sign in sign up
<<
Home , Ecological efficiency, Ecological stoichiometry, Herbivore

Light, nutrients, and -chain length constrain planktonic transfer efficiency across multiple trophic levels

Elizabeth M. Dickman1, Jennifer M. Newell1, María J. Gonza´ lez, and Michael J. Vanni2

Department of Zoology, Miami University, Oxford, OH 45056

Edited by David M. Karl, University of Hawaii, Honolulu, HI, and approved October 3, 2008 (received for review June 8, 2008) The efficiency of energy transfer through food chains [ , because their chemical composition is similar efficiency (FCE)] is an important function. It has been to that of their prey (8). Thus, it has been proposed that the hypothesized that FCE across multiple trophic levels is constrained /nutrient stoichiometry of primary producers constrains by the efficiency at which use energy, which energy transfer across multiple trophic levels, i.e., from primary depends on plant nutritional quality. Furthermore, the number of producers to carnivores (8, 9), but this hypothesis has not been trophic levels may also constrain FCE, because herbivores are less explicitly tested. efficient in using plant production when they are constrained by The stoichiometry of aquatic primary producers () often carnivores. These hypotheses have not been tested experimentally reflects the supply of nutrients and light (8, 10, 11). Algal cell in food chains with 3 or more trophic levels. In a field experiment carbon/ (C/N) and carbon/ (C/P) ratios de- manipulating light, nutrients, and food-chain length, we show that crease with increasing nutrients and decreasing light intensity FCE is constrained by algal food quality and food-chain length. FCE (10, 12, 13), and the of aquatic herbivores across 3 trophic levels (phytoplankton to carnivorous fish) was is often higher under low light and/or high nutrient conditions, highest under low light and high nutrients, where algal quality was when algal C/P is relatively low (14–16). However, other aspects best as indicated by taxonomic composition and nutrient stoichi- of algal food quality may also covary with stoichiometry, such as ometry. In 3-level systems, FCE was constrained by the efficiency at morphological features (e.g., size, shape, presence of gelatinous which both herbivores and carnivores converted food into pro- sheaths) and biochemicals (e.g., essential fatty acid and sterol duction; a strong nutrient effect on efficiency suggests a concentrations) (11, 17). Because algal differ in these carryover effect of algal quality across 3 trophic levels. Energy characteristics, phytoplankton taxonomic identity may be a transfer efficiency from algae to herbivores was also higher in surrogate of food quality. 2-level systems (without carnivores) than in 3-level systems. Our We explored the general hypothesis that FCE depends on light results support the hypothesis that FCE is strongly constrained by intensity, nutrient supply, and food-chain length. This field study light, nutrients, and food-chain length and suggest that carryover explicitly quantifies how light and nutrients interactively regulate effects across multiple trophic levels are important. Because many FCE in systems with 3 trophic levels. In a field experiment using environmental perturbations affect light, nutrients, and food- aquatic mesocosms, we tested 3 specific hypotheses: (i) FCE (in chain length, and many ecological services are mediated by FCE, it food chains of equal length) is highest under low light/high will be important to apply these findings to various ecosystem nutrient conditions and lowest at high light/low nutrients; (ii) types. ecological efficiency is higher in food chains with just 2 trophic levels than with 3 trophic levels, because herbivores are ͉ ͉ ͉ ͉ ecological efficiency fish unconstrained by in 2-level systems; and (iii) because phytoplankton herbivores are more constrained than carnivores by food quality, FCE across 3 trophic levels is constrained by herbivore ecological lucidating the constraints on the efficiency of energy transfer efficiency (ratio of herbivore production to ). Ethrough food chains is necessary for understanding many ecological processes (1–6). Food chain efficiency (FCE), defined Results as the proportion of energy fixed by primary producers that is In treatments with 3 trophic levels, FCE increased with decreas- transferred to the top , depends on the ecological ing light (P Ͻ 0.0001) and increasing nutrients (P ϭ 0.0004), and efficiencies at each trophic coupling (1). FCE can regulate was highest in the low light/high nutrient treatment, as predicted attributes such as food-chain length and (7) and eco- by our first hypothesis (Fig. 1A). Across all treatments, herbivore system services such as fisheries production (2, 3), export of efficiency was affected by the main and interactive (P ϭ 0.0010) carbon from (4), and concentrations of contaminants effects of light and fish and was greater under low light condi- in (5). tions than under high light (P ϭ 0.0003). In support of our second Although FCE may regulate the number of trophic levels, the hypothesis, herbivore efficiency was much higher in the absence reverse may also be true: the number of trophic levels may of fish than in their presence (P Ͻ 0.0001; Fig. 1 B and C). determine FCE (1). With 3 trophic levels (, herbivores, and carnivores), herbivores may be held in check by carnivores and thus inefficiently consume plant biomass. However, with 2 (or 4) Author contributions: E.M.D., J.M.N., M.J.G., and M.J.V. designed research; E.M.D. and levels, herbivore biomass is unconstrained (or less constrained) J.M.N. performed research; E.M.D., J.M.N., M.J.G., and M.J.V. analyzed data; and E.M.D., by predation, possibly leading to higher herbivore production J.M.N., M.J.G., and M.J.V. wrote the paper. relative to primary production (1). The authors declare no conflict of interest. Ecological efficiencies often depend on food-quality attributes This article is a PNAS Direct Submission. such as edibility and nutritional quality. The ecological efficiency 1E.M.D. and J.M.N. contributed equally to this work. of herbivores often depends on plant nutrient stoichiometry 2To whom correspondence should be addressed. E-mail: [email protected]. (carbon/nutrient ratio) relative to the respiratory demands, This article contains supporting information online at www.pnas.org/cgi/content/full/ nutrient demands, and assimilation efficiency of the herbivore 0805566105/DCSupplemental. (6, 7). Stoichiometric constraints may be less important for © 2008 by The National Academy of Sciences of the USA

