University of Southern California Dissertations and Theses

Multiple generations of hybridization between populations of the intertidal copepod Tigriopus californicus

(USC Thesis Other)

Multiple generations of hybridization between populations of the intertidal copepod Tigriopus californicus

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

MULTIPLE GENERATIONS OF HYBRIDIZATION BETWEEN POPULATIONS OF
THE INTERTIDAL COPEPOD TIGRIOPUS CALIFORNICUS

by

AnnMarie S. Hwang

A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(BIOLOGY)

August 2009

Copyright 2009 AnnMarie S. Hwang

ii
Dedication

This work is dedicated to Joe, whose loving support and
patience helped bring this to completion.

iii
Acknowledgements

I owe many thanks to the mentorship of Suzanne Edmands, my committee chair
and graduate advisor. I would also like to thank lab members Lisa Handschumacher,
Scott Harrison, Sara Northrup, Dennis Peterson, Victoria Pritchard, Catherine Purcell,
Colin Rose, Daniel Rundle, Augustus Vogel, and Tina Weier. Many undergraduate
students provided culture maintenance and assistance with molecular techniques, and I
am particularly indebted to the efforts of Jamie Alexander, Mark Chinen, Asia De la
Torre, Alexandra Hubbell, Tigran Karamanukyan, Yuri Kim, Melisa Lee, Katherine Liu,
Phoebe Luong, Alcea Myrie, Shanassa Roen-Padilla, Zahira Salahuddin, Narine Sargsyan
and Helen Truong. The rest of the dissertation committee, Cheng-Ming Chuong, Andrew
Gracey, Dennis Hedgecock, and Magnus Nordborg, all provided suggestions through
multiple stages of completion.
I was supported by several teaching assistantships from the USC department of
biology, for which I am grateful for the support of Linda Bazilian, Bill Trusten and
Adolfo de la Rosa. Office support to present results of this work was frequently provided
by Don Bingham, Glen Smith and Keun Song. Additional financial support came from
USC Women in Science and Engineering, the Rose Hills Fellowship program through the
Wrigley Institute for Environmental Studies, the USC College Final Year Dissertation
Fellowship, and research assistantships from Suzanne Edmands. Funding for research
included a Doctoral Dissertation Improvement Grant from the National Science
Foundation.

iv
Table of Contents

Dedication ii

Acknowledgements iii

List of Tables vi

List of Figures xi

Abstract xvii

Introduction 1
INTRODUCTION REFERENCES 7

Chapter 1: Long-term experimental hybrid swarms between 12
moderately divergent Tigriopus californicus
populations: Hybrid inferiority in early generations
yields to hybrid superiority in later generations
CHAPTER 1 REFERENCES 54

Chapter 2: Long-term experimental hybrid swarms between nearly 59
incompatible Tigriopus californicus populations:
Fitness recovery and assimilation by the superior
population
CHAPTER 2 REFERENCES 95

Chapter 3: Environmental stress speeds recovery from outbreeding 99
depression in experimental hybrid swarm generations
of moderately divergent Tigriopus californicus
populations
CHAPTER 3 REFERENCES 157

Chapter 4: Stronger patterns of molecular repeatability observed for 162
hybrid swarm replicates under salinity stress
CHAPTER 4 REFERENCES 232

Conclusion 236
CONCLUSION REFERENCES 241

Bibliography 243

v
Appendix: Maladapted gene complexes within populations of the 254
intertidal copepod Tigriopus californicus?
APPENDIX REFERENCES 278

vi
List of Tables

Table 1-1: Mean proportional deviation from midparent for survival and 23
eight male morphometric characters in three-generation
controlled cross. Values are averages among replicates,
relative to midparent in that same generation, with standard
errors in parentheses. Means significantly greater than
midparent values (α= 0.05) according to planned linear
contrasts are in bold. Means significantly less than
midparent values are indicated in bold italics.

Table 1-2: Proportional deviation from midparent survivorship values for 28
each hybrid swarm replicate. Means significantly different
from midparent values (α = 0.05) according to independent
linear contrasts are in bold. Standard errors are in
parentheses, with a few exceptional cases where only one
individual from the replicate was measured.

Table 1-3: Mean phenotypic values for female morphometric characters 30
for all control and hybrid swarm treatments. Units are in
millimeters. Treatment means significantly different from
the midparent according to independent linear contrasts are
indicated in bold (α = 0.05). Values significantly greater
than the midparent are marked with a + sign while those
significantly less than the midparent are marked by a - sign.
Standard errors are in parentheses with the exception of
months 12 and 15 where only one replicate remained for
those treatments.

Table 1-4: Mean phenotypic values for male morphometric characters in 31
hybrid swarm replicates. Units are millimeters. Treatment
means significantly different from the midparent according
to independent linear contrasts are indicated in bold (α =
0.05). Values significantly greater than the midparent are
marked with a + sign while those significantly less than the
midparent are marked by a - sign. Standard errors are in
parentheses.

Table 1-5: Results of multivariate Wilks’ test. Significant tests are 32
indicated in bold.

vii

Table 1-6: Estimated population allele frequencies in (a) one 50:50 39
replicate after 12 months of free mating and (b) one 80:20
replicate after 30 months. Bold numbers indicate the
direction of deviation from additivity. Asterisks indicate a
statistically significant deviation (Chi-squared test, p <
0.05).

Table 1-7: Pairwise tests of linkage disequilibrium for 80:20 replicate. 43
Significant deviations are shown in bold. Asterisks next to
locus numbers indicate pairs of loci that are physically
linked. After bonferroni correction, no locus pairs show
significant LD. *p < 0.05, **p < 0.01, ***p < 0.001

Table 2-1: Survivorship comparisons among treatments by nested 72
ANOVA and Bonferroni post-hoc at each time point. For
each month, the difference between treatment means is
shown. Significant p-values (p < 0.05) are indicated in
bold. Dashes indicate comparisons that were not done
because one or both treatments died out.

Table 2-2: Mean phenotypic values for female morphometric characters 75
for all control and hybrid swarm treatments. Units are in
millimeters. Treatment means significantly different from
PM according to nested ANOVA and Bonferroni post-hoc
tests are indicated in bold (α = 0.05). Values significantly
greater than PM are marked with a + sign while those
significantly less than PM are marked by a - sign. Standard
errors among replicates are in parentheses with the
exception of the PA treatment at month 9 where only one
replicate remained.

Table 2-3: Mean phenotypic values for male morphometric characters in 77
hybrid swarm replicates. Units are millimeters. Treatment
means significantly different from the superior parent (PM)
according to nested ANOVA and Bonferroni post-hoc tests
are indicated in bold (α = 0.05). Values significantly
greater than PM are marked with a + sign while those
significantly less than PM are marked by a - sign. Standard
errors among replicates are in parentheses with the
exception of the PA treatment at months 3 and 9 where
only one replicate contained mature males.

Table 2-4: Results of multivariate Wilks’ test. Significant tests are 78
indicated in bold.

viii

Table 2-5: Microsatellite Allele Frequencies obtained at months 18 and 83
30.

Table 3-1: Proportional deviation from midparent survivorship values for 114
each hybrid swarm replicate. Means significantly different
from midparent values (α = 0.05) according to planned
linear contrasts are in bold. Standard errors are in
parentheses.

Table 3-2: Proportional deviation from midparent metamorphosis values 126
for each hybrid swarm replicate. Means significantly
different from midparent values (α = 0.05) according to
planned linear contrasts are in bold. Standard errors are in
parentheses.

Table 3-3: Mean phenotypic values for female morphometric characters 130
for all control and hybrid swarm treatments. Units are in
millimeters. Treatment means significantly different from
the midparent value according to nested ANOVA and
planned contrast tests are indicated in bold. The direction
of deviation from the additive expectation is indicated by a
+ or – sign. Means that are boxed are transgressive in that
they are significantly different from both parents
(Bonferroni post hoc test) and not intermediate to them.
Standard errors among replicates are in parentheses.

Table 3-4: Mean phenotypic values for male morphometric characters for 132
all control and hybrid swarm treatments. Units are in
millimeters. Treatment means significantly different from
the midparent value according to nested ANOVA and
planned contrast tests are indicated in bold. The direction
of deviation from the additive expectation is indicated by a
+ or – sign. Means that are boxed are transgressive in that,
given they are significantly different from both parents
(Bonferroni post hoc test) and not intermediate. Standard
errors among replicates are in parentheses.

Table 3-5: Difference in mean morpholometrics between generations 1 134
and 13 for females (a) and males (b). Units are in
millimeters. Positive values indicate an increase in size
between generations. Significant differences are indicated
in bold (two-tailed t-test, p<0.05).

ix

Table 4-1: F
IS
values calculated for each replicate using 11 microsatellite 180
loci with significant values noted in bold (*) (p < 0.05)
Dashes indicates values that could not be estimated.
Results shown for generation seven (a) and generation
thirteen (b).

Table 4-2: F
IS
values calculated for females (a) and males (b) for each 181
replicate using 11 microsatellite loci with significant values
noted in bold (*) (p < 0.05). Results shown for generations
seven and thirteen.

Table 4-3: Estimated effective population size (Ne) obtained using the 184
standard variance in allele frequency change (F) according
to the temporal method of Nb_HetEx.

Table 4-4: Weir and Cockerham’s pair-wise F
ST
for pairs of replicate 185
hybrid beakers. Replicates are labeled by salinity treatment
followed by replicate number. All values are significant at
p < 0.0001. Tables show generation seven (a), and
generation thirteen (b).

Table 4-5: Analysis of molecular variance (AMOVA) of replicate groups 190
using 11 microsatellite loci, computed by the distance
matrix in Arlequin (10,000 permutations). Groups for
generation 7 (a) are 35 ppt replicates vs. 53 ppt replicates.
Three different group combinations were run for generation
13 (b) to assess the variation between specific replicate
beakers.

Table 4-6: Locus-by-locus AMOVA calculated for 11 microsatellite loci 191
for (a) generation 7 and (b) generation thirteen.

Table 4-7: Mean deviation from expected morphological and molecular 202
hybrid index scores for generations seven (a) and thirteen
(b).

Table 4-8: Genotypic linkage disequilibrium for each locus pair and 204
associated p-values for replicates of generation seven (a)
and generation thirteen (b). Significant p-values are
indicated in bold, italicized font. P-values significant after a
Bonferroni correction are indicated by a (*).

Table 4-9: F
IS
values for P5CR and P5CS genes for each replicate with 212
significant values in bold (p < 0.05). Results shown for
generation seven (a) and generation thirteen (b).

x

Table 4-10: Deviation from expected RP allele frequencies for P5CR 213
(a) and P5CRS (b) genes for each replicate.

Table 4-11: COI and Cytochrome C genotypic frequencies. 215

Table A-1: Marker information: locus, code (chromosome_abbreviated 261
name), marker type (C, codominant; D, dominant), forward
and reverse primers and annealing temperature (T
a
,
Celsius).

Table A-2: Single-locus genotype data for F2 hybrid males in two 266
reciprocal crosses (LB cytoplasm and RP cytoplasm). Loci
are listed by chromosomes number_locus number. Only LB
homozygote frequencies are available for dominant locus
1555B as RP homozygotes are indistinguishable from LB-
RP heterozygotes. χ
2
tests compared genotype numbers to
expectations within crosses (3:1 for dominant locus 1555B
and 1:2:1 for the remaining codominant loci) and between
reciprocal crosses. Loci with significant differences
between reciprocals are shown in bold. For all loci, cases
where the LB homozygote is favored on the foreign
cytoplasmic background are marked by superscript a. For
codominant loci, cases where the homozygote class with
higher viability does not match the cytoplasmic background
are marked by superscript b. Mean sample size = 140.8
individuals per locus per cross.

xi
List of Figures

Figure 1-1: Survivorship in three-generation controlled cross. Values are 24
averages among replicates, relative to midparent assayed in
the same generation,
+
1 standard error. Dashed line
indicates the additive expectation for each cohort.

Figure 1-2: Census counts for hybrid swarm replicates over 15 months 26
of free mating. Replicates 1 thru 5 are beakers with the
initial ratio of 50% RP: 50% SD. 6 thru 10 were initiated
with a ratio of 80% RP: 20% SD. Large bold numbers
along x-axis indicate month of sampling. Dashed lines
indicate midparent values (average of the parental means)
determined using means of replicate beakers for specific
months.

Figure 1-3: Mean proportional deviation of each hybrid swarm replicate 27
from the midparent value over 15 months of free mating.

Figure 1-4: Principle components analysis of 6 morphometric 35
measurements for females after 3 months (a) and 15
months (b). At 3 months, PC1 and PC2 account for 40.9%
and 21.9% of the variance respectively. At month 15, the
first two components account for 48.5% and 21.7% of the
variance. PCA of 8 morphometric measurements for males
after 3 months (c) and 15 months (d). At 3 months PC1
and PC2 account for 25% and 13.8% of the variance
respectively. At month 15 the first two components account
for 29.5% and 15.1% of the variance.

Figure 1-5: Frequencies of hybrid indices for genotyped replicates. 41
(a) 50:50, 12 months. (b) 80:20 30 months. Vertical dashed
lines indicate expected mean hybrid index: 0.5 for (a) and
0.2 for (b).

Figure 2-1: Proportional deviation from midparent values for clutch 69
size(a) and survivorship(b) in two-generation controlled
crosses. Values are averages among individual clutches
relative to the midparent assayed in the same generation,
+

1 standard error. The dashed line indicates the additive
expectation for each cohort.

xii

Figure 2-2: Mean census counts among all five replicates for each 70
treatment over 21 months of free mating. One replicate
from PA population remained at month nine but died out
before month twelve, while 80PA:20PM replicates did not
survive past month six.

Figure 2-3: Mean proportional deviation of each hybrid swarm replicate 73
from the superior parent value over 21 months of free
mating.

Figure 2-4: Principle components analysis of 6 morphometric 79
measurements for PM and 50:50 females after 3 months (a)
and 21 months (b). At 3 months PC1 and PC2 account for
45.9% and 22.5% of the variance respectively. At month 21
the first two components account for 49.0% and 24.3% of
the variance. PCA of 8 morphometric measurements for
PM and 50:50 males after 3 months (c) and 21 months (d).
At 3 months PC1 and PC2 account for 33.0% and 14.8% of
the variance respectively. At month 21 the first two
components account for 29.9% and 17.6% of the variance.

Figure 3-1: Pilot test of growth rate measured as the length of two-week 107
old juveniles reared in varying salinity. All individuals
assayed were from the Royal Palms population. Mean
length of individuals reared in 53ppt seawater is
significantly different from the control treatment of 35 ppt
(ANOVA with Bonferroni post hoc test, p < 0.001).

Figure 3-2: Mean proportional deviation from midparent for hybrid 113
replicate survivorship over thirteen discrete generations.
(a) 35 ppt treatment (b) 53 ppt treatment. Asterisks
indicate means significantly different from the midparent
according to planned linear contrasts (p < 0.05).

Figure 3-3: Index of potency (h
p
= Q/L) for survivorship over thirteen 116
generations. (a) 35 ppt treatment (b) 53 ppt treatment.
Asterisks indicate means significantly different from the
most extreme parent according to planned contrasts (*p <
0.05, **p<0.01, ***p<0.001).

Figure 3-4: Mean proportional deviation from midparent for 118
survivorship of individual 50:50 replicates over thirteen
discrete generations. (a) 35 ppt treatment (b) 53 ppt
treatment. Bars display one standard error.

xiii

Figure 3-5: Mean proportional deviation from midparent for hybrid 119
replicate metamorphosis over thirteen discrete generations.
(a) 35 ppt treatment (b) 53 ppt treatment. Asterisks
indicate significant deviation from the midparent according
to planned linear contrasts (p<0.05).

Figure 3-6: Index of potency (h
p
= Q/L) for metamorphosis over thirteen 121
generations. (a) 35 ppt treatment (b) 53 ppt treatment.
Asterisks indicate means significantly different from the
most extreme parent according to planned contrasts (*p <
0.05, **p<0.01, ***p<0.001).

Figure 3-7: Nauplii present at day 7 for all treatments at each generation. 123

Figure 3-8: Number of clutches with nauplii still present seven days 124
after hatching, shown as the mean hybrid proportional
deviation from the midparent. (a) 35 ppt (b) 53 ppt.

Figure 3-9: Proportion of clutches with nauplli present at day seven. 125
Shown as the difference of treatment means. Negative
values indicate that the 53 ppt treatment had a higher
proportion of nauplii present than the 35 ppt treatment.

Figure 3-10: Proportional deviations of female 50:50 treatment means 127
from midparent values over 13 generations of free mating.
Y-axis: proportional deviation from midparent value, (Obs-
Exp)/Exp. X-axis: generation.

Figure 3-11: Proportional deviations of male 50:50 treatment means from 136
midparent values over 13 generations of free mating. Y-
axis: proportional deviation from midparent value, (Obs-
Exp)/Exp. X-axis: generation.

Figure 3-12: Mean morphological hybrid index among replicates over 140
thirteen generations. A score of 0 indicates RP-like
morphology whereas a score of 1 indicates SD-like
morphology. Bars indicate one standard error. Dashed line
indicates additive expectation of 0.5. Female hybrid index
scores are shown in 9a and differences between treatments
are only significant at generation seven ((p<0.05, two-tailed
t-test ). Male scores are shown in 9b and treatment means
are significantly different at generations 3, 7, 9, 11 and 13.

Figure 3-13: Three-week competitive fitness observed 13 generations 142
after hybrid swarm formation. (a) 35 ppt; (b) 53 ppt.

xiv

Figure 3-14: Generation 14 heat shock assay. 143

Figure 3-15: Generation 14 heat shock survivorship. Contributions of 144
nauplii and copepods.

Figure 4-1: RP allele frequencies for eleven microsatellite loci for each 176
replicate. Horizontal dashed line shows the expected
frequency of 0.5. (a) Generation 7, 35 ppt; (b) generation
7, 53 ppt; (c) generation 13, 35 ppt; (d) generation 13, 53
ppt.

Figure 4-2: Factorial correspondence analysis for generations 7(A) and 187
13(B) for 11 microsatellite loci. Each data point represents
an individual. Replicates are indicated by different colors.

Figure 4-3: Inferred population structure for generation 13 from 193
simulated runs in STRUCTURE using the no admixture
model and K=2 to7. Each color represents one population
out of K total. Each vertical line represents an individual
divided up into the probability that it comes from each of K
populations. Numbers 1, 2, and 3 correspond to 35 ppt
replicates 1, 4 and 5 respectively. Numbers 4, 5, 6 and 7
correspond to 53 ppt replicates 1, 2, 4 and 5 respectively.

Figure 4-4: F
ST
values estimated from 11 microsatellite loci against 195
heterozygosity plotted for generation seven 35 ppt
replicates (a); generation thirteen 35 ppt replicates (b);
generation seven 53 ppt replicates (c) and generation
thirteen 53 ppt replicates(d). Each point represents one
locus, with locus 1202 circled. Lines indicate 95%
quantiles of expected Fst given no selection.

Figure 4-5: Molecular hybrid index distributions for individual replicates 197
calculated with 11 microsatellite loci. A score of 0
indicates all RP alleles and a score of 1 indicates all SD
alleles. Each histogram represents one replicate beaker for
generations 7, 35 ppt (a), 53 ppt (b) and 13 35 ppt (c) and
53 ppt (d). The x-axis represents hybrid index score and
the y-axis is number of individuals.

xv

Figure 4-6: Mean and standard error for proportional deviation from 208
expected microsatellite genotype frequencies for each
replicate for four two-locus classes. Only physically,
unlinked loci are included for a total of 55 two-locus
combinations tested within each replicate. (a) Generation
seven, 35 ppt; (b) generation seven , 53 ppt; (c) generation
thirteen, 35 ppt and (d) generation thirteen, 53 ppt.
Asterisks indicate the significance of paired, two-tailed t-
tests of observed vs. expected genotype numbers (*p<
0.05). N = 74 two-locus combinations for each cross.

Figure 4-7: Proline biosynthesis genes, P5CR and P5CS. Mean and 214
standard error for proportional deviation from expected
two-locus genotype frequencies for four classes of hybrids
across replicate beakers at 35 ppt salinity treatment (a) and
53 ppt salinity (b).

Figure 4-8: Proportional deviation from expected frequencies of 217
cytonuclear genotypes at generation seven for 35 ppt (a)
and 53 ppt (b). No significant differences from expected
numbers were observed.

Figure 4-9: Mean and standard error for proportional deviation from 218
expected COI and microsatellite genotypic frequencies at
generation 7. Matched individuals have COI and
homozygous microsatellite genotypes from the same
population. Unmatched individuals are homozygous for
microsatellite alleles that do not match the COI genotype.
Asterisks indicate the significance of one-tailed t-tests of
observed vs. expected genotype numbers (*p< 0.05,
**p<0.01, ***p<0.001).

Figure 4-10: Mean and standard error for proportional deviation from 220
expected COI and P5CS genotype frequencies at generation
seven. Matched individuals have COI and homozygous
P5CS genotypes from the same population. Unmatched
individuals are homozygous for P5CS alleles that do not
match the COI genotype. Asterisks indicate the significance
of one-tailed t-tests of observed vs. expected genotype
numbers (*p< 0.05).

xvi

Figure A-1: Mean minimum development time (
+
1SE) in reciprocal 265
hybrids, with significance tested by unpaired, 2-tailed t-
tests (***P < 0.01). A) F2 males RPf x LBm (N = 335) vs.
LBf x RPm (N = 178). B) Backcross males (RPf x
LBm)F1f x LBm (N = 154) vs. (LBf x RPm)F1f x LBm (N
= 125). C) Backcross females (RPf x LBm)F1f x LBm (N =
166) vs. (LBf x RPm)F1f x LBm (N = 22).

Figure A-2: Mean and standard error for proportional deviation from 269
expected two-locus genotype frequencies ((Obs-Exp)/Exp)
for four classes of F2 hybrids in each of two reciprocal
crosses (LB cytoplasm and RP cytoplasm). Only
physically, unlinked loci are included. Asterisks denote the
significance of paired, one-tailed t-tests of observed vs.
expected genotype numbers (*P < 0.05). N = 74 two-locus
combinations for each cross.

xvii
Abstract

For the intertidal copepod Tigriopus californicus, outbreeding depression for a
variety of fitness measures is typically observed in early-generation interpopulation
hybrids. This dissertation is an experimental approach to look at morphological, fitness
and molecular outcomes of mixed populations following multiple generations of mating.
Each chapter expands upon our understanding of the long-term, multi-generational
outcomes of hybrid swarms that were initiated with populations showing different
degrees of incompatibility.
Chapters 1 and 2 examined early-generation controlled crosses and long-term
hybrid swarms. In Chapter 1, crosses between Royal Palms and San Diego, California,
showed that only F2 cohorts exhibited significant declines in fitness compared to
midparent values. Chapter 2 utilized more divergent populations from Playa Altamira
and Punta Morro in Baja California, Mexico. F1 and F2 hybrids showed large declines in
survivorship, while backcrosses produced no offspring. Long-term hybrid swarms in
Chapter 1 exhibited early fitness declines followed by rapid recovery and surpassing of
midparent fitness. Microsatellites revealed extensive introgression. In contrast, highly
divergent populations in Chapter 2 showed fitness recovery that was likely due to genetic
swamping by the superior parent.
Chapters 3 and 4 investigated the same populations as Chapter 1 but included two
environmental treatments: benign conditions versus salinity stress. Cultures were initiated
with equal numbers from each source population and allowed to mate freely while
generations were kept discrete. For both survivorship and metamorphosis, early

xviii
generation heterosis was followed by outbreeding depression and recovery, which
occurred up to two generations earlier in the high salinity treatment. Microsatellites and
male morphology both indicated that swarms became more RP-like over time. High
salinity replicates displayed stronger repeatability for both molecular and fitness
character. High salinity hybrids were more fit than benign replicates when exposed to a
novel stress. Evidence for nuclear-nuclear and nuclear-cytoplasmic coadaptation was
observed at generation seven.
This work supports the hypothesis that hybrid breakdown in early generations
may be a temporary phenomenon followed by the persistence of highly fit genotypes.
While selection may have a deterministic role in hybrid swarm evolution, it is likely to be
hampered by drift resulting from environmental fluctuations that produce frequent
bottleneck events.

1
Introduction

Conservation and evolutionary biologists alike have many reasons to be
concerned about the long-term outcome of hybrid populations. Human action has
resulted in both the accidental (Bowman et al. 2007, Randi 2008) and intentional
relocation of populations and species (Allendorf et al. 2001, Tallmon et al. 2004, Gow et
al. 2006, Taylor et al. 2006, Edmands 2007), resulting in secondary contact between
previously isolated lineages. Biodiversity can be lost due to extinction by introgressive
hybridization (Seehausen et al. 2008). An understanding of the long-term outcome of
hybrid swarm formation can enhance the success of efforts to conserve and manage
threatened species. Perhaps one of the most widely discussed, and yet unresolved
questions, regarding the mixing of gene pools, is whether outbreeding depression from
farmed-wild matings is a threat to wild fish populations (McGinnity et al. 2003,
McClelland and Naish 2007, Roberge et al. 2007, Tymchuk et al. 2007, Fraser et al.
2008, Hutchings and Faser 2008). Experimental approaches to understanding the long-
term outcomes of hybridization in species of concern are difficult due to long generation
times and requirements for captivity.
Tigriopus californicus is an intertidal copepod that has been used extensively as a
model to study outbreeding depression (e.g. Burton 1986, Edmands 1999, Ellison &
Burton 2008). T. californicus is an excellent model for studies of experimental evolution
because it has a short generation time (~23 days, Burton 1987), cultures are easily
maintained in the laboratory, and large sample sizes may be collected from the wild. Its
native habitat consists of rocky intertidal outcrops extending from Alaska down to central

2
Baja California, Mexico. Populations exhibit a wide range of genetic divergence, with
mitochondrial DNA differences that range from 0.2 to 23% (Burton and Lee 1994,
Edmands 2001). Because T. californicus inhabits the highest supralittoral pools, known
as the splash zone, populations undergo wide fluctuations in environmental conditions,
particularly temperature and salinity. The ranges of these conditions may vary from
population to population, but a short term study of temperature and salinity stress found
very little evidence for local adaptation (Edmands and Deimler, 2004).
It has been well established that crosses between T. californicus populations
exhibit hybrid breakdown in the second generation (eg. Burton 1987, Burton et al. 1999,
Edmands and Burton 1999, Edmands 1999). However, recent studies have also revealed
that beneficial hybrid recombinants may be generated from those same early generation
crosses (Willett 2006, Edmands 2008, Edmands et al. in press). This work was an effort
to investigate the processes that take place in hybrid populations throughout many
generations after the initial hybridization event. By observing transitory stages of hybrid
swarm formation, a major goal of this thesis is to reach a better understanding of the
processes and forces that influence the fate of hybrid swarms.
Because T. californicus populations span a wide range of divergence (Edmands
1999, 2001), it is possible to observe the outcomes of hybridization with populations of
varying levels of incompatibility. This was done in Chapters 2 and 3, where hybrid
swarms were initiated using two different pairs of populations: one pair was moderately
incompatible, producing relatively fit F1 and backcross cohorts, while the other pair was
almost completely incompatible, showing large fitness reductions in F1 and F2 cohorts
and no offspring in backcross cohorts. Chapter 1 is the first study in Tigriopus to collect

3
extensive experimental evidence that hybrid swarms can recover from the effects of
severe early-generation outbreeding depression. Chapter 2 demonstrates that, in the case
of highly divergent, nearly incompatible populations, the long-term outcome of a hybrid
swarm is likely to be swamping by the parental population with superior fitness in the
experimental environment.
While extensive introgression may take place to homogenize hybrid swarms,
genetic swamping may occur, where genes of one species increase in frequency from the
initial hybridization event until the genetic integrity of the second species is compromised
(Childs et al. 1996, Bleeker et al. 2007, Kothera et al. 2007). This has been recorded as a
particular threat when a rare species of concern comes into contact with a more common
congener (Ellstrand and Elam 1993, Childs et al. 1996, Rhymer and Simberloff 1996,
Levin et al. 1996). Chapters 2 and 3 address the question of whether the proportional
contribution of each parental population to the hybrid swarm influences the outcome in
the long run. Both chapters included one set of trials where swarm formation was
initiated with equal numbers from each parental population, while a second set of trials
contained one common population (80%) and one rare population (20%).
Another possible outcome is that hybridization might enhance genetic diversity by
introducing new combinations of genes (Grant and Grant 1992, Dowling and Secor 1997,
Arnold 2006). In the wild, hybrid species often occur in habitats more extreme than
those of congeners (Rieseberg et al. 2003, Gross et al. 2004) and studies focusing on
plants have indicated that hybrid speciation may be especially likely under stressful or
novel conditions (Johnston et al. 2004, Rieseberg et al. 2007). A commonly discussed
hypothesis for this is that hybridization may generate recombinants with extreme

4
phenotypes (in a positive or negative direction) compared to parentals, a phenomenon
known as transgressive segregation (Rieseberg et al. 1999, Seehausen 2004).
Transgressive segregation, while reported more commonly in plants, (Lexar et al. 2003,
Johnston et al. 2004, Johansen-Morris and Latta 2006) has been observed for a variety of
taxa including arthropods and fish (Rieseberg 1999, Ranganath and Aruna 2003,
Albertson and Kocher 2005). Chapter 1 offers initial evidence of the presence of
transgressive segregation in T. californicus populations, as hybrid replicates show fitness
that exceeds the superior parent after fifteen months. However, it is Chapter 3 where this
question is largely tested with experimental trials set up under two different
environments: one that is benign and one that imposes salinity stress.
In considering both management strategies and theories of hybrid speciation,
there is a large interest in the importance of contingency versus repeatability (Rieseberg
2000, Simoes et al. 2008). This work provides an animal alternative to the known
examples of plants that show repeated hybridization events leading to similar outcomes
(Rieseberg et al. 1996, Brochman et al. 2000, Schwarzbach and Rieseberg 2002). These
examples imply that strong deterministic forces are involved in driving the consequences
of hybridization. Studies of Tigriopus have previously found only partial concordance
among replicates of the same treatment for molecular measures (Edmands et al. 2005).
This question is touched upon in Chapters 2 and 3, but more thoroughly addressed in
Chapters 4 and 5, with replicate trials in different environmental regimes, and molecular
data for multiple replicates at different generations of hybridization.
With the effects of global climate change and anthropogenic activity on natural
habitats, environmental stress is an important concern for both conservation and

5
evolution of populations (Bijlsma and Loeschcke 2005, Frankham 2005). While many
studies find that inbreeding depression is aggravated by stress (Lynch and Walsh 1998)
other findings suggest outbreeding depression may be alleviated by stress (Hoffman and
Parsons 1991, Armbruster et al, 1997, Edmands and Deimler 2004). Chapters 4 and 5
address this question by comparing experimental swarms of two different environments:
normal seawater versus high salinity. Another question that received attention concerned
the extent to which selection for resistance to one stress resulted in resistance to other
forms of stress. Studies using Drosophila have repeatedly shown this type of correlation
(Hoffman and Parsons 1989, Hoffman and Parsons 1991, Bubliy and Loeschcke 2005).
Chapter 3 takes an exploratory look at this in Tigriopus by exposing multi-generation
hybrids to a novel, thermal stress.
Finally, this work examined the extent to which evidence of coadapted gene
complexes can be detected after multi-generation hybridization. T. californicus has been
studied extensively to further our understanding of cytonuclear coevolution in early
generation hybrids. Earlier work focused on components of the mitochondrial electron
transport system with studies showing that enzyme-substrate activity is higher when
mitochondrial and nuclear encoded subunits are from the matching populations (Edmands
and Burton 1999, Rawson and Burton 2002; Harrison and Burton 2006). Recent work has
demonstrated even stronger evidence for cytonuclear coevolution. Ellison and Burton
(2008) also showed that fitness reductions in the F3 are completely restored in maternal
backcrosses where parental nuclear and mitochondrial combinations are reformed. A
different study (Ellison and Burton 2006) confirmed that only electron transport enzymes
with both nuclear and mitochondrial subunits display reduced fitness in hybrids, and

6
hybrids with mtDNA and nuclear-encoded mtRNA polymerase from the same population
have higher fitness.
Two studies have looked at nuclear-nuclear interactions in F2 hybrid generations.
Edmands et al. (in press) found evidence for nuclear-nuclear coadaptation, seen by an
excess of parental double homozygotes and a deficit of non-parental double
homozygotes. Willett (2006) showed a different pattern of interactions for three nuclear,
protein coding genes of the electron transport system: significant excesses of double
parental homozygotes, double heterozygotes and double nonparental homozygotes
suggesting not only simple nuclear-nuclear coadaptation but also complex dominance or
epistatic relationships.
Despite the compelling evidence for coadapted complexes in T. californicus,
recent studies have reported nuclear alleles that are favored on the wrong cytoplasmic
background (Willett and Burton 2003, Willett 2006, Edmands et al. in press). It is
difficult to predict the extent to which long-term hybridization would result in selection
for coadapted complexes that have been tested out in previous environments, or for
highly fit recombinant gene combinations being exposed to selection for the first time.
Questions concerning the evidence for nuclear-nuclear coadaptation after multiple
generations are addressed briefly in Chapter 1 and both cytonuclear and nuclear-nuclear
coadaptation are discussed more extensively in Chapter 4.

7
INTRODUCTION REFERENCES

Albertson RC, Kocher TD (2005) Genetic architecture sets limits on transgressive
segregation in hybrid cichlid fishes. Evolution 59, 686-690.

Allendorf FW, Leary RF, Sprunell P, Wenburg JK (2001) The problems with hybrids:
setting conservation guidelines. Trends in Ecology and Evolution, 613-622.

Armbruster P, Bradshaw WE, Holzapfel CM (1997) Evolution of the genetic architecture
underlying fitness in the pitcher plant mosquito, Wyeomyia smithii. Evolution 51,
451-458.

Arnold M (2006) Evolution Through Genetic Exchange Oxford University Press, Oxford,
UK.

Bijlsma R, Loeschcke V (2005) Environmental stress, adaptation and evolution: an
overview. Journal of Evolutionary Biology 18, 744-749.

Bleeker W, Schmitz U, Ristow M (2007) Interspecific hybridization between alien and
native plant species in Germany and its consequences for native biodiversity.
Biological Conservation 137, 248-253.

Bowman J, Kidd AG, Gorman RM, Schulte-Hostedde AI (2007) Assessing the potential
for impacts by feral mink on wild mink in Canada. Biological Conservation 139,
12-18.

Brochmann C, Borgen L, Stabbetorp OE (2000) Multiple diploid hybrid speciation of the
Canary Island endemic Argyranthemum sundingii (Asteraceae). Plant Systematics
and Evolution 220, 77-92.

Bubliy OA, Leoschcke V (2005) Correlated responses to selection for stress resistance
and longevity in a laboratory population of Drosophila melanogaster. Journal of
Evolutionary Biology 18, 789-803.

Burton RS (1987) Differentiation and integration of the genome in populations of the
marine copepod Tigriopus californicus. Evolution 41, 504-513.

Burton RS, Lee B-N (1994) Nuclear and mitochondrial gene genealogies and allozyme
polymorphism across a major phylogeographic break in the copepod Tigriopus
californicus. Proceedings of the National Academy of Sciences USA 91, 5197-
5201.

Burton RS, Rawson PD, Edmands S (1999) Genetic architecture of physiological
phenotypes: Empirical evidence for coadapted gene complexes. Amer. Zool. 39,
451-462.

8
Childs MR, Echelle AA, Dowling TE (1996) Development of the hybrid swarm between
pecos pupfish (Cyprinodontidae: Cyprinodon pecosensis) and sheepshed minnow
(Cyprinodon variegatus): A perspective from allozymes and mtDNA. Evolution
50, 2014-2022.

Dowling TE, Secor CL (1997) The role of hybridization and introgression in the
diversification of animals. Annual Review of Ecology and Systematics 28, 593-
619.

Edmands S (1999) Heterosis and outbreeding depression in interpopulation crosses
spanning a wide range of divergence. Evolution 53, 1757-1768.

Edmands S (2001) Phylogeography of the intertidal copepod Tigriopus californicus
reveals substantially reduced population differentiation at northern latitudes.
Molecular Ecology 10, 1743-1750.

Edmands S (2007) Between a rock and a hard place: evaluating the relative risks of
inbreeding and outbreeding for conservation and management. Molecular
Ecology 16, 463 - 475.

Edmands S (2008) Recombination in interpopulation hybrids of the copepod Tigriopus
californicus: release of beneficial variation despite hybrid breakdown. Journal of
Heredity 99, 316-318.

Edmands S, Burton RS (1999) Cytochrome c oxidase activity in interpopulation hybrids
of a marine copepod: a test for nuclear-nuclear or nuclear-cytoplasmic
coadaptation. Evolution 53, 1972-1978.

Edmands S, Deimler JK (2004) Local adaptation, intrinsic coadaptation and the effects of
environmental stress on interpopulation hybrids in the copepod Tigriopus
californicus. Journal of Experimental Marine Biology and Ecology 303, 183-196.

Edmands S, Feaman HV, Harrison JS, Timmerman CC (2005) Genetic consequences of
many generations of hybridization between divergent copepod populations.
Journal of Heredity 92, 114-123.

Edmands S, Northrup SL, Hwang AS (in press) Maladapted gene complexes within
populations of the intertidal copepod Tigriopus californicus? Evolution.

Ellison CK, Burton RS (2006) Disruption of mitochondrial function in interpopulation
hybrids of Tigriopus californicus. Evolution 60, 1382-1391.

Ellison CK, Burton RS (2008) Interpopulation hybrid breakdown maps to the
mitochondrial genome. Evolution 62, 631-638.

9
Ellstrand NC, Elam DR (1993) Population genetic consequences of small population size:
Implications for plant conservation. Annual Review of Ecology and Systematics
24, 217-242.

Frankham R (2005) Stress and adaptation in conservation genetics. Journal of
Evolutionary Biology 18, 750-755.

Fraser DJ, Cook AM, Eddington JD, Bentzen P, Hutchings JA (2008) Mixed evidence for
reduced local adaptation in wild salmon resulting from interbreeding with escaped
farmed salmon: complexities in hybrid fitness. Evolutionary Applications 1, 501-
512.

Gow JL, Peichel CL, Taylor EB (2006) Contrasting hybridization rates between
sympatric three-spined sticklebacks highlight the fragility of reproductive barriers
between evolutionarily young species. Molecular Ecology 15, 739-752.

Grant PR, Grant BR (1992) Hybridization of bird species. Science 256, 193-197.

Gross BL, Kane NC, Lexar C, et al. (2004) Reconstructing the origin of Helianthus
deserticola: Survival and selection on the desert floor. The American Naturalist
164, 145-156.

Harrison JS, Burton RS (2006) Tracing hybrid incompatibilities to single amino acid
substitutions. Molecular Biology and Evolution 23, 559-564.

Harrison JS, Edmands S (2006) Chromosomal basis of viability differences in Tigriopus
californicus interpopulation hybrids. Journal of Evolutionary Biology 19, 2040-
2051.

Hoffman AA, Parsons PA (1991) Evolutionary Genetics and Environmental Stress
Oxford University Press, Oxford.

Hoffmann AA, Parsons PA (1989) Selection for increased desiccation resistance in
Drosophila melanogaster: additive genetic control and correlated responses for
other stresses. Genetics 122, 837-845.

Hutchings JA, Fraser DJ (2008) The nature of fisheries-and farming-induced evolution.
Molecular Ecology 17, 294-313.

Johansen-Morris A, Latta R (2006) Fitness consequences of hybridization between
ecotypes of Avena barbata: hybrid breakdown, hybrid vigor and transgressive
segregation. Evolution 60, 1585-1595.

Johnston JA, Donovan LA, Arnold ML (2004) Novel phenotypes among early generation
hybrids of two Louisiana iris species: flooding experiments. Journal of Ecology
92, 967-976.

10

Kothera L, Ward SM, Carney SE (2007) Assessing the threat from hybridization to the
rare endemic Physaria bellii Mullican (Brassicaceae). Biological Conservation
140, 110-118.

Levin DA, Francisco-Ortega J, Jansen RK (1996) Hybridization and the extinction of rare
plant species. Conservation Biology 10, 10-16.

Lexar C, Welch ME, Raymond O, Rieseberg LH (2003) The origin of ecological
divergence in Helianthus paradoxus (Asteraceae): selection on transgressive
characters in a novel hybrid habitat. Evolution 57, 1989-2000.

Lynch M, Walsh B (1998) Genetics and Analysis of Quantitative Traits Sinauer,
Sunderland, MA.

McClelland EK, Naish KA (2007) What is the fitness outcome of crossing unrelated fish
populations? A meta-analysis and an evaluation of future research directions.
Conservation Genetics 8, 397-416.

McGinnity P, Prodohl P, Ferguson A, et al. (2003) Fitness reduction and potential
extinction of wild populations of Atlantic salmon, Salmo salar, as a result of
interactions with escaped salmon. Proceedings of the Royal Society of London B
270, 2443-2450.

Randi E (2008) Detecting hybridization between wild species and their domesticated
relatives. Molecular Ecology 17, 285-293.

Ranganath H, Aruna S (2003) Hybridization, transgressive segregation and evolution of
new genetic systems in Drosophila. Journal of Genetics 82, 163-177.

Rawson PD, Burton RS (2002) Functional coadaptation between cytochrome c and
cytochrome c oxidase within allopatric populations of a marine copepod.
Proceedings of the National Academy of Sciences USA 99, 12955-12958.

Rhymer JM, Simberloff D (1996) Extinction by hybridization and introgression. Annual
Review of Ecology and Systematics 27, 83-109.

Rieseberg LH (2000) Crossing relationships among ancient and experimental sunflower
hybrid lineages. Evolution 54, 859-865.

Rieseberg LH, Archer MA, Wayne RK (1999) Transgressive segregation, adaptation and
speciation. Heredity 83, 363-372.

Rieseberg LH, Kim S-C, Randell RA, et al. (2007) Hybridization and the colonization of
novel habitats by annual sunflowers. Genetica 129, 149-165.

11
Rieseberg LH, Raymond O, Rosenthal DM, et al. (2003) Major ecological transitions in
wild sunflowers facilitated by hybridization. Science 301, 1211-1216.

Rieseberg LH, Sinervo B, Linder CR, Ungerer MC, Arias DM (1996) Role of gene
interactions in hybrid speciation: evidence from ancient and experimental
hybrids. Science 272, 741-745.

Roberge C, Normandeau E, Einum S, Guderley H, Bernatchez L (2008) Genetic
consequences of interbreeding between farmed and wild Atlantic salmon: insights
from the transcriptome. Molecular Ecology 17, 314-324.

Schwarzbach AE, Rieseberg LH (2002) Likely multiple origins of a diploid hybrid
sunflower species. Molecular Ecology 11, 1703-1715.

Seehausen O (2004) Hybridization and adaptive radiation. Trends in Ecology and
Evolution 19, 198-207.

Seehausen O, Takimoto G, Roy D, Jokela J (2008) Speciation reversal and biodiversity
dynamics with hybridization in changing environments. Molecular Ecology 17,
30-44.

Simoes P, Santos J, Fragata I, et al. (2008) How repeatable is adaptive evolution? The
role of geographical origin and founder effects in laboratory adaptation. Evolution
62, 1817-1829.

Tallmon DA, Luikart G, Waples RS (2004) The alluring simplicity and complex reality
of genetic rescue. Trends in Ecology and Evolution 19, 489-496.

Taylor EB, Boughman JW, Groenenboom M, et al. (2006) Speciation in reverse:
morphological and genetic evidence of the collapse of a three-spined stickleback
(Gasterosteus aculeatus) species pair. Molecular Ecology 14, 343-355.

Tymchuk WE, Sundstrom LF, Devlin RH (2007) Growth and survival trade-offs and
outbreeding depression in rainbow trout (Oncorhynchus mykiss). Evolution 61,
1225-1237.

Willett CS (2006) Deleterious epistatic interactions between electron transport system
protein-coding loci in the copepod Tigriopus californicus. Genetics 173, 1465-
1477.

Willett CS, Burton RS (2003) Environmental influences on epistatic interactions:
viabilities of cytochrome c genotypes in interpopulation crosses. Evolution 57,
2286-2292.

12
Chapter 1

Long-term experimental hybrid swarms between moderately incompatible
Tigriopus californicus populations: Hybrid inferiority in early generations yields to
hybrid superiority in later generations

CHAPTER 1 ABSTRACT

Outbreeding depression is the typical result observed in early-generation
interpopulation hybrids of Tigriopus californicus. We examined both controlled crosses
and long-term experimental hybrid populations. Of all reciprocal F1, F2, F3 and
backcross cohorts, only F2 cohorts showed significant declines in fitness compared to
midparent values. For long-term studies, four treatment groups were initiated using two
populations from southern California: 100% Royal Palms (RP), 100% San Diego (SD),
50% RP: 50%SD, and 80%RP: 20%SD. Replicates were surveyed at regular intervals for
morphometric, census and fitness measures. Fitness of hybrid treatments showed declines
in fitness relative to midparent values followed by rapid recovery. At month fifteen, two
out of four replicates significantly exceeded midparent fitness. Hybrid morphological
measures decreased over time relative to parentals. Female morphometric characters
after 15 months showed similar trends among 50:50 replicates, while significant
differences among treatments were detected. Microsatellites for a subset of samples
revealed extensive introgression in hybrid treatments. This adds to previous evidence

13
that hybrid breakdown in early generations may be a temporary phenomenon followed by
the persistence of highly fit recombinant genotypes.

CHAPTER 1 INTRODUCTION

Hybridization, whether natural or anthropogenically induced, raises a variety of
concerns that are relevant to species management. Invasive species may be a threat to
native populations if they hybridize and cause a reduction in the number of individuals
with native genotypes (Bowman et al. 2007, Randi 2008, Ellstrand and Schierenbeck
2006). The act of introducing individuals from a divergent population to rescue
endangered species from inbreeding depression can be an attractive management tool
(Tallmon et al. 2004), but there is much concern over the possibility that intentional
translocation could do more harm than good if outbreeding depression occurs (Allendorf
et al. 2001, Edmands 2007).
There is now a growing body of evidence showing outbreeding depression in
hybrids of commercially important fishes such as salmon and trout (McGinnity et al.
2003, Gilk et al. 2004, Tymchuk et al. 2007). Issues of utmost concern for these species
include impacts on both the stock abundance and evolutionary trajectory that may result
from anthropogenic introductions, as in the event of farmed individuals mating with wild
populations (Hutchings and Faser 2007, McClelland and Naish 2007, Roberge et al.
2007). Finally, anthropogenic impacts may lead to a loss of environmental heterogeneity
which could result in the hybridizing of populations that were previous products of

14
ecological speciation (Gow et al. 2006, Taylor et al. 2006). Biodiversity would be lost
due to extinction by introgressive hybridization (Seehausen et al. 2008). An
understanding of the long-term outcome of hybrid swarm formation can enhance the
success of efforts to conserve and manage threatened species.
Hybrid fitness in advanced generations is difficult to predict because the mixing
of gene pools can simultaneously create both beneficial and deleterious interactions
between loci. Hybridization may result in an increase in fitness as inbreeding depression
is alleviated by the masking of deleterious recessive alleles, or heterosis or hybrid vigor
may occur due to overdominance (Lynch 1991). Alternatively, outbreeding depression
may occur for a variety of reasons. Hybrid fitness problems have largely been discussed
in terms of the Dobzhansky-Muller incompatibility model in which isolated populations
accumulate neutral or advantageous mutations over time. Selection may then contribute
to the formation of interlocus combinations with positive epistatic interactions. When
hybridization occurs, recombination and segregation can break up coadapted gene
complexes resulting in decreased fitness (Orr 1996, Turelli et al. 2001). These new
deleterious combinations can involve both heterozygote-homozygote or homozygote-
homozygote interactions (Lynch 1991, Turelli and Orr 2000). Reduction in fitness of F1
hybrids may occur as a result of the disruption of local adaptation, underdominance,
negative heterozygote-heterozygote interactions or sex chromosome interactions (Lynch
1991, Edmands 2007). Often reductions in fitness are delayed until the F2 generation or
later when the breakup of coadapted gene complexes occurs (e.g. Burton 1987, Burton et
al. 1999, Edmands and Burton 1999).

15
Another possibility is that hybridization might enhance genetic diversity by
introducing new combinations of genes (Grant and Grant 1992, Dowling and Secor 1997,
Arnold 2006). These novel genotypes might correspond to extreme phenotypes that have
superior fitness compared to either parental population (Rieseberg 1999).
The intertidal copepod Tigriopus californicus is an excellent model for laboratory
studies of hybridization because of its short generation time (~23 days) and the minimal
care that cultures require. This species inhabits rocky intertidal outcrops extending from
Alaska down to central Baja California, Mexico. Despite a seemingly high potential for
dispersal, populations are genetically differentiated over short geographic distances, with
mitochondrial DNA differences that range from 0.2 to 23% (Burton and Lee 1994,
Edmands 2001). Interpopulation crosses typically result in enhanced F1 hybrid fitness
compared to parents and in reduced F2 hybrid fitness (e.g. Burton 1986, Burton 1987,
Burton 1990ab, Edmands 1999).
Little empirical data exist on the duration of outbreeding depression and few
studies go beyond the first few generations of hybridization. Some suggest that
outbreeding depression may be temporary, with rapid recovery from fitness declines
(Templeton 1986, Rieseberg et al. 1996, Erickson and Fenster 2006). Yet, for taxa such
as Tigriopus, which shows limited evidence of local adaptation, there are also reasons to
believe that outbreeding depression may be long lasting. For example, computer
simulations show that populations take longer to recover from the disruption of intrinsic
coadaptation than from the disruption of local adaptation (Edmands and Timmerman
2003).

16
In this study we used experimental hybrid swarm populations to assess the
magnitude and duration of outbreeding depression over fifteen months of free mating (a
maximum of about 20 generations). This the first multi-generation hybridization study in
Tigriopus to measure patterns of fitness, morphology and genetic composition, which
were assayed at 3-month intervals.

METHODS

Population Sampling
Populations were sampled from two southern California locations, Royal Palms,
CA (RP, 33° 42’ N, 118° 19’ W) and San Diego, CA (SD, 32° 45’ N, 117° 15’ W) in
June 2004. These two populations show approximately 18% mitochondrial cytochrome
oxidase I divergence (Edmands 2001). Samples were maintained as mass cultures in 400
ml beakers with filtered seawater (37 μm) containing finely ground Spirulina (0.2 mg/ml)
and housed in a 20ºC incubator with a 12 h light: 12 h dark cycle.

Tigriopus Biology
The reproductive biology of Tirgriopus californicus has been well-documented
(Egloff 1967, Vittor 1971). Mating and reproduction occur year round. Adult males use
their antennae to clasp virgin females and guard them until they are sexually mature.
Males have multiple matings while females mate only once and store sperm to fertilize
multiple broods of offspring (Burton 1985). Individual lifespan may be as long as 95 days

17
and females can produce up to 20 clutches of eggs. Clutch size varies from less than 10
nauplii to over 100 (pers obs, Vittor 1971). Minimum generation time is approximately
23 days at 20°C.

Controlled Crosses
Virgin females and sexually mature males were collected using fine needle probes
to tease apart clasped pairs. Reciprocal crosses were made by placing 10 virgin females
with 10 mature males in a large Petri dish. Four replicate dishes (A-D) were set up for
each of 4 types of crosses (all crosses listed as female x male): two controls (RP x RP
and SD x SD) and two reciprocal crosses (RP x SD and SD x RP). Each mating dish
contained approximately 35 ml filtered seawater and ground Spirulina (0.2 mg/ml). A
maximum of 40 pairs were formed for each cross. Mating dishes were observed three
times per week to identify females with eggs. Females that formed an egg sac were
isolated into a small Petri dish containing 15 ml of seawater and Spirulina mixture. Each
of these birthing dishes was monitored three times a week to check for the presence of
nauplii. On the day of hatching, 10 larvae were pipetted into a new small Petri dish.
Survivors were counted after 14 days and mature males were counted after 28 days. If
adult males were present at day 28, one was randomly selected and photographed for
morphometric analysis following procedures in Edmands and Harrison (2003; see
below). If no adult males were present, the dish was set aside and monitored for the
presence of an adult male one week later.
After the first clutch of eggs hatched, the adult female was transferred into a 400
ml pair-forming beaker to produce subsequent clutches. Females from mating dish RP x

18
RP A were combined into pair-forming beaker RP x RP A, females from mating dish RP
x RP B were combined into pair-forming beaker RP x RP B, and so on. Each pair-
forming beaker was monitored three times a week for pairs and, once a week, females
with eggs were removed. Pairs that were removed were split and placed in appropriate
mating dishes to produce the second generation. Four replicates of each of the following
second generation cohorts were established: RP x RP control, SD x SD control, RP x SD
F2, SD x RP F2, Backcross to RP and Backcross to SD. The replicates were designed to
avoid inbreeding (e.g. AxB, BxA, CxD and DxC). For the backcross cohorts, the
replicates were the four possible backcross types (e.g. F1 x RP, RP x F1, reciprocal F1 x
RP, RP x reciprocal F1) with no inbreeding. Two replicates of each of the following third
generation cohorts were established: RP x RP control, SD x SD control, RP x SD F3, and
SD x RP F3. Again, replicates were designed to avoid inbreeding (e.g. AB/BA x CD/DC
and CD/DC x AB/BA). Protocols for fitness and morphometric assays described above
were repeated for the second and third generation cohorts.
To adjust for any temporal changes in the culture environment or the copepods
themselves, all phenotypic values were assessed relative to the midparent (the average of
the two replicated parental controls) for that same generation. Analyses of variance
including planned linear contrasts among cohorts were done using Statistica 7.1 (StatSoft,
Tulsa, OK).

Long-Term Hybrid Swarms
Four different culture treatments were set up for this experiment: 100% RP,
100% SD, 50% RP:50% SD, and 80%RP:20%SD. Five replicates per treatment were

19
each initiated by placing 500 gravid females in 1000 ml beakers containing 800 mls live
algal culture (Platymonas and Monochrisis) supplemented with 0.16 g finely ground
Spirulina and Tetramin flakes. Beakers were housed together in one incubator at 20°C
set to a 12h light: 12 h dark cycle. Once every two weeks a 50% seawater change was
performed. At the same time beakers were also fed and rotated within the incubator.
Every 3 months a census estimate was taken for each replicate. This was performed by
pouring the contents of an entire beaker into a 1L plastic bottle. The bottle was gently
inverted several times to evenly distribute copepods, after which 200 ml of culture was
poured into a 600 ml transparent Gladware container. A light box was used for visual
assistance in sorting copepods into males, females with eggs, pairs and subadult
categories using a Pasteur pipet. Each category was counted and all individuals were
returned to source beakers. Every 3 months, 20 gravid females and 20 mature males
were removed from each replicate beaker and were used for morphometric assays.
Females were also used for fitness assays. All copepods were returned to their source
beakers after assays were completed. Replicates were maintained for up to 30 months, at
which point 20 males and 20 females from surviving beakers were frozen for later
molecular analyses.

