Nutrient limitation of primary producers affects planktivorous fish condition

Arne Michael Malzahn1

Alfred Wegener Institute for Polar and Marine Research, Biologische Anstalt Helgoland, Ostkaje 1118, 27498Helgoland, Germany

Nicole AberleLeibniz Institute of Marine Sciences at the University of Kiel (IFM-GEOMAR), Dusternbrooker Weg 20, 24105 Kiel,Germany; Alfred Wegener Institute for Polar and Marine Research, Biologische Anstalt Helgoland, Ostkaje 1118, 27498Helgoland, Germany

Catriona ClemmesenIFM-GEOMAR, Dusternbrooker Weg 20, 24105 Kiel, Germany

Maarten BoersmaAlfred Wegener Institute for Polar and Marine Research, Biologische Anstalt Helgoland, Ostkaje 1118, 27498 Helgoland,Germany; GKSS Research Centre, Institute for Coastal Research, Max-Planck-Strasse 1, 21502 Geesthacht, Germany

Abstract

We investigated whether nutrient limitations of primary producers act upward through food webs only interms of density effects or if there is a second pathway for nutrient limitation signals channelled upward to highertrophic levels. We used tritrophic food chains to assess the effects of nutrient-limited phytoplankters (thecryptophyte Rhodomonas salina) on herbivorous zooplankters (the calanoid copepod Acartia tonsa) and finallyzooplanktivores (larval herring Clupea harengus) living on the herbivores. The primary producers’ food qualityhad a significant effect on fish condition. Our experimental phosphorus-limited food chain resulted in larval fishwith a significantly poorer condition than their counterparts reared under nitrogen-limited or nutrient-sufficientconditions. Our results show that mineral nutrient requirements of consumers have to be satisfied first before fattyacids can promote further growth. This challenges the match/mismatch hypothesis, which links larval fish survivalprobability solely to prey availability, and could imply that reduced nutrient releases into the environment mayaffect fish stocks even more severely than previously believed.

Herbivores usually feed on low-quality food because thestoichiometry (i.e., the ratio of elements) of the plant dietdiffers strongly from the demand of the consumer (Sternerand Elser 2002). Plants generally contain relatively highconcentrations of carbon and lower concentrations of othermajor elements such as nitrogen and phosphorus. Further-more, the nutrient stoichiometry in many plants roughlyreflects that of their surroundings. In contrast, mostanimals maintain a greater level of homeostasis (i.e., theykeep a constant ratio of carbon to nutrients), and, hence,they have to deal not only with food of a constantly lowquality but additionally with fluctuations in this quality.

One of the best-studied plant-herbivore interactions inaquatic environments is the interface between microalgaeand zooplankton, but it remains unclear whether the

changes in quality of phytoplankton as food for zooplank-ters are direct, i.e., nutrient limitations (Urabe et al. 1997;Boersma 2000; Plath and Boersma 2001), or whetheraccompanying changes, such as changes in the fatty acidspectrum (Muller-Navarra 1995), are responsible for theobserved quality effects on consumers. Herbivores haveseveral mechanisms to deal with excess carbon, such asrespiring carbohydrates (Trier and Mattson 2003), thermo-genesis (Zanotto et al. 1997; Raubenheimer and Simpson2003), or adjusting digestion (Lee and Houston 1993; Sabatet al. 1999; Hilton et al. 2000). Zooplankters might reactwith an increase in appendage beat rate (Plath andBoersma 2001) or simply refuse to eat low-quality diets(Irigoien et al. 2005). Whatever the strategy might be,handling excess carbon creates costs, and these are usuallycompensated by decreased growth or reproduction(Boersma 2000; Boersma and Kreutzer 2002).

Herbivore homeostasis, combined with the regulatorymechanisms just described, has led to the commonlyaccepted assumption that quality effects in the herbivore-plant interface are not transported up the food chain, andthe only important signal of variation in food quality tohigher trophic levels is of a quantitative nature, asherbivores tend to grow and reproduce at lower ratesunder poor feeding conditions. Here, we challenge this

1 Corresponding author (arne.malzahn@awi.de).

AcknowledgmentsWe thank Christoph Petereit, Daniela Harrer, Nicole Kube,

and Michael Donner for the helping hands during the experimentsand Helgi Mempel for his helpful assistance in the laboratory.This study was part of the Deutsche Forschungsgemeinschaft-project FRISCA (AB 289/1-1) and the Helgoland FoodwebProject. We are grateful for the funding. This study complies withthe current German law on animal studies.

Limnol. Oceanogr., 52(5), 2007, 2062–2071

E 2007, by the American Society of Limnology and Oceanography, Inc.

2062

assumption based on the following evidence: homeostasisof zooplankters is far from perfect, and within-speciesvariation in carbon-to-nutrient ratios has been reported(Plath and Boersma 2001; Boersma and Kreutzer 2002).Consequently, their quality as food for their predatorsshould vary. Furthermore, large interspecific variationshave been reported for marine (Gismervik 1997) andfreshwater zooplankters (Hessen 1992). Hence, given theargument of Brett (1993), which says that stoichiometricneeds of secondary consumers and the stoichiometry ofprey are normally finely tuned, this means that there isample opportunity for quality differences between preyorganisms of different nutritional status to be transferred tothe next trophic level. Consequently, we do not expect thatnutrient limitation of primary production would act onlyon higher trophic levels in terms of density effects. Wetherefore hypothesize that nutrient limitations can also actupward through the food web in the form of food-qualitysignals, and, hence, bottom-up control mechanisms are notonly of a quantitative nature.

In order to test this hypothesis, we conducted twoexperiments to track the effects of nutrient limitationthrough a tritrophic food chain, from primary producers toprimary and secondary consumers.

Material and methods

In order to investigate upward-travelling nutrient(nitrogen [N] and phosphorus [P], respectively in thefollowing) effects through several trophic levels, algae,copepods, and larval herring were reared under threenutrient regimes (Fig. 1). To investigate the effect of foodquality rather than food quantity, we adjusted the food fedto the two consumer levels to equal amounts for the threetreatments.

