Influences of the Environment and Plankton Community

INFLUENCES OF THE ENVIRONMENT AND PLANKTON COMMUNITY

INTERACTONS ON TOXIC CYANOBACTERIAL BLOOMS IN

VANCOUVER LAKE, WASHINGTON, A TEMPERATE

SHALLOW FRESHWATER SYSTEM

TAMMY ANNE LEE

A dissertation submitted in partial fulfillment of the requirements for the degree of

DOCTOR OF PHILOSOPHY

WASHINGTON STATE UNIVERSITY School of the Environment

DECEMBER 2015

To the Faculty of Washington State University:

The members of the Committee appointed to examine the dissertation of

TAMMY ANNE LEE find it satisfactory and recommend that it be accepted.

Stephen M. Bollens, Ph.D., Co-Chair

Gretchen Rollwagen-Bollens, Ph.D., Co-Chair

Steven Sylvester, Ph.D.

Brian N. Tissot, Ph.D.

ACKNOWLEDGEMENT

Many thanks to my graduate committee, Drs. Steve Bollens, Gretchen Rollwagen-

Bollens, Steve Sylvester, and Brian Tissot for their invaluable support and patience. Thanks to the Vancouver Lake Sailing Club for their open access to their docks over the past several years.

Special thanks to Molli MacDonald for her expertise and training of phytoplankton identification, Alejandro Gonzalez for zooplankton identification, Jennifer Duer Boyer for her time and effort collecting samples from Vancouver Lake, and everyone else involved with

Vancouver Lake: Julie Zimmerman, Josh Emerson, Vanessa Rose, and friends and past lab members who have helped with the sampling at Vancouver Lake, rain or shine. Thanks are also due to Ruth Philips, Josh Faber-Hammond, and Jenefer DeKoning for their unyielding support, expertise, and sharing bench space with the molecular work; Andy Ford for his initial guidance and expertise in system dynamics modeling; and my writing group Rebecca Bellmore, Bridget

Deemer, Whitney Hasset, Mailea Miller-Pierce, James Moore, Felicia and Olmeta-Schult.

Additional thanks to my family: Sangkee, Heok, Norman, and Leah Lee, and close friends: Dan

De Vriend and Kara Goodwin for their emotional support. Much of this work could not have been done without the financial support of USGS through the State of Washington Water

Research Center (grant no. 06HQGR0126), the Vancouver Lake Watershed Partnership, NSF

ULTRA-EX (grant no. 09-48983), GK-12 Fellowship from NSF STEM Fellows (grant no.

DGE07-42561), and the Robert Lane Fellowship.

iii

INFLUENCES OF THE ENVIRONMENT AND PLANKTON COMMUNITY

INTERACTONS ON TOXIC CYANOBACTERIAL BLOOMS IN

VANCOUVER LAKE, WASHINGTON, A TEMPERATE

SHALLOW FRESHWATER SYSTEM

Abstract

by Tammy Anne Lee, Ph.D. Washington State University December 2015

Co-Chairs: Stephen M. Bollens, Gretchen Rollwagen-Bollens

Occurrences of cyanobacterial blooms are in freshwater systems are increasing in frequency and intensity largely in response to urbanization of landscapes, eutrophication, and climate change. Cyanobacterial blooms negatively affect water quality which leads to a broad range of environmental, social, and economic concerns. In particular, cyanobacteria are known to produce a suite of toxins that have been linked to changes to the aquatic food web, small animal mortality and illness, and adverse health risks to humans. Vancouver Lake, located in southwest

Washington state, is a tidally influenced shallow freshwater lake that exhibits annual summer cyanobacterial blooms that have been an on-going concern for public health and natural resource managers. Thus, the purpose of this project was to investigate the biotic and abiotic interactions associated with bloom events in a shallow, freshwater system. The main objectives were: 1) analyze phytoplankton community dynamics with an emphasis on the cyanobacterial community in relation to water quality factors; 2) identify and quantify toxin producing cyanobacterial

iv populations in relation to water quality factors; 3) assess the potential effects of cyanobacterial blooms on zooplankton community dynamics; and 4) develop a model on the potential effects of wind-driven waves on internal phosphorus loading as a potential mechanism contributing to seasonal cyanobacterial blooms. Our findings suggest that nutrients, dissolved inorganic nitrogen and dissolved inorganic phosphorus, significantly influence cyanobacterial blooms dynamics, and more specifically with toxin producing cyanobacteria. In relation to zooplankton community dynamics, we found that while cyanobacterial blooms may have some influence, but also non- native invasive crustacean zooplankton may interact with cyanobacterial blooms affecting the summer zooplankton community. In spite of Vancouver Lake being a model large shallow lake highly susceptible to wind driven sediment resuspension, simulations of seasonal orthophosphate availability did not support observed measurements, suggesting other mechanisms such as redox related processes and bioturbation, should be examined in assessing potential management considerations for restoring Vancouver Lake.

TABLE OF CONTENTS

Page

ACKNOWLEDGEMENTS ...... iii

ABSTRACT ...... iv

CHAPTER

1. INTRODUCTION ...... 1

2. THE INFLUENCE OF WATER QUALITY VARIABLES ON

CYANOBACTERIAL BLOOMS AND PHYTOPLANKTON COMMUNITY

COMPOSITION IN A SHALLOW TEMPERATE LAKE ...... 4

Abstract ...... 4

Introduction ...... 5

Methods...... 8

Results ...... 14

Discussion ...... 20

References ...... 31

Tables and Figures ...... 41

Supplementary Materials ...... 50

3. ENVIRONMENTAL INFLUENCE ON CYANOBACTERIA ABUNDANCE

AND MICROCYSTIN TOXIN PRODUCTION IN A SHALLOW

TEMPERATE LAKE ...... 52

Abstract ...... 52

Introduction ...... 53

Methods...... 55

Results ...... 62

Discussion ...... 66

References ...... 72

Figures...... 80

4. THE EFFECTS OF EUTROPHICATION AND INVASIVE SPECIES ON

ZOOPLANKTON COMMUNITY DYNAMICS IN A SHALLOW

TEMPERATE LAKE ...... 85

Abstract ...... 85

Introduction ...... 86

Methods...... 89

Results ...... 94

Discussion ...... 100

References ...... 111

Tables and Figures ...... 122

Supplementary Materials ...... 131

5. OVERESTIMATING THE POTENTIAL INFLUENCE OF INTERNAL

PHOSPHORUS LOADING DUE TO WIND-DRIVEN SEDIMENT

RESUSPENSION IN A TIDALLY INFLUENCED SHALLOW FRESHWATER

LAKE: A MODELING APPROACH ...... 133

Abstract ...... 133

Introduction ...... 134

Methods...... 147

Results ...... 142

vii

Discussion ...... 144

References ...... 150

Tables and Figures ...... 156

Supplementary Materials ...... 163

6. CONCLUSION ...... 164

viii

CHAPTER ONE

INTRODUCTION

In freshwater systems, occurrences of cyanobacterial blooms have been increasing in frequency and intensity. Cyanobacterial blooms negatively affect water quality which leads to a broad range of environmental, social, and economic concerns. For example, blooms increase oxygen demand which may lead to localized hypoxia or anoxia and fish kills. Surface blooms can also block light from reaching other primary producers, such as macrophytes, altering higher trophic levels and water quality. Social and economic issue include negative effects on recreational opportunities due to closure of affected areas, fish kills which may affect local fisheries, and increased water treatment costs.

A suite of interacting environmental factors are thought to be responsible for the cause, duration, and decay of cyanobacterial blooms. Lakes have been shown to be sensitive to excess pollution in the form of increased nutrient inputs, hydrodynamics (i.e. wind-drive waves, turbulence, and stratification), physical factors (i.e. light availability and temperature), and the effects of higher trophic levels. Understanding the significance of which factors contribute to cyanobacterial blooms is important in determining management strategies for preventing blooms and restoring lake health.

Cyanobacteria are known to produce a suite of secondary metabolites that include hepatotoxins, neurotoxins, and dermatotoxic compounds. These toxins have been linked to decreased water quality, inter- and intra-specific competition for resources, and detrimental effects on higher aquatic trophic levels including small animal mortality and illness and adverse health risks to humans. The same environmental factors influencing cyanobacterial blooms can also affect cyanobacteria toxin production.

Vancouver Lake is a tidally influence shallow freshwater lake located in southwest

Washington state. It is a local and regional recreational destination and serves as critical habitat for wildlife and migratory waterfowl. Cyanobacterial blooms have been officially documented there since 1971, although anecdotal evidence suggests blooms have occurred prior to 1971.

More recently, the lake has been subject to public closure due to intense bloom conditions and high toxin levels. Thus, Vancouver Lake serves as a model system for studying cyanobacterial blooms.

The purpose of my dissertation was to investigate and model biotic and abiotic interactions associated with cyanobacterial blooms. Specifically, the following objectives outline the main goals of this project and represent individual chapters: 1) investigate how water quality variables influence cyanobacterial blooms and phytoplankton community dynamics; 2) identify and quantify toxin and non-toxin producing cyanobacteria, and assess how environmental factors influence the cyanobacteria community and toxin production; 3) examine how cyanobacterial blooms and other abiotic and biotic variables influence zooplankton community dynamics; and

4) develop a mechanistic model to assess how wind driven waves in a shallow lake contributes to sediment resuspension and internal phosphorus loading.

The overall results from this project provide new information regarding toxic cyanobacterial blooms and other environmental stressors in shallow freshwater lake systems. In particular, identifying abiotic and biotic factors associated with cyanobacterial blooms will provide critical insight for public health officials and natural resource managers to make informed decisions and implement management strategies. Results from this project are also compared to other shallow freshwater systems from a regional to global scale to help elucidate broader patterns and underlying causes of cyanobacterial blooms.

The alternative dissertation format, approved by the Graduate School and Washington

State University, has been used for each chapter. The second chapter was published in

Environmental Monitoring and Assessment; the third chapter was published in Ecotoxicology and Environmental Safety; the fourth chapter is currently under review at Fundamental and

Applied Limnology; and the fifth chapter has been submitted to Lake and Reservoir

Management. The writing and formatting styles in each chapter reflect each journal’s stylistic requirements. Stephen Bollens and Gretchen Rollwagen-Bollens are co-authors on each manuscript as the execution and funding of each project was through their lab and as my co- advisors have been of assistance throughout the writing and editing process. Joshua Faber-

Hammond was a co-author for chapter three for his assistance with method development, genetic sequencing knowledge, and editing. Joshua Emerson was a co-author for chapter four for his assistance with data collection and identification of zooplankton, and editing assistance.

CHAPTER TWO

THE INFLUENCE OF WATER QUALITY VARIABLES ON CYANOBACTERIAL

BLOOMS AND PHYTOPLANKTON COMMUNITY COMPOSITION IN A

SHALLOW TEMPERATE LAKE

Tammy A. Lee*1, Gretchen Rollwagen-Bollens1,2, Stephen M. Bollens1,2

1School of the Environment and 2School of Biological Sciences, Washington State University,

14204 NE Salmon Creek Avenue, Vancouver, Washington 98686, USA

*Corresponding author: [email protected], phone: (360) 546-9134, fax: (360) 546-9064

Abstract

Cyanobacterial blooms and their detrimental effects on water quality have become a worldwide problem. Vancouver Lake, a tidally influenced shallow temperate freshwater lake in Washington state, USA, exhibits annual summer cyanobacterial blooms that are of concern to local resource managers. Our objectives were to describe changes in phytoplankton community composition in

Vancouver Lake over seasonal, annual, and interannual time scales, and to identify strong water quality predictors of phytoplankton community structure, with an emphasis on cyanobacterial blooms, from 2007 to 2010. Cluster analysis, indicator species analysis, and non-metric multidimensional scaling were used to identify significantly different phytoplankton community groupings and to determine which environmental factors influenced community changes. From

2007 through 2009, depletion of NO3-N followed by elevated PO4-P concentration was associated with increased biomass and duration of each cyanobacterial bloom. Time-lag analysis suggested that NO3-N availability contributed to interannual changes within the summer

4 phytoplankton community. Specifically, in summer 2010, a distinct cyanobacteria community was not present, potentially due to increased NO3-N and decreased PO4-P and NH4-N availability. Our study provides a comprehensive assessment of species-level responses to water quality variables in a shallow non-stratifying temperate lake, contributes to a better understanding of phytoplankton dynamics generally, and may aid in predicting and managing cyanobacterial blooms.

Key words: algal blooms; NMDS; nutrients; orthophosphate; water quality; shallow lake

1. Introduction

Cyanobacterial blooms are commonly associated with eutrophic freshwater and estuarine systems worldwide. Excessive abundance of cyanobacteria may have detrimental effects on water quality, including the development of surface scums and depleted oxygen levels

(Carmichael 1992; Sellner et al. 2003). In addition, many cyanobacteria species can produce potent toxins that negatively affect aquatic life and in particular cause harm or even death to humans and other mammals (Chorus et al. 2000; Hudnell 2010; Koreivienė et al. 2014). Harmful cyanobacterial blooms may also negatively affect freshwater ecosystem services by decreasing recreational opportunities, causing fish kills which may affect local fisheries, increasing water treatment costs, and decreasing aquatic biodiversity (Hoagland et al. 2002; Jacoby and Kann

2007; Paerl 2008).

A suite of factors is thought to influence the formation, persistence, and decay of cyanobacterial blooms in freshwater systems. Excess nutrient concentrations, mainly phosphorous and/or nitrogen, are commonly associated with harmful algal blooms (Anderson et

5 al. 2002; Heisler et al. 2008). Hydrodynamic factors, such as turbulence, wind-driven waves and residence time, as well as light availability and temperature, have also been shown to contribute to harmful algal blooms (Cloern 1991; Koseff et al. 1993; Thompson et al. 2008). These abiotic factors may also be coupled with biotic factors, such as competition for resources and trophic interactions, to influence changes in bloom dynamics (Cloern 1991; Boyer et al. 2011;

Rollwagen-Bollens et al. 2013).

Despite the clear public health concerns, effective public monitoring programs for lakes are rare and often limited in their temporal resolution and the breadth of information they can provide about cyanobacterial bloom dynamics. For instance, monitoring is typically scheduled to coincide with the peak summer growing season such that early spring or late fall blooms may not be captured. In other cases, monitoring may only begin once a bloom has already been established, thus missing key information about the environmental dynamics leading to bloom formation. This may be due to lack of financial resources or technical expertise; however, without information about changes to the phytoplankton community before, during, and after a bloom, it is difficult to understand or predict which environmental factors may influence cyanobacteria bloom dynamics. Indeed, examining phytoplankton community dynamics regularly over several years has been shown to provide a better understanding of changes in relation to water quality and the environment (Schindler 1987). Moreover, by focusing only on relatively brief algal bloom periods, longer-term effects of seasonal blooms may be severely underestimated (Cottingham and Carpenter 1998; Rothenberger et al. 2009).

There are a number of well known comprehensive studies examining seasonal succession of phytoplankton communities in inland waters: Windermere Lake in England (Neale et al. 1991;

Thackeray et al. 2008; Elliott 2012), Lake Biwa in Japan (Hsieh et al. 2010; Tsugeki et al. 2010),

Lake Baikal in Siberia (Hampton et al. 2008), several Great Lakes in the United States and

Canada (Makarewicz et al. 1999), and more recently Lake Tai in China (Guo 2007). In the U.S.

Pacific Northwest region, Lake Washington may be the most well-known and intensively studied lake (Edmondson 1994; Arhonditsis et al. 2003). Each of these lake systems has been affected by increased eutrophication due to changing anthropogenic influences, and has a phytoplankton community that shifts from dominance by diatoms in spring to dominance by cyanobacteria during stratification events.

Stratification has been identified to be one of the more important drivers of phytoplankton seasonal succession (Sommer et al. 1986; De Senerpont Domis et al. 2013), especially with the onset and duration of cyanobacteria blooms (Paerl and Huisman 2009; De

Senerpont Domis et al. 2013). Diazotrophic cyanobacteria species are thought to thrive under stratified conditions because the formation of a stable epilimnion isolates the phytoplankton from a nitrogen source (Reynolds et al. 2000; Kokocinski and Soininen 2008). In contrast, few observational studies have investigated how the phytoplankton community changes from a mixed phytoplankton to a cyanobacteria-dominated system in large, shallow, non-stratified lakes.

Vancouver Lake in Washington state, USA, is a very shallow, tidally influenced, well mixed, and non-stratifying system; thus it represents a relatively under-studied freshwater system in which to investigate cyanobacterial bloom dynamics. Since 2006, Vancouver Lake has been the focus of a monitoring and experimental program to quantify the range of biotic and abiotic factors that influence the timing, magnitude, diversity, and toxicity of cyanobacterial blooms.

With regard to biotic interactions, microzooplankton and mesozooplankton have been found to exert top-down effects on cyanobacteria populations in Vancouver Lake, particularly before and after the summer peak in cyanobacteria abundance (Boyer et al. 2011; Rollwagen-Bollens et al.

2013). In addition, the seasonal and interannual variability in the cyanotoxin microcystin has been quantified in the lake, and using qPCR techniques, the cyanobacteria taxon responsible for producing the toxin was identified as Microcystis sp., despite this species often being undetectable in microscopic examination of plankton samples (Lee et al. 2015).

Despite what has been shown about the role of grazing on cyanobacteria bloom dynamics, the dominant abiotic controls on the cyanobacteria populations in Vancouver Lake have yet to be addressed. Therefore, the objectives of this study were to 1) describe the variability in phytoplankton community composition over seasonal, annual, and interannual time scales, based on samples collected from 2007-2010; and 2) determine which water quality factors most strongly influenced changes in phytoplankton community structure, specifically in relation to cyanobacterial blooms.

2. Materials and method

2.1 Study site

Vancouver Lake is located adjacent to the Columbia River in Clark County, Washington,

USA, within the greater Portland, OR-Vancouver, WA metropolitan area (Fig. 1). It is a natural floodplain lake that is 930 ha in area, tidally influenced, non-stratified, and seasonally eutrophic.

The lake is shallow (mean depth ~1 m) and highly turbid. A flushing channel, located near the southwest corner, was constructed in 1983 for the purpose of restoring some measure of natural flushing from the Columbia River that had been blocked by land reclamation in the early 20th century.

Depending on the time of year and tidal flow, the Vancouver Lake watershed may be up to 260 km2 in area, draining into the lake through Burnt Bridge Creek, Salmon Creek, and Lake

River. Tidal inflow from the Columbia River is through the flushing channel and Lake River, whereas tidal outflow is only through Lake River (Foreman et al. 2014). The immediate surrounding land use around Vancouver Lake is predominantly undeveloped or preserved wildlife habitat along the northern, western, and southern perimeter. Several small housing developments exist along the eastern shoreline south of Burnt Bridge Creek.

Annual summertime cyanobacterial blooms have been documented in the lake since

1971, although anecdotal evidence suggests blooms occurred earlier (Bhagat and Orsborn 1971).

The blooms in Vancouver Lake have historically been dominated by the filamentous cyanobacteria species Aphanizomenon sp. and Anabaena sp.; however, there have been some years when Microcystis sp. was the dominant taxon (Bhagat and Orsborn 1971; Jacoby and Kann

2007).

2.2 Sampling methods

All sampling was conducted from a dock on the southeast shore of the lake (Fig. 1).

Quarterly sampling at eight broadly distributed sites (littoral, limnetic, and dock) was conducted in 2007. Kendall’s tau (τb) was used to examine concordance among sites and showed no significant spatial differences in plankton community composition (n=10 per site, p>0.05)

(Bollens and Rollwagen-Bollens 2009). Because of this, and ease of access throughout the year, the dock was chosen as the site for all subsequent sampling. Lake water samples were collected on a monthly (November through February), bi-weekly (March and October), or weekly (April through September) basis from February 2007 through September 2010.

On each sampling date three independent bucket samples were collected from the surface and subsampled for microplankton abundance and water quality. 250 mL was subsampled from

9 each bucket and preserved in 5% acid Lugol’s solution for subsequent enumeration of phytoplankton. An additional subsample of 50 mL from each bucket was filtered through a 0.45

µm Millipore filter into a plastic bottle and kept refrigerated or frozen for a few weeks until analysis. Samples were sent to the Marine Chemistry Lab at the University of Washington’s

School of Oceanography, and were analyzed for nitrite (NO2-N), nitrate (NO3-N), ammonium

(NH4-N), phosphate (PO4-P), and silicate (SiO4-Si) concentrations. In addition, total water column depth and Secchi depth were recorded, and vertical profiles of temperature and dissolved oxygen (DO) were obtained using either a YSI 91 or YSI 6920 data sonde.

Lugol’s preserved samples were concentrated in settling chambers (≤ 10 mL) and examined using an Olympus CK-40 inverted microscope. Species were identified to genus, and species where possible, according to Prescott (1978) and Wehr and Sheath (2002).

Phytoplankton enumeration and biomass calculations were determined as recommended by

APHA (2012). Briefly, counts were made at 400X magnification, and species were identified at

600X magnification when needed. Individual phytoplankton cells were enumerated, sized and identified from 10 to 20 ocular fields, or to a minimum of 200 cells during months when overall abundances were low (Lund et al. 1958). Average cell count per sample exceeded 1,500 individual cells, of which at least a minimum of 200 non-cyanobacteria cells were targeted during cyanobacteria blooms. To keep units consistent for comparison among all phytoplankton taxa, our counting unit was individual cells. Phytoplankton biovolumes were calculated based on geometric shape (Hillebrand et al. 1999) and biomass estimated using the empirically derived biovolume-biomass relationships described in Menden-Deuer and Lessard (2000). Biomass values are reported here and were used for all subsequent statistical analyses.

2.3 Statistical analyses

Phytoplankton communities can be described by identifying all taxa in the community, or by combining individual taxa into higher taxonomic groups (e.g. diatoms, chlorophytes, dinoflagellates). Alternatively, phytoplankton taxa can be classified into functional groups based on shared morphological or functional characteristics (Reynolds 2002; Kruk et al. 2010). The utility of each of these grouping approaches depends on the question being asked (Mieleitner et al. 2008). For instance, pooling species into higher taxonomic groups may allow for ease of analysis of diversity, but because individual taxa utilize resources differently and may have different requirements, understanding how communities change over time in response to varying environmental conditions may be underestimated. Moreover, this approach does not allow for targeting specific taxa (e.g. toxin producing phytoplankton). With respect to functional groupings, they may be useful for making comparisons across multiple aquatic systems, and for studies where little is known or reported about the particular environmental conditions at a site; however, this approach is limited in its utility for natural resource managers who need to understand how local environmental conditions influence plankton community dynamics in their system of interest (e.g. harmful cyanobacterial blooms, etc.).

In order to avoid the potential problems described above, and to provide an analysis that would be most useful for the management of Vancouver Lake, we chose to utilize the complete list of all phytoplankton taxa identified, and to define groups using a cluster analysis statistical approach (Clarke 1993). Cluster analysis uses the frequency and relative abundance of all taxa observed throughout the sampling period, and compares groupings at each sampling period to determine community similarities over time. This allowed us to objectively characterize the phytoplankton community at each sampling time independent of environmental conditions, and

11 then to monitor annual and interannual patterns of seasonal community succession in relation to these conditions.

Our statistical analysis included four steps. First, cluster analysis was used to identify phytoplankton groups using relative Euclidean distance measure and Ward’s method for group linkage with 75% of information retained. Relative Euclidian distance is the preferred choice for determining similarities among sample dates and to eliminate potential bias that may arise from using absolute biomass values (McCune and Grace 2002). We then used multiple response permutation procedure (MRPP) (Mielke et al. 1981; McCune and Grace 2002), using ranked

Sorenson’s distance, to test whether the groups resulting from cluster analysis were significantly different from one another. The MRPP test also allowed us to determine whether there were significant interannual differences in the phytoplankton community composition over the full 4- year dataset.

Next, we used indicator species analysis to determine which subsets of phytoplankton taxa were most strongly associated with each group determined by cluster analysis (Dufrene and

Legendre 1997). One thousand randomizations were used in a Monte Carlo test. Indicator species analysis identifies which taxa best represent a cluster group based on the abundance and frequency of occurrence of each taxon; thus, the resulting information describes the community make-up of each cluster group (McCune and Grace 2002). Taxa associated with a particular cluster that occurred five times more frequently than in other clusters were deemed “faithful” and significantly representative of that particular cluster.

Finally, we used an ordination technique to detect the relationship between phytoplankton assemblages and environmental data. Nonmetric multidimensional scaling (NMDS) was chosen because it can be used with non-normal and discontinuous distributions, and is currently

12 considered the most effective ordination technique in community ecology (McCune and Grace

2002).

All cluster analyses, indicator species analyses, MRPP tests and ordinations were performed using PC-ORD version 5.3 software. Any sample dates with incomplete data (i.e. missing nutrient concentration data or dissolved oxygen values unavailable due to equipment malfunction) were excluded. Out of 90 identified species, a total of 70 species were included in all analyses (species occurring in <5% of samples were excluded). In all cases data were log +1 transformed to allow for the visualization of both the highly abundant and rare species, and to dampen undue influence of either group on the overall patterns of biomass (McCune and Grace

2002).

Environmental variables included in the ordinations were total water depth, Secchi depth, temperature, dissolved oxygen (DO), NO2-N, NO3-N, NH4-N, PO4-P, SiO4-Si, and DIN:DIP.

DIN was calculated as the molar sum of NO2-N, NO3-N, and NH4-N; DIP included only PO4-P.

Due to the number of significant groups derived from cluster analysis and MRPP, as well as the large number of samples in the four-year dataset, separate NMDS analyses for each year were performed. These separate NMDS ordinations allowed for better resolution and comparison of phytoplankton communities by reducing ‘noise’ that might otherwise have masked subtle differences in the larger dataset (Clarke 1993; Rothenberger et al. 2009). Because data of a particular type were not always available throughout our entire sampling period (e.g., due to instrument failure), ordinations using subsets of data that included ambient light and light attenuation were also examined.

Shifts in phytoplankton community structure may be a rapid response to existing environmental fluctuations, but may also reflect a response to prior environmental conditions.

For example, phytoplankton responses to chronic or pulsed nutrient delivery range from days to weeks (Rantajarvi et al. 1998; Glibert et al. 2008; Heisler et al. 2008), thus there may not be an immediate correlation of phytoplankton biomass with an environmental variable (Anderson et al.

2002). Therefore, we also imposed time lags from one to five weeks on the environmental variables and included these in the ordinations. If an environmental variable was associated with a specific group at time (lag) zero and either remained or was no longer associated with the same group after lagging from one to five weeks, no additional interpretation was undertaken, as it implied a response to environmental conditions existing at that time. However, if a significant relationship between phytoplankton composition and an environmental variable was detected only after imposing a lag, then we interpreted this to mean that the phytoplankton community was responding to prior environmental conditions.

3. Results

3.1 Phytoplankton community composition

A total of 90 distinct phytoplankton taxa were identified from acid Lugol’s preserved samples (Table 1). Absolute and relative phytoplankton abundance and biomass from 2007-2010 are illustrated in Figures 2 and 3, respectively. Total phytoplankton abundance increased during the summer months and decreased during the winter months. From 2007 through 2009, summer abundances increased but then exhibited a sharp decrease in 2010. Similarly, phytoplankton biomass was lowest during winter months (November through February) and highest during the summer months (August through October) (Fig. 2a). The general seasonal pattern in phytoplankton community composition changed from a diatom-dominated spring community

(March through June) to a cyanobacteria-dominated summer community of Anabaena sp. and

Aphanizomenon sp. (which contributed to annual phytoplankton peak biomass [Fig. 2b]). An exception to this occurred in 2010, when the summer phytoplankton community was dominated by diatoms. Total phytoplankton abundance was higher in 2007 than in 2010 and total phytoplankton biomass was higher in 2010 than in 2007. The discrepancy between these two metrics is due to differences in size and abundance of different phytoplankton taxa; for example, in 2007 the summer community was dominated by cyanobacteria of higher abundance and smaller biomass, compared to the summer diatom community of 2010 which was dominated by higher biomass and lower abundance.

3.2 Cluster and indicator species analyses

Six significantly different clusters were identified based on phytoplankton biomass (A =

0.144, p < 10-8), numbered 1 through 6, with a set of indicator species identified for each cluster

(Table 2). Interannual differences in community succession were observed throughout the study

(Fig. 4). First, the duration and magnitude of cyanobacteria-dominated blooms increased each year from 2007 to 2009. In 2007, the spring was dominated by a mixture of diatom and euglenoid biomass (cluster 1), followed by a bloom of cyanobacteria (cluster 2) in late summer and early autumn. The phytoplankton community biomass during the winter of 2008 was again dominated by a diatom-euglenoid community (cluster 1).

In 2008 and 2009 there were similar community patterns: spring of both years was dominated by a chlorophyte-diatom community (cluster 3) which then transitioned to a diatom- dominated community (cluster 4) in the early summer prior to the late summer cyanobacterial blooms (cluster 2). The only differences observed between 2008 and 2009 were the winter

15 communities: in 2008 the winter community consisted of mixed-algal groups (cluster 1) while in

2009 the winter community was a chryptophyte-dinoflagellate mix (cluster 5).

The community in 2010 was distinctly different from 2007 through 2009, with a shift in species dominance occurring every two months from a mixed assemblage (cluster 3) in spring, to diatoms (cluster 6) in early summer, back to a mixed assemblage (cluster 3) in late summer, and again to diatoms (cluster 6) in autumn (Fig. 4). Although there was an increase in both cyanobacterial biomass and abundance in 2010 (Figs. 2 and 3), the 2010 summer community was significantly different than the cyanobacterial community observed in previous years. Thus, there was no statistically detectable cyanobacteria community in 2010. MRPP analysis of the biomass clusters indicated the communities in 2008 and 2009 were similar and significantly different from those of 2007 and 2010.

