 

 

 

 

 

 

u
v
u
v

w )‘3v
4‘ ‘

 

Oh‘ﬂr 'Iﬂ’ﬂ uh

mun-cm

,t- -. - . .. . ;........ 3E
«2&3. 4....u.+..n.,ﬂr€...... , . . .é ; _ :

. l
I [Iv-ll

 

THESig

This is to certify that the
dissertation entitled
MIXED COMPETITION/PREDATION INTERACTIONS IN

SIZE-STRUCTURED FI SH COMMUNITIES

presented by

Mark Harrison Olson

has been accepted towards fulﬁllment
of the requirements for

Ph .D . degree in Zoology

ﬂfm

Major professor

 

oat... A50, /7, M93

 

MSU is an Affirmative Action/Equal Opportunity Institution 0- 12771

 

 

 

Milliillilll‘lillliliii

3 1293 01022 4479

 

LIBRARY
Michigan State
University

 

 

 

PLACE ll RETURN BOX to romovo this checkout from your rocord.

TO AVOID FINES Mum on or bdoro dot. duo.

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MSU Is An Afﬁrmative Action/Equal Opportunlty lnotttdlon
m

i i

MIXED COMPETITION/PREDATION INTERACTIONS IN
SIZE-STRUCTURED FISH COMMUNITIES

By

Mark Harrison Olson

A DISSERTATION

Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of

DOCI‘OR OF PHILOSOPHY
WK. Kellogg Biological Station
and

Department of Zoology
1993

ABSTRACT
MIXED COMPETITION/PREDATION INTERACTIONS IN
SIZE-STRUCTURED FISH COMMUNITIES
By

Mark Harrison Olson

One consequence of ontogenetic niche shifts in size-structured populations is the potential
for interspecific interactions to change with size. Individuals of two species may compete
with one another when they overlap in size, but as the individuals diverge in size the
interaction may switch to predator/prey. In this dissertation, I examined this type of mixed
competitive/predatory interaction as it occurs between largemouth bass (MW

31mm and bluegill (HEELS Mm. Small bass competed with small bluegill
for invertebrates, but as bass got larger small bluegill became an important prey item.

Data collected on diets of Young-of-Year (Y OY) bass in four small lakes in Michigan
showed that bass typically shifted to feeding on bluegill in their first year. However, in
order to become piscivorous, bass required a size advantage over their prey. Because of
I this, bass were sensitive to several factors which affected the relative sizes of bass and
bluegill. Most importantly, this study demonstrated that grth rates in the invertebrate
feeding stage played a large role in determining the timing of the shift to piscivory. Growth
rates among the lakes studied were highly variable, and were strongly dependent on the

timing of the shift.

During the invertebrate feeding stage, YOY bass competed with small bluegills for
invertebrate prey. Results of an experimental manipulation demonstrated that this
competition was strong and was mediated by changes in the size-structure of invertebrates

in the open water and vegetated habitats. Furthermore, competition between bass and

bluegill was asymmetric: bluegill had a much stronger effect on bass. This effect appears
to be important to dynamics of bass populations in natural systems. Analysis of bass and
bluegill populations in a set of 7 lakes suggests that the two species do interact as both
competitors and as predator/prey, generating patterns of growth and density that differ from

predicted effects of competition or predation acting alone.

One reason bass were able to shift to piscivory in their ﬁrst year was due to the fact that
bass spawn before bluegill. The evolution and maintenance of asynchronous spawning of
predator and prey was examined using a game theory approach. With sufficient
environmental uncertainty early in the season, bass (and other predators) behave optimally
by spawning early whereas for bluegill and other prey the optimal reproductive strategy is
to wait until later in the season. However, when conditions are predictable, synchronous

spawning is predicted.

ACKNOWLEDGEMENTS

To begin, I wish to express my sincere gratitude to my advisor, Dr. Gary Mittelbach.
Without his insight, encouragement, and support, this work would not have been possible.
I would also like to thank Dr. Craig Osenberg for his valuable contributions to this
research, and to my development as a scientist. The rest of my committee, Kay Gross,
Don Hall, Susan Kalisz, and Alan Tessier provided support of all kinds throughout my
education. I am grateful for all they have done.

During my time at KBS, I beneﬁtted tremendously from my interactions with other
graduate students. In particular, I thank Kevin Geedey, Peter Smith, Casey Huckins,
Andy Turner, and Jenny Molloy for helping me in so many ways. Casey deserves a
special thank you for the use of his standpipe in the competition experiment. Barbara
Oelslager provided emotional support during the ﬁnal stages of this dissertation, at a time
when I needed it most. Shelby and Asia (my dogs) helped me keep things in perspective.
I would also like to acknowledge René Kane for her support throughout.

I am grateful to the Burdick, Champion, and Nelson families, and to Earl Werner for
providing access to their lakes for my research. John Gorentz and Stephan Oaninski were
extremely helpful in answering all of my computing questions, and I am indebted to

Carolyn Hammarskjold for her assistance in the library.

Financial support for this project was provided by a George H. Lauff Research Award,

The Theodore Roosevelt Fund of the American Museum of Natural History, Sigma Xi
Grants-in-Aid of Research, the MSU chapter of Sigma Xi, the department of zoology, the
graduate program in ecology and evolutionary biology, NSF grants BSR-9207892 and
BSR—9208824 to Mittelbach and Osenberg, and the KBS graduate Research Training
Group (RTG) funded by NSF grant DIR-9113598.

Finally, for all they have done for me, I would like to thank my parents Barbara and

Alton Olson.

TABLE OF CONTENTS

Page
LIST OF TABLES .............................................. viii
LIST OF FIGURES ............................................. xxi
CHAPTER 1 ONT‘OGENETIC NICHE SHIFTS IN LARGEMOUTH BASS:
VARIABILITY AND CONSEQUENCES FOR FIRST YEAR
GROWTH ......................................... 1
INTRODUCTION ........................................ 2
METHODS ............................................. 4
RESULTS ............................................. 7
DISCUSSION .......................................... 45
CHAPTER 2 COMPETITION BETWEEN PREDATOR AND PREY:
RESOURCE-BASED MECHANISMS AND IMPLICATIONS
FOR STAGE-STRUCTURED DYNAMICS ................. 62
INTRODUCTION ....................................... 63
METHODS ........................................... 67
Competition Experiment ................................ 67
Lake Study .......................................... 73
RESULTS ............................................ 74
Competition Experiment ................................. 74
Lake Study ......................................... 106
DISCUSSION ......................................... 113
Competition Experiment ............................... 1 13
Lake Study ......................................... 118
CHAPTER 3 ASYNCHRONOUS SPAWNING IN PREDATOR-PREY FISH
COMMUNITIES AND THE ROLE. OF ENVIRONMENTAL
UNCERTAINTY .................................... 127
INTRODUCTION ....................................... 128

Page

THE MODEL. ......................................... 130
PREDATOR PAYOFF MATRIX ............................. 132
PREYPAYOFFMATRIX.......... ....................... 135
PREDATOR STRATEGIES ................................ 139
PREY STRATEGIES .................................... 144
DISCUSSION ......................................... 149
LIST OF REFERENCES ......................................... 154

vii

Table

LIST OF TABLES

Page

Diet compositions of YOY bass in four lakes in southwest Michigan in 1991.
Values reported are mean percent contributions of each prey category to the

total dry mass of prey of an individual. Sample sizes for three size classes,

15-25, 25-35, and 35-45 mm are listed in parentheses under lake names.

Prey categories are as follows: S+A=§jmgg§phajg§ and amphipods;
Zmps=copepods and W Insect Nymphs=odonate and
ephemeropteran nymphs ..................................... 13

Adjusted mean prey mass in stomach contents for bass feeding in the
invertebrate feeding stage in 1991. Combined prey mass is the sum of the

two important prey categories: crustaceans and insect nymph prey. Only bass
that fed entirely on invertebrates were used in the analysis. The upper limit of
bass sizes used in this analysis was 45 mm SL: above that length most bass in
T13 had shifted to piscivory. Data were analyzed by ANCOVA, using lake as
the grouping variable and ﬁsh standard length as the covariate. There was no
significant heterogeneity of slopes for either prey category or total mass
(crustaceans: F3,119=2.60, p>0.05, insect nymphs: F3,119=2.22, p>0.05, total
prey: F3,119=2.41, p>0.05). Models without interactions were signiﬁcant for
lake and standard length effects in insect nymph and total prey mass, but not
for crustacean prey (crustaceans: lake: F3,122=1.03, p>0.05, standard length:
F1,122=0.72, p>0.05: Insect nymphs: lake: F3,122=5.98, p<0.001, standard
length: F1,122=14.68, p<0.0005: Total prey: lake: F3,122=8.33, p<0.0001,
standard length: F1,122=18.15, p<0.0001) ........................ 16

Regression equations of bass mass (M) vs. time (t) for the invertebrate feeding
phase of bass growth in 1991. Regressions are based on dates in which >50%
of the population was feeding on invertebrates. All regressions are signiﬁcant at
p<0.005: relatively low r2 are due to the fact that there was substantial variation
in bass size on any given date. Slopes represent growth rates in units of g/day.
Analysis of Covariance for bass mass over time is highly signiﬁcant

(F7 557:18784, p<0.0001). Main effects of lake and time, as well as the
interaction, were all signiﬁcant (Lake: F3 557:3.49, df=3, p<0.02, Time:
PMs-1:37.28, p<0.0001, Interaction: F3 557:1 1.91, p<0.0001) ......... 18

Table

Page

Regression equations of 1991 bass mass (M) vs. time (t) for the piscivorous
feeding phase of bass growth in Fig. 7. All regressions are signiﬁcant at
p<0.005: relatively low r2 are due to the fact that there was substantial

variation in bass size on any given date. Slopes represent grth rates in

units of g/day. Intersection dates are used as estimates of the time when bass
shift to piscivory based on the intersection of linear equations describing the
invertebrate feeding stage (Table 3) and equations for the piscivorous phase.
Observed switch dates represent the ﬁrst sample in which >50% of the

population was piscivorous. Lawrence Lake has no equation because the
majority (>50%) of the population never shifted to piscivory ............ 32

Comparison of early bass growth rates in 1991 and 1992. Growth rates are
estimated as slopes of regressions of mass of individual bass vs. time during
the invertebrate feeding stage. (i.e. until >50% of the population switched to
piscivory). Slopes for 1991 were taken from Table 3. All regressions are
signiﬁcant at p<0.0005 level. Analysis of Covariance using years as the
grouping variable and time as the covariate is highly signiﬁcant (F354=104.11,
p<0.0001). Year effects were insigniﬁcant, but both the main effect of time
and the interaction were signiﬁcant (Year: F154=1.22, p>0.10 ,Time:
F154=269.30, p<0.0001, Interaction: F154=37.85, p<0.0001). The lack of
a signiﬁcant year effect implies that bass were similar sized as of June 1 in the
two years. A signiﬁcant interaction term indicates average growth rates were
different between the two years. .............................. 36

Initial and ﬁnal treatment densities. Initial treatments represent the numbers

of neighbor ﬁsh stocked. In addition, all sections received 50 target bluegill

and 70 target bass. Final densities are total numbers recovered at the end of

the experiment. Sections were numbered consecutively, starting with the
southernmost section and moving clockwise ....................... 69

Taxonomic diet compositions of bass and bluegill at the end of the experiment.
Numbers represent mean percent contribution by massﬂ standard error for all
sections combined (n=8). Means were ﬁrst calculated for each section based
on the following numbers of bass, target bluegill, and neighbor bluegill
respectively: section 1: 15, 10, 0, section 2:7, 10, 20, section 3:5, 11, 0,
section 4:9, 24,20, section 5:12, 16, 0, section 6:5, 24, 20, section 7: 8, 29, 0,
section 8: 3, 17, 20. Small cladocerans represent Mom; and
931W. Bass prey represent the combined totals of Simocephalus,
dipteran pupae, and Baetid, Coenagrionid, Aeschnid, and Libellulid nymphs.

Snails were represented by two groups: m m and M spp.
Miscellaneous prey represent all other prey ....................... 82

Table

10

11

Page

Mean body lengths (:1 SE)of small bodied cladocerans and calanoid copepods,
the two dominant zooplankton taxa, at the end of the experiment. Body length

of both taxa were signiﬁcantly different among treatments (small cladocerans:
F3,4=41.40, p<0.002, calanoid copepods: F3,4=34.54, p<0.003). Different
letters indicate means were signiﬁcantly different at p<0.05 (Bonferroni

T test) ................................................. 92

Fmal mean body lengths (:1: 1 SE) of the ﬁve vegetation dwelling invertebrate
taxa that showed signiﬁcant treatment effects (baetid nymphs: F3,4=8.61,

p<0.05, leptocerid larvae: F3,4=7.27, p<0.05, tanypodid larvae: F3,4=9.23,
p<0.05, W: F3,4=41.47, p<0.05, and chironomid larvae which

were marginally signiﬁcant: F3,4=5.52, p<0.07) .................... 101

Growth rates and ﬁsh densities of bass and bluegill in seven lakes. All grth
rates are measured in g/year; bluegill densities are in #/100 m2 and bass

densities are in CPUE’s (#/seine haul). Growth rates and densities of small
(20-50 mm SL) and large (60-100 mm SL) bluegill were taken from Mittelbach
and Osenberg (1994). Growth rates of YOY bass were estimated as the average
mass at the end of their ﬁrst year. Growth rates of large bass were estimated

by first regressing log change in mass over the course of a year vs. log mass at
the start of the year for each lake. Then, growth rates were determined for the
mean mass of a 3 year old bass (85.73 g). Densities of bass were estimated as
catch-per-unit-efforts from beach seines .......................... 120

Correlations among ﬁsh densities and growth rates given in Table 5.

Correlations involving YOY bass denstiy have N=5 and are signiﬁcant at

p<0 0.5 when r < -0. 87 or r > 0.87, all others have N=7 and are signiﬁcant
(p<0..05)whenr<-075 orr>0.75 ............................ 122

LIST OF FIGURES

Figure Page

1 Back-calculated bass sizes ag age 1. Each point represents mean bass
size of at least 3 bass. Lake and year effects explain a signiﬁcant amount
of the observed variation in an analysis of variance. The overall model is
highly signiﬁcant (F42,591=15.48, p<0.0001). As well, lake, year, and
the interactions terms are also signiﬁcant (Lake: F6’691=52.46, p<0.0001,
Year: F6,591=2.93, p<0.008, Interaction: F30,691=2.81, p<0.0001) ......... 8

2 Typical diets of YOY bass, in terms of percent composition by weight.
Values reported are means of the 4 lakes, for six 10 mm size classes except
for the 15-25 mm size class which lacked data from TL3. Crustaceans
include all copepods, cladocerans (including W, and amphipods;
nymphs are odonate and ephemeropteran nymphs; ﬁsh are primarily
YOY bluegill but also include cyprinids ........................... 11

3 Frequency of piscivores in YOY bass cohorts over time in 1991. Frequencies
were calculated from diet analysis. For any given date, the piscivorous
component was assumed to be all bass in a cumulative frequency distribution
that were larger than the smallest bass with a ﬁsh in its stomach. The
appearance of YOY bluegill was estimated as the ﬁrst day in which bluegill

appeared in seines. The appearance of these bluegill was synchronous
among lakes .............................................. 19

4 Changes in YOY bluegill cohort over time in 1991. Standard lengths were
measured to the nearest mm on 40 haphazardly chosen individuals. Results
from an analysis of covariance showed that the overall model was highly
signiﬁcant (F7,1055=117.98, p<0.0001). This signiﬁcance can be partitioned
into a lake effect (F3,1055=6.00, p<0.0005), a time effect (F1,1055=721.84,
p<0.0001), and an interaction of these two effects (F3,1055=3.82, p<0.001). . . 22

5 YOY bluegill availability over time for 1991. Each bass in a cohort was
assigned an individual percent based on the proportion of the bluegill cohort
collected on the same date that was less that 40% the bass standard length.
Mean availabilities were calculated as the mean of all bass. Analysis of
covariance indicates that the overall model is signiﬁcant (F7,512=50.25,
p<0.001) as are both main effects (Lake: F3,512=5.18, p<0.002, Time:
F1,512=19.42, p<0.001). The interaction was not signiﬁcant
(F3,612=1.68, p>0.10) ...................................... 24

Figure Page

6 Percent of bluegill in YOY bass diets as a function of bluegill availability.
Data are based on bass stomach contents. For each bass, a value was
calculated to represent the % of the bluegill cohort that was less than 40%
of the bass size (i.e. % available to bass) from that lake and date. This
number was paired with the percentagbe of the bass diet comprised of
bluegill. Each point represents mean % bluegill in the diet for a 10% class
of bluegill availability. Analysis of covariance on means for each lake
show that the overall model is signiﬁcant, but does not differ among lakes
(overall model: F7 33:11.33, p<0.0001, lake effect: F3,23=0.34, p>0.10,
percent effect: F133=21.24, p<0.0001, interaction: F3 23:1.86, p>0.10).
A sigmoid function has been ﬁtted to the data using the equation:
%die.=Ymu/(1=exp(a+b*%mi.)), where Yum, a, and b are estimated
parameters (me=77.44, a=3.781, b=-0.1185) ..................... 27

7 Invertebrate and piscivorous growth phases of YOY bass in 1991. Solid lines
represent grth rates in the invertebrate stage, dashed lines represent the
piscivorous stage. Each point represents mean bass mass. Regression lines
are in Table 3 and Table 4 .................................... 30

8 Mean water temperatures of Lawrence Lake. Points represent average
temperature readings for 1, 2, and 3 m depths taken at biweekly (1991) or
monthly (1992) intervals. Quadratic equations have been ﬁt to each year
(1991: y=-0.002x2+0.36x+9.72, r2=0.971, 1992: y=-0.001x2+0.17x+
12.67, r2=0.943). Data were provided by CK. Geedey ............... 34

9 YOY bluegill availability over time for 1992. Data were collected and
presented as in Fig. 5. Analysis of covariance is signiﬁcant (F7 259:8720,
p<0.0001); however only the lake effect is signiﬁcant this year (lake effect:
F3259=5.88, p<0.001, time effect: F1359=0.69, p>0.10, interaction:
F3,269=0.53, p>0.10) ...................................... 37

10 Invertebrate and piscivorous grth phases of YOY bass in 1992. As in
Fig. 7, solid lines represent the invertebrate stage and dashed lines represent
the piscivorous stage. Regression equations for the invertebrate feeding phase
are found in Table 4. Only TL3 showed a piscivorous growth phase. The
equation for this line is M=-2.189+0.0070(t), r2=0.200, n=39, p<0.005.
Using this equation, the intersection of the two growth phases in TL3 is at
53.4 days after June 1; the observed diet switch occurred at 65 days ....... 4O

11 A) Bass size at age 1 versus bluegill size at age 1. Bass data are from Fig. 1.
Bluegill sizes were back-calculated for each of the lake/year combinations in
which there were data for bass. B) Bass size at age 1 versus % availability of
bluegill. Bluegill availability was calculated from the size-frequency distribution
generated for each lake/year combination. Each pont represents the average of
% availabilities from each bass, using 40% of their length as the maximum
vulnerable bluegill size ...................................... 43

xii

Figure Page

12 Schematic representation of changes in bluegill vulnerability with bass
size. Cumulative vulnerability is based on an hypothetical normally
distributed bluegill cohort. Bars represent the proportional change in bluegill
vulnerability for a 5 mm reduction of bass size. Three scenarios are

presented, representing reductions of large (descending diagonal bar),

intermediate (hatched bar), and small (ascending diagonal bar) bass. A 5

mm reduction in bass size corresponds to a 6.5%, 38.2%, and 6.5%

reduction in % bluegill available respectively ........................ 54

 

13 Schematic representation of the interaction between largemouth bass and
bluegill. Arrows connecting stages represent the processes of production
and recruitment. Arrows connecting ﬁsh to their resources represent
consumption .............................................. 65

14 Changes in mean ﬁsh mass (:1 SE) over time for bass (A) and target
bluegill (B). Filled circles represent the target only treatment; open squares
represent the bass treatment; ﬁlled triangles represent the low bluegill treatment;
and ﬁlled squares represent the high bluegill treatment. Initial and intermediate
masses were converted from standard length by regression. Final mass was
measured directly. Initial mass did not differ among treatments for either
'species (Bass: F3,4=3.39, p>0.10, Bluegill: F3,4=0.14, p>0.10). Final mass
differed signiﬁcantly for both species (Bass: F3,4=278.60, p>0.0001, Bluegill:
F3,4=241.00, p>0.0001). For bass, Bonferroni T tests show that all ﬁnal
masses in all four treatments are different from one another (p<0.05). Target
bluegill ﬁnal size was signiﬁcantly different among target, low bluegill and
high bluegill treatments (Bonferroni T test: p<0.05), but the low bass treatment
was not signiﬁcantly different from targets alone (Bonferroni T test: p>0.05). . . 76

15 Linear regressions of ﬁnal masses of all bass (A) and target bluegill (B)
vs. ﬁnal bluegill density (targets and neighbors). Data were log-transformed
prior to analysis. Symbols are as in Fig. 14. Regression lines are based on 6
points from the target, low bluegill and high bluegill treatments: A)
log10Y=1.34-0.437(log10X), r2=0.990, B) logloY=1.31-0.375(log10X),
r2=0.996. Open squares represent ﬁnal masses in bass treatments as a function
of bluegill and “neighbor” bass densities above the nominal target density of
bass. The location of these points relative to the regression line gives an
indication of the relative strengths of intra- and interspeciﬁc competition.
In A), low bass points fall on the line, and can not be classiﬁed as outliers
(based on an outlier test p>0.10). In B), growth rates in the bass treatment
are above the line and both points can be classiﬁed as outliers (p<0.05),
suggesting interspeciﬁc competition is weaker than intraspeciﬁc competition
in bluegill ............................................... 78

16 Changes in mean zooplankton density (:1 SE) over time for all taxa
combined.Symbols are as in Fig. 14. Initial densities did not differ
(F3,4=2.16, p>0.10). Final densities differed signiﬁcantly among the 4
treatments (F3,4=231.83, p<0.0001). Bonferroni T tests distinguished 3
groups: target and low bass treatments, low bluegill and high bluegill (p<0.05). 84

xiii

Figure Page

17: Changes in mean densities (:1 SE) of 4 important zooplankton categories
over time. Symbols are as in Fig. 14. A) Chaoborus sp, B) Ma pulex,
C) small bodied cladocerans Qiaphanosoma and Ceriodapﬂia, and D)

calanoid copepods. Note different scales on the Y-axes ................. 86

18 Changes in mean zooplankton body length (:1 SE) over time. Symbols are
as in Fig. 14. Means are based on all zooplankton regardless of species
identity. Initial sizes were similar among treatments (F3,4=1.02, p>0.10).
When the experiment ended, signiﬁcant differences existed among treatments
(F3,4=94.94, p<0.0005). Bonferroni T tests did not ﬁnd signiﬁcant
differences between target and low bass treatments, but the low and high
bluegill treatments were separate from that group and from one
another (p<0.05) .......................................... 89

19 Changes in mean net energy return rates (:1 SE) over time for a bluegill
feeding in the Open water. Symbols are as in Fig. 14. Energy return rates
were calculated in Joules/second (J/s) using a model developed by Mittelbach
(1981). Bluegill sizes used in calculations were chosen to represent bluegill
sizes on each date 7/09:36.5 mm, 7/19, 41.9 mm, 8/04, 41.9 mm. 8/24
48.6 mm) The same sized bluegill was used for all treatments. See text
for details. Treatments did not differ initially (F3,4=2.74, p>0.10). By
the end of the experiment, signiﬁcant differences had emerged (F3,4=9.14,
p<0.03) ................................................ 93

20 Mean body length of zooplankton consumed by ﬁsh on the last date as a
function of bluegill density. Data were log transformed prior to analysis.
Filled circles represent target bluegills, ﬁlled squares are neighbor bluegills,
and open squares are bass. Filled triangles represent mean body length of
zooplankton available to ﬁsh in the environment on the last sampling date.
Three regression lines are also presented. The solid line represents mean
body sizes of zooplankton available in the environment: the dashed line is
based on zooplankton found in target bluegill guts, and the dot-dashed line
represents mean prey size of zooplankton found in bass guts. All three lines
show a signiﬁcant (p<0.05) decline across a gradient of bluegill density
(environment: log10Y=-O.058-0.121(log10X), r2=0.90, bass: logloY=0.034-
0.081(log10X), r2=0.588, bluegill: log10Y=-0.077-0.081(log10X), r2=0.837. 95

21 Changes in the soft-bodied littoral invertebrate community over time.
Symbols are as in Fig. 14. A) Changes in mean density (:1 SE) of all
taxa combined. Initially, treatments did not differ (F3,4=1.37, p>0.10).
Final densities were also not signiﬁcantly different from one another
(F3,4=1.91, p>0.10). B) Changes in mean body length (:1 SE) of all taxa
combined. There were no signiﬁcant differences when the experiment began
(F3,4=0.25, p>0.10). When the experiment ended there were signiﬁcant
differences among treatments (F3,4=14.07, p<0.02) .................. 98

xiv

Figure Page

22 Changes in mean net energy return rates (:1 SE) over time for a bluegill
foraging in the littoral zone. Symbols are as in Fig. 14. Bluegill sizes are
the same as in calculations used for Fig. 6 (36.5, 41.9, and 48.6 mm SL
for the 3 sample dates). Initial foraging return rates did not differ among
treatments (F3,4=2.74, p>0.10). Final rates were signiﬁcantly different
(F 3,4=9.76, p<0.03) ....................................... 102

23 Mean sizes of littoral prey consumed by ﬁsh at the end of the experiment
as a function of bluegill density. Data were log transformed prior to analysis.
Symbols of points and regression lines are as in Fig. 20. All regressions are
signiﬁcant (p<0.05) and show a decline in prey size with increasing bluegill
density (environment: log10Y=0.698-0.077(log10X), r2=0.92, bass:
log10Y=0.747-0.100(logloX), r2=0.53, bluegill: log10Y=0.921-
0.112(log10X), r2=0.72 ..................................... 104

24 Mean sizes of important bass prey (Simocephalus, dipteran pupae, Baetid
nymphs, Coenagrionid nymphs, Libellulid nymphs, and Aeschnid nymphs)
consumed by ﬁsh at the end of the experiment as a function of bluegill density.
Data were log transformed prior to analysis. Symbols of points and regression
lines are as in Fig. 20. All regressions show a signiﬁcant (p<0.05) decline
in prey size with increasing bluegill density (environment: log10Y=0.630-
0.073(log10X), r2=0.89, bass: log10Y=0.747-0.100(logloX), r2=0.53,
bluegill: logloY=0.905-0.226(log10X), r2=0.87 .................... 107

25 Small bass growth rates in natural populations as a function of small bluegill
growth and density. Bass growth is expressed as mean mass at age 1.
Bluegill data were taken from Mittelbach and Osenberg 1994. A) Small bass
growth vs small bluegill growth (r=0.93, n=7, p<0.003). B) Small bass
growth vs. small bluegill density (r=-0.95, n=7, p<0.001) ............. 109

26 Large bass growth rates in natural populations as a function of small bluegill
denstiy. Bass growth rates are expressed as change in mass (g/yr), calculated
from mass speciﬁc growth rate regressions calculated for each lake. Growth
rates reported are estimate dfor a mean sized three year old bass (85.73 g).

