins and lipids, theliver d13C and d15N data provided different information.Consequently, we did not remove fat from the liversamples, but the d13C data will be regarded along withthe C/N ratio as a proxy for lipid content.

Stomach contents

An appropriate aliquot volume extracted from the 50-mlhomogenized stomach content sample was placed in aDolphuss cell and the prey species present were identi-fied and counted under a binocular microscope. Themain prey identified were Cladocerans (Daphnia sp.,Leptodora kindtii and Bythotrephes sp.), Copepods(essentially Cyclopoıds), and Chironomid larvae andnymphs. The volume of each prey was assessed bymultiplying the number of individuals by a prey-dependent volumetric coefficient (Hyslop 1980). Thevolumetric proportion of each prey was then calculatedfor each individual Whitefish.

Zooplankton and benthos sampling

Each month, zooplankton was collected from severalvertical hauls with a 200-lm-mesh net from a depth of50 m to the surface. Bulk zooplankton was kept in avacuum bottle and maintained in lake-filtered waterovernight. The main zooplankton taxa were isolatedmanually under the binocular microscope and frozen.

In autumn and winter, Whitefish move to littoralzones just before spawning, and might, for this time,feed on chironomids larvae. Chironomids were thensampled in December 2003 in littoral areas of LakeGeneva. They were collected using an Eckman grab,along a depth gradient (5 m; 7 m; 16 m; 20 m). The gutwas removed under the binocular microscope, and thebody rinsed with demineralized water prior to freezing.

599

Stable isotope analysis

Fish, benthos and zooplankton samples were freeze-dried. Fish and benthos samples were then ground to afine powder in a ball mill. One milligram fractions of thesamples were weighed into 6·4 mm tin cups forCF-IRMS analysis using a Europa Scientific ANCA-NT20-20 Stable Isotope Analyser with a NCA-NT Solid/Liquid Preparation Module (PDZ Europa Ltd., Crewe,UK). The analytical precision (S.D.) was 0.2&, as esti-mated from five standards analyzed along with thesamples. Working standards were 1 mg leucine preparedby freeze-drying 50 ml of a 20 mg/ml stock solution inthe tin cups, and calibrated against ‘Europa flour’ andIAEA standards N1 and N2. The isotope ratios wereexpressed in d as the parts per thousand differences (&)from the standard references, which were PeeDeeBelemnite for d13C and atmospheric nitrogen for d15N.

Data processing

Estimation of the isotope composition of the diet

Whitefish in Lake Geneva is mainly zooplanktivorous.However, the proportion of chironomids to Whitefishdiet varies annually and the isotope composition ofchironomids depends on their depth. Hence, only theisotope composition of the pelagic part of the diet(which makes up more than 85% of the diet) was esti-mated in order to compare the difference in the isotopecompositions of the planktonic food and that ofWhitefish. Copepods and Daphnia could be sampledeach month, and their isotope composition measured. Incontrast, Leptodora and Bythotrephes were sometimesnot sufficiently abundant in the monthly samples fortheir isotope composition to be measured. As linearregressions between the isotope compositions of Lepto-dora or Bythotrephes versus Daphnia provided a gooddescriptor of these relationships (see Results), theregression equation was used to estimate the missingisotope composition values for Bythotrephes andLeptodora from those for Daphnia. The d13C and d15Nvalues for the diet were assessed by a mixing model. Asall zooplanktonic taxa had similar C/N values, we didnot incorporate the C and N concentrations into themixing models (Phillips and Koch 2002). The isotopecomposition of the zooplanktonic food was thereforeequal to the sum of the volumetric contributions of eachtaxa present multiplied by their respective monthlyisotope signatures.

Detecting significant changes in the monthly tissued13C, C/N ratios and d15N values

Homoscedasticity of the monthly muscle and liver C/N,d13C and d15N values was previously tested and rejected(Levene tests). We then relied on non-parametricalstatistical methods for our further data analyses. Any

significant changes in the isotope composition forWhitefish muscle and liver over the seasonal patternwere detected using the multiple comparisons Kruskal–Wallis test. To perform post-hoc non-parametric tests, amodified form of the Kruskal–Wallis test was used. AnS-Plus script of this modified test, which was createdby F. Mueter (mueter@ims.alaska.edu), is availablefrom the S-News forum http://www.biostat.wustl.edu/archives/html/s-news/1999-03/msg00245.html.

