Download View - OhioLINK ETD

QUANTITATIVE GENETICS OF ZEBRAFISH ONTOGENY UNDER CHANGING
ENVIRONMENTAL CONDITIONS
A Dissertation
Presented to
The Graduate Faculty of The University of Akron
In Partial Fulfillment
of the Requirements for the Degree
Doctor of Philosophy
Chris Marks
May, 2012
i
QUANTITATIVE GENETICS OF ZEBRAFISH ONTOGENY UNDER CHANGING
ENVIRONMENTAL CONDITIONS
Chris Marks
Dissertation
Approved:
Accepted:
______________________________
Advisor
Brian Bagatto
______________________________
Department Chair
Monte Turner
______________________________
Co-Advisor
Francisco B.-G. Moore
______________________________
Dean of the College
Chand Midha
______________________________
Committee Member
Kevin P. Kaut
______________________________
Dean of the Graduate School
George R. Newkome
______________________________
Committee Member
Andrea Case
______________________________
Date
______________________________
Committee Member
Mark Kershner
ii
ABSTRACT
Quantitative genetics has provided a proximate tool for making evolutionary
predictions based on genetic and environmental sources of variation.
Unfortunately, quantitative models have failed to address the evolutionary
consequences of development in complex, changeable environments. This type
of knowledge can only be achieved by tracking unique genotypes across all
possible combinations of changing environmental factors.
The first part of my research focuses on the consequences of changing
environmental oxygen on zebrafish cardiovascular development. I found that
cardiac output at the 48th hour of development in a given oxygen environment
was conditional upon oxygen conditions during the first 24 hours of development.
These conditional responses varied across genotypes, resulting in interactions
between genotype, early, and later environments (G xExE).
The second part of my research focused on the consequences of
changing oxygen on zebrafish morphology. I found that body shape in zebrafish
in a given oxygen environment for days 6-90 was conditional upon oxygen
conditions for days 0-6. These conditional responses also varied across
genotypes.
The third part of my research focused on the consequences of changing
oxygen on zebrafish behavior, size, and physiology. I found that size and
iii
behavior in zebrafish was dictated by oxygen conditions for days 0-30.
Physiology, however, was influenced most by oxygen conditions for days 30-90.
The fourth part of my research focused on the consequences of changing
food rations of zebrafish size and swimming ability. I found that fish raised under
all possible combinations of high and low food treatments achieved the same
body size at 60 days. Swimming performance, however, was conditional on
interactions between feeding treatments. I also detected significant family-level
variation for these responses, indicating at least some heritable variation.
These studies comprise some of the first quantitative genetic studies to
track developmental outcomes across more than one instance of environmental
change. My work has shed light on the evolutionary potential in complex,
changeable environments as well as provided a tractable tool for partitioning the
proximate sources of development across discrete ontogenetic periods.
iv
TABLE OF CONTENTS
Page
LIST OF TABLES .............................................................................................. viii
LIST OF FIGURES ............................................................................................. x
CHAPTER
I. INTRODUCTION .............................................................................................. 1
Quantitative genetics in changing environment .............................................. 2
Predictability and plasticity ............................................................................. 8
Evolutionary predictions in leiu of multiple instances of environmental
Change........................................................................................................... 8
Environmental change and past selection ...................................................... 9
Incorporating G: GxExE ............................................................................... 10
From theory to practice: zebrafish under changing environmental conditions
..................................................................................................................... 11
Chapters II-VI: a roadmap ............................................................................ 14
Hypothesis ................................................................................................... 16
II. A QUANTITATIVE GENETIC ANALYSIS ON THE RELATIVE INFLUENCES
OF GENOTYPE, EARLY, AND LATER ENVIRONMENTS ON THE
DEVELOPING ZEBRAFISH CARDIOVASCULAR SYSTEM ............................ 17
Introduction .................................................................................................. 17
Methods ....................................................................................................... 23
Results ......................................................................................................... 27
v
Discussion…………………….........................................................................36
Conclusions and evolutionary implications ................................................... 40
III. ONTOGENETIC PROGRAMMING IN BODY SHAPE OF ZEBRAFISH: THE
ROLE OF CHANGING OXYGEN DURING DEVELOPMENT IN SHAPING THE
P MATRIX........................................................................................................ 46
Introduction .................................................................................................. 46
Methods ....................................................................................................... 49
Results ......................................................................................................... 56
Discussion .................................................................................................... 70
Conclusions and evolutionary implications ................................................... 77
Acknowledgments ........................................................................................ 77
IV. ONTOGENETIC OXYGEN CHANGES ALTER ZEBRAFISH BEHAVIOR,
PHYSIOLOGY, AND ORPHOLOGY ............................................................... 79
Introduction .................................................................................................. 79
Methods ....................................................................................................... 83
Results ......................................................................................................... 90
Discussion .................................................................................................. 105
Conclusions and evolutionary implications ……….…..………………………116
Acknowledgments ...................................................................................... 116
V. THE INFLUENCE OF ONTOGENETIC DIETARY CHANGES ON ZEBRAFISH
SIZE AND SWIMMING PERFORMANCE ..................................................... 118
Introduction ................................................................................................ 118
Methods ..................................................................................................... 120
Results ....................................................................................................... 123
Discussion .................................................................................................. 133
vi
Conclusions and evolutionary implications ................................................. 136
Acknowledgments ...................................................................................... 136
VI. Conclusions................................................................................................ 138
Ramifications of environmental complexity ................................................ 138
Variation, selection, and evolution.............................................................. 139
Future directions ........................................................................................ 141
LITERATURE CITED ...................................................................................... 143
vii
LIST OF TABLES
Table
Page
2.1
Analysis of variance for vessel structure and cardiac performance in
developing zebrafish embryos ................................................................ 30
3.1
Multivariate analysis of variance for zebrafish morphology and body shape
................................................................................................................ 58
3.2:
Analysis of variance results for univariate traits (body area, maximum
depth, eye area, operculum height, eye-operculum distance, PC1, and
PC2)....................................................................................................... 60
3.3
Results of pair wise Flury comparisons for all half sib families pooled by
developmental treatment ........................................................................ 62
3.4
Results of pair wise Flury comparisons for all half sib families divided
among all four developmental treatments ............................................... 63
3.5
Pairwise Procustes distances between environmental treatments. The first
row and column represents the treatment............................................... 67
3.6
Pairwise Procustes distances between all sires and environmental
treatments ............................................................................................... 71
4.1
Mixed model repeated-measures ANOVA analyzing the effects of trial
phase, sire, early oxygen (E0-30), later oxygen (E30-90), and all possible
interactions on behaviors ........................................................................ 92
4.2
Mixed model ANOVA analyzing the effects of sire, early oxygen (Env0-30),
later oxygen (Env30-90), and all possible interactions on blood glucose,
PC1 (size), and latency to shoal (latency).............................................. 93
5.1
ANOVA results for total length (TL), standard length (SL), maximum
depth (MD), and swim velocity .............................................................. 125
5.2
Total length (mm) for zebrafish at 30 days under high and low food
rations ................................................................................................... 126
5.3
Maximum depth and standard length (mm) for zebrafish at 60 days under
viii
all combinations of high and low food rations ...................................... 127
ix
LIST OF FIGURES
Figure
Page
1.1
Hypothetical reaction norm showing the response of three genotypes to
one instance of environmental change ..................................................... 3
1.2
Hypothetical reaction norms representing phenotypic responses to two
instances of environmental change during development .......................... 5
1.3
Hypothetical reaction norms representing phenotypic responses of three
unique genotypes to environmental change later in development ............ 7
2.1
Various strains of zebrafish..................................................................... 25
2.2
The influence of developmental oxygen on stroke volume ..................... 31
2.3
The influence of developmental oxygen on cardiac output for half-sib
families A-K............................................................................................. 32
2.4
The influence of developmental oxygen on arterial diameter .................. 33
2.5:
The influence of developmental oxygen on arterial heart rate ................ 34
2.6
The influence of developmental oxygen on venous diameter ................. 35
2.7
Cardiac output of all half sib families ...................................................... 42
2.8
Cardiac output of all half sib families raised under early normoxic
conditions................................................................................................ 43
2.9
Cardiac output of all half sib families raised under early hypoxic
conditions................................................................................................ 44
2.10
Coefficient of variation due to all four combinations of oxygen
Environments .......................................................................................... 45
3.1
Landmarks and morphological measurements on photographed
zebrafish ................................................................................................. 53
3.2
The influence of developmental oxygen on PC1 ..................................... 65
x
3.3
The influence of developmental oxygen on PC1 for half sib families
A-D.......................................................................................................... 66
3.4
Effect of developmental environment (HH, HN, NH, NN) on the canonical
variate scores ........................................................................................ 68
3.5
Comparison of body shape of HH vs NN fish.......................................... 69
3.6
Variance due to all four combinations of oxygen environments .............. 76
4.1
Overview of behavioral arena ................................................................. 85
4.2
Percentage time spent swimming before, during, and after stimulus
introduction as a function of four developmental environments .............. 94
4.3
Percentage time spent on the shoaling side before, during, and after
stimulus introduction as a function of four developmental
environments .......................................................................................... 95
4.4
Percentage time shoaling before, during, and after stimulus introduction
as a function of four developmental environments .................................. 97
4.5
Number of crosses between shoaling and non-shoaling sides of the
arena before, during, and after stimulus introduction as a function of four
developmental environments .................................................................. 99
4.6
Percentage time spent hiding before, during, and after stimulus
introduction as a function of four developmental environments ............ 100
4.7
Blood glucose levels (mg/dl) as a function of four developmental
environments ........................................................................................ 102
4.8
Principal component 1 (body size) as a function of four developmental
environments ........................................................................................ 104
4.9
Variance due to all four combinations of oxygen environments ............ 117
5.1
The influence of diet for days 0-30 on total length in four full-sib
families.................................................................................................. 124
5.2
The influence of changing diet on standard length in four full-sib
families.................................................................................................. 128
5.3
The influence of changing diet on maximum depth in four full-sib
xi
families.................................................................................................. 129
5.4
Swimming velocity as a function of four nutritional environments ......... 131
5.5
The influence of changing diet on swimming velocity in four full-sib
families.................................................................................................. 132
5.6
Variance due to all four combinations of feeding treatments ............... 137
xii
CHAPTER I
INTRODUCTION
Within a population, phenotypic variation provides the raw material for
natural selection to optimize phenotypes. Thus, understanding the proximate
causes of phenotypic variation are critical to understanding evolutionary
processes. Ultimately, phenotypic variation is reduced to genetic variation (G),
environmental variation (E), and their interaction (GxE). The genetics of
phenotypic variation has been extensively explored. Additive, dominant, and
epistatic interactions have been pinpointed as discrete subcomponents
contributing to overlying phenotypic variation (Atchley and Zhu 1997; Lynch and
Walsh 1998; Wang et al 2006). Unfortunately, studies addressing the nature of
environmental variation have not kept pace. Traditional studies examining the
influence of environmental variation on resulting phenotypic variation typically
consider the influence of environmental change at one discrete point in ontogeny.
This approach has been criticized as a gross oversimplification (Sih 2004). As
the environment can change multiple times during ontogeny, there arises the
need to consider multiple changes at different points during developmental time.
Studies examining the consequences of such changes thus offers not only a
deeper understanding the of proximate causes of variation, but also allows for
1
better informed predictions on evolutionary processes in changeable
environments.
Qunatitative genetics in changing environments
The environment plays a critical role in development (Pigluicci 1998). It
can permanently alter developmental trajectories at critical ontogenetic stages
(Burggren and Reyna 2011). For example, insulin sensitivity can be programmed
in utero based on maternal nutrition status (review in Jones and Ozanne 2009).
Thus, the entire ontogenetic history of the developing organism becomes critical
in determining adult fitness. If we partition ontogeny into discrete stages (i.e.
early and later), we can apply quantitative genetic approaches to weigh the
relative influences of environmental conditions during these stages in
determining phenotypic outcomes. Under this approach, phenotypic outcomes
can be due to early environment (Eearly), later environment (Elate), or their
interaction (Eearly x Elate). This approach not only elucidates which conditions are
most critical in shaping phenotypes, but also the consequences of applying these
conditions at various points of ontogeny.
The above approach helps elucidate the role environmental change can
play throughout development, however, the potential for evolutionary change
remains unclear. Genetic variation for plasticity (i.e. G x E interactions) predicts
that plasticity can evolve. Traditional reaction norms plot the slope of the
responses of separate genotypes to one instance of environmental change (Fig.
1.1) (Via and Lande 1985). Such plots can be modified to account for two
instances of environmental change. Under this modified approach,
2
Fig. 1.1. Hypothetical reaction norm showing the response of three genotypes to
one instance of environmental change.
3
rather than representing genotypes, individual lines represent early
environmental states. The x axis, which traditionally represented the single
instance of environmental change in traditional reaction norms, now represents a
later environmental change. Examples of such figures are shown in figure 1.2.
On this hypothetical figure, line type (dashed vs solid) represents two possible
early environmental states (Eearly; A vs B). The x axis represents two possible
later environmental states (Elate; A vs B). Thus, the responses of individuals to all
four possible combinations of subsequent environmental changes are visibly
quantified (AA vs AB vs BB vs BA). Figure 1.2A represents the case where early
environmental conditions alter the directional response to later environmental
conditions. Quantitatively, this would be an Eearly x Elate interaction. Figure 1.2B
represents the case where variation is due to only to the early environment.
Quantitatively, Eearly would be the only significant source of variation. Both early
and later environmental conditions influence variation in figure 1.2C. Therefore,
both Eearly and Elate would be significant sources of variation. Figure 1.2D
represents a special case. Similar to figure 1.2A, the direction of the responses to
multiple instances of environmental change differ. While this would also
represent a significant Eearly x Elate interaction, unlike figure 1.2A, the population
variance structure in environments A and B differ.
4
Fig. 1.2. Hypothetical reaction norms representing phenotypic responses to two
instances of environmental change during development. Line type (dashed vs
solid) represent unique environmental conditions at an early point during
development. The x axis (A vs B) represents unique environmental conditions
experienced during later development. Each panel represents hypothetical
responses of four different genotypes.
5
While these figures display variance contributed by both environments,
genotypic variance is difficult to visualize. We can divide the reaction norms by
early developmental conditions (A vs B) with the response of each genotype to
later environmental conditions (A vs B). Figure 1.3 depicts the response of
genotypic variants to later environments A and B. However, figure 1.3A
represents variants exposed to early environment A and 1.3B represents the
same variants exposed to early environment B. These are therefore similar to
traditional reaction norms, however, the influence of subsequent developmental
environments is apparent. Visualizing genetic variation is essential to making
informed evolutionary predictions. The change in variance expressed in each
environment influences that rate at which selection removes unfit variants from a
population (Gupta and Lewontin 1982, Remold and Lenski 2001). The variance
structures in figure 1.3 are altered by early environmental conditions. In figure
1.3A, there is no change in variance structure across later environments A and B.
The change in rank order across later environments A and B is also important in
maintaining genetic variation within variable populations. This is because there is
no superior genotype as the environments changes from A to B (Gupta and
Lewontin 1982, Remold and Lenski 2001). Whereas early environment A alters
the rank order of 2 of the 3 genotypes, early environment B does not. The
influence of early environment B also differs in that the genetic variance structure
differs in later environments A and B.
6
Fig. 1.3. Hypothetical reaction norms representing phenotypic responses of three
unique genotypes to environmental change later in development. Line type
(dashed vs solid) represent unique environmental conditions at an early point
during development. The x axis (A vs B) represents the later environment.
7
Predictability and plasticity
Theory predicts that adaptive plasticity can occur if environmental change
is predictable (Levins 1968, Via et al. 1995 Schlichting and Pigliucci 1998, Reed
et al. 2010). The basic tenant of this dissertation is that environmental change is
not always predictable. Although some factors are certainly predictable (photo
cycle, seasons, high and low tides), many factors are more stochastic. Factors
such as food availability, oxygen levels, temperature, pH, and predators can vary
spatiotemporally across an organism’s ontogeny. Phenotypic responses in
unpredictable environments (i.e. where the early environment fails to predict the
later environment) have thus far been characterized at the population level
(Relyea 2003, Marks et al. 2005, Hoverman and Relyea 2007, Monaghan 2008,
Kotrschal and Taborsky 2010). One shortcoming of these studies is that no
evolutionary inferences can be made. In order to make evolutionary predictions,
genetic variation for responses in unpredictable environments must be
characterized. The work in this dissertation embodies the first series of
experiments that charactrize genotypic variance in a model organism exposed to
changing, reversible conditions during ontogeny. By partitioning genotypic
variation across different conditions during different periods of development, this
work will shed light on how evolution proceeds in complex enviornments.
Evolutionary predictions in leiu of multiple instances of environmental change
Variation in a given environment provides potential insight into
evolutionary dynamics. As the classic breeder’s equation predicts, the response
to selection (R) equals the product of the heritability of a trait (h 2) and the
8
selection differential (S) (Lynch and Walsh 1998). Numerous studies have noted
an increase in the heritability of traits under environmental change (Sgro and
Hoffman 1998, Bubily et al 2001, Charmantier and Garant 2005, Ibarra and
Famula 2008). As heritability and phenotypic variance increases, the response to
selection would increase proportionally. Thus, the amount of observed variance
under a set of subsequent environmental conditions provides insight into the rate
of evolution under such environmental changes. In figures 1.2A-C, the variance
in both later environments remains the same regardless of early environmental
conditions. Therefore, the rate of evolution would be the same under
environmental sequences AA and AB as BB and BA. Figure 1.2D, however,
provides a unique example. While this figure is quantitatively identical to figure
1A (i.e. Eearly x Elate interactions), the observed variance changes and is
conditional on the particular order of sequential environments. Individuals from
sequential environments AA and BA exhibit limited phenotypic variance.
Individuals from sequential environments AB and BB exhibit increased variance.
Thus, the response to selection in populations where these two particular
combinations of environmental states potentially occur would be accelerated.
Environmental change and past selection
Variance not only provides insight into the rate of evolution, but can also
provide information on past selection. As it optimizes phenotypic responses,
selection would by definition reduce phenotypic variance (Lynch and Walsh
1998). In figure 1.2D, variance is reduced under subsequent environmental
9
states AA and BA. Thus, these particular combinations of environments may
have predominated in this population’s evolutionary history.
Incorporating G: G x E x E
Statistically, the unique responses of separate genotypes to one instance
of environmental change contributes to G x E interactions (Via 1985). Variation in
the magnitude and direction of responses of separate genotypes across the four
possible combinations of subsequent environmental changes would contribute to
a significant G x E x E interaction. Figures 1.2A-D represent cases where there is
variation in the quality of responses to four possible combinations of sequential
environments. These collective responses embody the three way interaction
between genotype and subsequent environmental states (G x E x E). Thus, in a
complex environment, where states can be A, B, switch from A to B, and from B
to A, the most optimal response to these collective combinations of sequential
environmental states would be selected. For example, perhaps the highest
fitness would be achieved if a larger body size is expressed when exposed to
early environmental state A regardless of later environmental state (A or B). And
assume in the same population it would also be advantageous to express a small
body size when exposed to early environmental state B regardless of later
environmental state (A or B). This is the scenario presented in figure 1.2B. In this
scenario, the fitness of genotype B would be superior to genotypes A, C, and D.
Thus, genotype B would thrive under changes between these environmental
states. A significant GxExE response at the population level thus indicates the
10
presence of multiple genotypic strategies to respond to changes between
environmental states.
From theory to practice: zebrafish under changing environmental conditions
Why zebrafish?
Zebrafish are a popular vertebrate model of development (Grunwald and
Eisen 2002). They are native to Southeast Asia and inhabit a wide variety of
habitats including rivers, streams, and stagnant rice fields (Engeszer et al.
2007b). The wide range of environments they encounter makes them a tractable
model for studying the effects of changing ecological conditions on development.
The zebrafish genome has also been extensively studied, making them a
promising model for linking ecological impacts with developmental end
evolutionary outcomes (http://www.ncbi.nlm.nih.gov/genome/guide/zebrafish/).
Why use a developmental framework?
Development played a major role in evolutionary theory before and after
Darwin’s major contributions (Gould 1977). Prior to the rediscovery of Mendel’s
particulate theory of inheritance (De Vries 1889, Corren 1900), embryonic
development was viewed as the primary target for evolution and taxonomic
diversity (Richards 1982). However, the rise of Mendelian genetics catalyzed the
demotion of developmental biology to the backwaters of evolutionary thought.
The modern synthesis emphasized genetic mutation and recombination as the
primary target for evolutionary novelty. The computational utility offered by
population (Haldane 1924, Wright 1932) and quantitative genetic (Fisher 1930)
models yielded predictive potential not previously offered by developmental
11
models. The molecular underpinnings of phenotypic expression coupled with
theories of selection and drift have become the dominant paradigm for
evolutionary biology over the past 80 years.
Following the modern synthesis, biologists have returned to development
to explain more rapid diversification both in the fossil record (Gould 1977) and
extant species (Waddington 1952). Developmental plasticity and genetic
assimilation gained momentum as more parsimonious explanations for
macroevolutionary patterns (Waddington 1953, West-Eberhard 2003). The recent
discovery of homologous developmental regulatory genes has unlocked the
potential for studying development in a population and quantitative genetic
framework (Carroll 1995). Quantitative genetic models have begun to incorporate
a developmental framework into their models (Cheverud 1983, Atchley 1987,
Wang 2006). These studies demonstrate the usefulness of quantitative genetics
in making evolutionary predictions by partitioning development into discrete
stages through time. What is necessary to complete this picture is a further
understanding of the role plasticity plays at various stages of development and its
potential to confound or redirect ontogenetic trajectories. Particularly, do early
environmental conditions promote a plastic response that is canalized (as in Fig.
1.2B)? Or do they alter the manner in which later plastic responses are
expressed (as in Figs. 1.2A, D)? Collective figures such as 1.2A-D have yet to be
established in any know biological system. This dissertation’s analysis of
reactions to subsequent environmental changes across discrete stages of
12
development among separate genotypes represents a novel contribution to the
fields of development, evolution, and quantitative genetics.
Why hypoxia?
The majority of my studies examine the effects of changing oxygen levels
during zebrafish development. Dissolved oxygen is an aquatic environmental
factor that can change seasonally, monthly, or even daily according to other
conditions. Factors such as light, temperature, turbidity, and depth can alter
oxygen availability (Boutilier 1990). Hypoxia via eutrophication is becoming
increasingly common as humans continue to impart their influence on their
surroundings (Diaz and Rosenbergt 1995; Rabalais and Turner 2001).Hypoxia is
a common stressor in fish and is known to alter gene expression, physiology, and
behavior (Wu 2002). The ability to extract oxygen from the environment is critical
for organisms to maintain routine metabolic functions. Evolutionarily, fish have
acquired physiological adaptations to hypoxia such as increased gill surface area
(Nilsson 2007), altered hemoglobin binding efficiency (Jensen and Weber 1982),
respiration (Saint-Paul 1984), and lowered tissue oxygen demands (Hopkins and
Powell 2001). Fish have also adopted numerous behavioral strategies to cope
with hypoxia (reviewed in Pollock et al 2007). Hypoxia is therefore a particularly
important factor in shaping fish evolution.
Why different windows of development?
As indicated above, the following studies analyze the effects of changing
environmental conditions at varying points during ontogeny. While the timing of
these changes was chosen somewhat arbitrarily, they each represent a snapshot
13
of the potential ramifications of switching environmental conditions during
development. While the likelihood of zebrafish encountering constant hypoxia for
the first 30 days of development followed by constant normoxic conditions for 60
days (as an example) is unlikely, this approach explores a level of environmental
complexity that is not well established in current plasticity studies. Of all four
studies, chapter III uses perhaps the most ecologically relevant approach.
Oxygen regimes were switched at day 6 of development. Zebrafish embryos
remain sessile for approximately 5 days after development, after which they can
seek more optimal conditions. Given the aforementioned stagnant conditions in
which they reside coupled with their ability to tolerate anoxia (Padilla and Roth
2001), it is therefore likely that they encounter hypoxia early in development.
