Transactions of the American Fisheries Society, vol. 110, iss. 2, (March 1981), pp. 197 – 209. http://afs.allenpress.com http://afs.allenpress.com/archive/1548-8659/110/2/pdf/i1548-8659-110-2-197.pdf Online ISSN: 1548-8659 Print ISSN: 0002-8487 DOI: 10.1577/1548-8659(1981)110<197:PBPTMO>2.0.CO;2 © American Fisheries Society TRANSACTIONS of the March 1981 AMERICAN VOLUME 110

FISHERIES SOCIETY NUMBER 2

Tranaactionaof the America• F•sheries Society 110:197-209, 1981 ¸ Copyrightby the AmericanFisheries Society 1981

Predation by Pellet-RearedTiger Muskellunge on Minnows and Bluegills in Experimental Systems

AL^• L. C. ILLEN OhioCooperative Fishery Research Unit •

RoY A.

Departmentof Zoology,The OhioState University

ROBERT F. OhioCooperative Fishery Research Unit, TheOhio State University 1735 Neil Avenue, Columbus,Ohio 43210

Abstract Studiesin Wisconsinlakes have shownthat stockedtiger muskellunge(F• hybridsof female muskellunge,Esox masquinongy x male northern pike, E. lucius)reared on livefood survive better than thosereared entirely on dry pelletfood. We evaluatedthe abilityof pellet-rearedhybrids to convertto a minnow(Notropis spp. and Pimephalespromelas) or bluegill(Lepomis macrochirus) diet in laboratoryaquaria and hatcheryponds. In aquaria,86-310-mm (total length) tiger muskellungeselected cyprinids that wereabout 40% of their ownlength and bluegillsthat were about30% of their length,sizes closely predicted by an optimalforaging construct (time from preycapture to completeprey ingestion + preydry weight).Using these prey sizes, we tested hybrids(130, 150,and 170mm long)in conversionexperiments in aquariaand ponds.During experiments,prey were maintained at a constantdensity and predators were sampled periodically to determinethe proportioneating fish. Tiger muskellungeconverted more slowly to bluegills than to minnowsin bothaquaria and ponds.In aquaria,85% of the hybridsconverted from pelletsto minnowsby day 3, whereasonly 68% convertedto bluegills.By day 5, conver- sionsto minnowsand bluegillswere 95% and 82%, respectively.In ponds,73% of the hybrids convertedto minnowsby day 5 and 89%by day 14. No hybridshad eaten bluegills by day3 and only 53% convertedby day 14. The apparentlylimited ability of pellet-rearedtiger muskellungeto switchto a bluegilldiet mayinfluence survival and growthof thesepredators in reservoirsdominated by a centrachidforage base.

