University of Southern California Dissertations and Theses

Annual pattern and response of the bacterial and microbial eukaryotic communities in an aquatic ecosystem restructured by disturbance

(USC Thesis Other)

Annual pattern and response of the bacterial and microbial eukaryotic communities in an aquatic ecosystem restructured by disturbance

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

ANNUAL PATTERN AND RESPONSE OF THE BACTERIAL AND MICROBIAL
EUKARYOTIC COMMUNITIES IN AN AQUATIC ECOSYSTEM RESTRUCTURED BY
DISTURBANCE

by

Adriane Clark Jones

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

December 2012

Copyright 2012 Adriane Clark Jones
i

Dedication

I dedicate my dissertation to my parents Rick and Sandy Jones, for their grand, perpetual,
obvious, and infinitely appreciated love and support. I love you both to the moon.
ii

Acknowledgments

I could not (would not) have done this alone and I am grateful to all who helped me along the
way. I first want to acknowledge my advisor David Caron for his intellectual support and
encouragement, and for providing opportunities for growth and adventure (Antarctica!). I
appreciate the guidance of my committee: Karla Heidelberg, John Heidelberg, Frank Corsetti,
and Jed Fuhrman. I am grateful to Dave Hambright at the University of Oklahoma for entrusting
me with such a fantastic dataset and to Bruce Rose and Fares Najar at the University of
Oklahoma genome center. In addition I am thankful to James Easton, Anne Easton, Rich Zamor
and Emily Remmel at the University of Oklahoma for sample collection and field measurements,
and to Vivian Liao at the University of Southern California who assisted with sample processing.

I feel fortunate to have been part of an intellectually and emotionally supportive community at
the University of Southern California and Los Angeles as a whole. I especially value the
friendship and support of Diane Kim, Beth Stauffer, Marie-Eve Garneau, Astrid Schnetzer, Pete
Countway, Anand Patel, Mahira Kakajiwala, Jill Sohm, Victoria Campbell, Ilana Gilg, Dan
Heller, Andrew Gaines and Daniela Barbosa Gaines, Ananias Chairez, and Alex Del Real.

And to my family; Dad, Mom, Adam, Donna, Grandparents, Andrew, Noelle, Jackson, and
Caterina I love you and am eternally grateful for your love and support.

The work in this dissertation was funded by a grant from the Oklahoma Department of Wildlife
Conservation through the Sport Fish Restoration Program (grant F-61-R) to K.D.H.

iii

Table of Contents

Dedication i

Acknowledgments ii

Abstract v

Chapter 1: Seasonality and disturbance: annual pattern and response of the bacterial 1
and microbial eukaryotic assemblages in a freshwater ecosystem

Abstract 1
Introduction 2
Methods 5
Results 12
Discussion 23
References 32
Tables 39
Figure Legends 40
Figures 43
Supplemental Information 49
Supplemental Figures 51

Chapter 2: Taxon-specific responses at multiple trophic levels (metazoa, 56
microbial eukaryotes, and bacteria) within a freshwater microbial community
restructured by disturbance events.

Abstract 56
Introduction 57
Methods 59
Results 65
Discussion 74
References 82
Figure Legends 90
Figures 93
Supplemental Information 101
Supplemental Figures 102

iv

Chapter 3: Ecological patterns and relationships among bacteria and microbial 104
eukaryotes derived from network analyses over an annual cycle in a low salinity lake.

Abstract 104
Introduction 105
Methods 106
Results 112
Discussion 121
References 126
Tables 133
Figure Legends 134
Figures 140
Supplemental Tables 147
Supplemental Information 164
Supplemental Figures 166

Bibliography 171
v

Dissertation Abstract

Disturbance events are structuring forces that affect the species richness and composition of
biological communities. Depending on their scale and duration, these events can have
tremendous and long lasting effects on food web structure and ecosystem function.
Consequences and examples of disturbance are well-documented for macrobial communities, but
the degree to which disturbance affects microbial community structure is poorly understood.
Pyrosequencing of 18s (v9 region) and 16s (v6 region) SSU rDNA genes was used to obtain
monthly snapshots of eukaryotic and bacterial diversity, community structure, and community
composition of the microbial assemblages in Lake Texoma, southwest United States. Microbial
eukaryotic (metazoan, protist, fungi, and chlorophytes) and bacterial assemblages were
characterized monthly at two locations (Lebanon Pool and Wilson Creek) for one year within the
lake, significantly affected by two disturbance events: 1) a localized and prolonged (4-month)
bloom of the toxic alga Prymnesium parvum and 2) a large (17 cm) rain event overlaid onto
gradual seasonal environmental change. Interactions and co-occurrence patterns of the microbial
taxa were examined using Spearman correlation analysis and visualized and quantified using
network analyses. Eukaryotic species richness as well as both eukaryotic and bacterial
community structure exhibited cyclical seasonal patterns, including distinct responses to the rain
event. Patterns of connectivity within the microbial association networks at both locations
revealed highly interconnected consortia of taxa and also negative correlations reflected the
seasonality of the lake.
The P. parvum bloom in Lebanon Pool but not Wilson Creek created a natural
experiment in which to directly explore and compare the effects of an Ecosystem Disruptive
Algal Bloom (EDAB) on the microbial community. Microbial species richness was unaffected
vi

by the bloom but the eukaryotic community structure (evenness) and the patterns of both
eukaryotic and bacterial community similarity at bloom and non bloom sites were statistically
distinct during the 4 months of the bloom. The two locations had contrasting taxonomic
compositions of the microbial assemblages over the course of the bloom. The haptophyte (P.
parvum) strongly dominated the eukaryotic community when it bloomed, although high
abundances of fungi and chrysomonads were also observed. In contrast, the community at the
non-bloom site (Wilson Creek) contained a wide diversity of common taxa: haptophytes, ciliates,
cercozoa, chlorophytes, crustaceans and rotifers. Differences in the bacteria were more subtle,
although the actinobacteria were largely absent during the P. parvum bloom.
A comparison of the microbial association networks from the two locations during the P. parvum
bloom disturbance revealed compositional differences, as well as differences within the
interconnectivity patterns of the taxa. A large rain event resulted in massive restructuring of the
microbial communities at both locations and rapid decreases or increases of particular microbial
groups (cercozoa, ciliates, and betaproteobacteria). Our results indicate that the eukaryotic and
bacterial assemblages were significantly structured by disturbance, overlaid on more gradual and
subtle seasonal changes. Blooms of P. parvum are examples of particularly disruptive events
with far-reaching consequences on microbial eukaryotic and bacterial community structure,
trophic composition and interconnectivity within the microbial web as a whole.

1

Chapter 1
Seasonality and disturbance: annual pattern and response of the bacterial and microbial
eukaryotic assemblages in a freshwater ecosystem

Abstract
High-throughput pyrosequencing of SSU rDNA genes was used to obtain monthly snapshots of
eukaryotic and bacterial diversity and community structure at two locations in Lake Texoma,
located in the south central United States over the course of a year. The lake experienced two
disturbance events 1) a localized bloom of Prymnesium parvum restricted to one of the locations
that lasted from January to April, and 2) a large, global (17 cm) rain event in the beginning of
May, overlaid onto seasonal environmental change. Eukaryotic species richness as well as both
eukaryotic and bacterial community similarity exhibited cyclical seasonal patterns, including
distinct responses to the rain event. The P. parvum bloom created a natural experiment in which
to directly explore the effects of an Ecosystem Disruptive Algal Bloom (EDAB) on the microbial
community separated from seasonal changes. Microbial species richness was unaffected by the
bloom, however, the eukaryotic community structure (evenness) and the patterns of both
eukaryotic and bacterial community similarity at bloom and non bloom sites were statistically
distinct during the 4 months of the bloom. These results indicate that the eukaryotic and
bacterial communities were structured by physical and biological disturbances as well as
seasonal environmental forces.

2

Introduction
Microbes [bacteria, archea, and microbial eukaryotes] form the foundation of food webs in
aquatic ecosystems, playing a variety of critical biogeochemical roles such as primary producers,
nutrient recyclers, and links to higher trophic levels. (Caron and Countway, 2009; Fuhrman,
2009). Natural microbial assemblages are typically composed of relatively few common taxa
that perform the majority of ecological roles at any given time, yet incredible taxonomic richness
is maintained in virtually every microbial community even in extreme environments (Amaral-
Zettler et al., 2002; Edgcomb et al., 2002; Lopez-Garcia et al., 2003). Rare taxa comprise most
of the species richness, and have collectively become known as the microbial rare biosphere
(Sogin et al., 2006). There has been considerable speculation and debate recently regarding how
these rare taxa are maintained in natural assemblages, and their potential to contribute to
ecosystem function in an environment (Caron and Countway, 2009; Pedros-Alio, 2006).
The dominant taxa present in microbial communities change rapidly and often as a
consequence of environment change (Heino and Soininen, 2010; Kim et al., 2011). These
changes can be rhythmic and repeating (e.g. in response to diurnal and seasonal cycles) (Gilbert
et al., 2009) or stochastic such as may occur in response to a major environmental disturbance
(Jones et al., 2008). We presently have little understanding of the rules governing the response
and reassembly of microbial communities to environmental change or the degree of resilience to
these forcing factors, in part because of the incredible diversity of these communities. Our
present inability to predict changes in species composition and biogeochemical processes make
observational studies of natural communities highly insightful. (Fuhrman 2009).
The effects of disturbance on communities of macroscopic species are well documented.
Numerous studies have demonstrated that ecosystem disturbance can lead to losses of species
3

