The Movement of Marine-Derived Nutrients from a Salmon Spawning River to a
Nursery Lake
by
Jacob A. Duros
BSc. University of Wisconsin-Whitewater, 2011

THESIS SUBMITTED IN PARTIAL FULFILLMENT OF
THE REQUIRMENTS FOR THE DEGREE OF
MASTER OF SCIENCE
IN
NATURAL RESOURCES AND ENVIRONMENTAL STUDIES

UNIVERSITY OF NORTHERN BRITISH COLUMBIA
December 2017
Jacob Duros, 2017

Abstract
Salmon play a key role in the redistribution of marine-derived nutrients (MDNs) in aquatic
and terrestrial ecosystems. Research conducted on the movement and storage of MDNs in aquatic
systems throughout the Pacific Northwest seem to vary in whether MDNs have a beneficial,
neutral, or detrimental impact. Using Horsefly Bay (Quesnel Lake), the mechanism and driving
factors for the delivery and dispersion of MDNs were evaluated. Higher concentrations of marinederived nitrogen and carbon were found to enter this nursery system in the fall spawning period.
However, due to the increased water discharge, it was found that the load of marine-derived
nitrogen and carbon was higher during the spring freshet study period. These increases in MDNs
were found to correlate with chlorophyll-a and fluorescence levels which indicate increases in
productivity. Increased production can support the growth and survivorship of juvenile salmon
rearing in this nursery system through bottom-up trophic transfer.

ii

Table of Contents
Acronyms .............................................................................................................................. v
List of Tables ........................................................................................................................ vi
List of Figures ...................................................................................................................... vii
Preface .................................................................................................................................. x
Acknowledgements .............................................................................................................. xi
Chapter 1: Marine-Derived Nutrients and Salmon; A Literature Review .................................. 1
1.1 Introduction .............................................................................................................................1
1.2 Salmon Decomposition .............................................................................................................3
1.3 Salmon versus Non-Salmon Effects ...........................................................................................4
1.4 Plankton Background................................................................................................................5
1.5 MDNs Effect on Water Chemistry..............................................................................................6
1.6 Movement and Behavior of Marine-Derived Nutrients ..............................................................7
1.6.1 MDNs in Nursery Lakes ................................................................................................................ 8
1.6.2 Primary Productivity & Nutrient Pulses ....................................................................................... 9
1.7 Plume Dispersion Effects at River Lake Interface ..................................................................... 10

Chapter 2: Timing and Delivery of MDNs from a Spawning Salmon River to a Nursery Lake The Horsefly River/Quesnel Lake system .............................................................................. 13
2.1 Introduction ........................................................................................................................... 13
2.2 Research Question ................................................................................................................. 13
2.3 Methods ................................................................................................................................ 14
2.3.1 Study Site ...................................................................................................................................14
2.3.1.1 Site Characteristics ............................................................................................................................ 14

2.3.2 Study Design ...............................................................................................................................15
2.3.3 Sampling Techniques (Collection & Analysis) ............................................................................16
2.3.3.1 Isotopes.............................................................................................................................................. 16
2.3.3.2 Concentration of Suspended Particulate Matter .............................................................................. 17
2.3.3.3 Chlorophyll-A ..................................................................................................................................... 18
2.3.3.4 Phosphorous ...................................................................................................................................... 19
2.3.3.5 Statistical Analysis.............................................................................................................................. 19

2.4 Results ................................................................................................................................... 21
2.4.1 Suspended Particulate Characteristics ......................................................................................21
2.4.1.1 Total Particulate................................................................................................................................. 21
2.4.1.2 Organic vs. Inorganic Particulates ..................................................................................................... 22
2.4.1.3 Stable Isotopes: Nitrogen and Carbon (Enrichment) ........................................................................ 24
2.4.1.4 Stable Isotopes: Nitrogen and Carbon (Mixing Model) ..................................................................... 24
2.4.1.4 Stable Isotopes: Nitrogen and Carbon (Specific Load) ...................................................................... 26
2.4.2.1 Chlorophyll-a & Fluorescence ............................................................................................................ 28
2.4.2.2 Phosphorus ........................................................................................................................................ 29
2.4.2.3 Comparison of Variables / Linear Models ......................................................................................... 30

2.5 Discussion .............................................................................................................................. 33
2.5.1 Suspended Particulate Characteristics / Load ...........................................................................34
2.5.2 Isotope Mixing Model ................................................................................................................37
2.5.3 Chlorophyll-a Concentrations & Fluorescence ..........................................................................38

iii

2.5.4 Parameter Comparisons ............................................................................................................39
2.6 Implications ........................................................................................................................... 40

Chapter 3: The Movement and Dispersal of a River Plume and its Contents in a Nursery Lake –
Quesnel Lake ....................................................................................................................... 41
3.1 Introduction ........................................................................................................................... 41
3.1.2 Research Question .....................................................................................................................42
3.2 Methods ................................................................................................................................ 42
3.2.1 Site Characteristics .....................................................................................................................42
3.2.2 Study Design ...............................................................................................................................43
3.2.3 Sampling Techniques (Collection and Analysis) .........................................................................44
3.2.3.1 Observations of Water Column Parameters ..................................................................................... 44
3.2.3.2 Observations of Water Column Parameters from the LISST ............................................................. 44

3.2.4 Analysis of Results ......................................................................................................................45
3.3 Results ................................................................................................................................... 46
3.3.1 Water Properties ........................................................................................................................46
3.3.1.1 Temperature & Thermal Structure .................................................................................................... 46
3.3.1.3 D50 ...................................................................................................................................................... 50

3.3.2 Movement of the Plume ............................................................................................................51
3.3.2.1 Overhead View of Turbidity ............................................................................................................... 51
3.3.2.2 Turbidity’s Vertical Profile and Distance from Delta ......................................................................... 53

level of approximately 3 NTU. This increase in turbidity covered the whole water column 30 meters in
depth to the bottom at 100 meters. .............................................................................................. 60
3.4 Discussion .............................................................................................................................. 60
3.4.2 The Dispersal of the Horsefly River Plume.................................................................................63
3.4.3 Implications ................................................................................................................................64

Chapter 4: Conclusions and Management Implications ......................................................... 65
4.1 Conclusions ............................................................................................................................ 65
4.2 Future Management Implications ........................................................................................... 66

References .......................................................................................................................... 68
Appendix 1: Marine-Derived Nitrogen Load Equation ........................................................... 76
Appendix 2: R Script for Data Processing .............................................................................. 78

iv

Acronyms
AIC - akaike information criterion

LOESS - curve fitting local polynomial regression

12C - lighter isotope (naturally occurring)

MDN - marine-derived nutrient

13C - heavier isotope (naturally occurring)

14N - lighter isotope (naturally occurring)

CFC - continuous flow centrifuge

15

Chl-a - chlorophyll-a

predominantly marine-derived)

CTD - Seabird water profiler (conductivity,

NH4 - ammonium

temperature, depth)

NO-3 - nitrate

D50 - median diameter particle size

OSP - organic suspended particulate

DOC - dissolved organic carbon

P - phosphorous

DP - dissolved phosphorous

PO4-3 - orthophosphate

HFB - Horsefly Bay

QRRC - Quesnel River Research Centre

HFR - Horsefly River

SIAR - stable isotope analyzes in R

HSD - honest significance difference

SIMMR - stable isotope mixing model in R

HVFS - high volume filtration system

SPM - suspended particulate matter

ISP - inorganic suspended particulate

TN - total nitrogen

LISST - laser in-situ scattering transmissometry

TP - total phosphorous

N - heavier isotope (naturally occurring,

UV-VIS - ultraviolet visible spectrophotometry

v

List of Tables
1.1

Table showing various studies throughout Alaska and British Columbia that suggest a
positive (+), negative ( - ), or neutral effect that MDNs have on future salmon stocks. ....2

1.2

Results from four studies looking at water quality and biotic differences among salmon
bearing and non-salmon bearing streams and lakes. (+) denotes there was higher levels
or populations in salmon bearing streams and lakes. (N) denotes there was no
significant difference found in salmon bearing streams and lakes. (NA) denotes that the
study did not mention parameter. Results with * were taken from same lake; however,
salmon bearing was at the outflow of a salmon spawning river and non-salmon bearing
was in the middle of lake away from any salmon spawning river
mouths.…………………………………………………………………………………………………………… ..............5

2.1

Salmon escapement from 2013-2016 from the Horsefly Watershed and the Horsefly
River (HFR) specifically ..................................................................................................... 14

2.2

Linear models created and selected by AIC to test which parameters drive the delivery
of  15N levels and the load of 15N to HFB......................................................................... 21

2.3

Results from a two-way ANOVA of spatial (study site) and temporal (study period) for
chlorophyll-a levels in HFB, from a one-way ANOVA of temporal (study period) for
phosphorus levels and forms in HFB, and a one-way ANOVA of temporal (study period)
for fluorescence levels at site S1, S3, and S5 in HFB. The phosphorus form interaction
term includes dissolved and total phosphorus ................................................................. 31

2.4

Statistics results performed by an AIC on which parameters drive the delivery of 15N
levels and the load of 15N to HFB. ..................................................................................... 32

2.5

Results from a linear model function showing interactions between factors and 15N
values in Horsefly Bay. ...................................................................................................... 32

vi

List of Figures
1.1

Trophic pathways influenced by anadromous salmon in freshwater systems and the
energy flow within these systems. .....................................................................................3

1.2

Depicts a river/lake interface ecosystem during situations when river waters are colder
and more sediment-laden than lake surface waters (A) and when river waters are
warmer and less sediment-laden than lake surface waters (B). Blue arrows signify flow
path. (Desloges & Gilbert, 1998; Boehrer, 2008) ............................................................. 11

2.1

Map displaying sampling sites in HFB. .............................................................................. 15

2.2

HFR hydrograph, data taken at the river delta every five minutes. Red lines designate
sampling periods in 2014 and 2015 (Environment Canada)............................................. 16

2.3

Suspended particulate matter concentrations and discharge over time at site S5. Vertical
lines denote the three study periods................................................................................ 22

2.4

Organic and inorganic suspended particulate matter concentrations covering the three
study periods at site S5 ..................................................................................................... 23

2.5

The concentration of 15N and 13C in SPM over three study periods collected at site S5.
Method used to create blue lines was LOESS ................................................................... 24

2.6

SIMMR mixing model, showing the 15N 13C values of SPM samples collected
throughout the freshet, summer, and spawning sample periods at site S5 .................... 25

2.7

Proportion of three sources (aquatic vegetation, terrestrial vegetation, and salmon)
found in SPM samples taken from site S5 during the spring freshet, summer, and fall
spawn. The line in the middle of the boxes for each source signifies the median point of
the data. Boxes above and below the median line represent the upper and lower
quartile. At the end of the solid vertical lines are the highest and lowest data values
excluding outliers which are the black dots in this plot ................................................... 26

2.8

Chlorophyll-a, delta 15N, 15N loading, TP, and SPM data levels collected in HFB over the
three study periods. All data used in this plot, except chlorophyll-a, was collected from
site S5. Chlorophyll-a data used was collected from S1 ................................................... 27

2.9

Chlorophyll-A levels at four sites throughout HFB over the three study periods. Lines
were created using LOESS ................................................................................................. 28

vii

2.10

Fluorescence levels at three sites throughout HFB over the three study periods. .......... 29

2.11

Phosphorus levels in three forms (dissolved, particulate, total) at site S5 over the three
study periods. Trend lines were made using LOESS ......................................................... 30

3.1

Map showing CTD sampling stations in HFB..................................................................... 43

3.2

Diagram of DFO mooring setup ........................................................................................ 45

3.3

Temperature values of the HFR and the surface water of HFB throughout the period of
record (2014 fall spawning period, 2015 spring freshet, summer, and fall spawning
periods). Method used to create trend lines in R was LOESS. River water temperature
was collected using the Hobo while HFB temperatures were collected using the DFO
mooring ............................................................................................................................. 47

3.4

Temperature profiles at site S3 throughout the three study periods. Data obtained using
the CTD. Lines created with LOESS ................................................................................... 48

3.5

Site S3 SPM concentrations at five depths over the three study periods (2014-2015).
Data was collected by using a LISST and the plotting method in R used to create trend
lines was LOESS ................................................................................................................. 49

3.6

Site S3 percent transmission levels at four depths over the three study periods. Data
was collecting by using a LISST and the plotting method in R used to create trend lines
was LOESS ......................................................................................................................... 50

3.7

Site S3 D50 levels at various depths over the three study periods. Data was collecting by
using a LISST and the plotting method in R used to create trend lines was LOESS.......... 51

3.8

Freshet study period overhead view maps of HFB on April 24, 2015 (A), May 29, 2015
(B), and June 19, 2015 (C). Colors show NTU levels of highest water turbidity across the
bay. Turbidity colors were generated using ArcMap’s interpolation kriging function..... 54

3.9

Summer study period overhead view maps of HFB from July 2, 2015 (A), July 24, 2015
(B), and August 7, 2015 (C). Colors show NTU levels of turbidity across the bay. Turbidity
colors were generated using ArcMap’s interpolation kriging function. ........................... 55

3.10

Fall spawn study period overhead view map of HFB from October 7, 2015 (A), and
October 20, 2015 (B). Colors show NTU levels of turbidity across the bay. Turbidity
colors were generated using ArcMap’s interpolation kriging function. ........................... 56

viii

3.11

Interpolated vertical profile plot of the HFR plume (turbidity) as it enters at the mouth of
the HFR and exits at the end of HFB. These plots represent data taken from April 24,
2015 through July 2, 2015. Graph was created in R using ggplot and data was
interpolated using bicubic spline interpolation. White contour lines denote changes in
turbidity............................................................................................................................. 58

3.12

Interpolated vertical profile plot of the HFR plume (turbidity) as it enters at the mouth of
the HFR and exits at the end of HFB. These plots represent data taken from July 24, 2015
through October 20, 2015. Graph was created in R using ggplot and data was
interpolated using bicubic spline interpolation. White contour lines denote changes in
turbidity............................................................................................................................. 59

4.1

2014 through 2017 hydrograph data taken from Environment Canada - station number
08KH031 in Horsefly River above Quesnel Lake. .............................................................. 68

ix

Preface
This thesis is divided into four chapters. Chapter 1 provides an overview of the importance
and process of the delivery and incorporation of marine-derived nutrients (MDNs) into nursery
systems across the Pacific Northwest. Specifically, this chapter illustrates the biological impacts
of the decomposition of salmon and the release and dispersal of nutrients into a nursery system.
Results in Chapter 2 examine the timing and transportation of MDNs into a nursery lake and the
driving factors behind this delivery. This chapter looks at temporal variations and the physical
characteristics of MDNs entering a nursery lake. Chapter 3 analyzes the movement of particulates
vertically and horizontally throughout the water column of the bay which receives discharge from
the salmon-spawning river. Tracking of these particles provides data to better predict the
dispersal of nutrients entering a nursery lake. Chapter 4 presents conclusions and management
implications based on the results that were presented in Chapters 2 and 3.

x

Acknowledgements
There are many people I would like to thank, as without them this thesis project would
have never been possible. First, I would like to thank my supervisor Ellen Petticrew for the
opportunity to pursue my Master’s degree. The knowledge, time, and support you have provided
have helped me in becoming more knowledgeable in the field of ecology and have shown me
what it takes to be successful. I will always be grateful that you allowed me to join your graduate
team and gave me the tools I need to succeed.
I would also like to thank my committee members Samuel Albers and Svein Vagle. Sam,
the statistical and R-coding help and knowledge you provided me will inevitably make me more
successful in pursuing jobs in the biology/ecology field. I appreciate the patience and time you
gave to me as I slowly progressed in my coding capability. I can’t imagine how much longer this
thesis would have taken if I didn’t have your guidance and help. Svein, allowing me to help you
deploy and collect moorings allowed me to gain a better understanding of water structure and
the technologies out there that can help interpret processes that are happening throughout the
water column. I thank you for allowing me to tag along and absorb the knowledge and experience
you possess. Once again, I would like to thank my entire committee and the external reviewer
Dr. Kyle Hodder. Each of you have contributed significantly to this project and I look forward to
possibly working with you all again in the future.
This project would not have been successful or completed without the people that helped
me in the field and lab. Special thanks go to Kristy Rasmus, who helped me tremendously in
collecting and analyzing data. Big thanks also go out to Caitlin Langford, Lazlo Enyedy, Alex Koiter
and Stephanie LeZertze, Kristen Kieta, Todd French, Will Proctor, Erwin Rehl, and Hossein
Kazemian.
Much appreciation also goes out to my family who have supported me and continue to
support me in achieving my life and career goals. Thanks to my parents: Jeffrey and Jacqueline
Duros, my sister: Bree Duros, my stepsons: Cohen and Keagan Reid, my in-laws: Bernard and
Rhonda Wiebe, and my daughter: Adelynn Duros. Your support has tremendously helped me
through this long process and I appreciate the time and love you have given me.
Finally, I need to thank my wife, Jaden Duros. Being pregnant with our daughter, working
long hours, and caring for our children when I was focused on sampling and sometimes being
gone for three to four days at a time was heroic. Words cannot describe how truly grateful and
lucky I am.

xi

Chapter 1: Marine-Derived Nutrients and Salmon; A Literature Review

Chapter 1: Marine-Derived Nutrients and Salmon; A Literature Review
1.1 Introduction
The movement of suspended particulates throughout watersheds is an important
hydrological, fluvial, and geomorphological process. These particulates composed of sediment
and nutrients become suspended in rivers and streams and make their way from high elevations
to lower elevations until they settle out into slower moving water bodies (Owens, 2005). This
results in larger particles settling out faster and finer particles staying suspended longer and
moving farther downstream. Nutrients and contaminants tend to bind more easily to finer
particles in the water column due to their larger surface area/volume ratios (Horowitz & Elrick,
1988). These finer inorganic particles have a propensity to bind available nutrients through the
process of flocculation and in turn modify the transportation of both materials through aquatic
ecosystems (Choles, 2004; McConnachie & Petticrew, 2006). This process is particularly
important in interior oligotrophic British Columbian lakes and rivers that are used for salmon
spawning and rearing as these systems are usually nutrient limited (Naiman et al., 2002).
Each year in these ecosystems millions of Pacific salmon migrate from the sea, where they
have accumulated more than 95% of their body mass, to their natal freshwater streams to spawn
and die (Naiman et al., 2002). The displacement and suspension of particulate matter by
spawning salmon, in rivers and streams, is an important process that supplies nursery lakes with
pulses of sediment, organic/inorganic matter, and nutrients (Petticrew et al., 2011 & Reisinger,
2013).
Since the early 1990’s salmon numbers observed on the west coasts, and most interior
rivers and streams of British Columbia and Alaska, have been in steady decline (Stouder et al.,
2012). Fewer salmon will result in smaller run sizes and subsequently less nutrient influx to
streams. Gresh et al. (2000) and Naiman et al. (2002) performed a study to compare historical
levels of biomass, nitrogen, and phosphorus in Pacific Northwest nursery lakes to what was
present in the early 1990’s and found that there was an 87%-93% reduction of biomass and
nutrients delivered to these systems. This decrease in nutrients will limit the amount of nutrient
and carbon sources available to organisms living in these nursery lakes (Larkin & Slaney, 1997).
1

Chapter 1: Marine-Derived Nutrients and Salmon; A Literature Review

At a larger scale, the decrease in any marine-derived nutrient (MDN) may inhibit many aquatic
systems in sustaining salmon populations (Bilby et al., 1996). Although this assumption has rarely
been tested, a better understanding of the behavior and distribution of MDNs in limnetic
ecosystems will allow a more informed approach in the conservation and management of future
salmon populations.
Nursery lakes are an important habitat for some species of juvenile salmon for the first
one to two years of their lives before they migrate to the ocean (Naiman et al., 2002). These lakes
are typically oligotrophic and primary productivity is severely nutrient-limited (Naiman et al.,
2002). The transfer of MDNs into these nursery lakes increases productivity and helps drive algae,
zooplankton, and fish populations in a bottom-up trophic transfer (Hyatt et al. 2004; DeVries,
2012). Although, with this increase in productivity due to MDNs, research has not demonstrated
a clear link between lake productivity and salmon-spawner population size (Naiman et al., 2002;
Holtgrieve & Schindler, 2011).
Table 1.1: Table showing various studies throughout Alaska and British Columbia that
suggest a positive (+), negative ( - ), or neutral effect that MDNs have on future salmon
stocks.
Study

Study Location

Stockner and Macisaac (1996)

BC

MDN Impacts of Future
Salmon Populations
+

Schmidt et al. (1998)

BC

+

Finney et al. (2000)

Alaska

+

Stocker (2003)

Alaska & BC

+

Schindler et al. (2005)

Alaska

Neutral

Uchiyama et al. (2008)

Alaska

Neutral

Walters (2014)

BC

-

Results of research on the impacts that spawning salmon have on ecosystems, such as the
disturbance and redistribution of stream bed sediment and the mass influx of MDNs, vary as to
what effect MDNs have upon their natal systems (Table 1.1). Some studies show that the influx
of MDNs help increase riverine benthic productivity as well as increase survivorship of juvenile
2

Chapter 1: Marine-Derived Nutrients and Salmon; A Literature Review

salmon in downstream rearing lakes (Wipfli et al., 1998; Naiman et al., 2002; Johnson et al., 2004)
while others show a detrimental or neutral impact with nutrient export by smolts exceeding
nutrient import from spawning salmon as well as an observed decline in riverine primary
productivity (Holtgrieve & Schindler, 2011; Walters, 2014). However, these varying results may
be credited to the lack of understanding of the spatial and temporal factors that affect the
behavior of MDNs and therefore may be more system-specific than a general overall observation.
It is the objective of this study to determine seasonal quantitative differences in the
amount of MDNs flushed into a nursery lake from a river system and the depth at which these
nutrients disperse in the water column over the course of a salmon-spawning event, spring
freshet, and a summer period.

1.2 Salmon Decomposition
Nutrients from salmon carcasses are utilized and incorporated into the surrounding biota
by direct consumption of dead salmon tissue (Cederholm et al., 1999), the uptake of soluble
nutrients by primary producers (Schindler et al., 2003), and the uptake of dissolved and
particulate organic matter by micro-fauna (invertebrates) in streambed substrate (Claeson et al.,

Figure 1.1: Trophic pathways influenced by anadromous salmon in freshwater
systems and the energy flow within these systems.
3

Chapter 1: Marine-Derived Nutrients and Salmon; A Literature Review

2006)(Figure 1.1). These nutrients have been associated with the enhancement of algae and
aquatic invertebrate populations near carcasses and the increase of phytoplankton and
chlorophyll-a concentrations in nursery lakes (Mathisen, 1972; Schmidt et al., 1998; Schindler et
al., 2003). Organic matter and nutrients released from these carcasses can be the most significant
source of annual nutrient input and may be important in the growth of juvenile salmon that
reside in these ecosystems (Naiman et al., 2002).
The ability to identify MDNs in these aquatic systems helps track the effects that salmon
carcasses and their nutrients have on an ecosystem. Isotopic signatures of marine-derived
nitrogen (15N) have been used to trace MDNs throughout watersheds (Hocking & Reynolds, 2011,
Albers & Petticrew, 2013). The nitrogen derived from an adult salmon has a higher level of 15N
than that found in freshwater aquatic and terrestrial sources, allowing for the identification of
MDNs (Uchiyama et al., 2008). The percent 15N in an adult Sockeye Salmon averages around 11%12% of total N, while terrestrial sources of 15N are closer to zero (Kline et al., 1993; Welchand &
Parsons, 1993; Satterfield & Finney, 2002). This allows the use of the isotopic signature of 15N to
be used to trace MDNs pathways in freshwater ecosystems (Kline et al., 1990).

