Transactions of the American Fisheries Society, 1992, v.121, n.4, p.494-507

ISSN: (Print 0002-8487) (Online 1548-8659)

doi: 10.1577/1548-8659(1992)121<0494:FCBLGS>2.3.CO;2

http://www.fisheries.org

http://afsjournals.org/loi/fitr

http://afsjournals.org/doi/pdf/10.1577/1548-8659%281992%29121%3C0494%3AFCBLGS%3E2.3.CO%3B2

© 1992 American Fisheries Society

Transactions of the American Fisheries Society 121:494-507, 1992<& Copyright by ihc American Fisheries Society 1992

Food Consumption by Larval Gizzard Shad: ZooplanktonEffects and Implications for Reservoir Communities

JOHN M. DETTMERS AND ROY A. STEINAquatic Ecology Laboratory, Department of Zoology

The Ohio State University. Columbus, Ohio 43210, USA

Abstract.—Because peak abundance of larval gizzard shad Dorosoma cepedianum occurs si-multaneously with the midsummer decline of macrozooplankton in Ohio reservoirs, we hypoth-esized that zooplanktivory by larval gizzard shad caused this decline. To test this hypothesis, wecompared larval food consumption with zooplankton productivity in two reservoirs. Larval gizzardshad began to influence zooplankton production in a reservoir with high zooplankton productivity(exceeding 125 mg-m 3-d ') only after peak zooplankton biomass occurred, even at high larvaldensities (38 shad-nr3). However, consumption by early juvenile (25-30-mm) gizzard shad se-verely reduced zooplankton in this reservoir. Conversely, relatively low densities of larval gizzardshad (3-7 shad-m ?) had variable effects on zooplankton in a reservoir with low zooplanktonproductivity (at most 4 mg-m-'-d"1). Depending on larval gizzard shad density and zooplanktonproduction, larval gizzard shad alone or in conjunction with early juveniles may control zooplank-ton assemblages, increasing competition for limited resources at a time when zooplankton arecritical to sport-fish recruitment.

Predation often influences community structure(OBrien 1987). Aquatic systems tend to be lessstructurally complex than terrestrial ecosystems,affording prey fewer refugia than their terrestrialcounterparts (Murdoch and Bence 1987). As a re-sult, aquatic communities may be strongly me-diated by predation. Competition in aquaticcommunities also is common and may arise fromsize-structured interactions where similar-sizedindividuals of different species compete for a lim-ited resource (Stein et al. 1988). Predation by aparticularly numerous or efficient species may di-rectly influence the competitive interactions amongits prey or with competitors at its trophic level.Two species that interact as predator and prey asadults may be competitors as larvae or juveniles.These mixed predation-competition interactionsmay play an important role in aquatic commu-nities.

Gizzard shad Dorosoma cepedianum are abun-dant in many reservoirs throughout the south-eastern USA, often dominating the fish commu-nity (Cramer and Marzolf 1970; Noble 1981;Johnson et al. 1988). As preferred prey for manypiscivores (Carline et al. 1986; Johnson et al. 1988;Matthews et al. 1988; Wahl and Stein 1988), theyhave been stocked in attempts to improve pred-ator growth rates (Noble 1981). Historically, how-ever, little effort has been directed toward quan-tifying the role of gizzard shad in freshwatersystems and the possible direct and indirect effectsof this species on aquatic communities (but seeGuest et al. 1990; DeVries et al. 1991).

A review by DeVries and Stein (1990) suggestedthat shad Dorosoma spp. may not be ideal foragefishes. Gizzard shad can consistently produce largenumbers of offspring from few adults (Miller 1960;Pierce 1977), and their larvae may compete withother fishes for zooplankton (DeVries and Stein1992). Furthermore, because gizzard shad growquickly (Bodola 1966), they often reach a size ref-uge from most predators by the end of their firstyear (Adams and DeAngelis 1987; Johnson et al.1988). Impressive larval production coupled withfast growth limits predator consumption to a max-imum of 30% of gizzard shad production, at leastin Ohio reservoirs (Carline et al. 1984; Johnsonet al. 1988). Most importantly, however, gizzardshad are opportunistic omnivores, feeding on zoo-plankton as larvae (Barger and KJlambi 1980;DeVries et al. 1991), but capable of switching tophytoplankton or detritus as juveniles and adults(Kutkuhn 1958; Miller 1960; Bodola 1966; Pierceet al. 1981). Consequently, gizzard shad can drivezooplankton to extinction, yet still survive andgrow to adulthood.

Because gizzard shad are quite fecund and spawnbefore many sport fishes (e.g., bluegill Lepomismacrochirus), their larvae may deplete zooplank-ton resources to the extent that sport-fish larvaemay face unfavorable conditions for growth andsurvival. Whereas adult and juvenile fishes con-sume a wide range of prey types, most fishes arezooplanktivorous as larvae (Siefert 1972; Keast1980; Beard 1982; Rajasilta and Vuorinen 1983;Michaletz et al. 1987), creating the potential for

494

ZOOPLANKTIVORY BY LARVAL GIZZARD SHAD 495

exploitative competition if the single food re-source, zooplankton, is limiting. Such a severe on-togenetic and competitive bottleneck (sensu Wer-ner and Gilliam 1984) could profoundly affectrecruitment of all fishes with zooplanktivorouslarvae.

Because gizzard shad larvae are very abundantin several Ohio reservoirs, they may be directlyresponsible for midsummer zooplankton declines(DeVries and Stein 1992). Such direct trophic in-teractions can lead to complex effects on the entirecommunity (Carpenter et al. 1987), including in-creased phytoplankton blooms due to increasedplanktivory (Carpenter et al. 1985; Scavia et al.1988), as well as reduced forage for predators dueto poor survival among competing zooplankti-vores. Several studies (Carpenter et al. 1987; Millset al. 1987; Scavia et al. 1988; Hall and Ehlinger1989) have indicated that strong effects originat-ing from one trophic level can cascade throughthe community for several years. Thus, gizzardshad may influence the entire aquatic communityin ways fisheries scientists historically have notthought was possible.

To evaluate the effect that larval gizzard shadmight have on the crustacean zooplankton re-source, we quantified larval food consumption bydetermining gut contents and evacuation rates, andwe determined crustacean zooplankton produc-tion from field samples in two Ohio reservoirs. Bycalculating the consumptive demand of larval giz-zard shad and comparing these values with zoo-plankton productivity, we could evaluate whetherlarval gizzard shad predation contributes substan-tially to midsummer zooplankton declines in smallOhio reservoirs.

MethodsStudy sites.—Knox and Kokosing lakes (Knox

County) are small, relatively shallow reservoirs ineast-central Ohio. Knox Lake has a surface areaof 225 hectares, about 19.2 km of shoreline, anda maximum depth of 9.6 m; Secchi depths rangedfrom 47 to 106 cm during May-June 1988. Witha surface area of 65 hectares, Kokosing Lake has7.5 km of shoreline, and a maximum depth of 4.9m; Secchi depths ranged from 32 to 101 cm duringMay-July 1988. Both lakes are in largely agricul-tural areas; Knox Lake drains about 81 km2 andKokosing Lake about 87 km2. Neither lake strat-ified thermally, but Knox Lake became anoxic be-low 4.5 m after June 14. Fish communities in bothlakes consisted primarily of largemouth bass Mi-cropterus salmoides. gizzard shad, crappies Po-

moxis spp., bluegill, channel catfish let alumspunctatus, and brown bullhead Ameiurus nebu-losus.

