Evolutionary Ecology Concepts and Case Studies ppt

Evolutionary Ecology
This page intentionally left blank
Evolutionary Ecology
Concepts
and
Case
Studies
Edited
by
CHARLES
W.
FOX,
DEREK
A.
ROFF,
AND
DAPHNE
J.
FAIRBAIRN
OXPORD
UNIVERSITY
PRESS
2001
OXFORD
UNIVERSITY
PRESS
Oxford
New
York
Athens Auckland Bangkok Bogota Buenos Aires Cape Town

Chennai
Dar es
Salaam Delhi Florence
Hong
Kong Istanbul Karachi
Kolkata
Kuala Lumpur Madrid Melbourne
Mexico
City Mumbai Nairobi
Paris
Sao
Paulo Shanghai Singapore Taipei Tokyo
Toronto
Warsaw
and
associated companies
in
Berlin
Ibadan
Copyright
©
2001
by
Oxford University Press
Published
by
Oxford University Press, Inc.
198
Madison Avenue,
New

York,
New
York 10016
Oxford
is a
registered trademark
of
Oxford University Press.
All
rights reserved.
No
part
of
this publication
may be
reproduced,
stored
in a
retrieval system,
or
transmitted,
in any
form
or by any
means,
electronic, mechanical, photocopying, recording,
or
otherwise,
without
the

prior permission
of
Oxford University Press.
Library
of
Congress
Cataloging-in-Publication
Data
Evolutionary ecology
:
concepts
and
case studies
/
edited
by
Charles
W.
Fox, Derek
A.
Roff,
and
Daphne
J.
Fairbairn.
p. cm.
Includes bibliographical
references
(p. ).
ISBN

0-19-513154-1;
0-19-513155-X
(pbk.)
1.
Ecology.
2.
Evolution
(Biology)
I.
Fox, Charles
W. II.
Roff,
Derek
A.,
1949-
.
III.
Fairbairn, Daphne
J.
QH541
.E86
2001
577—dc21
00-053758
Cover art:
A
female broad-tailed hummingbird
(Selasphoms
platycercus)
pollinating Delphinium

nuttallianum.
Drawing
by
Mary
V.
Price, University
of
California, Riverside.
987654321
Printed
in the
United States
of
America
on
acid-free
paper
Preface
E
volutionary biology
and
ecology share
the
goals
of
describing variation
in
natural systems
and
discovering

its
functional basis. Within this com-
mon
framework, evolutionary biologists empha-
size
historical
and
lineage-dependent processes
and
hence
often
incorporate phylogenetic reconstruc-
tions
and
genetic models
in
their analyses.
Ecolo-
gists, while cognizant
of
historical processes, tend
to
explain variation
in
terms
of the
contemporary
effects
of
biotic

and
abiotic environmental factors.
Evolutionary ecology spans these
two
disciplines
and
incorporates
the
full
range
of
techniques
and
approaches
from
both. Evolutionary ecologists
consider both historical
and
contemporary
influ-
ences
on
patterns
of
variation
and
study variation
at all
levels, from within-individual variation (e.g.,
ontogenetic, behavioral)

to
variation among com-
munities
or
major
taxonomic groups.
The
overlap
between evolutionary ecology
and
ecology
is so
broad
that
some
previous
treatments
of the
field
(e.g.,
Pianka 1994) have little
to
distinguish them
from
standard ecology textbooks. However, recent
advances
in
molecular genetics, quantitative genet-
ics
(e.g., multivariate models, analyses

of
quantita-
tive
trait loci), statistical methods
for
comparative
analyses,
and
computer-intensive genetic modeling
have
enabled evolutionary ecologists
to
more
ex-
plicitly
incorporate lineage-dependent processes
and
constraints into their research programs.
In
modern evolutionary ecology, both
the
adaptive
significance
and the
"evolvability"
of
traits
are
hypotheses
to be

tested, rather than
a
priori
as-
sumptions.
As the
chapters
in
this volume
attest,
contemporary evolutionary ecologists have assem-
bled
a
very diverse
and
effective
array
of
tech-
niques
and
approaches
to
test these hypotheses.
Our
primary
objective
in
organizing this book
was to

provide
a
collection
of
readings,
as an
alter-
native
to
readings
from
the
primary literature, that
would serve
as an
introduction
to
contemporary
research programs
in
evolutionary ecology. Having
taught undergraduate
and
graduate courses
in
evo-
lutionary ecology
at our
respective
institutions,

we
recognized
the
need
for
such
a
volume
and
discov-
ered that many
of our
colleagues, including some
of
the
contributors
to
this volume,
felt
the
same
way.
We
hope
that
this
book
will
fill
this

need
and
be
suitable either
as a
textbook
for
evolutionary
ecology courses
offered
at the
graduate
or ad-
vanced
undergraduate
level,
or as a
reader
for
graduate seminars
on
this
same
topic.
We
have
asked authors
to
write their chapters
for

this audi-
ence. When writing
the
first
part
of the
book, enti-
tled "Recurring Themes," authors have assumed
that
students
have
the
equivalent
of at
least
one
undergraduate course
in
ecology
and one
course
in
genetics,
but
they have
not
assumed
any
back-
ground

in
population
or
quantitative genetics,
or
in
evolutionary theory.
As
indicated
by the
title,
the
concepts introduced
in
this section recur
throughout
the
volume
and are
fundamental
to
most research
in
evolutionary ecology.
For the
sue-
vi
Preface
ceeding
chapters (parts

II-V),
we
assume
that
stu-
dents have
a
basic understanding
of the
evolution-
ary
processes
and
concepts discussed
in
part
I.
Authors
in all
sections have also assumed that stu-
dents have read
the
preceding chapters
in the
vol-
ume, allowing chapters
to
build
on
each other

without repeatedly redefining terms
or
redevel-
oping basic concepts. However,
we
realize
that
some readers will select only
a
subset
of
chapters,
and
therefore when
specific
information
from
a
previous chapter
is
necessary
for
understanding
a
concept
or
example,
we
have tried
to

ensure
that
a
reference
to the
appropriate preceding chapter
is
included.
The
chapters
in
this volume have each been
written
by a
different
author;
all
authors
are
lead-
ing
researchers
in
their
field.
Chapters thus repre-
sent
the
current stage
of

evolutionary ecology bet-
ter
than
any
single-authored textbook could,
and
the
diversity
of
authors introduces students
to the
diversity
of
ideas, approaches,
and
opinions that
are the
nature
of
science. However,
a
multiau-
thored
textbook
presents special challenges
for
stu-
dents,
just
as

team-taught lecture course presents
challenges.
Authors vary
in the
level
at
which they
present their material
and in the
amount
of
back-
ground
that
they expect students
to
have when
reading their chapter. Authors also vary
in
their
writing styles
and
vary somewhat
in the way
that
they
organize their chapters.
We
have attempted
to