Fitness Assays
At each 3 month interval, 20 gravid females were sampled from each replicate
and isolated into individual Petri dishes containing 11 ml filtered seawater supplemented
with ground Spirulina and Tetramin flakes. Females with red egg sacs (red eggs being
more mature and therefore closer to hatching) were preferred to those whose eggs were

20
still green in color. Each dish was monitored once daily until eggs hatched. Opon
hatching, 10 larvae per clutch were pipetted into a new dish with fresh seawater culture
medium. Fourteen days later individuals in each dish were counted to determine
survivorship.

Morphological Assays
Morphometric measurements were taken from digital images of adult copepods
following procedures in Edmands and Harrison (2003). At each 3 month interval, up to
20 females and 20 males were randomly chosen from each replicate. All measurements
were done at a magnification of 32X using a Leica MZ12 dissecting microscope. Digital
images were captured and morphological measurements were taken using Optimas 5.2.
Absolute size was calibrated using a stage micrometer. Eight measurements were taken
for males: cephalothorax length (CTL), cephalothorax width (CTW), urosome length
(UL), urosome width (UW), telson width (TW), caudal seta length (CSL), antennule
width (AW) and clasper width (CLW). Four of the same measurements were taken for
females (CTL, CTW, UW and AW). Egg sac length and area was also measured for each
female. Every three months up to 40 individuals were scored from each replicate.

Microsatellite Assays
The 11 microsatellite loci and primers used here were developed using an
enriched DNA library from the RP population (Harrison et al. 2004). DNA was extracted
from individual copepods using the lysis protocol previously described in Edmands et al.
2005. Individual copepods were incubated in 50 μl lysis buffer at 65°C for 1 hour

21
followed by 100°C for 15 minutes. Polymerase chain reactions were carried out in 12 μl
volumes containing 0.5 μl template DNA, 0.25 μM fluorescently labeled forward primer,
1 μM reverse primer and 2.5 mM MgCl
2.
Temperature cycling was as follows: 5 min
denaturation at 94°C; 35 cycles of 30 s at 94°C, 35 s at 55°C, and 30 s at 72°C; 5 min at
72°C. This was with the exception of locus 480 which required an annealing temperature
of 62°C. Fluorescently labeled PCR products were run on a Beckman-Coulter CEQ 8000
Capillary Sequencer according to commercially recommended protocols. Allele sizes
were scored by eye.
Standardized hybrid indices for each individual were calculated by assigning a 0
for each RP allele and a 1 for each SD allele and then dividing by the number of loci
scored. In this way hybrid indices ranged from 0 to 1, with an expected hybrid index of
0.2 based on the 80RP:20SD starting frequencies. 7 diagnostic loci were used to calculate
hybrid indices for one 50:50 replicate and 9 diagnostic loci were used for one 80RP:20SD
replicate.

Statistical Analyses
Analyses of morphological and fitness characters within and between
experimental population treatments were done using Statistica 7.1 (StatSoft, Tulsa, OK).
Nested analysis of variance (ANOVA) was used to quantify differences in measures
among the different treatment types. Multivariate analysis of variance (MANOVA),
followed by a Wilks’ test, was used to quantify combined measures. When appropriate,
Bonferroni post hoc tests were utilized to determine the statistical significance between
group means. To compare each experimental replicate as well as all replicates of a

22
population type to the midparent value, we conducted ANOVAs and contrast tests using
SAS (proc GLM, SAS Institute 2006). Principle components analysis was performed on
all morphometric measures for each time point using Statistica 7.1.
Calculation of allele and genotype frequencies, deviations from Hardy-Weinberg
equilibrium, and linkage disequilibrium was performed by Genepop 4.0 (Raymond and
Rousset 1995). To test for epistatic interactions between loci, Chi square tests were used
to compare observed two-locus genotype numbers to expected numbers determined by
multiplying single-locus ratios (Statistica 7.1). Correlations between hybrid indices and
fitness were assessed using Statistica 7.1.

RESULTS

Controlled Crosses

Survivorship and Morphology
Proportional survivorship (Table 1-1 and Figure 1-1) showed no significant
deviation from the midparent in the F1 or backcross cohorts. F2 cohorts, however,
showed large (35-45%) and significant declines. Both F3 cohorts recovered to
survivorship values 11% below the midparent, and neither of these deviations were
significant. Compared to the survivorship data, male morphometric characters (Table 1-1)
tended to show smaller deviations from the midparent. For F1 cohorts there were

23
Table 1-1. Mean proportional deviation from midparent for survival and eight male morphometric characters in three-generation
controlled cross. Values are averages among replicates, relative to midparent in that same generation, with standard errors in
parentheses. Means significantly greater than midparent values (α = 0.05) according to planned linear contrasts are in bold. Means
significantly less than midparent values are indicated in bold italics.

Cohort
Proportional
survival CTL UL CTW UW TW CSL AW CLW
RP x RP Control 0.033(0.030) 0.016(0.015) -0.008(0.013) -0.021(0.018) 0.005(0.014) -0.014(0.022) 0.002(0.023) 0.031(0.029 0.030(0.026)
SD x SD Control -0.033(0.076) -0.016(0.010) 0.008(0.019) 0.021(0.008) -0.005(0.020) 0.014(0.014) -0.002(0.018) -0.031(0.032) -0.030(0.014)
RP x SD F1 -0.002(0.027) 0.032(0.015) 0.115(0.012) 0.077(0.017) 0.084(0.026) -0.054(0.028) 0.037(0.015) 0.074(0.023) 0.081(0.011)
SD x RP F1 0.057(0.044) -0.044(0.024) -0.029(0.022) 0.012(0.011) 0.017(0.019) -0.001(0.038) -0.008(0.030) 0.052(0.037) 0.068(0.019)
RP-Backcross 0.194(0.300) -0.023(0.023) 0.021(0.017) 0.024(0.022) 0.019(0.012) -0.015(0.051) 0.031(0.033) 0.026(0.012) 0.000(0.036)
SD-Backcross 0.106(0.279) 0.077(0.029) 0.020(0.020) 0.035(0.003) 0.012(0.004) 0.062(0.024) 0.093(0.049) -0.008(0.031) 0.004(0.024))
RP x SD F2 -0.347(0.188) 0.015(0.019) -0.010(0.010) -0.001(0.024) -0.025(0.015) -0.095(0.041) -0.067(0.049) -0.025(0.016) -0.038(0.016)
SD x RP F2 -0.448(0.102) -0.051(0.025) -0.045(0.024) -0.070(0.021) -0.014(0.032) -0.082(0.091) -0.009(0.012) -0.035(0.028) -0.027(0.030)
RP x SD F3 -0.107(0.037) 0.001(0.019) 0.021(0.020) 0.009(0.001) -0.023(0.001) 0.029(0.100) -0.019(0.022) -0.022(0.021) -0.051(0.019)
SD x RP F3 -0.106(0.042) -0.031(0.024) -0.040(0.013) -0.005(0.001) -0.052(0.010) 0.036(0.014) -0.013(0.022) -0.011(0.003) -0.031(0.016)

24
Figure 1-1. Survivorship graphed as proportional deviation from the midparent in three-
generation controlled cross. Values are averages among replicates, relative to midparent
assayed in the same generation,
+
1 standard error. Dashed line indicates the additive
expectation for each cohort.

-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
Survivorship

RP RP-Backcross F1,F2,F3 SD-Backcross SD
F1
F2
F3

25
significant, moderate (5-12%) increases in morphometric characters in 7 out of 16 cases,
and a significant decrease (4%) in one case. For backcross cohorts there were two cases
with significant increases in individual measures (8-9%). F2 and F3 cohorts tended to be
smaller than the midparent in 26 out of 32 cases, but the deviation was significant in only
one case (4%).

Long-Term Hybrid Swarms

Census
Throughout the duration of the hybrid swarm experiment, census counts for
individual replicate beakers showed large fluctuations between time points (Figure 1-2).
The census taken for a beaker at any given point was not a good indicator of what the
population size would be in the next three months (r = 0.08, p > 0.05). With a few
exceptions, census counts from hybrid swarm beakers were lower than midparent values
for the first six months of the experiment. By month 15, replicates remaining with living
copepods were the following: two RP controls, two SD controls, four 50:50 swarms, and
two 80RP:20SD swarms. At that time point, two 50:50 replicates and one 80RP:20SD
replicate had higher census counts than the midparent and differed by at least 12-fold.

Survivorship
The 3-month time point shows evidence of heterosis due to the presence of early
generation hybrids (Figure 1-3). Two 50:50 replicates and one 80RP:20SD replicate
have survivorship values significantly greater than the midparent (contrast test, p = 0.003,

26
Figure 1-2. Census counts for hybrid swarm replicates over 15 months of free mating. Replicates 1 thru 5 are beakers with the initial
ratio of 50% RP: 50% SD. 6 thru 10 were initiated with a ratio of 80% RP: 20% SD. Large bold numbers along x-axis indicate month
of sampling. Dashed lines indicate midparent values (average of the parental means) determined using means of replicate beakers for
specific months.

0
50
100
150
200
250
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
N
subadults
pairs
males
females
MP = 109.8 MP = 47.6 MP = 21.7 MP = 51.0 MP = 18.2
3 6 9 12 15

27
Figure 1-3. Mean proportional deviation from the midparent for survivorship of each
hybrid swarm replicate over 15 months of free mating. Error bar indicate one standard
error.

-0.8
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
Month
Survivorship
50/50_1
50/50_2
50/50_3
50/50_4
50/50_5
80/20_1
80/20_2
80/20_3
80/20_4
80/20_5
3 6 9 12 15

28
Table 1-2. Proportional deviation from midparent survivorship values for each hybrid swarm replicate. (a) 50:50 treatment, (b) 80:20
treatment. Means significantly different from midparent values (α = 0.05) according to independent linear contrasts are in bold.
Standard errors are in parentheses, with a few exceptional cases where only one individual from the replicate was measured.
Replicates that died out are indicated by a “-.”

(a)
Month Replicate

1 2 3 4 5
Mean of Reps
3 0.67(0.0624) 0.965(0.0483) 0.8211(0.0595) 0.895(0.0478) 0.6632(0.073)
0.8028(0.0601)
6 0.7842(0.0542) 0.8316(0.0654) 0.7264(0.0656) 0.3458(0.097) 0.7842(0.0659)
0.6944(0.0887)
50:50 9 0.5381(0.0889) 0.7608(0.0676) 0.7944(0.0669) 0.7842(0.0473) 0.5632(0.0892)
0.6881(0.0565)
12 1(0.0408) 0.75(0.25) 0.6907(0.0678) 0.83(0.0578) 0.875(0.0351)
0.8291(0.0533)
15 0.8037(0.1564) - 0.8785(0.0416) 0.8926(0.046) 0.7
0.8187(0.0441)

(b)
Month Replicate

1 2 3 4 5
Mean of Reps
3 0.6421(0.0702) 0.91(0.0369) 0.7944(0.0602) 0.7383(0.0601) 0.7778(0.0712)
0.7725(0.0434)
6 0.7474(0.077) 0.6986(0.0827) 0.66(0.1006) 0.6454(0.061) 0.5927(0.0585)
0.6688(0.026)
80:20 9 0.7519(0.0703) - 0.7994(0.0468) - 0.5714
0.7076(0.085)
12 0.7533(0.0682) - 0.7346(0.0798) - -
0.744(0.0093)
15 - - - - 0.675(0.0507)
0.675

29
0.038 and 0.025 respectively) (Table 1-2). Evidence of outbreeding depression in both
types of hybrid swarms is not severe throughout the first few months of swarm formation.
In the most severe case, month 6, one 50:50 replicate is significantly less than the
midparent by approximately 55% (p = 0.038). At the same time point one 80RP:20SD
replicate is significantly less than the midparent by about 26% (p = 0.025). At month 9,
two 50:50 replicates are significantly less than the midparent value by about 26 and 30%
(p = 0.003, 0.038 respectively). There are no other significant negative deviations from
midparent values throughout the course of the experiment. By month 12 three 50:50
replicates and one 80RP:20SD replicate show significant positive deviations from the
midparent (Table 1-2). Finally by month 15 two 50:50 replicates are significantly greater
than the midparent by approximately 63 and 66% (p = 0.007, < 0.001 respectively). One
80RP:20SD replicate exceeds the midparent by about 25%, but this difference is not
significant.

Morphological Assays
Nested ANOVA followed by planned linear contrasts were performed across all
time points to assess morphological variation (SAS, proc glm) (Tables 2-3 and 2-4). A
multivariate test indicated a significant effect of both treatment and replicate for females
and males. Both month and treatment had a significant effect on all four morphological
characters measured in females as well as the two egg sac measurements. Of the eight
male morphological characters, six showed a significant month effect and six differed
significantly among treatments. One-way MANOVAs were run for each treatment by
month for all measurements (Table 1-5). Thirteen of sixteen possible female treatment

30
Table 1-3. Mean phenotypic values for female morphometric characters for all control and hybrid swarm treatments. Units are in
millimeters. Treatment means significantly different from the midparent according to independent linear contrasts are indicated in
bold (α = 0.05). Values significantly greater than the midparent are marked with a + sign while those significantly less than the
midparent are marked by a - sign. Standard errors are in parentheses with the exception of months 12 and 15 where only one replicate
remained for parental treatments.
CTL CTW UW AW ESA ESL
3 months RP 0.5152(0.0067) 0.3409(0.0038) 0.1619(0.0026) 0.0404(0.0011) 0.1181(0.0090) 0.5114(0.0279)
SD 0.5486(0.0052) 0.3638(0.0039) 0.1830(0.0023) 0.0432(0.0008) 0.1346(0.0031) 0.5399(0.0078)
50:50 0.5268(0.0108) 0.3633(0.0039)+ 0.1870(0.0027)+ 0.0426(0.0013) 0.1183(0.0082)- 0.4947(0.0207)-
80:20 0.5340(0.0097) 0.3598(0.0024)+ 0.1867(0.0033)+ 0.0413(0.0015) 0.1291(0.0104) 0.5244(0.0185)
midparent 0.5319 0.3524 0.1725 0.0418 0.1264 0.5257

6 months RP 0.4707(0.0121) 0.3221(0.0082) 0.1618(0.0076) 0.0399(0.0004) 0.0803(0.0044) 0.4203(0.0126)
SD 0.5185(0.0103) 0.3513(0.0047) 0.1742(0.0029) 0.0385(0.0036) 0.1099(0.0073) 0.4844(0.0192)
50:50 0.5274(0.0166)+ 0.3540(0.0091)+ 0.1729(0.0032)+ 0.0411(0.0010) 0.1303(0.0108)+ 0.5303(0.0243)+
80:20 0.5202(0.0108)+ 0.3525(0.0064)+ 0.1747(0.0031)+ 0.0406(0.0011) 0.1306(0.0115)+ 0.5347(0.0265)+
midparent 0.4946 0.3367 0.168 0.0392 0.0951 0.4524

9 months RP 0.4650(0.0107) 0.3141(0.0141) 0.1497(0.0069) 0.0371(0.0023) 0.0750(0.0087) 0.4030(0.0261)
SD 0.4866(0.0103) 0.3228(0.0039) 0.1635(0.0016) 0.0419(0.0007) 0.0913(0.0069) 0.4441(0.0194)
50:50 0.5044(0.0154)+ 0.3340(0.0075)+ 0.1636(0.0020)+ 0.0399(0.0009) 0.1152(0.0057)+ 0.4958(0.0162)+
80:20 0.4410(0.0138)- 0.3050(0.0028)- 0.1509(0.0042) 0.0376(0.0009) 0.0691(0.0068) 0.3768(0.0156)-
midparent 0.4758 0.3185 0.1566 0.0395 0.0832 0.4236

12 months RP 0.4937 0.3271 0.1612 0.0379 0.0916 0.4471
SD 0.4791 0.3477 0.1663 0.0433 0.0982 0.4814
50:50 0.4947(0.0133)+ 0.3346(0.0084) 0.1677(0.0039) 0.0396(0.0005) 0.1144(0.0059)+ 0.5053(0.0176)
80:20 0.4917(0.0086) 0.3221(0.0212) 0.1723(0.0060) 0.0359(0.0024) 0.0971(0.0142)+ 0.4601(0.0264)
midparent 0.4864 0.3374 0.1637 0.0406 0.0949 0.4642

15 months RP 0.4769 0.3314 0.1675 0.0376 0.0737 0.3989
SD 0.5222 0.3602 0.1759 0.0436 0.1193 0.5094
50:50 0.4634(0.0140)- 0.3259(0.0023)- 0.1555(0.0025)- 0.0419(0.0031) 0.0916(0.0156)- 0.4565(0.0482)+
80:20 0.4470(0.0788)- 0.3095(0.0236)- 0.1313(0.0256)- 0.0344(0.0097) 0.0833(0.0565)- 0.3978(0.1669)-
midparent 0.4995 0.3458 0.1717 0.0406 0.0965 0.4542

31
Table 1-4. Mean phenotypic values for male morphometric characters in hybrid swarm replicates. Units are millimeters. Treatment
means significantly different from the midparent according to independent linear contrasts are indicated in bold (α = 0.05). Values
significantly greater than the midparent are marked with a + sign while those significantly less than the midparent are marked by a -
sign. Standard errors are in parentheses with the exception of months 12 and 15 for which only one control replicate of each
population remained.
Month Treatment CTL UL CTW UW TW CSL AW CLW
3 RP 0.4716(0.0091) 0.3380(0.0008) 0.3169(0.0055) 0.1522(0.0025) 0.0331(0.0008) 0.6495(0.0122) 0.0490(0.0011) 0.0655(0.0008)
SD 0.5145(0.0045) 0.3612(0.0041) 0.3413(0.0028) 0.1619(0.0015) 0.0356(0.0014) 0.6456(0.0055) 0.0494(0.0007) 0.0658(0.0013)
50:50 0.5053(0.0045)+ 0.3583(0.0021) 0.3526(0.0044)+ 0.1557(0.0009) 0.0327(0.0006) 0.6966(0.0114)+ 0.0488(0.0005) 0.0665(0.0007)
80:20 0.4967(0.0050) 0.3534(0.0037) 0.3495(0.0037)+ 0.1626(0.0016)+ 0.0341(0.0011) 0.7330(0.0092)+ 0.0493(0.0011) 0.0679(0.0009)+
midparent 0.4931 0.3496 0.3291 0.1571 0.0343 0.6475 0.0492 0.0657

6 RP 0.4610(0.0050) 0.3250(0.0020) 0.3157(0.0045) 0.1629(0.0005) 0.0335(0.0016) 0.6673(0.0264) 0.0526(0.0013) 0.0651(0.0010)
SD 0.4954(0.0082) 0.3538(0.0113) 0.3420(0.0042) 0.1653(0.0038) 0.0353(0.0007) 0.7168(0.0174) 0.0516(0.0008) 0.0672(0.0008)
50:50 0.4747(0.0069) 0.3400(0.0066) 0.3379(0.0065)+ 0.1628(0.0032) 0.0340(0.0011) 0.7051(0.0136) 0.0495(0.0011)- 0.0668(0.0014)
80:20 0.4794(0.0054) 0.3480(0.0047) 0.3408(0.0051)+ 0.1588(0.0018)- 0.0330(0.0005) 0.7093(0.0153) 0.0504(0.0011) 0.0690(0.0009)
midparent 0.4782 0.3394 0.3289 0.1641 0.0344 0.692 0.0521 0.0661

9 RP 0.4496(0.0156) 0.3292(0.0179) 0.3329(0.0074) 0.1483(0.007) 0.0326(0.0006) 0.7159(0.0205) 0.0478(0.0016) 0.0652(0.0006)
SD 0.4857(0.008) 0.3485(0.0076) 0.3393(0.0033) 0.1607(0.0021) 0.0333(0.0007) 0.7217(0.0349) 0.0502(0.0013) 0.0671(0.0009)
50:50 0.4667(0.008) 0.3485(0.0039) 0.3347(0.005) 0.1557(0.0013) 0.0337(0.0006) 0.7415(0.0052) 0.0508(0.0004) 0.068(0.0006)
80:20 0.4469(0.0084)- 0.3229(0.0033)- 0.3299(0.0042)- 0.1525(0.0023)- 0.035(0.0017) 0.707(0.0212)- 0.048(0.0017) 0.0644(0.0006)
midparent 0.4676 0.3389 0.3361 0.1545 0.0329 0.7188 0.049 0.0662

12 RP 0.4631 0.36530 0.3470 0.14820 0.03530 0.73320 0.0570 0.07250
SD 0.4785 0.35810 0.34080 0.15610 0.03560 0.74970 0.0490 0.06670
50:50 0.4637(0.0128) 0.3360(0.0074)- 0.3314(0.0095) 0.149(0.0042) 0.0338(0.0005) 0.728(0.0055) 0.0496(0.001) 0.0649(0.0008)
80:20 0.4567(0.0019) 0.3256(0.0087)- 0.3408(0.0058) 0.1531(0.0009) 0.0345(0.0012) 0.7437(0.0038) 0.0472(0.0004)- 0.0714(0.0022)
midparent 0.4708 0.3617 0.3439 0.1521 0.0355 0.7415 0.053 0.0696

15 RP 0.467 0.34090 0.34020 0.15410 0.03540 0.71320 0.05720 0.07120
SD 0.4829 0.34920 0.35350 0.16230 0.03710 0.7830 0.04790 0.06690
50:50 0.4626(0.0025)- 0.3465(0.0064) 0.3352(0.0039)- 0.1502(0.0008)- 0.0325(0.0022) 0.7217(0.0116)- 0.0515(0.0008) 0.0699(0.0011)
80:20 0.4581(0.0131) 0.3278(0.0154) 0.3346(0.0102) 0.1472(0.0013)- 0.032(0.0024) 0.7346(0.0113) 0.0464(0.0031) 0.0653(0.0023)
midparent 0.4749 0.3451 0.3469 0.1582 0.0363 0.7481 0.0526 0.069

32
Table 1-5. Results of multivariate Wilks’ test. Significant tests are indicated in bold.

Month Females Males
F
Effect
df
Error
df p F
Effect
df
Error
df p
3 RP
2.350 32 326.12 0.000

2.350 32 326.12 0.000
SD
1.166 32 318.75 0.253

1.166 32 318.75 0.253
50:50
1.057 32 322.44 0.388

1.057 32 322.44 0.388
80:20
1.542 32 318.75 0.034
1.542 32 318.75 0.034

6 RP
1.954 12 46.00 0.052

0.927 24 84.71 0.566
SD
3.614 24 203.55 0.000
2.103 24 197.82 0.003
50:50
4.030 24 200.06 0.000
2.311 32 241.30 0.000
80:20
3.272 24 227.97 0.000
1.539 32 219.18 0.039

9 RP
2.200 18 71.20 0.010
2.405 24 110.81 0.001
SD
2.265 12 88.00 0.015
1.857 24 145.62 0.014
50:50
4.244 24 287.27 0.000
1.742 32 322.44 0.009
80:20
1.767 12 36.00 0.093

1.461 16 36.00 0.169

12 RP
one replicate remaining

one replicate remaining
SD
one replicate remaining

one replicate remaining
50:50
2.076 24 112.84 0.006
3.767 24 154.32 0.000
80:20
2.5089 12 48.00 0.012
0.5174 16 48 0.925

15 RP
one replicate remaining

one replicate remaining
SD
one replicate remaining

one replicate remaining
50:50
1.774 16 70 0.053

1.774 16 70 0.053
80:20
3.537 8 14 0.019
3.537 8 14 0.019

33
groups returned a significant p-value (Wilks’ test, p<0.05). Ten of those were highly
significant (p < 0.0001). For male replicates 10 of 16 different treatment groups had a
significant replicate effect, suggesting that the mean of all replicates for a population type
is not necessarily indicative of any particular replicate.
At 3 months into the experiment, two morphological measures (CTW and UW)
for mixed-population females were significantly larger than midparent values (Table 1-
3). By month 15, both of those measures as well as CTL were significantly smaller than
midparents. Egg sac area decreased over time for all populations. At month 6, mixed
population egg sacs were larger than the midparent value, but by month 15 both the 50:50
and 80RP:20SD populations had egg sacs with area measurements significantly less than
the midparent. For males at month 3, all significant differences observed between mixed
populations and midparent values were greater than the midparent (Table 1-4). At month
15, all significant differences were less than the midparent. There was an overall trend
across treatments for morphological measurements to decrease over time with the
exception of male caudal setae length (CSL) and the end of the experiment most
differences observed between midparents and hybrid swarms in both males and females
showed that hybrid individuals were smaller.

Morphological Principle Components Analysis
After 3 months of swarm formation, principle components analysis was
performed on an average of 99.3 females from each treatment. The first two principle
components accounted for approximately 40.9% and 21.9% of the total phenotypic
variation respectively. Principle component 1 (PC1) was largely influenced by

34
cephalothorax size and egg sac size while variation along principle component 2 (PC2)
pertained largely to body width and egg sac size. Morphological measures show that the
50:50 and 80RP:20SD treatments were generally intermediate or equivalent to RP and
SD individuals, although a few transgressive outliers were observed (Figure 1-4a).
ANOVA performed on principle components scores suggests that the RP distribution is
significantly different from all other population types along both PC1 and PC2
(Bonferroni post hoc test, p < 0.01 for all pair wise tests) while SD is different from the
50:50 but not the 80RP:20SD treatment along both PC1 and PC2 (p < 0.001 for both
tests). The 50:50 and 80RP:20SD treatments are significantly different along PC2 only (p
= 0.012).
After 15 months, PCA was performed on an average of 26.8 females from each
treatment (Figure 1-4b). Reduction in numbers was due to the die off of some replicates
during the course of the experiment. PC1 was largely influenced by cephalothorax size
and egg sac size while variation along PC2 pertained largely to body width and egg sac
size. Along PC1, hybrid female morphology for most individuals fell within the ranges
of RP and SD parental morphology. A few transgressive outliers were observed. RP and
SD had significantly distinct distributions along PC1 (p < 0.0001), which accounted for
approximately 48.5 % of the total variation. The 80RP:20SD distribution along this axis
was similar to that of the SD parentals while the 50:50 distribution was more like that of
the RP parentals. Along PC2, which accounted for approximately 21.7% of the total
variance, 50:50 and 80RP:20SD hybrid individuals were present that were transgressive
relative to both parental populations. RP and SD showed similar distributions along PC2
(ANOVA with Bonferroni post hoc test, p > 0.999) and both hybrid swarm types

35
Figure 1-4. Principle components analysis of 6 morphometric measurements for females
after 3 months (a) and 15 months (b). At 3 months, PC1 and PC2 account for 40.9% and
21.9% of the variance respectively. At month 15, the first two components account for
48.5% and 21.7% of the variance. PCA of 8 morphometric measurements for males after
3 months (c) and 15 months (d). At 3 months PC1 and PC2 account for 25% and 13.8%
of the variance respectively. At month 15 the first two components account for 29.5%
and 15.1% of the variance.

(a)
Females, 3 Months
-5.0
-4.0
-3.0
-2.0
-1.0
0.0
1.0
2.0
3.0
4.0
-7.0 -2.0 3.0 8.0
PC1
PC2
RP_1
RP_2
RP_3
RP_4
RP_5
SD_1
SD_2
SD_3
SD_4
SD_5
50/50_1
50/50_2
50/50_3
50/50_4
50/50_5
80/20_1
80/20_2
80/20_3
80/20_4
80/20_5

36
Figure 1-4, Continued
(b)
Females, 15 Months
-3.0
-2.0
-1.0
0.0
1.0
2.0
3.0
4.0
-6.0 -4.0 -2.0 0.0 2.0 4.0 6.0 8.0
PC1
PC2
RP_3
SD_1a
SD_1b
50/50_1
50/50_3
50/50_4
50/50_5
80/20_1
80/20_5

(c)
Males, 3 Months
-4.0
-3.0
-2.0
-1.0
0.0
1.0
2.0
3.0
4.0
-4.0 -2.0 0.0 2.0 4.0 6.0 8.0
PC1
PC2
RP_1
RP_2
RP_3
RP_4
RP_5
SD_1
SD_2
SD_3
SD_4
SD_5
50/50_1
50/50_2
50/50_3
50/50_4
50/50_5
80/20_1
80/20_2
80/20_3
80/20_4
80/20_5

37
Figure 1-4, Continued
(d)
Males, 15 Months
-3.0
-2.0
-1.0
0.0
1.0
2.0
3.0
4.0
-3.0 -2.0 -1.0 0.0 1.0 2.0 3.0 4.0 5.0
PC1
PC2
RP_3b
SD_1a
SD_1b
50/50_1
50/50_3
50/50_4
80/20_1
80/20_5

38
exceeded the ranges of either parent. Mean PC2 scores for individuals from 80RP:20SD
populations were significantly different from RP, SD and 50:50 populations (ANOVA
with Bonferroni post hoc test, p < 0.0001 for all three comparisons). All other post hoc
comparisons between treatments were not significant.
Across all time points, most males from hybrid replicates, as observed using PCA,
fell within the ranges of the two parent morphologies although a few hybrid individuals
were transgressive outliers (Figure 1-4c,d). At month 3, an average of 98.8 males were
measured from each treatment. ANOVA performed on principle components scores show
RP to be significantly different from all other treatments along PC1 (Bonferroni post hoc
test, p < 0.0001 for all pair wise tests) and different from SD and the 50:50 treatment
along PC2 (p ≤ 0.01). PC1 and PC2 represent 25% and 13.8% of the variance
respectively. At month 15 an average of 27.5 males per treatment were measured and
RP is significantly different from all other treatments along PC2 which accounts for
15.1% of the variance. Along PC1, which accounts for 29.5% of the variance and was
largely influenced by cephalothorax size, SD is significantly different from RP (p < 0.01)
and the 50:50 treatment (p < 0.0001).

Microsatellite Assays
After 12 months of hybridization, 31 individuals from one 50:50 swarm replicate
were scored for nine microsatellite loci. For all loci, alleles deviated from the expected
50:50 starting ratio toward increased frequencies of RP alleles, and seven out of nine of
those deviations were significant (Chi-squared test, p < 0.05) (Table 1-6a). None of the
loci scored were fixed for either population’s alleles. Heterozygote deficiency and excess

39
Table 1-6. Estimated population allele frequencies in (a) one 50:50 replicate after 12
months of free mating and (b) one 80:20 replicate after 30 months. Bold numbers
indicate the direction of deviation from additivity. Asterisks indicate a statistically
significant deviation (Chi-squared test, p < 0.05).

(a)
Locus Allele Frequencies
RP SD
228* 0.89 0.12
1814* 0.67 0.33
56J2 0.55 0.45
1203* 0.69 0.31
558* 0.68 0.32
197* 0.71 0.29
1555 0.62 0.38
62J8* 0.63 0.36
1202 0.61 0.39

(b)
locus Allele frequencies
RP SD
228 0.81 0.19
1814 0.83 0.18
56J2 0.79 0.21
1203 0.81 0.19
558* 0.70 0.30
197 0.74 0.26
1555 0.79 0.21
62J8* 0.65 0.35
480* 0.97 0.03
1202* 0.49 0.51
30 0.88 0.13

40
was measured by calculating F
IS
according to Weir and Cockerham (1984). Deviations
from Hardy-Weinberg equilibrium calculated in GENEPOP indicate trends toward
heterozygote excess at 8 of 9 loci, although none of these deviations were statistically
significant. Heterozygote excess was observed for 6 of 9 loci in females and 4 of 9 loci
in males, but none of these deviations were statistically significant according to an exact
HW test. Deviations from linkage disequilibrium were computed using GENEPOP web
version 3.4. Of a total of 36 locus pairs, the following showed significant linkage
disequilibrium: TC1202 and TC558, TC228 and TC558, TC197 and TC1814, and 62J8
and 1203 (total); TC1202 and 1555 (females); TC558 and TC197, TC197 and TC1814,
TC1814 and TC1203 (males). Following both Bonferroni correction and sequential
Bonferroni correction, only loci TC197 and TC1814, which are physically linked,
showed significant linkage disequilibrium when analyzed for males only as well as across
the total population. The only other physically linked loci were TC62J8 and TC1203.
The expected hybrid index for this swarm population in Hardy-Weinberg
equilibrium was 0.5, given 50% RP and 50% SD starting allele frequencies. Mean
female hybrid index (0.297, n = 13) and mean male hybrid index (0.331, n = 18) were
equivalent to each other (independent t-test). Overall the average hybrid index was 0.317
(n = 31) and significantly lower than the expected value (p < 0.0001, t-test against value
of 0.5) (Figure 1-5a). No significant correlation (r = 0.18) was found between hybrid
index and survivorship for this swarm replicate.
After 30 months of hybridization 40 individuals from one swarm replicate (initial
starting frequencies of 80% RP: 20% SD copepods) were scored for 11 microsatellite
loci. Overall, allele frequencies remained close to the ratio of 80% RP and 20% SD, with

41
Figure 1-5. Frequencies of hybrid indices for genotyped replicates. (a) 50:50, 12 months.
(b) 80:20 30 months. Vertical dashed lines indicate expected mean hybrid index: 0.5 for
(a) and 0.2 for (b).

(a)
0
2
4
6
8
10
12
14
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Hybrid Index
Frequency

(b)
0
2
4
6
8
10
12
14
16
18
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Hybrid Index
Frequency
Mean SE
0.235 0.017
0.236 0.026
0.235 0.023
total
females
males

42
four of the diagnostic loci showing deviations toward SD alleles and five loci tending
towards RP in character (Table 1-6b). Two loci, TC480 and TC1202, show particularly
extreme deviation from the expected value (Chi-squared test, p = 0.001 and p < 0.0001
respectively) with TC480 deviating toward more RP alleles than expected while TC1202
shows the opposite pattern. Also showing a significant deviations were loci TC558 and
TC62J8 which both had an excess of RP alleles.
Heterozygote deficiency and excess were measured and deviations from Hardy-
Weinberg equilibrium indicate trends toward heterozygote excess at most loci.
Significant heterozygote excess was observed in both females and the total population at
locus TC1202 (p
females
= 0.025, p
tot
= 0.021) and in females only at locus TC56J2 (p =
0.0006). After Bonferroni corrections were performed, only the heterozygote excess in
females at locus TC56J2 significantly deviated from Hardy-Weinberg equilibrium. A
multi-locus test of heterozygote excess (U-test, Rousset and Raymond 1995) indicated
that global heterozygote excess is significant for females (p = 0.003) but not in males (p =
0.788) or the total population (p = 0.100). Of 55 locus pairs, the following showed
significant linkage disequilibrium (Table 1-7): TCS228 and TC1814, TC62J8 and
TC1814, TCS228 and TCS197 (total); TCS228 and TCS197, TCS558 and TC1202,
TCS197 and TC1202 (females); TC1203 and TC62J8, TCS558 and TCS030 (males).
Following both Bonferroni correction and sequential Bonferroni correction no locus pairs
showed significant linkage disequilibrium. When two-locus genotypes were pooled into
four categories (parental homozygotes, nonparental homozygotes, homozygote-
heterozygotes and double heterozygotes), there was a trend toward excess double
heterozygotes but no significant deviations from Mendelian expectations.

43
Table 1-7. Pairwise tests of linkage disequilibrium for 50:50 and 80:20 replicates.
Significant deviations are shown in bold. Asterisks next to locus numbers indicate pairs
of loci that are physically linked. After bonferroni correction, no locus pairs show
significant LD. *p < 0.05, **p < 0.01, ***p < 0.001

50:50 replicate, 12 months 80:20 replicate, 30 months
Locus
1
Locus
2 females males total females males total
228 1814 ns ns ns ns ns *
62J8 1814 ns ns ns ns ns *
1203 1814 ns * ns ns ns ns
558 197 ns ** ns ns ns ns
228 197 ns ns ns * ns **
558 1202 ns ns ** * ns ns
197 1202 ns ns ns ** ns ns
1555 1202 * ns ns ns ns ns
558 228 ns ns * ns ns ns
*62J8 1203 ns ns ** ns * ns
*197 1814 ns *** *** ns ns ns
*558 30 ns ns ns ns * ns

44
The expected hybrid index for this swarm population in Hardy-Weinberg equilibrium
was 0.2, given 80% RP and 20% SD starting allele frequencies. Mean female hybrid
index (0.235, n = 20) and mean male hybrid index (0.235, n = 20) were equivalent to
each other. Overall the average hybrid index was 0.235 (n = 40) and slightly higher than
the expected value (p = 0.044, t-test against value of 0.2) (Figure 1-5b). No significant
correlation (r = 0.01) was found between hybrid index and survivorship for this particular
replicate.

DISCUSSION

Predictions from controlled crosses

For all controlled crosses, only F2s showed significant declines in survivorship
compared to midparent values (Figure 1-1). The large and significant F2 declines in
survivorship portend fitness problems in the earlier hybrid swarm generations. The lack
of depressed survivorship in backcross hybrids implicates homozygote-homozygote
interactions may be the primary source of fitness problems. Significant recovery between
the F2 and F3 cohorts suggests that these deleterious epistatic interactions can be
efficiently purged. Yet the actual time course of recovery in freely mating hybrid swarms
is difficult to predict because of overlapping generations and the persistence of fertilized
females with lifespans up to 95 days (Vittor 1971). Male morphometric characters (Table
1-1) for F2 and F3 cohorts tended to be smaller than the midparent but this was only

45
significant for one case. For F1 and backcross cohorts most significant differences
observed showed increases in size. For morphology, it is more difficult to predict what
may happen in the long term because controlled cross hybrids show both patterns of
increase and decrease in size and many deviations from the midparent are non-
significant.

Duration of Outbreeding Depression
Other studies have focused on the magnitude of outbreeding depression in
controlled crosses (e.g. Burton 1987, Edmands and Burton 1999, Edmands 1999, Fenster
and Galloway 2000), partly because knowing the genomic composition of any parent
generation allows for comparisons of observed versus expected genotypic frequencies.
This is the first multi-generation hybridization study in this species in which individuals
were allowed to freely choose their mates throughout the course of the experiment. In
doing so, this mimics a natural hybridization event that may occur if divergent
populations have the opportunity to mix.
With an understanding of fitness measures for early generation crosses we can
begin to assess what may be taking place after many generations of hybridization.
Survivorship data suggest that after only three months of free mating (up to 4 generations
in the lab) none of the replicate populations show effects of outbreeding depression
(Figure 1-3). There are several ways that this may occur: (1) Individuals mate
assortatively or hybrids are selected against so that only parental genotypes remain in the
population (2) genetic swamping occurs in which one population’s genes are selected for
while genes from the second population decline in frequency (3) the effects of

46
outbreeding depression may be decreased or delayed because fitness measures include
persisting parentals as well as both early generation heterosis and hybrid breakdown and
(4) within the first few generations of hybridization selection may have chosen highly fit
recombinant genotypes such that oubreeding depression is rapidly purged. Evidence
suggests that Tigriopus does not avoid outbreeding (Ganz and Burton 1995, Palmer and
Edmands 2000). This, combined with the fact that hybrid genotypes were observed at
later time points in the experiment (Figure 1-5), indicates that scenario number one is not
taking place. Genetic swamping could be easily identified, as most individuals would be
homozygous for one population’s alleles, but we do not see this signature in any of the
experimental swarms. Even with the limited amount of molecular data we have for this
study, long-term hybrid cultures utilizing these particular parental populations have never
shown evidence of genetic swamping (Edmands et al. 2005, Hwang et al. unpublished
data).
In the absence of assortative mating, overlapping generations or selection, it
would only take three generations (less than 3 months) to achieve that largest proportion
of F2 individuals: a maximum of 25% in a 50:50 replicate and 10% in an 80RP:20SD
replicate. Yet hybrid replicate survivorship declines through months six and nine (Figure
1-3). This finding would be consistent with scenario 3, in which the manifestations of
outbreeding depression are delayed due to the presence of a significant proportion of
parentals and F1s in early generations. It is possible that it could take up to 12
generations (9 months) for the replicate to reach its maximum capacity of F2 individuals.
It is not certain, however, whether the greatest effects of outbreeding depression have

47
taken place before month three, or if the decline in fitness observed at months six and
nine represents the deepest adaptive valley that the mixed populations may encounter.
What stands out as most remarkable is that, not only do mixed populations
recover to expected midparent values but, by month fifteen, two out of four surviving
replicates significantly exceed midparent fitness (Figure 1-2, Table 1-1). As seen from
the controlled crosses (Figure 1-1), the backcrosses produced by this particular pair of
populations showed high fitness and this may have largely contributed to the recovery
from outbreeding depression. Recent evidence shows that Tigriopus populations may
harbor an epistatic load of maladapted gene combinations so that hybridization may
simultaneously create gene combinations that are both better and worse than those of the
parental populations (Edmands et al. in press, Appendix). This, along with evidence of
transgressive morphometric measures (Fig. 4), offers support to the idea that
hybridization can generate novel genetic variants that exceed parental phenotypes. If this
is the case, it is likely that scenario number four may have taken place and that selection
is capable of choosing the highly fit recombinant genotypes to persist in the population.
Additionally, scenarios 3 and 4 are not mutually exclusive and may be taking place
simultaneously.

Morphological and Molecular Patterns
Morphological data indicate that, over 15 months or a maximum of ~20
generations, hybrids experience a trade-off between survivorship and size. For females
the means of all morphometric characters for all treatments decreased or remained
equivalent over 15 months. In addition to individuals becoming smaller over time, both

48
body size and egg sac area of hybrid swarms are significantly smaller than the midparent
at month 15, though survivorship is higher. Male morphometric characters show the
same patterns of decreased measures over time with the one exception being caudal setae
length (CSL) which either increases in size or remains unchanged for all treatments by
month 15 (Table 1-3). Although CSL increases in size over time, by month 15 the 50:50
treatment is still significantly smaller than the midparent for this metric. The 80RP:20SD
treatment is also smaller than the midparent but this difference is not significant. Further
investigation needs to be conducted to determine whether the lab environment may create
situations where smaller individuals are truly more fit, or if the most fit hybrid genotypes
are constrained to those phenotypes that trade body size for enhanced survivorship.
Principle components analysis at 15 months showed that females exhibited
transgressive phenotypes, consistent with the idea that hybridization can enhance genetic
diversity through the creation of novel genotypes. Female characters showed similar
trends for replicates of the same treatment (Figure 1-4), while each treatment is
significantly different from every other treatment along at least one principle component.
Treatments have taken different paths of morphological evolution and replicates of the
same treatment tend to be similar to each other. This would argue against any notion that
female morphology may be largely influenced by drift and would support the suggestion
that a populations’ genetic composition may have a large influence on optimal
morphologies.
Overall, microsatellite analysis shows extensive introgression for both swarm
treatments that were genotyped (Figure 1-5, Table 1-5). This is consistent with Edmands
et al. (2005) who showed that, after one year, hybridity increased in all experimental

49
replicates. Heterozygote advantage may be contributing to these results, as microsatellite
data for both populations that were genotyped show a trend toward heterozygote excess.
For both replicates genotyped, the number of loci showing heterozygote excess was
greater in females than in males. However, only one locus, TC56J2, for females of the
80RP:20SD population showed heterozygote excess that was statistically significant.
Greater heterozygote excess for females was also reported by Harrison and Edmands
(2006). This may be explained by reduced viability of homozygous females due to
gender-specific effects of inbreeding. If the isolated pools of T. californicus harbor a
genetic load, females may benefit more from the masking of deleterious alleles.
Alternatively, female sex determining factors might also have negative epistatic
interactions with homozygous loci. This would result in differential viability between
males and females if females are the heterogametic sex. While the heterogametic sex of
Tigrioupus is still unknown, the lack of recombination in females makes them the
stronger candidates. Another possibility for overall heterozygote excess is that there
could be a fitness advantage associated with specific combinations of the two parental
genomes. If beneficial epistasis is present such that multilocus hybrid genotypes exhibit
increased fitness, we may expect consistent patterns of LD, but after 30 months of free
mating, LD was not detected following correction for multiple tests. The data also show
that no significant correlations were found between hybrid index and fitness. It is possible
that markers may not be closely associated enough with loci under selection and that this
small sample size does not offer enough resolution to determine the extent of beneficial
epistasis.

50
For the 80RP:20SD replicate, when genotypes were pooled into four different
categories (parental double homozygote, homozygote-heterozygote, heterozygote-
heterozygote and nonparental double homozygote), significant deviations from
Mendelian expectations were not found. This is interesting compared to prior studies,
which have shown strong deviations of two-locus pooled frequencies in early generations
of hybridization (Edmands et al. in press, Willett 2006). It is possible that time alone
may have allowed for enough recombination to break up the linkage between marker loci
and coadapted loci, or that this small data set does not have the power to detect the
patterns found in the previous study.
Based upon fitness results from controlled crosses and the fact that RP alleles
increased in frequency at most loci examined, there may be enhanced fitness associated
with alleles from the RP population. Similar outcomes of increased RP alleles were
observed in Edmands et al. 2005. Yet three of the 50:50 replicates survived the duration
of the experiment compared to one remaining 80RP:20SD replicate, and the 50:50
replicates had the highest fitness (Figure 1-3). However, it was the 80RP:20SD replicate
that survived long past the duration of the experiment to month 30 while all other swarm
and parental replicates went extinct. Even if there is a fitness advantage associated with
increased RP alleles, it seems that the initial starting contributions of each source
population do not aid in predicting the final outcome of hybridization.

Is the outcome of hybridization repeatable?
We were particularly interested in assessing the nature of repeatability in this
hybridized system. In other words, given the same set of starting circumstances, does

51
evolution tend to arrive upon the same outcome? This knowledge would be useful for
managers attempting to restore multiple populations of the same species, especially since
there are several examples showing that repeated hybridization events have led to similar
results (Brochman et al. 2000, Schwarzbach and Rieseberg 2002). One of the strongest
examples comes from a study by Rieseberg et al. (1996) in which three different hybrid
sunflower species were experimentally created using the same two parental species.
Conservation of large linkage blocks and evidence for strong epistatic interactions were
observed among three synthetic hybrid species as well as a fourth ancient hybrid.
Chromosomal rearrangements that resist recombination in the parental species may have
contributed to the observed repeatability, but the overall implication is that strong
deterministic forces are involved in driving the consequences of hybridization.
Conversely, a study using Tigriopus found only partial concordance among four hybrid
replicates with each replicate taking a different pathway of molecular evolution to arrive
at a similar conclusion of increased RP allele frequencies (Edmands et al. 2005).
A thorough observation of the individual morphometric characters of each
experimental replicate in this study indicates that the path of morphological evolution is
not necessarily repeatable, as replicates have distinct trajectories. MANOVAs for
morphometric characters followed by mutivariate Wilk’s tests showed that there was a
significant effect of replicate within treatment for at least one of the two types of hybrid
swarms at all months.
Despite the many differences between replicate populations for several
morphometric characters, the first two principle components from PCA suggest that
populations of the same treatment arrived at similar morphologies by month fifteen (Fig.

52
4). Additionally, surviving 50:50 replicates all exceeded the midparent value and, in our
study, like that of Edmands et al. (2005), arrived at RP allele frequencies higher than
expected. Patterns of linkage disequilibrium were not concordant between the two
swarm replicates genotyped nor with the 2005 study. The individual replicates that were
able to overcome the threat of extinction were those with the potential to exceed fitness
values of parental controls. Further studies that integrate detailed accounts of both
molecular and phenotypic data would help to resolve the questions surrounding hybrid
repeatabiliy. It is possible that there are potential deterministic forces for morphological
and molecular evolution, but that they are hampered by drift that may take place as
population sizes fluctuate, which clearly occurred in this study (Figure 1-2). T.
californicus’ native habitat is composed of the highest tidepools that experience extreme
fluctuations in variables such as temperature and salinity so it is entirely possible that the
evolution of this species is strongly influenced by drift compared to selection. This
would be consistent with the little evidence of local adaptation found (Edmands and
Deimler 2004, Hwang et al. unpublished data) as well as levels of molecular subdivision
that exceed measures of quantitative subdivision (Edmands and Harrison 2003).

Conclusion
In conclusion, this study offers extensive experimental evidence that hybrid
swarms can recover from the effects of severe early-generation outbreeding depression.
It lends support to the notion that outbreeding depression may be a temporary
phenomenon (Carney et al. 2000, Christiansen 2008, Rieseberg et al. 1996, Templeton
1986) and offers a ray of hope for managers faced with situations where the introduction

53
of genetically divergent individuals may be the only remaining option to bolster a
dwindling population. However, it is worth noting that, for this pair of populations, both
reciprocal backcrosses had uncharacteristically high fitness which may have contributed
to the rapid recovery observed. Caution must be taken to assess whether the mixing of
populations is likely to result in hybrid persistence or genetic swamping, as the amount of
divergence between the populations at hand, as well as the fitness of the individual
populations in a particular environment, may have a large influence on the outcome of
hybridization (Hwang et al. in prep). Finally, while we have clearly demonstrated the
ability to purge the deleterious effects of outbreeding, the long-term outcome of any
particular hybrid swarm appears to be largely unpredictable with respect to the initial
frequency of individuals from source populations

54
CHAPTER 1 REFERENCES

Allendorf FW, Leary RF, Sprunell P, Wenburg JK (2001) The problems with hybrids:
setting conservaton guidelines. Trends in Ecology and Evolution, 613-622.

Arnold M (2006) Evolution Through Genetic Exchange Oxford University Press, Oxford,
UK.

Bowman J, Kidd AG, Gorman RM, Schulte-Hostedde AI (2007) Assessing the potential
for impacts by feral mink on wild mink in Canada. Biological Conservation 139,
12-18.

Brochmann C, Borgen L, Stabbetorp OE (2000) Multiple diploid hybrid speciation of the
Canary Island endemic Argyranthemum sundingii (Asteraceae). Plant Systematics
and Evolution 220, 77-92.

Burton RS (1985) Mating system of the intertidal copepod Tigriopus californicus. Marine
Biology 86, 247-252.

Burton RS (1986) Incorporation of 14C-bicarbonate into the free amino acid pool during
hyperosmotic stress in an intertidal copepod. The Journal of Experimental
Zoology 238, 55-61.

Burton RS (1987) Differentiation and integration of the genome in populations of the
marine copepod Tigriopus californicus. Evolution 41, 504-513.

Burton RS (1987) Differentiation and integration of the genome in populations of the
marine copepod Tigriopus californicus. Evolution 41, 504-513.

Burton RS (1990) Hybrid breakdown in developmental time in the copepod Tigriopus
californicus. Evolution 44, 1814-1822.

Burton RS (1990) Hybrid breakdown in physiological response: A mechanistic approach.
Evolution 44, 1806-1813.

Burton RS, Lee B-N (1994) Nuclear and mitochondrial gene genealogies and allozyme
polymorphism across a major phylogeographic break in the copepod Tigriopus
californicus. Proceedings of the National Academy of Sciences USA 91, 5197-
5201.

Burton RS, Rawson PD, Edmands S (1999) Genetic architecture of physiological
phenotypes: Empirical evidence for coadapted gene complexes. Amer. Zool. 39,
451-462.

55
Carney SE, Gardner KA, Rieseberg LH (2000) Evolutionary changes over the fifty-year
history of a hybrid population of sunflowers (Helianthus). Evolution 54, 462-474.

Christiansen FB (2008) Theories of Population Variation in Genes and Genomes
Princeton University Press, Princeton, NewJersey.

Dowling TE, Secor CL (1997) The role of hybridization and introgression in the
diversification of animals. Annual Review of Ecology and Systematics 28, 593-
619.

Edmands S (2001) Phylogeography of the intertidal copepod Tigriopus californicus
reveals substantially reduced population differentiation at northern latitudes.
Molecular Ecology 10, 1743-1750.

Edmands S (2007) Between a rock and a hard place: evaluating the relative risks of
inbreeding and outbreeding for conservation and management. Molecular
Ecology 16, 463 - 475.

Edmands S, Burton RS (1999) Cytochrome c oxidase activity in interpopulation hybrids
of a marine copepod: a test for nuclear-nuclear or nuclear-cytoplasmic
coadaptation. Evolution 53, 1972-1978.

Edmands S, Deimler JK (2004) Local adaptation, intrinsic coadaptation and the effects of
environmental stress on interpopulation hybrids in the copepod Tigriopus
californicus. Journal of Experimental Marine Biology and Ecology 303, 183-196.

Edmands S, Feaman HV, Harrison JS, Timmerman CC (2005) Genetic consequences of
many generations of hybridization between divergent copepod populations.
Journal of Heredity 92, 114-123.

Edmands S, Harrison JS (2003) Molecular and quantitative trait variation within and
among populatioins of the intertidal copepod Tigriopus californicus. Evolution
57, 2277-2285.

Edmands S, Timmerman CC (2003) Modeling factors affecting the severity of
outbreeding depression. Conservation Biology 17, 883-892.

Egloff DA (1966) Ecological aspects of sex ratio and reproduction in experimental and
field populations of the marine copepod Tigrioups californicus Dissertation,
Stanford University.

Ellstrand NC, Schierenbeck KA (2006) Hybridization as a stimulus for the evolution of
invasiveness in plants? Euphytica 148, 35-46.

56
Erickson DL, Fenster CB (2006) Intraspecific hybridization and the recovery of fitness in
the native legume Chamaecrista fasciculata. Evolution 60, 225-233.

Fenster CB, Galloway LF (2000) Inbreeding and Outbreeding Depression in Natural
Populations of Chamaecrista fasciculata (Fabaceae). Conservation Biology 14,
1406-1412.

Ganz HH, Burton RS (1995) Genetic differentiation and reproductive incompatibility
among Baja California populations of the copepod Tigriopus californicus. Marine
Biology 123, 821-827.

Gilk SE, Wang IA, Hoover CL, et al. (2004) Outbreeding depression in hybrids between
spatially separated pink salmon, Oncorhynchus gorbuscha, populations: marine
survival, homing ability, and variability in family size. Environmental Biology of
Fishes 69, 287-297.

Gow JL, Peichel CL, Taylor EB (2006) Contrasting hybridization rates between
sympatric three-spined sticklebacks highlight the fragility of reproductive barriers
between evolutionarily young species. Molecular Ecology 15, 739-752.

Grant PR, Grant BR (1992) Hybridization of bird species. Science 256, 193-197.

Harrison JS, Peterson DL, Swain JR, Edmands S (2004) Microsatellite DNA markers for
the intertidal copepod Tigriopus californicus. Molecular Ecology Notes 4, 736-
738.

Hutchings JA, Fraser DJ (2008) The nature of fisheries-and farming-induced evolution.
Molecular Ecology 17, 294-313.

Lynch M (1991) The genetic interpretation of inbreeding depression and outbreeding
depression. Evolution 45, 622-629.

McClelland EK, Naish KA (2007) What is the fitness outcome of crossing unrelated fish
populations? A meta-analysis and an evaluation of future research directions.
Conservation Genetics 8, 397-416.

McGinnity P, Prodohl P, Ferguson A, et al. (2003) Fitness reduction and potential
extinction of wild populations of Atlantic salmon, Salmo salar, as a result of
interactions with escaped salmon. Proceedings of the Royal Society of London B
270, 2443-2450.