The tiger muskellunge(F• hybrid of female 1978). It can be economicallycultured on arti- muskellunge,Esox masquinonffy x male north- ficialdiets and may reachlarge sizesafter stock- ern pike, E. lucius)is stockedin lakesfor sport- ing. However, field studiesin Wisconsinlakes fishingand control of undesirablefishes (Graff have shownthat survivalof hybridsreared on fish in hatcherieswas much higher than that of hybridsreared on pellets(Johnson 1978). Poor i The Unit isjointly supportedby the United States survivalcould result if hybridsdo not develop Fish and Wildlife Service,Ohio Department of Nat- their ability to stalk and catchlive prey by the ural Resources,and The Ohio State University. time they reachstocking size, or if the time for 197 198 GILLEN ET AL. this behavioralchange is so long that starvation length classin aquaria, and recorded the size reducestheir viability. We conductedexperi- classof the first prey that was eaten (usually mentsin the laboratoryand smallponds to de- within 5 minutes after prey introduction). termine how quicklyhybrids reared on pellets Twenty suchtrials for each hybrid length class convertedto a diet of minnows(Notropis spp. were conducted.During size-selectionexperi- and Pimephalespromelas) or bluegills(Lepomis ments, we measuredhandling times (N = 12) macrochirus). for eachpredator and eachprey length class. We dried l0 each of bluegills and minnows Methods from each length class(N = 60 per prey type) Hybrids, reared entirely on dry pellet food, for 48 hours at 80 C, and weighedthem (near- were obtainedfrom Wolf Lake Hatchery, Mich- est 1.0 rag). We compared handling time per igan and Kineaid Fish Farm, Ohio. We held prey dry weight (C/B) with the prey sizesmost these fish in 500-liter tanks at 15-17 C and frequentlychosen in size-selectionexperiments. fed them pellets.For laboratorytrials, individ- We measuredtotal length (to the nearest1.0 ualswere acclimated over 24 hoursto experi- mm) and greatestbody depth (nearest0.1 ram) mental temperatures.During the acclimation of 60 minnowsand 60 bluegills(10 per length period, fish were fed pelletsor minnowsor class)and establishedequations relating these blue•11s.Hereafter, we refer to fish that were variables.We usedthese relationships to estab- alwaysfed pelletsas naive,whereas those that lish if hybridsbased their prey selectionon werefed live prey for at least1 day weretermed lengthor body depth. experienced.Once a hybrid captured its first prey, its predatory efficiencydid not increase HybridBehavior during subsequentcaptures (see Results). We comparedpredatory efficiencies of naive and experiencedhybrids. Naive individuals Prey-SizeSelection were starvedfor 24 hours in 100-liter aquaria In preliminarytrials, we determinedthe ap- before minnows were introduced. Minnows propriate sizeof minnowsand bluegillsto use that were not eaten within 30 minutes were re- in subsequentexperiments based on optimal moved.This procedurewas repeated daily until foraging theory (Schoener1971; Werner 1974, the hybrid ate its first minnow. On the day the 1977; Werner and Hall 1974). Optimal prey first minnow was consumed, we recorded the size is operationallydefined as the size with a time from prey introductionto capture.Times minimalratio of handlingtime (time from prey to capturefor thesenaive fish were compared captureto completeprey ingestion,Ht, in sec- to timesto capturefor experiencedhybrids (1- onds) per unit of dry weightof prey (Wt, in 14 days on a fish diet). We also comparedthe milligrams); that is, with a minimal cost/benefit numberof strikesper capturefor 25 naiveand ratio, C/B = Ht/Wt (Werner 1974). This ap- 25 experiencedhybrids (170 ram) that were proach has been used successfullyto predict given 60-ram minnows. prey sizeschosen by predatorsin the laboratory and field (Werner 1974; Kisalaliogluand Gib- Diet Conversionin Aquaria son 1976; Stein 1977). During 5-day experiments in a 700-liter Size-selectionand optimal foraging experi- aquarium (19-21 C), 40 minnowsor bluegills ments were conducted in 100-liter aquaria at that spannedthe optimalsize were introduced water temperaturesof 19-23 C. Hybrids were to 40 naive hybrids (150 or 170 ram) that had separatedinto six totaldengthclasses (millime- been starved 24 hours. The smaller hybrids ters): 86-90, 126-130, 146-150, 166-170,246- were given 51-60-mm minnowsor 31-40-ram 250, and 306-310. Prey typesalso were divided bluegills;the larger hybridswere given 61-70- into six length classes(millimeters): 26-30, 36- mm minnowsor 41-50-ram bluegills.Every 12 40, 46-50, 56-60, 66-70, and 76-80. Through- hours,enough prey were added to maintaina out the paper, we identify theselength classes densityof 40 prey per tank (57/ma).Twenty of predator and prey by the highestvalue in hybridswere removed daily and the proportion each range. Experienced hybrids were accli- that had eatenfish was determined by pumping mated individually in aquaria and starved for (days2 and 4) or dissecting(days 1, 3, 5) the at least24 hours.We placedone prey from each stomachs.Fish whosestomachs were pumped PREY SELECTION BY TIGER MUSKELLUNGE 199

were returned to the aquarium;dissected fish • MINNOW HYBRID were replacedfrom stocktanks in the same 4- BLUEGILL SIZE CLASSES proportion (naive:experienced)revealed by 310ram stomachanalyses. This insured no changein >_ 5o group learning conditions.For 150-mm hy- z brids,there were three experimentswith min- Ld 0 nowsand two with bluegills;for 170-mmpred- 250 mm ators,there were two experimentswith minnows ,?, 5o and three with bluegills.In the first bluegillex- perimentconversion by hybridswas low; there- o fore, samplingwas extended to 9 daysin sub- z 50 sequentexperiments. ¸