1.3 Salmon versus Non-Salmon Effects
The introduction of salmon carcasses to spawning rivers and lakes can have many effects
on biota that live in these watersheds (Bilby et al., 2011). Understanding the differences in
nutrient availability and nutrient uptake between salmon bearing rivers and lakes and nonsalmon bearing rivers and lakes is important in understanding how salmon alter and possibly
enhance these ecosystems. Bilby et al. (2011), Claeson et al. (2011), and Wipfli et al. (2011) have
performed studies in which salmon carcasses were added to streams and compared to control
sites where no salmon carcasses were introduced. Biofilm and macroinvertebrate abundance
increased in natural streams where salmon carcasses were added, suggesting that marinederived nutrients were enhancing stream productivity (Wipfli et al., 2011). These findings were
consistent with results showing that the addition of inorganic nitrogen and phosphorous, and
organic matter increased biological activity throughout trophic levels (Peterson et al., 1993;
Lamberti, 1996; Stockner & MacIsaac, 1996). Bilby et al. (2011) found delta 15N values in resident
fish were elevated at treated sites, indicating that decomposing salmon were ingested by fish
4

Chapter 1: Marine-Derived Nutrients and Salmon; A Literature Review

living in these streams. In another study the introduction and presence of decomposing salmon
increased ammonium and chlorophyll-a levels as well as the abundance of macroinvertebrates
when compared to non-salmon streams (Claeson et al., 2011). A study on Quesnel Lake
phosphorous levels shows that P levels were higher closer to the mouth of a salmon spawning
river than sites located away from the river delta with higher levels occurring in the spring freshet,
summer, and fall spawn, respectively (Neil, 2006). These studies show that there are differences
in water chemistry and/or biotic populations when comparing salmon bearing and non-salmon
bearing rivers and streams (Table 1.2) and further research and analyses should take this into
consideration.
Table 1.2: Results from four studies looking at water quality and biotic differences among salmon
bearing and non-salmon bearing streams and lakes. (+) denotes there was higher levels or
populations in salmon bearing streams and lakes. (N) denotes there was no significant difference
found in salmon bearing streams and lakes. (NA) denotes that the study did not mention
parameter. Results with * were taken from same lake; however, salmon bearing was at the
outflow of a salmon spawning river and non-salmon bearing was in the middle of lake away from
any salmon spawning river mouths.
Resident
Study
Chl-a TP TN NH4 15N 13C Macroinvertebrates
Biofilm
Fish
Bilby et al.
NA
NA NA NA
+
+
+
+
NA
2011
Claeson et
al. 2011

+

NA NA

+

+

+

N

NA

NA

Wipfli et
al. 2011

NA

NA NA

NA

NA

NA

+

NA

+

Neil 2006

NA

+*

NA

NA

NA

NA

NA

NA

N*

1.4 Plankton Background
Juvenile Sockeye Salmon (Oncorhynchus nerka) make extensive use of freshwater lake
habitat as rearing areas. These fish typically spend the first one to two years of their lives living
in the limnetic zone of a lake before making their way to the ocean (Burgner, 1991; Naiman et
al., 2002). While inhabiting these freshwater systems, Sockeye Salmon are predominantly
planktivoruous and mainly feed on pelagic zooplankton (Burgner, 1991; Kyle et al., 1996;
Koenings & Kyle, 1997). Hume et al. (2005) conducted a study on Quesnel Lake, Shuswap Lake,

5

Chapter 1: Marine-Derived Nutrients and Salmon; A Literature Review

and Mara Lake in British Columbia Canada that found that Daphnia sp. made up more than fifty
percent of biomass found in the stomachs of rearing Sockeye Salmon.
Studies have been implemented to determine the relationship between MDNs and
plankton communities in oligotrophic interior British Columbian and Alaskan lakes where
productivity is very nutrient-limited. Higher trophic level productivity in lakes is assumed to be
dependent on larger populations of plankton, as larger populations of phytoplankton means
more food and faster growth-rates for Sockeye Salmon (Hyatt et al., 2011). Larger populations of
zooplankton within these nursery lakes could also promote internal recycling of nutrients
furthering the enhancement of both phytoplankton and fish productivity (Vanni, 1987).
Sweetman (2001) found that Sockeye spawner abundance was highly correlated with the
abundance of herbivorous zooplankton populations in 23 Alaskan nursery lakes. This correlation
implies that with larger populations of sockeye spawners in nursery systems zooplankton
communities will also increase because of the higher amounts of MDNs. This idea suggests that
MDNs delivered by spawning salmon could be directly related to productivity in nursery lakes.
However, Walters (2014) found that zooplankton populations do not show the same decreasing
trend as female spawners have over the past 50 years which suggests that returning salmon
numbers have no effect on plankton populations. Walter’s findings, that the amount of MDNs
being delivered to nursery systems by spawning salmon, do not have any effect on growth rates
and survivorship of zooplankton in nursery lakes, suggests that the contribution MDNs have no
to little significance in bottom up trophic transfer.

1.5 MDNs Effect on Water Chemistry
Many studies have looked at the effects of salmon decomposition and how MDNs affect
water parameters. Concentrations of ammonium and soluble reactive phosphorus (SRP) are
positively affected by the arrival of MDNs (Bilby et al., 1998; Hood et al., 2007; Wipfli et al., 2010)
due to the addition of the salmon’s waste products and their post-spawn decay (Hood et al.,
2007; McIntyre et al., 2008). This increase in ammonium may help contribute to the increased
productivity level of water bodies and could be the mechanism for more algal growth that
support plankton communities. Total phosphorous (TP) and phosphate (PO4-3) also increase with
the arrival of MDNs to spawning rivers/streams and nursery lakes (Bilby et al., 1998 & Hume et
6

Chapter 1: Marine-Derived Nutrients and Salmon; A Literature Review

al., 2005). Hume et al. 2005 supported earlier conclusions that TP measured in the spring was
highly correlated with the previous year’s spawner’s density in Quesnel and Shuswap Lakes
(Shortreed & Hume, In Prep). This may be an important relationship as this increase occurs near
the beginning of the open-water season when temperatures are increasing and chlorophyll
concentrations are peaking (Hume et al., 2005). The other form of this nutrient observed to
increase in concentration, phosphate (PO4-3), is important in these interior, oligotrophic nursery
systems because it is the most readily available form of phosphorous for uptake by primary
producers such that higher concentrations of PO4-3 generally coincide with algal blooms (Ministry
of Environment, 1988). Increased algal growth is measured as higher levels of primary
productivity which tends to result in the growth and survival rate of secondary producers.
Concentrations of nitrates (NO-3) and dissolved organic carbon (DOC) also increase with the
addition of MDNs to freshwater systems; however, studies suggest that these aquatic parameters
are not solely related to the arrival of salmon in these systems (Hood et al., 2007; Cak et al.,
2008). These findings in these two papers indicate that with the arrival of salmon, concentrations
of TP and PO4-3 increase promoting the growth of algae and play the more significant role in aiding
productivity in nursery systems.

1.6 Movement and Behavior of Marine-Derived Nutrients
The delivery of MDNs by salmon is significant to aquatic and terrestrial ecosystems
(Naiman et al., 2002). Spawning salmon generate important biophysical changes to their natal
streams by such actions as disturbing streambeds/ re-suspending sediment during redd
construction and the introduction of MDNs to the system from the decomposition of their
carcasses (DeVries, 2012). There are many variables that affect the behavior and movement of
MDNs from the location where they are introduced in aquatic systems such as the order of the
stream or river, magnitude of hydrologic discharge, numbers of spawning salmon, seasonal
climatic changes, and floodplain connectivity (Cederholm et al., 1999; Naiman et al., 2002). The
re-suspension and redistribution of MDNs and sediment by spawning salmon can alter
ecosystems at all levels as the flocculation of suspended particulates affects the downstream
flow of MDNs (Rex et al., 2008; Petticrew et al., 2011). Interior streams depend on the cycling of
these MDNs because of their distance from marine ecosystems (Johnson et al., 1997). The
7

Chapter 1: Marine-Derived Nutrients and Salmon; A Literature Review

storage (Albers & Petticrew, 2012) and movement of these MDNs may be particularly important
for juvenile salmon that rear in downstream nursery lakes for up to a year or more (Naiman et
al., 2002).
The construction of spawning redds and the decaying of salmon carcasses, which may
occur simultaneously, plays an important role in the movement and storage of nutrients in
salmon-spawning streams (Wipfli & Baxter, 2010). This combination of events provides favorable
biological conditions for flocculation (the formation of aggregates of organic and inorganic
matter) of fine sediment (>63 m) and organic matter originating from salmon (Rex & Petticrew,
2008; Petticrew et al., 2011). Larger flocs can have increased settling rates that enhance particle
delivery to gravel beds, which increases the availability of organic matter and nutrients to benthic
organisms (Wotton, 2007: Petticrew et al., 2011). This process is important to inland streams,
which experience low flow during spawning events, because it increases the potential of these
gravel-bed driven nutrients to be available for hyporheic nutrient recruitment and storage (Rex
et al., 2014). The storage of MDNs in gravel beds, hyporheic zones, and riparian areas during the
fall season’s low flow allows these nutrients to be resuspended and transported farther
downstream during spring’s higher flows (O’Keefe & Edwards, 2002; McConnachie & Petticrew,
2005; Moore et al., 2007). Therefore, seasonal storage of MDNs may be a significant factor in
primary and secondary production both locally in the stream and in downstream nursery lakes.
1.6.1 MDNs in Nursery Lakes
Sources of nutrients in many Pacific Northwest salmon-rearing lakes are limited and one
study shows that the influx of MDNs by spawning salmon represents a major nutrient source to
nursery lakes which may only happen once a year (Moore & Schindler, 2004). Other studies on
biota in these nursery ecosystems show that substantial amounts of MDNs are retained and
stored in recipient freshwater systems (Gende et al., 2002; Naiman et al., 2002). Data have
suggested that because salmon spend a large amount of time in these freshwater nursery
systems, the storage of MDNs from previous salmon runs, inter-annual flux, is important for the
productivity of future salmon populations (Finny et al., 2000; Stockner, 2003). Although there is
minimal research supporting the linkage of MDNs and future salmon populations, this
relationship has become an important concern in salmon management and recovery efforts of
8

Chapter 1: Marine-Derived Nutrients and Salmon; A Literature Review

impacted populations in the Pacific Northwest (O’Keefe & Edwards, 2002; Compton et al. 2006;
Holtgrieve & Schindler, 2011).
Physical limnology plays an important role in how nutrients move throughout the water
column in lakes (Carmack et al., 1979). The stratification of lakes restricts the exchange of
suspended particulates and nutrients among the hypolimnion, metalimnion, and epilimnion. The
solar energy entering the lake’s surface in summer months promotes stratification in most
temperate lakes, while in the fall the surface water cools decreasing the density gradient which
allows the entire water column to mix via wind (Wetzel, 2001). This mixing of water in the spring
and fall months allows nutrients to be recycled back up to the euphotic zone from the
hypolimnion promoting primary productivity (Brock et al., 2007). However, if a lake doesn’t mix
fully, nutrient supply to the epilimnion will be limited which reduces nutrient recycling and
primary productivity (Brock et al., 2007). These physical characteristics limit MDNs from moving
freely throughout the water column during times of stratification; however, allow MDNs that
may have settled to the bottom of nursery lakes to be redistributed during times of mixing.
Although, there has been no substantial evidence that supports that MDNs have a direct
effect on future salmon population numbers, the amount of MDNs should have a direct effect on
juvenile salmon rearing in nursery lakes because of the effect it has on bottom up trophic
transfer. Schindler et al. (2005) discovered that algal abundance in Alaskan nursery lakes was
positively linearly correlated with sedimentary 15N in sediment cores. This coupled with findings
that 15N is strongly correlated to the density of spawning salmon (Finney et al., 2000; Brock et al.,
2006), suggests that algal abundance increases with higher spawning salmon populations.
Zooplankton communities that are consumed and provide fuel for the growth of juvenile salmon
are dependent on algal abundance for food (Mazumder & Edmundson, 2002). Therefore, the
delivery of MDNs by spawning salmon to nursery lakes is expected to be a regulating factor in
the development of juvenile salmonids.
1.6.2 Primary Productivity & Nutrient Pulses
Spawning salmon can supply a substantial nutrient subsidy to nursery lakes as they
decompose and release MDNs into their natal spawning rivers that connect with these lakes
(Larkin & Slaney, 1997; Gresh et al., 2000). Previous studies show that nitrogen (N) from Sockeye
9

Chapter 1: Marine-Derived Nutrients and Salmon; A Literature Review

Salmon can account for a 25% increase in N to Alaskan nursery lakes (Naiman et al., 2002) and
90% of total phosphorus (P) increase to coastal freshwater lakes and rivers (Schmidt et al., 1998;
Naiman et al., 2002; Mitchell & Lamberti, 2005). Finney (1998) found that approximately a
million Sockeye Salmon spawners brought in an estimated 64,100 kg.year-1 of nitrogen to Karluk
Lake in Alaska, while only 43,200 kg.year-1 and 800 kg.year-1 of nitrogen were delivered by rivers
and rain, respectively. However, because the amount of MDNs depends on the number of
spawners present, and this varies each year relative to the background of N and P from other
non-salmon sources (Naiman, 2002), there is a significant annual and regional discrepancy
regarding the effects of MDNs on primary production in nursery lakes (Wipfli et al., 1998;
Scheuerell et al., 2005; Schindler et al. 2005).
Some studies have shown that MDNs, specifically marine-derived nitrogen (15N), from
spawning Sockeye Salmon do not get taken up quickly by the surrounding biota and instead are
circulated in a dissolved form throughout nursery lake systems before stimulating primary
production (Finney et al., 2000; Schindler et al., 2005; Brock et al., 2006). In a study done by Hume
et al. (2005), which aimed to differentiate stream versus lake stored sources of nutrients, they
found results which supported previous research (Shortreed & Hume, In Prep) that suggested
that total phosphorous levels in Quesnel Lake during the spring freshet were directly correlated
with the amount of salmon spawners the year previous. This would suggest that a significant
amount of phosphorous is stored in the river ecosystem and not delivered to this nursery lake
until the spring during high, flushing flows. This may be important because seasonal differences
in a lake’s physical structure can limit or influence where incoming nutrients will be available for
uptake by biota living in these systems.

1.7 Plume Dispersion Effects at River Lake Interface
In many nursery lake ecosystems, river and stream discharges lead to the creation of
plumes in these lakes due to differences in temperature, suspended particulate load, and/or
salinity (Grimes & Kingsford, 1996). These river plumes have different water characteristics than
the lake they enter; these differences can be chemical (nutrients), physical (suspended
particulate, temperature), and biological (organisms, biomass) and will typically be higher in
turbidity and nutrient concentrations (Morgan et al., 2005; Owens, 2005; Vanderploeg et al.,
10

Chapter 1: Marine-Derived Nutrients and Salmon; A Literature Review

2007; Reichert, 2009; Hodder, 2009). These differences depend greatly on location, topography,
land use, geomorphology, watershed characteristics, and seasonal weather patterns (Grimes &
Kingsford, 1996; Mallin et al., 2005).
River plumes will mix and disperse into lake systems differently depending on the relative
properties of both water bodies (Figure 1.2). Colder and more particulate laden river water will
plummet to deeper depths when entering a lake ecosystem with warmer surface

A

B

Figure 1.2: Depicts a river/lake interface ecosystem during situations when river
waters are colder and more sediment-laden than lake surface waters (A) and when
river waters are warmer and less sediment-laden than lake surface waters (B). Blue
arrows signify flow path. (Desloges & Gilbert, 1998; Boehrer, 2008).

waters, since water around 4C is denser than warmer water (above 4C) and colder water (below
4C) (Figure 1.2A). Warmer and less particulate laden river water will stay suspended in a lake
system when lake waters are colder (Figure 1.2B). This process will influence where a river plume
and its contents will disperse in the lakes water column. Therefore, depending on the season and
11

Chapter 1: Marine-Derived Nutrients and Salmon; A Literature Review

water temperatures of both salmon spawning rivers and nursery lakes, MDNs that are contained
in these river plumes could be readily available for plankton that live in the warmer surface
waters of a lake or could be lost to the depths and not utilized by primary producers.
Unfortunately, to date there hasn’t been published research identifying the seasonal
magnitude of MDNs entering nursery lake systems or the mechanisms that control the spatial
and temporal variability of primary and secondary productivity to nutrient pulses. To reiterate
the objective of this study is to address this lack of information by determining the seasonal
quantitative differences in the amount of MDNs flushed into a nursery lake from a river system
and the depth at which these nutrients disperse in the water column over the course of a salmon
spawning event, spring freshet, and a summer period.

12

Chapter 2: Timing and Delivery of MDNs from a Spawning Salmon River to a Nursery Lake

Chapter 2: Timing and Delivery of MDNs from a Spawning Salmon River
to a Nursery Lake - The Horsefly River/Quesnel Lake system
2.1 Introduction
Many studies have been undertaken to try to predict the relationship between MDNs and
future salmon populations (Schindler et al., 2005; Naiman et al., 2009; Holtgrieve & Schindler,
2011; DeVries, 2012). However, no real consensus has been achieved to say whether MDNs are
beneficial, neutral, or detrimental to future salmon spawners. Spawning salmon returning to
their natal streams from the ocean, are transporting nutrients against the usual downstream
energy river gradient, supplying rivers and lakes with pulses of nutrients that are essential for
many oligotrophic interior British Columbia lakes (Bilby et al., 1996). These rivers and nursery
lakes where these salmon spawn and die act like sinks as both soluble and insoluble nutrients
find their way into river beds and/or are deposited on the lake bottom. These nutrient pulses
could be important to primary and secondary production in these nursery lakes where juvenile
salmon live for the first 1-2 years of their lives.
Extensive amounts of information have been gained in studying the transportation and
storage of MDNs in salmon-spawning streams (Petticrew et al., 2011; Reisinger, 2013). However,
the timing of the delivery of these MDNs relative to seasonal plankton production remains
understudied. The objective of this chapter is to examine the timing of delivery of MDNs from a
salmon-spawning river to a nursery lake, comparing the salmon fall-spawn event versus the
subsequent spring freshet.

2.2 Research Question
When salmon migrate back to their natal rivers they spawn and die, creating an influx of
soluble nutrient and particulate pulses that can move downstream into nursery lakes.

The

magnitude and timing of the downstream delivery of MDN is the focus of this chapter.
Specifically, are there differences in the amount of sediment-associated MDNs transported from
the Horsefly River (HFR) system to Quesnel Lake during the fall spawn event versus the spring
freshet? Differences in the amount of MDNs that enter Horsefly Bay (HFB) during these two
events should identify if these nutrients are stored over winter in the spawning river or if the bulk
13

Chapter 2: Timing and Delivery of MDNs from a Spawning Salmon River to a Nursery Lake

delivery of MDNs are associated with the fall spawn. This question will be addressed by analyzing
stable isotopic ratios and nutrient levels over the course of a fall spawning event and a spring
freshet.

2.3 Methods
2.3.1 Study Site
The Horsefly watershed is located in the Cariboo region in interior British Columbia. The
HFR runs 131 kilometers from the Cariboo Mountains to Quesnel Lake where it empties into
Horsefly Bay. The HFR represents a unique ecosystem that historically supports one of the largest
Sockeye Salmon runs in the Fraser River Watershed and eventually empties into Quesnel Lake
which is one of the deepest fiord lakes in the world reaching depths of 510 meters (Stockner &
Shortreed, 1989; Hume et al., 1996).
Table 2.1: Salmon escapement from 2013-2016 from the Horsefly Watershed and the
Horsefly River (HFR) specifically.

Salmon Escapement
Year

Horsefly River

Horsefly Watershed

2013

69,937

95,784

2014
2015
2016

457,553
23,524
619

492,011
28,471
642

2.3.1.1 Site Characteristics
A hydrometric station (08KH031) which records water level data every five minutes was
installed in the HFR near the mouth (52° 26' 39'' N 121° 25' 05'' W) by the Water Survey of
Canada, Environment Canada. These data were used to obtain daily river discharge rates (m3/s).
A pressure transducer data logger (Onset Hobo - model #U20-001-04) was placed 1500 meters
upstream of the HFR delta to record water level data, as well as temperature from August 2014
through October 2014 and again from April 2015 through September 2015, to document the HFR
water levels (Figure 2.1).

14

Chapter 2: Timing and Delivery of MDNs from a Spawning Salmon River to a Nursery Lake

2.3.2 Study Design
Five sites (S1, S2, S3, S4, and S5) were monitored in HFB (Figure 2.1); four of the five were
placed near the mouth of the HFR (S2, S3, S4, S5), and the furthest site (S1) was located across
the bay to act as a reference site. Sampling was performed twice monthly from June 1, 2014
through July 31, 2014, and twice weekly from August 1st, 2014 through November 1st, 2014 to
monitor salmon pre-spawn, mid-spawn, and post-spawn activity. Sampling twice weekly
recommenced on April 15th, 2015 through August 1st, 2015 to evaluate the pre-freshet, mid
freshet, and post freshet flush.

Water Survey Station

Figure 2.1: Map displaying sampling sites in HFB.
The sampling can be categorized into 3 distinct periods: 1) the spring freshet event which
took place from the beginning of April through the end of 2014 & 2015; 2) the summer which
took place during July and August (2014 & 2015); and 3) the fall spawn which took place from the
beginning of September though the beginning of November 2014 includes both the digging of
reds and the decaying of carcasses (Figure 2.2).
15

Chapter 2: Timing and Delivery of MDNs from a Spawning Salmon River to a Nursery Lake

Spring Freshet

Summer

Fall Spawn

Spring Freshet

Summer

Figure 2.2: HFR hydrograph, data taken at the river delta every five minutes. Red lines
designate sampling periods in 2014 and 2015 (Environment Canada).
2.3.3 Sampling Techniques (Collection & Analysis)
2.3.3.1 Isotopes
Suspended particulates used for isotopic analysis were collected from three of the five
sites (S1, S3, S5) using two methods including a continuous flow centrifuge (US Centrifuge M512) (CFC) and a high-volume filtration system (HVFS). The CFC, used between August 1st, 2014
through July 15th, 2015, uses centrifugal force to trap suspended particulates contained in water
being sampled against the side of a plastic bowl which can be collected after sampling. Centrifuge
times ranged from 45 minutes to 90 minutes depending on turbidity levels in the water column.
The CFC was equipped with reinforced vinyl tubing and a submersible pump which was lowered
to a selected depth for sampling of the water column. Sampling depth during the fall-spawn
sampling period was three meters, while sampling depth during the spring-freshet sampling
period was dependent on the depth of the river plume. A handheld turbidity meter (YSI:
Nephelometric-Optical, 900 Scatter) was used to measure turbidity throughout the water column
during the spring freshet to identify where the river plume was located. The submersible pump
was then lowered to the depth with the highest turbidity reading. These collected CFC samples
were scraped from the centrifuge bowl into 5 mL polystyrene Eppendorf tubes and stored in
16

Chapter 2: Timing and Delivery of MDNs from a Spawning Salmon River to a Nursery Lake

storage box until arrival at the lab (~2-4 hours). Once at the lab, samples were stored at -20oC
until further analysis. From July 15th, 2015 until the end of the sampling season a high-volume
filtration system (HVFS) was used to collect suspended sediment. This filtration system used a
submersible pump to sample water from a selected depth into a collection chamber where water
was forced through a 5-micron mesh filter which trapped the suspended particles. Once the filter
was clogged with suspended material, the water in the chamber and particulates attached to the
filter were transferred to small buckets and left to settle for 24 hours. Supernatant was then
syphoned off and the remaining material was placed into 5 mL polystyrene Eppendorf tubes and
stored at -20oC.
Suspended particulate samples were put in a freeze drier and left for 48 hours and
transferred into clean 50 mL polystyrene Falcon tubes. A small ball bearing was used to
homogenize the samples. In between homogenizing samples, the ball bearing was washed with
tap water, acetone, and deionized water to ensure the ball bearing did not transfer sample
material from one tube to another. The homogenized samples then were prepared for delivery
to the University of California-Davis’s Isotope Lab where they were analyzed for isotope ratios of
δ14N- δ15N and δ12C- δ13C.
To calculate the load of marine-derived nutrients entering HFB from the HFR, an equation
was formulated (see Appendix 1) using the percent of MDNs relative to the total N in the samples
collected. A 15N value is needed to complete this equation.