minimize
the
variation among chapters
by
provid-
ing
guidelines
to
authors,
by
asking authors
to
communicate with each other while writing their
chapters,
and by
aggressively editing
and
revising
chapters
as
needed.
We
have also tried
to
minimize
overlap among chapters
and to
ensure that chap-
ters build
on one

another. Perhaps
our
major
chal-
lenge
as
editors
was to
keep
the
volume
to a
rea-
sonable length, given
28
independently written
chapters. Each author
was
asked
to
contribute
no
more than 8000 words
of
text,
using
no
more than
six
figures,

plus tables,
and 30
references. These
restrictions precluded comprehensive reviews
and
forced
each contributor
to
select only
a few key
references.
Nevertheless, each chapter does serve
as
a
good introduction
to the
research area
by
provid-
ing
leading references
to
other reviews, books,
and
seminal
papers.
As
editors,
we
take

full
responsi-
bility
for the
resulting (and necessary) omission
of
many
additional references, perhaps equally appro-
priate
as
examples
or
case studies.
We
hope
that
readers will
be
inspired
to
delve more
fully
into
at
least
some
of the
research areas
and
will thus have

the
opportunity
to
discover
the
vast
and
detailed
literature that
we
have been unable
to
include.
Although
this volume
is
intended
to be
suitable
as
a
text
for
advanced undergraduates
or
graduate
students,
we
also
had a

second
objective—to
pro-
duce
a
volume that
is
valuable
to all
researchers
in
ecology, evolution,
and
genetics.
It is
largely
for
this reason
that
we
opted
for a
multiauthored vol-
ume
rather than
a
traditional textbook style. This
volume
is a
collection

of
chapters
that
describe
the
modern state
of
evolutionary ecology, including
in-
formed
and
thoughtful insights into where this
field
is, or
perhaps should
be,
going. Researchers
should
find
the
chapters dedicated
to
their areas
of
expertise interesting food
for
thought, while
the
chapters covering more disparate areas should pro-
vide

effective
updates
and
insights that will,
we
hope, encourage cross-fertilization.
Evolutionary
ecology
is a
very broad
and di-
verse
field
that
includes much
of
modern ecology
and
evolutionary biology. Unfortunately,
we
have
only
one
volume within which
to
cover
the
field.
We
have tried

to
include
as
many topics
as
possible
in
the
space provided,
but of
necessity, some topics
had to be
covered only
briefly
or
omitted alto-
gether. Thus,
we
have made numerous editorial
de-
cisions concerning content;
we
hope
that
you,
as
readers, will agree with most
of
these.
The

most
substantial decisions involved what topics should
be
left
out of the
book. Undoubtedly,
our
personal
interests
and
biases have influenced some
of
these
decisions,
but
most omissions
are for
practical rea-
sons.
For
example,
we
have opted
not to
include
chapters
on
speciation because numerous edited
volumes
have been dedicated

to the
topic
and it is
generally
well covered
in
general evolution text-
books.
We
have also limited
our
coverage
of
statis-
tical
and
analytical techniques
to
introductions
and
brief
descriptions within individual chapters. Read-
ers
will
not
find
specific
chapters dedicated
to
methodology such

as
chapters
on
molecular meth-
ods, methods
of
phylogenetic analysis,
or
methods
for
measuring genetic variance components. These
are
important techniques
for us all to
understand
but are
best acquired
from
specialized volumes
(e.g.,
Brooks
and
McLennan 1991; Harvey
and Pa-
gel
1991; Avise 1994;
Roff
1997).
Instead,
we fo-

cus
on
conceptual problems
and
case studies
that
may
illustrate
why and
when particular methods
Preface
vii
and
techniques
are
useful
in
evolutionary ecology,
but we do not
provide detailed recipes
for
applica-
tion
of the
methods.
In
closing,
we
express
our

gratitude
to all of the
authors contributing
to
this volume. Writing
a
book chapter
is
often
a
thankless task,
and our
stringent requirements have made
the
task espe-
cially
difficult.
We
have been
uniformly
impressed
not
only
by the
very
high quality
of the
contribu-
tions,
but

also
by
each
author's
cheerful
willing-
ness
to
respond
to our
requests
for
revision.
Al-
most
all of
these requests were made
to
standardize
the
style
of the
chapters
and
increase
the
cohesive-
ness
of the
volume

as a
whole.
To the
extent
that
we
have succeeded
in
this,
and in our
overall goal
of
providing
a
state-of-the-art introduction
to
evo-
lutionary ecology,
we
must thank
the
individual
chapter authors.
Charles
W.
Fox
Derek
A.
Roff
Daphne

J.
Fairbairn
This page intentionally left blank
Contents
Contributors
xi
Part
I.
Recurring Themes
1.
Nature
and
Causes
of
Variation
3
Susan
J.
Mazer
John
Damuth
2.
Evolutionary Significance
of
Variation
16
Susan
J.
Mazer
John

Damuth
3.
Natural Selection
29
Daphne
J.
Fairbairn
Jeff
P.
Reeve
4.
Adaptation
44
David Reznick
Joseph
Travis
5.
Phenotypic
Plasticity
58
Massimo Pigliucci
6.
Population Structure
70
Leonard Nunney
7.
Inbreeding
and
Outbreeding
Nickolas

M.
Waser
Charles
F.
Williams
84
Part
II.
Life
Histories
8.
Age and
Size
at
Maturity
Derek
A.
Roff
9.
Offspring
Size
and
Number
Frank
J.
Messina
Charles
W. Fox
10.
Senescence

128
Marc Tatar
11.
Life
Cycles
142
Jan A.
Pechenik
99
113
12. Sex and
Gender
Turk
Rhen
David
Crews
154
13. Sex
Ratios
and Sex
Allocation
Steven
Hecht
Orzack
14.
Ecological Specialization
and
Generalization
177
Douglas

J.
Futuyma
165
x
Contents
Part
III. Behavior
15.
Mating Systems
193
Ann K.
Sakai
David
F.
Westneat
16.
Sexual Selection
207
Udo
M.
Savalli
17.
Cooperation
and
Altruism
222
David Sloan
Wilson
18.
Foraging Behavior

232
Donald
L.
Kramer
19.
The
Evolutionary Ecology
of
Movement
247
Hugh
Dingle
Marcel
Holyoak
Part
IV.
Interspecific Interactions
20.
Ecological
Character
Displacement
265
Dolph
Schluter
21.
Predator-Prey Interactions
277
Peter
A.
Abrams

22.
Parasite-Host Interactions
290
Curtis
M.
Lively
23.
Plant-Herbivore Interactions
303
May
Berenbaum
24.
Mutualisms
315
Judith
L.
Bronstein
25.
The
Geographic Dynamics
of
Coevolution
331
John
N.
Thompson
Part
V.
Adaptation
to