Their developmental regime therefore makes it likely that larval and adult oxygen
environments can differ.
Chapters II-VI: a roadmap
Given the usefulness of zebrafish as a vertebrate model as well as their
dynamic environmental history, the following chapters analyze the evolutionary
potential on the effects of changing environmental conditions during their
ontogeny.
Chapter II analyzes the effects of changing oxygen conditions (hypoxia vs
normoxia) on cardiovascular development at 48 hours. This developmental
window (particularly the first 24 hours) has not been well studied with regard to
the developing zebrafish cardiovascular system. We examine the impact of
14
switching oxygen conditions at 24 hours to determine the effect on subsequent
cardiovascular development at hour 48.
Chapter III analyzes the effects of changing oxygen on zebrafish
morphology and body shape at day 90. As zebrafish remain sessile for the first 56 days of development, we utilize this ecologically relevant time period to analyze
the impact of switching oxygen conditions during this particularly early window of
development on later adult body shape and morphology. Chapter IV analyzes the
effects of changing oxygen on zebrafish behavior, size, and physiology at day 90.
As organisms are comprised of suites of traits that can be functionally linked in
many ways (Roff 2000), we examine the influence of switching hypoxia and day
30 of development on zebrafish behavior, morphology, and physiology.
Many environmental factors can contribute to developmental variation.
Therefore, rather than using oxygen concentration, chapter V analyzes the
effects of changing food rations on zebrafish size and swimming ability at day 60.
We switched food rations between high and low treatments after 30 days to
examine the influence on zebrafish size and swimming performance.
Chapter VI discusses the consequences of each chapter’s findings on
making predictions of zebrafish evolution under changing environments. Each
chapter utilizes the modified reaction norm approach described earlier. In
general, the variance expressed in each later environment as a function of each
early environmental state will allow for predictions on which stages (early vs
later) are most important in making evolutionary predictions. Finally, chapter VI
will discuss the potential ramifications of the specific findings of each chapter.
15
At the end of each chapter, comparisons of selected traits will be made to
the modified reaction norms established in this chapter and the evolutionary
implication will be discussed. All chapters use breeding designs to estimate the
genetic contribution to plasticity as a function of changing developmental
environments. Since dams donate both genotypic and maternal effects, sire
variance is typically used as a proxy to estimate additive genetic variance in a
population (Lynch and Walsh 1998). Although maternal effects have important
implications on measuring selection (reviewed in Robinson 1981, Lynch and
Walsh 1998), they were not within the scope of this study. Therefore, although
dam effects were included in every analysis, only results including sire and
subsequent interactions are presented. All models have therefore accounted for
dam variance.
Hypothesis
The following chapters test the overall hypothesis that early environmental
conditions will constrain phenotypic variation in later environments. By tracking
separate genotypes across subsequent developmental environments, we further
hypothesize that there is significant additive genetic variation in the quality of
these environmentally-mediated responses.
16
CHAPTER II
A QUANTITATIVE GENETIC ANALYSIS ON THE RELATIVE INFLUENCES OF
GENOTYPE, EARLY, AND LATER ENVIRONMENTS ON THE DEVELOPING
ZEBRAFISH CARDIOVASCULAR SYSTEM
Introduction
Understanding the relationship between phenotypic variation, its
underlying causes, and evolutionary change remains one of the primary goals of
evolutionary biology. Recently, the reintegration of development and evolution
has revitalized the synthetic field of evolutionary developmental biology (EDB)
(Hall et al 2004). EDB views evolution as a change in developmental trajectories
(Atchley 1987, Carroll 1995, Carroll 2005). In recent decades, a more
developmental framework has been incorporated in quantitative genetic models.
These models demonstrate the utility of quantifying the underlying genetic
variation of developmental events occurring at discrete stages through time. One
finding is that the underlying genetic structure of traits (dominance, epistasis,
additive genetic variance) is dependent on the developmental stage studied
(Atchley and Zhu 1997, Wang et al 2006). As the environment changes,
however, developmental trajectories may be contingent on past conditions. It is
therefore necessary to examine the role plasticity plays at various stages of
development and its potential to constrain or redirect developmental trajectories.
17
Seminal to this effort is an understanding of how ecological change regulates
developmental trajectories. While the current environment can provide immediate
clues to the developing system (Schlichting and Smith 2002), the effects of past
environments can remain due to previous epigenetic reprogramming (Ellison
2010). Modeling the underlying quantitative genetics of development therefore
requires rigorous approaches that track all possible combinations of
environmental conditions experienced during development. This stems from
awareness that development is a sequential process in which early components
induce and alter the development of later components (Cowley and Atchley
1992). As the early developmental environment mitigates cell, hormone, and
protein activity, there lies the potential for altering the development of future
phenotypes. Hypoxia, for example, induces a cascade of responses with altered
cell signaling and protein expression (Ton et al 2003). Quantitative genetic
variation in these events provides one proximate source of variation in the quality
and timing of developmental events. As the environment changes, previous
courses of development may confound or redirect the expression of future
phenotypes.
Phenotypic development proceeds as a function of genotype,
environment, and their interaction (Scheiner 1993, Lynch and Walsh 1998,
Pigliucci 2001). While the genotype provides the basic instructions for
development, the environment can alter these instructions, resulting in alternative
phenotypes. This ability, termed developmental plasticity (West-Eberhard 2003),
underscores an organism’s ability to buffer the demands of environmental
18
change. Adaptive plasticity is believed to occur when environmental change is
predictable (Levins 1968, Via et al. 1995 Schlichting and Pigliucci 1998, Reed et
al. 2010). In a specific example, Sih and Moore (1993) found that salamanders
(Ambystoma texanum) delayed hatching in the presence of predators, resulting
in larger, more physically developed larvae. These larvae experienced less
predation than their non-exposed, smaller, less physically developed
conspecifics.
Studies partitioning the effects of environmental variation across
genotypes have revealed the presence of genetic variation for plastic responses
(Pigliucci 2001). Thus, plasticity has been considered a trait which selection can
optimize. Fisher (1918) first decomposed phenotypic variance (V P) into basic
causal components of genetic variance (VG), environmental variance (VE), and
their interaction (VGXE). This partitioning has provided that basis for modeling the
relative impacts of genotype and environment on phenotypic evolution and
development as well as making evolutionary predictions. More recently, Scheiner
and Goodnight (1984) defined plasticity variance (VPL) as the summation of
environmental variance (VE), and genotype by environment variance (VGXE).
Thus, plasticity can also be considered a quantitative trait that is reducible to
environmental (VE) and genetic (VGxE) influences and is capable of evolution.
While quantitative genetic studies have further partitioned genetic variance
into the effects of additive (VA), dominance (VD) and epistasis (VI) (Lynch and
Walsh 1998), there has been continued emphasis on only one level of
environmental variance (VE). The conventional partitioning of plasticity or
19
phenotypic variance across one level of environmental variance has not been
without criticism. Sih (2004) has argued that traditional GxE studies have
disregarded the true complexity of the environment. Indeed the environment is
likely to change during development. Variations in early environmental conditions
can occur at any time during development. Early conditions can have immediate,
lasting impacts on the course of the developing organism. For example, maternal
nutrition status can irreversibly influence offspring insulin sensitivity in a number
of animal models (Jones and Ozanne 2009). In sticklebacks, increasing maternal
stress resulted in offspring that later exhibited increased shoaling and antipredator behavior (Giesling et al. 2010). Vascular, endocrine, and metabolic
pathways are also established during early development (Hocher et al 2001) with
lasting effects on adult behavior and disease susceptibility in a number of animal
models (reviewed in Ellison 2010). These results suggest the need to partition
environmental influences beyond the current paradigm into early and later
environments.
The cardiovascular system is one of the first organ systems to develop in
vertebrates. This makes it particularly susceptible to early environmental
influences. In zebrafish (Danio rerio) cardiac output has been shown to increase
at 3 days post fertilization (d.p.f.) in response to chronic hypoxia (Jacob et al
2002). Bagatto (2005) found that chronic hypoxia shifted andrenergic responses
to occur earlier in zebrafish development. Moore et al (2006) found that cardiac
venous diameter, arterial diameter, heart rate, stroke volume, and cardiac output
in zebrafish at 4 d.p.f. was altered by hypoxia differentially across families. Moore
20
et al. were the first to establish genotype-specific plastic responses in
cardiovascular development in zebrafish. While more rigorous designs of the
quantitative genetics of zebrafish cardiovascular development have yet to be
performed, this result indicates at least some level of genetic variation in this
animal model. Collectively, these studies suggest that the early developing
cardiovascular system in zebrafish provide a tractable system to partition the
effects of genetic and environmental variation.
Here, we test the flexibility of the early zebrafish cardiovascular system to
changing oxygen conditions (i.e. all possible combinations of hypoxia and
normoxia). With the aid of a rigorous quantitative genetic design, we partition the
relative impacts of genotype (among sire variation), early (oxygen for hours 024), and later (oxygen for hours 24-48) conditions, as well as their inclusive
interactions. Our approach expands on traditional quantitative genetic studies by
partitioning the effects on developmental plasticity across discrete developmental
environments. Our approach allows for additive as well as interactive effects of
sequential developmental environments. In quantitative genetic terms, our
approach is as follows:
Traditional approach:
VPL = VE + VGxE
Where VPL = variance in plasticity, VE = variance due to the developmental
environment, and VGxE = variance due to genotype by environment interaction.
Our (expanded) approach:
VPL = VE1 + VE2 +VGxE1 + VGxE2+VE1xE2 + VGxE1xE2
21
Where VPL = variance in plasticity, VE1 = variance due to the first developmental
environment, VE2 = variance due to the second developmental environment,
VGxE1 = variance due to interaction between genotype and the first environment,
VGxE2 = variance due to interaction between genotype and the second
environment, VE1xE2 = variance due to interactions between both environments,
and VGxE1xE2 = variance due to the interactions between both environments and
genotype.
What knowledge can be gained from partitioning these sources? While
previous plasticity studies characterize genetic variation associated with one
instance of environmental change, they yield no predictive value on the
importance of early vs later environments in shaping developmental trajectories.
If early oxygen (hours 0-24) alters the response to later oxygen (hours 24-48), we
anticipate significant environment – environment (ExE) interactions between
early and later oxygen environments. Such results have been previously
demonstrated in a few models (Marks 2005; Kotrschal and Taborsky 2010). In
this study, however, we partition the responses to changing oxygen across
separate genotypes (sires). This will shed light on how early conditions alter or
constrain responses to later conditions across genotypes. If there is variation in
these responses, we anticipate significant GxExE interactions.
We analyzed particularly early windows of development (hours 0-24 and
24-48). This early window of development has yet to be explored with respect to
cardiovascular development. Oxygen sensors controlling cardiac performance
have been shown to be competent as early as 2-3 d.p.f. (Jacob et al 2002,
22
Pelster et al 2003). Hypoxia has also been shown to alter gene expression and
suppress morphological development at 48 h.p.f. (Kajimura et al. 2005). Although
the heart is not fully functional at 24 h.p.f., environmental cues have been shown
to later the onset of critical events in development (reviewed in Spicer and
Burggren 2003). Furthermore, hypoxia has been proven to influence primordial
germ cell migration which can influence the development of later organ systems
(Lo et al. 2011). We therefore hypothesize that oxygen changes during the first
24 hours of development will alter cardiovascular development.
Methods
Animals
Zebrafish (Danio rerio) are freshwater cyprinids native to India,
Bangladesh, and Myanmar (Barman 1991). They occupy a variety of habitats
ranging from stagnant, hypoxic ditches and rice fields to fully oxygenated waters
such as ponds and streams (Talwar and Jhingran 1991). They are an important
vertebrate model organism in the fields of development and genetics (Vascotto et
ali 1997, Grunwald and Eisen 2002). Their established usefulness in genetic and
developmental assays coupled with their exposure to a wide range of habitats
makes them an ideal candidate for the study of oxygen-induced developmental
plasticity.
Adult zebrafish (Danio rerio) were obtained from Aquatica Tropicals (Plant
City, FL). There are a number of strains (Fig. 2.1). In this study, we used strains
of wildtype (Fig. 1 WT) and albino (Fig. 1 A). and populations were maintained
23
according to standard procedures (Westerfield 1994). Adults were maintained
and bred at 26 °C with a 14L : 10D light cycle.
Breeding design
Wildtype males (n=11) were randomly selected and mated to randomly
selected albino females (n=55) in a nested mating design. Albino females were
used for their ease of sexing, large egg production, and established minimal
genetic variation. Such designs are typically used with among sire variance used
to estimate genotypic variance (Lynch and Walsh 1998). Each male was mated
with 5 unique females. After successfully mating, each female was removed from
the design. Successful mating was determined if at least 80 fertilized eggs were
acquired. Mating pairs were placed in 2 L containers supplied with a common
water source (Z-Mod housing system, Marine Biotech, Beverly, MA).
Developmental treatments
Eighty eggs from each clutch were removed before the 8 cell stage and
randomly divided between two developmental treatments. One treatment
consisted of normoxia (dissolved oxygen > 6.0 mg O2 l-1) and the other hypoxia
(dissolved oxygen 1.0 ± 0.2 mg O2 l-1). The apparatus for maintaining
developmental treatments is described in Marks et al. (2005). Following 24
hours, half of the eggs from each treatment were randomly placed in the opposite
environment. The remaining eggs were left to develop in the same environment.
This resulted in four treatment groups: larvae raised in hypoxia for 48 hours (HH),
normoxia for 48 hours (NN), normoxia for 24 hours followed by hypoxia for 24
hours (NH), and hypoxia for 24 hours followed by normoxia for 24 hours (HN).
24
Figure 2.1. Various strains of zebrafish. Taken from
http://www.pnas.org/content/vol105/issue3/images/large/zpq0010889800001.jpeg
25
Measurements
At 48 hours of development, high speed digital recording was used to
capture images of cardiovascular architecture. Since breeding designs were
staggered, only one clutch per day was measured. Measurements were made
between 47-50 hours of development. Measurements were randomized across
treatments to avoid a time bias. Larvae were recorded on a temperature
controlled stage (Harvard Apparatus) under and inverted microscope (Leica
DMIRB). The temperature controlled stage consisted of a thermal insert attached
to a dual channel bipolar temperature controller. The thermal insert surrounded a
small 3 cm x 2 cm x 2cm chamber filled with water from the subject’s most recent
treatment. A detailed schematic of the apparatus is available online at:
http://www.harvardapparatus.com/Live_Cell_Imaging_Brochure.pdf. Digital video
was recorded at 250 frames sec-1 with a mounted high speed digital video
camera (Red Lake MASD, San Diego, CA). Heart rate (beats/min), stroke volume
(nl) cardiac output (nl/min) , and caudal arterial and venous diameters (mm) were
measured according to Moore et al. (2006). All of these parameters showed
family-specific plasticity to hypoxia in Moore et al. (2006). Stroke volume was
calculated as the product of heart rate and cardiac output. Due
Statistics
A three-way nested ANOVA utilizing the proc glm procedure in the SAS
software package (SAS Institute, Cary, NC) was used to test which sources
contributed significantly to variation in cardiovascular architecture and
performance. Early environment, later environment, and their interaction were
26
treated as fixed factors. Sire and all inclusive interactions were treated as
random factors. All results conformed to normality assumptions. We therefore
used raw, untransformed data for all analyses.
Results
Sire
All measured variable varied significantly across sires (table 2.1). This
indicates that there is significant additive genetic variance for the traits
measured.
Eearly
Stroke volume and cardiac output both exhibited variance do to early
oxygen conditions (table 2.1). Fish raised in early hypoxia had higher overall
stroke volume and cardiac output. These parameters were highest on average
when the second developmental environment was hypoxia (Figs. 2.2, 2.3L)
Elate
Stroke volume, cardiac output, arterial diameter, and heart rate were
significantly altered by later oxygen. With the exception of heart rate, all
parameters increased on average when the second oxygen environment was
hypoxia (Figs. 2.2, 2.3L, 2.4, 2.5).
EearlyxElate
Arterial diameter was the only trait in which the response to later oxygen
was altered by early oxygen. This resulted in a significant interaction between
subsequent oxygen environments (table 2.1). While fish raised in early normoxia
27
increased arterial diameter when switched to hypoxia, fish raised in early hypoxia
decreased arterial diameter when switched to normoxia (Fig. 2.4).
Genotype-dependent plasticity to each environment: GxEearly and GxElate
Arterial diameter, venous diameter, and stroke volume showed variation in
responses to early oxygen that varied significantly across sires (Fig. 2.1). This
indicates significant additive genetic variance to the manner in which early
oxygen constrains or alters these parameters. The response to arterial diameter
to later oxygen also varied across sires (Fig. 2.1). This indicates significant
additive genetic variance to the manner in which later oxygen constrains or alters
arterial diamater.
Genotype-dependent conditional plasticity: GxEearlyxElate
Heart rate, stroke volume, and cardiac output all showed significant
interactions between sire, early, and later oxygen environments (table 2.1). This
indicates that the nature of the interplay between subsequent developmental
environments dependent on the genetic context. Cardiac output will be explored
as an example. While cardiac output generally increases in response to
developmental hypoxia for hours 24-48 (Fig. 2.3L), the responses vary across
families (Figs. 2.3A-K). While some genotypes remain flexible to oxygen
changes, others are heavily influenced by early oxygen conditions. Families B, C,
D, E, and K, for example, express higher cardiac output by individuals raised in
early hypoxia regardless of the later rearing conditions (Fig. 2.3). Families A and
F are also constrained by early oxygen conditions, however, individuals raised in
early normoxia consistently express higher cardiac output (Fig. 2.3). Families I
28
and J remain flexible in their responses to later oxygen. The direction of these
changes, however, is genotype specific. For family I, individuals tend to increase
cardiac output when switched to the opposite oxygen environment. Family J
shows the opposite trend, with individuals decreasing cardiac output (Fig. 2.3). It
is important to note that these assessments are strictly qualitative. We preformed
no post-hoc test to determine which environmental sources were significant
across sires (ex: reaction norms in fig. 2.3). However, clearly the differences in
direction and magnitude of responses seen across sires contributed to a
significant interaction between sire and subsequent developmental
environments. It is this higher-order interaction which was within the scope of this
study.
29
Table 2.1 Analysis of variance for vessel structure and cardiac performance in
zebrafish embryos at 48 h.p.f Sources of variation included sire, developmental
oxygen for hours 0-24 (Eearly), developmental oxygen for hours 24-48 (Elate), and
all possible interactions. For each variable, the first line represents the mean
square error and the second line represents the significance level.
Source
Sire
Eearly
Sire x
Eearly
Elate
Sire x
Elate
Eearly x
Elate
Sire x Eearly
x Elate
DF
Stroke Volume
(nl)
37
0.1024
<.0001
1
0.2516
0.0028
10
0.1178
<.0001
1
1.7414
<.0001
10
0.0427
0.1227
1
0.0088
0.5752
10
0.2092
<.0001
Cardiac Output
(nl/min)
3354.67
<.0001
4498.97
0.0058
2538.58
<.0001
31063.66
<.0001
1209.85
0.0255
492.40
0.3600
4759.73
<.0001
Arterial
Diameter (mm)
4.63
<.0001
9.91
0.2989
2.39
0.0041
1.58
<.0001
1.19
0.2270
4.01
0.0368
1.13
0.2658
Heart Rate
(beats/min)
3310.86
<.0001
202.76
0.4441
488.50
0.1700
2053.91
0.0150
1172.23
0.0002
563.83
0.2020
1130.80
0.0004
Venous
Diameter (mm)
7.20
<.0001
-5
-5
-6
-5
1.34
0.4267
-5
-5
4.60
0.0181
30
-4
-6
3.74
0.8249
-5
-5
2.53
0.2940
-5
-7
6.90
0.8565
-5
-5
1.48
0.7293
Stroke Volume (nl)
0.6
0.5
0.4
0.3
0.2
0.1
Normoxia
Hypoxia
Fig. 2.2. The influence of developmental oxygen on stroke voulme. Dashed and
solid lines represent fish raised in hypoxia and normoxia for hours 0-24,
respectively. The x axis represents the oxygen environment for hours 24-48.
Error bars represent standard errors.
31
40
20
100
80
I
60
40
20
Normoxia
Hypoxia
20
100
80
F
60
40
20
100
80
J
60
40
20
Normoxia
Hypoxia
C
80
60
40
20
100
80
G
60
40
20
100
80
K
60
40
20
Normoxia
Cardiac Output (nl/min)
40
Cardiac Output (nl/min)
60
100
Cardiac Output (nl/min)
60
80
Cardiac Output (nl/min)
E
B
Cardiac Output (nl/min)
Cardiac Output (nl/min)
80
100
Cardiac Output (nl/min)
20
100
Cardiac Output (nl/min)
40
Cardiac Output (nl/min)
60
Cardiac Output (nl/min)
A
80
Cardiac Output (nl/min)
C aCr d
a ir ad ci a cOOu ut pt pu ut t (( nn ll // m
m iInn) )
Cardiac Output (nl/min)
100
Hypoxia
100
D
80
60
40
20
100
80
H
60
40
20
100
80
L
60
40
20
Normoxia
Fig. 2.3. The influence of developmental oxygen on cardiac output for half-sib
families A-K. Dashed and solid lines represent fish raised in hypoxia and
normoxia for hours 0-24, respectively. The x axis represents the oxygen
environment for hours 24-48. Error bars represent standard errors. Figure 5L
represents the mean values across all families.
32
Hypoxia
Arterial Diameter (mm)
0.0210
0.0205
0.0200
0.0195
0.0190
0.0185
0.0180
Normoxia
Hypoxia
Fig. 2.4. The influence of developmental oxygen on arterial diameter. Dashed
and solid lines represent fish raised in hypoxia and normoxia for hours 0-24,
respectively. The x axis represents the oxygen environment for hours 24-48.
Error bars represent standard errors.
33
Heart Rate (beats/min)
162
160
158
156
154
152
150
148
Normoxia
Hypoxia
Fig. 2.5 The influence of developmental oxygen on heart rate. Dashed and solid
lines represent fish raised in hypoxia and normoxia for hours 0-24, respectively.
The x axis represents the oxygen environment for hours 24-48. Error bars
represent standard errors.
34
Venous Diameter (mm)
0.0246
0.0244
0.0242
0.0240
0.0238
0.0236
0.0234
0.0232
0.0230
Normoxia
Hypoxia
Fig. 2.6. The influence of developmental oxygen on venous diameter. Dashed
and solid lines represent fish raised in hypoxia and normoxia for hours 0-24,
respectively. The x axis represents the oxygen environment for hours 24-48.
Error bars represent standard errors.
35
Discussion
Phenotypic development is a dynamic, hierarchical process involving the
genesis of cells, hormones, tissues, and organs (Hall 2003). Environmental input
can occur at any stage of development. A specific stage of development may be
critical in determining the nature of the physiological response. For example,
Chan and Burggren (2005) showed that hypoxia altered chick morphological
development (eye size, beak length) as well as yolk mass differentially across
three stages of development. Slatkin (1987) proposed the first model integrating
quantitative genetics and development. This model examined the effects of
selection given quantitative genetic variation in growth rates and transition times
between developmental stages (Slatkin 1987). The quantitative genetics of
heterochrony provides a tractable means of partitioning the relative effects of
genetic and environmental variation in evolutionary developmental studies. While
genetic information may be lacking, environmental factors can be controlled and
manipulated at various critical windows of development. The partitioning of
developmental stages into discrete events provides the ability to partition the
relative effects of environments experienced throughout the organism’s
ontogeny.
