J. N. Am. Benthol. Soc., 1987, 6(2):92-1041987 by The North American Benthological Society

Effects of three herbivores on periphytoncommunities in laboratory streams

GARY A. LAMBERTI, LINDA R. ASHKENAS, ANDSTAN V. GREGORY

Department of Fisheries and Wildlife, Oregon State University,Corvallis, Oregon 97331 USA

ALAN D. STEINMAN

Department of Botany and Plant Pathology, Oregon State University,Corvallis, Oregon 97331 USA

Abstract. The effects of grazing on algal assemblages by three different stream herbivores, themayfly Centroptilum elsa, the snail Juga silicula, and the caddisfly Dicosmoecus gilvipes, were studiedduring a 48-d experiment in six laboratory streams. Compared with ungrazed control streams,grazing by Centroptilum (500/m2) modified algal community structure slightly but had little effecton periphyton biomass and chlorophyll a. Grazing by Juga (350/m2) reduced periphyton biomassand chlorophyll a by nearly 50%, but increased the rate of primary production by up to 25%. Jugaalso prevented significant accumulation of cyanophytes and some diatom species. Grazing by Di-cosmoecus (200/m2) reduced periphyton biomass and chlorophyll a to less than 5% of the ungrazedlevels, but primary production declined by only 50%. Only adnate algal cells and short filamentspersisted on substrates grazed by Dicosmoecus. Algal export rates were increased by all three her-bivores.

Modification of algal growth patterns by both consumption and dislodgement, and dampeningof temporal fluctuations were key mechanisms by which these herbivores altered periphyton com-munities. Primary production was stimulated by low rates of grazing by Juga in the laboratorystreams, possibly as a result of increased light intensity in lower strata of the periphyton or removalof senescent algal cells. Algal assemblages displayed both community-level responses (e.g., biomass,production) and species-level responses (e.g., taxonomic composition) that should be considered inother studies of stream herbivory.

Key words: herbivory, stream, periphyton, benthic algae, grazing, Centroptilum, Juga, Dicosmoecus,primary production, chlorophyll a, export.

Guilds of grazing invertebrates in streamsusually consist of a suite of species, each of whichuses a unique feeding morphology and forag-ing behavior to harvest algal resources (Gregory1983, Lamberti and Moore 1984). Interactionsof herbivores often make it difficult, or impos-sible, to ascribe periphyton grazing responsesto particular herbivore species. Consequently,most previous investigations of grazing effectson stream periphyton communities have fo-cused on a single species either in laboratorychannels (e.g., Gregory 1983, Kehde and Wilhm1972, Sumner and McIntire 1982) or, more rare-ly, in natural streams where a single species canbe experimentally manipulated (e.g., Hart 1985,Lamberti and Resh 1983, McAuliffe 1984). Thesestudies delimit the effects of a predominantgrazer on periphyton assemblages, but generatelittle information on the range of effects that

coexisting, but functionally dissimilar, speciescan exert on periphyton. An essential step inevaluating grazer guild effects on algal assem-blages is to assess how the specific abilities ofdifferent herbivores influence those assem-blages.

In this study, we examined the separate ef-fects of three functionally different herbivoreson periphyton assemblages in laboratorystreams. These invertebrate grazers were themayfly Centroptilum elsa Traver, the caddisflyDicosmoecus gilvipes (Hagen), and the snail Jugasilicula (Gould). This study was designed tocharacterize the range of responses to grazingin the algal assemblages, and to generate hy-potheses that could be tested in future experi-ments. Our specific objectives were (1) to de-scribe the effects of each species' grazingactivities on periphyton biomass, chlorophyll

92

1987] HERBIVORY IN LABORATORY STREAMS 93

a, metabolism, export, and taxonomic compo-sition, and (2) to compare the effects of the threeherbivores and relate observed differences tograzer functional morphology.

The Herbivores

The three herbivores used in this study areabundant in streams throughout the PacificNorthwest, where they commonly coexist. Be-cause of morphological differences in feedingstructures (see below) and possible differencesin ingestion rate, food selectivity, and foragingbehavior, we postulated that their individualgrazing activities should elicit different re-sponses by the algal assemblage.

The mayfly Centroptilum (Ephemeroptera:Bae-tidae) functions as a scraper and a collector-gatherer (Edmunds 1984). Centroptilum nymphshave mouthparts fringed with brushlike hairs,allowing them to browse periphyton from thesubstrate. The caddisfly larva Dicosmoecus (Tri-choptera:Limnephilidae) uses robust bladelikemandibles and other mouthparts, which havelimited setation, to scrape periphyton (Wiggins1984). The radula of the snail Juga (Gastropoda:Pleuroceridae) is equipped with many fine teethto scrape or rasp substrates for adherent algae.Thus, these herbivores represent contrastingfeeding morphologies and methods of food ac-quisition (browsing, scraping, or rasping) thatmay affect periphyton communities differently.

Centroptilum nymphs and Juga snails were col-lected from Oak Creek, a third-order stream inthe coastal mountains of Oregon, and Dicos-moecus larvae were collected from Big Elk Creek,a fourth-order stream also in the Oregon coastalmountains. We used late instar Centroptilumnymphs, snails of intermediate size (10-15 mmtotal shell length), and third and fourth instarDicosmoecus larvae in the laboratory streams. Theanimals were held for no more than two daysbefore introduction to the laboratory streams.