This test involves a Kruskal function modified toinclude multiple comparisons. The output includes asimple matrix of pairwise comparisons (only if theoverall test rejects the null hypothesis), where T indicateswhether the difference is significant at the given level ofalpha. The test is based on a multiple comparison testgiven in Conover (1980).

Significant monthly differences were then targeted,and adjoining groups of months with non-significantdifferences in Whitefish isotope compositions were cre-ated from the matrix of the post-hoc tests results. All thetests were performed on S-Plus (Insightful 2001).

Results

Gut contents

Whitefish in Lake Geneva relied on zooplanktonic preyfor 75–100% of their diet. In each of the years studied,this percentage was lowest in February and highest inMay–June and July (Fig. 1). Samples collected in Au-gust 2003 constituted an exception, as the Whitefish gutswere full of chironomids nymphs. Bythotrephes andDaphnia are the most abundant prey species. Bythotre-phes were predominant in the gut contents at the end ofwinter (January and February) and in the early summermonths (June–July), whereas Daphnia were predominantin spring. Benthic preys (chironomids) were consumedfrom February to April, when they accounted for 25%of the gut contents. The contribution of chironomids

Fig. 1 Whitefish gut contents during the 2002–2003 survey

600

was also considerable in the late summers of 2002 and2003 (Fig. 1). Benthic consumption accounted for 4.6%of the total consumption in 2002 and for 12.3% in 2003,and was significantly greater in 2003 than in 2002(p=0.03). More precisely, the benthic contribution inFebruary to April was twice as great in 2003 as in 2002,and in August 2003, the Whitefish fed almost exclusivelyupon benthos.

Comparative changes in d13C in zooplankton,and in Whitefish muscle and liver

Zooplankton and chironomid d13C values

Every month, all the zooplanktonic taxa displayed verysimilar d13C values (Fig. 2a). Linear regression wastherefore used to compare the d13C values of the zoo-planktonic taxa (Table 1). There were strong and sig-nificant correlations between the d13C values for all thetaxa. Missing d13C values for Bythotrephes longimanusand Leptodora kindtii were estimated from linear models(Fig. 2a). Then, the d13C of the diet was assessed with asimple mixing model.

The d13C value for the zooplankton in the diet rangedbetween �25& and �33& (Fig. 2a) between January2002 and September 2003. The pelagic prey d13C reachedits lowest value in early spring (March 2002 and Feb-ruary 2003), and its highest value in summer (August2002 and July 2003). The seasonal pattern was fairlysimilar for 2002 and 2003.

Chironomids had d13C values ranging from �19& to�27& (Fig. 2a). The d13C values of the benthic preyoverlapped the values for the late spring and summerpelagic diet, but were higher than the winter or earlyspring pelagic values. Benthic consumption was there-fore discernable from pelagic preys via d13C basis inFebruary to April, when benthos made a moderatecontribution to Whitefish diet. In contrast, the con-sumption of late summer nymphs was probably masked.

d13C of Whitefish muscle and liver

The d13C of Whitefish muscle showed significant differ-ences between different months (Fig. 2b) (Kruskal–Wallis v2=94.7088; df=18; p value=0). Homogeneousgroups of mean d13C values are shown in Fig. 2b.Minimum values were reached in April 2002 and April–May 2003. However, the lowest values reached in 2002were significantly lower than in 2003. The higher valueof the spring 2003 minimum, and the fact that it oc-