Recent studies have emphasized the need to examine developmental
plasticity within a more ecological framework (Gilbert 2001, Hoverman and
Raylea 2007). This arises from an awareness of the environment’s ability to alter
the course of development at any time during an organisms’ ontogeny. Specific
interest has focused on the consequences changing ecological factors have on
36
the course of developmental trajectories. Pigliucci (1998) suggested
environmental change can constrain developmental plasticity in two ways. One is
through ontogenetic drift, in which a maladaptive resource allocation pattern is
maintained after the developmental environment changes (Pigliucci 1998). A
second process involves a mismatch between developmental windows and
environmental cues (Pigliucci 1998). Developmental windows involve the onset
of functional regulatory systems which can competently respond to
environmental change. Adaptive plasticity would therefore require the onset of
these critical points in development. If early developmental conditions are
capable of altering the timing of these developmental windows, this may alter the
ability of an organism to competently respond to later environmental change in
an adaptive manner. Our work shows that the early developing cardiovascular
system of zebrafish responds to environmental change in manners dependent on
early oxygen conditions. This suggests that early oxygen can alter the timing or
nature of regulatory events in cardiovascular development and therefore mediate
later responses to environmental change.
Switching oxygen environments during early zebrafish development
clearly alters cardiovascular architecture and performance. These changes are
not simply a function of one particular oxygen environment. Switching oxygen
environments constrains and redirects cardiovascular performance uniquely
across genotypes. In this manner, cardiovascular performance is a function of
the interaction between early and later oxygen environments. The cardiovascular
system is essential in supplying developing tissues with adequate oxygen and
37
nutrients. Thus, the heart is the first organ to develop. Proper cardiac responses
are therefore essential for proper development and survival (Clark 1990). Under
normoxic conditions, bulk diffusion is adequate to supply developing zebrafish
tissues with oxygen for the first 2 weeks of development (Pelster and Burggren
1996). Under hypoxic conditions, however, convective transport may become
necessary much earlier to compensate for oxygen deficits (Jacob et al 2002).
Previous studies imposing hypoxia through incubation (Jacob et al 2002) and
exercise (Pelster et al 2003) have confirmed the presence of oxygen sensors
controlling cardiac performance as early as 2-3 d.p.f. Our research indicates that
the oxygen conditions during the first 24 hours of development can modulate
cardiac responses over subsequent days. Although the heart is not functioning or
fully developed at 24 h.p.f., conditions during the first few hours of development
can be critical in shaping early cardiac performance.
Our study revealed variation in the direction and magnitude of response to
oxygen changes. This genotypic variation spanning subsequent developmental
environments may provide the raw material for natural selection to shape
plasticity-induced developmental trajectories. Spicer and Burggren (2003) termed
plasticity in the timing of crucial developmental events as heterokairy. Oxygen
has displayed the capacity to both up regulate and down regulate development.
Although hypoxia has a tendency to slow most developmental events (Pelster
1997, Padilla and Roth 2001), it can accelerate others. Spicer and El-Gamal
(1999) showed that hypoxia accelerated the onset of regulated respiration in
brine shrimp. Bagatto (2005) found that chronic hypoxia shifted andrenergic
38
responses to occur earlier in zebrafish development. Our study shows that early
oxygen conditions can delay or alter cardiovascular responses in developing
zebrafish. Since the cardiovascular system is not functioning at 24 h.p.f., there
have been no previous attempts to assess the impact of this early window of
development on cardiovascular performance. This is the first study, to our
knowledge, to show that conditions during the first 24 hours of development can
influence subsequent physiological development in zebrafish.
Our study elucidates the evolutionary potential within species when early
developmental conditions switch between environmental states (i.e. hypoxia vs
normoxia). Switching oxygen during development constrained and redirected
physiological trajectories in ways that were conditional on the quality of
environments experienced throughout ontogeny. Variation across genotypes in
the quality of environmentally constrained and redirected responses highlights
the potential for selection to shape the fittest developmental trajectory in light of
changing environmental conditions. Current models do not account for potential
variation within populations. Models such as the ‘silver spoon’ (Grafen 1988) and
‘environmental matching’ (Monaghan 2008) consider average responses at the
population level. While such models may apply at larger scales, our results
underscore the potential for interspecific variation in the nature of
environmentally mediated responses within populations. Therefore, the most
beneficial developmental pathway resulting from to environmental change seen
at the population or species level may be selected. Specifically, adaptive
plasticity can occur in light of subsequent environmental changes.
39
Conclusions and evolutionary implications
The overreaching goal of this chapter was to characterize variation across
genotypes (sires) in response to changing oxygen condition during early
zebrafish development. Indeed, we found that the nature of the constraint
imposed by early oxygen conditions in the ability to respond to later oxygen
change varied significantly across genpotypes. In the example of cardiac output,
this resulted in a significant Sire x Eearly x Elate (G xE x E) interaction. While figure
2.3 is informative for visualizing the effects of developmental conditions for each
genotype, it is difficult to visualize the degree of genetic variance. Therefore, the
evolutionary implications are not tractable from this figure. The modified
approach of splitting these figures by early environmental conditions addresses
this issue. The following figures plot cardiac output in hypoxia and normoxia at 48
h.p.f. as a function of early hypoxia and normoxia. Figure 2.7 plots all genotypes
and early oxygen states on the same figure. Overall, it appears that variance
increases when later oxygen is normoxia. Which early environment contributes to
this variance? Figures 2.8 and 2.9 plot only fish exposed to early normoxia and
hypoxia, respectively. Figure 2.10 summarizes the variance structure across all
four treatments. They indicate that fish exposed to early hypoxia experience
increased variance when switched to later normoxia. Thus, early developmental
hypoxia appears to influence the amount of variance expressed in a later
environment. Selection would therefore have more variants with which to select
from when early conditions switch from hypoxia to normoxia. Interestingly, fish
exposed to consistent hypoxia display the lowest amount of variance (Fig. 2.10).
40
Indeed, zebrafish are likely to encounter such conditions (see chapter I). Given
they are tolerant of anoxia during early development (Padilla and Roth 2001), it is
likely they have adapted to hypoxia during early development. This suggests that
selection has strongly influenced development in early hypoxia. Overall, these
findings provide novel insight into the mechanisms of variance expression during
early development in changing environments.
41
Fig. 2.7. Cardiac output of all half sib families. Dashed and solid lines represent
fish raised in hypoxia and normoxia for hours 0-24, respectively. The x axis
represents the oxygen environment for hours 24-48. Each color represents each
genotype. Error bars represent standard errors.
42
Fig. 2.8. Cardiac output of all half sib families raised under early normoxic
conditions (hours 0-24). The x axis represents the oxygen environment for hours
24-48. Each color represents each genotype. Error bars represent standard
errors.
43
Fig. 2.9. Cardiac output of all half sib families raised under early hypoxic
conditions (hours 0-24). The x axis represents the oxygen environment for hours
24-48. Each color represents each genotype. Error bars represent standard
errors.
44
Coeffecient of Variation
(cardiac output)
2.2
2.1
2.0
1.9
1.8
Later oxygen
Early oxygen
H
N
H
H
N
N
Fig. 2.10. Coefficient of variation due to all four combinations of oxygen
environments.
45
CHAPTER III
ONTOGENETIC PROGRAMMING IN BODY SHAPE OF ZEBRAFISH: THE
ROLE OF CHANGING OXYGEN DURING DEVELOPMENT IN SHAPING THE
P MATRIX
Introduction
A fundamental issue in evolutionary biology is the interplay between
genotype and environment in determining phenotypic outcomes. Phenotypic
plasticity allows organisms with a given genotype to respond to environmental
change (Bradshaw 1965; Via 1984). Correlations between traits (i.e. phenotypic
correlations) have been targeted as important players in affecting the expression
and evolution of plastic responses (Schlichting 1986; Pigliucci et al. 1995).
Although plasticity studies have revealed that phenotypic integration can be
altered by environmental change (Gianoli and González-Teuber 2005; Tonsor
and Scheiner 2007), the genetics of development in changeable ontogenetic
environments remains unclear.
In lieu of genetic variation, phenotypic plasticity can evolve as an adaptive
response to changeable environments (Van Tienderen 1991). Developmental
plasticity has been proposed as a source of evolutionary novelty (West-Eberhard,
2003, Carroll 2005). Knowledge of critical windows of development that are
particularly sensitive to environmental change reveal that the timing of change
46
can be critical in determining the expression of plastic responses (Pigliucci,
2001). These critical periods, particularly early in development, can result in
alternative developmental trajectories (Chan and Burggren 2005), which can
have lasting effects on development (Relyea 2001). These trajectories, however,
are likely to proceed under subsequent environmental change. The adaptive
value of developmental plasticity must therefore be considered in more complex
developmental environments.
As development proceeds, organisms are likely to encounter
environmental variation. An altered developmental trajectory can be adaptive
only if the early environment predicts future conditions (Dudley and Schmitt
1996). The influence on development can occur relatively soon, or can manifest
itself in delayed effects not seen until later stages of development (Chen et al.
1996; Buckley et al. 2005). If plastic results are delayed, changing environmental
conditions can offset otherwise adaptive responses. Recent work has begun to
characterize the consequences of plastic responses within populations when the
environment changes during ontogeny (Relyea 2001; Hoverman and Relyea
2007; Criscuolo et al. 2008; Monaghan 2008). While these studies underscore
the importance of environmental variance, the adaptive meaning of
developmental plasticity in complex environments can only be assessed by
addressing genetic variance.
Organisms are composed of suites of traits which can be genetically,
developmentally, and/or functionally linked (Cheverud 1996). These links result in
covariances between traits and have been proposed as a source of evolutionary
47
constraint (Arnold 1992; Murren 2002). Quantitative genetics models the
multivariate response to selection as Δ z =Gβ, where z is the vector of change in
mean trait values following selection, G is the matrix of additive genetic variances
and covariances, and β is the vector of selection gradients (Lande 1979). Since
multivariate trait correlations can be modulated by environmental change
(Pigluicci et al. 1995; Tonsor and Scheiner 2007), plasticity in these correlations
may provide the raw material for selection to overcome evolutionary constraints.
Here, we test whether oxygen changes during development alter overall
morphology and phenotypic integration in zebrafish. Although hypoxia has been
shown to alter body shape in fish (Chapman et al 2008, Crispo and Chapman
2011), no study to date has analyzed if the influence of hypoxia is maintained
after switching oxygen environments. Our study is the first to characterize the
effect of changing developmental oxygen on body shape and integrated
morphology (P) in the zebrafish, Danio rerio. We track separate genotypes (sire
offspring) across multiple windows of environmental change during development
to help elucidate the relative impact of sequential developmental environments
as well as the potential for genotype-specific responses. The ability to detect the
genetic contribution in this study will further aid in understanding the potential for
populations to adapt to heterogeneous conditions during development. Since
hypoxia is known to influence body shape, we hypothesize that fish exposed to
early hypoxia will experience modified body shapes, however, these
modifications will be dependent on later oxygen conditions.
48
Methods
We analyzed the effects of changing developmental oxygen on phenotypic
integration in 409 zebrafish resulting from 4 paternal half sib and 4 maternal half
sib families. Fish were raised in hypoxia for 90 days, normoxia for 90 days,
hypoxia for 6 days followed by normoxia for 84 days, or normoxia for 6 days
followed by hypoxia for 84 days. While we generally chose our times somewhat
arbitrarily, we chose 6 days in this study to represent an ecologically relevant
scenario. Following fertilization, zebrafish embryos remain sessile for
approximately 5 days, after which they can seek more optimal conditions. Their
developmental regime therefore makes it likely that larval and adult oxygen
environments can differ. The early developmental window utilized in this
experiment (days 0-6) provides an ecologically relevant example of the impact of
opposing larval and adult oxygen environments. We used MANCOVA and Flury
matrix comparisons to partition the effects of genotype, early, and later
developmental environments on overall morphology and phenotypic integration,
respectively.
Animals
Adult zebrafish (Danio rerio) were obtained from Aquatica Tropicals (Plant
City, FL) and populations of wild-type and albino were maintained according to
standard procedures (Westerfield, 1994). Adults were maintained and bred at 26
°C with a 14L : 10D light cycle.
Breeding design
49
Approximately 50 eggs from each clutch were removed before the 8 cell
stage and equally divided between developmental treatments. Following 6 days
of development, half of the eggs from each treatment were randomly selected
and placed in the opposite environment. The remaining eggs were left to develop
in the same environment. The larvae were raised under these conditions for the
remainder of the experiment (up to 90 days). This resulted in four treatment
groups: larvae raised in hypoxia for 90 days (HH), normoxia for 90 days (NN),
normoxia for 6 days followed by hypoxia for 84 days (NH), and hypoxia for 6
days followed by normoxia for 84 days (HN).
Wildtype males were randomly selected and mated to randomly selected
albino females in a full factorial North Carolina II breeding design. Sixteen
clutches resulted from 4 paternal half sib and 4 maternal half sib families. Albino
females were used due to their large egg production, ease of sexing, and lower
genetic variance. The later detail allowed us to contribute most of the genetic
variation in this study to sires. Mating pairs were placed in 2 L containers
supplied with a common water source (Z-Mod housing system, Marine Biotech,
Beverly, MA).
Developmental treatments
Treatments consisted of normoxic water (dissolved oxygen > 6.0 mg O2 l1
) and hypoxic water (dissolved oxygen 1.0 ± 0.2 mg O2 l-1). Fish were raised in
154 x 35 x 23 cm tanks divided into 16 chambers. System water was supplied to
each chamber with screen covered holes connecting neighboring chambers. This
ensured homogenous environmental conditions with the exception of treatments
50
being maintained using a partially separated sump system that allowed mixing of
water between treatments. The apparatus for maintaining developmental
treatments is further described in Marks et al. (2005).
Data collection
At 90 days post fertilization, fish were euthanized with an overdose of ms222, weighed, and photographed on each side with a length standard in each
picture. Photography was conducted with a Nikon D300 camera with a standard
reference scale under standard lighting conditions with a Kodak 18 percent
reflectance grey background.
Phenotypic measurements
Integrated morphology - To characterize integrated morphology, we
measured 5 morphometric characters (Fig. 3.1). These included body area,
maximum depth, eye area, operculum height, and the distance from the anterior
eye socket to the posterior portion of the operculum (eye-operculum distance).
We chose these traits as fair representatives of the variety of morphological
measurements found in previous studies (Pankhurst et al. 1993; Machiels and
Wijsman 1996; Webb et al. 2001; Monna et al. 2011). Measurements were taken
5 times on each side and averaged across both sides. This removed any
potential effect of changing asymmetry. Since sex could not always reliably be
determined visually (i.e. large males vs small females), sexes were pooled.
Measurements were made in imageJ (Version 1.42, NIH).
Body shape - On the left side of each subject, 14 landmarks were scored
using the software tpsDig2 (Rohlf 2001). We used a generalized least square
51
Procustes fit to correct for variation in scale, position, orientation, and centroid
size (Rohlf and Slice 1990). To account for any influence of size on shape, we
regressed Procustes coordinates on centroid size and used the residuals for
subsequent analysis. To characterize variation in body shape among and
between treatment groups, we performed a canonical variate analysis (CVA) and
discriminate function analysis (DFA) on regression residuals. These are common
ordination techniques used to assess multi-dimensional variation (Klingenberg et
al. 2003). These analyses reduce the data to fewer dimensions and allow for
simpler, 2-dimensional analysis. These analyses were performed in MorphoJ v.
1.01c (Klingenberg 2008). We used the software tpsRelw v. 1.49 (Rohlf, 2010) to
calculate weight matrices consisting of partial warp scores. These weight
matrices were used for a multivariate analysis (described below).
52
Fig. 3.1. Landmarks and morphological measurements on photographed
zebrafish. Landmarks: (1) Anterior tip of upper jaw; (2) anterior base of dorsal fin;
(3) posterior base of dorsal fin; (4) dorsal base of caudal fin; (5) medial spot of
caudal fin(aligned with midpoint of medial stripe); (6) ventral base of caudal fin;
(7) posterior base of anal fin; (8) anterior base of anal fin; (9) base of left pelvic
fin; (10) ventral tip of operculum; (11) distal tip of angular; (12) posterior tip of
operculum; (13) posterior portion of eye socket; (14) anterior portion of eye
socket. Morphological measurements: (A) Eye area; (B) distance from dorsal to
ventral operculum opening(operculum height); (C) distance from anterior portion
of eye socket (Eye-operculum (EO) distance); (D) maximum depth; (E) body area
(excluding fins).
53
Analysis of means
Integrated morphology - A three-way MANCOVA was performed to test for
effects on integrated morphology. Sire, early environment (Eearly, days 0-6), later
environment (Elate, days 6-90), and all possible interactions were included as
potential sources of variation. Post hoc univariate ANOVAs tested the effects on
individual traits. Sire, dam, and all inclusive interactions were treated as random
factors. Early environment, later environment, and their interaction were treated
as fixed factors. To remove the effect of size, body length was included as a
covariate. Assumptions of normality were assessed and variables were log
transformed as needed.
Body shape - A three-way MANOVA was performed on weight matrices to
test for effects on body shape. Sire, early environment (Eearly, days 0-6), later
environment (Elate, days 6-90), and all possible interactions were included as
potential sources of variation. Both of these tests were performed in SAS
software package version 9.0 using PROC GLM (SAS Institute, Cary, NC). For
both tests, if early environment affects variation in adult phenotype, we expect to
find significant Eearly components. If later environment influences adult
phenpotype, we expect to find significant Elate effects. If there is genetic variation
for these environmentally mediated responses, we expect to find significant G x
Eearly and/or G x Elate interactions. Another possibility is that later responses to
developmental oxygen are altered by early oxygen conditions. In this case, we
expect to find significant Eearly x Elate interactions. If there is genetic variation in
54
the manner in which these cross-environmental responses manifest, we expect
to find significant G x Eearly X Elate interactions.
Matrix comparisons
In addition to analyzing means, we analyzed the effect of subsequent
developmental environments and genetic variance on phenotypic covariance
matrices (P). Numerous studies have found correlations between P (phenotypic)
and G matrices (Cheverud 1988; Reusch and Blanckenhorn 1998; Waitt and
Levin 1998). We therefore use the P matrix as a reasonable estimator of the G
matrix in this study. While there are many matrix comparison methods available,
the Flury method remains one of the most widely used (Phillips and Arnold 1999).
Whereas most methods compare matrices for equality, Flury considers a
hierarchy of possible differences. It compares the matrices based on five
possible levels of similarity: 1) equality; in which the matrices share eigenvectors
and eigenvalues; 2) proportionality, in which the matrices share eigenvectors and
all their eigenvalues differ by a proportional constant; 3) common principle
components, in which the matrices share eigenvectors but do not share
proportional eigenvalues; 4) partial common principal components, in which
matrices share some eigenvectors; and 5) unrelatedness, in which none of the
above criteria apply (Flury, 1988). We used this method to compare P matrices of
full-sib half-sib groups reared in all possible environmental combinations (Table
2). For example, AHN represents the covariance matrix for the half-sibs of sire A
raised in hypoxia for days 0-6 and normoxia for days 6-90. We also compared
the matrices for all families pooled in each combination of oxygen environments
55
(HH, HN, NH, NN). This tested the average effect of changing oxygen conditions
across all families. P matrices were calculated in JMP 8.0. Flury analysis was
performed with the program CPC (available at
http://darkwing.uoregon.edu/~pphil/programs/cpc/cpc.htm). Phillips and Arnold
(1999) present three methods for determining the relationship between matrices
in the Flury hierarchy. One is the step up approach, in which the model is
compared to the next lowest model in the hierarchy and a move up the hierarchy
is made only if the result is not significant. Another is the jump up approach. This
is similar to the step up, except that each model is compared to an unrelated
structure rather than the next lowest one. The final method is the Akaike
information criterion (AIC). Similar to a log likelihood model, AIC adjusts the
goodness of fit according to the numbers of parameters estimated (Flury 1988).
All three models have their advantages and disadvantages. Phillips and Arnold
(1999) recommend the jump up approach because it is most appropriate for
testing a single hypothesis and provides the simplest interpretation of results.
Results
Matings resulted in 16 full-sib, half-sib families consisting of 94 individuals raised
completely in hypoxia (HH), 107 individuals raised in hypoxia followed by
normoxia (HN), 100 individuals raised in normoxia followed by hypoxia (NH), and
108 individuals raised completely in normoxia (NN).
Integrated morphology
Many factors contributed significantly to variance in integrated phenotype.
Significant sire component indicates the presence of genetic variation for overall
56
morphology. Significant Eearly and sire x Eearly components indicate there is
plasticity in overall morphology due to early oxygen conditions and these plastic
responses are genotype specific. The same pattern holds true for Elate and sire x
Elate. While there was not a significant Eearly x Elate interaction, there was genetic
variation for cross-environmental responses (sire x Eearly x Elate) (Table 3.1).
57
Table 3.1. Multivariate analysis of variance for zebrafish morphology and body
shape. Sources of variation included sire, developmental environment for days 06 (Eearly), developmental environment for days 6-90 (Elate), and all possible
interactions. The first two rows depict the numerator and denominator degrees of
freedom.
Sire x
Sire x Eearly x
Length
Sire
Eearly
Eearly
Elate
Elate
Elate
Num/
69
23
69
23
69
23
Den DF
965.81
323
965.81
323
965.81
323
<.0001 0.1344 0.0576 <.0001 0.1744 0.1879
Shape
Num/
5
15
5
15
5
15
5
Den DF
338
933.47
338
933.47
338
933.47
338
Morphology <.0001 <.0001 0.0006 <.0001 <.0001 0.0002 0.0823
58
Sire x
Eearly x
Elate
69
965.81
0.0063
15
933.47
0.0454
Univariate means
All univariate traits showed significant genetic variation in the form of sire effects
(Table 3.2). Eye-operculum distance and PC2 were the only variables influenced
solely by early oxygen. Maximum depth, operculum size, eye-operculum
distance, and PC2 showed genetic variation in plastic responses due to early
oxygen. Body area, max depth, eye area, and PC2 exhibited plasticity due to
later oxygen (Table 3.2). Body area, maximum depth, operculum size, and PC2
exhibited significant genotypic specific variation in plastic responses due to later
oxygen (Table 3.2). Averaging across genotypes, maximum depth, eye area, and
PC1 showed plastic responses to later oxygen which were conditional on early
oxygen (Table 3.2). In all 3 instances, fish raised in early normoxia showed little
or no response to switching to hypoxia. For example, in early hypoxia, however,
increased PC1 when switched to later normoxia (Fig. 3.2). PC1 also showed
genotype-specific variation in altered responses across early and later oxygen
environments (sire x Eearly x Elate). Individuals from sires A and D reared in early
normoxia showed an increase in PC1 when switched to hypoxia. Individuals from
sire B showed a decrease. Individuals from from sire C remained unresponsive.
Individuals from sires A, C, and D reared in early hypoxia showed an increase in
PC1 when switched to later normoxia. Individuals from sire B showed the
opposite trend (Fig. 3.3).
59
Table 3.2. Analysis of variance results for univariate traits (body area, maximum
depth, eye area, operculum height, eye-operculum distance, PC1, and PC2).
Sources of variation included sire, dam, developmental environment for days 0-6
(Eearly), developmental environment for days 6-90 (Elate), and all possible
interactions. Where appropriate, the effect of size was removed by including
length as a covariate Log transformations were applied prior to analysis for all
variables with the exceptions of PC1 and PC2.
DF
Body
Area
Max
Depth
Eye Area
Operc.
Ht.
EO Dist.
PC1
PC2
Elate
1
Sire x
Elate
3
Eearly x
Elate
1
Sire x
Eearly x
Elate
3
Length
1
Sire
3
Eearly
1
Sire x
Eearly
3
<.0001
0.0022
0.2904
0.4141
0.0002
0.0190
0.6926
0.0892
<.0001
<.0001
<.0001
<.0001
0.3077
0.0983
0.0269
0.4495
<.0001
0.0350
0.0045
0.4264
0.0369
0.0060
0.5638
0.0404
<.0001
<.0001
<.0001
0.0284
0.0107
<.0001
0.4251
0.0134
0.6962
0.0109
0.0005
0.0029
0.2133
<.0001
0.5858
0.0501
0.1754
0.0491
0.0068
0.1497
0.6038
0.0209
0.2899
0.9880
0.0281
0.2660
0.5733
0.0683
0.0209
0.1193
60
Phenotypic covariances
For pooled families, the eigenstructure varied across developmental treatments
(Table 3.3). P matrices for NN fish were equal to HH and NH fish (P = 0.1970,
0.0649 respectively). NN fish also shared 3 common principal components with
HN fish (P = 0.0554). P matrices for NH fish were equal HH fish (P = 0.5006).
