PLACE IN RETURN BOX to remov
To AVOID FINES return

MAY BE RECALLED with earlier

e this checkout from your record.
on or before date due.
due date if requeSted.

 

DATE DUE

DATE DUE

DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6m cJCIRCJDatoOuo.p65—p.15

ASSESSMENT OF HISTORICAL AND CONTEMPORARY GENETIC DIVERSITY
OF STEELHEAD (ONCORH YNCH US MYKISS) IN THE LAKE MICHIGAN BASIN

By

Meredith Lynn Bartron

A DISSERTATION

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

DOCTOR OF PHILOSOPHY

Department of Fisheries and Wildlife
Ecology, Evolutionary Biology, and Behavior Program

2003

ABSTRACT

ASSESSMENT OF HISTORICAL AND CONTEMPORARY GENETIC DIVERSITY
OF STEELHEAD (ONCORHYNCH US MYKISS) IN THE LAKE MICHIGAN BASIN

By

Meredith Lynn Bartron

Steelhead (Oncorhynchus mykiss) were introduced into the Great Lakes in the late
1800’s. Subsequently, natural recruitment across the Lake Michigan basin has been
regularly supplemented by hatchery production of strains derived from widely dispersed
locales within the species’ native range along the west coast of the United States. We
used microsatellite markers to assess how the populations of steelhead in Lake Michigan
are genetically structured 1) spatially within and among tributaries to lake Michigan, 2)
temporally based on time of entry into spawning runs (fall versus spring) as well as
between historical and contemporary populations, and 3) between naturalized populations
and hatchery strains used for supplementation. Hierarchical analysis indicated signiﬁcant
genetic differentiation between the naturalized populations and hatchery strains
(950.060, P<0.05), among naturalized populations (®p=0.003, P<0.05), and among
hatchery strains (®p=0.105, P<0.05). However, few signiﬁcant pairwise genetic
differences among naturalized populations exist, no signiﬁcant differences in allele
frequencies between different spawning populations within rivers were observed, and no
signiﬁcant differences in allele frequency between sympatric fall and spring spawning
runs were observed.

Prior to 1983, hatchery supplementation of Lake Michigan steelhead populations

in Michigan utilized primarily on strain and was largely unsuccessful due to low survival

estimates (0.01%) of small (<120mm) hatchery yearlings to smolt stage. Accordingly,
contributions of hatchery ﬁsh to historical adult spawning runs in Michigan tributaries
were low (0-30%) across six major drainages. Large (>150mm) yearlings of multiple
hatchery strains have been stocked exclusively since 1983, increasing estimates of
survival (90%) to smolting. Consequently, the proportion of hatchery adults in spawning
runs increased to 13-79%. We examined the effects of changes in stocking practices on
straying of hatchery steelhead and to temporal changes in levels of genetic diversity and
relationships among populations for steelhead populations sampled for two time periods
(1983-1984 and 1998-1999). Measures of inter-population divergence (mean FST) were
not signiﬁcant for either time period. However, spatial genetic relationships among
historical and contemporary pOpulations were signiﬁcantly correlated with geographic
distance. Increased numbers of alleles in spawning adults from populations can be
attributed to alleles speciﬁc to recently introduced hatchery strains.

The increased contribution of hatchery origin individuals to spawning runs in
Michigan rivers increases the potential for introgressive hybridization between hatchery
and river origin individuals. Therefore, maintenance of genetic diversity within hatchery
strains is important to management. We empirically compared six mating strategies used
by hatcheries in the Great Lakes region to examine the effects of reproductive variance
on measures of genetic diversity. Treatments that minimized reproductive variance by
not pooling gametes from multiple individuals, and did not use individuals repeatedly for
matings resulted in the lowest estimates of coancestry (inbreeding) and highest effective

population size estimates.

ACKNOWLEDGEMENTS

Support for this project was provided by Michigan Department of Natural
Resources Fisheries Division through Federal Aid in Fisheries Restoration Act (Dingle-
Johnson Project F-80-R-3), and Great Lakes Restoration Act through the US. Fish and
Wildlife Service. This research would not have been possible without the assistance of
the Michigan Department of Natural Resources (MDNR), and I would like to particularly
thank the many creel clerks and biologists who provided samples. I am especially
grateful to Paul Seelbach, Jan Sapak, and Jory Jonas of MDNR for their support and
advice regarding this research. Dave Swank and numerous students at the Institute for
Fisheries Research provided invaluable ﬁeld assistance, and additional thanks to Dave for
reading scales. Patrick DeHaan, Kristine Bennett, and Sarah Wycoff aided with lab
work. Emily Szali and Michael Wilberg provided statistical advice. Alexander Topchy
helped with programming. I’m grateful to John Kubisiak, Tom Burzynski, and Brad
Eggold of Wisconsin Department of Natural Resources; John Dettmers of the Illinois
Natural History Survey; and J anel Palla of the Indiana Department of Natural Resources,
and Dr. Gene Rhodes of Purdue University for providing samples. I would also like to
thank Jan Sapak and her staff at the Michigan DNR Little Manistee Weir and Martha
Wolgamood at the MDNR Wolf Lake Hatchery for assistance with the hatchery mating
experiments. I’d like to thank my committee members: Dr. Kim Scribner, Dr. Tom
Coon, Dr. Mike Jones, Dr. Bryan Epperson, and Dr. Ed Rutherford for their support,
advice, and encouragement. My sincere thanks also go to my family and friends for their

continued love and support.

iv

TABLE OF CONTENTS

LIST OF TABLES .................................................................................. vii
LIST OF FIGURES .................................................................................. x
INTRODUCTION .................................................................................... 1
CHAPTER 1
SPATIAL GENETIC. STRUCTURE AMONG LAKE MICHIGAN STEELHEAD
(ONCORH YNCH US MYKISS) POPULATIONS ............................................... 11
Abstract ...................................................................................... 11
Introduction ................................................................................. 12
Methods ...................................................................................... l6
Stocking history ................................................................... 16
Sampling locations and tissue collection ....................................... 17
Scale pattern analysis (SPA) ..................................................... 19
Genetic analysis ................................................................... 19
Statistical analysis ................................................................. 20
Results ....................................................................................... 22
Stocking history ................................................................... 22
Genetic analysis ................................................................... 22
Contributions to genetic diversity ............................................... 25
Discussion ................................................................................... 26
Conclusion .................................................................................. 31
CHAPTER 2
GENETIC EVALUATION OF ALTERNATIVE HATCHERY MATING
STRATEGIES ....................................................................................... 32
Abstract ...................................................................................... 32
Introduction ................................................................................. 33
Methods ................. . .................................................................... 37
Gamete take/Fertilization ......................................................... 37
Experimental treatments ......................................................... 38
Genetic analysis ................................................................... 39
Parentage assignment ............................................................. 40
Statistical analysis ................................................................. 40
Results ....................................................................................... 44
Assessment cf reproductive variation .......................................... 44
Effects of reproductive variation on measures of genetic diversity... ......45
Discussion ................................................................................... 48
CHAPTER 3

TEMPORAL COMPARISONS OF GENETIC DIVERSITY IN LAKE MICHIGAN
STEELHEAD (ONCORH IWCH US MYKISS) POPULATIONS: EFFECTS OF

HATCHERY SUPPLEMENTATION ............................................................ 52
Abstract ...................................................................................... 52
Introduction ................................................................................. 53
Methods ...................................................................................... 56

Stocking history ................................................................... 56
Sample collection .................................................................. 58
Scale pattern analysis (SPA) ..................................................... 59
Genetic analysis ................................................................... 60
Statistical analysis ................................................................. 61
Results ....................................................................................... 62
Hatchery contributions to spawning runs ...................................... 62
Genetic analysis ................................................................... 63
Discussion ................................................................................... 66
CHAPTER 4

METHODOLOGICAL BIAS IN ESTIMATES OF STRAIN COMPOSITION AND
STRAYING OF HATCHERY-PRODUCED STEELHEAD IN LAKE MICHIGAN

TRIBUTARIES ...................................................................................... 70
Abstract ...................................................................................... 70
Introduction ................................................................................. 71
Methods ...................................................................................... 74

Stocking history ................................................................... 74
Sampling locations and tissue collection ....................................... 76
Scale pattern analysis (SPA) ..................................................... 76
Genetic analysis ................................................................... 77
Statistical analysis ................................................................. 78
Determination of hatchery contribution ........................................ 79
Results ....................................................................................... 81
Stocking history ................................................................... 81
Genetic identiﬁcation of hatchery strains ...................................... 81
Estimates of strain-speciﬁc contribution based on observation ............. 82

Estimates of strain-speciﬁc contribution based on genotype and SPA. ....82
Combined direct and indirect estimates of strain-speciﬁc contribution. . .83
Straying of hatchery strains into Michigan rivers ............................. 84
Discussion ................................................................................... 86

APPENDIX I

SALdPLE SIZES (N) AND ALLELE FREQUENCIES AT EACH OF THE SIX
MICROSATELLIT E LOCI FOR SIX MICHIGAN STEELHEAD POPULATIONS
SAMPLED DURING EACH TIME PERIOD (1983-4 AND 1998-9), AND EACH
HATCHERY STRAIN. HATCHERY INDIVIDUALS WERE EXCLUDED FROM
RIVER SAMPLES ................................................................................. 91

APPENDIX II

vi

TABLES AND FIGURES ......................................................................... 94

LITERATURE CITED ........................................................................... 126

vii

LIST OF TABLES

Table l. Stocking histories for each of the ten naturalized populations examined.
Numbers of individuals stocked represents the average number of juvenile steelhead
stocked into each river between 1993-1997. The estimates of smolt-equivalents stocked
are based on survival estimates for different size, age, and location of stocking described
by Rand et al. (1993). Strains used for stocking were Michigan strain (M) and Skamania
strain (S) .............................................................................................. 95

Table 2. Sample size (N), number of alleles per locus (A), alleleic richness (Ar), observed
heterozygosity (Ho), and expected heterozygosity (He) for each population by locus and
mean values over all loci. Additionally, mean F (inbreeding coefﬁcient) is provided over
all loci. Because the Little Manistee population is used for propagation of the Michigan
strain, it is included in the hatchery population group. Natural populations are deﬁned as
follows: MN (Manistee River), PM (Pere Marquette River), PL (Platte River), MU
(Muskegon River), BE (Betsie River), WH (White River), SJ (St. Joseph River), BL
(Black River), and TH (Thompson Creek). Hatchery strains are deﬁned as follows: LM
(Little Manistee River/Michigan strain), SK (Skamania strain), CC (Chambers Creek
strain), and GA (Ganaraska strain) ............................................................... 96

Table 3. Hierarchical partitioning of genetic variation (Weir 1996) within and among the
naturalized populations and hatchery strains is provided for each locus and mean over
seven loci. Naturalized populations include the Manistee River, Pere Marquette River,
Platte River, Muskegon River, Betsie River, White River, St. Joseph River, Black River,
and Thompson Creek. Hatchery strains include the Little Manistee River/Michigan
strain, Skarnania strain, Chambers Creek strain, and Ganaraska strain .................... 99

Table 4. Pairwise comparisons of inter-population variance in allele frequency (mean 9
across seven loci). Signiﬁcant (P<0.05) comparisons are italicized. Mean pairwise 6)
among naturalized populations (including the Little Manistee River) is 0.003. Mean
pairwise 6 among the hatchery strains (including the Little Manistee River) is 0.095..101

Table 5. Estimates of the mean and variance number of offspring produced by males and
females for each mating treatment. The number of individual males and females (N)
used for each treatment is also provided ....................................................... 102

Table 6. Total number of males and females used for matings (N), and summary
measures of genetic diversity including the percent unrelated offspring, coancestry (O),
and effective population size calculated using the number of males and females (Ne),
reproductive variance (N eV), and coancestry (Nee) for each of the six hatchery mating
strategies used ..................................................................................... 103

Table 7. Number of j uvenile Michigan strain steelhead annually stocked into Michigan
rivers of the Lake Michigan basin, estimates of the number of juveniles that would be

viii

expected to survive to smolt stage, and percentage of the total number stocked based on
estimated survival ................................................................................. 104

Table 8. Estimates of the mean (i 95% CI) number of juvenile steelhead stocked into
Michigan tributaries of Lake Michigan that were expected to survive to smolt (i.e.
number of smolt-equivalents), and the proportion of wild-ori gin adults in the spawning
runs of six rivers in Michigan estimated from two time periods prior to and following
changes in juvenile stocking practices ......................................................... 105

Table 9. Sample sizes (N), number of alleles per locus (A), allelic richness (Ar), and
observed and expected (1-10 and IL) heterozygosity for each steelhead population during
each time period (1983-4 and 1998-9 or 2000), and for each hatchery strain. Hatchery
individuals were excluded from naturalized populations .................................... 106

Table 10. Pairwise estimates of PST between historic (above diagonal) and contemporary
(below diagonal) populations. Pairwise F51» estimates along the diagonal (italicized)
represent comparisons between time periods for the same population. Comparisons were
not statistically signiﬁcant ........................................................................ 108

Table 11. 0. Estimates of the yearly mean number (i SE) of juvenile steelhead stocked
and proportion clipped by each of the four states bordering Lake Michigan between 1993
and 1997. Juvenile year classes are representative of expected returns to Lake Michigan
tributaries during fall 1998 and spring 1999. Numbers represent averages of each strain
stocked into Lake Michigan, based on the actual numbers stocked (total), and numbers
adjusted for differential size— and age-speciﬁc survival (Rand et a1. 1993). b. Average
number of steelhead of each hatchery strain stocked into each of four Michigan rivers
between 1993 and 1997 ........................................................................... 109

Table 12. Pairwise estimates of inter-strain variance in allele frequency (mean F ST over
seven rrricrosatellite loci) for four hatchery strains of steelhead stocked into Lake
Michigan ............................................................................................ 111

Table 13. Estimates of classiﬁcation accuracy for likelihood-based assignment tests
(Blanchong et a1. 2002). Percentages on the diagonal are the proportion of individuals of
that strain correctly re-classiﬁed to strain of origin based on seven microsatellite loci..112

Table 14. Numbers and percentage of steelhead from each hatchery strain identiﬁed
based on ﬁn clip observations and using a combination of genetic analysis and scale
pattern analysis (SPA) to identify unclipped hatchery-produced steelhead. Strains
include M (Michigan), Sk (Skamania), G (Ganaraska), and C (Chambers Creek) ....... 113

ix

LIST OF FIGURES

Figure 1. Depiction of inbreeding and outbreeding on offspring ﬁtness depending on
relatedness of parents .............................................................................. 1 14

Figure 2. A map representing all naturalized populations and hatchery strains sampled in
this study. Naturalized populations are numbered as follows: 1) Thompson Creek, 2)
Black River, 3) Platte River, 4) Betsie River, 5) Manistee River, 6) Little Manistee River,
7) Pere Marquette River, 8) White River, 9) Muskegon River, and 10) St. Joseph River.
Hatchery strains sampled are denoted as follows: C) Chambers Creek strain, G)
Ganaraska strain, S) Skamania strain, M) Michigan strain .................................. 115

Figure 3. This map depicts locations sampled within the Manistee River, Pere Marquette
River, and Muskegon River to determine whether genetically unique populations existed
within river drainages. Populations sampled within the Manistee River were 1) Bear
Creek, and 2) the mainstem of the Manistee River. Within the Pere Marquette River,
populations sampled were 3) Baldwin River, 4) Middle Branch, 5) Little South Branch,
and 6) the Pere Marquette River mainstem. Populations sampled within the Muskegon
River were 7) Bigelow Creek and 8) the Muskegon River mainstem ..................... 116

Figure 4. Estimated contributions of each population to total diversity observed based on
the diversity and divergence contributed by each population. The upper graph represents
the unbiased contribution to total diversity of each p0pulation, and the lower graph
represents the contribution to total diversity by each population incorporating allelic
richness ............................................................................................. 1 17

Figure 5. Diagrams of the six mating treatments used. Each circle represents the
combination of gametes from representative males and females. Sex ratios are listed
following the number of female to the number of males. Each treatment was repeated
following the prescribed mating strategy until all males and females allotted for that
treatment were used. Equal volumes of milt and eggs for each male and female were
used for each individual for each treatment .................................................... 1 18

Figure 6. Number of offspring produced by each male and female in each treatment. . . l 19

Figure 7. Predicted generational changes in population of coancestry (O) for populations
maintained using one of the six mating strategies over successive generations.
Coancestry values of 0.25 represent the relationship between full siblings ............... 121

Figure 8. Demonstration of how the effective population size for captive populations
maintained using six different mating can impact the total effective size of two wild
populations of different size and varying contributions of the captive population,
following Ryman & Laikre (1991) .............................................................. 122

Figure 9. Geographic locations where historical and contemporary populations of
steelhead were sampled from Michigan tributaries of Lake Michigan. Samples were

obtained from throughout the river drainages. Numbers on maps corresponding to river
names are as follows: 1) Betsie River, 2) Manistee River, 3) Little Manistee River, 4)
Pere Marquette River, 5) White River, 6) Muskegon River. The bars that intersect both
the Manistee River and Muskegon River correspond to dams that prevent upstream
migration of adult steelhead. Sampling locations for these rivers were distributed
throughout the portion of the rivers downstream from the dams ............................ 123

Figure 10. Unrooted neighbor-joining trees based on Cavalli-Sforza & Edwards (1967)
chord distance, demonstrating the genetic relationships among each of the six populations
for two time periods (a) 1983-1984 and (b) 1998-2000. Bootstrap values at nodes
represent the percentage of 500 trees where the populations past the nodes were grouped
together ............................................................................................. 124

Figure 11. Geographic locations where populations of steelhead were sampled from
Michigan tributaries of Lake Michigan. Numbers on maps corresponding to river names
are as follows: 1) Platte River, 2) Bear Creek (tributary to the Manistee River), 3) Little
Manistee River, and 4) Pere Marquette River. The bar that intersects the Manistee River
corresponds to Tippy Darn, which prevents upstream migration of adult steelhead.
Sampling locations for these rivers were distributed throughout the portion of the rivers
downstream from the dams. Hatchery strains sampled are denoted as follows: C)
Chambers Creek strain, G) Ganaraska strain, S) Skamania strain, M) Michigan strain..125

xi

INTRODUCTION

Steelhead (Oncorhynchus mykiss) are native to the west coast of North America.
Steelhead are anadromous, but differ from other ocean-mi grating Paciﬁc salmon due to
their ability to spawn multiple times, whereas other Oncorhynchus species are
semelparous. Steelhead home to their natal stream, though low levels of straying occurs
(Quinn 1993). High levels of natal homing (Quinn 1993) and local adaptation (Taylor
1991) have contributed to the spatial and temporal patterns of genetic variation within
and among steelhead populations in their native range (Busby et a1. 1996).

Steelhead populations or stocks are often deﬁned by drainages in which they
spawn (Busby et a1. 1996). For larger river systems, multiple genetically distinct
populations may be present (Beacham et a1. 1999; Heath et a1. 2002). Typically,
populations spawning in close proximity are more similar genetically than more

geographically distant populations (Busby et a1. 1996).

Steelhead in Lake Michigan

Steelhead were initially introduced to the Great Lakes in the late 1800’s
(MacCrimmon and Gots 1972). Within Lake Michigan, spawning primarily occurs in
Michigan tributaries (Seelbach and Whelan 1988). Due to the lack of suitable spawning
habitat, natural reproduction contributing to natural recruitment in other areas is believed
to be negligible (Seelbach and Whelan 1987). Historically, the state of Michigan stocked
juvenile steelhead as fall ﬁngerlings (<110 mm; Seelbach 1987) with an estimated
survival to smolt stage of 0.01% (Rand et a1. 1993). Following a change in stocking

practices in 1984, large yearlings (>150 mm; Seelbach 1987a) with higher estimated

survival to smolt stage (90%; Rand et a1. 1993) were used for stocking. Prior to this
change, average contributions of hatchery ﬁsh to spawning runs in six Michigan rivers
was generally low (12%; Seelbach & Whelan 1988). Following increases in size at
stocking, the average contribution of hatchery ﬁsh to spawning runs increased (40%;
Bartron & Scribner, in press).

Approximately 1.8 million juvenile steelhead are stocked into the Lake Michigan
basin annually. Four hatchery strains, including Skamania, Michigan, Chambers Creek,
and Ganaraska are currently stocked into the basin. The Skamania strain originated from
the Washougal River in southwest Washington state. The Skamania strain was
introduced into the Great Lakes in 1975 by the Indiana Department of Natural Resources.
The Skamania strain is maintained by annual gamete takes from the St. Joseph River to
propagate the strain stocked by Indiana, and through separate broodstock rivers in
Wisconsin. The Chambers Creek strain originated ﬁom the Puget Sound region of
Washington State, was introduced to Lake Michigan in 1987, and is stocked by
Wisconsin. The Ganaraska strain (also stocked by Wisconsin) originated from
naturalized reproduction in the Ganaraska River, a tributary to Lake Ontario, and was
also introduced to Lake Michigan in 1987. Both the Ganaraska and Chambers Creek
hatchery strains are maintained by annual gamete takes from broodstock rivers in
Wisconsin. Michigan and Indiana stock the Michigan steelhead strain. The Michigan
strain is maintained from gametes taken annually from a naturalized population in the
Little Manistee River, a tributary to Lake Michigan.

Ecological, morphological or genetic differences among steelhead strains have

emerged over long periods of time in their native environments. Population differences

in these strains are of adaptive signiﬁcance, and are maintained in nature by the balance
between evolutionary forces of selection, genetic driﬁ, and gene ﬂow. The four strains of
steelhead used for supplementation in waters of the Lake Michigan basin exploit strain-
speciﬁc heritable differences in the timing of adult returns to rivers for spawning to
increase recreational ﬁshing opportunities for anglers. Additional strains of non-
anadromous rainbow trout and other non-native salmonid species have and are continuing
to be stocked widely throughout the basin. Stocking oﬁen is conducted without directed
efforts to assess whether numbers stocked are justiﬁed relative to returns to the ﬁshery.
There has been little effort to assess the proportional contributions of hatchery ﬁsh to the
spawning runs in the major tributaries of Lake Michigan or of the potential impacts
hatchery ﬁsh may have on naturalized populations either competitively or via
introgressive hybridization. Intro gression between individuals from different hatchery
strains will reduce between-strain differences in allele frequency, although measures of
genetic diversity within strains (or populations) will increase. Of concern to management
is the decrease in between-strain genetic differences, and the potential for the loss of the
heritable differences in run-timing among strains. This thesis research focuses explicitly
on these questions.

There is widespread confusion among ﬁshery managers as to how management
actions related to stocking can compromise the genetic integrity of naturalized
populations. One widely held perception is that because the stock or strain composition
of steelhead and most other introduced species is highly “contrived” or artiﬁcial relative
to areas from the species native range, ﬁsheries managers need not worry about levels of

genetic variation or of introgression. Is the management goal to manage the resource for

sustainable yields for the public? Year to year variation in the number and strain
composition of ﬁshes stocked is widely believed to leave little opportunity for ﬁshes to
adapt (or evolve) within the environment. If there is no adaptation and survival fails to
meet management prescriptions then managers can simply restock with a different strain.
Using genetic markers we seek to provide empirical data that demonstrates that
management actions can affect the genetic characteristics of populations within a highly
manipulated lake ecosystem.

There is much confusion in ﬁsheries management as to what levels of genetic
variation within and among populations are optimal. If inbreeding is to be discouraged
due to the negative effects of expression of deleterious recessive alleles in homozygous
condition in offspring (and lower ﬁtness; Figure 1), then should we not be encouraging
outbreeding? Would not introgression of genetically different ﬁshes (e. g., from different
hatchery strains or between hatchery and wild ﬁsh) result in hybrid vigor in progeny?
However, introgression between genetically different ﬁshes could also potentially
negatively reduce the ﬁtness or progeny. Outbreeding depression is the decrease in
ﬁtness of offspring resulting from matings between highly genetically differentiated
individuals (Figure 1).

By quantifying how genetic diversity is partitioned within and among hatchery
strains and naturalized populations of steelhead in the Lake Michigan basin, we provide
recommendations that minimized threats of inbreeding and outbreeding given existing

constrains imposed by management.

Stocking strategies: inbreeding versus outbreeding

Two main concerns exist regarding genetic diversity within and among steelhead
populations in Lake Michigan. The ﬁrst concern relates to inbreeding depression, and the
associated decreased ﬁtness of offspring resulting from matings among highly related
individuals (Figure 1). In general, high levels of alleleic diversity and heterozygosity
within most of the hatchery strains and naturalized populations do not indicate inbreeding
depression is currently of concern to steelhead in Lake Michigan.

Due to the highly signiﬁcant genetic differences among hatchery strains and
between hatchery strains and naturalized populations, outbreeding depression and
associated potential for decrease in ﬁtness of offspring resulting from matings between
genetically distinct populations should be a concern for managers. Hatchery gamete-take
practices that use large numbers of adult male and female individuals for spawning,
rrrinimize re-use of adults for matings, and do not pool gametes from more than two adult
males (to minimize sperm competition) act to minimize coancestry (inbreeding) among

resulting offspring.

Spatial comparison of genetic diversity

In this study, we used molecular microsatellite markers to quantify levels of
genetic diversity within and the magnitude of variation among steelhead populations
across the Lake Michigan basin. We found that the major component of genetic diversity
among Lake Michigan steelhead populations was attributed to differences among the

hatchery strains (mean F51=0.095), whereas genetic differentiation among naturalized

populations was low (mean Fs1=0.003). We found no evidence for signiﬁcant
differences in allele frequency between steelhead spawning in different locations within
rivers. Additionally, we found no evidence for signiﬁcant differences in allele
frequencies between steelhead sampled from fall and spring-spawning runs within rivers,
though previous studies (Seelbach 1993) documented differences in adult sex ratios

between runs.

Temporal comparison of genetic diversity

Assessments of the magnitude of changes within and inter-relationships among
populations across the Lake Michigan basin would proﬁt from knowledge of historical
benchmarks of genetic characteristics of steelhead prior to changes in hatchery
management. Reisenbichler & Phelps (1989) hypothesized that introgression between
native steelhead and hatchery steelhead widely stocked across large areas led to lower
inter-population variance in allele frequency between steelhead populations in
Washington State. Reisenbichler & Rubin (1999) suggested that although inter-breeding
may occur between hatchery and wild individuals, the potential reproductive contribution
of hatchery ﬁsh to wild spawning populations could be quite low due to lower
reproductive success of hatchery individuals. Increased survival of juvenile steelhead
stocked into the Lake Michigan watershed can in turn increase the proportion of hatchery
origin adults in spawning runs, and increase the potential for introgression between
steelhead of hatchery and natural origin.

Comparisons of allele frequencies assessed from samples collected during two

time periods allowed assessment of changes in genetic diversity following changes in

stocking practices. Inter-population variance in allele frequency across six naturalized
populations assessed before (1983-1984) and after (1998-1999) changes in stocking
practices decreased (mean F51: 0.006 versus 0.002) respectively. The correlation
between inter-population geographic distance and genetic distance also decreased over
time (for historical populations R=0.55, P<0.01; for contemporary populations R= 0.39,
P<0.02; Bartron & Scribner in press). Additionally, alleles unique to hatchery strains
currently stocked in the basin were not present in naturalized populations sampled prior
to the change in management, but were found in unmarked spawning adults in
contemporary populations. The presence of alleles unique to hatchery steelhead currently
stocked provides evidence that introgression has occurred between steelhead of natural

and hatchery origin.

Straying of hatchery individuals

The four states surrounding Lake Michigan use identifying marks (e. g. ﬁn clips,
maxillae clips, or combinations thereof) to identify hatchery-ori gin steelhead. However,
not all states clip all hatchery steelhead prior to release. Clip data are used to estimate
hatchery strain-speciﬁc rates of straying, and to estimate the contribution of both
hatchery- and naturally-produced steelhead to the recreational ﬁshery in Lake Michigan
and its tributaries. Accurate assessments of straying and relative contribution of hatchery
ﬁsh are likely compromised due to duplication of speciﬁc clips for more than one strain,
and incomplete clipping of hatchery-ori gin steelhead prior to stocking. Further,
abundance of certain hatchery strains are likely overestimated due to confusion of strain-

speciﬁc clips (i.e., maxillae clips) with booking injuries. We observed a bias in the

estimates of contribution of hatchery individuals to spawning runs in Michigan rivers
when estimates were obtained solely based on observation of hatchery clips in
comparison to estimates obtained by scale pattern analysis and genetic identiﬁcation to
strain. Because each method identiﬁes two separate components of the hatchery
contribution (clipped and not clipped), we combined the results from methods to better
estimate the strain-speciﬁc proportion of hatchery individuals in the spawning runs of

Michigan rivers.