Orr HA (1996) Dobzhansky, Bateson and the genetics of speciation. Genetics 144, 1331-
1335.

57
Palmer CA, Edmands S (2000) Mate choice in the fact of both inbreeding and
outbreeding depression in the intertidal copepod Tigriopus californicus. Marine
Biology 136, 693-698.

Randi E (2008) Detecting hybridization between wild species and their domesticated
relatives. Molecular Ecology 17, 285-293.

Randi E (2008) Detecting hybridization between wild species and their domesticated
relatives. Molecular Ecology 17, 285-293.

Raymond M, Rousset F (1995) GENEPOP (version 1.2) population genetic software for
exact tests and ecumenicism. Journal of Heredity 86, 248-249.

Rieseberg LH, Archer MA, Wayne RK (1999) Transgressive segregation, adaptation and
speciation. Heredity 83, 363-372.

Rieseberg LH, SInervo B, Linder CR, Ungerer MC, Arias DM (1996) Role of gene
interactions in hybrid speciation: evidence from ancient and experimental
hybrids. Science 272, 741-745.

Roberge C, Normandeau E, Einum S, guderley h, Bernatchez L (2008) Genetic
consequences of interbreeding between farmed and wild Atlantic salmon: insights
from the transcriptome. Molecular Ecology 17, 314-324.

Schwarzbach AE, Rieseberg LH (2002) Likely multiple origins of a diploid hybrid
sunflower species. Molecular Ecology 11, 1703-1715.

Seehausen O, Takimoto G, Roy D, Jokela J (2008) Speciation reversal and biodiversity
dynamics with hybridization in changing environments. Molecular Ecology 17,
30-44.

Tallmon DA, Luikart G, Waples RS (2004) The alluring simplicity and complex reality
of genetic rescue. Trends in Ecology and Evolution 19, 489-496.

Taylor EB, Boughman JW, Groenenboom M, et al. (2006) Speciation in reverse:
morphological and genetic evidence of the collapse of a three-spined stickleback
(Gasterosteus aculeatus) species pair. Molecular Ecology 14, 343-355.

Templeton AR (1986) Coadaptation and Outbreeding Depression. In: Conservation
Biology, The Science of Scarcity and Diversity (ed. Soule ME), pp. 105-116.
Sinauer Associates, INC, Sunderland, Massachusetts.

Turelli M, Barton NH, Coyne JA (2001) Theory and speciation. Trends in Ecology and
Evolution 16, 330-343.

58
Turelli M, Orr HA (2000) Dominance, epistasis and the genetics of postzygotic isolation.
Genetics 154, 1663-1679.

Tymchuk WE, Sundstrom LF, Devlin RH (2007) Growth and survival trade-offs and
outbreedind depression in rainbow trout (Oncorhynchus mykiss). Evolution 61,
1225-1237.

Vittor BA (1971) Effects of the environment on fitnbess-related life history character in
Tigriopus californicus Dissertation, University of Oregon.

Willett CS (2006) Deleterious epistatic interactions between electron transport system
protein-coding loci in the copepod Tigiropus californicus. Genetics 173, 1465-
1477.

59
Chapter 2

Long-term experimental hybrid swarms between nearly incompatible Tigriopus
californicus populations: Fitness recovery and assimilation by the superior
population

CHAPTER 2 ABSTRACT

For the intertidal copepod Tigriopus californicus, outbreeding depression for a
variety of fitness measures is typically observed in early-generation interpopulation
hybrids . We examined both controlled crosses and long-term experimental hybrid
populations. For populations from Playa Altamira (PA) and Punta Morro (PM) in Baja
California, Mexico. F1 and F2 hybrids showed large and significant declines in
survivorship compared to parentals. Reciprocal backcrosses produced no offspring. For
long-term studies, four treatment groups were initiated: 100% PA, 100% PM, 50% PA:
50%PM, and 80%PA: 20%PM. Replicates were surveyed at regular intervals for
morphometric, census and fitness measures. All PA and 80PA:20PM treatments went
extinct within the first few months of experimental culturing but 50:50 and PM parental
treatments persisted throughout the duration of the experiment. Fitness of the 50:50
treatment showed declines relative to the PM parental, but this was followed by rapid
recovery and surpassing of PM fitness by month 9. At month 21, all 50:50 replicates
declined in survivorship below the PM parent. Hybrid morphological measures
decreased over time but months 15 through 21 showed fewer significant differences

60
between treatments. Genotyping of diagnostic microsatellites indicated genetic swamping
by the PM population. This adds support to previous evidence that hybrid breakdown in
early generations may be temporary, and suggests that in the case of nearly incompatible
populations with different fitness levels, only the more fit parental population may
persist.

CHAPTER 2 INTRODUCTION

Human activity is increasingly resulting in disturbance of natural habitats as well
as the relocation of species by both intentional and accidental means. Invasive species
are now seen to have significant consequences for changes in biodiversity (Bleeker et al.
2007, Gilchrist and Lee 2007), given their potential for the exchange of genetic material
with native species. Additionally, the indirect effects of hybridization and introgression
have been considered to be a threat to the biodiversity of native species (Barilani et al.
2005, Halis and Morley 2005). Anthropogenic or natural causes may result in habitat loss
that brings together formerly allopatric species (Allendorf et al. 2001), and the collapse of
multispecies assemblages into hybrid swarms may result in extinction (Rhymer and
Simberloff 1996, Seehausen et al. 2008). This phenomenon has been frequently referred
to as genetic swamping, where genes of one species increase in frequency from the initial
hybridization event until the genetic integrity of the second species is compromised
(Bleeker et al. 2007, Childs et al. 1996, Kothera et al. 2007). A loss of biodiversity due

61
to swamping may be a particular danger when the species of concern comes into contact
with a more common congener (Ellstrand and Elam 1993, Levin et al. 1996).
There are many factors that can impact the fitness of hybridizing populations
including genetic divergence, reproductive compatibility and the fitness of each
contributing parental group relative to each other. Experimental work often focuses on
the effects of hybridization on early-generation crosses (Edmands 2007). Hybrid fitness
can be difficult to predict because the mixing of gene pools, followed by generations of
recombination, can result in the formation of both beneficial and detrimental gene
interactions. In some cases, hybrids may exhibit an increase in fitness termed heterosis or
hybrid vigor, but hybridization may also result in a decrease in fitness known as
outbreeding depression. This might occur in F1 hybrids due to the disruption of local
adaptation, underdominance or epistatic interactions. Fitness declines may also be
delayed until the F2 or backcross generations, when deleterious heterozygote x
homozygote or homozygote x homozygote interactions arise (Lynch 1991, Turelli and
Orr 2000).
The mixing of the two previously isolated populations may result in a hybrid
swarm, in which overlapping hybrid generations coexist in the same population. Over
time many different outcomes are possible including lasting outbreeding depression,
stabilization, assimilation, or the creation of superior recombinant genotypes. In a
previous study, using two moderately incompatible Tigriopus californicus populations
(Chapter 1), fitness and morphology of hybrid swarms were monitored over many
generations. Surprisingly, after 15 months of free mating, hybrid fitness surpassed
midparent fitness. The current study tests whether the same outcome occurs when the

62
experiment is performed on a pair of populations with greater genetic divergence, higher
incompatibility in early generation hybrids and greater fitness differences under lab
conditions.
Tigriopus californicus is an excellent system for studying experimental
hybridization because it has a short generation time (~23 days) and is easily reared in the
lab. This species inhabits rocky intertidal outcrops extending from Alaska down to
central Baja California, Mexico. Despite a seemingly high potential for dispersal,
populations are genetically differentiated over short geographic distances, with
mitochondrial DNA differences that range from 0.2 to 23% (Burton and Lee 1994,
Edmands 2001). We used nearly incompatible T. californicus populations from Punta
Morro and Playa Altamira in Baja, California, Mexico. These two populations show
approximately 21% mitochondrial cytochrome oxidase I divergence (Edmands 2001) and
partial incompatibility in that only one reciprocal cross will produce viable, fertile
offspring. It could be argued that there are grounds for calling theses two populations
sibling species (Ganz and Burton 1995). However, there is no evidence for prezygotic
isolation between these populations, and viable F1 and F2 offspring are produced for one
of the two reciprocal crosses (Ganz and Burton 1995). Many studies have shown that
interpopulation crosses result in F1 heterosis while F2 hybrids show reduced fitness (e.g.
Burton 1986, Burton 1987, Burton 1990ab, Edmands 1999), but the population pair used
in this study does not conform to previous pattern and, instead, exhibits severe fitness
reductions in both F1 and F2 hybrids. We assessed the outcome of hybridization for two
different mixed-population starting ratios over the course of 21 months of free mating (a
maximum of about 27 generations). Patterns of fitness, morphology and genetic

63
composition were assayed at 3-month intervals. Parental populations differed greatly in
their ability to thrive in the laboratory environment and, unlike the outcome of the
previous experiment using less diverged populations, the final outcome of hybridization
was either swamping by the superior population or extinction of the entire swarm.

METHODS

Population Sampling
Populations were sampled from two locations in Baja California, Mexico: Punta
Morro (PM, 31° 52’ N, 116° 40’ W) and Playa Altamira (PA, 28° 32’ N, 114° 5’ W) in
May 2003 (animals used for controlled crosses) and in June 2004 (animals used for long-
term hybrid swarms). Samples were maintained as mass cultures in 400 ml beakers with
filtered seawater (37 μm) containing finely ground Spirulina (0.2 mg/ml) and housed in a
20ºC incubator with a 12 h light: 12 h dark cycle.

Tigriopus Biology
The reproductive biology of Tigriopus has been well-documented (Egloff 1966,
Vittor 1971). Mating and reproduction occurs year round. Adult males use their
antennae to clasp virgin females and mate guard them until they are sexually mature.
Males have multiple matings while females mate only once and store sperm to fertilize
multiple broods of offspring (Burton 1985). Individual lifespan may be as long as 95 days
and with females producing up to 20 clutches of eggs. Clutch size varies from less than

64
10 nauplii to over 100 (pers obs, Vittor 1971) and minimum generation time is
approximately 23 days at 20°C (Burton 1987).

Controlled Crosses
Cultures were maintained in the laboratory for 12 months before controlled
crosses began. All crosses were conducted using coarsely filtered sea water (37 μm)
containing 0.2g ground Spirulina per liter. Virgin females and sexually mature males
were collected by pipeting clasped pairs onto a piece of filter paper, and using fine needle
probes to tease males and females apart. Control crosses (PA x PA and PM x PM) and
hybrid crosses (PA female x PM male and PM female x PA male) were established by
uniting 5 virgin females and 5 mature males in each Petri dish, with a total of 30-50 pairs
for each of the four crosses. These mating dishes were checked every 2-3d. When
females with egg sacs were observed, they were isolated and placed individually into new
petri dishes. These dishes were then examined every 2-3d. When the first clutch of
larvae hatched, the mother was transferred to a new dish and the number of live larvae
was counted by pipeting each larva into a new dish. These dishes were set aside and the
number of survivors was counted 14d later. Subsequent clutches of larvae were collected
and all larvae for a particular cross type (e.g. PA x PA F1) were pooled into a 500 ml
beaker. Beakers were monitored every 2-3d for the appearance of clasped pairs. Pairs
were dissected apart and males and females placed in mating dishes for either F2 crosses
(F1 x F1) or backcrosses (F1 x parental). Again, dishes were monitored every 2-3d and
females with egg sacs were transferred to their own individual Petri. The number of

65
larvae in the first clutch was counted and the number of survivors was recounted 14d
later. In this way, generations were made discrete.

Long-Term Hybrid Swarms
Four different culture treatments were set up for this experiment: 100% PA,
100% PM, 50% PA:50%PM, and 80%PA:20%PM. Five replicates per treatment were
each initiated by placing 500 gravid females in 1000 ml beakers containing 800 mls
culture medium (400 mls Platymonas algal culture and 400 mls Monochrysis algal
culture, supplemented with 0.16 g finely ground Spirulina and Tetramin flakes). Beakers
were housed together in one incubator at 20°C set to a 12h light: 12 h dark cycle. Once
every two weeks a 50% culture medium change was performed. At the same time
beakers were also rotated within the incubator. Every 3 months a census estimate was
taken for each replicate. This was performed by pouring the contents of an entire beaker
into a 1L plastic bottle. The bottle was gently inverted several times to evenly distribute
copepods, after which 200 ml of culture was poured into a 600 ml transparent Gladware
container. A light box was used for visual assistance in sorting copepods into males,
females with eggs, pairs and subadult categories using a Pasteur pipet. Each category
was counted and all individuals were returned to source beakers. Every 3 months, up to
20 gravid females and 20 mature males were removed from each replicate beaker and
were used for morphometric assays. Females were also used for fitness assays. All
copepods were returned to their source beakers after assays were completed. Replicates
were maintained for up to 30 months, at which point 20 males and 20 females from
surviving beakers were frozen for later molecular analyses.

66

Fitness Assays
At each 3 month interval, 20 gravid females were sampled from each replicate
and isolated into individual Petri dishes containing 11 ml filtered seawater supplemented
with ground Spirulina and Tetramin flakes. Females with red egg sacs (red eggs being
more mature and therefore closer to hatching) were preferred to those whose eggs were
still green in color. Each dish was monitored once daily until eggs hatched. On the day
of hatching, 10 larvae per clutch were pipetted into a new dish with fresh seawater culture
medium. Fourteen days later, individuals in each dish were counted to determine
survivorship.

Morphological Assays
Morphometric measurements were taken from digital images of adult copepods
following procedures in Edmands and Harrison (2003). At each 3-month interval, up to
20 females and 20 males were randomly chosen from each replicate. All measurements
were done at a magnification of 32X using a Leica MZ12 dissecting microscope. Digital
images were captured and morphological measurements were taken using Optimas 5.2.
Absolute size was calibrated using a stage micrometer. Eight measurements were taken
for males: cephalothorax length (CTL), cephalothorax width (CTW), urosome length
(UL), urosome width (UW), telson width (TW), caudal seta length (CSL), antennule
width (AW) and clasper width (CLW). Four of the same measurements were taken for
females (CTL, CTW, UW and AW). Egg sac length and area was also measured for each
female. Every three months up to 40 individuals were scored from each replicate.

67

Microsatellite Assays
Three microsatellite loci (Harrison et al. 2004) were found to be diagnostic for
populations PA vs. PM. These three loci were screened in hybrid replicates at months 18
and 30. DNA was extracted from individual copepods using the lysis protocol described
in Edmands et al. 2005. Individual copepods were incubated in 50 μl lysis buffer at 65°C
for 1 hour followed by 100°C for 15 minutes. Polymerase chain reactions were carried
out in 12 μl volumes containing 0.5 μl template DNA, 0.25 μM fluorescently labeled
forward primer, 1 μM reverse primer and 2.5 mM MgCl
2.
Temperature cycling was as
follows: 5 min denaturation at 94°C; 35 cycles of 30 s at 94°C, 35 s at 55°C, and 30 s at
72°C; 5 min at 72°C. Fluorescently labeled PCR products were run on a Beckman-
Coulter CEQ 8000 Capillary Sequencer according to commercially recommended
protocols. Allele sizes were compared to the manufacturer-produced 400 b.p. size
standard and scored by eye.

Statistical Analyses
Analyses of morphological and fitness characters within and between
experimental population treatments were done using Statistica 7.1 (StatSoft, Tulsa, OK).
Nested analysis of variance (ANOVA) and multivariate analysis of variance (MANOVA)
were used to quantify differences in measures among the different treatment types. When
appropriate, Bonferroni post hoc tests were utilized to determine the statistical
significance between group means. Principle components analysis was performed on all
morphometric measures for each time point using Statistica 7.1.

68
For each locus that appeared to be fixed for PM alleles, a binomial probability
calculator (http://faculty.vassar.edu/lowry/binomialX.html) was used to determine the
power of detecting PA alleles (1-β) if they were present at frequencies observed in other
replicates or at other loci.

RESULTS

Controlled Crosses
Fitness (Hatching and Survivorship)
Proportional clutch size (Figure 2-1a) showed large (77-99%) and significant
(p<0.0001 for both, planned linear contrasts) declines from both parental groups in the F1
and F2 cohorts. Both reciprocal backcross cohorts produced no offspring. Of the
individuals that did hatch (3 clutches out of 63), outbreeding depression was not observed
in the survivorship counts of the F1 and F2 cohorts (Figure 2-1b). Planned linear
contrasts showed that differences between F1 and midparent as well as F2 and midparent
survivorship were not significant (p = 0.670 and p=0.962).

Long-Term Hybrid Swarms
Census
Overall, census counts for individual replicate beakers showed large fluctuations
between time points (Figure 2-2). Census averages included dead replicates with a count
of zero. By month 12, all PA parental replicates as well as all 80PA:20PM replicates had

69
Figure 2-1. Proportional deviation from midparent values for clutch size (a) and
survivorship (b) in two-generation controlled crosses. Values are averages among
individual clutches relative to the midparent assayed in the same generation,
+
1 standard
error. Hybrid individuals are offspring of PMO females crossed with PA males, as the
reciprocal cross produces no offspring. The dashed line indicates the additive expectation
for each cohort.

(a)
Clutch Size
-1.2
-1
-0.8
-0.6
-0.4
-0.2
0
0.2
Proportional Deviation from Midparent
PA PA-Backcross PM-Backcross PM
F1
F2

(b)
Survivorship
-1.2
-1
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
Proportional Deviation from Midparent
PA PA-Backcross PM-Backcross PM
F2
F1

70
Figure 2-2. Mean census counts among all five replicates for each treatment over 21
months of free mating. One replicate from PA population remained at month nine but
died out before month twelve, while 80PA:20PM replicates did not survive past month
six. Error bars represent one standard error.

-60
40
140
240
340
440
540
3 6 9 12 15 18 21
Month
Individuals/200 ml
PA
PM
50:50
80:20

71
gone extinct. When the total census taken for a replicate at any given point was
compared to the census count of the same replicate during the following month, there was
a significant correlation (r = 0.34, p = 0.006). Census counts from both hybrid swarm
treatments were lower than PM means at month 3. All hybrid replicates, with the
exception of 50:50 replicate 3, also had census counts below midparent values at month
3. At months 6 through 18 at least two surviving replicates have census counts greater
than the PM replicate mean. At month 18 replicates 4 and 5 exceed the superior PM
parent mean by 4 and 5 fold but, at month 21, hybrid census counts fall back below those
for PM.

Survivorship
Nested ANOVA was run for treatments at each time point. After a Bonferroni
post-hoc test both hybrid treatments had survivorship significantly below the PM superior
parent for months three and six (Table 2-1). Treatments were not significantly different
for months nine, twelve and eighteen. At month 15, all PA parental and 80A:20M
replicates had died out and the 50:50 treatment had significantly greater survivorship than
the PM parentals (Figure 2-3). By the end of the experiment, at month 21, the 50:50
treatment had significantly lower survivorship than the superior PM population.
Survivorship of individual hybrid replicates was also compared to the superior
parent. A one-way ANOVA was performed with all parental replicates of the same
treatment grouped together. Bonferroni post-hoc tests were used to determine whether
hybrid replicates differed in survivorship from parental treatments. In all significant
comparisons, hybrid replicates had lower survivorship than the superior parent, but, for

72
Table 2-1. Survivorship comparisons among treatments by nested ANOVA and
Bonferroni post-hoc at each time point. For each month, the difference between treatment
means is shown. Significant p-values (p < 0.05) are indicated in bold. Dashes indicate
comparisons that were not done because one or both treatments died out.

Month PM - PA
PA -
50:50
PA -
80:20
PM -
50:50
PM -
80:20
50:50 -
80:20
3 0.4260 -0.1857 -0.0769 0.2403 0.3491 0.1088
6 0.2353 -0.0169 0.1446 0.2184 0.3799 0.1615
9 -0.1594 -0.0177 - -0.1771 - -
12 - - - -0.0022 - -
15 - - - -0.0827 - -
18 - - - 0.0225 - -
21 - - - 0.2525 - -

73
Figure 2-3. Mean proportional deviation of each hybrid swarm replicate from the
superior parent value over 21 months of free mating. Error bars show one standard error.

Survivorship
-0.9
-0.7
-0.5
-0.3
-0.1
0.1
0.3
3 6 9 12 15 18 21
Month
Proportional
deviation from
superior parent
50/50_1
50/50_2
50/50_3
50/50_4
50/50_5
80/20_1
80/20_2
80/20_3
80/20_4

74
the 50:50 treatment, more replicates showed equivalent survivorship to the superior
parent. For the 80A:20M treatment, all three replicates were significantly less than the
superior parent at month 3 (p = 0.023, p = 0.018 and p < 0.0001) , and one out of two was
less than the superior parent at month 6 (p < 0.0001). After month six, 80A:20M
replicates crashed and survivorship could not be assayed. At both months 3 and 6 one
out of four of the 50:50 treatment replicates was significantly less than the overall PM
mean (p < 0.001, p < 0.0001). From month nine through 18 all 50:50 replicates were
equivalent to PM for survivorship. At month 21 one 50:50 replicate had lower
survivorship (p = 0.015).

Morphological Assays
Nested ANOVA was performed at each time point to assess morphological
variation (Statistica 7.1), with treatment means compared to the superior parent (PM). For
females (Table 2-2) at the first timepoint (3 months) both the PA treatment and the 50:50
treatment were significantly smaller for 4 characters (CTL, CTW, ESA and ESL). At the
same timepoint, the 80:20 treatment was smaller for one of these characters (CTL) but
larger for two of them (ESA and ESL). At the final timpoint (21 months), both the PA
and 80:20 treatments had died out, and the 50:50 treatment had become smaller for one
character (UL) and larger for two characters (ESA and ESL). For the PM and 50:50
treatments which persisted until month 21, egg sac measures ESA and ESL decreased
over time. For the 50:50 and PM treatments, the changes observed in female morphology
between months 3 and 21 were all statistically significant with the one exception being
ESA for the 50:50 treatment.

75
Table 2-2. Mean phenotypic values for female morphometric characters for all control and hybrid swarm treatments. Units are in
millimeters. Treatment means significantly different from PM according to nested ANOVA and Bonferroni post-hoc tests are
indicated in bold (α = 0.05). Values significantly greater than PM are marked with a + sign while those significantly less than PM are
marked by a - sign. Standard errors among replicates are in parentheses with the exception of the PA treatment at month 9 where only
one replicate remained.

Month CTL CTW UW AW ESA ESL
3 PM 0.5270(0.0052) 0.3447(0.0043) 0.1708(0.0026) 0.0440(0.0012) 0.1084(0.0075) 0.4954(0.0171)
PA 0.4797(0.0054)- 0.3323(0.0241)- 0.1631(0.0032) 0.0425(0.0033) 0.1033(0.0147)- 0.4913(0.0527)-
50:50 0.5037(0.0079)- 0.3313(0.0082)- 0.1669(0.0046) 0.0409(0.0004) 0.0930(0.0044)- 0.4579(0.0125)-
80:20 0.5012(0.0099)- 0.3373(0.0078) 0.1609(0.0008) 0.0408(0.0026) 0.1276(0.0132)+ 0.5395(0.0306)+

6 PM 0.4956(0.0075) 0.3137(0.0044) 0.1613(0.0038) 0.0450(0.0024) 0.0956(0.0038) 0.4831(0.0111)
PA 0.5007(0.0077) 0.3369(0.0119)+ 0.1650(0.0090) 0.0419(0.0024)- 0.0801(0.0034) 0.4285(0.0075)-
50:50 0.5059(0.0078) 0.3299(0.0103) 0.1683(0.0109) 0.0423(0.0005)- 0.1124(0.0084)+ 0.5085(0.0225)
80:20 0.4931(0.0004) 0.3249(0.0047) 0.1604(0.0049) 0.0431(0.0011) 0.1068(0.0015)+ 0.4977(0.0022)

9 PM 0.5127(0.0047) 0.3392(0.0062) 0.1681(0.0023) 0.0433(0.0006) 0.1268(0.0167) 0.5477(0.0404)
PA 0.5063 0.3399 0.1609 0.0399 0.1381+ 0.56+
50:50 0.5324(0.0174)+ 0.3514(0.0223)+ 0.1685(0.0096) 0.0410(0.0035) 0.1198(0.0103) 0.5120(0.0180)

12 PM 0.5206(0.0077) 0.3441(0.0063) 0.1600(0.0023) 0.0379(0.0032) 0.1142(0.0083) 0.4660(0.0403)
50:50 0.4861(0.0169)- 0.3178(0.0051)- 0.1506(0.0005)- 0.0416(0.0006) 0.0821(0.0110)- 0.4184(0.0246)-

15 PM 0.5034(0.0150) 0.3292(0.0106) 0.1542(0.0057) 0.0421(0.0011) 0.0931(0.0049) 0.4550(0.0102)
50:50 0.5068(0.0169) 0.3290(0.0098) 0.1565(0.0032) 0.0412(0.0008) 0.0952(0.0144) 0.4516(0.0326)

18 PM 0.5008(0.0098) 0.3288(0.0070) 0.1377(0.0040) 0.0352(0.0004) 0.0938(0.0041) 0.3847(0.0096)
50:50 0.5003(0.0075) 0.3250(0.0063) 0.1408(0.0016)+ 0.0361(0.0002) 0.0914(0.0063) 0.3794(0.0161)

21 PM 0.4996(0.0084) 0.3231(0.0060) 0.1453(0.0044) 0.0330(0.0007) 0.0721(0.0072) 0.3247(0.0184)
50:50 0.4897(0.0052) 0.3153(0.0074) 0.1383(0.0015)- 0.0322(0.0006) 0.0910(0.0070)+ 0.3800(0.0233)+

76
For males at month 3 (Table 2-3) all measurements that were significantly different from
the superior parent were smaller. By month 21, when only the 50:50 and PM treatments
remained, the only significant difference from PM was smaller antennule width (AW).
With the exceptions of antennule width and urosome width (UW), there was an overall
trend across treatments for male morphological measurements to increase over time.
These increases were significantly different between months 3 and 21 for 5 out of 6
characters (everything but telson width).
Nested ANOVA was performed across all time points to assess morphological
variation. A multivariate test indicated a significant effect of both treatment and replicate
for females and males. Both month and treatment had a significant effect on all four
morphological characters measured in females as well as the two egg sac measurements.
Out of the eight male morphological characters, all eight showed a significant month
effect and seven differed significantly among treatments. One-way MANOVAs were run
for each treatment by month for all measurements (Table 2-4). Fourteen out of eighteen
possible female treatment groups returned a significant p-value (Wilks’ test, p<0.05).
Twelve of those were highly significant (p ≤ 0.0001). For male replicates 13 out of 17
different treatment groups had a significant replicate effect.

Morphological Principle Components Analysis
For each time point, principle components analysis (PCA) was performed on an
average of 80 females from the PM treatment and 58 from the 50:50 treatment (Figures
3-4a and 3-4b). At month three, the first two principle components accounted for
approximately 45.9% and 22.5% of the total phenotypic variation respectively. T-tests

77
Table 2-3. Mean phenotypic values for male morphometric characters in hybrid swarm replicates. Units are millimeters. Treatment
means significantly different from the superior parent (PM) according to nested ANOVA and Bonferroni post-hoc tests are indicated
in bold (α = 0.05). Values significantly greater than PM are marked with a + sign while those significantly less than PM are marked
by a - sign. Standard errors among replicates are in parentheses with the exception of the PA treatment at months 3 and 9 where only
one replicate contained mature males.

Month CTL UL CTW UW TW CSL AW CLW
3 PM 0.4676(0.0036) 0.3203(0.0026) 0.3376(0.0017) 0.1578(0.0027) 0.0341(0.0004) 0.7224(0.0152) 0.0495(0.0010) 0.0689(0.0012)
PA 0.4659 0.3234 0.3295 0.1547 0.0319 0.6813 0.0492 0.063-
50:50 0.4453(0.0156)- 0.2992(0.0195) 0.3105(0.0114)- 0.1422(0.0055)- 0.0293(0.0032) 0.6939(0.0398)- 0.0457(0.0020) 0.0645(0.0032)
80:20 0.4515(0.0025) 0.3153(0.0063) 0.3281(0.0079) 0.1516(0.0027) 0.0338(0.0008) 0.6992(0.0155) 0.0513(0.0015) 0.0657(0.0004)

6 PM 0.4551(0.0059) 0.3156(0.0053) 0.3179(0.0009) 0.1395(0.0093) 0.0290(0.0055) 0.7113(0.0407) 0.0507(0.0003) 0.0714(0.0038)
PA 0.4919(0.0068)+ 0.3416(0.0004) 0.3381(0.0111) 0.1581(0.0047) 0.0365(0.0018) 0.7425(0.0537) 0.0511(0.0029) 0.0648(0.0010)
50:50 0.4671(0.0046) 0.3324(0.0023) 0.3179(0.0048) 0.1483(0.0042) 0.0352(0.0010) 0.6905(0.0303) 0.0461(0.0015) 0.0686(0.0016)
80:20 0.4752(0.0092) 0.3507(0.0034)+ 0.3259(0.0052) 0.1522(0.0011) 0.0339(0.0019) 0.7296(0.0243) 0.0478(0.0006) 0.0619(0.0026)

9 PM 0.4770(0.0082) 0.3460(0.0045) 0.3363(0.0108) 0.1564(0.0020) 0.0370(0.0006) 0.7677(0.0197) 0.0521(0.0018) 0.0688(0.0014)
PA 0.4527 0.3615 0.3109- 0.1277- 0.0372 0.7626 0.039- 0.0575-
50:50 0.49310(0.0011) 0.3555(0.0060)+ 0.3592(0.0024)+ 0.1621(0.0042) 0.0370(0.0003) 0.7600(0.0221) 0.0502(0.0002) 0.0723(0.0013)

12 PM 0.4865(0.0043) 0.3428(0.0070) 0.3360(0.0030) 0.1506(0.0022) 0.0354(0.0006) 0.7649(0.0184) 0.0484(0.0019) 0.0696(0.0014)
50:50 0.4642(0.0066)- 0.3316(0.0055)- 0.3268(0.0122) 0.1497(0.0040) 0.0323(0.0009)- 0.7190(0.0163)- 0.0482(0.0007) 0.0666(0.0020)-

15 PM 0.4814(0.0061) 0.3363(0.0054) 0.3351(0.0071) 0.1499(0.0035) 0.0349(0.0009) 0.7633(0.0125) 0.0485(0.0016) 0.0688(0.0009)
50:50 0.4766(0.0045) 0.3425(0.0145) 0.3310(0.0022) 0.1453(0.0019) 0.0335(0.0002) 0.7530(0.0051) 0.0473(0.0021) 0.0658(0.0015)-

18 PM 0.4691(0.0061) 0.3374(0.0027) 0.3245(0.0030) 0.1313(0.0024) 0.0307(0.0004) 0.7637(0.0193) 0.0457(0.0012) 0.0634(0.0011)
50:50 0.4706(0.0046) 0.3460(0.0028)+ 0.3265(0.0098) 0.1267(0.0028)- 0.0313(0.0005) 0.7785(0.0150) 0.0444(0.0003)- 0.0626(0.0019)

21 PM 0.4848(0.0100) 0.3478(0.0086) 0.3354(0.0073) 0.1397(0.0032) 0.0337(0.0012) 0.7899(0.0401) 0.0443(0.0009) 0.0654(0.0015)
50:50 0.4758(0.0068) 0.3479(0.0066) 0.3299(0.0033) 0.1342(0.0019) 0.0328(0.0008) 0.7918(0.0156) 0.0427(0.0007)- 0.0651(0.0016)

78
Table 2-4. Results of multivariate Wilks’ test. Significant tests are indicated in bold.

Month Females Males
F Effect df Error df p F Effect df Error df p
3 PA
1.838 6 11.00 0.1807
PA
one replicate measured
PM
3.067 24 245.41 0.0000 PM
1.663 32 322.44 0.0159
50:50
2.481 24 168.66 0.0004 50:50
3.021 24 145.62 0.0000
80:20
6.328 12 94.00 0.0000 80:20
1.462 16 52.00 0.1506

6 PA
2.455 6 25.00 0.0527
PA
3.006 8 14.00 0.0345
PM
0.805 24 70.98 0.7181
PM
0.608 8 11.00 0.7549
50:50
6.512 18 144.74 0.0000 50:50
1.794 24 116.61 0.0217
80:20
1.410 6 21.00 0.2572
80:20
1.428 8 20.00 0.2452

9 PA
one replicate measured
PA
one replicate measured

PM
3.473 24 168.66 0.0000 PM
2.212 16 94.00 0.0095
50:50
18.676 6 31.00 0.0000 50:50
1.449 8 31.00 0.2160

12 PM
14.631 18 147.56 0.0000 PM
1.634 32 200.74 0.0231
50:50
7.870 12 102.00 0.0000 50:50
2.566 16 70.00 0.0036

15 PM
5.227 24 287.27 0.0000 PM
2.430 32 233.93 0.0001
50:50
10.966 12 88.00 0.0000 50:50
2.679 16 98.00 0.0015

18 PM
3.769 24 315.18 0.0000 PM
4.133 32 304.00 0.0000
50:50
1.974 12 104.00 0.0339 50:50
3.663 16 98.00 0.0000

21 PM
3.634 24 203.55 0.0000 PM
3.743 32 300.31 0.0000
50:50
3.908 12 104.00 0.0001 50:50
1.949 18 96.00 0.0203

79
Figure 2-4. Principle components analysis of 6 morphometric measurements for PM and
50:50 females after 3 months (a) and 21 months (b). At 3 months PC1 and PC2 account
for 45.9% and 22.5% of the variance respectively. At month 21 the first two components
account for 49.0% and 24.3% of the variance. PCA of 8 morphometric measurements for
PM and 50:50 males after 3 months (c) and 21 months (d). At 3 months PC1 and PC2
account for 33.0% and 14.8% of the variance respectively. At month 21 the first two
components account for 29.9% and 17.6% of the variance.

(a)
Females, Month 3
-3.0
-2.0
-1.0
0.0
1.0
2.0
3.0
4.0
5.0
6.0
-4.0 -3.0 -2.0 -1.0 0.0 1.0 2.0 3.0 4.0 5.0
PC1
PC2
PM_1
PM_2
PM_3
PM_4
PM_5
50/50_1
50/50_2
50/50_3
50/50_4
50/50_5

80
Figure 2-4, Continued
(b)
Females, Month 21
-5.0
-4.0
-3.0
-2.0
-1.0
0.0
1.0
2.0
3.0
4.0
-6.0 -4.0 -2.0 0.0 2.0 4.0 6.0
PC1
PC2
PM_1
PM_3
PM_3b
PM_4
PM_5
50/50_3
50/50_4
50/50_5

(c)
Males, Month 3
-4.0
-3.0
-2.0
-1.0
0.0
1.0
2.0
3.0
4.0
5.0
-5.0 -4.0 -3.0 -2.0 -1.0 0.0 1.0 2.0 3.0 4.0 5.0 6.0
PC1
PC2
PM_1
PM_2
PM_3
PM_4
PM_5
50/50_1
50/50_2
50/50_3
50/50_4

81
Figure 2-4, Continued

(d)
21 Months
-2.5
-2.0
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
2.0
2.5
-5.0 -4.0 -3.0 -2.0 -1.0 0.0 1.0 2.0 3.0 4.0 5.0 6.0
PC1
PC2
PM_1
PM_3
PM_3b
PM_4
PM_5
50/50_3
50/50_4
50/50_5

82
performed on principle components scores showed that the treatments were significantly
different from each other for size, PC1 (p < 0.0001). This same pattern occurs at month 6.
Morphology during months 9, 15 and 18 does not differ significantly along PC1 or PC2.
Month 12 was the only time point at which morphology between treatments along both
PC1 and PC2 was significantly different (p < 0.0001, p = 0.006). The pattern observed at
month 21 is more similar to that of month 3, in which treatments differ along only one
principle component, PC2 (p < 0.0001). Variable contributions did not indicate that any
particular measurement had a stronger influence on either PC1 or PC2.
PCA for male morphometric characters was performed on an average of 75 PM
and 54 50:50 individuals at each time point (Figures 3-4c and 3-4d). Across all time
points the combined total variance represented by the first two principle components
ranged from 43.5% to 55.8% with a mean of 48.5%. Months 9, 12, 15, and 18 all showed
significantly different distributions between the 50:50 and PM treatments along one of
the first two principle components (independent t-tests). There was no evidence for
significant differences of size and shape between treatment distributions at months 6 and
21. Male morphology did exhibit significantly different body size and shape at month 3.
As with female measures, variable contributions did not suggest that any particular
morphometric had a relatively large influence over principle component scores.

Microsatellite Identification
At month 18, three loci were genotyped for a small (< 20) number of individuals
from each of 50:50 replicates 3 and 5. In replicate 3 only locus 197 showed the presence
of PA alleles at a frequency of 0.2 (Table 2-5). All alleles scored for replicate 5 were

83
Table 2-5. Microsatellite Allele Frequencies for 50:50 replicates obtained at months 18
and 30. 1-β is the percent chance that a PA allele would be detected at least once given a
frequency of 0.2 at 18 months and 0.03 at 30 months.

Month Replicate locus n Pop (allele) frequency 1-β
3 1555 16 PM (168) 1 97.2
197 10 PM (190) 0.8 89.3
18 PA (202) 0.2
1203 13 PM (202) 1 94.5
5 197 16 PM (190) 1 97.2
1203 14 PM (202) 1 95.6
1202 16 PM (181) 1 97.1
3 197 32 PM (190) 0.97 62.3
PA (202) 0.03
30 1202 33 PM (202) 1 63.4
4 197 34 PM (190) 0.96 64.5
PA (202) 0.04

84
from the PM population. For replicates and loci at month 18 where no PA alleles were
genotyped, sample sizes corresponded to a 94.5 – 97.2 percent chance of detecting at lest
one PA allele if PA alleles were present at a frequency of 0.2. At month 30, loci 1202
and 197 were genotyped for 33 individuals from 50:50 replicate number 3. Locus 1202
was fixed for PM alleles while PA alleles were present at locus 197 at a frequency of
0.03. Only locus 197 was scored for replicate 4 at month 30. PA alleles were present at a
frequency of 0.04. For locus 1202 at month 30, where no PA alleles were genotyped,
ther was a 63.4 percent chance of detecting at lest one PA allele if PA alleles were
present at a frequency of 0.03. No homozygotes for PA alleles were found in any
observed replicate for both months 18 and 30.

DISCUSSION

Controlled crosses
The large and significant F1 and F2 declines in survivorship, as well as the lack of
offspring in reciprocal backcrosses, indicate severe fitness problems in early hybrid
swarm generations. This is similar to the findings of Ganz and Burton (1995) who
demonstrated that only crosses between PM females and PA males produced F1 hybrids
and some F1 hybrids reached maturity and produced F2 but few F2 hybrids reached
maturity. While Ganz and Burton were not able to determine if the F2 were fertile, this
study showed that the few F2 clutches that were produced exhibited survivorship
equivalent to the midparent. The more extreme fitness reduction in backcross hybrids

85
compared to the significant decline in hatching number for F2s, implicates homozygote-
heterozygote interactions as a primary contributor to fitness problems for this extremely
divergent cross. Without the ability to select mates of the same population origin,
combined with the extreme reductions in offspring numbers for hybrid matings, it could
be possible for swarm populations to decrease drastically in numbers and be driven to
extinction simply by forces of demographic stochasticity. This may be what happened in
two of our experimental replicates, but three out of five replicates actually recovered to
superior parental fitness values. In the controlled crosses, survivorship of the few F1 and
F2 clutches that hatched shows no significant declines compared to the expected value.
The small portion of viable offspring that are produced from these hybrid generations
may be those with highly fit recombinant genotypes which are likely a key source of
population recovery within the next few generations. Our data show that even the severe
fitness declines in hybrid survivorship are not enough to drive the population to
extinction, perhaps because Tigriopus is fecund enough to allow for a few highly fit
recombinants to persist each generation. The actual time course of recovery in freely
mating hybrid swarms is difficult to predict due to overlapping generations and fertilized
females that can produce multiple broods throughout their lifespan.

Does rarity alone lead to assimilation or swamping?
Several cases have been reported where the rarer species are considered
threatened by hybridization with a more common species (Childs et al. 1996, Levin et al
1996, Rhymer and Simberloff 1996). In most cases exact numbers of rare and common
species are not known or reported, but in considering native versus alien populations,

86
studies have reported relative abundance (Bleeker et al. 2007), meaning simply that one
species has a larger population size than the second species. In the case of Tigriopus,
rarity alone (defined as an abundance of 20% of the total population) did not lead to
swamping by the more common population. Rather, the chance of genetic swamping
occurring was most likely determined by the compatibility of the two populations as well
as the fitness of each population in the lab environment. In Chapter 1, an experimental
population initiated with a common (80%) type and a rare (20%) type revealed that, after
15 months of free mating, there was no evidence of outbreeding depression for
survivorship and, after 30 months, replicates of the same mixed population represented
both parental populations at the molecular level. However, in this study of highly
divergent populations which show severe outbreeding depression in early generation
hybrids, the addition of the highly fit PM population (20%) to the struggling PA
population (80%) did not result in introgression or fitness recovery. Instead, like the PA
controls, these mixed replicates eventually went extinct. In contrast, 60% of the 50:50
replicates survived to month 21.

Timing of recovery from outbreeding depression
Conclusions about the duration of outbreeding depression in the population
depend on the way that recovery is defined. If we were worried about the species as a
whole going extinct due to outbreeding depression exhibited in hybrids of nearly
incompatible populations, then the data are somewhat encouraging in that recovery to full
parental fitness was attained. Even though hybrid fitness declined again by the end of
month 21, assays performed at month 30, after the duration of the planned experiment,

87
showed that fitness was equivalent to the superior parent. On the contrary, if managers
are concerned with the loss of biodiversity from genetic swamping then this study
suggests that there is a much larger threat when either (1) the swarm is composed of
highly incompatible populations or (2) the hybridizing populations have differential
fitness, and a combination of both factors is likely responsible for the outcome observed.
After nine months of free mating, survivorship of the surviving 50:50 replicate
populations was equal to the superior parent (Fig. 3). There are a few avenues by which
this may have occurred: (1) Positive assortative mating occurred and one population was
selected against so that eventually only parental genotypes of the superior population
remained (2) genetic swamping occurred so that one population’s genes were selected for
while genes from the second population declined in frequency or (3) within the first few
generations of hybridization selection may have chosen highly fit recombinant genotypes
such that oubreeding depression was rapidly purged and the population contained
individuals with a mosaic of the parental populations’ genomes. There are previous
studies showing that Tigriopus does not avoid outbreeding (Ganz and Burton 1995,
Palmer and Edmands 2000) so it is unlikely that scenario number one occurred.
Molecular data offer valuable insight for distinguishing between outcomes two
and three. Although microsatellite data are few, there is a clear signature of genetic
swamping, as PA allele frequencies are much lower than would be expected for all
samples taken, and most individuals are homozygous for PM alleles. Decreases in fitness
due to outbreeding depression in the 50:50 treatment only last for the first few months
and are followed by rapid recovery to superior parent values by month nine. The signal
given by morphology also supports swamping by the superior PM population as there are

88
fewer observed differences between PM and 50:50 for both males and females after
month 12.
If PM was the species of concern and there was a danger of less fit migrants
entering the gene pool, the results of this study would suggest that there is little threat.
Even with a 50% introduction, the PM population persisted in the long term (30 months).
However, when PM individuals only composed 20% of the swarm population, the entire
population was driven to extinction. This is consistent with other studies showing that
rarity increases the chances of extinction. If instead it was actually PA that was the
species of concern, introducing individuals with higher fitness did not alleviate the
problems that the PA population was experiencing in the lab environment.

Wild populations vs. observed patterns of morphology and fitness
Wild hybridization is often detected by the presence of morphological
intermediates (Gompert et al. 2006, Norman 2009, Ureta et al. 2008). However, if
swamping occurs, it becomes more difficult to detect an ancestral hybridization event
with subsequent generations, because individuals with hybrid ancestry may become
indistinguishable in morphology from the superior parent. Four female morphological
characters differed significantly between the parental populations at month three, but two
of those were significantly different at month six and the PA parent did not survive
beyond that point. This makes it difficult to assess whether or not morphological
intermediates persisted throughout the experiment. Morphological measurements for
females decreased over time, just as in Chapter 1. At month 21 50:50 replicates had
larger egg sacs compared to the PM parent, as well as lower survivorship. This is

89
concordant with previous data indicating that, in the lab environment, females of long-
term mixed populations showed higher survivorship but had smaller egg sacs (Chapter 1).
However, the same pattern was not observed for males as it was in the previous study.
Male measurements either increased or remained the same over 21 months.
Principle components analysis at month 21 indicates that female characters
showed similar trends for replicates of the same treatment (Figure 2-4), while PM and
50:50 treatments are significantly different from each other along one principle
component. Treatments took slightly different paths of morphological evolution but
replicates of the same treatment had distributions similar to each other. It may be that
female morphological characters are more deterministic with respect to their population
of origin, and that the balance between selection and drift is different for females versus
males. Males did not show the same pattern where distributions along PCs varied
throughout the experiment. Distributions of male characters for the PA and 50:50
treatments along the first two principle components were both significant at month three
but, by month 21, distributions of the two treatments were similar, in agreement with data
from individual morphological measurements suggesting that hybrid swarms are
equivalent to the superior parent by the end of the experiment (Figure 2-4).
Overall, the morphometric signal for both males and females suggests that 50:50
morphology became more like that of the PM parental over time, with this pattern being
stronger in males. Unlike early hybrid generations in which morphological intermediates
may be evidence of hybridization, populations show few differences from the superior
parent after 21 months. In contrast, even after 30 months there is still a molecular
signature of the original hybridization event, so even though mixed populations look like

90
the PM parent population and have similar fitness values, one would still be able to
perform genetic analyses to determine that the population had descended from a
hybridization event in the past. This result is encouraging if management decisions are
required to preserve the genetic integrity of individual populations or species. Even in
the event of genetic swamping, it may be possible to determine which populations are
descended from hybrids, provided that candidate parental populations can be identified.

Is the outcome of hybridization repeatable?
For managers attempting to restore multiple populations of the same species, it
would be particularly useful to know how often similar hybridization events might result
in the same outcome and how those patterns might differ among taxa. The experimental
trials performed here provide an animal alternative to the known examples of plants that
show repeated hybridization events leading to similar outcomes (Brochman et al. 2000,
Schwarzbach and Rieseberg 2002). These cases may involve specific characteristics such
as chromosomal rearrangements, observed in hybrids of the genus Helianthus, that resist
recombination in the parental species and thus contribute to observed repeatability
between both synthetic lineages (Rieseberg et al. 1996) and natural and synthetic hybrids
(Rieseberg et al. 2003). The overall implication from these studies is that strong
deterministic forces are involved in driving the consequences of hybridization. Studies of
Tigriopus have previously found only partial concordance among replicates of the same
treatments for both molecular (Edmands et al. 2005) and morphological (Chapter 1)
measures. It is important to note, however, that very large effective population sizes (N
e
),
of approximately 2 million, have been estimated for species of Helianthus (Strasburg and

91
Rieseberg 2008). Studies of synthetic hybrids have mainly focused on controlled crosses
descending from a few individuals, but the repeatable outcome observed in natural hybrid
species could be a result of either higher N
e
or stronger selection on chromosomal
rearrangements.
Individual morphometric characters of each experimental replicate in this study
indicate that the path of morphological evolution is not necessarily repeatable. When
morphological measurements are assessed together, replicates have distinct trajectories.
Significant replicate effects were seen within treatments for all treatments after month
nine and replicate effects were significant for the 50:50 treatment for all months.
Significant replicate effects were also observed for morphometric measures and
survivorship for all months except month 18, indicating that the mean of all replicates for
a treatment type may not be indicative of any particular replicate.
Despite the many differences between replicate populations for several
morphometric characters, PCA suggests that, for females, populations of the same
treatment arrived at similar morphologies by month fifteen (Figure 2-4). Additionally,
two of the three 50:50 replicates had surprisingly parallel survivorship measures
throughout the duration of the experiment in that, for all time points, no differences were
detected between the two replicates. In concordance with the Chapter 1 study, the
individual replicates that were able to overcome the threat of extinction were those with
the fitness values greater than the superior parental controls at month 15. This
experiment monitored the 50:50 treatment further (21 months instead of only 15) and
fitness dropped to levels significantly lower than the superior parent but additional data
collected at month 30 revealed mean fitness as equivalent to the superior parent. Further

92
studies that integrate detailed accounts of both molecular and phenotypic data are
warranted in order to resolve questions surrounding determinism and repeatability.
Overall the results from this experiment agree with those of Chapter 1 in that
deterministic forces for morphological and molecular evolution may exist but be
hampered by drift in the small populations used in this experiment, just as drift may
impact natural populations experiencing extreme environmental fluctuations. A stronger
influence of drift compared to selection would be consistent with the little evidence of
local adaptation found (Edmands and Deimler 2004, Hwang et al. unpublished data) as
well as levels of molecular subdivision that exceed measures of quantitative subdivision
(Edmands and Harrison 2003).

Conclusions
Depending on how one defines recovery from outbreeding depression, this study
adds to the growing body of evidence that hybrid swarm populations can persist long
after the effects of severe early-generation outbreeding depression become evident.
Experimental results are also in agreement with the notion that outbreeding depression
may be a transient period in a population’s evolution (Carney et al. 2000, Christiansen
2008, Rieseberg et al. 1996, Templeton 1986). Where the introduction of genetically
divergent individuals may be the only remaining option to bolster a dwindling
population, managers may want to take an extra step of caution and assess the amount of
divergence between the populations of choice, as well as the fitness of the individual
populations in the protected environment. Our results suggest that both of these factors
may have an influence over whether the mixing of populations is likely to result in hybrid

93
persistence or genetic swamping. While we have demonstrated here that the deleterious
effects of outbreeding may be purged over time, with nearly incompatible, highly
divergent populations such as these, the long-term outcome of a hybrid swarm may not
necessarily be the conservation of the desired genetic composition.

95
CHAPTER 2 REFERENCES

Allendorf FW, Leary RF, Spruell P, Wenburg JK (2001) The problems with hybrids:
setting conservaton guidelines. Trends in Ecology and Evolution, 613-622.

Barilani M, Deregnaucourt S, Gallego S, et al. (2005) Detecting hybridization in wild
(Coturnix c. coturnix) and domesticated (Coturnix c. japonica) quail populations.
Biological Conservation 126, 445-455.

Bleeker W, Schmitz U, Ristow M (2007) Interspecific hybridisation between alien and
native plant species in Germany and its consequences for native biodiversity.
Biological Conservation 137, 248-253.

Brochmann C, Borgen L, Stabbetorp OE (2000) Multiple diploid hybrid speciation of the
Canary Island endemic Argyranthemum sundingii (Asteraceae). Plant Systematics
and Evolution 220, 77-92.

Burton RS (1985) Mating system of the intertidal copepod Tigriopus californicus. Marine
Biology 86, 247-252.

Burton RS (1986) Evolutionary consequences of restricted gene flow in the copepod
Tigriopus californicus. Bulletin of Marine Science 39, 526-535.

Burton RS (1987) Differentiation and integration of the genome in populations of the
marine copepod Tigriopus californicus. Evolution 41, 504-513.

Burton RS (1990) Hybrid breakdown in developmental time in the copepod Tigriopus
californicus. Evolution 44, 1814-1822.

Burton RS (1990) Hybrid breakdown in physiological response: A mechanistic approach.
Evolution 44, 1806-1813.

Burton RS, Lee B-N (1994) Nuclear and mitochondrial gene genealogies and allozyme
polymorphism across a major phylogeographic break in the copepod Tigriopus
californicus. Proceedings of the National Academy of Sciences USA 91, 5197-
5201.

Carney SE, Gardner KA, Rieseberg LH (2000) Evolutionary changes over the fifty-year
history of a hybrid population of sunflowers (Helianthus). Evolution 54, 462-474.

Childs MR, Echelle AA, Dowling TE (1996) Development of the hybrid swarm between
pecos pupfish (Cyprinodontidae: Cyprinodon pecosensis) and sheepshed minnow
(Cyprinodon variegatus): A perspective fom allozymes an mtDNA. Evolution 50,
2014-2022.

96
Christiansen FB (2008) Theories of Population Variation in Genes and Genomes
Princeton University Press, Princeton, NewJersey.

Edmands S (1999) Heterosis and outbreeding depression in interpopulation crosses
spanning a wide range of divergence. Evolution 53, 1757-1768.

Edmands S (2001) Phylogeography of the intertidal copepod Tigriopus californicus
reveals substantially reduced population differentiation at northern latitudes.
Molecular Ecology 10, 1743-1750.

Edmands S (2007) Between a rock and a hard place: evaluating the relative risks of
inbreeding and outbreeding for conservation and management. Molecular
Ecology 16, 463 - 475.

Edmands S, Deimler JK (2004) Local adaptation, intrinsic coadaptation and the effects of
environmental stress on interpopulation hybrids in the copepod Tigriopus
californicus. Journal of Experimental Marine Biology and Ecology 303, 183-196.

Edmands S, Feaman HV, Harrison JS, Timmerman CC (2005) Genetic consequences of
many generations of hybridization between divergent copepod populations.
Journal of Heredity 92, 114-123.

Edmands S, Harrison JS (2003) Molecular and quantitative trait variation within and
among populatioins of the intertidal copepod Tigriopus californicus. Evolution
57, 2277-2285.

Edmands S, Harrison JS (2003) Molecular and quantitative trait variation within and
among populatioins of the intertidal copepod Tigriopus californicus. Evolution
57, 2277-2285.

Egloff DA (1966) Ecological aspects of sex ratio and reproduction in experimental and
field populations of the marine copepod Tigrioups californicus Dissertation,
Stanford University.

Ellstrand NC, Elam DR (1993) Population genetic consequences of small population size:
Implications for plant conservation. Annual Review of Ecology and Systematics
24, 217-242.

Ganz HH, Burton RS (1995) Genetic differentiation and reproductive incompatibility
among Baja California populations of the copepod Tigriopus californicus. Marine
Biology 123, 821-827.

Gilchrist GW, Lee CE (2007) All stressed out and nowhere to go: does evolvability limit
adaptation in invasive species? Genetica 129, 127-132.

97
Gompert Z, Fordyce JA, Forister ML, Shapiro AM, Nice CC (2006) Homoploid Hybrid
Speciation in an Extreme Habitat. Science 314, 1923-1925.

Halis RS, Morley K (2005) Genes invading new populations: a risk assessment
perspective. Trends in Ecology and Evolutio 20, 245-252.

Harrison JS, Peterson DL, Swain JR, Edmands S (2004) Microsatellite DNA markers for
the intertidal copepod Tigriopus californicus. Molecular Ecology Notes 4, 736-
738.

Kothera L, Ward SM, Carney SE (2007) Assessing the threat from hybridization to the
rare endemic Physaria bellii Mullican (Brassicaceae). Biological Conservation
140, 110-118.