15 𝑝𝑝𝑡

𝑁
15𝑁 % = 100 – 100 /(0.003676 (1 + ( 1000
)) + 1)

(1)

15 % × 𝑇𝑜𝑡𝑎𝑙 𝑁 𝜇𝑔

𝜇𝑔

𝑁
) × (𝑆𝑢𝑠𝑝𝑒𝑛𝑑𝑒𝑑 𝑃𝑎𝑟𝑡𝑖𝑐𝑢𝑙𝑎𝑡𝑒 𝑀𝑎𝑡𝑡𝑒𝑟 )) ×
𝐿𝑜𝑎𝑑 𝑃𝑎𝑟𝑡𝑖𝑐𝑢𝑙𝑎𝑡𝑒 15𝑁 = (( 𝑇𝑜𝑡𝑎𝑙
𝜇𝑔 𝑆𝑎𝑚𝑝𝑙𝑒
𝐿

𝐿

(2)

(𝐹𝑙𝑜𝑤 𝑅𝑎𝑡𝑒 𝑆)

2.3.3.2 Concentration of Suspended Particulate Matter
Samples were collected to determine suspended particulate concentrations along with
organic/inorganic concentrations from an influent hose that was connected to the CFC or the
17

Chapter 2: Timing and Delivery of MDNs from a Spawning Salmon River to a Nursery Lake

HVFS. These were collected in 1.5-liter polystyrene Nalgene bottles from three sites (S1, S3, S5).
Two surface grab samples were also collected at sites S2 and S4. Samples were stored at 4oC in
the dark, until filtration of the samples was initiated, which generally was 1 to 3 days.
To obtain suspended sediment and organic/inorganic concentrations, a known volume of
sample was filtered through a pre-ashed /pre-weighed 0.7m Whatman glass microfiber filter
(GF/F). These filters then were dried in a 60oC oven for 24 hours and then weighed to obtain total
particulate mass. Filters were then placed in a muffle oven at 550oC for one hour to burn off
organic matter. Filters once again were weighed to determine organic and inorganic mass. The
organic particulate mass volume, ∅ (g/L-1), the inorganic particulate volume density, 𝜒 (g/L-1), and
the total particulate mass volume 𝜓 (g/L-1) were obtained as,
1000𝑚𝐿/𝐿

∅ = (𝐷𝑟𝑖𝑒𝑑 𝐹𝑖𝑙𝑡𝑒𝑟 𝑊𝑒𝑖𝑔ℎ𝑡 − 𝐴𝑠ℎ𝑒𝑑 𝐹𝑖𝑙𝑡𝑒𝑟 𝑊𝑒𝑖𝑔ℎ𝑡) × 𝑚𝐿 𝑆𝑎𝑚𝑝𝑙𝑒 𝐹𝑖𝑙𝑡𝑒𝑟𝑒𝑑
1000𝑚𝐿/𝐿

(4)

1000𝑚𝐿/𝐿

(5)

𝜒 = (𝐴𝑠ℎ𝑒𝑑 𝐹𝑖𝑙𝑡𝑒𝑟 𝑊𝑒𝑖𝑔ℎ𝑡 − 𝑃𝑟𝑒 𝐹𝑖𝑙𝑡𝑒𝑟 𝑊𝑒𝑖𝑔ℎ𝑡) × 𝑚𝐿 𝑆𝑎𝑚𝑝𝑙𝑒 𝐹𝑖𝑙𝑡𝑒𝑟𝑒𝑑
and

(3)

𝜓 = (𝐷𝑟𝑖𝑒𝑑 𝐹𝑖𝑙𝑡𝑒𝑟 𝑊𝑒𝑖𝑔ℎ𝑡 − 𝑃𝑟𝑒 𝐹𝑖𝑙𝑡𝑒𝑟 𝑊𝑒𝑖𝑔ℎ𝑡) × 𝑚𝐿 𝑆𝑎𝑚𝑝𝑙𝑒 𝐹𝑖𝑙𝑡𝑒𝑟𝑒𝑑
respectively.
2.3.3.3 Chlorophyll-A

Water samples that were tested for chlorophyll-a levels were collected at a depth of three
meters at four sites (S1, S2, S3, S4). Site S5 was not sampled for chlorophyll-A because of higher
flows of water which would hinder phytoplankton from free floating at this site. There also would
be higher concentrations of SPM at certain parts of the year which also could interfere with the
amounts of chlorophyll-a analytical technique. A discrete depth water sampler (Van Dorn) was
used to collect all chlorophyll-a samples and these samples were placed in 1.5-liter polystyrene
Nalgene bottles. Chlorophyll-a samples were filtered through a 0.7 m Whatman glass microfiber
filter (GF/F) using a vacuum pump. After filtration took place, filters were folded, wrapped in
aluminum foil, and frozen at -20oC.
To determine chlorophyll-a levels for each sample, frozen filters were unfolded, placed
into 50 mL beakers, and fully submerged in a 90% acetone solution for 24 hours at -20oC in the
18

Chapter 2: Timing and Delivery of MDNs from a Spawning Salmon River to a Nursery Lake

dark. Samples then were placed into a centrifuge and centrifuged for 15 minutes at 3000 rpm.
Samples were transferred to a 1cm wavelength ultraviolet spectrometer (UV-VIS) cuvette using
a micro-pipet. A Biochrom Ultraspec 2100 Pro UV-VIS was used to analyze samples at five
wavelengths (750, 663, 645, 630, 664). The recorded numbers were used to calculate
uncorrected values for chlorophyll-a (Equation 6). Samples were also acidified (0.1 mL of
hydrochloric acid 1N) and run through a UV-VIS at wavelengths 750 and 665 approximately 90
seconds after the addition of the acid. These numbers were used to calculate the corrected
pheophytin value for chlorophyll-a using Strickland and Parson’s equation (1972).
𝐸𝑥𝑡𝑟𝑎𝑐𝑡 𝑉𝑜𝑙𝑢𝑚𝑒 (𝑚𝐿)

𝜇𝑔

𝐶ℎ𝑙𝑎 (10𝑚𝐿) = (11.6 × 𝑂. 𝐷.664 − 1.31 × 𝑂. 𝐷.645− 0.14 × 𝑂. 𝐷.630 )( 𝐶𝑢𝑣𝑒𝑡𝑡𝑒 𝑊𝑖𝑑𝑡ℎ (𝑐𝑚) ) (6)

2.3.3.4 Phosphorous
Water samples were collected to analyze levels of dissolved phosphorous (DP) and total
phosphorous (TP). These water samples were collected bi-weekly throughout the fall sampling
season at sites S1, S3, and S5. During the spring sampling season water samples were collected
weekly at sites S1 and S3 while water samples were collected twice weekly at S5. Two samples
of 40 mL water for each DP and TP were collected from an influent hose that was connected to
either the CFC or the HVFS. The samples for DP were filtered using a 0.45-micron filter to remove
suspended particulates. Samples were frozen and kept at -20oC until further analysis took place.
For analysis, samples were concentrated by allowing 10 mL of well mixed, representative subsample to evaporate to approximately 3 mL. Digestion acids were added to top up all samples to
5 mL. An Agilent Technologies ICP-OES (model #: 5100ICPOES) was used to read levels of both DP
and TP in each sample. To calculate for particulate phosphorus, DP was subtracted from TP.
2.3.3.5 Statistical Analysis
To assess differences among the parameters tested (SPM, organic/inorganic, isotope
enrichment, isotope load, phosphorus) a one-way ANOVA was used assuming there was
normality, equality, and independence between the samples. Although all parameters were not
normally distributed when using a Shapiro-Wilks test, an ANOVA is not very sensitive to moderate
deviations. Statistical studies using a variety of non-normal distributed data have shown that the
19

Chapter 2: Timing and Delivery of MDNs from a Spawning Salmon River to a Nursery Lake

false positive rate is not affected by this assumption (Glass et al., 1972, Harwell et al., 1992, Lix
et al., 1996). If significant differences were found through the one-way ANOVA, a Tukey’s Honest
significance difference (HSD) test was used to determine differences among sample periods. A
two-way ANOVA was used to assess chlorophyll levels using sample site and study period as fixed
effects. A post hoc test, Tukey’s HSD, was used to find where the significant differences lie in this
two-way ANOVA.
A stable isotope mixing model written in R (version 3.2.2), SIMMR (Parnell, 2016), which
is an upgrade to the stable isotope analyzes (SIAR) program in R, was used to determine what
proportions of samples collected were made up of which source material (salmon, terrestrial
vegetation, aquatic vegetation). This package solves mixing models within a Bayesian framework.
Other sources such as plankton were collected to be source material. However, since samples
collected for isotope data were only used from site S5, where the flow would be too fast for
plankton to stay suspended, only salmon, terrestrial vegetation, and aquatic vegetation were
used. Boxplots showing proportions of sources materials in samples in each study period, and a
boxplot of the proportion of salmon through the three study periods was also created within this
package.
Multiple linear models were created and selected with akaike information criterion (AIC)
to try to determine what river characteristics were driving the delivery of delta 15N values and
the load of MDNs (Table 2.2). With the delta 15N values as the variable of interest; river
temperature, SPM concentrations, phosphorous concentrations, discharge data, and study
period variables were included in the first linear model created to be tested by an AIC (1). This
linear model tested all and only first order terms. The second linear model (2) tested SPM
concentrations, discharge data, and study period parameters. Phosphorus concentrations and
river temperature were left out as they were not significant in the first linear model and logically
should have little effect on when MDNs were delivered to HFB. The third linear model (3) tested
all first and second order terms, which were included in the second linear model. The fourth
linear model (4) tested interactions of all first order and second order terms that were significant
in the summary output for the third linear model.

20

Chapter 2: Timing and Delivery of MDNs from a Spawning Salmon River to a Nursery Lake

This same method and creation of linear models (1, 2, 3, 4) was applied to the second
variable of interest, the load of MDNs, to create linear models (5, 6, 7, 8). Once again, all variables
were used for the fifth linear model tested as the sixth, seventh, and eighth linear models
included only SPM concentrations, discharge data, and study period data, as phosphorous and
river temperatures were non-significant in the fifth linear model and shouldn’t impact when the
load of MDNs is delivered to HFB.
Table 2.2: Linear models created and selected by AIC to test which parameters drive
the delivery of  15N levels and the load of 15N to HFB.
Model
Interest Variable

Variable Selection

Model #

Flow + River Temp + SPM + TP + Study Period

1

Flow + SPM + Study Period

2

Flow2 + SPM2 + Study Period2

3

 15N Levels

Flow x SPM + Flow x Study Period + SPM x Study
4
Period
Flow + River Temp + SPM + TP + Study Period

5

Flow + SPM + Study Period

6

Flow2 + SPM2 + Study Period2

7

Load of 15N

Flow x SPM + Flow x Study Period + SPM x Study
8
Period

2.4 Results
2.4.1 Suspended Particulate Characteristics
2.4.1.1 Total Particulate
The timing of suspended particulate matter (SPM) entering HFB from the HFR both in
2014 and 2015 reflects historical observations in that with higher flow comes higher
21

Chapter 2: Timing and Delivery of MDNs from a Spawning Salmon River to a Nursery Lake

concentrations of SPM (McConnachie & Petticrew, 2006). There appears to be a correlation
between discharge and SPM data shown in Figure 2.3. Kendall’s correlation equation was used
to check this relationship, yielding a correlation coefficient of 0.68.
Concentrations of SPM rise at the end of April and beginning of May 2015 to the highest
recorded concentration of SPM, 12.5 mg/L, on May 28, 2015. Concentrations then decrease and
remain between 5-10 mg/L until mid-July when concentrations drop to below 2 mg/L which
coincides with decreasing flow rates. In late September and early October of 2014, there is a
gradual increase in SPM concentrations during that year’s salmon spawning event.
It was found that there was a significant difference (p<0.000) in-between study periods
and a post-hoc test, Tukey HSD, revealed that SPM concentrations were significantly lower during
the summer (p<0.000) and salmon spawn (p<0.000) periods than during the spring freshet.
However, there was no significant difference in SPM concentrations between the summer and
the salmon spawn with a p-value of 0.78.

Summer

Fall Spawn

Spring Freshet

Summer

Figure 2.3: Suspended particulate matter concentrations and discharge over time at site
S5. Vertical lines denote the three study periods.
2.4.1.2 Organic vs. Inorganic Particulates
The dispersal of organic and inorganic suspended particulates is greatly influenced by flow
rates. Organic suspended particulate (OSP) concentrations stay below 1 mg/L in the summer
study period (Figure 2.4). During the salmon spawn and die-off, there is an increase in the amount
22

Chapter 2: Timing and Delivery of MDNs from a Spawning Salmon River to a Nursery Lake

of OSP, which more than doubles to 2.3 mg/L at the end of October 2014, but for most of this
period the concentration remains more consistent around 1 mg/L. However, the spring freshet
period has elevated concentrations of OSP with a mean of 1.45 mg/L, but with peaks above 2.1
mg/L and 2.2 mg/L in May and June, respectively. Also, at the end of June 2015 the concentration
increases to a high of 4.3 mg/L.
When analyzing OSP concentrations, a 1-way Anova was used and a p value of <0.000 was
found, which suggests that there is a significant difference in the concentrations of OSP over the
three study periods. A post-hoc test, Tukey’s HSD, showed that OSP concentrations were
significantly lower during the summer (p<0.000) and salmon spawn (p<0.01) periods than during
the spring freshet. However, there was no significant difference between the summer and
salmon spawn (p<0.23).
Inorganic particulate concentrations (ISP) were found to have a significant difference
(p<0.000) over the three study periods. A post-hoc test, Tukey’s HSD, revealed that ISP
concentrations were significantly lower during the summer (p<0.000) and salmon spawn
(p<0.000) periods than during the spring freshet. However, there was no significant difference in
ISP concentrations between the summer and salmon spawn periods with a p-value of 0.58.
Spring Freshet

Summer

Fall Spawn

Spring Freshet

Summer

Figure 2.4: Organic and inorganic suspended particulate matter concentrations
covering the three study periods at site S5.

23

Chapter 2: Timing and Delivery of MDNs from a Spawning Salmon River to a Nursery Lake

2.4.1.3 Stable Isotopes: Nitrogen and Carbon (Enrichment)
In Figure 2.5, the delta values of 15N show a steady increase starting in early September
2014 and continued to rise until the end of that study period in middle to late October 2014. The
delta values of 13C show the same increasing trend but starts later in the spawning study period.
Its lowest delta values occur mid-September right as salmon begin to spawn; however, it then
increases in late September and continues through the end of the study period in middle to late
October 2014. Both 15N and 13C values start to drop at the beginning of November 2014.
Conversely, 15N values during the spring freshet and summer stay consistent at approximately
2 ppt, which is considerably less than values observed during the fall salmon spawning period.
13C during the spring freshet and summer stays consistent between -26 ppt and -28 ppt.
Summer

Fall Spawn

Spring Freshet

Summer

Figure 2.5: The concentration of 15N and 13C in SPM over three study periods collected
at site S5. Method used to create blue lines was LOESS.
The delta values associated with nitrogen were analyzed using a one-way ANOVA in which
a p-value of 6.08e-11 was found. This number shows that there was a significant difference in
delta values between study periods. A post-hoc test (Tukey’s HSD) found that nitrogen delta
values were significantly lower during the summer (p<0.000) and spring freshet (p<0.000) periods
than the salmon fall spawn. However, there was no significant difference in nitrogen delta values
between the summer and spring freshet with a p-value of 0.99. The delta values associated with
carbon between study periods were also analyzed using a one-way ANOVA; however, the p-value
(0.125) indicates that there was no significant difference between periods.
2.4.1.4 Stable Isotopes: Nitrogen and Carbon (Mixing Model)
The three source materials (aquatic vegetation, terrestrial vegetation, and salmon) are
shown in Figure 2.6 along with the isotopic values for the SPM collected seasonally. The results
24

Chap
t
e
r2
:T
im
in
gandD
e
l
i
v
e
r
yo
f MDNsf
romaSpawn
in
gSa
lmonR
i
v
e
rt
oaNu
r
s
e
r
yLa
k
e

showth
a
tth
esumm
e
randf
re
sh
e
tSPMs
amp
l
e
sw
e
re mo
reh
e
a
v
i
l
ycompo
sedo
faqu
a
t
i
c
ve
ge
t
a
t
ionandt
e
r
re
s
t
r
i
a
lve
ge
t
a
t
ionth
ano
fs
a
lmon
.How
e
ve
r
,f
a
l
lsp
awnSPMs
amp
l
e
sshowa
15Nv
t
rendo
fin
c
re
a
s
in
g 15Nv
a
lu
e
s wh
i
che
x
c
e
edthe 
a
lu
e
so
fbo
tht
yp
e
so
fve
ge
t
at
ion
,

ind
i
c
a
t
in
gth
a
tth
e
ycon
t
a
in
edmo
resou
r
c
em
a
t
e
r
i
a
lf
roms
a
lmontha
nth
eo
th
e
rtwos
amp
l
in
g
p
e
r
iod
s
.Th
e
sef
ind
in
g
ssuppo
r
tse
c
t
ion2
.4
.1
.3
.Summ
e
randf
re
sh
e
tSPMa
l
soa
recompo
sedo
f
mo
ret
e
r
re
s
t
r
ia
lve
g
e
t
a
t
ionth
a
naqu
a
t
i
cve
ge
t
a
t
ion
.How
e
ve
r
,th
ise
ven
sou
tdu
r
in
gth
es
a
lmon
sp
awnp
e
r
iod.
Fo
cu
s
in
go
nth
esp
r
in
gf
re
sh
e
t
(
Fi
gu
re2
.
7)
,sou
r
c
em
a
t
e
r
i
a
lth
a
tw
a
scon
t
a
in
ed
inth
eSPM
s
amp
l
e
sf
roms
a
lmonw
e
relowa
tapp
ro
x
im
a
t
e
l
yfou
rp
e
r
c
en
t
,wh
i
l
esou
r
c
em
a
t
e
r
i
a
lth
a
tw
a
s
con
t
a
in
ed
inth
eSPMs
amp
l
e
sf
romaqu
a
t
i
candt
e
r
re
s
t
r
i
a
lve
ge
t
a
t
ionw
e
reapp
ro
x
im
a
t
e
l
y37%
and59%re
sp
e
c
t
i
ve
l
y
.Th
esumm
e
rp
e
r
iod(
F
i
gu
re2
.
7
)show
ss
im
i
l
a
rre
su
l
t
stowh
a
tw
a
sfound
du
r
in
gth
esp
r
in
gf
re
sh
e
tp
e
r
iod.SPMs
amp
l
e
sco
l
l
e
c
t
eddu
r
in
gth
i
ss
tud
yp
e
r
iodw
e
recompo
sed
o
fapp
ro
x
im
a
t
e
l
y4%s
a
lmon
,37%aqu
a
t
i
cve
ge
t
a
t
ion
,and59%t
e
r
re
s
t
r
i
a
lve
ge
t
a
t
ion
.A
l
thou
gh
,
th
e
se p
e
r
c
en
t
a
ge
s a
re th
e s
am
e a
s th
e sp
r
in
g f
r
e
sh
e
t p
e
r
iod, th
e in
t
e
r
12
.5

10
.0

)
d15N(‰

Summ
rS
PM
SPM
1
grpe
SPM
r
p
2S
g
F
r
e
s
h
e
t
PM

7
.5

SPM
p
gr
Sp
a
w
n3
SPM
Aq
u
c
V
.
eg
a
t
i
A
q
u
a
t
i
c
V
e
g
.

5
.0

Te
r
s
r
a
Ve
g
i
l
t
.
r
e
T
e
r
r
e
s
t
r
i
a
l
V
e
g
.
Sal
mon

S
a
lmon

2
.5

0
.0

-3
0

-2
7

-2
4

-2
1

13
d
C(‰
)

15N
13
F
i
gu
re2
.6
:S
IMMR
m
i
x
in
g mod
e
l
,show
in
gth
e
Cv
a
lu
e
so
fSPMs
amp
l
e
s
co
l
l
e
c
t
edth
rou
gh
ou
tth
ef
re
sh
e
t
,summ
e
r
,andsp
awn
in
gs
amp
l
ep
e
r
iod
sa
ts
i
t
eS5
.

qu
a
r
t
i
l
er
an
ge
i
ssm
a
l
l
e
rdu
r
in
gthesp
r
in
gf
re
sh
e
tth
andu
r
in
gth
esumm
e
rs
amp
l
in
gp
e
r
iod
.Th
e
s
a
lmonsp
awn
in
ge
ven
tshow
sth
eh
i
gh
e
s
tp
ropo
r
t
iono
fs
a
lmondu
r
in
gth
eth
re
es
amp
l
in
g
p
e
r
iod
s
,w
i
thape
r
c
en
t
a
gea
round37(
Fi
gu
re2
.
7)
.Th
i
ss
amp
l
in
gp
e
r
ioda
l
soshow
sth
a
tSPM
25

Chapter 2: Timing and Delivery of MDNs from a Spawning Salmon River to a Nursery Lake

samples were composed of more salmon material than aquatic (~32%) or terrestrial vegetation
(~31). The proportion of terrestrial vegetation in the SPM samples dropped from approximately
59% to 31% during the salmon spawn. Comparing the three study periods, the SPM samples
collected are significantly comprised of more salmon during the salmon spawn period than the
other two study periods. SPM samples were made up of approximately 40% salmon source during
the salmon spawn, 3% during the summer, and 4% during the spring freshet.

Figure 2.7: Proportion of three sources (aquatic vegetation, terrestrial vegetation,
and salmon) found in SPM samples taken from site S5 during the spring freshet,
summer, and fall spawn. The line in the middle of the boxes for each source signifies
the median point of the data. Boxes above and below the median line represent the
upper and lower quartile. At the end of the solid vertical lines are the highest and
lowest data values excluding outliers which are the black dots in this plot.
2.4.1.4 Stable Isotopes: Nitrogen and Carbon (Specific Load)
The specific load of marine-derived nitrogen (15N) (mg/s) is much higher during the month
of May 2015 through the middle of June with the highest amount of 15N collected, 8.8 mg/s, on
May 15, 2015 (Figure 2.8(C)). During July 2014, the specific load of 15N drops and stays below 2.5
mg/s until a slight rise to approximately 3.75 mg/s at the end of the fall spawn study period.
The specific load of 15N was found to be significantly different (p<0.0001) over the three
study periods. A post-hoc test, Tukey’s HSD, revealed that 15N loads were significantly lower
during the summer (p<0.001) and salmon spawn (p<0.001) periods than during the spring
freshet. However, there was no significant observed difference in 15N loading between the
26

Chapter 2: Timing and Delivery of MDNs from a Spawning Salmon River to a Nursery Lake

summer and salmon spawn periods with a p-value of 0.78. When observing Figures 2.3 and 2.8
(C), the specific loading of 15N appears to be dependent on the flow rate of the HFR, as flow

A

B

C

D

E

Figure 2.8: Chlorophyll-a, delta 15N, 15N loading, TP, and SPM data levels collected in HFB
over the three study periods. All data used in this plot, except chlorophyll-a, was
collected from site S5. Chlorophyll-a data used was collected from S1.
peaks during the spring freshet and falls during the summer, as do the amount of 15N loads. The
discharge of the HFR greatly affects the load of 15N, as concentrations of marine-derived N and 
15N values are highest in the fall spawn when flows are low; however, the load of marine-derived

N is highest in the spring freshet when discharge is high.

27

Chapter 2: Timing and Delivery of MDNs from a Spawning Salmon River to a Nursery Lake

2.4.2 Water Analysis
2.4.2.1 Chlorophyll-a & Fluorescence
Figure 2.9, illustrates the changes in chlorophyll-a levels throughout the three study
periods. Chlorophyll-a increases at the beginning of August 2014 at sites S2, S3, and S4 but
decreases at site S1. All four sites however see a further increase in chlorophyll-a by the middle
of October. Considering 2015, there is an increase in chlorophyll-a at sites S1 and S3 but a
decrease at sites S2 and S4. Three sites (S2, S3, S4) also see an increase in chlorophyll-a during
the middle of June, however, all sites see a drop in the beginning of July.
The interaction between the variables study period and study site was not significant for
chlorophyll-a through the three study periods (Table 2.3). There also was no significant difference
among study sites. However, there was a significant difference between study periods when
assessing chlorophyll-a levels (p-value 0.004). Using a post hoc test, Tukey’s HSD,

Spring Freshet

Summer

Fall Spawn

Spring Freshet

Summer

Figure 2.9: Chlorophyll-A levels at four sites throughout HFB over the three study
periods. Lines were created using LOESS.
it was found that there was a significant difference between the summer and the spring freshet
(p-value 0.025) periods as well as significant differences between summer and fall spawn periods
(p-value 0.005). However, there was no significant difference found between the fall spawn and
the spring freshet (p-value 0.85).

28

Chapter 2: Timing and Delivery of MDNs from a Spawning Salmon River to a Nursery Lake

Fluorescence levels peak in both the 2014 fall spawn and 2015 spring freshet at sites S1,
S3, and S5 (Figure 2.10). The biggest peaks at S1 and S3 occur in the 2014 fall spawn while the
highest peak at S5 occurs during the 2015 spring freshet. During both summer periods,
fluorescence shows to be consistent between 0.5-1 RFU. There were significant differences at
site S5 (p-value 0.00736) when comparing study periods.
Summer

Fall Spawn

Spring Freshet

Summer

Figure 2.10: Fluorescence levels at three sites throughout HFB over the three study periods.
2.4.2.2 Phosphorus
At the start of the first study period in May 2014, total phosphorus levels were up to 0.026
mg/L while dissolved phosphorus levels were approximately half that value at 0.014 mg/L (Figure
2.11). Both total and dissolved phosphorus concentrations drop to approximately equal values
(shown by the trend line) in the middle of July, begin to increase in the beginning of September,
and peak at the beginning of October, with TP reaching 0.034 mg/L. After this peak both forms
of phosphorus decrease and continue to decease throughout the rest of the sampling period. In
2015, there is a slight increase in dissolved phosphorus throughout the spring freshet, followed
by a decrease as the summer period begins. Total phosphorus also increases during the spring
freshet but has a much steeper rise at the end of May. When comparing phosphorus between
the dissolved and particulate form, the data shows that throughout all study periods dissolved
phosphorus was in higher concentrations than particulate bound phosphorus except on three

29

Chapter 2: Timing and Delivery of MDNs from a Spawning Salmon River to a Nursery Lake

sampling

dates:
Spring Freshet

May

14,

2014,

Summer

April
Fall Spawn

21,

2014,

and

Spring Freshet

June

29,

2015.

Summer

Figure 2.11: Phosphorus levels in three forms (dissolved, particulate, total) at site S5
over the three study periods. Trend lines were made using LOESS.
The interaction between study period and phosphorus form was not significant through
the three study periods (Table 2.3). However, both study period and form were significant on
their own. A post hoc, Tukey’s HSD, test showed that there was a significant difference between
the fall spawn and spring freshet (p-value 0.002) and the summer (p-value 0.0008). There was,
however, no significant difference in phosphorus levels between the summer and spring freshet
periods (p-value 0.789). There was a significant difference between the total form of phosphorus
and the dissolved form with a p-value of 8.36e-05.
2.4.2.3 Comparison of Variables / Linear Models
In Figure 2.8, five characteristics of HFB were plotted above one another to visually assess
trends over the two sampling years. Except for the specific load of 15N, Chlorophyll-a, delta 15N,
TP, and SPM all increase and peak at the beginning of October 2014. During April, May, and June
of the 2015 sampling season, there is a similar visual trend in SPM, load of 15N, and TP with the
highest peaks during that study period.