However, they shared only one principal component with HN fish (P = 0.0701).
HN fish shared common principal components with HH fish (P = 0.0665).
Partitioning across all half sib families, there were a total of 120 matrix
comparisons, 40 which showed no shared structure (Table 3.4). This matrix
provides insight into which specific factors contributed to unequal variance
structure in this study. For example, individuals from sire C exposed to continual
hypoxia (CHH) equaled individuals from the same sire exposed to hypoxia
followed by normoxia (CHN). However, They shared no common variance
structure with individuals from the same sire raised under normoxia followed by
hypoxia (CNH). Thus, the first 6 days of oxygen proved more influential in
altering this particular genotype’s variance structure.
61
Table 3.3. Results of pair wise Flury comparisons for all half sib families pooled
by developmental treatment. The first row and column represents the treatment.
HH = hypoxia for 90 days, HN = hypoxia for days 0-6 and normoxia for days 690, NH = normoxia for days 0-6 and hypoxia for days 6-90, and NN = normoxia
for 90 days. For comparison results, E = matrix equality, C = common principal
components, and 1-3 = the number of shared principal components.
HN
NH
NN
HH
C
E
E
HN
NH
1
3
E
62
Table 3.4. Results of pair wise Flury comparisons for all half sib families divided
among all four developmental treatments. The first row and column represents
the sire, the second row and column represents oxygen conditions for days 0-6,
the third row and column represents oxygen for days 6-90, and the fourth row
and column represents the number of fish sampled. For comparison results, E =
matrix equality, P = proportionality, C = common principal components, 1-4 = the
number of shared principal components, and N = no shared structure.
A
A
A
B
B
B
B
C
C
C
C
D
D
D
D
H
N
N
H
H
N
N
H
H
N
N
H
H
N
N
N
H
N
H
N
H
N
H
N
H
N
H
N
H
N
23
20
24
25
31
31
30
23
23
24
29
27
28
27
25
A
H
H
19
2
E
E
N
N
2
1
N
1
1
1
1
2
1
1
A
H
N
23
A
N
H
20
A
N
N
24
B
H
H
25
B
H
N
31
B
N
H
31
B
N
N
30
C
H
H
23
C
H
N
23
C
N
H
24
C
N
N
29
D
H
H
27
D
H
N
28
D
N
H
27
1
C
1
1
1
N
E
E
N
C
1
N
C
1
E
1
C
E
E
1
1
N
N
E
N
1
N
E
C
P
E
C
E
N
E
E
C
4
N
2
1
E
1
E
N
N
1
N
1
N
E
E
E
2
N
N
1
N
C
N
E
E
1
N
E
1
N
N
N
E
C
N
C
1
C
C
C
E
N
E
N
C
E
C
N
E
N
2
1
N
N
N
N
N
N
E
E
E
N
N
P
N
N
N
1
63
Shape
Multivariate analysis revealed that body shape showed significant genotypic
variation (sire). Shape also showed plasticity to due later oxygen E late. There was
also significant genotype-specific cross-environmental responses to oxygen
changes (Table 3.1). Procustes distances comparing individuals pooled by
environments revealed that HH fish differed in shape compared to HN and NN
fish. NH fish also significantly differed in shape compared to HN and NN fish
(Table 3.5). In a specific comparison, NN fish differed from HH fish by dorsal
transpositions in landmarks 2, 3, and 12 (Fig. 3.5).
64
0.8
0.6
PC1
0.4
0.2
0.0
-0.2
-0.4
-0.6
Normoxia
Hypoxia
Fig. 3.2. The influence of developmental oxygen on PC1. Dashed and solid lines
represent fish raised in hypoxia and normoxia for days 0-6, respectively. The x
axis represents the oxygen environment for days 6-90. Error bars represent
standard errors.
65
1.5
1.0
PC1
0.5
Sire A
Sire B
Sire C
Sire D
0.0
-0.5
-1.0
-1.5
Normoxia
Hypoxia
Fig. 3.3. The influence of developmental oxygen on PC1 for half sib families A-D.
Dashed and solid lines represent fish raised in hypoxia and normoxia for days 06, respectively. The horizontal axis represents the oxygen environment for days
6-90. Error bars represent standard errors.
66
Table 3.5. Pairwise Procustes distances between environmental treatments. The
first row and column represents the treatment. HH = hypoxia for 90 days, HN =
hypoxia for days 0-6 and normoxia for days 6-90, NH = normoxia for days 0-6
and hypoxia for days 6-90, and NN = normoxia for 90 days. Asterisks indicate
level of significance (* = <0.05, † = <0.01).
HH
HN
NH
HN 0.0098*
NH 0.0081 0.0107†
NN 0.0117† 0.0062 0.0128†
67
0.8
NH
CV2 (16.1%)
0.6
0.4
0.2
0.0
HN
NN
-0.2
HH
-0.4
-0.6
-1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2
CV1 (76.8%)
Fig. 3.4. Effect of developmental environment (HH, HN, NH, NN) on the
canonical variate scores. Mean + SE of the canonical scores are shown.
68
Fig. 3.5. Comparison of body shape of HH vs NN fish. Vectors depict the
components of shape change based on a discriminant analysis.
69
Procustes distances comparing individuals pooled by sire and
environments revealed sire, Eearly, Elate, and all combinations of factors
contributed significantly to variation in body shape. Of the 120 comparisons, 82
differed in shape (Table 3.6). Variation in shape due to Elate was evident along
the first canonical axis (76.8% explained variance). Variation along the second
canonical axis (16.1% explained variation) was due to Eearly (Fig. 3.4).
Landmark comparisons reveal more specific effects of developmental
treatments. Across treatments, shape variation was mainly attributed to
landmarks 2, 3, 8, and 9. When Elate was switched to normoxia, individuals
tended to produce deeper profiles. This is evident in HH vs. HN, HH vs. NN, and
NH vs. NN comparisons. HH vs. NH and HN vs. NN comparisons reveal that the
effects of different early environments were not as pronounced (appendix).
Completely opposite developmental regimes (HN vs. NH) resulted in ventral
transpositions at landmarks 8 and 9 and a slight dorsolateral transposition in
landmark 3.
70
71
Discussion
The changing oxygen conditions applied during this experiment are within
those experienced during the ontogeny of zebrafish. Zebrafish are subjected to
variation in a number of environmental factors including oxygen, temperature,
and turbidity (Talwar and Jhingran 1991). Their association with rice fields
(Spence at al., 2008) makes them likely to experience anthropogenic hypoxia.
Laboratory tests using native soils as substrate have shown that conditions reach
anoxia with 96 hours (Strecker 2011). Padilla and Roth (2001) demonstrated that
zebrafish embryos can survive and recover from anoxia for 24 hours of
development. These findings make it reasonable to assume that zebrafish larvae
are adapted to developmental hypoxia. Following fertilization, zebrafish embryos
remain sessile for approximately 5 days, after which they can seek more optimal
conditions. Their developmental regime therefore makes it likely that larval and
adult oxygen environments can differ. The early developmental window utilized in
this experiment (days 0-6) therefore provides an ecologically relevant example of
the impact of opposing larval and adult oxygen environments.
Ontogenetic oxygen changes clearly alter body shape and phenotypic
integration in zebrafish. For many of these changes, the response to later
developmental oxygen is contingent upon the oxygen condition early during
development. These results are seen in significant Eearly x Elate interactions for
eye area, maximum depth, and PC1 (Table 3.2). Furthermore, numerous traits
show significant genetic variation for these responses to oxygen changes in the
form of sire x Eearly x Elate interactions (Tables 3.1 & 3.2). For zebrafish, which
72
inhabit a wide variety of habitats (Talwar and Jhingran 1991), this may have
profound implications on morphological evolution.
Oxygen is known to have direct effects on fish morphology. Hypoxia has
been shown to alter gill morphology in sea bass (Rinaldi et al. 2005), carp (Sollid
et al., 2003), African cichlids (Chapman et al. 2008), and sailfin mollies
(Timmerman and Chapman 2004). Chapman et al (2008) found skull variation
correlated with gill plasticity in African cichlids. Although zebrafish gills are not
fully developed by day 14 (Vulesevic and Perry 2006), neural epithelial cells in
primordial gill filaments are actively expressed at 5 d.p.f. with associated
hyperventilatory responses to hypoxia (Jonz and Nurse 2005). We found
variation in jaw and gill covering morphology across treatments (Fig. 3.5;
landmarks 10, 11, 12, & 13). While we did directly assess skeletal morphology,
our results indicate the presence of oxygen-mediated morphological plasticity.
The influence of oxygen, however, was contingent on the unique combination of
oxygen environments.
On average, variation in body shape was more influenced by later oxygen
(Table 3.1). Hypoxia has been shown to decrease feeding in bass (Pichavant et
al. 2001) salmon (Jobling, 1994) and turbot (Pichavant et al. 2000). Metabolic
alterations including reduced aerobic respiration (Mohamed and Kutty, 1983;
Pincetich et al. 2005), reduced protein synthesis (Storey 1988), and reduced
lipolytic rates (Van Raaij et al. 1996) are common responses to hypoxia across
species. While early oxygen did not solely significantly alter body shape (Table
3.1), partitioning across genotypes reveals that some genotypes proved more
73
sensitive to early oxygen conditions. Individuals from sires A and D differed in
body shape despite sharing a common later oxygen environment (AHH vs ANH,
DHH vs DNH) (Table 3.5). The significant sire x Eearly x Elate interaction for body
shape (Table 3.1) highlights the importance of both early and later oxygen
environments in determining body shape. The sire component, however,
indicates genetic variation for these responses. In other words, some genotypes
were more influenced by early than later oxygen. This is evident in maximum
depth, where the direction and magnitude of the response to switching
environments were genotype specific (Fig. 3.6; appendix).
Flury comparisons revealed a number of P matrices that shared no
common structure (Table 3.4). These inequalities resulted from all sources of
variation, including their various combinations. These results indicate that
phenotypic (and likely genetic) covariance structures within populations can be
altered by numerous sources of input throughout ontogeny. Plasticity in character
correlations such as these may provide the raw material for natural selection to
optimize integrated phenotypes.
G matrices have been found to be fairly stable across closely related
species and different at higher taxanomic levels (Reviewed in Steppan et al.
2002). The observed variation covariance structure in our study provides novel
insight into the role of ecology and development in guiding micro and
macroevolutionary patterns. Ecological input, as a guiding force of development,
can redirect or canalize integrated phenotypic development in manners
dependent upon the contexts of both early and later developmental conditions
74
contingent upon specific genotypes. PC1, for example, displays both plastic and
canalized responses (Fig. 3.7). While individuals raised in early normoxia exhibit
no response when switched to hypoxia, individuals raised in early hypoxia
increase eye area when switched to hypoxia. Partitioning across genotypes,
however, reveals that the level of canalization varies across genotypes (Fig. 8).
Given the diverse habitats of zebrafish as well as their wide distribution (Talwar
and Jhingran 1991), our results indicate the potential for G and P to evolve as
developmental plasticity varies within populations.
Recent studies have addressed the consequences early developmental
conditions have on adult fitness (Hales and Barker 2001; Relyea 2003;
Hoverman and Relyea 2008). Plastic responses to early conditions have been
shown to increase adult fitness when the inducing environment predicts future
conditions (Moran 1992; Kingslover and Huey 1998; Relyea 2011). Recent
interest has focused on the ramifications of mismatches between juvenile and
adult developmental environments (Monaghan 2008). Hypotheses have been
proposed modeling fitness outcomes when these conditions prevail. The silver
spoon hypothesis predicts that individuals raised in optimal conditions will always
have a fitness advantage regardless of later conditions (Grafen 1988). The
environmental matching hypothesis predicts that individuals will always perform
best in the environment in which they were raised and do poorly in the opposite
environment (Monaghan 2008). While these models may apply to populations in
general, they do not account for potential variation within populations. Our results
demonstrate that genotypes respond uniquely to mismatches between juvenile
75
and adult oxygen environments. For instance, while PC1 of some genotypes is
more canalized by early oxygen, for others PC1 responds to later oxygen. The
direction and magnitude of plastic response, however, is contingent upon unique
combinations of subsequent developmental environments (Fig. 3.3). This
indicates that knowledge of both early and later developmental environments is
critical in determining proximate causes of observed phenotypic outcomes. Most
importantly, genotype can dictate the how canalized or plastic development in
changeable environments can be.
We found that zebrafish body shape and integrated morphology are
modified by changes in developmental oxygen during development and the
nature of these modifications is contingent on the organism’s genotype. The
interaction between sire, early, and later oxygen environments results in one of
the most early noted examples of G x E x E interactions. Genetic variation
therefore plays a critical role in determining the level of canalization within a
population during environmental change. Methodologically, we highlight the
importance of incorporating dynamic environmental histories into quantitative
genetic studies. Variation in the way genotypes respond to environmental change
during development provides critical insight into the evolution of phenotypes in
complex environments. While this study analyzes morphological responses,
further studies should seek to test hypotheses regarding fitness and adaptive
responses.
76
Conclusions and evolutionary implications
Figure 3.3 (PC1; overall morphology) represents a modified reaction norm
described in chapter 1. While it demonstrates that that nature of interactions
between subsequent developmental environments varies across sires, it also
yields insight into the evolutionary potential of zebrafish morphology under
changing oxygen conditions. Since the reaction norms are largely crossing with
changes in rank order across environments, selection would have just as much
variation in both later hypoxia and normoxia with which to select optimal
morphologies. While the variance structures of PC1 look qualitatively similar in
figure 3.5, they are actually quite different (Fig. 3.6). Plotting them indicates that
variance is highest when oxygen switches from hypoxia to normoxia. This result
is similar to chapter II, in which variance in cardiac output was highest when the
environment switched from hypoxia to normoxia. Also similar to chapter II, the
lowest variance is expressed in consistent hypoxia. This again implies that
hypoxia has been a major force in shaping morphology in zebrafish.
Acknowledgements
We than Tim Astrop for his help with analyzing morphology data.
77
Variance (PC1)
2.2
2.1
2.0
1.9
1.8
Later oxygen
Early oxygen
H
H
N
H
N
N
Figure 3.6. Variance due to all four combinations of oxygen environments.
78
CHAPTER IV
ONTOGENETIC OXYGEN CHANGES ALTER ZEBRAFISH BEHAVIOR,
PHYSIOLOGY, AND MORPHOLOGY
Introduction
In a given environment, the fitness of an organism is determined by the
manner in which it budgets many potential behaviors in response to
environmental conditions. Behavioral requirements such as predator avoidance,
grouping, foraging, and mating all present unique combinations of potentially
conflicting external factors that require context-specific assessment strategies
(Lima and Dill 1990; Vainikka et al. 2005). Behavioral responses are the result of
the underlying physiological properties of an organism’s sensory, cognitive, and
motor abilities (Alais et al. 2010). Examining the factors that influence the
development of these modules is critical to further understanding biological
complexity.
Development proceeds from cells to tissues to organs to systems, all of
which underlie the prerequisites for complex behaviors. While genetics provides
the basic instructions for this unfolding, environmental input may alter the course
of development (Pigliucci 1998). Developmental plasticity provides one
mechanism for organisms to buffer environmental change (West-Eberhard 2003).
Genotype-environment interactions provide the basis for natural selection to
79
optimize the most appropriate plastic responses (Schlichting and Pigliucci 1998).
Environmental conditions can potentially change at any point during
development. Recent studies have begun to characterize phenotypic
development as a consequence of changing environments. While some traits
display seemingly infinite responsiveness (Sollid and Nilsson 2006) , others may
be permanently “programmed” according to early developmental conditions
(Jones and Ozanne 2009). Both of these findings suggest a dynamic relationship
between environment and development. Similar developmental mechanisms
across vertebrates pose tractable sources of investigation for studying genetic
and environmental sources of both physiological development and evolution
(Burggren 2000).
Organisms are comprised of suites of traits which may be correlated
(Gould and Lewontin 1979). These correlations can be due to genetic,
developmental, and functional links (Pigliucci 2003). Underlying physiological
mechanisms provide a proximate target for understanding the integrated nature
of biological development. Hormones, for example, may co-regulate the
physiology, morphology, and behavior of developing organisms (Ketterson et al.
2009). Hormone action can be modulated through external programming during
early development. The hypothalamic-pituitary-adrenal axis (HPA), for example,
can be irreversibly programmed via early nutrition, alcohol exposure, or stressful
conditions (Hellemans et al. 2010; Jones and Ozanne 2009; Tanaka et al. 2010).
While the implications of early epigenetic reprogramming on adult behavior and
80
physiology have been well documented (Kapoor et al. 2006), the role of genetic
variation in these mechanisms requires further elucidation.
Hypoxia is a ubiquitous environmental factor affecting the development of
a wide range of fish species. In fish, hypoxia has been shown to alter genetic,
physiological, and behavioral outcomes (Wu 2002). Genetic responses include
the induction of glycolytic genes and the down regulation of structural and
metabolic proteins (Ton et al. 2003). Physiological responses include reduced
ATP concentration, increased androgen production, increased blood glucose,
and altered Na+-K+ concentrations (Smit and Hattingh 1978; Wu 2002; Silkin
and Silkina 2005). Behaviorally, fish reduce foraging, alter swimming behavior,
and reduce aggressive encounters (Bejda et al. 1987). Chronically hypoxic fish
also tend to be smaller and experience delayed development (Shang and Wu
2004). While the overall effect of hypoxia has been well documented, the
developmental consequences of changing oxygen have only recently been
explored. Developmentally hypoxic zebrafish, for example, spend more time
hiding in the presence of a mirror image regardless of the oxygen environment
they are tested in (Marks et al. 2005). Given the far-reaching effects of hypoxia
on the development, it is important to consider multiple traits spanning many
biological hierarchies.
Studies examining the role of changeable environments in directing
phenotypic development have only recently begun. On average, individuals can
compensate for poor early developmental conditions by accelerating
development if conditions become more favorable. The costs of these responses,
81
however, remain poorly understood (Metcalfe and Monaghan 2001). Qualitative
models predicting the consequences of such changes have been proposed
(Monaghan 2008), however, the evolutionary consequences of these models
cannot be examined until the underlying genetics of these responses have been
explored. One promising way to explore these relationships involves studying the
effects of shifting environmental conditions across separate genotypes within a
population. While a few studies have examined the potential for interactive
effects across developmental environments (Yang 1993; Marks et al. 2005;
Taborsky 2006; Kotrschal and Taborsky 2010) , none have included a genetic
component. Genotype-environment studies have provided quantitative
predictions on the evolution of plasticity. Recent environment-environment
studies, however, have provided only qualitative predictions about plasticity. If
there is genetic variation for environment-environment responses, this would
provide natural selection with the raw material to select the response with the
lowest cost. It is therefore essential to add a genetic component to the growing
literature of environment-environment interactions.
In this study, we examine the effects of changing developmental oxygen
on zebrafish behavior, physiology, and morphology. We utilize a modified
quantitative genetic approach to assess the relative contributions of genotype,
and early and later oxygen conditions on integrated phenotypic development.
Zebrafish are native to Southeast Asia and inhabit a wide variety of habitats
including rivers, streams, and stagnant rice fields (Engeszer et al. 2007b). The
wide range of oxygen concentrations they encounter makes them a tractable
82
model for studying the effects of shifting oxygen conditions on development.
Zebrafish respond genetically and physiologically to hypoxia early in
development ( Ton et al. 2003; Bagatto 2005; Moore et al. 2006;). These
alterations can have lasting effects on adult physiology and physical performance
(Widmer et al. 2006). Our approach yields insights into role that changing
ecological parameters can play in shaping integrated adult morphology,
physiology, and behavior, as well as the underlying genetic contributions.
Materials and Methods
Animals
Adult zebrafish (Danio rerio) were obtained from Aquatica Tropicals (Plant
City, FL) and populations of wildtype and albino strains were maintained
according to standard procedures (Westerfield, 1994). Adults were maintained
and bred at 26 ± 0.5°C with a 14L : 10D light cycle.
Breeding design
Four wildtype males were randomly selected and mated to four randomly
selected albino females in a 4x4 full factorial design. Albino females were used
due to their large egg production and ease of sexing. Mating pairs were placed in
2 L containers lined with a marble substrate and supplied with a common water
source (Z-Mod housing system, Marine Biotech, Beverly, MA).
Developmental treatments
Approximately 50 eggs from each clutch were removed before the 8 cell
stage and randomly divided between developmental treatments. One treatment
consisted of normoxic water (dissolved oxygen > 6.0 mg O2 l-1) and the other
83
consisted of hypoxic water (dissolved oxygen 1.0 ± 0.2 mg O2 l-1). The apparatus
for maintaining developmental treatments is described in Marks et al. (2005).
Fish were raised in 154 x 35 x 23 cm tanks divided into 16 chambers. System
water was supplied to each chamber with screen covered holes providing
homogeneous conditions between neighboring chambers. Since food could not
travel among the chambers or reach the tank’s outflow, multiple daily feedings
were found to be excessive. Fish were therefore fed once daily with a custom
diet of ground standard zebrafish food with nutrient additives. Excess food was
siphoned from the bottom of each chamber daily. Following 30 days of
development (E0-30), half of the eggs from each treatment were randomly
selected and placed in the opposite environment. The remaining eggs were left
to develop in the same environment. Fish were left under these conditions for the
remainder of the experiment (90 days; E30-90). This resulted in four treatment
groups: fish raised in hypoxia for 90 days (HH), normoxia for 90 days (NN),
normoxia for 30 days followed by hypoxia for 60 days (NH), and hypoxia for 30
days followed by normoxia for 60 days (HN).
Data collection
Behavioral assay – We combined previous behavioral paradigms which
assessed fish shoaling (Godin and Morgan 1985; Krause and Godin 1994) and
novel stimulus response (Frost et al. 2007; Sneddon et al. 2003; Wright et al.
2003). At 90 days, fish were subjected to a behavioral assay which measured
their swimming and shoaling behavior before, during, and after the presentation
of a novel stimulus.
84
Fig. 4.1. Overview of behavioral arena. The starting area consists of a 5 cm
diameter PVC tube. After a one minute accommodation period, each subject was
released to explore the arena. The upper half represents the shoaling half with
shoaling tank containing 5 conspecifics. Lower half contains a plastic plant for
refuge.
85
For each behavioral session, subjects were randomly netted and placed in
a 5 cm PVC tube which was immersed vertically in the center of a 30.5 x 30.5 x
30.5 cm white, opaque polyethylene tank. At one side of the tank was a smaller
16 x 7.5 x 7.5 cm clear polyethylene tank containing 5 randomly selected
conspecifics. The conspecifics were chosen randomly each day from a separate
system raised under standard conditions. Along the opposite wall was a plastic
aquarium plant (see Fig. 4.1). Following a 1 minute accommodation period, the
PVC tube was lifted and the subject was allowed to freely explore the arena. The
behavioral experiment was conducted in 3 phases: 1) pre-stimulus, 2) stimulus
introduction, and 3) stimulus removal. Behavior was recorded with a Samsung
SC-HMX10C digital camcorder that was mounted 50 cm above the tank. An LED
light illuminated the tank from below. The arena was continuously supplied with
system water from the subject’s most recent developmental environment. Flow
was adjusted to supply fresh water to the arena without disturbing the behavior of
the fish.