Levin DA, Francisco-Ortega J, Jansen RK (1996) Hybridization and the extinction of rare
plant species. Conservation Biology 10, 10-16.

Lynch M (1991) The genetic interpretation of inbreeding depression and outbreeding
depression. Evolution 45, 622-629.

Norrmann GA (2009) Natural hybridization in the Andropogon lateralis comples
(Andropogoneae, Poaceae) and its impact on taxonomin literature. Botanical
Journal of the Linnean Society 159, 136-154.

Palmer CA, Edmands S (2000) Mate choice in the fact of both inbreeding and
outbreeding depression in the intertidal copepod Tigriopus californicus. Marine
Biology 136, 693-698.

Rhymer JM, Simberloff D (1996) Extinction by hybridization and introgression. Annual
Review of Ecology and Systematics 27, 83-109.

Rieseberg LH, SInervo B, Linder CR, Ungerer MC, Arias DM (1996) Role of gene
interactions in hybrid speciation: evidence from ancient and experimental
hybrids. Science 272, 741-745.

Schwarzbach AE, Rieseberg LH (2002) Likely multiple origins of a diploid hybrid
sunflower species. Molecular Ecology 11, 1703-1715.

Seehausen O, Takimoto G, Roy D, Jokela J (2008) Speciation reversal and biodiversity
dynamics with hybridization in changing environments. Molecular Ecology 17,
30-44.

Templeton AR (1986) Coadaptation and Outbreeding Depression. In: Conservation
Biology, The Science of Scarcity and Diversity (ed. Soule ME), pp. 105-116.
Sinauer Associates, INC, Sunderland, Massachusetts.

98
Turelli M, Orr HA (2000) Dominance, epistasis and the genetics of postzygotic isolation.
Genetics 154, 1663-1679.

Ureta MS, Cantamutto M, Carrera A, Delucchi C, Poverene M (2008) Natural hybrids
between cultivated and wild sunflowers (Helianthus spp.) in Argentina. Genet
Resour Crop Evol 55, 1267-1277.

Vittor BA (1971) Effects of the environment on fitnbess-related life history character in
Tigriopus californicus Dissertation, University of Oregon.

99
Chapter 3

Environmental stress speeds recovery from outbreeding depression in experimental
hybrid swarm generations of moderately divergent Tigriopus californicus
populations

CHAPTER 3 ABSTRACT

Despite well established hybrid breakdown in early generations, hybrid swarms of
Tigriopus californicus have shown evidence of recovery coupled with extensive
introgression. In this study experimental hybrid swarm replicates between populations
from Royal Palms (RP) and San Diego (SD), California, were initiated with equal
contributions from each source population and maintained for fourteen generations.
Cultures were allowed to mate freely while generations were kept discrete. For both
survivorship and metamorphosis, early generation heterosis was followed by outbreeding
depression and recovery, which occurred up to two generations earlier in the high salinity
treatment. At benign salinity, female hybrids showed evidence of transgressive
segregation for three characters at the end of the experiment: Cephalothorax length
(CTL), egg sac length (ESL) and egg sac area (ESA). Male morphometrics increased
between generations one and thirteen for most treatments, except for high salinity
hybrids, which showed a decrease for almost all measures. Hybrid index scores indicated
that males became more RP-like in nature over time, with a stronger effect in the high
salinity environment. High salinity replicates also displayed stronger repeatability for

100
survivorship and hybrids were more fit than benign replicates when exposed to a novel
stress.

CHAPTER 3 INTRODUCTION

The conservation and protection of species depends on preserving the natural
genetic diversity in populations. This involves efforts to minimize both inbreeding and
outbreeding depression. Yet, studying these patterns in species of concern can be
difficult, especially with species that have long generation times or dwindling population
sizes. Experimental studies of hybrid genotypes have important implications for
conservation biology, as translocation is being implemented as a method of alleviating
inbreeding depression in reduced populations (Allendorf et al. 2001). Efforts to utilize
this type of genetic rescue rely on the addition of new genes to a population in order to
increase the populations’ fitness (Tallmon et al. 2004), but the actual outcome of
hybridization may range from hybrid fitness that exceed the original populations
(Rieseberg et al. 1999) to outbreeding depression (Edmands 2007). While it is difficult to
predict whether mixing of gene pools will increase or decrease fitness in hybrids, the
effective use of conservation methods, such as translocation, will be largely dependant
upon how well we are able assess the potential outcomes of hybridizing populations.
Hybrid fitness is difficult to predict since both beneficial and deleterious
genotypes may be created from the hybridization event. A phenomenon known as
heterosis occurs if hybrids show an increase in fitness compared to parentals. This could

101
be due to the masking of deleterious recessive alleles, overdominance or beneficial
epistasis. Contrastingly, outbreeding depression may occur. F1 outbreeding depression
can be a result of underdominance, overdominance for alleles that decrease
morphological or fitness traits, disruption of local adaptation or epistatic interactions
(Lynch 1991). Often times a reduction in fitness is delayed until the F2, in which hybrid
breakdown can be attributed to recombination and the breakup of coadapted genomic
regions (Turelli and Orr 2000). One study of a North American legume revealed that
outbreeding depression was not evident until the F3 (Fenster and Galloway 2000),
suggesting thatt an extra generation of recombination was necessary to break up tightly
linked coadaptation. Interpopulation hybridization in Tigriopus typically results in F1
heterosis and F2 breakdown correlated with genetic divergence (Burton 1990, Edmands
1999). Beyond the F2, it is possible for fitness to decline with rounds of recombination
that disrupt more tightly linked coadapted complexes, or fitness may increase as selection
promotes favorable hybrid combinations. Several studies have collected evidence of both
nuclear-nuclear and nuclear-mitochondrial coadaptation (Burton 1987, Edmands 1999,
Willett and Burton 2001, Rawson and Burton 2002). Additionally, studies of non-
recombinant backcrosses reveal a deficit of parental homozygotes for the majority of
linkage groups indicating a combination of beneficial dominance and deleterious
homozygote-heterozygote interactions (Harrison and Edmands, 2006).
In order to understand thoroughly the consequences of mixing gene pools, it is
important to study hybridizing populations over multiple generations. Much work has
been done on natural hybrid zones established in the wild (Burke et al. 1998, Rieseberg et
al. 1998, Fitzpatrick and Bradley 2004), but these studies explore hybrid populations that

102
are relatively stable. Because the intertidal copepod Tigriopus californicus has a short
generation time (~23 days in an optimal lab environment), we can observe the trajectories
of experimental hybrid populations and determine the genetic composition of the
population in transitory states of evolution.
A pilot study of long term hybridization (Edmands et al. 2005) began with
backcross individuals from RP and SD populations that were allowed to mate freely for
one year (up to 16 generations). Hybridity in all four replicates increased (from 25% RP
alleles: 75% SD alleles to approximately 46% RP alleles: 54% SD alleles) over multiple
generations, suggesting that hybrid fitness problems in early generations are a weak
barrier to introgression. This finding was supported by an experiment described in
Chapter 1 of this dissertation, where replicate swarms of two divergent populations
persisted through 15 months of free mating and showed fitness measures higher than the
expected midparent value. Chapter 2 showed that two nearly incompatible populations
were not driven to extinction by severe early-generation fitness reductions but, after 21
months, had recovered to represent the superior parental population in morphology and
molecular characteristics.
Because Tigriopus inhabits the highest supralittoral tide pools, in what is known
as the splash zone, these copepods undergo wide fluctuations in environmental
conditions, particularly temperature and salinity. Tidepool temperatures have been found
to vary from 4˚C to 35˚C (Edmands and Deimler 2004) and bouts of evaporation, as well
as inputs of fresh rain water, can cause large fluctuations in salinity. The ranges of these
conditions may vary from population to population, but a short term study of
environmental stress on hybrids found very little evidence for local adaptation (Edmands

103
and Deimler, 2004). Edmands and Deimler did, however, find that F2 breakdown was
partially alleviated under thermal stress. Willett and Burton (2003) showed a substantial
effect of temperature/light regime on selection at the cytochrome c locus in
interpopulation hybrids that is consistent with environmental effects on cytonuclear
epistasis. Little is known about the environmental effects on populations hybridizing for
many generations.
With the impacts of global climate change and habitat fragmentation,
conservation biology should explore cases of adaptation and response to environmental
stress. Stress is an important concern for both conservation biology and evolution
(Bijlsma and Loeschcke 2005, Frankham 2005). Hybridization can contribute to
adaptative evolution in two ways: 1) introgression may result in the transfer of alleles
that increase fitness from one population to another or 2) new evolutionary lineages may
be formed from relatively fit hybrid genotypes (Burke and Arnold 2001). While many
studies find that inbreeding depression is aggravated by stress (Lynch and Walsh 1998)
other findings suggest outbreeding depression may be alleviated by stress (Armbruster et
al, 1997, Hoffman and Parsons 1991, Edmands and Deimler 2004). Additional studies
have suggested that novel environmental conditions might supply hybrid genotypes with
new niches for which they are better fit than parental genotypes (Rieseberg et al. 1999).
Hybridization may result in generating extreme phenotypes (in a positive or negative
direction) compared to parentals, known as transgressive segregation (Rieseberg et al.
1999, Seehausen 2004). In the wild, hybrid species often occur in habitats more extreme
than those of congeners (Rieseberg et al. 2003, Gross et al. 2004) and recent studies have
revealed that extreme phenotypes can be generated from massive recombination events.

104
Transgressive segregation has been demonstrated to occur largely in plants (Lexar et al.
2003, Johansen-Morris and Latta 2006, Johnston et al. 2004) although it has also been
observed in a fish and other taxa (Albertson and Kocher 2005, Ranganath and Aruna
2003, Rieseberg 1999). These extreme phenotypes might adapt well to novel
environments which would otherwise be considered stressful to the parental population,
and while studies focusing on plants have indicated that hybrid speciation may be
especially likely under stressful or novel conditions (Johnston et al. 2004, Rieseberg et al.
2007), less is known about its potential in animals.
In this study we monitored experimental hybrid swarm populations over thirteen
discrete generations of free mating. Because hybridization was not forced and
individuals were allowed to choose their mates freely, we sought to better mimic a
natural hybridization event in the wild while still partitioning the effects of hybridization
at specific generations. Prior to this research, a pilot study determined that T. californicus
growth rate is decreases under high salinity conditions. Experimental replicates were
exposed to both high salinity and benign conditions to assess the effects of a stressful or
novel environment. The magnitude and duration of outbreeding depression was
determined by collecting measurements of morphological characters and multiple fitness
components.

105
METHODS

The Tigriopus californicus study system
There are several characteristics that make T. californicus an excellent model for
studies of long-term experimental hybridization. Population samples can be easily
maintained in culture or used for studies that require breeding manipulation. Mature
males as well as gravid and virgin females can be easily identified because males use
their enlarged claspers to mate-guard females until they are mature enough to be
fertilized. T. californicus has a short generation time (approximately 23 days at 20˚C;
Burton 1987) which makes it appropriate for long-term hybridization studies.
Studies using a variety of molecular markers, including allozymes, DNA
sequences and microsatellites, show that, despite potential for high gene flow,
populations of T. californicus within as little as 500 m remain genetically differentiated
(Burton 1997, Edmands 2001, Edmands and Harrison 2003). Mitochondrial COI has
been assayed in populations extending from Alaska to Baja California, Mexico and show
that populations differ from 0.2 to 23% (Burton and Lee 1994, Edmands 2001). Genetic
distance is correlated with geographic distance, with decreased interpopulation genetic
divergence in the northernmost regions (Edmands 2001).

Population Sampling
Populations were sampled from two southern California locations, Royal Palms,
CA (RP, 33° 42’ N, 118° 19’ W) and San Diego, CA (SD, 32° 45’ N, 117° 15’ W) in
December 2005. These two populations show approximately 18% mitochondrial

106
cytochrome oxidase I divergence (Edmands 2001). Samples were maintained as mass
cultures in 400 ml beakers with 350 ml filtered seawater (37 μm) and algal food
supplements and housed in a 20ºC incubator with a 12 h light: 12 h dark cycle.
Long-Term Hybrid Swarms
Three different culture treatments were initiated (100% RP, 100% SD, and 50%
RP:50% SD) by placing 100 gravid females in 400 ml beakers full of filtered seawater
containing 50 mls live Platymonas culture. Each beaker was supplemented with finely
ground Spirulina and Tetramin flakes at a concentration of 0.2 mg/ml. In January 2006
12 replicates of each culture treatment were set up for each of two salinities: 35 ppt
(benign conditions) and 53 ppt (salinity stress). Prior to the setup of the experiment a 14-
day test of algal growth rate was performed to confirm that Platymonas grown under lab
conditions did not have significantly different growth rates in the two chosen salinities. A
pilot study was also done to determine if T. californicus growth and/or development (size
at two weeks) is compromised under high salinity conditions (Figure 3-1). Beakers were
housed together in one incubator at 20 °C set to a 12h light: 12 h dark cycle.
Beakers were monitored each week for the presence of copepodids (juvenile
copepods). Early stage copepods were distinguished from adult females based upon size
and color. If juveniles were present, all adult females were removed from the beaker.
One week later, all adult females were again removed. This was done as a way to double
check that all females of the previous generation were removed. Beakers were then
monitored once a week and, as soon as adult females with eggs were observed, they were
transferred to new beakers with fresh algae-seawater-dry food mixture to start the next
generation. Adult gravid females were transferred to the new generation beaker for the

107
Figure 3-1. Pilot test of growth rate measured as the length of two-week old juveniles
reared in varying salinity. All individuals assayed were from the Royal Palms population.
Mean length of individuals reared in 53ppt seawater is significantly different from the
control treatment of 35 ppt (ANOVA with Bonferroni post hoc test, p < 0.001).

0.35
0.40
0.45
0.50
0.55
0.60
0.65
0.70
0.75
0.80
35 18 53 70
Salinity (ppt)
Length (um)

108
next two or three weeks depending on the presence of absence of new juveniles. When
juveniles were observed in the new beaker, adult females were removed. In this way,
generations were maintained as discrete. Once a week, all beakers were fed and rotated
within the incubator in order to homogenize light exposure.
For selected generations, 20 gravid females and 20 mature males were removed
from each replicate beaker and were used for morphometric assays. Females were also
used for fitness assays. All copepods were either returned to their source beakers after
assays were completed, or were frozen for molecular analysis. Replicates were
maintained for 14 generations which corresponded to 18 months.

Metamorphosis and Survivorship Assays
For each generation scored (generations 1, 2, 3, 5, 7, 9, 11 and 13), 20 gravid
females were sampled from each replicate and isolated into individual Petri dishes
containing 11 mls filtered seawater supplemented with ground Spirulina and Tetramin
flakes. Females with red egg sacs (red eggs being more mature and therefore closer to
hatching) were preferred to those whose eggs were still green in color. Each dish was
monitored three times a week until eggs hatched. Once hatching was observed, 10 larvae
per clutch were pipetted into a new petri dish with fresh seawater culture medium. T.
californicus experiences increased development time when reared under harsh conditions
(Figure 3-1 and pers. obs.). In order to be able to obtain a measure of development time
for such a large number of samples, we recorded two different measures: number of
copepods metamorphosed after one week and presence or absence of nauplii after one
week (presence of nauplii indicating slow development). Seven days after hatching,

109
metamorphosed individuals in each dish were counted and the presence or absence of
nauplii was noted. Fourteen days after hatching, individuals in each dish were counted to
determine survivorship.

Competitive Fitness Assay
After 13 generations of hybridization, the competitive survivorship of remaining
replicates was determined to gain a more comprehensive measure of overall fitness.
Because individuals from different populations are indistinguishable by eye, and hybrid
populations will share alleles with parental populations, it was not possible to complete
different treatments directly against each other. A third, divergent population from Santa
Cruz, CA (36° 57’ N, 122° 03’ W) was used for this competition assay. Females from
selected replicates were isolated until their clutch hatched. In a petri dish containing
55ml seawater of the experimental replicate salinity, one SC clutch and one experimental
clutch were combined. Thirty plates were set up from each treatment for a total of 180
plates. Plates were incubated in a 20ºC incubator with a 12 h light: 12 h dark cycle and
fed once a week by adding 5 drops of algal food mixture (0.2 g finely ground Spirulina
and Tetramin in 100 ml filtered seawater) with a Pasteur pipet. After three weeks, 10
surviving individuals from each plate were frozen for microsatellite identification.
Five diagnostic microsatellites (Loci 1203, 558, 30, 1555, and 228 described in
Harrison et al. 2004) were utilized to identify the population of origin for each sampled
individual after three weeks. DNA was extracted using the lysis protocol previously
described in Edmands et al. 2005. Individual copepods were incubated in 50 μl lysis
buffer at 65°C for 1 hour followed by 100°C for 15 minutes. For each individual, one of

110
the five microsatellites was PCR amplified in 12 μl volumes containing 0.5 μl template
DNA, 0.25 μM fluorescently labeled forward primer, 1 μM reverse primer and 2.5 mM
MgCl
2
. The following temperature cycling conditions were used: 5 min denaturation at
94°C; 35 cycles of 30 s at 94°C, 35 s at 55°C, and 30 s at 72°C; 5 min at 72°C. The use
of multiple loci with different fluorescent labels allowed for pooling of several PCR
products into one sample run. Labeled products were run on a Beckman-Coulter CEQ
8000 Capillary Sequencer according to commercially recommended protocols. Allele
sizes were scored by eye.

Heat Shock Assay
Heat shock assays were performed on generation 14 nauplii. Prior to performing
this heat shock assay a pilot stress test was done to determine that 20 minutes at 36°C
reduced survivorship but was not 100% lethal. In order to also collect data for a separate
experiment designed to measure heritability, one male and two virgin females from a
generation 13 treatment were placed in a petri dish together. One-hundred fifty of these
families were set up for each treatment. Dishes were monitored until males fertilized
both females. When egg sacs were observed, females were isolated to their own dish.
When eggs hatched, 10 nauplii from each clutch were isolated to a new dish. Seven days
later dishes were exposed to a heat shock of 36°C for 20 minutes. Dishes were returned
to incubation temperatures for one hour and survivors were counted.

111
Morphological Assays
Morphometric measurements were taken from digital images of adult copepods
following procedures in Edmands and Harrison (2003). As each sampled generation
matured, up to 20 females and 20 males were randomly chosen from each replicate. All
measurements were done at a magnification of 32X using a Leica MZ12 dissecting scope.
Digital images were captured and morphological measurements were taken using
Optimas 5.2. Absolute size was calibrated using a stage micrometer. Eight
measurements were taken for males (cephalothorax length, cephalothorax width, urosome
length, urosome width, telson width, caudal seta length, antennule width and clasper
width) and four were taken for females (cephalothorax length, cephalothorax width,
urosome width, and antennule width). Egg sac length and area was also measured for
each female.

Statistical Analyses
Analyses of phenotypic and fitness characters within and between experimental
population treatments were done using Statistica 7.1 (StatSoft, Tulsa, OK). Nested
analysis of variance (ANOVA) and multivariate analysis of variance (MANOVA) were
used to quantify differences in measures among the different experimental population
treatments and replicates. When appropriate, Bonferroni post hoc tests were utilized to
determine the statistical significance between group means. To compare each
experimental replicate as well as all replicates of a population type to the midparent
value, we conducted ANOVAs followed by planned contrast tests. Correlations were
performed between combinations of fitness, morphology and variance components.

112
Survivorship and metamorphosis were also analyzed by pooling all replicates of
each treatment together and scoring each offspring (up to 10 from each female) as dead
or alive. These data were categorical and binomial, but the normal distribution provides a
good approximation for the binomial distribution if the interval [μ ± 3σ], where μ=np and
σ= [np(1 – p)]
1/2
, lies between 0 and n (Hedgecock et al. 2007). Hybrid fitness declines
were then quantified by calculating an index of potency, h
p
= Q/L < 1, where Q is twice
the deviation of a hybrid from the midparent value and L is the absolute difference
between the mean fitness values of the two parental controls (Griffing 1990). These
contrasts were estimated from ANOVA followed by appropriate tests (SAS, proc glm).

RESULTS

Survivorship
While generation 1, composed of only parentals and F1s, showed no deviation
from parental means, generation 2, reared in the 35 ppt environment, showed enhanced
fitness most likely due to the presence of early generation hybrids (Figure 3-2). This was
not seen until generation 3 in replicates of the 53 ppt treatment, perhaps because the
formation of additional F1s was necessary to observe a positive deviation from parentals.
At these generations, which showed increased hybrid fitness, two hybrid replicates in
benign salinity were significantly greater than the midparent while one replicate had
significantly lower survivorship (planned contrast) (Table 3-1). Maximum outbreeding
depression occurs earlier in high salinity treatments, as generation 5 was the timepoint of
lowest mean survivorship among 53 ppt hybrid replicates, but replicates reared in benign

113
Figure 3-2. Mean proportional deviation from midparent (± one standard error) for hybrid
replicate survivorship over thirteen discrete generations. (a) 35 ppt treatment (b) 53 ppt
treatment. Asterisks indicate means significantly different from the midparent according
to planned linear contrasts (p < 0.05).

(a)
35 ppt
-0.5
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
Generation
Proportional Deviation from Midparent
1 2 3 5 7 9 11 13
*
*
*
*
*

(b)
53 ppt
-0.5
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
Generation
Proportional Deviation from Midparent
1 2 3 5 7 9 11 13
*
*
*

114
Table 3-1. Proportional deviation from midparent survivorship values for each hybrid swarm replicate. Means significantly different from midparent values (α =
0.05) according to planned linear contrasts are in bold. Standard errors are in parentheses.

Treatment Generation Replicate
1 2 3 4 5 Mean
1 -0.172(0.129) 0.330(0.033) 0.350(0.042) -0.446(0.118) 0.304(0.056) 0.073(0.162)
2 0.447(0.189) 0.230(0.215) 0.123(0.180) 1.012(0.165) -0.658(0.136) 0.231(0.270)
3 0.359(0.137) -0.060(0.22) 0.142(0.184) -0.811(0.115) - -0.093(0.254)
35 ppt 5 0.157(0.131) -0.568(0.122) -0.31(0.179) -0.179(0.148) - -0.225(0.151)
7 -0.382(0.076) -0.469(0.086) -0.046(0.081) -0.525(0.087) - -0.356(0.107)
9 -0.082(0.068) -0.002(0.063) -0.39(0.092) -0.221(0.092) - -0.174(0.085)
11 0.014(0.093) 0.237(0.104) 0.127(0.068) -0.385(0.164) - -0.002(0.136)
13 -0.073(0.079) -0.559(0.106) -0.226(0.119) -0.093(0.097) - -0.237(0.112)

1 -0.119(0.071) -0.125(0.087) 0.231(0.010) -0.121(0.090) - -0.034(0.088)
2 -0.122(0.076) 0.060(0.042) 0.021(0.058) 0.021(0.058) - -0.005(0.040)
3 0.086(0.156) 0.382(0.198) 0.166(0.172) 0.289(0.167) - 0.231(0.065)
53 ppt 5 -0.245(0.103) -0.135(0.097) -0.701(0.086) -0.179(0.083) - -0.315(0.131)
7 -0.037(0.046) -0.035(0.072) -0.292(0.057) -0.196(0.075) - -0.140(0.063)
9 0.249(0.015) -0.014(0.073) -0.217(0.043) 0.141(0.040) - 0.040(0.101)
11 0.074(0.053) 0.108(0.043) 0.058(0.045) -0.037(0.078) - 0.051(0.031)
13 -0.200(0.091) -0.096(0.054) 0.053(0.021) 0.036(0.038) - -0.052(0.060)

115

conditions showed a timepoint of lowest mean survivorship at generation 7. Replicates
of both salinity treatments reached a similar magnitude of outbreeding depression, falling
to 35% below midparent values in the 35 ppt environment and 32% in the 53 ppt
environment at the point of lowest relative fitness. By generation 11, both salinity
treatments showed recovery to midparent values, although at that time point one 35 ppt
replicate was significantly greater than the midparent and one was significantly less.
Generation 13 showed a decline in fitness of the hybrids in benign salinity. In addition to
the mean of all replicates being less than the midparent, each individual replicate showed
lower survivorship than the midparent. The high salinity hybrid treatment did not show a
decrease in survivorship compared to the parents, and the one replicate that was
significantly different from parental expectations showed higher survivorship compared
to the midparent.
When survivorship was scored as yes or no for each nauplius counted, and all
replicates were pooled together for each treatment, the index of potency showed a similar
trend of an increase in fitness for early generations followed by significant declines in
hybrid fitness at generation five and recovery by generation eleven (Figure 3-3). The
index of potency allows for comparison to the parental values, as -1 < h
p
< 1 indicates the
range of parental means. The main differences seen between h
p
(Figure 3-3), where
replicates are combined, and nested ANOVA (Figure 3-2), which accounts for individual
replicates, are the significant decline in hybrid survivorship observed at generation
thirteen for the high salinity treatment and the difference in magnitude of deviations from
expectations.

116

Figure 3-3. Index of potency (h
p
= Q/L) for survivorship over thirteen generations. (a) 35
ppt treatment (b) 53 ppt treatment. Asterisks indicate means significantly different from
the most extreme parent according to planned contrasts (*p < 0.05, **p<0.01,
***p<0.001). Dashed lines contain the range of parental means.

(a)

(b)
53 ppt
-6
-5
-4
-3
-2
-1
0
1
2
1 2 3 5 7 9 11 13
Generation
Index of Potency
***
*

117

To assess repeatability among replicates, one-way ANOVA was performed for
50:50 replicates for each treatment and generation. Significant replicate effects were
seen in all generations for replicates reared at 35 ppt, but generations 2, 3 and 11 for the
53 ppt environment did not show significant differences among replicates. The variance
among 50:50 replicates of the same salinity was averaged over all generations. Mean
variance was 0.04 for replicates at 35 ppt and 0.01 for 53 ppt replicates and the difference
in means was statistically significant (p=0.007, two-tailed t-test). This is also evident
when observing the trajectories of individual replicate beakers (Figure 3-4).

Metamorphosis and Presence of Nauplii at Day 7
Across all generations and treatments, metamorphosis and survivorship were
significantly correlated among individuals (r = 0.524, p < 0.001). Correlation
coefficients calculated within generations were all highly significant with r varying
between 0.444 and 0.598. When replicate means were compared, survivorship and
metamorphosis showed a slightly higher correlation (r=0.628) that was also highly
significant (p<0.0001).
Metamorphosis measures are concordant with survivorship in that we see
evidence of heterosis at generation 2 for the 35 ppt treatment, but not until generation 3
for the 53 ppt environment (Figure 3-5). At these generations showing increased hybrid
fitness, two hybrid replicates in benign salinity were significantly greater than the
midparent while one replicate had significantly lower survivorship (planned contrast)
(Table 3-1). Mirroring survivorship, generation 5 was the timepoint of lowest mean
survivorship among 53 ppt hybrid replicates, but replicates reared in benign conditions

118

Figure 3-4. Mean proportional deviation from midparent for survivorship (± one standard
error) of individual 50:50 replicates over thirteen discrete generations. (a) 35 ppt
treatment (b) 53 ppt treatment. Bars display one standard error.

(a)
35 ppt Survivorship
-1
-0.5
0
0.5
1
0 2 4 6 8 10 12 14
Generation
Proportional deviation from midparent
Replicate 1
Replicate2
Replicate 5
Replicate 4
Replicate 3

(b)
53 ppt Survivorship
-1
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
0 2 4 6 8 10 12 14
Generation
Proportional deviation from midparent
Replicate 1
Replicate 2
Replicate 4
Replicate 5

119

Figure 3-5. Mean proportional deviation from midparent (± one standard error) for
hybrid replicate metamorphosis over thirteen discrete generations. (a) 35 ppt treatment
(b) 53 ppt treatment. Asterisks indicate significant deviation from the midparent
according to planned linear contrasts (p<0.05).

(a)
35 ppt
-0.6
-0.5
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
Generation
Proportional Deviation from Midparent
1 2 3 5 7 9 11 13
* *
*

(b)
53 ppt
-0.7
-0.5
-0.3
-0.1
0.1
0.3
0.5
0.7
Generation
Proportional Deviation from Midparent
1 2 3 5 7 9 11 13
*
*
*

120

showed a time point of lowest mean survivorship at generation 7. The two salinity
treatments reached different magnitudes of outbreeding depression for metamorphosis,
though this difference was not significant. Hybrids of the 35 ppt salinity treatment fell
to37% below midparental fitness while a reduction of 45% was observed for hybrids in
the 53 ppt environment at the point of lowest relative fitness. By generation 11 and
through generation 13 both salinity treatments showed recovery to midparent
metamorphosis measures.
When metamorphosis was scored as yes or no for each nauplius counted, and all
replicates were pooled together for each treatment, the index of potency showed a similar
trends to those seen when nested ANOVA was performed. Generations of maximum
outbreeding depression and recovery occur up to two generations earlier in high salinity
(Figure 3-6). Unlike measures of survivorship, metamorphosis shows complete recovery
in both salinities by generation 11 and is maintained through generation 13.
ANOVA showing fewer replicate effects for 53 ppt generations, as well as a
significantly different mean variance between salinities, was not observed for
metamorphosis. Individual replicate trajectories showed no obvious patterns of
repeatability like those observed for survivorship (Figure 3-4).
ANOVA followed by planned contrasts were conducted on the proportions of
clutches with nauplii present at day seven for each generation. No significant differences
were detected, but this was probably due to the small sample available for this
measurement (five or fewer replicates per treatment). Across all generations and
treatments, number of nauplii present and metamorphosis were significantly negatively
correlated among replicates (r = -0.068, p < 0.05). There was no significant correlation

121

Figure 3-6. Index of potency (h
p
= Q/L) for metamorphosis over thirteen generations. (a)
35 ppt treatment (b) 53 ppt treatment. Asterisks indicate means significantly different
from the most extreme parent according to planned contrasts (*p < 0.05, **p<0.01,
***p<0.001). Dashed lines contain the range of parental means.

(a)
35 ppt
-5
-4
-3
-2
-1
0
1
1 2 3 5 7 9 11 13
Generation
Index of Potency
***
***
***

(b)
53 ppt
-4
-3
-2
-1
0
1
2
3
4
5
1 2 3 5 7 9 11 13
***
***

122

between survivorship and number of nauplii present. For most experimental generations
and both salinity treatments, hybrids showed a higher frequency of clutches with nauplii
present at day seven compared to the midparent (Figures 3-7 and 3-8). For all
generations, hybrid replicates reared in 53 ppt seawater had a much higher proportion of
nauplii present compared to hybrids at 35 ppt, although this difference was only
statistically significant for generation three. To observe the effect of salinity, the
difference in treatment means was calculated (Figure 3-9). The difference in means
increases for all three population types after generation 5, suggesting adaptation to high
salinity. This occurs at a faster rate for SD than RP, while the 50:50 treatment shows no
particular pattern of increase but is equivalent to parentals at generation thirteen.

Morphology
Nested ANOVA was performed across combined generations to assess
morphological variation with replicate held as a random effect. Both generation and
treatment had a significant effect on all 6 body and egg sac characters measured in
females. Nested ANOVA was also performed for individual generations (Table 3-2).
For each generation there was a significant effect of both treatment and replicate for all
six female characters with the exception of treatment for egg sac area at generation 3 and
both effects on antennule width at generation 13. At generation 13, hybrid females reared
in the benign salinity showed significantly increased sizes for almost all measures (Figure
3-10). This coincides with a significant decrease in fitness measures at the same time
point.

123

Figure 3-7. Nauplii present at day 7 (± one standard error) for all treatments at each
generation.

Nauplii at Day 7
0
0.2
0.4
0.6
0.8
1
1.2
1 2 3 5 7 9 11 13
Generation
Proportion of Clutches with Nauplii at Day 7
RP(35ppt)
SD(35ppt)
50:50(35ppt)
RP(53ppt)
SD(53ppt)
50:50(53ppt)

124

Figure 3-8. Number of clutches with nauplii still present seven days after hatching (± one
standard error), shown as the mean hybrid proportional deviation from the midparent. (a)
35 ppt (b) 53 ppt.

(a)
Clutches with Nauplii present at Day 7, 35 ppt
-1
-0.5
0
0.5
1
1.5
2
1 2 3 4 5 6 7 8 9 10 11 12 13
Generation
Proportional Deviation from
Midparent

(b)
Clutches with Nauplii Present at Day 7, 53 ppt
-1
-0.5
0
0.5
1
1.5
2
1 2 3 4 5 6 7 8 9 10 11 12 13
Generation
Proportional Deviation from
Midparent

125

Figure 3-9. Proportion of clutches with nauplli present at day seven. Shown as the
difference of treatment means. Negative values indicate that the 53 ppt treatment had a
higher proportion of nauplii present than the 35 ppt treatment.

-0.6
-0.5
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
1 2 3 5 7 9 11 13
Generation
Difference of Treatment Means
RP
SD
50:50

126
Table 3-2. Proportional deviation from midparent metamorphosis values for each hybrid swarm replicate. Means significantly different from midparent values (α
= 0.05) according to planned linear contrasts are in bold. Standard errors are in parentheses.

Treatment Generation Replicate
1 2 3 4 5 Mean
1 -0.420(0.152) 0.312(0.139) 0.247(0.125) -0.795(0.112) 0.560(0.066) -0.019(0.253)
2 0.876(0.255) 0.247(0.278) -0.176(0.267) 0.876(0.255) -0.790(0.110) 0.206(0.319)
3 0.017(0.186) -0.042(0.247) 0.031(0.215) -0.684(0.149) - -0.169(0.172)
35 ppt 5 -0.209(0.142) -0.403(0.155) -0.219(0.183) -0.280(0.119) - -0.278(0.045)
7 -0.349(0.087) -0.599(0.085) 0.038(0.098) -0.573(0.087) - -0.371(0.147)
9 0.092(0.097) -0.116(0.143) -0.957(0.043) -0.429(0.128) - -0.352(0.228)
11 0.013(0.135) -0.016(0.196) 0.124(0.153) -0.453(0.201) - -0.083(0.127)
13 0.042(0.146) -0.479(0.145) -0.053(0.198) 0.288(0.169) - -0.051(0.160)

1 -0.824(0.098) -0.059(0.181) 0.985(0.028) -0.049(0.216) - 0.013(0.371)
2 -0.543(0.195) -0.778(0.136) 0.064(0.18) 0.064(0.180) - -0.298(0.215)
3 -0.600(0.284) 1.697(0.743) 1.198(0.762) 0.498(0.535) - 0.698(0.498)
53 ppt 5 -0.817(0.077) -0.286(0.139) -0.891(0.066) 0.207(0.369) - -0.447(0.256)
7 0.086(0.144) 0.279(0.191) -0.906(0.067) -0.680(0.139) - -0.305(0.288)
9 1.373(0.070) -0.315(0.263) -1.000(0.000) -0.719(0.154) - -0.165(0.532)
11 0.240(0.228) -0.303(0.200) -0.253(0.165) 0.710(0.171) - 0.098(0.238)
13 -0.820(0.124) 0.098(0.15) -0.025(0.181) 0.395(0.138) - -0.088(0.260)

127
Figure 3-10. Proportional deviations of female 50:50 treatment means from midparent
values over 13 generations of free mating. 35 ppt replicates are represented by black bars
and 53 ppt replicates are represented by gray bars.
(a)

Cephalothorax Length
-0.10
-0.08
-0.06
-0.04
-0.02
0.00
0.02
0.04
0.06
0.08
1 2 3 5 7 9 11 13

(b)

Cephalothorax Width
-0.10
-0.08
-0.06
-0.04
-0.02
0.00
0.02
0.04
0.06
0.08
0.10
1 2 3 5 7 9 11 13

(c)

Urosome Width
-0.10
-0.05
0.00
0.05
0.10
1 2 3 5 7 9 11 13

Proportional Deviation from Midparent Value
Generation
Generation
Generation

128
Figure 3-10, Continued

(d)

Antennule Width
-0.10
-0.05
0.00
0.05
0.10
0.15
0.20
1 2 3 5 7 9 11 13

(e)

Egg Sac Length
-0.15
-0.10
-0.05
0.00
0.05
0.10
0.15
1 2 3 5 7 9 11 13

(f)

Egg Sac Area
-0.30
-0.20
-0.10
0.00
0.10
0.20
0.30
1 2 3 5 7 9 11 13

Generation
Generation
Generation
Proportional Deviation from Midparent Value

129
Characters were considered to be transgressive if the hybrid mean was
significantly different from both parents and not intermediate to parentals. Transgressive
characters were only observed in the 53 ppt hybrids for the first three generations (CTW
in generation one and CTL and ESA in generation three), all of which were greater than
the parental measures (Table 3-3). At generation five, CTW and AW both exceeded
parentals in the 35 ppt hybrids and CTL and CTW were found to be greater in size than
parentals in the 53 ppt environment. Generation seven showed the highest frequency of
transgressive measurements for the 35 ppt treatment, with five of six hybrid female
measures significantly smaller than the smallest parent. Two characters (UW and ESA)
were also significantly smaller than the smallest parent in the high salinity treatment.
After generation 7, transgressive characters are not observed until generation 13, at which
only the benign salinity environment contains measures that are significantly different
and larger than both parental groups.
Differences in mean female measurements between generations one and thirteen
indicated no particular trends at 35 ppt (Table 3-5a), except for egg sac size which
exhibited a significant increase (two-tailed t-test, p<0.05). RP females at 53 ppt showed
significant decreases for all measures while SD at 53 ppt showed only significant
increases (for three out of six measures). Five out of six hybrid 50:50 measures
decreased and three of those decreases were statistically significant.
The same analyses were also performed for male morphological data. Nested
ANOVA with replicate held as a random effect was performed across combined
generations. Generation, treatment and replicate all had a significant effect on all eight
morphological characters. Nested ANOVA was also performed for individual

130
Table 3-3. Mean phenotypic values for female morphometric characters for all control and hybrid swarm treatments. Units are in
millimeters. Treatment means significantly different from the midparent value according to nested ANOVA and planned contrast tests
are indicated in bold. The direction of deviation from the additive expectation is indicated by a + or – sign. Means that are boxed are
transgressive in that they are significantly different from both parents (Bonferroni post hoc test) and not intermediate to them.
Standard errors among replicates are in parentheses.
Generation Treatment CTL CTW UW AW ESL ESA
RP-35 0.4964(0.0196) 0.3223(0.0139) 0.1452(0.0069) 0.0353(0.0012) 0.4200(0.0128) 0.1114(0.0059)
SD-35 0.5226(0.0090) 0.3413(0.0054) 0.1545(0.0028) 0.0364(0.0008) 0.4446(0.0166) 0.1154(0.0068)
1 50:50-35 0.5237(0.0136)+ 0.3419(0.0112)+ 0.1555(0.0059)+ 0.0375(0.0012)+ 0.4342(0.0136) 0.1124(0.0068)
RP-53 0.5509(0.0078) 0.3506(0.0046) 0.1591(0.0019) 0.0373(0.0007) 0.4676(0.0217) 0.1334(0.0095)
SD-53 0.5375(0.0034) 0.3485(0.0041) 0.1620(0.0007) 0.0367(0.0005) 0.4385(0.0092) 0.1208(0.0038)
50:50-53 0.5606(0.0102)+ 0.3637(0.0071)+ 0.1634(0.0013) 0.0373(0.0004) 0.4752(0.0241)+ 0.1402(0.0117)+

RP-35 0.5118(0.0146) 0.3280(0.0101) 0.1424(0.0043) 0.0343(0.0008) 0.4519(0.0119) 0.1244(0.0063)
SD-35 0.5404(0.0050) 0.3507(0.0052) 0.1549(0.0028) 0.0347(0.0004) 0.4821(0.0135) 0.1369(0.0056)
2 50:50-35 0.5339(0.0087)+ 0.3455(0.0052)+ 0.1505(0.0023) 0.0343(0.0004) 0.4370(0.0118)- 0.1180(0.0042)-
RP-53 0.5322(0.0100) 0.3361(0.0046) 0.1502(0.0032) 0.0359(0.0006) 0.4433(0.0162) 0.1231(0.0072)
SD-53 0.5246(0.0072) 0.3361(0.0046) 0.1518(0.0010) 0.0348(0.0003) 0.4356(0.0094) 0.1128(0.0039)
50:50-53 0.5260(0.0137) 0.3419(0.0088)+ 0.1470(0.0037)- 0.0354(0.0013) 0.4365(0.0120) 0.1302(0.0105)+

RP-35 0.5150(0.0071) 0.3277(0.0066) 0.1437(0.0030) 0.0321(0.0006) 0.4359(0.0102) 0.1172(0.0030)
SD-35 0.5505(0.0061) 0.3542(0.0032) 0.1602(0.0010) 0.0343(0.0009) 0.4701(0.0080) 0.1284(0.0039)
3 50:50-35 0.5593(0.0117)+ 0.3621(0.0059)+ 0.1623(0.0022)+ 0.0356(0.0005)+ 0.4802(0.0147)+ 0.1414(0.0067)
RP-53 0.5427(0.0104) 0.3371(0.0062) 0.1510(0.0036) 0.0342(0.0008) 0.4667(0.0149) 0.1387(0.0168)
SD-53 0.5546(0.0110) 0.3565(0.0060) 0.1627(0.0033) 0.0358(0.0008) 0.4700(0.0170) 0.1311(0.0084)
50:50-53 0.5690(0.0019)+ 0.3583(0.0018)+ 0.1560(0.0023) 0.0363(0.0005)+ 0.4947(0.0166) 0.1441(0.0051)+

RP-35 0.5317(0.0125) 0.3319(0.0085) 0.1487(0.0042) 0.0333(0.0010) 0.4225(0.0186) 0.1200(0.0084)
SD-35 0.5684(0.0119) 0.3637(0.0087) 0.1631(0.0035) 0.0362(0.0008) 0.4710(0.0187) 0.1410(0.0107)
5 50:50-35 0.5744(0.0125)+ 0.3746(0.0080)+ 0.1670(0.0037)+ 0.0397(0.0004)+ 0.4635(0.0139)+ 0.1338(0.0089)
RP-53 0.5426(0.0074) 0.3441(0.0071) 0.1516(0.0034) 0.0357(0.0011) 0.4410(0.0156) 0.1228(0.0069)

131
Table 3-3, Continued
SD-53 0.5640(0.0045) 0.3621(0.0026) 0.1617(0.0010) 0.0374(0.0003) 0.4595(0.0096) 0.1373(0.0046)
50:50-53 0.5823(0.0088)+ 0.3698(0.0061)+ 0.1628(0.0038)+ 0.0374(0.0012) 0.4691(0.0225)+ 0.1425(0.0105)+

RP-35 0.5961(0.0016) 0.3748(0.0023) 0.1647(0.0015) 0.0350(0.0006) 0.4973(0.0022) 0.1532(0.0077)
SD-35 0.5643(0.0179) 0.3676(0.0116) 0.1636(0.0042) 0.0373(0.0009) 0.4694(0.0135) 0.1471(0.0087)
7 50:50-35 0.5391(0.0025)- 0.3455(0.0014)- 0.1515(0.0014)- 0.0356(0.0005) 0.4394(0.0083)- 0.1238(0.0036)-
RP-53 0.5609(0.0209) 0.3500(0.0118) 0.1579(0.0058) 0.0348(0.0008) 0.4633(0.0318) 0.1434(0.0140)
SD-53 0.5799(0.0096) 0.3745(0.0044) 0.1700(0.0021) 0.0390(0.0006) 0.4764(0.0033) 0.1435(0.0020)
50:50-53 0.5633(0.0086)- 0.3516(0.0069)- 0.1524(0.0035)- 0.0368(0.0004) 0.4465(0.0074)- 0.1300(0.0048)-

RP-35 0.5296(0.0110) 0.3474(0.0040) 0.1482(0.0024) 0.0364(0.0006) 0.4424(0.0200) 0.1197(0.0089)
SD-35 0.5608(0.0079) 0.3573(0.0023) 0.1575(0.0025) 0.0384(0.0006) 0.4776(0.0174) 0.1362(0.0063)
9 50:50-35 0.5469(0.0057) 0.3614(0.0042)+ 0.1514(0.0023) 0.0384(0.0010) 0.4526(0.0159) 0.1299(0.0063)
RP-53 0.5350(0.0090) 0.3364(0.0068) 0.1436(0.0043) 0.0367(0.0007) 0.4457(0.0152) 0.1258(0.0068)
SD-53 0.5779(0.0031) 0.3672(0.0026) 0.1540(0.0018) 0.0383(0.0010) 0.4873(0.0063) 0.1445(0.0022)
50:50-53 0.5517(0.0061) 0.3469(0.0030) 0.1463(0.0020) 0.0369(0.0007) 0.4577(0.0266) 0.1335(0.0106)

RP-35 0.5514(0.0058) 0.3507(0.0010) 0.1502(0.0018) 0.0375(0.0008) 0.4767(0.0097) 0.1355(0.0057)
SD-35 0.5252(0.0174) 0.3487(0.0097) 0.1494(0.0055) 0.0373(0.0011) 0.4623(0.0282) 0.1293(0.0128)
11 50:50-35 0.5335(0.0098) 0.3460(0.0054) 0.1514(0.0025) 0.0364(0.0012) 0.4470(0.0088)- 0.1283(0.0044)
RP-53 0.5419(0.0082) 0.3439(0.0044) 0.1516(0.0033) 0.0374(0.0008) 0.4542(0.0121) 0.1279(0.0058)
SD-53 0.5670(0.0029) 0.3666(0.0040) 0.1652(0.0011) 0.0387(0.0005) 0.4716(0.0081) 0.1342(0.0025)
50:50-53 0.5640(0.0051)+ 0.3576(0.0068) 0.1555(0.0023)- 0.0379(0.0010) 0.4778(0.0025)+ 0.1439(0.0049)+

RP-35 0.5133(0.0041) 0.3266(0.0017) 0.1414(0.0016) 0.0333(0.0006) 0.4086(0.0114) 0.1127(0.0048)
SD-35 0.5098(0.0061) 0.3420(0.0038) 0.1530(0.0020) 0.0395(0.0050) 0.4106(0.0180) 0.1089(0.0045)
13 50:50-35 0.5316(0.0065)+ 0.3438(0.0022)+ 0.1543(0.0020)+ 0.0346(0.0011) 0.4563(0.0150)+ 0.1332(0.0080)+
RP-53 0.5413(0.0052) 0.3429(0.0053) 0.1496(0.0032) 0.0355(0.0006) 0.4340(0.0094) 0.1246(0.0029)
SD-53 0.5560(0.0063) 0.3601(0.0016) 0.1628(0.0010) 0.0366(0.0006) 0.4504(0.0127) 0.1304(0.0063)
50:50-53 0.5474(0.0057) 0.3498(0.0029) 0.1561(0.0021) 0.0379(0.0013) 0.4604(0.0061)+ 0.1364(0.0044)+

132
Table 3-4. Mean phenotypic values for male morphometric characters for all control and hybrid swarm treatments. Units are in
millimeters. Treatment means significantly different from the midparent value according to nested ANOVA and planned contrast tests
are indicated in bold. The direction of deviation from the additive expectation is indicated by a + or – sign. Means that are boxed are
transgressive in that, given they are significantly different from both parents (Bonferroni post hoc test) and not intermediate. Standard
errors among replicates are in parentheses.
Gen Treatment CTL UL CTW UW TW CSL AW CLW
RP-35 0.4444(0.0143) 0.3423(0.0096) 0.3112(0.0114) 0.1323(0.0043) 0.0347(0.0007) 0.7248(0.0459) 0.0434(0.0020) 0.0604(0.0023)
SD-35 0.4641(0.0080) 0.3535(0.0073) 0.3239(0.0061) 0.1397(0.0027) 0.0361(0.0004) 0.7323(0.0360) 0.0448(0.0008) 0.0627(0.0010)
1 50:50-35 0.4749(0.0099)+ 0.3664(0.0115)+ 0.3342(0.0091)+ 0.1418(0.0035)+ 0.0376(0.0012)+ 0.8200(0.0469)+ 0.0463(0.0010)+ 0.0638(0.0022)+
RP-53 0.4915(0.0051) 0.3743(0.0046) 0.3392(0.0033) 0.1387(0.0007) 0.0359(0.0005) 0.8280(0.0140) 0.0484(0.0005) 0.0658(0.0009)
SD-53 0.4879(0.0054) 0.3707(0.0045) 0.3394(0.0051) 0.1402(0.0022) 0.0379(0.0002) 0.8232(0.0111) 0.0462(0.0010) 0.0621(0.0004)
50:50-53 0.4983(0.0083)+ 0.3777(0.0038) 0.3532(0.0071)+ 0.1438(0.0033)+ 0.0385(0.0007)+ 0.8453(0.0229) 0.0469(0.0011) 0.0632(0.0012)

RP-35 0.4731(0.0103) 0.3617(0.0044) 0.3265(0.0081) 0.1357(0.0023) 0.0333(0.0005) 0.7952(0.0185) 0.0454(0.001) 0.0658(0.0009)
SD-35 0.5005(0.0059) 0.3855(0.0031) 0.3477(0.0066) 0.1447(0.0025) 0.0361(0.0006) 0.8533(0.0197) 0.0458(0.0003) 0.0681(0.0004)
2 50:50-35 0.4987(0.0067)+ 0.3752(0.0074) 0.3438(0.0075)+ 0.1431(0.003)+ 0.0372(0.0008)+ 0.8549(0.0147)+ 0.0455(0.0005) 0.0675(0.0011)
RP-53 0.4875(0.0050) 0.3639(0.0034) 0.3315(0.0045) 0.1381(0.0011) 0.0342(0.0005) 0.8041(0.0244) 0.0499(0.0065) 0.0661(0.0009)
SD-53 0.4856(0.0052) 0.3662(0.0058) 0.3334(0.0044) 0.1354(0.0015) 0.0355(0.0004) 0.7844(0.0206) 0.0439(0.0005) 0.0643(0.0007)
50:50-53 0.4874(0.0097) 0.3634(0.0033) 0.3384(0.0097)+ 0.1379(0.0034) 0.0370(0.0004)+ 0.7733(0.0397) 0.0453(0.0011) 0.0670(0.0012)+

RP-35 0.4677(0.0131) 0.3572(0.0050) 0.3237(0.0049) 0.1384(0.0017) 0.0345(0.0006) 0.808(0.0314) 0.0454(0.0011) 0.0652(0.0012)
SD-35 0.5108(0.0054) 0.3944(0.0047) 0.3526(0.0039) 0.1448(0.0016) 0.0381(0.0002) 0.897(0.0108) 0.048(0.0008) 0.0701(0.0013)
3 50:50-35 0.5042(0.0048)+ 0.3764(0.0023) 0.3502(0.0029)+ 0.1478(0.0009) 0.0362(0.0008) 0.908(0.0105)+ 0.0471(0.0005) 0.0697(0.0004)
RP-53 0.4975(0.0059) 0.3677(0.0046) 0.3332(0.0058) 0.1378(0.0017) 0.0347(0.0004) 0.8251(0.0185) 0.0462(0.0010) 0.0656(0.0012)
SD-53 0.5019(0.0062) 0.3689(0.0040) 0.3469(0.0046) 0.1424(0.0018) 0.0365(0.0005) 0.8588(0.0181) 0.0474(0.0004) 0.0667(0.0006)
50:50-53 0.5052(0.0043) 0.3693(0.0011) 0.3517(0.0028)+ 0.1456(0.0012)+ 0.0362(0.0006) 0.8249(0.0131) 0.0463(0.0003) 0.0666(0.0014)

RP-35 0.4916(0.0064) 0.3739(0.0068) 0.3330(0.0093) 0.1375(0.0030) 0.0355(0.0007) 0.8841(0.0148) 0.0472(0.0008) 0.0660(0.0018)
SD-35 0.5108(0.0054) 0.3944(0.0047) 0.3526(0.0039) 0.1448(0.0016) 0.0381(0.0002) 0.8970(0.0108) 0.0480(0.0008) 0.0701(0.0013)
5 50:50-35 0.5186(0.0082)+ 0.3895(0.0034) 0.3649(0.0057)+ 0.1540(0.0024)+ 0.0377(0.0007) 0.9288(0.0049)+ 0.0500(0.0007)+ 0.0717(0.0011)+
RP-53 0.5032(0.0079) 0.3774(0.0060) 0.3381(0.0084) 0.1389(0.0037) 0.0349(0.0009) 0.8575(0.0168) 0.0493(0.0011) 0.0676(0.0020)

133
Table 3-4, Continued
SD-53 0.5222(0.0047) 0.3916(0.0035) 0.3585(0.0047) 0.1486(0.0017) 0.0387(0.0005) 0.9079(0.0066) 0.0502(0.0007) 0.0680(0.0007)
50:50-53 0.5238(0.0078)+ 0.3879(0.0047) 0.3599(0.0065)+ 0.1462(0.0027)+ 0.0381(0.0012)+ 0.9062(0.0178)+ 0.0506(0.0008) 0.0704(0.0014)+

RP-35 0.5169(0.0072) 0.3964(0.0071) 0.3487(0.0040) 0.1560(0.0106) 0.0380(0.0011) 0.8838(0.0189) 0.0505(0.0020) 0.0710(0.0016)
SD-35 0.5202(0.0125) 0.3936(0.0107) 0.3589(0.0080) 0.1485(0.0037) 0.0388(0.0016) 0.9084(0.0416) 0.0489(0.0010) 0.0703(0.0020)
7 50:50-35 0.4932(0.0051)- 0.3791(0.0072)- 0.3404(0.0026)- 0.1408(0.0008)- 0.0376(0.0011) 0.8542(0.0164)- 0.0491(0.0012) 0.0693(0.0008)
RP-53 0.5090(0.0177) 0.3827(0.0115) 0.3382(0.0105) 0.1415(0.0043) 0.0360(0.0010) 0.8938(0.0235) 0.0490(0.0011) 0.0683(0.0014)
SD-53 0.5247(0.0113) 0.3968(0.0079) 0.3626(0.0076) 0.1505(0.0028) 0.0387(0.0009) 0.9581(0.0192) 0.0516(0.0008) 0.0699(0.0009)
50:50-53 0.5108(0.0054)- 0.3773(0.0081)- 0.3442(0.0027)- 0.1419(0.0014) 0.0363(0.0005)- 0.9199(0.0117) 0.0492(0.0006) 0.0670(0.0005)-

RP-35 0.4960(0.0048) 0.3621(0.0030) 0.3472(0.0046) 0.1404(0.0013) 0.0341(0.0004) 0.8132(0.0155) 0.0464(0.0009) 0.0660(0.0008)
SD-35 0.5034(0.0049) 0.3742(0.0029) 0.3487(0.0019) 0.1420(0.0013) 0.0359(0.0001) 0.8496(0.0139) 0.0465(0.0008) 0.0653(0.0005)
9 50:50-35 0.5032(0.0073) 0.3693(0.0058) 0.3476(0.0074) 0.1416(0.0024) 0.0365(0.0015)+ 0.8583(0.0050)+ 0.0468(0.0004) 0.0675(0.0011)+
RP-53 0.4931(0.0080) 0.3606(0.0035) 0.3336(0.0045) 0.1342(0.0010) 0.0332(0.0005) 0.8167(0.0036) 0.0454(0.0004) 0.0646(0.0007)
SD-53 0.5164(0.0032) 0.3840(0.0024) 0.3554(0.0022) 0.1402(0.0013) 0.0357(0.0004) 0.8714(0.0086) 0.0458(0.0005) 0.0654(0.0005)
50:50-53 0.5044(0.0026) 0.3780(0.0044) 0.3389(0.0020)- 0.1344(0.0013)- 0.0340(0.0006) 0.8164(0.0205)- 0.0446(0.0008) 0.0639(0.0007)

RP-35 0.4904(0.0143) 0.3695(0.0100) 0.3366(0.0079) 0.1377(0.0028) 0.0344(0.0004) 0.8182(0.0283) 0.0489(0.0013) 0.0657(0.0012)
SD-35 0.4939(0.0083) 0.3629(0.0027) 0.3457(0.0045) 0.1401(0.0022) 0.0366(0.0008) 0.8415(0.0064) 0.0473(0.0004) 0.0656(0.0006)
11 50:50-35 0.4884(0.0075) 0.3727(0.0093) 0.3382(0.0040) 0.1359(0.0027)- 0.0356(0.0006) 0.8287(0.0197) 0.0449(0.0013)- 0.0643(0.0013)
RP-53 0.4913(0.0037) 0.3715(0.0031) 0.3375(0.0024) 0.1389(0.0015) 0.0348(0.0002) 0.8499(0.0092) 0.0483(0.0006) 0.0643(0.0006)
SD-53 0.5115(0.0020) 0.3947(0.0014) 0.3545(0.0019) 0.1455(0.0014) 0.0382(0.0005) 0.8667(0.0193) 0.0508(0.0007) 0.0644(0.0006)
50:50-53 0.4978(0.0048) 0.3740(0.0056)- 0.3371(0.0038)- 0.1405(0.0024) 0.0363(0.0004) 0.8204(0.0165)- 0.0472(0.0006)- 0.0639(0.0015)

RP-35 0.4692(0.0049) 0.3560(0.0009) 0.3249(0.0038) 0.1349(0.0005) 0.0345(0.0004) 0.8091(0.0090) 0.0466(0.0002) 0.0622(0.0015)
SD-35 0.4743(0.0084) 0.3574(0.0072) 0.3322(0.0046) 0.1374(0.0028) 0.0371(0.0003) 0.7949(0.0128) 0.0480(0.0006) 0.0635(0.0004)
13 50:50-35 0.4814(0.0021) 0.3636(0.0030) 0.3310(0.0049) 0.1385(0.0013) 0.0371(0.0008) 0.8270(0.0204) 0.0460(0.0008) 0.0656(0.0015)+
RP-53 0.5046(0.0141) 0.3786(0.0115) 0.3417(0.0093) 0.1436(0.0018) 0.0368(0.0009) 0.8554(0.0189) 0.0499(0.0012) 0.0648(0.0012)
SD-53 0.5069(0.0026) 0.3872(0.0030) 0.3498(0.0012) 0.1457(0.0014) 0.0384(0.0005) 0.8694(0.0137) 0.0494(0.0002) 0.0652(0.0008)
50:50-53 0.4978(0.0048)- 0.3740(0.0056)- 0.3371(0.0038)- 0.1405(0.0024)- 0.0363(0.0004)- 0.8204(0.0165)- 0.0472(0.0006)- 0.0639(0.0015)

134
Table 3-5. Difference in mean morpholometrics between generations 1 and 13 for females (a) and males
(b). Units are in millimeters. Positive values indicate an increase in size between generations. Significant
differences are indicated in bold (two-tailed t-test, p<0.05).