3.3 Environmental variables

Total water depth ranged from 0.8 m during the summer to 4.5 m during the spring freshet (Fig. 5a). Secchi depth ranged from 0.1 m during the summer to 1.6 m during the spring, with increased water clarity during seasonal winter rains and spring freshets (Fig. 5a).

Temperature varied from as low as 0°C during the winter (December 2009) to as warm as 28°C during the summer (Fig. 5b). Dissolved O2 ranged from 0.69 mg/L to 24 mg/L at the lake surface

(Fig. 5b).

SiO4-Si concentrations throughout the sampling period ranged from 150 to 11,000 µg/L, and were lowest during late spring and early summer months. A slight decrease in SiO4-Si concentration was observed during the winter months, followed by an increase in SiO4-Si during the early spring and late summer months of each year (Fig. 5c). NO2-N concentrations were

16 variable (<0.06 to 20 µg/L) and did not show any consistent seasonal or interannual variation

(Fig. 5c). However, NO3-N showed increased concentrations during the winter and early spring months of each year (mean highs ranged from 220 to 370 µg/L). NO3-N did not show annual variation; however, the timing of peak NO3-N levels became progressively later by one month each year (Fig. 5d). Although NH4-N did not show any significant seasonal or annual patterns, peak concentrations were observed during the summer months of 2007, 2008, and 2009. In 2010

NH4-N concentrations were variable throughout the year (Fig. 5d).

Similar to trends observed with NH4-N, PO4-P concentrations increased during the summer months of each year (49 to 230 µg/L), with an additional small increase in PO4-P observed during each winter (19 to 22 µg/L). Finally, DIN:DIP did not show consistent patterns of significant seasonal or annual variability; however, there were sustained periods of DIN:DIP less than 16:1 observed during summer months (Fig. 5e).

3.4 Association of phytoplankton groupings with environmental conditions

We examined the full four-year phytoplankton data set to elucidate how different phytoplankton community groups were associated with corresponding environmental variables.

The full data set was then parsed into three subsets of data based on the MRPP results described above (2007, 2008-2009, and 2010) to examine which environmental variables were most strongly associated with each annual phytoplankton community, and to identify relationships that might explain the MRPP-defined temporal groupings. The four sets of the phytoplankton composition and environmental data used in the ordinations were: 1) May 2007 – September

2010 (the full dataset); 2) May – December 2007; 3) January 2008 – December 2009; and 4)

January – September 2010. Species data from February – April 2007 were not included in any of

17 the NMDS analyses due to a lack of corresponding environmental data. Environmental vector cutoff values (r2), observed variance of each axis, stress and instability for each NMDS analysis are reported in Table 3. The direction of environmental vectors indicates its association with a particular community. The correlation of each environmental vector with a particular axis or axes indicates the strength of its association with a particular community. Separate, additional ordinations that included subsets of data on light availability and light attenuation at the water surface were not associated with any axes or phytoplankton groups (data not shown).

Results from NMDS analyses of phytoplankton biomass from 2007 – 2010 (Fig. 6a) showed diatom biomass (clusters 4 and 6) were associated with high lake water levels and water clarity, and high DIN:DIP ratios (>16:1). Total lake depth, Secchi depth, and DIN:DIP were inversely related to cyanobacteria biomass (cluster 2) during the blooms in 2007–2009.

Cyanobacteria blooms (cluster 2) were positively associated with an increase in SiO4-Si, PO4-P, and, NH4-N. In 2008 and 2009, SiO4-Si, PO4-P and, NH4-N were lowest during the late spring months, as well as during late spring and summer 2010, when diatoms (clusters 4 and 6) were dominant.

In 2007, there were only two phytoplankton communities, clusters 1 and 2. The ordination from this specific subset of data showed that the diatom-euglenoid community (cluster

1) was associated with high DIN:DIP, increased NO3-N, total lake depth, and Secchi depth during late spring and early summer, and with NO3-N during winter (Fig. 6b). Dissolved O2 and temperature increased and peaked in early stages of the summer cyanobacteria bloom (cluster 2) just before a rapid increase in NH4-N. Increased phytoplankton biomass and diversity during late summer months were associated with increased availability of PO4-P and SiO4-Si. Increased cyanobacteria biomass during late summer was associated with increased PO4-P availability. In

18 particular, the high biomass of cyanobacteria (cluster 2) was largely due to the taxa

Aphanizomenon sp. and Anabaena sp. (Online Resource 1).

The ordination of the combined years of 2008 and 2009 indicated that the euglenoid- diatom and mixed phytoplankton communities (clusters 1 and 3, respectively) were associated with dissolved O2 and the annual increase of NO3-N (Fig. 6c). Increased temperature, PO4-P and

SiO4-Si were associated with the annual summer cyanobacterial blooms (cluster 2). High

DIN:DIP, deeper lake depth and deeper Secchi depth were associated with the late spring diatom community (cluster 4).

In 2010, DIN:DIP, NO3-N and SiO4-Si were associated with the mixed phytoplankton community (cluster 3) during early spring and late summer months. Dissolved O2, NH4-N, temperature and total lake depth were associated with the chlorophyte-diatom community

(cluster 6) during late spring and early summer months (Fig. 6d).

Time lag analysis was applied to the full dataset (2007-2010) to assess if there were any delayed responses within the phytoplankton community to environmental variables that would otherwise not have been apparent. Only a three-week time lag analysis between environmental conditions and phytoplankton species biomass resulted in significant relationships, and showed that NO3-N was associated with the mixed phytoplankton community and diatom-chlorophyte community (clusters 6 and 3, respectively) (data not shown). NO3-N was highest during winter and early spring months in 2008, 2009, and 2010, when phytoplankton biomass was lowest. As

NO3-N levels decreased, phytoplankton biomass began to increase, particularly the diatoms and flagellates, suggesting the spring communities relied on the increased winter storage of NO3-N conditions. In 2010, NO3-N persisted at higher levels during summer months compared to previous years, which may have contributed to the significant difference in summer

19 phytoplankton community observed between the diatom-chlorophyte group (cluster 6) and cyanobacteria (cluster 2).

4. Discussion

Our statistical analyses relating water quality variables to the phytoplankton community composition in Vancouver Lake from 2007 to 2010 demonstrated that the timing and magnitude of several key environmental variables were significantly associated with changes within the phytoplankton community, most notably with summer blooms of cyanobacteria.

4.1 Environmental variables associated with summer cyanobacteria blooms

With the exception of 2010, summer cyanobacterial blooms were associated with increased concentrations of SiO4-Si, PO4-P, and NH4-N, and the depletion of NO3-N.

Silicate. SiO4-Si decreased during the spring as diatom biomass increased and SiO4-Si was taken up. As the diatom populations crashed, an increase in SiO4-Si was observed. While

SiO4-Si did not appear to influence the summer cyanobacteria blooms directly, it did signify a shift in community composition. Other studies have also observed an association between SiO4-

Si and shifts from a spring community to a summer community. In a study consisting of 31 shallow and artificial lakes in southeast England, Bennion and Smith (2000) showed that silicate decreased when Chl-a increased during the spring, but then silicate increased during a shift in the phytoplankton community from a diatom-dominated assemblage to a mixed phytoplankton community. In Lake Washington, Arhonditsis et al. (2003) observed a shift from a diatom- dominated community to a chlorophyte-cyanobacteria community when silicate and PO4-P availability increased. In a set of laboratory incubation experiments, Spears et al. (2008) found

20 that increased SiO4-Si coincided with increased PO4-P, mainly due to the dissolution of diatoms.

In none of these systems did increased SiO4-Si contribute to a cyanobacteria bloom. Instead, increased silicate concentration was associated with a shift in phytoplankton community composition from a diatom-dominated community to a mixed-phytoplankton community. In

Vancouver Lake, however, summertime increases in SiO4-Si were associated with a shift towards a cyanobacteria-dominated community.

Phosphate. PO4-P was also strongly associated with the summer cyanobacteria blooms in

Vancouver Lake, except in 2010. From 2007 to 2009, PO4-P availability during the summer increased each year, and the timing of peak PO4-P availability corresponded to peak cyanobacteria biomass. This suggests that PO4-P was associated with the magnitude and duration of the blooms, but not associated with their initiation. Previous studies have indicated that

Anabaena, Aphanizomenon, and Microcystis are poor PO4-P competitors; however, with ample

PO4-P availability, these species are able to dominate the cyanobacteria community (Riddolls

1985; Davis et al. 2010;).

The differences between an Anabaena-dominated and Aphanizomenon-dominated community could be attributed to Aphanizomenon being a poor P competitor compared to

Anabaena (De Nobel et al. 1995). De Nobel et al. (1997) showed that Anabaena growth rates and biomass production were greater than Aphanizomenon when an inorganic N source was unavailable and P was limited. However, in the presence of ample PO4-P and NH4-N, Anabaena still had a competitive advantage, although at a lower rate. The dominance of Anabaena in

Vancouver Lake may be explained by increased PO4-P and NH4-N availability, but the co- occurrence and eventual dominance of Aphanizomenon could be due to light-limiting conditions

(De Nobel et al. 1998), or other environmental factors (i.e. micronutrients and/or selective grazing) that have not yet been investigated.

Ammonium. NH4-N availability influenced phytoplankton biomass across all groups, but it may have contributed to the dominance of cyanobacteria in particular. Several lab and field studies have demonstrated cyanobacteria preference for NH4-N over other inorganic forms of N because the former is more energetically favorable. For instance, NH4-N availability can promote twice as much carbon production in cyanobacteria when compared to other oxidized forms of N

(Blomqvist et al. 1994; Turpin et al. 1985), and Ferber et al. (2004) showed that diazotrophic species, including Anabaena and Aphanizomenon, preferred and utilized NH4-N when the community shifted from non-diazotrophic cyanobacteria in the early part of the bloom to diazotrophic cyanobacteria in the latter half of the bloom in Shelburne Pond, Vermont, USA.

Sharp increases in both PO4-P and NH4-N occurred around the same time in Vancouver

Lake: in 2007 maximum NH4-N levels occurred in July and PO4-P levels peaked in August. In

2008 and 2009, both nutrients peaked in August, and to a lesser extent they also peaked together in August 2010. This might be explained by a combination of several chemical and biological factors. During the same period when PO4-P and NH4-N increased, dissolved O2 levels near the sediment water interface were as low as 3.0 mg/L, and on some dates <2.0 mg/L (data not shown). As sinking and decomposition of the spring phytoplankton community occurs, oxygen can become depleted, creating anoxic conditions at the sediment-water interface. PO4-P is released from iron-bound particles due to redox-mediated processes: as conditions at the sediment-water interface become anoxic, Fe (III) reduces to Fe (II), releasing PO4-P into the water column (Mortimer 1942; Caraco et al. 1993; Søndergaard et al. 2003). NH4-N accumulates and is released into the water column due to the lack of oxygen inhibiting the nitrification

22 process (Lehman 2011). This suggests that anoxic conditions may have contributed to PO4-P and

NH4-N release into the water column in Vancouver Lake.

In most lake systems where this pattern is observed, the release of PO4-P and NH4-N usually occurs in the hypolimnion (Krivtsov and Sigee 2005; Lehman 2011). However,

Vancouver Lake does not have a hypolimnion due to its extremely shallow depths and strong mixing, and may therefore be more comparable to a large sedimentation pond. Sedimentation ponds are components of water treatment complexes that serve a similar function as wetlands in

P removal, and act as small shallow lakes that can be mediated by similar physical and biological processes (Palmer-Felgate et al. 2011). Palmer-Felgate et al. (2011) showed in sedimentation ponds that under certain conditions, NH4-N and total reactive phosphate concurrently increased and peaked, as did NH4-N and soluble reactive phosphate just below the sediment water interface. Their study, along with Cook et al. (2010), conducted in the Grippsland Lakes in southeast Australia, suggested that the observed increase in inorganic phosphate was due to decomposition of the spring phytoplankton community, causing decreased concentrations of dissolved O2 and release of inorganic phosphate. Similarly, in both field and modeling studies, increased NH4-N levels have been attributed to macrophyte die-off, which then contributed to increased phytoplankton biomass (Landers 1982; Farnsworth-Lee and Baker 2000)

Nitrate. Finally, although NO3-N was not significantly associated with the summer cyanobacteria blooms in Vancouver Lake, it may have contributed to the initiation of each bloom. NO3-N was reduced during the spring, when there was an increase in both diatoms and chlorophytes. Several studies have indicated that N depletion is necessary for cyanobacteria bloom formation since cyanobacteria, especially diazotrophic species, have a physiological advantage for N-storage and N-fixation, thereby making them better N competitors (Dodds and

Priscu 1990; Blomqvist et al. 1994; Cook et al. 2010). In Vancouver Lake, NO3-N levels, in addition to NO2-N and NH4-N, were depleted to trace amounts for several weeks during the transition between the late spring/early summer phytoplankton community and late summer cyanobacteria bloom (Figs. 2a and 3a). Notably, in a separate study of zooplankton grazing in

Vancouver Lake conducted during 2008 and 2009, Rollwagen-Bollens et al. (2013) showed that during this 3-4 week period prior to the cyanobacteria bloom when dissolved inorganic nitrogen was low, a cascading effect among planktonic grazers likely resulted in the selective removal of diatoms and thus reduced the competition with cyanobacteria for scarce nutrients, which may have allowed the cyanobacteria bloom to begin.

4.2 Environmental variables associated with winter/spring phytoplankton communities

In general, our results indicated that winter and early spring phytoplankton communities were associated with increased total lake depth, increased water clarity (Secchi depth), high

DIN:DIP, and high NO3-N. Total lake depth and Secchi depth were important environmental variables associated with periods when biomass was at an annual low and DIN:DIP levels were

>16:1. Although NO3-N was not an important variable influencing species biomass across our full data set, results from subsequent ordinations of “annual” subsets of data and time-lag analysis indicated otherwise.

Ordination results from yearly subsets of biomass data showed NO3-N was associated with winter communities of diatoms and chlorophytes (cluster 1) and flagellates (cluster 5), when

NO3-N levels were highest and overall biomass was lowest. The effects of peak NO3-N levels were delayed by one month each year from 2008 through 2010; this may have contributed to the absence of NO3-N as an important environmental variable in the overall NMDS analysis. Yet,

24 time lag analysis suggested that initial establishment of mixed algae (cluster 3) and chlorophytes- diatoms (cluster 6) were associated with increased NO3-N availability during the winter.

High levels of NO3-N during winter months may indicate that other conditions, such as low temperature and low light availability, limited phytoplankton growth and nutrient uptake during this time, which may also explain the time-lag response to NO3-N in species biomass.

Bennion and Smith (2000) found that 29 out of 31 shallow ponds investigated in England exhibited increased NO3-N levels during the winter, concurrent with reduced primary productivity. In Lake Washington, in Washington state, USA, Arhonditisis et al. (2003) observed a similar pattern of high NO3-N availability during the winter and low diatom abundance and phytoplankton biomass (chl a), until increasing temperatures and day length facilitated increased diatom abundance and biomass and depletion of NO3-N.

Our results suggest that in Vancouver Lake, the diatom-chlorophyte bloom in spring may be a significant cause of NO3-N depletion, as indicated by an increase in absolute abundance and biomass of phytoplankton, and depletion of SiO4-Si. Phytoplankton biomass decreased again shortly after the depletion of NO3-N. However, the collapse of the spring diatom-chlorophyte bloom may not be explained solely by nutrient limitation, as zooplankton grazing (Arhonditsis et al. 2003; Boyer et al. 2011; Rollwagen-Bollens et al. 2013), sedimentation, and flushing may also significantly affect spring phytoplankton communities in lakes (Reynolds 2006). Indeed,

Boyer et al.(2011) and Rollwagen-Bollens et al. (2013) both found that the grazing effect of zooplankton on phytoplankton (including cyanobacteria) was as high as 70% of total biomass on a daily basis in the weeks immediately following the bloom peaks in Vancouver Lake during

2008 and 2009.

4.3 Interannual variability in cyanobacteria blooms

The 2010 phytoplankton community provided a contrast to the three other years of our study. Specifically, a cyanobacterial bloom community was not detected during the summer, even though a small and brief increase in cyanobacterial biomass was observed. Instead, the summer 2010 phytoplankton community was largely dominated by two different diatom communities. During 2007 – 2009, NO3-N availability was consistently near zero in the summer months. However, in 2010, NO3-N availability during the summer was much more variable.

Blomqvist et al. (1994) suggested that NO3-N availability may decrease the competitive ability of cyanobacteria in the presence of other phytoplankton. Diatoms, and eukaryotes more generally, prefer NO3-N over other forms of N (Blomqvist et al. 1994; McCarthy et al. 2009).

This may have allowed eukaryotes to continue thriving under these conditions, thereby muting the cyanobacteria response to PO4-P availability.

The phytoplankton community in 2010 was dominated by diatoms. In 2007-2009, SiO4-

Si levels rebounded after each spring diatom bloom, during which diatom biomass decreased. In

2010, SiO4-Si availability remained low throughout the summer, suggesting continual recycling of SiO4-Si as diatom biomass and dominance increased. Cluster analysis indicated a shift between two different diatom communities — from a community of Fragilaria and Melosira during the spring, to a second diatom community consisting of Aulacosira and Stephanodiscus during the summer. Changes in N availability (from NO3-N to NH4-N) and increased temperature may have caused a shift within the diatom community due to differences in resource uptake.

Decreased PO4-P and NH4-N availability may also have contributed to the observed decrease in biomass of cyanobacteria in summer 2010. Dissolved O2 levels near the sediment

26 water interface remained >5.0 mg/L in 2010, whereas during 2007-2009 summertime dissolved

O2 levels were usually much lower (<3.0 mg/L). Similarly, in 2010 PO4-P and NH4-N levels were lower than in previous years, which suggest that low oxygen conditions observed in 2008 and 2009, and to a lesser extent in 2007, may have strongly influenced PO4-P and NH4-N release into the water column. Hupfer and Lewandowski (2008) suggested that microbial recycling of organic matter can also contribute to observed inorganic phosphate release. Other studies suggest that microbial cycling of nutrients (Gachter et al. 1988; Davelaar 1993), along with other mechanisms such as bioturbation by zooplankton and fish (Fukuhara and Sakamoto 1987;

Adámek and Maršálek 2013), excretion by zooplankton and then remineralization by bacteria

(Axler et al. 1981; Hambright et al. 2007), may also contribute to N and P availability.

4.4 Summary and management implications

Overall, phytoplankton community dynamics were strongly influenced by nutrient availability and the phytoplankton community succession patterns observed in Vancouver Lake – a spring diatom bloom followed by a summer cyanobacteria bloom – were similar to patterns observed in deep, stratifying, eutrophic systems. However, the mechanisms underlying these patterns were likely different in this shallow, non-stratifying system (e.g. internal nutrient loading due to sediment resuspension), and might have been overlooked if sampling had occurred less frequently.

The environmental variables most closely associated with changes in the phytoplankton community from 2007 through 2009 were SiO4-Si concentrations, which were associated with the shift from a spring diatom community to a summer cyanobacteria community; high NO3-N concentrations, which were associated with the spring phytoplankton communities; low NO3-N,

NO2-N, and NH4-N availability in early summer that influenced the initiation of the summer cyanobacteria blooms; and finally, high NH4-N and PO4-P concentrations, which were associated with summer cyanobacteria blooms. The effects of NO3-N depletion and increased PO4-P availability on the growth, magnitude, and duration of cyanobacterial blooms suggests that management tactics that control for these two nutrients might help mitigate the severity of seasonal cyanobacteria blooms.

The same environmental variables that were important in 2007-2009 were also important in 2010, with one important exception: PO4-P and NH4-N levels were drastically reduced in

2010, which may have influenced the observed change in phytoplankton community composition that year. In a typical stratifying lake, the availability of PO4-P and NH4-N from the hypolimnion can be attributed to a mixing event that breaks down the thermocline; however, the timing and pattern of the observed increase in PO4-P and NH4-N in Vancouver Lake suggests that there may be physical-chemical interactions at the sediment-water interface that provide for the availability of these two nutrients during the summer, but not at other times of the year. Thus, further studies investigating internal nutrient loading in shallow non-stratifying lakes are warranted as it may influence management decisions aimed at controlling nutrient inputs.

Few previous studies have intensively monitored and characterized phytoplankton community dynamics and associated environmental conditions over four continuous years in a shallow temperate lake. Our study demonstrates the importance of consistent monitoring to capture interannual variation of both the phytoplankton community and corresponding environmental variables, but also the benefit of high frequency (weekly) sampling during spring and summer, which allowed us to identify the dynamic relationships between environmental variables and cyanobacteria blooms that vary on short time scales. This approach may allow

28 better prediction and management of cyanobacteria blooms, and should be considered when planning future monitoring and mitigation programs. For example, in Vancouver Lake monthly monitoring of nutrients might have missed the dramatic pulse in PO4-P availability observed during our weekly sampling. Thus, the effects of nutrients on the persistence and growth of cyanobacterial blooms during the summer might have been overlooked and might have led to potentially less effective management strategies.

Finally, future studies of cyanobacteria bloom dynamics should include examination of rate processes such as growth, grazing, flushing, sinking and resuspension, which will be necessary to achieve a full understanding of the causal mechanisms underlying cyanobacteria blooms in shallow temperate lakes. Such a causal understanding of bloom dynamics will become increasingly important in the future, given recent studies that have highlighted the relationship between climate change and increased cyanobacteria blooms in shallow lakes (Kosten et al.

2011; Taranu et al. 2012).

Acknowledgements

We thank M. McDonald, A. Gonzalez, and J. Zimmerman for help with sampling and data collection, and the Vancouver Lake Sailing Club for lake access. This research was partially supported by Grant# 06HQGR0126 from the United States Geological Survey (USGS) to G.R.B and S.M.B, through the State of Washington Water Research Center. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the USGS.

Additional support came from the Vancouver Lake Watershed Partnership and NSF ULTRA-EX grant 09-48983 to S.M.B and G.R.B. The lead author also received additional funding from the

Robert Lane Fellowship in Environmental Science through Washington State University, and a

GK-12 Fellowship from the National Science Foundation STEM Fellows in K-12 Education grant (DGE 07-42561) awarded to G.R.B. and S.M.B.

References

Adámek, Z., & Maršálek, B. (2013). Bioturbation of sediments by benthic macroinvertebrates

and fish and its implication for pond ecosystems: A review. Aquaculture International,

21, 1–17.

Anderson, D. M., Glibert, P. M., & Burkholder, J. M. (2002). Harmful algal blooms and

eutrophication: Nutrient sources, composition, and consequences. Estuaries and Coasts,

25, 704–726.

American Public Health Association (APHA), American Water Works Association, Water

Environment Federation, (2012). Standard Methods for the Examination of Water and

Wastewater , 22nd edition. American Public Health Association: Washington, D.C.

Arhonditsis, G., Brett, M. T., & Frodge, J. (2003). Environmental control and limnological

impacts of a large recurrent spring bloom in Lake Washington, USA. Environmental

Management, 31, 603–618.

Axler, R. P., Redfield, G. W., & Goldman, C. R. (1981). The importance of regenerated nitrogen

to phytoplankton productivity to phytoplankton productivity in a subalpine lake. Ecology,

62, 345–354.

Bennion, H., & Smith, M. A. (2000). Variability in the water chemistry of shallow ponds in

southeast England, with special reference to the seasonality of nutrients and implications

for modelling trophic status. Hydrobiologia, 436, 145–158.

Bhagat, S. K., & Orsborn, J. F. (1971). Summary report on water quantity and quality studies of

Vancouver Lake, Washington. Pullman, Washington: Washington State University.

http://aquaticcommons.org/1457/.

Blomqvist, P., Petterson, A., & Hyenstrand, P. (1994). Ammonium-nitrogen: A key regulatory

factor causing dominance of non-nitrogen-fixing cyanobacteria in aquatic systems.

Archiv für Hydrobiologie, 132, 141–164.

Bollens, S., & Rollwagen-Bollens, G. (2009). Biological Assessment of the Plankton in

Vancouver Lake, WA. Washington State University Vancouver.

http://sites.cityofvancouver.us/vancouverlake/MapsMaterials/WSUYr1Report_Text.pdf.

Boyer, J., Rollwagen-Bollens, G., & Bollens, S. M. (2011). Microzooplankton grazing before,

during and after a cyanobacterial bloom in Vancouver Lake, Washington, USA. Aquatic

Microbial Ecology, 64, 163–174.

Caraco, N. F., Cole, J. J., & Likens, G. E. (1993). Sulfate control of phosphorus availability in

lakes. Hydrobiologia, 253, 275–280.

Carmichael, W. W. (1992). Cyanobacteria secondary metabolites - the cyanotoxins. Journal of

Applied Bacteriology, 72, 445–459.

Chorus, I., Falconer, I. R., Salas, H. J., & Bartram, J. (2000). Health risks caused by freshwater

cyanobacteria in recreational waters. Journal of Toxicology and Environmental Health

Part B: Critical Reviews, 3, 323–347.

Clarke, K. (1993). Nonparametric multivariate analyses of changes in community structure.

Australian Journal of Ecology, 18, 117–143.

Cloern, J. E. (1991). Tidal stirring and phytoplankton bloom dynamics in an estuary. Journal of

Marine Research, 49, 203–221.

Cook, P. L. M., Holland, D. P., & Longmore, A. R. (2010). Effect of a flood event on the

dynamics of phytoplankton and biogeochemistry in a large temperate Australian lagoon.

Limnology and Oceanography, 55, 1123.

Cottingham, K. L., & Carpenter, S. R. (1998). Population, community, and ecosystem variates as

ecological indicators: Phytoplankton responses to whole-lake enrichment. Ecological

Applications, 8, 508–530.

Davelaar, D. (1993). Ecological significance of bacterial polyphosphate metabolism in

sediments. Hydrobiologia, 253, 179–192.

Davis, T. W., Harke, M. J., Marcoval, M. A., Goleski, J., Orano-Dawson, C., Berry, D. L., &

Gobler, C. J. (2010). Effects of nitrogenous compounds and phosphorus on the growth of

toxic and non-toxic strains of Microcystis during cyanobacterial blooms. Aquatic

Microbial Ecology, 61, 149–162.

De Nobel, W. T., Huisman, J., Snoep, J. L., & Mur, L. R. (1997). Competition for phosphorus

between the nitrogen-fixing cyanobacteria Anabaena and Aphanizomenon. FEMS

Microbiology Ecology, 24, 259–267.

De Nobel, W. T., Matthijs, H. C. P., von Elert, E., & Mur, L. R. (1998). Comparison of the light-

limited growth of the nitrogen-fixing cyanobacteria Anabaena and Aphanizomenon. New

Phytologist, 138, 579–587.

De Nobel, W. T., Staats, N., & Mur, L. R. (1995). Competition between nitrogen-fixing

cyanobacteria during phosphorus-limited growth. Water Science and Technology, 32, 99–

101.

De Senerpont Domis, L. N., Elser, J. J., Gsell, A. S., Huszar, V. L. M., Ibelings, B. W., Jeppesen,

E., Kosten, S., Mooij, W. M., Roland, F., Sommer, U., Van Donk, E., Winder, M., &

Lürling, M. (2013). Plankton dynamics under different climatic conditions in space and

time. Freshwater Biology, 58, 463–482.

Dodds, W. K., & Priscu, J. C. (1990). A comparison of methods for assessment of nutrient

deficiency of phytoplankton in a large oligotrophic lake. Canadian Journal of Fisheries

and Aquatic Sciences, 47, 2328–2338.

Dufrene, M., & Legendre, P. (1997). Species assemblages and indicator species: The need for a

flexible asymmetrical approach. Ecological Monographs, 67, 345–366.

Edmondson, W. T. (1994). Sixty years of Lake Washington: A curriculum vitae. Lake and

Reservoir Management, 10, 75–84.

Elliott, J. A. (2012). Predicting the impact of changing nutrient load and temperature on the

phytoplankton of England’s largest lake, Windermere. Freshwater Biology, 57, 400–413.

Farnsworth-Lee, L. A., & Baker, L. A. (2000). Conceptual model of aquatic plant decay and

ammonia toxicity for shallow lakes. Journal of Environmental Engineering, 126, 199–

207.

Ferber, L. R., Levine, S. N., Lini, A., & Livingston, G. P. (2004). Do cyanobacteria dominate in

eutrophic lakes because they fix atmospheric nitrogen? Freshwater Biology, 49, 690–708.

Foreman, J. R., Marshall, C. A., & Scheibley, R. W. (2014). Discharge, water temperature, and

selected meterological data for Vancouver Lake, Washington, water years 2011-13. U.S.

Geological Survey Data Series 849, 52 p., http://dx.doi.org.10.3133/ds849.

Fukuhara, H., & Sakamoto, M. (1987). Enhancement of inorganic nitrogen and phosphate release

from lake sediment by tubificid worms and chironomid larvae. Oikos, 48, 312–320.

Gachter, R., Meyer, J. S., & Mares, A. (1988). Contribution of bacteria to release and fixation of

phosphorus in lake sediments. Limnology and Oceanography, 33, 1542–1558.