The relationship between large bass growth and small bleugill dneisty is
signiﬁcant (r=0.93, n=7, p<0.003) ............................. 111

27 Predator payoff matrix. The two player's potential strategies (spawning
early or late) are crossed, giving 3 2X2 matrix. Each entry represents the
predator's relative payoff resulting from a given combination of predator and
prey strategies. When predators spawn early, the resulting payoff must be
weighted by p, the proportion of good years. The letters A through D
represent different potential outcomes ........................... 133

28 Prey payoff matrix. This matrix is set up identically to Figure 27. The
letters W through Z represent different relative payoffs for the prey ....... 136

Figure

29

30

Page

Relative predator payoffs across a gradient of p values. The two lines
represent the potential payoffs when prey spawn early or late. A) payoffs
when prey are spawning early. The two lines cross when p = (C-D)/(B-D).

At this point, strategies should switch from late to early. B) payoffs when
prey are spawning late. The p value where the strategy changes is

p = (B-D)/(A-D) ......................................... 0

Relative prey payoffs across a gradient of p values. The two lines represent

the potential payoffs when prey spawn early or late.A) payoffs when predators
are spawning early. Prey should switch from late to early spawning when p =
(W-Z)/(W+X-Y-Z). B) payoffs when predators are spawning late. The

critical value where the strategy changes is p = (X-Z)/(W-X) ............ 145

CHAPTER 1
ONTOGENETIC NICHE SHIFTS IN LARGEMOUTH BASS: VARIABILITY
AND CONSEQUENCES FOR FIRST YEAR GROWTH

2

INTRODUCTION

Ontogenetic niche shifts, deﬁned as changes in resource or habitat use during the course
of an individuals lifetime, are a prevalent feature in the ecology of many size-structured
populations (Werner and Gilliam 1984). For example, bluegill sunﬁsh (Lemmis
macrochirus) switch from feeding in the vegetated littoral zone to the open-water limnetic
zone as they grow (Mittelbach 1981). This shift is related to a change in the risk of
predation: individuals avoid the more proﬁtable but riskier limnetic zone until they reach a
size where they are relatively immune to predation (Werner et al. 1983 a,b). In other
systems, size related diet shifts are the result of size-speciﬁc changes in foraging ability,
which enable individuals to capture and consume progressively larger prey (Wilson 1975,
Keast 1977, Werner 1986, Polis and McCormick 1986, Osenberg and Mittelbach 1989,
Persson 1990).

When ontogenetic niche shifts are relatively discrete, a population can potentially be
divided into separate size classes (or stages) based on diet or habitat use (Nisbet et al.
1989, Osenberg et al. 1993). Associated changes in the nature and intensity of intra- and
interspeciﬁc interactions such as competition or predation can result in each stage playing a
different role in community or ecosystem level processes to a point where each stage is a
functionally different “ecological species.” However, separate stages are not independent
of one another: they are linked together by processes of production and recruitment.
Factors such as resource productivity or predation intensity that affect one stage are
therefore transmitted to other stages, and eventually feed back to the original stage
(Tschumy 1982, Mittelbach and Chesson 1987, Osenberg et al. 1992, Mittelbach and
Osenberg 1994). As a corollary, dynamics within a stage are affected not only by
processes operating in that stage, but by processes operating in other stages as well.

Consequently, overall p0pulation dynamics in stage-structured populations are inﬂuenced

3

by a wide variety of mechanisms, each acting on a subset of the population.

The complexity created by the division of a population into distinct stages presents a
formidable challenge to the study of these types of organisms (Werner 1988, Osenberg et
al. 1993). Recent work on stage-structured populations has focussed on factors that
operate speciﬁcally within a single life-stage (Polis and McCormick 1987, Wilbur 1988,
Jones 1988, Persson and Greenburg 1990) However, with a few exceptions (e.g.
Mittelbach et al. 1988, Neill 1988, McCauley et al. 1991, Mittelbach and Osenberg 1994),
our understanding of how these factors influence transitions between stages only comes
from theoretical work (Tschumy 1986, Prout 1986, Mittelbach and Chesson 1987).
Clearly, there is a need for more empirical studies on stage-structured populations, with an

emphasis on factors that affect the shift from one niche/stage to the next.

The largemouth bass (Micropterus salmoides, henceforth bass) provides an excellent
opportunity to study niche shifts in a size-structured population. Adult bass are the top
predators in lakes and ponds throughout eastern North America. However, bass begin
their lives feeding on invertebrates and typically make the transition to piscivory sometime
in their ﬁrst year. In Michigan lakes, the shift to piscivory occurs when young-of-year
(Y OY) bluegill become available in the littoral zone (bluegill dominate the biomass of ﬁsh
communities in these lakes and are a major prey species for bass Swingle and Smith 1940,
Dillard and Novinger 1975, Werner et al. 1977 ). Because bluegill spawn after bass and
bluegill fry spend their ﬁrst few weeks of life offshore before migrating into the littoral
zone (Werner 1967, Werner and Hall 1988), there is generally a 1-2 month difference

between the time bass are born and YOY bluegill arrive inshore.

After YOY bluegill arrive in the littoral habitat, bass can only shift to piscivory if the

4

relative sizes of bass and bluegill exceeds a critical value. Bass are gape-limited predators
and thus require a size advantage over their prey in order to consume them (T immons et al.
1980, Hambright 1991). Otherwise, YOY bass do not shift and instead continue to feed on
invertebrates (Keast and Eadie 1985). It is commonly believed, although few data exist,
that the shift to piscivory increases ﬁrst year growth of bass (Timmons et al. 1980, Adams
and DeAngelis 1987). This is important because overwinter survival in bass is strongly size
dependent, (Davies et al. 1982, Gutreuter and Anderson 1985) and small changes in early
growth rates can have profound implications for individual survivorship, and for the
dynamics and size-structure of the population as a whole. In this study, I found that
growth rates of bass cohorts were highly variable among lakes and years, and that this
variation in grth was very sensitive to the timing of the shift to piscivory. Several
factors were found which inﬂuenced the size advantage of bass when bluegill ﬁrst became
available, and thus affected the shift. As a result, subtle differences among lakes or years

lead to substantial differences in bass size at the end of their ﬁrst growing season.
METHODS

Patterns of YOY bass growth and diet were examined in a set of seven natural lakes in
southwest Michigan (Culver, Deep, Lawrence, Palmatier, Three Lakes 11 (henceforth
TL2), Three Lakes III (TL3) and Warner). All seven lakes are located within 30 km of the
Kellogg Biological Station and are similar in water chemistry and productivity (e.g. pH,
total phosporous: see Osenberg et al. 1988, Mittelbach and Osenberg 1994). These lakes
are also similar in size (5-26 ha) and depth (10-16 m), except for TL3 which has a depth of
4 m. Because TL3 is so shallow, it is the only lake that dOes not thermally stratify in
summer. Despite this, mean epilimnetic and littoral water temperatures are very similar

among lakes (lake temperatures differed by < 2° C on any given date: Olson unpublished

5

data). All lakes are dominated by ﬁshes in the family Centrarchidae, especially bluegill,
although cyprinids are also numerically abundant (Werner et al. 1977). Largemouth bass

are by far the dominant piscivore, representing well over 90% of the guild (Werner and

Hall 1988, Olson unpublished data).

Among-lake variation in YOY bass growth was determined by back-calculation from
scale annuli. Bass (1 to 5 years in age) were collected by seining and/or angling each lake
from May to September between 1990 and 1992. Standard lengths (the length from snout
to caudal terminus of the vertebral column) were measured to the nearest millimeter and 5
scales were collected just posterior to the depressed left pectoral ﬁn. Impressions of these
scales were made on cellulose acetate strips which were then projected with a microﬁche
reader. Each bass was aged to determine its year of birth, and distances from the focus to
the ﬁrst annulus and scale edge were measured from 1 non-regenerated scale per ﬁsh. First
annulus lengths were converted to standard lengths at age 1 by the Fraser-Lee method
(Tesch 1968) using 16.5 as the intercept of the standard length-scale length regression.
Mean standard lengths at age 1 were calculated for each lake/year combination when n23.

Means were based on an average of 19.1 ﬁsh, with a range of 3 to 52.

Bluegill are the major prey species for bass (Swingle and Smith 1940, Bennett 1948,
Dillard and Novinger 1975, Keast 1985) and their vulnerability to bass predation is
strongly size-dependent. Therefore, I also determined bluegill sizes after their ﬁrst
growing season. The scale length intercept used in calculations was 11.9 (taken from
Osenberg et al. 1988). Mean bluegill sizes at age 1 and cumulative size frequency
distributions were generated for each combination of lakes and years for which there were
bass data. Cumulative size frequency distributions were then used to quantify bluegill

availabilities for individual bass. All bluegill less than 40% the bass standard length were

6

considered to be vulnerable to predation (Lawrence 1958, Werner 1977, Timmons et al.
1980). Because bluegill were much more abundant than bass, large sample sizes could be

collected for each lake/year combination 07:80; rangez4-563).

To better understand how growth rates of YOY bass differed among lakes, cohort size
distributions were tracked through the summer in four of the seven study lakes in 1991 and
1992 (Lawrence, TL’Z, TL3 and Warner). YOY bass were collected with a beach seine
(23m X 1.8m; 3.2 mm mesh) at roughly 10 day intervals from the time of ﬁrst spawning
until mid-September. Mean bass size for each collection was calculated based on standard
length measurements (to the nearest mm) of an average of 24.1 ﬁsh (range 4-50). Most of
the bass were released after being measured. Regressions were generated to convert
standard lengths to wet masses. The regressions for each lake were:

LawrenCe: Mass=0.00002566(SLZ-955): N=87, r2=0.99

TL2: Mass=0.00002375(SL2.977): N=59, r2=0.99

TL3: Mass=0.00002019(SL3-020): N=70, r2=0.99

Warner: Mass=0.00003475(SL2-876): N=83, r2=0.99.

Growth rates for bass cohorts in 1991 and 1992 were estimated as the slope of the

appropriate linear regressions of mass vs. time for each lake.

A subsample of the bass collected in 1991 and 1992 were preserved in 10% neutral
formalin, and later analyzed for stomach contents. Prey were identiﬁed to the lowest
taxonomic level possible (typically to genus or family), counted, and a subset were
measured (up to 20 haphazardly chosen individuals). Lengths were used to convert prey
counts to dry mass using length-mass regressions for each taxonomic category (Mittelbach
and Osenberg, unpublished data; Olson, unpublished data). Diet compositions are

expressed as percents of total prey mass.

7

Size distributions of YOY bluegill cohorts were measured in 1991 and 1992 starting
when they appeared in the littoral zone until mid-September. Standard lengths (to the
nearest mm) were recorded for bluegill collected on the same days as YOY bass were
collected. Mean sizes and cumulative size frequency distributions were calculated based on
40 haphazardly chosen individuals (sample sizes in T12 in 1992 were smaller (mean=8.3)

because YOY bluegill were very rare).
RESULTS

Analysis of bass sizes at the end of their ﬁrst year showed pronounced variation among
lakes and strong year-to-year variation within some of the lakes (Fig 1). The effects of
lakes and years were highly signiﬁcant as was their interaction (Fig. 1). This interaction
arose because lakes differed in their sensitivity to year effects. The ranking of lakes,
however, was fairly consistent: bass were always largest in Palmatier and TL3, followed
by Culver, Deep, and Lawrence (although the rankings within the three varied) and ﬁnally
T12 and Warner, which typically had the smallest bass.

The observed lake-to—lake variance in size of year 1 bass is potentially the result of three
factors: 1) differential growth, 2) size-selective mortality, or 3) different spawning times.
Variation in growth rates among lakes appears to be the major factor driving the pattern
observed in Fig. 1. Variation in selective mortality of small bass could have created
different mean sizes among lakes even if all lakes started with identical size distributions.
However, if this was the only mechanism operating, we would expect a negative
correlation between mean and variance: instead a weak, but positive correlation was
observed (r=0.162, p=0.31) Variation in bass size was not the result of different spawning

times by adults. Since bass spawning is temperature dependent (Scott and Crossman

Figure 1. Back-calculated bass sizes ag age 1. Each point represents mean bass
size of at least 3 bass. Lake and year effects explain a signiﬁcant amount of the
observed variation in an analysis of variance. The overall model is highly
signiﬁcant (F42,591=15.48, p<0.0001). As well, lake, year, and ther interactions
terms are also signiﬁcant (Lake: F6,691=52.46, p<0.0001, Year: F5,591=2.93,
p<0.008, Interaction: F30,591=2.81, p<0.0001).

Nmma

seam wwaa ewaa
. .

vwau

 

 

 

Lucas? .11i$11
a 5.1....

“WEE 1110111
saga in..-
8553 '0'.

 

 

moon— 1 10 1 1
.526 1.14.1.1
8:3

 

 

Q \
.\
\..\ II a
1\. I I .s.
V I I \.\
. I .o
.s I .I .s.
\
.s o a. s.
I I s.
s I ‘
s I
s I
.s a
s
s. I
.~ I
a a
4 .
I
D I
\ 1 o 0
ﬂ \\ ’0
000 \o I
.. MD". .I
000 so \ II 0
00 I‘§\Q 0' ’
o s s c! \o 000 a
U
‘ .\‘ '0' UUUUUUU ’0.
q 4.. .9
‘.
.7.
.4010
00/0

on

1mm

1 1 1
E S 8
(In!!!) I 98v 12 11131171 pmpmns 8828

f
V1
[\

 

 

Figure 1

10

1979), lakes that warm up quickly would have an earlier onset of reproduction and
offspring in these lakes could be larger simply due to a longer growing season. However,
within this set of lakes there was little variation in littoral zone water temperatures during
the spawning season (<2° C). Furthermore, the variation was opposite of expectations:
mean bass sizes at the end of year 1 were negatively correlated with water temperatures in
late May (r=-0.67, p=0.10). The similarity among lakes in temperature also precludes the
possibility that growth rate variation was due to differences in average temperatures among
lakes. The most likely explanation for the observed differences in YOY bass growth are

among lakes differences in prey availability and energy intake.

To investigate the role of prey resources in determining YOY bass growth, I followed in
detail the diets and grth of YOY bass in 4 of the study lakes (Lawrence, TL2, TLB, and
Warner) in 1991 and 1992. Patterns differed between years, and I discuss the results from
1991 first. The typical pattern observed was for bass to change diets with size (Fig. 2).
The most dramatic diet shift occurred when bass were 40-50 mmvSL. Above that size,
bass were predominately piscivorous, feeding especially on YOY bluegill which made up
80% of the ﬁsh prey. The diets of bass below that 40-50 mm SL were composed primarily
of invertebrates. Within this invertebrate feeding stage, two prey types were most
important (Fig. 2). Small crustaceans, including copepods, pelagic cladocerans,
Simocephalus, and amphipods, dominated the diets of small bass, but by the time bass
reached 30 mm SL insect nymphs (primarily odonates and ephemeropterans) made up the
majority of the diet. Chironomid larvae also appeared in small amounts in diets. This prey
category was most important to the smallest size classes of bass, but never made up more

than 20% of the diet.

This general dietary pattern in the invertebrate feeding stage was found in all 4 study

11

Figure 2. Typical diets of YOY bass, in terms of percent composition by weight.
Values reported are means of the 4 lakes, for six 10 mm size classes except for the
15-25 mm size class which lacked data from TL3. Crustaceans include all
copepods, cladocerans (including Simocephalus), and amphipods; nymphs are
odonate and ephemeropteran nymphs; ﬁsh are primarily YOY bluegill but also
include cyprinids.

12

 

 

 

 

 

 

 

 

    

 

100
-—O— Crustaceans
(ZOOpIankton, Amphipods -
mamas.) I
---I--- Chironomids
75 - ----A---- Nymphs
- - O- - Fish
I ’ .~ -
I I I ’ T .
£5. ’
a T .'
‘fg 50 ,A.
s 1'-
s .
£ I
o "‘
.5. 1
o . : ~21—
10 20 30 40 50 60 70

Standard Length (mm)

 

13

Table 1: Diet compositions of YOY bass in four lakes in southwest Michigan in
1991. Values reported are mean percent contributions of each prey category to the
total dry mass of prey of an individual. Sample sizes for three size classes, 15-25,
25-35, and 35-45 mm are listed in parentheses under lake names. Prey categories
are as follows: S+A=Simocephalus and amphipods; Zoops=copepods and Daphnia
galeata; Insect Nymphs=odonate and ephemeropteran nymphs.

14

Woe md od
Ewe m6 wdﬁ
ﬁg 56 1mm

w.Nm HA m.mm

23:32 mqooN <+m
835 888680

SE 3.2

v.3 mdv 9.?
ed... md aﬁN
an»... ad mém

v.9. ca Ndm

mean—>2 SOON <+w
.03.:— 288620

88 3.3

mum—U oﬁm

ed Ewm ad
<2 <2 <2
3: mA wen
HSN wd hem
mamamz aeoN <+m
.82: 83053.0
88 3-3

@336
3:83

3.2.8
3...
9.33
S...

E .3 .3
35:33

83

Table 1

15

lakes (Table 1). The only qualitative difference in diets among lakes was in Warner
Lake when bass fed on crustaceans. Unlike the other 3 lakes, in this lake bass
rarely consumed Simocephalus or amphipods. Instead, small bass fed on
zooplankton, especially mania galLata, which was the dominant prey item of bass
up to 35 mm SL where insect nymphs became important (Table 1). In general,
however, bass diets in the invertebrate feeding stage were fairly similar among
lakes: diets were quite narrow and tended towards specializing on only a limited

range of prey types (mean diet breadth, using Pianka’s Index (1973) was 4.26).

In all lakes, bass consumed more invertebrate prey as they got larger (the main
effect of size on total invertebarate prey biomass: Table 2). Up to lengths of 45
mm, however, bass in TL3 consistently consumed a higher total biomass of prey.
This was the result of their having more crustacean and especially insect nymph
prey in their stomachs. Bass in T12 and Lawrence consumed less invertebrate
biomass across this size range. Although they ate similar amounts of Simocephalus
and amphipod prey compared to TL3 (their adjusted means were lower, but there
were no significantdifferences among lakes for this prey category: Table 2), they
ate far less from the insect nymph category and consequently had a lower total prey
biomass (Table 2). Bass in Warner Lake had only a small amount of prey in their
stomachs. Consumption of I_)_. gm compensated for the lack of Simocephalus or
amphipods to keep bass in Warner close to the other lakes in terms of crustacean
prey biomass. However, they ate far less from the insect nymph category, and as a
result consistently had the lowest biomass of invertebrate prey in their stomachs

(Table 2).

These observed differences in prey consumption among lakes were reﬂected in

16

Table 2: Adjusted mean prey mass in stomach contents for bass feeding in the
invertebrate feeding stage in 1991. Combined prey mass is the sum of the two
important prey categories: crustaceans and insect nymph prey. Only bass that fed
entirely on invertebrates were used in the analysis. The upper limit of bass sizes
used in this analysis was 45 mm SL: above that length most bass in TL3 had shifted
to piscivory. Data were analyzed by ANCOVA, using lake as the grouping variable
and ﬁsh standard length as the covariate. There was no signiﬁcant heterogeneity of
slopes for either prey category or total mass (crustaceans: F3,119=2.60, p>0.05,
insect nymphs: F3,119=2.22, p>0.05, total prey: F3,119=2.41, p>0.05). Models
without interactions were signiﬁcant for lake and standard length effects in insect
nymph and total prey mass, but not for crustacean prey (crustaceans: lake:
F3,122=1.03, p>0.05, standard length: F1,122=0.72, p>0.05: Insect nymphs: lake:
F3,122=5.98, p<0.001, standard length: F1,122=14.68, p<0.0005: Total prey: lake:
F3,122=8.33, p<0.0001, standard length: F1,122=18.15, p<0.0001)

Adjusted Mean Prey Mass
Lake Crustaceans Insect Nymph Combined Mass
TL3 0.543 1.323 1.866
Lawrence 0.364 1.026 _ 1.390
TL2 0.420 0.850 1.270

Warner 0.285 0.199 0.484

17

patterns of YOY bass growth. Analysis of covariance of ﬁsh mass over the time
when most bass fed on invertebrates showed signiﬁcant effects of time and lakes,
and a significant interaction of lake by time (Table 3). The interaction term
demonstrated that slopes were heterogenous, or that grth rates differed among
lakes. Growth rates (deﬁned as the slope of the mass vs. time regression)
calculated for bass feeding on invertebrate resources were highest in TL3, followed
by Lawrence, TL2, and then Warner (Table 3). Because of these differences, bass
sizes quickly diverged among lakes. By early July, bass were at least 55% larger in
TL3 relative to the other lakes (1.63 g compared to 0.90, 0.68 and 0.66 in TL2,
Lawrence and Warner, respectively). Furthermore, bass in T12 and Lawrence took
10 to 15 days longer to reach 1.63 g, and in Warner Lake bass did not reach that
mass until well after July 24, over 3 weeks later than bass in TL3 had reached that
size. Note also that while bass in Warner Lake and Lawrence Lake were similar in
size in early July (0.66 g and 0.68 g), Warner Lake bass grew much more slowly
thereafter, probably because of the lack of insect nymphs in their diet. Adjusted
mass of zooplankton prey in Warner Lake was comparable to the mass of
Simocephalus/amphipod prey in Lawrence Lake when bass in both lakes were
small. Once nymphs became a major component of bass diets, however, Warner
Lake bass had much less prey in their stomachs, and their growth rates were
reduced relative to the other lakes. The result was a clear divergence of bass sizes

through the invertebrate feeding stage.

The growth rate differences established during the invertebrate feeding stage had
dramatic effects on the ability of bass to shift to piscivory in the different lakes.
Owing to their rapid growth while feeding on invertebrates, bass in TL3 were able

to make a successful transition to piscivory as soon as YOY bluegill began to appear

18

Table 3: Regression equations of bass mass (M) vs. time (t) for the invertebrate
feeding phase of bass growth in 1991. Regressions are based on dates in which
>50% of the population was feeding on invertebrates. All regressions are
signiﬁcant at p<0.005: relatively low r2 are due to the fact that there was substantial
variation in bass size on any given date. Slopes represent growth rates in units of
g/day. Analysis of Covariance for bass mass over time is highly signiﬁcant

(F7 557:18784, p<0.0001). Main effects of lake and time, as well as the
interaction, were all signiﬁcant (Lake: F3 557:3.49, df=3, p<0.02, Time:
Pas-1:37.28, p<0.0001, Interaction: F3 57:11.91, p<0.0001).

Lake Equation Standard Error r2 N
of Slope

TL3 M=-0.203+0.061(t) 0.010 0.433 48

Lawrence M=-0.583+0.047(t) 0.002 0.657 ' 194

T12 =-0.413+0.045(t) 0.006 0.360 104

Warner * =0.240+0.029(t) 0.002 0.499 219

19

Figure 3. Frequency of piscivores in YOY bass cohorts over time in 1991.
Frequencies were calculated from diet analysis. For any given date, the piscivorous
component was assumed to be all bass in a cumulative frequency distribution that
were larger than the smallest bass with a ﬁsh in its stomach. The appearance of
YOY bluegill was estimated as the ﬁrst day in which bluegill appeared in seines.
The appearance of these bluegill was synchronous among lakes.

20

1991

 

 

 

 

—0— Lawrence

T12

TL3
--I-- Warner

“*uom

 

 

   

 

 

 

100

u
5
7

$3 88383 mo >05:on

Sept.

August ‘

July 1

Appearance of YOY Bluegill

21

in the littoral zone (Fig. 3). In contrast, bass in the other lakes did not make the switch to
piscivory until later in the season. In T12, bass shifted to piscivory 1-2 weeks later than in
TL3 even though YOY bluegill appeared in the littoral zone at the same time. This effect
was even more pronounced in Warner Lake, where the shift from invertebrates to fish did
not occur until the end of August. Thus, differences in YOY bass growth while feeding on
invertebrates affected the time at which they made the shift to feeding on YOY bluegills.
Because bass cohorts among lakes differed in the timing of the switch to piscivory, the size
at which they made the shift also differed. In T13, bass switched while they were smaller
compared to T12 or Warner (42 mm vs. 46 mm in T12 and 56 mm in Warner). In
Lawrence Lake, bass never made a clear shift to piscivory and invertebrates dominated the
diet throughout the season (Fig. 3). In addition, those ﬁsh consumed by Lawrence Lake
bass were cyprinids; YOY bluegill were never found in bass stomachs. This result was

somewhat surprising given that bass in Lawrence were similar in size compared to bass in

T12.

The size advantage required for bass to shift to piscivory depends not only on bass
growth rates while feeding on invertebrates, but also on the size distribution of YOY
bluegill. Although YOY bluegill arrived in the littoral zone in early July in all 4 lakes, mean
bluegill sizes among lakes were very different (Fig. 4). In addition, the lakes varied in
YOY bluegill growth rates which enhanced the divergence in YOY bluegill size among
lakes. The largest YOY bluegill were in Lawrence lake, where they reached 25 mm SL
close to 1 month before any other lake. In T12, on the other hand, the bluegill were
smaller than in any other lake from the time they ﬁrst arrived in the littoral zone and through

the rest of the season.

These differences in YOY bluegill size among lakes, combined with the differences in

22

Figure 4. Changes in YOY bluegill cohort over time in 1991. Standard lengths
were measured to the nearest mm on 40 haphazardly chosen individuals. Results
from an analysis of covariance showed that the overall model was highly signiﬁcant
(F7,1056=117.98, p<0.0001). This signiﬁcance can be partitioned into a lake effect
(F3,1055=6.00, p<0.0005), a time effect (F1,1055=721.84, p<0.0001), and an
interaction of these two effects (F3,1056=3.82, p<0.001).

- 1-- __-_.__.l

 

23

1991

 

 

27.5 -

 

 

 

i?

Bluegill Standard Length (mm)
E3
UI
l

20-

 

 

 

 

1 7.5 | 1
July August Sept. OCt‘

Figure 4

24

Figure 5. YOY bluegill availability over time for 1991. Each bass in a cohort was
assigned an individual percent based on the proportion of the bluegill cohort
collected on the same date that was less that 40% the bass standard length. Mean
availabilities were calculated as the mean of all bass. Analysis of covariance
indicates that the overall model is signiﬁcant (F7,612=50.25, p<0.001) as are both
main effects (Lake: F3,612=5.18, p<0.002, Time: F1,612=19.42, p<0.001). The
interaction was not signiﬁcant (F3,612=1.68, p>0.10).