Phase 1: Pre-stimulus - Positive shoaling behavior was initiated once the
subject spent 3 or more consecutive seconds within a 3 cm perimeter of the
conspecifics’ tank (see Fig. 4.1). If the subject did not shoal within 3 minutes of
filming, the experiment was terminated and the subject was labeled a nonshoaler. If the subject initiated shoaling, the subject was allowed to shoal or
explore the arena for 30 seconds and experiment proceeded to phase 2.
Phase 2: Stimulus introduction – Thirty seconds after shoaling was
initiated, a novel stimulus was introduced to the shoaling area. Stimuli of various
86
shapes and colors have been used as novel fright stimuli in previous fish studies
(Frost et al. 2007; Sneddon et al. 2003). Our stimulus consisted of five 14.2 g
lead weights wrapped in orange rubber, and was suspended just above the water
line prior to delivery. The experimenter manually released the weights, which
were all tied to a common releasing mechanism, resulting in simultaneous
deployment into the tank. The releasing mechanism was outside the tank and not
in the field of view of the subject. The weights were allowed to free fall and
designed to rest 2 cm above the arena floor. The weights were distributed evenly
around the perimeter of the shoaling tank, with 3 in the front and 1 on each side.
Following 30 seconds, the experiment proceeded to phase 3.
Phase 3: Stimulus removal - The experimenter slowly returned the weights
above the water line via the external mechanism. The subject was recorded for
another 60 seconds, at which time the behavioral assay was terminated.
Video analysis - On the video screen, a 3 cm perimeter was drawn around
the conspecific tank and defined as the shoaling area (see Fig. 4.1). A horizontal
line was drawn at the midpoint of the arena, dividing the arena into two sides.
The shoaling side was defined as the side with the shoaling tank and the other
side contained the plastic plant. Recorded dependent variables included whether
the subject shoaled (phase 1), latency to shoal (phase 1), percentage time
swimming (phases 1, 2, 3), percentage time on shoaling side of arena (phases 1,
2, 3), percentage time shoaling (phases 1, 2, 3), number of re-entries (per
second) to the shoaling side of the arena (phases 1, 2, 3), percentage time hiding
in the plant (phases 1, 2, 3), and reaction to the stimulus (phase 2). We defined
87
reaction to the stimulus as any sudden burst swimming (i.e. suddenly exiting of
the shoaling area) that occurred immediately following deployment of the
stimulus. Re-entry to the shoaling side of the arena was recorded if the entire
body length of the subject crossed the midpoint line and entered the plant side of
the arena followed by re-entry to the shoaling side. We define this behavior
herein as “crossing”.
Morphological measurements – Immediately following the behavioral
assay, the subject was euthanized with MS-222( 300 mg/ml tricaine methane
sulfonate buffered to a neutral pH with sodium bicarbonate), weighed, and
photographed on the subject’s right side with a length standard in each picture.
Photography was conducted with a Nikon D300 camera under standard lighting
conditions with a Kodak 18 percent reflectance grey background. We measured
standard length from most anterior point to the base of the hypural plate at
caudal flexion. Maximum depth was measured as the maximum dorsal-ventral
distance measured along the flank. All measurements were made using imageJ
(Version 1.42, NIH). Measurements were made 5 times and the mean was
recorded.
Blood glucose – Glucose measurements were made with a Nova Max
blood glucose monitor (Bedford, MA) according to manufacturer’s instructions.
Immediately following photography, the subject was bisected through the
pectoral girdle and the exposed dorsal artery at the posterior portion of the head
was applied to the test strip.
88
Statistics
Behavior – A three way mixed model repeated measures ANOVA was
performed to test for effects on overall shoaling behavior. Two individuals were
removed from this analysis due to failure of the alarm stimulus to properly deploy.
For purposes of this study, we were interested in quantifying genetic variation for
cross-environmental responses. Since sires donate purely genotypic effects and
dams donate both genetic and maternal effects, we dropped dam from all
models. This allowed us to assess purely additive genetic variance. Sire, early
oxygen for days 0-30 (E0-30), later oxygen for days 30-90 (E30-90), and all possible
interactions were included as potential sources of variation. Post hoc TukeyKramer analyses provided adjusted p values estimating the significance of all
factors at each stage of the experiment. Residuals were analyzed for normality
and transformations were applied as needed. We used chi-squared tests to
determine associations between both stimulus reactivity and shoaling tendency
and developmental treatments.
Morphology – To assess the effects of genetic and environmental variation
on overall size, we reduced the dimensionality of the data using principal
components analysis. Mass, maximum depth, and body length were included as
dependent variables. Subsequent scores from the first principal component were
used in a three way mixed model ANOVA. Sire, early oxygen for days 0-30 (E030),
later oxygen for days 30-90 (E30-90), and all possible interactions were
included as potential sources of variation.
89
Blood Glucose – Blood glucose did not differ between shoaling and nonshoaling fish (T-test P=0.28). For shoaling fish, no behavioral variables exhibited
a significant correlation with blood glucose. We therefore pooled shoaling and
non-shoaling fish for blood glucose analysis. We regressed all continuous
dependent variables (including mass) on blood glucose. Blood glucose correlated
positively with mass (P <0.0001) and negatively with percentage time spent
swimming (P=0.0006). These two factors were therefore included as covariates.
This removed the effects of mass and swimming on blood glucose in determining
genetic and environmental sources of variation. A three way mixed model
ANCOVA was used to estimate the effects of sire, early oxygen for days 0-30 (E030),
later oxygen for days 30-90 (E30-90), and all possible interactions on blood
glucose. All data analyses were performed in SAS software package (SAS
Institute, Cary, NC) using PROC MIXED (ANOVAs) and PROC FREQ (chisquared).
Results
Survival
After starting with equal egg numbers across treatments, we ended with
100 fish in HH, 100 fish in HN, 88 fish in NH, and 101 fish in NN. (Chi-squared
test revealed no association between treatments and mortality (chi-square =
1.86, P = 0.602)
Behavior
Shoaling tendency - Of the 389 fish in the study, 60% initiated shoaling
behavior. Of these, 28% were raised under consistent hypoxia (HH), 28% under
90
hypoxia followed by normoxia (HN), 23% under normoxia followed by hypoxia
(NH), and 21% under consistent normoxia (NN). Developmental oxygen
significantly altered shoaling tendency (chi-square = 8.76, P = 0.0326), with early
developmentally hypoxic fish (HH) shoaling most often. Latency to shoal was not
affected by developmental oxygen, however, we did detect significant variation
among genotypes (Table 4.2).
Swimming - Swimming significantly increased over time (see ANOVA
Table 4.1, Fig. 4.2) and there was variation in overall swimming time across
genotypes (Table 4.1). Early oxygen (E0-30) altered overall swimming time (Table
4.1), with early developmentally normoxic fish (NN & NH) swimming more than
early developmentally hypoxic fish (HH and HN) (refer to Fig. 4.2). Post hoc
Tukey-Kramer comparisons revealed significant effects of developmental oxygen
between and within phases of the experiment. Both early (E0-30) and later (E30-90)
oxygen altered swimming behavior. Overall, NH fish swam more than HN fish
(P=0.0203; Fig. 4.2). Both early normoxic (NN & NH) and early hypoxic (HH &
HN) fish increased swimming in presence of the stimulus (P=0.0018, 0.0279; Fig.
4.2). Later developmentally hypoxic fish (HH & NH) also increased swimming in
the presence of the stimulus (P=0.0067; Fig. 4.2). Later normoxic fish (NN, HN)
also increased swimming in response to stimulus presence (P=0.0098; figure
4.2).
91
Table 4.1. Mixed model repeated-measures ANOVA analyzing the effects of trial
phase, sire, early oxygen (E0-30), later oxygen (E30-90), and all possible
interactions on behaviors. Dependent variables included percentage time spent
swimming (swim), percentage time spent on shoaling side of the arena
(shoalside), percentage time spent shoaling (shoal), number of crosses between
shoaling and non-shoaling side of the arena per second (cross), and percentage
time spent hiding (hide). Since there were no significant interactions with other
factors, only the overall effect of trial phase is presented. Degrees of freedom
(numerator/denominator) are listed below each factor. F values are presented for
each factor. Significance level is indicated by notation; *<.05, **<.01, ***<.0001.
DF (N/D)
Swim
ShoalSide
Shoal
Cross
Hide
Phase
2/18
21.10***
7.91**
10.46**
2.5
1.56
Sire
3/23
5.54**
1.03
0.92
2.95
3.77*
E0-30
1/9
8.25*
0.24
0.00
9.44*
4.67
92
Sire x
E0-30
3/20
2.52
1.03
1.09
3.05
1.78
E30-90
1/8
6.26*
0.46
0.05
1.91
4.32
Sire x
E30-90
3/16
2.8
1.79
3.81*
0.89
1.36
E0-30 x
E30-90
1/8
0.15
0.00
0.02
2.78
0.60
Sire x
E0-30 x
E30-90
3/11
1.57
1.37
1.53
2.15
0.32
Table 4.2. Mixed model ANOVA analyzing the effects of sire, early oxygen (Env030),
later oxygen (Env30-90), and all possible interactions on blood glucose, PC1
(size), and latency to shoal (latency). Covariates for blood glucose included mass
and total time spent swimming (swimT). Degrees of freedom
(numerator/denominator) are listed below each factor. F values are presented for
each factor. Significance level is determined by notation; *<.05, **<.01, ***<.0001.
Sire x
Sire x
Sire x E0-30 x E0-30 x
Source
Mass SwimT Sire E0-30 E0-30 E30-90 E30-90 E30-90 E30-90
DF (N/D) 1/184 1/184 3/184 1/184 3/184 1/184 3/184 1/184 3/184
Glucose 21.39*** 4.91* 2.70* 1.68 1.35 19.37*** 0.63 1.68 0.65
DF (N/D)
3/372 1/372 3/372 1/372 3/372 1/372 3/372
PC1
0.82 32.38*** 0.80 23.82*** 0.30 0.02 0.42
DF (N/D)
3/214 1/214 3/214 1/214 3/214 1/214 3/214
2.95* 0.70 0.38 0.81 2.09 0.45 0.57
Latency
93
Fig. 4.2. Percentage time spent swimming before, during, and after stimulus
introduction as a function of four developmental environments. Solid circles
represent fish raised in early and later hypoxia (HH), open circles represent fish
raised in early hypoxia followed by later normoxia (HN), solid triangles represent
fish raised in early normoxia followed by later hypoxia (NH), and open triangles
represent fish raised in early and later normoxia (NN). For environmental factors,
early developmental oxygen (E0-30) is represented by shapes (circles = hypoxia,
triangles = normoxia). Shading indicates later developmental oxygen (E 30-90; solid
= hypoxia, open = normoxia). Error bars represent one standard error. Overall
differences between developmental oxygen are noted below symbol legend.
Significant differences between various combinations of developmental oxygen
treatments between phases are noted. Significance level is determined by
notation; *<.05, **<.01, ***<.0001.
94
Fig. 4.3. Percentage time spent on the shoaling side before, during, and after
stimulus introduction as a function of four developmental environments. Solid
circles represent fish raised in early and later hypoxia (HH), open circles
represent fish raised in early hypoxia followed by later normoxia (HN), solid
triangles represent fish raised in early normoxia followed by later hypoxia (NH),
and open triangles represent fish raised in early and later normoxia (NN). For
environmental factors, early developmental oxygen (E0-30) is represented by
shapes (circles = hypoxia, triangles = normoxia). Shading indicates later
developmental oxygen (E30-90; solid = hypoxia, open = normoxia). Error bars
represent one standard error. Significant differences between specific phases of
the trial are noted. Significance level is determined by notation; *<.05, **<.01,
***<.0001.
95
Shoaling - Developmental oxygen did not influence latency to shoal,
however, genotype did effect shoaling tendency (Table 4.1). As noted in table 4.1
and represented in figures 4.3 and 4.4, fish significantly increased the time spent
on the shoaling side of the arena and within the shoaling zone throughout the 3
phases of the experiment. E0-30 had no overall effect on the time spent on the
shoaling side during the experiment, yet it was noted that E30-90 altered time
spent shoaling as a function of genotype (see Table 1).
96
Fig. 4.4. Percentage time shoaling before, during, and after stimulus introduction
as a function of four developmental environments. Solid circles represent fish
raised in early and later hypoxia (HH), open circles represent fish raised in early
hypoxia followed by later normoxia (HN), solid triangles represent fish raised in
early normoxia followed by later hypoxia (NH), and open triangles represent fish
raised in early and later normoxia (NN). For environmental factors, early
developmental oxygen (E0-30) is represented by shapes (circles = hypoxia,
triangles = normoxia). Shading indicates later developmental oxygen (E30-90; solid
= hypoxia, open = normoxia). Error bars represent one standard error. Significant
differences between specific phases of the trial are noted. Significance level is
determined by notation; *<.05, **<.01, ***<.0001.
97
Reaction to stimulus – Of the fish that shoaled, 58% reacted to the
stimulus. Chi- squared results indicated no association between developmental
oxygen and reactivity. In other words, fish from all 4 developmental oxygen
environments were just as likely to react to the stimulus.
Crossing – Averaged across all treatment groups, crossing behavior did
not change throughout the 3 phases of the experiment (Table 4.1, Fig. 4.5).
However, E0-30 influenced the tendency to cross between shoaling and nonshoaling sides (Table 4.1), with early developmentally normoxic fish (NN & NH)
crossing more than early developmentally hypoxic fish (HH & HN) (Fig. 4.5). For
later developmentally normoxic fish (NN & HN), only those exposed to early
hypoxia decrease in crossing behavior (Tukey-Kramer adjusted P=0.047; Fig.
4.5). In response to the stimulus introduction, early developmentally normoxic
fish (NN & NH) crossed more than early developmentally hypoxic fish (HH, HN)
(Tukey-Kramer adjusted P=0.0386; Fig. 4.5).
98
Fig 4.5. Number of crosses between shoaling and non-shoaling sides of the
arena before, during, and after stimulus introduction as a function of four
developmental environments. Solid circles represent fish raised in early and later
hypoxia (HH), open circles represent fish raised in early hypoxia followed by later
normoxia (HN), solid triangles represent fish raised in early normoxia followed by
later hypoxia (NH), and open triangles represent fish raised in early and later
normoxia (NN). For environmental factors, early developmental oxygen (E0-30) is
represented by shapes (circles = hypoxia, triangles = normoxia). Shading
indicates later developmental oxygen (E30-90; solid = hypoxia, open = normoxia).
Error bars represent one standard error. Overall differences between
developmental oxygen are noted below symbol legend. Significant differences
between developmental oxygen treatments within specific phases of the trial are
noted. Significance level is determined by notation; *<.05, **<.01, ***<.0001.
99
Fig. 4.6. Percentage time spent hiding before, during, and after stimulus
introduction as a function of four developmental environments. Solid circles
represent fish raised in early and later hypoxia (HH), open circles represent fish
raised in early hypoxia followed by later normoxia (HN), solid triangles represent
fish raised in early normoxia followed by later hypoxia (NH), and open triangles
represent fish raised in early and later normoxia (NN). For environmental factors,
early developmental oxygen (E0-30) is represented by shapes (circles = hypoxia,
triangles = normoxia). Shading indicates later developmental oxygen (E30-90; solid
= hypoxia, open = normoxia). Error bars represent one standard error.
100
Hiding – Overall hiding behavior was not influenced by developmental
oxygen (Table 4.1, Fig. 4.6). We did, however, detect significant genetic variation
in overall hiding behavior (Table 4.1).
Blood glucose
Blood was successfully extracted from 202 fish. Following the behavioral
assay, later oxygen (E30-90) significantly altered blood glucose levels. Later
developmentally normoxic fish (NN & HN) had on average higher glucose levels
than later developmentally hypoxic fish (HH & NH) (Table 4.2, Fig. 4.7). Although
early developmentally normoxic fish (NN & NH) maintained higher overall blood
glucose regardless of later oxygen conditions (Fig. 4.6), the effect of E0-30 was
not significant (Table 4.2). Blood glucose levels did not differ between fish that
shoaled and fish that did not shoal (T-test; P=0.28).
101
Fig. 4.7. Blood glucose levels (mg/dl) as a function of four developmental
environments. Solid symbols and line represent fish raised in early normoxia.
Open symbols and dashed line represent fish raised on early hypoxia. The x axis
represents later oxygen conditions. Error bars represent one standard error.
102
Morphology
Mass, maximum depth, and body length all loaded positively in the first
principal component (PC1). PC1 accounted for 90% of the variation. Early
developmentally normoxic (NN & NH) fish were significantly larger regardless of
E30-90 (Table 2, Fig. 4.8). Later developmentally normoxic (NN & HN) fish were
also significantly larger than later developmentally hypoxic (HH & NH) (Table 4.2,
Fig. 4.8).
103
Fig. 4.8. Principal component 1 (body size) as a function of four developmental
environments. Solid symbols and line represent fish raised in early normoxia.
Open symbols and dashed line represent fish raised on early hypoxia. The x axis
represents later oxygen conditions. Error bars represent one standard error.
104
Discussion
Environmental conditions during early development have been shown to
permanently alter developmental outcomes. Environmental mismatches (i.e.
when developmental conditions fail to accurately predict future conditions) can
yield pathological results. One excellent illustration lies in early metabolic
reprogramming. Fetuses exposed to malnutrition have been shown develop
reduced insulin sensitivity, and upon exposure to improved nutrition later in life,
reduced insulin action leads to physiological pathologies such as obesity and
diabetes (Jones and Ozanne 2009). Numerous examples have been outlined in
both human and animal models (review in Jones and Ozanne 2009). While the
environmental and physiological aspects of such responses are well-studied, the
underlying genetics of these responses remain poorly understood. Specifically,
are these responses the same across all genotypes? Are some genotypes more
constrained by early conditions than others? Here, we have attempted to address
these questions by analyzing the effects of early and later developmental
conditions on phenotypic development across genotypes.
Shoaling
In fish, shoaling offers many advantages, including but not limited to
avoiding predators and optimizing food intake (Godin and Morgan 1985; Pitcher
et al. 1982). Since zebrafish inhabit a wide variety of habitats, it is likely that
shoaling behavior has been highly selected upon across a wide range of oxygen
environments (Engeszer et al. 2007b). While shoaling steadily increased
throughout the experiment across all treatments, we found little influence of
105
genotype or developmental oxygen on shoaling behavior. This concurs with
previous findings in zebrafish. Wright et al. (2003) also found no variation in
shoaling behavior of zebrafish across different populations. We did, however, find
significant genotype-environment responses to shoaling behavior resulting from
E30-90. These results indicate the presence of genetic variation in plastic
responses to more recent oxygen conditions. In this sense, E0-30 plays no role in
shaping shoaling behavior. E30-90, however, effect shoaling differently across
genotypes. This indicates shoaling behavior is not constrained by early
environment and exhibits plasticity to later conditions. While zebrafish begin
shoaling during the early larvae stage, shoaling preferences remain plastic until
later development (Engeszer et al. 2007a). Our results confirm the existence of
later windows of plasticity for shoaling behavior.
Swimming
Fish exposed to early normoxia spent the most time swimming, most
noticeably during stimulus introduction. Interestingly, early normoxic fish exposed
to later hypoxia (NH) spent the most time swimming both during and after
stimulus introduction. Hypoxia has been shown to increase swimming activity in a
number of species (Bejda et al. 1987; Domenici et al. 2000; Parsons and Carlson
1998), which might explain in part why NH fish maintained the highest swimming
levels during the experiment. Increased swimming in this group correlates with
acute responses to hypoxia in previous studies (Bejda et al. 1987; Domenici et
al. 2000).
106
Crossing
Crossing behavior was used as an index of responsiveness to the novel
stimulus. Shoaling, darting, and freezing are all intraspecific behaviors occurring
in fish in response to alarm stimuli (Brown and Godin 1999). Therefore, during
stimulus introduction, we expected variance in the tendency to retreat (cross) or
remain shoaling during stimulus presence. While crossing behavior remained
steady across treatments before stimulus introduction, crossing tendencies
became more divergent as the trial progressed. Developmentally normoxic fish
(NN) exhibited significantly increased crossing behavior in the presence of the
novel stimulus. This group did so while maintaining shoaling times consistent
with the other three groups during this phase. These results indicate quite a
dynamic behavioral regimen. As with increased swimming, the increased
crossing activity seen in developmentally normoxic fish would also be
metabolically costly. The underlying physiologies of developmentally normoxic
vs hypoxic fish may provide insight into these divergent behaviors, as discussed
later
Hiding
Curiously, developmental oxygen did not alter hiding behavior. We
previously demonstrated that fish raised in chronic hypoxia spent more time
hiding than those from chronic normoxia regardless of the oxygen environment
which they were tested in (Marks et al. 2005). The fish from the current
experiment, however, were switched to the opposite oxygen environment much
earlier during development. These results indicate that timing of environmental
107
change may be critical in determining the nature of adult phenotypes. In our
previous study (2005), we hypothesized that fish raised in chronic hypoxia
engaged in more hiding behavior as an energy saving strategy. Concurrent with
our lab’s findings that fish raised in chronic hypoxia attain lower swimming
speeds and produce lower lactate levels (Widmer et al. 2006), we hypothesized
that hypoxia elicits metabolic depression in zebrafish. In our current study,
however, hiding time did not differ across treatments. Also, hypoxic fish that
began development in normoxia (NH) spent the most time swimming during the
experiment. These results indicate the potential for early favorable environments,
such as normoxia, to impart optimal developmental trajectories regardless of
later environmental conditions. One such theory, the silver spoon hypothesis,
states that individuals raised in optimal conditions maintain fitness advantages
throughout life, regardless of later conditions (Grafen 1988). Compared with our
previous findings, our current results highlight the potential for such responses to
be contingent upon timing of environmental change.
Proximate mechanisms of developmental plasticity
Phenotypic variation underlies the basic target for evolution by natural
selection to optimize traits within populations (Darwin 1859). More recently,
selection on environmentally-induced variation has gained notoriety (Hall 2001;
Schlichting and Smith 2002; Lande 2009). Zebrafish inhabit a wide range of
habitats with numerous variable ecological factors (Engeszer et al. 2007b).
Ecological factors such as oxygen content, community structure, and habitat
complexity can influence brain size in fish (Chapman and Hulen 2001; Pollen et
108
al. 2007; Shumway 2008). Furthermore, variation in hypoxia-induced brain size
plasticity among populations of cichlids has been detected (Chapman et al.
2008). Links between brain morphology and behavior in fish are well documented
(Kotrschal et al. 1998; Healy and Rowe 2007). For example, trout with larger
telencephalons exhibit more active foraging strategies (Wilson and McLaughlin
2010). Links between physiology and behavior have also been studied in fish
(Scott and Sloman 2004). In a specific example, bold carp exhibit lower glucose,
lactate, and metabolic rates than timid carp (Huntingford et al. 2010). While we
did not assess brain morphology, physiological variation may provide one
proximate mechanism for the observed behavioral variation. In our study,
developmental oxygen altered stimulus responsiveness (crossing behavior) and
time spent swimming. Inter-population variation in responsiveness to novel
stimuli have been observed in zebrafish (Wright et al. 2003). Such differences
among populations would require genetic or environmentally-induced variation
for selection to act upon. The observed ability of developmental oxygen to alter
behavior in this study embodies one such proximate example.
Zebrafish, like many other fish species, utilize external fertilization.
Developing embryos are therefore subject to environmental insults such as
hypoxia. We routinely raise zebrafish embryos in oxygen conditions as low as 1.5
mg/L (approximately 17% dissolved oxygen at 25˚C and 1000 m above sea level)
without severe pathologies (Marks et al. 2005; Widmer et al. 2006). Although little
is known about the natural conditions and their associated variation during
zebrafish development, laboratory tests using native zebrafish sediments as
109
substrate revealed conditions nearing anoxia within 96 hours post fertilization
(hpf) (Strecker et al. 2011). Additionally, zebrafish embryos have been shown to
be able to survive anoxia for up to 24 hours (Padilla and Roth 2001). These
findings suggest that zebrafish could be experiencing extremely low oxygen
levels during development. Chronic hypoxia has been shown to elicit a two-fold
increase in whole-body cortisol concentrations in larval trout (Fuzzen et al. 2011).