18408–18412 ͉ PNAS ͉ November 25, 2008 ͉ vol. 105 ͉ no. 47 www.pnas.org͞cgi͞doi͞10.1073͞pnas.0805566105 Downloaded by guest on September 29, 2021 Fish Present Fish Absent A Food Chain Efficiency (Phytoplankton to Fish) 0.07 A 300 B 350 a a 0.06 250 300 a 250 0.05 200 b 0.04 200 b b 150 c 0.03 150 d c

Seston C:P 100 0.02 100 c

Fish production / c 0.01 c 50 50 phytoplankton production 0.00 High Low High Low 0 0 Nutrients Nutrients Nutrients Nutrients High Light Low Light 14 14 C D a 12 12 Herbivore Efficiency (Phytoplankton to Zooplankton) a a 0.20 2.0 10 a a 10 b Fish Present Fish Absent b b B C 8 8 0.15 1.5 a a 6 6

b Seston C:N 0.10 1.0 4 4 b b 2 2 0.05 0.5 b 0 0 b b Zooplankton production / phytoplankton production 0.00 0.0 High Low High Low High Low High Low E 1.5 F 1.5 Nutrients Nutrients Nutrients Nutrients Nutrients Nutrients Nutrients Nutrients a High Light Low Light High Light Low Light a 1.0 1.0 Carnivore Efficiency (Zooplankton to Fish) ab D 1.5 b b b D b

food quality 0.5 0.5 1.2 b a 0.9 a

Phytoplankton compositional 0.0 0.0 0.6 High Low High Low High Low High Low Nutrients Nutrients Nutrients Nutrients Nutrients Nutrients Nutrients Nutrients ab

