Ecological stoichiometry in the microbial food web: a test of the light: nutrient hypothesis

AQUATIC MICROBIAL ECOLOGYAquat Microb Ecol

Vol. 31: 49–65, 2003 Published February 13

INTRODUCTION

The microbial food web is widely recognized as akey conduit of energy and material flows in pelagicecosystems (Azam et al. 1983, Sherr & Sherr 1991,Hwang & Heath 1997, Cotner & Biddanda 2002). Forexample, phytoplankton exude labile organic matter(as dissolved organic carbon, DOC) and thus supporttrophic pathways involving bacteria, bacterivorous fla-gellates, and eventually larger consumers (Azam et al.1983, Güde 1989, Guerrini et al. 1998). While suchmicrobial connections are omnipresent, their impor-tance relative to the conventional ‘grazing chain’

appears to vary among systems for reasons that arepoorly understood. Bacteria and algae are oftenthought to exist in a loose mutualism, where algae pro-vide DOC to support bacterial growth while bacteriaremineralize organically bound nutrients to supportphytoplankton production. However, the mutualisticscenario between bacteria and phytoplankton is com-plicated by the fact that, in many cases, bacteria arenot the primary remineralizers of organically boundnutrients (Azam et al. 1983, Ferrier Pages & Ras-soulzadegan 1994, Vadstein 2000) and bacteria cancompete with algae for inorganic nutrients (Currie &Kalff 1984, Cotner & Wetzel 1992). In addition, it

© Inter-Research 2003 · www.int-res.com*Email: [email protected]

Ecological stoichiometry in the microbial food web:a test of the light:nutrient hypothesis

J. J. Elser1,*, M. Kyle1, W. Makino2, T. Yoshida2, 3, J. Urabe2

1Department of Biology, Arizona State University, Tempe, Arizona 85287-1501, USA2Center for Ecological Research, Kyoto University, Kamitanakami Hirano, Otsu 520-2113, Japan

3Present address: Department of Ecology and Evolutionary Biology, Cornell University, Ithaca, New York 14853, USA

ABSTRACT: The ‘light:nutrient hypothesis’ (LNH) states that the importance of the microbial foodweb relative to grazing impacts by macrozooplankton and the nature of the relationship betweenalgae and bacteria (competition or commensalism) are affected by the balance of light and nutrientsexperienced by phytoplankton. We tested this hypothesis in field enclosures by manipulating irradi-ance and nutrient supply in a P-limited lake in Ontario, Canada. Shading and P-enrichment had littleeffect on standing biomass of small suspended particles (<1 µm) but both decreased C:P ratio in thissize fraction. P-fertilization had no effect on algal biomass or bacterial abundance but shading signif-icantly lowered algal biomass and the ratio of algal biomass to bacterial abundance. Shading had noeffect on heterotrophic nanoflagellates (HNF) but HNF abundance declined strongly with P-enrich-ment under both shaded and unshaded conditions, coincident with large increases in macrozoo-plankton biomass that accompanied P-fertilization. Shading and nutrient enrichment also affectedresource limitation as indicated by dilution bioassays. Increased algal light limitation in the enclo-sures was associated with lower seston C:P ratio, as expected under the LNH. In addition, the C:Pratio of new seston produced in bioassay bottles was also affected by enclosure treatment; unshadedenclosures generally produced seston with moderately high C:P ratio while shaded enclosures pro-duced new seston at low C:P under low ambient light but at very high C:P ratio when algae wereincubated at higher light. Our data provide support for some aspects of the LNH, including predictedimpacts on the microbial loop and on resource limitations and C:P stoichiometry. However, someresponses were not consistent with the LNH, indicating that the hypothesis needs to be modified toincorporate potential indirect effects.

KEY WORDS: Microzooplankton · Stoichiometry · Light · Phosphorus · Bacteria

Resale or republication not permitted without written consent of the publisher

Aquat Microb Ecol 31: 49–65, 2003

appears that in many situations bacterial production isnot limited by DOC but by the supply of key nutrients,such as N or especially P (Morris & Lewis 1992, Elser etal. 1995b, Cotner et al. 1997, Vadstein 2000, Carlsson &Caron 2001). Under such conditions the relationshipbetween bacteria and phytoplankton is not mutualistic(+, +) but competitive (–, –). Indeed, in chemostatexperiments bacteria rapidly outcompete phytoplank-ton for inorganic nutrients when the supply of labileDOC is high relative to the supply of nutrients (Bratbak& Thingstad 1985, Grover 1997, 2000).

In addition to the intensive study of factors influenc-ing the resource interactions of algae and bacteria inthe pelagic zone, aquatic studies (Güde 1989, Sanders& Wickham 1993, Jürgens 1994, Pace & Cole 1996,Cole 1999, Langenheder & Jürgens 2001) have quanti-fied the impact of microzooplankton on both algae andbacteria and attempted to understand the factors influ-encing the relative importance of ‘conventional’trophic links in the pelagic food web (algae to macro-zooplankton to fish) relative to the ‘microbial loop’(algal DOC and detritus to bacteria to small protozoa tomacrozooplankton to fish). The dominance of these 2pathways is known to vary with trophic structure; forexample, in lake food webs where macrozooplanktonare suppressed by fish predation, microzooplanktonplay a dominant role in applying grazing pressure onbacteria and small phytoplankton (Jürgens 1994, Pace& Cole 1996). Trophic status also appears to be ofimportance, as the microbial food web appears to playa proportionately larger role in trophic interactions inoligotrophic relative to eutrophic conditions (Cotner &Biddanda 2002). Finally, microbial interactions play akey role in nutrient cycling in the pelagic zone andindeed, it now appears that in many situations nutri-ents are remineralized primarily by bacterivorous fla-gellates (Rothaupt 1992, Vadstein 2000) and not bybacteria themselves, which is consistent with thestrong tendency of bacteria to experience nutrientlimitation. What emerges from these developments is apicture of a complex set of internal and external inter-actions connecting pelagic trophic interactions withnutrient cycling and resulting in considerable variationamong ecosystems in the relative importance of vari-ous pathways and dominant players.

To explain this variation in the context of thesecomplex factors, Sterner et al. (1997) proposed a‘light:nutrient’ hypothesis (LNH) pertaining to the rela-tionships between algae and bacteria, the relativeimportance of the microbial food web, as well as theproduction and community composition of metazoanzooplankton in lakes. They proposed that variations inthe supply balance of solar radiation and key inorganicnutrients (such as PO4) alter the severity of nutrientlimitation of phytoplankton growth, with subsequent

effects on metazoan herbivores and the microbial foodweb. More specifically, high light:nutrient supplyratios should accentuate algal nutrient limitation, asincreased light increases phytoplankton demand fornutrients (by raising algal potential growth rate, µm;Shuter 1979) that may not be matched by ambient sup-ply processes. One consequence of these responses isincreased biomass C:nutrient ratio in algal biomass.Conversely, low light:nutrient ratios should promotephytoplankton light limitation, leading to low µm andmore easily allowing nutrient supply to keep pace withalgal demand. As a result, algal C:nutrient ratiosshould be low and roughly in Redfield proportions.Such changes in the severity of algal nutrient limitationshould affect the microbial food web in 2 ways. First,algae with high C:nutrient ratios are poor food forzooplankton (Sterner & Hessen 1994, Elser et al. 2001).Thus, macrozooplankton production should be im-paired, which should increase the importance ofmicrobial trophic links as protozoa and ciliates wouldno longer be suppressed due to consumption bymacrozooplankton. Second, severely nutrient-limited(especially P-limited) phytoplankton tend to increasetheir exudation of labile organic carbon (Lancelot1983, Obernosterer & Herndl 1995, Guerrini et al.1998, Berman-Frank & Dubinsky 1999, Corzo et al.2000), which increases DOC availability to bacteriaand shifts the resource limiting bacterial growth fromDOC to inorganic nutrients, thus shifting the competi-tive advantage from algae to bacteria. Therefore,bacteria should have greater relative abundance,compared to algae, in systems with high light:nutrientratio (Sterner et al. 1997).

To date, various components of the LNH have beentested, primarily in laboratory settings. For example,effects of light:nutrient balance on Daphnia growth(Urabe & Sterner 1996) and population dynamics(Sterner et al. 1998) have been demonstrated in labo-ratory settings involving PO4 limitation of algae. Inthese experiments, increased light inhibited Daphniaperformance by raising algal C:P ratios above optimallevels. In the study of Urabe & Sterner (1996), a clearinteraction between light and PO4 supply was ob-served, as higher light intensities were required toinhibit Daphnia growth as PO4 availability increased.Such effects have now been demonstrated for naturalconditions (Urabe et al. 2002b) in an experiment inwhich a 10-fold reduction in incoming solar radiationto field enclosures resulted in a 5-fold increase inzooplankton production due to an improvement instoichiometric food quality. This experiment alsodemonstrated a light-nutrient interaction, as loweringirradiance at high PO4 inputs resulted in decreasedzooplankton production due to lowered overall abun-dance of food. Thus, the predictions from the LNH for

50

Elser et al.: Light:nutrient balance and microbes

macrozooplankton dynamics seem well-supported byavailable data.

