TRANSACTIONS of the AMERICAN FISHERIES SOCIETY

Volume 116 September 1987 Number 5

Transactionsof the American FisheriesSociety 116:677-691, 1987 ¸ Copyright by the American FisheriesSociety 1987

Diet and Selection of Major Prey Species by Lake Michigan Salmonines, 1973-1982

DAVID J. JUDE, FRANK J. TESAR, SCOTT F. DEBOE, AND TIMOTHY J. M•I•I•eR Great Lakes ResearchDivision, Universityof Michigan Ann Arbor, Michigan 48109, USA

Abstract.-- To elucidateprey preferences,we quantifiedstomach contents of 1,231 salmonines collectedfrom inshore(21 m or shallower)southeastern Lake Michigan during 1973-1982. Pred- ators ate 12 speciesof fish. Alewife Alosa pseudoharengusmade up 48-79% by weight of the diet of brown trout $almo trutta, chinook salmon Oncorhynchustshawytscha, coho salmon Oncorhyn- chuskisutch, lake trout Salvelinusnamaycush, and rainbow trout Salmo gairdneri. Alewives eaten rangedfrom 23 to 245 mm total length;46% were 150-200 mm. Rainbow smelt Osmerusmordax eaten were 21-245 mm long and made up 4-11% of the salmonines'prey. There were significant direct linear relationshipsbetween lengths of alewivesand rainbow smelt eaten and lengthsof the five predators.Alewives are currently declining in Lake Michigan. If their population collapses, there should be a shift to alternative prey species.We have seen no such shift through 1982, althoughmore recentdata of other investigatorsshow a declinein the importanceof alewife in salmoninediets. Diets of the midwater-feedingchinook and coho salmonwere heavily dominated by the pelagicalewife, whereasbrown and lake trout diets were more diverse.This suggeststhat trout shouldhave better survival and growth than salmon, becausetrout would be able to utilize the more benthic yellow perch, Percafiavescens, rainbow smelt, and, to some degree,bloater Coregonushoyi. The latter speciesare becomingmore abundantwith the declinein alewife.Under the current salmonine stocking regime, alewives will continue to supply a lower and variable portion of the salmoninediet, and predatorypressure on alewifeshould lead to increasesin endemic prey species'populations.

The present Great Lakes salmonine fishery is (Stewart et al. 1981). Information on the diet of very important to sport fishermen. The five sal- salmonines is important for successful manage- monine species that inhabit most of the Great ment of the predator-prey system in the Great Lakes, lake trout Salvelinus natnaycush,rainbow Lakes. In the case of Lake Michigan, such diet trout Saltno gairdneri, brown trout Saltno trutta, information can help in the monitoring of any coho salmon Oncorhynchuskisutch, and chinook substantial changein prey stocks or in predicting salmon Oncorhynchus tshawytscha, are main- the way salmonines may respond to such a change tained largely by stocking in quantities unrelated (Stewart et al. 1981). Diet information, along with to the abundances of prey on which they depend. energeticsmodels, can be used by fishery man- The potential for predator-prey destabilization is agersto help selectthe most appropriate predator great if stocksof alewifeAlosa pseudoharengus, to exploit available prey should the alewife pop- the major prey species, collapse (Jude and Tesar ulation collapse. 1985), or if predators outstrip their food supply During 1973-1982, the period that this paper