Fish production / 0.3 High Light Low Light High Light Low Light

zooplankton production b 0.0 High Low High Low Fig. 2. Quality of phytoplankton as a food based on cell stoichi- Nutrients Nutrients Nutrients Nutrients High Light Low Light ometry (A–D) and taxonomic composition (E and F). Phytoplankton responses were measured in each mesocosm throughout the study, and each point Fig. 1. Light, nutrient, and fish effects on FCE (2-way ANOVA, n ϭ 12, P ϭ represents a mesocosm mean, averaged over the experiment. Horizontal lines 0.0009) (A), herbivore efficiency (3-way ANOVA, n ϭ 23, P ϭ 0.0003) (B and C), represent treatment means, and letters indicate treatments that are signifi- and carnivore efficiency (2-way ANOVA, n ϭ 12, P ϭ 0.0138) (D). Each point cantly different from each other (Tukey post hoc test). Treatments with fish represents the efficiency for an individual mesocosm, obtained by using present and absent were analyzed separately. production rates averaged over the experiment for each trophic level. Hori- zontal lines represent treatment means, and letters indicate treatments that are significantly different from each other (Tukey post hoc test). For herbivore Nutrients decreased seston C/P at low light (as expected), but efficiency, all 8 treatments (fish absent and fish present) were analyzed increased C/P at high light (Fig. 2 A and B and Table S1). It is together, although they are graphically depicted separately. Note that the unclear why C/P increased in response to nutrients in the high scale differs in fish present vs. fish absent treatments for herbivore efficiency. In 2 mesocosms, carnivore efficiency exceeded 1, possibly because toward the light treatments; perhaps the marked increase in phytoplankton end of the experiment fish consumed some benthic algae, although zooplank- biomass (Fig. 3) caused a depletion in nutrients that led to low ton still made up the majority of their diet (see SI Text for more details). Note cell P content. Supportive of this idea, in the high light treat- that herbivore efficiency was Ͼ1 in some low light mesocosms, possibly ments, soluble reactive P concentrations (SRP) were not signif- because of some consumption by zooplankton of other than phyto- icantly different in treatments with and without nutrient addi- plankton, such as other zooplankton (), periphyton, and tion; in contrast, at low light, nutrient addition significantly . The relative contributions of the production of bacteria and per- increased SRP (19). Among-treatment differences in seston C/N iphyton, relative to PPr, were higher in the low-light treatments (unpublished were qualitatively similar to those for C/P (Fig. 2 C and D and data). Diagrams represent efficiencies depicted in the graphs. Note the dif- ference in scale between fish absent and fish present treatments. Table S1). Phytoplankton compositional food quality was generally higher at low light (than at high light) and at high nutrients (than Because fish had strong effects on herbivore efficiency, and we at low nutrients); thus, compositional quality was highest in the wanted to explore FCE in 3-level systems, we also separately low-light/high-nutrient treatments (Fig. 2 E and F and Table S1). analyzed treatments with and without fish. In treatments without Cryptomonads, other small flagellates, and diatoms, typically fish, increased light intensity decreased herbivore efficiency (P ϭ considered high-quality foods because of their edible size and 0.0041; Fig. 1C). Our third hypothesis, that herbivore efficiency high nutritional content (18), dominated phytoplankton in the constrains FCE, was partially supported, because FCE was also low-light/high-nutrient treatments (Fig. 3). In contrast, in the associated with carnivore efficiency, i.e., the ratio of carnivore high-light/low-nutrient treatments, phytoplankton biomass was production to herbivore production (Fig. 1D). Specifically, in- comprised mostly of cyanobacteria (poor food quality) and creased light intensity decreased both herbivore efficiency and chlorophytes (intermediate quality; Fig. 3). Differences in phy- FCE (Fig. 1 A–C), whereas increased nutrients increased both toplankton composition (especially cryptomonads vs. cyanobac- carnivore efficiency and FCE (Fig. 1 A and D). The observation teria) between treatments were more pronounced in the pres- that nutrients increased FCE and carnivore efficiency, but not ence of fish than in the absence of fish (Fig. 3). Thus, in the herbivore efficiency, suggests that carnivore efficiency at least 3-level systems, FCE was highest in the low-light/high-nutrient partially constrained FCE. treatment, where phytoplankton compositional and stoichiomet- Among-treatment differences in ecological efficiencies were ric food quality were also highest. accompanied by large differences in phytoplankton quality, We used stepwise multiple regressions to better understand based on both phytoplankton stoichiometry and a food-quality the factors mediating efficiencies. In the 3-level systems, the best index based on phytoplankton taxonomic composition (which is model (P ϭ 0.0052; Table S2) explained 98% of variation in FCE associated with food quality (18) and size [supporting informa- and included phytoplankton compositional food quality (which tion (SI) Table S1 and Fig. 2]. Phytoplankton (seston) C/P was itself explained 85% of variance), seston stoichiometry, and lower in low light treatments in both 2- and 3-level systems. zooplankton composition as independent variables.

