a.
‘x;

3.... .5 P
$m®i fa.
.. :..r‘l¢Mf
132‘ .11! I! s.
3."..3... .5”: .5
5. 3.4.14.1“.

. is}. ..
%. tniini
15...»..1

95...? .
1.3

3‘!
t}:
n :

 

ivy...
1.1..

A, .(vdu
3.1 I} 1...
stain-7.31.

I19 q.
II! P‘ ‘ «I... .1
‘9? till"

Liv .S‘tuLI.

 

 

 

 

.i,

....;\ Crag“. .. ...;.H.w.k. 5wa

 

 

 

 

LIBRARY
Michigan State

 

Universnty

 

 

This is to certify that the
thesis entitled

ENVIRONMENTAL DEPENDENCE OF NON-
CONSUMPTIVE EFFECTS IN PREDATOR-PREY
INTERACTIONS

presented by

Katrina A. Button

has been accepted towards fulfillment
of the requirements for the

degree In Fisheries and Wildlife

 

:Sazf)

Major‘> rofessor’ Signature
4/? 30 OE“
/0

ate

 

MSU is an aflinnative-action, equal-opportunity employer

—‘— .n-l-l-l-I-I-l-I_-‘_-_L_-U-l-U-I-I‘1-|-I-—-. ~

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/07 p:/CIRC/DaleDue.indd-p.1

 

 

 

ENVIRONMENTAL DEPENDENCE OF NON-CONSUMPTIVE EFFECTS IN
PREDATOR-PREY INTERACTIONS

By

Katrina A. Button

A THESIS

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

MASTER OF SCIENCE
Department of Fisheries and Wildlife

2008

 

ABSTRACT

ENVIRONMENTAL DEPENDENCE OF NON-CONSUMPTIVE EFFECTS IN
PREDATOR-PREY INTERACTIONS

By
Katrina A. Button
Predators affect the population dynamics of their prey in two ways, through direct
consumption (consumptive effect) or through inducing trait changes (non-consumptive
effect). Studies have shown that the trait changes induced by predators reduce fitness
enough to make up a large portion of the total effect of the predator. However, these
studies were conducted under a finite set of natural conditions. Little is known about
how variation in biotic and abiotic parameters affects trait expression or the importance
of non-consumptive effects. I investigated the importance of the non-consumptive effect
of Bythotrephes longimanus (predator) on Daphnia mendotae (prey) in terms of
population growth rate (instantaneous measure) and population dynamics (longer-term),
using three modeling techniques. D. mendotae respond to B. longimanus, a visual
predator, by migrating down in the water column into cooler, darker water to avoid
predation, however, this behavior incurs a reproductive cost associated with the cooler
water. I found that the non-consumptive effects are important over a large range of
conditions. Interestingly, changes in water temperature and B. Iongimanus density had a
large non-linear impact on the importance of non-consumptive effects due to changes in
vertical migration behavior. Understanding how B. longimanus effects change seasonally
will provide better predictions of impacts in the Great Lakes and the many other systems
which it has invaded. This approach is also key to predicting species effects in different

scenarios, such as those predicted with Global Warming.

 

To Mom and Dad

iii

 

ACKNOWLEDGEMENTS

I would like to recognize all of the people who have helped make my graduate
school experience fun, fiilfilling and ultimately successful.

I would especially like to thank my advisor, Scott Peacor, and my lab mates
Kevin Pangle and Andrea Jaeger-Miehls. I would also like to thank my committee
members, Andrew McAdam and Charles Ofria. The faculty and graduate students in the
Department of Fisheries and Wildlife, the Ecology, Evolutionary Biology and Behavior
program and at Kellogg Biological Station are a wonderful group of people. I have really
enjoyed my interactions with them over the past few years and will miss their company
and conversation.

I would also like to thank my family for all the support and encouragement they

have given me.

iv

 

TABLE OF CONTENTS

 

 

 

LIST OF TABLES vii
LIST OF FIGURES viii
CHAPTER 1: INTRODUCTION 1
State of the field ............................................................................................................... 2
Importance of non-consumptive effects across systems .......................................... 2
Importance of non-consumptive effects across environmental gradients ................ 4
Needs for further study ............................................................................................. 6
Thesis research ................................................................................................................ 6

CHAPTER 2: ENVIRONMENTAL FACTORS IMPACT THE RELATIVE
IMPORTANCE OF CONSUMPTIVE AND NON-CONSUMPTIVE EFFECTS IN

 

 

LAKE MICHIGAN ZOOPLANKTON 9
Abstract ............................................................................................................................ 9
Introduction ................................................................................................................... 10
Methods ......................................................................................................................... 14

Description of System ............................................................................................ 14
Overview of modeling techniques .......................................................................... 15
Adaptive Trait Expression ...................................................................................... 16
Net Effect, Consumptive Effect and Non-consumptive Effect .............................. 19
Parameterizing the models ..................................................................................... 20
Examining the effect of the Environment .............................................................. 26
Results ........................................................................................................................... 27
Is vertical migration adaptive? ............................................................................... 28
Impact of environmental variability ....................................................................... 36
Discussion ...................................................................................................................... 35
Tables ............................................................................................................................. 41
Figures ....................................................................................................... ' .................... 4 3

CHAPTER 3: SEASONAL CHANGES IN THE IMPORTANCE OF NON-

CONSUMPTIVE EFFECTS IN LAKE MICHIGAN ZOOPLANKTON 47
Abstract .......................................... . ............................................................................... 47
Introduction ................................................................................................................... 48
Methods ......................................................................................................................... 52

Description of System ............................................................................................ 52
D. mendotae Population Dynamics Model ............................................................. 53
Describing the different steps of the model ............................................................ 56
Determining how B. longimanus predation affects D. mendotae population
dynamics metrics .................................................................................................... 60
Results ........................................................................................................................... 61
Discussion ...................................................................................................................... 65
Figures ........................................................................................................................... 69
CHAPTER 4: SUMMARY AND IMPLICATIONS 76

 

 

Summary of thesis ......................................... ‘ ................................................................ 76

Motivation for thesis research ................................................................................ 76

The effect of environmental variation .................................................................... 77
Translating fitness correlate results for seasonal population dynamics ................. 78
Needs for further study ........................................................................................... 79
Implications of thesis work ............................................................................................ 81
LITERATURE CITED _ 92

 

vi

 

LIST OF TABLES

Table 2.1 A summary of the Environmental Parameters included in the population
growth rate model including Parameter, Parameter description, field or lab data range and
parameter space investigated 41

 

Table 2.2 A description of additional predation curves added to the Population Growth
Rate model, their biological rationale and summary of the results 42

vii

 

LIST OF FIGURES

Figure 2.1 Comparing model predictions and field observations 43

 

Figure 2.2 Including a swimming speed constraint for D. mendotae makes anticipating
day-break necessary. 44

 

Figure 2.3 The effect of variation in surface temperature on the composition of the
predator effect. 45

 

Figure 2.4 The effect of B. Iongimanus density on overall predator effect. 46

 

Figure 3.1 Flow diagram for Population Dynamic Model. 69

Figure 3.2 D. mendotae population dynamics over a 5 month season from July to
November with fixed B. Iongimanus population based on field data. 70

 

Figure 3.3 Breaking down the net effect of the predator into the proportion due to non-
consumptive effects and the proportion due to consumptive effects. 71

 

Figure 3.4 D. mendotae population dynamics over a 5 month season from July to
November with a responsive, slow growing, B. longimanus population. , 72

 

Figure 3.5 D. mendotae population dynamics over a 5 month season from July to
November with a responsive, fast growing, B. Iongimanus population. 73

 

Figure 3.6 Comparing consumptive and non-consumptive effects for present and future
climate over a season. 74

 

Figure 3.7 Comparing the non-consumptive effect for future climate conditions over a
season with dynamic B. longimanus population densities or forced natural densities ....... 75

viii

 

CHAPTER 1
INTRODUCTION

One main area of study within ecology is the interactions between populations of
organisms and their environments. Food-webs are a useful model for studying the
interactions of populations in ecological communities. Food-webs are graphical
representations of the populations in a community and some of the interactions between
these populations, in particular, who-eats-whom. Food-webs are built from knowledge of
predator-prey interactions. They provide information on how energy flows through a
community, as well as information on basic community structure that many scientists use
to better understand ecological communities. However, our basic building blocks for
food-webs, predator-prey interactions are not entirely understood. There is still much
that is not known about how predator induced changes in prey traits might impact prey
populations.

Traditionally it has been thought that the only important effect predators have on
their prey is through direct consumption. The evidence that predator induced changes in
prey traits have important fitness consequences for prey populations has been growing.
Examples of prey modifying traits in response to predator risk are ubiquitous in predator-
prey relationships (reviewed in Lima and Dill 1990, Lima 1998, Tollrain and Harvell
1999). These trait changes reduce direct consumption, but also reduce the fitness of the
prey in other ways (Harvell 1990). Prey face a tradeoff between survival and
reproduction in trying to limit the overall effect of a predator: avoid being eaten or
produce as many offspring as possible. When prey modify traits (behavioral,

morphological, physiological or life-history) in order to simultaneously minimize

 

predation and maximize fitness, the prey population incurs some direct mortality due to
predation and some cost to fitness through decreased growth or fecundity. Thus, the total
effect of the predator on the prey population can be broken down into multiple
components including: the effect due to direct consumption, which is termed the
consumptive effect (also density-mediated or lethal effect); and the effect of changes in
prey traits on reproduction, which is termed the non-consumptive effect (also trait-
mediated or non-lethal effect). Many studies have shown that non-consumptive effects
can be very important (reviewed in Peacor and Werner 2004a, Preisser at al. 2005).
Little research has been done to determine how contributions of consumptive
effects and non-consumptive effects to predator-prey interactions might vary with
changes in the environment. We expect individual trait expression to vary with
environmental parameters and if the cost associated with these trait changes is not
constant, we would also expect the magnitude of non-consumptive effects to vary with
environmental conditions. The goal of this thesis is to explore how trait expression and
the magnitude and importance of non-consumptive effects changes with variation in

environmental conditions.

State of the field

Importance of non-consumptive effects across systems

Theoretical studies (Abrams 1982, 1984, 1993, Ives & Dobson 1987, Bolker et a1.
2003, Peacor & Werner 2004b) and empirical studies (reviewed in Peacor and Werner
2004a, Preisser et a1. 2005) have shown that non-consumptive effects are an important
component of the net effect of the predator in a number of systems. It is not surprising

that increasingly, researchers are finding non-consumptive effects to be important in a

 

wide range of systems given the number of systems where predators induce trait changes
in their prey (reviewed in Lima and Dill 1990, Lima 1998, Tollrain and Harvell 1999,
Agrawal 2001). These changes in trait expression often have an associated cost in terms
of growth and fecundity (Harvell 1990) resulting in large non-consumptive effects
(reviewed in Peacor and Werner 2004a and Preisser et al. 2005).

The number of systems where researchers have found large non-consumptive

effects of predators on prey fitness surrogates (individual growth, fecundity, population 3
growth, etc.) is growing each year. Recently, several major reviews have been published

that summarize how prey trait changes induced by predators effect prey populations l:

(Peacor and Werner 2004a, Preisser et al. 2005, Preisser et al. 2007, Cresswell 2008) and :

 

propagate through the community to affect populations of other organisms (Werner and
Peacor 2003, Bolker et al. 2003). The majority of these studies have been done in aquatic
systems, but there is growing emphasis on terrestrial systems. In a meta-analysis of 49
published studies, Preisser et a1. (2005) found a similar strength in consumptive and non-
consumptive effects, with particularly strong and sometimes non—intuitive effects
documented in aquatic systems. For example, Peacor (2002) found that dragonfly larvae
caused a net increase in small bull frog growth through a behavioral response (shift in
foraging level). The majority of the terrestrial studies to date have been done on
invertebrates. For example, Nelson (2004) found that surgically altered damsel bugs
cause a large non-consumptive effect (30% reduction) on pea aphid population growth.
There have been few studies on terrestrial vertebrates. Some predator exclusion
experiments have been done on mammals and have identified non-consumptive effects of

predators on prey body condition or growth rates. For example, Karels et al (2000) found

that various predators caused non-consumptive effects on arctic ground squirrels body
weight at parturition and litter size. Birds have complicated anti-predator behaviors
(Caro 2005), but few studies of the costs of this behavior have documented both
consumptive and non-consumptive effects. Thus, the relative importance of non-
consumptive effects is not well understood for birds (Cresswell 2008).

The majority of the studies mentioned above have focused on quantifying non-
consumptive effects in terms of individual grth (Peacor and Werner 2004a). Recently,
researchers have begun attempting to quantify the reproductive cost of modifying traits in
terms of population-level responses partially in an effort to quantify non-consumptive

effects and consumptive effects in a common currency. The systems where this has been

 

9’, '4.-

done extend from aquatic zooplankton (Boeing et al. 2005, Pangle et al. 2007) to stream
macroinvertebrates (Peckarsky 1993, McPeek and Peckarsky 1998) to insects in
agricultural systems (Nelson 2004).
Importance of non-consumptive effects across environmental gradients

The studies mentioned above only give us an understanding of the importance of
non-consumptive effects under very specific conditions, a snapshot of the possible biotic
and abiotic environmental conditions. We expect plastic traits to change as a result of
changes in the environment. If the cost of trait expression is not constant, we would also
expect the relative contribution of consumptive and non-consumptive effects to vary
across environmental conditions as well. However, few studies have attempted to
determine how interactions with the environment and with other species might influence
trait expression and the relative importance of non-consumptive effects. A few studies
have investigated how predator induced trait changes are influenced by the environment

(biotic and abiotic), focusing on exploring the impacts of resource levels (Werner and

Anholt 1996, Fiksen et al 1997, Boscarino et al. 2007), temperature gradients (Boscarino
et al. 2007) and density and types of predators (Werner and Anholt 1996, F iksen, 2006,
Boscarino et al. 2007). Boscarino found that the vertical distribution of Mysis in the
water column depended on food availability, temperature gradient and the level of
predator kairomones.

These documented changes in traits suggest we should find that the environment
affects the relative contributions of consumptive and non-consumptive effects. Initial
studies that have looked for an impact of interactions with other organisms on the relative
contributions of consumptive and non-consumptive effects have focused on examining

the effect of resource levels (Eklov and Halvarsson 2000, Peacor and Werner 2004b,

 

Turner 2004,) competition (Peacor and Werner 2000, 2004b, 2006, Turner 2004) and
predator type (Eklov and Werner 2000, Preisser 2007). These studies found that non-
consumptive effects are dependent on resource levels and interactions with other
organisms.

These initial studies have shown that changes in the enviromnent (in terms of
biotic interactions) can influence the relative contributions of consumptive and non-
consumptive effects. However, we as ecologists still do not have a good understanding
of how abiotic environmental conditions might affect trait expression or the importance
of non-consumptive effects. More work is needed to understand the breadth of
conditions over which trait expression changes and non-consumptive effects are
important. In order to understand the generality of the importance of including plastic

trait expression in our theoretical constructs of ecological communities, we need to

investigate how non-consumptive effects and consumptive effects depend on
environmental variation.
Needs for further study

Improving our understanding of the generality of the importance of trait-mediated
interactions will allow us to determine how influential they may be in community
dynamics. We still lack a good understanding of how the importance of non-
consumptive effects might change due to environmental variation or in scaling to long-
term situations. Bolker et al. 2003 identified scaling from short-term to long-terrn

responses of communities as one of “the most critical needs” to determining whether

 

trait-mediated interactions affect community dynamics at scales that will influence
practical management decisions. I attempt to fill this gap in our knowledge by addressing
the following questions:

1. Do changes in environmental conditions affect the importance of trait-
mediated interactions? Would the ratio of the trait effect to the net effect of the predator
change with variation in different environmental conditions (perhaps in a different field,
stream or pond or with changes associated with establishment of invasive species, climate
change or extreme weather)?

2. How might the importance of non-consumptive effects vary over longer time
frames that incorporate variation in environmental conditions?

Thesis research

In this thesis, I use multiple modeling techniques to examine one system

extensively, determining the importance of non-consumptive effects under a wide range

of environmental conditions. In the second chapter, I determined how environmental

conditions influence adaptive behavior and the relative contributions of consumptive and
non-consumptive effects, which I measured in terms of prey population growth rate. In
the third chapter, I determined how findings of non-consumptive effects on a fitness
correlate (population growth rate) translate to longer—term, multi- generational population
dynamics. In the final chapter I review my main findings and put them in context of the

literature, discussing what data would be necessary to repeat a similar analysis in other

'1'!"

systems. I also reviewed the literature to determine if this analysis is currently possible
with the available data. The second and third chapters are being prepared for submission

with coauthors Scott Peacor and Kevin Pangle.