677 678 JUDE ET AL. addresses,the prey fish speciessustaining the sal- removal from the nets. Fish less than 1 kg were monines in Lake Michigan were mainly alewife weighed to the nearest 0.1 g; larger fish were and rainbow smelt Osmerus mordax and, occa- weighed to the nearest 25 g. Fish were measured sionally, sculpinsCottus spp. From 1980 to 1982, to the nearest millimeter total length (TL), and the alewife population declined 86% in south- their sex was determined. Details on methods are eastern Lake Michigan, while abundances of yel- describedby Judeet al. (1979) and Jude and Tesar low perch Percafiavescens and rainbow smelt in- (1985). creased (Jude and Tesar 1985). Populations of Stomach processing.-- Stomachswere removed yellow perch and, to some degree, rainbow smelt from juvenile salmonines(< 20 cm long) collected are believed to be depressed when alewives are mostly with seinesin 1973 and 1974, but no stom- abundant (Smith 1968). Starting in 1977, the pop- achs were examined quantitatively in 1975-1982. ulation ofbloaters Coregonushoyi expanded (Jude However, fish collected in later years had their and Tesar 1985), presumably as a result of both stomachs opened and examined qualitatively as the 1976 closure of commercial fishing for this part of our routine fish processing.No dramatic speciesin Lake Michigan and the decline of ale- changes in diet were noted from 1973-1974 ob- wife. In light of these events, our objectives were servations. Because these fish ate insects almost to establish the composition of the diet of the five exclusively, lack of quantitative data over the lat- most important salmonines inhabiting Lake ter period of the study when alewife populations Michigan, to determine what size and speciesof declined should not affect conclusions drawn about prey they selected by season, and to relate what the diet of small salmonines in the near-shore zone. was eaten to the sizes and abundancesof prey fish Stomachs were preserved in 10% formaldehyde populationspresent in Lake Michigan. Our results solution. Stomach contents were later examined, did not indicate that a shift to other prey occurred and total volumes (cm3) were determined volu- in the face of the alewife decline from 1980 to metrically, then converted to grams on a one-to- 1982. one basis. Percent composition of prey items eaten was estimated visually and was used to calculate Methods the weight (g) contributed by each food group. Field sampling. --Prey and predatory fish were Food items were grouped into four major cate- collected from southeasternLake Michigan at two gories: fish, invertebrates, fish eggs (mostly ale- sites,one 16 and the other 24 km south of the St. wife), and plant materials (macrophytes and al- Joseph River mouth. Sand was the predominant gae).The highestpercentage that fisheggs and plant bottom type in the area. Stations were established material contributed in any season or size group at water depths of 1, 6, 9, and 21 m. was 4.4% for rainbow trout. Three gear types were deployed to collect prey Predators longer than 20 cm were examined in and predator fish by day and night once a month each study year. Individual prey were removed from April to November 1973-1982. Seines were from stomachs,identified if possible,then weighed used at three 1-m stations, and trawling and gill- to the nearest0.1 g. Lengths (to nearestmm) were netting were conducted at two 6- and two 9-m recorded for whole prey fish and estimated for stations. Supplementary gillnetting and trawling partially digestedfish. Data from detailed analysis were occasionally conducted separately at a 6-, a of fish less than 20 cm collected in 1973-1974 are 9-, and a 21-m station. Fish were also seinedand pooled in the tables and figures with data from gillnetted during December and January-March fish 30 cm or shorter collected in 1973-1982. on an irregular basis. The beach seine was 38 x Data analysis. --For each predator species,food 1.8 m (6.35-mm mesh) and was hauled twice in data were analyzed by percent weight composition the same direction parallel to shore for nonover- and percent frequency of occurrence. Data were lapping 61-m stretchesat each station. Gill nets grouped for comparisons according to season were 160 x 1.8 m with 12 bar-mesh sizes from (spring: March-May; summer: June-August; fall: 1.3 to 10.2 cm and were set for approximately 12 September-November). Only 18 predators had h during the day and 12 h at night. The trawl had eaten during winter so only brief summaries of a 4.9-m headrope and 1.9-cm-bar mesh, and was these data are presented. Preliminary analysis of towed at a speedof 5 km/h. Tows were duplicated all data by predator-length interval showed a and were 10 min in duration. Most prey fish col- change in diet at about 30 cm, so we combined lected were frozen as soon as possible, whereas data into "small fish" (-<30 cm) and "large fish" salmonineswere usually processed within 4 h after (> 30 cm) categories.To examine changesin feed- FEEDING BY LAKE MICHIGAN SALMONIDS 679 ing activity of predators over seasons,we calcu- (Table I). Fish were collectedevery month of the lated the percentageof the total number of each year, including winter, when 18 fish with food in size group captured with food present, and then their stomachs were taken (not in Table 1). used chi-square tests of independence to deter- mine any significantdifferences. If differenceswere Brown Trout found, pairwise tests were performed to find dif- Of 331 small brown trout stomachs examined, ferences among seasons. 71% contained food. Average lengths of those fish Measures of food intake relative to fish size were with food were similar among seasons(P = 0.717). determined by calculating an "index of fullness" A significantly lower proportion of small brown (weight of stomach contents as a percentageof trout contained food in spring than in fall or sum- total body weight of each predator: Hyslop 1980). mer (P = 0.003; Table 1). Fish made up the ma- Index values were log•0-transformedand subject- jority of the total weight of prey eaten for each ed to analysis of variance performed to ascertain season, averaging 90% in spring; invertebrates differences in feeding intensity among seasons, dominated remaining weights (Figures 1, 2). Ale- Only large lake trout, brown trout, and coho salm- wives made up 7-37% of the diet over all three on provided sufficient sample sizes for this anal- seasons. Rainbow smelt and spottail shiner were ysis. For each salmoninc species,regression anal- preyed on in summer and sculpins were prey in ysis was used to establish a relationship between spring (Table 2; Figure 1). Fish eggswere eaten length of predator and length of alewife and rain- during all three seasons. bow smelt prey (or average length if more than Of large brown trout, 56% contained food, the one was found in a stomach; this was done to highest percentage of all large salmonines. This maintain independence among samples). In ad- proportion was significantly lower in fall than in dition, an analysis of covariance was performed spring or summer (chi-square, P = 0.001; Table (predator length was the covariate) to determine 1). Indexes of fullnessfor each seasonwere similar any significantdifferences in mean size of alewife (P = 0.122; Table 1). Fish accounted for almost prey among seasonsfor large lake trout, brown 100% of the total weight of food eaten, and in- trout, and coho salmon (sample sizes were too vertebrates (chironomid adults, crayfish, Gam- small for chinook salmon and rainbow trout). If marus sp., and snails) made up the remainder a significantdifference was found, Scheft%pairwise (Figures 1, 2). comparisons were used to determine which sea- Among identified fish eaten, alewives contrib- sons' mean lengths were significantly different. uted approximately 78% of the total weight in Mean numbers ofalewife prey per stomach were spring and 90O/oin summer and fall (Figure 1). also calculated for each season,and chi-square tests Most remaining fish prey were rainbow smelt, of independencewere used to ascertainsignificant which decreased in importance from spring differences.We grouped data into four categories: throughfall. Some young-of-the-yearbloaters were stomachs containing one, two, three, and four or eaten in fall. In winter, seven large and two small more alewives. Again, if a dependence was found brown trout ate gizzard shad and rainbow smelt. across seasons, pairwise tests were performed. In spring, rainbow smelt were 13% of the total Similar tests were performed for data on rainbow weight of brown trout diets; sculpins made up 5% smelt prey but data were pooled for all large sal- (Figure 1). Brown trout diets contained much monines because of the small sample size. For higher proportions of sculpins, spottail shiner, tests of seasonaldependence on number of prey bloater, and johnny darter than occurred in the per stomach, only three intervals (1, 2, 3+) were diets of other salmonines(Table 2; Figure 2). used.To addressselectivity and availability of prey Average length of alewives consumed by large fish, length frequencieswere compared for alewife brown trout was significantly smaller in fall (61 and rainbow smelt in predator stomachs (for all mm) becausemore young of the year were eaten salmonines combined) and for those captured in then than in spring (139 mm) and summer (P = field gear for each season(pooled over 10 years). 0.001; Table 3). In addition, the average number of 5.3 alewives per stomach was significantlyhigher Results (P = 0.02) in fall than in spring or summer. During 1973-1982, we collected 4,894 salmo- nines, among which food was present in 1,065 Chinook Salmon large and 1,352 small fish; contents were quanti- Of the small chinook salmon captured during fied for 848 (80%) and 365 (27%), respectively spring and summer, 93-95% had food in their 680 JUDE ET AL.

TABLE1.--Sample sizesand percentagesof fish with food in stomachsfor small and largesalmonines caught in southeasternLake Michigan, 1973-1982. Mean indexesof stomachfullness (% of body weight) are given for large fish where sample sizes were adequate. Seasonalvalues for each speciesfollowed by the same letter are not significantlydifferent (chi-square tests and analysesof variance; P < 0.05).

Fish -<30 cm total length Fish >30 cm total length Stomach contents Meanweight (meanñSD)a Stomachs (g) of stomach Stomachs % with pro- contents % with pro- Weight (g) % of body Season N food cessed (mean_+SD)a N food cessed per stomach weight

Brown trout Spring 139 62z 25 2.0+3.8 183 62z 81 37.5_+43.2 1.8+1.6 z Summer 136 76y 52 1.8_+2.8 94 72z 52 19.5ñ22.4 1.2ñ1.1 z Fall 56 84 y 26 1.0ñ0.9 152 38 y 27 17.2ñ22.1 1.4_+1.6 z

Chinook salmon Spring 301 95z 20 0.3ñ0.3 168 14z 15 45.1ñ57.8 Summer 493 93 z 58 0.3ñ0.4 25 56 y 10 6.0_+4.1 Fall 98 67 y 30 2.5ñ2.1 52 17 z 5 12.7ñ9.0

Coho salmon Spring 85 92z 25 0.5_+0.4 307 44z 119 17.1ñ16.6 1.4ñ1.2 z Summer 44 61y 23 1.0ñ0.7 67 87y 54 38.5ñ27.8 2.0ñ1.2 y Fall 5 40 I 8.0 58 21 x 9 9.9ñ7.9