Dickman et al. PNAS ͉ November 25, 2008 ͉ vol. 105 ͉ no. 47 ͉ 18409 Downloaded by guest on September 29, 2021 Fish Present Fish Absent whereas compositional food quality exhibited a unimodal trend 8 8 in response to increasing PPr (Fig. 4 D and E). The nonlinear a A B 7 7 a relationship between PPr and fish production probably reflects /mL)

3 interactive effects of light and nutrients on phytoplankton 6 b c 6 ab

m c b b compositional and stoichiometric quality. Indeed, in terms of 5 5 phytoplankton quality (and fish production), the 4 treatments log(Biovolume) ( 4 4 were rather distinct from each other, another indication that particular combinations of light and nutrients drove the responses. 100 100 Phytoplankton C D To further evaluate how fish performance responded to Cya Bg Cya manipulations of light and nutrients, we compared fish body 75 Cryp 75 Dia nutrient contents and ratios at the end of the experiment. Fish

50 Fla mass was used as a covariate because body nutrients often vary Dia 50 % composition Grn Grn with fish size (20). Fish body C and P contents were significantly

25 25 different among treatments but responded in opposite ways (Fig. 5). Fish body P was highest, and fish body C was lowest, in Oth 0 0 Oth HL, HN, +F HL, LN, +F LL, HN, +F LL, LN, +F treatments with nutrients added; both of these responses were High Low High Low HighHL, HN LowHL, LN HighLL, HN LowLL, LN Nutr. Nutr. Nutr. Nutr. Nutr. Nutr. Nutr. Nutr. greater at low light than at high light (Fig. 5). As a consequence, High Light Low Light High Light Low Light fish body C/P and C/N were also lowest in the low-light/high-

3 3 nutrient treatment. Such strong effects on fish body C and P were a E ab b F somewhat unexpected, because fish body stoichiometry (includ- a a a c g/L) 2 2 ing that of Dorosoma cepedianum) seems to vary predictably with a ontogeny (21), implying a possible insensitivity to environmental 1 1