Glibert, P. M., Kelly, V., Alexander, J., Codispoti, L. A., Boicourt, W. C., Trice, T. M., &

Michael, B. (2008). In situ nutrient monitoring: A tool for capturing nutrient variability

and the antecedent conditions that support algal blooms. Harmful Algae, 8, 175–181.

Guo, L. (2007). Doing battle with the green monster of Taihu Lake. Science, 317, 1166.

Hambright, K. D., Hairston, N. G., Schaffner, W. R., & Howarth, R. W. (2007). Grazer control

of nitrogen fixation: synergisms in the feeding ecology of two freshwater crustaceans.

Archiv für Hydrobiologie, 170, 89–101.

Hampton, S. E., Izmes’eva, L. R., Moore, M. V., Katz, S. L., Dennis, B., & Silow, E. A. (2008).

Sixty years of environmental change in the world’s largest freshwater lake - Lake Baikal,

Siberia. Global Change Biology, 14, 1947–1958.

Heisler, J., Glibert, P. M., Burkholder, J. M., Anderson, D. M., Cochlan, W., Dennison, W. C., et

al. (2008). Eutrophication and harmful algal blooms: A scientific consensus. Harmful

Algae, 8, 3–13.

Hillebrand, H., Dürselen, C. D., Kirschtel, D., Pollingher, U., & Zohary, T. (1999). Biovolume

calculation for pelagic and benthic microalgae. Journal of Phycology, 35, 403–424.

Hoagland, P., Anderson, D. M., Kaoru, Y., & White, A. W. (2002). The economic effects of

harmful algal blooms in the United States: Estimates, assessment issues, and information

needs. Estuaries and Coasts, 25, 819–837.

Hsieh, C. H., Ishikawa, K., Sakai, Y., Ishikawa, T., Ichise, S., Yamamoto, Y., et al. (2010).

Phytoplankton community reorganization driven by eutrophication and warming in Lake

Biwa. Aquatic Sciences-Research Across Boundaries, 72, 467–483.

Hudnell, H. K. (2010). The state of U.S. freshwater harmful algal blooms assessments, policy

and legislation. Toxicon, 55, 1024–1034.

Hupfer, M., & Lewandowski, J. (2008). Oxygen controls the phosphorus release from lake

sediments – A long-lasting paradigm in limnology. International Review of

Hydrobiology, 93, 415–432.

Jacoby, J. M., & Kann, J. (2007). The occurence and response to toxic cyanobacteria in the

Pacific Northwest, North America. Lake and Reservoir Management, 23, 123–143.

Kokocinski, M., & Soininen, J. (2008). Temporal variation in phytoplankton in two lakes with

contrasting disturbance regimes. Archiv für Hydrobiologie, 171, 39–48.

Koreivienė, J., Anne, O., Kasperovičienė, J., & Burškytė, V. (2014). Cyanotoxin management

and human health risk mitigation in recreational waters. Environmental Monitoring and

Assessment, 186, 4443–4459.

Koseff, J. R., Holen, J. K., Monismith, S. G., & Cloern, J. E. (1993). Coupled effects of vertical

mixing and benthic grazing on phytoplankton populations in shallow, turbid estuaries.

Journal of Marine Research, 51, 843–868.

Kosten, S., Huszar, V. L. M., Bécares, E., Costa, L. S., Donk, E., Hansson, L. A., Jeppesen, E.,

Kruk, C., Lacerot, G., & Mazzeo, N. (2011). Warmer climates boost cyanobacterial

dominance in shallow lakes. Global Change Biology, 18, 118–126.

Krivtsov, V., & Sigee, D. C. (2005). Importance of biological and abiotic factors for

geochemical cycling in a freshwater eutrophic lake. Biogeochemistry, 74, 205–230.

Kruk, C., Huszar, V. L. M., Peeters, E. T. H. M., Bonilla, S., Costa, L., Lürling, M., Reynolds,

C., & Scheffer, M. (2010). A morphological classification capturing functional variation

in phytoplankton. Freshwater Biology, 55, 614–627.

Landers, D. H. (1982). Effects of naturally senescing aquatic macrophytes on nutrient chemistry

and chlorophyll a of surrounding waters. Limnology and Oceanography, 27, 428–439.

Lee, T. A., Rollwagen-Bollens, G., Bollens, S. M., & Faber-Hammond, J. J. (2015).

Environmental influence on cyanobacteria abundance and microcystin toxin production

in a shallow temperate lake. Ecotoxicology and Environmental Safety, 114, 318–325.

Lehman, J. T. (2011). Nuisance cyanobacteria in an urbanized impoundment: interacting internal

phosphorus loading, nitrogen metabolism, and polymixis. Hydrobiologia, 661, 277–287.

Lund, J. W. G., Kipling, C., & Le Cren, E. D. (1958). The inverted microscope method of

estimating algal numbers and the statistical basis of estimations by counting.

Hydrobiologia, 11, 143–170.

Makarewicz, J. C., Lewis, T. W., & Bertram, P. (1999). Phytoplankton composition and biomass

in the offshore waters of Lake Erie: Pre- and post-Dreissna introduction (1983-1993).

Journal of Great Lakes Research, 25, 135–148.

McCarthy, M. J., James, R. T., Chen, Y., East, T. L., & Gardner, W. S. (2009). Nutrient ratios

and phytoplankton community structure in the large, shallow, eutrophic, subtropical

Lakes Okeechobee (Florida, USA) and Taihu (China). Limnology, 10, 215–227.

McCune, B., & Grace, J. B. (2002). Analysis of Ecological Communities. MjM Software Design.

Menden-Deuer, S., & Lessard, E. J. (2000). Carbon to volume relationships for dinoflagellates,

diatoms, and other protist plankton. Limnology and Oceanography, 45, 569–579.

Mieleitner, J., Borsuk, M., Bürgi, H.-R., & Reichert, P. (2008). Identifying functional groups of

phytoplankton using data from three lakes of different trophic state. Aquatic Sciences, 70,

30–46.

Mielke, P., Berry, K., & Brier, G. (1981). Application of multi-response permutation procedures

for examining seasonal-changes in monthly mean sea-level pressure patterns. Monthly

Weather Review, 109, 120–126.

Mortimer, C. H. (1942). The exchange of dissolved substances between mud and water in lakes.

Journal of Ecology, 30, 147–201.

Neale, P. J., Talling, J. F., Heaney, S. I., Reynolds, C. S., & Lund, J. W. G. (1991). Long time

series from the English Lake District: Irradiance-dependent phytoplankton dynamics

during the spring maximum. Limnology and Oceanography, 36, 751–760.

Paerl, H. (2008). Nutrient and other environmental controls of harmful cyanobacterial blooms

along the freshwater–marine continuum. In H. K. Hudnell (Ed.), Cyanobacterial Harmful

Algal Blooms: State of the Science and Research Needs (pp. 217–237). New York:

Springer.

Paerl, H. W., & Huisman, J. (2009). Climate change: a catalyst for global expansion of harmful

cyanobacterial blooms. Environmental Microbiology Reports, 1, 27–37.

Palmer-Felgate, E. J., Mortimer, R. J. G., Krom, M. D., Jarvie, H. P., Williams, R. J., Spraggs, R.

E., & Stratford, C. J. (2011). Internal loading of phosphorus in a sedimentation pond of a

treatment wetland: Effect of a phytoplankton crash. Science of the Total Environment,

409, 2222–2232.

Prescott, G., Bamrick, J., Cawley, E., & Jaques, W. (1978). How to Know the Freshwater Algae

(3rd ed.). McGraw-Hill Science/Engineering/Math.

Rantajarvi, E., Gran, V., Hallfors, S., & Olsonen, R. (1998). Effects of environmental factors on

the phytoplankton community in the Gulf of Findland - Unattended high frequency

measurements and multivariate analyses. Hydrobiologia, 363, 127–139.

Reynolds, C., Dokulil, M., & Padisák, J. (2000). Understanding the assembly of phytoplankton

in relation to the trophic spectrum: Where are we now? Hydrobiologia, 424, 147–152.

Reynolds, C. S. (2002). Towards a functional classification of the freshwater phytoplankton.

Journal of Plankton Research, 24, 417–428.

Reynolds, C. S. (2006). Ecology of phytoplankton. New York: Cambridge University Press.

http://site.ebrary.com/id/10129230.

Riddolls, A. (1985). Aspects of nitrogen fixation in Lough Neagh. I. Acetylene reduction and

frequency of Aphanizomenon flos-aquae heterocysts. Freshwater Biology, 15, 289–297.

Rollwagen-Bollens, G., Bollens, S. M., Gonzalez, A., Zimmerman, J., Lee, T., & Emerson, J.

(2013). Feeding dynamics of the copepod Diacyclops thomasi before, during and

following filamentous cyanobacteria blooms in a large, shallow temperate lake.

Hydrobiologia, 705, 101–118.

Rothenberger, M. B., Burkholder, J. A. M., & Wentworth, T. R. (2009). Use of long-term data

and multivariate ordination techniques to identify environmental factors governing

estuarine phytoplankton species dynamics. Limnology and Oceanography, 54, 2107.

Schindler, D. W. (1987). Detecting ecosystem responses to anthropogenic stress. Canadian

Journal of Fisheries and Aquatic Sciences, 44(Suppl. 1), 6–25.

Sellner, K. G., Doucette, G. J., & Kirkpatrick, G. J. (2003). Harmful algal blooms: Causes,

impacts and detection. Journal of Industrial Microbiology & Biotechnology, 30, 383–

406.

Sommer, U., Gliwicz, Z. M., Lampert, W., & Duncan, A. (1986). The PEG-model of seasonal

succession of planktonic events in fresh waters. Archiv für Hydrobiologie, 106, 453–471.

Søndergaard, M., Jensen, J. P., & Jeppesen, E. (2003). Role of sediment and internal loading of

phosphorus in shallow lakes. Hydrobiologia, 506, 135–145.

Spears, B. M., Carvalho, L., Perkins, R., & Paterson, D. M. (2008). Effects of light on sediment

nutrient flux and water column nutrient stoichiometry in a shallow lake. Water Research,

42, 977–986.

Taranu, Z. E., Zurawell, R. W., Pick, F., & Gregory-Eaves, I. (2012). Predicting cyanobacterial

dynamics in the face of global change: The importance of scale and environmental

context. Global Change Biology, 18, 3477–3490.

Thackeray, S. J., Jones, I. D., & Maberly, S. C. (2008). Long-term change in the phenology of

spring phytoplankton: species-specific responses to nutrient enrichment and climatic

change. Journal of Ecology, 96, 523–535.

Thompson, J. K., Koseff, J. R., Monismith, S. G., & Lucas, L. V. (2008). Shallow water

processes govern system-wide phytoplankton dynamics: A field study. Journal of Marine

Systems, 78, 153–166.

Tsugeki, N. K., Urabe, J., Hayami, Y., Kuwae, M., & Nakanishi, M. (2010). Phytoplankton

dynamics in Lake Biwa during the 20th century: Complex responses to climate variation

and changes in nutrient status. Journal of Paleolimnology, 44, 69-83.

Turpin, D. H., Layzell, D. B., & Elrifi, I. R. (1985). Modeling the C economy of Anabaena flos-

aquae: Estimates of establishment, maintenance, and active costs associated with growth

- on NH3, NO3 , and N2. Plant Physiology, 78, 746–752.

Wehr, J. D. (2002). Freshwater Algae of North America: Ecology and Classification. Academic

Press.

Tables and Figures

Table 1. Vancouver Lake phytoplankton taxa present in >5% of samples from 2007-2010.

Bacillariophyceae Chlorophyta continued Achnanthes spp. Monoraphidium spp. Asterionella spp. Oocystis spp. Aulacoseira spp. Pediastrum spp. Aulacoseira spp. (spiral morphology) Scenedesmus acuminatus Cocconeis spp. Scenedesmus dimorphus Cyclotella spp. Scenedesmus linearis Cymbella spp. Scenedesmus quadricauda Epithemia spp. Scenedesmus spp. Fragilaria spp. Schroederia spp. Gyrosigma spp. Sphaerocystis spp. Melosira spp. Staurastrum spp. Navicula spp. Tetrabaena spp. Nitzschia spp. Tetraedron spp. Nitzschia fruticosa Tetrastrum spp. Stephanodiscus spp. Treubaria spp. Surirella spp. Unknown Chlorophyte Synedra spp. Cyanobacteria Unidentified Pennate spp. Anabaena spp. Dinophyta Aphanizomenon spp. Ceratium hirundinella Aphanocapsa spp. Unknown prolate thecate dinoflagellate Aphanothece spp. Unknown round athecate dinoflagellate Chroococcus spp. Unknown prolate thecate dinoflagellate Coelosphaerium spp. Unknown round thecate dinoflagellate Merismopedia spp. Chrysophyta Microcystis spp. Dinobryon spp. Synechococcus spp. Cryptophyte Synura spp. Cryptomonad spp. Euglenozoa Euglenoid Synuraphyceae Mallomonas spp. Chlorophyta Actinastrum spp. Ankistrodesmus spp. Asterococcus spp. Characium spp. Coelastrum spp. Cosmarium spp. Crucigenia spp. Dictyosphaerium spp. Eudorina spp. Golenkinia spp. Gonium spp. Kirchneriella spp.

Table 2. Groups delineated from cluster analysis of phytoplankton biomass. Species associated with each cluster were determined from indicator species analysis (p<0.05). Cyanobacteria species are in bold. Underlined taxa are considered faithful and strongly associated with the cluster.

Cluster Taxa Month/Year 1 Achnanthes, Cyclotella, Euglenoid, Gonium March – June 2007, December 2007 – March 2008

2 Actinastrum, Anabaena, Ankistrodesmus, July – October 2007, June – November 2008, Aphanizomenon, Aphanocapsa, Aphanothece, July – November 2009 Coelospherium, Cosmarium, Golenkinia, Microcystis, Nitschia, Nitschia fruticosa, Scenedesmus dimorphus, Scenedesmus sp., Synechococcus, Tetraedon,

3 Athecate dinoflagellate, Characium, March – April 2008, April – May 2009, March – Dictyospherium, Fragellaria, Kirchnerella, April 2010, mid June – mid August 2010 Melosira, Monophoridum, Scenedesmus quadrauca, Schroederia, Tetrastrum,

4 Aulacosira May 2008, June 2009

5 Cryptomonad, Thecate dinoflagellate December 2008 – March 2009, December 2009 – February 2010

6 Aulacosira(spiral), Crucigenia, Oocystis, May – mid July 2010, mid August – September Stephanodiscus 2010

Table 3. Sample size (n), vector cutoff for environmental variables (r2), and stress values for each of five NMDS analyses performed. The % of total variance that is explained is provided parenthetically below each axis. A three dimensional solution was reached for each NMDS, except for 2007 and 2008-2009 biomass analyses. The instability value for each analysis was

10-5. The r2 value for each environmental variable associated with each axis, and which met the cutoff criteria, are provided. Although NO2-N was included as an environmental variable, it never met the cutoff criteria for any of the analyses.

2007-2010 2007-2010 2007 2008 - 2009 2010 three week lag

n=25, r2>0.2, n=65, r2>0.2, n=118, r2>0.1, stress=17 stress=16 stress=18 n=29, r2>0.2, stress=12 n=87, r2>0.1, stress=14

Axis Axis Axis Axis Axis Environmental 1 2 3 1 2 1 2 1 2 3 1 2 3 variable (14%) (39%) (27%) (61%) (21%) (34%) (50%) (34%) (28%) (21%) (28%) (16%) (42%)

Total lake depth - 0.36 0.39 0.40 0.32 - 0.56 - 0.37 - 0.18 - 0.22 Secchi depth - 0.40 0.39 0.56 - - 0.72 - - - 0.17 - 0.36 Temperature - - 0.14 - 0.29 0.25 0.26 0.36 - 0.37 - - 0.40 DO - - - - 0.38 - - - 0.29 - - - -

PO4-P - 0.26 - 0.36 - - 0.32 - - - 0.13 - 0.16

SiO4-Si - 0.40 - 0.51 - - 0.29 0.29 - 0.39 0.13 - -

NO3-N - - - 0.27 - 0.20 - - - 0.30 0.10 - -

NH4-N - 0.12 - 0.49 - - 0.20 - - 0.24 - - 0.20 DIN:DIP - 0.17 0.21 0.33 - - 0.37 - - 0.36 - - 0.11

Figure 1: Vancouver Lake is located in Vancouver, Washington, within the greater Portland,

OR-Vancouver, WA metropolitan area. Inflows to Vancouver Lake are the flushing channel

(located southwest ‘corner’), Burnt Bridge Creek (located southeast ‘corner’), Lake River

(located at the northern end), and Salmon Creek (flows into Lake River). The star represents the sampling station at Vancouver Lake Sailing Club

Figure 2: a Absolute and b relative phytoplankton abundance from March 2007 through

September 2010. Each species was classified into one of five different groups: diatoms

(Bacilliriophyceae), dinoflagellates (Dinophyta), flagellates (Chrysophyta, Cryptophyta,

Euglenozoa, and Synuraphyceae), chlorophytes (Chlorophyta), or cyanobacteria.

Figure 3: a Absolute and b relative phytoplankton biomass from March 2007 through

September 2010. Each individual species was classified into one of five different groups (as described in Fig. 2).

Figure 4: Seasonal succession of each phytoplankton cluster determined from cluster analysis.

Each blocked area represents the time period over which a specific cluster was identified.

Figure 5: Environmental data from May 2007 through September 2010: a total water depth and

Secchi depth; b temperature and DO; c SiO4-Si and NO2-N; d NO3-N and NH4-N; e PO4-P and molar DIN:DIP; Boxed areas represent cyanobacteria bloom events delineated from cluster analysis of phytoplankton abundances

Figure 6: Results of each ordination examining relationships between phytoplankton groups and environmental variables. Each point represents a specific sampling date and is demarcated by cluster group. The distance between points represents the amount of similarity or dissimilarity in phytoplankton community composition. Vectors are environmental variables associated with each cluster. a Ordination of the entire dataset (2007 – 2010). b Ordination of 2007 representing clusters 1, 2, and 6. c Ordination representing 2008 – 2009. d 2010 ordination results.

The influence of water quality variables on cyanobacterial blooms and phytoplankton

community composition in a shallow temperate lake

Tammy A. Lee*1, Gretchen Rollwagen-Bollens1, Stephen M. Bollens1

1School of the Environment, Washington State University, 14204 NE Salmon Creek Avenue,

Vancouver, Washington 98686, USA

*Corresponding author: [email protected], phone: (360) 546-9134, fax: (360) 546-9064

Environmental Monitoring and Assessment - Electronic Supplementary Material

a 100

Relative biomassRelative 20

0 100 b

Relative abundanceRelative 20

Jul-07 Jul-08 Jul-09 Jul-10 Apr-07May-07 Jun-07 Aug-07Sep-07 Oct-07 Nov-07 Dec-07 Jan-08 Feb-08 Mar-08 Apr-08 May-08 Jun-08 Aug-08Sep-08 Oct-08 Nov-08 Dec-08 Jan-09 Feb-09 Mar-09 Apr-09 May-09 Jun-09 Aug-09Sep-09 Oct-09 Nov-09 Dec-09 Jan-10 Feb-10 Mar-10 Apr-10 May-10 Jun-10 Aug-10Sep-10 Oct-10

Anabaena Aphanizomenon Aphanocapsa Aphanothece Chroococcus Coelosphaerium Merismopedia Microcystis Synechoccus

Online Resource 1 Cyanobacteria genera relative abundance a, and biomass b, from March

2007 through September 2010.

CHAPTER THREE

ENVIRONMENTAL INFLUENCE ON CYANOBACTERIAL ABUNDANCE AND

MICROCYSTIN TOXIN PRODUCTION IN A SHALLOW

TEMPERATE LAKE

Tammy A. Lee#1, Gretchen Rollwagen-Bollens1, Stephen M. Bollens1, Joshua J. Faber-

Hammond1,2

1School of the Environment, Washington State University, 14204 NE Salmon Creek Avenue,

Vancouver, Washington 98686, USA

2Current Address: Department of Biology, Portland State University, PO Box 751, Portland OR

97207

#Corresponding Author. Email: [email protected] Phone: 1 360 546 9134

Abstract

The increasing frequency of harmful cyanobacterial blooms in freshwater systems is a commonly recognized problem due to detrimental effects on water quality. Vancouver Lake, a shallow, tidally influenced lake in the flood plain of the Columbia River within the city of

Vancouver, WA, USA, has experienced numerous summertime cyanobacterial blooms, dominated by Aphanizomenon and Anabaena. Cyanobacteria abundance and toxin (microcystin) levels have been monitored in this popular urban lake for several years; however, no previous studies have identified which cyanobacteria species produce toxins, nor analyzed how changes in environmental variables contribute to the fluctuations in toxic cyanobacteria populations. We used a suite of molecular techniques to analyze water samples from Vancouver Lake over two

52 summer bloom cycles (2009 and 2010). Both intracellular and extracellular microcystin concentrations were measured using an ELISA kit. Intracellular microcystin concentrations exceeded WHO guidelines for recreational waters several times throughout the sampling period.

PCR results demonstrated that Microcystis sp. was the sole microcystin-producing cyanobacteria species present in Vancouver Lake, although Microcystis sp. was rarely detected in microscopical counts. qPCR results indicated that the majority of the Microcystis sp. population contained the toxin-producing gene (mcyE), although Microcystis sp. abundance rarely exceeded

1% of overall cyanobacteria abundance. Non-metric multidimensional scaling (NMDS) revealed that PO4-P was the main environmental variable influencing the abundance of toxic and non- toxic cyanobacteria, as well as intracellular microcystin concentrations. Our study underscores the importance of using molecular genetic techniques, in addition to traditional microscopy, to assess the importance of less conspicuous species in the dynamics of harmful algal blooms.

Keywords: qPCR; cyanobacteria; ELISA; NMDS; MIC 16S; mcyE

1. Introduction

Harmful cyanobacterial blooms are an increasingly common occurrence in eutrophied aquatic systems, and may result in a broad range of environmental, social, and economic consequences. For example, cyanobacterial blooms can increase oxygen demand, which may lead to localized incidents of hypoxic or anoxic conditions that lead to fish kills (Anderson et al.,

2002; Paerl, 2008). Surface blooms can also block light from reaching benthic primary producers, which may then adversely affect food web dynamics dependent on lake-bottom habitat (Bricelj and Lonsdale, 1997; Gallegos and Bergstrom, 2005). Social and economic

53 impacts of cyanobacterial blooms include negative effects on recreational opportunities due to closure of affected areas, reduced local fisheries, and increased water treatment costs (Hoagland et al., 2002; Paerl, 2008).

In addition to their broader ecological impacts, cyanobacteria are known to produce a suite of secondary metabolites that include hepatotoxins, neurotoxins, and dermatotoxic compounds. These toxins have been linked to decreased water quality and detrimental effects on higher trophic levels (Leonard and Paerl, 2005; Ferrão-Filho et al. 2009), as well as small animal mortality and illness (Boyer, 2007; Jacoby and Kann, 2007), and adverse health risks to humans

(Paerl, 2008). A complex suite of interacting environmental factors influence cyanobacteria bloom dynamics and toxin production. These include abiotic factors such as nutrient inputs

(Downing et al., 2001; Elliott, 2012; Heisler et al., 2008), hydrodynamics (i.e. wind-driven currents, turbulence, and stratification) (Fortin et al., 2010; Hotto et al., 2007), light availability

(Cires et al., 2011; Renaud et al., 2011), and temperature (Cires et al., 2011; Davis et al., 2009), but also biotic interactions such as competition and grazing (Elser, 1999; Ger et al., 2010; Gobler et al., 2007).

Not all cyanobacteria species produce toxins, therefore being able to identify and quantify the abundance of toxin and non-toxin-producing cyanobacteria is essential to understanding toxic cyanobacteria bloom dynamics. In particular, toxin and non-toxin producing individuals of the same species often coexist at the same time (Davis et al., 2009; Otsuka et al.,

1999; Rantala et al., 2006), and toxin producing individuals cannot be morphologically distinguished from non-toxin producing individuals (Dittmann et al., 1997). Moreover, cyanotoxins are not species specific: several different species are known to produce the same toxin and the regulation of toxin production varies considerably as environmental conditions

54 change. Without identifying the toxin-producing species, managing for toxic cyanobacteria bloom effects on water resources may be ineffective.

Vancouver Lake is a large, temperate, shallow, non-stratifying lake located in the floodplain of the Columbia River, within the city limits of Vancouver, Washington, USA.

Vancouver Lake is a local recreational destination serving residents throughout the Portland

(OR)-Vancouver (WA) Metropolitan area. It also serves as critical habitat for wildlife and migratory waterfowl. Cyanobacterial blooms have been documented in Vancouver Lake since

1979, although anecdotal references suggest blooms also occurred prior to that (Bhagat and

Orsborn, 1971). In recent years the lake has often been closed during late summer due to cyanobacterial blooms and the presence of toxins. Although the overall plankton community of

Vancouver Lake has been studied and documented (Boyer et al. 2011, Rollwagen-Bollens et al.

2103), no studies have investigated the identity of the microcystin-producing cyanobacteria and the potential water quality variables associated with toxin production.

More broadly, we are interested in Vancouver Lake as a model system to elucidate cyanobacteria bloom dynamics in shallow, temperate lakes generally. Thus the objectives of this study were to use molecular genetic techniques to identify any microcystin-producing species in Vancouver Lake, and to investigate which environmental variables influence changes in microcystin concentration and the toxin-producing cyanobacteria population.

2. Materials and methods

2.1. Field collection

Sampling for the current study was conducted from the Vancouver Lake Sailing Club dock (Figure 1) on a weekly basis during the periods of May through October 2009, and June through September 2010. A prior study examining the spatial distribution of plankton communities in Vancouver Lake showed no significant differences in plankton abundance or taxonomic composition among eight study sites throughout the lake (Bollens and Rollwagen-

Bollens, 2009). Briefly, eight different sampling sites representing both littoral and limnetic zones were sampled on a quarterly basis for one year to measure spatial and temporal variability in plankton. Although there were significant seasonal differences, there were no significant spatial differences among sites (Kendall’s tau, p>0.5), thus we consider the dock site to be representative of the lake as a whole. On each sampling date, vertical profiles of temperature, dissolved oxygen, pH, and turbidity were measured every 0.2 m from the surface to the bottom using a YSI 91 probe. Total water column depth and Secchi depth were also recorded. Because the lake is very shallow and well-mixed during the summer, we collected surface water samples using a clean bucket, and triplicate subsamples were taken for later nutrient analyses.

Subsamples for DNA extraction and toxin analysis were also collected and kept chilled in a cooler for transport and subsequent analysis in the laboratory. Triplicate 200 mL subsamples were also obtained and preserved in 5% acid Lugol’s solution for microscopical analysis to determine abundance and taxonomic composition of phytoplankton.

2.2. Nutrient analyses

Nutrient samples were filtered through 0.45 µm Millipore disposable filter capsules, stored in plastic bottles and kept refrigerated until analysis. Dissolved nitrate (NO3-N), nitrite

(NO2-N), ammonium (NH4-N), phosphate (PO4-P), and silicate (SiO4-Si) concentrations were

56 analyzed to assess the effects of nutrients on the phytoplankton community generally and the cyanobacteria community specifically. Samples were analyzed by the Marine Chemistry Lab at the University of Washington’s School of Oceanography.

2.3. Microscopical analyses

Lugol’s preserved phytoplankton samples were concentrated in settling chambers (≤ 10 mL) and at least 200 cells were enumerated using an Olympus CK-40 inverted microscope

(400x) and identified at least to genus, and to species whenever possible (Prescott et al., 1978;

Wehr, 2002).

2.4. DNA extraction and purification

For DNA extraction, 250 mL aliquots of lake water were filtered onto 0.45 µm GF/F filters at the time of collection. Filters were kept frozen in polyethylene centrifuge tubes at -

80°C until analyzed. Filters were initially suspended in Buffer ATL (Qiagen, Valencia, CA) and disrupted by three cycles of freezing at -20°C and thawing at 70°C. Proteinase K was added to a final concentration of 50 µg/mL and incubated at 50°C for two hours. DNA was extracted by first adding phenol/chloroform/isoamyl alcohol (25:24:1) (v/v), and then a second chloroform/isoamyl alcohol (24:1) extraction (v/v). DNA was precipitated overnight at -20°C after adding 95% ethanol (v/v) and 9M sodium acetate (0.1% v/v). DNA was collected the following day by centrifugation (16,000 x g, 30 min). A second rinse step was performed using

75% ethanol (v/v) and then centrifuged (16,000 x g, 15 min). DNA pellets were air-dried and resuspended in sterile water. DNA was further purified using GeneClean II (MP Biomedicals).

Concentration and purity of extracted DNA was measured spectrophotometrically (SmartSpec

Plus Spectrophotometer, BioRad).

2.5. PCR

Samples were initially examined to assess the presence of cyanobacteria by targeting 16s rDNA (Neilan et al., 1997), and microcystin-producing cyanobacteria by targeting genera- specific microcystin synthetase gene mcyA (Hisbergues et al., 2003) and genera-specific microcystin synthetase gene mcyE (Rantala et al., 2004). Each assay was carried out in 10 µL reactions and contained 400 µM of each primer, 400 µM dNTPs, 400 µM BSA, 10x buffer

(10%v/v), and 1 unit Taq polymerase (Genscript). PCR protocol for cyanobacteria 16S rDNA was as follows: initial denaturation step of 95°C for 2 min, 35 cycles of 95°C for 30 s, 60°C for

45 s, 72°C for 45 s, and a final extension step at 72°C for 7 min, and then held at 4°C.