Diet Conversion in Pon& • o Ld 5O Two diet conversionexperiments were con- ducted in 0.05-hectareponds at London Fish o• 0 Farm, Ohio. Pondswere 1.5 m deep and con- Ld 5O tained a moderate(about 18%) bottom coverof macrophytes(average density, about 40 stems/ m2).Reed canary grass (Phalaris arundacea) and >_ 0 90 mm terrestrialgrasses grew out of the water within Ld 5O 1 m of pond margins;Chara spp. was widely distributedthroughout the ponds.Secchi disc 0 readingsalways exceeded 1.5 m (N = 10) and 50 40 50 60 70 80 oxygenwas always above 12 rag/liter. A maxi- mum-minimum thermometer, submergedto a MEAN PREY LENGTH (mm) depthof 1 m, wasread everyother day. Mean FIGUREl.•Proportion of time.•experienced tiger muskel- watertemperatures were 23 C during the first lungechose different sizes of minnowsor bluegillsduring experiment (termed the August experiment) 20 trialsrun in 100-1iteraquaria. For eachpredator size and 17 C for the secondexperiment (termed class,hybrid• were given a choiceof sixprey-length classes of minnowsor bluegills. the Octoberexperiment). We stockedeach of five pondswith 125 tiger muskellungebetween 120 and 130 mm at the both experiments.This pond was used to de- beginningof the Augustexperiment and with termine the ability of tiger muskellungeto con- 124 between 160 and 175 mm at the beginning vert from pelletsto minnowsin the presenceof of the October experiment.Two ponds con- particularlyabundant prey. At least 100 prey tainedbluegills; three pondscontained fathead frowneach pond were measuredat the begin- minnows(P. promelas).For each prey type, two ning and end of each 14-dayexperiment. pondswere stockedwith 250 prey per pond Naive hybrids, obtained from Kincaid Fish everyother day; densityincreased from 0.25 to Farm, were stockedin the ponds; prey were 1.9 prey/m3 during the courseof the experi- added the next day. The percentageof hybrids ment. These densities were somewhat higher convertingto fish was determined from a sam- than inshoreprey densitiesin Ohio reservoirs ple of 15 per pond, obtainedby seiningon days (R. F. Carline and R. A. Stein, unpublished 1, 3, 5, 7, 11, and 14. Samplefish were weighed, data). Length rangesof minnowsand bluegills measured, and dissected. Gut contents were spannedthe optimalsizes fbr hybrids.For the identified; minnows and bluegills were mea- Augustand Octoberexperiments, respectively, sured. mean lengthsof minnowswere 53 mm (range Results 36-60 mm) and 59 mm (31-90 mm), and mean lengths of bluegillswere 32 mm (21-45 mm) Prey-SizeSelection and 37 mm (21-55 mm). The fifth pond con- Sizeof prey selectedby experiencedhybrids rained at least 10,000 minnows (range 16-80 increasedwith predator size (Fig. 1). Lengths mm), or about 10 prey/m3, for the durationof of prey offered were not large enoughto ade- 200 G1LLENET AL.

l A I- qLMEDIAN, 95% C.I. N=20 PER SIZE CLASS z 75. ß BLUEGILL Ld

50 ÷ ßMINNOW

I i I I I I I

I 20

o •0

I i I I i I I I I I I I00 200 300 PF•F'DATOF• lENGTH (turn) FZOU•,E2.---Median length.; and body depths of minnow.;and bluegills chosen by different sizes of experiencedtiger mu.;kellungein size-selection experiments. To convert body lengths (BL) to bodydepths (BD) weused the equation: minnows,BD = 1.13 + O.14BL,N = 60, r = 0.91; bluegilL;,BD = 3.38 + 0.36BL, N = 60, r = 0.98. ConfidenceintervaL;for median.; were calculated bymeans of Walshaverages (Hollander and Wolfe 1973). quately determine preferred sizesof 250- and plainingsize-selection results. Cost of handling 310-mm hybrids. Predators, 90 to 170 mm, preyincreased with prey length; prey of a given most frequentlyselected minnows 37-43% of sizewere handledin lesstime by large preda- their meanlength and bluegills25-30% of their torsthan by smallones (Fig. 3). Handling times length. However, when prey size selectedwas were longer for bluegillsthan for minnowsof measuredin termsof body depth rather than the samelength. When thesehandling times length, hybridsselected larger bluegillsthan are incorporated into cost/benefitcalculations minnows(Fig. 2). While prey lengthprovides (Ht Wt), it is apparentthat the range of "prof- a convenientvariable for measuringprey size itable"prey sizes(low handlingcosts per food selection,body depth may be equallyimportant return) broadensas thesepredators grow (Fig. in influencingpredator choice. 4). The lowestpoint on eachcurve, which rep- Cost/benefitanalysis (H,/Wt) wasuseful in ex- resents the theoretical optimal prey size in PREY SELECTION BY TIGER MUSKELLUNGE 201

- 90mm 130mm 150mm 400 ! INTER- 300 ! QUARTILE

200 - •- RANGE ,'