richness among plant and animal communities resulting in changes in the trophic structure and
energy flow within environments, with potentially devastating effects. The effects of these
changes on species dominance can radiate throughout food webs, reducing overall biodiversity
and altering the ability of communities to maintain desirable food chains or even conduct
essential ecosystem processes (Caraco et al., 1997; Rothschild et al., 1994). Characterizing the
impacts resulting from reductions of biodiversity among plants and animals on ecosystem
structure and function has become a topic of major concern and ecological research (Estes et al.,
2011). In contrast to this situation for assemblages of large organisms, we presently have a poor
understanding of the potential impact of disturbance on the diversity and community structure of
microbial communities and their subsequent emergent properties.
Harmful algal blooms (HABs; the dominance of a single algal taxon possessing toxic or
noxious properties) are well-documented disruptions of planktonic ecosystems that result in the
contamination, disruption or collapse of aquatic food webs (Buskey et al., 2001; Gobler et al.,
2005) . The detrimental effects of these blooms have been well documented at higher trophic
levels, including impacts on zooplankton, invertebrates, fish, birds and mammals (Scholin et al.,
2000), yet there is limited knowledge of the effects of these events on the structure of the
microbial communities (Michaloudi et al., 2008; Vigil et al., 2009).
Prymnesium parvum is a haptophyte alga that forms harmful algal blooms in brackish
waters across the world, including the United States where such blooms appear to be an
emerging issue. Blooms of P. parvum were first detected in Lake Texoma, south central U.S.
during 2004 and this species has since formed annual blooms that can last for months (Roelke et
al., 2010). The factors contributing to and controlling blooms of P. parvum are not well known,
but this species does have several physiological characteristics that equip it to grow quickly and
4

in a variety of habitats. It has a wide salinity tolerance allowing it to establish in brackish waters.
P. parvum also produces a suite of toxins that negatively affect gill breathing animals and can
lead to massive fish kills in the environment. In the laboratory, these toxins have been shown to
effect zooplankton grazers, and other members of the plankton (Evardsen and Imai, 2006;
Fistarol et al., 2003; Skovgaard and Hansen, 2003). P. parvum is also mixotrophic, combining
phototrophy with a well-developed ability to capture, subdue and ingest a wide variety of
microbial species. This physiological duality presumably provides an ecological advantage
relative to other algae and heterotrophic protists, and thus may play a role in bloom formation
and maintenance (Tillmann, 1998; Tillmann, 2003). Large scale HAB events such as the ones
formed by P. parvum constitute major biological disturbance events in pelagic ecosystems that
can dramatically affect food web structure and function and have been termed Ecosystem
Disruptive Algal Blooms (EDABs) (Gobler and Sunda, 2012; Sunda et al., 2006)
Eukaryotic and bacterial community structure was examined in monthly samples
collected at two locations in Lake Texoma, Oklahoma, USA for one year. Cyclical patterns in
eukaryotic species richness as well as eukaryotic and bacterial community similarity were
observed at both locations that indicated strong seasonal environmental forcing. Two episodic
disturbances were overlaid on this seasonal cycle; a large spring rain event at both locations in
May and a bloom of P. parvum at one of the sampling locations but not the other during the
winter. These events resulted in major changes in microbial community structure and
community similarity that were comparable in magnitude or greater than responses brought
about by seasonal changes. Despite these major shifts in microbial community structure to short-
term, environmental or biological disturbances, community composition returned to a cyclical
annual pattern within the one-year study.
5

Methods
Site description, sample collection and laboratory processing
Lake Texoma (Fig. 1) is located along the Texas and Oklahoma boarder. Salinity in the lake is
generally high. Practical salinity units (PSU) averaged ~0.8 between from 1959 to 2008, and
values can range from 0.5 to 2.5 PSU (Hambright et al., 2010).
Near-shore water samples were collected monthly from November 2008 to October 2009
at two locations (Lebanon Pool and Wilson Creek; Fig. 1) for temperature, specific conductance,
chlorophyll a, nutrients, dissolved oxygen, pH, Prymnesium parvum cell counts, and DNA for
subsequent sequencing of 18s and 16s rRNA genes. Wilson Creek was sampled twice on the
same day in December and this additional sample served as a true replicate to examine the
variance associated with sample collection and processing. The frequency of sampling for
temperature, salinity, chlorophyll a, and P. parvum counts was greater than once a month for part
of the year.
Temperature and salinity were measured in-situ at the time of water collection.
Processing of all other water samples occurred in the laboratory within hours of collection.
Aliquots of 50-100 ml of lake water were filtered onto GF/F filters (Whatman) and stored at -
20°C for later chlorophyll a analysis by fluorometry after acetone extraction (Parsons et al.,
1984) Dissolved oxygen was measured with a YSI (6820 Multi-parameter Water Quality
Monitor) or Hydrolab (H2O Submersible Water Quality Data Transmitter) sonde. pH was
measured with a VWR 8025 pH meter. Water samples were filtered (GF/F) for nutrient
determinations by flow injection auto analysis (Lachat 8500 Quickchem FIA). Soluble reactive
phosphate, ammonia, and nitrate were determined on filtered samples; total dissolved
phosphorus and total dissolved nitrogen on filtered samples digested in acid (P) and alkaline (N)
6

persulfate at 120 °C for 1 h; and total phosphorus and total nitrogen on unfiltered samples
digested as above for total dissolved phosphorus and nitrogen. P. parvum cells were
immediately counted from whole unpreserved water samples using a hemocytometer (minimum
of six subsamples) and a compound microscope at 200-400 x magnification (Hambright et al.,
2010). Water samples of 100-300 ml were filtered onto 47mm GF/F filters (Whatman) and
stored at -80˚C for later DNA extraction, PCR amplification and sequencing.
DNA was extracted from one quarter of each filter as described in (Countway et al.,
2007). DNA was then PCR-amplified with a cocktail of primers targeting the v9 region (~120
bases) of the 18s rRNA gene for eukaryotes and the v6 region (~90 bases) of the 16s rRNA gene
for bacteria (Amaral-Zettler et al., 2009; Sogin et al., 2006). The 18s and 16s rRNA gene
fragments were amplified for each of the 25 samples (two locations for 12 months plus one
duplicate) in 7-10 individual 25 µl PCR reactions each. All reactions were performed according
to the following protocol: 0.45 µM cocktail of each forward and reverse primers, 1x GoTaq Flexi
Reaction Buffer (Promega, Madison WI), 2.5mM magnesium chloride, 200 mM dNTPs, 0.8
µg/ml bovine serum albumin (BSA), 2.5 Units of GoTaq Flexi DNA Polymerase (Promega) and
10 ng of total DNA. The following touchdown PCR protocol was used: 1 cycle (95°C for 2 min),
10 cycles (95°C for 30 seconds, 65°C-decreasing 1°C per cycle for 30 seconds, 72°C for 30
seconds), 18 cycles (95°C for 30 seconds, 55°C for 30 seconds 72°C 30 for seconds), and a final
extension of 72°C for 7 minutes. The multiple PCR reactions were pooled, checked for a band
of appropriate size on an agarose gel, and then cleaned and concentrated (Zymo Research). The
amplicons (~500ng total) were then amended with 454 adaptors and DNA barcodes according to
Roche manufacturing protocols and sequenced with Roche Life Sciences 454 Titanium
chemistry and platform. The 50 libraries (two genes at two locations for twelve months plus one
7

duplicate) were multiplexed on 2 one-quarter sequencing plates. Sequence barcodes and
identifier keys were removed after sequencing using Roche software (Margulies et al., 2005).

Bioinformatic analyses of 18s and 16s rDNA sequence tags
The free software package MOTHUR v1.18.0 (http://www.mothur.org) was used for sequence
processing, taxonomic identification, alignments, OTU calling and alpha diversity analyses
(Schloss et al., 2009). The sequences were screened for quality and only the sequences that met
the following criteria were retained: an average quality score of 25 or higher, zero ambiguous
bases, less than 8 homopolymers, and exact matches to the proximal and distal primers. (Huse et
al., 2007) Primer sequences were then trimmed from the sequence reads passing this quality
assurance.
Plastids were removed from the dataset to avoid biasing bacterial community structure
with eukaryotic plastid sequences. Each 16s sequence was compared against the 16s SILVA
taxonomy reference database within MOTHUR. Sequences identified as cyanobacteria were
manually checked (blastn) against the web-based NCBI (nt) database (Zheng Zhang, 2000) to
ensure they were not plastids. Removing the plastids from the 16s dataset reduced the total
number of 16s sequences by approximately 10%.
The total number of high quality eukaryotic (18s rDNA) sequences/sample ranged from
7,678 to 24,934 for the 25 samples analyzed. Bacterial sequences/sample ranged from 3,597 to
18,017. Each library (all months at both locations) in the 18s (including metazoa; approx 5% of
the sequences) and 16s (without plastids) was randomly subsampled to the month/site with the
fewest number of sequences (7,678, and 3,597 for 18s and 16s, respectively). All 25 subsampled
libraries of each type were then pooled for calling operational taxonomic units (OTUs). The
8

subsampling approach was carried out to avoid artifacts arising from estimating diversity indices
from libraries containing different numbers of sequences (Gihring et al., 2012).
The subsampled pooled datasets were each aligned to a reference SILVA alignment (18s
rDNA or 16s rDNA). A Needleman-Wunsch algorithm (kmer=8) was used with the following
criteria: match (+1), mismatch (-1), gap opening (-2), and gap extension (-1). Sequences that did
not align well to the V9 region of the 18s or V6 region of the 16s were removed from each
dataset (average <2% for 18s and <8% for 16s from any library). We performed a pre-clustering
step on the sequences to condense those with single base pair differences that may have been the
result of sequencing errors (Huse et al., 2010). OTUs were called at 97% similarity for each of
the two subsampled datasets using an average neighbor method. A representative sequence (one
with the smallest total distance to all the other sequences within an OTU) was extracted from
each OTU and compared to the SILVA small subunit ribosomal (v108) using stand alone
BLAST+. Each OTU was assigned an identification based on a best match of the representative
sequence in SILVA. A best match was defined as greater than 95% sequence similarity over
95% of the query sequence.
Prior to normalization, each library (month and location) contained an unequal number of
sequences that differed by as much as 80% among the libraries. OTU calling as described above
was also performed on 18s and 16s datasets prior to subsampling in order to assess the effect that
subsampling might have on the community similarity measurements (see following section).
Rarefaction curves with confidence intervals were generated for each of the libraries before
subsampling via a random sampling technique within MOTHUR.