Hatchery propagation

A large proportion of realized steelhead recruitment to the creel and to spawning
populations is provided by hatcheries. There is a high probability that hatchery
individuals inter-breed with naturalized individuals. Therefore, it would be prudent to
ensure that hatchery ﬁsh are produced with a goal to maintain genetic diversity. Genetic
evaluation of alternative mating strategies used for propagation of hatchery strains is
important to the long-term maintenance of genetic variation in steelhead across the basin.

Recommendations to improve hatchery management oﬁen focus on effective
population size which is affected by different mating strategies (Kincaid 1995; Allendorf
& Ryman 1987), sex ratios of adults (Simon et a1. 1986), and total numbers of adults
bred. Genetic guidelines have frequently been based on long-standing theory (Wright
1931) that unequal numbers of males and females can lead to reduced effective
population size (N), when deﬁned as the probability that two randomly selected genes in
the current generation are copies of the same parental gene. Though Wright’s

formulation of Nc is an important conceptual basis for management, the use of N, as

described by Wright (1931) does not account for variance in reproductive success among
males and females.

Rigorous quantitative and comparative evaluations of the effects of alternative
hatchery mating strategies on measures of genetic diversity of progeny have not been
conducted. We compared six mating strategies commonly used in the Great Lakes basin
for hatchery propagation. We estimated the magnitude of variation in male and female
reproductive success associated with each mating strategy, and the effects reproductive
variance had on the genetic diversity of resulting progeny. We found that mating
strategies which pooled gametes from multiple individuals resulted in a higher mean
coancestry and lower estimates of effective population size due to disproportional
reproductive contributions of relatively few of the males spawned. Strategies based on
matings of one or two males per female resulted in the highest percentage of unrelated

offspring and highest estimates of effective population size.

Conclusion

Why should genetic diversity be considered in any comprehensive management
program for steelhead in Lake Michigan? Based on the results of this dissertation
research, steelhead in the Lake Michigan basin do not represent a single, panmictic
population. Signiﬁcant differences in allele frequencies exist between hatchery and
naturalized populations. Among naturalized populations, few signiﬁcant differences
exist, potentially due to stocking practices and straying of hatchery individuals. Straying
of hatchery individuals stocked elsewhere in the Lake Michigan basin into Michigan

rivers occurs. The proportional contribution of hatchery individuals to spawning runs has

increased following changes in stocking practices that increased the survival of stocked
juveniles. Due to differences in management practices used for marking hatchery origin
individuals by management agencies around the Lake Michigan basin, identiﬁcation of
hatchery individuals can over- or underestimate the contribution of hatchery individuals
in spawning runs. Additionally, mating strategies used for hatchery propagation
demonstrably effect levels of coancestry, and the probability of increase in inbreeding,
due to variance in male reproductive success most likely resulting ﬁom sperm
competition. Due to the potential for continued introgression between hatchery and
natural origin individuals, it is important to maintain genetic diversity within hatchery

strains.

10

Chapter 1
SPATIAL GENETIC STRUCTURE AMONG LAKE MICHIGAN STEELHEAD
(ONCORH YNCH US MYKISS) POPULATIONS
ABSTRACT
Steelhead were originally introduced to the Great Lakes in the late 1800’s, and

large self-sustaining naturalized populations have existed in Michigan tributaries to Lake
Michigan since introductions began. Despite high levels of natural recruitment,
widespread supplemental stocking of hatchery ﬁsh occurs in almost all major rivers in
Michigan, and throughout the Lake Michigan basin. We seek to learn how the
populations of steelhead in Lake Michigan are genetically structured 1) spatially within
and among Lake Michigan tributaries, 2) based on time of entry into spawning streams
(fall vs. spring), and 3) between naturalized populations and hatchery strains used for
supplementation. Sampling included adults from tributaries from fall and spring
spawning runs, and multiple spawning locations within three of the 10 naturalized
populations. Hierarchical analysis indicated there was signiﬁcant genetic differentiation
between the naturalized populations and hatchery strains (®s=0.060, P<0.05), among
naturalized populations (®p=0.003, P<0.05), and among hatchery strains (®p=0.105,
P<0.05). Analyses of speciﬁc pairwise relationships between populations revealed that
differences in allele ﬁ'equencies are more evident between hatchery strains rather than
naturalized populations. We compare levels of genetic variation for Michigan steelhead
populations to populations of comparable geographic spatial scales to within their native

range.

11

INTRODUCTION

Paciﬁc sahnonids spawn as discrete populations despite multiple year residence
times across wide expanses of open-ocean. Within each species, high natal philopatry to
spawning areas (Scheer 1939; Ricker 1972; Quinn 1993), adaptive differences in life
history characters such as timing of spawning, utilization of different spawning habitats,
and varying residence times in rivers, lakes, and ocean environments have contributed to
reproductive isolation and concomitantly in genetic differentiation even over
rnicrogeographic scales (National Research Council 1996). Adaptations to speciﬁc
stream environments are inﬂuenced to some degree by genetic differences among
individuals and populations (Taylor 1991). Genetic diversity is evident within and
among salmon populations at a variety of spatial and temporal scales (i.e., from within
drainage to large regional levels and among different spawning runs, respectively) (see
review in Allendorf and Waples 1996).

Management of Paciﬁc sahnonids in their native range has historically recognized
regional and population differences in phenotypic, genotypic or ecological characters
(Moulton 1939; Hard et a1. 1996; Busby et al. 1996; Weitkamp et a1. 1995), and strives to
balance conservation (e. g., preservation of stock structure) with commercial and
recreational exploitation (Waples 1991a). Management of naturalized populations of
Paciﬁc salmonids introduced into United States and Canadian waters of the Great Lakes
differs appreciably. While the abundance and distribution of non-native salmonid species
are considered in Great Lakes ﬁsheries community plans and objectives (Eschenroder et

a1. 1995), management focuses on resource utilization by recreational ﬁsheries. Other

12

management issues based on population structure are generally not considered, as
abundance is assumed to be primarily based on hatchery production, even though many
species and populations in the Great Lakes are believed to be self-sustaining (e. g.
Eschenroder et a1. 1995; Seelbach and Whelan 1988).

Steelhead (Oncorhynchus mykiss) are an anadromous salmonid native to the west
coast of North America. Steelhead populations have evolved variations in life history
traits such as differences in mm timing and time of spawning (Leider et a1. 1984; Busby et
a1. 1996). Within their native range, local adaptation associated with stream
environments (Taylor 1991), combined with natal ﬁdelity (Ricker 1972; Schroeder et al.
2001) have contributed to genetic differentiation among steelhead populations at both
levels of macro and micro-geographic spatial scales (Parkinson 1984; Reisenbichler and
Phelps 1989; Reisenbichler et a1. 1992; Beacharn et a1. 1999; Nielsen 1999; Nielsen and
Fountain 1999; Osterberg and Thorgaard 1999). However, in the presence of hatchery
supplementation, stocking and subsequent introgression of hatchery ﬁsh with native
populations has been shown to reduce genetic differences between steelhead populations
(e. g., in the state of Washington; Reisenbichler and Phelps 1989).

Steelhead were the ﬁrst non-indigenous naturalized species purposely'introduced
into the Great Lakes (MacCrimmon and Gots 1972), and have been stocked periodically
in Lake Michigan since initial introductions began in the late 1800’s (Biette et al. 1980;
Fielder 1987). Steelhead used for introductions originated from locations in California
and Washington, though additional sources have been used since initial introduction
(MacCrimmon and Gots 1972). As a result, a large fraction of the genetic diversity that

historically existed in geographically widely distributed spawning populations across the

13

Paciﬁc Coast of North America was introduced into the Great Lakes. Previous genetic
studies of steelhead populations in tributaries of the Great Lakes have shown that
steelhead spawning in geographically widely dispersed populations around Lake Superior
are genetically differentiated (Kruger and May 1987). Adjacent tributaries in Lake
Ontario (Dueck and Danzmann 1996) were found to differ in mitochondrial DNA
haplotypes. Genetic differentiation among steelhead populations in the Great Lakes
could have accrued due to 1) retention of spatial dispersion of genetic variation reﬂecting
different geographic sources of populations initially used for stocking or 2) local
adaptation and genetic drift that occurred subsequent to initial stocking.

Steelhead abundance and distribution across the Great Lakes have been
maintained by natural reproduction and extensive hatchery supplementation (Seelbach
and Whelan 1988). The majority of natural reproduction of steelhead in the Lake
Michigan basin occurs in Michigan drainages due to the lack of favorable spawning
habitat elsewhere in the basin (Seelbach and Whelan 1988). A total of four hatchery
strains are currently stocked into Lake Michigan and its tributaries by the four states
surrounding the basin, representing an average of 1.8 million juvenile steelhead stocked
into Lake Michigan per year since 1993 (Michigan ﬁsh stocking records, Burzynski
1999; Palla 1999; Table 1). These strains have been used for supplementation to provide
a variety of angling opportunities taking advantage of strain-speciﬁc heritable differences
in the timing of return of adults to rivers to spawn.

Within the state of Michigan, two hatchery strains (Michigan strain and the
Skamania strain) are currently stocked (Table 2). The Michigan strain originates from

the naturalized population in the Little Manistee River in Michigan. The Skamania strain

14

originated from the Washougal River in southwest Washington State. This strain was
introduced into the Great Lakes in 1975 by the Indiana Department of Natural Resources,
and in Indiana is maintained by annual gamete takes from adults of hatchery origin
returning to the St. Joseph River. Wisconsin also maintains a separate Skamania strain
which was obtained from Indiana. Additional strains stocked into the Lake Michigan
basin include the Ganaraska strain (stocked by Wisconsin), which originated from
naturalized reproduction in the Ganaraska River, a tributary to Lake Ontario, and the
Chambers Creek strain (also stocked by Wisconsin), which originated from the Puget
Sound region of Washington State. Both the Ganaraska and Chambers Creek hatchery
strains are maintained by annual gamete takes from broodstock rivers in Wisconsin.

Recent changes (mid 1980’s) in hatchery stocking practices resulted in increased
survival of hatchery-raised juveniles (Seelbach 1987a) and higher return rates of hatchery
steelhead adults (Seelbach and Miller 1993). Consequently, the proportion of hatchery
individuals to the steelhead spawning runs in Michigan signiﬁcantly increased (Bartron
and Scribner in press). Increased contribution of individuals of hatchery origin to
spawning runs has resulted in introgression between steelhead of natural and hatchery
origin (Bartron et al. in press). Although maintaining genetic differences among
naturalized steelhead populations has not been stated as a management goal, the
homogenization of gene frequencies among populations may act to effectively negate any
adaptive evolution that may have evolved during the species approximate 120 year tenure
in the basin (Bartron and Scribner, in press).

Our objective was to determine how genetic variation is partitioned within and

among Lake Michigan tributaries. To assess how genetic variance was partitioned within

15

and among naturalized populations and hatchery strains, we sampled ten naturalized
steelhead populations in Michigan, which represent the majority of the natural
reproduction contributing to the Lake Michigan steelhead populations. Within larger
drainages supporting natural reproduction, multiple tributaries were sampled to determine
if signiﬁcant differences in allele frequency were apparent among discrete spawning
areas in different tributaries within rivers. Additionally, for a subset of rivers,
comparisons were made between adults sampled during the fall and spring spawning runs
to determine whether time of entry into spawning streams conferred signiﬁcant levels of
reproductive isolation. The null hypothesis was that no spatial heterogeneity in gene
frequencies exist across populations of steelhead spawning across Michigan tributaries of
Lake Michigan. Results will be tied directly to extensive current and historical data on
stocking intensity, distribution, and success. Findings will be compared to known levels

of spatial and temporal genetic diversity within the species native range

METHODS
Stocking history
Stocking records for Michigan rivers were obtained for the years between 1993
through 1997, representing the time period that steelhead returning to rivers in 1998-1999
and the spring of 2000 would have been stocked. Speciﬁcally, information was collected
regarding the hatchery strain stocked, the number of individuals stocked, size and age of
the hatchery juveniles stocked, and location of stocking. Stocking records were obtained

from http://www.michigangov/dnr (Michigan Department of Natural Resources).

16

Sampling Locations and Tissue Collection

Samples (total N=667) were collected from adult steelhead from ten rivers in
Michigan, representing the majority of the naturalized steelhead populations in the Lake
Michigan basin (Figure 1). Rivers sampled for adult steelhead included Thompson
Creek, Black River, Platte River, Betsie River, Manistee River, Little Manistee River,
Pere Marquette River, White River, Muskegon River, and St. Joseph River. Sampling
occurred multiple times throughout the spawning runs.

To determine if fall and spring-run steelhead collected from the same stream
represented genetically different populations, samples were obtained from the fall and
spring spawning runs of the Little Manistee River (fall N=57; spring N=60), Manistee
River (fall N=52), Pere Marquette River (fall N=29), Platte River (fall N=50; spring
N=55), and Muskegon River (fall N=37). Individuals from the fall run were collected
from the mainstem of each river through creel surveys during September 1998 to
December 1998. Individuals from the spring run were collected by electroshocking and
creel surveys during March 1999 to May 1999. Spring-run steelhead were sampled on
spawning areas, identiﬁed by the presence of gravel substrate, spawning redds, and
observations of spawning behavior.

To examine the partitioning of variance in allele frequency within drainages, adult
steelhead from multiple spawning locations were sampled during the 1999 spring run in
the Pere Marquette River, the Manistee River, and the Muskegon River (Figure 2).
Tributaries sampled in the Pere Marquette River included the Little South Branch

(N=25), the Middle Branch (N=25), and Baldwin River (N=25; Figure 2). Tributaries to

17

the Pere Marquette River were sampled by electroshocking. Spring-run steelhead from
the Manistee River included the river mainstem (N=55) and Bear Creek (N=50; Figure
2). Sampling was completed by creel survey and electroshocking, respectively. Spring-
run steelhead samples from the mainstem (N=13) of the Muskegon River were collected
by creel survey. Samples from Bigelow Creek (N=25) were obtained by electroshocking.

Steelhead from additional populations naturalized populations in the Lake
Michigan basin were also sampled. The St. Joseph River (N=15) was sampled in the fall
of 1998 by electroshocking. The Black River (N=43), and Thompson Creek (N=12) were
sampled in the spring of 1999 by electroshocking. The Betsie River (N=25) and White
River (N= 17) were sampled in the spring of 2000 by electroshocking.

Adult steelhead from hatchery strains sampled included the Michigan strain
(obtained from the Little Manistee River), Skamania strain, Chambers Creek, and
Ganaraska strain. Wisconsin maintains and stocks steelhead ﬁom the Skamania strain
(N=52), Chambers Creek strain (N=60), and Ganaraska strain (N=60), and provided ﬁn
clips from individuals of each strain from the spring spawning run of 1998 (Figure 1).
Indiana maintains steelhead from the Skamania strain (N=60), which was sampled in
1998. Because the Wisconsin and Indiana Skamania strains have been managed
separately (each state collects broodstock from different rivers), samples were obtained
for each states collection.

Small ﬁn clips from the upper lobe of the caudal ﬁn were taken from each
individual and stored either in urea storage buffer (4M Urea, 2M NaCl, 0.1M Tris-HCl,
0.5% Sarcosine, lOmM EDTA), 95% ethanol, or were dried and stored in scale

envelopes. Buffer and ethanol preserved ﬁn clips were stored at -20° C until DNA

18

extraction. Samples were taken only from steelhead that did not have hatchery strain-

speciﬁc marks.

Scale pattern analysis (SPA)

Scale samples were also taken from each individual sampled for genetic analysis
to determine river or hatchery origin. Origin was determined by scale grth pattern
analysis (ratio 23; Seelbach and Whelan, 1988). Ratio 23 analyzes the differences in
grth between the ﬁve circuli preceding the ﬁrst annulus and the ﬁve circuli following
the ﬁrst annulus (Seelbach and Whelan 1988). Individual steelhead identiﬁed as hatchery
origin by scale pattern analysis (SPA) were removed prior to analysis [see Bartron et al.

(in review) for further discussion of the SPA technique].

Genetic analysis

DNA was extracted using the PurGene® extraction kit (Gentra Systems) and
Qiagen DNeasy® Tissue Kit (Qiagen Inc.) DNA concentrations were determined using
ﬂuorimetery and working stocks of 20ng/ul were made for each sample. Seven
microsatellite loci were used to estimate allele frequencies and population levels of
genetic variability. Loci included Ogola and Ogo4 (Olsen et a1. 1998), OneulO and
Oneull (Scribner et a1. 1996), Omy77 (Morris et a1. 1996), Ots103 (Beacham et al.
1998), and Oki200 (Beacham et a1. 1999).

PCR reactions for Ogola, OneulO, Oneul 1, and Omy77 were conducted in 25 ul
reaction volumes using 100 ng DNA, 2 ul 10x PCR Buffer (0.1 M Tris-HCl, ph 8.3,

0.015 M MgClz, 0.5 M KCl, 0.1% gelatin, 0.1% NP-40, 0.1% Trition-X 100), 0.2 mM

19

dNTPs, 0.6 uM ﬂuorescently labeled forward primer, 0.6 uM unlabeled reverse primer,
and 0.3 Units Taq Polymerase. PCR reactions for Ogo4 were conducted in 25 in
volumes, with 2.5 mM MgC12, and primer concentrations were reduced to 0.5 uM. PCR
reactions for Otle3 and Oki200 were conducted in 10 ul volumes using 40 ng DNA, 1
ul 10x PCR Buffer (0.1 M Tris-HCI, pH 8.3, 0.015 M MgClz, 0.5 M KCl, 0.1% gelatin,
0.1% NP-40, 0.1% Triton-X 100), 0.2 mM dNTP’s, 0.4 uM ﬂuorescent labeled forward
primer, 0.4 uM unlabeled reverse primer, and 0.3 units Taq Polymerase. PCR reactions
for all loci utilized an initial denaturing step at 94° C for 2 minutes, followed by 30
cycles of 94° C for 1 minute, annealing temperature for one minute, and extension at 72°
C for 1 minute. A ﬁnal extension period of 2 rrrinutes, 30 seconds was followed by
storage at 6° C until electrophoresis. Annealing temperatures for Ogola, Ogo4, Omy77,
OneulO, Oneul l, Otle3, and Oki200 were 56° C, 54° C, 54° C, 52° C, 62° C, 50° C,
and 50° C respectively. Denaturing acrylamide gels (6%) were used for electrophoresis.
Genotypes were visualized using a Hitachi FM-BIO II scanner and LI—COR® IR2 Global
Edition DNA Sequencer. Molecular weight standards and individuals of known genotype

were run on each gel to standardize scoring.

Statistical analysis

Allele ﬁequencies for each population at each locus and estimates of observed
and expected heterozygosity were generated using the program FSTAT (Goudet 1995;
ver. 2.9.3.2). Exact tests (1000 iterations) for deviations from Hardy-Weinberg
equilibrium (Weir 1990, Guo and Thompson 1992) were conducted using the program

GENEPOP (Raymond and Rousset 1995). Pairwise FST (Weir & Cockerharn 1984)

20

comparisons were used to determine if differences in allele frequency existed among
tributaries within drainages, between fall and spring spawning runs within drainages, and
between naturalized populations and hatchery strains. Estimates of pair-wise and overall
population differentiation were summarized using F-statistics, implemented in the
program FSTAT (Goudet 1995) v.2.9.3.1, and norrrinal alpha levels were adjusted for
multiple comparisons using Bonferroni corrections (Rice 1989). If pairwise comparisons
of F31~ among tributaries within drainage were not signiﬁcant, the samples were pooled.
If pairwise comparisons of F31- between fall and spring spawning runs within drainage
were not signiﬁcant, then the samples were pooled.

Hierarchical analyses of variation in allele frequency were used to determine how
variance was partitioned within and among naturalized populations or hatchery strains.
Variance was examined among alleles within individuals (F), among individuals within
populations (f), among populations within groups (naturalized or hatchery; GP), and
among groups ((9,) were conducted using program Genetic Data Analysis (GDA; Lewis
and Zaykin 2001).

The program CONTRIB (Petit et a1. 1998) was used to calculate allelic diversity
or “richness”, and relative contributions of each population to the total genetic diversity
(Ct). Allelic richness is used to estimate the numbers of alleles for a given sample size
(El Mousadic and Petit 1996), and allows for comparisons between numbers of alleles
between populations when sample sizes differ. Contributions of individual populations to
overall diversity were also useful in understanding spatial patterns of diversity.
Contributions of an individual population to total diversity were described two ways. Ct

was estimated based on an individual populations contribution to the total diversity. C,

21

was estimated based on allelic diversity or richness. Measures of Cr and Ca were
composed of two parts: the contribution of an individual population to total diversity
based on its own diversity, and based on its degree of divergence from the other

populations.

RESULTS

Stocking history

Of the ten naturalized populations examined, only three rivers were not stocked
during 1993-1997 (Table 1). The rivers not supplemented were Thompson Creek, Black
River, and Platte River (Table 1). Of the seven rivers stocked, an average of 44,5 1 5
juvenile steelhead were stocked into each river each year. The smolt-equivalent
adjustments described by Rand et a1. (1993) were used to standardize juvenile numbers to
account for differential juvenile survival based on size, age, and stocking location. An
average of 28,044 smolt-equivalent juvenile steelhead were stocked per year into each of
the seven stocked rivers (Table 1). The Betsie River, Little Manistee River, White River,
and Muskegon River were stocked with individuals from the Michigan strain (Table 1).
The Manistee River, Pere Marquette River, and St. Joseph River were stocked with

individuals from both the Michigan strain and the Skamania strain (Table 1).

Genetic analysis
Estimates of degree of differentiation in allele frequency (pairwise F ST) among
spring-run spawning populations collected from different tributaries sampled within the

Muskegon River, Manistee River, and Pere Marquette River were not signiﬁcant

22

(P>0.05). Samples from different spawning populations within drainages were therefore
combined for further analyses. Estimates of pairwise F ST between adults sampled from
fall and spring spawning runs from the Manistee River, Pere Marquette River, Platte
River, Muskegon River, and Little Manistee River were also not signiﬁcant (P>0.05).
Individuals from different temporal segments of the spawning run were combined within
each drainage for further analyses. Pairwise comparison of mean F51 between samples
from the Indiana Skamania strain and the Wisconsin Skamania strain was not signiﬁcant,
and samples from the two strains were combined.

Deviations from Hardy-Weinberg equilibrium were observed in the Manistee
River, Pere Marquette River, White River, and Chambers Creek hatchery strain.
Deviations were determined by Hardy-Weinberg exact tests (1000 iterations), most likely
result from pooling populations even though no signiﬁcant pairwise differences in allele
frequency were observed. Mean observed heterozygosity over seven loci ranged 0.651 in
the Betsie River to 0.513 in the Skamania strain (Table 2). Inbreeding values (F) were
estimated to vary between -0.073 in the Betsie River to 0.083 in the St. Joseph River
(Table 2). Mean allelic richness over seven loci ranged from 3.90 in the Skamania
hatchery strain to 5.8 in the St. Joseph River (Table 2). The portion of returning Indiana
uses adult steelhead to the St. Joseph River, which ﬂows through Michigan and Indiana
for egg takes to propagate the Skamania strain. The discrepancy between allelic richness
between the naturalized population and the hatchery strain derived from that population
may be due to hatchery spawning practices which utilize a small number of males and

females for mating, and the mating strategy employed to fertilize gametes.

23

Pairwise comparisons of differences in allele frequency between naturalized
populations and hatchery strains were signiﬁcant (P<0.05; Table 3), and overall
difference between the naturalized populations and hatchery strains, accounted for a
signiﬁcant portion of the total variance in allele frequency observed across the Lake
Michigan basin (950.060; P<0.05; Table 2). Pairwise comparisons of F51 between the
Ganaraska strain and each of the naturalized populations with the exception of the St.
Joseph River were signiﬁcant (P<0.05; Table 3). The Michigan strain (source from the
Little Manistee River), which has been widely stocked in Michigan, differed signiﬁcantly
from steelhead sampled in the Manistee River and Black River (P<0.05; Table 3).

Overall, the variance in allele frequencies among populations within hatchery
strains ((9p = 0.105; P<0.05; Table 3) and among naturalized populations was also
statistically signiﬁcant (G)p = 0.003; P<0.05; Table 3). Pairwise comparisons of FST
between each of the hatchery strains were signiﬁcant (P<0.05; Table 3), and the mean F ST
among hatchery strains over seven loci was 0.095. In contrast, pairwise comparisons of
population differentiation indicate few naturalized populations signiﬁcantly differ in
allele frequency from other naturalized populations, and the mean F51 among naturalized
populations over seven loci is 0.003 (Table 3). Only the Black River was signiﬁcantly
different from the Manistee River, Pere Marquette River, and Muskegon River (p<0.05;
Table 3). The Black River was not stocked between 1993-1997 (Table 1), and is
geographically distant from the Manistee River, Pere Marquette River, and Muskegon

River (Figure 1).

24

Contributions to genetic diversity

The two of the four hatchery strains and the St. Joseph River population
contributed the most to the total observed diversity among the populations examined in
this study (Ct; Figure 2). For example, the St. Joseph River had the highest contribution
to total diversity (Ct=0.014; Figure 2), due to the diversity observed within the population
(Cs=0.016; Figure 2). The Skamania hatchery strain also contributed to the total diversity
(Ct=0.013; Figure 2), due to the divergence of the hatchery strain from the other
populations (Cd=0.017). The Skamania strain has been derived from adult steelhead
returning to the St. Joseph River, therefore similarities in their contribution to the total
diversity of steelhead in the Lake Michigan basin is not unexpected. Also, similar
contributions to the total diversity of steelhead in the basin is important in that it shows
there is not overwhelming levels of straying of other ﬁsh into the St. Joseph River. The
naturalized populations contributed proportionally less to total diversity relative to the
hatchery strains. This ﬁnding reﬂects the lack of signiﬁcance of inter-population variance
in allele frequency among the naturalized populations, and signiﬁcant allelic variation
values between the Skamania strain and each naturalized population. The Michigan
strain and the naturalized populations (with the exception of the St. Joseph River) did not
contribute to total diversity (Cr), likely due to the lack of divergence of this strain, which
is widely stocked (Table 1), from other populations (Cd) and similarity in
presence/absence of allele and allele frequency across populations (Cs).

Comparisons of individual population (or strain) contribution to total diversity
were based on the allelic richness of each of the populations. Proportional contributions

of Skamania and Chambers Creek hatchery strains to the total diversity were higher than

25

proportional contributions for other populations (Cm: 0.030 and 0.033 respectively; _
Figure 2), due primarily to their divergence (presence of alleles not shared with other
populations) from the other populations. Most of the allelic diversity is apportioned

among hatchery strains.

DISCUSSION

Steelhead have been present in the Great Lakes for approximately 100 years, or
36 generations. Using data from populations in the species (or other sahnonids) native
range as comparisons, this is a sufﬁcient period of time for populations to genetically
differentiate. Given that Hendry et a1. (2000) found that signiﬁcant genetic differences
between two different sockeye (Oncorhynchus nerka) populations in geographically close
proximity had evolved within 56 years (approximately 13 generations), steelhead
populations in Lake Michigan may have had sufﬁcient time for spatial genetic structuring
to become apparent. Another component potentially contributing to differences in allele
frequency among the Lake Michigan basin steelhead populations were the use of
different source populations (MacCrimmon and Gots 1972). Differences in genotypic
diversity among strains used for stocking reﬂecting different phylogeographic origin may
have introduced high amounts of genotypic diversity into the Lake Michigan basin.
Limited survival of hatchery-produced ﬁsh (Seelbach 1987a) may have acted to maintain
genetic population structure that may have evolved since initial introduction, and the
more recent concurrent stocking of four signiﬁcantly genetically differentiated hatchery
strains may have provided an additional inﬂux of genotypic diversity into populations

found in the basin.