0 170mm 250mm 310mm 400

300

200 , •00

0 I i I • I • I ,50 7'0 30 ,50 7'0 30 ,50 7'0

MEAN PREY LENGTH (mm) F/CURE3.--Handling time as a J•tnction of thetotal length of bluegillsand minnows for sixlength classes of experienced tigermuskellunge (indicated in theupper right-hand corner of eachpanel). terms of energetics, was about 40% of total bridscaptured minnows more quickly than did lengthof tiger muskellungefor minnowsand their naive counterparts(Mann-Whitney U- about30% of predatortotal length for bluegills. test, P < 0.025, N = 90). Median time to cap- The shapeof a C/B curve providesa relative ture by experienced hybrids was 75 seconds measureof diet breadth(Werner 1974).Tight- (range5-600 seconds,N = 77) coinparedwith ly folded curvessuggest a narrow range of 540 seconds(range 10-740 seconds,N = 13) profitable prey sizesor narrow diet breadth, for naivepredators. Thus, somelearning may whereas broad C/B curves indicate a wide diet have occurredas thesepredators converted to breadth.Clearly, for sinall(90, 130 mm) pred- live food. Experienced hybrids had fewer ators,the curvewas tightly folded, and a rela- strikesper capturethan naiveones, although tivelysinall increase in prey lengthbeyond the thisdifference was not quite significant(mean, minimum Ht/Wt resultedin a large increasein 1.3 and 2.0 strikes/capture,respectively, t-test, handling costsper food return. In hybrids P = 0.15). Once hybrids had eaten their first longer than 170 mm, C/B ratiosremained low fish, strikesper capturein subsequentpreda- over a wide range of prey lengths.Thus, diet tion actsdid not change(Mann-Whitney U-test, breadthincreased gradually as predator sizein- P > 0.05). Overall,experienced predators cap- creased. tured prey more efficientlythan naive ones.

Hybrid Behavior Diet Conversion Diet historyinfluenced predatory behavior of In laboratoryexperiments a large portionof tiger muskellungein aquaria.Experienced hy- naivehybrids converted to minnowsor bluegills 202 GILLEN ET AL.

90mm 130mm 150mm 5.0

>- 2.0 rr

Z I..0

0 1TOm,m 250mm 310mm 2.0

Z 1.0

0 Z 30 50 70 50 50 70 50 50 70

Z MEAN PREY LENGTH (mm)

Fmu.E 4.•Cost/benefitcurves (handling time/dry weighO, as a functionof minuowor blueg•lllength, for six length classesof experiencedtiger muskellunge (indicated in theupper right-hand corner of eachpanel). after 5 to 9 days, but conversionto minnows ponds,minnow densitiesranged from 0.25 to was more rapid than to bluegills(arcsin trans- 1.9 prey/mø;in the third pond, densitieswere formation, t-test, P < 0.05; Fig. 5). By day 5, 10 prey/mø. Thus, prey densitieswithin this 90-95% of both hybrid sizeshad converted to range did not influenceconversion rate of naive minnows,whereas only 70-85% had converted hybridsto minnows.In both experiments,naive to bluegills(Fig. 5). By day 7, 80% of the 150- tiger muskellungeconverted slowly to bluegills mm hybridshad convertedto bluegillsand by and no differencesexisted between ponds with- day 9, nearly 100% of 170-mm predators had in either experiment (Friedman's test, P > converted.The conversionby 170-mm hybrids 0.05; Fig. 6). No predator ate a bluegill before wasgenerally faster than that of 150-mm fish. day 3. Lessthan 10% had convertedby day 5 In both the August and October pond ex- and only 28-53% by day 14. periments,naive hybridsconverted rapidly to In both experiments,conversion to bluegills minnows. Because conversion rates did not dif- was higheston day 14, after ponds had been fer among the three minnow ponds (Fried- partly drained overnight. This reduction in man's test, P > 0.05; Fig. 6), data were com- pond volume prevented accessby bluegillsto bined for each experiment. More than 70% of inshorecover, possiblyincreasing their vulner- the hybridshad convertedto minnowsby day ability to predation. Percentagesof hybrids 5 and more than 90% by day 14. In two of the convertingto bluegillswere much lower than PREY SELECTION BY TIGER MUSKELLUNGE 203

80 > z o 60 i•-•- • MINNOW 40 20 ' -.- E: 150mm 0

80 o 6O 95% CONFIDENCE 40 z INTERVALS 20 HYBRID 81ZE• 170mm 0 0 z TI ME (doys)