Amplification for mcyA and mcyE genes was similar, except that the annealing temperatures were set at 51°C and 55°C, respectively. PCR products were visualized on 1.5% agarose gels stained with 1% GelRed.

Cultures of microcystin-producing Microcystis aeruginosa (UTEX 2385) and non- microcystin-producing Microcystis aeruginosa. (UTEX 2386) were used as positive and negative controls for initial target screening of lake water samples (University of Texas Culture

Collection). These cultures were grown and maintained using 3N Bold liquid medium on a

12:12 hour light:dark light regime. PCR products isolated from lake water samples from three different sampling dates, for each of the mcyA and mcyE procedures (for a total of six products), were sequenced on an ABI3130xl Sanger sequencer (according to the manufacturer’s protocol)

58 and compared among sequences available in GenBank. Sequences were uploaded to GenBank with accession numbers KC603867 and KC603858.

2.6. qPCR

Based on results from initial screening of lake water samples using conventional PCR,

UTEX 2385 (microcystin-producing Microcystis aeruginosa) and UTEX 2386 (non-microcystin- producing Microcystis aeruginosa) were used as external standards to determine cyanobacteria

16S rDNA (Baxa et al., 2010) and Microcystis sp. 16S rDNA (Neilan et al., 1997) gene copy numbers. UTEX 2385 was used as the external standard for determining Microcystis sp.-specific mcyE gene copy numbers (Vaitomaa et al., 2003). Gene copy numbers were calculated according to Viatomaa et al. (2003), however we did not assume two 16s rDNA gene copies per cell. All gene copy numbers for both Microcystis sp. and cyanobacteria genera are reported here as gene copy per ng DNA, due to extraction efficiencies and variable rRNA operon copy number. Throughout the extraction and subsequent clean-up process, we could not assume the final amount of DNA extracted from the filters represented the initial volume of lake water used.

We also did not assume an average of four 16s rDNA copies per cell for quantifying the cyanobacteria population, nor did we assume an average of two 16s rDNA copies per cell for

Microcystis sp. Environmental conditions influence 16s rRNA operon copy numbers in bacteria, unfavorable environmental conditions for some species exhibiting low growth rates will result in fewer rRNA operons, since expression from multiple rRNA operons incurs a metabolic expense of slower growing cells (Klappenbach et al., 2000). Also, using cultured strains derived from a different aquatic system has been shown to differ in operon copy number compared to the same species found in a different environment (Acinas et al., 2004). We assumed one Microcystis sp.

16s rDNA copy number per cell and mcyE copy per cell. Although this can lead to inaccurate estimates of cell abundances (Crosby and Criddle, 2003; Rinta-Kanto et al., 2009, 2005), relative changes in population abundance can still be determined, and the utility of qPCR in monitoring potentially toxin producing Microcystis sp. populations remains high (Martins and Vasconcelos,

2011).

A series of 10-fold dilutions were prepared for standard curve calculations. Standard curves used to calculate gene copy numbers in lake water samples met the minimum requirement of efficiencies ranging from 90 – 110% and r2>0.99. Reactions (20 µL ) were prepared using 1

µL of DNA from extracted standard strains or 2 – 20 ng of DNA from lake water samples, 400

µM BSA, 400 nM of each primer, and 10 µL of Power SYBRGreen (Invitrogen), and then run on an ABI 7500 instrument (Applied Biosystems, SDS software v.2.0.5). We used the same amplification protocol from our conventional PCR for our qPCR. To avoid errors caused by primer dimers we also added a disassociation step from 65 to 95°C after the final extension step for each assay. All lake water samples were run in triplicate.

2.7. Microcystin toxin analyses

Lake water subsamples for toxin analysis were filtered in 50 mL aliquots through 0.22

µm polycarbonate filters. Filters were kept in glass scintillation vials and the filtrates were stored in polyethylene centrifuge tubes. Both filter and filtrate were then kept frozen (-80°C) until analyzed. Intracellular and extracellular microcystin toxin concentrations were quantified using a commercial enzyme-linked immunosorbent assay (ELISA) kit (Beacon Analytical

Systems, Saco, ME). Although the ELISA kit does not distinguish between microcystin-LR

(calibrator) and other variants of microcystin (i.e., cross reactivities among different microcysin

60 variants are possible and noted by the manufacturer), this method is acceptable for monitoring microcystin concentrations and has been used in previous studies (Gurbuz et al., 2012).

Extracellular toxin analysis was performed on the filtrate according to the manufacturer’s protocol. Intracellular toxin analysis was conducted according to a modification of the procedures described in Sevilla et al. (2009). Briefly, intracellular toxins were extracted from phytoplankton cells by submerging the polycarbonate filters for 30 min in a buffer mixture of

80% methanol, 0.1% Tween, and 0.1% trifluoroacetic acid. Aliquots (50 µL) of the supernatant were then diluted 1:20 with 5% phosphate buffered saline and then analyzed with the ELISA kit according to the manufacturer’s protocol.

2.8. Statistical analyses

Non-metric multidimensional scaling (NMDS) was used to examine associations among water quality variables, intra- and extra-cellular microcystin concentrations, and gene copy numbers of each assay. We used NMDS because it is an unconstrained ordination method that does not assume linear relationships and is robust in handling discontinuous and non-normal data

(McCune and Grace, 2002). All ordinations were performed using PC-ORD version 5.0 software. A total of five “species” were included in the analysis: cyanobacteria 16s rDNA gene copy numbers (CYA 16S), Microcystis sp. 16S gene copy numbers (MIC 16S), Microcystis sp.- specific mcyE gene copy numbers (mcyE), intracellular microcystin concentration, and extracellular microcystin concentration. Environmental variables included in the analysis were total water depth, Secchi depth, temperature, dissolved O2, NO2-N, NO3-N, NH4-N, PO4-P, SiO4-

Si, turbidity, and pH. Species abundances were log (x+1) transformed. We used Euclidean distance since we had two different types of variables in representing “species” data: gene copy

61 numbers and microcystin concentrations (Clarke and Warwick, 2001). Due to missing data points of various environmental variables, the NMDS was performed on a subset of data May –

October 2009 and June – September 2010 (n=26).

3. Results

3.1. Physical and nutrient characteristics

Total water depth ranged from 0.91 – 3.1 m in 2009, and 0.97 – 3.2 m in 2010 (Figure

2a). Secchi depth ranged from 0.10 – 0.99 m in 2009, and 0.2 – 0.66 m in 2010 (Figure 2a).

Summer temperatures in 2009 reached as high as 28°C, but in 2010 only reached 23°C (Figure

2b). On average, dissolved O2 was lower in 2009 (5.8 mg/L) than in 2010 (10.4 mg/L) (Figure

2b). In 2009, pH ranged from 8.0 – 10, and ranged between 7.9 – 9.9 in 2010 (Figure 2c).

Turbidity ranged from 11 – 170 NTU in 2009, and in 2010 turbidity ranged from 23 – 130 NTU

(Figure 2c).

During the sampling period in 2009, PO4-P concentrations showed greater variation (1.1

– 240 µg/L) than in 2010 (2.1 – 50 µg/L) (Figure 2d). Similarly, the range of NH4-N concentrations was greater in 2009, when it peaked at 460 µg/L, compared to 2010 when the highest concentration was 160 µg/L (Figure 2d). NO3-N concentrations were lower in 2009, with a maximum concentration of 20 µg/L, than in 2010, with a maximum concentration of 95

µg/L (Figure 2e). NO2-N levels remained relatively low in both 2009 and 2010 (Figure 2e).

4 SiO4-Si availability was greater in 2009, with the maximum concentration of 1.1 x 10 µg/L, than in 2010, when the maximum was 6.3 x 103 µg/L (Figure 2f).

3.2. Microscopic enumeration of cyanobacteria community

In a separate analysis Lee et al. (unpublished) showed that the summer phytoplankton communities in Vancouver Lake were significantly different between 2009 and 2010, largely due to the presence of a distinct cyanobacteria bloom in 2009 that did not occur in 2010. More specifically, in 2009 the cyanobacteria bloom was initially dominated by Anabaena, but then progressed to an Aphanizomenon-dominated cyanobacteria community for the rest of the bloom

(Figure 3a). In 2010, although there was a brief period with increased Aphanizomenon abundance in July and August, the summertime phytoplankton community was dominated by chlorophytes and diatoms (Figure 3a). In both 2009 and 2010 a wide range of cyanobacteria species were observed; however, Anabaena and Aphanizomenon together made up >80% of the relative abundance as determined through traditional light microscopy (Figure 3b).

3.3. Identification and quantification of toxin and non-toxin producing cyanobacteria species

DNA sequence results from PCR amplicons using genera specific mcyA and mcyE primers indicated that a species of Microcystis was the only potential microcystin producer detected in Vancouver Lake. Yet, Microcystis sp. populations were rarely observed microscopically throughout the sampling period (Figure 3b). Because we were unable to reliably detect or monitor Microcystis sp. using traditional microscopy, we were unable to correlate the qPCR results with cell counts. In addition, on several dates we observed Anabaena to account for >50% of the cyanobacteria community; however, we did not detect any molecular signature of Anabaena using either of the genera specific mcyA or mcyE primers. Although only one microcystin-producing cyanobacteria species was detected throughout our samples, we do not

63 exclude the possibility of other cyanobacteria producing microcystin existing in Vancouver Lake

Gene copy numbers for total cyanobacteria, Microcystis sp., and potentially microcystin- producing Microcystis sp. were variable throughout the sampling period in 2009 and 2010

(Figure 3c). qPCR results comparing cyanobacteria 16S rDNA gene copy numbers and

Microcystis sp. 16S rDNA gene copy numbers indicated that Microcystis sp. made up <1% of the total cyanobacteria community throughout the sampling period. Overall, Microcystis sp. gene copy numbers in 2010 were less than in 2009, as were total cyanobacteria gene copy numbers.

Notably, potentially toxin-producing Microcystis sp. was present in every lake water sample except one during the 2009 and 2010 sampling periods, and in most cases made up >50% of the total Microcystis sp. population (Figure 3d).

3.4. Intracellular and extracellular microcystin concentrations

Intracellular microcystin concentrations ranged from undetectable to 15 µg/L in 2009

(Figure 4). As mcyE abundance increased in 2009, intracellular microcystin concentration also increased. Further, intracellular microcystin decreased as mcyE abundance decreased at the beginning of October 2009, and then increased in mid-October when mcyE abundances also increased. In 2010, intracellular microcystin concentrations fluctuated from undetectable to 12

µg/L (Figure 4). Similar to 2009, in 2010 intracellular microcystin increased as mcyE increased.

However, overall intracellular microcystin concentration and mcyE abundances were less in 2010 than in 2009.

Extracellular microcystin concentrations ranged from undetectable to 0.3 µg/L in 2009 and from undetectable to 0.5 µg/L in 2010 (Figure 4). No discernible or significant trends were

64 observed between extracellular microcystin concentrations and intracellular microcystin concentrations or mcyE abundance.

3.5. NMDS results comparing species abundance and environmental variables

Results from NMDS analyses indicated that SiO4-Si, PO4-P, and turbidity were strongly associated with increased toxin and non-toxin producing cyanobacteria populations, and intracellular microcystin concentration (Figure 5). In 2009, SiO4-Si levels increased during the summer and plateaued for the duration of the bloom. In 2010, SiO4-Si levels increased somewhat at the beginning of the period of elevated cyanobacteria concentration, but were variable over the duration of the sampling period. PO4-P levels peaked in August 2009 and gradually decreased throughout that year’s bloom, but in summer 2010 PO4-P levels were lower in magnitude and duration.

In addition, total lake water depth and Secchi depth were inversely associated with toxin and non-toxin producing cyanobacteria populations. As total lake water depth and Secchi depth decreased, total cyanobacteria, Microcystis sp., and microcystin-producing Microcystis sp. gene copy numbers increased. Additionally, NMDS results showed total cyanobacteria gene copy numbers and extracellular microcystin concentrations were not strongly associated with any other variable; intracellular microcystin was most strongly associated with microcystin- producing Microcystis sp., and toxic and non-toxic Microcystis sp. were positively associated with each other.

4. Discussion

In Vancouver Lake, microcystin concentrations have been monitored for several years; however, the cyanobacteria species responsible for producing the toxin had not been positively identified until now. Our previous microscopic examinations of Vancouver Lake plankton assemblages have shown that, since 2006, the annual cyanobacterial blooms have been dominated by Aphanizomenon and Anabaena, with rare observations of Microcystis sp. (Lee at al. unpublished). Surprisingly, in this two-year study using molecular methods, Microcystis sp. was the only microcystin-producing cyanobacteria species detected. Prior to our own studies,

Microcystis sp. was visually observed to be dominant during one summer bloom in 2003, when microcystin levels ranged from <0.5 ug/L to 18.5 ug/L (Jacoby and Kann, 2007), but despite microcystin levels having been detected and reported since then, Microcystis sp. has rarely been observed visually. During the summer of 2011, the Clark County Public Health Department measured microcystin levels periodically throughout the monitoring season, and in one instance levels were high enough to trigger a lake closure for recreational uses, despite the absence of

Microcystis sp. in microscopic counts (Clark County, 2011). This suggests that there can be significant adverse effects on water quality regardless of the absolute or relative abundance of microcystin-producing Microcystis sp.

Therefore, our results demonstrate the utility of using molecular tools to detect and monitor microcystin-producing cyanobacteria. Microcystis sp. was rarely detected when quantified using light microscopy, yet our PCR results indicated that potentially toxin-producing

Microcystis sp. was present during almost all sampling dates. The discrepancy between qPCR and microscopy results may be due to several factors, including preservation and counting methods, and the potential amplification of other species using Microcystis sp.-specific 16S

66 rDNA primers. However, the total MIC 16S gene copy numbers represented <1% of the total

CYA 16S gene copy numbers, suggesting the discrepancy between qPCR and microscopy results may instead be due to the scarcity of the species and therefore a low probability of accurately enumerating the species through microscopy. Although we only monitored one microcystin gene, mcyE, qPCR results suggested the potential microcystin-producing Microcystis sp. population was highly variable.

Our results are similar to observations in Upper Klamath Lake, OR, where an

Aphanizomenon-dominated cyanobacteria bloom (>90% by volume) was shown to contain microcystin-producing Microcystis sp., even though Microcystis sp. was rarely detected via microscopy (Eldridge et al., 2012; Saker et al., 2007). In addition, in Microcystis sp.-dominated cyanobacterial blooms, Rinta-Kanto et al. (2009) showed the proportion of toxin-producing

Microcystis sp. to non-toxin producing Microcystis sp. to be highly variable in Lake Erie, ranging from 0 – 60%. Kardinaal et al. (2007) found microcystin-producing Microcystis sp. dominated total Microcystis sp. populations in two unstratified lakes located in the Netherlands.

Finally, Baxa et al. (2010) showed that microcystin-producing Microcystis sp. makes up less than

20% of the total Microcystis sp. population in the San Francisco Estuary.

Our analyses showed PO4-P was associated with increased populations of toxin- and non- toxin producing Microcystis sp., total cyanobacteria, and intracellular microcystin concentration in Vancouver Lake. Assimilation of phosphorus is critical for cellular function and influences toxin-producing cyanobacteria dynamics, including growth, life cycle stages, and toxin production (Dyhrman, 2008; Jacoby et al., 2000). Previous studies have shown that increased

PO4-P availability positively influences cyanobacteria populations (Rinta-Kanto et al., 2009; Xu et al., 2010) and more specifically toxin and non-toxin producing Microcystis sp. populations

(Davis et al., 2009; Li et al., 2012; Xu et al., 2010). Increased PO4-P availability has also been shown to influence microcystin concentrations in other lake systems (Jacoby et al., 2000; Rinta-

Kanto et al., 2009), and in some cases higher inorganic phosphorous levels have been shown to yield higher microcystin content per cell (Rapala et al., 1997). To better understand the relationship between microcystin and microcystin producing cyanobacteria populations we recommend that future experiments be done to measure gene expression for microcystin- producing cyanobacteria under varying PO4-P availability.

Although our analysis also showed turbidity and SiO4-Si concentrations to be associated with toxin and non-toxin producing Microcystis sp., total cyanobacteria, and intracellular microcystin concentration, we do not suggest that turbidity or SiO4-Si directly influence these relationships. Instead, we suggest these two factors broadly describe the environmental conditions associated with the summer blooms. The lake is very well mixed during the summer months, when total lake depth is at a minimum (<1.5 m). This, along with increased cyanobacteria abundance during the summer, contributes to increased turbidity. With respect to silicate, Rinta-Kanto et al. (2009) showed that SiO4-Si was positively correlated with microcystin concentrations in Lake Erie. Similarly, Aboal et al. (2005) found that increased intracellular microcystin levels were associated with increased silicate concentrations in a river in southeast

Spain. However, neither study has shown nor even suggested that SiO4-Si directly contributes to microcystin production. Instead, SiO4-Si has been linked to the dissolution of diatoms

(Spears et al., 2008), with increased SiO4-Si availability signifying a shift in phytoplankton community composition away from a diatom-dominated community to a cyanobacteria- dominated community (Bennion and Smith, 2000; Aboal et al., 2005).

Total lake depth and Secchi depth were inversely related to SiO4-Si and PO4-P. Secchi depth and lake water depth were strongly associated with each other – as lake water depth decreased, Secchi depth also decreased. These two variables describe seasonal lake hydrology where lake water depth and Secchi depth decrease during the summer months and increase during the winter and spring months during the rainy season and spring snow melt.

Intracellular microcystin concentrations were strongly associated with the abundance of microcystin-producing Microcystis sp., and to a lesser extent associated with the total

Microcystis sp. population (Figure 5). As toxic Microcystis sp. populations increased, intracellular microcystin concentrations also increased. Our results are similar to those of Rinta-

Kanto et al. (2009) in Lake Erie, who found intracellular microcystin concentrations to be correlated with both the total Microcystis sp. population and microcystin-producing Microcystis sp. Davis et al. (2009) also showed microcystin-producing Micocystis sp. were significantly correlated with intracellular microcystin in several lakes in New York state, USA.

Intracellular microcystin concentrations in Vancouver Lake were two to three orders of magnitude higher than extracellular concentrations. This difference may be due to how microcystin is released into the water column and the rate of microcystin degradation.

Microcystin is primarily released through biotic means, including cell lysis, and to a lesser extent by active transport through the cell wall (Rapala et al., 1997). Once released into the water column, microcystin can rapidly degrade (average half-life of one day) due to heterotrophic bacteria (Christoffersen et al., 2002) and other chemical interactions (Tsuji et al., 1994), and is thus unlikely to accumulate in the water column (Christoffersen et al., 2002; Lahti et al., 1997).

However, other research indicates that microcystin can persist in the water column for up to several weeks (Harada et al. 1996; Gągala and Mankiewicz-Boczek, 2012).

In conclusion, Microcystis sp. was the only microcystin-producing cyanobacteria species detected in Vancouver Lake, a species that had previously gone unnoticed due to the limitations of traditional microscopical analysis and the numerical dominance of other cyanobacteria genera, namely Aphanizomenon and Anabaena. Cell counts and species diversity determined by microscopy may provide a general understanding of the phytoplankton community composition; however, without the use of molecular and genetic techniques it is effectively impossible to quantitatively monitor changes in toxic populations that are relatively low in overall abundance.

PO4-P was the main environmental variable associated with increased toxin and non- toxin producing Microcystis sp. gene copy numbers, and intracellular microcystin concentrations. These results are consistent with other Microcystis sp. dominated systems, although studies of lakes of similar physical and phytoplankton community characteristics to that of Vancouver Lake remain underrepresented. While using 16S rDNA to monitor changes in

Microcystis sp. and other cyanobacteria populations may not provide absolute abundance, qPCR still remains an important tool in assessing relative changes in potential toxic cyanobacteria populations. Quantifying mcyE gene copy number, rather than relying solely on visual cell counts, may be a better explanatory metric for overall toxin concentration. As natural resource and public health managers continue to monitor regional and local freshwater systems for toxic cyanobacterial blooms, a multifaceted approach is needed to understand cyanobacteria bloom dynamics on both species-specific and community levels.

Acknowledgements

We thank Drs. Jenefer DeKoning, Steve Sylvester, and Ruth Philips for their invaluable support and resources throughout this project, Julie Zimmerman for helping with the sampling

70 and data collection, and the Vancouver Lake Sailing Club for lake access. This research was partially funded through Grant# 06HQGR0126 from the United States Geological Survey

(USGS) to G.R.B and S.M.B, through the State of Washington Water Research Center. The contents of this manuscript are solely the responsibility of the authors and do not necessarily represent the official views of the USGS. The lead author received additional funding as an NSF

GK-12 Fellow, as part of a STEM Fellows in K-12 Education (GK-12) Fellowship grant (DGE

07-42561) awarded to G.R.B and S.M.B, and a Robert Lane Fellowship in Environmental

Science through Washington State University.

References

Aboal, M., Puig, M.A., Asencio, A.D., 2005. Production of microcystins in calcareous

Mediterranean streams: The Alharabe River, Segura River basin in south-east Spain. J.

Appl. Phycol. 17, 231–243.

Acinas, S.G., Marcelino, L.A., Klepac-Ceraj, V., Polz, M.F., 2004. Divergence and redundancy

of 16S rRNA sequences in genomes with multiple rrn operons. J. Bacteriol. 186, 2629–

2635.

Anderson, D.M., Glibert, P.M., Burkholder, J.M., 2002. Harmful algal blooms and

eutrophication: nutrient sources, composition, and consequences. Estuaries Coasts 25,

704–726.

Baxa, D.V., Tomofumi, K., Ger, K.A., Lehman, P.W., Teh, S.J., 2010. Estimating the abundance

of toxic Microcystis in the San Francisco Estuary using quantitative real-tim PCR.

Harmful Algae 9, 342–349.

Bennion, H., Smith, M.A., 2000. Variability in the water chemistry of shallow ponds in southeast

England, with special reference to the seasonality of nutrients and implications for

modelling trophic status. Hydrobiologia 436, 145–158.

Bhagat, S.K., Orsborn, J.F., 1971. Summary report on water quantity and quality studies of

Vancouver Lake, Washington [WWW Document]. URL

http://aquaticcommons.org/1457/ (accessed 2.12.13).

Boyer, G.L., 2007. The occurrence of cyanobacterial toxins in New York lakes: Lessons from

the MERHAB-Lower Great Lakes program. Lake Reserv. Manag. 23, 153–160.

doi:10.1080/07438140709353918

Bricelj, V.M., Lonsdale, D.J., 1997. Aureococcus anophagefferens: Causes and ecological

consequences of brown tides in U.S. mid-Atlantic coastal waters. Limnol. Oceanogr.,

Part 2: The ecology and oceanography of harmful algal blooms 42, 1023–1038.

Christoffersen, K., Lyck, S., Winding, A., 2002. Microbial activity and bacterial community

structure during degradation of microcystins. Aquat. Microb. Ecol. 27, 125–136.

Cires, S., Wörmer, L., Timon, J., Wiedner, C., Quesada, A., 2011. Cylindrospermopsin

production and release by the potentially invasive cyanobacterium Aphanizomenon

ovalisporum under temperatuer and light gradients. Harmful Algae 10, 668–675.

Clarke, K.R., Warwick, R.M., 2001. Change in Marine Communities: An Approach to Statistical

Analysis and Interpretation. Plymouth Marine Laboratory.

Crosby, L.D., Criddle, C.S., 2003. Understanding bias in microbial community analysis

techniques due to rrn operon copy number heterogeneity. BioTechniques 34, 790–803.

Davis, T.W., Berry, D.L., Boyer, G.L., Gobler, C.J., 2009. The effects of temperature and

nutrients on the growth and dynamics of toxic and non-toxic strains of< i>

Microcystis during cyanobacteria blooms. Harmful Algae 8, 715–725.

Dittmann, E., Neilan, B.A., Erhard, M., von Dohren, H., Börner, T., 1997. Insertional

mutagenesis of a peptide synthetase gene that is responsible for hepatotoxin production in

the cyanobacterium Microcystis aeruginosa PCC 7806. Mol. Microbiolgoy 26, 779–787.

Downing, J.A., Watson, S.B., McCauley, E., 2001. Predicting cyanobacteria dominance in lakes.

Can. J. Fish. Aquat. Sci. 58, 1905–1908.

Dyhrman, S.T., 2008. Molecular approaches to diagnosing nutritional physiology in harmful

algae: Implications for studying the effects of eutrophication. Harmful Algae 8, 167–174.

Eldridge, S.L.., Wood, T.M., Echols, K.R., 2012. Spatial and Temporal Dynamics of

Cyanotoxins and Their Relation to Other Water Quality Variables in Upper Klamath

Lake, Oregon, 2007–09.

Elliott, J.A., 2012. Predicting the impact of changing nutrient load and temperature on the

phytoplankton of England’s largest lake, Windermere. Freshw. Biol. 57, 400–413.

Elser, J.J., 1999. The pathway to noxious cyanobacteria blooms in lakes: The food web as the

final turn. Freshw. Biol. 42, 537–543.

Ferrão-Filho, A.S., Soares, M.C.S., de Freitas Magalhães, V., Azevedo, S.M.F.O., 2009.

Biomonitoring of cyanotoxins in two tropical reservoirs by cladoceran toxicity bioassays.

Ecotoxicol. Environ. Saf. 72, 479–489. doi:10.1016/j.ecoenv.2008.02.002

Fortin, N., Aranda-Rodriguez, R., Jing, H., Pick, F., Bird, D., Greer, C.W., 2010. Detection of

microcystin-producing cyanobacteria in Missisquoi Bay, Quebec, Canada, using

quantitative PCR. Appl. Environ. Microbiol. 76, 5105–5112.

Gallegos, C.L., Bergstrom, P.W., 2005. Effects of a< i> Prorocentrum minimum bloom on

light availability for and potential impacts on submersed aquatic vegetation in upper

Chesapeake Bay. Harmful Algae 4, 553–574.

Ger, K.A., Teh, S.J., Baxa, D.V., Lesmeister, S., Goldman, C.R., 2010. The effects of dietary

Microcystis aeruginosa and microcystin on the copepods of the upper San Francisco

Estuary. Freshw. Biol. 55, 1548–1559. doi:10.1111/j.1365-2427.2009.02367.x

Gobler, C.J., Davis, T.W., Coyne, K.J., Boyer, G.L., 2007. Interactive influences of nutrient

loading, zooplankton grazing, and microcystin synthetase gene expression on

cyanobacterial bloom dynamics in a eutrophic New York lake. Harmful Algae 6, 119–

133. doi:10.1016/j.hal.2006.08.003

Heisler, J., Glibert, P.M., Burkholder, J.M., Anderson, D.M., Cochlan, W., Dennison, W.C.,

Dortch, Q., Gobler, C.J., Heil, C.A., Humphries, E., Lewitus, A., Magnien, R., Marshall,

H.G., Sellner, K., Stockwell, D.A., Stoecker, D.K., Suddleson, M., 2008. Eutrophication

and harmful algal blooms: A scientific consensus. Harmful Algae 8, 3–13.

doi:10.1016/j.hal.2008.08.006

Hisbergues, M., Christiansen, G., Rouhiainen, L., Sivonen, K., Börner, T., 2003. PCR-based

identification of microcystin-producing genotypes of different cyanobacterial genera.

Arch. Microbiol. 180, 402–410.

Hoagland, P., Anderson, D.M., Kaoru, Y., White, A.W., 2002. The economic effects of harmful

algal blooms in the United States: estimates, assessment issues, and information needs.

Estuaries Coasts 25, 819–837.

Hotto, A.M., Satchwell, M.F., Boyer, G.L., 2007. Molecular characterization of potential

microcystin-producing cyanobacteria in Lake Ontario embayments and nearshore waters.

Appl. Environ. Microbiol. 73, 4570–4578.

Jacoby, J.M., Collier, D.C., Welch, E.B., Hardy, F.J., Crayton, M., 2000. Environmental factors

associated with a toxic bloom of Microcystis aeruginosa. Can. J. Fish. Aquat. Sci. 57,

231–240.

Jacoby, J.M., Kann, J., 2007. The occurence and response to toxic cyanobacteria in the Pacific

Northwest, North America. Lake Reserv. Manag. 23, 123–143.

Kardinaal, W.E.A., Janse, I., Kamst-van Agterveld, M., Meima, M., Snoek, J., Mur, L.R.,

Huisman, J., Zwart, G., Visser, P.M., 2007. Microcystis genotype succession in relation

to microcystin concentrations in freshwater lakes. Aquat. Microb. Ecol. 48, 1–12.

Klappenbach, J.A., Dunbar, J.M., Schmidt, T.M., 2000. rRNA operon copy number reflects

ecological strategies of bacteria. Appl. Environ. Microbiol. 66, 1328–1333.

Lahti, K., Rapala, J., Färdig, M., Niemelä, M., Sivonen, K., 1997. Persistence of cyanobacterial

hepatotoxin, microcystin-LR in particulate material and dissolved in lake water. Water

Res. 31, 1005–1012.

Leonard, J.A., Paerl, H.W., 2005. Zooplankton community structure, micro-zooplankton grazing

impact, and seston energy content in the St. Johns river system, Florida as influenced by

the toxic cyanobacterium Cylindrospermopsis raciborskii. Hydrobiologia 537, 89–97.

doi:10.1007/s10750-004-2483-9

Li, D., Kong, F., Shi, X., Ye, L., Yu, Y., Yang, Z., 2012. Quantification of microcystin-

producing and non-microcystin producing< i> Microcystis populations during the

2009 and 2010 blooms in Lake Taihu using quantitative real-time PCR. J. Environ. Sci., J

Environ Sci 24, 284–290.