 

The second chapter uses modeling techniques to determine adaptive behavior and
measure consumptive and non-consumptive effects on a prey fitness correlate, population
growth rate. The system I modeled includes an invasive invertebrate predator,
Bythotrephes Iongimanus that feeds on zooplankton (small aquatic grazers), in particular
Daphnia mendotae. This is a well studied system where large non-consumptive effects
have been documented (Pangle et al. 2007). Specifically I modeled the decrease in per
capita population growth rate of D. mendotae due to predation by B. longimanus over a
range of environmental conditions to examine the generality of the importance of non-
consumptive effects. I used two modeling techniques: an optimality model in which I
maximized a differential equation for per-capita population growth rate, and a state-
dependent dynamic optimality model in which I optimized lifetime reproductive success
to determine if the behavioral response of D. mendotae to B. Iongimanus is adaptive and
to estimate the magnitude and the importance of non-consumptive effects. Using the

optimality model and state-dependent dynamic optimality model, I determined whether

the behavioral response of D. mendotae and the instantaneous importance of non-
consumptive effects were general over a range of biotic and abiotic conditions: B.
longimanus density, D. mendotae density, B. longimanus distribution, D. mendotae
distribution, predation risk from additional predators (including size selective predators),
competition, surface light availability, light attenuation and temperature.

The third chapter extends these instantaneous results to the impact of D. mendotae
trait changes due to B. Iongimanus predation on the population dynamics of D. mendotae l
over an entire season. I used a stochastic simulation model, to understand changes in
adaptive behavior and to estimate the magnitude and the importance of the trait effect
over the course of an entire season. The stochastic simulation model allowed the
extension of these instantaneous results into long-term multi-generational dynamics
accounting for seasonal variation in abiotic conditions. I was also able to investigate how
changes in abiotic and biotic conditions due to climate change and exotic invasions might
influence population dynamics.

The fourth chapter discusses my findings in chapters 2 and 3 and puts them in
context of the literature. I also explored whether these results might be expected to
extend to other systems. I explored other systems for which researchers have tried to
quantify non-consumptive effects. For these systems, I explored the available data for
how consumptive and non-consumptive effects might change with biotic and abiotic
parameters. I was unable to find any other systems for which sufficient data existed to

repeat a similar analysis.

CHAPTER 2
ENVIRONMENTAL FACTORS IMPACT THE RELATIVE
IMPORTANCE OF CONSUMPTIVE AND NON-CONSUMPTIVE
EFFECTS IN LAKE MICHIGAN ZOOPLANKTON

Abstract

Daphnia mendotae migrate vertically into deeper, cooler regions of Lake Michigan to
avoid predation by an invasive planktivore, Bythotrephes longimanus. Since B. .-
Iongimanus is a visual predator, D. mendotae vertical migration reduces direct

consumption of D. mendotae by B. longimanus (consumptive effect), however, inhabiting

 

the colder and less productive regions of the water column has a reproductive cost for D.
mendotae (non-consumptive effect). These types of tradeoffs are ubiquitous to predator-
prey interactions, and studies of various systems have shown that non-consumptive
effects can be important relative to consumptive effects. However, these studies are
snapshots, encompassing a small range of the natural variation in biotic and abiotic
conditions that might affect prey trait expression and ultimately the importance of non-
consumptive effects. Here I evaluated how seasonal changes in the pelagic environment
can influence B. Iongimanus's density and non-consumptive effects on D. mendotae using
differential equation and dynamic optimization models. Both models were parameterized
with data from Lake Michigan and lab experiments. I found that predicted adaptive
behavior for D. mendotae varies in response to changes in environmental conditions and
that some environmental factors can strongly affect the importance of non-consumptive
effects. For example, I found that, both, water temperature and B. longimanus density
had a large non-linear impact on the importance of non-consumptive effects. Under

certain conditions, the models predict that D. mendotae should remain at the surface.

There are threshold values for surface temperature and B. longimanus densities at which
point behavior shifts drastically and non-consumptive effects change from being non-
existent (no vertical migration) to being very important (vertical migration). An
understanding of how B. longimanus effects change seasonally will provide better
predictions of its impact in the Great Lakes and the many other systems which it has

. invaded. This approach is also key to predicting species effects in different scenarios,

such as those predicted with Global Warming.

Introduction

_-_ _._ _——_'T

Predation can have a large impact on prey populations through direct
consumption, but also through a suite of phenotypic changes that they induce in
individual prey. For many years it was thought that predators affect prey populations
primarily through consumptive effects (also lethal or density effects). Ecologists have
been aware for some time that prey respond to predation risk by modifying their
phenotype to minimize the effect of the predator. It is only recently that ecologists have
begun to understand how important a role these phenotypic modifications play in
predator prey interactions. In modifying their traits, prey shift the effect of predators on
prey fitness from direct mortality to other sources, minimizing the total effect to the
predator on prey fitness. Prey incur costs in terms of fitness surrogates such as individual
growth, reproduction or population growth rates, this cost is termed the non-consumptive
effect (also non-lethal or trait-mediated effect). Prey change their traits in a number of
ways in response to predators, including behavioral, morphological and physiological
adaptations that make them less likely to be consumed (Lima 1998). These trait changes

reduce direct consumption, but can also reduce the reproductive success of the prey

10

(Harvell 1990). If prey forage less or put resources into growing larger in order to

reproduce earlier or into growing defensive features, they have less resources for growth
and reproduction or may face mortality from other sources. Thus, prey face a tradeoff in
trying to limit the overall effect of a predator: avoid being eaten while maximizing

fecundity. Studies have shown that predators induce trait changes in their prey in a large
number of systems (reviewed in Dill 1987, Lima and Dill 1990, Lima 1998) and in many
of these systems, these trait changes result in large non-consumptive effects (reviewed in

Peacor and Werner 2004a, Preisser et al. 2005).

 

There is reason to believe that the importance of non-consumptive effects will be 2
affected by abiotic environmental parameters and interactions with other organisms. An
organisms’ phenotype, or level of trait expression is a product of their environment and is
affected not only by predation risk, but by environmental parameters and interactions
with other species. We know that differing levels of trait expression amount to differing
costs in terms of reproduction and that the relationship between trait expression and costs
may change with changes in the environment. Thus, it follows that the magnitude of
non-consumptive effects might be influenced by environmental parameters and
interactions with other species as well. However, very few studies have addressed the
question of how levels of trait expression and/or the magnitude of non-consumptive
effects are affected by the environment or by interactions with other species. A few
studies have investigated how predator induced trait changes are influenced by the
environment (biotic and abiotic), focusing on exploring the impacts of resource levels
(Werner and Anholt 1996, Fiksen et al 1997, Boscarino et al. 2007), temperature

gradients (Boscarino et al. 2007) and density and types Of predators (Werner and Anholt

ll

1996, Fiksen, 2006, Boscarino et al. 2007) on trait expression. For example, Boscarino
found that the vertical distribution Of Mysis in the water column depended on food
availability, temperature gradient and the level of predator kairomones.

Studies that explore the importance of environmental factors and interactions with
other organisms in altering the relative contributions of consumptive and non-
consumptive effects to predator-prey interactions have focused on examining the effect of
resource levels (Eklov and Halvarsson 2000, Peacor and Werner 2004b, Turner 2004,)
competition (Peacor and Werner 2000, 2004b, 2006, Turner 2004) and predator type
(Eklov and Werner 2000, Preisser 2007). These studies found that non-consumptive
effects are dependent on resource levels and interactions with other organisms. However,
we still do not have a good understanding of how abiotic environmental conditions might
affect trait expression or the importance of non-consumptive effects. More work is
needed to understand the breadth of conditions over which non-consumptive effects are
important.

Through this study I hope to be able to answer the following questions about the
generality of the importance of non-consumptive effects. Would non-consumptive
effects be less important in a different field, stream, or pond with slightly different abiotic
and biotic conditions? Does inter—annual and intra—annual variability in environmental
conditions affect trait expression and/or the importance of non-consumptive effects?
Might abiotic and biotic changes associated with establishment of invasive species,
climate change or extreme weather events affect trait expression and/or the importance of

non-consumptive effects?

12

 

In order to answer questions about the generality of the importance of non-
consumptive effects I employed different modeling techniques to examine one system
extensively, determining the importance of non-consumptive effects under a wide range
of conditions spanning intra-annual and inter-annual variability. I used modeling
techniques to determine adaptive trait expression and to quantify the instantaneous
impact of predation on prey population growth rate in terms of direct consumption
(consumptive effect) and decrease in reproduction due to predator induced trait changes
in the prey (non-consumptive effect). I used a well studied system, D. mendotae and B.
longimanus, in which non-consumptive effects have been shown to be important (Pangle
et al. 2006 & 2007). D. mendotae change their vertical migration behavior by migrating
deeper in the water column during the day in an attempt to avoid direct predation
(consumptive effect) by an invasive predator B. longimanus but incur a reproductive cost
(non-consumptive effect) from being in colder water. I modeled the decrease in per
capita population growth rate of D. mendotae due to predation by B. longimanus over a
range of environmental conditions to examine the generality of the importance of non-
consumptive effects.

I used two modeling techniques, differential equations and dynamic optimality
modeling to determine if the behavioral response of D. mendotae to B. Iongimanus is
adaptive and to estimate the magnitude and the importance of non-consumptive effects.
Using the differential equation model of per-capita population growth rates I determined
whether the behavioral response of D. mendotae and the instantaneous importance of
non-consumptive effects are general over a range of biotic and abiotic conditions: B.

Iongimanus density, D. mendotae density, B. longimanus distribution, D. mendotae

l3

 

distribution, predation risk from additional predators, competition, surface light
availability, light attenuation and temperature. The dynamic optimality model allowed us
to add size-selective predation, which is known to be important for gape-limited
predators such as young of year fish, which prey on smaller D. mendotae (Hansen and
Wahl 1981, Mayer and Wahl 1997, Mehner et al. 1998, Graeb et al. 2004, Hulsmann et
al. 2004, Gelinas et a1. 2007) and may be important for B. longimanus predation on larger

D. mendotae (Schulz and Yurista 1999).

Methods

Description Of System:

B. longimanus, a predatory cladoceran, is a recent invader to Great Lakes region
and is thought to be having a large impact on the ecological communities of the Great
Lakes, particularly on a key prey item of commercially important fish species, the
zooplankton D. mendotae (Lehman and Caceres 1993; Vanderploeg et al. 2002, Barbiero
and Tuchman 2004). D. mendotae are small grazers that feed on algae and bacteria and
are a main food source for small and young fish. B. longimanus is a visual predator,
meaning the risk to D. mendotae is greatest near the surface where light levels are high
(Muirhead & Sprules, 2003). As light attenuates in the water column, B. longimanus are
less effective predators decreasing the risk to D. mendotae. D. mendotae reproduce at
higher rates in warmer surface waters; their fecundity falls off with the change in
temperature of the epilimnion to the hypolimnion (Pangle et al. 2006). D. mendotae
balance reproductive cost and predation by migrating within the water column to
minimize overall fecundity loss due to predation (Lehman & Caceres 1993, Pangle et al.

2006).

14

 

The B. longimanus - D. mendotae system is an ideal system in which to study the
dependence of non-consumptive effects on the environment for two reasons. First, the
importance of trait-mediated interactions has already been demonstrated in the field for
this system. Second, this predator-prey interaction has been studied extensively and field
and laboratory studies have been done that give us information on how per-capita birth
and death rate change as a function of trait expression and biotic and abiotic
environmental parameters.

Overview of modeling techniques

I used two modeling techniques, an Optimality model where I maximized per-

 

capita population grth rate (Abrams 1987, Peacor and Werner 2004) and a state-
dependent dynamic optimality model where I maximized life-time reproductive success
(Mangel and Clark 1988), that together allowed me to determine how environmental
parameters and interactions with other organisms influence trait expression and, in turn,
the relative contribution of non-consumptive effects. The optimality model which
maximized per-capita population growth rate was used to determine how trait expression
and the importance of the non-consumptive effects varies over a wide range of biotic and
abiotic conditions: B. longimanus density, D. mendotae density, B. longimanus
distribution, D. mendotae distribution, predation risk from additional predators,
competition, surface light availability, light attenuation and temperature. The state-
dependent dynamic optimality model was used to examine the effect of size-selective
predation on trait expression, which was not possible with the per-capita population
growth rate model. Size-selective predation is known to be important for 1) B.

longimanus predation on larger Daphnids and may be important for larger D. mendotae

15

and 2) for predation by young of year fish. I then uSed the predictions from the state-
dependent dynamic Optimality model in a per-capita population growth rate model to
determine the relative contribution of consumptive and non-consmnptive effects.

In the following sections I discuss each of these models in detail. First I describe
how I determined adaptive trait expression for the per-capita population growth rate
model and the state-dependent dynamic optimality model. Next, I use my results on
adaptive trait expression to calculate consumptive and non-consumptive effects. In the
third section I describe how each model was parameterized from field surveys and

experiments. Finally, I describe how I examined the influence of different environmental

 

factors. a _
Adaptive Trait Expression

I determined how different biotic and abiotic factors impacted. trait expression; in
this system the trait in question is vertical migration depth. To find an adaptive vertical
migration depth under a set of conditions, I assumed that D. mendotae maximize their
fitness, which I measured in terms of per capita population growth rate as a function Of
trait expression (as in Abrams 1984 and 1987) or lifetime reproductive success (as in
Mange] and Clark 1988). To determine the effect of B. longimanus density, D. mendotae
density, B. longimanus distribution, D. mendotae distribution, predation risk from
additional predators, competition, surface light availability, light attenuation and
temperature on adaptive vertical migration depth, per capita population growth rate was
used as a surrogate for fitness. To determine the effect of size-selective predation on
adaptive vertical migration depth, fitness was measured in terms of lifetime reproductive

success.

16

Optimality Model: maximizing population growth rate

Optimality models, where a measure Of fitness is optimized, are Often used in
behavioral ecology to determine adaptive trait expression (Gore and Paranjpe 2001). I
used per-capita population grth rate as a measure of fitness, a metric that has been
used previously to determine adaptive trait expression (e. g., Abrams 1984, 1987, Peacor

and Werner 2004). The general equation for per capita population growth rate, g, is

 

_ l dND(Z)_ _
g<z)— ND dt —b(z) d<z> (2.1)

where ND is D. mendotae density, 2 is depth in water column, b is per capita birth rate, d

is per capita death rate and t is time. I find the vertical migration depth between 0 and 60

 

m, that maximizes equation 2.1, by differentiating with respect to z and setting equal to
zero and solving to find extrema (as in Abrams 1984, Gore and Paranjpe 2001 , Chapter
7). This process is repeated for each hour in a 24 hour period over a variety of different
biotic and abiotic parameters (table 1). Thus for a 24 hour period there are 24 values for
adaptive migration depth, each for a specific time, which I denote, 2”. The model
assumes that D. mendotae migrate to each of these adaptive vertical migration depths
during a 24 hour period.

In order to determine whether D. mendotae behave adaptively (as predicted by my
model) in nature, my model predictions were compared to field data. Field data from
Pangle et al. (2007, figure 3) for surface light levels, attenuation coefficients and
temperature were used to predict adaptive migration depths. I compared model
predictions for both mid-day and mid-night to average D. mendotae depth from field

surveys conducted in Lake Michigan (see Pangle et al. 2007 for details of field surveys).

17

Dynamic Optimality Model: Maximizing lifetime reproductive success

Dynamic Optimality models can be used to model adaptive traits that are
dependant on state variables of the particular organism (Mangle and Clark 1988). This
allows us to determine how adaptive behavior should change depending on both state
variables (D. mendotae size) and the environment (size-selective predation) which was
not possible with a non state—dependent model. In this section, we are interested in
determining how adaptive vertical migration depth (and non-consumptive effects) will
change as a ftmction of D. mendotae size when predation risk is size-dependent. I built a
dynamic optimality model with D. mendotae size as a state variable and predation risk as
a function of size.

I found the adaptive vertical migration depth as a function of time, abiotic factors
(surface light levels, light attenuation, temperature profile), biotic factors (predation) and
state (size) of an individual D. mendotae using a recursive formula for a measure Of

fitness, life-time reproductive success,

(b(w,z,t) = max [b(z) + e_dF(w’z’t)—d3 (W’z’t) x (I>(w+ wsg(z),z,t +1)J. (2.2)

0<z<60
where z is depth, 1 is time in hours, w is D. mendotae size, (D is lifetime reproductive
success, b(z) is birth rate, sg somatic growth, dB is size-selective predation due to B.
longimanus and dp is size-selective predation due to young fish. This equation is used
recursively to determine a series of adaptive vertical migration depths (between 0 and 60
m) for each hour, over the course of a 30 day period, an appropirate life-span for D.

mendotae.

l8

 

Net Effect, Consumptive Effect and Non-consumptive effect

I calculated the net effect, non-consumptive effect and consumptive effect of B.
Iongimanus on D. mendotae in the same currency, all in terms of a reduction in D.
mendotae per capita population growth rate due to predation by B. longimanus. I define
the net effect of B. longimanus on D. mendotae as the depression in D. mendotae per

capita population grth rate due to predation by B. longimanus;
NE = gnp (2) - g p (2) (2.3)

This equation is equivalent to D. mendotae birth rate in predator absence minus birth rate

in predator presence, plus the death rate in predator presence.
NE = [b(znp,h) — b(zp,h)J+ d(zp,h) (2.4)

I then average over 24 hours to account for the effect Of diel changes in adaptive vertical

migration depth.