Anthropogenic
Change
26.
Pesticide Resistance
347
John
A.
McKenzie
27.
Predicting
the
Outcome
of
Biological
Control
361
Judith
H.
Myers
28.
Evolutionary Conservation Biology
371
Philip
W.
Hedrick
References
385
Index
415
Contributors

Abrams, Peter
A.
Department
of
Zoology,
University
of
Toronto,
25
Harbord St.,
Toronto,
Ontario
M5S
3G5
Canada.
Berenbaum, May. Department
of
Entomology,
University
of
Illinois,
505 S.
Goodwin Ave.,
Urbana, Illinois
61801-3795
USA.
Bronstein,
Judith
L.
Department

of
Ecology
and
Evolutionary
Biology, University
of
Arizona,
Tucson,
Arizona
85721
USA.
Crews,
David.
Section
of
Integrative Biology,
University
of
Texas,
Austin, Texas
78712
USA.
Damuth,
John.
Department
of
Ecology, Evolution
and
Marine Biology, University
of

California,
Santa Barbara, California
93106
USA.
Dingle,
Hugh.
Department
of
Entomology,
University
of
California, Davis, California
95616
USA.
Fairbairn,
Daphne
J.
Department
of
Biology,
University
of
California, Riverside, California
92521
USA.
Fox,
Charles
W.
Department
of

Entomology,
S-225 Agricultural Science Center
North,
University
of
Kentucky, Lexington, Kentucky
40546-0091
USA.
Futuyma, Douglas
J.
Department
of
Ecology
and
Evolutionary Biology, State University
of New
York, Stony Brook,
New
York
11794-5245
USA.
Hedrick,
Philip
W.
Department
of
Biology,
Arizona State University,
Tempe,
Arizona

85287-
1501 USA.
Holyoak,
Marcel.
Department
of
Environmental
Science
and
Policy, University
of
California,
Davis, California
95616
USA.
Lively,
Curtis
M.
Department
of
Biology, Indiana
University,
Bloomington, Indiana
47405
USA.
Kramer,
Donald
L.
Department
of

Biology,
McGill University, 1205 Avenue Docteur
Penfield,
Montreal, Quebec
H3A 1B1
Canada.
Mazer,
Susan
J.
Department
of
Ecology,
Evolution
and
Marine Biology, University
of
California,
Santa Barbara, California
93106
USA.
McKenzie,
John
A.
Center
for
Environmental
Stress
and
Adaptation
Research,

Department
of
Genetics, University
of
Melbourne, Parkville,
VIC
3052
Australia.
xii
Contributors
Messina, Frank
J.
Department
of
Biology, Utah
State University, Logan
UT
84322-5305
USA.
Myers,
Judith
H.
Department
of
Zoology,
University
of
British Columbia, Vancouver,
British
Columbia,

V6T 1Z4
Canada.
Nunney, Leonard. Department
of
Biology,
University
of
California, Riverside, California
92521
USA.
Orzack,
Steven
Hecht.
Fresh Pond Research
Institute,
64
Fairfield
St., Cambridge,
Massachusetts
02140
USA.
Pechenik,
Jan A.
Department
of
Biology, Tufts
University, Medford, Massachusetts
02155
USA.
Pigliucci,

Massimo.
Department
of
Botany,
University
of
Tennessee,
Knoxville, Tennessee
37996-1100
USA.
Reeve,
Jeff
P.
Department
of
Biology, Concordia
University, 1455
de
Maisonneuve
Blvd.
West,
Montreal, Quebec,
H3G
IMS
Canada.
Reznick, David. Department
of
Biology,
University
of

California, Riverside, California
92521
USA.
Rhen,
Turk.
Laboratory
of
Signal Transduction,
National Institute
of
Environmental Health
Sciences,
National
Institute
of
Health,
Research
Triangle Park,
North
Carolina
27709
USA.
Roff,
Derek
A.
Department
of
Biology, University
of
California,

Riverside,
California
92521
USA.
Sakai,
Ann K.
Department
of
Ecology
and
Evolutionary Biology, University
of
California,
Irvine, California,
92697
USA.
Savalli,
Udo M.
Department
of
Entomology,
University
of
Kentucky, Lexington, Kentucky
40546-0091
USA.
Schluter,
Dolph.
Department
of

Zoology,
University
of
British Columbia,
6270
University
Blvd.,
Vancouver,
British Columbia,
V6T 1Z4
Canada.
Tatar,
Marc.
Department
of
Ecology
and
Evolutionary Biology, Brown University,
Box
G-W,
Providence, Rhode Island
02912
USA.
Thompson,
John
N.
Department
of
Ecology
and

Evolutionary Biology,
Earth
and
Marine
Sciences
Building,
University
of
California, Santa Cruz,
California
96064
USA.
Travis,
Joseph.
Department
of
Biological Sciences,
Florida State University, Tallahassee, Florida
32306
USA.
Waser,
Nickolas
M.
Department
of
Biology,
University
of
California, Riverside, California
92521

USA.
Westneat,
David
F.
Center
for
Ecology, Evolution
and
Behavior, School
of
Biological Sciences,
101
Morgan Building, University
of
Kentucky,
Lexington, Kentucky
40506-0225
USA.
Williams, Charles
F.
Department
of
Biological
Sciences, Idaho State University, Pocatello, Idaho
83209
USA.
Wilson,
David Sloan. Department
of
Biological

Sciences, State University
of New
York,
Binghamton,
New
York
13902-6000
USA.
PART
I
RECURRING
THEMES
This page intentionally left blank
1
Nature
and
Causes
of
Variation
SUSAN
J.
MAZER
JOHN
DAMUTH
T
he
field
of
evolutionary ecology
is at its

core
the
study
of
variation within individuals, among
in-
dividuals,
among populations,
and
among species.
For
several reasons, evolutionary ecologists need
to
know
the
causes
and the
effects
of
variation
in
traits
that
influence
the
performance, behavior,
longevity,
and
fertility
of

individuals
in
their natu-
ral
habitats. First,
to
determine whether
the
condi-
tions
for
evolution
by
natural selection
of
traits
of
interest
are
fulfilled,
we
need
to
know
the
degree
to
which
the
phenotype

of a
trait
is
determined
by
the
genetic constitution
(or
genotype)
of an
indi-
vidual
and by the
environment
in
which
an
individ-
ual
is
raised. Second,
to
predict whether
and how
natural selection will cause
the
mean phenotype
of
a
trait

in a
population
to
change
from
one
genera-
tion
to the
next,
we
must understand
the
ways
in
which
an
individual's phenotype (for this trait)
in-
fluences
its
genetic contribution
to
future
genera-
tions
(i.e.,
its
fitness).
Third,

to
understand
why
the
phenotype
of a
given trait
influences
an
indi-
vidual's
fitness,
we
need
to
know
how the
trait
af-
fects
an
individual's ability
to
garner resources
for
growth
or
reproduction,
to
avoid predation,