(a)
Treatment CTL CTW UW AW ESL ESA
RP-35 0.0169 0.0043 -0.0038 -0.0020 -0.0114 0.0012
SD-35 -0.0127 0.0007 -0.0015 0.0031 -0.0339 -0.0065
50:50-35 0.0079 0.0019 -0.0012 -0.0029 0.0221 0.0208
RP-53 -0.0096 -0.0077 -0.0095 -0.0018 -0.0336 -0.0088
SD-53 0.0185 0.0116 0.0008 -0.0001 0.0119 0.0096
50:50-53 -0.0132 -0.0138 -0.0073 0.0006 -0.0148 -0.0037

(b)
Treatment CTL UL CTW UW TW CSL AW CLW
RP-35 0.0249 0.0137 0.0137 0.0026 -0.0002 0.0844 0.0032 0.0018
SD-35 0.0102 0.0039 0.0083 -0.0023 0.0010 0.0626 0.0031 0.0008
50:50-35 0.0065 -0.0029 -0.0032 -0.0033 -0.0005 0.0070 -0.0003 0.0019
RP-53 0.0130 0.0043 0.0025 0.0050 0.0009 0.0274 0.0015 -0.0010
SD-53 0.0190 0.0165 0.0104 0.0056 0.0005 0.0463 0.0031 0.0030
50:50-53 -0.0005 -0.0037 -0.0161 -0.0033 -0.0022 -0.0249 0.0002 0.0006

135
generations (Table 3-4). For each generation, there was a significant effect of both
treatment and replicate seen for all eight male characters with the exception of both
effects on antennule width at generation 2 and treatment on clasper width at generation
11.
Overall, hybrids reared in high salinity that had male characters differing
significantly from the midparent were larger for generations one through five, but smaller
in generations seven through thirteen (Table 3-4 and Figure 3-11). This same pattern was
not seen for males reared in benign conditions. Hybrid males from the 35 ppt salinity
with characters differing significantly from the midparent were larger with the exception
of generations seven and 11. Generation seven is the point of lowest fitness for the 35
ppt replicates and generation 11 is the point of full fitness recovery to midparent values.
For males, all observed transgressive characters up until generation five were from the 35
ppt treatment and in the positive direction. This was seen for four measurements in
generation one plus CTW and UW in generation five. CTL, CTW and CSL, also from
the 35 ppt salinity, were transgressive at generation seven in the negative direction.
Differences in mean male morphometrics between generations one and thirteen
for RP and SD show only significant increases (two-tailed t-test, p<0.05) and this is seen
at both salinity treatments (Table 3-5b). 50:50 males at 35 ppt showed decreases in six
out of eight characters but only one (UW) was significant. 50:50 males at 53 ppt also
showed decreases in six out of eight characters and three of those were significant (CTW,
TW, and CSL).

136
Figure 3-11. Proportional deviations of male 50:50 treatment means from midparent
values over 13 generations of free mating. 35 ppt replicates are represented by black bars
and 53 ppt replicates are represented by gray bars.

(a)

Cephalothorax Length
-0.06
-0.04
-0.02
0.00
0.02
0.04
0.06
1 2 3 5 7 9 11 13

(b)

Urosome Length
-0.06
-0.04
-0.02
0.00
0.02
0.04
0.06
0.08
1 2 3 5 7 9 11 13

Generation
Generation
Proportional Deviation from Midparent Value

137
Figure 3-11, Continued

(c)

Cephalothorax Width
-0.06
-0.04
-0.02
0.00
0.02
0.04
0.06
0.08
1 2 3 5 7 9 11 13

(d)

Urosome Width
-0.10
-0.05
0.00
0.05
0.10
0.15
1 2 3 5 7 9 11 13

(e)

Telson Width
-0.06
-0.04
-0.02
0.00
0.02
0.04
0.06
0.08
0.10
1 2 3 5 7 9 11 13

Generation
Generation
Proportional Deviation from Midparent Value

138

Figure 3-11, Continued

(f)

Caudal Seta Length
-0.10
-0.05
0.00
0.05
0.10
0.15
0.20
1 2 3 5 7 9 11 13

(g)

Antennule Width
-0.08
-0.06
-0.04
-0.02
0.00
0.02
0.04
0.06
0.08
1 2 3 5 7 9 11 13

(h)

Clasper Width
-0.06
-0.04
-0.02
0.00
0.02
0.04
0.06
0.08
1 2 3 5 7 9 11 13

Generation
Generation
Generation
Proportional Deviation from Midparent Value

139
A common pattern seen across both male and female morphological measures
was the relative decrease in hybrid measures compared to parentals at generation five
followed by a relative increase at generation seven (Figures 3-10 and 3-11).
Hybrid index scores were determined such that a score of 0 indicates RP-like
morphology while a score of 1 indicates SD-like morphology. Mean female hybrid index
showed both an increase and then a decrease over time at both salinities, but did not
change significantly between generations one and thirteen (Figure 3-12a). Mean hybrid
index at generation 13 for the 35 ppt replicates was 0.551 while the mean index for 53 ppt
replicates was 0.498, indicating overall intermediate morphology for hybrids of both
salinity treatments. Mean male hybrid index showed a decrease over time in both
salinities (Figure 3-12b). At generation 13, mean hybrid index for 35 ppt replicates was
not statistically different from generation 1. In comparison, mean hybrid index for 53 ppt
had significantly decreased from generation 1 (p= 0.011, two-tailed t-test), suggesting
that male morphology had become more RP-like in the high salinity environment, while
males in benign conditions retained more intermediate measures. Mean scores are
significantly different between salinities at generations 3, 7, 9, 11 and 13. The hybrid
index variance among 50:50 replicates of the same salinity was averaged over all
generations. Difference in mean variance between salinities was not statistically
significant for either males or females.

Competitive Fitness
After thirteen generations the RP population exhibited lower survivorship in the
high salinity environment relative to its survivorship under benign conditions. This was

140
Figure 3-12. Mean morphological hybrid index among replicates over thirteen
generations. A score of 0 indicates RP-like morphology whereas a score of 1 indicates
SD-like morphology. Bars indicate one standard error. Dashed line indicates additive
expectation of 0.5. (9a) Females; (9b) Males. Asterisks denote significant differences
between treatments (p<0.05, two-tailed t-test ).

(a)
Females
0.3
0.35
0.4
0.45
0.5
0.55
0.6
0.65
0.7
0 1 2 3 4 5 6 7 8 9 10 11 12 13
Generation
35 ppt
53 ppt
*
Morphological Hybrid Index

(b)
Males
0.3
0.35
0.4
0.45
0.5
0.55
0.6
0.65
0 1 2 3 4 5 6 7 8 9 10 11 12 13
Generation
35 ppt
53 ppt
*
*
* *
*
Morphological Hybrid Index

141
in contrast to the beginning of the experiment where it was the SD population that
experienced a reduction in fitness at high salinity. RP and SD were competitively
equivalent under benign conditions while RP was the inferior parent at high salinity
(p=0.063 and p<0.0001 respectively, two-tailed t-test). Under normal salinity conditions
hybrids had competitive survivorship measures that were equivalent to parentals (Figure
3-13a). Hybrids reared in the high salinity treatment (Figure 3-13b), however, showed
survivorship that was significantly greater than the midparent according to ANOVA and
a planned linear contrast (p = 0.0303). Hybrids under normal salinity performed more
like the inferior parent, while those reared in high salinity resembled the superior parent.
This was consistent with individual clutch survivorship measures at generation 13 which
were lower for hybrid replicates in the 35 ppt salinity treatment compared to those at 53
ppt. However, when comparing hybrids to the midparent, the 35 ppt salinity survivorship
only hybrids reared in benign conditions show decreased survivorship. Competitive
fitness, as opposed to individual clutch survivorship, shows that both treatments have
recovered from outbreeding depression by generation 13, and those raised under salinity
stress exceeded midparent fitness.

Heat Shock Tolerance
For replicates reared in 35 ppt seawater, hybrids had significantly lower heat
shock survivorship than parentals (Figure 3-14). In contrast, for replicates reared at 53
ppt, hybrid survivorship after heat shock was equivalent to parentals (Figure 3-14). When
survivorship was accounted for separately for nauplii and metamorphosed copepodids
(Figure 3-15), a highly significant salinity effect was observed for hybrid nauplii

142
Figure 3-13. Three-week competitive fitness observed 13 generations after hybrid swarm
formation. (a) 35 ppt; (b) 53 ppt. Error bars show one standard error.

(a)
15
25
35
45
55
65
% non-SC survivors after 3 weeks
RP 50:50 SD

(b)

15
25
35
45
55
65
% non-SC survivors after 3 weeks
RP 50:50 SD

143
Figure 3-14. Generation 14 heat shock assay performed for replicates reared under both
35 ppt and 53 ppt. Error bars indicate one standard error.

45
50
55
60
65
70
75
RP
SD
50:50
35 ppt 53 ppt
% Individuals Surviving Heat Shock

144
Figure 3-15. Generation 14 heat shock survivorship (± one standard error).
Contributions of nauplii and copepods.

0
10
20
30
40
50
60
70
80
RP 35ppt RP 53ppt SD 35ppt SD 53ppt Hyb 35ppt Hyb 53ppt
Percent Survival
Copepods
Nauplii

145
(p<0.0001) but not for surviving hybrid copepodids (p = 0.112). In all three treatments
(RP, SD and 50:50), the ratio of survivors reaching the copepodid stage compared to
those that did not metamorphose is significantly reduced in high salinity, while the total
number of survivors increased at high salinity for hybrids (p = 0.500, independent t-test),
but not for either parental treatment.

DISCUSSION

Duration and severity of outbreeding depression
Interbreeding between divergent populations has been known to result in F1
heterosis followed by hybrid breakdown in the second or later generations (Edmands
1999, Edmands 2007). Controlled crosses from a previous study (Chapter 1) utilized the
same populations as this experiment and found that survivorship showed no significant
deviation from the midparent in the F1 or backcross cohorts while F2 cohorts showed
large and significant declines. F3 cohorts recovered to midparent survivorship. The
fitness outcomes of controlled F1, F3 and backcross generations in this particular
interpopulation cross offer reasons to believe that hybrid swarms can recover from early
generation outbreeding depression, depending on the magnitude of hybrid breakdown
generated by recombinant F2s. The actual time course of recovery in natural hybrid
swarms would be difficult to predict due to overlapping generations and the persistence
of fertilized females with lifespans up to 95 days (Vittor 1971). For this reason, we chose
to maintain our experimental swarms as distinct generations. This being the case, in a
randomly mating population we would expect the largest proportion of F2 individuals to

146
be present at generation 2 at a frequency of 25%. This may not, however, be the swarm
generation of lowest fitness, as 75% of the population should be composed of highly fit
backcross, F1 and parental individuals. Additionally, if coadapted gene complexes are
composed of tightly linked genes, the full effects of outbreeding depression may not be
observed until after many generations of recombination have occurred (Fenster and
Galloway 2000). Although the possibilities of positive assortative mating or genetic
swamping may exist, previous studies have shown that T. californicus does not avoid
outbreeding (Ganz and Burton 1995, Palmer and Edmands 2000), and molecular data
from Chapter 4 of this dissertation show that hybrid genotypes were maintained into the
thirteenth generation of this experiment.
Ultimately the magnitude and duration of outbreeding depression will be
determined by the types of recombinants that are generated and selected for or against in
subsequent generations, and we were particularly interested in observing the effects of a
stressful environment on these parameters. While it has been suggested that stress may
intensify both F1 heterosis and F2 hybrid breakdown (Vetukhiv and Beardmore 1959), a
number of studies suggest that outbreeding depression may be alleviated by stress
(Hoffman and Parsons 1991, Armbruster et al. 1997, Edmands and Deimler 2004).
Another finding is that novel environmental conditions might supply hybrid genotypes
with new niches for which they are better fit than parental genotypes (Johansen-Morris
and Latta 2006, Rieseberg et al. 1999).
Our data showed that, for survivorship, early generation heterosis was followed
by both outbreeding depression and recovery, which occurred up to two generations
earlier in high salinity treatments. Generation five was the timepoint of lowest mean

147
survivorship among 53 ppt hybrid replicates, but replicates reared under benign
conditions showed lowest mean survivorship at generation seven (Table 3-1). Both
salinity treatments reached a similar magnitude of outbreeding depression. These
patterns were concordant with metamorphosis measures in that we see evidence of early
generation heterosis followed by hybrid breakdown. Like survivorship, generation five
exhibited lowest mean metamorphosis for 53 ppt hybrid replicates, but replicates reared
in benign conditions showed lowest mean metamorphosis at generation seven (Table 3-
2). The two salinity treatments reached different magnitudes of outbreeding depression
for metamorphosis with high salinity hybrids showing a greater decrease from midparent
metamorphosis, though this difference was not significant. As with survivorship,
outbreeding depression and recovery both occurred up to two generations earlier in the
high salinity environment. Overall, the stressful environment altered the time course of
outbreeding depression, but not necessarily the magnitude or duration of it. However, it
could be interpreted that outbreeding depression was decreased in the high salinity
treatment because replicates recovered to parental fitness within fewer generations.
In comparison to this data, which was collected for discrete generations, 50:50
hybrid swarms from Chapter 1 recovered to midparent values by month 9 of free mating,
and surpassed midparental fitness by month 12. For benign salinity swarms where
generations were kept discrete, recovery from outbreeding depression occurred at
generation eleven, which corresponded to approximately 12 months of experimental time.
The partitioning of generations increased the time to recovery from outbreeding
depression and produced no evidence suggesting that hybrids can surpass parent

148
population fitness. This may be because unique recombinant genotypes may be created
at a much faster rate when matings can form between hybrids of different generations.
For most experimental generations and both salinity treatments, 50:50 replicates
showed higher frequency of clutches with nauplii present at day seven compared to the
midparent (Figures 7 and 8). For all generations, 50:50 replicates reared in 53 ppt
seawater had a higher proportion of nauplii present compared to those at 35 ppt, although
this difference was only statistically significant for generation three. The presence of
nauplii at day seven observed over all thirteen generations suggests slower development
under high salinity. Trade offs between stress resistance and developmental rate have
been identified in other organisms (Chippindale et al. 1998, Harshman et al. 1999) as
well as Tigriopus (Edmands and Deimler 2004). Despite the possibility of this
physiological trade off, salinity stress did not alter the duration of outbreeding depression,
and only mildly affected the severity of metamorphosis measures in the more stressful
environment.
A simple model by Christiansen (2008) predicted that hybrids swarms showing
early generation heterosis followed by hybrid breakdown should eventually converge to
the mean fitness of the parental populations at a stable equilibrium. While our results are
generally consistent with this model, we do observe that fitness measurements for one
salinity treatment drop below the midparent at generation 13, the last generation where all
fitness measures were taken. However, competitive fitness measured at generation 13 did
not show effects of outbreeding depression. It is also possible that, after the initial
population recovery, rounds of recombination and segregation result in a population

149
fitness that oscillates in a decreasing fashion about the mean parental fitness as less fit
recombinant phenotypes are exposed and selected against.

Morphological evolution and transgressive characters
While female hybrids showed increased measures over time, three of them (CTL,
ESL and ESA) being transgressive, there was little overall temporal patterns of female
morphometrics between generations one and thirteen. In general, male morphometrics
increased between generations one and thirteen for all treatments except for high salinity
hybrids which showed a decrease for almost all measures. Hybrid index scores indicated
that males became more RP-like in nature over time, this effect being stronger in the high
salinity environment. The fact that specific morphometrics were not significantly
different between parental treatments in each generation imposes limits on the ability to
distinguish morphological classes of hybrids well. This type of analysis can be enhanced
by the use of molecular data, since it has been established that molecular subdivision
exceeds morphological subdivision (Edmands and Harrison 2003), and is employed in
Chapter 4 of this work.
While the progeny of these divergent T. californicus populations demonstrate
outbreeding depression in both environmental treatments, we were also interested in
exploring whether hybridization may serve a creative role in the generation of novel
phenotypes. This phenomenon, known as transgressive segregation, has been observed in
several taxa (Albertson and Kocher 2005, Lexar et al. 2003, Rieseberg 1999) although it
has been studied most extensively in plants. These new combinations of genes could
supply the variation that allows for adaptation to new environments (Rieseberg et al.

150
2007, Latta et al. 2007). In Chapter 1, survivorship of 50:50 replicates composed of the
same parental populations exhibited transgressive survivorship after 15 months of
hybridization. Mixed population treatments had a mean survivorship of 87 percent while
parental population treatments RP and SD had mean survivorships of 52 and 56 percent
respectively. Fifteen months corresponds to approximately 11 generations in this study
and, at generation eleven, we did not find any evidence of transgressive survivorship in
either salinity treatment. However, the discrepancy seen in the two studies may be
explained in part by the fact that, in the Chapter 1, beneficial recombinants in early
generations were allowed to persist and contribute offspring to multiple subsequent
generations.
In the current study transgressive morphological characters were exposed during
the generations where 50:50 replicates exhibit lowest fitness. Even though recombinant
phenotypes that exceed parental values were created, our data would suggest that they did
not confer a fitness advantage and were perhaps selected against prior to subsequent
generations. One exception was egg sac size, which showed positive transgression at
generation thirteen but only for benign conditions (although it was also significantly
larger than the midparent in high salinity). This is one example hybridization may have
generated variation for increased egg sac size which was selected for over time. No
transgressive characters were observed in late generations for the high salinity
experiment. Our data do not provide evidence for the hypothesis that transgressive
hybrid morphology is favored in a novel or stressful environment. Instead, because a
specific character may not confer the same fitness advantage in both environments, the
selective forces that exist in the high salinity environment might act in such a way as to

151
limit the advantages of transgressive characters, as seen with the outcome of egg sac size.
While transgressive phenotypes are certainly created in a hybridizing T. californicus
swarm (Tables 3 and 4), the production of variants that are actually selected for over time
is apparently rare. It is likely that the fitness advantages conferred by most transgressive
phenotypes do not result in a selective force strong enough to overcome the forces of drift
in these finite population sizes.

Individual vs. comprehensive fitness measures
Competitive fitness in generation 14 showed that hybrid replicates recovered to
midparent fitness values in benign salinity, and exceeded midparent fitness in high
salinity This is consistent with individual clutch survivorship at generation 13 in that
hybrids in high salinity performed better than hybrids in the benign salinity, but it does
not give any indication of persistent effects of outbreeding depression in benign salinity
that were observed for individual clutch survivorship. In the same sense, competitive
survivorship both corroborates and contests the information obtained from
metamorphosis measurements. While metamorphosis at generation 13 also suggests that
outbreeding depression was purged in both salinity treatments, it does not show the
significantly higher fitness of high salinity replicates compared to the midparent. Even
though RP seems to be the superior parent, according to measures of survivorship and
morphology, it is competitively equivalent at normal salinity and competitively inferior at
high salinity. This may be due to smaller clutch sizes in RP, as found by Edmands and
Harrison (2003) and Edmands et al. (2005). Despite the lack of evidence for competitive
superiority of RP, morphological hybrid indices show hybrid males becoming

152
increasingly RP-like over time, demonstrating that neither parent is the clear winner
across all fitness and morphometric characters.

Is the outcome of hybridization more or less repeatable in a stressful environment?
As was stated previously in Chapter 1, we were particularly interested in
assessing the repeatability of hybrid replicates in order to determine whether evolutionary
outcomes are likely to be the same, given the same set of initial conditions. Studies exist
demonstrating that repeated hybridization events in plants have resulted in similar
outcomes (Brochman et al. 2000, Rieseberg 1996, Schwarzbach and Rieseberg 2002).
Rieseberg et al. 1996 showed that strong deterministic forces associated with
chromosomal rearrangements and strong epistatic interactions are involved in driving the
consequences of sunflower hybridization. The first long-term hybridization study in T.
californicus found only partial concordance among four hybrid replicates with each
replicate taking a different pathway of molecular evolution to arrive at a similar
conclusion of increased RP allele frequencies (Edmands et al. 2005). In Chapter 1 we
observed the outcomes of experimental swarms using the same populations for which
generations were not discrete. Observations of individual morphological characters did
not indicate repeatable outcomes, but principle components analysis was suggestive of
similar trends in morphological change among replicates of the same treatment.
Additionally, molecular data from surviving mixed population replicates arrived at higher
than expected RP allele frequencies as reported by Edmands et al. 2005.
In order for similar outcomes to occur, replicates must be subjected to similar
selection pressures and the changes that occur due to these pressures must have a

153
heritable genetic basis (Falconer and Mackay 1996). Otherwise, divergent paths of
evolution would be observed (Schwarzbach and Rieseberg 2002, Wade and Goodnight
1998). Even though strong evidence for deterministic forces was uncovered in studies of
Helianthus, Schwarzbach and Rieseberg also noted that there is an element of
contingency, in this case the environment, that is involved in producing hybrid
speciation. Our experimental study allows for us to impose identical selection pressures
on replicates and ask whether changes are associated with their hybrid genotypic
constitution.
In order to examine the morphological repeatability among replicates with the
same initial conditions, ANOVA of mean hybrid index was conducted for each salinity at
each generation for both males and females. In almost all generations of both sexes and
both salinities, a significant replicate effect was detected. When the variance of hybrid
index among 50:50 replicates of the same salinity was averaged over all generations,
there was no statistically significant difference in mean variance for both male and
female hybrid index scores. For morphological measures, there was no evidence for a
difference of repeatability between the two salinity treatments and significant p-values
for ANOVA at most generations indicates that replicates are not identical in their
trajectories of morphological evolution. For survivorship, significant replicate effects
were seen in three more generations for replicates reared at 35 ppt compared to the 53 ppt
environment and high salinity had a statistically smaller mean variance over all
generations. This outcome of greater repeatability among high salinity replicates can also
be seen by examining the trajectories of individual beakers over time (Figure 3-4). This
would suggest that survivorship is more deterministic and predictable for replicates

154
reared in more stressful environments. The same was not observed for metamorphosis,
perhaps due to little variation in fitness for development rate but stronger selection on
other fitness phenotypes of mature individuals in the high salinity treatment. Overall,
while patterns of repeatability are not strong, the more stressful environment produced a
more consistent pattern of repeatability for survivorship.
The deterministic forces that may exist in hybrid swarm evolution are likely to be
hampered by drift. T. californicus’ native habitat is composed of supralittoral pools that
experience extreme fluctuations in variables such as temperature and salinity, so it is
entirely possible that the evolution of this species is strongly influenced by drift
compared to selection. However, it may be the case that selection in the high salinity
environment was strong enough to overcome the force of drift, resulting in stronger
determinism for the 53 ppt replicates.

Heat Shock Resistance
After 14 generations of mating, effects of outbreeding depression were observed
in hybrids reared in the benign salinity environment. However, hybrids with a history of
being reared under high salinity stress for multiple generations displayed equivalent
survivorship to parentals. This suggests the presence of adaptive genetic variation in
initial hybrid generations that may allow for resistance to evolve and raises the question
of whether selection for resistance to one stress can confer resistance to other forms of
environmental stress.
Studies on animal species, particularly Drosophila, have repeatedly shown that
resistance to one stress often correlates with resistance to other stresses (Bubliy and

155
Loeschcke 2005, Hoffman and Parsons 1989, Hoffman and Parsons 1991). In the case of
this model species, discussions about the mechanism of resistance to multiple stresses
have centered around a reduction in metabolic rate, the accumulation of metabolic energy
reserves or the response of heat shock proteins (Bubliy and Loeschcke 2005). In one
example, starvation resistance was associated with increased glycogen levels (Djawdan et
al. 1998) and Chippindale et al. (1998) hypothesized that glycogen may also function to
bind extra water for increased desiccation resistance.
Harshman et al. (1999) proposed that a general response to stress selection may be
to ameliorate the immediate physiological impact of the stress factor. Tigriopus is known
to do this in the case of osmotic stress by synthesizing osmolytes (Goolish and Burton
1989., Burton and Feldman 1983, Willett and Burton 2002) and upregulating
glutathionine reductase, an enzyme with antioxidant function (Seo et al. 2006). Proline,
the primary osmolyte made by Tigriopus during hyperosmotic stress, is also known to
accumulate in E. coli in response to other environmental changes including drought, low
temperature and SO
2
(Le Rudelier et al 1984). However, correlations between osmotic
and thermal stress, such as we have observed here, on a physiological basis, have yet to
be explored. It has also be shown in yeast that slow growth is a stress itself and that slow-
growing cells exhibit resistance to heat shock (Lu et al. 2009). T. californicus individuals
reared in a high salinity environment do show slower growth rates and it is possible that
is selection for resistance to the stress of a depressed developmental rate, and not
necessarily resistance to salinity stress, that also confers resistance to heat shock.

156
Conclusions
The examination of multiple generations of hybridization over an extended period
of time has allowed us to examine some of the transitory stages of hybridization that
occur before a swarm has stabilized. By partitioning the fitness outcomes of early hybrid
generations we were able to determine that outbreeding depression is equally as severe
for T. californicus in a stressful or novel environment. However, the generations of
maximum outbreeding depression and recovery both occur up to two generations earlier
in high salinity replicates. We also determined that patterns of repeatability are stronger
in a more stressful environment. For managers hoping to make predictions about the
result of mixing populations this could offer an animal model to contrast with known
plant examples. Although transgressive phenotypes were generated as a result of
hybridization, they did not show any evidence of increased fitness compared to parentals
when exposed to a high salinity environment. Thus, the ability for hybridization to act as
a creative force to generate new recombinant phenotypes with enhanced fitness compared
to parentals may be limited in this species. The more stressful environment did, however,
result in hybrid swarm replicates that were more fit than their benign condition
counterparts when exposed to a novel stress, revealing another metric by which we can
conclude that stress may alleviate some of the effects of outbreeding depression.

157
CHAPTER 3 REFERENCES

Albertson RC, Kocher TD (2005) Genetic architecture sets limits on transgressive
segregation in hybrid cichlid fishes. Evolution 59, 686-690.

Allendorf FW, Leary RF, Sprunell P, Wenburg JK (2001) The problems with hybrids:
setting conservation guidelines. Trends in Ecology and Evolution, 613-622.

Armbruster P, Bradshaw WE, Holzapfel CM (1997) Evolution of the genetic architecture
underlying fitness in the pitcher plant mosquito, Wyeomyia smithii. Evolution 51,
451-458.

Bijlsma R, Loeschcke V (2005) Environmental stress, adaptation and evolution: an
overview. Journal of Evolutionary Biology 18, 744-749.

Brochmann C, Borgen L, Stabbetorp OE (2000) Multiple diploid hybrid speciation of the
Canary Island endemic Argyranthemum sundingii (Asteraceae). Plant Systematics
and Evolution 220, 77-92.

Bubliy OA, Leoschcke V (2005) Correlated responses to selection for stress resistance
and longevity in a laboratory population of Drosopila melanogaster. Journal of
Evolutionary Biology 18, 789-803.

Burke JM, Arnold ML (2001) Genetics and the Fitness of Hybrids. Annual Review of
Genetics 35, 31-52.

Burke JM, Carney SE, Arnold ML (1998) Hybrid fitness in the Louisiana irises:
Analysis of parental and F1 performance. Evolution 52, 37-43.

Burton RS (1987) Differentiation and integration of the genome in populations of the
marine copepod Tigriopus californicus. Evolution 41, 504-513.

Burton RS (1990) Hybrid breakdown in developmental time in the copepod Tigriopus
californicus. Evolution 44, 1814-1822.

Burton RS (1997) Genetic evidence for long term persistence of marine invertebrate
populations in an ephemeral environment. Evolution 51, 993-998.

Burton RS, Feldman MW (1983) Physiological effects of an allozyme ploymorphism:
Glutamate-pyruvate transaminase and response to hyperosmotic stress in the
copepod Tigriopus californicus. Biochemical genetics 21, 239-251.

Burton RS, Lee B-N (1994) Nuclear and mitochondrial gene genealogies and allozyme
polymorphism across a major phylogeographic break in the copepod Tigriopus
californicus. Proceedings of the National Academy of Sciences USA 91, 5197-
5201.

158
Chippindale AK, Gibbs AG, Sheik M, et al. (1998) Resource acquisition and the
evolution of stress resistance in Drosophila melanogaster. Evolution 52, 1342-
1352.

Christiansen FB (2008) Theories of Population Variation in Genes and Genomes
Princeton University Press, Princeton, NewJersey.

Djawdan M, Chippindale AK, Rose MR, Bradley TJ Metabolic reserves and evolved
stress resistance in Drosophila melanogaster. Physiological Zoology 71, 584-594.

Edmands S (1999) Heterosis and outbreeding depression in interpopulation crosses
spanning a wide range of divergence. Evolution 53, 1757-1768.

Edmands S (2001) Phylogeography of the intertidal copepod Tigriopus californicus
reveals substantially reduced population differentiation at northern latitudes.
Molecular Ecology 10, 1743-1750.

Edmands S (2007) Between a rock and a hard place: evaluating the relative risks of
inbreeding and outbreeding for conservation and management. Molecular
Ecology 16, 463 - 475.

Edmands S, Deimler JK (2004) Local adaptation, intrinsic coadaptation and the effects of
environmental stress on interpopulation hybrids in the copepod Tigriopus
californicus. Journal of Experimental Marine Biology and Ecology 303, 183-196.

Edmands S, Feaman HV, Harrison JS, Timmerman CC (2005) Genetic consequences of
many generations of hybridization between divergent copepod populations.
Journal of Heredity 92, 114-123.

Edmands S, Harrison JS (2003) Molecular and quantitative trait variation within and
among populatioins of the intertidal copepod Tigriopus californicus. Evolution
57, 2277-2285.

Falconer DS, Mackay TFC (1996) Introduction to quantitative genetics, 4 edn. Longman
Science and Technology, Harlow, UK.

Fenster CB, Galloway LF (2000) Population differentiation in an annual legume: genetic
architecture. Evolution 54, 1157-1172.

Fitzpatrick BM, Bradley H (2004) Environment-dependent admixture dynamics in a tiger
salamander hybrid zone. Evolution 58, 1282-1293.

Frankham R (2005) Stress and adaptation in conservation genetics. Journal of
Evolutionary Biology 18, 750-755.

159
Ganz HH, Burton RS (1995) Genetic differentiation and reproductive incompatibility
among Baja California populations of the copepod Tigriopus californicus. Marine
Biology 123, 821-827.

Goolish EM, Burton RS (1989) Energetics of osmoregulation in an intertidal copepod:
effects of anoxia and lipid reserves on the pattern of free amino acid
accumulation. Functional Ecology 3, 81-89.

Griffing B (1990) Use of a controlled-nutrient experiment to test heterosis hypotheses.
Genetics 126, 753-767.

Gross BL, Rieseberg LH (2004) The ecological genetics of homoploid hybrid speciation.
Journal of Heredity 96, 241-252.

Harrison JS, Edmands S (2006) Chromosomal basis of viability differences in Tigriopus
californicus interpopulation hybrids. Journal of Evolutionary Biology 19, 2040-
2051.

Harrison JS, Peterson DL, Swain JR, Edmands S (2004) Microsatellite DNA markers for
the intertidal copepod Tigriopus californicus. Molecular Ecology Notes 4, 736-
738.

Harshman LG, Hoffmann AA, Clark AG (1999) Selection for starvation resistance in
Drosophila melanogaster: physical correlates, enzyme activities and multiple
stress responses. Journal of Evolutionary Biology 12, 370-379.

Hedgecock D, Lin J-Z, DeCola S, et al. (2007) Transcriptomic analysis of growth
heterosis in larval Pacific oysters (Crassostrea gigas). Proceedings of the
National Academy of Sciences 104, 2313-2318.

Hoffman AA, Parsons PA (1991) Evolutionary Genetics and Environmental Stress
Oxford University Press, Oxford.

Hoffmann AA, Parsons PA (1989) Selection for increased dessication resistance in
Drosophila melanogaster: additive genetic control and correlated responses for
other stresses. Genetics 122, 837-845.

Johansen-Morris A, Latta R (2006) Fitness consequences of hybridization between
ecotypes of Avena barbata: hybrid breakdown, hybrid vigor and transgressive
segregation. Evolution 60, 1585-1595.

Johnston JA, Donovan LA, Arnold ML (2004) Novel phenotypes among early generation
hybrids of two Louisiana iris species: flooding experiments. Journal of Ecology
92, 967-976.

160
Latta RG, Gardner KM, Johansen-Morris AD (2007) Hybridization, recombination and
the genetic basis of fitness variation across environments in Avena Barbata
Genetica 129, 167-177.

Lexar C, Welch ME, Raymond O, Rieseberg LH (2003) The origin of ecological
divergence in Helianthus paradoxus (Asteraceae): selection on transgressive
characters in a novel hybrid habitat. Evolution 57, 1989-2000.

Lu C, Brauer MJ, Botstein D (2009) Slow growth induces heat-shock resistance in
normal and repiatory-deficient yeast. Molecular Biology of the Cell 20, 891-903.

Lynch M (1991) The genetic interpretation of inbreeding depression and outbreeding
depression. Evolution 45, 622-629.

Lynch M, Walsh B (1998) Genetics and Analysis of Quantitative Traits Sinauer,
Sunderland, MA.

Palmer CA, Edmands S (2000) Mate choice in the fact of both inbreeding and
outbreeding depression in the intertidal copepod Tigriopus californicus. Marine
Biology 136, 693-698.

Ranganath H, Aruna S (2003) Hybridization, transgressive segregation and evolution of
new genetic systems in Drosophila. Journal of Genetics 82, 163-177.

Rawson PD, Burton RS (2002) Functional coadaptation between cytochrome c and
cytochrome c oxidase within allopatric populations of a marine copepod.
Proceedings of the National Academy of Sciences USA 99, 12955-12958.

Rieseberg LH, Archer MA, Wayne RK (1999) Transgressive segregation, adaptation and
speciation. Heredity 83, 363-372.

Rieseberg LH, Baird SJE, Desrochers AM (1998) Patterns of mating in wild sundlower
and hybrid zones. Evolution 52.

Rieseberg LH, Kim S-C, Randell RA, et al. (2007) Hybridization and the colonization of
novel habitats by annual sunflowers. Genetica 129, 149-165.

Rieseberg LH, SInervo B, Linder CR, Ungerer MC, Arias DM (1996) Role of gene
interactions in hybrid speciation: evidence from ancient and experimental
hybrids. Science 272, 741-745.

Rieseberg LH, Widmer A, Arntz AM, Burke JM (2003) The genetic architecture
necessary for transgressive segregation is common in both natural and
domesticated populations. Philosophical Transactions of the Royal Society of
London B 358, 1141-1147.

161
Rudulier DL, Strom AR, Dandekar AM, Smith LT, Valentine RC (1984) Molecular
biology of osmoreguation. Science 224, 1064-1068.

Schwarzbach AE, Rieseberg LH (2002) Likely multiple origins of a diploid hybrid
sunflower species. Molecular Ecology 11, 1703-1715.

Seehausen O (2004) Hybridization and adaptive radiation. Trends in Ecology and
Evolution 19, 198-207.

Seo JS, Lee K-W, Rhee J-S, et al. (2006) Environmental stressors (salinity, heavy metals,
H
2
O
2
)modulate expression of glutathione reductase (GR) gene from the intertidal
copepod Tigriopus japonicus. Aquatic Toxicology 80, 281-289.

Tallmon DA, Luikart G, Waples RS (2004) The alluring simplicity and complex reality
of genetic rescue. Trends in Ecology and Evolution 19, 489-496.

Turelli M, Orr HA (2000) Dominance, epistasis and the genetics of postzygotic isolation.
Genetics 154, 1663-1679.

Vetukhiv MO, Beardmore JA (1959) Effect of environment upon the manifestation of
heterosis and homeostasis in Drosophila pseudoobscura. Genetics 44, 759-768.

Vittor BA (1971) Effects of the environment on fitnbess-related life history character in
Tigriopus californicus Dissertation, University of Oregon.

Wade MJ, Goodnight CT (1998) The theories of Fisher and Wright in the context of
metapopulations: When nature does many small experiments. Evolution 52, 1537-
1553.

Willett CS, Burton RS (2001) Viability of cytochrome c genotypes depends on
cytoplasmic backgrounds in Tigriopus californicus. Evolution 55, 1592-1599.

Willett CS, Burton RS (2002) Proline biosynthesis genes and their regulation under
salinity stress in the euryhaline copepod Tigriopus californicus. Comparative
Biochemistry and Physiology Part B 132, 739-750.
Willett CS, Burton RS (2003) Environmental influences on epistatic interactions:
viabilities of cytochrome c genotypes in interpopulation crosses. Evolution 57,
2286-2292.

162
Chapter 4

Stronger patterns of molecular repeatability observed for hybrid swarm replicates
under salinity stress

CHAPTER 4 ABSTRACT

When moderately incompatible Tigriopus populations mix, the outcome is likely
to be recovery from outbreeding depression with extensive introgression in advanced
generations of hybridization. For this study, discrete generations of experimental hybrid
swarm replicates were initiated with equal numbers of gravid females from Royal Palms
and San Diego, California. Hybrid swarm replicates were genotyped at generation seven
and generation thirteen for eleven microsatellite loci, two proline biosynthesis genes,
cytochrome oxidase I and cytochrome c. Microsatellites showed an overall trend of
increased RP alleles from expected starting frequencies. Replicate beakers for the 35 ppt
treatment, with the same initial starting conditions, diverged greatly from each other and
from the high salinity replicate beakers while high salinity replicates showed an overall
pattern of repeatability. Strong evidence for both nuclear-nuclear and nuclear-
cytoplasmic coadaptation was observed at generation seven. Weaker evidence at
generation thirteen may be due to the breakup of linkage between marker loci and loci
under selection. This study, showing that stronger selection under stressful conditions
overcame drift, is concordant with previously reported data where increases in RP-like
morphology and stronger repeatability of survivorship were observed in a higher-stress

163
environment. While this work shows that deterministic forces play a role in hybrid swarm
evolution, they may be minimized due to the contingency associated with small, isolated
populations.

CHAPTER 4 INTRODUCTION

An understanding of the genetic mechanisms that determine hybrid fitness is of
great concern to the fields of evolutionary and conservation biology. Anthropogenic
activities can result in both the intentional translocation of individuals with the intention
of diversifying gene pools (Allendorf et al 2001, Edmands 2007), and the accidental
relocation of species such that once-isolated populations face the risk of collapsing into
hybrid swarms (Seehausenn et al 2008). The indirect effects of hybridization and
introgression have been considered to be a threat to the biodiversity of some native
species (Barilani et al. 2005, Halis and Morley 2005). In addition to offering insight for
species management, the ability to determine the fitness of mixed populations has
implications for understanding the role of hybridization in reproductive isolation or
speciation.
Hybrid fitness is difficult to predict since both beneficial and deleterious
genotypes may be created from the hybridization event. Heterosis, where hybrids show
an increase in fitness compared to parentals, may occur due to the masking of deleterious
recessive alleles, overdominance or beneficial epistasis. Another outcome might be
outbreeding depression where hybrid offspring show a decrease in fitness compared to

164
parents. Outbreeding depression in the first generation of hybridization can be a result of
underdominance, disruption of local adaptation or epistatic interactions (Lynch 1991).
Often times a reduction in fitness is delayed until the F2 resulting from recombination
and the breakup of coadapted genomic regions. The most common model used to explain
this is one described by Dobzhansky and Muller, who proposed that natural selection
within isolated populations will favor sets of alleles showing positive epistasis which may
result in the development of coadapted gene complexes (Turelli and Orr 2000). If
isolated populations hybridize, new combinations of alleles that have not been previously
tested together may result in deleterious interactions. One study (Fenster and Galloway
2000) of a legume demonstrated that outbreeding depression was not evident until the F3,
and concluded that an extra generation of recombination was necessary to break up
tightly linked coadaptation.
Interpopulation hybridization in Tigriopus typically results in F1 heterosis and F2
hybrid breakdown for several fitness characters (Burton 1986, 1987, 1990) with the level
of outbreeding depression observed in the F2 being correlated with genetic divergence
(Edmands 1999). Fitness beyond the F2 is a challenge to predict because rounds of
recombination that disrupt more tightly linked coadapted complexes could result in
declines, or fitness may increase as selection promotes favorable hybrid combinations. In
order to assess the merits of mixing gene pools, it is important to study hybridizing
populations over multiple generations. The intertidal copepod Tigriopus californicus has
a short generation time (~23 days, Burton 1987) facilitating determination of the genetic
composition of the population in transitory states of evolution.

165
Much work has been done using T. californicus as a model for understanding
cytonuclear coevolution in early generation hybrids. Some of the earlier work focuses on
the mitochondrial and nuclear encoded complex Cytochrome c oxidase (COX) and the
nuclear encoded locus cytochrome c (CYC), both components of the electron transport
system. Studies examining enzyme-substrate activity have observed higher COX activity
when CYC and COX are from the matching populations (Rawson and Burton 2002;
Harrison and Burton 2006). Ellison and Burton (2008) demonstrated that low F3 hybrid
fitness is restored in maternal backcrosses, which have a full haploid nuclear genome
matching the mitochondrial genome, but this same pattern of recovery is not observed in
paternal backcrosses. A different study (Ellison and Burton 2006) also confirmed that
only electron transport enzymes with both nuclear and mitochondrial subunits display
reduced fitness in hybrids, and that hybrids with mtDNA and nuclear-encoded mtRNA
polymerase from the same population have higher fitness. In the past several years, work
has also show that T. californicus may harbor maladapted gene combinations, as
interpopulation crosses show nuclear alleles favored on the wrong cytoplasmic
background (Willett and Burton 2001, Willett 2006, Edmands et al. in press, Appendix).
The extent to which long-term hybridization results in selection for coadapted complexes
that have been tested out in previous environments, or for highly fit recombinants being
exposed for the first time, is difficult to determine. Outcomes may result in variable
combinations of both, and stochastic events may play a significant role in determining
which loci or gene combinations could become fixed.
Some studies have produced evidence for both nuclear-nuclear and nuclear-
mitochondrial coadaptation (Burton 1987, Edmands 1999, Willett and Burton 2001,

166
Rawson and Burton 2002). Willett (2006) showed that, for F2 hybrids, double parental
homozygotes, double heterozygotes and double nonparental homozygotes tended to be
present in numbers greater than Mendelian expectations, suggesting not only nuclear-
nuclear coadaptation but also complex dominance or epistatic relationships. After many
generations, as with cytonuclear coadaptation, selection may favor fit parental genotypes,
or newly generated recombinant genotypes that have greater fitness than parentals.
Tigriopus inhabits the highest supralittoral tidepools, in what is known as the
splash zone, and these copepods undergo wide fluctuations in environmental conditions,
particularly temperature and salinity. Tidepool temperatures have been found to vary
from 4˚C to 35˚C (Edmands and Deimler 2004) and both evaporation and fresh rain water
can cause large fluctuations in salinity. Following an increase in environmental salinity,
Tigriopus is known to accumulate intracellular proline, alanine and glycine (Burton and
Feldman 1982). with proline showing the largest increase in concentration (Goolish and
Burton 1989). Both the pyrroline-5-carboylate-reductase (P5CR) and pyrroline-5-
carboxylate-synthetase (P5CS) genes are involved in the pathway that synthesizes proline
from glutamate. These genes may be candidates for nuclear-nuclear coadaptation.
The ranges of different environmental conditions may vary from population to
population, but a short-term study of environmental stress on hybrids found very little
evidence for local adaptation and that F2 breakdown was partially alleviated under
thermal stress (Edmands and Deimler 2004). Willett and Burton (2003) showed a
substantial effect of temperature/light regime on selection at the cytochrome c locus in
interpopulation hybrids that is consistent with environmental effects on cytonuclear

167
epistasis, but little is known about the environmental effects on populations hybridizing
for many generations.
The first long-term hybridization experiment on Tigriopus beginning with
backcross individuals (Edmands et al. 2005) showed that, after one year, hybridity in four
replicates increased, suggesting that hybrid fitness problems in early generations are a
weak barrier to introgression. This result was supported by the findings in Chapter 1 of
this dissertation, where replicate swarms of two divergent populations persisted through
15 months of free mating and showed fitness measures higher than the expected
midparent value. Chapter 2 showed that two nearly incompatible populations were not
driven to extinction by severe early-generation fitness reductions but resembled superior
parent equivalents in morphology and molecular characteristics. Molecular data beyond
the initial study by Edmands et al. 2005 has been limited in providing a detailed account
of the long term molecular evolution of these swarms.
This study provides the most detailed account to date of the molecular
consequences of long-term hybridization in T. californicus. We looked at the molecular
composition of hybrid swarms reared under both benign and stressful environments.
Repeatability among replicate trials was assessed within and between two different
salinity environments. Evidence for coadapted gene complexes was examined in both
nuclear-nuclear and nuclear-cytoplasmic protein coding genes as well as noncoding
microsatellite markers, covering a total of 9 of the 12 Tigriopus chromosomes.

168
METHODS

The Tigriopus californicus study system
T. californicus is an excellent model for studies of long-term experimental
hybridization for multiple reasons. Large sample sizes are easily collected from the wild
and samples can be easily maintained in culture or used for studies that require breeding
manipulation. Mature males as well as gravid and virgin females can be easily identified
because males use their enlarged claspers to guard females until they are mature enough
to be fertilized and gravid females carry an egg sac that is visible to the naked eye. T.
californicus has a short minimum generation time (approximately 23 days at 20˚C;
Burton 1987) which makes it appropriate for long-term hybridization studies.
Studies using a variety of molecular markers including allozymes, DNA
sequences and microsatellites show that, despite potential for high gene flow, populations
of T. californicus within as little as 500 m remain genetically differentiated (Burton 1997,
Edmands 2001, Edmands and Harrison 2003). Mitochondrial COI DNA sequences have
been assayed in populations extending from Alaska to Baja California, Mexico and show
that populations differ from 0.2 to 23% (Burton and Lee 1994, Edmands 2001) and full
mitochondrial genome sequences are available for some populations (Burton et al. 2007).
Genetic distance is correlated with geographic distance, with decreased interpopulation
genetic divergence in the northernmost regions (Edmands 2001).

169
Population Sampling
Populations were sampled from two southern California locations, Royal Palms,
CA (RP, 33° 42’ N, 118° 19’ W) and San Diego, CA (SD, 32° 45’ N, 117° 15’ W) in
December 2005. These two populations show approximately 18% mitochondrial
cytochrome oxidase I divergence (Edmands 2001). Crosses between them are known to
produce F1s with a slight increase in fitness measures, F2s with decreased fitness and
variable patterns of backcross fitness (Edmands et al. 2005, Chapter 1). Samples were
maintained as mass cultures in 400 ml beakers with 350 ml filtered seawater (37 μm) and
algal food supplements and housed in a 20ºC incubator with a 12 h light: 12 h dark cycle.

Long-Term Hybrid Swarms
Three different culture treatments were initiated (100% RP, 100% SD, and 50%
RP:50% SD) by placing 100 gravid females in 400 ml beakers full of filtered seawater
containing 50 mls live Platymonas culture. Each beaker was supplemented with finely
ground Spirulina and Tetramin flakes at a concentration of 0.2 mg/ml. In January 2006
12 replicates of each culture treatment were set up for each of two salinities: 35 ppt
(benign conditions) and 53 ppt (salinity stress). Prior to the setup of the experiment a 14-
day test of algal growth rate was performed to confirm that Platymonas grown under lab
conditions did not have significantly different growth rates in the two chosen salinities. A
pilot study was also done to determine that T. californicus growth and/or development
(size at two weeks) is compromised under high salinity conditions (Chapter 3). Beakers
were housed together in one incubator at 20 °C set to a 12h light: 12 h dark cycle.

170
Beakers were monitored each week for the presence of copepodids (juvenile
copepods). Early stage copepods were distinguished from adult females based upon size
and color. If juveniles were present, all adult females were removed from the beaker.
One week later, all adult females were again removed. This was done as a way to double
check that all females of the previous generation were removed. Beakers were then
monitored once a week and, as soon as adult females with eggs were observed, they were
transferred to new beakers with fresh algae-seawater-dry food mixture to start the next
generation. Adult gravid females were transferred to the new generation beaker for the
next two or three weeks depending on the presence of absence of new juveniles. When
juveniles were observed in the new beaker, adult females were removed. In this way,
generations were maintained as discrete. Once a week all beakers were fed and rotated
within the incubator in order to homogenize light exposure.
For selected generations, 20 gravid females and 20 mature males were removed
from each replicate beaker and were used for morphometric assays. Females were also
used for fitness assays. All copepods were either returned to their source beakers after
assays were completed, or were frozen for molecular analysis. Replicates were
maintained for 14 generations which corresponded to 18 months.

Microsatellite Assays
Eleven microsatellite loci were scored for hybrid replicates at generation 7 and
13. These microsatellites were developed using an enriched DNA library from the RP
population (Harrison et al. 2004). DNA was extracted from individual copepods using
the lysis protocol previously described in Edmands et al. 2005. Individual copepods were

171
incubated in 50 μl lysis buffer at 65°C for 1 hour followed by 100°C for 15 minutes.
Polymerase chain reactions were carried out in 12 μl volumes containing 0.5 μl template
DNA, 0.25 μM fluorescently labeled forward primer, 1 μM reverse primer and 2.5 mM
MgCl
2.
Temperature cycling was as follows: 5 min denaturation at 94°C; 35 cycles of 30
s at 94°C, 35 s at 55°C, and 30 s at 72°C; 5 min at 72°C. This was with the exception of
locus 480 which required an annealing temperature of 62°C. Fluorescently labeled PCR
products were run on a Beckman-Coulter CEQ 8000 Capillary Sequencer (Beckman
Coulter, Fullerton, CA USA) according to commercially recommended protocols. Allele
sizes were scored by eye.

Genotyping of Cytochrome Oxidase I and Cytochrome c
A segment of the COI gene was amplified using primers COIVHel (5’-
TACACCTCAGGATGTCCAAAAAATCA-3’) and COIVLel (5’ –
GAGGGGCTACGAACCACAAAGATA-3’) modified from Folmer et al. (1994) to
match Tigriopus COI sequence. Polymerase chain reactions were carried out in 12 μl
volumes containing 0.5 μl template DNA, 1 μM forward primer, 1 μM reverse primer and
2.5 mM MgCl
2.
Temperature cycling was as follows: 5 min denaturation at 94°C; 40
cycles of 30 s at 94°C, 35 s at 50°C, and 1 min at 72°C; followed by 5 min at 72°C. PCR
products were restriction enzyme digested by HinfI at 60°C for 1 hour. HinfI cleaves a
restriction site that is present in the RP COI sequence, but not the SD sequence.
Restriction digests were run out electrophoretically on 1.2% agarose gels for one hour at
100 V.

172
Primers for the cytochrome c gene developed by Willett and Burton (2001) were
used to amplify and score population specific PCR products from hybrid individuals.
Primers ccAB.f and cyt14.r which were designed to amplify a 625 base pair fragment in
the Abalone Cove population also amplify the same fragment in the RP population used
in this study. Primers ccSD.f and cyt14r amplify a 394 bp fragment in the SD population.
We did not see successful amplification using the multiplexed protocol of Willett and
Burton, so two PCRs (one for each population’s allele) were run for each individual.
Temperature cycling conditions for both primer sets were as follows: 5 min denaturation
at 94°C; 35 cycles of 30 s at 94°C, 35 s at 60°C, and 1 min at 72°C; followed by 5 min at
72°C. Samples were run out electrophoretically on 1.2% agarose gels for one hour at
100 V.