30

Chapter 2: Timing and Delivery of MDNs from a Spawning Salmon River to a Nursery Lake

Table 2.3: Results from a two-way ANOVA of spatial (study site) and temporal (study period)
for chlorophyll-a levels in HFB, from a one-way ANOVA of temporal (study period) for
phosphorus levels and forms in HFB, and a one-way ANOVA of temporal (study period) for
fluorescence levels at site S1, S3, and S5 in HFB. The phosphorus form interaction term
includes dissolved and total phosphorus.
Chlorophyll-a
Source of Variance
Study Period
Study Site
Study Period X Study Site
Residuals

DF
2
3
6
138

Sum Sq.
4.09
1.04
3.34
50.56

Mean Sq.
2.0467
0.3471
0.5571
0.3664

Pr(>F)
.00465
0.41973
0.17580

Source of Variance
Study Period
Form
Study Period X Form

DF
2
1
2

Sum Sq.
0.0004713
0.0004557
0.0000214

Mean Sq.
0.0002357
0.0004557
0.0000214

Pr(>F)
0.000318
8.14e-05
0.450113

Source of Variance
Study Period
Residuals
Study Period
Residuals
Study Period
Residuals

DF
2
34
2
36
2
45

Sum Sq.
0.2666
1.5120
0.595
3.420
3.108
12.741

Mean Sq.
0.13331
0.04447
0.29741
0.09499
1.5541
0.2831

Pr(>F)
0.0632

Phosphorous

Fluorescence
S1
S3
S5

0.0558
0.00736

Using an AIC, linear models were tested to determine which linear model best predicted
higher delta 15N values. It was found that the model with the lowest AIC value was the linear
model (Model 1) that looked at only first order terms (Table 2.4). Table 2.5, shows a summary of
the linear model that had the lowest AIC value and suggests that the study period factor plays
the most important role in finding what drives higher delta 15N levels with a P-value of 0.0009.
Linear models were also used to determine the factors that drive the load of MDNs in HFB (Table

31

Chapter 2: Timing and Delivery of MDNs from a Spawning Salmon River to a Nursery Lake

2.4). However, three models that were tested produced AIC values that were within two AIC
units. Summaries of these models also revealed little interaction between factors.
Table 2.4: Statistics results performed by an AIC on which parameters drive the delivery of 15N
levels and the load of 15N to HFB.
Interest Variable Model #
K- # Parameters
AIC
R2
1

6

57.99872

0.8473

2

4

71.75135

0.7785

3

4

77.09301

0.7851

4

4

77.09301

0.7851

5

6

10.95368

0.7804

6

4

6.34016

0.7343

7

4

7.08423

0.7907

8

4

7.08423

0.7907

 15N Levels

Load of 15N

Table 2.5: Results from a linear model function showing interactions between factors
and 15N values in HFB.
Source of Variance

Std. Error

P Value

Flow

0.02014

0.426639

River Temperature

0.07609

0.859301

Suspended Particulate Matter 0.18102

0.512272

Total Phosphorus

67.03843

0.314964

Study Period

0.52869

0.000922

32

Chapter 2: Timing and Delivery of MDNs from a Spawning Salmon River to a Nursery Lake

2.5 Discussion
The productivity of an aquatic system is influenced by nutrient availability and quantity
as well as the hydrological and physical characteristics of the system (Bilby et al., 1996; Larkin &
Slaney, 1997; McConnachie & Petticrew, 2006). All samples that were analyzed for MDNs were
collected from the HFR’s plume to ensure the ability to compare three distinct study periods and
to understand the transport of sediment-associated MDNs to a nursery lake system in HFB. The
spatial abundance and delivery patterns of MDNs from the HFR to HFB were mostly impacted by
the presence, disturbance, and decay of spawning salmon (salmon disturbance regime; Albers,
2010) and the HFR’s discharge. These two processes dominate the productivity of HFB and the
timing of when MDNs are available to the biota in the bay. This conclusion is supported by
increasing trends that are shown in chlorophyll-a, isotope enrichment / load, phosphorus levels,
OSP, ISP, and SPM during the spring freshet and fall spawn study periods. Although some results
(chlorophyll-a, SPM, OSP, ISP) were consistent with previous studies (McConnachie & Petticrew,
2006; Hume et al., 2005), there were other parameters (phosphorus, chlorophyll-a) that were
not (Hume et al., 2005). These differences could be due to Quesnel Lake’s unique structure being
one of the deepest and largest lakes in North America and therefore system-specific.
To assess the timing of MDNs delivery to HFB, the present research was split into three
study periods (spring freshet, summer, and fall spawn). The spring freshet study period was
meant to capture the early spring flush of snow melt at low elevation and the later high elevation
melt. The channel scouring flows can deliver material stored over winter and determining how
these conditions affected the source and amount of nutrients entering HFB was an objective of
this research. The summer study period was intended to determine the amount and source of
nutrients entering the bay during low flow periods before the salmon arrived to spawn. The fall
spawn study period was designed to evaluate the movement of nutrients into the nursery system
while salmon were present in their natal stream, creating redds, spawning, and subsequently
decomposing. The spring freshet was characterized by higher flow volumes, higher SPM/ISP
concentrations, high 15N load, higher chlorophyll-a levels, and higher phosphorus levels. The
summer was shown to have the lowest flows which coincided with lower values of phosphorous,
chlorophyll-a, SPM, and 15N load. The fall spawn exhibited higher SPM concentrations, high
33

Chapter 2: Timing and Delivery of MDNs from a Spawning Salmon River to a Nursery Lake

isotope enrichment of 15N and 13C, a higher proportion of salmon particulate, and high
chlorophyll-a / phosphorus levels.
2.5.1 Suspended Particulate Characteristics / Load
The high flows that accompany the spring freshet are well documented as increasing SPM
concentrations and for flushing inorganic and organic material to downstream lake systems
(Owens, 2005; McConnachie & Petticrew, 2006). The present study supports these findings.
During the summer and fall spawning periods, the HFR’s flow ranged between five and fortyeight cubic meters per second and concentrations of SPM ranged from 1 to 6 milligrams per liter
(Figure 2.3). As the flow increased during the spring freshet, from forty to ninety-five cubic
meters per second, particles that had been stored in the river bed and banks were re-suspended
and flushed down stream creating SPM concentrations between 2.5 and 12.5 milligrams per liter.
Highest concentrations were recorded during peak flow and then started to decline as the river
flow rate plateaued and subsequently decreased. The flow reached and maintained its peak rate
from the middle of June through the beginning of July; however, there was an observed drop in
SPM concentration during this plateau period. Although, the flow plateaued and remained
relatively high, it is presumed that the supply of available SPM would have been flushed by the
flows in the rising limb of the hydrograph such that the material had already been transferred
down river, leaving heavier particulate, too heavy to erode, behind in the river bed and banks.
The data shown in Figure 2.3 also show a strong relationship between the concentration of SPM
and the river’s discharge rate with a Kendall’s correlation of 0.68. This suggests that there is a
relationship between flow and SPM concentration; however, there is some discrepancy between
peaks which could be due to sediment/nutrient source exhaustion.
The trend in ISP is consistent with what was observed in the SPM concentrations (Figure
2.4). ISP concentrations were around one milligram per liter at the beginning of first sampling
period in 2014. This concentration rose as salmon started arriving and began constructing redds
in the HFR, and peaked in middle to late October which coincided with fall storms and an increase
in river discharge (Figure 2.2). The observed OSP mirrors this trend, but on a smaller scale. ISP
and OSP started the spring freshet 2015 sampling period at around 6.8 milligrams per liter and
1.3 milligrams per liter respectively. The OSP concentrations were as high in the spring freshet as
34

Chapter 2: Timing and Delivery of MDNs from a Spawning Salmon River to a Nursery Lake

they were during the fall spawning period. This suggests that salmon organic material could have
been bound to inorganic particles creating flocs (Petticrew & Rex, 2008) that entered HFB in the
fall increasing OSP concentrations. On the other hand, the observed increased OSP
concentrations in the spring could be a result of higher river discharge flushing and overbank
flows contributing riparian vegetation into HFB as evidenced by the results of the mixing model.
The use of 15N and 13C isotopes to trace MDNs elucidates food web processes at the
aquatic and terrestrial levels (Rinella, 2010). This method was also used in this research to
determine the enrichment levels of SPM and the amount of MDNs that entered HFB. In Figure
2.5, the delta values of marine-derived nitrogen (15N) and carbon (13C) are approximately 1 parts
per thousand and -28.5 parts per thousand respectively at the mouth of the HFR. The delta values
of 15N and 13C stay relatively consistent until the salmon spawning event when delta levels
increase through the end of the study period. This increase could only be contributed to arrival
of the salmon as 15N signals are marine-derived. At the start of the 2015 spring freshet, the delta
values of 15N and 13C are approximately 2 parts per thousand and -27 parts per thousand. These
levels are consistent through the spring freshet and summer. These findings suggest that SPM
that is enriched in 15N and 13C is delivered to HFB only during the fall salmon spawn and may not
be available to be re-suspended during the following spring freshet. However, due to climate
change, British Columbian winters are becoming warmer and causing melts to happen earlier in
the year (Schnorbus et al., 2012). In 2015, there was a significant low-elevation snowmelt event
during the middle of January that may have caused MDNs to be re-suspended in the HFR and
flushed downstream into HFB (Figure 2.2). No data were collected at this time due to lack of
access and therefore there is no way to confirm if MDNs were delivered into HFB at this time.
The MDNs that may have been flushed with this abnormal snowmelt in January would otherwise
have been flushed with the spring melt at the end of April. Therefore, the idea that the highest
enrichment levels of 15N and 13C only happen during the fall salmon spawn cannot be assumed
until this can be tested in future years.
Although, delta values of 15N were highest during the fall spawn event, the load of 15N
was significantly higher during the spring freshet. This result was heavily influenced by the
discharge rate of the HFR as the average flow of water was 30 cubic meters per second during
35

Chapter 2: Timing and Delivery of MDNs from a Spawning Salmon River to a Nursery Lake

the fall spawn and tripled to approximately 90 cubic meters per second during the spring freshet.
Consequently, SPM coming into HFB during the fall spawn contains a greater concentration of
MDNs which could allow easier uptake of these nutrients as flow levels are low and plankton are
able to stay suspended in these nutrient rich waters. However, a larger load of MDNs is delivered
to HFB during the spring freshet but is more diluted with other suspended particulates. During
this time, the water discharge rate is high, possibly making it difficult for plankton to take up
these nutrients in HFB as it is more difficult for these organisms to stay suspended in fastermoving waters. These nutrients would eventually spread out or settle out of the faster moving
waters for uptake, delivering MDN to surface waters outside of the plume and/or deeper depths
farther out into Quesnel Lake.
Neil (2006) found that phosphorous levels were higher in HFB than levels in Quesnel Lake
outside of HFB throughout the year. As the HFR empties into Quesnel Lake at HFB, these levels
of phosphorus can be directly attributed to the HFR. Although, it cannot be directly related to the
presence of salmon, this delivery of phosphorous is still important. His research also suggests
that the spring melt produces the highest concentration of TP and the fall spawn produces the
least. The findings in this research contradict Neil’s, as TP concentration was highest during the
fall spawning event. This could be due to differences in sampling site, as Neil sampled out in the
middle of HFB and this research was conducted at the mouth of the HFR. However, the increase
in phosphorous during the fall is associated with the arrival of salmon, as the increase of TP occurs
when discharge is at low flow.
Most phosphorus input to HFB was in dissolved form occurring during the fall. Particulate
phosphorus, like the dissolved form, also had a smaller peak during the spring freshet, suggesting
that particulate phosphorus was stored in the HFR during winter months. Only four samples
collected throughout the study exhibited more particulate than dissolved P. Three of these
samples were collected during the spring freshet and Figure 2.11 shows dissolved and particulate
phosphorus levels are much closer during this study period. This is likely due to the ability to bind
to a higher concentration of suspended particulates as the increased discharge from the HFR
contains more suspended particulates. Given that dissolved phosphorus is more readily available
for uptake by primary producers than particulate bound phosphorus (Bricker et al., 2007),
36

Chapter 2: Timing and Delivery of MDNs from a Spawning Salmon River to a Nursery Lake

biological uptake of this nutrient in HFB is enhanced during the fall spawn event when the higher
proportion of phosphorus is in the dissolved form.
2.5.2 Isotope Mixing Model
The SIMMR mixing model in R used three source materials to represent the dominant
supply of isotopic C and N in the system. Salmon were chosen because the nutrients they
contribute to this nursery system are the focus of this research. Terrestrial vegetation and aquatic
vegetation were also chosen as they are at the base of the food web and are composed of
different levels of 13C which allows the model to differentiate sample sources more efficiently.
Plankton, from the open water areas of the bay were also collected and would be a good source
material for a model in HFB. However, as the SPM mixturesamples were collected at site S5,
which is situated at the mouth of the HFR, the open water algal source materials were not used
in this model. The current speeds varied significantly at S5, where the sampling focused on river
effluent such that pelagic plankton would not be represented in these samples. It would also be
difficult to characterize the river plankton from these samples for use as a potential source
material as the separation of plankton and suspended sediment would be difficult.
Figure 2.6 shows that SPM collected during the summer and spring freshet were
dominantly comprised of aquatic and terrestrial vegetation with the majority being of terrestrial
origin. The fall spawn period SPM had an increased level of 15N, signifying that it is made up of
more salmon source than during the other two periods. This supports the previous assumption
(Section 2.51), that MDNs were entering HFB in a higher concentration during the fall spawning
event than in the other two study periods. In Figure 2.7, the median terrestrial vegetation
contribution is 58% of the SPM samples, and aquatic vegetation and salmon contribute 38% and
4% respectively. During the spring freshet study period, flow rates and water levels were at their
highest, scouring river banks and flushing large amounts of detrital riparian vegetation from both
the floodplain and gravel bed storage down river. Terrestrial plant detritus was also the
dominant source in the summer period suggesting that the supply continued, it is not clear where
it was coming from but it exceeded aquatic vegetation by approximately 15%. During the fall
spawning event, salmon were the main contributors to the makeup of SPM contributing
approximately 37% while terrestrial and aquatic vegetation contributed 31% and 32%
37

Chapter 2: Timing and Delivery of MDNs from a Spawning Salmon River to a Nursery Lake

respectively (Figure 2.7). The various instream processes associated with the spawn are the key
contributors to this change of SPM make up. With their construction of redds, the salmon are
disturbing the gravel beds and removing biofilms on the surface rocks which is reflected in the
increase in the proportion of aquatic plant signal in the fall spawn SPM samples. SPM samples
that were collected in the fall spawn are also made up of approximately nine times as much
salmon material as they were in spring freshet and summer sampling periods. This result supports
the hypothesis that most salmon-derived 15N (MDNs) is being delivered to HFB during the fall
spawn sampling period.
2.5.3 Chlorophyll-a Concentrations & Fluorescence
Chlorophyll-a is used widely in ecological studies to show productivity levels in aquatic
systems throughout the Pacific Northwest (Hume et al., 2005; Schindler et al., 2005). Most
studies aim to understand the effect MDNs have on productivity and most conclude the nursery
lake’s productivity levels are reliant on spawner numbers (Uchiyama et al., 2008; Larkin & Slaney,
1997). Chlorophyll-a levels throughout the three study periods varied significantly by study
season within HFB (p-value<0.0047), but statistical analysis showed no significant differences
among sites (p-value<0.42). Claeson et al. (2011) found there to be higher chlorophyll-a levels in
salmon bearing systems than in non-salmon bearing systems during the fall spawn. Findings from
this study support this research, as peaks of chlorophyll-a are present across all sites during the
fall spawn, which could be directly attributed to the higher concentrations of MDNs and
phosphorus that were being delivered into HFB (Figures 2.9). This implies that MDNs entering
HFB during the fall spawn directly increased phytoplankton populations throughout the bay.
Smaller peaks of chlorophyll-a were found throughout the spring freshet period, which again
could be attributed to large quantities of MDNs entering the bay during this period. Hume et al.
(2005) found similar results as they saw chlorophyll-a increases during June at the peak of the
freshet. However, they also found no temporal trends in chlorophyll-a levels which contradicts
the results of this research. In Figure 2.8, there isn’t a definitive trend at sites S2, S3, and S4. Site
S1 has a wider variation in chlorophyll-a which suggests that location is critical when sampling for
chlorophyll-a levels. Sites S2, S3, and S4 are situated in the flow path of the HFR while S1 is not.

38

Chapter 2: Timing and Delivery of MDNs from a Spawning Salmon River to a Nursery Lake

Therefore, different results were found depending on location which could explain why Hume et
al. (2005) came to different conclusions.
Fluorescence differed significantly when comparing study periods at site S5; however
there were no significant differences among sites. In Figure 2.10, peaks are shown to occur at
sites S1, S3, and S5 at the end of the spawning study period. These increases could be directly
related to the presence of MDN’s that are being delivered to HFB from salmon carcasses up the
HFR. These peaks also correlate with peaks found in chlorophyll-a levels which suggests higher
productivity in HFB during this study period. There are also similar trends between chlorophyll-a
and fluorescence at the end of the 2015 spring freshet. This may be due to nutrients and
particulates being flushed into HFB from the HFR during higher discharge.
2.5.4 Parameter Comparisons
Figure 2.8 allows for the visual comparison among variables measured in HFB throughout
the study periods. Individual parameter observations can be found in their respective sections
throughout this chapter. However, Figure 2.8 suggests that there may be correlations between
chlorophyll-a values, delta 15N levels, and TP during the fall spawn. The increase and correlations
among these three variables is likely due to the arrival, spawning, and decomposition of salmon
during this period. During the spring freshet, variables that show a common upward trend include
MDN load, SPM, and TP. With higher discharge comes higher amount of SPM and nutrients that
would have been stored in the river bed over the winter period.
AIC values were calculated for linear models created to identify what parameters are
driving higher delta 15N values and the load of MDNs in HFB. The best model created in assessing
delta 15N values, based on an AIC, shows that the most important parameter was study period
(Table 2.5). This result indicates that the levels of delta 15N values are dependent on the study
period and the fall spawning period was found to contain the highest delta 15N values which
coincides with the arrival, spawning, and decomposition of salmon. Therefore, it can be assumed
that delta 15N concentrations entering HFB will always be highest during the fall spawning period.
The method of comparing models was also used when determining what parameters drive the
load of MDNs. However, as the three linear models that were created only varied in AIC values
by less than two units, it is concluded that there are no differences in the models that were
39

Chapter 2: Timing and Delivery of MDNs from a Spawning Salmon River to a Nursery Lake

produced (Stoica & Selen, 2004). This indicates that the data is insufficient to support selecting a
single model.

2.6 Implications
The predominant delivery of MDNs to HFB took place during the spring freshet and the
fall spawning periods. The pulses of MDNs in each study period possessed different
characteristics which affected both the availability and utilization of these nutrients. The spring
freshet was found to deliver the larger load of MDNs but in a more diluted and heavily sedimentladen flush. With more particulates available to bind to and higher concentrations of OSP,
phosphorus levels in particulate form increased and in a few sampling days were higher than the
dissolved form. With the increase in particulate phosphorus, the amount of phosphorus readily
available for uptake by organisms in HFB could have been more limited than if it had been in a
dissolved state. This suggests that MDNs entering HFB during the spring freshet are flushed and
carried out into deeper parts of Quesnel Lake.
The fall spawning study period was found to provide a more concentrated / enriched
delivery of MDN and a higher amount of dissolved phosphorus to HFB. Chlorophyll-a levels were
found to increase during this flush of MDNs and dissolved phosphorus. Hume et al. (2005) found
that Daphnia populations within Quesnel Lake peaked during late September and early October,
coinciding with this flush of MDNs and subsequent increase of chlorophyll-a. This suggests that
Daphnia populations, which make up 99% of juvenile salmonids diet (Hume et al., 2005) in
Quesnel Lake are utilizing the fall flush of MDNs which results in higher chlorophyll-a levels. These
findings support the hypothesis that future populations of salmonids are dependent on the fall
flush and delivery of MDNs to nursery lake systems.

40

Chapter 3: The Movement and Dispersal of a River Plume and its Contents in a Nursery Lake

Chapter 3: The Movement and Dispersal of a River Plume and its
Contents in a Nursery Lake – Quesnel Lake
3.1 Introduction
The delivery of suspended particulates from streams and rivers into downstream lake
systems is an important ecological process which aids in organic matter cycling and the delivery
of nutrients. These particulates containing inorganic and organic sediments as well as marinederived nutrients (MDNs) and possible contaminants become suspended in rivers and streams
and make their way to receiving lake systems where they eventually settle out (Owens, 2005).
Being able to track and determine where suspended particulates travel and settle is important
for the management of waterways as it influences many ecological processes. Larger or heavier
particles settle out of the water column faster near river mouths, while finer or less dense
particles stay suspended longer, which allows them to move out into deeper areas of lakes. Finer
particles have the ability to adsorb nutrients and contaminants due to their larger surface areas
(Horowitz & Elrick, 1988). These finer inorganic particles can combine with MDN through the
process of flocculation and in turn modify the transportation of both materials through aquatic
ecosystems (Choles, 2004; McConnachie & Petticrew, 2006; Hodder, 2009). As well, the
displacement and suspension of particulate matter by spawning salmon, in rivers and streams, is
an important process that supplies nursery systems with pulses of sediment (organic/inorganic
material), and nutrients (Petticrew et al., 2011; Reisinger, 2013). These processes, in interior
oligotrophic British Columbian nursery lakes, is thought to be especially important for
maintenance of salmon spawning and rearing habitats (Petticrew et al., 2011).
The objective of this study is to assess seasonal changes in the movement, composition,
and distribution of suspended particulates and their associated MDNs emptying into Horsefly Bay
(HFB) from the Horsefly River (HFR) on Quesnel Lake in British Columbia. In excess of 457,000
spawning salmon returned to the Quesnel system in the fall of 2014, compared to 69,000 the
year before. Sampling in the fall 2014 spawn and die-off period during this peak year of Sockeye
returns allowed us to asses a potentially strong fall MDN signal and as well the 2015 freshet was
expected to provide a significant contribution to the lake if river channel storage was an
41

Chapter 3: The Movement and Dispersal of a River Plume and its Contents in a Nursery Lake

important process. Capturing these two events was intended to represent the “peak” spawn
endmembers for evaluation of the process and effects of suspended particulates and their
associated MDNs moving through the watershed. Understanding how these sediments and
nutrients disburse into nursery systems is important for the future management of juvenile
salmon as, depending on their pathway and fate, these particulates may play a significant role in
their growth and development.
3.1.2 Research Question
Determining the depth at which sediment associated MDNs are delivered to HFB is
important in assessing if these nutrients are available for use by biota. If MDNs are delivered to
and stay suspended in the photic zone, lake surface water that receives light, they remain
available for primary and secondary production in nursery ecosystems. However, if the density
difference of the river water relative to the lake water forces MDNs to plunge to deeper aphotic
regions of the lake, the attenuated sunlight may restrict uptake by biota limiting primary
production but could potentially be available for benthic invertebrates in the deeper portions of
the bay and lake (secondary production). The specific question addressed in chapter 3 is:
•

Is there a difference in the movement and distribution of MDNs in the water column of
Horsefly Bay among the spring freshet, summer, and spawning event periods?

3.2 Methods
3.2.1 Site Characteristics
Water quality parameters were sampled from HFB on Quesnel Lake, British Columbia. A
more detailed description of the study site and its characteristics can be found in section 2.2.
Water temperature, conductivity, and water current patterns in HFB were monitored by a
mooring deployed by the Department of Fisheries and Oceans where the HFR empties into
Horsefly Bay in approximately 20 meters of water (Figure 2.1). This mooring was deployed in the
spring and recorded data through the fall each year. A more detailed description of this mooring
and its instrumentation can be found in section 3.2.3.1.

42

Chapter 3: The Movement and Dispersal of a River Plume and its Contents in a Nursery Lake

3.2.2 Study Design
Sample site locations were created by generating 50 points throughout Horsefly Bay with
an equal distance of 50 meters in-between each point (ArcMap 10). These sites ranged in distance
from the mouth of the HFR to the bay’s connection with Quesnel Lake. The techniques and
methods used in this research limited the number of locations that could be sampled in a
reasonable time. Therefore, 28 out of the 50 points were selected to encompass and represent
the different regions of HFB (Figure 3.1). This grid of sample locations as well as sites S1, S3, and
S5 (Section 2.3.2) were used in this study.

Figure 3.1: Map showing CTD sampling stations in HFB.

43

Chapter 3: The Movement and Dispersal of a River Plume and its Contents in a Nursery Lake

3.2.3 Sampling Techniques (Collection and Analysis)
3.2.3.1 Observations of Water Column Parameters
Water column profile properties including depth, turbidity, transmissivity, and
temperature were collected by a water profiling instrument (Seabird SB19 Plus) (CTD) which also
had some additional sensors attached to this unit. The CTD uses a 100x gain cable to amplify
signals in low turbidity settings. A five point transect was initially used to track the plume moving
through HFB. However, it was determined that a larger grid would need to be implemented to
better illustrate the movement of the HFR plume. A 28-point grid was added to this research
project near the end of the fall spawn in 2014; therefore, the CTD grid was used only once during
the 2014 spawning event. During 2015, however, sampling was performed weekly from April
15th, 2015 through August 8th.
This instrument was turned on, lowered into the water until submerged, and held at the
surface for one minute to allow the pump to purge standing water and air from the system. The
CTD was then lowered through the water column at 0.5-meters per second until within one meter
of the lake bottom. Data from the descending cast were used for analyses.
3.2.3.2 Observations of Water Column Parameters from the LISST
Suspended sediment concentrations, particle size profiles, and optical water properties
in the water column were determined using an in-situ laser diffraction (Sequoia Scientific - laser
in-situ scattering transmissometry -100x) (LISST). Laser diffraction is advantageous because the
size distribution of a large population of particles can be obtained from the pattern of scattered
light. This allows the LISST to record the particle-size distribution at discrete depths in the water
column as well as calculate a volume concentration of particles at any given depth. This
instrument was also used to calculate transmission data or the amount of light that passes
through a specific water volume, which aids in determining how turbid the water is at certain
depths. The LISST was lowered into the water until submerged, and held at the surface for one
minute to allow the instrument to warm up. At sites S1, S3, and S5 the LISST was lowered to a
pre-determined depth and retrieved in five-meter increments stopping one minute at each
depth. This was performed twice weekly from September 1st, 2014 through November 1st, 2014
and recommenced April 15, 2015 until the LISST malfunctioned on July 15th, 2015.
44

Chapter 3: The Movement and Dispersal of a River Plume and its Contents in a Nursery Lake

3.2.3.3 Mooring Observations of Water Column Parameters
For continuous monitoring of water properties in HFB a subsurface mooring was deployed
at the mouth of the HFR in approximately 20 meters of water. This mooring had sensors that
recorded water temperature, conductivity, water current movements, and depth (Figure 3.2). A
RBR TDR-2050 recorder that measured temperature
and pressure every 20 seconds was placed alongside a
RBR CT recorder that measured conductivity and
temperature at a depth of five meters every 40

5m

5

TDR205
0

C
T

TD C

seconds. A Teldyne RD Workhorse Sentinel 1.2 MHz

15m

1

Acoustic Doppler Current Profiler (ADCP) was anchored
at the bottom of the mooring and recorded horizontal
water current speeds and direction every 1m from the
bottom to approximately 4m below the surface every 1
to 18 minutes. A RBR TR-1050 was attached to the

A

ADCP

TR1050

T

Figure 3.2: Diagram of DFO
mooring setup.