9

Statistical analyses of diversity, community structure and seasonal patterns
The parametric Chao1 (Chao, 2005; Chao and Lee, 1992) and non-parametric Catchall (Bunge,
2011) richness estimators, as well as the Simpson index of diversity (1-D) which takes into
account taxonomic evenness and richness (Simpson, 1949) were computed with upper and lower
95% confidence intervals using MOTHUR for the subsampled 18s libraries and 16s libraries
without plastids. The Simpson index of diversity was also calculated for the 18s Lebanon Pool
dataset following removal of sequences attributed to Prymnesium, and the dataset subsampled.
All linear regressions, fitting of trend lines and simple correlation analyses were computed and
plotted in SigmaPlot v11.
Cumulative dominance plots, also known as k-dominance plots, were generated for the
18s libraries and 16s libraries using the software package PRIMER v6.1.7. Cumulative
dominance plots show the distributional trends of sequence membership in OTUs. OTUs are
plotted by rank abundance along the X axis, and the cumulative contribution of each successive
OTU is plotted on the Y axis. A perfectly even community (with the X values plotted on a log
scale) would fit an exponential curve, while a community highly dominated by one or a few taxa
would have a steeper slope. These plots are used for comparative purposes to describe and
compare the distribution and dominance of taxa within samples. (Clarke, 1990; Lambshead et
al., 1983; Warwick, 1991) An analysis of similarity (ANOSIM) test was used to statistically test
for differences between the curves. The ANOSIM test computes a test statistic R between 0
(accept the null) or 1 (strongly reject the null) and an associated p-value (Clarke and Warwick,
2001).
The multivariate software package PRIMER v6.1.7 was used to compute Bray-Curtis
(Bray and Curtis, 1957) and Jaccard (Jaccard, 1908) similarity values between each month and
10

location, perform permutation-based statistical tests, and build cluster and non-metric
multidimensional scaling (MDS) plots (Clarke, 1993; Clarke and Warwick, 2001). Bray-Curtis
similarity takes into account shared taxa and the relative contribution of each taxon and does not
take into account shared absences. Bray-Curtis similarity matrices were built for each dataset
(18s and 16s libraries) pre-treated in the following three manners: 1. Subsampled (as noted
above), with OTU abundances converted to relative abundances; 2. Not subsampled, with OTU
abundances converted to relative abundance; and 3. Subsampled, with OTU abundances
converted to relative abundances and square root transformed. In order to assess the similarity of
these data treatments, Spearman correlations between the matrices were computed using second
stage MDS. The subsampled and not subsampled libraries yielded high correlation coefficients
(0.99) for both the 18s libraries and 16s libraries. Comparison of the square-root transformed and
untransformed libraries also yielded high Spearman correlations (0.97) for both datasets. Based
on these high correlations, the remainder of the analyses were conducted using the subsampled,
untransformed OTUs converted to relative abundances.
Jaccard is a presence/absence measure that does not include shared absences. Similarity
matrices with this index were built for each dataset (18s and 16s libraries) pre-treated in the
following two manners: 1. Subsampled (as noted above), and 2. Not subsampled. In order to
assess the similarity of these data treatments, Spearman correlations between the matrices were
computed using second stage MDS. The subsampled and not subsampled libraries yielded high
correlation coefficients (0.99) for the 18s libraries and (0.98) 16s libraries.
The ANOSIM procedure was used to test for differences in community similarities
between the two locations, and between the months within each location. We treated the P.
parvum bloom (documented in Lebanon Pool from January to April) as a factor and used
11

ANOSIM to test for differences in community similarities between the locations during bloom
and non-bloom conditions.
Cluster diagrams were constructed from the Bray Curtis and Jaccard similarity matrices,
and the hierarchical clusters of samples were organized based on group average scores. The
SIMPROF permutation test was used to identify statistically significant samples within the
clusters. The same similarity matrices used to build the dendrograms were converted into ranks
and then used to construct 2-dimensional MDS plots with a Kruskal fit scheme of 1 and 25
restarts. The placement of the samples is considered a good fit when the stress is less than 0.1.
The annual trajectory of the samples was depicted using arrows to connect sequential months for
each location on the plots.
Cyclical patterns at each location within the eukaryotic or bacterial communities were
examined using Spearman correlations to a model matrix. Relationships between the 18s and
16s datasets were examined using Spearman correlations between matrices. Relationships
between the two indices, the Bray-Curtis and Jaccard, were tested by computing Spearman
correlations between the similarity matrices. These correlations were statistically tested using the
RELATE permutation test. (Clarke and Warwick, 2001; Clarke et al., 2006)
The global BEST procedure was used to identify the environmental variables or sets of
variables that best matched the patterns detected in the biological community similarity analyses,
it returns a test statistic and a permuted p-value (Clarke et al., 2008). The BEST routine was
performed with the following environmental parameters that were square root transformed and
normalized: temperature, PSU, dissolved oxygen, pH, total dissolved nitrogen, total nitrogen,
total dissolved phosphorous, total phosphorous, and both total and dissolved molar N:P ratios.
12

The BEST routine was performed a second time without the biological variables of cell counts,
chlorophyll and dissolved oxygen.

Results
Annual patterns of temperature, salinity, pH, dissolved oxygen, nutrients, chlorophyll, and cell
abundances.
Trends in the physical, and chemical parameters of Lake Texoma reflected the seasonality and
two disturbance events of the system 1) Lebanon Pool experienced a 4-month bloom of the alga
Prymnesium parvum (Fig. 2F), and 2) both locations experienced a large (17 cm) and anomalous
rain event on May 3rd 2009 (Supplemental Fig. 1) (McPherson et al., 2007) .
Lebanon Pool and Wilson Creek exhibited similar patterns of surface water temperature,
salinity, pH, dissolved oxygen, and dissolved nutrients over the length of the study period (Fig.
2A, B, C, D, G and H). Water temperature (Fig. 2A), had winter minima of approximately 5°C
during January followed by progressive warming to summer maxima of approximately 30°C
during July, and subsequent cooling through the fall. Seasonal changes in temperature at both
locations displayed good fit (r
2
= 0.8) to sinusoidal regressions with set periods of 365 days
(Supplemental Fig. 2 B and C). The patterns of salinity, pH and dissolved oxygen measurements
each had the highest values from December to April, precipitously decreased at the time of the
rain event in May, and gradually increased through the summer and fall (Fig. 2 B-D). Salinities
(Fig. 2B) at both sites, ranged from 2.5-3 PSU and were relatively constant from November
through April. Salinities decreased to <0.5 PSU in May, gradually increased through September,
and were approximately 1 PSU from August to October. pH values (Fig. 2C) in both locations
ranged from 8 to 8.5 and were relatively constant from November through April. pH decreased
13

to 7.5 in May, and increased to 8 through September. Dissolved oxygen (Fig. 2D) values at both
sites ranged from 10 -13 mg/l from November to April, decreased to 5 or 3 mg/l in May and then
increased to 6 mg/l during June. Dissolved oxygen values in Lebanon Pool increased from June
(6 mg/l) through October (8 mg/l) while values in Wilson Creek remained fairly constant at 6
mg/l through October. Dissolved nitrogen (Fig. 2G solid lines) at the two locations had similar
values (25-75 µM) and fluctuations throughout the year. Dissolved phosphorous measurements
(Fig. 2H solid lines) in the two sites were mostly constant (1-2 µM) throughout the year, with
two major increases (to 3-5 µM) in May and September.
Trends in chlorophyll a (Fig. 2E) concentrations were similar at both study sites, but were
not identical. Month-to-month changes in chlorophyll concentration in Lebanon Pool were
highly variable but overall showed a modest increase (from ~50 to ~90 µg/L) from late-January
to April. Chlorophyll values in Wilson Creek exhibited a less variable, progressive increase
(maximum of 120 µg/L) from mid-March to mid-April. Chlorophyll concentrations at both
locations decreased during May to approximately 4 µg/L, coinciding with the timing of the rain
event (Supplemental Fig. 1) and the decreased values of salinity, pH and dissolved oxygen at the
two locations (Fig. 2B-D). Chlorophyll concentrations at both sites increased steadily from June
onward and reached maxima of 160 and 190 µg/l during early fall (Fig. 2E).
Microscopical counts of P. parvum (Fig. 2F) differed drastically between the two study
sites. Lebanon Pool experienced a large bloom of P. parvum during winter-spring. Cell
abundances increased steadily from January to mid-February at this site, sustained maximal
abundances of 1.5-2x10
5
cells/ml into April, and then decreased rapidly from April to May. The
demise of the P. parvum bloom began prior to rain event at the beginning of May (Supplemental
Fig. 1). The P. parvum bloom coincided with a relatively modest (and highly variable) increase
14

in chlorophyll concentrations during the winter in Lebanon Pool (Fig. 2F). P. parvum cells were
detected at Wilson Creek during the same period of time, but never exceeded 800 cells/ml (250x
lower than the maximum observed in Lebanon Pool). The haptophyte was undetected by routine
microscopical examination at both locations from May until the end of the study (Fig. 2F).
Total nutrient concentrations in the lake were high. Total nitrogen (Fig. 2G dotted lines)
ranged from 75 to 100 µM in Lebanon Pool, and from 75 to over 150 µM in Wilson Creek. Total
nitrogen values were larger in Wilson Creek (100- 150 µM) compared to Lebanon Pool
(approximately 75 µM) in March and April and from August to October. Total phosphorous
(Fig. 2H dotted lines) ranged from 4 to 10 µM in Lebanon Pool, and from 4 to 15 µM in Wilson
Creek. Total phosphorous values were greater in Wilson Creek (approximately 2x) compared to
Lebanon Pool for March, April, and September. The ratios of total nitrogen to total phosphorous
(Fig. 2I) ranged from 10 to 25 with averages for the year of 18.5 in Lebanon Pool and 16 in
Wilson Creek. N:P ratios were higher in Lebanon Pool (approximately 1.5x) compared to Wilson
Creek from February to April and in July.