Fig. 2 Monthly variations of d13C values of: a Daphnia sp.,Bythotrephes longimanus (Byth.), Copepods, Leptodora kindtii(Lept.), and Chironomids (Chiro.), b Whitefish muscle and cWhitefish liver. d Monthly variations in the liver C/N ratio, used asa fat content proxy. Horizontal bars indicate homogeneous groupsof monthly means (Post-hoc test adapted from Conover (1980))

c

601

curred later may result from the higher proportion ofchironomids in the diet. The highest d13C values formuscle were reached in July, and remained unchangeduntil the following February, even though the d13C ofthe zooplankton food source fell steadily from �24& to�32&. The d13C values of the zooplanktonic prey andof the muscle exhibited parallel seasonal patterns fromFebruary to July each year, with the d13C pattern formuscle lagging 1 or 2 months behind that of the zoo-planktonic diet. In contrast, during the autumn andwinter of 2002–2003, muscle remained at the d13C valuereached during summer, whereas the d13C of the zoo-plankton prey were still decreasing. During this period(autumn and winter 2002–2003), the C/N ratios inWhitefish muscle did not show any significant change(KW test, v2=6.1435, df=6, p=0.40).

Liver d13C also varied significantly over the year (KWtest v2=140.20; df=18; p value <10�3). Liver d13Cfluctuations spanned a range of 3.5&, and were twice asgreat as those for muscle (Fig. 2c). Minimum valueswere reached in April 2002 and March–May 2003, andmaximum values in summer. The pattern in spring andsummer 2003 included some fluctuations that might re-flect the high proportion of benthic prey species in thediet during this period. Like the d13C for muscle, that forliver remained unchanged from the summer until thenext February, even though the d13C of the food sourcefell throughout the autumn and winter. However, unlikemuscle, the C/N ratio of the liver underwent majorchanges during this period (KW test, v2=18.94, df=3,p=0.003). The liver C/N ratio was greatest during thegrowth period each year and lowest at the end of winter(Fig. 2d). Lipids stored during the growth period weretherefore being consumed during the winter.

Comparative changes of the d15N values in zooplanktonand in Whitefish muscle and liver

d15N of the zooplanktonic food

The zooplanktonic taxa exhibited differing d15N values,but they all followed the same pattern of seasonal vari-ation (Fig. 3a). Linear models were used to describerelationships between the d15N values for the differenttaxa (Table 2, Fig. 3b). There were strong and signifi-cant correlations between the d15N values for all thetaxa. Missing d15N values for Bythotrephes and Lepto-dora were estimated from linear models.

General trends in the d15N of the zooplanktonic preydisplayed very similar seasonal trends in 2002 and 2003(Fig. 4a). The d15N of the pelagic prey was highest inwinter (17&) in both years, and lower in spring andsummer (12&), with a transient minimum in early spring(as low as 5&). The extremes values, reached in winteror late spring, are the same for both years surveyed. Themain difference is the steeper and earlier decrease ind15N values in 2003.

d15N values of chironomids ranged from 5.2& to 11.7& along a depth gradient and overlapped those forzooplankton in spring and summer.

Whitefish tissue d15N

The monthly means of the Whitefish muscle d15N variedsignificantly throughout the year (KW v2=117, df=18,p<10�3) (Fig. 4b), but only by 2&. In contrast, levels inthe prey spanned a range of 12&. Whitefish muscle d15Nwas greatest at the end of winter in both years, anddecreased in spring. Minimum values, reached as earlyas May, remained constant until September. However,

Table 1 Parameters of the linear regression model between d13C values for zooplankton taxa and probabilities associated with Student’stests for Lake Geneva. t0 is the t-value of the Student’s test comparing a to 0 and t1 is the t-value of the Student’s test comparing b to 0

Regression model r2 a (± SE) P(>t0) b (± SE) P(>t1)

d13CBythotrephes=a. d13CDaphnia + b 0.84 0.86 (± 0.11.) <10�3 �3.44 (± 3.32) 0.32d13CCopepods=a. d13CDaphnia + b 0.74 0.80 (± 0.13) <10�3 �5.74 (± 3.71) 0.14d13CLeptodora=a. d13CDaphnia + b 0.54 1.24 (± 0.43) 0.02 8.12 (± 11.78) 0.51

Fig. 3 a Zooplankton taxa d15N during the 2002–2003 survey.b Linear relationships between the d15N values for Bythotrephes,Leptodora and copepods and that of Daphnia

602

in January 2003, the muscle d15N was higher than that inNovember 2002, indicating an increase (of about 1&) inmuscle during the winter. Like the d15N of the zoo-

planktonic diet, the spring decrease occurred earlier in2003 than in 2002. The seasonal pattern of d15N forWhitefish muscle was similar throughout the survey,although benthic consumption in 2003 was double thatin 2002. However d15N values for benthic prey over-lapped those for zooplankton, which made it impossibleto identify benthic consumption from the Whitefishmuscle d15N.