24
Zlb<znp,h)—b<zp,h)]+ dept)

NE = h=l 2.5
24 ( )

 

where NE is the net effect of B. longimanus on D. mendotae, z,.,,,;, is the adaptive
migration depth at time h without B. longimanus present, 2,”, is the adaptive migration
depth at time h with B. Iongimanus present, b(zanJ is per capita population birth rate in
the absence of B. longimanus and b(zpfi) and d(zp,;,) are respectively the per capita
population birth rate and death rate at depth 2p), with B. longimanus present.

The non-consumptive effect (N CE) is the portion of the net effect of B.
longimanus predation on D. mendotae per capita population growth rate that is due to the
reproductive cost of vertical migration, which is the difference of D. mendotae ’s birth

rate in the absence and presence of B. Iongimanus.

l9

24
Zb(znp,h) _ b(Zer)

NCE = ”=1 (2.6)
24

 

where 2",”, ZN» b(zmh) and b(zM) are as above.

The consumptive effect (CE) is the portion of the net effect of B. longimanus
predation on D. mendotae per capita population growth rate that is due to direct
consumption of D. mendotae by B. longimanus V

24
Zd(zp,h)

CE=h_=_'_—__. 2.7
24 ( )

Parameterizing the models

Population growth rate as a function of depth (no size-selective predation)

The per capita birth rate of D. mendotae as a function of daily vertical migration

depth, b(z), was estimated using the egg-ratio method (Palheimo 1974).

rn(E +1)
b(z): ӣ0 (2.8)

 

This method combines field data on the number of D. mendotae eggs (E) and D.
mendotae individuals (ND) into an egg—ratio (E/ND), with a relationship for how egg
development duration is affected by temperature (D). Values for the number of D.
mendotae eggs and individuals from the field survey of Lake Michigan were combined to
give an egg-ratio with an average value of 0.7 but ranged as high as 1.02 (Pangle, KL
unpublished data). I used this egg ratio method because it is a surrogate for resource
levels. It allowed us to incorporate information on resources without explicitly including
resources in the model and allowed us to investigate competition by altering a single

parameter, the egg ratio. For D. mendotae, the relationship between water temperature

20

 

and egg developmental duration has been derived from laboratory experiments and can
be estimated using the equation:

1
[0.0004172 + 0.0108T — 0.0163]

 

(2.9)

(Edmonson & Litt 1982). Temperature gradients spanning the whole season were

obtained from a model developed by researchers at GLERL (Schwab and Bedford 1999),

these data were fit (R v2.5.1, nonlinear regression, p<0.001) with a sigmoidal equation of I
the form
T=t - + ""f (210)
mm 5:2 '
1 + e ’5 i

 

where 1mm, Idlf tc and ts are parameters of the sigmoidal equation. Here, tnu'n gives a
minimum temperature for the hypolimnion, tdlf the difference in temperature between the
hypolimnion and epilimnion, tc gives the location (depth) of the thermocline and tse the
steepness of the thermocline.

Given the temperature dependence of egg development, and how depth affects
temperature, I combined equations 2.9 and 2.1.0 in equation 2.8 to give per capita birth
rate, b(z), as a function Of depth
_ 2 1

E ’d' "d'
b(z)=ln(—IQ—D—+1) 0.00041 rm," +——izf:; +0.0108 1min +——izf_—tc— -0.0163 .

 

 

 

 

L 1+e ’5 1+e ’3

(2.11)
Per-capita death rate as a function of depth, (1(2), due to B. longimanus predation

was determined by combining field data on B. Iongimanus and D. mendotae distributions,

21

densities and light gradients with lab data on predator efficiency at different light levels,
and the functional response. The predation rate of B. longimanus on D. mendotae is
given by a type II functional response, F (2) which gives an attack rate multiplied by B.

longimanus density, N 3(2)

19(2) = N 3(Z)F (N 0(2)) ' (2-12)

This predation rate is then modified by a term that accounts for the decrease in predation

rate as light levels decrease with depth, P(z) I
D<z>=P<z>NB(z)F(ND(z» (2.13) A

I then divided by D. mendotae density, ND(z) to get a per capita predation rate

 

“.L ‘2‘:-

 

: P(Z)NB(Z)F(ND(Z)) (2.14)

61(2) N13(2)

The functional response of B. longimanus is given by (K. L. Pangle unpublished

data).

2.67ND

2.15
5.31+ND ( )

F(ND)=

I found that the distributions of B. Iongimanus and D. mendotae in Lake Michigan (from

Pangle et al. 2007) could be well described by a normal distribution (Rv2.5.1, nonlinear

 

 

regression, p<0.001 )
N3(2)=chjEe 2"le (2.16)
ND(z)= CD e 2002 (2.17)

22

where ND(z) is D. mendotae density at depth 2, N 3(2) is B. longimanus density at depth 2,
pp and 0'1), are mean depth and standard deviations of D. mendotae distributions from
field surveys, p3 and 03 are mean depth and standard deviations of B. longimanus
distributions from field surveys, and cD and CB are scaling constants.

Light levels (light level in microeinsteins/mzs) affect predation risk according to

the relationship (K. L. Pangle, S. D. Peacor, and H. A. Vanderploeg, unpublished data).

L !

P(L)=11.5+L (2'18)

 

Light attenuates as it moves through the water column at a rate K from a surface light

level of 10 (Beers Law) such that the light level is described by

 

L(z) = [De—2K . (2.19)
Combining equations 2.18 and 2.19 gives a function for how per capita predation rate is

impacted by light attenuation

[Ge-'ZK
10(2) = K (2.20)
11.5 + 10e—z

 

Equations 2.15, 2.16, 2.17 and 2.20 were combined into equation 2.14 for per capita

predation rate as a function Of depth

23

 

 

 

 

 

 

 

 

' 7
_(z-MJ)2
CD 20‘ 2
_ - 2.27 e D
. _(z-#B)2 Urn/27!
1.3510e_ZK c3 8 2032
13.5+10e‘ZK_ 0372” _(Z-#D)2
CD 20' 2
r ‘ 5.3l+ e D
O'DVZJZ'
d(z)= - 2 - (2.21)
_(Z-IID)
CD 2002

 

Lifetime reproductive success (size-selective predation)

The parameterization for the first term of equation 2.2, b(z), is described above
(equation 9). To parameterize the second term of equation 2.2, we need functions for
predation risk due to young of year fish and B. longimanus. Predation rate due to B.
longimanus, d3(w,z), was estimated using data for B. longimanus distribution and density
(equation 2.12), typical D. mendotae distribution and density (equation 2.13), functional
response (equation 2.11) and light levels (equation 2.16) and data from Schulz and
Yurista (1999) to account for size preference of B. longimanus, SP(w). I used data from
experiments conducted by Schulz and Yurista (1999) on the size preferences of B.
Iongimanus for D. pulicaria (a slightly larger species of Daphnia) to estimate SP(w) in
terms of number of D. mendotae eaten per B. mendotae per hour

SP(w) = 0.73w + 0.53. (2.22)

I modify equation 2.14 from above to incorporate size-selective B. longimanus predation
by multiplying by a term that describes how per-capita predation rate is modified by D.

mendotae size

24

___ SP(w)P<z)NBF(ND)

dB(W,Z) N
D

 

(2.23)

Predation risk due to young of year fish, dp(w,z), was estimated as in Fiksen 1997 using
data from Gelinas et al. (2007). Predation rate was assumed to be proportional to fish
density Np, the area of image of D. mendotae, A(w) and light level as a function of depth,
L(z) (equation 2.19 from above).

NFL(z)A(w)
Afr?)

 

dp(w,z) = (2.24)

where A(W) is the area of an individual D. mendotae with median size. The area of image
of D. mendotae was estimated by assuming spherical body shape.
Population growth rate (size-selective predation)

As described above, I used equations 2.5, 2.6 and 2.7 along with adaptive
behavior predicted by the state-dependent dynamic optimality model (equation 2.2) to
determine the magnitude of the net, trait-mediated and density-mediated effects of B.
longimanus predation on D. mendotae. I used a modified version of equation 2.1, such
that the death rate was dependent on both depth and D. mendotae size

1 dND(z)
ND dt

 

g(z) = = b(z) — d(z, w) (2.25)

b(z), remained the same (equation 2.9). I modeled predation rate as a function of depth
and D. mendotae size, as the sum of the predation rate Of B. longimanus, d3(z,w)

(equation 2.23 from above) and of fish, d3(z) (equation 2.24 from above).

d(Z, W) = dB (W, Z) + dF (W, Z)
d(z, w) = SP(W)P(Z)NBF(ND) + NFL(Z_){4(W) (2.26)
ND A(w)

 

25

Examining the effect of the Environment

I determined adaptive levels of trait expression and the net, trait and consumptive
effects as described above for a range of different conditions, varying surface light
intensity, attenuation rate, water temperature profile, B. longimanus distribution and
density, functional response, D. mendotae density, additional risk of death due to other
predators, competition and size-selective predation. Table 1 gives ranges for the different
parameters investigated. Predation risk from additional predators was simulated through
changing d(z) with generic types of predation risk curves with biological rational behind
them. The types of curves and rationale are described in the first two columns of Table 2.
Interspecific competition was modeled by varying the egg-ratio, which is an indirect
measure of resources in the field. Increases in competition would lead to lower egg-
ratios and decreases in competition would lead to high egg-ratios. I used my dynamic
optimality model of lifetime reproductive success to investigate negative size-selective
predation due to young of year fish and positive size-selective predation due to B.
longimanus. To do this I ran the model described above with no size-selective predation,
with only B. Iongimanus size-selective predation (or positive size-selective predation)
and with only fish size-selective predation (or negative size-selective predation) and both
B. longimanus and fish size-selective predation.

I also investigated the impact of adding swimming speed constraints to the model.
I limited the range of depths for 2,"), over which I maximized equation 2.1 to 5 m :l: 2",, h- 1,
constraining movement of D. mendotae to within 5 m above or below their previous

location. I chose a swimming speed constraint of 5m/hr because it represents a

26

biologically relevant constraint to D. mendotae vertical migration speeds (Dawidowicz

and Loose 1992),

Results

Is vertical migration adaptive?

The model predicts that D. mendotae will stay near the top of the water column in
B. longimanus absence. In B. Iongimanus presence, in contrast, the model predicts that D.
mendotae will stay near the surface at night, and migrate to deeper waters during the day
to avoid B. longimanus predation. There was also close quantitative agreement between
model and field observations. At mid-day my model predicted an optimal migration
depth of 33 m, this agreed with mean D. mendotae depths observed in the field, 30.3 m
(sd=5.04 m) (Pangle et al. 2007, figure 3). At mid-night the model predicted that D.
mendotae should remain at the surface, but the field survey showed D. mendotae at an
average depth of 11.3 m (sd=7.47 m) (Pangle et al. 2007, figure 3).

The discrepancy between model predictions and field observations at night may
arise from a combination of two factors, the shape of the fitness landscape and additional
cost or constraints not included in the model. The fitness surface is very flat as a function
of depth high in the water column; changes in depth have little effect on growth (Figure
2.1 panel A). Above a depth of 12 m the change in per capita population growth rate
associated with a decreasing depth is very small, less than 5% (Figure 2.1 panel A). This
difference in fitness may be so small that even a very small cost or constraint not
included in the model could account for this discrepancy. Constraint on migration speed
may be one such factor. Adding this migration constraint makes it necessary for D.

mendotae to anticipate daybreak and begin migrating early to avoid an area of very high

27

predation risk. The mid-night field data were actually taken at 1 AM, and by this time,
D. mendotae, have already begun migrating (figure 2.2 panellb). Panels 1a and lb of
Figure 2.2 are fitness landscapes that show D. mendotae per capita population growth
rate as a function of time Of day on the x-axis and depth Of the y-axis. There is a large
area of negative fitness between the surface and 25 m down in the water column from 7
hours to 19 hours (Figure 2.2 panels 1a and 1b). With a movement constraint of Sm/hr,
D. mendotae would have to move through this area of negative fitness (Figure 2.2 panel
la), and in doing so would incur a large survival cost, such that it is better for D.
mendotae to remain at the surface in this case (Figure 2.2 panel 2a). However if D.
mendotae are allowed to anticipate daybreak and begin moving early they can avoid this
area of negative fitness (Figure 2.2 panel lb) and vertical migration is still adaptive
(Figure 2.2 panel 2b). This delayed migration means that D. mendotae would barely
make it to the surface by midnight, combining this with the flatness of the fitness curve
gives a plausible reason for the observed discrepancy between the model and field data.
The model predictions were therefore in strong agreement, both qualitatively and
quantitatively with field observations. This close match Observed between daytime
vertical migration in the field and model predictions suggest the vertical migration

observed in the field is adaptive.

Impact Of environmental variability

General Mechanism

There are two main mechanisms that determine the effect of environmental
variability on the relative contributions of consumptive and non-consumptive effects. A

change in an environmental variable can change g(z), through changes in either b(z) or

28

d(z) or both which has two possible effects. First, a change in either b(z) or d(z) or both,
can alter the adaptive vertical migration depth, zp), which leads to changes in non-
consumptive effects and consumptive effects. Second, a change in b(z) or d(z) or both,
can directly effect the consumptive and non-consumptive effects by changing the factors
that affect them, b(znp,;.), b(zpy.) and d(znh).

Changes in environmental conditions can potentially influence both of these
factors, and thereby the contribution of non-consumptive effects. Large changes in the
relative contributions of non-consumptive effects due to changes in adaptive vertical
migration depth were due to switches between adaptive vertical migration depth being at
the surface (no migration) to an adaptive vertical migration depth greater than 22 m.
These types of switches resulted in switches from dominance by consumptive effects to
non-consumptive effects. Smaller changes in adaptive vertical migration depth (within
the range between 22 and 40 m) caused little change in the relative importance of non-
consumptive effects, unless there was an accompanying change in b(zpfl) or d(zM).
Changes in b(znpyJ, b(zM) and d(zpy.) without an accompanying change in vertical
migration depth also resulted in changes in the relative contributions of consumptive and
non-consumptive effects.

Effects of particular environmental variables

Typical levels of variation in environmental parameters experienced by D.
mendotae were predicted by the model to have a large effect on adaptive behavior and the
relative contributions of consumptive and non-consumptive effects. Over the range of
environmental conditions I investigated, adaptive vertical migration behavior ranged
from staying at the surface to migrating to depths of between 22 m and 40 m. Predation

risk during daylight hours creates an area Of negative fitness that extends from the surface

29

to 21 m, leading to an absence of adaptive vertical migration depths between 1 and 21 m.
The model indicated that the relative importance of non-consumptive effects is strongly
dependent on multiple environmental parameters. In particular, temperature, B.
longimanus density and the presence of additional predators all strongly affected the
relative contribution of non-consumptive (nonlethal or trait-mediated) effects to the net
effect of B. longimanus, while other variables had less pronounced effects (as reviewed
below). Importantly, for the majority of the ranges tested for each of these parameters,
the magnitude of the non-consumptive effects remained equal to or greater than the
magnitude Of the consumptive effects.

The absolute value of surface water temperatures had a marked, nonlinear, effect
on the contribution of trait-mediated interactions (Figure 2.3). The model indicates that a
temperature threshold exists, above which D. mendotae do not migrate, but below which
it is advantageous for D. mendotae to migrate and avoid predation. For high surface
temperatures, above a threshold value (of ~25°C) the advantages of staying in the warm
water outweigh the risks of direct predation and it is best for D. mendotae to stay at the
surface. Thus small changes in temperature can have a profound influence on the
contribution of non-consumptive effects. The mechanism at work here is a change in b(z)
and an alteration in adaptive vertical migration depth. Above the threshold, the net effect
of the predator is minimized by remaining at the surface. Since D. mendotae do not
migrate under these conditions, non-consumptive effects are non-existent. But as
temperature falls below the threshold, vertical migration (to a depth of at least 22 m)

becomes adaptive and non-consumptive effects dominate.

30

 

 

The model predicts that the threshold temperature described above is dependent
on B. Iongimanus density. As B. longimanus density increases, the threshold temperature
increases as well. As B. longimanus density increases d(z) increases meaning that the
threshold surface temperature must also increase in order for the advantages Of staying in
the warm water at the surface to outweigh the risks of direct predation.

The density of the predator, B. longimanus also had a marked, non-linear effect on
the relative contribution of non-consumptive effects (Figure 2.4). For low B. Iongimanus

densities, below a threshold of l individual/m3, it was adaptive for D. mendotae to remain

 

at the surface because predation risk was not high enough to make vertical migration

 

beneficial (Figure 2.4). At and above that threshold, adaptive behavior shifted to an
adaptive vertical migration depth of at least 22 m, predation rates were now high enough
at the surface that it was best for D. mendotae to migrate. The mechanism here, like
above, is a shift in adaptive behavior, however unlike temperature which affect b(z),
predator density affects d(z).