However, evidence regarding effects of light:nutri-ent balance on components of the microbial food webis limited. Sterner et al. (1998) reported no effect ofaltered irradiance on bacterial abundance and produc-tion in their artificial ecosystem experiment. In labora-tory experiments involving mixed bacteria assem-blages and the green alga Scenedesmus, Gurung et al.(1999) showed that bacteria resource limitation wasinfluenced by both light and PO4 supplies, with DOC-limitation generally strongest at low light and high PO4

supply, while PO4-limitation was strongest at high lightand low PO4 inputs. Furthermore, they showed that thedominance of bacteria relative to algae decreased withincreasing light (contrary to the LNH) but increasedwith increasing PO4 supply at low irradiance (consis-tent with the LNH). While these results provide provi-sional support for some aspects of the LNH, theyinvolved laboratory settings with unialgal cultures inthe absence of higher trophic levels, including proto-zoa and macrozooplankton. Thus, to date there havebeen few tests of microbial aspects of the LNH undernatural conditions and over ecologically significanttime scales. However, an observational study hasrecently appeared (Chrzanowski & Grover 2001), onbacterial resource limitation in 2 Texas reservoirs.They reported results generally consistent with theLNH, in that bacterial growth was more likely to be P-than DOC-limited during periods when epilimeticlight:nutrient ratios were high. To examine howchanges in environmental light and nutrient suppliesaffect food web structure in pelagic ecosystems, weperformed a large in situ enclosure experiment in aCanadian Shield lake. Since major responses of totalseston stoichiometry and zooplankton production inthis experiment have already been presented else-where (Urabe et al. 2002b), here we report responsesof microbial food web components to achieve a morecomplete assessment of the responses of the wholeplanktonic food web to these factors.

MATERIALS AND METHODS

Study site. Experimental manipulations were main-tained for 26 d from 8 July to 3 August 1999 in Lake239 (L239 hereafter), a relatively large (56 ha, 30 mmaximum depth) oligotrophic lake (TP = ~0.17 µM) atthe Experimental Lakes Area (ELA), northwesternOntario, Canada (93° 30’ to 94° 00’ W, 49° 30’ to49° 45’ N, altitude 360 to 380 m above sea level). Con-centrations of dissolved inorganic nitrogen and silicaare relatively high in lakes of the ELA and thus P haslong been recognized as the key limiting nutrient in

these systems (Schindler 1977, Schindler et al. 1978).Consistent with overall scarcity of P, seston C:P ratiosin the euphotic zone of ELA lakes are high and havebeen directly linked to mineral (P) limitation of zoo-plankton growth (Elser et al. 2001). Macrozooplanktonin L239 comprise various calanoid and cyclopoid cope-pods (including Diaptomus minutus, Diacyclops bicus-pidatus thomasi) and a mixed cladoceran assemblage(Bosmina longirostris, Holopedium gibberum, andDaphnia mendotae). L239 phytoplankton are diverse,with substantial contributions of, in declining order ofrelative dominance, chrysophytes, dinoflagellates,cyanophytes, diatoms, and cryptophytes (Elser et al.2001, D. Findlay unpubl. data). Little previous workhas examined the lake’s microbial components.

Design and execution of mesocosm experiment.Details of the experimental enclosures and overallexecution of the experiment are described by Urabe etal. (2002b). Briefly, enclosures consisted of 1 m dia-meter, 4 m deep, clear polyethylene tubes (closed onthe bottom, total volume 3.1 m3). At the time of theexperiment, the mixed-layer depth in L239 was 4.5 m.Enclosures were filled by pumping water from a depthof 2 m and passing it through 125 µm mesh Nitex. Livezooplankton from L239 were collected using horizontaltows of a 125 µm zooplankton net and combined withcollections of Daphnia dentifera from nearby Lake 979.This species was added to assure the presence of alarge-bodied daphnid in all enclosures. Subsamples ofthis collection were added to each enclosure to bringthe initial macrozooplankton biomass to a level similarto that found in L239. To manipulate irradiance wecovered the top and sides of 6 of 12 enclosures withneutral density screening (shaded treatment) and leftthe remaining enclosures uncovered (unshaded treat-ment). The screening lowered photosyntheticallyactive radiation (PAR) to 7% of ambient levels. Thisreduction in incident light is roughly equivalent to thatexperienced by a lake during a period of deeply over-cast skies. Each enclosure was also covered with a1.25 m square of OP3 Plexiglas to eliminate incidentultraviolet (UV) radiation and thus prevent confound-ing effects of lowering total irradiance with potentialeffects of also lowering UV. Direct measurements ofunderwater PAR using a Li-Cor 2π underwater sensoralong with a 2π deck cell indicated that intensities at1 m depth were 20 to 40% of surface radiation inunshaded enclosures but ranged from 3 to 6% inshaded enclosures. Effects of nutrients were examinedby creating a gradient of nutrient inputs in both shadedand unshaded enclosures. These nutrient gradientswere created by adding PO4 at 6 concentrations (0,0.024, 0.048, 0.097, 0.19, and 0.39 µM, with 1 enclosureper nutrient treatment within each irradiance treat-ment). This range of PO4 enrichment was chosen based

51


on previous assessment of seston C concentrations inL239 so that we could expect seston C:P ratios todecrease substantially to levels lower than those likelyto be limiting to daphnids (e.g. <250:1, Urabe & Watan-abe 1992). PO4 was added only once (as KH2PO4) at thebeginning of the experiment and produced a rapiddecrease in total seston C:P (Urabe et al. 2002b). PO4

fertilization treatments were not replicated in order toachieve a broad range of experimental PO4 levels; sta-tistical analyses rested on regression analyses (to testfor effects of added PO4) and pair-wise t-tests (to testfor effects of light), as described below and in Urabe etal. (2002b). Inorganic N was also added (as NH4NO3) toall enclosures at an atomic ratio of 30 to ensure that Premained the limiting nutrient in the enclosures. Theexperiment ran for 26 d. While somewhat long relativeto time scales of microbial response, this time span wasnecessary to permit potential changes in macrozoo-plankton to occur (Urabe et al. 2002b). Calculation oftotal P at the end of the experiment to determine ifadded P was present in water column pools (total dis-solved, seston, zooplankton) indicated that, whilethere were no differences between shaded andunshaded enclosures in the relative importance of‘missing’ P (i.e. P that is not accounted for in water col-umn P pools), the relative amount of missing Pincreased with P-fertilization level in both shaded andunshaded enclosures (from negligible in enclosuresreceiving less than 0.097 µM P to more than 2⁄3 in thehighest fertilization treatment). We cannot determine ifthis missing P was associated with accumulated mate-rials at the bottom of the enclosures, which we proba-bly under-sampled despite enclosure mixing, or withwall growth. However, observable wall growth wasminimal during the experiment and thus we suspectthat an increasing fraction of P was sedimented in thehigh P enclosures. Thus, in keeping with general rec-ommendations for interpreting mesocosm experimentsgiven potential enclosure effects (Schindler 1998),results for our high fertilization treatments should betreated with some caution given that the fate of addedP is relatively uncertain in these bags. However, theecologically relevant results that pertain most closelyto field conditions at the ELA are for the low-enrich-ment enclosures; in these enclosures, we were able toaccount for >85% of the added P.

Response variables. Abundance and stoichiometryof plankton constituents and dissolved nutrients:Enclosures were mixed every day by pulling a 0.3 mperforated PVC disk several times through the lengthof the enclosure. Monitoring for a variety of parametersoccurred at 5 d intervals. Methods of monitoring ofmacrozooplankton are described by Urabe et al.(2002b). Water samples for quantification of hetero-trophic nanoflagellates (HNF, as an index of overall

microzooplankton abundance), algae, bacteria, seston,and dissolved nutrient analysis were taken after eachenclosure was mixed. HNF samples were preservedwith 4% glutaraldehyde and kept refrigerated untilenumeration by epifluorescent microscopy after stain-ing with primulin (Caron 1983). Algal samples werefixed with Lugol’s solution for later microscopic enu-meration. Algal biomass was estimated by combiningcell counts with taxon-specific cell-volume estimatesfollowing Findlay & Kling (1998). Biovolume for algaltaxa with variable morphology or colony sizes wasestimated independently for each sample while singleconstant values were used for taxa that did not exhibitsubstantial variation in size and shape. Bacteria sam-ples were preserved with 2% filtered formaldehydeand kept refrigerated until enumeration by epifluor-escent microscopy using DAPI (4’6-diamidino-2-phenylindole) staining following Porter & Feig (1980).Thus our data pertain to responses of the entire micro-bial assemblage to experimental conditions. As previ-ous studies have shown that manipulations of nutrientsor grazing pressure can induce physiological and taxo-nomic shifts within the overall microbial assemblage(e.g. Jürgens et al. 1999, Larsen et al. 2001), this meansthat our results pertaining to microbial responses tolight and nutrient manipulations will be essentiallyconservative; some potentially important responsesmay therefore have been missed.

Water samples for seston analysis were size-fraction-ated using a 1 µm polycarbonate filter after which sus-pended particles in whole-water and <1 µm waterwere filtered onto pre-combusted GF/F glass-fiber fil-ters and dried for later analysis (Elser et al. 1995a). Ses-tonic C concentration (seston [C] hereafter, units: mg Cl–1) was determined using infrared analysis of CO2 inheadspace after filters were oxidized in sealed jarscontaining concentrated persulfate (Elser et al. 2001).Seston P was then analyzed colorimetrically (APHA1992) and C:P ratios of the 2 seston size fractions werecalculated and expressed as atomic ratios. Samples foranalysis of dissolved chemical constituents were fil-tered through a 0.2 µm polycarbonate filter and ana-lyzed immediately for ammonium-N (NH4

+, NH4-Nhereafter, Solorzano 1969) or kept frozen for lateranalysis of nitrate-N (NO3

–, NO3-N hereafter) and sol-uble reactive phosphorus (SRP) according to standard-ized methods (Stainton et al. 1977). In our analysis, thedetection limit for SRP was in the range of 0.065 µM.