Methods

Laboratory streams

This experiment was conducted in six fiber-glass laboratory streams, each 3 m long, 0.5 mwide, and 0.2 m deep (ca. 2 m 2 surface area,including sides). Each recirculating stream hadtwo parallel channels (each 0.25 m wide) sep-

arated by a centerboard that was open at bothends. A stainless steel paddlewheel, located atone end and connected by a steel shaft to avariable speed gearmotor, circulated water con-tinuously at a velocity of 10 cm / s at the sub-stratum. Well water was supplied continuouslyat a rate of 1.5 L / min, which replaced waterthat drained from the streams through a stand-pipe that maintained water depth at 10 cm. To-tal stream volume (0.15 m 3) was replaced aboutonce every 100 min and water temperature waskept at 13 ± 1°C. Light energy was provided bysix 1000-watt HID lamps (Sylvania Metalarc),which generated a photon flux density of 400AEinsteins m-2 s-' at the water surface. We useda photoperiod of 8L:16D. Nutrient concentra-tions in the well water were high (NO3-N: 6.499mg / L; NH,-N: 0.002 mg / L; P0,-P: 0.096,mg/L).

The bottom of each stream was lined with7.4 x 7.4-cm unglazed ceramic tiles (55 cm2each), which served as surfaces for algal growthand as sampling units. One out of six tiles hadan upturned end ("cove" tile) to generate mi-cro-current heterogeneity. The stream sides werelined with 15 x 15-cm tiles to ensure uniformsubstrate throughout the stream, though thesetiles were not sampled.

Experimental design

On the first day of the experiment, 7 April1985, periphyton was scraped from rocks col-lected from four streams in Benton County, Or-egon. The scrapings were homogenized for 30s in a Waring blender, brought to a volume of6 L with water, and a 1-L aliquot was added toeach of the six experimental streams. Herbi-vores were introduced into three of the streamsnine days later, 16 April, leaving three flumesas ungrazed controls. Each of the grazed streamswas stocked with one herbivore species at thefollowing density: 1) 500 Centroptilum / m 2 (ca.0.5 g ash-free dry mass/ m2), 2) 350 Juga /m2 (ca.4.0 g AFDM / m 2), or 3) 200 Dicosmoecus I m 2 (ca.2.5 g AFDM / m 2). These densities roughly cor-responded to field densities observed at the col-lection sites. The experiment was terminated on25 May, 48 d after inoculation.

Chlorophyll a standing crop was measured at4, 8, 12, 16, 24, 32, and 48 d after inoculation.Periphyton biomass was measured at 8, 12, 16,24, 32, and 48 d. Rates of primary production

94

GARY A. LAMBERTI ET AL. [Volume 6 1987]

and respiration were measured at 9, 16, 34, and48 d, and taxonomic structure of periphytonwas determined at 8, 16, 32, and 48 d. Exportwas measured at 7, 11, 15, 23, and 46 d, one totwo days before any disturbance resulting fromroutine sampling.

Three tiles were randomly selected from eachstream on each sampling date for measurementof chlorophyll a and periphyton biomass. Pe-riphyton was removed from the upper surfaceof each tile with a razor blade, homogenizedwith a small-volume blender, and split into twoequal portions. Each portion was filtered ontoa separate Millipore filter (0.45 Am pore size).For chlorophyll a analysis, one filter was groundand pigments extracted in buffered 90% acetonefor 3 hr. The scraped tile was also soaked in 90%acetone for 24 hr to extract any residual pig-ment. Chlorophyll a was measured with a spec-trophotometer using the trichromatic method(Strickland and Parsons 1968). For determina-tion of periphyton biomass as ash-free dry mass,the pre-weighed second filter was dried at 55°Cfor 24 hr, weighed, combusted at 500°C for 24hr, and reweighed.

Primary production of periphyton was de-termined using recirculating chambers similarin design to those of McIntire and Wulff (1969).Three randomly selected tiles from a stream wereplaced in the 2-L plexiglass chamber. The cham-ber was sealed and water was circulated with asubmersible pump. Three water samples weredrawn from the chamber after each of two con-secutive 2-3-hr incubations: community respi-ration (dark run) and net community primaryproduction (light run). Algal assemblages wereallowed to adjust to the light for 30 min beforethe primary production incubation began. Oxy-gen concentration in each sample was measuredwith an Orbisphere oxygen meter. Hourly rateof gross primary production was calculated byadding the loss of dissolved oxygen in the darkto the net production of dissolved oxygen inthe light and dividing by the incubation time.Daily gross primary production and daily netcommunity primary production were deter-mined based on 8 hr of primary production and24 hr of community respiration per day. Lightenergy, nutrient concentrations, current veloc-ity, and temperature were maintained as closeas possible to conditions in the laboratorystreams. Two complete incubations with differ-ent sets of tiles were conducted for each stream.

Macroinvertebrates were removed from the tilesbefore each incubation. Chlorophyll a was mea-sured for each set of tiles and used in calcula-tions of assimilation number.

Taxonomic structure of algal assemblages wasdetermined by scraping the surfaces of two ran-domly selected tiles from each stream into asingle pooled sample. Taxonomic compositionwas determined quantitatively according to themethod described in detail in Steinman andMcIntire (1986). In brief, periphyton assem-blages were fixed in Lugol's solution, settled in50-m1 chambers, and 500 algal units per samplewere counted at 400 x with a Nikon MS in-verted microscope. All algae were identified tospecies in this step, except diatoms which werecounted and identified at 1250 x with a ZeissRA microscope. An algal unit was an individualcell or valve if the taxon was unicellular, or anindividual filament or colony if it was multi-cellular. Mean biovolume of each algal taxonwas determined from measurements of at least10 cells per taxon using standard geometric for-mulae. The biomass of each algal taxon was es-timated by multiplying the taxon's proportionof the community biovolume by the total com-munity biomass.

Export of particulate organic matter (POM >10 Am) from the streams was measured by plac-ing a 10-Am mesh bag under the standpipe,which filtered all effluent water for a given timeinterval (30-60 min; 45-90 L volume). Exportwas measured simultaneously for all streams atmid-morning (0900-1000 hr) on each samplingdate. Each sample was analyzed for ash-free drymass as described for epilithon samples. In-fluent well water contained no POM.