Variation in the temperature profile had a large impact on the magnitude and
relative importance of the non-consumptive effects. As the difference in temperature
between the epilimnion and hypolimnion decreased the cost of migration also decreased,
leading to trait-mediated effects of smaller magnitudes and a decrease in the relative
importance of trait-mediated interactions (Figure 2.3). The change in the temperature
gradient caused a change in b(z) and the difference between b(znp,h) and b(zm) to
decrease. However, the depth of the temperature gradient remains fairly constant, except
in the extreme case where there is no temperature gradient, so there is very little change

in adaptive migration depth, zp_;,. In the extreme, when there is no temperature gradient,

31

the magnitude of non-consumptive effects is zero and consumptive effects make up the
entire net effect of B. longimanus predation on D. mendotae.

Interestingly, I found that some parameters, such as surface light levels and the
light attenuation coefficient had large effects on adaptive vertical behavior, but not much
Of an impact of the relative importance of non-consrnnptive effects. Variation in surface
light levels and the rate at which light attenuates caused changes in the shape of d(z),
causing the adaptive vertical migration depths during daylight hours to range from 22 m
to 40 m but had very little influence on the magnitude of the non-consumptive effect.
These parameters have no influence over b(z) and the changes they caused in adaptive
vertical migration depth were in a range (below the thermocline) where b(z) varied little
and thus had little no impact on non-consumptive effects (see equation 2.6). Both
parameters did affect d(z), although the change was primarily above a depth of 20 m and
since adaptive vertical migration depths were always below 20 m the magnitude of the
consumptive effects were barely affected by these changes to d(z). Thus, neither, surface
light levels or light attenuation had much of an impact on the relative importance of non-
consumptive effects.

Variation in predation risk as a function of depth (due to the inclusion of
additional predators) influenced the importance of non-consumptive effects, both
increasing and decreasing the relative importance of non-consumptive effects depending
on the type of predator involved (summarized in table 2). Adding a linear predation
curve, constant or increasing/decreasing with depth, led to non-consumptive effects
becoming relatively less important, however, only with large predation risk at deeper

depths do I see consumptive effects begin to dominate non-consumptive effects. For

32

example, for a constant curve where risk is independent of depth, adaptive vertical
migration depth and the magnitude of the non-consumptive effect remained constant,
however the magnitude of the consumptive effect increased, decreasing the relative
importance of non-consumptive effects. Interestingly, under some conditions such as
linear predation risk with a very large negative slope (predation risk increases with
depth), staying at the surface is adaptive, and under these conditions there is no trait-
mediated effect. Here, the model predicts a similar threshold effect to what I have seen
with surface temperature and predator density. As predation rate increases deep in the
water column, the relative contribution of consumptive effects increase do to increasing
d(znp,h), until a threshold above which vertical migration is no longer adaptive and D.
mendotae do not migrate and consumptive effects make up the entirety of the net effect
of B. longimanus on D. mendotae. Adding a non-linear predation curve, where predation
risk decreases as a function of depth (such as for a Visual predator or temperature
sensitive tactile predator) affects both adaptive migration depth and slightly influences
the importance of non-consumptive effects. Adding either an additional visual predator
or a tactile (temperature sensitive) predator would either have no impact on adaptive
vertical migration depth or would slightly increase adaptive vertical migration depth.
The addition of either curve would have no impact on the magnitude of the non-
consumptive effect, since b(z) does not change the value of equation 2.4, however, the
magnitude of the consumptive effect could increase causing non-consumptive effects to
become relatively less important.

Variation in D. mendotae densities and competition had little to no effect on

adaptive migration depth or the importance of trait-mediated interactions. Variation in D.

33

mendotae densities changed d(z) through the functional response, however, this effect
was very small compared to the effect of light attenuation and had very little impact on
zp), or d(zM). Variation competition changed b(z), but had no impact on the shape and so
did not impact 2,”. This change in b(z) did impact b(znpy.) and b(zpfi), but there was still a
large cost to migrating and hence a large non-consumptive effect.

Including size-selective predation caused adaptive migration depth to change over
time as D. mendotae became larger, increasing for positive size-selective predation and
decreasing for negative size-selective predation. Including: B. Iongimanus size-selective
predation for larger D. mendotae led to optimal migration depth becoming deeper over
time as size increased. The optimal migration depth for larger D. mendotae is deeper
than for smaller D. mendotae under positive size-selective predation. This is because
with increasing size d(z) becomes larger, leading to a deeper adaptive migration depth
and a slightly larger consumptive effects. Adding a gape limited predator (such as a
small fish) that preferentially feeds on small D. mendotae led the Opposite effect,
adaptive vertical migration depth decreased over time. As D. mendotae became larger,
the depth to which D. mendotae migrated decreased and the magnitude of the
consumptive effects decreased slightly. This is because as D. mendotae size increased,
d(z) became smaller, leading to shallower adaptive migration depths and slightly smaller
consrunptive effects. Adding both positive size-selective predation and negative size-
selective predation increased the vertical migration depth slightly for large and small D.
mendotae, thus vertical migration depth did not change over time as with only negative or

positive size-selective predation.

34

Increasing migration depth due to size-selective predation had no impact on the
magnitude of non-consumptive effects, as b(z) did not change with the addition of size-
selective predation. The magnitude of the consumptive effect increased slightly with
both positive and negative size-selective predation. There was only a slight change
because both B. longimanus and young of year fish are visual predators and most of the
change in d(z) was well above the adaptive vertical migration depths. Thus there was
very little change in the relative importance of non-consumptive effects with size-

selective predation.

Discussion

My results show that trait expression (in this case vertical migration depth) and
the importance of trait-mediated interactions are highly dependent on environmental
conditions. The model predicts big changes in adaptive vertical migration depth and the
relative contributions of consumptive and non-consumptive effects for some variables
(surface temperature, temperature gradient, B. longimanus density and preSence of
additional predators) and not others (D. mendotae density, competition, surface light
levels and light attenuation rate). Some of the changes I observed were gradual, for
example, seasonal changes in the temperature gradient caused a gradual decrease in the
magnitude of non-consumptive effects. Other variables, such as B. longimanus density
and surface water temperatures caused strongly nonlinear changes. The variation in
impact on the relative importance of non-consumptive effects can be explained by
examining how the factors affect birth and death magnitudes as a function of depth, and

in turn how this affects the adaptive trait change.

35

Given the ubiquity of trait changes and the dependence of trait changes on the
environment, I expected to find the large effects that I observed. Others have
investigated the effect of resource levels (Eklov and Halvarsson 2000, Peacor and Werner
2004b, Turner 2004,) competition (Peacor and Werner 2000, 2004b, 2006, Turner 2004)
and predator type (Eklov and Werner 2000, Preisser 2007). These initial studies point to
the importance of considering the effects of environmental variation on the relative
importance of consumptive and non-consumptive effects. I have expanded on this initial
work, showing that environmental variation can have large effects on the relative
importance of non-consumptive effects with both gradual and strong, non-linear changes.

The predation curvature results show that different types of predators will have
different types of effects. I found that visual and tactile predators, which are very
common in Lake Michigan, increase the magnitude of Consumptive effects, slightly
decreasing the relative importance of non-consumptive effects. Other studies have found
that trait expression changes with number of predators present or the type of predator
present. In particular, Preisser et al. 2007, found that the importance of non-consumptive
effects was highly dependent on predator hunting mode. My results imply that the
importance of non-consumptive effects depends on the specific predators present and
how predation risk changes as a function of trait expression for the assemblage of
predators present.

I found that predation rates were not proportional to predator density. Rather, I
Observed a marked, non-linearity in this relationship. As predator density increased to
above 1 B. longimanus per m3, the model predicts a shift in adaptive behavior from no

vertical migration (remaining at the surface) to migrating to a depth of at least 22 m. This

36

behavior shift leads to a shift from dominance of consumptive effects to dominance of
non-consumptive effects resulting in a large decrease in the actual predation rate. These
results therefore show that if prey modify traits adaptively in response to predation, the
death rate of prey due to direct consumption by the predator does not increase
proportional to predator density (Abrams 1993), rather there is a strong non-linear
decrease in predation risk due to a threshold where adaptive behavior flips, having a huge
impact on the relative contribution of non-consumptive effects.

I also found strong nonlinear responses as a function of temperature. As surface
temperatures increase there was a gradual increase in the importance of non-consumptive

effects, until a threshold, at and above this threshold the model predicts a shift in adaptive

 

behavior similar to the threshold described above. Surface temperatures become so high
that it is no longer adaptive for D. mendotae to migrate, non-consumptive effects are
nonexistent and consumptive effects make up the entirety of the effect of the predator.
This has huge implications in terms of climate change. The value of this threshold is
dependent on B. longimanus density. At natural B. longimanus densities, the threshold is
at 25°C. Currently, surface temperatures in Lake Michigan during August are typically
below this threshold (Pothoven et al. 2001), but with the changes in temperature
predicted under climate change, we may see temperatures rising above this threshold on
some days during July and August. This would mean a period in late summer when non-
consumptive effects are non-existent and consumptive effects dominate, currently non-
consurnptive effects are at their highest levels during this time period. The implications

of climate change are discussed further below. This also has implications across similar

37

systems, implying that we may see large variation in the relative importance of non-
consumptive effects depending on lake temperature.

The model predicts that changes in the Great Lakes due to climate change could
increase the time period over which non-consumptive effects are important. Climate
change scenarios indicate shorter winters with longer periods of thermal stratification
with some predictions placing fall turnover up to 2 months later (Brooks and Zastrow,
2002). If stratification were to persist longer into the winter it would add to the period of
time that non-consumptive effects dominate. However, if surface water temperatures
increase much above current levels (increasing by as little as 2 °C) the model indicates
that D. mendotae no longer need to migrate. These results, taken together indicate that
we may begin to see a brief period during the summer months when density changes
dominate due to higher surface temperatures, transitioning to a longer period where non-
consumptive effects dominate due to longer periods of stratification. More work is
needed to clarify the non-consumptive effects given the simultaneous operation of
different mechanisms if we integrate these two results over time.

My findings, that non-consumptive effects are dependent on environmental
conditions, have implication for aquatic ecology and community ecology. New studies
on the importance of non-consumptive effects in specific systems need to consider how
environmental conditions might influence non-consumptive effects based on the biology
of the system. Otherwise, they run the risk of under or over-estimating the impact of the
predator on the prey. Further, in interpreting thelscope of the influence of studies, my
results indicate that ecologists need to be careful about interpreting how general finding

of the importance of non-consumptive effects are with respect to variation in

38

 

environmental conditions. For example, consider previous work that has shown non-
consumptive effects to be very important for this system in July and. August (Pangle et a1
2007), to extend these results to other times of the year or other lakes we need to consider
how environmental variability might influence adaptive vertical migration and the
relative contribution of consumptive and non-consumptive effects. My results indicate
that non-consumptive effects are dependent on environmental conditions, but that over
the majority of the range of environmental conditions, non-consumptive remained a large
and important portion of the net effect of B. longimanus. Thus we can make predictions
about other lakes surrounding the Great Lakes that have already been invaded by B.
longimanus or are under risk of invasion. My results indicate the non-consumptive
effects will likely dominate the net effect of B. longimanus on D. mendotae per capita
population growth rate, unless the temperature gradient is unusually small or the lake
contains a D. mendotae predator that is more efficient in cold or dark water or B.
longimanus densities are very low, below 1 individual per m3. We can also make
predictions about whether non-consumptive effects remain important over a whole
season. My results indicate that as the temperature gradient decreases the magnitude and
relative importance of non-consumptive effects decreases as well.

Extending the generality of the importance of trait-mediated interactions over a
wide range of environmental conditions has implications for community ecology. Theory
has shown us that, at least in simple commrmities, incorporating changes in trait
expression and the resulting fitness costs, can stabilize communities, change the relative
strengths of direct and indirect interactions (reviewed in Bolker et al. 2003) and even help

repel invasion (Peacor et al. 2006). It is unclear how including the effect of the

39

 

environment on trait expression might impact these results. However, the strong changes
I observed in trait expression and the relative contributions of consumptive and non-
consumptive effects indicate that considering the effect of environmental variation is
impOrtant.

Understanding how trait expression and trait-mediated interactions can vary with
environmental conditions has implications for ecological communities in terms of trait-
mediated indirect interactions. We know that including plastic trait expression and the
resulting cost in terms of fitness can change the relative strengths of direct and indirect
interactions (Abrams 1995). Knowing that the importance of non-consumptive effects
changes as the temperature gradient changes in Lake Michigan or as B. lonimanus
densities increase could have an impact the rest of the community. B. longimanus are
having a large impact on the community in Lake Michigan and understanding the
mechanisms behind the changes that are occurring in the community is very important
(Lehman and Caceres 1993; Vanderploeg et al. 2002, Barbiero and Tuchman 2004). D.
mendotae are an important food source for young of year piscivorous fish and for
planktivorous fish. If non-consumptive effects are important, then the reduction in D.
mendoth population growth rate is due to lower birth rates rather than consumption by
B. longimanus. This means that energy is not flowing up the food web into B.
longimanus, rather the resources that would be consumed by D. mendotae may be
available to competitors that may or may not be good food sources for fish. Further work
should look into the trait-mediated indirect interactions within Lake Michigan and
determine what impact variation in trait expression and non-consumptive effects has on

other species, particularly important fisheries species such as yellow perch.

40

Tables

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Parameter Parameter description Field or lab data Parameter space
range investigated

E/N Egg-ratio 0.7 0.35 to 1.05

tmin Deep water temperature 4-6°C 4°C - 6°C

tdir Surface temperature 4°C - 30°C 4°C - 30°C

Tc Depth Of thermocline 18.7 m 14m to 24 m

Ts Steepness of thermocline 2.8 1 to 6

[.13 Center of B. longimanus 11.6 m +/- 5
Distribution

up Center of D. mendotae 11.0 m, 31.1 m +/- 5
Distribution

CB Constant determining height 169 +/- 50
Of B. longimanus distribution

CD Constant determining height 26060, 28610 +/- 5000
of D. mendotae distribution

0'3 st. dev. of B. longimanus 6.6 m +/- 3
distritubion

on st. dev. of D. mendotae 7.2 In, 4.6 m +/- 3
distribution

10 Surface Light Levels 0 to 2500 uE

K Attenuation Coefficient 0 to 1.5

 

Table 2.1. A summary of the Environmental Parameters included in the population

 

growth rate model including Parameter, Parameter description, field or lab data range and
parameter space investigated.

41

 

 

 

 

 

 

 

 

 

Curve type Biological Summary Of Results
rationale

Constant Tactile predators where DVM to same depth, but consumptive
predator movement is effects become more important as
independent of temp predation risk increases
(endothermic)

Linear (risk increases light or temp decreases DVM depth, under some

w/ depth) dependent but not conditions best to stay at surface,
directly proportional consumptive effects“ become more

important

Linear (risk decreases Somewhat light or Still best to migrate, increases DVM

w/ depth) temp dependent but not depth, non-consumptive effects
directly proportional continue to dominate

Non-linear, predation Temperature dependent Still best to migrate, increases DVM

decreasing with depth tactile predator depth, non-consumptive effects

(following ' continue to dominate

temperature gradient)

Non-linear, predation Visual predator Migration depth remains the same,

decreasing with depth consumptive effects become gradually

(following light more important as the efficiency of the

gradient) predator increases

 

Table 2.2. A description of additional predation curves added to the Population Growth
Rate model, their biological rationale and summary of the results.

42

 

Figures

 

 

 

 

 

 

 

 

 

 

 

 

Per capita Population Growth Comparing model predictions
Rate as a function of Vertical for adaptive vertical migration
Migration Depth (m) o depths and field data
0 ‘0 ---1 -
,0 ._
o 8 ‘1
m ..
E
E e — s ..
8
o
C
4% a - a .
:6»
2
a
g a - g 4
> T
o —-4
4. 5:3 ..
O -l _—
I l I I I l l O " “_.-
-2.5 4.5 -0.5 0.5 mid-day rm'd-nimt
per capita population growth rate Trme of Day

Figure 2.1. Comparing model predictions and field observations.

Per-capita population growth rate as a function of depth for mid-day and mid-night are
given in Panel A. This fitness curve (panel A) gives predictions for adaptive vertical
migration depth, the depths that maximize per-capita population growth rate (shown in
black bars on panel B). Average vertical migration depth from a field survey conducted
in Aug. 2005 are shown in panel B (white bars), the mid-day data are from 2pm and the
mid-night data are from 1 pm. The error bars represent standard deviations. The model
predictions were made for the same light and temperature conditions and times. The
mid-day actual depths and predictions agree very well. There is a discrepancy between
the mid-night predictions and field data, but this can partially be explained by the flatness
of the fitness curve near the surface (between 0 m and 12 m).