to
find
mates,
and to
reproduce
successfully.
Finally,
to
evaluate
whether
the
phenotypic
differences
we
observe
among populations
and
species
may
rep-
resent
the
long-term outcome
of
evolution
by
natu-
ral
selection,
we

must understand
how
different
phenotypes perform under
different
environmental
conditions.
In
sum, with
an
understanding
of the
causes
and
consequences
of
phenotypic variation
within
and
among populations,
we can
detect evo-
lutionary processes operating
at a
variety
of
eco-
logical levels: within random-mating populations;
within
and

among subpopulations distributed over
a
species' geographic range;
and
even
among
mul-
tispecies
associations. These goals, however,
re-
quire
a
clear understanding
of the
nature
of
pheno-
typic
variation.
The aim of
this chapter
and the
next
is to
illus-
trate
that
the
richness
of

evolutionary ecology
has
increased
in
direct
proportion
to our
understand-
ing of the
multiple causes
of
intraspecific
pheno-
typic
variation.
Before
reviewing these sources
of
variation,
it is
worth
considering
briefly
a
funda-
mental question:
What
kind
of
variation

is
evolu-
tionarily
significant?
Any
trait
whose phenotype
is
reliably transmit-
ted
from
parents
to
offspring
over multiple genera-
tions
has the
potential
to
evolve.
The
rules
of
Men-
delian genetics tell
us
that
traits
whose phenotypes
are

determined
by
nuclear genes
operating
in an
additive
manner
(i.e.,
alleles
whose
effects
are
inde-
pendent
of the
genetic background
in
which they
are
expressed)
are
most
likely
to
fulfill
this crite-
rion. Indeed,
the
importance
of

this kind
of
inheri-
tance
has
been considered
so
great
that
the
propor-
tion
of
total
phenotypic variance
in a
trait
that
is
due to the
additive
effects
of
nuclear genes
is
given
a
special term:
heritability.
But

what about traits
that
are
partly
or
largely
3
4
Recurring Themes
influenced
by
nonnuclear genes, nonadditive inter-
actions among
alleles
or
loci,
the
maternal environ-
ment,
an
individual's current environment,
the in-
teraction between
an
individual's genotype
and its
environment,
or the age or
developmental stage
of

the
organism exhibiting
them?
Over
the
last dec-
ade,
it has
become clear
not
only that such traits
are
ubiquitous
in
natural
populations,
but
that
they
can
evolve
as
well. Unlike Mendelian traits,
however,
the
evolutionary
trajectory
of
traits sub-
ject

to
these
effects
can be
difficult
to
predict.
The
rate
or
direction
of
their evolution
can
depend
on
the
degree
and
nature
of
population
structure,
in-
teractions among individuals,
and
nonrandom
mating.
In
addition, genetic

drift
can
take
on
spe-
cial importance
in
promoting
the
differentiation
of
populations when nonadditive sources
of
variation
are
prevalent.
Consequently,
for
evolutionary ecologists inter-
ested
in
predicting evolutionary change
in a
partic-
ular
trait
in a
given
population,
it is

important
not
only
to
determine whether traits
are
transmitted
from
parents
to
offspring,
but
whether they
are
transmitted
in a
predictable fashion.
An
under-
standing
of all
potential sources
of
phenotypic
variation
in a
trait
helps
to
achieve

this
goal.
This
chapter reviews
the
kinds
of
variation
that
interest
evolutionary
ecologists
and
notes their relevance
to
particular evolutionary questions.
In
addition,
causes
of
variation within individuals
are
intro-
duced.
Chapter
2
considers components
of
varia-
tion among individuals. Together, chapters

1 and
2
consider
the
causes
and
evolutionary conse-
quences
of
variation
in
both unstructured
and
structured populations,
and we
highlight
our
view
that
new
insights into
the
potential
for
natural
se-
lection
to
cause phenotypic change
in

wild species
will come
from
the
study
of
subdivided popula-
tions
in
which mating
is
anything
but
random (see
also Nunney, this volume).
Modes
of
Expression
of
Variation
Predicting
the
outcome
of
natural selection
on
eco-
logically
important traits depends
on

being able
to
determine
the
quantitative relationships between
phenotype, genotype,
and
fitness.
If the
phenotypic
variation
in a
trait
is
genetically based
and
corre-
lated with
the
fitness
of
individuals
in a way
that
can be
expressed mathematically, then
it is
possible
to
predict

the
direction
in
which
the
trait
should
evolve
(Fairbairn
and
Reeve, this
volume).
The
pattern
of
variation expressed
by a
trait,
however,
has
a
strong
influence
on the
quantitative
and ex-
perimental methods used
to
detect
and to

measure
this relationship.
Discrete
Traits
Traits
whose phenotypes
can be
assorted into dis-
tinct, nonoverlapping classes exhibit discrete varia-
tion.
The
phenotypic
frequency
distributions
of
such
categorical
or
qualitative traits
are
usually
de-
picted
as
histograms,
where
the
phenotypic catego-
ries
are

indicated
on the
#-axis,
and the
number
or
proportion
of
sampled individuals identified
in
each
category
is
indicated
on the
y-axis
(figure
1.1).
Often, such traits
are
simple Mendelian traits
controlled
by a
single locus. While
the
color
and
surface
texture
of

Mendel's peas
and the
wing
color
of the
peppered moth
(Biston
betularia)
pro-
vide excellent
if
time-worn examples
for
introduc-
tory biology students, many other discretely inher-
ited
traits provide evidence
for the
potential
for (or
the
limitations
of)
natural selection
to
mold genetic
variation.
When
the
frequencies

of
multiple morphs
are
high enough that their abundances cannot
be ac-
counted
for by
mutation alone, this
is
identified
as
a
polymorphism
of
considerable evolutionary
in-
terest.
In
such cases,
it
would appear that natural
selection
may not be
effective
at
eliminating
an in-
ferior
genotype. Alternatively, either
the

morphs
may
be
identical with respect
to
both survivorship
and
reproduction,
or the
morphs
may
enjoy
equal
fitnesses
because where one, say,
has an
advantage
in
fertility,
another
has an
advantage
in
survivor-
ship. Other possibilities
are
that each
morph
enjoys
a

fitness
advantage
in a
particular microenviron-
ment (where
the
population
occupies
a
heteroge-
neous habitat), that
the
relative performance
of the
morphs varies over time,
or
that
the
performance
of
a
given morph
is
gender-specific. Detecting
the
processes responsible
for the
maintenance
of
such

polymorphisms
in
natural
populations
is a
chal-
lenge
for
evolutionary ecologists,
and
numerous
studies have aimed
to do so.
In
many animals,
for
example, body color
is a
discrete trait that varies among individuals
and af-
fects
their vulnerability
to
their predators.
For ex-
ample,
the
adder,
Viperus
berus, exhibits

two
dor-
sal
color patterns: black
and
zig-zag.
In a
6-year
mark-recapture study
of
island
and
mainland pop-
Figure
1.1
Examples
of
frequency distributions
of
discrete
traits
found
in
natural populations.
(A)
Fre-
quencies
of
alternative style morphs
in one

