EXTIRPATION OF A NATIVE POPULATION FOLLOWING A STOCKING
PROGRAM: GENETIC POPULATION STRUCTURE AND DEMOGRAPHICS OF
KOKANEE (ONCORHYNCHUS NERKA) IN A LARGE IMPOUNDED WATERSHED
by
Paige Nicole Wilson
B.Sc., University of Louisville, 2016

THESIS SUBMITTED IN PARTIAL FULFILLMENT OF
THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE
IN
NATURAL RESOURCES AND ENVIRONMENTAL STUDIES
(BIOLOGY)

UNIVERSITY OF NORTHERN BRITISH COLUMBIA
October 2021
© Paige Nicole Wilson, 2021

ABSTRACT
Kokanee, the non-anadromous life history form of Oncorhynchus nerka, use lacustrine
habitat in watersheds draining into the north Pacific Ocean. Kokanee have also been widely
introduced into reservoirs following impoundment of rivers due to the construction of dams.
Locally-adapted subpopulations of Kokanee, however, should be identified and evaluated when
implementing watershed-level management strategies. In Chapter 1, I examined fork length,
condition factor, and age at maturity for Kokanee in the Williston watershed of northern British
Columbia to identify potential spatial and temporal trends in demographic structure following a
large-scale stocking program that occurred in the 1990s. Adult spawning Kokanee that were
native to the reservoir and collected prior to stocking events were significantly larger and
maintained higher condition factors than Kokanee stocked from the Columbia River sampled in
any year after 1991. Introduced Kokanee sampled in 2018 and 2019 were the smallest spawners
and were significantly smaller than all fish collected between 1989 and 2018; the condition
factors of these fish were also significantly lower than native Kokanee and the first spawning
cohorts of Columbia-origin fish. The average age at maturity did not change across spatial or
temporal scales (3 yrs.). My results indicate an ongoing trend of decreasing spawner size and
condition factor for Kokanee in the Williston Reservoir since introduction events in the early
1990s. In Chapter 2, I analyzed the genetic population structure of Kokanee in the Williston
watershed, including from the reservoir before stocking Columbia-origin fish and native
populations from headwaters of the Williston Reservoir: Thutade, Arctic, and Tacheeda Lakes.
Using microsatellite markers, I identified that all fish collected from 2006 to 2019 were
introduced Columbia-origin genotypes, and there was no evidence of genetic divergence by
spawning location. Native populations in Arctic and Tacheeda Lakes remained entirely separate

ii

from the reservoir populations, and there was no indication of past or current introgression with
introduced stock. I identified that native Williston Reservoir Kokanee diverged from the Thutade
Lake population; as native Williston fish have not been sampled since 2000, it is likely that this
population has been extirpated in the reservoir by the successful Columbia-origin lineage. My
results highlight an unfortunate consequence of underinformed management practices that failed
to recognize the native Williston Kokanee as a distinct population. Strategies that incorporate
knowledge of subpopulations of Williston watershed Kokanee, such as genetic populations or
reproductive ecotypes, should be prioritized to conserve locally-adapted genetic diversity.

iii

TABLE OF CONTENTS
ABSTRACT.................................................................................................................................... ii
TABLE OF CONTENTS ............................................................................................................... iv
LIST OF TABLES ......................................................................................................................... vi
LIST OF FIGURES ..................................................................................................................... viii
LIST OF APPENDICES................................................................................................................ xi
ACKNOWLEDGMENTS ............................................................................................................ xii
PROLOGUE ................................................................................................................................... 1
CHAPTER 1: Historic and contemporary morphometrics and life history characteristics
of Oncorhynchus nerka in the Williston Reservoir of northern B.C....................................... 7
Introduction ..................................................................................................................................... 7
Methods........................................................................................................................................... 9
Sampling and Data Sources ............................................................................................. 9
Otolith Extraction and Ageing........................................................................................ 12
Data Analysis.................................................................................................................. 12
Results ........................................................................................................................................... 14
Among Year Fork Length Comparisons ......................................................................... 14
Within-Cohort Fork Length Comparisons...................................................................... 15
Among Year Condition Factor Comparisons ................................................................. 15
Within-Cohort Condition Factor Comparisons.............................................................. 16
Between Year Age at Maturity Comparisons ................................................................. 17
Discussion ..................................................................................................................................... 23
Size, Condition Factor, and Age at Maturity ................................................................. 23
Factors Affecting Growth ............................................................................................... 25
Environmental and Anthropogenic Variables ................................................................ 27
Genetic Effects on Size, Condition Factor, and Age ...................................................... 30
Conclusions .................................................................................................................... 31
CHAPTER 2: Genetic population structure of introduced and native lineages of
Oncorhynchus nerka in a large, impounded watershed ........................................................ 32
Introduction ................................................................................................................................... 32
Methods......................................................................................................................................... 35
Sampling ......................................................................................................................... 35
Microsatellite Markers ................................................................................................... 38
Genetic Variability ......................................................................................................... 45
Genetic Structure ............................................................................................................ 46
iv

Results ........................................................................................................................................... 48
Microsatellite Markers ................................................................................................... 48
Genetic Variability ......................................................................................................... 49
Genetic Structure ............................................................................................................ 51
Discussion ..................................................................................................................................... 68
Status of Native Williston Watershed Kokanee .............................................................. 68
Interactions Between Columbia-Type and Native Fish .................................................. 71
Population Structure of Tributary Spawners ................................................................. 75
Conclusions .................................................................................................................... 76
EPILOGUE ................................................................................................................................... 78
REFERENCES ............................................................................................................................. 87
APPENDIX A ............................................................................................................................... 99

v

LIST OF TABLES
Table 1.1. Number of Williston watershed Kokanee examined in analysis of fork length
(LF) and condition factor (KFL) by collection location and year. The number of
Kokanee assessed per measurement was dependent on recorded information or
extracted otoliths (presented in brackets). In 2000, data was separated based on
genotyping for native Williston (2000W) and Columbia-origin (2000 C) fish........................ 20
Table 1.2. Average fork length (LF, cm) and condition factor (KFL; [g/cm3] ✕ 100) for
spawning Kokanee sampled in the Williston Reservoir watershed by year. Data are
presented as means ± standard error (range in parentheses). Samples collected in
1989–1991were native Williston Reservoir Kokanee; 1994 fish were Columbiaorigin Kokanee (3 native Williston Kokanee were removed based on genotype); in
2000, data was separated based on genotyping for native Williston (2000W) and
Columbia-origin (2000C) fish; and after 2000, genotyping revealed that all fish were
Columbia-origin. ................................................................................................................... 21
Table 1.3. Pairwise comparison table of the P values of Kruskal-Wallis tests and post hoc
Dunn tests comparing fork length (cm; above diagonal) and condition factor (KFL;
below diagonal) of Kokanee collected from the Williston watershed across years.
Significantly different values (P < 0.05) are bolded. ............................................................ 21
Table 1.4. Average fork length (cm) and condition factor (KFL; [g/cm 3] ✕ 100) for
spawning Kokanee repeatedly sampled from three tributaries to the Williston
Reservoir. Data are presented as means ± standard error (range in parentheses). ................ 22
Table 1.5. Pairwise comparison table of the P values of a two-way ANOVA with
interaction and post hoc Tukey’s HSD test comparing fork length (cm; above
diagonal) and condition factor (KFL; below diagonal) of Kokanee collected from
three tributaries to the Williston Reservoir in 2006 (06) and 2018 (18). Significantly
different values (P < 0.05) are bolded. .................................................................................. 22
Table 1.6. Collection year, number of samples (n), and age of Williston Reservoir
Kokanee for which otoliths were examined. Data are means ± standard error (range
in parentheses). No significant differences in age estimates between years were found
(Kruskal-Wallis test, P = 0.08).............................................................................................. 22

vi

Table 2.1. Forward and reverse sequences, annealing temperature (Ta), and allelic size
ranges of 14 microsatellite loci for Oncorhynchus nerka. A delineates the number of
alleles observed across all sample groups of the comprehensive dataset of 1,870
genotypes. .............................................................................................................................. 43
Table 2.2. Sample groups of Williston watershed Kokanee genotypes by sample location
and year. Rsvr. 2000 is retained as a separate group because of an excess of
homozygotes that indicates a Wahlund effect (Khrustaleva et al. 2017). Genotypes
are organized into four datasets: comprehensive (C; n = 1,870; groups 1–21);
tributary (T; n = 905; groups 5–17); time-series (TS; n = 347; groups 8, 10, and 13
organized by sample year 2006, 2018, 2019); and native (N; n = 340; groups 1–2,
20–21). Donor populations are groups 3–4; fish collected from the body of the
reservoir are groups 18–20. ................................................................................................... 44
Table 2.3. Results of multiple AMOVA between groups of Kokanee sampled from the
same locations, given as FST and associated P value. ........................................................... 56
Table 2.4. Measurements of genetic variation of Williston watershed Kokanee by sample
group. A represents the total number of alleles across all loci per group, and AR is the
rarefied allelic counts across all loci per sample group. Expected heterozygosity and
observed heterozygosity are represented by He and Ho, respectively. Ho did not
depart from He for any particular group across all loci, but Ho was significantly lower
than expected for locus Ots3 (P = 0.016). For each parameter, standard errors are
given in parentheses. ............................................................................................................. 57
Table 2.5. F-statistics (Weir and Cockerham 1984) of 21 groups of Kokanee sampled
from the Williston watershed in northern British Columbia, Canada. .................................. 59

vii

LIST OF FIGURES
Figure P.1. The Williston Reservoir as formed by the construction of the W.A.C. Bennett
Dam in 1968 and subsequent impoundment of the Upper Peace River. Three main
reaches are relevant to this study: Finlay (north and west), Peace (east), and Parsnip
(south)...................................................................................................................................... 6
Figure 1.1. The Williston Reservoir and associated reaches of the watershed. Red symbols
signify the original stocking locations of Columbia-origin Kokanee in the watershed
(Langston 2012). Blue symbols are locations of sample sites of Kokanee collected in
1989 (n = 32), 1990 (n = 119), 1991 (n = 56), 1994 (n = 117), 2006 (n = 350), 2018,
(n = 284) and 2019 (n = 123). Orange symbols are the approximate locations of
within-reservoir sampling as they occurred in 2000 (n = 34; Pillipow and Langston
2002). ..................................................................................................................................... 11
Figure 1.2. Frequency of length classes (LF; cm) for a preliminary analysis (n = 952) of
Kokanee caught in the Williston Reservoir and tributaries to the reservoir, compiled
by year (a) and by location per year (b–f). Kokanee collected in 2000 (d) were
isolated by genotype: native Williston (2000W) and Columbia-origin (2000C). All fish
included in this analysis were considered mature and/or were actively spawning at
the time of capture. ................................................................................................................ 18
Figure 1.3. Light microscope images from mature Kokanee (Oncorhynchus nerka)
spawners caught in the Williston Reservoir watershed. A) native Williston Reservoir
Kokanee (29.2 cm, 340.0 g) caught in the lower Finlay River in 1989. B) Columbiaorigin Kokanee (22.1 cm, 125.6 g) caught in Aley Creek in 2018. C) Columbiaorigin Kokanee (21.9 cm, 99.1 g) caught in Russel Creek in 2019. D) Columbiaorigin Kokanee (22.2 cm, 128.4 g) caught in Aley Creek in 2019. Red dots indicate
annuli. .................................................................................................................................... 19
Figure 2.1. The Williston Reservoir and associated reaches of the watershed. Red symbols
signify the original stocking locations of Columbia-origin Kokanee in the watershed
(Langston 2012). Blue symbols are locations of sample sites of Kokanee that spawn
in tributary streams to the reservoir. Purple symbols are locations where native
Kokanee were found in the Parsnip Reach; pink symbols indicate native Kokanee in
the Finlay Reach. Orange symbols are the approximate locations of within-reservoir
viii

sampling as they occurred in 2000 (Pillipow and Langston 2002). Samples collected
from the G.M. Shrum intake towers in 2016 and 2019 were from the site closest to
the W.A.C. Bennett Dam. ..................................................................................................... 37
Figure 2.2. Estimates of mean inbreeding coefficients (F) represented as bars for
individual Kokanee in Arctic and Tacheeda groups (top), Native Rsvr. and Thutade
groups (middle), and tributary groups (bottom) of Kokanee collected from the
Williston watershed. The dashed line represents the overall mean F for each group:
Tacheeda (0.246), Arctic (0.234), Native Rsvr. (0.205), Thutade (0.182), and
tributary (0.165). ................................................................................................................... 58
Figure 2.3. Neighbor-joining tree constructed using Cavalli-Sforza and Edwards (1967)
chord distance (DC) inferred from variation at fourteen microsatellite loci in 21
groups of Williston watershed Kokanee. Numbers represent percentage of 10,000
bootstrap replicates. Percentages below 50% are not reported. ............................................ 59
Figure 2.4. STRUCTURE analysis for the comprehensive dataset of Kokanee assayed at
fourteen microsatellite loci. Each fish is represented by a vertical line which exhibits
the proportional composition of each fish’s genome across genetic clusters (K = 1–6). ...... 60
Figure 2.5. STRUCTURE analysis for the tributary dataset of Kokanee collected in
tributaries to the Williston reservoir. Genetic clusters spread across all sample groups
with increasing K, suggesting that these fish belong to one genetic population (K =
1). ........................................................................................................................................... 61
Figure 2.6. STRUCTURE analysis for the time-series dataset of Kokanee collected from
Germansen, Pelly, and Russel and grouped by sample year. Increasing admixture and
a lack of clustering between sample years indicates a single genetic population (K =
1). ........................................................................................................................................... 62
Figure 2.7. STRUCTURE analysis for the native dataset (Arctic, Tacheeda, Native Rsvr.,
and Thutade) of Kokanee, and those collected in 2000 (Rsvr. 2000), which exhibited
evidence of a mixed-stock sample group. ............................................................................. 63
Figure 2.8. STRUCTURE analysis for the Rsvr. 2000 group of Kokanee that were sampled
from the body of the reservoir by Pillipow and Langston (2002). The analysis clearly
delineates two genetic clusters, which, in the context of the comprehensive dataset,
are in-line with Hill-type (blue) and native-type (orange) Kokanee. .................................... 64

ix

Figure 2.9. Plots of discriminant analysis of principal components (DAPC) for fourteen
microsatellites across the comprehensive dataset of Williston watershed Kokanee.
Individual genotypes in the scatterplot (a) are represented by dots, and individual
membership probabilities (b) are presented as vertical bars; genetic clusters are colorcoded. The predominant genetic population of each cluster are as follows: Cluster 1
= Hill; Cluster 2 = Meadow; Cluster 3 = Native Rsvr. and Thutade; Cluster 4 =
Arctic and Tacheeda. ............................................................................................................. 65
Figure 2.10. A subsection of the DAPC for the comprehensive dataset across fourteen
microsatellite loci. Identifying the individuals by sample group reveals that the Rsvr.
2000 group (▲) contains individuals that align with Cluster 1 (Hill), Cluster 2
(Meadow), and Cluster 3 (native-type). The Native Rsvr. group (●), comprised of
Kokanee collected from 1988 to 1990 (Native Rsvr.) entirely aligns with Cluster 3.
The Rsvr. Forebay group (

) were assigned to Cluster 1 and Cluster 2, with one

outlier aligning with Cluster 3. Thutade fish (+) were assigned mostly to Cluster 3,
but several aligned with Cluster 2. ........................................................................................ 66
Figure 2.11. A principal component analysis (PCA) for four groups of Williston
watershed Kokanee (Rsvr. Forebay, Rsvr. 2000, Native Rsvr., and Thutade). The
Rsvr. 2000 is shown not to be a distinct genetic cluster, but rather a mixed-stock
sample group containing individuals that cluster with Columbia-type (Rsvr. Forebay)
and native (Native Rsvr. and Thutade) genetic populations. ................................................ 67

x

LIST OF APPENDICES
A2.1. Total number (n) of Williston watershed Kokanee genotypes from each location
and collection year. Hill Creek and Meadow Creek (Columbia River) samples are
included as source stocks. ..................................................................................................... 99
A2.2. Evidence of null alleles due to homozygote excess (Aley; Thutade) and likely a
Wahlund effect (Rsvr. 2000). Statistically significant values (P < 0.05) are indicated
with *. .................................................................................................................................. 101
A2.3. Pairwise Nei's (1987) FST genetic distance estimates between native
Kokanee sampled from Arctic and Tacheeda Lakes, the body of the reservoir before
stocking events, and Thutade Lake are represented below the
diagonal. HIERFSTAT bootstrapping over loci confidence intervals are presented in
brackets above the diagonal. Bolded values are significantly different from zero. ............ 102
A2.4. Pairwise Weir and Cockerham (1984) ! genetic distance estimates between
Williston watershed Kokanee sample groups are represented below the

diagonal. HIERFSTAT bootstrapping over loci confidence intervals are presented in
brackets above the diagonal. Bolded values are significantly different from zero ............. 103

A2.5. Evanno table output of the STRUCTURE analysis of the comprehensive dataset, as
obtained through STRUCTURE HARVESTER. A K of 2 was most supported by the
Evanno protocol, but this represents an oversimplification of the population structure
(Janes et al. 2017). ............................................................................................................... 104
A2.6. Evanno table output of the STRUCTURE analysis of the tributary dataset, as obtained
through STRUCTURE HARVESTER. A K of 2 was most supported by the Evanno
protocol, but this represents an oversimplification of the population structure (Janes
et al. 2017). .......................................................................................................................... 104
A2.7. Plot of the Bayesian Information Criterion (BIC) values from which the lowest
optimal number of clusters (K) was chosen. A K of 4 was selected for the
comprehensive dataset......................................................................................................... 105
A2.8. A principal component analysis (PCA) for the tributary dataset showing the lack of
cluster patterns exhibited by Kokanee from these sample locations. .................................. 106

xi

ACKNOWLEDGMENTS
I would like to express my deep gratitude to the Tsay Keh Dene, McLeod Lake Indian Band,
Kwadacha, Nak’azdli Whut’en, Saulteau, and also the Lheidli T’enneh peoples for their generosity in
hosting me in their traditional territories while I was conducting my thesis research. The Williston
Reservoir is a lasting legacy of harm and colonial violence against many First Nations in British
Columbia, and these communities continue to advocate on behalf of the fish and wildlife under their
stewardship.
I would like to thank Dr. Mark Shrimpton for remembering me when this research opportunity
presented itself and for inviting me back to UNBC. I have felt incredibly lucky to have had you as a
thoughtful supervisor and a wonderful mentor. Thank you for always checking in and being available for
all of my questions. For the record, BIOL 307 truly is the best course in western North America and I’m
very proud I got to be a part of it.
This project was made possible with generous financial support: the Fish and Wildlife
Compensation Program – Peace Region, and the FWCP Aquatic Research Award; the UNBC Research
Project Award; and the UNBC Special Graduate Support Award in Research.
I am grateful to Ruth Withler, Mike Wetklo, and the laboratory team at the Fisheries and Oceans
Canada Pacific Biological Station for genotyping the tissue samples and providing binned data so
expediently. I’d like to thank Andrew McDermot-Fouts and the field crew at DWB Consulting Services
Ltd. for collecting Kokanee, conducting enumeration flights, and delivering the frozen specimens to me in
Tsay Keh Dene—a very fond memory for me. A big thank-you to Dr. Steven Cooke, Dirk Algera, and
Taylor Ward of Carleton University who, with Randy Zemlak of BC Hydro, provided me with Kokanee
from the G.M. Shrum intake towers. Thank you to Chelsea Coady of the FWCP and Nikolous Gantner,
Matt Neufeld, and Steven Arndt of the Province of B.C. Ministry of Forests, Lands, National Resource
Operations and Rural Development for providing me with archived scale samples and even more
Kokanee. I would also be remiss if I did not thank Ken Ambrose and North/South Consultants, Inc. for
the otolith ageing results and methodology. All contributions were very much appreciated.
I am exceedingly grateful to my committee members, Dr. Fiona Johnston and Dr. Brent Murray,
for your comments, feedback, enthusiasm, and support. A massive thank-you to Dr. Michael Russello for
your thoughtful questions and recommendations as my external examiner. I’m proud of what we have
accomplished together.
To Luc Turcotte and Cale Babey—thank you for being such welcoming lab mates and great fish
bros. I miss seeing you in the hall of Building 4 every day. To Cherie McKeeman—thank you for your
mentorship in TA-ing, lab protocols, and all of your time and feedback with my writing. How can I ever
thank you for inviting me to Trench Brewing my first week in PG? To Jeannine Randall—your solidarity
and friendship continue to mean the world to me. A giant thank-you to everyone who ever kept me
company in the lab: Luc, Dan Larson, and Brittney Reichert for helping me process fish; Cale and Erin
Hall for counting endless eggs; and the fall 2020 class of BIOL 406 and my dear friend Joe Bottoms for
dutifully ageing otoliths. The friendship and support I experienced from my lab mates, close friends, and
fellow NRES students made my masters truly extraordinary. Thank you!
And it’s not hyperbole to say that I could not have made it through this experience without the
love and encouragement from my parents—who believed in me enough to help me move across the
continent. Unending gratitude goes to David, for the laughs, solace, QGIS maps, Tuesday tea, and Friday
sushi. UNBC was just the beginning.

xii

PROLOGUE
Effective conservation and management strategies require an understanding of population
trends, genetic structure, and mechanisms influencing compensatory responses (Andrusak et al.
2000; Askey 2016). Ill-informed management decisions could result in unexpected or
unanticipated consequences (Peck et al. 2019; Bassett et al. 2020; Warnock et al. 2021). For
instance, fish populations that are locally adapted to a region could be eradicated through genetic
introgression (Veale and Russello 2016), intrapopulation competition (Crowder 1984), densityrelated predation pits (Warnock et al. 2021), or through the degradation of key spawning habitats
(Frazer and Russello 2013). Rather than adopting broad or static regulations, as is typical of
traditional fisheries systems, care should be taken to consider locally-adapted subpopulations
when implementing management strategies at the watershed level (Taylor et al. 2000; Askey
2016).
Kokanee (Oncorhynchus nerka), the non-anadromous freshwater form of Sockeye
Salmon, exhibit distinct intraspecific and subpopulation diversity in reproductive behavior
(Taylor et al. 1997; Lemay and Russello 2015). A pronounced homing ability in the species
allows for adaptation to specific environments and the development of divergent reproductive
ecotypes within the same watershed or river system (Taylor et al. 2000; Wood et al. 2008;
Whitlock et al. 2018). This species’ potential for diverse genetic structure is of particular
importance in the context of fisheries management and fish conservation; careful consideration
should be taken to avoid unintentional losses of genetic diversity over the course of conservation,
recovery, or supplementation efforts (Nelson et al. 2003). Run timing (Beacham et al. 2014),
abundance patterns for harvest and escapement goals (Beacham et al. 2010), population-specific
responses to environmental and density variables (Rieman and Myers 1992; Martinez and

1

Wiltzius 1995), and the preservation of genetic and ecotype variability (Nichols et al. 2016) are
all dependent on an understanding of the population structure of the target species.
The Williston Reservoir was identified as an optimal study area to investigate the effects
of introducing a large number of Kokanee to a highly-impacted, ultra-oligotrophic lentic system
(Stockner et al. 2005). Located at approximately 56° N latitude, 124° W longitude, the Williston
Reservoir is the largest body of water in British Columbia and was created by the impoundment
of the Upper Peace River by the W.A.C. Bennett Dam in 1968 (Langston and Murphy 2008;
Robinson 2020). The watershed is composed of three main sub-watersheds or “reaches” that
center on the flooded river systems of the Peace (east), Parsnip (south), and Finlay (west and
north) (Blackman et al. 1990; Robinson 2020). The substantial annual drawdown (11 m) and
frequent flushing (2.2 years) of the reservoir have effectively prevented the formation of a
functional littoral zone and, combined with high dissolved oxygen levels, have contributed to the
reservoir’s current status as ultra-oligotrophic (Blackman et al. 1990; Wilson and Langston 2000;
Harris et al. 2005; Stockner et al. 2005). Species that were adapted to the pre-impoundment lotic
environments, notably Arctic Grayling (Thymallus arcticus) and Mountain Whitefish
(Prosopium williamsoni), were slowly replaced with those more suited to reservoir conditions,
such as Kokanee and Burbot (Lota lota) (Sebastian et al. 2009; Langston 2012). In an effort to
develop a sport fishery and increase productivity in the Williston Reservoir, the B.C. Ministry of
Environment and the Fish and Wildlife Compensation Program – Peace Region (FWCP) jointly
developed an initiative to introduce a new lineage of Kokanee to supplement the existing native
population in the Finlay Reach and thereby establish self-sustaining Kokanee runs in
southeastern (i.e., accessible) regions of the reservoir (Blackman et al. 1990). From 1990–1998,
over 3 million juvenile Kokanee from Hill Creek (Arrow Lakes Reservoir) and Meadow Creek

2

(Kootenay Lake) of the Columbia River were introduced into five tributaries to the Williston
Reservoir: Carbon Creek, Davis River, Dunlevy Creek, Manson River, and Nation River
(Langston and Murphy 2008; Langston 2012). Since their introduction, the Columbia-origin
Kokanee expanded rapidly from their five initial stocking tributaries and have now been
observed spawning in as many as 68 tributaries to the reservoir (Langston 2012; Robinson 2020).
The original objectives of the Williston Lake [sic] Management Plan included directives
to conserve, enhance, supplement, and maximize fish abundance and species diversity within the
Williston watershed (Blackman et al. 1990). Reservoir-wide, a particular emphasis was placed
on prioritizing enhancement activities over pre- and post-program inventory evaluations
(Blackman et al. 1990). For Kokanee specifically, the stocking program was considered to be
crucial for the immediate establishment of a sport fishery in streams accessible to recreational
anglers; the enhancement of native populations in the Finlay Reach was secondary and would
only be prioritized if stocked Kokanee did not show adequate spawning success (Blackman et al.
1990; Langston and Murphy 2008). The existing populations of Kokanee in Thutade Lake
(Finlay River headwaters) and Arctic and Tacheeda Lakes (Parsnip River headwaters), and their
potential interactions with Kokanee in the reservoir, were entirely omitted by the original
management plan (Blackman et al. 1990). Ongoing enumeration surveys have documented the
widespread population growth of Columbia-origin Kokanee in the years following the stocking
program (Langston 2012; McDermot-Fouts 2019; Robinson 2020), and the current FWCP
Reservoirs Action Plan has now identified the stocked Kokanee as a potential source of
reservoir-wide community effects (FWCP 2020). However, recent enumeration flights have
reported a marked decrease in spawner escapement compared to a decade prior (McDermotFouts 2019; Robinson 2020).