ADCP to also measure temperature at the bottom of the mooring every 20 seconds. The mooring
was deployed in mid to late April every year and retrieved at the same time the following year.
3.2.4 Analysis of Results
Unlike in Chapter 2, there was no statistical analyses run on Chapter 3 data to determine
significant differences because the aim of this chapter was primarily to track the river plume in
the lake as a function of time and depth. Four variables (temperature, transmissivity, sediment
concentration, and D50) were plotted using ggplot (ggplot2) in R (version 3.3.2) to show the
movement of the HFR plume through HFB.
Spatial analysis was performed on the data from the twenty-eight point CTD grid using
linear interpolation. Data points that were chosen throughout HFB were unevenly spaced and
therefore the “maximum point technique” (spatial analyst tool) in ArcMap was used to
interpolate in-between points. The river plume’s movement through HFB has a directional bias
because it flows from the mouth of the HFR out into Quesnel Lake, in a general east to west
direction. Therefore, the general direction of the plume is known which better suits the

45

Chapter 3: The Movement and Dispersal of a River Plume and its Contents in a Nursery Lake

interpolation option of “kriging” because it assumes that the distance or direction between
sample points reflects a spatial correlation that can be used to explain the variation in the surface.
To assess CTD casts performed to characterize the vertical profile of HFB at various points
throughout the bay, a R code was created to generate a 3-dimensional plot that could be
assessed. The R code (version 0.6-2) used a package called akima, that uses bicubic spline
interpolation to generate values for spaces in-between points. Bicubic spline interpolation is an
extension of cubic interpolation and used to plot data points on a 2-dimensional grid. This
interpolation method is preferred in this situation over bilinear interpolation of nearest-neighbor
interpolation as it generates a smoother interpolation product (Johnston et al., 2003).

3.3 Results
Many studies have focused on the relationship between riverine sediment storage and
resuspension/transportation of particles as a means of evaluating whether, and for what time
period, nutrients are delivered and retained by stream beds. A retention and storage period
exceeding hours or days would retard the delivery of MDNs to a downstream nursery lake
(McConnachie & Petticrew, 2006; Rex & Petticrew, 2006; Rex et al., 2014) and enhance uptake
of MDN by riverine organisms. Other research has studied physical dynamics of nursery lakes
including water thermal structures and mixing parameters (Carmack et al., 1978) which influence
the trophic structure and diets of organisms that live within these nursery lakes (Hume et al.,
2005). However, there are few studies that demonstrate the linkage between physical limnology
and nutrient flow. This study attempts to bridge this gap to determine if the lake and river’s water
variable density structure, at different times of the year, regulates the distribution and therefore
the availability of these nutrients. Evaluating the spatial pattern of MDNs delivery with various
combinations of temperature/density differences between the river and lake should indicate if
and when MDNs are available for both primary and secondary production in the photic zone or
are involved in pelagic and/or benthic productivity.
3.3.1 Water Properties
3.3.1.1 Temperature & Thermal Structure
Water temperature values in the HFR, measured near the river mouth by the Hobo are
compared to the surface water temperatures measured in HFB on the mooring over the period
46

Chapter 3: The Movement and Dispersal of a River Plume and its Contents in a Nursery Lake

of this study in Figure 3.3. Throughout the spring freshet, the HFR temperatures were 1.3C
higher than temperatures in HFB. The summer water temperatures were 1.5oC above the surface
temperatures in HFB. However, at the beginning of August 2015, the river temperature decreases
faster than in the bay. Due to this decrease, which was observed during both the 2014 and 2015
spawning periods, the measured river temperatures were an average of 1.3C and 3.2C below
the surface waters in the bay in 2014 and 2015, respectively. This seasonal cooling pattern in the
river does not hold for the whole period of study but shows some fluctuations. For example,
there is a noticeable increase in temperature during the 2014 fall spawning period (mid-August)
in both the HFR and HFB and another in HFB at the end of October.

HFB
HFR

Fall Spawn
Spawn

Spring Freshet

Summer

Fall

Figure 3.3: Temperature values of the HFR and the surface water of HFB throughout
the period of record (2014 fall spawning period, 2015 spring freshet, summer, and fall
spawning periods). Method used to create trend lines in R was LOESS. River water
temperature was collected using the Hobo while HFB temperatures were collected
using the DFO mooring.

47

Chapter 3: The Movement and Dispersal of a River Plume and its Contents in a Nursery Lake

The thermal structure of HFB can be seen in Figure 3.4 which shows temperature profiles
at site S3 which is approximately 150 m from the river mouth in 15 m of water depth. During the
spring months of May and June the thermocline, a denser layer of water that can act to inhibit
particles from settling to deeper depths, is situated in the warmer photic waters above 8 meters
in depth. During July and August, the thermocline is deeper at around 10 to 12 meters and during
the months of September and October it is even deeper at approximately 13 to 15 meters.

Temperature (Celsius)
Figure 3.4: Temperature profiles at site S3 throughout the three study periods. Data
obtained using the CTD. Lines created with LOESS.
3.3.1.2 Turbidity & Transmission of Light
To assess where the HFR plume travels once it enters HFB, Celsius LISST data was used to
estimate SPM concentrations at various depths at site S3, which is in 15 meters of water and near
the mouth of the HFR. In Figure 3.5, SPM concentrations at the beginning of the summer sampling
period (mid-July 2014) tended to be highest at the surface (49 g/L). However, surface
concentrations dropped at the end of the summer at which time the highest concentrations,
approximately 20 g/L and 16 g/L, were found at 10 and 15-meter depths, respectively. During
the middle of September, SPM concentrations at all depths decreased to near zero, while at the
48

Chapter 3: The Movement and Dispersal of a River Plume and its Contents in a Nursery Lake

end of October, SPM concentrations at the depths of 5, 10, and 15 meters all increased again.
During the 2015 spring freshet, SPM concentrations were similar through the beginning and
middle of May at all depths. However, at the end of May, SPM concentrations increased at the
surface and at the five-meter depth. There were similar trends in SPM concentrations in both the
2014 and 2015 summer sampling periods. Concentrations at the 10 and 15-meter depths started
to increase as the surface and five-meter depth concentrations decreased towards the end of the
summer.

Summer

Fall Spawn

Spring Freshet

Figure 3.5: Site S3 SPM concentrations at five depths over the three study periods (20142015). Data was collected by using a LISST and the plotting method in R used to create
trend lines was LOESS.
Percent transmission values throughout the 2014 summer and salmon spawning periods
tended to stay consistent, ranging between 80 and 100 percent transmission (Figure 3.6). There
are points that dropped to as low as 60 percent transmission at depths of 10 and 15-meters
during the beginning and middle of September, respectively. The 2015 spring freshet sampling
period had larger and sharper decreases in transmission over time than the summer and salmon
spawning periods in 2014. At the end of May, there was a drop-in transmission at the surface and
at five-meters depth. However, the surface transmittance percentage leveled off at
49

Chapter 3: The Movement and Dispersal of a River Plume and its Contents in a Nursery Lake

approximately 80 percent as the five-meter transmission continued to drop to approximately 22
percent at the beginning of July. During the middle of June, transmission at the ten-meters depth
dropped and continued to drop to approximately 48 percent in the beginning of July.

Summer

Fall Spawn

Spring Freshet

Figure 3.6: Site S3 percent transmission levels at four depths over the three study
periods. Data was collecting by using a LISST and the plotting method in R used to create
trend lines was LOESS.
3.3.1.3 D50
The D50 or median diameter of the particle-size distribution at site S3 throughout the
three study periods is shown in Figure 3.7. The D50 during the 2014 summer study period started
off with the largest median particle-size distribution, approximately 340, 280, and 280 microns,
being at the surface, 10-meters deep, and at 15-meters deep respectively. The surface D 50
decreased as the salmon spawning study period began; dropping from approximately 280
microns to 20 microns. Although, the D50 at the five-meter depth started off the summer study
period at around 280 microns, it dropped to approximately 5 microns and did not increase again
until mid-October, when it peaked at approximately 250 microns. At the ten-meter depth we
see a constant slope which continued into the fall spawning study period. Although there are
outliers, one being in the middle of July and one at the beginning of the fall spawn study period.
50

Chapter 3: The Movement and Dispersal of a River Plume and its Contents in a Nursery Lake

In the 2015 spring freshet sampling period, the D50 at the surface of HFB started at approximately
95 microns and increased to approximately 245 microns where it remained consistent
throughout the rest of the 2015 study periods. D 50 at the five and 10 meter depths appear to
have the same trend in which they have peaks at the beginning and end of May and a decrease
in the beginning of July. The D50 at the 15-meter depth also decreased once the fall spawning
study period began due to two sampling dates in the beginning of September where the D 50 was
below 100 microns. However, after these two sampling days, D50 at the 15-meter depth exhibited
a median particle size of approximately 260 microns through into the beginning of October. The
D50 at the 15-meter depth trend stays consistent around 240 microns; however, there are outliers
at approximately 75 and 340 microns during the end of June and beginning of July respectively.

Summer

Fall Spawn

Spring Freshet

Figure 3.7: Site S3 D50 levels at various depths over the three study periods. Data was
collecting by using a LISST and the plotting method in R used to create trend lines was
LOESS.
3.3.2 Movement of the Plume
3.3.2.1 Overhead View of Turbidity
River water carrying suspended sediment enters HFB with a turbidity of approximately 3
NTU and hugs the northern shoreline until it enters Quesnel Lake with a turbidity of
51

Chapter 3: The Movement and Dispersal of a River Plume and its Contents in a Nursery Lake

approximately 2.5 NTU (Figure 3.8, A). Although the highest turbidity water stays along the north
shore of HFB, the HFR plume does affect and increase the turbidity across the whole bay until it
disperses into Quesnel Lake. The waters with the highest turbidity remain along the north shore
of HFB as they enter Quesnel Lake.
On May 29, 2015 NTU levels of the water entering HFB from the HFR reached upwards of
9.8 in the HFR delta (Figure 3.8, B). This plume moved through the bay in a similar fashion to the
plume observed on April 24 (Figure 3.8, A). The highest turbidity once again was found at the
river delta and hugged the northern shoreline as it moved through HFB and entered Quesnel
Lake. This plume also increased turbidity levels across the entire bay. Unlike, the plume on April
24, the plume on May 29th carried more turbid waters out into Quesnel Lake, spanning the full
extent of the HFB as it entered Quesnel Lake.
The HFR Plume on June 19, 2015 entered HFB at approximately 6 NTU and exhibited a
more direct path into the lake (Figure 3.8, C). The most turbid water was located at site b which
is shown on Figure 3.8 (C), even though sites a and c are closer to the mouth of the HFR. As the
plume moves out farther into HFB the route looks to be more localized, as it doesn’t span the
whole bay, and moves from the northern shore across the bay to the southeastern shore as it
makes its way out into Quesnel Lake.
On July 2, 2015, the HFR plume enters HFB with a turbidity of 3.1 NTU and travels along
the northern shore until it enters Quesnel Lake (Figure 3.9, A). After spreading along the north
shore, it spreads southward and eastward. As the plume spreads towards Quesnel Lake it flows
and exists on the northern shore.
In Figure 3.9 (B), the highest turbidity in HFB is approximately 2.8 NTU and is found at the
mouth of the HFR on July 24, 2015. Compared to previous sample dates, this turbidity level is
low. The HFR plume enters HFB and turns south until it hits the southern shore. In this case, the
turbid waters do not make it out into Quesnel Lake and end up diminishing as they move towards
the southern shoreline.
On August 7, 2015 (Figure 3.9, C), turbidity values in HFB are also found to be low with
the highest turbidity reading being approximately 2.8 NTU around the mouth of the HFR. As the
plume enters HFB it spreads out across the bay and uniformly dissipates as it moves towards
52

Chapter 3: The Movement and Dispersal of a River Plume and its Contents in a Nursery Lake

Quesnel Lake. At the mouth of the bay, turbidity levels are approximately 1 NTU and the HFR
plume looks to be non-existent.
In Figure 3.10 (A) from October 7, 2015, turbidity levels across HFB are less than 1 NTU.
Although the figure suggests differences in turbidity from the mouth of the HFR out into Quesnel
Lake, the differences are very small, indicating that the HFR plume is non-existent.
On October 20, 2015 (Figure 3.10, B), the highest turbidity level recorded in HFB was
approximately 4 NTU. Point b in Figure 3.15, shows where the highest turbidity was observed in
HFB, even though points a and c, which are closer to the mouth of the HFR show lower NTU
levels. Turbidity levels dropped as the HFR plume moved farther out into HFB and the plume was
not detected where HFB empties into Quesnel Lake.
When assessing the HFR plume’s movement and characteristics throughout the three study
periods, turbidity levels were highest during the spring freshet and lowest during the summer
and the salmon-spawning event. There was an increase in turbidity near the mouth of the HFR at
the end of the salmon spawning event. This increase was approximately 4 NTU higher than three
weeks earlier. The HFR plume was found travel and enter Quesnel Lake via the northern shore of
HFB.
When the spring freshet is ending and the summer is starting, the HFR plume appears to be
more focused and not as dispersed. During this time period, the HFR plume also changes direction
and is found along the southern shoreline.
3.3.2.2 Turbidity’s Vertical Profile and Distance from Delta
Figure 3.11 shows the evolution of the plume as it expands through the HFR delta and HFB at
four different time periods during 2015. On April 24 turbidity levels reached levels of
approximately 4 NTU in the top ten meters of the water column and extended approximately 700
meters into HFB (Figure 3.11, A). Turbidity levels decreased as the HFR plume exited HFB and was
approximately 3 NTU at the location where the plume entered Quesnel Lake. levels decreased
away from the river delta and were observed to be 5 NTU where the plume exited HFB. Again,
most of the particulate matter was contained within the top 10-15 meters of the water column.
However, as the plume moved eastward and exited the bay, the more turbid waters are found
higher in the water column and are contained within the top five meters.
53

A

B

54

C

a
Figu
b

c

Figure 3.8: Freshet study period overhead view maps of HFB
on April 24, 2015 (A), May 29, 2015 (B), and June 19, 2015
(C). Colors show NTU levels of highest water turbidity across
the bay. Turbidity colors were generated using ArcMap’s
interpolation kriging function.

A

B

55

C

Figure 3.9: Summer study period overhead view maps
of HFB from July 2, 2015 (A), July 24, 2015 (B), and
August 7, 2015 (C). Colors show NTU levels of
turbidity across the bay. Turbidity colors were
generated using ArcMap’s interpolation kriging
function.

A

B

a
b
c

Figure 3.10: Fall spawn study period overhead view map of HFB from
October 7, 2015 (A), and October 20, 2015 (B). Colors show NTU levels
of turbidity across the bay. Turbidity colors were generated using
ArcMap’s interpolation kriging function.

56

Chapter 3: The Movement and Dispersal of a River Plume and its Contents in a Nursery Lake

On June 19, the observed turbidity levels were lower than at the end of May (Figure 3.11,
C). The HFR plume entered HFB with a turbidity level of approximately 4 NTU and was contained
to the top five meters of the water column. The turbidity of the plume on this sampling day
decreased to approximately 2 NTU once it had traveled 1000 meters from the delta.
The data from July 2, showed a similar trend to that of June 19 (Figure 3.11, D). Once
again, the turbidity levels of the water entering the bay were approximately 4 NTU and the turbid
waters remained within the top five meters of the water column. However, during this sampling
period the HFR plume did not dissipate once it had traveled 1000 meters from the HFR delta but
continued traveling east and exited HFB with a NTU level of approximately 4.
On July 24 (Figure 3.12) there was increased turbidity near the HFR delta, with levels
reaching nearly 2 NTU (Figure 3.12, A). As the plume traveled farther from the delta, the turbidity
gradually decreased and the vertical extent of the plume also decreased.
On August 7, the turbidity in the HFB was found at deeper depths than previously observed
(Figure 3.12, B). The HFR plume entered HFB with turbidity levels of approximately 2 NTU. The
turbidity level increased with depth as the plume moved eastward and reached 6 NTU at a depth
of 27 meters, approximately 300 meters from the HFR delta. At a distance of 600 meters the
turbidity levels remained constant while the depth of the center of the plume increased to a
depth of almost 90 meters. The plume was not detectable beyond this point.
On October 7, 2014, there is no evidence of the HFR plume and the measured turbidity levels are
relatively constant throughout the bay (Figure 3.12, C). However, on October 20 an increase in
turbidity levels at deeper depths in HFB was detected (Figure 3.12, D). The observations from this
day show that turbidity levels reached approximately 4 NTU at a depth of 40 meters, which then
spread throughout the water column as the plume moved farther away from the HFR delta. This
plume started around 400 meters from the delta and traveled about 150 meters. As this plume
traveled eastward, both its turbidity level and the depth of maximum turbidity increased. Once
the plume had traveled approximately 550 meters, the depth of this plume spanned between 25
and 80 meters; even though, the turbidity level of the plume had decreased to approximately 2
NTU. There was another increase in turbidity at 1100 meters from the HFR delta with a turbidity

57

A

B

C

D

58
Figure 3.11: Interpolated vertical profile plot of the HFR plume (turbidity) as it enters at the mouth of the HFR and exits
at the end of HFB. These plots represent data taken from April 24, 2015 through July 2, 2015. Graph was created in R
using ggplot and data was interpolated using bicubic spline interpolation. White contour lines denote changes in
turbidity.

A

B

C

D

59
Figure 3.12: Interpolated vertical profile plot of the HFR plume (turbidity) as it enters at the mouth of the HFR and
exits at the end of HFB. These plots represent data taken from July 24, 2015 through October 20, 2015. Graph was
created in R using ggplot and data was interpolated using bicubic spline interpolation. White contour lines denote
changes in turbidity.

Chapter 3: The Movement and Dispersal of a River Plume and its Contents in a Nursery Lake

level of approximately 3 NTU. This increase in turbidity covered the whole water column 30
meters in depth to the bottom at 100 meters.

3.4 Discussion
The movement and dispersal of river water and its contents entering a large body of water
such as a lake are affected by the physical characteristics of both the river and lake waters
(Carmack et al., 1979). The sampling stations used during this research were spread throughout
HFB to ensure the ability to compare the movements of the HFR plume across a spring freshet, a
summer, and a salmon spawning period. The movement patterns of the HFR plume are driven by
differing temperatures and turbidities, which modify densities in the HFR and HFB and the HFR’s
flow. Trends in D50, concentration, and the spatial mapping suggest that the movement of the
HFR Plume and its contents differ depending on time of year, which could affect where MDNs
may be available to organisms living in this nursery system. Although some results obtained from
this research were consistent with previous studies performed (Carmack et al., 1979; Reichert,
2009), there were some that were not (Carmack et al., 1979). As stated in the previous chapter,
these research findings could be system-specific.
3.4.1 Characteristics of Horsefly Bay
Temperature is one of the main factors when trying to predict how a river’s water and
contents will interact with that of a lake (Hilton et al., 1986). At temperatures above ~4oC the
colder a water body is, the denser it is. Therefore, colder river water will enter and stay below
warmer lake water allowing the contents of the river water to settle to the bottom of the lake
(Wetzel, 2001). On the other hand, if the river’s water is warmer than that of the lake’s water,
then the river water and its contents, if the suspended material does not modify the density
differences significantly, will stay suspended in the lake’s water column. In Figure 3.4, the HFR
water temperature is warmer than that of HFB during the spring freshet and most of the summer
study periods (May-beginning of August). Therefore, river water and its contents that are
entering HFB from the HFR would most likely stay suspended higher in the water column
assuming the SPM concentration is minimal. The contents of the HFR, which could contain MDNs
would be more readily available for uptake by plankton living in these warmer surface waters
(Baines & Pace, 1994). Evidence to corroborate this is seen in the results of chlorophyll-a and
60

Chapter 3: The Movement and Dispersal of a River Plume and its Contents in a Nursery Lake

fluorescence measurements, which both exhibit peaks at the end of the spring freshet. In the
beginning of August, the HFR water temperature dropped and became colder than HFB’s water
temperature. The river water would become denser than that of HFB’s water, and upon entering
the bay its contents would stay below the thermocline promoting settling to the bottom of the
bay or further offshore in Quesnel Lake. These findings are consistent with what was found with
suspended particulates entering Quesnel Lake from the Mount Polley Mine spill staying
suspended below the thermocline and spreading out over deeper depths of the lake (Petticrew
et al., 2015).
Water heavily laden with suspended particulates will also affect where river water and its
contents will go once they enter a lake system. Assuming equivalent temperatures, waters that
are less laden with particulates are less dense than water that contains more suspended
particulates. At the end of the summer and throughout the salmon spawn period, the most turbid
waters are located at around the 10 to 15 meter depths, while the river water is colder than that
of lake surface water. Water and particulate contents during this study period were entering HFB
below the thermocline and staying suspended in the hypolimnion until settling out on the bottom
of HFB, as the thermocline was located at approximately 15 meters in depth. During the spring
freshet, the most turbid waters are found to be at the surface and at the 5-meter depth. This is
due to warmer river water entering a cooler HFB. Even when the freshet exhibits the highest
SPM concentrations for the study period, the amount of particulates does not result in a denser
plunging plume (Figure 3.3). Water that enters HFB from HFR during this study period is
remaining suspended in the surface epilimnetic water because of the depth of the thermocline.
Another way to look at a waterbody’s particle concentration, is by recording transmission
or light attenuation data. Transmission is the measurement of how much light can pass through
a certain volume of water. The more turbid a waterway, the less light transmission there is. In
Figure 3.6, transmission data trends are similar to what was seen in the turbidity data in the
previous section. Lower levels of transmission are observed at deeper depths, around 10-15
meters, during the end of the summer and throughout the salmon spawning period. However,
during the spring freshet the upper 5 m of water tend to have lower transmission, meaning they
contain higher amounts of suspended particulates than water at the 10 or 15-meter depth. These
61

Chapter 3: The Movement and Dispersal of a River Plume and its Contents in a Nursery Lake

results indicate that when the HFR plume and its particles enter HFB from August through the
end of October, they enter below the thermocline which restricts mixing into the epilimnion and
enhances settling through the hypolimnion. On the other hand, evidence here indicates when
the HFR plume enters HFB from April through July, it enters above the thermocline and remains
suspended in the warmer surface waters.
Analyzing the size of the particulates coming out of the HFR and whether they stay
suspended in the water column are important in understanding the movement of the particulates
that are transported via the HFR. Suspended particulates entering HFB were largest during the
spring freshet and the beginning of the summer at all depths. There is a decrease in the
suspended particulate median size at the end of summer and throughout the salmon spawning
period. This suggests that with higher flow periods, larger particles can be transported down river
and out into HFB. In the spring freshet, larger particles tend to be suspended between the 5 and
10-meter range. However, at the start of June we see a drop in D 50 at the 5 to 10-meter range
and a slight increase in D50 at the surface which contains the highest D50 until the start of the
summer period. The increase in D50 in surface waters could be due to a rise in chlorophyll-a levels
as we see an increase during the 2014 spawning period and during June 2015 (Figure 2.8). These
small plankton communities would affect D 50 levels at shallower depths of the water column as
the thermocline would trap these smaller organisms in the photic surface waters. Two studies
previously looking at suspended particulate median size in O’Ne-eil Creek and the HFR
(McConnachie & Petticrew, 2006; Albers & Petticrew, 2012) found that larger-sized particles
were found to be suspended in the river during spawning and post-spawning events. Most likely
bioturbation caused by spawning salmon had re-suspended larger particles that may had been
trapped in the river beds. However, results found in this study show that larger particles
suspended in HFB are found to be present in the spring freshet with lower median-sized particles
being found during the spawning event. This suggests that although larger particles are resuspended by bioturbation, they never make their way to HFB and are only transported by the
higher flow event of the spring freshet. This assumption is supported by Liu et al. (2011) who
found particle size and load in suspended and bed sediment to be highly dependent on discharge.