Martins, A., Vasconcelos, V., 2011. Use of qPCR for the study of hepatotoxic cyanobacteria

population dynamics. Arch. Microbiol. 193, 615–627.

McCune, B., Grace, J.B., 2002. Analysis of Ecological Communities. MjM Software Design.

Neilan, B.A., Jacobs, D., Blackall, L.L., Hawkins, P.R., Cox, P.T., Goodman, A.E., 1997. rRNA

sequences and evolutionary relationships among toxic and nontoxic cyanobacteria of the

genus Microcystis. Int. J. Syst. Bacteriol. 47, 693–697.

Neilan, B.A., Stuart, J.L., Goodman, A.E., Cox, P.T., Hawkins, P.R., 1997. Specific

Amplification and Restriction Polymorphisms of the Cyanobacterial rRNA Operon

Spacer Region. Syst. Appl. Microbiol. 20, 612–621. doi:10.1016/S0723-2020(97)80033-

Otsuka, S., Suda, S., Li, R., Watanabe, M., Oyaizu, H., Matsumoto, S., Watanabe, M.M., 1999.

Phylogenetic relationships between toxic and non-toxic strains of the genus

Microcystis based on 16S to 23S internal transcribed spacer sequence. FEMS

Microbiol. Lett. 172, 15–21. doi:10.1111/j.1574-6968.1999.tb13443.x

Paerl, H., 2008. Nutrient and other environmental controls of harmful cyanobacterial blooms

along the freshwater–marine continuum. Cyanobacterial Harmful Algal Blooms State

Sci. Res. Needs 217–237.

Prescott, G., Bamrick, J., Cawley, E., Jaques, W., 1978. How to Know the Freshwater Algae, 3rd

ed. McGraw-Hill Science/Engineering/Math.

Rantala, A., Fewer, D.P., Hisbergues, M., Rouhiainen, L., Vaitomaa, J., Börner, T., Sivonen, K.,

2004. Phylogenetic evidence for the early evolution of microcystin synthesis. Proc. Natl.

Acad. Sci. U. S. A. 101, 568–573.

Rantala, A., Rajaniemi-Wacklin, P., Lyra, C., Lepisto, L., Rintala, J., Mankiewicz-Boczek, J.,

Sivonen, K., 2006. Detection of microcystin-producing cyanobacteria in Finnish lakes

with genus-specific micorcystin synthetase gene E (mcyE) PCR and associations with

environmental factors. Appl. Environ. Microbiol. 72, 6101–6110.

Rapala, J., Sivonen, K., Lyra, C., Niemelä, S.I., 1997. Variation of microcystins, cyanobacterial

hepatotoxins, in Anabaena spp. as a function of growth stimuli. Appl. Environ.

Microbiol. 63, 2206–2212.

Renaud, S.L.B., Pick, F.R., Fortin, N., 2011. Effect of light intensity on the relative dominance

of toxigenic and nontoxigenic strains of Microcystis aeruginosa. Appl. Environ.

Microbiol. 77, 7016–7022.

Rinta-Kanto, J.M., Konopko, E.A., DeBruyn, J.M., Bourbonniere, R.A., Boyer, G.L., Wilhelm,

S.W., 2009. Lake Erie< i> Microcystis: Relationship between microcystin

production, dynamics of genotypes and environmental parameters in a large lake.

Harmful Algae 8, 665–673.

Rinta-Kanto, J.M., Ouellette, A.J.A., Boyer, G.L., Twiss, M.R., Bridgeman, T.B., Wilhelm,

S.W., 2005. Quantification of toxic Microcystis spp. during the 2003 and 2004 blooms in

western Lake Erie using quantitative real-time PCR. Environ. Sci. Technol. 39, 4198–

4205.

Saker, M.L., Welker, M., Vasconcelos, V.M., 2007. Multiplex PCR for the detection of toxigenic

cyanobacteria in dietary supplements produced for human consumption. Appl. Microbiol.

Biotechnol. 73, 1136–1142.

Sevilla, E., Smienk, H., Razquin, P., Mata, L., Peleato, M.L., 2009. Optimization of intracellular

microcystin-LR extraction for its analysis by protein phosphatase inhibition assay. Water

Sci. Technol. 60, 1903–1909.

Spears, B.M., Carvalho, L., Perkins, R., Paterson, D.M., 2008. Effects of light on sediment

nutrient flux and water column nutrient stoichiometry in a shallow lake. Water Res. 42,

977–986.

Tsuji, K., Naito, S., Kondo, F., Ishikawa, N., Watanabe, M.F., Suzuki, M., Harada, K., 1994.

Stability of microcystins from cyanobacteria: effect of light on decomposition and

isomerization. Environ. Sci. Technol. 28, 173–177.

Vaitomaa, J., Rantala, A., Halinen, K., Rouhiainen, L., Tallberg, P., Mokelke, L., Sivonen, K.,

2003. Quantitative real-time PCR for determination of microcystin synthetase E copy

numbers for Microcystis and Anabaena in lakes. Appl. Environ. Microbiol. 69, 7289–

7297.

Wehr, J.D., 2002. Freshwater Algae of North America: Ecology and Classification. Academic

Press.

Xu, Y., Wang, G., Yang, W., Li, R., 2010. Dynamics of the water bloom-forming Microcystis

and its relationship with physicochemical factors in Lake Xuanwu (China). Environ. Sci.

Pollut. Res. 17, 1581–1590.

Figures

Figure 1: Vancouver Lake, Washington, is located just north of Portland, Oregon. The filled circle represents the sampling station at Vancouver Lake Sailing Club.

Figure 2: Environmental variables for 2009 and 2010: (a) total water depth and Secchi depth,

(b) temperature and dissolved O2 (DO), (c) pH and turbidity, (d) PO4-P and NH4-N, (e) NO3-N and NO2-N, (f) SiO4-Si. Boxed areas represent the sampling period and data used for 2009 and

2010 analyses. *Although this value falls outside the analyzed sampling period, it is recognized as a potentially erroneous value.

Figure 3: Vancouver Lake water samples: (a) the absolute abundance of cyanobacteria in relation to other phytoplankton groups during the 2009 and 2010 sampling period based on microscopical counts, (b) relative abundance of cyanobacteria species found in Vancouver Lake,

(c) gene copy numbers for MIC 16S, MCYE, and CYA 16S, (d) relative abundances of MIC 16S and MCYE gene copy numbers. Boxed areas represent the sampling period for 2009 and 2010.

Figure 4: Intracellular and extracellular microcystin concentrations for 2009 and 2010.

Figure 5: NMDS ordination of gene copy numbers of toxic (MCYE) and non-toxic Microcystis sp. (MIC 16S), cyanobacteria (CYA 16S), and intra- (Intra) and extra-cellular (Extra) microcystin concentrations. Each point represents a sampling date. Axis 1 represented 56.5% of the observed variance and axis 2 represented 43.3% of the observed variance. Vectors are environmental variables associated with species points (r2>0.2).

CHAPTER FOUR

THE EFFECTS OF EUTROPHICATION AND INVASIVE SPECIES ON

ZOOPLANKTON COMMUNITY DYNAMICS IN A

SHALLOW TEMPERATE LAKE

Tammy A. Lee*1, Stephen M. Bollens1,2, Gretchen Rollwagen-Bollens1,2, and Joshua E.

Emerson1

*Corresponding author: [email protected]

1 School of the Environment, Washington State University, 14204 NE Salmon Creek Avenue,

Vancouver, Washington 98686, USA

2 School of Biological Sciences, Washington State University, 14204 NE Salmon Creek Avenue,

Vancouver, Washington 98686, USA

Keywords

Invasive zooplankton, cyanobacterial bloom, water quality, environmental stressors, shallow lake

Abstract

Eutrophication (and associated cyanobacterial blooms) and biological invasions are increasingly common problems in aquatic ecosystems, yet their effects on zooplankton community dynamics are not well understood. We examined zooplankton community dynamics from 2005 to 2011 in a tidally-influenced shallow temperate lake (Vancouver Lake, Washington,

USA), with particular emphasis on the effects of eutrophication and biological invasions. Cluster analysis, indicator species analysis, and non-metric multidimensional scaling analyses were used to explore interactions between the zooplankton community and multiple environmental

85 stressors. Our results suggest that interannual differences in seasonal zooplankton community succession may be influenced directly by turbidity, cyanobacterial blooms, predatory zooplankton, and invasive crustacean zooplankton, and indirectly by PO4-P availability and temperature. Based on these results, we suggest that two separate management goals – alleviating eutrophication and managing the spread of invasive species – may be in conflict. We recommend future studies of competition between native and non-native species to better understand the effects of cyanobacterial blooms on the success of non-native species, and the potential long-term consequences of non-native species invasions on zooplankton community dynamics.

Introduction

Freshwater ecosystems are under assault by multiple stressors, such as eutrophication arising from changing land use practices, and invasions of non-native species arising from increased commercial shipping and recreational boating. Over the past decade or more, there have been many studies investigating the effects of increased nutrient availability on freshwater food web dynamics and restoration efforts (Jeppesen et al. 1997, Boersma et al. 2008, Glibert et al. 2011). One prominent consequence of eutrophication is the increased intensity and frequency of cyanobacterial blooms (Paerl 2008). Cyanobacterial blooms can in turn affect zooplankton community dynamics by adversely altering the quantity and quality of food available

(Christoffersen et al. 1990, Gulati & Demott 1997, Davis et al. 2012, Rollwagen-Bollens et al.

2013) and shifting predation pressure on zooplankton by higher trophic levels (e.g. by altering planktivore feeding behavior and intensity) (Jeppesen et al. 1997, Engström-Öst et al. 2006).

Thus, eutrophication can have profound and far-reaching effects on zooplankton communities.

Aquatic invasive species can also significantly alter trophic interactions in planktonic food webs (Hooff & Bollens 2004, Strecker & Arnott 2008, Strecker et al. 2011). For example, the invasion of the cladoceran Bythotrephes longimanus in Canadian lakes resulted in an overabundance of planktivorous predators which led to a decline in zooplankton prey diversity

(Strecker & Arnott 2008, Kelly et al. 2013b). Likewise, the introduction of several species of

Asian copepods has caused vast changes in the plankton communities of the San Francisco

Estuary (Bollens et al. 2002, 2011, 2014, Hooff & Bollens 2004). Yet another example of an aquatic invader is the ctenophore Mnemiopsis leidyi, whose introduction to the Black Sea in the

1980s and expansion throughout the Caspian Sea in the 1990s caused declines in the abundance and diversity of native plankton (Roohi et al. 2008, 2010) and zooplanktivorous fish (Shiganova et al. 2004, Daskalov & Mamedov 2007). Thus, invasive zooplankton can have substantial effects on native zooplankton communities that may ramify throughout aquatic food webs.

Zooplankton are important intermediaries between primary producers and higher trophic level organisms in aquatic ecosystems, by mediating algal growth through grazing and being consumed by zooplanktivorous invertebrates and fishes. Therefore, changes in zooplankton communities over time may serve as important indicators of the effects of eutrophication

(Søndergaard et al. 2005, Winder et al. 2009, Jeppesen et al. 2011), as well as the effects of biological invasions (Kelly et al. 2013b, Palmer & Yan 2013).

Variation in zooplankton community composition may also be an important factor to consider in the monitoring and management of freshwater ecosystems (Dijkstra et al. 2011).

Freshwater systems are experiencing dramatic changes in land use and development activities, at the same time as climate shifts influence environmental conditions of these systems and their surrounding watersheds. Observational studies and field surveys are essential to investigating

87 and understanding the cumulative effects of multiple stressors on aquatic communities (Dodds et al. 2012, Palmer & Yan 2013), and are a critical complement to, and extension of, experimental programs that are often limited to testing the effects of a single stressor. Moreover, multi-year observational studies contribute to a better understanding of ecosystem dynamics by identifying relationships and interactions among different measured stressors which can then serve as a springboard for developing future hypothesis-driven experiments (Dodds et al. 2012, Palmer &

Yan 2013).

Vancouver Lake, located in the Columbia River floodplain in southwest Washington state, USA, is an ideal model system in which to study the effects of multiple stressors on freshwater communities. Vancouver Lake is a large, shallow, tidally influenced, non-stratifying lake that was historically flushed each spring by the lower Columbia River, but due to diking and reclamation is now hydrologically connected to the river only via a large engineered flushing channel and one small creek. It is a eutrophic lake subject to seasonal toxic cyanobacterial blooms resulting partly from increased phosphorus availability (Lee et al. 2015a) and has been the focus of previous studies to assess the biotic and abiotic influences on phytoplankton dynamics, bloom formation, and cyanotoxin production (Boyer et al. 2011, Rollwagen-Bollens et al. 2013, Lee et al. 2015b). Additionally, the spatial and temporal distributions of several invasive zooplankton taxa in the nearby lower Columbia River have recently been documented

(Cordell et al. 2008, Bollens et al. 2012, Smits et al. 2013, Breckenridge et al. 2015, Dexter et al.

2015, Emerson et al. 2015). But to date the interactions among multiple stressors on Vancouver

Lake as an interconnected system within the lower Columbia River have not been well studied.

Thus, the specific objectives of this 6.5-year observational study (October 2005 through

September 2011) in Vancouver Lake were to 1) characterize seasonal zooplankton community

88 succession, and 2) examine how multiple stressors — especially eutrophication and zooplankton invasions — are associated with and likely affecting zooplankton community dynamics.

Methods

Study site

Vancouver Lake is located in southwest Washington state, USA (Fig. 1). It is a shallow

(mean depth ~1 m), tidally influenced (but with no salinity intrusions), non-stratifying natural lake within the Columbia River floodplain. Vancouver Lake has two distinct hydrological connections with the Columbia River. First, the flushing channel, located on the southwest side of the lake, contains a one-way floodgate allowing water from the Columbia River to enter into

Vancouver Lake when water levels in the river are higher than in the lake. In addition, Lake

River connects the north end of Vancouver Lake to the Columbia River ~16 km downstream, and has a bi-directional flow depending on the tidal stage: as the flood tide raises the water level in the Columbia River, Lake River flows into Vancouver Lake, and as the tide ebbs, Lake River flows out of Vancouver Lake and into the Columbia River. Vancouver Lake also receives freshwater from two small creeks (Burnt Bridge Creek and Salmon Creek) that run through urban, suburban, and semi-rural areas of southwest Washington, representing a drainage area of up to 262 km2 of the surrounding watershed.

Previous studies of the aquatic biota of Vancouver Lake have been very limited, but include an in depth analysis of the phytoplankton community composition (including cyanobacteria biomass) from February 2007 through October 2010 (Lee et al. 2015a), as well as studies of microzooplankton grazing (Boyer et al. 2011) and mesozooplankton grazing

(Rollwagen-Bollens et al. 2013). The fish community has only been described for the summer of

89 one year (Caromile et al. 2000) and no studies of zooplanktivory have been performed in

Vancouver Lake.

Field collections

A study of spatial variability of plankton in Vancouver Lake was undertaken in 2007, with a total of eight littoral and limnetic sites, and a dock station, sampled 10 times each.

Kendall’s tau (τb) was then used to examine concordance among sites and demonstrated no significant spatial differences in plankton community composition (n=10 per site, p>0.05)

(Bollens & Rollwagen-Bollens, 2009). Additionally, the depth at the end of the dock sampling site has been shown to be representative of the mean depth of the lake (Sheibley et al. 2014).

Based on these results, and ease of access, all subsequent sampling was conducted from the end of the dock located just south of Burnt Bridge Creek (Fig. 1), and all previous samples collected there were deemed representative of the lake as a whole.

From October 2005 through February 2007, lake water and plankton samples were collected monthly. From March 2007 through September 2010, lake water and plankton samples were collected monthly (November through February), bi-weekly (March and October), or weekly (April through September). From May 2011 through October 2011, lake water and plankton samples were collected weekly. During each sampling effort a YSI 91 or 6920 was used to measure temperature and dissolved oxygen (DO) from the surface to the bottom, at 0.3 m intervals. Lake water depth and Secchi depth were also measured.

A clean bucket was used to collect surface water samples in triplicate for chlorophyll a

(chl a) and kept on ice until transported back to the laboratory. A 15 - 25 mL aliquot of lake

90 water from each replicate was then filtered onto GF/F filters and kept frozen at -20°C until fluorometric analysis using a Turner Model 10 AU fluorometer (Strickland & Parsons 1972).

At each sampling time, triplicate zooplankton samples were collected using a 0.5 m diameter, 73 µm mesh net towed vertically from 0.15 m above the lake bottom to the surface.

Zooplankton samples were preserved in 5 - 10% formalin. Well mixed subsamples were taken from whole samples with a Hensen-Stempel pipette, and a minimum of 300 zooplankton individuals were quantified and identified to the lowest taxon possible using a Nikon SMZ 1500 microscope at 10X – 40X magnification. Copepod nauplii were often highly abundant, and were not included in the organism counts because they can overwhelm and dampen overall community composition patterns; however, nauplii of the calanoid copepod Pseudodiaptomus forbesi were included in order to examine any significant changes associated with this invasive species. Zooplankton densities were calculated based on volumes filtered (individuals m-3).

From February 2007 through October 2010, additional surface subsamples of 50 mL for analysis of nitrite (NO2-N), nitrate (NO3-N), ammonium (NH4-N), orthophosphate (PO4-P), and silicate (SiO4-Si) were filtered through a 0.45 µm Millipore filter into a plastic bottle and kept on ice until returned to the laboratory. Samples were then frozen and sent to the Marine Chemistry

Lab at the University of Washington’s School of Oceanography for analysis.

Statistical analyses

Prior to all analyses, zooplankton abundances were log+1 transformed. All nauplii, except for those of P. forbesi, and those taxa occurring in <3% of all samples were excluded. In two cases, taxa were combined to family (e.g. Bosminidae, see explanation in results) and order

(e.g. Calanoida) to form a group that included rarer taxa that might have been excluded from the analysis or would otherwise have overestimated the influence of rare taxa.

In order to characterize significantly distinct groupings within the zooplankton community, we used cluster analysis (Clarke 1993), multiple response permutation procedure

(MRPP) (Mielke et al. 1981, McCune & Grace 2002), and indicator species analysis (Dufrene &

Legendre 1997). Clusters of distinct zooplankton communities were defined by using relative

Euclidian distance measure and Ward’s method for cluster linkage with 75% of the information retained. This allowed us to examine how similar/dissimilar communities were clustered together and how they changed over time based on the relative abundances and occurrence of zooplankton taxa. MRPP was then used to test the significance of the resulting zooplankton clusters, as well as to determine if there were significant interannual differences in the groupings, using ranked Sorenson’s distance. Indicator species analysis was then applied to each zooplankton group to determine which zooplankton taxa were most strongly associated with each cluster. Indicator species analysis specifically identifies taxa that best represent each of the resulting zooplankton communities derived from cluster analysis. Taxa that occurred five times or more in one cluster compared to other clusters were considered significantly representative of that particular cluster and deemed “faithful.”

We used non-metric multidimensional scaling (NMDS) to detect relationships between zooplankton species abundance and environmental conditions. NMDS is a robust ordination technique that can be used with non-normally distributed and discontinuous data (McCune et al.

2000). Any sample dates with incomplete data (i.e. missing physical or nutrient data) were excluded. Due to the varying sampling regimes and detection of invasive zooplankton, three

NMDS ordinations spanning different time periods were performed to determine the extent of

92 abiotic (physical and nutrient) and biotic (cyanobacteria biomass and non-native zooplankton) influences on changes in zooplankton community composition. All NMDS ordinations were performed using the Sorenson distance measure.

In the first NMDS ordination, we examined how zooplankton community composition varied between October 2005 through September 2010, and May 2011 through October 2011, in response to changes in lake depth, Secchi depth, temperature, season, and chl a. Season was calculated as sin(365/360(Julian day)), with December 1 set as 0, March 1 as 1, June 1 as 0, and

September 1 as -1. For the second ordination, we examined how zooplankton community composition varied over a shorter period of time for which we had more extensive environmental data: from February 2007 through September 2010, zooplankton composition was assessed in association with lake depth, Secchi depth, DO, nutrient availability, cyanobacteria biomass, and chl a. A detailed description of our process of enumeration of cyanobacteria and other phytoplankton can be found in Lee et al. (2015a). Briefly, cyanobacteria biomass calculations were determined as recommended by APHA (2012); biomass was estimated (Menden-Deuer &

Lessard 2000) after calculating biovolume based on geometric shape (Hillebrand et al. 1999).

Finally, in our third ordination, we examined the zooplankton communities and associated environmental data (lake depth, Secchi depth, chl a, temperature, and season) using the combined periods of pre- and post-introduction of invasive zooplankton, i.e., October 2005 -

February 2007 and March 2010 - October 2011, respectively. In this third ordination, taxa that were initially combined to a higher classification (i.e. species to family or order) were separated to their lowest taxonomic resolution because they no longer occurred in <3% of samples. All statistical analyses were performed using PC-ORD version 5.33 software (McCune & Grace

2002).

Results

Environmental variables

Over the 6.5-year study, total lake depth ranged from 0.80 m to 4.9 m and fluctuated seasonally, with increased depth during the spring freshet and decreased depth during the late summer/early autumn (Fig. 2a). Secchi depth ranged from 0.15 m to 2.1 m and followed a similar seasonal pattern as total lake depth (Fig. 2a). Temperature ranged from 2.2°C to 28°C, and showed a seasonal pattern of warmer temperatures in the summer and cooler temperatures in the winter (Fig. 2b). The warmest summer was observed in 2009 and the coolest summer in 2010

(peaking at 23°C). DO ranged from very low levels (<2.0 mg L-1) during the summer months to supersaturated conditions (>14 mg L-1) during the winter and spring months (Fig. 2b). Chl a ranged from 4.1 µg L-1 to 820 µg L-1 and exhibited intense seasonal peaks during the summer, concurrent with increased temperatures and the onset of cyanobacteria blooms, with the highest values observed in summers of 2007 and 2009 (Fig. 2c). Nutrient data are described in detail in

Lee et al. (2015a), but briefly, PO4-P exhibited relatively stronger peaks during the onset of cyanobacteria blooms during the summer months (Fig. 2d). NH4-N fluctuated throughout the year, but similar to PO4-P, concentrations were greater during the summer months (Fig. 2d).

NO3-N availability was greatest during the winter months, when chl a was lowest (Fig. 2e).

SiO4-Si showed seasonal fluctuations with concentrations greater during the summer and winter, and lower during the spring (Fig. 2e).

Dominant zooplankton groups

A total of 28 zooplankton taxa, excluding nauplii and copepodites, were identified from

171 samples in Vancouver Lake (Table 1). It should be noted that our 73 µm mesh net may have missed some smaller rotifer species. Two seasonal peaks of total zooplankton abundance were observed during most years – one during late spring (May) and one during late summer (August/

September) (Fig. 3a, in color Supplementary data, Fig. S1a). We observed only one peak in total zooplankton abundance in 2006 (spring) and 2011 (summer), which might be attributed to a somewhat reduced sampling frequency in those years (monthly throughout 2006 and monthly from May through October 2011, vs. weekly during the summer of other years). Cladocera dominated the late spring peak each year, whereas the dominant group observed during late summer peaks varied among cyclopoid copepods, rotifers, and cladocera (Fig. 3b, in color

Supplementary data, Fig. S1b).

Of the copepods, cyclopoids were the most common taxon, dominated nearly exclusively by Diacyclops thomasi. In addition, calanoid copepods were observed throughout the sampling period (except in 2008); however, they did not contribute substantially to total zooplankton abundance except in 2011. Due to the rare occurrence (<3%) of adults of the calanoids

Leptodiaptomus sp., Skistodiaptomus sp., and Pseudodiaptomus forbesi, these taxa were subsequently combined into “adult calanoid copepods” so as not to exclude them from the analysis. Finally, harpacticoid adults were distinguished, but were rarely present. Among rotifers, six taxa were identified but only five were included in the analysis due to the rarity of one taxon.

Polyarthra sp., Asplanchna sp., and Brachionus sp. were present each year; Keratella sp. was present each year from 2007 through 2011, but not in 2005 or 2006; and Kellicottia sp. was present only in 2010. With respect to cladocerans, Daphnia retrocurva was the most abundant of a total of nine cladoceran species identified (Table 1).

Seasonal succession and zooplankton community composition

Results from cluster analysis defined four significant clusters (A=0.208, p<10-8), suggesting a recurring seasonal succession, as well as significant interannual variability

(A=0.176, p<10-8) (Fig. 3a). Seasonal succession was observed in 2007, 2009 and 2010, with cluster 1 occurring during the winter, clusters 2 and 3 representing the spring, cluster 4 representing the summer, and cluster 2 representing the autumn. Zooplankton assemblages between October 2005 and December 2006 were defined only by clusters 1 and 4, when overall zooplankton abundances were lowest. In contrast, during 2008, clusters 1 and 4 were completely absent, with cluster 2 representing the winter and spring, and cluster 3 representing the summer and early autumn. Additionally, with the absence of cluster 4 in 2008, overall abundances of clusters 2 and 3 were higher than in previous years. In 2011, cluster 1 represented the entire spring and summer, with overall abundances higher than in 2007-2010 (but not 2005 and 2006), and with abundances of other clusters dramatically suppressed (Fig. 3a).

Based on indicator species analysis, cluster 1 was characterized by calanoid copepods, harpacticoid copepods, and all non-daphnid and bosminid cladocerans (Fig. 4a). With the exception of 2008 and 2011, cluster 1 occurred during winter months. From 2005 through 2006, the cladocerans Ceriodaphnia sp. and Chydorus sp. were especially abundant. The rare occurrence of cluster 1 in 2007 was dominated by harpacticoid copepods. Cluster 1 was the only group represented in 2011 and was dominated by calanoid copepods, with overall abundances higher than in the previous four years.

Cluster 2 consisted primarily of the cladoceran family Bosminidae, and the rotifers

Polyarthra sp., Asplancha sp., Keratella sp., and Kellicotia sp. (Fig. 4b). Abundances of these

96 taxa were low during the winter and variable throughout spring, summer, and autumn. A late spring and late summer bimodal distribution of cluster 2 species abundance was particularly pronounced in 2007, 2008, and 2010. Based on indicator species analysis, Kellicottia sp. was the species that best defined this cluster, however it was only detected in 2010.

Cluster 3 was represented by the cladoceran, D. retrocurva, and other daphnids (Fig. 4c).

Daphnids were largely absent during the winter (as were other zooplankton taxa) and exhibited two peaks in abundance during the late spring and late summer, with the exception of 2011, when daphnids were notably absent during the summer.

Cluster 4 consisted of the cyclopoid copepod D. thomasi, cyclopoid copepodites, the rotifer Brachionus sp., and the carnivorous cladoceran Leptodora kindtii (Fig. 4d). As with clusters 2 and 3, there were striking interannual differences in overall abundance and occurrence of cluster 4 taxa. Cluster 4 defined the summer community each year, with the exception of 2008 and 2011. In 2008, L. kindtii was absent during the summer and the zooplankton community was characterized by cluster 3. In summer 2011, although abundance and occurrence of D. thomasi and copepodites were lower compared to previous years, and Brachionus sp. exhibited uncommonly high abundances, the zooplankton community was defined by cluster 1.

Abiotic and biotic factors associated with zooplankton community composition

The first NMDS ordination of zooplankton community composition, including the longer time series (2005-2011) but more limited set of environmental variables, resulted in a three dimensional solution of which only axis two was strongly correlated with three of the five environmental variables (stress=15.28, r2>0.3): Secchi depth, season, and chl a (Fig. 5a). Cluster

1 was associated with increased water clarity (deeper Secchi depth) and season. Ordination of

97 sample points representing cluster 1 were more dispersed than other clusters and was largely driven by Chydorus sp., and calanoid adults and copepodites (r2>0.2), even though other cladoceran taxa are included in this cluster (Fig. 5b). Seasonal summer peaks of chl a showed a strong relationship with Cluster 4. Clusters 2 and 3 occupied an intermediate space of the ordination, suggesting that these clusters exist as transitional conditions across each environmental gradient.

The second ordination, using data from February 2007 through September 2010, was performed to examine whether including additional environmental variables, not available in other years (i.e.., DO, nutrients, and cyanobacteria biomass), would yield additional insight into the dynamics of the zooplankton community (Fig. 6). A three dimensional solution

(stress=14.95) was reached and included several additional environmental factors associated with zooplankton community composition (r2>0.2). Interannual differences were more apparent with this subset of data. Season was associated with clusters 2 and 3 (Fig. 6a) from 2008 and 2010

(Fig. 6b). All annual summer zooplankton communities (clusters 4 and 2) were associated with increased levels of PO4-P and chl a, warmer temperatures, and higher cyanobacteria biomass.

High cyanobacteria biomass and chl a defined the annual phytoplankton blooms in Vancouver

Lake, specifically from 2007 through 2009, which have been correlated with seasonally eutrophic conditions (as described by increased PO4-P); however, in 2010 a significant cyanobacterial bloom was not observed, even though there was an increase in PO4-P (Lee et al.

2015a).

Introduction of non-native zooplankton

Two invasions of non-native zooplankton were detected in Vancouver Lake over the course of our study. The non-native Asian calanoid copepod Pseudodiaptomus forbesi was first identified in Vancouver Lake in June 2006 as nauplii, and then in November 2006 as adults (Fig.