To conclude, Whitefish muscle d15N displayed aseasonal pattern that mirrored that of the pelagic foodsource, with higher values in winter and lower ones inspring and summer. Whitefish muscle d15N increasedfrom October 2002 to February 2003, as did that of thediet.

Liver d15N also displayed significant changes over theyear (KW v2=175, df=18, p<10�3) but, unlike that ofthe muscle, the liver d15N varied all year long (Fig. 4c).The maximum d15N, was reached at the end of winter.There was a sharp decrease in d15N (�5&), and theminimum value of d15N was reached in May. The d15Nof the liver generally lagged one month behind that ofthe zooplankton, and the amplitude of liver d15N vari-ations was three times higher than that of muscle. As inzooplankton, this transient minimum was followed by arise of d15N up to 13& in summer. As in muscle, d15Nvalues also increased in winter (by about 2&). Changesin liver isotope composition reflected the same seasonalpattern as the pelagic food sources.

Discussion

We found that the isotope composition of the Whitefishfood sources varied continuously over our survey. Thereasons for these variations, and especially for seasonalvariations in the isotope compositions of the zooplank-tonic taxa, will not be addressed here. As d15N onlytracks proteins, and as lipids were not removed from oursamples, d15N is a more adequate tool for comparingtissue isotope turnover than d13C, which tracks bothproteins and lipids which are known to have differentisotope compositions and turnovers. The seasonal nat-ure of the zooplankton d15N was reflected by the con-sumer’s tissues after an interval of 1 month in the case ofthe liver and of 4–5 months in that of the muscle.Moreover, if liver and muscle have similar turnovers, assuggested by Hesslein et al. (1993), then the amplitudesof d15N variations in these two organs would have beenexpected to be the same. In contrast, we found that theamplitudes of d13C and d15N variation in the liver were

Table 2 Parameters of the linear regression model between the d15N of zooplankton taxa and probabilities associated with Student’s testsfor Lake Geneva. t0 is the t-value of the Student’s test comparing a to 0 and t1 is the t-value of the Student’s test comparing b to 0

Regression model r2 a (± SE) P(>t0) b (± SE) P(>t1)

d15NBythotrephes=a. d15NDaphnia + b 0.85 0.80 (± 0.10.) <10�3 5.46(± 0.91) <10�3

d15NCopepods=a. d15NDaphnia + b 0.85 0.90 (± 0.10) <10�3 3.26 (± 0.91) <10�3

d15NLeptodora=a. d15NDaphnia + b 0.90 0.88 (± 0.11) <10�3 3.93 (± 0.88) <10�3

Fig. 4 Monthly variations of d15N values of: a Daphnia sp.,Bythotrephes longimanus (Byth.), Copepods, Leptodora kindtii(Lept.), and Chironomids (Chiro.), b Whitefish muscle andc Whitefish liver. Horizontal bars indicate homogeneous groupsof monthly means (Post-hoc test adapted from Conover (1980))

603

three times greater than those in muscle suggesting thereis a difference in fish liver and muscle isotope turnoverrates, as previously shown both in birds (Hobson andClark 1992) and in mammals (Tieszen et al. 1983).Contradictions with the data previously reported byHesslein et al. (1993), who admitted that the conclusionsdrawn should be tempered because of limited samplesize, prompt us to confirm these results by furtherexperiments.