Lake trout Spring 23 70 z 9 0.5ñ0.4 278 73 z 172 51.3ñ43.8 1.7ñ 1.3 z Summer 30 70 z 12 0.9_+0.6 227 66 z 143 29.0ñ24.8 1.0ñ0.8 y Fall 10 80 z 5 3.2_+3.5 1,627 12 y 148 23.7_+19.4 0.8ñ0.7 y Rainbow trout Spring 63 84 z 39 1.6_+1.3 32 12 z 5 12.6ñ 11.5 Summer 54 96 z 23 2.0 ñ 1.6 9 44 z 3 34.3ñ40.4 Fall 46 83z 17 3.5_+4.7 32 34z 5 13.2ñ15.7

a Stomachs with food only. stomachs, but the percentagedeclined significant- erage length (138 mm) and number per stomach ly to 67% in fall (Table 1). Diets were composed were highest in spring (Table 3). of fish and invertebrates(mostly chironomid adults and terrestrial insects) in spring and summer, Coho Salmon whereasfish dominated in fall (Figure 1). The main Eighty percent of all small coho salmon cap- reason for this (and for differencesin percentage tured contained food. The percentagewas signif- of fish stomachs containing food) is most likely icantly higher in spring than summer or fall (P = that average predator length was much greater in 0.001), although sample size was small during fall fall than in spring or summer. Chinook salmon (Table 1). The dominant food group in the diet ate rainbow smelt and spottail shiner in springand during spring and summer was invertebrates (ter- summer (Table 2). Chinook salmon that were eat- restrial insects and chironomid pupae), but fish ing fish in fall preyed most heavily upon alewives; contributed 40% by weight (Figure 1). Occurrence rainbow smelt and trout-perch were also eaten. of fish prey in the diet increased from spring to Of all large salmonines,chinook salmon had the summer (Table 2), as did average length of small highestpercentage of empty stomachs(79%). Per- coho salmon, which was significantly greater in centage of feeding fish was significantly higher in summer than in other seasons(P = 0.001). summer than in spring or fall, but mean weight For the 10-year period, 48% of the large coho per stomach was much lower in summer (Table salmon contained food. Significantly more fish 1). Stomach contentswere exclusivelyfish (Figure were feeding in summer than in spring or fall sea- 2); alewives and rainbow smelt were the only iden- sons,and significantlymore fed in spring than in tified fish prey (Table 2). Rainbow smelt made up fall (P < 0.001; Table 1). Mean values for index a greaterproportion of the weight of identified fish of fullnesswere significantlyhigher in summerthan prey in summer than spring. Alewives were the in spring(Table 1). Alewives made up a large pro- only identified fish prey found in fall. Alewife av- portion of the diet in all seasons(Figure 1) and FEEDING BY LAKE MICHIGAN SALMONIDS 681

SPRING SUMMBR FALL SPRING SUMMBR PALL BROWN N=25 N=52 N--26 N--81 N---52 N---27 TROUT

LAKE N:=9 N=12 N=5 N=172 N=143 N=148 TROUT

RAINBOW N=59 N=23 N=17 N=5 N--3 N=5 TROUT

CHINOOK N--58 N=30 N=15 N=10 N=5 SALMON

COHO N--23 N=I N=-119 N=54 N=9 SALMON

RAINBOWSMELT • INVERTEBRATES • SPOTTAILSHINER l ALEWIFE • COTTIDS • UNIDENTIFIED • MISC. FIGURE1.--Seasonal percent composition by weightof the diet of five speciesof salmoninescollected from southeasternLake Michigan. Data were pooled over 1973-1982. Left panel includes predators 30 cmlong or less; rightpanel includes fish longer than 30 cm.Miscellaneous (misc.) includes yellow perch, trout-perch Percopsis omiscomaycus,bloater, johnny darter Etheostoma nigrum, gizzard shad Dorosoma cepedianum, ninespine stick- lebackPungitius pungitius, bluegill Lepomis macrochirus, fisheggs, and plants. Spottail shiner is Notropis hudsonius.

overallcontributed 78% by weightto thediet (Fig- Table1). As withlarger lake trout, the total weight ure 2). The remainderof the diet was predomi- of preywas made up almost entirely of fish(Figure nantly rainbowsmelt, particularly in spring.In- 2). Someinvertebrates were eaten in springand frequentoccurrences ofspottail shiner and sculpins summer.Small lake trout were significantly longer in spring were also noted. in fall than in springor summer. The averagelengths of alewivesin largecoho In spring,small lake trout ate mostlyalewives salmon stomachs were significantly different and some rainbow smelt (Figure 1). In fall, ale- among all seasons(P < 0.001); alewives eaten in wivesdominated (97%), whereas, in summer,53% summer were the longest(Table 3). Alewives con- of the diet was rainbow smelt; the remainder of sumedin fall werethe shortest,and the average the diet included ninespine stickleback, trout- number per stomachwas highest.However, the perch,and johnny darter (Figure 1). number per stomach was statistically similar Among 2,132 largelake trout captured,26% had among seasons(P = 0.093). eatenrecently. Feeding incidence was significantly Lake Trout lowerin fall thanin springand summer (Table 1), becausespawning lake trout in fall seldom fed. Of 63 smalllaketrout examined,70% contained The indexof fullnessfor springwas significantly food;there were no significantdifferences in pro- higherthan it wasfor summeror fall (Table 1). portionof feedingfish among seasons (P = 0.807; Amongall salmonines,large lake trout generally 682 JUDE ET AL.

N=105

AL 20.2• N=160

._.•INV SM 11.4•[ BROWN 11.9•[ 20.5•[ • MISC TROUT $P 2.0•[

0,5% kilSC UNID 0.8•[ uc __/ 28.4•[ 9.3•[ N=26 N=465 UNID 8.3•[

UC LAKE 3.7x kil$C TROUT aisc 2.ex 8,o•[ INV 2.7• UNID UNID 9.5•[ 8.2• Ski

19.2% UC N=15 8.4•[Ski

RAINBOW • • • • kilSC [.•____kilSC TROUT ••--- 3.1X INV 0.2X AL• 12.4•[ Ski __ UNID 2.4..I•[ 5.5•[ 33.7•, N=108 N=,.•O

1.1%

CHINOOK uc SALMONIavkilSC14.8•, ' • • t.4x 1.7% UNID

ski • "-•-ZZ/--• UN•D Ski 12.1X 28.3• I0.5•[ N=49 N=182 COHO 14N7•

UNID Ski • •'•Y--•-/--'• UNID 13.5% 13.9% 25.8•[ SM 7.6•[ FIGURE2.--Percent compositionby weight of the diet of five speciesof salmoninescollected from southeastern Lake Michigan. Data were pooled over 1973-1982. Left panel includespredators 30 cm long or less;fight panel includesfish longer than 30 cm. Misc. (miscellaneous)is definedin Figure 1 and includesall categoriesthat composed lessthan 1% of the total. Prey species:AL = alewife; SM = rainhow smelt; SP = spottail shiner; UC = sculpins (Cottus spp.); UNID = unidentified fish; INV = invertebrates. FEEDING BY LAKE MICHIGAN SALMONIDS 683

T^BLE 2.--Seasonal values for percent frequency of occurrenceof fish prey in stomachsof small and large salmoninescaptured in southeasternLake Michigan, 1973-1982. Empty cells mean 0.

Fish <30 cm total length Fish >30 cm total length Sample size and fish prey Spring Summer Fall Spring Summer Fall

Brown trout Number of stomachs 11 25 19 78 52 27 Fish prey (% frequency) Alewife 18 16 37 62 69 63 Bluegill 9 Johnny darter 3 Rainbow smelt 8 18 14 7 Spottail shiner 8 5 4 Unidentified cot fid 18 18 2 Unidentified fish 64 84 63 35 38 44 Bloater 4

Chinooksalmon

Number of stomachs 8 9 24 15 10 5 Fish prey (% frequency) Alewife 46 73 40 80 Rainbow smelt 38 33 8 33 30 Spottail shiner 12 Trout-perch 4 Unidentified fish 62 67 50 20 30 40

Cohosalmon

Number of stomachs 10 13 I 116 54 9 Fish prey (% frequency) Alewife 20 8 100 65 78 67 Rainbow smelt 38 17 9 11 Spottail shiner 2 Unidentified cottid 2 2 Unidentified fish 70 54 100 35 35 33 Yellow perch 10 2

Lake trout Number of stomachs 6 9 4 171 143 146 Fish prey (% frequency) Alewife 50 75 70 70 63 Gizzard shad 4 Johnny darter 11 Ninespine stickleback I 1 Rainbow smelt 17 56 18 5 8 Spottail shiner 4 2 Trout-perch 11 Unidentified cotfid 11 5 2 1 Yellow perch I I 1 Unidentified fish 33 33 25 37 40 38 Bloater 1

Rainbow trout

Number of stomachs 9 8 4 3 2 Fish prey (% frequency) Alewife l l 75 67 100 Rainbow smelt 12 50 Spottail shiner 25 Unidentified cottid 25 Unidentified fish 89 38 5O 33

contained the highestmean weight of food during Remaining fish eaten included rainbow smelt in all three seasons. springand fall and gizzard shad in fall (Figure 1). Almost exclusively,fish were eaten by large lake The average length of alewife prey was signifi- trout in all seasons(Figure 2). Alewives dominat- canfly shorter in fall than in spring or summer ed the diet, particularlyduring summer (Figure 1). (P < 0.0001), but significantly higher numbers of 684 JUDE ET AL.