7a). With the exception of a single occurrence in April 2009, P. forbesi was not observed again until 2010 (Fig. 7a). In 2011 it was abundant in all life history stages (Fig. 7a).

The non-native bosminid cladoceran, Bosminia coregoni, was first observed in

Vancouver Lake in March 2007. B. coregoni first appeared in the Columbia River as a single occurrence in September 2006, before a population boom was observed in 2008 (Dexter et al.

2015). All Bosminidae species in Vancouver Lake were initially identified by us as B. coregoni, due to their dominance beginning in March 2007. However, in 2010 we determined that both B. coregoni and the native B. longirostris were present in Vancouver Lake. Due to the uncertainty in identifications between 2007 and 2009, B. coregoni and B. longirostris occurring during this period were combined into a single taxonomic group (family Bosminidae). Bosminids in 2010 and 2011 samples, however, were reliably identified as either of two separate taxa: B. coregoni or B. longirostris.

The third and final NMDS ordination was employed for the purposes of examining possible environmental associations with zooplankton communities both pre- and post- introduction of non-native zooplankton. The pre-introduction period (October 2005 – February

2006) was defined by the absence of B. coregoni and the lack of an established P. forbesi population. The post-introduction period (March 2010 - October 2011) was defined by the increased occurrence and abundance of both invasive zooplankters, P. forbesi and B. coregoni

(Fig. 7b).

The NMDS ordination of pre- and post-introduction showed distinctly different groups

(stress=14.29) (Fig. 8). With the exception of two samples, the pre-introduction group consisted of cluster 1 taxa and was strongly associated with season. The post-introduction group predominantly consisted of clusters 1, 2 and 4. After the introduction of both invasive zooplankters, cluster 1 was principally associated with P. forbesi and other calanoid copepods.

Cluster 2 was largely comprised of rotifers and B. longirostris. In contrast, cluster 4 was predominantly represented by B. coregoni. Of the post-introduction samples, summer samples of cluster 4 occurred during 2010, and summer samples of cluster 1 occurred during 2011, and both were most strongly associated with temperature. The native B. longirostris (cluster 2) was inversely associated with B. coregoni (cluster 4).

Discussion

The results of our study suggest that multiple stressors – including cyanobacterial blooms

(due to eutrophic conditions) and invasive zooplankton species – influenced seasonal and interannual variability in zooplankton community dynamics in Vancouver Lake.

Effects of eutrophication on zooplankton community dynamics

Environmental variables that strongly influenced zooplankton community dynamics in

Vancouver Lake included common characteristics of eutrophic conditions such as increased turbidity, increased PO4-P, and cyanobacterial blooms. In particular, PO4-P availability in

Vancouver Lake may have had an indirect influence on summertime zooplankton community composition through the stimulation of cyanobacteria blooms. Increased orthophosphate concentrations during summer months in Vancouver Lake were most likely due to a combination

100 of several processes, including wind driven sediment resuspension (Sheibley et al. 2014), anoxia at the sediment-water interface (Lee et al. 2015a), and possibly the release of PO4-P due to grazing (Frost et al. 2004). Recently, Lee et al. (2015a) demonstrated that peaks in PO4-P concentration were significantly associated with the occurrence and duration of large cyanobacteria blooms that occurred in Vancouver Lake during each summer from 2007 to 2009.

These authors further found that a very muted bloom dominated more by diatoms than cyanobacteria in 2010 was associated with a lower peak in PO4-P concentration relative to previous summers.

Our results from analysis of zooplankton abundance and taxonomic composition over the same time period (2007-2010) show that the zooplankton communities in 2007 through 2009 were generally similar to each other (defined by rotifer-dominated cluster 2 and D. thomasi- dominated cluster 4) and strongly associated with eutrophic conditions of increased cyanobacteria biomass, chl a, temperature, and PO4-P. However, the community in 2010 was significantly different (represented by cluster 1 dominated by calanoid copepods) and not strongly associated with either seasonality or eutrophication. Although an annual increase in chl a was observed during the 2010 and 2011 summers, these were noticeably lower than the blooms in 2007-2009. More specifically, we observed a zooplankton community shift from cluster 4 to cluster 1 at the end of the 2010 summer, and the entire 2011 summer community was also defined by cluster 1. This suggests a strong link between zooplankton community structure and seasonally eutrophic conditions, in particular the occurrence (or not) of cyanobacterial blooms.

Effects of cyanobacterial blooms on zooplankton community dynamics have been well studied elsewhere (see review by Ger et al. 2014). Meta-analyses of laboratory studies examining effects of toxic cyanobacteria on zooplankton fitness suggested that although cyanobacteria are

101 not ideal food items, zooplankton responses to cyanobacterial blooms are species-specific and may not be as severely detrimental as previously thought (Wilson et al. 2006, Tillmanns et al.

2008). For example, observational studies of Lake Tai showed that increases in cladoceran populations correlated with an increase in cyanobacteria biomass, but that copepods varied independently from cyanobacteria biomass (Sun et al. 2012). Other studies have shown that smaller cladocerans and copepods are better competitors than larger cladocerans during cyanobacterial blooms (Deng et al. 2008, Wang et al. 2010, Sun et al. 2012), and that larger cladocerans may respond by a decrease in body size while still maintaining a positive growth rate (Sarnelle et al. 2010). Daphnids have been reported to become tolerant to cyanobacterial toxins when repeatedly exposed to cyanobacteria (Hairston et al. 2001, Sarnelle & Wilson 2005), suggesting that cyanobacteria may act as a selection force on some zooplankton taxa (Hairston et al. 2001). Thus, the composition of the summer zooplankton community in Vancouver Lake may be changing due to an interaction of repeated annual exposure to summer cyanobacterial blooms and the greater prominence of cyanobacteria-tolerant zooplankton taxa.

Predation pressure by carnivorous zooplankton may also affect summer zooplankton community composition (Wojtal et al. 1999, Pichlova & Brandl 2003, Lesutiene et al. 2012).

Each year of our study in Vancouver Lake, cladocerans and rotifers tended to dominate the zooplankton community during the spring. Subsequently, in every summer (except 2008) the carnivorous cladoceran, L. kindtii, was observed, concomitant with decreased daphnid and rotifer abundances. L. kindtii are voracious predators that can cause severe declines in prey populations generally (Wojtal et al. 1999, Chang & Hanazato 2003, Lesutiene et al. 2012), and have been shown to significantly alter zooplankton community structure by preying upon cladocerans and rotifers specifically (Wagner & Benndorf 2007, Wagner et al. 2013). The absence of L. kindtii in

102

Vancouver Lake in 2008 may have contributed to the observed increase in abundance of rotifers and daphnids that summer (although our collection methods [e.g. sample volume and net diameter], combined with the evasive capabilities of L. kindtii, almost certainly caused us to underestimate the abundance of this predator [e.g., Wojtal et al. 1999]). Direct effects of cyanobacteria blooms on L. kindtii are not well known, although a few studies have documented their co-occurrence and the continued predation by L. kindtii on available prey (Patoine et al.

2006, Perga et al. 2013). While we observed seasonal and interannual changes in zooplankton community composition in relation to eutrophic conditions in Vancouver Lake, the relative abundance of predatory zooplankton may also have had an effect on zooplankton community composition.

Finally, we observed zooplankton abundances to be positively associated with increased

Secchi depth (i.e. water clarity), except during winter months when temperatures were not conducive for zooplankton growth. This relationship could have been due to food availability and/or predation pressure (although the latter was not measured by us). For instance, concomitant with increased Secchi depth, chl a levels were very low, suggesting decreased food availability and/or increased grazing. Conversely, when Secchi depth was low, chl a levels were very high, implying increased food availability. Another possible interpretation, however, is that low Secchi depths represent periods of high turbidity that may provide zooplankton with refuge from visual predation (Engström-Öst & Mattila, 2008, Schulze 2011). Thus, either increased food availability or decreased visual predation may have contributed to the higher zooplankton abundances we observed during summer months. Finally, we acknowledge the potential of fish planktivory to structure zooplankton communities (Brooks & Dodson 1965, Gliwicz et al. 2010,

103

Schulze 2011); however, it was impossible for us to evaluate here, because studies on seasonal fish abundance and zooplanktivory have not been conducted in Vancouver Lake.

Effects of invasive zooplankton

We observed two non-native zooplankton species in Vancouver Lake in the later years of our study: the bosminid cladoceran, B. coregoni, and the calanoid copepod, P. forbesi. In comparing the pre- (2005-2006) and post-invasion (2010-2011) time periods, temperature was the main environmental variable that was associated with both B. coregoni and P. forbesi nauplii, suggesting that warmer temperatures was conducive for these two species. However, temperature alone does not explain the population abundance patterns of B. coregoni observed in 2010 and

2011. Specifically, bosminids were also negatively associated with chl a (Fig. 5b).

Based on native habitat conditions and the spatial and temporal occurrence of P. forbesi and B. coregoni (in east Asia and Eurasia, respectively), the mechanisms of introduction to

Vancouver Lake were most likely different for these two species. P. forbesi has been well established throughout the lower Columbia River and upstream in several impoundments since

2005, and has been shown to dominate the zooplankton community of these waters in the late summer/early autumn (Cordell et al. 2008, Bollens et al. 2012, Breckenridge et al. 2015, Dexter et al. 2015, Emerson et al. 2015). Because of its presence throughout the lower Columbia River, introduction of P. forbesi to Vancouver Lake was most likely through natural flows from nearby

“source” waters, e.g. through the flushing channel or Lake River. Yet, while the introduction of

P. forbesi in 2010 significantly altered the summer zooplankton community in Vancouver Lake when it first arrived, this Asian copepod did not dominate the zooplankton community of the

Lake, as it has in various parts of the Columbia River. It remains to be seen if further re-

104 introductions of P. forbesi in Vancouver Lake will continue to affect zooplankton community composition, including the possibility that this invasive copepod might eventually dominate the summer zooplankton community in the lake.

In contrast, B. coregoni has been an established freshwater invader in North America since the 1960s (Deevey & Deevey 1971) via multiple invasions (DeMelo & Hebert, 1994), but its introduction to the U.S. Pacific Northwest has been more recent and the effects of its invasion on zooplankton community dynamics and trophic interactions are not well known (Smits et al.

2013, Dexter et al. 2015). B. coregoni first appeared in the Columbia River as a single occurrence in September 2006, prior to a population boom in 2008 (Dexter et al. 2015).

However, a recurring population of B. coregoni in Vancouver Lake was first identified in March

2007, before its detection in the Columbia River estuary, Lake Washington and several other lakes in eastern Washington state (Smits et al. 2013). The patchy occurrence and distribution of

B. coregoni throughout the state of Washington suggests that its introduction and dispersal may be largely attributed to recreational boating and migratory fowl (Smits et al. 2013).

Overall bosminid cladoceran abundances in Vancouver Lake were high during the spring, and then drastically reduced during the summer. The noticeable increase in bosminid abundances observed in 2010 coincided with a truncated and mild cyanobacterial bloom, with the native B. longirostris dominating during the early summer and the invasive B. coregoni dominating during the late summer. At present, no studies have examined competitive interactions between B. longirostris and B. coregoni. Under laboratory conditions, cyanobacteria blooms have been demonstrated to suppress populations of larger cladocerans, while allowing the opportunity for smaller cladocerans, such as B. longirostris, and copepods to persist (Lampert 1987, Fulton &

Paerl 1988). Yet, B. coregoni is larger than B. longirostris, and field studies have demonstrated

105 that B. coregoni in its native range was positively correlated with cyanobacteria (Deng et al.

2008, Sun et al. 2012), suggesting that B. coregoni may be tolerant of cyanobacteria blooms.

Similar to daphnids, bosminids have also been shown to adapt to repeated exposure to toxic cyanobacteria (Jiang et al. 2013). Our results indicate that Bosminids (which were dominated by

B. coregoni) may be adversely affected by cyanobacteria blooms in Vancouver Lake as they are negatively associated with increased levels of chl a (Fig. 5); however, this does not preclude the possibility that B. coregoni may eventually become tolerant to cyanobacterial blooms in

Vancouver Lake.

Calanoid copepods have been shown to be more sensitive to toxic cyanobacteria compared to other types of crustacean zooplankton (DeMott et al. 1991). In particular, P. forbesi in the San Francisco estuary has been shown to be more susceptible to microcystin than other copepods and cladocerans (Ger et al. 2009), yet can nevertheless co-exist with toxic blooms of

Microcystis sp. because of its selective feeding behavior (Ger et al. 2010). However, no studies have examined the effects of cyanobacteria blooms on different life history stages of P. forbesi.

In Vancouver Lake, P. forbesi consisted mainly of nauplii, with copepodites and adults usually very rare or absent (indeed, adult female P. forbesi were never observed by us), although the limited occurrence of P. forbesi copepodites and adults in our samples may be due to the truncated sampling regimes in 2010 and 2011, which ceased at the end of September and

October, respectively. In the Columbia River, adult P. forbesi peak in abundance in late summer and early autumn (Bollens et al. 2012, Dexter et al. 2015, Emerson et al. 2015) and thus we may have missed peak abundances of adult P. forbesi in Vancouver Lake in 2010 and 2011.

Alternatively, the generally low abundances of P. forbesi in Vancouver Lake could have been the result of deleterious effects of cyanobacteria, as has been shown for this and other calanoid

106 copepods (DeMott et al. 1991, Ger et al. 2009). Our results suggest that further investigation of the effects of cyanobacterial blooms on different life stages of P. forbesi is warranted.

An established P. forbesi population may have detrimental effects on native zooplankton communities. Studies within the Columbia River watershed have shown P. forbesi to consume ciliates and diatoms, and may compete for similar prey items as Diacyclops thomasi, a seasonally dominant cyclopoid copepod in Vancouver Lake (Rollwagen-Bollens et al. 2013, Bowen et al. submitted). Although we rarely observed P. forbesi adults in Vancouver Lake, the increased abundance of calanoid copepods in general in the late spring of 2011 coincided with an overall decrease of other copepods, cladocerans, and rotifers. A similar effect was observed during the late summer of 2011, but a substantial increase of Brachionus sp. abundance and the presumed resulting competition for resources might also have contributed to the drastic declines of non- calanoid copepods, cladocerans, and other rotifers compared to previous summers.

Havel et al. (2005) predicted that landscapes consisting of lakes and reservoirs, such as the Columbia River basin, are more susceptible to invasions by non-native species. The

Columbia River is a major commercial waterway and can be considered a “hotline” in that it acts as a biodiversity corridor through several different ecosystems throughout the northwest United

States and Canada (Décamps 2011). Vancouver Lake is also a recreational destination for sailing, rowing, and fishing, and is directly connected with the Columbia River. Recreational boating has been demonstrated to be an effective vector for spreading invasive plankton (Kelly et al. 2013a). Moreover, the lower Columbia River hosts several other invasive zooplankton species that have not yet been detected in Vancouver Lake (Cordell et al. 2008, Bollens et al. 2012,

Breckenridge et al. 2015, Dexter et al. 2015). This suggests that future invasions of Vancouver

107

Lake are likely, lending greater urgency to the need to examine interactions between invasive and native zooplankton, and their implications for higher trophic levels and food web dynamics.

Management implications for Vancouver Lake: a model shallow lake system

Intensive observational studies of shallow, tidally influenced freshwater ecosystems, such as Vancouver Lake, are uncommon. Thus, Vancouver Lake serves as a model system for examining multi-stressor effects in shallow lakes, and the ecosystem consequences and management implications of these stressors.

Eutrophication, as measured by increased turbidity, nutrient and chl a concentrations, and cyanobacteria biomass – common problems facing freshwater systems worldwide – significantly influenced the composition and abundance of the zooplankton community in Vancouver Lake, indicating that management to reduce nutrient addition may also impact zooplankton assemblages with potential effects on planktonic food web dynamics. The presence of invasive zooplankton taxa also resulted in changes to the overall zooplankton community in Vancouver

Lake, with potential management implications. Furthermore, there may be a negative interaction in Vancouver Lake between eutrophication (e.g. cyanobacteria blooms) and invasive zooplankton. For instance, in 2010 and 2011 there was a numerical increase in invasive zooplankton species at a time of year (late summer) when cyanobacteria had bloomed in previous years, but did not occur in these two years. While speculative, this suggests that the absence of cyanobacteria blooms in 2010 and 2011 may have facilitated the numerical increase of invasive zooplankton. If cyanobacterial blooms negatively affect invasive zooplankton, management actions aimed at reducing cyanobacterial blooms may contribute to the success and establishment of introduced species such as B. coregoni and P. forbesi. Thus, further research is

108 needed to examine interactions between cyanobacterial blooms and invasive zooplankton species.

Finally, long term investigations of understudied systems such as Vancouver Lake are needed to better understand the effects of multiple environmental stressors to help guide future management and restoration efforts, particularly under conditions of climate warming. For example, climate change has been indicated as the cause of changes in zooplankton community dynamics (Gyllström et al. 2005) and increases in temperature have been demonstrated to alter zooplankton community composition by selecting for life history characteristics more resilient to warmer temperatures (Dijkstra et al. 2011). Additionally, asynchronous systems, where the timing of the presence of predator and prey are mismatched, can occur as a result of changes in seasonal meteorological conditions (Straile 2000, Anneville et al. 2010, Wagner et al. 2013).

This may be exacerbated if either predator or prey species become replaced by non-native species. Understanding possible interactions among multiple stressors (e.g., eutrophication, invasive species, and climate change) will be important to better managing impaired, interconnected freshwater ecosystems.

Acknowledgments

We thank J. Duerr Boyer, M. McDonald, and J. Zimmerman for help with sampling and data collection; A. Gonzalez for zooplankton identification and sample processing; and the

Vancouver Lake Sailing Club for lake access. This research was partially supported by Grant No.

06HQGR0126 from the United States Geological Survey (USGS) to G.R.B. and S.M.B, through the State of Washington Water Research Center. Its contents are solely the responsibility of the

109 authors and do not necessarily represent the official views of USGS. Additional support included a grant from the Vancouver Lake Watershed Partnership and awards from the National Science

Foundation’s ULTRA-EX program (09-48983) and Graduate STEM Fellows in K-12 Education program (07-42561) to S.M.B. and G.R.B.

110

References

American Public Health Association, American Water Works Association, & Water

Environment Federation (APHA), 2012: Standard Methods for the Examination of Water

and Wastewater 22 edition. Rice, E. W., Baird, R. B., Eaton, A. D. & Clesceri, L. S.

(eds.) American Water Works Association. – Washington, D.C.

Anneville, O., Molinero, J. C., Souissi, S. & Gerdeaux, D., 2010: Seasonal and interannual

variability of cladoceran communities in two peri-alpine lakes: Uncoupled response to

the 2003 heat wave. – J. Plankton Res. 32: 913–925.

Boersma, M., Aberle, N., Hantzsche, F. M., Schoo, K. L., Wiltshire, K. H. & Malzahn, A. M.,

2008: Nutritional limitation travels up the food chain. – Int. Rev. Hydrobiol. 93: 479–

488.

Bollens, S., Breckenridge, J., Cordell, J., Simenstad, C. & Kalata, O., 2014: Zooplankton of tidal

marsh channels in relation to environmental variables in the upper San Francisco Estuary.

– Aquat. Biol. 21: 205–219.

Bollens, S. M., Breckenridge, J. K., Cordell, J. R., Rollwagen-Bollens, G. & Kalata, O., 2012:

Invasive copepods in the Lower Columbia River Estuary: Seasonal abundance, co-

occurrence and potential competition with native copepods. – Aquat. Invasions 7: 101–

109.

Bollens, S. M., Breckenridge, J. K., Hooff, R. C. & Cordell, J. R., 2011: Mesozooplankton of the

lower San Francisco Estuary: Spatio-temporal patterns, ENSO effects and the prevalence

of non-indigenous species. – J. Plankton Res. 33: 1358–1377.

111

Bollens, S. M., Cordell, J. R., Avent, S. & Hooff, R. C., 2002: Zooplankton invasions: A brief

review, plus two case studies from the northeast Pacific Ocean. – Hydrobiologia 480: 87–

110.

Bollens, S. M. & Rollwagen-Bollens, G., 2009: Biological Assessment of the Plankton in

Vancouver Lake, WA. – Washington State University Vancouver.

Bowen, A., Rollwagen-Bollens, G., Bollens, S. M. & Zimmerman, J., Submitted: Feeding of the

invasive copepod Pseudodiaptomus forbesi on natural microplankton assemblages within

the lower Columbia River. – J. Plankton Res.

Boyer, J., Rollwagen-Bollens, G. & Bollens, S. M., 2011: Microzooplankton grazing before,

during and after a cyanobacterial bloom in Vancouver Lake, Washington, USA. – Aquat.

Microb. Ecol. 64: 163–174.

Breckenridge, J. K., Bollens, S. M., Rollwagen-Bollens, G. & Roegner, G. C., 2015: Plankton

assemblage variability in a river-dominated temperate estuary during late spring (high-

flow) and late summer (low-flow) periods. – Estuar. Coasts. 38: 93-103.

Brooks, J. L. & Dodson, S. I., 1965: Predation, body size, and composition of plankton. –

Science 150: 28–35.

Caromile, S. J., Meyer, W. R. & Jackson, C. S., 2000: 1998 warm water fish survey of

Vancouver Lake – Clark County Washington Department of Fish and Wildlife, Olympia,

Washington.

Chang, K. H. & Hanazato, T., 2003: Seasonal and spatial distribution of two Bosmina species (B.

longirostris and B. fatalis) in Lake Suwa, Japan: Its relation to the predator Leptodora. –

Limnology 4: 47–52.

112

Christoffersen, K., Reimann, B., Hansen, L. R., Klysner, A. & Sørensen, H. B., 1990: Qualitative

importance of the microbial loop and plankton community structure in a eutrophic lake

during a bloom of cyanobacteria. – Microb. Ecol. 20: 253–272.

Clarke, K., 1993: Nonparametric multivariate analyses of changes in community structure. –

Aust. J. Ecol. 18: 117–143.

Cordell, J. R., Bollens, S. M., Draheim, R. & Sytsma, M., 2008: Asian copepods on the move:

Recent invasions in the Columbia–Snake River system, USA. – ICES J. Mar. Sci. J.

Cons. 65: 753–758.

Daskalov, G. M. & Mamedov, E. V., 2007: Integrated fisheries assessment and possible causes

for the collapse of anchovy kilka in the Caspian Sea. – ICES J. Mar. Sci. J. Cons. 64:

503–511.

Davis, T. W., Koch, F., Marcoval, M. A., Wilhelm, S. W. & Gobler, C. J., 2012:

Mesozooplankton and microzooplankton grazing during cyanobacterial blooms in the

western basin of Lake Erie. – Harmful Algae 15: 26–35.

Décamps, H., 2011: River networks as biodiversity hotlines. – C. R. Biol. 334: 420–434.

Deevey, E. S. & Deevey, G. B., 1971: The American species of Eubosmina seligo (Crustacea,

Cladocera). – Limnol. Oceanogr. 16: 201–218.

DeMelo, R. & Hebert, P. D., 1994: Founder effects and geographical variation in the invading

cladoceran Bosmina (Eubosmina) coregoni Baird 1857 in North America. – Heredity 73:

490–499.

DeMott, W. R., Zhang, Q. X. & Carmichael, W. W., 1991: Effects of toxic cyanobacteria and

purified toxins on the survival and feeding of a copepod and three species of Daphnia. –

Limnol. Oceanogr. 36: 1346–1357.

113

Deng, D., Xie, P., Zhou, Q., Yang, H., Guo, L. & Geng, H., 2008: Field and experimental studies

on the combined impacts of cyanobacterial blooms and small algae on crustacean

zooplankton in a large, eutrophic, subtropical, Chinese lake. – Limnology 9: 1–11.

Dexter, E., Bollens, S. M., Rollwagen-Bollens, G., Emerson, J. & Zimmerman, J., 2015:

Persistent vs. ephemeral invasions: 8.5 years of zooplankton community dynamics in the

Columbia River. – Limnol. Oceanogr. 60: 527-537.

Dijkstra, J. A., Westerman, E. L. & Harris, L. G., 2011: The effects of climate change on species

composition, succession and phenology: A case study. – Glob. Change Biol. 17: 2360–

2369.

Dodds, W. K., Robinson, C. T., Gaiser, E. E., Hansen, G. J., Powell, H., Smith, J. M., Morse, N.

B., Johnson, S. L., Gregory, S. V., Bell, T., Kratz, T. K. & McDowell, W. H., 2012:

Surprises and insights from long-term aquatic data sets and experiments. – BioScience

62: 709–721.

Dufrene, M. & Legendre, P. 1997: Species assemblages and indicator species: The need for a

flexible asymmetrical approach. – Ecol. Monogr. 67: 345–366.

Emerson, J. E., Bollens, S. M. & Counihan, T. D., 2015: Seasonal dynamics of zooplankton in

Columbia–Snake River reservoirs, with special emphasis on the invasive copepod

Pseudodiaptomus forbesi. – Aquat. Invasions 10: 25–40.

Engström-Öst, J., Karjalainen, M. & Viitasalo, M., 2006: Feeding and refuge use by small fish in

the presence of cyanobacteria blooms. – Environ. Biol. Fishes 76: 109–117.

Engström-Öst, J. & Mattila, J., 2008: Foraging, growth and habitat choice in turbid water: An

experimental study with fish larvae in the Baltic Sea. – Mar. Ecol. Prog. Ser. 359: 275–

281.

114

Frost, P. C., Xenopoulos, M. A. & Larson, J. H., 2004: The stoichiometry of dissolved organic

carbon, nitrogen, and phosphorus release by a planktonic grazer, Daphnia. – Limnol.

Oceanogr. 49: 1802–1808.

Fulton, R. S. & Paerl, H. W., 1988: Effects of the blue-green alga Microcystis aeruginosa on

zooplankton competitive relations. – Oecologia 76: 383–389.

Ger, K. A., Hansson, L. A. & Lürling, M., 2014: Understanding cyanobacteria-zooplankton

interactions in a more eutrophic world. – Freshw. Biol. 59: 1783–1798.

Ger, K. A., Teh, S. J., Baxa, D. V., Lesmeister, S. & Goldman, C. R., 2010: The effects of

dietary Microcystis aeruginosa and microcystin on the copepods of the upper San

Francisco Estuary. – Freshw. Biol. 55: 1548–1559.

Ger, K. A., Teh, S. J. & Goldman, C. R., 2009: Microcystin-LR toxicity on dominant copepods

Eurytemora affinis and Pseudodiaptomus forbesi of the upper San Francisco Estuary. –

Sci. Total Environ. 407: 4852–4857.

Glibert, P. M., Fullerton, D., Burkholder, J. M., Cornwell, J. C. & Kana, T. M., 2011: Ecological

stoichiometry, biogeochemical cycling, invasive species, and aquatic food webs: San

Francisco estuary and comparative systems. – Rev. Fish. Sci. 19: 358–417.

Gliwicz, Z. M., Wursbaugh, W. A. & Szymanska, E., 2010: Absence of predation eliminates

coexistence: experience from the fish–zooplankton interface. – Hydrobiologia 653: 103–

117.

Gulati, R. & DeMott, W., 1997: The role of food quality for zooplankton: Remarks on the state-

of-the-art, perspectives and priorities. – Freshw. Biol. 38: 753–768.

Gyllström, M., Hansson, L. A., Jeppesen, E., Garcia-Criado, F., Gross, E., Irvine, K., Kairesalo,

T., Kornijow, R., Miracle, M. R., Nykänen, M., Nõges, T., Romo, S., Stephen, D., Van

115

Donk, E. & Moss, B., 2005: The role of climate in shaping zooplankton communities of

shallow lakes. – Limnol. Oceanogr. 50: 2008–2021.

Hairston, N. G., Holtmeier, C. L., Lampert, W., Weider, L. J., Post, D. M., Fischer, J. M.,

Cáceres, C. E., Fox, J. A. & Gaedke, U., 2001: Natural selection for grazer resistance to

toxic cyanobacteria: Evolution of phenotypic plasticity? – Evolution 55: 2203–2214.

Havel, J. E., Lee, C. E. & Vander Zanden, J. M., 2005: Do reservoirs facilitate invasions into

landscapes? – BioScience 55: 518–525.

Hillebrand, H., Dürselen, C. D., Kirschtel, D., Pollingher, U. & Zohary, T., 1999: Biovolume

calculation for pelagic and benthic microalgae. – J. Phycol. 35: 403–424.

Hooff, R. C. & Bollens, S. M., 2004: Functional response and potential predatory impact of

Tortanus dextrilobatus, a carnivorous copepod recently introduced to the San Francisco

Estuary. – Mar. Ecol. Prog. Ser. 277: 167–179.

Jeppesen, E., Jensen, J. P., Søndergaard, M., Lauridsen, T., Pedersen, L. J. & Jensen, L., 1997:

Top-down control in freshwater lakes: The role of nutrient state, submerged macrophytes

and water depth. – Hydrobiologia 342-343: 151–164.

Jeppesen, E., Nõges, P., Davidson, T. A., Haberman, J., Nõges, T., Blank, K., Lauridsen, T. L.,

Søndergaard, M., Sayer, C., Laugaste, R., Johansson, L. S., Bjerring, R. & Amsinck, S.

L., 2011: Zooplankton as indicators in lakes: A scientific-based plea for including

zooplankton in the ecological quality assessment of lakes according to the European

Water Framework Directive (WFD). – Hydrobiologia 676: 279–297.