Growth limitation of phytoplankton and bacteria:Resource limitation of phytoplankton (P vs light) andbacterial (P vs DOC) growth was assessed in eachenclosure at the end of the experiment using dilutionbioassays (Sterner 1994, Elser et al. 1995b). Dilutionreduces the encounter rate between algae and micro-zooplankton and thus permits a more direct measure of

52


actual algal or bacterial growth rate. In both bacterialand phytoplankton assays, mixed enclosure water wasfiltered through 0.2 µm cartridge filters and then wholeenclosure water was combined with filtered water in aproportion of 1:3 (75% dilution).

For phytoplankton assays, 2 sets of 4 clear PVCbottles (1 l) were then filled with diluted enclosurewater. For each set, 2 bottles were enriched with P(+2 µM as NaHPO4, P hereafter; a concentration of2 µM was chosen to assure that saturation of algaluptake and growth kinetics was achieved) and theother 2 left as unenriched controls (C). One set of 4bottles was incubated at 1 m depth in the enclosure (Ehereafter) from which the original sample was takenwhile the other set (L) was incubated in L239 at a depthof ~0.5 m. Algal growth was assessed by measuring, forboth initial diluted water and for water in bottles at theend of the incubation, seston [C] for materials capturedon GF/C grade glass-fiber filters, which probably passmost bacterial cells (Chrzanowski et al. 1995, Elser etal. 1995a). Thus, we were able to assess the degree ofalgal P-limitation under the light conditions of theenclosure by comparing growth in control and +Pbottles held in the enclosure (EC vs EP). We could alsoassess potential P-limitation under ample light by con-sidering the bottles incubated at 0.5 m in L239 (LC vsLP). Finally, the approach allowed us to assess algallight limitation by comparing growth during incuba-tion in the enclosure (where light might have beenlimiting, especially in shaded enclosures) to growthduring incubation at elevated light intensities (EC vsLC). We were also able to assess whether responses tolight were affected by P-enrichment. To assess how P-enrichment and light manipulation in bioassaysaffected algal C:P stoichiometry, we also measuredseston [P] as described above. This allowed us to esti-mate, for each enclosure, the C:P ratio of new algalbiomass produced during the bioassay incubationperiod in different P and light manipulation bioassaytreatments.

Bacterial growth response was measured using thesame 75% diluted enclosure water prepared asdescribed above. Diluted water was sub-sampled (induplicate) for later enumeration of initial bacterialdensity and then used to fill eight 500 ml bottles thatwere then subjected to a factorial enrichment of DOC(added as glucose and acetate in 1:1 molar proportionto achieve a final concentration of 250 µM C) and P(added as NaHPO4 to a concentration of 2 µM P). Tworeplicates of each treatment combination were pre-pared and there were 4 treatment combinations in all:‘controls’ that received no DOC or P, +P that receivedPO4 only, +DOC that received DOC only, and +DOC&Pthat received both DOC and P. Bottles were held atlake temperature by suspending them at 0.5 m in L239.

Duplicate 10 ml sub-samples were preserved formicroscopic enumeration as described above. Bacterialgrowth was assessed by changes in bacteria numbersdetermined by epifluorescent microscopy.

Growth rates (µ) of algae and bacteria during thebioassays were calculated assuming exponentialgrowth as:

µ (d–1) = ln(C2/C1)/dt

where C2 and C1 represent abundance (algal [C] orbacteria numbers) at the end and beginning of theincubation, respectively, and dt is the duration of theincubation. Based on prior experience in these lakes(Sterner et al. 1995, Chrzanowski et al. 1997), incuba-tions lasted 3 d for algae and 1 d for bacteria.

Data analysis. Analyses for monitored variables inthe enclosures involved data from the final 3 samplingdates (last 2 wk) of the experiment. Examination oftemporal data for major parameters such as seston [C]and macrozooplankton biomass indicated that tempo-ral dynamics during this period were relatively modest(Urabe et al. 2002b), suggesting that the enclosureshad reached some kind of quasi-equilibrium duringthe focal period considered. To indicate the extent oftemporal variation of various response parametersduring the final 2 wk of the experiment, we presenterror bars as ±1 SE of the mean values observed duringthose final sampling dates. However, only the meanvalues of the data from the final sampling dates, andnot the individual measurements in time, were used instatistical tests. Effects of enclosure P-enrichment wereevaluated by regressing the mean values of each vari-able over the final sampling dates versus enclosure P-enrichment level. Prior to regression, P-enrichmenttreatment was generally transformed as ln(P + 1),where P is the P-enrichment level for that enclosure.Effects of light were assessed by performing a singlepaired t-test for each parameter. This test evaluatedthe hypothesis that, across all P-enrichment levels, thedifference between unshaded and shaded enclosuresfor each particular P-enrichment level for that para-meter was zero.

Each bioassay experiment (algal response to lightand P, bacteria response to DOC and P) was analyzedby 2-way analysis of variance (ANOVA) to evaluatethe statistical significance of main effects and inter-actions for each enclosure’s experiment (degrees offreedom: 1 for main effects and interactions, 4 for theerror term). Effects of enclosure irradiance treatment(e.g. shaded vs unshaded) were evaluated by a pairedt-test of the particular enrichment bioassay responsefor each level of enclosure P-enrichment, as describedabove for routinely monitored variables. Here we con-sidered the mean value of bioassay response in eachshaded enclosure’s assay relative to the unshaded

53


enclosure at the same enrichment level and tested thenull hypothesis, across all enclosures, that the differ-ence between the values for shaded and correspond-ing unshaded enclosures was zero. Effects of enclosureP-enrichment treatment (e.g. 0 to 0.39 µM P) wereassessed by regressing bioassay responses vs enclo-sure P-enrichment treatment separately for eachenclosure irradiance treatment; again, the data weregenerally transformed [ln(P + 1)] prior to regression.

RESULTS

Dissolved nutrients

At the end of the experiment, concentrations of NO3-N and NH4-N increased significantly with increasinglevel of fertilizer addition in both unshaded and shadedenclosures (Fig. 1A,B). However, this increase wasmuted in unshaded enclosures, especially for NO3-N(for NO3-N, slopes of the regression lines for shadedand unshaded enclosures were significantly different, p< 0.05) and in general shaded enclosures had higherconcentrations of NH4-N and NO3-N than unshaded

enclosures (p < 0.04 for NH4-N, p < 0.09 for NO3-N, p <0.01 for total inorganic nitrogen, TIN = NH4-N + NO3-N; data not shown). In contrast to strong increases in in-organic nitrogen with enclosure fertilization level, ef-fects of increasing fertilizer input on SRP were modest(Fig. 1C) and statistically significant only in the shadedenclosures. Thus, phytoplankton and other microbeswere able to efficiently sequester added PO4 under un-shaded conditions, a result that is consistent with thestrongly P-limited conditions of the lakes of the ELA(Schindler et al. 1978, Sterner et al. 1995). Because en-richment effects were stronger for NO3-N and NH4-Nthan for SRP, the N:P ratio of dissolved inorganic nutri-ents (as indexed by TIN:SRP) was generally higher athigh enrichment levels (Fig. 1D; however, this tendencywas statistically significant only for the shaded enclo-sures. A paired t-test indicated that TIN:SRP ratioswere significantly higher in shaded relative to un-shaded enclosures. In total, these data indicate that dis-solved nutrient concentrations were affected not onlyby the experimental level of nutrient addition to the en-closures but also by irradiance, as decreased light lev-els allowed higher concentrations of nutrients to bemaintained in the dissolved pool.

54

Fig. 1. Effects of shading and nutrient enrichment on (A) NO3-N, (B) NH4-N, (C) soluble reactive phosphorus (SRP), and (D) totalinorganic nitrogen (TIN; NH4-N + NO3-N):SRP ratio in the enclosures. Error bars indicate ±1 SE of observations on the final 3sampling dates. Lines indicate regression lines fit to the data (shown when statistically significant) for shaded enclosures (----) orunshaded enclosures (___). In all figures, data for shaded enclosures were displaced upwards slightly on the x-axis so that con-fidence limits for the shaded and unshaded enclosures can be more easily discerned at a given nutrient enrichment level


Seston abundance and C:P stoichiometry

Changes in total seston abundance and C:P ratiowere reported by Urabe et al. (2002b); we summarizethose data here. Total seston C concentration ([C]hereafter) was generally unaffected by enclosure P-fertilization but, at each P-enrichment level, shadedenclosures had lower total seston [C]. Total seston C:Pdecreased strongly with increasing P enrichment inboth unshaded and shaded enclosures. Shading gener-ally lowered total seston C:P at all P-enrichment levelsbut the effect was larger at low P-enrichment levels(Urabe et al. 2002b). Seston in small particles (the ‘bac-teria’ fraction, <1 µm) was also affected by fertilizationand light (Fig. 2). As for total seston [C], P enrichmenthad no effect on <1 µm [C] (Fig. 2A; regression statis-tics non-significant, p > 0.10). However, unlike theresponse for total seston reported by Urabe et al.(2002b), shading had no effect on <1 µm seston [C](Fig. 2; paired t-test, p > 0.3). However, both enrich-ment and shading appeared to affect C:P stoichiometryin the <1 µm fraction (Fig. 2B). Similar to the significantdecline in total seston C:P reported by Urabe et al.(2002b), P-enrichment produced marginally significantdeclines in <1 µm seston C:P in both unshaded andshaded sets of enclosures (p = 0.051 for shaded enclo-sures, with P-enrichment level transformed as ln(P+ 1);p = 0.089 for unshaded enclosures, no transforma-tions). Also similar to the response of whole seston(Urabe et al. 2002b), <1 µm seston C:P was consistentlylower in shaded than in unshaded enclosures (Fig. 2B;paired t-test, p < 0.01). Thus, both P-enrichment andshading affected the P-content of bacteria-sizedparticles in the enclosures.