TABLE 3.--Seasonal mean number and mean length Rainbow Trout (mm) of alewives eaten by large salmonines(>30 cm). Among small rainbow trout, 88% had food in Values for each predator followed by the same letter are not significantlydifferent betweenseasons (analyses of their stomachs and no significant differences covariance or chi-square tests; P < 0.05). Numbers of among seasonswere found in this ratio. In spring alewives measured are in parentheses. Hypothesis tests and exclusively in summer, invertebrates domi- were not performed for chinook salmon and rainbow nated the diet (Figure 1). Fish were predominantly trout becauseof insufficient sample sizes. eaten in fall, although invertebrates occurred more

Alewife prey frequently than fish then. Fish eggsand plant ma- terial composed 6% of the spring diet. Alewives Number Mean number of per stomach ñ (65%) and rainbow smelt (2%) were the only iden- Season stomachs SD Mean lengthñSD tified fish prey eaten. Brown trout Over the 10-year study, 26% of large rainbow Spring 48 2.6ñ 1.9 zy 139+49 z (120) trout contained food; this percentagewas similar Summer 36 1.6ñ 1.0 z 131 ñ49 z (49) among seasons(Table 1). Fish, mostly alewives, Fall 17 5.3ñ5.7 y 61 ñ45 y (83) made up a large part of the diet over all seasons Chinook salmon (55%; Figure 2), whereas a few invertebrates (ex- Spring 11 2.2+ 1.8 138+50 (22) clusively terrestrial insects) were eaten in spring Summer 4 1.0ñ0.0 119ñ53 (4) Fall 4 1.8ñ0.9 92+41 (6) and fall (5-8%; Figure 1).

Coho salmon Spring 75 2.1ñ1.8 z 122ñ42 z (130) Totals by Species Summer 42 1.9ñ 1.0 z 169ñ28 y (59) Fall 6 4.2_+4.3 z 51+_22 x (20) When data on fish consumedby all salmonines

Lake trout were pooled for each predator by small (<-30 cm) Spring 120 2.1 ñ 1.5 z 157+36 z (233) and large (>30 cm) size groups (Figure 2), the Summer 100 1.6ñ 1.0 y 160ñ29 z (140) dominance of alewives in the diet was shown Fall 92 1.7ñ1.7 y 138ñ43 y (136) clearly. For small salmonines, alewives made up Rainbow trout from 11 to 61% by weight of the diet (Figure 2). Summer 2 9.5_+9.2 104ñ37 (19) Exclusive of unidentified fish (8-28%), rainbow Fall 2 16.0+12.7 53_+6 (32) smelt was the second- or third-most important component of the fish diet of all predators except rainbow trout. Diets of large lake trout, coho alewives were found in stomachs in spring than salmon, and chinook salmon all were about 80% in summer or fall (P = 0.0032; Table 3). The low- alewife by weight. Brown trout and rainbow trout est yearly weight fraction of dietary alewives oc- ate lesser amounts. Unidentified fish or rainbow curred in 1982 (data not shown)when the alewife smelt were next in importance for all large pred- population was at its lowest level over the 1O-year ators. Brown trout (6%) and lake trout and rain- study (Jude and Tesar 1985). We also examined bow trout (3%) had the greatest proportions of the our data (not shown) for changesin the size of more uncommonly eaten species,such as slimy alewives eaten between 1980-1982 (the time of sculpin Cottus cognatus, spottail shiner, gizzard the alewife population decline) and 1973-1979. shad, bloater, yellow perch, and johnny darter. We found no significant change (chi-square test) Sculpins made up almost 4% of the diet of brown in the size (mode at 170 mm) of large alewives trout and 2% of the diet of rainbow trout. Chinook eaten by all predators. However, during 1973- and coho salmon ate alewives and rainbow smelt 1979, young-of-the-year fish (20-60 mm) com- almost exclusively; only 0.8% of the coho salm- posedabout 20% of the diet numerically, but fish on's diet was other fish species. of this size were seldom eaten in 1980-1982 and Average number of rainbow smelt eaten by all were replaced by 80-mm yearlings in the diet. large predators combined was similar among sea- There was a dramatic decline in young-of-the-year sons (mean, 1.7; N = 107; Kruskal-Wallis test, fish abundance starting in 1980 and concomi- P = 0.67). Averagelength of rainbow smelt preyed tantly an expanded presenceof yearlings in the upon by all large predatorscombined was signif- nearshorezone in May-July; we had not observed icantly greater in spring (144 mm; N = 103) than such expansion prior to 1980 (Tesar and Jude in summer (118 mm; N = 28) or fall (96 mm; N = 1985). Yearlingsgenerally inhabit the thermocline 34; analysis of covariance, P = 0.002). Examina- far offshore (Brown 1972). tion of plots (not shown)of length of alewife prey FEEDING BY LAKE MICHIGAN SALMONIDS 685

T^BLE4.--Regression equations relating predator length(mm) to prey length.YAL and YSM = lengthofalewife and rainbow smelt prey, respectively;X = lengthof one of the five salmoninepredators studied, _+SE. Asterisks denote P _< 0.05* or P _< 0.001'*.

Predator Regressionequation r2 N

Brown trout YAL - 4.34--+ 18.0 + 0.257+_0.0365X 0.340** 98 YSM - 40.3+33.1 + 0.213-+0.0689X 0.314' 23 Chinook salmon YAL = 7.96 +_ 14.7 + 0.213+-0.0286X 0.716'* 24 YSM = 18.6_+ 10.5 + 0.206+-0.0290X 0.807** 14 Coho salmon a YAL = -49.6+-30.4 + 0.370+-0.0576X 0.314'* 92 Lake trout YAL = 75.5-+ 12.9 + 0.125+-0.0194X 0.128'* 284 YSM = 50.3-+20.4 + 0.158+-0.0332X 0.312'* 52 Rainbow trout a YAL = 18.34+- 25.2 + 0.153 +-0.0537X 0.504* 10

a NO significantrelationship was found for rainbow smelt.