Sampling methods. — We sampled larval fishweekly in both lakes for at least 6 weeks imme-diately after larval gizzard shad first appeared in1988. Larvae were collected by towing a 0.75-m-diameter, metered ichthyoplankton net (500-Mmmesh) along the surface at 1.0-1.5 nvs~ l for 3-5min. Larvae were preserved immediately in 10%formalin, returned to the laboratory, and identi-fied, counted, and measured (up to 50/species) tothe nearest millimeter in total length (TL). Foreach sample date, we estimated larval density,mean larval size, diet composition, food con-sumption, and gastric evacuation rate. Density oflarvae used to compute larval food consumptionwas always calculated as the mean of all samplestaken between 2200 and 2400 hours becausenighttime larval densities were always at least twicedaytime estimates. To estimate food consump-tion, larvae were sampled every 1.5 h from sunriseto sunset; at sunset, we sampled larvae every 15min for at least 3 h to quantify evacuation rates.Because larval gizzard shad did not feed after dark,their evacuation rate could be quantified by mon-itoring gut contents beginning immediately aftersunset.

We calculated gastric evacuation rates and foodconsumption of larval and early juvenile gizzardshad (5-30 mm TL) with a model developed byElliott and Persson (1978) and used for severalfishes (Elliott and Persson 1978; Persson 1979,1982; Cochran and Adelman 1982). This modelrequires assumptions that (1) food consumptionis constant during the interval between samples,and (2) evacuation rate is exponential. Thus, rateof change in gut contents can be expressed by thedifferential equation

dS/dt = F - RS\ (1)S = weight of stomach contents (g stomach

contents-g body weight'1);t = time(h);

F = rate of food consumption (g-g"' • h ~ ') ;R = evacuation rate (g-g- ' -h 1 ) .

Using the above equation, we could solve for foodamounts in stomachs at time /:

5, = So*'*' + (F/R) (1 - e Rl). (2)However, when no food is consumed over a timeinterval, the above equation reduces to

S, = So*-*'. (3)

496 DETTMERS AND STEIN

Thus, we solved for R with equation (3) after wedetermined the stomach contents at consecutivesample times as well as the time interval betweensamples. To determine total food consumption (Q(measured in mg-g-'-d"1), we rearranged equa-tion (2) to solve for F, the hourly rate of foodconsumption

F=R(St - Soe-Rt)/(l - e-*). (4)Then, we solved for total food consumption withthe equation

C • Ft. (5)To estimate the weight of gut contents for larval

gizzard shad, we used a technique developed byPost and McQueen (1988), by which the empty-gut weight is subtracted from the combined weightof the gut contents and gut. To accomplish this,we first regressed empty-gut weight against bodyweight. Empty-gut weight was determined by re-moving and weighing entire guts from fish col-lected at night after evacuation was complete. Bodyweight was measured by blotting excess water froman individual fish and then immediately weighingit (nearest 0. 1 ing). To determine food biomass inlarval guts, we dissected 10 larvae for each sampletime. Because the gut at this developmental stageis a simple tube, we weighed the contents of theentire digestive tract from esophagus to anus. Wealso quantified diet composition of larval gizzardshad (N = 5 fish/sample time). Crustacean zoo-plankton were identified to genus; rotifers werelumped.

To evaluate prey selection, we compared larvalgizzard shad diets with zooplankton samples usingChesson's alpha (Chesson 1978, 1983). The for-mula for this index is

r/ = the proportion of prey item / in the fish'sdiet;

Pi = the proportion of prey item / in the en-vironment.

For each zooplankton taxon, we compared ourcalculated a (±95% confidence interval, CI) to theexpected a if prey were eaten in proportion to theiravailability. Thus, a prey taxon was selected if awas greater than the reciprocal of the number ofprey types in the environment (Chesson 1978,1983).

Zooplankton were sampled twice weekly in bothlakes. One sample was taken simultaneously withlarval gizzard shad collections; a second sample

was collected 3-5 d later. Each zooplankton sam-ple included two replicate day hauls and two rep-licate night hauls on dates when larval fish werecollected. Crustacean zooplankton and rotiferswere sampled with a 0.3-m-diameter plankton net(70-Aim mesh) towed vertically through the entirewater column and preserved in 5% sucrose for-malin (Haney and Hall 1973). We counted all in-dividuals in rare taxa (<200 individuals per sam-ple); for abundant taxa, 5% subsamples werecounted until at least 200 individuals had beencounted. Up to 50 individuals of each taxon in asample were measured (nearest 0.01 mm) with adigitizing tablet viewed through a drawing tubeattached to a microscope. These lengths were con-verted to biomass with taxon-specific, length-dryweight regressions (Culver et al. 1985).

We evaluated crustacean zooplankton produc-tion (P) by calculating the population biomass (B)for each taxon and estimating P, following themethods of Bean (1980). We determined temper-ature-specific P/B ratios by regressing P/B ratioon temperature for taxa of interest, using data fromLake Erie (D. A. Culver, Ohio State University,personal communication). We then multiplied theexisting biomass of each taxon on each date bythe temperature-dependent P/B ratio for those taxato determine taxon-specific production. We sim-ply added the production of each taxon to deter-mine total crustacean zooplankton production foreach date. With this technique, we assumed thattaxon-specific P/B ratios vary slightly across arange of trophic states (Plante and Downing 1989).In addition, crustacean zooplankton productionwas calculated explicitly (sensu Culver and DeMott1978) on three dates in each lake to evaluate theaccuracy of estimating P/B ratios.

To determine if larval gizzard shad substan-tially reduced crustacean zooplankton biomass, weused a simple mass-balance equation:

Bk - kC\ (6)

Bk = crustacean zooplankton biomass at theend of the interval of k days (mg-rn 3,wet weight);

BQ = crustacean zooplankton biomass at thebeginning of each interval (mg-m'3, wetweight);

k = number of days in each interval (N = 3or 4);

P = crustacean zooplankton production at thebeginning of the interval (mg-m-'-d'1,wet weight);

C = larval gizzard shad consumption of crus-

ZOOPLANKT1VORY BY LARVAL GIZZARD SHAD 497

5 12JULY

DATEFIGURE 1.—Density of larval gizzard shad and of crus-

tacean zooplankton biomass (dry weight) in (A) Koko-sing Lake and (B) Knox Lake in Ohio, May-July 1988.Note different ordinate scales, reflecting a 5-fold greaterdensity of larval gizzard shad and up to a 10-fold greatercrustacean zooplankton biomass in Kokosing Lake. Dataare means ± SEs.

tacean zooplankton, calculated weekly-^d-1, wet weight).

With this equation we could determine if ob-served larval gizzard shad densities accounted forthe observed decline in crustacean zooplankton,as well as predict the density of larvae or juvenilesneeded to account for the observed zooplanktondecline. When larval gizzard shad consumed preyother than crustacean zooplankton, we estimatedcrustacean zooplankton consumption by subtract-ing the weight of zooplankton, determined bylength-dry weight regression, from the total weightof food in the gut. Because biomass of zooplank-ton consumed was expressed as wet weight, andresource estimates of zooplankton biomass wereexpressed as dry weight, we converted dry weightto wet weight with a 1 : 1 0 (dry : wet) ratio (Bal vay1987).

ResultsLarval gizzard shad were first collected in Ko-

kosing and Knox lakes in late May; larval densi-

3.0

2.0

• KOKOSING LAKE* KNOX LAKE

28MAY

11 18JUNE

25 2JULY

FIGURE 2.—Growth of larval gizzard shad in Koko-sing and Knox lakes, Ohio, May-July 1988. Ordinate islog,, (length).

ties peaked in both lakes in mid-June (Figure I).Nighttime densities were greater than daytimedensities on each sampling date in both lakes (re-peated-measures analysis of variance: F = 8.05;df = 1, 5; /> = 0.04 in Kokosing Lake and F =7.54; df = 1, 5; P = 0.04 in Knox Lake); thus, weused nighttime density as an estimate of larvalabundance. Peak density differed between lakes:38 ± 0.62 fish-nr3 (mean ± SE) in KokosingLake as compared with 7 ± 0.28 fish-m-3 in KnoxLake (two sample /-test: P = 0.0002). Larval den-sities declined rapidly after the peak. This declinelikely was a combination of mortality and gearavoidance, larvae larger than 20 mm TL avoidingthe net.