eastern
Ontario
population
of the
tristylous perennial plant,
Decodon
vertidllatus
(Lythraceae). This population exhibits
a
marked deficiency
of
mid-style morphs
that
is
persistent over time (Eckert
and
Barrett 1995).
(B)
Mean frequencies
of
style morphs
in the
tristylous
perennial
herb
Lythrum
salicaria
(purple loosestrife; Lythraceae) sampled from
populations
in

northern
and
central Sweden, where
it is
native,
and in
Ontario,
where
it has
been introduced.
The
regional
differences
in
morph
frequencies
are
thought
to be the
result
of
both
fitness
differences
between
the
morphs
and
historical factors
(Agren

and
Ericson 1996).
N
represents
the
number
of
populations
sampled from each region.
(C)
Diploid genotype frequencies
for the
malate dehydrogenase locus sampled
in
three Eastern Ontario populations
of
Decodon vertidllatus (Eckert
and
Barrett
1993).
The
genotypic
frequencies
do not
differ
significantly from
Hardy-Weinberg
expectations.
(D)
Frequencies

of
banded,
intermediate,
and
unhanded
body patterns
in
Lake Erie water snakes, Nerodia sipedon
insularum,
in
different
age
classes.
The
relative abundances
of the
different
morphs were
not
found
to
differ
signifi-
cantly among
age
classes, suggesting
that
there
is no
strong selection

on
body color patterns
in
this
sample
(King
1987).
6
Recurring Themes
ulations
in and
near Sweden, Forsman (1995) found
that
the
relative performance
of
these
two
morphs
differed
between males
and
females.
In
male snakes,
the
black form
suffered
lower survival (apparently
due

to
increased predation) than
the
zig-zag form,
but in
females,
the
pattern
was
reversed. Whether
this pattern
of
gender-specific relative
fitness
ac-
counts
for the
maintenance
of the
polymorphism
is
not
certain. However,
the
relationship between body
color
and
survivorship
is
clearly gender-specific, sug-

gesting that
the
fitness
of a
given morph depends
on
the
behavior
of the
individual exhibiting
it.
Individ-
uals
of the
Lake Erie water snake,
Nerodia
sipedon
insularum,
also exhibit qualitative variation
in
body
color pattern
(King
1987).
Differences
in the
relative
abundances
of
banded

and
unbanded
morphs
be-
tween mainland
and
island populations,
and
among
young-of-the-year,
juvenile,
and
adult individuals,
suggest
that
the
probability
of
predation
is
both
en-
vironment-specific
and
morph-specific.
Electrophoretic variation
is
less
convenient
to

assess
than visible polymorphisms,
but
several
studies
suggest
that
selection acts
on
allozymes
1
(or
on
closely linked genes)
that
appear
to
influence
fitness
through
their
physiological
effects.
Carter
(1997)
found
that
among populations
of
gilled

ti-
ger
salamanders
(Ambystoma
tigrinum
nebulosum)
living
in
ephemeral ponds, there
is a
positive corre-
lation between
the
frequency
of
homozygotes
for
the
"slow"
allele
at the
alcohol dehydrogenase
lo-
cus
(Adh-SS
genotypes)
and
oxygen availability
in
the

ponds. This geographic pattern
is
consistent
with environment-specific selection under controlled
conditions
and
with
temporal patterns
of
selection
within ponds; relative
to the
Adh-FF
(Fast/Fast
ho-
mozygotes)
and
Adh-FS
(Fast/Slow
heterozygotes)
genotypes, Adh-SS genotypes
are
selected against
under low-oxygen
conditions.
Discrete traits controlled
by one or a few
loci
are of
special interest

to
evolutionists because pop-
ulation genetics theory allows
the
derivation
of
precise predictions
of
allele frequency changes
from
generation
to
generation.
If the
genetic basis
of
the
phenotypic categories
is
well understood,
al-
lele
frequencies
can be
precisely measured
and
tracked, providing
a
powerful tool
for

testing evo-
lutionary
hypotheses.
Quantitative
Traits
In
contrast
to
discrete traits
are
those
for
which
the
phenotype
varies along
a
continuum
and is
deter-
mined
by
alleles
at
multiple loci.
The
frequency
distribution
of
such quantitative

or
"metric"
traits
is
often
illustrated
as a
histogram
(as in the
case
of
discrete traits),
but
here
the
x-axis
is
arbitrarily
divided
into convenient categories
that,
in
sum,
il-
lustrate
the
shape
of the
distribution
(figure

1.2).
(Indeed,
the
width
of the
categories
can
have
a
very
strong
effect
on the
shape
of the
resulting distribu-
tion.)
The
y-axis
shows
the
proportion
or
number
of
all
sampled individuals whose "phenotypic
value"
for the
trait

falls
within
the
boundaries
of
each category. Within populations, such polygenic
traits
often
exhibit
a
normally distributed fre-
quency
distribution, although
for
many traits
the
raw
values
of the
individuals' phenotypes must
be
transformed (e.g., log-
or
arcsine-transformed)
to
provide
a
scale that will generate
a
bell-shaped

curve.
Where
the
value
of a
trait must
be
expressed
as an
integer (e.g.,
the
number
of
anthers
per
flower),
the
boundaries between
the
categories
are
less
arbitrary,
but the
trait
may
nevertheless behave
as
a
quantitative trait

and be
controlled
by
multi-
ple
loci.
The
alleles
and
loci that
influence
a
quantitative
trait
may
each
contribute
additively
to
phenotype,
whereby
the
change
in the
phenotypic value
of a
trait caused
by an
allelic replacement
at a

locus
is
independent
of
both
the
other allele
at
that locus
and the
genotypes expressed
at
other loci. Alterna-
tively, alleles
and
loci
may
interact
so
that
the ef-
fect
on
phenotype
of an
allelic change
at a
locus
depends
on the

alleles
or
genotypes
at
this
or
other
loci (i.e., dominance
and
epistasis; Mazer
and Da-
muth,
chapter
2,
this volume). Only
in the
absence
of
dominance
and
epistasis,
and
where mating
is
random,
can
precise predictions
be
made concern-
ing

the
similarity between parents
and
offspring
or
the
phenotypic response
to
selection.
Many evolutionary ecologists focus exclusively
on the
evolution
of
quantitative traits simply because
so
many traits
of
known ecological importance
and
with strong
effects
on
fitness
are
continuously
distributed
or
known
to
behave

as
quantitative
ge-
netic
traits.
These include life-history traits such
as
germination time,
juvenile
growth rate,
age of
first
reproduction, clutch size, number
of
reproductive
bouts,
longevity, fecundity,
and
fitness itself;
be-
havioral traits such
as
running speed, foraging
rate,
and the
rate
of
resource acquisition; physio-
logical traits, such
as

photosynthetic rate
or
meta-
bolic
efficiency;
and
fitness-related morphological
traits, such
as
size
at
birth
and
adulthood, bill size
Nature
and
Causes
of
Variation
Figure
1.2
Frequency distributions
of
quantitative
traits found
in a
greenhouse-raised experimental
population
of the
annual salt marsh plant