43

Daphnia Population growth rate Dag‘hnia Population growth rate
wr wt 3 migrafi

 

   

 

 

 

 

 

 

 

 

13 h a migratron constraint and on constraint and
no antrcrpatron of sunrise anticrpation of sunrise
10 - .
A 20 - ° .
5. ° °
£3“ .. ~..
o
o
40 ~
50 <
60 l t I l l l l I
0 5 10 15 20 10 15 20
Time (hr) Time (hr)
, W -. r - ,,
1b Decom 2b

Decomposin the predator effect
with a mlgra 'on constraint and
anticrpation of sunrise

posing the predator effect
withou migration constraint and
anticipation of sunnse

 

Depression of D. mendotae per

capita population growth ra
Depressron of D mendotae per
capita population growth rate

   

I
Figure 2.2. Including a migration constraint makes anticipating day-break necessary.
1a and 2a show the same fitness landscape with different migration scenarios. The
fitness landscape shows a region of negative population growth rate during daylight hours
from 0m to 25m. 1b and 2b compare the net predator effect with and without vertical
migration (DVM) and break the net predator effect into non-consumptive and
consumptive effects. The impact of B. longimanus is measured in terms of the depression
of the per capita population growth rate due to the predator, no DVM refers to the total
predator effect without D. mendotae migration, DVM refers to the total predator effect
with migration. The no DVM and DVM bars are compared to determine whether
remaining at the surface (no DVM) or migrating (DVM) is adaptive. The net effect is the
total effect of B. longimanus on D. mendotae for the adaptive strategy. 1a assumes a
biologically realistic swimming speed constrain, in this scenario, D. mendotae would
have to migrate through the area of negative fitness incurring a huge mortality cost. 1b
shows that the effect of the predator is minimized by remaining at the surface (no DVM).
The migration scenario shown in 2a assume the same migration constraint but also
assume that D. mendotae begin migrating early avoiding the large cost of moving through
relatively shallow water with high B. longimanus density during daylight. 2b shows that
migrating minimizes the total predator effect and that the non-consumptive effect makes
up the majority of the total predator effect.

 

 

44

 

l Erect of Variation in Average Surface Vthter Terrperature on the ‘
conposition of the predator effect of B. Iongirranus on D. rmndotae

-----------------------

 

6a,,

 

 

 

 

 

   

 

  

 

 

 

 

 

 

- 8»
N
a:
g c 0.25 y WMJI ............ .:‘ 3 Rage; of .1
. u - l U em :‘
53 ‘— -ncx1-oorsurr.t|\e; l ......... ..y—s-. ‘ by monthp 5‘
8% 02 dfects l "" ‘ lu‘:......' ..... fif‘JUSt L l r l l I
3 3 l- ' 'mmm 1 l ‘ September 4 ‘ l
a; . dfects ‘ . ..................... ‘Wl / L l L i
gé 0-15 ”W October Tf ‘lfh ' ' , r
\ I.
g5. 01 ‘ l 4 &l _L ‘ ii ‘
~ - r l 7 A, a
5 III December l l I ‘ l l l
. 5.2 ‘ 1 l ' 1 1 l l
a g 0% k a .4, 1-“? 1 T .71 w
_‘u t 1 ‘ ‘
é o I'l'r'i'".LLl4L
g 5 1o 15 20 25 sol

 

Surface Water Temperature (°C) l

 

Figure 2.3. The effect of variation in surface temperature on the composition of the
predator effect.

The ranges of average surface temperatures for Lake Michigan by month are shown by
the amount of the x axis covered by the five boxes near the top of the graph. As the lake
transitions into fall and approaches the fall turnover, the relative importance of the non-
consumptive effect decreases and the relative importance of the consumptive effect
increases, although its magnitude remains fairly constant. There is a strong non-linear
effect at higher surface temperatures. Above a threshold temperature of 25 °C, D.
mendotae do not migrate, non-consumptive effects are zero and consumptive effects
make up the entire net effect of the predator.

45

 

Depression of D. mendotae per capita Population

 

 

B. longimanus density influences the
importance of non-consumptive effects

 

 

 

 

 

 

 

   
 

0.25

0.2_fi_~__ _;___, ::——:
l—i - no vertical migration l

0.15 —— A a »— Net effect 7

 

— — Non-consumptive effect .

 

 

 

0.1 — ---—-——1—-— “—1
'- - - - -Consumptive effect
005 ,. ____.-_Lr_..__._._..;‘ :: .3
I I I
0 __ : ______________________

 

 

Growth Rate due to B. Ionglmanus predation
_o

 

1 10
B. longimanus Density (m'3)

 

Figure 2.4 The effect of B. longimanus density on overall predator effect.

At low densities, the predators have very little impact on D. mendotae per capita
population grth rate and it is best for D. mendotae to stay at the surface so
consumptive effects make up the entirety of the predator effect since there is no trait
change. As B. longimanus densities increase, above a level of 1 individual/m3 vertical
migration is the best strategy and the net effect of B. longimanus predation on D.

mendotae is almost entirely made up Of the non-consumptive effect.

46

CHAPTER 3
SEASONAL CHANGES IN THE IMPORTANCE OF NON-
CONSUMPTIVE EFFECTS IN LAKE MICHIGAN ZOOPLANKTON

Abstract

Daphnia mendotae migrate vertically into deeper, darker, cooler regions of Lake
Michigan to avoid predation by an invasive planktivore, Bythotrephes longimanus. Since
B. longimanus is a visual predator, D. mendotae vertical migration reduces direct
consumption of D. mendotae by B. longimanus (consumptive effect), however, inhabiting
the colder and less productive regions of the water column has a reproductive cost for D.
mendotae (non-consumptive effect). These types of tradeoffs are ubiquitous to predator-
prey interactions, and studies of various systems have shown that non-consumptive
effects can be important relative to consumptive effects. In the B. longimanus — D.
mendotae system, non-consumptive effects are important relative to consumptive effects,
but this relative importance is dependent on environmental conditions. Here I translate
my previous findings on consumptive and non-consumptive effects into longer-term,
multi-generational changes in population dynamics which in this system can be observed
with-in a single season. I built a stochastic population dynamics model to
explore how changes in the pelagic environment (changes in abiotic conditions such as
temperature and light availability along with changes in biotic conditions such as B.
longimanus density, D. mendotae density, B. longimanus distribution, D. mendotae
distribution, predation risk from additional predators and competition) over the course of
a season will influence D. mendotae population dynamics. My results Show that the
importance Of non-consumptive effects is large compared to consumptive effects over the

course Of a season. I also found that predicted changes in the temperature of Lake

47

Michigan due to climate change impacts the importance of non-consumptive effects. An
understanding of how B. longimanus effects change seasonally will provide better
predictions of its impact in the Great Lakes and the many other systems which it has
invaded. Additionally, understanding how previous findings impact population dynamics
is key to predicting species effects in different scenarios, such as those predicted with

Global Warming

Introduction i

Predation can have a large impact on prey populations through direct

 

consumption, but also through a suite of phenotypic changes that they induce in

individual prey. For many years it was thought that predators affect prey populations

l
1...: P
l

primarily through consumptive effects (also lethal or density-mediated effects).
Ecologists have been aware for some time that prey respond to predation risk by
modifying their phenotype to minimize the effect of the predator. It is only recently that
ecologists have begun to understand how important a role these phenotypic modifications
play in predator prey interactions. In modifying their traits, prey incur costs in terms Of
fitness surrogates such as individual growth, reproduction or population growth rates.
This cost is termed the non-consumptive effect (also non-lethal or trait-mediated effect).
Prey change their traits in a number of ways in response to predators, including
behavioral, morphological and physiological adaptations that make them less likely to be
consumed (Lima 1998). These trait changes reduce direct consumption, but also reduce
the reproductive success of the prey (Harvell 1990). If prey forage less or put resources
into growing larger in order to reproduce earlier or into growing defensive features, they

have less resources for growth and reproduction or may face mortality from other

48

 

sources. Thus, prey face a tradeoff in trying to limit the overall effect of a predator:
avoid being eaten while maximizing fecundity. Studies have shown that predators induce
trait changes in their prey in a large number of systems (reviewed in Lima and Dill 1990,
Lima 1998, Tollrain and Harvell 1999, Agrawal 2001) and in many of these systems,
these trait changes result in large non-consumptive effects on prey growth, fecundity,
population growth rates, etc. (reviewed in Peacor and Werner 2004a, Preisser et al. 2005).

There is evidence from both empirical work and theory that suggests/predicts
non-consumptive effects will be important over long time scales. Empirical studies have
been slowly moving towards measuring consumptive and non-consumptive effects on
longer-time scales, but researchers are limited in what they can accomplish by the
restraints of biological systems, relatively long generation times and funding limits. The
question of long-term importance of non-consumptive effects has been addressed in
theoretical studies in terms of long-term community stability (Ives and Dodson 1987,
Abrams 1995), and community structure. However, these studies are not based on data,
rather, they are based on hypothetical relationships between functional responses or
trade-offs and the degree of trait expression, rather than relationships supported with data.
These studies illuminate the possible long-term implications of important non-
consumptive effects. However, it is unclear how these results will apply to real-world
scenarios.

The empirical studies mentioned so far measured non-consumptive effects in
terms of fitness correlates. There is a gap in our real-world knowledge of how these
fitness correlate results will translate into effects on population densities and dynamics.

There has been a general progression towards measuring non-consumptive effects in

49

terms of population level measures of fitness, but we do not know the importance of non-
consurnptive effects in terms of longer-term multigenerational measures, such as
population dynamics. Initial studies of non-consumptive effects focused on quantifying
costs of trait changes at the level of the individual, for instance individual growth rates,
fecundity or mortality (reviewed in Lima 1998). Recently there has been a trend towards
measuring non-consumptive effects on population level responses such as population
growth rate (r) and geometric population growth rate (A), these studies extend from
aquatic zooplankton (Boeing et al. 2005, Pangle 2007) to stream macro-invertebrates
(Peckarsky 1993, McPeek and Peckarsky 1998) to agricultural systems (Nelson 2004).
Nelson measured the non-consumptive effects of damsel bug predators on pea aphid
population growth in the field in small cages over a single generation and saw large
effects (up to 30% decrease in population growth rate). There is a need to determine how
changes in trait expression will affect prey population dynamics. Now that we know
non-consumptive effects to be important in a large range of predator-prey interactions,
what does this mean in terms of population dynamics?

Bolker et al. (2003) reviewed the needs of the field in 2003 and identified scaling
from short-term to long-term responses of communities as one of “the most critical
needs” to determining whether trait-mediated interactions affect community dynamics at
scales that will influence practical management decisions. There is considerable
ecological theory predicting that non-consumptive effects are likely to be important in the
short term (Abrams 1984, Peacor and Werner 2004b) or to long term population stability
(Ives and Dodson 1987, Abrams 1995), and to food web properties such as susceptibility

to species invasions (Peacor et a1 2006, Sih et al. 1985). These results, while important

50

and interesting, are based on hypothetical curves that relate functional responses or trade-
offs to degree Of trait expression. The relationship between trait expression and trade-
offs is unknown for many systems, thus, it is unclear whether the curves used in these
theoretical studies represent real communities. We need to know how non-consumptive
effects affect population dynamics at seasonal scales, the scale at which many practical
management decisions are made.

Managers and ecologists need to be able to translate previous findings of the
relative importance of consumptive and non-consumptive effects into longer-term
multigenerational dynamics in order to understand the implications of large non-
consmnptive effects in ecological communities. In the previous chapter of this thesis I
determined that non-consumptive effects were dependent on environmental conditions
with a short-term study. In this chapter I translate these results into longer-term
multigenerational dynamics. In our system, transitioning from short-term to longer-term
time scales means including environmental variability that would be experienced over
these longer time scales. The range of parameters I investigated in the previous chapter
encompasses all environmental variation that would occur within a season, a time frame
that for this system includes multigenerational dynamics. This chapter expands on our
previous broad range, short-term results, integrating them in order to study longer-term
multi-generational population dynamics of D. mendotae. I used a stochastic population
dynamics model that allows translation of previous results on consumptive and non-
consumptive effects into longer-term, multi-generational population dynamics. With this
model I explore how changes in the pelagic environment (changes in abiotic conditions

such as temperature and light availability along with changes in biotic conditions such as

51

B. longimanus density, D. mendotae density, B. longimanus distribution, D. mendotae
distribution, predation risk from additional predators and competition) over the course of
a season will influence D. mendotae population dynamics. This allows us to investigate
how changes in abiotic and biotic conditions due to climate change and exotic invasions

may influence population dynamics.

Methods

Description Of System:

B. longimanus, a predatory cladoceran, is a recent invader to Great Lakes region
and is thought to be having a large impact on the ecological communities of the Great
Lakes (Lehman and Caceres 1993; Vanderploeg et al. 2002, Barbiero and Tuchman
2004). One of the main prey of Bythotrephes longimanus is a zooplankton called
Daphnia mendotae (Lehman and Caceres 1993; Vanderploeg et al. 2002, Barbiero and
Tuchman 2004). D. mendotae are small grazers that feed on algae and bacteria, they are
a main food source of small and young fish, some of which are important commercially.
B. longimanus is a visual predator, meaning the risk to D. mendotae is greatest near the
surface where light levels are high (Muirhead & Sprules, 2003). As light attenuates in
the water column, B. longimanus are less effective predators decreasing the risk to D.
mendotae. D. mendotae reproduce at higher rates in warmer surface waters, their
fecundity decreases with the decrease in temperature from the epilimnion to the
hypolimnion (Pangle et al. 2006). D. mendotae balance reproductive cost and predation
by migrating within the water column to minimize overall fecundity loss due to predation
(Lehman & Caceres 1993, Pangle et al. 2006). In the last chapter, it was demonstrated

that environmental factors, such as surface temperature, B. longimanus density and the

52

presence of additional predators all can have large effects on the relative importance of
consumptive and non-consumptive effects. Because these factors change through time,
the relative importance of consumptive and non-consumptive effects will likely change
over the course of a season and impact D. mendotae population dynamics over a season

The B. longimanus — D. mendotae system is an ideal system in which to study the
importance of non-consumptive effects on longer-term multigenerational population
dynamics for several reasons. First, the importance of trait-mediated interactions has
already been demonstrated in the field for this system. Second, this predator-prey
interaction has been studied extensively and extensive field and laboratory studies have
been done that give us information on how per-capita birth and death rate change as a
function of trait expression and biotic and abiotic environmental parameters. Third, the
effect of environmental variation on the importance of non-consumptive effects has been
documented for short-term population level responses. Finally, using a zooplankton
based system allows the study Of long-term multigenerational population dynamics on
relatively Short time scales that encompasses seasonal variation in environmental
parameters that will influence multiple generations.
D. mendotae Population Dynamics Model

I built a population dynamic model to allow the extension of the instantaneous
population growth rate results from the previous chapter to population dynamics,
investigating the importance of trait and consumptive effects over a single season, i.e.
from June to Nov, which encompass multiple generations of D. mendotae and B.
longimanus. Due to the biology of the D. mendotae, B. longimanus system, the changes

in population density over multiple generations can be investigated within a single

53

season. With this model I investigate how changes in environmental conditions might
affect the importance of non-consumptive effects over the course of a season where
abiotic conditions vary naturally, and changes in biotic conditions are controlled by
model dynamics.

The model determines how D. mendotae population densities change over time
and how non-consumptive effects contribute to D. mendotae population dynamics. I
examine two different scenarios of B. longimanus density and dynamics, a responsive B.
longimanus population scenario and a fixed-data B. longimanus scenario. The fixed-data.
B. longimanus scenario uses field data to set B. longimanus densities throughout the
season. The responsive B. longimanus scenario assumes that B. longimanus population
dynamics are mainly driven by D. mendotae since B. longimanus has no other prey in the
model. I include both types of B. longimanus population dynamics to investigate the two
extremes of possible predator prey interactions. The first scenario assumes that D.
mendotae are the primary prey of B. longimanus and drive B. longimanus dynamics. The
second scenario assumes that B. longimanus feed on other prey as well as D. mendotae
and that with D. mendotae migration to avoid B. longimanus predation, that other prey
primarily drive B. longimanus densities. The model also allows no B. longimanus
predation as a comparison to determine the overall effect of B. longimanus predation on
D. mendotae.

The model is composed of a loop that repeats every time step, determining how
D. mendotae and B. longimanus grow, reproduce and die (Figure 3.1). When the model
is run with the data-fixed B. longimanus population, only D. mendotae grow, reproduce

and die, B. longimanus population densities are set to natural levels as described below.

54

Specifically, there are eight steps that determine population dynamics and are repeated
each time step (Figure 3.1, each step described in more detail below):

1. determine adaptive vertical migration depth for D. mendotae

2. distribute D. mendotae around this depth

3. determine how many new D. mendotae are born based on distribution and

current densities

A

. B. longimanus catch and eat D. mendotae

5. B. longimanus grow and reproduce

6. B. longimanus die of other causes (background death and old age)

7. D. mendotae die of other causes (background death and old age)

8. B. longimanus population is redistributed
When the B. longimanus population is set to natural densities the densities are treated as a
time-dependent model input and read into the model and the main loop consists of steps
1, 2, 3, 4, and 7 from above (Figure 3.1). When there is no predator, the B. longimanus
population is set to zero and D. mendotae die due to other causes, such as background
death. Each iteration, the model outputs data on D. mendotae and B. longimanus
population size as well as how many D. mendotae are born and eaten by B. longimanus,
these data are used to quantify the net, non-consumptive and consumptive effect of B.
longimanus on D. mendotae (Figure 3.1). The ongoing model inputs (temperature
gradient, light levels, light attenuation and under some conditions, B. longimanus density)
are based on field data and vary over time, they are read into the model main loop each
time step (figure 3.1). The initial model inputs (population densities and distributions)

are only entered into the model once, before the first iteration. The initial population

55

 

densities and distributions are used to populate the model with individual D. mendotae
and B. longimanus. After the first iteration, population densities and distributions are
determined by model dynamics (Figure 3.1). All portions of the model will be described
in greater detail below. ’

Describing the different steps of the model

D. mendotae adaptive vertical migration depth

Each time step, the model chooses one adaptive vertical migration depth by
maximizing a fitness function that gives projected population growth rates at each depth.