Genetic signatures of Prymnesium cells
The bloom of P. parvum in Lebanon Pool that was documented with microscopical counts (Fig.
2F) was also apparent in the 18s and 16s ribosomal sequence data. The 18s rDNA sequences
from Lebanon Pool identified as Prymnesium (Fig. 3A) (see Methods) comprised 30-50% of the
total number of sequences for the months of January, February, March and April. Tags identified
as Prymnesium in Wilson Creek during the same time period were detected but constituted no
more than 2% of the total number of sequences in those samples, consistent with the low number
of cells detected by microscopy (Fig. 2F insert). 16s rDNA sequences identified as Prymnesium
15

plastids (Fig. 3B) (see Methods) comprised approximately 20% of the total number of sequences
for the months of January, February, March and April in Lebanon Pool, whereas those sequences
were detected in Wilson Creek but remained at low abundances (approximately 2% of total
sequences). 18s rDNA sequences attributed to Prymnesium were detected at low relative
abundances at both sampling locations from June to October (Fig. 3A inset) when the alga was
not detected by microscopy (Fig. 2F). The timing and duration of the high abundances of 18s and
16s rDNA sequences attributed to Prymnesium were consistent with the patterns of cell
abundances of P. parvum obtained by microscopical examination for the bloom months (Fig.
2F). Further, each location had excellent spearman correlations [0.90 and 0.89 in Lebanon Pool
(p-values <<0.001) and 0.85 and 0.83 in Wilson Creek (p-values <<0.001)] between relative
abundances of sequence tags and cell counts.

Bacterial and eukaryotic operational taxonomic units:
OTU calling at 97% similarity was performed on the subsampled 18s and 16s (plastids removed)
datasets (see Methods). The grand total of OTUs at both sites and across all months was 9,770
for the 18s and 8,392 for the 16s sequences. Singletons (OTUs represented by a single sequence)
comprised 59% (5753) of the eukaryotic 18s OTUs, but these sequences represented only 3% of
the total number of sequences (Table 1). Singletons in the 16s dataset comprised 64% (5401) of
the total number of OTUs but represented only 6% of the total number of sequences (Table 1).
The number of sequences per library was normalized within the 18s and 16s datasets,
however, the sampling effort for the 18s dataset (7,678 sequences) was over twice that of the 16s
dataset (3,597 sequences). Despite this difference in the number of tags collected, the numbers
of detected 18s and 16s OTUs were similar (Table 1). Rarefaction curves constructed for the
16

three 16s libraries from December (Supplemental Fig. 3B) were overall steeper than those built
for the three 18s libraries (Supplemental Fig. 3A), implying higher diversity among the bacterial
communities. Our normalized datasets of approximately 7,500 and 3,500 yielded total numbers
of sequences that were not in the steepest portion of the rarefaction curves. Neither sets of curves
(18s nor 16s) approached a maximum, implying we had yet to reach sampling saturation in the
samples (Supplemental Fig. 3).
The numbers of observed OTUs per month during the winter months of January to April
were lower than the numbers observed during the rest of the year in both datasets (18s and 16s)
at both sampling sites (Table 1; Fig. 4A and B). Seasonal differences in the numbers of
eukaryotic OTUs were greater than seasonal differences in the numbers of bacterial OTUs. The
number of 18s OTUs observed each month declined from November to January, and remained
low (250-327 per sample for Lebanon Pool and 425-539 per sample for Wilson Creek) through
April. The numbers of eukaryotic OTUs detected in Lebanon Pool during the P. parvum bloom
(Fig. 2F), (January to March) were 1/3 the number of eukaryotic OTUs detected in Wilson Creek
during the same period. Both sites experienced marked increases in the number of observed
OTUs during May, reached maxima during August (~1200) and declined steadily through the
fall. The number of eukaryotic OTUs observed each month in both Lebanon Pool and Wilson
Creek showed a yearly seasonal pattern, with good agreement to a sine wave function (r
2
= 0.8)
with a set period of 365 days (Fig. 4A and (Supplemental Fig. 4A and C). The numbers of 16s
OTUs per month in both locations were lowest from January to April, but otherwise the values
remained relatively high (approximately 750) and constant (Table 1; Fig. 4B). The lowest
number of OTUs observed in each site occurred in Lebanon Pool during February (400) and
Wilson Creek during January (394). The largest month-to-month increases in the numbers of
17

bacterial OTUs occurred from April to May at both study sites, similar to trends for the 18s data.
The largest numbers of OTUs per sample were observed in Lebanon Pool during May (871) and
Wilson Creek during June (911) (Table 1; Fig. 4B).
In general, the dominant eukaryotic OTU in each sample constituted a greater percentage
of the total number of sequences in that sample (median of 12.5%) relative to the contribution of
sequences of the dominant bacterial OTU in each sample (median of 6%) (Table 1). The
dominant 18s OTU in Lebanon Pool for the four months during the P. parvum bloom was
identified as a Prymnesiales. Sequences in the Prymnesiales OTU constituted 29-49% of the
total sequences for those four months (Table 1). These percentages are similar to the percentages
of the individual 18s sequence reads identified as Prymnesium (Fig. 3A). This correspondence
indicates that the sequence reads identified as Prymnesiales condensed at 97% similarity into a
single unique OTU. This degree of dominance of a single OTU was not observed at any other
time in the study. The dominant bacterial OTU during the P. parvum bloom in Lebanon Pool was
a freshwater Synechococcus in 3 of the 4 months (Table 1).
True replicates (two separate samples from Wilson Creek during December obtained on a
single sampling date, and processed in parallel) were examined to establish the reproducibility of
the sampling, processing and sequencing procedures. These two samples yielded similar values
for the total number of OTUs, singletons and taxonomic assignment of OTUs for the 18s dataset
and for the numbers of non-singleton OTUs and taxonomic assignment of OTUs for the 16s
dataset (Table 1).

18

Diversity of 18s and 16s OTUs
Annual trends in the non parametric Chao 1 (Fig. 4C) and parametric Catchall (Fig. 4E) species
richness estimates for the 18s rDNA data were similar for the two sampling locations (mostly
overlapping confidence intervals). The richness estimators were lowest (<750 and <2500 for
Chao I and Catchall, respectively) during the winter-spring months of January to April and
highest (~3000 and ~7000, respectively) during the summer months of June and July (Fig. 4C
and E). These changes reflected the trends in the observed 18s OTUs (Fig. 4A), and the two
richness estimators had excellent correlations (spearman correlations > 0.93 (p-values <0.001)
with the observed 18s richness at both locations. The strong dominance of the Prymnesiales
OTU in Lebanon Pool during the months of January to April (Table 1) did not cause a noticeable
difference between the richness estimates obtained at the bloom (Lebanon Pool) and non-bloom
(Wilson Creek) sites (Fig. 4C and E).
The annual patterns for the Simpson Index of Diversity for the eukaryotic community
(Fig. 4E) was markedly different between the two locations for part of the year. This index was
significantly lower (≈0.7-0.8) for Lebanon Pool samples collected during the months of January
to April compared to values obtained for the rest of the year in Lebanon Pool and also compared
to all of the values obtained for Wilson Creek (values >0.9). The timing of the low values
coincided with the P. parvum bloom in Lebanon Pool (Fig. 2F). The index was relatively
constant throughout the year for Wilson Creek samples (average = 0.96) (Fig. 4E). When
sequences attributed to the Prymnesiales OTU were removed from the dataset, the sequences
subsampled again to normalize the number of sequences in each sample, and the Simpson index
recalculated (Fig. 4E dotted line), the values were still low in 3 (February, March and April) of
the 4 months in which the haptophyte bloomed (Fig. 2F, 3A and B).
19

The Chao and Catchall values for bacterial richness (Fig. 4D and F) indicated a trend
towards higher values during the summer at both sampling locations, but these differences were
not significant because of the high variances. Species richness in Lebanon Pool displayed the
lowest values (~750 and <2000 for Chao and Catchall, respectively) during January and
February and highest values during May or June (~2250 and ~7000, respectively; Fig. 4D and F
black lines). Values of estimated species richness for Wilson Creek fluctuated from
approximately 800 to 2250 (Chao index; Fig. 4D) and from 3000 to 6000 (Catchall index; Fig.
4F) throughout the year. The Chao estimator (Fig. 4D) showed good correlations (spearman
values of 0.75 and p-values <0.01) to the observed OTU richness at both locations (Fig. 4B). The
catchall richness estimators did not correlate well to the observed 16s richness.
Values of the Simpson Index determined for the bacterial assemblages (Fig. 4F) were
remarkably similar, constant and high (0.98) at both locations throughout the year-long study.
The one exception to this generality was the February sample in Lebanon Pool, which was
characterized by a single, abundant bacterial OTU (Flavobacteria) that constituted 24% of all
sequences (Table 1).

OTU dominance in 18s and 16s datasets
Cumulative dominance plots for 18s and 16s datasets were constructed for the two sampling sites
throughout the annual cycle (Fig. 5, and data not shown). Dominance plots of the 18s OTUs
(Fig. 5A) were clearly distinct between the Lebanon Pool and Wilson Creek samples for the
months of January through April, the time of the P. parvum bloom in Lebanon Pool (Fig. 2F).
Eleven or fewer taxa comprised 80% of the total number of the 18s sequences in the Lebanon
Pool samples during these months, compared with 25-40 taxa in Wilson Creek for the same
20

period. This site-specific difference during these months was statistically supported by the
ANOSIM test (R test statistic = 0.92, and p-value = 0.03). Differences between locations were
not found before or after the bloom (December, May and data not shown). The curves for the 16s
data (Fig. 5B) were overall less steep compared to the 18s data and the two sites were not
statistically different from each other. Only one sample (February, Lebanon Pool) had a
distinctive cumulative dominance plot among the 16s assemblages. That sample was the one
sample strongly dominated by a single bacterial OTU (Flavobacteria) that constituted 24% of all
16s sequences (Table 1).