25

 

 

 

—0— Lawrence

---O--T12

 

TL3
—-l—- Warner
‘0

....‘....

 

 

1
Sept.

August

 

 

ﬁ - u

m m m
cog—88m $3 2 03805.5 Ewes—m >O> mo cuss—020m

 

July

Figure 5

26

bass growth rate, had important consequences on the availability of bluegill as prey for
bass (Fig. 5). The high growth rate of YOY bluegill in Lawrence Lake resulted in their
being too large to be eaten when they arrived in the littoral zone. Although bass grew faster
than bluegill, most of the YOY bluegill stayed in the size refuge through the course of the
season. The combination of slightly larger bass and much smaller bluegill in T12 led to a
very different situation. The average bass in T12 was capable of consuming around 40%
of the bluegill cohort from the time prey arrived in the littoral zone until the end of the
season. A similar pattern was observed in T13. High growth rates of bass in the
invertebrate stage afforded them a considerable size advantage over YOY bluegill. Even
though YOY bluegill in TL3 were intermediate in size compared to the other three lakes,
close to 60% were available when they ﬁrst migrated inshore. In Warner Lake, bluegill
were invulnerable to has predation when they ﬁrst appeared inshore. Despite being
relatively small, these prey were too large for bass because of the poor growth rates of the
predator during their invertebrate feeding stage. As the season progressed, however, bass
in this lake were able to grow fast enough to gain a size advantage over bluegill and

increase the proportion available as prey to close to 30% by September (Fig. 5).

As the percentage of the YOY bluegill cohort that were vulnerable to YOY bass increased,
the percent of bass that were piscivorous increased (Fig. 6). This relationship did not vary
among the 4 lakes. The lack of a lake effect implied that successful piscivory depended
only on the proportion of the YOY bluegill cohort that was available as food, and was
independent of absolute densities of predator and prey. In T12 and TL3, most bluegill
were vulnerable as soon as they arrived and bass made early transitions to piscivory. The
shift took longer in Warner Lake because it took time for bass to outgrow their prey. In
Lawrence Lake, has never fed on YOY bluegill. The relative sizes of predator and prey

were too close, making most bluegill unavailable to their predator, and bass were never

27

Figure 6. Percent of bluegill in YOY bass diets as a function of bluegill availability.
Data are based on bass stomach contents. For each bass, a value was calculated to
represent the % of the bluegill cohort that was less than 40% of the bass size (i.e. %
available to bass) from that lake and date. This number was paired with the
percentagbe of the bass diet comprised of bluegill. Each point represents mean %
bluegill in the diet for a 10% class of bluegill availability. Analysis of covariance
on means for each lake show that the overall model is signiﬁcant, but does not
differ among lakes (overall model: F7 38:1 1.33, p<0.0001, lake effect:

F3,23=O.34, p>0.10, percent effect: F133=21.24, p<0.0001, interaction:
F333=1.86, p>0.10). A sigmoid function has been ﬁtted to the data using the
equation: %dier=Ym,/(1=exp(a+b*%,Vai|)), where Yum, a, and b are estimated
parameters (me=77.44, a=3.781, b=-0.1185).

 

Percent Bluegill in Diet

28

 

 

 

 

100
o
75 q
o 0 .
50 -
25 -
o
0 l l . l
0 25 50 75

Percentage of Bluegill Cohort Available
Figure 6

 

100

29

able to outgrow bluegill and make the niche shift from invertebrates.

In lakes where bass successfully shifted to piscivory, YOY bass growth followed a two-
phase grth pattern (Fig. 7). The ﬁrst growth phase occurred while the majority of bass
in a cohort bass fed on invertebrates. During this stage, bass growth in TL3 was higher
than the other lakes; Lawrence and T12 were intermediate and in Warner Lake bass grew
slowest (Table 3). The second phase featured enhanced growth rates and started after bass
became pisciVorous. The initiation of this phase (estimated for each lake as the intersection
of the separate growth rate regressions for the invertebrate feeding stage and the
piscivorous stage) closely matched the ﬁrst date in which >50% of the cohort was
piscivorous (Table 4). In TL3, the switch took place in early July. These bass grew at the
enhanced rate during the piscivorous stage for most of the season and therefore reached the
largest ﬁnal size of all 4 study lakes. Bass in T12 shifted a little later, but still experienced
an increased growth rate for most of the season. In Warner Lake, the invertebrate growth
phase lasted most of the season. A jump in growth rates came very late in the season, after
bass ﬁnally grew large enough to consume YOY bluegill. The majority of bass in
Lawrence Lake never shifted to piscivory, and as a result never experienced a second
growth phase. Instead, they continued to grow at a single rate through their ﬁrst year and

ended up smaller than in the other lakes (Fig. 7).

In addition to the variation in bass growth rates observed among lakes, Fig. 1 shows that
there was substantial year-to-year variation in some lakes. This variation was due to both a
main effect of years and an interaction between lakes and years. The interaction term
suggests that lakes differed in their sensitivity to climatic or other year-related differences in
1991 and 1992. This response can be illustrated by comparing dynamics of bass growth in

the same 4 study lakes in 1992 with 1991. Both air and water temperatures were much

30

Figure 7. Invertebrate and piscivorous growth phases of YOY bass in 1991. Solid
lines represent growth rates in the invertebrate stage, dashed lines represent the
piscivorous stage. Each point represents mean bass mass. Regression lines are in
Table 3 and Table 4.

31

 

 

 

 

 

 

 

 

 

1991
10
0 Lawrence
0 TL2 ,-
A TL3 ,’ ‘
I Warner ‘ ’I’
7.5 q XI
’33
:2 5 -
a
2
2:5-
0
June July August Sept.

Figure 7

32

Table 4: Regression equations of 1991 bass mass (M) vs. time (t) for the piscivorous
feeding phase of bass growth in Fig. 7. All regressions are significant at p<0.005:
relatively low r2 are due to the fact that there was substantial variation in bass size on any
given date. Slopes represent growth rates in units of g/day. Intersection dates are used as
estimates of the time when bass shift to piscivory based on the intersection of linear
equations describing the invertebrate feeding stage (Table 3) and equations for the
piscivorous phase. Observed switch dates represent the ﬁrst sample in which >50% of the
population was piscivorous. Lawrence Lake has no equation because the majority (>50%)
of the population never shifted to piscivory.

Standard Error Predicted Observed
Lake Equation of Slope r2 Date of Switch Date of Switch
TL3 M=-1.773+0.105(t) 0.011 0.332 35.2 34
Lawrence NA NA NA NA NA
T12 =-1.975+0.088(t) 0.018 0.174 40.4 53

Warner M=-7.551+0.126(t) 0.040 0.265 76.5 91

0V:
lino

bass

00min

33

colder in 1992 than 1991. Mean monthly air temperatures from May to August were about
5°C cooler (unpublished data from the WK. Kellogg Biological Station). Epilimnetic
water temperatures from Lawrence Lake reﬂected these differences and were also about

5°C cooler in 1992 than 1991 during much of the summer (Fig. 8).

One effect of lower water temperatures in 1992 was to delay the onset of spawning in
bass. Young-of-year bass ﬁrst appeared in beach seines 10 days later in 1992 than 1991
(bass first appeared in 1991 on June 4 compared to June 14 in 1992). In addition, YOY
bass were much smaller when they ﬁrst appeared in 1992 because early growth rates were
reduced. In all four lakes, growth rates in the invertebrate feeding stage were signiﬁcantly
lower in 1992 than 1991, and again the lakes differed in growth rates during this stage
(Table 5). Bass in TL3 had the highest growth rates followed by Lawrence, Warner, and
TL2, where bass grew slowest (Table 5). As was the case in 1991, these differences
matched observed differences in prey mass in bass stomachs. In TL3, bass had an average
of 1.10 g of prey, whereas in Lawrence, Warner, and TLZ, bass had 0.86, 0.70, and 0.58
g respectively (calculated using adjusted means from an analysis of covariance: see Table 2
for a description of a similar analysis done for 1991 data). Interestingly, growth rates
decreased more in TL2 (by 70%) than in the other lakes, which incurred reductions of

around 50%.

The reduced bass grth in 1992 had strong effects on the relative size advantage of bass
over YOY bluegill later in the season. Despite the fact that YOY bluegill appeared in the
littoral zone one month later in 1992, bluegill in 3 of 4 lakes were essentially unavailable to
bass when they ﬁrst appeared (Fig. 9). Furthermore, vulnerabilities did not change over
time. Bluegill grew well enough to maintain a relatively constant ratio to has sizes and

continued to be unavailable in 3 of the lakes for the entire season. This was especially true

34

Figure 8. Mean water temperatures of Lawrence Lake. Points represent average
temperature readings for 1, 2, and 3 m depths taken at biweekly (1991) or monthly
(1992) intervals. Quadratic equations have been ﬁt to each year (1991: y:-
0.002x2+0.36x+9.72, r2=0.971, 1992: y=-0.001x2+0.17x+12.67, r2=0.943).
Data were provided by CK. Geedey.

Mean Epilimnetic Water Temperature (0C)

’35

 

 

 

 

 

 

 

 

 

30
O 1991
O 1992
25 1
20 ..
15 -
10 j l I I I
May June July August Sept. Oct. Nov.

Figure 8

36

Table 5: Comparison of early bass growth rates in 1991 and 1992. Growth rates are
estimated as slopes of regressions of mass of individual bass vs. time during the
invertebrate feeding stage. (i.e. until >50% of the population switched to piscivory).
Slopes for 1991 were taken from Table 3. All regressions are signiﬁcant at p<0.0005
level. Analysis of Covariance using years as the grouping variable and time as the
covariate is highly signiﬁcant (F354=104.11, p<0.0001). Year effects were insigniﬁcant,
but both the main effect of time and the interaction were signiﬁcant (Year: F154=1.22,
p>0.10 ,Time: F154=269.30, p<0.0001, Interaction: F154=37.85, p<0.0001). The lack
of a signiﬁcant year effect implies that has were similar sized as of June 1 in the two
years. A signiﬁcant interaction term indicates average growth rates were different between
the two years.

 

Growth Rate Growth Rate
Lake in 1991 (g/day) in 1992 (g/day) r2 (1992) N(1992)
TL3 0.061 0.034 0.433 48
Lawrence 0.047 0.028 0.657 194
TL2 0.045 0.014 0.360 104

Warner 0.029 0.015 0.499 219

37

Figure 9. YOY bluegill availability over time for 1992. Data were collected and
presented as in Fig. 5. Analysis of covariance is signiﬁcant (F7 269:8720,
p<0.0001); however only the lake effect is signiﬁcant this year (lake effect:

F3 359:5.88, p<0.001, time effect: F1369=0.69, p>0.10, interaction: F3359=0.53,
p>0.10).

1

gill Cohort Availablee

% of YOY Blue

38

 

 

 

 

 

 

 

 

 

 

1992
00
—0— Lawrence
----~O----- T'L2
---I-- Warner
.',." .................... ‘3
75 - .........
‘o" k“
50 -
25 ..
0 d ——k “T“j—h...
A089“ Sept.

Figure 9

1|

Aﬂm
Ltm

f0;

Ba

mid.

39

in T12 and Warner, where less than 10% of the bluegill size distribution was vulnerable to
bass predation. Bass were somewhat larger in Lawrence, and as a result they were capable
of consuming around 20% of the YOY bluegill. TL3 was the only lake in which the
majority of YOY bluegill were vulnerable to predation. Inthis lake, bass again grew well
in the invertebrate stage and when YOY bluegill arrived, they were capable of consuming a

large proportion of the bluegill cohort (Fig. 9).

Because bass in T13 were the only ones with a clear size advantage over their prey, this
was the only lake in which the bass exhibited a clear shift to piscivory in 1992. As in
1991, as soon as bluegill began migrating to the littoral zone in TL3, over 70% of the bass
switched to feeding on ﬁsh and this pattern continued through the season. In Lawrence
Lake, only the largest bass were piscivorous on the last 2 sampling dates in September.
The majority (Mean=64.6%) fed on insect nymphs throughout the entire growing season.
No niche shifts were observed in either TL2 or Warner Lake. Young-of-year bluegill were

too large and as a result bass in these two lakes remained in the invertebrate feeding stage.

As was observed in 1991, a shift to piscivory lead to an increased in growth rates of
YOY bass (Fig. 10). Unlike 1991, however, this shift only occurred in TL3. As in 1991,
the initiation of the second (higher) phase of growth in TL3 closely matched the observed
date of a switch to piscivory (predicted date=53.4 days after June 1, observed date=65
days: Fig 10). Bass in the other lakes grew at a constant rate set by invertebrate resources
for the entire season. Therefore, Lawrence Lake bass ﬁnished the year larger than bass in

TL2, which were slightly larger than those in Warner.

Bass in all four lakes had substantially lower ﬁnal sizes in 1992 compared to 1991. By

mid-September, mean bass mass was 2.91 g (or 35.1%) less in TL3 and 2.17 g (or

40

Figure 10. Invertebrate and piscivorous growth phases of YOY bass in 1992. As
in Fig. 7, solid lines represent the invertebrate stage and dashed lines represent the
piscivorous stage. Regression equations for the invertebrate feeding phase are
found in Table 4. Only TL3 showed a piscivorous growth phase. The equation for
this line is M=-2.189+0.0070(t), r2=0.200, n=39, p<0.005. Using this equation,
the intersection of the two growth phases in TL3 is at 53.4 days after June 1; the
observed diet switch occurred at 65 days. ‘

41

 

 

 

 

 

 

 

1992

6

0 Lawrence

0 T12

A TL3 I
5 - I Warner ,’

’I
I
I
I"
I
4 - I,
I
A”
I
’I
3 -
’I
’I
A o
I
2 - 1’
’I
O
1 -
‘ A
I
O "' I I I
June July August Sept.

Figure 10

 

42

46.9%) less in Lawrence Lake. These reductions were much smaller than observed in
T12 or Warner, where ﬁnal bass mass decreased by 4.26 and 4.53 g in the 2 lakes
(representing declines of 74.7% and 76.3% in TL2 and Warner respectively). This
underlies the importance of the niche shift for bass growth. In TL3, bass shifted in both
years: in Lawrence bass were not able to shift in either year. The pecent reductions in these
two lakes were similar and show a direct effect of year-to-year variation on bass growth
when niche shifts were not affected. In T12 and Warner, bass were piscivorous in 1991
but not in 1992. In these 2 lakes, grth rate reductions were much larger and were driven
by two mechanisms. First, strong year effects (e.g. climate) directly reduced grth rates
in the invertebrate feeding stage as they did in TL3 and Lawrence. Because of this
reduction, however, bass were too small when YOY bluegills arrived, and thus were
unable to shift to piscivory. As a result, bass experienced no increase in growth rates, and

consequently were much smaller size at the end of 1992 than in the previous year.

Strong effects of early growth on later shifts to piscivory can also explain the variation in
bass size found in Fig. 1. If growth rates of bass during the invertebrate feeding stage did
not vary among lakes or years, a negative correlation between ﬁnal bass and bluegill size
(i.e. at the end of the season) would be expected: smaller bluegill would be more available
as prey, and bass would grow better. However, this relationship was not observed.
Instead, bass and bluegill sizes for different lake/year combinations varied independently of
one another (Fig 11a). There was, however, a strong positive correlation between bass
size and the percentage of the bluegill distribution available to bass (Fig. 1 1b). Early
growth rates of bass likely played a strong role in determining bluegill vulnerability, and in
the subsequent shift to pisvivory. Among lakes, growth during the invertebrate feeding
stage varied. In lakes where grth rates were high (like TL3), bass were capable of

feeding on most of the bluegill as soon as they arrived in the littoral zone. Bass in these

43

Figure 11. A) Bass size at age 1 versus bluegill size at age 1. Bass data are from
Fig. 1. Bluegill sizes were back-calculated for each of the lake/year combinations
in which there were data for bass. B) Bass size at age 1 versus % availability of
bluegill. Bluegill availability was calculated from the size-frequency distribution
generated for each lake/year combination. Each pont represents the average of %
availabilities from each bass, using 40% of their length as the maximum vulnerable
bluegill size.

A)

Bass Standard Length (mm)

3)

Bass Standard Length (mm)

 

 

80 1
r=-0.03 '
o o
75 - . . .
Q I
70 - : '
' o
65 4
+
o
' o
60 - . ' 0 .°
. ' o o
. C C
55 "' . .
o

 

 

 

 

 

 

 

Figure 1 1

l I
25 50 75

% of Bluegill Cohort Available

 

100

45

lakes could make successful shifts to piscivory and therefore continue to grow well through
the rest of the season. Among years, bass and bluegill sizes varied independently. Bass
sizes probably depended on early climatic conditions. If conditions were favorable for bass
early in the season (through early spawning and/or enhanced growth rates), then bass
would be larger when bluegill appeared and would be capable of consuming a large
proportion of the prey distribution. Regardless of how well bluegill grew, the early growth
advantage of bass would have enabled them to shift diets. Because of this, bass growth
rates would increase in the piscivorous stage, and the end result would be a good year for

ﬁrst year bass growth.
DISCUSSION

One important consequence of ontogenetic niche shifts is that populations can potentially
be divided into ecologically distinct stages. Through these size-related changes in diet
and/or habitat use, individuals can have very different effects on community and ecosystem
level processes during their lifetime. For example, many organisms move up trophic levels
as their diets change (Stein et al. 1988). Therefore, their effect on a given trophic level
switches from a direct negative effect to an indirect positive effect; a result of changing
from feeding on a resource to feeding on organisms that feed on that resource (Miller and
Kerfoot 1987). In other cases, ontogenetic shifts in habitat create a link between separate
habitats though energy and nutrient ﬂow and recruitment dynamics (Lodge et al. 1988,
Osenberg et al. 1993). As a ﬁrst step towards understanding these stage-structured
interactions, it is critical that we study factors that affect transitions from one stage to the

next, and the effect that these niche shifts have on overall population dynamics.

Successful niche shifts from invertebrates to ﬁsh were very important to ﬁrst year growth

glow

46

in bass. Variation in growth rates is ecologically important because survival through the
ﬁrst winter is highly size-dependent in bass and other ﬁsh. Small individuals are much
more susceptible to overwinter starvation than are large individuals due to higher speciﬁc
metabolic rates and fewer stored lipids (Aggus and Elliot 1975, Eipper 1975, Oliver et al.
1979). Bass grew at a higher rate after switching to piscivory as a result of feeding on a
higher quality resource (Paloheimo and Dickie 1966) and consuming higher amounts of
food (Olson, unpublished data). Consequently, there was a clear divergence in size
between populations in which most of the bass shifted to piscivory compared to
populations in which the majority did not shift and thus grew at a lower rate for the entire
season. Similar results have been observed in other populations of largemouth bass
(Summerfelt 1975, Shelton et al. 1979, Timmons et. al 1980, Keast and Eadie 1985).

, Over time, bimodal size distributions developed from initially unirnodal distributions. -
Larger bass in one mode were all piscivorous. These bass grew so much faster than those
that remained in the invertebrate feeding stage that they were actually able to grow into a
distinct size distribution (Timmons et al. 1980, Keast and Eadie 1985). In addition to the
ability to shift niches, the timing of the shift is also important to bass growth. When bass
make an early switch to piscivory, as they did in TL3 in 1991, they experience enhanced
growth rates for a long period of time and ﬁnish their ﬁrst season much larger than bass in

other lakes or years.

Successful niche shifts are very dependent on bass having a size advantage over their
primary ﬁsh prey, bluegill. Because bass spawn before bluegill. they have a period of 1-2
months to grow until YOY bluegill begin to appear in the littoral zone. If growth in this
stage is good and bass are large, they are likely to be capable of consuming much of the
incoming bluegill cohort, and therefore will make a successful niche shift. However, if

growth is poor and bass are small relative to their prey, then most bluegill will be

47

invulnerable and instead of shifting to piscivory, bass will continue to feed on

invertebrates.

In this regard, the predator-prey interaction between bass and bluegill can be viewed as a
race in size over time. Bass get a headstart by spawning earlier, but also need a substantial
size advantage over bluegill to be successful predators (Lawrence 1958 Hambright 1991).
Growth rates in this early stage are critical in determining the magnitude of the size
advantage when bluegill ﬁrst become available, which in turn dictates whether or not bass
can make the niche shift to piscivory. Growth rates continue to be important as the season
progresses: because bluegill are also growing it is crucial for bass to continue to grow well
in order to maintain their size advantage and remain piscivorous. A size race between
predator and prey appears to be a fairly common phenomenon in systems where both
species grow (Wilbur 1988). For bass feeding on shad (Shelton et al. 1979, Adams and
DeAngelis 1987), walleye on perch (Madenjian and Carpenter 1991), and ﬂounder on spot
(Rice et al. 1993), successful predation has been shown to depend on the predator having a
size advantage over their growing prey. Because the outcome of a size race hinges on the
balance of predator and prey size, it will be sensitive to many different factors. In this
study, three factors appear to be important for bass and bluegill: growth rates of bass
during the invertebrate feeding stage, year-to-year varition in growing conditions, and size

distributions of the YOY bluegill cohort.

The invertebrate feeding stage is very important to bass despite the fact that it does not
last very long. Growth rates in this stage determine bass size when bluegill ﬁrst arrive, and
thus whether or not bass can shift to piscivory at that time. During the invertebrate feeding
stage, bass diets change rapidly with size. These changes are associated with increases in

prey size: small bass diets are dominated by small crustacean prey and as they grow, larger

48

odonate and ephemeropteran nymphs become more important. A similar progression of
invertebrate prey in bass diets has been found by Applegate and Mullen (1968) and Gilliam
(1982) except that in the former study bass fed extensively on Chironomids after they had
shifted from crustaceans. At any given size, bass diets in the present study were quite
consistent among lakes. Bass had narrow diets, and tended to specialize on a few active,
mobile prey taxa. These prey are probably easiest for bass to capture. Bass are
morphologically adapted for pursuing prey in open water rather than extracting them from
sediments or vegetation (Werner 1977, Gilliam 1982). Changes in taxonomic diet

composition develop as bass become able to catch larger and larger prey.

Although bass in the four study lakes consumed similar kinds of prey, they differed in
the amounts of prey consumed, leading to differences in growth rates. Bass in TL3
consistently had the most food in their stomachs throughout the invertebrate stage and had
the highest growth rates of the 4 lakes studied. Lawrence Lake bass were intermediate in
stomach fullness and growth, followed by TL2 and Warner. Within this stage, both
crustacean and insect prey are important for bass to grow well. Bass in TL2 grew much
better in 1991 than 1992, and in the ﬁrst year they had much more crustacean prey in their
stomachs. That was also the year they made a successful niche shift and spent the second
half of the season feeding on ﬁsh. Results from Warner Lake demonstrate the importance
of insect nymphs. Even though they consumed similar amounts of crustacean prey as in
other lakes, they fed very little on insect nymphs. Consequently, bass in this lake grew
poorly in the invertebrate feeding stage and were much smaller than bass in other lakes.
This kept bass from shifting to piscivory when bluegill ﬁrst arrived; instead bass continued

to feed on invertebrates until September in 1991 and for the entire season in 1992.

Climatic factors, in this case water temperature, can also inﬂuence bass niche shifts in a

on
ICC
Cm

7776

We

The
Popul

49

manner analogous to invertebrate resources: by changing predator size when prey arrive.
One way this can happen is through an alteration of predator and prey spawning times.
Because spawning in ﬁsh is temperature dependent (Carlander 1977), the rate at which
waters warm can potentially lengthen or shorten the duration of a predator’s headstart.
Adams and DeAngelis (1987) use an individual based model to show that bass are more
successful piscivores on 2 species of shad (Dorosoma spp.) when they have a longer
headstart, because it gives them more of a size advantage over their prey. On the other
hand, when shad spawn early and shorten the headstart, bass are too small to be capable of
piscivory. While changes in spawning times may be important in other size races, it is not
a factor in the race between bass and bluegill. In 1992, bass had a 2 month headstart on
bluegill, compared to only 1 month in 1991. Nevertheless, fewer bass shifted in 1992. A
potential reason for this is that the advantage gained by having a longer headstart was
negated by climate induced reductions in bass growth rates. Despite having an extra month
to grow while feeding on invertebrates, growth rates were much lower in 1992; when
bluegills ﬁnally arrived inshore bass were large enough to switch in only 1 of the 4 study
lakes. Spawning time is still undoubtedly important, but it is clear that climatic factors have
a more complicated role through their additional affect on growth rates as well. In
comparing growth rate reductions from 1991 to 1992, it appears that the effect of climate
on growth rates is lake speciﬁc. This suggests that temperature does not produce a simple
reduction in feeding or metabolic rates in all bass. Instead, temperature alters the
environment, perhaps by changing rates of secondary production of invertebrate prey.
Then, through reductions in early growth rates, bass have a difﬁcult time switching to

piscivory.

The third factor revealed to affect niche shifts in bass is size structure of the prey

population. Since piscivory depends on sizes of both predator and prey, growth rates of

50

YOY bluegill can play as critical a role as early predator growth in determining availabilities
(Rice et al. 1993). Unfortunately, very little is known about factors that affect YOY
bluegill growth. Climate will probably play a role, through its effect on spawning times
and/or growth rates. As well, resources in both the limnetic and littoral zones will likely be
a factor (Devries et. al 1991). Pelagic cladocerans and copepods are used extensively by
bluegill in their ﬁrst stage (Werner 1967, Mittelbach 1981), and the quality of this resource
could inﬂuence sizes of individuals as they initially arrive inshore. Bluegill size at this time
is important in determining whether or not bass can immediately shift to piscivory. In
1991, bass in Lawrence and TLZ were essentially the same size when bluegill arrived, but
because bluegill were large in Lawrence and small in TL2, a niche shift only occurred in
TL2. Once bluegill are in the littoral zone, subsequent growth on resources in this habitat
will affect their vulnerability for the rest of the season. In Warner Lake in 1991, bluegill
grew very slowly in the littoral zone, and over time bass were actually able to gain a size
advantage. In Lawrence Lake, on the other hand, bluegill grew well enough to stay
invulnerable to YOY bass for the entire season. Bluegill’s ability to remain invulnerable in
this lake was aided by their size when they initially migrated inshore. By being’large as
they ﬁrst arrived, bluegill kept bass from shifting, and thus bass never experienced the
jump in growth that accompanies a transition to piscivory. Without this jump in bass
growth rates, bluegill in Lawrence were able to maintain sufﬁcient growth rates to prevent a
shift to piscivory in bass for the rest of the season. This underscores another reason why
niche shifts are so important to bass. Once bass switch, the increased growth rates that
follow enable bass to maintain their size advantage more easily; bluegill would have to
grow extremely well to escape to a size refuge. However, if bass do not shift, bluegill can

grow at a lower rate and still remain invulnerable to their slow growing predator.