1 CIA/0(2)
ND df

 

g(Z) = = 13(2) - d (Z)

As in the previous chapter, b(z) (equation 2.11) is per-capita birth rate as a function and
d(z) (equation 2.21 ) is per-capita predation rate as a function of depth. The 60 m water
column in this model is divided into 1m3 intervals. Choosing the adaptive vertical
migration depth requires calculating g(z) for the center of each interval and comparing to
find the highest value.

D. mendotae distribution

After an adaptive depth is chosen, D. mendotae are distributed around this depth in a
normal distribution with u = current adaptive migration depth and o = 4.6 m and 7.2 m
for day and night respectively. These values for 6 were obtained from non-linear
regressions of field data (Rv2.5. l , nonlinear regression, p<0.001, data from Pangle et al.

2007) in order to emulate field distributional patterns.

56

D. mendotae reproduce

Each iteration the model determines how many D. mendotae are born and add the
new individuals to the population. The per-capita birth rate, b(z) is determined for each

depth using the egg-ratio method described in chapter 2 (equations 2.8 and 2.9).

b(z) =1n(%,D +1)[0.00041T2 + 0.0108T — 00163]

Where b(z) is per-capita birth rate, E/N is the egg-ratio and T is water temperature.
Information on the birth rate as a function of depth, b(z) is combined with the current
distribution of D. mendotae to determine how many new D. mendotae are born

B(Z) = N D (Z)b(Z)

Next, B(z) is summed over the entire water column to determine how many new D.

mendotae to add to the population
60 E 2

B = N In +1 [00004]?" + 0.0108T — 0.0163]
20 0(2) (AD >

The new D. mendotae individuals are added to the population, and distributed following a
normal distribution with the same mean and variation as the present population.

Predation by B. longimanus

The number of D. mendotae eaten per iteration by each B. longimanus is a
function of depth that depends on light levels and the number of D. mendotae and B.
longimanus at a particular depth. For each lm3 cell, of depth 2, the number of D.
mendotae eaten, D(z), is determined by an attack rate, which is the number of D.
mendotae eaten per B. longimanus. This attack rate is modified as a function of D.
mendotae density (type II functional response, equation 2.15) and then multiplied by the

number of B. longimanus in that particular cell (equation 2.12). This attack rate is

57

modified to decrease as light intensity decreases (equation 2.13). D(z) is divide by D.
mendotae density, ND(z) to get a per capita predation rate (equation 2.14)

( _(Z-flD)2 \

2
2.27 CD e 200

—2K
[11 51081 _ZK ]N3(z) \ j
. + 08

 

 

 

 

 

5.31+ CD e 200

 

 

 

 

 

(“2) = —ND(Z)

To determine if an individual D. mendotae is eaten, a random number from 0 to 1 is
chosen, if it is less than or equal to the predation rate for the current depth of the
individual in question, the individual is eaten and removed from the population. For each
D. mendotae eaten, one individual B. longimanus is randomly chosen from the current
depth to eat the unlucky D. mendotae and this information is stored and used later for B.
longimanus growth and reproduction.

B. longimanus grow and reproduce

B. longimanus growth and reproduction depends on the number of D. mendotae
consumed (as determined above). Allocation energy gained from consuming D. mendotae
towards growth or reproduction is a fimction of the life history stage (i.e. instar) of each
B. longimanus. Using a growth model developed by Yurista and Schulz (1999) the D.
mendotae consumed are converted to grth for instar 1 and 2 individuals and to eggs for
instar 3 individuals. The model utilizes consumption, ingestions efficiency, assimilation

efficiency, respiration and molting.

58

B. longimanus and D. mendotae die of other causes (background death)

Death due to additional factors was included in the model as a background death
rate. Background death is implemented by choosing a random number between 0 and 1
and comparing to the background death rate, if the random number is lower than the
background death rate, the individual dies and is removed from the population. The
background death rates for both D. mendotae and B. longimanus were varied between 0

and 0.1 to determine the effect of including additional mortality on population dynamics.

B. longimanus population is redistributed

To mimic natural distributions of B. longimanus, the population was redistributed
using a normal distribution with p. = 11.6 m and o = 6.6 m , these values were obtained
from a non-linear regression of field data (Rv2.5. l , nonlinear regression, p<0.001, data

from Pangle et al. 2007).

Initial Model Inputs

Initial population densities and distributions were based on field data from Pangle

et al. 2007 (as described above).

Time-dependent Model Inputs

The model inputs that varied over the course of a season and need to change with
each iteration included water temperature, surface light levels, light attenuation rates and
B. longimanus densities (when B. longimanus population densities are set at natural
levels). Water temperature data for the entire season were Obtained from a GLERL
model of temperature profiles for Lake Michigan and data from Pangle et al. 2007.

Surface light levels for each hour were obtained from the GLERL real-time

59

meteorological network for the Muskegon site (2005 and 2006 data). Light attenuation
rates were obtained from Pangle et al. 2007 and varied stochastically throughout the
season. B. longimanus natural densities were set following data from Pothoven et a1.
2001 with low initial densities that grew to peak between late July and September and

then decreased again.

Model Outputs

Each iteration, the model outputs data on D. mendotae and B. longimanus
population size as well as how many D. mendotae are born and how many D. mendotae
are eaten by B. longimanus. These data were used to quantify the net, non-consumptive

and consumptive effect of B. longimanus on D. mendotae (Figure 3.1).

Determining how B. longimanus predation affects D. mendotae population
dynamics metrics

The effect of a predator on prey population dynamics can be measured in terms of
biomass (standing crop) or energy flow (population turnover). I used both of these
methods to determine the effect of B. longimanus predation on D. mendotae population
dynamics. The change in biomass due to predation was determined by taking the
difference between D. mendotae density in B. longimanus absence and B. longimanus
presence. The change in energy flow due to B. longimanus predation was determined by
finding the difference between the daily D. mendotae population density change in B.
longimanus absence and presence.

Determining the net effect

The net effect was measured each time step and is called the instantaneous net

effect. The instantaneous net effect is determined by calculating how large the

60

population of D. mendotae would grow in a single time step without B. longimanus
predation if they remained at the surface. To determine this value, I take the population
density at the beginning of a time step and calculate how many D. mendotae would be
born if all current D. mendotae were to remain at the surface for the time step, I then
subtract the actual population density at the end of the time step (after cost of migration
and direct predation have been factored in).

Determining the consumptive effect

The consumptive effect is the change in population density due to direct
mortality. I isolate the consumptive effect each time step by keeping track of the number
of D. mendotae that are consumed by B. longimanus.

Determining the non-consumptive effect

The non-consumptive effect is the decrease in the number of D. mendotae born
due to the cost of vertical migration. I isolate the non-consumptive effect each time step
by keeping track of the number of D. mendotae that are born and comparing that to the
number of D. mendotae that would have been born if they all were at the surface (i.e.

position in B. longimanus absence).

Results

B. longimanus predation had a large impact on D. mendotae population dynamics
(Figure 3.2). When B. longimanus densities are fixed at natural densities (figure 3.2a), B.
longimanus predation substantially decreases D. mendotae biomass. Comparing D.
mendotae density in the presence and absence of B. longimanus (figure 3.2b) elucidates
the impact of B. longimanus predation in terms of changes in D. mendotae biomass. The
percent decrease in D. mendotae biomass due to predation of B. longimanus is large,

ranging from 90% early in the season to 10% later in the season (figure 3.2c). B.

61

longimanus predation also affects D. mendotae population dynamics through the turnover
rate Of the population, the energy flow (figure 3.2c). The amount of energy flow through
the food web due to turnover in D. mendotae populations changes over the course of the
season, quickly increasing from 10% of the total D. mendotae population to 30% of the
population and the gradually decreasing over the course of the season (figure 3.2c). The
changes D. mendotae biomass and energy flow are mainly comprised of non-
consumptive effects (Figure 3.2c).

The non-consumptive effect of B. longimanus predation on D. mendotae
population growth rate is very large compared to the consumptive effect (figure 3.2d).
Non-consumptive effects make up majority of the net effect of the predator, ranging from
60 to 90 percent of the total effect of B. longimanus predation on D. mendotae population
dynamics (figure 3.3). The model predicts that non-consumptive effects dominate
consumptive effects throughout the entire season, although there are seasonal differences
in the magnitude of both consumptive and non-consumptive effects. Over the course of
the season, both the consumptive and non-consumptive effect change, although the non-
consurnptive effects changes are much larger than the changes in the consumptive effect.

Seasonal changes had a strong influence on the relative magnitude of non-
consurnptive effects and a smaller influence in the relative importance of consumptive
effects. Generally, the net effect of the predator increases initially and then decreases
over the remaining portion of the season (Figure 3.2d). Changes in B. longimanus
density contribute to the observed changes in the non-consumptive effect. The changes
are of small magnitude because the magnitude of the consumptive effects is smaller.

Both, changes in surface water temperatures and B. longimanus density (figure 3.2a)

62

contribute to the changes in non-consumptive effects. Comparing B. longimanus
dynamics (figure 3.2a) and non-consumptive effects (figure 3.2d), it is evident that the
importance of the non-consumptive effect closely follows B. longimanus density.
However, during July, August and September, there appears to be another factor
influencing D. mendotae population dynamics. The non-consumptive effect curve (figure
3.2d) is comparatively less steep than the B. longimanus density curve (figure 3.2d)
during the earlier portion of the season, however it matches the shape of the temperature
curve very well. The model predicts that during the beginning of the season, surface
water temperature plays a large role in the interaction between B. longimanus and D.
mendotae and that the influence of B. longimanus density increases as the season
progresses.

Interestingly, the model predicts two distinct, consistent patterns of population
dynamics for both D. mendotae and B. longimanus densities when a responsive B.
longimanus population is included. The first pattern of dynamics is remarkably similar to
the fixed-data case described above (figure 3.3). After an initial increase, D. mendotae
density remains fairly constant and gradually begins to decline (figure 3b). B.
longimanus densities increase gradually for the first month, slowly peak and then begin to
decrease (figure 3a) showing very similar dynamics to natural B. longimanus density data
(figure 3.2a) although with some small fluctuations in density. The second pattern is
drastically different, B. longimanus densities increase quickly over the first month
causing a crash in the D. mendotae population (figure 3.4b), which leads to a Sharp peak
and large decrease in B. longimanus densities (figure 3.4a). Interestingly, as B.

longimanus population density increases and D. mendotae population densities begin to

63

fall, consumptive effects become more important that non-consumptive effects for a brief
period of time (figure 3.4d). The initial rate at which the B. longimanus population grows
seems to determine dynamics. Varying both B. longimanus and D. mendotae background
death rates led to these different patterns emerging. For low B. longimanus background
death rates, or low D. mendotae background death rates, initial B. longimanus population
growth rates were high leading to dynamics that followed the second pattern. For high B.
longimanus background death rates, or high D. mendotae background death rates, where
initial B. longimanus population grth was much slower, population growth followed
the first pattern (similar to fixed-data results).

B. longimanus and D. mendotae population dynamics changed drastically with
changes in B. longimanus background death rate. As B. longimanus background death
rates decrease B. longimanus populations grow at a fast rate until a threshold density
(peak in figure 3.4,b) after which the D. mendotae population crashes down to refuge
levels and then begins to grow again once B. longimanus population density has
decreased. At lower B. longimanus background death rates, where B. longimanus has
higher growth rates, predation has a drastic effect on D. mendotae dynamics that is made
up almost entirely of changes in D. mendotae population density (with the percentage
decreases is density due to predation ranging from 30% to almost 100% under some
conditions).

Under climate change predictions water temperatures will increase and fall
turnover will occur later (Brooks and Zastrow, 2002), extending the interval for which
non—consumptive effects are important (Figure 3.5). We do not see changes in adaptive

vertical migration behavior and consequently large non-linear changes in the magnitude

64

of the non-consumptive effect at water temperatures above 25°C, as was predicted in the
last chapter (Figure 3.5). This seems to be due to changing B. longimanus densities that
are tied to D. mendotae densities. If B. longimanus densities are fixed at natural levels
under climate change predictions we see a decrease in non-consumptive effect in July,

August and September when there are some days that reach above 25°C (Figure 3.6).

Discussion

My results indicate that non-consumptive effects are an important component of
changes in seasonal dynamics. The changes observed at this multigenerational timescale
were seen in D. mendotae density and turnover rates. Population dynamics were
influenced by both B. longimanus densities and surface water temperature. Surface water
temperature had the greatest impact early in the season and the influence of B.
longimanus densities increased over time. Non-consumptive effects were very large and
made up the majority of the net effect of the predator. As the season progressed and the
hypolimnion cooled, non-consumptive effects became less important, but still dominated
consumptive effects. Increasing surface temperature and the period of thermal
stratification (as is predicted for Lake Michigan under climate change) leads to an
increase in the magnitude and importance of non-consumptive effects.

My results point out that non-consumptive effects are very important on seasonal,
multigenerational time-scales. Other studies have shown non-consumptive effects to be
important in short-term experiments and field studies up to time-frames of a single
generation (Nelson 2004, Peckarsky 1993). Theory predicts that non-consumptive effects
are likely to be important in the short term (Abrams 1984, Peacor and Werner 2004) or to

long term population stability (Ives and Dodson 1987, Abrams 1995), and to food web

65

properties such as susceptibility to species invasions (Peacor et a1 2006, Sih et al. 1985).
However, these theoretical studies used made up functional responses and so it is difficult
to understand how they might apply in real systems. My results build on these previous
results by using field and laboratory data to make predictions about the longer-term
importance Of non-consumptive effects. It may be impractical to attempt to determine if '
non-consumptive effects are important over longer time scales in many systems through
long term experiments since many systems of interest have species with long generation
times making long-term experimentation prohibitive. However, as I have shown in this
chapter, it is possible to make longer-term predictions of the importance of non-
consumptive effects if one understands how biotic and abiotic environmental conditions
affect the importance of non-consumptive effects and how these conditions vary naturally
over the time frame of interest. Additionally, the D. mendotae, B. longimanus system
does not have this generational time limitation, thus multi-generational dynamics can be
observed in a single season. Further work should test the predictions made in this chapter
in the field.

Previous empirical studies have shown non-consumptive effects to be important
relative to consumptive effects (reviewed in Peacor and Werner 2004 and Preisser et al.
2005) and theoretical work has shown non-consumptive effects to have important
implications for long-term community stability (Abrams 1995, Ives and Dobson 1987).
These studies have helped to answer questions about the importance of non-consumptive
effects. However, these are not the scales at which management decisions are made. For
managers to understand that non-consumptive effects are very likely to be important in

their systems, it is necessary to show that non-consumptive effects are important on

66

 

scales at which practical management decisions are made, on seasonal dynamics. Our
results show that non-consumptive effects have more of an impact on seasonal dynamics
than consumptive effects.

Results from this model support some of the predictions made in the last chapter.
Results reported in the previous chapter Show a strong non-linear relationship between
surface temperature and non-consumptive effects for high temperatures that is affected by
B. longimanus density. As B. longimanus density increases, the threshold temperature
increases to a level above what is predicted with climate change and we no longer see a
shift in behavior; non-consumptive effects dominate for the whole summer and fall. Our
model probably over estimates the dependence of B. longimanus on D. mendotae since it
dose not include any other prey. In natural communities, B. longimanus prey on other
species and fluctuations in their population density do not follow D. mendotae population
density fluctuation as closely. Thus, it still might be possible for the temperature
threshold to fall within climate change predictions. Whether we see days in July and
August in the future where non-consumptive effects are non-existent because D.
mendotae do not migrate will depend on B. longimanus densities as well as surface
temperature.

Our model predicts that changes in the Great Lakes due to climate change could ’
increase the time period over which non-consumptive effects are important. Climate
change scenarios indicate shorter winters with longer periods of thermal stratification
with some predictions placing fall turnover up to 2 months later than current dates
(Brooks and Zastrow, 2002). If stratification were to persist longer into the winter it

would add to the period of time that non-consumptive effects dominate. Increasing

67

temperatures over the season also increase the magnitude of the non-consumptive effects
for later months. It is still unclear what might happen in July and August. ‘If
temperatures are consistently above the temperature threshold, then D. mendotae should
remain at the surface and non-consumptive effect will be non-existent. It is unclear if this
will happen, since the threshold value depends on B. longimanus density as B.
longimanus density increases, the threshold for the behavior shift also increases. My
model indicates that observing this behavioral transition in the field will likely depend on
how quickly B. longimanus densities respond to increases in D. mendotae densities. If
there are days with low B. longimanus densities and high temperatures then we expect to
see D. mendotae remain at the surface and non-existent non-consumptive effects. Since
this behavior shift is dependent on B. longimanus densities, we might expect to see
particular days where D. mendotae do not migrate, however, we probably will not
Observe long periods where D. mendotae do not migrate. This would lead to non-
consumptive effects being less important overall during late summer but still dominating
consumptive effects, transitioning to a longer period where non-consumptive effects
dominate due to longer periods of stratification but become increasing less important.
However, I am making an assumption here that the mechanism by which D. mendotae
respond to B. longimanus presence is also attuned to temperature and predator density.
This is probably not the case so we probably will not see D. mendotae remaining at the
surface on hot August days. More work is needed to clarify the non-consumptive effects
given the simultaneous operation of different mechanisms if we integrate these

temperature results over time.