Phytoplankton—A stock culture of Rhodomonas salinawas cultivated in enriched seawater, following Guillard andRyther (1962). For our experiments, we cultivated algae at18uC under a 16 : 8 h light : dark (LD) light regime inenriched natural seawater (salinity ,15), as well as undernutrient (here, P and N) limitation. Prior to the experiment,all of the water used for algae cultures during the

experiments was filtered using a 0.2-mm sterile filter inone effort and stored cool and dark until use. The firsttreatment consisted of f/2-enriched seawater (f/2 contains8.83 1024 mol L21 NaNO3 and 3.63 1025 mol L21

NaH2PO4), as described by Guillard and Ryther (1962).The algae of the two limitation treatments were alsoenriched after Guillard and Ryther (1962), but without theaddition of the limiting nutrient (no P or N addition; 2Pand 2N in the following), so they could therefore onlyutilize the natural P or N sources present in the seawater atthe moment of filtration. This kind of enrichment ensureda limitation of the desired nutrient by enriching the mediawith the other nutrients (i.e., N or P, vitamins, and metals)in excess. Several tests on algal growth rates wereconducted prior to the experiments to detect the ‘‘carryingcapacity’’ of the three different media and to define theduration until the algae were properly limited by theelement of choice in the different treatments. Concentra-tions of algae were determined using fluorometric measure-ments (Turner, 10-AU-005-CE) at an excitation wavelengthof 436 nm and an emission wavelength of 680 nm. Prior tothe experiments, individual cell number–fluorescence re-lationships were set up using flow cytometer counts atseven concentrations ranging from 0.05 to 1.7 3106 cells mL21 for all three differently grown algae.

To ensure constant food quality, new cultures of each ofthe treatments were inoculated every day with roughly 0.23 106 cells mL21 for the 2N treatment and 0.33 106 forthe 2P and f/2 treatments. Algae were harvested atdensities of ,0.5, 1.0, and 1.3 3 106 cells mL21 (2N,2P, and f/2, respectively) after the predefined growth phaseof six days for 2N and seven days for f/2 and 2P. Thesewere the maximum densities possible with the natural N orP sources contained in the seawater. To prevent food-quantity effects, the copepods were fed with the sameamount of algal cells at each treatment.

Zooplankton—Eggs of the calanoid copepod Acartiatonsa were produced in 200-liter cylindrical tanks, wherethe animals were cultivated at 18uC at a 12 : 12 LD cycle.Copepods were fed on a mixture of the algae Rhodomonassalina, Dunaliella sp., and the flagellate Oxhyrris sp. Eggswere siphoned from the bottom of the tanks daily and

Fig. 1. Simplified illustration and main details of the experimental procedure. Numbers in brackets indicate the total number ofcultures ran during experiments one and two.

Nutrient limitation affects fish condition 2063

stored in seawater at 4uC for later use. No eggs older thanthree months were used during the experiments.

The stored eggs were incubated in fresh seawater. Thehatch rate of the eggs was around 20%. The copepods usedfor the experiments were reared from the egg to the fifthnaupliar stage in 5-liter plastic bags at 18uC at densities of1,500 individuals L21. The cultured algae were fed to thecopepods at ,1.0 mg carbon L21 d21 for five consecutivedays. In order to guarantee that the algal deficiencies werenot changed by uptake of nutrients during the incubationswith the copepods, the copepods were reared in artificial,N- and P-free seawater that was adjusted to a salinity of 18(salt: hw-Marinemix, www.hw-wiegandt.de). Copepodswere first fed 48 h after addition of the eggs to the rearingbags, assuming two days for the development from the eggto the first feeding second naupliar stage. For each day ofthe feeding experiment, three new copepod bags (one foreach treatment) were started to ensure a constant foodquality for the secondary consumers, the larval herringClupea harengus. Copepods were harvested after seven daysof cultivation when ,80% of all the animals were in theirfifth naupliar stage and ,20% were in the fourth. Thesecopepods were then fed to the larval herring at densities of1 individual mL21 for all treatments.

Fish—Herring larvae were obtained by means ofartificial fertilization. Adult ripe Baltic herring Clupeaharengus were purchased from a local fisherman. The fishwere transported immediately to the laboratory and kepton ice the whole time. Female fish were strip-spawned onglass plates. The eggs were applied in single rows to ensurea good oxygen supply during the incubation phase, and theglass plates were placed in a plastic box. Milt was strippedonto the eggs and activated by the addition of seawater.Fertilization was allowed to take place for five minutes;afterward, the eggs were washed and transferred to theincubation containers. Eggs were incubated in a flow-through system, using 4-mm prefiltered natural seawater at13uC. The first hatch took place on the night of day 10;peak hatch took place the following night. Only larvaefrom the hatch peak were used during the experiments.Larvae were transferred to cylindrical 200-liter stock tanks.The stock tanks were operated as a flow-through systemand gently supplied with prefiltered water. Experiment onewas started using 4-d-posthatch larvae, which had nofeeding experience. The larvae in the stock tanks were fedfrom day 4 on the rotifer Brachionus plicatilis reared on thealgae Nannochloropsis sp. B. plicatilis were taken fromroutine cultures of the facility. Experiment two was startedusing 9-d-posthatch larvae.

Twenty larvae each were randomly transferred to 1-literglass beakers filled with GF/F filtered seawater. The fourtreatments (2P, 2N, f/2, and starving) were replicated 10times (with 20 larvae each) in experiment one and threetimes in experiment two. Both experiments were conductedat 15uC, and larval herring were reared for six days. Withinthe six days, larvae were fed the quality-manipulated dietsfive times. Feeding took place at noon. The food quantityfed to the herring was kept constant between the treatmentsand over the experimental period (1 copepod mL21 d21).

This prey concentration is higher than the densities that areusually reported for food-saturated growth in larval herring(Clemmesen 1994). This was corroborated by the fact thatin all of the experimental containers, copepods were stillpresent after one day of feeding. More than 80% of thewater was replaced daily before feeding to ensure that thevast majority of uneaten prey organisms stayed in theexperimental container for a maximum of 24 h. This wasessential in order to avoid alterations of the bodycomposition of the copepods due to starvation and henceto assure a constant food quality over the experimentalperiod. The experiments were terminated on the morningafter the fifth day of herring feeding.