For YOY bluegill, mortality rates over the course of a season will be lower when bass

Oﬁc
Min.

Struc

51

can not make an early transition to piscivory. Therefore, factors which reduce the size
advantage of bass when bluegill ﬁrst arrive inshore will increase survivorship within the
YOY bluegill cohort. For example, if growth rates in the limnetic stage increase, YOY
bluegill will be larger and consequently less vulnerable when they arrive inshore (Devries et
al. 1991). The size advantage of bass will also be reduced when the duration of the
headstart is shortened. From an evolutionary perspective, therefore, bluegill might beneﬁt
by spawning earlier in the season and closer to the spawning time of their predator.
However, spawning early also carries a cost of potentially spawning in unfavorable
conditions that impose heavy mortality on eggs and larvae, which would negate any
advantage of spawning at that time (Summerfelt 1975, Eipper1975). Chapter 3 develops a
simple model that addresses this tradeoff between reducing losses to predators and
spawning in risky conditions. The model suggests that, for a wide range of environmental
conditions, the Optimal reproductive strategy for prey is to spawn later in the season after

their predator, but when conditions are more benign.

Small changes in early growth rates of predators can have profound impacts on the ability
of predators to shift niches. Because of this, predators could be very susceptible to the
effects of competition (both intra- and inter-speciﬁc) during their early life stage.
Interspeciﬁc competition is possible because small predators often overlap with other
species in diet and habitat use (Werner 1986). High diet overlaps are a consequence of
similar constraints which force both species to converge on small prey when they are small
(Keast 1977b). Shared habitat use arises as a result of small individuals of both species
often being restricted to a common refuge in order to avoid predators (Werner et al. 1983a,
Mittelbach 1984). Competitive interactions within a stage can be very important in
structured populations because they can also inﬂuence dynamics of other life history stages

through changes in recruitment rates (Tschumy 1982, Mittelbach and Chesson 1987). This

52

is especially true for small predators that ultimately prey on the other species once they get
large. In these systems, the interaction changes from competitive to predatory during the
predators ontogeny, and interspeciﬁc effects on predators switch from negative to positive
(Gilliam 1982, Persson 1988). If competition is strong enough, prey can impose a
recruitment bottleneck on their predator and limit or even reduce the density of their

predator (Persson 1988).

The interaction between bluegill and bass is an example of a mixed competition/predation
interaction (Chapter 2). Like bass, bluegill undergo niche shifts in their ontogeny
(Mittelbach 1981, 1984). After migrating inshore from the limnetic zone as YOY, bluegills
spend 34 years (until they reach ~75mm SL) in the littoral zone which is a refuge from
predators before moving back into the open water (Mittelbach 1981, Werner and Hall
1988). During this littoral stage, bluegill feed on many of the same prey items as YOY
bass (Olson unpublished data). The density and growth rates of bluegill vary over the
seven study lakes, (Osenberg et al. 1988, Mittelbach and Osenberg 1994) and these
patterns suggest that bluegill generate a gradient in invertebrate resources which might also
affect bass growth. Indeed, bass size at age 1 declines as small bluegill density increases
among these seven lakes (r=-0.99, p<0.001). Furthermore, growth rates of small bluegill
are positively correlated to bass size (r=0.88, p<0.01), as would be expected if both are
using shared resources (Chapter 2). Strong competitive effects were also demonstrated
experimentally; YOY bass growth rates declined in response to a gradient in small bluegill
density (Chapter 2). This occurred despite substantial niche partitioning between the two
species. Bluegills fed primarily on Chironomids, but also ate many other prey taxa
including several types important to bass. Although these prey were not major components
of bluegill diets, they were consumed enough to reduce the availability of large prey for

bass. As a result, with increasing bluegill density, bass fed on smaller prey and

53

consequently grew much more slowly. As an added consequence of competition, the
reduction in growth rates can prevent bass from shifting to piscivory, in which case
competition for invertebrates continues. Prolonging the competitive interaction with small
bluegill, combined with the potential competition between bass and YOY bluegill, Creates a
strong negative effect of bluegill on bass, and this leads to very different dynamics from

typical predator-prey interactions (Chapter 2).

The invertebrate feeding stage in bass could also be important because of its inﬂuence on a
population’s sensitivity to year to year variation. In lakes where invertebrates are
abundant, bass grow well and are capable of feeding on most of the bluegill size
distribution. Unless conditions are extremely poor, these bass will grow well enough even
in unfavorable years to shift to piscivory (i.e. they would still be on the plateau of Fig. 6).
Furthermore, because YOY bluegill typically have a normal size distribution (Olson,
unpublished data), changes in bass size have only minor effects on the proportion of
bluegill available to bass (Fig. 12). Thus, small changes in early growth rates will have
little inﬂuence on the shift to piscivory, and overall YOY growth rates will be relatively
insensitive to yearly variation. When bass have intermediate growth rates in the
invertebrate feeding stage, the situation is very different. As bluegill arrive in the littoral
zone, bass typically are able to feed from the lower half of the distribution. Hence, the
proportion of the distribution that is vulnerable to predators will change dramatically with
small changes in bass size (Fig. 12) These changes will in turn affect whether or not bass
can make a successful niche shift because bass are likely to be in the transition zone of Fig.
6 (i.e. close to the inﬂection point). As a result, small changes in early growth will have a
strong inﬂuence on the niche shift, and make bass size very dependent on climate. If bass
growth while feeding on invertebrates is low, bass will only be able to feed from the lower

tail of the bluegill size distribution, if at all. Most of the bluegill prey will be invulnerable,

54

Figure 12. Schematic representation of changes in bluegill vulnerability with bass
size. Cumulative vulnerability is based on an hypothetical normally distributed
bluegill cohort. Bars represent the proportional change in bluegill vulnerability for
a 5 mm reduction of bass size. Three scenarios are presented, representing
reductions of large (descending diagonal bar), intermediate (hatched bar), and small
(ascending diagonal bar) bass. A 5 mm reduction in bass size corresponds to a

6.5 %, 38.2%, and 6.5 % reduction in % bluegill available respectively.

55

 

\\\\\ \\\\\\\\\\\ \\“‘"'

75d

P
‘

 

(I,(I’li’l’TrIIIIITI'IIIII
\\\\\\\\\\\\\\\\\\\\\\\\\\
IIIIIIIIIIIIIIIIIIIIIIIII
\\\\\\\\\\\\\\\\\\\\\\\\\\
IIIIIIIIIIIIIIIIIIIIIIIII
'5‘\\\\\\\\\\\\\\\\\\\\\\\
III/IIIIIIIIIIIIIIIIIIIII
\\\\\\\\\\\\\\\\\\\\\ \\
IIIIIIIIIIIIIIIIIIIIIIII
\\\\\\\\\\\\\\\\\\\\\ \
III/IIIIIIIIIIIIIIIIIIII
\\\\\\\\\\\\\\\\\\\\\\\\
III/IIIIIIIIIIIIIIIIIIII
\\\\\\\\\\\\\\\\\\\\\\\\
IIIIIIIIIIIIIIIIIIIIIII
\\\\\\\\\\\\\\\\\\\\\\\
IIIIIIIIIIIIIIIIIIIIIII '
\\\\\\\\\\\\\\\\\\\\\\\
IIIIIIIIIIIIIIIIIIIIIII
\\\\\\\\\\\\\\\\\\\\\\\
IIIIIIIIIIIIIIIIIIIIII
\\\\\\\\\\\\\\\\\\\\\
III/IIIIIIIIIIIIIIIIII
\\\\ \\\\\\\\\\\\\\\\\
III/IIIIIIIIIIIIIIIIII
\\\\\\\\\\\\\\\\\\\\\\
IIIIIIIIIIIIIIIIIIIII
\\\\\\\\\\\\\\\\\\\\\

III/IIIIIIIIIIIIIIIII
\sxxssxxsxsssxsssxxxs
I’ll/IIIIIIIIIIIIIII
\\\\\\\\\\\\\\\\\\\\
IIIIIIIIIIIIIIIIIIII
\\\\\\\\\\\\\\\\\\\\
IllklIIIIIIIIAIIIIIII

Cumulative Vulnerability of YOY Bluegill (%)
8
l

4

 

[ll/ﬂ”

 

0 I I
20 30 40

Bass Size (mm SL)

Figure 12

50

 

56

and as was the case for bass feeding on the upper end of the distribution, the percentage of

bluegill vulnerable is relatively insensitive to changes in bass size (Fig. 12).

The differential sensitivity of bass populations to year-to-year variation will lead to a
hump shaped function describing variation in bass size vs. growth rates on invertebrate
resources. Through variation in the ability to shift to piscivory, lakes with intermediate
bass growth rates will be most variable from year to year, while lakes with better or worse
growth will be more stable. There is some evidence to support this hypothesis. Of the
seven lakes studied, proportions of bluegill available to bass have the largest range and
highest variance in Culver, Deep, and Lawrence, the three lakes where bass sizes are
intermediate (Fig. 1). Furthermore, these three lakes also have the most year to year
variance in bass size at age 1. As a result, there is a marginally signiﬁcant positive
correlation among the seven study lakes between the coefﬁcient of variation in bass size
and variance in percent prey available (r=0.72, p=0.10). In lakes where early bass growth
is intermediate, variation in size at the end of the growing season could lead to wide
ﬂuctuations in overwinter survival rates. Consequently, these lakes may also show the

greatest variation in bass recruitment.

The shift to piscivory by bass is sensitive to the relative sizes of predator and prey; small
changes in predator or prey size can drastically change the timing of a niche shift, which in
turn has important consequences on bass growth rates. Therefore, a variety of mechanisms
can generate variation in YOY bass grth rates by inﬂuencing the ability of bass to shift
from invertebrates to ﬁsh. This variation in the shift to piscivory puts largemouth bass at
one end of a continuum of variability in the timing of ontogenetic niche shifts. Among
populations or years, the size and age at which bass shift to piscivory changes dramatically

depending on environmental conditions. At the other end of the continuum are species with

57

genetically determined niche shifts. For example, many groups of insects that undergo
metamorphosis do so at a preset size and/or age (Werner 1988). Note, however, that there
are examples of species that undergo metamorphosis which exhibit a great deal of
variability in size at metamorphosis (see Wilbur and Collins 1973). Molluscivorous ﬁshes,
like wrasses or pumpkinseed sunﬁsh, represent an intermediate point on the continuum.
Niche shifts in these ﬁshes are determined by a combination of morphological and
environmental factors (Wainwright 1988, Wainwright et al. 1991 Osenberg et al. 1992).
With increasing size, jaw muscles develop which enable these ﬁsh to crush larger, thicker
shells. Although crushing ability increases smoothly with size, niche shifts are quite abrupt
because the size distributions of molluscans typically provide few prey for small ﬁsh but
abundant prey for large ﬁsh. As well, these distributions are fairly constant, making shift

sizes quite general among populations (Osenberg et al. 1993).

In contrast, the timing of ontogenetic niche shifts in bass are highly variable because they
feed on a more dynamic resource whose size-structure changes considerably during the
season. Young-of-year bluegill migration to the littoral zone is highly pulsed, and
following their arrival they continue to grow and change size quickly through time.
Consequently, for a given sized bass there is only a narrow window of time in which
bluegill are available. This window closes quickly as prey grow, and once it is closed there
is no other chance for bass to become piscivorous at that size. In order for that bass to shift
to piscivory, it must get larger over time and re-establish a size advantage that will reopen
the window. The ephemeral nature of the prey resource makes niche shifts in bass a
function of both size and time. Size and time dependence also creates a feedback loop in
the niche shift. If only size were important, then reductions in growth rate during the ﬁrst
stage would only prolong that stage until individuals grew to the requisite size. However,

since prey are growing, that requisite size increases over time. Therefore, reductions in

58

early growth also increase the size at which a bass can switch to piscivory, which feeds
back to prolong the stage even more. Subtle effects on early growth, like climatic factors

or competition, will be especially inﬂuential through this feedback.

There are many ecological implications of a highly variable niche shift. For organisms on
this end of the continuum, the role of their ﬁrst stage in community level processes
becomes much more variable. For example, predator impacts on resources will be
sensitive to environmental conditions. When resources are abundant, predators grow well
and shift off the initial prey resource quickly. The result is a minimal impact on the
resource. However, if resources are rare and growth is poor, the impact is much greater.
Predators will deplete this resource much more because growth rates are reduced and the
stage is prolonged. A similar result occurs in poor climatic conditions: reductions in
predator growth rates will intensify their effect on resources. Variable impacts on
resources lead to an unpredictable competitive effect of predators on other organisms that
also use this resource. These competitive effects will also have a feedback loop. As
resources become more limiting, competitive effects become much stronger.
Consequently, competitive effects can range from negligible to very strong depending on

environmental and climatic conditions.

Further ecological consequences arise in the stage that follows the niche shift.
Recruitment rates into the second stage are strongly dependent on early growth rates and
their subsequent effect on the timing of niche shifts: late shifts severely decrease
recruitment. This will be true if mortality rates are uniform across size simply because
individuals will start the second stage later and more will have died before shifting. For
many organisms, however, mortality rates decrease with size (Ricker 1979, Peters 1983).

In such cases, size and time dependent shifts are even more devastating because low

59

growth rates force individuals to shift at a larger size, and even fewer will survive to that
size. As a result, when there is variation in the niche shift population densities in the
second stage will ﬂuctuate widely. This can potentially have important consequences for
the resources used by that stage. For bass and many other species of piscivorous ﬁsh, they
are in the top trophic level when they reach their second stage (Carpenter et al. 1987, Mills
and Fomey 1988). Variation in their density could create a cascading trophic interaction
(mu Carpenter et al. 1985) that will change abundances in all subsequent trophic levels.
Variation in survivorship to the second stage can also eventually affect densities of
reproductive adults. Therefore, production rates of new offspring could be extremely
variable among populations and years. In years of high production, intraspeciﬁc
competition among juveniles in the ﬁrst stage could reduce growth and limit later
recruitment of that cohort, generating large oscillations in overall population density and

making species with variable niche shifts even more unpredictable.

In addition to ecological implications, evolutionary responses to niche shifts will be
affected by a population’s location on the continuum. Werner (1977) hypothesized that
many organisms face a tradeoff in performance (e.g. foraging efﬁciency, vulnerability to
predators) at different life history stages. The optimal strategy to deal with these conﬂicting
demands (without a major reorganization through metamorphosis: see below) is a
compromise in adaptations for different stages so as to maximize overall ﬁtness (Werner
1988). While it is easy to envision selection reaching a compromise when niche shifts are
predictable, it is much more difﬁcult when shifts vary. If early conditions are favorable,
virtually all individuals can make the shift regardless of their level of adaptation to that
stage. In that case, the optimal strategy is to specialize more in the second stage. On the
other hand, unfavorable conditions early make the ﬁrst stage very important: only the best

suited individuals for that stage will grow well enough to be able to shift. Thus, changing

the
9X11
fact
Pl'SCl

Chan;

60

conditions create a variable environment for selection, sometimes favoring the second stage
and other times favoring the ﬁrst. Analogous variation on a spatial scale (i.e. when
resources in the ﬁrst stage differ) can potentially lead to dramatic divergence in localized

adaptations among populations, especially if conditions remain relatively stable over time.

When the timing of niche shifts are variable, there is also potential for rapid changes in
population genetic structure. In many size-structured populations, a few individuals
contribute disproportionately to the potential gene pool of the next generation (Kirkpatrick
1988). This is due to the fact that reproductive output usually scales with size, and only a
few individuals reach large sizes (Peters 1983, Kirkpatrick 1988). With variable niche
shifts, these effects are compounded. Adult populations sizes vary over time, as do
survival and recruitment rates of their offspring. Under a particular set of circumstances, a
few adults could produce a very strong year class in a year when offspring make an early
niche shift and recruit successfully. Then, the majority of a population would be the
product of a small number of adults. P0pulations with variable niche shifts would
therefore pass through periodic genetic bottlenecks, which can have major impacts on the
genetic structure of a population by restricting variance and forcing the population through

repeated founding events (Mayr 1963).

One challenge to the study of structured populations is to understand processes that affect
transitions from one stage to the next (Osenberg et al. 1993). Furthermore, the inﬂuence of
these processes on overall population dynamics are poorly understood, but may be
extremely important. Largemouth bass are an example of an organism in which many
factors can profoundly inﬂuence the niche shift from an invertebrate feeding stage to a
piscivorous stage. Subtle changes in early growth rates of bass or bluegill led to dramatic

changes in the timing of the niche shift, which resulted in substantial effects on grth

 

rat:
sen

fun

im;

61

rates of bass through the rest of the growing season. Because bass show this extreme
sensitivity to factors which change early growth rates, they may prove very useful for
future studies of ontogenetic niche shifts, and for elucidating some of the mechanisms

important to the ecology and evolution of stage-structured populations in general.

CHAPTER 2
COMPETITION BETWEEN PREDATOR AND PREY:
RESOURCE-BASED MECHANISMS AND IMPLICATIONS
FOR ST AGE-S'I'RUCI'URED DYNAMICS

62

63

INTRODUCTION

A common pattern in sizeestructured populations is for individiuals to show substantial
changes in diet or habitat use as they grow (Werner and Gilliam 1984). Because of these
ontogenetic niche shifts, populations are often split into distinct stages based on size
(Nisbet et. al.1989, Osenberg et al.1993). The separation of a population into stages can
have important consequences for many ecological processes. Interspeciﬁc interactions,
such as competition or predation, may be strong in some stages of a population but very
weak for other stages. Nevertheless, dynamics of these other stages will be affected
indirectly through the processes of production and recruitment (Mittelbach and Chesson
1987). For example, individuals of different species often use similar resources when they
are small, but use very different resources as they get larger (Keast 1977, Mittelbach 1984,
Polis 1988). Therefore, even though most members of the two populations do not interact
with one another directly, the two species may become linked through a competitive stage
early in their life history (Osenberge et al. 1988, Mittelbach et al. 1988).

In many cases, ontogenetic shifts in diet are the result of increases in the ability to capture
larger prey (Wilson 1975, Werner 1986). Because body size is typically correlated with
trophic level (especially in aquatic systems: Stein et al.1988), these niche shifts often
correspond to changes in position along the food chain. Consequently, the nature of
interspeciﬁc interactions will change with ontogeny. When individuals of different species
overlap in size and occupy the same trophic level, competition can be an important process
(Mittelbach 1988). However, as an individual of one species grows, it may move up a
trophic level and prey on species that had previously been competitors. Thus, the overall
interaction is a mixture of competition and predation, and prey may have both a negative

and a positive effect on their predator (Persson 1988). If competition is strong enough,

64

prey can even impose a bottleneck on their predator, and restrict recruitment to the
predatory stage (Werner 1977, Werner and Gilliam 1984, Persson 1988). Bottlenecks may
be quite common because prey dentities are usually much higher than predator densities,
and survivorship of predators through early life-history stages is very sensitive to changes
in growth rates (Werner and Gilliam 1984). As a result, population dynamics of these two
species may be very different from the dynamics of simple non-structured predator-prey

systems.

largemouth bass (Micropterus salmoides) and bluegill (Lemmis macrochirus) are two
species that appear to interact through competitive and predatory stages (Fig. 13). Adult

bass are the top predators in small lakes and ponds across much of eastern North America.
However, young-of-year (Y OY) bass initially pass through a stage in which they feed
predominately on invertebrates in the littoral zone (Applegate and Mullen 1966, Heidinger
1975, Gilliam 1982, Chapter 1). Late in their ﬁrst year, bass switch to piscivory, and from
that point on rely heavily on small bluegill for food (Hackney 1975, Heidinger 1975, Keast
1985). Bluegill are the dominant ﬁsh species in many lakes in terms of biomass and are
much more abundant than bass (Werner et al.1977, Olson unpublished data). Bluegill also
undergo an ontogenetic niche shift driven by a tradeoff between foraging rate and predation
risk (Werner et al.1983, Turner and Mittelbach 1990). Small bluegill (up to 75 mm
standard length) feed on littoral invertebrates while they use the vegetation in this habitat as
a refuge from predators (Mittelbach 1981a, Werner et al.1983). Above 75 mm, bluegill are
essentially invulnerable to predators and are able to feed in the open water on more
energetically proﬁtable zooplankton (Mittelbach 1984, Werner and Hall 1988, Mittelbach
and Osenberg 1994). When bass and bluegill are in their ﬁrst dietary stages, the two
species overlap in both habitat and resource use and may go through a competitive stage.

However, the interaction changes to predator-prey when bass become piscivorous.

65

Figure 13: Schematic representation of the interaction between largemouth bass and
bluegill. Arrows connecting stages represent the processes of production and
recruitment. Arrows connecting ﬁsh to their resources represent consumption.

66

 

_ cot—SwaooN _

/

 

 

Teﬁﬁctgﬁ .82...— _

 

 

532m :25

53:5 emeﬁ

 

N /

 

 

 

 

3.5 :25

»

 

 

«
mmmm owes

 

 

Figure 13

ch
bl:

Cori
00m
”Pt
Pfed
Were

neig}

67

Therefore, bass are potentially linked to small bluegill in both stages of their life history
(Fig. 13).

In this study, I show that the interactions between bass and bluegill are strong. Results
of a pond experiment in which densities of small bass and bluegill were independently
manipulated show that competition occurs between these two species. Competitive effects
are mediated by changes in the composition of invertebrate prey. This conclusion is
supported by data collected in small lakes, which suggests that competition with bluegill
affects growth rates of YOY bass in natural populations. Small bluegill also have a positive
effect on large bass in these lakes that counters the negative effects of competition.
Therefore, the nature of the interaction between bass and bluegill is stage-dependent: small
bluegill compete with small bass but are prey for large bass. Because of this ontogenetic
change in the interaction, patterns of growth and density in natural populations of bass and

bluegill differ from non-structured populations.

METHODS

COMPETITION EXPERIMENT

I used a target-neighbor design (gem Goldberg and Werner 1983) to evaluate the
competitive interaction between small (<75 mm) bluegill and YOY bass. In this design, a
constant density of “target” individuals (of both bass and bluegill) were assigned to each
experimental unit, and competitive environments (treatments) were created by adding
predetermined densities of “neighbor” individuals to different units. Competitive effects
were assessed by comparing the performance (growth) of targets across the range of

neighbor densities. To examine the relative strengths of intra- and interspeciﬁc competion,

68

I assigned equal densities of either bass or bluegill neighbors to separate treatments (Table
6). Intra- and interspeciﬁc effects were then assessed for each species by comparing
growth rates of targets in the absence of any neighbors versus in the presence of equal
densities of conspeciﬁc or heterospeciﬁc neighbors, respectively. In addition, because
bluegill densities are typically 5-6 times higher than bass densitites in study lakes (Werner
1977, Olson unpublished data), I also examined the effect of this higher bluegill density on
YOY bass growth. Thus, the experimental design had a total of 4 treatments (Table 6):
target bass and bluegills with no additional neighbors (henceforth referred to as the target
treatment), targets plus bass neighbors (bass treatment), targets plus bluegill neighbors at
the same density as the previous treatment (low bluegill treatment), and targets plus bluegill

neighbors at ﬁve times the density of the bass treatment (high bluegill treatment).

The experiment was conducted in a circular pond (29 m in diameter and 2 m deep) located
at the Kellogg Biological Station (KBS), Michigan State University. The pond was ringed
by a continuous border of cattails (1‘1th spp.), beyond which a thick mat of §h_ar_§
covered the pond bottom and left no bare patches. 1 divided the pond into 8 pie-shaped

sections (88 in2 each) using partitions of mesh nylon netting (3.2 mm mesh), suspended 30

cm above the water from cables anchored to shore and to a center post. Partition bottoms
were attached to chain and buried into the sediment. Once deployed, the partitions were

quickly colonized by periphyton and water ﬂow between sections was minimal.

Juvenile bluegill were collected from nearby Crooked Lake, while YOY bass were
collected from a brood pond on site. Bass ranged in size from 16-23 mm standard length
(SL: as measured from snout to caudal terminus), averaged 192910.16 mm (X11 SE),

and represented a size range typical of YOY bass cohorts within local lakes. Potential

69

Table 6: Initial and ﬁnal treatment densities. Initial treatments represent the numbers of
neighbor ﬁsh stocked. In addition, all sections received 50 target bluegill and 70 target
bass. Final densities are total numbers recovered at the end of the experiment. Sections
were numbered consecutively, starting with the southernmost section and moving
clockwise.

Treatment Sections Initial Treatment Final Densities '
_ Bass Target Bluegill Neighbor Bluegill

Target 3 No Additions 9 29 -
7 No Additions 11 32 -
Bass 1 + 150 Bass 33 25 -
5 + 150 Bass 23 34 - -
Low Bluegill 2 + 150 Bluegill 15 27 117
4 + 150 Bluegill 10 33 121
High Bluegill 6 + 750 Bluegill 8 29 ' 659
8 + 750 Bluegill 5 21 652

70

bluegill competitors come from a wider size range (25-55 mm SL: Mittelbach 1981a,

1984), and encompass those individuals feeding predominantly on vegetation dwelling
invertebrates. Neighbor bluegill encompassed much of this size range (Rangez36-55 mm
SL: X=47.13:0.36 mm SL), while target bluegill were chosen from a narrower,
numerically dominant subset of this range (31-41 mm SL; X=36.45:t:0.24 mm SL) in order
to better characterize their growth response. Target bluegill were distinguished from
neighbors by clipping their right pelvic ﬁn. Initial stocking densities of bluegill and bass
are summarized in Table 6. For both bluegill and bass I used a fairly large number of target
individuals (50 and 70 per section, respectively) because I anticipated losses due to natural
mortality (the number of target bass used was higher than that of bluegill because mortality
was expected to be greater on these smaller ﬁsh). The density of bluegill neighbors used in
the low density treatment fell well within the range of densities observed in nearby lakes
(Mittelbach 1988, Osenberg et al.1988), while the density of bluegills used in the high
density treatment was at the upper end of natural densities. Bass were added on July 10,

1992, neighbor bluegill were added on July 11, and target bluegill on July 12.

Initial mortality due to handling stress was evaluated by snorkeling through each section
and walking around the pond edge each day for the ﬁrst week. Initial mortality was
estimated at 18% for target bluegills and 8.3% for neighbors; very few dead bass were
observed (estimated mortality rate: 0.4%) but due to their small size, I probably recovered
only a small fraction of the bass that had died. Target and neighbor bluegills were replaced
with similar sized ﬁsh collected from Crooked Lake on July 17 and 20; no bass were
replaced.

Mean sizes of bass and target bluegill were estimated two times during the course of the

experiment. On July 29, 17 days into the experiment, ﬁsh were collected from 2 seines per

71

section, identiﬁed, measured and released. On August 26, 45 days after the experiment
began, each section was seined 3 times, and all ﬁsh caught were measured, weighed, and
preserved in 10% neutral formalin for later diet analysis. Remaining ﬁsh were recovered
over the next 2 days as the pond was drained, and ﬁnal lengths and masses of all targets
were recorded. Initial and intermediate masses were estimated by length-mass regressions

based on ﬁnal sizes. The regression for bass mass was M=O.000020(SL3-018), r2=0.987,

and for target bluegill was M=0.0000098(SL3311), r2=0.962.