Monthly changes in tissue d13C and d15N wereinvestigated, and we confirmed that spring and summerd13C and d15N patterns were integrated into the tissues:liver and muscle d13C and d15N patterns matched thoseof the zooplankton. Liver and muscle d13C matched theearly spring decrease in zooplankton in 2002 and 2003and the increase that followed in the late spring andsummer to reach a balance with the maximal summervalues. In 2003, spring and summer muscle and liverd13C values were all higher than in 2002, probablyreflecting the higher benthic contribution to the diet inearly spring. Liver d15N matched the spring diet d15Ndecrease and the following summer increase to reachequilibrium with the diet d15N summer values. Becauseof its slower turnover, muscle d15N matched only thespring decrease in diet d15N. However, at first sight, thedata concerning food isotope composition of the twoisotopes seem to contradict each other during autumnand winter. Whitefish muscle and liver d13C remainedconstant from July to February, while the Whitefishwere still feeding, and zooplanktonic d13C was stilldecreasing. Neither muscle nor liver then seemed to re-flect the d13C of the zooplankton during autumn andwinter. On the contrary, muscle and liver d15N actuallyincreased even in winter, as did the zooplankton d15N,which would support the assumption of a continuousisotope turnover in both tissues.

Under the assumption of a continuous incorporationof diet isotope composition in muscle, the absence ofsignificant variations in Whitefish muscle and liver d13Cthroughout autumn and winter could occur only if thed13C of their diet had also not changed during this time,i.e., had remained equal to the summer value of �26&.This could occur if the Whitefish consumed enough13C-enriched chironomids to compensate for the d13Cdecrease in zooplankton during this period. We then

assessed the proportion of chironomids that would beneeded for the diet d13C to settle around �26& duringautumn and winter (Table 3). As the d13C of the chir-onomids depended on depth, we determined minimumand maximum percentages depending on the depth ofthe chironomids potentially consumed (Table 3). Withthis aim, a simple linear mixing model was used. Linearmixing models are usually the topic of controversy be-cause their key assumption is that C and N isotopesfrom each dietary source have similar isotopic routing,which might not be realistic unless C and N composi-tions of the dietary sources are not substantially different(Phillips and Koch 2002). However, zooplankton duringautumn and winter and chironomids had similar Ccontents (average C content for zooplankton: 42%;chironomids: 44%, U-test, p value=0.12), which enablesus to consider as realistic the assumption of similarC-isotopic routing of the nutrients provided by thezooplankton and chironomids dietary sources.

The observed proportion of chironomids in the gutduring October 2002 was within the range of thatestimated. However, for all the following months, thehypothetically required proportion of chironomids wasfar higher than the range actually observed in thisstudy or in previous surveys (Ponton 1986; Gerdeauxand Hamelet 2001). Chironomid preys were only sam-pled once and so seasonal variability of chironomidsd13C and d15N values between autumn and late winterwas assumed to be minor. Isotope variability of chir-onomids in cold temperate lakes is related to methane-derived carbon incorporation (Grey et al. 2004). Sea-sonal variability in chironomids d13C values was highin eutrophic lakes that experimented substantial anoxiain summer. In contrast, variability was lower than 1&(SD) in the mesotrophic lake studied, because sedi-ments did not undergo long periods of anoxia. LakeGeneva is mesotrophic and does not undergo anoxiaanytime. Whitefish feed on chironomids larvae in lateautumn and winter when they move to littoral areas tospawn. These littoral areas are always oxygenated andno methanotrophy occurs at these depths. Moreover,when chironomids incorporate methane-derived car-bon, the pattern of d13C decrease with depth occursalong with a d15N decrease. In contrast, in our dataset,the chironomids d13C gradient with depths is negatively

Table 3 Theoretical minimum and maximum percentages of benthic prey species in the gut contents required for the diet d13C in autumnand winter to remain equal to its summer 2002 value of �26.0& and the percentages actually recorded

Date d13C of thepelagic diet

Minimumproportion ofchironomids (&)

Maximum proportionof chironomids(%)

Proportion ofchironomidsrecorded (%)

October 2002 �26.7 10 64 25November 2002 �27.7 23 82December 2002 �28.9 32 88 0–5% (Ponton 1986;

Gerdeaux and Hamelet 2001)January 2003 �30.0 40 91 0–3February 2003 �31.5 47 93 20–40March 2003 �32.0 50 94 10–20

604

correlated with the d15N gradient, supporting the ideathat chironomids isotope variations with depth resultsfrom a littoral-pelagic carbon gradient rather thanfrom increasing incorporation of methane-derived car-bon with depths (but see further). It is then unlikelythat chironomids d13C values dramatically changedover autumn and winter.