3

There are considerable knowledge gaps concerning the population-specific adaptive
responses and genetic impacts of such an intensive Kokanee stocking program in the Williston
Reservoir. My research examined the status and trends of native Williston and Columbia-origin
populations of Kokanee in the watershed, including the reservoir, tributaries to the reservoir, and
nearby Thutade, Artic, and Tacheeda Lakes (Figure P.1). In Chapter 1, I examined fork length,
condition factor, and age at maturity for Kokanee in the watershed to identify potential trends in
demographic structure over time. I analyzed differences in adult size and condition factor
between native and the introduced stock, as well as among spawning locations within the same
year and cohort. My results were used to assess the potential influences of environmental factors
and fish density on Kokanee size, condition factor, and age at maturity.
In Chapter 2, I conducted an in-depth spatial and temporal analysis of the genetic
population structure of Williston watershed Kokanee. DNA was extracted from tissue and scale
samples of Kokanee collected from 1988 to 2019 and microsatellite loci were examined to assess
the degree of genetic introgression, population differentiation, and persistence of genetic lineages
of native and introduced Kokanee.
The Williston watershed is a highly complex and modified freshwater system and the
reservoir has been a site of ongoing, large-scale habitat change since its formation (FWCP 2020).
My research contributes to the growing body of work that highlights the importance of
identifying management units at the genetic or subpopulation level (Burger and Spearman 1997;
Moreira and Taylor 2015; Nichols et al. 2016; Veale and Russello 2016), especially when
managing or mitigating human-caused habitat alterations (Taylor et al. 2014; Jensen et al. 2017).
Proper identification of independent population structure and at-risk management units may

4

prevent unintentional loss of genetic variability and could ultimately help guide future restoration
and conservation actions in the reservoir.

5

Figure P.1. The Williston Reservoir as formed by the construction of the W.A.C. Bennett Dam in 1968 and subsequent
impoundment of the Upper Peace River. Three main reaches are relevant to this study: Finlay (north and west), Peace
(east), and Parsnip (south).

6

CHAPTER 1: Historic and contemporary morphometrics and life history characteristics of
Oncorhynchus nerka in the Williston Reservoir of northern B.C.
Introduction
The Williston Reservoir, formed by the construction of the W.A.C. Bennett Dam on the
Upper Peace River in 1968, is the largest body of water in British Columbia and the 7th largest
reservoir in the world by volume (Figure 1.1) (Chao et al. 2008; FWCP 2020). Since its initial
formation, the reservoir has followed a pattern typical of northern lentic freshwater systems: a
decade-long period of high productivity, followed by a systematic decline into an ultraoligotrophic state of remarkably low primary production (Harris et al. 2005; Stockner et al.
2005). A large-scale Kokanee (Oncorhynchus nerka) stocking program was implemented in the
1990’s to generate a sport fishery and provide a source of prey for large piscivorous fish species,
such as native Bull Trout (Salvelinus confluentus) and proposed Lake Trout (Salvelinus
namaycush) and Gerrard Rainbow Trout (Oncorhynchus mykiss) stocking initiatives (Blackman
et al. 1990; Langston 2012; FWCP 2020). Over 3 million juvenile Kokanee sourced from the
Columbia River were stocked into tributaries to the Williston Reservoir from 1990 to 1998
(Blackman et al. 1990; Langston and Murphy 2008), and by 2006 Kokanee abundance was
reported to have expanded to up to 9 million fish, with spawners observed in at least 68
tributaries across the watershed (Sebastian et al. 2009; Langston 2012).
Ongoing enumeration studies have monitored the distribution of Kokanee from the
1990’s to the present (Langston 2012; McDermot-Fouts 2019; Robinson 2020). Enumeration
flights conducted in 2018 and 2019 noted a substantial reservoir-wide decrease in spawner
abundance during the peak spawning period for both years (McDermot-Fouts 2019; Robinson
2020). Because Kokanee are known to exhibit density-dependent changes in size, body

7

condition, and age at maturity, the populations in the Williston Reservoir may have undergone
physical changes in response to changes in fish density and environmental productivity (Hyatt
and Stockner 1985; Rieman and Myers 1992).
Monitoring the body size of Kokanee is important for population management, because
the size of mature adults—primarily females—directly impacts the number and mass of eggs
produced, as well as the ability of the individual to construct and successfully defend redd sites
(Kaeriyama et al. 1995; McGurk 2000). The condition of a fish, often characterized as an index
of “plumpness”, is a commonly monitored factor of fish health, favorable habitat conditions, and
food availability (Moyle and Cech 2004; Froese 2006; Blackwell et al. 2010). The direct
relationship between length and weight allows for more robust predictions of fecundity,
population growth and abundance, gonadal and somatic growth, and overall health and wellbeing than either length or weight measurements alone (Moyle and Cech 2004; Blackwell et al.
2010). Condition factor (KFL), in particular, has been used extensively in fisheries management
and is widely considered to be a useful measurement of isometric fish growth within a selected
species (Heinke 1908; Froese 2006; Nash et al. 2006; Blackwell et al. 2010).
Similar to body size and condition factor, age at maturity is an important metric to
monitor for effective species and population management (Patterson et al. 2008). Spawning
cohorts that are younger at maturity may indicate they are faster-growing, which in turn can be
influenced by both genetic and environmental factors. Fish density, growth, and mean age at
maturity for a spawning cohort can also be influenced by management decisions such as altered
harvest regulations (Patterson et al. 2008).
I examined morphometric data of mature spawning Kokanee collected over a period of
30 years to identify potential changes in body size, condition, and age at maturity of fish of the
8

Williston Reservoir. I used recorded historical and contemporary field measurements to directly
compare fork length, condition factor, and otolith annuli count between spawner groups.
Changes in morphometrics and age over time may be indicative of environmental influences,
genetic or cohort lineages, or adaptive responses to changes in fish density (Kendall et al. 2014).
My work will aid the ability to predict and assess factors such as population growth and
escapement, which may have cascading effects throughout the managed Williston Reservoir
system (van Poorten et al. 2018; Oke et al. 2020).
Methods
Sampling and Data Sources
Almost 2,000 Kokanee were collected from the Williston Reservoir and tributaries to the
reservoir between 1989 and 2019; of these, 1,115 were included in analyses of fork length (LF,
cm) and condition factor (KFL), calculated as mass divided by fork length cubed (g/cm3) ✕ 100
(Table 1.1). I used fork length and mass (g) data recorded on scale collection envelopes donated
by the Fish and Wildlife Compensation Program for Kokanee gill netted in the watershed in
1989 (n = 32), 1990 (n = 119), 1991 (n = 56), and 1994 (n = 117), before and shortly after the
initiation of the stocking program; where able, I corroborated these collection envelope
measurements with their respective FWCP – Peace Region reports (Fielden 1991; Fielden 1992;
Langston and Zemlak 1998; Pillipow and Langston 2002).
Collection envelope and report data from mature Kokanee gill netted from six locations
in the Williston Reservoir between August 25–September 3 of 2000 (n = 34) were also included,
and unpublished field collection data for Kokanee that were collected in 2006 (n = 350) were
provided by Randy Zemlak of BC Hydro. Whole frozen spawners collected in 2006 (n = 113),
2016 (n = 23), and 2017 (n = 10) were also donated to UNBC by Dr. Nikolaus Gantner of the
9

B.C. Ministry of Forests, Lands, Natural Resource Operations and Rural Development
(MoFLNRORD) for purposes of otolith extraction and age determination.
Kokanee were also collected for this study by DWB Consultants, Inc. in the spawning
period of 2018 (n = 282) and 2019 (n = 123). Capture was accomplished predominately through
the use of a backpack electrofishing unit (Model 12-B, Smith-Root Inc., Vancouver, WA, USA)
and stunned fish were collected with a seine or dip net. Fish were anesthetized in a solution of
100 mg · L–1 tricaine methane sulfonate (MS-222) buffered with sodium bicarbonate, euthanized
by concussion and exsanguination, and frozen at -20ºC until otolith extraction.
I analyzed field values of fork length and weight to avoid potential bias introduced by the
effects of freezing (Armstrong and Stewart 1997; Leonard et al. 2021).

10

Figure 1.1. The Williston Reservoir and associated reaches of the watershed. Red symbols signify the original stocking
locations of Columbia-origin Kokanee in the watershed (Langston 2012). Blue symbols are locations of sample sites of
Kokanee collected in 1989 (n = 32), 1990 (n = 119), 1991 (n = 56), 1994 (n = 117), 2006 (n = 350), 2018, (n = 284)
and 2019 (n = 123). Orange symbols are the approximate locations of within-reservoir sampling as they occurred in
2000 (n = 34; Pillipow and Langston 2002).

11

Otolith Extraction and Ageing
Otoliths were included in collection envelopes for Kokanee collected from the Finlay
River in 1989 (n = 13) and from Philip Creek in 1994 (n = 27). Whole Kokanee spawners that
were frozen (2006–2019) were partially thawed before I extracted otoliths. A total of 464 otoliths
were examined for age (Table 1.1). I viewed each whole otolith under a dissecting microscope
(Leica S9i Stereomicroscope; Opti-Tech, Toronto, ON, Canada) and captured the image with the
integrated 10 M.P. complementary metal-oxide semiconductor (CMOS) digital camera to
visualize annuli. Three individuals and one consultant group independently viewed whole
otoliths or digital images and assigned an age based on the number of visible annuli; independent
age assessments were performed to reduce potential bias and incorporate estimates from
researchers with prior aging experience and expertise. A subsample of otoliths (n = 100) from
spawning Kokanee sampled in 2006 and 2018 were submitted to North/South Consultants, Inc.
(Winnipeg, MB, Canada) for age and ageing method verification, and otoliths were further
assessed by J. Bottoms (UNBC; n = 226), Dr. M. Shrimpton (UNBC; n = 450), and myself
(UNBC; n = 464).
Data Analysis
I performed a preliminary analysis to assess whether sizes of Kokanee at maturity have
changed over time by visually assessing fork length data (n = 952), grouped by year in length
frequency plots using DataGraph version 4.6.1 (Visual Data Tools, Inc., Chapel Hill, NC, USA).
To identify potential changes in size or age at maturity over time, a one-factor Analysis of
Variance (ANOVA) was conducted with R v4.0.3 to test for differences in morphometric data
(LF and KFL) or age by collection year, respectively. LF, KFL, and age estimates were grouped by
year, with the exception of LF and KFL data from 2000, which were included as two separate

12

groups based on genotype: native Williston (2000W) and Columbia-origin (2000 C; Chapter 2). I
assessed the post-ANOVA residuals for normality and homoscedasticity with Levene’s tests and
Shapiro-Wilk tests. Given a significant failure of normality, I conducted a Kruskal-Wallis test
and post hoc Dunn’s test with the false discovery rate (FDR) correction procedure of Benjamini
and Hochberg (1995) to determine significant differences in morphometric and age data among
collection years. Mean values are presented as means ± standard error.
To assess potential variances in size, condition factor, and age at maturity that could be
attributed to specific spawning locations, I also grouped the morphometric and age data into a
subset examining only three tributaries to the Williston Reservoir: Germansen River, a tributary
to the Omineca River on the west side of the reservoir; Pelly Creek, a tributary to the Ingenika
River on the northwest side of the reservoir; and Russel Creek, a tributary to the Finlay River to
the north of the reservoir (Figure 1.1). All three of these locations were sampled in both 2006
and 2018. I used a random number generator to select equal sample sizes (40 fish) for each
sample group. Assuming a generation time of four years, these sample years should have
theoretically fallen into the same spawning cohort, and any apparent difference in size or
maturation of fish in these spawning locations could potentially have been attributed to
interspecific population genetics, site-specific environmental differences, or other factors. I used
a two-factor ANOVA to test for differences in LF, KFL, and age between spawning location,
among collection years, and the interaction between spawning location and collection year. All
three datasets were tested for normality and homoscedasticity, and mean values are presented as
means ± standard error.

13

Results
Among Year Fork Length Comparisons
Sizes of Kokanee at maturity clearly grouped by year in a length-frequency assessment
(Figure 1.2), and individual fork lengths (LF) ranged from 16.1 cm (2000W) to 33.8 cm (1990;
Table 1.2). A one-factor Kruskal-Wallis test found a significant decrease in fork length at
maturity over time (P < 0.001). Mature native Williston fish collected in 1989 (29.17 cm ± 0.24),
1990 (29.72 cm ± 0.17), and 1991 (28.62 cm ± 0.26) did not differ significantly in length
between the three collection years (Table 1.3). Fish caught in 1990 were on average the largest,
and spawners in all three years of 1989, 1990, and 1991 were significantly larger than Kokanee
collected in 1994 and after. The first cohort of mature Columbia-origin stocked Kokanee,
sampled in 1994, were significantly smaller (26.79 cm ± 0.09; P < 0.05 for all comparisons) than
the native reservoir fish, but they were on average 2 to 4 cm larger than fish caught between
2006 (24.85 cm ± 0.10) and 2019 (21.81 cm ± 0.09; P < 0.001 for all comparisons; Table 1.3).
Kokanee sampled from the reservoir in 2000 and 2006 occupied an intermediate position
between the larger native and first-cohort Columbia fish and the smaller Kokanee that have been
sampled since 2006. The 21 fish that were identified as mature native reservoir Kokanee
(Chapter 2) were smaller (24.16 cm ± 0.74) on average than the 13 Columbia-origin Kokanee
(25.87 cm ± 0.87), but this distinction was not significant (P = 0.085). 2000C Kokanee fork
lengths were not significantly different than those of Kokanee collected in 1994 (P = 0.388), but
2000W measurements were significantly smaller than those of 1994 fish (P = < 0.001). Both
2000W and 2000C fork lengths were not significantly different from measurements of 2006 fish
(P = 0.218 and P = 0.251, respectively).

14

Kokanee collected in 2018 (22.08 cm ± 0.08) and 2019 (21.81 cm ± 0.09) were the
smallest spawners and were significantly smaller than all fish collected between 1989 and 2018
(P < 0.001 for all comparisons; Table 1.2 and Table 1.3). There was no statistical difference in
LF between 2018 and 2019 spawners.
Within-Cohort Fork Length Comparisons
Mature Kokanee collected from Germansen River, Pelly Creek, and Russel Creek in 2006
and 2018 were grouped by collection year and fork length data were log-transformed to induce
normality (Levene’s test P = 0.083; Shapiro-Wilk test P = 0.660). Year, location, and the
interaction between year and location were all determined to have a significant effect on fork
length (P < 0.001 for all), and a Tukey’s HSD test revealed significant differences among the
pairwise comparisons (Table 1.4). Fork lengths of individual Kokanee collected in 2006 ranged
from 21.7 cm (Pelly) to 32.4 cm (Germansen; Table 1.4). Kokanee collected in 2018 were
generally smaller, with individual LF ranging from 19.1 cm (Russel) to 24.4 cm (Pelly).
Germansen fish sampled in 2006 (27.27 cm ± 0.27) were the largest of the six groups examined
and were significantly larger than Pelly and Russel 2006 fish (P < 0.001 for both comparisons),
as well as fish sampled from all three locations in 2018 (P < 0.001; Table 1.5). Kokanee
collected from Pelly and Russel in 2006 did not differ in fork length significantly (P = 0.999),
but both were significantly larger than fish from all three locations in 2018 (P < 0.001 for all
comparisons). All Kokanee collected in 2018 did not significantly differ from one another in fork
length. The smallest group from this dataset was Russel in 2018 (21.66 cm ± 0.22).
Among Year Condition Factor Comparisons
Individual condition factor (KFL) values of mature spawning Kokanee grouped by
collection year ranged from 0.52 (2018) to 1.69 (1991; Table 1.2). An overall decrease in KFL

15

from 1989 to 2019 was found with a one-factor Kruskal-Wallis test (P < 0.001; Table 1.3). KFL
values generally followed the trend of LF, with a few exceptions. While the fork lengths of
Williston Kokanee collected in 2000 (2000W) were on average shorter than Columbia-origin
Kokanee collected in the same year (2000C), the KFL values of 2000W fish (1.29 ± 0.03) were not
statistically different from their 2000C counterparts (1.19 ± 0.02; P = 0.082). 2000W KFL values
also showed no difference from Kokanee collected pre-2000 (P > 0.10 for all comparisons), but
were significantly greater than all fish collected between 2006 and 2019 (P < 0.001 for all
comparisons).
Kokanee sampled from 2006 to 2019 showed the lowest KFL values of all fish analyzed in
this study (Table 1.2). Kokanee collected in 2006 did not significantly differ in condition factor
from 2000C spawners (P = 0.522). 2018 and 2019 spawners had significantly lower average KFL
values than every other sample year (P < 0.001 for all comparisons), and the Kokanee in 2018
had the statistically-lowest K FL values (1.05 ± 0.01; P < 0.05 for all comparisons).
Within-Cohort Condition Factor Comparisons
Condition factor of mature Kokanee collected from Germansen River, Pelly Creek, and
Russel Creek in 2006 and 2018 were grouped and assessed with a two-factor ANOVA; KFL data
did not necessitate log-transformation to induce normality (Levene’s test P = 0.178; ShapiroWilk test P = 0.909). Location did not significantly affect KFL, but location and the interaction
between year and location both were significant factors for KFL (P < 0.001 for both). KFL values
of individual Kokanee collected in 2006 ranged from 0.81 (Russel) to 1.52 (Pelly; Table 1.4).
Kokanee collected in 2018 had on average lower KFL values, with individual values ranging from
0.76 (Russel) to 1.32 (Germansen and Russel). A Tukey’s HSD test revealed that spawner K FL
values in each respective year did not significantly differ among sample locations, with the

16

exception of Germansen fish showing significantly higher KFL than Pelly in 2018 (P = 0.046;
Table 1.5). All fish collected in 2006 showed significantly greater KFL values than all fish
collected in 2018 (P < 0.01 for all comparisons). Pelly fish sampled in 2006 (1.25 ± 0.02) had the
greatest average KFL of all analyzed groups. The spawners with the lowest average KFL from this
dataset was Pelly in 2018 (1.02 ± 0.02).
Between Year Age at Maturity Comparisons
Annuli of collected otoliths were clearly visible for many, but not all, of the whole mount
images for Kokanee sampled from the Williston watershed (Figure 1.3). A one-factor KruskalWallis test of age data for Kokanee collected from 1989 to 2019 indicated there were no
significant differences in age between sampled years (P = 0.08). As with fork length data, age
data for Kokanee collected in 2006 and 2018 from Germansen, Pelly, and Russel were grouped
and log-transformed to induce normality (Levene’s test P = 0.55; Shapiro-Wilk test P < 0.001).
A two-factor ANOVA did not indicate any significant differences between year (P = 0.78),
location (P = 0.15), or interactions between year and location (P = 0.63). Based on the age
estimates of fish examined for this study, there was no indication of significant change in age at
maturity over time.
Individual age estimates ranged from 1 year (2018) to 5 years (1994; Table 1.6). The
grand mean of ages across all sample years and locations was 2.83 years ± 0.02. This indicates
that most Kokanee in the reservoir spawn at 3 years of age and supports a generation time, from
fertilization to sexual maturation, of 4 years for both native Williston and Columbia-origin fish.
While not statistically significant, Kokanee collected in 1989 (3.08 years ± 0.14) and 1994 (3.07
years ± 0.09) were marginally older than fish collected between 2006 and 2019 (Table 1.6).

17

Figure 1.2. Frequency of length classes (LF; cm) for a preliminary analysis (n = 952) of Kokanee caught in the
Williston Reservoir and tributaries to the reservoir, compiled by year (a) and by location per year (b–f). Kokanee
collected in 2000 (d) were isolated by genotype: native Williston (2000W) and Columbia-origin (2000C). All fish
included in this analysis were considered mature and/or were actively spawning at the time of capture.

18

Figure 1.3. Light microscope images from mature Kokanee (Oncorhynchus nerka) spawners caught in the Williston
Reservoir watershed. A) native Williston Reservoir Kokanee (29.2 cm, 340.0 g) caught in the lower Finlay River in
1989. B) Columbia-origin Kokanee (22.1 cm, 125.6 g) caught in Aley Creek in 2018. C) Columbia-origin Kokanee
(21.9 cm, 99.1 g) caught in Russel Creek in 2019. D) Columbia-origin Kokanee (22.2 cm, 128.4 g) caught in Aley
Creek in 2019. Red dots indicate annuli.

19

20

104

56

31 (27 )
1

1

(131)

151

131

211

50
2

(103)

(103)

2017

4

41 (41 )
4

40

4

4

40 (40 )
4

4

41 (40 )
4

404 (394)

414

41
4

2018

4

43 (41 )
4

4

40 (37 )
4

404 (404)

2019

Total

49 (13)

(20)

84 (82)

50

40

130 (116)

50

280

91 (79)

(13)

130 (114)

41

31 (27)

Otoliths extracted from whole frozen spawners donated by Nikolaus Gantner (B.C. MoFLNRORD)

Field data and otoliths obtained from spawners collected by DWB Consultants, Inc. for the FWCP – Peace Region

3

4

20

Total
32 (13)
119
56
117 (27)
13
21
350 (113) (23)
(10)
284 (160) 123 (118) 1115 (464)
1
Field data (or preserved otoliths) recorded on scale collection envelopes and reports provided by Chelsea Coady (FWCP – Peace Region); Fielden 1991; Fielden 1992;
Langston and Zemlak 1998; Pillipow and Langston 2002
2
Field data provided by Randy Zemlak (BC Hydro; unpublished data)

Aley Creek
Stevenson
Creek
Reservoir

Bower Creek

Cutoff Creek

50 (39 )
3

Russel Creek
2

50
2

100
2

3

Tsaydiz Creek

Finlay River

2

(133)

2016

76

1

2006

23

2000W

63

2000C

631

1994

50 (39 )
1

1991

Pelly Creek
1

1990

502 (353)

1

12

1

1989

Manson River
Germansen
River
Osilinka River

Philip Creek

Dunlevy Creek

Carbon Creek

Location

Table 1.1. Number of Williston watershed Kokanee examined in analysis of fork length (LF) and condition factor (KFL) by collection location and year. The
number of Kokanee assessed per measurement was dependent on recorded information or extracted otoliths (presented in brackets). In 2000, data was separated
based on genotyping for native Williston (2000W) and Columbia-origin (2000C) fish.