62

Chapter 3: The Movement and Dispersal of a River Plume and its Contents in a Nursery Lake

3.4.2 The Dispersal of the Horsefly River Plume
Assessing the dispersal of the HFR plume into HFB is most easily observed through spatial
and temporal mapping of turbidity. Such analysis shows that the HFR plume brings the most
turbid waters into HFB during the spring freshet period with turbidity levels reaching
approximately 11 NTU. There is a slight increase in turbidity during the salmon spawn period,
which is most likely due to bioturbation and redd construction (Naiman et al., 2002; Moore et al.,
2007; Albers, 2010) as well as a late October storm event (Figure 2.2). These results support
findings that were documented in chapter 2. One common trend that is shown in these maps and
is consistent with previous research studies (Carmack et al., 1979; Saetre & Ljoen, 1972; Hilton
et al., 1986), is the movement of the HFR plume seems to be affected by the Coriolis effect. Seven
out of the eight days sampled, the HFR plume makes its way into HFB and turns right towards the
southernmost shoreline. May 29th, 2015 is the only sampling date that the plume appears to exit
HFB along the northern shoreline. This date also exhibits the highest turbidity levels observed
during this research period and supports the findings of Saetre & Ljoen (1971) who state that the
Coriolis force has less effect on plumes with higher velocities and higher water densities.
The HFR plume stays suspended during the spring freshet and beginning of summer and
drops to lower depths at the end of the summer and throughout the salmon spawning period.
Turbidity again was shown to be highest during the spring freshet and lowest during the summer
and salmon spawning periods. In the months of April, May, June, and the beginning of July, the
HFR plume stays in the upper 10 m of the lake, and the densest water is at a depth of 10 to 15
meters (Figure 3.17). This dense water layer is likely the reason that the HFR plume and its
contents are staying suspended above the 10 m depth. In the months of August, September and
October, there is little evidence on certain sampling days that a plume exists in HFB. However,
when there are higher turbidity levels during this time period, those areas of high turbidity exist
below the 20 to 25-meter depth. This suggests that if a HFR plume enters HFB during this period
that it and its contents will enter below the thermocline and would not be able to rise in the
water column through warmer surface waters. These findings and assumptions are consistent
with Petticrew et al. (2015), where the water column thermal structure restricted high turbid
waters from mixing with epilimnetic waters.
63

Chapter 3: The Movement and Dispersal of a River Plume and its Contents in a Nursery Lake

3.4.3 Implications
The movement and dispersal of the HFR plume to HFB is dependent on the time of year,
which affects the relative differences in temperature of the two “water systems”. Warmer river
waters during the spring freshet allows river water and its contents to stay suspended in the
warmer surface lake water, while the colder river water in the fall spawn period plunges down
below the surface lake water and is trapped beneath the thermocline. These different processes
could directly affect how sediment-associated nutrients, that are carried in this HFR plume, are
utilized and taken up by biota living in HFB. Nutrients that are suspended in the warmer surface
waters are more readily utilized by plankton populations living in these waters, and will aid in
enhanced primary and secondary production. Nutrients that enter below the thermocline are
less available for uptake by plankton communities, limiting primary production and may only be
available to secondary production. In Figures 2.8 and 2.9, there are peaks in both chlorophyll-a
and fluorescence at the end of both the 2014 fall spawn and the 2015 spring freshet. This could
indicate that MDNs are available and being utilized during these study periods and increasing
productivity. This implies that particulates entering HFB during the fall season are still entering
above the thermocline but at deeper depths seen in the freshet and summer periods.
This study indicates that the direction of the plume is affected by the Coriolis force, such
that suspended particulates that are transported by the plume can be deposited on the southernmost banks of HFB and not into the deeper parts of Horsefly Bay or Quesnel Lake. The deposition
of these suspended particulates in shallow waters will expose themto subsequent re-suspension
processes of wind-generated waves or higher river flows associated with the spring freshet.
These processes can thenincrease the distribution of nutrients that werenot previously utilized
allowing these nutrients to be delivered into deeper lake regions.
Linking the information in this chapter, with evidence presented in chapter 2 of this thesis
can help determine where nutrients are being delivered, and when they are available to biota
living in HFB. These findings support the idea that future populations of salmonids are dependent
on the fall flush and delivery of MDNs to nursery lake systems.

64

Chapter 4: Conclusions and Management Implications

Chapter 4: Conclusions and Management Implications
4.1 Conclusions
This thesis demonstrates how seasonal and temporal differences in discharge and water
properties affect timing of the movement and the range dispersal of marine-derived nutrients
(MDNs) from a salmon spawning river to a nursery lake. Moreover, this thesis supplements
existing literature which explores and aims to determine whether MDNs are beneficial, neutral,
or detrimental to juvenile salmonid rearing habitat. The results presented in Chapter 2 and 3
show that MDNs do enter Horsefly Bay (HFB) during both the spring freshet, during which
nutrients remain suspended, as well as the fall spawning period, during which nutrients fall to
deeper depths. These natural and behavioral observations provide further insight into the
movement of MDNs through freshwater systems and aid in determining the degree to which they
contribute to lake productivity.
The findings in Chapter 2 support past literature, which states that MDNs have a positive
effect on chlorophyll-a, phosphorous, fluorescence, total nitrogen, and total carbon levels that
could potentially stimulate nursery lake productivity-- helping drive algae, zooplankton, and fish
populations in a bottom-up trophic transfer (Hyatt et al., 2004; DeVries, 2012). While these
findings outline benefits to the productivity of HFB, it’s important to consider this research does
not specifically address whether lake productivity has an effect on salmon-spawner populations.
However, as productivity is positively linked to the increase of bottom-up trophic transfer, one
may conclude that growth and survival of juvenile salmon willbenefit from the delivery of MDNs.
Understanding the underlying mechanisms of a biological process is important in
assessing how particular processes can affect organisms. Both the discharge rates and presence
of salmon (spawn and decay) played critical roles in regulating MDN load and MDN
concentration, respectively. The productivity of HFB benefited from the increased load of MDNs
delivered during the spring freshet and the high concentrations of MDNs entering in the fall
spawn, as shown by increased chlorophyll-a and fluorescence levels during these two study
periods. Although it was identified that nutrients entering HFB during the fall spawning period
enter below the thermocline and settle to the bottom based on differences in river and lake
65

Chapter 4: Conclusions and Management Implications

water temperatures, there was still an increase in productivity during this period. The D50
distribution throughout the water column during the fall spawning period identifies smaller
particles suspended higher in the water column, around the five-meter mark, while larger
particles are found in deeper water, around 15 meters. This shows that larger particles entering
HFB are entering below the thermocline and dropping to deeper depths as is shown in Figure
3.12, whereas lighter particles are entering above the thermocline and staying suspended-aiding in productivity as can be seen in Figures 2.9 and 2.10. These findings indicate that MDNs
entering HFB in the fall spawning period are spanning across the entire water column and could
be aiding in both primary and secondary production. This also implies that MDNs are available to
support the survival of benthic lake invertebrates, as shown by Wiplfli et al. (2011) and Bilby et
al. (2011) in rivers and streams supplied with salmon carcasses.
Overall, the spring freshet and fall spawning periods are equally important in the delivery
of MDNs to nursery systems. The spring freshet delivered an increased but diluted load of MDNs
to HFB, fueling chlorophyll-a, phosphorous, and fluorescence levels and, in turn, increased
productivity. The fall spawn delivered a more concentrated load of MDNs to HFB, supplying both
the epilimnion and hypolimnion with nutrients that could be used in the growth of algae,
plankton populations, and benthic communities.

4.2 Future Management Implications
The results summarized throughout this thesis could help researchers better understand
the timing and movement of MDNs entering HFB from the HFR. In Chapter 2, data shows that the
load of marine-derived N is driven by higher discharge due to the spring melt. With climates
changing across Central British Columbia, it’s important to question how earlier spring melts may
affect the timing and load of MDNs that are traditionally delivered to HFB in the months of April,
May, and June. In Figure 4.1, early melting periods are evident in 2015 and 2017. These melts
start as early as January and greatly affect the discharge during previously normal freshet time
periods. Colder weather could potentially impact the utilization of MDNs in primary production,
limiting the uptake by plankton as growth rates and populations decrease during winter months
(Weslawski et al., 1991). MDNs entering HFB at this time could settle on the bottom and become
re-suspended in later, high-flow periods or be lost entirely to productivity. The results found in
66

Chapter 4: Conclusions and Management Implications

this thesis, combined with further research into winter MDNs delivery and movement, may help
in understanding how earlier melts might affect productivity and juvenile salmon growth rates in
nursery systems.

Figure 4.1: 2014 through 2017 hydrograph data taken from Environment Canada station number 08KH031 in Horsefly River above Quesnel Lake.
Within the last few years, salmon return numbers have been lower than expected (Cone,
2017). If returns continue to decrease, the amount of MDNs being delivered to these nursery
systems will also decrease, limiting productivity (Bilby et al., 1996). Past studies have shown the
benefits of artificially fertilizing rivers and streams with salmon carcasses to increase productivity
throughout the system (Bilby et al., 2011; Claeson et al., 2011; Wipfli et al., 2011). Results from
Chapters 2 and 3 demonstrate both timing and delivery tendencies and mechanisms; these
results may help in determining when and where artificial fertilization can be implemented-- for
example, in systems where low salmon return numbers are limiting primary and secondary
production. Although a few published studies project that the increase of imported nutrients to
nursery systems is detrimental to future female spawner populations (Holtgrieve & Schindler,
2011; Walters, 2014), there has been no data collected to support these predictions. However,
alternative studies show that salmon carcasses, and the nutrient input they provide to nursery
systems, are beneficial and of ecological value (Johnson et al., 2004; Wipfli et al., 2010). The
research conducted for this thesis, supports the idea that MDNs do enhance productivity in this
nursery system and in turn should increase the survivorship of salmon living in this nursery lake.

67

References
Albers, S. 2010. The Salmon disturbance regime effects on biofilm, sediment and water;
Horsefly Spawning Channel, Horsefly, B.C. Master’s Thesis, University of Northern British
Columbia.
Albers, S., & Petticrew, E. 2012. Ecosystem response to a salmon disturbance regime:
Implications for downstream nutrient fluxes in aquatic systems. Limnology and
Oceanography, 57(1), 113-123.
Albers, S., & Petticrew, E. 2013. Biogeomorphic impacts of migration and disturbance:
implications of salmon spawning and decay. Geomorphology, 202, 43-50.
Baines, S., & Pace, M. 1994. Relationships between suspended particulate matter and sinking
flux along a trophic gradient and implications for the fate of planktonic primary production.
Canadian Journal of Fisheries and Aquatic Sciences, 51, 25-36.
Bilby, R., Fransen, B., & Bisson, P. 1996. Incorporation of nitrogen and carbon from spawning
Coho salmon into the trophic system of small streams: Evidence from stable isotopes.
Canadian Journal of Fisheries and Aquatic Sciences, 53, 164–173.
Boehrer, B., & Schultze, M. 2008. Stratification of lakes. Reviews of Geophysics. 46, RG2005.
doi:10.1029/2006RG000210.
Bricker, S., Longstaff, B., Dennison, W., Jones, A., Boicourt, K., Wicks, C., & Woerner, J. 2007.
Effects of nutrient enrichment in the nation's estuaries: a decade of change. NOAA Coastal
Ocean Program Decision Analysis Series, No. 26. National Centers for Coastal Ocean Science,
Silver Spring, MD. 322.
Brock, C., Leavitt, P., Schindler, D., Johnson, S., & Moore, J. 2006. Spatial variability of stable
isotopes and fossil pigments in surface sediments of Alaskan coastal lakes: Constraints on
quantitative estimates of past salmon abundance. Limnology and Oceanography, 51, 16371647.
Brock, C., Leavitt, P., Schindler, D., & Quay, P. 2007. Variable effects of marine-derived nutrients
on algal production in salmon nursery lakes of Alaska during the past 300 years. Limnology
and Oceanography, 52(4), 1588-1598.
Burgner, R. 1991. Life history of sockeye salmon (Oncorhynchus nerka). In Pacific Salmon Life
Histories. University of British Columbia Press, Vancouver, 1-117.

68

Cak, A., Chaloner, D., & Lamberti, G. 2008. Effects of spawning salmon on dissolved nutrients
and epilithon in coupled stream-estuary systems of southeastern Alaska. Canadian Journal
of Fisheries and Aquatic Sciences, 70, 169–178. doi:10.1007/s00027-008-8090-5.
Carmack, E., Gray, C., Pharo, C., & Daley, R. 1979. Importance of lake-river interaction on
seasonal patterns in the general circulation of Kamloops Lake, British Columbia. Limnology
and Oceanography, 24(4), 634-644.
Cederholm, J., Kunze, M., Murota, T., & Sibatani, A. 1999. Pacific salmon carcasses: essential
contribution of nutrients and energy for aquatic and terrestrial ecosystems. Fisheries,
24(10), 6-15.
Choles, J. 2004. Sediment Sources and Movement in Lesser Slave Lake. Alberta Environment,
Publication NO. T/790. ISBN: 0-7785-4004-9.
Claeson, S., Li, J., Compton, J., & Bisson, P. 2006. Response of nutrients, biofilm, and benthic
insects to salmon carcass addition. Canadian Journal of Fisheries and Aquatic Sciences, 63,
1230-1241.
Compton, J., Anderson, C., Phillips, D., Brooks, R., Jonhson, M., Church, R., Hogsett, W., Cairns,
M., Rygiewicz, P., McComb, B., & Shaff, C. 2006. Ecological and water quality consequences
of nutrient addition for salmon restoration in the Pacific Northwest. Frontiers in Ecology
and the Environment, 4(1), 18-26.
Cone, T. 2017. Email Correspondence, Department of Fisheries and Oceans, Salmon
Escapement in Fraser River System Numbers.
Desloges, J. & Gilbert, R. 1998. Sedimentation in Chilko Lake: a record of the geomorphic
environment of the eastern Coast Mountains of British Columbia, Canada.
Geomorphology, 25, 75–91.
Devries, P. 2012. Salmonid influences on rivers: a geomorphic fish tail. Geomorphology, 157,
66-74.
Finney, B., Gregory-Eaves, I., Sweetman, J., Douglas, M., & Smol, J. 2000. Impacts of climatic
change and fishing on Pacific salmon abundance over the past 300 years. Science, 290, 795799.
Gende, S., Edwards, R., Willson, M., & Wipfli, M. 2002. Pacific salmon in aquatic and terrestrial
ecosystems. BioScience, 52, 917-928.
Glass, G., Peckham, P., & Sanders, J. 1972. Consequences of failure to meet assumptions
underlying fixed effects analyses of variance and covariance. Review of Educational
Research, 42, 237-288.
69

Gregory, S., Swanson, F., McKee, W., & Cummins, K. 1991. An ecosystem perspective of riparian
zones. BioScience, 540–551.
Gresh, T., Lichatowich, J., & Schoonmaker, P. 2000. An estimation of historic and current levels
of salmon production in the Northeast Pacific ecosystem: evidence of a nutrient deficit in
the freshwater systems of the Pacific Northwest. Fisheries, 25, 15-21.
Grimes, C. & Kingsford, M. 1996. How do riverine plumes of different sizes influence fish larvae:
do they enhance recruitment? Marine and Freshwater Research, 47, 191-208.
Harwell, M., Rubinstein, E., Hayes, W., & Olds, C. 1992. Summarizing Monte Carlo results in
methodological research: the one- and two-factor fixed effects ANOVA cases. Journal of
Education Statistics, 17, 315-339.
Hicks, B., Wipfli, M., Lang, D., & Lang, M. 2005. Marine-derived nitrogen and carbon in
freshwater-riparian food webs of the Copper River Delta, southcentral Alaska. Oecologia,
144(4), 558-569.
Hocking, M. & Reynolds, J. 2011. Impacts of salmon on riparian plant diversity. Science,
331(6024), 1609-1612.
Hodder, K. 2009. Flocculation: a key process in the sediment flux of a large, glacier-fed lake.
Earth Surface Processes and Landforms, 34(8), 1151-1163.
Holtgrieve, G. & Schindler, D. 2011. Marine-derived nutrients, bioturbation, and ecosystem
metabolism: reconsidering the role of salmon in streams. Ecology, 92(2), 373-385.
Hood, E., Fellman, J., & Edwards, R. 2007. Salmon influences on dissolved organic matter in a
coastal temperate brown-water stream: an application of fluorescence spectroscopy.
Limnology and Oceanography, 52, 1580–1587.
Horowitz, A. & Elrick, K. 1988. Interpretation of bed sediment trace metal data: methods of
dealing with the grain size effect. Chemical and biological characterization of sludges,
sediments, dredge spoils, and drilling muds. American Society for Testing and Materials,
114-128.
Hume, J., Shortreed, K., & Morton, K. 1996. Juvenile sockeye rearing capacity of three lakes in
the Fraser River system. Canadian Journal of Fisheries and Aquatic Sciences, 53, 719-733.
Hume, J., Shortreed, K., & Whitehouse, T. 2005. Sockeye fry, smolt, and nursery lake monitoring
of Quesnel and Shuswap lakes in 2004. Fisheries & Oceans Canada.
Hyatt, K. & Stockner, J. 1985. Responses of sockeye salmon (Oncorhynchus nerka) to
fertilization of British Columbian Coastal Lakes. Canadian Journal of Fisheries and Aquatic
Sciences, 42, 320-321.
70

Hyatt, K., McQueen, M., Shortreed, K., & Rankin, D. 2004. Sockeye salmon (Oncorhynchus
nerka) nursery lake fertilization: review and summary of results. Environmental Reviews,
12(3), 133-162.
Johnson, N., McDonald, J., Hall, K., & Tchaplinski, P. 1997. A preliminary study of the role of
sockeye salmon (Oncorhynchus nerka) carcasses as carbon and nitrogen sources for benthic
insects and fishes in the “Early Stuart” stock spawning streams, 1050 km from the ocean.
Fisheries Project Report No. RD55, British Columbia Ministry of Environment, Lands and
Parks, Victoria, British Columbia.
Johnston, K., Ver Hoef, J., Krivoruchko, K., & Lucas, N. 2003. ArcGIS 9 - Using ArcGIS
Geostatistical Analyst. ESRI.
Johnston, N., MacIsaac, E., Tschaplinksi, P., & Hall, K. (2004) Effects of the abundance of
spawning sockeye salmon (Oncorhynchus nerka) on nutrients and algal biomass in forested
streams. Canadian Journal of Fisheries and Aquatic Science, 61, 384–403.
Kline, T., Goering, J., Mathisen, O., Poe, P., & Parker, P. 1990. Recycling of elements transported
upstream by runs of Pacific salmon: 15N and 13C evidence in Sashin Creek, Southeastern
Alaska. Canadian Journal of Fisheries and Aquatic Sciences, 47, 136–144.
Kline, T., Goering, J., Mathisen, O., Poe, P., Parker, P., & Scalan, R. 1993. Recycling of elements
transported upstream by runs of Pacific salmon: II. 15N and 13C evidence in the Kvichak River
watershed, Bristol Bay, Southwestern Alaska. Canadian Journal of Fisheries and Aquatic
Sciences, 50, 2350–2365.
Kline, T., Goering, J., & Piorkowski, R. 1997. The effect of salmon carcasses on Alaskan
freshwaters. In Freshwaters of Alaska: ecological syntheses. Ecological Studies 119.
Springer- Verlag. New York, 179-204.
Kyle, G. 1996. Stocking sockeye salmon (Oncorhynchus nerka) in barren lakes of Alaska: effects
on the macrozooplankton community. Fisheries Research, 28, 29- 44.
Kyle, G., Koenings, J., & Edmundson, J. 1997. An overview of Alaska lake- rearing enhancement
strategy: nutrient enrichment and juvenile stocking. In Freshwaters of Alaska: ecological
syntheses. Springer-Verlag, New York.
Larkin, G., & Slaney, P. 1997. Implications of trends in marine-derived nutrient influx to South
Coastal British Columbia salmonid production. Fisheries, 22, 16-24.
Liu, Y., Metivier, F., Gaillardet, J., Ye, B., Meunier, P., Narteau, C., Lajeunesse, E., Han, T., &
Malverti, L. 2011. Erosion rates deduced from seasonal mass balance along the upper
Urumqi River in Tianshan. Solid Earth, 2, 283-301.
71

Lix, L., Keselman, J., & Keselman, H. 1996. Consequences of assumption violations revisited: a
quantitative review of alternatives to the one-way analysis of variance F test. Review of
Educational Research, 66, 579-619.
Mallin, M., Cahoon, L., & Durako, M. 2005. Contrasting food-web support bases for adjoining
river-influenced and non-river influenced continental shelf ecosystems. Estuarine, Coastal
and Shelf Science, 62, 55-62.
Mathisen, A. 1972. Biogenic enrichment of sockeye salmon lakes and stock productivity.
Verhandlungen der Internationalen Vereinigung fur Theoretische und Angewandte
Limnologie, 18, 1089–1095.
Mazumder, A. & Edmundson, J. 2002. Impact of fertilization and stocking on trophic
interactions and growth of juvenile sockeye salmon (Oncorhynchus nerka). Canada Journal
of Fisheries and Aquatic Sciences, 59, 1361-1373.
McConnachie, J., & Petticrew, E. 2005. Tracing organic matter sources in riverine suspended
sediment: implications for fine sediment transfers. Geomorphology, 79, 13-26.
McKinnel, S., Curchitser, E., Groot, C., Kaeriyama, M., & Myers, K. The decline of Fraser River
Sockeye Salmon Oncorhynchus nerka (Steller, 1743) in relation to marine ecology. Technical
Report 4.
Ministry of Environment. 1998. Guidelines for interpreting water quality data. Resources
Inventory Committee. Version 1.0.
https://www.for.gov.bc.ca/hts/risc/pubs/aquatic/interp/intrptoc.htm
Moore, J., & Schindler, D. 2004. Nutrient export from freshwater ecosystems by anadromous
sockeye salmon (Oncorhynchus nerka). Canadian Journal of Fisheries and Aquatic Sciences,
61, 1582-1589.
Moore, J., Schindler, D., Carter, J., Fox, J., Griffiths, J., & Holtgrieve, G. 2007. Biotic control of
stream fluxes: spawning salmon drive nutrient and matter export. Ecology, 88, 1278-1291.
Morgan, C., Robertis, A., & Zabel, R. 2005. Columbia River plume fronts. Hydrography,
zooplankton distribution, and community composition. Marine Ecology Progress Series, 299,
19-31.
Naiman, R., Bilby, R., Schindler, D., & Helfield, J. 2002. Pacific salmon, nutrients, and the
dynamics of freshwater and riparian ecosystems. Ecosystems, 5, 399–417.
Neil, A. 2006. Horsefly River nutrient loading and dispersion in Quesnel Lake. 2006 preliminary
assessment and water quality monitoring. Ministry of Environment, Canada.

72

Noakes, D., Beamish, R., & Kent, M. 2000. On the decline of Pacific salmon and speculative links
to salmon farming in British Columbia. Aquaculture, 183, 363-386.
Northcote, T., & Hartman, G. 2008. Fishes and Forestry: Worldwide Watershed Interactions and
Management. John Wiley & Sons.
O’Keefe, T., & Edwards, R. 2002. Evidence for hyporheic transfer and removal of marine-derived
nutrients in a sockeye stream in Southwest Alaska. American Fisheries Society Symposium,
33, 99-107.
Owens, P. 2005. Conceptual models and budgets for sediment management at the river basin
scale. Journal of Soils & Sediments, 5(4), 201-212.
Parnell, A. 2016. Stable Isotope Mixing Model. Package Version 0.3. R Version 3.2.2.
https://cran.r-project.org/package=simmr.
Petticrew, E., Rex, J., & Albers, S. 2011. Bidirectional delivery of organic matter between
freshwater and marine systems: the role of flocculation in Pacific salmon streams. Journal of
the North American Benthological Society, 30, 779-786.
Petticrew, E., Albers, S., Baldwin, S., Carmack, E., Dery, S., Gantner, N., Graves, K., Laval, B.,
Morrison, J., Owens, P., Selbie, D., & Vagle, S. 2015. The impact of a catastrophic mine
tailings impoundment spill into one of North America’s largest fjord lakes: Quesnel Lake,
British Columbia, Canada. Geophysical Research Letters, 42(9), 3347-3355.
Quinn, T. 2005. The Behavior and Ecology of Pacific Salmon and Trout. University of Washington
Press (April 30, 2005).
Reichert, J. 2009. River plume effects on yellow perch growth, survival, and recruitment in Lake
Erie. Electronic Theses and Dissertations. University of Windsor.
Reisinger, A., Chaloner, D., Ruegg, J., Tiegs, S., & Lamberti, G. 2013. Effects of spawning Pacific
salmon on the isotopic composition of biota differ among southeast Alaska streams.
Freshwater Biology, Blackwell Publishing Ltd.
Rex, J., & Petticrew, E. 2008. Delivery of marine-derived nutrients to streambeds to Pacific
salmon. Nature Geoscience, 1, 840-843.
Rex, J. 2009. The flocculation feedback loop: Delivery of marine derived nutrients in Pacific
salmon streams. Ph.D. thesis, University of Northern British Columbia.
Rex, J., Petticrew E., Albers, S., & Williams, N. 2014. The influence of Pacific salmon decay
products on near-field streambed sediment and organic matter dynamics: a flume
simulation. Earth Surface and Processes Landforms, 39, 1378-1385.
73

Rinella, D. 2010. Marine-derived nutrients in riverine ecosystems: developing tools for tracking
movement and assessing effects in food webs on the Kenai Peninsula, Alaska. University of
Alaska Fairbanks.
Saetre, R. & Ljoen, R. 1972. The Norwegian coastal current. The Technical University of Norway.
Schindler, D., Scheuerell, M., Moore, J., Gende, S., Francis, T., & Palen, W. 2003. Pacific salmon
and the ecology of coastal ecosystems. Frontiers in Ecology and the Environment, 1, 31–37.
Schindler, D., Leavitt, P., Brock, C., Johnson, S., & Quay, P. 2005. Marine-derived nutrients,
commercial fisheries, and production of salmon and lake algae in Alaska. Ecology, 86 (12),
3225-3231.
Schnorbus, M., Werner, A., & Bennett, K. 2012. Impacts of climate change in three hydrologic
regimes in British Columbia, Canada. Hydrological Processes, 28(3), 1170-1189.
Schmidt, D., Carlson, S., & Kyle, G. 1998. Influence of carcass-derived nutrients on sockeye
salmon productivity of Karluk Lake, Alaska: importance in the assessment of an escapement
goal. North American Journal of Fisheries Management, 18, 743–763.
Shortreed, K. & Hume, J. In prep. Effects of increased sockeye escapements on productivity of a
large British Columbia lake.
Stockner, J. & Shortreed, K. 1989. Algal picoplankton production and contribution to food-webs
in oligotrophic British Columbia lakes. Hydrobiological, 173, 151- 166.
Stockner, J. 2003. Nutrients in salmonid ecosystems: sustaining production and biodiversity.
American Fisheries Society Symposium 34. American Fisheries Society, Bethesda, Maryland,
USA.
Stoica, P. & Selen, Y. 2004. Model-order selection, a review of information criterion rules. Ieee
Signal Processing Magazine. July 2004.
http://www.sal.ufl.edu/eel6935/2008/01311138_ModelOrderSelection_Stoica.pdf
Stouder, D., Bisson, P., & Naiman, R. 2012. Pacific Salmon & their Ecosytems: Status and Future
Options. Springer Science & Business Media.
Strickland, J. & Parsons, T. 1972. A practical handbook of seawater analysis. Journal of Fisheries
and Research Board of Canada, 167(2), 310.