In our mixing model, we assessed d13C values forchironomids in autumn and winter 2002–2003 fromthose of the individuals sampled in December 2003.Chironomids species are uni- or bi-voltine species inlittoral areas of Lake Geneva (B. Lods-Crozet, personalcommunication) and there might be inter-annual varia-tions in chironomids isotope compositions. We howeverbelieve that chironomids d13C values were similar be-tween years. d13C values of chironomids displayeddepth-dependent variations over a 8& range with lowervalues in deeper areas. The pelagic baseline d13C wasestimated, from Dreissena polymorpha sampled at thesame time, at �28.3& (SD=0.8; n=10). With increas-ing depths, chironomids values tend then to match thepelagic baseline values. Benthic primary producers’ d13Cusually displayed higher d13C values than pelagic ones(France 1995). Hence, the gradient in chironomids d13Cvalues with depths is very likely to result from a varia-tion in the contribution of periphytic to phytoplanktoncarbon in chironomids diets. Then, chironomids d13Cvalues strongly depend on those of phytoplankton andperiphyton. Zooplankton was shown to feed mainly onphytoplankton in Lake Geneva and patterns of zoo-plankton seasonal variations are fairly similar betweenthe 2 years (Perga, unpublished), suggesting there wereno strong variations in baseline d13C values between theyears 2002 and 2003. We then considered that the chir-onomids d13C variations with depths were likely toencompass the between-year variability in their isotopecomposition.

According to our results, it is unlikely that benthicconsumption could compensate for the decrease inzooplankton d13C from November 2002 to February2003, which implies that the d13C of the diet was notbeing recorded in muscle during autumn and winter. Ifthis is the case, the increase in d15N observed in muscleduring winter would not reflect the d15N of the diet. Inautumn and wintertime, diet resources are limited andnutrients in the muscle could be remobilized to supportgonadic growth or basal metabolism when all the lipidsreserves have been consumed, as has been reported inbirds (Hobson et al. 2000) and scallops (Lorrain et al.2002). Nutrient remobilization is known to increasetissue d15N, suggesting that in the context of starvationor of limited resources, animals catabolize their owntissue proteins (Frazer et al. 1997; Adams and Sterner2000; Olive et al. 2003). In this situation, changes inmuscle d15N during the winter could result from changesin physiology rather than in diet. This means that theisotope composition of muscle cannot provide reliableinformation about the food consumed during seasonswhen there is no somatic growth.

In the case of the liver, the apparent d13C patternsalso support the hypothesis that diet d13C is not reflectedby this organ during the winter, whereas the d15N of theliver matched that of the zooplanktonic all year round.For the liver, fat should also be taken into consideration.While muscle fat content remained low and constantthroughout the autumn and winter, the liver stored fatduring summer and consumed it during autumn andwinter. Lipids have lower levels of 13C than proteins (DeNiro and Epstein 1978). During autumn and winter, theconsumption of 13C-depleted lipids should have led toan increase in liver d13C, unless this increasing d13C wasoffset by the integration of proteins with lower d13Cvalues coming from the zooplankton. In the liver then,the integration of the zooplanktonic d13C signal could bemasked by the consumption of the fat stored in the liverduring summer. Our findings for liver may therefore beconsistent with the hypothesis that there is a continuousturnover all year round. Should this assumption beverified by further experiments, then food consumedoutside the somatic growth period could be studied fromSIA performed on lipid-free liver samples.

Conclusion

The data support our initial predictions. Muscle doesindeed exhibit a slow and discontinuous turnover.Whitefish dorsal muscle provides a long-term integratedimage of the isotope composition of the food consumedfrom March to September, 7 months out of 12, duringwhich nutrients are allocated to growth. The Whitefishstomachs were not empty for the other 5 months, but asnutrients from the food consumed over the other5 months were allocated to basal metabolism and togonadic growth, their isotope composition was not re-flected in the muscle. Food consumed during nearly halfof the year cannot be detected by SIA of dorsal muscle.The Whitefish growth period is even shorter in boreallakes (Brusle and Quignard 2001) and consequently theperiod during which SIA of muscle provides reliableinformation about diet is probably even shorter. In en-ergy flow studies, stable isotopes composition of animalstissues were usually claimed to provide information onthe assimilated food, rather than on ingested preys, forwhich gut contents analysis has long been criticized.However, nutrients provided by preys consumed inautumn and winter were assimilated but allocated toenergy production and gonadic growth rather than tosomatic growth. So, if gut contents can be misleading forenergy flows studies, stable isotope analysis may as wellbe deceptive as muscle isotope composition reflects onlythe isotope composition of the food which nutrientscontributed in muscle tissue synthesis and not that of thewhole assimilated food.