versus length of predator showed that, around a young of the year consumed in fall 1979 and 1981 predator length of 60-70 cm and longer, alewife by brown and lake trout. prey of maximum lengths (200-220 ram) were Alewife length frequenciesfrom field catchesand eaten.For rainbow smelt,predators 40-50 cm long predator stomachs showed similar spring distri- were eating maximum-size rainbow smelt (220- butions over 1973-1982 (Figure 3). Peaks in both 245 ram). The relationship between length of curves occurred at around 70 mm (yearlings) and predator and length of alewife prey was linear and 170 min. In summer, peak catches were observed significantfor all five salmonines(Table 4), where- for young-of-the-year (30-40 mm), yearling (70- as for rainbow smelt it was only significant for 80 ram), and adult (170 ram) alewives. However, brown trout, chinook salmon, and lake trout. salmoninespreyed only on the latter two groups. Differences in percentageof stomachscontain- In fall, there was one mode in field catches at 30- ing food were examined by speciesand sex; no 40 mm (young of the year) and one at 150-160 significant differences were found between sexes mm (obscured on Figure 3 because of the large for large chinook salmon and brown trout. Sig- catches of young of the year); fish of both sizes nificantly more large rainbow trout and coho were preyed on heavily by salmonines.In winter, salmon males than females contained food (P = salmonines only preyed on 170-180-ram adults, 0.0031 and 0.0157, respectively),whereas, for lake despite the presenceof a large group of yearlings. trout, significantly more females contained food Plots of field catches of rainbow smelt and those than males (P < 0.001). preyed on by salmoninesin spring(Figure 4) both revealedpeaks in abundanceat 60 mm (youngof Selectivityand Vulnerability the year) and 140-150 ram, but the predators were The numerical percentagesof prey fish in the pooled salmonine diets and in the field catch T^BLE5.--Percent composition(by number) of those showed that alewives were eaten in even higher identified fish eaten by salmoninepredators and of fish proportions (78%) than would be predicted from collectedin southeasternLake Michiganduring monthly field catchcomposition (61%; Table 5), despitethe sampling, 1973-1982. near-shore bias in prey collections. A similar pat- Eaten by tern was noted for rainbow smelt and sculpins. predators Field catch Spottail shiners made up 21% of the field catch, Species (N-1,656) (N= 1,129,606) yet were only represented sparsely (2%) in sal- Alewife 78.5 61.2 toonine diets. Bloaters,historically part of the sta- Spottail shiner 1.8 20.7 Yellow perch 0.4 7.2 ple of Lake Superior lake trout diets (Dryer et al. Rainbow smelt 14.3 6.8 1965), made up an average of 1.7% of our field Trout-perch 0.2 1.9 catch; recently their populations have increasedin Bloater 0.5 1.7 Lake Michigan, and in 1981 and 1982 they made Johnny darter 0.4 0.2 Gizzard shad 0.9 0.1 up 19 and 21% of our field catches,respectively Cottu$ spp.a 2.9 0. I (Tesar and Jude 1985). Despite this higher abun- Ninespine stickleback 0.1 <0.1 dance, salmonines seldom ate bloaters, which rep- Bluegill 0.1 <0.1 resentedonly 0.5% of their diets. Those eaten were a Includesslimy and mottled sculpins(C. bairdl). 686 JUDE ET AL.

SPRING

40 40-

30-

• •o' SUMMER FIELDCATCH 50 SUMMER FIELDCATCH 40 • PREY

0 FALL

2 4 6 6 10 12 14 16 18 20 22 24 26 0 2 4 6 8 10 12 14 16 18 20 22 24 26 TOTALLENGTH (cm) TOTALLENGT. (ore) F•OVRE3.-- •ngth-frequency distributions (% of total FIGURE4.--Length-frequency distributions (% of total number) for alewives collected during field s•eys and number) for rainbow smeltcollected during field surveys for thoseeaten by five speciesof salmoninesd•ng spring, and for thoseeaten by five speciesof salmoninesduring summer, and fall in southeasternLake Michigan. Data spring,summer, and fall in southeasternLake Michigan. were pooled over 1973-1982. N = 691,80• for field Data were pooled over 1973-1982. N = 77,204 for field catches and 1,128 for number of alewives eaten. catches and 215 for number of rainbow smelt eaten.

selective for the larger prey. This pattern was also cupying preferred colder temperatures offshore, as found for summer and fall comparisons, but se- was found by Enge! and Magnuson (1976). lectivity was not as consistent or pronounced. Ware (1971, 1972), in discussingthe risk of prey to a piscine predator, cited handling time, prey Discussion exposure, food size, hunger, and previous expe- The near-shore zone is an important nursery rience as factors affecting which prey are eaten. area for juvenile salmonines, except for lake trout, For most prey speciesin Lake Michigan, we felt which are known generallyto inhabit water deeper exposure and predator experience were most im- than adults do (Martin 1951). We found that the portant. Spottail shiners were the second-most diet of juvenile lake trout was over 60% alewife abundant fish collectedin the study area (Jude and and 19% rainbow smelt by weight. In Lake Mich- Tesar 1985), yet few were eaten. They are gener- igan, Eck and Wells (1986) found that age-1 and -2 ally demersalfish, residingwithin the 9-m contour lake trout ate young-of-the-year alewives in De- where presumably waters are so warm that most cember and slimy sculpin in March 1982. Juvenile predators were excluded. This pattern was also lake trout in Lake Ontario ate mainly slimy scul- noted by Martin (1951) for Algonquin Park lake pins and, secondarily, alewives and rainbow smelt trout, which had to eat zooplanktonduring sum- (Elrod 1983). Juvenilesof other salmonincspecies mer stratificationbecause most prey fish were re- in our studywere often collectedin relatively warm siding inshore and above the thermocline. Slimy (> 20øC)beach-zone waters and were usually feed- sculpin in our study were eaten in spring; they ing on terrestrial insectsconcentrated on the sur- appear to be more susceptibleto predation during face. English (1983) and Kwain (1983) found that their migrationscloser to shorefor spawningand juvenile rainbow trout and chinook salmon ate nest guarding activities. They were the third-most floating insectprey. The near-shorezone provided important prey eaten and made up 32% of the these young salmonines with food and, perhaps, diets of small brown trout in spring. They were a warm-water refugium from larger predatorsoc- also fed on extensivelyin Lake Michigan during FEEDING BY LAKE MICHIGAN SALMONIDS 687