FIGURE5.•Percentage of naivetiger muskellunge converting to minnowsand bluegillsthrough time in a 700-liter aquarium.The partition in thebluegill cu7•e (panel B) separatesdata taken during regular sampling (l•t ofpartition) anddata taken from extendedsampling (right of partition).For cuT•esAe andB, eachvalue plotted represenLs two experiments( N = 2OZ,'ampleday),' for cu•esA• andBe (l•t ofpartition), each value plotted represents three experiments (N = 20/sampleday). For Be to the right of the partition Nro•,• = 38 hybrids.Open symbot• (¸, •) representconversion determinationbyinspection orstomach pumping; closed symbols ( •, &) representdetermination byd•ection. percentagesof those convertingto minnows of these predators to convert to bluegills or during both pond experiments(Friedman's minnows. test, P < 0.001). Even though a large propor- We stocked ponds with bluegills and min- tion of hybridsin bluegillponds were not eating nowswhose length range spannedsizes most fish, they did consumebenthic invertebrates, frequently selectedby hybridsin laboratory ex- including mayfly nymphs (Baetis),dragonfly pariments,so that conversionrates by hybrids naiads,snails, leeches, algae, cladocerans,and would not be biased by prey length. In both detritus. Growth of hybrids was greater in experiments,hybrids consumedbluegills that pondsstocked with minnowscompared to those were smaller than the most abundant size stocked with bluegills in both experiments groups (Fig. 8). In contrast,hybrids ate min- (nonparametricmultiple range test, P < 0.05, nows that were generally similar in size to the Zar 1974;Fig. 7). No differencesin thesevari- mostabundant size groups. These distributions ableswere found among the three minnow or do not provide the besttest of prey size selec- betweenthe two bluegill ponds,within either tion by predators,because predator lengthsin- experiment(Friedman's test, P • 0.05);growth creasedduring experiments. differencesappeared to be relatedto the ability To determine if hybridsof a given sizewere 204 GILLEN ET AL.

• 100- A AUGUST EXPERIMENT T-M SlZE:130mmm

6o v • 95%CONFIDENCE 4o m 20

0 I I I I ] I I I I I I [ I I 0 I00 _ B OCTOBEREXPERIMENT T-MSIZE•170mm• ao z o 60

a_ 40 z -*- BLUEGILL / • 20 / / //• :MINNOW / o .___.,___ +--- + ...... ß

o ,5 I0 15 TIME (days) FIGU,E6.--Percentages oftwo sizes (130, 170 ram) of naive tiger muskellunge (T-M) convertingtominnows and bluegill• in O.05-hectarepond•. For the minnow ponds, each point represents 45 predators (15 perpond); for thebluegill ponds, eachpoint represents 30 predators (15 perpond). Lines were fitted by inspection. selectinga given sizegroup of prey, we plotted mentsdivided by mean sizeof stockedprey and hybridlength against the ratio of lengthof prey plotted these on the same graph. If hybrids consumedto mean length of prey stocked.If chosesimilar sizesof minnowsin the laboratory hybrids selected prey at random, this ratio and ponds, data points should be clustered shouldremain at unity irrespectiveof predator about those lines (Fig. 9, lines A and B). length. We used only data from ponds with Lengthsof minnowseaten by the smallerhy- high densitiesof minnows(about 10/ma) be- brids (149-152 ram) in the August experiment causethese wide length rangesof prey provid- approximatedpredicted ratios; however, larger ed the best opportunity for size-selectivepre- deviationsmight have resultedif a wider size dationby hybrids.Length ratios (prey consumed range of minnowswere available.Larger hy- to prey available) increased with predator brids (161-197 ram) consumedprey that were length when data from both experimentswere smaller than predicted from the laboratory combined,suggesting that as hybridsincreased data; 39 of 41 length ratios were less than in length they choselarger prey (Fig. 9). We would be expectedbased on size-selectionex- then recomputed ratios using length of prey periments. most frequentlyselected in laboratoryexperi- These resultswere not unexpectedbecause PREY SELECTION BY TIGER MUSKELLUNGE 205

A AUGUSTEXPERIMENT 20- INITIALT-MSIZE: 15-

I0

__._- 95% CONFIDENCE ....+ 5- INTERVALS --e-- MINNOW -*- BLUEGILL

0 I I I i I I I I i I I I i B OCTOBEREXPERIMENT 35- z 30- INITIALT-MSIZE:170rn• 25-

20 ß I I I I I I I I [ I I I I I 0 5 I0 15

TI ME (doys) FmURE7.--Gr0wth of tiger muskellunge (T-M) in O.05-hectare ponds with minnows and bluegills asprey. For the minnow ponds,each point represent; 45 predato•:;(15 perpond); for thebluegill ponds, each point represents 30 predators (15 perpond). Lines were fitted by inspection. cost/benefitcurves for hybrids less than 150 theoryhas occurred in a varietyof other studies mm were tightlyfolded (Fig. 3), suggestingthat (Kisalalioglu and Gibson 1976; Stein 1977; only a restrictedsize range of prey is profitable. Werner 1977). Thus, laboratory-derivedcost/ Cost/benefitcurves for hybridsover 170 mm benefitratios may be usefulin predictingop- were rather fiat, indicating that these larger timal prey sizeand type. predators would derive approximatelyequal However, not all of our laboratoryresults benefit from a relativelywide range of prey were borne out in the field. Bluegill Ht/Wt sizes.While theseresults indicate good corre- curvesdid not predict the small sizesthat were spondencebetween predictions from optimal chosenby tiger muskellungein ponds.Because foragingtheory and field results,theory could hybridsconverted so slowlyto bluegills,their havebeen more rigorouslytested in our exper- hungerlevels were probably much higher than imentsif a wide rangeof prey sizegroups had those in the minnow ponds. As hunger in- been stockedin similarproportions. creases,predators generally become less selec- tive in their choiceof prey (Emlen 1968; Ware Discussion 1972);this couldexplain the lackof sizeselec- tion for bluegills.In addition, if pursuit and Prey Selection searchcontributed substantially to the costterm Pond experimentsbroadly confirm labora- for tiger muskellungepreying on bluegills,Ht/ tory findingsconcerning optimal foragingby Wt may not accuratelypredict optimal size. In tiger muskellunge.The congruenceis particu- similarexperiments in smallponds, Schneider larly markedfor naivehybrids presented with (1975) found that walleyes(Stizostedion vitreum) minnowprey. Similar confirmation of foraging reared on artificial diets also chose the smallest 206 GILLEN ET AL.