Larval growth differed between lakes (analysisof covariance: homogeneity of intercept, P = 0.017;homogeneity of slope, P = 0.032; Figure 2). Ko-kosing Lake larvae were larger (11.44 ± 0.34 mm,mean ± SE) than Knox Lake larvae (5.16 ± 0.64mm) when first sampled on May 28, suggestingearlier spawning by gizzard shad in Kokosing Lake.Kokosing Lake larvae were always at least 1.5 mm,but at most 7 mm, larger than Knox Lake larvaeduring late May through early July; however, Ko-kosing Lake larvae grew at only half the rate ofKnox Lake larvae.

Crustacean zooplankton composition andabundance also differed markedly between lakes.Whereas Bosmina spp. dominated Kokosing Lake,copepods were dominant in Knox Lake (Figure3). Crustacean zooplankton abundance increasedquickly and peaked at the end of May in KokosingLake, declining to near zero by June 21 (Figure1A). Crustacean zooplankton in Knox Lake didnot exhibit the typical spring peak in biomass(Figure IB); instead, a bloom of Diaphanosoma

498 DETTMERS AND STEIN

UJQQ

900

400

900

200

100

too

75

50

25

Daphnia A. KOKOSIN6 LAKE

B. KNOX LAKE

Diaphonosoma -

31 7 14 21MAY JUNE

28 5JULY

FIGURE 3.—Crustacean zooplankton abundance andcomposition in (A) Kokosing Lake and (B) Knox Lakein Ohio, May-July 1988. Abbreviations for taxa: eye —cyclopoid copepods; cal = calanoid copepods.

spp. caused zooplankton biomass to peak near theend of June, although copepod nauplii numbersalso increased dramatically (Figure 3B).

Macrozooplankton biomass in Kokosing Lake(652 ± 84 [SE] mg-rn-3) exceeded that in KnoxLake (3.55 ± 0.94 mg-m-3) on May 31, beforelarval gizzard shad were near peak abundance (twosample /-test, P = 0.041). Crustacean zooplanktonproduction, estimated by PIB ratios, peaked inKokosing Lake at the end of May. Production fellsteadily thereafter, reaching a minimum on June23. Crustacean zooplankton production in KnoxLake remained low until June 26. Production in-creased after June 28 due to a Diaphanosomabloom. These production estimates were com-pared with explicitly calculated production values(as per the methods of Culver and DeMott 1978)on May 31, June 3, and June 19 (Table 1). In allcases, calculated production values did not differfrom values estimated by PIB ratios (paired /-test,P > 0.30).

Because food consumption and diet composi-tion changed as larvae grew, at least in KokosingLake, we analyzed diet composition and food con-sumption for small (5-17 mm TL) and large (18-24 mm TL) larvae. Diet composition and esti-

TABLE 1.—Comparison of calculated versus estimatedcrustacean zooplankton production (P) in Kokosing andKnox lakes, Ohio, May 31, June 2, and June 19, 1988.B is biomass.

Date

May 31Jun 2Jun 19

May 31Jun 2Jun 19

P estimatedfrom

PIB ratios(mg-m 3-d" J)Kokosing Lake

131.7181.50

2.92Knox Lake

0.290.480.34

P calculatedfrom zoo-

plankton data(mg-m~3-d ')

188.4386.02

2.87

0.310.450.28

mated food consumption were quantified for earlyjuveniles (25-30 mm TL) in Kokosing Lake only.In Kokosing Lake, small larval gizzard shad pri-marily consumed cyclopoid copepodites and nau-plii on or before June 9 (Table 2), when larvaereached their peak abundance. After June 9, lar-vae consumed smaller proportions of cyclopoids(Mest, P < 0.0001), Daphnia (P = 0.0031), andnauplii (P < 0.0001) but a larger proportion ofrotifers (P < 0.0001). Large larvae primarily con-sumed cyclopoid copepodites, Bosmina. andDaphnia on or before June 9 (Table 2); however,these large larvae ate smaller proportions of cy-clopoids (P = 0.015), Bosmina (P = 0.028), andDaphnia (P = 0.036), but more rotifers (P <0.0001) after June 9. Before peak larval densitieson June 9, large larvae consumed greater propor-tions of Bosmina (P = 0.012) and Daphnia (P =0.04), but a smaller proportion of nauplii (P <0.0001) than smaller larvae. After the larval peak,large larvae ate proportionately more Bosminathan small larvae (P =* 0.05). By July 1, all larvaefed entirely on rotifers and detritus in the absenceof crustacean zooplankton.

Prey selection by small larval gizzard shad, de-termined via Chesson's alpha (Chesson 1978,1983), reinforces this dietary pattern. Small larvaeselected (1) copepod nauplii and cyclopoid cope-podites before crustacean zooplankton popula-tions crashed (May 28-June 2; Table 3), (2) cy-clopoid copepodites during the crash (June 9-16),and (3) no crustacean zooplankton after the crash(June 23). Large larvae preferred (1) cyclopoid co-pepodites before the crash, (2) Bosmina andDaphnia during the crash, and (3) no crustaceanzooplankton after the crash. Rotifer electivitieswere not calculated because we did not accuratelyestimate rotifer density.

ZOOPLANKTIVORY BY LARVAL GIZZARD SHAD 499

TABLE 2.—Diet composition of larval gizzard shad (N = 25 small and 15 large larvae per date) described as theproportion by number (mean ± SE) of each zooplankton taxon consumed on each date in Kokosing Lake, Ohio,May-July 1988. Calanoid copepods, although present, contributed proportions of less than 0.01 to diets of larvalgizzard shad. Proportions ofDaphnia spp. were important only on June 9, when they were 0.31 ± 0.12 in the dietsof large larvae. TL is total length.

Dietary proportions of:

Date

May 28J u n 2J u n 9Jun 16Jun23

Jun 2Jun 9Jun 16Jun 23

Total preycounted

8421,1451.2161.3472.630

2.6903.429

10.3786,819

Cyclopoidcopepodites Bosmina spp.

Small larvae (5-17 mm TL)0.60 ± 0.03 0.0 1 ± 0.0 10.62 ± 0.06 0.06 ± 0.020.51 ± 0.05 0.03 ± 0.010.08 ± 0.04 0.03 ± 0.02

<0.01 0.02 ± 0.01Large larvae (18-24 mm TL)

0.69 ± 0.14 0.25 ± 0.120.22 ± 0.09 0.39 ±0.130.05 ± 0.02 0.16 ± 0.05

0 <0.01

Copcpodnauplii

0.34 ± 0.030.27 ± 0.060.25 ± 0.040.04 ± 0.02

<0.01

<0.010.01 ± 0.010.04 ± 0.02

0

Rotifers

0.04 ± 0.0 10.03 ± 0.010.20 ± 0.060.85 ± 0.060.98 ± 0.01

0.06 ± 0.030.06 ± 0.040.75 ± 0.080.99 ± 0.01

Diets did not differ between small and large lar-vae in Knox Lake (Mest, P > 0.43). Larval dietsin Knox Lake included up to 32% crustacean zoo-plankton, and the remainder was composed of ro-tifers (Table 4), most likely as a result of low crus-tacean zooplankton biomass (Figure 1).

Larval gizzard shad in Knox Lake ate crusta-cean zooplankton in proportion to their avail-ability during May through July (Table 4). Larvaenever selected Diaphanosoma, although it wasabundant during the zooplankton peak in late June(Figure 3).