Table 1.2. Average fork length (LF, cm) and condition factor (KFL; [g/cm3] ✕ 100) for spawning Kokanee
sampled in the Williston Reservoir watershed by year. Data are presented as means ± standard error
(range in parentheses). Samples collected in 1989–1991were native Williston Reservoir Kokanee; 1994
fish were Columbia-origin Kokanee (3 native Williston Kokanee were removed based on genotype); in
2000, data was separated based on genotyping for native Williston (2000W) and Columbia-origin (2000C)
fish; and after 2000, genotyping revealed that all fish were Columbia-origin.
Year
1989
1990
1991
1994
2000C
2000W
2006
2018
2019

LF
29.17 ± 0.24 (25.5–31.9)
29.72 ± 0.17 (21.7–33.8)
28.62 ± 0.26 (19.0–31.7)
26.79 ± 0.09 (23.7–29.9)
25.87 ± 0.87 (17.3–29.0)
24.16 ± 0.74 (16.1–29.2)
24.85 ± 0.10 (21.6–32.4)
22.08 ± 0.08 (19.1–28.4)
21.81 ± 0.09 (19.3–24.2)

KFL
1.37 ± 0.02 (1.09–1.54)
1.31 ± 0.01 (1.02–1.55)
1.28 ± 0.02 (1.03–1.69)
1.23 ± 0.02 (0.77–1.53)
1.19 ± 0.02 (1.07–1.35)
1.29 ± 0.03 (0.75–1.47)
1.15 ± 0.01 (0.69–1.52)
1.05 ± 0.01 (0.52–1.44)
1.08 ± 0.01 (0.76–1.40)

Table 1.3. Pairwise comparison table of the P values of Kruskal-Wallis tests and post hoc Dunn tests
comparing fork length (cm; above diagonal) and condition factor (KFL; below diagonal) of Kokanee
collected from the Williston watershed across years. Significantly different values (P < 0.05) are bolded.
1989
1989
1990
1991
1994
2000C
2000W
2006
2018
2019

0.152
0.023
<0.001
0.002
0.188
<0.001
<0.001
<0.001

1990
0.744
0.170
<0.001
0.017
0.725
<0.001
<0.001
<0.001

1991
0.557
0.252
0.142
0.122
0.568
<0.001
<0.001
<0.001

1994
0.007
<0.001
0.013
0.414
0.113
<0.001
<0.001
<0.001

2000C
0.016
0.004
0.033
0.388
0.082
0.522
0.004
0.045

2000W
<0.001
<0.001
<0.001
<0.001
0.085
<0.001
<0.001
<0.001

2006
<0.001
<0.001
<0.001
<0.001
0.251
0.218
<0.001
<0.001

2018
<0.001
<0.001
<0.001
<0.001
<0.001
<0.001
<0.001

2019
<0.001
<0.001
<0.001
<0.001
<0.001
<0.001
<0.001
0.307

0.044

21

Table 1.4. Average fork length (cm) and condition factor (KFL; [g/cm3] ✕ 100) for spawning Kokanee
repeatedly sampled from three tributaries to the Williston Reservoir. Data are presented as means ±
standard error (range in parentheses).
Location

Year

n

LF

KFL

Germansen
Pelly
Russel
Germansen
Pelly
Russel

2006
2006
2006
2018
2018
2018

40
40
40
40
40
40

27.27 ± 0.27 (24.4–32.4)
25.05 ± 0.28 (21.7–29.1)
25.14 ± 0.22 (23.0–28.4)
22.09 ± 0.16 (20.4–24.0)
21.81 ± 0.15 (19.5–24.4)
21.66 ± 0.22 (19.1–24.3)

1.19 ± 0.01 (0.99–1.44)
1.25 ± 0.02 (1.01–1.52)
1.21 ± 0.02 (0.81–1.49)
1.10 ± 0.02 (0.86–1.32)
1.02 ± 0.02 (0.82–1.32)
1.03 ± 0.02 (0.76–1.25)

Table 1.5. Pairwise comparison table of the P values of a two-way ANOVA with interaction and post hoc
Tukey’s HSD test comparing fork length (cm; above diagonal) and condition factor (KFL; below diagonal)
of Kokanee collected from three tributaries to the Williston Reservoir in 2006 (06) and 2018 (18).
Significantly different values (P < 0.05) are bolded.
Germansen 06 Pelly 06
Germansen 06
<0.001
Pelly 06
0.241
Russel 06
0.989
0.615
Germansen 18 0.012
<0.001
Pelly 18
<0.001
<0.001
Russel 18
<0.001
<0.001

Russel 06 Germansen 18
<0.001
<0.001
0.999
<0.001
<0.001
0.001
<0.001
0.046
<0.001
0.096

Pelly 18
<0.001
<0.001
<0.001
0.917

Russel 18
<0.001
<0.001
<0.001
0.595
0.991

1.000

Table 1.6. Collection year, number of samples (n), and age of Williston Reservoir Kokanee for which
otoliths were examined. Data are means ± standard error (range in parentheses). No significant
differences in age estimates between years were found (Kruskal-Wallis test, P = 0.08).
Year
1989
1994
2006
2016
2017
2018
2019

n
13
27
113
23
10
160
118

Age
3.08 ± 0.14 (2–4)
3.07 ± 0.09 (2–5)
2.79 ± 0.05 (2–4)
2.74 ± 0.13 (2–4)
2.80 ± 0.13 (2–3)
2.77 ± 0.04 (1–4)
2.87 ± 0.04 (2–4)

22

Discussion
The results of my study provide a comprehensive view on the shifting population
demographics and trends exhibited by Kokanee in an ultra-oligotrophic reservoir environment.
Based on the morphometric data in the timeframe examined in this study, there has been a
significant decrease in fork length and condition factor at maturity since the introduction of
Columbia-origin Kokanee in the early 1990’s. The native Williston fish collected in the late
1980’s and the first mature cohorts of introduced Kokanee were both significantly larger and
maintained a higher length-weight relationship at sexual maturity than most Kokanee collected
since 2006, and this trend appears to be continuing in contemporary spawning cohorts. Age at
maturity, conversely, has not significantly changed over time and all examined cohorts have
maintained a generation time of 4 years. These factors, combined with both historic and recent
escapement estimates, have implications for future fisheries management decisions to ensure the
productive survival of Kokanee in the Williston watershed.
Size, Condition Factor, and Age at Maturity
Kokanee in the Williston Reservoir have undergone a striking change in size at maturity
over time. Native Kokanee collected in the late 1980s and early 1990s—immediately prior to
stocking of Kokanee from the Columbia River—were significantly larger than fish collected in
2018 and 2019 by an average of almost 8 cm. The initial cohorts of Columbia-origin Kokanee,
which first matured in 1994, were also significantly smaller than native Kokanee. A comparison
of a small number of mature native and introduced-lineage Kokanee collected in 2000 suggests
an interesting dichotomy—the Columbia fish were on average larger but exhibited lower
condition factor (Table 1.2). However, collections from the spawning periods of 2006 to 2019
reveal a significant decrease in both fork length and condition factor in mature adult Kokanee;

23

spawners collected in 2019 were the smallest and had the lowest condition factor of all spawner
groups examined in my study.
Large variation (>5 cm) in spawner size in consecutive years is not unheard of for
Kokanee populations in British Columbia. Assessments of spawner size and escapement at the
Hill Creek Spawning Channel of the upper Columbia River revealed an increase of 4.7 cm in
average spawner fork length between 2005 and 2006 cohorts (Andrusak 2007). However, this
was believed to be due to the overall decrease in spawner abundance in 2006 as a within-lake
compensatory mechanism (Andrusak 2007). It was also predicted that the 2007–2009 spawners
would exhibit a significant shift in age at maturity as a compensatory response to low
recruitment from 2003 to 2005, but this was not verified with age analyses in subsequent years
(Andrusak 2007). In the context of the Williston Reservoir, both recent (Coxson et al. 2017;
Wilson and Shrimpton 2020) and subsequent (Wilson and Shrimpton 2021) analyses involving
Kokanee have confirmed similar sizes for spawners collected in 2016, 2017, and 2020, which
suggests the decline in size at maturity is not an anomalous event but rather a sustained trend.
The age estimates of mature Kokanee did not significantly differ across the timeframe,
and all examined populations exhibited a mature age of 3 years and a generation time of 4 years.
This result was unexpected, as an increase in age at maturity and generation time is a known
adaptive response to factors that contribute to decreased growth rates and size at maturity,
including limited resources, acute stress, or restricted environmental conditions (Grover 2005;
2006). In some situations, the reverse is also true; poor growth conditions can also cause earlier
maturation in environments where adult survival is disproportionately impacted (Grover 2005).
No significant change in age or generation time while undergoing a significant decrease in length
and condition factor, as I found to be the case for Williston watershed Kokanee, may indicate

24

that the time allocated to somatic growth has not been affected by growth conditions or fish
density in the reservoir.
Factors Affecting Growth
Density-dependent interactions in an environment of low productivity may be influencing
the size and condition factor of Kokanee spawners in the Williston watershed (Sebastian et al.
2003; Stockner et al. 2005; Sebastian et al. 2009). Kokanee spawner abundance and distribution
across the Williston watershed have been repeatedly assessed, and data are available from 2002,
2003, 2006, 2007, 2010, 2018, and 2019 (Langston 2012; McDermot-Fouts 2019; Robinson
2020). Observed spawner escapement increased rapidly from less than 3,000 adults in four of the
original five stocking tributaries in 1994 to a peak observed spawner escapement of over 598,000
fish in 28 tributaries in 2010 (Langston and Zemlak 1998; Langston 2012; Robinson 2020).
Spawner escapement began to decline after 2010, and standardized comparisons of spawner
enumerations indicated a roughly 3.2-fold decrease in Kokanee spawner abundance in 2018
compared to 2010, and a 4.2-fold decrease in 2019 compared to 2003 (McDermot-Fouts 2019;
Robinson 2020). Across selected streams that were flown for escapement estimates in 2019,
spawner abundance had decreased to less than 41,000 adults (Robinson 2020). Increasing the
frequency of enumeration flights is recommended to accurately assess spawner escapement
moving forward, as the gap between 2010 and 2018 spawner surveys could have better informed
current escapement trends. Nevertheless, this dramatic increase and subsequent decline may
suggest a linkage to density-dependent factors and increased competition that could be limiting
the growth of Kokanee in the ultra-oligotrophic reservoir environment.
The negative correlation between body size and population density has been welldocumented in salmonids, and Kokanee populations experiencing changes in fish density are

25

known to exhibit strong adaptive growth responses (Hyatt and Stockner 1985; Rieman and
Myers 1992; Askey 2016). Productivity of the lacustrine environment particularly influences the
range of size at maturity for Kokanee spawners in different lakes or reservoirs (Askey 2016). The
status of the Williston Reservoir as ultra-oligotrophic (Stockner et al. 2005) is likely more of a
contributing factor to the size and density of Kokanee than density-dependent factors alone.
The mechanisms influencing compensatory growth responses in Kokanee populations are
dynamic and complex (Kendall et al. 2014) and may occur fairly quickly. In other salmonids,
such as Brown Trout (Salmo trutta), levels of sustainable spawning and recruitment were reached
in 2–6 years after an intense population decline (Jenkins et al. 1999). In a study of Kokanee in
Bucks Lake, California, a longstanding (25+ year) density-induced trend of decreasing size at
maturity was reversed in the 5-year period following the construction of a spawning barricade,
which prevented adults from successfully spawning and reduced fish density overall (Grover
2006). Additionally, this increase in size was effectively re-reversed in the years following the
removal of the spawning barricade and the subsequent increase in Kokanee abundance: adult size
was reduced to pre-barricade levels within 8 years (Grover 2006).
For other systems, even exceptional growth conditions do not affect Kokanee populations
as management efforts intended. Two long-term British Columbia nutrient restoration programs,
in Kootenay Lake (since 1992) and in the Upper and Lower Arrow Lakes Reservoir (since 1999),
have resulted in abundant zooplankton resources and entirely different compensatory growth
responses. In the Kootenay Lake system, the Kokanee that survived to sexual maturity in recent
years reached record sizes (>36 cm), but the overall population remained in a state of nearcollapse due to intensive top-down predation from piscivorous species (Peck et al. 2019;
Warnock et al. 2021). Density-dependent growth responses in the form of larger sizes at maturity

26

were apparent, but also coincided with a shift to later maturation (age 4 spawners) rather than the
predicted rapid growth and age 2 spawners (Peck et al. 2019). For the Arrow Lakes Reservoir,
the situation was reversed: spawner escapement and overall density remained low in recent years
and density-dependent growth was not apparent, but spawners appeared to mature early, at age 2
(Bassett et al. 2020).
Generally-speaking, however, if growth conditions are favorable we would expect to see
an increase in salmonid size at maturity over time following a period of limited recruitment;
conversely, a smaller size at maturity would be indicative of a high-density or low-productivity
environment (Johnston and Post 2009). In the context of an ultra-oligotrophic environment, the
combination of a significant decrease in fork length and condition factor over time, no significant
change in age over time, and an observed decline in spawner abundance over the last 15 years
may be indicative of a reservoir-wide population decline following a period where the upper
threshold density may have been reached (Rieman and Myers 1992; Martinez and Wiltzius
1995). The lack of typical adaptive responses to density suggests the oligotrophic Williston
Reservoir and limited food resources are more influential on Kokanee size, condition factor, and
age than density-dependent factors.
Environmental and Anthropogenic Variables
The mechanisms behind the observed trend of declining size of Kokanee at maturity is
not fully understood for the Williston Reservoir, and further research is therefore required to
effectively manage this species. Egg size and overall fecundity of Kokanee are directly related to
length—namely, females that reach a larger size at maturity also produce a greater number and
generally larger eggs (McGurk 2000). In the case of Williston Kokanee, a significant decrease in
fork length at maturity over the period examined in this study could be correlated with a

27

shrinking population as each reproductive cohort produces fewer eggs. There is also evidence to
suggest a significant decrease in spawner length and fecundity with increasing latitude; this
could suggest that Columbia-origin Kokanee, which were sourced from around 50°N, may be
substantially more fecund than native Kokanee of the Williston reservoir, which developed
around 55°N (McGurk 2000). Examinations of the reproductive efforts and fecundity of native
Williston watershed Kokanee and Columbia-origin Kokanee, both pre-stocking and
contemporary populations, have confirmed that fecundity has decreased from historic (1990 and
1991) measurements for fish in the Williston Reservoir and future population growth may
therefore be affected (Wilson and Shrimpton 2021).
As an ultra-oligotrophic system, the Williston Reservoir is subjected to high annual
drawdown, outflow, and sedimentation (Stockner et al. 2005; FWCP 2020). Low levels of
limiting nutrients and a scarcity of zooplankton have resulted in an environment of poor
productivity, which may not support large populations of fish for extended periods of time
(Harris et al. 2005; Stockner et al. 2005; FWCP 2020). Logistical difficulties surrounding such a
large and complex lacustrine ecosystem have prevented widespread or frequent limnological
assessments, in-depth monitoring, and nutrient enhancement programs (Harris et al. 2005;
Stockner et al. 2005; FWCP 2020). Nutrient restoration programs in other large British Columbia
lentic systems such as Kootenay Lake and Arrow Lakes Reservoir have successfully bolstered
levels of zooplankton and primary productivity to support Kokanee fisheries (Askey 2016; Peck
et al. 2019), and similar programs are being investigated for the Williston Reservoir (FWCP
2020). However, both systems are now monitoring continued Kokanee population collapse due
to related factors, namely top-down predation effects (Askey 2016; Redfish Consulting Ltd.
2016; Peck et al. 2019; Warnock et al. 2021).

28

A recent 4-year acoustic telemetry study in the Williston Reservoir has focused on the
dramatic expansion of Lake Trout (S. namaycush) that occurred concurrently with Kokanee
population increases following stocking efforts (Culling et al. 2020). Stomach contents and field
observations indicated that Kokanee were opportunistically and actively targeted by Lake Trout,
and ongoing collaborative projects are examining the role of Kokanee occupancy in directing the
movement patterns of piscivorous species such as Lake Trout and Bull Trout (S. confluentus;
Culling et al. 2020). Members of the Tsay Keh Dene Nation have reported that while Bull Trout
numbers in the reservoir and tributaries have remained stable, the fish have noticeably decreased
in size (Pearce and Abadzadesahraei 2019). A prevalence of predator species such as Bull Trout
reaching smaller sizes has been identified as a characteristic of the ongoing Kokanee collapse in
Kootenay Lake, and may suggest a predator-prey imbalance (Warnock et al. 2021). While the
direct impacts of Lake Trout predation on Williston Kokanee have not been quantified, it is
possible that top-down predation effects are having an impact on Kokanee growth or seasonal
movement patterns across the reservoir (Peck et al. 2019; Culling et al. 2020; Hagen, FWCP,
personal communication).
Similar to predation effects, angler harvest and fisheries-induced changes can be
important mechanisms affecting growth and reproduction (Martinez and Wiltzius 1995;
Andrusak et al. 2000; Kendall et al. 2014). Recent surveys of local First Nations and recreational
and professional anglers have suggested that Kokanee are not prioritized in the Williston
Reservoir, due to their small size and low catchability (Coxson et al. 2017; Pearce and
Abadzadesahraei 2019; Pearce and Morgan 2019). The B.C. MoFLNRORD proposed a change
in Kokanee quota regulation for the Williston Reservoir in January of 2021, to increase the daily
quota from 4 to 10 fish (B.C. MoFLNRORD 2021). The potential for fishery-induced

29

overharvest or selective pressure from angler activity is therefore not of great concern at this
time, and has likely not been a substantial influence on the declining spawner size in the
watershed.
Genetic Effects on Size, Condition Factor, and Age
The results of my study may support a genetic component of age and size at maturity in
populations of Kokanee, in addition to adaptive responses to poor growth conditions in the
reservoir (Patterson et al. 2008). Native Williston fish captured in 1989, 1990, and 1991 were
significantly larger at maturity than all Columbia-origin cohorts, and it is unknown whether this
trend continues. Fish sampled from the reservoir in 2000 and genotyped as native Williston fish,
however, did not differ in size from the Columbia-origin fish caught at the same time, thereby
suggesting the reservoir environment may have been more of an influence on size at maturity
than genetic lineage alone. Nevertheless, I also found location to be a significant contributing
factor to differences in fork length and condition factor among sample years, even within the
same reproductive lineage (2006 and 2018). Germansen spawners, for instance, were
significantly larger than Pelly and Russel counterparts in 2006, and maintained a higher
condition factor than Pelly spawners in 2018. Differences in morphometrics between spawning
location may be indicative of shifting population genetics or local adaptation within the reservoir
(Kendall et al. 2014).
Additionally, there is a knowledge gap of the reproductive strategies used by Kokanee
populations in the Williston watershed. For instance, “silver” females that engaged in
prespawning waiting in a 2006 study of Meadow Creek Kokanee were found to have a shorter
fork length and were younger at maturity than females of the normal “red” phenotype (Morbey
and Guglielmo 2006). This suggested the utilization of a conditional reproductive strategy with

30

two discrete thresholds for size at maturity, both of which roughly equivalent in terms of lifehistory trade-offs (Morbey and Guglielmo 2006). Kokanee from the Meadow Creek Spawning
Channel were included in the Columbia stock that were introduced to the Williston Reservoir in
the 1990s, and such reproductive strategies may persist in contemporary Columbia stock. Further
research is therefore imperative to assess the importance of genetic and environmental factors on
the population trends of Williston Kokanee.
Conclusions
The purpose of this study was to analyze the observed trends of Kokanee size, condition
factor, and age at maturity in the Williston Reservoir, particularly following an extensive
stocking program in an ultra-oligotrophic environment. My results indicate that Kokanee
populations have experienced a decrease in length and condition factor at maturity over time, but
have not undergone a shift in generation time or overall age at maturity. Enumeration flights in
recent years have also documented what may be a reservoir-wide decrease in spawner
escapement compared to the first decade of the 2000s, and these factors combined may suggest
an overall decline in Kokanee following a period of overpopulation in the watershed. Further
enumeration flights and escapement estimates on a yearly basis are necessary to establish
appropriate data baselines, as Kokanee spawners have been known to undergo dramatic
fluctuations in escapement among cohorts and spawning periods (Andrusak 2007). My results
also indicate that spawning location and genetic lineage influenced spawner size and condition.
Further investigation into the population genetics of native and introduced stock could provide
information on the effects of genetic lineage on current Kokanee growth and survival and may
improve the understanding of the morphometric trends observed in my study.