Analytical procedures—The nutritional condition oflarval herring was assessed by means of the analysis ofthe ratio between ribonucleic acid (RNA) and deoxyribo-nucleic acid (DNA) content in the organisms, a methodcommonly used in larval fish ecology and fisheries research.The analysis of RNA and DNA concentrations of in-dividual herring larvae was performed using a modificationof the method by Clemmesen (1993). Larval herring werethawed, and standard length was measured using a stereo-microscope. Larvae were freeze-dried to constant weight(16 h, using a Christ Alpha 1–4 freeze-dryer at 251uC) andwere weighed to the nearest 0.0001 mg (Sartorius micro-balance SC2). The freeze-dried larvae were rehydrated inTris-SDS buffer (Tris 0.05 mol L21, NaCl 0.01 mol L21,ethylenediaminetetracetic acid (EDTA) 0.01 mol L21, so-dium dodecyl sulfate (SDS) 0.01%) for 15 min. Cells weredisrupted by shaking in a cell-mill with different-sized glassbeads (diameter 2.0 mm and 0.17–0.34 mm) for 15 min.The homogenate was then centrifuged at 6,000 rpm at 0uCfor 8 min, and the supernatant was used for the analysis.The amount of nucleic acids was measured fluorometricallyin a microtiter fluorescence reader (Labsystems, FluorescanAscent) using the fluorophor ethidiumbromide. Totalnucleic acids were measured first, and, subsequently,RNAse was applied to the sample in order to digest theRNA. After the enzyme treatment (30 min at 37uC), theremaining DNA was measured. The RNA fluorescence wascalculated by subtracting the DNA fluorescence from thetotal nucleic acid fluorescence. RNA calibrations were setup every measurement day. The DNA concentrations werecalculated using the relationship between RNA and DNAfluorescence described by Le Pecq and Paoletti (1966).

For the analysis of carbon and nitrogen contents of thealgae, an estimated amount of 150 mg algal carbon wasfiltered on precombusted Whatman GF/F filters. For theanalysis of copepod carbon and nitrogen, 500 individualswere counted into tin capsules. The elemental analyses weredone using a Fison 1500N CHN analyzer. Phosphorus wasanalyzed as orthophosphate after acidic oxidative hydro-lysis with 5% H2SO4 (Grasshoff et al. 1999). Unfortunately,algae P data are only available for experiment 1. C : N,C : P, and N : P ratios are given as the molar ratio.

The fatty acids of algae and copepods were measured asfatty acid methyl esters (FAMEs). Lipids were extractedfrom the samples by dichloromethane : methanol (2 : 1vol : vol) in an ultrasound bath for 30 min. Water-soluble

2064 Malzahn et al.

fractions were removed after centrifugation by washingwith 0.88% KCl buffer. The water phase was removed, andthe organic remainder was evaporated using nitrogen gas.The esterification was done using methanolic sulfuric acidat 70uC for 30 min. The FAMEs were washed from themethanolic sulfuric acid using n-Hexane. Excess n-Hexanewas evaporated using nitrogen gas. All chemicals used weresuprasolv or gas chromatography (GC) grade. FAMEswere analyzed by gas chromatography using a Varian CP8400 gas chromatograph equipped with a DB-225 column(J&W Scientific, 30-m length, 0.25-mm inner diameter [ID],0.25-mm film). The injector temperature was set to 250uC.The column oven was set to 60uC, which was held for 1 minafter injection. The oven was heated to 150uC at15uC min21, then to 170uC at 3uC min21, and finally to220uC at 1uC min21, which was held for 21 min. Thecarrier gas was helium at a constant pressure of 82.737Pa.The flame ionization detector was set to 300uC. Injection ofthe 1-mL aliquots of the samples was done in a split-lessmode. FAMEs were quantified using calibrations set up foreach fatty acid separately and a known amount of C 23:0was added at the first step of the preparation as an internalstandard.

Statistics—Fish standard length, dry weight, andRNA : DNA ratios were statistically analyzed by meansof a two-factorial analysis of variance (ANOVA) withtreatment and experiment as independent variables andRNA : DNA ratio as the dependent variable. Phytoplank-ton and zooplankton fatty acids, and C : P, C : N, and N : Pratios were analyzed by one-factorial ANOVA, usingtreatment as the independent factor and the various fattyacid proxies or the nutrient ratios as the dependentvariables. Phytoplankton C : N was analyzed in a two-factorial ANOVA, using experiment and treatment as theindependent factors and the C : N ratio as the dependentvariable. Due to the lack of phytoplankton P data forexperiment two, phytoplankton C : P and N : P ratios wereanalyzed by a one-factorial ANOVA, using treatment asthe independent factor and C : P or N : P as the dependentvariable. Stoichiometric measures of zooplankton wereanalyzed in a two-factorial ANOVA, using experiment andtreatment as independent factors and the stoichiometricmeasures as dependent variables. Phytoplankton andzooplankton fatty acid measures were analyzed by meansof one-factorial ANOVA. Fish response variables to thedifferent treatments (RNA : DNA, final standard length,and final dry weight) were analyzed in a two-factorialdesign with experiment and treatment as independentfactors and the fishes’ RNA : DNA ratio, standard length,and dry weight as the dependent variables. Duncan’s testswere used as post-hoc tests in all cases.

Results

Phytoplankton—Algal molar C : N ratios differed signif-icantly between treatments in both experiments. We wereable to create similar values in both experimental runs, sothat the factor ‘‘Experiment’’ had no significant influencein the two-way ANOVA. Algae grown under N limitation

showed the highest C : N ratios, while f/2 and P-limitedalgae showed significantly lower C : N ratios (,10, ,7.5,and ,8.0 respectively; p , 0.05, Fig. 2A). Due to the lackof P data for experiment two, we were not able toinvestigate the effect of the experimental run. However,the C : P ratio differed significantly among treatments (f/2,230, 2N ,180, and 2P ,580; Fig. 2B). The samepattern was found in the N : P ratio, where the P-limitedtreatment showed the highest value, and the N-limitedtreatment showed the lowest values (f/2 ,25, 2N ,15, and2P ,78; Fig. 2C).

The different nutrient-limitation treatments causeddifferent fatty acid spectra in the algae (Table 1; Fig. 3A).The 2P treatment showed not only the highest concentra-tion of fatty acids (mg FA mg C21) but also the highestamount and proportion of unsaturated fatty acids, omega-3, and omega-6 fatty acid concentrations, as well asthe lowest percentage of saturated fatty acids in theexperiment (all p , 0.05 to f/2 and no differences comparedto 2N).