log(Biomass) ( factors such a nutrients or light. 0 0 Discussion

100 100 Our results support the hypothesis that increased nutrients G H Zooplankton Cla Cla and/or decreased light intensity enhances FCE. The mechanisms 75 75 accounting for these responses are difficult to discern because of Adult potential covariation between different aspects of phytoplank- 50 cop 50 ton food quality (11, 16, 18, 22, 23). Based on our regressions, Adult composition % Nau cop phytoplankton compositional food quality had the strongest 25 25 effects on FCE (presumably because taxonomic identity is Rot Nau 0 0 related to edibility and nutritional quality), but considering HighHL, HN, +F LowHL, LN, +F HighLL, HN, +F LowLL, LN, +F HighHL, HN LowHL, LN HighLL, HN LowLL, LN herbivore efficiency alone, stoichiometric quality had much Nutr. Nutr. Nutr. Nutr. Nutr. Nutr. Nutr. Nutr. High Light Low Light High Light Low Light stronger effects (Table S2). In 3-level systems, high FCE was associated with a relative of cryptomonads and Fig. 3. Biomass and community composition responses of phytoplankton and zooplankton. (A, B, E, and F) Phytoplankton (A and B) and zooplankton diatoms, which dominated under low-light/high-nutrient condi- (E and F) biomass. Each point represents the mean for a mesocosm, averaged tions. These taxa have high concentrations of essential fatty acids over the experiment. Horizontal lines represent treatment means, and letters (23) and are high-quality food for zooplankton (18). In our indicate treatments that are significantly different from each other (Tukey experiment, these taxa were also associated with low seston C/P. post hoc test). (C and D) Relative biovolume of major phytoplankton taxo- In contrast, cyanobacteria, which were most abundant in 3-level nomic groups for each treatment, averaged across the experiment. Cya, systems with high light and low nutrients, have low fatty acid and cyanobacteria; Cryp, cryptomonads; Dia, diatoms; Grn, green algae; Fla, small sterol concentrations, represent the lowest food quality of any flagellates; Oth, phytoplankton groups that made up Ͻ10% of total phyto- taxonomic group, and were associated with high C/P. In addition plankton biovolume when averaged across the study within a treatment. (G to the low fatty acid content of cyanobacteria, recent studies and H) The relative biomass of major zooplankton taxonomic groups for each treatment was averaged across the experiment. Cla, cladocerans; Adult cop, indicate that the absence of sterols in cyanobacteria may con- adult copepods; Nau, nauplii; Rot, rotifers. tribute to their low quality as food for zooplankton (17). Chlorophytes are generally of intermediate quality in terms of these indicators (18). In whole phytoplankton assemblages, Seston stoichiometry explained 45–67% of the variation in correlations between species shifts and seston C/P may reflect the herbivore efficiency, depending on whether analyses considered different optimal cell stoichiometries of different taxa (24), but all treatments or subsets with and without fish (Table S2). additional study is needed to disentangle these interactions (11). Carnivore efficiency was positively related to percentage of FCE was mediated by both herbivore efficiency and carnivore biomass of rotifers and adult copepods (Fig. 3) and phytoplank- efficiency. Herbivore efficiency and FCE were both higher at low ton compositional quality, which together explained 69% of the light, and both carnivore efficiency and FCE were higher with elevated nutrient supply (Fig. 1). The response of both carnivore variation in carnivore efficiency (Table S2). One caveat of these efficiency and FCE to nutrients may represent a ‘‘carryover analyses is that seston stoichiometry and compositional food effect’’ across 3 trophic levels. When algal cells have high r2 ϭ quality were moderately correlated with each other (e.g., concentrations of P and/or essential fatty acids, zooplankton also 0.41 for the relationship between seston C/P and compositional may exhibit high concentrations of these resources and thus quality). Thus, it is difficult to determine the causal effect of represent higher-quality food for fish, which have high P re- specific independent variables. quirements for rapid growth (16, 23, 25). In laboratory experi- Interactions between light and nutrients produced a variety of ments, differences in zooplankton P content, resulting from relationships between phytoplankton production (PPr) and fish differences in phytoplankton P status, have been shown to affect responses and indicators of algal quality. Fish survival increased fish growth (16). In our experiment, fish body P content in- linearly with PPr, but the mean mass of surviving fish was lowest creased when nutrients were added, suggesting that fish may with high PPr; as a result, fish production was unimodally related have been P-limited. P limitation of fish growth might be to PPr (Fig. 4 A–C). Seston C/P was highest with high PPr, expected in rapidly growing larvae, because P is needed for RNA

18410 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0805566105 Dickman et al. Downloaded by guest on September 29, 2021 0.25 180 8 2 2 A r =0.84 B r2=0.45 C r =0.21 p<0.0001 160 p=0.0693 7 p=0.3412 0.20 140 6

120 gC/L/day) 5

0.15 µ 100 4 0.10 80 3 High nutrients 60 2 0.05 Low nutrients High light Proportional fish survival High nutrients 40 1 Low light Fish production ( Low nutrients 0.0 Median fish dry weight (mg/fish) 20 0 0 100 200 300 400 500 0 100 200 300 400 500 0 100 200 300 400 500