31

CHAPTER 2: Genetic population structure of introduced and native lineages of
Oncorhynchus nerka in a large, impounded watershed
Introduction
A basic understanding of genetic population structure is a critical component of assessing
the status, trends, and distinct units for fisheries management (Taylor et al. 2014). Since the
completion of the W.A.C. Bennett Dam in 1968 and the subsequent formation of the Williston
Reservoir in northern British Columbia, Canada, the Upper Peace River watershed has been a
site of ongoing mitigation efforts—including a large-scale stocking program aimed at expanding
the presence and abundance of Kokanee (Oncorhynchus nerka) in the reservoir, primarily to
generate a successful sport fishery and also provide a prey source for large piscivorous fish
species (Blackman et al. 1990; Langston 2012; FWCP 2020). Columbia River stream-spawning
Kokanee from Arrow Lakes Reservoir (Hill Creek) and Kootenay Lake (Meadow Creek) were
stocked in tributaries to the Williston Reservoir from 1990 to 1998 (Langston and Murphy
2008); 72% (n = 2,370,741) of the total number of stocked Kokanee were sourced from Hill
Creek between 1990 and 1995, and the remaining 28% (n = 932,474) were sourced from
Meadow Creek from 1996 to 1998 (Langston and Murphy 2008). Over 3 million juvenile
Kokanee were stocked into five tributaries: Carbon Creek (n = 1,015,338), Dunlevy Creek (n =
152,487), Davis River (n = 75,000), Manson River (n = 1,396,305), and Nation River (n =
664,085; Figure 2.1); three systems on the east side of the reservoir and two rivers that flow into
the southwest portion of the reservoir, all of which are considered to be accessible to anglers
(Blackman et al. 1990; Langston and Murphy 2008). Ongoing enumeration studies have
documented a widespread distribution since the 1990’s (Langston 2012; McDermot-Fouts 2019;
Robinson 2020). In 2006, Kokanee were reported to spawn in at least 68 rivers and streams from
the Parsnip River watershed (southernmost reach) to the Finlay River watershed (northernmost
32

reach; Figure 2.1; Langston 2012). By 2008, it was estimated that the Kokanee population could
be up to 9 million fish, overtaking Lake Whitefish as the most abundant species in Williston
Reservoir (Sebastian et al. 2009). This is also a substantial increase in the estimated Kokanee
population from 1 million fish in the reservoir, as assessed by hydroacoustic measurements made
in 2000 (Sebastian et al. 2003). The rapid expansion of Columbia-origin Kokanee in the
Williston watershed offers an opportunity to examine the degree of genetic population
structuring of a new lineage of fish introduced to a watershed with historically low numbers of
native Kokanee.
Populations of native Kokanee exist in the Williston watershed currently and were
documented before impoundment of the Peace River. In the headwaters of the Parsnip River,
Arctic and Tacheeda Lakes have shore-spawning Kokanee (Figure 2.1) (Langston and Zemlak
1998). Thutade Lake in the headwaters of the Finlay River also has a population of Kokanee that
appears to be comprised of shore-spawning fish (Figure 2.1) (Blackman et al. 1990; Fielden
1991; McLean and Blackman 1991; Langston and Zemlak 1998). A native population of
Kokanee was established in the Williston Reservoir, and appeared to be growing in the years
prior to the stocking program; changes in relative abundance of Kokanee in the reservoir
increased from 0.1% of total fish gill netted in 1974, to 2.3% of captures in 1988 (Blackman et
al. 1990). Native Kokanee collected from the Finlay River in 1990 were morphologically distinct
from the stocked Columbia-origin Kokanee, with the former exhibiting an overall coloration of
dark reddish brown, and the latter displaying the bright red bodies and green heads consistent
with the source populations in Hill and Meadow Creek (Langston and Zemlak 1998). The origin
of these native Kokanee in the reservoir may be from one or both of the two potential sources

33

existing at the southern and northern regions of the watershed, and the extent of their current
distribution and abundance is unknown.
Interactions between introduced and native Kokanee may occur at a few key points in the
watershed. In the Parsnip reach, passage between the reservoir and Arctic and Tacheeda Lakes is
possible given adequate water levels (Langston 2012). In the Finlay reach, fish from the reservoir
are unable to move upstream over Cascadero falls into Thutade Lake, but native Kokanee from
Thutade Lake are capable of passing over the falls into the reservoir (Figure 2.1) (Fielden 1992;
Langston and Zemlak 1998). This partial lack of physical barriers, in conjunction with reported
upstream expansion of Columbia-origin spawners in years of high escapement (Langston 2012),
presents the potential for genetic introgression between native Kokanee in the Williston
Reservoir watershed and the introduced Columbia-origin fish.
Admixture or concurrent gene flow between introduced and native stock in a water
system is of particular interest in the context of fisheries management decisions. Introgression of
introduced genes into indigenous populations following stocking programs has been documented
in Kokanee populations (Yamamoto et al. 2011; Whitlock et al. 2018). However, populations of
both anadromous and non-anadromous forms of O. nerka have circumvented genetic admixture
following stocking programs due to mechanisms such as temporal isolation (Young et al. 2004;
Whitlock et al. 2018), spatial isolation or associative mating (Foote et al. 1989; Wood and Foote
1996), or a combination of both temporal and spatial restrictions (Burger and Spearman 1997). A
preliminary review of Kokanee in the Williston Reservoir was undertaken in 2007 (Withler
2007), but a comprehensive understanding of the contemporary population structure of Kokanee
across the Williston watershed is required for appropriate and ongoing species management. In

34

particular, the identification of unique or diverging populations is critical for the conservation of
genetic diversity in highly-managed systems (Taylor and Gow 2010).
I examined the genetic population structure of Kokanee sampled from across a broad
temporal and spatial range of the Williston watershed. I assessed the current status of native
Kokanee, the extent of interactions between Columbia-type and native Kokanee, and the degree
of differentiation between groups of introduced Columbia Kokanee spawning in tributaries to the
reservoir. I used microsatellite DNA to determine the genetic differentiation within and among
sampling groups. Introgression between native and introduced lineages of Kokanee in the
Williston watershed would not be inherently harmful to the success of the species as a whole, but
may indicate a threat to the persistence of native genetic signatures (Veale and Russello 2016).
Genetic distinction between spawning groups of Columbia-origin Kokanee could inform whether
management decisions would be more effective at the level of genetic units or as a species
collective (Taylor and Gow 2010; Taylor et al. 2014; Moreira and Taylor 2015). My work
provides important insights into the fine-scale genetic impacts of past stocking programs
implemented in this large, impounded watershed.
Methods
Sampling
A total of 1,870 Kokanee samples were collected from the Williston reservoir, tributaries
to the reservoir, surrounding lakes, and from the source populations in the Columbia River
watershed between 1988 to 2019 (Table A2.1). In the sampling that occurred in 2018 and 2019,
ten tributary streams and three lakes that support native Kokanee populations were selected to
represent the full geographic range of Kokanee in the Williston watershed, including areas where
genetic introgression with native populations may occur (Figure 2.1). Mature spawners were

35

collected by DWB Consultants during the peak spawning season in September of 2018, and in
September and late October in 2019. Capture method varied depending on the water body type;
stream sampling was accomplished predominately through the use of a backpack electrofishing
unit (Model S-R 12-B, Smith-Root Inc., Vancouver, WA, USA) and stunned fish were collected
with a seine or dip net. The shore-spawning Kokanee of Tacheeda and Arctic Lakes in the
Parsnip River watershed were sampled with sinking and floating gill nets (McDermot-Fouts
2019; Robinson 2020). Fish were anesthetized in a solution of 100 mg · L –1 tricaine methane
sulfonate (MS-222) buffered with sodium bicarbonate, euthanized by concussion and
exsanguination, and frozen at -20°C until further processing (McDermot-Fouts 2019; Robinson
2020). I collected muscle tissue or adipose fin samples from Kokanee individuals, stored them in
95% ethanol at -20°C, and submitted them for genotyping to the Pacific Biological Station (PBS)
Molecular Genetics Lab, Nanaimo, BC.
Additionally, scale samples from Kokanee gill netted in the reservoir in the late 1980s
before the stocking program, in the early 1990s shortly after the stocking program was initiated,
and in 2000, a decade after stocking, were provided to UNBC by the FWCP – Peace Region.
Individuals collected from the G.M. Shrum intake towers of the W.A.C. Bennett Dam (Figure
2.1) in 2016 and 2019 were donated by Carleton University and Mr. R. J. Zemlak for inclusion in
this study. Genotypes of Kokanee specimens that were collected by the FWCP from spawning
tributaries to the reservoir in 2003–2006 and 2010 were provided by Ruth Withler of PBS.

36

Figure 2.1. The Williston Reservoir and associated reaches of the watershed. Red symbols signify the original stocking
locations of Columbia-origin Kokanee in the watershed (Langston 2012). Blue symbols are locations of sample sites of
Kokanee that spawn in tributary streams to the reservoir. Purple symbols are locations where native Kokanee were
found in the Parsnip Reach; pink symbols indicate native Kokanee in the Finlay Reach. Orange symbols are the
approximate locations of within-reservoir sampling as they occurred in 2000 (Pillipow and Langston 2002). Samples
collected from the G.M. Shrum intake towers in 2016 and 2019 were from the site closest to the W.A.C. Bennett Dam.

37

Microsatellite Markers
For the archived scale samples, I conducted an experimental round of DNA extraction
from Kokanee scales from fish caught in 2006 that had been archived in freezers by the B.C.
Ministry of Forests, Lands, Natural Resource Operations and Rural Development
(MoFLNRORD) and from fish caught in 2018 by DWB. I determined via the below procedure
that an optimal number of 4–8 scales was necessary to obtain a target concentration of DNA (10–
100 ng/µl) for polymerase chain reaction (PCR) amplification. Four to eight scales were
collected from each archival envelope with no fewer than 4 scales used for any one sample. I
performed DNA extractions from the scale samples using QIAamp DNA Micro® kits (Thermo
Fisher Scientific, Waltham, MA, USA) following the manufacturers protocol, with the
modification of overnight (approx. 12 hrs) incubation at 37°C. I procured two elutions of 50 μl
per sample, with the first elution submitted to PBS for genotyping and the second reserved at 20°C at UNBC for any future analyses. To ensure that my protocols were appropriate, I also
conducted a check of DNA quantity and quality, described below.
To assess the quantity of DNA obtained from scales, I measured the extraction products
of four samples of each sample group from 1988 to 2000 with a Qubit® Fluorometer (Thermo
Fisher Scientific, Waltham, MA, USA) dsDNA HS assay. To assess quality, I amplified DNA
using PCR for these four samples with primers for the microsatellite loci Ogo2 (Olsen et al.
1998). I conducted these PCR reactions in volumes of 10 µl to keep the amount of DNA required
at a minimum, and reaction conditions for each were as follows: 3 µl Milli-Q PCR H2O; 5 µl 2 X
Type-it Mix; 1 µl Ogo2 primer cocktail (0.07 µM each of forward and reverse primers); and
approximately 20–200 ng of DNA in 1 µl were added as a template for each reaction. PCR was
conducted in an MJ Research thermal cycler (MJ Research, Inc., Watertown, MA, USA) with
38

one cycle of 5 min at 94°C; one cycle of 30 sec at 94°C; one cycle of 90 sec at annealing
temperature (Ta) 58°C; one cycle of 30 sec at 72°C; a repeat of cycles 2–4 an additional 37
times; one cycle of 30 min at 60°C; followed by an indefinite holding temperature of 4°C. PCR
products, negative controls, and a MassRuler Express LR Forward DNA Ladder Mix were
electrophoresed in a 2% agarose gel at 100 V for 1.5–2 hours. I compared product fragment sizes
with the DNA ladder to ascertain the quality of these older scale samples before the bulk of the
DNA elutions were submitted to PBS.
For the DNA extraction of muscle tissue at PBS, a Chelex® (Bio-Rad Laboratories, Inc.,
Hercules, CA, USA) protocol was used. Approximately 2.5 mg of tissue was digested in 1 X TE
(10 mM Tris-HCl, 1 mM EDTA pH 8.0) containing 50 mg/ml of Chelex resin and 0.2 mg/ml of
Proteinase K. The tissue and Chelex solution was incubated at 50°C for 15 min, followed by
95°C for 15 min. 10 µl of the resulting DNA elution was diluted with 70 µl of 1 X TE containing
16 µg of Proteinase K and incubated at 50°C for 15 min, followed by 95°C for 5 min.
PCR was conducted in Biometra TAdvanced thermal cyclers (Biometra GmbH,
Göttingen, Germany) for 14 established microsatellites that are effective in distinguishing
populations of O. nerka genetic distance estimates (Beacham et al. 2010; Beacham and Withler
2017). Primer sequences and Ta are given in Table 2.1. The reverse primers for each locus were
modified with a GTTT consensus tail. Microsatellite loci were amplified singly or multiplexed in
6 µl volumes, containing 1 X Qiagen PCR buffer, 84–168 µM of each dNTP, 1.6–3.7 mM
MgCl2, 117–755 nM each of forward and reverse primers, 0.2 U Qiagen Hotstar Taq, and 1 µl of
1:8 diluted DNA. PCR cycling conditions were as follows: one cycle of 15 min at 95°C; one
cycle of 20–30 sec at 94°C; one cycle of 30–45 sec at Ta; 31–38 cycles of 30–60 sec at 70°C;

39

followed by a holding period of 10 min at 72°C. Amplified microsatellite fragments were sizefractioned under denaturing conditions with the 3730 Capillary Electrophoresis DNA Analyzer
(Applied Biosystems, Foster City, CA, USA) and allele sizes were determined using
GeneMapper 5.0 software (Applied Biosystems, Foster City, CA, USA).
Microsatellite genotypes were tested for duplicates at PBS using Microsatellite Toolkit
(Park 2001). Duplicated genotypes in the Russel Creek 2019 (n = 1) and Thutade Lake 2003 (n =
9) groups were excluded from all analyses as they were most likely the result of duplicated
sampling or cross-sample contamination. The duplicated genotypes in Tacheeda Lake 2004 (n =
17) were retained because of the overall low genetic diversity of the sample group; these were
thought to be the result of family structure in the group.
Many locations were sampled multiple or consecutive years. To determine significant
differences between sample years I performed an analysis of molecular variance (AMOVA) in
GenAlEx version 6.51b2 (Peakall and Smouse 2012), using 999 permutations. I combined
groups by sample location in instances of low FST. The total number of sample groups was
reduced from 42 to 21, provided in Table 2.2. Note that the terminology “group” will be used to
refer to these 21 samples, and “population” will be reserved for discussions of genetic population
structure. Unless otherwise specified, “native groups” refers to four locations where Kokanee
were sampled: Arctic Lake (Arctic), Tacheeda Lake (Tacheeda), fish sampled in the reservoir
before stocking with Columbia River Kokanee (Native Rsvr.), and Thutade Lake (Thutade).
Likewise, “tributary groups” refers to the fish collected from 13 tributaries to the reservoir where
mature spawning Kokanee were collected (groups 5–17 in Table 2.2).
Microsatellites are often prone to null alleles that result from base pair mutations at the
PCR priming site (Banks et al. 1999; Holm et al. 2001). Both large allele dropout and the
40

presence of null alleles may contribute to inaccurate amplification of certain loci between
different populations (Banks et al. 1999). I tested for the presence of null alleles with the
program MICRO-CHECKER version 2.2.3 (van Oosterhout et al. 2004). Across all loci and sample
groups, there was evidence of 3 significant null alleles present due to homozygote excess: locus
Ots3 in the Aley group (0.086, P < 0.001); and Oki10 (0.057, P < 0.01) and Ots3 (0.080, P <
0.001) in the Thutade group. Frequencies below 0.10 were maintained for all 3 null alleles. There
was no evidence for scoring error (stuttering) or allele dropout at any locus.
The Rsvr. 2000 group indicated nonsignificant evidence for 6 null alleles (row 19 of
Table A2.2). As Kokanee in this group were collected from the reservoir rather than a discrete
spawning channel (Pillipow and Langston 2002), there is a possibility that this evidence of null
alleles may indicate a Wahlund effect; when samples are collected from a mixed catch, such as
spawning lakes or along spawning migration routes, the resulting analysis can show a Wahlund
effect with varying gene frequencies and an overabundance of homozygotes (Khrustaleva et al.
2017). For salmon populations with relatively short evolutionary timeframes (10,000–15,000
years), null alleles may not hinder analyses of genetic population structure (Small et al. 1998).
More null alleles may also be found when analyzing populations that are more genetically
distinct, rather than local subpopulations (Small et al. 1998), as may be the case for this mixed
sample group. Huang et al. (2016) advised that all loci with null allele frequencies greater than
0.50 should not be used in genetic analyses regardless of applied corrections, as the genetic
information provided by such loci would be less than half of a typical microsatellite. However,
null allele frequencies equal to or less than 0.10 may be acceptable to use without applied
corrections (Huang et al. 2016). The 3 significant null alleles maintained frequencies below 0.10,

41

and the 6 nonsignificant Rsvr. 2000 null alleles showed frequencies between 0.06–0.16.
Therefore, all loci were maintained in subsequent genotypic analyses.
I tested for linkage disequilibrium between pairs of loci within each group in ARLEQUIN
v3.5.2.2 (Excoffier and Lischer 2010) and for significant deviations from Hardy-Weinberg
equilibrium in each group using an exact test based on 1,000 Monte Carlo permutations of alleles
in GENEPOP v1.1.7 package for R v4.0.3 (Guo and Thompson 1992; Rousset 2008). Corrections
were applied for multiple simultaneous tests using the modified false discovery rate (FDR)
procedure of Benjamini and Hochberg (1995), as recommended by Narum (2006).

42

Banks et al. 1999

Beacham and Margolis 1998; Nelson and Beacham 1999

Smith et al. 1998; Nelson et al. 2003

Scribner et al. 1996

Morris et al. 1996

2

3

4

5

CTGTTCTGCTCCAGTGAAGTGGA
GGCCTTCCAGCAGAGAGTTA
CTGTTCTGCTCCAGTGAAGTGGA
TGCAGTGAAGCCTTAAAGAC

CTGTTCTGCTCCAGTGAAGTGGA
CAACTAGACCCAGCCTCACAG
AACATTCTGGGATGACAGGGGTA
CGTTCTCTACTGAGTCAT

Oki163
Oki293
One84
Omy775

1

CACCCATAATCACATATTCAGA
AACAGACAGCTAATGCAGAACG
CTGTTCTGCTCCAGTGAAGTGGA

AGGATGGCAGAGCACCACT
TCAACAGATAGACAGGTGACACA
CTGTTCTGCTCCAGTGAAGTGGA

Oki1b3
Oki63
Oki103

ATAGAGACCTGAATCGGTA
TGGCAAGGAGAGAGACAGAGGG
CACCCATAATCACATATTCAGA

CAGTGTAAGGATATTAAA
CTTCCATTGTGATTCT
ACGGACGTGCCAGTGAG
GACATAGCGTTCAGCACAG

ACACCTCACACTTAGA
CACACTCTTTCAGGAG
TGAACATGAGCTGTGTGAG
AGGCTCTGGGTCCGTG

Ots21
Ots31
Ots1002
Ots1032

Ots1072 ACAGACCAGACCTCAACA
Ots1082 TTTCTATTAGTCTGTCACTAC
Oki1a3 AGGATGGCAGAGCACCACT

Reverse Sequence

Forward Sequence (5'–3')

Locus

55
50
62
58

55
50
55

55
52
55

62
55
56
55

Ta (°C)

265–468
290–469
174–291
87–135

129–181
82–177
152–750

79–155
100–294
92–130

129–235
64–122
125–337
105–340

Size Range

52
41
44
21

11
38
94

16
32
9

30
27
34
31

A

43

Table 2.1. Forward and reverse sequences, annealing temperature (Ta), and allelic size ranges of 14 microsatellite loci for Oncorhynchus nerka.
A delineates the number of alleles observed across all sample groups of the comprehensive dataset of 1,870 genotypes.

Arctic Lake

Tacheeda Lake

Hill Creek

Meadow Creek

Carbon Creek

Dunlevy Creek

Manson River

Germansen River

Osilinka River

Pelly Creek

Finlay River

Tsaydiz Creek

Russel Creek

Cutoff Creek

Bower Creek

Aley Creek

Stevenson Creek

Rsvr. Forebay

Rsvr. 2000

Native Rsvr.

Thutade Lake

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

C, N

C, N

C

C

C, T

C, T

C, T

C, T

C, T, TS

C, T

C, T

C, T, TS

C, T

C, T, TS

C, T

C, T

C, T

C, Donor

C, Donor

C, N

C, N

Dataset

36

36
1

1988

34

34
1

1989

15

15

1

1990

86

23
1

631

1994

Tissue samples from whole fish collected by Chu Cho Environmental for the FWCP – Peace Region

3

Scale samples provided by Chelsea Coady (FWCP – Peace Region)
2
Genotypes of spawners collected by the FWCP – Peace Region; provided by Ruth Withler (PBS)

1

Total

Location

Group

283

87

2

1962

2003

150

1002

502

2004

389

50

2

49

2

49

2

92

2

49

2

50
2

502

2006

60

602

2010

134

95

37

20

3

7
6

10

3

10
3

3

2017

3

16

13

3

2016

40

402

41

5

40

5

40

5

40

5

40

6

5

5

199

40

42

39

20

5

1,870

107

85

45

135

17

109

50

40

128

49

112

89

13

130

41
5

41

64

63

345

100

80

68

Total

5
5

185

2019

5

41

494

404

305

2018

44

Tissue samples from spawners collected by Matt Neufeld and Steven Arndt (B.C. MoFLNRORD)
5
Tissue samples from spawners collected by DWB Consultants, Inc. for the FWCP – Peace Region
6
Tissue samples collected by Randy Zemlak (BC Hydro), Steven Cooke, Dirk Algera, and Taylor Ward
(Carleton University)

4

45

45

1

2000

Table 2.2. Sample groups of Williston watershed Kokanee genotypes by sample location and year. Rsvr. 2000 is retained as a separate group because of an excess of
homozygotes that indicates a Wahlund effect (Khrustaleva et al. 2017). Genotypes are organized into four datasets: comprehensive (C; n = 1,870; groups 1–21); tributary
(T; n = 905; groups 5–17); time-series (TS; n = 347; groups 8, 10, and 13 organized by sample year 2006, 2018, 2019); and native (N; n = 340; groups 1–2, 20–21).
Donor populations are groups 3–4; fish collected from the body of the reservoir are groups 18–20.

Genetic Variability
I organized and evaluated genotypes in four datasets, summarized in Table 2.2:
comprehensive (n = 1,870; groups 1–21); tributary (n = 905; groups 5–17); time-series (n =
347; tributary locations grouped by sample year 2006, 2018, 2019); and native (n = 340;
groups 1–2, 20–21). To evaluate genetic variation, I performed basic population and genetic
descriptive statistics in R. I calculated expected heterozygosity (He), observed heterozygosity
(Ho), and mean allelic richness (AR) using HIERFSTAT v0.5–7 package (Goudet et al. 2020). I
determined the number of alleles (A) with ADEGENET v2.1.3 package (Jombart 2008; Jombart
and Ahmed 2011). I also used ADEGENET to assess the average inbreeding coefficient (F) for
the native groups, and compared these with the F estimates for the tributary groups.
To assess population subdivision of the sample groups, I constructed a neighborjoining tree (Saitou and Nei 1987) using Cavalli-Sforza and Edwards’ (1967) chord distance
(DC) as a measure of genetic distance between populations. The analysis incorporated 10,000
bootstrap replications using TREEFIT v1.2 (Kalinowski 2009) and the tree was visualized in
FIGTREE v1.4.4 (Rambaut 2018). The use of DC requires no assumptions about evolutionary

models or rates of mutation, and has been shown to consistently perform well in obtaining
phylogenies (Takesaki and Nei 1996). However, the infinite-alleles model associated with
Nei’s distance measurements may be more appropriate for populations that are not as closely
related, and for which genetic drift may not be the driving factor of population differentiation
(Felsenstein 1985; Excoffier 2003; DoĞAn and DoĞAn 2016). Therefore, I calculated Nei’s
(1987) pairwise distance estimates (FST) for the native dataset. For the comprehensive
dataset, I evaluated FST, estimated as ! (Weir and Cockerham 1984), in HIERFSTAT v0.5–7
package for R. Analyzing the tributary dataset separately produced identical results to the

45

comprehensive dataset. I used 10,000 bootstraps with a lower and upper quantile for
confidence intervals of 0.025 and 0.975 to determine significant genetic distances between
sample groups (function boot.ppfst). I calculated global F-statistics for all groups with
GENEPOP v1.1.7 package for R and FSTAT v2.9.4 (Goudet 1995; 2003).