Genotyping of Proline Biosynthesis Genes
Sequences for δ1-pyrroline-5-carboylase-synthase (P5CS) and δ1-pyrroline-5-
carboylase-reductase (P5CR) from the San Diego population (Willett and Burton, 2002)
were used to design primers that would amplify a segment of these genes in both the RP
and SD populations. These genes had been previously characterized by Willett and
Burton (2002). Primers P5CS_FE (5’-GCGTCATGCACAACTTCAAT-3’) and
P5CS_RC (5’-TCGTTGACGCTCATGATTTC-3’) amplify a 436 bp fragment of the
P5CS gene and primers P5CR_FC (5’-ACACCACAAGACAACGCAAC-3’) and
P5CR_RC(5’-ACATGCATTCCTCATCCGTA-3’) amplify a 570 bp fragment of the
P5CR gene. Polymerase chain reactions were carried out in 20 μl volumes containing 1 μl
template DNA extraction, 1 μM forward primer, 1 μM reverse primer and 2.5 mM

173
MgCl
2.
Reaction temperature conditions were as follows for P5CR: 94°C for 5 min
followed by 35 cycles of 30 s at 94°C, 35 s at 57°C, and 1 min at 72°C; completed by a 5
min extension step at 72°C. Temperature cycling conditions for locus P5CS were
identical with the exception of a 61°C annealing temperature. PCR products were run out
on 1.2% agarose gels to confirm the presence of amplified product, after which fragments
were sequenced on an ABI capillary sequencer. Alleles for the P5CR locus were
identified by scoring three different population-specific SNPs. Alleles for the P5CS
locus were also identified based upon three different population-specific SNPs, and
heterozygotes for P5CS were easily identified by the presence or absence of a 1bp
insertion in the San Diego population.

Molecular Analyses
Deviations from Hardy-Weinberg Equilibrium were detected using the F
IS

statistic, and genotypic disequilibrium was calculated for each pair of loci in each
replicate using GENEPOP 1.2 (Raymond and Rousset 1995) with 10,000
dememorization steps, 20 batches, and 5,000 iterations. Weir and Cockerham’s overall
FST and pair-wise FSTs were calculated for all replicates using 10,000 permutations in
GENETIX 4.04 (Belkir et al. 2000). A factorial correspondence analysis (FCA) was also
conducted for all groups in GENETIX (Belkir et al. 2000). Locus-by-locus analysis of
molecular variance (AMOVA) was run with Arlequin 3.1 (Excoffier et al. 2006) using
10,000 permutations. The program STRUCTURE (Pritchard et al. 2000), a Bayesian
assignment test, was used to test the number of populations (K) that had the highest
posterior probability. The admixture model was used with burn-in set to 50,000 and

174
Monte Carlo Markov Chains (MCMC) steps set to 85,000, with K set to the number of
replicate groupings (2 to 7). Each run was conducted 3 times. The effective population
size was estimated for each replicate for which microsatellite data was collected at both
generations seven and thirteen. Estimates were obtained with Nb_HetEx (Zhdanova and
Pudovkin 2008) using the temporal method of Waples (1989). The expected relationships
between F
ST
and heterozygosity using coalescent simulations of an island model were
obtained using Fdist2 (Beaumont and Nichols 1996). The procedure was performed
separately for each salinity. Simulations were repeated 20,000 times under the infinite
alleles model and an interval in which 95% of the F
ST
estimates are expected to lie was
constructed. Loci with F
ST
values outside of the 95% interval were marked as candidates
for selection.
Hybrid index scores for each individual were calculated by assigning a 0 for each
RP allele and a 1 for each SD allele. Alleles shared between populations (found in loci
558, 197 and 56J2), were given a score equal to the probability that the allele came from
SD. Individual hybrid index was determined by averaging over the number of loci scored
for each individual.
Calculation of allele and genotype frequencies, 2-locus contingency tables and
linkage disequilibrium was performed by Genepop 4.0 (Raymond and Rousset 1995). To
further test for epistatic interactions, observed two-locus genotype numbers were
compared to expected numbers determined by multiplying single-locus ratios. Paired
two-tailed t-tests were used to compare observed vs. expected genotype numbers.

175
RESULTS

Summary statistics
In order to provide an estimate of allele frequencies in each population sample, 40
individuals, sampled within a week from each population collection, were genotyped for
all eleven microsatellite loci. A maximum of 40 individuals from select (one replicate
per treatment) parental population control replicates were scored for the eleven
microsatellite loci at generations one and thirteen. Ten individuals from all other
replicates were scored at generation thirteen to confirm that culturing errors had not
resulted in cross contamination of replicate beakers.
At generations seven and thirteen a maximum of 40 individuals per hybrid
replicate were scored for each of eleven microsatellites. Tests of genic differentiation
across all loci returned highly significant p-values (p<0.0001) for all replicate pairs,
indicating that the allelic distribution is not identical between any two replicates at either
generation seven or generation thirteen. In the same manner, tests of genotypic
differentiation across all loci returned highly significant p-values (p<0.0001) for all
replicate pairs, indicating that genotypic distribution was not identical between any two
replicates.
At generation seven for 35 ppt replicates, RP alleles had increased from the
expected starting frequency of 0.5 at all loci in replicate two, but decreased at almost all
loci (21 out of 22 cases) in replicates one and five (Figure 4-1a). Replicate five showed
particularly extreme reduction in RP allele frequencies, with no RP alleles scored for
locus 480. Replicate four was intermediate to the other three replicates, as seven out of

176
Figure 4-1. RP allele frequencies for eleven microsatellite loci for each replicate.
Horizontal dashed line shows the expected frequency of 0.5. (a) Generation 7, 35 ppt; (b)
generation 7, 53 ppt; (c) generation 13, 35 ppt; (d) generation 13, 53 ppt.

(a)
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
1202(11)
1814(9)
558(1)
1203(5)
228(2)
197(9)
56J2(7)
62J8(5)
480(8)
30(1)
1555(10)
RP Allele Frequency
1
2
4
5

177
Figure 4-1, Continued

(b)
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
1202(11)
1814(9)
558(1)
1203(5)
228(2)
197(9)
56J2(7)
62J8(5)
480(8)
30(1)
1555(10)
RP Allele Frequency
1
2
4
5

(c)
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
1202(11)
1814(9)
558(1)
1203(5)
228(2)
197(9)
56J2(7)
62J8(5)
480(8)
30(1)
1555(10)
RP Allele Frequency
1
4
5

178
Figure 4-1, Continued

(d)
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
1202(11)
1814(9)
558(1)
1203(5)
228(2)
197(9)
56J2(7)
62J8(5)
480(8)
30(1)
1555(10)
RP Allele Frequency
1
2
4
5

179
eleven loci showed increases in RP alleles. At generation seven in the 53 ppt replicates,
replicate five showed fewer than expected RP alleles at all loci (Figure 4-1b). For the
other three high salinity replicates, 31 out of 33 cases showed increases in RP alleles. At
generation thirteen 35 ppt replicate number five continued to show extreme deviation
from expected allele frequencies in the SD direction, with three loci (1814, 1203 and
1555) indicating fixation for SD alleles (Figure 4-1c). Of the remaining two benign
salinity replicates, 13 out of 22 cases exhibited increases in RP alleles. For high salinity
replicates, RP alleles increased from the expected starting frequency in all cases except
two (locus 1202 in replicate 2 and locus 480 in replicate 5) (Figure 4-1d) and replicate
four had two cases of potential fixation for RP alleles (loci 1814 and 197).
F
IS
estimates were calculated for each replicate (Table 4-1) and for males and
females separately (Table 4-2). No single locus showed a significant deviation from
HWE in all replicates of a salinity treatment. However some loci, such as 1202 in high
salinity replicates and 1814 in benign salinity replicates, showed significant deviations in
three out of four replicates. The frequency of heterozygote excess and deficit was similar
for the two salinity treatments at generation seven. When replicates for each salinity
treatment were calculated separately for males and females there are many more cases of
heterozygote deficit seen in males at generation seven. Cases of heterozygote excess are
equal between males and females. The large difference in heterozygote deficit between
makes and females is not observed in generation thirteen.
U-tests (Raymond and Rousset 1995) were used to test for deviations from HWE
across all loci and all replicates. For generation seven at 35 ppt, significant deficits were
found for replicate one and two loci (1202 and 1813) while significant excesses were

180
Table 4-1. F
IS
values calculated for each replicate using 11 microsatellite loci with
significant values noted in bold (*) (p < 0.05) Values significant after a Bonferroni
correction are shown in italics. Dashes indicate values that could not be estimated.
Results shown for generation seven (a) and generation thirteen (b).

(a)
35 ppt 53 ppt
Chromosome_Locus 1 2 4 5 1 2 4 5
1_30 +0.296 -0.121 -0.010 -0.013 -0.101 -0.198 -0.276 -0.239
1_558 -0.118 -0.101 -0.010 -0.013 -0.013 -0.070 -0.139 -0.530*
2_228 -0.010 -0.088 -0.093 -0.056 -0.226 +0.295 -0.350* +0.409*
5_1203 -0.393* -0.211 -0.014 - +0.069 +0.153 +0.110 +0.378*
5_62J8 -0.331* -0.245* -0.037 -0.082 -0.382* +0.020 -0.118 +0.370*
7_56J2 +0.102 -0.144 -0.240* -0.027 +0.230 +0.151 +0.131 +0.005
8_480 -0.082 +0.158 +0.024 - +0.058 +0.011 +0.173 -0.028
9_197 +0.226 +0.001 -0.083 -0.018 -0.146 -0.033 -0.076 +0.163
9_1814 +0.089* +0.226* -0.203* -0.136 -0.086 0.000 -0.150 +0.204
10_1555 -0.307 +0.047 -0.033 -0.027 -0.116 +0.010 -0.236 -0.018
11_1202 +0.248* +0.316* 0.000 -0.032 +0.450* +0.220* +0.347* -0.035

(b)
35 ppt 53 ppt
Chromosome_Locus 1 4 5 1 2 4 5
1_30 -0.200 -0.164 -0.044 -0.091 -0.056 +0.282 +0.098
1_558 -0.088 -0.066 - +0.164 -0.086 +0.202 +0.228
2_228 -0.162 +0.108 - +0.379* -0.206 +0.057 -0.124
5_1203 +0.097 +0.002 +1.000* -0.125 +0.002 +0.025 -0.319*
5_62J8 +0.119 +0.172 +0.177 +0.052 +0.034 -0.298* -0.327*
7_56J2 -0.021 0.427* -0.180 -0.021 -0.143 +0.038 +0.106
8_480 +0.043 +0.131 -0.059 +0.066 -0.015 -0.096 +0.055
9_197 +0.131 +0.149 - +0.154 -0.036 +0.324* -0.008
9_1814 +0.575* +0.424* - -0.044 -0.040 - +0.175
10_1555 -0.377* -0.450* - -0.125 +0.069 -0.105 -0.162
11_1202 -0.125* -0.125 -0.157 -0.017 +0.122* -0.191 -0.117

181
Table 4-2. F
IS
values calculated for females and males for each replicate using 11
microsatellite loci with significant values noted in bold (*) (p < 0.05). Dashes indicate
values that could not be estimated. Results shown for generations seven (a) and thirteen
(b).

(a)
Females
35 ppt 53 ppt
Chromosome_Locus 1 2 4 5 1 2 4 5
1_30 +0.123 -0.125 -0.226 - - -0.286 -0.259 -0.345
1_558 -0.188 -0.091 +0.066 - - -0.059 -0.079 -0.579
2_228 -0.393 -0.161 -0.520* -0.029 -0.241 -0.059 -0.619* +0.163
5_1203 -0.307 -0.212 -0.141 - +0.212 -0.053 -0.248 -0.053
5_62J8 -0.215 -0.269 -0.217 -0.059 -0.385 +0.065 -0.107 -0.032
7_56J2 -0.221 +0.027 -0.138 -0.029 +0.131 +0.150 +0.022 +0.232
8_480 -0.097 +0.122 +0.045 - +0.174 -0.290 +0.050 -0.040
9_197 +0.294 -0.044 -0.390* -0.014 -0.036 -0.111 -0.057 +0.003
9_1814 +0.044 +0.103 -0.538* -0.205 -0.084 -0.169 -0.092 -0.042*
10_1555 -0.462 -0.101 -0.137 - -0.385 +0.254 -0.259 -0.270
11_1202 +0.211 +0.432* +0.021 -0.164 +0.406* +0.124 +0.183 -0.179
Males
35 ppt 53 ppt
Chromosome_Locus 1 2 4 5 1 2 4 5
1_30 +0.487* -0.091 +0.217 -0.027 -0.188 -0.103 -0.267 -0.148
1_558 -0.079 -0.086 -0.077 -0.027 -0.027 -0.073 -0.166 -0.484*
2_228 +0.397 -0.041 +0.395 -0.056 -0.188 +0.639* -0.100 +0.701*
5_1203 -0.484* -0.188 +0.141 - -0.152 +0.349 +0.431* +1.000*
5_62J8 -0.448* -0.200 +0.124 -0.08 -0.357 0.000 -0.138 +1.000*
7_56J2 +0.469* -0.270 -0.318* - +0.348 +0.174 +0.166 -0.338*
8_480 -0.097 +0.184 -0.010 - -0.065 +0.162 +0.282* -0.052
9_197 +0.178 +0.060 +0.193 - -0.242 -0.017 -0.134 +0.229
9_1814 +0.146* +0.328 +0.035 -0.056 -0.063 +0.183* -0.182 +0.300
10_1555 -0.148 +0.234 +0.080 -0.027 +0.240 -0.229 -0.208 +0.163
11_1202 +0.277* +0.187 -0.011 +0.105 +0.505* +0.303* +0.537* +0.035

182
Table 4-2, Continued

(b)
Females
35 ppt 53 ppt
Chromosome_Locus 1 4 5 1 2 4 5
1_30 -0.226 -0.267 -0.063 -0.056 -0.059 +0.064 -0.086
1_558 +0.088 +0.093 - +0.315 -0.091 +0.370 +0.240
2_228 -0.357 +0.053 - +0.240 -0.125 -0.133 -0.125
5_1203 +0.395 -0.171 +1.000* -0.086 +0.297 -0.097 -0.188
5_62J8 -0.111 +0.016 +0.125 -0.029 -0.205 -0.234 -0.407
7_56J2 -0.141 +0.568* -0.143 +0.119 -0.151 +0.141 +0.004
8_480 -0.067 +0.103 -0.032 +0.214 +0.047 -0.124 +0.050
9_197 +0.406* +0.026 - +0.194 -0.140 +0.570* -0.169
9_1814 +0.914* +0.434* - - -0.131 - +0.139
10_1555 -0.175 -0.438 - -0.118 +0.297 -0.097 -0.166
11_1202 -0.044 -0.094 -0.239 -0.163 -0.076 -0.172 -0.149
Males
35 ppt 53 ppt
Chromosome_Locus 1 4 5 1 2 4 5
1_558 -0.241 -0.188 - -0.029 -0.056 +0.073 +0.240
1_30 -0.152 -0.056 - -0.118 -0.027 +0.397 +0.240
2_228 -0.016 +0.159 - +0.655 -0.267 +0.240 -0.192
5_1203 -0.166 +0.114 - -0.152 -0.267 -0.005 -0.438
5_62J8 +0.368* +0.283* +0.200 +0.030 +0.166 -0.390* -0.229
7_56J2 +0.089 +0.335* -0.188 -0.214 -0.173 -0.092 +0.210
8_480 +0.134 +0.157 -0.056 -0.136 -0.063 -0.086 -0.120
9_197 -0.144 +0.227 - -0.056 -0.051 +0.098 +0.079
9_1814 +0.248* +0.395 - -0.056 +0.040 - +0.161
10_1555 -0.579* -0.462 - -0.118 -0.188 -0.086 -0.148
11_1202 -0.201* -0.172 -0.084 -0.076 +0.287* -0.190 -0.100

183
found for loci 1203 and 62J8. In the high salinity treatment significant deficits were
observed in replicate number two as well as for three loci (1202, 1203 and 56J2).
Significant excess in high salinity were found for loci 558 and 30. At generation thirteen
the 35 ppt replicates collectively showed heterozygote excess for 3 loci (1202, 30 and
1555) and heterozygote deficit for two loci (1814 and 1203). Overall heterozygote deficit
was also detected in replicate number four. Additionally, heterozygote deficit was
significant when tested across all loci and all 35 ppt replicates. No significant
deficitswere detected for high salinity replicates while one locus (62J8) showed an
overall heterozygote excess.
The estimated effective population size (N
e
) was 29 and 23 for replicates one and
four at benign salinity (Table 4-3). Replicate five, which contained mostly SD alleles
was estimated to have N
e
=63, perhaps because allele frequencies were less likely to
fluctuate between the six generations tested. High salinity replicates had N
e
estimates
that ranged from 17 to 38.

Divergence Among Replicate Hybrid Beakers
Weir and Cockerham’s F
ST
was calculated for both generations. The overall F
ST

among all beakers was 0.228 for generation seven and 0.363 for generation thirteen and
both of these were highly significant.
Pair-wise F
ST
estimates for all replicate hybrid beakers within each treatment were
calculated for each generation using the method of Weir and Cockerham (Table 4-4). F
ST

estimates between all replicate pairs were highly significant (p<0.0001). The highest F
ST
values were between pairings with 35 ppt replicate number five, which was shown by

184
Table 4-3. Estimated effective population size (Ne) obtained using the standard variance
in allele frequency change (F) according to the temporal method of Nb_HetEx.

Salinity Replicate F Ne(F) 95% CI
35 ppt 1 0.1284 29.2 14.9 - 52.7
4 0.1590 22.5 12.1 - 38.2
5 0.0738 63.1 22.9 - 182.6
53 ppt 1 0.1250 30.6 13.9 - 60.8
2 0.1074 37.5 17.5 - 75.4
4 0.1271 29.8 16.5 - 50.6
5 0.2021 17 9.2 - 28.6

185
Table 4-4. Weir and Cockerham’s pair-wise F
ST
for pairs of replicate hybrid beakers.
Replicates are labeled by salinity treatment followed by replicate number. All values are
significant at p < 0.0001. Tables show generation seven (a), and generation thirteen (b).

(a)
35 ppt replicates 53 ppt replicates
1 2 4 5 1 2 4
2 0.2241
35 ppt replicates 4 0.0409 0.1504
5 0.2582 0.6376 0.3827
1 0.2551 0.0596 0.1777 0.6761
53 ppt replicates 2 0.1005 0.0592 0.0442 0.5027 0.0879
4 0.1241 0.1039 0.0612 0.5129 0.0869 0.0367
5 0.0314 0.2713 0.0787 0.2476 0.3018 0.1338 0.1338

(b)
35 ppt replicates 35 ppt replicates
1 4 5 1 2 4
35 ppt replicates 4 0.1382
5 0.4308 0.5142
1 0.2132 0.2335 0.7228
53 ppt replicates 2 0.1447 0.1989 0.6430 0.0812
4 0.2074 0.2037 0.6666 0.1035 0.0668
5 0.0746 0.1304 0.5610 0.1267 0.0336 0.0820

186
other measures (allele frequencies and hybrid index scores) to be composed mostly of SD
alleles. When replicate five was not considered, pair-wise F
ST
at generation seven ranged
from 0.04 to 0.38 for 35 ppt replicates and 0.03 to 0.30 for 53 ppt replicates. F
ST
between
replicates of different salinities ranged from 0.03 to 0.27. Therefore, the measure of
variation between replicates of different salinities was similar to variation among
replicates of the same salinity. At generation thirteen, large F
ST
values (0.56 to 0.72)
were also seen in pairings with 35 ppt replicate number five. With the exception of
replicate five, F
ST
between the two other 35 ppt replicates at generation thirteen was 0.14
while F
ST
values for high salinity replicates ranged from 0.03 to 0.13. F
ST
between
replicates of different salinities ranged from 0.07 to 0.23. This shows that there was less
variation among replicates at 53 ppt.
Factorial correspondence analysis (FCA) of microsatellite allelic composition was
run at both generations. At generation seven (Figure 4-2a), 22.2% of the variation could
be displayed on two factors. Factors one, two and three represented 13.0%, 5% and 4.2%
of the genetic variation respectively. While 35 ppt replicates two and five appeared to
diverge from the rest of the group, there was no distinct separation among replicates or
between the salinity treatments. By generation thirteen, the molecular distribution of
individuals was completely different. Figure 4-2b shows the variation among individuals
graphed on the first three factors of the FCA. Factor 1 represents 15.01% of the variation
while Factors 2 and 3 represent 6.60% and 4.33% respectively. Here, each 35 ppt
replicate clustered on its own, with replicate number five being largely separated from the
rest. In contrast, all 53 ppt replicates clustered together into one group.

187
Figure 4-2. Factorial correspondence analysis for generations 7(A) and 13(B) for 11
microsatellite loci. Each data point represents an individual. Replicates are indicated by
different colors.

(a)

188
Figure 4-2, Continued

(b)

189
Analysis of molecular variance (AMOVA) was conducted at each generation with
the different salinity treatments representing different groups (Table 4-5). At generation
seven, less than one percent of the variation was found among groups, but at generation
thirteen, variation among groups amounted to more than 18%. To further assess the
variation among replicates and salinity treatments at generation thirteen, AMOVA was
conducted for different replicate groupings (Table 4-5b). Variation among groups was
maximized (31.6%) when each 35 ppt replicate was pulled out as an individual group
while all 53 ppt replicates were grouped together as one. Separating 53 ppt replicates into
different groups actually decreased the amount of between-group variation, indicating
that high salinity replicates do not show large amounts of divergence from each other. F
ST

measures and FCA both suggest that benign salinity replicates have diverged from one
another as well as the high salinity treatment, while high salinity replicates have diverged
from benign salinity replicates but not from each other.
Locus-by-locus AMOVA showed similar contributions of all loci at generation
seven (Table 4-6). At generation thirteen a large percentage of the variation can be
attributed to loci 1814 and 197 which are both on chromosome nine.
The number of distinct populations at generation thirteen was estimated using the
program Structure, which applies a Bayesian clustering algorithm, to assess microsatellite
data. Replicated runs for K=2-7 resulted in the highest probabilities for seven
populations. At K=7, individual membership to a particular replicate can generally be
estimated based on multilocus genotype, though it is difficult to distinguish between
replicates two and five of the high salinity treatment (Figure 4-3). When K=2, structure
is first pulled out for the SD-swamped replicate. As K increases, the next groups of

190
Table 4-5. Analysis of molecular variance (AMOVA) of replicate groups using 11
microsatellite loci, computed by the distance matrix in Arlequin (10,000 permutations).
Groups for generation 7 (a) are 35 ppt replicates vs. 53 ppt replicates. Three different
group combinations were run for generation 13 (b) to assess the variation between
specific replicate beakers.

(a)
Source of Variation d.f.
Sum of
squares
Variance
Components
%
Variation Fixation Index p-value
Among groups 1 64.75 0.02397 Va 0.75 FCT=0.00753 0.33693
Among populations
within groups 6 343.468 0.70018 Vb 22 FSC=0.22169 0.00000
within populations 618 1519.157 2.45818 Vc 77.24 FST=0.22755 0.00000
Total 625 1927.375 3.18233

(b)
Source of Variation d.f.
Sum of
squares
Variance
Components
%
Variation Fixation Index p-value
(1) All 35 ppt replicates; (2) All 53 ppt
replicates
Among groups 1 212.853 0.61112 Va 18.2 FCT=0.18204 0.02802
Among populations
within groups 5 246.633 0.60775 Vb 18.1 FSC=0.22132 0.00000
within populations 537 1148.251 2.13827 Vc 63.69 FST=0.36307 0.00000
Total 543 1607.737 3.35713
(1) 35 ppt replicates 1 and 4; (2) 35 ppt replicate 5; (3) All 53
ppt replicates
Among groups 2 373.37 1.05904 Va 30.73 FCT=0.37963 0.00091
Among populations
within groups 4 86.116 0.24945 Vb 7.24 FSC=0.10447 0.00000
within populations 537 1148.251 2.13827 Vc 62.04 FST=0.30726 0.00000
Total 543 1607.737 3.44676
(1) 35 ppt replicates 1; (2) 35 ppt replicate 4; (3) 35 ppt replicate 5; (4) All
53 ppt replicates
Among groups 3 410.395 1.07461 Va 31.63 FCT=0.37067 0.02644
Among populations
within groups 3 49.091 0.1848 Vb 5.44 FSC=0.07955 0.00000
within populations 537 1148.251 2.13827 Vc 62.93 FST=0.31628 0.00000
Total 543 1607.737 3.39768
(1) All 35 ppt replicates; (2) 53 ppt replicate 1; (3) 53 ppt replicate 2; (4) 53 ppt replicate 4; (5) 53 ppt
replicate 5
Among groups 4 261.944 -0.31711 Va -10.4 FCT= -0.10397 0.05923
Among populations
within groups 2 197.542 1.22874 Vb 40.29 FSC=0.36493 0.00000
within populations 537 1148.251 2.13827 Vc 70.11 FST=0.29890 0.00000
Total 543 1607.737 3.04989

191
Table 4-6. Locus-by-locus AMOVA calculated for 11 microsatellite loci for (a) generation 7 and (b) generation thirteen.

(a)

Among
Groups Among Populations Within Populations
Chromosome_Locus SSD d.f. Va % variation SSD d.f. Vb % variation SSD d.f. Vc % variation
11_1202 9.091 1 0.017 4.331 21.796 6 0.042 10.441 212.022 616 0.344 85.228
9_1814 6.917 1 0.005 1.693 31.532 6 0.064 20.436 150.796 616 0.245 77.870
1_558 12.776 1 0.020 7.519 39.423 6 0.082 31.065 100.045 614 0.163 61.416
5_1203 1.318 1 -0.016 -6.022 37.779 6 0.078 29.484 124.950 618 0.202 76.538
2_228 2.106 1 -0.013 -5.003 36.191 6 0.075 29.782 116.567 618 0.189 75.222
9_197 8.196 1 0.015 4.492 21.844 6 0.045 13.275 164.647 592 0.278 82.233
7_56J2 4.879 1 -0.004 -1.120 35.814 6 0.074 23.504 149.299 612 0.244 77.616
5_62J8 0.836 1 -0.014 -4.918 31.074 6 0.063 22.398 143.901 616 0.234 82.520
8_480 3.165 1 -0.007 -2.025 29.006 6 0.067 19.356 155.306 542 0.287 82.669
1_30 8.864 1 0.008 3.102 38.334 6 0.080 31.211 103.411 616 0.168 65.687
10_1555 6.214 1 0.003 1.048 31.970 6 0.066 24.293 124.411 614 0.203 74.659

Fixation indices
Chromosome_Locus FSC P-value FST
P-
value FCT P-value
11_1202 0.109 0 0.148 0 0.043 0.059
9_1814 0.208 0 0.221 0 0.017 0.318
1_558 0.336 0 0.386 0 0.075 0.222
5_1203 0.278 0 0.235 0 -0.060 0.685
2_228 0.284 0 0.248 0 -0.050 0.574
9_197 0.139 0 0.178 0 0.045 0.166
7_56J2 0.232 0 0.224 0 -0.011 0.506
5_62J8 0.213 0 0.175 0 -0.049 0.740
8_480 0.190 0 0.173 0 -0.020 0.546
1_30 0.322 0 0.343 0 0.031 0.282
10_1555 0.246 0 0.253 0 0.010 0.256

192
Table 4-6, Continued

(b)

Among
Groups Among Populations Within Populations
Chromosome_Locus SSD d.f. Va % variation SSD d.f. Vb % variation SSD d.f. Vc % variation
11_1202 8.786 1 0.019 4.528 18.997 5 0.045 10.803 186.897 533 0.351 84.670
9_1814 47.734 1 0.168 45.055 18.357 5 0.046 12.325 83.952 529 0.159 42.620
1_558 25.346 1 0.076 27.693 26.950 5 0.069 25.055 68.564 529 0.130 47.252
5_1203 18.456 1 0.048 16.161 28.878 5 0.073 24.538 93.579 533 0.176 59.301
2_228 16.135 1 0.042 16.739 25.632 5 0.065 26.159 75.244 531 0.142 57.102
9_197 37.775 1 0.134 33.972 12.330 5 0.029 7.389 122.607 531 0.231 58.639
7_56J2 7.446 1 0.011 3.974 22.664 5 0.057 19.997 113.267 519 0.218 76.029
5_62J8 15.028 1 0.045 14.462 15.637 5 0.038 12.084 121.716 529 0.230 73.454
8_480 3.895 1 -0.008 -2.164 29.536 5 0.076 20.274 156.542 511 0.306 81.891
1_30 9.052 1 0.008 3.976 34.023 5 0.086 41.136 61.722 535 0.115 54.889
10_1555 24.487 1 0.077 27.383 19.613 5 0.049 17.246 83.605 535 0.156 55.371

Fixation indices
Chromosome_Locus FSC P-value FST
P-
value FCT P-value
11_1202 0.113 0 0.153 0 0.045 0.094
9_1814 0.224 0 0.574 0 0.451 0.029
1_558 0.347 0 0.527 0 0.277 0.061
5_1203 0.293 0 0.407 0 0.162 0.196
2_228 0.314 0 0.429 0 0.167 0.031
9_197 0.112 0 0.414 0 0.340 0.027
7_56J2 0.208 0 0.240 0 0.040 0.253
5_62J8 0.141 0 0.265 0 0.145 0.029
8_480 0.198 0 0.181 0 -0.022 0.718
1_30 0.428 0 0.451 0 0.040 0.286
10_1555 0.237 0 0.446 0 0.274 0.029

193
Figure 4-3. Inferred population structure for generation 13 from simulated runs in
STRUCTURE using the no admixture model and K=2 to7. Each color represents one
population out of K total. Each vertical line represents an individual divided up into the
probability that it comes from each of K populations. Numbers 1, 2, and 3 correspond to
35 ppt replicates 1, 4 and 5 respectively. Numbers 4, 5, 6 and 7 correspond to 53 ppt
replicates 1, 2, 4 and 5 respectively.

K=2
K=3
K=4
K=5
K=6
K=7
1 2 3 4 5 6 7
1 2 3 4 5 6 7
1 2 3 4 5 6 7
1 2 3 4 5 6 7
1 2 3 4 5 6 7
1 2 3 4 5 6 7

194
molecular structure detected are the two different salinity treatments, followed by
replicates of benign salinity.
For all trials performed using Fdist2, the data fit the model well, with most loci
falling within the simulated 95% confidence interval (Figure 4-4). Locus 1202 falls
outside of the simulated distribution for both generations when all replicates of a
generation are run together. When high salinity and benign salinity replicates are
modeled separately, locus 1202 is an outlier for 35 ppt replicates at both generations but
only at generation seven for 53 ppt replicates.

Molecular hybrid index
Molecular hybrid index was calculated for each replicate by scoring each allele
with the probability that it came from the SD population and then averaging over the
number of loci scored. A score of 0 indicates an individual with all RP alleles and a
score of 1 indicates all SD alleles. At generation seven, mean hybrid index for 35 ppt
beakers was 0.597, 0.221, 0.482 and 0.959 indicating no specific direction of change
from the expected hybrid index of 0.5 (Figure 4-5a). Fifty-three ppt replicates had mean
hybrid index scores of 0.201, 0.334, 0.343 and 0.627 (Figure 4-5b). At generation
thirteen three surviving 35 ppt replicates were genotyped (Figure 4-5c). Replicate
number five contained mostly SD alleles and had a mean hybrid index of 0.950. The
other two replicates showed decreases in mean hybrid index from generation seven. All
four 53 ppt replicates at generation thirteen showed decreases in mean hybrid index from
generation seven with all four replicates having mean hybrid index scores below the
expected value of 0.5 (Figure 4-5d).

195
Figure 4-4. F
ST
values estimated from 11 microsatellite loci against heterozygosity
plotted for generation seven 35 ppt replicates (a); generation thirteen 35 ppt replicates
(b); generation seven 53 ppt replicates (c) and generation thirteen 53 ppt replicates(d).
Each point represents one locus, with locus 1202 circled. Lines indicate 95% quantiles of
expected Fst given no selection.

(a)

(b)

196
Figure 4-4, Continued

(c)

(d)

197
Figure 4-5. Molecular hybrid index distributions for individual replicates calculated with
11 microsatellite loci. A score of 0 indicates all RP alleles and a score of 1 indicates all
SD alleles. Each histogram represents one replicate beaker for generations 7, 35 ppt (a),
53 ppt (b) and 13 35 ppt (c) and 53 ppt (d). The x-axis represents hybrid index score and
the y-axis is number of individuals.

(a)

C1
0
1
2
3
4
5
6
7
8
9
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

C2
0
2
4
6
8
10
12
14
16
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

C4
0
2
4
6
8
10
12
14
16
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

C5
0
5
10
15
20
25
30
35
40
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Hybrid Index Score
Number of Individuals

198
Figure 4-5, Continued

(b)

F1
0
2
4
6
8
10
12
14
16
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
F2
0
2
4
6
8
10
12
14
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

F4
0
2
4
6
8
10
12
14
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

F5
0
2
4
6
8
10
12
14
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Hybrid Index Score
Number of Individuals

199
Figure 4-5, Continued

(c)

C1
0
5
10
15
20
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

C4
0
5
10
15
20
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

C5
0
5
10
15
20
25
30
35
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Hybrid Index Score
Number of Individuals

200
Figure 4-5, Continued

(d)

F1
0
5
10
15
20
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

F2
0
2
4
6
8
10
12
14
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

F4
0
5
10
15
20
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

F5
0
2
4
6
8
10
12
14
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Hybrid Index Score
Number of Individuals

201
Molecular and morphological hybrid indices are concordant with each other in that both
show changes in the same direction from the expected index of 0.5 (Table 4-7).
However, the deviations from expected index scores are larger for the molecular hybrid
index.

Nuclear-nuclear interactions
Pairwise tests of linkage disequilibrim were performed for each replicate (Table
4-8). At generation seven for physically unlinked loci after Bonferroni correction, linkage
disequilibrium was significant for two pairs at 35 ppt and for 16 pairs at 53 ppt. This
large discrepancy between the two salinity treatments can be attributed to the large
amount of linkage disequilibrium seen in replicate five of the high salinity treatment. All
16 pairs found to be in significant LD are from replicate five. At generation thirteen,
following Bonferroni correction, linkage disequilibrium in both salinity treatments was
significant only between physically linked loci.
Additional tests for nuclear-nuclear interactions were performed by comparing
two-locus genotypic combinations to Mendelian expectations. Genotypes were pooled
into four different categories across all physically unlinked loci: parental double
homozygotes, homozygote-heterozygotes, heterozygote-heterozygotes and nonparental
double homozygotes. There was a significant overall excess of parental homozygotes
and a significant overall deficit of nonparental homozygotes in twelve out of sixteen
cases at generation seven (Figure 4-6a and 4-6b). Two cases of significant heterozygote-
heterozygote excess as well as two cases of significant heterozygote-homozygote deficit
were observed at each salinity treatment. For generation thirteen (Figure 4-6c and 4-6d),

202
Table 4-7. Mean deviation from expected morphological and molecular hybrid index scores for generations seven (a) and thirteen
(b).

(a)
Females Males Females and Males
Replicate
Morphological
Hybrid Index (SE)
Molecular Hybrid
Index (SE)
Morphological
Hybrid Index (SE)
Molecular Hybrid
Index (SE)
Molecular Hybrid
Index (SE)
1 +0.102 (0.015) +0.095 (0.031) +0.071 (0.008) +0.099 (0.047) +0.097 (0.028)
35 ppt 2 +0.140 (0.017) -0.277 (0.027) +0.137 (0.016) -0.280 (0.033) -0.279 (0.021)
4 +0.093 (0.016) -0.007 (0.028) +0.073 (0.011) -0.029 (0.031) -0.018 (0.020)
5 +0.099 (0.020) +0.459 (0.008) +0.062 (0.012) +0.455 (0.011) +0.459 (0.008)
Total +0.108 (0.011) +0.067 (0.152) +0.086 (0.017) +0.061 (0.153) +0.065 (0.153)

1 -0.038 (0.014) -0.286 (0.023) +0.001 (0.013) -0.310 (0.019) -0.299 (0.015)
53 ppt 2 +0.029 (0.013) -0.158 (0.053) -0.017 (0.019) -0.174 (0.035) -0.166 (0.031)
4 -0.090 (0.021) -0.123 (0.030) -0.059 (0.018) -0.187 (0.036) -0.157 (0.024)
5 -0.067 (0.026) +0.023 (0.050) -0.057 (0.015) +0.227 (0.047) +0.127 (0.038)
Total -0.042 (0.026) -0.136 (0.064) -0.033 (0.015) -0.111 (0.117) -0.124 (0.090)

203
Table 4-7, Continued

(b)
Females Males Females and Males
Replicate
Morphological
Hybrid Index (SE)
Molecular Hybrid
Index (SE)
Morphological
Hybrid Index (SE)
Molecular Hybrid
Index (SE)
Molecular Hybrid
Index (SE)
1 +0.024 (0.015) -0.067 (0.022) +0.041 (0.013) -0.032 (0.028) -0.050 (0.018)
35 ppt 2 +0.036 (0.015) - +0.014 (0.010) - -
4 +0.118 (0.017) -0.124 (0.018) +0.005 (0.009) -0.117 (0.021) -0.121 (0.014)
5 +0.029 (0.016) +0.447 (0.012) +0.003 (0.009) +0.453 (0.009) +0.450 (0.008)
Total +0.051 (0.022) +0.085 (0.157) +0.016 (0.009) +0.101 (0.154) +0.093 (0.156)

1 +0.027 (0.013) -0.396 (0.021) -0.128 (0.016) -0.344 (0.016) -0.368 (0.013)
53 ppt 2 -0.049 (0.012) -0.291 (0.025) -0.145 (0.019) -0.248 (0.027) -0.269 (0.019)
4 +0.022 (0.015) -0.327 (0.023) -0.008 (0.018) -0.284 (0.018) -0.304 (0.015)
5 -0.008 (0.017) -0.238 (0.026) -0.126 (0.015) -0.155 (0.022) -0.197 (0.018)
Total -0.002 (0.017) -0.313 (0.033) -0.102 (0.032) -0.258 (0.040) -0.284 (0.036)

204
Table 4-8. Genotypic linkage disequilibrium for each locus pair and associated p-values for replicates of generation seven (a) and
generation thirteen (b). Significant p-values are indicated in bold, italicized font. P-values significant after a Bonferroni correction are
indicated by a (*).
(a)
35 ppt replicates 53 ppt replicates
1 2 4 5 1 2 4 5
locus 1 locus 2 p-value p-value p-value p-value locus 1 locus 2 p-value p-value p-value p-value
11_1202 9_1814 0.43406 0.74870 0.09469 0.03087 11_1202 9_1814 0.52412 0.25576 0.82270 0.00108
11_1202 1_558 0.90311 0.65550 0.48080 0.42408 11_1202 1_558 0.78103 0.01982 0.55675 0.51555
9_1814 1_558 0.46818 0.06209 0.41465 0.62160 9_1814 1_558 1.00000 0.22004 0.56703 0.52852
11_1202 5_1203 0.02776 0.89742 0.71608 1.00000 11_1202 5_1203 0.55260 0.06465 0.61727 0.00003*
9_1814 5_1203 0.80468 0.75127 0.33475 0.02643 9_1814 5_1203 0.30866 0.44514 0.78896 0.40410
1_558 5_1203 0.77706 0.10677 0.99299 1.00000 1_558 5_1203 0.12449 0.06980 0.46555 0.04589
11_1202 2_228 0.15984 0.15557 0.65618 0.28060 11_1202 2_228 0.43503 0.13382 0.16451 0.00067*
9_1814 2_228 0.93540 0.02581 0.01177 0.30778 9_1814 2_228 0.34433 0.10460 0.55302 0.04344
1_558 2_228 0.18563 1.00000 0.84167 0.24272 1_558 2_228 0.51417 0.15886 0.58040 0.01502
5_1203 2_228 0.49502 0.16275 0.07587 1.00000 5_1203 2_228 0.58328 0.32072 0.00568 0.00000*
11_1202 9_197 0.46603 0.38271 0.92433 0.66510 11_1202 9_197 0.15615 0.09892 0.10268 0.00056*
9_1814 9_197 0.19664 0.00000* 0.00000* 0.09081 9_1814 9_197 0.50902 0.01525 0.00000* 0.00000*
1_558 9_197 0.07722 0.40912 0.99882 1.00000 1_558 9_197 0.58368 0.49992 0.16632 0.02371
5_1203 9_197 0.47419 0.32062 0.62983 0.07696 5_1203 9_197 0.32575 0.05609 0.36278 0.01415
2_228 9_197 0.76010 0.67318 0.00969 0.13949 2_228 9_197 0.09565 0.46357 0.68377 0.00002*
11_1202 7_56J2 0.22632 0.04670 0.86176 0.71143 11_1202 7_56J2 0.58383 0.07597 0.08387 0.08566
9_1814 7_56J2 0.73291 0.02659 0.18592 0.22867 9_1814 7_56J2 0.05694 0.98642 0.24877 0.08864
1_558 7_56J2 0.20811 0.33383 0.88386 1.00000 1_558 7_56J2 0.66370 0.33570 0.74691 0.22775
5_1203 7_56J2 0.11008 0.40059 0.04677 1.00000 5_1203 7_56J2 0.72408 0.48866 0.09494 0.00000*
2_228 7_56J2 0.14410 0.88883 0.11045 1.00000 2_228 7_56J2 0.41566 0.12098 0.36386 0.00000*
9_197 7_56J2 0.48999 0.11543 0.13462 1.00000 9_197 7_56J2 0.69188 0.66688 0.10377 0.08009
11_1202 5_62J8 0.00970 0.62858 0.56672 0.17256 11_1202 5_62J8 0.25728 0.19643 0.68687 0.00049*
9_1814 5_62J8 0.81091 0.04650 0.66294 0.20450 9_1814 5_62J8 0.01050 0.08347 0.88535 0.25365
1_558 5_62J8 0.71860 0.02658 0.98016 0.21112 1_558 5_62J8 0.49534 0.08644 0.94336 0.09302
5_1203 5_62J8 0.00001* 0.00000* 0.00000* 0.10065 5_1203 5_62J8 0.00002* 0.00000* 0.00171 0.00000*
2_228 5_62J8 0.19555 0.03050 0.12376 0.17863 2_228 5_62J8 0.75215 0.22449 0.58399 0.00000*

205
Table 4-8a, Continued

9_197 5_62J8 0.81480 0.31659 0.62122 0.12751 9_197 5_62J8 0.71283 0.22235 0.38890 0.00479
7_56J2 5_62J8 0.24382 0.00080* 0.79387 1.00000 7_56J2 5_62J8 0.08309 0.11576 0.11704 0.00005*
11_1202 8_480 0.21351 0.79130 0.83000 - 11_1202 8_480 0.11008 0.14085 0.80147 0.00032*
9_1814 8_480 0.47559 0.01761 0.64713 - 9_1814 8_480 0.16747 0.70827 0.99920 0.31360
1_558 8_480 0.68846 0.55082 0.51087 - 1_558 8_480 0.80626 0.63007 0.29236 0.26815
5_1203 8_480 0.59366 0.52856 0.58670 - 5_1203 8_480 0.90147 0.15724 0.60732 0.00000*
2_228 8_480 0.34155 0.38006 0.66725 - 2_228 8_480 0.88948 0.97295 0.83468 0.00000*
9_197 8_480 0.20554 0.21072 0.59637 - 9_197 8_480 0.97308 0.42628 0.78593 0.00000*
7_56J2 8_480 0.49251 0.34904 0.83078 - 7_56J2 8_480 0.05947 0.94404 0.90041 0.00000*
5_62J8 8_480 0.59451 0.85158 0.15548 - 5_62J8 8_480 0.37976 0.51740 0.37865 0.00000*
11_1202 1_30 0.38922 0.61355 0.59146 0.42592 11_1202 1_30 0.57825 0.00863 0.29963 0.92146
9_1814 1_30 0.37038 0.06958 0.92991 0.62295 9_1814 1_30 0.72237 0.19065 0.98648 0.98653
1_558 1_30 0.00444 0.00000* 0.00000* 0.00137 1_558 1_30 0.03787 0.00000* 0.00000* 0.05522
5_1203 1_30 0.97849 0.06912 0.40560 1.00000 5_1203 1_30 0.41416 0.01211 0.21312 0.11891
2_228 1_30 0.40670 0.76403 0.46037 0.24277 2_228 1_30 1.00000 0.06132 0.24365 0.03976
9_197 1_30 0.31949 0.92644 0.82177 1.00000 9_197 1_30 0.68887 0.57401 0.09861 0.44759
7_56J2 1_30 0.31146 0.21377 0.11523 1.00000 7_56J2 1_30 0.48897 0.79836 0.47039 0.23646
5_62J8 1_30 0.55335 0.01919 0.23340 0.21172 5_62J8 1_30 1.00000 0.09113 0.37693 0.24198
8_480 1_30 0.07926 0.56129 0.97428 - 8_480 1_30 0.18051 0.79338 0.41694 0.23358
11_1202 10_1555 0.93589 0.86723 0.00022* 0.94380 11_1202 10_1555 0.50766 0.17519 0.92911 0.01846
9_1814 10_1555 0.98733 0.46312 0.22094 0.58635 9_1814 10_1555 0.11984 0.53126 0.38569 0.05244
1_558 10_1555 0.08754 0.07781 0.32293 0.14939 1_558 10_1555 0.53613 0.40889 0.69057 0.99664
5_1203 10_1555 0.58768 0.09680 0.70368 1.00000 5_1203 10_1555 0.08541 0.08505 0.17892 0.00660
2_228 10_1555 0.46372 0.13166 0.85927 0.03732 2_228 10_1555 0.70183 0.64953 0.04530 0.02822
9_197 10_1555 0.07617 0.77758 0.42359 1.00000 9_197 10_1555 0.99032 0.55472 0.28674 0.09890
7_56J2 10_1555 0.25929 0.24439 0.26569 1.00000 7_56J2 10_1555 0.01555 0.67069 0.00508 0.06723
5_62J8 10_1555 0.37327 0.00208 0.39747 0.76216 5_62J8 10_1555 0.24623 0.18985 0.17048 0.03176
8_480 10_1555 0.03255 0.89746 0.98667 - 8_480 10_1555 0.16388 0.68196 0.19678 0.13150
1_30 10_1555 0.21954 0.09474 0.69137 0.15087 1_30 10_1555 0.17346 0.43269 0.92203 0.96516

206
Table 4-8, Continued
(b)
35 ppt replicates 53 ppt replicates
1 4 5 1 2 4 5
locus 1 locus 2 p-value p-value p-value locus 1 locus 2 p-value p-value p-value p-value
11_1202 9_1814 0.65920 0.00198 0.50888 11_1202 9_1814 0.63387 0.02969 - 0.47061
11_1202 1_558 0.08607 0.65611 - 11_1202 1_558 0.43631 0.91612 0.25758 0.43575
9_1814 1_558 0.84908 0.50301 - 9_1814 1_558 1.00000 0.87509 - 0.00114
11_1202 5_1203 0.80718 0.04142 1.00000 11_1202 5_1203 0.73126 0.43285 0.81506 0.02087
9_1814 5_1203 0.53396 0.01517 - 9_1814 5_1203 0.55429 0.02852 - 0.35749
1_558 5_1203 0.44773 0.11828 - 1_558 5_1203 0.66270 1.00000 0.47251 0.65664
11_1202 2_228 0.02218 0.40191 0.49757 11_1202 2_228 0.17301 0.82175 0.63488 0.74849
9_1814 2_228 0.52312 0.03242 1.00000 9_1814 2_228 1.00000 0.81994 - 0.00267
1_558 2_228 0.66399 0.46957 - 1_558 2_228 0.76734 0.70082 0.82388 0.06513
5_1203 2_228 0.78817 0.19812 1.00000 5_1203 2_228 0.50589 0.09713 0.55757 1.00000
11_1202 9_197 0.08317 0.05137 - 11_1202 9_197 0.12937 0.13081 0.39600 0.14645
9_1814 9_197 0.00468 0.00000* - 9_1814 9_197 0.18934 0.03153 - 0.00000*
1_558 9_197 0.66094 0.54608 - 1_558 9_197 0.85499 0.41601 0.03710 0.09685
5_1203 9_197 0.27009 0.10648 - 5_1203 9_197 0.34283 0.34581 0.12033 0.34177
2_228 9_197 0.15859 0.12743 - 2_228 9_197 0.85686 0.59281 0.65493 0.12278
11_1202 7_56J2 0.08547 0.91308 0.36874 11_1202 7_56J2 0.34972 0.00278 0.67683 0.53751
9_1814 7_56J2 0.01972 0.56780 1.00000 9_1814 7_56J2 0.69308 0.04461 - 0.01177
1_558 7_56J2 0.19538 0.97731 - 1_558 7_56J2 0.44149 0.11334 0.28360 0.08233
5_1203 7_56J2 0.03065 0.76661 0.34335 5_1203 7_56J2 0.83114 0.24416 0.20402 0.69702
2_228 7_56J2 0.03617 0.90075 1.00000 2_228 7_56J2 0.35936 1.00000 0.38417 0.03265
9_197 7_56J2 0.16860 0.15816 - 9_197 7_56J2 0.05997 0.76592 0.33425 0.20481
11_1202 5_62J8 0.79223 0.06823 0.04988 11_1202 5_62J8 0.32970 0.22616 0.34666 0.36983
9_1814 5_62J8 0.98661 0.02596 1.00000 9_1814 5_62J8 0.29040 0.25840 - 0.19258
1_558 5_62J8 0.84507 0.33340 - 1_558 5_62J8 0.71238 0.58851 0.31531 0.09189
5_1203 5_62J8 0.22379 0.19534 0.24736 5_1203 5_62J8 0.00000* 0.00006* 0.00338 0.00023*
2_228 5_62J8 0.02458 0.10520 1.00000 2_228 5_62J8 0.38694 0.69377 0.65824 0.75290
9_197 5_62J8 0.51055 0.01754 - 9_197 5_62J8 0.06357 0.00678 0.89681 0.67171
7_56J2 5_62J8 0.15685 0.99525 0.11940 7_56J2 5_62J8 1.00000 0.93481 0.30053 0.21987

207
Table 4-8b, Continued

11_1202 8_480 0.72233 0.23426 0.01640 11_1202 8_480 0.12899 0.24535 0.54922 0.70251
9_1814 8_480 0.94781 0.34977 1.00000 9_1814 8_480 0.41500 0.34991 - 0.21037
1_558 8_480 0.60076 0.41599 - 1_558 8_480 0.01280 0.09378 0.62140 0.37758
5_1203 8_480 0.09992 0.34386 1.00000 5_1203 8_480 0.40921 0.02516 0.26657 0.67694
2_228 8_480 0.93659 0.88009 0.13522 2_228 8_480 0.05873 0.27016 0.58723 0.83519
9_197 8_480 0.26752 0.70496 - 9_197 8_480 0.05773 0.01525 0.83062 0.56866
7_56J2 8_480 0.92164 0.39020 1.00000 7_56J2 8_480 0.89179 0.82100 0.32394 0.36994
5_62J8 8_480 0.43475 0.61537 0.30065 5_62J8 8_480 0.19323 0.05737 0.27988 0.93154
11_1202 1_30 0.15875 0.02504 0.03269 11_1202 1_30 0.73758 0.97292 0.19786 0.99742
9_1814 1_30 0.97478 0.00204 1.00000 9_1814 1_30 0.57072 1.00000 - 0.01372
1_558 1_30 0.00123 0.49678 - 1_558 1_30 0.00025* 0.00002* 0.05527 0.00026*
5_1203 1_30 0.63594 0.00674 1.00000 5_1203 1_30 0.16004 1.00000 0.06322 0.28101
2_228 1_30 0.31254 0.01087 0.10724 2_228 1_30 0.26472 0.63619 0.38630 0.41201
9_197 1_30 0.58439 0.10581 - 9_197 1_30 0.21193 0.24858 0.00862 0.03053
7_56J2 1_30 0.80744 0.81170 0.29061 7_56J2 1_30 1.00000 0.49469 0.04441 0.20879
5_62J8 1_30 0.16560 0.00425 0.44045 5_62J8 1_30 0.57309 0.40663 0.52134 0.24308
8_480 1_30 0.27776 0.02246 1.00000 8_480 1_30 0.00813 0.21265 0.21031 0.26678
11_1202 10_1555 0.49983 0.12567 - 11_1202 10_1555 0.91687 0.12200 0.95875 0.29494
9_1814 10_1555 0.73494 0.21538 - 9_1814 10_1555 1.00000 0.13454 - 0.26185
1_558 10_1555 0.49488 0.71605 - 1_558 10_1555 0.66377 0.19588 0.65177 0.66291
5_1203 10_1555 0.13539 0.19399 - 5_1203 10_1555 1.00000 0.00118 0.33848 0.64911
2_228 10_1555 0.16552 0.16287 - 2_228 10_1555 0.60023 0.38389 0.10691 0.63096
9_197 10_1555 0.46968 0.00432 - 9_197 10_1555 0.73984 0.15737 0.04350 0.15930
7_56J2 10_1555 0.13926 0.56505 - 7_56J2 10_1555 0.68775 0.89096 0.14914 0.44473
5_62J8 10_1555 0.55661 0.33190 - 5_62J8 10_1555 0.60882 0.00186 0.37253 0.16118
8_480 10_1555 0.36807 0.87771 - 8_480 10_1555 0.19635 0.05885 0.24044 0.08357
1_30 10_1555 0.30348 0.12263 - 1_30 10_1555 0.65680 0.30835 0.01121 0.65900

208
Figure 4-6. Mean and standard error for proportional deviation from expected
microsatellite genotype frequencies for each replicate for four two-locus classes. Only
physically, unlinked loci are included for a total of 55 two-locus combinations tested
within each replicate. (a) Generation seven, 35 ppt; (b) generation seven , 53 ppt; (c)
generation thirteen, 35 ppt and (d) generation thirteen, 53 ppt. Asterisks indicate the
significance of paired, two-tailed t-tests of observed vs. expected genotype numbers (*p<
0.05). N = 74 two-locus combinations for each cross.