Spergu-
laria
marina (Caryophyllaceae) (Mazer
et
al.
1999).
(A)
The
frequency
distribution
of the
mean num-
ber of
ovules produced
per
flower.
(B)
Distribution
of
the
mean number
of
developmentally normal
anthers produced
per
flower.
(C)
Frequency distri-
bution
of the

mean area
of all
petals produced
by
a
flower.
N=
1179
individuals
for all
traits.
The
normal distribution corresponding
to the
mean
and
variance
of
each
trait
is
superimposed
on the
histogram
of the
actual
frequency
distribution
of
phenotypic values.

in
birds, wing length relative
to
body size,
and the
expression
of
secondary sexual characters such
as
tail
length, flower size
and
color,
and the
size
of
color spots
on
bodies, wings,
or
petals.
One
appealing attribute
of
normally distributed
traits
is
that their statistical properties
are
well

known
(figure
1.3; Falconer
and
MacKay
1996).
This means that
it is a
simple matter
to ask
wheth-
er
population
or
species means
differ
significantly,
potentially
reflecting
the
direct outcome
of
evolu-
tion
by
natural selection
(figure
1.4; Mazer
and
Lebuhn 1999; Reznick

and
Travis,
this
volume).
It
is
also possible
to
conduct controlled breeding
experiments
from
which
to
estimate
the
proportion
of
total phenotypic variance that
is due to
environ-
mental versus genetic sources (e.g., Reznick
and
Travis, this volume).
In
addition, statistical meth-
ods
have been derived
to
predict
accurately

the re-
sponse
to
artificial
and
natural selection
on
nor-
mally
distributed traits (Fairbairn
and
Reeve, this
volume).
Currently, agriculturalists
and
evolution-
ary
ecologists alike
use
these methods both
to
esti-
mate genetic
and
environmental causes
of
phe-
notypic variance
and to
predict

how the
mean
phenotype
of
populations will
(or
should) change
from
generation
to
generation
in
response
to
artifi-
cial
or
natural selection.
The
study
and
description
of
quantitative traits
require
some
familiarity
with
the
statistical param-

eters
that
can be
measured given
a
sample
of
data
representing
a
continuous variable.
Any set of ob-
servations
of a
quantitative trait
can be
summa-
rized
by its
mean
(or
average), variance, standard
deviation, standard error
of the
mean,
and
coeffi-
cient
of
variation (among others)

(figure
1.3).
The
mean
and
standard deviation
are
parameters
re-
ported
in the
units
in
which they
are
measured;
the
variance
is a
function
of the
square
of
these units
and so is
generally reported simply
as a
number.
One
problem with using these parameters

to
char-
acterize
a
trait emerges when
one
aims
to
compare
the
variability
of two or
more
traits.
This
is
often
a
first
step
when
attempting
to
predict
which
traits
may
most
easily respond
to

natural selection.
All
else
being equal, traits exhibiting high levels
of
phenotypic
and
genetic variation have
a
higher
po-
tential
to
undergo evolutionary change than those
that
do
not.
However,
given that
the
units
in
which
variation
is
measured
are
often trait-specific (with
mass reported
in

grams, length
in
linear units, vol-
ume in
cubed units, color
in
wavelengths, etc.),
it
is
often
meaningless
to
use, say,
the
standard devi-
7
Figure
1.3
Statistical properties
of
quantitative traits. Top:
The
shape
of a
normal distribution,
for a
hypothetical trait whose
values
range between
40 and 180

units.
The
following parame-
ters
can be
estimated
for all
quantitative traits measured
on a
sample
of
individuals representing
a
laboratory
or
field
popula-
tion:
the
sample mean, variance,
and
standard deviation. Bot-
tom: Frequency distributions
of
floral spur length
for two
species
of
Aquilegia
collected from

field
populations
in the
Bishop Creek
Drainage
(Inyo
County, California). Flowers
of
Aquilegia for-
mosa
(N=
129) were sampled between 1950
and
2780
m
in
ele-
vation; flowers
of A.
pubescence
(N
-
236) were collected
at
elevations
of
between
3400
and
3950

m. One
flower
per
plant
was
sampled (Scott
Hodges,
unpublished data).
The
mean spur
lengths
of the two
species
differ
significantly.
8
Figure
1.4
Variation among populations
in the
frequency
distribution
of
two
quantitative traits.
The
frequency
distributions
of
ovule production

and
developmentally
normal anthers
per
flower
are
shown
for
each
of
four
greenhouse-raised
populations
of
Spergularia
marina. Each population
was
derived
from
seeds collected
from
a
distinct wild population. There
is
signif-
icant variation among populations with respect
to the
mean number
of
ovules

and
anthers produced
per
flower under greenhouse
conditions (Ma-
zer
and
Delesalle
1996).
Arrows indicate
the
phenotypic mean
of the
trait
for
each
frequency
distribution.
9
10
Recurring
Themes
ation
to
compare
the
variability
of two or
more
traits.

Various solutions have been
proposed
to
solve
this problem, including
the use of
dimension-
less
parameters such
as the
coefficient
of
variation.
Because
the
statistical
and
mathematical prop-
erties
of
normal distributions
are
well known
and
tractable, theoretical models
of the
evolution
of
quantitative traits (for which
a

normal distribution
is
assumed) have been well developed.
The
success
of
these models
in
predicting
the
response
of a
trait
to
selection, based
on its
heritability
and on the
strength
of
selection,
affirms
the
appropriateness
of
a
quantitative genetic model
of
inheritance
for

many
continuously distributed traits (Falconer
and
MacKay
1996).
A
summary
of the
statistical meth-
ods
used
to
estimate
the
heritability
of
quantitative
traits
and to
predict their evolutionary trajectories
is
beyond
the
scope
of
this chapter
but is
available
in
several recent volumes (Falconer

and
MacKay
1996;
Roff
1997; Lynch
and
Walsh 1998).
It is
worth mentioning, however, what
is
perhaps
the
most
useful
theoretical
and
practical
contribution
to
evolutionary ecologists
by
quantitative geneti-
cists—the
analysis
of
variance.
The
statistical anal-
yses
now

conducted routinely
to
detect
the
causes
of
variation
in
quantitative traits
and
their
effects
on
fitness
and on
each
other,
and to
identify
pat-
terns
of
temporal
and
geographic variation would
not be
possible without
Sir
Ronald Fisher's inven-
tion

of the
analysis
of
variance
(Fisher
1925).
Threshold
Traits
A
special case
of
discrete inheritance
is
represented
by
threshold
traits.
These
are
traits
for
which
the
phenotypes
can be
assigned
to one of two or
more
distinct classes,
but