68

 

 

Population Dynamics Model Flow

Model Outputs: Time-dependent Model Initial Model Inputs
D. mendotae optimal depth, Inputs (biotic factors)
D. mendotae density, Light levels, light Initial B.
# of D. mendotae born, attenuation rate, surface longimanus
# of D. mendotae eaten by temperature and distribution and
B. longimanus, and temperature gradient density, and initial
B. longimanus density (B. longimanus density) D. mendotae density

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

A
D. mendotae choose
optimal depth
Redistribute B. Distribute D.
longimanus (randomly, mendotae around
based on normal dist.) optimal depth
1 I
B. longimanus die due D ' mendotae
to background death reproduce based on
water temperature
1 t
D. mendotae die due B. longimanus catch
to background death and eat D. mendotae
\ B_ longimanus grow /
and reproduce
(bioenergetics model)

 

 

 

 

Figure 3.1: Flow diagram for Population Dynamic Model

The general flow for the Population Dynamic Model describing the model inputs (green),
the basic model loop (blue) and the outputs (red). The main loop for the model version
that includes a B. longimanus population that is dynamically tied to D. mendotae follows
the black arrows and includes the rectangles with white and gray backgrounds. The main
loop for the model version where B. longimanus densities are fixed at natural levels
includes only the rectangles with white backgrounds (skip over the gray backgrounds).
Initial model inputs include D. mendotae density, D. mendotae distribution, B.
longimanus density and distribution, these are parameters that are initially supplied to the
model, but for all subsequent model loops are provided by the previous iteration. Tirne-
dependent model inputs are abiotic parameters that are used each iteration and may vary
with time, such as surface light levels, light attenuation, temperature profile and with the
fixed B. longimanus populations, B. longimanus density. Each iteration of the model the
main loop of the model runs and D. mendotae and B. longimanus grow and reproduce
and B. longimanus density, D. mendotae density, optimal depth, and number of D.
mendotae births and deaths are outputted.

69

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

C
«n r: ‘
525 we l
as =50 1
Eu 5 20° 3
t E a 150 l
g“ 3 100
m 8.100- 50 . _
7 ..7A 7 7 7 7 7 . 7, 7 7 7 .7 7 7 7 L.
E c D. mendotae dynamics wrth aindrwlthout B. longimanus am
. E g o D. n’eriflfléflersityw‘thou B. lorgirrans
E,- E 350000 ___________ _7, _______ 7 D. nerdotaedersityw’th B. lorgirrarus 5
o 3 300000 , , 7...; ~ 4 ANetefieddB-Wmmpmmeflner
‘3 ‘5 250000 7 , T. . . .. 7 T.... T 77 TMt
.u 3 m . x 7‘: . 77 7 _7 77777 7 7 ”31L"“Fifi-“5:."1‘17;:.~::.::,;::prawn-314,1.1.6: “:3”.
E 5 150000 7 ,‘7 7 7 7 7 , 7 7 ,
E :3 100000 lit 15.2 ~ 77 7, 7» 7 — ,7 ,7 , 7 777
d 0 50000 if 9
U, J 7 7
q C Effect of B. longimanus predation on D. mendotae 320
dynamics (population density and turnover rate) 7 7 ,
d 1 A %irueaseifD.rrerdotaetum1erdteto .
.E a .. a l B. lorgirrarls predation l;
3&3 0‘8 . 4'. 7 if" i . %deaeaseinD.merr:lotaedersitydLeto ‘i
g a; 0-6 1’ g ' T 7.7.... lg B. lor‘girrgns predafionir , "71::
'5 E C 0.4 '. 7 77777 77 ~ ,7 ~ . 7 77..
.\° 02
770 l
d Comparing the consumptive and non-consumptive effects 32°
C Julamay I . 7 7 7*
l "3% ‘
o E—
l 2.5 3
:1 o 5
2% l
3‘3 .
g 5: i
2 5.2 l
, One-o ‘ ‘ ‘
8 3 180 200 20 240 260 280 300 320
” Julian Day ‘

 

Figure 3.2: D. mendotae population dynamics with fixed B. longimanus population
based on field data.

Figure 3.2a shows B. longimanus population densities. Figure 3.2b shows D. mendotae
population dynamics with and without B. longimanus predation and the cumulative net
effect of B. longimanus predation on D. mendotae population density. Figure 3.2c shows
the effect of B. longimanus predation on D. mendotae density and turnover rate. Figure
3.2d shows the instantaneous net effect of the predator, the non-consumptive effect and
the consumptive effect.

70

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Proporfionofthemteffectduetooonsumtiveandnon-
eonsmptiveeffects
1 ......-...---...._.,.7 77-7- 7.- -7
0.9 _H’WWT’TW 7
i 0.8~ ——:— ’ 7 7 .. ’won»
! 07+ 937 _7 7 7 _ L 7 ___77_ __
. . . _
g 05 {7 7 7 . 7 7 7 _7 .7. 7 ._ oPtopufionmnoomurpfiwefl‘ad
‘5 05777 ___777 77, 7 77 , 777__.____7 7 __{_f Proporlimooremptixeeflect
E 0.4 7_ 77 -7 77 7 77.77-. _7 77 7. 7__ 77
I
0.3 171.7777 -——- —--
33:7? 77 7 _ e _ ::'T-.~«3-:~7W- '
o - . .
1&3 2C!) 220 240 260 2&3 30) 320
.uhnmy

 

 

 

Figure 3.3: Breaking down the net effect of the predator into the proportion due to non-
consumptive effects and the proportion due to consumptive effects.

71

 

1
1
l

B. longimanus dynamics

 

 

 

 

 

 

   

 

 

 

 

a
C
to :
§§§ 6000
.83 3 5000 7 77 :7 7 7 7 7
Em 3 4000 77 7 «o 77 77 7 7 7 7
c E a o
g3g30007 7 7: . 77 7 77 7
l“tn-'15 2000 7 7 o’ w 7 7 7 7 77 7 7
1000 7 7 7 77°.7 7 77 7 77 7 7
7, L 7 , 7 \.....~
b D. mendotae dynamics with and without D. longimanus
NE E o D. mendotae density without B. longimanus
.. 3 D. mendotae density with B. longimanus ‘ 1
1 S. 3 400000 -7 777-7-777-777- A net effect ofB. longimanus on D. mendotae overtime 1—— ‘
° L 77 7 7 7777 - 77- 77 7 77 4
I ‘3 #3 300000 77 77 7 7 7
1 ‘5 3 A l
e = 200000 A 7
1 o 2 A
5_ *5 100000 7 7
1 Q 3 l
l0 E ect of B. longimanus predation on D. mendotae 1

dynamics (population density and turnover rate)

 

 

 

Q 1

g 0

:8 a; 0.8 7 7- ‘0 ° 71 07,7 7 _ 77 15-7 . .7 1

D81: . . . A: Increase In D. mendotae . 1

==E 06 'A ,7 .. . ' tumver . ., ,7" ,,, , 1

I! 0 :I o . 1 o 1

5 E c 0'4 .f ‘ o % decrease in D. mendotae‘ "'. ,, ’ r l

32 0.2 A 7 7 7:71 Wdensity 7 1 A fi,: 1
D , - _ . . 7 . . , .

 

 

d Comparing the consumptive and non-consumptive effects 1
E

 

 

 

 

._ 7,- . B
O 1 .

E5 20000 I Instantaneous t‘let effect W!
as 8 00000 -7 .1. , ‘ :0mns2mphgéfied : 1
Egg 80000 7 _-' 777 “sump V” 77 7 1

l 5.8; 60000 777 . , W
C, t!— .z--- 4

‘ a“, 0 40000 7 77 7 777 7 7
N E = i

1 2 __9 20000 . ‘ 7 , ,.
8°23 Mm... M

“ 180 200 220 240 260 280 300 320
Julian Day

Figure 3.4: D. mendotae population dynamics with a responsive, slow growing, B.
longimanus population.

Fiugre 4a shows B. longimanus population dynamics. Figure 3.4b shows D. mendotae
population dynamics with and without B. longimanus predation and the cumulative net
effect of B. longimanus predation on D. mendotae population. Figure 3.4c shows the
effect of B. longimanus predation on D. mendotae density and turnover rate. Figure 3.4d
shows the instantaneous net effect of the predator, the consumptive effect and the non-
consumptive effect.

72

B. longimanus dynamics

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

C

1 ‘0:

~ :35 .0. 1

‘ E88 4‘13 ’ i w i *‘i ’*r’lr 3
.-m ‘

1 guinea 77 7‘" 7 ~A~f\ 7 77777777 l
.51, .m u..- WW 1
mD-‘a 1007 7 7 7 777 7 77 7 777 7 7 7 7777- 777 77 1

1

. , 7 7, 41

1 E c D. mendotae dynamics with and without B. longimanus1 32° 1

E E , Julia» Dy

‘ h 3 o D. nerdotaedersitywittm B. lorgirrans

1 a 8 350000 WM” _ -_ ..._._‘ D. mendotaedersitywithB. lorgirrans 1__ l

a g 3000007 7 ANeteflbctofB. lorgIIransonDrrerdolaeovertIn'E« 7

l :g; ; 250000 7 __ .. 1

1 c 5: 2000007 7 7 7 1

a: 0 150000 77 7 7 7 7 77 7

1E=3100000 7777 77 77777777 1

Q‘ a 50000 . 7 7 l

111*, f1 ,4 , m , 7’7,#7 7

1 C Effect of B. longimanus predation on D. mendotae 320

dynamics (population density and turnover rate)

l d 1 WWW“_"WWWWMWWJ . %imeaseifD.rrerdotaetum\erdLeto l
g: 0. N ‘ B. lorgin'arls predation 1
:50: 0.8 '77 77 77 71° 7
mg: . 0, o/odecreaseinD nerrblaedersitydueto1

‘ 5:; 06—: 7.. * fl B lorgm'arlspredaton 1 *
fig: Q4 t W, o ,,,___.7 ,, "D, :V :. :,;_

1 .\°

d Comparing the consumptive and non-consumptive effects 32° 1

1 all“ ll-My 7 ,7
‘5 g nNetEfiéd 1
gwg% . Nari-corsurpfivedfectH 1
.50 $3. 5"” F ' effect .

1 gm; mo _.- tr...u"{...fl"m.n&.,mr ,, Mm1 1

1 2%“; 30000 73‘ all M. 777 77 7 77 l

1 "gas zoom 7 77 7 7 1

1 5 ...°_ 0 1 . . . .. ’ 1
8°11
o 3 180 200 220 240 260 230 300 320 1

‘1 JulianDay 1

 

Figure 3.5: D. mendotae population dynamics with a responsive, fast growing, B.
longimanus population.

Figure 3.5a shows B. longimanus population dynamics. Figure 3.5b shows D. mendotae
population dynamics with and without B. longimanus predation and the cumulative net
effect of B. longimanus predation on D. mendotae population density. Figure 3.5c shows
the effect of B. longimanus predation on D. mendotae density and turnover rate. Figure
3.5d shows the instantaneous net effect of the predator, the consumptive effect and the
non-consumptive effect.

73

 

Coranptiveeffectofprledamr

 

  
   
 
  

 

 

 

1.5 7000
5!
:1 u . 1
§§m 7 7 1Ifttueclirrae1
1§§m Barrertclirrae1
$754000
6:;
923000
113,3
g 2030
E
1 1(1)!)
0
My August Seria'rber October Marta 1
7 7 7 7 7 7 71
Non-oa‘mnptiveefiectofpredamr 1
. 1
‘E 50000
450m 7 -77 77-
1§ Em i Alfilu'eclirrde 1
O
1%fim 7E1611mclin‘ag17
gem " ,7
$525000
$320000
0
dg15000 7 7
.-E"1000077
13 5000
3 0

 

 

Jay Angst Septerrber October Natenber

Bi , , ._ ,1

Figure 3.6: Comparing consumptive and non-consumptive effects for present and future
climate over a season.

Both the consumptive and non-consumptive effects are measured in terms of decrease in
D. mendotae population density. Each bar represents an average value for the month.
Non-consumptive effects are much larger than consumptive effects, although the
magnitude of consumptive effects decreases overtime.

 

74

1 Effect of dynamic vs. fixed B. longimanus density on magnitude of
non-consumptive effects

 

50000 1 . . . . 1 1
I B. longimanus dynamics primarily

45000 " " ' ’ ’ i ’ i '1 driven by D. mendotae ‘

W ~ 7 7 ~ r_'] B. longimanus population fixed at 1

,1 natural levels

   

 

density due to cost of migration

   

 

 

Decrease in D. mendotae Population

July August September October November

Figure 3.7: Comparing the non—consumptive effect for future climate conditions over a
season with dynamic B. longimanus population densities or forced at natural densities.
The non-consumptive effects are measured in terms of decrease in D. mendotae
population density. Each bar represents an average value for the month. Non-
consumptive effects are much larger for dynamic B. longimanus populations that for the
populations set at natural densities and not dynamically tied to D. mendotae. These
differences are due to D. mendotae not migrating on some days in July and August when
temperatures are above 25°C and non-consumptive effects being zero.

 

75

CHAPTER 4
SUMMARY AND IMPLICATIONS

Summary of thesis
Motivation for thesis research

As I began my thesis, there was a substantial body of research showing that a
large variety of prey modify their phenotypes in an attempt to minimize predation
(reviewed in Lima and Dill 1990, Lima 1998, Tollrain and Harvell 1999, Agrawal 2001).
A growing number of studies have shown that when prey modify their phenotype, they
incur a cost that effects their fitness through changes in growth rate, fecundity, etc. and
that these costs can be as large or larger than the fitness cost of direct mortality (reviewed
in Peacor and Werner 2004a and Preisser et al. 2005). These trait modifications, while
maximizing overall fitness, shift the cost from mortality due to predation (consumptive
effect) to other costs (non-consumptive effects) such as growth and fecundity. Many
studies have quantified these costs on fitness correlates (such as individual growth,
fecundity, population growth rate) and found non-consumptive effects to be large in a
variety of systems (both aquatic and terrestrial).

Ecologists have made great strides towards understanding the importance of non-
consumptive effects, but many questions still remain regarding the general importance of
non-consumptive effects. Particularly, little is know about how interactions with the
environment and with other species might influence trait expression and the relative
importance of non-consumptive effects. We expect plastic traits to change as a result of
changes in the environment. If the cost of trait expression is not constant, we would also

expect the relative contribution of consumptive and non-consumptive effects to vary

76

 

across environmental conditions as well. However, very few studies have attempted to
determine how environmental variability affects the relative importance of consumptive
and non-consumptive effects.

The majority of studies that have quantified non-consumptive effects so far, have
measured them in terms of fitness correlates. We know that prey trait modification in
response to predators has a large impact on prey growth and fecundity and in some more
recent studies, population growth rates. However, we do not know how these
instantaneous effects translate into longer-term, multi-generational population dynamics.
These longer-term dynamics are the scales at which management decisions are made. If
we truly want to understand how much of an impact prey trait modification has on
communities, we need to understand how these instantaneous results translate to longer
time scales.

The effect of environmental variation

I used the models presented in chapter 2 to determine how environmental
conditions affect trait expression and the relative importance of consumptive and non-
consumptive effects. This required a thorough knowledge of how birth and death rates
depend on trait expression, and how this relationship is affected by changes in both biotic
and abiotic environmental parameters. I used a zooplankton predator-prey system (B.
longimanus and D. mendotae). In this system the trait in question was vertical migration.
The environmental conditions I investigated are B. longimanus density, D. mendotae
density, B. longimanus distribution, D. mendotae distribution, predation risk from
additional predators, competition, surface light availability, light attenuation, surface

water temperature and the thermal profile of Lake Michigan.

77

My results show that trait expression and the importance of trait-mediated
interactions are highly dependent on environmental conditions. The models predict large
changes in adaptive vertical migration depth and the relative contributions of
consumptive and non-consumptive effects for some variables (surface temperature,
temperature gradient, B. longimanus density and presence of additional predators) and
not others (D. mendotae density, competition, surface light levels and light attenuation
rate). Some of the changes we observed were gradual. For example, seasonal changes in
the temperature gradient caused a gradual decrease in the magnitude of non-consumptive
effects. Other variables, such as B. longimanus density and surface water temperatures
caused strong, nonlinear changes. The variation in impact on the relative importance of
non-consumptive effects can be explained when we examine how the factor affects birth
and death magnitudes as a function of depth, and in turn how this affects the adaptive
trait change.