Statistical analyses

The goal of our experiment was to obtain abroad view of the grazing process by evaluatingthree different herbivores and to generate test-able hypotheses for future experiments. Be-cause this approach sacrificed treatment repli-cation for breadth (i.e., different herbivorestreams were not replicated), grazing effectscould not be examined with inferential statis-tics. However, replication of the ungrazedstreams (n = 3) provided a basis for comparingstreams that received the same treatment (i.e.,degree of inter-stream variability), and indicat-ed that there was little difference among those

60A

50 -E

40

4

1.4B

1.2-

1.0 -

0.4

0.2 -

0 • Ar".....40

FIG. 1. Peripbiomass, (B) chiPOM export. Va

streams (see 11system have vceiving the saegal assemblage1987). As a bavariation, wemeasured perilrophyll a, gro:nity respiratio:mary statisticchlorophyll ratassimilation nt.ual algal taxa),4are treated as sipiing date.

Taxonomic cin different stnmeasure of cowMoore 1977). Tto 1, where 0 ino species in ccidentical in bo,ative abundancalgal biovolum■

•4. 11

•

504010 20 30TIME (days)

C240

200 -

es160 -

120 -

58a 0 -

40-

60

30-

20-

0

tf

0

FIG. 1. Periphyton dynamics in three replicate streams containing no herbivores, presented as (A) algalbiomass, (B) chlorophyll a, (C) biomass : chlorophyll ratio (jointly measured beginning on day 8), and (D)POM export. Values presented are means (± 1 SE; n = 3, for A, B) of within-stream samples.

95[Volume 6 1987] HERBIVORY IN LABORATORY STREAMS

streams (see Results). Previous studies in oursystem have verified that different streams re-ceiving the same treatment produce similar al-gal assemblages (Steinman and McIntire 1986,1987). As a basis for representing intra-streamvariation, we present the standard errors ofmeasured periphyton attributes (biomass, chlo-rophyll a, gross primary production, commu-nity respiration) on each sampling date. Sum-mary statistics based on means (biomass :chlorophyll ratio, production : respiration ratio,assimilation number), pooled samples (individ-ual algal taxa), or time-interval samples (export)are treated as single measurements on each sam-pling date.

Taxonomic composition of algal assemblagesin different streams was compared by the SIMImeasure of community similarity (McIntire andMoore 1977). The SIMI value can range from 0to 1, where 0 indicates two communities haveno species in common, and 1 indicates they areidentical in both species composition and rel-ative abundance. SIMI values were based onalgal biovolume (Steinman and McIntire 1986).

Results

Periphyton development in ungrazed streams

Three streams were used to assess develop-ment of benthic algal assemblages in the ab-sence of herbivores. Development of the algalassemblage in terms of algal biomass was con-sistent with a logistic growth model (Fig. 1A).The initial rate of accumulation was slow (days0-8), followed by a period of rapid biomass ac-cumulation (logarithmic growth phase; days 8-24), and finally a period of relatively constantbiomass (asymptotic phase; days 24-48). Vari-ability among replicates streams in biomass ac-cumulation was low, at least until day 32. Afterday 32, algal senescence occurred, as shown bychloroplast degradation and decline in mat co-hesiveness. This deterioration of the periphy-ton community resulted in spatial heteroge-neity or patchiness in the assemblage (seeSteinman and McIntire 1987) and probably ac-counted for differences in mean biomass amongthe ungrazed streams by day 48.

red from the tilesphyll a was mea-used in calcula-

assemblages wasfaces of two ran-h stream into aMc compositionaccording to the

Steinman andriphyton assem-kution, settled inunits per samplei Nikon MS in-ere identified tooms which were)x with a Zeissas an individualnicellular, or anif it was multieach algal taxonitents of at leastd geometric for-Al taxon was es-on's proportiony the total corn-

matter (POM >easured by plac-the standpipe,

for a given timerolume). Exportor all streams att each samplingfor ash-free dryIn samples. In-POM.

was to obtain aby evaluating

o generate test-petiments. Be-reatment repli-rent herbivoregrazing effectsferential statis-the ungrazedfor comparingtreatment (i.e.,V), and indicat-e among those

96

GARY A. LAMBERTI ET AL. [Volume 6 1

ti

Standing crops of chlorophyll a peaked in allthree streams at day 24, then declined rapidlythrough day 48 (Fig. 1B). This decline in chlo-rophyll a probably reflects the senescence of thealgal assemblage. The biomass-to-chlorophyllratio remained relatively constant until day 32(Fig. 1C), then increased markedly through day48. Allocation of photosynthate into storageproducts, lower photosynthetic activity, and de-creased pigment synthesis are typical of latestages of algal community development (Wetz-el and Westlake 1969), and are all processes thatcould result in increased ratios of biomass tochlorophyll a.

Export measured POM >10 Am that was car-ried with the current and passed out of thestreams. Some algae (e.g., single cells < 10 Am)undoubtedly passed through the collection net,which would result in an underestimate of ex-port. However, most taxa were probably re-tained by the net either because they had afilamentous or colonial growth form, were in-tertwined with other algal taxa, or because in-dividual cells were large. Consequently, under-estimates of algal export were probably minor.Export rates generally increased with biomassaccumulation in the ungrazed streams (Fig. 1D).Rates of export varied more than biomass orchlorophyll a among replicate streams, possiblybecause of occasional sloughing of large algalclumps during the 30-60 min sampling period.

Grazer effects on periphyton biomass,pigment, and export

All three types of grazers, the mayfly Cen-troptilum, snail Juga, and caddisfly Dicosmoecus,depressed periphyton biomass to some degreerelative to ungrazed algal assemblages (Fig. 2A).Grazing by Centroptilum slowed the rate of bio-mass accumulation after day 16, resulting inslightly less biomass at the asymptotic phasethan that of the ungrazed streams; this differ-ence (about a 20% reduction) persisted through-out the experiment. Juga reduced the rate ofbiomass accumulation soon after introduction.Grazing by the snails maintained periphytonbiomass at 15-20 g /m2 after day 16, about onehalf of the biomass in the ungrazed streams.Dicosmoecus larvae reduced algal biomass im-mediately after their introduction, and, unlikethe other herbivores, resulted in a net decline

in periphyton biomass to very low levels ( <1g /m2) for the duration of the experiment.