26

However, the widespread use of hatchery supplementation within the Lake
Michigan basin and subsequent introgression between hatchery and naturalized
individuals may have reapportioned genetic variation, effectively decreasing variation
among populations due to elevated levels of gene ﬂow (natural straying and purposeful
and directed stocking), while increasing diversity within populations. Reisenbichler and
Phelps (1989) demonstrated that introgression between hatchery and native steelhead in
Washington may have resulted in the homogenization of allele ﬁ'equencies among
populations. Recently, survival of stocked hatchery ﬁsh has greatly increased following
changes in stocking practices by the Michigan Department of Natural Resources in 1984
(Seelbach 1987a). Signiﬁcant increases in the proportion of hatchery individuals to the
spawning runs of naturalized populations in Michigan indicated an increased potential for
introgression between hatchery and naturalized individuals (Bartron and Scribner in
press).

Introgression between hatchery individuals and individuals from Michigan
steelhead populations has been possible through indirect straying of hatchery individuals
returning as adults to spawn or directed stocking events (of j uveniles) in rivers.
Estimated rates of straying to rivers other than their natal river for spawning by adult
steelhead in the Great Lakes range from 3-10% (Biette et a1. 1981). Individuals of
hatchery origin were found in each of the naturalized populations sampled for this study,
including rivers that were not recently stocked (Bartron et al. in review; Bartron
unpublished data). Adult steelhead originating from both Indiana and Wisconsin
hatcheries identiﬁed through both strain-speciﬁc ﬁn clips and genetic assignment to

strain (Bartron et al. in review) have been observed in both the fall and spring spawning

27

runs in Michigan rivers. Therefore, straying of adult hatchery steelhead was not limited
to straying into neighboring river drainages. Rather, straying occurred around the Lake
Michigan basin.

Stocking of hatchery steelhead occurs widely throughout the Lake Michigan
basin. In total, approximately 1.8 million juvenile hatchery steelhead are stocked in the
Lake Michigan basin annually (Bartron et al. in review). Within Michigan, both the
Skamania strain and Michigan strain are stocked. Considering the contribution of the
hatchery strain to spawning runs have increased, the potential for homogenization of gene
frequencies among populations stocked with Michigan strain steelhead resulting from
introgression also increased. Signiﬁcant pairwise differences between each naturalized
population examined and the Skamania, Chambers Creek, and Ganaraska strains (Table
3) indicate relatively little if any intro gression has occurred between the naturalized
populations and thOSe hatchery strains. Bartron and Scribner (in press) demonstrated that
alleles unique to the Skamania, Chamber Creek, and Ganaraska strains were not present
within naturalized populations in Michigan prior to the introduction of those strains to the
Lake Michigan basin. However, alleles unique to the Skamania, Chambers Creek, and
Ganaraska hatchery strains are currently present in most of the naturalized populations
examined in this study (Bartron and Scribner in press), indicate some amount of
introgression has occurred.

Most likely, introgression between naturalized populations and the Michigan
strain has had a greater impact on the partitioning of genetic variation among naturalized
populations. Evidence for introgression between the Michigan strain and naturalized

populations in Michigan includes few signiﬁcant pairwise comparisons of allele

28

frequency among the Michigan strain and naturalized populations (Table 2), and low
contribution to total genetic diversity observed within the basin due to the divergence of
these populations (Figure 3). The only populations representing signiﬁcant pairwise
differentiation in allele frequencies with the Michigan hatchery strain (Little Manistee
River) were the Manistee River and Black River. The Manistee River is heavily stocked
(Table l), but the Black River has not been recently stocked. Additional signiﬁcant
pairwise differences in allele frequency between the Black River and other naturalized
populations support the consideration of the Black River as a genetically distinct
population, whereas no other pairwise comparisons between natural populations and the
Manistee were signiﬁcant. Therefore, in the absence of stocking, and relative geographic
isolation, genetic differentiation of naturalized populations in the Lake Michigan basin is
possible.

Levels of population differentiation for steelhead in their native range on the West
Coast of North America provide insightful comparisons for Lake Michigan steelhead
populations. Lack of signiﬁcant differentiation between sympatric within river fall and
spring-spawning runs of the naturalized populations examined in this study was
consistent with comparisons of winter and summer-run steelhead in their native range
(Utter and Allendorf 1977; Chilcote et al. 1980; Leider et al. 1984). Although sex ratios
of fall and spring run steelhead in Michigan differ (Seelbach and Whelan 1988),
signiﬁcant differences in allele ﬁ'equencies were not observed (Table 3). Estimates of the
mean population differ in allele frequency among between naturalized populations,
including the Little Manistee River population, were quite low (mean F51: 0.003; Table

3), compared to mean F51 values between geographically proximate populations of

29

steelhead in their native range along the West Coast of North America. Between
steelhead populations in northern and southern California, Nielsen (1999) found mean
Fs1=0.064. Mean FST (over seven loci) among tributaries within the Skeena River and
Nass River in British Columbia, Canada were 0.026 and 0.024 respectively (Beacham et
a1. 2000). However, Beacharn et a1. (1999) did not ﬁnd signiﬁcant differentiation among
neighboring populations within the same drainage of the Thompson River in British
Columbia, Canada.

The mean estimate of the degree of population differentiation among the hatchery
strains stocked into the Lake Michigan basin is quite high (F5T=0.095; Table 3), greater
than differences between geographically distant populations. Pairwise comparisons of
differences in allele frequencies between steelhead populations in Alaska and California
ranged from 0025-0029 (Nielsen 1999). The observance of such high allelic variance
between the hatchery strains is most likely representative of the differences in
phylogeographic origin of the hatchery strains.

Most of the genetic diversity among Lake Michigan steelhead populations is
partitioned among the hatchery strains. The hatchery strains used for stocking into the .
Lake Michigan basin were chosen to increase angling opportunities due to heritable life
history traits, primarily the timing of the retum of adults to rivers to spawn. Straying of
adults around the basin increases the potential for introgression between strains due to
different management practices by each of the various agencies (Bartron et al. in review).
Introgression between strains would decrease the among-strain differences in genetic

variation, and could change the heritable life history characteristics important to

30

management goals, and also reduce the total genetic diversity observed in the Lake

Michigan basin.

CONCLUSION

Most of the genetic diversity in steelhead in the Lake Michigan basin was
attributed to the differences among the hatchery strains used for supplementation. We
found no evidence for signiﬁcant genetic structuring among spawning populations within
drainages, nor between fall and spring spawning runs within drainages. Levels of genetic
differentiation among drainages in Michigan were greatly reduced in comparison to those
found among drainages in the native range of steelhead. Widespread stocking of
hatchery ﬁsh and introgression with the Michigan strain was likely the major factor
contributing to the low degree of genetic differentiation among naturalized populations of

Lake Michigan steelhead.

31

Chapter 2

GENETIC EVALUATION OF ALTERNATIVE HATCHERY MATING STRATEGIES

ABSTRACT

Hatchery supplementation of natural populations is increasingly used in ﬁsh
conservation efforts to restore or repatriate populations. Diverse mating strategies used
by hatcheries employ different numbers of males and females, varying sex ratios, and
strategies to mix gametes. Due to sex ratio skew and variation in reproductive success,
often the effective number of individuals contributing gametes to subsequent generations
is signiﬁcantly less than the total number spawned. Accordingly, mating strategies affect
levels of genetic diversity, coancestry, and concomitantly long-term ﬁtness of hatchery
stocks or natural populations into which hatchery ﬁsh are placed. Using steelhead
(Oncorhynchus mykiss) we mimicked six commonly employed hatchery mating regimes.
We determined parentage using highly variable microsatellites. We calculated the mean
and variance of reproductive success for each parent used in each mating regime. We
determined summary measures of percentage unrelated offspring, coancestry, and
effective population size for each mating regime that could be used to provide guidelines
more consistent with conservation goals for maintaining gene diversity. We found that
male reproductive variance in numbers of progeny produced was highest when male
gametes were pooled prior to fertilization (387-2333). Mating strategies using single
pair matings (1:1) and those that used gametes from two unique males for each female
(1 :2) resulted in comparatively lower mean coancestry and the percentage of related
offspring while maximizing variance effective population size relative to other gamete

pooling treatments. To examine long-term effects of each mating regime to domestic

32

broodstocks or closed populations, we project estimates of coancestry and effective
population size over time. We show that 1:1 and 1:2 mating regimes are most efﬁcient at
minimizing coancestry and maintaining genetic diversity in hatchery or other closed

populations.

INTRODUCTION

Numerous complex mating strategies have evolved in natural systems to
maximize ﬁtness, maintain genetic diversity, and adaptive potential (DeWoody & Avis
2001). In wild self-sustaining populations, male and female reproductive success varies
greatly due to selection for traits that increase ﬁtness (Fleming & Gross 1994). Size,
fecundity, and other phenotypic, behavioral, or life history characteristics contribute to
inter-individual variance in mating opportunities and reproductive success (Fleming &
Gross 1993; Fleming & Gross 1994).

In artiﬁcial environments, such as ﬁsh hatcheries, selection of mates is imposed
by hatchery personnel and is often a function of availability and synchrony in maturation
of male and female gametes. In addition, selection during hatchery mating and rearing of
juveniles may not necessarily confer adaptive advantage once in the wild (Heggberget et
al. 1993; Fleming et a1. 1996; Negus 1999; Reisenbichler & Rubin 1999; Lynch &
O’Hely 2001; Ford 2002).

Natural populations are increasingly augmented by releases of hatchery offspring
for many reasons. Reasons for use of hatchery supplementation as a management
prescription include augmentation of existing stocks (Levin .et a1. 2001 ), mitigation for

dam construction (Waples 1991a), enhancement of ﬁshing opportunities through

33

introduction of desirable sport ﬁsh, and conservation (Waples 1991a; Ryman 1991). Due
to the large numbers of ﬁsh produced in hatcheries and stocked into natural
environments, there is great concern about the potential impacts of hatchery ﬁsh on
native populations (Waples 1991a; Thomas & Mathisen 1993). Increasing evidence for
negative impacts of hatchery stocks to wild populations has prompted much discussion
and research regarding the efﬁciency of hatchery use for conservation and management
(Waples 1991a; Ryman 1991; Hilbom 1992; Ryman et a1. 1995; Busack & Currens 1995;
Philippart 1995; Waples 1999). Interactions between hatchery and wild populations have
been widely studied both from biological and genetic perspectives to determine potential
for introgression, biological interactions, and threats to conservation (Hindar et a1. 1991;
Waples 1991a; Hutchings 1991; Washington & Koziol 1993; Heggberget et a1. 1993;
Thomas & Mathisen 1993; Gharrett 1994; Fleming et a1. 1996; Reisenbichler & Rubin
1999)

Concerns regarding the potential interactions of hatchery and wild populations
have led to evaluations of hatchery practices as they relate to the maintenance of genetic
diversity (Allendorf & Phelps 1980; Kincaid 1983; Allendorf 1993). Guidelines have
been presented to improve hatchery management. Recommendations focus on effective
population size related to assessment of mating strategies (Kincaid 1995; Allendorf &
Ryman 1987), sex ratios (Simon et a1. 1986), and total numbers of adults bred. Genetic
guidelines have frequently been based on long-standing theory (Wright 1931) that
unequal numbers of males and females can lead to reduced effective population size (N),
when deﬁned as the probability that two randomly selected genes in the current

generation are copies of the same parental gene. Though Wright’s formulation of N6 is

34

an important conceptual basis for management, the use of Nc as described by Wright
(1931) does not account for variance in reproductive success among males and females.
N,- has been used to suggest guidelines for hatchery management (Ryman 1991; Ryman
et a1. 1995; Allendorf & Ryman 1987). Quantitative and comparative evaluations of
hatchery mating strategies currently in practice and those suggested as alternatives have
not been rigorously tested. While theory dictates that factors related to Ne are important,
there have been few empirical experimental studies that investigate the effects of
different hatchery gamete-take practices on measures of genetic diversity that are directly
linked to long-term probabilities of population persistence.

Variation in reproductive success has been known to occur in situations where the
gametes of males have been pooled prior to fertilization (Gharrett & Shirley 1985; Simon
et a1. 1986; Danzmann & Ferguson 1988; Withler 1988; Gile & Ferguson 1990; Fleming
& Gross 1993; Gile & Ferguson 1995). In hatchery matings either milt or eggs are oﬁen
pooled in order to maximize possibility of fertilization. Pooling of gametes increases the
possibility of competition among the gametes (i.e. sperm competition) that could increase
reproductive variation. Due to variation in reproductive success, the use of pooled
matings may bias estimates of Ne derived solely on the basis of total numbers of
individuals bred because of high potential for unequal parental contribution (Simon et a1.
1986). Inclusion of reproductive variance into estimates of NC (Lande & Barrowclough
1987; Wood 1987; Crow & Denniston 1988) can provide a more accurate assessment of
the effective number of individuals contributing to subsequent generations than is likely
realized using estimates of NC based solely on total numbers of adult males and females

spawned.

35

Relatedness among progeny provides another measure useful in evaluations of
hatchery mating regimes. Kinship has been used as a surrogate measure of relatedness in
situations where actual pedigree and actual relationships among individuals are unknown
(Bentzen et al. 2001; Ruzzante et al. 2001). When the pedigree of individuals is known,
measures of the probability of identity by decent of alleles among individuals, or
coancestry can be empirically determined. (Cockerham 1967; Cockerham 1969;
Cockerham 1973; Chesser 1991a; Chesser 1991b). However, measurements of
coancestry are rarely available for assessments of hatchery spawning practices because
parentage resulting from hatchery matings is not often determined as a direct measure of
individual parental reproductive success. In closed populations and depending on the
mating system, mean population coancestry can be a direct measure of potential for
future inbreeding.

It is desirable to minimize the increase in coancestry. In hatcheries, this can be
accomplished in several ways. The mating strategy determines how coancestry will be
accumulated with each subsequent generation. Though the importance of maintaining
genetic diversity is universally recognized, the emphasis placed on speciﬁc variables to
offer in management recommendations vary. Further, different mating regimes have
been promoted to aquaculturists and hatchery managers on the basis of the same
population genetic theory. Unfortunately, the quantitative and comparative analyses of
the efﬁciency of alternative spawning regimes have not been rigorously pursued. To
empirically determine the effects of hatchery mating strategies on measures of genetic
diversity, we examined the effects of different mating strategies used by management

agencies. We genetically determined parentage for offspring produced from six different

36

mating strategies to assess individual adult male and female reproductive success,
coancestry, percentage of unrelated offspring, and several measures of effective
population size. We evaluated each mating strategy and project generational changes in
population levels of coancestry, and the impacts resulting from stocking juveniles

propagated using each mating strategy onto existing populations (Ryman & Laikre 1991).

METHODS

Gamete take/Fertilization

We examined six mating strategies used widely by hatcheries to propagate many
ﬁsh species. Gametes were obtained from 10 female and 20 male sexually mature adult
steelhead (Oncorhynchus mykiss) captured at the Little Manistee Weir on the Little
Manistee River in Michigan. To standardize the number of gametes used from each
individual in all experimental matings measured volumes (50 ml of eggs from each
female and 1.5 ml of milt from each male) were used for each treatment. Before
fertilization, eggs from one female were placed in a bowl and mixed with eggs from
additional females according to treatment. Male nrilt was subsequently added and
thoroughly mixed (as required by each protocol; see below). Fertilization was completed
by cold-water hardening the eggs and milt for 5 minutes before pooling with the other
fertilized eggs for each treatment. After fertilization was completed, egg lots were kept
separate by treatment. Eggs were transported to the Wolf Lake Hatchery (Michigan
Department of Natural Resources) and incubated in vertical stack ﬂowing water trays
separately by treatment. Once the fry had reached swim-up stage and the yolk sac was

completely absorbed, all progeny were sacriﬁced and stored in 95% ethanol.

37

Experimental treatments

Treatments 1-4 used equal volumes of nrilt from the same 10 males, and
treatments 5 and 6 used an additional 10 males (Figure 5), also with an equal volume of
milt. Treatment 1 consisted of 10 replicates of 1:1 female to male crosses. Eggs from
female 1 (F 1) were placed in a container and fertilized with milt from male 1 (M;). This
was repeated for each female (N=10) and the ﬁrst ten males. Treatment 1 also provided a
measure of the viability of each male and female used for the experimental crosses.
Treatment 2 consisted of 10 replicates of 1 :2 female to male crosses, where each male
was used to fertilize two females. Eggs from F1 were fertilized by the combined milt of
M1 and M2, as were the eggs from F2, but in a separate container. Treatment 3 was
completed in four containers: eggs from females 1 through 5 were combined, mixed, and
split into two lots and eggs from females 6 through 10 were combined, mixed, and split
into two lots. The ﬁrst lot of eggs from each of the two egg groups was fertilized by the
combined milt of males 1 through 5, and the second lot was fertilized by the combined
milt of males 6 through 10. Treatment 4 combined the eggs from all females (1-10),
which were fertilized by the combined rrrilt of males 1 through 10. Crosses is treatments
5 and 6 both consisted of a 1:2 female to male ratio, using each male only once (i.e. F1
was fertilized with the milt of M1 and M2, F2 was fertilized were the milt of M3 and M).
In treatment 5, milt was mixed prior to fertilization; in treatment 6 the milt from each

male was added sequentially.

38

 

Genetic analysis

We used microsatellite loci to genotype each parent and a large random
subsample of offspring from each experimental treatment. DNA was extracted from ﬁn
clips of all male and female parents (N=30) and progeny (N=220 per treatment)
following the PurGene (Gentra Inc.) protocol. Fin clips from parents were stored in 95%
ethanol. Microsatellite loci Ogola (Olsen et al. 1998), Otsl (Banks et a1. 1999), OtleO
(Nelson et a1. 1998), and Omy77 (Morris et a1. 1996) were used for all treatments. Locus
Og02 (Olsen et a1. 1998) was also used for Treatments 3 and 4 to increase the statistical
power of likelihood-based parental assignment when the pool of potential parents was
increased. PCR reactions for Ogola, Ots 1, and Omy77 were conducted in 25 ul reaction
volumes using 100 ng DNA, 2 ul 10x PCR Buffer (0.1 M Tris-HCl, ph 8.3, 0.015 M
MgCl2, 0.5 M KCl, 0.1% gelatin, 0.1% NP-40, 0.1% Trition-X 100), 0.2mM dNTP’s, 0.6
uM ﬂuorescently labeled forward primer, 0.6 uM unlabeled reverse primer, and 0.3 Units
Taq Polymerase. PCR reactions for Og02 were conducted in 25 ul volumes, 3.5 mM
MgCl2, and primer concentrations were reduced to 0.5 uM. PCR reactions for OtleO
used 12.5 M MgCl2. PCR reactions for all loci utilized an initial denaturing step at 94°
C for 2 minutes, followed by 30 cycles of 94° C for 1 minute, locus-speciﬁc annealing
temperature for one minute, and extension at 72° C for 1 minute. A ﬁnal extension
period of 2 minutes, 30 seconds was followed by storage at 6° C until electrophoresis.
PCR annealing temperatures varied by loci. Annealing temperatures for Ogola, Og02,
Omy77, Otsl and OtleO were 56° C, 54° C, 54° C, 54° C, and 58° C respectively. 6%
non-denaturing acrylamide gels were used for electrophoresis and gels were visualized on

a Hitachi FM-BIO H scanner. Allele sizes were determined based on a ﬂuorescently

39

labeled ladder from BioVentures and individuals of known genotype were run on each

gel.

Parentage assignment

Parentage was assigned for all offspring in each treatment using the program
PROBMAX (Danzmann 1997). Pedigree relationships for all offspring from each of the
six treatments were empirically determined by genetic determination of parentage. Data
were used to examine the effects of each mating regime on measures of genetic diversity.
Parentage for all offspring used for analyses were based on total exclusions of all but one
maternal and paternal parent. Parentage was assigned to 194 offspring from Treatment
1, 178 offspring from Treatment 2, 162 offspring from Treatment 3, 171 offspring from
Treatment 4, 181 offspring from Treatment 5, and 189 offspring from Treatment 6.

Offspring were randomly chosen within each treatment for genetic analysis.

Statistical analysis

Based on the number of offspring genetically assigned per adult, we were able to
determine the mean and variance of male and female reproductive success for each
treatment. Variances were transformed using the Box-Cox power transformation because
they greatly varied between treatments (Rao 1998), and uniformity among variances was
tested using SAS software (SAS Institute). We estimated the percentage of all offspring
in each treatment that were determined to be unrelated (i.e. not full or half siblings). To
calculate the number of related individual pairs for both full and half siblings, the number

of offspring and their relation to each other was determined by the establishment of

40

family groups based on the genetics-based parentage assignment. The total number of
pairs was determined by factorial analysis to sample without replacement (Lindgren
1976)

Percent contribution of male parents to offspring production was calculated for
each treatment to examine inter-treatment variation in male reproductive success. Milt
and eggs from the same males and females were used in all treatments. For treatment 1
(1 male: 1 females), each of ten males would be expected to contribute 10% of the total
progeny sampled. Results from treatment 1 form the basis for comparisons of the
treatments, representing the natural background level of variation likely encountered due
to female differences in reproductive condition and egg quality and size and niunber.

Coancestry (O) values for each treatment were calculated following Cockerham
(1969),

o = n, (0.25) + nh, (0.125) + n, (0)

n

 

(1)

where nf, is the number of full-sib pairs in each treatment, mu. is the number of half-sib
pairs, nu is the number of unrelated pairs, and n, is the total number of pairs of offspring
in the treatment. The number of individuals for each sibling relationship as determined
by the genetically determined pedigree was multiplied by the appropriate coancestry
value for each pairwise relationship (i.e., 0.25 for full sibling and 0.125 for half-sibling
pairwise relationships). The estimates of relatedness between full siblings calculated by
coancestry differ from those calculated by rxy coefﬁcients because coancestry calculates
the probability of identity by descent.

To place different coancestry levels resulting from each mating strategy in an

applied context, such as the maintenance of a captive broodstock, we used a recursive

41

model to estimate and compare the rates of accumulation ofcoancestry over time for a
simple scenario of a captive broodstock population. We assumed the population was
closed, generations did not overlap, and the only reproduction that occurred was through
the hatchery (artiﬁcial) mating. Six separate initial populations were started using the
progeny of unrelated individuals, and the progeny were created following each of the six

mating regimes evaluated in this study.
(9!
91+: = 9: +(Grr)(1——k_) (2)

Initial coancestry estimates among parents used for mating in each treatment were 0.
Initial levels of coancestry for population derived and maintained using each mating
regime were empirically determined from the parentage analyses. The coancestry
estimates resulting from each mating strategy was deﬁned as (r), the rate of increase in
coancestry for each successive generation. Coancestry estimates for successive
generations (0w) were a function of the coancestry estimate for the previous generation,
(0,), the rate of increase in coancestry as a function of the mating strategy (r), and the
asymptotic value of coancestry (k, or 1).

Effective populations size (Ne) based solely on the numbers of males (Nm) and
females (Nf) used were determined following Wright (1931) as

4NmN/

Ne =(—-—-)Nm +N, (3)

An alternative and more fully parameterized measure of effective population size was
calculated by ﬁrst calculating the effective numbers of males (Nem) and females (Nef) for
each treatment (Lande & Barrowclough 1987) utilizing our empirical estimates of male

and female reproductive variance.

42

N = M M (4)

 

and

(5)

 

respectively, where h is the mean number of progeny produced by either the males (km)
or female (kf) for each treatment, and 02 is the variance in the number of progeny for the

males or females. The variance population size (Nev) for each treatment was then

calculated (Lande & Barrowclough 1987)

N..= [ 1 +4] (6)

 

N N

em ef
We. also calculated coancestral effective population size (N99) for each of the size
mating strategies. Deﬁned by Chesser et al. (1993), Neg is the number of breeding
individuals consistent with observed accrual of gene correlations over generations. Due
to our ability to calculate coancestries of progeny arrays resulting from each mating
strategy (assuming (9:0 for adults spawned to produce the progeny in all mating
strategies), we calculated the instantaneous coancestral effective size (Chesser et a1.

1993)

N... = —— (7)

43

To compare relative input of reproductive variance on N, for each of the six
mating regimes for situations where hatchery ﬁsh are used to supplement a wild

population, we followed the model

2 2
= +(I—x)
N

C W

I x

 

(8)

2

_1_

N e
used by Ryman and Laikre (1991) where x and (l-x) are the relative contribution of the
offspring of captive (x) and wild (1-x) parents, where the probability that two random
individuals were derived from the same group (captive or wild) were estimated to be x2
and (l-x)2 respectively. Estimates of Ne, NC, and NW were the effective populations sizes
of the total, captive, and wild populations respectively. Variance effective population
sizes were calculated for each treatment and were used to determine total effective size
given a range of relative contributions by hatchery ﬁsh to a wild population as per Ryman
& Laikre (1991). Initial effective population sizes for the total population were 40 and

400 individuals.

RESULTS

Assessment of reproductive variation

The viability of the gametes of each individual male and female was determined
by the single paired (1 :1) matings in treatment 1 (Figure 5). Mean numbers of offspring
produced by each female did not signiﬁcantly differ between treatments. Tests of
heterogeneity of variance revealed mean female reproductive output did not differ
signiﬁcantly among treatments (P>0.05). In contrast, comparisons of male reproductive
output among all treatments revealed signiﬁcant differences between treatment 1 and

treatment 5 (P<0.05). The mean number of offspring produced per male also did not

44

signiﬁcantly differ among treatments 1 through 4, but was reduced in treatments 5 and 6
to reﬂect the increased number of males used for fertilization (Table 5). Variance in
number of offspring per male differed greatly among treatments. High variances reﬂect
non-uniform contribution of males to reproduction, and were greatest in treatment 3
(Table 5).

Estimated contributions varied among males for each treatment (Figure 6). In
treatment 1, the estimated contribution of progeny ranged from 14.4% for males 1 and 9,
and 4.2% for male 7 (Figure 6). Male contribution for treatments 5 and 6 (n=20) would
be expected to be 5%, but estimated contribution varied greatly. Males 6 and 8
contributed the majority of progeny (18.5% and 16.7% respectively) for treatment 3 (5
males: 5 females; Figure 6). In treatment 4 (10 males: 10 females), males 4, 8, and 9
contributed 10.5%, 16.4%, and 29.8% respectively of the progeny (Figure 6). In
treatments 3 through 6, male 10 did not contribute any progeny (Figure 6). Female
contribution was relatively consistent among individuals and among treatments (Figure

6).

Eﬂ'ects of reproductive variation on measures of genetic diversity

We determined the effects of each mating treatment on summary measures of
genetic variation for progeny sampled. The controlled nature of all experimental
treatments allowed us to estimate all measures empirically. Mating regimes resulting in
the lowest mean levels of coancestry are desirable, as lower average levels of gene
correlations among progeny indicate a lower rate of allele sharing and lower levels of

inbreeding in progeny of the next generation. In treatment 1, a higher proportion of all

45

offspring surveyed were unrelated (90.2%; Table 6) compared to progeny from all other
treatments. In contrast, treatment 3 had the lowest proportion of unrelated offspring
(74.4%; Table 6). Progeny produced in treatments 1, 5, and 6 exhibited the lowest mean
coancestry values (mean ®=0.024, 0.023, and 0.023 respectively; Table 6). Estimates of
coancestry of progeny from treatments 2, 3, and 4 were comparatively higher (mean
®=0.029, 0.034, and 0.033 respectively; Table 6) due to greater reproductive skew of
males.

Estimates of effective population size varied by treatment and method of
estimation. Ne estimated based solely on numbers of males and females used in each
mating regime were consistent for treatments 1 through 4 due to the use of 20 individuals.
Increased estimates of Ne for treatments 5 and 6 reﬂect the use of 10 additional males
(total N=30). A more fully parameterized estimate of variance effective size (Nev)
incorporated male and female reproductive variance for each treatment. Across all
treatments, estimates of Nev were lowest for treatments 3 and 4 (Nev=14.1 and 14.7
respectively; Table 6). Estimates of Nev for the treatments employing 20 parents were
highest for treatment 1, being most representative of the actual number of parents mated
(Nev=19.2; Table 6). The highest estimate of Nev were documented for treatments 5 and
6, but unlike Treatment 1, estimates were not reﬂective of the actual number of
individuals used (Total N=30; Nev=21.8, 21.7 respectively; Table 6). Data conclusively
show that estimates of Nc offered solely based on numbers and sex ratio of parents mated,
without inclusion of reproductive variance are decidedly upwardly biased.