Zooplankton—Copepod C : N ratios showed significantdifferences among treatments. In experiment one, copepodsthat were fed on N- as well as P-limited algae showedsignificantly higher C : N ratios in comparison with thosecopepods fed on nonlimited algae. In experiment two, nodifferences where found due to the unexpectedly low C : Nratios in the N-limited treatment (Fig. 2A). Significantlydifferent C : P ratios in zooplankton were created by thedifferent phytoplankton diet (f/2 ,180, 2N ,186, and 2P,280). The 2P treatment created significantly higher C : Pand N : P ratios in copepods than the f/2 and the 2Ntreatments did (both p , 0.01). The latter two did not differsignificantly (Fig. 2B). Exactly the same pattern in fattyacid measures was found for zooplankton reared ondifferent algae treatments as in phytoplankton (Table 1;Fig. 3B). The 2P treatment showed not only the highestconcentration of fatty acids (mg FA mg C21) but also thehighest amount and proportion of unsaturated fatty acids,omega-3, and omega-6 fatty acid concentrations, as well asthe lowest percentage of saturated fatty acids in theexperiments (all p , 0.05 to f/2 and no differencescompared to 2N). This implies that for both the copepodsas well as the fish, the best food quality in terms of fattyacids was produced under P limitation, followed by the N-limited conditions. The f/2-cultivated phytoplanktonseemed to be of the poorest food quality when only fattyacid spectra were considered. On the other hand, innutrient stoichiometric terms, f/2 produced the highestfood quality (lowest C : N and C : P ratios), and thelimitations created food of much lower (i.e., higher C : Pand C : N ratios) quality. This opposing effect enabled us todistinguish differential effects of nutrient- and biochemical-originated food quality.

Fish—The age of the larvae had a significant effect onthe RNA : DNA ratio, where older larva showed generallya better condition (Table 2). For young larvae (experimentone), no significant differences among the treatments weredetected. All fed treatments showed significantly different

Nutrient limitation affects fish condition 2065

values than the starved larvae of experiment one. Olderlarvae of experiment two showed the same pattern as thoseof experiment one, but these patterns were more pro-nounced, and differences were significant, despite thelower level of replication. All fed treatments differedsignificantly in their RNA : DNA ratios from each other(Duncan’s test p , 0.05). The N-limited treatment showedthe highest RNA : DNA ratios, followed by the f/2treatment. The 2P treatment showed the poorest nutri-tional condition of the fed groups. All three fed treatmentsdiffered significantly from the corresponding food-de-prived groups (Fig. 4A). Growth virtually provided thesame pattern at either the final standard length (Fig. 4B)or the final dry weight (Fig. 4C) of the larval herring.Again, the N-limited food chain produced the longest andheaviest larvae, followed by the f/2 and the 2P treatment

Fig. 2. (A) Molar C : N, (B) C : P, and (C) N : P ratios ofphytoplankton cultured under nutrient-sufficient (f/2), N-limited(2N), and P-limited (2P) conditions and the molar (A) C : N, (B)C : P, and (C) N : P ratios of the zooplankton reared on thedifferent phytoplankton cultures. Error bars: 1 standard de-viation. Asterisk marks significant differences (p , 0.05) from theother treatments of the given species and experiment.

Fig. 3. Summary of several fatty acid proxies of (A)phytoplankton cultured under nutrient-sufficient (f/2), N-limited(2N), and P-limited (2P) conditions and (B) zooplankton rearedon the different phytoplankton cultures. Error bars: 1 standarddeviation. Asterisk marks significant differences (p , 0.05) fromthe other treatments, and ‘‘1’’ marks significant differencesbetween 2P and f/2 only.

2066 Malzahn et al.

Table

1.

Fa

tty

aci

d(F

A)

com

po

siti

on

of

ph

yto

pla

nk

ton

gro

wn

un

der

nu

trie

nt-

suff

icie

nt

(f/2

),N

-lim

ited

(2N

),a

nd

P-l

imit

ed(2

P)

con

dit

ion

sa

nd

the

fatt

ya

cid

com

po

siti

on

of

zoo

pla

nk

ton

rea

red

on

dif

fere

nt

ph

yto

pla

nk

ton

trea

tmen

ts.

Va

lues

are

mea

ns

of

19

sam

ple

sp

era

lga

ltr

eatm

ent

an

d1

0sa

mp

les

per

cop

epo

dtr

eatm

ent.

Va

lues

are

giv

enin

mg

FA

mg

C2

1(s

tan

da

rdd

evia

tio

ns

inp

are

nth

eses

)a

nd

as

ap

erce

nta

ge

of

the

tota

lfa

tty

aci

dco

nte

nt.

AR

A5

ara

chid

on

ica

cid

,D

PA

5d

oco

sap

enta

eno

ica

cid

.

Rh

od

om

on

as

sali

na

Aca

rtia

ton

sa

f/2

2N

2P

f/2

2N

2P

(mg

FA

mg

C2

1)

(%)

(mg

FA

mg

C2

1)

(%)

(mg

FA

mg

C2

1)

(%)

(mg

FA

mg

C2

1)

(%)

(mg

FA

mg

C2

1)

(%)

(mg

FA

mg

C2

1)

(%)

C1

6:0

19

.06

(4.4

2)

34

.58

68

.82

(42

.25

)4

3.7

53

6.7

2(1

3.7

1)

28

.93

47

.26

(13.8

4)

60

.83

67

.98

(32

.52

)5

6.7

17

0.4

3(3

2.0

2)

49

.47

C1

6:1

n7

0.6

6(0

.32

)1

.18

3.3

8(6

.71

)2

.21

7.1

2(8

.76

)3

.40

.59

(0.2

7)

0.7

61.0

8(0

.5)

0.9

12

.83

(1.6

8)

1.8

8C

16

:2n

40

00

00

00

00

00

0.0

0C

17

:00

.47

(0.2

3)

0.8

40

.94

(0.5

2)

0.6

20

.58

(0.6

8)

0.4

72

.41

(1.1

1)

3.0

61

.69

(1.0

9)

1.3

42

.64

(1.4

7)

1.8

1C

17

:1n

70

.1(0

.08

)0

.18

0.2

6(0

.31

)0

.18

0.1

4(0

.17

)0

.13

0.2

6(0

.27

)0

.30

0.4

2(0

.61

)0

.41

0.6

9(0

.66

)0

.51

C1

8:0

2.9

2(1

.14

)5

.31

8.5

1(7

.01

)5

.53

.25

(0.8

)3

.49

10

.47

(3.7

5)