Compared with standing crops of pigment inthe ungrazed streams, the amount of chloro-phyll a was also reduced by each type of grazer(Fig. 2B). Centroptilum and Juga reduced the max-imum level of chlorophyll a by about 40% and60%, respectively, whereas grazing by Dicos-moecus reduced chlorophyll a by over 95%. Thedecline in chlorophyll a abundance observedin the ungrazed streams after day 24 also oc-curred in two of the grazed systems (Centrop-tilum, Juga). In these latter cases, senescence ofthe algal assemblage apparently contributed tothe dynamics of chlorophyll a.

The biomass-to-chlorophyll ratio was similarin the ungrazed, Centroptilum, and Juga streams,generally ranging from 50 to 80 during the first32 d of the experiment (Fig. 2C). Following thedeterioration in the algal community, whenchlorophyll a declined but biomass remainedrelatively constant, this ratio increased to >100in these three systems. The biomass-to-chloro-phyll ratio was consistently low (<30) in thestream grazed by Dicosmoecus.

Export rates of periphyton increased in allthree streams soon after herbivore introduction(Fig. 2D). By day 15, export rates in the grazedstreams were several times higher than exportin the ungrazed systems. Patterns of export re-flected the extent of reduction of algal abun-dance by herbivores; export was highest for theDicosmoecus treatment, indicating that these an-imals were most disruptive of the algal assem-blage, and intermediate for the Juga and Cen-troptilum systems. Export rates declined betweendays 15 and 46 in the grazed streams, but didnot change appreciably in the ungrazed streams(Fig. 2D). Biomass remained relatively constantin all streams over this same period (Fig. 2A),indicating that biomass-specific export rates (i.e.,export per unit attached biomass) also declinedin the grazed streams.

Grazer effects on periphyton metabolism

Rates of primary production and respirationwere measured for the ungrazed, Juga, and Di-cosmoecus streams. Rates of gross primary pro-duction (GPP) increased in all three systemsduring the course of the experiment (Fig. 3A).Grazing by Dicosmoecus reduced GPP by about50% as compared with the ungrazed streams,

NS

cacti

butaboafte

Cung16 aDicaday:dowheP/Rperithe .thethanCorn:diffedaystrewand

Padispiabunstancgreat

1987]

HERBIVORY IN LABORATORY STREAMS 97

2. Periphyton dynamics in streams containing either Centroptilum mayflies, Juga snails, or Dicosmoecuscaddisflies, compared with a mean of the three ungrazed streams. A-D as in Figure 1. Arrow shows date ofanimal introduction.

20 30ISSE (days)

0 40 50

1.2B

50

210

3 140

as 150

60

3

50

40 -

30 -

204

10

0

but grazing by Juga actually stimulated GPP byabout 25% over that in the ungrazed streamsafter day 16.

Community respiration (CR) increased in theungrazed and Juga streams between days 9 and16 and then remained steady (Fig. 3B); in theDicosmoecus stream CR gradually increased fromdays 16 to 48. The production-to-respiration ra-tio (P/R) was slightly greater than one (1.06)when the animals were introduced (Fig. 3C).P/R increased in all treatments during the ex-periment. By day 48, GPP was more than doublethe respiratory demands of the community inthe ungrazed and Dicosmoecus streams, and morethan triple the CR in the Juga stream. Daily netcommunity primary production (the absolutedifference between daily GPP and daily CR) onday 48 was 3.6 g 02 m - 2 d -' in the ungrazedstreams, 5.2 g 02 m- 2 d-' in the Juga stream,and 1.7 g 02 m- 2 d-' in the Dicosmoecus stream.

Patterns of algal production (Fig. 3A) did notdisplay the extreme differences shown in algalabundance (Fig. 2A). For example, thoughstanding crops of chlorophyll a were muchgreater in the ungrazed streams than in the Di-

cosmoecus stream, rates of GPP in the ungrazedstreams were only slightly more than twice theproduction in the Dicosmoecus stream. This re-sponse reflects the influence of grazing on therate of primary production per mass of chlo-rophyll a (chlorophyll-specific primary produc-tion) in the algal assemblage. At the light in-tensities used in this experiment (400 AEinsteinsm-2 s-') photosynthesis was almost certainlysaturated (Jasper and Bothwell 1986), so chlo-rophyll-specific primary production representsassimilation number. Assimilation number wasalways higher in the grazed streams than in theungrazed streams (Fig. 3D). Thus, reductions inalgal abundance did not cause equivalent de-clines in primary production. In fact, reductionsin biomass and chlorophyll a in the Juga streamwere accompanied by increases in gross pri-mary production.

Grazer effects on algal community structure

A total of 46 algal species was identified dur-ing the experiment. The number of taxa wasrelatively consistent for each treatment: 33

tme 6

Is (<1t.ent inltloro-FrazerL` max-% andDicos-I. Theerved

oc-ntrop-Ice ofted to

imilarearns,e first4; thewhenwined>100tloro-n the

in allfictionrazedxportart re-tbun-w thew an-stem-Cen-

weenLt didearnsstant2A),(i.e.,

lined

ationd Di-pro-

terns3A).bout21118,

98 GARY A. LAMBERTI ET AL. [Volume 6 198

8N 1

6

6 5

40a.

•

3

2

1

o

z220a 0

9 18 34TIME (days)

48 0 9 16 34TIME (days)

48

3

E

2-

z

1-

4

N g

N0

N

2

rc

6 1

3

1.