TABLE 3.—Prey selection by larval gizzard shad (N =25 small and 15 large larvae per date) described by Ches-son's alpha (mean ± SE) for each zooplankton taxonpresent in larval diets in Kokosing Lake, Ohio, May-July 1988. Values of alpha greater than 0.17 (the recip-rocal of the number of taxa present in the lake) indicatepositive selection; smaller values indicate nonpreferredtaxa. Calanoid copepods were never selected by larvalgizzard shad; Daphnia spp. were preferred only on June9, when large larvae strongly selected them (alpha = 0.69± 0.09). TL is total length.

DaleCyclopoid

copepodites Bosmina spp.Copepodnauplii

Small larvae (5-17 mm TL)May 28Jun 2Jun 9Jun 16Jun 23

0.69 ± 0.030.77 ± 0.050.81 ± 0.030.20 ± 0.070.05 ± 0.05

0.01 ± 0.0040.01 ± 0.010.01 ± 0.0030.07 ± 0.030.23 ± 0.09

0.30 ± 0.030.18 ± 0.050.16 ± 0.030.05 ± 0.030.03 ± 0.03

Large larvae (18-24 nun TL)Jun 2Jun 9Jun 16Jun 23

0.93 ± 0.040.23 ± 0.060.28 ± 0.08

0

0.06 ± 0.040.08 ± 0.050.30 ± 0.04

<0.01

<0.01<0.01

0.09 ± 0.040

Food consumption was calculated by couplingthe amount of food in larval guts through timewith larval evacuation rates. Based on gut analy-ses, larvae fed continuously at a constant rate dur-ing daylight (regression analysis: slope no differentfrom zero, F-test; 1, 4 df; P > 0.24 in all cases),but never fed after sunset (Figures 4, 5). Becausegut evacuation is strongly temperature dependent,we first calculated six larval evacuation rates acrossthe range of T = 2G-28°C. We then regressed evac-uation rate (mg-g'^h-1) against temperature (R= -21,024 + 1,019 T\ r2 = 0.98; F= 209.01; df= 1, 4; P < 0.001) to determine evacuation ratesfor June-July 1988.

Mean daily food consumption of individual lar-val gizzard shad was similar in Kokosing and Knoxlakes for a given temperature. Generally, individ-ual consumption was lower in Kokosing Lake dueto cooler water temperatures. Food consumptionby the entire larval population increased throughJune 9 and declined thereafter in Kokosing Lake(Table 5). In Knox Lake, the larval gizzard shadpopulation consumed an order of magnitude morefood in late June than in late May because of in-creased larval density and size. Although individ-ual food consumption was similar between lakes,the population of larval gizzard shad consumedmore food in Kokosing Lake than in Knox Lakesimply because they were more abundant.

In Kokosing Lake, larval gizzard shad con-sumed more than 100% of the crustacean zoo-plankton production only after zooplankton hadcrashed. Larval gizzard shad in Knox Lake con-sumed at least 80% of daily crustacean zooplank-ton production except on June 7, when larvae con-

500 DETTMERS AND STEIN

TABLE 4.—Diet composition and prey selection (mean ± SE) of larval gizzard shad ()V = 25 fish/date) in KnoxLake, Ohio, May-July 1988. Diet composition is presented as the proportion by number of each zooplankton taxonin larval guts on each date, whereas prey selection is described by Chesson's alpha. To evaluate electivity, valuesgreater than 0.17 (the reciprocal of the number of taxa present in the lake) indicate positive selection; smaller valuesindicate nonpreferred taxa. Calanoid copepods, Bosmina spp., and Diaphanosoma spp., although present, neveraccounted for dietary proportions greater than 0.0 J and were never preferred by larval gizzard shad. Electivitiesfor rotifers were not calculated because they were not efficiently sampled.

Proportion or electivity of:

Date

May 31J u n 7Jun 14J u n 2 lJ u l S

May 31Jun 7Jun 14Jun 21Jul 5

Total preycounted

70147212

2,37414.189

70147212

2,37414,189

Cyclopoidcopcpodites

Diet composition0.11 ± 0.060.04 ± 0.020.04 ± 0.030.04 ± 0.020.02 ± 0.004

Prey selection0.18 ± 0.090.24 ±0.130.22 ±0.140.21 ± 0.120.23 ± 0.03

Copepodnauplii

0.21 ± 0.080.12 ± 0.05

0<0.01

0.0 1 ± 0.004

0.26 ±0.100.25 ±0.12

0<0.01

0.02 ± 0.01

Rotifers

0.680.840.960.950.96

± 0.10± 0.05± 0.03± 0.06± 0.01

sumed 55% of crustacean zooplankton production,and on July 5, when larval densities were extreme-ly low. Larvae consumed more than 450% of crus-tacean zooplankton production on June 14, whenlarval densities peaked.

Gizzard shad larvae do not exert a strong influ-ence on crustacean zooplankton in Kokosing Lake;however, larvae may be an important source ofmortality in Knox Lake. To fully appreciate theirinfluence, we compared larval food consumptionwith the sum of crustacean zooplankton biomassand production (Table 6). Existing densities of lar-val gizzard shad could not account for observedcrustacean zooplankton decline in Kokosing Lake;only unreasonably high larval densities (60-212larvae-m-3) could be expected to consume enoughzooplankton to cause the observed decline.

However, juvenile gizzard shad occasionallywere captured in ichthyoplankton tows, first ap-pearing on June 2. Enough juveniles were collect-ed to determine diet composition and amount offood in the digestive system during daylight hours.Juvenile diets were similar to those of large larvaeduring June 2-16 (two-sample /-test, P > 0.14).Assuming that evacuation rates for juvenile giz-zard shad 25-30 mm TL declined by 15% as com-pared with larval evacuation rates (Mills and For-ney 1981), we estimated individual juvenile foodconsumption (Table 5) and determined the num-ber, of juvenile gizzard shad required to cause theobserved zooplankton decline. Fitting the ob-served zooplankton decline to estimated con-

sumption by larvae plus juveniles, we calculatedthat 1.3-17.8 juveniles-m~3 would account for thecrustacean zooplankton declines in Kokosing Lake(Table 6). These are not unreasonable juveniledensities for Kokosing Lake because Johnson etal. (1988) quantified mean densities at 12.8 ju-veniles-m~3 over the growing season and a max-imum density of 20 juveniles-m"3. Although lar-val consumption alone was not responsible for thecrustacean zooplankton decline in Kokosing Lake,additional consumption by early juveniles, com-bined with larval consumption, likely drove zoo-plankton to low levels.

In Knox Lake, gizzard shad larvae had variableeffects on zooplankton biomass (Table 6). Duringthe week of May 31, larval food consumption couldhave accounted for zooplankton changes, but itdid not prevent zooplankton biomass from in-creasing during the week of June 7. From June 14to 22, existing larval densities were more thansufficient to account for changes in zooplanktonbiomass. Zooplankton biomass increased fromJune 22 to 28 despite larval gizzard shad preda-tion. After June 28, unreasonably high larval andjuvenile densities were required to account for theobserved crustacean zooplankton decline.

To summarize, high densities of larvae couldnot account for the decline in zooplankton whencrustacean zooplankton production and biomasswere great in Kokosing Lake. Only consumptionby juvenile gizzard shad could account for the ob-served zooplankton decline. However, zooplank-

ZOOPLANKTIVORY BY LARVAL GIZZARD SHAD 501

70

503010

^ 60

0> 40

200*E

o

C/)Ul

amaoo

60

40

20

0

60

40

20

0

60

40

20

UAY28

JUNE 2

JUNE 16

JUNE 23

1200150016002100000 300 600 900

TIME (h)FIGURE 4.—Diel patterns of gut fullness (means ±

SEs) for gizzard shad larvae in Kokosing Lake, Ohio,May-June 1988. Bars along the abscissa represent darkperiods.