Population projections of the accumulation of mean coancestry over time

assuming closure for p0pulation (or domestic broodstock) demonstrated that the expected

46

rate of increase in coancestry over successive generations was dependent on the hatchery
mating regime employed. Coancestry values accrued at higher rates in populations
maintained by breeding regimes characterized by induced polygrry (individual males
mated with multiple females; Figure 7). Mating regimes used in treatments 2, 3, and 4
resulted in higher coancestry values than treatments 1, 5, and 6 (Table 6). Accordingly
coancestry values associated with those mating regimes increased more rapidly over
successive generations than was observed for other treatments. Mating regimes that
resulted in progeny showing the slowest increase in coancestry were those that did not
use either parent repeatedly during spawning, or those that pooled gametes prior to
fertilization (Table 6; Figure 7). The use of each parent once throughout the treatments
decreased the number of sibling relationships through the reduction in the number of
half-sibling relationships (data not shown). Mating regimes that resulted in progeny with
the highest mean coancestry also reached the level of full sibling relationships (®=0.25)
over the shortest time interval (Figure 7).

The effects of supplementation of a wild population on the total N, also varied by
hatchery mating strategy. When total population size was small (e.g., N=40; Figure 8a),
supplementation can demonstrably increase the total Ne ranged over a relatively broad
range of hatchery contributions. This trend was observed with each mating strategy. The
treatments that produced the highest NW (treatments 1, 5, and 6; Table 6) had the largest
range of relative contribution resulting in an increase in total Nc (Figure 8a). When the
total population size of is large (Ne=400), only contributions of less than 10% of

hatchery-produced individuals (for all treatments) resulted in an increase in total Nc

(Figure 8b).

47

DISCUSSION

The use of different guidelines for hatchery mating strategies has emphasized the
importance of maintaining genetic diversity (Waples et al. 1990; Kincaid et al. 1993).
The need for analysis of hatchery mating strategies is increasing as the role of hatcheries
changes ﬁ'om supplementing to sustaining ﬁsh populations (Ryman 1991). By
experimentally creating artiﬁcial crosses in a hatchery environment, we were able to
quantify the effects of six commonly employed mating regimes on measures of genetic
diversity, and offer recommendations regarding the relative success by which each
mating regime can achieve goals for the production of progeny with high levels of
genetic diversity.

Effective population size is commonly used predictive measure of generational
changes in allele frequency, heterozygosity, and inbreeding (Ryman & Laikre 1991;
Kincaid 1995; Tringali & Bert 1998). In the absence of information on male and female
reproductive success, estimates of Ne offered based on evaluations of hatchery mating
strategies are typically derived using the numbers and sex ratio of adults spawned. We
document that estimates of Ne based on the numbers and sex ratio of parents used
overestimated the effective population size when estimates of reproductive variance were
not incorporated. Previous studies have documented variance in reproductive success
(particularly of males) in mating scenarios occurred when gametes from multiple
individuals were combined prior to fertilization (Gharrett & Shirley 1985; Withler 1988;

Gile & Ferguson 1990; Gile & Ferguson 1995). With the exception of the single pair

48

mating treatrnent(1 male: 1 female), all other mating strategies we examined spawned
females with multiple males.

Male reproductive variance has often been attributed to variation in male potency
(Gharrett & Shirley 1985), timing of the application of the sperm to the eggs (Gharrett &
Shirley 1985; Gile & Ferguson 1995), number of males used (Gile & Ferguson 1990),
other characteristics such as biochemical interactions with the sperm and egg (Aas et a1.
1991), or to sperm competition. The effects of reproductive variance on levels of genetic
variation and relatedness of progeny, while potentially adaptive in the wild (through
selection on traits associated with ﬁtness), were found to result in decreased genetic
variation and increased relatedness among hatchery-produced offspring (Table 6).

Coancestry has not been commonly used to evaluate relatedness among progeny
produced in hatcheries. Many studies have used kinship (er) as a surrogate measure of O
to estimate genetic variation within populations (Blouin et al. 1996; Norris et a1. 2000;
Bentzen et a1. 2001; Ruzzante et al. 2001). We used coancestry rather than estimates of
r,‘y to measure the degree of relatedness among individuals because we were able to
determine parentage of each individual offspring directly, and subsequently determine the
relationships among all pairwise combinations of progeny. Empirical documentation of
coancestry among all progeny allowed us to predict population level increases in
relatedness over time assuming the parents used for initial matings were not related. By
incorporating the coancestry values resulting ﬁom each of the mating strategies examined
into the models described in Chesser (1991a; 1991b) and Chesser & Baker (1996), we
were able to demonstrate that mating strategies differently affect the rate at which gene

correlations accrue in a closed population. Maintaining low levels of mean coancestry

49

over successive generations is important because higher mean coancestry of potential
parents leads to higher levels of inbreeding in progeny in the next generation, and
concomitant declines in ﬁtness (Lynch & O’Hely 2001; Ford 2002). The management
and conservation implication of long-term maintenance of genetic variation applies to
situations where a broodstock or captive population is maintained in order to act as a
source for supplementation or restoration of a wild population (see Hedrick et al. 2000).

Hatchery supplementation of wild populations has been a commonly used
management strategy, and is increasingly used for purposes of conservation (Waples
1991a; Ryman et a1. 1995; Hedrick et a1. 2000; Hansen et al. 2001). We apply the
Ryman and Laikre (1991) model to evaluate whether effective population size of wild
populations can be enhanced through hatchery supplementation. We incorporate an
additional comparison based on the expected effects of different hatchery mating
regimes. Even if the genetic diversity of a captive population is maximized (by choosing
a strategy that maximizes the effective population size), the relative contribution of the
hatchery population can greatly impact the total effective population size for natural
p0pulations. Realized increases in total effective population size only occurs at very
small stocking rates, particularly when there is a large discrepancy between the effective
population sizes of the hatchery and wild populations as was the case across the six
mating regimes.

This study is the ﬁrst to empirically examine the impacts of multiple hatchery-
mating strategies used by management agencies on multiple measures of genetic
diversity. We expand upon earlier contributions (e.g., Ryman & Laikre 1991) by

incorporating more fully parameterized models for derivation of Ne contrast different

50

mating strategies. Although reproductive variance is known to occur when gametes are
pooled, the extent to which this variance would impact estimates of effective population
size or coancestry were not known, neither have projections of long-term retention of
genetic diversity resulting from each mating strategy been possible without empirical
demonstrations of expected reproductive variation. Mating regimes that minimize
reproductive variance, maximize the percentage of unrelated offspring, minimize
coancestry, and maximize variance effective population size are recommended for
hatchery use of maintenance of genetic variation is a management goal. Of the mating
regimes we used, single-pair matings between one male and one female, or those that
mated two unique males per female, best met these criteria. We offer these mating
strategies be adopted as the preferred method of propagation. Summary measures of
genetic diversity (e. g. proportions of unrelated offspring and mean coancestry) can
provide valuable insight into likely trajectories of generational change in population
levels of genetic diversity when different hatchery mating regimes are emphasized, and

should, when possible, be incorporated into hatchery management.

51

Chapter 3
TEMPORAL COMPARISONS OF GENETIC DIVERSITY IN LAKE MICHIGAN
STEELHEAD (ONCORH YNCH US MYKISS) POPULATIONS: EFFECTS OF
HATCHERY SUPPLEMENTATION
ABSTRACT

Steelhead (Oncorhynchus mykiss) were ﬁrst introduced into the Great Lakes in
the late 1800’s. Subsequently, natural recruitment across the Lake Michigan basin has
been regularly supplemented by primarily one hatchery strain. Recently, multiple strains
derived from locations across the species native range along the west coast of the United
States have also been stocked by different management agencies. Prior to 1983, hatchery
supplementation of Lake Michigan steelhead populations in Michigan was largely
unsuccessful due to low smolting rates of small (<120mm) hatchery yearlings (estimated
survival 0.01%). Accordingly, contributions of hatchery ﬁsh to historical adult spawning
runs in Michigan tributaries were low (0-30%) across six major drainages. Large
yearlings of different hatchery strains (>150mm) have been stocked exclusively since
1983, increasing estimates of survival to smolting (90%). Consequently, the proportion
of hatchery adults in spawning runs increased to l3-79%. We examined the effects of
changes in stocking practices on straying of hatchery steelhead and to temporal changes
in levels of genetic diversity and relationships among populations. We used
microsatellite loci to estimate allele frequencies for six populations sampled for two time
periods (1983-1984 and 1998-1999). Measures of inter-population divergence (mean
F31) were not signiﬁcant for either time period. However, spatial genetic relationships
among historical and contemporary populations were signiﬁcantly correlated with

geographic distance; a result not expected if gene ﬂow (natural straying) among

52

populations was mediated solely by hatchery supplementation. Increased numbers of
alleles in spawning adults from populations can be attributed to alleles speciﬁc to

recently introduced hatchery strains.

INTRODUCTION

Molecular markers have been used widely in studies of Paciﬁc salmon species
(Oncorhynchus spp.) to document interactions between natural populations and between
ﬁsh from natural and hatchery origins. For steelhead (0. mykiss), molecular markers
have identiﬁed genetic population structure at micro- and macro-geographic spatial scales
(Parkinson 1984; Reisenbichler & Phelps 1989; Reisenbichler et a1. 1992; Beacharn et al.
1999; Nielsen 1999; Nielsen & Fountain 1999; Osterberg & Thorgaard 1999). Changes
in population estimates of allele frequency and genetic diversity have been used to
examine how levels of genetic variation within and among populations have changed
over time (Heath et a1. 2002), and have provided insight into factors (both natural and
anthropogenic) contributing to temporal change (Waples 1998; Garant et a1. 2000).

Hatchery supplementation has been used widely for salmon populations.
Concerns regarding the potential for interactions between individuals of wild and
hatchery-mi gin have increased the awareness of the need to re-evaluate hatchery
management practices (Ryman 1991; Waples 1991a; Lynch & O’Hely 2001).
Theoretical and empirical evidence (Washington & Koziol 1993; Gharrett 1994; Lynch
1997; Reisenbichler & Rubin 1999) has focused on potential consequences of
introgression and breakdown of physiological or biochemical compatibilities between

genes (co-adapted gene complexes) in wild and/or naturalized populations (Lynch &

53

O’Hely 2001). One hypothesized mechanism for outbreeding and potential ﬁtness
reduction in salmonids relates to increased levels of gene flow (straying) between
hatchery and native or naturalized populations leading to the breakdown of locally
adapted gene complexes.

Paciﬁc salmon generally exhibit low levels of straying to non-natal spawning
grounds (Quinn 1993). However, increased levels of gene ﬂow between hatchery and
native or naturalized populations may occur as hatchery ﬁsh exhibit higher rates of
straying from rivers into which they were stocked (see Waples 1991a). While co-
occurrence of hatchery and native ﬁshes has been widely reported (largely ﬁom direct
observations of tags or ﬁn clips), evidence for introgression and consequences to
population viability are not widely lmown. Assessments of the magnitude of genetic
change within and among populations would proﬁt from knowledge of historical
benchmarks of genetic characteristics prior to outbreeding events. Reisenbichler &
Phelps (1989) hypothesized that introgression between native steelhead and hatchery
steelhead widely stocked across large areas led to lower inter-population levels of genetic
diversity. However, Reisenbichler & Rubin (1999) suggest that although interbreeding
may occur between hatchery and wild individuals, the potential reproductive contribution
of hatchery ﬁsh to wild spawning populations could be quite low (due to lower
reproductive success of hatchery individuals).

Steelhead were introduced into the Great Lakes in the late 1800’s (MacCrimmon
& Gots 1972). Natural reproduction primarily occurs in Michigan rivers due to the lack
of suitable spawning habitat elsewhere in the Lake Michigan basin (MacCrimmon &

Gots 1972). Widespread natural reproduction in the Great Lakes led to development of

54

self-sustaining populations. However, natural recruitment has been supplemented by
hatchery-production (Biette et a1. 1981; Seelbach 1987a). Management agencies around
the Lake Michigan basin stock multiple steelhead strains to take advantage of strain-
speciﬁc variation in life history characteristics (i.e., run timing) that are valued by
recreational anglers. Because naturalized steelhead spawn in drainages in only a portion
of the basin, stocking acts to increase the number of ﬁsh present in rivers during
spawning runs, and to increase the distribution of steelhead in open-water areas
throughout the basin. One hatchery strain (the Michigan strain, derived ﬁ'om the Little
Manistee River) has historically been used across the basin, and is presently used for
supplementation by the state of Michigan. Recently released hatchery strains include the
Chambers Creek and Skamania strains originating from Washington, and the Ganaraska
strain established from the naturalized population in the Ganaraska River in Ontario.
Currently, genetic differences between the four hatchery strains represent the major
component of genetic diversity in Lake Michigan steelhead (Bartron et al. in review).
Introgression among hatchery strains stocked by other agencies across the Great Lakes
basin may lead to increasing levels of straying, outbreeding, and loss of strain-speciﬁc
hatchery traits.

Changes in hatchery management practices have increased the relative
contribution of hatchery steelhead to naturalized populations in Michigan. The state of
Michigan changed the size and age of stocked juvenile steelhead in 1983, from stocking
fall ﬁngerlings (<110mm) to large yearlings (>150mm; Seelbach 1987a). Larger
juveniles had higher rates of smolting (mean 48.2% vs. 0.5-2.9%; Seelbach 1987a) and

survival to smolt stage (90% vs. 0.01%; Rand et a1. 1993) compared to juveniles stocked

55

at smaller sizes. Prior to changes in age and size at stocking, the average contribution of
hatchery ﬁsh to spawning runs in six Michigan rivers was generally low (0-30%;
Seelbach & Whelan 1988). Following the change in stocking practices, the contribution
of hatchery ﬁsh in the same six rivers during the 1998-1999 spawning run substantially
increased (13-79%; Bartron et al. in review).

In this study, we compared allele frequencies and levels of genetic diversity for
two different time periods (1983-1984 and 1998-1999) for each of six populations before
and after changes in juvenile age at stocking. Comparisons should show whether
increased survival of hatchery juveniles and abundance of adult hatchery steelhead led to
increased levels of introgression as evidenced by altered genetic characteristics within,
and inter-relationships among naturalized populations. Our two main objectives were 1)
to determine if time series data on survival and stocking histories for steelhead in the
Lake Michigan basin were suggestive of increasing threats of introgression between
hatchery and wild individuals, and 2) using time-series data of population allele
frequencies and estimates of genetic diversity, to determine if the increased presence of
hatchery individuals during the spawning period had demonstrable effects on population

genetic characteristics of naturalized populations.

METHODS
Stocking history
Original introductions of steelhead into the Great Lakes in the late 1800’s utilized
gametes taken from rivers in California (primarily McCloud River, Klamath River, and

Redwood Creek), and the Willamette and Rogue rivers in Oregon (MacCrimmon & Gots

56

1972). Within Lake Michigan speciﬁcally, steelhead (or non-anadromous rainbow trout)
from the McCloud River were ﬁrst stocked in 1896 (MacCrimmon & Gots 1972).
Naturalized populations were well established and widely distributed around the Lake
Michigan basin by the 1920’s although primary reproduction occurred in Michigan due to
the abundance of spawning habitat (MacCrimmon & Gots 1972). Supplemental stocking
within Lake Michigan did not occur again until the mid 1950’s following a decline in the
lake-wide steelhead population (MacCrimmon & Gots 1972). Until the mid-1980’s,
stocking primarily utilized one hatchery strain, derived from returning adults captured at
the Little Manistee weir in Michigan’s Little Manistee River (Figure 9). Starting in the
mid 1980’s, additional hatchery strains have been used for stocking by states around the
Lake Michigan basin. These additional hatchery strains include: the Chambers Creek
strain from the Puget Sound region of Washington; the Skamania strain, from the
Washougal River in Washington; the Ganaraska strain, ﬁ'om the Ganaraska River in
Ontario, Canada. The Ganaraska and Michigan strains originate from naturalized
populations within the Great Lakes. Each of the four hatchery strains stocked into Lake
Michigan are maintained by gametes taken from adults returning to weirs located on four
rivers around the Lake Michigan basin each year. No speciﬁc broodstocks are
maintained. Thus, unintentional straying and introgression of ﬁsh of multiple strains
could compromise the genetic integrity of each strain.

Stocking records of Michigan hatchery strain steelhead for the Lake Michigan
basin (Table 7) and speciﬁcally for the Lake Michigan tributaries examined in this study
(Table 8) were obtained from the Michigan Department of Natural Resources for two

time periods (1979-1982 and 1993-1997). Time periods represent years preceding the

57

two sampling events of adult steelhead spawning runs, and reﬂect cohorts that may
contribute to the spawning runs sampled. Smolt-equivalents were calculated from
estimates of survival rates of juvenile steelhead stocked into Lake Michigan based on

age, size, and stocking location (Rand et a1. 1993).

Sample collection
Samples from adult steelhead spawning populations were obtained from six

populations for each of two time periods, before (1983-1984) and after (1998-1999,
2000) changes in stocking practices. Adult steelhead ﬁom both time periods were
sampled from the Betsie River, Manistee River, Little Manistee River, Pere Marquette
River, White River, and Muskegon River (Figure 9). Historical samples were based on
archived scale samples held at the Michigan Department of Natural Resources Institute
for Fisheries Research (Ann Arbor, Michigan) that had been obtained by creel surveys
and hook and line sampling of adult steelhead. Archived scales were sampled from
adults during the fall 1983 and spring 1984 spawning runs in each river (Seelbach &
Whelan 1988). Among historical populations, fall 1983 and spring 1984 samples were
obtained for the Betsie, Little Manistee, and White rivers. The Manistee, Pere Marquette,
and Muskegon rivers were only sampled in the fall of 1983. Contemporary populations
in the Manistee, Little Manistee, Pere Marquette, and Muskegon rivers were sampled
during both fall 1998 and spring 1999. The White and Betsie rivers were sampled in the
spring of 2000. Among historical populations, the Betsie (fall run N=l7; spring run

=33), Little Manistee (fall run N=29; spring run N=29), and White (fall run N=7; spring

run N=6) rivers were sampled during both fall and spring runs. Among contemporary

58

populations, the Manistee (fall run N=53; spring run N=105), Little Manistee (fall run
N=57; spring run N=60), Pere Marquette (fall run N=29; spring run N=73), and
Muskegon (fall run N=37; spring run N=39) rivers were sampled during both fall and
spring runs. Contemporary populations were sampled by electroﬁshing and creel surveys
of adult steelhead returning to Michigan rivers throughout the fall of 1998, spring 1999,

and spring of 2000, to include samples from the entire spawning period.

Scale pattern analysis (SPA)

Scale pattern analysis (SPA) was used to determine hatchery or natural origin of
each adult steelhead (Seelbach & Whelan 1988). SPA for ﬁsh sampled from both
historic and contemporary time periods was performed using the ratio 23 method
(Seelbach & Whelan 1988). Ratio 23 quantiﬁes differential winter and spring growth
rates in juvenile steelhead through comparison of the width of the ﬁve intercirculus
spaces prior to the ﬁrst annulus to the width of the ﬁve intercirculus spaces that follow
the ﬁrst armulus (Seelbach & Whelan 1988). For contemporary populations, juvenile
stream resident time (Bartron et al. in review) was also used to identify hatchery-ori gin
individuals. Juvenile stream resident time was used when the ratio 23 value was between
0.7 and 0.8 (0.7 being the threshold for detemrination of hatchery or natural-origin).
Hatchery yearlings migrate downstream to Lake Michigan within a year after stocking,
whereas juvenile steelhead produced in the wild may reside in the stream for more than
one year prior to smolting (Seelbach 1987a). The use of an additional technique to
identify hatchery-ori gin individuals was necessary due to changes in growth patterns

resulting ﬁ'om the increased size and age-at-stocking of hatchery ﬁsh. The likelihood of

59

classiﬁcation errors resulting from the ratio 23 technique is discussed in Seelbach &
Whelan (1988). All steelhead identiﬁed as originating in hatcheries based on SPA were

removed from the genetic analysis.

Genetic analysis

After removal of hatchery individuals from each population, sample sizes for
historical and contemporary populations for each river were: Betsie River (N=49 and
N=25), Manistee River (N=52 and N=107), Little Manistee River (N=57 and N=116),
Pere Marquette River (N=10 and N=102), White River (N=15 and N=18), and Muskegon
River (N=7 and N=75). DNA was obtained from scales and ﬁn clips. DNA ﬁom scales
was extracted using Qiagen DNeasy® Tissue Kit (Qiagen Inc.). DNA from ﬁn clips was
extracted using PurGene® (Gentra Inc.) protocols. Individuals were genotyped at six
rrricrosatellite loci, including Ogola and Ogo4 (Olsen et a1. 1998), Omy77 (Morris et a1.
1996), OneulO (Scribner et a1. 1996), Otle3 (Beacham et a1. 1998), and Oki200
(Beacham et a1. 1999). PCR reactions for Ogola, Omy77, Oneu10, and Ogo4 followed
protocols described in Bartron et al. (in review). PCR reactions for Otle3 and Oki200
were conducted in 10 ul volumes using 40 ng DNA, 1 pl 10x PCR Buffer (0.1 M Tris-
HCl, pH 8.3, 0.015 M MgCl2, 0.5 M KCl, 0.1% gelatin, 0.1% NP-40, 0.1% Triton-X
100), 0.2 mM dNTP’s, 0.4 uM ﬂuorescent labeled forward primer, 0.4 uM unlabeled
reverse primer, and 0.3 units Taq Polymerase. PCR reactions for all loci utilized an initial
denaturing step at 94° C for 2 min., followed by 30 cycles for tissue-derived DNA and 40
cycles for scale-derived DNA of 94° C for 1 min, annealing temperature for 1 min., and

extension at 72° C for l min., and a ﬁnal extension period of 2.5 min. The annealing

60

temperature for OtleB and Oki200 was 50° C. Genotypes for Ogola, Ogo4, Omy77,
and OneulO were visualized on a Hitachi F M-BIO® II scanner. Genotypes for Otle3
and Oki200 were visualized on a LI-COR® IR2 Global Edition DNA Sequencer.
Molecular weight standards and individuals of known genotype were nm on each gel to

standardize scoring.

Statistical analysis

Samples were taken over two portions of the spawning run (fall and spring) from
most populations. We used Fisher’s exact test (GENMOD procedure) in SAS software
(SAS Institute 1999) to determine if proportions of hatchery and river-ori gin individuals
differed between fall and spring runs, and between historical and contemporary time
periods.

Estimates of observed and expected heterozygosity and allele frequencies for each
population were calculated using BIOSYS-l (Swofford & Selander 1981). Signiﬁcance
of deviations of genotypic frequencies ﬁom Hardy-Weinberg expectations were
determined for each population using Markov chain methods of Guo & Thompson (1992)
implemented in program GENEPOP (Raymond & Rousset 1995). Allelic richness values
for each population for each locus (Petit et a1. 1998) were calculated using FSTAT
(Goudet 1995) v2.9.3.1 to standardize population measures of allelic diversity for
differences in sample size. Mean allelic richness estimates for each population were
presented as the sum of the allelic richness values assayed.

Pairwise F51 (Weir & Cockerham 1984) comparisons were used to determine if

genetic differences existed between fall and spring runs for those rivers that were

61

sampled for both runs. In the absence of differences, fall and spring samples were
combined for analyses of population differences. Estimates of pair-wise and overall
population differentiation were summarized using F-statistics, implemented in the
program F STAT (Goudet 1995) v.2.9.3.1. Nominal alpha values were corrected for
multiple pair-wise comparisons using Bonferroni corrections (Rice 1989). Cavalli-Sforza
& Edwards (1967) chord distances were estimated for all population comparisons during
both time periods using BIOSYS-l (Swofford & Selander 1981) and were used to
construct neighbor-joining (Saitou & Nei 1987) dendrograms. Neighbor-joining trees
and associated bootstraps (500 iterations) were generated using PHYLIP v3.5
(F elsenstein 1993). The resulting trees were displayed using TREEVIEW (Page 1996).
Generalized Mantel tests (Smouse et a1. 1986) were used to test for correlations
between genetic relationships (inter-population genetic distance) and population
geographic proximity, providing measures of spatial autocorrelation in allele frequency.
Geographic distances (km) were estimated between river mouths along the Lake

Michigan shoreline.

RESULTS
Hatchery contributions to spawning runs
Stocking histories for steelhead of the Michigan hatchery strain into rivers in
Michigan revealed a decline in the number of steelhead stocked over time (Table 7). On
average, 1,045,737 Michigan-strain steelhead were stocked into Lake Michigan each year
between 1979 and 1982, whereas an average of 524,895 were stocked each year between

1993 and 1997 (Table 7 ). Following the change in age and size at stocking, the estimated

62

number of stocked juvenile steelhead that survive to smolting increased from 4.1% to
83.2% between the two time periods (Table 7). Survival estimates were comparable to
those of river-origin steelhead in Lake Michigan, which ranged ﬁ'om 13-90% for
presmolt winter survival (Seelbach 1987b). Between 1978 and 1982, the estimated
yearly mean number of Michigan strain juveniles stocked into Lake Michigan surviving
to smolt stage was 33,509 (Table 7), compared to a mean of 437,061 between 1993 and
1997 (Table 7). Between 1983-1984 and 1998-1999, the mean number of Michigan
strain steelhead stocked into the six rivers examined in this study that survived to smolt
stage increased ﬁom 3,054 to 28,269 (Table 8). Estimates of natural recruitment are not
available for the six rivers, however the ten-fold increase in hatchery contributions that
survive to the smolt stage suggests an increased potential for straying of adult hatchery
ﬁsh into natural spawning populations across the Lake Michigan basin.

We observed an increase in the proportion of hatchery-ori gin individuals found in
adult steelhead spawning runs for six rivers in Michigan between the 1983-1984 and
1998-1999 spawning runs (Table 8). The contribution of hatchery steelhead in the six
rivers examined averaged 12% during 1983-1984 spawning run, and 40% during the
1998-1999 spawning run. Signiﬁcant differences were observed in hatchery
contributions to the spawning run of each river between the historical and contemporary
populations for the Little Manistee River (def=|=37.86; P<0.0001), Manistee River
(dee1=15-54; P<0.0001), Muskegon River (def=1=10.49; P<0.0012), and White River

(x’df=1=5.38; P<0.02).

Genetic analysis

63

The mean number of alleles per locus across all loci increased over time
(comparisons of historical and contemporary populations; Table 9). The mean number of
alleles per locus ranged from 42-47 for the historical populations and 5.3-7.8 for the
contemporary populations (Table 9; speciﬁc allele frequencies are provided in Appendix
1). When measured by allelic richness to correct for sample size, the mean number of
alleles per locus increased between time periods for three of the six rivers (Table 9).
Alleles not found in historic populations were observed in contemporary populations, half
of which are found in one or all of the Skamania, Chambers Creek, or Ganaraska
hatchery strains (Appendix 1). Five alleles (in loci Ogo4 and Omy77) were found in the
contemporary populations, and in three of the hatchery strains, but not in the historic
populations (Appendix 1). One allele (in locus Ogola) was absent ﬁ'om historic
populations except for the Little Manistee, but was found in three of the other ﬁve
contemporary naturalized populations (Appendix 1). Given that the majority of common
alleles are shared across strains and naturalized populations, estimates of introgression
based solely on the appearance of rare alleles are likely under-estimates. Mean observed
heterozygosity for the historical populations ranged from 0.563 to 0.632 and from 0.555
to 0.686 for the contemporary populations (Table 9). Deviations from Hardy-Weinberg
equilibrium were not observed in any population sampled during either time period
(P>0.05).

We found no evidence for signiﬁcant differences in allele ﬁ'equency between fall
and spring portions of spawning runs within rivers. Variance in allele frequency as
estimated using FST values for comparisons between fall and spring runs sampled within

each river for each time period did not indicate evidence for signiﬁcant genetic

64

differentiation (P>0.05 for mean FST values across all pair-wise comparisons).
Consequently, samples from each run were combined for those populations that included
samples from the fall and spring runs. Mean F51 values when estimated across all six
populations for each time period were not statistically signiﬁcant (mean F ST for historical
and contemporary populations were 0.006 and 0.002 respectively; Table 10).