13

.23

12.1

5(8

.5)

9.5

21

3.3

(9.5

3)

8.8

1C

18

:1n

94

.46

(2.7

7)

7.2

31

3.1

8(7

.91

)8

.96

12

.91

(7.7

7)

8.2

46

.41

(2.6

)8

.53

11

.43

(4.7

1)

10

.15

12

.39

(5.2

2)

9.4

6C

18

:1n

70

00

00

00

00

00

.58

(1.8

2)

0.5

9C

18

:2n

64

.03

(3.1

1)

6.0

27

.93

(4.3

8)

5.1

11

9.9

6(1

8.6

3)

9.6

50

.81

(1.0

6)

0.8

81

.78

(3.3

8)

1.1

46

.56

(4.3

5)

4.2

1C

18

:2n

90

.4(0

.43

)0

.76

0.6

2(0

.64

)0

.39

0.3

9(0

.49

)0

.39

0.2

(0.1

8)

0.3

30

.25

(0.1

8)

0.2

10

.78

(0.4

7)

0.5

2C

18

:3n

31

1.1

5(7

.98

)1

6.5

31

8.1

8(1

2.5

6)

12

.43

40

.78

(31

.37

)2

0.5

22

.43

(2.6

1)

2.7

05

.1(4

.56

)3

.84

10

.45

(7.4

5)

6.6

7C

18

:3n

60

.85

(0.8

8)

1.2

31

.16

(0.6

)0

.76

2.1

2(1

.9)

1.0

90

.12

(0.1

9)

0.1

30

.27

(0.2

9)

0.2

00

.57

(0.4

)0

.37

C1

8:3

n4

00

00

00

00

00

00

C1

8:4

n3

00

00

00

00

00

00

C2

0:0

0.2

1(0

.09

)0

.38

0.2

8(0

.15

)0

.24

0.2

(0.1

1)

0.2

0.5

7(0

.15

)0

.75

0.7

5(0

.6)

0.5

70

.7(0

.36

)0

.49

C2

0:1

n9

0.0

7(0

.09

)0

.14

0.2

3(0

.21

)0

.14

0.1

6(0

.18

)0

.09

0.0

3(0

.06

)0

.03

0.2

1(0

.22

)0

.15

0.2

1(0

.15

)0

.14

C2

0:2

n6

0.0

5(0

.05

)0

.08

0.0

8(0

.08

)0

.06

0.0

5(0

.06

)0

.03

0.0

9(0

.12

)0

.09

0.1

1(0

.09

)0

.09

0.3

2(0

.19

)0

.21

C2

0:3

n3

0.0

3(0

.07

)0

.03

0.1

1(0

.09

)0

.08

0.1

4(0

.15

)0

.07

0.0

7(0

.08

)0

.08

0.0

3(0

.03

)0

.03

0.0

3(0

.05

)0

.02

C2

0:3

n6

0.1

(0.0

9)

0.1

80

.14

(0.1

6)

0.0

90

.23

(0.4

9)

0.6

0.0

2(0

.03

)0

.03

0.0

4(0

.03

)0

.04

0.0

3(0

.06

)0

.02

C2

0:4

n6

(AR

A)

1.1

6(0

.71

)2

.01

1.2

(0.9

2)

0.8

20

.96

(0.8

1)

0.8

20

.6(0

.37

)0.8

00

.78

(0.6

)0

.73

1.0

3(0

.56

)0

.77

C2

0:4

n3

00

00

00

00

00

00

C2

0:5

n3

(EP

A)

8.8

3(7

.31

)1

2.7

81

0.7

(5.7

8)

7.5

41

6.2

(10

.65

)9

.06

2.2

3(2

.79

)2

.44

3.3

9(3

.59

)2

.65

5.8

1(3

.35

)3

.90

C2

1:0

00

00

00

0.0

4(0

.06

)0

.05

0.0

7(0

.07

)0

.06

0.0

2(0

.03

)0

.02

C2

2:0

0.0

8(0

.03

)0

.13

0.1

1(0

.07

)0

.08

0.0

9(0

.06

)0

.09

0.3

1(0

.13

)0

.40

0.4

6(0

.66

)0

.29

0.4

(0.1

9)

0.2

8C

22

:1n

90

.06

(0.0

6)

0.1

10

.01

(0.0

3)

0.0

10

.01

(0.0

2)

0.0

10

.17

(0.1

4)

0.2

00

.07

(0.0

7)

0.0

50

.23

(0.1

1)

0.1

6C

22

:2n

60

00

00

00

00

.03

(0.0

4)

0.0

30

.01

(0.0

3)

0.0

1C

22

:5n

3(D

PA

)0

00

00

00

00

00

0C

22

:6n

3(D

HA

)6

.62

(4.9

2)

9.7

11

6.5

2(1

7.3

)1

0.7

82

3.8

2(3

9.0

8)

12

.31

3.2

5(3

.34

)3

.71

13

.06

(12

.3)

10

.15

12

.52

(8.2

5)

9.1

3C

24

:00

.27

(0.1

4)

0.4

70

.29

(0.1

3)

0.2

30

.28

(0.1

8)

0.3

90

.42

(0.1

8)

0.5

80

.56

(0.5

7)

0.4

20

.46

(0.2

5)

0.3

2C

24

:1n

90

.05

(0.1

2)

0.1

10

.03

(0.0

5)

0.0

20

(0.0

1)

0.0

10

.09

(0.0

8)

0.1

00

.38

(0.2

1)

0.3

20

.33

(0.1

6)

0.2

3

To

tal

FA

57

.54

(22

.65

)1

44

.75

(55

.93

)1

46

.16

(90

.22

)7

8.0

5(2

5.1

8)

12

0.3

1(5

3.4

4)

13

6.7

4(5

8.7

5)

Su

msa

tura

ted

23

(5.2

)4

1.7

17

8.9

6(4

9.4

3)

50

.41

41

.11

(14

.16

)3

3.5

86

1.4

7(1

8.1

1)

78

.90

83

.66

(43

.19

)6

8.9

18

7.9

4(4

2.7

)6

1.2

0S

um

un

satu

rate

d3

4.5

4(2

2.0

5)

58

.29

65

.79

(32

.15

)4

9.5

91

05

.05

(78

.6)