In sticklebacks, elevated cortisol exposure during development elicited lasting
effects on later metabolism and behavior (Giesing et al. 2010). Early
developmentally hypoxic fish (HH & HN) displayed significantly different behavior
than early normoxic fish (NN & NH). In addition to fewer crossings between sides
of the arena, they also consistently swam less time than NN and NH fish. Given
previous studies, we feel the relationship between developmental hypoxia,
cortisol production, and their influence adult behavior can provide promising
insight into proximate causes of animal behavior.
While the overreaching effect of developmental stressors has been
established in a number of animal models (McArdle et al. 2006), the role genetic
variation plays during development has received little attention. In this study, we
used crossing behavior as an indicator of reactivity to novel stimulus introduction.
We found that early developmentally hypoxic fish (HH & HN) crossed among
arena sides less often and were thus more less to the stimulus. E 0-30 thus altered
the reactivity of fish to a novel stimulus in this study. Our work highlights the
potential for early developmental conditions to alter adult behavioral phenotypes.
110
Developmental oxygen clearly altered circulating glucose concentrations
in this study. The overall mean glucose was 65 mg/dl. A previous study reported
means between 60-73 mg/dl in untreated zebrafish fish (Eames et al. 2010).
Increased blood glucose is a reliable indicator of stress in fish (Iwama et al.
1995). In our study, shoalers and non-shoalers did not differ in blood glucose
levels. Non-shoalers were not exposed to the novel stimulus during the
behavioral trials. Therefore, we deduce that the novel stimulus did not elicit a
significant stress response. The only correlates with glucose in this study were
mass (positive) and total time spent swimming (negative). E30-90 significantly
altered blood glucose levels, with glucose increasing in normoxia. Despite the
effect of E30-90, HN fish did not catch up to NN fish in terms of blood glucose
(figure 7). This suggests that E0-30 fundamentally altered glucose metabolism.
Upon acute exposure to hypoxia, fish typically mobilize more blood glucose for
anaerobic ATP production (Chippari-Gomes et al. 2005; MacCormack and
Driedzic 2007; O’Connor et al. 2011). Chronic hypoxia in this study, however,
lowered blood glucose. Chronic developmental hypoxia in zebrafish has been
shown to alter physical performance and underlying physiology. Developmentally
hypoxic zebrafish not only attain lower swimming velocities than their normoxic
counterparts, but they also produce less lactate (Widmer et al. 2006). These
collective findings (lower glucose, lower lactate, lower stimulus reactivity) indicate
unique physiological responses contrary to previous studies (Chippari-Gomes et
al. 2005; MacCormack and Driedzic 2007; O’Connor et al. 2011) and require
furtther elucidation. While shoaling as much as their developmentally normoxic
111
counterparts, hypoxic fish exhibited lowered responsiveness to the novel
stimulus. Therefore, while routine shoaling and swimming behaviors may be
physiologically compliant, the reactivity to potential stressors such as the novel
stimulus may be too energetically costly.
Morphologically, NN and NH fish were significantly larger than HH and HN
fish. Fish exposed to later normoxia (NN & HN) were also on average larger than
fish exposed to later hypoxia (HH & NH). These results indicate a steadily
steeper developmental trajectory defined by both E0-30 and E30-90. Small body
size may be an adaptive feature of hypoxic survival. Smaller individuals have
been shown to be more tolerant to hypoxia than their larger counterparts in both
bass and sea bream (Burleson et al. 2001; Cerezo and García García 2004).
Indeed the general consensus is that smaller fish are more tolerant to hypoxia
(Doudoroff 1970; Lowe-Jinde and Niimi 1983). This is due to a number of
reasons. One factor is that increasing ventilation is more costly for larger fish
(Jones 1971). Secondly, studies have noted an inverse relationship between
body size and hematocrit (Lowe-Jinde and Niimi 1983; Zanuy and Carrillo 1985).
Smaller fish also have more relative gill area vs body weight than larger fish
(Hughes 1984). Smaller fish are also observed to be outcompeted by larger fish
for more favorable habitats (Coutant 1987). Smaller fish may also be forced to
less favorable waters to avoid predation (Werner and Hall 1988). These
collective factors indicate that the smaller overall size of HH and HN fish in this
study may have been a result of past selection for adaptive developmental
responses to hypoxia. Whether smaller body size in this case is due to selection
112
(i.e. adaptive) or physiological constraint (i.e. non-adaptive) is not within the
scope of this study and warrants further investigation.
It is worth noting that the effect of E0-30 on size remained, with HH and HN
fish never “catching up” to NN and NH fish in terms of size. This was true despite
plentiful food (daily excess food removal was required for all treatments). Fish
have been shown to fully or even overcompensate for growth deficits when
switched from unfavorable to favorable conditions (Noltie and Wang 1997).
Periods of rapid growth, however, can be energetically costly, reduce fitness, and
may therefore be selected against (Ali et al. 2003). The tendency to retain a
steady growth pattern in this study may likely be a function of changing
conditions during development. While we characterize development as a function
of one environmental change, the reality is environments can change frequently
during development. Dissolved oxygen can fluctuate between normoxia and
hypoxia due to daylight photosynthesis and night time respiration (Tyler 2007). If
individuals starting out in hypoxia began increasing growth when switched to
normoxia, the response would be unfavorable if the environment changed back
to hypoxia. Simply put, it may be costly to continually respond to environmental
change. A general developmental trajectory, established during early
development, may be the best strategy. While shrimp respond to actuate
hypoxia, gene expression is shown to return to pre-hypoxia levels upon exposure
to cyclic hypoxia (Brown-Peterson et al. 2008). This may provide direct evidence
for the costs associated with repeated reactions to environmental change.
113
Fish size may have also influenced shoaling behavior. Across species,
fish are observed to shoal with similar sized fish in both lab and field populations
(Hoare et al. 2000). Since conspecific shoaling fish were chosen at random from
lab stocks each day, we had no way to quantify the relationship between their
size and test subject size. Also, since different conspecifics were randomly
chosen each day, any effect of size would have been balanced throughout the
experiment. When we regressed mass against individual behavioral variables,
however, mass showed no significant correlation with any behaviors. This
indicates that, once shoaling behavior was initiated, fish size had no influence on
subsequent behavior. With no observed covariation between mass and any of
the measured variables, we are confident the variation in observed behaviors
was influenced primarily by genotype and developmental oxygen.
There was a significant positive correlation between mass and blood
glucose. Larger individuals had higher blood glucose values. Larger fish typically
have higher glycogen stores, particularly in white muscle tissue (Ferguson et al.
1993) . This provides larger fish with an increased capacity to engage in
anaerobically supported burst swimming (Kieffer 2000). In an endurance study,
developmentally nornoxic fish produced more lactate than their hypoxic
counterparts, suggesting an increased capacity for anaerobic respiration for
normoxic fish (Widmer et al. 2006). Reduced feeding may have also influenced
body size and blood glucose. Hypoxia has been shown to decrease feeding in
bass (Pichavant et al. 2001) salmon (Jobling, 1994) and turbot (Pichavant et al.
2000). These responses, however, tend to occur with acute exposure to hypoxia.
114
Since fish exposed to early normoxia (NN and NH) were significantly larger than
HH and HN fish, it is likely that the observed correlation between high glucose
and high body mass is driven primarly by E0-30 .
Conclusion – We have provided insight into the consequences of
changing oxygen during zebrafish development. Our unique design incorporating
all possible combinations of developmental environments reveals the relative
impact of sequential environmental conditions in shaping developmental
outcomes. Furthermore, by considering the effect of environmental fluctuation
across separate genotypes, we add insight into the evolutionary potential of
behavioral, physiological, and morphological plasticity in complex environments.
Our approach offers the flexibility to explore the consequences of shifting
environmental factors at any point during development. This can address
important questions regarding critical windows of development. We found
morphological (body size) and behavioral (time spent swimming) factors to be
constrained by early oxygen conditions. These results are indicative of the
numerous studies highlighting the lasting effects of early developmental
conditions (McArdle et al. 2006). Epigenetic reprogramming, for example, is
believed to occur as a result of certain stimuli occurring during a critical point in
development (Clark 1997). Despite early conditions, shoaling behavior and blood
glucose levels remained reactive to E30-90. This insight was available due to the
partitioning of the developmental environment into discrete components. We
propose the advancement of this technique to analyze the influence of early
115
developmental conditions on further plastic responsiveness in other traits and
systems.
Conclusions and evolutionary implications
Contrary to the previous chapters, we found little evidence of interactions
between subsequent developmental environments. Variation in time spent
swimming, stimulus reactivity, and size was constrained by early oxygen
conditions. Fish raised in early normoxia, for example, maintained larger body
sizes regardless of later oxygen conditions. This implies the early environment
can be critical in determining zebrafish size and behavior. Furthermore, with no
significant interaction between sire and early oxygen, this indicates that there is
little genotypic variation in plastic responsiveness due to early oxygen. Given that
zebrafish are likely to encounter changing oxygen during development, this
indicates that selection has likely optimized plasticity in these traits. The variance
structures across all four treatments confirms this observation. Figure 4.9 shows
little change in variance due to later oxygen. Rather, early oxygen contributes to
the changes in variance, with variance highest in early normoxia. Similar to
chapters II and III, hypoxia contributes to decreased variance. This suggests that
hypoxia is also a strong selective force in shaping zebrafish size.
Acknowledgements
We thank Kevin P. Kaut for commenting on drafts of this manuscript.
116
2.0
Variance (PC1)
1.8
1.6
1.4
1.2
1.0
0.8
0.6
Later oxygen
Early oxygen
H
H
N
H
N
N
Figure 4.9. Variance due to all four combinations of oxygen environments.
117
CHAPTER V
THE INFLUENCE OF ONTOGENETIC DIETARY CHANGES ON
ZEBRAFISH SIZE AND SWIMMING PERFORMANCE
Introduction
Phenotypic flexibility is critical in determining fitness. As conditions change
during ontogeny, continued responsiveness is necessary to meet the demands of
the environment. Phenotypic responses can be irreversibly sculpted during
critical points of ontogeny (Burggren and Reyna 2011). For example, we
previously that long term constraints on aggression, swimming performance, and
lactate production are a product of early oxygen environment (Marks et al. 2005;
Widmer et al. 2006). We found that fish tested in normoxia displayed phenotypes
that were altered or constrained by development in early hypoxia.
Studies demonstrating genetic variation for responses to environmental
change are ubiquitous (DeWitt and Scheiner 2004). This variation is an important
factor in shaping evolutionary dynamics (West-Eberhard 2005). Much of our
current understanding comes from studies tracking genotypes across one
instance of environmental change. As the environment can change multiple times
during ontogeny, it becomes increasingly important to characterize the role
genetic variation plays in more complex environments.
118
Food availability is a critical factor in shaping animal development. Food
manipulation studies have demonstrated significant effects of dietary change on
fish metabolism and swimming performance (Alsop and Wood 1997; Beamish et
al. 1989). Food type and availability vary and such changes are associated with
seasons (Wu and Culver 1992), presence of competitors (Osenberg et al. 1992),
and microhabitat use (García-Berthou 1999). Ontogenetic dietary shifts have
been demonstrated to increase cognitive performance in the cichlid Simochromis
pleurospilus (Kotrschal and Taborsky 2010). Fish that were switched from high
ration to low ration and low ration to high ration diets outperformed conspecifics
maintained on steady high or low ration diets. While dietary change clearly
influences developmental outcomes in fish, the role genetic variation plays in
influencing these altered responses in more complex environments remains
unclear.
To elucidate the role genetic variation plays in determining responses to
ontogenetic dietary changes, we examined body size and swim performance in
the zebrafish Danio rerio. Four full sib families were fed either consistent high or
low food rations, or a combination of the two for 60 days. We analyzed sources
of variation on phenotypic outcomes as functions of family (F), early diet (days 030; Diet0-30), later diet (days 30-60; Diet30-60), interactions between dietary
environments (Diet0-30 x Diet30-60), genetic variation to either dietary environment
(F x Diet0-30, F x Diet30-60), and genetic variation in interactions between dietary
environments (F x Diet0-30 x Diet30-60). With many potential sources of variation,
we made no a priori hypotheses on these sources.
119
Methods
Animals
This experiment performed under approval by The University of Akron’s
Institutional Animal Care and Use Committee. Adult zebrafish (Danio rerio) were
obtained from Aquatica Tropicals (Plant City, FL) and populations of wildtype
strains were maintained according to standard procedures (Westerfield, 1994).
Adults were maintained and bred at 26 ± 0.5°C with a 14L:10D light cycle.
Breeding design
Males and females were randomly paired resulting in four full-sib families.
Mating pairs were placed in 2 L containers lined with a marble substrate and
supplied with a common water source (Z-Mod housing system, Marine Biotech,
Beverly, MA).
Treatments
Siblings were raised together in 2 L containers with a common water
source (Z-Mod housing system, Marine Biotech, Beverly, MA) and were
maintained at 26 ± 0.5°C with a 14L : 10D light cycle for the duration of the
experiment. Food consisted of pulverized Zeigler™ adult zebrafish diet
supplemented with equal parts of <100 and 100-150 micron Zeigler™ larval diet
(1:1:1). After 30 days, the <100 and 100-150 micron supplements were replaced
with 150-250 and 250-450 micron supplements. For all feedings, 500 mg of food
was mixed with 250 ml of system water. From this solution, fish were fed at 0.1
ml/fish (low ration treatment) and 0.2 ml/fish (high ration treatment). We chose
these rations based on a standard dry food recipe from a protocol available at the
120
Zebrafish International Resource Center
(http://zebrafish.org/zirc/documents/protocols/pdf/Fish_Feeding/Flake_Food/Dry_
Food_Recipes.pdf). We assigned 0.1 ml/fish to the low ration treatment since it
was the minimum designated amount according to this protocol. Therefore, the
terms “high” and “low” are relative and apply only to the confines of this particular
study. Feedings were conducted once daily and excess food was removed
before each feeding. After two weeks, we noted all food was being consumed
within 24 hours. After 30 days, half of the individuals from each treatment were
switched to a separate 2 L tank and subjected to the opposite feeding treatment
for the remainder of the experiment. This resulted in four feeding treatments: high
food rations for 60 days (HH), low food rations for 60 days (LL), high food rations
for 30 days followed by low food rations for 30 (HL), and low food rations for 30
days followed by high food rations for 30 (LH).
Measurements
Body size - To ensure feeding treatments had an initial effect, we
measured the total length (TL) of each subject at 30 days. Fish were placed
individually in a small petri dish filled with system water. A standard length was
included in each picture. Photography was conducted with a Nikon D300 camera
under standard lighting conditions. We measured total length from most anterior
point to the posterior point of the caudal fin. We observed no damage to caudal
fins at any point in the study. At 60 days, each subject was euthanized with MS222 (300 mg/ml tricaine methane sulfonate buffered to a neutral pH with sodium
bicarbonate) and photographed on the subject’s right side with a length standard
121
in each picture. We measured standard length (SL) from most anterior point to
the base of the hypural plate at caudal flexion. Maximum depth (MD) was
measured as the maximum dorsal-ventral distance measured along the flank. All
measurements were made using imageJ (Version 1.42, NIH). Measurements
were made 5 times on each subject and the mean was recorded.
Swim velocity - Prior to terminal measurements (above), swimming
velocity was measured according to Widmer et al (2006). Briefly, individual fish
were placed in a clear acrylic flume (44.7 mm inner diameter by 30 cm long)
which drew system water. Baffles placed at the anterior portion of the chamber
maintained consistent laminar flow throughout the length of the flume. With an
initial flow rate of 3.8 l/min, flow (Blue-White Industries, Huntington Beach, CA,
U.S.A.; flow rate meter F-1000-RB) was incrementally increased by 1.9 l/min until
the fish spent >50% of the time increment touching the back mesh of the
chamber (Brett 1964). Maximum swim velocity was calculated based on the inner
diameter of the tube and the final flow measurement, and was determined based
on the standard length of the fish tested.
Statistics
To test for the effects of feeding for days 0-30 on TL, we used a two way
ANOVA. Family, treatment, and their interaction were included as sources of
variation. To test for the effects of feeding throughout the experiment on SL and
swim velocity, we used a three way ANOVA. Family, food treatment for days 030 (Diet 0-30), food treatment for days 30-60 (Diet 30-60), and all possible
122
interactions were included as potential sources of variation. TL and SL were log
transformed to meet normality assumptions.
Results
Survival
Ninety-three subjects survived the experiment. Chi-squared tests revealed
that survival shared no contingencies with feeding treatments (X2=0.56, P=0.91)
or families (X2=2.91, P=0.41).
Body size
At 30 days, family (F) and diet were both significant factors in influencing
TL (Table 5.1). Individuals from the high food treatment were significantly longer
(Fig. 5.1). At 60 days, however, the only significant source of variation on SL and
MD was family (Table 1; Figs. 5.2 & 5.3). Untransformed means are presented in
tables 2 and 3.
123
Fig. 5.1. The influence of diet for days 0-30 on total length in four full-sib families.
Error bars represent standard errors.
124
Table 5.1. ANOVA results for total length (TL), standard length (SL), maximum
depth (MD), and swim velocity. For TL, factors included family, diet for days 0-30
(Diet0-30), and their interaction. For SL, MD, and swim velocity, factors included
family, diet for days 0-30 (Diet0-30), diet for days 30-60 (Diet30-60), and all possible
interactions. TL, SL, and MD were log transformed for analyses.
Variable
Total length
Source
DF
MS
Family
3
16.4850
Diet0-30
1
19.8648
Family x Diet0-30
3
0.7649
Error
128 2.1122
Standard length Family
3
3.4353
Diet0-30
1
0.0123
Family x Diet0-30
3
0.0017
Diet30-60
1
0.0424
Family x Diet30-60
3
0.0013
Diet0-30 x Diet30-60
1
0.0003
Family x Diet0-30 x Diet30-60 3
0.0078
Error
77
0.0140
Maximum depth Family
3
2.6945
Diet0-30
1
0.0066
Family x Diet0-30
3
0.0055
Diet30-60
1
0.0370
Family x Diet30-60
3
0.0043
Diet0-30 x Diet30-60
1
0.0009
Family x Diet0-30 x Diet30-60 3
0.0074
Error
77
0.0111
Swim velocity
Family
3 491.9777
Diet0-30
1
7.1416
Family x Diet0-30
3
5.0020
Diet30-60
1
51.2372
Family x Diet30-60
3
28.1016
Diet0-30 x Diet30-60
1
56.8658
Family x Diet0-30 x Diet30-60 3
28.3586
Error
77
9.0594
125
F
P
7.8045 <0.0001
9.4046 0.0026
0.3621 0.7805
245.04 <0.0001
0.87
0.3526
0.12
0.9477
3.02
0.0860
0.09
0.9645
0.02
0.8861
0.56
0.6438
242.74 <0.0001
0.60
0.4426
0.50
0.6858
3.33
0.0718
0.39
0.7612
0.08
0.7724
0.67
0.5728
54.31
0.79
0.55
5.66
3.10
6.28
3.13
<0.0001
0.3774
0.6482
0.0199
0.0315
0.0143
0.0304
Table 5.2. Total length (mm) for zebrafish at 30 days under high and low food
rations. Data are presented as untransformed arithmetic mean ± SEM.
H
L
Total Length 4.68±0.20 3.87±0.17
126
Table 5.3. Maximum depth and standard length (mm) for zebrafish at 60 days
under all combinations of high and low food rations. Data are presented as
untransformed arithmetic mean ± SEM.
HH
Maximum Depth
HL
LH
LL
5.86±0.87 6.13±0.90 5.84±0.79 5.64±0.82
Standard Length 6.20±0.96 5.32±0.79 7.17±0.97 6.10±0.94
127
Fig. 5.2. The influence of changing diet on standard length in four full-sib families.
Line type (open vs solid) represents diet ration for days 0-30. The x axis
represents food ration for days 30-60. Error bars represent standard errors.
128
Fig. 5.3. The influence of changing diet on maximum depth in four full-sib
families. Line type (open vs solid) represents diet ration for days 0-30. The x axis
represents food ration for days 30-60. Error bars represent standard errors.
129
Swimming velocity
Many factors contributed to variation in swimming velocity. Besides
variation among families (F), diet for days 30-60 also contributed to variation in
swimming velocity (Diet30-60) and this effect varied significantly across families (F
x Diet30-60; Table 5.1). Fish fed low rations for days 30-60 attained higher
velocities on average than those fed high rations (Fig. 5.4). Early diet (Diet0-30)
also contributed to variation in swimming performance through its interaction with
later diet (Diet0-30 x Diet30-60; Table 5.1). Fish raised on low rations for the duration
of the experiment (LL) maintained similar velocities to those switched from low to
high rations (LH; Fig. 5.4). Interestingly, fish switched from high to low rations
(HL) attained higher swimming velocities than thoses maintained on high rations
(HH; Fig. 5.4). The interaction between family and both dietary periods indicates
that the quality of interactions between dietary treatments varied across families.
This resulted in a three way interaction between family, early, and later food
treatments (F x Diet30-60 x Diet30-60; Table 5.1). Variation in swimming
performance for families A and B remained consistent across diet treatments.
Families C and D, however, showed variation due to both early (Diet 0-30) and
later (Diet30-60) food treatments. Fish from these families raised on low rations for
the duration of the experiment (LL) maintained similar velocities to those
switched from low to high rations (LH). Fish switched from high to low rations
(HL), however, attained higher swimming velocities than this maintained on high
rations (HH; Fig. 5.5).
130
Fig. 5.4. Swimming velocity as a function of four nutritional environments. Open
lines and symbols represent low food rations days 0-30 and solid lines and
symbols represent high food rations for days 0-30. The x axis represents food
ration for days 30-60. Error bars represent standard errors.
131
Fig. 5.5. The influence of changing diet on swimming velocity in four full-sib
families. Line type (open vs solid) represents diet ration for days 0-30. The x axis
represents food ration for days 30-60. Error bars represent standard errors.
132
Discussion
Low diet fish were significantly smaller than high diet fish at 30 days. This
confirms our feeding treatment significantly altered development prior to
switching feeding treatments. Interestingly, neither nutritional environment
contributed to variation in SL or MD at the end of the experiment. In other words,
while early diet initially influenced size, subsequent nutritional change resulted in
equal sizes across all four treatments. This indicates some compensatory growth
mechanism for fish exposed to low food rations for days 0-30. Compensatory
growth is generally not without cost. Following exposure to suboptimal
conditions, compensatory growth has been show to be associated with costs in a
number of physiological, morphological, life history, and performance traits
(reviewed in Metcalfe and Monaghan 2001).
The performance trait addressed in this study was maximum swim
velocity. Although they exhibited compensatory growth, fish raised on early low
food rations maintained similar swimming velocity regardless of later dietary
rations. This indicates negligible cost of compensatory growth on swimming
performance in our study. Previous studies have demonstrated a tradeoff
between accelerated growth and physical performance in salmon (Farrell et al.
1997) and sticklebacks (Álvarez and Metcalfe 2007). In the case of sticklebacks,
however, the associated tradeoff was present in stream rather than pond
populations. This result indicates that local selective pressures can alter tradeoff
trajectories among populations. Zebrafish inhabit a wide variety of habitats
ranging from active streams to stagnant rice fields (Spence et al. 2008). It is
133
therefore likely that ecological variation has shaped tradeoff trajectories in this
species.
Interestingly, swim velocity was highest in fish switched from high to low
food rations. Thus, dietary change enhanced swimming performance, but only for
fish started on high rations. Fish started on low rations did not increase
swimming velocity when switched to high rations. Thus, although HL and LH fish
attained similar size, their physical abilities differed. Phillips (2004) performed a
similar study with mussels and found similar results. While mussels switched
from high to low rations equaled those switched from low to high in terms of shell
size, they differed in terms of lipid content. This result suggests that dietary order
can be more critical in shaping physiological rather than morphological
outcomes. Thus, the underlying physiologies of the subjects in our study may
have been affected. Specifically, fish switched from high to low rations proved
physiologically superior to fish from other treatments.