I

COCO

oCDQOO

60

40

20

0

60

40

E 20V^X 0

D 60

40

20

0

60

40

20

0

60

40

20

MAY 31

HJUNE?

id.JUNE 14

HJUNE 21

JULYS

I—i—i—W.1200150018002100000 300 600 900

TIME (h)FIGURE 5.—Diel patterns of gut fullness (means ±

SEs) for gizzard shad larvae in Knox Lake, Ohio, May-July 1988. Bars along the abscissa represent dark peri-ods.

ton consumption when larval densities were lowwas often sufficient to account for the zooplanktondynamics in Knox Lake, where production andbiomass were low during May-June 1988.

DiscussionLarval Food Consumption and Diets

To sustain their metabolic and growth require-ments, larval gizzard shad feed continuously atconstant ration levels throughout daylight hoursduring late May through early July. With contin-uous and constant feeding, larval gizzard shad wereideal animals with which to quantify food con-sumption via the Elliott-Persson model (Elliottand Persson 1978).

Gizzard shad larvae do not feed at night; thus,

larval gizzard shad are likely visual, paniculatefeeders, unlike juveniles and adults (Drenner et al.1982a, 1982b). By measuring the decline in gutcontents through time after dark, we could deter-mine evacuation rates in the field. Rarely canevacuation be directly determined from field col-lections; however, this was a distinct advantagebecause larval gizzard shad are difficult to collectand manipulate experimentally in the laboratory.

Larval gizzard shad ate crustacean zooplanktonwhen it was abundant during late May-early July.When crustacean zooplankton became scarce, lar-vae fed on rotifers. Small larvae selected copepodnauplii and copepodites when crustacean zoo-plankton were abundant. Although large larvaepreferred copepodites, they ate few nauplii andconsumed more cladocerans than small larvae.

502 DETTMERS AND STEIN

TABLE 5.—Calculated food consumption by individualsmall gizzard shad larvae (5-17 mm total length; N =60 fish/date) and large larvae (18-24 mm; N = 30 fish/date), estimated food consumption by individual juve-nile gizzard shad (25-30 mm), and estimated total con-sumption by the population of larval gizzard shad in

TABLE 6.—Comparison of observed crustacean zoo-plankton biomass (see Figure 1) and production with thepredicted number of larval and early juvenile gizzardshad required to account for the observed changes incrustacean zooplankton abundance in Kokosing andKnox lakes, Ohio, May-July 1988.

»Individual consumption

( m g . g ' d »)by: Populationrr»nciiT«p1ion

Small Large Juve- (mg m 3-d !),Date larvae larvae niles all larvae

Kokosing Lake*May 29 2,383 NP NP 146Jun2 736 994 728 158Jun9 358 540 672 263Jun 16 693 831 1,520 115Jun23 1,510 1,743 NCb 37

Knox Lake8

May 31 2,995 NP NP 13Jun 7 656 NP NP 11Jun 14 2.777 NP NP 133Jun 21 3,351 3,753 NP 189Ju l5 851 978 NP 32

' NP = larval and juvenile fish were not present or not collected.b NC - juveniles were not consuming macrozooplankton.

Diaphanosoma composed the bulk of zooplank-ton in Knox Lake after June 28, yet larvae did notselect them. Nevertheless, fish larvae do preferDiaphanosoma (Van Den Avyle and Wilson 1 980;Mallin et al. 1987; DeVries et al. 1991), which isnot surprising because Diaphanosoma is larger thanmost zooplankters in these reservoirs, has no de-fensive spines, and has poorer escape abilities thancopepods (Drenner et al. 1978; Drenner andMcComas 1980). Diaphanosoma likely was notselected because deep Knox Lake (maximumdepth, 10 m) provided a spatial refuge for thisvertically migrating zooplankter (Carter et al. 1 980;Nero and Sprules 1986).

Based on electivity measures, larvae selectedprimarily copepods; however, electivity did notalways reflect those prey most common in the diet.Large larvae in Kokosing Lake selected Bosminaon only one date; yet more than 25% of their dietwas Bosmina on dates when the cladoceran wasnot preferred. Caution must be used when elec-tivities are interpreted; if preferred prey are rare,they may only contribute slightly to the diet. How-ever, prey that are simply consumed in proportionto their abundance may be an extremely impor-tant energy source if they form a substantial pro-portion of the prey community. In food con-sumption studies, actual diet composition is morerelevant than diet preference simply because con-

Larvae Juve-re- niles

Zooplankton Existing quired8 re-

(mg-irr3- density ber- (number-Date d~ ' ,wet) (number - m 3 ) m 3) m 3)

Kokosing LakeMay 28-31 815 ± 92 17.5 ± 1.7 BI NPMay 31-

Jun2 1.317 ±97 17.5 ± 1.7 211.8 17.8Jun 2-5 806 ± 81 20 ± 1.7 180.5 13.9Jun 5-9 803 ± 83 20 ± 1.7 155.8 11.7Jun 9-12 460 ± 126 38 ± 2.1 91.8 3.8Jun 12-16 253 ± 61 38 ± 2.1 59.8 1.5Jun 16-19 130 ± 15 11 ± 0.6 122.0 5.2Jun 19-23 29 ± 3.5 1 1 ± 0.6 37.9 1.3

Knox LakeMay 31-

Jun 2 3 ± 0.8 2 ± 0.2 2 NPJun 2-7 4 ± 1.5 2 ± 0.2 2 NPJun 7-1 2 4 ± 1.8 3 0.1 5.6 NPJun 12-15 5 ± 2.1 3 0.1 29.7 NPJun 15-19 2 ± 0.5 7 0.3 1.0 NPJun 19-22 3 ± 0.6 7 0.3 7.0 NPJun 22-26 9 ± 2.9 5 0.1 BI NPJun 26-28 10 ± 2.2 5 0.1 BI NPJun 28-

Jul5 41 ± 6.7 2 0.1 16.3 27.4Jul5-9 29 ± 5.9 1 0.1 85.2 47.6a BI ™ zooplankton biomass increased during the interval despite

predation by larval gizzard shad.b NP = juveniles were not present in the lake.

sumption estimates are based on diets, not on foodchoice.

All zooplankton taxa, with the exception ofDiaphanosoma in Knox Lake, were vulnerable togizzard shad during at least part of that species9

larval development. Because zooplankton werevulnerable, larval gizzard shad have the potentialto strongly influence plankton community dy-namics if their consumption exceeds crustaceanzooplankton production. Based on comparisonsbetween larval food consumption and the sum ofcrustacean zooplankton biomass plus productionduring late May through early July, we concludethat gizzard shad larvae at low densities likelyconsumed enough crustacean zooplankton in KnoxLake, where zooplankton biomass and productionwere low, to prevent crustacean zooplankton fromincreasing until after June 2 1 . Conversely, in Ko-kosing Lake, where zooplankton biomass and pro-duction were high, even high densities of larval

ZOOPLANKTIVORY BY LARVAL GIZZARD SHAD 503

gizzard shad had only a minor effect on zooplank-ton assemblages. Instead, although the source ofthe initial decline in crustacean zooplankton is un-clear, once zooplankton began to decrease, zoo-plankton consumption by early juvenile gizzardshad may have accounted for the observed declinein zooplankton.

Low densities of larval gizzard shad in KnoxLake had a greater effect on zooplankton than didhigh densities in Kokosing Lake for two reasons.First, zooplankton production and biomass werethree orders of magnitude lower in Knox Lake onMay 31, whereas peak larval densities were 20%of those in Kokosing Lake. Thus, Knox Lake lar-vae initially exerted a larger predatory pressure ontheir zooplankton resource than did Kokosing Lakelarvae.