Balance of bacteria and phytoplankton

Average algal biomass determined by microscopicexamination over 3 of the final sampling dates (Days 200,210, and 215; samples for Day 205 were lost) generallyincreased with increasing P-fertilization (Fig. 3A) butthese trends were not statistically significant (p ≈ 0.20).However, algal biomass was generally lower in shadedrelative to unshaded enclosures at each P-enrichmentlevel (paired t-test, p < 0.04). Thus, algal biomass re-sponded in a similar manner qualitatively to overallpatterns of response of total seston abundance reportedby Urabe et al. (2002b). Neither P-enrichment nor shad-ing affected bacterial abundance in the enclosures (p >0.70; Fig. 3B). Visual examination indicated that therewere no systematic differences in bacterial cell size as afunction of nutrient enrichment or irradiance and so bac-terial numbers reasonably approximate bacterial bio-mass in the enclosures. The balance of bacteria relative

to algae was assessed by calculating an index of algaldominance as the ratio of algal biomass to bacteria abun-dance (A:B ratio) for each sampling date. Mean values ofA:B for 3 of the final sampling dates (Days 200, 210 and215) were then analyzed with respect to effects of P-fer-tilization and irradiance manipulations. Contrary to ex-pectations based on the LNH, there was no apparent ef-fect of P-enrichment on the relative abundance of algaeversus bacteria (Fig. 3C). However, despite relativelyhigh variance in A:B during the final sampling dates,shading produced a consistent reduction in A:B (p < 0.03in paired t-test; Fig. 3C), primarily due to the relativelylarge decreases in algal biomass in shaded enclosures.This result is inconsistent with expectations based on theLNH (Sterner et al. 1997), which predicts that increasedlight under nutrient limitation should enhance bacteriaat the expense of algae due to increased supply of labileDOC from nutrient-limited algae, allowing bacteria tooutcompete bacteria for inorganic P. However, the factthat differences in algal:bacteria ratio were due entirelyto changes in algal biomass and that there were notreatment effects on bacterial numbers suggests that thisoutcome should be considered with some caution.

55

Fig. 2. Effects of experimental manipulations on (A) theabundance (µM C) and (B) C:P ratio of ‘bacteria-sized’ seston(passing through a 1.0 µm membrane filter but captured on

a GF/F glass-fiber filter). Confidence limits as in Fig. 1


Balance of micro- and macrozooplankton

Data for total macrozooplankton biomass in shadedand unshaded enclosures were analyzed and pre-sented by Urabe et al. (2002b). In brief, total zooplank-ton biomass increased with P-fertilization in bothunshaded and shaded enclosures. However, macro-zooplankton biomass was significantly depressed inunshaded enclosures relative to shaded enclosures atlow P-enrichment due to poor food quality. In contrast,at high P-enrichment zooplankton in unshaded enclo-sures exceeded zooplankton in shaded enclosures,likely due to lower total algal abundance and produc-

tion under low light. As in our other analyses, data forHNF abundances for the final 3 sampling dates (Days205, 210 and 215) were used. Visual examination indi-cated that there were no systematic differences in HNFcell size as a function of nutrient enrichment or irradi-ance and so HNF densities reasonably approximateHNF biomass in the enclosures. Unlike the observedincreases in macrozooplankton biomass with P-fertil-ization, the abundance of HNF declined significantlywith P-fertilization (Fig. 4A) in both shaded (p < 0.004,R2 = 0.91) and unshaded (p < 0.04, R2 = 0.72) enclosures(P-enrichment was ln + 1 transformed prior to regres-sion to stabilize the variance around the regression).There was no consistent effect of light treatment onHNF abundance (Fig. 4A); indeed, the relationshipsbetween HNF abundance and P-enrichment level inunshaded and shaded treatments had nearly identicalslopes (–89 vs –84, respectively) and intercepts (303 vs305). To evaluate potential shifts in the balance of

56

Fig. 3. Effects of experimental manipulations on (A) algalbiomass (as determined by direct microscopic enumeration),(B) bacterial numbers (as determined by direct epifluorescentenumeration), and (C) the relative abundance of algae and

bacteria. Confidence limits as in Fig. 1

Fig. 4. Effects of experimental manipulations on (A) the abun-dance of heterotrophic nanoflagellates (HNF) and (B) theratio of HNF abundance and macrozooplankton biomass.Data for macrozooplankton came from Urabe et al. (2002b).

Confidence limits as in Fig. 1


microbial and grazing food chains in the enclosures asa function of light and enrichment treatments, wecalculated the ratio of HNF abundance to macro-zooplankton biomass for each sampling date and ana-lyzed the mean value of that parameter over 2 of thefinal 3 sampling dates (Days 205 and 210; data formacrozooplankton were not available for Day 215).Reflecting the simultaneous decrease in HNF abun-dance (Fig. 4A) and increase in macrozooplankton bio-mass (Urabe et al. 2002b) with P-fertilization, theHNF:macrozooplankton ratio decreased significantlywith P-enrichment, especially in the unshaded enclo-sures (Fig. 4B; p < 0.05, R2 > 0.76 for regressions withuntransformed and natural-log transformed data forHNF:macrozooplankton ratio). The effect of light onHNF:macrozooplankton ratio appeared to depend onP-fertilization treatment, as this ratio was higher inunshaded enclosures at low P-enrichment levels butsomewhat lower in unshaded enclosures at high P-enrichment levels (this is somewhat difficult to discernon the linear y-axis in Fig. 4B). This shift in theHNF:macrozooplankton ratio with light primarilyreflected the changing impact of light on macrozoo-plankton biomass. The effect of P-fertilization on abun-dance of microflagellates was independent of irradi-ance (Fig. 4A) but, as reported by Urabe et al. (2002b),reducing irradiance led to strong increases in macro-zooplankton abundance at low P-enrichment (due tofood quality effects) but modest decreases at high P-enrichment. Thus, shaded enclosures had lowerHNF:macrozooplankton ratios than unshaded enclo-sures at low P-enrichment but higher ratios at high P-enrichment.

Bioassay responses of algae

Growth bioassays indicated that algal resource limi-tation was affected by the light and P manipulationsimposed on the enclosures. The nature of algalresponse and the statistical significance of light and Ptreatments and light × P interactions on algal growthare summarized in Table 1. The results reveal somesurprises. In the unshaded enclosures, adding P inbioassays significantly stimulated algal growth in only1 enclosure, the highest P-enrichment treatment(+0.30 µM P enclosure). However, manipulation oflight in the bioassay experiment had a positive effect in3 unshaded enclosures (significance was marginal in 2cases) across the range of enclosure P-enrichment andwas inhibitory to algal growth in 1 case (+0.024 µM Penclosure). Thus, light paradoxically appeared to havea stronger stimulatory effect on algal growth than P inthe unshaded enclosures. Responses to light in theshaded enclosures were less surprising, as bioassays

revealed significant increases in growth due to in-creased light in all 6 experiments performed (Table 1).However, responses to P-enrichment were frequent inthe shaded enclosures despite the apparent existenceof light limitation: 5 of 6 experiments in shaded enclo-sures resulted in algal growth response to P-enrich-ment (2 of these were marginally significant, 0.05 < p <0.10). Thus, while light clearly had a dominant role asa limiting factor for algae in the shaded enclosures, Pappeared to be more important as a limiting factor inthe shaded than in the fully illuminated enclosures, asurprising result.

To evaluate trends in relative growth limitation ofalgae we calculated an index of the degree of algalgrowth limitation by light or by P in each enclosure.Following Gurung et al. (1999), an index of P limitation(LP) was calculated as the average of µLP – µLC and µEP –µEC while an index of light limitation (LL) was calcu-lated as the average of µLC – µEC and µLP – µEP. Thisneglects possible influences of light × P interactions inthe bioassays but makes the trends easier to analyzegraphically. In general, growth responses in the bioas-says were dominated by a single treatment main effectand the few interaction effects were relatively modest.Thus, this analysis captures how algal growth limita-tion shifted with enclosure P-enrichment and shadingtreatments. The plots indicate that, as expected, LL wasgreater for shaded versus unshaded enclosures(Fig. 5), a result supported by a paired t-test comparingLL for unshaded and shaded enclosures at particular P-enrichment levels (p < 0.01). In addition, algal lightlimitation in shaded enclosures increased with enclo-sure P-enrichment (Fig. 5; p < 0.001, R2 = 0.96). How-ever, there was no tendency for LL to change withincreasing enclosure fertilization level in unshadedenclosures (p > 0.85). Reflecting the surprising resultsof statistical analyses (Table 1), LP was significantlyhigher in shaded than in unshaded enclosures (Fig. 5B;p < 0.03 in paired t-test) while LP increased (p = 0.05)with increasing enclosure P fertilization only in un-shaded enclosures (Fig. 5B). Finally, as expected, thestrength of light limitation relative to P limitation ofalgal growth (LL – LP) was generally higher in shadedenclosures (compare the relative magnitudes of LL andLP for shaded and unshaded enclosures in Fig. 5), espe-cially for P-enrichments greater than 0.048 µM P. How-ever, this difference was only marginally significant(p < 0.09 in a paired t-test). The relative importance oflight limitation (again, as indicated by the difference LL

– LP) increased with enclosure P-fertilization in shadedenclosures (p < 0.0001, R2 = 0.97) but was not affectedby P-fertilization in the unshaded enclosures (p > 0.45).