(a)
Generation 7, 35 ppt
-1
-0.5
0
0.5
1
1.5
2
2.5
Deviation from expectations
C1
C2
C4
C5
***
***
*
*
*
*
*
**
***
***
Parental Heterozygote Heterozygote Nonparental
Homozygotes Heterozygotes Homozygotes Homozygotes

209
Figure 4-6, Continued

(b)

Generation 7, 53 ppt
-1
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
Deviation from expectations
F1
F2
F4
F5

Parental Heterozygote Heterozygote Nonparental
Homozygotes Heterozygotes Homozygotes Homozygotes
***
***
***
***
***
***
***
***
***
**

(c)
Generation 13, 35 ppt
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
Deviation from expectations
C1a
C4
C5

Parental Heterozygote Heterozygote Nonparental
Homozygotes Heterozygotes Homozygotes Homozygotes

210
Figure 4-6, Continued

(d)
Generation 13, 53 ppt
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
Deviation from expectations
F1
F2
F4
F5

Parental Heterozygote Heterozygote Nonparental
Homozygotes Heterozygotes Homozygotes Homozygotes
*
*
**
*

211
with the exception of high salinity replicate number two which showed a significant
excess or parental homozygotes and double heterozygotes, as well as a deficit of
heterozygote-homozygotes, there were no other significant deviations from expectations.
F
IS
calculations for each proline biosynthesis gene showed that for locus P5CR in the 35
ppt treatment at generation 7, there were two replicates with heterozygote excess and one
replicate with heterozygote deficit (Table 4-9). No other significant deviations from
HWE were detected for either of the two salinities or generations. Locus P5CS showed
increases in the RP allele from the expected frequency of 0.5 in all generations and
salinities, and this increase was much greater for high salinity replicates (Table 4-10a).
Locus P5CR showed a decrease in RP allele frequency at 35 ppt, but an increase at 53 ppt
(Table 4-10b).
When two-locus genotypes were pooled into four different categories (parental
double homozygote, homozygote-heterozygote, heterozygote-heterozygote and
nonparental double homozygote), no significant differences from expected numbers were
found in any generation or at any salinity (Figure 4-7).

Cytonuclear Interactions
COI genotypic frequencies showed deviations from an expected 50:50 ratio for all
replicates at generation seven (Table 4-11a). Due to inferior extraction quality, COI
genotypes were not obtained for 35 ppt replicate 4 at generation thirteen and were only
obtained for five individuals for 53 ppt replicate 1 at generation seven. At generation
thirteen, all 53 ppt replicates became fixed for one COI genotype (Table 4-11b). For
replicates that were fixed for COI genotype, cytochrome c did not show any significantly

212
Table 4-9. F
IS
values for P5CR and P5CS genes for each replicate with significant values
in bold (p < 0.05). Dashes indicate values that could not be estimated. Results shown for
generation seven (a) and generation thirteen (b).

(a)
35 ppt 53 ppt
locus 1 2 4 5 1 2 4 5
P5CR -0.4000 -0.1111 -0.5652 0.7719 -0.0476 0.0000 0.4359 0.0148
P5CS -0.1179 -0.2984 0.0690 - 0.1852 0.2143 -0.2031 0.1667

(b)
35 ppt 53 ppt
locus 1 4 5 1 2 4 5
P5CR 0.0217 -0.2456 - -0.1000 -0.0323 -0.0333 -0.1525
P5CS 0.1705 -0.2195 - -0.0732 -0.3435 0.1429 -0.2446

213
Table 4-10. Deviation from expected RP allele frequencies for P5CR (a) and P5CRS (b)
genes for each replicate.

(a)
Generation 7 Generation 13
35 ppt 53 ppt 35 ppt 53 ppt
Replicate RP Replicate RP Replicate RP Replicate RP
1 (n=32) -0.062 1 (n=24) 0.438 1 (n=36) 0.347 1 (n=32) 0.406
2 (n=38) 0.355 2 (n=36) 0.250 4 (n=26) 0.192 2 (n=33) 0.439
4 (n=37) 0.095 4 (n=33) 0.424 5 (n=24) -0.500 4 (n=34) 0.441
5 (n=17) -0.324 5 (n=37) -0.095 5 (n=35) 0.357
mean 0.016 mean 0.254 mean 0.013 mean 0.411

(b)
Generation 7 Generation 13
35 ppt 53 ppt 35 ppt 53 ppt
Replicate RP Replicate RP Replicate RP Replicate RP
1 (n=36) 0.069 1 (n=36) 0.236 1 (n=33) 0.136 1 (n=28) 0.411
2 (n=39) 0.154 2 (n=38) 0.105 4 (n=40) 0.037 2 (n=37) 0.027
4 (n=40) 0.050 4 (n=38) 0.039 5 (n=33) -0.500 4 (n=35) 0.186
5 (n=38) -0.461 5 (n=39) -0.295 5 (n=37) 0.000
mean -0.047 mean 0.021 mean -0.109 mean 0.156

214
Figure 4-7. Proline biosynthesis genes, P5CR and P5CS. Mean and standard error for
proportional deviation from expected two-locus genotype frequencies for four classes of
hybrids across replicate beakers at 35 ppt salinity treatment (a) and 53 ppt salinity (b).

(a)

(b)
53 ppt
-0.7
-0.2
0.3
0.8
1.3
1.8
parental
homozygotes
homozygote
heterozygotes
double
heterozygotes
nonparental
homozygotes
Proportional deviation from expected
Gen 7
Gen 13

215
Table 4-11. COI and Cytochrome C genotypic frequencies for generation seven (a) and
generation thirteen (b).

(a)
Replicate N
COI
Genotype
Frequency
Cyt. C RP
allele
frequency
Cytochrome C Genotype
Frequencies
Significance
of χ2
RP SD RP-RP RP-SD SD-SD
1 37 0.14 0.87 0.31 0.16 0.30 0.54 ***
35 ppt 2 36 0.56 0.44 0.82 0.72 0.19 0.08 ***
4 38 0.74 0.26 0.36 0.21 0.29 0.50 **
5 39 0.00 1.00 0.00 0.00 0.00 1.00 ***

1 5 0.40 0.60 0.50 0.20 0.60 0.20 ns
53 ppt 2 36 0.86 0.14 0.56 0.42 0.28 0.31 *
4 34 0.71 0.29 0.49 0.38 0.21 0.41 **
5 39 0.23 0.77 0.45 0.28 0.33 0.38 ns

(b)
Replicate N
COI
Genotype
Frequency
Cyt. C RP
allele
frequency
Cytochrome C Genotype
Frequencies
Significance
of χ2
RP SD
RP-
RP RP-SD SD-SD
1 31 0.19 0.81 0.48 0.23 0.52 0.26 ns
35 ppt 4 34 - - 0.56 0.29 0.53 0.18 ns
5 36 0.00 1.00 0.00 0.00 0.00 1.00 ***

1 32 1.00 0.00 0.81 0.63 0.38 0.00 ns
53 ppt 2 14 1.00 0.00 0.68 0.50 0.36 0.14 ns
4 28 0.00 1.00 0.45 0.18 0.54 0.29 ns
5 12 1.00 0.00 0.58 0.25 0.67 0.08 ns

216
distorted genotypic ratios, except for 35 ppt replicate 5 which became fixed for SD alleles
at both COI and cytochrome c loci. At generation seven allele frequencies suggested both
possible advantage and disadvantage when matching the cytoplasm for normal salinity
replicates only. Replicate one showed an excess of both SD cytoplasm and cyt c alleles
while replicate two revealed the same pattern for RP. Replicate four showed an excess of
RP cytoplasm but a deficit of RP alleles. Mean RP allele frequencies were 0.50 (0.02
SE) for high salinity replicates. Observed vs. expected cytonuclear genotype frequencies
were compared for COI and cytochrome c at generation seven. (This was not done for
generation thirteen because most replicates were fixed for one cytoplasmic genotype.)
No significant deviations from expected cytonuclear genotypes were detected for either
salinity, although trends toward parental cytonuclear genotypes are consistent with
coadaptation (Figure 4-8).
Microsatellite-COI cytonuclear genotypic frequencies were analyzed for two
classes: matched homozygotes and unmatched homozygotes. Thirty-five ppt replicate
five was not included in microsatellite cytonuclear calculations due to evidence of genetic
swamping. Significant deviations from expected numbers were only found at generation
seven (Figure 4-9). One benign salinity replicate and three high salinity replicates
showed significant excess of matched cytonuclear genotypes and a significant deficit of
unmatched genotypes.
Observed vs. expected cytonuclear genotype frequencies were also compared for
P5CR and P5CS to determine if a population’s proline biosynthesis alleles were favored
on its own cytoplasmic background. Significant deviations from expectations were not
found for locus P5CR. A one-tailed t-test showed that, for locus P5CS, only high salinity

217
Figure 4-8. Proportional deviation from expected frequencies of cytonuclear genotypes
at generation seven for 35 ppt (a) and 53 ppt (b). No significant differences from
expected numbers were observed.

(a)
35 ppt
-1
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
1.2
RP COI
SD COI
RP/RP RP/SD SD/SD
Cytochrome C Genotype
Proportional deviation from expected

(b)
53 ppt
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
RP COI
SD COI
RP/RP RP/SD SD/SD
Cytochrome C Genotype
Proportional deviation from expected

218
Figure 4-9. Mean and standard error for proportional deviation from expected COI and
microsatellite genotypic frequencies at generation 7. Matched individuals have COI and
homozygous microsatellite genotypes from the same population. Unmatched individuals
are homozygous for microsatellite alleles that do not match the COI genotype. Asterisks
indicate the significance of one-tailed t-tests of observed vs. expected genotype numbers
(*p< 0.05, **p<0.01, ***p<0.001).

-0.6
-0.5
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
Proportional deviation from expectations
1_35
2_35
4_35
2_53
4_53
5_53
Matched Unmatched
*
*
***
***
***
***
**
**

219
replicates at generation seven had significant increases from expectations for
homozygotes on their own cytoplasmic background and decreases from expectations for
homozygotes on unmatched cytoplasmic backgrounds (Figure 4-10). Normal salinity
replicates at generation 7 had marginally non-significant deviations at p=0.059.

DISCUSSION

Stronger selection for RP alleles in high salinity
Microsatellites showed an overall trend of increased RP alleles from expected
starting frequencies. Additionally, morphological hybrid index scores calculated in
Chapter 3 suggest that several replicates became more RP-like over time. Edmands et al.
(2005) showed that selection was required to explain a decrease in mean hybrid index
from 10.5 to 7.2 (calculated for four replicates over seven loci). We observed a decrease
in mean hybrid index of 0.12 and 0.28 for 53 ppt replicates at generations seven and
thirteen respectively. If the Edmands et al. hybrid index was standardized to our
measure, they would have observed a change of 0.24, so we also have strong evidence for
selection occurring on RP alleles, at least for the 53 ppt replicates at generation thirteen.
The simplest explanation for the observed increases in RP alleles is RP
superiority, and RP replicates do outperform SD for fitness measures at most generations
(Chapter 3), although morphologically they tend to be smaller. However, RP superiority
is certainly not always the case, even for other experiments on highly correlated fitness

220
Figure 4-10. Mean and standard error for proportional deviation from expected COI and
P5CS genotype frequencies at generation seven. Matched individuals have COI and
homozygous P5CS genotypes from the same population. Unmatched individuals are
homozygous for P5CS alleles that do not match the COI genotype. Asterisks indicate the
significance of one-tailed t-tests of observed vs. expected genotype numbers (*p< 0.05).

-0.6
-0.5
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
Proportional deviation from expectations
ppt 35
ppt 53
*
*
Matched Unmatched

221
characters that were tested under the same laboratory conditions (see Edmands et al.
2005). Additionally, Chapter 1 suggested no strong fitness differences between RP and
SD in the laboratory. This is the third experiment for which this cross has been used and
the outcome has resulted in more RP alleles than expected despite RP’s relative
performance to SD. It is likely that there is a component of genetic determinism for
hybrid populations of this particular cross that involves RP advantage as well as the
interactions between hybrid genotypes involving SD alleles.
According to calculations of linkage disequilibrium (Table 4-8), 53 ppt replicate
five, also seems to have undergone a bottleneck event at some time before generation
seven. Large amounts of LD may result from an excess of parental gene combinations at
generation seven. Perhaps this replicate experienced the same conditions that resulted in
increased SD alleles, but, because selection is stronger in the more stressful environment,
SD did not drift to fixation for any locus. Instead, RP alleles increased so that, while at
generation seven, this replicate had a hybrid index score 67% above the mean of all high
salinity replicates, but at generation thirteen HI score had decreased so that it was only
40% above the mean. The estimated N
e
for this replicate was smaller than all others
(Table 4-3) which is also consistent with stronger directional selection acting between
generations seven and thirteen.
Numbers of surviving replicates were similar throughout generations for all
treatments, so there was no evidence to suggest that a particular starting genotypic
composition was more likely to go extinct. The only replicate that went extinct between
generations seven and thirteen was normal salinity replicate two. This replicate had a
larger excess of RP alleles and cytochrome c RP homozygotes than other replicates of the

222
same salinity treatment. It also had the lowest hybrid index score at generation seven.
Even though the overall data show that more RP alleles than expected are favored over
time, the replicate with the greatest proportion of RP alleles at generation seven is the one
that went extinct. If this was not a simple result of stochastic processes, then this finding
also suggests that the high frequencies of RP alleles observed may not be due to simple
RP advantage but could be a result of beneficial epistasis, such that RP alleles confer a
fitness advantage if specific interactions occur with some SD alleles.
Most replicates, for both salinity treatments, show increases in RP alleles over
time. Yet, the only replicate that shows fixation for multiple loci (35 ppt replicate five) is
composed almost entirely of SD alleles. This may have been propagated by demographic
stochasticity resulting in RP as the rare population early on. Because replicates were
maintained under common garden conditions, it is difficult to postulate how such a
dramatically different outcome could occur in one beaker, but there are some variables
that are not completely controlled. For example, endemic prokaryotes and algae in
collected seawater may pass through the filtering process. In replicate five the RP
population may simply have been more susceptible to the effects of a microorganism that
multiplied in that particular beaker. Alternatively, natural microorganism abundance may
have provided replicate five with exceptional food sources such that population size grew
rapidly and RP was outcompeted at an early stage of swarm formation. Even though
there may be selection present for some RP alleles or traits, RP may have established
itself as the rare population in this replicate in early generations. This outcome is similar
to what may be described in Carney et al (2000), where the sunflower H. bolanderi may
be going extinct in a hybrid zone, despite apparent selection for some of its traits.

223
Divergence between replicates and salinity treatments
Analysis of microsatellite data repeatedly suggests that there is minimal
differentiation among replicates at generation seven but considerable variation among
salinities and among replicates of benign salinity at generation thirteen, with less
variation among replicates of high salinity. This is evident according to several different
molecular measures: AMOVA showing that the maximum divergence among groups
occurs when benign salinity replicates are analyzed separately and high salinity replicates
were grouped together (Table 4-5), pairwise F
ST
values that are higher for 35 ppt
replicates (Table 4-4), hybrid index scores that that are more concordant among high
salinity replicates (Figure 4-5), and factorial correspondence analysis which shows
greater separation among 35 ppt replicates at generation thirteen (Figure 4-2). Finally,
simulated runs of Structure with K set between 2 and 7 show that structure is first pulled
out for the for the SD-swamped replicate, then the two different salinity treatments,
followed by replicates of benign salinity.
Replicate beakers for the 35 ppt treatment, with the same initial starting
conditions, diverged greatly from each other and from the high salinity replicate beakers.
In contrast, high salinity replicates show an overall pattern of repeatability of outcomes,
both in fitness (Chapter 3) and molecular characters. Rice and Hostert (1993) concluded
that divergence resulting in reproductive isolation could occur even between sympatric
lab populations if divergent selection is strong relative to gene flow. Being that there is
no possibility of migration to homogenize our replicates, and at least in the case of high
salinity replicates, repeatability suggests minimal effects of drift, it is reasonable to infer

224
that selection may be at work to diversify the two salinity treatments while maintaining
the patterns of repeatability that we see among 53 ppt replicates.
It could either be directional selection particular to the high salinity environment, or
stabilizing selection maintaining specific frequencies at high salinity, while drift causes
treatments to diverge at a greater rate.
Another possible explanation is that benign salinity replicates may have
experienced larger or more frequent fluctuations in population size than their high salinity
counterparts, causing the effects of drift to be enhanced. This would be consistent with
the finding that swamping by one populations’ alleles occurred for one 35 ppt replicate,
but not for any of the 53 ppt replicates. Even though a higher salinity may be more
stressful for individual fitness, if it resulted in a more homogeneous environment over
time, this could explain the greater repeatability observed among trials. Census counts
were not taken throughout the course of this experiment, though population density did
appear to remain high for 53 ppt replicates throughout the entire experiment, while subtle
changes in density were noted for 35 ppt replicates (personal observation). Still, although
it is possible, and certainly may have occurred for 35 ppt replicate 5, we did not observe
any replicates that seemed to be undergoing severe bottlenecking during culture
maintenance.
According to the methods of Beaumont and Nichols (1996), we determined that
there is a consistent pattern of greater differentiation for locus 1202 which may be due to
selection on a closely linked locus (Table 4-4). It turns out that locus 1202, unlike most
other loci, shows a deficit of RP alleles at benign salinity for all replicates at both
generations seven and thirteen with the exception of replicate 2 which went extinct before

225
generation thirteen. At 53 ppt, 1202 shows an excess of RP alleles for two out of four
replicates for both generations. It also shows significant heterozygote deficit for five out
of eight replicates at generation seven and one out of seven replicates at generation
thirteen. One benign salinity replicate at generation thirteen shows significant excess of
heterozygotes. There are two possible explanations of selection for the anomaly detected.
Either there is positive selection for the same 1202 variants, generated from
hybridization, in all replicates or divergence among replicates is taking place at all loci
except for 1202. This warrants further investigation of this region of the genome. While
it is not yet known if there are any functional genes tightly linked to locus 1202, this
marker was also one of the few loci in Harrison and Edmands (2006) that did not show a
significant excess of htereozygotes in backcross hybrids. The development and mapping
of EST-based SNPs is well underway and will aid in offering better resolution of the
functional areas of T. californicus chromosomes.

Cytonuclear interactions
Studies of mitochondrial versus nuclear encoded genes in T. californicus show
repeated and compelling evidence for cytonuclear coadaptation, with particular attention
being given to the mitochondrial electron transport system (Edmands and Burton 1999,
Rawson and Burton 2002, Harrison and Burton 2006). Most notable is a recent study by
Ellison and Burton (2008) demonstrating that fitness reductions in the F3 are completely
restored in maternal backcrosses where parental nuclear and mitochondrial combinations
are reformed. However, cases have also been reported where nuclear alleles are favored
on the wrong cytoplasmic background (Willett and Burton 2003, Willett 2006, Edmands

226
et al. in press, Appendix). Because Tigriopus experiences extreme fluctuations in its
environment, it is known to undergo repeated bottlenecks in the wild (Dybdahl 1994,
Burton 1997). This may result in a significant genetic load, producing hybrids with
alleles more fit on a foreign background. Because the mitochondrial genome is haploid,
it has a much smaller effective population size, and will additionally be more easily
influenced drift. Our data reflect this in that, by generation thirteen, mitochondrial
genotype was fixed in five out of the six replicates scored for COI, but three were fixed
for RP while two were fixed for SD (Table 4-11).
Results of this study lend support to the presence of cytonuclear coadaptation.
Evidence for this can been detected after up to seven generations of hybridization. This
was seen for both the P5CS gene (Figure 4-10) and nuclear microsatellites (Figure 4-9).
The fact that Edmands et al. (in press, Appendix) found significant evidence for
cytonuclear maladaptation in an F2 generation but that the same patterns were not
detected in our studies of advanced generations could mean that most newly formed
hybrid complexes do not offer the highest fitness advantage and, over multiple
generations, more coadapted complexes are likely to persist. Another explanation for the
discrepancy seen in the two studies is that crosses between different population pairs, or
even different collections of the same populations, may harbor distinct genetic loads due
to drift in small populations. Therefore, there will be enhanced population and locus
specificity for any particular interaction in any particular hybridization experiment.

227
Nuclear-nuclear interactions
Two studies in particular have looked at nuclear-nuclear interactions in F2 hybrid
generations. Edmands et al. (in press, Appendix) found evidence for nuclear-nuclear
coadaptation in a Royal Palms and Laguna Beach cross. When two-locus genotypes were
combined, both reciprocal crosses showed an excess of parental double homozygotes and
a deficit of non-parental double homozygotes, and both deviations were significant for
one of the reciprocals. A study by Willett (2006) showed a different pattern of
interactions for a cross between San Diego and Abalone Cove. Three nuclear, protein
coding genes of the electron transport system showed significant excesses of double
parental homozygotes, double heterozygotes and double nonparental homozygotes. After
seven generations of recombination, our results show some similarities to both studies.
Like Edmands et al. (in press, Appendix) we observed a significant excess of double
parental homozygotes and a significant deficit of double nonparental homozygotes. Three
out of four pooled categories showed similar patterns to those observed by Willet (2006),
the exception being nonparental homozygotes for which we observed a significant deficit.
Additionally, a non-recombinant backcross study by Harrison and Edmands (2006) found
an excess of heterozygotes and a deficit of parental homozygotes.
Even though a significant decrease of homozygote-heterozygotes was observed in
this study, as in Willett (2006), these results can still be explained by nuclear-nuclear
coadaptation under conditions of dominance, consistent with Edmands et al. (in press,
Appendix). Because recombination can occur each generation and break up superior
hybrid gene combinations, selection should have more opportunities to operate on
dominant loci that confer a fitness advantage. This may explain why evidence for

228
epistasis is weak according to our observations at later generations of hybridization. By
generation thirteen, signatures of coadaptation may be less likely to be observed due to
many generations of recombination that have broken up linkage between coadapted loci
and microsatellite markers. In concordance with these findings, Chapter 1, which
assessed one hybrid replicate after 30 months (a maximum of 39 generations) also found
no significant nuclear-nuclear deviations from Mendelian expectations.

Stress and selection
While we have already discussed stronger selection for RP alleles at high salinity,
the data also show a signal of stronger selection for both nuclear-mitochondrial nuclear-
nuclear coadaptation under stressful conditions. The larger magnitude of coadaptation
observed at high salinity at generation seven is consistent with the fact that high salinity
replicates have made it past their generation of lowest mean fitness (generation 5) while
benign salinity replicates are still exhibiting significant decreases in fitness compared to
parentals due to hybrid breakdown (Chapter 3). A signature of nuclear-nuclear
coadaptation was still present after thirteen generations in the 53 ppt environment, but
was completely absent in the 35 ppt environment. Willett and Burton (2003) showed that
fitness of cytonuclear genotypes was highly dependent on environmental (temperature)
conditions. If extrinsic selection is important, then environment should influence the
specific epistatic interactions underlying hybrid breakdown. Cytonuclear coadaptation
was stronger at generation seven for 53 ppt replicates compared to 35 ppt replicates. This
indicates that selection for coadapted complexes may be stronger for populations
experiencing environmental stress.

229
The protein-coding genes P5CR and P5CS are actively involved in
osmoregulation. In response to hyperosmotic stress a common cellular response is to
increase organic osmolytes (Yancey et al. 1982). In Tigriopus, Burton and Feldman
(1982) found that intercellular osmolyte accumulation primarily involves proline, alanine
and glycine, with proline showing the largest increase in concentration under
hyperosmotic stress (Goolish and Burton 1989). Under conditions of constant salinity,
glutamate (proline’s metabolic precursor) is synthesized requiring the activity of P5CR
and P5CS (Willett and Burton 2002). These genes were selected for study because they
were potential candidates for nuclear-nuclear coadaptation, though significant evidence
for coadaptation between these two loci was not observed. Evidence for cytonuclear
coadaptation was detected for locus P5CS only, and only in high salinity replicates,
corroborating the fact that selection for coadapted complexes is higher under salinity
stress.
Genomic studies have suggested that the genetic basis of adaptation to stressful
and divergent habitats requires only a few loci (Lexar and Fay 2005, Wilding et al 2001),
and the fairly quick recovery observed in high salinity replicates may argue for few loci
underlying hybrid breakdown. At the same time, Orr (1995) notes that hybrid
incompatibilities are likely to involve more than two loci and Edmands and Timmerman
(2003) showed that populations take longer to recover from the disruption of intrinsic
coadaptation than from the disruption of local adaptation, suggesting that many loci
underlie hybrid breakdown in T. californicus. Considering both of these principles,
selection for a few strong extrinsic interactions in the high salinity environment may have
been much greater than selection for several weaker intrinsic interactions under benign

230
conditions, resulting in faster time to purge outbreeding depression in replicates under
stress (Chapter 3).

Conclusions
The fate of moderately incompatible mixed populations is likely to be recovery
from initial outbreeding depression with extensive introgression in advanced generations
of hybridization. The possibility of genetic swamping due to stochastic effects, however,
does exist, and the direction of swamping, as seen in this experiment, could not be
predicted based on the relative fitness of the parental populations. Although strong
genetic determinism may not be the primary driver of evolution among populations
subject to a substantial amount of drift, stronger selection should result in higher
repeatability among replicates, and this is what was observed for replicates reared under
more stressful conditions. While we cannot determine whether the patterns we observe
among replicate trajectories are due to stronger drift, as a result of environmental
stochasticity under benign conditions, or stronger selection under stressful conditions, an
important finding from this study is that repeatability for both molecular and fitness
(Chapter 3) characters, appears to be largely dependent on an environmental component.
Managers might benefit from additional studies investigating the form of environmental
conditions (benign vs. mildly stressful vs. highly stressful) that may result in higher
repeatability, and thus greater predictability, of the outcomes of hybridization. Evidence
was also observed for both nuclear-nuclear and cytonuclear coadaptation after seven
generations of mating, with limited evidence for epistasis which has been suggested by
studies of early generation hybrids (Harrison and Edmands 2006, Willett 2006). After

231
seven generations of mating, deleterious epistasis generated early on may be purged,
leading to the recovery from outbreeding depression observed in Chapter 3.

232
CHAPTER 4 REFERENCES

Allendorf FW, Leary RF, Sprunell P, Wenburg JK (2001) The problems with hybrids:
setting conservation guidelines. Trends in Ecology and Evolution, 613-622.

Barilani M, Deregnaucourt S, Gallego S, et al. (2005) Detecting hybridization in wild
(Coturnix c. coturnix) and domesticated (Coturnix c. japonica) quail populations.
Biological Conservation 126, 445-455.

Beaumont MA, Nichols RA (1996) Evaluating loci for use in the genetic analysis of
population structure. Proceedings of the Royal Society of London B 263, 1619-
1626.

Belkhir K, Borsa P, Chikhi L, Raufaste N, Bonhomme F (2004) GENETIX 4.05, logiciel
sous Windows TM pour la genetique des populations. In: Laboratoire Genome,
Populations, Interactions. CNRS UMR 5171, Universite de Montellier II,
Montpellier, France.
Burton RS (1986) Incorporation of 14C-bicarbonate into the free amino acid pool during
hyperosmotic stress in an intertidal copepod. The Journal of Experimental
Zoology 238, 55-61.

Burton RS (1987) Differentiation and integration of the genome in populations of the
marine copepod Tigriopus californicus. Evolution 41, 504-513.

Burton RS (1990) Hybrid breakdown in developmental time in the copepod Tigriopus
californicus. Evolution 44, 1814-1822.

Burton RS (1997) Genetic evidence for long term persistence of marine invertebrate
populations in an ephemeral environment. Evolution 51, 993-998.

Burton RS, Byrne RJ, Rawson PD (2007) Three divergent mitochondrial genomes from
California populations of the copepod Tigriopus californicus. Gene 403, 53-59.

Burton RS, Feldman MW (1982) Changes in free amino acid concentrations during
osmotic response in the intertidal copepod Tigriopus californicus. Comparative
Biochemistry and Physiology 73A, 441-445.

Burton RS, Lee B-N (1994) Nuclear and mitochondrial gene genealogies and allozyme
polymorphism across a major phylogeographic break in the copepod Tigriopus
californicus. Proceedings of the National Academy of Sciences USA 91, 5197-
5201.

Carney SE, Gardner KA, Rieseberg LH (2000) Evolutionary changes over the fifty-year
history of a hybrid population of sunflowers (Helianthus). Evolution 54, 462-474.

233
Dybdahl MF (1994) Extinction, recolonization and the genetic structure of tidepool
copepod populations. Evolutionary Ecology 8, 113-124.

Edmands S (1999) Heterosis and outbreeding depression in interpopulation crosses
spanning a wide range of divergence. Evolution 53, 1757-1768.

Edmands S (2001) Phylogeography of the intertidal copepod Tigriopus californicus
reveals substantially reduced population differentiation at northern latitudes.
Molecular Ecology 10, 1743-1750.

Edmands S (2007) Between a rock and a hard place: evaluating the relative risks of
inbreeding and outbreeding for conservation and management. Molecular
Ecology 16, 463 - 475.

Edmands S, Deimler JK (2004) Local adaptation, intrinsic coadaptation and the effects of
environmental stress on interpopulation hybrids in the copepod Tigriopus
californicus. Journal of Experimental Marine Biology and Ecology 303, 183-196.

Edmands S, Feaman HV, Harrison JS, Timmerman CC (2005) Genetic consequences of
many generations of hybridization between divergent copepod populations.
Journal of Heredity 92, 114-123.

Edmands S, Harrison JS (2003) Molecular and quantitative trait variation within and
among populatioins of the intertidal copepod Tigriopus californicus. Evolution
57, 2277-2285.

Edmands S, Northrup SL, Hwang AS (in press) Maladapted gene complexes within
populations of the intertidal copepod Tigriopus californicus? Evolution.
Edmands S, Timmerman CC (2003) Modeling factors affecting the severity of
outbreeding depression. Conservation Biology 17, 883-892.

Ellison CK, Burton RS (2006) Disruption of mitochondrial function in interpopulation
hybrids of Tigriopus californicus. Evolution 60, 1382-1391.

Ellison CK, Burton RS (2008) Genotype-dependent variation of mitochondrial
transcriptional profiles in interpopulation hybrids. Proceedings of the National
Academy of Sciences 105, 15831-15836.

Ellison CK, Burton RS (2008) Interpopulation hybrid breakdown maps to the
mitochondrial genome. Evolution 62, 631-638.

Excoffier L, Laval G, Schneider S (2005) Arlequin ver. 3.0: An integrated software
package for population genetics data anlysis. Evolutionary Bioinformatics Online
1, 47-50.

234
Fenster CB, Galloway LF (2000) Inbreeding and Outbreeding Depression in Natural
Populations of Chamaecrista fasciculata (Fabaceae). Conservation Biology 14,
1406-1412.

Folmer O, Black M, Hoeh W, Lutz R, Vrijenhoek R (1994) DNA primers for
amplification of mitochondrial cytochrome c oxidase subunit I from diverse
metazoan invertebrates. Molecular Marine Biology and Biotechnology 3, 294-
299.

Goolish EM, Burton RS (1989) Energetics of osmoregulation in an intertidal copepod:
effects of anoxia and lipid reserves on the pattern of free amino acid
accumulation. Functional Ecology 3, 81-89.

Halis RS, Morley K (2005) Genes invading new populations: a risk assessment
perspective. Trends in Ecology and Evolution 20, 245-252.

Harrison JS, Burton RS (2006) Tracing hybrid incompatibilities to single amino acid
substitutions. Molecular Biology and Evolution 23, 559-564.

Harrison JS, Peterson DL, Swain JR, Edmands S (2004) Microsatellite DNA markers for
the intertidal copepod Tigriopus californicus. Molecular Ecology Notes 4, 736-
738.

Lexar C, Fay MF (2005) Adaptation to environmental stress: a rare or frequent driver of
speciation? Journal of Evolutionary Biology 18, 893-900.

Lynch M (1991) The genetic interpretation of inbreeding depression and outbreeding
depression. Evolution 45, 622-629.

Orr HA (1995) The population genetics of speciation: the evolution of hybrid
incompatibilities. Genetics 139, 1805-1813.

Pritchard JK, Stephens M, Donnelly P (2000) Inference of population structure using
multilocus genotypic data. Genetics 155, 945-959.

Rawson PD, Burton RS (2002) Functional coadaptation between cytochrome c and
cytochrome c oxidase within allopatric populations of a marine copepod.
Proceedings of the National Academy of Sciences USA 99, 12955-12958.

Raymond M, Rousset F (1995) GENEPOP (version 1.2) population genetic software for
exact tests and ecumenicism. Journal of Heredity 86, 248-249.

Rice WR, Hostert EE (1993) Laboratory experiments in speciation: What have we
learned in 40 years? Evolution 47, 1637-1653.

235
Seehausen O, Takimoto G, Roy D, Jokela J (2008) Speciation reversal and biodiversity
dynamics with hybridization in changing environments. Molecular Ecology 17,
30-44.

Turelli M, Orr HA (2000) Dominance, epistasis and the genetics of postzygotic isolation.
Genetics 154, 1663-1679.

Waples RS (1989) A generalized approach for estimating effective population size from
temporal changes in allele frequency. Genetics 121, 379-391.

Wilding CS, Butlin RK, Grahame J (2001) Differential gene exchange between
parapatric morphs of Littorina saxatilis detected using AFLP markers. Journal of
Evolutionary Biology 14, 611-619.

Willett CS (2006) Deleterious epistatic interactions between electron transport system
protein-coding loci in the copepod Tigiropus californicus. Genetics 173, 1465-
1477.

Willett CS, Burton RS (2001) Viability of cytochrome c genotypes depends on
cytoplasmic backgrounds in Tigriopus californicus. Evolution 55, 1592-1599.

Willett CS, Burton RS (2001) Viability of cytochrome c genotypes depends on
cytoplasmic backgrounds in Tigriopus californicus. Evolution 55, 1592-1599.

Willett CS, Burton RS (2002) Proline biosynthesis genes and their regulation under
salinity stress in the euryhaline copepod Tigriopus californicus. Comparative
Biochemistry and Physiology Part B 132, 739-750.

Willett CS, Burton RS (2003) Environmental influences on epistatic interactions:
viabilities of cytochrome c genotypes in interpopulation crosses. Evolution 57,
2286-2292.

Zhdanova OL, Pudovkin AI (2008) Nb_HetEx: A program to estimate the effective
number of breeders. Journal of Heredity Advance Access.

236
Conclusion

The examination of multiple generations of hybridization over an extended period
of time has allowed us to observe some of the transitory stages of hybridization that occur
before a swarm has stabilized. For conservation efforts that seek to make predictions
about the result of mixing populations this could offer an animal model to contrast with
known plant examples. Natural populations of T. claifornicus hybridize very rarely, but
this may serve as an extreme example for other taxa that hybridize more frequently. For
example, if recovery from outbreeding depression is observed in this species, despite
extreme genetic divergence and evidence for intrinsic coadaptation involving multiple
regions of the genome, then hopeful potential exists for recovery in other taxa.
Magnitude and duration of outbreeding depression. While it is well established
that T. californicus populations show hybrid breakdown in early hybrid generations (e.g.
Burton 1987, 1990, Burton et al. 1999, Edmands 1999) this thesis was concerned with
determining the duration and severity of outbreeding depression in multi-generation
swarms. All chapters confirmed an overall pattern of population recovery following
initial decreases in fitness. Chapter 1 provided the first experimental account of
Tigriopus swarm fitness at regular time intervals and demonstrated that hybrid swarms
can quickly recover from the effects of early-generation outbreeding depression and even
exceed midparent fitness. In Chapter 2 initial fitness declines were followed by rapid
recovery to superior parental fitness. Chapter 3 revealed that outbreeding depression is
equally as severe in either a stressful or novel environment. However, the generations of

237
maximum outbreeding depression and recovery both occur earlier in a stressful
environment.
Differing levels of interpopulation incompatibility. Because populations span a
range of divergence (Edmands 1999, 2001) the outcome of swarms that were initiated
between moderately incompatible populations (Chapter 1) could be compared to swarms
initiated with nearly incompatible populations (Chapter 2). While populations from
Chapter 1 showed uncharacteristically high fitness in backcross cohorts, and populations
from Chapter 2 produced backcrosses that had no offspring, both experiments showed
rapid recovery to parental fitness values. When source populations exhibited similar
fitness (Chapter 1), extensive introgression resulted, but if one population displayed a
much higher fitness in the experimental environment (Chapter 2), the likely outcome was
swamping. While hybridization between populations that show severe outbreeding
depression will not likely result in extinction of both groups, the long-term outcome may
not be the conservation of the desired group. Chapter 4 raised the possibility that genetic
swamping could also result due to stochastic effects so the direction of swamping, as seen
in this experiment, may not be easily predicted based on the relative fitness of the
parental populations. Again, there is evidence for very little gene flow among
populations of T. claifornicus in the wild. However, if this is taken as an extreme
example compared to other taxa such as salmon, where the introduction of unfit farmed
individuals to wild populations is a concern, then the persistence of the highly fit
population, despite the introduction of highly unfit individuals, is an encouraging
outcome.

238
Transgressive segregation. Morphological and fitness characters were examined
for transgressive segregation to determine if there was any evidence for highly fit novel
recombinants, which have previously been found a variety of taxa, most being plants
(Johansen-Morris and Latta 2006, Lexar et al. 2003, Ranganath and Aruna 2003,
Rieseberg et al. 2007). Although both Chapter 1 and Chapter 3 produced evidence for the
generation of transgressive phenotypes, there was no evidence of increased fitness or
extreme morphology compared to parentals after many generations of exposure to a novel
environment. Thus, the ability for hybridization to act as a creative force, resulting in
recombinants with enhanced fitness compared to parentals, may be limited in this species.
Contingency versus repeatability. Portions of this work uncovered some evidence
for deterministic forces at work following Tigriopus hybridization events. Multiple
experiments have arrived at the same outcome of increased RP alleles in RP-SD hybrid
swarms (Chapter 1, Chapter 4, Edmands et al. 2005) even when RP was not the superior
parent for all fitness measures. In Chapter 1, female PCA indicated that replicates of the
same treatment had similar morphologies even though different treatments were
somewhat distinct from each other. It is likely that there are potential deterministic forces
for morphological, fitness and molecular measures, but that they are limited by drift that
occurs as small populations fluctuate in size. An important finding from both Chapters 4
and 5 is that repeatability for both molecular and fitness characters, appears to be largely
dependent on an environmental component: stronger repeatability was observed for high
salinity replicates for both fitness and molecular measures. Even among high salinity
replicates showing strong patterns of repeatability after many generations, one high
salinity replicate showed evidence for a bottleneck event prior to generation seven. The

239
presence of a significant stochastic component in swarm evolution is further supported by
the evidence of maladapted cytonuclear (but not nuclear-nuclear) complexes observed in
Edmands et al. (in press, Appendix). The mitochondrial genome has a four-fold smaller
effective size and Chapter 4 showed that COI drifted to fixation in multiple replicates.
This same outcome was not observed for nuclear markers and supports the hypothesis
that cytonuclear complexes may be more likely to end up on suboptimal fitness peaks.
Environmental stress. Although strong genetic determinism may not be the
primary driver of evolution among populations subject to a substantial amount of drift,
stronger selection should result in higher repeatability among replicates, and this is what
was observed for replicates reared under more stressful conditions in Chapter 4. Chapter
4 also identified a candidate locus for selection, which could be investigated in the future
with a more high-density chromosome map or studies of gene expression. The more
stressful environment also resulted in hybrid replicates that were more fit than those
under benign conditions when exposed to a novel stress (Chapter 3), suggesting that
stress may reduce some of the impacts of outbreeding depression.
It is worth emphasizing that T. californicus is an extreme example of outbreeding
depression in which hybrid breakdown might be expected to be long lasting due to the
extreme genetic divergence between populations (on the order of groups that are
considered separate species) and evidence that multiple gene regions appear to contribute
to fitness reductions (Harrison and Edmands 2006). Therefore, if recovery occurs rapidly
in this system, it might be expected to occur at least as quickly in other taxa. This work
strongly implies that mixed populations are likely to recover from early-generation
outbreeding depression with a high chance of extensive introgression in advanced

240
generations of hybridization. If one population has significantly reduced fitness compared
to the second population, it is likely to be swamped by the superior population, but the
additional possibility of genetic swamping due to stochastic effects does exist. The
overarching results from all chapters of this dissertation support the notion that
outbreeding depression is likely to be a transitional phase in a population’s evolution
(Carney et al. 2000, Christiansen 2008, Rieseberg et al. 1996, Templeton 1986). Even for
populations showing significant outbreeding depression, recovery occurred rapidly in
multiple replicate trials and was hastened in a stressful environment. These findings do
offer hope for managers faced with situations where the long-term effects of outbreeding
may be a concern for the persistence of a species. Environmental or historical
contingency may limit our ability to accurately predict the outcome of mixing gene pools.
However, if hybrid populations can survive the fitness problems invoked in early
generations, the chance of recovery from outbreeding depression in later generations is
somewhat promising.

241
CONCLUSION REFERENCES
Burton RS (1987) Differentiation and integration of the genome in populations of the
marine copepod Tigriopus californicus. Evolution 41, 504-513.

Burton RS (1990) Hybrid breakdown in physiological response: A mechanistic approach.
Evolution 44, 1806-1813.

Burton RS, Rawson PD, Edmands S (1999) Genetic architecture of physiological
phenotypes: Empirical evidence for coadapted gene complexes. Amer. Zool. 39,
451-462.

Carney SE, Gardner KA, Rieseberg LH (2000) Evolutionary changes over the fifty-year
history of a hybrid population of sunflowers (Helianthus). Evolution 54, 462-474.

Christiansen FB (2008) Theories of Population Variation in Genes and Genomes
Princeton University Press, Princeton, New Jersey.

Edmands S (1999) Heterosis and outbreeding depression in interpopulation crosses
spanning a wide range of divergence. Evolution 53, 1757-1768.

Edmands S (2001) Phylogeography of the intertidal copepod Tigriopus californicus
reveals substantially reduced population differentiation at northern latitudes.
Molecular Ecology 10, 1743-1750.

Edmands S, Feaman HV, Harrison JS, Timmerman CC (2005) Genetic consequences of
many generations of hybridization between divergent copepod populations.
Journal of Heredity 92, 114-123.

Edmands S, Northrup SL, Hwang AS (in press) Maladapted gene complexes within
populations of the intertidal copepod Tigriopus californicus? Evolution.

Harrison JS, Edmands S (2006) Chromosomal basis of viability differences in Tigriopus
californicus interpopulation hybrids. Journal of Evolutionary Biology 19, 2040-
2051.

Johansen-Morris A, Latta R (2006) Fitness consequences of hybridization between
ecotypes of Avena barbata: hybrid breakdown, hybrid vigor and transgressive
segregation. Evolution 60, 1585-1595.

Lexar C, Welch ME, Raymond O, Rieseberg LH (2003) The origin of ecological
divergence in Helianthus paradoxus (Asteraceae): selection on transgressive
characters in a novel hybrid habitat. Evolution 57, 1989-2000.

242
Ranganath H, Aruna S (2003) Hybridization, transgressive segregation and evolution of
new genetic systems in Drosophila. Journal of Genetics 82, 163-177.

Rieseberg LH, Kim S-C, Randell RA, et al. (2007) Hybridization and the colonization of
novel habitats by annual sunflowers. Genetica 129, 149-165.

Rieseberg LH, SInervo B, Linder CR, Ungerer MC, Arias DM (1996) Role of gene
interactions in hybrid speciation: evidence from ancient and experimental
hybrids. Science 272, 741-745.

Templeton AR (1986) Coadaptation and Outbreeding Depression. In: Conservation
Biology, The Science of Scarcity and Diversity (ed. Soule ME), pp. 105-116.
Sinauer Associates, INC, Sunderland, Massachusetts.

243
Bibliography

Albertson RC, Kocher TD (2005) Genetic architecture sets limits on transgressive
segregation in hybrid cichlid fishes. Evolution 59, 686-690.

Allendorf FW, Leary RF, Sprunell P, Wenburg JK (2001) The problems with hybrids:
setting conservation guidelines. Trends in Ecology and Evolution, 613-622.

Armbruster P, Bradshaw WE, Holzapfel CM (1997) Evolution of the genetic architecture
underlying fitness in the pitcher plant mosquito, Wyeomyia smithii. Evolution 51,
451-458.

Arnold M (2006) Evolution Through Genetic Exchange Oxford University Press, Oxford,
UK.

Barilani M, Deregnaucourt S, Gallego S, et al. (2005) Detecting hybridization in wild
(Coturnix c. coturnix) and domesticated (Coturnix c. japonica) quail populations.
Biological Conservation 126, 445-455.

Beaumont MA, Nichols RA (1996) Evaluating loci for use in the genetic analysis of
population structure. Proceedings of the Royal Society of London B 263, 1619-
1626.

Belkhir K, Borsa P, Chikhi L, Raufaste N, Bonhomme F (2004) GENETIX 4.05, logiciel
sous Windows TM pour la genetique des populations. In: Laboratoire Genome,
Populations, Interactions. CNRS UMR 5171, Universite de Montellier II,
Montpellier, France.

Bijlsma R, Loeschcke V (2005) Environmental stress, adaptation and evolution: an
overview. Journal of Evolutionary Biology 18, 744-749.

Bleeker W, Schmitz U, Ristow M (2007) Interspecific hybridization between alien and
native plant species in Germany and its consequences for native biodiversity.
Biological Conservation 137, 248-253.

Bowman J, Kidd AG, Gorman RM, Schulte-Hostedde AI (2007) Assessing the potential
for impacts by feral mink on wild mink in Canada. Biological Conservation 139,
12-18.

Brochmann C, Borgen L, Stabbetorp OE (2000) Multiple diploid hybrid speciation of the
Canary Island endemic Argyranthemum sundingii (Asteraceae). Plant Systematics
and Evolution 220, 77-92.

244
Bubliy OA, Leoschcke V (2005) Correlated responses to selection for stress resistance
and longevity in a laboratory population of Drosopila melanogaster. Journal of
Evolutionary Biology 18, 789-803.

Burke JM, Arnold ML (2001) Genetics and the Fitness of Hybrids. Annual Review of
Genetics 35, 31-52.

Burke JM, Carney SE, Arnold ML (1998) Hybrid fitness in the Louisiana irises:
Analysis of parental and F1 performance. Evolution 52, 37-43.

Burton RS (1985) Mating system of the intertidal copepod Tigriopus californicus. Marine
Biology 86, 247-252.

Burton RS (1986) Evolutionary consequences of restricted gene flow in the copepod
Tigriopus californicus. Bulletin of Marine Science 39, 526-535.

Burton RS (1986) Incorporation of 14C-bicarbonate into the free amino acid pool during
hyperosmotic stress in an intertidal copepod. The Journal of Experimental
Zoology 238, 55-61.

Burton RS (1987) Differentiation and integration of the genome in populations of the
marine copepod Tigriopus californicus. Evolution 41, 504-513.

Burton RS (1990) Hybrid breakdown in developmental time in the copepod Tigriopus
californicus. Evolution 44, 1814-1822.

Burton RS (1990) Hybrid breakdown in physiological response: A mechanistic approach.
Evolution 44, 1806-1813.

Burton RS (1997) Genetic evidence for long term persistence of marine invertebrate
populations in an ephemeral environment. Evolution 51, 993-998.

Burton RS, Byrne RJ, Rawson PD (2007) Three divergent mitochondrial genomes from
California populations of the copepod Tigriopus californicus. Gene 403, 53-59.

Burton RS, Feldman MW (1982) Changes in free amino acid concentrations during
osmotic response in the intertidal copepod Tigriopus californicus. Comparative
Biochemistry and Physiology 73A, 441-445.

Burton RS, Feldman MW (1983) Physiological effects of an allozyme ploymorphism:
Glutamate-pyruvate transaminase and response to hyperosmotic stress in the
copepod Tigriopus californicus. Biochemical genetics 21, 239-251.

245
Burton RS, Lee B-N (1994) Nuclear and mitochondrial gene genealogies and allozyme
polymorphism across a major phylogeographic break in the copepod Tigriopus
californicus. Proceedings of the National Academy of Sciences USA 91, 5197-
5201.

Burton RS, Rawson PD, Edmands S (1999) Genetic architecture of physiological
phenotypes: Empirical evidence for coadapted gene complexes. Amer. Zool. 39,
451-462.

Carney SE, Gardner KA, Rieseberg LH (2000) Evolutionary changes over the fifty-year
history of a hybrid population of sunflowers (Helianthus). Evolution 54, 462-474.

Childs MR, Echelle AA, Dowling TE (1996) Development of the hybrid swarm between
pecos pupfish (Cyprinodontidae: Cyprinodon pecosensis) and sheepshed minnow
(Cyprinodon variegatus): A perspective fom allozymes an mtDNA. Evolution 50,
2014-2022.

Chippindale AK, Gibbs AG, Sheik M, et al. (1998) Resource acquisition and the
evolution of stress resistance in Drosophila melanogaster. Evolution 52, 1342-
1352.

Christiansen FB (2008) Theories of Population Variation in Genes and Genomes
Princeton University Press, Princeton, NewJersey.

Djawdan M, Chippindale AK, Rose MR, Bradley TJ Metabolic reserves and evolved
stress resistance in Drosophila melanogaster. Physiological Zoology 71, 584-594.

Dowling TE, Secor CL (1997) The role of hybridization and introgression in the
diversification of animals. Annual Review of Ecology and Systematics 28, 593-
619.

Dybdahl MF (1994) Extinction, recolonization and the genetic structure of tidepool
copepod populations. Evolutionary Ecology 8, 113-124.

Edmands S (1999) Heterosis and outbreeding depression in interpopulation crosses
spanning a wide range of divergence. Evolution 53, 1757-1768.

Edmands S (2001) Phylogeography of the intertidal copepod Tigriopus californicus
reveals substantially reduced population differentiation at northern latitudes.
Molecular Ecology 10, 1743-1750.

Edmands S (2007) Between a rock and a hard place: evaluating the relative risks of
inbreeding and outbreeding for conservation and management. Molecular
Ecology 16, 463 - 475.

246
Edmands S (2008) Recombination in interpopulation hybrids of the copepod Tigriopus
californicus: release of beneficial variation despite hybrid breakdown. Journal of
Heredity 99, 316-318.

Edmands S, Burton RS (1999) Cytochrome c oxidase activity in interpopulation hybrids
of a marine copepod: a test for nuclear-nuclear or nuclear-cytoplasmic
coadaptation. Evolution 53, 1972-1978.

Edmands S, Deimler JK (2004) Local adaptation, intrinsic coadaptation and the effects of
environmental stress on interpopulation hybrids in the copepod Tigriopus
californicus. Journal of Experimental Marine Biology and Ecology 303, 183-196.

Edmands S, Feaman HV, Harrison JS, Timmerman CC (2005) Genetic consequences of
many generations of hybridization between divergent copepod populations.
Journal of Heredity 92, 114-123.

Edmands S, Harrison JS (2003) Molecular and quantitative trait variation within and
among populatioins of the intertidal copepod Tigriopus californicus. Evolution
57, 2277-2285.

Edmands S, Northrup SL, Hwang AS (in press) Maladapted gene complexes within
populations of the intertidal copepod Tigriopus californicus? Evolution.

Edmands S, Timmerman CC (2003) Modeling factors affecting the severity of
outbreeding depression. Conservation Biology 17, 883-892.

Egloff DA (1966) Ecological aspects of sex ratio and reproduction in experimental and
field populations of the marine copepod Tigrioups californicus Dissertation,
Stanford University.

Ellison CK, Burton RS (2006) Disruption of mitochondrial function in interpopulation
hybrids of Tigriopus californicus. Evolution 60, 1382-1391.

Ellison CK, Burton RS (2008) Genotype-dependent variation of mitochondrial
transcriptional profiles in interpopulation hybrids. Proceedings of the National
Academy of Sciences 105, 15831-15836.

Ellison CK, Burton RS (2008) Interpopulation hybrid breakdown maps to the
mitochondrial genome. Evolution 62, 631-638.

Ellstrand NC, Elam DR (1993) Population genetic consequences of small population size:
Implications for plant conservation. Annual Review of Ecology and Systematics
24, 217-242.

247
Ellstrand NC, Schierenbeck KA (2006) Hybridization as a stimulus for the evolution of
invasiveness in plants? Euphytica 148, 35-46.

Erickson DL, Fenster CB (2006) Intraspecific hybridization and the recovery of fitness in
the native legume Chamaecrista fasciculata. Evolution 60, 225-233.

Excoffier L, Laval G, Schneider S (2005) Arlequin ver. 3.0: An integrated software
package for population genetics data anlysis. Evolutionary Bioinformatics Online
1, 47-50.

Falconer DS, Mackay TFC (1996) Introduction to quantitative genetics, 4 edn. Longman
Science and Technology, Harlow, UK.

Fenster CB, Galloway LF (2000) Inbreeding and Outbreeding Depression in Natural
Populations of Chamaecrista fasciculata (Fabaceae). Conservation Biology 14,
1406-1412.

Fenster CB, Galloway LF (2000) Population differentiation in an annual legume: genetic
architecture. Evolution 54, 1157-1172.

Fitzpatrick BM, Bradley H (2004) Environment-dependent admixture dynamics in a tiger
salamander hybrid zone. Evolution 58, 1282-1293.

Folmer O, Black M, Hoeh W, Lutz R, Vrijenhoek R (1994) DNA primers for
amplification of mitochondrial cytochrome c oxidase subunit I from diverse
metazoan invertebrates. Molecular Marine Biology and Biotechnology 3, 294-
299.

Frankham R (2005) Stress and adaptation in conservation genetics. Journal of
Evolutionary Biology 18, 750-755.

Fraser DJ, Cook AM, Eddington JD, Bentzen P, Hutchings JA (2008) Mixed evidence for
reduced local adaptation in wild salmon resulting from interbreeding with escaped
farmed salmon: complexities in hybrid fitness. Evolutionary Applications 1, 501-
512.

Ganz HH, Burton RS (1995) Genetic differentiation and reproductive incompatibility
among Baja California populations of the copepod Tigriopus californicus. Marine
Biology 123, 821-827.

Gilchrist GW, Lee CE (2007) All stressed out and nowhere to go: does evolvability limit
adaptation in invasive species? Genetica 129, 127-132.

Gilk SE, Wang IA, Hoover CL, et al. (2004) Outbreeding depression in hybrids between
spatially separated pink salmon, Oncorhynchus gorbuscha, populations: marine

248
survival, homing ability, and variability in family size. Environmental Biology of
Fishes 69, 287-297.

Gompert Z, Fordyce JA, Forister ML, Shapiro AM, Nice CC (2006) Homoploid Hybrid
Speciation in an Extreme Habitat. Science 314, 1923-1925.

Goolish EM, Burton RS (1989) Energetics of osmoregulation in an intertidal copepod:
effects of anoxia and lipid reserves on the pattern of free amino acid
accumulation. Functional Ecology 3, 81-89.

Gow JL, Peichel CL, Taylor EB (2006) Contrasting hybridization rates between
sympatric three-spined sticklebacks highlight the fragility of reproductive barriers
between evolutionarily young species. Molecular Ecology 15, 739-752.

Grant PR, Grant BR (1992) Hybridization of bird species. Science 256, 193-197.

Griffing B (1990) Use of a controlled-nutrient experiment to test heterosis hypotheses.
Genetics 126, 753-767.

Gross BL, Kane NC, Lexar C, et al. (2004) Reconstructing the origin of Helianthus
deserticola: Survival and selection on the desert floor. The American Naturalist
164, 145-156.

Gross BL, Rieseberg LH (2004) The ecological genetics of homoploid hybrid speciation.
Journal of Heredity 96, 241-252.

Halis RS, Morley K (2005) Genes invading new populations: a risk assessment
perspective. Trends in Ecology and Evolutio 20, 245-252.