March 1982, when lake trout apparently over- another 8-34% of the diet was unidentified fish. lapped their distribution, whereas alewives, which Alewives dominated the diets of southern Lake were farther offshore, were not eaten to any sub- Michigan salmonines in 1970 (McComish and stantial degree (Eck and Wells 1986). Christie et Miller 1976) and of Lake Ontario salmonines in al. (1987) maintained that slimy sculpin were the 1983-1984 (Brandt 1986); rainbow smelt were preferredprey of Lake Ontario lake trout, and Van second in importance (which we also found). Oosten and Deason (1938) showed that Cottidae Stomachs of small rainbow trout had the highest were important food for southern Lake Michigan proportions of invertebrates (17-100% over all lake trout before their populations collapsed. three seasons);salmonine foraging behavior work Only 0.4% (six fish) of all fish eaten by salmo- done by J. Savitz (Loyola University, Chicago, nines were yellow perch, a reflection of their re- personal communication) showed that rainbow duced abundance in the 1970s. Recently, more are trout were the slowest of all salmonines studied, being consumed by salmonines, particularly dur- only catching prey in pelagic waters when reactive ing times of thermal overlap between predator and distances were small. prey during spring, during upwellings, and espe- Alewife, a marine exotic, entered Lake Michi- cially during fall (Hagar 1984; N. Kevern, Mich- gan in the 1940s (Miller 1957) in the almost total igan State University, personal communication), absence of top predators after lake trout collapsed the time of maximum energetic demands on pred- (Smith 1968; Christie 1974) and with very little ators (Edsall et al. 1974; Stewart et al. 1981). Mar- competition from the historical deepwater plank- tin (1951) found that considerable spatial overlap tivores (the "chub complex": Smith 1964). Stock- betweenlake trout and yellow perch occurreddur- ing of predators started first with lake trout in ing stratification of an Algonquin Park lake where 1965, followed by brown trout, rainbow trout, and yellow perch, especially young of the year, ex- coho and chinook salmon. For the newly stocked tended out to 12øCwater and were fed upon heavi- salmonines, alewives were abundant and easy to ly by lake trout. catch, and searchimages (Holling 1959; Marcotte Bloaters have increased dramatically in abun- and Browman 1986) were undoubtedly formed. dance, starting in 1978 (Jude and Tesar 1985). These piscivores selectedalewives over other prey Only a few were eaten during this study, in con- in Lake Michigan, a pattern that Holling (1973) trast to pre-sea lamprey Petromyzon marinus days referred to as contagious predation. We feel the (Van Oosten and Deason 1938; Dryer et al. 1965), exotic alewives are more vulnerable prey than en- when much of the lake trout diet was composed demic species such as the bloater. Alewives have of coregonines, including bloaters. Despite their not fully adapted to fresh water, as exemplified by extremely disparate population abundances, al- iodine deficiencies (Colby 1973) and by severe most twice as many gizzard shad as bloaters were mortalities in spring and during cold winters (Eck eaten by salmonines during our study. In more and Brown 1985). In contrast, cold water is op- recent studies, bloaters have made up increasing timal habitat for salmonines. Salmon are pelagic, proportions of salmonine diets, especially in fall continuously swimming predators (Lackey 1970; (J. Janssen,Loyola University, Chicago, personal Speirs 1974); thus alewife schools suspended in communication), but still not to the degree ex- midwater present vulnerable targets for these fish. pected. Bloaters may have taken up a more ben- While scuba diving in Lake Michigan, we have thic existencein responseto competition from ale- observed large schools of alewives, which were wives (Crowder and Crawford 1984), which would suspendedin the entire water column at 9 m and place them out of the pelagic zone where most were visible from long distances as a large shim- salmon search for food. In addition, alewives may mering sheet.Behavioral work by Savitz (personal still be buffering bloaters. Formation of a prey communication) showed that chinook salmon had searchimage, as Ware (1971) suggestedoccurred the highest swimming speedsof salmonines tested for rainbow trout, may be delayed, perhaps ex- and that pelagic alewives were easily detected and plaining why salmonines are still preying heavily dispatched, in contrast to demersal yellow perch, on alewives despite the increased abundance of which were often ignored. bloaters. Recently, alewife populations in southeastern Alewives were the preferred prey of all five sal- Lake Michigan have declined dramatically (Jude monines collected, making up 55-82% by weight and Tesar 1985). Implications of this decline could of the diet of large fish. The actual percentagesare be serious for the sport fishery becauseof the im- undoubtedly higher than we measured because portance of this prey in salmonine diets. Stewart 688 JUDE ET AL. et al. (1981) warned of the dangersof overstocking wives were also heavily preyed upon by lake trout salmonine predators that could eventually out- in Lake Michigan in December (Eck and Wells strip the alewife prey base. Concurrent with the 1986). The predation pattern was similar for rain- alewife decline, bloaters and yellow perch have bow smelt, except that young of the year were fed increased (Jude and Tesar 1985). We examined upon during summer and fall, a reflection of this our data for evidence that salmonines were species'more offshore nursery grounds that over- switching (Murdoch 1969; Murdoch et el. 1975) lap with the predators' distributions. The mean to alternate prey species,especially to bloater, but, lengthof rainbow smelt eaten was longestin spring despite an 86% decline in alewife abundance, a when the prey appeared to be more susceptibleto 10-fold increase in bloater abundance, and a 5-fold predation during onshore spawning runs; mean increasein yellow perch abundance, no diet shift prey length declined through summer and fall as had occurredas of 1982. Apparently, alewives were more juveniles were eaten. still being eaten selectively (depensatorymortali- Plots of predator length versus prey length for ty), buffering bloaters from increased predation. both prey species showed that predators above Lake trout in Lake Opeongo switched from yellow 60-70 cm consumed maximum sizes of prey. Sal- perch, when the latter declined, to lake whitefish monine predators greater than 70 cm could eat and lake herring Coregonusartedi (Martin 1970). much larger prey if the latter were available and In 1984, Lake Erie coho salmon switched from susceptible.A direct relationship between preda- rainbow smelt to yellow perchwhen rainbow smelt tot and prey sizes was found for lake trout and declined and yellow perch formed a strong year their prey in Lake Opeongo (Martin 1970) and in class (A. Timmerman, Lake Erie Fisheries As- southernLake Michigan in 1970 (McComish and sessmentUnit, personal communication). Chris- Miller 1976). In Lake Ontario, Brandt (i 986) found tie et el. (1987) noted that stocked lake trout in a significant relationship for only one of five sal- the Kingston basin of eastern Lake Ontario ate monine species and its alewife prey. In Lake increasing amounts of alewife as rainbow smelt Michigan, the only speciesthat could fill this void and slimy sculpin became less abundant. They would be large bloaters, rainbow smelt, and per- concluded that alewife was the third choice, as haps lake whitefish Coregonus clupeaformis, a lake trout preferentially ate slimy sculpin first, then specieseaten by lake trout in Lake Opeongo (Mar- rainbow smelt. tin 1970). A currently invading species,pink salm- Salmonine predators fed on alewives of all sizes, on Oncorhynchusgorbuscha (Kwain and Lawrie but selected different sizes at different times dur- 1981), reaches sizes to 390 mm fork length; it is ing the year. Modes in sizesof alewives eaten dur- pelagic and mostly planktivorous. This species ing spring and summer occurred at around 80 and could provide another larger prey fish. In the mid- 170 mm. These modes corresponded well with 1980s, during the period of low alewife abundance modes in our field catches, except that young of (Jude and Tesar 1985), researchersdid not ob- the year, which generally occupiedthe near-shore serve a significant relationship between predator zone (6 m or less) where water temperatures are and prey sizes because more of the smaller ale- highest (Jude et el. 1979), remained inaccessible wives were eaten (Hagar 1984; Kevern, personal to salmonine predators during summer months, communication). Offshore, young of the year and except during upwellings. These data suggestthat yearlings inhabit warmer, more pelagic waters, salmonineswere eating prey in proportion to their whereas adults tend to be closer to the bottom abundance in the field only when preferred tem- (Wells 1968; Brown 1972; Brandt et el. 1980), peratures ofpredator and prey overlapped (for ex- thus making the smaller fish more vulnerable to ample, see Martin 1951; Engel and Magnuson predatory salmon residing at middepths. As fur- 1976). Elrod (1983) also found that the incidence ther evidence, lake trout, which tend to be more of prey fish in Lake Ontario trawl catches was demersal predators, ate the largestalewives of all similar to the occurrence of these fish in lake trout predators (with one exception) during all three diets. Christie et el. (1987) found that lake trout seasons. in the Kingston basin, Lake Ontario, ate smaller The decline of alewives in Lake Michigan has prey on averagethan those trawled, but predators cascaded through other parts of the ecosystem generally ate fish of sizes that were available to (Carpenteret el. 1985). Zooplankton size structure them. During fall in our study, young-of-the-year did not changeinshore as yellow perch predation alewives were heavily preyed upon along with replaced alewife predation, but larger offshore adults around 160 mm. Young-of-the-year ale- zooplankton increasedin abundance and Daphnia FEEDING BY LAKE MICHIGAN SALMONIDS 689 pulicaria becamedominant (Evans and Jude 1986). Lake Michigan fish