Swimming ability is a critical trait in fish. Its implications on prey capture,
predator avoidance, and social interactions are evident (Domenici and Blake
1997). Thus, as swimming ability is sensitive to environmental change, the
ontogenetic history of fish becomes critical in shaping their fitness. At the
population level, variation in environmentally-altered developmental trajectories
provides the raw material for natural selection to optimize fitness in changing
environments. Zebrafish inhabit a wide variety of habitats throughout Southeast
Asia (Spence et al. 2008). Their association with a number of different habitats
throughout seasonal changes makes it likely that factors such as temperature,
134
oxygen, and food availability can vary during their ontogeny. Thus, it is likely that
there is some ecological component to variation in swimming ability in zebrafish.
Given their small size, zebrafish are prolific swimmers that display remarkably
low associated physiological costs (Plaut and Gordon 1994). The selective
factors that have shaped these abilities require further elucidation.
The role environmental complexity plays in shaping ontogenetic
trajectories is receiving increasing attention (Monaghan 2008). Few studies to
our knowledge have quantitatively demonstrated significant interactions between
subsequent ontogenetic periods of development (Kotrschal and Taborsky 2010;
Marks et al. 2005). Even less clear is the role genetic variation plays in
determining the quality of phenotypic outcomes under complex conditions. Our
study not only demonstrates that subsequent dietary conditions can interact in
shaping zebrafish physical performance, but the quality of these effects is family
specific. This result indicates at least some role of genetic variation in shaping
plastic responses under complex conditions. This variation provides critical
insight into how populations evolve in complex environments.
In summary, we found a significant interaction between dietary
environments (Diet0-30 x Diet30-60) for swimming velocity. Overall, fish switched
from high to low food rations attained the highest swimming velocity. Fish started
on low food rations attained similar swimming velocities regardless of later food
rations. The quality of responses to dietary change varied across families,
resulting in a significant Family x Diet0-30 x Diet30-60 interaction. Although early
food rations influenced size at the midway point of the experiment, fish achieved
135
equal sizes across all food treatments at the end of the experiment. These
results suggest that plastic responsiveness to subsequent environmental
changes can be trait specific and vary significantly within populations. The
specific order of environmental conditions can also be critical in determining
performance outcomes.
Conclusions and evolutionary implications
Figure 5.5 depicts a modified reaction norm similar to the one described in
chapter I and employed in chapters II and III. While it also demonstrates that the
quality of interaction between subsequent environments is conditional on
genotype, it also yields potential evolutionary implications. This reaction norm
demonstrates that variance in swimming velocity for fish exposed to later low
food rations is highest when fish start on high food rations. Figure 5.6 confirms
this assertion. Therefore, fish exposed to these subsequent conditions would
exhibit more phenotypic variants with which selection can act upon. Swim
velocity is undoubtedly an important trait. Faster swimming fish would be more
likely to escape predators and would therefore incur a fitness advantage. This
result therefore demonstrates that changes in dietary conditions can have
consequences on fitness.
Acknowledgements
We thank Choose Ohio First Tiered Mentoring Program for funding this project.
We also thank Steven Lombardo and Kristie Formanik for their help collecting
data.
136
Coeffecient of Variation
(swim velocity)
0.70
0.68
0.66
0.64
0.62
0.60
0.58
0.56
Later food
Early food
H
H
L
H
L
L
Figure 5.6. Variance due to all four combinations of feeding treatments.
137
CHAPTER VI
CONCLUSIONS
We tested the overall hypothesis that the early environment can constrain
or alter later phenotypic responses in a given environment. We also explored the
role of genetic variation in this process. The previous chapters clearly
demonstrated that environmental modulation of phenotypic variance in a given
environment is conditional on early environmental conditions. In chapter II, for
example, cardiovascular performance at 48 h.p.f. was not only conditional on the
oxygen level of the first 24 hours of development, but the quality of these
conditional responses varied across genotypes. In chapter III, the variance of
integrated morphology showed the same trend. Chapter V showed that variation
in swim velocity was greatest in fish exposed to a low ration diet, but only for fish
started out on a high ration diet. However, in chapter IV we found no variation in
the manner in which the early environment altered zebrafish shoaling, stimulus
reactivity, and size. Thus, while our hypothesis was confirmed in many cases,
these responses were specific to the trait being studied.
Ramifications of environmental complexity
With the likelihood that organisms experience changing conditions during
development, it is therefore likely that the degree of observable phenotypic
variance within any population is a function of early developmental conditions.
138
Therefore, when assessing fitness and evolution within natural populations, it is
critical to account for the entire environmental history.
The environment is extremely complex with nearly infinite combinations of
various biotic and abiotic factors that can vary spatiotemporally throughout
ontogenetic time. These methods could therefore be adapted to explore the
influence of changing conditions in any model. However, caution should be used
in over-factorializing studies. Just one more subsequent environmental change in
our study would have further halved sample sizes in each treatment. Maintaining
sample sizes large enough to execute multifactorial studies would be difficult to
employ. However, while these collective studies represent relatively simple
scenarios, they still provide insight into the consequences of changing
environmental conditions during development. The current paradigm analyzes
plasticity as a function of one environmental change. We have demonstrated that
this view is not only overly simplistic, but provides an incomplete picture of the
proximate causes of phenotypic variance in populations. We have demonstrated
that early conditions can permanently constrain zebrafish size and behavior
regardless of later conditions (chapter IV). We have also demonstrated that early
conditions can alter responses to later ones and that the nature of these
conditional responses vary significantly across genotypes (chapters II, III, V).
Variation, selection, and evolution
The factors that promote and maintain phenotypic variation in the natural
world are of central importance to biologists. It is this variation that continually
provides the raw material for natural selection to optimize fitness and promote
139
speciation. Our collective studies demonstrate that the degree of variation
expressed in a population can be a function how variants from a particular early
environment respond to later conditions. Thus, the context of subsequent
developmental environments can be critical in determining the rate at which unfit
variants are removed. In other words, the tempo of evolution in any given
population can be conditional on the series of conditions its individuals
experience.
Plotting the variance structures yielded interesting insight into the
environmental factors shaping phenotypic variance. In chapter II, III, and IV, the
variance in all examples (Cardiac output, morphology, and overall size) are
lowest in fish exposed to consistent hypoxia (Figs. 2.10, 3.6, 4.9). This reduced
variance relative to other combinations of subsequent environments suggests
that hypoxia has been a major force in shaping the evolution of these traits in
zebrafish. This not surprising given the conditions of their native habitat
discussed in chapter I.
Future directions
Proximate mechanisms of plasticity: quantitative epigenetics
Biologists have qualitatively known that early conditions can shape
developmental trajectories and determine phenotypes regardless of later
environmental conditions. Barker (1998) noted that adult phenotypes such as
disease susceptibility could be correlated with maternal nutrition, drug use, and
stress levels. He speculated that these stressors irreversibly altered the course of
development. Thus, he coined the term ‘fetal programming’. Since then biologists
140
have pinpointed the proximate mechanisms of this epigenetic phenomenon
(Reviewed in Mulder et al 2002). One well-documented case involves the
rearchitecturing of the developing hypothalamic-pituitary-adrenal (HPA) axis.
During instances of maternal stress, corticoid receptor density is permanently
down-regulated, resulting in offspring with compromised negative feedback
systems (reviewed in Mulder et al 2002). Thus, while the epigenetics of the
influence of early development has been elucidated, genetic variation in this trait
has not been characterized. I advance the hypothesis that there is variation in the
quality of epigenetic remodeling and this may contribute to interactions between
subsequent developmental environments. For example, while maternal stress
may cause receptor density to be down regulated on average within a population,
some genotypes may not respond accordingly. Some genotypes may develop
the same receptor density regardless of maternal stress levels. They would
develop a normal negative feedback system. Alternatively, other phenotypes may
increase receptor density. This would promote an individual with an increased
negative feedback system. Therefore, a quantitative genetic assessment of
epigenetic phenomenon would shed critical insight into the proximate
mechanisms of environmental modulation of phenotypic outcomes.
Quantitative genetics and the future of evolutionary biology
Pigliucci and Muller emphasized the need to advance the work of the
modern synthesis by incorporating aspect of development and ecology (Pigliucci
and Muller 2010). As these ideas to develop, the need will arise to further
synthesize ecology and development into a more refined theory of evolution.
141
Quantitative genetics has proved a powerful tool for making evolutionary
predictions (Roff 2007). Our results clearly demonstrate the utility of quantitative
genetics in addressing the evolution of complex, developmentally plastic traits in
heterogeneous environments. Our approach offers the flexibility of manipulating
ecological factors while exploring specific windows of development. Development
is a dynamic, hierarchical process involving the genesis and interaction of cells,
hormones, tissues, and organs (Hall 2003). The multifactorial framework we
employ can provide a powerful tool in exploring the complex relationship between
genes, developing systems, phenotypic variation, and dynamic environments.
Subsequent studies should not only focus on elucidating proximate mechanisms
of plasticity, but also the relative impacts of sequential environmental conditions
of more fitness related traits (i.e. fecundity).
142
LITERATURE CITED
Alais D., Newell F., and Mamassian, P. 2010. Multisensory Processing in
Review: from Physiology to Behaviour. Seeing and Perceiving 23: 3-38.
Alderman S. L., and Bernier N.J. 2009. Ontogeny of the corticotropin-releasing
factor system in zebrafish. Gen Comp Endocrinol 164: 61-69.
Ali M., Nicieza, A., and Wootton, R. J. 2003. Compensatory growth in fishes: a
response to growth depression. Fish and Fisheries 4: 147-190.
Alsop D., and Wood, C. 1997. The interactive effects of feeding and exercise on
oxygen consumption, swimming performance and protein usage in
juvenile rainbow trout (Oncorhynchus mykiss). J Exp Biol 200: 2337-2346.
Álvarez D., and Metcalfe, N. B. (2007). The tradeoff between catch-up growth
and escape speed: variation between habitats in the cost of
compensation. Oikos 116: 1144-1151.
Arnold S. J. 1992. Constraints on phenotypic evolution. Am Nat 140: S85–S107.
Atchley W. R. 1987. Developmental quantitative genetics and the evolution of
ontogenies. Evolution 41:316-330.
Atchley W. R., and Zhu J. 1997. Developmental quantitative genetics, conditional
epigenetic variability and growth in mice. Genetics 147: 765-776.
Bagatto B. 2005. Ontogeny of cardiovascular control in zebrafish (Danio rerio):
effects of developmental environment. Comp Biochem Physiol B Biochem
Mol Biol 141: 391-400.
Barman R. P. 1991. A taxonomic revision of the Indo-Burmese species of Danio
rerio. Record of the Zoological Survey of India Occasional Papers 137:1–
91.
Bejda A. J., Studholme A. L., and Olla B. L. 1987. Behavioral responses of red
hake, (Urophycis chuss), to decreasing concentrations of dissolved
oxygen. Environ Biol Fish 19: 261-268.
143
Beamish F. W. H., Howlett J. C., and Medland T. E. 1989. Impact of diet on
metabolism and swimming performance in juvenile lake trout, Salvelinus
namaycush. Can J Fish Aquat Sci 46: 384-388.
Boutilier, R.G. 1990. Respiratory gas tensions in the environment. In: Advances
in comparative and environmental physiology. Boutilier, R.G. (ed.).
Springer-Verlag. Chapter 1.
Bradshaw A. D. 1965. Evolutionary significance of phenotypic plasticity in plants.
Adv Genet 13: 115–155.
Brett J. R. 1964 The respiratory metabolism and swimming performance of
young sockeye salmon. J Fish Res Board Can 21: 1183–122.
Brown-Peterson N.J., Manning C. S., Patel, V., Denslow, N. D., and Brouwer, M.
2008. Effects of cyclic hypoxia on gene expression and reproduction in a
grass shrimp, (Palaemonetes pugio). Biol Bull 214: 6-16.
Brown G. E. and Godin, J.-G. J. 1999. Chemical alarm signals in wild Trinidadian
guppies (Poecilia reticulata). Can J Zool 77: 562-570.
Bubliy O. A., Loeschcke V., and Imasheva, A. G. 2001. Genetic variation of
morphological traits in Drosophila melanogaster under poor nutrition:
isofemale lines and offspring-parent regression. Heredity 86:363–369.
Buckley C. R., Michael S. F. and Irschick, D. J. 2005. Early hatching decreases
jumping performance in a direct-developing frog, Eleutherodactylus coqui.
Funct Ecol 19: 67–72.
Burggren W.W. 2000. Developmental physiology, animal models, and the August
Krogh principle. Zool Anal Complex Syst 102: 148-156.
Burggren W. W., and Reyna K. S. 2011. Developmental trajectories, critical
windows and phenotypic alteration during cardio-respiratory development.
Respir Physiol Neurobiol 178: 13-21.
Burleson M. L., Wilhelm, D. R., and Smatresk, N. J. 2001. The influence of fish
size size on the avoidance of hypoxia and oxygen selection by largemouth
bass. J Fish Biol 59: 1336-1349.
Carroll S. B. 1995. Homeotic genes and the evolution of arthropods and
chordates. Nature 376:479-485.
Carroll. S.B. 2005 Endless Forms Most Beautiful: The New Science of Evo
Devo and the Making of the Animal Kingdom. W.W. Norton, New York.
144
Cerezo J., and García García, B. 2004. The effects of oxygen levels on oxygen
consumption, survival and ventilatory frequency of sharpsnout sea bream
(Diplodus puntazzo Gmelin, 1789) at different conditions of temperature
and fish weight. J Appl Ichthyol 20: 488-492.
Chan T., and Burggren, W. W. 2005. Hypoxic incubation creates differential
morphological effects during developmental critical windows in the embryo
of the chicken (Gallus gallus). Resp Physiol Neurobi 145:251-263.
Chapman L. J., Albert J., and Galis, F. 2008. Developmental Plasticity, Genetic
Differentiation, and Hypoxia-induced Trade-offs in an African Cichlid Fish.
The Open Evolution Journal 2: 75-88.
Chapman L.J., and Hulen, K.G. 2001. Implications of hypoxia for the brain size
and gill morphometry of mormyrid fishes. J Zool 254: 461-472.
Charmantier A., and Garant D. 2005. Environmental quality and evolutionary
potential: lessons from wild populations. Proc Roy Soc B-Biol Sci
272:1415 –1425.
Chen D. S., Dykhuizen, G. V., Hodge J., and Gilly W. F. 1996. Ontogeny of
Copepod Predation in Juvenile Squid (Loligo opalescens). Biol Bull 190:69 81.
Cheverud, J. M., Rutledge, W., and Atchley, W. R. 1983. Quantitative genetics of
development: Genetic correlations among age specific trait values and the
evolution of ontogeny. Evolution 37:895-905.
Cheverud J. M. 1988. A comparison of genetic and phenotypic correlations.
Evolution 42: 958–968.
Cheverud J. M. 1996. Developmental Integration and the Evolution of Pleiotropy
Amer Zool 36: 44-50.
Chippari-Gomes A.R., Gomes L. C, Lopes N.P., Val A. L., and Almeida-Val V. M.
F. 2005. Metabolic adjustments in two Amazonian cichlids exposed to
hypoxia and anoxia. Comp Biochem Physiol B 141: 347-355.
Clark E. B. 1990. Growth, morphogenesis and function: The dynamics of cardiac
development. In: Moller, J.H., Neal, W., and Lock, J. eds. Fetal, neonatal
and infant heart disease. New York: Appleton-Century-Crofts.
Clark P.M. 1997. Programming of the hypothalamo-pituitary-adrenal axis and the
fetal origins of adult disease hypothesis. Eur J Pediatr 157: S7-S10.
145
Correns, C. G. 1900. Mendel's Regel über das Verhalten der Nachkommenschaft
der Rassenbastarde. Berichte der deutschen botanischen Gesellschaft
18:158–168.
Coutant C. 1987. Thermal preference: when does an asset become a liability?
Environ Biol Fish 18: 161-172.
Cowly D. E., and Atchley W. R. 1992. Quantitative genetic models for
development, epigenetic selection, and phenotypic evolution. Evolution
46: 495-518.
Criscuolo F., Monaghan, P., Nasir L., and Metcalfe N. B. 2008. Early nutrition
and phenotypic development: ‘catch-up’ growth leads to elevated
metabolic rate in adulthood. P Roy Soc B-Biol Sci 275: 1565-1570.
Crispo, E. and L. J. Chapman. 2011. Hypoxia drives divergence in cichlid body
shape. Evol Ecol 25:949-964.
Darwin C. 1859. On the origin of species by means of natural selection. London:
J. Murray.
De Vries, H. 1889. Intracellulare Pangenesis. Jena. Gustav Fisher.
DeWitt T. J., and Scheiner S. M., eds. 2004 Phenotypic Plasticity. Functional
and Conceptual Approaches. New York: Oxford University Press.
Diaz R. L., and Rosenberg R., 1995. Marine benthic hypoxia: a review of its
ecological effects and the behavioural responses of benthic macrofauna.
Oceanogr. Mar Biol Ann Rev 33:245–303.
Domenici P., and Blake R. 1997. The kinematics and performance of fish faststart swimming. J Exp Biol 200:1165-1178.
Domenici P., Steffensen, J. F., and Batty, R. S. 2000. The effect of progressive
hypoxia on swimming activity and schooling in Atlantic herring. J Fish Biol
57: 1526-1538.
Doudoroff P. S., and Shumway D. L. 1970. Dissolved oxygen requirements of
freshwater
fishes. FAO Technical Paper 86.
Dudley, S. A., and Schmitt, J. 1996. Testing the adaptive plasticity hypothesis:
density-dependent selection on manipulated stem length in Impatiens
capensis. Am Nat 147:445–465.
146
Eames S. C., Philipson, L. H., Prince, V. E., and Kinkel, M. D. 2010. Blood
Sugar Measurement in Zebrafish Reveals Dynamics of Glucose
Homeostasis. Zebrafish 7: 205-213.
Eidietis L., Forrester T. L., and Webb, P. W. 2002. Relative abilities to correct
rolling disturbances of three morphologically different fish. Can J Zoolog
80: 2156-2163.
Ellison P. T. 2010. Fetal programming and fetal psychology. Infant Child Dev 19:
6-20.
Engeszer R. E., Da Barbiano L., Ryan M., and Parichy, D. 2007a. Timing and
plasticity of shoaling behaviour in the zebrafish, Danio rerio. Anim Behav
74: 1269-1275.
Engeszer R.E., L.B. Patterson, A.A. Rao, and D.M. Parichy. 2007b. Zebrafish in
the wild: a review of natural history and new notes from the field. Zebrafish
4: 21-40.
Espmark Å.M., Eriksen, M. S., Salte R., Braastad B. O., and Bakken, M. 2008. A
note on pre-spawning maternal cortisol exposure in farmed Atlantic
salmon and its impact on the behaviour of offspring in response to a novel
environment. Applied Anim Behav Sci 110: 404-409.
Farrell A. P., Bennett W., and Devlin R. H. 1997. Growth-enhanced transgenic
salmon can be inferior swimmers. Can J Zool 75: 335-337.
Ferguson R. A., Kieffer J. D., and Tufts B. L. 1993. The effects of body size on
the acid–base and metabolite status in the white muscle of rainbow trout
before and following exhaustive exercise. J Exp Biol 180: 195–207.
Fisher R. A. 1918. The correlation between relatives on the supposition of
Mendelian Inheritance. T Royal Soc Edin 52: 399-433.
Fisher, R.A. 1930. The Genetical Theory of Natural Selection. Clarendon Press,
Oxford.
Flury, B. 1988. Common Principal Components and Related Multivariate Models.
New York: Wiley and Sons..
Frost A. J., Winrow-Giffen A., Ashley P. J., and Sneddon L. U. 2007. Plasticity in
animal personality traits: does prior experience alter the degree of
boldness? Proc Roy Soc B-Biol Sci 274: 333-339.
Fuzzen M. L. M., Alderman S. L., Bristow E. N., and Bernier N. J. 2011.
Ontogeny of the corticotropin-releasing factor system in rainbow trout and
147
differential effects of hypoxia on the endocrine and cellular stress
responses during development. Gen Comp Endocrinol 170: 604-612.
García-Berthou E. 1999. Food of introduced mosquitofish: ontogenetic diet shift
and prey selection. J Fish Biol 55: 135-147.
Gianoli E., and González-Teuber M. 2005. Environmental heterogeneity and
population differentiation in plasticity to drought in Convolvulus chilensis
(Convolvulaceae) Evol Ecol 19: 603-613.
Giesing E. R., Suski C. D., Warner R. E., and Bell A. M. 2010. Female
sticklebacks transfer information via eggs: effects of maternal experience
with predators on offspring. Proc Roy Soc B-Biol Sci 278: 1753-1759.
Gilbert S. F. 2001. Ecological developmental biology: Developmental biology
meets the real world. Dev Biol 233:1-12.
Glover V. 2011. Prenatal stress and the origins of psychopathology: an
evolutionary perspective. J Child Psychol Psych 52: 356-367.
Godin J.-G. J., and Morgan M. J. 1985. Predator avoidance and school size in a
cyprinodontid fish, the banded killifish (Fundulus diaphanus Lesueur).
Behav Ecol Sociobiol 16: 105-110.
Gould S. J., and Lewontin R. C. 1979. The Spandrels of San Marco and the
Panglossian Paradigm: A Critique of the Adaptationist Programme. Proc R
Soc Lond B Biol Sci 205: 581-598.
Gould, S. J. 1977. Ontogeny and Phylogeny. The Belknap Press, Cambridge.
Grafen A. 1988. On the uses of data on lifetime reproductive success. In:
Reproductive Success. Clutton-Brock T. editor. Chicago: University of
Chicago Press.
Grunwald, D. J., and Eisen, J. S. 2002. Headwaters of the zebrafish –
emergence of a new model vertebrate. Nat Rev Genet 3:717-724.
Gupta A. P., and Lewonti, R. C. 1982 Evolution (Lawrence, Kans.) 36:934–948.
Haldane, J. B. S. 1924. A mathematical theory of natural and artificial selection.
Part I. Trans Camb Phil Soc 23:19-41.
Hales C. N., and Barker D. J. P. 2001. The thrifty phenotype hypothesis. Br. Med.
Bull. 60:5–20.
148
Hall B. K. 2001. Organic Selection: Proximate Environmental Effects on the
Evolution of Morphology and Behaviour. Biology and Philosophy 16: 215237.
Hall, B. K. 2003. Evo-Devo: Evolutionary developmental mechanisms. Int J Dev
Biol 47: 491-495.
Hall B. K., Pearson R. D.,and Müller G. B. 2004. Environment, development, and
evolution. Cambridge: MIT Press.
Healy S. D., and Rowe C. 2007. A critique of comparative studies of brain size.
Proc Roy Soc B-Biol Sci 274: 453-464.
Hellemans K. G. C., Sliwowska J.H., Verma P., and Weinberg J. 2010. Prenatal
alcohol exposure: Fetal programming and later life vulnerability to stress,
depression and anxiety disorders. Neurosci Biobehav Rev 34: 791-807.
Hoare D. J., Krause J., Peuhkuri N., and Godin J.-G. J. 2000. Body size and
shoaling in fish. J Fish Biol 57: 1351-1366.
Hopkins S. R., and Powell F. L. 2001. Common themes of adaptation to hypoxia:
insights from comparative physiology. Adv Exp Med Biol 502:153–167.
Hocher B., Slowinksi T., Bauer C., and Halle H. 2001. The advanced fetal
programming hypothesis. Nephrol Dial Transplant 16: 1298-1299.
Hoverman J. T., and Reylea R. A. 2007. How flexible is phenotypic plasticity?
Developmental windows for trait induction and reversal. Ecology 88:693705.