Harrison JS, Burton RS (2006) Tracing hybrid incompatibilities to single amino acid
substitutions. Molecular Biology and Evolution 23, 559-564.

Harrison JS, Edmands S (2006) Chromosomal basis of viability differences in Tigriopus
californicus interpopulation hybrids. Journal of Evolutionary Biology 19, 2040-
2051.

Harrison JS, Peterson DL, Swain JR, Edmands S (2004) Microsatellite DNA markers for
the intertidal copepod Tigriopus californicus. Molecular Ecology Notes 4, 736-
738.

Harshman LG, Hoffmann AA, Clark AG (1999) Selection for starvation resistance in
Drosophila melanogaster: physical correlates, enzyme activities and multiple
stress responses. Journal of Evolutionary Biology 12, 370-379.

249
Hedgecock D, Lin J-Z, DeCola S, et al. (2007) Transcriptomic analysis of growth
heterosis in larval Pacific oysters (Crassostrea gigas). Proceedings of the
National Academy of Sciences 104, 2313-2318.

Hoffman AA, Parsons PA (1991) Evolutionary Genetics and Environmental Stress
Oxford University Press, Oxford.

Hoffmann AA, Parsons PA (1989) Selection for increased desiccation resistance in
Drosophila melanogaster: additive genetic control and correlated responses for
other stresses. Genetics 122, 837-845.

Hutchings JA, Fraser DJ (2008) The nature of fisheries-and farming-induced evolution.
Molecular Ecology 17, 294-313.

Johansen-Morris A, Latta R (2006) Fitness consequences of hybridization between
ecotypes of Avena barbata: hybrid breakdown, hybrid vigor and transgressive
segregation. Evolution 60, 1585-1595.

Johnston JA, Donovan LA, Arnold ML (2004) Novel phenotypes among early generation
hybrids of two Louisiana iris species: flooding experiments. Journal of Ecology
92, 967-976.

Kothera L, Ward SM, Carney SE (2007) Assessing the threat from hybridization to the
rare endemic Physaria bellii Mullican (Brassicaceae). Biological Conservation
140, 110-118.

Latta RG, Gardner KM, Johansen-Morris AD (2007) Hybridization, recombination and
the genetic basis of fitness variation across environments in Avena Barbata
Genetica 129, 167-177.

Levin DA, Francisco-Ortega J, Jansen RK (1996) Hybridization and the extinction of rare
plant species. Conservation Biology 10, 10-16.

Lexar C, Fay MF (2005) Adaptation to environmental stress: a rare or frequent driver of
speciation? Journal of Evolutionary Biology 18, 893-900.

Lexar C, Welch ME, Raymond O, Rieseberg LH (2003) The origin of ecological
divergence in Helianthus paradoxus (Asteraceae): selection on transgressive
characters in a novel hybrid habitat. Evolution 57, 1989-2000.

Lu C, Brauer MJ, Botstein D (2009) Slow growth induces heat-shock resistance in
normal and repiatory-deficient yeast. Molecular Biology of the Cell 20, 891-903.

Lynch M (1991) The genetic interpretation of inbreeding depression and outbreeding
depression. Evolution 45, 622-629.

250
Lynch M, Walsh B (1998) Genetics and Analysis of Quantitative Traits Sinauer,
Sunderland, MA.

McClelland EK, Naish KA (2007) What is the fitness outcome of crossing unrelated fish
populations? A meta-analysis and an evaluation of future research directions.
Conservation Genetics 8, 397-416.

McGinnity P, Prodohl P, Ferguson A, et al. (2003) Fitness reduction and potential
extinction of wild populations of Atlantic salmon, Salmo salar, as a result of
interactions with escaped salmon. Proceedings of the Royal Society of London B
270, 2443-2450.

Norrmann GA (2009) Natural hybridization in the Andropogon lateralis comples
(Andropogoneae, Poaceae) and its impact on taxonomin literature. Botanical
Journal of the Linnean Society 159, 136-154.

Orr HA (1995) The population genetics of speciation: the evolution of hybrid
incompatibilities. Genetics 139, 1805-1813.

Orr HA (1996) Dobzhansky, Bateson and the genetics of speciation. Genetics 144, 1331-
1335.

Palmer CA, Edmands S (2000) Mate choice in the fact of both inbreeding and
outbreeding depression in the intertidal copepod Tigriopus californicus. Marine
Biology 136, 693-698.

Pritchard JK, Stephens M, Donnelly P (2000) Inference of population structure using
multilocus genotypic data. Genetics 155, 945-959.

Randi E (2008) Detecting hybridization between wild species and their domesticated
relatives. Molecular Ecology 17, 285-293.

Ranganath H, Aruna S (2003) Hybridization, transgressive segregation and evolution of
new genetic systems in Drosophila. Journal of Genetics 82, 163-177.

Rawson PD, Burton RS (2002) Functional coadaptation between cytochrome c and
cytochrome c oxidase within allopatric populations of a marine copepod.
Proceedings of the National Academy of Sciences USA 99, 12955-12958.

Raymond M, Rousset F (1995) GENEPOP (version 1.2) population genetic software for
exact tests and ecumenicism. Journal of Heredity 86, 248-249.

Rhymer JM, Simberloff D (1996) Extinction by hybridization and introgression. Annual
Review of Ecology and Systematics 27, 83-109.

251
Rice WR, Hostert EE (1993) Laboratory experiments in speciation: What have we
learned in 40 years? Evolution 47, 1637-1653.

Rieseberg LH (2000) Crossing relationships among ancient and experimental sunflower
hybrid lineages. Evolution 54, 859-865.

Rieseberg LH, Archer MA, Wayne RK (1999) Transgressive segregation, adaptation and
speciation. Heredity 83, 363-372.

Rieseberg LH, Baird SJE, Desrochers AM (1998) Patterns of mating in wild sundlower
and hybrid zones. Evolution 52.

Rieseberg LH, Kim S-C, Randell RA, et al. (2007) Hybridization and the colonization of
novel habitats by annual sunflowers. Genetica 129, 149-165.

Rieseberg LH, Raymond O, Rosenthal DM, et al. (2003) Major ecological transitions in
wild sunflowers facilitated by hybridization. Science 301, 1211-1216.

Rieseberg LH, Sinervo B, Linder CR, Ungerer MC, Arias DM (1996) Role of gene
interactions in hybrid speciation: evidence from ancient and experimental
hybrids. Science 272, 741-745.

Rieseberg LH, Widmer A, Arntz AM, Burke JM (2003) The genetic architecture
necessary for transgressive segregation is common in both natural and
domesticated populations. Philosophical Transactions of the Royal Society of
London B 358, 1141-1147.

Roberge C, Normandeau E, Einum S, Guderley H, Bernatchez L (2008) Genetic
consequences of interbreeding between farmed and wild Atlantic salmon: insights
from the transcriptome. Molecular Ecology 17, 314-324.

Rudulier DL, Strom AR, Dandekar AM, Smith LT, Valentine RC (1984) Molecular
biology of osmoreguation. Science 224, 1064-1068.

Schwarzbach AE, Rieseberg LH (2002) Likely multiple origins of a diploid hybrid
sunflower species. Molecular Ecology 11, 1703-1715.

Seehausen O (2004) Hybridization and adaptive radiation. Trends in Ecology and
Evolution 19, 198-207.

Seehausen O, Takimoto G, Roy D, Jokela J (2008) Speciation reversal and biodiversity
dynamics with hybridization in changing environments. Molecular Ecology 17,
30-44.

252
Seo JS, Lee K-W, Rhee J-S, et al. (2006) Environmental stressors (salinity, heavy metals,
H
2
O
2
)modulate expression of glutathione reductase (GR) gene from the intertidal
copepod Tigriopus japonicus. Aquatic Toxicology 80, 281-289.

Simoes P, Santos J, Fragata I, et al. (2008) How repeatable is adaptive evolution? The
role of geographical origin and founder effects in laboratory adaptation. Evolution
62, 1817-1829.

Tallmon DA, Luikart G, Waples RS (2004) The alluring simplicity and complex reality
of genetic rescue. Trends in Ecology and Evolution 19, 489-496.

Taylor EB, Boughman JW, Groenenboom M, et al. (2006) Speciation in reverse:
morphological and genetic evidence of the collapse of a three-spined stickleback
(Gasterosteus aculeatus) species pair. Molecular Ecology 14, 343-355.

Templeton AR (1986) Coadaptation and Outbreeding Depression. In: Conservation
Biology, The Science of Scarcity and Diversity (ed. Soule ME), pp. 105-116.
Sinauer Associates, INC, Sunderland, Massachusetts.

Turelli M, Barton NH, Coyne JA (2001) Theory and speciation. Trends in Ecology and
Evolution 16, 330-343.

Turelli M, Orr HA (2000) Dominance, epistasis and the genetics of postzygotic isolation.
Genetics 154, 1663-1679.

Tymchuk WE, Sundstrom LF, Devlin RH (2007) Growth and survival trade-offs and
outbreeding depression in rainbow trout (Oncorhynchus mykiss). Evolution 61,
1225-1237.

Ureta MS, Cantamutto M, Carrera A, Delucchi C, Poverene M (2008) Natural hybrids
between cultivated and wild sunflowers (Helianthus spp.) in Argentina. Genet
Resour Crop Evol 55, 1267-1277.

Vetukhiv MO, Beardmore JA (1959) Effect of environment upon the manifestation of
heterosis and homeostasis in Drosophila pseudoobscura. Genetics 44, 759-768.

Vittor BA (1971) Effects of the environment on fitnbess-related life history character in
Tigriopus californicus Dissertation, University of Oregon.

Wade MJ, Goodnight CT (1998) The theories of Fisher and Wright in the context of
metapopulations: When nature does many small experiments. Evolution 52, 1537-
1553.

Waples RS (1989) A generalized approach for estimating effective population size from
temporal changes in allele frequency. Genetics 121, 379-391.

253
Wilding CS, Butlin RK, Grahame J (2001) Differential gene exchange between
parapatric morphs of Littorina saxatilis detected using AFLP markers. Journal of
Evolutionary Biology 14, 611-619.

Willett CS (2006) Deleterious epistatic interactions between electron transport system
protein-coding loci in the copepod Tigriopus californicus. Genetics 173, 1465-
1477.

Willett CS, Burton RS (2001) Viability of cytochrome c genotypes depends on
cytoplasmic backgrounds in Tigriopus californicus. Evolution 55, 1592-1599.

Willett CS, Burton RS (2002) Proline biosynthesis genes and their regulation under
salinity stress in the euryhaline copepod Tigriopus californicus. Comparative
Biochemistry and Physiology Part B 132, 739-750.

Willett CS, Burton RS (2003) Environmental influences on epistatic interactions:
viabilities of cytochrome c genotypes in interpopulation crosses. Evolution 57,
2286-2292.

Zhdanova OL, Pudovkin AI (2008) Nb_HetEx: A program to estimate the effective
number of breeders. Journal of Heredity Advance Access.

254
Appendix

Maladapted gene complexes within populations of the intertidal copepod Tigriopus
californicus?

Suzanne Edmands
1,2
, Sara L. Northrup
1,3
and AnnMarie S. Hwang
1,4
1
Department of Biological Sciences, University of Southern California, Los Angeles,
California 90089-0371
2
E-mail: sedmands@usc.edu
3
Present address: Department of Zoology, University of British Columbia, Vancouver,
BC V6T 1Z4, Canada; E-mail: snorthru@zoology.ubc.ca
4
E-mail: achinen@usc.edu

APPENDIX ABSTRACT
The prevalence of F2 hybrid breakdown in interpopulation crosses of the marine
copepod Tigriopus californicus can be explained by disruption of coadapted gene
complexes. This study further dissects the nature of hybrid gene interactions, revealing
that parental populations may also harbor maladapted gene complexes. Diagnostic
molecular markers (14) were assayed in reciprocal F2 hybrids to test for gene
interactions affecting viability. Results showed some evidence of nuclear-nuclear
coadaptation. While there were no significant examples of pairwise linkage
disequilibrium between physically unlinked loci, one of the two reciprocal crosses did

255
show an overall excess of parental double homozygotes and an overall dearth of non-
parental double homozygotes. In contrast, the nuclear-cytoplasmic data showed a
stronger tendency toward maladaptation within the specific inbred lines used in this
study. For three out of four loci with significant frequency differences between reciprocal
F2, homozygotes were favored on the wrong cytoplasmic background. A separate study
of reciprocal backcross hybrids between the same two populations (but different inbred
lines) revealed faster development time when the full haploid nuclear genome did not
match the cytoplasm. The occurrence of such suboptimal gene complexes may be
attributable to effects of genetic drift in small, isolated populations.

APPENDIX INTRODUCTION
New alleles arising in a lineage through mutation or migration must be tested
against the background of alleles already present. Selection for harmonious interactions
may therefore promote coadapted gene complexes whose joint action increases the fitness
of the whole organism (Wright 1969, Mayr 1970, Wallace 1981). The prevalence of
different coadapted gene complexes within species has been much debated (see Wallace
1991), due largely to the vulnerability of all but the most tightly linked gene
combinations to breakdown by recombination.
The threat of disruption by recombination is minimized for conspecific
populations connected by low levels of gene flow. In this case, both local selection and
genetic drift may promote different sets of beneficial epistatic interactions in each
population. Evidence for different sets of harmoniously interacting genes in
geographically isolated populations comes from numerous examples of reduced fitness

256
(and increased variance) in second-generation interpopulation hybrids (see reviews in
Endler 1977 and Edmands 2007). It is in these second generation hybrids that genomic
coadaptation is first disrupted by meiotic recombination, leading to the phenomenon of
hybrid breakdown.
Investigations into the evolution of beneficial gene interactions have been central
to our understanding of postzygotic reproductive isolation, including the mechanisms
underlying Haldane’s Rule and the Large X Effect (e.g. Coyne and Orr 2004). Beneficial
epistasis and its breakdown by hybridization are also critical issues for conservation as
populations and species are increasingly being hybridized through both accidental
introductions and intentional translocations (Allendorf et al. 2001; Edmands 2007).
Predictions of the severity and duration of hybrid breakdown depend on a better
understanding of the numbers of loci involved and linkage relationships among them.
The tidepool copepod Tigriopus californicus is becoming an established model
for the genetic basis of hybrid breakdown. Despite a seemingly high potential for
dispersal, populations are genetically differentiated over short geographic distances, with
mitochondrial DNA differences exceeding 20% (Burton and Lee 1994, Edmands 2001).
Interpopulation crosses typically result in F1 hybrids with similar or higher fitness than
their parents, and F2 and backcross hybrids with reduced fitness (e.g. Burton 1986,
Burton 1990ab, Edmands 1999). While work on hybrid breakdown in other systems has
focused largely on nuclear-nuclear coadaptation, much work in Tigriopus has focused on
nuclear-mitochondrial coadaptation, particularly those interactions necessary for
mitochondrial energy production (e.g. Burton et al. 2006). The likelihood of different
nuclear-mitochondrial gene complexes in different populations is enhanced by the

257
particularly rapid rate of mitochondrial DNA evolution in T. californicus (protein-coding
mitochondrial genes evolve approximately 25- to 40-fold faster than protein-coding
nuclear genes, Willett and Burton 2004), and the prospect for detailed study of these
nuclear-mitochondrial complexes is enhanced by the availability of full mitochondrial
genome sequences for several populations (Burton et al. 2007).
This study uses reciprocal F2 and backcross hybrids to assess both nuclear-
nuclear and nuclear-cytoplasmic interactions. Fourteen diagnostic molecular markers
distributed throughout the genome are used to look at gene interactions affecting
viability. The simple prediction is that coadaptation within natural populations should
promote parental gene combinations which have higher fitness than hybrid gene
combinations.

MATERIALS AND METHODS
Study Species
Tigriopus californicus is a harpacticoid copepod whose reproductive biology has
been well studied (e.g. Egloff 1967: Vittor 1971; Burton 1985). Sexes are separate,
outcrossing is obligatory, and even the most divergent populations have not shown
evidence of prezygotic isolation (Ganz and Burton 1995, Palmer and Edmands 2000).
Adult males use their geniculate first antennae to clasp immature females and guard them
until the female reaches sexual maturity. Males then mate with the females before
releasing them. Virgin females can therefore be obtained by placing a clasped pair on a
piece of filter paper and teasing the copepods apart under a dissecting microscope using a
fine probe. Females mate only once and use stored sperm to fertilize multiple broods of

258
offspring. Inbred lines are easily created by isolating a single gravid female and allowing
full siblings and their subsequent progeny to mate freely. Development involves six
naupliar (larval) stages and 5 copepodid (juvenile) stages before reaching adulthood.
Adult males can be distinguished by the hooks at the end of their first antennae.
Recombination in this species is restricted to males (Ar-rushdi 1963, Burton et al. 1981)
and mitochondria are inherited maternally (Lee 1993).

F2 Crosses
Breeding design
T. californicus were collected from intertidal rock pools at two sites in southern
California: Laguna Beach (LB, 33º 33’N, 117º 47’ W, November 2001) and Royal Palms
(RP, 33º 42’N, 118º 19’W, October 2000). These two populations have been found to be
~17% divergent in mitochondrial DNA (COI; Edmands 2001 and D. Peterson, unpub.
data). All cultures were kept in a 20ºC incubator with a 12 h light: 12 h dark cycle. Stock
cultures were maintained in 400 ml beakers in natural seawater supplemented with
commercial flake-type fish food and Spirulina algae.
Isofemale lines from the two populations were created by placing a single gravid
female in a Petri dish with filtered seawater (37 mm) containing 0.2 mg finely ground
Spirulina per ml. Lines were maintained for 6-8 months before reciprocal F2 crosses (RP
female x LB male and LB female x RP male) were initiated in October 2002. Given that
the species has overlapping generations and a minimum generation time of 23 days at
20ºC (Burton 1987), this 6-8 months of inbred line maintenance corresponds to a
maximum of ~10 generations of inbreeding. Crosses were begun by uniting 5 virgin

259
females from the first population with 5 adult males from the second population in each
Petri dish. A total of approximately 25 pairs were established for each of the two crosses.
Dishes were checked 3 times per week. When females formed egg sacs they were
transferred to a new dish with new filtered seawater and Spirulina. After the female laid
several clutches she was again transferred to a new dish. When F1 offspring formed
clasped pairs the pairs were dissected apart and reunited with partners descending from a
different original dish in order to avoid additional inbreeding. Five F1 virgin females and
five adult F1 males were placed in each dish. These dishes were monitored 7 days a
week. When females formed egg sacs they were transferred to a new dish.
F2 assays utilized the fastest developing male from each brood. Mature males are
easier to recognize than mature females, and minimum male development time was easily
determined by recording the date that F2 larvae hatched and the date that the first male
matured (formed clasping antennae) for the first clutch of eggs for each F1 female.
Development time is expected to be tightly correlated with fitness for continuously
breeding species like Tigriopus (Lewontin 1974). Individual males were rinsed briefly
with diH
2
O, blotted dry on filter paper and frozen at -70º C for subsequent molecular
analysis.
Molecular assays
DNA was extracted by placing individual frozen copepods in 50ml lysis buffer
(10mM Tris pH 8.3, 50mM KCl, 0.5% Tween 20 and 200 μg/ml Proteinase K) and
incubating at 65º C for 1h followed by 100º C for 15 min. Individuals were scored for 13
diagnostic microsatellite loci using methods previously described (Harrison et al. 2004;
Edmands et al. 2005). These 13 loci were previously mapped to chromosome using a

260
non-recombinant backcross (Harrison et al. 2004; Edmands et al. 2005). Individuals
were also scored for population-specific fragments of the cytochrome-c oxidase gene
using methods developed for this study. Cytochrome c was mapped to chromosome in
the same non-recombinant backcross panel used in previous studies (Harrison et al. 2004,
Edmands et al. 2005). Primers and assay conditions for all markers are listed in Table 1.
For all PCR the forward primers were fluorescently labeled and run on a CEQ 8000
capillary sequencer (Beckman Coulter, Fullerton, CA, USA).
For screening of cytochrome c, forward primers, ccAB.fZip and ccSD.fZip, were
developed by modifying primers from Willett & Burton, 2001. A 25 bp ZipCode
sequence (Chen et al. 2000) was added to the 5’end of each forward primer to allow for
more economical assays. Fluorescently labeled ZipCode primers were used for
amplification of fragments containing these modified forward primers. Reverse primer
cyt14.r (Willett and Burton, 2001) was used with each of the forward primers. Primers
ccAB.fZip and cyt14.r amplify a 650 bp fragment from the LB population and primers
ccSD.fZip and cyt14.r amplify a 416 bp fragment from the SD population. Separate
PCRs for each primer set were performed on each individual (annealing temperatures for
the population-specific primers are in Table A-1, annealing temperature for the ZipCode
primer was 57 ºC). Reactions were carried out in a 12 μl volume and contained 1X PCR
buffer, 0.25 mM dNTPs, 2.5 mM MgCl
2,
0.33 μmol forward ZipCode primer, 1 μmol
reverse primer, 0.67 μmol fluorescently labeled ZipCode and 0.4 units of Taq
polymerase. Cycling conditions were as follows: 4 min at 94 ºC; 22 cycles of 30 s at 94
ºC, 35 s at 57-60 ºC, 1 min at 72 ºC; 15 cycles of 30 s at 94 ºC, 40 s at 57 ºC, 1.5 min 72
ºC; 10 min at 72 ºC.

261
Table A-1. Marker information: locus, code (chromosome_abbreviated name), marker
type (C, codominant; D, dominant), forward and reverse primers and annealing
temperature (T
a
, Celsius).

Locus Code Marker
type
Forward primer
(5’-3’)
Reverse primer
(5’-3’)
T
a

TCS030 1_30 C cattcccgaacgaagac
g
ttaaaagaaccaaa
cgcacg
55
TCS558 1_558 C cgagaacataacttcaa
acgaaac
gtacatctgtgcatg
gtccac
55
TCS228 2_228 C aatcgagttggcatcctt
aga
ggtcatatcttgcga
ttgaga
55
TC1203 5_1203 C gcgttcaactctcgaaat
ca
tccttatctcctcatc
ccataga
55
TC62J8 5_62J8 C acggtcatctcaatgctg
aa
ggtgaaaaatcgg
aaaacca
55
TC1555B 6_1555B D gatttggtgttggagacg
cc
gatcgacaaatcac
acacac
55
TCCytC-
LB
6_cytC C gatgatcgacgagacac
tctcgccaccgacagac
gggcaaggcctct
ggaatgtactttttg
gggttcg
60
TCCytC-
SD
6_cytC C gatgatcgacgagacac
tctcgccaccacaaaat
gcccctcgctcg
ggaatgtactttttg
gggttcg
58

262
Table A-1, Continued
TC56J2 7_56J2 C ctcccaacgctggtatta
gg
aagatggggcaaa
ggaattt
55
TCS480 8_480 C gctgtccacccaaccaa
c
tgaaactgccaaca
agatccatac
61
TCS061 8_61 C ccaacgactgacgggtc
c
atccgcgagtcga
gagtatg
55
TCS197 9_197 C tgtttgccaaccaaagtg
aa
cacagtatgaagaa
gccagtcc
55
TC1814 9_1814 C tttttctgctcgagcgtttt ccgtcgtctcgagc
tcttt
55
TC1555 10_1555 C gatttggtgttggagacg
cc
gatcgacaaatcac
acacac
55
TC1202 11_1202 C gcgatgcgtgtaataatg
g
tgatttgttacctcgc
ctga
55

263
Data analyses
Analysis of development times within and between reciprocal crosses done using
Statistica 7.1 (StatSoft, Tulsa, OK). Map Manager QTX version b20 (Manly et al. 2001)
was used to determine linkage associations for each of the two crosses separately, using
the Kosambi mapping function and a linkage criterion of P = 0.05. Calculation of allele
and genotype frequencies, 2-locus contingency tables and linkage disequilibrium was
performed by Genepop 4.0 (Raymond and Rousset 1995). Chi square tests were used to
compare single locus genotype numbers to Mendelian expectations for each cross, and to
compare genotype numbers between reciprocal crosses. To test for epistatic interactions
affecting survival, observed two-locus genotype numbers were compared to expected
numbers determined by multiplying single-locus ratios.

Backcrosses
Reciprocal backcrosses between RP and LB were previously analyzed for the
effects of recombination (Edmands 2008). Here, we reanalyze the non-recombinant
crosses only to test the effects of cytoplasmic background. Experimental details are given
in Edmands 2008. In brief, isofemale lines for the RP and LB populations (not the same
lines used for F2 crosses) were maintained for 2-4 months before crosses began.
Reciprocal F1 hybrids (RP female x LB male and LB female x RP male) were
backcrossed to LB. Each clutch of backcross offspring was maintained in a separate Petri
dish and monitored daily. Minimum male development time was defined as described
above. Minimum female development time was defined as the time from hatching until
the first female in the clutch extruded an egg sac. Because we focused on only the fastest

264
developing individuals from each brood, both the F2 and backcross studies targeted the
"best" genotypic combinations (at least in terms of development time), whether they were
parental or nonparental combinations. In this way we did not sample highly dysfunctional
genotypes that are least likely to contribute to future evolution.

RESULTS
F2 Hybrids
Development time was significantly faster in F2 hybrid males with RP cytoplasm
(15.8
+
0.2d) than in those with LB cytoplasm (17.3
+
0.2d) (Figure A-1A). Fourteen
diagnostic markers were scored in an average of 140.8 F2 hybrid individuals per locus
per cross. Linkage analyses show that 8 of these markers are unlinked. Loci 1_30
(chromosome_locus) and 1_558 were found to be between 22.7 cM apart (cross with LB
cytoplasm) and 25.1 cM apart (cross with RP cytoplasm). Loci 5_62J8 and 5_1203 were
found to be between 24.6 cM apart (cross with LB cytoplasm) and 25.1 cM apart (cross
with RP cytoplasm). Loci 9_197 and 9_1814 were found to be between 12.6 cM apart
(cross with LB cytoplasm) and 19.1 cM apart (cross with RP cytoplasm). Loci 6_1555B
and 6_cytC were not significantly linked in the current F2 data set, but are known to
reside on the same chromosome based on data for non-recombinant backcross hybrids.
Single-locus genotype data (Table A-2) showed distorted ratios at 9 of 14 loci in
each of the reciprocal crosses. For the 13 codominant loci there were seven cases of
significant distortion within both reciprocals. In 4 of these cases LB homozygotes had
higher viability than RP homozygotes on both cytoplasmic backgrounds and in 1 case RP
homozygotes had higher viability than LB homozygotes on both backgrounds, indicating

265
Figure A-1. Mean minimum development time (
+
1SE) in reciprocal hybrids, with
significance tested by unpaired, 2-tailed t-tests (***P < 0.01). A) F2 males RPf x LBm
(N = 335) vs. LBf x RPm (N = 178). B) Backcross males (RPf x LBm)F1f x LBm (N =
154) vs. (LBf x RPm)F1f x LBm (N = 125). C) Backcross females (RPf x LBm)F1f x
LBm (N = 166) vs. (LBf x RPm)F1f x LBm (N = 22).

266
Table A-2. Single-locus genotype data for F2 hybrid males in two reciprocal crosses (LB
cytoplasm and RP cytoplasm). Loci are listed by chromosomes number_locus number.
Only LB homozygote frequencies are available for dominant locus 1555B as RP
homozygotes are indistinguishable from LB-RP heterozygotes. χ2 tests compared
genotype numbers to expectations within crosses (3:1 for dominant locus 1555B and
1:2:1 for the remaining codominant loci) and between reciprocal crosses. Loci with
significant differences between reciprocals are shown in bold. For all loci, cases where
the LB homozygote is favored on the foreign cytoplasmic background are marked by
superscript a. For codominant loci, cases where the homozygote class with higher
viability does not match the cytoplasmic background are marked by superscript b. Mean
sample size = 140.8 individuals per locus per cross.

Locus Cross Genotype frequencies Significance of χ
2

LB-LB LB-RP RP-RP Within crosses Between crosses
1_30 LB cytoplasm 0.12
a
0.66 0.22
b
*** ns
RP cytoplasm 0.16
ab
0.69 0.15 ***
1_558 LB cytoplasm 0.16
a
0.64 0.20
b
** ns
RP cytoplasm 0.26
ab
0.56 0.18 ns
2_228 LB cytoplasm 0.37 0.56 0.07 *** ns
RP cytoplasm 0.33
b
0.56 0.11 ***
5_1203 LB cytoplasm 0.35 0.41 0.24 ns *
RP cytoplasm 0.17 0.55 0.28 ns
5_62J8 LB cytoplasm 0.17
a
0.66 0.17 *** ns
RP cytoplasm 0.19
ab
0.63 0.18 **
6_1555B LB cytoplasm 0.24
a
NA NA ns *
RP cytoplasm 0.35
a
NA NA ***
6_cytC LB cytoplasm 0.36
a
0.35 0.29 ns ns
RP cytoplasm 0.40
ab
0.28 0.32 **

267
Table A-2, Continued
7_56J2 LB cytoplasm 0.24
a
0.56 0.20 ns ns
RP cytoplasm 0.30
ab
0.46 0.24 ns
8_61 LB cytoplasm 0.11
a
0.25 0.64
b
*** ns
RP cytoplasm 0.13
a
0.20 0.67 ***
8_480 LB cytoplasm 0.48
a
0.45 0.07 *** *
RP cytoplasm 0.62
ab
0.26 0.12 ***
9_1814 LB cytoplasm 0.22
a
0.56 0.22 ns ns
RP cytoplasm 0.33
ab
0.47 0.20 ns
9_197 LB cytoplasm 0.29
a
0.55 0.16 * ns
RP cytoplasm 0.41
ab
0.45 0.14 ***
10_1555 LB cytoplasm 0.36 0.42 0.22 ** ns
RP cytoplasm 0.29
b
0.46 0.25 ns
11_1202 LB cytoplasm 0.21
a
0.61 0.18 * ***
RP cytoplasm 0.40
ab
0.46 0.14 ***
***P < 0.001, **P < 0.01, *P < 0.05, ns P ≥ 0.05; NA

268
additive superiority of alleles linked to these markers. In one case the homozygote with
higher viability did not match the cytoplasm in both reciprocals and in another case the
favored homozygote did not match the background in one reciprocal and the two
homozygotes were equal in the other reciprocal. In no case did the favored homozygote
class match the cytoplasmic background in both reciprocals.
For all 14 loci (including the dominant locus), there were 4 cases where genotypic
ratios were significantly different between reciprocal crosses. In three of these four cases,
surviving LB homozygotes were more frequent on the foreign cytoplasmic background
than on their native background. Allele frequencies also showed no advantage to alleles
matching the cytoplasm. For the 13 codominant loci, mean LB allele frequencies were
0.52 (0.03 SE) for the cross with LB cytoplasm and 0.54 (0.03 SE) for the cross with RP
cytoplasm. Significant heterozygote excesses (8) were more frequent than significant
heterozygote deficits (5), but the overall heterozygote frequency for the two crosses
(50.3%) was very close to the expected frequency of 50%.
To test for potential nuclear-nuclear interactions affecting survival, two-locus
genotypic combinations were compared to expected numbers determined by multiplying
single-locus ratios. After Bonferroni correction, linkage disequilibrium was significant
only between physically linked loci. However, across all physically unlinked loci when
genotypes were pooled into four different categories (parental double homozygote,
homozygote-heterozygote, heterozygote-heterozygote and nonparental double
homozygote), the cross with LB cytoplasm showed a significant overall excess of
parental homozygotes and a significant overall deficit of nonparental homozygotes
(Figure A-2).

269
Figure A-2. Mean and standard error for proportional deviation from expected two-locus
genotype frequencies ((Obs-Exp)/Exp) for four classes of F2 hybrids in each of two
reciprocal crosses (LB cytoplasm and RP cytoplasm). Only physically, unlinked loci are
included. Asterisks denote the significance of paired, one-tailed t-tests of observed vs.
expected genotype numbers (*P < 0.05). N = 74 two-locus combinations for each cross.

270
Backcross Hybrids
Both male and female backcross hybrids developed significantly faster when the
full haploid nuclear genome did not match the cytoplasmic background (Fig. A-1B,C).
When F1 hybrid females were backcrossed to paternal males ((RPf x LBm)F1f x LBm))
minimum male development time was 18.1
+
0.2d, as compared to 21. 5
+
0.5d in the
maternal backcross ((LBf x RPm)F1f x LBm)). Similarly, minimum female development
time was 28.1
+
0.4d in the paternal backcross, and 32.3
+
1.2d in the paternal backcross.

DISCUSSION
Transmission Ratio Distortion
Distorted ratios were found at 64% of loci tested in each of the two reciprocal F2
hybrids. Previous work on interpopulation hybrids in this species shows skewed
genotypic ratios in adults (Burton 1987; Willett and Burton 2001, 2003; Harrison and
Edmands 2006; Willett 2006; Willett and Berkowitz 2007), but not in newly hatched
larvae (Willett and Burton 2001, Willett 2006, Willett and Berkowitz 2007), suggesting
that these distortions are due to differential zygote viability rather than meiotic drive.
The current study includes the same two populations (but different inbred lines)
and the same microsatellite markers as an earlier study (Harrison and Edmands 2006) in
which the lack of recombination in females (Ar-rushdi 1963; Burton et al. 1981) was
used to create backcross hybrids with intact parental chromosomes. For the 12
microsatellite loci common to both sets of crosses, heterozygote frequencies in male LBf
x RPm backcross hybrids were highly correlated with heterozygote frequencies in both
the LBf x RPm F2 hybrids (r = 0.61, P = 0.034) and the RPf x LBm F2 hybrids (r = 0.79,

271
P = 0.002). This demonstrates that relative heterozygote viabilities for a small number of
loci marking pieces of recombinant chromosomes are a reasonable proxy for relative
heterozygote viabilities of those same loci marking intact chromosomes, suggesting that
at least some of these markers must lie near genes of large effect. The two studies differ
however in that the non-recombinant backcross hybrids showed heterozygote excess
across most loci (mean heterozygosity 59.4%) while mean heterozygosity in the two F2
crosses (50.3%) was much closer to the expected frequency of 50%. The reduced
viability of completely homozygous chromosomes in non-recombinant backcross hybrids
could be caused by both beneficial dominance, and detrimental homozygote-heterozygote
interactions.

Nuclear-Nuclear Interactions
The current study showed some evidence for nuclear-nuclear coadaptation. There
were no examples of linkage disequilibrium between pairs of unlinked loci. However,
when all two-locus genotypes were combined, both reciprocal crosses showed an excess
of parental double homozygotes and a deficit of non-parental double homozygotes, and
both deviations were significant for the LBf x RPm cross. Frequencies of homozygote-
heterozygote and heterozygote-heterozygote genotypes were close to expectations. This
small number of largely non-functional markers therefore reveals the general pattern of
nuclear-nuclear coadaptation expected under conditions of dominance (e.g. Turelli & Orr
2000).
A study by Willett (2006) revealed a very different pattern of nuclear-nuclear
interactions in a different pair of T. californicus populations (AB and SD). In this study

272
F2 hybrids were screened for three unlinked nuclear genes involved in the electron
transport system. In almost all cases, double parental homozygotes, double heterozygotes
and double nonparental homozygotes were all favored, while homozygote-heterozygote
combinations were disfavored. This is certainly not consistent with simple nuclear-
nuclear coadaptation, and suggests complex dominance/epistatic relationships between
these three specific functional genes, or loci closely linked to them.
These complicated patterns of nuclear-nuclear interaction contrast with the more
expected patterns typically reported in other taxa. For example, numerous introgressions
between Drosophila species show nonparental double homozygotes to be more
deleterious than homozygote-heterozygote combinations (Turelli and Orr 2000).
Similarly, crosses between closely related marine bivalve species show that the least fit
genotype was always one or the other nonparental double homozygote (Bierne et al.
2006). Such patterns fit with expectations of the dominance theory of postzygotic
isolation (Turelli and Orr 2000).

Nuclear-Cytoplasmic Interactions
Results show little evidence of nuclear-cytoplasmic coadaptation, and indeed
there are several instances of maladaptation, where nuclear markers fare significantly
worse on their own cytoplasmic background than they do on a highly differentiated,
foreign cytoplasmic background. It should be noted that both the F2 and backcross
studies used a single isofemale line per population, and thus sample only a single
cytoplasmic type. However the level of intra-population mitochondrial variation is so low
in this species (F
ST
for mitochondrial COI is 0.98, Edmands 2001) that a single

273
cytoplasmic type is likely to be a good representative of the whole population. There is
still the possibility that the sampled cytoplasmic types contained deleterious
mitochondrial mutations that altered nuclear-cytoplasmic interactions, but this is
mitigated by the fact that the F2 and backcross studies used different isofemale lines.
For the F2 crosses, codominant loci that were significantly distorted in both
reciprocals showed patterns consistent with additive effects in 5 cases, and maladaptive
interactions in one case. For three out of the four loci showing significant differences
between reciprocal crosses, LB homozygotes had higher viability on the wrong
cytoplasmic background. Development time in the backcross hybrids also provided no
evidence for nuclear-cytoplasmic coadaptation. Offspring of reciprocal F1 hybrid females
backcrossed to the same parental male are expected to have identical nuclear composition
but different cytoplasmic backgrounds. In this study, higher fitness (faster development
time) for both males and females was found in the cross in which the full haploid nuclear
genome did not match the cytoplasmic background ((RPf x LBm)F1f x LBm), in direct
contrast to expectations under cytonuclear coevolution (e.g. Rand et al. 2004). Note that
both the F2 and backcross studies were done under a single set of environmental
conditions (20ºC, normal oceanic salinity, 12h light: 12h dark), and that previous work
has shown fitness of cytonuclear genotypes to be highly dependent on environmental,
particularly temperature, conditions (Willett and Burton 2003). Still, the higher fitness of
mismatched cytonuclear genotypes found in the current study under environmental
parameters well within those experienced by both tested populations is not an expected
result of cytonuclear coevolution.

274
There are at least three potential explanations for faster development in the
mismatched ((RPf x LBm) F1f x LBm) backcross hybrids, given that RPf x LBm F2
hybrids also develop faster than LBf x RPm F2 hybrids. One explanation is that the RP
cytoplasm itself confers faster development time. A second explanation is that RP
grandmothers are phenotypically superior to LB grandmothers, perhaps due to
differences in age or nutritional status, and that these differences extend across
generations (e.g. Hercus and Hoffmann 2000, Magiafoglou and Hoffmann 2003). A third
explanation is that LB nuclear alleles have greater negative interactions with LB
cytoplasm than RP cytoplasm (i.e. maladaptation). There is also the possibility that RP
nuclear alleles have greater negative interactions with LB cytoplasm than RP cytoplasm
(i.e. coadaptation), but such effects should be reduced in the backcross hybrids which are
expected to have only 25% RP nuclear alleles, and yet the difference between reciprocals
is even greater than in the F2 hybrids expected to have 50% RP nuclear alleles. Clean
distinction among potential explanations will require additional crosses, but the current
data on development time certainly do not provide strong evidence for nuclear-
cytoplasmic coadaptation.
Evidence for nuclear-cytoplasmic coadaptation in previous studies of this species
is somewhat equivocal. Much of the work on coadaptation in T. californicus has focused
on nuclear and mitochondrial components of the electron transport system (ETS), with
particular emphasis on interactions between cytochrome-c (CYC, encoded in the nucleus)
and cytochrome-c oxidase (COX, contains both nuclear and mitochondrial encoded
subunits) in ETS complex IV. Cytonuclear hybrids created by repeated backcrossing
exhibit COX activity levels consistent with nuclear-mitochondrial coadaptation in only a

275
subset of interpopulation crosses (Edmands and Burton 1999). In vitro studies show
higher COX activity when CYC and COX are from the same population (Rawson and
Burton 2002; Harrison and Burton 2006), but the highest COX activity occurs with CYC
variants (generated by site-directed mutagenesis) that are a mosaic of amino acids from
populations that do and do not match the cytoplasmic background (Harrison and Burton
2006). In addition to ETS complex IV, Ellison and Burton (2006) also showed that
nuclear-mitochondrial mismatch reduces activity of ETS complexes I and III, as well as
ATP production when a series of interpopulation crosses are pooled together.
Nevertheless, segregation ratios in hybrids frequently show that nuclear alleles are
favored on the wrong cytoplasmic background (Willett and Burton 2001, 2003; Willett
2006). Similarly, a recent study found that in 2 out of 6 interpopulation crosses,
mismatched hybrids (mitochondrial DNA from one population, nuclear-encoded
mitochondrial RNA polymerase from a different population) had significantly greater
capacity to up-regulate mitochondrial genes in response to osmotic stress (Ellison and
Burton 2008a). Finally, strong evidence for nuclear-cytoplasmic coadaptation in
Tigriopus comes from a recent study (Ellison and Burton 2008b) in which the low fitness
of F3 hybrids is restored in maternal backcrosses, which have a full haploid nuclear
genome matching the mitochondrial genome, but not in paternal backcrosses, which have
greater nuclear-mitochondrial mismatch. This is quite different from patterns found in the
present study, and more work is needed to determine if these differences are due to the
fitness components measured (minimum male development time vs. mean fecundity and
survivorship), the specific populations or isofemale lines used, and/or the contrast
between F1 and F3 backcrosses. The effect of the chosen fitness component is a

276
particularly interesting subject for future study. It may be that the 'best' (fastest
developing) mismatched genotype could beat the 'best' matched genotype, even when the
mean mismatched genotype is not superior.

Causes and Consequences of Maladapted Gene Complexes
Interpopulation hybrids in T. californicus exhibit F2 hybrid breakdown for a
broad range of fitness and physiological components (e.g. Burton 1986, 1990ab;
Edmands 1999; Burton et al. 2006), implying that coadaptation predominates. However
further dissection of gene interactions in the present study reveals that parental
populations also harbor a surprising number of maladapted gene complexes, particularly
nuclear-cytoplasmic complexes. Coyne & Orr (2004) noted that “…it is a hard to imagine
that (a gene) would often work better (on a related genetic background) than on its own
background”. While this is true, it is somewhat less difficult to imagine in Tigriopus
where isolated populations experience repeated bottlenecks (Dybdahl 1994; Burton 1997)
likely to impede the efficiency of selection and leave populations stranded on suboptimal
adaptive peaks (sensu Wright 1932). The predominance of drift over selection in this
species is consistent with the limited evidence for local adaptation to salinity and
temperature (Edmands and Deimler 2004; but see also Willett and Burton 2003) as well
as the unusually high F
ST
/Q
ST
ratio (Edmands and Harrison 2003). Accumulation of
maladaptive mitochondrial alleles may be more likely than nuclear, due to the four-fold
lower effective population size for mitochondrial loci (Wright 1969).
The accumulation of maladaptive gene combinations in small populations prone
to genetic drift may create an epistatic load, in addition to the better-understood genetic

277
load of deleterious recessives. Hybridization in such situations may simultaneously create
gene combinations that are both better and worse than parental gene combinations. The
creation of superior hybrid gene combinations in Tigriopus is evidenced by the beneficial
effects of recombination on F2 hybrids (Edmands 2008), and by replicated long-term
hybrid swarms showing an increase in molecular hybridity (Edmands et al. 2005) and
fitness levels surpassing parental controls (Hwang et al., unpub. data). For small
populations suffering from inbreeding depression a current management dilemma is
whether translocation from a genetically and demographically healthier population
should be used to cure the inbreeding depression, or whether this will incur outbreeding
depression (e.g. Tallmon et al. 2005, Edmands 2007). The existence of both beneficial
and detrimental epistasis in small, inbred populations further complicates this issue, but
introduces the hopeful scenario that translocation could in some cases create beneficial
new gene combinations that aid in population recovery.

ACKNOWLEDGMENTS
This work was funded by grants to S.E. from the U.S. National Science
Foundation (DEB-0077940 and DEB-0316807) and the USC Women in Science &
Engineering Program. We gratefully acknowledge J. Curole and M. Voigt for advice on
the ZipCode protocol. The manuscript also benefited from the comments of two
anonymous reviewers and Associate Editor W. Owen McMillan.

278
APPENDIX REFERENCES

Allendorf, F.W., R.F. Leary, P. Spruell, and J.K.Wenburg. 2001. The problem with
hybrids: setting conservation guidelines. Trends Ecol. Evol. 16:613-622.

Ar-Rushdi, A.H. 1963. The cytology of achiasmatic meiosis in the female Tigriopus
(Copepoda). Chromosoma 13: 1754-1763.

Bierne, N., F. Bonhomme, P. Boudry, M. Szulkin, and P. David. 2006. Fitness landscapes
support the dominance theory of post-zygotic isolation in the mussels Mytilus edulis and
M. galloprovincialis. Proc. Roy Soc. B 273:1253-1260.

Burton, R.S. 1985. Mating system of the intertidal copepod Tigriopus californicus. Mar.
Biol. 86:247-252.

Burton, R.S. 1986. Evolutionary consequences of restricted gene flow among natural
populations of the copepod Tigriopus californicus. Bull. Mar. Sci. 39:526-535.

Burton, R.S. 1987. Differentiation and integration of the genome in populations of the
marine copepod Tigriopus californicus. Evolution 41:504-513.

Burton, R.S. 1990a. Hybrid breakdown in physiological response a mechanistic
approach. Evolution 44:1806-1813.

Burton, R.S. 1990b. Hybrid breakdown in development time in the copepod Tigriopus
californicus. Evolution 44:1814-1822,

Burton RS. 1997. Genetic evidence for long-term persistence of marine invertebrate
populations. Evolution 51:993-998.

Burton, R.S., R.J. Byrne, and P.D. Rawson. 2007. Three divergent mitochondrial
genomes from California populations of the copepod Tigriopus californicus. Gene
403:53-59.

Burton, R.S., C.K. Ellison and J.S. Harrison. 2006. The sorry state of F-2 hybrids:
Consequences of rapid mitochondrial DNA evolution in allopatric populations. Am. Nat.
168:S14-S24.

Burton, R.S., M.W. Feldman and S.G. Swisher. 1981. Linkage relationships among five
enzyme-coding gene loci in the copepod Tigriopus californicus: A genetic confirmation
of achiasmatic meiosis. Biochem. Genet. 19: 1237-1245.

Burton, R.S., and B-N. Lee. 1994. Nuclear and mitochondrial gene genealogies and
allozyme polymorphism across a major phylogeographic break in the copepod Tigriopus
californicus. Proc. Nat. Acad. Sci. USA 91:5197-5201.

279
Chen, J., M.A. Iannone, M-S. Li, J.D. Taylor, P. Rivers, A.J. Nelsen, K.A. Slentz-Kesler,
A. Roses, and M.P. Weiner. 2000. A microsphere-based assay for multiplexed single
nucleotide polymorphism analysis using single base chain extension. Genome Research
10:549-557.

Coyne, J.A., and H.A. Orr. 2004. Speciation. Sinauer Associates, Sunderland, MA.

Dybdahl, M.F. 1994. Extinction, recolonization, and the genetic structure of tidepool
copepod populations. Evol. Ecol. 8:113-124.

Edmands, S. 1999. Heterosis and outbreeding depression in interpopulation crosses
spanning a wide range of divergence. Evolution 53: 1757-1765.

Edmands, S. 2001. Phylogeography of the intertidal copepod Tigriopus californicus
reveals substantially reduced population differentiation at northern latitudes. Mol. Ecol.
10:1743-1750.

Edmands, S. 2007. Between a rock and hard place: evaluating the relative risks of
inbreeding and outbreeding for conservation and management. Mole. Ecol. 16:463-475.

Edmands, S. 2008. Recombination in interpopulation hybrids of Tigriopus californicus:
release of beneficial variation despite hybrid breakdown. J. Hered. 99:316-318.

Edmands, S. and J.K. Deimler, 2004. Local adaptation, intrinsic coadaptation and the
effects of environmental stress on interpopulation hybrids in the copepod Tigriopus
californicus. J. Exp. Mar. Biol. Ecol. 303:183-196.

Edmands, S., H.V. Feaman, J.S. Harrison and C.C. Timmerman, 2005. Genetic
consequences of many generations of hybridization between divergent copepod
populations. J. Hered. 96:114-123.

Edmands, S. and J.S. Harrison, 2003. Molecular and quantitative trait variation within
and among populations of the intertidal copepod Tigriopus californicus. Evolution
57:2277-2285.

Egloff, D.A. 1967. Ecological aspects of sex ratio and reproduction in experimental and
field populations of the marine copepod Tigriopus californicus. Ph.D. Dissertation,
Stanford University, Palo Alto, CA.

Ellison, C.K., and R.S. Burton. 2006. Disruption of mitochondrial function in
interpopulation hybrids of Tigriopus californicus. Evolution 60:1382-1391.

Ellison, C.K., and R.S. Burton. 2008a. Genotype-dependent variation of mitochondrial
transcriptional profiles in interpopulation hybrids. Proc. Nat. Acad. Sci. USA 105:15831-
15836.

280
Ellison, C.K., and R.S. Burton. 2008b. Interpopulation hybrid breakdown maps to the
mitochondrial genome. Evolution 62:631-638.

Endler, J.A. 1977. Geographic variation, speciation, and clines. Princeton University
Press, Princeton, NJ.

Ganz, H.H., and R.S. Burton. 1995. Genetic differentiation and reproductive
incompatibility among Baja California populations of the copepod Tigriopus californicus.
Mar. Biol. 123:821-827.

Harrison, J.S. and R.S. Burton. 2006. Tracing coadapted gene complexes to single amino
acid substitutions. Molec. Biol. Evol. 23:559-564.

Harrison, J.S., D.L. Peterson, J.R. Swain and S. Edmands, 2004. Microsatellite DNA
markers for the intertidal copepod Tigriopus californicus. Molec. Ecol. Notes 4:736-738.

Harrison, J.S. and S. Edmands, 2006. Chromosomal basis of viability differences in
Tigriopus californicus interpopulation hybrids. J. Evol. Biol. 19-2040-2051.

Hercus, M.J., and A.A. Hoffmann. 2000. Maternal and grandmaternal age influence
offspring fitness in Drosophila. Proc. Royal Soc. Lond. B 267:2105-2010.

Lee, B.-N. 1993. Genetic structure of Tigriopus californicus populations inferred from
mitochondrial cytochrome oxidase I sequences. Ph.D. dissertation, University of
Houston, TX.

Lewontin, R.C. 1974. The genetic basis of evolutionary change. Columbia University
Press, N.Y.

Magiafoglou, A. and A.A. Hoffmann. 2003. Cross-generation effects due to cold
exposure in Drosophila serrata. Funct. Ecol. 16:664-672.

Manly, K.F., R.H. Cudmore and J.M. Meer. 2001. Map manager QTX, cross-platform
software for genetic mapping. Mammalian Genome 12:930-932.

Mayr, E. 1970. Populations, species, and evolution. Harvard University Press,
Cambridge, MA.

Palmer, C.A., and S. Edmands. 2000. Mate choice in the face of both inbreeding and
outbreeding in the intertidal copepod Tigriopus californicus. Mar. Biol. 136:693-698.

Rand, D.M., R.A. Haney, and A.J. Fry. 2004. Cytonuclear coevolution: the genomics of
cooperation. Trends Ecol. Evol. 19:645-653.

281
Raymond, M. and F. Rousset. 1995. GENEPOP (version 1.2): population genetics
software for exact tests and ecumenicism. J. Hered. 86:248-249.

Rawson, P.D. and R.S. Burton. 2002. Functional coadaptation between cytochrome c and
cytochrome c oxidase within allopatric populations of a marine copepod. Proc. Nat.
Acad. Sci. USA 99:12955-12958.

Turelli, M. and H.A. Orr. 2000. Dominance, epistasis and the genetics of postzygotic
isolation. Genetics 154:1663-1679.

Vittor, B.A. 1971. Effects of the environment on fitness-related life history characters in
Tigriopus californicus. PhD Dissertation. University of Oregon, Eugene, OR.

Wallace, B. 1981. Basic population genetics. Columbia University Press, NY.

Wallace, B. 1991. Coadaptation revisited. J. Heredity 82:89-95.

Willett, C.S. 2006. Deleterious epistatic interactions between electron transport system
protein-coding loci in the copepod Tigriopus californicus. Genetics 173:1465-1477.

Willett, C.S. and J.N. Berkowitz. 2007. Viability effects and not meiotic drive cause
dramatic departures from Mendelian inheritance for malic enzyme in hybrids of
Tigriopus californicus populations. J. Evol. Biol. 20:1196-1205.

Willett, C.S. and R.S. Burton. 2001. Viability of cytochrome c genotypes depends on
cytoplasmic backgrounds in Tigriopus californicus. Evolution 55:1592-1599.

Willett, C.S. and R.S. Burton. 2003. Environmental influences on epistatic interactions:
Viabilities of cytochrome c genotypes in interpopulation crosses. Evolution 57:2286-
2293.

Willett, C.S. and R.S. Burton. 2004. Evolution of interacting proteins in the
mitochondrial electron transport system in a marine copepod. Molec. Biol. Evol. 21:443-
453.

Wright, S. 1932. The role of mutation, inbreeding, crossbreeding and selection in
evolution. Proceedings of the Sixth International Congress on Genetics 1:356-366.

Wright, S. 1969. Evolution and the genetics of populations. Volume 2. The theory of
gene frequencies. Chicago: University Press.

Abstract (if available)

Abstract For the intertidal copepod Tigriopus californicus, outbreeding depression for a variety of fitness measures is typically observed in early-generation interpopulation hybrids. This dissertation is an experimental approach to look at morphological, fitness and molecular outcomes of mixed populations following multiple generations of mating. Each chapter expands upon our understanding of the long-term, multi-generational outcomes of hybrid swarms that were initiated with populations showing different degrees of incompatibility.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

The evolution of pollution tolerance in the marine copepod Tigriopus

PDF

Phylogeography, reproductive isolation, and the evolution of sex determination mechanisms in the copepod Tigriopus californicus

PDF

Application of evolutionary theory and methods to aquatic ecotoxicology

Asset Metadata

Creator Hwang, AnnMarie S. (author)

Core Title Multiple generations of hybridization between populations of the intertidal copepod Tigriopus californicus

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Biology

Publication Date 07/30/2009

Defense Date 05/12/2009

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag fitness,hybrid breakdown,interpopulation hybridization,microsatellite loci,multiple generation,OAI-PMH Harvest,outbreeding depression

Place Name Baja California (states), California (states), Mexico (countries), San Diego (city or populated place)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Edmands, Suzanne (committee chair), Chuong, Cheng-Ming (committee member), Gracey, Andrew (committee member), Hedgecock, Dennis (committee member), Nordborg, Magnus (committee member)

Creator Email achinen@usc.edu,annmariehwang@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-m2423

Unique identifier UC1494234

Identifier etd-Hwang-3012 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-569405 (legacy record id),usctheses-m2423 (legacy record id)

Legacy Identifier etd-Hwang-3012.pdf

Dmrecord 569405

Document Type Dissertation

Rights Hwang, AnnMarie S.

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Repository Name Libraries, University of Southern California

Repository Location Los Angeles, California

Repository Email cisadmin@lib.usc.edu