Second, zooplankton composition differed be-tween lakes; Kokosing Lake was rich in cladoc-erans whereas Knox Lake was dominated by co-pepods. Given differences in turnover rate betweenthe two taxa and that small larvae prefer copepodsover cladocerans, this difference is important.Generation times are 3-5 d for cladocerans (Gu-lyas 1980) but 2-3 weeks for copepods (Gulyas1980; Webb and Parsons 1988) across our ob-served temperature range. Thus, the copepod-dominated zooplankton assemblage in Knox Lakehad more difficulty maintaining itself in the faceof mortality (e.g., predation) than the cladoceran-dominated assemblage of Kokosing Lake. In ad-dition, the entire zooplankton assemblage in KnoxLake was not only available to, but was also pre-ferred by, larval gizzard shad. Thus, Knox Lakelarvae could maintain predatory pressure suffi-cient to prevent zooplankton accumulation untilDiaphanosoma appeared.

Although gizzard shad larvae were quite abun-dant in Kokosing Lake, their effect on zooplank-ton was insignificant because they primarily atecopepodites and nauplii. Copepods were respon-sible for less than 6% of crustacean zooplanktonproduction; thus, although larvae most likely re-duced copepod abundance (from 20% of zoo-plankton on May 28 to 6% on June 9), overallproduction and biomass did not decline until largelarvae and juveniles began consuming cladoceranson June 2.

Gizzard shad larvae may select nauplii and co-pepodites in Kokosing Lake because of gape lim-itations, even though the crustacean zooplanktonassemblage was dominated by cladocerans (53-80% of the zooplankton during May 28-June 9).Gape size influences diet and perhaps success of

larval fishes (Hunter and Kimbrell 1980; Da-browski et al. 1984; Michaletz et al. 1987). Forexample, larval pollan Coregonus pollan feed onnauplii and copepodids until they reach 20 mm,when they switch to larger zooplankters, includingadult copepods and Daphnia (Dabrowski et al.1984). Maximum prey size consumed by pollansmaller than 13 mm did not differ from mean preysize. Conversely, larval pollan larger than 13 mmate prey smaller than their gape size would dictate,suggesting that pollan were no longer gape-limit-ed. Increasing gape size permitted larval pollan toshift from small to large prey.

Similarly, when feeding on crustacean zoo-plankton, larval gizzard shad 5-17 mm TL fromKokosing Lake predominately consumed cope-pods (nauplii and copepodites) during May 28-June 9, whereas larvae 18-24 mm TL primarilyconsumed cladocerans during June 9-16. In anassessment of the importance of zooplankton sizeto survival of larval fishes, M. Bremigan (OhioState University, personal communication) deter-mined gape size of 17-mm (TL) larval gizzard shadto be 0.45 mm in diameter. If larval gizzard shadeat prey 26% of their gape (as 12-mm larval pollando: Dabrowski et al. 1984), a 17-mm larva couldconsume zooplankters only 0.12 mm in the lim-iting dimension. Presumably, smaller larval giz-zard shad would have an even more stringent preysize restriction. In Kokosing Lake on June 9, preyhad the following mean widths: copepodites, 0.15mm; nauplii, 0.10 mm; Bosmina, 0.22 mm; Daph-nia, 0.33 mm. Thus, only copepodites and naupliihave size distributions that put large proportionsof their populations at risk from larval gizzardshad predation. Hence, because of interactions be-tween gape size and the zooplankton assemblage,gizzard shad larvae may strongly influence plank-ton dynamics in copepod-dominated Knox Lake,but not in cladoceran-dominated Kokosing Lake.

Larval gizzard shad had less effect than juve-niles in Kokosing Lake simply because their con-sumptive demand was so much smaller, owing toallometric growth patterns (Peters 1983). As lar-vae, gizzard shad grow primarily in length. Atmetamorphosis (about 25 mm TL), however, in-dividuals add considerably more weight per unitof length added. For each 20% length increase aslarvae, weight increases by a factor of 1.8; as lar-vae metamorphose, weight increases by a factorof 5 for a similar length increase (J. M. Dettmers,unpublished data). Because food consumption isdirectly related to weight, consumptive demandincreases dramatically as larvae become early ju-

504 DETTMERS AND STEIN

veniles. Thus, consumption by juveniles may haveoverwhelmed the reproductive capacity of thecrustacean zooplankton community, driving zoo-plankton biomass to low levels.

This interpretation agrees with that presentedby DeVries and Stein (1992), whose experimentcontained fishless exclosures and gizzard shad en-closures. Stocked into bags at 18 mm and growingto 40 mm during the 4-week experiment, gizzardshad drove zooplankton to extinction. Converse-ly, in fishless exclosures, zooplankton maintainedhigh densities. Our estimates of zooplankton con-sumption by gizzard shad in the same lake andyear, combined with this experiment, stronglysuggest that larval, and especially juvenile, gizzardshad are a major source of mortality contributingto the observed decline of crustacean zooplank-ton.

Although consumption by larval and juvenilegizzard shad may explain crustacean zooplanktondynamics during May-July in both study lakes,other factors likely play important roles in regu-lating zooplankton dynamics. In Kokosing Lake,the initial phase of the zooplankton decline (May31-June 5) cannot be fully attributed to gizzardshad. Larval densities were probably too low toproduce the densities of juveniles needed to ac-count for the decline at the end of May. A possibleexplanation for the crustacean zooplankton de-cline is consumption by other planktivorous fishes(e.g., adult gizzard shad, bluegills, and crappies).This scenario is difficult to evaluate because wedid not determine the diet composition of adultgizzard shad and abundance of bluegill and crap-pie was low (personal observation).

Alternatively, the initial crustacean zooplank-ton decline may have been due to bottom-up ef-fects mediated by nutrients (McQueen et al. 1986).If the algal composition changed to favor inedibletypes, zooplankton could not graze effectively, re-sulting in reduced reproductive rates and biomass(Lampert 1978; Lampert et al. 1986). Once zoo-plankton reproductive rates were reduced, the im-mense predatory pressure by early juvenile giz-zard shad likely could facilitate the collapse of thecrustacean zooplankton population.

In Knox Lake, the late-season peak and declineof crustacean zooplankton also is not explainedby young-of-year gizzard shad predation. At thetime of the zooplankton peak, larval gizzard shadconsumption could not accommodate daily zoo-plankton production; as a result, zooplankton bio-mass should have increased. However, the mag-nitude of the increase was surprising. The peak

and decline of zooplankton biomass appeared tobe largely the result of Diaphanosoma populationdynamics. The mechanism responsible for thisdramatic increase, followed by a rapid decline, isunclear. Likely explanation include a hatch ofresting eggs, a change in algal composition, changein some physical factor (e.g., temperature or flowrate), and increased bacterial production (DeMott1989).Gizzard Shad and Reservoir CommunityDynamics

Early life stages of gizzard shad may stronglyinfluence crustacean zooplankton assemblages inOhio reservoirs. By consuming zooplankton, giz-zard shad may drive many interactions within thecommunity. For example, gizzard shad likely in-fluence recruitment of other fishes (Guest et al.1990; DeVries et al. 1991). By dramatically re-ducing zooplankton in spring, gizzard shad larvaeand early juveniles may create a competitive bot-tleneck through which other fishes, whose young-of-year rely on zooplankton, must recruit. Mostyoung-of-year fishes common to Ohio reservoirsare zooplanktivorous (Siefert 1972; Michaletz etal. 1987; DeVries 1989), increasing the likelihoodof such bottlenecks. Severe bottlenecks across sev-eral years likely would reduce recruitment ofplanktivorous sport fishes (e.g., bluegill and crap-pies), conceivably leading to indirect effects oncommunity dynamics.