--ß--OUT N6:12 60 • PONDNp=99 HIGH MINNOW DENSITY (AUGUST)

DENSITY N6 = 33 (AUGUST) Np =100 40 ÷ • BLUEGILLLOWMINNOW N6 = 57 A, / / IAUGUSTINp=l13

20- '/ ', •/ ;,/

HIGH MINNOW DENSITY (OCTOBER) • 60' --ß-- GUTN6=,48 N6=63 N6 = 41 c:• + PONDNp=187 Np=159 Np = 147 BLUEGILL(OCTOBER) i:• 40 ILOW MINNOWDENSITY(OCTOBER

2o

o 20 40 20 40 60 80 20 40 60 80 PREY LENGTH (mm) FIGURE8.--Length-frequency distributions' of bluegills and minnowsat the,•tart o./' the August and Octoberpond experi- mentsand of fish removed from hybrid guts. Sample sizesJbr available prey and those in gutsare represented byNe and N•, respectively.Arrows indicate the lengthof preymost,#equently selected in laboratoryexperiment,•, 25% of mean hybridlength for blueg•illsand 37%for minnow,•.Horizontal adjacent bars under the arrows represent ñSD o] mean hybridlength times the appropriate proportion, 25 or 37%. bluegills available (14-26% of their total Northern pike (110-335 ram) prefer soft-rayed length). Thus, high hunger levelsof predators fish over spiny-rayedfish in laboratoryaquaria in the pond environments,coupled with no (Beyerleand Williams1968), swimmingpools considerationof the costsof pursuitor search, (Mauck and Coble 1971), and hatcheryponds may explain disagreementsbetween our labo- (Mauck and Coble 1971; Weithman and An- ratory-generatedoptimal size predictionsand derson1977). Long handling times may explain pond results. why esocidfood studiesshow that spiny-rayed Tiger muskellunge captured and handled fishare lessdesirable than soft-rayedones. Spi- minnowsmore efficientlythan they did blue- ny-rayed fish are more costlyprey, in terms of gills.This wasdue, in part, to the differencein both energyrequired for ingestionand risk of morphologybetween the two prey types.Min- mortality;spines can puncturethroat or stom- nows are cylindricaland soft-rayed;bluegills achlinings (of 290 hybridsfeeding on bluegills, are deep-bodiedand spiny-rayed.Body length 14 died). and bodydepth appearedto be importantfac- tors in selectingand handling prey. In related Conversionby Tiger Muskellunge work, Johnson (1969) concluded that body Experiments in laboratory aquaria and depthof bluegillsinfluenced selection by north- hatchery ponds both demonstrated that naive ern pike. In preliminary experiments, we tiger muskellungeconverted more rapidly to found that hybridshandled spineless bluegills minnowsthan to bluegills.Most hybrids were nearlyas efficiently as minnows, suggesting that able to convert within the first 7 days in the spines contribute to the costsof predation. laboratory.Confinement of prey in aquaria PREYSELECTION BY TIGER MUSKELLUNGE 207

E o AUGUST EXPERIMENT / N=33 ß /

ß OCTOBERN=41 EXPERIMENT LINE B 7// ß ß ------PREDICTED RATIOS /

c• Ld 0:2 G_ • 1.0 Z z o H Ill T ß Y:O.20+O.OOSX •' • o o ß • 0.82 r =0.60 o Z n o N =74 z • 0.6-

I i I ' I ' I 120 140 160 180 200 PREDATOR LENGTH (mm) FIOURE9.•Relationship between predator length (X) andthe ratio (Y) ofprey length in predatorguts to mean prey lengthin pondsfor theAugust and October pond experiments. Only data from high-density minnow ponds are given. Lines'A andB werecalculated./?om datadeffved from pr• dze-selectionexperiments when ratios equalled 0.37 times predatorlength divided by mean prey length in ponds.Lines' A andB areoffset because mean prey length in pond• differedbetween experiments (tee Fig. 8).

