‘1 b» s . 5 z.» 5 1 1;" g-‘:’ a t '§ I . h ~‘ 5 5 .‘ ' 5‘5 ' l b‘ “ h.‘ ‘ . 5 5 2; a .v x g ‘ 5.. ‘3 u. . - .... ‘ I“. ...... . . $051,591.53. 50.: .. Elf.-. 1513.91.15.45 1 z .3... .4338? é}? ...st5....o. ‘z‘glt; N‘vzi.§\ti§«§. A O in...‘;YO§ttfi-i3\tt-L in 55~3Q9hh5555xu12 5 . fi.3..!u...fl\nk4iounnn. t... . . .95.”: Q. ht~51§u1§515ittikghfi 5...... t ‘6 q » .an. 4‘D.x:¥..fil‘»l . . .... $3512 .191. , 15 s! 51.1.1115 . . k o ‘sttxn k. , ...... 2.2. $4.}. 55...! .stfl. .2.1.n. .3 5 .5 5 . z‘AI . ‘ . gaunt-33 '59.!!91 $.13.va “a t l.§fil§1 Ev! fig}! .5 1.1!!! 91mm”: “(*{XEIE . . .. 1.513. 2.1113: I .Hu‘I \l"). V 11“ , . u I .... A n v. A . b1 4 V 5.01 it} «5‘55! § ....iév‘tztiibbthi . ....1 51.905... » V N39. 52x3". “1“." fl . . 1 “LC-"‘1‘“: _ i...“ . zlxfi§gxrfififin \ A . ‘ 7|)... 2¥1\s. Lasuuzifl‘g .3 . 52.1. $34.81!.) 3?. .v , 551.511.15.235‘1501159511» xéhil...‘mmm& aszszizizlvfiuhhall gwggfibséfihhw {Pg 2‘551333Hfi .. til-nth. . . 11.5..) . 3 .. 5.115355%"...55 than“, .. : u 1; a ...... V71. 2:3: 3.15.3.3 ‘uzuvhl 5 ". l‘zgt5 1‘5“! DLVI)\7"\.U.JDM9§)VI~ Q. luii‘tvb Slui‘?‘ .ofl.l)..“hhwhk.§u:vhi.u.§g\.:§.ur “Vii-22%... . \ ca Ii!“ 71...»..955 $13.51.! x. . . \\.\ A I 0.” hmhl‘ibiuivfivfll‘h , I.) L“ . ._ v 1.! x..- 5” tin-v.35. In! 0‘ w‘\ fitviiaunvpténiz I‘DE . . .311132“)! 71... )é‘i‘: \ . ¥I$\lifi\zz§$“1‘itlil!§uxt\5 ‘fi if! ‘EQ'II‘~"“‘: . §§| it k “U‘“ “nu U u “ “.“15-|\l,' l)\‘ll“(~l\‘|‘t‘\-kHKHu \‘fi, I\l\- , \I an. ‘ . a ti) VQIQV 1. gang)“ 51.535510! Vail).- lftix...\\..u.\~.. .7ii...ll.)\.)\1\h\ 3) 132.1%}.5 {01.115333. any. . 1.51:5! 2161'! Souk-19.1.. ‘Nu m? I n '- u ~111.(.x!.|u!l.\. ‘\\J. I]! {1.2931095 ).\3’.\.‘\l..)u§.\\~\ 1355.21? {yizgwlb . . W l s I) L .W t . 5 . $0.355 3““. .u‘ilvli -. ._ . Kraun 1 .u 11“. X1.£ll~a\.1\v\lil\V‘-.K\:LLIJII .ir%.5,1hha x\.\\.\s.\\.uo\h:.\b.\)1)\‘.\1 \ \HNVQI I5 )1.)- u.» asbfirifilii‘i‘é :5... 1r...r41$l.lsu 1.x Mafia“?15.51r545551hms5xw‘01515..§25..\ap&41,§ 51321 {551552 5:11.155. 1531. H x . 5 ‘5 x . DH .hiflqn‘llrit§$il‘§L‘~§vz§l ..H‘ . \) 4 ‘xx . $01!. 5.12.53.23) ”infiT‘ I! J“. .1 i6 ”.59 5...... 15:). \5 éfi'llll‘..h ‘ 5 '0‘ . . . [uh-x135\.‘\\\t).i\.. 1.11.01.» 1&1b51 ‘11.. 5011112.. 5“, $1M \ . V 51 n .l. s 7 I" Y . I. . 1:7 My!“ . . 2| 5!... . 515.3121‘5915 “fistflflu-ln‘. u 5.531..“ {5.. .5351)! Wu:\.{.u.t.!i$\tt:\)\3\x\:)s.. is ‘1‘“; Eight.) 1.1.112.) 0..» 4. 1.. 5'0. 5.. 15.355131 3... 5 . 555115555515... 11.3.11 liksiyawxss‘lsszthhxv. Illanaiyi~§1113 ‘Hshiflz §l§gik§y5i \ . 1555.5 .....h .55....“ “a“! 5.155.156.5555 «1.55.551» 451.15.. 1. 3.5...1355. 1.1.». 50.15}.(\.. “flowivunhfihfv‘? gfi.2ss\1hlis a» \Iuflv ‘ .51 )3” $111]. In} hv5§n§~x ... C‘s... s u 1’1““. 55!. . ZEE.E§QEA§ Ili\)5)ll.)ll. ‘11)}. IA. 11‘ 3 ll” ‘51!\3§ ‘HMUUUI‘LI-I1\s ‘3}. \Eins .l u £511.11 ‘5‘“ l.»-$9.1ah2 75$!) dut‘fkautuii . .t. : an. . I¥.u.hur~t§i\.z5t.¥vl. ......q... iii~3¥lk§§1¥3fial§s§iiflltl§ . 1511:7555 111115.255. ii}?! . .2525 3» 1&1:ng .1.4\<5..L§ » o $1.2! 3 5‘... ..m “"1 1 \:~ I fififlw“.¥b§.x1}>“lx,v 51".“ ”4‘1; ‘Zl‘c‘xrn‘KK-‘h "‘u‘tz‘ V1.\.\.‘.§t \I‘lr‘ir. v. V‘nl“! .7 t5! \l‘E‘hIK" 5. . . 9.th $951K. iiuwaufia: .. c . 21. uixifihv¥§1§1\\séxiuufl.ifisy :2. .lxtizli « .t . h. 2.55. :19-.. “a v stt‘gvl 41.50.50.111 5.1311133... r. 31.71.51. 1.51.44.st Ill Uhxitni)\.\.)57\.g)}a_m.\s)u:x\v\i\x . \. $9... . $855,». _ fig. 55% I 3‘ v} u .v~7!~oo~5\ .5‘; I'lvbvlvf‘vy 1.6.0; , ‘vv v In 5111. 11151245 $116.51: «Vthllfifiigtllifxl 1‘ If: .5 In. 5 in}... aka...»...fi,..&........;..§z.§. 2.. z 1155, 13112.5. 2. 3.. .. .5....1,....u.155¥§ 52.55%... 5251...; 3...... 52.2... . z s , .55, .25 s. .1353. - i. v .1": I Q V}. “LI . ‘ , .I I V . 1‘ .n‘ Li‘- .H1\\.\n)\.|n‘4l1 5.\ {311‘ . ~ u. ....... L o. . tin. 2. .35 ...-.. a 1V a. 51%“ 3-511%?“ axuw.wr(bul<.1..ist.£%\“51{a (dun. 13W!) .1 a y. r ...Li‘lrbé! [1531‘ L.)\\ 11553711110311} 5131\‘iisti13w‘ixix, )3. Y . I. 1.x: 5 h 7 n ‘siE‘ll ‘1 l’Jtl‘lI} 1g 2|i“?.\lz.i‘l\ [I I‘l . 55 531151.111. .. .11.... 51511252 1151511 $153115: 1555595 ..alzsslii. 1.» 1 . 1... 1:51 . ... .‘xltlYilIn. 3 1 \ll «‘I‘U‘wli 51)... 5.15211... .... 1...; 1.10M.\T1-2$11¢y 35V. g‘i7‘li‘: tin!) V3 (\‘5 g1". . 505.555? 5.1. 11151 uvlSuxutiVQ1.5).!Zhuifiaaxsbiiwlwzsl.$shufluuhhbi(\l~tll. n . . , 1.x}. iynflihshfini515§nflills~54 \mfilfiu 58.355, 12., AJ‘ ‘).Il‘z‘l‘vnl'i“‘\“\‘l\nl\‘l 1,) ‘3, git 1‘ 1112‘.) v .91.!)115151. i . . ....u...u..wv.2u...t.3...nvflis : .... e .. . i‘l’l‘: . \ ‘tia' ...-{....5$35..$.¢o\ . .....55 . 55.1.51... «5.5.51! ilxvyxhtfi‘ti l . 1......»‘5155‘1 v1“... r'viiiifv ‘31 . {11539531. 51., 55:55:51.. 53511553351555 5115525? 5 .. v1\£nfi‘§?ilbi:.n . :0) ‘51., iltisinilltlplz. . x55 1. 5 . lab... fighgwlfisfiil....IJMsxiiaxéfl‘nmN . 131353155193?:hefixfilsxslslty-3al?l\xsii.i>.zs\h\ a R l .. . 1?}1.525\1v§. It. Ill), 111301. . . 5) Kkhflanbixxulmivls synics sh. 9.41.5515 x . ...UH’I. hill?!) V1513? i\\\.fi1\\ ‘51“: . Hui-Qt: xslxiltriflizifiuflnhflxlilai, 1555 32% 5%.. . nan“. : I. ~il‘p‘t‘i 1 . 3.15“ I 1“ ..h ~'h hfi‘tl'l YII‘ , '1‘. sightifiln u . .. . v ‘5flfllerlv..uw.lau.vun l. i igorcvilvib y 9%:11.’ u I ll‘\.l. . ~ §Ofi ...!)iviifiifiu. . 9 it 1 :5 ‘5‘ ‘5‘!!!) 5...... filth/“Visit. a§ 2‘ ailiili‘. . . 7 illafilirhvvvflnflvfli s.) m). 511)). E‘hfiilnlil?iv)§> \fl. - u .EEEA-‘IQ‘ . Y 11,113“. it; ._ .15\ .dfl“l.h1.)$fi\¢vl\lov “duty-M 311111... H5 (5)5 .. 5L? 4‘ («“553 55? ~31. $11)“..a‘ n)“ \ . 3-252%.5‘fi1maufifikiinbv‘ «vagina magi}, 1915\111 Vii-2‘1“. 55353331933351? .1\ 1‘11}? ‘1‘89151‘ \ 11. :11. ~51... iii‘lfiag . . .... . ‘Ht‘: - BUN—laaétu “LE.“Es 1|. .1133“ u . :12‘Ju}%§v¥ifin) y “1.1313 .xnv‘.\\$\0w1i%.l!t Li \1 s ‘7‘; §~‘~§§~ 5" \‘Qt 383...!) 5 \s . Illtlruxfisi . . ...35 .. . . $2 -x .23 ~ *zC \ . v... X 85 V s . \Iilscld‘trflai zilgvt! l. 1.5!}! .1. 15103.“! .‘11‘.\ \.4\‘5”H..1 5‘ It“. 1.5011... 8.0.5.1.! \ 1.1 ... .\ \LiAlchl... v4. :3)! 31.1.): v ; .. i . 111‘ 1Q§.iiw%i «(11‘3“ n\sul1s.l‘.\ xii Elléili 11“; Lil-tightnx‘t ffi’gz“ _ 1%” W0“ ””0335...“ gt] . 2155115.... . £111.)... Elli-T}...- 351 I! s \L .th 1.19.6505 “1555153. .1 11311.11 1.5. 1‘. )1 . x \- éf. \\.)~I( 511 § . 0% ‘5‘... 111" . {45‘ [3513‘ I %\ ‘ukfié‘l‘ .01“... 1’35}? {I \9.In(l.5.nb§8 in is“ 1%‘ufi 1‘12! (61 . {‘5 ‘9 '- i1;\;2 LIL 5 ~Ll .55 i ..55‘553}. tux I v t. y 5% Vi. infliii Kg 5 {.1 £525.13"? ‘ 5%..tlt 2...! 5 (58 .la? 5...¢\ Lt {1‘42 5555.11.55?” A .5 s. i L. hikn‘nisthvfils .55) 5.16.7.1! 5§NMME~ . 55...??? .5115 Ltd. 1.1 (I ) I1 {1“ . 313”“.1lfi 5‘15! Eii‘glg ti 1‘ . c. it“. it. . ‘7 l . , .0 . ll"§‘n] 1’ . itifit It]. :1 I‘l‘ 9 15.5.ifibtflfig. . 15:..ng Etfiig‘ss‘ 1.x 1 .ltl‘fiiti “UNI; 55.5».1. i..it.:~5.){.k $0.1 {1 1 514.1(5‘3I.-. 15.5.“... 51ir‘.L.-\.A§\.!II it¥:5st 5.. {Rikki-5‘; v.5!!\'3\..\.‘1!lnv'\ slgtl-VXL $.95. 251555.}?! . , h {l‘ffik‘i‘lofix‘ . i ilt~$|t 5%.}. Pl‘qzkkv: >5 szr...:‘5u .\.?|.\.(A .3943 5““: 595;».th IN (43.5.51... s ovh... I. uh. . :1 Hvfihovvk‘mbni .\ Ll“; ‘ .95.. . u \u ‘ Al. . u. . . :‘ru..:.ht{bb$¥‘ .0910: I I»: ‘ ... . Q9553 :otélv...vlr\5 \fitxb‘iy‘tiit 655.; ill-1‘9“,“ . a x . 1‘ 5\% éflfinfixl 5515115221: . i.‘ o I... . . J. . '0‘. . P t N 5 ssmfinnuosi.in.\.. . . - - . . ., +x3..!.5.§hfl..\.a.¥h;55v . 0» .\ .6 r155 . ~ . .5 » »: v15. ‘ . [1‘11 . .. 5 .. . .152; , n. t 5 0 VA r hr ikfi‘ghwnc‘55riut .Vlv . 35“}‘1 it; .3, 5 whfluuhfivl. - ..t(\\.5.5.vH§..l)£Ls5vvl»Vt.!sbst5.. , .115. .... a). ...» . v W“Whh”hflw."hh§uld55~nuu“iivv§yfloflhfl .%“\5§3E 515‘ \' ‘Oli5v‘vil‘io15‘. i‘ \ III: 1:16.55! ‘1‘» .. if :q ‘iktu-O‘Elh‘i’l‘ti , . EA; 1 13‘3".“ . 5 . , , V5 . u . . v.5 i 7....“Iut‘ivi viiit 22115? ‘1‘ KtLA ‘ “It. firfivu... . $5» $355.. ,Q‘vvvfib5’gufinvv . 1211514. .5 § . 1-..: I'M, .... .. . 5'; (End! . . .- iv . . Eat-5:25: ER (9.0.1:)!!! 5 15-54.» Fran-rmtréo .Iv .... Tu \- (Alix)... . 1411;11:55‘“ . 5 . p!..i\..l A a. .5 . 4 v . . ...1'5 .\ «a El . M. ..I‘vau.5mmmv 5 .. . .. l. ...,» .-.T. . . 5 5.2.7.7.. 1335..‘.51{11552% 3‘x‘iil‘51i. Irv?! . 5 . v I: . x ((5.)? {s I o.~.. .\§:. 151.. 12“}: ($19331 . . 55.5%“ ( . . . .urtul. u 5 .. :kC. a”. ‘\l.1\.1)\§‘§(‘| :i‘t‘i‘“ t . . . .155S5w..£ 4: %E%§;EEQII.Y‘555;K%5 9 I . . . ( all . . . 5 I... 5 1‘3. (1.1831 t+..tl.{t\\.t|.t¥v$v£c§%|\..ll\. V51 , 1.32... . . , .....k. . .. . . , ... .. :ihiikgfiu . . .‘\..V\5\..2 , _. 5 . . . y. 5‘ 1. . x..ru\ I . ‘1 3x10)... M335. ‘ . 55L. .... , . i In: . \sth\\5\n5§' .55... . , A. . . it 5 l. \ v5.15... 52....1: ,.,£.A.A.¢t.( .. . 15!.» E 1259.. . 5.05: . L .. . . u . t . Quinn .... .. .... 32.1.32. ...umuuunflauunfiqu: .. ,\ ... u. T. .... . .... .5.’.u.. U THESIS This is to certify that the dissertation entitled The Foraging Behavior of Largemouth Bass in Structured Environments presented by Owen Anderson has been accepted towards fulfillment of the requirements for Ph . D degree in Zoology Major professor / Date {/23/?3 / / 0-12771 MSU is an Affirmative Action/Equal Opportunity Institution )V‘ESI.J RETURNING MATERIALS: Place in book drop to LlBRARlES remove this checkout from .—5—_ your record. FINES will be charged if book is returned after the date stamped below. NOT CIR ilLATE R OMUSEO Y 0 Hill (I (ULATE 5.9m. _. _ THE FORAGING BEHAVIOR OF LARGEMOUTH BASS IN STRUCTURED ENVIRONMENTS By /§’2"~ ix" 7 Owen Anderson A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Zoology 1983 ABSTRACT THE FORAGING BEHAVIOR OF LARGEMOUTH BASS IN STRUCTURED ENVIRONMENTS By Owen Anderson The specific effects of environmental structural complexity on foraging behavior have been poorly understood. A laboratory study using largemouth bass (Micropterous salmoides) and three prey types demonstrated that variation in structural complexity significantly influenced prey encounter rates, handling times, and values, and the ability of predators to learn about prey. Prey encounter rates gen— erally declined with increased structure, but the effect was non—linear, with greater changes occurring between high and intermediate structures than between intermediate and low. In one case, using a schooling prey, encounter rates increased from low to medium structure. Two morphologically similar insect prey types had drastically different encounter rates, and structure modified their encounter rates differ— ently. Structural complexity and prey handling times were positively related. Mechanisms producing handling time changes, including the relative frequencies of multiple captures and shortened pursuits, were examined. Decrements in overall energy intake associated with longer handling times were larger in less structured environments. Prey value, traditionally viewed as dependent on physical Owen Anderson attributes of predator and prey, declined with increased structure and was also found to be influenced by prey density and distribution. Structure had minimal effects on prey capture success and the ener- getic costs of handling and searching for prey. I suggest that var- iation in structural complexity will have substantial effects on pred— ators' breadths of diets and that because of the complexity of the relationship between structure and diet breadth, generalizations relat— ing the two variables will be difficult to make. Evidence from the laboratory and field corroborates these assumptions. An optimal foraging model was utilized to predict prey selec- tion by largemouth bass in two laboratory environments with identical prey communities but different densities of macrophytes. Patterns of prey selection by the bass corresponded closely with the predictions of the model, with bass in low structure being more specialized. Finally, growth rates and diet breadths of largemouth bass were examined in stream environments with different levels of macro~ phytes. In the field, largemouth bass had the greatest final weights and narrowest diets at intermediate levels of structure. Bass in both low structure, where the resource community was impoverished, and high structure, where prey were more effectively shielded by vegeta— tion, had reduced growth rates and broader diets than their medium vegetation counterparts a ACKNOWLEDGMENTS I would like to thank my major professor, William Cooper, for his advice, enthusiasm, and support during the course of this research. Don Hall made many valuable contributions both during the planning and implementation of this research. Jack King was a con— stant source of encouragement and help. Rich Merritt, Earl Werner, Bob Robbins, Craig Osenberg, Jim Wetterer, and Ed Turancik made many valuable comments which contributed to this work. Jack Arthur and his staff at the EPA outdoor laboratory in Monticello, Minnesota, allowed me to use their pole barn for a laboratory. Tom Henry instructed me in the fine art of pumping Mississippi River water into the lab. Scott Cooper, Mike Pease, and Brent Danielson helped develop the inner workings of the laboratory in Minnesota, and Julie Bebak, Scott Cooper, Mike Pease, and Gary and Mike Unger helped with various parts of the field work. Rangnar Anderson assisted in the building of a laboratory in Michigan. Sue Anderson, Cori Anderson, Elena Ugarte, Amy Outman, Neil Anderson, Craig Anderson, and Oren Anderson helped collect damselflies in the field. Squeaky and Cindy, my two "catfish", helped greatly by staying out of the laboratory aquaria. This work was supported by a National Science Foundation predoctoral fellowship to the author and by EPA grant CR807555 to Drs. William Cooper and Robert Boling. LIST OF TABLES LIST OF FIGURES INTRODUCTION PART 1. OPTIMAL TABLE OF CONTENTS 0 u o o o o o u o u u c o FORAGING BY LARGEMOUTH BASS IN STRUCTURED ENVIRONMENTS . . . . . . Introduction . . . . Methods . . . . . . Results of the Preliminary Experiments Results of the Optimal Foraging Experiments . . . . . Discussion . . . . . . . . Literature Cited . . . . . 2. THE EFFECTS OF STRUCTURAL COMPLEXITY ON FORAGING BEHAVIOR AND DIET BREADTH Introduction . . . . . . . . Methods . . . . . . . . . . Results . . . . . . . . . . Encounter Rates . . . . . Handling Times . . . Rates of Movement and Energetic Costs . . . . . . . . Diet Breadths . . . . . . . Discussion . . . . . . . . Literature Cited . . . . . 3. PREY SELECTION AND GROWTH OF LARGEMOUTH BASS AT DIFFERENT MACROPHYTE DENSITIES . . APPENDIX Methods . . . . . . . . . . Results . . . . . . . . . Discussion . . . . . . . . . Literature Cited . . . . . . . Page iv viii 21 31 4O 46 47 52 58 58 7O 78 81 92 107 112 121 124 139 143 146 PART 1. Table LIST OF TABLES Search and handling time relations for two prey of largemouth bass at two structural levels. Mean times per prey item (+ 1 SE) are given in seconds. Values of F and the probability of Type I error are given for each foraging parameter (ANOVA). N = the number of prey captures . . . . . . Relationship between capture order within foraging trials for guppies and the mean search times (in seconds) required to detect guppies in low and high structures. Welch's t' statistic is given based on comparisons of means for each capture number, with prob- ability of Type I error . . . . . . . Mean velocities (+ 1 SE) of 76 mm largemouth bass while searching for or handling either damselflies or guppies in two differently structured environments. Velocities are expressed in cm/s. Values of F and the probability of Type I error are given for each comparison between structures (ANOVA) . Calculation of handling cost and E /T when pur— suit time is a large or small fraction of total handling time or when an average overall handling velocity is used. P = pursuit time; H 2 handling time not including pursuit. Y 2 oxygen or energy consumption . . Patterns of prey selection by largemouth bass in low and high structure for the first seven days of the optimal foraging experiments. L: low structure, H = high structure, F : fish prey, and D : damselfly prey. Prey are repre— sented in the order in which they were cap— tured during a foraging bout . . . . . iv Page 20 22 Table PART 2. Table 1. 2. 3. 4. Prey selection by largemouth bass on days 8—14 of the optimal foraging experiments. Fishes 1—3 had been switched to high struc— ture, fishes 4—6 to low structure . . . Average rates of energy acquisition (J/s) after consumption of two prey by fishes 1—3 in high structure. An * indicates that the first two prey eaten were gup— pies, + indicates that a guppy and a damselfly were the first two prey, and 0 indicates the first two prey were dam— selflies . . . . . . . . . . . Fitted parameters (+ 1 SE) of the linear regression equation describing encounter rates (B) by largemouth bass with mayflies in three structurally different environ— ments. The form of each equation is B = a + b x, with x = mayfly density (number per aquarium) . . . . . . . . . Fitted parameters (+ 1 SE) of the linear regression equations describing encounter rates (B) by largemouth bass with fathead minnows in three structurally different environments. The form of each regression equation is B = a + b x, with x = fathead minnow density (number per aquarium) . . Predicted mean encounter rates (captures per second of search with 95% C. I.) by large— mouth bass with Callibaetis and fathead m'n— now prey at different prey densities (#/m ) in environments with different amounts of structural complexity. L — low structure, M — medium structure, and H — high structure. Encounter rates and intervals were calculated using the linear regression relationships described in Tables 1 and 2. . . . . . Comparisons of search times (+ 1 SE, in seconds) required by largemouth bass—to detect Calli— baetis mayflies and Coenagrionid damselflies in differently structured environments . V Page 23 32 Page 61 67 69 Table 5. 6. 7. Percentage of capture attempts which are suc- cessful for largemouth bass with three prey types at three structural levels. L — low, M — medium, and H — high levels of structure. N is greater than 90 for each combination of structure and prey type . . . . . . . Mean handling times (+ 1 SE) in seconds for largemouth bass with three different prey types in structurally complex environments — low (L), medium (M), or high (H) . . . . . . . . Tests of significance for structure—related differences in handling times by bass with three different prey types. For each prey type, two mutually orthogonal contrasts are shown. For each contrast, delta k (the dif— ference between the averages of the observa— tions on either side of the contrast), the t statistic, and probability of Type I error are given. H : high structure, M 2 medium structure, and L = low structure . . . . Categorization of handling times (in seconds) for largemouth bass with Callibaetis. Two contrasts are presented as well as the 95% confidence interval for the differences be— tween the means on opposite sides of the con— trast (Scheffe's Interval) . . . . . . . Average velocities (cm/s + 1 SE) of largemouth bass while searching fof or handling two types of prey in three differently structured envi— ronments. L — low structure, M — medium struc- ture, and H — high structure . . . . . Calculated values (cal/s) of three different prey of largemouth bass at three levels of structural complexity. L, M, and H stand for low, medium, and high levels of structure . . Hypothetical rates of energy acquisition (E /T) in cal/s for largemouth bass foraging forn mayflies in three structured environments, low medium, or high, with three different handling times. E /T is calculated for each combination of handlng time and structure using the optimal foraging model described in the text and the encounter rates given in Table 3 (with density =300/m3>............. vi Page 71 72 73 77 82 100 PART 3. Table Page 1. Expected patterns of diet breadth in relation to macrophyte density based on four possible underlying mechanisms. Mechanisms are num— bered as in the text. L : low, M 2 medium, and H = high macrophyte density in this and all other tables . . . . . . . . . . . 118 2. Prey community characteristics in stream areas with different amounts of vegetation. Mean values + 1 SE are given . . . . . . . . 125 3. Mean final total lengths (: 1 SE, in mm) of largemouth bass at different levels of veg— etation . . . . . . . . . . . . . . 127 4. Prey consumption patterns of largemouth bass in stream habitats with different amounts of vegetation. Mean values are given + 1 SE . . . . . . . . . . . . ". . . . 129 5. Relative utilization of four different prey types by largemouth bass at different levels of vegetation . . . . . . . . . . . . 131 6. Representation of Baetid mayflies in the prey communities and the diets of largemouth bass in three habitat types . . . . . . . 136 7. Representation of Coenagrionid damselflies in the prey communities and diets of largemouth bass in three habitat types . . . . . . . 138 8. Representation of amphipods in the prey com— munities and diets of largemouth bass in three habitat types . . . . . . . . . . 140 PART 1. Figure 1. 2. PART 2. Figure 1. LIST OF FIGURES Page Consumption of fish prey by largemouth bass in low (LV) and high (HV) structure during 14—day optimal foraging experi- ments . . . . . . . . . . . . . . 24 Number of damselfly captures per ten seconds of search (+ 1 SE) for bass in high struc— ture at various stages of the foraging experiments. A = damselfly capture rate during the first forty seconds of search when bass foraged for damselflies alone. B = damselfly capture rate during first forty seconds of search on days 5—7 of the optimal foraging experiments (both prey present). C = damselfly capture rate on days 5—7 once two guppies had been eaten . . 28 Page Proposed mechanism by which diet breadth increases as prey encounter rates decline. The solid line plots net energy gained (En) vs. time spent foraging (T) for a predator when prey availability is high. The dashed line is for the same relationship but with low prey availability. Symbols (stars, closed circles, and open circles) show net energy gained (E_) from eating an individual of a particular prey type vs. the handling time (Hi) for that prey type. Stars represent prey types eaten under both conditions of prey availability; prey represented by closed circles are not consumed when the availability of preferred prey is high (the solid line) be— cause such consumption would lower E /T. Prey represented by open circles are exclHded from the diet in both cases . . . . . . . . 50 viii Figure Page 2. Mean encounter rates (# captures/second of search, + 1 SE) for largemouth bass in rela— tion to Ehe density of Callibaetis mayflies in laboratory environments with low (L), medium (M), and high (H) amounts of struc— ture. Multip y mayfly densities by 8.33 to obtain #/m . Lines were drawn from the regression equations in Table 1 . . . . . . 6O 3. Mean encounter rates (# captures/second of search, + 1 SE) for largemouth bass in rela— tion to fhe density of fathead minnows (Pime— phales promelas) in laboratory environments with low (L), medium (M), and high (H) amounts of structural complexity. Multiply prey den— sities by 8.33 to obtain #/m . Lines were drawn using the regression equations in Table 2 . . 64 4. Handling times (R + 1 SE) in seconds for large— mouth bass feeding on Callibaetis mayflies at different mayfly densities in low structure. Multiply mayfly densities by 8.33 to obtain #/m3. The form of the regression equation is Y = 5.8 — .O8(x) . . . . . . . . . . . 76 5. Graphical representation of the predicted for— aging modes of 70 mm largemouth bass at differ— ent densities of two prey types, fathead min— nows (Pimephales promelas) and Callibaetis may— flies, in environments with low (L), medium (M), or high (H) levels of structure. Specialization on the preferred prey, Pimephales, is predicted in the hatched areas. Open areas of the graphs represent combined prey densities at which bass should be generalists, taking both prey as en— countered. Fathead dry weigh : 9.9 mg, mayfly dry weight = 2.0 mg . . . . . . . . . . 86 6. Graphical representation of predicted foraging modes utilized by 70 mm largemouth bass at dif— ferent densities of fathead minnows (Pimephales) and Coenagrionid damselflies in three differently structured environments. Specialization on Pime— phales is predicted in the hatched areas. Both prey should be eaten if their combined densities lie in open areas of the graphs. Fathead dry mass = 9.9 mg, damselfly dry mass : 2.3 mg. . . . . 89 ix Figure 7. 8. 9. Graphical representation of predicted diets of 70 mm largemouth bass at different densi— ties of Callibaetis mayflies and Coenagrionid damselflies in low, medium and highly struc— tured environments. Specialization on Calli— baetis is predicted in hatched areas; gen— eralization (taking both prey) is energetic— ally advantageous in open portions of the graphs. Mayfly dry weight = 2.0 mg, damsel— fly dry weight : 2.3 mg . . . . . . . Relationship between structural complexity (macrophyte stems/m ) and prey weighting fac— tors for two prey types. Prey weighting fac— tor = the number which must be multiplied by the prey density at any level of structure so that the encounter rate by largemouth bass with that prey type at that level of structure is equal to the encounter ra e at the lowest level of structure (200 stems/m ). Dashed line repre— sents plot of weighting factor vs. structure for Coenagrionid prey, dash—dot line is for Callibaetis. Solid line represents plot of factors by which prey densities are actually increased across changes in structure (from Gerking 1957) . . . . . . . . . . Relationship between prey density and density of of aquatic macrophytes in Bryant's Creek Lake, Indiana (from Gerking 1957). The straight line is drawn from the linear regression equation: Y = 281.9 + 19.96(x). r2 = .44 . . . 10. Schematic representation of the alteration in value of one prey type due to the change in distribution of another prey species. Indi— viduals of the two prey species are represented by dots labelled M and N. Curved line X des— ignates the limit of the fish's reactive field when it is in the position shown. Y designates the reactive field limit once the fish has reached prey type M. In 103, species N is at low density or is not clumped; in 10b the den— sity or clumping of N has increased . . . Page 91 95 98 103 Page Representation of mechanism by which diet breadth increases at high or low macrophyte densities. The solid line plots net energy gained(E ) vs. time spent foraging (T) for a fish predator in an environment with intermediate amounts of vegetation. The dashed line is for the same relationship at high or low macrophyte densities. Symbols show net energy gained from eating an individual of a particular prey type (E.) vs. the handling time for that prey type. Staré repre- sent prey types selectively preyed upon at all macrophyte densities; prey represented by closed Circles are not consumed in intermediate vege— tation because consumption of such prey would lower E /T. Open circles represent prey types which are not eaten at any macrophyte level . . 116 Plots of the number of Baetids or Coenagrionids eaten per fish at three levels of vegetation and actual Baetid or Coenagrionid densities vs. level of vegetation. L = low, M 2 medium, and H = high vegetation. Dashed lines and open dots characterize Coenagrionid consumption or density. Closed dots, solid lines are for Baetids . . . . . . . . . . . . . . 135 xi INTRODUCTION Interactions between predators and their prey are usually not carried out in homogeneous arenas but take place in environments with varying amounts of physical structure. Despite the pervasiveness of physical structure, large gaps have existed in our understanding of the effects of environmental structural complexity on foraging beha— vior. Although ecologists have been aware that variation in structure can alter prey encounter rates and the energetic costs of searching for and handling prey, little has been known concerning the effects of structure on prey handling times, prey values, the ability of pred— ators to learn about prey, and predator diet breadths. Such effects, if present, should influence the intensity and outcome of predator— prey interactions as well as the extent of diet overlap between poten— tially competing predator species. The research described in this dissertation represents a quantitative examination of the foraging behavior of young largemouth bass (Micropterous salmoides) in relation to changes in structural complexity. Young bass are active foragers in the well—structured littoral regions of North American lakes and encounter wide variation in structural complexity both within lakes and throughout their nat- ural range. The research is divided into three separate papers. The first manuscript describes a situation in which bass are free to select prey in environments with identical prey communities but different levels of structure. The predictions of an optimal foraging model are com- pared with actual prey consumption of the bass. The second paper de- scribes in quantitative terms the changes in bass foraging behavior produced by variation in structure. Both of these studies were car— ried out in the laboratory. Finally, the third portion of this the— sis analyzes the growth rates and diet breadths of bass in stream environments with different levels of macrophytes. The foraging beha— vior of bass in the field is compared with predictions based on the laboratory studies. OPTIMAL FORAGING BY LARGEMOUTH BASS IN STRUCTURED ENVIRONMENTS Owen Anderson Zoology Department Michigan State University East Lansing, MI 48824 INTRODUCTION Vegetation, rocks, and debris are pervasive features of lit— toral zone environments, yet the role of such structures in determining the foraging behavior of fish has been poorly understood. Although increased amounts of structure have been correlated with lowered rates of prey capture by fish predators (Glass 1971, Ware 1973, Vince et al. 1976, Stein 1977, Stoner 1982), the effects of structure on prey handl- ing times and on the specific energetic costs of both searching for and handling different prey types have been unknown. Such effects, if pre— sent, may be quite important since the actual composition of fish diets has been demonstrated to depend on the relative values of the available prey (Mittelbach 1981, Werner and Mittelbach 1981). Prey value, in turn, incorporates both the time required to handle a prey item and the net energy gained from that prey individual (Charnov 1976). Increased structure has been shown to have opposite effects on the diet breadths of fish predators. Vince et al. (1976) documented increases in diet breadth at higher levels of structure while Stoner's (1982) work implied narrower diets in highly structured portions of the environment. Greater diet breadth could occur at higher structural levels if encounter rates with prey types declined uniformly and preda— tors became less selective. This negative relationship between prey availability and diet breadth has been documented repeatedly (Ivlev 1961, Werner and Hall 1974, Curio 1976). Conversely, if increased structure influenced the availability of prey types differentially, cer— tain prey might gain almost complete refuge from predation in high structure (Stoner 1982). Other, evasive prey might have high pursuit costs or low capture probabilities in high structure (Glass 1971). These effects could produce an inverse relationship between structure and diet breadth. Knowledge of the relative importance of such factors is rudi— mentary. Indeed, there has been only one attempt to predict quantitatively the composition of predators' diets in relation to changes in structural complexity (Mittelbach 1981). However, Mittelbach's experimental manipu— lations altered both the structural complexity of the foraging environ— ment and the nature of the prey community. In this study I quantify parameters of an optimal foraging model proposed by Charnov (1976) as functions of structural complexity in an attempt to predict diet choice by predators in environments with identical prey communities but differ— ent amounts of structure. Optimal foraging theory was employed because of its recent successes in predicting resource utilization by animals in the field (Belovsky 1978, Mittelbach 1981) and laboratory (Werner and Hall 1974, Krebs, Ryan and Charnov 1974, Cowie 1977, Cook and Cockrell 1978. Small (76 mm TL) largemouth bass (Micropterous salmoides) were used as predators because they are active foragers in the vegetatively structured littoral regions of North American lakes (Heidinger 1975). Furthermore, largemouth bass are distributed widely (MacCrimmon and Rob- bins 1975) and generally represent a significant portion of the total fish biomass in lakes in which they are found (Heidinger 1975). METHODS The tests were carried out using six largemouth bass which were 74—78 mm in total length. The fish were obtained from the Frank— fort National Fish Hatchery, Frankfort, Kentucky when they were 30 mm in length and were fed small aquatic invertebrates in the laboratory until they reached the appropriate size. The bass were housed separately in two 110—L aquaria, each of which had been divided into three compart— ments with the insertion of fiberglass screen dividers mounted on wooden frames. The fish were divided into two groups. Initially, three fish foraged only in an environment with a small amount of structure (vege— tation); the other three fish foraged in a highly structured environ— ment. The fish were tested one at a time, and a complete record of the foraging behavior of each bass was carefully maintained. The low structure environment consisted of a 208—L aquarium with 5 plastic Elodea plants (12 stems/plant) spaced uniformly and anchored in a sand substrate. The 208-L high structure tank con— tained 17 plants. Stem densities were ZOO/m2 in low structure and 670/ m2 in high. These stem densities are similar to values found commonly in the field (Gerking 1957). Each 208—L aquarium was divided into two unequal sections with the insertion of a sliding Plexiglas door. The larger 184-L portion contained the vegetation and any experimental prey which had been introduced. The small 24—L section served as an acclimation cham— ber for each bass prior to a foraging bout. A specific foraging trial proceeded in this manner: follow— ing transfer from the holding tank, an individual bass acclimated for thirty minutes in the 24-L chamber. The Plexiglas door was then raised, allowing the bass to commence searching for prey in the larger volume. From behind a blind, an observer recorded the foraging behavior on a voice tape recorder, categorizing the time spent by the bass as either search or handling time and noting the success or failure of each capture attempt. At the end of each day's six foraging trials, the tape was replayed and a Heuer 410 microsplit digital stopwatch was employed to determine actual search and handling times for each prey item. Handling times which did not end in prey capture were added to the next success- ful handling time for purposes of determining average handling time per captured prey item. A bass searched for prey by swimming at a steady rate through the water, frequently changing its direction and adjusting the position of the eyes. Handling behavior, which included the pursuit, capture, and swallowing of a prey item, plus a pause afterward, was easily dis- tinguishable from searching. The beginning of handling time (pursuit) involved an accelerated, unidirectional movement toward a prey item which could usually be seen by the observer. At the end of the handl— ing period for a particular prey individual, a bass was usually station— ary, with the gill covers flaring in and out in a pumplike manner. Search was reinitiated when the bass stopped the pronounced gill cover movement and resumed swimming with a strong thrust of the caudal fin. While many authors have separated pursuit and handling time, the two are combined in this study because together they represent a time invest— ment which must be made by a predator to gain energy from a prey type once it is detected. For purposes of determining an optimal diet, com— bining pursuit with the rest of handling time is essential (see Equa- tion 3, which follows). The aquaria were marked horizontally and vertically with 5 cm markings and the third spatial co-ordinate was estimated by the observer based on the known width of an aquarium so that the spatial positions of the bass could be recorded throughout the timed trials. Since the distance traveled by a bass for each search or handle could then be cal— culated as it moved through a 3—dimensional grid, it was possible to compute average velocities of the bass for each activity. These velo- cities were then used to determine the costs (in J/s) of searching for and handling prey, using the relationship described by Glass (1971), (1) Y = RM + bV where Y is the oxygen consumption in mg/hour, RM is the rate of oxy— gen consumption at zero velocity (routine metabolic level or Ymin)’ V is the velocity of the bass in cm/s, and b and c are constants. For 76 mm bass at an ambient temperature of 20°C., the appropriate equa- tion is (Glass 1971), (2) Y = 2.08 + (.026)(V)1'7 In this study water temperatures actually varied from 19—210C., but this produces only slight changes in oxygen use. Oxygen consumption was then converted to joules expended/second using the relationship established by Elliot and Davison (1975), where 1 mg oxygen consumed equals 13.6 J. Non-swimming handling costs, including the energy required to mouth and swallow prey, are unknown for bass, and no attempt was made to estimate them. Two prey types were utilized in the experiments - uniform sizes of Coenagrionid damselflies and female guppies (Lebistes reticulatus). In order to determine encounter rates, handling times, and costs of searching for and handling each prey type, guppies or damselflies were initially presented alone in the foraging environments at densities of 30 damselflies (210/m3) or 4 guppies (28/m3). Bass in each structural type were allowed to eat either all 4 guppies or eight damselflies dur— ing a foraging bout (after 4 guppies or 8 damselflies, 76 mm bass begin to become satiated, and handling times increase). For the optimal for- aging experiments, both prey were present, and bass in high and low structure were allotted five minutes total foraging time. To predict the foraging behavior of the bass when both prey were available, an optimal foraging model similar to that described by Charnov (1976) and utilized by Mittelbach (1981) was employed. In the model, the net rate of energy intake of a predator can be formu— lated as, 10 (3) E /T = n where EH 2 the net energy gained while foraging (J), T = the time spent foraging (3), Bi 2 the encounter rate with prey type i (# cap- tures/second of search), Ei = the net energetic gain associated with prey type i (J), CS 2 the cost of searching (J/s), and Hi 2 the time required to handle prey type i (5). Further, Ei’ the net energetic gain associated with prey type i, can be defined as, (4) E. = Ae. — CH. where ei = the actual energetic content of prey type i (J), A = the fraction of the prey's energy content which can be assimilated by the bass, and Ch 2 the cost of handling prey type i (J/s). The assimi— lable fraction of energy ingested (A) was assumed to be .7 (Elliot 1976). The ei's were estimated by drying the guppies and damselflies in a Fisher Isotemp 501 drying oven at 45°C. for 48 hours and then weighing the prey to the nearest .1 mg. Dry masses were then con- verted to joules by assuming that 1 mg dry mass = 21.3 J for damsel- flies (Cummins and Wuycheck 1971), and 1 mg : 20.9 J for the guppy prey (Adams 1975). Mean dry masses of damselflies and guppies were 11 7.6 I .3 and 26.1 i .7 mg, respectively (n=3O for each prey type). Equations 2 and 4 were used to compute the net energetic gain (Ei) associated with each prey type: 113.3 J for damselflies, 381.8 J for guppies (structure had no significant effect on E1 — see Results). Encounter rates, handling times, and costs of searching and handling were determined only after the bass had become acclimated to the laboratory environments and had maximized their encounter rates and minimized their handling times for each prey type. This initial pro- cess took from 3 to 7 days while the bass foraged for damselflies and three days for the guppy prey. There followed six consecutive days during which the bass foraged for damselflies, then six days with gup— pies only as prey. Prior to the optimal foraging tests, the two prey were presented on alternating days for six days (each prey type pre— sented three times). Thus 108 total foraging trials were available for analysis and predictive purposes prior to the optimal foraging experiments. The format for the Optimal foraging work was as follows: both damselflies and guppies were available to the bass at the pre— scribed densities for seven consecutive foraging bouts. During this period the three bass which had been trained only in low structure continued to forage there, and the high structure bass foraged in high structure. Seven days (foraging bouts) has been shown to be an adequate time for fish to maximize their foraging efficiency while learning to forage in a novel situation (Werner et al. 1981). After seven days the two groups of fish were switched, i.e., on the eighth day the original low structure bass foraged in high structure, and 12 vice—versa. This marked the first time that the fish in either group had been exposed to the alternate environment. The bass were then allowed to forage in their new environments for seven consecutive days. This changeover design was utilized so that more fish could be tested in each environment and so that potential residual effects resulting from foraging in the alternately structured environment could be deter- mined. Bass in high and low structure consumed approximately 1630 net joules per day during the optimal foraging experiments. Since the fish were given no other food, bass in both treatments began their foraging trials in approximately the same motivational state. While foraging, bass in low and high structure acquired energy at much dif— ferent rates, however. Fish in low structure initially had higher rates of prey capture and then consumed few prey during the latter stages of foraging bouts; fish in high structure captured prey at a slow but fairly steady rate throughout the five minute trials. The result was a similar overall net energy intake per day (foraging trial). 76 mm bass require about 840 J/day for routine metabolism; thus the bass grew slightly during the experiments - about 1 mm per week. RESULTS OF THE PRELIMINARY EXPERIMENTS In all cases, increased structural complexity significantly lengthened the time required by the bass to search for and handle prey 13 items (Tables 1 and 2). ANOVA could not be utilized to compare search times for guppies between high and low structure due to heterogeneous variance. Furthermore, in high structure there was a strong dependence of search time on capture order, with the third and fourth guppies taken in a foraging bout requiring much longer search times than the first two (Table 2). Therefore, Welch's solution (Gill 1978) was used to test for search time differences between structures for a given capture number (where 1 is the first guppy taken during a bout and 4 is the fourth and last). For each capture, there is a significant difference in the search time between structures (Table 2). Structure had an effect on the swimming speed of the bass while searching for damselflies and while handling damselflies and guppies (Table 3). While searching for damselflies, bass swam 1.6 times faster in low structure than in high (6.1 cm/s vs. 3.8 cm/s). The mean search time required to detect a damselfly was 23.1 s in low structure and 37.5 s in high (Table 1). These search times translate into encounter rates (# captures per second of search) of .O43/s and .027/s, respec— tively. Thus, the encounter rate in low structure is .O43/.027 = 1.6 times greater than the rate in high and can apparently be directly related to the differences in search velocities between structures. Handling velocity, the average rate of movement of the bass while handling prey, was elevated in low structure for both prey types (Table 3). Handling velocity consisted of the average rate of movement during the entire handling period and thus incorporated the potentially different velocities of the bass during the various subcomponents of 14 TABLE 1. Search and handling time relations for two prey of large- mouth bass at two structural levels. (i 1 SE) are given in seconds. Mean times per prey item Values of F and the probability of type I error are given for each foraging parameter (ANOVA). N = the number of prey captures. Low Structure High Structure N F Damselfly Search Time 23.1 i 3.1 37.5 i 4.4 144 7,3** Damselfly Handling Time 6.6 i .5 8.9 i .6 144 14.3*** Guppy Handling Time 5.9 i .6 7.9 i .7 120 4.8* 7': P<.O5 71‘7“ P< .01 kid: P < o 001 15 TABLE 2. Relationship between capture order within foraging trials for guppies and the mean search times (in seconds) required to detect guppies in low and high structures. Welch's t' statistic is given based on comparisons of means for each capture number, with probability of Type I error. Search Time Search Time Capture Number t' Low Structure High Structure 1 3.8 i 1.0 15.8 i 2.7 4,2** 2 2.4 I .6 17.0 I 4.4 3.1* 3 3.2 i .9 113.1 : 14.7 5.2** 4 12.0 + 2.8 134.6 + 13.8 6.0** * P‘=.005 ** P < .0005 16 TABLE 3. Mean velocities (: 1 SE) of 76 mm largemouth bass while searching for or handling either damselflies or guppies in two differently structured environments. Velocities are expressed in cm/s. Values of F and the probability of Type I error are given for each comparison between structures (ANOVA). Low Structure High Structure N F Damselfly $$* Search Velocity 6‘1 i ’2 3'8 i -2 144 16.2 Damselfly ¢¢¢¢ Handling Velocity 5'2 i '4 3-0 i .2 144 22.4 GUPpy * Search Velocity 6‘8 i '6 6'4 : °6 60 .6 GUPPY 10-5 i .9 6.4 t .6 60 8.4** Handling Velocity * NS P < .05 9:9:7': P < . 02 5 Satin? P< .001 17 handling (pursuing, capturing, swallowing, pausing). An increase in pursuit velocities probably accounted for the higher handling veloci— ties in low structure. In low structure bass can detect prey items at greater distances, and a positive relationship between pursuit velo— city and distance from prey at which pursuit is begun has been docu— mented by Nyberg (1971). Actual pursuit velocities were not calculated, however, and it is also possible that average handling velocities were greater in low structure because pursuit represented a larger propor— tion of total handling time there. Pursuits in high structure were usually initiated at a closer distance to the prey than in low struc— ture. For a complete discussion of the effects of structure on handling time, see Anderson (1983). The differences in costs incurred by bass in the two environ— ments were small. For example, the bass in low structure swam 1.6 times as fast as their high structure counterparts (6.1 vs. 3.8 cm/s, Table 3) while searching for damselflies yet experienced only a small incre- ment in cost (from Equation 2, .010 J/s vs. .009 J/s). Similarly, despite the fact that bass in low structure increased their average rate of movement while handling guppies from 6.4 to 10.5 cm/s compared to high structure bass, this represented an increase of only .003 J/s, from .010 to .013 J/s. Since the difference in handling time between low and high structure was two seconds (Table 1), this represented a miniscule alteration (.006 J) in the net energetic content (E1) of the prey. Structure had little influence on capture success. Greater than 98% of all damselfly and 90% of all guppy capture attempts were successful at each structural level. 18 Using this preliminary information, it was possible to use Equation 3 to calculate the net energetic returns in each habitat for each foraging mode. In low structure the optimal foraging behavior involved eating guppies until there were none left. This strategy had a predicted En/T of 35.6 J/s compared with 31.2 J/s for eating guppies and damselflies as encountered and 3.8 J/s for eating only damselflies. Thus, a bass specializing on guppies could have increased its rate of energy intake by 13% compared to a generalized foraging mode (taking guppies and damselflies as encountered) and 844% compared to special— ization on damselflies. The prediction was not so simple in high struc— ture. The strong inverse relationship in high structure between num- ber of guppies already captured and search time required to detect another guppy (Table 2) made it necessary to reapply the optimal for— aging model after each guppy capture in order to predict the best pat- tern of prey selection. The question then could be stated: Was the encounter rate with the more valuable prey type, guppies, ever low enough in high structure so that bass could increase their energy intake rate by including damselflies in the diet? The value (Bi/Hi) of a dam- selfly in high structure was 113.3 J/8.9 s = 12.7 J/s. Using the Optimal foraging model (Equation 3) and the appropriate parameters for guppy prey in high structure (E1 = 381.8 J, Hi 2 7.9 s), it can be shown that if the encounter rate with guppies fell to below .O45/s, the net rate of energy acquisition (En/T) while foraging for guppies would be less than 12.7 J/s and it would be energetically favorable to include damselflies in the diet. Thus the critical guppy search time 19 in high structure was 1/.O45 = 22.2 s. If a bass encountered guppies less often than every 22.2 seconds, generalization was predicted. From Table 2 it can be seen that the critical search time was exceeded, on average, for the third and fourth guppies taken during a foraging bout. Thus, the optimal strategy in high structure was to initially specialize, taking two guppies, and then to generalize thereafter, taking any prey item encountered. It should be noted that the calculated value of En/T is quite insensitive to possible variation in energetic costs incurred by the bass while foraging. The lack of sensitivity of En/T to changes in cost is important, because, as mentioned, an average handling velocity was utilized to estimate energetic costs of bass while pursuing and cap- turing prey. Such averaging may underestimate the cost of handling prey, since cost is not linearly related to velocity (Equation 2) and pursuit (at higher than average handling velocity) represented an unknown frac- tion of handling time. Table 4 summarizes the changes in cost per han— dled prey item and the resultant changes in En/T when pursuit time is allowed to be a varying portion of total handling time. Data is pre— sented for a bass foraging in high structure for the first two guppy prey, with a search time of 16.4 s and handling time = 7.9 s. In case A, the average handling velocity is used to determine handling cost and En/T° Case B assumes that pursuit time (PT) occupied all but one second of total handling time. Case C assumes that pursuit time required only 2 seconds. Note that changes in total cost per handled prey item and En/T are small. Similar results are obtained if low structure bass are ana- lyzed. Also, differences in the costs of searching for damselflies vs. 20 Hmcfiwwuoz ‘—I smeo. Name. am.w m.mm o.N sage Ha mNOA.mH ooHH. omao.AnHH e no oeso. Shoo. wo.N o.o o.m size I wmko. ono. qw.N m.a 9.0 saco em Amok.ma hawov\. moHoc\\\\ "m ///oaoo. ////oaoo. wo.N o.o o.a aaco A: O C O O O O O H‘ H; O mmON ma oowo NoHo NoHo 00 N a o o n k m + I .< A.\4v Axua.aac\ev A.\av : umoo mwmnm>< Am\hv Auc\wEv Am\EoV Amv e\ m Hmuoe emuewaaz s » auaooam> «sue ammo .coHumEDmcoo xwuocm no mowhxo H » .uHSmusm wCMvSHocH no: wEHu wcflfipcm: H H: mmEflu ufismusa n 9m .pmm: ma >ufiuofio> wcflapcmg Hfimuo>o owmuo>w cm :653 no memo mafifipcm: HmuOu mo cowuomuw HHmEm no owumfi m me meu uwamusa Goa: H\:m can umoo wcfifipcms mo cofiumfisofimo .¢ mqmmo m >mo a zmm m zen N mum H mmo .uson wcwaHOM m wcflnsp coumm new pmusuamo mums %m£u scans SH nopuo ozu CH vmqumouamu mum zoum .hmua kaHomEmp n a cam .xoum :mflm n m .ousuosuum nwflc n m .musuosnum 30H u A .mucmEHummxo wcmepom HmEHuao ecu mo mxmp co>mm umuww mg“ now munuosuum cwwa paw 30H CH mmmn susoEmwumH kn cofiuomfiwm >opa mo mcuouuwm .m mqm1E4mm2M, M:>H, P<:.05). Thus, for a given Callibaetis density, the encounter rates can be ranked: L M H. A quite different relationship was observed with the fathead minnow, Pimephales promelas, as prey. Here the slope of the regression 59 FIGURE 2. Mean encounter rates (# captures/second of search, i 1 SE) for largemouth bass in relation to the density of Callibaetis mayflies in laboratory environments with low (L), medium (M), and high (H) amounts of structure. Multiply mayfly densities by 8.33 to obtain #/m3. Lines were drawn from the regression equations in Table 1. 60 CV om ON or 35ch mzwmflzmo CV Om ON or u _ 0v Om ON or — q .N mmDUHm ‘Q‘t‘YQQ‘QV‘V v-' T- v- r- (s/#) 9193 Jaiunooua 61 TABLE 1. Fitted parameters (+ 1 SE) of the linear regression equation describing encounter rates (B) by largemouth bass with may— flies in three structurally different environments. The form of each equation is B = a + b x, with x = mayfly density 1 (number per aquarium). Environment a b Significance n 1 level Low structure —.06 i .21 .040 i .008 .001 83 Medium structure .04 + .12 .015 + .004 .005 84 High structure -.02 + .03 .004 + .001 .005 85 62 line relating encounter rate to prey density was actually increased in medium structure compared to low or high structure (P‘=.05) and there was no significant difference between low and high structures (Figure 3, Table 2). Higher encounter rates with fathead minnows in medium structure were the result of the combined effects of struc— tural complexity and the behavior of the prey. The fathead minnows tended to aggregate in schools, doing so most dramatically in the low structure environment and to a much lesser extent in the more struc- tured aquaria. The bass avoided dense aggregations of fatheads, low- ering the encounter rate in low structure. Although the fatheads schooled little in high structure, the bass had difficulty finding them because of the increased amount of structure. As a result, the encounter rates were highest in medium vegetation. The bass in low structure seemed to search for solitary fat— head minnows, bypassing aggregated groups of the prey. Many studies have demonstrated that schooling reduces the susceptibility of fish prey to predators (Radakov 1958, Neill 1970 and 1974, Shaw 1978). In this study no quantitative record was kept relating capture success to the spatial position of a prey item relative to other prey, but qualitatively it did seem that fewer captures resulted on occasions when bass attacked a group of fatheads. Also, the regression of handl— ing time on fathead density in low structure had a positive slope and was moderately significant (P<=.10). Prey density should be positively related to the probability of schooling and the average number of prey individuals in a group. Since handling time increased with density, there was an indication that schooled prey had reduced value (Bi/Hi)° 63 FIGURE 3. Mean encounter rates (# captures/second of search, + 1 SE) for largemouth bass in relation to the density of fathead min— nows (Pimephales promelas) in laboratory environments with low (L), medium (M), and high (H) amounts of structural complexity. Multiply prey densities by 8.33 to obtain #lm3. Lines were drawn using the regression equations in Table 2. 64 I: u T a E 4 lLJlll “QY‘Q‘Yfi (S/#) 3193 Jaiunooua 2O 30 1O 20 30 10 20 3O Pimephales Density 1O FIGURE 3. 65 TABLE 2. Fitted parameters (: 1 SE) of the linear regression equa— tions describing encounter rates (B) by largemouth bass with fathead minnows in three structurally different environments. The form of each regression equation is: B = a + blx, with x = fathead minnow density (number per aquarium). , Significance Env1ronment a b n 1 level Low structure —.04 i .07 .010 i .003 .005 83 Medium structure .07 + .06 .018 + .003 .001 99 High structure —.02 + .05 .010 + .003 .001 93 66 The positive relationship between handling time and fathead minnow density did not occur in medium or high structure, where schooling behavior was reduced. The magnitude of the effect that structure has on encounter rates should be quite prey—specific. Larger, more mobile prey should have encounter rates which, relative to smaller, more sedentary prey, are less influenced by changes in the quantity of environmental struc— ture. Thus it was expected that bass would have higher encounter rates with actively swimming, larger Pimephales than with sedentary Calli- baetis, and that the changes in encounter rates associated with dif— ferences in structure would be smaller for Pimephales. Indeed, for a given prey density, the highest encounter rate (in medium structure) was only 1.8 — 2 times greater than the lowest encounter rates (in low and high structure) for Pimephales, whereas for Callibaetis there was a tenfold increase in the encounter rate from high to low structure (Table 3). At medium and high structures, the encounter rate was higher with Pimephales than with Callibaetis at any given density, as expected (Table 3). But in low structure, mayfly encounter rates were higher, due to the fathead schooling effect and an additional factor. In low structure, where bass obtained unobstructed views of their surround— ings, it was possible for there to be several mayflies within the reac— tive field (that region of space within which the bass can detect prey). Thus, several prey could be captured without the necessity of inter— vening search, increasing the mayfly encounter rate, which was ex— pressed in captures per second of search. This phenomenom could not 67 oH. H mm. NH. H NN. «H. H 04. so. H NH. mH. + NH. NN. H NN.H com mo. H Nm. oH. H Ho. oH. H am. No. H o_. 00. H OH. NH. H «H.H omN 00. H 0N. No. H om. mo. H wN. No. H mo. 00. H OH. NH. H 00. ooN no. H oN. no. H am. No. H NN. mo. H mo. NH. H Hm. oN. H 00. cmH 00. H HH. as. H aN. mo. H OH. Ho. H mo. OH. H NN. NN. H Ne. 00H 00. H mo. mo. H NH. oH. H OH. 00. H «00. oN. H NH. mm. H NH. an m 2 A m z A muflmcoo mmNmaameHm HHHHHBHHHHO mmHuoucfi cam moumu Houcaoocm .musuosuum swfiz I : tam 30H I 4 .kuwonQEoo HmNSHQDHum wo mausoEm ucmuommwc LHHB kmua ucwuowmww um houa AMMMMMNMHMMV soCCHE wmonumm paw %n A.H .o fimo cuws Loummm wo wcooom pom mousudmov moumu ummcHH msu chm: woumasoamo mNoB .wusuosuum Esfiuoe I z .ounuosuum mucoeconfl>cm CH AmE\%V mofiuwmcop mwummnwfifimo auws mmmn LuzoEmwumH HQUCDOOCQ #2me Umuowflmhm om MmefiH 68 occur with fatheads, which fled from sites of prey capture and had to be searched for anew, nor could it occur at higher structural levels where the shielding effect of the vegetation reduced the probability of there being multiple prey simultaneously in view. Thus prey beha— vior and spatial distribution, in addition to size, are important fac- tors in determining prey encounter rates at different levels of struc— ture. Bass predators had much higher encounter rates with Callibaetis than with Coenagrionid damselflies at all structural levels (Table 4). Despite the fact that individuals of the two prey types were very sim- ilar in size, the two prey were quite different behaviorally. The Coenagrionids were climbers, always positioning themselves on the veg— etation, whereas many Callibaetis individuals were found on the sand substrate. As a result, in low structure the Coenagrionids were hid— den by the stems and leaves of the small amount of vegetation that was present while Callibaetis individuals were relatively unprotected. Average Coenagrionid search times were eleven times greater than Calli- baetis search times in low structure (Table 4). As the amount of struc- ture increased and Callibaetis gained protection from the interposed vegetation, the differences in average search times between the two prey became proportionately smaller but were still highly significant. Welch's t' statistic (Gill 1978) was employed to make comparisons between prey types because of heterogeneous variance. The large observed differences in the search times for these similar sized prey indicate that the specific effects of structure will depend on whe— ther structure intervenes between a predator and its prey (as it did 69 mooo. 0H m.q H.0H + o.mw o.e + N.HH an: +l mooo. mm H.m o.m + o.NN m. w.m Esfipmz mooo. om o.m o.N H w.HH N. H H.H 304 m > .u mENH Loummm >awamm8ma oENH noumwm >Hw>mz wusuosuum .comwuma IEoo some you co>ww mum uopuo H oa%9 mo xuwfiwnmnoua paw .Eovmmhm mo mmmuwow .ofiumflumum .u m.£onB mo mmSHm> .mE\0H¢ vmfimsvo muwmcmv kHWHowEmp .mE\mmm mmB xuwmcov %Hw%mz .mufimecou IH>am wousuosuum >auCoumMMNc SH mmHHmHmmEmn vflcowummcmoo paw mmwfim%me mwummnwafimo noouwv cu mmmn nanoEowumH kn vmuflncou Amvcooom cw .mm H “V mmEHu goumom mo mcomwudeoo .q mqm I l 1 1 1 1O 20 30 4O Callibaetis Density FIGURE 4. 77 TABLE 8. Categorization of handling times (in seconds) for large- mouth bass with Callibaetis. Two contrasts are presented as well as the 95% confidence interval for the differences between the means on opposite sides of the contrast (Scheffe's Interval). Handling Time Characteristic Mean Handling Time : 1 SE Isolated 5.8 i .3 Following 3.6 i .3 Cut short 2.6 i .2 Contrast 95% Confidence Interval Isolated vs. Following 2.2 i 1.0 + \0 Following vs. Cut short 1.0 _ 78 (Table 8). Isolated handles are the longest, as one would expect. Following handles take less time, presumably because of shortened pur— suits, and cut short handles are the shortest of all, a full three seconds less than isolated handles. Apparently, the stimulus provided by a nearby prey item is the strongest factor in shortening handling times, with decreases in pursuit distance (time) also important. For each contrast between the means of the handling types, a 95% confidence interval is given. Scheffé's interval was utilized since the contrasts were selected postdata (Scheffé 1953). RATES OF MOVEMENT AND ENERGETIG COSTS Structure also affected the rates of movement of bass while searching for or handling prey (Table 9). A threshold effect was apparent. Rates of searching for Callibaetis and Pimephales‘were sim- ilar, with a marked decline in the rate of search for each prey type in high structure. Handling velocities were higher for Pimephales than Callibaetis, especially in high structure; greater pursuit vel— ocities were required to capture the fish prey. The average Pimephales handling velocity increased in high structure compared to medium and low (Table 9). Apparently a high velocity of pursuit was needed in high structure to overcome the fat- heads' propensity to escape by abruptly and erratically turning while moving through the dense vegetation. In contrast, the average Calli— baetis handling velocity dropped in high structure. Since metabolic 79 TABLE 9. Average velocities (cm/s : 1 SE) of largemouth bass while searching for or handling two types of prey in three differ— ently structured environments. L — low structure, M - medium structure, and H — high structure. Search Velocities Prey Type L M H Callibaetis 6.2 i .5 6.1 i .5 3.6 i .3 Pimephales 6.8 I .6 6.7 i .4 3.8 I .4 Handling Velocities Callibaetis 3.2 i .3 3 1 i .3 1 7 i .2 Pimephales 4.7 i 4 4.7 i 3 5.5 i .5 80 cost is related to activity level, the cost of handling Callibaetis decreased while the cost of handling Pimephales increased in high structure, implying that Ei’ the net energetic gain associated with a particular prey item, declines for Pimephales in high structure and is enhanced for Callibaetis, relative to less structured environments. The differences in costs associated with the changes in velo— city in high structure are small, however. By increasing its Pimepha- leg handling velocity from 4.7 cm/s in medium structure to 5.45 cm/s in high structure, a 70 mm largemouth bass increases its cost from .0022 cal/s to .0023 calls (from Equation 4). Thus bass in medium and high structure would have to handle fatheads for almost three hours a day for there to be even a 1 calorie savings in cost in medium struc- ture. Similarly, bass handling Callibaetis in high structure reduce their cost of handling by .0001 cal/s compared to medium structure. If handling times and encounter rates were similar between structures, with handling time set at six seconds, the bass would have to handle almost 1700 mayflies/day in order for bass in high structure to have a 1 calorie/day advantage based on reduced cost. Thus the small differences in costs of handling prey associ- ated with different structural levels have little effect on the expres— sion describing net energetic gain per prey item: Since the bass are able to capture fairly large prey, the equation is swamped by the ei term, the total energy content of the prey. ei is 81 expressed in calories, while C the cost of handling, is expressed h, in calories X 10_3/s. Accordingly, the increased Pimephales handling cost in high structure results in an Ei of 34.6167 in high structure compared to 34.6231 cal in medium, a reduction of only 6.4 X 10_3 cal per fathead. Since handling time (Hi) is multiplied by the cost of handling term, changes in handling time also have minimal effects on Ei. For example, increasing the Pimephales handling time to 20 sec- onds in high structure reduces E1 to 34.6039 cal, a .019 calorie reduc— tion in energy per fathead compared to medium structure. The significant differences in growth observed between bass in differently structured field environments (Anderson 1983b) are thus not caused by structure related changes in energetic cost per prey item. It is doubtful whether differences in search costs are important, either. Bass in high structure might have to search longer to obtain a given amount of energy than bass at intermediate structures because of lowered prey encounter rates in high structure. But this extended searching would only tend to balance out the energy costs between structural types, since it is actually slightly cheaper to forage at higher structures, with lower search velocities. DIET BREADTHS Even though structure has a miniscule effect on the net energy gained per prey item, structure should have important effects on prey selection. Structure alters prey encounter rates (Figures 1 and 2, Table 4) and can have strong effects on prey values (Table 10), two 82 TABLE 10. Calculated values (cal/s) of three different prey of large— mouth bass at three levels of structural complexity. L, M, and H stand for low, medium, and high levels of structure. Prey L M H Fathead minnows 3.4 2.9 2.4 Mayflies 1.8 1.6 1.0 Damselflies 1.1 1.0 .9 83 factors which are important in determining patterns of prey selection (Mittelbach 1981a). The value of a prey item (Ei/Hi) should be an indicator of its desirability for a predator. For the three prey types utilized in this study, the major shift in value across structures occurs for Callibaetis (Table11», with its value dropping almost in half from low to high structure. This is because Callibaetis handling times are so sensitive to structure, almost doubling over the range of structures used in these experiments. There are moderate changes for Pimephales. Note that it has a value 1.8 times greater than Callibaetis in low structure and is worth 2.5 times as much in high structure, a trend which is opposite to what one would expect based on the reduced Pime- phales capture success in high structure. Coenagrionids are relatively resistant to changes in value because their handling times change lit— tle with structure. The result is that Coenagrionids and Callibaetis become very similar in value in high structure. Thus, if mayflies are included in the diet in high structure, slightly larger damselflies should also be included whereas in low structure the probability of damselflies and mayflies occurring together in the diet is lower. In fact, in a field study done with bass (Anderson 1983b), damselflies were absent from the diet in low and medium structures (even though present in the environment at high densities) and strongly represented in the diet in high structure (mayflies were frequent prey in all hab- itats. In a separate laboratory study, an optimal foraging model 84 proved to be an adequate predictor of the foraging behavior developed by bass in differently structured environments (Anderson 1983a). Therefore, the energy—maximizing model described earlier (Equation 1) was employed, using the foraging parameters determined in these exper— iments, to predict general patterns of diet breadth for bass in struc— tured environments. Figure 5 represents predicted diet breadths of bass at three structural levels with a simple prey community of fathead minnows and Callibaetis mayflies. Under natural conditions bass would never encounter such a simple prey community, but the relationships shown in Figure 5 are meant to portray structure—related diet breadths of bass when fish and invertebrate prey types are present. The hatched areas represent combined densities of fatheads and mayflies wherein a 70 mm bass can maximize its rate of energy intake by specializing on the more valuable prey — fathead minnows. Note that the highest probability of specialization occurs in medium structure, where bass have the high— est encounter rates with minnows. Despite the fact that encounter rates with fathead minnows are similar in low and high structures, bass in low structure should be generalists at minnow densities up to 58/m3, whereas in high structure bass should begin specializing at a fathead density of 25/m3 (Figure 5). This difference is due to the increased value of Callibaetis in low structure. Thus, when the prey community consists of large individuals which change little in value with struc— ture (the fatheads in this case) and smaller prey types which change more dramatically, it is quite possible that greater foraging special- ization will occur at higher levels of structure instead of in low structure. This may be the mechanism underlying the greater consumption of large prey by bluegills at higher structural levels in Crowder and 85 FIGURE 5. Graphical representation of the predicted foraging modes of 70 mm largemouth bass at different densities of two prey types, fathead minnows (Pimephales promelas) and Callibaetis mayflies, in environments with low (L), medium (M), or high (H) levels of structure. Specialization on the preferred prey, Pimephales, is predicted in the hatched areas. Open areas of the graphs represent combined prey den— sities at which bass should be generalists, taking both prey as encoun— tered. Fathead dry weight = 9.9 mg, mayfly dry weight = 2.0 mg. 86 .m wusmfim Amati 36:8 5235:. On 00 Om OV Om ON 0—. 0 ON 00 Om O? on ON or O Oh 000m O? on ON 0— O . . _ O u _ . a H H . . \._ JOF _ ION \_ \u 1 \i ”H lOm 10v [Om (SUI/4H Misued sueeqmeo 87 Cooper's experimental work (1982). Note that the graphical represen— tation neglects the positive relationship between Callibaetis density and value in low structure. Such an effect increases the probability of specialization on Pimephales at low Callibaetis densities and makes inclusion of Callibaetis more likely at high densities. Thus a verti— cal line separating regions of specialization and generalization would be inappropriate. Using a simple prey community of Pimephales and another inver- tebrate prey, Coenagrionid damselflies (Figure 6), predicted regions of specialization are greater than in the Callibaetis case because the Coenagrionids were less valuable. Generalization is predicted in medium structure only when fathead densities drop almost to zero. There is a slight difference in the predicted pattern of resource utilization between low and high structures. The Pimephales encounter rates are similar in the two environments but specialization is predicted at a slightly lower minnow density in high structure (15 vs. 23/m3) because of the small decrease in Coenagrionid value with structure. If one looks at an invertebrate community (Figure 7) of may— flies and damselflies, the simple trend of greater generalization with increased structure finally holds. This is because for any given den— sity of the preferred prey, Callibaetis, the encounter rate is decreased as structure is increased, making it more likely that Coenagrionids would be an energetically acceptable part of the diet. In high struc— ture generalization is predicted over the entire range of Callibaetis densities examined (0—700/m3). Note, however, that the patterns of predicted diet breadth would change markedly if the damselflies were 88 FIGURE 6. Graphical representation of predicted foraging modes utilized by 70 mm largemouth bass at different densities of fat— head minnows (Pimephales) and Coenagrionid damselflies in three differ— ently structured environments. Specialization on Pimephales is pre— dicted in the hatched areas. Both prey should be eaten if their com- bined densities lie in open areas of the graphs. Fathead dry mass = 9.9 mg, damselfly dry mass = 2.3 mg. 89 ON. Om om 0v Om ON or O Amer: 37.50 8.232.. Oh Om Om O? on ON 0— O Y _ on om Om ov Om ON or . o mmDOHm loN [on 10? 10m [Om .Oh (Em/m Aigsuaa pguogifieueoo 90 FIGURE 7. Graphical representation of predicted diets of 70 mm largemouth bass at different densities of Callibaetis mayflies and Coenagrionid damselflies in low, medium, and highly structured environ— ments. Specialization on Callibaetis is predicted in hatched areas; generalization (taking both prey) is energetically advantageous in open portions of the graphs. Mayfly dry weight = 2.0 mg, damselfly dry weight = 2.3 mg. 91 005 00m 00m OOV OOO OON OO_. O q u q q - rust: 36:8 $38.60 005 000 00m 00». OOm OON OO_. O A\_ .m mmonm OON. 00m 00m 00? Con OON 00? O V d u _ q O ... OO_. 1 CON 1 COM 1 00¢ 1 00m ... OOO i ooh (cw/1;) Aigsuea pguoufieuaog 92 somewhat larger (more valuable) than the mayflies. In that case, the sensitivity of mayfly value to structure would produce larger regions of predicted generalization in low structure. In fact, increased gen- eralization should occur in low structure whenever there are high num- bers of prey which are not associated with (hidden by) structure and which have values dependent on their densities. The validity of the predicted patterns of Figure 7 have been examined under field conditions using largemouth bass as predators and a prey community which consisted entirely of invertebrates and had Callibaetis mayflies as the most valuable common prey (Anderson 1983b). Despite the presence of more than 20 prey taxa, diet breadths of bass in the field corresponded with the generalizations permitted from exam— ination of Figure 7. Bass in the highly structured field environment had broader diets than their medium vegetation counterparts, even though prey densities were significantly greater in high vegetation. Bass in low structure also foraged in a more generalized manner than bass in medium vegetation. Corresponding with Figure 7, Baetid mayfly densities in the low vegetation field habitat were close to 100/m3, and most may— flies were smaller (less valuable) than those used in this study. DISCUSSION At a given density of either of the two invertebrate prey types, encounter rates decreased as the amount of structure increased. However, under natural conditions prey densities are generally 93 enhanced with increased amounts of structure (Gerking 1957). Poten— tially, such increases in prey densities might be sufficient to make prey encounter rates in highly structured environments comparable to those found in less structured environments. The relationship between structural complexity and the actual adjustments which must be made in prey densities to equalize encounter rates across structural levels are illustrated in Figure 8. The prey weighting factors are simply numbers which must be multiplied by the density of a prey type at a given level of structure so that the encoun— ter rate by bass with that prey type is equal to the encounter rate at the lowest level of structure used in this study (200 stems/m2). The weighting factors are estimated from the slopes of the regression lines relating encounter rates to prey densities for Callibaetis (Table 1) and from the relationship between structure and mean search times for Coenagrionid prey (Table 4). For example, from Table 1 the slope of the line relating encounter rate to prey density is ten times greater in low structure than in high (.04 vs. .004). Thus, for a given den— sity in low structure, prey density in high structure must be ten times greater for encounter rates to be equal. Only one density of Coena- grionid damselflies was utilized, so there is no regression of encoun— ter rate on density for this prey type; however, prey weighting factors can still be computed by using the inverse of search times (Table 4) for encounter rates and by assuming a direct linear relationship be- tween prey density and encounter rate. Note (Figure 8) that prey den- sity must increase exponentially as a function of structure to equalize encounter rates across structural levels. 94 FIGURE 8. Relationship between structural complexity (macro— phyte stems/m2) and prey weighting factors for two prey types. Prey weighting factor = the number which must be multiplied by the prey density at any level of structure so that the encounter rate by large— mouth bass with that prey type at that level of structure is equal to the encounter rate at the lowest level of structure (200 stems/m2). Dashed line represents plot of weighting factor vs. structure for Coe— nagrionid prey, dash—dot line is for Callibaetis. Solid line repre— sents plot of factors by which prey densities are actually increased across changes in structure (from Gerking 1957). Prey Weighting Factor ANQAWGVODLOO I 95 FIGURE 8. J 200 400 600 Stems/ M 2 96 Gerking (1957) examined the relationship between prey and vegetation biomass in Bryant's Creek Lake, Indiana. His vegetation and prey dry weights can be converted to number of stems and number of prey individuals, respectively, by assuming that the average dry mass of a stem of aquatic vegetation is 400 mg (Anderson, pers. obs.) and that average prey dry weight is .03 mg (Mittelbach 1981b). One can then plot prey density vs. stem density (Figure 9). A simple linear regression model fits the available data better (linear model r2 = .44) than either exponential (r2 = .41), logarithmic (r2 = .41), or power functions (r2 = .38), indicating that there is a relatively constant number of prey individuals per stem regardless of stem density. While prey densities do increase with structure (the straight line in Figure 8), the higher densities are inadequate to compensate for the loss of foraging efficiency in highly structured environments. Predators in such environments can be expected to acquire energy at reduced rates. Although the analysis of handling times provided a mechanism underlying the reduction of handling time in low structure, it left one with an unsettling question. That is, if bass have the ability to shorten handling times (the ”cut short" handles), why do they not always do so? Such reductions can only increase the rate of energy acquisition. Furthermore, the two hypotheses put forth to explain the minimization of handling time did not completely account for the differ- ence in handling times between low and high structures, since isolated handles in low structure were 5.8 5, still less than the 7.1 s aver- age in high. In addition, damselfly handling times increased steadily with structure (Tables 6 and 7) yet damselflies, concealed on the 97 FIGURE 9. Relationship between prey density and density of aquatic macrOphytes in Bryant's Creek Lake, Indiana (from Gerking 1957). The straight line is drawn from the linear regression equation: Y = 28109 + 1.9096(X). r2 = 044', P <0050 98 l l l l l l ] 17— 07‘ E14— (‘9 O 1- 1‘. 0,11— 5 ‘1? z < 8— o a: O 5— FIGURE 9. 300 380 460 540 620 700 780 STEMS/M2 99 vegetation, were captured one at a time - thus no multiple capture effects could have been working. The only other explanation that can be given is that the ener- getic penalties associated with increasing the handling time for a par— ticular prey type are much greater in low structure than in high (Table 11). In high structure, if an extra second or two is spent handling a prey item instead of searching for new prey, little energy is lost because prey encounter rates are low in high structure and the proba— bility of detecting another prey item in that short time interval is small. On the other hand, extra seconds spent handling prey in low structure can have a more profound effect on the rate of energy acqui- sition, since encounter rates in low structure are high — thus the ben— efits of using even small amounts of time for searching are greater. From Table 11, one can see that increasing the handling time from five to seven seconds for mayfly prey in low structure results in a greater than 25% drop in the rate of energy return. The same handling time change in high structure produces only an 11.5% reduction in energy intake. The observed fluctuations in handling times between densities and structures are of concern when considering the value of a prey item to a predator, when prey value = Ei/Hi° Ecologists are used to think- ing of a particular prey type as having a relatively fixed value; yet with handling times changing so markedly with density and structure such a view seems untenable. Furthermore, the optimal foraging model deveIOped by Charnov (1976) and utilized by Mittelbach (1981a) postulates that whether a prey item should be included in a predator's diet depends 100 TABLE 11. Hypothetical rates of energy acquisition (En/T) in calls for largemouth bass foraging for mayflies in three structured environments, low, medium, or high, with three different handling times. En/T is calculated for each combination of handling time and structure using the Optimal foraging model described in the text and the encounter rates given in Table 3 (with density = 300/m3). The mayflies are assumed to be worth 7 calories net. E/T n Handling Time L M H 55 1.22 1.04 .52 65 1.04 .91 .49 7s .90 .80 .46 101 only on its value to the predator, not its density (availability). This View was defended by Sih (1979). However, if prey value is de— pendent on density, this hypothesis is fallacious. The handling time — density dependency suggests that the val— ue of a prey type can be affected by the density or distribution of other prey types as well. In Figure 10, a largemouth bass is depicted foraging in environments with two prey types, M and N. Arc X repre— sents the limit of the reactive field of the bass when it is in the position shown. Arc Y defines the field's limit when the bass has reached prey M. Note that in Figure 10b, the increased density (or increased clumping) of prey N will lead to a shorter pursuit time for prey M because there is a type N individual between the bass and M; it will also cut short the handling time on M, since the individual of prey type N between arcs X and Y will be visible from the capture site of M. Thus the value of prey type M is increased in environment 10b, even though its density is unchanged from 10a. Obviously, the kind of relationship described here will work best in relatively unstructured environments, or with prey which are not strongly associated with structure. In this study the Coenagrio— nids, always found on the vegetation, had handling times that were not affected by density even in low structure; mayfly handling times were sensitive to density in low but not in high structure. Overall, the multiple prey in view phenomenom, by increasing the value of certain prey, should tend to make predators' diets in unstructured environments more generalized than might otherwise be expected. Structure should also affect the ability of predators to 102 FIGURE 10. Schematic representation of the alteration in value of one prey type due to the change in distribution of another prey species. Individuals of the two prey species are represented by dots labelled M and N. Curved line X designates the limit of the fish's reactive field when it is in the position shown. Y designates the reac— tive field limit once the fish has reached prey type M. In 103, spe— cies N is at low density or is not clumped; in 10b the density or clumping of N has increased. 103 a X 0N 13C>~-—> M b X Y 0N M’?.oN FIGURE 10. 104 change their patterns of prey selection as familiar prey decline in abundance or new prey become available. At first glance it would seem that predators in low structure would be at an advantage in tracking changes in resource levels; they have higher encounter rates with prey and a better view of what prey may be present. In a set of experiments designed to investigate the interaction between structure and learning, bass in low structure which had been trained in the laboratory to for- age only for damselflies were much quicker at detecting and capturing novel fish prey than were their high structure counterparts (Anderson 1983c). On the other hand structure can have surprising effects on the ability of fish to adjust their diets. In another experiment, bass in low and high structure were given the opportunity to forage for a sim— ple prey community of fish and damselflies. The low structure bass specialized on fish; the high structure bass generalized. Then a third prey was added — Anax dragonflies of sufficient size to be an energeti- cally favorable prey in both structure types. In this case the bass in high structure added the new prey to the diet more quickly. The poten— tial mechanism underlying the unpredictability of the effect of struc— ture on learning could be that at high levels of structure all prey types become spatially associated with structure and searching for prey becomes a simple process of scanning vegetation. In high vegeta- tion new prey will eventually be detected even if they are encountered at a low rate. At lower levels of structure, the environment may be split into discrete parts — vegetation, open water, and sediments in the aquatic case — and specialization on a prey type found in one of the environmental subdivisions may make predators less able to track 105 changes in resources elsewhere. In the example given, low structure bass temporarily continued to forage for fish in open water even though more profitable prey were located in the vegetation. Apparently, it will be difficult to make sweeping generaliza— tions regarding the effects of structural complexity on predators' for— aging behavior and patterns of prey selection. Since structural com— plexity can influence almost all aspects of the foraging process, in— cluding the ability of predators to encounter, handle, and learn about prey as well as prey value and behavior, and since the magnitudes of these effects are prey—specific, it may be more prudent to analyze par— ticular ecological settings and make concrete appraisals of the effects of structure on predators and their prey. Indeed, experimental investi— gations of the relationship between structure and diet breadth have thus far yielded somewhat contradictory results. Vince et al. (1976) and Anderson (1983a) observed broader diet breadths at high structural lev— els; in contrast Stoner (1982) obtained narrower predator diets in high structure, where one prey type gained almost complete refuge from preda— tion. The relationship between the foraging behavior of predators and structural complexity may be of critical importance in understand- ing the mechanisms which regulate the community ecology of littoral zone areas. Nelson (1979) has argued that predation by fishes and decapods is the most important factor in determining the distribution, diversity, species composition, and abundance of amphipods associated with macrophytes in an eelgrass community. Of course the intensity of such predation will be governed by the structural complexity of the 106 environment. In turn, specific foraging behaviors and patterns of resource utilization should determine the amount of competition between predator species. The recent successes of optimization theory in predicting such patterns (Belovsky 1978, Mittelbach 1981a) have created hope that the extent of interspecific interactions can be quantitatively predicted and mechanistically described. The present study suggests that optimal foraging models should be used with great care. A key element in such models is the time required by predators to handle prey items. Handl— ing time, traditionally viewed as dependent on physical attributes of the predator and its prey (Werner 1977), can also be influenced by prey density and the spatial distribution of prey and may to a certain extent be under the predator's control — witness the difference in handling time between low and high structure even when multiple prey in view phenomena have been accounted for. Such control of prey value should not be surprising. Staddon (1977) has described cases wherein the rap— idity with which animals become responsive (search for food) following reinforcement is proportional to their expectancy of future reward. Nonetheless the dependency of prey value on so many variables is trou— blesome since prey value constitutes such an important part of the optimal foraging model. Either the notion of prey value may have to be redefined or else the ranking of prey items and the precise predic— tions of predators' diets under complex field conditions will be exceedingly difficult. 107 LITERATURE CITED Adams, C. 1975. The nutritive value of foods. United States Depart— ment of Agriculture. Washington, D. C. Anderson, 0. 1983a. Optimal foraging by largemouth bass in structured environments. Ecology 64:000-000. Anderson, 0. 1983b. Prey selection and growth of largemouth bass in differently structured stream environments. Submitted to Animal Behaviour. Anderson, 0. 1983c. The interaction between structural complexity and the composition of the prey community in determining changes in the foraging behavior largemouth bass. In prepa— ration. Belovsky, G. E. 1978. Diet optimization in a generalist herbivore: the moose. Theoretical Population Biology 14:105—134. Charnov, E. L. 1976. Optimal foraging: attack strategy of a mantid. American Naturalist 110:141—151. Crowder, L. B., and W. E. Cooper. 1982. Habitat structural complexity and the interaction between bluegills and their prey. Ecology 63:1802—1813. Cummins, K. W., and J. C. Wuycheck. 1971. Caloric equivalent for investigations in ecological energetics. Internationale Vereinigung ffir Theoretische und Angewandte Limnologie, Mitteilungen 18:1—158. 108 Curio, E. 1976. The ethology of predation. Springer—Verlag, New York, New York, USA. Elliot, J. M. 1976. The energetics of feeding, metabolism and growth of brown trout (S2192 trutta L.) in relation to body weight, water temperature, and ration size. Journal of Animal Ecology 45:923-948. Elliot, J. M., and W. Davison. 1975. Energy equivalents of oxygen consumption in animal energetics. Oecologia 19:195—201. Elner, R. W., and R. N. Hughes. 1978. Energy maximization in the diet of the shore crab, Carcinus maenas. Journal of Animal Ecology 47:103—116. Emlen, J. M. 1966. The role of time and energy in food preference. American Naturalist 100:611—617. Gerking, S. D. 1957. A method of sampling the littoral macrofauna and its application. Ecology 38:219—225. Gill, J. L. 1978. Design and analysis of experiments in the animal and medical sciences. Iowa State University Press, Ames, Iowa, USA. Glass, N. R. 1971. Computer analysis of predation energetics in the largemouth bass. Pages 325—363 in B. C. Patten, editor. Systems analysis and simulation in ecology. Volume 1. Aca- demic Press, New York, New York, USA. Goss—Custard, J. D. 1977. Optimal foraging and the size selection of worms by redshank, Tringa totanus. Animal Behavior 25:10—29. Ivlev, V. W. 1961. Experimental ecology of the feeding of fishes. Yale University Press, New Haven, Connecticut, USA. 109 Krebs, J. R., J. T. Erichsen, M. I. Webber, and E. L. Charnov. 1977. Optimal prey selection in the great tit (£3533 £3125). Animal Behavior 25:30-38. MacArthur, R. H., and E. R. Pianka. 1966. On optimal use of a patchy environment. American Naturalist 100:603—609. Mittelbach, G. G. 1981a. Foraging efficiency and body size: a study of optimal diet and habitat use by bluegills. Ecology 62: 1370—1386. Mittelbach, G. G. 1981b. Patterns of invertebrate size and abundance in aquatic habitats. Canadian Journal of Fisheries and Aquatic Science 38:896—904. Neill, S. R. St.J. 1970. A study of antipredator adaptations in fish with special reference to silvery camouflage and shoaling. Ph.D. Thesis, Oxford University. 159 pp. . 1974. Experiments on whether schooling by their prey affects the hunting behavior of cephalopods and fish. Journal of Zoology, London, 549—569. Nelson, W. G. 1979. An analysis of structural pattern in an eelgrass (Zostera marina L.) amphipod community. Journal of Experimen— tal Marine Biology and Ecology 39:231—264. Radakov, D. V. 1958. On the adaptive significance of shoaling of young coalfish (Pollachius virens). Vop. Ikhtiol 11:69—74. Rice, J. A., and J. E. Breck. 1982. Synthesizing fish respiration data for bioenergetics modeling; examples using largemouth bass and striped bass. Submitted to Ecology. 110 Scheffe, H. 1953. A method of judging all contrasts in the analysis of variance. Biometrika 40:87—104. Schoener, T. W. 1971. Theory of feeding strategies. Annual Review of Ecology and Systematics 2:369-404. Shaw, E. 1978. Schooling fishes. American Scientist 66:166—175. Sih, A. 1979. Optimal diet: The relative importance of the parame— ters. American Naturalist 113:460—463. Staddon, J. E. R. 1977. The handbook of operant behavior. J. E. R. Staddon and Werner K. Honig, eds. Prentice—Hall, Englewood Cliffs, New Jersey, USA. Stoner, A. W. 1982. The influence of benthic macrophytes on the foraging behavior of pinfish, Lagodon rhomboides. Journal of Experimental Marine Biology and Ecology 58:271—284. Vince, S., Valiela, I., Backus, N., and J. M. Teal. 1976. Predation by the salt marsh killifish Fundulus heteroclitus in relation to prey distribution and abundance. Journal of Experimental Marine Biology and Ecology 23:255—266. Ware, D. M. 1973. Risk of epibenthic prey to predation by rainbow trout (Salmo gairdneri). Journal of the Fisheries Research Board of Canada 30:787—797. Werner, E. E. 1977. Species packing and niche complementarity in three sunfishes. American Naturalist 111:553-578. Werner, E. E., and D. J. Hall. 1974. Optimal foraging and the size selection of prey by the bluegill sunfish (Lepomis macrochirus). Ecology 55:1042—1052. 111 Werner, E. E., and G. G. Mittelbach. 1981. Optimal foraging: field tests of diet choice and habitat switching. American Zoologist 21:813—829. PREY SELECTION AND GROWTH OF LARGEMOUTH BASS AT DIFFERENT MACROPHYTE DENSITIES Owen Anderson Zoology Department Michigan State University E. Lansing, MI 48824 112 113 Littoral zone aquatic macrophytes can have important effects on the interactions between fish predators and invertebrate prey. Macro— phytes provide attachment surfaces for periphyton and harbor a variety of invertebrate organisms which serve as food for fish populations. Increased amounts of aquatic vegetation have been shown to be associa— ted with greater biomasses and numbers of invertebrates (Gerking 1957, Hruska 1961). Although macrophyte and prey densities are positively related, macrophytes reduce the efficiency of fish predators by decreasing the velocity with which they search for prey, by making prey more difficult to detect, and by making some prey more difficult to capture (Anderson 1983). When resource levels are held constant, increased macrophyte levels have been shown to lower the rate of energy vauisition of fish predators (Vince et al. 1976, Stoner 1982, Anderson 1983). In fact, most prey populations would have to increase exponentially as a function of macrophyte density in order for fish predators to have equal prey encounter rates at all macrophyte levels (Anderson 1983). However, prey density appears to be linearly related to macrophyte density (Gerking 1957, Anderson 1983). Thus, fish predators should have steadily declining prey encounter rates as macrophyte levels are increased except that, in areas with small numbers of macrophytes, prey communities tend to be impoverished. Thus, Crowder and Cooper (1982) predicted that fish would have highest growth rates at inter— mediate macrophyte densities, a prediction which was corroborated in their experimental work with bluegill sunfish. 114 Crowder and Cooper (1982) hypothesized that fish which for- aged in intermediately vegetated environments would have narrower diet breadths than fish in low or high vegetation. The rationale for this argument can be understood by studying Figure 1. Basically, as the rate of energy acquisition declines from intermediate to high or low levels of vegetation due to decreased encounters with prey, more prey types become acceptable parts of the diet (consumption of such prey can in— crease the overall rate of energy intake). Ivlev (1961) and Werner and Hall (1974) have documented increases in diet breadth with lowering of resource levels for fish. However, Crowder and Cooper (1982) generally observed narrow— est diet breadths in low vegetation despite the fact that greatest growth occurred in medium vegetation, suggesting that a simple energetic expla— nation of diet breadth may be inappropriate. The factors which influ— ence diet breadth as a function of macrophyte density may be quite com— plex and interrelated. In addition to the energetic feedback mechanism proposed by Crowder and Cooper, at least three other models can be pro— posed: (1) First, the manner in which fish search for prey might differ as a function of vegetational density. In littoral zones with modest amounts of vegetation, the overall environment can be more readily par— titioned into discrete subdivisions of vegetation, substrate, and open water. Fish which search for prey in one of these microhabitats may have reduced efficiencies in exploiting prey found in the other micro— habitats (Werner et al. 1981). Increased specialization would then be expected at reduced macrophyte levels. At high macrophyte densities, 115 FIGURE 1. Representation of mechanism by which diet breadth increases at high or low macrophyte densities. The solid line plots net energy gained (En) vs. time spent foraging (T) for a fish preda— tor in an environment with intermediate amounts of vegetation. The dashed line is for the same relationship at high or low macrophyte densities. Symbols show net energy gained from eating an individual of a particular prey type (Bi) vs. the handling time for that prey type. Stars represent prey types selectively preyed upon at all macrophyte densities; prey represented by closed circles are not con- sumed in intermediate vegetation because consumption of such prey would lower En/T° Open circles represent prey types which are not eaten at any macrophyte level. 116 FIGURE 1. 117 all prey types would be associated with vegetation, and fish predators which simply scanned the vegetation for prey might have broad diets. This mechanism could account for the patterns observed by Crowder and Cooper (1982). In their study, bluegills in low vegetation special— ized to the greatest extent on zooplankton found in open water., (2) Alternately, high macrophyte levels might lessen competition be— tween invertebrate species and promote prey community diversity because of either greater productivity or increased opportunities for spatial partitioning of the environment by invertebrates. Fish diets might then tend to expand at higher macrophyte levels because there would be more prey species available and/or fewer rare, seldom encountered prey types. (3) Finally, prey species might be differ~ entially shielded from predation at high macrophyte densities. Small prey species could gain complete refuge and be unavailable in densely vegetated environments. Substrate prey would have a protective blan— ket of vegetation covering them. Diet breadth might then be negatively related to the abundance of macrophytes, due to the reduced number of prey species actually available to fish at high macrophyte densities. Hypothetical patterns of diet breadth in relation to macro— phyte density are summarized in Table 1. The energetic feedback model proposed by Crowder and Cooper for structured environments is number 4 in the table. Note that models 1 and 2 predict the same pattern of diet breadth. It is possible to differentiate between the two, how— ever, because model 1 predicts that fish in sparse vegetation will tend to consume prey types from only one microhabitat whereas model 2 says 118 TABLE 1. Expected patterns of diet breadth in relation to macrophyte density based on four possible underlying mechanisms. Mechanisms are numbered as in the text. L = low, M = medium, and H = high macrophyte density in this and all other tables. M=>H means that diet breadth is greater at medium than at high macrophyte levels. Mechanism Predicted Diet Breadth (1) Reduced macrophyte levels produce Hi’M=:L greater microhabitat (open water, sub— strate, or vegetation) specialization by fish. Fewer prey types are avail— able to fish foraging in only one microhabitat. (2) Macrophytes promote prey diversity; H:>M=>L fish predators are relatively non-selective. (3) Macrophytes offer almost complete L=>M:>H refuge from predation for some prey types; fish predators are relatively non-selective. (4) Macrophytes alter rate of energy H,L::M intake while foraging; fish exclude prey which lower energy return. 119 only that fish in low vegetation will eat fewer prey types but does not specify that they be from one subhabitat. Note also that model one applies to the diet breadths of individual fish rather than an entire population of fish. Obviously, a fish population as a whole might be quite generalized in terms of prey consumption while indi— viduals within the population were specialized if there were signifi— cant differences between fish in the ability to exploit different microhabitats. Unlike models one and four, models two and three imply no feedback between rate of energy intake and prey selection. In models two and three, prey are eaten as encountered, with little active selec— tion or rejection of prey occurring. If selection of prey does occur, the same criterion of prey acceptability is used at each vegetation level. If only prey above a certain size are eaten, that same size requirement is utilized at each macrophyte density. Mechanisms one and four imply a causal relationship between the rate of energy acqui— sition and prey selection. Fish avoid certain microhabitats (model one) or certain prey types (model four) because their foraging efficiency would be compromised by utilizing such resources. Models two and three are not mutually exclusive. In fact, if the two mechanisms acted in concert, one would expect little change in diet breadth in relation to macrophyte density. Furthermore, the macrophyte effects of models two and three could be coupled with the energy sensitive fish of model four to produce additional patterns of diet breadth. An examination of how the models' predictions compare with actual prey consumption patterns requires the collection of 120 accurate quantitative information regarding how changes in fish energy intake rates, relative prey encounter rates and energetic values, and prey community diversities are associated with alterations in macro— phyte densities. The relationship between diet breadth and macrophyte density is important for several reasons: (1) If diet breadth changes with vegetational density, interspecific competition among fish and fish predator—prey relationships may be affected by the level of macrophytes found in a given lake. (2) Effective management of a given fish species may require knowledge of that species‘ ability to use resources in rela— tion to macrophyte density. (3) Analysis of the relationship between macrophyte levels and diet breadth can provide insights into the mecha— nisms underlying prey selection. The manner in which predators select prey has been a subject of much recent interest to ecologists (Krebs 1978, Werner and Mittelbach 1981). In this study I quantify growth rates and patterns of prey selection for young largemouth bass (Micropterous salmoides) in field environments in which macrophyte densities have been manipulated. Largemouth bass were studied because of their ubiquity and ecological importance in North American lakes (see Heidinger 1975). Furthermore, after an initial short period in the open water, largemouth bass spend the early parts of their lives foraging in aquatic vegetation, first for invertebrates and then for other fish prey (McCammon et al. 1964). 121 METHODS The study was carried out from June — September, 1979, in an experimental stream channel at an EPA outdoor laboratory in Monticello, Minnesota. The stream, which contained a full complement of natural prey types (see Appendix), was 2.5 m wide and 523 m in length and fea— tured an alternating pattern of 30.7 m long riffles and pools. The riffles had a stony substrate and no vegetation; the pools had muddy bottoms and varying levels of aquatic macrophytes, primarily Elodea canadensis and Ceratophyllum demersum. Current velocities, measured at the surface, averaged 6 cm/s in the pools and 11 cm/s in the riffle areas. Stream temperatures varied from 11 to 26°C. over the course of the summer. In the stream, three different habitat types were created with either low, medium, or high amounts of vegetation in the pools. Each habitat was 122.8 m long and contained two pools, each preceded by a riffle area. Pool macrophytes were manipulated from the stream banks using long handled rakes. Seven observers independently rated the vegetation with regard to both horizontal density and vertical height, and the macrophyte densities were altered until there was uniform observer agreement on differences between habitat types. The high vegetation habitat had no areas of bare substrate, and the dense vegetation extended vertically almost to the water surface (1 m). The medium vegetation habitats were thinned lightly; vegetation extended about 2/3 of the way to the water surface, stems were slightly further apart, and there were small areas of exposed substrate. In 122 low vegetation, approximately 1/3 of the substrate had no vegetation, and the vertical extent was limited to 1/2 m. Vegetation was main— tained at the appropriate level in each treatment throughout the remain— der of the summer. Quantitative samples taken at the end of the summer revealed that stem densities in low, medium, and high vegetation were approximately 200, 450, and 700 stems/m2, respectively. Barriers were created at the ends of each 123 m long habitat by attaching rigid hardware screen to extended railroad ties, placing one end of each tie on opposite stream banks, and anchoring the screen to the stream substrate using large boulders, sand, and gravel. The screen permitted free movement of water, but not fish, between habitat types. 200 largemouth bass and 200 smallmouth bass (Micropterous d212— migg) were placed in each of the three habitats — low, medium, and high vegetation — on June 26, 1979. The foraging behavior and growth of the smallmouth bass will be described in a forthcoming paper. Initial largemouth total length was 41.8 i .4 mm (n = 63); the smallmouth were 38.0 i .4 mm (n = 133). Both species were generously provided by the New London National Fish Hatchery, New London, Minnesota. The bass density (26.4 kg/ha) utilized was well within the limits found in natural bass populations (Hackney 1975). At approximately weekly intervals from 7/3/79 to 9/10/79, prey samples were taken in each of the treatments by tying together ten Elodea stems, weighting each vegetation clump with lead sinkers, and placing the vegetation in the center of a collapsed 23 cm diameter blind end plankton net (# 10 mesh, 153 um) which was then lowered into 123 place on the pool substrate. On each date, two samples were placed at random locations in each pool (4 samples/treatment/date). Each clump was left in the stream for one week and then removed by pulling up monofilament lines which connected bobbers at the stream's surface with the collapsed plankton nets on the stream bottom. Invertebrates were washed from the vegetation, preserved in 10% formalin, and were later placed into one of twenty—six taxonomic categories. The inver— tebrates were counted and measured, and their weights were calculated using established length—weight regressions (E. Werner, D. Hall, D. Laughlin, unpublished). Beginning on 7/18/79, approximately six fish were taken at ten day intervals from each treatment to determine weights and examine gut contents. The fish were trapped using Plexiglas boxes 50 x 75 x 90 cm fitted on three sides with fibreglass screening and possessing on the fourth side two sliding Plexiglas doors arranged in a V — pat— tern. These doors extended 40 cm in front of each trap, with the open part of the V facing outward. The narrow part of the V was left open 3 cm so that bass could swim into the box. On each date a trap was placed in each pool shortly after dawn and then removed two hours later. Three largemouth were randomly selected from each pool (six/treatment) and were weighed, measured, and sacrificed. The stomach contents were preserved in formalin for later analysis, and 85 total stomachs were analyzed. Prey organisms found in the stomachs were measured and placed into one of the twenty—six taxonomic categories. To examine the breadths 124 of diet of the bass in the different treatments, the taxa were not fur- ther subdivided into prey categories by weight because most of the prey consumed by the bass were small (‘=1 mg dry weight). Such small prey should differ by less than 5 cal in energetic value (Cummins and Wuy— check 1971), and over this size range the time required for bass to pursue, capture, and swallow individuals of a particular prey type should change very little (Anderson, pers. obs.). Repeated seinings of the riffle areas of the streams at var— ious times of day throughout the summer yielded many smallmouth but no largemouth bass. Thus it was assumed that the largemouth actually for— aged for prey in the vegetated pools, not the riffle areas, and that dif— ferences in their growth and foraging behavior across treatments were related to the vegetational differences between pools. On 9/10/79 the stream was treated with rotenone and the bass were removed. The bass were measured to the nearest mm (total length) and preserved in formalin. RESULTS The bass had little impact on the prey communities in the extremely productive stream; prey communities in the different treat- ments did not decline in numbers or total biomass over the course of the summer. Therefore, prey samples taken on different dates were combined to provide an estimate of the nature of the prey community in each habitat type (Table 2). As expected, both the total number 125 TABLE 2. Prey community characteristics in stream areas with differ— ent amounts of vegetation. Mean values + 1 SE are given. LV MV HV Prey Degsity 31,833.3 56,360.0 83,779.4 (#/m ) :3,058.4 :3,304.0 :10,122.2 Prey Biomass 6781.5 13,776.6 16,729.9 (mg dry weight/m3) :691.9 :1,058.2 :1,805.6 Average Individual .21 .24 .20 Prey Biomass (mg dry weight) Number of Taxa 15.8:.6 17.5 i 1.0 16.5 i .7 per Sample Mean Diversity 1.32 1.89 1.07 (H') 126 and biomass of prey were positively related to the amount of vegetation in a treatment. However, within the stream, there was little difference between vegetation types in the average biomass/prey individual or in the total number of taxa present. Largemouth bass grew better in medium vegetation than they did in low and high vegetation (ANOVA, p-=.05, see Table 3). Additionally, these length differences translated into greater average biomasses (wet weights) for bass in medium vegetation. The regression of log weight vs. total length had a greater slope for bass in medium vegetation than in low and high (p<:.10, r2:=.95 for each regression). Thus, a 115 mm bass had a greater biomass in medium than in low or high vegetation. Initial mortality of the fish after stocking was much higher in low and high vegetation habitats. For example, in the first week after stocking the stream, 26 largemouth were discovered dead in the low vegetation habitat, 14 were dead in high vegetation, and only 4 were found dead in the medium vegetation habitat. All dead fish were replaced with similar sized bass, but the number of bass ultimately recovered from the medium vegetation habitat exceeded the numbers in the other two habitats (Table 3 — recall that 200 largemouth were ori— ginally placed in each habitat). Thus, bass grew better in medium vegetation even though their density was probably higher there. The size difference between bass in medium and low vegetation is even greater if one looks at median instead of mean size differences. The median length in medium vegetation was 118 mm; the median in low 127 TABLE 3. Mean final total lengths (: 1 SE, in mm) of largemouth bass at different levels of vegetation. LV MV HV Number of Bass 82 112 85 Recovered Total Length (mm) 115.1 119.0 116.0 .0 . 128 vegetation was only 112 mm. Mean size of largemouth in low vegetation was increased by the presence of a few very large individuals. Eight individuals larger than 134 mm were recovered from low vegetation; only two bass of that size were found in medium. Presumably, those indivi— duals were cannibalistic, a phenomenon which is not uncommon with largemouth bass. There was a tendency for the distribution of bass lengths in low vegetation to depart from normality, although the depar— ture was not significant. 41% of the length values fell between one standard deviation below the mean and the mean, and the distribution was slightly skewed in the direction of large sizes. Four prey types — Simocephalus, a cladoceran; Hyallela azteca, an amphipod; Coenagrionid damselflies; and Baetid mayflies - were the most common prey found in the bass stomachs. In all treatments, these four prey categories accounted for greater than 90% of the total prey consumed by the bass. As the summer progressed, two trends became apparent, both of which were expected. First, as the bass grew, they consumed fewer Simocephalus individuals, which were quite small (c.04 mg dry weight). Second, as the bass became larger, there tended to be more food present in the stomach. However, both trends were apparent in all treatments, and so stomach content data within each treatment were pooled for all dates to make comparisons between the habitat types. A number of pat- terns emerged (Table 4): (1) Largemouth bass in medium vegetation had the fewest prey items/gut; bass in low vegetation had the most. (2) Bass in medium vegetation consumed larger prey items than bass in high and low vegetation. (3) Bass in low vegetation had the greatest total 129 TABLE 4. Prey consumption patterns of largemouth bass in stream habi— tats with different amounts of vegetation. Mean values are given : 1 SE. LV MV HV Number of Stomachs 33 25 27 Examined Number of Prey 38.7 12.7 26.9 Items/Stomach :7.8 :2.8 +4.4 Prey Biomass/Stomach 12.3 6.5 6.5 (mg dry weight) :3.6 :1.8 :1.5 Biomass/Individual .32 .51 .24 Prey Item in Cut :.09 +.05 +.O7 (mg dry weight) — — Number of Taxa/Stomach 4.2 2.9 3.8 :.3 +.4 + .3 130 biomass of prey per gut; bass in medium and high vegetation had about the same total weight of prey. (4) Bass in medium vegetation were most specialized, consuming fewer prey types/fish than their counterparts in high or low vegetation. The average total prey biomass/stomach was highest for bass in the low vegetation habitat because of the presence of very large prey in several of the guts. Bass in low vegetation had a greater ten— dency than bass in medium or high vegetation to include terrestrial prey, which were usually fairly large (='5 mg dry weight). For example, 9 of the bass in low vegetation had a total of 32 ants (mean ant dry mass : 6.6 mg) in their stomachs whereas only 4 bass from medium vegetation had terrestrial prey (4 total ants) in their stomachs and 3 bass from high vegetation consumed terrestrial prey (2 ants, 1 spider). Appar— ently, it was easier for bass in low vegetation to detect terrestrial prey on the surface of the water. Despite the presence of many large terrestrial prey in the stomachs, low vegetation bass still on average consumed smaller prey items than fish in medium vegetation (Table 4) because they ate about 5 times as many Simocephalus individuals (Table 5), which averaged less than .05 mg dry weight, and because Coenagrionids and Baetids, which averaged about .5 mg dry weight, comprised a smaller proportion of their diets (Table 5). If one excludes the 9 bass from low vegeta— tion which had ants in their stomachs, the total prey biomass/stomach for bass in low vegetation drops to 5.8 i 1.2 mg, less than the bio— mass of prey found in the stomachs of bass from medium and high 131 + | H. N.H m.o as Hmauo m.H H N.¢ ¢.mH NHH mchoHmecwoo w. H m.N H.o we weHHmmm N.H H m.oH o.wm mwN HHHHHHN: ANQHN HHHOH NNNO H.m H H.w 0.0m wHN msfimamwooeHm coHumewm> cmH: m. H o.H N.w 0N Hwauo m. H n. ~.m wH mchoHmecooo H.H + o.H o.om so meHHmmm o.H H o.N 0.0H mo HHHHHHH: ANmHm Hmuoe NHNO H.N H m.e m.mm NHH msHmamwooeHm coHHmHmwm> asHewz o. H o.m N.o wHH Hmauo m. H o. o.H HN meHcoHHmmcmoo m. H H.H w.N om meHHmmm H.N H H.0H w.oN Nam meHHmN: AHHHN Hmuoe wNNHV w.N H H.mN m.am HON msHmammooeHm :oHHmewm> sea Amm HHV :mHH Ham coumm HOQEDZ :mumm woum Hauoa mo N :wumm Honfisz mm%H zmnm HHHHQH: me>oH ucouomme Hm mmmn :usoEomeH >3 mmmzu wean ucoummew .coHumuomo> wo Hsom Ho :oHHmNHHHHs H>HHmHom .m mamM=~H) did not occur due to changing patterns of prey selectivity by bass in the different habitat types (bass in medium vegetation were more selective). Model 1 did not come close to matching actual foraging behavior of the bass — bass in low vegetation, which should have been most specialized according to this model, included the most prey types in their diets. Bass in low vegetation did occasion— ally include large numbers of a prey type — substrate dwelling Chirono— mids — which were seldom eaten by bass in medium and high vegetation. However, bass in low vegetation which ate 10 or more Chironomids aver- aged 5.3 : 1.0 prey types per stomach, greater than other bass in low vegetation which had not eaten Chironomids (Table 4). There was no tendency toward microhabitat specialization. Only model 4 fit the data. Diet breadths were narrowest in medium vegetation, and close parallels were observed between degree of selectivity and inferred energy intake. The results of this study are in agreement with the findings of Crowder and Cooper (1982) regarding fish growth rates at different macrophyte densities but differ with regard to diet breadths. Crowder and Cooper observed little change in diet breadths across macrophyte levels, but there was a tendency for their low vegetation fish to be most specialized. Their low vegetation community, however, was not im- poverished — it had high prey densities early in the summer. Further- more, they utilized bluegill sunfish, which are better equipped than bass for specializing on small zooplankton, which they apparently did in the 142 open water, low vegetation habitat. In this study, largemouth bass in medium vegetation were at an advantage compared to bass in low or high vegetation. Their growth rates were greater and they were more selective while foraging, taking fewer but larger prey than fish in low or high vegetation and consuming fewer prey taxa per fish. With regard to the management fish populations, questions related to the appropriate amount of vegetation needed to max- imize production of a certain species can be put in simple logical terms. Basically, too little vegetation will not provide adequate prey numbers for maximal fish growth; too much vegetation will provide ample prey but hinder foraging. Moderate amounts of vegetation appear to be best, but the quantitative definition of what constitutes ”moderate" vegetation will depend on the unique foraging abilities of the fish species of interest. Further, any benefits associated with efficient foraging at a given macrophyte density may be negated if that level of vegetation is also associated with increased predation risk. In such cases, fish may avoid risky areas and forage ”suboptimally” in locations where pre— dation risk is reduced (Mittelbach 1981). 143 LITERATURE CITED Anderson, 0. 1983. Optimal foraging by largemouth bass in structured environments. Ecology 64:000—000. Anderson, 0. 1983. The effects of structural complexity on foraging behavior and diet breadth. Submitted to Oecologia. Charnov, E. L. 1976. Optimal foraging; attack strategy of a mantid. American Naturalist 110:141—151. Crowder, L. B., and W. E. Cooper. 1982. Habitat structural complexity and the interaction between bluegills and their prey. Ecology 63:1802—1813. Cummins, K. W., and J. C. Wuycheck. 1971. Caloric equivalent for investigations in ecological energetics. Internationale Vereinigung fur Theoretische und Angewandte Limnologie, Mitteilungen 18:1—158. Gerking, S. D. 1957. A method of sampling the littoral macrofauna and its application. Ecology 38:219-225. Hackney, P. A. 1975. Bass populations in ponds and community lakes. Pages 131—139 in Henry Clepper, editor. Black bass biology and management. Sport Fishing Institute, Washington, D. C. Heidinger, R. C. 1975. Life history and biology of the largemouth bass. Pages 11—20 in Henry Clepper, editor. Black bass biology and management. Sport Fishing Institute, Washington, D. C. 144 Hruska, V. 1961. An attempt at a direct investigation of the influ— ence of the carp stock on the bottom fauna of two ponds. Internationale Vereiningung fur Theoretische und Angewandte Limnologie, Verhandlungen 14:732—736. Ivlev, V. W. 1961. Experimental ecology of the feeding of fishes. Yale University Press, New Haven, Connecticut, USA. Krebs, J. R. 1978. Optimal foraging: decision rules for predators. Pages 23—63 in J. R. Krebs and N. B. Davies, editors. Behavioural ecology. Sinauer Associates, Sunderland, Massa- chusetts, USA. McCammon, G. W., LaFaunce, D., and C. M. Seeley. 1964. Observations on the food of fingerling largemouth bass in Clear Lake, Lake County, California. California Fish and Game 50:158—169. Mittelbach, G. G. 1981. Foraging efficiency and body size: a study of optimal diet and habitat use by bluegills. Ecology 62: 1370—1386. Stoner, A. W. 1982. The influence of benthic macrophytes on the for— aging behavior of pinfish, Lagodon rhomboides. Journal of Experimental Marine Biology and Ecology 58:271—284. Vince, S., Valiela, I., Backus, N., and J. M. Teal. 1976. Predation by the salt marsh killifish Fundulus heteroclitus in relation to prey size and habitat structure: consequences for prey distribution and abundance. Journal of Experimental Marine Biology and Ecology 23:255-266. 145 Werner, E. E., and D. J. Hall. 1974. Optimal foraging and the size selection of prey by the bluegill sunfish (Lepomis macro- chirus). Ecology 55:1042—1052. Werner, E. E., and G. G. Mittelbach. 1981. Optimal foraging: field tests of diet choice and habitat switching. American Zoologist 21:813-829. Werner, E. E., Mittelbach, G. G., and D. J. Hall. 1981. Foraging profitability and the role of experience in habitat use by the bluegill sunfish. Ecology 62:116—125. APPENDIX 146 Appendix. Invertebrates identified from vegetation samples and largemouth bass stomachs. CRUSTACEA: Copepoda; Ostracoda; Isopoda; Cladocera - Chydorus, Simo— cephalus, Daphnia; Amphipoda - Hyallela. INSECTA: Ephemeroptera — Baetidae, Caenidae, Heptageniidae; Odonata — Aeschnidae - A335, Libel- lulidae, Lestidae, Coenagrionidae; Hemiptera — Corixidae, Notonectidae, Belastomatidae; Trichoptera; Coleoptera — Dytiscidae, Haliplidae; Dip- tera — Chironomidae — Tanypodinae, Tanytarsus; Ants - Formica; Mega- loptera; Homoptera; ANNELIDA: Hirudinea. ”'ili'liiiiiiijiifiltflifliflitiiflifliilif