According to the LNH and prior work on interactionsbetween light and nutrient limitation (Smith 1983,Healey 1985, Goldman 1986), algal C:nutrient ratio

57


should vary negatively with the degree of algal lightlimitation. To examine this, we plotted initial sestonC:P in each bioassay versus LL assessed for that bioas-say, expecting a negative correlation. As anticipated,seston C:P in the enclosures was negatively associatedwith the degree of apparent light limitation of phyto-plankton growth (Fig. 6; p < 0.003, R2 = 0.67, after nat-ural log-transformation of LL). Above an LL value of~0.1, seston C:P reached values approximating theRedfield ratio. In addition, the relationship appeared tobe non-linear, with steep increases in seston C:P at lowlevels of light limitation (LL < 0.1). Thus, it appears thatconsiderable alleviation of light limitation is requiredbefore seston C:P increases substantially. For shadedenclosures, increasing enclosure P-fertilization wasassociated with a shift from left to right in this figurebut there was no indication in unshaded enclosuresthat nutrient enrichment per se influenced the positionof different enclosures on this plot.

Because we also measured seston C and P in eachbioassay bottle at the end of the incubations, we werealso able to estimate the C:P of new seston produced(C:Pnew) during the bioassay. Bioassay P enrichment in-evitably lowered C:Pnew (Table 1) to values of ~30 to 75;this response is not surprising given the high per capitaP-supply in the P-enriched dilution bottles. In contrast,raising light (LC vs EC, LP vs EP) generally led to in-creased C:Pnew (Table 1). This effect was significant in 5

of 6 experiments for shaded enclosures but only for 1 of5 experiments involving unshaded enclosures. To con-sider the effect of enclosure irradiance and bioassay ir-radiance on the stoichiometry of algal production, wecalculated mean C:Pnew for each unenriched bioassaytreatment combination for each enclosure (that is, themean values of C:Pnew for LC and EC treatments). Wepooled all enclosure data for various enclosure enrich-ment levels for each irradiance treatment and per-formed a 2-way ANOVA (C:Pnew was natural log-trans-formed to stabilize the variance) to evaluate the effectsof enclosure irradiance (shaded vs unshaded) andbioassay irradiance (in enclosure vs in lake) manipula-tions on C:Pnew. Data for C:Pnew for bottles incubated in-side the enclosure clearly show that algae in fully illu-minated enclosures were forming new biomass at ahigher C:P ratio relative to algae growing in shaded en-closures (Fig. 7). Not surprisingly, C:Pnew in unshadedenclosures did not respond very strongly to the modestincrease in irradiance when enclosure plankton wereincubated at 0.5 m depth in L239. However, C:Pnew forshaded enclosures increased dramatically when enclo-sure plankton were incubated in L239 (Fig. 7), reachingvalues even higher than those seen for unshaded enclo-sures. Thus, in the 2-way ANOVA for C:Pnew, enclosuretreatment (unshaded vs shaded) and enclosure-bio-assay treatment interactions were statistically signifi-cant (p < 0.05). Examination of the data for individual

58

P Light Growth Final seston C:P New seston C:Ptreatment treatment L P L × P L P L × P L P L × P

+0 µM Unshaded (+) ns ns ns – ns (+) – ns+0.024 µM Unshaded – ns ns ns – A (–) – ns+0.048 µM Unshaded (replicate lost; no statistics possible)+0.097 µM Unshaded ns ns ns ns – ns ns – ns+0.19 µM Unshaded + ns ns ns – ns + – ns+0.39 µM Unshaded (+) + B + – ns ns – C+0 µM Shadeded + + (D) + – ns + – ns+0.024 µM Shadeded + + E + – ns + – ns+0.048 µM Shadeded + + ns + – (F) + – ns+0.097 µM Shadeded + ns ns ns – ns + – ns+0.19 µM Shadeded + (+) G + – (H) + – ns+0.39 µM Shadeded + + (I) + – ns + – ns

Table 1. Statistical analyses of algal growth responses and changes in overall seston C:P ratio and the C:P ratio of new sestonproduction in enrichment bioassays performed at the end of the enclosure experiment. Responses to bioassay irradiancemanipulation are indicated under L (incubation in the enclosure vs incubation at 0.5 m in the lake itself) while responses to bio-assay P-enrichment (unamended control or enriched with 2 µM PO4) are indicated under P. Significant treatment effects (p < 0.05)are indicated by ‘+’ when the response parameter increased in response to the manipulation of that factor or by ‘–’ when the re-sponse parameter decreased in response to manipulation of that factor. Marginally significant responses and interaction terms(0.05 ≤ p < 0.10) are indicated by parentheses. ns = non-significant (p > 0.10). Occurrence of significant (p < 0.05) light × P-enrichment interaction terms are indicated under L × P. Other significant interaction terms: A, Light lowered seston C:P in con-trols but increased C:P in +P; effect was modest. B, Effect of P was slightly stronger when incubated in lake. C, New C:P de-creased slightly with light in controls but increased in +P bottles. D, Effect of P was (marginally) stronger when bottles wereincubated in the lake. E, Effect of P was stronger when bottles were incubated in the lake. F, Effect of light was (marginally)stronger in controls relative to +P bottles. G, Effect of P was negative in enclosure but positive in the lake. H, Effect of light was(marginally) larger in controls than in +P bottles. I: Effect of P was (marginally) stronger when bottles were incubated in the lake


enclosures suggested that the responsiveness of C:Pnew

to incubation location (enclosure vs L239) for shadedenclosures increased with enclosure P-enrichment (p =0.09, R2 = 0.55). Thus, it appeared that algae growing inenclosures with the greatest likelihood of light limita-tion (shaded, high P-enrichment) developed a carbonfixation overcapacity that resulted in stoichiometricallyunbalanced production of new biomass when irradi-ance was increased. In summary, these responsesclearly show that the C:P stoichiometry of production ofnew biomass in L239 is sensitive to both P supply andirradiance experienced by the phytoplankton.

59

Fig. 5. Shifts in algal growth responses to (A) light enrichment(LL; incubation at 0.5 m depth in Lake 239 relative to incuba-tion at 1 m depth in their respective enclosure) and (B) to Penrichment (LP; +2 µM PO4) at the end of the enclosure exper-iment. LL calculated as average difference in algal growthwhen bottles incubated at relatively high light in lake vs inenclosure; average involved responses in unenriched bottles(µLC – µEC) and responses in P-enriched bottles(µLP – µEP). LP for each enclosure was calculated as averagedifference in algal growth for P-enriched vs unenrichedbottles; average involved data from bottles incubated atrelatively high light in lake (µLP – µLC) and from bottles incu-bated in enclosure (µEP – µEC). Results of statistical tests ofalgal responses to light or P-enrichment are given in Table 1.(----) Results of regression analysis of LL for shaded enclo-sures and LP for unshaded enclosures, the only relationshipsthat were statistically significant (enrichment treatment was

ln + 1 transformed to normalize the variance)

Fig. 6. Correlation between initial seston C:P ratio in bioassaybottles and the degree of light limitation (LL) exhibited duringthe bioassay. LL calculated as average increase in growth rateof algal seston (caught on GF/C filters) in elevated vs ambientirradiance treatments. P-fertilization treatment for each datapoint is shown. (----) Results of regression analysis on ln-transformed data for initial seston C:P (p < 0.003, R2 = 0.61)

Fig. 7. C:P ratio of ‘new’ seston produced in bioassay bottlesincubated under different conditions of illumination usingsample water from unshaded and shaded enclosures. Data forall P-enrichment treatments were pooled (there was no statis-tically significant effect of P enrichment on C:Pnew in eitherirradiance treatment). Error bars indicate ±1 SE. Two-wayANOVA indicated significant effect of bioassay incubationtreatment (incubation in enclosure vs in lake), no effect ofenclosure treatment (unshaded vs shaded), but significantbioassay incubation × enclosure treatment interaction. Letterson each bar reflect results of Scheffe’s multiple comparisontest following the ANOVA; different letters indicate means

significantly different from each other (p < 0.05)


Bioassay responses of bacteria

Bacteria in the enclosures responded to addition ofboth DOC and PO4 in various enclosures but therewere few significant DOC × PO4 interactions in theexperiments (Table 2). Following the approach foralgal bioassay response, for each bioassay an index ofbacterial DOC limitation (LDOC) was calculated as theaverage of µDOC – µC and µDOC+P – µP while a similarindex of bacterial P-limitation (LP) was calculated asthe average of µP – µC and µDOC+P – µDOC. In shadedenclosures, addition of DOC had a significant or mar-ginally significant effect on bacterial growth rate in 4of the 6 bioassays but DOC effects were observed inonly 1 of the 5 bioassays in the fully illuminated enclo-sures (Table 2). However, a paired t-test comparingLDOC for shaded and unshaded enclosures was not sta-tistically significant (p > 0.85), largely due to high vari-ation in LDOC among the unshaded enclosures. Statisti-cally significant stimulation of bacterial growth byadded P occurred in 4 of the 6 experiments in theshaded enclosures (Table 2) but in only 2 of the 6experiments in the unshaded enclosures (there wasalso a significant decrease in growth to P-enrichedbottles in the +0.30 µM P enclosure). There was no

clear tendency of bacterial P-limitation to change as afunction of P-fertilization level (regression analyses ofLP for bacteria against P-fertilization were non-signifi-cant, p > 0.20) nor was there any significant effect ofirradiance on bacterial P response (paired t-test, p >0.65). Visual examination indicated no consistenteffect of bioassay treatments on bacterial cell size;thus, treatment effects on growth rates determined bycell counts probably reflected changes in overallbacterial production.