Acknowledgements We would like to thank the participants at the‘‘Stable Isotope Ecology Conference’’ and two anonymous refereeswhose questions and comments definitely helped improve this

605

manuscript. We acknowledge Blake Matthews for his commentsand editing corrections of the manuscript. This work was sup-ported by the ‘ACI ecologie quantitative ’program from the FrenchMinistry of Research.

References

Adams TS, Sterner RW (2000) The effects of dietary nitrogencontent on the trophic level d15N enrichment. Limnol Oceanogr45:601–607

Brusle J, Quignard J-P (2001) Biologie des poissons d’eaux douceseuropeens. Technique et Documentation, Londres Paris

Caranhac F (1999) Modelisation de la dynamique de populationspiscicoles exploitees integrant la variabilite annuelle de crois-sance : applications aux coregones (Coregonus lavaretus) du lacd’Annecy. Thesis Claude Bernard University, Lyon, p 276

Conover WJ (1980) Practical nonparametric statistics, 2nd edn.Wiley, New York

De Niro M, Epstein S (1978) Influence of the diet on the distri-bution of carbon isotopes in animals. Geochim CosmochimActa 42:495–506

Doucett RR, Booth RK, Power G, McKinley RS (1999) Effects ofthe spawning migration on the nutritional status of anadro-mous Atlantic salmon (Salmo salar): insights from stable-iso-tope analysis. Can J Fish Aquat Sci 56:2172–2180

Dufour E (1999) Implications paleoenvironnementales et paleoal-imentaires des abondances isotopiques en Carbone et Azote despoissons teleosteens. Thesis Pierre and Marie Curie University,Paris, p 197

Frazer TK, Ross RM, Quetin LB, Montoya JP (1997) Turnover ofcarbon and nitrogen during growth of larval krill, Euphausiasuperba Dana: a stable isotope approach. J Exp Mar Biol Ecol212:259–275

Gannes LZ, O’Brien DM, Martinez Del Rio C (1997) Stableisotopes in animal ecology: assumptions, caveats, and a call formore laboratory experiments. Ecology 78:1271–1276

Gerdeaux D, Hamelet V (2001) Regime alimentaire des coregoneset des gardons du Leman, en milieu pelagique, campagne 2000.In: CIPEL (ed) Rapports sur les etudes et recherches entreprisesdans le bassin lemanique, Lausanne, pp 137–143

Herzka SZ, Holt GJ (2000) Changes in isotopic composition of reddrum (Sciaenops ocellatus) larvae in response to dietary shifts:potential applications to settlement studies. Can J Fish AquatSci 57:137–147

Hesslein RH, Hallard KA, Ramlal P (1993) Replacement of sulfur,carbon and nitrogen in tissue of growing broad Whitefish(Coregonus nasus) in response to a change in diet traced by delta34S, delta 13C and delta 15N. Can J Fish Aquat Sci 50:2071–2076

de la Higuera M, Akharbach H, Hidalgo MC, Peragon J, LupianezJA, Garcia-Gallego M (1999) Liver and white muscle proteinturnover rates in the European ell (Anguilla anguilla): effects ofdietary protein quality. Aquaculture 179:203–216

Hobson KA, Bairlein F (2003) Isotopic fractionation and turnoverin captive Garden Warblers (Sylvia borin): implications fordelineating dietary and migratory associations in wild passe-rines. Can J Zool 81:1630–1635

Hobson KA, Clark RW (1992) Assessing avian diets using stablecarbon and nitrogen isotopes I turnover of carbon-13. Condor94:181–188