Seasonal and site-specific patterns in microbial community similarities
Community similarities built from Bray-Curtis and Jaccard indices for the eukaryotic and
bacterial assemblages were examined across all months, and between sampling sites. The Bray-
Curtis and Jaccard similarity values for the eukaryotic and bacterial assemblages revealed an
annual cyclical pattern of change over the course of the year at both sites, as well site-specific
differences during a portion of the year. The average Bray-Curtis similarity across all monthly
samples at each location was low (<8% for the 18s data and < 18% for the 16s data). The global
analysis of variance test (ANOSIM) confirmed differences between the months: (R test statistics
of 0.53 and p-value = 0.002 for the 18s data, and 0.79 and p-value = 0.001 for the 16s data). The
average Jaccard similarity across the year at each location was also low (<9% for the 18s data,
and <9% for the 16s data). The ANOSIM test confirmed that there were differences between the
months: (R test statistics of 0.73 and p-value = 0.001 for the 18s data, and R test statistic of 0.75
and p-value = 0.001 for the 16s data). The eukaryotic or bacterial assemblages between the two
locations (Lebanon Pool and Wilson Creek) could not be distinguished with a global ANOSIM
21

test when all months were considered together. However, when the samples were grouped based
on months with high or low cell abundances of P. parvum (see Methods), there was a strong and
significant difference observed between the two locations [R test statistics of 0.98 (Bray-Curtis)
and 1 (Jaccard) for the 18s data, and R test statistics of 0.80 (Bray-Curtis) and 0.88 (Jaccard) for
the 16s data; (p-values < 0.03)]. In both the 18s and 16s datasets we did not detect a difference
between the locations for the months of November, December and May to October.
The two dimensional MDS representations of the eukaryotic and bacterial Bray-Curtis
(Fig. 6A and B) or the Jaccard similarity matrices (Fig. 6C and D) each had low stress values
≤0.10 and indicated that these communities experienced a similar seasonal progression
throughout the year. The MDS representations also reflected the hierarchical cluster diagrams
built from group average scores for the 18s and 16s Bray-Curtis and Jaccard similarity matrices
(Supplemental Fig. 5A-D). The strong site-specific differences detected with the ANOSIM test
for the months of January to April (the period of the P. parvum bloom in Lebanon Pool) were
conspicuous on the MDS plots. The spatial separation was dramatic for the 18s data but was also
clearly represented in the 16s data. These separations were apparent when relative abundance
(Bray-Curtis; Fig. 6A and C) or presence/absence (Jaccard; Fig. 6B and D) were used to
calculate the similarities. The 18s and 16s MDS arrangements of the samples from Lebanon
Pool and Wilson Creek for the month of May (the time of the rain event) were similar and also
spatially distinct from all other samples. The eukaryotic or bacterial communities at the two
locations remained closely grouped from July through October and returned to the proximity of
the November samples in the previous year (i.e. the beginning of the study period; Fig. 6A-D). A
statistical test (see Methods) for annual cyclicity of the 18s and 16s datasets at both Lebanon
Pool and Wilson Creek showed good Spearman correlations to a model cyclical matrix (0.52 to
22

0.70 with p-values = 0.001). The specific Spearman correlations to the model matrix for each
location and each measure were: 18s Bray-Curtis, 0.53 and 0.54; 18s Jaccard, 0.62 and 0.59; 16s
Bray-Curtis, 0.64 and 0.52; and 18s Jaccard, 0.70 and 0.62). The similarity matrices built from
the Bray-Curtis relative abundance measures and the Jaccard presence/absence measures were
remarkably similar (Spearman correlations of >0.9) within the 18s and 16s datasets.
The placement of the eukaryotic and bacterial samples in the MDS plots were spatially
similar for the Bray-Curtis (Fig. 6A vs. C) and Jaccard indices (Fig. 6B vs. D). This similarity
was statistically supported with strong Spearman correlations between the 18s and 16s Bray-
Curtis similarity matrices (0.73 p-value = 0.001) and between the 18s and 16s Jaccard similarity
matrices (0.87 p-value = 0.001).
The replicability of the 18s and 16s sequence datasets was high. The duplicate samples
collected from Wilson Creek during December and processed in parallel were highly similar and
not statistically different on the basis of the SIMPROF test. Bray-Curtis similarity values for the
two duplicate samples were 85% for the 18s and 75% for the 16s datasets. Jaccard values for the
two duplicate samples were 30% for the 18s and 23% for the 16s datasets (Supplemental Fig. 5).
The two duplicate samples occupied close physical space on the MDS plots (Fig. 6A-D). The
duplicate samples also provided an experimental maximum of the Bray-Curtis and Jaccard
similarity measures. No two 18s samples were more similar than the values for the duplicate
December samples of 85% or 30% (Supplemental Fig. 5A and B), no two 16s Bray-Curtis values
were more similar than 75% (Supplemental Fig. 5 C), and only 4 pairs of Jaccard values were
equal to or greater than 23% similar (Supplemental Fig. 5D). Complete hierarchical cluster
diagrams based on group average scores for the Bray-Curtis and Jaccard similarities for the 18s
and 16s datasets are presented in Supplemental Fig. 5A-D).
23

The two BEST analyses returned the following sets of environmental variables that best
correlated with the detected patterns in 18s or 16s community similarity for the entire dataset
(both locations). The 18s communities were best matched (P of 0.844, p-value = 0.01) with
temperature, salinity, and P. parvum cell abundances. The 16s community patterns were best
correlated (P of 0.738, p-value = 0.01) with temperature, salinity, dissolved oxygen, pH, and P.
parvum cells. The 18s communities were best correlated (P of 0.663, p-value = 0.01) with
temperature, salinity, and pH when the biological variables were removed, and the 16s
community patterns were best correlated (P of 0.661, p-value 0.01) with temperature, salinity,
and pH when the biological variables were removed.

Discussion
High-throughput sequencing of ribosomal DNA fragments was applied in this study to
characterize changes in the eukaryotic and bacterial assemblages of a freshwater ecosystem over
an annual cycle. Pyrosequencing has been increasingly employed to examine species richness
and the relative abundances of microbial taxa in natural samples from a wide variety of
ecosystems (Amaral-Zettler et al., 2009; Behnke et al., 2011; Edgcomb et al., 2011; Gilbert et
al., 2009). We applied a high level of quality control and normalization to our sequence data in
an effort to reduce potential artifactual results or conclusions. Towards this end, we obtained
good replication of our sequencing approach, as well as good agreement between traditional and
molecular approaches, as evidenced by the behavior of true duplicates collected in December
(similar numbers of OTUs, estimates of diversity and community relatedness for 18s and 16s
datasets), and good agreement between absolute abundances of P. parvum by microscopical
counts and relative abundances of sequence tags attributed to the alga (Fig. 2F, Fig. 3A and B,
24

Tables 1 and 2). Moreover, multiple analyses at various levels of stringency (i.e. subsampling
and transformations, and the comparison of results using relative abundances of tags vs.
presence/absence) indicated that the relationships observed within our dataset were highly robust
(Fig. 6 and Supplemental Fig. 5). Our approach cannot address some inherent caveats regarding
the use of sequence information for ecological investigations (e.g. variations in gene copy
number, PCR bias, etc.) (Kunin et al., 2010; Medinger et al., 2010) but pyrosequencing greatly
improved our ability to detect taxa present at low abundances in samples, and thus our ability to
characterize a greater percentage of the microbial community of Lake Texoma.
Changes in the eukaryotic and bacterial diversity (species richness and evenness),
community membership and structure observed over an annual cycle at two locations within
Lake Texoma reflected distinct and clearly identifiable responses to (1) seasonal environmental
forcing factors, (2) a shared episodic physical disturbance (a major rain event) and (3) a
prolonged and localized ecosystem disruptive harmful algal bloom (of the toxic alga, P. parvum).
The latter two responses were overlaid on what appeared to be an annual cyclicity of
environmental forcing that acted to ‘reset’ the microbial community structure close to the same
starting point at the beginning and end of the year-long study. Overall, our results were
consistent with the hypothesis that environmental perturbations (physical, chemical and
biological forcing factors) result in significant and rapid reassembly of the microbial community
(Caron and Countway, 2009) to maintain biological community structure.

Seasonal and episodic features of Lake Texoma
The trends and absolute values of temperature, salinity, nutrients, P. parvum cell counts and
chlorophyll concentrations observed throughout this year-long study period (November 2008 to
25

October 2009) were representative of these parameters at the two littoral study sites and from
other localities within the lake (Hambright et al., 2010). Seasonal trends in surface water
temperature (Fig. 2A and Supplemental Fig. 2B and C) and day length (Supplemental Fig. 2A)
over the year at both locations showed excellent fit to a yearly sinusoidal regression (period of
365 days), and together showed the strong overall environmental seasonality of the system.
Spring rain events are stochastic in nature and have, in the past, resulted in large-scale and rapid
replacement of lake water in its littoral environments (Hambright et al., 2010). Both sampling
locations experienced a dramatic rain event in this study (Supplemental Fig. 1) and resulted in
related perturbations including decreases in salinity, pH and dissolved oxygen (Fig. 2 B-D) in the
beginning of May. Localized blooms of P. parvum in Lebanon Pool have become annually
repeating biological features of Lake Texoma since 2004, while Wilson Creek has not typically
experienced blooms (Hambright et al., 2010; Zamor et al., 2012). We documented one of the
largest blooms to date of P. parvum (>10
5
cells/ml) in Lebanon Pool that lasted from January to
April. P. parvum cells were present in Wilson Creek during this study but abundances were
generally 100-fold lower than in Lebanon Pool (Fig 2F and Fig. 3A and B).
Dissolved nitrogen and phosphorous concentrations (Fig. 2 G and H solid symbols) were
similar at both locations and fluctuations in this constituents showed no obvious relationship to
seasonality or disturbance. Total nitrogen and phosphorous values were high and qualify the lake
for eutrophic status (Wetzel, 2001). The ratios of total N:P (Fig. 2I) for the year were, on
average, close to Redfield (Redfield, 1958). The differences between the two locations could
have been due to differences in the biological community composition, but most likely reflected
localized differences in the source of particulate and total nutrient concentrations to the two
locations (Fig. 2G and H open symbols). The nutrient parameters did not correlate with the P.
26

parvum bloom nor did they describe patterns of diversity (richness, evenness, or community
structure and similarity).
Changes in chlorophyll concentrations (a proxy for algal standing stock; Fig. 2E) at both
locations broadly indicated the responses of the microbial communities to the three levels of
environmental change in the lake. Both locations reflected seasonal influences (minor increases
in early spring and massive increases in early fall), and short-term, event-driven responses (rapid
decreases and low values that coincided with the rain event in the beginning of May). The P.
parvum bloom in Lebanon pool resulted in a moderate, prolonged increase in chlorophyll that
occurred over the time span of the bloom and contributed to the slight differences in chlorophyll
concentrations between the two locations.
Together these features (temperature, salinity, pH, dissolved oxygen, nutrients, P. parvum
cell abundances and chlorophyll concentrations) set the environmental context for exploring the
combined importance of gradual seasonal forcing and episodic events (massive rain events and a
harmful algal bloom) in shaping microbial communities.