74

Vanderploeg, H., Johengen, T., Lavrentyev, P., Chen, C., Lang, G., Agy, M., Bundy, M., Cavaletto,
J., Eadie, B., Liebig, J., Miller, G., Ruberg, S., & McCormick, M. 2007. Anatomy of the
recurrent coastal plume in Lake Michigan: the importance of turbulence, suspended
sediments, and zebra mussels on nutrient and plankton distributions. Journal of
Geophysical Research, 112. doi:10.1029/2004JC002379.
Vanni, M. 1987. Effects of food availability and fish predation on a zooplankton community.
Ecological Monographs, 57, 61-68.
Walters, C. 2014. Has the Quesnel Lake sockeye stock collapsed due to accumulation of
nutrients from spawner carcasses? Fisheries Centre, University of British Columbia.
Welch, D. & Parsons, T. 1993. 13C and 15N values as indicators of trophic position and
competitive overlap for Pacific salmon (Oncorhynchus spp.). Fisheries Oceanography, 2, 11–
23.
Weslawski, J., Kwasniewski, S., & Wiktor, J. 1991. Winter in a Svalbard Fiord Ecosystem. Artic,
44(2), 115-123.
Wetzel, R. 2001. Limnology: Lake and River Ecosystems, 3rd ed. San Diego, CA: Academic Press,
2001.
Wipfli, M., Hudson, J., & Caouette, J. 1998. Influence of salmon carcasses on stream
productivity: response of biofilm and benthic macroinvertebrates in southeastern Alaska,
USA. Canadian Journal of Fisheries and Aquatic Science, 55, 1503–1511.
Wipfli, M. & Baxter, C. 2010. Linking ecosystems, food webs, and fish production: subsides in
salmonid watersheds. Fisheries, 35, 373-387.
Wotton, R. 2007. Do benthic biologists pay enough attention to aggregates formed in the water
column of streams and rivers? Journal of the North American Benthological Society, 26, 111.
Uchiyama, T., Finney, B., & Adkison, M. 2008. Effects of marine-derived nutrients on population
dynamics of sockeye salmon (Oncorhynchus nerka). Fisheries and Aquatic Science, 65, 16351648.

75

Appendix 1: Marine-Derived Nitrogen Load Equation
Delta 15N=3.45 & Amount of N (g) =152.58 (These numbers were provided by UC-Davis Isotope
Laboratory).
𝛿15𝑁 = 1000 × (

𝑅 𝑆𝑎𝑚𝑝𝑙𝑒−𝑅 𝑆𝑡𝑎𝑛𝑑𝑎𝑟𝑑
𝑅 𝑆𝑡𝑎𝑛𝑑𝑎𝑟𝑑

(1)

)

UC-Davis Air Standard vs Sample= 99.6337% 14N and 0.3663% 15N
0.3663%
99.6337%

(2)

= 0.003676

Equation
3.45 = 1000 × (

𝑅 𝑆𝑎𝑚𝑝𝑙𝑒−0.003676
0.003676

(3)

)

Divide both sides by 1000 to remove 1000 on right side of =
0.00345 =

𝑅 𝑆𝑎𝑚𝑝𝑙𝑒−0.003676

(4)

0.003676

Multiply both sides by 0.003676 to remove 0.003676 in denominator
(5)

0.00001268 = 𝑅 𝑆𝑎𝑚𝑝𝑙𝑒 − 0.003676
Add 0.003676 to both sides to remove 0.003676 to isolate R Sample
0.00368868 = 𝑅 𝑆𝑎𝑚𝑝𝑙𝑒
15

15

𝑅 𝑆𝑡𝑎𝑛𝑑𝑎𝑟𝑑 = 14𝑁 so 14𝑁 would equal the sample.
𝑁

𝑁

Overall equation
𝑅 𝑆𝑎𝑚𝑝𝑙𝑒 = 0.003676(1 + (
To find 15N%
15𝑁
= 0.00368868
14𝑁
76

𝛿15𝑁
))
1000

Multiply through by 14N to get 15N alone
15𝑁 = 0.00368868(14𝑁 )
Assuming 14N g + 15N g would equal my total N so…
100% 𝑁 = 0.00368868(14𝑁 ) + 14𝑁
Solve this equation
100% 𝑁 = 14𝑁 (0.00368868 + 1)
Divide both sides by (0.00368868 + 1) to get 14N alone
14𝑁 = 99.6324876%
Which means
15𝑁 = 0.3675124%
Overall equation
15𝑁 % = 100 −

100
𝑅 𝑆𝑎𝑚𝑝𝑙𝑒 + 1

Take the % and multiply by concentration of N to find the amount of 15N in g.

77

Appendix 2: R Script for Data Processing
September 4, 2017

Chapter 2
Packages Used
library(dplyr)
library(simmr)
library(tibble)
library(tidyr)
library(viridis)
library(ggplot2)
library(Stack)
library(PerformanceAnalytics)
library(lubridate)

R-code used to analyze marine-derived nitrogen and carbon using simmer mixing models.These
Scripts were taken from Parnell and Inger (2016) and edited and added to by Jacob Duros and
Samuel Albers on August 23, 2017.
Vegetation and salmon delta N/C values at site S5
mix = matrix(c(-27.04, -27.04, -27.04, -27.04, -27.04, -27.37, -27.37, -27.74
, -26.68, -26.68, -28.55, -27.74, -27.84,
-29.74, -29.5, -26.97, -27.88, -29.88, -25, -30.38, -22.86, -2
2.83, -22.88, -22.83, -22.88, -24.52, -22.01,
-22.15, -23.04, -23.74, -28.18, -27.02, -26.56, -29.19, -26.53
, -26.45, -26.11, -26.29, -26.43, -26.19, -25.95, -26.07,
-26.22, -26.95, -27.59, -26.53, -26.83, -26.74, -26.65, -26.48
, -26.31, -24.77, -25.54, -26.09, -26.75, 2.52, 2.89,
2.65, 1.79, 2.34, 1.9, 1.76, 2, 2.45, 2.18, 1.33, 1.6, 2.42, 3
.05, 3.45, 5.6, 4.72, 2.54, 5.33, 3.02, 6.83, 6.68, 6.68,
6.36, 6.36, 5.11, 7.6, 7.38, 6.54, 6.55, 2.15, 2.1, 1.99, 1.4,
1.64, 1.6, 1.76, 2.01, 2.11, 2.14, 2.4, 2.44, 2.41, 2.55,
1.95, 2.41, 2.29, 2.52, 2.14, 2.07, 2.24, 2.37, 2.63, 2.32, 2.
57), ncol=2, nrow=55)

Naming columns and sources
colnames(mix) = c('d13C','d15N')
s_names = c("Aquatic Veg.", "Terrestrial Veg.","Salmon")

Entering means and standard deviations calculated from data
s_means = matrix(c(-21.55, -30.26, -23.52, 2.33, 2.01, 11.43), ncol=2, nrow=3
)
s_sds = matrix(c(2.59, 2.02, 1.27, 0.57, 1.6, 0.76), ncol=2, nrow=3)

78

Grouping into Study Periods (Summer=1, Spring Freshet=2, Fall Spawn=3)
grp = as.integer(c(1, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 2, 3, 3, 3, 3, 3, 3
, 3, 3, 3, 3, 3, 3, 3, 3,
3, 3, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1
, 1, 2, 2, 2, 2, 2, 2))
simmr_groups = simmr_load(mixtures=mix,
source_names=s_names,
source_means=s_means,
source_sds=s_sds,
group=grp)

Plotting Mixing Model Output
plot(simmr_groups,group=1:3,xlab=expression(paste(delta^13, "C (\u2030)",sep=
"")),
ylab=expression(paste(delta^15, "N (\u2030)",sep="")),
title='',mix_name='SPM')

Running Simmr's statisical analysis
simmr_groups_out = simmr_mcmc(simmr_groups)
##
## Running for group 1
##
## Compiling model graph
##
Resolving undeclared variables
##
Allocating nodes
## Graph information:
##
Observed stochastic nodes: 52
##
Unobserved stochastic nodes: 5
##
Total graph size: 210
##
## Initializing model
##
##
## Running for group 2
##
## Compiling model graph
##
Resolving undeclared variables
##
Allocating nodes
## Graph information:
##
Observed stochastic nodes: 26
##
Unobserved stochastic nodes: 5
##
Total graph size: 158
##
## Initializing model
##
##

79

## Running for group 3
##
## Compiling model graph
##
Resolving undeclared variables
##
Allocating nodes
## Graph information:
##
Observed stochastic nodes: 32
##
Unobserved stochastic nodes: 5
##
Total graph size: 170
##
## Initializing model

Viewing statistics generated
summary(simmr_groups_out,type=c('quantiles','statistics'),group=c(1:3))
##
## Summary for group 1
##
2.5%
25%
50%
75% 97.5%
## Aquatic Veg.
0.320 0.369 0.394 0.419 0.471
## Terrestrial Veg. 0.504 0.551 0.574 0.596 0.639
## Salmon
0.009 0.022 0.031 0.041 0.063
## sd_d13C
0.011 0.118 0.256 0.428 0.869
## sd_d15N
0.006 0.068 0.144 0.253 0.529
##
mean
sd
## Aquatic Veg.
0.394 0.038
## Terrestrial Veg. 0.573 0.034
## Salmon
0.032 0.014
## sd_d13C
0.301 0.234
## sd_d15N
0.176 0.141
##
## Summary for group 2
##
2.5%
25%
50%
75% 97.5%
## Aquatic Veg.
0.245 0.330 0.369 0.409 0.495
## Terrestrial Veg. 0.471 0.549 0.584 0.621 0.691
## Salmon
0.012 0.030 0.043 0.059 0.095
## sd_d13C
0.020 0.253 0.535 0.922 1.893
## sd_d15N
0.010 0.108 0.235 0.400 0.880
##
mean
sd
## Aquatic Veg.
0.370 0.062
## Terrestrial Veg. 0.584 0.056
## Salmon
0.046 0.022
## sd_d13C
0.646 0.509
## sd_d15N
0.286 0.233
##
## Summary for group 3
##
2.5%
25%
50%
75% 97.5%
## Aquatic Veg.
0.123 0.252 0.318 0.381 0.513
## Terrestrial Veg. 0.127 0.247 0.306 0.367 0.489
## Salmon
0.285 0.346 0.375 0.406 0.472

80

## sd_d13C
1.917 2.497 2.881 3.361 4.520
## sd_d15N
0.952 1.286 1.499 1.770 2.404
##
mean
sd
## Aquatic Veg.
0.317 0.098
## Terrestrial Veg. 0.307 0.091
## Salmon
0.376 0.047
## sd_d13C
2.976 0.686
## sd_d15N
1.553 0.384

Arranging data into data frame.
str(simmr_groups_out)
## List of 2
## $ input :List of 12
##
..$ mixtures
: num [1:55, 1:2] -27 -27 -27 -27 -27 ...
##
.. ..- attr(*, "dimnames")=List of 2
##
.. .. ..$ : NULL
##
.. .. ..$ : chr [1:2] "d13C" "d15N"
##
..$ source_names
: chr [1:3] "Aquatic Veg." "Terrestrial Veg." "Sa
lmon"
##
..$ source_means
: num [1:3, 1:2] -21.55 -30.26 -23.52 2.33 2.01 .
..
##
..$ source_sds
: num [1:3, 1:2] 2.59 2.02 1.27 0.57 1.6 0.76
##
..$ correction_means
: num [1:3, 1:2] 0 0 0 0 0 0
##
..$ correction_sds
: num [1:3, 1:2] 0 0 0 0 0 0
##
..$ concentration_means: num [1:3, 1:2] 1 1 1 1 1 1
##
..$ group
: int [1:55] 1 1 1 1 1 1 1 2 2 2 ...
##
..$ n_obs
: int 55
##
..$ n_tracers
: int 2
##
..$ n_sources
: int 3
##
..$ n_groups
: int 3
##
..- attr(*, "class")= chr "simmr_input"
## $ output:List of 3
##
..$ :List of 4
##
.. ..$ : mcmc [1:1000, 1:5] 0.389 0.401 0.423 0.432 0.415 ...
##
.. .. ..- attr(*, "dimnames")=List of 2
##
.. .. .. ..$ : NULL
##
.. .. .. ..$ : chr [1:5] "Aquatic Veg." "Terrestrial Veg." "Salmon" "sd_
d13C" ...
##
.. .. ..- attr(*, "mcpar")= num [1:3] 1010 11000 10
##
.. ..$ : mcmc [1:1000, 1:5] 0.447 0.342 0.415 0.406 0.371 ...
##
.. .. ..- attr(*, "dimnames")=List of 2
##
.. .. .. ..$ : NULL
##
.. .. .. ..$ : chr [1:5] "Aquatic Veg." "Terrestrial Veg." "Salmon" "sd_
d13C" ...
##
.. .. ..- attr(*, "mcpar")= num [1:3] 1010 11000 10
##
.. ..$ : mcmc [1:1000, 1:5] 0.363 0.362 0.385 0.398 0.431 ...
##
.. .. ..- attr(*, "dimnames")=List of 2
##
.. .. .. ..$ : NULL

81

##
.. .. .. ..$ : chr [1:5] "Aquatic Veg." "Terrestrial Veg." "Salmon" "sd_
d13C" ...
##
.. .. ..- attr(*, "mcpar")= num [1:3] 1010 11000 10
##
.. ..$ : mcmc [1:1000, 1:5] 0.406 0.337 0.367 0.426 0.375 ...
##
.. .. ..- attr(*, "dimnames")=List of 2
##
.. .. .. ..$ : NULL
##
.. .. .. ..$ : chr [1:5] "Aquatic Veg." "Terrestrial Veg." "Salmon" "sd_
d13C" ...
##
.. .. ..- attr(*, "mcpar")= num [1:3] 1010 11000 10
##
.. ..- attr(*, "class")= chr "mcmc.list"
##
..$ :List of 4
##
.. ..$ : mcmc [1:1000, 1:5] 0.448 0.502 0.277 0.365 0.308 ...
##
.. .. ..- attr(*, "dimnames")=List of 2
##
.. .. .. ..$ : NULL
##
.. .. .. ..$ : chr [1:5] "Aquatic Veg." "Terrestrial Veg." "Salmon" "sd_
d13C" ...
##
.. .. ..- attr(*, "mcpar")= num [1:3] 1010 11000 10
##
.. ..$ : mcmc [1:1000, 1:5] 0.425 0.387 0.282 0.341 0.396 ...
##
.. .. ..- attr(*, "dimnames")=List of 2
##
.. .. .. ..$ : NULL
##
.. .. .. ..$ : chr [1:5] "Aquatic Veg." "Terrestrial Veg." "Salmon" "sd_
d13C" ...
##
.. .. ..- attr(*, "mcpar")= num [1:3] 1010 11000 10
##
.. ..$ : mcmc [1:1000, 1:5] 0.393 0.288 0.278 0.375 0.282 ...
##
.. .. ..- attr(*, "dimnames")=List of 2
##
.. .. .. ..$ : NULL
##
.. .. .. ..$ : chr [1:5] "Aquatic Veg." "Terrestrial Veg." "Salmon" "sd_
d13C" ...
##
.. .. ..- attr(*, "mcpar")= num [1:3] 1010 11000 10
##
.. ..$ : mcmc [1:1000, 1:5] 0.312 0.406 0.337 0.316 0.373 ...
##
.. .. ..- attr(*, "dimnames")=List of 2
##
.. .. .. ..$ : NULL
##
.. .. .. ..$ : chr [1:5] "Aquatic Veg." "Terrestrial Veg." "Salmon" "sd_
d13C" ...
##
.. .. ..- attr(*, "mcpar")= num [1:3] 1010 11000 10
##
.. ..- attr(*, "class")= chr "mcmc.list"
##
..$ :List of 4
##
.. ..$ : mcmc [1:1000, 1:5] 0.291 0.473 0.232 0.171 0.317 ...
##
.. .. ..- attr(*, "dimnames")=List of 2
##
.. .. .. ..$ : NULL
##
.. .. .. ..$ : chr [1:5] "Aquatic Veg." "Terrestrial Veg." "Salmon" "sd_
d13C" ...
##
.. .. ..- attr(*, "mcpar")= num [1:3] 1010 11000 10
##
.. ..$ : mcmc [1:1000, 1:5] 0.393 0.376 0.468 0.44 0.443 ...
##
.. .. ..- attr(*, "dimnames")=List of 2
##
.. .. .. ..$ : NULL
##
.. .. .. ..$ : chr [1:5] "Aquatic Veg." "Terrestrial Veg." "Salmon" "sd_
d13C" ...
##
.. .. ..- attr(*, "mcpar")= num [1:3] 1010 11000 10
##
.. ..$ : mcmc [1:1000, 1:5] 0.292 0.326 0.237 0.345 0.204 ...

82

##
.. .. ..- attr(*, "dimnames")=List of 2
##
.. .. .. ..$ : NULL
##
.. .. .. ..$ : chr [1:5] "Aquatic Veg." "Terrestrial Veg." "Salmon" "sd_
d13C" ...
##
.. .. ..- attr(*, "mcpar")= num [1:3] 1010 11000 10
##
.. ..$ : mcmc [1:1000, 1:5] 0.35 0.336 0.249 0.189 0.351 ...
##
.. .. ..- attr(*, "dimnames")=List of 2
##
.. .. .. ..$ : NULL
##
.. .. .. ..$ : chr [1:5] "Aquatic Veg." "Terrestrial Veg." "Salmon" "sd_
d13C" ...
##
.. .. ..- attr(*, "mcpar")= num [1:3] 1010 11000 10
##
.. ..- attr(*, "class")= chr "mcmc.list"
## - attr(*, "class")= chr "simmr_output"
out_all_1 = as_tibble(do.call(rbind, simmr_groups_out$output[[1]][,1:simmr_gr
oups_out$input$n_sources]))
out_all_1$Season = "Summer"
out_all_2 = as_tibble(do.call(rbind, simmr_groups_out$output[[2]][,1:simmr_gr
oups_out$input$n_sources]))
out_all_2$Season = "Spring"
out_all_3 = as_tibble(do.call(rbind, simmr_groups_out$output[[3]][,1:simmr_gr
oups_out$input$n_sources]))
out_all_3$Season = "Fall"
out_all = rbind(rbind(out_all_1, out_all_2), out_all_3)
df = gather(out_all, Source, Proportion, -Season)
df$Season_f=factor(df$Season, levels=c("Spring","Summer","Fall"))

Ploting Boxplots
ggplot(df, aes(x = Source, y = Proportion)) +
geom_boxplot(notch = TRUE) +
#coord_flip() +
#scale_colour_viridis(discrete = TRUE)+
theme_bw()+
facet_grid( . ~ Season_f)

Using SPM data to see if there are significant differences between study periods. All scripts
were written by Jacob Duros on August 23, 2017.
Inputting and Organizing Data
SPM <- read.csv("~/Desktop/Thesis Stuff/ISP.csv")
SPM<-stack(SPM)
names(SPM)<-c("SPM","Period")

Performing a 1-way ANOVA
SPM1<-aov(SPM~Period,data=SPM)
summary(SPM1)

83

##
Df Sum Sq Mean Sq F value
Pr(>F)
## Period
2 127.6
63.80
31.03 1.07e-09 ***
## Residuals
54 111.0
2.06
## --## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Performing a Tukey's HSD to determine significant differences among study periods.
tk1<-TukeyHSD(SPM1)
tk1
##
Tukey multiple comparisons of means
##
95% family-wise confidence level
##
## Fit: aov(formula = SPM ~ Period, data = SPM)
##
## $Period
##
diff
lwr
upr
p adj
## Spawn-Freshet -2.9173306 -4.038444 -1.7962169 0.0000002
## Summer-Freshet -3.3797887 -4.500902 -2.2586750 0.0000000
## Summer-Spawn
-0.4624581 -1.583572 0.6586556 0.5837308

Using organic and inorganic SPM data to see if there are significant differences between study
periods. All scripts were written by Jacob Duros on August 23, 2017.
Inputting Data
ISPS <- read.csv("~/Desktop/Thesis Stuff/ISP.csv")
OSPS <- read.csv("~/Desktop/Thesis Stuff/OSP.csv")

Organzing Inorganic SPM Data
ISP<-stack(ISPS)
names(ISP)<-c("IPM","Period")

Performing a 1-way ANOVA
ISP1<-aov(IPM~Period,data=ISP)
summary(ISP1)
##
Df Sum Sq Mean Sq F value
Pr(>F)
## Period
2 127.6
63.80
31.03 1.07e-09 ***
## Residuals
54 111.0
2.06
## --## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Performing a Tukey's HSD to determine significant differences among study periods.
tk2<-TukeyHSD(ISP1)
tk2

84

##
Tukey multiple comparisons of means
##
95% family-wise confidence level
##
## Fit: aov(formula = IPM ~ Period, data = ISP)
##
## $Period
##
diff
lwr
upr
p adj
## Spawn-Freshet -2.9173306 -4.038444 -1.7962169 0.0000002
## Summer-Freshet -3.3797887 -4.500902 -2.2586750 0.0000000
## Summer-Spawn
-0.4624581 -1.583572 0.6586556 0.5837308

Organzing organic SPM Data
OSP<-stack(OSPS)
names(OSP)<-c("OPM","Period")

Performing a 1-way ANOVA
OSP1<-aov(OPM~Period,data=OSP)
summary(OSP1)
##
Df Sum Sq Mean Sq F value
Pr(>F)
## Period
2 6.877
3.439
10.86 0.000109 ***
## Residuals
54 17.097
0.317
## --## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Performing a Tukey's HSD to determine significant differences among study periods.
tk3<-TukeyHSD(OSP1)
tk3
##
Tukey multiple comparisons of means
##
95% family-wise confidence level
##
## Fit: aov(formula = OPM ~ Period, data = OSP)
##
## $Period
##
diff
lwr
upr
p adj
## Spawn-Freshet -0.5354107 -0.9753675 -0.09545379 0.0134250
## Summer-Freshet -0.8403784 -1.2803353 -0.40042152 0.0000752
## Summer-Spawn
-0.3049677 -0.7449246 0.13498915 0.2257706

Using Isotope data to see if there are significant differences between study periods. All scripts
were written by Jacob Duros on August 23, 2017.
Inputting Data
Stable.Isotope <- read.csv("~/Desktop/Thesis Stuff/Stable Isotope.csv")
Stable.Isotope$Date<-mdy(Stable.Isotope$Date)

Organize data for Nitrogen
85

IsoN <- Stable.Isotope %>%
filter(Site=="S5") %>%
select(d15N, Group) %>%
rename(`Period`=Group)

Performing a 1-way ANOVA
NI<-aov(d15N~Period,data=IsoN)
summary(NI)
##
Df Sum Sq Mean Sq F value
Pr(>F)
## Period
2 112.63
56.32
43.37 6.08e-11 ***
## Residuals
42 54.54
1.30
## --## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Performing a Tukey's HSD to determine significant differences among study periods.
tk4<-TukeyHSD(NI)
tk4
##
Tukey multiple comparisons of means
##
95% family-wise confidence level
##
## Fit: aov(formula = d15N ~ Period, data = IsoN)
##
## $Period
##
diff
lwr
upr
p adj
## Spawn-Freshet
3.23976608 2.329184 4.150348 0.0000000
## Summer-Freshet 0.03546053 -1.131326 1.202247 0.9969990
## Summer-Spawn
-3.20430556 -4.380656 -2.027955 0.0000002

Organize data for Carbon
IsoC <- Stable.Isotope %>%
filter(Site=="S5") %>%
select(d13C, Group) %>%
rename(`Period`=Group)