300 1.5 D E 2 r2=0.85 r =0.13 p=0.5247 250 p=0.0002

1.0 200

150 0.5

Seston C:P (molar) Seston C:P 100 Compositional food quality 50 0.0 0 100 200 300 400 500 0 100 200 300 400 500 PPr (µgC/L/day)

Fig. 4. Relationships between PPr and phytoplankton food quality and fish responses. Each point represents the value for an individual mesocosm. Except in A and B, values are those averaged over the experiment. (A) Proportion of fish surviving to the end of the experiment. (B) Mean fish dry mass at the end of the experiment. (C) Fish production. (D) Phytoplankton C/P stoichiometric food quality. (E) Phytoplankton compositional food quality. Phytoplankton compositional and stoichiometric food quality data are shown only for treatments with fish present. Best fit lines, r2, and P values are given for each relationship, using either linear or polynomial regression, whichever gave the best fit. High light treatments are represented by squares, and low light treatments are indicated by circles. High nutrient treatments are depicted with closed symbols, and low nutrient treatments are depicted with open symbols.

synthesis (necessary for rapid growth) and development environmental perturbations. For example, eutrophication in- (because are comprised of calcium phosphate) (16, 21). creases nutrient supply and decreases light intensity (26), and However, we cannot exclude the possibility that increased car- change can affect trophic couplings by altering nutrients, nivore efficiency under low-light/high-nutrient conditions is light, and temperature (27, 28). Fishery yields, which constitute caused by increased fatty acid contents of zooplankton resulting one of the most globally important ecosystem services, are from shifts in phytoplankton taxonomic composition (23). constrained by FCE (3). Thus, understanding how light and Our study shows that FCE is constrained by light and nutrient nutrient supplies mediate FCE may help explain some of the supply, resources that are greatly affected by many global observed variation in the relationship between primary produc-

A Carbon B Nitrogen C Phosphorus ECOLOGY 0.50 0.13 0.05

0.04 0.45 0.12 0.03

g fish dry mass 0.02 µ g fish dry mass µ 0.40 0.11 µ g fish dry mass g N/ g C/ 0.01 µ µ Treatment: p<0.0001 Treatment x weight: p=0.0173 µ g P/ Treatment: p=0.0033 Treatment x weight: p=0.0478 0.35 0.10 0 01234 01234 01234

C:N C:P N:P D E F High nutrients High nutrients High light 5.0 60 14 Low nutrients High nutrients Low light 50 12 Low nutrients 4.5 10 40 8 4.0 30 6 20 4 Fish C:P (molar) Fish N:P (molar) Fish C:N (molar) 3.5 Treatment: p=0.0002 10 Treatment: p=0.0006 2 Treatment: p=0.0030 Treatment x weight: p=0.0157 3.0 0 0 01234 01234 01234 fish wet mass (g)

Fig. 5. Body C (P Ͻ 0.0001), N (P ϭ 0.0485), and P (P ϭ 0.0034) (A–C) and C:N (P Ͻ 0.0001), C:P (P ϭ 0.0007), and N:P (P ϭ 0.0038) (D–F) of larval gizzard shad, regressed against fish wet mass at the end of the experiment (ANCOVA, n ϭ 31 for all variables). Significant (P Ͻ 0.05) ANCOVA results are indicated.

Dickman et al. PNAS ͉ November 25, 2008 ͉ vol. 105 ͉ no. 47 ͉ 18411 Downloaded by guest on September 29, 2021 tion and fishery yields (29). Conversely, overfishing may affect We quantified phytoplankton, zooplankton, and limnological parameters functional food-chain length (2, 30, 31) possibly shifting food before (May 30, 2005) and for 8 weeks after light, nutrient, and fish treatments chains between odd- and even-number systems, with additional were applied. We quantified biomass, community composition, and produc- tion of all 3 trophic levels, and the nutrient contents and ratios (C/N/P) of implications for FCE. Given the manifold importance of FCE, phytoplankton and fish. We attempted to quantify zooplankton C/N/P. How- it is imperative that future studies elucidate the particular ever, we were unable to feasibly separate phytoplankton from zooplankton mechanisms by which it is mediated by light, nutrients, and assemblages because they overlapped in size, and it was impractical to quan- food-chain length, especially in systems where ecosystem ser- tify body C/N/P on individual zooplankton taxa with the number of replicates vices depend on this efficiency. and sampling dates used in our experiment (see SI Text). Production of each trophic level was measured in units of ␮gofCLϪ1⅐dϪ1 to allow us to calculate Methods FCE. We assessed primary production by using 14C uptake, zooplankton pro- duction by using temporal dynamics in body size and egg production, and fish We manipulated light and nutrient supply and the presence or absence of production by using initial and final biomass (see SI Text for details on sample planktivorous fish in 24 polyethylene 5000L mesocosms containing local as- collection and production methods). FCE was quantified as production of the semblages of phytoplankton and zooplankton. We used a full-factorial de- top trophic level divided by PPr. In addition, we quantified ecological effi- sign, with 2 levels of light (high and low), 2 levels of nutrients (high and low), ciencies at each trophic coupling: herbivore efficiency ϭ zooplankton produc- and 2 or 3 trophic levels (presence or absence of planktivorous fish), for a total tion/PPr, and carnivore efficiency ϭ fish production/zooplankton production. of 8 treatments with 3 replicates of each. Light was manipulated by using lids A detailed description of methods is presented in SI Text. with or without 90% light reduction shade cloth. Nutrients were added to half of the mesocosms 3 times per week as NH4NO3 (50 ␮g of N/L) and as ACKNOWLEDGMENTS. We thank 2 anonymous reviewers and A. Babler, M. NaH2PO4⅐H2O(5␮g of P/L). Larval gizzard shad (D. cepedianum) were added Boone, J. Duncan, L. Hagenbuch, M. Horgan, L. Knoll, F. Rowland, H. Stevens, to half of the mesocosms at a density of 250 fish per mesocosm at the and C. Williamson for comments on the manuscript; J. Elder and P. Levi for field and laboratory assistance; and R. Kolb (Miami University Ecology Research beginning of the study. We began the experiment with a low, but realistic, Center) for logistical support. This research was supported by the Cooperative biomass of fish to allow for quantifiable fish production. Although gizzard State Research, Education, and Extension Service, U.S. Department of Agri- shad are omnivorous across all life stages, larvae are carnivores, feeding on culture, under Award OHOR-2003-01756, National Science Foundation Grant zooplankton (31). DEB 0235755, and the Miami University Ecology Research Center.

1. Hairston NG, Hairston NG (1993) Cause–effect relationships in energy-flow, trophic 17. Martin-Creuzburg D, von Elert E, Hoffmann KH (2008) Nutritional constraints at the structure and interspecific interactions. Am Nat 142:379–411. cyanobacteria– magna interface: The role of sterols. Limnol Oceanogr 53:456– 2. Pauly D, Christensen V, Dalsgaard J, Froese R, Torres F (1998) Fishing down marine food 468. webs. Science 279:860–863. 18. Brett MT, Muller-Navarra DC, Park SK (2000) Empirical analysis of the effect of phos- 3. Ware DM, Thomson RE (2005) Bottom-up ecosystem trophic dynamics determine fish phorus limitation on algal food quality for freshwater zooplankton. Limnol Oceanogr production in the northeast Pacific. Science 308:1280–1284. 45:1564–1575. 4. Richardson TL, Jackson GA (2007) Small phytoplankton and carbon export from the 19. Dickman EM (2007) Light, nutrients, and planktivory effects on phytoplankton com- surface ocean. Science 315:838–840. munity and stoichiometric response and food chain efficiency. Masters thesis (Miami 5. Vander Zanden MJ, Rasmussen JB (1996) A trophic position model of pelagic food University, Oxford, OH). webs: Impact on contaminant in trout. Ecol Monogr 66:451–477. 20. Hendrixson HA, Sterner RW, Kay AD (2007) Elemental stoichiometry of freshwater 6. Elser JJ, et al. (2000) Nutritional constraints in terrestrial and freshwater food webs. fishes in relation to phylogeny, and ecology. J Fish Biol 70:121–140. Nature 408:578–580. 21. Pilati A, Vanni MJ (2007) Ontogeny, diet shifts, and nutrient stoichiometry in fish. Oikos 7. Frost PC, et al. (2006) Threshold elemental ratios of carbon and phosphorus in aquatic 116:1663–1674. consumers. Ecol Lett 9:774–779. 22. Muller-Navarra DC (1995) Biochemical versus limitation in Daphnia. Limnol 8. Sterner RW, Elser JJ (2002) Ecological Stoichiometry: The of Elements From Oceanogr 40:1209–1214. Molecules to the Biosphere (Princeton Univ Press, Princeton). 23. Brett MT, Muller-Navarra DC, Ballantyne AP, Ravet JL, Goldman CR (2006) Daphnia 9. Sterner RW, Clasen J, Lampert W, Weisse T (1998) Carbon/phosphorus stoichiometry fatty acid composition reflects that of their diet. Limnol Oceanogr 51:2428–2437. and food chain production. Ecol Lett 1:146–150. 24. Klausmeier CA, Litchman E, Daufresne T, Levin SA (2004) Optimal nitrogen-to- 10. Sterner RW, Elser JJ, Fee EJ, Guildford SJ, Chrzanowski TH (1997) The light/nutrient phoshorus stoichiometry of phytoplankton. Nature 429:171–174. ratio in : The balance of energy and materials affect ecosystem structure and 25. Schindler DE, Eby LA (1997) Stoichiometry of fishes and their prey: Implications for process. Am Nat 150:663–684. nutrient recycling. Ecology 78:1816–1831. 11. Hall SR, Leibold MA, Lytle DA, Smith VH (2007) Grazers, producer stoichiometry, and 26. Schindler DW (2006) Recent advances in the understanding and management of the light/nutrient hypothesis revisited. Ecology 88:1142–1152. eutrophication. Limnol Oceanogr 51:356–363. 12. Hall SR, Leibold MA, Lytle DA, Smith VH (2004) Stoichiometry and planktonic grazer 27. Richardson AJ, Schoeman DS (2004) Climate impact on plankton ecosystems in the composition over gradients of light, nutrients, and predation risk. Ecology 85:2291– Northeast Atlantic. Science 305:1609–1612. 2301. 28. Schindler DW, Curtis PJ, Parker BR, Stainton MP (1996) Consequences of climate 13. Dickman EM, Vanni MJ, Horgan MJ (2006) Interactive effects of light and nutrients on warming and lake acidification for UV-B penetration in North American lakes. phytoplankton stoichiometry. Oecologia 149:676–689. Nature 379:705–708. 14. Urabe J, et al. (2002) Reduced light increases herbivore production due to stoichio- 29. Downing JA, Plante C, Lalonde S (1990) Fish production correlated with primary metric effects of light/nutrient balance. Ecology 83:619–627. , not the morphodeaphic index. Can J Fish Aqua Sci 47:1929–1936. 15. Urabe J, Sterner RW (1996) Regulation of herbivore growth by the balance of light and 30. Myers RA, Baum JK, Shepherd TD, Powers SP, Peterson CH (2007) Cascading effects of nutrients. Proc Natl Acad Sci USA 93:8465–8469. the loss of apex predatory sharks from a coastal ocean. Science 315:1846–1850. 16. Malzahn AM, Aberle N, Clemmesen C, Boersma M (2007) Nutrient limitation of primary 31. McIntyre PB, Jones LE, Flecker AS, Vanni MJ (2007) Fish extinctions alter nutrient producers affects planktivorous fish condition. Limnol Oceanogr 52:2062–2071. recycling in tropical freshwaters. Proc Natl Acad Sci USA 104:4461–4466.

18412 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0805566105 Dickman et al. Downloaded by guest on September 29, 2021