Response of the microbial community to seasonal environmental forcing
The eukaryotic assemblages in the monthly samples collected in Lake Texoma exhibited clearly
discernible, yearly (periods of 365 days) sinusoidal changes in the number of observed and
predicted OTUs with highs in the summer and lows in the winter (Fig. 4A,C, and E). Overall,
the annual patterns were similar for both locations and suggest that eukaryotic taxa richness
responds directly to seasonally fluctuating environmental parameters such as day length or water
temperature (Sommer et al., 1986). The annual patterns in the observed numbers of bacterial
OTUs were similar at the two locations and the magnitude and cyclicity of their seasonal
27

fluctuations were less dramatic and evident compared to the eukaryotic richness (Fig. 4A vs B, C
vs D and E vs F). This suggests that bacterial taxonomic richness as measured in this study, was
less sensitive to seasonal forcing factors and responded more to localized environmental inputs
or processes.
The microbial communities in the lake maintained relatively high species evenness
throughout the study, despite fluctuations in taxonomic richness and seasonal changes in the
chemical/physical environment, with one exception the P. parvum bloom in Lebanon Pool. The
Simpson index of diversity reflects species richness, but is strongly affected by the relative
abundances of taxa within an assemblage. Contrary to results for the species richness indices
(Chao, Catchall), community evenness estimates for the eukaryotic and bacterial assemblages
showed very little response to seasonal environmental changes (Fig. 4 E and F). Moreover,
values of the Simpson index were relatively high and invariant for the Wilson Creek samples
throughout the year, and showed significant decreases for the microbial assemblage in Lebanon
Pool only during the P. parvum bloom, and then only for one of those months for the bacteria.
Examination of community similarities among the monthly samples using the Bray-
Curtis and Jaccard indices demonstrated significant annual cycles for the eukaryotic and bacterial
assemblages at both sampling sites. The yearly trajectory of the months in the MDS plots
indicated a statistically supported cyclical seasonal progression, and a return of community
structure to the approximate starting point at each location at the end of the year-long study (high
Spearman correlation coefficients; Fig. 6 A-D). Similar patterns were detected with both the
Bray-Curtis (which takes into account shared taxa and relative abundances of OTUs), and the
Jaccard (which considers presence/absence and thus common and rare taxa are given equal
weight) indicating that both the community membership patterns and patterns in community
28

structure (the catalogue of taxa present) were affected in parallel by seasonal environmental
forcing factors. The conserved pattern of cyclical change implies an active seasonal response
among the less dominant taxa in the microbial communities.

Microbial community response to an ecosystem disruptive harmful algal bloom
The P. parvum bloom that occurred in Lebanon Pool and not Wilson Creek created the
conditions for a natural experiment in which to explore the effects of an ecosystem disruptive
harmful algal bloom on eukaryotic and bacterial microbial community structure. The overall
effect of this harmful alga on food web structure derives from the direct effects of toxins
produced by this species on a wide range of organisms (Skovgaard and Hansen, 2003), as well as
its well-characterized phagotrophic ability (Tillmann, 1998). Toxins produced by P. parvum
have been shown to negatively affect phototrophic and heterotrophic species of the microbial
plankton, and its phagotrophic capabilities allow the ingestion of taxa ranging in size from
bacteria to zooplankton that can be significantly larger than the alga (Tillmann, 2003).
The P. parvum bloom in Lebanon Pool had a surprisingly limited impact on microbial
taxonomic richness. The number of observed eukaryotic OTUs per month was lower over the
time span of the bloom in Lebanon Pool compared to Wilson Creek (Fig. 4A), but similar
estimates of total eukaryotic species richness were observed at both sites throughout the bloom
using the Catchall and Chao richness estimators (Fig 4C and E). This result presumably indicates
that features of the eukaryotic assemblage structure on which these species richness estimators
are based were similar at both study sites during the P. parvum bloom in Lebanon Pool. Negative
effects were not apparent on the observed number of bacterial OTUs or total bacterial species
richness estimates as a consequence of the P. parvum bloom (Fig. 4B and D). Despite the
29

pronounced dominance of a single taxon during the bloom, a diverse suite of eukaryotic and
bacterial microbes remained present as subdominant or rare taxa.
The P. parvum disturbance event had a strong negative effect on the taxonomic structure
(depressed evenness and altered distributional trends) of the eukaryotic assemblage.
The Simpson Index of Diversity was markedly decreased in Lebanon Pool from January to April
(Fig. 4E) (the timing and location of the P. parvum bloom; Fig 2C and 3A and B). The low
evenness was not solely due to the over-dominance of P. parvum because low values were also
obtained when P. parvum sequences were removed from the dataset and the dataset was re-
analyzed (dotted line in Fig. 4E). Low Simpson values derived using the latter dataset were
relegated to the end of the P. parvum bloom, implying that community structure was strongly
skewed towards fewer, highly dominant species and provided evidence that the effects of bloom
were radiating into the eukaryotic community selecting for physiologies that were tolerant to, or
benefited from, the presence of the toxic haptophyte. The cumulative dominance plots mirror the
trends in the disrupted taxonomic evenness. The steep and distinct slopes of the four bloom
months samples reflect altered structure compared to the non-bloom months at Lebanon Pool and
all months (including the bloom months) at Wilson Creek (Fig. 5A). Bacterial taxonomic
evenness and distributions were not affected by the bloom (Fig. 4F and Fig. 5B).
Analysis of community similarities also indicated a strong impact of the P. parvum
bloom on both the eukaryotic and bacterial assemblages during the study (Fig. 6). The patterns in
the annual trajectories of community similarities were highly consistent between the bacterial
and the eukaryotic assemblages, including the divergence between Lebanon Pool and Wilson
Creek during the P. parvum bloom. This statistically supported separation (see Results) was
observed in both the Bray-Curtis and Jaccard plots, and suggests that the effects of the bloom on
30

the microbial communities were far-reaching, affecting the most abundant as well as rare
eukaryotic and bacterial taxa (Fig. 6A-D). Therefore, while bacterial species richness and
evenness were minimally affected by the P. parvum bloom, the composition of the bacterial
assemblage changed in a manner that paralleled changes in the eukaryotic assemblage.

Episodic physical disturbance
The major rain event in the beginning of May (Supplemental Fig. 1) and resulting decreases in
salinity, pH and dissolved oxygen (Fig. 2 B-D) led to dramatic changes in the eukaryotic and
bacterial assemblages at both study sites. Community compositional changes were apparent as
dramatic shifts in the May samples on the MDS plots of community similarity. The samples
from May plotted closely together for the two sampling sites, but apart from all other samples,
for both the bacterial and eukaryotic assemblages (i.e. low similarity to all other samples; Fig. 6).
The rain event also brought an end to differences in microbial community similarities
between the two sampling locations caused by the P. parvum bloom in Lebanon Pool, and thus
acted to ‘reset’ the eukaryotic and bacterial communities at these locales back to a high level of
similarity (albeit with community composition that was different from any other month). It
would be anticipated that the rain and runoff and consequent changes in chemical/physical
parameters would introduce and/or select for a unique suites of microbial taxa. Indeed, the
largest increase in observed eukaryotic and bacterial OTUs between months at both locations
occurred between April (pre-rain event) and May (post-rain event; Fig 4A and B).

31

Conclusions
The microbial community at two sampling sites in Lake Texoma experienced two episodic
disturbance events overlaid on seasonal environmental forcing. Both the eukaryotic and the
bacterial assemblages responded significantly to these environmental influences via changes in
community structure (species richness and dominance) and taxonomic composition
(demonstrated through community similarity analyses). Eukaryotic taxonomic richness
demonstrated clear seasonal patterns of change, with highest richness in the summer and lowest
richness in the winter, however, the bacterial taxa did not show such distinct patterns. Seasonal
effects were evident as a cyclical, annual pattern in microbial community similarity for the
eukaryotic and bacterial assemblages at both locations in most months, implying shared or
coordinated responses to seasonal environmental forcing factors at both locations. A bloom of
the toxic alga, Prymnesium parvum, in Lebanon Pool (but not Wilson Creek) during January-
April resulted in strongly divergent microbial community structure for the two sites during those
months. Composition of the eukaryote assemblage was strongly and directly affected by the
presence of the P. parvum bloom. Divergence of the bacterial assemblages at the two sites also
occurred in response to the bloom, presumably indicating the close relationships and
dependencies between the eukaryotic and bacterial assemblages. Recovery of the microbial
assemblages from the divergent state caused by the bloom was facilitated by a physical
perturbation in the form of a major rain event that reestablished highly similar microbial
communities and common seasonal trajectories at both sampling sites. Community composition
at the two sites changed during subsequent months but remained highly similar to each other
(Fig. 6A-D) eventually attaining community compositions that were similar to starting conditions
observed in the previous fall.
32

References
Amaral-Zettler, L.A., Gomez, F., Zettler, E., Keenan, B.G., Amils, R., and Sogin, M.L. (2002)
Microbiology: eukaryotic diversity in Spain's River of Fire. Nature 417: 137-137.
Amaral-Zettler, L.A., McCliment, E.A., Ducklow, H.W., and Huse, S.M. (2009) A method for
studying protistan diversity using massively parallel sequencing of V9 hypervariable
regions of small-subunit ribosomal RNA genes. PLoS One 4: e6372.
Behnke, A., Engel, M., Christen, R., Nebel, M., Klein, R.R., and Stoeck, T. (2011) Depicting
more accurate pictures of protistan community complexity using pyrosequencing of
hypervariable SSU rRNA gene regions. Environ Microbiol 13: 340-349.
Bray, J., and Curtis, J. (1957) An ordination of the upland forest communities of southern
Wisconsin. Ecol Monogr 27: 325-349.
Bunge, J. (2011). Estimating the number of species with CatchAll. Paper presented at:
Proceedings of Pacific Symposium on Biocomputing (Kohala Coast, Hawaii).
Buskey, E.J., Liu, H., Collumb, C., Guilherme, J., and Bersano, J.G.F. (2001) The decline and
recovery of a persistent Texas brown tide algal bloom in the Laguna Madre (Texas,
USA). Estuaries 24: 337-346.
Caraco, N., Cole JJ, Raymond PA, Strayer DL, Pace ML, Findlay SEG, and DT, F. (1997) Zebra
mussel invasion in a large, turbid river: phytoplankton response to increased grazing
ecology. Ecology 78: 588-602.
Caron, D.A., and Countway, P.D. (2009) Hypotheses on the role of the protistan rare biosphere
in a changing world. Aquat Microb Ecol 57: 227-238.
Chao, A. (2005). Species estimation and applications, Vol 12 (New York: Wiley).
33