O

0

F 120

r

•

too

° scro80

240

20

B

FIG. 3. Metabolism of periphyton grazed by Juga snails, Dicosmoecus caddisflies, or ungrazed, for fivesampling dates during the experiment, presented as (A) rate of gross primary production (1 + 1 SE; n = 2), N IS(B) community respiration, (C) 24-hr production : respiration ratio, and (D) assimilation number. Animalswere introduced after measurements on day 9. 121species were found in the ungrazed stream, 35species in the Centroptilum stream, 37 species inthe Juga stream, and 29 species in the Dicosmoe-cus stream.

Greater than 90% of the periphytic materialby mass consisted of identifiable algae. Exami-nation of live and fixed samples by light mi-croscopy revealed very few bacterial cells orfungal hyphae. In addition, observation of tileswith scanning electron microscopy failed to re-veal significant amounts of non-algal biomass(Steinman and McIntire 1986, 1987). This maybe related to a lack of microbial inocula in thewater supply. The small amount of detrituspresent was therefore of algal origin. Assuminga detrital composition similar to the live algalcommunity, specific biomasses could be esti-mated from individual biovolumes and totalcommunity biomass (Fig. 4).

The three dominant diatom taxa respondeddifferently to the experimental treatments.Nitzschia oregona and Achnanthes lanceolata weremost abundant in the ungrazed streams, where

they increased throughout the experiment (Figs.4A, B). These taxa also increased in the Cen-troptilum and Juga systems until day 32 and thendeclined; both taxa were sparse in the Dicos-moecus stream after day 16. The diatom Synedraulna showed high biomass in the ungrazed andCentroptilum streams, intermediate in the Juga,and low in the Dicosmoecus (Fig. 4C). Synedraulna was dominant in early algal assemblagedevelopment in all treatments, peaking at day16 and then declining.

The cyanophyte Phormidium tenue and chlo-rophytes Scenedesmus spp. showed little accu-mulation until day 32, after which P. tenue ac-cumulated substantial biomass in the ungrazedand Centroptilum streams (Fig. 4D). Scenedesmusincreased rapidly in all but the Dicosmoecustreatment (Fig. 4E), representing one third toone half of total community biovolume by day48 (Table 1).

The biomass of filamentous chlorophytes,predominantly Stigeoclonium tenue, peaked at day32 in the ungrazed, Centroptilum, and Juga treat-

I 6

3.

0'

FIG.may&shows

mentsmulatiowingflametotal cday 48succesnizatitStigeocsucceadesmus

rtiE

yH

ata

1.2 -

0 10 20 30

TIME (days)0 10 20 30

TAE (days)40 50 40 50

HERBIVORY IN LABORATORY STREAMS

2.4

Phormidium tenue

1987) 99

aF Filamentous

Chlorophytes

20-

1.6 -

0.8

0.4 -

FIG. 4. Biomass accumulation for the six dominant algal taxa in streams containing either Centroptilummayflies, Juga snails, or Dicosmoecus caddisflies, compared with a mean of the three ungrazed streams. Arrowshows date of animal introduction.

ments (Fig. 4F). Though no algal taxon accu-mulated much biomass in the Dicosmoecus streamowing to heavy grazing pressure (Fig. 2A), thefilamentous forms did account for >50% of thetotal community biovolume in that system onday 48 (Table 1). In general, the pattern of algalsuccession in the streams included early colo-nization and dominance by Synedra, growth ofStigeoclonium, Achnanthes, and Nitzschia at mid-successional stages, and late growth of Scene-desmus and Phormidium.

SIMI measures of algal community similarityat day 8 showed that there was a high degreeof similarity (SIMI > 0.99) among the four treat-ments before grazer introduction (Table 2). Byday 16, algal community structure in the Dicos-moecus stream began to diverge, but the othertreatments remained similar. On day 32, algalassemblages grazed by Centroptilum and Juga re-mained quite similar to each other (SIMI = 0.96)but differed from the control streams (SIMI <0.40). The algal community grazed by Dicos-

tine 6

or five2),

nimals

(Figs.Cen-

i thenDicos-vnedrad and

Juga,piedra'Wageit day

chlo-accu-re ac-razedesmusioecusird to

day

tytes,it daytreat-

TABLE 1. Relative abundance, as percent of totalcommunity biovolume, of the six dominant algal taxain each treatment at day 48.

Taxon

Relative Abundance (%)

Un-grazed

Centro',tilum Jura

Dicos-moecus

Nitzschia oregona 5.1 0.5 0.5 0.1Achnanthes lanceo-

lata 9.0 1.8 3.0 8.5Synedra ulna 19.6 22.1 8.5 7.5Phormidium tenue 3.1 7.2 0.0 19.0Scenedesmus spp. 32.3 47.0 51.6 0.0Filamentous chlo-

rophytes 5.1 2.3 4.8 57.2All other algal taxa 25.8 19.1 31.6 7.7

Total 100.0 100.0 100.0 100.0

moecus was distinct from all other streams(SIMI < 0.35). At day 48, similarity increasedbetween both the Juga and Centroptilum treat-ments and the ungrazed streams (SIMI > 0.70).Algal community structure in the Dicosmoecussystem continued to be distinct throughout the48 d (SIMI < 0.20).

Discussion

Laboratory stream experiments cannot repro-duce the degree of natural variation and envi-ronmental heterogeneity imposed by naturalsystems. However, laboratory studies profit fromthe ability to maintain certain environmentalparameters constant (such as light, nutrients,flow, and substrate in our study) while manip-ulating key features of interest (type of herbi-vore in our study). In natural streams where theherbivore community may consist of manyspecies, it may be difficult to ascribe responsesin the periphyton community to individualgrazer species. The laboratory streams allowedus to study a single species in isolation, to ac-curately describe its grazing effects, and to iden-tify important relationships for examination insubsequent experiments.