66

.41

6.5

8(9

.57

)2

1.1

03

6.6

5(2

1.5

4)

31

.09

48

.8(1

7.8

2)

38

.80

Sa

t/u

nsa

t0

.96

(0.6

5)

0.8

51

.68

(1.5

4)

1.4

10

.98

(1.6

1)

0.9

14

.39

(1.7

8)

4.2

72

.62

(1.1

3)

2.5

81

.82

(0.4

2)

1.6

3n

32

6.6

3(1

9.8

6)

39

.06

45

.51

(26

.65

)3

0.8

48

0.9

6(6

7.4

9)

41

.97

7.9

8(8

.78

)8

.93

21

.58

(19

.71

)1

6.6

62

8.8

1(1

3.9

1)

19

.72

n6

6.1

9(4

.4)

9.5

21

0.5

2(5

.28

)6

.83

23

.32

(20

.85

)1

2.1

91

.64

(1.4

8)

1.9

33

.01

(3.4

7)

2.2

28

.52

(5.1

5)

5.6

0n

3/n

64

.26

(1.7

6)

4.2

65

.28

(5.2

5)

5.2

86

.47

(10

.36

)6

.47

4.5

5(2

.24

)4

.55

10

.1(7

.31

)1

0.1

03

.83

(1.5

5)

3.8

3

Nutrient limitation affects fish condition 2067

(Table 2). All differences in weight and length weresignificant.

Discussion

We hypothesized that nutrient limitation acts upwardthrough the food web in the form of food-quality signals,and, hence, bottom-up control mechanisms are not only ofa quantitative nature. Our results support this hypothesis,as nutrient-limitation signals from phytoplankters weretransferred via zooplankters to secondary consumers,despite the fact that larval fish were fed ad libitum, and,thus, density-dependent prey effects were excluded. Thecopepods in our experiments clearly dampened thevariation of carbon : nutrient ratios of their food, but onlyto a certain degree. Hence, zooplankton homeostasisshould be regarded as relaxed in nature rather than strict(Plath and Boersma 2001; Boersma and Kreutzer 2002).This relaxed homeostasis led to limited growth ofsecondary consumers, in our case larval herring.

Phytoplankton—In our study, the cryptophyte Rhodo-monas salina produced significantly more total fatty acidsand unsaturated fatty acids, such as ecosapentaenoic acid(20:5n3, EPA) and docosahexaenoic acid (22:6 n3, DHA),under both N and P limitation, and thus in our case, thelimited algae showed enhanced food quality compared withthe algae grown under excess nutrient conditions withrespect to the fatty acid composition. Nutrient limitation isknown to be reflected in alterations of algal biochemistry.Typically, under N limitation, the protein content of algaeis reduced, while under P limitation, phospholipidsconcentration is lower (Klausmeier et al. 2004). The totallipid content seems to be positively affected by nutrient-limiting growth conditions, although there are largeinterspecific variations (Sterner and Hessen 1994). Theenhanced lipid levels of the nutrient-limited algae alteredthe carbon : nutrient ratios of the limited phytoplankton tohigher carbon : nutrient ratios, and as a result, froma nutrient point of view, these phytoplankters must beconsidered as food of lower quality for consumers (Sternerand Elser 2002).

Table 2. Summary of all analyses of variance (ANOVA). Phytoplankton C : N was analyzed in a two-factorial ANOVA, usingexperiment and treatment as independent factors and the C : N ratio as dependent variable. Phytoplankton C : P and N : P were analyzedby a one-factorial ANOVA, using treatment as the independent factor and C : P or N : P as the dependent variable. Stoichiometricmeasures of zooplankton were analyzed in a two-factorial ANOVA, using experiment and treatment as independent factors and thestoichiometric measures as dependent variables. Fish data were analyzed in a two-factorial design with experiment and treatment asindependent factors and the fishes RNA : DNA ratio, standard length, and dry weight as dependent variables.

Trophic level Factor MS df F p

Phytoplankton C : N Experiment 10.175 1 3.726 n.s.Treatment 58.681 2 21.491 ,0.01Exp. x Treat. 2.094 2 0.767 n.s.Error 2.731 73

C : P Treatment 591,618 2 41.0142 ,0.01Error 14,425 27

N : P Treatment 9,018.76 2 46.0649 ,0.01Error 195.78 27

Zooplankton C : N Experiment 1.4003 1 11.667 ,0.01Treatment 1.2187 2 10.154 ,0.01Exp. x Treat. 0.5320 2 4.433 ,0.05Error 0.1200 20

C : P Experiment 24,045 1 6.3839 ,0.05Treatment 39,732 2 10.5489 ,0.01Exp. x Treat. 841 2 0.2233 n.s.Error 3,766 20

N : P Experiment 414.23 1 3.3791 n.s.Treatment 800.12 2 6.5269 ,0.01Exp. x Treat. 45.21 2 0.3688 n.s.Error 122.59 20

Fish RNA : DNA Experiment 0.57254 1 16.8299 ,0.01Treatment 0.54261 3 15.9501 ,0.01Exp. x Treat. 0.20525 3 6.0332 ,0.01Error 0.03402 44

Standard length Experiment 15.945 1 144.73 ,0.01Treatment 1.516 3 13.76 ,0.01Exp. x Treat. 0.708 3 6.42 ,0.01Error 0.110 44

Dry weight Experiment 12,961.0 1 230.427 ,0.01Treatment 3,190.3 3 56.718 ,0.01Exp. x Treat. 1,559.7 3 27.729 ,0.05Error 56.2 44

2068 Malzahn et al.

Zooplankton—Long-chained, unsaturated fatty acidslike EPA and DHA cannot be synthesized de novo bymost copepods and, hence, have to be taken up in the food.Consequently, the copepod fatty acid compositions re-flected those of their diets well, a pattern which is in goodcorrespondence with previous studies (Sargent and Falk-Petersen 1988). The copepods reared on P-limited algaerepresented the best food offered to larval herring in termsof fatty acids. We found significantly higher C : P ratios incopepods grown on P-limited algae but unaffected C : Nratios of copepods grown on N-limited algae. Similarfindings of upward-travelling effects of nutrient-limitedphytoplankton have been reported by Sterner (1993),DeMott et al. (1998), and Plath and Boersma (2001), but,to our knowledge, no one has made the next step andinvestigated the effects of a relaxed homeostasis ofzooplankters on growth of those who feed on thezooplankters.