Jiang, X., Li, Q., Liang, H., Zhao, S., Zhang, L., Zhao, Y., Chen, L., Yang, W. & Xiang, X.,

2013: Clonal variation in growth plasticity within a Bosmina longirostris population: The

potential for resistance to toxic cyanobacteria. – PLoS ONE 8: e73540.

116

Kelly, N. E., Wantola, K., Weisz, E. & Yan, N. D. 2013a: Recreational boats as a vector of

secondary spread for aquatic invasive species and native crustacean zooplankton. – Biol.

Invasions 15: 509–519.

Kelly, N. E., Yan, N. D., Walseng, B. & Hessen, D. O., 2013b: Differential short- and long-term

effects of an invertebrate predator on zooplankton communities in invaded and native

lakes. – Divers. Distrib. 19: 396–410.

Lampert, W., 1987: Laboratory studies on zooplankton‐cyanobacteria interactions. – N. Z. J.

Mar. Freshw. Res. 21: 483–490.

Lee, T. A., Rollwagen-Bollens, G. & Bollens, S. M., 2015a: The influence of water quality

variables on cyanobacterial blooms and phytoplankton community composition in a

shallow temperate lake. – Environ. Monit. Assess. Advanced online publication.

doi:10.1007/s10661-015-4550-2.

Lee, T. A., Rollwagen-Bollens, G., Bollens, S. M. & Faber-Hammond, J. J., 2015b:

Environmental influence on cyanobacteria abundance and microcystin toxin production

in a shallow temperate lake. – Ecotoxicol. Environ. Saf. 114: 318–325.

Lesutiene, J., Semenova, A., Griniene, E., Gasiunaite, Z. R., Savickyte, V. & Dmitrieva, O.,

2012: Abundance dynamics and functional role of predaceous Leptodora kindtii in the

Curonian Lagoon. – Cent. Eur. J. Biol. 7: 91–100.

McCune, B. & Grace, J. B., 2002: Analysis of Ecological Communities. – MjM Software

Design.

McCune, B., Rosentreter, R., Ponzetti, J. M. & Shaw, D. C., 2000: Epiphyte habitats in an old

conifer forest in western Washington, USA. – The Bryologist 103: 417–427.

117

Menden-Deuer, S. & Lessard, E. J., 2000: Carbon to volume relationships for dinoflagellates,

diatoms, and other protist plankton. – Limnol. Oceanogr. 45: 569–579.

Mielke, P., Berry, K. & Brier, G., 1981: Application of multi-response permutation procedures

for examining seasonal-changes in monthly mean sea-level pressure patterns. – Mon.

Weather Rev. 109: 120–126.

Paerl, H., 2008: Nutrient and other environmental controls of harmful cyanobacterial blooms

along the freshwater–marine continuum. – In: Hudnell,H.K. (ed): Cyanobacterial

Harmful Algal Blooms: State of the Science and Research Needs, Advances in

Experimental Medicine and Biology. – Springer New York, pp. 217–237.

Palmer, M. E. & Yan, N. D., 2013: Decadal-scale regional changes in Canadian freshwater

zooplankton: the likely consequence of complex interactions among multiple

anthropogenic stressors. – Freshw. Biol. 58: 1366–1378.

Patoine, A., Graham, M. D. & Leavitt, P. R., 2006: Spatial variation of nitrogen fixation in lakes

of the northern Great Plains. – Limnol. Oceanogr. 51: 1665–1677.

Perga, M. E., Domaizon, I., Guillard, J., Hamelet, V. & Anneville, O., 2013: Are cyanobacterial

blooms trophic dead ends? – Oecologia 172: 551–562.

Pichlova, R. & Brandl, Z., (2003) Predatory impact of Leptodora kindtii on zooplankton

community in the Slapy Reservoir. – Hydrobiologia 504: 177–184.

Rollwagen-Bollens, G., Bollens, S.M., Gonzalez, A., Zimmerman, J., Lee, T. & Emerson, J.,

2013: Feeding dynamics of the copepod Diacyclops thomasi before, during and following

filamentous cyanobacteria blooms in a large, shallow temperate lake. – Hydrobiologia

705: 101–118.

118

Roohi, A., Kideys, A. E., Sajjadi, A., Hashemian, A., Pourgholam, R., Fazli, H., Khanari, A. G.

& Eker-Develi, E., 2010: Changes in biodiversity of phytoplankton, zooplankton, fishes

and macrobenthos in the Southern Caspian Sea after the invasion of the ctenophore

Mnemiopsis leidyi. – Biol. Invasions 12: 2343–2361.

Roohi, A., Yasin, Z., Kideys, A.E., Hwai, A. T. S., Khanari, A. G. & Eker-Develi, E., 2008:

Impact of a new invasive ctenophore (Mnemiopsis leidyi) on the zooplankton community

of the Southern Caspian sea. – Mar. Ecol. 29: 421–434.

Sarnelle, O., Gustafsson, S. & Hansson, L. A., 2010: Effects of cyanobacteria on fitness

components of the herbivore Daphnia. – J. Plankton Res. 32: 471–477.

Sarnelle, O. & Wilson, A. E., 2005: Local adaptation of Daphnia pulicaria to toxic

cyanobacteria. – Limnol. Oceanogr. 50: 1565–1570.

Schulze, P. C., 2011: Evidence that fish structure the zooplankton communities of turbid lakes

and reservoirs. – Freshw. Biol. 56: 352–365.

Sheibley, R. W., Foreman, J. R., Marshall, C. A. & Welch, W. B., 2014: Water and nutrient

budgets for Vancouver Lake, Vancouver, Washington, October 2010-October 2012. –

U.S. Geological Survey Scientific Investigations Report 2014-5201.

http://dx.doi.org/10.3133/sir20145201.

Shiganova, T. A., Dumont, H. J., Sokolsky, A. F., Kamakin, A. M., Tinenkova, D. & Kurasheva,

E. K., 2004: Population dynamics of Mnemiopsis leidyi in the Caspian Sea, and effects on

the Caspian ecosystem. – In Dumont, H. J., Shiganova, T. A., and Niermann,U. (eds):

Aquatic Invasions in the Black, Caspian, and Mediterranean Seas: The Ctenophores

Mnemiopsis leidyi and Beroe in the Ponto-Caspian and Other Aquatic Invasions. –

Kluwer Academic Publishers, Dordrecht.

119

Smits, A., Litt, A., Cordell, J., Kalata, O. & Bollens, S., 2013: Non-native freshwater cladoceran

Bosmina coregoni (Baird, 1857) established on the Pacific coast of North America. –

BioInvasions Rec. 2: 281–286.

Søndergaard, M., Jeppesen, E. and Jensen, J. P., 2005: Pond or lake: Does it make any

difference? – Arch. Für Hydrobiol. 162: 143–165.

Straile, D., 2000: Meteorological forcing of plankton dynamics in a large and deep continental

European lake. – Oecologia 122: 44–50.

Strecker, A. L. & Arnott, S. E., 2008: Invasive predator, Bythotrephes, has varied effects on

ecosystem function in freshwater lakes. – Ecosytems 11: 490–503.

Strecker, A. L., Beisner, B. E., Arnott, S. E., Paterson, A. M., Winter, J. G., Johannsson, O. E. &

Yan, N. D., 2011: Direct and indirect effects of an invasive planktonic predator on

pelagic food webs. – Limnol. Oceanogr. 56: 179–192.

Strickland, J. D. H. & Parsons, T. R., 1972: A Practical Manual for Seawater Analysis 2nd ed. –

Fisheries Research Board of Canada. – Ottawa, Canada.

Sun, X., Tao, M., Qin, B., Qi, M., Niu, Y., Zhang, J., Ma, Z. & Xie, P., 2012: Large-scale field

evidence on the enhancement of small-sized cladocerans by Microcystis blooms in Lake

Taihu, China. – J. Plankton Res. 34: 853–863.

Tillmanns, A. R., Wilson, A. E., Pick, F. R. & Sarnelle, O., 2008: Meta-analysis of

cyanobacterial effects on zooplankton population growth rate: Species-specific responses.

– Arch. Für Hydrobiol. 171: 285–295.

Wagner, A. & Benndorf, J., 2007: Climate-driven warming during spring destabilises a Daphnia

population: A mechanistic food web approach. – Oecologia 151: 351–364.

120

Wagner, A., Huelsmann, S., Horn, W., Schiller, T., Schulze, T., Volkmann, S. & Benndorf, J.,

2013: Food-web-mediated effects of climate warming: Consequences for the seasonal

Daphnia dynamics. – Freshw. Biol. 58: 573–587.

Wang, X., Qin, B., Gao, G. & Paerl, H. W., 2010: Nutrient enrichment and selective predation by

zooplankton promote Microcystis (cyanobacteria) bloom formation. – J. Plankton Res.

32: 457-470.

Wilson, A. E., Sarnelle, O. & Tillmanns, A. R., 2006: Effects of cyanobacterial toxicity and

morphology on the population growth of freshwater zooplankton: Meta-analyses of

laboratory experiments. – Limnol. Oceanogr. 51: 1915–1924.

Winder, M., Schindler, D. E., Essington, T. E. & Litt, A. H., 2009: Disrupted seasonal clockwork

in the population dynamics of a freshwater copepod by climate warming. – Limnol.

Oceanogr. 54: 2493–2505.

Wojtal, A., Frankiewicz, P. & Zalewski, M., 1999: The role of the invertebrate predator

Leptodora kindtii in the trophic cascade of a lowland reservoir. – Hydrobiologia 416:

215–223.

121

Table and Figures

Table 1: Names and occurrences (% of total samples) of zooplankton taxa collected in

Vancouver Lake. Taxa occurring in <3% of samples were not included in statistical analyses.

*Bosminidae includes both B. longirostris and B. coregoni. **Nauplii includes all copepod nauplii except for P. forbesi.

Taxa % Occurrence Cladocera Bosminidae* 93.57 Daphnia retrocurva 88.89 Other Daphnia sp. 42.69 Chydorus sp. 15.80 Leptodora kindtii 15.79 Ceriodaphnia sp. 9.36 Diaphnosoma sp. 5.26 Eurycercus sp. 4.09 Alona sp. 1.17 Rotifera Polyarthra sp. 86.55 Brachionus sp. 74.27 Asplanchna sp. 70.18 Keratella sp. 67.25 Kellicottia sp. 7.02 Filina longiseta 0.58 Copepoda Nauplii ** 100 Cyclopoida Copepodites 99.41 Diacyclops thomasi 97.06 Harpacticoid 3.5 Calanoid Copepodites 20.47 Leptodiaptomus sp. 16.96 Pseudodiaptomus forbesi Nauplii 16.37 Copepodites 2.34 Adult 1.75 Skistodiaptomus sp. 4.68 Other Oligochaeta 2.94 Chironomidae 2.35 Nematoda 1.18

122

Arachnidae 0.59 Chironomidae 0.59 Polychaeta 0.59

Figure 1: Vancouver Lake, a 630 ha shallow, eutrophic lake located in Clark County, southwest

Washington state. Inflows into Vancouver Lake are via the flushing channel, Burnt Bridge

Creek, Lake River, and Salmon Creek. Lake River is the only outflow from Vancouver Lake.

The filled circle represents the sampling station located at the end of a dock at the Vancouver

Lake Sailing Club.

123

a 0 1 2 3 4 Depth (m) Depth Station depth 5 Secchi depth

306 b Temperature (oC) 25 DO (mg L-1) 20 15 10 5

100000 25000

c )

Chl a -1 ) 20000 1000

-1

Cyanobacteria biomass g L 15000 

g L

 100

(

a 10000

Chl Chl 10 5000

Biomass (C (C Biomass 351 0 d 30 PO4-P 25 NH4-N 20

M  15 10 5 0 e 1200 60 1000 SiO4-Si 50

M) NO -N 3 M)

 800 40

 600 30

-Si ( -Si

-N ( -N

4 400 20

SiO 200 10 0 0

Oct-05Dec-05 Feb-06 Apr-06 Jun-06 Aug-06 Oct-06 Dec-06 Feb-07 Apr-07 Jun-07 Aug-07 Oct-07 Dec-07 Feb-08 Apr-08 Jun-08 Aug-08 Oct-08 Dec-08 Feb-09 Apr-09 Jun-09 Aug-09 Oct-09 Dec-09 Feb-10 Apr-10 Jun-10 Aug-10 Oct-10 Dec-10 Feb-11 Apr-11 Jun-11 Aug-11 Oct-11

Figure 2: (a) Station depth and Secchi depth, (b) DO and temperature, (c) chl a concentration in

Vancouver Lake from October 2005 through September 2010, and May 2011 through October

2011; and cyanobacterial biomass from February 2007 through September 2010. Concentrations of (d) PO4-P and NH4-N, and (e) SiO4-Si and NO3-N measured from May 2007 through

September 2010.

124

Figure 3. (a) Distribution of dominant clusters and total zooplankton abundance (individuals m-3), and (b) relative zooplankton abundance (%), October 2005 through October 2011, grouped by order.

125

Figure 4: Panels (a) through (d) show abundance of zooplankton taxa representing each cluster determined by indicator species analysis (p<0.05). Taxa in bold are exclusively faithful to the cluster.

126

Cluster a Cluster b Cluster Cluster 1 1 1 2 1 2 2 3 2 3 3 4 3 4 4 4

Secchi depth Daphnia spp Seasonality Daphnia spp Daphnia spp Asplanchna D. retrocurva Bosminidae Keratella Chydorus Copepodites Asplanchna D. retrocurva Asplanchna D. retrocurva Cal.cope Bosminidae Bosminidae Brachionous Keratella KeratellaChl a Chydorus Chydorus L. kindtii

Copepodites Copepodites Axis3(20.9%)

Axis3(20.9%) Cal.adult Cal.cope Cal.cope Brachionous Brachionous

L. kindtii L. kindtii Axis3(20.9%) Cal.adult Axis3(20.9%) Cal.adult

Axis 2 (44.0%) Axis 2 (44.0%)

Figure 5: NMDS ordination of zooplankton samples and environmental conditions from October

Axis 2 (44.0%) 2005 through SeptemberAxis 2 (44.0%) 2011. Axis 1 (not shown) represented an additional 22.5% of explained

variance among samples. Joint plot cut offs for both environmental vectors (a) and species

associations (b) were set at r2>0.2. The magnitude and orientation of the vector represents the

strength and direction (positive or negative), respectively, of the association.

127

Cluster Year a Cluster b 1 Year 2007 1 2 2007 2008 2 3 2009 4 2008 2010 3 2009 4 2010

PO4.P PO4.P Chl.a Chl.a TempCyanoBio Temp

CyanoBio Axis2(34.4%) PO4.P Axis2(34.4%) PO4.P Chl.a Chl.a

SeasonalTempCyanoBio Temp

SeasonalCyanoBio

Axis2(34.4%) Axis2(34.4%)

Seasonal Seasonal

Axis 1 (24.5%) Axis 1 (24.5%)

Figure 6: NMDS ordination of zooplankton samples and environmental variables, including

cyanobacteriaAxis 1 (24.5%) biomass, in Vancouver Lake. Axis 2 (notAxis shown) 1 (24.5%) represented an additional 34.4%

of explained variance among samples. Joint plot cut offs for environmental vectors were set to

r2>0.2. Environmental vectors are shown in relation to sampling points defined by zooplankton

community clusters (a) and year (b).

128

Figure 7: Occurrence and abundance of non-native zooplankton taxa in Vancouver Lake pre- introduction (October 2005 – February 2007) and post-introduction (March 2010 – October

2011). (a) Abundance of P. forbesi of different life stages; (b) abundances of B. longirostris and

B. coregoni.

129

Figure 8: NMDS ordination of zooplankton samples representing pre-introduction (solid triangles) and post-introduction (circles) of non-native zooplankton. Axis 3 (not shown) represented 19.9% of the explained variance among samples. Joint plot cut offs for environmental and species vectors were r2>0.2.

130

Supplementary Information

The effects of eutrophication and invasive species on zooplankton community dynamics in a shallow temperate lake

Tammy A. Lee*1, Stephen M. Bollens1,2, Gretchen Rollwagen-Bollens1,2, and Joshua E.

Emerson1

*Corresponding author: [email protected]

1 School of the Environment, Washington State University, 14204 NE Salmon Creek Avenue,

Vancouver, Washington 98686, USA

2 School of Biological Sciences, Washington State University, 14204 NE Salmon Creek Avenue,

Vancouver, Washington 98686, USA

131

a 3e+6 5 Cladocera 3e+6 Calanoid Cyclopoid 4 Harpacticoid 2e+6 Nauplii 3 -3 Rotifera 2e+6 Other

Cluster Ind m Cluster 2 1e+6

1 5e+5

0 0 100 b 80

40 20

Relative abundance (%) abundance Relative

Oct-05 Jan-06 Apr-06 Jul-06 Oct-06 Jan-07 Apr-07 Jul-07 Oct-07 Jan-08 Apr-08 Jul-08 Oct-08 Jan-09 Apr-09 Jul-09 Oct-09 Jan-10 Apr-10 Jul-10 Oct-10 Jan-11 Apr-11 Jul-11 Oct-11

Figure S1: (a) Distribution of dominant cluster and total zooplankton abundance (individuals m-

3), and (b) relative zooplankton abundance (%) in Vancouver Lake from October 2005 through

October 2011, grouped by order.

132

CHAPTER FIVE

OVERESTIMATING THE POTENTIAL INFLUENCE OF INTERNAL PHOSPHORUS

LOADING DUE TO WIND-DRIVEN SEDIMENT RESUSPENSION IN A TIDALLY

INFLUENCED SHALLOW FRESHWATER LAKE: A MODELING APPROACH

Tammy A. Lee*1, Gretchen Rollwagen-Bollens1,2, Stephen M. Bollens1,2

1School of the Environment, Washington State University, 14204 NE Salmon Creek Avenue,

Vancouver, Washington 98686, USA

2School of Biological Sciences, Washington State University, 14204 NE Salmon Creek Avenue,

Vancouver, Washington 98686, USA

*Corresponding author: [email protected]

Running title: Modeling wind-driven sediment resuspension

Abstract

Wind-driven sediment resuspension is often the predominant process contributing to increased turbidity and suspended solids in shallow lake systems, and may also indirectly lead to internal phosphorus loading. Vancouver Lake, located in southwest Washington state, is a tidally-influenced shallow freshwater lake with a mean depth of ~1 m during the summer, a period when cyanobacterial blooms often result in lake closure. Increased concentrations of orthophosphate, ranging from 50 – 200 µg L-1, have been observed during the late summer months. We hypothesized that due to Vancouver Lake’s size, shallow depth, and surrounding topography, wind-driven sediment resuspension may significantly contribute to the internal

133 loading of phosphorus. Using wave theory and conservative and known estimates of sediment characteristics we developed a mechanistic model to test our hypothesis by comparing simulated data and observational data from 2007 through 2012. In addition, we examined how altering fetch affects wind-driven sediment resuspension as a potential management option for reducing suspended sediments. The results of our model demonstrated that resuspended sediments were highly variable. More interestingly, when observed lake depth was shallowest, during late summer months, simulated wind-driven sediment resuspension concentrations were lowest. This suggests that the observed increases in orthophosphate during the late summer months are most likely due to processes other than wind-driven sediment resuspension. Thus, while creating artificial barriers to reduce fetch distances may decrease the effects of wind-driven sediment resuspension throughout the year, it may not be the most effective solution for reducing internal phosphorus loading. Other mechanisms responsible for internal phosphorus loading, such as bioturbation and redox conditions, are recommended for further study.

Key words: mechanistic model, eutrophic, phosphorus, wind-driven waves, sediment resuspension

Introduction

As harmful algal blooms and other consequences of eutrophication become more frequent, understanding nutrient dynamics in shallow aquatic systems has become one of the main challenges in water quality management and restoration practices (Ibelings et al. 2007;

Gulati et al. 2008; Søndergaard et al. 2013). In both shallow lakes and estuaries internal loading of nutrients, specifically phosphorus (P), can be a significant source and sink, i.e., nutrients can

134 accumulate in sediments and also be released back into the water column under certain environmental conditions (Koski-Vähälä & Hartikainen 2001; Søndergaard et al. 2003; Lake et al. 2007). In estuaries, the amount of external nutrient sources do not always account for observed primary production, and thus primary producers must be relying, at least in part, on internal loading (Corbett 2010). In systems with long residence times (e.g. lakes, ponds, and reservoirs), reducing internal pools of phosphorus can be a management challenge (Gurkan et al.

2006), especially since sediment-bound phosphorus can account for up to 100 times the amount in the water column and can potentially become reintroduced into the water column for biological uptake (Søndergaard et al. 2003). Such resuspension of sediment-bound phosphorous can provide continual resources for harmful algal blooms for upwards of 20 years (Lewis et al.

2007)

Vancouver Lake, located in southwest Washington state, USA, is a tidally-influenced, shallow, eutrophic freshwater lake that is hydrologically connected to the lower Columbia River.

Over the past century the lake has slowly filled in to a mean depth of ~1 m. Due to decreased water quality and increased frequency of cyanobacterial blooms, a flushing channel and dredging operation were commissioned in the early 1980s. The benefits of these efforts were short-lived, as high sediment loads and cyanobacterial blooms continued to be a problem. In a two-year study of nutrient and water budgets in Vancouver Lake (October 2010 through October 2012), net export of total suspended solids (TSS), total phosphorus (TP), and orthophosphate (PO4-P) were observed during summer months, concomitant with monthly observations of increased TSS, TP, and PO4-P concentrations (Sheibley et al. 2014). Previous research has described dramatic increases of dissolved PO4-P concentrations during the summer months, from 2007 through

2010, ranging from 50 to 200 times greater than at other times of the year (Rollwagen-Bollens et

135 al., 2013; Lee et al., 2015; Rose et al., in review). These results suggest that there is a potentially significant internal contribution of phosphorus in Vancouver Lake, although the exact processes responsible for this remain unclear.

In general, internal loading of phosphorus occurs when phosphorus-laden sediments accumulate within lake beds and then are re-released into the water column via organic matter mineralization, changes in redox status, differences in concentration gradient between the sediment and water column, and sediment resuspension (Istvánovics et al. 2004). In shallow lakes, sediment resuspension can be a dominant cause of elevated water column phosphorus concentrations via increased diffusion from changes in concentration gradients as phosphorus- bound sediment is introduced into the water column, as physical mixing of pore water with lake water, and physical release of particulate and labile phosphorus from sediments into the water column (Boström et al. 1988; Havens et al. 2007) . Resuspension may be caused by bioturbation, recreational-related activities such as boat motors, and wind-driven waves. While relationships between wind-driven waves and sediment resuspension have been well studied (Carper &

Bachmann 1984; Hamilton & Mitchell 1997; Penning et al. 2013), there are few prescriptive lake ecosystem models to assess the effects of different mechanisms of internal phosphorus loading

(Søndergaard et al. 2003).

Most lakes lack long-term, high-resolution monitoring data, so mechanistic models to describe internal P loading are a critical tool for natural resource managers and ecosystem ecologists. Models can provide additional insight by simulating conditions based on available data and theory, which can highlight knowledge gaps and serve as a springboard for future hypothesis-driven research. Thus, our study focused on the development and application of a model representing the potential effects of wind-driven sediment resuspension on internal

136 phosphorus loading to help direct future management actions. We applied wave theory in conjunction with a six-year dataset of hydrologic, weather, and phosphorus data to assess the potential contribution of wind-driven sediment resuspension on internal phosphorus. This approach of coupling observational data with fundamental lake processes for model development permits greater flexibility in estimating the contribution of each variable and mechanism (e.g. sediment resuspension) (Burger et al. 2008; Afshar et al. 2012).

Methods

Study site

Vancouver Lake (Fig. 1) is a 930 ha shallow, non-stratifying, tidally-influenced (but without salt water intrusion), freshwater lake located in southwest Washington state, with a mean depth of ~1 m and a drainage area of up to 262 km2. The area immediately surrounding the lake is predominantly undeveloped with a preserved wildlife habitat along the northern, western, and southern perimeter. Along the eastern shoreline, south of Burnt Bridge Creek, are several small housing and recreational developments. Of the four major hydrological flows into the lake, two are connected directly to the Columbia River via a one-way flushing channel at the southwest edge of the lake and Lake River located at the north end of the lake. Depending on tidal flow,

Lake River either flows into Vancouver Lake (flood tides) or empties into the Columbia River

(ebb tides). The remaining two freshwater inputs are from Salmon Creek (to the north) and Burnt

Bridge Creek (to the southeast); however, because Salmon Creek is connected to Lake River, inputs from Salmon Creek also depend on tidal flow. Water and nutrient budgets for Vancouver

Lake are well described by (Sheibley et al. 2014). The lake bottom is mainly unconsolidated fine sediment, and virtually no macrophyte growth in or around the lake edge. Previous studies of

137

Vancouver Lake include in-depth analyses on plankton community dynamics (Lee, Rollwagen-

Bollens, Bollens, et al. 2015; Lee, Rollwagen-Bollens, & Bollens 2015), and studies on microzooplankton grazing (Boyer et al. 2011) and mesozooplankton grazing (Rollwagen-Bollens et al. 2013).

Model approach

Our model was initially developed based on the general understanding of feedbacks characterizing internal P loading processes, and then subsequently constrained using conditions specific to Vancouver Lake to assess internal phosphorus load due to wind-driven sediment resuspension (Fig. 2). Specifically, we used wave theory to calculate potential sediment resuspension events linking total phosphorus (TP) in the sediment to TP and orthophosphate

(PO4-P) in the water column. Model building and simulations were done in STELLA v10.0.2

(ISEE Systems, Lebanon, NH, USA). All environmental parameters used in the model are summarized in Table 1.

Model formulations

To estimate the effects of wind-driven sediment resuspension we followed (Chapra 1997) to calculate wave period (T) (eq. 1) and wave height (Hw) (eq. 2):

(1)

(2)

138

Where F is the effective fetch (m), g is gravity (9.8 m s-2), h is lake depth (m), and U is wind speed (m s-1). Because Vancouver Lake is very shallow, wavelength was calculated following

(Fenton & McKee 1990) using the deep water wavelength calculation:

(3)

(4)

Where Lo is the deep water wavelength calculated using wave period, and L is the resulting wavelength in a shallow lake.

Total bottom shear stress can be calculated as the sum of wave and current shear stress, but in shallow water environments, where wave stress can be considerably greater (Luettich et al.

1990; Hawley & Lesht 1992; Teeter et al. 2001), we approximated total bottom shear stress as wave shear stress (τ) (Luettich et al. 1990):

(5)

Where ρ is water density, fw is the friction factor, and Ub is the maximum wave orbital velocity.

The friction factor was calculated according to (Dyer 1986):

139

(6)

(7)

(8)

The kinematic viscosity of water (v) was 0.9 m2 s-1.

Rate of resuspension, or erosion rate (E), was calculated as a function of τ (Partheniades,

1965):

(9)

-2 -1 Where M is the erodibility coefficient (kg m s ), and τce is the critical shear stress for erosion

(N m-2). The rate or deposition (D), was calculated as:

(10)

-1 -3 Where ws is the settling velocity (m s ), C is the sediment concentration (g cm ), and τcd is the critical shear stress for deposition (N m-2). To calculate lake volume we used a linear regression

140 using known daily lake volume (L) and daily stage height (m), where lake volume =

9.55x109*stage-3.20x108 (n=711, r>0.99, p<0.001).

Model calibration and sensitivity analysis

Model calibration was performed using lake stage, wind speed, and TSS data collected by

Shiebley et al. (2014) from January 2011 through August 2012. Mean lake stage was recorded daily, and was used to represent lake depth (h) (herein referred to as lake depth) in calculating wave characteristics and sediment resuspension. Relevant physical time scales in shallow lakes and estuaries are on the order of seconds to minutes, thus we used 20-s gust speeds over five minute time intervals. Gust speeds were obtained from a nearby weather station (Wunderground station KWAVANCO27, using http://oco-carbon.com) located ~2 km northeast of the lake and adjusted to velocity at 10 m above lake elevation. Additional model calibration parameters included fetch, erodibility coefficient (M), critical shear stress of erosion (τce) and deposition

(τcd), and settling velocity (ws). To calibrate the model, fetch was set at 3000 m to represent the approximate maximum fetch length from west to east in Vancouver Lake, and northwest to southeast (to avoid the island; Fig. 1). Fetch length was estimated with the measuring tool in

ArcGIS v.10.0.2 (ESRI, Redlands, CA, USA). Because no sediment resuspension experiments have been performed in Vancouver Lake, we chose the most conservative parameter values from a range of published values from lakes having similar characteristics (Table 1). All simulations were run with 5-minute time steps. Observed TSS concentrations were compared against corresponding simulated suspended sediment concentrations.

We also performed a sensitivity analysis on our model to examine the potential effects of reduced fetch under management conditions, including floating wetlands and artificial islands

141

(Gulati & Van Donk 2002; Kelderman et al. 2012). Fetch values were set to 500, 1,000, and

2,000 m.

To assess the potential contribution of sediment resuspension to phosphorus loading, we used the phosphorus content analysis of (Sheibley et al. 2014), who reported that within the first five centimeters of lake bed sediment (~7.1x108 kg) there is a mean of 1,100 mg TP kg-1 of sediment, of which approximately three percent is available inorganic phosphorus. We then applied the calibrated model to assess if wind-driven sediment resuspension could explain the observed annual increases of orthophosphate from 2007 through 2010 (Lee, Rollwagen-Bollens,

& Bollens 2015). The model was run from July through October 2007, and April through

September of 2008, 2009 and 2010, when lake depth and orthophosphate measurements were collected weekly and wind gust velocity was available.