Our primary objective was to compare therelative effects of three different grazers onstream periphyton communities. We used threeherbivore species that are widely distributedand frequently abundant in streams of the Pa-cific Northwest. In addition, two of these species(Dicosmoecus gilvipes and Juga silicula) have been

TABLE 2. Matrices of similarity values (SIMI) ofalgal community structure for grazing treatments bysampling date. A single randomly selected stream wasused to represent the ungrazed treatment on eachdate.

Un-grazed

Centrop-tilum Juga

Dicos-moecus

Day 8Ungrazed 1.000Centroptilum 0.998 1.000Juga 0.998 0.999 1.000Dicosmoecus 0.995 0.999 0.999 1.000

Day 16Ungrazed 1.000Centroptilum 0.955 1.000Juga 0.969 0.993 1.000Dicosmoecus 0.863 0.840 0.882 1.000

Day 32Ungrazed 1.000Centroptilum 0.393 1.000Juga 0.380 0.958 1.000Dicosmoecus 0.299 0.324 0.296 1.000

Day 48Ungrazed 1.000Centroptilum 0.763 1.000Juga 0.748 0.899 1.000Dicosmoecus 0.129 0.148 0.101 1.000

identified previously as key herbivores in west-ern streams (Gregory 1983, Hart 1981, Hawkinsand Furnish 1987, Sumner and McIntire 1982),and functionally analogous species can be foundin other geographical areas (Kehde and Wilhm1972, McAuliffe 1984, Mulholland et al. 1983).

Previous investigations have demonstratedthat stream herbivores can affect periphytonbiomass (Lamberti and Resh 1983, McAuliffe1984, Power et al. 1985), rates of production(Gregory 1983, Lamberti and Resh 1983), taxo-nomic structure (Hart 1985, Sumner and Mc-Intire 1982), export (Mulholland et al. 1983,Sumner and McIntire 1982), and nutrient cy-cling (Mulholland et al. 1983). However, mostprevious studies have documented the responsein only one or two periphyton parameters, suchas biomass accumulation or pigment concentra-tion, rather than examining the array of featuresthat characterize plant-herbivore interactions(Gregory 1983). This study has demonstratedthat measuring a variety of features in the pe-riphyton community may be necessary to ade-

19871

quatelyA singlyductivitonly a sof the ayton bicfrom wtexport,

Grazer re

The tFhad diffigal matslightlythe ungithat indiyton fro,immediatid may8algal assGrazing Istandingcommunimediate 1cosmoecusby Jugasions (HalaboratorMcIntiremoecus impigment tany accuning Trichta prevailber systemResh 1983

The witment obsealgal userstreams diJuga and Lalgal assetcame senein the bioDicosmoecutions duriblage deveble standiexperimenments nevieral, thein grazing

100

GARY A. LAMBERTI ET AL. [Volume 6

1987] HERBIVORY IN LABORATORY STREAMS 101

quately characterize algal response to grazing.A single feature (e.g., biomass, pigment, pro-ductivity, or taxonomy) would have describedonly a small fraction of the observed responsesof the algal assemblage. For example, periph-yton biomass alone was not a suitable measurefrom which to infer chlorophyll a, metabolism,export, or community structure.

Grazer regulation of algal biomass and export

The three herbivores examined in this studyhad differing effects on the accumulation of al-gal material. Grazing by Centroptilum onlyslightly reduced algal biomass below levels inthe ungrazed streams. However, we observedthat individual Centroptilum could clear periph-yton from small patches ( <2 cm in diameter)immediately around themselves, and other bae-tid mayflies produce similar patches in benthicalgal assemblages (Wiley and Kohler 1984).Grazing by Juga greatly reduced the maximumstanding crop attained and resulted in an algalcommunity with biomass and pigment inter-mediate between that of Centroptilum and Di-cosmoecus. Similar reductions of algal biomassby Juga have been observed in stream diver-sions (Hawkins and Furnish 1987) and in otherlaboratory channels (Gregory 1983, Sumner andMcIntire 1982). Grazing by the caddisfly Dicos-moecus immediately reduced algal biomass andpigment to extremely low levels and preventedany accumulation of algal material. Some scrap-ing Trichoptera have been shown to have sucha prevailing effect on algal accumulation in oth-er systems as well (Hart 1981, Lamberti andResh 1983, McAuliffe 1984).

The wide fluctuations in biomass and pig-ment observed during the asymptotic phase ofalgal assemblage development in the ungrazedstreams did not occur in the streams grazed byJuga and Dicosmoecus. After day 24, the benthicalgal assemblage of the ungrazed streams be-came senescent, as shown by a rapid increasein the biomass: chlorophyll ratio. Grazing byDicosmoecus and Juga dampened these fluctua-tions during the asymptotic phase of assem-blage development and produced relatively sta-ble standing crops of algae throughout theexperiment. Algal assemblages in these treat-ments never became thick mats of algae. In gen-eral, the three herbivores produced a gradientin grazing pressure ranging from strong re-

duction of algal biomass by Dicosmoecus to therelatively weak influence of Centroptilum.

All three grazers substantially increased therate of algal export, because the animals me-chanically dislodged algae as they moved acrossthe substrate. Dicosmoecus larvae, in particular,disrupted algae by vigorously clawing at thealgal material with their front tarsi and by drag-ging their stone cases as they moved. At timesin the laboratory streams, periphyton dislodgedmay exceed that consumed (Lamberti, unpub-lished data). In natural streams, dislodgementmay be an important mechanism by whichstream herbivores regulate periphyton com-munities while generating high-quality foodresources for filter-feeding and deposit-collect-ing organisms. Herbivore-dislodged benthic al-gae may be important in the bioenergetics ofdownstream communities, just as phytoplank-ton released in the outflows of impoundmentscan determine the production of filter-feedingcaddisflies and blackflies below dams (Ward1984). Untidy feeders such as Dicosmoecus maycontribute to the pool of algal seston in a man-ner disproportionate to their abundance.