Chao, A., and Lee, S.M. (1992) Estimating the number of classes via sample coverage. Journal
of the American Statistical Association 87: 210-217.
Clarke, K.R. (1990) Comparisons of dominance curves. J Exp Mar Biol Ecol 138: 143-157.
Clarke, K.R. (1993) Non-parametric multivariate analyses of changes in community structure.
Australian Journal of Ecology 18: 117–143.
Clarke, K.R., and , and Warwick, R.M. (2001). Change in Marine Communities: An Approach to
Statistical Analysis and Interpretation, 2nd edn. edn (Plymouth, UK).
Clarke, K.R., Somerfield, P.J., and Chapman, M.G. (2006) On resemblance measures for
ecological studies, including taxonomic dissimilarities and a zero-adjusted Bray–
Curtis coefficient for denuded assemblages. J Exp Mar Biol Ecol 330: 55-80.
Clarke, K.R., Somerfield, P.J., and Gorley, R.N. (2008) Testing of null hypotheses in exploratory
community analyses: similarity profiles and biota-environment linkage. J Exp Mar
Biol Ecol 366: 56-69.
Countway, P.D., Gast, R.J., Dennett, M.R., Savai, P., Rose, J.M., and Caron, D.A. (2007)
Distinct protistan assemblages characterize the euphotic zone and deep sea (2500 m)
of the western N. Atlantic (Sargasso Sea and Gulf Stream). Environ Microbiol 9:
1219-1232.
Edgcomb, V., Orsi, W., Bunge, J., Jeon, S., Christen, R., Leslin, C., Holder, M., Taylor, G.T.,
Suarez, P., Varela, R., et al. (2011) Protistan microbial observatory in the Cariaco
Basin, Caribbean. I. Pyrosequencing vs Sanger insights into species richness. ISME J
5: 1344-1356.
34

Edgcomb, V.P., Kysela, D.T., Teske, A., de Vera Gomez, A., and Sogin, M.L. (2002) Benthic
eukaryotic diversity in the Guaymas Basin hydrothermal vent environment. PNAS
USA 99: 7658-7662.
Estes, J.A., Terborgh, J., Brashares, J.S., Power, M.E., Berger, J., Bond, W.J., Carpenter, S.R.,
Essington, T.E., Holt, R.D., Jackson, J.B.C., et al. (2011) Trophic downgrading of
planet Earth. Science 333: 301-306.
Evardsen, B., and Imai, I. (2006). The ecology of harmful flagellates within Prymnesiophyceae
and Raphidophyceae. In Ecology of Harmful Algae, E. Granéli, and J. Turner, eds.
(Netherlands: Springer-Verlag), pp. 67-79.
Fistarol, G.C., Legrand C., and E, G. (2003) Allelopathic effect of Prymnesium parvum on a
natural plankton community. Mar Ecol Prog Ser 255: 115–125.
Fuhrman, J.A. (2009) Microbial community structure and its functional implications. Nature
459: 193-199.
Gihring, T.M., Green, S.J., and Schadt, C.W. (2012) Massively parallel rRNA gene sequencing
exacerbates the potential for biased community diversity comparisons due to variable
library sizes. Environ Microbiol 14: 285-290.
Gilbert, J.A., Field, D., Swift, P., Newbold, L., Oliver, A., Smyth, T., Somerfield, P.J., Huse, S.,
and Joint, I. (2009) The seasonal structure of microbial communities in the Western
English Channel. Environ Microbiol 11: 3132-3139.
Gobler, C.J., Lonsdale DJ, and GL, B. (2005) A review of the causes, effects, and potential
management of harmful brown tide blooms caused by Aureococcus anophagefferens
(Hargraves et Sieburth). Estuaries 28: 726-749.
35

Gobler, C.J., and Sunda, W.G. (2012) Ecosystem disruptive algal blooms of the brown tide
species, Aureococcus anophagefferens and Aureoumbra lagunensis. Harmful Algae
14: 36-45.
Hambright, K.D., Zamor, R.M., Easton, J.D., Glenn, K.L., Remmel, E.J., and Easton, A.C.
(2010) Temporal and spatial variability of an invasive toxigenic protist in a North
American subtropical reservoir. Harmful Algae 9: 568-577.
Heino, J., and Soininen, J. (2010) Are common species sufficient in describing turnover in
aquatic metacommunities along environmental and spatial gradients? Limnol
Oceanogr 55: 2397-2402.
Huse, S.M., Huber, J.A., Morrison, H.G., Sogin, M.L., and Welch, D.M. (2007) Accuracy and
quality of massively parallel DNA pyrosequencing. Genome Biol 8: R143.
Huse, S.M., Welch, D.M., Morrison, H.G., and Sogin, M.L. (2010) Ironing out the wrinkles in
the rare biosphere through improved OTU clustering. Environ Microbiol 12: 1889-
1898.
Jaccard, P. (1908) Nouvelles recherches sur la distribution florale. Bulletin de la Société
Vaudoise des Sciences Naturelles 44: 223-270.
Jones, S.E., C., C., Kratz, T.K., Wu, J.T., Shadeand, A., and and McMahon, K.D. (2008)
Typhoons initiate predictable change in aquatic bacterial communities. Limnol
Oceanogr 53: 1319-1326.
Kim, D.Y., Countway, P.D., Gast, R.J., and Caron, D.A. (2011) Rapid shifts in the structure and
composition of a protistan assemblage during bottle incubations affect estimates of
total protistan species richness. Microb Ecol 62: 383-398.
36

Kunin, V., Engelbrektson, A., Ochman, H., and Hugenholtz, P. (2010) Wrinkles in the rare
biosphere: pyrosequencing errors can lead to artificial inflation of diversity estimates.
Environ Microbiol 12: 118-123.
Lambshead, P.J.D., Platt, H.M., and Shaw, K.M. (1983) The detection of differences among
assemblages of marine benthic species based on an assessment of dominance and
diversity. J Nat His 17: 859-874.
Lopez-Garcia, P., Philippe, H., Gail, F., and Moreira, D. (2003) Autochthonous eukaryotic
diversity in hydrothermal sediment and experimental microcolonizers at the Mid-
Atlantic Ridge. PNAS USA 100: 697-702.
Margulies, M., Egholm, M., Altman, W.E., Attiya, S., Bader, J.S., Bemben, L.A., Berka, J.,
Braverman, M.S., Chen, Y.J., Chen, Z., et al. (2005) Genome sequencing in
microfabricated high-density picolitre reactors. Nature 437: 376-380.
McPherson, R.A., Fiebrich, C.A., Crawford, K.C., Elliott, R.L., Kilby, J.R., Grimsley, D.L.,
Martinez, J.E., Basara, J.B., Illston, B.G., and Morris, D.A. (2007) Statewide
monitoring of the mesoscale environment: A technical update on the Oklahoma
Mesonet. J Atmos Ocean Tech 24: 301-321.
Medinger, R., Nolte, V., Pandey, R.V., Jost, S., Ottenwalder, B., Schlotterer, C., and Boenigk, J.
(2010) Diversity in a hidden world: potential and limitation of next-generation
sequencing for surveys of molecular diversity of eukaryotic microorganisms.
Molecular ecology 19 Suppl 1: 32-40.
Michaloudi, E., Moustaka-Gouni, M., Gkelis, S., and Pantelidakis, K. (2008) Plankton
community structure during an ecosystem disruptive algal bloom of Prymnesium
parvum. J Plankton Res 31: 301-309.
37

Parsons, T.R., Maita, Y., and Lalli, C.M. (1984). A manual of chemical and biological methods
for seawater analysis (Oxford: Pergamon Press).
Pedros-Alio, C. (2006) Marine microbial diversity: can it be determined? Trends in microbiology
14: 257-263.
Redfield, A.C. (1958) The biological control of chemical factors in the environment. American
Scientist 46: 205-211.
Roelke, D.L., Grover, J.P., Brooks, B.W., Glass, J., Buzan, D., Southard, G.M., Fries, L., Gable,
G.M., Schwierzke-Wade, L., Byrd, M., et al. (2010) A decade of fish-killing
Prymnesium parvum blooms in Texas: roles of inflow and salinity. J Plankton Res
33: 243-253.
Rothschild, B.J., Ault, J.S., Goulletquer, P., and Héral, M. (1994) Decline of the Chesapeake Bay
oyster population: a century of habitat destruction and overfishing Mar Ecol Prog
Ser 111.
Schloss, P.D., Westcott, S.L., Ryabin, T., Hall, J.R., Hartmann, M., Hollister, E.B., Lesniewski,
R.A., Oakley, B.B., Parks, D.H., Robinson, C.J., et al. (2009) Introducing mothur:
open-source, platform-independent, community-supported software for describing
and comparing microbial communities. Appl Environ Microbiol 75: 7537-7541.
Scholin, C., Gulland F, Doucette GJ, Benson S, Busman M, Chavez FP, Cordaro J, DeLong R,
De Vogelaere A, Harvey J, et al. (2000) Mortality of sea lions along the central
California coast linked to a toxic diatom bloom. Nature 403: 80-84.
Simpson, E.H. (1949) Measurement of Diversity. Nature 163: 688.
Skovgaard, A., and Hansen, P.J. (2003) Food uptake in the harmful alga Prymnesium parvum
mediated by excreted toxins. Limnol Oceanogr 48: 1161-1166.
38