community toward one that Because these large Daphnia pulicaria are more has less dependence on exotic, variably recruited efficient at removing particles, including phyto- prey speciesto one that has historically abundant plankton, from the water column than was the and endemic species,such as yellow perch, bloater, former copepod-dominated zooplankton, dra- deepwatersculpin Myoxocephalus thompsoni, and, matic increasesin water clarity were observed in more recently, emerald shiner Notropis atheri- the early 1980s (Scavia et at. 1986). In addition, noides (Great Lakes Research Division, Univer- current studies (1983-1986) of satmonine diets sity of Michigan, unpublisheddata), provided that (Hagar 1984; Janssen, personal communication; these regain numerical abundance. These popu- Kevern, personal communication), have docu- lations are inherently more stable because they mented increased diet diversity, more empty evolved in Lake Michigan, they should be some- stomachs than in the early 1980s, and a shift to a what easier to manage, and they can act as im- high proportion of young-of-the-year alewives portant commercial, sport, and prey fish. There rather than adults in stomachs.However, alewives should be a reduction in transfer of toxic sub- still make up a higher-than-expectedproportion stancesto upper levels in the food chain if species of the diet. The lake should be able to sustain a such as yellow perch make up a higher proportion higher biomassof young-of-the-yearalewives than of salmonine diets. Alewives are very inefficient of older age groups, becausethe amount of food in food conversion and should concentrate more consumed that is converted to body biomass of toxic substancesthan do other specieswhich have young of the year (5%) is much higher than that higher food conversion efficiencies (Flath and ofadutts (1.3-2.8%) (Stewartand Binkowski 1986). Diana 1985; Stewart and Binkowski 1986). Christie et at. (1987) also alluded to this for ale- Eventually, salmonines may switch to the now wives in the Kingston Basin, Lake Ontario; these more abundant bloater, which should be closely authors suggestedthat there may have been changes monitored. Bloaters are more important com- in secondary production caused by size-selective mercially than alewives, so some conflict may de- predation on zooplankton that enhanced produc- velop in allocation of the bloater resource. Those tion of lower trophic levels, thus favoring growth salmonines with a more diverse diet should be of alewives. able to take advantage of the now more abundant Fish managers may now be in a unique position non-alewife prey. These include brown trout, to forge changesin prey populationsthat will ben- whose overall non-alewife prey made up 18% by efit the Lake Michigan fish community and the weight of the diet of large fish in our study, and sport and commercial fisheries, and that will pro- lake trout (11%). Both speciesappear to be more vide direction and forewarning elsewhere. Man- demersal predators than the salmon, and will en- agers must increasingly manage ecosystemswith counter more of the now abundant, underutilized uncoupled predators and prey, so the history of prey species, such as yellow perch, trout-perch, changesin Lake Michigan will provide a suite of and, to a lesser degree, btoater. There is a large possible scenarios that may occur. The alewife population of spottail shinersin near-shoresouth- population has fluctuated wildly since entering eastern Lake Michigan that was seldom preyed Lake Michigan, acting like most exotic speciesthat upon by salmonines. Yellow perch are so abun- attain dominance in new ecosystems,first increas- dant now that their growth has been depressed ing, then crashing, then maintaining some equi- through density-dependent effects (Wells 1985; librium (Smith 1970; Brown 1972). Christie et at. Great LakesResearch Division, unpublisheddata). (1987) consideredthe alewife to be integrated into Walleye might be a speciesto introduce to utilize the Lake Ontario system, because it has been an this currently untapped resource. The largely important part of the lake trout diet sincethe 1920s. unexploited deepwater scutpin also has greater po- O'Gorman and Schneider(1986) maintained that tential to be consumed by lake trout, because alewives there are very resilient and will be able deepwater scutpins were important prey for lake to withstand considerablepredation pressure.Still, trout before that predator became extinct in Lake alewife populations are difficult to manage and Michigan (Dryer et at. 1965). present substantial risk to fish managerswho must depend on this prey base for the extensive sat- Acknowledgments monine stocking programs now in place in the We thank the many members of the Great Lakes Great Lakes. Managers now have the opportunity ResearchDivision's FisheryLaboratory, who, over to continue this predatory pressure and force the the 10 years of the study, helped with data collec- 690 JUDE ET AL. tion. Particular thanks go to senior members D. Edsall, T., E. Brown, Jr., T. Yocum, and R. Wolcott. Bimber, D. Einhouse, M. Enk, J. Dorr III, G. Go- 1974. Utilization of alewives by coho salmon in dun, C. Madenjian, P. Mansfield, P. Rago, H. Tin, Lake Michigan. U.S. Fish and Wildlife Service,Great N. Thurber, and J. Tomlinson; L. Flath did some Lakes Fishery Laboratory, Administrative Report, Ann Arbor, Michigan. early processingof data. Text processingwas han- Elrod, J.H. 1983. Seasonalfood of juvenile lake trout dled by J. Huhn, M. Sweeney, and B. McClellan. in U.S. waters of Lake Ontario. Journal of Great We thank G. Stoermer, C. Schelske, N. Ringler, Lakes Research 9:396-402. D. Stewart, and two anonymous reviewers for Enõel, S., and J. J. Magnuson. 1976. Vertical and hor- constructive comments. The study was funded by izontal distributions ofcoho salmon (Oncorhynchus the American Electric Power Service Corporation, kisutch),yellow perch (Percafiavescens), and cisco Indiana and Michigan Power Company, to whom (CoregonusartediD in Pallette Lake, Wisconsin. Journal of the Fisheries Research Board of Canada we are indebted. This is Great Lakes Research 33:2710-2715. Division Contribution 468. English, K. 1983. Predator-prey relationships for ju- venile chinook salmon, Oncorhynchustshawytscha, References feeding on zooplankton in "in situ" enclosures.Ca- Brandt, S. 1986. Food of trout and salmon in Lake nadian Journal of Fisheries and Aquatic Sciences Ontario. Journal of Great Lakes Research 12:200- 40:287-297. 205. Evans, M., and D. J. Jude. 1986. Recent shifts in Brandt, S., J. Magnuson, and L. Crowder. 1980. Ther- Daphnia community structurein southeasternLake mal habitat partitioningby fishesin Lake Michigan. Michigan: a comparison of the inshore and offshore. Canadian Journal of Fisheriesand Aquatic Sciences Limnology and Oceanography31:56-67. 37:1557-1564. Flath, L. E., and J. S. Diana. 1985. Seasonal energy Brown, E. H., Jr. 1972. Population biology ofalewives, dynamics of the alewife in southeasternLake Mich- Alosa pseudoharengus,in Lake Michigan, 1949- igan. Transactions of the American Fisheries Soci- 1970. Journal of the Fisheries Research Board of ety 114:328-337. Canada 29:477-500. Haõar, J. 1984. Diets of Lake Michigan salmonids: an Carpenter, S., F. Kitcheil, and J. Hodgson. 1985. Cas- assessmentof the dynamics of predator-prey inter- cading trophic interactions and lake productivity-- actions. Master's thesis. University of Wisconsin, fish predationand herbivorycan regulatelake eco- Madison. systems. BioScience 35:634-639. Holling, C. S. 1959. Some characteristics of simple Christie, W.J. 1974. Changesin the fish speciescom- types of predation and parasitism. Canadian Ento- position of the Great Lakes. Journal of the Fisheries mologist 91:385-398. Research Board of Canada 31:827-854. Holling, C.S. 1973. Resilience and stability of ecolog- Christie, W. J., K. A. Scott, P. G. Sly, and R. H. Strns. ical systems. Annual Review of Ecology and Sys- 1987. Recent changesin the aquatic food web of tematics 4:1-23. eastern Lake Ontario. Canadian Journal of Fisheries Hyslop, E.J. 1980. Stomach content analysis--a re- and Aquatic Sciences44 (Supplement2):37-52. view of methods and their application. Journal of Colby, P. J. 1973. Response of the alewives, Alosa Fish Biology 17:411-429. pseudoharengus,to environmental change. Pages Jude, D. J., and six coauthors. 1979. Inshore Lake 163-198 in W. Chavin, editor. Responsesof fish to Michigan fish populations near the D.C. Cook Nu- environmental changes.Thomas, Springfield, Illi- clear Power Plant during preoperational years-- nois. 1973-1974. University of Michigan, Great Lakes Crowder, L., and H. Crawford. 1984. Ecologicalshifts Research Division, Special Report 71, Ann Arbor. in resource use by bloaters in Lake Michigan. Jude, D. J., and F. J. Tesar. 1985. Recent changesin Transactionsof the American FisheriesSociety 113: the inshoreforage fish of Lake Michigan. Canadian 694-700. Journal of Fisheries and Aquatic Sciences42:1154- Dryer, W. R., L. F. Erkkila, and C. L. Tetzloff. 1965. 1157. Food of lake trout in Lake Superior. Transactions Kwain, W. 1983. Downstream migration, population of the American Fisheries Society 94:169-176. size, and feeding of juvenile rainbow trout. Journal Eck, G. W., and E. H. Brown, Jr. 1985. Lake Michi- of Great Lakes Research 9:52-59. gan's capacity to support lake trout (Salvelinus Kwain, W., and A. H. Lawrie. 1981. Pink salmon in namaycush)and othersalmonines: an estimatebased the Great Lakes. Fisheries(Bethesda) 6(2):2-6. on the statusof prey populationsin the 1970s. Ca- Lackey, R.T. 1970. Seasonaldistribution of Atlantic nadian Journal of Fisheries and Aquatic Sciences salmon, brook trout, alewives, and American smelt 42:449-454. in a small lake. Journal of the Fisheries Research Eck, G. W., and L. Wells. 1986. Depth distribution, Board of Canada 27:1656-1661. diet, and overwintergrowth of lake trout (Salvelinus Marcotte, B., and H. Browman. 1986. Foraging be- namaycush)in southeasternLake Michigan sam- havior in fishes:perspectives on variance. Environ- pled in December 1981 and March 1982. Journal mental Biology of Fishes 16:25-33. of Great Lakes Research 12:263-269. Martin, N.V. 1951. A study of the lake trout Salveli- FEEDING BY LAKE MICHIGAN SALMONIDS 691