Genetic Structure
The Bayesian methods in the program STRUCTURE v2.3.4 (Pritchard et al. 2000) are
the most widely used to evaluate population structure across datasets with multiple loci (Earl
and vonHoldt 2011). Sampling from multiple potential source populations allows for the
assessment of the number of clusters (K) that most likely represents real world population
structure, reported with assigned values of likelihood (Pritchard et al. 2000). STRUCTURE
sequentially imposes population structure and groups genotypes by clusters that maintain
assumptions of Hardy-Weinberg equilibrium and unlinked loci, and allows for the
assignment of admixed individuals to a particular population (Pritchard et al. 2000). I used
STRUCTURE to determine the number of clusters and inferred population structure of all

genotypes. The models I used were correlated allele frequencies and admixture, with a burnin period of 100,000 followed by 300,000 iterations. This duration can be considered
medium-length among salmonid population genetics studies (Hansen and Mensberg 2009;
Yamamoto et al. 2011; Frazer and Russello 2013; Aunins et al. 2014; Moreira and Taylor
2015). These protocols were followed by Withler (2007), and were chosen here with the
understanding that substantially increasing the number of iterations may not proportionally
change the results (Vaha and Primmer 2006).
For the comprehensive dataset I tested K of 1–21 and replicated each K-value ten
times in order to survey all potential structure at the genetic population level. I used sample

46

location priors in an attempt to analyze population structure with slight geographic bias to
assist with assigning K. I identified the most effective value of K (DK) ad hoc via the webbased program STRUCTURE HARVESTER (Evanno et al. 2005; Earl and vonHoldt 2011). I
visualized cluster results and likelihoods with Cluster Markov Packager Across K (CLUMPAK;
Kopelman et al. 2015). I evaluated the tributary, time-series, and native datasets
independently and without location priors in STRUCTURE to evaluate potential subpopulation
structure, genetic drift between different spawning locations, or evidence of introgression
with native groups.
Analyzing genetic information with programs such as STRUCTURE in conjunction with
a Principle Component Analysis (PCA) and other multivariate models (Discriminant
Analysis) without assumptions may be the most effective combined method for accurately
recreating real-world population structure (Porras-Hurtado et al. 2013). PCA allows for the
assessment of populations without a priori knowledge of the structure of biological
populations, and it does not require assumptions of Hardy-Weinberg equilibrium or linkage
disequilibrium. A Discriminant Analysis of Principal Components (DAPC) replicates
Bayesian clustering methods by assigning populations to particular discriminated clusters,
and provides a visual representation of the pattern of population partitioning (Jombart et al.
2010). I visualized the genetic differentiation of the comprehensive dataset with a DAPC to
assess structure patterns and identify clusters with maximized among-population variation. In
addition, I separately analyzed the native dataset to identify the genetic signature of native
fish that historically occupied the reservoir (Native Rsvr.) and to assess the degree of relation
to native fish that persist in Thutade, Arctic, and Tacheeda Lakes. I analyzed the tributary
dataset to discern potential differences between Kokanee homing to different spawning

47

locations. Finally, I examined the time-series dataset, in which Columbia-type Kokanee in
the Williston watershed were grouped by collection year: 2006 (Germansen, Pelly, and
Russel), 2018 (Germansen, Pelly, and Russel), and 2019 (Germansen and Russel). These
groups were restricted to Germansen, Pelly, and Russel to maximize potential temporal
changes and minimize spatial differences. I assessed all datasets with ADEGENET v2.1.3
package for R, and the optimum number of principal components (PCs) was cross-validated
with the lowest root mean squared error (function xval), as recommended by Jombart and
Collins (2015).
Results
Microsatellite Markers
All groups with multiple sample years per location (Table A2.1) did not show
substantial genetic deviation; all FST values were close to zero or negative (FST ≤ 0.011),
indicating random mating under Hardy-Weinberg equilibrium (Table 2.3). I retained the
Rsvr. 2000 group separately from the rest of the reservoir groups (FST = 0.082), while those
from before (1988–1990; Native Rsvr.) and after (2016, 2019; Rsvr. Forebay) were
respectively combined (Table 2.2). Group sample sizes ranged from 13 (Osilinka) to 345
(Meadow) individuals.
Tests for linkage disequilibrium between pairs of loci within each group and
corrected for Benjamini-Hochberg FDR resulted in statistically significant departures in 37
out of 1,911 tests (FDR ɑ = 0.00088), but departures were not concentrated on specific locus
pairs (Kruskal-Wallis test, P = 0.93) and were less frequent than theoretically-expected.
Therefore, each locus likely represents an independent measure of genetic variation and
divergence. Multiple tests (294) were conducted to identify deviations from Hardy-

48

Weinberg, and after corrections 35 tests deviated from Hardy-Weinberg equilibrium by way
of heterozygote deficiency (FDR ɑ = 0.0045). However, these deviations were not
concentrated on specific loci (Kruskal-Wallis test, P = 0.36) or groups (ANOVA, P = 0.85).
All loci were polymorphic and retained in analyses.
Genetic Variability
The mean He (± standard error) averaged across all 14 loci and 21 groups of the
comprehensive dataset was 0.66 ± 0.022 (Table 2.4), and ranged from 0.38 ± 0.064
(Tacheeda) to 0.72 ± 0.049 (Hill) and 0.72 ± 0.047 (Osilinka). Of the tributary groups, mean
He was 0.70 ± 0.002. The mean Ho for all loci and groups was 0.65 ± 0.021. Tacheeda
reported the lowest Ho of 0.38 ± 0.066, with 0.71 ± 0.049 (Hill) and 0.71 ± 0.043
(Germansen) on the higher end. The mean allelic richness (AR) across loci and groups was
5.99 ± 0.258, and ranged within groups from 2.82 ± 0.454 (Tacheeda) to 6.97 ± 0.944
(Meadow). The AR for tributary groups all exceeded 6.00, while the native groups averaged
3.93 ± 0.701 AR. The greatest number of alleles for a group was sampled from Meadow
(215), with Germansen (178) and Thutade (152) reporting the greatest number of alleles for
the tributary groups and native groups, respectively. The native groups consistently reported
the lowest genetic diversity among all the groups, and the putative source populations of Hill
and Meadow exhibited some of the highest genetic diversity and allelic richness. Thutade had
higher mean allelic richness and expected heterozygosity (AR = 5.93 ± 0.793; He = 0.64 ±
0.059) than other native groups, the values of which being more in-line with tributary groups
than the other native Kokanee. The genetic diversity of the tributary groups fell close to but
slightly lower than the values of the Columbia River source populations.

49

I estimated the mean inbreeding coefficient (F) of native and tributary groups to
investigate the low genetic diversity observed in the former (Figure 2.2). Tacheeda reported
the highest mean F (0.246), followed by Arctic (0.234). The Native Rsvr. (0.205) and
Thutade (0.182) groups had slightly lower F values, and the combined 13 tributary groups
averaged the lowest inbreeding coefficient value (0.165).
The neighbor-joining tree of Cavalli-Sforza and Edwards’ chord distances (DC) for
the comprehensive dataset clustered the tributary groups largely together and with Hill and
Meadow (84%; Figure 2.3). The Osilinka and Stevenson groups appeared somewhat apart
from the rest of the tributary groups (73%), but this could have been due to their low sample
sizes (Osilinka = 13 individuals; Stevenson = 17 individuals) rather than a true reflection of
genetic distance. The Rsvr. Forebay group, comprised of Kokanee that were collected from
the G.M. Shrum intake towers near the W.A.C. Bennett Dam in 2016 and 2019, also
clustered with the tributary groups. Notably, the Rsvr. 2000 group was placed outside of the
tributary groups (100%) but also separate from the nearby Thutade and Native Rsvr. groups
(95%). Arctic and Tacheeda clustered together apart from the rest of the samples (100%).
The global FST across all loci and groups was 0.1017 (Table 2.5) for the
comprehensive dataset. Examining the genetic distances in the native dataset with Nei’s
(1987) pairwise FST, all distances were significantly different from zero (Table A2.3). The
largest genetic distance was between Tacheeda and Native Rsvr. (0.404), and the smallest
was between Arctic and Tacheeda (0.049). The genetic distance between Thutade and the
Native Rsvr. group was 0.080, and these two groups were more genetically similar to each
other than either were to Arctic and Tacheeda (Table A2.3).

50

The genetic distance values between native groups in the comprehensive dataset were
slightly different in the context of Weir and Cockerham’s (1984) ! estimates, but the

relationships remained consistent (Table A2.4). The genetic distances between Arctic or
Tacheeda groups and all Columbia-type groups were significantly different from zero and
ranged from 0.330 (Arctic–Germansen; Arctic–Finlay; Arctic–Russel) to 0.410 (Tacheeda–
Stevenson). Likewise, the Meadow, Rsvr. 2000, Native Rsvr., and Thutade distance values
were all significantly different from zero. Most striking, however, was the lack of genetic
distance between all tributary groups, the ! estimates of which ranged from 0.000 to 0.004.
These groups were generally more similar to Hill than to Meadow (Table A2.4). The !

estimates of the Rsvr. 2000 group were shown to be the largest between Arctic (0.336) and
Tacheeda (0.347), and roughly equidistant between the source groups (0.048–0.083),
tributary groups (0.036–0.053), and the Native Rsvr. (0.048) and Thutade (0.051) groups.
Genetic Structure
The STRUCTURE analysis of the comprehensive dataset indicated a DK value of 2
clusters (Table A2.5) and a Markov clustering (MCL) similarity of 0.994 to describe the
population structure of the Williston watershed Kokanee. In a two-cluster system, samples
from Arctic, Tacheeda, Thutade, and Native Rsvr. clustered together as one genetic
population, and all other samples comprised the other genetic population (Figure 2.4). K = 2
was rejected as an instance of oversimplification that can be prevalent in studies using
STRUCTURE and the Evanno protocol (Janes et al. 2017). The next most-supported cluster

probability, K = 4 (MCL = 0.969), segregated Arctic and Tacheeda to one cluster, Thutade
and Native Rsvr. fish to a separate cluster, and differentiated between the two putative source
populations of Columbia-origin Kokanee. The Native Rsvr. fish were differentiated from

51

Thutade at the third most-supported cluster probability, K = 5 (MCL = 0.959). STRUCTURE
plots showing a range of K are included to demonstrate the emergence of informative
patterns with increasing K (Figure 2.4).
Examining K = 5, the samples from Hill creek showed signs of admixture with
Meadow creek signatures (Figure 2.4). Since initial stocking in the Williston reservoir,
Meadow creek signatures did not appear to have persisted as a distinct genetic population in
the watershed; while the tributary groups contained different proportions of Meadow-type
signatures, the large majority of these groups were dominated by Hill-type signatures. The
Rsvr. Forebay group also contained only Hill-type individuals. However, the Carbon group,
containing Kokanee that were collected in 1994, shortly after the stocking program began,
contained three Native Rsvr. individuals. This was the only apparent indication via
STRUCTURE of native signatures present in a group related to a tributary. However, these fish

were gill netted in the embayment approximately 1 km from the mouth of Carbon Creek.
In the tributary dataset, K of 1–4 were evaluated. A DK value of 2 clusters (Table
A2.6) and a MCL similarity of 0.997 was most supported by the Evanno protocol (2005).
However, with increasing K the tributary groups did not cluster into distinct genetic
populations but instead behaved as a single group; clusters were spread across all tributary
groups as one genetic population (Figure 2.5). Therefore, the most appropriate number of
clusters to assign to this subset was K = 1 (MCL = 1.000).
In the time-series dataset, K of 1–4 were assessed. Similar to the tributary dataset,
these groups did not cluster into different genetic populations, but rather one genetic
population was further admixed across all groups with increasing number of clusters (Figure
2.6). A value of K larger than 1 (MCL = 1.000) could not be supported.
52

A K of 3 or 4 were equally supported (MCL = 0.997) in the separate STRUCTURE
analysis of the native dataset (Arctic, Tacheeda, Native Rsvr., Thutade, and the Rsvr. 2000
group; Figure 2.7). However, based on the results of the comprehensive dataset, the Rsvr.
2000 group clearly contained individuals of both introduced Columbia-type signatures and
Native Rsvr. genotypes. Therefore, a K of 4 best described the native dataset, and the Rsvr.
2000 group was further analyzed independently. I tested Rsvr. 2000 with a range of K from
1–5; in this instance, a K of 2 clusters (MCL = 0.999), rather than K = 3 (MCL = 0.985), was
both the most-supported and the most likely to be representative of true genetic structure
(Figure 2.8). With K increasing past 2 clusters, the individuals that appeared to be Columbiaorigin splintered into admixed signatures rather than separate genetic populations.
For the DAPC analysis of the comprehensive dataset, the value of K associated with
the lowest Bayesian Information Criterion (BIC) value was somewhat subjective across a
range of values (Figure A2.7). Therefore, a number of K values (K = 2–10) were assessed. Of
the possible 300 PCs that could be used to assess variance, I retained 80 PCs and 3
discriminant functions that explained approximately 90% of the dataset.
Similar to the K = 4 STRUCTURE analysis for the comprehensive dataset, the Native
Rsvr. Kokanee signature was masked by Thutade genotypes in a four-cluster system. With
trials of increasing K from 4–8, Hill-type Kokanee split into multiple overlapping clusters
before the Thutade and Native Rsvr. were differentiated. Therefore, a K of 4 was retained as
the most parsimonious option. Genetic signatures for the Hill group and all tributary groups
were largely contained to Cluster 1; the Meadow group aligned with Cluster 2. Samples from
Thutade and Native Rsvr. clustered together in Cluster 3, while Arctic and Tacheeda fish
partitioned to a combined Cluster 4 (Figure 2.9). In this analysis, the Rsvr. 2000 group was

53

also shown to be a mixture of individuals assigned to Clusters 1–3 (Figure 2.10). Hill-type
signatures dominated the tributary groups, but Meadow-type signatures were more
widespread among tributary groups than was reflected in the STRUCTURE analysis. In addition
to the Carbon group, Germansen, Russel, Bower, and Aley all contained individuals that
were partially or predominately assigned to Cluster 3 (Native Rsvr. / Thutade). It should also
be noted that an individual that clustered with Native Rsvr. / Thutade was found in the
Meadow group—which was sampled from the Columbia River (Figure 2.9). This could
therefore indicate a statistical similarity of signatures between Meadow and Thutade / Native
Rsvr. genotype sequences. Additionally, an individual in the Rsvr. Forebay group was
assigned to Cluster 3 (Figure 2.10)—but a review of the membership probability values
assigned by DAPC indicated this individual’s cluster composition was 43.2% Cluster 1
(Hill), 53.4% Cluster 2 (Meadow), and 3.4% Cluster 3 (Thutade / Native). The remainder of
the individuals in the Rsvr. Forebay group showed membership probabilities for Cluster 3
that were less than 0.005%. I therefore considered this assignment erroneous, and perhaps
likewise associated with a statistical similarity between Meadow and Thutade / Native Rsvr.
genotypes.
A separate DAPC was conducted for Rsvr. 2000, Native Rsvr., Thutade, and Rsvr.
Forebay to assess the extent to which these groups would differentiate apart from the
comprehensive dataset and to distinguish to which genetic clusters Rsvr. 2000 would be
assigned. A maximum of K = 12 clusters was examined with 40 PCs and 2 discriminant
functions that explained approximately 90% of the overall variance; K = 3 aligned with the
lowest BIC value. In both the PCA and the DAPC, three clusters were easily differentiated
(Figure 2.11), confirming that Thutade and Native Rsvr. groups form distinct genetic clusters

54

(despite the K of 4 for the comprehensive dataset). The Rsvr. Forebay group clustered apart
from Thutade and Native Rsvr. groups, which supports the STRUCTURE result that this group
was comprised of only Columbia-type Kokanee, and is not a mixed group despite being
sampled in the body of the Williston reservoir. The Rsvr. 2000 group contained members that
aligned with Columbia-origin, Thutade, and Native Rsvr. clusters—a further confirmation
that this sample group was a mixture of different genetic populations in the Williston
reservoir.
Kokanee organized into the tributary and time-series datasets were analyzed,
respectively. In both instances, a PCA showed an entirely scattered dataset (Figure A2.8)
with no optimal BIC value that would be indicative of meaningful population structure. In
multiple trials, the K values that I selected based on knowledge of the Williston system
resulted in indiscriminate and random collections of individuals in each cluster. The notion of
genetic differentiation revealed by spawning location or reproductive cohort was, therefore,
not supported by the DAPC analyses.

55

Table 2.3. Results of multiple AMOVA between groups of Kokanee sampled from the same
locations, given as FST and associated P value.

Group
Arctic
Tacheeda
Hill
Meadow
Dunlevy
Germansen
Pelly
Finlay
Russel
Aley
Stevenson
Rsvr. Forebay
Native Rsvr.
Thutade
1

Years
2006, 2019
2004, 2018
2010, 2018
2003, 2004, 2018
1994, 2018
2006, 2018, 2019
2006, 2018
20061, 2019
2006, 2018, 2019
2016, 2017, 2018, 2019
2016, 2017
2016, 2019
1988, 1989, 1990
2003, 2017

FST
-0.008
0.011
0.001
0.000
0.006
0.003
0.002
0.000
0.002
0.002
-0.018
0.003
0.005
0.009

Kokanee were collected from a slough and side channel of the Finlay River in 2006.

P value
0.962
0.040
0.243
0.482
0.067
0.087
0.235
0.372
0.177
0.126
0.942
0.062
0.149
0.017

56

Table 2.4. Measurements of genetic variation of Williston watershed Kokanee by sample group. A
represents the total number of alleles across all loci per group, and AR is the rarefied allelic counts
across all loci per sample group. Expected heterozygosity and observed heterozygosity are
represented by He and Ho, respectively. Ho did not depart from He for any particular group across
all loci, but Ho was significantly lower than expected for locus Ots3 (P = 0.016). For each
parameter, standard errors are given in parentheses.

Group
Arctic
Tacheeda
Hill
Meadow
Carbon
Dunlevy
Manson
Germansen
Osilinka
Pelly
Finlay
Tsaydiz
Russel
Cutoff
Bower
Aley
Stevenson
Rsvr. Forebay
Rsvr. 2000
Native Rsvr.
Thutade

A
56
49
168
215
150
143
138
178
103
173
169
136
170
131
136
169
97
184
133
89
152

AR
3.12 (0.526)
2.82 (0.454)
6.83 (0.903)
6.97 (0.944)
6.62 (0.890)
6.27 (0.827)
6.50 (0.951)
6.48 (0.827)
6.92 (0.897)
6.54 (0.826)
6.51 (0.823)
6.18 (0.789)
6.44 (0.793)
6.32 (0.840)
6.27 (0.814)
6.38 (0.828)
6.10 (0.864)
6.56 (0.832)
6.22 (0.795)
3.84 (0.577)
5.93 (0.793)

He
0.40 (0.064)
0.38 (0.064)
0.72 (0.049)
0.70 (0.050)
0.71 (0.044)
0.69 (0.050)
0.70 (0.047)
0.71 (0.044)
0.72 (0.047)
0.71 (0.043)
0.71 (0.045)
0.70 (0.042)
0.71 (0.043)
0.70 (0.046)
0.70 (0.044)
0.70 (0.045)
0.69 (0.044)
0.71 (0.043)
0.69 (0.053)
0.52 (0.064)
0.64 (0.059)

Ho
0.40 (0.067)
0.38 (0.066)
0.71 (0.049)
0.70 (0.047)
0.69 (0.051)
0.68 (0.057)
0.68 (0.050)
0.71 (0.043)
0.69 (0.049)
0.70 (0.039)
0.70 (0.046)
0.70 (0.044)
0.70 (0.045)
0.68 (0.049)
0.69 (0.045)
0.69 (0.046)
0.66 (0.045)
0.68 (0.043)
0.61 (0.054)
0.53 (0.064)
0.64 (0.056)

57

Figure 2.2. Estimates of mean inbreeding coefficients (F) represented as bars for individual Kokanee in Arctic and
Tacheeda groups (top), Native Rsvr. and Thutade groups (middle), and tributary groups (bottom) of Kokanee
collected from the Williston watershed. The dashed line represents the overall mean F for each group: Tacheeda
(0.246), Arctic (0.234), Native Rsvr. (0.205), Thutade (0.182), and tributary (0.165).

58

Figure 2.3. Neighbor-joining tree constructed using Cavalli-Sforza and Edwards (1967) chord distance (DC)
inferred from variation at fourteen microsatellite loci in 21 groups of Williston watershed Kokanee. Numbers
represent percentage of 10,000 bootstrap replicates. Percentages below 50% are not reported.
Table 2.5. F-statistics (Weir and Cockerham 1984) of 21 groups of Kokanee sampled from the
Williston watershed in northern British Columbia, Canada.

FST
FIT

0.1017
0.1134

FIS

0.0130

59

Figure 2.4. STRUCTURE analysis for the comprehensive dataset of Kokanee assayed at fourteen microsatellite loci.
Each fish is represented by a vertical line which exhibits the proportional composition of each fish’s genome
across genetic clusters (K = 1–6).

60

Figure 2.5. STRUCTURE analysis for the tributary dataset of Kokanee collected in tributaries to the Williston
reservoir. Genetic clusters spread across all sample groups with increasing K, suggesting that these fish belong to
one genetic population (K = 1).

61

Figure 2.6. STRUCTURE analysis for the time-series dataset of Kokanee collected from Germansen, Pelly, and
Russel and grouped by sample year. Increasing admixture and a lack of clustering between sample years indicates
a single genetic population (K = 1).

62

Figure 2.7. STRUCTURE analysis for the native dataset (Arctic, Tacheeda, Native Rsvr., and Thutade) of Kokanee,
and those collected in 2000 (Rsvr. 2000), which exhibited evidence of a mixed-stock sample group.

63

Figure 2.8. STRUCTURE analysis for the Rsvr. 2000 group of Kokanee that were sampled from the body of the
reservoir by Pillipow and Langston (2002). The analysis clearly delineates two genetic clusters, which, in the
context of the comprehensive dataset, are in-line with Hill-type (blue) and native-type (orange) Kokanee.

64

Figure 2.9. Plots of discriminant analysis of principal components (DAPC) for fourteen microsatellites across the
comprehensive dataset of Williston watershed Kokanee. Individual genotypes in the scatterplot (a) are represented
by dots, and individual membership probabilities (b) are presented as vertical bars; genetic clusters are color-coded.
The predominant genetic population of each cluster are as follows: Cluster 1 = Hill; Cluster 2 = Meadow; Cluster 3
= Native Rsvr. and Thutade; Cluster 4 = Arctic and Tacheeda.

65

Figure 2.10. A subsection of the DAPC for the comprehensive dataset across fourteen microsatellite loci.
Identifying the individuals by sample group reveals that the Rsvr. 2000 group (▲) contains individuals that align
with Cluster 1 (Hill), Cluster 2 (Meadow), and Cluster 3 (native-type). The Native Rsvr. group (●), comprised of
Kokanee collected from 1988 to 1990 (Native Rsvr.) entirely aligns with Cluster 3. The Rsvr. Forebay group ( )
were assigned to Cluster 1 and Cluster 2, with one outlier aligning with Cluster 3. Thutade fish (+) were assigned
mostly to Cluster 3, but several aligned with Cluster 2.

66

Figure 2.11. A principal component analysis (PCA) for four groups of Williston watershed Kokanee (Rsvr.
Forebay, Rsvr. 2000, Native Rsvr., and Thutade). The Rsvr. 2000 is shown not to be a distinct genetic cluster, but
rather a mixed-stock sample group containing individuals that cluster with Columbia-type (Rsvr. Forebay) and
native (Native Rsvr. and Thutade) genetic populations.

67

Discussion
The results of this study provide new insights into the expansion of an introduced
lineage of Kokanee in a large, impounded watershed, including the impacts of such a
widespread stocking program on native populations of Kokanee. I found a distinct lack of
population structure among Kokanee spawning in tributary streams to the Williston reservoir,
and no instances of substantial introgression between Columbia-type and native populations.
The gradual elimination of native genotypes through hybridization may not be of lasting
concern, as native populations have persisted despite mass stocking efforts in other systems
(Young et al. 2004). However, the extirpation of the Kokanee population that was native to
the Williston reservoir remains a possibility. My findings have implications for future
fisheries management decisions when considering large-scale stocking programs, and for
prioritizing genetic diversity and variability within populations of Williston watershed
Kokanee.
Status of Native Williston Watershed Kokanee
As expected, Kokanee from Arctic and Tacheeda Lakes in the Parsnip region of the
Williston watershed are very closely related and exhibit overall low heterozygosity and
allelic richness. There is limited documentation as to the source of Kokanee in these lakes,
but it is generally thought that Tacheeda Lake individuals are relatively recent transplants
from Arctic Lake, which likewise would have originated from Fraser River Sockeye Salmon
(Nelson 1968; Langston and Zemlak 1998). My results reveal that the small population sizes
of these lake stocks have led to a higher inbreeding coefficient relative to other Kokanee
populations in the Williston watershed. A mean estimate inbreeding coefficient of 0.246, as
seen in the Tacheeda group, indicate that at the time of collection adult fish had almost 25%

68

less genetic diversity than their genetic founders (Hedrick and Kalinowski 2000; Kalinowski
et al. 2012). This estimation does, however, require confirmation through assessment of
genetic information from the presumed source population in the Fraser River, as well as
integrated information on when the Tacheeda fish “migrated” or were otherwise stocked
from Arctic Lake. Regardless, Arctic and Tacheeda genetic signatures are highly distinct
from the other native populations of Kokanee in Thutade Lake and the reservoir. While it
was thought to be possible for Columbia-type Kokanee to enter the headwaters of the Parsnip
Reach and hybridize with native fish (Langston 2012), there is no evidence of individuals
from Arctic and Tacheeda Lakes having been sampled elsewhere in the reservoir, nor is there
any indication of interspecific hybrid genotypes between these native groups and Columbiatype Kokanee.
Native Kokanee were reported to exist in the Williston reservoir prior to stocking
activity in the early 1990’s, and were observed spawning in the Finlay River between
November and January (Blackman et al. 1990; Fielden 1991; McLean and Blackman 1991;
Langston and Zemlak 1998). This population was an established, if not overabundant, fishery
and the population was considered to be expanding based on gillnet percentages in 1988
(Blackman et al. 1990). Thutade was thought to be the likely source of Kokanee found in the
Williston reservoir, and is comprised of a lineage of anadromous O. nerka from the Skeena
River system (Fielden 1992; Langston and Zemlak 1998). Cascadero Falls functions as a
physical barrier to upstream movement that prevents individuals from returning to Thutade
Lake to spawn. My results support the related nature of these two groups, but also suggests
that the reservoir Kokanee could be classified as a distinct genetic population. Evidence of
rapid reproductive isolation has been documented in O. nerka before; the study by Hendry et