Of the ﬁsh collected on August 26, all bass and target bluegill and 20 randomly selected
neighbor bluegill were analyzed for stomach contents. Prey were identiﬁed to the lowest
taxonomic level possible (typically to family or genus), enumerated, and measured (up to
20 haphazardly chosen individuals per prey category). For damaged prey, body lengths
were estimated from other body dimensions. All lengths were converted to dry masses
with unpublished length-mass regressions (Mittelbach unpublished data). I used prey mass
to quantify diet composition by mass, and to calculate diet overlaps between groups of ﬁsh

among sections.

To explore the effects of ﬁsh on their resources and how resource availability affected
ﬁsh growth, I sampled invertebrates in the open water and vegetation habitats throughout
the experiment. Zooplankton were sampled on 4 different dates: July 9 (before ﬁsh were
added), July 19, August 4, and Aug 24 (just before the experiment was terminated). On
each date, three samples were collected from each section beginning 1/2 hour after sunset.
Zooplankton were collected at a depth of 1.0 m using a 19 litre Schindler trap with an 80
um mesh bucket, and were immediately preserved in cold 4% sucrose formalin. For each
sample, zooplankton were identiﬁed to genus or species, counted mm , and measured

(up to 50 haphazardly chosen individuals). In all analyses, I used average densities and

72

mean sizes of the 3 samples per section to estimate treatment effects.

Vegetation dwelling invertebrates were sampled from each section three different times:
July 9 (2 samples were taken per section before stocking ﬁsh), July 27 (3 samples/section)
and August 24-25 (3 samples/section). Samples were collected from the Qhﬂ vegetation
by a diver using a modiﬁed Gerking sampler (Mittelbach 1981b). Invertebrates were
separated from vegetation by washing samples onto a 0.5 mm mesh sieve. All
invertebrates retained on the sieve were then removed manually and preserved in 10%
neutral formalin. Invertebrate taxa were identiﬁed (typically to family or genus level)
counted, and measured (up to 50 haphazardly chosen individuals per taxa). All samples
from a given section were pooled to estimate treatment effects on invertebrate densities and

size structure.

In order to gain a more mechanistic understanding of how changes in zooplanktongand
vegetation- dwelling invertebrates affected resource quality, I used foraging models
developed speciﬁcally for bluegill to calculate predicted energetic return rates separately for
open water and vegetation habitats (Mittelbach 1981a). Resources were summarized by
dividing the size-density distribution of all cladocerans for the open water and soft-bodied
invertebrates in the vegetation into discrete size classes (10 size classes for zooplanktbn and
6 size classes for vegetation prey). The model then used size-speciﬁc encounter rates,
handling times, and energetic contents to calculate foraging returns of the optimal diet for a
given sized bluegill. This model has been used numerous times in the past and has been
successful in predicting growth rates in natural and experimental populations of bluegill
(Werner et al. 1983b, Osenberg et. al 1988, Mittelbach and Osenberg 1993, Mittelbach and
Osenberg 1994). Unfortunately, I do not have a foraging model for largemouth bass.

However, it is likely that bass will perceive habitats in qualitatively the same manner as

73

bluegill,although the exact values may differ.
LAKE SURVEY

To examine the effects of bluegill on bass in natural lakes, 1 estimated densities and
growth rates of small and large bass in a set of 7 lakes in southwest Michigan (all within 30
km of IGBS). These lakes are typical of hardwater lakes in the region and are similar in size
(5-26 ha) and depth (10.16 m except for Three Lakes III which is 4 m: see Osenberg et
al.(1988) for a description of the lakes). The ﬁsh communities in these lakes are dominated ,
by ﬁshes of the family Centrarchidae, particularly bluegill and bass which are the 2 most
abundant species in terms of biomass. Previous work has focussed on patterns of growth
and density in the bluegill populations of these lakes (Osenberg et 31.1988, Mittelbach and
Osenberg 1994).

Bass densities were quantiﬁed as catch-per-unit-effort (CPUE) in 5 of the study lakes for
small bass, and in all 7 lakes for large bass. CPUE’s (numbers caught per seine) were
estimated from an average of 31.6 :I: 8.6 beach seines (23 m x 1.8 m; 3.2 mm mesh) taken

from June to September 1990-1992 (there were no year effects: F3,92=0.69, p>0.10 for
small bass and F255=0.99, p>0.10 for large bass). Growth rates of small and large bass

were back-calculated from scales of 1 to 5 year old bass collected by seining and/or angling
all 7 lakes from 1990 to 1992. Bass were weighed (to the nearest g) and measured (to the
nearest mm SL), and 5 scales were taken just posterior to the depressed left pectoral ﬁn.
Impressions of these scales were made with acetate strips, which were projected with a
microﬁche reader. Distances from the focus to each annulus and scale edge were recorded
from one non-regenerated scale per ﬁsh. Fish lengths at each age were back-calkculated

from these distances using the Fraser-Lee method (Tesch 1968) with an intercept of 16.5

74

mm SL. Lengths at age were converted to live mass using length-mass regressions
developed for each lake (Olson, unpublished data). Growth rates of small bass were
estimated as mean mass at the end of their ﬁrst year for bass born between 1987 and 1990
(averages of each year, based on X=19.1 ﬁsh, were averaged to get a single number/lake).
Growth rates of large bass were estimated by ﬁrst regressing In change in mass through a
year vs. ln mass at the start of that year for each lake (average sample size: 105 bass/lake;
year to year variation in growth rates were ignored). Mass speciﬁc growth rates were then

estimated from these regressions using the mean size of a 3 year old bass (85.73 g).

RESULTS

COMPETITION EXPERIMENT

There was signiﬁcant mortality on both bluegill and bass during the experiment. In
particular, only 14% of the stocked bass survived to the end of the experiment (Table 6).
This low recovery rate was probably a result of initial handling stress and/or predation by
large invertebrates (e.g. Aug) while bass were small. Mortality of bass was density
independent, and I recovered similar proportions of bass in all treatments (F3,4=2.06,
p>0.10). Recovery rates of bluegill were higher than bass (Table 6: 58% for targets and
83% for neighbors). The replacement of dead bluegill at the start contributed to this result,
as did the fact that bluegill were larger initially and handling mortality was therefore lower.
As was the case for bass, recovery rates of target bluegill were independent of density
(F3 ,4=0.54, p>0.10). Although mortality of bass and bluegill was higher than I expected,
I could still assess competitive effects through comparisons of growth rates among targets

because the mortality of each species was independent of density.

I]:

sh

111:

mm

75

Increasing neighbor densities had strong negative effects on the growth rates of both bass
and bluegill (Fig. 14). For bass, initial mass did not differ among treatments, but by the
end of the experiment bass masses were signiﬁcantly different in all 4 treatments (Fig.
14a). Thus, bass growth rates were reduced through both intra- and interspeciﬁc
competition. Signiﬁcant treatment effects were also observed on growth rates of target
bluegill (Fig. 14b). However, bluegill were affected primarily by intraspeciﬁc competition.
Although bluegill were smaller in the bass treatment compared to the target treatment when
the experiment ended, this difference was not signiﬁcant (Bonferroni T test: p>0.05).
Signiﬁcant competitive effects were only seen between target, low bluegill, and high

bluegill treatments.

Iow recovery rates of bass made direct comparison of the relative strengths of intra- and
interspeciﬁc competition problematic (i.e. neighbor densities in bass and low bluegill
treatments were no longer equal). Therefore, as an alternative to ANOVA I used a
regression approach to develop predicted growth rates of each species at different densities
of competitors (i.e. Goldberg and Werner 1983). Competitive effects of bluegill were
assessed by regressing ﬁnal bass and bluegill masses against bluegill density (targets and
neighbors combined) for target, low bluegill and high bluegill treatments (bass densities
were ignored because they were similar in all 6 experimental units). These relationships
show the effect of a change in bluegill density on bass and bluegill growth (Fig. 15). Final

masses and ﬁsh densities were log transformed to make the regressions linear.

To estimate the relative competitive effect of bass compared to bluegill, I plotted ﬁnal
bass and bluegill masses in the two pond sections with added bass (low bass treaMent) and
compared these points relative to the regression lines of bass and bluegill growth vs.

bluegill density (Fig. 15). If the points lie above the line, then bass have less of a

76

Figure 14: Changes in mean ﬁsh mass (:1 SE) over time for bass (A) and target
bluegill (B). Filled circles represent the target only treatment; open squares
represent the bass treatment; ﬁlled triangles represent the low bluegill treatment; and
ﬁlled squares represent the high bluegill treatment. Initial and intermediate masses
were converted from standard length by regression. Final mass was measured
directly. Initial mass did not differ among treatments for either species(Bass:
F3,4=3.39, p>0.10, Bluegill: F3,4=0.14, p>0.10). Final mass differed
signiﬁcantly for both species (Bass: F3,4=278.60, p>0.0001, Bluegill:
F3,4=241.00, p>0.0001). For bass, Bonferroni T tests show that all ﬁnal masses
in all four treatments are different from one another (p<0.05). Target bluegill ﬁnal
size was signiﬁcantly different among target, low bluegill and high bluegill
treatments (Bonferroni T test: p<0.05), but the low bass treatment was not
signiﬁcantly different from targets alone (Bonferroni T test: p>0.05).

77

em semis
h

em :3
_

 

 

 

 

 

385 Am

en :33
-

 

 

 

I
N

l
M

 

 

83 £38093 3

(3) 888w 1%

Figure 14

78

Figure 15: Linear regressions of ﬁnal masses of all bass (A) and target bluegill (B)
vs. ﬁnal bluegill density (targets and neighbors). Data were log-transformed prior
to analysis. Symbols are as in Fig. 14. Regression lines are based on 6 points from
the target, low bluegill and high bluegill treatments: A) log10Y=1.34-
0.437(log10X), r2=0.990, B) logloY=1.31-0.375(log10X), r2=0.996. Open
squares represent ﬁnal masses in bass treatments as a function of bluegill and
“neighbor” bass densities above the nominal target density of bass. The location of
these points relative to the regression line gives an indication of the relative
strengths of intra- and interspeciﬁc competition. In A), low bass points fall on the
line, and can not be classiﬁed as outliers (based on an outlier test p>0.10). In B),
grth rates in the bass treatment are above the line and both points an be
classiﬁed as outliers (p<0.05) , suggesting interspeciﬁc competition is weaker than
intraspeciﬁc competition in bluegill.

gill (B)
d prior
llS from

d
mom of

m the
ll 8).

:r than

79

 

#

 

B) Bluegill

1000

Bluegill Density (#/Quadrant)

 

 

6

V

M

(3) ssew lama lllﬁanla

10

 

 

A) Largemouth Bass

 

1000

 

 

Figure 15

N
(3) ssew Wald ssea

E

3

as,
-g-E‘

"Eb

3

m
‘2

bll

l0i

In

feet

mg:

80

competitive effect than an equivalent density of bluegill: if points fall below the line the

. conclusion is that equal densities of bass have stronger competitive effects. As Fig. 15a
shows, ﬁnal bass sizes in bass treatments fell very near the line describing bluegill
competitive effects, indicating that the effects of intra- and interspeciﬁc competition on bass
growth were similar in magnitude. For bluegill, interspeciﬁc competition with bass had
less of an effect than intraspeciﬁc competition because ﬁnal masses in the presence of 9
additional bass were clearly above the line (Fig. 15b). Therefore, the interspeciﬁc
interaction appeared to be asymmetric: competitive effects of bluegill on bass were stronger

than vice versa.

In natural systems, prey are generally more abundant than their predators, and in local
lakes bluegill outnumber bass by an average of 5 to 1 (Werner et al. 1977, Olson
unpublished data). I can examine the magnitudes of the “population level effects” of
bluegill on YOY bass growth by comparing bass growth in the low bluegill treatment to
bass growth in the bass treatment (representing a six-fold difference in neighbor densities)
Bass growth was much lower in the low bluegill treatment (Fig. 14a), suggesting that
differences in YOY bass growth rates among natural systems would thus be driven
primarily by variation in densities of bluegill rather than bass (because the two species have
equivalent competitive effects and bluegill are much more abundant than bass). Growth of
bluegill, on the other hand, is affected most strongly by bluegill (Fig. 14b) due to both the
low density of bass and the smaller effect of bass on bluegill growth (Fig. 15b).

Interspeciﬁc competitive effects were observed despite the fact that the two species were
feeding on very different prey. Overlap indices, calculated by Schoener’s Index (1975)
based on stomach contents at the end of the experiment, averaged 0.32 between bass and

target bluegill and 0.29 for bass and neighbor bluegill. These values are slightly higher

81

than overlap indices between natural populations of bass and bluegill (mean overlap=0.19:
Olson unpublished data). Interestingly, overlaps did not change among treatments (bass

and targets: F3,4=0.16, p<0.10, bass and neighbors: F1,3=2.92, p>0.10). Across a

gradient in ﬁsh density, the percentage composition by mass of different prey taxa were
very consistent for both bass and bluegill (the percent composition of only one prey taxa,
Chm changed across ﬁsh densities: p<0.001 using the sequential Bonferroni
technique of Holm (1979) cited in Rice (1989)).

In both the open water and vegetation habitats, bass and bluegill consumed prey in very
different proportions (Table 7). When feeding in open water, bass relied heavily on
calanoid copepods (which made up 88% of their open water prey), while bluegill fed more
on cladocerans (64% for targets, 80% for neighbors). Calanoid copepods were also
consumed by bluegill, but at a much lower proportion (Table 7). In the vegetation, bass
had relatively narrow diets. They utilized a total of 11 types of prey, and of those 6 made
up the majority of their diet. Insect nymphs (of the families Baetidae, Coenagrionidae,
Aeschnidae, and Libellulidae), W and dipteran pupae together composed 91%
of the prey eaten by bass in this habitat (Table 7). Bluegill had much broader diets when
feeding in vegetation, consuming 10 of the 11 prey types found in bass diets as well as 6
others. Target and neighbor bluegill had very similar diets (Schoener’s Index averaged
0.81). Chironomid larvae dominated diets of both bluegill groups, and no other prey taxa
comprised as much as 10% of the total prey mass. Only about 20% of bluegill diets came

from the 6 categories that were most abundant in bass stomachs (Bass Prey: Table 7).

In order to better understand how bluegill exerted a strong competitive effect on bass

despite relatively little overlap in resource use, I examined changes in prey abundances

 

TX 5 c n SEIIIII.‘\IIII]

82

Table 7: Taxonomic diet compositions of bass and bluegill at the end of the experiment.
Numbers represent mean percent contribution by massﬂ standard error for all sections
combined (n=8). Means were ﬁrst calculated for each section based on the following
numbers of bass, target bluegill, and neighbor bluegill respectively: section 1: 15, 10, 0,
section 2:7, 10, 20, section 3:5, 11, 0, section 4:9, 24, 20, section 5:12, 16, 0, section
6:5, 24, 20, section 7:8, 29, 0, section 8: 3, 17, 20. Small cladocerans represent
Diaphanosoma and Ceriodaphnia. Bass prey represent the combined totals of
Simocephalus, dipteran pupae, and Baetid, Coenagrionid, Aeschnid, and Libellulid
nymphs. Snails were represented by two groups: Gyraulus parvus and Physa spp.
Miscellaneous prey represent all other prey.

lit—82m 133—SS. MM _JL_N¢i borBltlsalll
Small Cladocerans 2.10 (0.69) 10.01 (2.97) 6.15 (1.47)
Calanoid Copepods 15.18 (6.09) 5.69 (2.73) 1.53 (1.01)
Bass Prey 72.11 (6.27) 17.19 (1.90) 21.83 (3.57)
Chironomids 4.14 (1.14) 52.15 (3.58) 47.20 (2.17)
Snails 0.00 (0.00) 0.50 (0.21) 1.05 (0.27)

Miscellaneous 6.56 (1.34) 14.69 (354) 22.33 (3.69)

 

doc

lhel

and:

mean

[11610

83

through time in the pond. Prey density and size structure were quantiﬁed for both open
water and vegetation dwelling invertebrates in each section. Strong treatment effects were
observed through time on invertebrates found in open water. As the experiment
progressed, total densities of zooplankton diverged among treatments and by the end of the
experiment, zooplankton densities were signiﬁcantly lower in low and high bluegill
treatments compared to target and bass treatments (Flg. 16). Zooplankton densities also
varied through time as a result of changes in species composition within the zooplankton
community (Fig. 17). Initially, densities of most cladocerans were kept low by
Chaoborus, an abundant invertebrate predator. After ﬁsh were introduced, Chaoborus
were quickly eliminated from all treatments (Fig. 17a), which released other zooplankton
from predation. One of the ﬁrst taxa to respond was M gm. Their densities
increased rapidly following ﬁsh stocking, particularly in target and bass treatments (Fig.
17b). This peak was followed by a rapid decline, and D. M were replaced by two
species of small-bodied cladocerans, Diaphanosoma and Ceriodaphnia (Fig. 17c). When
the experiment ended, these small cladocerans and calanoid copepods dominated the
community (Fig. 17c, d), and the observed differences in total density among treatments

were a result of strong ﬁsh effects on these taxa (small cladocerans: F3,4=41.40, p<0.002,

calanoid copepods: F3,4=34.54, p<0.003).

The replacement of D. m by the smaller Wm and W lead to a
decrease in mean zooplankton body size over time in all treatments (Fig. 18). As well,

there was a divergence in mean size among treatments. By the end of the experiment,
zooplankton were smallest in the high bluegill treatment, intermediate with low bluegill,
and they were largest in target and bass treatments (Fig. 18). The observed differences in
mean size at the end of the experiment were due to changes in taxonomic composition of

the zooplankton community (Fig. 17) as well as changes in mean size within individual

Figure 16: Changes in mean zooplankton density (:1 SE) over time for all taxa
combined. Symbols are as in Fig. 14. Initial densities did not differ (F3,4=2.16,

p>0.10). Final densities differed signiﬁcantly among the 4 treatments
(F3,4=231.83, p<0.0001). Bonferroni T tests distinguished 3 groups: target and
low bass treatments, low bluegill and high bluegill (p<0.05).

85

 

 

I—l

 

400+
300-
i
.3‘
E
0
‘3
$200-
3'
100-
0 l
’ July9

Figure 16

J.
0'.“.
‘1
1.
.L N
l I
August 4 August 24

 

86

Figure 17: Changes in mean densities (:1 SE) of 4 important zooplankton
categories over time. Symbols are as in Fig. 14. A) Chaoborus sp, B) Daphnia
pulex, C) small bodied cladocerans Diaphanosoma and Ceriodaphnia, and D)
calanoid copepods. Note different scales on the Y-axes.

87

       

 

«N semis a 349%

a be” e be.
_

 

 

  

 

 

 

3: «man Am

(l/#) M9090 m

vm «unmet

 

4 .392. 2 :3 a a:
_

 

 

Ia

ﬁm

 

 

and?

(l/#) Kllsnao SM 09 3

Figure 17

8 games. a games 2 :3 e be
h _ . .

 

88

 

83.33. Ease. E :3 e be.
- — n -

 

 

 

 

 

 

 

o
loll! Ill
I a... / a .
-2 .
-2:
la S
m
H . . a
o m D
a. . -3 W
/ a. saw
m
coo. l—I no. m. m
o.- .. 1- cm pm, I“...
- [Li a .m m
I_I ac. [In M. W I 8” W
o. a H Nu. I (
l .8
can use. . l m
c 83. m
rm.
cm W.
8888 22.26 a 888820 ensem— =aam 0 m
m.
F

 

89

Figure 18: Changes in mean zooplankton body length (:1 SE) over time. Symbols
are as in Fig. 14. Means are based on all zooplankton regardless of species
identity. Initial sizes were similar among treatments (F3,4=1.02, p>0.10). When
the experiment ended, signiﬁcant differences existed among treatments
(F3,4=94.94, p<0.0005). Bonferroni T tests did not ﬁnd signiﬁcant differences
between target and low bass treatments, but the low and high bluegill treatments
were separate from that group and from one another (p<0.05).

90

 

0.7 -

P
at
J

9
at
L

Mean Zooplankton Body Length (mm)

0.4 -

 

 

 

Figure 18

.‘x
I
1:...
as. \l*
I. T
.o. I
a... .L
.l -------------------- *4.“
T ---- g
.L
\:
.L
I I I I
July 9 July 19 August 4 August 24

 

91

taxa (Table 8). Small cladocerans and calanoid copepods were both signiﬁcantly smaller

in the presence of higher ﬁsh densities in the bluegill addition treatments.

Changes in the density and size structure of zooplankton had important consequences on
resource quality over time and among treatments. A foraging model developed for bluegill
demonstrated a general decline in predicted energetic return rates (E/I') over time in all
treatments (Fig. 19). This decline was driven primarily by changes in zooplankton size;
E/I‘ values decreased in some cases even when zooplankton densities increased (compare
density (Fig. 16) and E/T responses from July 19 to August 4). When the experiment
ended, there were signiﬁcant differences in E/I' between bluegill addition treatments and
target and bass treatments (Bonferroni T tests separated treatments into 2 groups: p<0.05).
Therefore, bluegill had a signiﬁcant negative effect on the quality of their resource. ,
Presumably, they had a similar effect on planktonic resources used by bass. For example,
in addition to their effect on cladocerans, increasing bluegill density also lead to a decrease
in mean size of calanoid copepods (Table 8), which were an important prey item in bass
diets (Table 7).

The effects of a reduction in available prey size were directly reﬂected in ﬁsh diets. Bass
and bluegill always fed on largest prey available, and prey size decreased for both species
across the gradient in bluegill density (Fig. 20). like their effect on growth rates,
increasing bluegill density lead to a non-linear decline in prey sizes in the environment and
in ﬁsh stomachs. Thus, per-capita impacts of bluegill predation on zooplankton size
structure were not constant, but were strongest at low density and decreased as density

increased.

In addition to their impact on zooplankton, ﬁsh also affected invertebrates associated with

92

Table 8: Mean body lengths (:1 SE)of small bodied cladocerans and calanoid copepods,
the two dominant zooplankton taxa, at the end of the experiment. Body length of both taxa
were signiﬁcantly different among treatments (small cladocerans: F3,4=41.40, p<0.002,

calanoid copepods: F3,4=34.S4, p<0.003). Different letters indicate means were
signiﬁcantly different at p<0.05 (Bonferroni T test).

Prey Type Target Bass Low Bluegill High Bluegill
Small Cladocerans 0.5848(0.002) 0.5508(0.017) 0.479b(0.003) 0.466b(0.003)

Calanoid Copepods 0.9398(0.038) 0.8853(0.017) 0.659b(0.018) 0.629b(0.028)

93

Figure 19: Changes in mean net energy return rates (:1 SE) over time for a bluegill
feeding in the open water. Symbols are as in Fig. 14 Energy return rates were
calculated in Joules/second (J/s) using a model developed by Mittelbach ( 1981).
Bluegill sizes used in calculations were chosen to represent bluegill sizes on each
date 7/09:36.5 mm, 7/19, 41.9 mm, 8/04, 41.9 mm. 8/24 48.6 mm) The same
sized bluegill was used for all treatments. See text for details. Treatments did not
differ initially (F3,4=2.74, p>0.10). By the end of the experiment, signiﬁcant
differences had emerged (F3,4=9.14, p<0.03).

94

 

 

 

 

 

 

0.3 -

I u. I‘ll.
_. lTkl
u _
2 J
0 0

$5 38803—0 8m 8am Eﬁom 38cm 32

 

July 19 August 4 August 24

July 9

Figure 19

95

Figure 20: Mean body length of zooplankton consumed by ﬁsh on the last date as a
function of bluegill density. Data were log transformed prior to analysis. Filled
circles represent target bluegills, ﬁlled squares are neighbor bluegills, and open
squares are bass. Filled triangles represent mean body length of zooplankton
available to ﬁsh in the environment on the last sampling date. Three regression
lines are also presented. The solid line represents mean body sizes of zooplankton
available in the environment: the dashed line is based on zooplankton found in
target bluegill guts, and the dot-dashed line represents mean prey size of
zooplankton found in bass guts. All three lines show a signiﬁcant (p<0.05) decline
across a gradient of bluegill density (environment: log10Y=-0.058-0.121(log10X),

r2=0.90, bass: logloY=0.034-0.081(logloX), r2=0.588, bluegill: log10Y=-0.077-
0.081(log10X), r2=0.837.

96

 

0.9

0.8

0.7

0.6

Mean Zooplankton Body Length (mm)

0.4

 

 

 

 

10

Figure 20

I
100
Bluegill Density (#lQuadrant)

1000

97

the vegetation. These effects were limited to soft-bodied prey. Bluegills occasionally fed
on snails (Table 7), but they had no observable effect on either density or mean size of this
prey category (density: F3,4=1.46, p>0.10, size: F3,4=1.17, p>0.10). Because of this, all
subsequent analyses on vegetation dwelling prey were restricted to soft-bodied

invertebrates.

Treatment effects on vegetation dwelling invertebrates were observed primarily through
changes in size structure; ﬁsh had little effect on prey density (Fig. 21a). Initial densities
were similar among treatments for almost all taxa: Chironomid larvae were the only taxon

that differed among treatments, but were most abundant in the high bluegill treatment
(F3,4=34.06, p<0.003) and stayed that way through the experiment. Only one prey group,
amphipods, showed signiﬁcant treatment effects on their density, and 2 more
(Simocephalus and mites) were marginally signiﬁcant (amphipodst3,4=13.96, p<0.02,
W: F3,4=4.76, p<0.09, mites: F3,4=4.74, p<0.09). Because the majority of
invertebrate taxa in the vegetation showed no response, combined densities of all soft-

bodied prey were not signiﬁcantly different among treatments when the experiment ended
(Fig. 21a).

Changes in prey size structure were reﬂected in reductions in mean size. Invertebrate
body lengths decreased over time, and also diverged among treatments (Fig. 21b). When
the experiment ended, prey were smallest in the high bluegill treatment, intermediate with
low bluegill densities, and largest in target and bass treatments (Bonferroni T test separated
treatments into these 3 groups: p<0.05). These differences were driven by changes in the
abundance of large prey. Invertebrates > 6.0 mm in length were signiﬁcantly less abundant

in low and high neighbor bluegill treatments compared to target and bass treatments

98

Figure 21: Changes in the soft-bodied littoral invertebrate community over time.
Symbols are as in Fig. 14. A) Changes in mean density (:1 SE) of all taxa
combined. Initially, treatments did not differ (F3,4=1.37, p>0.10). Final densities
were also not signiﬁcantly different from one another (F3,4=1.91, p>0.10). B)
Changes in mean body length (:1 SE) of all taxa combined. There were no
significant differences when the experiment began (F3,4=0.25, p>0.10). When the
experiment ended there were signiﬁcant differences among treatments (F3,4=14.07,
p<0.02).