,nayhave facilitated capture by hybrids;pred- 1967). In addition to superiormaneuverability, atorsoften trapped prey in aquariumcorners. bluegills reduce their vulnerability by swim- Hybrids convertedto minnowsat similarrates ming in schoolsdeep in the watercolumn, near in bothlaboratory aquaria and ponds;however, vegetated areas (Hall and Werner 1977), conversionto bluegillswas far slowerin the whereasminnows in our experimentsremained pondsthan in the laboratory.From preliminary in open water, seekinglittle refuge. Becauseof observations,bluegills appear more difficultto these habitat differences, bluegills may have capturein the pondsituation; in thelaboratory, been lessavailable to predators than minnows. hybridsrequired fewer strikes/capture (mean = Consistentwith thisdescription of behaviorand 2.0, range 1-13, N = 30) than in the pond habitatuse by bluegillswas the markedincrease (mean= 4.3, range,1-5, N = 12).In turn, dif- in predationrate on day 14 of the pond exper- ferences for minnows between laboratory iments.On day 13, pondswere partly drained, (mean= 1.3, range 1-3, N = 30) and pond densemacrophytes along the pond margin no (mean = 1.5, range 1-2, N = 5) were not as longer providedrefuge ibr bluegills,and the great.Apparently, confinement in smallaquar- percentageof tiger muskellungeconverting on ia modifiedthe escapetactics of bluegillsmore the last day increasedfrom 13 to 28 in the Au- dramaticallythan thoseof minnows.In ponds, gust experiment and 0 to 53 in the October bluegillsmay well haveavoided predators ow- experiment. ingto theirsuperior maneuverability compared Becausethese predators converted more rap- to the more fusifbrm minnows (Alexander idly to minnowsthan to bluegills,they gFew 208 GILLEN ET AL. fasterin minnowponds than in bluegillponds. may exhibit poor survivaldue to their failure Predators with food in their stomachs had to convertto live forage. We believethat naive either fish or invertebrates, seldom both; those hybrids stockedin systemswith a cyprinid or not eating fish had small organismsin their clupeid forage base would quickly convert to stomachs, such as cladocerans and benthie in- live forageand survivalwould be governedby sects,and stomachs were less full than those of other factors. Stockingprograms using small piscivores.Beyerie (1978) found that age-If predatorsmust be preciselycoordinated with northern pike in bluegill lakes fed largely on sizeand abundanceof prey owing to the nar- insectlarvae, crayfish, and tadpolesdespite an row sizerange of vulnerableprey. Agenciescan abundanceof edible-sizedbluegills. Our results afford to be lessprecise in their choiceof stock- suggestthat when centrarchidsdominate the ing sitesfor large esocids;wide diet breadths foragebase in reservoirs,tiger muskellungewill coupledwith reduced susceptibilityto preda- be unsuccessfulpredators during the first 2 tors improves the likelihood of high rates of weeksafter stocking.Under stressfulconditions growth and survival. these fish may have lower survival than those stockedinto lakes with a cyprinid or clupeid Acknowledgments forage base. We thank M. Headrick and T. Nagel for their In experimentssimilar to ours, when pellet- assistancein pond experimentsat LondonFish reared walleyeswere combinedwith bluegills Farm; D. B. Apgear for coordinatingactivities or minnowsin 0.3-hectareponds, 70-90% of with the Ohio Department of Natural Re- the mortality occurredwithin 6 monthsafter sources;P. J. Pfister(Manager of KincaidFish stockingand appeared to be associatedwith Farm) for supplying tiger muskellunge for handlingand adaptationto a new environment pond experiments;J. Copeland(Manager of (Schneider 1975). Annual survivalin bluegill Wolf Lake Hatchery) for supplyingtiger mu- and minnow ponds was 53 and 79%, respec- skellungefor laboratoryexperiments; A. Stoltz tively. After adjustmentof growth rates for (Managerof Hebron NationalHatchery, Ohio) density,walleyes were 41% heavier in the min- for supplyingbluegills for pond experiments; now pond than in the bluegillpond 1 year after and G. Beyerie,E. Crossman,and T. C. Grubb, stocking(Schneider 1975). Thus, our experi- Jr. for criticallyreviewing the manuscript.This ments support Schneider'sfindings in suggest- researchwas supportedin part by funds from ing that pellet-reared (naive) fishesmay have the Federal Aid in Fish Restoration Act under difficulty preying on bluegills. Dingell-JohnsonProject F-57-R. In additionto prey type, conversionof naive predators to live forage will be influencedby References prey size.Cost/benefit calculations suggest that ALEX^•)ER, R. McN. 1967. Functional design in larger hybrids would consumea wider size fishes.Hutchinson, London, England. range of prey than smallerones once stocked BE¾•RL•,G. B. 1978. Survival, growth, and vulner- in the wild. Food habit studiesof young-of-year ability to angling of northern pike and walleyes stockedas fingerlings in smalllakes with bluegills walleyes(60-390 mm) in Lake Erie supportthis or minnows.American FisheriesSociety Special contention. Large walleyesconsume a wider Publication 11:135-139. range of prey sizesthan smallerones (Parsons B•¾•RLE,G. B., ^• j. E. WILLIAMS. 1968. Some 1971).With more prey availableto them,larger observationsof food selectivityby northern pike hybridmuskellunge should enjoy better growth in aquaria. Transactionsof the American Fish- and survivalthan their smallercounterparts. In eries Society97:28-31. EMLE•,J. M. 1966. The role of time and energy in supportof this suggestion,stocking success of food preference. The American Naturalist 50-mm northern pike appeareddirectly related 100:611-617. to availabilityof smallforage fish, whereas these FLICKINGER,S. A., ANDJ. H. CLARK.1978. Manage- correlations were not evident with 377-mm ment evaluationof stockednorthern pike in Col- northern pike (Flickingerand Clark 1978). orado's small irrigation reservoirs. American Resultsfrom our study and others suggest FisheriesSociety Special Publication 11:284-291. GRAFF,D. R. 1978. Intensive culture of esocids:the that prey type and sizemay influencestocking current state of the art. American Fisheries So- successof predators. Naive tiger muskellunge cietySpecial Publication 11:195-201. stocked in centrarchid-dominated reservoirs H^LL,D. J., ^•) E. E. W•R•ER. 1977. Seasonaldis- PREY SELECTION BY TIGER MUSKELLUNGE 209