69

al. (2000) discussed the possibility of partial reproductive isolation between ecotypes of O.
nerka after fewer than 13 generations, and without significant geographical barriers. If the
assumption is that native fish in the reservoir began to diverge from Thutade only after the
completion of the W.A.C. Bennett Dam in 1968, the reproductive divergence would have had
roughly 5 generations (4-year generation time) to develop before adults were collected in
1988. This estimate implies an exceptionally rapid divergence that does not incorporate the
possibility of native Kokanee existing apart from Thutade Lake in the lotic environment of
the watershed before impoundment, which is more likely given the short timeframe of 5
generations (Adkison 1995; Hendry et al. 2000). The overall low genetic diversity, low
allelic richness, and higher inbreeding coefficient exhibited by native Kokanee in this
analysis as compared to the Thutade group would suggest an ongoing divergence that
probably began before the completion of the dam (Selkoe and Toonen 2006; Iwamoto et al.
2012).
The high degree of relation between Thutade and native reservoir fish likely caused
the slight discrepancy between the ideal number of clusters (K) between STRUCTURE and
DAPC analyses; my DAPC of the comprehensive dataset did not differentiate Thutade and
native reservoir groups until a K of 8. The STRUCTURE analysis of the same dataset supported
a K of 4 (0.969), with Thutade and native reservoir fish combined to a single cluster, over a K
of 5 (0.959) in which they clustered separately. However, when isolated from the Columbiatype source and tributary populations, Thutade and native reservoir fish readily clustered into
their own respective and statistically-supported genetic populations ( STRUCTURE K = 4, MCL
= 0.997). Coupled with the geographic isolation imposed by Cascadero Falls on the Finlay
River, it is therefore plausible that the native Kokanee population found in the body of the

70

Williston reservoir was in the process of genetically diverging from its source population in
Thutade Lake.
According to my analyses, none of the 1,600+ fish collected after the sampling
performed by Pillipow and Langston (2002) were identified as native Williston Kokanee. The
20-year gap in collecting native genotypes could be indicative of local extirpation of this
genetic population, but limitations of the sampling methodologies employed for this study
may have introduced sampling biases—both spatially and temporally—that could have
inadvertently overlooked native Kokanee. Rather than focusing on spawning tributaries,
trawl or gill netting surveys could be used as an approximate random sampling technique that
could minimize biases; and STRUCTURE-based assignment tests performed on mixed trawl
samples could elucidate the question of extirpation. Further sampling, preferably on a
consistent basis, is required to investigate the possibility of persistence, as the native
Williston population should be considered one of two major management units that could
exist within the reservoir: native fish and introduced Columbia-origin Kokanee.
Interactions Between Columbia-Type and Native Fish
Examining the entirety of the sampled Kokanee from the Williston system, a few
trends become apparent. The native groups from Arctic and Tacheeda Lakes, pre-stocking
reservoir fish, and Thutade lake are still quite distinct from Columbia-type Kokanee that
spawn in tributaries to the reservoir. Tributary groups strongly cluster together, with Hilltype signatures largely persisting over those of the stock population from Meadow Creek.
Assuming a generation time of 4 years, the Columbia-origin Kokanee have existed in the
watershed for almost 8 generations, and in that time have rapidly expanded from their five
initial stocking locations (Langston 2012; McDermot-Fouts 2019; Robinson 2020). Previous

71

studies have reported that Columbia-type Kokanee spawn around mid-September,
substantially earlier than native Kokanee in the reservoir (McLean and Blackman 1991). This
combination of widespread straying and mismatched timing of spawning events may have
insulated the introduced Kokanee from developing considerable population structure between
spawning locations, as well as prevented introgression with native populations. Salmonids,
including non-anadromous Kokanee, have been shown to develop or maintain sympatric
genetic populations based on reproductive ecotype rather than solely geographic barriers to
gene flow (Foote et al. 1989; Hendry et al. 2000; Taylor et al. 2000; Jensen et al. 2017).
Reproductive ecotypes that differ by spawning location (“beach-spawning” vs “streamspawning”) or spawning month (September vs November, in this case) can undergo or
maintain genetic differentiation while existing as sympatric populations (Withler et al. 2000;
Lemay and Russello 2015). In some cases, temporal isolation alone may be a stronger driver
for genetic divergence than site-specific isolation (Young et al. 2004; Whitlock et al. 2018).
Nevertheless, the loss of distinct native genotypes may be a legitimate concern regardless of
ecotypic barriers to gene flow (Veale and Russello 2016).
My results from STRUCTURE show that some individuals with proportions of native
Kokanee signatures were collected in 1994 from Carbon Creek. This sample group is likely
one of the first cohorts of Columbia-type Kokanee to reach sexual maturity after juveniles
and fertilized eggs from the source populations were initially stocked in 1990 (Langston and
Murphy 2008). With a generation time of 4 years, hybrids between Columbia-type and native
Kokanee would not have been sampled as mature spawners until approximately 1998. As
such, the timeframe does not support previous introgression with native stock before
maturation in 1994.

72

Native individuals collected from the embayment of Carbon Creek in the Peace
region in August of 1994 were considered to be in spawning condition (Langston and Zemlak
1998). This could be a departure from known spawning locations and timing of native
reservoir Kokanee, which were previously observed spawning in the Finlay region and
months after Columbia-type adults had concluded spawning activity (Fielden 1991; Fielden
1992; Langston and Zemlak 1998). However, as these fish were collected via a floating net
and were not actively spawning, it cannot be definitively said that the three native individuals
were preparing to enter Carbon Creek to spawn (Langston and Zemlak 1998). Additionally,
as there is no genetic evidence of hybridization across almost 30 years of sampling, this
instance in and of itself is not cause for concern on behalf of the genetic diversity of
Williston watershed Kokanee. There is some suggestion that post-mating selection against
"hybrids" of reproductive forms of O. nerka (Sockeye Salmon and Kokanee) may keep gene
flow limited between sympatric populations where they spawn at the same time and location
(Taylor et al. 1996; Wood and Foote 1996). The potential for a similar selection mechanism
against native-introduced "hybrid" Kokanee should be further investigated in the Williston
system. Based on the samples collected for this analysis, however, introgression does not
appear to have occurred at any examined location or spawning period.
It should be recognized that in my DAPC and some individual STRUCTURE analyses,
the Thutade group appears to cluster closer to tributary groups (namely Meadow genotypes)
than does the Native Rsvr. group. This result is unexpected given that the tributary fish are
clearly endemic to the Columbia River and are not assumed to share an immediate (i.e., postglacial) common lineage with Skeena River anadromous O. nerka, which are thought to be
the genetic source of the Thutade population (Fielden 1992; Wood et al. 1994; Langston and

73

Zemlak 1998). However, large-scale surveys of O. nerka populations across their North
Pacific range have shown an overlap of Skeena River populations with the three other major
genetic populations of O. nerka identified in British Columbia (Southern Rivers, Coastal
Islands, and Coastal Mainland groups), suggesting that the populations in these regions may
have been founded by ancestors from multiple glacial refugia (Wood et al. 1994; Taylor et al.
1996). Wood et al. (1994) also described shared rare alleles between populations from both
the Skeena and Columbia drainages, indicating past gene flow between populations that are
now highly geographically separated.
It is also possible that the apparent Thutade-Meadow genetic relatedness in my results
could be due to my choice of molecular marker. Microsatellites are well-equipped and highly
informative when used for fine-scale examinations of population structure between closelyrelated populations (Banks et al. 1999). The ability of a molecular marker to identify the
degree of differentiation between highly-related Kokanee spawning in different tributary
streams within the same watershed was a main factor in choosing microsatellites over single
nucleotide polymorphisms (SNPs) (O’Reilly and Wright 1995; Beacham and Wood 1999). In
contrast to microsatellites, which are selectively neutral within the genome, SNPs allow for
the examination of selective genes, which may offer high resolution stock identification
where applicable (Habicht et al. 2010; Dann et al. 2012). Additionally, as conservative
markers, SNPs retain evidence of shifts in allele frequency distribution that occurred in
relatively ancient times more so than microsatellites (Khrustaleva et al. 2017). This may
make SNPs more effective in phylogenetic reconstructions than microsatellites, which have
not been historically successful in analyses of populations that are not closely related
(Goldstein and Pollock 1997). More contemporary studies have suggested that the ideal

74

procedure for genetic stock identification is a combination of the most informative
microsatellites and SNPs (Narum et al. 2008; Beacham et al. 2010). Such an approach may
better inform the degree of relatedness between Thutade Lake (Skeena River) and Meadow
Creek (Columbia River) Kokanee in the context of a broader examination.
Population Structure of Tributary Spawners
A major focus of this study was to evaluate the degree of differentiation, genetic
patterns, or signs of population structure among the Columbia-type Kokanee that spawn in
tributaries to the Williston reservoir. Hill-type signatures have largely persisted over
Meadow-type signatures, and evidence of admixture between Hill and Meadow genotypes
reflects the stocking history of the Hill creek spawning channel, which has been periodically
stocked with Kokanee from Meadow creek (Langston and Murphy 2008). My results suggest
that there is no discernable genetic structure for these spawning groups at spatial or temporal
scales. This is not unexpected given the relatively short timeframe and expansive straying
that occurred after the juvenile Kokanee were introduced in 1990. In instances where O.
nerka populations are genetically distinct, spatial variation has been shown to exceed
temporal variation (Beacham and Margolis 1998; Beacham and Wood 1999; Beacham et al.
2005). Additionally, salmonids that experience regular low levels of gene flow due to
straying may experience stabilizing effects and reduced temporal genetic variation (Taylor et
al. 2000; Walter et al. 2009; Taylor and Gow 2010; 2010). Heavily managed sport fisheries
that stock large numbers of Kokanee may not experience a great degree of reproductive
isolation between spawning groups (Whitlock et al. 2018). Because Columbia-type Kokanee
in this system are known to have strayed significantly from their five original stocking
locations (Langston 2012; McDermot-Fouts 2019; Robinson 2020), substantive genetic

75

divergence between Kokanee spawning locations has likely been prevented thus far. The
Columbia-origin population in the Williston Reservoir should therefore be considered a
single management unit for purposes of fisheries monitorization and governance.
Conclusions
The aim of this study was to examine the expansion of an introduced lineage of
Kokanee in a large, impounded watershed, including the impacts of a widespread stocking
program on native populations of Kokanee. While my results indicate that the introduced
Columbia-type Kokanee have neither diverged by spawning location nor have they
hybridized with native populations of Kokanee, there are troubling implications for the
lasting survival of native genetic signatures in the reservoir. Although gillnetting from the
body of the reservoir has occurred several times in the past twenty years (2000, 2016, 2019),
no native reservoir Kokanee signatures have been detected since the sampling performed by
Pillipow and Langston (2002) in 2000. Further gillnetting following the timing (August and
September) and procedures put forth by Pillipow and Langston (2002) is recommended in
order to intercept any native Kokanee that may be persisting in the body of the reservoir. As
it stands currently, it appears as though the native population of Kokanee that diverged from
Thutade Lake has been extirpated due to the widespread introduction of Columbia-type
Kokanee. This erosion of locally-adapted genetic diversity in the watershed was not due to
introgression with introduced Kokanee, as was of concern to management (Langston 2012;
McDermot-Fouts 2019; Robinson 2020). I recommend further investigation into other factors
of extirpation, such pelagic competition between genetic populations of Kokanee, that may
have occurred after the widespread stocking program was initiated in this highly-oligotrophic
reservoir (Olsen et al. 2017). My findings have implications for the strategic direction of

76

fisheries management decisions, and the particular need for identifying management units at
the genetic population level within the Williston watershed.

77

EPILOGUE
My thesis research examined the trends of Kokanee demographics and genetic
population structure within the context of intensive stocking efforts in a complex impounded
watershed. My work demonstrated strong adaptive growth responses exhibited by Kokanee
in environments of poor productivity (Hyatt and Stockner 1985; Rieman and Myers 1992;
Askey 2016), and highlighted the remarkable habitat plasticity that has allowed Kokanee to
repeatedly colonize regions in a pattern of parallel adaptive radiation since the last glaciation
period (Wood et al. 1994; Taylor et al. 1996; Beacham and Withler 2017). Understanding the
degree to which native and introduced Kokanee populations have responded to changes in
fish density and the addition of exogenous stocks is crucial for effective fisheries
management at a time when many populations of Kokanee in British Columbia are
experiencing human-related population declines or outright collapse (Askey and Johnston
2013; Taylor et al. 2014; Peck et al. 2019; Bassett et al. 2020; Grant et al. 2021; Warnock et
al. 2021).
The mechanisms affecting compensatory growth responses in Kokanee populations
are dynamic and complex, and Kokanee are known to show strong adaptive responses to
changes in fish density and lacustrine productivity (Kendall et al. 2014; Askey 2016). In
Chapter 1, I quantified the combined trends of significant decreases in fork length and
condition factor of mature Kokanee over time, as well as no significant change in spawner
age over time, all in the presence of a likely decline in spawner abundance over the last 15
years in the Williston Reservoir (Langston 2012; McDermot-Fouts 2019; Robinson 2020).
My results could indicate a genetic component to differences in morphometrics, as native
Kokanee that were collected prior to introductions from the Columbia River were

78

significantly larger than mature individuals collected in any other year after stocking efforts
were initiated. However, the lack of compensatory responses to recent changes in fish density
point to poor growth conditions in the reservoir, which could have caused or contributed to
the differences in adult size at maturity between introduced and native fish. I also found
spawning location to be a significant contributing factor to differences in fork length and
condition factor, even within the same spawning cohort across years. Regular monitoring of
spatial and temporal differences in morphometrics is an important aspect of adaptive
management practices; changes in phenotype may be indicative of shifting population
genetics or local adaptative divergence within a reservoir environment (Kendall et al. 2014;
Whitlock et al. 2018).
Hybridization or introgression between native and introduced stocks is not inherently
harmful (Veale and Russello 2016), and in the case of Kokanee in the Williston Reservoir,
my results indicate that no genetic homogenization has occurred. However, while native
populations have persisted in other systems in spite of mass stocking efforts (Young et al.
2004), the widespread expansion of Columbia-origin Kokanee throughout the Williston
watershed may constitute a substantial threat to the lasting survival of native genetic
signatures in the region. In Chapter 2, I examined the genetic population structure of
Kokanee that spawn contemporarily in and around the reservoir, as well as temporal
comparisons with native fish collected before Kokanee introductions and source populations
of the introduced Columbia stock. The introduced Columbia-type Kokanee represented the
entirety of sample groups collected from 2006 to 2019, and there was no evidence of genetic
divergence by spawning location. Native populations in Arctic and Tacheeda Lakes remain
entirely separate from reservoir populations and the threat of introgression with introduced

79

stock appears to be quite low. Native Williston Reservoir Kokanee, which I showed to have
diverged from the population that exists in Thutade Lake, have not been definitively
intercepted in the reservoir since sampling performed by Pillipow and Langston (2002) in
2000; it is likely that this population has been extirpated or otherwise out-competed by
Columbia-type Kokanee. My findings represent an unfortunate consequence of the expedited
measures outlined by past management guidelines; particularly, the emphasis on immediately
creating a recreational sport fishery without “spend[ing] a disproportionate amount of effort
evaluating and conducting inventories” (Blackman et al. 1990). This management philosophy
explicitly prioritized introduced stock over the known native populations, and intensive
investigations into native Williston Kokanee conducted at the time did not result in directives
to conserve and enhance locally-adapted populations (Blackman et al. 1990; Fielden 1991;
Fielden 1992).
My findings are particularly relevant for managing Kokanee fisheries in British
Columbia. Anthropogenic environmental alterations, namely habitat loss and degradation,
continue to be the greatest threat to species persistence in Canada (Dextrase and Mandrak
2006; Taylor et al. 2014; Grant et al. 2021). Hydroelectric dam construction and subsequent
inundation and reservoir formation are exceptionally impactful on freshwater fish species
(Stockner et al. 2005; Taylor et al. 2014). The Williston Reservoir is a particularly immense
man-made lacustrine system, and the impacts of its formation extend to a multitude of
interconnected physiochemical and biological processes (Stockner et al. 2005). The
watershed’s inherent complexity has made analytical and monitoring protocols a distinct
challenge; however, it is this complexity that necessitates the appropriate use of tools and
resources to ensure the viability of both native and introduced populations.

80

An emphasis should be placed on understanding reservoir-wide relationships among
independent populations of Kokanee, to not only implement remedial measures but to
ultimately develop adaptive and flexible management philosophies; retrospective analyses
and regime-shift predictions have limited functionality in managing population trends as they
occur (Andrusak et al. 2000; Askey and Johnston 2013). For many biological processes,
conservation efforts, and mitigation strategies, the most appropriate management unit is
based on the scale of proposed impacts (McElhany et al. 2000). For salmonids, it is often
most useful to focus on fish at the level of “independent population” or “stock” rather than a
species in its entirety for a given region (McElhany et al. 2000). An independent population
is considered to be a breeding unit with low levels of gene flow with migrant individuals, and
may be considered a viable salmonid population if it is not at immediate risk of extinction
from changes in population demographics, altered environmental factors, or genetic structure
(McElhany et al. 2000). In the case of Williston watershed Kokanee, my research has
identified at least four independent populations based on microsatellite markers: native
Thutade Lake Kokanee; native Williston Reservoir Kokanee (most likely arising from
Thutade Lake individuals); native Arctic and Tacheeda Lakes Kokanee; and introduced
Columbia-origin Kokanee. Within the reservoir itself, native Williston Reservoir Kokanee
and introduced Columbia-origin Kokanee are two management units that I identified through
my analyses. It is unknown whether the native Williston Kokanee would have been
designated a viable salmonid population under the definitions outlined by McElhany et al.
(2000), but this population was documented to be increasing in number in the years
immediately prior to the stocking program (Blackman et al. 1990). These native Kokanee
were known to spawn in side channels and sloughs to the lower Finlay River and also later in
the season than Columbia fish, and displayed different color patterns at maturity (e.g. dark
81

reddish brown) (Langston and Zemlak 1998). Assumptions about stock composition without
appropriate phenotypic, demographic, and genetic evidence can lead to management
decisions that entirely overlook genetically-distinct Kokanee populations (Young et al.
2004). Conversely, a thorough understanding of genetic markers and phenotypic variation
among reproductive ecotypes could appropriately direct enhancement efforts to benefit one
or multiple independent populations (Frazer and Russello 2013; Whitlock et al. 2018).
Other large lakes and reservoir systems in British Columbia are contending with
similar challenges in preserving genetic variability and propagating healthy Kokanee
fisheries. The Arrow Lakes Reservoir system in southwestern B.C. has been extensively
modified by the construction of three dams, which negatively affected local Kokanee
populations and prompted in-depth monitoring beginning in the 1990’s (Taylor and Gow
2010; Bassett et al. 2020). Microsatellite analyses revealed weak within-basin population
structure but significant genetic divergence between populations of different basins, and at
least one instance of introgression between native and donor stocks (Taylor and Gow 2010).
The ongoing nutrient restoration initiative in the reservoir has not affected the exceptionally
low juvenile survival rates, and the adult Kokanee have shifted to early (age 2) and small (21
cm; 129 g) maturation since 1999 (Bassett et al. 2020). A similar nutrient restoration program
in the Kootenay Lake system has successfully increased the size of harvested Kokanee (>36
cm), but intensive top-down predation by piscivorous Rainbow Trout (Oncorhynchus mykiss)
and Bull Trout (Salvelinus confluentus) has kept the Kokanee population in a state of
collapse (Peck et al. 2019; Warnock et al. 2021). It has been theorized that favorable nutrient
and rearing conditions elevated the Kokanee population past the carrying capacity of the
Kootenay Lake system (Warnock et al. 2021). Okanagan Lake and Wood Lake in the south-

82

central region of B.C. have also experienced dramatic declines in Kokanee populations in
recent years (Taylor et al. 2000; Askey and Johnston 2013; Askey 2016; Ward et al. 2019).
Both systems are composed of multiple reproductively-isolated Kokanee ecotypes, and
current management protocols are compelled to recognize these ecotypes as discrete
management units to ensure the recovery and persistence of Kokanee as a whole (Taylor et
al. 2000; Ward et al. 2019). The preservation of spawning habitat specific to each ecotype in
conjunction with the implementation of adaptive harvest regulations has been valuable in the
recovery plans for these populations (Taylor et al. 2000; Askey 2016; Ward et al. 2019).
The results of my work in Chapter 1 have raised questions regarding the factors
influencing compensatory growth responses of Kokanee in the Williston Reservoir. Although
the age at maturity has shown no significant change over time, my results demonstrate
significant declines in the sizes of mature fish since introductions of Columbia-origin stock;
these declines in-turn have implications for female fecundity, reproductive potential, and
overall population growth (McGurk 2000; Wilson and Shrimpton 2021). Mechanisms behind
the rapid changes in spawner size and condition factor may be associated with densitydependent responses, environmental productivity, or within-species phenotypic plasticity
(Hyatt and Stockner 1985; Rieman and Myers 1992; Grover 2006; Kendall et al. 2014). The
decline in number of spawning Kokanee indicated by recent aerial flight counts (Langston
2012; McDermot-Fouts 2019; Robinson 2020) suggest that the number of Kokanee may have
declined in the reservoir. Such a decline in spawner escapement could be linked to a change
or reduction in productivity and, consequently, carrying capacity. This could result in greater
intraspecific competition between Columbia-origin Kokanee and native Williston Kokanee,
if the latter still persist in the watershed. Further pelagic fish surveys similar to those

83

previously conducted in the reservoir (Sebastian et al. 2003; Sebastian et al. 2009) could be
helpful in determining the causes of the morphometric changes that were identified in my
study.
Genetic tools allow for highly informative monitoring strategies that can assess the
degree to which native and introduced genetic signatures persist in a managed system (Young
et al. 2004). Based on my results in Chapter 2, there are troubling implications for the lasting
survival of native Kokanee in the Williston Reservoir. The lack of native Williston Kokanee
genotypes among any of the fish sampled in 2006, 2016, 2017, 2018, or 2019 suggest that
native fish may have been extirpated from the reservoir. Sampling of adults at spawning
locations in mid-September, as was the procedure for fish collection in my study, may have
inadvertently excluded native individuals; this population was not previously observed to
spawn at the same time or locations as Columbia-type Kokanee (Fielden 1991; Fielden 1992;
Langston and Zemlak 1998). My genetic analysis included tissue samples of fish collected in
2016 and 2019 from the G.M. Shrum intake towers close to the W.A.C. Bennett Dam, in the
Forebay region of the reservoir—yet all of these fish (n = 135) were of Columbia-origin and
no native genetic signatures were detected. Further sampling in the body of the reservoir
where pelagic mixing of genetic populations occurs, and timed to duplicate surveys
conducted in August of 2000 (Pillipow and Langston 2002), is imperative to assess whether
native Kokanee are still found within the reservoir.
Considerable phenotypic and genetic differences exist between the native Williston
Kokanee and the introduced Columbia-origin Kokanee. The genetic source of the Thutade
and native Williston populations is thought to be a lineage of Skeena River anadromous O.
nerka (Fielden 1992; Wood et al. 1994; Langston and Zemlak 1998), which is supported by

84

an observed darker coloration at maturity that is commonly associated with Kokanee derived
from post-glaciation residual Sockeye Salmon (Ricker 1938; Ricker 1959; Craig and Foote
2001). This is in contrast with the Columbia-type Kokanee in the reservoir, which display
bright red bodies and green heads at maturity—a phenotype characteristic typically found in
predominantly non-anadromous lineages (Ricker 1938; Ricker 1959). Furthermore, Beacham
and Withler (2017) provided evidence for a monophyletic origin for Kokanee from the
Columbia River; adaptive radiation from a glaciation refuge is genetically-supported for
Kokanee from the upper Columbia region, rather than recurrent independent evolution of
Kokanee from anadromous O. nerka.
My work also indicates distinctive phenotypic differences between introduced and
native Williston Kokanee which may have had an effect on the survival of locally-adapted
fish experiencing mixed-stock pelagic competition. I found significant differences in fork
length and condition factor, and it is possible that other morphological features may also have
differed. Mean gill raker numbers—a heritable trait that has been shown to be stable over
time—are significantly higher in Kokanee than Sockeye Salmon, and may also suggest that
Kokanee are more efficient at eating small zooplankton compared to their anadromous form
(Wood and Foote 1996; Foote et al. 1999). As the native Williston Kokanee may be
phenotypically and genetically more similar to anadromous Skeena River O. nerka, they may
also have fewer gill rakers than the stocked Columbia-origin fish of monophyletic Kokanee
lineage. Competition induced by introductions may result in phenotypic changes, including
reduced numbers of gill rakers (Crowder 1984). However, any differences in morphology or
feeding efficiency between native and introduced Kokanee are not known. Morphometric
analysis of native populations from Thutade Lake, Arctic Lake, and Tacheeda Lake could be

85

used to investigate differences between native and donor populations (Hill Creek and
Meadow Creek) from the Columbia River. Additionally, if morphometric differences
between the two lineages are found, gill raker count of pelagic species that have been
historically successful in the Williston Reservoir, such as Lake Whitefish (Coregonus
clupeaformis) (Blackman et al. 1990; Sebastian et al. 2009), may be informative as to why
one lineage of Kokanee was more successful than the other.
My thesis research demonstrates the remarkable compensatory growth responses,
rapid rates of straying and colonization, and maintained genetic differentiation among
sympatric populations of Kokanee in the Williston watershed. I identified marked declines in
spawner size and condition factor over the past 30 years, as well as the existence of four
independent genetic populations in and around the reservoir. The survival of locally-adapted
genetic and ecotype diversity is dependent on management practices that recognize and
monitor the structure and demographic independence of Kokanee populations in the
Williston Reservoir. Strategies that incorporate knowledge of the four discrete populations of
Williston watershed Kokanee, such as spawning habitat enhancement for diverse ecotypes,
control of piscivorous populations (e.g., Lake Trout), or reach-specific harvest limits should
be prioritized moving forward.