that
are
determined
by
alleles
at
multiple loci.
The
loci
affecting
a
threshold trait
each
have
a
relatively small
effect
on
some under-
lying
trait
that varies continuously, such
as the
concentration
of a
chemical product,
the
rate
of
development,

or
metabolic
efficiency.
Genotypes
expressing less than some critical
(or
"threshold")
value
of
this underlying
trait
will exhibit
one
phe-
notypic value, while those expressing more than
the
critical value will exhibit
an
alternative pheno-
typic state.
In
other
words,
there
are
discontinu-
ities
in the
phenotypic expression
of a

continuous
underlying
variable.
Two-class (dimorphic) threshold traits
may ex-
hibit
the
expected
(3:1)
Mendelian ratios
in the F2
generation produced
by
crosses among
the
Fl
progeny
of
parents representing
the two
pheno-
typic
classes,
but the
expected ratios
do not
appear
when conducting backcrosses.
The
underlying trait

on
which
the
threshold
is
based
is
inherited
as a
quantitative trait
and is
termed
the
liability.
The
heritability
of a
discrete threshold
trait
is
therefore
estimated
as the
heritability
of its
underlying lia-
bility.
The
relationships between
a

threshold trait
and
quantitative
traits
(such
as
size, fecundity,
and
fit-
ness)
are
often
nonlinear,
and the use of
highly con-
trolled breeding designs
and
selection experiments
to
evaluate these relationships
can
strengthen con-
clusions
concerning their inheritance
and
covaria-
tion
(Roff
et
al.

1999).
For
example,
Roff
et
al.
(1999)
used both approaches
to
examine
the ef-
fects
of a
wing dimorphism
(a
threshold trait)
on
fecundity
(a
quantitative trait)
in
female
sand
crickets
(Gryllus
firmus).
They
found
that
short-

winged
flightless
females
have smaller
flight
mus-
cles
but
higher
fecundity
than
the
long-winged
morph,
and
that
both wing morph
and
fecundity
have
a
quantitative genetic basis. Moreover,
artifi-
cial
selection experiments confirmed
the
interpreta-
tion
that
there

is an
intrinsic negative correlation
between wing length
and
fecundity. Selection
fa-
voring individuals with high
(or
low) fecundity
re-
sulted
in a
direct increase
(or
decrease)
in
fecundity
and a
correlated increase
(or
decrease)
in the
pro-
portion
of
flightless
females.
Both wing morphs
persist
in

natural populations because spatial
and
temporal heterogeneity
in
habitat persistence con-
tinually
shifts
the
balance between selection
for
movement among patches
(flight)
and
rapid popu-
lation growth within patches
(flightlessness).
As
in the
case
of
quantitative traits, dimor-
phisms
or
polyphenisms (where multiple pheno-
typic states exist)
that
behave
as
threshold traits
can be

subject
both
to
strong environmental
and
genetic
influences.
The
beetle
Onthophagus
taurus
provides
an
example
of
dietary
effects
on
pheno-
type (Moczek
1998).
Males
of
this
species
are di-
morphic with respect
to
horn development;
the

presence
or
absence
of
horns
is
determined
by the
quality
and
quantity
of the
food they receive
from
their parents.
In
other species, dimorphisms
are
strongly
associated with
a
highly heritable
trait.
Quantitative genetic analyses
of
juvenile
hormone
esterase
in the
crickets Gryllus firmus

and G.
as-
similis
indicate that this enzyme (which degrades
juvenile
hormone)
is
highly heritable, responds
Nature
and
Causes
of
Variation
11
rapidly
to
selection,
and is a
strong determinant
of
wing morph (Fairbairn
and
Yadlowoski 1997;
Roff
et
al.
1997; Zera
et
al.
1998).

Sexually
Dimorphic
Traits
In
organisms with separate sexes,
it is
common
to
observe
that
many traits
differ
between males
and
females.
Sexually dimorphic traits often play
a
role
in
attracting
or
competing
for
mates
or in
raising
offspring.
Where behavior
and its
concordant risks

of
mortality
are
highly gender-specific,
one may
expect traits that
influence
fitness
to
evolve
differ-
ently
in
males
and
females. Such traits
may
include
body color, body mass, pheromone production,
and
mating calls;
the
expression
of
secondary sex-
ual
traits such
as
physical ornaments, flower size,
or

nuptial
gifts;
and
parental care (Andersson
1994; Fairbairn
and
Reeve, this volume; Savalli,
this volume). Dimorphisms
may
also evolve where
the
sexes
differ
in
other social behaviors
or in
habi-
tat
preferences.
In
either case, gender-specific traits
are
usually interpreted
as
being
the
direct
or
indi-
rect result

of
gender-specific patterns
of
sexual
or
natural selection.
Traits
favored
in
males
due to
their positive
ef-
fects
on
mating success
are not
necessarily favored
among
females,
and
vice versa. Where
female
choice
is a
major component
of
male reproductive
success,
sexual selection favoring elaborate court-

ship
behaviors
or
visually attractive traits will
be
restricted
to
males. Where
the
outcome
of
direct
competition among males determines their repro-
ductive
success,
sexual
selection
favoring large size
or
aggressive behavior
may be
stronger
in
males
than
in
females.
Similarly,
where there
are

differences
in the be-
havior
of
males
and
females unrelated
to
mating,
the
optimum
phenotype
may
differ
between
the
sexes.
For
example, where males spend more time
foraging
than females, natural selection favoring
cryptic coloration
may be
stronger among
the
for-
mer. Traits
for
which
the

phenotype favored
in one
sex is
actually selected against
in the
other
sex are
termed sexually antagonistic characters
(Rice
1984).
When
the
direction
of
selection
is
gender-specific,
if
there
is a
genetic mechanism (such
as
X-linkage)
that
permits
the
expression
of a
trait
to be re-

stricted
to one
sex, natural selection
can
result
in
significant
differences
between
the
sexes
in
either
discrete
or
quantitative traits.
Hundreds
of
cases
of
sexual dimorphism have
been
documented
in
animals
and in
plants.
Recent
studies
provide evidence

that
the
dimorphism
is the
result
of
gender-specific
patterns
of
sexual
or
natu-
ral
selection (e.g., Fairbairn
and
Reeve, this vol-
ume; Savalli, this volume).
For
example, Grether
(1996)
reports evidence
that
the red
wing spots
re-
stricted
to
male rubyspot
damselflies
are the

result
of
selection operating during competition among
males
for
mating territories, rather than
the
result
of
female
choice. Bisazza
and
Marin (1995) argue
that
the
sexual size dimorphism observed
in the
eastern mosquitofish
Gambusia
holbrooki,
in
which
the
males
are
smaller than
the
females,
is the
result

of the
mating advantage
enjoyed
by
rela-
tively
small males during most
of the
reproductive
season. Gwynne
and
Jamieson
(1998)
suggest
that
the
evolution
of the
huge mandibles
in
male alpine
wetas
(Hemideina
maori,
Orthoptera)
of New
Zealand represent "cephalic weaponry"
that
have
evolved

in
response
to
male-male battles
for
access
to
females.
Similarly,
the
body size dimorphism
in
marine iguanas
(Amblyrhynchus
cristatus)
appears
to be the
result
of
sexual selection favoring larger
size
more strongly
in
males than
in
females
(Wikel-
ski
and
Trillmich