Mantel tests were conducted using genetic distances among populations from
each time period and for correlations between inter-population genetic distances arnon g
populations. The Mantel tests indicated that genetic relationships among populations
were consistent between the two time periods (R=0.65; P<0.01; Figure 10). Correlations
between the geographic and genetic distances between each spawning population were
signiﬁcant for historic populations (R=0.55; P<0.01), indicating signiﬁcant spatial
autocorrelation in allele frequencies between population (isolation by distance)
relationships. Relationships between genetic and geographic distance were also evident
for contemporary populations, though the relationship was less strongly supported
(R=0.39; P<0.02). Neighbor-joining tree topology representing inter-population variation
in allele frequency across loci as viewed in the Cavalli-Sforza & Edwards (1967)
distances among historic populations generally grouped populations with geographic
neighbors, but topology may have been biased by populations with small sample sizes
(i.e. Muskegon, Pere Marquette, and White river populations). Among contemporary
populations grouping was more random, and again may be biased by populations with
small sample sizes (i.e. Betsie and White rivers; Figure 10). Nodes for both trees were

weakly supported by bootstrap values (Figure 10).

65

DISCUSSION

Widespread stocking of hatchery ﬁsh has raised concerns regarding the potential
for interactions between hatchery and wild populations of many ﬁsh species. Particular
emphasis has been placed on salmonid populations in their native range due to their
socio-econornic importance and to dramatic declines in population numbers and
distribution. Individuals of hatchery origin commonly co-occur with wild individuals,
which pose increased risks of introgression. Outbreeding can lead to the loss of genetic
adaptations within populations (Lynch & O’Hely 2001). Loss of naturally evolved co-
adapted gene complexes may not be a pressing issue in introduced and artiﬁcially
maintained systems, such as those that currently exist in the Great Lakes. However,
phenotypic differences (e. g. variation in run timing) are often exploited by managers to
provide increased recreational ﬁshing opportunities for anglers. In Lake Michigan,
multiple hatchery strains with evolved tendencies to enter rivers at different times of the
year are stocked to extend the duration of once seasonal ﬁsheries. Concurrent stocking of
multiple strains increase risks of introgression among strains and between individuals of
hatchery and natural origin. Variation among strains in phenotype or life history,
whether adaptive or not, is useful to managers and is at risk of loss through outbreeding.

Although large numbers of juvenile steelhead are stocked yearly (Table 7), natural
reproduction provides a substantial component of the steelhead production to Lake
Michigan, and hence to the basin-wide recreational ﬁshery. Rand et a1. (1993) estimated
that steelhead smolts emigrating from the Little Manistee River contribute 13-21% of the
total basin-wide smolt yield. Although the total number of hatchery steelhead stocked in

Michigan has decreased since the early 1980’s (Table 7), survival of hatchery ﬁsh to

66

smolt-stage has increased (Table 8). As a consequence, the proportion of hatchery ﬁsh
present in spawning runs increased signiﬁcantly between the two time periods (Table 8).
Clearly, increased abundance of hatchery adults in spawning runs increases the potential
for introgression between the hatchery and naturalized populations of steelhead in
tributaries across the Lake Michigan basin. Previously, there was a lack of information
regarding how the threat of increased introgression translated to actual outbreeding. .

We did not observe strong evidence of increased levels of gene ﬂow on the basis
of changes in magnitude of inter-population variance in allele frequency as seen from F-
statistics. However, our ability to detect effects of potential introgression if occurring
(i.e. statistical power) was low, as historical populations were not signiﬁcantly diverged
genetically. We did however observe signiﬁcant correlations between genetic distance
and geographic distance among populations, suggesting that steelhead populations in
Michigan (both historical and contemporary) do not represent a single panmictic
population, but rather are structured as a function of geographic distance among
populations. Decreased correlations between genetic and geographic distance in
contemporary populations as compared to historical populations are suggestive of some
level of changes in patterns of gene ﬂow or homogenization of allele frequencies due to
increased introgression as would be expected between hatchery and naturalized
individuals.

Recent introduction of new genetically differentiated hatchery strains by states
across the Lake Michigan basin has let to increasing incidence of these strains in
Michigan rivers (Bartron et al. in review). Introgression most likely has contributed to

increased levels of within-population diversity for most contemporary populations

67

surveyed in this study. Spatial variance in allele frequency represents a small component
of the total genetic variation apportioned across the Lake Michigan basin. Genetic
diversity within populations was higher in contemporary populations than in historical
populations due to increased mean number of alleles per locus. The appearance of alleles
not found in historical populations is due to the recently introduced hatchery strains
(Appendix 1). Since all ﬁsh surveyed are of wild origin, the formerly unseen alleles can
only be attributed to past introgression.

The absence of signiﬁcant genetic structure among historical populations was not
unexpected because of the proximity of natural populations to each other, and the
relatively short time steelhead have been present in Lake Michigan. Differentiation over
relatively short geographic distances was detected among steelhead populations within
the Skeena River (Beacham et a1. 2000). On a larger spatial scale, Krueger & May
(1987) analyzed steelhead populations within the Lake Superior basin and found greater
levels of population differentiation (F ST=0.026). Observation of signiﬁcant spatial
structure by Krueger & May (1987) was potentially due to the sampling of juveniles, and
spatial structure could reﬂect family groups rather than population differences (Allendorf
& Phelps 1981).

Despite lack of signiﬁcant spatial genetic differentiation, genetic characteristics
observed within steelhead populations in Lake Michigan (such as the lack of signiﬁcant
genetic differences between fall and spring run steelhead) are consistent with
comparisons of winter and summer-run steelhead in their native range (Chilcote et al.
1980; Leider et al. 1984). Heath et a1. (2002) found temporal comparisons of genetic

diversity among steelhead populations to be much greater over a shorter time period than

68

was found among Michigan populations. The lack of greater temporal genetic
divergence among Michigan populations could be attributed in part to disparities in
sample size obtained for populations in each time period. However, three of the historic
populations had large sample sizes for both time periods and pair-wise comparisons
between those p0pulations and their contemporary counterpart did not indicate signiﬁcant
differences.

Within regional spatial contexts such as the Great Lakes, gene ﬂow among ﬁsh
from naturalized populations may be high due to natural levels of straying. However, the
Great Lakes basins share a number of characteristics of other multi-jurisdictional
ﬁsheries resources that are increasingly at risk to accelerated levels of outbreeding due to
increased reliance on hatchery supplementation. In the Great Lakes, different agencies
stock ﬁsh of different genetic backgrounds. Due to higher rates of straying by hatchery
ﬁsh (Waples 1991a), the potential to interbreed with native or naturalized ﬁsh from
locally adapted gene pools increases as seen by levels of gene diversity within and among
steelhead populations in Lake Michigan. Concomitantly, apportionment of genetic
diversity may be altered within and among populations. Findings of signiﬁcant
correlations of geographic and genetic distance in contemporary populations suggest that
contributions to recruitment from naturalized ﬁsh continue to be proportionally higher
than is realized by ﬁsh of hatchery origin despite changes in management practices used -

for hatchery supplementation.

69

Chapter 4
METHODOLOGICAL BIAS IN ESTIMATES OF STRAIN COMPOSITION AND
STRAYING OF HATCHERY-PRODUCED STEELHEAD IN LAKE MICHIGAN
TRIBUTARIES
ABSTRACT
Steelhead (Oncorhynchus mykiss) were ﬁrst introduced into the Great Lakes in

the late 1800’s. Subsequently, natural recruitment of steelhead from spawning runs in
streams across the basin has been regularly supplemented by hatchery production of
strains derived from widely dispersed locales within the species’ native range. Estimates
of hatchery contributions to spawning runs of naturalized populations may be under-
represented by observations of clipped ﬁsh, as not all hatchery ﬁsh are marked prior to
release. To assess the potential bias to estimates of hatchery contribution to steelhead
spawning runs in four major rivers in Michigan, we used scale pattern analysis (SPA) to
identify non-clipped hatchery ﬁsh, and multi-locus genotypes to estimate proportional
contributions of each hatchery strain to spawning runs. The largest component to genetic
diversity in Lake Michigan steelhead is among the ﬁve strains currently stocked (mean
Fs1=0.077), making strain-speciﬁc identiﬁcation using likelihood-based assignment tests
possible. The main cause for differences between direct (clip observations) and indirect
(SPA and genetic analysis) estimates of hatchery contribution were due to variations in
percentage of hatchery ﬁsh clipped by states prior to release and the potential for
conﬁrsion of certain marks with injuries. Combining direct and indirect assessment
methodologies, we estimated that the percentage of hatchery ﬁsh retuming to four rivers
ranged from 13-31% of total spawning runs. The large contribution of hatchery ﬁsh to

the non-stocked rivers differed signiﬁcantly from expectations of strain-speciﬁc stocking

70

rates across the Lake Michigan basin and for individual streams, indicating a high rate of

straying into Michigan streams.

INTRODUCTION

Many ﬁsheries exist in waters that extend across multiple state, provincial,
national boundaries, and are jointly managed by several management agencies. Different
agencies often use different assessment methodologies, making coordination and multi-
jurisdictional ﬁsheries data comparison difﬁcult. Such an example is found for steelhead
(Oncorhynchus mykiss) ﬁsheries in Lake Michigan, where the four states surrounding the
lake use identifying marks (e. g. ﬁn clips, maxillae clips, or a combination thereof) to
identify hatchery-origin steelhead. All states stocking steelhead around the Lake
Michigan basin clip a portion of the hatchery steelhead prior to release (Table 11). Clip
data are used to estimate hatchery strain-speciﬁc rates of straying, and to estimate the
contribution of both hatchery- and naturally-produced steelhead to the recreational
ﬁshery in Lake Michigan and its tributaries. Accurate assessments of straying and
relative contribution of hatchery ﬁsh are likely compromised due to duplication of
speciﬁc clips for more than one strain, and because not all individuals of hatchery-origin
are clipped by each state prior to stocking. Additionally, overestimates of certain
hatchery strains are likely, as general strain-speciﬁc clips may be confused with booking
injuries (i.e., maxillae clips) resulting in upward bias in abundance estimates.

The use of multiple assessment techniques to identify strain of p0pulation
contributions to mixtures has been widely used in ﬁsheries management. Traditional

techniques (scale pattern analysis, coded wire tags, ﬁn clips, otolith analysis) have been

71

useﬁrl to place individuals to geographic locations (Candy and Beacham 2000), hatchery
strain (Burzynski 1999), or stocking location (Thedinga et a1. 2000; Hard and Heard
1999). Genetic stock identiﬁcation techniques are also commonly used to discrirrrinate
among stock contributions to ﬁsheries (Pella and Milner 1987; Scribner et a1. 1998;
Beacham et a1. 2000; Hansen et a1. 2001; Potvin and Bematchez 2001).

Molecular techniques have become a common tool in ﬁsheries management.
Molecular analysis of degree of population structure and assessment of strain or
population contribution to ﬁsheries (Marshal et a1. 1991; Scribner et a1. 1998; Beacham et
al. 2000) represent just a few applications. In the absence of physical marks, the ability
to assign individuals to population or strain of origin is based on degree of population
differences in allele frequencies for a suite of genetic markers (Comuet et a1. 1999), and
has been particularly useful in ﬁsheries assessments. Molecular markers, when used in
association with more traditional ﬁsh identiﬁcation techniques can increase classiﬁcation
accuracy and effectively deﬁne the potential for hatchery and wild interactions.

Steelhead were ﬁrst introduced into Lake Michigan in the late 1800’s (Biette et a1.
1981). Adult steelhead return in the fall and spring to natal rivers, and spawn from early
to late spring. Spawning habitat in Wisconsin, Indiana, and Illinois is marginal or limited
in distribution (Seelbach 1986). Accordingly, reproduction in rivers primarily occurs in
Michigan tributaries due to the abundance of favorable spawning habitat. After hatching,
juveniles spend one to three years in the river before emigrating as smolts to the Lake
Michigan (Seelbach 1993). Juveniles spend one to three years in the lake or ocean
enviromnent before returning to spawn in their natal river (Seelbach 1993). Adults are

believed to stray to rivers other than the natal river for spawning, but at low rates (Quinn

72

1993). Estimated straying rates were historically around 6% in Lake Michigan (Stauffer
1955) and 3-10% in Lake Huron (Dodge 1972).

Presently, spawning runs of steelhead in many tributaries to Lake Michigan are
maintained by both natural reproduction and hatchery supplementation. Hatchery
production averages 1.8 million annually (Table 11), and is used to supplement natural
recruitment to enhance recreational river and lake ﬁsheries. High levels of stocking and
changes in age at release (fall ﬁngerling to large yearling; Seelbach 1987a) in some states
have likely increased juvenile survival, and led to greater returns of adult hatchery
steelhead (Seelbach 1987a; Seelbach and Miller 1993; Rand et a1. 1993).

Stocking location, size, and age of j uvenile steelhead can produce large variations
in survival to smolt-stage for juvenile steelhead and in straying rates of spawning adults.
Stocking locations of steelhead have been widely distributed around the Lake Michigan
basin. Estimated probability of survival of stocked juveniles in Lake Michigan can vary
from 0.0001 over two years for a ﬁngerling stocked into a marginal river (not optimal
trout habitat) to 0.9 for a large yearling (>150 mm) stocked into a trout river (Rand et al.
1993). Rates of straying of anadromous salmonids have been found to be impacted by
stocking location within a river (upstream versus river mouth; Thedinga et a1. 2000), date
of release (Pascual et a1. 1995), size of juveniles at time of release (Pascual et a1. 1995),
and if stocking location differed from the location where juveniles were raised (Pascual et
a1. 1995).

Our objectives were to assess degree of bias in assessments of hatchery strain
composition of spawning runs and to determine whether strain-speciﬁc rates of straying

were consistent with levels of stocking and origins of release. We examined several

73

hypotheses pertaining to straying patterns of hatchery steelhead for Lake Michigan
tributaries. The null hypothesis being tested was that steelhead of hatchery origin stocked
into Lake Michigan represent a single panmictic population, and the abundance of adults
of all hatchery strains was proportional to the rates at which they were stocked. Each
population would be expected to be composed of adults of natural origin and hatchery
ﬁsh present in proportions reﬂecting relative stocking proportions. Support for this
general hypothesis comes from the fact that Lake Michigan represents a relatively small
basin in comparison to the Paciﬁc Ocean, and steelhead have been found to be widely
distributed throughout open-water habitats in the Lake Michigan basin. We further
wished to examine potential differences in hatchery strain composition in each of four

rivers surveyed.

METHODS

Stocking history

Stocking records for the Lake Michigan basin were obtained for the years
between 1993 and 1997; representing the time period that hatchery steelhead returning to
rivers in 1998-1999 would have been stocked. Speciﬁcally, for each state around the
Lake Michigan basin, information was collected regarding the hatchery strain stocked,
the number of individuals stocked (marked and not marked), size and age of the hatchery
juveniles stocked, location of stocking, and type of mark used. The speciﬁc stocking
records were obtained from Burzynski (1999), J. Palla (Indiana Department of Natural

Resources, personal communication), http://wwwmichigangov/dnr (Michigan

74

Department of Natural Resources), and S. Krueger (Illinois Department of Natural
Resources, personal communication).

Hatchery marks observed in each of the four rivers examined were identiﬁed to
strain of origin using stocking records obtained from each state agency that has stocked
steelhead into the Lake Michigan basin (Table 11). Due to the use of a single ﬁn clip to
designate both the Michigan strain and the Indiana Skamania strain, all ﬁsh with clips
identifying individuals to one of those two strains were classiﬁed as Michigan strain. Due
to the small portion of Indiana Skamania steelhead marked prior to release (15%; Table
11), the bias introduced by combining the information would be small. Therefore,
individuals identiﬁed as Skamania strain by clip observations represented only the
Wisconsin Skamania strain. To determine the hatchery composition of the non-clipped
ﬁsh, we collected ﬁn clips for genetic analysis and scale samples (for SPA) from all non-
clipped individuals.

We developed expectations for proportional contributions of adults of hatchery
origin to spawning runs based on total numbers of juveniles stocked. However, there is
likely to be increased mortality of hatchery steelhead stocked as ﬁngerlings in
comparison to those stocked at a larger size in Lake Michigan (Seelbach 1987a). To
account for differential mortality to smolt stage (smolt equivalents) based on age and size
at stocking (i.e., small yearlings <150 mm versus large yearlings Z 150 mm), and
stocking location, the number of hatchery juveniles stocked were adjusted as described
by Rand et a1. (1993). These adjustments allowed estimated abundances of hatchery ﬁsh
stocked into the Lake Michigan basin to be standardized and to more accurately represent

the proportional abundance of each strain. Adjustments were made by multiplying the

75

Ill.‘

$1

(I)

number stocked by an estimate of percent survival to smolt stage. The resulting numbers
of smolt-equivalents were used as the total expected abundance of each hatchery strain of

steelhead in each river.

Sampling locations and tissue collection

Adult steelhead were sampled from four Michigan rivers during the fall 1998 and
spring 1999 spawning runs. Samples were obtained by electro-shocking and creel
sampling on spawning grounds in the Pere Marquette River, Bear Creek, and Platte
River, and at a weir for the Little Manistee River (Figure 11). Samples were collected on
multiple dates over the course of the spawning season to ensure representative samples
were obtained from spawning adults. The Pere Marquette River and Little Manistee
River were sampled during the fall and spring, while Bear Creek and Platte River were
sampled only in the spring. Due to absence of genetic differences between the fall and
spring run and no difference in the proportional contribution of individuals of hatchery
origin to the fall run and spring run within the Pere Marquette and Little Manistee
(Bartron and Scribner, in press), fall and spring samples for these river systems were

combined.

Scale pattern analysis (SPA)

We used scale pattern analysis to estimate stream growth and residence time of
juveniles to determine if non-marked individuals were of wild or hatchery origin.
Speciﬁcally, we used Seelbach and Whelan’s (1988) ratio 23 metric of scale growth.

Ratio 23 is the ratio of the distance from the ﬁrst stream annulus to the ﬁfth circulus

76

me
the

193

ori
inc
rat
int
aft
for

JU‘

pr

51

a!

measured towards the focus (band 2), and to the distance from the ﬁrst stream annulus to
the ﬁfth circulus measured towards the scale margin (band 3; Seelbach and Whelan
1988). Distributions of ratio values from ﬁsh of known origin (hatchery or river) were
used to determine a point at which hatchery-ori gin ﬁsh could be differentiated from river-
origin ﬁsh (Seelbach and Whelan 1988). We also considered the number of years an
individual spent in a river to improve the determination of origin for individuals whose
ratio 23 value fell within the area of overlapping distributions of known origin
individuals. Hatchery yearlings migrate downstream to Lake Michigan within a year
after stocking, whereas juvenile steelhead produced in the wild may reside in the stream
for more than one year prior to smoltiﬁcation (Seelbach 1987a). Incorporation of
juvenile stream age likely reduced errors in misidentiﬁcation of stream origin ﬁsh in
situations where estimates of ratio 23 were close to 0.70. Therefore, any steelhead with a
ratio 23 value 3 0.80 that also had a stream age of 2 or greater was classiﬁed as being of
wild origin. To test whether the observed and expected numbers of hatchery steelhead
present in each river system differed signiﬁcantly based on the proportions of steelhead
stocked into Lake Michigan, the expected number of steelhead was based on smolt-
equivalent estimates to account for differential survival juveniles stocked at different size,

age, and location into the Lake Michigan basin.

Genetic analysis
Fin clips from each of the four hatchery strains currently stocked around the basin
were obtained for genetic analysis and served as baseline. Small samples for genetic

analysis were taken from the caudal ﬁn of adult steelhead returning to each of four rivers

77

in Michigan. Only individuals with no hatchery identifying marks were sampled for
genetic analysis and SPA. Samples were preserved in 95% ethanol, or were dried in
scale envelopes. DNA was extracted using PurGene® (Gentra Inc.) protocols.
Microsatellite loci used for analysis were Ogola and Ogo4 (Olsen et al. 1998), OneulO
and Oneull (Scribner et al. 1996), Omy77 (Monis et a1. 1996), Otsl (Banks et a1. 1999),
and OtleO (Nelson et a1. 1998). PCR reactions were conducted in 25 ul reaction
volumes using 100 ng DNA, 2 ul 10x PCR Buffer (0.1 M Tris-HCl, ph 8.3, 0.015 M
MgCl2, 0.5 M KCl, 0.1% gelatin, 0.1% NP-40, 0.1% Trition-X 100), 0.2mM dNTP’s, 0.6
uM ﬂuorescently labeled forward primer, 0.6 uM unlabeled reverse primer, and 0.3 Units
Taq Polymerase. PCR reactions for Ogo4 were conducted in 25 ul volumes, following
the above protocol with an additional 2.5 mM MgCl2, and primer concentrations were 0.5
uM. PCR reactions for OtleO were conducted in 25 ul volumes, following the above
protocol with an additional 12.5 mM MgCl2, with primer concentrations of 0.5 uM. PCR
reactions for all loci utilized an initial denaturing step at 94° C for 2 minutes, followed by
30 cycles of 94° C for 1 minute, annealing temperature for one minute, and extension at
72° C for 1 minute, and a ﬁnal extension period of 2 minutes, 30 seconds. Annealing
temperatures for Ogola, Ogo4, Omy77, OneulO, Oneul 1, Otsl and OtleO were 56° C,
54° C, 54° C, 52° C, 62° C, 54° C, and 58° C respectively. Non-denaturing 6%
acrylamide gels were used for electrophoresis. Genotypes were visualized on a Hitachi

FM-BIO II scanner. Molecular weight standards and individuals of known genotype

were run on each gel to standardize scoring.

Statistical analysis

78

Pairwise estimates of degree of inter-strain differentiation in allele frequency
(F St) were determined using F STAT (ver. 2.9.3.2), and nominal alpha levels were
adjusted for multiple comparisons using Bonferroni corrections (Rice 1989). Individuals
were assigned to strain of origin using likelihood-based assignment tests (Rannala and
Mountain 1997). Estimates of assignment accuracy were based on resampling of the
strain baseline samples using the “leave-one-out” technique (Efron 1983). Estimates of
statistical conﬁdence in individual assignment decisions were based on posterior

probabilities (Pritchard et al. 2000; Blanchong et a1. 2002).

Determination of hatchery contribution

We determined the percentage of adult spawners of wild origin from each river
based on clip observations by summing the total number of clipped hatchery ﬁsh
collected in each river and dividing by the number of ﬁsh collected. Strain totals for the
genetics/SPA method were based on individuals without clips. Conﬁdence intervals for
wild contributions for each methodology were based on 95% conﬁdence intervals (Sokal

and Rohlf 1995)

1.96”“ ——p(l'p)
n

where p is the proportion of wild ﬁsh and n is the number of ﬁsh sampled. Direct (clip)
and indirect (SPA and genetic) methods of hatchery strain identiﬁcation were combined
by summing the numbers of individuals assigned to each of the hatchery strains based on
both methods. Estimates of hatchery strain-speciﬁc contributions derived by genetics and
SPA for the Little Manistee were extrapolated when combined with clip observations as

only a subsample of the fall and spring runs were analyzed with genetic methods.

79

Expected strain-speciﬁc counts of stocked ﬁsh were based on the yearly average
of numbers stocked strain into Lake Michigan between 1993 and 1997 for both the Lake
Michigan basin (Table 11a) and each river (Table 11b). The yearly averages of the
numbers stocked were adjusted for age and size- speciﬁc survival to calculate numbers of
smolt-equivalents (Rand et a1. 1993). The Ganaraska and Chambers Creek strains are not
stocked into Michigan rivers; therefore individuals belonging to these strains would not
be expected to be found in any of the rivers we examined for this analysis. Comparisons
were only made between the observed and expected numbers based on stream-speciﬁc
and strain-speciﬁc stocking for Michigan strain individuals found in the Little Manistee
River, and Michigan and Skamania strain individuals found in the Pere Marquette, Bear
Creek, and Platte rivers.

Three tests were conducted to determine how observed hatchery strain
composition compared to expected counts based on levels of stocking. The numbers of
hatchery individuals by strain were ﬁrst compared among rivers to test for differences in
proportional abundance by river and by strain. Fisher’s exact tests were performed to test
among-stream heterogeneity using the SAS software (SAS Institute). Second, observed
counts of hatchery steelhead by strain were compared (chi-square test) to expected
numbers based on the proportion that each strain was stocked into the Lake Michigan
basin. Finally, the observed counts of hatchery steelhead by strain were compared using
chi-square tests to expected numbers based on the yearly average number of steelhead

stocked into each river examined.

80

RESULTS
Stocking history

Approximately 1.8 million juvenile steelhead were stocked annually between
1993 and 1997 into the Lake Michigan basin (Table l 1). Adjustments to the total number
of juveniles stocked into the basin according to age, size, and location of stocking (smolt
equivalents) reduced the annual number of juvenile steelhead stocked to 640,517 (Table
11). The proportion of j uvenile steelhead marked prior to release from a hatchery varied
by state for the time period between 1993 and 1997. Beginning in 1995, Michigan began
clipping all hatchery origin individuals. Illinois also marks 100% of juvenile steelhead
prior to release from hatcheries (Table 11). Wisconsin marks approximately 33% of the
hatchery juveniles, and Indiana marks approximately 15% of hatchery juveniles.

Of the four rivers examined in this study, only the Platte River was not stocked
between 1993 and 1997 (Table 11). The Little Manistee River was stocked with a small
number of individuals of the Michigan strain. Portions of the Pere Marquette River and
Manistee River drainages (which Bear Creek is a tributary) were stocked large numbers
of both Michigan and Skamania hatchery strains (Table 11). However, Bear Creek is not
directly stocked, and only one tributary of the Pere Marquette (Ruby Creek, located near

the mouth) is stocked.

Genetic identiﬁcation of hatchery strains
Signiﬁcant differences in allele frequency were observed among the four hatchery

strains of steelhead stocked in the Lake Michigan basin (P<0.05; Table 12).

Differentiation between the Skamania strains produced by Indiana and Wisconsin was

81

low and
contribu
contribu
differen'
F550. 1
Chambe
dilieren

Classiﬁt

ESIlmQ;

steelhe
SAmple
if €Xtr;
Spawn
Platte
the Pe
Plane

creek

ESlim

low and not signiﬁcant (P>0.05; data not shown). Accordingly, estimates of
contributions of Skamania strain steelhead based on genotypic data represent cumulative
contributions from both states. Inter-strain comparisons indicated that the greatest
differentiation was between the Michigan and Skamania hatchery strains (mean

F 51:0.127; Table 12). The smallest difference in allele frequencies was between the
Chambers Creek and Skamania strains (mean FST=0.045; P<0.05; Table 12). Levels of
differentiation among strains were sufﬁciently large to provide accurate individual

classiﬁcation for each strain (range 98.2-86.7%; Table 13).

Estimates of strain-specific contribution based an observation

Direct estimates of the contribution of individual hatchery strains to adult
steelhead spawning runs ﬁ'om four Michigan rivers ranged from 1% to 15% of the
samples collected from each spawning run. Total contributions across all hatchery strains
if extrapolated across an entire spawning rim varied between 9% of the fall and spring
spawning runs in the Pere Marquette River to 27% of the spring spawning run in the
Platte River (Table 14). Clips identifying each of four hatchery strains were observed in
the Pere Marquette River and Bear Creek (Table 14). The Little Manistee River and
Platte River both contained individuals from the Michigan, Skamania, and Chambers

Creek hatchery strains.

Estimates of strain-speciﬁc contribution based on genotype and SPA

We estimated that of the non-clipped ﬁsh collected, contributions of hatchery

strains ranged from 3% to 26% of the total spawning nm (Table 14). Hatchery

82

CC

contribution of non-clipped ﬁsh for all four strains ranged from 3% in the Pere Marquette
River to 26% in the Little Manistee River (Table 14). The Michigan strain was present in
all four rivers. Individuals of each of the four hatchery strains were found in the Little
Manistee River (Table 14). Individuals of the Michigan and Ganaraska strains were

found in the Pere Marquette River and Bear Creek (Table 14).