Biomass-specific export rates declined overtime in all of the grazed streams. This may havebeen due to declining rates of feeding or move-ment by the animals, or to shifts in algal com-position or physiognomy to forms that resistedexport. For example, in the Dicosmoecus stream,algal physiognomy shifted from large overstorycells and filaments to adnate forms that prob-ably resisted detachment by herbivore activity.

Grazer regulation of periphyton metabolism

Increases in the production per unit biomass(P /B) of periphyton by grazing have been ob-served frequently in habitats ranging from lakelittoral zones (Flint and Goldman 1975) to nat-ural streams (Lamberti and Resh 1983) and lab-oratory flumes (Gregory 1983). This increase inalgal P / B appears to occur over a wide range ofgrazing pressure (Lamberti and Moore 1984).However, stimulation of gross primary produc-tion of periphyton may occur only under rel-atively low grazing pressure (Cooper 1973, Flintand Goldman 1975, Hargrave 1970). Moderateto high levels of grazing most frequently resultin a decline in gross primary production, as hasbeen demonstrated in ponds (Cuker 1983, Seale1980) and streams (Lamberti and Resh 1983).

102 GARY A. LAMBERTI ET AL. [Volume 6

Grazing pressure in most streams is sufficientto reduce absolute rates of gross or net primaryproduction, a condition that has been termed"overgrazing" in aquatic systems (Lamberti andMoore 1984) or "undercompensation" in ter-restrial habitats (Belsky 1986). Our experiment,however, suggests that some levels of grazing(e.g., that applied by Juga) may increase ratesof gross and net primary production.

Grazing by both Dicosmoecus and Juga in-creased the production per unit biomass of pe-riphyton. However, Dicosmoecus substantiallyreduced gross primary production comparedwith ungrazed streams. Only the grazing byJuga increased gross primary production overthat of the ungrazed algal assemblage. Most ex-planations for production stimulation by graz-ing center around nutrient renewal by algal celldisruption and excretion by grazers, or reducedcompetition for those nutrients (Gregory 1983,Lamberti and Resh 1983). Because of high nu-trient concentrations in the laboratory streams,it is doubtful that nutrient supply could explainthe enhanced production (i.e., nutrients wereprobably not limiting at any time). More likelyin our system, biomass reduction by the grazersincreased light levels in the lower strata of theassemblage. If this were accompanied by re-moval of senescent cells, either by consumptionor dislodgement, then increased rates of pri-mary production could result. Further study isneeded to identify the precise mechanisms op-erating in our system.

Grazer regulation of community structure

Several previous studies have shown that thetaxonomic structure of periphyton communi-ties can be altered by either grazing caddisflies(e.g., Hart 1981, Lamberti and Resh 1983,McAuliffe 1984) or snails (e.g., Cuker 1983,Gregory 1983, Sumner and McIntire 1982), andmay be modified, at least at a local level, bygrazing mayflies (e.g., Wiley and Kohler 1981).The herbivores in our study also influenced thetaxonomic structure of the algal community. Byday 32, there was little similarity in communitystructure between any pair of treatments exceptthe Centroptilum and Juga streams.

Taxonomic structure appeared to be relatedto the type of grazer and the intensity of har-vest. For example, at day 48 the algal assemblagegrazed by Dicosmoecus was dominated by basal

cells and short erect filaments of Stigeocloniumtenue, resulting in a relatively homogeneous al-gal monolayer. In contrast, ungrazed assem-blages at that time included large amounts ofScenedesmus spp. and several diatom species.Apparently, Dicosmoecus, a scraper that possess-es robust mandibles, could remove most algaeexcept for basal structures of S. tenue. Juga feedswith an efficient fine-toothed radula that wasparticularly effective at removing Synedra ulna,a large rosette-forming diatom, and Phormidiumtenue, a cyanophyte whose loose filamentousform may make it susceptible to grazing or dis-lodgement. Assemblages grazed by Centroptilumhad lower abundances of some diatoms, butcomparable amounts of filaments, as ungrazedassemblages. The delicate, brushlike mouth-parts of the mayflies may be more effective atharvesting diatoms than at removing long fil-aments. In general, assemblages grazed by Cen-troptilum and Juga were more heterogeneous inappearance than those grazed by Dicosmoecus,possibly reflecting the lower intensity of graz-ing by the mayflies and snails that allowed un-interrupted development of certain algalpatches.

Closing remarks

Our study demonstrated that each type ofherbivore can alter periphyton assemblages,though the effects differ and may be detectableonly as community-level responses (e.g., bio-mass, production) in some cases, or species-levelresponses (e.g., taxonomic shifts) in other cases.Two of the three herbivores (Dicosmoecus andJuga) substantially modified patterns of algalgrowth (community-level response), but thisdegree of influence was not demonstrated byCentroptilum. All three herbivores modified taxo-nomic composition (species-level response) to somedegree and increased algal export. Dislodge-ment of periphyton into export is a potentiallyimportant mechanism of algal community mod-ification that has significant implications for thenutrition of downstream consumers in naturalstreams. Because this study examined only sin-gle densities of three types of herbivores, ex-tension of these results to other systems is lim-ited. Further investigation is needed todetermine the effects of a range of densities ofdifferent types of herbivores on periphyton as-semblages. For example, grazing pressure can

1987] HERBIVORY IN LABORATORY STREAMS 103

then be varied to bracket the response in pri-mary production and to test hypotheses con-cerning potential stimulation of primary pro-duction by stream herbivores.

Acknowledgements

We thank Jim Fairchild, Karan Fairchild, JudyLi, and Randy Wildman for their assistance inseveral phases of this research. Dave McIntireparticipated in the design of the experimentand in interpretation of the algal taxonomic data.The manuscript benefited from constructive re-views by Amy Ward, Win Fairchild, and ananonymous reviewer. This study was supportedby research grant BSR-8318386 from the Na-tional Science Foundation.

Literature CitedBELSKY, A. J. 1986. Does herbivory benefit plants? A

review of the evidence. American Naturalist 127:870-892.