Gizzard shad also may influence lower trophiclevels. By substantially reducing zooplankton, lar-val and early juvenile gizzard shad contribute tototal algal biomass by removing crustacean graz-ers (Carpenter et al. 1987; McQueen et al. 1989)and providing nutrients through remineralization(Lazzaro et al., in press). Thus, gizzard shad maycompromise water quality by increasing algal bio-mass.

The degree to which gizzard shad may influencecommunity dynamics in Ohio reservoirs requiresadditional field tests. However, larval and es-pecially early juvenile gizzard shad likely have thecapability to strongly mediate crustacean zoo-plankton population dynamics within the boundsset by bottom-up processes. This capability sug-gests that gizzard shad, depending on zooplanktonbiomass and larval density, can play an importantrole in a larger series of reservoir community in-teractions than previously demonstrated.

AcknowledgmentsWe thank J. Miner, C. Habicht, L. Einfalt, K.

Jones, and T. Wickerham, for their help in the

ZOOPLANKTIVORY BY LARVAL GIZZARD SHAD 505

field and laboratory. The comments of J. E. Breck,D. A. Culver, D. R. DeVries, and T. H. Hether-ington substantially improved earlier drafts of thismanuscript. This work was supported by FederalAid in Sport Fish Restoration, Project F-57-R,administered through the Ohio Division of Wild-life. We also thank the Ohio Cooperative Fish andWildlife Research Unit, sponsored jointly by theU.S. Fish and Wildlife Service, Ohio Departmentof Natural Resources, The Ohio State University,and the Wildlife Management Institute, for logis-tical support.

ReferencesAdams, S. M., and D. L. DeAngelis. 1987. Indirect

effects of early bass-shad interactions on predatorpopulation structure and food web dynamics. Pages103-117 i/t W. C. Kerfoot and A. Sih, editors. Pre-dation: direct and indirect impacts on aquatic com-munities. University Press of New England, Han-over, New Hampshire.

Balvay, P. G. 1987. Equivalence entre quelques para-metres estimatifs de 1'abondance du zooplanctontotal. Swiss Journal of Hydrobiology 49:75-84.

Barger, L. E., and R. V. Kilambi. 1980. Feeding ecol-ogy of larval shad, Dorosoma, in Beaver Reservoir,Arkansas. U.S. Fish and Wildlife Service BiologicalServices Program FWS/OBS-80/43:136-145.

Bean, D. J. 1980. Crustacean zooplankton productionin Lake Erie, 1970. Master's thesis. The Ohio StateUniversity, Columbus.

Beard, T. D. 1982. Population dynamics of young-of-the-year bluegill. Wisconsin Department of NaturalResources Technical Bulletin 127.

Bodola, A. 1966. Life history of the gizzard shad, Dor-osoma cepedianum (LeSueur), in western Lake Erie.U.S. Fish and Wildlife Service Fishery Bulletin 65:391-425.

Carline, R. F., B. L. Johnson, and T. J. Hall. 1984.Estimation and interpretation of proportional stockdensity for fish populations in Ohio impoundments.North American Journal of Fisheries Management4:139-154.

Carline, R. F., R. A. Stein, and L. M. Riley. 1986.Effects of size at stocking, season, largemouth basspredation, and forage abundance on survival of tigermuskellunge. American Fisheries Society SpecialPublication 15:151-167.

Carpenter, S. R., J. F. Kitchell, and J. R. Hodgson.1985. Cascading trophic interactions and lake pro-ductivity. BioScience 35:634-639.

Carpenter, S. R., and nine coauthors. 1987. Regulationof lake primary productivity by food web structure.Ecology 68:1863-1876.

Carter, J. C. H., M. J. Dadswell, J. C. Roff, and W. G.Sprules. 1980. Distribution and zoogeography ofplanktonic crustaceans and dipterans in glaciatedeastern North America. Canadian Journal of Zo-ology 58:1355-1387.

Chesson, J. 1978. Measuring preference in selectivepredation. Ecology 59:211-215.

Chesson, J. 1983. The estimation and analysis of pref-erence and its relationship to foraging models. Ecol-ogy 64:1297-1304.

Cochran, P. A., and I. R. Adelman. 1982. Seasonalaspects of daily ration and diet of largemouth bass,Micropterus salmoides. with an evaluation of gastricevacuation rates. Environmental Biology of Fishes7:265-275.

Cramer, J. D., and G. R. Marzolf. 1970. Selective pre-dation on zooplankton by gizzard shad. Transac-tions of the American Fisheries Society 99:320-332.

Culver, D. A., M. M. Boucherle, D. J. Bean, and J. W.Fletcher. 1985. Biomass of freshwater crustaceanzooplankton from length—weight regressions. Ca-nadian Journal of Fisheries and Aquatic Sciences42:1380-1390.

Culver, D. A., and W. R. DeMott. 1978. Productionof zooplankton at nearshore stations in Lakes On-tario and Erie. Internationale Vereinigung fur theo-retische und angewandte Limnologie Verhandlun-gen 20:252-256.

Dabrowski, K~, E. Murawska, J. Terlecki, and S. Wiel-gosz. 1984. Studies on the feeding of Coregonuspollan (Thompson) alevins and fry in Lough Neagh.Internationale Revue der Gesamten Hydrobiologie69:529-540.

DeMott, W. R. 1989. The role of competition in zoo-plankton succession. Pages 195-252 in U. Somer,editor. Plankton ecology. Springer-Verlag, NewYork.

DeVries, D. R., and R. A. Stein. 1990. Manipulatingshad to enhance sport fisheries in North America:an assessment. North American Journal of Fisher-ies Management 10:209-223.

DeVries, D. R., and R. A. Stein. 1992. Complex in-teractions between fish and zooplankton: quantify-ing the role of an open-water planktivore. CanadianJournal of Fisheries and Aquatic Sciences 49:1216-1227.

DeVries, D. R., R. A. Stein, J. G. Miner, and G. G.Mittelbach. 1991. Stocking ihreadfin shad: con-sequences for young-of-year fishes. Transactions ofthe American Fisheries Society 120:368-381.

Drenner, R. W., and S. R. McComas. 1980. The rolesof zooplankter escape ability and fish size selectivityin the selective feeding and impact of planktivorousfish. Pages 587-593 in W. C. Kerfoot, editor. Evo-lution and ecology of zooplankton communities.University Press of New England, Hanover, NewHampshire.

Drenner, R. W., W. J. O*Brien, and J. R. Mummert.1982a. Filter-feeding rates of gizzard shad. Trans-actions of the American Fisheries Society 111:210-215.

Drenner, R. W., F. DeNoyelles, Jr., and D. Kettle.1982b. Selective impact of filter-feeding gizzardshad on zooplankton community structure. Lim-nology and Oceanography 27:965-968.

Drenner, R. W., J. R. Stockier, and W. J. O'Brien. 1978.Capture probability: the role of zooplankter escape

506 DETTMERS AND STEIN

in the selective feeding of planktivorous fish. Jour-nal of the Fisheries Research Board of Canada 35:137O-1373.

Elliott, J. M., and L. Persson. 1978. The estimation ofdaily rates of food consumption for fish. Journal ofAnimal Ecology 47:977-991.

Guest, W. C, R. W. Drenner, S. T. Threlkeld, F. D.Martin, and J. D. Smith. 1990. Effects of gizzardshad and thread fin shad on zooplankton and young-of-year white crappie production. Transactions ofthe American Fisheries Society 119:529-536.

Gulyas, P. 1980. The effect of temperature on the mostfrequent Cladocera and Copepoda species in LakeVelence. Aquaeultura Hungarica 2:2155-2170.