Combined direct and indirect estimates of strain-speciﬁc contribution

To estimate the total hatchery contribution to spawning runs for the four rivers, a
combination of hatchery marks observations and SPA/ genetics analysis were used to
examine the composition of two separate groups of steelhead (clipped and not clipped).
Individual strain contributions to spawning runs ranged from 0% to 23% (Table 14).
With the exception of the Platte River, all of the four hatchery strains were present in the
rivers examined. Total hatchery contribution (combined over all strains) ranged from
31% in the Little Manistee River to 13% in the Pere Marquette River (Table 14). In Bear
Creek, hatchery individuals contributed 27% to the spawning run (Table 14), and in
Platte River, hatchery individuals contributed 29% to the spawning run (Table 14).
Estimates of Michigan strain hatchery contribution in the Pere Marquette River and the
Platte River were similar using both clip observation and genetics/SPA (Table 14).
Individuals of Skamania strain origin were present in all rivers as estimated by clip
observations (Table 14). Genetics/SPA analysis identiﬁed Skamania strain only in the
Little Manistee River, in a greater abundance to the fall and spring spawning runs than

estimated by clip observations (4% versus 1% respectively; Table 14).

83

Observed numbers of steelhead with the Ganaraska strain clip represented 1% of
the spawning runs in the Pere Marquette River and Bear Creek (Table 14). Genetic/SPA
identiﬁed individuals of the Ganaraska strain in the Pere Marquette River (3% of the total
run; Table 14), Little Manistee River (3% of the total run; Table 14), and Bear Creek (5%
of the total run; Table 14) spawning runs. The Chambers Creek strain contributed to the
spawning runs of Bear Creek and Platte River (9% and 15% respectively; Table 14)
estimated by clip observations whereas this strain was not present in the genetic/SPA
analysis of the non-clipped individuals. Estimated contributions by genetic/ SPA analysis
did not detect the Chambers Creek strain in the Pere Marquette River, compared to
estimates of contribution based on clip observations (4%; Table 14). Estimates of
Chambers Creek strain contribution to the Little Manistee River were comparable for the
direct and indirect identiﬁcation methods (1%; Table 14).

Of the rivers examined in this study, only the Platte River was not stocked with
hatchery individuals, and the Little Manistee River was stocked with very low numbers
of individuals. However, sizable portions of hatchery-derived adults were observed
during spawning. This includes appreciable numbers of hatchery strains stocked in other
states around the basin. The Little Manistee River had the highest contribution of
hatchery steelhead of the two non-stocked rivers (31%; Table 14). The Platte River also

has a large hatchery component (29%; Table 14).

Straying of hatchery strains into Michigan rivers

Comparisons of the observed and expected contributions of hatchery-ori gin

individuals to the spawning runs of the four Michigan rivers provided evidence of

84

ho:

lhc'

51¢

51

homing in Lake Michigan steelhead. We found signiﬁcant differences among rivers in
the strain composition of hatchery-origin individuals (x2=181.4; df=9; P<0.0001). Chi-
square tests between the observed and expected contribution of each hatchery strain
pooled over all rivers indicated a signiﬁcant difference (xz= 1249.2; df =3; P<0.001) in
the number of hatchery ﬁsh observed based on the proportion of smolt-equivalents
stocked into the Lake Michigan basin. River-speciﬁc chi-square tests indicated
signiﬁcant difference in the observed and expected contribution of hatchery-ori gin
steelhead relative to the proportions stocked into Lake Michigan for the Pere Marquette
River (x2=5.3; df =1; P<0.05), Little Manistee River (x2= 1015.7; df =3; P<0.001), Bear
Creek (x2= 5.0; df =1; P<0.05), and Platte River (x2= 10.5; df =1; P<0.05). Due to the
low contribution of Ganaraska and Chambers Creek to the total percentage of stocked
smolt-stage steelhead into Lake Michigan, and the low number of hatchery origin
individuals observed in Pere Marquette River, Bear Creek, and Platte River, no
individuals from these two strains were expected in these rivers. Therefore chi-square
comparisons were not made for three of the four rivers between the observed and
expected numbers of Chambers Creek and Ganaraska strain hatchery ﬁsh relative to what
is stocked into Lake Michigan.

The observed number of Michigan and Skamania strain individuals differed from
the number of Michigan and Skamania strain individuals stocked into each river. There
was a signiﬁcant difference in the proportions observed in the spawning run compared to
the proportions stocked in of the Pere Marquette River (x2= 5.9; df =1; P<0.05), but not
in Bear Creek. As the Michigan and Skamania strains are the only hatchery steelhead

strains stocked into the Pere Marquette and Bear Creek drainages, statistical comparisons

85

were not made for the Chambers Creek or Ganaraska strains despite observations of

individuals with clips speciﬁc to those strains.

DISCUSSION

Comparisons of estimated straying rates derived from clip observations with rates
based on genetic and SPA methods indicated that potential biases exist when clip data
alone are used to estimate hatchery contributions to steelhead spawning runs. We
documented non-uniformity in proportions of steelhead juveniles who received clips
among states around Lake Michigan (Table 11). As such, assessment information based
solely on clips may misrepresent strain-speciﬁc estimates of abundance or contribution to
ﬁsheries.

The use of hatchery clips (e.g., ﬁn clips, maxillary clips, and or a combination of
marks) that are easily confused with injuries such as those resulting from hooking may
bias estimates by overestimating the contribution of those strains that use those clips.

The Wisconsin Skamania and Chambers Creek hatchery strains are identiﬁed by clipped
left or right maxillae (Burzynski 1999). Estimates for the left maxillae clipped Chambers
Creek strain were consistently higher than the Ganaraska strain, which can be identiﬁed
by having both maxillae clipped (sometimes additional clips are used in combination with
the maxillae clips; Burzynski 1999). Estimates of Skamania strain contribution primarily
consist of steelhead from Wisconsin, although Indiana and Michigan both stock
Skamania and clip a portion of the stocked ﬁsh (Table 11). Because only one-third of
Wisconsin’s hatchery ﬁsh are clipped prior to release, estimates of the non-clipped

steelhead are expected to be approximately two times greater than estimates based on clip

86

observations alone. However, analysis of non-clipped individuals estimated smaller
contributions of the single maxillae clipped steelhead. Assuming clipping does not affect
rates of straying, a portion of single maxillae clipped individuals are not representative of
the Chambers Creek or Skamania strains, but rather represent river origin individuals
with booking injuries. Support for the bias in estimates for the Chambers Creek strain
due to clip and injury confusion was found in the observation of multiple individuals with
the single maxillae clip in Bear Creek and Platte River (Table 14), but the lack of support
for these observations using the genetics and SPA techniques.

Straying (or returning to spawn in a river other than the natal river) is an adaptive
life history characteristic of sahnonids, including steelhead. Therefore, the presence of
individuals of hatchery origin belonging to strains not stocked into a particular system
was expected. However, the extent of straying was not expected (Table 14). It has been
observed that hatchery individuals may stray at higher rates than found in naturally
occurring populations (Waples 1991b). The Little Manistee River has been stocked with
very low numbers of individuals and Platte River is not currently stocked. However, the
mouths of these drainages are in close geographic proximity to heavily stocked systems,
resulting in an increased potential for straying of ﬁsh during fall and spring adult
spawning migrations. The contribution of steelhead straying from Wisconsin,
speciﬁcally the Chambers Creek and Ganaraska strains (the Wisconsin and Indiana
Skamania strains are not genetically distinct in this analysis) differ between direct and
indirect observations.

We found signiﬁcant differences among rivers in the strain-speciﬁc composition

of hatchery-ori gin steelhead. Among-river heterogeneity in strain contribution to

87

spawning runs indicates adult hatchery contribution was not randomly distributed
between each river system. Additionally, signiﬁcant differences in the observed and
expected numbers of individuals of hatchery-origin to spawning runs for all rivers
indicated non-uniform distribution or survival of hatchery strains. Proportions of
hatchery-origin individuals found in the Little Manistee River, Bear Creek, and Platte
River were signiﬁcantly different from proportions of each strain stocked into Lake
Michigan. However the proportion of hatchery-origin individuals did not differ in the
Pere Marquette River in comparison to the proportion stocked into Lake Michigan. This
result may be biased by the relatively small proportion of the Chambers Creek and
Ganaraska strains to the Lake Michigan hatchery-origin steelhead population, and the
relatively low observed number of hatchery-origin individuals in the Pere Marquette
River.

There are positive and negative implications associated with ﬁndings of high
proportions of hatchery ﬁsh in Michigan spawning runs derived from stocking by other
states. On one hand, Michigan experiences an “embarrassment of riches”. If strain
composition of spawning adults is consistent with strain-contributions to the creel, then
Michigan anglers are reaping the beneﬁts of resources expended by other states.
However, introgression between hatchery strains will lead to the breakdown of among-
strain genetic differences, potentially resulting in the loss of the differences in heritable
life history traits (run-timing) that currently exist between the strains and which provide
managers with greater management options.

The large contribution of certain hatchery strains to the spawning runs in

Michigan increased the potential for introgression among hatchery strains and between

88

naturalized and hatchery populations. Reisenbichler and Phelps (1989) hypothesized that
reduction in among stream genetic diversity of Washington steelhead populations was
due to the widespread stocking of hatchery steelhead and resulting introgression with the
native populations. Introgression between the naturalized populations in Michigan and
the Michigan strain has been proposed to be primary cause for lack of signiﬁcant genetic
differentiation among drainages (Bartron and Scribner in press). Of particular concern is
the contribution of hatchery origin individuals to the fall and spring spawning runs of the
Little Manistee River. The Little Manistee River serves as the source for gametes for

. propagation of the Michigan strain (Seelbach 1987a).

When accounting for differential survival of the various ages and stages of ﬁsh
stocked into the Lake Michigan basin, we found that only approximately 39% (Table 11)
of the stocked juveniles survive to the smolt stage. Due to the small size and ﬁngerling
stage at stocking, low survival rates indicate that hatchery stocking programs are
producing more steelhead for stocking purposes than needed, and could better devote
resources to producing few larger, older juveniles that have better survival rates (Rand et
a1. 1993; Seelbach 1987a). Concerted management efforts to utilize hatchery marks that
are not easily misidentiﬁed with injuries, and to mark all hatchery-produced individuals
may reduce bias in hatchery-mark derived information. Additionally, despite the large
number of steelhead stocked into the Lake Michigan basin, the contribution of natural
recruitment to the spawning runs for the four Michigan rivers examined was larger than
expected. Further quantiﬁcation of the amount of natural reproduction may lead to a
reduction in stocking levels while the total population would be maintained at current

levels.

89

As varying portions of hatchery-produced steelhead are marked each year prior to
release, estimates of hatchery ﬁsh in spawning runs based solely on counts of clipped
individuals may not adequately estimate hatchery contribution. Using additional
techniques to estimate hatchery contribution, such as SPA to determine river or hatchery
origin and genetic identiﬁcation of hatchery strains provide a better representation of the

contribution of hatchery individuals.

90

ooo.o

 

 

 

 

 

 

 

do _ .o ooo.o omod owod m mod mid mood ooNd mmod ooo.o Sod wood ovod Nwod
vmod owi .o Nwod Cod Vmod Sod ooo.o imod mmo. o oNod omod mmod oiod oNod Nmod
ooo.o ooo. d ooo. o ooo.o ooo.o ooo.o ooo.o .mod ooo. o ooo.o ooo.o ooo.o ooo.o ooo.o ooo.o
mwod ooo. o ooo o Nomd mmNd oVNd omNd mmmd mmN. o momd ooNd dNNd ohNd ooNd mom. o
Cod ooo. o ooo. o oood ooo.o oood Kod ooo.o oood mood oo_ .o Edd iNod oNod ooo. o
vi .o ooo. o ooo.o vmod mood oood ooo.o imod ooo.o oNod omod mood _No.o ooo.o So. o
on on 3 c: on mm N. i 2 oi oi voi mv wN av
Nvod ooo.o ooo.o wood wiod ooo.o ooo. o imo. o omo. o mood ooo.o oood ooo.o ooo.o ooo.o
mood emod iod ooo.o ooo.o oood ooo. o ooo. o ooo. o oNod ooo.o ooo.o ooo.o ooo.o ooo. o
ooo.o ooo.o god ooo.o ooo.o ooo.o ooo.o ooo. o ooo. o mood ooo.o mood ooo.o ooo.o ooo. o
N iod vmod N2 .o m _ od NNod ooo.o ooo.o .mod omod dmod omod mNod oNod oNod ooo.o
m mod wnod :. i .o oNod NNod dood mid _ mod omod omod ooo.o mo _ .o wwod owod Sod
52d omNd SNd did woNd oid mid mN_.o m_Nd mid ooid mwod ooo. o of .o VNNd
ooo.o Nmod Nmod ooo.o ooo.o ooo.o ooo.o ooo.o ooo. o o_od ooo.o ooo.o ooo. o ooo. o ooo.o
ooo.o oood ooo.o ooo.o ooo.o hood ooo.o ooo.o ooo. o ooo.o ooo.o ooo.o ooo. o ooo.o ooo.o
ooo.o ooo.o ooo.o ooo.o ooo.o ooo.o ooo.o ooo.o ooo. o ooo.o ooo.o mood ooo. o ooo.o ooo.o
mnod oood ooo.o Nmod mvod mmod :od mood omod omod ooo.o mNod mmo. o ooid God
omod oood ooo.o wood ooo.o iod ooo.o ooo.o ooo.o mood ooo.o ooo.o oiod ooo.o oNod
oo. .o hid mmid momd ommd ovmd iNd imNd omNd momd oomd omNd dNNd omNd omNd
wmmd nmmd mid onmd wvmd mnmd ded wmvd mdmd mnmd ommd mnvd odvd oomd mde
oo mm :— m: om Nu m i i oo_ o_ oo_ Nm N 3.
S: .o Cod d—od :md oomd Nde i Nd ded mm _ .o thd omvd mmNd :md mvmd immd
SNd owid Sid VmNd mde mde mid dmN. o mom. o mNNd omNd SNd Nde dmNd med
ooo.o ooo.o ooo.o ooo.o ooo.o hood ooo.o ooo. o ooo. o ooo.o ooo.o ooo.o oood ooo.o ooo.o
oomd and ind :vd oomd mmmd mvod N3. o oom. o movd oomd owed ommd mivd m :d
ooo.o oood ooo.o wiod ooo.o oNod ooo.o ooo. o ooo. o ovod ooo.o dNod ooo.o oood ooo.o
ooo.o oood ooo.o ooo.o ooo.o ooo.o ooo.o oood ooo. o mood ooo.o ooo.o ooo.o ooo.o ooo.o
oo an moi i _ mm 2. N. 2 mi ioi o. we Nm mN S.
8.3850 3080 SEE dmddi Ymmdi Qmoow Ymmdi dmddi Ymmdi dmdd— Ymmd— 9on medi dmdd— immdi
Eon—EEO -96 ovarian: gauge; 012:? 3.2682 oBWEaE gum
235 Sam
ggm rep—8a:
conﬂsnom

 

oo—
vo—
No—
oo—
mo
Nm

2 Ease

Ni
oi
wmi
omi
vmi
Nmi
om—
wNi
0N—
N—
N—
oN_
w:

Si
mi
_i
am—
N—
m:

Ewe

z 2an

.383

 

dog 83.— an woos—88 one? $316365 3058.5 .583 Conan: some 98 Admdd— 23 med:
cocoa 08.5 some $53 3388 macaw—soon ©3538 5&322 5m 5.. moo— oEEameoE in 05 do :08 3 885.63% £3? 28 :6 85m £95m ._ x8593

91

N._o.o mNNd iood m io.o oood N.No.o _N.od ooo.o ooo.o mood ooo.o NNod NNod iNod ooo.o oo—
N.mN.o o_N.o _ww.o omN. o wNN. o N.w_d ~N.o. o NNN. o oom. o moN. o wwwd omNd ooNd oid wNNd mo
ooo.o ooo.o ooo.o ooo. o ooo. o ooo.o ooo o ooo. o ooo. o ooo. o ooo.o ooo.o ooo.o ooo.o ooo.o oo
mNod eNd Nmid owe. o mwo. o mmod N.wN.. o NNN.. o ooN.. o o_N.. o own. o wiod oood owod woN.d wo
Nwod moNd :md mmod mwod mN.o.o ooo. o ooo. o ooo. o owod ooo. o ooo.o wwo. o mwod iwod No
Ntiod ooo.o ooo.o ooo.o oood m _od _N.o. o No. o ooo. o o_od ooo.o ooo.o ooo. o mwod oiod od
Nwod w _o.o ooo.o m mod w _o.o N.wod ooo. o wNod ooo.o oNo.o ooo.o NNod NNo. o iNod iNod om
do N.m oo— m: hm N. N. mi 2 oo— o— ow ow wN ow Z oom.:NQ

ooo.o ooo.o ooo.o wood wwod ooo.o ooo. o ooo.o ooo.o ooo.o ooo.o ooo.o ooo.o ooo.o ooo.o N.m _
WNod mNod ooo. o dmod mmod oNo.o ooo. o imo. o Nbo. o mmod ooo. o mmod omod Nwo. o oiod m m _
mmod mmi .o ooo. o mood ooo.o oNo.o ooo.o ooo. o ooo. o :od ooo. o io.o oiod ooo. o ooo.o mm—
omNd N._od ooo.o dmmd imd ond hmmd imN. o SN. o did :— d wNNd mNNd mN.m. o wwmd _m_
oom.o _wm.o wind dmwd wmwd ond oom.o wmwd oom.o omwd NNN.d iowd oowd wmwd owwd oN—
Noid Nid wwid o: .o did ooNd mid omNd mmid N.mNd Nbid mid ooid woid Noid N.N_
ooNd Nid woo. o woo. o wio. o oNo.o ooo. o ooo.o mmod mwod ooo.o God oNo.o iNod imod mNi
ooo.o ooo.o ooo. o ooo. o ooo. o ooo.o ooo.o ooo.o ooo.o ooo.o ooo.o ooo.o oNo.o ooo.o ooo.o _N_
oo oo moi m: N.m N. N. 3 2 no o— boi in N ow Z 3130

ooo. o ooo.o ooo.o m_od oood m 3. o _N.od ooo.o ooo.o wood ooo.o io.o oiod oNo.o o_od Ni
ooo. o ooo.o ooo.o ooo.o oood ooo. o ooo.o ooo.o ooo.o o_od ooo.o ooo.o oiod ooo.o ooo.o cm _
woo. o owod ooo.o oood omod oNo. o ooo.o ooo.o mmod owod owo. o ooo.o iNod owod oNo.o wm—
ooo. o N._od 2 id omod N.No.o N.Nod ooo.o mood ooo.o oNo.o ooo. o oNo.o _Nod owod oNo.o Nm _
woo. o wmid wowd owod omod NNid _N.od mN_d ioNd mid oo_ d _o_d Nmod owod Nood om—
mood wmod wm _ .o mood wmid wmi d _N.od ooo. o Nbod omo. o omo. o woo. o woo. o ooo. o Nood dN—
oood N._o.o ooo. o ooo.o ooo. o ooo. o ooo.o ooo. o ooo.o ooo. o ooo. o ooo. o ooo. o ooo. o ooo.o oN—
oood wid iwo. o ooo.o ooo. o ooo.o ooo.o ooo.o ooo.o ooo.o ooo.o ooo. o ooo.o ooo.o ooo.o wN_
owod ooo.o ooo.o ooo.o owo. o moo. o mi. o oood ooo.o mwo. o oo— .o wmod Nwo. o oo—d mod N—
ooo.o oNNd did wood ooo.o ooo. o ooo. o ooo.o ooo.o ooo. o ooo. o wNod ooo. o owod ooo.o oN—
wmo. o wood ooo. o Nmod mmod N.wo. o NN.o. o _mod mmod ooo. o omo. o NrN.o.o moo. o owo. o Nwod o:
ooo. o wood ooo. o m_od ooo.o m 5d ooo. o imod mmod m _o.o ooo. o wNod ooo. o ooo. o ooo.o i —
N.N~ .o owod ooo.o Nwod owod mm—d Nwo. o ww—d hood mood on i d oN_d on i .o oN_ .o Nwod N:
mm—d Cod ooo.o owod mwod mmod ooo.o imod oo—d oi. o oo— d mwod mid ooo.o _wod o:
wood N._od ooo.o wood ooo.o ooo. o ooo.o ooo. o ooo.o ooo. o ooo. o io.o ooo.o oNo. o oood wo—

 

.G....ae _ 528.3

ooo.o
mwN..o
who. o
ooo. o
_N_ .o
ooo. o

mm

ooo.o
oom.o
N.— _d
mNo. o
ooo. o
oood

mw

owod
mN.N..o
mN-d
owod
ooo. o
ooo. o

moi

ooo.o
Nwwd
mwod
ooo.o
mmo. o
ooo. o

m:

oood
owwd
Sod
ooo.o
wwod
ooo.o

Nh

oood
iowd
iwod
ooo.o
woo. o
ooo. o

mN.

oood
hmwd
mid
ooo.o
ooo. o
ooo. o

N.

ooo.o
io.o
omo. o
ooo. o
No. o
ooo. o

w—

mmod
_owd
mmo. o
ooo. o
mm_ o
ooo. o

m—

ooo.o
Nood
wwo. o
ooo. o
wwo. o
ooo. o

No—

ooo.o
oood
ooo. o
ooo. o
oo_ .o
ooo.o

o—

io.o
mnwd
woo. o
ooo. o
Nwo. o
ooo. o

oo

ooo.o
mowd
wwo. o
ooo. o
wwo. o
ooo. o

Nm

ooo.o
owwd
owo. o
ooo. o
owo. o
ooo. o

N

ooo.o
nowd
imod
ooo.o
Nwod
ooo.o

ow

 

.618 _ 52 <

mm
3
E.
mN.
N.m
mm
2 2:an

93

APPENDIX 11

TABLES AND FIGURES

94

Table 1. Stocking histories for each of the ten naturalized populations exarrrined.

Numbers of individuals stocked represents the average number of juvenile steelhead
stocked into each river between 1993-1997. The estimates of smolt-equivalents stocked
are based on survival estimates for different size, age, and location of stocking described
by Rand et a1. (1993). Strains used for stocking were Michigan strain (M) and Skamania

 

 

 

slain (S).

Total Smolt- Average smolt- Strains
River # stocked equivalent equivaylentper year stocﬂ
Thompson 0 0 0 -
Black 0 0 0 -
Platte 0 0 0 -
Betsie 247,973 223,176 44,635 M
Manistee 357,739 321,965 64,393 M, S
Little Manistee 500 450 90 M
Pere Marquette 70,242 63,218 12,644 M, S
White 112,100 100,890 20,178 M
Muskegon 292,683 263,415 52,683 M
St. Joseph 476.787 429.108 85.822 M. S

95

0

 

 
 

 

 

oom.o wemd mth wad Edd oidd iodd ond wdmd Ngd oemd mmwd eewd :
ooo.o oom.o eoN.d wwnd Edd N.mm.o mmod wmod oNo.o NwN.d mde onN.d de a:
emd med ww.m wmd o.: m.: dd_ dmd w.: id m_.o_ wwd Nod h<
m_ m_ e N.— : ei Ni 2 mi mi 2 e 2 <
oe oe moi e: Ni mw mi ei mN mu moi ooi imi Z
ESQ
omed meN..o miwd Nohd ewN.d imhd ewN.d omnd mend :Nd mwN.d me.d oded um
oom.o mmnd womd mmod mmwd mamd mmnd Nemd dend NNNd amed owed Eed om
eme mN.e nwd mew com doe mme awe emm Now N.w.m mum nmm J.
m d m m m N. h N e N. d : oi <
oe oe N: 5: Ni Nw mi ei mN NN. moi ooi mmi 2
was
mmmd ded N._Nd eeed wmed N.wed wN.ed oN.ed eeed N.med oN.ed owed Nmed um
mmwd neNd mNNd Nied mmmd wded mmnd med omnd oNN.d mmed mwed Need om
3N em.N om.N emm com mNm com com com iwm omm eN..m Nmm .<
m m m w m w m m m w w m w <
oe oe N: e: Ni mw mi 2 mN mm moi ioi wmi Z
Somme
<0 00 vim 2..— EN. 4m 8 m3 mm DE .3 SE 7:2 22?.
magmas—3 O BNSQNBNZ «390%

anaemia—3 O nor—Qua:

 

 

aged <0 2... .aaaaala

$3828 00 A583 35:8me Mm A383 gwﬁoééuga 863.32 25.: 55 36:8 3 Scene 93 mcﬁbm Concomm .3020
8882: E 632 38.8 .E Nei :38. do a .95 33% E» .35 ease mm .85 Eggs: :2 Area 33E
.5 Aug?“ 033332 0.5% SE fuzz 8dean 22 “mac—EN as deemed Pa «does—anon down—«5:2 .3on gem—noon began:

05 E does—o5 mm x .585 53:32 05 No aoemwmooa So com: mm seem—30o 83:32 235 05 omswoom .62 =a No.6 wow—>95

mm 956508 mag—595:5 m 50.: 53523.4. .62 =~ 85 82?» SSE one «:02 .3 seem—soon :80 SN 3.: badmxmeao:

usage Ba 65 somewhere: 8288 do asset 0:21.. As 252 ea ea:a No 25:: do 3a 295
. m .N 2%...

96

 

eved
need
we...‘

de
bend
Nemd
5N6
we
ded
wemd
dNN
me
eEd
:wd
woe

de

dend
vad
owe.

me
wvmd
oom.o
evN
mv

_ Sud
ddvd
emN
de
meed
mwed
dwé

de

e eed
need
36

ma
Smd
mgd
dim
E:
Red
e emd
mud
a _
dvvd
hmvd
mdN

d:

mmed
NNed
ewe

m ~ ~
Emd
oom.o

wvd
e~ _
emwd
ﬁnd
and

e:
eeed
dded

mad

e:

med
5 Sad
ode

Nd
dddd
dddd

do;

9
owed
mwed

dd.m

Nd
dbed
mwed

dd.e

~—

eNed
hmed
deé
e

:V

eNNd
ead
mwN
v

3‘

dVNd
Nde
dud.

mm
need
ehed

Ed

3.

emed
ddvd
deé
e

e _

vad
ueNd
3N
m

e ~

eHed
mmed
dd.m

ed
wdnd
heed
die

e~

dvvd
eeed
med.
v

w—

deﬁd
be_d
ee.~
m

w_

eemd
ehmd
exam

e~
wwed
de
end

e~

deed
wdnd
dwé
e

vm

hwmd
dwmd
nwd

e N
dmvd
dvvd

dd.N
em
deed
wdnd
mmé

vm

mmed
dwﬁd
hmé
e

en

Ndmd
ave—d
_e.N

mm
ﬁmvd
_evd

mud

Z.
ewed
owed

mmé

eh

eeed
ee e .d
woe
e

do

Smd
dad
KN.

on
wevd
:vd
bwd
X:
mwed
edbd
Nvé

em

mevd
devd
wvd
e

ddﬁ

mwﬁd
ead
vim
m

we

~evd
eNed
odd
e

3

Nhed
edhd
eeé

ed

~eed
heed
ddé
e

v2

Numd
hemd
dwd
v

3:

ch
dwvd
and
m

w:

End
dded
dve
w

ee—

0

z<€mm

O

ddNEO

om

 

9:53 m 23..

97

 

io.o mood io.o wwod mmod mmod mwod mwod mwod- wmod wmod wood Rod
and the ommd Rho and mend mood mmmd wood mwmd and ﬁnd wood
mwmd 30o m—md and oom.o ovﬁo 39o mmmd Sod oom.o mwmd Nnmd hwmd
and «We cad wmé cwé cwé end 96 mwé 51V mud $6 $6
$6 and Rd cwd cwé ado hmw cod :6 36 oo.“ 96 Nmﬁ
ow hm we 0: 3 3 a C vm mm 8 3 mg
.83 an .B\§m>~

O

31:13:;

z<<“

 

.5288 N 2an

98

Table 3. Hierarchical partitioning of genetic variation (Weir 1996) within and among the
naturalized populations and hatchery strains (populations) is provided for each locus and
mean over seven loci. Naturalized populations include the Manistee River, Pere
Marquette River, Platte River, Muskegon River, Betsie River, White River, St. Joseph
River, Black River, and Thompson Creek. Hatchery strains include Little Manistee

River/Michigan strain, Skamania strain, Chambers Creek strain, and Ganaraska strain.