99

mm .m=m=<
-

R .22
p

w :3
-

 

 

 

 

 

.8

(ma!) uﬁm Kpoa 919199119491 mm

3 same:
—

R :3
-

 

 

 

83

s

(gm/it) M91190 amqauawr

é

Icccv

 

 

E

Figure 21

100

(F3,4=9.77, p<0.03. Bonferroni T test separated treatments into 2 groups at p<0.05),
creating a truncated size frequency distribution and thus a lower mean. Many different prey
taxa contributed to this result. Simocephalus, baetid nymphs, and larval leptocerids,
tanypodids, and chironomids were all smaller in treatments with additional bluegill
neighbors (Table 9). One consequence of this reduction in prey size in low and high
bluegill treatments was to lower the quality of vegetation dwelling invertebrates as a
resource for bluegill and presumably bass (Fig. 22). Net energetic return rates were
signiﬁcantly different among treatments when the experiment ended (Fig. 22). Overall E/l‘
values were very sensitive to the abundance of large prey, and as bluegill densities

increased these prey became increasingly rare.

When available prey sizes decreased, there was a corresponding reduction in the sizes of
prey eaten by bass and bluegill. Mean body lengths of invertebrates in stomachs of both
species declined across the gradient of bluegill density at the end of the experiment (Fig.
23). Similar to the observed pattern in zooplankton, the reduction in available and
consumed prey size was not linear, but rather declined exponentially with ﬁsh density.
Again, bluegill did not have a constant per-capita effect; instead their effect decreased as
density increased.

Results from diet analysis also showed that bluegill ate much larger littoral prey than
bass, and that bass fed on the same sized prey as was available (Fig. 23). This suggests
that bluegill were more effective at feeding on large prey than bass. However, bass fed on
only a subset of available prey, and on these prey the pattern of size selection was very
different. I deﬁned important bass prey as those groups making up 92% of bass diets in
the littoral zone (i.e. W dipteran pupae, and nymphs of baetids, coenagrionids,
libellulids, and aeschnids). Mean sizes of these prey available in the environment decreased

101

Table 9: Final mean body lengths (: 1 SE) of the ﬁve vegetation dwelling invertebrate taxa
that showed signiﬁcant treatment effects (baetid nymphs: F3,4=8.61, p<0.05, leptocerid

larvae: F3,4=7.27, p<0.05, tanypodid larvae: F3,4=9.23, p<0.05, Simocephalus:
F3,4=41.47, p<0.05, and chironomid larvae which were marginally signiﬁcant: F3,4=5.52,
p<0.07).

Prey Type Target Bass Low Bluegill High Bluegill
Baetid Nymphs 3.30 (0.18) 3.32 (0.02) 3.12 (0.02) 2.73 (0.02)
Leptocerid Larvae 3.75 (0.05) 3.61 (0.15) 3.19 (0.23) 2.89 (0.09)
Tanypodid Larvae 5.37 (0.16) 5.10 (0.09) 4.60 (0.27) 4.23 (0.06)
W 1.88 (0.05) 1.93 (0.01) 1.76 (0.03) 1.49 (0.03)

Chironomid Larvae 8.38 (0.21) 8.38 (0.52) 7.01 (0.44) 6.71 (0.25)

102

Figure 22: Changes in mean net energy return rates (:1 SE) over time for a bluegill
foraging in the littoral zone. Symbols are as in Fig. 14. Bluegill sizes are the same
as in calculations used for Fig. 6 (36.5, 41.9, and 48.6 mm SL for the 3 sample
dates). Initial foraging return rates did not differ among treatments (F3,4=2.74,

p>0.10). Final rates were signiﬁcantly different (F3,4=9.76, p<0.03).

103

 

 

 

 

' I
JUIY 27 August 25

July 8

 

 

0.05

0.04 -

I -
3 2
o 0.
0 0
33 85— 530m Egon—m— .02

0.01 -

Figure 22

104

Figure 23: Mean sizes of littoral prey consumed by ﬁsh at the end of the experiment
as a function of bluegill density. Data were log transformed prior to analysis.
Symbols of points and regression lines are as in Fig. 20. All regressions are
signiﬁcant (p<0.05) and show a decline in prey size with increasing bluegill density
(environment: logloY=0.698-0.077(logloX), r2=0.92, bass: logloY=0.747-

0.100(logloX), r2=0.53, bluegill: long=0.921-0.112(logloX), r2=0.72.

Mean Invertebrate Body Length (mm)

105

 

 

 

 

7
6
I
5 ‘ ‘ ~ ‘ ~ ~ \
. \ U ‘ ~
. .
4
3
I
10 100
Bluegill Density (#lQuadrant)

Figure 23

 

1000

106

across bluegill densities (F3,4=53.53, p<0.0001); however, bass had a higher mean prey
size in their stomachs than what was found in either the environment or bluegill stomachs
(Fig. 24). Furthermore, bluegill fed on prey smaller than those available, particularly in the
low and high bluegill treatments (Fig. 24). Therefore, bass were more selective for large
prey than bluegill on the subset of prey types consumed by bass, suggesting that bass may
feed more efficiently on these prey types. However, bluegill were still able to reduce prey
size structure, and through that mechanism had a strong competitive effect on the growth

rates of bass.

IAKE SURVEY

Data collected from populations in local lakes are consistent with the observation that
mall bluegill have strong competitive effects on YOY bass. As would be expected if the
two species compete for invertebrate prey in natural systems, growth rates of small bass
(expressed as mass at the end of their first growing season) and small bluegill (in g/yr: see
Mittelbach and Osenberg 1994) were positively correlated among lakes (Fig. 25a).
Furthermore, bass size at age 1 declined with bluegill density (Fig. 25b). Interestingly,
these effects were observed despite the fact that diet overlaps in natural populations of bass
and bluegill were even lower than observed in the experiment (mean overlap index was

0.19 : 0.03: Olson, unpublished data).

While increasing densities of small bluegill had a negative effect on small bass, large
piscivorous bass beneﬁtted from an increase in the abundance of these potential prey.
Mass speciﬁc grth rates of large bass showed a signiﬁcant positive correlation with
small bluegill density (Fig. 26). Therefore, as bass got larger, bluegill switched from

having a negative effect when they were competitors to a positive effect when they became

107

Figure 24: Mean sizes of important bass prey (Simocephalus, dipteran pupae,
baetid nymphs, coenagrionid nymphs, libellulid nymphs, and aeschnid nymphs)
consumed by ﬁsh at the end of the experiment as a function of bluegill density.
Data were log transformed prior to analysis. Symbols of points and regression
lines are as in Fig. 20. All regressions show a signiﬁcant (p<0.05) decline in prey
size with increasing bluegill density (environment: logloY=0.630-0.073(long),
r2=0.89, bass: log10Y=0.747-0.100(log10X), r2=0.53, bluegill: log10Y=0.905-
0.226(log10X), r2=0.87.

Mean Bass Prey Body Length (mm)

108

 

 

 

 

 

10

Figure 24

I
100
Bluegill Density (#/Quadrant)

1000

109

, Figure 25: Small bass growth rates in natural populations as a function of small
bluegill growth and density. Bass growth is expressed as mean mass at age 1:
Bluegill data were taken from Mittelbach and Osenberg 1994. A) Small bass
growth vs small bluegill growth (m0.93, n=7, p<0.003). B) Small bass growth
vs. small bluegill density (r=-0.95, n=7, p<0.001).

110

29 as: 5.55 amnam =eam Awe 83: sense :85 :28

 

 

 

 

 

 

 

 

 

h c m e m c
n n — p N. N
e
Iv I
e e
e a a
o m m
w w
I0 I
m e m
e e
e
I” I”
e
x
m
2 S .m
E F

111

Figure 26: Large bass growth rates in natural populations as a function of small
bluegill denstiy. Bass growth rates are expressed as change in mass (g/yr),
calculated from mass speciﬁc growth rate regressions calculated for each lake.
Growth rates reported are estimated for a mean sized three year old bass (85 .73 g).
The relationship between large bass growth and small bluegill dneisty is signiﬁcant
(r=0.93, n=7, p<0.003).

Large Bass Growth Rate (g/yr)

110

112

 

 

 

 

 

Figure 26

-

I I

10 15
Small Bluegill Density (#/100 m2)

20

113

prey. As a result of this ontogenetic change in the interaction between bass and bluegill,
the two life history stages of bass responded very differently to changes in bluegill density.
Across a gradient in bluegill density, growth rates of small bass were negatively correlated

with the growth rates of large bass (r=-0.93, p<0.003).

DISCUSSION

CDMPETITION EXPERIMENT

Interspeciﬁc competition between predator and prey is potentially common in size-
structured populations. The ontogeny of predators is typically marked by a succession of
niche shifts to progressively larger prey (Werner and Gilliam 1984). Since body size is
often correlated with trophic level, a consequence of these shifts in prey size is that a
predator may pass through several different trophic levels during its lifetime (Stein et
al.1988). Therefore, a predator may occupy the same trophic level with other species when
they overlap in size, but as the predator grows it will move up a trophic level and prey on
those same species. This type of mixed competition/predation interaction has been
suggested in many ﬁeld studies which have shown substantial diet overlap between at least
some size classes of predator and prey (see Polis et al.1989). However, few studies have
actually demonstrated competition to be an important process in these systems (Wilbur
1980, Morin 1983, Tonn and Pasczkowski 1986, Tonn et al.1986, Persson and Greenburg
1990).

Strong competitive effects were observed in the growth rates of small bass and small
bluegill when the two species fed on aquatic invertebrates. Similar results were found by

Gilliam (1982) in another experimental manipulation of bass and bluegill densities.

114

Furthermore, in natural systems competitive effects of bluegill on bass will be strengened
by a negative feedback loop (Chapter 1). If competition is weak and bass grow well on
invertebrates, they quickly become large enough to consume YOY ﬁsh (which is a size
dependent process: Lawrence 1958, Timmons et al.1980) and shift out of the invertebrate
feeding stage. Thus, weak competition leads to an early shift to piscivory, whichfurther
enhances growth rates (Chapter 1). However, if competition with bluegill is strong, it can
reduce growth rates to a point where bass are not large enough to consume ﬁsh. Instead,
bass are forced to remain in the invertebrate feeding stage, and continue to compete with
bluegill. In this case, competition prolongs the duration of the invertebrate-feeding stage,
and as a result bass are smaller when the season ends. Piscivory Was not possible in our
experiment (i.e. YOY bluegill were not present), and all bass fed on invertebrates for the
entire experiment regardless of size. Therefore, competitive effects were underestimated
because the invertebrate-feeding stage lasted the same length of time in all treatments. If
large bass in low density sections had been allowed to shift to piscivory, competitive

effects of bluegill on bass growth would have been even more pronounced.

Bluegill had a strong competitive effect on themselves and bass through their impacts on
invertebrates in both open water and vegetation habitats. This was driven primarily by
changes in invertebrate size-structure (particularly in the vegetation where bluegill had no
observable effect on invertebrate densities). Prey size is an important determinant of
resource quality for ﬁsh (Paloheimo and Dickie 1966, Werner 1986). Increasing densities
of bluegill decreased the abundance of the largest, most proﬁtable invertebrates, creating a
- strong reduction in net energetic return rates for bluegill and presumably bass feeding in
either habitat. Bass and bluegill were forced to feed on smaller prey which lead to a strong

reduction in growth rates of both species across a gradient in bluegill density.

115

Results of this experiment also suggest that interspeciﬁc competition was asymmetric.
While bluegill had strong effect on bass, the reciprocal effect of bass on bluegill was much
weaker. The experiment actually found no signiﬁcant effect of bass on bluegill. However,
low recovery rates of bass created a small effect size between target and bass treatments that
made it difﬁcult to detect treatment effects (mean recovery rates in target sections was 10
bass and for bass sections was 28 bass). Regression analysis of target bluegill growth
rates suggests there was an effect of bass, but it was weaker than the intraspeciﬁc effect of
bluegill on themselves (Fig. 15). In contrast, eﬁ‘ects of bass and bluegill on bass were

similar (Fig. 15).

The observed asymmetry in bass-bluegill competition is consistent with model
predictions of Polis et al.(1989) for mixed competition/predation interactions (which they
term intra-guild predation). As long as predators can survive on the shared resource
(which bass can do: Hodgson and Kitchell 1987), Polis et al.( 1989) assert that asymmetry
is required for coexistence of predator and prey. Otherwise, prey will be excluded by the
dual forces of competition and predation. Asymmetric competition between predator and
prey is also predicted based on laboratory feeding performance studies (between bass and
bluegill, Werner 1977; between eurasian perch (Legal ﬂuviatilis) and roach (Mills
DEM), Persson 1987, 1988). Results of these performance studies suggest that ﬁsh face
a morphological tradeoff in feeding on prey that are large or small relative to their own
body size. Because predators spend much of their lives feeding on relatively large prey (up
to half their own body length: Timmons et al.1980), they are not morphologically adapted
for capturing small prey when they are small. Prey ﬁsh (like bluegill or roach) differ in that
they feed from a narrower range of relative prey sizes through their ontogeny and thus can
be morphologically adapted to this prey, making them much more efﬁcient than their
predator. Based on these differences in efﬁciency, bluegill are predicted to be superior

116

competitors to bass (Werner 1977).

Limitations set by bass morphology may restrict their diets to a subset of available prey in
the invertebrate community. In this experiment, bass fed on fewer types of prey than
bluegill, and tended to specialize on a small number of prey taxa (e.g. calanoid copepods,
Simocephalg, ephemeropteran and odonate nymphs, etc.). These are generally active,
mobile prey for which bass morphology is well suited. Bass have a fusiform body shape
and large mouth which enable them to rapidly pursue and engulf elusive prey (Werner
1977, Winemiller and Taylor 1987). Interestingly, on this subset of prey bass may be
more effective than bluegill. Bass ate larger prey in all treatments, and there is evidence to
suggest that encounter rates on some of these prey (baetid nymphs) are higher for bass than
bluegill (Gilliam 1982). Bluegill use a different foraging strategy from bass that is slower
and more deliberate (Mittelbach 1981a, Gilliam 1982). This mode of foraging may be less

effective for feeding on elusive prey.

Therefore, the mechanism of competitive asymmetry is not the same as predicted by
Werner (1977). Rather than being driven by differences in efﬁciency when feeding on
shared prey, asymmetric competition appears to be the result of differences in diet (niche)
overlap (Colwell and Fuentes 1975). The bass niche is included within the bluegill’s:
around 20% of bluegill prey came from those categories important to bass, and almost all
of the prey types found in bass diets were also found in bluegill. Bluegill, on the other
hand, have many prey types (e.g. chironomids, leptocerids, tanypodids) that are rarely or
never consumed by bass, and thus offer bluegill a competitive refuge. Chironomids are
especially important because they are the single dominant prey type in bluegill diets and are
one of the most abundant invertebrate taxa in the littoral zone (Mittelbach 1981b).

Asymmetric competition occurs because bass can only affect a subset of available

117

invertebrates which make up a small percentage of bluegill diets, while bluegill affect all
invertebrates important to bass. An analogous situation occurs between bluegill and their
congener, the green sunﬁsh (Lemmis cyanellus: Werner and Hall 1977). Green sunﬁsh
are habitat specialists that outcompete bluegill when both species are restricted to the
vegetation. However, when other habitats are available, bluegill expand their niche to
include open water and bare sediments (in addition to vegetation), and the asymmetry of
competition is reversed (Werner and Hall 1979). Bass differ from green sunﬁsh in that
they specialize on prey types rather than habitat types. While bass might outcompete
bluegill if those were the only prey available, they are inferior competitors when a diverse

assemblage of prey exists.

Even though bluegill fed on smaller shared prey than bass, they still had strong impacts
on the size-structure of these invertebrates. In this regard, bluegill should be considered
“effect competitors” (Goldberg 1990). Through the occasional consumption of large prey,
or by eating prey while prey were small and preventing their recruitment to larger sizes (see
Taylor 1980, Barry and Tegner 1991), bluegill can reduce the abundance of large prey for
bass. Bass are very sensitive to changes in prey size; growth rates in the experiment were
much lower when large prey were rare. Unlike bluegill, bass had very little effect on
invertebrate size-structure, particularly on invertebrates important to bluegill (which were
rarely consumed by bass). Instead, these invertebrates were affected predominately by
bluegill. This agrees with Mittelbach (1988), who also found intraspeciﬁc competition in
bluegill using a cage experiment in a natural lake. In his experiment, bluegill were strong
effect competitors through their impact on the abundance of large prey, which had
important consequences for their growth rates. Results of the present experiment suggest

the same mechanism also reduces growth rates of bass.

118

LAKE SURVEY

Results of the competition experiment suggest that interspeciﬁc competition for littoral
invertebrates can be an important process affecting growth rates of YOY bass. Competition
was particularly strong at the population level where bluegill were ﬁve times more abundant
than bass. This suggests that in natural systems, interspeciﬁc competition with bluegill can
potentially have strong affects on grth rates of YOY bass. Data collected from the seven
study lakes supports this prediction: bass size at the end of their ﬁrst year showed a striking
decline across a gradient in small bluegill density. As well, growth rates of YOY bass and
small bluegill were positively correlated, which suggests that the two species respond

similarly to a gradient in resources.

However, the negative effect of bluegill was only observed for small bass: growth rates
of large bass were greatest in lakes with high densities of small bluegill, as expected based
on observations that bluegill are an important prey item in large bass diets (Swingle and
Smith 1940, Dillard and Novinger 1975, Keast1985). Therefore, the bass population is
split into two stages that responded very differently to changes in bluegill abundance. The
ﬁrst stage is made up of young-of-year bass that are negatively affected by bluegill as a
result of direct competition and by preventing bass from shifting to piscivory. The second
stage consists of larger, older bass that have shifted to piscivory. Like bass, bluegill
populations are also stage-structured. Small bluegill feed on invertebrates in the littoral
zone (Mittelbach 1981a, 1984) where they compete with bass. As well, bluegill in this
stage are vulnerable to predation by large bass. Large bluegill, on the other hand, feed on
zooplankton in the open water and have no direct interaction with either stage of the bass
population. Therefore, bass and bluegill populations are linked by two stage-speciﬁc

processes: competition among small ﬁsh for littoral invertebrates, and predation by large

119

bass on small bluegill (Fig. 13).

Among lake variation in growth rates and densities of bass and bluegill density suggest
that the interactions between these two species are strong (Table 10,11). Variation in
growth and density of bluegill is driven by variation in zooplantkon productivity (Osenberg
et al.1988, Mittelbach et al.1988, Mittelbach and Osenberg 1994): lakes with higher
zooplankton productivity have higher densities of both small and large bluegill. I Increased
densities of small bluegill intensiﬁes interspeciﬁc competition for littoral invertebrates,
which lowers growth rates of YOY bass and prolongs the duration of the ﬁrst stage (Table
11). Beeause overwinter mortality is size-dependent (Davies et al.1982, Gutreuter and
Anderson 1985), increased competition can potentially limit per-capita recruitment rates of
bass to the piscivorous stage. This cost is countered by the beneﬁt of small bluegill to the
grth rates of large bass. An increase in large bass growth is predicted to result in
increased production rates of bass, because fecundity is strongly correlated with size in ﬁsh
(Carlander 1977, Baganel 1978). Large bass densities still increase, suggesting the stock-
recruitrnent curve is monotonic (Mittelbach and Chesson 1987), but the density response is
dampened by a recruitment bottleneck in the ﬁrst stage, and densities only partially respond
to increase in bluegill density (Table 11).

In this set of study lakes, both bass and bluegill show stage-structured responses to an
increase in the productivity of the second stage’s resource. These increases enhance
growth and fecundity of adults, which leads to increased interspeciﬁc competition between
bass and bluegill in the ﬁrst stage. As a result, growth rates of both small bass decrease.
The divergent responses between stages generate patterns that are very different from non-
structured populations. Decreased recruitment rates (a result of lower growth and higher

mortality in the ﬁrst stage) decouple large ﬁsh from their resources and prevent a complete

120

Table 10: Growth rates and ﬁsh densities of bass and bluegill in seven lakes. All
growth rates are measured in g/year; bluegill densities are in #/100 m2 and bass
densities are in CPUE’s (#/seine haul). Growth rates and densities of small (20-50
mm SL) and large (60-100 mm SL) bluegill were taken from Mittelbach and
Osenberg (1994). Growth rates of YOY bass were estimated as the average mass at
the end of their ﬁrst year. Growth rates of large bass were estimated by ﬁrst
regressing log change in mass over the course of a year vs. log mass at the start of
the year for each lake. Then, growth rates were determined for the mean mass of a
3 year old bass (85.73 g). Densities of bass were estimated as catch-per-unit-
efforts from beach seines.

121

EA
de
ewd
omd
hmd
bed
and

So on: .3 So on: >3

86
mod
Nmﬁ
<2

5N6
3%
<2

mmde
bwﬁc
vmﬁca
mode
mode
chmw
Mada

so 25 .3 .5 £5 >0».

med
med
cm.m
2k.
8%

ﬂaw

mm.v

m3
2.:
Nam
Fro
e5
23
one
So mm .3

WE mdu
ed wda
EMH wen
WN de
5w céa
ma H Non
QNH v.2

1m
HS
mum
w.m
o4.
wﬁ
Ev

3:83

a
g
Hosea—am
ooaogﬂ

moon—

33:0

.eoo mm .am .5 mm .3 .5 mm .am one:

Table 10

122

Table 11: Correlations among ﬁsh densities and growth rates given in Table 5.
Correlations involving YOY bass denstiy have N=5 and are signiﬁcant at p<0.05

when r < -0.87 or r > 0.87, all others have N=7 and are signiﬁcant (p<0.05) when
r < -0.75 or r > 0.75.

123

mh.o+
.36...

36.
who...
med...
No.0,:

wed-

.eoo £5 .3

cad...
med.
Hmd+
wwd...
hvd+
ad-
.59 2:5 >0?

ewd-
emd+
Had...
med...
ad.

.5 £5 .3

and-
3.?
end.
med...

.HO 2:3 >0?

cod...
Ned+
and.

duo on: >0»

.5 on: .3

so on: >0»

con 3 .3

- don mm .Em
m2: - .5 mm .3
was. and- so 3 .8m

593 .3 con 3 .am .6 .mm .3

Table 11

124

density response. Instead, densities partially respond, and bass and bluegill both
show the unusual pattern of a positive correlation between growth and density in
the second stage (unlike the ﬁrst stage where growth declines with density for both
species). As well, decreased recruitment dampens a consumer’s impact on their
resource and as a result, overall densities (and biomasses) of bass and bluegill

increase in parallel (Table 11).

Positive correlations in biomass of adjacent trophic levels differ from predicted
patterns for simple non-structured food chains. If populations have no size— or
stage-structure, a model devel0ped by Oksanen et al.(1981) predicts increases only
in alternate trophic levels, (starting with the top level), and no correlation in
biomass of adjacent trophic levels. In the bass-bluegill system, however, there is
not a simple food chain. Increased productivity intensiﬁes competition between
small bass and bluegill and prevents a complete response by the t0p trophic level.
Bass biomass does show an increase, but not enough to drive bluegill down to a
constant level, and biomasses of both stages increase. McCauley et al.(1988) and
Mittelbach et al.(1988) show that this is an expected result when predator death
rates increase (via decreased recruitment rates or other factors) across a gradient in
productivity. Stage-structure in bluegill also generates a positive correlation
between bluegill and zooplankton biomass (expressed as E/I‘: Mittelbach et al.1988,
Mittelbach and Osenberg 1994). As a result, three consecutive trophic levels all

show increased biomasses across a gradient in productivity.

Positive correlations in biomass are not an automatic result of a competitive stage
between adjacent trOphic levels. If the negative effect of competition is stronger

than the positive effect of predation, a negative correlation would be expected.

125

Negative correlations have been found between biomasses of roach and their
predator, eurasian perch, in Scandinavian Lakes (Persson et al.1988, 1992).

Roach compete with small perch directly for zooplankton and also have an indirect
negative effect by intensifying intraspeciﬁc competition in larger perch (Persson,
1987, Persson and Greenburg 1990). With increasing lake productivity, roach
become more abundant and their negative effect on perch imposes a severe
bottleneck on recruitment that actually reduces the number of piscivorous perch. As
a consequence, overall perch biomass decreases with productivity. One reason the
negative effect outweighs the positive effect is that perch spend several years in the
invertebrate feeding stage before they are large enough to switch to piscivory. Bass
differ from perch in that they are piscivorous after their ﬁrst year (if nOt sooner).
Most of their lives are spent as predators, so it is not surprising that the net effect of
bluegill is positive. Nevertheless, the negative effect is still important, and is

responsible for dynamics that are very different from typical predator-prey systems.

Through the competitive stage, bluegill impose a bottleneck on bass recruitment to
the piscivorous stage, and thus limit densities of the eventual predator. This is very
different from non-structured systems, where predators are an important force
structuring prey populations from the top of the food chain. Note that bass are still
important; it is the risk of predation that initially generates the stage-structure in
bluegill. Regardless, results of this study ﬁt within a developing framework which
views bluegill as the key player organizing lake communities (Mittelbach and
Osenberg 1994, Osenberg et al.1993). Because of the niche shift in their life
history, bluegill affect resources in both limnetic and littoral zones, and potentially
link these two important habitats of lakes through their population dynamics.

Furthermore, impacts on these resources and species that rely on these resources

126

are likely to be strong because bluegill are the dominant ﬁsh species in many lakes,
making up to 80% of the total ﬁsh biomass (Werner et al.1977). Previous studies
have shown that bluegill have strong effects on another sunﬁsh, the pumpkinseed
(Lemmis gibbosus: Mittelbach 1984, 1988, Osenberg et al.1988). Like bass and
bluegill, pumpkinseeds also change diets as they grow, shifting from soft-bodied
invertebrates to snails. Competition occurs between bluegill and pumpkinseeds in
the ﬁrst stage, which affects recruitment rates of pumpkinseeds to the
molluscivorous stage (Osenberg et al.1988, 1992). Consequently, dynamics of
large pumpkinseeds and their snail prey also become linked to bluegills, even

though there is virtually no direct interaction (Osenberg et al.1992).

The present study suggests that bluegill strongly inﬂuence largemouth bass
through stage-structured interactions. Through their effect on the size-structure of
invertebrate prey, bluegill substantially reduce prey quality for small bass. This
leads to a reduction in growth rates of bass, which in turn inﬂuences subsequent
recruitment rates into the piscivorous stage. Therefore, dynamics of the
piscivorous stage become linked to competition, resulting in unique patterns in
growth rates and densities of predator and prey. Because of ontogenetic niche
shifts in bluegill and bass, the overall interaction is a mixture of competitive and
predatory stages. Interspeciﬁc effects become very size dependent, and as a result
the interaction between bass and bluegill is very different from either competition or

predation alone.