COOPER, D. C. 1973. Enhancement of net primaryproductivity by herbivore grazing in aquatic lab-oratory microcosms. Limnology and Oceanog-raphy 18:31-37.

CUKER, B. E. 1983. Grazing and nutrient interactionsin controlling the activity and composition of theepilithic algal community of an arctic lake. Lim-nology and Oceanography 28:133-141.

EDMUNDS, G. F. 1984. Ephemeroptera. Pages 94-125in R. W. Merritt and K. W. Cummins (editors).An introduction to the aquatic insects of NorthAmerica. 2nd edition. Kendall/Hunt PublishingCompany, Dubuque, Iowa.

FLINT, R. W., AND C. R. GOLDMAN. 1975. The effectsof a benthic grazer on the primary productivityof the littoral zone of Lake Tahoe. Limnologyand Oceanography 20:935-944.

GREGORY, S. V. 1983. Plant-herbivore interactions instream systems. Pages 157-189 in J. R. Barnes andG. W. Minshall (editors). Stream ecology. PlenumPress, New York.

HARGRAVE, B. T. 1970. The utilization of benthicmicroflora by Hyalella azteca (Amphipoda). Jour-nal of Animal Ecology 39:427-437.

HART, D. D. 1981. Foraging and resource patchiness:field experiments with a grazing stream insect.Oikos 37:46-52.

HART, D. D. 1985. Grazing insects mediate algal in-teractions in a stream benthic community. Oikos44:40-46.

HAWKINS, C. P., AND J. K. FURNISH. 1987. Are snailsimportant competitors in stream ecosystems? Oi-kos (in press).

JASPER, S., AND M. L. BOTHWELL. 1986. Photosyn-thetic characteristics of lotic periphyton. Cana-dian Journal of Fisheries and Aquatic Sciences43:1960-1969.

KEHDE, P. M., AND J. L. WILHM. 1972. The effects ofgrazing by snails on community structure of pe-riphyton in laboratory streams. American Mid-land Naturalist 87:8-24.

LAMBERTI, G. A., AND J. W. MOORE. 1984. Aquaticinsects as primary consumers. Pages 164-195 inV. H. Resh and D. M. Rosenberg (editors). Theecology of aquatic insects. Praeger Publishers,New York.

LAMBERTI, G. A., AND V. H. RESH. 1983. Stream pe-riphyton and insect herbivores: an experimentalstudy of grazing by a caddisfly population. Ecol-ogy 64:1124-1135.

MCAULIFFE, J. R. 1984. Resource depression by astream herbivore: effects on distributions andabundances of other grazers. Oikos 42:327-333.

MCINTIRE, C. D., AND W. W. MOORE. 1977. Marinelittoral diatoms: ecological considerations. Pages333-371 in D. Werner (editor). The biology ofdiatoms. University of California Press, Berkeley.

McINrotE, C. D., AND B. L. WULFF. 1969. A laboratorymethod for the study of marine benthic diatoms.Limnology and Oceanography 14:667-678.

MULHOLLAND, P. J., J. D. NEWBOLD, J. W. ELWOOD, ANDC. L. Hom. 1983. The effect of grazing intensityon phosphorus spiralling in autotrophic streams.Oecologia (Berlin) 58:358-366.

POWER, M. E., W. J. MATTHEWS, AND A. J. STEWART.1985. Grazing minnows, piscivorous bass, andstream algae: dynamics of a strong interaction.Ecology 66:1448-1456.

SEALE, D. B. 1980. Influence of amphibian larvae onprimary production, nutrient flux, and compe-tition in a pond ecosystem. Ecology 61:1531-1550.

STEINMAN, A. D., AND C. D. MCINTIRE. 1986. Effectsof current velocity and light energy on the struc-ture of periphyton assemblages in laboratorystreams. Journal of Phycology 22:352-361.

STEINMAN, A. D., AND C. D. MCINTIRE. 1987. Effectsof irradiance on the community structure andbiomass of algal assemblages in laboratorystreams. Canadian Journal of Fisheries andAquatic Sciences (in press).

STRICKLAND, J. D. H., AND T. R. PARSONS. 1968. Apractical handbook of seawater analysis. Fisher-ies Research Board of Canada, Ottawa.

SUMNER, W. T., AND C. D. MCINTIRE. 1982. Grazer-periphyton interactions in laboratory streams.Archiv filr Hydrobiologie 93:135-157.

WARD, J. V. 1984. Ecological perspectives in themanagement of aquatic insect habitat. Pages 558-577 in V. H. Resh and D. M. Rosenberg (editors).The ecology of aquatic insects. Praeger Publish-ers, New York.

104 GARY A. LAMBERT' ET AL. [Volume 6

WETZEL, R. G., AND D. F. WESTLAKE. 1969. Periphy-ton. Pages 33-40 in R. A. Vollenweider (editor).A manual on methods for measuring primaryproduction in aquatic environments. Interna-tional Biological Programme Handbook 12.Blackwell Scientific Publications, Oxford.

WIGGINS, G. B. 1984. Trichoptera. Pages 271-311 inR. W. Merritt and K. W. Cummins (editors). Anintroduction to the aquatic insects of NorthAmerica. 2nd edition. Kendall/Hunt PublishingCompany, Dubuque, Iowa.

WILEY, M. J., AND S. L. KOHLER. 1981. An assessment

of biological interactions in an epilithic streamcommunity using time-lapse cinematography.Hydrobiologia 78:183-188.

WILEY, M. J., AND S. L. KOHLER. 1984. Behavioraladaptations of aquatic insects. Pages 101-133 inR. W. Merritt and K. W. Cummins (editors). Anintroduction to the aquatic insects of NorthAmerica. 2nd edition. Kendall/Hunt PublishingCompany, Dubuque, Iowa.

Received: 5 January 1987Accepted: 21 April 1987