 

 

(o (9..)
(F) Among Among
Alleles individuals populations ((95)
within within within Among
Log. individuals populations gromgs groups
Ogol a
Naturalized populations 0.019 0.020 -0.001
Hatchery strains 0.236 0.099 0.152
Combined groups 0.100
Ogo4
Naturalized populations 0.059 0.056 0.003
Hatchery strains 0.089 0.011 0.078
Combined groups 0.037
0my7 7
Naturalized populations 0.067 0.062 0.005
Hatchery strains 0.178 0.060 0.126
Combined groups 0.062
One/11 0
Naturalized populations -0.012 -0.014 0.002
Hatchery strains 0.111 0.005 0.106
Combined groups 0.045
Onepl I
Naturalized populations -0.016 -0.018 0.002
Hatchery strains 0.122 0.122 0.000
Combined groups 0.003
Ots] 03
Naturalized populations 0.068 0.064 0.002
Hatchery strains 0.012 -0.008 0.021
Combined groups 0.019
Ots] 00
Naturalized populations 0.020 0.018 0.002
Hatchery strains 0.141 -0.032 0.168
Combined groups 0.115

99

Eble 3 (cont’dL

Mean for all loci
Naturalized populations
Hatchery strains
Combined groups

95 % Conﬁdence
Naturalized populations
Upper bounds
Lower bounds
Hatchery strains
Upper bounds
Lower bounds
Combined
Upper bounds
Lower bounds:

0.030
0.135

0.053
0.003

0.176
0.092

0.028
0.034

0.050
0.001

0.072
-0.020

100

0.003
0.105

0.004
0.001

0.137
0.063

0.060

0.084
0.032

awe 025 $8“ machomeom 355.53 memo: comics modvm .

 

 

 

ooo.o ooo.o ooo.o mmoo ooo.o :oo ooo.o ooo.o ooo.o ooo.o ooo.o ooo.o ﬁnesse
ooo.o ooo.o oNo.o ooo.o ooo.o mono ooo.o ooo.o ooo.o ooo oNo.o oueoocaou
«2o 3o 33 ooo.o “2o ooo.o ooo.o $3 $3 5o nausea

ooo.o ooo.o _oo.o ooo.o ooo.o ooo.o ooo.o ooo.o ooo.o .55.:

io.o ooo.o ooo.o ooo.o ooo.o ooo.o ooo.o ooo.o 385.5

. ooo.o ooo.o _oo.o ooo.o ooo.o ooo.o Boo x85

ooo.o ooo.o ooo.o ooo.o ooo.o ooo.o 2083.0.

ooo.o ooo.o ooo.o ooo.o ooo.o 323

ooo.o ooo.o ooo.o ooo.o 23$

_oo.o ooo.o ooo.o Swans:

ooo.o Nooo 23o

ooo.o on: 28

0o Mm s: E om a :3 mm :2 .3 2o 22 33m

 

 

.390 mm ﬂuid ooommcwE 235 of “Eva—oz: minim—ohm 30:22.3 oﬁwcoao ® aﬂoﬁmwd

:82 .890 mm Quid oBmEoE 2:5 05 wEonﬁv macro—omen Bum—«.53: mecca ® 35:3 :82 .voﬁozo: 08 303383
Amodvnc Boo—mama A62 :38 388 ® 525 5:33.: 223 E 3593 cows—snoadooﬁ mo mcoaoaﬁoo corian— .v 033.

101

Table 5. Estimates of the mean and variance number of offspring produced by males and
females for each mating treatment. The number of individual males and females (N)
Qed for each treatment is also provided.

 

 

 

Males Females
Treatment N mean gr. N mean vaL
1 (1:1) 10 19.4 34.0 10 19.4 34.0
2 (1:2, rep.) 10 17.8 144.0 10 17.8 25.3
3 (5:5) 10 16.2 233.3 10 16.2 16.6
4(10:10) 10 17.1 192.5 10 17.1 45.7
5 (1:2, mixed) 20 9.1 38.7 10 18.1 65.0
6 (1:2, seq.) 20 9.5 43.0 10 18.9 71.0

 

102

Table 6. Total number of males and females used for matings (N), and summary
measures of genetic diversity including the percent unrelated offspring, coancestry ((9),

and effective population size calculated using the number of males and females (Ne),

reproductive variance (Nev), and coancestry (Neg) for each of the six hatchery mating
strategies used.

 

% unrelated

 

 

Treatment N offspring 01 N; N913 N2 94
1 (1:1) 20 90.2% 0.0244 20 19.2 19.3
2 (l :2, rep.) 20 83.3% 0.0291 20 16.4 17.2
3 (5:5) 20 74.4% 0.0344 20 14.1 14.5
4 (10:10) 20 75.6% 0.0329 20 14.7 15.2
5 (1:2, mix) 30 88.7% 0.0225 26.7 21.8 22.2
6 (1:2, sea.) 30 88.7% 0.0226 26.7 21.7 22.1
1 Cockerham 1969

2 Wright 1931

3 Lande & Barrowclough 1987

4

Chesser et a1. 1993

103

Table 7. Number of juvenile Michigan strain steelhead annually stocked into Michigan
rivers of the Lake Michigan basin, estimates of the number of juveniles that would be
expected to survive to smolt stage, and percentage of the total number stocked based on
estimated survggl.

 

Smolt- Estimated

Yiar Number stocked equivalent1 percent survival
1978 data not available - -

1979 1,092,273 21,091 1.9
1980 1,325,177 19,942 1.5
1981 674,343 77,508 11.5
1982 1,091,154 15,493 1.4
Mean 1,045,737 33,509 4.1
1993 475,139 388,765 81.8
1994 532,688 439,297 82.3
1995 544,530 448,243 82.3
1996 527,329 426,517 80.9
1997 544,791 482,485 88.6
Min 524,895 437,061 83.2

 

l Smolt equivalents were calculated as described by Rand et a1. (1993). Estimated

survival to smolt stage was based on juvenile size and age at stocking, and stocking
location.

104

Table 8. Estimates of the mean (i 95% CI) number of juvenile steelhead stocked into
Michigan tributaries of Lake Michigan that were expected to survive to smolt (i.e.
number of smolt-equivalents), and the proportion of wild—ori gin adults in the spawning
runs of six rivers in Michigan estimated from two time periods prior to and following

changes in juvenile stocking practices.

 

 

1983-1984 1998-1999

Mean number Proportion Mean number Proportion
River stocked per year wild1 stogied per year wild2
Betsie 2,588 0.82 (i010) 44,635 0.65 (i018)
Manistee 826 0.88 (i008) 42,781 0.65 (i006)
Little Manistee 1,400 0.98 (i003) 90 0.67 ($0.00)
Pere Marquette 113 1.00 (i000) 9,252 0.87 (i001)
White 4,880 0.88 ($0.16) 20,178 0.55 (i017)
Muskegon 8,514 0.70 ($0.28) 52,683 0.21 (i004)
ﬂan 3.054 0.88 28,269 0.60

 

1Data from Seelbach & Whelan (1988)
2Data from Bartron et al. (in review)

105

Nowd ccwd Nth wad 3%.: ~56 ~36 _Nw.c :wd mwwd mmod v56 cgd wawd vowd 0:
m _ ad N _ 56 35.6 556 Had 53.6 536 wmad Swd Stud co; thd med 9.36 536 or
ones woos howé 3N6 2mg. N36 256 Sad «.36 m—mﬁ 32m N056 at; 03.x QVmH u<
2 Q o C 2 m— a 2 o— 2 Z 3 2 m. 9 <
am an 3 0: cm mu m m: 2 2: o— X: we mm 3 Z RAEO
$36 End 536 N25 whwd 5nd ownd omhd and wmnd cued hood mwod mend vmhd 0:
oom.o @556 N36 mwod . 3.06 Nth End mend hmwd owod oomd owed Vmod oom.o mwod 03

mod N2 .n mmod mucé mmmé Gmé 89m tad mwod mmhé 39m 309 howé mwhé mNmé h<
a a w w \t h m N. N. I v 2 h o v <

co mm :_ m: cm Nb N. 3 E 2: S 2: mm mm we 2 womb
mwmd mmmd mmmd hood :bd hwcd ocmd owod w 5d 506 2.06 wwod mwod ooo.o mmod om
mwvd :Nd ﬁnd 306 3nd 996 3 Nd vamd coed mwod ooo.o awed wmod and wand om
wmod ms _ .N NmON co~.m 2 ﬁn ammd cod maad vmad mm: oood 80m momN oaod god h<

 

 

 

m m m v v w m m m m m v m m m <
co an m3 <2 mm mm N. C 2 2: 3 2: mm mm 5. Z 2&0
£8ng x35 «HQ: a-woo_ vtmwmﬁ mm?— mea 8cm «1me 932 mea 932 «1&2 coca v.33 383
83820 -85 coma—2 come—m3; BE? 0325312 83:31: 650m
3%..— Bum
imﬁabm Eczemm

 

 

meets: on pea—8:32

 

 

“macaw—sag BE.§..:3. 89a nova—88 203 «344339: inseam .Ealbm E223 some he.“ 28.483 8 .332 end v.33 v perceive: ﬂag
32:33 :08 no.“ biomanobuo: Aom v5 05 @8898 can 9:5on 98 .15 manager 6:3? A<V v.32 Ba 3.0—? me 585.: .26 money. 0.9—Em 6 28¢.

106

 

 

N86 89° 82 28.0 23 22 82 £2 23 Nana $3 23. 823 N86 N86
:25 $3 $2 83 82 £2 82 23 NM: 53 Nmmd £2 23 £3 Nana
33. N2... 23 No.3 83. 83. 22. £3 83. $3 NNN.N 3N2. 28.2. 83. 2.3
3 we 3. «N N.N Na 3. on Na 3 Ne NN Na 3 3
N93 :3 $2 :3 :3 83 $3 8:. N25 $3 a: .o aNNd $3 $3 3N6
None :3 N23 8N5 a; 3; oood N25 8:. 8:. 83 8N5 :3 83 £3
ENN 2 .N MEN 2 _.N 2NN.N N2 .N SoN NNoN NomN NSN :2 82 SNN ”va moNN
m N v N v N N N v N. N v N N N

mm 2. we 2_ NW NN N 2 2 N2 2 8 NW “N a.
2.2 CNS 89o NNS :28 22 83 $3 .23 N28 MN? 22 $3 and 023
None N33 23 NNE an; 8:. NNvd $2 2.3 O23 N23 82 83 8:. 22
83 $3 8?. NmNN 23 3.; 83. SNN 382 $3 82 83 No; :3. SNN
o N m c c c 2. 2. N c N e m o m

8 Nm 2: N: Nm mN N 2 2 82 2 a 2. «N a.
mNNd mod 32 83 £3 23 2.3 was was NN§ 3% ES ES 83 :86
:3 $3 2.3 83. 2.3 83 N23 22. 23 83 33 Na: 33 8N6 NNod
23. 8N6 ONNN 33 N8... NNNN 83 N23 NNS. N84“ 82 2.2. $3. NSN awn
e e N N r e o N v m o N N N m m

8 8 82 22 Nm mN N 2 2 a 2 NE a .N a.

< :32

<
Z SEQ

<
Z QQNQQ

z 235

 

.Gqcoﬁ m oEdH

107

Table 10. Pairwise estimates of PST between historic (above diagonal) and contemporary
(below diagonal) populations. Pairwise F51 estimates along the diagonal (italicized)
represent comparisons between time periods for the same population. Comparisons were
not statistically/signiﬁcant.

 

 

Population
Little Pere
Population Betsie Manistee Manistee Mara. White Musk_egon
Betsie 0.000 0.007 0.000 0.021 0.001 0.000
Manistee 0.002 0.001 0.000 0.002 0.002 0.000
Little Manistee 0.000 0.007 0. 000 0.013 0.000 0.000
Pere Marquette 0.003 0.005 0.004 0. 007 0.016 0.032
White 0.000 0.000 0.000 0.000 0. 000 0.000
Mu_skegon 0.000 0.007 0.001 0.004 0.000 0. 000

 

108

 

 

 

 

 

 

 

 

 

 

:02005 .o: 055
3:62 25.5%
55.9. www.mn coax” _ N mnoﬁmw 53:22 23 _ -33 00.252:
:0 50m
ca 2: cm: 8m 53:22 32 02252 0:2:
8:: 25885 52.33
mmwd wwo.: oom.o: cow: m 53:22 hog-mom: 0:02.52 0:05
50230:- 50> 20% :0200-5 :020otam :02813. :86on :0209m E50> 53¢
:02”:on .550 20:53: 0w80>< .560 5:832 $2285
-295 0wS0>< -585 .:
050—0: 9 :25
:0258 505003
22 28. $2. 028V bases: 2
n _ adv: gad— wJ 50» :50 :00?on 505008 .3 5:82: 0w80><
8S: $3.5:
ad v3.6 0: mdeGN x x x 31m mmmdom x x 20 205525
28.: 63.2 :
_._ Nwmd v.3 Soﬁa x x x Nwmd 3563 x x 525550
5N5 82.5 aNNNV 32.3 82.: :52 8:3: 82.2:
5.3 9:de 5.9m 32.9: www.mi oNN-dam mam Nowdm NwN-é Ehdm — amt—v owmdh 25.55
63.3 ASN.0: :3.va 6N5:
v5: :19? NHM 2:65 o_ 3 w_m._m_ x x x x GAE-me momémm 5wE22
32 E25250 :82 :58- .>:=-U0-:oEm :29 .>_=w0-=oEIm .33 250-2385 :89 .>:=w0-:oEm :82 :55
:
:o .x. :o 2 «565 22:: 528...; 5&ng Eggs:
5
50> :0: 052m :50 >: :30on 505003 :0 20:5: 0 80>< d
.33 :5

32 503.0: 20>: 53:22 .55 .«o :50 22 :02008 .55 20:05: :50 .3 505005 mo 20:89: 05:22 .: .23: :5 :0 :55 55:8 05005.05 :5
-02m 32:80.5: 8.: :2555 20:5: :5 .588: :38: 80:5: 525 05 no :05: .5w._:22 0:3 2.: :2,—02m 288 :50 .8 «05:05 2:00.050: 20:83:
d2: 52% :5 mg: :5 mats: 8535: 52:22 025: 9 2288 :20050 .3 0585850: 05 5253 50» 05:02:. .83 :5 32 50.50: 52:22

023 mar—0:8: 0.053. 55 05 .«o :50 z: :2550 5:355 :5 :02003 505005 0555...? am 3 5:5: 508 350» 0:: Mo 55:55 6 .2 2:3-

109

.:0:::0 :o: 0:03 :2: :05: .3 058080: a :5 3:: 5 :50: 0.88:8: $2258 032:0:5 02500: $8: 5:: 80: 2 25> 2:5.
0

.5w::22 a: :0286 0005—0 020 20800:: c: 03:20: 02.802:

m:::0w8.: 5:80 .:o 5:838 :o 005: 58050 5:3: 80:0: 225: :5 582:5 825023 :5 20:88: wan—00:0 :23?“ :5 :80: 503:0: 55:2: 093
:
.83 :5 co:— 8 585885 :02080 225: a

.3550: : _ 03$

 

110

Table 12. Pairwise estimates of inter-strain variance in allele
frequency (mean Fs-r over seven microsatellite loci) for four
hatchm strains of steelhead stoclgd into La_l_(e Michigan.

Hatchery Strain
Chambers
Hatchm Strain Skamania Cree_lg_ Ganaraska
Skamania -
Chambers Creek 0.045* -

Ganaraska 0.1 17* 0.060* -
Michigan 0.127* 0.080* 0.037*

 

* P<0.05

111

Table 13. Estimates of classiﬁcation accuracy for likelihood-based assignment tests
(Blanchong et a1. 2002). Percentages on the diagonal are the proportion of individuals of
that strain correctly re-cla_ssiﬁed to stigin of origin based on seven microsatellite loci.

Strain re-classiﬁed a_s

 

 

 

 

Chambers
Strain of origin Micggan Skamania Creﬂc Ganaraska
Michigan 86.7 % - - 13.3 %
Skamania - 98.2 % 1.8 % -
Chambers Creek - 5.0 % 95.0 % -
Ganaraska 6.7 % - 1.7 % 91.7%

 

112

.0353 2:5 2 3.83.535 wean—8555 Bo? <mm\mozo=ow 95.: 33828 acousebag 958%-Enbm £8508 <mm\mozo=ow
98 @2820on 30 SC 0:383 2 BUS E .m_§E>€E 05 no £9538 a co max—£8 ozone» cantata 3.8 :5 632239: 33:0 =a Bingo 03

n

.33: and no woman an ﬁning: 3:03:80
a

 

32 bone“: 2 so. 3:. PE. :. co

 

 

 

 

 

 

 

 

 

 

«on gm :3 as? «N $2 $2 $3 $3 oi $2 $2. 523.6 .x.
a N2 3N NNN_ _ a 0 3 an N on owN 38.
5:: $2 $2 «a. God. 3m :3: $2 $2 .x:
x: Na o w m 2 33 o o o N mm as? N_ o w _ E d 3.5
c <8 <3 see so». 93 95.8 axon gm :3: «a .x; 0% oi.
.me o e N o 2 9.3 o m o N on $3 0 _ N m E .6 sum,
3.8 =E-M=.:Am.
:2: .x; 0on x... 32 :98 ox: can £3. £2 83: oi § § 3:52
$3 8 m2 mNN oNN_ n32 x: _ v 0 2 5 $8 2 o 2 NNN ENm 2::
398 0% X». can :3 A83 can ox: «A83 oi. as; can eon 05%;:
5m m N v m E £3 o N o _ 2: $8 m _ v N E 28
32.358 =E-m=.:&m.\=ek
:8 o 0 call 2 a :2 u 0 Wm 2 a :8 o o xmés 2 a 83x
25 .x. 5:22 .x. 25 .x. 5:32 .x. , 25 .x. 3222 .x.
“38:2 89:52 59:52
Conga: ~32. <mm\mluzo:00 3N8 gocatomno 9:0 5,.—
.3020 neon—:86.

U 98 Ana—«8.239 O Auras—85V am .Aﬁwﬁzzv 2 3.203 953m 6350on 339:9bon32 coma—=25 @282 9 Eng mix—28 5839 38m 98
$9253 ozone» mo cougﬁﬁoo a man: 28 98.33030 93 ca .8 v8.5 coca—82 Baum bone“: 53 Sod @850on .8 cassava 23 33:52 .3 033.

113

Figure l. Depiction of inbreeding and outbreeding on offspring ﬁtness depending on
gelatedness of parents.

 

 

     

Outbreeding

  
  

Inbreeding

   
  

Relative offspring ﬁtness

 

 

Sibling Different species

 

Relatedness of parents

114

Figure 2. A map representing all naturalized populations and hatchery strains sampled in
this study. Naturalized populations are numbered as follows: 1) Thompson Creek, 2)
Black River, 3) Platte River, 4) Betsie River, 5) Manistee River, 6) Little Manistee River,
7) Pere Marquette River, 8) White River, 9) Muskegon River, and 10) St. Joseph River.
Hatchery strains sampled are denoted as follows: C) Chambers Creek strain, G)

Ganaraska strain, S) Skamania strain, M) Michigan strain.

 

 

 

 

 

 

 

Michigan 1 2
w O
. o
g f — 0
‘ 3
4
5
Wisconsin 6
7
C, G, S 8 M, s
' 9
Michigani
Lake
Michigan
10
Illinois
S lnI a
(l) 100”!“
6 106m

115

 

 

Figure 3. This map depicts locations sampled within the Manistee River, Pere Marquette
River, and Muskegon River to determine whether genetically unique populations existed
within river drainages. Populations sampled within the Manistee River were 1) Bear
Creek, and 2) the mainstem of the Manistee River. Within the Pere Marquette River,
populations sampled were 3) Baldwin River, 4) Middle Branch, 5) Little South Branch,
and 6) the Pere Marquette River mainstem. Populations sampled within the Muskegon
River were 7) Bigelow Cred; and 8) the Muskegon River mainstem.

 

 

 

Michigan
0
D
1 °
1
1
. . 2
Wisconsm 3
4
6
5
8 7
Lake Michigan
Michigan
Illinois
Indiana —I'—"

 

 

 

O 100 IMics

116

 

Figure 4. Estimated contributions of each population to total diversity observed based on
the diversity and divergence contributed by each population. The upper graph represents
the unbiased contribution to total diversity of each population, and the lower graph
represents the contribution to total diversity by each population incorporating allelic
richness.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

o.o15~ T g
53:33:10, %
g .
°\o 0.005~ %
5 H g
0.000 . , [—1 , . / /
gaaa a a
-o.oos- .
%
-o.01o , . T . . . . r . 1 . .
MAPM PL MU BEWHSJ BA TH LM SK cc GA
0 0.02~ .

0.01 -

&\\\\\\\\\\\\\\\\\\\\\\\\\.\\\\\

 

 

 

 

 

 

 

 

I l

MAPMPLMUBEWHSJBATHLMSKCCGA

 

 

Naturalized populations Hatchery strains

117

Figure 5. Diagrams of the six mating treatments used. Each circle represents the
combination of gametes from representative males and females. Sex ratios are listed
following the number of female to the number of males. Each treatment was repeated
following the prescribed mating strategy until all males and females allotted for that
treatment were used. Equal volumes of milt and eggs for each male and female were
used for each individﬁual for each merit.

Treatment 1 Treatment 2

Sex ratio: 1:1 Sex ratio: 1:2, repeated
No. females: 10 No. females: 10

No. males: 10 No. males: 10

No. matings: 10 No. matings: 10

Treatment 3
Sex ratio: 5:5 F1 F2 F3 F4 F5 F6 F7 F3 F9 Fm
No. females: 10
No. males: 10
No. matings: 4

M] b42 Rd3 M4 M5 M6 M7 M8 M9 M10

 

 
  
  

Treatment 4
Sex ratio: 10:10
No. females: 10
No. males: 10
No. matings: 1

  
      
 

F1 F2 F3 F4 F5
. F6 F7 F8 F9 Fm
‘ Ml M2 M3 Ma M5
M6 M7 M8 M9 M10

Treatment 5 Treatment 6

Sex ratio: 1:2, mixed Sex ratio: 1:2, sequential
No. females:10 No. females: 10

No. malesz20 No. males: 20

No. matings: 10 No. matings: 10

118

‘

 

 

each male and femgle in each treatment.

T

6. Ngmber of offspring produced by

 

 

 

 

 

 

 

Figure

 

 

O
1
9
8
1 7
m
G 6
m s
.4 4
m a
M
2
4|
0
1
9
8
1 7
t
n
m 6
m 5
M 4
ﬂ 3
m 2
1

 

 

 

 

 

 

 

 

Female ID

 

 

 

I

 

 

 

w
9
8
2 7
t
m 6
m s
m
T 4
o
m a
M 2
w w w m m 0
>562: .o 83:52
N
9
8
2 7
m m
6i.
m 5 ..
. m
M 4
ﬂ 3
m 2

 

 

 

 

 

mamwmso
>509... .o 85:52

 

 

 

 

0 l

  

 

 

 

 

9

8
3 7
m 6
m D

5 u
.m .. .2.
m 2
M 2

mwwwmmo
scone...— .o 80:52

W.

9

8
3 7
m m
. 6 e
m .I.
m s m
T F
. 4
m
a 3
m 2

 

 

s o 2.. m s m
2 2 4| 1
acouoiuoinEaz

 

 

119

Fi ure 6 cont’d).

 

 

 

 

 

 

 

0
1
9
8
4 7
m
9 6
m s m
D
.m 4 m
m. M
h 3
a
M 2
1
m :00 0 0 0 0 o
4 3 2 1
>533 .0 32:52
ll.'|lllll 1. l7
0
1
9
8
4 7
m m
9 6 O
m m
m 5 o
T F
r 4
lo
a 3
F 2
1
0 0 0 0 0
4 3 2 1
>533 .0 82:52

 

 

 

 

 

 

 

 

 

 

 

 

9
1
7
1
5
1
5 w
m
e H
m m
9
e
m m
Tr 7 M
.w
a 5
M
3
1
5 0 5 o 5 0
2 2 1 1
>533 .o 82:52
0
1
9
8
5 .
7
m m
e 6 e
m m
n 5 e
.d F
s 4
h .
a 3
m
e
F 2
1
0 0 0 0 0
4 3 2 1
>583 to 33:52

 

 

 

 

 

7
9
7 1
7
7 1
, 5
7 1
, 6 B
.n.
6 HD
7 m w
93
7 m M
7 .4 7
, a
7 a s
, M
7 3
7 1
5 o 5 0 5 0
, 2 2 1 1
7 >533 .o 53:52
7 l l l lull
6
m m
o to
m m
a
m n
.4
8
h
a
m
o
F

 

 

 

 

 

120

Figure 7. Predicted generational changes in population of coancestry ((9) for populations
maintained using one of the six mating strategies over successive generations.

Coancesz values of 0.25 rgpresent the relationship between full siblings.

 

 

  

   

 

—O—T4 (10:10)
—l—-T 5 (1:2. mix)
+T6(1:2. seq)

1.0

g 0.8 ~

3

; 06 l +T1(1:1) 1
h +T2(1:2.rep) l
8 +Ta(5:5) l
° 1
c

a

O

U

 

 

 

 

 

0 4O 80 120 160 200 240 280 320 360 400 440 480

Generations

121

Figure 8. Demonstration of how the effective population size for captive populations
maintained using six different mating can impact the total effective size of two wild
populations of different size and varying contributions of the captive population,
following Ryman & La_i_kre (1991).

a.

Total population size=40

 

 

70
60 .

4o~
30~
20~

Total effective size

10*

 

 
 
 
 
  

+T1(1:1)
+T2 (1:2. rep)
l—o—T3(5:5)
—O-T4 (10:10)
—l—T5 (1:2, mix)
—x—T6(1:2. seq)

 

 

 

 

450

400 ~

350

150
100

Total effective size

0 0.1 0.2 0.3

300 ,
250 a
200 4

T

T

0.4 0.5 0.6 0.7 0.8 0.9 1

Relative captive contribution

Total population size N=400

 

.1

.1

-1

 

 

 

 

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Relatlve captive contribution

122

 

Figure 9. Geographic locations where historical and contemporary populations of
steelhead were sampled from Michigan tributaries of Lake Michigan. Samples were
obtained from throughout the river drainages. Numbers on maps corresponding to river
names are as follows: 1) Betsie River, 2) Manistee River, 3) Little Manistee River, 4)
Pere Marquette River, 5) White River, 6) Muskegon River. The bars that intersect both
the Manistee River and Muskegon River correspond to dams that prevent upstream
migration of adult steelhead. Sampling locations for these rivers were distributed
throughout the portion of the rivers downstream from the dam.

 

 

 

 

 

 

 

Michigan
0
a
I ° ‘
l
1
2
Wisconsin 3
4
5
6
Michigan
Lake
Michigan
Illinois
Indiana
(I) 100w“
6 106m

123

 

Figure 10. Unrooted neighbor-joining trees based on Cavalli-Sforza & Edwards (1967)
chord distance, demonstrating the genetic relationships among each of the six populations
for two time periods. a) 1983—1984 and b) 1998-2000. Bootstrap values at nodes
represent the percentage of 500 trees where the populations past the nodes were grouped
tggether.

a. b. White
Manistee

    
  

Pere Marquette

    
 
  

Betsie

Manistee

 

39
42 Muskegon

Muskegon

 

Little Manistee

Little Manistee Betsie
White
Pere Marquette

124

 

Figure 11. Geographic locations where populations of steelhead were sampled from
Michigan tributaries of Lake Michigan. Numbers on maps corresponding to river names
are as follows: 1) Platte River, 2) Bear Creek (tributary to the Manistee River), 3) Little
Manistee River, and 4) Pere Marquette River. The bar that intersects the Manistee River
corresponds to Tippy Darn, which prevents upstream migration of adult steelhead.
Sampling locations for these rivers were distributed throughout the portion of the rivers
downstream from the dams. Hatchery strains sampled are denoted as follows: C)
Chambers Creek strain, G) Ganaraska strain, S) Skamania strain, M) Michigan strgin.