1997).
Above
a
given body size,
female
marine iguanas allocate resources
to ad-
ditional
egg
production rather than
to
increased
growth, although both sexes grow
to be
larger than
the
apparent naturally selected optimum. Balmford
et
al.
(1994) provide comparative data suggesting
that sexual dimorphism
in
wing length among
57
species
of
sexually dimorphic long-tailed birds
is
the
result

of
natural selection occurring concur-
rently
with
sexual
selection
on
tail
length.
They
ar-
gue
that
the
secondary evolution
of the
wing size
dimorphism serves
to
offset
the
functional costs
in-
curred
by the
sexually selected tails.
Evidence
for
dimorphisms
due to

sexual selec-
tion
is not
restricted
to
animals with complex
so-
cial interactions. Dioecious plant species also
ex-
hibit
marked sexual dimorphism
in
traits related
to
mating success
(Delph
et al.
1996).
Gender-specific
sexual
selection
may be
expected
in
species
in
which male success
in
delivering pollen
to

females
is
mediated
by
pollinators. Among male plants,
re-
productive success
(or at
least pollen removal)
of-
ten
increases linearly with visitation
by
pollinators;
males
benefit
from
multiple visits
per
flower,
as
only
a
fraction
of
their pollen
is
removed
by any
single

visit.
By
contrast,
reproductive success
by fe-
males
is
often
limited
not by
pollen
but by
other
12
Recurring Themes
resources;
the
result
is
that
relatively
few
pollinator
visits
are
required
to
achieve maximum seed set,
and
females

do not
benefit
from
investing
in at-
tractants
beyond
those
necessary
to
achieve
this
maximum.
The
evolutionary outcome
of
this disparity
is
that
traits
favored
in
males (highly
conspicuous,
relatively
large,
or
profusely
flowered inflores-
cences)

differ
from
those favored
in
females
(smal-
ler-
or
fewer-flowered inflorescences). Accordingly,
the
smaller
but
more numerous flowers
in the in-
florescences
of
male relative
to
female
Silene
lati-
folia
(Meagher 1992)
may be the
result
of
competi-
tion among males
to
attract

pollinators. Similarly,
the
more numerous flowers
of
male relative
to fe-
male
plants
or
inflorescences
of
Wurmbea
dioica,
Salix
myrsinifolia-phylicifolia,
and
Ilex
opaca
ap-
pear
to be due to
gender-specific selection. Analo-
gous
to
Balmford
et
al.'s
(1994) observations con-
cerning
the

evolution
of
sexually dimorphic wing
size
in
birds,
the
dimorphism
in
sexually selected
traits
in
plants seems
to
result
in the
evolution
of
gender-specific
life-history traits. Male
and
female
plants have also been found
to
differ
in
growth
rates, phenology, frequency
of
reproduction,

and
plant height. These
differences
may
represent
the
outcome
of
selection
in
females, which usually
ex-
hibit
a
much higher absolute investment
in
repro-
duction (due
to
fruit
and
seed production)
than
do
males.
Causes
of
Variation
and
Their

Evolutionary
Consequences
Regardless
of the
kind
of
variation exhibited
by a
trait,
predicting
its
evolutionary
trajectory
requires
knowledge
of its
environmental
and
genetic basis.
In
the
remainder
of
this chapter,
we
consider
the
causes
and
evolutionary consequences

of
pheno-
typic variation within individuals. Then
(in
chapter
2),
we
consider
the
causes
and
consequences
of
variation
among individuals
in
random-mating,
unstructured populations,
and we
describe recent
conceptual advances concerning
the
consequences
of
population structure.
In
structured populations,
genotypes
do not
interact

at
random,
and the re-
lationship between genotype
and
fitness
can be
highly sensitive
to the
identity
of the
genotypes
with which
an
individual interacts (Nunney, this
volume; Wilson, this volume).
Variation
within Individuals
Ontogenetic
Variation
Ontogenetic variation
is the
component
of
phenotypic
variation
in a
trait
ex-
pressed

as an
individual ages
or
progresses through
sequential developmental stages. Ontogenetic varia-
tion
can be
expressed
in two
ways.
First
are
age-
related changes
in an
individual's
phenotype.
Traits
such
as
body size, growth rates, hormone produc-
tion,
pigmentation,
metabolic
rate,
and
behavior
are
usually age-dependent. Second
are

changes
in
the
phenotype exhibited
by
sequentially produced
or
"modular"organs.
For
example,
the
size
of se-
quentially
produced leaves, flowers,
or
fruits
may
change over time. Similarly,
the
size
or
number
of
seeds
or
eggs produced
in
successive
fruits

or
clutches
may
change over time.
For
traits that exhibit either type
of
Ontogenetic
variation, genetic
and
temporal sources
of
varia-
tion will
be
confounded unless special measures
are
taken
to
separate their
effects.
If
Ontogenetic
variation comprises
a
sufficiently
large
proportion
of
total phenotypic variance,

the
proportion
of
phenotypic variance that
is
genetically based
may
be
obscured
to the
point
of
being undetectable
un-
less
ontogenetic sources
of
variation
are
taken into
account.
Consider
a
population
of
individuals
for
which
a
trait's

phenotype varies over time
(figure
1.5).
When sampling this population,
the
magnitude
of
phenotypic variance
may
depend
on the age
struc-
ture
of the
population sampled.
If a
cohort
of
indi-
viduals
is
measured repeatedly over time,
and if all
individuals
change their phenotype
in the
same
way
at the
same

rate,
then
the
phenotypic
variance
will
not
change over time
(figure
1.5A).
On the
other hand,
if
individuals
differ
in
their ontoge-
netic
trajectories,
the
phenotypic variance exhib-
ited
by the
cohort
may
either increase
or
decrease
(figures
1.5B-D:

phenotype versus time, with lines
converging
or
diverging over time).
In
this
case,
the
magnitude
of
phenotypic variation detected
in a
population will depend
on
both
the
pattern
of on-
togenetic
change exhibited
by its
members
and the
ages
of the
individuals included
in the
sample.
Where
different

genotypes exhibit
different
pat-
terns
of
ontogenetic variation,
the
amount
of
inter-
genotypic variation
may
also vary over time (fig-
ures
1.5B-D).
An
example
of
this phenomenon
is
observed
in
floral
traits among successively produced flowers
of
a
short-lived,
self-fertilizing,
annual species
in