CHAPTER 3
ASYNCHRONOUS SPAWNING IN PREDATOR-PREY FISH COMMUNITIES
AND THE ROLE OF ENVIRONMENTAL UNCERTAINTY

127

128

INTRODUCTION

A commonly observed pattern in temperate freshwater ﬁsh is for piscivores to spawn
before their prey (Keast 1985). This pattern holds for specialized piscivore families (e.g.
Esocidae and Lepisosteidae, Scott and Crossman 1979) as well as piscivorous species of
more generalized families (e.g. Stizostedion F: Percidae and Micropterus F: Centrarchidae,
Scott and Crossman 1979). For example, within the sunﬁshes (Centrarchidae), the
piscivorous largemouth and smallmouth bass (Microp_terus salmoides and_lVL dolomieui
respectively) generally spawn 4 to 5 weeks before their common prey ﬁshes, the bluegill

(Lemmis macrochirus) and other Lemmis species (Scott and Crossman 1979, Keast
1985)

For the piscivore, the adaptive value of early breeding is obvious; spawning before prey
gives young-of-the-year (Y OY) piscivores time to develop a size advantage over YOY prey
ﬁsh. Piscivores require a size advantage in order to successfully consume their prey
(Lawrence 1958, Timmons et. al. 1980). Early growth rates can play a critical role in
determining this size advantage (Chapter 1). As well, the duration of the “headstart” before
prey spawn will affect the relative sizes of predator and prey (Chapter 1). The ability to
shift from a diet of invertebrates to YOY ﬁsh in the ﬁrst year is vital for rapid growth
(Paloheimo and Dickie 1966, Aggus and Elliot 1975, Keast and Eadie 1985, Madenjian
and Carpenter 1991, Chapter 1)), which in turn affects the ability of piscivores to survive
through their ﬁrst winter (Oliver et al. 1979, Adams and DeAngelis 1987). Shelton et al.
(1979) and Timmons et al. (1980) found that overwinter survival within a cohort of
largemouth bass was limited to those ﬁsh which were piscivorous as YOY; mortality of

smaller ﬁsh that were unable to shift to piscivory was near complete. Studies of walleye

 

(Stizostedion vitreum vitreum; Maloney and Johnson 1957, Mills et al. 1987) and
largemouth bass (Aggus and Elliot 1975, Hackney 1975) have also demonstrated a strong

129

relation between the recruitment success of YOY piscivores and the abundance of YOY
prey ﬁsh. Clearly, there is a selective advantage for piscivorous species to spawn before

their prey.

However, simply recognizing the adaptive value of asynchronous spawning for
piscivores does not explain why the pattern exists. We must also know why prey spawn
late (Keast 1985). It is easy to envision an advantage for prey that spawn early. By
reproducing when piscivores do, the size advantage of YOY piscivores would be
minimized. Without a size advantage, piscivores are not able to shift to predation.
Furthermore, predator growth rates are typically much lower when they feed on alternate
food, making it relatively easy for prey to maintain a size refuge through the growing
season (Keast and Eadie 1985, Chapter 1). Selection for this behavior should be strong
because predation by YOY predators can be a substantial source of mortality (Fomey 1971,
Olson unpublished data). Yet, early spawning by prey is rare. Given the apparent
advantage of spawning earlier in the season, delayed spawning by prey is a puzzling

behavior.

Several hypotheses have been proposed to explain asynchronous spawning of predator
and prey (Keast 1985). One possibility is that phylogenetic constraints may prevent prey
from spawning any earlier than they already do. Alternatively, there may be hidden costs
which would make early spawning a poor strategy. For example, prey may delay
spawning until aquatic macrophytes have grown enough to provide their offspring with a
refuge from predators. Or, food for YOY ﬁsh may not be very abundant early in the
reproductive season, so if prey spawned early their offspring would die of starvation,
rather than be consumed by YOY predators.

130

Another possible explanation for the pattern is that asynchronous spawning reﬂects
different responses by predators and prey to an ecological trade-off between biotic and
abiotic processes. Fish that reproduce early run the risk of subjecting their offspring to
periods of reduced temperature and/or strong winds that create turbulent waters. Both of
these factors are known to inﬂict substantial mortality on eggs and larvae of ﬁsh
(Summerfelt 1975, Eipper 1975, Clady 1976). The advantages of spawning early must
therefore be tempered with the potential for'poor weather conditions. For prey, the costs
may outweigh the beneﬁts, and for them it might be better to delay reproduction.
However, this conservative strategy may not be possible for piscivores. Without a size
advantage over their prey, YOY piscivores will not be able to capture ﬁsh very effectively
(Werner 1977, Hoyle and Keast 1987,1988, Hambright 1991), and low growth rates
resulting from this would lead to high overwinter mortality (Oliver et al 1979, Adams and
DeAngelis 1987).

Below, I formalize this hypothesis using a simple game theoretic approach and evaluate
how environmental uncertainty effects the relative spawning times of predators and prey.
The problem can be viewed as a game between iteroparous species of predator and prey,

each trying to maximize the number of offspring surviving through their ﬁrst winter.
THE MODEL

Within the framework ofthis model, the success of a particular spawning strategy (for a
ﬁxed reproductive effort) will depend on three factors: 1) when the player spawns, 2) when
its opponent spawns, and 3) what the environmental conditions experienced by the
offspring of each player are like. Although spawning can occur continuously for extended

periods of time, as a simpliﬁcation only two times will be considered, early and late. The

131

distinction between these categories is that ﬁsh which spawn early risk poor environmental
conditions, whereas late spawners experience benign conditions (i.e. weather is only a
factor early). Early and late refer to relative spawning times. In absolute terms, both

strategies occur early enough in the season to potentially allow YOY ﬁsh to survive through

the winter.

Environmental conditions are likewise represented by two distinct states; good and bad.
In a good year, conditions are optimal for spawning, and YOY mortality is minimal. In
this case, there is no difference in mortality between spawning early or late. Bad years, on
the other hand, are years of near complete YOY mortality (or year class failure for the entire
population). Although species may differ in their tolerance to poor environmental
conditions, this effect will not be considered here. Bad years lead to heavy YOY mortality
for both species. It is important to note that bad years are not necessarily entire seasons of
poor weather. All that is required is a period of harsh conditions (temperature or wind)
after spawning has occurred, resulting in heavy mortality of eggs and larvae (see Eipper
(1975) and Clady (1976) for examples).

The reproductive success (i.e. number of offspring surviving through their ﬁrst winter)
resulting from each player's spawning time, given the spawning time of its opponent, can
be represented by a single value. If either or both players spawn early, this payoff must be
weighted by the frequencies of good and bad years. In the model developed below, these
payoffs are expressed as values relative to the other combinations of spawning times for
each player. However, absolute values of reproductive success could easily be used.
Because this is a two player game, each payoff represents the relative reproductive success
of an individual. If there is no strong frequency dependence, these individual payoffs also

represent relative year class strength for an entire population. For each player, the best

132

strategy to use is the one which yields the highest average payoff (Maynard Smith 1982).
When the payoffs for early and late spawning are determined, two separate calculations are

necessary; one for each possible strategy adopted by the opponent.
PREDATOR PAYOFF MATRD(

The predator’s payoff matrix in Figure 27 shows four different payoffs, denoted A
through D. These payoffs do not, however, correspond directly to the 4 combinations of
spawning times. They also depend on the environmental conditions of the early season. A
good or bad year will lead to very different payoffs for the predator and prey. In all
calculations, p is used to represent the proportion of years which have favorable early

breeding conditions (i.e. good years), while (1-p) equals the proportion of bad years.

In Figure 27, A represents the best possible outcome for predators; they successfully
spawn early and prey spawn late. In this case, all YOY predators have a size advantage
over their prey, and consume them easily. As a result, YOY predators enjoy high growth

rates and in turn, high overwinter survival.

When both species spawn simultaneously (either early or late), the success of YOY
predators is severely reduced. Without a size advantage, most prey are too large to be
captured. Assuming that environmental effects have created a distn'bution of both predator
and prey offspring sizes, only the largest predators are large enough to consume prey, and
they are restricted to capturing only the smaller prey. This combination of spawning times
is represented by B and it occurs in 2 places; when both species spawn early in favorable
years, or when both species spawn late (Fig. 27). Although the absolute spawning time

133

Figure 27: Predator payoff matrix. The two player‘s potential strategies (spawning
early or late) are crossed, giving a 2X2 matrix. Each entry represents the predator's
relative payoff resulting from a given combination of predator and prey strategies.
When predators spawn early, the resulting payoff must be weighted by p, the
proportion of good years. The letters A through D represent different potential
outcomes.

134

 

 

Prey
Early Late
Early p(B)+(1-p)D p(A)+(1-p)D
Predator
Late C B

 

 

 

Figure 27

 

135

may be an important factor in determining size at the end of a growing season, this effect is
not considered here. Both early and late spawning are assumed to be early enough to

insure adequate growth, so only the relative spawning times matter.

In Figure 27, C represents the payoff when the predator spawns late, and prey spawn
early. When this combination occurs, prey are actually larger than their predator, and are
therefore invulnerable to predation. Because of this, the success of prey reproduction is
unimportant. In either case, they are unavailable as food for YOY predators. Instead,
YOY predators are forced to utilize alternate prey, which results in lower growth rates
through their ﬁrst year (Chapter 1). Low growth rates lead to small size as ﬁsh enter

winter, which results in high overwinter mortality.

If A represents the largest beneﬁt of spawning early, D represents the largest cost;
spawning early in unfavorable years. This payoff is equivalent to near complete
reproductive failure and loss of a cohort. In Figure 26, D occurs when predators spawn
early and conditions are bad, and does not depend upon the strategy used by the prey.
Although it is likely that the payoff would be somewhat higher when prey spawn late
versus early, this difference would be negligible compared to the differences between other
payoffs, and is therefore ignored. Regardless of the prey's strategy, spawning early in

poor years gives the lowest payoff.

PREY PAYOFF MATRIX

As was the case for the predator, the payoff matrix for the prey (Fig. 27) also has 4
possible payoffs (W through Z). Once again, the payoffs represent the relative reproductive

success for a particular combination of spawning times and environmental conditions. It is

136

Figure 28: Prey payoff matrix. This matrix is set up identically to Figure 27. The
letters W through Z represent different relative payoffs for the prey.

137

 

 

Predator
Early Late
Early p(X>+(1-p)z p(W)+(1-p)z
Prey
we p<Y)+(1-p)w x

 

 

 

 

Figure 28

138

important to note that the rankings of payoffs are not identical to the predators, nor are they
opposite (i.e. it is a non-cooperative non-zero-sum game; Maynard Smith 1982). The

payoffs for different combinations are instead unique to the prey.

The highest payoff, W, occurs when there are no losses to YOY predators. This payoff
arises in 2 different situations; when prey spawn successfully before predators, or when
prey spawn after a failed reproductive attempt by predators. Interestingly, W can occur
when prey spawn early or late, depending on what the predator does and what the

environment is like. In either case, YOY prey suffer no mortality by YOY predators.

The letters X and Y in Figure 28 represent increasing amounts of mortality due to
predation. The payoff X occurs when predators and prey spawn simultaneously, either
early or late. Predation is low and is restricted to the smallest members of the cohort. Once
again, the absolute spawning time is unimportant. In either case, the same range of the
distribution is vulnerable to predators. All members of a cohort are vulnerable, however,
when prey spawn after predators have successfully spawned. YOY predators are larger
than YOY prey, and can feed upon them effectively. Therefore, mortality of YOY prey is
substantial, and is represented by Y (Fig. 28).

The payoff for prey when they spawn early and conditions are bad is the lowest payoff
considered (2 in Figure 27). As was the case for predators, spawning in poor conditions
results in heavy mortality of eggs and larvae, and thus low survivorship. The same value,
2, is used regardless of the strategy of the predator. The possible effect of the predator's
spawning strategy is assumed to be negligible relative to the differences between 2 and

other payoffs.

139

Now that the payoff matrices for both predator and prey are complete, the next step is to
determine the optimal spawning times across a gradient of environmental conditions, where
the frequency of good years (p) ranges from 0 to 1. The optimal strategy is to spawn at the
time which yields the highest payoff, for a given opponent strategy. The process of
deciding upon the optimal strategy consists of 3 steps. First, the payoffs for a given
strategy are computed for both possible strategies of the opponent. Then, the strategy of
the opponent is determined. Finally, the optimal strategy, given the opponents spawning
time, is determined. There is a risk of circularity involved when each species choice of
strategy depends upon the strategy of the opponent. However, this is not a problem here.
In all cases, at least one player's strategy is independent of its opponent's strategy. At no

time is the choice of each species' strategy simultaneously dependent on the other's.
PREDATOR STRATEGIES

If prey spawn early, predators should also spawn early when the following inequality is
satisﬁed:

p(B) + (1-p)D > c (1).

When the direction of this inequality is reversed, predators should spawn late. From
equation (1), it is clear that the optimal strategy is dependent upon p, the frequency of good
years (Fig. 29a). llf p=0 predators should spawn late and if p=1 predators should spawn
early. At some value of p between 0 and 1, the optimal strategy changes from late to early.
Setting equation (1) to an equality and rearranging shows that the critical value of p

(proportion of good years) where the optimal strategy changes is:

140

Figure 29: Relative predator payoffs across a gradient of p values. The two lines
represent the potential payoffs when prey spawn early or late. A) payoffs when
prey are spawning early. The two lines cross when p = (C-D)/(B-D). At this
point, strategies should switch from late to early. B) payoffs when prey are
spawning late. The p value where the strategy changes is p = (B-D)/(A-D).

141

 

 
     

 

 

Predator Payoffs

   

 

 

 

 

 
 
   
 

 

Predator Payoffs

   

 

 

 

A) Prey Spawn Early
B
Predators spawn early
C I
I
: x A
: Predators spawn late
D I
I
I
I
I
I
I
9.1.).
0 B-D
Proportion of Good Years (p)
B) Prey Spawn Late
A
Predators spawn early
B
I
' \
I
I
: Predators spawn late
:
D i
' I
I
0 B-D
A-D

Figure 29 ' Proportion of Good Years (p)

142

GB
p=—

B-D

The precise value of p satisfying this equation of course depends on the values of B, C,

and D (Fig. 29a).

When prey are spawning late, the decision rule for the predator to spawn early is:

P(A) + (1'13)D > B (2);

otherwise, spawn late. The decisions here are very similar to those when prey spawn early
and are summarized in Figure 29b. If p=0, spawning late results in the largest payoff, and
when p=1 predators should spawn early. The optimal spawning time changes from late to

early when:

B-D

A-D

Values for the payoffs A-D must be speciﬁed in order to calculate what proportion of
good years (p) are required for the predator's strategy to change from early to late breeding.
However, by making two simple assumptions we can visualize the general optimal strategy
for the predator. The ﬁrst assumption is that the reduction in success from C (successful
spawning but no piscivory) to D (reproductive failure) is smaller than the difference
between B (simultaneous spawning and some piscivory) and C. That is, the inability of

YOY predators to consume ﬁsh in their first year results in low growth rates and thus high

143

overwinter mortality, so C is only marginally superior to D. Thus, the difference between
these two payoffs is small relative to the difference between some YOY predators
consuming ﬁsh and none (i.e. (B-C) > (C-D)). The second assumption is that spawning
before prey (A) results in much higher success than spaWning simultaneously (B). To
represent this, A is assumed to be at least twice as successful as B (i.e. A > 2B). This
assumption is conservative because it represents the difference between conditions in which
all piscivores can consume all prey, and when only the largest piscivores can consume only

the smallest prey. In all likelihood, A will be much more than twice B.

Using these two assumptions, we can see that when p is 0.5 or larger, the payoffs for
spawning early exceed the payoffs for spawning late in both equations (1) and (2). Thus,
regardless of the spawning strategy of prey, predators should spawn early when the
frequency of good years (p) is greater than or equal to 0.5, and late when p=0.
Somewhere within the range 0 < p < 0.5 the strategy~~will switch from late to early. The
precise value of p where the strategy changes will depend upon the values of A through D
and on the prey's strategy. This can be illustrated by using a numerical example. For a
hypothetical set of relative payoffs {A=9, B=4, C=2, D=1}, we can see that if prey are
spawning early, predators should spawn early if 1 z p > 0.33, and late if 0.33 > p z 0.
Similarly, if prey are spawning late, predators should switch from early to late when
p=0.375. Considering both possible prey strategies together, predators should spawn
early, regardless of the prey's spawning time when p is larger than 0.375. And when p <
0.33, predators should spawn late for both prey strategies. Between 0.33 and 0.375,
however, the predators strategy becomes conditional upon the strategy used by the prey;
they should spawn early if prey are early, and late if prey are late. To determine which

strategies are used within this range, we must know what strategy the prey are using.

144

PREY STRATEGIES

Now that we know which predator strategies are optimal for a given value of p, the next
step is to determine the optimal strategies for prey. If predators are spawning early, then

prey should spawn early when:
1300 + (l-p)Z > p(Y) +(1-p)W (3)-

Once again, the optimal strategy changes as p increases from 0 to 1 (Fig. 30a). When
p=0, prey should spawn late. Prey should continue to spawn late when p=0.5, because
1/2(X+Z) < 1/2(Y+W). However, when p=1, the optimal time to spawn is early.
Somewhere in the range 1 > p > 0.5, the strategy switches from late to early. The critical

value of p can be found by rearranging equation (3) and solving for p as an equality as

follows:

W-Z
p = *
W+X-Y-Z.

As in the case of the predators, speciﬁc values of W-Z are needed to determine the value of

p that satisﬁes the above equation.
If predators are spawning late, prey will spawn early if:

P(W) + (11))2 > X (4);

otherwise, they will spawn late. The decision rules in this case are similar to those when

predators spawn early. In Figure 30b, we see that when p=0 prey should reproduce late,

145

Figure 30: Relative prey payoffs across a gradient of p values. The two lines
represent the potential payoffs when prey spawn early or late.A) payoffs when
predators are spawning early. Prey should switch from late to early spawning
when p = (W-Z)/(W+X-Y-Z). B) payoffs when predators are spawning late. The
critical value where the strategy changes is p = (X-Z)/(W-X). -

 

146

 

     
 
      

 

 

 

 

   
 

 

     

A) Predators Spawn Early
W
Prey spawn late
a x *
o
>5
(6
9..
>‘ I
O I
a: Y I
. I
\ .
I
Z Prey spawn early :
l
I
I
I
I
0 W—Z
W+X-Y-Z 1
Proportion of Good Years (p)
B) Predators Spawn late
W
l
Prey spawn late
I!)
““5 X .
:- I
g! I
>5 I
0 I
a: Y ;
\ .
I
I
Pre s awn earl '
Z Y P Y :
I
I
I
I
I

 

 

 

0 2g 1
W-

Figure 30 Pr0portion of Good Years (p)

147

and when p=1 prey should spawn early. From equation (4), the prey's optimal strategy

will change from late to early when:

x-z
p=—

W-Z.

Without numerical values for W, X, and Z, it is not possible to specify where within the
range of 0 < p < 1 the prey's strategy will switch from late to early spawning. However,
the range of potential environmental conditions can be narrowed if one assumes that (X-Z)
> (W-X). The payoff X occurs when predator and prey spawn at the same time, and only
small individuals are consumed by predators. This payoff is only marginally inferior to W,
where no individuals are consumed by predators. The difference between these two
payoffs is assumed to be smaller than the difference between X and 2 (When the
environment destroys almost all of the offspring). When this intuitive assumption is
considered, prey will spawn late for all values p _<_ 0.5. As was the case when predators
spawned early, if predators are spawning late, prey should switch from late spawning to

early spawning somewhere between 0.5 and 1.

These predictions can be demonstrated using a numerical example; the set {W=4, X=3,
Y=2, Z=1}. In this case, prey should spawn early if p > 0.75, and late when p < 0.75 if
predators are spawning early. If predators are spawning late, prey should shift from early
to late spawning when p=0.67. For p values between 0.67 and 0.75, the optimal strategy
is conditional upon the strategy of the predator. Below 0.67, prey should spawn late,
irrespective of the predators strategy. And when p > 0.75, prey beneﬁt most by spawning
early, for both predator spawning strategies.

148

For both predator and prey, the proportion of good years (p) where the optimal strategies
change are dependent on the strategy of the other player. Therefore, predators and prey
both have a small range of p values where their strategy is conditional on the strategy of the
other species. .But in both cases, only one player's decision is conditional, the other
players strategy is set. When p is between C-D/B-D (the critical value for predators when
prey are spawning early) and B-D/A-D (the critical value for predators when prey are
spawning late), we know that the predators strategy depends on when the prey are
spawning. Under the above assumptions, this range falls somewhere below p < 0.5. For
all values of p < 0.5, however, prey will spawn late regardless of the predators strategy.
Therefore, predators will spawaiatewhen the value of p falls between the two critical
values. As a result, when B—D/A-D > p z 0, predators will spawn late and when 1 _>_ p >
B-D/A-D, predators will spawn early. For the prey, their optimal strategy is conditional on
the predators strategy for values of p between X-Z/W-X and W-Z/W+X-Y-Z. Both of
these critical values are greater than 0.5, and when p > 0.5, predators will spawn early
irrespective of the prey's strategy. Because of this, prey will switch from late spawning to
early spawning when p = W-Z’W-t-X-Y-Z, the critical value for prey when predators are

spawning early.

Across the entire gradient of p values from 0 to 1, the strategies of predator and prey both
shift from late to early. However, the values of p at which they switch differ markedly
between the two players. Prey require the frequency of good years to be greater than
W-Z/W-t-X-Y-Z in order to spawn early. This value falls somewhere in the range of 0.5 <
p < 1. Predators, on the other hand, will switch from late spawning to early spawning
when p = B-D/A-D, which occurs within the range of 0 < p < 0.5. So, when good years
occur quite frequently (p > W-ZJW-l-X-Y-Z) both species will spawn early. The risk of

poor conditions is sufficiently low that both species do best by spawning early. For

149

intermediate values of p, predators do best when spawning early, while prey achieve
maximum success by spawning late. Finally, when good years are generally rare (p <
B-D/A-D), both species should spawn late. The frequency of bad years is too great to

make early spawning a viable strategy for either predator or prey.
DISCUSSION

According to the model presented here, the decision of whether to spawn early or late
depends upon the proportion of years that are favorable for,early spawning. If conditions
are always (or almost always) good, both predator and prey will spawn early; if conditions
are predominately poor, both species will delay reproduction. However, if there is an
intermediate probability, of poor conditions early, predators and prey will spawn
asynchronously. Predators will adopt the risky strategy. Spawning early is so vital to their
success that they will risk the reproductive failure that results from spawning in bad years.
Prey, on the other hand, will use a more conservative strategy and spawn late. The
consequences of spawning late are not as drastic for prey, and they are better off avoiding

the risk of spawning early.

While the preceeding argument makes conclusions consistent with patterns observed in
many groups of ﬁshes (see Introduction), the possibility remains that asynchronous
spawning occurs for other reasons. For example, prey may simply be incapable of
spawning earlier than they do. This could result from a lack of genetic variation, or maybe
prey can spawn early but the eggs and larvae are physiologically incapable of surviving in
the early conditions. If these constraints are important, we would expect to see clear
phylogenetic divergence of predator and prey. However, asynchronous spawning occurs

between predators and prey of the same families. In the family Centrarchidae, the black

150

bases (Micropterus spp.) spawn earlier than their sunﬁsh prey (Lapgmis spp.; Dillard and
Novinger 1975); in the family Percidae, walleye spawn earlier than their principal prey, the
yellow perch (Perca ﬂavescenS' Maloney and Johnson 1957). The close relation between

 

predator and prey in these cases suggests that phylogenetic constraints are not the sole

cause of the pattern.

Another possible explanation for late spawning in prey is that there is a lack of suitable
food for YOY prey early in the season. However, YOY piscivores often initially feed on
the same resources as their prey (Gilliam 1982, Persson 1988, Chapter 1) and they are able
to capture adequate amounts of food early in the season. In addition, there is evidence
indicating that invertebrate prey are actually more abundant early in summer rather than later
on (Mittelbach 1981, Mills and Fomey 1988, Geedey 1990). If this is true, it makes
delayed spawning in prey even more puzzling. Spawning early would not only reduce
mortality to predators, it would also enable YOY prey to take advantage of an abundant

resource base.

Early spawning by prey may also be avoided if there is no cover to protect YOY prey
from older age classes of predators at that time (Keast 1985). The presence of cover has
been found to severely affect a predators ability to capture prey (Savino and Stein 1982).
Therefore, prey might be better off waiting until macrophytes have grown enough to serve
as a refuge. In order for the presence/absence of cover to explain asynchronous spawning,
it requires differential vulnerability of YOY piscivores and prey to predation. Otherwise,
both species should spawn late. This explanation may be important in some systems,

however, it is unlikely to be the sole explanation in all cases of asynchronous spawning.

The major assumption employed by the present model is that poor weather conditions do

151

indeed lead to year class failure. Support for this comes from numerous studies which
demonstrate wide ﬂuctuations in year class strength for many species of piscivores
(Kramer and Smith 1962, Franklin and Smith 1963, Hassler 1970, Miller and Kramer
1971, Wahlberg 1972, Summerfelt 1975, Nelson and Wahlberg 1977, Aggus 1979). In
these species, little correlation between size of parental stock and year class strength exists.
Rather, variation in year class strength is better explained by low air temperatures or high
winds (which cause high turbulence) during the spawning and early growing season. Even
stronger evidence comes from studies by LeCren (1955) and Fry and Watt (1957). They
found, for eurasian perch (Perca ﬂuviatilis) and smallmouth bass respectively, that
populations in different lakes within a region showed synchronous ﬂuctuations in cohort
population size. This evidence suggests a common climatic factor is responsible. Fry and
Watt (1957) also found that when air temperatures were below average, populations in all
lakes had poor year classes. Finally, Doan (1942) found that when air temperatures around
Lake Erie were low, several species suffered poor year classes. Smith and Krefting (1953)
report similar synchrony between species in Red Lake, Minnesota. It seems unlikely that
some other factor could produce this pattern across species. Low temperatures, causing

heavy mortality in eggs and larvae, is the most likely mechanism.

All of this evidence suggests that harsh environmental conditions can cause high mortality
in YOY ﬁsh. When the early environment has these conditions at a moderate frequency,
the model predicts that predators should spawn before prey. Conversely, when there is no
uncertainty in the environment, the model predicts that predators and prey will spawn
simultaneously. Synchronous spawning by predatOrs and prey is commonly observed for
many tropical ﬁsh populations in large lakes (Lowe-McConnell 1975). In these lakes, both
predator and prey spawn continuously, so the temporal overlap in spawning time is unity.