 

 

 

 

 

 

 

Michigan
0
a
1 °
2
Wisconsin 3
4
C, G, S
M, S
Lake
Mimiga" Michigan
Illinois
8 Indiana
0 100 Mllos
i 2 J
0 100 KM

125

 

LITERATURE CITED

Aas, G. H., T. Refstie, B. Gjerde. 1991. Evaluation of milt quality of Atlantic salmon.
Aquaculture 95: 125-132.

Allendorf, F.W. 1993 Delay of adaptation to captive breeding by equalizing family size.
Conservation Biology 7: 416-419.

Allendorf, F.W. and N. Ryman. 1987. Genetic management of hatchery stocks. in N.
Ryman and F. Utter, editors. Population Genetics and Fishery Management.
University of Washington Press, Seattle.

Allendorf, F.W. and R. S. Waples. 1996. Conservation and genetics of salmonid ﬁshes. in
J. C. Avise and J. L. Hamrick, editors. Conservation Genetics: Case Histories
From Nature. Chapman and Hall, New York, New York.

Allendorf, F. W. and S. R. Phelps. 1980 Loss of genetic variation in a hatchery stock of
cutthroat trout. Transactions of the American Fisheries Society 109: 537-543.

Allendorf, F .W. and SR Phelps. 1981. Use of allelic frequencies to describe population
structure. Canadian Journal of Fisheries and Aquatic Science. 38: 1507-1514.

Banks, M. A., M. S. Blouin, B. A. Baldwin, V. K. Rashbrook, H. A. Fitzgerald, S. M.
Blankenship, and D. Hedgecock. 1999. Isolation and inheritance of novel

microsatellites in chinook salmon (Oncorhynchus tschawytscha). Journal of
Heredity 90: 281 -288.

Bartron, M. L. and K. T. Scribner. in press. Temporal comparisons of genetic diversity in
Lake Michigan steelhead (Oncorhynchus mykiss) populations: effects of hatchery
supplementation. Environmental Biology of Fishes.

Bartron, M. L., D. R. Swank, E. S. Rutherford, and K. T. Scribner. in review.
Methodological bias in estimates of strain composition and straying of hatchery-
produced steelhead in Lake Michigan tributaries. North American Journal of
Fisheries Management.

Beacham, T. D., S. Pollard, and K. D. Le. 2000. Microsatellite DNA population structure
and stock identiﬁcation of steelhead trout (Oncorhynchus mykiss) in the Nass and
Skeena Rivers in northern British Columbia. Marine Biotechnology 2:587-600.

Beacham, T. D., L. Margolis, and R. J. Nelson. 1998. A comparison of methods of stock
identiﬁcation for sockeye salmon (Oncorhynchus nerka) in Barkley Sound,
British Columbia. North Paciﬁc Anadromous Fish Commission Bulletin 1: 227 -
239.

126

 

Beacham, T. D., S. Pollard, and K. D. Le. 1999. Population structure and stock
identiﬁcation of steelhead in southern British Columbia, Washington, and the
Columbia River based on microsatellite DNA variation. Transactions of the
American Fisheries Society 128: 1068-1084.

Bentzen P., J. B. Olsen, J. E. McLean, T. R. Seamons, and T. P. Quinn. 2001. Kinship
analysis of Paciﬁc salmon: insights into mating, homing, and timing of
reproduction. Journal of Heredity 92: 127-136.

Biette, R. M., D. P. Dodge, R. L. Hassinger, and T. M. Stauffer. 1981. Life history and
timing of migrations and spawning behavior of rainbow trout (Salmo gairdnerz')
populations of the Great Lakes. Canadian Journal of Fisheries and Aquatic
Science 38: 1759-1771.

Blanchong, J. A., K. T. Scribner, and S. R. Winterstein. 2002. Assignment of individuals
to populations: Bayesian methods and multi-locus genotypes. Journal of Wildlife
Management 66:321-329.

Blouin, M. S., M. Parsons, V. Lacaille, and S. Lotz. 1996. Use of microsatellite loci to
classify individuals by relatedness. Molecular Ecology 5:393-401.

Burzynski, T. 1999. Wisconsin’s Lake Michigan salmonid stocking program. Wisconsin
Department of Natural Resources, Great Lakes Research Facility, Milwaukee,
Wisconsin.

Busack, C. A. and K. P. Currens. 1995. Genetic risks and hazards in hatchery operations:
fundamental concepts and issues. American Fisheries Society Symposium 15:71-
80.

Busby, P. J ., T. C. Wainwright, G. J. Bryant, L. Lierheimer, R. S. Waples, F. W.
Waknitz, and I. V. Lagomarsino. 1996. Status review of west coast steelhead from
Washington, Idaho, Oregon, and California. NOAA Tech. Memo. NMFW-
NWFSC-27. U. S. Dep. Commerce.

Candy, J. R. and T. D. Beacham. 2000. Patterns of homing and straying in southern

British Columbia coded-wire tagged chinook salmon (Oncorhynchus
tshauytscha) populations. Fisheries Research 47:41-56.

Cavalli-Sforza, LL. and A.W.F. Edwards. 1967. Phylogenetic analysis: models and
estimation procedures. Evolution 21:550-570.

Chesser, R. K. 1991a Gene diversity and female philopatry. Genetics 127: 43 7-477.

Chesser, R. K. 1991b Inﬂuence of gene ﬂow and breeding tactics on gene diversity
within populations. Genetics 129:573-583.

127

 

Chesser, R. K., O. E. Rhodes, D. E. Sugg, and A. Schnabel. 1993 Effective sizes for
subdivided populations. Genetics 135: 1221-1232.

Chilcote, M.W., B.A. Crawford, and S.A. Leider. 1980. A genetic comparison of
sympatric populations of summer and winter steelheads. Transactions of the
American Fisheries Society 109:203-206.

Cockerham, C. C. 1967. Group inbreeding and coancestry. Genetics 56:89-104.
Cockerham, C. C. 1969. Variance of gene frequencies. Evolution 23:72-84.
Cockerham, C. C. 1973. Analysis of gene frequencies. Genetics 74:679-700.

Comuet, J. M., S. Piry, G. Luikart, A. Estoup, and M. Solignac. 1999. Comparison of
methods employing multilocus genotypes to select or exclude populations as
origins of individuals. Genetics 153 : 1 989-2000.

Crow, J. F. and C. Denniston. 1988. Inbreeding and variance effective numbers.
Evolution 42:482-495.

Danzmann, R. G. 1997. PROBMAX: A computer program for assigning unknown
parentage in pedigree analysis from known genotypic pools of parents and
progeny. Journal of Heredity 88:333.

Danzmann, R. G. and M. M. Ferguson. 1988. Temperature-dependent genotypic selection
and embryonic survival of rainbow trout. Biochemical Genetics 26269-81.

DeWoody, J. A. and J. C. Avise.2001. Genetic perspectives on the natural history of ﬁsh
mating systems. Journal of Heredity 92:167-172.

Dodge, D. P. 1972. Comparative bio-ecology of rainbow trout (Salmo gairdneri,
Richardson) of three tributaries to the Owen Sound, Lake Huron. Doctoral
dissertation. University of Guelph, Ontario.

Dueck, L. A., and R. G. Danzmann. 1996. Matriarchal population structure of introduced
rainbow trout (Oncorhynchus mykiss) in the Lake Ontario watershed. Canadian
Journal of Fisheries and Aquatic Sciences 53:2100-2114.

El Mousadik, A. and R. J. Petit. 1996. High levels of genetic differentiation for allelic
richness among populations of the argan tree [Argan sinosa (L.) Skeels] endemic
of Morocco. Theoretical and Applied Genetics 92:832-839.

Efron, B. 1983. Estimating the error rate of a prediction rule: Improvement on cross-
validation. Journal of the American Statistical Association 78:316-320.

Eshenroder, R. L., M. E. Holey, T. K. Gorenﬂo, and R. D. Clark Jr. 1995. Fish-

128

 

community objectives for Lake Michigan. Great Lakes Fisheries Commission
Special Publication 95-3. Ann Arbor, Michigan.

F elsenstein, J. 1993. Phylip: phylogeny inference package. Ver 3.5c. Department of
Genetics, University of Washington, Seattle.

Fielder, D. G. 1987. An assessment of the introduction of summer steelhead into
Michigan. Michigan Department of Natural Resources, Fisheries Research Report
1948. Lansing, Michigan.

Fleming, 1. A. and M. R. Gross. 1993. Breeding success of hatchery and wild coho
salmon (Oncorhynchus kisutch) in competition. Ecological Applications 3:230-
245.

Fleming, I. A. and M. R. Gross. 1994. Breeding competition in a Paciﬁc salmon (coho:
Oncorhynchus kisutch): measures of natural and sexual selection. Evolution
48:63 7-657.

Fleming, 1. A., B. J onsson, M. R. Gross, and A. Lamberg. 1996. An experimental study
of the reproductive behaviour and success of farmed and wild Atlantic salmon
(Salmo salar). Journal of Applied Ecology 33:893-905.

Ford, M. J. 2002. Selection in captivity during supportive breeding may reduce ﬁtness in
the wild. Conservation Biology 16:815-825.

Garant, D., J. J. Dodson, and L. Bematchez. 2000. Ecological determinants and temporal
stability of the within-river population structure in Atlantic salmon (Salmo salar
L.). Molecular Ecology 9: 615-628.

Gharrett, A.J. 1994. Genetic dynamics of a small population system: a model applicable
to interactions between hatchery and wild ﬁsh. Aquaculture and Fisheries
Management 25 (Suppl. 2):79-92.

Gharrett, A. J. and S. M. Shirley. 1985. A genetic examination of spawning methodology
in a salmon hatchery. Aquaculture 47:245-256.

Gile, S. R. and M. M. Ferguson. 1990. Crossing methodology and genotypic diversity in
a hatchery strain of rainbow trout (Oncorhynchus mykiss). Canadian Journal of
Fisheries and Aquatic Sciences 47:719-724.

Gile, S. R. and M. M. Ferguson. 1995. Factors affecting male potency in pooled gamete
crosses of rainbow trout, Oncorhynchus mykiss. Environmental Biology of Fishes
42:267-275.

Goudet, J. 1995. F-STAT v. 1.2: A computer program to calculate F -Statistics. Journal of
Heredity 86:485-486.

129

 

 

Guo, S. W. and E. A. Thompson. 1992. Performing the exact test of Hardy-Weinberg
proportions for multiple alleles. Biometrics 48: 361-372.

Hansen, M. M., E. Kenchington, and E. E. Nielsen. 2001. Assigning individual ﬁsh to
populations using microsatellite DNA markers. Fish and Fisheries 2:93-112.

Hansen, M. M., D. E. Ruzzante, E. E. Nielsen, and K-L. D. Mensberg. 2001. Brown trout
(Salmo trutta) stocking impact assessment using microsatellite DNA markers.
Ecological Applications 1 1 :148-160.

Hard, J. J. and W. R. Heard. 1999. Analysis of straying variation in Alaskan hatchery
chinook salmon (Oncorhynchus tshawytscha) following transplantation.
Canadian Journal of Fisheries and Aquatic Sciences 56:578-589.

Hard, J. J ., R. G. Kope, W. S. Grant, P. W. Waknitz, L. T. Parker, and R. S. Waples.
1996. Status review of pink salmon from Washington, Oregon, and California.
NTIS No. PB96-l62607.

Heath, D. D, C. Busch, J. Kelly, and D. Y. Atagi. 2002. Temporal change in genetic
structure and effective population size in steelhead trout (Oncorhynchus mykiss).
Molecular Ecology 11:197-214.

Hedrick, P. W., D. Hedgecock, S. Hamelberg, and S. J. Croci. 2000 The impact of
supplementation in winter-run chinook salmon on effective population size.
Journal of Heredity 91:112-116.

Heggberget, T. G., B. O. Johnsen, K. Hindar, B. Jonsson, L. P. Hansen, M. A. Hvidsten,

and A. J. Jensen. 1993. Interactions between wild and cultured Atlantic salmon: a
review of the Norwegian experience. Fisheries Research 18: 123-146.

Hendry, A. P., J. K. Wenburg, P. Bentzen, E. C. Volk, and T. P. Quinn. 2000. Rapid
evolution of reproductive isolation in the wild: evidence from introduced salmon.
Science 290:516-518.

Hilbom, R. 1992. Hatcheries and the future of salmon in the northwest. Fisheries 17:5-8.

Hindar, K., N. Ryman, and F. Utter. 1991. Genetic effects of cultured ﬁsh on natural ﬁsh
populations. Canadian Journal of Fisheries and Aquatic Sciences 48:945-957.

Hutchings, J. A. 1991. The threat of extinction to native populations experiencing
spawning intrusions by cultured Atlantic salmon. Aquaculture 98:119-132.

Kincaid, H. L. 1983. Inbreeding in ﬁsh populations used for aquaculture. Aquaculture
33:215-227.

130

 

 

Kincaid, H. L. 1995. An evaluation of inbreeding and effective population size in
salmonid broodstocks in federal and state hatcheries. American Fisheries Society
Symposium 15:193-204.

Kincaid, H. L., C. C. Krueger, and B. May. 1993. Preservation of genetic variation in the
Green Lake strain lake trout derived from remnant domestic and feral populations.
North American Journal of Fisheries Management 13:318-325.

Krueger, CC. and B. May. 1987. Genetic comparison of naturalized rainbow trout
populations among Lake Superior tributaries: differentiation based on allozyme
data. Transactions of the American Fisheries Society 116:795-806

Lande, R. and G. F. Barrowclough. 1987. Effective population size, genetic variation, and
their use in population management. in M. E. Soule, editor. Viable populations for
conservation. Cambridge University Press, Cambridge.

 

Leider, S. A., M. W. Chilcote, and J. J. Loch. 1984. Spawning characteristics of
sympatric populations of steelhead trout (Salmo gairdneri): evidence for partial

reproductive isolation. Canadian Journal of Fisheries and Aquatic Sciences
41:1454-1462.

Levin, P. S., R. W. Zabel, J. G. Williams. 2001. The road to extinction is paved with
good intentions: negative association of ﬁsh hatcheries with threatened salmon.
Proceedings of the Royal Society B 268:1153-1158.

Lewis, P. O. and D. Zaykin. 2001. Genetic Data Analysis: computer program for the
analysis of alleleic data. Version 1.0 (d16c).
http://lewis.eeb.uconn.edu/lewishome/softwarehtml

Lindgren, B. W. 1976. Statistical theory, 3rd edn. Macmillan Publishing Co., New York.

Lynch, M. 1997. Inbreeding depression and outbreeding depression. pp. 59-67. in W.S.
Grant editor. Genetic effects of straying of non-native hatchery ﬁsh into natural
populations: Proceedings of the Workshop. US. Dept. Comm, NOAA Tech
Memo. NMFS-NWFSC-30.

Lynch, M. and M. O’Hely. 2001. Captive breeding and the genetic ﬁtness of natural
populations. Conservation Genetics. 2: 363-378.

Marshall, A. R., M. Miller, C. Busack, and S. R. Phelps. 1991. Genetic stock
identiﬁcation analysis of three 1990 Washington ocean and Strait of Juan de Fuca
chinook salmon ﬁsheries. Washington Department of Fisheries, GSI Summary
Report 91-1, Olympia, Washington.

MacCrimmon, H. R. and B. L. Gots. 1972. Rainbow trout in the Great Lakes. Ontario
Ministry of Natural Resources. Toronto, Ont.

131

Michigan ﬁsh stocking record. Fisheries Division, Michigan Department of Natural
Resources, Lansing, Michigan. www.dnr.ste_rte.mi.us/ﬁshstockasp

Morris, D. B., K. R. Richard, and J. M. Wright. 1996. Microsatellites from rainbow trout
(Oncorhynchus mykiss) and their use for genetic study of sahnonids. Canadian
Journal of Fisheries and Aquatic Sciences 53: 120-126.

Moulton, F. R. 1939. The migration and conservation of salmon. Publication of the
American Association for the Advancement of Science. Number 8.

National Research Council. 1996. Upstream: salmon and society in the Paciﬁc
Northwest. National Academy Press, Washington DC.

Nelson, R. J. and T. D. Beacham. 1999. Isolation and cross species ampliﬁcation of
microsatellite loci useful for study of Paciﬁc salmon. Animal Genetics 30: 228-
229.

Nelson, R. J ., T. D. Beacham, and M. P. Small. 1998. Microsatellite analysis of the
population structure of a Vancouver Island sockeye salon (Oncorhynchus nerka)
stock complex using nondenaturing gel electrophoresis. Molecular Marine
Biology and Biotechnology 7:312-319.

Negus, M. T. 1999. Survival traits of naturalized, hatchery, and hybrid strains of
anadromous rainbow trout during egg and fry stages. North American Journal of
Fisheries Management 19:930-941.

Nielsen, J. L. 1999. The evolutionary history of steelhead (Oncorhynchus mykiss) along
the US Paciﬁc Coast: developing a conservation strategy using genetic diversity.
ICES Journal of Marine Science 56:449-458.

Nielsen, J. L. and M. C. Fountain. 1999. Microsatelite diversity in sympatric reproductive
ecotypes of Paciﬁc steelhead (Oncorhynchus mykiss) from the Middle Fork Eel
River, California. Ecology of Freshwater Fishes 8: 159-168.

Norris, A. T., Bradley, D. G. and E. P. Cunningham. 2000. Parentage and relatedness
determination in farmed Atlantic salmon (Salmo salar) using microsatellite
markers. Aquaculture 182:73-83.

Olsen, J. B., P. Bentzen, and J. E. Seeb. 1998. Characterization of seven microsatellite
loci derived from pink sahnon. Molecular Ecology 7:1087-1089.

Osterberg, C. O. and G.H. Thorgaard. 1999. Geographic distribution of chromosome and

microsatellite DNA polymorphisms in Oncorhynchus mykiss native to Western
Washington. Copeia 1999:287-298.

132

 

 

Palla, J. S. 1999. 1999 Lake Michigan creel survey. Indiana Department of Natural
Resources, Division of Fish and Wildlife, Michigan City, Indiana.

Page, K. S. 1996. TreeView: an application to display phylogenetic trees on personal
computers. Computer Applications in Biological Science 12:357-35 8.

Parkinson, E. A. 1984. Genetic variation in populations of steelhead trout (Salmo
gairdneri) in British Columbia. Canadian Journal of Fisheries and Aquatic
Sciences 41 : 1412-1420.

Pascual, M. A., T. P. Quinn, and H. Fuss. 1995. Factors affecting the homing of fall
chinook salmon from Columbia River hatcheries. Transactions of the American
Fisheries Society 124:308-320.

Pella, J. J. and G. B. Milner. 1987. Use of genetic marks in stock composition analysis.
Pages 247-276 in N. Ryman and F. Utter, editors. Population Genetics and
Fishery Management. Washington Sea Grant, Seattle, Washington.

Petit, R. J ., A. El Mousadik and O. Pons. 1998. Identifying populations for conservation
on the basis of genetic markers. Conservation Biology 12:844-855.

Philippart, J. C. 1995. Is captive breeding an effective solution for the preservation of
endemic species? Biological Conservation 72:281-295.

Potvin, C. and L. Bematchez. 2001. Lacustrine spatial distribution of landlocked Atlantic
salmon populations assessed across generations by multilocus individual
assignment and rrrixed-stock analyses. Molecular Ecology 10:2375-23 88.

Pritchard, J. K., M. Stephens, and P. Donnelly. 2000. Inference of population structure
using multilocus genotype data. Genetics 155:945-959.

Quinn, T. P. 1993. A review of homing and straying of wild and hatchery-produced
salmon. Fisheries Research 18:29-44.

Rand, P. S., D. J. Stewart, P. W. Seelbach, M. L. Jones, and L. R. Wedge. 1993.
Modeling steelhead population energetics in Lakes Michigan and Ontario.
Transactions of the American Fisheries Society 122:977-1001.

Rannala, B. and J. L. Mountain. 1997. Detecting immigration using multilocus
genotypes. Proceedings of the National Academy of Sciences 94: 9197-9201.

Rao, P.V. 1998. Statistical research methods in the life sciences, Brooks/Cole Publishing
Company, Paciﬁc Grove, CA.

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

133

 

.113

Reisenbichler, R. R. and Phelps, S. R. 1989. Genetic variation in steelhead (Salmo
gairdneri) from the north coast of Washington. Canadian Journal of Fisheries and
Aquatic Sciences 46:66-73.

Reisenbichler, R. R., J. D. McIntyre, M. F. Solazzi, and S. W. Landino. 1992. Genetic
variation of steelhead of Oregon and northern California. Transactions of the
American Fisheries Society 131:158-169.

Reisenbichler, R. R. and S. P. Rubin. 1999. Genetic changes from artiﬁcial propagation
of Paciﬁc salmon affect the productivity and viability of supplemented
populations. ICES Journal of Marine Science 56:459-466.

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

Ricker, W.E. 1972. Hereditary and environmental factors affecting certain salmonid
populations. Pages 27-160 in R. C. Simon and P. A. Larkin, editors. The stock
concept in Paciﬁc salmon. H. R. MacMillan Lectures in Fisheries, University of
British Columbia Press. Vancouver, BC.

Ruzzante, D. E., M. M. Hansen, and D. Meldrup. 2001. Distribution of individual
inbreeding coefﬁcients, relatedness and inﬂuence of stocking on native
anadromous brown trout (Salmo trutta) population structure. Molecular Ecology
10:2107-2128.

Ryman, N. 1991. Conservation genetics considerations in ﬁshery management. Journal of
Fisheries Biology 39:211-224.

Ryman, N., F. Utter, and K. Hindar. 1995. Introgression, supportive breeding, and genetic
conservation. Pages 341-365 in J. D. Ballou, M. Gilpin, and T. J. Foose, editors.
Population management for survival and recovery: analytical methods and
strategies in small population conservation. Columbia University Press, New
York.

SAS Institute. 1999. SAS/STAT Version 8.0. SAS Institute, Cary, North Carolina.

Saitou, N. and M. Nei. 1987. The neighbor-joining method: A new method for
reconstructing phylogenetic trees. Molecular Biology and Evolution 4:406-425.

Scheer, B. T. 1939. Homing instinct in salmon. Quarterly Reviews in Biology 14:408-
430. ‘

Schroeder, R. K., R. B. Lindsay, and K. R. Kenaston. 2001. Origin and straying of

hatchery winter steelhead in Oregon coastal rivers. Transactions of the American
Fisheries Society 130:431-441.

134

 

 

Scribner, K. T., J. R. Gust, and R. L. Fields. 1996. Isolation and characterization of novel
salmon microsatellite loci: cross-species ampliﬁcation and population genetic
applications. Canadian Journal of Fisheries and Aquatic Sciences 53:833-841.

Scribner, K. T., P. A. Crane, W. J. Spearrnan, and L. W. Seeb. 1998. DNA and allozyme
markers provide concordant estimates of population differentiation: analyses of
US. and Canadian populations of Yukon River fall-run chum salmon
(Oncorhynchus keta). Canadian Journal of Fisheries and Aquatic Sciences
55:1748-1758. ‘

Seelbach, P. W. 1986. Population characteristics and dynamics of wild steelhead in a
Michigan tributary of Lake Michigan. Doctoral dissertation. University of
Michigan, Ann Arbor, Michigan.

Seelbach, P. W. 1987a. Smolting success of hatchery-raised steelhead planted in a
Michigan tributary of Lake Michigan. North American Journal of Fisheries
Management 7:223-23 1 .

Seelbach, P.W. 1987b. Effect of winter severity on steelhead smolt yield in Michigan: an
example of the importance of environmental factors in determining smolt yield.
American Fisheries Society Symposium 1:411-450.

Seelbach, P. W. and G. E. Whelan. 1988. Identiﬁcation and contribution of wild and
hatchery steelhead stocks in Lake Michigan tributaries. Transactions of the
American Fisheries Society 117:444-451.

Seelbach, P. W. and B. R. Miller. 1993. Dynamics in Lake Superior of hatchery and wild
steelhead emigrating from the Huron River, Michigan. Michigan Department of
Natural Resources, Fisheries Division, Research Report 1993, Ann Arbor,
Michigan.

Simon, R. C., J. D. McIntyre, A. R. Hemmingsen. 1986. Family size and effective
population size in a hatchery stock of coho salmon (Oncorhynchus kisutch).
Canadian Journal of Fisheries and Aquatic Sciences 43:2434-2442.

Smouse, P. E., J. Neel, and R. R. Sokal. 1986. Multiple regression and correlation
extensions of the Mantel test of matrix correspondence. Systematic Zoology
35:627-632.

Sokal, R. R. and F. J. Rohlf. 1995. Biometry. 3rd ed. Freeman, San Francisco, CA.
Stauffer, T. M. 1955. Migration and grth of tagged, immature rainbow trout in the
Black River, Mackinac County, and in Lake Michigan, 1951-1954. Michigan

Department of Conservation, Institute of Fisheries Research Report 1441, Ann
Arbor, Michigan.

135

Swofford, D. L. and R. B. Selander. 1981. BIOSYS-l: a FORTRAN program for the
comprehensive analysis of electrophoretic data in population genetics and
systematics. Journal of Heredity 72:281-283.

Taylor, E. B. 1991. A review of local adaptation in Salrnonidae, with particular reference
to Paciﬁc and Atlantic salmon. Aquaculture 98: 185-207.

Thedinga, J. F., A. C. Wetherimer, R. A. Heintz, J. M. Maselko, and S. D. Rice. 2000.
Effects of stock, coded-wire tagging, and transplant on straying of pink salmon
(Oncorhynchus gorbuscha) in southeastern Alaska. Canadian Journal of Fisheries
and Aquatic Sciences 57:2076-2085.

Thomas, G. L. and O. A. Mathisen. 1993. Biological interactions of natural and enhanced
stocks of salmon in Alaska. Fisheries Research 18:1-17.

Tringali, M. D. and T. M. Bert. 1998. Risk to genetic effective population size should be
an important consideration in ﬁsh stock-enhancement programs. Bulletin of
Marine Science 62:641-659.

Utter, F. M. and F. W. Allendorf. 1977. Determination of the breeding structure of
steelhead populations through gene frequency analysis. Pages 44-54 in T. J.
Hassler and R. R. Van Kirk, editors. Proceedings of the genetic implications of
steelhead management symposium, May 20-21, 1977, Arcata, CA. Special Report
77-1. California Cooperative Fisheries Research Unit.

Waples, R. S. 1999. Dispelling some myths about hatcheries. Fisheries 24:12-21.

Waples, R. S. 1991a. Genetic interactions between hatchery and wild sahnonids: lessons
from the Paciﬁc Northwest. Canadian Journal of Fisheries and Aquatic Sciences
48 (Suppl. 1):124—133.

Waples, R. S. 1991b. Deﬁnition of “Species” under the Endangered Species Act:
Application to Paciﬁc Salmon. NOAA Technical Memorandum NMFS F/NWC-
194. US. Department of Commerce, National Oceanic and Atmospheric
Administration, National Marine Fisheries Service.

Waples, R. S. 1998. Separating the wheat from the chaff: patterns of genetic
differentiation in high gene ﬂow species. Journal of Heredity 89:438-450.

Waples, R.S., G. A. Winans, F. M. Utter, and C. Mahnken. 1990. Genetic approaches to
the management of Paciﬁc salmon. Fisheries 15:19-25.

Washington, P. M. and A.M. Koziol. 1993. Overview of the interactions and

environmental impacts of hatchery practices on natural and artiﬁcial stocks of
salmonids. Fisheries Research 18:105-122.

136

Weir, B. S. 1990. Genetic Data Analysis. Sinauer Publishing Company, Sunderland, MA.

Weir, B. S. and C. C. Cockerham. 1984. Estimating F-Statistics for the analysis of
population structure. Evolution 38:1358-1370.

Weitkamp, L. A., T. C. Wainwright, G. J. Bryant, G. B. Miller, D. J. Teel, R. G. Kope,
and R. S. Waples. 1995. Status review of coho salmon from Washington, Oregon,
and California, NTIS No. PB96-106554.

Withler, R. E. 1988. Genetic consequences of fertilizing chinook salmon (Oncorhynchus
tshauytscha) eggs with pooled milt. Aquaculture 68: 15-25.

Wood, J. W. 1987. The genetic demography of the Gainj of Papua New Guinea, 2.
Determinants of effective population size. American Naturalist 129:165-187.

Wright, S. 1931. Evolution in Mendelian populations. Genetics 16297-159.

137