Fish—The RNA : DNA ratio is a valid predictor fornutritional condition of larval fish. This has been validatedin many laboratory and field studies (Clemmesen 1994;Malzahn et al. 2003, in press). The RNA : DNA ratio oflarval herring in our first experiment showed no significantdifferences between the fed groups, but significant differ-ences were observed between all fed groups and thestarvation control. These observations could be attributedto the simultaneous exploitation of internal (yolk) andexternal energy sources (copepods). Suneetha et al. (1999)also reported less serious starvation effects in smallerherring larvae (Clupea harengus) than in bigger ones andattributed this finding to benefits derived from leftoveryolk. In older herring, which were exclusively dependent onexternal energy sources, the N-limited food chain producedthe best conditioned larvae. This is contrary to thepredictions that could be drawn from the fatty acid profilesalone, where the P-limited food chain displayed the bestfood quality. High levels of fatty acids are known to favorgrowth in larval fish. St. John et al. (2001) were able todemonstrate significant effects of dietary EPA as well asDHA levels on larval cod growth, and Izquierdo et al.(2000) stressed not only the major role of high fatty acidconcentrations but also the importance of ratios betweencertain fatty acids for the growth of larval fish. Neverthe-less, the herring larvae reared under the highest lipid supplyshowed the lowest nutritional condition. Under theassumption that the same processes are acting not only atthe primary producer–herbivore interface but also atinterfaces between higher trophic levels, this could well beattributed to the unfavorable high C : P ratio of thezooplankters of the P-limited food chain. This inevitablyleads to the conclusion that the herring’s growth was P-

Fig. 4. Response variables of larval fish reared on zooplank-ton that was fed on phytoplankton cultured under nutrient-sufficient (f/2), N-limited (2N), and P-limited (2P) conditions.(A) RNA : DNA ratio, (B) larval standard length (mm), and (C)larval dry weight (mg). Error bars: 1 standard deviation.Significant differences (p , 0.05) between treatments are indicated

r

with different letters for experiment two. In both experiments,RNA : DNA ratios of all fed treatments were significant different(p , 0.05) from the corresponding starving group. No differenceswithin experiment one were present in standard length anddry weight.

Nutrient limitation affects fish condition 2069

limited. Considering the best herring growth performancereared on the N-limited food chain, the C : P ratio and thelipid measures suggest that biochemical components likefatty acids can further promote growth, provided thatnutrient requirements are satisfied.

Further support for the notion that fish growth can be P-limited is derived from aquaculture studies (e.g., Vielma etal. 2002), which have shown that it is quite possible forgrowth in fish to be limited by a shortage in P supply. Infact, the study by Boersma and Elser (2006) summarizedmany findings stating that secondary production typicallyreaches a maximum when the P content of the food isaround 0.5–1.0% of dry weight. This translates to a molarC : P ratio of 125–250 (assuming C is 50% of dry weight),above which secondary production is limited by P. In ourstudy, the C : P ratio of the P-limited copepods was wellabove 300; therefore, P limitation for the fish was verylikely. Despite this fact, almost no studies exist thatinvestigate this in an ecological context (but see Hood etal. 2005). Schindler and Eby (1997) demonstrated thepossibility of P limitation on fish growth rates by means ofmodel exercises. P-limited growth rates were presented for 3out of 186 cases, while another 5 cases where probablyclose to limitation. The three cases they reported were forzooplanktivorous fish, where limitation of primary pro-duction is more likely to have an effect than in benthic foodwebs. Contrasting to the dominant P-rich cladocerans infreshwater systems, marine systems are dominated mainlyby copepods, which have a much higher N : P ratio thancladocerans (50 for calanoid copepods and 15 for Daphnia;Elser et al. 1996). In terms of P content, planktivorous fishin marine systems like herring should therefore moreregularly face food of lower quality (in terms of P) thantheir freshwater relatives are. Schindler and Eby’s models(Schindler and Eby 1997) (and others) assume homeostasisin fish body nutrients. However, it is likely that larval fishbody nutrient content is not homeostatic, especially for P,due to strong ontogenetic shifts like the shift frompredominantly muscle growth to the additional develop-ment of bones and fin rays later in the larval development,which have increasing P demands. This implies that thereare phases in a fish’s life that are more vulnerable to Plimitation. Similar processes have been shown for daph-nids, where differing P demands occur in young and olderdaphnids, suggesting nonhomeostatic body compositionbetween ontogenetic stages also for zooplankters (Boersma2000).

In summary, we found that nutrient-limitation signalscan travel up the food chain, not only as a decreasedquantity of the food (zooplankton) for secondary produ-cers, but also the quality of the zooplankton as food for fishcan change. This finding may have implications for thepredictions of fish population dynamics. Larval fish mighthave lower growth rates than the actual prey availabilitywould suggest, especially in situations where nutrients arelimited, such as at the end of a phytoplankton bloom.Hence, larval fish growth in the field is not likely to belimited by food quantity only, but also by food quality, andit could well be that given the current re-oligotrophicationin many systems, these effects will become stronger in years

to come. When considering match-mismatch situationsbetween predators and their prey, such quality-relatedeffects should be attended more specifically in futurestudies.

References

BOERSMA, M. 2000. The nutritional quality of P-limited algae forDaphnia. Limnol. Oceanogr. 45: 1157–1161.

———, AND J. J. ELSER. 2006. Too much of a good thing: Onstoichiometrically balanced diets and maximal growth.Ecology 87: 1325–1330.

———, AND C. KREUTZER. 2002. Life at the edge: Is food qualityreally of minor importance at low quantities? Ecology 83:2552–2561.

BRETT, M. T. 1993. Comment on ‘‘Possibility of N or P limitationfor planktonic cladocerans: An experimental test’’ (Urabe andWatanabe) and ‘‘Nutrient element limitation of zooplanktonproduction: (Hessen). Limnol. Oceanogr. 38: 1333–1337.

CLEMMESEN, C. 1993. Improvements in the fluorimetric determi-nation of the RNA and DNA content of individual marinefish larvae. Mar. Ecol. Prog. Ser. 100: 177–183.

———. 1994. The effect of food availability, age or size on theRNA/DNA ratio of individually measured herring larvae:Laboratory calibration. Mar. Biol. 118: 377–382.