Sogin, M.L., Morrison, H.G., Huber, J.A., Welch, D.M., Huse, S.M., Neal, P.R., Arrieta, J.M.,
and Herndl, G.J. (2006) Microbial diversity in the deep sea and the underexplored
"rare biosphere". PNAS USA 103: 12115-12120.
Sommer, U., Gliwicz, Z.M., Lampert, W., and Duncan, A. (1986) The PEG-model of seasonal
succession of planktonic event in fresh waters. Arch Hydrobiol 106: 433-471.
Sunda, W.G., Graneli, E., and Gobler, C.J. (2006) Positive feedback and the development and
persistence of ecosystem disruptive algal blooms. J Phycol 42: 963-974.
Tillmann, U. (1998) Phagotrophy by a plastidic haptophyte, Prymnesium patelliferum. Aquat
Microb Ecol 14: 155-160.
Tillmann, U. (2003) Kill and eat your predator: a winning strategy of the planktonic flagellate
Prymnesium parvum. Aquat Microb Ecol 32: 73-84.
Vigil, P., Countway, P.D., Rose, J., Lonsdale, D.J., Gobler, C.J., and Caron, D.A. (2009) Rapid
shifts in dominant taxa among microbial eukaryotes in estuarine ecosystems. Aquat
Microb Ecol 54: 83-100.
Warwick, R.M.C.K.R. (1991) A Comparison of some methods for analysing changes in benthic
community structure. Journal of Marine Biology 71: 225-244.
Wetzel, R.G. (2001). Limnology: Lake and River Ecosystems, Vol 3rd (Philadelphia: Academic
Press).
Zamor, R.M., Glenn, K.L., and Hambright, K.D. (2012) Incorporating molecular tools into
routine HAB monitoring programs: Using qPCR to track invasive Prymnesium.
Harmful Algae 15: 1-7.
Zheng Zhang, S.S., Lukas Wagner, and Webb Miller (2000) A greedy algorithm for aligning
DNA sequences. Journal of Computational Biology 7: 203-214.
39

40

Figure Legends
Figure 1: Map of Lake Texoma, south central U.S. with the location of the two sampling sites:
Wilson Creek and Lebanon Pool.

Figure 2: An annual cycle at Lebanon Pool (Black Circles) and Wilson Creek (Gray Circles) in
Lake Texoma of surface water temperature (A), salinity (B), pH (C), dissolved oxygen (D),
extracted chlorophyll a, (E) Prymnesium parvum cell abundances determined by microscopical
counts (F), dissolved nitrogen (G-solid symbols) and total nitrogen (G-open symbols), dissolved
phosphorous (H-solid symbols) and total phosphorous (H-open symbols), and molar ratios of
total nitrogen to total phosphorous (I). Data and samples were collected from November 2008 to
October 2009. A duplicate sample collected in Wilson Creek during December 2008 is shown as
a dark gray circle in (A, B, C, E, and F). The inset in (F) illustrates the same data with a
narrower range of the y axis showing low but detectable abundances of P. parvum in Wilson
Creek during late winter and spring. The horizontal dashed lines in (I) represent the average
values for Lebanon Pool (black) and Wilson Creek (gray).

Figure 3: The percent contribution of 18s sequence tags in samples from Lebanon Pool (black
circles), and Wilson Creek (gray circles) in Lake Texoma attributed to genus Prymnesium (A),
and percent contribution of 16s sequence tags attributed to a Prymnesium plastid (B). Samples
were collected from November 2008 to October 2009. A duplicate sample collected in Wilson
Creek during December 2008 is shown as a dark gray circle. The insets illustrate the same data
with a narrower range of the y axis showing low but detectable abundances of Prymnesium
sequences in Wilson Creek during late winter and spring.
41

Figure 4: Diversity indices determined over an annual cycle (November 2008 to October 2009)
in Lebanon Pool (black circles), Wilson Creek (gray circles) and a duplicate sample from Wilson
Creek in December (dark gray circles) for eukaryotic OTUs (A,C,E,G) and bacterial OTUs
(B,D,F,H). Indices included the observed number of OTUs in each sample (A and B), the Chao 1
estimation of species richness (C and D), the CatchAll estimation of species richness (E and F),
and the 1-Simpson Index of Diversity estimate (G and H). Plastid sequences were removed from
the bacterial dataset prior to calculating diversity estimates. Error bars represent the computed
upper and lower 95% confidence intervals. The dotted line in (G) represents the 1-Simpson
Index of Diversity estimate, for the Lebanon Pool dataset with Prymnesium OTUs removed and
the data re-analyzed.

Figure 5: Cumulative dominance plots for eukaryotic (A) and bacterial (B) assemblages in
Lebanon Pool (black symbols) and Wilson Creek (gray symbols) for the months of December
(crosses), January (triangles), February (squares), March (diamonds), April (circles) and May
(vertical lines). The percent contribution of OTUs to total sequences are plotted by rank on a log
scale along the x axis and the cumulative percent of sequences on the Y axis. Plastid sequences
were removed from the bacterial dataset prior to generating the plots in (B).

Figure 6: Annual patterns of eukaryotic (A,C) and bacterial (B and D) community structure in
Lebanon Pool (black triangles), Wilson Creek (gray circles) and a duplicate sample from Wilson
Creek in December (gray x) based on two-dimensional non-metric multidimensional scaling
representation of Bray Curtis similarity values (A and B) and Jaccard similarity values (C and
42

D) calculated from non-transformed absolute OTU abundances. Sequential months have been
linked with lines to indicate annual trajectories of the similarities of the assemblages.
43

44

45

46

47

48

49

Supplemental Information
Supplemental Figure 1: An annual cycle of daily rainfall at the Madill OK, station (Black
Circles). The sampling dates of this study are represented with black triangles. Rainfall data is
taken from: http://www.mesonet.org.

Supplemental Figure 2: An annual cycle of day length in minutes (A) and surface water
temperature in Lebanon Pool (B) and Wilson Creek (C) plotted against the Julian Day. Dotted
lines represent the data fitted to a 4 parameter sinusoidal trend line with a set period of 365 days.
r
2
values for each curve are presented. Day length data is taken from
http://aa.usno.navy.mil/data/.

Supplemental Figure 3: Representative Rarefactions Curves for the non-subsampled December
samples from the 18s (A) and 16s (B) data. Curves and 95% confidence intervals were computed
for the three December samples: Lebanon Pool (black), Wilson Creek (light gray) and Wilson
Creek Duplicate (dark gray) in each dataset. Vertical dotted lines represent the level of
subsampling for each dataset analyzed in this study. Note that the three samples yielded different
numbers of sequences, as indicated by the different lengths of the lines.

Supplemental Figure 4: Annual cycle of the detected 18s (A and C) and 16s (B and D) OTUs
per month plotted against the Julian Day for Lebanon Pool (A and B) and Wilson Creek (C and
D). Dotted lines represent the data fitted to a 4-parameter sinusoidal trend line with a set period
of 365 days. r
2
values for each curve are presented.

50

Supplemental Figure 5: Cluster Diagrams of eukaryotic (A,C) and bacterial (B,D) community
structure in Lebanon Pool (black triangles), Wilson Creek (gray circles) and a duplicate sample
from Wilson Creek in December (gray x) based on group average clustering of Bray Curtis
similarity values (A and B) and Jaccard similarity values (C and D) calculated from non
transformed absolute OTU abundances. Dotted lines represent clusters of samples that were not
significantly different from one another on the basis of the SIMPROF permutation test. Vertical
gray dashed lines represent the level of similarity for the December Wilson Creek replicates.
51

52

53

54

55

56

Chapter 2
Taxon-specific responses at multiple trophic levels (metazoa, microbial eukaryotes, and
bacteria) within a freshwater microbial community restructured by disturbance events.

Abstract
Disturbance events are mechanisms leading to changes in communities of macroscopic species,
but the degree to which disturbance affects microbial community structure is poorly understood.
We characterized the monthly taxonomic composition of the microbial eukaryotic (18s rDNA)
and bacterial (16s rDNA) assemblages at two locations (Lebanon Pool and Wilson Creek) for
one year in a low salinity (0.5-3.5) lake, significantly affected by two disturbance events; a
localized and prolonged (4 month) bloom of the toxic alga Prymnesium parvum and a large rain
event overlaid onto more gradual seasonal environmental forcing. Pyrosequencing was used to
document microbial diversity, assign high-level identifications (phylum-to-supergroup), and
monitor changes in the relative abundances of taxa. The P. parvum bloom in Lebanon Pool
resulted in a major dichotomy between the microbial communities at the two sites. The
haptophyte strongly dominated the eukaryotic community when it bloomed, although high
abundances of fungi and chrysomonads were also observed. In contrast, the community at the
non-bloom site (Wilson Creek) contained a wide diversity of common taxa: haptophytes, ciliates,
cercozoa, chlorophytes, crustaceans and rotifers. Differences in the bacteria were more subtle,
although the actinobacteria were largely absent during the P. parvum bloom. A large rain event
resulted in massive restructuring of the microbial communities at both locations and rapid
decreases or increases of particular microbial groups (cercozoa, ciliates, and betaproteobacteria).
Collectively, unique high-level ecological trends among the metazoan, microbial eukaryotic and
57

bacterial communities were demonstrable, as well as altered trophic structure, as a consequence
of the two types disturbance.

Introduction
Aquatic communities, including microbial communities, are naturally structured and restructured
by biological, chemical, and physical forces. (Caron and Countway, 2009; Fuhrman, 2009).
Disturbance events encompass episodic extremes in these structuring forces and can be
biological (introduced or eliminated species), chemical (addition of nutrients or pollutants), or
physical (severe weather events, or advection) in nature. Depending on their frequency and
magnitude, disturbance events can result in completely altered ecosystems (Estes et al., 2011), or
act as mechanisms for maintaining biological diversity and ecosystem function within an
environment (Connell, 1978). The consequences and properties of disturbance have been well
documented within macro-organismal communities, and are becoming more widely recognized
as important factors affecting the structure of aquatic microbial communities (Allison and
Martiny, 2008; Floder and Sommer, 1999; Reynolds et al., 1993; Shade et al., 2012).
Harmful algal blooms arising from the dominance of species with toxic or noxious
properties, and mixing or flushing associated with strong meteorological storm events, are two
examples of disturbance events that can significantly alter freshwater, estuarine and brackish
aquatic systems (Buskey et al., 2001; