MICROFILMED 1990 INFORMATION to USERS the Most Advanced Technology Has Been Used to Photo Graph and Reproduce This Manuscript from the Microfilm Master

■ V . • r MICROFILMED 1990 INFORMATION TO USERS The most advanced technology has been used to photo graph and reproduce this manuscript from the microfilm master. UMI films the text directly from the original or copy submitted. Thus, some thesis and dissertation copies are in typewriter face, while others may be from any type of computer printer. The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleedthrough, substandard margins, and improper alignment can adversely affect reproduction. In the unlikely event that the author did not send UMI a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion. Oversize materials (e.g., maps, drawings, charts) are re produced by sectioning the original, beginning at the upper left-hand corner and continuing from left to right in equal sections with small overlaps. Each original is also photographed in one exposure and is included in reduced form at the back of the book. These are also available as one exposure on a standard 35mm slide or as a 17" x 23" black and white photographic print for an additional charge. Photographs included in the original manuscript have been reproduced xerographically in this copy. Higher quality 6" x 9" black and white photographic prints are available for any photographs or illustrations appearing in this copy for an additional charge. Contact UMI directly to order.

University Microtilms International A Bell & Howell Information Company 300 North Z eeb Road. Ann Arbor, Ml 48106-1346 USA 313/761-4700 800/521-0600

Order Number 0014410

Complex interactions in aquatic communities: Trophic-level interactions, ontogenetic niche shifts, and the role of an open-water planktlvore

DeVries, Dennis Robert, Ph.D. The Ohio State Unlvercity, 1989

UMI 300 N. Zeeb Rd. Ann Aibor, MI 48106

COMPLEX INTERACTIONS IN AQUATIC COMMUNITIES:

TROPHIC-LEVEL INTERACTIONS, ONTOGENETIC NICHE SHIFTS,

AND THE ROLE OF AN OPEN-WATER PLANKTIVORE

DISSERTATION

Presented in Partial Fulfillment of the Requirements for

the Degree Doctor of Philosophy in the Graduate

School of the Ohio State University

Dennis R. DeVries, B .S ., M.S.

*****

The Ohio State University

1989

Dissertation Committee Approved by

Roy A. S te in

Peter L. Chesson f Advisor William M, Masters DepanLnent of Zoology

Gary G. M lttelbach To Tammy, Stephen, and Zachary

11 ACKNOWLEDGMENTS

Z would like to thank a great number of people that have had an

Influence on me during my tenure here at Ohio State University. Roy

Stein has influenced me in a great many ways, in particular by providing an excellent role model for me. I thank him for the help that he has given to me through the years and for the influence that he has had on my development as a scientist. He has served as both a mentor and a friend. I also want to thank the other members of my dissertation committee, Drs. Feter L. Chesson, W. Mitch Masters, and

Gary G. Mittelbach, for their helpful comments throughout my dissertation research. Many thanks go to Dr. David A. Culver, who provided much needed help throughout this project, including loaning me an ichthyoplankton net after ours took a permanent dive to tho bottom of Stonellck Lake, providing insight as to how we should sample zooplankton and phytoplankton, and contributing to our efforts to

Interpret the results of phytoplankton counts. A multitude of

interactions, arguments, and discussions with Jeffrey G. Miner, B illie

L. Kerans, Steve P. Klosiewski, Martha E. Mather, David H, Wahl, John

M. Dettmers, and the rest of the "EDIT" gang have improved the quality of my research as well as my understanding of research in ecology and fisheries In general. I want to acknowledge the technical assistance of Ed Lewis, Lori Ryan, Holly Irvin, Craig Mallison, and

Kathy Jones, both in the field and in the lab. In addition to

i l l assisting and organizing trips in the field, conducting behavioral observations during lab experiments, counting and measuring larval fish, and maintaining field equipment, Ed Lewis counted more zooplankton during the last two and one-half years than most people want to count in a lifetim e. Dr. Susan Munch counted all phytoplankton samples included in this work. Funding for this work was provided In part by an Ohio State University Graduate Alumni

Research Award, Rational Science Foundation project BSR-8705518 to Roy

A. Stein, BSR-8715730 to Gary G. Mittelbach, and Federal Aid in Fish

Restoration Project F-57-R to Roy A. Stein, administered through the

Ohio Division of U ildllfe. Additional support, in the form of laboratory space, use of vehicles and boats was provided by .the Ohio

Cooperative Fish and Uildlife Research Unit. I thank my parents for instilling in me an appreciation of the Importance of education. Most of all, I want to express my appreciation for the understanding and patience of my wife, Tammy, and my sons, Stephen and Zachary. Tammy has withstood the rigors of a longer*than-expected graduate program on top of the responsibilities of new found parenthood. Although Stephen and Zachary do not yet realize what these years have meant to us, I hope that someday they will understand the sacrifices they have made.

iv VITA

25 September 1959 ...... Bom • Chicago, Illinois

1979*1982 ...... Research •Assistant, Purdue U n iv ersity , F ort Wayne; Indiana

1982 ...... B.S., Purdue University, F o rt Wayne, Indiana

1982-1985 ...... Graduate Teaching and Research Associate, The Ohio State University, Columbus, Ohio

1985 ...... M.S., The Ohio State University, Columbus, Ohio

1985-present ...... Graduate Teaching and Research Associate, The Ohio State University, Columbus; Ohio

PUBLICATIONS

Brown, K. M. and D. R. DeVries. 1985. Predation and the distribution and abundance of a pulmonate pond snail. Oecologla 66:93-99.

Brown, K. M., D. R. DeVries, and B. K. Leathers. 1985. Causes of llf e - h ls to r y v a r ia tio n in th e freshw ater s n a il Lvmnaea elodes (Say). Malacologla 26:191-200.

DeVries, D. R., R. A. Stein, and .P. L. Chess on. 1989. Sunfish foraging among patches: the patch-departure decision. Animal Behaviour 37:455-464.

FIELDS OF STUDY

Major Field: Aquatic Ecology

v TABLE OF CONTENTS

ACKNOWLEDGMENTS...... i i i

VITA ......

LIST OF TABLES...... v i l i

LIST OF FIGURES...... x

INTRODUCTION ...... 1

CHAPTER

I . MANIPULATING SHAD TO ENHANCE SPORT FISHERIES IN NORTH AMERICA: AN ASSESSMENT...... 4

Introduction ...... 4 Methods ...... 5 R esults ...... 6 D is c u s s io n ...... 9 Conclusions ...... 22 Literature Cited ...... 23

I I . THREADFIN SHAD AS SUPPLEMENTARY FORAGE: CONSEQUENCES FOR YOUNG-OF-YEAR F IS H E S ...... 1 ...... 45

Introduction ...... 45 Methods ...... 47 R esu lts ...... 50 D is c u s s io n ...... 56 Conclusions and Management Recommendations ...... 64 Literature C ited ...... 66

I I I . COMPLEX INTERACTIONS AMONG FISH, ZOOPLANKTON, AND PHYTOPLANKTON AS INFLUENCED BY AN OPEN-WATER PLANKTIVORE...... 88

Introduction ...... 88 M ethods ...... ■...... 91 R e s u l t s ...... 95 D i s c u s s i o n ...... 101 Literature Cited ...... 110

vl IV. HABITAT USE BY BLUEGILL IN LABORATORY POOLS: WHERE IS THE REFUGE WHEN MACROPHYTES ARE SPARSE AND ALTERNATIVE PREY ARE PRESENT?...... 147

Introduction ...... 147 M eth o d s ...... 149 R esu lts ...... 151 D i s c u s s i o n ...... ■...... 153 Literature Cited ...... 157 I SUMMARY...... 171

APPENDIX A. COMPARING THREE ZOOPLANKTON SAMPLERS: A TAXON-SPECIFIC ASSESSMENT...... 176

Introduction ...... 176 M eth o d s ...... 177 Results and Discussion ...... 178 Literature Cited ...... 182

v l i LIST OF TABLES

TABLE

1. Studies of shad introduction that we reviewed. A subset of these studies (I.e., those meeting criteria 1 and 2) were used to generate our primary conclusions. Numbers in the column labeled study number indicate studies meeting criteria 1 and 2 and refer to the lakes on Figure 1 ...... 34

2. Studies of shad removal that wo reviewed. A subset of these studies (I.e., those meeting both criteria) were used to generate our primary co n clu sio n s. Numbers in th e column la b eled study number Indicate studies meeting criteria 1 and 2 and refer to the lakes on Figure 1 ...... 37

3. Food selection (using Chesson's alpha, Chesson 1978, 1983) by larval bluegill collected offshore in Clark Lake before (1987) and after (1988) threadfin shad Introduction. Data are presented as means ± 95% confidence interval. Values greater than neutral selection (i.e ., between 0.09 and 0.13, the reciprocal of the number of prey items In the lake) indicate positive selection; values less than this level indicate avoidance. Bluegill £ 13.0 mm were not collected during 1988 .... 71

4. Food selection (using Chesson's alpha, Chesson 1978, 1983) by larval bluegill collected offshore in Stonellck Lake before (1987) and after (1988) threadfin shad Introduction. Data are presented as means ± 95% confidence interval. Values greater than neutral selection (i.e., between 0.09 and 0.13, the reciprocal of the number of prey items in the lake) indicate positive selection; values less than this level indicate avoidance ...... 72

v l i l Food selection (using Chesson's alpha, Chesson 1978, 1983) by larval threadfin shad in Clark and Stonelick lakes during 1988. Data are presented as means ± 95X confidence Interval. Values greater than neutral selection (I.e., the reciprocal of the number of prey items in the lake, here between 0.09 and 0.13) indicate positive selection; values less than this level indicate avoidance. Dash (-) - fish of the indicated size not present. Threadfin shad <7.0 mm were not present in Clark Lake ......

Food selection (using Chesson's alpha, Chesson 1978, 1983) by larval bluegill collected offshore in Kokoslng Lake during 1987 and 1988. Data are presented as means ± 95X confidence Interval. Values > 0.08*0.13 (the reciprocal of the number of prey items in the lake) indicate positive selection; values less than this indicate avoidance. Bluegill £ 13.0 mm were not collected during 1987. NP - zooplankton taxon not present in the lake at the same time as fish of the indicated size. Sample sizes indicated in parentheses ......

Food selection (using Chesson’s alpha, Chesson 1978, 1983) by larval gizzard shad collected offshore in Kokoslng Lake during 1987 and 1988. Data are presented as means ± 95X confidence interval. Values > 0.10*0.14 (the reciprocal of the number of prey items in the lake) indicate positive selection; values less than this indicate avoidance. Gizzard shad < 10 mm did not contain food 1987. NP - zooplankton taxon not present with fish of the Indicated size. Sample sizes indicated in parentheses ......

Results of the MAN0VA testing density of zooplankton within each of 10 taxa collected by three gears in Stonelick Lake, Ohio. Because gears were compared across all 10 taxa, alpha level required for significance was [0.05/10]-0.005. Legend: N - vertical net haul, S - Schindler*Patalas trap, and T - tube sampler . . LIST OF FIGURES

FIGURE

1. Hap representing the distribution of water bodies where gizzard shad (GS) and threadfin shad (TS) were introduced or removed within the United States. Superimposed on the map Is the natural range of these two species (from Trautman 1981). Lake numbers correspond to numbers used in Tables 1*2 and Figures 2 -3 ...... 60

2. Proportion of studies of shad introductions that caused negative, neutral, positive, or mixed impacts on crapple, largemouth bass, and bluegill. To apportion studies into these categories we used author-generated conclusions and interpretation. Total number of introductions is given in parentheses. Numbers within bars refer to the lakes and introductions referenced in Figure 1 and Table 1 ...... 42

3. Proportion of studies of shad removals that caused negative, neutral, positive, or mixed impacts on crapple, largemouth bass, and bluegill. To apportion studies into these categories we used author-generated conclusions and interpretation. Total number of removals is given In parentheses. Numbers w ith in b ars r e f e r to th e lakes and removals in Figure 1 and Table 2 ...... 44

4. Density (mean ± 1 SE) of larval threadfin shad in Clark and Stonelick lakes, Ohio following adult stocking in April 1988 ...... 75

5. Density (mean ± 1 SE) of larval bluegill in offshore larval tows and inshore seine samples in Clark and Stonelick lakes, Ohio before (1987, dotted line) and after (1988, solid line) threadfin shad were Introduced ...... 77

x 6. Diets of young-of-year bluegill collected inshore in Clark and Stonelick lakes, Ohio during 1987 and 1988. Prey were divided into limnetic zooplankton, litto ral prey, and cyclopold copepods (which can be found in either habitat) (Mittelbach 1981). Results were combined across samples sites and are presented as a percentage of the total biomass of prey found in all fish guts on that date. Sample sizes were 5*15 fish for each date .... 79

7. Density (mean ± 1 SE) and length of young-of-year largemouth bass in Clark and Stonelick lakes, Ohio. Dotted and solid lines represent estimates from before (1987) and after (1988) threadfin shad Introduction, respectively ...... 81

8. Diets of young-of-year largemouth bass collected Inshore in Clark and Stonelick lakes, Ohio during 1987 and 1988. Prey categories Included fish, zooplankton, corlxids, and other littoral invertebrates. Results were combined across sample sites and are presented as a percentage of the total biomass of prey present in all fish guts on that date. Sample sizes were 2-7 fish for each date, usually 5 ...... 83

9. Density (mean ± 1 SE) of macrozooplankton in Clark and Stonelick lakes, Ohio before (1987) and after (1988) threadfin shad introduction ...... 85

10. Density of macrozooplankton taxa in Clark and • Stonelick lakes, Ohio before (1987) and after (1988) threadfin shad. Abbreviations for taxa are ca-calanoid copepods, cy-cyclopoid copepods, na-copepod nauplll, da-Daohnla. bo-Bosmlna. ce-Cerlodaohnla. dl-Dlaphanosoma ...... 87

11. Abundance of larval gizzard shad (fish/m-*, A, D), macrozooplankton (number/1, B, E), and rotifers (number/1, C, F) in Kokoslng Lake, Ohio during 1987 and 1988. Larval fish data are means + 1 SE from two replicate surface tows each week; zooplankton data are means ± 1 SE from three replicate integrated tube samples each week. Note different Y-axis scales among panels ...... 124

x i 12. Mean fecundity of Bosmlna (open squares) and Daohnla (open circles) in Kokoslng Lake during the 1987 (A) and 1988 (B) zooplankton d e c lin e s . Mean fecundity was the product of the proportion of individuals carrying eggs (n - 50) and the mean number of eggs per individual carrying eggs (n - 20). Data are means i 1 SE from three replicate s a m p l e s ...... ‘ ...... 126

13. Final size (mm total length) of gizzard shad (GS) from the 1988 bag experiment in Kokoslng Lake. Solid circles represent data from enclosures (n - 6)'and open circles are from "exclosures" (n - 2 of 6 exclosures) that had fish at the end of the e x p e rim e n t ...... 128

14. Density (number/1) of macrozooplankton (A), size (mm) of macrozooplankton (B), and density (number/1) of rotifers (C) in enclosures (solid circles) and exclosures (open circles) from the 1988 bag experiment in Kokoslng Lake. Data are means + 1 SE from six rep licates ...... 130

15. Size (mm) of six macrozooplankton taxa in enclosures (solid lines) and exclosures (dotted lines) during the 1988 bag experiment In Kokoslng Lake. Data are-means ± 1 SE from six replicates. Note that the horizontal lines.do not represent zero values along the Y -axis ...... 132

16. Mean fecundity of Bosmina (A) and Daphnia (B) in enclosures (solid symbols) and exclosures (open symbols) during the 1988 experiment in Kokoslng Lake. Mean fecundity is the product of the proportion of individuals carrying eggs and the mean number of eggs per individual carrying eggs. Data are means ± 1 SE from six re p lic a te s ...... 134

17. Biovolume (ml/m^) of edible and Inedible phytoplankton in enclosures (solid bars) and exclosures (open bars) at the beginning (7 June 1988) and middle (21 June 1988) of the bag experiment in Kokosing Lake, Data are means ± 1 SE from three replitates ...... 136

18. Density (fish/m^) of larval bluegill (solid diamond) and larval gizzard shad (open triangle) for 1986-1989. Note scale differences for the ordinates among panels ...... 138

x i i 19. Density of larval bluegill (fish/m^, solid diamond) and macrozooplankton (number/1, open diamond) during 1987 (A) and 1988 (B) in Kokoslng Lake. Data are means ± 1 SE from two replicate larval fish tows from three replicate integrated tube samples. Note scale differences for the ordinates between panels ...... 140

20. Mean biomass (mg dry weight of prey/fish) of prey (A) and proportion (by numbers) of macrozooplankton (B) in the diets of IS-mm gizzard shad (n - 5 fish/date) before, during, and after the 1987 (open c ir c le s ) and 1988 (open squares) zooplankton declines in Kokoslng Lake ...... 142

21. Regression of number of larval bluegill captured per minute of larval fish tow (total number of bluegill caught per total sampling time) on the peak larval gizzard shad density (fish/m^) and during 1986-1989 In Kokoslng Lake ...... 144

22. Suggested spring food web in Kokoslng Lake, in co rp o ratin g In te ra c tio n s among la rv a l fish , zooplankton, and phytoplankton generated from our field sampling and enclosure/exclosure experiment . . . 146

23. Proportion of observations during which bluegill were In the shallow (< 20 cm deep) and deep (> 20 cm deep) areas of a 3-m diameter pool. Treatments were small bluegill (BG) with all combinations of an adult largemouth bass (LMB) and young-of-year gizzard shad (GS). Nine replicates were run for each treatm ent...... 162

24. The proportion of observations during which bluegill were in the shallow-center (SC; < 20 cm deep, > 10 cm from edge), shallow -edge (SE; < 20 cm deep, < 10 cm from edge), deep-center (DC; > 20 cm deep, > 10 cm from edge), and deep-edge (DE; > 20 cm deep, < 10 cm from edge) areas of the laboratory pool. Treatments were small bluegill (BG) with all combinations of an adult largemouth bass (LMB) and young-of-year gizzard shad (GS). Nine replicates were run for each treatment ...... 164

25. The proportion of observations during which gizzard shad were in the four areas of the pool. See Figure 24 for definitions ...... 166

x i i i 26. The mean p ro p o rtio n of o b serv atio n s during which bluegill were found in the shallow (< 20 cm deep) end of the pool as a function of the proportion of observations during which the largemouth bass were active in the deep end of the pool (from the 1-min observations). Open circles are mean proportions from experiments with bluegill, gizzard shad, and largemouth bass, and open squares are from experiments with bluegill and largemouth bass. Nine replicates were run for each, treatm ent ...... 168

27. The proportion of observations during which bluegill were schooled as a function of the proportion of observations during which the largemouth bass was active (from the. 1-min observations). Symbols are as in Figure 4. Nine replicates were run for each treatm ent ...... 170

28. Total zooplankton (a), cladocerans (b), copepods (c), and rotifers (d) collected by an integrated tube sampler (dotted line), a Schindler-Patalas trap (dashed line), and a vertical net tow (solid line) during 24 h (N-3 replicate samples) in Stonelick Lake, Ohio on 23-24 June 1987. The bar along the x-axls indicates the period of darkness . . . 186

29. Vertical distribution of rotifers sampled with a Schindler-Patalas trap at seven times during 24 h in Stonelick Lake, O h io ...... 188

x iv INTRODUCTION

In the following four chapters I deal with the impact that open- water planktivores (gizzard shad, Dorosoma cenedlanum and threadfin shad, fi. petenense) can have on reservoir community structure, quantifying effects on pisclvores, other planktivores, zooplankton, and phytoplankton. In addition, in Appendix A, I present the results of a comparison of three zooplankton collection devices, including the tube sampler used in much of the work described in the dissertation.

Here, I describe the content of each chapter so as to provide an orientation to the overall organization of the document.

Chapter I contains the results of a literature review dealing with manipulations (both introductions and removals) of gizzard shad and threadfin shad. Manipulating forage fish populations is a popular practice in fisheries management, Intended to enhance sport fisheries.

This chapter evaluates the success of this management practice. In general, the results of shad manipulations on predators and presumed competitors were not consistent, with the entire range of results from negative to positive observed, depending on the study. A lack of statistical treatment of the data made individual studies difficult to interpret. In this chapter I discuss ways to improve design of studies of forage fish manipulation, suggesting a community-oriented approach to understanding the impact of forage fishes on aquatic community structure.

1 2

One conclusion from the results of the review presented In

Chapter I is that adult fishes form the focus for most studies of forage fish manipulations. However, these manipulations obviously have consequences for other life stages of resident fishes. In

Chapter II I quantify the effects that threadfin shad introduction into two reservoirs has had on young-of-year bluegill fLeoomls macrochlrus) and largemouth bass fMlcrooterus salmoldes). Because the goal of forage fish manipulations Is an improved fishery for adult fishes, the response of young-of-year fishes typically is not studied.

In this chapter I quantify direct and indirect influence of larval threadfin shad on larval and juvenile bluegill and juvenile largemouth bass, as mediated through zooplankton. In two lakes, stocked with adult threadfin shad in April, larval threadfin shad densities peaked in August, minimizing competition with earlier-spawning bluegill. In one of the lakes, zooplankton density crashed subsequent to the larval threadfin shad peak, reducing bluegill recruitment and prey available to young-of-year largemouth bass.

In Chapter III I document the pattern of larval fish and zooplankton abundance in a lake containing gizzard shad, suggesting that larval gizzard shad may contribute to the mid-summer decline observed in most systems. Using the results of a field enclosure/exclosure experiment designed to quantify the response of a zooplankton community to predation by larval gizzard shad, I demonstrate that not only can shad reduce macrozooplankton density, they also can dramatically reduce the density of rotifers and even edible phytoplankton. Simultaneous competition and predation by 3

larval shad on zooplankton leads to dramatic declines in zooplankton

abundance. Ultimately, through decreased consumption caused by

depressed zooplankton densities, bluegill recruitment is reduced.

In addition to the potential for negative effects it is possible

that shad, being the preferred prey for many plscivores, could

actually buffer predation pressure from resident plscivores on other

fish species. To test this hypothesis I conducted a set of

experiments in laboratory pools to assess whether small bluegill would

alter their habitat use (deep vs. shallow areas of the pool) in

response to the presence of shad and/or a largemouth bass predator (in

Chapter IV). Though bluegill did not alter their habitat use In the

presence or absence of gizzard shad, they did reduce their use of deep

areas of the pool when a largemouth bass predator was present relative

to when it was absent. This suggests that shallow areas of reservoirs

may provide refuge from predation when macrophytes are not present (as

would occur due to high turbidity or water*level fluctuations).

Finally, In the Appendix I present the results of a taxon-

speclflc, multivariate comparison of three zooplankton sampling gears*

a vertical net tow, a Schindler*Fatalas trap, and an integrated tube

sampler. This comparison is necessary because, although other

comparisons have been conducted, none has incorporated multivariate * * statistical approach into the design, as is required given the

interdependence of species densities. As a result of this study, I used a tube sampler throughout my work. CHAPTER I

MANIPULATING SHAD TO ENHANCE SPORT FISHERIES IN NORTH AMERICA:

AN ASSESSMENT

Introduction

Approaches used by fisheries biologists to manage aquatic systems for sport fish production have evolved during the past AO-SO years.

Initially, management practices emphasized slngle-specles approaches, such as stocking game fish, limiting catch, and instituting length lim its and seasons. Since the 1950s, however, managers have also manipulated forage fishes to enhance predator growth. These practices have become widespread (see reviews in Ney 1981; Noble 1981; Wydoski

t and Bennett 1981; Noble 1986), contributing to the variety of tools available for managing lakes and reservoirs. Though these manipulations sometimes positively Influence target species, negative results also have been documented, suggesting the need for an assessment of the overall benefits of this practice. Herein we review results of forage fish manipulations to determine whether the overall outcome has been beneficial. Our ultimate goal is to evaluate whether forage fish manipulations tend to improve a fishery or possibly contribute to its decline.

Given that forage fish manipulation may produce responses in sport fishes that are specific to the prey being manipulated, we chose

A 5 to lim it our review to manipulations involving related forage fishes.

Manipulations of a number of taxa are represented in the literature, in clu d in g members o f A th erin id ae, C atostom idae, C lupeidae,

Coregonldae, Cyprinidae, and Percidae (e.g., see references in Ellis

1978, Ney 1981, Noble 1981, Vydoskl and Bennett 1981, Moyle 1986,

Werner 1986, Li and Moyle in press). However, our search revealed that shad (both gizzard shad Dorosoma ceoedianum and threadfin shad D. petenense) were manipulated at least ten times more often than any other group. Because we wanted to draw conclusions as to the overall effect of forage fish manipulations across a wide range of systems, the number of published studies was an important consideration.

Furthermore, both species of shad are Important forage species in much . of North America (Lewis and Helms 1964; Aggus 1973; Carllne et nl.

1986; Johnson et al. 1988). Thus, we lim it our review to studies of manipulations Involving shad.

Methods

We considered an Introduction to be the addition of a shad to a system not containing that species. Introductions Included deliberate stockings, accidental introductions (e.g., shad inadvertently introduced when another species was stocked), or additions due to unknown causes (e.g., unauthorized persons stocking fish, fish moving downstream into a lake). Removals were any measurable reduction of a shad’population, including deliberate removals as well as reductions due to winterkill. 6

All studies were evaluated against objective criteria to determine whether a paper would make a meaningful contribution to our ability to generalize from the review. Though we considered a variety of criteria, only two determined whether a study would be included.

First, we required documentation of the success of an Introduction or removal. Shad abundance estimates from pre- as well as post*treatment years were necessary to permit valid comparisons across years, before and after the manipulation. Without these data, changes in the target species might reflect normal year-to-year variability, rather than changes resulting directly from the manipulation. For shad introductions, this criterion was loosely applied; If shad were sampled from the lake after introduction, we considered the introduction successful. We Included studies with either explicit sampling for shad or the presence of shad In diets of predator fishes.

Similarly, sampling after a removal quantifying a reduction in abundance or complete absence of shad documented removal success. Our second criterion was that the parameter of Interest (e.g., length at age, harvest) had to be quantified both before and after the manipulation. Post-treatment data had to be compared explicitly to pre-treatment data collected in a similar way, to attribute changes to the manipulation. For our purposes, backcalculated growth rates were acceptable. Studies meeting both criteria are discussed below.

R esu lts

Shad manipulations have occurred in a broad band though the midwestem United States, from Oklahoma and Kansas east to Virginia 7

(Fig. 1). Fewer manipulations occurred in the southeastern and southwestern portions of the country, and none In the upper midwest, the Pacific northwest, or the northeast. Almost no manipulations occurred within the native range of the threadfin shad (Fig. 1).

Manipulations outside the native range of threadfin shad could seemingly only be introductions; however, 20Z of threadfin manipulations outside of their range were removals, including studies of reduction due to winterkill and to predator addition. Only one threadfin shad manipulation involved a deliberate removal. In contrast, about 80Z of the gizzard shad manipulations Involved removal rather than addition (Fig. 1). No reports were found where gizzard shad were introduced to lakes outside their native range. Thus, th re a d fin shad were stocked more commonly as a forage f is h whereas g izzard shad were removed more commonly (F ig . 1 ). T his o b serv atio n reflects two facts: 1) threadfin shad have a much smaller native range than gizzard shad and 2) threadfin shad is viewed as a more desirable forage species than gizzard shad due to Its smaller maximum size.

Of 60 papers dealing with shad manipulations, 43 dealt with shad introductions and 17 with shad removals (Tables 1 and 2). Forty-four percent of the introductions and 33X of the removals met both our criteria. Target species examined in studies that met both criteria were almost exclusively sport fishes. Most often studied were largemouth bass Microoterus salmoides (in 16 introductions and 6 removals), white crappie Poraoxis annularis and black crapple £. nigromaculatus (14 introductions and 6 removals), and blueglll Leoomts macrochlrus (12 Introductions and 5 removals). Lesser examined 8 species Included walleye Stlzostedion vltrQUa vltreun (3

Introductions), smallmouth bass Microoterus dolotaieui (3 introductions), green sunfish Lepomis cyanellus (1 introduction), spotted bass Hicrooterus punctatus (1 introduction), and redear sunfish Lepomis microloohus (1 removal). In addition, changes in the zooplankton community due to' shad manipulation were examined in three introductions.

The parameter used most often to determine shad effects, in both

Introductions and removals, was some estimate of growth, such as length at age (13 introductions, 3 removals). Catch, catch per unit effort, and total harvest (9 introductions and 1 removal), condition factor and relative weight (6 Introductions and 1 removal), and diet

(4 introductions) also were evaluated.

Responses to Manipulations

Because largemouth bass, crappie, and blueglil were clearly the most frequent target species, we chose to evaluate effects of shad manipulation on these species. The potential Influence of shad on these three species follows from their functional role in aquatic systems. We expected growth of piscivores, such as largemouth bass and crappie, to increase when shad were added. Blueglil, as a potential competitor with shad, should grow more slowly after shad were added. In turn, removing shad should reduce placivore growth and increase that of its competitors. We categorized the overall outcome of each manipulation as being positive, neutral, negative, or mixed.

For example, when all monitored parameters associated with a target species were positive in response to shad, then the-overall result 9

for that speclea In that atudy was positive. If sone parameters

indicated improvement whereas others did not (i.e ., they were either

neutral or negative), then results were mixed. Because most studies

(i.e ., 803C of introductions and 71% of removals meeting both of our

criteria) did not provide statistical analysis or estimates of variation around measured parameters, we could not objectively

determine whether the response of a target species was significant.

Thus, we had to rely on author-generated conclusions from each manipulation.

As expected, the response of crappie to shad addition was skewed

toward positive effects (seven studies), but two studies documented negative effects, two documented neutral effects, and three revealed mixed responses (Fig. 2). Similarly, response of largemouth bass to shad introduction was skewed toward positive effects (seven studies), but nine studies demonstrated negative, neutral, or mixed responses

(Fig. 2). Blueglil, as a potential competitor with shad, had fewer positive responses to shad addition than did crappie or largemouth bass (one study), but only three studies revealed negative effects

(Fig. 2). Shad removal led to highly variable results for all three species (Fig. 3). All species experienced positive, neutral, and negative effects from shad removal, with each category containing one or two responses, making generalizations Impossible.

D iscussion

Unfortunately, we were not able to generalize about how shad influence sport fishes. Although blueglil appeared more negatively 10 affected by shad than crappie, with the response of largemouth baas being intermediate, we cannot draw such conclusions because the entire range of results occurred for each target species. Why do we not see clear, generallzable results? Below we address two possible reasons for the lack of generalization, after which we discuss several ideas that can Improve our understanding of studies of species manipulations. Finally, we present an alternative approach for studies designed to evaluate the overall impact of species manipulations.

Whv no Generalizations?

Changes in response variables due to shad manipulation were difficult to interpret because most investigators did not analyze their data statistically. He were unable to determine whether

"trends" in pre- versus post-manipulation data represented real differences or Just random variation with no differences in mean values. Thus, we had to rely on the author's conclusions in determining the overall effect of shad on target species, introducing investigator bias as a factor that might be responsible, in part, for the skew in our results. If an Investigator expects shad to enhance predators (or negatively influence competitors), then positive (or negative) results may be accepted, even though response variables do not differ statistically between pre- and post-shad sampling periods.

In fact, it is possible that among the range of responses that were obtained were some studies with significant differences, whereas others had only non-significant "trends". If the significant n differences were all positive or negative for a given species, with the remainder being neutral "trends", then results might be generalizable. Valid conclusions, only can be reached by statistically comparing data from before and after the treatment.

In addition, most studies focused on the effects of shad on adult fish, the life stage of interest to fisheries biologists. Other fish life stages can be affected by these manipulations and their responses w ill Influence the ultimate outcome of the manipulation. Young-of- year fishes feed on zooplankton (e.g., Cramer and Marzolf 1970; fiarger and Kllambi 1980; Van Dan Avyle and Wilson 1980; K east 1980, 1985a;

Lemly and Dimmlck 1982; Mathias and Li 1982) and may compete with zooplanktlvorous shad. Several studies included in-our review

(Prophet 1982, 1985, 1988; Ziebell et al. 1986; seb also DeVries 1989) document how shad can severely depress zooplankton abundance. This effect on the zooplankton may reduce recruitment of such sport fishes as blueglil and crappie, whose limnetic larvae feed on zooplankton for as long as 4*6 weeks (Werner 1967,1969; Keast 1980; Beard 1982). The negative impact of reduced zooplankton abundance on sport fish recruitment would not be detected for several years in studies focused on adult sport fish. Almost no studies reviewed here included data on young-of-year sport fish abundance (but see Kirk and Davies 1985; Kirk et al. 1985); if they had, negative effects of shad may have been more ev id en t.

How Can.we Improve Studies of Forage Fish Manipulations?

Assessment of forage fish manipulations can be improved by careful attention to initial study design and additional data 12 collection. Though we do not argue that the following is the formula for a perfect study, we do believe these features should be considered when contemplating forage fish manipulations.

Describe the site. Basic descriptive information, Including nutrient levels, predator population size structure, etc., should be reported to permit comparison among systems. For example, availability of structure such as macrophytes or turbidity might influence vulnerability of shad to its predators (e.g., Vlnyard and O'Brien

1976; Savino and Stein 1982, 1989a,b). In addition, monitoring sport* fish harvest is critical because Increased angling mortality could obscure long-term population responses. Both abiotic and biotic variables must be monitored through time.

Manipulate only one variable. Often, more than one manipulation occurred simultaneously within a single lake (e.g., introducing brook sllversldes and shad simultaneously; size lim its imposed during a shad manipulation). Clearly, such multiple management strategies confound interpretation. However, given that the goals of fisheries management

(e.g., improved angling) typically differ from that of fisheries research (an understanding of the mechanisms by which management manipulations effect changes in a fishery), manipulation of more than one variable may be necessary. Unfortunately, though we may accomplish management goals with this approach, we learn little of underlying reasons for success or failure. Consequently, our ability to generalize about how particular manipulations influence the fish community has been compromised. Further, because the response of a fishery to simultaneous management strategies will likely involve 13

Interaction among strategies, interpretation of the role of any one

strategy w ill be impossible. When possible, we recommend

manipulating only a single variable.

Document manipulation success. The influence of stocking or removing

forage fishes depends on the extent of the manipulation itself. An

increase from 0 to 5 prey/m^* is certainly a substantive change that

should permit predators to respond. In contrast, how much must a shad population be reduced to reduce competition between shad and sport

fishes? Forage-fish population size must be estimated to allow managers to judge the extent of the response against the magnitude of

the manipulation. In addition, to assure that predators directly benefit from an introduction, pre- and post-manipulation predator diets should be quantified. Only in this fashion can we explicitly

link changes in predator growth with the' Introduction. Finally, in studies of shad removal, the effects of the removal process on the

target species, as well as its effects on the forage species, must be considered. If the target species population is reduced, then improved growth might occur due to reduced Intraspecific, rather than

Interspecific, competition.

Additions vs. removals. Strong, sustained manipulations are required to detect system responses (Walters 1986; Carpenter 1989).

Consequently, removals (partial removals in practice) are far less powerful than additions (Carpenter 1989). Because shad are extremely fecund, a SOX reduction in the adult shad population may only provide a single season during which to collect data, followed by a large forage fish year class, which returns the population to its original 14 size. Though the risk of nonestablishment occurs with shad additions, quantifying this result is far easier than attempting to estimate proportion removed. The difficulty of shad removal is reinforced by our review. Though results from shad additions were skewed (possibly influenced by investigator bias), no trends were apparent from removals. All target species experienced positive, neutral, and negative effects upon shad removal. Consequently, if we are attempting to assess shad effects on ecosystem processes, additions are more likely to succeed than removals.

Include a reference system. Studies of forage-fish manipulations typically Involve data collection across several years before and after the manipulation. To permit unambiguous interpretation of these manipulations, an unmanipulated reference lake (Likens 1985) should be monitored such that changes in the treatment lake can be explicitly ' attributed to the manipulation and riot'to year-to-year environmental variation. Though we recognize that selection and use of a reference lake may be unrealistic in large (> 200 ha) systems, their use is critical to interpretation of study results and may be reasonable in smaller systems; consequently, we discuss their use in this context.

The process by which an appropriate reference lake is selected is critical to ultimate interpretation. Lakes to be manipulated and those to be used as a reference site must be matched as closely as possible in terms of both biotic and abiotic variables (e.g., lake morphometry, shoreline development, abundance and type of vegetation, predator-prey complex, etc.). Toward this end, multivariate techniques, such as principal component analysis or discriminant 15

analysis, could be used to identify appropriate reference lakes. £ven with the use of a reference lake, caution must be exercised when

drawing conclusions from reference versus manipulated comparisons (see

Valters et al. 1988).

All studies we reviewed included comparisons of response variables within the same system from before and after the manipulation. Hurlbert (1984) termed this "temporal

pseudoreplication" in that one can not determine if changes in predator growth rate (as an example) resulted from the manipulation or

from some other unrelated cause, such as weather. Hurlbert (1984) has a dismal view of this study design, arguing that a "situation where a single control area and a single impact area are available" Is intractable statistically. In contrast, Stewart-Oaten et al. (1986) argue that the effects of this type of manipulation can be evaluated statistically using data taken simultaneously and replicated in time before the manipulation, After the manipulation, at both £ontrol

(i.e ., reference site, Likens 1985) and impact sites (termed BACI design). Ve suggest that with growth data, back-calculated measures will suffice, but investigators must monitor a reference lake as well as the manipulated one. Manipulation effects are quantified by comparing the difference in response variables between reference and manipulated sites through time (Stewart-Oaten et al. 1986).

Often year-to-year variation in the response variable within a system may be large, influencing the ability of BACI design to detect changes due to the manipulation. For example, limnetic larval blueglil abundance varied by an order of magnitude before and after 16 threadfin shad introduction In an Ohio lake, but the reduction was not due to the manipulation (DeVries 1989). Consequently, variation across years produced changes that could have incorrectly been attributed to the manipulation If only 2 years of data were used with

BACI design. In such cases several years of data collection, before and after the manipulation, may be required to quantify annual v a ria tio n .

To achieve adequate replication, multiple systems should be manipulated. However, given the extreme cost of monitoring whole systems in response to a manipulation, Carpenter (1989) and Carpenter et al. (1989) have carefully developed the case for unreplicated, paired system, experiments. Using randomized intervention analysis,

Carpenter et al. (1989) argue that tlme-series data, such as that suggested by Stewart-Oaten et al. (1986), can be tested for changes after a manipulation.

Use statistics. As discussed earlier, It is critical that statistics, or at least estimates of variation, be used in any analysis of a forage fish manipulation. Without this crucial component, meaningful conclusions are not possible. Often growth rates from before and after a management manipulation are compared using backcalculated length-at-age data (see Summerfelt and Hall 1987). Because fish growth is a complex response to age and year/environment effects, estimates of variance apportioned to each of these contributing factors are required for managers to assess the value of a particular management manipulation. Consequently, we support an approach in which variance is apportioned into components attributed to age and 17

the nanipulatlon (tfelsberg 1986; tfelsberg and Frle 1987). The

technique baalcally employs a two-way analysis of variance, though

single formulas for this explicit technique do not apply (Welsberg and

Frle 1987). Rather, a more general technique employing general linear

models, such as the GLM procedure In SAS (SAS Institute Inc. 1985)

must be used (tfelsberg 1986; tfelsberg and Frle 1987), Ulth this

technique, standard statistical tests can be applied with the explicit

goal of determining whether growth changes result from the manipulation Itself or from random year-to-year effects.

Do large manipulations In a management context. By definition, adding

forage fishes to water bodies results In a whole-system manipulation.

Because of the tremendous Insight gained from this sort of technique, aquatic ecologists have embraced this approach (see Individual chapters In Carpenter 1988a), though with several caveats. First, manipulations should be quite strong, Increasing or decreasing the abundance of the species of Interest substantially (tfalters 1986;

Carpenter et al. 1989; Kitchell et al. 1988). tfeak manipulations often lead to misleading Interpretations of the role of a species.

Honltoring effort is wasted If the manipulation Is so weak as to provide no effects. In these Ill-fated experiments, pre- and post manipulation data sets provide essentially no insight Into system function (yet require much time and effort).

Our own experience In this regard may prove instructive, tfhlle planning a threadfin shad Introduction to two reservoirs, we extensively debated the merits of high versus low stocking rates.

Initially, low stocking rates (25-50 adults/ha) appeared desirable, 18 for they permitted economical hauling of adults, I.e., stocking rates • * of 250/ha could not be maintained except on an experimental basis. In addition, low stocking rates would minimize competitive effects between shad and young life stages of sport fishes. In coptrast, because hauling mortality and post-stocking predation by largemouth bass could be substantial, high stocking rates might have made sense.

Additionally, If the Introduction failed, at least 1 full year would be lost, Including all of the monitoring effort that year entailed.

This conflict was resolved by stocking at a high rate with the proviso that future work be directed toward back-titrating to a lower, more economical stocking rate. Ue believe this example typifies the conservative nature of scientists and managers. Uhereas much can be gained from a strong manipulation, Initial tendencies are to design manipulations that lie within the historical experience of the

Investigators, setting up experiments to "fine tune" management policy before we know the lim its to the system. Ue commend the advice of

(Walters 1986) who Invokes an adaptive management approach, In which management manipulations (as well as research ones) involve testing the limits of the system. We embrace the perspective that argues for participating In more daring experimental policy with uncertain outcomes (I.e., larger manipulations in this context) than In policies that maintain the status quo (Walters 1986).

Finally, if possible, small-scale, mechanistic experiments should be done concurrently with the large-scale manipulation. In so doing, underlying mechanisms for observed phenomena can be tested and generalizations for management potential should emerge (Carpenter 19 v 1988b, Kitchell et al. 1988). Coupling these experiments with strong manipulations in a management context should yield information that is sufficient for evaluating new management strategies.

An Alternative Approach

Though forage*fish management has moved fishery biology from a simple consideration of a single*species management to an appreciation of both predator and pray, additional system complexity confounds even this approach (Noble 1986). First, as Is obvious, lake communities are composed of many trophic levels, all of which interact to some degree (Carpenter et al, 1985; McQueen et al. 1986; Crowder et al.

1988; Stein et al. 1988). For example, if threadfin shad are introduced into a largemouth bass-bluegill system, threadfin shad should Increase prey resources for largemouth bass. But, because threadfin shad feed on zooplankton, 'as do larval and adult blueglil, then blueglil and threadfin shad may compete, leading to a negative effect of shad on blueglil (DeVries 1989; also suggested by Kirk and

Davies 1985).

System complexity increases, as well, because changes occur in diet and habitat use as fish grow (the ontogenetic niche, Vemer and

Gilliam 1984). For example, blueglil, crappie, and largemouth bass all feed on zooplankton as larvae (tferner 1967; Keast 1980; Beard

1982; Lemly and Dlmmick 1982; Keast 1985a,b), and potentially compete at this life stage. However, as they grow, diets change dramatically,

0 with blueglil diets shifting to littoral invertebrates and zooplankton

(Turner 1955; Seaburg and Moyle 1964; Keast 1978; Beard 1982; Keast 20

1985b), crappie diets changing fron zooplankton to fish (Ellison 1984;

O'Brien et al. 1984; Keast 1985a), and largemouth bass diets changing even more quickly than crappie from zooplankton to fish (Applegate and

Mullan 1967; Hamilton and Powles 1979; Keast and Eadie 1985; Keast

1980, 1985a,b). Thus, shad and crappie can compete at one life stage and interact as predator and prey as crappie grow.

Finally, we must consider how spatial heterogeneity affects community interactions in lakes and reservoirs. Because lakes and reservoirs have somewhat distinct littoral and limnetic habitats, any consideration of interactions among fishes must take into account the habitats occupied. For example, if threadfin shad feed in the limnetic zone while largemouth bass feed in the littoral zone, then largemouth bass might not experience any positive effects due to the increased food abundance because of habitat segregation.

The importance of system complexity in affecting the overall outcome of forage manipulations is further illustrated by the results from Introductions of Hvsls rellcta (recently reviewed by Lasenby et al. 1986). In one of the earliest documented introductions, Hvsls was introduced into Kootenay Lake, British Columbia in 1949 in an effort to improve growth rates of rainbow trout Oncorhvnchus mvkiss. facilitating the switch in diet from feeding on zooplankton to fish

(Sparrow et al. 1964). Although trout growth increased slightly, they did not feed extensively on Hvsls (Northcote 1973). In fact, eutrophlcatlon occurring after the Hvsls Introductions could have been responsible for this growth increase (Northcote 1972, 1973). However, growth of another species, the kokanea salmon Oncorhvnchus nerka. 21 increased substantially, encouraging further use of Hvsls as a forage supplement (Northcote 1970, 1973). Mvsls exhibits dlel migration, influencing the benthos during the day and limnetic zooplankton at night (Lasenby et al. 1986). Because trout rely on limnetic zooplankton for several years, any impact on this resource by Hvsls

(as documented in numerous lakes, reviewed in Lasenby et al. 1986) could negatively affect trout growth. Thus, the very organism intended to improve adult trout growth could lead to a decline in

Juvenile trout growth through competition for zooplankton.

In a forage fish manipulation, Ll et al. (1976) monitored growth of crappie before and after introduction of inland sllverside Henldia bervlllna. They found that piscivorous adult crappie growth increased significantly. However, growth through their first 2 years was significantly reduced, likely due to competition with silversldes for zooplankton. If reduced growth of smaller crappie leads to reduced survival, the overall effect of sllverside introduction may actually be negative. In fact, the ultimate outcome from this manipulation could be some combination of these opposing effects on different sizes of the target species.

Results presented here indicate that several aspects of whole systems must be quantified if we are to assess how forage fish manipulations affect a fish community. Based on our review of manipulations involving gizzard shad and threadfin shad, potential competition and predation, spatial refuges, indirect effects through common predators and prey, and the influence of ontogenetic shifts in habitat and diet of the target and introduced species are all critical 22

to being able to predict the influence of a forage fish nanlpulatlon on a target species.

Conclusions

We feel that a change to a community-oriented approach to fisheries management is necessary if we are to make the most rapid advances in our understanding of how to use forage fishes to Improve a fishery. In future studies better documentation (including statistical treatment of the data) is required if we are to gain an understanding of these manipulations. Only with studies that include well-designed experiments and appropriate analysis can management myths be prevented from perpetuating themselves.

We have the sense that most fishery biologists believe that manipulating forage fishes will produce positive results, Increasing growth of predators with additions and increasing competitor growth with removals. In support of this notion, we note the many ongoing shad manipulations across the country. Our literature review suggests that the documentation underlying these assertions Is lacking.

Consequently, until carefully designed field studies are completed, managers should not assume that shad manipulations w ill generate positive effects. Whether one is adding Mvsls. silversldes, or shad, unanticipated system-wide effects can dramatically undermine even the best lntentioned efforts. 23

Literature Cited

Aggus, L. R. 1973. Food of angler harvested largemouth, spotted, and smallmouth bass in Bull Shoals Reservoir. Proceedings of the Annual Conference of the Southeastern Association of Game and Fish Commissioners 27:501-505.

Anderson, W. M. 1983. Effect of stocking gizzard shad on the fishery in L ittle Dixie Lake, Missouri. Pages 58-77 In D* Bonneau and G. Radonski, editors. Pros and cons of shad. Iowa Conservation Commission, Des Moines.*

Applegate, R. L., and J. W. Mullan. 1967. Food of young largemouth bass, Mlcropterus salmoldes. in a new and old reservoir. Transactions of the American Fisheries Society 96:74-77.

Barger, L. E., and R. V. Kilambi, 1980. Feeding ecology of larval shad, Dorosoma. in Beaver Reservoir, Arkansas. Pages 136-145 In L. A. Fuiman, editor. Proceedings of the Fourth Annual Larval Fish Conference. U.S. Fish and W ildlife Service, Ann Arbor, M ichigan.

Bartholomew, J. P. 1966. The effects of threadfin shad on white crappie growth in Isabella Reservoir, Kern County, California. California Department of Fish and Game, Inland Fisheries Administration Report 66-6.

Beard, T. D. 1982. Population dynamics of young-of-the-year blueglil. Wisconsin Department of Natural Resources, Technical Bulletin 127.

Beers, G. D. 1965. Effects of a threadfin shad introduction upon black crappie and smallmouth buffalo populations in Roosevelt Lake. Master's thesis, University of Arizona, Tuscon.

Beers, G. D., and W. J. McConnell. 1966. Some effects of threadfin shad introduction on black crappie diet and condition. Journal o f the* A rizona Academy o f Science 4:71-74.

Brummett, K. 1983. Effects of gizzard shad Introduction on the fishery of Lake Paho, Missouri. Pages 77-86 An D. Bonneau and G. Radonski, editors. Pros and cons of shad. Iowa Conservation Commission, Des Moines.

Carline, R.F., R.A. Stein, and L.M. Riley. 1986, Effects of size at stocking, season, largemouth bass predation, and forage abundance on survival of tiger muskellunge. American Fisheries Society Special Publication 15:151-167.

Carpenter, S. R. 1988a. Complex interactions in lake communities, Sprlnger-Verlag, New York. 24

Carpenter, S. R. 1988b. Transmission of variance through lake food webs. Pages 119*135 In S. R. Carpenter, editor. Complex interactions in lake communities. Springer-Verlag, New York.

Carpenter, S. R. 1989. Replication and treatment strength In whole* lake experiments. Ecology 70:453-463.

Carpenter, S. R., T. M. Frost, D. Heisey, and T. K. Kratz. 1989. Randomized intervention analysis and the interpretation of whole- ecosystem experiments. Ecology 70:1142-1152.

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

Cramer, J. D., and G. R. Harzolf. 1970. Selective predation on zooplankton by gizzard shad. Transactions of the American Fisheries Society 99:320-332.

Crowder, L. B ., R. W. Drenner,-W. C. K erfoot, D. J . McQueen, E. L. M ills, U. Sommer, C. N. Spencer, and M. J. Vannl. 1988. Food web interactions in lakes. Pages 141-160 in S. R. Carpenter, editor. Complex interactions in lake communities. Springer-Verlag, New York.

DeVries, D. R. 1989. Complex interactions in aquatic communities: trophic-level Interactions, ontogenetic niche shifts, and the role of an open-water planktivore. Doctoral Dissertation, The Ohio State University, Columbus,

Davies, U, D., B. W. Smith, and W. L. Shelton. 1979. Predator-prey relationships in management of small Impoundments. Pages 449-458 In R. H. Stroud and H. Clepper, editors. Predator-prey systems in fisheries management. Sport Fishing Institute, D istrict of Columbia.

DeVries, D. R, 1989, Complex interactions in aquatic communities: trophic-level interactions, ontogenetic niche shifts, and the role of an open-water planktivore. Doctoral Dissertation, The Ohio State University, Columbus.

Dietz, E., and K. C. Jurgens. 1963. An evaluation of selective shad control at Medina Lake, Texas. Inland Fisheries Report 5, Texas Parks and W ildlife Department, Austin.

Domrose, R. J. 1963. Evaluation of threadfin shad introductions. Final Report, Federal Aid Project F-5-R-8, Commission of Game and Inland Fisheries, Richmond, Virginia. 25

Eder, S. 1983. Effects of gizzard ahad introductions on the fisheries In three snail lakes In northwest Hissourl. Pages 100* 107 1 q D. Bonneau and G. Radonski, editors. Pros and cons of shad. Iowa Conservation Commission, Des Hoines.

Ellis, F. 1978. The effect of threadfin shad introductions In small Impoundments. A literature review. Project WC-5-1 Job Completion Report. Georgia Department of Natural Resources, Game and Fish Division, Atlanta.

Ellis, F. 1981. The effect of winterkill of threadfin shad in Lake Jackson. Final Report, Project VC-8 to Georgia Department of Natural Resources, Game and Fish Division, Atlanta.

Ellison, D. G. 1984. Trophic dynamics of a Nebraska black crappie and white crappie population. North American Journal of Fisheries Management 4:355-364.

Fast, A. V., L. H. Bottroff, and R. L. Miller. 1982. Largemouth bass, Mlcropterus salmoldes. and blueglil, Lepomis macrochlrus. growth rates associated with artificial destratification and threadfin shad, Dorosoma petenense. introductions at El Capitan reservoir, California. California Fish and Game 67:4-20.

Fetterolf, Jr., G. M. 1956. Stocking as a management tool in Tennessee reservoirs. Proceedings of the Annual Conference of the Southeastern Association of Game and Fish Commissioners 10:275-285.

Hamilton, J. G., and P. M. Powles. 1979. Feeding habits and growth of young-of-the-year largemouth bass (Mlcropterus salmoldes) near its northern lim it, Nogles Creek, Ontario. Canadian Journal of Zoology 57:1431-1437.

Heldlnger, R. C. 1977. Potential of the threadfin shad as a forage fish in mldwestem power cooling reservoirs. Transactions of the I l l i n o i s Academy o f Science 70:15-25.

Heidinger, R., and F. Imboden. 1974. Reproductive potential of young-of-the-year threadfin shad (Dorosoma petenense) in southern I l l i n o i s la k e s. T ransactions o f the I l l i n o i s Academy o f Science 67:397-401.

Heisey, P. G,, D. Mathur, and N. C. Magnusson. 1980. Accelerated growth of smallmouth bass in a pumped storage system. Transactions of the American Fisheries Society 109:371-377.

Hepworth, D., and S. P. Gloss. 1976. Food habits and age-growth of walleye in Lake Powell, Utah-Arizona, with reference to introduction of threadfin shad. Publication 76-15, Utah State Division of Wildlife Resources, Salt Lake City. 26

Hepworth, D., and T. D. Pettenglll. 1980. Age and growth of largemouth bass and black crappie in Lake Powell, Utah-Arizona, with reference to threadfin shad introduction. Publication 80- 22, Utah State Division of Wildlife Resources, Salt Lake City.

Huish, M. T. 1958a. Studies of gizzard shad reduction at Lake Beulah, Florida. Proceedings of the Annual Conference of the Southeastern Association of Fish and W ildlife Agencies 12:66-70.

Huish, M. T. 1958b. Gizzard shad removal In Deer Island Lake, Florida. Proceedings of the Annual Conference of the Southeastern Association of Fish and W ildlife Agencies 12:312-318.

Hurlbert, S.H. 1984. Pseudoreplication and the design of ecological field experiments. Ecological Monographs 54:187-211.

Jackson, S. W., Jr. 1965. Summary of fishery management activities on lakes Eucha and Spavinaw, Oklahoma, Proceedings of the Annual Conference of the Southeastern Association of Game and Fish Commissioners 19:315-342. » Jahn, L. 1983. Effects of gizzard shad and the fishery in a 250 acre Illinois lake. Pages 30-38 in D. Bonneau and G. Radonski, editors. Pros and cons of shad. Iowa Conservation Commission, Des Moines.

Johnson, B. M., R. A. Stein, and R. F. Carline 1988. Use of a quadrant rotenone technique and bloenergetics modeling to evaluate prey availability to stocked plscivores. Transactions of the American Fisheries Society 117:127-141.

Keast, A. 1978. Feeding interrelations between age-groups of pumpklnseed (Lepomis ylbboaus) and comparisons with blueglil (L. macrochirus). Journal of the Fisheries Research Board of Canada 35:12-27.

Keast, A. 1980. Food and feeding relationships of young fish in the first weeks after the beginning of exogenous feeding in Lake Oplnlcon, Ontario. Environmental Biology of Fishes 5:305-314.

Keast, A. 1985a. The plsclvore feeding guild of fishes in small freshwater ecosystems. Environmental Biology of Fishes 12:119-129.

Keast, A. 1985b. Development of dietary specializations in a summer community of juvenile fishes. Environmental Biology of Fishes 13:211-224.

Keast, A., and J. HcA. Eadie. 1985. Growth depensation in year-0 largemouth bass: the influence of diet. Transactions of the American Fisheries Society 114:204-213. 27

Kinsey, J. B., R. H. Hagy, and G. V. McCamaon. 1957, Progress report on the Mississippi threadfin shad in the Colorado River for 1956. California Department of Fish and Cane, Inland Fisheries Administration Report 57-23.

Kirk, J. P. 1986. Competitive influences of gizzard shad introductions on balanced largemouth bass/bluegill populations. Doctoral Dissertation, Auburn University, Auburn, Alabama.

Kirk, J.P. and V.D. Davies. 1985. Competitive influences of gizzard shad on largemouth bass and blueglil in small impoundments. Proceedings of the Annual Conference of the Southeastern Association of Fish and W ildlife Agencies 39:116-124,

K irk, J . P . , W.D. Davies, and K. Park. 1985. Response o f some members of the fish community to gizzard shad removal from Chambers County Public Fishing Lake, Alabama. North American Journal of Fisheries Management 6:252-255.

Kltchell, J. F., S. M. Bartell, S. R. Carpenter, D. J. Hall, D. J. McQueen, W. B. N e ill, D. Scavia, and E. E. Werner. 1988. Epistemology, experiments and pragmatism. Pages 263-280 in S. R. Carpenter, editor. Complex Interactions in lake communities. Springer-Verlag, New York.

Kline, D. 1983. Gizzard shad - Lake Geode. Pages 38-64 In D. Bonneau and G. Radonski, editors. Pros and cons of shad. Iowa Conservation Commission, Des Moines.

LaFaunce, D. A., J. B. Klmsey, and H. K. Chadwick. 1964. The fishery at Sutherland Reservoir, San Diego County, California. California Fish and Game 50:271-291.

Lasenby, D. C., T. C. Northcote, and M. Furst. 1986, Theory, practice, and effects of Hvsis relicts Introductions to North America and Scandinavian lakes. Canadian Journal of Fisheries and Aquatic Sciences 43:1277-1284.

Lemly, A. D,, and J. F. Dimmlck. 1982. Growth of young-of-the-year and yearling centrarchlda in relation to zooplankton in the litto ral zone of lakes. Copela 1982:305-321.

Lewis, V. M., and D.R. Helms. 1964. Vulnerability of forage organisms to largemouth bass. Transactions of the American Fisheries Society 93:315-318

LI, H.W. and P.B. Hoyle, in press. Management of introduced fishes. Iq C. Kohler, and W. Hubert, editors. Inland Fisheries Management in North America. American Fisheries Society Special Publication, American Fisheries Society, Bethesda, Maryland. 28

LI, H. V., P. B. Hoyle, end R. L. Garrett. 1976. E£fect of the introduction of the H isiiiiippi silveraide (Menldla audens) on the growth of black crappie (Pomoxls niaromaculatus) and white crappie (£. annularis> in Clear Lake, California. Transactions of the American Fisheries Society 105:404-408.

Likens, G. E. 1985. An experimental approach for the study of ecosystems. Journal of Ecology 73:381-396.

Mathias, J. A., and S. Li. 1982. Feeding habits of walleye larvae and Juveniles: comparative1 laboratory and field studies. Transactions of the American Fisheries Society 111:722-735.

•May, B. E., and C. Thompson. 1974. Impact of threadfin shad (Dorosoma petenense) Introduction on food habits of four centrarchids in Lake Powell. Proceedings of the Annual Conference of the Western Association of Game and Fish Commissioners 54: 317-343.

May, B. E., C. Thompson, and S. P. Gloss. 1975. Impact of threadfin shad (Dorosoma petenense) introduction on food habits of four centrarchids. Utah State Division of Wildlife Resources Publication 75-4, Salt Lake City.

McConnell, W. J., and J. H. Gerdes. 1964. Threadfin shad, Dorosoma petenense as food of yearling centrarchids. California Fish and Game 50:170-175.

McGee, M. V,, J. S. Griffith, and R. H. McLean. 1979. Prey selection by sauger in Watts Bar Reservoir, Tennessee, as affected by cold-induced mortality of threadfin shad. Proceedings of the Annual Conference of the Southeastern Association of Fish and W ildlife Agencies 33:404-411.

McHughi J. J. 1980. Northeast Virginia sport fishery studies. Completion Report, Project F-37-R, Virginia Commission of Game and Inland Fisheries, Richmond.

McHugh, J. J. 1983. Evaluation of threadfin shad stocking in two Virginia lakes. Pages 124-133 In D. Bonneau and G. Radonski, editors. Pros and cons of shad. Iowa Conservation Commission, Des Moines.

McLean, R. B ., J . S. G r if f ith , and M. V. McGee, 1985. T hreadfin shad, Dorosoma petenense Gunther, mortality: causes and ecological implications in a South-eastern United States reservoir. Journal of Fish Biology 27:1-12.

McQueen, D. J . , J . R. P o st, and E. L. M ills . 1986. Trophic relationships in freshwater pelagic ecosystems. Canadian Journal of Fisheries and Aquatic Sciences 43:1571-1581. 29

M illar, E. E. 1971. The age and growth of centrarchid fishes in Millerton and Fine Flat reservoirs, California. California Department of Fish and Cane, Inland Fisheries Administration Report 71-4.

Morris, D. J., and B. J. Foilis. 1978. Effects of striped bass predation upon shad in Lake E. V. Spence, Texas. Proceedings of the Annual Conference of the Southeastern Association of Fish and W ildlife Agencies 32:697-702.

Mosher, T. D. 1983. Responses of white crappie and black crappie to threadfin shad introductions in a lake containing gizzard shad. Kansas Fish and Game Comprehensive Planning Option Project FW-9-P-2.

Mosher, T. D. 1984a. An evaluation of threadfin shad Dorosoma petenense introduction in Kansas. Kansas Fish and Game Comprehensive Planning Option Project FW-9-P-2.

Mosher, T. D. 1984b. Responses of white crappie and black crappie to threadfin shad introductions in a lake containing gizzard shad. North American Journal of Fisheries Management 4:365-370.

Moss, J. L,, and W. C, Reeves, 1983. Blueglil responses to threadfin shad elimination in a 130 acre Impoundment. Pages 150-168 In D. Bonneau and G. Radonski, editors. Pros and cons of shad. Iowa Conservation Commission, Des Moines.

Moyle, P.B, 1986. Fish Introductions into North America: patterns and ecological Impact. Pages 27-43 In H.A. Mooney and J.A. Drake, editors. Ecology of Biological Invasions of North America and Hawaii, E cological S tu d ies Number 58, S pringer-V erlag, New York.

Myhr, A. I., Ill, 1971. A study of the white bass, Morone chrvsops (Raflnesque), in Dale Hollow Reservoir, Tennessee, Kentucky. Masters thesis, Tennessee Technological University, Cooksville.

Neuswanger, D. J. 1983, Effects of gizzard shad introductions on the fishery of Lake Paho, Missouri. Pages 87-99 In D. Bonneau and G. Radonski, editors. Pros and cons of shad. Iowa Conservation Commission, Des Moines.

Ney, J. J. 1981. Forage-fish management In the United States. Transactions of the American Fisheries Society 110:725-728.

Noble, R. L. 1981*. Management of forage fishes in impoundments of the southern United States. Transactions of the American Fisheries Society 110:738-750. 30

Noble, R. L. 1986. Predator-prey Interactions in reservoir communities, Pages 137*143 In G. E. Hall and M. J. Van Den Avyle, editors. Reservoir fisheries managenent: strategies for the 80*s. Reservoir Committee, Southern Division, American Fisheries Society, Bethesda, Maryland.

Northcote, T. G. 1970. Advances In management of fish in natural lakes of western North America. Pages 129*139 in N* G. Benson, editor. A century of fisheries in North America. American Fisheries Society Special Publication 7, Bethesda, Maryland.

Northcote, T. G. 1972. Some effects of mysld introduction and nutrient enrichment on an oligotrophic lake and its salmonlds. Internationale Verelnlgung fuer Theoretlsche und Limnologle Verhandlungen 18:1096*1106.

Northcote, T. G. 1973. Some Impact of man on Kootenay Lake and its salmonlds. Great Lakes Fisheries Commission Technical Report 25.

Norwat, D. H. 1978. Evaluation of threadfin shad introduction into Little Dixie Lake. Masters thesis. University of Missouri, Columbia.

O'Brien, V. J., B. Loveless, and D. Wright. 1984. Feeding ecology of young white crappie in a Kansas reservoir. North American Journal of Fisheries Management 4:341-349.

Prophet, C. W. 1982. Zooplankton changes in a Kansas lake 1963*1981. Journal of Freshwater Ecology 1: 569*575.

Prophet, C. W. 1985. Calanold population structure in a Kansas lake after introduction of threadfin shad. Southwestern Naturalist 30:162-163.

Prophet, C. W. 1988. Changes in seasonal population structures of two species of Dlaotomus (Calanolda, Copepoda) subsequent to introductions of threadfin and gizzard shad. The Southwestern Naturalist 33:41*53.

Putnam, T. 1983. Factors leading to renovation of two small Iowa public fishing lakes. Pages 134-141 In D. Bonneau and G. Radonski, editors. Pros and cons of shad. Iowa Conservation Commission, Des Moines.

Range, J. D. 1973. Growth of five species of game fishes before and after Introduction of threadfin shad into Dale Hollow Reservoir. ' Proceedings of the Annual Conference of the Southeastern Association of Game and Fish Commissioners 25:510*518. 31

Rasmussen, J. L., and S. H. Hlchaelson. 1974. Attempts to prevent largemouth bass overharvest in three northwest Missouri lakes. Pages 69*83 In J. Funk, editor. Symposium on overharvest and management of largemouth bass in small impoundments. Special Publication 3, North Central Division of the American Fisheries Society, Bethesda, Maryland.

Rose, E. T. 1957. Results of fish management at Blackhavk Lake. Iowa Conservation Commission Quarterly Reports 9(1):6-11.

Russell, K. 1983. The effect of gizzard shad on the blueglil fishery of Argyle Lake. Pages 175*177 In D. Bonneau and G. Radonski, editors. Pros and cons of shad. Iowa Conservation Commission, Des Moines.

Sandoz, 0. 1956. Changes in the fish population of Lake Murray following the reduction of gizzard shad numbers. Proceedings of the Oklahoma Academy of Science 37:174-181.

SAS Institute Inc. 1985. SAS Users Guide: Statistics. Version 5 ed. SAS Institute Inc., Cary, North Carolina.

Savino, J.F.and R.A. Stein. 1982. Predator-prey interaction between largemouth bass and bluegllls as influenced by simulated, submerged vegetation. Transactions of the American Fisheries Society 111:255-266.

Savino, J.F ., and R.A. Stein. 1989a. Behavioral interactions between fish predators and their prey: effects of plant density. Animal Behaviour 37:311-321.

Savino, J.F ., and R.A. Stain. 19B9b. Behavior of fish predators and their prey: habitat choice between open water and dense vegetation. Environmental Biology of Fishes 24:287-293.

Seaburg, K.G., and J.B, Moyle. 1964. Feeding habits, digestion rates, and growth of soma Minnesota warm water fishes. Transactions of the American Fisheries Society 93:269-285.

Smith, W. A., Jr. 1958. Shad management in reservoirs. Proceedings of the Annual Conference of the Southeastern Association of Game and Fish Commissioners 12:143-147.

Sparrow, R. A. H., P. A. Larkin, and R. A. Rutherglen. 1964. Successful introduction of Hvsis relicta Loven into Kootenay Lake, British Columbia. Journal of the Fisheries Research Board of Canada 21:1325-1327. 32

Stein, R. A., S. T. Threlkeld, C. D. Sandgren, W. G. Sprules, L. Feraaon, E. E. Werner, W. B. Neill, and S. I. Dodaon. 1988. Size-structured Interactions In lake communities. Pages 161-179 In S. R. Carpenter, editor. Complex Interactions In lake communities. Sprlnger-Verlag, New York.

Stewart-Oaten, A., W. W. Murdoch, and K. R. Parker. 1986. Environmental impact assessment: "pseudorepllcatlon” In time? Ecology 67:929-940.

Summerfelt, R. C., and G. E. Hall, 1987. Age and Growth of Fish. Iowa State University Press, Ames.

Trautman, H. 1981. The Fishes of Ohio. The Ohio State University Press, Columbus, Ohio.

Turner, W. R. 1955. Food habits of the blueglll Leoomls macrochlrus macrochirus (Rafinesque), in eighteen Kentucky farm ponds during A p ril and May. T ran sactio n s o f the Kentucky Academy o f Science 16:98-101.

Van Den Avyle, M. J., and J. R. Wilson. 1980. Food habits and feeding selectivity of larval Dorosoma spp. In Center Hill Reservoir. Pages 146-156 In L. A. Fuiman, editor. Proceedings of the Fourth Annual Larval Fish Conference, U.S. Fish and W ildlife Service, Ann Arbor, Michigan.

Vinyard, G. L., and W. J. O'Brien. 1976. Effects of light and turbidity on the reactive distance of blueglll (Lapointe macrochlrus). Journal of the Fisheries Research Board of Canada 33:2845-2849.

VonGeldern, C. E., Jr. 1971. Abundance and distribution of flngerling largemouth bass, MIcropterua salmoldes. as determined by electrofishing, at Lake Naclmlento, California. California Fish and Game 57:228-245.

VonGeldern, C. E., Jr., and D. F. Mitchell. 1975. Largemouth bass and threadfin shad in California. Pages 436-449 In Black bass biology and management. Sport Fishing Institute, D istrict of Columbia,

Walters, C. J. 1986. Adaptive Management of Renewable Resources. M cM lllian P ub lish in g Company, New York.

Walters, C. J., J. S. Collie, and T. Webb. 1988. Experimental designs for estimating transient responses to management disturbances. Canadian Journal of Fisheries and Aquatic Sciences 45:530-538. 33

Welsbarg, S. 1986. A linear model approach to backcalculatlon of fish length. Journal of the Anerican Statistical Association 81:922-929.

Weisberg, S., and R. V. Frie. 1987. Linear models for the growth of fish. Pages 127-143 In R. C. Sumaerfelt and G. B. Hall, editors. Age and growth of fish. Iowa State University Press, Ames.

Werner, E.E. 1986. Species interactions in freshwater fish communities. Pages 344-358 In J* Diamond and T.J. Case, editors. Community Ecology. Harper and Row, New York.

Werner, R. G. 1967. Intralacustrine movements of bluegill fry in Crane Lake, Indiana. Transactions of the American Fisheries Society 96:416-420.

Werner, R. G. 1969. Ecology of limnetic bluegill (Lepomls macrochlrus) fry in Crane Lake, Indiana. American Midland N aturalist 81:164-181.

Werner, E. E., and J. F. Gilliam. 1964. The ontogenetic niche and species interactions in size-structured populations. Annual Review of Ecology and Systematics 15:393-425.

Wyatt, H, N., and H. D. Zeller. 1962. Fish population dynamics following a selective shad kill. Proceedings of the Annual Conference of the Southeastern Association of Game and Fish Commissioners 16:411-418.

Wydoski, R. S., and D. H, Bennett 1981. Forage species in lakes and reservoirs of the western United States. Transactions of the American Fisheries Society 110:764-771.

Zeller, H. D., and H. N. Wyatt. 1967. Selective shad removal in southern reservoirs. Pages 405-414 Jji C.E. Lane, Jr., editor. Reservoir Fishery Resources Symposium. Reservoir Committee of the Southern Division, American Fisheries Society, Athens, Georgia.

Ziebell, C. D., J. C. Tash, and R. L. Barefield. 1986. Impact of threadfln shad on microcrustacean zooplankton in two Arizona lakes. Journal of Freshwater Ecology 3:399-406. 34

Table 1- Studies of shad Introduction that ve reviewed. A subset of

these studies (I.e., those meeting criteria 1 and 2) were used to

generate our primary conclusions. Numbers in the column labeled study

number indicate studies meeting criteria 1 and 2 and refer to the

lakes on Figure 1.

C r ite ria ^ ___ Study Lake 1 2 2 k number R eference Alamo Lake, AZ Y Y N N 9 Z ie b e ll e t a l . 1986 Argyle Lake, IL Y Y N N 31 Russell 1983 Auburn U. ponds, AL NYYN • Davies et al. 1979 Auburn U. ponds, AL Y Y Y Y 56 Kirk 1984 Kirk and Davis 1985 Blackshear Lake, GA YNYN Wyatt and Zeller 1962 Zeller and W yatt 1967 Carvin Cove Res., VA NAC Y Y NA • Doarose 1963 Center H ill Res,, TN YN NN - Fetterolf 1956 Cheatham Lake, TN YNNN • F e tte r o lf 1956 Cherokee Lake, TN YNNN ■ F e tte r o lf 1956 Claytor Lake, VA Y Y Y N 61 Doarose 1963 Conowingo Pond, PA Y Y N Y 66 Helsey et al. 1980 Crosavllle Cty Rs.TN N NYN * F e tte r o lf 1956 Dale Hollow Res., TN Y NYN « F e tte r o lf 1956 Y N NY • Hyhr 1971 YY Y Y 46 Range 1973 Douglas Lake,' TN YNYN - Fetterolf 1956 El Capitan Res,, CA NY NY• Fast et al. 1982 Geary S t. F ish . L k ,, KS YYYN 18 Hosher 1984a G reat F a lls , TN (CS) YN NN - F e tte r o lf 1956 Great Falla, TN (IS) YN YN m F e tte r o lf 1956 Hunnawell Lake, HO NY YN m Brunmett 1983 Isabella Res., CA YYY N 4 Bartholomew 1966 Janesport Coma Lake, HO N YYN Rasmussen and Hlchaelson 1974 Eder 1983 Kerr Res., VA NA Y Y NA • Domrose 1963 Lake B rittle, VA YY Y N 59 HcHugh 1980 HcHugh 1983 Lake Burke, VA YY Y N 60 HcHugh 1980 HcHugh 1983 Table 1- continued

Lake Geode, IA - Y N N 24 K line 1983 Lake Havasu, AZ Y Y Y N 8 Kinsey e t a l . 1957 Lake Nacieniento, CA N N Y N VonGeldern 1971 Lake of Egypt, IL Y Y Y N 34 Heidinger 1977 Lake Paho, MS YY Y N 33 Neusvanger 1983 Lake Pow ell, UT-AZ YY N Y 7a Hepworth and Gloss 1976 YY N N 7b Hepworth and Pettengill 1980 May and Thompson 1974 May e t a l. 1975 Lake Woods, TN YN Y N Fetterolf 1956 Little Dixie Lk, M0 Y Y Y N 30a Norwat 1978 YY Y N 30b Anderson 1983 Little Grassy Lk, IL Y N Y N Heidinger and Imboden 1974 Lyon S t. Fish. Lk., KS Y Y Y N 19a Mosher 1984a (- Lake Reading, KS) xd Y x N 19b Prophet 1962 Prophet 1985 Prophet 1988 McMinnville City Lk, TN NN Y N Fetterolf 1956 Millerton Res., CA N Y N N M iller 1971 VonGeldern and M itch ell 1975 Muddy Run, PA Y Y NY 66 Heisey et al. 1980 Norris Lake, TN Y N Y N Fetterolf 1956 Old Hickory Res., TN Y N N N Fetterolf 1956 Osage S t. Fish. Lk., KS Y Y Y N 20 Mosher 1983 Mosher 1984a Mosher 1984b Patagonia Lake, AZ YY N N 12 Z ieb e ll et al. 1986 Pena Blanca Lake, AZ N N N N McConnell and Gerdes 1964 Philpott Lake, VA YY Y N 62 Domrose 1963 Pine Flat Res., CA NY N N M iller 1971 Pony Express Lk, MO N Y Y N Eder 1983 Roosevelt Lake, AZ YN N N Beers 1965 Beers and McConnell 1966 South Holston R es., TN N N Y N Fetterolf 1956 Spring Lake, IL Y Y N N 32 Jahn 1983 Sutherland Res., CA N Y Y N LaFaunce et al. 1964 36

Table 1- continued

Vatauga Res., TN (GS) Y NNN • Fetterolf 1956 Watauga Res., TN (TS) N N YN • F e tte r o lf 1956 Worth Co Comm Lake, HO NYY N - Rasmussen and Mlchaelson 1974 Eder 1983

PERCENTAGE OF Y'S: 72X 63X 63X 13X 43 papers/58 In tro d u ctio n s a The criteria are as follows; did the study:

1 - document success of the Introduction?

2 - present pro-Introduction data?

3 — document the actual Introduction conditions?

4 — present some estimate of variation or use any statistical

te s ts ?

b The study number refers to the number of the lake on Figure 1. A

dash indicates that the study did not meet criteria 1 and 2 and

was not considered in the results of the review.

c "NA" Indicates that the attempted Introduction was not

successful.

d An "x" indicates that the study examined an Introduction reported

in a previous paper. Table 2- Studies of shad removal that we reviewed. A subset of these

studies (i.e., those meeting both criteria) were used to generate our

primary conclusions. Numbers in the column labeled study number

indicate studies meeting criteria 1 and 2 and refer to the lakes on

Figure 1.

Criteria* Study Lake 1 1 1 k number Reference Blackhawk Lake, IA YN YN - Rose 1957 Blackshear Lake, GA YY Y N 57 Wyatt and Zeller 1962 Zeller and Wyatt 1967 Carpenter's Lake, ICY YNYN Smith 1958 Chambers Co. L k ., AL N Y YN - K irk 1984 K irk e t a l. 1985 Deer Island Lake, FL YN Y N Hulsh 1958b Deer Ridge Comm. Lk.MO Y Y YN 28 Brummett 1983 Dewey Lake, KY Y Y YN 39 Smith 1958 Herrington Lake, KY YN Y N - Smith 1958 Lake Ahquabi, IA N Y YN - Putnam 1983 Lake Beulah, FL YN Y N - Hulsh 1958a Lake E.V. Spence, TX Y N NN - M orris and Follls 1978 Lake Eucha, OK N N YN Jackson 1965 Lake Jackson, GA YY Y Y 58 E llis 1981 Lake Hurray, OK YYYN 15 Sandoz 1956 Lee Co. Lake, AL N Y Y N - Moss and t Reeves 1983 Hedlna Lake, TX YY Y Y 14 D ietz and Jurgens 1963 Prairie Rose Lk, IA NYYN Putnam 1983 Shanty Hollow Lk, KY Y Y YN 36 Smith 1958 Spavinaw Lake, OK N N Y N - Jackson 1965 Watts Bar Res, TN NY Y N - McGee et al. 1979 xc Y X Na - McLean e t a l . 1985

PERCENTAGE OF Y 'S: 65X 62X 85X 10X 17 p apers/20 removals 38

Table 2- continued

a The criteria are as follows; did the study;

1 - document success of the removal?

2 - present pre-removal data?

3 - document the actual* removal conditions?

A - did the study present some estimate of variation or use any

statistical tests?

b The study number refers to the number of the lake on Figure 1. A

dash indicates that the study did not meet criteria 1 and 2 was

not considered in the results of the review.

c An "x" indicates that the study examined a removal reported in a

previous paper. Figure 1. Map representing the distribution of water bodies where gizzard shad (GS) and threadfln shad (TS) were Introduced or removed within the United States. Superimposed on the map is the natural range of these two species (from Trautman 1981). Lake numbers correspond to numbers used in Tables 1*2 and Figures 2-3.

39 AO

TS'

/ I,- nO-gi.

S ' / / / SS \ -’, S : • InnMton ■ \>&y ■ Ramovala o 105 » o s s '- /-/ /•' / • / ' ' *• , OluardShad \ ,iu-ga J Thraadhn Shad

wmooucnoNS 1 MlOarlon Raaarvoir ta 33 LMa Qraaay laka ta 13 Laka E.V. Spanca otH 2 Pna Flal Raaarvoir ta 34 Laka ot Egypl ta 14 Madina Laka 0* 3 Laka Nadamianto u 35 LakaPaho 04 15 Laka Murray gi 4 laabaka Raaarvoir la 40 Oran Fan* Laka ga/ta 10 Spavnaw Laka 0* 5 Sulhartand Raaarvoir ta 41 ChaafiamLaka ta 17 LakaEucha 9* B G Capdan Raaarvoir ta 42 CroaavM Cdy Raaarvoir la 21 PralrtaRoaalaka pt 7 LakaPowad ta 43 lakaWooda ta 22 LakaAhquabl o« 1 LakaHavaau ta 44 McMmwida Cay laka ta 23 Blackhawk Laka 9* 9 Mtmo Lain ta 45 Camar H4I Raaarvoir ta 20 Daar Ridga Community Laka g> 10 Rooaavad Laka ta 40 Oaia HoOow Raaarvow la 30 Shanty Hollow Laka 9» 11 Pana Blanca Laka ta 47 Ok] Hickory Raaarvoir la 37 Carpanlari Laka 9* 12 Paiagomalaka la 4B Horn* Laka ta 30 Harrington Laka gt IB Oaary Suta Flahlng Laka la 49 Chatokaalaka ta 39 OawayLaka 9> I i

f Douglai Laka 19 1 taga 51 ta 50 WaOa Bat Raaanioir tt 20 Otaga Siaia Fialvng Laka ta 52 Souai Hotakan Reaarvoar ta 54 Chambara County Raaarvoir 0t 24 LakaOaoda 9* S3 Watauga Raaarvoir garta 55 Laa County Laka t* 25 Worth County Community laka 0* SO Auburn Ponda taiga ST Bteckstew Likt 9% 20 HunnawadLaka gala 59 lakaBrtda ta SO Liki Jtckton u 27 Jamaaport Community Laka 04 00 LakaBurka ta 05 D M rttlM U ki Qt 29 Pony Eapraaa laka 0* 01 ClaylorLaka ta 30 LKtla Du* Laka gata 02 PhdpoOLaka ta 31 Aiyytalaka 0* 03 Carvin Cova Raaarvoa ta 32 Spring Laka 04 04 KarrRaaarvoa ta 60 Muddy Run Pond 04

Figure 1 Figure 2. Proportion of studies of shad introductions that caused negative, neutral, positive, or nixed impacts on crappie, largemouth bass, and bluegill. To apportion studies into these categories we used author*generated conclusions and interpretation. Total number of introductions Is given in parentheses. Numbers within bars refer to the lakes and Introductions referenced in Figure 1 and Table 1.

41 gr 2 igure F Proportion of studies 0.0 0.2 0.0 0.2 0.2 0.0 0.4 0.4 - (n -16)L a r g e m o u t h b a s s (n = C r a p p i e ( w h i t e a n d b l a c k c o m b i n e d ) Bluegill (n -14) 12 Negative Neutral Mixed ) 35 24 31 24 56 60 59 8 (negative, O v e r a l l d i r e c t i o n o f r e s p o n s e neutral)

56 32 31

30b 30a 19a 30a 62 4b 62

20 I (neutral, positive) MPo i x es d i t i v e 0 7b 30a 46 34 32

30b 19a 31 20 61 7b 4

Figure 3. Proportion of studies of shad removals that caused negative, neutral, positive, or mixed impacts on crappie, largemouth bass, and bluegill. To apportion studies into these categories we used author-generated conclusions and interpretation. Total number of removals is given In parentheses. Numbers within bars refer to the lakes and .removals in Figure 1 and Table 2.

43 gr 3 igure F Proportion of studies 0.2 0.0 0.4 0.2 0.2 0.0 0.0 N B l u e g i l l ( n = 5 ) C r a p p i e ( w h i t e a n d b l a c k c o m b i n e d ) ( n = 6 ) L a r g e m o u t h b a s s ( n « 6 ) e g a t i v e M i x e d 39 39 57 57 39 58 (negative, neutral) O v e r a l l d i r e c t i o n o f r e s p o n s e Neutral 28 15 15 57 15 positive) Mixed o i i e Mixed3ositive 28 14 14 <4 # positive) 58 CHAPTER I I

THREADFIN SHAD AS SUPPLEMENTARY FORAGE:

CONSEQUENCES FOR YOUNG-OF-YEAR FISHES

Introduction

Forage fish manipulation has become a popular technique for

Increasing production of sport fishes (Ney 1981; Noble 1981; Uydoskl and Bennett 1981; Noble 1986; DeVries and Stein, in press).

Manipulations Involve both prey Introduction to increase the. food base and removal of planktivores to reduce competition. Numerous taxa have been used in these manipulations, including members of Atherinldae,

Catostomldae, Clupeidae, Coregonidae, Cyprinidae, and Percldae (see references In DeVries and Stein, in press). These manipulations, based primarily on the predator-prey Interaction between introduced prey and adults of the target species, seek to enhance a target sport species. However, enhancement is not always the result. In a recent review, DeVries and Stein (in press) documented that manipulations involving two clupeid species (gizzard shad, Dorosoina ceoedianum and threadfin shad, JQ. petenense) and four target species (crapple,

PvniVXlg annularis and £. nleromaculatus. largemouth bass, Hlcropterus salmoldes. and blueglll, Lopomls raacrochirus). resulted in the target species experiencing positive, neutral, and negative effects. Thus,

Increasing forage for predators through introductions or reducing

45 46 presumed competitors through removals may not produce the desired result (i.e., increased adult growth) and other, unexpected consequences may derive £rom prey manipulation.

Because large sport fish interest anglers, the adult stage of these species has been most studied in relation to prey manipulations.

Other life stages (larvae, Juveniles) of prey and predator have been largely ignored (but see Kirk 1984; Kirk and Davies 1985). Fish, however, often change their diets and/or habitats as they grow (i.e., the ontogenetic niche, Verner and Gilliam 1984). Thus, species that interact as predator and prey as adults may compete at earlier life stages (e.g., largemouth bass and bluegill, Gilliam 1982; European perch, ferca flUYlfltllla, and roach, Rutllua rutllua. Persson 1988).

Hany fishes have limnetic, zooplanktivorous larvae (e.g., Werner 1967;

Barger and Kilambl 1980; Keast 1980; Beard 1982; H ills et al. 1987), creating the situation where introduced forage fishes may compete with their subsequent predators during this life stage. Because year-class strength is set early in life (Cushing 1975; Lasker 1975; Ware 1980;

Hills and Forney 1988), Interactions during this critical period may have major effects on recruitment and subsequent harvest. Clearly, interactions at all life stages may determine the success of a forage fish manipulation, with the overall outcome dependent on some combination of the potential positive (improved forage availability to adults) and negative (competition among larvae and juveniles) e f f e c ts .

Here, we quantify how larval bluegllls respond to the introduction of threadfin shad in two Ohio lakes, concentrating on the 47

potential for pegatlva effecta due to interactions among larvae.

Because first-feeding bluegill are limnetic for several weeks, then

stove into the litto ral zone (between 10-25 mm SL; Werner 1967; Storck

1978; Werner and Hall 1988), we also determine how the offshore

interaction with threadfin shad might manifest itself on littoral

blueglll, and perhaps largemouth bass recruitment.

Methods

Study Lakes

Clark and Stonellck lakes are shallow, turbid reservoirs in

southwestern Ohio. Clark Lake (40 ha, Clark County, Ohio) has 4.5 km of shoreline, a maximum depth of 2 m, and Secchi depths 25 to 75 cm.

Stonellck Lake (69 ha, Clermont County, Ohio) has 16 km of shoreline, a maximum depth o f 4 m, and Secchi depths 22 to 119 cm. Submersed vegetation was never abundant in either lake and emergent vegetation occupied about 25X of the shoreline of Clark Lake (Xxohfl) and 70X of

the shoreline of Stonellck Lake (both Tvoha spp. and water willow,

Justlcla amerlcana). Neither lake stratified and dissolved oxygen levels fell below 3 mg/1 only within 0.5 m of the bottom In Clark Lake and below 2 m In Stonellck Lake. Fish communities In both lakes consisted primarily of largemouth bass, white crappie, bluegill, longear sunflsh (Leoomis menalotisl, brown bullhead (Ictlobus nebulosus), and carp (Cyprinus carplo).

In collaboration with personnel from the Departments of Natural

Resources of both Ohio and Kentucky, adult threadfin shad from

Herrington Lake (Mercer County, Kentucky) were stocked at densities of 48

48 fish/ha in Clack Laka and 59 fish/ha in Sconelick Lake on 14-18

April 1988. Transport mortality was visually estimated at < IX during

both stockings. Post-stocking predatory mortality was monitored for 3

days; one largemouth bass of 22 collected in Clark Lake and one

largemouth bass of 103 collected in Stonellck Lake had eaten one

threadfin shad each. Additional details concerning collection,

transport, and stocking procedures can be found in Buynak et al.

(1989) and Austin and Hurley (1989).

Sampling Methods

We sampled during April through October in both pre- (1987) and

post-shad (1988) years. Larval fish were collected offshore once per

week in two replicate surface tows with a 3/4-m diameter

lchthyoplankton net (mesh size - 500 micron) towed in the limnetic

zone at £1.5 m*a~^. A flow meter mounted in the mouth of the net

allowed calculation of the 'total volume of water filtered. Juveniles

were collected biweekly at four (Clark Lake) or five (Stonellck Lake)

sites in the littoral zone of each lake using a 9-m bag seine (4-mm

mesh). Both larvae and juveniles were preserved and returned to the

laboratory where they were identified, measured (nearest mm, up to 50 per species), and their diets (up to 1 0 per species per date) quantified. Diets were quantified from weekly samples for larval fish and from monthly samples for fishes collected in the littoral. We

identified prey to the lowest taxonomic category possible (> 80X to genus) and measured them (nearest 0.1 mm). Prey lengths were converted to biomass using taxon-specific length-dry weight regressions. 49

Integrated zooplenkton samples (two tube hauls per sample, three replicate samples per date) were collected using a 2 -m tube sampler

(7.30 cm Inside dla, mesh size - 54 micron) at the same time that larval fish were collected. Samples were preserved in 5X sucrose formalin (Haney and Hall 1973). Zooplankton taxa with < 200 individuals per sample were counted in their entirety; subsamples were taken for more abundant taxa until at least 2 0 0 individuals were counted. At least 20 individuals of each taxon in a sample were measured (total body length excluding spines, helmets, caudal rami) using an ocular micrometer.

We evaluated the degree of resource overlap (and thus potential for competition) between larval bluegill and larval threadfin shad when present in the limnetic, using Schoener's (1970) index based on the average proportion that each prey taxon contributed to total biomass in larval fish diets (i.e., the proportion was calculated within a fish and averaged across fish within a date; Wallace 1981).

The formula for this index is

(X - 1 - 0.5 (£ .|rxl - ryl|), 1-1 3

where rx^ - the proportion of prey taxon i in the diet of species x and n - the number of prey categories. This index ranges from 0 to 1, with values near zero indicating little overlap, and overlap increasing as the index approaches 1, Because larval fish and 50 zooplankton communities changed through time, we treated overlap measures calculated for each sample date as replicates.

To evaluate prey selection by fishes, we compared larval fish diets with zooplankton samples using Chesson's alpha (Chesson 1978,

1983), treating individual fish within a size range as replicates.

The formula for this index is

^ _ —CiZEi— * 2 - i where p^ - the proportion of prey Item i in the environment (I.e., the lake) and r^ - the proportion of prey item i in the fish's diet.

Calculation of this index was based on the number of prey in individual fish guts and in the zooplankton community.

R esu lts

Larval Fish Abundance

Although larval threadfin shad ware first collected offshore in

Clark and Stonellck lakes in mid-May, densities remained low (< 0.20 flsh*m*3) until peaks of 8 . 8 fish’m*^ in Clark Lake and 2.3 fish'm*^ in Stonellck Lake occurred in August 1988 (Figure 4). Larvae were collected through September 1988 in both study lakes (Figure 4).

Larval bluegill collected offshore in Clark Lake peaked at

3.6 fish'm’3 in 1987, whereas in 1986 several peaks occurred, but density was always £ 0.2 fish*m*3 (Figure 5A). Abundance differed between years (repeated-measures ANOVA, F^^ “ 130.24, p - 0.008), although the year x time interaction term was significant (repeated- 51 measures ANOVA, - 80.23, p - 0.007), due Co protracted spawning during 1988. As would be expected from abundance of larval blueglll between years, more juvenile bluegill were caught in litto ral seine hauls during 1987 than in 1988 (Figure SB; repeated-measures ANOVA,

F^ ' 6 - 50.42, p - 0.0004). As with offshore larvae, the protracted spawn during 1988 led to a significant year x tine interaction term for Inshore blueglll abundance (repeated-measures ANOVA, F^q,60 “

16.69, p - 0.0001). In spite of the overall between-year differences in abundance, no difference in littoral bluegill densities was detectable after August (ANOVA, F ^ q “ 0.06, p - 0.81).

In Stonellck Lake, larval blueglll continuously recruited to the open water during May through,August in both years (Figure 5C).

Because bluegill spawned at different times between years, the year x time interaction term was significant (repeated-measures ANOVA, Fj o ^ o

- 29.02, p - 0.006); however, overall offshore larval blueglll abundances did not vary between years (repeated-measures ANOVA, F^(2 “

0.01, p - 0.92). This continual recruitment of larval bluegill abundance to the limnetic led to increasing catches in the littoral zone in the pre-threadfln shad year. In contrast, just a single peak occurred in mid-July after threadfin shad introduction (Figure

SD), suggesting some influence of threadfin shad (repeated-measures

ANOVA, year effect, Fj^g - 7.00, p - 0.03; year x time interaction term, Fj^go - 4.76, p - 0.006). Urral Flah Diets Larval bluegill collected offshore consumed only zooplankton in both lakes during 1987 and 1988. Preference for various zooplankton 52

taxa was determined by comparing alpha values (± 95Z Cl) for a taxon

with the alpha value expected when prey were eaten in proportion to

their availability (i.e., the reciprocal of the number of prey types

in the environment). Alpha values exceeding the reciprocal indicate

positive selection for a prey taxon.. Because soft-bodied rotifers

were digested quickly and rarely identified in larval fish guts, we

only included hard-bodied rotifer genera (Brachlonus and Keratella) in

these analyses. Four taxa (Chvdorus. Molna. Slmocephalus. and

ostracoda) were sampled but were never eaten; they are not discussed

below.

In Clark Lake during 1987, larval blueglll began feeding at 4-7

mm TL by selecting copepod nauplii and Dlaphanosoma (Table 3). As

blueglll grew, selection for copepod nauplii decreased while

increasing for DlaphanoBoma (Table 3). Among other taxa, calanold

copepods and Ceriodaphnla were avoided by all size classes; Booming.

Daphnla. cyclopoid copepods, and rotifers either were eaten in

proportion to their availability or avoided. In 1988, all sizes of

larval blueglll selected Dlaphanosoma (Table 3). Other prey taxa

either were eaten in proportion to their availability or avoided

(Table 3 ).

During 1987 in Stonellck Lake, first-feeding blueglll selected copepod nauplii, then shifted to cyclopoid copepods and Dlaphanosoma as bluegill grew (Table 4). Prey selection patterns were similar in

the post-shad year, with copepod nauplii selected by first-feeding larvae, and Dlaphanosoma and calanold copepods selected by larger 53

larvae (Table 4). Ocher prey taxa were either eaten in proportion to

their availability or avoided (Table 4).

Larval threadfin shad in both lakes fed entirely on limnetic

zooplankton, as did limnetic larval bluegill. In Clark Lake,

threadfin shad selected Diaohanosoma. calanold and cyclopoid copepods, whereas in Stonellck Lake first-feeding threadfin shad selected

copepod nauplii, then Diaohanosoma and calanold copepods (Table 5).

■ Using Schoenor's overlap index (Schoener 1970), which varies from

0 (no overlap) to 1 (complete overlap), diet overlap between larval

threadfin shad and blueglll collected offshore was 0.37 ± 0.20 (mean ±

95X Cl, n - 8 ) In Clark Lake and 0.43 ± 0.08 (mean ± 95X Cl, n-13) in

Stonellck Lake. Diet overlap values exceeded 0.50 on two of eight dates in Clark Lake and on five of thirteen dates in Stonellck Lake.

Bluegill collected by seining nearshore were feeding predominantly on litto ral prey, although some limnetic prey were also consumed. In Clark Lake, Inshore blueglll fed on prey from both litto ral and limnetic habitats during both pro* and poat-shad years

(F igures 6A,B). On five of six dates in 1987 and two of four dates In

1988, the majority of their diets came from the littoral zone (e.g., chlronomld larvae and pupae), with the remainder primarily cyclopoid copepods. However, limnetic zooplankton remained an important diet component (> 20X by dry weight) on two of six dates in 1987 and one of four dates in 1988 (Figures 6A,B).

In Stonellck Lake, Inshore bluegill diets differed dramatically between 1987 and 1988 (Figures 6C,D). In 1987, blueglll fed as in

Clark Lake, consuming primarily litto ral prey on four of six dates, 54

and limnetic prey on the other two dates (Figure 6C). In 1988,

limnetic prey never contributed more than 2X of bluegill diets (never

more than 10X if cyclopoid copepods were considered exclusively

open-water) (Figure 6D). Across all dates during 1988, > 90X of

bluegill diet originated in the littoral zona,

Largemouth Bass Abundance. Growth, and Diet

During both years in both lakes young-of-year largemouth bass

were collected only in the littoral zone. In Clark Lake, young-of-

year largemouth bass abundance differed marginally between years

(repeated-measures ANOVA, year effect F^g - 5.42, p - 0.06), although

the year x time interaction term was significant (repeated-measures

ANOVA F^i'gg - 9.34, p - 0.006), as a result of peak abundance being a

4 month later in 1988 (Figure 7A). As a consequence of later spawning

in 1988, young-of-year largemouth bass were smaller in June-July 1988

than in 1987, but were similar-sized by the beginning of August

(Figure 7B), with no differences across years (two-way ANOVA, F^^gg -

0,84, p - 0.36). In contrast to 1987, in which young-of-year

largemouth bass in Clark Lake fed on fish In the litto ral zone when

available, young-of-year bass did not feed on fish during 1988 (c .f,,

Figures 5 and 8 ). However, prey biomass in young-of-year largemouth

bqss stomachs, measured as dry weight of prey per gram of fish, did

not differ across pre- and post-shad years (two-way ANOVA, F ^ ^ -

2.06, p - 0.16),.and led to similar growth between years (Figure 7B).

In Stonellck Lake, young-of-year largemouth bass were -20 times

more abundant at their peak in 1988 than in 1987 (repeated-measures

ANOVA: overall year effect, Fj^g - 9.67, p - 0.01); however, by late 55

summer abundance was s im ila r between yeaca (F igure 7C). In a d d itio n ,

young-of-year largemouth bass in Stonelick Lake grew more slowly in

1988 than in 1987 (Figure 7D; two-way ANOVA, year x time interaction,

^4,380 **9*29, p •* 0.0001; year effect, Fj^gg — 74.07, p - 0.0001).

Young-of-year largemouth bass primarily ate fish during 1987, but only

ate fish on one date during 1988 (Figures 8C,D). Although larval

th re a d fin shad were p re se n t during much o f th e summer in 1988 (see

Figure 4B), young-of-year largemouth bass never consumed them. In

addition, biomass of prey in bass stomachs (measured as dry weight of

prey per g of fish) was lower in 1988 than in 1987 (two-way ANOVA,

Fl , 35 - 1.4.01, p - 0.0007).

Zooplankton Abundance

Total zooplankton abundance in Clark Lake fluctuated during both

1987 and 1988 (Figures 9A,B), but did not differ across years

(repeated-measures ANOVA, year effect F^^ - 2.00, p - 0.23; year x

time interaction Fj^g^ - 8.22, p - 0.004). Further, species

composition did not appear' to differ between years (Figure 10A,B).

Density did not decline during late-August and early-September 1988,

after larval threadfin shad abundance peaked (Figure 9B). If

threadfin shad were to reduce zoopiankton abundance, this reduction

should have occurred after mid-August (see Figure 4A).

In Stonellck Lake, as in Clark Lake, zooplankton density

fluctuated during both 1987 and 1988 (Figures 9C,D), but was lower during Nay through July 1987 than during this period in 1988

(repeated-measures ANOVA, year x time interaction Fj^yg - 29.49, p -

0.0001; year effect F^^ - 61.00, p - 0.002). As in Clark Lake, zooplankton species composition did not appear to differ between years

(Figure 10C,D). During August 1988, zooplankton density declined,

with all taxa except copepod nauplii essentially eliminated (Figure

10D). To determine whether this decline could be explained in part by

decreased zooplankton birth rates (i.e., reduced egg production, as an

indicator of reduced resource availability for zooplankton), as

opposed to increased death rates (i.e., due to increased predation) we

regressed mean fecundity (the product of the proportion of individuals

carrying eggs and the mean number of eggs per individual carrying

eggs) of Daohnla. Bosmlna. and Cerlodaohnla (the only abundant

cladoceran taxa present in August) on time during August (i.e., the

time of the zooplankton decline). Mean cladoceran fecundity did not

change during the zooplankton decline (p > 0.16 for all three taxa).

Similarly, during the threadfin shad peak in Clark Lake (mid-August

through mid-September), mean fecundity of Bosmlna and Cerlodaphnia did not change (p > 0.31), while mean fecundity of Daohnla increased (p -

0 .0 3 ).

Discussion

Threadfin shad- Blueglll Interactions

Spawning by threadfin shad peaked during August in both Clark and

Stonellck lakes. This is in contrast to spring spawning documented for it and its congener, gizzard shad in other systems (Baglln and

Kllambl 1968; Mayhew 1977; Barger and Kllambl 1980; Van Den Avyle and

Uilson 1980; Downey and Toetz 1983; Tlsa et al. 1985; W illis 1987).

In Clark lake larval bluegill occurred in the limnetic zone before, 57

but not during or after peak larval threadfln shad abundance, thus

reducing the potential for competition between these species. In

Stonelick Lake, however, larval blueglll spawned during May through

August, with peak abundances before and during the peak larval

threadfln shad abundance. Thus, competition between larval shad and

blueglll was possible while each species occupied the limnetic zone in

Stonelick Lake.

The late summer peak in larval threadfln shad abundance observed

in this study could be due to several factors; e.g., delayed spawning

could have resulted from stocking stress, drought conditions which reduced water levels during 1988, or spawning of young-of-year

threadfln shad that matured by late summer (as documented by Heidinger and Imboden 1974). Whether late spawning is a predictable feature of

threadfln shad introduction la critical to determining its overall value as a forage fish. If spawning time of introduced forage fishes can be altered (e.g., by stocking such that young-of-year fishes spawn during late summer), then negative effects resulting from larval competition can be minimized. Whether a late spawn by introduced threadfln shad is a predictable phenomenon requires additional re sea rc h .

Because both threadfln shad and blueglll rely on limnetic zooplankton during the critical larval stage (e.g., Werner 1967;

Mayhew 1977; Mallln et al, 1985), competition due to reduced zooplankton abundance during this time may dramatically Influence these larvae. In Clark Lake, zooplankton density did not change during peak threadfln shad abundance. Thus, while offshore larval 58

blueglll abundance in Clark Lake declined overall during 1988 relative

to 1987, it Is unlikely that competition from threadfln shad (either

larvae or adults) was responsible. Further, larval threadfln shad

abundances did not peak in Clark Lake until after blueglll had moved

to the littoral zone. Other factors that may be responsible for the

reduction In larval blueglll* densities between 1987 and 1988 include

possible predation by adult threadfln shad; however, fish never » occurred in stomachs of threadfln shad from either lake during 1988

(N-41 adult shad stomachs examined). Alternatively, because drought

conditions in 1988 reduced lake levels in Clark Lake by "1 o during

blueglll nesting, blueglll spawning may have been disrupted (see

recent review in Ploskey 1986), reducing production of'larvae during

1988, ~

Though larval threadfln shad peaked late in summer, they co

occurred with limnetic larval blueglll during Hay through September in

Clark Lake. Both species ate only limnetic zooplankton, and diet

overlap v alu es ty p ic a lly remained below 0.50. Once b lu e g lll moved

Inshore, the potential for competition with limnetic threadfln shad

fell greatly. Diets of blueglll collected in the littoral zone did

not change between 1987 and 1988 in Clark Lake; all prey types (i.e .,

litto ral, limnetic, and cyclopold copepods) were represented. Though

collected inshore, blueglll apparently moved far enough offshore to

consume limnetic prey; yet, inclusion of littoral prey in their diets

reduced overlap with threadfln shad.

Very different responses to threadfln shad introduction occurred

in Stonelick Lake. Zooplankton declined precipitously during August, 59

Just as larval threadfln shad and blueglll peaked. All

macrozooplankton vere eliminated, with only cyclopold copepods and

copepod nauplll at densities > 1.5 organisms*11te r b y late August.

Because birth rates did not decline across cladoceran taxa, falling

abundance was not due to reduced zooplankton reproduction (I.e.,

resources becoming limiting to the zooplankton) but, rather, due to

Increased death rates. Of the planktlvores In open water (e.g.,

blueglll, threadfln shad, and white crapple, Pomoxls annularis1). only

larval blueglll and threadfln shad Increased during August,

Implicating these two species in the decline. Because larval blueglll

were absent after 9 August and zooplankton continued to decline

through August, threadfln shad, present through 7 September, were the

most likely factor responsible.

Differential zooplankton responses In Clark and Stonelick lakes

could have derived from two obvious system differences. First, larval blueglll abundance peaked simultaneously with larval threadfln shad in

Stonelick Lake but not Clark Lake, However, as discussed above, blueglll probably did not cause the zooplankton decline in Stonelick

Lake. Second, turbidity was higher In Clark than in Stonelick lake;

Secchl depths during August ranged from 27 to 41 cm In Clark Lake and

80 to 81 cm in Stonelick Lake. Increased turbidity (within this range) can cause larval fish to concentrate In surface waters

(Swenson and Matson 1976, Matthews 1984) and, given that we sampled larval fish from the top 0.75 m of the water column In both lakes, larval threadfln shad abundance may be overestimated in Clark Lake.

In addition, larval densities in Clark Lake may be overestimated 60 because larval fish are more susceptible to collection at decreased light levels (i.e., at night or at Increased turbidity). In support ■ of this, night sampling (using a Tucker trawl) by Ohio Department of

Natural Resources personnel produced similar peak threadfln shad densities between lakes (Austin and Hurley 1989). Finally, it is unlikely that larger planktivores (i.e., adult threadfln shad and white crapple) were responsible for the observed differences, given that sampling conducted in collaboration with Ohio Department of

Natural Resources did not suggest any differences between lakes.

Zooplankton declines as in Stonelick Lake have been documented following threadfln or gizzard shad introductions (VonGeldem and

Mitchell 1975; Prophet 1982, 1985, 1988; Zlebell et al. 1986); however, none of these studies presents larval fish densities. Thus, the extent to which zooplanktlvory by larval flsheB was related to zooplankton declines cannot be determined. In lakes with resident shad populations, correlative relationships have been suggested between the occurrence of larval shad and zooplankton decline (Cramer and Harzolf 1970; Johnson 1970; Mayhew 1977; Kashuba and Matthews

1984; Klsslck 1988). Manipulating larvae in 2 -ra^ bags, one of us

(DeVries 1989) has demonstrated that larval gizzard shad ("19 flsh’m*

3) can reduce macrozooplankton density from 700 to < 10 animals•liter*

w ith in 2 weeks, a decline similar in magnitude to that observed in

Stonelick Lake. Coupled with this result, we believe that, at least in Stonelick Lake, pjredatlon by larval threadfln shad did result in a significant reduction in zooplankton densities.

In Stonelick Lake, offshore larval blueglll abundance did not 61

differ between pre- and poet*threadfln shad years and several peaks

occurred during both years. Fish from the protracted spawn during

1987 continued to migrate to the littoral zone, as evidenced by

Increasing inshore abundance through summer. In 1988, fish from the

early-June peak migrated to the litto ral zone during July; however,

fish from the August peak never appeared Inshore. These fish were

lost to our offshore gear 2 weeks after their peak abundance; yet,

they were never collected in the littoral zone. As suggested earlier,

predation by adult threadfln shad probably Is not responsible.

However, because threadfln shad reduced zooplankton abundance during

August, offshore larval blueglll likely experienced poor survival.

First-feeding limnetic larval blueglll and threadfln shad both

selected copepod nauplli in Stonelick Lake, after which blueglll

switched to calanold copepods and Dlaphanosoma and larval threadfln

shad fed on a variety of prey taxa, including calanold and cyclopoid

copepods and Dlaphanosoma. As In Clark Lake, overlap values were

typically below 0.5, although both species selected Dlaphanosoma. As

densities of all zooplankton declined In August, overlap remained

moderately high (range - 0.45 • 0.64), perhaps leading to competitive

effects that reduced blueglll survival and recruitment.

There is little data in the literature to Indicate whether larval

threadfln shad and blueglll compete for limnetic zooplankton. Diets of blueglll and shad (gizzard shad or threadfln shad) have rarely been quantified and interpretation of published results remains tenuous.

There is some agreement that first-feeding shad and blueglll select copepod nauplli (Hayhew 1977; Nallln et al. 1985; Klssick 1988); 62

consequently, they could compete If they co * occurred at this stage,

In our lakes, as threadfln shad gre w, they preferred larger prey, such as Dlaphanosoma. This conflicts wi th published results that shad

(both gizzard and threadfln shad) feed on progressively smaller prey as they grow, I.e., rotifers and flfl going (Barger and Kllambl 1980; Van

Den Avyle and Wilson 1980). Howeve r, without zooplankton availability data, prey preference by shad In these studies cannot be evaluated. For example, If zooplankton abundance declined, as In

Stonelick Lake In August and as doc juaented In other lakes containing shad, then shad may be forced to fe ad on smaller prey Items as the

fish grow; they may simply be consua lng zooplankton In proportion to

their availability, rather than actively selecting smaller prey items.

Our results agree with those of sevsral studies documenting larval fish selection of Dlaphanosoma (Van Den Avyle and Wilson 1980; Mallln et al. 198S), although this Is certainly not a universal result (e.g.,

Cramer and Marzolf 1970). Consequently, though competition between larval blueglll and larval threadfln shad may be suggested, particularly during August in Stonelick Lake, the outcome of interactions between limnetic larvaa of these species across lakes ultimately depends on both abundance and species composition of the zooplankton community.

As In Clark Lake, diets of blueglll collected inshore in

Stonelick Lake included both littoral and limnetic prey, reducing competition with threadfln shad onca blueglll became littoral. During

1987, blueglll collected In the lltuoral zone fed on littoral and limnetic prey types, apparently mov mg far enough offshore to consume 63

limnetic prey. However, in 1988 limnetic prey were an extremely small

component of their diet. Zooplankton abundance was reduced to such a

point that blueglll fed only on littoral prey, particularly after

August 1.

Complex_Effects_on_JYoung^of^vear Largemouth Bass

Because young-of-year largemouth bass remained in the litto ral

zone in both lakes, they did not compete with limnetic threadfln shad.

Furthermore, owing to this spatial segregation, young-of-year

largemouth bass did not eat larval threadfln shad. Thus, the apparent

effect of larval shad on young-of-year bass was complex and

unexpected. In Clark Lake, largemouth bass spawned several weeks

later in 1988 than 1987; consequently, when juvenile blueglll moved

inshore, they were too large for largemouth bass to consume. In spite

of this, in 1988, largemouth bass were able to compensate with other

foods (primarily Corixidaey, and prey biomass in their stomachs did not differ across years. In Stonelick Lake during 1987 small fish

(primarily Centrarchldae) were a major component of bass diets, due in part to the continual recruitment of larval blueglll Inshore. In

1988, there was essentially no recruitment of Juvenile bluegills to the litto ral zone after mid-July. Young-of-year largemouth bass were also extremely abundant in early summer in 1988. This high density of young-of-year largemouth bass, coupled with the lack of recruitment of

Juvenile bluegills as forage fish In 1986, Is likely responsible to the strong reduction in largemouth bass growth rates in 1988 compared to 1987. Thus, interaction between larval threadfln shad and larval blueglll may have a pronounced negative effect on young-of-year 64

largemouth bass growth, if it leads to a reduction in Juvenile

recruitment to the littoral zone. Additionally, reduced growth may

influence largemouth bass recruitment, if overwinter survival depends

on body size and fat reserves (Adams et al. 1982a,b). As a

consequence, the very management practice intended to enhance the

fishery for adult piscivores may reduce recruitment of the target

species. Although these negative effects are not direct (as are the

positive effects of increased forage availability), the long-term

(i.e ., over several years) outcome of such a management manipulation,

from the combination of direct and indirect effects, could be

significant.

Conclusions and Management Recommendations

Because most sport fishes change their diets and habitat use

dramatically during their ontogeny, the total impact of an introduced

forage species on all phases of the "target” species life history must be considered. For example, although growth of adult largemouth bass and white crapple increased in two studies following forage fish

introduction (Fast et al. 1982, Li et al. 1976), growth of younger

fish was reduced, and was suggested to be due to competition with the

introduced forage fish. Also, while Mvals has occasionally been

Introduced to provide supplemental forage for predators, their impact on benthlc organisms during the day and macrozooplankton at night

(Lasenby et al. 1986) may result in competition with the very species that they are expected to enhance. Clearly, if we hope to predict the outcome of such manipulations, we must consider effects of 65 interactions aaong all life stages of all species within the community, not just the introduced and target species. Indirect effects, mediated through other species, or even through other trophic levels, are critical to determining the ultimate outcome of the manipulation. 66

Literature Cited

Adams, S. M., R. B. McLean, and M. M. Huffman. 1982a. S tru c tu rin g o f a predator population through temperature-mediated effects on prey availability. Canadian Journal of Fisheries and Aquatic Sciences 39:1175-1184,

Adams, S. M., R. B. McLean, and J . A. P a rro tta . 1982b. Energy partitioning in largemouth bass under conditions of seasonally fluctuating prey availability. Transactions of the American Fisheries Society 111:549-558.

Austin, M. R. and S. T. Hurley. 1989. Evaluation of a threadfln shad Introduction on crappie growth in an Ohio lake(s). Annual Performance Report, Project F-29-R-28, Study 21. Ohio Department of Natural Resources, Columbus, Ohio, USA.

Baglin, R. E., Jr., and R. V. Kilambi. 1966. Maturity and spawning periodicity of the gizzard shad, Dorosoma capedlanum. (LeSueur), in Beaver R eserv o ir. Arkansas Academy o f Science 22:38-43.

Barger, L. E., and R. V. Kilambi. 1980. Feeding ecology of larval shad, Dorosoma. in Beaver reservoir, Arkansas. Pages 136-145 In L. A. Fulman, editor. Proceedings of the Fourth Annual Larval Fish Conference. United States Fish and U lldlife Service, Ann Arbor, Michigan, USA.

Beard, T. D. 1982. Population dynamics of young-of-the-year b lu e g lll. T echnical B u lle tin Number 127, W isconsin Department o f Natural Resources, Madison, Wisconsin, USA.

Buynak, G. L,, B. T. Kinman, and R. V. Jackson. 1989. Purse seining to capture large numbers of adult threadfln shad. North American Journal of Fisheries Management 9:121-123.

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

Chesson, J. 1983. The estimation and analysis of preference and Its relationship to foraging models. Ecology 64:1297-1304.

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

Cushing, D. H. 1975. Marine ecology and fisheries. Cambridge 'University Press, Cambridge, England.

DeVries, D. R. 1989. The influence of an open-water planktivore on reservoir communities: the importance of trophic-level interactions and ontogenetic niche shifts. Doctoral Dissertation, The Ohio State University, Columbus, Ohio, USA. 67

DeVries, D. R., and R. A. Stein. In press. Manipulating shad to enhance sport fisheries in North Anerlca: an assessment. North American Journal of Fisheries Management.

Downey, P., and D. Toetz. 1983. Distribution of larval gizzard shad fDorosoma cepedlanura) in Lake Carl Blackwell, Oklahoma. American Midland N aturalist 109:23-33.

Fast, A. W., L. H. Bottroff, and R. L. Miller. 1982. Largemouth bass, Mlcrooterus salmoldes. and blueglll, Lepomls macrochirus. growth rates associated with artificial destratlflcatlon and threadfln shad, Dorosoma petenense introductions at El Cap!tan Reservoir, California. California Fish and Game 67:4-20.

Gilliam, J. F. 1982. Foraging under mortality risk in size-structured populations. Doctoral dissertation, Michigan State University, East Lansing, Michigan, USA.

Haney, J. F., and D. J. Hall. 1973. Sugar-coated Daohnla: a preservation technique for Cladocera. Limnology and Oceanography 18:331*333.

Heldinger, R.-, and F. Imboden. 1974. Reproductive potential of young-of-the-year threadfln shad (Dorosoma petenense) in southern Illinois lakes. Transactions of the Illinois Academy of Science 67:397-401.

Johnson, J. E. 1970. Age, growth, and population dynamics of threadfln shad, Dorosoma petenense (Gunther), in central Arizona reservoirs. Transactions of the American Fisheries Society 99:739-753.

Kashuba, S. A., and W. J. Matthews. 1984. Physical condition of larval shad during sprlng-summer in a southwestern reservoir. Transactions of the American Fisheries Society 113:199*204.

Keast, A. 1980. Food and feeding relationships of young fish in the first weeks after the beginning of exogenous feeding in Lake Oplnicon, Ontario. Environmental Biology of Fishes 5:305-314.

Kirk, J. P. 1984. Competitive influences of gizzard shad introductions on balanced largemouth bass/bluegill populations. Doctoral Dissertation, Auburn University, Auburn, Alabama, USA.

Kirk, J. P. ,and W. D. Davies. 1985. Competitive influences of gizzard shad on largemouth bass and blueglll in small impoundments. Proceedings of the Annual Conference of the Southeastern Association of Fish and U ildlife Agencies 39:116-124. 68

Klsslck, L. A. 1988. Early life history of the gizzard shad in Acton Lake, Ohio: feeding ecology and d rift of stream-spawned larvae. Master's thesis, Miami University, Oxford, Ohio, USA.

Lasenby, D. C., T. G. Northcote, and M. Furst. 1986. Theory, practice, and effects of Mvsls relicta introductions to North America and Scandinavian lakes. Canadian Journal of Fisheries and Aquatic Sciences 43:1277-1284.

Lasker, R. 1975. Field criteria for survival of anchovy larvae: the relation between Inshore chlorophyll maximum layers and successful first feeding. Fishery Bulletin 73:453-462.

Li, H. V., P. B. Moyle, and R. L. Garrett. 1976. Effect of the introduction of the Mississippi silverslde (Menldla audens) and the growth of black crapple (Pomoxla nlgromaculatus) and white crappie (£. annularis) in Clear Lake, California. Transactions of the American Fisheries Society 105:404-408.

Mallin, M, A., L. J. Blrchfleld, and .W. Warren-Hicks. 1985. Food habits and diet overlap of larval Leoomls spp. and gizzard shad in a Piedmont reservoir. Proceedings of the Annual Conference of the Southeastern Association of Fish and W ildlife Agencies 39:146-155.

Matthews, W. J. 1984. Influence of turbid inflows on vertical distribution of larval shad and freshwater drum. Transactions of the American Fisheries Society 113:192-198.

Mayhew, J. 1977. The effects of flood management regimes on larval fish and fish food organisms at Lake Rathbun. Iowa Fisheries Technical Series 77-2, Iowa Conservation Commission, Des Moines, Iowa, USA.

Mills, E. L., J. L. Forney, and K. J. Wagner. 1987. Fish predation and its cascading effect on the Oneida Lake food chain. Pages 118-131 Jjq W, C. Kerfoot and A. Slh, editors. Predation: direct and Indirect impacts on aquatic communities. University Press of New England, Hanover, New Hampshire, USA.

Mills, E. L., and J. L. Forney. 1988. Trophic dynamics and development of freshwater pelagic food webs. Pages 11-29 in S. R. Carpenter, editor. Complex interactions in lake communities. Springer-Verlag, New York, New York, USA.

Ney, J. J. 1981. Evolution of forage-fish management in lakes and reservoirs. Transactions of the American Fisheries Society 110:725-728.

Noble, R. L. 1981. Management of forage fishes in impoundments of the southern United States. Transactions of the American Fisheries Society 110:738-750. 69

Noble, R. L. 1986. Predator-prey Interaction* in reservoir communities. Page* 137-143 In C. E. Hall and M. J. Van Dan Avyle, editors. Reservoir Fisheries Management: Strategies for the 80*8. Reservoir Committee, Southern Division American Fisheries Society, Bethesda, Maryland, USA.

Persson, L. 1988. Asymmetries in competitive and predatory in te ra c tio n * in f is h p o p u latio n s. Pages 203-218 I n Ebenman and L. Persson, editors. Size-structured populations. Springer- Verlag, New York, New York, USA.

Ploskey, G. R. 1986. Effects of water-level changes on reservoir ecosystems, with implications for fisheries management. Pages 86-97 I n G. E. H all and M. J . Van Den Avyle, e d ito r s . R eservoir Fisheries Management: Strategies for the 80's. Reservoir Committee, Southern Division American Fisheries Society, Bethesda, Maryland, USA.

Prophet, C. V. 1982. Zooplankton changes in a Kansas lake 1963-1961. Journal of Freshwater Ecology 1:569-575.

Prophet, C. V. 1985. Calanold population structure in a Kansas lake after Introduction of threadfln shad. The Southwestern N aturalist 30:162-163.

Prophet, C. U, 1988. Changes in seasonal population structures of two species of Diaotomus (Calanolda, Copepods) subsequent to introductions of threadfln and gizzard shad. The Southwestern Naturalist 33:41-53. ‘

Schooner, T. W. 1970. Non-synchronous spatial overlap of lizards in patchy habitats. Ecology 51:408-418.

Storck, T. W. 1978. The distribution of limnetic fish larvae In a flood control reservoir in central Illinois. Transactions of the American Fisheries Society 107:419-424.

Swenson, V. A. and M. L. Matson. 1976. Influence of turbidity on survival, growth, and distribution of larval herring (Corezonua artedlD . Transactions of the American Fisheries Society 105:541-545.

Tisa, M. S., J. J. Ney, and D. K. Whitehurst. 1965. Spatial and temporal distribution of larval alewives and gizzard shad in a Virginia reservoir. Proceedings of the Annual Conference of the Southeastern Association of Fish and W ildlife Agencies 39:65-73.

Van Den Avyle, M. J., and J. R. Wilson. 1980. Food habits and feeding selectivity of larval Dorosoma spp. in Center Hill Reservoir. Pages 146-156 In L. A. Fuiman, editor. Proceedings of the Fourth Annual Larval Fish Conference. United States Fish and W ildlife Service, Ann Arbor, Michigan, USA. 70

VonGeldom, C., Jr. and D. F. Mitchell. 1975. Largemouth baa a and threadfln shad In California. Pages 436-449 in R. H. Stroud and H. Clepper, editors. Black bass biology and management. Sport Fishing Institute, Washington, D.C., USA.

Wallace, R. K., Jr. 1981. An assessment of dlet-overlap indexes. Transactions of the American Fisheries Society 110:72-76.

Ware, D. M. 1980. Bioenergetics of stock and recruitment. Canadian Journal of Fisheries and Aquatic Sciences 37:1012-1024.

Werner, E. E., and J. F. Gilliam.' 1984. The ontogenetic niche and species interactions in size-structured populations. Annual Review of Ecology and Systematics 15:393-425,

Werner, E. E. and 0. J. Hall. 1988. Ontogenetic habitat shifts in blueglll: the foraging rate-predation risk trade-off. Ecology 69:1352-1366.

Werner, R. C. 1967. Intralacustrine movements of blueglll fry in Crane Lake, Indiana. Transactions of the American Fisheries Society 96:416-420.

W illis, D. W. 1987. Reproduction and recruitment of gizzard shad in Kansas reservoirs. North American Journal of Fisheries Management 7:71-80.

Wydoskl, R. S., and D. H. Bennett. 1981. Forage species in lakes and reservoirs of the western United States. Transactions of the American Fisheries Society 110:764-771.

Ziebell, C. D., J. C. Tash, and R. L. Barefield. 1986. Impact of threadfln shad on microcrustacean zooplankton in two Arizona lakes. Journal of Freshwater Ecology 3:399-406. Table 3. Food selection (using Chessan's alpha, Chesson 1978, 1983) by larval blueglll collected offshore in' Clark lake before (1987) and after (1988) threadfin shad introduction. Data are presented as moans + 95% confidence interval. Values greater than neutral selection (i.e ., between

0.09 and 0.13, the reciprocal of the number of prey items in the lake) indicate positive selection; values less than this level indicate avoidance. Bluegill > 13.0 nm were not collected during 1988. s iz e range calanoid cyclcpoid copepod fwa TL) Bosmina Oeriodachnia Danhnia Piaphanosotna copeDods ocoepods nauplli rotifers

PRE-SHAD (1987)

4.0- 6.9 0.00+0.01 0 0.06+0.04 0.36+0.23 0 0.06+0.09 0.49+0.22 0.03+0.05

7.0- 9.9 0.00+0.01 0.00+0.01 0.03±0.01 0.80+0.15 0 0.04+0.03 0.06+0.10 0.08+0.10

10.0-12.9 0.02+0.04 0.02+0.02 0.02+0.02 0.58+0.14 0.01+0.01 0.24+0.13 O.O^+O.Ol 0.09+0.06

> 13.0 0.06+0.11 0.03+0.01 0 0.78+0.20 0.00+0.01 0.06+0.08 0 0.07+0.08

POST-SHAD (1988)

4.0- 6.9 0.09+0.21 0.19+0.24 0.14+0.22 0.43±0.31 0 0.14+0.22 0.02+0.03

7.0- 9.9 0.00+0.01 0 0.01+0.01 0.93±0.08 0.02+0.03 0 0 0.04+0.07

10.0-12.9 0.01+0.01 0.00+0.01 0.01+0.01 0.94+0.04 0.04+0.04 0 Table 4. Pood selection (using Chesscn's alpha, Chesson 1978, 1983) by larval bluegill collected offshore in Stonelick Lake before (1987) and after (1988) threadfln shad introduction. Data are presented as means + 95% confidence interval. Values greater than neutral selection (i.e ., between

0.09 and 0.13, the reciprocal of the number of prey items in the lake) indicate positive selection; values less than this level indicate avoidance.

s i z e ran ge calanold cyclopoid copepod fn«n TIA Bosmina Oeriodachnia Parhnia Piaphanosaia copepods copepodspaupjU r o t if e r s ERE-SHAD (1987)

4.0- 6.9 0.01±0.03 0 0 0.12+0.19 0 0.02+0.02 0.84+0.12 0.07+0.08

7 . 0 - 9 .9 0 0 0.02+0.02 0.71+0.15 0.02+0.02 0.18+0.09 0.18+0.11 0.04+0.05

10.0-12.9 0.0110.01 0 0.03+0.02 0.52+0.16 0.02+0.02 0.36+0.14 0.05+0.03 0.01+0.01

> 1 3 .0 0 0.03+0.03 0.1Q+0.05 0.44+0.20 0.04+0.04 0.35+0.19 0.03+0.03 0

POST-SHAD (1988)

4.0- 6.9 0.08+0.11 0.04+0.08 0 0.23+0.22 0.12+0.13 0.07+0.10 0.50+0.20 0.04+0.08

7.0- 9.9 0.08+0.08 0.02+0.04 0.02+0.04 0.40+0.17 0.39+0.15 0.10+0.08 0.03+0.05 0.05+0.05

10.0-12.9 0.24+0.14 0.04+0.06 0.05+0.09 0.36+0.29 0.38+0.18 0.01+0.01 0.03+0.04 0.01+0.01

> 13.0 0.13+0.20 0.00+0.01 0.03+0.05 0.76+0.24 0.07+0.05 0.02+0.05 0.01+0.03 0.03+0.07 r o Table 5. Food selection (using Chesson's alpha, Chesson 1978, 1983) by larval threadfln shad in dark and Stonelick lakes during 1988. Data are presented as means + 95% confidence interval.

Values greater than neutral selection (i.e ., the reciprocal of the number of prey items in the lake, here between 0.09 and 0.13) indicate positive selection; values less than this level indicate avoidance. Dash (-) ■* fish of the indicated size not present. Threadfln shad < 7.0 ran were not present in dark lake.

s i z e ran ge calanoid cyclopoid copepod fmn TL) Bnmina Qeriodarhnia Dachnia Diarhanosctna conepods ccpeoods raup l i l r o t if e r s CLARK LAKE

7.0- 9.9 0.14+0.35 0 0.06+0.15 0.50+0.57 0 0.13+0.30 0.13+0.29 0.18±0.30

10.0-12.9 0 0 0.05+0.09 0.31±0.21 0.11+0.13 0.40+0.22 0.05+0.09 0.09+0.13

> 13.0 0.01+0.01 0 0.02+0.02 0.42+0.08 0.35+0.08 0.19+0.07 0.01+0.01 0.02+0.02

STONEUCK LAKE

4.0- 6.9 0 0 0 0 0 0 1 . 00+0 .0 0 0

7.0- 9.9 0 0 0 0.18+0.20 0.08+0.11 0.14+0.12 0.56+0.19 0.11±0.11

10.0-12.9 0.03+0.05 0 0 0.33+0,16 0.35+0.15 0.11±0.09 0.12+0.09 0.11+0.10

> 13.0 0.02+0.02 0.02+0.03 0.02+0.03 0.24+0.11 0.54+0.10 0.09+0.05 0.09+0.06 0.06+0.05 Figure 4, Density (mean ± 1 SE) of larval threadfin shad in Clark and

Stonelick lakes, Ohio following adult stocking in April 1988.

74 iue 4 Figure n THREADFIN SHAD • E . STONELICK B. LAKE " 2 . LR LAKE CLARK A. UAY UEJULY JUNE AUO SEPT Figure 5. Density (mean ± 1 SE) of larval blueglll in offshore larval

tows and inshore seine samples in Clark and Stonelick lakes, Ohio before (1987, dotted line) and after (1988, solid line) threadfln shad were Introduced.

76 CLARK LAKE STONEUCK LAKE

A. OFFSHORE C. OFFSHORE 3

2 - I pre-TS (1987) poat-TS (1988) I ?198^ \ port—IS (1988) , / ' * i V f o A j - , B. INSHORE D. INSHORE

4 -

JULY AUG SEPT OCT MAY JUNE JULY AUG Figure 6 . Diets of young-of-year blueglll collected inshore in Clark and Stonelick lakes, Ohio during 1987 and 1988. Prey were divided into limnetic zooplankton, litto ral prey, and cyclopoid copepods

(which can be found in either habitat) (Mlttelbach 1981). Results were combined across samples sites and are presented as a percentage of the total biomass of prey found in all fish guts on that date.

Sample sizes were 5-15 fish for each date.

78 PERCENT OF TOTAL DIET 100 100T 40 80 60 0 2 0 2 80 40 60 . 1988 B. . 1987 A. JUNE JULY LITTORAL LR LAKE CLARK LITTORAL UUT ET OCT SEPT AUGUST mne ic et n im l CYCLOPOIOS UMNEDC IPOIOS . 1988 D. UEJULY JUNE . 1987 C. LIMNETIC TNLC LAKE STONELICK UMNETIC UUT ET OCT SEPT AUGUST Figure 7. Density (mean 1 1 SG) and length of young*of*year largemouth bass in Clark and Stonelick lakes, Ohio. Dotted and solid lines represent estimates from before (1987) and after (1988) threadfln shad introduction, respectively.

80 H*

0 .4

post—T5 (1988)

0.0----- £ 1 0 0 B E IT' 75

MAY JUNE JULY AUG SEPT OCT MAY JUNE JULY AUG SEPT OCT t-*CO Figure 8 . Diets of young-of-year largemouth bass collected inshore in

Clark and Stonelick lakes, Ohio during 1987 and 1988. Frey categories included fish, zooplankton, corixids, and other littoral invertebrates. Results were combined across sample sites and are presented as a percentage of the total biomass of prey present in all fish guts on that date. Sample sizes were 2-7 fish for each date, u su a lly S.

82 PERCENT OF TOTAL PREY 100 100 40 80 60 0 2 40 60 80 . 1988 B. . 1987 A. JUNE FISH JULY LR LAKE CLARK UUT ET OCT SEPT AUGUST C0RIX1DS INVERTS . 1987 C. JUNE ? 8 9 1 FISH ZP TNUK LAKE STONEUCK JULY UUT ET OCT SEPT AUGUST C0RJX1DS Figure 9. Density (mean ± 1 SE) of macrozooplankton in Clark and

Stonelick lakes, Ohio before (1987) and after (1988) threadfin shad introduction.

84 •sM CP CLARK LAKE STONEUCK LAKE

A. 1 9 8 7 C. 1 9 8 7 1500

2 1000 - ■

5 0 0

B. 1988 D. 19 8 8 O 1500

< 1000

500

MAY JUNE JULY AUG SEPT OCT MAY JUNE JULY AUG SEPT OCT Figure 10. Density of macrozooplankton taxa in Clark and Stonelick lakes, Ohio before (1987) and after (1988) threadfin shad.

Abbreviations for taxa are ca-calanold copepods, cy-cyclopoid copepods, na-copepod nauplli, da-Daohnla. bo-Bosmlna. ce-Cerlodaphnia. dl-Dlaphanosoma.

86 (g « CP *1 NUMBER-LITER" 1500 1000 1500 1000 500 PI MAY APRIL JUNE LR LAKE CLARK UY U SP OT PI MY UE UY U SP OCT SEPT AUG JULY JUNE MAY APRIL OCT SEPT AUG JULY TNUK LAKE STONEUCK 09 CHAPTER I I I

COMPLEX INTERACTIONS AMONG FISH, ZOOPLANKTON, AND PHYTOPLANKTON

AS INFLUENCED BY AN OPEN-WATER PLANKTIVORE

Introduction

Predators can dramatically Influence aquatic communities, both directly and indirectly (see chapters in Kerfoot and Sih 1987;

C arpenter 1988; Ebenman and Persson 1988). By removing prey, predators directly alter community species composition. Indirect effects of predators can take several forms, leading to positive

(e.g., Kerfoot 1987) as well as negative (e.g., Mittelbach and Chesson

1987) effects for prey. The current focus within ecology on blomanlpulatlon (Shapiro and Wright 1984) and cascading trophic interactions (Carpenter et al. 1985, 1987), as tools to influence water quality, emphasizes the indirect impacts that predators can have on entire lake ecosystems, with effects of predation occurring beyond adjacent trophic levels.

System-wide effects of predators have been well-described for planktlvore-plankton interactions (see reviews in Hall et al. 1976,

Greene 1985, Lazzaro 1987, Northcote 1988, Gliwicz and Pijanowska

1989), In general, vertebrate planktivores remove large zooplankters

(e.g., Brooks and Dodson 1965) and invertebrates remove small forms

(e.g., Wong 1981, Dodson 1974). Conventional wisdom suggests that

88 89 when planktlvorous fish occur in a lake, large zooplankters will be selectively removed and smaller forms w ill dominate (e.g., Brooks and

Dodson 1965, Wells 1970, Warshaw 1972, Evans 1986). Furthermore, because vertebrate predators select large prey, they also will remove invertebrate predators, which tend to be large, reducing their predation on small zooplankton and reinforcing the dominance of small forms (Dodson 1970, 1974). Without fish, large plankters may outcompete small ones (but see Smith and Cooper 1982, Bengtsson 1987) and, because invertebrate predators can survive and remove small zooplankters (Lynch 1979), large forms predominate.

Studies detailing these interactions typically have involved adult fishes; however, larvae of most fishes also feed on zooplankton

(e.g., Applegate and Mullan 1967, Werner 1969, Keast 1980, Rajasilta and Vuorinen 1983, Whiteside et al. 1985a,b, Whiteside 1989, Confer and O'Bryan 1989). Seasonality in fish spawning produces pulses of larval planktlvores that vary in their seasonal impact on the zooplankton. In addition, fishes exhibit dramatic ontogenetic niche shifts (Werner and Gilliam 1984); by shifting to and from planktlvory

(e.g., Mlttelbach 1981, Werner et al. 1983), they provide a temporally variable predation pressure on plankton. In turn, because year-class strength in most fishes is set during early life history (Cushing

1975, Lasker 1975, Ware 1980, H ills and Forney 1988), survival through the larval stage is critical to recruitment. Consequently, understanding how larval fishes and zooplankton interact contributes to understanding the mechanisms underlying adult 90

population size (as well as the role of larval fishes In biomanipulation).

How larval fish Influence zooplankton community structure is not well understood. As gape-limited predators, larvae remove small p la n k te rs (Vong and Ward 1972, Z aret 1980, Hanson and Wahl 1981). As

fish grow, their gape increases until they can capture large plankters

(R osenthal and Hempel 1970, Wong and Ward 1972). How th is seasonal pulse of size-selective larval planktlvores Influences plankton dynamics has received little attention. In Lake Itasca, Minnesota, larval yellow perch (Perea flavescans) spend several weeks in the limnetic zone, then migrate to the littoral zone, where they may reduce zooplankton density (Doolittle 1982, Whiteside et al. 1985b,

Whiteside 1989). Yellow perch in Oneida Lake, New York exhibit similar migration patterns, depressing the density of large limnetic

Daphnia in those years when yellow perch biomass exceeds 2 0 kg/ha in

August (Mills et al. 1987). In Alderfen Broad, England, roach

(RuSilua rutllus) interact with zooplankton, exhibiting a 2 -year cycle. During years of good recruitment, young-of-year fish reduce zooplankton density, competing with and depressing fecundity of older conspecifics for the subsequent year (Cryer et el. 1986). Thus, where larval fish have been studied, they have been capable of influencing zooplankton abundance.

As planktlvores, gizzard shad (Dorosoraa cepedlanunO and threadfin shad (£. petenense) can Influence zooplankton dynamics. In lakes in

Arizona (Ziebell et al. 1986) and in Kansas (Prophet 1982, 1985,

1988), zooplankton densities declined subsequent to threadfin shad 91

introductions. In ponds and laboratory pools, Drenner and his

colleagues document that adult gizzard shad reduce densities of most

easily-captured zooplankters (Drenner et al. 1978, 1982 a,b, Drenner

and McComas 1980), as well as influence phytoplankton community

structure (Drenner et al, 1984, 1986), All studies to date, however,

have dealt with adult fishes, providing no insight Into effects of

la rv a e .

Herein, we quantify how larval gizzard shad Influence reservoir

community structure, as mediated through their influence on

zooplankton. We first document zooplankton and larval gizzard shad

abundance patterns In an Ohio reservoir, suggesting two hypotheses to

explain these patterns. Then, we describe and enclosure/exclosure

experiment to evaluate the role of larval gizzard shad in controlling

zooplankton dynamics in Ohio reservoirs.

Methods

Study lake

Kokoslng Lake (Knox County, central Ohio) is a shallow, turbid

flood control impoundment with 7.5 km of shoreline, a maximum depth of

4.5m, and secchl depths typically < 1 m. Neither submersed nor emergent vegetation were abundant. The fish community consisted primarily of largemouth bass (Hlcroptcrus salmotdes) , white crappie

(PomPKla a n n u la ris ) . b lu e g lll (Leoorals raacrochira. B ailey and Robins

1988), and carp (Cvnrinus carnlo). For additional descriptive details, see Johnson et al. (1988). 92

F ie ld .samplin g methods

Larval fish were collected In two replicate surface tows with a

3/4-ra diameter ichthyoplankton net (mesh size - 500 micron) towed in

the limnetic zone at £ 1.5 m‘S*^ once per week during April through

September 1986*1988. We used a flow meter mounted in the mouth of the

net to calculate volume of water filtered. Larvae were preserved and

returned to the laboratory where they were Identified, measured * (nearest mm, up to 50 per species), and their diets (up to 10 per

species per dote) quantified. Prey were Identified to the lowest

taxonomic category possible (> 802 to genus) and measured (nearest 0 . 1

mm). Prey lengths were converted to biomass using taxon-specific

length-dry weight regressions (G. G. Hittelbach, personnel

communication). Integrated zooplankton samples (four hauls per

sample, three replicate samples per date) were collected each week

during April through September 1987 and 1988 (zooplankton samples were

not collected during 1986) using a 2-m tube sampler (7.30 cm inside

dia, mesh size - 54 micron) simultaneous with larval fish collection

(DeVries and Stein 1990b). Samples were preserved in 5X sucrose

formalin (Haney and Hall 1973) and returned to the laboratory. When <

2 0 0 individuals per taxon were captured, all were counted in each

taxon; subsamples were counted for abundant taxa until at least 2 0 0

individuals were counted. At least 20 individuals of each taxon in a sample were measured using an ocular micrometer. Cladocerans were measured from the anterior portion of the carapace to the base of the basal spine; copepods were measured from the anterior portion of the carapace to the base of the caudal rami. Number of individuals 93 carrying egga ouc of SO randomly chosen individuals and the fecundity

for 20 individuals carrying eggs were recorded for Bosmlna and Daohnia

(the dominant cladocerans) In each sample.

To evaluate the potential for competition between larval bluegill and larval gizzard shad that co-occurred in the limnetic, we used

Schooner's overlap index (Schoener 1970) based on the average proportion that each prey taxon contributed to total biomass in larval fish diets (i.e., the proportion was calculated within each fish and averaged across fish within a date, Wallace 1981). The formula for this index is

n OC - 1 - 0-5 <£.|rxi * ryi|), i - 1 7

where rxj - the proportion of prey type 1 in the diet of species x,

4 and n - the number of food categories. Because larval fish and zooplankton communities changed through time, we treated overlap measures calculated for each sample date as replicates.

To evaluate prey selection by fishes, we compared larval fish diets with zooplankton samples using Chesson's alpha (Chesson 1978,

1983), treating individual fish as replicates within each date. The formula for this index is

—JCiZfii—

Eneloaure/excloflure_exDerlment

To quantify the impact of larval gizzard shad on zooplankton community structure, we conducted an enclosure/exclosure experiment in

Kokoslng Lake during 1988. Twelve polyethylene bags (1,13 m dia, 2 ra deep, 2 m^ volume).were filled with lake water from a depth of 1 m using a large water pump (3.8 cm dla, 189 l*mln~^) on 16-17 May 1988

(just be'fore larval gizzard shad were first collected in the limnetic). On 6 June 1988 larval gizzard shad were collected at night by attracting them to lights (as In Gregory and Powles 1985), and 120

Individuals stocked into each of six randomly-chosen bags. To estimate survival of stocked larval gizzard shad, we introduced 50 fish into two containers (about 200 1) floating next to the bags. The next day these containers were drained; surviving fish were counted and measured. To control for nutrient release in enclosures due to larval shad mortality (as in Threlkeld 1987), 100 dead larval shad

(i.e ., the number we expected to not survive in enclosure) were added to each exclosure. On 7 June 1988 and weekly thereafter, zooplankton were sampled from the bags with a 2 -ra integrated tube sampler.

Procedures for zooplankton sampling and processing were as for lake samples. Phytoplankton samples were collected from three randomly- chosen enclosures and exclosures on 7 and 21 June 1989 by taking 100 ml of water from an Integrated sample (collected using the 2 -m tube sampler), and preserved with Lugol's iodine. Samples were concentrated by settling for 24-48 h and two replicate transects across a Sedgewlck-Rafter cell counted for each sample. On 5 July

1988 bags were drained using the large water pump; all fish were 95 collected and preserved. Data vara analyzed using ANOVA and repeated- measures ANOVA techniques (GLM procedure, SAS Institute, Inc. 1985).

R esults

Field pattern? larval gizzard shad and_zooplankton

Larval gizzard shad abundance peaked during late-Hay to early-

June across years (Fig. 11A,D). Larvae were first captured during mid* to late-May, with peak densities occurring within 1 week of first capture. Peak density varied across years, ranging from 14 fish*m*3 during 1988 to 84 fish'm’^ during 1986.

Although macrozooplankton (i.e ., net zooplankton minus rotifers) abundance d iffe re d during 1987 and 1988 (repeated*m easuras ANOVA, F i ,4

- 64.49, p - 0.001) and the timing of the peak differed across years

(time x year interaction, F2Q(80 " 15.58, P “ 0.0002), the overall pattern of a single peak during spring, followed by a dramatic decline, was similar across years (Fig. 11B,E). Rotifer abundance patterns, as well as absolute densities, varied across years (year effect, F]^ - 15.69, p - 0.02; time x year Interaction, Fjq^O "

18.92, p - 0.0002). However, rotifer density did not consistently decline across years as did the macrozooplankton (Fig. 11C,F). In both years, periods of most rapid macrozooplankton decline occurred immediately following (within 2 weeks) peak larval gizzard shad density, suggesting these predators influence macrozooplankton d e n s itie s .

However, declining zooplankton abundance could be caused by decreased zooplankton birth rates as well as by increased predation 96 pressure by gizzard shad. To evaluate how zooplankton fecundity was related to declining densities, we regressed mean fecundity

(calculated as the product of the proportion of individuals that carried eggs and the mean number of eggs per individual carrying eggs) of Bosmlna and Daphnla. the dominant cladocerans, on time during the decline in zooplankton density in both years (11 May* 10 June 1987, 17

May- 27 June 1988; Fig. 11B,E). Neither taxa changed in fecundity during 1987 (Fig. 12A; linear regression, p > 0.16 for both Bosmina and Daphnla). During 1988 fecundity of both taxa decreased during the macrozooplankton crash (Fig. 12B; linear regression, p < 0.004 for both taxa).

Enclosure/exclosure experiment

Survival of stocked gizzard shad larvae, based on survival in the holding containers, was 32.4X (range 29.4-35.4X, n-2), yielding a stocking density of 19.4 fish/m^, similar to the peak density (13.9 fish/m^) in Kokosing Lake during 1988 (see Fig 11D). Mean size of stocked fish was 18.4 ± 0 .4 mm TL (mean ± 1 SE, n -3 8 ).

Fish survival varied across enclosures, yielding densities at experiment's end of 6 to 15.5 fish/m^ (mean ± 1 SE - 10.9 + 1.4 fish/m^). In addition, two exclosures contained fish at the end of the experiment (densities - 1.5 and 5.5 fish/m^), Because these bags had no holes at draining, fish must have been introduced as exclosures were filled. Because fish were predicted to reduce zooplankton density, their presence in "exclosures" generates a conservative test; thus, data from these exclosures were included in all analyses as fish exclosure data. At experiment's end, fish averaged 40.7 + 2.1 mm 97

total length (mean ± 1 SE), and final length was negatively related to

final gizzard shad density (Fig. 13; linear regression, Fj^g-16.88,

p-0.006). This relationship was significant regardless of whether

data from the two exclosures with fish were included.

Macrozooplankton density declined in enclosures relative to

exclosures (Fig. 14A; repeated-measures ANOVA, treatment effect

- 76.00, p - 0.0001; time x treatment interaction F 4 4 Q " 35.74, p -

0.0001). Treatment differences were nearly significant 1 week after

larval shad introduction (ANOVA, Fj^iq - 6.91, p - 0.025; p required

for signlflcance-[0.05/5]-0.01). After 2 weeks, zooplankton density

In enclosures was reduced by 95X (ANOVA, Fj^iq - 39.55, p - 0.0001).

Though in itial macrozooplankton density was marginally higher in bags with fish than in the lake (ANOVA, F j^ - 7.99, p - 0.03, p required for significance - [0.05/5] - 0.01), density did not differ between the lake and bags with fish' and declined similarly (Fig. 14A; repeated-measures ANOVA, treatment effect F ]^ - 1.71, p - 0,23).

Higher initial density in bags versus the lake was due to bags excluding all planktlvores for the 3 weeks between filling and fish addition. Rotifer densities also declined in enclosures relative to exclosures (Fig. 14C; repeated-measures ANOVA, treatment effect F^^q

-26.76, p - 0.0004, treatment x time interaction F 4 4 0 " 19*58, p -

0.0001). However, rotifer density did not differ between treatments until 2 weeks after fish introduction (Fig. 14C; ANOVA F^^q - 6.75, p

- 0.027); this difference became larger through time,

Macrozooplankton size differed between treatments (Fig. 14B; repeated-measures ANOVA, treatment effect, F^iO - 7.88, p - 0.02; 98

treatment x time interaction, F4,40 " 82, p - 0.002). Size did not

differ between treatments at the beginning or end of the experiment

(ANOVAs, p > 0,11), but was smaller in enclosures 2 and 3 weeks after

fish addition (ANOVAs, p < 0.003). However, zooplankton size within a

taxon did not differ (Fig. IS; repeated-measures ANOVA, treatment

effects all p > 0.07). The time x treatment interaction term was

significant only for Bosmlna (p - 0.01; for all other taxa p > 0.18);

however, without a significant treatment effect for Bosmlna (p >

0.66), gizzard shad did not consistently Influence Bosmlna size

through time.

As discussed earlier, declining zooplankton abundance could be

caused by decreased zooplankton birth rates as well as by Increased

predation pressure by gizzard shad. Mean fecundity of Bosmlna

differed between treatments (repeated-measures ANOVA, F^ iq - 16.42, p 4 *

- 0 . 0 0 2 ), although the time x treatment Interaction was also significant (F 4 (4 q - S.25, p < 0.01) (Fig. 16A). The only date on which mean fecundity differed between treatments was on 27 June 1988, when fecundity was higher In exclosures (Fig. 16A). Mean fecundity of * Daohnla declined sim ilarly in both treatments (repeated-measures

ANOVA, time x treatment Interaction, F 4 (4 q - 0 .8 5 , p > 0 .4 8 ), b u t was lower in enclosures than in exclosures (treatment effect, F^^q -

6.07, p - 0.03) (Fig. 16B).

Edible phytoplankton volume was reduced in enclosures, relative to exclosures (Fig. 17; repeated-measures ANOVA; treatment effect,

F i ' 4 - 12.55, p - 0.02; time x treatment interaction, Fj ^ 4 - 12.67, p

- 0.02). However, shad did not influence volume of inedible 99

phytoplankton (p > 0.17 for both the treatment effect and the time x

treatment interaction). Because 80-9SX of edible phytoplankton, at

least on 7 June and In exclosures on 21 June, was Crvotomonas. the

response of edible phytoplankton was due to a reduction in Crvotomonas

in enclosures relative to exclosures.

Field patterns

Bluegill spawned for up to 3 months during 1986-1989, as

documented by presence of limnetic larvae. Larval bluegill and

gizzard shad typically overlapped in the limnetic for only 2-4 weeks during 1986-1989 (Fig. 18). However, because bluegill spawned later

than gizzard shad, larval bluegill arrived in the limnetic as macrozooplankton were declining precipitously (Fig. 19).

Small larval bluegill preferred small prey (nauplil and rotifers) during both years; bluegill £ 1 0 mm did not feed preferentially on any prey taxa during 1987, but selected Dlaphanosoma during 1988 (Table

1). Variability in alphas was high, in part, because sample sizes were small, given that few bluegill were collected in offshore larval tows, particularly during 1987.

During 1987, diets of larval gizzard shad < 10 mm were not quantified; fish > 1 0 mm either were nonselective or-chose rotifers

(Table 2). During 1988 small gizzard shad selected rotifers, followed by copepod nauplil, rotifers, and cyclopold copepods as they grew

(Table 2). In addition, phytoplankton was observed in larval gizzard shad‘diets, though poor preservation techniques for fish prohibited us from quantifying or identify algae. Phytoplankton was never observed

In larval bluegill diets. Overlap between larval bluegill and gizzard 100

shad, measured using Schoener's index, and based only on zooplankton

in fish diets, was 0.52 ± 0.16 (mean ± 1 SE, n-4) during 1987 and 0,52

± 0 .0 5 (mean ± 1 SE, n-3) during 1988.

To evaluate how larval fish responded to the decline of

macrozooplankton, we quantified diets of 15-mm gizzard shad (size held

constant to control for changes in diet due to fish size) before,

during, and after the decline during 1987 and 1988. Diets were

quantified from five dates during both years: two during the

macrozooplankton peak, two during the decline, and one after the

crash. Biomass of prey consumed per fish differed across sample dates

(F4,40 “ 5 *97, P *■ 0.0007), and differed marginally across years (Fig.

20A; two-way ANOVA; year effect, Fj^ q “ 3.39, p - 0.07; year x date

interaction, F ^ ^ q - 4.23, p - 0.006). During 1987, prey biomass was

greatest during peak macrozooplankton abundance, and decreased with

macrozooplankton density (ANOVA, F4(20 “ 5*27, p - 0.005); however,

during 1988 prey biomass did not differ across sample dates (ANOVA,

F*4 20 *" 1*92, p - 0.15). Peak prey biomass differed marginally across ’ *- years (Fj^g ” 4.46, p - 0.07). Further, proportion of

macrozooplankton in gizzard shad diets (the remainder being rotifers)

decreased during and after the zooplankton crash in both years (Fig.

20B).

Based on these results, we predicted that high peak densities of

gizzard shad would lead to reduced macrozooplankton, which, in turn, would reduce available food, growth, and ultimately, recruitment of

larval bluegill. Using data from 4 years of sampling (1986-1989),

total larval bluegill catch-per-effort (total number of bluegill 101

collected divided by total sanpllng tine) was marginally significantly

related to peak larval gizzard shad density (Fig. 21; “ 13.72, p

- 0 .0 6 6 ).

D iscussion

Larval gizzard shad-zooplankton interactions

Complex interactions among larval gizzard shad, zooplankton,

phytoplankton, and larval bluegill drive reservoir community dynamics

during spring in Kokoslng Lake (Fig. 22). Via an enclosure/exclosure

experiment, we demonstrated that larval gizzard shad dramatically

reduce macrozooplankton, with rotifer decline occurring about a week after the macrozooplankton, in enclosures. In exclosures, densities of all zooplankton remained high. Patterns In the lake followed those

in enclosures suggesting we mimicked lake conditions in this

treatment. Shad In enclosures also reduced edible phytoplankton.

Because phytoplankton occurred in guts of lake fish, then shad most likely controlled zooplankton dynamics in Kokoslng Lake, by direct predation and by exploitative competition through phytoplankton h e rb iv o ry .

Previous work suggests that gizzard shad > 35 mm selectively consume smaller prey items, such as phytoplankton and detritus as they grow (Kutkuhn 1957, Cramer and Marzolf 1970, Baker and Schmitz 1971,

Barger and Kilambi 1980; reviewed by H iller 1960; Bodola 1966).

However, without documentation of zooplankton densities, we can not ascribe these changes in diet to changes in feeding modes. Without macrozooplankton, gizzard shad may be forced to feed on small prey. 102

Two pieces of evidence suggest thee large gizzard shad can feed on macrozooplankton. First, adult gizzard shad modify zooplankton species composition in laboratory experiments, via filter feeding

(Drenner et al. 1978, 1982a,b, Drenner and McComas 1980), removing more easily caught taxa. Second, young-of*year gizzard shad > 50 nun prey on macrozooplankton during summer In Lake Erie, consuming only trace amounts of phytoplankton and detritus (Wilder and Vondracek

1988, Parrish and Vondracek 1989). If macrozooplankton remain abundant, large gizzard shad might continue to be zooplanktlvorous, rather than switching to herblvory and detritivory.

In addition to predatory mortality, birth rates of the dominant cladocerans also decreased during the zooplankton decline, particularly during 1988. When zooplankton become sufficiently abundant,.they overgraze phytoplankton with a commensurate decline in birth rate (Lamport et a l.'1986, reviewed in Sommer et al. 1986).

However, this overgrazing seems unlikely in our experiment. In fact, fecundities in exclosures exceeded those in enclosures, where zooplankton density was lower. Predation, via selective removal of the more visible, egg>bearing individuals by larval gizzard shad, can reduce apparent birth rates (Gllwlcz 1981). Additionally, given that gizzard shad in the lake consumed phytoplankton, falling birth rates

In the bags could be the result of reduced edible phytoplankton, driven to low levels by gizzard shad. Our fecundity data do not allow us to separate these hypotheses. However, it is clear that larval gizzard shad can simultaneously be predators on, and competitors with, zooplankton. Though increased planktlvore abundance typically leads 103 to Improved resources for zooplankton (Drenner et al. 1984, Lazzaro

1987), the duel effects of predation and competition by larval gizzard shad on zooplankton eliminated any indirect positive effects for phytoplankton. Typically, mixed competitlon-predatlon interactions typically Involve ontogenetic changes in Interactions between species

(e.g., redslde shiner, Rlchardsonlus balteatus-rainbow trout, Salmo gardnerl. Larkin and Smith 1954, Johannes and Larkin 1961; largemouth bass•bluegill, Gilliam 1982; perch, Perea fluvlatllls-roach. Rutllus rutllus. Persson 1988). Here, however both competition and predation occur simultaneously. Thus, reduced zooplankton birth rates are responsible, in part, for the zooplankton crash. Larval gizzard shad reduce birth rates and contribute to increased death rates, due to competitive and predatory interactions with zooplankton.

Though peak larval gizzard shad densities in Kokoslng Lake seem high relative to published values (e.g., 20-25 fish/m^, Hayhew 1977;

10 fish/m^, Downey and Toetz 1983; '4 fish/m^, Matthews 1984; 0.5 fish/m^, Tlsa et al. 1985), final densities in the experiment (6.0-

15.5 fish/m^), all of which produced zooplankton crashes, lie within published ranges. Macrozooplankton in "exclosures" with fish densities < 6 fish/m^ did not crash (macrozooplankton densities at experiment's end were 96 organlsms/1 with 5.5 fish/m^, and 493 organisms/1 with 1.5 fish/m^). Consequently, at least in Kokoslng

Lake, gizzard shad densities must exceed this threshold to drive zooplankton to extinction (as for yellow perch, Mills et al. 1987).

Variations in year-class strength of gizzard shad coupled with complex interactions among larval gizzard shad, zooplankton, and phytoplankton 104 yield fluctuations In gizzard shad abundance. These fluctuations around the threshold density necessary for a zooplankton decline, nay lead to system-wide effects and certainly contributes to the tremendous system variability so characteristic of small reservoirs.

Effects _on larval bluegill

Because only few larval bluegill could be collected in the limnetic zone before and during the macrozooplankton crash, we could not assess whether they consumed less food and grew slowly in response to depressed zooplankton resources. As documented for gizzard shad, bluegill probably consumed less, grew more slowly, and ultimately suffered high mortality with shad. Given that appropriate-sized prey of a specific minimum density are required for larval survival (for examples, see Lasker 1975, Noble 1975, Werner and Blaxter 1980, Mills and Forney 1981), exploitative competition by gizzard shad may have reduced prey to below the threshold required by bluegill.

Furthermore, slow growth rates could increase the period which larvae are vulnerable to gape-limited predators (Rice et al. 1987a,b). In either case, the observed inverse relation between bluegill and gizzard shad abundance probably depends on relative spawning times.

If bluegill spawned 2-4 weeks earlier, their presence in the limnetic would occur during peak macrozooplankton abundance, perhaps leading to coexistence with gizzard shad. Conversely, if gizzard shad spawned 2-

4 weeks earlier, they would arrive in the limnetic before macrozooplankton peaked, possibly reducing gizzard shad success, as well as their impact on zooplankton and larval bluegill. Larval gizzard shad density, per capita consumption by gizzard shad and 105

bluegill, peak macrozooplankton density, spawning times, and the

minimum zooplankton density required for survival by bluegill and

gizzard shad all interact to determine year-class strengths of gizzard

shad and bluegill.

Despite the dramatic Impact of larval gizzard shad on

zooplankton, and perhaps ultimately on bluegill recruitment, bluegill

and gizzard shad diet overlap was not high. During 1987 overlap values increased during May through June as macrozooplankton became

sparse and rotlfors became the major diet item for both species; this pattern did not occur during 1988 when bluegill and gizzard shad co occurred on only three sampling dates. Clearly, these data do not

indicate that.competition Is not important, or that these species have little effect on one another, but rather they simply describe the degree to which larval gizzard shad and bluegill share resources while co-occurring (Abrams 1980, 1982). In fact, because gizzard shad growth in the enclosure/exclosure experiment was density dependent (at densities common in Kokoslng Lake), zooplankton is lim iting and gizzard shad and bluegill likely compete. Mallln et al. (1985) suggest that overlap between bluegill and gizzard shad is greatest when both are small (both feeding on copepod nauplil and rotifers) and decreases with fish size; however, their analysis assumes simultaneous spawning. Because bluegill spawned after gizzard shad in Kokoslng

Lake, gizzard shad always were larger than bluegill and, as such, ate a diversity of prey items while first-feeding bluegill were restricted to copepod nauplil and rotifers. Thus, gizzard shad appear to influence bluegill via exploitative competition, driving 106

macrozooplankton to extinction before bluegill arrive In the limnetic.

Competition between gizzard shad and bluegill occurs even though

adult diets do not typically overlap; gizzard shad eat detritus (Dendy

1966, Pahl and Maurer 1962, Baker and Schmitz 1971, Mundahl and

Wlsslng 1987) whereas bluegill eat zooplankton and littoral

Invertebrates.(Keast 1978, Mlttelbach 1984). In fact, coexistence

between these species may be facilitated by the storage effect

(Chesson and Warner 1981), because adult bluegill persist during those

years when larval gizzard shad density is high and bluegill

recruitment Is low. However, interactions between young of these

species have dramatic consequences for coexistence. Similar

ontogenetic shifts occur In other species (Werner and Gilliam 1984).

For example, adult bluegill and pumpklnseed (Lepomls glbbosus) feed on

different resources. However, because their juveniles are restricted

to the littoral zone by risk of predation, they compete in this

protective habitat (Mlttelbach and Chesson 1987, Mlttelbach 1988).

Recruitment bottlenecks also can occur In Chaoborus. where competition between Daphnla and early instars of Chaoborus lim its recruitment of

Chaoborus to older instars that prey upon Daphnla (Neill 1988). As

larvae are Included In our study of interactions among fishes, more complex Interactions such as these will be discovered.

Historically abiotic factors have been emphasized relative to control of recruitment in fishes (Busch et al. 1975, Clady and

Hutchinson 1975, Clady 1976, Beam 1983). However, recent work suggests that biotic factors contribute substantially to recruitment variability (Lasker 1975, Forney 1976, Werner and Blaxter 1980, Rice 107

at al. 1987a,b, Post and Prankaviclus 1987). In Kokoslng Lake 87Z of

the year-to-year variability in bluegill recruitment was explained by

variation in a biotic factor: peak larval gizzard shad density. In

reality, several biotic mechanisms operate simultaneously to influence bluegill recruitment. Larval gizzard shad prey upon zooplankton,

competing directly with larval bluegill. In addition, larval gizzard

shad compete with zooplankton by consuming edible phytoplankton,

similarly Influencing bluegill by further reducing macrozooplankton.

Thus, though larval gizzard shad and bluegill rarely co-occur in the

limnetic and adult diets do not overlap, gizzard shad still strongly reduce bluegill recruitment.

The consequences of Interaction between larval gizzard shad and bluegill extend beyond bluegill recruitment, particularly because both gizzard shad and bluegill are forage species for piscivores. Shad recruitment varies independently of adult stock (Stock 1971), producing, we would argue, annual variation in zooplankton abundance, shad growth rates, and bluegill recruitment. Two factors that influence the susceptibility of prey to piscivores, bluegill recruitment and gizzard shad growth rates, vary with shad density, producing annual fluctuations in availability of prey to young-of- year largemouth bass. Because overwinter survival of young-of-year largemouth bass depends on body size and fat reserves in the fall

(Adams et al. 1982), variations in prey availability ultimately will affect predator recruitment (see example in DeVries 1989). Variation in recruitment alters adult population size, which, in turn, because fish can be long-lived, will influence the aquatic community for many 108

years, even decades in the £uture (Kitchell et al. 1988).

Implications for blomanioulatlon

Gizzard shad often dominate the forage fish biomass in lakes

within the United States (Jenkins 1957, 1967, Noble 1961, Wydoski and

Bennett 1981; recently reviewed in DeVries and Stein 1990a). Fishery

managers have attempted to remove shad from many systems, thus

reducing the abundance of a presumed sport*fish competitor (DeVries

and Stein 1990a). Based on predictions of the cascading trophic

interactions/biomanlpulation hypothesis (Shapiro and Uright 1984,

Carpenter et al. 1985, 1987), gizzard shad reduction (perhaps due to

predator addition) should enhance zooplankton, reduce phytoplankton,

and extend the typical "clear-water" phase (Sommer et al. 1986).

However, if trophic cascadlng/biomanlpulation is to work, linkages

between adjacent trophic levels must be both direct and strong (Paine

1980, Kerfoot 1987). Because larval gizzard shad prey on and compete with zooplankton, they "short-circuit" the herbivore-phytoplankton

linkage, simultaneously acting as both planktlvore and herbivore.

Additionally, because gizzard shad have extremely high fecundities * (approaching '500,000 eggs/female, Bodola 1966, Wilder and Vondracek

1988, Parrish and Vondracek 1989), few adults can produce a large year class (e.g., Smith 1958, Dietz and Jurgens 1963, Brummett 1983).

Dense populations of larvae can potentially negate increased zooplankton density due to adult removal by piscivorous predators.

Thus, clear predictions of the effects of shad removal on water quality are not possible. If we are to understand the effects that shad have on water quality, we must quantify their effects on all 109 * trophic levels. Though ouch effort Is required to quantify responses of all trophic levels, our ability to generate testable predictions from these complex interactions depends on oiir understanding of linkages between trophic levels.

Conclusions

Though gizzard shad < 25 mm may constitute a small fraction of the population biomass (< 0.01X, Crinstead et al. 1976), their impact on reservoir community structure is both impressive and complex.

Larval gizzard shad effects extend to zooplankton (via competition and predation) and phytoplankton (herbivory). Their effects also extend up the food web to other forage fishes (i.e., bluegill, via * competition) and ultimately to largemouth bass (DeVries 1989). By quantifying both direct and indirect effects of larval gizzard shad on the food web, we can quantify the relative strengths of important linkages, ultimately determining the role of larval fish in structuring aquatic communities. 110

Literature Cited

Abrams, P. 1980. Some comments on measuring niche overlap. Ecology 61:44-49.

Abrams, P. 1982. Reply to a comment by Hurlbert. Ecology 63:253- 254.

Adams, S. M., R. B. McLean, and M. M. Huffman. 1982. S tru c tu rin g o f a predator population through temperature-mediated effects on prey availability. Canadian Journal of Fisheries and Aquatic Sciences 39:1175-1184.

Applegate, R. L. and J. W. Mullan. 1967. Food of young largemouth b a ss, MlSHPP.tqrva salraoldes. In a new and o ld re s e rv o ir. Transactions of the American Fisheries Society 96:74-77.

B ailey, R. M. and C. R. Robins. 1988. Changes In North American fis h names, especially as related to the International Code of Zoological Nomenclature, 1985, Bulletin of Zoological Nomenclature 45:92-^03.

Baker, C. D. and E. H. Schmitz. 1971. Food habits of adult gizzard and threadfin shad In two Ozark reservoirs. Pages 3-11 in G. E. Hall, editor. Reservoir Fisheries and Limnology. Special P u b lica tio n Number 8 , American Fisheries Society, Washington, D.C., USA.

Barger, L. E., and R. V. Kllambl. 1980. Feeding ecology of larval shad, Dorosoma. In Beaver reservoir, Arkansas. Pages 136-145 in L. A. Fulman, editor. Proceedings of the Fourth Annual Larval Fish Conference. United States Fish and Wildlife Service, Ann Arbor, Michigan, USA.

Beam, J. H, 1983. The effect of annual water level management on population trends of white crapple in Elk City Reservoir, Kansas. North American Journal of Fisheries Management 3:34-40.

Bengtsson, J. 1987. Competitive dominance among Cladocera: are single-factor explanations enough? Hydroblologla 145:245-257.

' Bodola, A. 1966. Life history of the gizzard shad, Dorosoma ceoedianum (LeSueur) in western Lake Erie. U.S. Fish and W ildlife Service Fishery Bulletin 65:391-425.

Brooks, J. L. and S. I. Dodson. 1965. Predation, body size and composition of the plankton. Science 150:28-35.

Brummett, K. 1983. Results of a selective renovation using fintrol on Deer Ridge Community Lake, M issouri. Pages 168-174 i n D. Bonneau and G. Radonskl, editors. Pros and cons of shad. Iowa Conservation Commission, Des Moines, Iowa, USA. I l l

Busch, W. D. N., R. L. Scholl, and W. L. Hartman. 1975. Environmental factors affecting the strength of walleye (Stlzostedlon vltreum vltreua) year classes on western Lake Erie, 1960-70. Journal of the Fisheries Research Board of Canada 32:1733-1743.

Carpenter, S. R. 1988. Complex interactions in lake communities. Sprlnger-Verlag, New York, New York, USA.

Carpenter, S. R., J. F. Kitchell, and J. R. Hodgson. 1985. Cascading trophic Interactions and lake productivity. Bioscience 35:634-639.

Carpenter, S. R., J. F. Kitchell, J. R. Hodgson, P. A, Cochran, J. J. Elser, M. M. Elser, D. M. Lodge, D. Kretchmer, X. He, and C. N. von Ende. 19B7. Regulation of lake primary productivity by food web structure. Ecology 68:1863-1876.

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

Chesson, J. 1983. The estimation and analysis of preference and its relationship to foraging models. Ecology 64:1297-1304.

Chesson, P. L. and R. R. Warner. 1981. Environmental variability promotes coexistence in lottery competitive systems. American Naturalist 117:923-943.

Clady, M. D. 1976, Influence of temperature and wind on the survival of early stages of yellow perch, Perea flavescens. Journal of the Fisheries Research Board of Canada 33:1887-1893.

Clady, M. D. and B. Hutchinson. 1975. Effect of high winds on eggs of yellow perch, Perea flavescens. in Oneida Lake, New York. Transactions of the American Fisheries Society 104:524-525.

Confer, J. M. and L. M. O'Bryan. 1989. Changes in prey rank and preference by young planktlvores for short-term and long-term ingestion periods. Canadian Journal of Fisheries and Aquatic Sciences 46:1026-1032.

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

Cryer, M., G. Peirson, and C. R. Townsend. 1986. Reciprocal interactions between roach, Rutilus rutllus. and zooplankton in a small lake: prey dynamics and fish growth and recruitment. Limnology and Oceanography 31:1022-1038.

Cushing, D. H. 1975. Marine ecology and fisheries. Cambridge University Press, Cambridge, England. 112

Dendy, J. S. 1946. Food of several species of fish, Norris R eserv o ir, Tennessee. Jo u rn a l o f Che Tennessee Academy o f Science 21:105-127.

DeVries, D. R. 1989. Complex interactions in aquatic communities: trophic-level interactions, ontogenetic niche shifts, and the role of an open-water planktivore. Doctoral Dissertation, The Ohio State University, Columbus, Ohio, USA.

DeVries, D. R. and R. A. Stein. 1990. Manipulating shad to enhance sport fisheries in North America: an assessment. North American Journal of Fisheries Management (in press).

DeVries, D. R. and R. A. Stein. 1990b. Comparing three zooplankton samplers; a taxon-specific assessment, submitted to Limnology and Oceanography.

Dietz, E. and K. C. Jurgens. 1963. An evaluation of selective shad control in Medina Lake, Texas. Inland Fisheries Report 5, Parks and W ildlife Department, Austin, Texas, USA.

Dodson, S. I. 1970. Complementary feeding niches sustained by size- selective predation. Limnology and Oceanography 15:131-137.

Dodson, S. I. 1974. Zooplankton competition and predation; an experimental test of the size-efficiency hypothesis. Ecology 55:605-613.

Doolittle, W. L. 1982. Thfe nature and cause of the midsummer decline of litto ral zooplankton in Lake Itasca,'Minnesota. Ph.D. dissertation, University of Tennessee, Knoxville, Tennessee, USA.

Downey, P. and D. Toetz. 1983. Distribution of larval gizzard shad fDorosoma ceoedianum) in Lake Carl Blackwell, Oklahoma. American Midland N aturalist 109:23-33.

Drenner, R. W., deNoyelles, F., Jr., and D. Kettle. 1982b. Selective impact of filter-feeding gizzard shad on zooplankton community structure. Limnology and Oceanography 27:965-968.

Drenner, R. U. and S. R. McComas. 1980. The roles of zooplankter escape ability and fish size selectivity in the selective feeding and impact of planktivorous fish. Pages 587-593 J j q W. C. Kerfoot, editor. Evolution and ecology of zooplankton communities. The University Press of New England, Hanover, New Hampshire, USA.

Drenner, R. W., J. R. Hummert, F. DeNoyelles, Jr., and D. Kettle. 1984. Selective particle Ingestion by a filter-feeding fish and its impact on phytoplankton community structure. Limnology and Oceanography 29:941-948. 113

Drenner, R. W., W. J . O 'B rien, and J . R. Mummert. 1982a. Filter*feeding rates of gizzard shad. Transactions of the American Fisheries Society 111:210-215.

Drenner, R. W., J. R. Strickler, and W. J. O'Brien. 1978. Capture probability: the role of zooplankter escape in the selective feeding of planktlvorous fish. Journal of the Fisheries Research Board of Canada 35:1370-1373.

Drenner, R. tf., S. T. Threlkeld, and M. D. McCracken. 1986. Experimental analysis of the direct and indirect effects of an omnivorous filter-feeding clupeid on plankton community structure. Canadian Journal of Fisheries and Aquatic Sciences 43:1935-1945.

Ebenman, B. and L. Persson, 1988. Size-structured populations. Springer-Verlag, New York, New York, USA.

Evans, M. S. 1986. Recent major declines in zooplankton populations in the inshore region of Lake Michigan: probable causes and Implications. Canadian Journal of Fisheries and Aquatic Sciences 43:154-159.

Forney, J. J. 1976. Year-class formation in the walleye (Stlzostedlon vltreum) population of Oneida Lake, New York, 1966- 1973. Journal of the Fisheries Research Board of Canada 33:783- 792.

Gilliam, J. F. 1982. Foraging under mortality risk in size-structured populations. PhD dissertation, Michigan State University, East Lansing, Michigan, USA.

Gllwicz, Z. M. 1981. Food and predation in limiting clutch size of cladocerans. Internationale Vereinigung fuer Theoretische und Angewandte Llmnologle Verhandlungen 21:1562-1566. + Gllwicz, Z. M. and J. Pijanowska. 1989. The role of predation in zooplankton su ccessio n . Pages 253-296 in U. Sommer, e d ito r. Plankton ecology: succession in plankton communities. Springer- Verlag, New York, New York, USA.

Greene, C. H. 1985. Planktivore functional groups and patterns of prey selection in pelagic communities. Journal of Plankton Research 7:35-40.

Gregory, R. S. and P. M. Powles. 1985. Chronology, distribution, and sizes of larval fish sampled by light traps in macrophytic Chemung Lake. Canadian Journal of Zoology 63:2569-2577. 114

G rinstead, B. G ., R. M. Gennings, G. R. Hooper, C. A. S ch u ltz, and D. A. Vhorton. 1976. Estimation of standing crop of fishes in the predator-stocklng-evaluatlon reservoirs. Proceedings of the Annual Conference of the Southeastern Association of Fish and W ildlife Agencies 30:120-130.

Hall, D. J., S. T. Threlkeld, C. W. Burns, and P. H. Crowley. 1976. The slze-efflclency hypothesis and the size structure of zooplankton communities. Annual Review of Ecology and Systematlcs 7:177-208.

Haney, J. F. and D. J. Hall. 1973. Sugar-coated Daphnia: a preservation technique for Cladocera. Limnology and Oceanography 18:331-333.

Hansen, H. J. and D. H. Wahl. 1981. Selection of small Daphnia pulex by yellow perch in Oneida Lake, New York. Transactions of the American Fisheries Society 110:64-71.

Jenkins, R. H. 1957. The effect of gizzard shad on the fish of a small Oklahoma lake. Transactions of the American Fisheries Society 85:58-74.

Jenkins, R. M. 1967. The influence of some environmental factors on standing crop and harvest of fishes in U.S. reservoirs. Pages 298-321 In The American Fisheries Society, editor. Reservoir fishery resources symposium. American Fisheries Society, Bethesda, Maryland, USA.

Johannes, R. E. and P. A. Larkin. 1961. Competition for food between redslde shiners fRlchardsonius balteatus) and rainbow trout fSalrao galrdneri) in two British Columbia lakes. Journal of the Fisheries Research Board of Canada 18:203-220.

Johnson, B. H., R, A. Stein, and R. F. Carline. 1988. Use of quadrat rotenone technique and bioenergetlcs modeling to evaluate prey availability to stocked piscivores. Transactions of the American Fisheries Society 117:127-141.

Keast, A. 1978. Feeding interrelationships between age groups of puapklnseed sunflsh (Leoomls glbbosusl and with the bluegill aunfish (L. macrochlrusl . Journal of the Fisheries Research Board of Canada 35:12-27.

Keast, A. 1980. Food and feeding relationships of young fish in the first weeks after the beginning of exogenous feeding in Lake 'Opinlcon, Ontario. Environmental Biology of Fishes 5:305-314.

Kerfoot, W. C. 1987. Cascading effects and indirect pathways. Pages 57-70 in W, C. Kerfoot and A. Sih, editors. Predation: direct and indirect impacts on aquatic communities. University Press of New England, Hanover, New Hampshire, USA. 115

Kerfoot, V. C. and A. Sih. 1987. Predation: direct and Indirect Impacts on aquatic communities. University Press of New England, Hanover, New Hampshire, USA.

Kltchell, J. F., S. M. Bartell, S. R. Carpenter, D. J. Hall, D. J. McQueen, W. E. N e ill, D. S cavla, and E. E. Werner. 1988. Eplstemology, experimentation, and pragmatism. Pages 263-280 in S. R. Carpenter, editor. Complex Interactions In lake communities. Springer-Verlag, Berlin, Germany.

Kutkuhn, J. H. 1957. Utilization of plankton by Juvenile gizzard shad In a shallow prairie lake. Transactions of the American Fisheries Society 87:80-103.

Lamport, W., W. Fleckner, H. Ral, and B. E. Taylor. 1986. Phytoplankton control by grazing zooplankton: a study on the spring clear-water phase. Limnology and Oceanography 31:478-490.

Larkin, P. A. and S. B. Smith. 1954. Some effects of Introduction of the redslde shiner on the Kamloops trout In Paul Lake, British Columbia. Transactions of the American Fisheries Society 83:161-175.

Lasker, R. 1975. Field criteria for survival of anchovy larvae: the relation between Inshore chlorophyll maximum layers and successful first feeding. Fishery Bulletin 73:453-462.

Lazzaro, X. 1987. A review of planktlvorous fishes: their evolution, feeding behaviours, selectlvltles, and Impacts. Hydroblologla 146:97-167.

Lynch, M. 1979. Predation, competition, and zooplankton community structure: an experimental study. Limnology and Oceanography 24:253-272.

Mallin, M. A., L. J. Birchfield, and W. Warren-Hlcks. 1985. Food habits and diet overlap of larval Leoomls spp. and gizzard shad in a Piedmont reservoir. Proceedings of the Annual Conference of the Southeastern Association of Fish and W ildlife Agencies 39:146-155.

Mayhew, J. 1977. The effects of flood management regimes on larval fish and fish food organisms at Lake Rathbun. Iowa Fisheries Research T echnical S eries Number 77-2, Iowa C onservation Commission, Des Moines, Iowa, USA.

Matthews, W. J. 1984. Influence of turbid inflows on vertical distribution of larval shad and freshwater drum. Transactions of the American Fisheries Society 113:192-198. 116

McQueen, D. J . ( J . R. P o st, end E. L. M ills. 1986. tro p h ic relationships in freshwater pelagic ecosystems. Canadian Journal of Fisheries and Aquatic Sciences 43:1571*1581.

Miller, R. R. 1960. Systematica and biology of the gizzard shad (Dorosoma ceoedlanum) and related fishes. Fishery Bulletin 173:371-392.

Mills, E. L. and J. L. Forney. 1981. Energetics, food consumption, and growth of young yellow perch in Oneida Lake, New York. Transactions of the American Fisheries Society 110:479-488.

Mills, E. L. and J. L. Forney. 1988. Trophic dynamics and development of pelagic food webs. Pages 11-30 in S. R. Carpenter, editor. Complex interactions in lake communities. Springer-Verlag, Berlin, Germany.

Mills, E. L., J. L. Forney, and K. J. Vagner. 1987. Fish predation and its cascading effect on the Oneida Lake food chain. Pages 118-131 in V. C. Kerfoot and A. Sih, editors. Predation: direct and indirect Impacts on aquatic communities. University Press of New England, Hanover, New Hampshire, USA.

Mittelbach, G. G. 1981. Foraging efficiency and body size: a study of optimal diet and habitat use by bluegills. Ecology 62:1370- 1386.

Mittelbach, G. G. 1984. Predation and resource partitioning in two sunflshes (Centrarchldae). Ecology 65:499-513.

Mittelbach, G. G. 1988. Competition among refuging sunflshes and effects of fish density on littoral zone invertebrates. Ecology 69:614-623.

Mittelbach, G. G. and P. L. Chesson. 1987. Predation risk: indirect effects on fish populations. Pages 315-332 In V. C, Kerfoot and A. Sih, editors. Predation: direct and indirect impacts on aquatic communities. University Press of New England, Hanover, New Hampshire, USA,

Mundahl, N. D. and T. E. Visaing. 1987. Nutritional importance of detritivory in the growth and condition of gizzard shad in an Ohio reservoir. Environmental Biology of Fishes 20:129-142.

N e ill, V. E. 1988. Community responses to experim ental n u trie n t perturbations in oligotrophic lakes: the importance of bottlenecks in size-structured populations. Pages 236-255 In B. Ebenman and L. Persson, e d ito rs . S iz e -s tru c tu re d in te ra c tio n s . Springer-Verlag, Berlin. 117

Noble, R. I. 1975. Growth of young yellow perch (Efiififl flavescens) In relation to zooplankton populations. Transactions of the American Fisheries Society 104:731-741.

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

Northcote, T. G. 1988. Fish In the structure and function of freshwater ecosystems: a "top-down" view. Canadian Journal of Fisheries and Aquatic Sciences 45:361-379.

Fahl, B. G. and J. Maurer. 1962. Food habits of the young gizzard shad. Proceedings of the Minnesota Academy of Science 31:20-21.

Paine, R, T, 1980, Food webs, linkage interaction strength, and community Infrastructure. Journal of Animal Ecology 49:667-685,

Parrish, D. L. and B. Vondracek. 1989. Population dynamics and ecology of Lake Erie gizzard shad. Project F-61-R-2 Annual Performance Report, Ohio Department of Natural Resources, Columbus, Ohio, USA.

Person, L, 1988. Asymmetries in competitive and predatory interactions in fish populations. Pages 203-218 in B. Ebenman and L. Persson, editors. Size-structured populations. Springer-Verlag, Berlin.

Prophet, C. W. 1982. Zooplankton changes in a Kansas lake 1963-1981. Journal of Freshwater Ecology 1:569-575.

Prophet, C. W. 1985. Calanoid population structure in a Kansas lake after introduction of threadfin shad. The Southwestern N aturalist 30:162-163.

Post, J. R. and A, B. Prankeviclus. 1987. Size-selective mortality in young-of-the-year yellow perch (Perea flavescens): evidence from otolith microstructure. Canadian Journal of Fisheries and Aquatic Sciences 44:1840-1847. ,

Prophet, C. W. 1988. Changes in seasonal population structures of two species of Diaptomus (Calanoida, Copepoda) subsequent to introductions of threadfin and gizzard shad. The Southwestern Naturalist 33:41-53.

Rajasllta, M. and I. Vuorlnen. 1983, A field study of prey selection in planktivorous fish larvae. Oecologia 59:65-68. 118

Rice, J. A., L. B. Crowder, and M. E. Holey. 1987a. Exploration of mechanisms regulating larval survival In Lake Michigan bloater: a recruitment analysis based on characteristics of individual larvae. Transactions of the American Fisheries Society 116:703-718.

Rice, J. A., L. B. Crowder, and F. P. Binkowski. 1987b. Evaluating potential sources of mortality for larval bloater fCoregonus hovll : starvation and vulnerability to predation. Canadian Journal of Fisheries and Aquatic Sciences 44:467-472.

Rosenthal, H. and G. Hempel. 1970. Experimental studies in feeding and food requirements of herring larvae fCluoea harencus L.). Pages 344-364 In J. H. Steel, editor. Marine food chains. University of California Press, Berkeley, California, USA.

SAS Institute Inc. 1985. SAS Users Guide: Statistics. Version 5 ed. SAS Institute Inc., Cary, North Carolina, USA.

Schoener, T. W. 1970. Non-synchronous spatial overlap of lizards In patchy habitats. Ecology 51:408-418.

Shapiro, J. and D. 1. Wright. 1984. Lako restoration by blomanlpulatlon: Round Lake, Minnesota, the first two years. Freshwater Biology 14:371-383.

Smith, D. W. and S. D. Cooper. 1982. Competition among Cladocera, Ecology 63:1004-1015.'

Smith, W. A., Jr. 1958. Shad management in reservoirs. Proceedings of the Annual Conference of the Southeastern Association of Game and Fish Commissioners 12:143-147.

Sommer, U., Z. M. Gllwicz, W. Lamport, and A. Duncan. 1986. The PEG model of seasonal succession of planktonlc events in fresh waters. Archive fur Hydroblologie 106:433-471.

Stock, J. N. 1971. A study of some effects of population density on gizzard shad and bluegill growth and recruitment in ponds. Masters thesis. University of Missouri, Columbia, Missouri, USA.

Threlkeld, S. T. 1987. Experimental evaluation of trophic-cascade and nutrient-mediated effects of planktlvorous fish on plankton community structure. Pages 161-173 In W. C. Kerfoot and A. Sih, editors. Predation: direct and Indirect impacts on aquatic communities. University Press of New England, Hanover, New Hampshire, USA.

Tlsa, M, S., J. J. Ney, and D. K. Whitehurst. 1985. Spatial and temporal distribution of larval alewives and gizzard shad in a Virginia reservoir. Proceedings of the Annual Conference of the Southeastern Association of Fish and W ildlife Agencies 39:65-73. 119

Wallace, R. K., Jr. 1981. An assessment of dlet-overlap indexes. Transactions of the American Fisheries Society 110:72-76.

Ware, D. H. 1980. Bioenergetics of stock and recruitment. Canadian Journal of Fisheries and Aquatic Sciences 37:1012-1026.

Warshaw, S. J. 1972. Effects of alewives (Alosa PBeudoharengus) on the zooplankton of Lake Wononskopomuc, Connecticut. Limnology and Oceanography 17:816-825.

Wells, L. 1970. Effects of alewife predation on zooplankton populations in Lake Hlchigan. Limnology and Oceanography 15:556- 565.

Werner, E. E., and J. F. Gilliam. 1984. The ontogenetic niche and species Interactions in size-structured populations. Annual Review of Ecology and Systematica 15:393-425.

Werner, E. E., J. F. Gilliam, D. J. Hall, & G. G. Mittelbach. 1983. An experimental test of the effects of predation risk on habitat use In fish. Ecology 64:1540-1548.

Werner, R. G. 1969. Ecology of limnetic bluegill (Lepomla macrochlrus) fry in Crane Lake, Indiana. American Midland Naturalist 81:164-181.

Werner, R. G. and J. H. S. Blaxter. 1980. Growth and survival of larval herring fCluoea harangue} in relation to prey density. Canadian Journal of Fisheries and Aquatic Sciences 37:1063-1069.

Whiteside, M. C. 1989. 0+ fish as major factors affecting abundance patterns of littoral zooplankton. Internationale Vereinlgung fuer Theoretische und Angewandte Limnologle Verhandlungen 23:1710-1714.

Whiteside, M. C., W. L. Doolittle, and C. M. Swindoll. 1985a. Zooplankton as food resource for larval fish. Internationale Vereinlgung fuer Theoretische und Angewandte Limnologle Verhandlungen 22:2523-2526.

Whiteside, M. C,, C. M. Swindoll, and W. L. Doolittle. 1985b. Factors affecting the early life history of yellow perch, Perea flavescens. Environmental Biology of Fishes 12:47-56.

Wilder, I. B. and B. Vondracek. 1988. Population dynamics and ecology of Lake Erie gizzard shad. Project F-61-R-1 Annual 'Performance Report, Ohio Department of Natural Resources, Columbus, Ohio, USA. 120

Wong, B. and F. J. Ward. 1972. Size selection of Daohnla nullcarla by yellow perch (Efilfifl flavescena) fry In West Blue Lake, Manitoba. Journal of the Fisheries Research Board of Canada 29:1761-1764.

Wong, C. K. 1981. Predatory feeding behavior of Eolschura lacustris (Copepoda: calanolda) and prey defense. Canadian Journal of Fisheries and Aquatic Sciences 38:275-279.

Wydoskl, R. S., and D. H. Bennett 1981. Forage species In lakes and reservoirs of the western United States. Transactions of the American Fisheries Society 110:764-771.

Zaret, T. M. 1980. Predation and freshwater communities. Yale University Press, Hew Haven, Connecticut, USA.

Z le b e ll, C. D ., J . C. Tash, and R. L. B a re fie ld . 1986. Impact of threadfin shad on macrocrustacean zooplankton In two Arizona Lakes. Journal of Freshwater Ecology 3:399-406. Table 6. Food selection (using Chesson's alpha, Chesson 1978, 1983) by larval bluegill collected offshore in Kokosing Late during 1987 and 1988. Data are presented as means + 95% oonfiderr^ interval. Values > 0.08-0.13 (the reciprocal of the number of prey items in the late) indicate positive selection; values less than th is indicate avoidance. Bluegill > 13.0 mn were not collected during 1987. NP = zooplankton taxon not present in the late at the same time as fish of the indicated size. Sample sizes indicated in parentheses. size range calanoid cyclcpoid copepod -iPB.Tfr) . Bosroina OeriodaphniaDaphnia Diaphanosqna ocpepods ocpepods nauplii rotifers

(1987)

4.0- 6.9 (8) 0 0 0 0 0.12+0.29 0 0.11+0.26 0.77+0.36

7.0- 9.9 (3) 0.63+1.35 0 0 0 0 0 0 0.37+1.35

10.0-12.9 (3) 0 0 HP 0 .5 8 + 1 .2 8 0 0.12+0.52 0 0.09+0.39

(1988) r

4.0- 6.9 (16) 0.06+0.13 0 0 0 0.88+0.18 0.06+0.13

7.0- 9.9 (13) 0.46+0.31 0 0.08+0.18 0.50+0.57 0 0 0.08+0.17 0.15+0.23

10.0-12.9 (14) 0.27+0.22 0 0 0.53+0.28 0 .0 3 + 0 .0 5 0 0.06+0.12 0.07+0.07

> 13.0 (6) 0 0 0.16+0.41 0.5610.28 0 .2 8 + 0 .1 7 0 121 Table 7. Food selection (using Chesson's alpha, Chesson 1978, 1983) by larval gizzard chart collected offshore in Kotosing late during 1987 and 1988. Data are presented as means + 95% omfidgnr** interval. Values > 0.10-0.14 (the reciprocal of the nunfcer of prey items in the late) indicate positive selection; values less than this indicate avoidance. Gizzard shad < 10 mn did not contain food 1987. NP = zooplankton taxcn not present with fish of the indicated size. Sample sizes indicated in parentheses.

size range calanoid cyclcpoid ccpepod (wn T U i Boanina Oeriodaphnia Dachnia Piaphanosoma oooepods ocpepods nauplii rotifers

(1987)

10.0-12.9 (7) 0.14+0.35 0 0.01+0.03 0 0.13+0.27 0.51±Q.41 0.05+0.07 0.15r+0.35

> 13.0 (25) 0.21±0.14 0 0.03+0.03 0.02+0.04 0.02+0.02 0.31+0.16 0.01+0.01 0.37+0.18

(1988)

4.0- 6.9 (2) 0 NP 0 NP 0 0 0.05+0.70 0.95+0.70

7.0- 9.9 (5) 0 NP 0 NP 0 0.41+0*37 0.52+0.40 0.07+0.13

10.0-12.9 (13) 0.00+0.01 0 0.07+0.12 NP 0 0.47+0.17 0.36+0.14 0.10+0.17

> 13.0 (38) 0.18+0.10 0 0.03+0.04 0 0 0.25+0.12 0.18+0.09 0.37+0.14

toISO Figure 11. Abundance of larval gizzard shad (fish/nr*, A, D), macrozooplankton (number/1, B, E), and rotifers (number/1, C, F) in

Kokosing Lake, Ohio during 1987 and 1988. Larval fish data are means

± 1 SE from two replicate surface tows each week; zooplankton data are means ± 1 SE from three replicate Integrated tube samples each week.

Note different Y-axis scales among panels.

123 1987 1988 100

zoo

200 too ISO

too 200

MAY JUNE JULY AUG SEPT IKY JUNE JULY AUG SEPT OCT

Figure 11 Figure 12. Mean fecundity of Bosmina (open squares) and Daphnia (open circles) in Kokosing Lake during the 1987 (A) and 1988 (B) zooplankton d e clin es (11 May- 10 Jun? 1987 and 17 Hay- 27 June 1988; see Fig,

11B,E). Mean fecundity was the product of the proportion of

Individuals carrying eggs (n - 50) and the mean number of eggs per individual carrying eggs (n - 20). Data are means ± 1 SE from three replicate samples.

125 iue 12 Figure MEAN FECUNDITY (eg gs/individual) 6 23 16 MAY 0 1 20 13 6 30 omn □—□ — □ Bosmina ahi o—o — o Daphnia \ UNE N JU . 8 8 9 1 B. A. 1987

126 Figure 13. Final sizo (nun total length) of gizzard shad (GS) from the

1988 bag experiment in Kokoslng Lake. Solid circles represent data from enclosures (n - 6 ) and open circles are from "exclosures" (n - 2 o f 6 exclosures) that had fish at the end of the experiment.

127 ^ ^ u I-* o n e* ♦*1 C

FINAL GS TOTAL LENGTH ( m E 70 70 50 50 30 60 40 40 5 0 15 10 5 0 IA G DNIY fsh/m h (fis DENSITY GS FINAL exclosures enclosures 4 7 . 0 = 2 r = 59.87-1.6X 67X . 1 - 7 8 . 9 5 = Y = 0.006 0 0 . 0 = p 3 ) Figure 14. Denslcy (number/1) of macrozooplankcon (A), size (mm) of macrozooplankcon (B), and densicy (number/1) of rotifers (C) in enclosures (solid circles) and exclosures (open circles) from Che 1988 bag experiment in Kokoslng Lake. Daca are means ± 1 SE from six replicates.

129 130

^ 750 WITHOUT SHAD

500'

250- WITH SHAD

O Ld 0 .2 N M

400'

f= 200-

Figure 14 Figure IS. Size (nun) of six raacrozooplankton taxa in enclosures

(solid lines) and exclosures (dotted lines) during the 1988 bag experiment In Kokoslng Lake. Data are means ± 1 SE from six replicates. Note that the horizontal lines do not represent zero values along the Y-axis.

131 Figure E E

ZOOPLANKTON SIZE ( 0.19 o.io 0.3 0.6 0.6 0.B 0.2 0.4 0.9 0.7 0.3 0.9 1 Ceriodaphnia Bosmina cyclopoids 6 UE JULY JUNE 19 calanoids EXCLOSURES ENCLOSURES 22 nauplii 0 2 Daphnia 132 6 Figure 16. Mean fecundity o f Bosmlna (A) and Daohnla (B) In enclosures (solid symbols) and exclosures (open symbols) during the

1988 experiment In Kokoslng Lake. Mean fecundity Is the product of the proportion of individuals carrying eggs and the mean number of eggs per individual carrying eggs. Data are means ± 1 SE from six replicates.

133 134

A. Bosmina o 3 WITHOUTr SHAD / ?« T 3 t c 0.2 + ^^iVwiTH SHAD/l CO cr» a > o.o-

B. Dophnia 1-5 * Q :z Z) 1.0- O LU *-*- 0.5 +

8 15 22 29 JUNE JULY

Figure 16 Figure 17. Blovolurae (ml/m^) of edible and inedible phytoplankton In enclosures (solid bars) and exclosures (open bars) ac the beginning (7

June 1988) and middle (21 June 1988) of the bag experiment in Kokoslng

Lake. Data are means ± 1 SC from three replicates.

135 ENCLOSURE EXCLOSURE □

edible phytoplankton inedible phytoplankton

JUNE 7 JUNE 21 Figure 18. Density (fish/m^) of larval bluegill (solid line, solid diamond) and larval gizzard shad (dashed line, open triangle) for

1986-1989. Note scale differences for the Y-axes among panels.

137 100 • T: 1 A. 1986 C r o 1 C. 1988 II 73 0.13 15 /I 1

; l 1i 50 *2 Q0.10 r 1 0 Q ||| 1 1 \ 1 1 1 \ 25 N gj 0 05 I 5 1 1 \ O , . h 4 l 0.00 .1 \ ± A f L - ....

>100 B. 1987 D. 1989

yi 0.2 GIZZARD SHAD/m3 GIZZARD SHAD/mr5 GIZZARD

MAY JUNE JULY AUG SEPT MAY JUNE JULY AUG SEPT

U> 00 Figure 19. Density of larval bluegill (flsh/m^, solid diamond) and macrozooplankcon (number/1, open diamond) during 1987 (A) and 1988 (B) in Kokoslng Lake. Data are means ± 1 SE from two replicate -larval fish tows from three replicate Integrated tube samples. Note scale differences for the Y*axes between panels.

139

0.02-- o r iue 19 Figure 0.06' • 0.06'

BLUEGILL/m ■ 0.04- 0.00 0 " 5 .1 0 05" 5 .0 0 0.00 . 10 " A JN JL AG SEPT AUG JULY JUNE MAY . 1987 A. . 1988 B. HO I Figure 20. Mean biomass (mg dry weight of prey/flsh) of prey (A) and proportion (by numbers) of macrozooplankcon (B) In tha diets of 15-mm gizzard shad (n - 5 fish/date) before, during, and after the 1987

(open circles) and 1988 (open squares) zooplankton declines in

Kokoslng Lake.

141 CO o o PREY/FISH 00 mg o O CD

a» proportion MACROZOOPLANKTON MACROZOOPLANKTON r i\H i/ rlon Vl m

Figure 20 Figure 21. Regression of number of larval bluegill captured per minute of larval fish tow (total number of bluegill caught per total sampling time) on the peak larval gizzard shad density (fish/m^) and during 1986-1989 in Kokoslng Lake.

143 00 o r o e

LARVAL BLUEGILL CPUE 50 .5 0 0 1 . . 0 0 5 2 + -- -- EK AVL IZR SA DENSITY SHAD LARVAL GIZZARD PEAK #/ ) /m (# + 1. 2 8 .2 1 + X 6 1 1 0 . 0 - 87 .8 0 = 2 r = 0. 7 .0 0 = p 90 Figure 22. Suggested spring food web in Kokosing Lake, incorporating

Interactions among larval fish, zooplankton, and phytoplankton generated from our field sampling and enclosure/exclosure experiment.

145 9VI

LARVAL ROTIFERS GIZZARD SHAD EDIBLE PHYTOPLANKTON MACRO- LARVAL BLUEGILL ZOOPLANKTON

Figure 22 CHAPTER IV

HABITAT USE BY BLUEGILL IN LABORATORY POOLS: WHERE IS THE REFUGE

WHEN MACROPHYTES ARE SPARSE AND ALTERNATIVE PREY ARE PRESENT?

Introduction

« The importance of predation in structuring freshwater fish

communities has been well-documented (see review in Sih et al, 1985).

Ultimately predation takes two forms: (1) the direct effect due to

removal of prey individuals, and ( 2 ) alteration of prey behavior

simply due to the presence of predators (e.g., Sih 1982, Hlttelbach

1981, 1984, Power 1984, Power et al. 1985, Werner et al. 1983).

Although the direct impact of predation Is unquestionably an Important

force In determining the structure of freshwater fish communities, the

behavioral shifts that prey undergo in the presence of a predator have

also been shown to be Important (e.g., Mlttelbach & Chesson 1987,

Turner 1987; see review in Werner & Gilliam 1984). It is these

behavioral shifts that form the focus of this study.

In freshwater lakes, the growing paradigm is that habitat choice

by bluegill, Lenomls macrochirus. is determined by changes in growth

rate and predation risk as a function of habitat use and fish size.

As juveniles, bluegill are constrained to the vegetated littoral zone,

even when the limnetic zone is more profitable energetically, because

the risk of being eaten, relative to the growth rate, is greater in

147 148

the limnetic than in the litto ral zone (Gilliam 1982, Werner & Gilliam

1984, Werner et al. 1983, Werner & Hall 1988). As bluegill grow,

limnetic predation risk decreases to a point where the ratio of

mortality rate/growth rate in the limnetic zone falls below that in

the littoral zone, allowing bluegill to forage in the limnetic zone

(Mlttelbach 1981, Werner & Gilliam 1984, Werner & Hall 1988).

Most studies examining habitat choice by small bluegill have been

conducted in clear*water lakes with vegetated litto ral zones and fish

communities dominated by centrarchlds. In lakes in the southern and western United States turbidity is often greater than in more northern

lakes, and these systems have open-water fish communities dominated by planktlvores such as shad (both gizzard shad, Dorosoma cooedlanum. and

threadfln shad, petenense) (Jenkins 1955, 1967). Both factors have

the potential to modify habitat use by small bluegill.

In addition to its obvious direct effects on predator*prey

interactions (e.g., Vlnyard & O'Brien 1976), turbidity can indirectly affect predator-prey interactions by modifying the refuge available to prey fishes. In most studies examining habitat choice by small bluegill, macrophytes form the refuge that is provided for the prey

(Savlno & Stein 1982, 1989a, 1989b, Werner 6i Hall 1988). However, in systems where turbidity decreases the abundance of litto ral vegetation, often eliminating it completely (Chambers & Kalff 1985,

Duarte et al, 1986), this refuge is no longer available. In these systems, it may actually be the shallow areas of the littoral zone that provide refuge to small fish (for a recent example of the use of shallow areas as a refuge, see Mclvor & Odum 1988), In addition, 149 water*level fluctuations, particularly In nan*Influenced lnpoundnents, can prevent growth of aquatic macrophytes, similarly eliminating this refu g e.

Although available evidence suggests that shad have a negative impact on bluegill (through competition for zooplankton, especially as larvae; Ziebell et al. 1986, Prophet 1982, 1985, 1988), shad could have a positive effect on bluegill, by acting as a buffer to predation

(as suggested by Lagler & Applegate 1942). This effect could occur because predators generally select shad as prey when present (Lewis &

Helms 1964, Aggus 1973, Johnson et al. 1988, Carllno et al. 1986), possibly reducing the risk of predation to other prey species. The reduction in predation risk, especially In the open*wator habitat, could decrease the ratio of mortality rate/growth rate in the limnetic zone, such that this becomes a more profitable habitat than the littoral zone. In this paper I examine how small bluegill alter their habitat use in response to predatory largemouth bass and the presence of gizzard shad as a potential predation buffer, when littoral macrophyte refuges are not available.

Methods

Experiments were run in large (3*m dla, 0.5 m deep) pools that were tilted at a 15° angle to provide a simulated shallow, shoreline area and a deeper offshore area. This created a 1.7 m wide shoreline

In the shallow end of the pool, and an area that was 0.5 m deep In the deep end of the pool, Fools were categorically divided into deep and shallow regions by the 20 cm depth contour. I chose 20 cm because 150 this was the nlninun depth into which an adult largemouth bass (of the size used in these experiments) would actively pursue prey, as revealed In preliminary experiments. Thus, fish at depths less than

2 0 cm were probably safer from predation than those at depths > 2 0 cm

(see also suggestion of a 30 cm division by Werner & Hall 1988). Four treatments were used, consisting of Juvenile bluegill with all combinations of young-of-year gizzard shad and an adult largemouth bass. Treatments were presented in a randomized block design, with replicate experiments blocked by time.

One hour before the start of an experiment prey were Introduced to the pool. Frey were 20 bluegill (mean size ± 1 se - 40.6 + 0.2 mm), with or without 12 gizzard shad (mean size + 1 se ■ 69.4 ± 0.3 mm). These densities lie within the range of densities found in Ohio reservoirs (Johnson et al. 1988). Prey were not reused across experiments. Thirty minutes before the start of an experiment one bass (from a pool of five bass, size range - 205-265 mm) was placed into a cage (70 x 40 x 40 cm) in the deep end of the pool. In non bass treatments, an empty cage was used. All bass were used at least once in each treatment. At the start of each experiment, the cage was turned over and behavioral observations begun. The locations of the predator and all prey were quantified at 5-min intervals for 30 mln.

In addition to the definition of shallow (< 20 cm depth) and deep (>

2 0 cm depth) areas, these areas were further divided Into edge (< 1 0 cm from edge) and center (> 10 cm from edge). During predator treatments, I also recorded the location and activity (active-mobile and orienting toward prey, inactive-stationary and not orienting 151 toward prey) o£ the bass and whether bluegill were schooled (fish as part of a group that acted as a unit) or dispersed (fish not associating strongly with one another) at l*min intervals.

Observations were made from behind a plastic curtain for nine replicates of each treatment. For all analyses, multiple observations across time within an experiment were collapsed to a single value that was then treated as a replicate. Proportion data were arcsin- transformed before analysis.

R esu lts

Bluegill spent more time in the deep habitat in the absence of largemouth bass than in their presence (Fig. 23, randomized-block

ANOVA, F^'g - 8.71, p - 0.02), but did not change habitat use in response to gizzard shad (Fig. 23, F^g - 0.00, p - 0.98). In addition, the interaction term was not significant (gizzard shad x largemouth bass interaction, F^g - 0.00, p - 0.99), indicating that gizzard shad did not affect the manner in which bluegill responded to largemouth bass.

Uhen bluegill location was divided between center and edge areas, largemouth bass significantly affected bluegill mlcro-habltat choice

(Fig. 24, randomized-block MANOVA, F 3 6 - 20.61, p - 0.002), but neither the gizzard shad effect nor the bass x gizzard shad interaction term were significant (Fg^g - 0.03, p - 0.99 for gizzard shad, F3 tg - 0,10, p — 0.96 for the interaction term). Bluegill used the shallow edge more with largemouth bass than without it

(randomized-block ANOVA, Fj^g - 17.64, p - 0.003, alpha required for 152

significance - 0.0125) and had marginally significant: increases in

their use of both deep and shallow-center areas (randomized-block

ANOVAs, Fl(g - 8.75, p - 0.02 for deep-center; F^g - 6.78, p - 0.03 for shallow-center, alpha required for significance - 0.0125).

Although gizzard shad were always present in combination with bluegill, shad did not alter*their habitat use in response to the presence of largemouth bass (Fig. 25, randomized-block MAN0VA, Fgg -

0.61, p - 0.63). They used the deep habitat during 85-95X of the observations, spending most of this time in the deep-edge habitat

(F ig . 25).

Though bluegill used the shallow habitat with largemouth bass, they still occurred in the deep habitat about 50X of the time. In fact, time spent in the deep-edge habitat in the presence or absence of the predator did not differ (randomized-block ANOVA, F^g - 3.62, p

- 0.09). To determine if this could be explained by differences in predator activities across experiments, 1 reg ressed the mean proportion of bluegill observed in the shallow edge of the pool against the proportion of observations during which the largemouth bass was active in the deep end of the pool (from the 1 -min observations). For both predator treatments, the regression was not significant (bluegill+largemouth bass+glzzard shad, p-0.57; bluegill+largemouth bass, p-0.98; treatments combined, p-0.62, Fig.

26). To determine if schooling, another response to predation risk

(Seghers 1974, Sullivan & Atchinson 1978), was related to bass activity, I regressed the proportion of observations during which bluegill were schooled against the proportion of 1 -min observations 153 during which the largemouth bass was active. Apparently, some level of predator activity was required before bluegill schooled (Fig. 27),

Indicating the need for a threshold model, such as the logistic.

Because sample sizes for the two predator treatments were small, I combined them for this analysis to allow sufficient data points both above and below the threshold. The regression was significant (data transformed as in Neter et al. 1983, p — 0.03), Indicating that largemouth bass activity influenced bluegill schooling behavior according to a logistic model. At low predator activity levels (i.e.,

< 10X of observations In these experiments) bluegill tended to remain dispersed, while at higher predator activity levels (£ 25X of observations in these experiments), bluegill were almost entirely schooled.

* D iscussion

Bluegill altered their habitat use in laboratory pools in response to the presence of a largemouth bass predator by reducing their use of deeper water. Without predators, bluegill occurred in

• « deep water about 80-85X of the time. With a predator, bluegill spent approximately equal amounts of time in deep and shallow areas of the pool. In similar experiments where artificial macrophytes provided refuge, Savino & Stein (1989a) also found that bluegill altered their habitat use in the presence of largemouth bass, with 35X of the bluegill using the risky (unvegetated) habitat in the absence of the predator and OX using it in the presence of the bass. In these experiments, nearly all time spent in the deep water with a predator 154 was spent along the pool edge. Because this edge provides refuge

(Moody et al. 1983), bluegill using this area of the pool nay actually be relatively safe from predation. That the deep end of the pool should be riskier than the shallow end is supported by the results of

Mclvor and Odum (1988), who documented that Fundulus spp. experienced decreased predation risk in shallow relative to deep habitats. In addition, in my experiments, predators, when active, were observed in the deep end of the pool 70 ± 5X (X ± 1 s.e., n-16) of the time.

Thus, the deep end of the pool likely represents increased predation risk relative to the shallow end.

Gizzard shad did not influence bluegill habitat use under any circumstances. Thus, although the presence of shad may reduce the number of predator attacks on other prey (as described by Chitwood

1976), their presence did not lead to short-term changes in habitat use by bluegill. In addition, in the presence of bluegill, shad did not alter their habitat use in the presence of a predator from that observed in the absence of the predator, spending about 90X of their time in the deep end of the pool.

Though bluegill did not respond to the presence of shad during these short-term experiments, longer-term changes may occur, via apparent competition (Holt 1977, Abrams 1987). Given that shad are a preferred prey, their presence may increase predator abundance, ultimately increasing predation pressure on bluegill, by increasing predator population size, unless control is exerted on the predator

(e.g., by a fishery for that species). Predicted outcomes of these 155 complex Interactions require the use of theoretical models, as in

Abrams (1987).

Although bluegill responded to the predator, their habitat use was not related to predator activity, suggesting that the choice of deep versus shallow habitat is related only to presence or absence of the predator, rather than to some estimator of predator activity.

* Schooling by bluegill, however, was related to bass activity, according to a logistic model, suggesting some threshold of predator activity is required for bluegill to begin schooling.

Bluegill may be responding to the presence of a predator in different ways at different scales, Uhen a predator is initially detected, bluegill may move into the shallow areas of the littoral zone regardless of whether the predator is active. As long as the predator is relatively Inactive, bluegill remain dispersed in the shallow area. As the bass becomes more active, bluegill do not alter their habitat use, but rather they respond by schooling in the shallow a re a.

If bluegill are restricted to shallow habitats, even when the macrophyte refuge is removed, then predictions about bluegill habitat use generated from work in more northern systems may s till apply to southern lakes. In the current study, Increased turbidity levels were only examined in terms of their indirect potential Impact on predator- prey interactions via the removal of the macrophyte refuge. However, turbidity can have important direct impacts on lakes and reservoirs by altering both system productivity and predator-prey interactions

(e.g., Vlnyard & O'Brien 1976), and these areas require additional 156

study before predictions as to their effects on bluegill habitat

choice can be made.

In addition to the Indirect effects examined here, shad can have

Important direct effects on reservoir community structure. Because

shad feed on zooplankton, at least as larvae and juveniles (e.g.,

Cramer & Marzolf 1970, Barger & Kllambi 1980, Van Den Avyle & Wilson

1980), they can Increase the ratio of mortality risk to foraging gain

in the open water, altering predictions of bluegill habitat choice.

These effects of shad must be documented before predictions of bluegill habitat use in the presence of shad can be made.

In the current study I have shown that even when macrophytes are

eliminated and a preferred alternate prey is present, bluegill

continue to use shallow areas to avoid predation. Thus, predictions based on work In clear-water systems, where macrophytes provide refuge may not need to be altered when macrophytes are not present. 157

Literature Cited

Abrams, P. 1987. Indirect interactions between species that share a predator: varieties of indirect effects, pp. 38*54. £ q : W. C. Kerfoot & A. Sih (ed.) Predation: Direct and Indirect Impacts in Aquatic Communities. University Press of New England, Hanover.

Aggus, L. R. 1973. Food of angler harvested largemouth, spotted, and smallmouth bass in Bull Shoals Reservoir. Proc. Ann. Conf. SE. Assoc. Game Fish Comm. 26:501-505.

Barger, L. E. & R. V. Kllambi. 1980. Feeding ecology of larval shad, Dorosoma. in Beaver Reservoir, Arkansas, pp, 136*145. In: L. A. Fuiman (ed.) Proceedings of the Fourth Annual Larval Fish Conference, U.S. Fish and W ildlife Service, Ann Arbor.

Carline, R. F., R. A. Stein, & L. M. Riley. 1986. Effects of size at stocking, season, largemouth bass predation, and forage abundance on survival of stocked tiger muskellunge. Amor. Fish. Soc. Spec. Pub. 15:151-167.

Chambers, P. A. & J. Kalff. 1985. Depth distribution and biomass of submersed macrophyte communities in relation to Secchi depth. Can. J. Fish. Aquat. Scl. 42:701-709.

Chitwood, J. B. 1976. The effects of threadfln shad as a forage species for largemouth bass in combination with blueglll, redear, and other forage species. MS Thesis, Auburn University, Auburn. 28 pp.

Cramer, J. D. & G. R. Harzolf. 1970. Selective predation on zooplankton by gizzard shad. Trans. Amer. Fish. Soc. 99:320-332.

Duarte, C. M., J. Kalff, & R. H. Peters. 1986. Patterns in biomass and cover of aquatic macrophytes in lakes. Can. J. Fish. Aquat. Sci. 43:1900-1908.

Gilliam, J. F. 1982. Foraging under mortality risk in size-structured populations, Ph.D. Dissertation, Michigan State University, East Lansing. 107 pp.

Holt, R. D. 1977. Predation, apparent competition, and the structure of prey communities. Theor. Pop. Biol. 12:197-229.

Jenkins, R. M. 1955. The effect of gizzard shad on the fish population of a small Oklahoma lake. Trans, Amer. Fish. Soc. 85:58-74.

Jenkins, R. M. 1967. The influence of some environmental factors on standing crop and harvest of fishes in U.S. reservoirs, pp 298- 321. Iq : The American Fisheries Society (ed.) Reservoir Fishery Resources Symposium, American Fisheries Society, Bethesda. 158

Johnson, B. M., R. A. S te in , & R. F. C arlin e. 1988. Use o f a quadrat rotenone technique and bioenergetics modeling to evaluate prey availability to stocked pisclvores. Trans. Amer. Fish. Soc. 117:127-141.

Lagler, K. F. & V. C. Applegate. 1942. Age and growth of the gizzard ‘ shad, Dorosoma cenedlanum (LeSueur), with a discussion of its value as a buffer and as forage of game fishes. Invest. Ind. Lakes Streams 2:99*110.

Lewis, V. M. & D. R. Helms. 1964. Vulnerability of forage organisms to largemouth bass. Trans. Amer. Fish. Soc. 93:315-318.

H clvor, C. C. & V. E. Odum. 1988. Food, p red atio n r is k , and microhabitat selection in a marsh fish assemblage. Ecology 69:1341-1351.

Mlttelbach, G. G. 1981. Foraging efficiency and body size: a study of optimal diet and habitat use by bluegllls. Ecology 62:1370- 1386.

Mlttelbach, G. G. 1984. Predation and resource partitioning in two sunfishes (Centrarchldae). Ecology 65:499-513.

Mlttelbach, G. G. & P. L. Chesson. 1987. Predation risk: indirect effects on fish populations, pp 315-332. In: W. C. Kerfoot & A. Sih (ed.) Predation: Direct and Indirect Impacts in Aquatic Communities. University Press.of New England, Hanover.

Moody, R. C ., J . M. H olland, & R. A. S te in . 1983. Escape ta c tic s used by bluegllls and fathead minnows to avoid predation by tiger muskellunge. Env. Biol. Fish. 8:61-65.

Neter, J., W. Wasserman, & M. H. Kutner. 1983. Applied linear regression models. Irwin Publishers, Homewood. 547 pp.

Power, M. E. 1984. Depth distribution of armored catfish: predator- induced resource avoidance? Ecology 65:523-529,

Power, M. E., U. J. Matthews, & A. J. Stewart. 1985. Grazing minnows, piscivorous bass, and stream algae: dynamics of a strong interaction. Ecology 66:1448-1456.

Prophet, C. V. 1982, Zooplankton changes in a Kansas lake 1963-1981. J. Freshwater Ecol. 1:569-575.

Prophet, C. W. 1985. Calanold population structure in a Kansas lake after introduction of threadfln shad. The Southwestern Naturalist 30:162-163. 159

Prophet, C. V. 1988. Changes in seasonal population structures of two species of Dlaotomus (Calanoida, Copepoda) subsequent to Introductions of threadfln and gizzard shad, The Southwestern Naturalist 33:41*53.

Savino, J. F. & R. A. Stein. 1982. Predator-prey interactions between largemouth bass and bluegllls as Influenced by simulated, submersed vegetation. Trans. Amer. Fish. Soc. 111:255*266.

Savino, J. F. & R. A. Stein. 1989a. Behavior of fish predators and their prey: habitat choice between open water and dense vegetation. Env. Biol. Fish. 24:287*293.

Savino, J. F. & R. A. Stein. 1989b. Behavior of fish predators and their prey: effects of macrophyte density. Anita. Behav. 37:311- 321.

Seghers, B. H. 1974. Schooling behavior in the guppy (Poecllia reticulata!: an evolutionary response to predation. Evolution 28:486*489.

Slh, A. 1982. Foraging strategies and the avoidance of predation by an aquatic Insect, Notonecta hoffmannl. Ecology 63:786-796.

Sih, A., P. Crowley, H. McPeek, J. Petranka, & K. Strohmelr. 1965. Predation, competition, and prey communities: a review of field experiments. Ann. Rev. Ecol. Syst. 16:269*311.

Sullivan, J. F. & C. J, Atchlnson. 1978. Predator-prey behavior of fathead minnows, Plraenhales promotes, and largemouth bass, Mlcropterus salmoldes. in a model ecosystem. J. Fish Biol. 13:249*253.

Turner, A. M. 1987. Predator avoidance by juvenile sunfish; effects on the plankton. M.S. thesis, The Ohio State University, Columbus. 40 pp.

Van Den Avyle, M. J. & J. R. Wilson. 1980. Food habits and feeding selectivity of larval Dorosoma spp. in Center Hill Reservoir, pp 146-156. Id : L. A. Fuiman (ed.) Proceedings of the Fourth Annual Larval Fish Conference, U.S. Fish and W ildlife Service, Biological Services Program, Ann Arbor.

Vinyard, G. L. & W. J. O'Brien. 1976. Effects of light and turbidity on the reactive distance of blueglll sunfish (Leoorals macrochirus!. J. Fish. Res. Board Can. 33:2845-2849.

Werner, E. E. & J. F. Gilliam. 1984. The ontogenetic niche and species Interactions in size-structured populations. Ann. Rev. Ecol. Syst. 15:393-425. 160

Werner, E. E., J. F. Gillian, D. J. Hall, & G. G. Mlttelbach. 1983. An experimental test of the effects of predation risk on habitat use In fish. Ecology 64:1560-1548.

Werner, E. E. & D. J. Hall. 1988. Ontogenetic habitat shifts in blueglll: the foraging rate-predation risk tradeoff. Ecology 69:1352-1366.

Ziebell, C. D., J. C. Tash, & R. L. Barefield. 1986. Impact of threadfln shad on microcrustacean zooplankton in two Arizona lakes. J. Freshwater Ecol. 3:399-406. Figure 23. Proportion of observations during which blueglll wore in the shallow {< 2 0 cm deep) and deep (> 2 0 cm deep) areas of a 3-m diameter pool. Treatments were small blueglll (BG) with all combinations of an adult largemouth bass (LMB) and young-of-year gizzard shad (GS). Nine replicates were run for each treatment.

161 Proportion of observations 0.2 .4 0 0.6 0.0 0.8 1.0 (< 20 cm) (> 20 cm) (> 20 cm) (< 20 hlo deep shallow GLBG BG+LMB BG+LMB+GS J T GG BG BG+GS T T

s. . .e s 1 ± n a e m 162 Figure 24. The proportion of observations during which blueglll were in the shallow-center (SC; < 20 cm deep, > 10 cm from edge), shallow- edge (SE; < 20 cm deep, < 10 cm from edge), deep-center (DC; > 20 cm deep, > 10 cm from edge), and deeip-edge (DE; > 20 cm deep, < 10 cm from edge) areas of the laboratory pool. Treatments were small blueglll (BG) with all combinations of an adult largemouth bass (LMB) and young-of-year gizzard shad (GS). Nine replicates were run for each treatment.

163 ■sM na 0.8 f O x > CO mean ± 1 s.e. c o J +- SE a > 0.6 V- DE Q> CO -Q O M— 0 .4 11 o C .9 o 0.2 Q. O V. DC CL SC 0.0 _ £ i i J*1 A t BG+LMB+GS BG+LMB BG+GS BG

j ON I Figure 25. The proportion of observations during which gizzard shad were in the four areas of the pool. See Figure 24 for definitions.

165 a t r c

Proportion of observations .4 0 0.6 0.8 0.0 0.2 1.0 x l ■ ^ +G+MB +LM S+BG G SC Q SE DC I DE I V a s. . .e s 1 ± ean m j £ i GS+BG a f 4

Figure 26. The mean p ro p o rtio n o f o b serv atio n s during which b lu e g ill were found in the shallow (< 2 0 cm deep) end of the pool as a function of the proportion of observations during which the largemouth bass were active in the deep end of the pool (from the l>mln observations).

Circles are mean proportions from experiments with blueglll, gizzard shad, and largemouth bass, and squares are from experiments with blueglll and largemouth bass. Nine replicates were run for each treatm en t.

167 00 ~ c

Mean proportion bluegill in shallow end 0.50 = 0.50 0.006 - 0.25 0.75P* 1.00 .0 .5 .0 .5 1.00 0.75 0.50 0.25 0.00 rprin bevtos rdtr cie n ep end deep in active predator observations Proportion a ■ — l BG+LMB+GS

BG+LMB 168 Figure 27. The proportion of observations during which blueglll were schooled as a function of the proportion of observations during which the largemouth bass was active (from the 1-mln observations),. Symbols are as in Figure 4. Nine replicates were run for each treatment.

169 Proportion of observations bluegill schooled 0.00x 0.50 .5 „ 0.25 0.75 1.00 .0 .5 .0 0.75 0.50 0.25 0.00 0 ------□

rprin f bevtos rdtr active predator observations of Proportion - o - o - 1 ------1 ------□ □ □ 1 ------o a -o

1.00 170 SUMMARY

Planktivores, such as shad, can Influence dramatically reservoir

community dynamics. Herein, I discuss predictions regarding the

Interactions between threadfln shad and three life stages of

largemouth bass and blueglll; I then summarize the results of each

chapter, evaluating these predictions. To visualize the potential

Impact of shad on these fishes, consider the following predictions of

the effects of shad on larval, juvenile, and adult largemouth bass

and blueglll.

L ife Stage

Soeclcs Larvae Juveniles Adults Overall

Largemouth bass 0 + + +

Blueglll - - or +

In the literature review (Chapter I), I reviewed the evidence for

an overall effect of shad on these species. This review demonstrated

that shad introductions can enhance growth of predators, such as crappie and largemouth bass, and negatively affect presumed competitors such as blueglll. However, responses were often

inconsistent within a species, with some studies documenting negative responses of predators or positive responses of competitors to shad

introductions. Across all studies, all target species experienced

171 172 negative, neutral, and positive effects due to shad removal, making generalizations impossible. An absence of statistical treatment of the data, coupled with several problems associated with study design, further complicated interpretation of outcomes. Additionally, because resident predators feeding on introduced prey comprise only a portion of the complex of interactions that occur in a lake, factors such as ontogenetic shifts in diet and habitat and spatial heterogeneity must be considered when attempting to predict the outcome of these manipulations. Only by considering the complexities of lakes and reservoirs will we be able to predict the ultimate outcome of forage fish manipulations.

Because diet and habitat use of fishes change with ontogeny, interactions at all life stages can influence the ultimate outcome of management manipulations. Consequently all life stages must be examined when evaluating the success of management practices. To evaluate the impact that interactions among larval fishes could have on reservoir community structure, I evaluated how larval threadfln shad influence a potential competitor (young-of-year blueglll) and predator (largemouth bass) during threadfln shad introductions into two lakes. Because shad spawned in August In both lakes, larval shad overlapped little with earlier-spawned blueglll. However, In one lake, zooplankton abundance declined precipitously following peak larval threadfln shad abundance. Data on cladoceran birth rates indicate that this decline was due to increased predation. The virtual elimination of zooplankton during this period was likely responsible for reduced blueglll recruitment, which, in turn, 173

contributed to reduced growth of young-of-year largemouth bass (due to

reduced forage fish availability) following threadfin shad

introduction. Competition among larval fish for zooplankton had a

predictable negative effect on young-of-year blueglll, as well as an

unforseen negative impact on young-of-year largemouth bass.

Potential indirect negative effects of management manipulations on the

very species to be enhanced must be considered when manipulating prey

or predator species.

The influence of larval gizzard shad on zooplankton, and

ultimately on blueglll In Kokoslng Lake was equally dramatic.

Macrozooplankton density declined to near zero immediately after

larval gizzard shad abundance peaked during 1987 and 1988. This decline was attributed to reduced zooplankton birth rates (most probably owing to selective predation by larval shad on egg-bearing

females) and Increased death rates, due to predation. In a larval gizzard shad enclosure/exclosure experiment, I found that larval gizzard shad significantly reduced densities of macrozooplankton, rotifers, and edible phytoplankton within 2 weeks of introduction.

Consequently, larval gizzard shad simultaneously compete with (by directly consuming edible algae) and prey upon zooplankton, causing them to crash. Because larval blueglll migrate to the limnetic during or shortly after the zooplankton crash, food consumed by blueglll and their ultimate recruitment were reduced by gizzard shad.

Through direct predation and consumption of edible algae, gizzard shad can drive zooplankton to extinction, reducing sport-fish recruitment. 174

The story on shad nay not, however, be all bad news. A potential

benefit of shad Is that, as a preferred prey, their presence may

reduce predation rates on less preferred prey types, providing a type

of "buffer". In laboratory pools blueglll used depths £ 20 cm more

often when largemouth bass were present than without them, indicating

that shallow areas of the pool provided refuge for blueglll. However,

the presence of a preferred alternate prey, gizzard shad, did not

affect this behavior. Though blueglll did not alter their habitat use

in response to bass activity, they did school more as predator

activity Increased. Even In the presence of a preferred alternate

prey when macrophytes were not present, blueglll continued to use

shallow "littoral" areas to avoid predation. Consequently, gizzard

shad do not appear to provide a predation buffer, at least in terms of habitat choice by young-of-year blueglll.

My original predictions were reasonably well'Supported; however,

the indirect effects of threadfln shad on young-of-year largemouth bass (In Stonellck Lake) and the complex Interactions among larval

gizzard shad, zooplankton, phytoplankton, and larval blueglll (in

Kokoslng Lake) indicate the importance of "surprises" in the outcome of interactions involving open-water planktlvores. These results emphasize that all life stages of interacting species and food-web structure be considered if we are to make accurate predictions of the outcome of management manipulations. What is clear is that both gizzard and threadfln shad are extremely important components of our lakes and reservoirs. As a consequence, we must continually strive to 175 understand their role In these systems and the Impact they have on all species and li£e stages of organisms across all trophic levels. APPENDIX A

COMPARING THREE ZOOPLANKTON SAMPLERS:

A TAXON-SPECIFIC ASSESSMENT

Introduction

A variety of devices are available for sampling zooplankton,

including plankton nets (see examples in Lind 1979), plankton traps

(e.g.t Schindler 1969; Ackefors 1971; Kankaala 1984), pumps (e.g.,

Waite and O'Grady 1980; Farquhar and Geiger 1984), and tube samplers

(e.g., George and Owen 1978; Lewis and Saunders 1979; Solomon et al.

1982; Graves and Morrow 1988), When density estimates integrated

across depth, are required, samples taken at discrete depths with a plankton trap must be combined, resulting in wasted effort. Vertical net tows provide integrated samples, but suffer from reduced

filtration efficiency due to clogging, avoidance of the net by some

taxa, and loss of plankton through the mesh (Tranter and Smith 1968;

Clutter and Anraku 1968; Vannucci 1968). Tube samplers represent an alternative to vertical net tows for collecting integrated samples (at least for depths £ 5 in).

To assess a new gear type, zooplankton densities collected by it and "standard" gear types must be compared. Historically, these comparisons often are done for a few taxa or the plankton is arbitrarily divided into copepods, cladocerans, and rotifers (but see

Solomon et al. 1982). Given that zooplankton exhibit taxon-specific

176 177 behavior (e.g., Meyers 1980, Lamport 1987), gear cypes should be evaluated by comparing densities of individual taxa across gear types.

Here we use this approach to compare zooplankton densities collected by an integrated tube sampler, a vertical net haul, and a

Schindler-Fatalas trap. Ve determine whether tube samplers collect plankton densities comparable to more traditional gear types.

Methods

We collected zooplankton during 24 h in Stonelick Lake (Clermont

County, Ohio) using a vertical net haul, a Schindler-Patalas trap, and a tube sampler. Samples were taken from the surface to 2 m. Because

Stonelick Lake is shallow (mean depth - 2.3 m, max depth -4m), maximum depth at all sites was between 2.5-3.0 m. We collected three replicate.samples at seven times (0000, 0400, 0630, 0930, 1430, 1730, and 2130) on 23-24 June 1987. Because the Schindler-Patalas trap was

"0.5 m tall and samples were collected at O.S-rn intervals from the surface to 2.0 m, the entire water column was sampled. A tube sampler, made of clear plastic (0.16 cm thick, 7.3 cm inside dla), had a cap fitted to the top with a hole (0.18 cm dia). After it was lowered to 2 m, a stopper was put in the hole, the sampler raised, and water filtered through a plankton net. Mesh size in all cases was 54 urn. Volumes collected were 179 1 for the zooplankton net (35 cm dia; volume filtered confirmed with flow meter readings), 24.3 1 for the

Schindler-Patalas trap (22 cm x 23 cm x 47 cm), and 16.8 1 for the tube sampler (two hauls per sample). Samples were preserved in 5X sucrose formalin (Haney and Hall 1973). At least 200 individuals were 178

counted for each taxon in each sample. For abundant taxa, 10X

subsamples were taken until at least 2 0 0 individuals were counted; for

rare taxa, all individuals were counted. Numbers were converted to

densities and log*transformed. Data were analyzed using SAS GLM

techniques (SAS Institute 1985).

Results and Discussion

Total zooplankton density differed across gear, with density

ordered: tube > Schindler-Patalas trap > net (Fig. 28a; randomized

block ANOVA, p-0.0001). The block (i.e., time) effect was marginally

significant (p-0.04), but the gear x time interaction term was not

(p -0 .4 2 ).

When grouped by cladocerans, copepods, and rotifers, a

significant gear effect still occurred, though order of the gear types

* differed (Figure 28b-d; randomized block ANOVAs, gear effect: p<0.01

for all taxa). The tube sampler collected higher densities of

copepods and cladocerans than did the Schindler-Patalas trap and the

net (p<0,009). The Schindler-Patalas trap collected more copepods

than the net (p-0 . 0 0 0 1 ), but these two gear types did not differ for

cladocerans (p-0.70). The Schindler-Patalas trap captured more

rotifers (p<0.003) than the net, which captured more than the tube

sampler (p-0.04). The gear x time interaction term was not

significant across taxa (p>0.14).

To compare across taxa, we used a m ultivariate ANOVA, where

densities of zooplankton taxa captured by each gear type were dependent variables. To obtain sufficient degrees of freedom to run 179

the MANOVA, we did not include Chvdorus. Molna. or ostracods because

these taxa never accounted for more than 0.3Z of the sample.

Densities of all herbivorous rotifers were combined, but Asplanchna

was considered separately due to its predatory food habits.

- The MANOVA yielded a significant gear effect (p-0.0003) and block

(time) effect (p- 0 . 0 0 0 1 ), and the gear x time interaction term was

marginally significant (p-0.04). Of the 10 taxa, four did not have a

significant gear effect (i.e., p was not < [0.05/10J-0.005), five had

a significant gear effect (calanold copepods, nauplii, Bosmlna.

Diaphanosoma. and herbivorous rotifers), and one had a significant

gear x time interaction term fChaoborus: however, the gear effect for

Chaoborus was not significant, p-0.01).

Gear types did not differ for cyclopoid copepods, Daphnia.

Cerlodaphnla. and Asplanchna (Table 8 ). Of the five taxa differing

across gear types, the tube sampler collected densities of calanold copepods, nauplii, and Bosmlna at least as high as those collected by

the trap and the net (Table 8 ). The Schindler-Patalas trap collected highest densities of herbivorous rotifers; the net collected highest densities of Diaphanosoma (Table 8 ).

Because rotifer density increased with depth to 2 m during 24 h

(randomized block ANOVA, test for main effect of depth F 3 \g-13.21, p<0.0001, Fig. 29), water loss from the bottom of the tube sampler as

it was brought above the surface may have reduced numbers collected.

In fact, when rotifer density was greatest at depths < 2 m (I.e., 0930 and 1430, Figure 29), differences between the trap and tube were less

than when rotifer density peaked at 2 m (c.f., Figs. 28 and 29). A 180 valve could be placed on the end of che sampler Co prevenC losses of water as che Cube is raised (Lewis and Saunders 1979; Solomon et al.

1982; Graves and Morrow 1988).

Why Diaphanosoma should be less susceptible Co the Cube and Crap

Chan che net is not clear. Because Diaphanosoma is easy to catch, at lease relative to copepods (Drenner and McComas 1980), and che tube collected copepods as efficienCly as the nec (Table 8 , Fig. 28), gear avoidance Is not likely. Given Chat Che Schindler-Patalas trap typically is considered Che standard for comparison, and it also collected significantly less Diaphanosoma than did the net, the lower efficiency of these two gear types relative to the net is an enigma.

That the tube collected more zooplankton than the trap, which collected more zooplankton than the net, is not surprising. The tube typically compares 'favorably with the net (Lewis and Saunders 1979;

Graves and Morrow 1988) and traps (George and Owen 1978; Solomon et al. 1982); in turn, Craps compare favorably with nets (Schindler 1969;

Kankaala 1984), However, these studies had problems: including lumping across taxa, absence of variation, and conducting multiple statistical tests without adjusting alpha levels. Whether study conclusions would be altered, by using taxon-specific, multivariate analyses, is not obvious.

A taxon-specific approach is required when evaluating gear types.

Though tube samplers were best when comparing total zooplankton, it was poorest for rotifers, though s till best for cladocerans and copepods. Within Cladocera, the tube was best for collecting Bosmlna. but was worst for collecting Diaphanosoma. Thus, before sampling 181 efficiency can be assessed, conparisons across taxa are necessary.

Because tube samplers Integrate the entire water column, providing zooplankton densities commensurate with traditional gear, it

Is a time-saving alternative gear for sampling zooplankton in shallow la k e s . 182 Literature Cited

Ackefors, H. 1971. A quantitative plankton sampler. Olkos 22:114- 118.

Clutter, R. I. and H. Anraku. 1968. Avoidance of samplers, p. 57- 76. In D. J. Tranter [ed.], Reviews on Zooplankton Sampling Methods. UNESCO Monographs on Oceanographic Methodology #2, Geneva.

Drenner, R. W. and S. R. HcComas. 1980. The ro le s o f zooplankter escape ability and fish size selectivity in the selective feeding and impact of planktlvorous fish, p. 587-593. In W. C. Kerfoot [ed.], Evolution and Ecology of Zooplankton Communities. Univ. Press of New England, Hanover.

Farquhar, B, W. and J. C, Geiger. 1984. Portable zooplankton sampling apparatus for hatchery ponds. Prog. Fish-Cult. 46:209- 211.

George, D. G. and G. H. Owen. 1978. A new tube sampler for crustacean zooplankton. Llmnol. Oceanogr. 23:563-566.

Graves, K. G. and J. C. Morrow. 1988. Tube sampler for zooplankton. Prog. Fish-Cult. 50:182-183.

Haney, J. ,F. and D. J. Hall. 1973. Sugar-coated Daohnia: a preservation technique for Cladocera. Llmnol. Oceanogr. 18:331- 333.

Kankaala, P. 1984. A quantitative comparison of two zooplankton sampling methods, a plankton trap and a towed net, In the Baltic. Int. Revue ges. Hydroblol. 69:277-287.

Lewis, W. M., Jr. and J. F. Saunders, III. 1979. Two new integrating samplers for zooplankton, phytoplankton, and water chemistry. Arch. Hydroblol. 85:244-249.

Lamport, W. 1987. Vertical migration of freshwater zooplankton: indirect effects of vertebrate predators on algal communities, p. 291-299. In W. C. Kerfoot and A. Slh [eds.], Predation: Direct and Indirect Impacts on Aquatic Communities. Univ. Press of New England, Hanover.

Lind, 0. T. 1979, Handbook of Common Methods in Limnology, 2nd edition. C. V. Mosby Co., St. Louis, Missouri, USA.

Meyers, D. G. 1980. Diurnal vertical migration in aquatic mlcrocrustacea: light and oxygen responses of litto ral zooplankton, p. 80-90. In W. C. Kerfoot [ed.], Evolution and Ecology of Zooplankton Communities. Univ. Press of New England, Hanover. 183

SAS Institute Inc. 1985. SAS User's Guide: Statistics, Version 5. SAS Institute Inc., Cary, North Carolina, USA.

Schindler, D. W. 1969. Two useful devices for vertical plankton and water sampling. 'J. Fish. Res. Board Can. 26:1948-1955.

Solomon, K. R., K. Smith, and G. I. Stephenson. 1982. Depth integrating samplers for use in limnocorrals. Hydroblol, 94:71- 75.

Tranter, D. J. and P. E. Smith. 1966. Filtration performance. Pages 27-56 in D. J. Tranter, ed. Reviews on Zooplankton Sampling Methods. UNESCO Monographs on Oceanographic Methodology #2, Geneva, Switzerland.

Vannuccl, M. 1968. Loss of organisms through the meshes, p. 77-86. In D. J. Tranter [ed.], Reviews on Zooplankton Sampling Methods. UNESCO Monographs on Oceanographic Methodology #2, Geneva.

Waite, S. W. and S. M. 0 'Grady. 1980, Description of a new submersible fllter-pump apparatus for sampling plankton. Hydroblol. 74:187-191. 184

Table 8 . Results of the HANOVA testing density of zooplankton within

each of 10 taxa collected by three gears In Stonellck Lake, Ohio.

Because gears were compared across all 10 taxa, alpha level required

for significance was [0.05/10]-0.005. Legend: N - vertical net haul,

S - Schindler-Patalas trap, and T - tube sampler.

laxon p for gear effect direction of_response

C0PEF0DS

calanolds 0.0001 T > S > N

cyclopolds 0.09 ------

nauplil 0.0001 T - S > N

CLADOCERANS

Daphnla 0.009 -----

Bosmlna 0.001 T > N - S

Dlaphanosoma 0.0001 N > T - S

Cerlodaohnla 0.008

ROTIFERS

Asplanchna 0.01------

herbivorous rotifers 0.0006 S > N - T

OTHER

Chagbgpjg 0.01 Figure 28. Total zooplankton (a), cladocerans (b), copepods (c), and rotifers (d) collected by an Integrated tube sampler (dotted line), a

Schlndler-Patalas trap (dashed line), and a vertical net tow (solid line) during .24 h (N-3 replicate samples) in Stonelick Lake, Ohio on

23*24 June 1987. The bar along the x-axis indicates the period of d ark n ess,

185 186

m i 1000

000 tuba trap

300 not

300 i tuba cn 100 not

SP trap

u 2

SP trap

not

I k. «> o not e SP trap 2 tubo S3 e

0000 0400 0800 1200 1600 2000 2400 Figure 28 time Figure 29. Vertical distribution of rotifers sampled with a # Schindler-Patalas trap at seven times during 24 h in Stonelick Lake,

Ohio.

187 1730 + + ] - i ----- 50 0 50 0 50 0 -H 0930 1430

) } rotifers • liter rotifers • 0630

3 + X + SE + SE X 0400

] r 50 0 50 0 50 -H— 0000 3 3 - 3 a. o E 1 TJ r

Figure 29 bottom surface

MICROFILMED 1990 INFORMATION to USERS the Most Advanced Technology Has Been Used to Photo­ Graph and Reproduce This Manuscript from the Microfilm Master

MICROFILMED 1990 INFORMATION to USERS the Most Advanced Technology Has Been Used to Photo Graph and Reproduce This Manuscript from the Microfilm Master