DEMOTT, W. R., R. D. GULATI, AND K. SIEWERTSEN. 1998. Effectsof phosphorus-deficient diets on the carbon and phosphorusbalance of Daphnia magna. Limnol. Oceanogr. 43: 1147–1161.

ELSER, J. J., D. R. DOBBERFUHL, N. A. MACKAY, AND J. H.SCHAMPEL. 1996. Organism size, life history, and N : Pstoichiometry. BioScience 46: 674–684.

GISMERVIK, I. 1997. Stoichiometry of some marine planktoniccrustaceans. J. Plankton Res. 19: 279–285.

GRASSHOFF, K., K. KREMLING, AND M. ERHARDT. 1999. Methodsof seawater analysis, 3rd ed. Wiley-VCH.

GUILLARD, R., AND J. RYTHER. 1962. Studies of marine planktonicdiatoms. Can. J. Microbiol. 8: 229–239.

HESSEN, D. O. 1992. Nutrient element limitation of zooplanktonproduction. Am. Nat. 140: 799–814.

HILTON, G. M., R. W. FURNESS, AND D. C. HOUSTON. 2000. Theeffects of diet switching and mixing on digestion in seabirds.Funct. Ecol. 14: 145–154.

HOOD, J. M., M. J. VANNI, AND A. S. FLECKER. 2005. Nutrientrecycling by two phosphorus-rich grazing catfish: Thepotential for phosphorus-limitation of fish growth. Oecologia146: 247.

IRIGOIEN, X., K. J. FLYNN, AND R. P. HARRIS. 2005. Phytoplanktonblooms: A ‘loophole’ in microzooplankton grazing impact? J.Plankton Res. 27: 313–321.

IZQUIERDO, M. S., J. SOCORRO, L. ARANTZAMENDI, AND C. M.HERNANDEZ-CRUZ. 2000. Recent advances in lipid nutrition infish larvae. Fish Physiol. Biochem. 22: 97–107.

KLAUSMEIER, C. A., E. LITCHMAN, T. DAUFRESNE, AND S. A. LEVIN.2004. Optimal nitrogen-to-phosphorus stoichiometry of phy-toplankton. Nature 429: 171–174.

LEE, W. B., AND D. C. HOUSTON. 1993. The effect of diet qualityon gut anatomy in British voles (Microtinae). J. Comp.Physiol. B V163: 337.

LE PECQ, J. B., AND C. PAOLETTI. 1966. A new fluorometricmethod for RNA and DNA determination. Ann. Biochem.17: 100–107.

MALZAHN, A. M., M. BOERSMA, K. H. WILTSHIRE, C. CLEMMESEN,AND S. LAAKMANN. 2007. Comparative nutritional conditionof larval dab and lesser sandeel in a highly variableenvironment. Mar. Ecol. Prog. Ser. 334: 205–212.

2070 Malzahn et al.

———, C. CLEMMESEN, AND H. ROSENTHAL. 2003. Temperatureeffects on growth and nucleic acids in laboratory-reared larvalcoregonid fish. Mar. Ecol. Prog. Ser. 259: 285–293.

MULLER-NAVARRA, D. C. 1995. Biochemical versus minerallimitation in Daphnia. Limnol. Oceanogr. 40: 1209–1214.

PLATH, K., AND M. BOERSMA. 2001. Mineral limitation ofzooplankton: Stoichiometric constraints and optimal forag-ing. Ecology 82: 1260–1269.

RAUBENHEIMER, D., AND S. J. SIMPSON. 2003. Nutrient balancing ingrasshoppers: Behavioural and physiological correlates ofdietary breadth. J. Exp. Biol. 206: 1669–1681.

SABAT, P., J. A. LAGOS, AND F. BOZINOVIC. 1999. Test of theadaptive modulation hypothesis in rodents: Dietary flexibilityand enzyme plasticity. Comp. Biochem. Phys. A 123: 83.

SARGENT, J. R., AND S. FALK-PETERSEN. 1988. The lipid bio-chemistry of calanoid copepods. Hydrobiologia 167/168:101–114.

SCHINDLER, D. E., AND L. A. EBY. 1997. Stoichiometry of fishesand their prey: Implications for nutrient recycling. Ecology78: 1816–1831.

STERNER, R. W. 1993. Daphnia growth on varying quality ofScenedesmus: Mineral limitation of zooplankton. Ecology 74:2351–2360.

———, AND J. J. ELSER. 2002. Ecological stoichiometry: Thebiology of elements from molecules to the biosphere.Princeton University Press, Princeton, New Jersey.

———, AND D. O. HESSEN. 1994. Algal nutrient limitation and thenutrition of aquatic herbivores. Ann. Rev. Ecol. Sys. 25: 1–29.

ST. JOHN, M. A., C. CLEMMESEN, T. LUND, AND T. KOSTER. 2001.Diatom production in the marine environment: Implicationsfor larval fish growth and condition. ICES J. Mar. Sci. 58:1106–1113.

SUNEETHA, K. B., A. FOLKVORD, AND A. JOHANNESSEN. 1999.Responsiveness of selected condition measures of herring,Clupea harengus, larvae to starvation in relation to ontogenyand temperature. Environ. Biol. Fish. 54: 191–204.

TRIER, T. M., AND W. J. MATTSON. 2003. Diet-induced thermo-genesis in insects: A developing concept in nutritionalecology. Environ. Entomol. 32: 1–8.

URABE, J., J. CLASEN, AND R. W. STERNER. 1997. Phosphoruslimitation of Daphnia growth: Is it real? Limnol. Oceanogr.42: 1436–1443.

VIELMA, J., J. KOSKELA, AND K. RUOHONEN. 2002. Growth, bonemineralization, and heat and low oxygen tolerance inEuropean whitefish (Coregonus lavaretus L.) fed with gradedlevels of phosphorus. Aquaculture 212: 321–333.

ZANOTTO, F. P., S. M. GOUVEIA, S. J. SIMPSON, D. RAUBENHEIMER,AND P. C. CALDER. 1997. Nutritional homeostasis in locusts: Isthere a mechanism for increased energy expenditure duringcarbohydrate overfeeding? J. Exp. Biol. 200: 2437–2448.

Received: 5 December 2006Accepted: 18 April 2007

Amended: 2 May 2007

Nutrient limitation affects fish condition 2071