nus namaycushin two Algonquin Park, Ontario, brook trout in a Maine lake after introduction of lakes. Transactions of the American Fisheries So- landlocked alewives. Transactions of the American ciety 81:111-137. FisheriesSociety 103:396-399. Martin, N. V. 1970. Long-term effects of diet on the Stewart,D. J., and F. P. Binkowski. 1986. Dynamics biology of the lake trout and the fishery in Lake of consumptionand food conversionin Lake Mich- Opeongo, Ontario. Journal of the Fisheries Re- igan alewives: an energetics-modelingsynthesis. search Board of Canada 27:125-146. Transactionsof the American FisheriesSociety 115: McComish, T., and W. Miller. 1976. Notes on the 643-661. biology of the lake trout and other Salmonidac in Stewart, D. J., J. F. Kitcheil, and L. B. Crowder. 1981. Indiana waters of Lake Michigan. Proceedingsof Prey fishes and their salmonid predators in Lake the Indiana Academy of Science85:161-169. Michigan. Transactionsof the American Fisheries Miller, R.R. 1957. Origin and dispersalof the alewife, Society 110:751-763. Alosa pseudoharengus,and the gizzard shad, Dor- Tesar, F. J., and D. J. Jude. 1985. Adult and juvenile osoma cepedianum, in the Great Lakes. Transac- fishpopulations of inshoresoutheastern Lake Mich- tions of the American Fisheries Society 86:97-111. igan near the Cook nuclear power plant, 1973-82. Murdoch, W.W. 1969. Switchingin generalpredators: University of Michigan, Great Lakes ResearchDi- experimentson predator specificityand stability of vision, SpecialReport 106, Ann Arbor. prey populations.Ecological Monographs 39:335- Van Oosten, J., and H. Deason. 1938. The food of the 354. lake trout (Cristivomernamaycush namaycush) and Muralocr, W. W., S. Avery, and M. Smyth. 1975. of the lawyer (Lota maculosa)of Lake Michigan. Switchingin predatoryfish. Ecology56:1094-1105. Transactionsof the American FisheriesSociety 67: O'Gorman, R., and C. Schneider. 1986. Dynamics of 155-177. alewives in Lake Ontario following a mass mortal- Ware, D. 1971. Predation by rainbow trout: the effect ity. Transactionsof the American FisheriesSociety of experience. Journal of the Fisheries Research 115:1-14. Board of Canada 28:1847-1852. Scavia, D., G. Fahnenstiel, M. Evans, D. Jude, and J. Ware, D. 1972. Predation by rainbow trout: the influ- Lehman. 1986. Influence of salmoninc predation enceof hunger,prey density,and prey size.Journal and weather on long-term water quality trends in of the Fisheries Research Board of Canada 29:1193- Lake Michigan. Canadian Journal of Fisheries and 1201. Aquatic Sciences43:435-443. Wells, L. 1968. Seasonaldepth distribution of fish in Smith, S. H. 1964. Statusof the deepwatercisco pop- southeasternLake Michigan. U.S. Fish and Wildlife ulations of Lake Michigan. Transactions of the Service Fishery Bulletin 67:1-15. American FisheriesSociety 93:155-163. Wells, L. 1985. Changesin Lake Michigan's prey fish Smith, S.H. 1968. Speciessuccession and fisheryex- populations with increasing salmonid abundance, ploitation in the Great Lakes. Journal of the Fish- 1962 to 1984. Pages 13-25 in R. L. Eshenroder, eries Research Board of Canada 25:667-693. editor. Great Lakes Fishery Commission Special Smith, S.H. 1970. Speciesinteractions of the alewife Publication 85-3, Ann Arbor, Michigan. in the Great Lakes. Transactions of the American Received March 22, 1986 FisheriesSociety 99:754-765. AcceptedDecember 21, 1987 Speirs, G. 1974. Food habits of landlocked salmon and