Performing a 1-way ANOVA
CI<-aov(d13C~Period,data=IsoC)
summary(CI)
##
## Period
## Residuals

Df Sum Sq Mean Sq F value Pr(>F)
2 18.78
9.391
2.19 0.125
42 180.15
4.289

Performing a Tukey's HSD to determine significant differences among study periods.
tk5<-TukeyHSD(CI)
tk5

86

##
Tukey multiple comparisons of means
##
95% family-wise confidence level
##
## Fit: aov(formula = d13C ~ Period, data = IsoC)
##
## $Period
##
diff
lwr
upr
p adj
## Spawn-Freshet
1.3771053 -0.277873 3.0320835 0.1194931
## Summer-Freshet 0.2333553 -1.887273 2.3539836 0.9614030
## Summer-Spawn
-1.1437500 -3.281761 0.9942609 0.4032868

Using isotope loading data to see if there are significant differences between study periods. All
scripts were written by Jacob Duros on August 23, 2017.
Import data file
IsoAnova <- read.csv("~/Desktop/Thesis Stuff/IsoAnova.csv")

Organize nitrogen 15 loading data
ISO<-stack(IsoAnova)
names(ISO)<-c("N","Period")

Performing a 1-way ANOVA
ISO1<-aov(N~Period,data=ISO)
summary(ISO1)
##
Df Sum Sq Mean Sq F value
Pr(>F)
## Period
2 85.07
42.54
13.46 2.89e-05 ***
## Residuals
43 135.91
3.16
## --## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 11 observations deleted due to missingness

Performing a Tukey's HSD to determine significant differences among study periods.
tk6<-TukeyHSD(ISO1)
tk6
##
Tukey multiple comparisons of means
##
95% family-wise confidence level
##
## Fit: aov(formula = N ~ Period, data = ISO)
##
## $Period
##
diff
lwr
upr
p adj
## Spawn-Freshet -2.943265 -4.433864 -1.4526663 0.0000582
## Summer-Freshet -2.483722 -4.075041 -0.8924034 0.0013294
## Summer-Spawn
0.459543 -1.211891 2.1309775 0.7835296

87

Using isotope loading data and discharge data to see if there is a correlation betweent he two
variables. All scripts were written by Jacob Duros on August 23, 2017.
Organzing data
isoload<-read.csv("~/Desktop/Thesis Stuff/isoload.csv")
coriso <- isoload %>%
select(Flow,N15)

Using Kendall's Correlation method
cor(coriso, use="complete.obs", method="kendall")
##
Flow
N15
## Flow 1.0000000 0.6805221
## N15 0.6805221 1.0000000
chart.Correlation(coriso)

Using chlorophyll-a data to see if there are differences between study periods and study sites.
All scripts were written by Jacob Duros on August 23, 2017.
Importing data
Chlorophyll <- read.csv("~/Documents/Graduate Research/Lab Data/Chlorophyll.c
sv")
HFB <- read.csv("~/Desktop/Thesis Stuff/HFB.csv")

Organizing and formatting data
Chlorophyll$Date<-mdy(Chlorophyll$Date)
HFB$Date<-mdy(HFB$Date)
CHL <- Chlorophyll %>%
full_join(HFB, by=c("Date","Site")) %>%
select(Date,Site,Chlorophyll.x, Study.Period)%>%
dplyr::rename("Chlorophyll"=Chlorophyll.x) %>%
filter(Site %in% c("S1", "S2","S3","S4"))

Using fluorescence data to see if there are differences between study periods and study sites.
All scripts were written by Jacob Duros on August 23, 2017.
Importing data
Fluorescence <- read.csv("~/Desktop/Fluorescence.csv")

Organizing and formatting data
Fluorescence$Date<-mdy(Fluorescence$Date)
#S5
FLU5 <- Fluorescence %>%
filter(Site %in% c("S5"))

88

#S3
FLU3 <- Fluorescence %>%
filter(Site %in% c("S3"))
#S1
FLU1 <- Fluorescence %>%
filter(Site %in% c("S1"))

S5 Performing a 1-way ANOVA
FLUANOVA5<-aov(lm(fluorescence ~ Period, FLU5))
summary(FLUANOVA5)
##
Df Sum Sq Mean Sq F value Pr(>F)
## Period
2 3.108 1.5541
5.489 0.00736 **
## Residuals
45 12.741 0.2831
## --## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

S5 Performing a Tukey's HSD to determine significant differences among study periods.
TukeyHSD(FLUANOVA5)
##
Tukey multiple comparisons of means
##
95% family-wise confidence level
##
## Fit: aov(formula = lm(fluorescence ~ Period, FLU5))
##
## $Period
##
diff
lwr
upr
p adj
## Freshet-Fall
0.53136697 0.1215039 0.94123001 0.0081799
## Summer-Fall
0.02246989 -0.5099578 0.55489760 0.9942496
## Summer-Freshet -0.50889708 -1.0568734 0.03907923 0.0735650

S3 Performing a 1-way ANOVA
FLUANOVA3<-aov(lm(fluorescence ~ Period, FLU3))
summary(FLUANOVA3)
##
Df Sum Sq Mean Sq F value Pr(>F)
## Period
2 0.595 0.29741
3.131 0.0558 .
## Residuals
36 3.420 0.09499
## --## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

S3 Performing a Tukey's HSD to determine significant differences among study periods.
TukeyHSD(FLUANOVA3)
##
Tukey multiple comparisons of means
##
95% family-wise confidence level
##
## Fit: aov(formula = lm(fluorescence ~ Period, FLU3))

89

##
## $Period
##
diff
lwr
upr
p adj
## Freshet-Fall
-0.2125573 -0.4876402 0.06252550 0.1565530
## Summer-Fall
-0.2985613 -0.6161996 0.01907696 0.0689931
## Summer-Freshet -0.0860040 -0.4036423 0.23163430 0.7869023

S1 Performing a 1-way ANOVA
FLUANOVA1<-aov(lm(fluorescence ~ Period, FLU1))
summary(FLUANOVA1)
##
Df Sum Sq Mean Sq F value Pr(>F)
## Period
2 0.2666 0.13331
2.998 0.0632 .
## Residuals
34 1.5120 0.04447
## --## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

S1 Performing a Tukey's HSD to determine significant differences among study periods.
TukeyHSD(FLUANOVA1)
##
Tukey multiple comparisons of means
##
95% family-wise confidence level
##
## Fit: aov(formula = lm(fluorescence ~ Period, FLU1))
##
## $Period
##
diff
lwr
upr
p adj
## Freshet-Fall
-0.15681081 -0.3488377 0.03521605 0.1273443
## Summer-Fall
-0.19454117 -0.4207689 0.03168661 0.1033620
## Summer-Freshet -0.03773036 -0.2667512 0.19129046 0.9143121

Using phosphorus data to see if there are differences between study periods. All scripts were
written by Jacob Duros on August 23, 2017.
Importing data
PLevels <- read.csv("~/Documents/Graduate Research/Thesis/PLevels.csv")
P_Work <- read.csv("~/Documents/Graduate Research/Thesis/P Work.csv")

Organizing and formatting data
P_Work$Date<-mdy(P_Work$Date)
Phos <- P_Work %>%
full_join(PLevels, by=c("ID")) %>%
select(-ID)%>%
filter(Site=="S5")

Data frame was exported into excel to add DP data column
write.csv(Phos, file="Phos.csv")

90

File imported back into R
PhosP <- read.csv("~/Desktop/Thesis Stuff/PhosP.csv")

Organizing and formatting data
PhosP$Date<-mdy(PhosP$Date)
Phos1 <- PhosP %>%
gather(variable, value, -Date)
Phos1<-Phos1 %>%
filter(variable=="Total")

Performing a 1-way ANOVA
PAOV <- read.csv("~/Desktop/Thesis Stuff/PAOV.csv")
PhosANOVA<-aov(lm(Concentration ~ Period * Form, PAOV))
summary(PhosANOVA)
##
Df
Sum Sq
Mean Sq F value
Pr(>F)
## Period
2 0.0004713 0.0002357
8.886 0.000318 ***
## Form
1 0.0004557 0.0004557 17.183 8.14e-05 ***
## Period:Form 2 0.0000427 0.0000214
0.806 0.450113
## Residuals
83 0.0022011 0.0000265
## --## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Performing a Tukey's HSD to determine significant differences among study periods.
TukeyHSD(PhosANOVA)
##
Tukey multiple comparisons of means
##
95% family-wise confidence level
##
## Fit: aov(formula = lm(Concentration ~ Period * Form, PAOV))
##
## $Period
##
diff
lwr
upr
p adj
## Spawn-Freshet
0.0044447619 0.001387006 0.007502518 0.0023724
## Summer-Freshet -0.0008952381 -0.004152309 0.002361833 0.7895239
## Summer-Spawn
-0.0053400000 -0.008705669 -0.001974331 0.0008338
##
## $Form
##
diff
lwr
upr
p adj
## total-dissovled 0.004566646 0.002371678 0.006761614 8.36e-05
##
## $`Period:Form`
##
diff
lwr
upr
## Spawn:dissovled-Freshet:dissovled
0.0054095238 0.0004240619 0.010394986
## Summer:dissovled-Freshet:dissovled 0.0001452381 -0.0050383983 0.005328875
## Freshet:total-Freshet:dissovled
0.0063452381 0.0011616017 0.011528875
## Spawn:total-Freshet:dissovled
0.0087809524 0.0035973160 0.013964589

91

## Summer:total-Freshet:dissovled
0.0037395238 -0.0020327294 0.009511777
## Summer:dissovled-Spawn:dissovled
-0.0052642857 -0.0107623624 0.000233791
## Freshet:total-Spawn:dissovled
0.0009357143 -0.0045623624 0.006433791
## Spawn:total-Spawn:dissovled
0.0033714286 -0.0021266481 0.008869505
## Summer:total-Spawn:dissovled
-0.0016700000 -0.0077262090 0.004386209
## Freshet:total-Summer:dissovled
0.0062000000 0.0005216108 0.011878389
## Spawn:total-Summer:dissovled
0.0086357143 0.0029573251 0.014314103
## Summer:total-Summer:dissovled
0.0035942857 -0.0026260780 0.009814649
## Spawn:total-Freshet:total
0.0024357143 -0.0032426749 0.008114103
## Summer:total-Freshet:total
-0.0026057143 -0.0088260780 0.003614649
## Summer:total-Spawn:total
-0.0050414286 -0.0112617923 0.001178935
##
p adj
## Spawn:dissovled-Freshet:dissovled 0.0254926
## Summer:dissovled-Freshet:dissovled 0.9999995
## Freshet:total-Freshet:dissovled
0.0075875
## Spawn:total-Freshet:dissovled
0.0000571
## Summer:total-Freshet:dissovled
0.4155433
## Summer:dissovled-Spawn:dissovled
0.0685494
## Freshet:total-Spawn:dissovled
0.9961739
## Spawn:total-Spawn:dissovled
0.4784423
## Summer:total-Spawn:dissovled
0.9659459
## Freshet:total-Summer:dissovled
0.0241000
## Spawn:total-Summer:dissovled
0.0003903
## Summer:total-Summer:dissovled
0.5450094
## Spawn:total-Freshet:total
0.8100450
## Summer:total-Freshet:total
0.8248692
## Summer:total-Spawn:total
0.1809817

Performing a correlation using phosphorus, chlorophyll-a, discharge, isotope, and river
temperature data to see if there are trends among variables. This data was also used to
perform the AIC statistical analysis below to see what variables drive the delivery of delta 15N
and the load of 15N. All scripts were written by Jacob Duros on August 23, 2017.
Importing data sets that haven't already been imported
Hobo_Complete <- read.csv("~/Documents/Graduate Research/Hobo/Hobo Complete.c
sv")
#Discharge
HFR.Daily.Flow<-read.csv("~/Documents/untitled folder/untitled folder 3/Water
Flow Data/HFR Daily Flow.csv")
Thesis.data<-read.csv("~/Desktop/Thesis Stuff/Thesis Data.csv")
Hobo_Complete$Date<-mdy(Hobo_Complete$Date)
HFR.Daily.Flow$Date<-mdy(HFR.Daily.Flow$Date)
Thesis.data$Date<-mdy(Thesis.data$Date)

Organizing and pulling all data into same dataframe
FD<-HFR.Daily.Flow%>%
select(Date,Mean)%>%

92

full_join(Hobo_Complete, by=c("Date")) %>%
full_join(Stable.Isotope, by=c("Date"))%>%
full_join(Thesis.data, by=c("Date"))%>%
select(Date, Mean, RiverTemp,d15N,SPM.mg.L,N.Amount..ug.)%>%
full_join(Phos1, by=c("Date"))%>%
full_join(CHL, by=c("Date"))%>%
distinct(Date, .keep_all = TRUE)%>%
select(-Date,-Chlorophyll)%>%
rename("Flow"=Mean,"SPM"=SPM.mg.L,"NLoad"=N.Amount..ug.)

Creating linear models to see what drives the delivery of delta 15N to HFB.
1st linear model includes variables: Discharge, River Temp, SPM, TP, and Study Period. 1st order
terms only.
lm1<-lm(FD$d15N~FD$Flow+FD$RiverTemp+FD$SPM+FD$TP+FD$Study.Period)
summary(lm1)
##
## Call:
## lm(formula = FD$d15N ~ FD$Flow + FD$RiverTemp + FD$SPM + FD$TP +
##
FD$Study.Period)
##
## Residuals:
##
Min
1Q
Median
3Q
Max
## -1.76512 -0.23015 -0.00297 0.38294 1.18630
##
## Coefficients:
##
Estimate Std. Error t value Pr(>|t|)
## (Intercept)
-0.37431
1.31105 -0.286 0.77976
## FD$Flow
0.03234
0.01510
2.142 0.05175 .
## FD$RiverTemp
-0.04320
0.05734 -0.753 0.46468
## FD$SPM
0.03807
0.06547
0.581 0.57088
## FD$TP
23.07878
29.07055
0.794 0.44151
## FD$Study.PeriodSpawn
2.32161
0.76561
3.032 0.00962 **
## FD$Study.PeriodSummer 2.23920
1.01357
2.209 0.04572 *
## --## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.7668 on 13 degrees of freedom
##
(343 observations deleted due to missingness)
## Multiple R-squared: 0.566, Adjusted R-squared: 0.3657
## F-statistic: 2.826 on 6 and 13 DF, p-value: 0.05497

2nd linear model includes variables: Discharge, SPM, and Study Period. 1st order terms only.
lm2<-lm(FD$d15N~FD$Flow+FD$SPM+FD$Study.Period)
summary(lm2)
##
## Call:

93

## lm(formula = FD$d15N ~ FD$Flow + FD$SPM + FD$Study.Period)
##
## Residuals:
##
Min
1Q
Median
3Q
Max
## -1.76584 -0.31446 0.06707 0.41460 1.08260
##
## Coefficients:
##
Estimate Std. Error t value Pr(>|t|)
## (Intercept)
-0.092860
0.579138 -0.160 0.873901
## FD$Flow
0.024062
0.008633
2.787 0.010008 *
## FD$SPM
0.014761
0.010186
1.449 0.159718
## FD$Study.PeriodSpawn
2.164635
0.471438
4.592 0.000107 ***
## FD$Study.PeriodSummer 1.668899
0.507773
3.287 0.003002 **
## --## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.6547 on 25 degrees of freedom
##
(333 observations deleted due to missingness)
## Multiple R-squared: 0.4908, Adjusted R-squared: 0.4093
## F-statistic: 6.024 on 4 and 25 DF, p-value: 0.001548

3rd linear model includes variables: Discharge, SPM, and Study Period. 1st and 2nd order terms
are used.
lm3<-lm(FD$d15N~(FD$Flow+FD$SPM+FD$Study.Period)^2)
summary(lm3)
##
## Call:
## lm(formula = FD$d15N ~ (FD$Flow + FD$SPM + FD$Study.Period)^2)
##
## Residuals:
##
Min
1Q
Median
3Q
Max
## -1.75745 -0.40912 0.06429 0.40796 1.06129
##
## Coefficients:
##
Estimate Std. Error t value Pr(>|t|)
## (Intercept)
0.240941
1.696565
0.142
0.888
## FD$Flow
0.016118
0.037439
0.431
0.671
## FD$SPM
-0.060180
0.338559 -0.178
0.861
## FD$Study.PeriodSpawn
1.801568
1.686708
1.068
0.298
## FD$Study.PeriodSummer
2.643794
2.077979
1.272
0.218
## FD$Flow:FD$SPM
0.001918
0.008574
0.224
0.825
## FD$Flow:FD$Study.PeriodSpawn
0.010856
0.036204
0.300
0.767
## FD$Flow:FD$Study.PeriodSummer -0.081255
0.101828 -0.798
0.434
## FD$SPM:FD$Study.PeriodSpawn
0.038245
0.258042
0.148
0.884
## FD$SPM:FD$Study.PeriodSummer
0.364011
0.634177
0.574
0.572
##
## Residual standard error: 0.7125 on 20 degrees of freedom
##
(333 observations deleted due to missingness)

94

## Multiple R-squared: 0.5175, Adjusted R-squared: 0.3004
## F-statistic: 2.384 on 9 and 20 DF, p-value: 0.05074

4th linear model includes variables: Discharge, SPM, which are interacting with Study Period
which was the only significant term in the 3rd model.
lm4<-lm(FD$d15N~(FD$Flow*FD$Study.Period)+(FD$SPM*FD$Study.Period))#first ord
er with interaction between signifcant from above
summary(lm4)
##
## Call:
## lm(formula = FD$d15N ~ (FD$Flow * FD$Study.Period) + (FD$SPM *
##
FD$Study.Period))
##
## Residuals:
##
Min
1Q
Median
3Q
Max
## -1.75966 -0.39337 0.02412 0.40694 1.05627
##
## Coefficients:
##
Estimate Std. Error t value Pr(>|t|)
## (Intercept)
-0.104752
0.684539 -0.153
0.8798
## FD$Flow
0.024152
0.010351
2.333
0.0297 *
## FD$Study.PeriodSpawn
2.128367
0.824086
2.583
0.0174 *
## FD$Study.PeriodSummer
2.950071
1.527524
1.931
0.0671 .
## FD$SPM
0.015522
0.011074
1.402
0.1756
## FD$Flow:FD$Study.PeriodSpawn
0.004849
0.023727
0.204
0.8400
## FD$Flow:FD$Study.PeriodSummer -0.087914
0.095152 -0.924
0.3660
## FD$Study.PeriodSpawn:FD$SPM
-0.017744
0.061468 -0.289
0.7757
## FD$Study.PeriodSummer:FD$SPM
0.339249
0.610157
0.556
0.5841
## --## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.6962 on 21 degrees of freedom
##
(333 observations deleted due to missingness)
## Multiple R-squared: 0.5163, Adjusted R-squared: 0.332
## F-statistic: 2.802 on 8 and 21 DF, p-value: 0.02785

Performing an AIC on linear models 1, 2, 3, & 4.
AIC(lm1,lm2,lm3,lm4)
##
df
AIC
## lm1 8 53.51926
## lm2 6 66.25595
## lm3 11 74.63790
## lm4 10 72.71289

Creating linear models to see what drives the delivery of the load of 15N to HFB.

95

5th linear model includes variables: Discharge, River Temp, SPM, TP, and Study Period. 1st
order terms only.
lm5<-lm(FD$NLoad~FD$Flow+FD$RiverTemp+FD$SPM+FD$TP+FD$Study.Period)
summary(lm5)
##
## Call:
## lm(formula = FD$NLoad ~ FD$Flow + FD$RiverTemp + FD$SPM + FD$TP +
##
FD$Study.Period)
##
## Residuals:
##
Min
1Q
Median
3Q
Max
## -130.839 -64.389
4.416
27.570 230.098
##
## Coefficients:
##
Estimate Std. Error t value Pr(>|t|)
## (Intercept)
394.168
167.603
2.352
0.0351 *
## FD$Flow
-4.211
1.930 -2.181
0.0481 *
## FD$RiverTemp
2.594
7.331
0.354
0.7291
## FD$SPM
-1.139
8.370 -0.136
0.8938
## FD$TP
726.663
3716.341
0.196
0.8480
## FD$Study.PeriodSpawn
-78.445
97.875 -0.801
0.4373
## FD$Study.PeriodSummer -232.485
129.574 -1.794
0.0961 .
## --## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 98.02 on 13 degrees of freedom
##
(343 observations deleted due to missingness)
## Multiple R-squared: 0.4822, Adjusted R-squared: 0.2432
## F-statistic: 2.018 on 6 and 13 DF, p-value: 0.1359

6th linear model includes variables: Discharge, SPM, and Study Period. 1st order terms only.
lm6<-lm(FD$NLoad~FD$Flow+FD$SPM+FD$Study.Period)
summary(lm6)
##
## Call:
## lm(formula = FD$NLoad ~ FD$Flow + FD$SPM + FD$Study.Period)
##
## Residuals:
##
Min
1Q
Median
3Q
Max
## -167.518 -60.850
5.248
35.821 240.319
##
## Coefficients:
##
Estimate Std. Error t value Pr(>|t|)
## (Intercept)
400.9295
100.4172
3.993 0.000505 ***
## FD$Flow
-3.6205
1.4969 -2.419 0.023191 *
## FD$SPM
-0.5429
1.7662 -0.307 0.761084
## FD$Study.PeriodSpawn -60.0858
81.7431 -0.735 0.469145

96

## FD$Study.PeriodSummer -60.7687
88.0433 -0.690 0.496418
## --## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 113.5 on 25 degrees of freedom
##
(333 observations deleted due to missingness)
## Multiple R-squared: 0.3073, Adjusted R-squared: 0.1964
## F-statistic: 2.772 on 4 and 25 DF, p-value: 0.0492

7th linear model includes variables: Discharge, SPM, and Study Period. 1st and 2nd order terms
are used.
lm7<-lm(FD$NLoad~(FD$Flow+FD$SPM+FD$Study.Period)^2)#all 1st and 2nd order te
rms
summary(lm7)
##
## Call:
## lm(formula = FD$NLoad ~ (FD$Flow + FD$SPM + FD$Study.Period)^2)
##
## Residuals:
##
Min
1Q
Median
3Q
Max
## -227.986 -45.666
-7.244
36.211 241.226
##
## Coefficients:
##
Estimate Std. Error t value Pr(>|t|)
## (Intercept)
288.9964
287.7664
1.004
0.327
## FD$Flow
-1.0805
6.3504 -0.170
0.867
## FD$SPM
21.9452
57.4254
0.382
0.706
## FD$Study.PeriodSpawn
20.2377
286.0945
0.071
0.944
## FD$Study.PeriodSummer
183.4968
352.4609
0.521
0.608
## FD$Flow:FD$SPM
-0.5698
1.4542 -0.392
0.699
## FD$Flow:FD$Study.PeriodSpawn
-0.9348
6.1408 -0.152
0.881
## FD$Flow:FD$Study.PeriodSummer -3.3715
17.2717 -0.195
0.847
## FD$SPM:FD$Study.PeriodSpawn
-12.6004
43.7684 -0.288
0.776
## FD$SPM:FD$Study.PeriodSummer -87.5118
107.5673 -0.814
0.425
##
## Residual standard error: 120.9 on 20 degrees of freedom
##
(333 observations deleted due to missingness)
## Multiple R-squared: 0.3719, Adjusted R-squared: 0.08925
## F-statistic: 1.316 on 9 and 20 DF, p-value: 0.2895

8th linear model includes variables: Discharge, SPM, which are interacting with Study Period.
lm8<-lm(FD$NLoad~(FD$Flow*FD$SPM)+(FD$Flow*FD$Study.Period)+(FD$SPM*FD$Study.
Period))#first order with interaction between signifcant from above
summary(lm8)
##
## Call:

97

## lm(formula = FD$NLoad ~ (FD$Flow * FD$SPM) + (FD$Flow * FD$Study.Period) +
##
(FD$SPM * FD$Study.Period))
##
## Residuals:
##
Min
1Q
Median
3Q
Max
## -227.986 -45.666
-7.244
36.211 241.226
##
## Coefficients:
##
Estimate Std. Error t value Pr(>|t|)
## (Intercept)
288.9964
287.7664
1.004
0.327
## FD$Flow
-1.0805
6.3504 -0.170
0.867
## FD$SPM
21.9452
57.4254
0.382
0.706
## FD$Study.PeriodSpawn
20.2377
286.0945
0.071
0.944
## FD$Study.PeriodSummer
183.4968
352.4609
0.521
0.608
## FD$Flow:FD$SPM
-0.5698
1.4542 -0.392
0.699
## FD$Flow:FD$Study.PeriodSpawn
-0.9348
6.1408 -0.152
0.881
## FD$Flow:FD$Study.PeriodSummer -3.3715
17.2717 -0.195
0.847
## FD$SPM:FD$Study.PeriodSpawn
-12.6004
43.7684 -0.288
0.776
## FD$SPM:FD$Study.PeriodSummer -87.5118
107.5673 -0.814
0.425
##
## Residual standard error: 120.9 on 20 degrees of freedom
##
(333 observations deleted due to missingness)
## Multiple R-squared: 0.3719, Adjusted R-squared: 0.08925
## F-statistic: 1.316 on 9 and 20 DF, p-value: 0.2895

Performing an AIC on linear models 5, 6, 7, & 8.
AIC(lm5,lm6,lm7,lm8)
##
df
AIC
## lm5 8 247.5500
## lm6 6 375.5889
## lm7 11 382.6505
## lm8 11 382.6505

98