86

REFERENCES
Adkison, M. D. 1995. Population differentiation in Pacific salmon: local adaptation, genetic
drift, or the environment? Canadian Journal of Fisheries and Aquatic Sciences
52:2762–2777.
Andrusak, G. 2007. Hill Creek Spawning Channel Kokanee Adult Escapement - 2006.
Columbia Basin Fish and Wildlife Compensation Program, 22 pp.
Andrusak, H., D. C. Sebastian, I. McGregor, S. Matthews, D. Smith, K. Ashley, S. Pollard,
G. H. Scholten, J. G. Stockner, P. Ward, R. Kirk, D. Lasenby, J. Webster, J. Whall, G.
Wilson, and H. Yassien. 2000. Okanagan Lake Action Plan Year 4 (1999) Report.
Fisheries Management Branch, Ministry of Agriculture, Food and Fisheries, BC,
Report No. RD 83, 314 pp.
Armstrong, J. D., and D. C. Stewart. 1997. The Effects of Initial Length and Body Curvature
on Shrinkage of Juvenile Atlantic Salmon During Freezing. Journal of Fish Biology
50:903–905.
Askey, P. J. 2016. Managing Dynamic Fisheries with Static Regulations: an Assessment of
Size-Graded Bag Limits for Recreational Kokanee Fisheries. North American Journal
of Fisheries Management 36:241–253.
Askey, P. J., and N. T. Johnston. 2013. Self-Regulation of the Okanagan Lake Kokanee
Recreational Fishery: Dynamic Angler Effort Response to Varying Fish Abundance
and Productivity. North American Journal of Fisheries Management 33:926–939.
Aunins, A. W., J. T. Petty, T. L. King, M. Schilz, and P. M. Mazik. 2014. River mainstem
thermal regimes influence population structuring within an appalachian brook trout
population. Conservation Genetics 16(1):15-29.
B.C. MoFLNRORD. 2021. Kokanee quota change on Williston Lake and Peace River. L.
B.C. Ministry of Forests, Natural Resource Operations and Rural Development
Angling, Hunting and Trapping Engagement, editor.
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). The American
Genetic Association 90:281-288.
Bassett, M., S. Arndt, and E. Schindler. 2020. Arrow Lakes Reservoir Nutrient Restoration
Program. Columbia Basin Fish and Wildlife Compensation Program, 20 pp.
Beacham, T. D., J. R. Candy, B. McIntosh, C. MacConnachie, A. Tabata, K. Kaukinen, L.
Deng, K. M. Miller, R. E. Withler, and N. Varnavskaya. 2005. Estimation of Stock
Composition and Individual Identification of Sockeye Salmon on a Pacific Rim Basis
Using Microsatellite and Major Histocompatibility Complex Variation. Transactions
of the American Fisheries Society 134:1124-1146.
87

Beacham, T. D., S. Cox-Rogers, C. MacConnachie, B. McIntosh, and C. G. Wallace. 2014.
Population Structure and Run Timing of Sockeye Salmon in the Skeena River, British
Columbia. North American Journal of Fisheries Management 34:335-348.
Beacham, T. D., M. Hansen, B. McIntosh, and C. Wallace. 2010. A comparison of stock and
individual identification for Sockeye salmon (Oncorhynchus nerka) in British
Columbia provided by microsatellites and single nucleotide polymorphisms.
Canadian Journal of Fisheries and Aquatic Sciences 67:1274-1290.
Beacham, T. D., and L. Margolis. 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.
Beacham, T. D., and R. E. Withler. 2017. Population structure of sea-type and lake-type
Sockeye salmon and Kokanee in the Fraser River and Columbia River drainages.
PLoS One 12:e0183713.
Beacham, T. D., and C. C. Wood. 1999. Application of microsatellite DNA variation to
estimation of stock composition and escapement of Nass River Sockeye salmon
(Oncorhynchus nerka). Canadian Journal of Fisheries and Aquatic Sciences 56:297310.
Benjamini, Y., and Y. Hochberg. 1995. Controlling the False Discovery Rate: A Practical
and Powerful Approach to Multiple Testing. Journal of the Royal Statistical Society.
Series B (Methodological) 57:289–300.
Blackman, B. G., D. Jesson, D. Ableson, and T. Down. 1990. Williston Lake Fisheries
Compensation Program Management Plan. Peace/Williston Fish and Wildlife
Compensation Program, Report No. 60, 38 pp.
Blackwell, B. G., M. L. Brown, and D. W. Willis. 2010. Relative Weight (Wr) Status and
Current Use in Fisheries Assessment and Management. Reviews in Fisheries Science
8:1–44.
Burger, C., and W. Spearman. 1997. Genetic Differentiation of Sockeye Salmon
Subpopulations from a Geologically Young Alaskan Lake System. Transactions of
the American Fisheries Society 126:926–938.
Cavalli-Sforza, L. L., and A. W. F. Edwards. 1967. Phylogenetic Analysis: Models and
Estimation Procedures. American Journal of Human Genetics 19:233-257.
Chao, B. F., Y. H. Wu, and Y. S. Li. 2008. Impact of Artificial Reservoir Water
Impoundment on Global Sea Level. Science 320:212-214.
Coxson, D., D. Huber, and J. M. Shrimpton. 2017. Ecosystem impact of nutrient enrichment
by Kokanee in the Williston Reservoir Watershed. PWFWCP, PEA-F17-F-1471, 40
pp.

88

Craig, J. K., and C. J. Foote. 2001. Countergradient variation and secondary sexual color:
phenotypic convergence promotes genetic divergence in carotenoid use between
sympatric anadromous and nonanadromous morphs of Sockeye salmon
(Oncorhynchus nerka). Evolution 55:380-391.
Crowder, L. B. 1984. Character Displacement and Habitat Shift in a Native Cisco in
Southeastern Lake Michigan: Evidence for Competition? Copeia 1984:878–883.
Culling, B., T. Euchner, T. Ward, and S. J. Cooke. 2020. Peach Reach Lake Trout
Movements Final Report - Year 4 - PUBLIC. Peace/Williston Fish and Wildlife
Compensation Program, Project No. PEA-F20-F-2948, 104 pp.
Dann, T. H., J. R. Jasper, H. A. Hoyt, H. Hildebrand, and C. Habicht. 2012. Western Alaska
Salmon Stock Identification Program Technical Document 6: Selection of the 96 SNP
Marker Set for Sockeye Salmon. Alaska Department of Fish and Game, Regional
Information Report No. 5J12-11, 113 pp, Anchorage, Alaska.
Dextrase, A. J., and N. E. Mandrak. 2006. Impacts of Alien Invasive Species on Freshwater
Fauna at Risk in Canada. Biological Invasions 8:13–24.
DoĞAn, İ., and N. DoĞAn. 2016. Genetic Distance Measures: Review. Turkiye Klinikleri
Journal of Biostatistics 8:87-93.
Earl, D. A., and B. M. vonHoldt. 2011. STRUCTURE HARVESTER: a website and program
for visualizing STRUCTURE output and implementing the Evanno method.
Conservation Genetics Resources 4(2):359-361.
Evanno, G., S. Regnaut, and J. Goudet. 2005. Detecting the number of clusters of individuals
using the software STRUCTURE: a simulation study. Molecular Ecology 14:26112620.
Excoffier, L. 2003D. J. Balding, M. Bishop, and C. Cannings, editors. Handbook of
Statistical Genetics. John Wiley & Sons, Ltd.
Excoffier, L., and H. E. Lischer. 2010. Arlequin Suite Ver 3.5: A New Series of Programs to
Perform Population Genetics Analyses Under Linux And Windows. Molecular
Ecology Resources 10:564-567.
Felsenstein, J. 1985. Phylogenies from Gene Frequencies: A Statistical Problem. Systematic
Zoology 34:300-311.
Fielden, R. J. 1991. Finlay River Kokanee (Oncorhynchus nerka) Spawning Survey 1990.
Peace/Williston Fish and Wildlife Compensation Program, Report No. 59, 40 pp.
Fielden, R. J. 1992. Finlay River Kokanee (Oncorhynchus nerka) Spawning Survey, 1991.
Peace/Williston Fish and Wildlife Compensation Program, Report No. 63, 27 pp.

89

Foote, C. J., K. Moore, K. Stenberg, K. J. Craig, J. K. Wenburg, and C. C. Wood. 1999.
Genetic Differentiation in Gill Raker Number and Length in Sympatric Anadromous
and Nonanadromous Morphs of Sockeye Salmon, Oncorhynchus nerka.
Environmental Biology of Fishes 54:263–274.
Foote, C. J., C. C. Wood, and R. E. Withler. 1989. Biochemical Genetic Comparison of
Sockeye Salmon and Kokanee, the Anadromous and Nonanadromous forms of
Oncorhynchus nerka. Canadian Journal of Fisheries and Aquatic Sciences 46:149158.
Frazer, K. K., and M. A. Russello. 2013. Lack of Parallel Genetic Patterns Underlying the
Repeated Ecological Divergence of Beach and Stream-Spawning Kokanee Salmon.
Journal of Evolutionary Biology 26:2606-2621.
Froese, R. 2006. Cube Law, Condition Factor and Weight-Length Relationships: History,
Meta-Analysis and Recommendations. Journal of Applied Ichthyology 22:241–253.
FWCP. 2020. Fish and Wildlife Compensation Program - Peace Region: Rivers, Lakes, &
Reservoirs Action Plan. BC Hydro, 39 pp.
Goldstein, D. B., and D. D. Pollock. 1997. Launching Microsatellites: A Review of Mutation
Processes and Methods of Phylogenetic Inference. Journal of Heredity 88:335-342.
Goudet, J. 1995. FSTAT (Version 1.2): A Computer Program to Calculate F-Statistics. The
Journal of Heredity 86:485-486.
Goudet, J. 2003. FSTAT (version 2.9.4), a Program (for Windows 95 and Above) to Estimate
and Test Population Genetic Parameters.
Goudet, J., T. Jombart, Z. N. Kamvar, E. Archer, and O. Hardy. 2020. hierfstat: Estimation
and Tests of Hierarchical F-Statistics.
Grant, S. C. H., B. L. MacDonald, D. Lewis, N. Wilson, J. L. Boldt, J. King, T. Ross, R. I.
Perry, D. A. Patterson, K. A. Robinson, D. T. Selbie, C. G. Hannah, and A. VelezEspino. 2021. Canada’s 2020 Report on Recent Pacific Salmon Abundance and
Ecosystem Trends. North Pacific Anadromous Fish Commission, Report No. 1984, 9
pp.
Grover, M. C. 2005. Changes in size and age at maturity in a population of Kokanee
Oncorhynchus nerka during a period of declining growth conditions. Journal of Fish
Biology 66:122-134.
Grover, M. C. 2006. Evaluation of a Negative Relationship between Abundance during
Spawning and Size at Maturity in Kokanee. Transactions of the American Fisheries
Society 135:970–978.
Guo, S., and E. A. Thompson. 1992. Performing the Exact Test of Hardy-Weinberg
Proportion for Multiple Alleles. Biometrics 48:361-372.
90

Habicht, C., L. W. Seeb, K. W. Myers, E. V. Farley, and J. E. Seeb. 2010. Summer–Fall
Distribution of Stocks of Immature Sockeye Salmon in the Bering Sea as Revealed by
Single-Nucleotide Polymorphisms. Transactions of the American Fisheries Society
139:1171-1191.
Hagen, J., FWCP, personal communication.
Hansen, M. M., and K. L. D. Mensberg. 2009. Admixture analysis of stocked Brown trout
populations using mapped microsatellite DNA markers: indigenous trout persist in
introgressed populations. Biology Letters 5:656-659.
Harris, S., A. R. Langston, J. G. Stockner, and L. Vidmanic. 2005. A limnological
assessment of four Williston Reservoir embayments, 2004. Peace/Williston Fish and
Wildlife Compensation Program, Report No. 305, 48 pp.
Hedrick, P. W., and S. Kalinowski. 2000. Inbreeding Depression in Conservation Biology.
Annual Review of Ecology, Evolution, and Systematics 31:136–162.
Heinke, F. 1908. Bericht über die Untersuchungen der Biologischen Anstalt auf Helgoland
zur Naturgeschichte der Nutzfische. Die Beteiligung Deutschlands an der
Internationalen Meeresforschung 5:67–155.
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(5491):516-519.
Holm, L. E., V. Loeschcke, and C. Bendixen. 2001. Elucidation of the molecular basis of a
null allele in a Rainbow trout microsatellite. Marine Biotechnology 3:555-560.
Huang, K., K. Ritland, D. W. Dunn, X. Qi, S. Guo, and B. Li. 2016. Estimating Relatedness
in the Presence of Null Alleles. Genetics 202:247-260.
Hyatt, K. D., and J. G. Stockner. 1985. Responses of Sockeye Salmon (Oncorhynchus nerka)
to Fertilization of British Columbia Coastal Lakes. Canadian Journal of Fisheries and
Aquatic Sciences 42:320–331.
Iwamoto, E. M., J. M. Myers, and R. G. Gustafson. 2012. Resurrecting an extinct salmon
evolutionarily significant unit: archived scales, historical DNA and implications for
restoration. Molecular Ecology 21:1567-1582.
Janes, J. K., J. M. Miller, J. R. Dupuis, R. M. Malenfant, J. C. Gorrell, C. I. Cullingham, and
R. L. Andrew. 2017. The K = 2 Conundrum. Molecular Ecology 26:3594-3602.
Jenkins, T. M., S. Diehl, K. W. Kratz, and S. D. Cooper. 1999. Effects of Population Density
on Individual Growth of Brown Trout in Streams. Ecology 80:941–956.

91

Jensen, A. J., L. P. Hansen, B. O. Johnsen, and S. Karlsson. 2017. Rapid Evolution of
Genetic and Phenotypic Divergence in Atlantic Salmon Following the Colonisation of
Two New Branches of a Watercourse. Genetics Selection Evolution 49:1–22.
Johnston, F. D., and J. R. Post. 2009. Density-dependent life-history compensation of an
iteroparous salmonid. Ecological Applications 19:449-467.
Jombart, T. 2008. adegenet: a R package for the multivariate analysis of genetic markers.
Bioinformatics 24:1403-1405.
Jombart, T., and I. Ahmed. 2011. adegenet 1.3-1: new tools for the analysis of genome-wide
SNP data. Bioinformatics 27:3070-3071.
Jombart, T., and C. Collins. 2015. A Tutorial for Discriminant Analysis of Principal
Components (DAPC) using adegenet 2.0.0.
Jombart, T., S. Devillard, and F. Balloux. 2010. Discriminant analysis of principal
components: A new method for the analysis of genetically structured populations.
BMC Genetics 11:94.
Kaeriyama, M., S. Urawa, and M. Fukuwaka. 1995. Variation in Body Size, Fecundity, and
Egg Size of Sockeye and Kokanee Salmon, Oncorhynchus nerka, Released from
Hatchery. Scientific Reports of the Hokkaido Salmon Hatchery 49:1-9.
Kalinowski, S. T. 2009. How well do evolutionary trees describe genetic relationships among
populations? Heredity 102:506-513.
Kalinowski, S. T., D. M. Van Doornik, C. C. Kozfkay, and R. S. Waples. 2012. Genetic
Diversity in the Snake River Sockeye Salmon Captive Broodstock Program as
Estimated from Broodstock Records. Conservation Genetics 13:1183–1193.
Kendall, N. W., U. Dieckmann, M. Heino, A. E. Punt, and T. P. Quinn. 2014. Evolution of
Age and Length at Maturation of Alaskan Salmon Under Size-Selective Harvest.
Evolutionary Applications 7:313–322.
Khrustaleva, A. M., N. V. Klovach, and J. E. Seeb. 2017. Genetic variability and population
structure of Sockeye salmon from the Asian Coast of Pacific Ocean. Russian Journal
of Genetics 53:1126-1136.
Kopelman, N. M., J. Mayzel, M. Jakobsson, N. A. Rosenberg, and I. Mayrose. 2015.
Clumpak: A Program for Identifying Clustering Modes and Packaging Population
Structure Inferences Across K. Molecular Ecology Resources 15:1179-1191.
Langston, A. R. 2012. Williston Watershed Kokanee Spawner Distribution and Enumeration
Surveys (2002 – 2006). Peace/Williston Fish and Wildlife Compensation Program,
Report No. 357, 11 pp.

92

Langston, A. R., and E. B. Murphy. 2008. The History of Fish Introductions (to 2005) in the
Peace/Williston Fish and Wildlife Compensation Program Area. Peace/Williston Fish
and Wildlife Compensation Program, Report No. 325, 60 pp.
Langston, A. R., and R. J. Zemlak. 1998. Williston Reservoir Stocked Kokanee Spawning
Assessment, 1994. Peace/Williston Fish and Wildlife Compensation Program, Report
No. 176, 13 pp.
Lemay, M. A., and M. A. Russello. 2015. Genetic evidence for ecological divergence in
Kokanee salmon. Molecular Ecology 24:798-811.
Leonard, K. C., N. Worden, M. L. Boettcher, E. Dickinson, and A. Hartstone-Rose. 2021.
Effects of Freezing and Short-Term Fixation on Muscle Mass, Volume, and Density.
The Anatomical Record 2021:1–10.
Martinez, P. J., and W. J. Wiltzius. 1995. Some Factors Affecting a Hatchery-Sustained
Kokanee Population in a Fluctuating Colorado Reservoir. North American Journal of
Fisheries Management 15:220–228.
McDermot-Fouts, A. 2019. Williston Watershed Kokanee Spawner Distribution and Aerial
Enumeration Surveys (2018). Peace/Williston Fish and Wildlife Compensation
Program, 31 pp.
McElhany, P., M. H. Rucklelshaus, M. J. Ford, T. C. Wainwright, and E. P. Bjorkstedt. 2000.
Viable Salmonid Populations and the Recovery of Evolutionarily Significant Units.
U.S. Department of Commerce, NOAA Technical Memorandum, NMFS-NWFSC42, 156 pp.
McGurk, M. D. 2000. Comparison of fecundity-length-latitude relationships between
nonanadromous (Kokanee) and anadromous Sockeye salmon (Oncorhynchus nerka).
Canadian Journal of Zoology 78:1791-1805.
McLean, A. R., and B. G. Blackman. 1991. Williston Watershed Aerial Kokanee
(Oncorhynchus nerka) Spawning Survey 1990. Peace/Williston Fish and Wildlife
Compensation Program, Report No. 62, 10 pp.
Morbey, Y. E., and C. G. Guglielmo. 2006. Evidence of a novel reproductive tactic in female
kokanee Oncorhynchus nerka. Journal of Fish Biology 69:1731–1743.
Moreira, A. L., and E. B. Taylor. 2015. The origin and genetic divergence of “black”
Kokanee, a novel reproductive ecotype of Oncorhynchus nerka. Canadian Journal of
Fisheries and Aquatic Sciences 72:1584-1595.
Moyle, P. B., and J. J. Cech. 2004. Fishes: An Introduction to Ichthyology, 5th edition,
University of California, Davis.
Narum, S. R. 2006. Beyond Bonferroni: Less Conservative Analyses for Conservation
Genetics. Conservation Genetics 7:783–787.
93

Narum, S. R., M. Banks, T. D. Beacham, M. R. Bellinger, M. R. Campbell, J. Dekoning, A.
Elz, C. M. Guthrieiii, C. Kozfkay, K. M. Miller, P. Moran, R. Phillips, L. W. Seeb, C.
T. Smith, K. Warheit, S. F. Young, and J. C. Garza. 2008. Differentiating salmon
populations at broad and fine geographical scales with microsatellites and single
nucleotide polymorphisms. Molecular Ecology 17:3464-3477.
Nash, R. D. M., A. H. Valencia, and A. J. Geffen. 2006. The Origin of Fulton’s Condition
Factor—Setting the Record Straight. Fisheries 31:236–238.
Nei, M. 1987. Molecular Evolutionary Genetics. Columbia University Press, New York.
Nelson, J. S. 1968. Distribution and Nomenclature of North American Kokanee,
Oncorhynchus nerka. Journal of the Fisheries Research Board of Canada 25:409-414.
Nelson, R. J., C. C. Wood, G. Cooper, C. T. Smith, and B. Koop. 2003. Population Structure
of Sockeye Salmon of the Central Coast of British Columbia: Implications for
Recovery Planning. North American Journal of Fisheries Management 23:703-720.
Nichols, K. M., C. C. Kozfkay, and S. R. Narum. 2016. Genomic signatures among
Oncorhynchus nerka ecotypes to inform conservation and management of endangered
Sockeye Salmon. Evolutionary Applications 9:1285-1300.
O’Reilly, P., and J. M. Wright. 1995. The evolving technology of DNA fingerprinting and its
application to fisheries and aquaculture. Journal of Fish Biology 47:29-55.
Oke, K. B., C. J. Cunningham, P. A. H. Westley, M. L. Baskett, S. M. Carlson, J. Clark, A. P.
Hendry, V. A. Karatayev, N. W. Kendall, J. Kibele, H. K. Kindsvater, K. M.
Kobayashi, B. Lewis, S. Munch, J. D. Reynolds, G. K. Vick, and E. P. Palkovacs.
2020. Recent Declines in Salmon Body Size Impact Ecosystems and Fisheries. Nature
Communications 11:1–13.
Olsen, J. B., P. Bentzen, and J. E. Seeb. 1998. Characterization of seven microsatellite loci
derived from Pink salmon. Molecular Ecology 7:1087-1089.
Olsen, J. B., J. K. Wenburg, S. A. Pavey, J. L. Miller, and T. R. Hamon. 2017. The Time of
Origin and Genetic Diversity of Three Isolated Kokanee Populations in Southwest
Alaska. Transactions of the American Fisheries Society 146:1212-1222.
Park, S. 2001. Microsatellite Toolkit v. 3.1, Animal Genomics Lab, University College,
Dublin, Ireland.
Patterson, S. D., D. L. Scarnecchia, and J. L. Congleton. 2008. Sexual Maturation in Kokanee
Oncorhynchus nerka. Northwest Science 82:30-47.
Peakall, R., and P. E. Smouse. 2012. GenAlEx 6.5: genetic analysis in Excel. Population
genetic software for teaching and research - an update. Bioinformatics 28:2537-2539.

94

Pearce, T., and S. Abadzadesahraei. 2019. First Nations Information Gathering on Kokanee,
Bull Trout and Arctic Grayling: Report for the Tsay Keh Dene Nation.
Peace/Williston Fish and Wildlife Compensation Program, 24 pp.
Pearce, T., and J. Morgan. 2019. First Nations Information Gathering on Kokanee, Bull Trout
and Arctic Grayling: Report for the Saulteau First Nations. Peace/Williston Fish and
Wildlife Compensation Program, 23 pp.
Peck, K., D. Johner, M. Bassett, T. Weir, and R. Fox. 2019. Kootenay Lake Nutrient
Restoration Program North Arm and South Arm 2017 and 2018 Report. Columbia
Basin Fish and Wildlife Compensation Program, Report No. 164, 68 pp+.
Pillipow, R. A., and A. R. Langston. 2002. Williston Reservoir Fish Assessments 2000,
Pelagic Netting Summary. Peace/Williston Fish and Wildlife Compensation Program,
Report No. 261, 12 pp.
Porras-Hurtado, L., Y. Ruiz, C. Santos, C. Phillips, A. Carracedo, and M. V. Lareu. 2013. An
overview of STRUCTURE: applications, parameter settings, and supporting software.
Frontiers in Genetics 4:1-13.
Pritchard, J. K., M. Stephens, and P. Donnelly. 2000. Inference of population structure using
multilocus genotype data. Genetics 155:945-959.
Rambaut, A. 2018. FigTree v1.4.4. Institute of Evolutionary Biology, University of
Edinburgh, Edinburgh.
Redfish Consulting Ltd. 2016. Kootenay Lake Action Plan - 2016. The Ministry of Forests,
Lands and Natural Resource Operations, Nelson, BC, 35 pp+, Nelson, BC.
Ricker, W. E. 1938. “Residual” and Kokanee salmon in Cultus lake. Journal of the Fisheries
Research Board of Canada 4:192-218.
Ricker, W. E. 1959. Additional Observations Concerning Residual Sockeye and Kokanee
(Oncorhynchus nerka). Journal of the Fisheries Research Board of Canada 16:897–
902.
Rieman, B. E., and D. L. Myers. 1992. Influence of Fish Density and Relative Productivity
on Growth of Kokanee in Ten Oligotrophic Lakes and Reservoirs in Idaho.
Transactions of the American Fisheries Society 121:178–191.
Robinson, M. 2020. Williston Watershed Kokanee Spawner Distribution and Aerial
Enumeration Survey (2019). Peace/Williston Fish and Wildlife Compensation
Program, 42 pp.
Rousset, F. 2008. Genepop'007: A Complete Re-Implementation of the Genepop Software
for Windows and Linux. Molecular Ecology Resources 8:103-106.

95

Saitou, N., and M. Nei. 1987. The Neighbor-joining Method: A New Method for
Reconstructing Phylogenetic Trees. Molecular Biology and Evolution 4:406-425.
Sebastian, D. C., G. Andrusak, G. H. Scholten, and A. R. Langston. 2009. Williston Fish
Index. Peace/Williston Fish and Wildlife Compensation Program, Report No.
GMSMON #13, 89 pp.
Sebastian, D. C., G. H. Scholten, and P. E. Woodruff. 2003. Williston Reservoir Fish
Assessment: Results of Hydroacoustic, Trawl and Gill Net Surveys in August, 2000.
Peace/Williston Fish and Wildlife Compensation Program, Stock Management Report
No. 16, 62 pp.
Selkoe, K. A., and R. J. Toonen. 2006. Microsatellites for ecologists: a practical guide to
using and evaluating microsatellite markers. Ecology Letters 9:615-629.
Small, M. P., T. D. Beacham, R. E. Withler, and R. J. Nelson. 1998. Discriminating Coho
salmon (Oncorhynchus kisutch) populations within the Fraser River, British
Columbia, using microsatellite DNA markers. Molecular Ecology 7:141-155.
Stockner, J. G., A. R. Langston, D. C. Sebastian, and G. Wilson. 2005. The Limnology of
Williston Reservoir: British Columbia’s Largest Lacustrine Ecosystem. Water Quality
Research Journal of Canada 40:28-50.
Takesaki, N., and M. Nei. 1996. Genetic Distances and Reconstruction of Phylogenetic Trees
from Microsatellite DNA. Genetics 144:389-399.
Taylor, E. B., C. J. Foote, and C. C. Wood. 1996. Molecular genetic evidence for parallel
life-history evolution within a Pacific Salmon (Sockeye salmon and Kokanee,
Oncorhynchus nerka). Evolution 50:401-416.
Taylor, E. B., and J. L. Gow. 2010. An Analysis of Population Structure in Kokanee
(Oncorhynchus nerka) from Tributaries of the Upper and Lower Arrow Lakes, West
Kootenay Region, British Columbia using Microsatellite DNA. BC Hydro, 29 pp,
Revelstoke, BC.
Taylor, E. B., S. Harvey, S. Pollard, and J. Volpe. 1997. Postglacial Genetic Differentiation
of Reproductive Ecotypes of Kokanee Oncorhynchus nerka in Okanagan Lake,
British Columbia. Molecular Ecology 6:503–517.
Taylor, E. B., A. Kuiper, P. M. Troffe, D. J. Hoysak, and S. Pollard. 2000. Variation in
developmental biology and microsatellite DNA in reproductive ecotypes of Kokanee,
Oncorhynchus nerka: Implications for declining populations in a large British
Columbia lake. Conservation Genetics 1:231-249.
Taylor, E. B., M. M. Yau, and A. B. Mattock. 2014. Population Structure in Three Species of
Co-Distributed Salmonid Fishes in the Peace River and Tributaries near a Major
Proposed Hydroelectric Development in Northeastern British Columbia, Canada.
River Research and Applications 30(9):1120-1133.
96

Vaha, J. P., and C. R. Primmer. 2006. Efficiency of model-based Bayesian methods for
detecting hybrid individuals under different hybridization scenarios and with different
numbers of loci. Mol Ecol 15(1):63-72.
van Oosterhout, C., W. F. Hutchinson, D. P. M. Wills, and P. Shipley. 2004. MICROCHECKER: software for identifying and correcting genotyping errors in
microsatellite data. Molecular Ecology Notes 4:535-538.
van Poorten, B., S. Harris, and A. Hebert. 2018. Evaluating benefits of stocking on sockeye
recovery projections in a nutrient-enhanced mixed life history population. Canadian
Journal of Fisheries and Aquatic Sciences:1-47.
Veale, A. J., and M. A. Russello. 2016. Sockeye salmon repatriation leads to population reestablishment and rapid introgression with native Kokanee. Evolutionary
Applications 9:1301-1311.
Walter, R. P., T. Aykanat, D. W. Kelly, J. M. Shrimpton, and D. D. Heath. 2009. Gene Flow
Increases Temporal Stability of Chinook Salmon (Oncorhynchus tshawytscha)
Populations in the Upper Fraser River, British Columbia, Canada. Canadian Journal
of Fisheries and Aquatic Sciences 66:167-176.
Walter, R. P., J. M. Shrimpton, and D. D. Heath. 2010. Reply to the comment by Beacham
and Withler on “Gene flow increases temporal stability of Chinook salmon
(Oncorhynchus tshawytscha) populations in the Upper Fraser River, British
Columbia, Canada”. Canadian Journal of Fisheries and Aquatic Sciences 67:206-208.
Ward, H. G. M., P. J. Askey, T. Weir, K. K. Frazer, and M. A. Russello. 2019. Genetic Stock
Identification Reveals That Angler Harvest Is Representative of Cryptic Stock
Proportions in a High‐Profile Kokanee Fishery. North American Journal of Fisheries
Management 39:415–425.
Warnock, W., J. Thorley, S. Arndt, T. Weir, M. Neufeld, J. Burrows, and G. Andrusak. 2021.
Kootenay Lake Kokanee (Oncorhynchus nerka) Collapse into a Predator Pit.
Canadian Journal of Fisheries and Aquatic Sciences Just-IN
https://doi.org/10.1139/cjfas-2020-0410.
Weir, B. S., and C. C. Cockerham. 1984. Estimating F-Statistics for the Analysis of
Population Structure. Evolution 38:1358-1370.
Whitlock, S. L., M. R. Campbell, M. C. Quist, and A. M. Dux. 2018. Using Genetic and
Phenotypic Comparisons to Evaluate Apparent Segregation among Kokanee
Spawning Groups. Transactions of the American Fisheries Society 147:43-60.
Wilson, G., and A. R. Langston. 2000. Williston Reservoir zooplankton analysis program,
1999. Peace/Williston Fish and Wildlife Compensation Program, Report No. 215, 21
pp.

97

Wilson, P., and J. M. Shrimpton. 2020. Genetic Population Structure and Demographics of
Kokanee Introduced to the Williston Reservoir. Peace/Williston Fish and Wildlife
Compensation Program, Project No. PEA-F19-F-2870-DCA, 18 pp.
Wilson, P., and J. M. Shrimpton. 2021. Genetic Population Structure and Demographics of
Kokanee Introduced to the Williston Reservoir. Unpublished report. Peace/Williston
Fish and Wildlife Compensation Program. Project No. PEA-F20-F-3143-DCA, 41 pp.
Withler, R. E. 2007. Genetic analysis and relationships of Kokanee (Oncorhynchus nerka) in
the Peace River drainage. Pacific Biological Station, 3 pp, Nanaimo, British
Columbia.
Withler, R. E., K. D. Le, J. S. Nelson, K. M. Miller, and T. D. Beacham. 2000. Intact genetic
structure and high levels of genetic diversity in bottlenecked Sockeye salmon
(Oncorhynchus nerka) populations of the Fraser River, British Columbia, Canada.
Canadian Journal of Fisheries and Aquatic Sciences 57:1985-1998.
Wood, C. C., J. W. Bickham, R. John Nelson, C. J. Foote, and J. C. Patton. 2008. Recurrent
evolution of life history ecotypes in sockeye salmon: implications for conservation
and future evolution. Evolutionary Applications 1:207-221.
Wood, C. C., and C. J. Foote. 1996. Evidence for Sympatric Genetic Divergence of
Anadromous and Nonanadromous Morphs of Sockeye Salmon (Oncorhynchus
nerka). Evolution 50:1265-1279.
Wood, C. C., B. E. Riddell, D. T. Rutherford, and R. E. Withler. 1994. Biochemical Genetic
Survey of Sockeye Salmon (Oncorhynchus nerka) in Canada. Canadian Journal of
Fisheries and Aquatic Sciences 51:114-131.
Yamamoto, S., S. Kitamura, H. Sakano, and K. Morita. 2011. Genetic structure and diversity
of Japanese Kokanee Oncorhynchus nerka stocks as revealed by microsatellite and
mitochondrial DNA markers. Journal of Fish Biology 79:1340-1349.
Young, S. F., M. R. Downen, and J. B. Shaklee. 2004. Microsatellite DNA data indicate
distinct native populations of kokanee, Oncorhynchus nerka, persist in the Lake
Sammamish Basin, Washington. Environmental Biology of Fishes 69:63-79.

98

APPENDIX A
A2.1. Total number (n) of Williston watershed Kokanee genotypes from each location and
collection year. Hill Creek and Meadow Creek (Columbia River) samples are included as source
stocks.

Group
1
1
2
2
3
3
4
4
4
5
6
6
7
8
8
8
9
10
10
11
11
11
12
13
13
13
14
15
16
16
16
16
17

Location
Arctic Lake
Arctic Lake
Tacheeda Lake
Tacheeda Lake
Hill Creek
Hill Creek
Meadow Creek
Meadow Creek
Meadow Creek
Carbon Creek
Dunlevy Creek
Dunlevy Creek
Manson River
Germansen River
Germansen River
Germansen River
Osilinka River
Pelly Creek
Pelly Creek
Finlay River - Slough
Finlay River - Side
Channel
Finlay River
Tsaydiz Creek
Russel Creek
Russel Creek
Russel Creek
Cutoff Creek
Bower Creek
Aley Creek
Aley Creek
Aley Creek
Aley Creek
Stevenson Creek

Year
2006
2019
2004
2018
2010
2018
2003
2004
2018
1994
1994
2018
2018
2006
2018
2019
2016
2006
2018
2006

n
50
18
50
30
60
40
196
100
49
63
23
41
41
50
40
40
13
49
40
45

2006

47

2019
2006
2006
2018
2019
2018
2006
2016
2017
2018
2019
2016

20
49
49
40
39
40
50
16
10
41
42
10

99

A2.1 Continued

17
18
18
19
20
20
20
21
21

Stevenson Creek
Reservoir Forebay
Reservoir Forebay
Reservoir
Reservoir
Reservoir
Reservoir
Thutade Lake
Thutade Lake

2017
2016
2019
2000
1988
1989
1990
2003
2017
Total

7
95
40
45
36
34
15
87
20
1,870

100

n

68

80

100

345

63

64

41

130

13

89

112

49

128

40

50

109

17

135

45

85

107

Group

Arctic

Tacheeda

Hill

Meadow

Carbon

Dunlevy

Manson

Germansen

Osilinka

Pelly

Finlay

Tsaydiz

Russel

Cutoff

Bower

Aley

Stevenson

Rsvr. Forebay

Rsvr. 2000

Native Rsvr.

Thutade

Ots107

0.061

Ots108

0.078

Ots100

0.057*

Oki10

0.157

Oki16

Oki1a

0.139

Oki1b

Oki29

Oki6

0.135

Omy77

0.113

One8

Ots103

Ots2

101

0.080*

0.086*

Ots3

A2.2. Evidence of null alleles due to homozygote excess (Aley; Thutade) and likely a Wahlund effect (Rsvr. 2000). Statistically significant
values (P < 0.05) are indicated with *.

A2.3. Pairwise Nei's (1987) FST genetic distance estimates between native Kokanee sampled from
Arctic and Tacheeda Lakes, the body of the reservoir before stocking events, and Thutade Lake are
represented below the diagonal. HIERFSTAT bootstrapping over loci confidence intervals are presented
in brackets above the diagonal. Bolded values are significantly different from zero.

Arctic
Tacheeda
Native Rsvr.
Thutade

Arctic
0.049
0.398
0.327

Tacheeda
(0.030–0.075)
0.404
0.332

Native Rsvr.
(0.286–0.496)
(0.281–0.524)
0.080

Thutade
(0.229–0.393)
(0.230–0.415)
(0.050–0.114)

102

0.049

0.330

0.323

0.349

0.359

0.375

0.328

0.392

0.340

0.331

0.363

0.331

0.374

0.365

0.341

0.398

0.331

0.336

0.395

0.317

Hill

Meadow

Carbon

Dunlevy

Manson

Germansen

Osilinka

Pelly

Finlay

Tsaydiz

Russel

Cutoff

Bower

Aley

Stevenson

Rsvr.
Forebay

Rsvr. 2000

Native
Rsvr.

Thutade

Arctic

Tacheeda

Arctic

Group

(0.014—
0.106)

(0.013—
0.099)

0.325

0.403

0.347

0.338

0.407

0.349

0.375

0.382

0.337

0.374

0.337

0.348

0.401

0.337

0.386

0.368

0.360

0.098

0.163

0.048

0.003

0.013

0.004

0.007

0.002

0.004

0.006

0.004

0.006

0.003

0.003

0.004

0.005

0.004

0.111

0.185

0.083

0.043

0.069

0.047

0.056

0.050

0.049

0.050

0.049

0.053

0.043

0.045

0.048

0.053

0.050

(0.001—
0.012)

(0.000—
0.009)

0.026

(0.008—
0.051)

0.329

0.337

(0.262—
0.456)

(0.236—
0.424)

0.104

0.152

0.036

0.000

0.000

0.001

0.000

0.000

0.001

0.003

0.000

0.000

0.000

0.000

0.001

0.000

0.108

0.163

0.042

0.000

0.000

0.001

0.000

0.000

0.001

0.004

0.000

0.000

0.000

0.001

0.000

(0.000—
0.001)

(0.263—
0.470)

(0.261—
0.446)

(0.243—
0.432)

(0.254—
0.432)

(0.232—
0.406)

Dunlevy

(0.241—
0.408)

Carbon

(0.030—
0.075)

Meadow

Hill

Tacheeda

0.121

0.181

0.049

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

(0.000—
0.004)

(0.000—
0.004)

(0.019—
0.086)

(0.002—
0.007)

(0.279—
0.490)

(0.272—
0.469)

Manson

0.106

0.159

0.045

0.000

0.000

0.000

0.002

0.000

0.000

0.000

0.000

0.000

0.002

(0.000—
0.002)

(0.000—
0.003)

(0.000—
0.001)

(0.015—
0.090)

(0.000—
0.007)

(0.249—
0.421)

(0.241—
0.404)

Germansen

0.101

0.173

0.042

0.000

0.000

0.004

0.000

0.000

0.001

0.001

0.000

0.000

(0.000—
0.010)

(0.000—
0.010)

(0.000—
0.006)

(0.000—
0.008)

(0.001—
0.102)

(0.000—
0.016)

(0.282—
0.518)

(0.274—
0.497)

Osilinka

0.115

0.177

0.053

0.000

0.000

0.002

0.000

0.000

0.000

0.001

0.000

(0.000—
0.005)

(0.000—
0.001)

(0.000—
0.003)

(0.000—
0.002)

(0.000—
0.002)

(0.013—
0.108)

(0.000—
0.014)

(0.252—
0.439)

(0.246—
0.420)

Pelly

0.107

0.162

0.045

0.000

0.000

0.000

0.000

0.000

0.000

0.001

(0.000—
0.000)

(0.000—
0.010)

(0.000—
0.000)

(0.000—
0.000)

(0.000—
0.001)

(0.000—
0.002)

(0.016—
0.094)

(0.001—
0.007)

(0.244—
0.424)

(0.241—
0.408)

Finlay

0.110

0.176

0.050

0.001

0.000

0.002

0.004

0.000

0.000

(0.000—
0.004)

(0.000—
0.006)

(0.000—
0.008)

(0.000—
0.003)

(0.000—
0.002)

(0.000—
0.008)

(0.000—
0.008)

(0.017—
0.099)

(0.000—
0.014)

(0.272—
0.474)

(0.261—
0.456)

Tsaydiz

0.107

0.165

0.050

0.000

0.000

0.001

0.002

0.000

(0.000—
0.001)

(0.000—
0.000)

(0.000—
0.002)

(0.000—
0.011)

(0.000—
0.000)

(0.000—
0.002)

(0.000—
0.004)

(0.000—
0.002)

(0.016—
0.097)

(0.000—
0.010)

(0.246—
0.423)

(0.241—
0.409)

Russel

0.110

0.173

0.046

0.000

0.000

0.000

0.000

(0.000—
0.000)

(0.000—
0.004)

(0.000—
0.000)

(0.000—
0.000)

(0.000—
0.009)

(0.000—
0.001)

(0.000—
0.000)

(0.000—
0.000)

(0.000—
0.000)

(0.016—
0.098)

(0.000—
0.007)

(0.274—
0.488)

(0.270—
0.468)

Cutoff

0.115

0.173

0.049

0.002

0.001

0.003

(0.000—
0.000)

(0.000—
0.005)

(0.000—
0.009)

(0.000—
0.003)

(0.000—
0.002)

(0.000—
0.001)

(0.000—
0.006)

(0.000—
0.003)

(0.000—
0.003)

(0.000—
0.001)

(0.016—
0.110)

(0.002—
0.014)

(0.271—
0.475)

(0.267—
0.451)

Bower

0.113

0.165

0.047

0.000

0.000

(0.000—
0.008)

(0.001—
0.005)

0.123

0.188

0.047

0.000

(0.000—
0.006)

(0.000—
0.006)

(0.000—
0.005)

(0.000—
0.005)

(0.000—
0.002)

(0.000—
0.005)

(0.000—
0.011)

(0.000—
0.006)

(0.000—
0.007)

(0.000—
0.006)

(0.000—
0.005)

(0.019—
0.134)

(0.002—
0.025)

(0.281—
0.532)

(0.274—
0.516)

Stevenson

(0.000—
0.000)

(0.000—
0.004)

(0.000—
0.005)

(0.000—
0.001)

(0.000—
0.004)

(0.000—
0.015)

(0.000—
0.000)

(0.000—
0.000)

(0.000—
0.004)

(0.000—
0.004)

(0.018—
0.085)

(0.002—
0.007)

(0.258—
0.436)

(0.253—
0.417)

Aley

0.110

0.165

0.046

(0.000—
0.005)

(0.000—
0.000)

(0.000—
0.004)

(0.000—
0.001)

(0.000—
0.002)

(0.000—
0.006)

(0.000—
0.000)

(0.000—
0.002)

(0.000—
0.006)

(0.000—
0.000)

(0.000—
0.002)

(0.000—
0.004)

(0.000—
0.001)

(0.014—
0.082)

(0.000—
0.006)

(0.250—
0.418)

(0.245—
0.403)

Rsvr.
Forebay

0.051

0.048
0.079

103

(0.050—
0.113)

(0.055—
0.201)

(0.062—
0.171)

(0.055—
0.188)

(0.054—
0.174)

(0.028—
0.083)

(0.110—
0.242)

(0.026—
0.068)

(0.056—
0.164)

(0.028—
0.072)

(0.110—
0.224)

(0.031—
0.071)

(0.053—
0.174)

(0.055—
0.171)

(0.110—
0.249)

(0.026—
0.079)

(0.054—
0.166)

(0.106—
0.227)

(0.105—
0.224)

(0.027—
0.066)

(0.054—
0.186)

(0.028—
0.067)

(0.110—
0.247)

(0.029—
0.081)

(0.041—
0.180)

(0.113—
0.269)

(0.110—
0.244)

(0.023—
0.065)

(0.057—
0.160)

(0.022—
0.075)

(0.103—
0.218)

(0.028—
0.064)

(0.062—
0.189)

(0.107—
0.227)

(0.112—
0.254)

(0.027—
0.073)

(0.048—
0.176)

(0.109—
0.241)

(0.097—
0.235)

(0.021—
0.066)

(0.053—
0.166)

(0.028—
0.067)

(0.094—
0.214)

(0.021—
0.055)

(0.046—
0.188)

(0.052—
0.150)

(0.229—
0.414)

(0.227—
0.391)

Thutade

(0.027—
0.074)

(0.113—
0.264)

(0.109—
0.220)

(0.279—
0.521)

(0.283—
0.495)

Native
Rsvr.

(0.044—
0.132)

(0.031—
0.067)

(0.245—
0.448)

(0.244—
0.420)

Rsvr.
2000

A2.4. Pairwise Weir and Cockerham (1984) ! genetic distance estimates between Williston watershed Kokanee sample groups are represented
below the diagonal. HIERFSTAT bootstrapping over loci confidence intervals are presented in brackets above the diagonal. Bolded values are
significantly different from zero

A2.5. Evanno table output of the STRUCTURE analysis of the comprehensive dataset, as obtained through
STRUCTURE HARVESTER. A K of 2 was most supported by the Evanno protocol, but this represents an
oversimplification of the population structure (Janes et al. 2017).
K

Reps

Mean LnP(K)

Stdev LnP(K)

Ln'(K)

|Ln''(K)|

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21

10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10

-93900.99
-87409.52
-89542.9
-83008.5
-82695.23
-82653.93
-83466.39
-83515.39
-83647.52
-84441.33
-84017.47
-85150.13
-84621.81
-84886.79
-84754.09
-85575.96
-85448.15
-84981.69
-86313.02
-86499.75
-85629.25

0.1287
26.2144
14096.1117
50.3545
861.6011
152.2466
1017.1268
565.1225
630.6098
1325.4609
755.015
1562.0687
1048.3229
1048.4832
862.9525
1269.9443
1101.7455
1095.4726
1668.0079
2814.3084
909.1458

6491.47
-2133.38
6534.4
313.27
41.3
-812.46
-49
-132.13
-793.81
423.86
-1132.66
528.32
-264.98
132.7
-821.87
127.81
466.46
-1331.33
-186.73
870.5

8624.85
8667.78
6221.13
271.97
853.76
763.46
83.13
661.68
1217.67
1556.52
1660.98
793.3
397.68
954.57
949.68
338.65
1797.79
1144.6
1057.23
-

DK
329.011303
0.614906
123.546765
0.315657
5.607746
0.750605
0.147101
1.04927
0.918677
2.061575
1.063321
0.756732
0.379291
1.106168
0.747812
0.307376
1.641109
0.686208
0.375662
-

A2.6. Evanno table output of the STRUCTURE analysis of the tributary dataset, as obtained through
STRUCTURE HARVESTER. A K of 2 was most supported by the Evanno protocol, but this represents an
oversimplification of the population structure (Janes et al. 2017).
K
1
2
3
4

Reps
10
10
10
10

Mean LnP(K)
-42556.69
-42657.88
-42927.23
-43512.67

Stdev LnP(K)
0.1449
12.4035
80.05
179.79

Ln'(K)
-101.19
-269.35
-585.44

|Ln''(K)|
168.16
316.09
-

DK
13.55749
3.948658
-

104

A2.7. Plot of the Bayesian Information Criterion (BIC) values from which the lowest optimal number of clusters
(K) was chosen. A K of 4 was selected for the comprehensive dataset.

105

A2.8. A principal component analysis (PCA) for the tributary dataset showing the lack of cluster patterns exhibited
by Kokanee from these sample locations.

106