tribution and abundance of fishes in the littoral port 1833. MichiganDepartment of Natural Re- zone of a Michigan lake. Transactionsof the sources,Ann Arbor, Michigan, USA. American FisheriesSociety 106:545-555. SGHOENER,t. W. 1971. Theory of feeding strategies. HOLLANDER,M., ANDD. A. WOLFE.1973. Nonpara- Annual Review of Ecology and Systematics metric statisticalmethods. John Wiley and Sons, 2:369-404. New York, New York, USA. STEIN,R. A. 1977. Selectivepredation, optimal forag- JOHNSON,L. D. 1969. Food of angler-caughtnorth- ing, and the predator-preyinteraction between ern pike in Murphy Flowage.Wisconsin Depart- fish and crayfish.Ecology 58:1237-1253. ment of Natural Resources Technical Bulletin WARE,D. M. 1972. Predation by rainbowtrout (Sal- 42. mogairdneri): the influenceof hunger, prey den- JOHNSON,L. D. 1978. Evaluationof esocidstocking sity, and prey size.Journal of the FisheriesRe- program in Wisconsin.American FisheriesSo- search Board of Canada 29:1193-1201. ciety SpecialPublication 11:298-301. WEITHMAN, A. S., AND R. O. ANDERSON. 1977. Sur- KISALALIOGLU,g., AND R. N. GIBSON. 1976. Prey vival, growth, and prey of Esocidaein experi- 'handlingtime' and its importancein food selec- mental systems.Transactions of the American tion by the 15-spinedstickleback, Spinachia spi- FisheriesSociety 106:424-430. nachia.Journal of Experimental Marine Biology WERNER,E. E. 1974. The fish size, prey size, han- and Ecology25:151-158. dling time relation in severalsunfishes and some MAUCK,W. L., ANDD. W. COBLE.1971. Vulnerability implications.Journal of the FisheriesResearch of somefishes to northernpike (Esox lucius) pre- Board of Canada 31:1531-1536. dation. Journal of the FisheriesResearch Board WERNER,E. E. 1977.Species packing and niche com- of Canada 28:957-969. plementarityin three sunfishes.The American PARSONS,J. W. 1971. Selectivefood preferencesof Naturalist 111:553-578. walleyesof the 1959 year classin Lake Erie. WERNER,E. E., ANDD. J. HALL. 1974. Optimal for- Transactionsof the American FisheriesSociety agingand the sizeselection of the bluegillsunfish 100:474-485. (Lepomis macrochirus ). Ecology 55:1042-1052. SCHNEIDER,J. C. 1975. Survival, growth, and food ZAR,J. H. 1974. Biostatisticalanalysis. Prentice-Hall, of 4-inch walleyesin pondswith invertebrates, EnglewoodCliffs, New Jersey,USA. sunfishes, or minnows. Fisheries Research Re-