Hoverman J. T., and Reylea R. A. 2008. Temporal environmental variation and
phenotypic plasticity: a mechanism underlying priority effects. Oikos
117:23-32.
Hughes G. M. 1984. Measurement of gill area in fishes: practices and problems.
J Mar Biol Assoc UK 64: 637-655.
Huntingford F.A., Andrew G., Mackenzie S., Morera D., Coyle S. M., Pilarczyk
M., and Kadri S. 2010. Coping strategies in a strongly schooling fish, the
common carp (Cyprinus carpio). J Fish Biol 76: 1576-1591.
Ibarra A. M., and Famula, T. R. 2008. Genotype by environment interaction for
adult body weights of shrimp Penaeus vannamei when grown at low and
high densities. Genet Sel Evol 40:541–551.
149
Iwama G.K., Morgan J.D., and Barton B.A. 1995. Simple field methods for
monitoring stress and general condition of fish. Aquaculture Research 26:
273-282.
Roff, D. A. 2007. A centennial celebration for quantitative genetics. Evolution 61,
1017-1032 (2007).Jobling, M. 1994. Fish Bioenergetics. Volume 13: 1-44.
London: Chapman and Hall.
Jensen F. B. and Weber R. E. 1982. Respiratory properties of tench blood and
hemoglobin: adaptation to hypoxichypercapnic water. Mol Physiol 2:235–
250.
Jones D. R. 1971. Theoretical analysis of factors which may limit the maximum
oxygen uptake of fish: The oxygen cost of the cardiac and branchial
pumps. J Theor Biol 32: 341-349.
Jones R. H., and Ozanne S. E. 2009. Fetal programming of glucose-insulin
metabolism. Mol Cell Endocrinol 297: 4-9.
Jonz M. G., and Nurse C. A. 2005. Development of oxygen sensing in the gills of
zebrafish. J Exp Biol 208:1537-1549.
Kajimura, S., Aida, K., and Duan, C. 2005. Insulin-like growth factor-binding
protein-1 (IGFBP-1) mediates hypoxia-induced embryonic growth and
developmental retardation. Proc Nat A Sci USA 102, 1240 –1245.
Kapoor A., Dunn E., Kostaki A., Andrews M. H., and Matthews S.G. 2006. Fetal
programming of hypothalamo-pituitary-adrenal function: prenatal stress
and glucocorticoids. J Physiol 572: 31-44.
Ketterson E. D., Atwell J. W., and McGlothlin J. W. 2009. Phenotypic integration
and independence: Hormones, performance, and response to
environmental change. Integr Comp Biol 49: 365-379.
Kieffer J. D. 2000. Limits to exhaustive exercise in fish. Comp Biochem Phys A
126: 161-179.
Kingsolver J. G., and Huey R. B. 1998. Evolutionary analyses of morphological
And physiological plasticity in thermally variable environments. Amer.
Zool. 38:545- 560.
Klingenberg C. P., Barluenga M., and Meyer A. 2003. Body shape variation in
cichlid fishes of the Amphilophus citrinellus species complex. Biol J Linn
Soc 80:397–408.
Klingenberg, C. P. 2008. MORPHOJ. Faculty of Life Sciences, University of
150
Manchester, UK. http://www.flywings.org.uk/MorphoJ_page.htm.
Kotrschal K., Van Staaden M. J., and Huber R. 1998. Fish Brains: Evolution and
Environmental Relationships. Rev Fish Biol Fish 8: 373-408.
Kotrschal A., and Taborsky B. (2010). Environmental change enhances cognitive
abilities in fish. PLoS Biol 8, e1000351.
Krause J., and Godin J.-G. J. 1994. Shoal Choice in the Banded Killifish
(Fundulus diaphanus, Teleostei, Cyprinodontidae): Effects of Predation
Risk, Fish Size, Species Composition and Size of Shoals. Ethology 98:
128-136.
Lande R. 1979. Quantitative genetic analysis of multivariate evolution, applied to
brain:body size allometry. Evolution 33:402-416.
Lande R. 2009. Adaptation to an extraordinary environment by evolution of
phenotypic plasticity and genetic assimilation. J Evol Biol 22: 1435-1446.
Levins, R. 1968 Evolution in Changing Environments. Princeton University Press,
Princeton.
Lima S. L., and Dill L. M. 1990. Behavioral decisions made under the risk of
predation: a review and prospectus. Can J Zool 68: 619-640.
Lo K. H., Hui M. N., Yu R. M., Wu R. S., and Cheng S. H. 2011 Hypoxia Impairs
Primordial Germ Cell Migration in Zebrafish (Danio rerio) Embryos. PLoS
One 6:e24540
Lowe-Jinde L., and Niimi A. J. 1983. Influence of sampling on the interpretation
of haematological measurements of rainbow trout (Salmo gairdneri). Can
J Zool 61: 396-402.
Lynch M., and Walsh B. 1998. Genetics and Analysis of Quantitative Traits.
Sunderland: Sinauer.
MacCormack T. J., and Driedzic W. R. 2007. The impact of hypoxia on in vivo
glucose uptake in a hypoglycemic fish (Myoxocephalus scorpius). Am J
Physiol Regul Integr Comp Physiol 292: R1033-R1042.
Machiels, M. A. M., and Wijsman J. 1996. Size-selective mortality in an exploited
perch population and the reconstruction of potential growth. Ann Zool
Fenn 33: 397-401.
Marks, C., West, T. N., Bagatto, B., and Moore, F. B-G. 2005. Developmental
environment alters conditional aggression in zebrafish. 2005: 901-908.
151
McArdle H. J., Andersen H. S., Jones H., and Gambling L. 2006. Fetal
Programming: Causes and Consequences as Revealed by Studies of
Dietary Manipulation in Rats – A Review. Placenta 27, Supplement: 56-60.
Metcalfe N. B., and Monaghan, P. 2001. Compensation for a bad start: grow
now, pay later? Trends Ecol Evol 16: 254-260.
Mohamed, M. P., and Kutty M. N. 1983. Influence of hypoxia on metabolism and
activity in Puntius sarana (Hamilton) (Pisces: Cyprinidae). Proc Indiana
Acad Sci Anim Sci 92: 215-220.
Monna, F., Camizuli E., Revelli P., Biville C., Thomas C., Losno R., Scheifler R.,
Bruguier O., Baron S., Chateau C., Ploquin A., and Alibert P. 2011.
Wild brown trout affected by historical mining in the C vennes National
Park, France. Envir Sci Tech 45: 6823-6830.
Moore F. B.-G., Hosey M., and Bagatto B. 2006. Cardiovascular system in larval
zebrafish responds to developmental hypoxia in a family specific manner.
Front Zool 3: 4.
Monaghan P. 2008. Early growth conditions, phenotypic development and
environmental change. Philos Trans R Soc B 363: 1635-1645.
Moran, N. A. 1992. The evolutionary maintenance of alternative phenotypes. Am
Nat 139: 971-989.
Murren C. J. 2002. Phenotypic integration in plants. Plant Spec Biol 17:89–99.
Nilsson, G. E. 2007. Gill remodeling in fish: a new fashion or an ancient secret? J
Exp Biol 210:2403–2409.
Noltie D. B., and Wang N. 1997. Use of compensatory growth to double hybrid
sunfish growth rates. T Am Fish Soc 126: 316-322.
O’Connor E. A., Pottinger T. G., and Sneddon L. U. 2011. The effects of acute
and chronic hypoxia on cortisol, glucose and lactate concentrations in
different populations of three-spined stickleback. Fish Biochem Physiol 37:
461-469.
Osenberg C. W., Mittelbach G. G., and Wainwright P. C. (1992). Two-stage life
histories in fish: the interaction between juvenile competition and adult
performance. Ecology 73: 255-267.
Padilla P.A., and Roth M.B. 2001. Oxygen deprivation causes suspended
animation in the zebrafish embryo. Proc Nat A Sci 98: 7331-7335.
152
Pankhurst, P. M., Pankhurst N. W., and Montgomery J. C. 1993. Comparison of
behavioural and morphological measures of visual acuity during ontogeny
in a teleost fish, Forsterygion varium, tripterygiidae (Forster, 1801). Brain
Behav Evolut 42: 178-188.
Parsons G. R., and Carlson J. K.. 1998. Physiological and behavioral responses
to hypoxia in the bonnethead shark (Sphyrna tiburo): routine swimming
and respiratory regulation. Fish Biochem Physiol 19: 189-196.
Pelster B., and Burggren, W. W. 1996. Disruption of hemoglobin oxygen
transport does not impact oxygen-dependent physiological processes in
developing embryos of zebra fish (Danio rerio). Circ Res 79:358–362.
Pelster, B. 1997. Oxygen, temperature, and pH influences on the development of
nonmammalian embryos and larvae. In: Development of Cardiovascular
Systems. Burggren W. W., and Keller B. B. eds. , Cambridge: Cambridge
University Press.
Pelster B., Sanger A. M., Siegele M., Schwerte T. 2003. Influence of swim
training on cardiac activity, tissue apillarization, and mitochondrial density
in muscle tissue of zebrafish larvae. Am J Physiol 285: R339-R347.
Phillips P. C., and Arnold S. J. 1999. Hierarchical comparison of genetic
variance-covariance matrices. I. Using the Flury hierarchy. Evolution 53:
1506-1515.
Phillips N. E. 2004. Variable timing of larval food has conseuqnces for early
juvenile performance in a marine mussel. Ecology 85: 2341-2346.
Pichavant K., Person-Le-Ruyet J., Le Bayon N., Severe A., Le Roux A.,
Quemener L., Maxime V., Nonnotte G., and Boeuf G. 2000. Effects of
hypoxia on growth and metabolism of juvenile turbot. Aquaculture
188:103-114.
Pichavant K., Person-Le-Ruyet J., Le Bayon N., Severe A., Le Roux A.,
Quemener L., Maxime V., Nonnotte G., and Boeuf G. 2001. Comparative
effects of long-term hypoxia on growth, feeding and oxygen consumption
in juvenile turbot and European sea bass. J Fish Biol 59: 875-883.
Pigliucci M., Whitton J., and Schlichting C. D. 1995. Reaction norms of
Arabidopsis. I. Plasticity of characters and correlations across
water,nutrient and light gradients. J Evol Biol 8: 421–438.
153
Pigliucci M. 1998. Developmental phenotypic plasticity: Where internal
programming meets the external environment. Curr Opin Plant Biol 1: 8791.
Pigliucci M. 2001. Phenotypic Plasticity: Beyond Nature and Nurture. Baltimore:
Johns Hopkins University Press.
Pigliucci M. 2003. Phenotypic integration: studying the ecology and evolution of
complex phenotypes. Ecol Lett 6: 265-272.
Pincetich, C. A., Viant M. R., Hinton D. E., and Tjeerdema R. S. 2005. Metabolic
changes in Japanese medaka (Oryzias latipes) during embryogenesis and
hypoxia as determined by in vivo P-31 NMR. Comp Biochem Physiol C
Toxicol Pharm 140: 103-113.
Pigliucci, M. & Muller, G. (Eds.) 2010. Evolution: The Extended Synthesis. MIT
Press, Cambridge.
Pitcher T. J., Magurran A. E., and Winfield I. J. 1982. Fish in larger shoals find
food faster. Behav Ecol Sociobiol 10: 149-151.
Plaut I., and Gordon M. 1994. Swimming metabolism of wild-type and cloned
zebrafish Brachydanio rerio. J Exp Biol 194: 209-223.
Pollen A. A., Dobberfuhl A. P., Scace J., Igulu M. M., Renn S. C. P., Shumway C.
A., and Hofmann H. A. 2007. Environmental Complexity and Social
Organization Sculpt the Brain in Lake Tanganyikan Cichlid Fish. Brain
Behav Evol 70: 21-39.
Pollock M. S., Clarke L. M. J, and Dubé, M. G. 2007. The Effects of Hypoxia on
Fishes: From Ecological Relevance to Physiological Effects. Environ Rev
15: 1–14.
Rabalais, N. N., and Turner R. E. (eds) Coastal hypoxia: consequences for living
resources and ecosystems. American Geophysical Union, Washington,
DC.
Reed, T. E., Waples, R. S., Schindler, D. E., Hard, J. J., and Kinnison, M. T.
2010. Phenotypic plasticity and population viability: the importance of
environmental predictability. Proc of the Roy Soc B Biol Sci 277:3391 –
3400.
Reylea, R. A. 2001. The lasting effects of adaptive plasticity: Predator-induced
tadpoles become longer-legged frogs. Ecology 82:1947-1955.
154
Relyea, R. A. 2003. Predators come and predators go: The reversibility of
predator-induced traits. Ecology 84: 1840-1848.
Relyea, R. A. 2011. Adaptive plasticity in predator-induced defenses in a
Common freshwater snail: altered selection and mode of predation due to
prey phenotype. Evol Ecol 25: 189-202.
Remold, S. K., Lenski, R. E. 2001. Contribution of individual random mutations
to genotype-by-environment interactions in Escherichia coli. Proc Natl
Acad Sci USA 98:11388-11393
Reusch T., and Blanckenhorn W. U. 1998. Quantitative genetics of the dung fly
Sepsis cynipsea: Cheverud’s conjecture revisited. Heredity 81: 111–119.
Richards, R. 1992. The Meaning of Evolution. University of Chicago Press,
Chicago.
Rinaldi L., Basso P., Tettamanti G., Grimaldi A., Terova G., Saroglia M. and
DeEguileor M. 2005. Oxygen availability causes morphological changes
and a different VEGF/Flk-1/HIF-2 expression pattern in sea bass gills. Ital
J Zool 72: 103-111.
Robison, O. W. 1981. The influence of maternal effects on the efficiency of
selection: A
review. Livest Prod Sci 8:121-137.
Roff, Derek. 2000. The Evolution of the G Matrix: Selection or Drift? Heredity
84:135–142.
Roff, D. A. A. 2007. A centennial celebration for quantitative genetics. Evolution
61:1017-1032.
Rolf F. J., and Slice D. 1990. Extensions of the Procustes method for the optimal
superimposition of landmarks. Syst Zool 39: 40-59.
Rohlf, F. J. 2001. tpsDig2: a program for landmark development and analysis.
http://life.bio.sunysb.edu/morph.
Rohlf, F. J. 2010. tpsRelw, version 1.49. Stony Brook, NY: Department of
Ecology and Evolution, State University of New York.
Saint-Paul, U. 1984. Physiological adaptation to hypoxia of a neotropical
characoid fish Colossoma macropomum, Serrasalmidae. Environ Biol Fish
11: 53–62.
155
Scheiner, S. M.1993. Genetics and evolution of phenotypic plasticity. Annu Rev
Ecol Syst 24:35–68
Schlichting, C. D. 1986. The evolution of phenotypic plasticity in plants. Annu
Rev Ecol Syst 17: 667–693.
Schlichting C. D., and Pigliucci M. 1998. Phenotypic evolution : a reaction norm
perspective. Sunderland: Sinauer.
Schlichting C. D. and Smith H. 2002. Phenotypic plasticity: linking molecular
mechanisms with evolutionary outcomes. Evol Ecol 16: 189-211.
Scott G. R., and Sloman K. A. 2004. The effects of environmental pollutants on
complex fish behaviour: integrating behavioural and physiological
indicators of toxicity. Aquat Toxicol 68: 369-392.
Sgro, C. M., and Hoffmann, A. A. 1998. Effects of stress combinations on the
expression of additive genetic variation for fecundity in Drosophila
melanogaster. Gen Res 72:13-18
Shang E. H. H., and Wu R. S. S. 2004. Aquatic Hypoxia Is a Teratogen and
Affects Fish Embryonic Development. Environ Sci Technol 38: 4763-4767.
Shumway C. A. 2008. Habitat Complexity, Brain, and Behavior. Brain Behav Evol
72: 123-134.
Sih A., and Moore R. 1993. Delayed hatching of salamander eggs in response to
enhanced larval predation risk. Am Nat 142: 947-960.
Silkin Y. A., and Silkina E. N. 2005. Effect of Hypoxia on PhysiologicalBiochemical Blood Parameters in Some Marine Fish. J Evol Biochem
Phys 41: 527-532.
Slatkin M. 1987. Quantitative genetics of heterochrony. Evolution 41: 799-811.
Smit G. L., and Hattingh J. 1978. The effect of respiratory stress on carp
haemoglobin. Comp Biochem Physiol A Physiol 59: 369-374.
Sneddon L. U., Braithwaite V. A., and Gentle M. J. 2003. Novel object test:
examining nociception and fear in the rainbow trout. J of Pain 4: 431-440.
Sollid J., De Angelis P., Gundersen K., and Nilsson G. E. 2003. Hypoxia induces
adaptive and reversible gross morphological changes in crucian carp gills.
J Exp Biol 206: 3667–3673,
156
Sollid J., and Nilsson G. 2006. Plasticity of respiratory structures - Adaptive
remodeling of fish gills induced by ambient oxygen and temperature.
Respir Physiol Neurobiol 154: 241-251.
Spence R., Gerlach G., Lawrence C., and Smith C. 2008. The behaviour and
ecology of the zebrafish, Danio rerio. Biol Rev 83: 13-34.
Spicer J. L., and El-Gamal M. M. 1999. Hypoxia accelerates the development of
respiratory regulation in brine shrimp – but at a cost. J exp Biol 202: 3637–
3646.
Spicer J. L., and Burggren, W. W. 2003. Development of physiological regulatory
systems: altering the timing of crucial events. Zoology 106:91–99.
Stamatiou D., and Liew C. 2003. Gene expression profile of zebrafish exposed to
hypoxia during development. Physiol Genomics 13: 97-106.
Steppan S. J., Phillips P. C., and Houle D. 2002. Comparative quantitative
genetics: evolution of the G matrix. Trends Ecol Evol 17: 320–327.
Storey, K. B. 1988. Suspended animation-the molecular basis of metabolic
depression. Can J Zool 66: 124-132.
Strecker R., Seiler T.-B., Hollert H., and Braunbeck T. 2011. Oxygen
requirements of zebrafish (Danio rerio) embryos in embryo toxicity tests
with environmental samples. Comp Biochem Phys C 153: 318-327.
Taborsky B. 2006. The influence of juvenile and adult environments on lifehistory trajectories. Proc Roy Soc B-Biol Sci 273: 741-750.
Talwar P. K., and Jhingran A. G. 1991. Inland fishes of India and adjacent
countries. Calcutta.: Oxford & I. B. H. Publishing.
Tanaka K., Osako Y., and Yuri K. 2010. Juvenile social experience regulates
central neuropeptides relevant to emotional and social behaviors.
Neuroscience 166: 1036-1042.
Timmerman, C. M., and Chapman L. J. 2004. Hypoxia and interdemic variation in
the sailfin molly (Poecilia latipinna). J Fish Biol 65: 635-650
Ton C., Stamatiou D., and Liew C. C. 2003. Gene expression profile of zebrafish
exposed to hypoxia during development. Physiol Genomics 13: 97-106.
Tonsor S. J., and Scheiner S. M.. 2007. Plastic trait integration across a CO2
Gradient in Arabidopsis thaliana. Am Nat 169: E119–E140.
157
Tyler R. M., and Targett T. E. 2007. Juvenile weakfish, (Cynoscion regalis),
distribution in relation to diel-cycling dissolved oxygen in an estuarine
tributary. Mar Ecol Prog Ser 333: 257-269.
Vainikka A., Jokelainen T., Kortet R., and Ylönen H. 2005. Predation risk
allocation or direct vigilance response in the predator interaction between
perch (Perca fluviatilis L.) and pike (Esox lucius L.)? Ecol Freshw Fish 14:
225-232.
Van Tienderen, P. H. 1991. Evolution of generalists and specialists in spatially
heterogeneous environments. Evolution 45: 1317–1331.
Van Raaij, M. T. M., Van den Thillart G., Vianen G. J., Pit D. S. S., Balm P. H. M.,
and Steffens B.. 1996. Substrate mobilization and hormonal changes in
rainbow trout (Oncorhynchus mykiss, L.) and common carp (Cyprinus
carpio, L.) during deep hypoxia and subsequent recovery. J Comp Physiol
B 166: 443-452.
Via S. 1984. The quantitative genetics of polyphagy in an insect herbivore. 1.
Genotype-environment interaction in larval performance on different host
plant species. Evolution 38: 881–895.
Via S., and Lande R. 1985. Genotype-Environment interaction and the evolution
of phenotypic plasticity. Evolution 39: 505-522.
Via S., Gomulkiewicz R., De Jong G., Scheiner S. M.., Schlichting, CD. and Van
Tienderen, P. H. 1995. Adaptive phenotypic plasticity: consensus and
controversy. Trends Ecol Evol 10:212–217.
Vascotto S. G., Beckham Y., and Kelly G. M. 1997. The zebrafish’s swim to fame
as an experimental model in biology. Biochem Cell Biol 75:479–485.
Vulesevic B., and Perry S. F. 2006. Developmental plasticity of ventilatory control
in zebrafish, Danio rerio. Resp Physiol Neurobi 154: 396-405.
Waddington, C. H.1952. Selection of the genetic basis for an acquired character.
Nature 169:278.
Waddington, C. H. 1953. Genetic assimilation of an acquired character. Evolution
7:118-126.
Waitt D. E., and Levin D. A. 1998. Genetic and phenotypic correlations in plants:
a botanical test of Cheverud’s conjecture. Heredity 80:3 10–319.
Wang C., Li, S., Liu, Z. Xiang S., Wang J., Pang Z., Duan J. 2006.
Developmental quantitative genetic analysis of body weight and
158
morphological traits in red common carp, Cyprinus carpio L. Aquaculture
251: 219-30.
Webb P.W., and Fairchild A. G. 2001. Performance and maneuverability of three
species of teleostean fishes. Can J Zool 79: 1866-1877.
Werner E.E., and Hall D.J. 1988. Ontogenetic Habitat Shifts in Bluegill: The
Foraging Rate-Predation Risk Trade-off. Ecology 69: 1352-1366.
West-Eberhard, M. J. 1989. Phenotypic plasticity and the origins of diversity. Ann
Rev Ecol Syst 20:249-278.
West-Eberhard M. J. 2003. Developmental plasticity and evolution. New York:
Oxford University Press.
West-Eberhard M. J. 2005. Developmental plasticity and the origin of species
differences. Proc Natl Acad Sci U.S.A. 102: 6543-6549.
Westerfield M. 1994. The Zebrafish Book. Eugene: University of Oregon Press
Widmer S., Moore F. B-G., and Bagatto B. 2006. The effects of chronic
developmental hypoxia on swimming performance in zebrafish. J Fish Biol
69: 1885-1891.
Wilson A. D. M., and McLaughlin R. L. 2010. Foraging behaviour and brain
morphology in recently emerged brook charr (Salvelinus fontinalis). Behav
Ecol Sociobiol 64: 1905-1914.
Wright, S. 1932. The roles of mutation, inbreeding, crossbreeding and selection
in evolution. Proc 6th Int Cong Genet 1:356–66.
Wright D., Rimmer L. B., Pritchard V. L., Krause J., and Butlin R. K. 2003. Inter
and intra-population variation in shoaling and boldness in the zebrafish
(Danio rerio). Naturwissenschaften 90: 374-377.
Wu L., and Culver D. A. 1992. Ontogenetic diet shift in Lake Erie age-0 yellow
perch (Perca flavescens): a size-related response to zooplankton density.
Can J Fish Aquat Sci 49: 1932-1937.
Wu R. S. S. 2002. Hypoxia: from molecular responses to ecosystem responses.
Mar Pollut Bull 45: 35-45.
Yang A. P. 1993. The influence of changing resources on life-history patterns of
allocation and plasticity in female guppies. Ecology 74: 2011-2019.
Zanuy S., and Carrillo M. 1985. Annual cycles of growth, feeding rate, gross
159
conversion efficiency and hematocrit levels of sea bass (Dicentrarchus
labrax L.) adapted to two different osmotic media. Aquaculture 44: 11-25.
160