Results

Modeled resuspended sediment versus observed TSS concentration

Environmental characteristics influencing wind-driven sediment resuspension included wind gust velocity and lake stage. Wind gusts were variable and did not seem to exhibit any seasonal patterns (Fig 3a). Lake depth exhibited seasonal fluctuations (Fig. 3b), where annual increases in depth (1.8 – 4.9 m) occurred during winter rainy periods and spring months, which coincided with the spring freshet from snowmelt draining into the Columbia River. Annual decreases in depth occurred during the late summer, with a mean of ~1 m.

The relationship between lake depth and wind gust velocity on sediment resuspension was made very apparent when we examined the mass and concentration of resuspended sediments. Simulated resuspended sediment varied from January 2011 through August 2012 and

142 ranged from zero to ~16,000 kg of sediment (Fig. 4). Modeled resuspended sediment concentrations tended to be less than corresponding observed TSS concentrations (Fig. 4). This is not unexpected because TSS does not differentiate the type or source of particulate matter which includes other substances such as phytoplankton. Although we were able to compare monthly measured TSS concentrations with corresponding simulated resuspended sediment concentrations, due to the frequency discrepancy between observed and simulated measurements, we were unable to assess how the highly dynamic simulated response of wind- driven sediment resuspension compared to observations. For example, between the two sampled time points of November 2011 and December 2011, there was a drastic increase in simulated resuspended sediment concentrations, yet at both sampled points the simulated results were less than the observed results. Conversely, when lake depth was shallowest in September 2011, simulated suspended sediment concentration was also lowest when TSS was highest.

Effect of depth and fetch on sediment resuspension

The simulated variability of resuspended sediments can be described as a function of gust velocity and lake depth under constant fetch conditions. From a management perspective, altering sediment characteristics (e.g. M, τce, τce) is not feasible, but reducing fetch (e.g. creating floating wetlands, artificial islands) and increasing lake depth (e.g. dredging) are, at least in theory, possible. We applied a sensitivity analysis of three additional fetch lengths of 500, 1,000, and 2,000 m. In general, reducing fetch length decreased the effect of wind gust on sediment resuspension showing a dramatic decrease in the amount of resuspended sediments (Fig. 5). In particular, at shallower lake depths more energy via wind gust was needed to resuspend sediments at shorter fetch. Alternatively, increasing lake depth via dredging may be just as

143 effective in reducing sediment resuspension. For example, dredging the lake to 4.5 m with a fetch of 3,000 m requires a minimum wind gust of ~10 m s-1 to cause sediment resuspension

(Fig. 5a); and under these conditions at higher gust velocities, the amount of resuspended sediments is less than reducing lake fetch to 500 m at mean lake depth of ~1 m (Fig. 5d).

Potential contribution of wind-driven sediment resuspension on internal phosphorus loading

To assess the potential contribution of wind-driven sediment resuspension on internal phosphorus loading, we compared PO4-P concentration data from late spring-early summer of

2007 through 2010 (Lee et al. 2015a) with simulated model results for each summer from 2007 through 2010. Similar to 2011 through 2012 (Sheibley et al. 2014), resuspended sediments were highly variable and ranged from zero to ~8,000 kg throughout the simulation for summer months from 2007 through 2010. Under simulated conditions orthophosphate concentrations due to wind-driven sediment resuspension were several orders of magnitude less than observed summer orthophosphate concentrations from 2007 through 2010 (Fig. 6). In comparison to observed seasonal peaks ranging from 50 to over 200 µg L-1 during the late summer months, simulated orthophosphate concentrations were highly variable each year and did not exhibit any seasonal peaks in concentration (Fig. 6).

Discussion

Several mechanistic models exist for accurately predicting the effects of wind-driven sediment resuspension (Chung et al. 2009); (de Vicente et al. 2006). However, many of these models require a priori information, including a combination of high resolution spatial and/or temporal data, and in situ or experimental measurements of related sediment characteristics (e.g.

144

M, τce, τce,ws) that may not be readily available. In the absence of sediment characteristics specific to Vancouver Lake, we chose conservative values based on published literature of similar lakes to estimate wind-driven sediment resuspension. Our goal was not to accurately predict temporal variation in wind-driven sediment resuspension, but rather, to assess if wind-driven sediment resuspension could ever be a dominant mechanism for internal phosphorus loading in Vancouver

Lake. Our findings suggest that wind-driven sediment resuspension, although highly variable throughout a year, does not explain the seasonally high orthophosphate concentrations observed during the summer.

Resuspended sediments

Vancouver Lake is a hydrologically dynamic system with lake levels fluctuating throughout the year, from a minimum depth of approximately one meter to a maximum depth of approximately five meters. We expected that the influence of wind-driven sediment resuspension would be most apparent during the summer months, when lake levels were lowest. However, this was not the case — simulated results from 2011 through 2012 indicated that wind-driven sediment resuspension was greatest during spring and autumn months. Surprisingly, the contribution of wind-driven sediment resuspension predicted by the model was dramatically less than expected during the summer months (<2.0 mg L-1), when the lake is observed to be very turbid and to have relatively high TSS concentrations (72 mg L-1). Similarly, simulated results of resuspended sediments during the summer months of 2007 through 2010 were very low, and were not enough to explain extremely high observed turbidity (e.g., very low Secchi depths of

~0.30 m; Lee et al., 2015). Incidentally, no significant relationship was found between empirical wind and turbidity data for Vancouver Lake, as both factors were highly variable (Sheibley et al.

145

2014). Thus the increase in TSS during the late summer must be attributed to causes other than wind-driven sediment resuspension, such as increased phytoplankton biomass or increases in suspended sediments due to other mechanisms.

Given the relatively high residence time (mean of 13 days), and variable sediment loading of Vancouver Lake (Sheibley et al. 2014), variable flows and tidal currents can also contribute to resuspended sediments. Transport of sediment via variable inflows and outflows of sediments from external sources (e.g. Burnt Bridge Creek, Lake River, and the flushing channel) can alter suspended sediment concentrations that are not captured in this model (Luettich et al.

1990; Hamilton & Mitchell 1997; Chao et al. 2008). Less studied are the effects of tidal currents on sediment resuspension in a shallow lake system; however, the effects of tidal currents in other aquatic systems such as shallow estuaries have been shown to effect sediment resuspension (van

Maren & Hoekstra 2004; Venier et al. 2014). Another, less commonly investigated but potentially significant source for sediment resuspension is bioturbation by fish through foraging within the benthos (Scheffer et al. 2003). Brown bullhead, white crappie, and common carp make up the majority of fish biomass present in Vancouver Lake (Caromile et al. 2000). Both white crappie and common carp have been shown to tolerate poor water quality conditions of low dissolved oxygen and high turbidity (Goetz et al. 2015; Benito et al. 2015). We suggest that fish bioturbation may also be a significant source of sediment resuspension in Vancouver Lake and we recommend that further studies be undertaken to assess this behavior.

Climate change is yet another factor that might influence suspended sediments in the

Columbia River Basin generally, and Vancouver Lake specifically. Climate projections for the

Columbia River Basin predict earlier spring freshets and decreased flows during the spring (Naik

& Jay 2011). Our results for Vancouver Lake suggest that a reduced spring freshet, which is the

146 predominant cause of increased lake depth, would enhance the effect of wind-driven sediment resuspension. If the lake were to maintain a year-round depth of only 1.0 to 2.0 m, and a fetch between 2,000 and 3,000 m, wind gusts greater than 5.0 m s-1 would stimulate sediment resuspension (Fig. 5a and b).

Internal phosphorus loading due to wind-driven sediment resuspension and other mechanisms

Our simulated results indicated little potential contribution from wind-driven sediment resuspension to affect inorganic phosphorus loading. Based on the phosphorus content in the top five centimeters of Vancouver Lake sediments, there is approximately 22,720 kg of available inorganic phosphorus. Given that mean lake depth during the summer is ~1.0 m, and assuming a completely homogenous system, an estimate of 2.5 mg L-1 orthophosphate could be released into the water column. In comparison to observed peak orthophosphate, the modeled summertime availability of orthophosphate in the water column was several orders of magnitude less. Thus, wind-driven sediment resuspension does not seem to be a significant process contributing to internal phosphorus loading in Vancouver Lake.

Other potential mechanisms that may influence sediment phosphorus release include bioturbation and various biogeochemical pathways. Similar to sediment resuspension, bioturbation caused by macrobenthic fauna can release a significant amount of sediment bound phosphorus to the water column, where it is estimated that benthic flux of nutrients from the sediment to water column is on the order of several magnitudes (Kuwabara et al. 2009). In The

Norfolk Broads, England, bioturbation due to macrobenthic fauna was shown to significantly delay restoration efforts because of its contribution towards internal P loading. After reducing external nutrient sources and biomanipulation of removing fish, the benthic chironomid

147 population increased contributing to significant internal P loading (Phillips et al. 1994).

Biogeochemical processes, including microbial metabolic processes at the sediment water interface, (Boström et al. 1988; Gächter et al. 1988) and changes in redox conditions mediated by chemical or microbial pathways, have also been shown to increase orthophosphate concentration in the water column (Woszczyk et al. 2011; Smith et al. 2011; Penn et al. 2014). Specifically in

Vancouver Lake, during late summer months dissolved oxygen (DO) levels <3.5 mg L-1 were observed near the sediment water interface and throughout the lake water column, concurrent with increased orthophosphate concentrations (Lee et al., 2015; Rose et al., in review)

(Supplemental Fig. 1). At least one study has shown that DO as high as 3.5 mg L-1 can mediate phosphorus release via redox reactions from the sediment to water column (Nürnberg et al.

2013). Thus we recommend further studies on redox conditions and anoxic environments in shallow non-stratifying lakes to better elucidate the effects of this process on internal phosphorus loading.

Management considerations and future directions

Effects of wind-driven waves are often cited as the predominant mechanism for sediment resuspension and its potential influence on internal phosphorus loading in shallow lakes (Gulati et al. 2008; de Vicente et al. 2010; Corbett 2010). We hypothesized that when Vancouver Lake was shallowest, during the late summer months, wind-driven sediment resuspension would be the primary cause of the observed increase in orthophosphate concentrations, which occur coincident with harmful cyanobacterial blooms. We also explored the effect of reducing fetch on sediment resuspension as a potential management action. Comparing the simulated results from our mechanistic model with observations, the contribution of sediment resuspension due to wind-

148 driven waves was very small, indicating that internal phosphorus loading by this mechanism was not enough to support observed orthophosphate concentrations. We also determined that while a reduction in fetch decreased the effect of wind-driven sediment resuspension, it would seem to have a minimal effect in reducing internal phosphorus loading in Vancouver Lake. Our results suggest that future management plans for mitigating internal phosphorus loading will require further studies to examine other physical and biogeochemical processes that mediate the release of sediment bound phosphorus, such as bioturbation, variable inflows and currents, and climate change induced alterations in the timing of spring freshets and changes in lake depth.

Acknowledgements

We would like to thank Rich Sheibley with the United States Geological Survey for data provided in support of the model, and J. Ryan Bellmore for advice on model development.

149

References

Afshar A, Saadatpour M, Marino M. 2012. Development of a complex system dynamic

eutrophication model: Application to Karkheh reservoir. Environ Eng Sci. 29:373–385.

Benito J, Benejam L, Zamora L, García-Berthou E. 2015. Diel Cycle and Effects of Water Flow

on Activity and Use of Depth by Common Carp. Trans Am Fish Soc. 144:491–501.

Boström B, Andersen JM, Fleischer S, Jansson M. 1988. Exchange of phosphorus across the

sediment-water interface. Hydrobiologia. 170:229–244.

Boyer J, Rollwagen-Bollens G, Bollens SM. 2011. Microzooplankton grazing before, during and

after a cyanobacterial bloom in Vancouver Lake, Washington, USA. Aquat Microb Ecol.

64:163–174.

Burger DF, Hamilton DP, Pilditch CA. 2008. Modelling the relative importance of internal and

external nutrient loads on water column nutrient concentrations and phytoplankton

biomass in a shallow polymictic lake. Ecol Model. 211:411–423.

Caromile SJ, Meyer WR, Jackson CS. 2000. 1998 warmwater fish survey of Vancouver Lake,

Clark County. Olympia, Washington: Washington Department of Fish and Wildlife.

Carper GL, Bachmann RW. 1984. Wind resuspension of sediments in a prairie lake. Can J Fish

Aquat Sci. 41:1763–1767.

Chao X, Jia Y, Shields FD, Wang SSY, Cooper CM. 2008. Three-dimensional numerical

modeling of cohesive sediment transport and wind wave impact in a shallow oxbow lake.

Adv Water Resour. 31:1004–1014.

Chapra SC. 1997. Surface Water-Quality Modeling. New York: McGraw-Hill.

150

Chung EG, Bombardelli FA, Schladow SG. 2009. Modeling linkages between sediment

resuspension and water quality in a shallow, eutrophic, wind-exposed lake. Ecol Model.

220:1251–1265.

Corbett DR. 2010. Resuspension and estuarine nutrient cycling: Insights from the Neuse River

Estuary. Biogeosciences. 7:3289–3300.

Dyer KR. 1986. Coastal and Estuarine Sediment Dynamics. [place unknown]: Wiley.

Fenton JD, McKee WD. 1990. On calculating the lengths of water waves. Coast Eng. 14:499–

513.

Gächter R, Meyer JS, Mares A. 1988. Contribution of bacteria to release and fixation of

phosphorus in lake sediments. Limnol Oceanogr. 33:1542–1558.

Goetz D, Miranda LE, Kröger R, Andrews C. 2015. The role of depth in regulating water quality

and fish assemblages in oxbow lakes. Environ Biol Fishes. 98:951–959.

Gulati RD, Dionisio Pires LM, Van Donk E. 2008. Lake restoration studies: Failures, bottlenecks

and prospects of new ecotechnological measures. Limnol - Ecol Manag Inland Waters.

38:233–247.

Gulati RD, Van Donk E. 2002. Lakes in the Netherlands, their origin, eutrophication and

restoration: state-of-the-art review. Hydrobiologia. 478:73–106.

Gurkan Z, Zhang J, Jørgensen SE. 2006. Development of a structurally dynamic model for

forecasting the effects of restoration of Lake Fure, Denmark. Ecol Model. 197:89–102.

Hamilton D, Mitchell S. 1997. Wave-induced shear stresses, plant nutrients and chlorophyll in

seven shallow lakes. Freshw Biol. 38:159–168.

Havens K, Jin K-R, Iricanin N, James R. 2007. Phosphorus dynamics at multiple time scales in

the pelagic zone of a large shallow lake in Florida, USA. Hydrobiologia. 581:25–42.

151

Hawley N, Lesht BM. 1992. Sediment resuspension in Lake St. Clair. Limnol Oceanogr.

37:1720–1737.

Ibelings BW, Portielje R, Lammens EHRR, Noordhuis R, Berg MS van den, Joosse W, Meijer

ML. 2007. Resilience of Alternative Stable States during the Recovery of Shallow Lakes

from Eutrophication: Lake Veluwe as a Case Study. Ecosystems. 10:4–16.

Istvánovics V, Osztoics A, HOnti M. 2004. Dynamics and ecological significance of daily

internal load of phosphorus in shallow Lake Balaton, Hungary. Freshw Biol. 49:232–252.

Kelderman P, De Rozari P, Mukhopadhyay S, Ang’weya RO. 2012. Sediment dynamics in

shallow Lake Markermeer, The Netherlands: field/laboratory surveys and first results for

a 3-D suspended solids model. Water Sci Technol. 66:1984–1990.

Koski-Vähälä J, Hartikainen H. 2001. Assessment of the risk of phosphorus loading due to

resuspended sediment. J Environ Qual. 30:960–966.

Kuwabara JS, Topping BR, Lynch DD, Carter JL, Essaid HI. 2009. Benthic nutrient sources to

hypereutrophic Upper Klamath Lake, Oregon, USA. Environ Toxicol Chem. 28:516–524.

Lake BA, Coolidge KM, Norton SA, Amirbahman A. 2007. Factors contributing to the internal

loading of phosphorus from anoxic sediments in six Maine, USA, lakes. Sci Total

Environ. 373:534–541.

Lee TA, Rollwagen-Bollens G, Bollens SM. 2015. The influence of water quality variables on

cyanobacterial blooms and phytoplankton community composition in a shallow

temperate lake. Environ Monit Assess. 187:6.

Lee TA, Rollwagen-Bollens G, Bollens SM, Faber-Hammond JJ. 2015. Environmental influence

on cyanobacteria abundance and microcystin toxin production in a shallow temperate

lake. Ecotoxicol Environ Saf. 114:318–325.

152

Lewis GN, Auer MT, Xiang X, Penn MR. 2007. Modeling phosphorus flux in the sediments of

Onondaga Lake: Insights on the timing of lake response and recovery. Ecol Model.

209:121–135.

Luettich RA, Harleman DR, Somlyody L. 1990. Dynamic behavior of suspended sediment

concentrations in a shallow lake perturbed by episodic wind events. Limnol Oceanogr.

35:1050–1067. van Maren D., Hoekstra P. 2004. Seasonal variation of hydrodynamics and sediment dynamics in

a shallow subtropical estuary: the Ba Lat River, Vietnam. Estuar Coast Shelf Sci. 60:529–

540.

Naik PK, Jay DA. 2011. Distinguishing human and climate influences on the Columbia River:

Changes in mean flow and sediment transport. J Hydrol. 404:259–277.

Nürnberg GK, LaZerte BD, Loh PS, Molot LA. 2013. Quantification of internal phosphorus load

in large, partially polymictic and mesotrophic Lake Simcoe, Ontario. J Gt Lakes Res.

39:271–279.

Penn C, McGrath J, Bowen J, Wilson S. 2014. Phosphorus removal structures: A management

option for legacy phosphorus. J Soil Water Conserv. 69:51A–56A.

Penning W, Genseberger M, Uittenbogaard R, Cornelisse J. 2013. Quantifying measures to limit

wind-driven resuspension of sediments for improvement of the ecological quality in some

shallow Dutch lakes. Hydrobiologia. 710:279–295.

Rollwagen-Bollens G, Bollens SM, Gonzalez A, Zimmerman J, Lee T, Emerson J. 2013.

Feeding dynamics of the copepod Diacyclops thomasi before, during and following

filamentous cyanobacteria blooms in a large, shallow temperate lake. Hydrobiologia.

705:101–118.

153

Schaaff E, Grenz C, Pinazo C, Lansard B. 2006. Field and laboratory measurements of sediment

erodibility: A comparison. J Sea Res. 55:30–42.

Scheffer M, Portielje R, Zambrano L. 2003. Fish facilitate wave resuspension of sediment.

Limnol Oceanogr. 48:1920–1926.

Sheibley RW, Foreman JR, Marshall CA, Welch WB. 2014. Water and nutrient budgets for

Vancouver Lake, Vancouver, Washington, October 2010-October 2012.

http://dx.doi.org/10.3133/sir20145201

Smith L, Watzin MC, Druschel G. 2011. Relating sediment phosphorus mobility to seasonal and

diel redox fluctuations at the sediment–water interface in a eutrophic freshwater lake.

Limnol Oceanogr. 56:2251–2264.

Søndergaard M, Bjerring R, Jeppesen E. 2013. Persistent internal phosphorus loading during

summer in shallow eutrophic lakes. Hydrobiologia. 710:95–107.

Søndergaard M, Jensen JP, Jeppesen E. 2003. Role of sediment and internal loading of

phosphorus in shallow lakes. Hydrobiologia. 506:135–145.

Teeter AM, Johnson BH, Berger C, Stelling G, Scheffner NW, Garcia MH, Parchure TM. 2001.

Hydrodynamic and sediment transport modeling with emphasis on shallow-water,

vegetated areas (lakes, reservoirs, estuaries and lagoons). Hydrobiologia. 444:1–23.

Venier C, D’Alpaos A, Marani M. 2014. Evaluation of sediment properties using wind and

turbidity observations in the shallow tidal areas of the Venice Lagoon. J Geophys Res

Earth Surf. 119:1604–1616. de Vicente I, Cruz-Pizarro L, Rueda FJ. 2010. Sediment resuspension in two adjacent shallow

coastal lakes: Controlling factors and consequences on phosphate dynamics. Aquat Sci.

72:21–31.

154 de Vicente I, Moreno-Ostos E, Amores V, Rueda F, Cruz-Pizarro L. 2006. Low predictability in

the dynamics of shallow lakes: Implications for their management and restoration.

Wetlands. 26:928–938.

Woszczyk M, Bechtel A, Cieśliński R. 2011. Interactions between microbial degradation of

sedimentary organic matter and lake hydrodynamics in shallow water bodies: insights

from Lake Sarbsko (northern Poland). J Limnol. 70:293–304.

155

Table and Figures

Table 1: List of given and calculated model parameters, where values in bold represent calibrated values.

Symbol Description Units Value Wave Characteristics and Resuspension

Ab Maximum bottom wave excursion amplitude Cm -- kg m-2 s- E Rate of resuspension 1 -- F Fetch M 1000, 2000, 3000 fw Friction coefficient -- -- g Gravity m s-2 9.8 h Stage, lake depth m a,b Hw Wave height m -- L Deep water wavelength m -- Lo Shallow water wavelength m -- kg m-2 s- 3.4 x 10-6, 7.5 x M Erodibility coefficient 1 10-6 c T Wave period s -- U Wind speed m s-1 d -1 Ub Maximum horizontal velocity cm s -- v Kinematic viscosity of water m-2 s-1 0.9 ρ Water density mg mL-1 1 τ Bottom shear stress N m-2 -- -2 e τcd Critical shear stress of deposition N m 0.0, 0.1, 0.18 0.008, 0.01, 0.02 -2 c,e τce Critical shear stress of erosion N m f ws Settling velocity m s-1 0.002, 0.02 a (Sheibley et al. 2014) b (Lee, Rollwagen-Bollens, & Bollens 2015) c (Schaaff et al. 2006) d http://oco-carbon.com e (Chao et al. 2008)

156

Figure 1: Vancouver Lake, Washington, is located within the Portland, Oregon – Vancouver,

Washington metropolitan area. The star represents the sampling location for January 2007 through October 2010 conducted by Lee, et al. (2015) and one of the sampling sites conducted by Sheibley et al., (2014). The triangle represents a second sampling location for October 2010 through September 2012 conducted by Sheibley, et al. (2014).

157

Wind Atmospheric deposition Outflows Inflows

Sediment resuspension

Sedimentation Diffusive flux, groundwater Permanent burial

Figure 2: A conceptual model showing phosphorus dynamics in a shallow lake. Inflows, outflows, groundwater seepage, sedimentation, and diffusive flux were calculated on a monthly scale, as described in Sheibley, et al. (2014). The contribution of phosphorus due to wind-driven sediment resuspension (bold) was modeled in our study.

158

Figure 3. Daily maximum wind gust velocity per five minute interval (a) and mean lake stage for

January 2011 through August 2012 (b) (Sheibley et al., 2014).

159

Figure 4. Simulated resuspended sediment (a) and suspended sediment concentration (b) for

January 2011 through August 2012. Monthly measured TSS and corresponding simulated suspended sediment concentration (c).

160

Fetch = 3000 m Fetch = 2000 m

a b

20000 20000

15000 15000

10000 10000

5000 20 5000 20 -1 ) -1 )

15 15

Resuspended sediment (kg) sediment Resuspended (kg) sediment Resuspended 0 10 0 10 5.0 4.5 5.0 4.5 4.0 3.5 5 4.0 3.5 5 3.0 Wind gust (m s 3.0 Wind gust (m s 2.5 2.0 2.5 2.0 Lake depth (m) 1.5 0 Lake depth (m) 1.5 0 1.0 1.0

0 5000 10000 15000 20000 Fetch = 1000 m Fetch = 500 m

c d

20000 20000

15000 15000

10000 10000

5000 20 5000 20 -1 ) -1 )

15 15

Figure 5. Relationship of wind gust velocity and lake depth on sediment resuspension under calibrated conditions of a 3,000 m fetch (a). The light gray mesh area highlights the difference of resuspended sediments under calibrated conditions and reduced fetch conditions of 2,000 (b),

1,000 (c), and 500 m (d).

161

Figure 6. Simulated (top) and observed (bottom) orthophosphate concentrations from July through October of 2007, and April through October of 2008, 2009, and 2010. Observed data were from Lee et al. (2015a).

162

a 8

mol

 4

-P

PO 3

0 b 2

)

-1 4

Depth (ft), DO (mg L (ft), Depth DO 12

Jan-07Apr-07 Jul-07 Oct-07 Jan-08 Apr-08 Jul-08 Oct-08 Jan-09 Apr-09 Jul-09 Oct-09 Jan-10 Apr-10 Jul-10 Oct-10 Jan-11 Apr-11 Jul-11 Oct-11 Jan-12 Apr-12 Jul-12 Oct-12 Jan-13 Apr-13 Jul-13 Oct-13 Jan-14

0 2 4 6 8 10 >12

Supplemental Figure 1. Measured orthophosphate concentration for May 2007 through October

2010 collected by Lee et al. (2015a), November 2010 through October 2012 collected by

Sheibley et al. (2014), and April 2013 through September 2013 collected by Rose et al. (in

Review) (a), and corresponding DO profiles (color scale representing concentration) and lake depth (b).

163

CHAPTER SIX

CONCLUSIONS

Cyanobacterial blooms and associated adverse effects on water quality has become an increasingly global concern. Locally, Vancouver Lake is of regional concern serving as a regional recreational destination and wildlife habitat, thus making it an ideal model system for examining cyanobacterial blooms in a shallow freshwater lake. Specifically, the following objectives were addressed in each successive chapter: 1) assess how water quality factors affect phytoplankton community dynamics focusing on the cyanobacteria community; 2) identify and quantify toxin producing cyanobacterial populations and how water quality variables influence toxin producing cyanobacteria; 3) investigate the potential effects of cyanobacterial blooms on zooplankton community dynamics; and 4) develop a mechanistic model to examine if wind- driven sediment resuspension contributes to internal phosphorus loading, which may significantly influence cyanobacterial blooms. Results from each objective provided new information critical towards understanding and managing cyanobacterial blooms.

The second chapter showed that nutrients were the primary variables associated with the initiation, intensity, and duration of cyanobacterial blooms, and influence seasonal phytoplankton community dynamics. The depletion of NO3-N followed by elevated PO4-P concentration was associated with increased biomass and duration of cyanobacterial blooms, and

NO3-N availability contributed to interannual changes within the summer phytoplankton community. Although these result may be similar compared to other lakes, the mechanisms underlying the observations are likely different. Interestingly, these results also highlight the

164 monitoring methodology of high frequency sampling. Courser sampling periods might have missed the dramatic pulse in PO4-P availability observed during weekly sampling efforts.

The third chapter identified Microcystis as the toxin producing cyanobacteria species present in Vancouver Lake, although Microcystis was rarely detected microscopically. Using qPCR, Microcystis made up for less than 1% of the cyanobacteria community, and toxin- producing Microcystis were even smaller. This small population of toxin-producing cyanobacteria was responsible for high levels of intracellular microcystin suggesting that the effects of a small overlooked species could potentially have greater consequences. Additionally, toxin-producing Microcystis population and intracellular microcystin concentration were found to be associated high levels of PO4-P concentration. As public health officials and natural resource managers continue monitoring toxic cyanobacterial blooms, a multifaceted approach is crucial towards understanding cyanobacterial bloom dynamics.

Cyanobacterial blooms can affect higher trophic levels, thus the fourth chapter examined how cyanobacterial blooms and associated eutrophic factors influence zooplankton community dynamics. Interestingly, additional stressors, specifically non-native invasive zooplankton, were found to also influence zooplankton community composition. Results from this study indicate that the interannual differences in seasonal zooplankton community succession may be directly influenced by interacting effects of turbidity, cyanobacterial blooms, predatory zooplankton, and invasive crustacean zooplankton, and indirectly by PO4-P availability and temperature. These results suggest that two separate management goals – de-eutrophication and managing invasive species – may be in conflict. Additionally, these results underscore the need for future studies examining the effects of native and non-native species composition, how cyanobacterial blooms

165 affect the success of non-native species, and the potential long-term consequences of non-native species invasions on zooplankton community dynamics.

Because PO4-P was shown to be a potentially significant environmental variable influencing cyanobacterial blooms, the fifth chapter was focused on developing a mechanistic model that could explain the observed seasonal pulse of PO4-P concentrations. Based on the literature, hydrology, and geomorphology of Vancouver Lake, it was hypothesized that wind- driven sediment resuspension is the most likely mechanism for internal phosphorus loading. If wind-driven waves were the main mechanism for internal phosphorus loading and high turbidity, then possible mitigation measures could be decreasing fetch (e.g. artificial islands, floating wetlands) or increasing lake depth (e.g. dredging). The results of the model demonstrated that although resuspended sediments were highly variable, it did not explain the increase in summer

PO4-P concentration. Creating artificial barriers to reduce fetch may decrease wind-driven sediment resuspension throughout the year; however, it may not be the most effective solution for reducing internal phosphorus loading. These results highlight the need to examine other mechanisms, such as bioturbation and redox conditions, to better understand and ultimately provide effective management strategies for restoring Vancouver Lake.

In conclusion, managing Vancouver Lake requires a multifaceted approach, from molecular techniques to mechanistic modeling, due to multiple stressors affecting ecosystem dynamics.

166