Hobson KA, Sirois J, Gloutney ML (2000) Tracing nutrient allo-cation to reproduction with stable isotopes: a preliminaryinvestigation using colonial waterbirds of great Slave Lake.Auk 117:760–774

Hyslop EJ (1980) Stomach content analysis—a review of methodsand their application. J Fish Biol 17:411–429

Insightful (2001) S-Plus 6. Insightful Corporation, Seattle, Wash-ington

Johnson BM, Martinez PJ, Stockwell JD (2002) Tracking trophicinteractions in coldwater reservoirs using naturally occurringstable isotopes. Trans Am Fish Soc 131:1–13

Kelly JF (2000) Stable isotopes of carbon and nitrogen in the studyof avian and mammalian trophic ecology. Can J Zool 78:1–27

Kurle CM (2002) Stable isotope ratios of blood components fromcaptive northern fur seals (Callorhinus ursinus) and their diet:applications for studying the foraging ecology of wild otarids.Can J Zool 80:902–908

Kurle CM, Worthy GAJ (2002) Stable nitrogen and carbon isotoperatios in multiple tissues of the northern fur seal Callorhinusursinus: implications for dietary and migratory reconstructions.Mar Ecol-Prog Ser 236:289–300

Lorrain A, Paulet Y-M, Chauvaud L, Savoye N, Donval A, SaoutC (2002) Differential d13C and d15N signatures among scalloptissues: implications for ecology and physiology. J Exp MarBiol Ecol 275:47–61

Mac Avoy SE, Macko SA, Garman GC (2001) Isotopic turnover inaquatic predators: quantifying the exploitation of a migratoryprey. Can J Fish Aquat Sci 58:923–932

Mc Connaughey T, Mc Roy CP (1979) Food-web structure and thefractionation of carbon isotopes in the Bering Sea. Mar Biol53:257–262

Mc Millan DN, Houlihan DF (1989) Short-term responses ofprotein synthesis to re-feeding in rainbow trout. Aquaculture79:37–46

Mookerji N, Heller C, Meng HJ, Burgi HR, Muller R (1998) Dieland seasonal patterns of food intake and prey selection byCoregonus sp. in re-oligotrophicated Lake Lucerne, Switzer-land. J Fish Biol 52:443–457

O’Brien DM, Schrag DP, del Rio CM (2000) Allocation toreproduction in a hawkmoth: A quantitative analysis usingstable carbon isotopes. Ecology 81:2822–2831

Olive PJW, Pinnegar JK, Polunin NVC, Richards G, Welch R(2003) Isotope trophic-step fractionation: a dynamic equilib-rium model. J Anim Ecol 72:608–617

Peragon J, Barroso JB, Garcia-Salguero L, de la Higuera M, Lu-pianez JA (1994) Dietary protein effects on growth and frac-tional protein synthesis and degradation rates in liver and whitefish muscle of rainbow trout (Oncorhynus mykiss). Aquaculture124:35–46

Phillips DL, Koch PL (2002) Incorporating concentration depen-dence in stable isotope mixing models. Oecologia 130:114–125

Pinnegar JK, Polunin NVC (1999) Differential fractionation ofd13C and d15N among fish tissues: implications for the study oftrophic interactions. Func Ecol 13:225–231

Ponton D (1986) Croissance et alimentation de deux poissonsplanctonophages du lac Leman : le Coregone (Coregonus sp.) etle gardon (Rutilus rutilus). Thesis Claude Bernard University,Lyon, p 156

Roth JD, Hobson KA (2000) Stable carbon and nitrogen isotopicfractionation between diet and tissue of captive red fox: impli-cations for dietary reconstruction. Can J Zool 78:848–852

Tieszen LL, Boutton TW, Tesdahl KG, Slade NA (1983) Fraction-ation and turnover of stable carbon isotopes in animal tissues:Implications for delta-13 C analysis of diet. Oecologia 57:32–37

Zohary T, Erez J, Gophen M, Berman-Frank I, Stiller M (1994)Seasonality of stable carbon isotopes within the pelagic foodweb of Lake Kinneret. Limnol Oceanogr 39:1030–1043

606