Translating fitness correlate results to seasonal population dynamics

To translate previous findings on consumptive and non-consumptive effects into
longer-term, multigenerational changes in p0pulation dynamics, I used a population
dynamic model presented in chapter 3. I used the Daphm'a mendotae, Bythotrephes
longimanus system to explore how changes in the pelagic environment (i.e., changes in
abiotic conditions such as temperature and light availability along with changes in biotic
conditions such as B. longimanus density, D. mendotae density, B. longimanus
distribution, D. mendotae distribution, predation risk from additional predators and
competition) over the course of a season influence D. mendotae population dynamics.

The focus was on seasonal dynamics because it is a scale at which management decisions

78

are made and in this system a season incorporates environmental variability and
encompasses many generations of both D. mendotae and B. longimanus.

My results indicate that non-consumptive effects are an important component of
changes in seasonal dynamics. The changes we observed at this multi-generational
timescale were mostly due to seasonal changes in water temperature and B. longimanus
densities. As the season progressed and the hypolimnion cools, non-consumptive effects
became less important. As the season progressed, the influence of B. longimanus
densities on D. mendotae dynamics increased. As B. longimanus densities decrease and
surface temperature decreases later in the season, the effect of B. longimanus densities
becomes more important. This study shows that it is possible to make longer-term
predictions on the importance of non-consumptive effects if one understands how biotic
and abiotic environmental conditions affect the importance of non-consumptive effects
and how these conditions vary naturally over the time frame of interest.

Needs for further study

The apparent universality of prey modifying their phenotype to minimize the
effect of predators and the dependence of expression of phenotypic traits on
environmental conditions, suggests that the relative importance of consumptive and non-
consumptive effects will generally depend on the environment. However, aside from
some preliminary studies and this in depth study of the B. longimanus, D. mendotae
system, there is not a lot of evidence to support this idea. Therefore, the analysis
performed in my study needs to be repeated in additional systems. I put a large effort
into applying my approach to additional systems. However, was unable to find the

system in which sufficient data were available, even when combining data from multiple

79

laboratories and contacting authors for unpublished data. This suggests that whereas
many researchers are demonstrating the short term effects of non-consumptive effects,
they are not collecting the information required to make longer term inferences. In order
to apply the analysis performed in my study to other systems the following information is
required

1. birth and death rate dependence on trait expression

2. how these birth and death rate functions vary with changes in environmental

conditions

3. extensive data on how environmental conditions vary over time
I reviewed studies that have looked at the impact of some environmental conditions
(predators, resources and competitors) and studies that attempted to measure both
consumptive and non-consumptive effects in an attempt to repeat this study and
determine if my main result, the dependence of the relative importance of non-
consumptive and consumptive effects on the environment, is true for other systems.
Through this literature search, I learned that the necessary data are not available for many
systems and was unable to find the data needed to repeat such a study. In particular,
there is little understanding of how birth and death rates change as a function of trait
expression.

My study indicates that further work on the B. longimanus, D. mendotae system
would improve our understanding of the impact variation in trait expression and non-
consumptive effects has on other species, particularly important fisheries species. Theory
predicts that including plastic trait expression and the resulting cost in terms of fitness

can change the relative strengths of direct and indirect interactions (Abrams 1995) and

80

 

many empirical studies have shown trait—mediated indirect interactions to be important
(reviewed in Werner and Peacor 2003). Previous studies have shown that B. longimanus
are having a large impact on the community in Lake Michigan and understanding the
mechanisms behind the changes that are occurring in the community is very important
(Lehman and Caceres 1993; Vanderploeg et al. 2002, Barbiero and Tuchman 2004). D.
mendotae are an important food source for young piscivorous fish and for planktivorous
fish. If non-consumptive effects are important as my study indicates, then the reduction
in D. mendotae population growth rate is due to lower birth rates rather than consumption
by B. longimanus. This means that energy is not flowing up the food web into B.
longimanus, rather the resources that would be consumed by D. mendotae may be .
available to competitors that may or may not be good food sources for fish. We need to
understand how these large non-consumptive effects propagate through the Lake
Michigan community. We also need to understand that these non-consumptive effects
are large, but change over the course of the season and that this can influence the rest of

the community.

Implications of thesis work

The implications of the work presented in this thesis are far reaching. It is now
clear that changes in phenotypic trait expression due to environmental conditions have a
large impact on predator-prey interactions, including prey population dynamics. This
work has implications for community ecology, aquatic ecology, particularly
understanding the Lake Michigan pelagic community and understanding how climate

change and invasive predators will likely impact aquatic communities.

81

 

It is clear that researchers need to consider that the importance of non-
consumptive effects depends on environmental conditions when deciding whether results
from specific systems regarding the importance of non-consumptive effects will hold for
their particular set of conditions. The work presented in this thesis has implications for
generalizing from empirical studies that show important non-consumptive effects. My
work has shown that the relative importance of consmnptive and non-consumptive effects
depends on environment conditions making it clear that environmental conditions should
be considered when trying to extend results within a system to different environmental 8
conditions (temporally or spatially) or between similar systems. An understanding of the
effects of the environment will be particularly important when considering past and
future changes to ecosystems from invasive species and climate change.

Understanding how trait expression and non-consumptive effects can vary with
environmental conditions has implications for ecological communities in terms of trait-
mediated indirect interactions. Theory indicates that including plastic trait expression
and the resulting cost in terms of fitness can change the relative strengths of direct and
indirect interactions (Abrams 1995), indicating that changes in trait expression due to
changes in environmental conditions may be very important. Explicitly including
environmental factors in addition to predation risk may change how we understand
communities. These observed changes in consumptive and non-consumptive effects as a
function of the environment do not only affect the predator and prey populations, but also
propagate through the community, as indirect interactions, affecting additional prey,
competitors and predators of our two focal species. Thus, plastic traits and non-

consumptive effects may have large impacts on biological communities.

82

 

My results indicate that these large instantaneous non-consumptive effects are
important on longer time scales and do affect population dynamics. In fact, non-
consurnptive effects make up a large portion of the effect of B. longimanus on D.
mendotae population dynamics. The contribution of non-consumptive effects to changes
in prey dynamics indicates that managers need to seriously consider the importance of
changes in trait expression and the resulting non-consumptive effects in interactions
between species. Non-consumptive effects are large enough that they have a substantial
impact on prey population dynamics through both changes in prey biomass and energy
flow. Potentially, non-consumptive effects on prey population dynamics could
drastically impact the community through trait-mediated indirect interactions.

Lastly, the work I have presented in my thesis adds to the growing understanding
that adaptive behavior needs to be included in a general understanding of biological
communities. Specifically, this thesis adds to a growing body of research that shows that
adaptive changes in trait expression have a profound effect on communities. We, as
ecologists, cannot hope to understand ecological communities without including dynamic
traits. It is becoming clear that in order to understand biological communities, the fields
of ecology, evolution and behavior must come together. The body of research on the
importance of adaptive trait expression and non-consumptive effects indicates that
ecological models need to include plastic traits or risk severely underestimating the

importance of species interactions and the mechanisms that underlie these interactions.

83

LITERATURE CITED

Abrams, P. A. 1982. Functional responses of optimal foragers. American Naturalist
1202382-390.

Abrams, P. A. 1984. Foraging time optimization and interactions in food webs.
American Naturalist 124:80-96.

Abrams, P. A. 1987. Indirect interactions between species that share a predator: varieties
of indirect effects. Pp. 38-54. in Kerfoot, WC. and A. Sih, eds. Predation: direct and
indirect impacts on aquatic communities. University Press of New England. Hanover.

Abrams, RA. 1993. Why predation rate should not be proportional to predator density.
Ecology, 74:726-733.

Abrams, RA. 1995. Implications of dynamically variable traits for identifying,
classifying and measuring direct and indirect effects in ecological communities.
American Naturalist, 146:1 12-134.

Abrams, PA. 1997. Variability and adaptive behavior: implications for interactions
between stream organisms. Journal of the North American Benthological Society,
1 6:3 5 8-3 74.

Agrawal, AA. 2001. Phenotypic plasticity in the interactions and evolution of species.
Science 294: 321—326.

Barbiero RP. and ML. Tuchman. 2004. Changes in the crustacean communities of
Lakes Michigan, Huron, and Erie following the invasion of the predatory cladoceran
Bythotrephes longimanus. Canadian Journal of Fisheries and Aquatic Sciences, 61,
2111—2125.

Boeing, W.J., Wissel, B. and C. W. Ramcharan. 2005. Costs and benefits of Daphnia
defense against Chaoborus in nature. Canadian Journal of Fisheries and Aquatic
Sciences, 62: 1286—1294.

Bolker, B., Holyoak, M., Krivan, V., Rowe, L., and O. Schmitz. 2003. Connecting
theoretical and empirical studies of trait-mediated interactions. Ecology, 8421101-
1114.

Boscarino, B.T., Rudstam, L.G., Mata, 8., Gal, G., Johannsson, O.E., and EL. Mills.
2007. The effects of temperature and predator-prey interactions on the migration

behavior and vertical distribution of Mysis relicta. Limnology and Oceanography,
52:1599-1623.

84

 

Brooks, AS, and J.C. Zastrow. 2002. The Potential Influence of Climate Change on
Offshore Primary Production in Lake Michigan. Journal of Great Lake Resources,
28:597-607.

Caro, TM. 2005. Antipredator Defenses in Birds and Mammals. Chicago: University of
Chicago Press.

Cresswell, W. 2008. Non-lethal effects of predation on birds. Ibis, 150:3-17.

Dawidowicz, P., and C. J. Loose. 1992. Cost of swimming by Daphnia during diel
vertical migration. Limnology and Oceanography 37:665—669.

Dawidowicz, P. and M. Wielanier. 2004. Cost of predator avoidance reduce competitive
ability of Daphnia. Hydobiologia 526:165-169.

Dill, L.M. 1987. Animal decision making and its ecological consequences: the future of
aquatic ecology and behavior. Canadian Journal of Zoology, 652803—811

Edmonson, WT, and AH. Litt. 1982. Daphnia in Lake Washington. Limnology and
Oceanography. 27:272-293.

Eklov, P., and BE. Werner. 1999. Multiple predator effects on size-dependent behavior
and mortality in two anuran species. Oikos, 88:250-25 8.

Eklov, P., and C. Halvarsson. 2000. The trade-offs between foraging activity and
predation risk for Rana temporaria in different food environments. Canadian Journal

of Zoology, 78:734-739.

Fiksen, O., 1997. Allocation patterns and diel vertical migration: modeling the optimal
Daphnia. Ecology 78: 1446-1456.

Fiksen, O., Eliassen, S., and J. Titleman. 2005. Multiple predators in the pelagic:
modeling behavioural cascades. Journal of Anmical Ecology 74:423-429.

Gelinas, M., Pinel-Alloul B., and M. Slusarczyk. 2007. Alternative antipredator
responses of two coexisting Daphnia species to negative size selection by YOY
perch. Journal of Plankton Research 29:775-789.

Gore, A., and S. Paranjpe. 2001. A course in mathematical and statistical ecology.
Kluwer Academic Publishers.

Harvell, CD. 1990. The ecology and evolution of inducible defenses. Quarterly Review
of Biology, 65:323—340.

Ives, A. R., and A. P. Dobson. 1987. Antipredator behavior and the population dynamics
of simple predator-prey systems. American Naturalist 130:431-447.

85

 

Karels, T.J., Byrom, A.E., Boonstra, R. and C.J. Krebs. 2000. The interactive effects of
food and predators on reproduction and overwinter survival of Arctic Ground
Squirrels. Journal of Animal Ecology, 69: 235—247.

Lehman, J.T. & Caceres CE. 1993. F cod-webs response to species invasion by a

predatory invertebrate: B. longimanus in Lake Michigan. Limnology and
Oceanography, 38, 879—891.

Lima, S.L. (1998) Non-lethal effects in the ecology of predator—prey interactions: What
are the ecological effects of anti-predator decision-making? Bioscience, 48, 25—34.

Lima, S.L., and L.M. Dill. 1990. Behavioral decisions made under the risk of predation.
Canadian Journal of Zoology 68:619-640

 

Mangel M, Clark C (1988) Dynamic modeling in behavioral ecology. Princeton
University Press, Princeton, NJ

McPeek, M.A., and B. L. Peckarsky. 1998. Life histories and the strengths of species
interactions: Combining mortality, growth and fecundity effect. Ecology 79:867—879.

Muirhead J. and WC. Sprules. 2003. Reaction distance of B. longimanus longimanus,
encounter rate and index of prey risk for Harp Lake, Ontario. Freshwater Biology,
48: 1 3 5—146.

Nelson, E. 1H., C. E. Matthews, and J. A. Rosenheim. 2004. Predators reduce prey
populations growth by inducing changes in prey behavior. Ecology 85: 1853—1858.

Nelson, E. H. 2007. Predator avoidance behavior in the pea aphid: costs, frequency and
population consequences. Oecologia. 151: 22-32.

Palheimo, J .E. 1974. Calculations of instantaneous birth rate. Limnology and
Oceanography. 19:692-694.

Pangle, K. L, and SD. Peacor. 2006. Non-lethal effect of the invasive predator
Bythotprehes longimanus on Daphnia mendotae. Freshwater Biology 51:1070-1078.

Pangle, K.L., Peacor, SD, and O. E. Johannsson. 2007. Large nonlethal effects of an
invasive invertebrate predator on zooplankton population growth rate. Ecology,
88:402-412.

Peacor, SD, and EB. Werner. 2000. Predator effects on an assemblage of consumers
through induced changes in consumer foraging behavior. Ecology, 81 : 1998-2010.

Peacor, SD, and E. E. Werner. 2004a. How dependent are species-pair interaction
strengths on other species in the food web? Ecology 85(10):2754-2763.

86

Peacor, SD, and E. E. Werner. 2004b. Context dependence of non-lethal effects of a
predator on prey growth. Israel Journal of Zoology, 50:139-167.

Peacor, SD. and BE. Werner. 2006. Lethal and nonlethal predator effects on an
herbivore guild mediated by system productivity. Ecology, 87: 347-361.

Peacor S.D., Allesina S., Riolo R.L., and M. Pascual. 2006. Phenotypic plasticity
Opposes species invasions by altering fitness surface. PLoS Biology,4:2112-2120.

Pekarsky, B.L., Cowar, C.A., Penton, MA. and C. Anderson. 1993. Sublethal
Consequences of stream-dwelling predatory stoneflies on mayfly growth and
fecundity. Ecology, 74: 1836-1846.

Pothoven, S.A., Fahnenstiel, G.L., and HA. Vanderploeg. 2001. Population dynamics
of Bythotrephes cederstroemii in south-east Lake Michigan 1995-1998. Freshwater
Biology, 46:1491-1501.

Preisser, E.L., Orrock, J .L., and OJ. Schmitz. 2007. Predator hunting mode and habitat
domain alter nonconsumptive effects in predator-prey interactions. Ecology,
88:2744-2751.

Preisser, E.L., Bolnick, DI, and M.F. Benard. 2005. Scared to death? The effects of
intimidation in predator-prey interactions. Ecology, 501-509.

Sih, A., Crowley, P., McPeek, M., Petranka, J. and K. Strohmeier. 1985. Predation,
competition and prey commmrities — a review of field experiments. Annual Review of
Ecology and Systematics. 16: 269—3 1 1.

Schulz, KL., and RM. Yurista. 1999. Implication of an invertebrate predator’s

(Bythotprehes cederstroemi) atypical effects on a pelagic zooplankton community.
Hydobiologia 380: 179-1 93.

Schwab, D. J., and K. W. Bedford. 1999. The Great Lakes Forecasting System. In
Coastal Ocean Prediction, Coastal and Estuarine Studies 56, C.N.K. Mooers (Ed),
American Geophysical Union, Washington, DC, pp. 157-173.

Tollrain, R. and CD. Harvell. 1999. The ecology and evolution of inducible defenses.
Princeton University Press, Princeton, N. J.

Trussell, G. C. 2000. Predator-induced morphological tradeoffs in latitudinally-separated
populations of Littorina obtusata. Evolutionary Ecology Research 22803—822.

Turner, A.M. 2004. Non-lethal effects of predators on prey growth rates depend on prey
density and nutrient additions. Oikos 104:561-569.

87

 

Vanderploeg H.A., Nalepa T.F., Jude D.J., Mills E.L., Holeck K.T., Liebig J .R.,
Grigorovich LA. and H. Oj aveer. 2002. Dispersal and emerging ecological impacts
of Ponto-Caspian species in the Laurentian Great Lakes. Canadian Journal of
Fisheries and Aquatic Sciences, 59:1209—1228.

Werner, EB. and B.R. Anholt. 1993. Ecological consequences of the tradeoff between
growth and mortality rates mediated by foraging activity. American Naturalist
142:242-272.

Werner, BE, and SD. Peacor. 2003. A review of trait-mediated indirect interactions in
ecological communities, Ecology 84: 1083—1 100.

Yurista, RM. and KL. Schulz. 1995. Bioenergetics analysis of prey consumption by

Bythotrephes cederstroemi in Lake Michigan. Candian Journal of Fisheries and
Aquatic Sciences, 52:141-150.

88

 

       

S

11

 

1111111111111111111