DISCUSSION

The main results of our study are summarized in adiagram that semi-quantitatively depicts the status ofmajor planktonic pools and regulatory interactionsunder contrasting conditions of light and P inputimposed in our enclosures (Fig. 8). Here we summarizethe 4 scenarios illustrated there:

(1) Under high light and low P supply (i.e. unshadedenclosures receiving no experimental nutrient input),a pelagic community with high C:P ratio in overallseston as well as in bacteria-sized particles occurred(Fig. 8A). Algal biomass was high relative to bacterialbiomass and neither algae nor bacteria responded sig-nificantly to bioassay enrichments of nutrients orenergy (light, DOC). Macrozooplankton biomass waslow and representation of high P taxa such as Daphniawas diminished (Urabe et al. 2002b); therefore, C:Pratio of zooplankton biomass was high. Macrozoo-plankton growth was constrained by stoichiometricfood quality with low growth efficiency in terms of C(Urabe et al. 2002b). Hence, macrozooplankton graz-ing was weak, microzooplankton (as indexed by HNFin our study) probably experienced modest grazinglosses, and thus had relatively high biomass despitepotentially poorer quality (high C:P) of bacterial prey.As a result, a dominant trophic pathway in this systemprobably involved DOC transfer from algae to bacteriaand subsequent consumption of bacteria by microzoo-plankton with strong dissipation of excess C by micro-zooplankton due to stoichiometric imbalance with theirbacteria prey.

(2) Under low P but low light (Fig. 8B), shifts in bothstoichiometric aspects and biomass pools wereobserved. Algal biomass was decreased somewhat butwas also more P-rich. Signs of algal light limitationwere observed, as expected. However, somewhatparadoxically, growth rates of both algae and bacteriaresponded to P-enrichment, perhaps reflecting adecrease of overall availability of P to phytoplanktonand bacteria due to the large increase in the biomassand P-content of the macrozooplankton community(Urabe et al. 2002b). Stoichiometric limitation of

60

P Light Growthtreatment treatment DOC P DOC × P

+0 µM Unshaded ns ns A+0.024 µM Unshaded ns + ns+0.048 µM Unshaded ns ns ns+0.097 µM Unshaded (+) ns ns+0.19 µM Unshaded ns + ns+0.39 µM Unshaded + (-) (B)+0 µM Shadeded ns + ns+0.024 µM Shadeded (+) + ns+0.048 µM Shadeded ns ns ns+0.097 µM Shadeded + + ns+0.19 µM Shadeded (+) ns ns+0.39 µM Shadeded + + ns

Table 2. Statistical analyses of bacterial growth responses inenrichment bioassays performed at the end of the enclosureexperiment. Responses to DOC addition (unamended controlor enriched with 250 µM C as glucose and acetate) are indi-cated under DOC, while responses to bioassay P-enrichment(unamended control or enriched with 2 µM PO4) are indicatedunder P. Significant treatment effects (p < 0.05) are indicatedby ‘+’ when the response parameter increased in response tothe manipulation of that factor or by ‘–’ when the responseparameter decreased in response to manipulation of that fac-tor. Marginally significant responses and interaction terms(0.05 ≤ p < 0.10) are indicated by parentheses. ns = non-sig-nificant (p > 0.10). Occurrence of significant (p < 0.05) DOC ×P-enrichment interaction terms are indicated under DOC × P.Other significant interaction terms: A, DOC stimulatory withP but inhibitory without P. B, P inhibitory without DOC but

stimulatory with DOC


macrozooplankton production was relieved by shad-ing, resulting in greater zooplankton production andgrowth efficiency in terms of C (Urabe et al. 2002b).Despite higher macrozooplankton biomass that proba-bly imposed high loss rates, microzooplankton biomasswas unchanged (relative to unshaded enclosures), per-haps due to more efficient growth in response toincreased P-content of bacterial prey. Important path-ways of trophic flow involved both direct links and

indirect links (i.e. via bacteria and microzooplankton)to macrozooplankton.

(3) At high rates of P-input and ambient light(Fig. 8C), strong shifts in both biomass and stoichio-metry were also observed relative to ambient light andnutrient conditions. Inorganic P levels, as indicated bySRP, were higher than at low P-input. The absoluteand relative abundances of algae and bacteria werelargely unchanged compared to unenriched conditions

61

Fig. 8. Changes in major componentsof the planktonic food web under con-trasting conditions of light and nutri-ent balance. Scenarios summarizedata from this study and for whole ses-ton and macrozooplankton fromUrabe et al. (2002b). (A–D) Sizes ofboxes indicate approximate magni-tude of biomass of particular func-tional groups; absolute sizes of boxesfor different functional groups (e.g.macrozooplankton vs macrozooplank-ton under high light, low P) are notmeant to be compared directly, as wedo not have direct measurements ofthe absolute biomass levels of all ofthe functional groups depicted. Darkshading within each box indicates rel-ative P-content of that pool. A greybox is used to indicate microzoo-plankton because we do not havedata on their P-content. Indications oflikely relative magnitude of particularfluxes and of efficiency of trophictransfers (in terms of gross growthefficiency, or the fraction of ingestedC that is transformed to new biomassproduction) are included. Also, indi-cation is given of primary limitingfactors for algae (light, phosphorus)and bacteria (DOC, phosphorus). Z =macrozooplankton; M = microzoo-plankton; A = algae; B = bacteria;

P = PO4


but seston C:P ratios, both overall and in bacteria-sizedparticles, were significantly decreased. While algalgrowth surprisingly responded only to P-enrichmentdespite elevated P-input levels, bacterial growth rateswere not DOC-limited, as expected. Again, perhapsthe increased sink of P in massively larger biomass ofP-rich macrozooplankton (Urabe et al. 2002b) wasresponsible for continued P-limited algal growth ratedespite increased P-inputs. Large increases in macro-zooplankton due to relief of stoichiometric constraintsand increased growth efficiency were accompanied bystrongly decreased abundance of microzooplankton.Thus, in this condition the dominant trophic links pri-marily involved direct macrozooplankton consumptionof algal and bacterial production.

(4) To a large degree these responses to increased Pinput were even more pronounced at low light input(Fig. 8D). Here, algal biomass was decreased due toshading but seston C:P ratios were quite low. Algaeexperienced strong light limitation (along with aweaker response to P-enrichment) while bacteriaresponded to both DOC and P-amendment. Zooplank-ton biomass was somewhat lower than in the fertilized,unshaded enclosures, perhaps due to lower overallfood abundance. Microzooplankton abundance wasalso low relative to low enrichment enclosures. Thus,dominant trophic pathways probably involved directlinks to macrozooplankton under these conditions.

Our study represents the first large-scale field exper-iment to test microbial aspects of the LNH of Sterner etal. (1997) and has produced the following primary find-ings consistent with these aspects:

(1) The relative abundance of macrozooplanktonover microzooplankton was significantly affected byboth irradiance and nutrient supply. In particular, fer-tilization led to increases in macrozooplankton abun-dance (Urabe et al. 2002b), decreases in HNF abun-dance (Fig. 4A), and thus to a major increase in therelative abundance of macrozooplankton versus HNFas P-enrichment increased (Fig. 4B). It appears thatrelieving stoichiometric constraints on macrozooplank-ton production in L239 induced a massive shift in thepathways of organic matter processing between thegrazing chain and the microbial food web, implyingthat in L239 and similar unproductive lakes,light:nutrient balance affects microzooplankton pri-marily through ‘top down’ effects. P-enrichment didnot lead to increased abundances of heterotrophicnanoflagellates in the enclosures. On the contrary,abundance of these microzooplankton declined dra-matically in high P enclosures, in concert with largeincreases in macrozooplankton induced by theimproved stoichiometric food quality due to P-fertiliza-tion (Urabe et al. 2002b). This suggests that microzoo-plankton abundance in L239 is strongly determined by

the proliferation of macrozooplankton and that theeffects of light:nutrient balance on microzooplanktonare indirect and occur mainly via the impacts oflight:nutrient balance on food quality for larger graz-ers. However, there is some suggestion that HNF suc-cess was also influenced by stoichiometric food quality,as has been suggested in laboratory studies (Nakano1994, Rothaupt 1996). That is, in low P (+0 µM P,+0.024 µM P) enclosures, HNF abundances in shadedand unshaded enclosures were nearly identical(Fig. 4A), despite the fact that macrozooplankton bio-mass was 2 to 4 times higher in shaded enclosures(Urabe et al. 2002b) and that HNF in those enclosuresprobably experienced higher loss rates. Thus, HNFgrowth rates were probably higher in shaded enclo-sures to compensate for those losses, perhaps inresponse to a more suitable food quality (lower C:P inoverall seston and in bacteria-sized <1 µm particles)induced by shading. Unfortunately, we cannot assessthis possibility in the absence of direct assessments ofHNF growth rates.

(2) Both irradiance and P-fertilization generallyaffected algal resource limitation (Fig. 5) and the C:P ofnew algal biomass formed (Fig. 7) in expected ways(but see below), in that decreased light resulted inincreased prevalence of light limitation and lowerC:Pnew while P-enrichment increased the magnitude oflight limitation (in shaded enclosures) and loweredC:Pnew. Enclosure shading interacted with P-fertiliza-tion to have a large effect on the responsiveness ofalgal stoichiometry to light; algae from shaded enclo-sures under high P-enrichment greatly increased theirC:P when incubated in full sunlight (Fig. 7). Finally,one component of the LNH of Sterner et al. (1997)relies on an argument that decreased irradiancedecreases the overall severity of algal growth limita-tion, and thus C:nutrient ratio, by reducing the physio-logical demand for nutrients established by algalgrowth capacity (µm). Thus, we expected that algae inlow light enclosures would exhibit low µm whenreleased from potential nutrient limitation. To assessthis we used growth rate in the P-enriched bottlesincubated in the enclosures (i.e. µEP) to estimate µm.Indeed, values of µEP were nearly twice as high in theunshaded enclosures as in shaded enclosures (0.13 vs0.07 d–1, data not shown; p < 0.04 in a pairwise t-test).Enclosure P-enrichment level had no effect on µm (p >0.50 in regression of µm vs P-enrichment treatment).

(3) Irradiance and P-fertilization affected bacteriamostly by altering their physiological state (as indexedby the relative frequency of bioassay responses and by<1 µm seston C:P in shaded and unshaded enclosures),not by affecting their population size or standing bio-mass. More specifically, shading appeared to amelio-rate bacterial P limitation, as indicated by lower <1 µm

62


seston C:P in the shade and by the increased fre-quency of DOC responses (relative to responses to P)in bioassays in shaded enclosures (Table 2). In addi-tion, P-enrichment lowered <1 µm seston C:P in bothunshaded and shaded enclosures. The impact of lighton bacterial C:P is particularly interesting, as it empha-sizes a close linking between the nutritional states ofalgae and bacteria in the enclosures. However, manip-ulations of light and P supply had relatively modestimpact on the magnitudes and relative importance ofDOC and P limitation of bacterial growth. The relativeconstancy of bacterial abundance despite strongmanipulations of light and nutrients may reflect thenature of our assessment of the microbial community,which involved total direct counts and did not evaluateany physiological or phylogenetic shifts within theassemblage. Previous studies using direct epifluores-cent counts also report only relatively modest re-sponses of total bacterial abundance to manipulationsof nutrients because grazing on bacteria can maskchanges in growth (Pace & Cole 1996); studies usingmore sophisticated techniques capable of assessingdifferent strains or taxa of microbes in the communitywould give a more sensitive assessment of the micro-bial response to alteration in light:nutrient balance.However, such effects are unlikely to have influencedour data related to microbial limiting factors, as ourgrowth bioassays involved dilutions which reducedgrazing on bacteria by 75% and thus we were proba-bly able to observe effects on growth rate more directlyfor at least the dominant members making up thebacterial assemblage.

Nevertheless, some of our observations are at oddswith predictions derived from the LNH. Specifically,decreasing light or increasing nutrient supply (or thecombination of the two) did not decrease the domi-nance of bacteria relative to algae as expected (Sterneret al. 1997). On the contrary, lowering irradianceresulted in reductions in algal biomass but no changein bacterial abundance; thus, algae had greater rela-tive abundance in the fully illuminated enclosures,irrespective of P enrichment level. This discrepancy isparticularly interesting, as our bioassay and C:P datasuggest that algal and bacterial physiological condi-tion were affected in ways generally consistent withthe LNH. Bacteria seemed to be more consistentlylimited by DOC in shaded enclosures while algae weremore strongly limited by light in shaded enclosuresand under high P-enrichment. The discrepancy poten-tially arises because predictions from the LNH aboutrelative algal and bacterial dominance consider onlythe qualitative direction of the outcome of competitiverelations between the 2 functional groups and do notconsider any other indirect effects on trophic pathwaysthat might be induced simultaneously by shifts in

light:nutrient balance. However, our light and nutrientmanipulations produced massive changes in theabsolute and relative abundances of macro- andmicrozooplankton (Urabe et al. 2002b, Fig. 4) and thusthere were also likely to have been major changes ingrazing losses experienced by algae and bacteria.Unfortunately, we do not have data to evaluate theserelative impacts. The original formulation of the LNHdid not incorporate such indirect effects mediated bychanges in loss processes to algae and bacteria; ourdata suggest that a more complicated formulationincluding these ‘top-down’ effects is probably neces-sary to reflect how the relative abundances of algaeand bacteria shift with light:nutrient balance.

Also at odds with the LNH are some aspects of algalgrowth responses to P-enrichment in the bioassayexperiments. First, lowering irradiance did not amelio-rate algal responsiveness to P-enrichment as expected,even as the bioassay response to light and overall ses-ton C:P declined in expected ways as discussed above.On the contrary, algal response to P addition in bio-assays was consistently stronger in shaded enclosuresthan in fully illuminated enclosures at each enclosureP-fertilization level (Fig. 5). This apparent increase inthe overall scarcity of P under strongly diminished irra-diance is somewhat paradoxical, especially given thecoincident responses in seston C:P stoichiometry andlight limitation. One explanation for this outcome isthat increased biomass of P-rich macrozooplanktonunder low light and high P-inputs (Urabe et al. 2002b)increased the sink strength for P in this pool (Andersen1997), thus removing circulating P from the planktoncommunity. Alternatively, increased algal P-limitationunder low light suggests a strong interaction betweenirradiance and P demands in which lower light raisesthe P-requirement of the algae. However, there is littleevidence in the existing literature on algal physiologi-cal ecology that would lead one to expect such aresponse.

In summary, the bulk of our data support variousaspects of the hypothesis of Sterner et al. (1997), thatvariation among lakes in seston C:P stoichiometry andthus in pelagic food web structure is driven by changesin light:nutrient balance. Manipulations of light and Psupply altered algal and bacterial physiological stateand C:P stoichiometry; these changes had major con-sequences for the planktonic food web in the form ofreciprocal impacts on macrozooplankton (Urabe et al.2002b) and microzooplankton. However, some aspectsof the LNH were not supported, such as shifts in therelative dominance of algae versus bacteria biomass.Factors responsible for the considerable inertia of thisbalance remain enigmatic. Nevertheless, documenta-tion of effects of light:nutrient balance on food webdynamics has now been accomplished in experimental

63


systems ranging in scale from laboratory flasks (Urabe& Sterner 1996), indoor microcosms (Sterner et al.1998, Urabe et al. 2002a), and field enclosures withnatural plankton communities (Urabe et al. 2002b, thisstudy). It remains to be seen whether the applicabilityof these mechanisms continues to hold at increasingscales, as one extrapolates to whole ecosystems wherea myriad of factors vary both spatially and temporally.However, field studies involving multi-lake compara-tive analysis (Sterner et al. 1997) and temporal dynam-ics at fixed stations (Karl 1999) suggest that hydrody-namic alterations affecting the relative supplies of lightand nutrient supply impinge at least on seston C:N:Pstoichiometry. Ramifications for the food web at theselarger scales remain generally unexplored.

Acknowledgements. We thank P. Frost, J. Clasen, A.Waggener, C. Yoshimizu and the staff of the ExperimentalLakes Area, Freshwater Institute, Canada, for their technicaland logistical support. J. Cotner and 2 anonymous reviewersprovided invaluable comments on the manuscript. This workwas performed as a Japan-USA joint science study and sup-ported by grant-in-aid from JSPS and IGBP-MESSC-JPN toJ.U. and NSF grant DEB-9725867 to J.J.E.

LITERATURE CITED

Andersen T (1997) Pelagic nutrient cycles: herbivores assources and sinks. Springer-Verlag, Berlin

APHA (American Public Health Association) (1992) Standardmethods for the examination of water and wastewater,18th edn. APHA, Washington, DC

Azam F, Fenchel T, Field JG, Grey JS, Meyer-Reil LA,Thingstad F (1983) The ecological role of water-columnmicrobes in the sea. Mar Ecol Prog Ser 10:257–263

Berman-Frank I, Dubinsky Z (1999) Balanced growth inaquatic plants: myth or reality? BioScience 49:29–37

Bratbak G, Thingstad TF (1985) Phytoplankton-bacteria inter-actions: an apparent paradox? Analysis of a model systemwith both competition and commensalism. Mar Ecol ProgSer 25:23–30

Carlsson P, Caron DA (2001) Seasonal variation of phospho-rus limitation of bacterial growth in a small lake. LimnolOceanogr 46:108–120

Caron DA (1983) Technique for enumeration of heterotrophicand phototrophic nanoplankton, using epifluorescencemicroscopy, and comparison with other procedures. ApplEnviron Microbiol 46:491–498

Chrzanowski TH, Grover JP (2001) The light:nutrient ratio inlakes: a test of hypothesized trends in bacterial nutrientlimitation. Ecol Lett 4:453–457

Chrzanowski TH, Sterner RW, Elser JJ (1995) Nutrient enrich-ment and nutrient regeneration stimulate bacterioplank-ton growth. Microb Ecol 29:221–230

Chrzanowski TH, Kyle M, Elser JJ, Sterner RW (1997)Element ratios and growth dynamics of bacteria in anoligotrophic Canadian shield lake. Aquat Microb Ecol 11:119–125

Cole JJ (1999) Aquatic microbiology for ecosystem scientists:new and recycled paradigms in ecological microbiology.Ecosystems 2:215–225

Corzo A, Morillo JA, Rodriguez S (2000) Production of trans-

parent exopolymer particles (TEP) in cultures of Chaeto-ceros calcitrans under nitrogen limitation. Aquat MicrobEcol 23:63–72

Cotner JB, Biddanda BA (2002) Small players, large role:microbial influence on auto-heterotrophic coupling andbiogeochemical processes in aquatic ecosystems. Ecosys-tems 5:105–121

Cotner JB Jr, Wetzel RG (1992) Uptake of dissolved inorganicand organic phosphorus compounds by phytoplanktonand bacterioplankton. Limnol Oceanogr 37:232–243

Cotner JB, Ammerman JW, Peele ER, Bentzen E (1997) Phos-phorus-limited bacterioplankton growth in the SargassoSea. Aquat Microb Ecol 13:141–149

Currie DJ, Kalff J (1984) The relative importance of bacterio-plankton and phytoplankton in phosphorus uptake infreshwater. Limnol Oceanogr 29:311–321

Elser JJ, Chrzanowski TH, Sterner RW, Schampel JH, FosterDK (1995a) Elemental ratios and the uptake and release ofnutrients by phytoplankton and bacteria in three lakes ofthe Canadian Shield. Microb Ecol 29:145–162

Elser JJ, Stabler LB, Hassett RP (1995b) Nutrient limitation ofbacterial growth and rates of bacterivory in lakes andoceans: a comparative study. Mar Microb Food Webs 9:105–110

Elser JJ, Hayakawa H, Urabe J (2001) Nutrient limitationreduces food quality for zooplankton: Daphnia response toseston phosphorus enrichment. Ecology 82:898–903

Ferrier Pages C, Rassoulzadegan F (1994) N-remineralizationin planktonic protozoa. Limnol Oceanogr 39:411–418

Findlay DL, Kling HJ (1998) Protocols for monitoring bio-diversity: phytoplankton in fresh waters. Ecological Moni-toring and Assessment Network Web Site; available at:www.cciw.ca/eman-temp/research/protocols/freshwater/phyto/

Goldman JC (1986) On phytoplankton growth rates and par-ticulate C:N:P ratios at low light. Limnol Oceanogr 31:1358–1363

Grover JP (1997) Resource competition. Chapman & Hall,London

Grover JP (2000) Resource competition and community struc-ture in aquatic microorganisms: experimental studies ofalgae and bacteria along a gradient of organic carbon toinorganic phosphorus supply. J Plankton Res 22:1591–1610

Güde H (1989) The role of grazing on bacteria in planktonsuccession. In: Sommer U (ed) Plankton ecology: succes-sion in plankton communities. Springer-Verlag, NewYork, p 337–364

Guerrini F, Mazzotti A, Boni L, Pistocchi R (1998) Bacterial-algal interactions in polysaccharide production. AquatMicrob Ecol 15:247–253

Gurung TB, Urabe J, Nakanishi M (1999) Regulation of therelationship between phytoplankton Scenedesmus acutusand heterotrophic bacteria by the balance of light andnutrients. Aquat Microb Ecol 17:27–35

Healey FP (1985) Interacting effects of light and nutrient lim-itation on the growth rate of Synechococcus linearis(Cyanophyceae). J Phycol 21:134–146

Hwang SJ, Heath RT (1997) The distribution of protozoaacross a trophic gradient, factors controlling their abun-dance and importance in the plankton food web. J Plank-ton Res 19:491–518

Jürgens K (1994) Impact of Daphnia on planktonic microbialfood webs — a review. Mar Microb Food Webs 8:295–324

Jürgens K, Pernthaler J, Schalla S, Amann R (1999) Morpho-logical and compositional changes in a planktonic bacter-ial community in response to enhanced protozoan grazing.

64


Appl Environ Microbiol 65:1241–1250Karl DM (1999) A sea of change: biogeochemical variability in

the North Pacific Subtropical Gyre. Ecosystems 2:181–214Lancelot C (1983) Factors affecting phytoplankton extracellu-

lar release in the Southern Bight of the North Sea. MarEcol Prog Ser 12:115–121

Langenheder S, Jürgens K (2001) Regulation of bacterial bio-mass and community structure by metazoan and proto-zoan predation. Limnol Oceanogr 46:121–134

Larsen A, Castberg T, Sandaa RA, Brussaard CPD and 6others (2001) Population dynamics and diversity of phyto-plankton, bacteria and viruses in a seawater enclosure.Mar Ecol Prog Ser 221:47–57

Morris DP, Lewis WM (1992) Nutrient limitation of bacterio-plankton growth in Lake Dillon, Colorado. LimnolOceanogr 37:1179–1192

Nakano S (1994) Carbon:nitrogen:phosphorus ratios andnutrient regeneration of a heterotrophic flagellate fed onbacteria with different elemental ratios. Arch Hydrobiol129:257–271

Obernosterer I, Herndl GJ (1995) Phytoplankton extracellularrelease and bacterial growth: dependence on the inor-ganic N:P ratio. Mar Ecol Prog Ser 116:247–257

Pace ML, Cole JJ (1996) Regulation of bacteria by resourcesand predation tested in whole-lake experiments. LimnolOceanogr 41:1448–1460

Porter KG, Feig YS (1980) The use of DAPI for identifyingand counting aquatic microflora. Limnol Oceanogr 25:943–948

Rothaupt KO (1992) Stimulation of phosphorus-limited phyto-plankton by bacterivorous flagellates in laboratory exper-iments. Limnol Oceanogr 37:750–759

Rothaupt KO (1996) Utilization of substitutable carbon andphosphorus sources by the mixotrophic chrysophyteOchromonas sp. Ecology 77:706–715

Sanders RW, Wickham SA (1993) Planktonic protozoa andmetazoa: predation, food quality, and population control.Mar Microb Food Webs 7:197–223

Schindler DW (1977) Evolution of phosphorus limitation inlakes. Science 195:260–262

Schindler DW (1998) Replication vs. realism: the need forecosystem-scale experiments. Ecosystems 1:323–334

Schindler DW, Fee EJ, Ruszczynski T (1978) Phosphorus inputand its consequences for phytoplankton standing crop andproduction in the Experimental Lakes Area and in similarlakes. J Fish Res Board Can 35:190–196

Sherr EB, Sherr BF (1991) Planktonic microbes — tiny cells at

the base of the ocean’s food webs. Trends Ecol Evol 8:50–54

Shuter B (1979) A model of physiological adaptation in uni-cellular algae. J Theor Biol 78:519–552

Smith VH (1983) Light and nutrient dependence of photosyn-thesis by algae. J Phycol 19:306–313

Solorzano L (1969) Determination of ammonia in naturalwaters by the phenolhypochlorite method. LimnolOceanogr 14:799–801

Stainton MP, Capel MJ, Armstrong FAJ (1977) The chemicalanalysis of fresh water, 2nd edn. Can Fish Mar Serv MiscSpec Publ 25:vii + p 166

Sterner RW (1994) Seasonal and spatial patterns in macro andmicro nutrient limitation in Joe Pool Lake, Texas. LimnolOceanogr 39:535–550

Sterner RW, Hessen DO (1994) Algal nutrient limitation andthe nutrition of aquatic herbivores. Annu Rev Ecol Syst 25:1–29

Sterner RW, Chrzanowski TH, Elser JJ, George NB (1995)Sources of nitrogen and phosphorus supporting thegrowth of bacterio- and phytoplankton in an oligotrophicCanadian shield lake. Limnol Oceanogr 40:242–249

Sterner RW, Elser JJ, Fee EJ, Guildford SJ, Chrzanowski TH(1997) The light:nutrient ratio in lakes: the balance ofenergy and materials affects ecosystem structure and pro-cess. Am Nat 150:663–684

Sterner RW, Clasen J, Lampert W, Weisse T (1998)Carbon:phosphorus stoichiometry and food chain produc-tion. Ecol Lett 1:146–150

Urabe J, Sterner RW (1996) Regulation of herbivore growthby balance of light and nutrients. Proc Natl Acad Sci USA93:8465–8469

Urabe J, Watanabe Y (1992) Possibility of N or P limitation forplanktonic cladocerans: an experimental test. LimnolOceanogr 37:244–251

Urabe J, Elser JJ, Kyle M, Sekino T, Kawabata Z (2002a) Her-bivorous animals can mitigate unfavorable ratios ofenergy and material supplies by enhancing nutrient recy-cling. Ecol Lett 5:177–185

Urabe J, Kyle M, Makino W, Yoshida T, Andersen T, Elser JJ(2002b) Reduced light increases herbivore production dueto stoichiometric effects of light:nutrient balance. Ecology83:619–627

Vadstein O (2000) Heterotrophic, planktonic bacteria andcycling of phosphorus: phosphorus requirements, compet-itive ability, and food web interactions. Adv Microb Ecol16:115–167

65

Editorial responsibility: David Karl,Honolulu, Hawaii, USA

Submitted: July 10, 2002; Accepted: September 10, 2002Proofs received from author(s): January 30, 2003

Documents

Ecological stoichiometry in the microbial food web: a test of the light: nutrient hypothesis