If we assume that these tropical systems possess little uncertainty, this result is consistent

152

with the sequence predicted by the model. However, these systems may also lack
seasonality, so the model may not apply to this situation. Some tropical systems do show
pronounced seasonal ﬂuctuations in water level, effectively creating a seasonal environment
for resident ﬁsh populations (with the low water season being equivalent to winter). Fish
in these environments often are seasonal breeders (Lowe-McConnell 1975). It would be
interesting to see if there is asynchrony in predator-prey spawning times in these systems,

and if there is uncertainty in the timing or duration of high water levels.

Although this model was developed speciﬁcally to explain asynchronous spawning in
predator-prey interactions, the concept can be generalized. It can also potentially explain
the existence of asynchrony in intra- and inter-speciﬁc competitive interactions, where one
competitor reproduces before another and as a result gains a size advantage over its
competitors. In plants, many competitively inferior plants germinate before their
competitors (Grime 1977, Silvertown 1981). The inferior competitor can be thought of as
being equivalent to the predator in this model. This species can compensate for its
inferiority by starting before its competitor. Starting earlier gives this species a size
advantage, and if competitive ability scales positively with size (as it does in competition
for light; Weiner and Thomas 1986, Tilman 1988), the asymmetry of competition can be
negated or even reversed. Starting early does, however, carry the risk of reproducting
when environmental conditions are poor (V enable 1984). The inferior competitor must risk
this for any chance of success, like the predator who must be larger in order to consume
prey. On the other hand, the superior competitor may be better off by waiting until

conditions are favorable; a size advantage is not as critical for their success.

To represent asymmetric spawning/germination between competitors, the model

developed here must be modiﬁed. The payoff matrices for both competitors will be of the

153

same format as the matrix of the prey (Fig. 28). However, the magnitudes of the payoffs
W-Z will differ for the two competitors. For the superior competitor W, X, and Y will be
fairly similar in magnitude. But for the inferior competitor, W will be much larger than X
or Y, to emphasize the reduction in success when there is no size advantage. The different
payoffs will lead to different values of p where strategies change. The inferior competitor

will switch from late to early at a much smaller value of p than the superior competitor.

For predator-prey interactions in freshwater ﬁsh communities, this study concludes that
uncertainty in the early environment can result in asynchronous spawning of predator and
prey. This reﬂects differing responses to an ecological trade-off. Spawning before an
opponent gives one player a clear advantage in the biotic interactions that will occur
between the players. However, early spawning carries a risk of abiotic conditions that can
eliminate offspring before any biotic interactions occur. For the predator, the advantages of
early spawning are so critical, that they will risk harsh conditions and spawn early. The
prey, on the other hand, will use the more conservative strategy. Qualitatively, this model
is supported in the literature; asynchronous spawning is seen in many temperate ﬁsh
communities. Within these systems, many studies have reported failures in piscivore year
classes as a result of poor environmental conditions. A stronger test of the model would
involve comparisons of communities that exhibit asynchronous versus synchronous
spawning. In this case, the potential role of environmental uncertainty in structuring these

patterns could be directly addressed.

LIST OF REFERENCES

154

LIST OF REFERENCES

Adams, SM. and D.L DeAngelis. 1987. Indirect effects of early bass-shad interactions on
predator population structure and food web dynamics. pp.103-1 17. In: W.C. Kerfoot and

A. Sih. (eds), Predation: direct and indirect impacts on aquatic communities. University
Press of New England, Hanover, NH.

Aggus, LR. 1979. Effects of weather on freshwater ﬁsh predator-prey dynamics. pp.
47-56. In: H. Clepper (ed.), Predator-prey systems in ﬁsheries management. Sport Fishing
Institute, Washington, D.C

Aggus, LR. and G.V. Elliot. 1975. Effects of cover and food on year-class strength of
largemouth bass. pp. 317-322. In: R.H. Stroud and H. Clepper (eds.), Black bass biology
and management. Sport Fishing Institute, Washington, D.C.

Applegate, R.L. and J.W. Mullen. 1966. Food of young largemouth bass, Micropteggs
salmoides in a new and old reservoir. Transactions of the American Fisheries Society
96:75-77.

 

Baganel, T.B. 1978. Aspects of ﬁsh fecundity. p.75-101. In: S.D. Gerking (ed.), Ecology
of freshwater ﬁsh production. J Wiley and Sons, New York. NY.

Barry, J. P. and MJ. Tegner. 1990. Inferring demographic proceSses from size-frequency
distributions: simple models indicate speciﬁc patterns of growth and mortality. Fisheries
Bulletin 88: 13- 19.

Bennett, G.W. 1948. The bass-bluegill combination in a small artiﬁcial lake. Bulletin of the
Illinois Natural History Survey 24:378-412.

Carlander, K.D. 1977. Handbook of freshwater ﬁshery biology. Iowa State University
Press. Aames, IA.

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. Dretchmer, X. He, and C. N. Von Ende.1987. Regulation of lake
primary productivity by food web structure. Ecology 68: 1863-1876.

Clady, MD. 1976. Inﬂuence of temperature and wind on the survival of early stages of
yellow perch, Eerca ﬂavescens. Journal of the Fisheries Research Board of Canada
33:1887-1893.

Colwell, R.K. and ER. Fuentes. 1975. Experimental studies of the niche. Annual Review
of Ecology and Systematics 6:281-310.

155

Davies, W.D., W.L. Shelton, and SP. Malvestuto. 1982. Prey-dependent recruitment of
largemouth bass: a conceptual model. Fisheries 7:12-15.

DeVries, D.R, R.A. Stein, J.G. Miner, and G.G. Mittelbach. 1991. Threadﬁn shad as
supplementary forage: consequences for young-of-year ﬁshes. Transactions of the
American Fisheries Society 120:368-381.

Dillard, J .G. and GD. Novinger. 1975. Stocking largemouth bass in small
impoundments. p.459-479. In: R.H. Stroud and H. Clepper (eds.), Black bass biology
and managment. Sport Fishing Institute, Washington, D.C.

Doan, KB. 1942. Some meteorological and lirnnological conditions as factors in the
abundance of certain ﬁshes in Lake Erie. Ecological Monographs 12: 293-314.

Eipper, AW. 1975. Environmental inﬂuences on the mortality of bass embryos and larvae.
p.295-305. In: R.H. Stroud and H. Clepper (eds.), Black bass biology and managment.
Sport Fishing Institute, Washington, D.C.

Fomey, J .L. 1971. Development of dominant year classes in a yellow perch population.
Transactions of the American Fisheries Society 100: 739-749.

Franklin, DR. and L.L.Jr. Smith. 1963. Early life history of the northern pike, Esox
lucius L., with special reference to the factors inﬂuencing the numerical strength of year
classes. Transactions of the American Fisheries Society 92: 91-1 10.

Fry, F. EJ. and KB. F. Watt. 1957. Yields of year classes of the smallmouth bass hatched
in the decade of 1940 in Manitoulin Island waters. Transactions of the American Fisheries
Society 85:135-143.

Geedey, CK. 1990. The seasonal dynamics of Daphnia pulex in ﬁsh and ﬁshless ponds.
M.S. Thesis. Michigan State University, East Lansing, MI. 98 pp.

Gilliam, J .F. 1982. Habitat use and competitive bottlenecks in size-structured ﬁsh
populations. Ph.D. Dissertation. Michigan State University, East Lansing, MI. 107 pp.

Goldberg, D.E. and P.A. Werner. 1983. Equivalence of competitors in plant communities,
a null hypothesis and a ﬁeld experimental approach. American Journal of Botany 70:1098-
1 1 04.

Goldberg D.E. 1990. Components of resource competition in plant communities. p.27-49
In: 1.8. Grace and D. Tilman (eds.), Perspectives on plant competition. Academic Press.
San Diego CA.

Grime, J .P. 1979. Plant strategies and vegetation processes. John Wiley and Sons Ltd.
New York. 222 pp.

Gutreuter, SJ. and R.O. Anderson. 1985. Importance of body size to the recruitment

process in largemouth bass populations. Transactions of the American Fisheries Society
114:317-327.

156

Hackney, P.A. 1975. Bass populations in ponds and community lakes. p.131-139 In:
R.H. Stroud and H. Clepper (eds.), Black bass biology and management. Sport Fishing
Institute, Washington DC.

Hambright, KB. 1991. Experimental analysis of prey selection by largemouth bass: role
of predator mouth width and prey body depth. Transactions of the American Fisheries
Society 120: 500-508.

Hassler, TJ. 1970. Environmental inﬂuences on early development and year-class strength
of northern pike in Lakes Oahe and Sharpe, South Dakota. Transactions of the American
Fisheries Society 99: 369-375.

Heidinger R.C. 1975. Ler history of the largemouth bass. p. 11-20 In: R.H. Stroud and
H. Clepper (eds.), Black bass biology and management. Sport Fishing Institute,
Washington, DC.

Hodgson, J .R. and J .F. Kitchell. 1987. Opportunistic foraging by largemouth bass
(Micropterus salmoides). American Midland Naturalist 1 18:323-336.

Holm, S. 1979. A simple sequentially rejective multiple test procedure. Scandinavian
Journal of Statistics 6:65-70.

Hoyle, J.A. and A. Keast. 1987. The effect of prey morphology and size on handling time

in a piscivore, the largemouth bass (Micropterus salmoides). Canadian Journal of Zoology
65: 1972-1977.

Hoyle,-J.A. and A. Keast. 1988. Prey handling time in two piscivores, Esox americanus
vermiculatus and Micropterus salmoides with contrasting mouth morphologies. Canadian
Journal of Zoology 66:540-542.

 

Jones, G.P. 1987. Competitive interactions among adults and juveniles in a coral reef fish.
Ecology 68:1534-1547. ‘

Keast, A. 1977a. Diet overlaps and feeding relationships between the year classes in the
yellow perch (Perca ﬂavescens). Environmental Biology of Fishes 2:53-70.

Keast, A. 1977b. Mechanisms expanding niche width and minimizing intraspeciﬁc
competition in two centrarchid ﬁshes. Evolutionary Biology 10:333-395.

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

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

Kirkpatrick, M. 1988. The evolution of size in size-structured populations. p.13-28 In: B.
Ebenman and L. Persson (eds.), Size-structured populations: ecology and evolution.
Springer-Verlag, Berlin.

Kramer, R.H. and L.L.Jr. Smith. 1962. Formation of year classes in largemouth bass.
Transactions of the American Fisheries Society 91: 29-41.

UMwm nu MWL

hmT hP

157

Lawrence, J .M. 1958. Estimated sizes of various forage ﬁshes largemouth bass can
swallow. Proceedings of the Annual Conference of the southeast Game Fish Commision
1 1 :220-225.

LeCren, ED. 1955. Year to year variation in the year-class strength of Perca ﬂuviatilis.
Verh. Intemat. Ver. Limnol. 13: 187-192.

Lodge, D.L., J.W. Barko, D. Strayer, J.M. Melack, G.G. Mittelbach, R.W. Howarth, B.
Menge, and IE. Titus. 1988. Spatial heterogeneity and habitat interactions in lake
communities. p. 181-208. In: S.R. Carpenter (ed.), Complex interaction in lake
communities Springer-Verlag. New York.

Lowe-McConnell, R.H. 1975. Fish communities in tropical freshwaters. Longman Group
Ltd. London. 337 pp.

Madenjian, GP. and SR. Carpenter. 1991. Individual-based model for growth of

young-of-the-year walleye: a piece of the recruitment puzzle. Ecological Applications
1 :268-279.

Maloney, IE. and EH. Johnson. 1957. Life histories and inter-relationships of walleye
and yellow perch, especially during their ﬁrst summer in two Minnesota lakes.
Transactions of the American Fisheries Society 85: 191-202.

Maynard Smith, J. 1982. Evolution and the Theory of Games. Cambridge University
Press. Cambridge, MA. 224 pp.

Mayr, E. 1963. Animal species and evolution. Harvard University Press. Cambridge

McCauley, E., W.W. Murdoch, and S. Watson. 1988. Simple models and variation in
plankton densities among lakes. American Naturalist 132:383-403.

McCauley, E., W.W. Murdoch, R.M. Nisbet, and W.S. Gurney. 1991. The physiological

ecology of Daphnia: development of a model of growth and reproduction. Ecology 71:703-
715.

Miller, KD. and R.H. Kramer. 1971. Spawning and early life history of largemouth bass
(Micropterus salmoides) in Lake Powell. pp. 73-83. In: G.E. Hall (ed.), Reservoir
limnology and ﬁsheries. American Fisheries Society Special Publication. No. 8.

Miller, TE. and W.C. Kerfoot. 1987. Rediﬁning indirect effects. p. 33-37, In: W.C.
Kerfoot and A. Sih (eds.), Predation: direct and indirect impacts on aquatic communities.
University Press of New England. Hanover, NH.

Mills, EL. and J .L. Fomey, 1988. Trophic dynamics and development of freshwater
pelagic food webs. p.11-30. In: S.R. Carpenter (ed.), Complex interactions in lake
communities. Springer-Verlag. New York, NY.

Mills, E.L, J.L. Fomey, and KJ. Wagner. 1987. Fish predation and its cascading effect
on the Oneida Lake food chain. pp. 118-131. In: W.C. Kerfoot and A. Sih. (eds.),
Predation: direct and indirect impacts on aquatic communities. Univeristy Press of New
England. Hanover, NH.

158

Mittelbach, G.G. 19813. Foraging efﬁciency and body size: a study of optimal diet and
habitat use by bluegills. Ecology 62:1370-1386.

Mittelbach, G.G. 1981b. Patterns of invertebrate size and abundance in aquatic habitats.
Canadian Journal of Fisheries and Aquatic Sciences 38:896-904.

Mittelbach, G.G. 1984. Predation and resource partitioning in two sunﬁshes
(Centrarchidae). Ecology 65:499-513.

Mittelbach, G.G. 1988. Competition among refuging sunﬁshes and effects of fish density
on littoral zone invertebrates. Ecology 69:614-623.

Mittelbach, G.G. and RI. Chesson. 1987. Predation risk: indirect effects on ﬁsh
populations. p.315-332. In: W.C. Kerfoot and A. Sih (eds.), Predation: direct and indirect
impacts on aquatic communities. University Press of New England. Hanover, NH.

Mittelbach, G.G. and CW. Osenberg. 1994. Stage-structured interactions in bluegill:
consequences of adult resource variation. Ecology (in press).

Mittelbach, G.G., C.W. Osenberg, and MA. Leibold. 1988. Trophic relations and
ontogenetic niche shifts in aquatic ecosystems. p.219-235. In: B. Ebenman and L. Persson
(eds.), Size-structured populations: ecology and evolution. Springer-Verlag, Berlin.

Morin, PJ. 1983. Competitive and predatory interactions in natural and experimental
populations of Notophthalmus viridescens dorsal is and Ambystoma tigg’num. Copeia
1983:628-639.

Neill, W.E. 1988. Community responses to experimental nutrient perturbations in
oligotrophic lakes: the importance of bottlenecks in size-structured populations. p. 236-
255. In: B. Ebenman and L. Persson (eds.), Size-structured populations: ecology and
evolution. Springer-Verlag, Berlin.

 

Nelson, W.R. and CH. Wahlberg. 1977. Population dynamics of yellow perch (Perca
ﬂavescens), sauger (Stizostedion canadense), and walleye (S. vitreum vitreum) in four
main stem Missouri River reservoirs. Journal of the Fisheries Research Board of Canada
34:1748-1763.

 

Nisbet, R.M, W.S. Gurney, W.W. Murdoch, and E. McCauley. 1989. Structured
population models: a tool for linking effeCts at individual and population levels. Biological
Journal of the Linnean Society 37:79-99.

Oksanen, L., S.D. Fretwell, J. Arruda and P. Niemela. 1981. Exploitation ecosystems in
gradients of primary productivity. American Naturalist 118:240-261.

Oliver, T.D., G.F. Holeton, and KB. Chua. 1979. Overwinter mortality of ﬁngerling
smallmouth bass in relation to size, relative energy stores, and environmental temperature.
Transactions of the American Fisheries Society 108:130-136.

Osenberg, C.W., Mittelbach, G.G. and P.C. Wainwright. 1992. Two-stage life histories
in ﬁsh: the interaction between juvenile comeptition and adult performance. Ecology 73:255-
267.

159

Osenberg, C.W., M.H. Olson, and G.G. Mittelbach. 1993. Stage structure in ﬁshes:
resource productivity and competition gradients. In: DJ. Stouder, KL. Fresh and RJ.
Feller (eds.), Theory and application of ﬁsh feeding ecology. University of South Carolina
Press. Columbia SC (In press).

Osenberg, C.W., E.E. Werner, G.G. Mittelbach, and DJ. Hall. 1988. Growth patterns in

bluegilll (Lemmis macrochirus) and pumpkinseed (_L_.. gibbosus) sunfish: environmental
variation and the importance of ontogenetic niche shifts. Canadian Journal of Fisheries and

Aquatc Sciences 45:17-26.

Paloheimo, J.E. and LM. Dickie. 1966. Food and growth in ﬁshes 11: effects of food and
temperature on the relation between metabolism and body weight. Journal of the Fisheries
Research Board of Canada 23:869-908.

Persson, L. 1987. Effects of habitat and season on comeptitive interactions between roach
(Rutilus rutilus) and perch (Perca ﬂuviatilus). Oecologia 73:170-177.

 

 

Persson, L. 1988. Asymmetries in competitive and predatory interactions in ﬁsh
populations. p. 203-218. In: .B. Ebenman and L. Persson (eds.), Size-structured
populations: ecology and evolution. Springer-Verlag. Berlin.

Persson, L. 1990. Predicting ontogenetic niche shifts in the ﬁeld: what can be gained by
foraging theory? p.303-321. In: R.N. Hughes (ed.), Behavioural mechanisms of food
selection. Springer-Verlag, Berlin.

Persson, L, G. Andersson, S.F. Hamrin, and L. Johansson. 1988. Predator regulation
and primary production along the productivity gradient in temperate lake ecosystems. p. 45-

68. In: S. Carpenter (ed.) Complex interaction in lake communities. Springer-Verlag New
York, NY.

Persson, L., S. Diehl, L. Johansson, G. Anderson, and SF. Hamrin. 1992. Trophic
interactions in temperate lake ecosystems: a test of food chaing theory. American Naturalist
140:59-84.

Persson, L. and LA. Greenburg. 1990. Competitive juvenile bottlenecks: the perch (Perca
ﬁuviatilis)—roach (Rutilus rutilus) interaction. Ecology 71:44-56.

Peters, R. H. 1983. The ecological implications of body size. Cambridge University Press,
Cambridge. 329 p.

Pianka, ER. 1973. The structure of lizard communities. Annual Review of Ecology and
Systematics 4:53-74.

Polis, G.A. 1988. Exploitation competition and the evolution of interference, cannibalism,
and intraguild predation in age/size structured populations. p. 185-202. In: B. Ebenman
and L. Persson (eds.), Size-structured populations: ecology and evolution. Springer-
Verlag. Berlin.

Polis, GA. and SJ. McCormick. 1986. Patterns of resource use and age structure among
species of desert scorpion. Journal of Animal Ecology 55:59-73.

160

Polis, G.A. and S.J.McCormick. 1987. Intraguild predation and competition among desert
scorpions. Ecology 68:332-343.

Polis, G.A., C.A. Myers, and RD. Holt. 1989. The ecology and evolution of intraguild
predation: potential competitors that each each other. Annual Review of Ecology and
Systematics 20:297-330.

Prout, T. 1986. The delayed effect on fertility of preadult competition: two-species
population dynamics. American Naturalist 127:809-818.

Rice, W.R. 1989. Analyzing tables of statistical tests. Evolution 43:223-225.

Rice, J.A., T.J. Miller, K.A. Rose, L.B. Crowder, E.A. Marschall, A.S. Trebitz, and
D.L. DeAngelis. 1993. Growth rate variation and larval survival: inferences from an
individual-based size-dependent predation model. Canadian Journal of Fisheries and
Aquatic Sciences 50:133-142.

Ricker, W.B. 1979. Growth rates and models.p.677-743. In:W.S. Hoar, D.J. Randall,
and J.R. Brett (eds.), Fish Physiology.Vol.VIII. Academic Press. New York, NY.

Savino, J.F. and RA. Stein. 1982. Predator-prey interaction between largemouth bass and
bluegills as inﬂuenced by simulate, submersed vegetation. Transactions of the American
Fisheries Society 1 1 1: 255-266.

Schoener, T.W. 1975. Presence and absence of habitat shift in some widespread lizard
species. Ecological Monographs 45:233-258. '

Scott, W.B. and EJ. Crossman, 1979. Freshwater ﬁshes of Canada. Fisheries Research
Board of Canada Bulletin 184. ,

Shelton, W.L., W.D. Davies, T.D. King, and T.J. Timmons. 1979. Variation in the
growth of the initial year class of largemouth bass in West Point Reservoir, Alabama and
Georgia. Transactions of the American Fisheries Society 108:142-149.

Silvertown, J.W. 1981. Seed size, life span, and germination date as coadapted features of
plant life history. American Naturalist 118:860-864.

Smith, L.L. Jr. and LW. Krefting. 1953. Fluctuations in production and abundance of
commercial species in the Red Lakes, Minnesota, with special reference to the walleye
population. Transactions of the American Fisheries Society 83:131-160.

Stein, R.A., S.T. Threlkeld, C.D. Sandgren, W.G. Sprules, L. Persson, E.E. Werner,
W.B. Neill, and SJ. Dodson. 1988. Size-structured interactions in lake communities.
p.161-180 In: S.R. Carpenter (ed.), Complex interactions in lake communities. Springer-
Verlag. New York, NY.

Summerfelt, R.C. 1975. Relationship between weather and year-class strength of
largemouth bass. p. 166-174. In: R.H. Stroud and H. Clepper (eds.), Black bass biology
and managment. Sport Fishing Institute. Washington, D.C.

161

Swingle, HS. and E.V. Smith. 1940. Experiments on the stocking of ﬁsh ponds.
Transactions of the Northa American Wildlife Conference 5:267-276.

Taylor, BE. 1980. Size-selective predation on zooplankton. p.377-387. In: W.C. Kerfoot
(ed.), Evolution and ecology of zooplankton communities. University Press of New
England. Hanover, NH.

Tesch, F.W. 1968. Age and Growth. p.93-123. In: W.B. Ricker (ed.), Methods for
assessment of ﬁsh production in fresh. waters. Blackwell Scientiﬁc Publishers. Oxford.

Tilman, D. 1982. Resource competition and community structure. Monographs in
Population Biology No. 17. Princeton University Press. Princeton, NJ. 296 pp.

Tilman, D. 1988. Plant strategies and the dynamics and structure of plant communities.
Monographs in Population Biology No.26. Princeton University Press. Princeton, NJ.
360 pp.

Timmons, T.J., W.L. Shelton, and W.D. Davies. 1980. Differential growth of largemouth
bass in West Point Reservoir, Alabama-Georgia. Transactions of the American Fisheries
Society 109:176-186.

Tonn, W.M and CA. Paszkowski. 1986. Size-limited predation, winterkill, and the
organization of Umbra-Perca assemblages. Canadian Journal of Fisheries and Aquatic
Sciences 43:194-202.

Tonn, W.M, C.A. Paszkowski, and T.C. Moermond. 1986. Competition in Umbra-Perca
assemblages: experimental and ﬁeld evidence. Oecologia 69:126-133.

Tschumy, W.O. 1982. Competition between juveniles and adults in ate-structured
populations. Theoretical Population Biology 21:255-268.

Venable, D.L. 1984. Using intraspeciﬁc variation to study the ecological signiﬁcance and
evolution of plant life-histories. pp. 166-187. In: R. Dirzo and J. Sarukhan. (eds.),
. Perspectives in plant population ecology. Sinauer Associates. Sunderland, MA.

Wahlberg, CH. 1972. Some factors associated with ﬂuctuation in year-class strength of
Sanger, Lewis and Clark Lake, South Dakota. Transactions of the American Fisheries
Society 101: 311-316.

Wainwright, P.C. 1988. Morphology and ecology: functional basis of feeding constraints
in Caribbean labrid ﬁshes. Ecology 69:635-645.

Wainwright, P.C., C.W. Osenberg, and G.G. Mittelbach. 1991. Trophic polymorphism in
the pumpkinseed sunﬁsh m gibbosus Linnaeus): effects of environment on
ontogeny. Functional Ecology 5:40-55.

Weiner, J. and SC. Thomas. 1986. Size variability and competition in plant monocultures.
Oikos 47:211-222.

Werner, E.E. 1977. Species packing and the niche complementarity in three sunﬁshes.
American Naturalist 111:553-578.

162

Werner, E.E. 1986. Species interactions in freshwater ﬁsh communities. p.344-358 In: J.
Diamond and T.J. Case (eds.), Community Ecology. Harper and Row Publishers. New
York NY.

Werner, E.E., 1988. Size, scaling, and the evolution of complex life cycles. p. 60-81. In:
B. Ebenman and L. Persson (eds.), Size-structured populations: ecology and evolution.
Springer-Verlag, Berlin.

Werner, E.E. and J .F. Gilliam. 1984. The ontogenetic niche and species interactions in
size-structured populations. Annual Review of Ecology and Systematics 15:393-426.

Werner, E.E., J.F. Gilliam, D.J. Hall, and G.G. Mittelbach, 1983a. An experimental test
of the effects of predation risk on habitat use in ﬁsh. Ecology 62:1540-1548.

Werner, E.E. and DJ. Hall. 1977. Competition and habitat shifts in two sunﬁshes
(Centrarchidae). Ecology 58:869-876.

Werner E.E. and DJ. Hall. 1979. Foraging efﬁciency and habitat switching in competing
sunﬁshes. Ecology 60:256-264.

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

Werner, E.E., D]. Hall, D.R. Laughlin, D.J. Wagner, L.A. Wilsmann and P.C. Funk.
1977. Habitat partitioning in a freshwater ﬁsh community. Journal of the Fisheries
Research Board of Canada 34:360-370.

Werner, E.E., G.G. Mittelbach, D.J. Hall, and J.F. Gilliam. 1983b. Experimental tests of
optimal habitat use in ﬁsh: the role of relative habitat proﬁtability. Ecology 64:1525-1539

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

Wilbur, HM. 1980. Complex life cycles. Annual Review of Ecology and Systematics
1 1:67-93.

Wilbur, H. M. 1988. Interactions between growing predators and growing prey. p. 157-
172 In: B. Ebenman and L. Persson (eds), Size-structured populations: ecology and
evolution. Springer-Verlag, Berlin.

Wilbur, H. M. and J. P. Collins. 1973. Ecological aspects of amphibian metamorphosis.
Science 182. 1305- 1314.

Wilson, D.S. 1975. The adequacy of body size as a niche difference. American Naturalist
109:769-784.

Winemiller, K0. and D.H. Taylor. 1987. Predatory behavior and competition among
laboratory-housed largemouth and smallmouth bass. American Midland Naturalist 117:148-
166.

“llllilllliiilllll