Hall, D. J., and T. J. Ehlinger. 1989. Perturbation,planktivory, and pelagic community structure: theconsequence of winterkill in a small lake. CanadianJournal of Fisheries and Aquatic Sciences 46:2203-2209.

Haney,J.F.,andD. J. Hall. 1973. Sugar-coated Daph-ma: a preservation technique for Cladocera. Lim-nology and Oceanography 18:331-333.

Hunter, J. R., and C. A. Kimbrell. 1980. Early lifehistory of Pacific mackerel. Scomberjaponicus. U.S.National Marine Fisheries Service Fishery Bulletin78:89-101.

Johnson, B. M., R. A. Stein, and R. F. Carline. 1988.Use of a quadrat rotenone technique and bioener-getics modeling to evaluate prey availability tostocked piscivores. Transactions of the AmericanFisheries Society 117:127-141.

Keast, A. 1980. Food and feeding relationships of youngfish in the first weeks after the beginning of exoge-nous feeding in Lake Opinicon, Ontario. Environ-mental Biology of Fishes 5:305-314.

Kutkuhn, J. H. 1958. Utilization of plankton by giz-zard shad in a shallow prairie lake. Transactions ofthe American Fisheries Society 87:80-103.

Lampert, W. 1978. A field study on the dependenceof fecundity of Daphnia spec, on food concentra-tion. Oecologia (Berlin) 36:363-369.

Lampert. W., W. Fleckner, H. Rai, and B. E. Taylor.1986. Phytoplankton control by grazing zooplank-ton: a study on the spring clear-water phase. Lim-nology and Oceanography 31:478-490.

Lazzaro, X. L., R. W. Drenner, R. A. Stein, and J. D.Smith. In press. Planktivores and plankton dy-namics: effects offish biomass and planktivore type.Canadian Journal of Fisheries and Aquatic Sci-ences.

Mallin, M. A., L. J. Birchfield, and W. Warren-Hicks.1987. Food habits and diet overlap of larval Lc-pomis spp. and gizzard shad in a Piedmont reser-voir. Proceedings of the Annual Conference South-eastern Association of Fish and Wildlife Agencies39(1985):146-155.

Matthews, W. J., L. G. Hill, D. R. Edds, J. J. Hoover,and T. G. Heger. 1988. Trophic ecology of stripedbass, Morone saxatilis, in a freshwater reservoir(Lake Texoma, U.S.A.). Journal of Fish Biology 33:273-288.

McQueen, D. J., M. R. Johannes, J. R. Post, T. J. Stew-

art, and D. R. S. Lean. 1989. Bottom-up and top-down impacts on freshwater pelagic communitystructure. Ecological Monographs 59:289-309.

McQueen, D. J., J. R. Post, and E. L. Mills. 1986.Trophic relationships in freshwater pelagic ecosys-tems. Canadian Journal of Fisheries and AquaticSciences 43:1571-1581.

Michaletz, P. H., D. G. Unkenholz, and C. C. Stone.1987. Prey size selectivity and food partitioningamong zooplanktivorous age-0 fishes in Lake Fran-cis Case, South Dakota. American Midland Natu-ralist 117:126-139.

Miller, R. R. 1960. Systematics and biology of thegizzard shad (Dorosoma cepedianum) and relatedfishes. U.S. Fish and Wildlife Service Fishery Bul-letin 173:371-392.

Mills, E. L., and J. L. Forney. 1981. Energetics, foodconsumption, and growth of young yellow perch inOneida Lake, New York. Transactions of the Amer-ican Fisheries Society 110:479-488.

Mills, E. L., J. L. Forney, and K. J. Wagner. 1987. Fishpredation and its cascading effect on the OneidaLake food chain. Pages 118-131 in W. C. Kerfootand A. Sih, editors. Predation: direct and indirectimpacts on aquatic communities. University Pressof New England, Hanover, New Hampshire.

Murdoch, W. W., and J. Bence. 1987. General pred-ators and unstable prey populations. Pages 17-30i/i W. C. Kerfoot and A. Sih, editors. Predation:direct and indirect impacts on aquatic communi-ties. University Press of New England, Hanover,New Hampshire.

Nero, R. W., and W. G. Sprules. 1986. Predation bythree glacial opportunists on natural zooplanktoncommunities. Canadian Journal of Zoology 64:57-64.

Noble, R. L. 1981. Management of forage fishes inimpoundments of the southern United States.Transactions of the American Fisheries Society 110:738-750.

O'Brien, W. J. 1987. Planktivory by freshwater fish:thrust and parry in the pelagia. Pages 3-16 in W.C. Kerfoot and A. Sih, editors. Predation: direct andindirect impacts on aquatic communities. Univer-sity Press of New England, Hanover, New Hamp-shire.

Persson, L. 1979. The effects of temperature and dif-ferent food organisms on the rate of gastric evacu-ation in perch (Percafluviatilis). Freshwater Biology9:99-104.

Persson, L. 1982. Rate of food evacuation in roach(Rut i I us rut i I us) in relation to temperature, and theapplication of evacuation rate estimates for studieson the rate of food consumption. Freshwater Biol-ogy 12:203-210.

Peters, R. H. 1983. The ecological implications of bodysi/e. Cambridge University Press, New York.

Pierce, R. J. 1977. Life history and ecological ener-getics of the gizzard shad (Dorosoma cepedianum)in Acton Lake, Ohio. Doctoral dissertation. MiamiUniversity, Oxford, Ohio.

Pierce, R. J., T. E. Wissing, and B. A. Megrey. 1981.

ZOOPLANKTIVORY BY LARVAL GIZZARD SHAD 507

Aspects of the feeding ecology of gizzard shad inActon Lake, Ohio. Transactions of the AmericanFisheries Society 110:391-395.

Plante, C, and J. A. Downing. 1989. Production offreshwater invertebrate populations in lakes. Ca-nadian Journal of Fisheries and Aquatic Sciences46:1489-1498.

Post, J. R., and D. J. McQueen. 1988. Ontogeneticchanges in the distribution of larval and juvenileyellow perch (Perca flavescens): a response to preyor predators? Canadian Journal of Fisheries andAquatic Sciences 45:1820-1826.

Rajasilta, M, and I. Vuorinen. 1983. A field study ofprey selection in planktivorous fish larvae. Oecolo-gia (Berlin) 59:65-68.

Scavia. D., G. A. Lang, and J. F. Kitchell. 1988. Dy-namics of Lake Michigan plankton: a model eval-uation of nutrient loading, competition, and pre-dation. Canadian Journal of Fisheries and AquaticSciences 45:165-177.

Siefert, R. E. 1972. First food of larval yellow perch,white sucker, bluegill, emerald shiner, and rainbowsmelt. Transactions of the American Fisheries So-ciety 101:219-225.

Stein, R. A., and seven coauthors. 1988. Size-struc-tured interactions in lake communities. Pages 161-179 in S. R. Carpenter, editor. Complex interactionsin lake communities. Springer-Verlag, New York.

Van Den Avyle, M. J., and J. R. Wilson. 1980. Foodhabits and feeding selectivity of larval Dorosomaspp. in Center Hill Reservoir. U.S. Fish and Wild-life Service FWS/OBS-80/43:146-156.

Wahl, D. H., and R. A. Stein. 1988. Selective preda-tion by three esocids: the role of prey behavior andmorphology. Transactions of the American Fish-eries Society 117:142-151.

Webb, D. G., and T. R. Parsons. 1988. Empirical anal-ysis of the effect of temperature on marine harpac-ticoid copepod development time. Canadian Jour-nal of Zoology 66:1376-1381.

Werner,E. E.,andJ. F.Gilliam. 1984. The ontogeneticniche and species interactions in size-structuredpopulations. Annual Review of Ecology and Sys-tematics 15:393-425.

Received May 3, 1991Accepted November 19, 1991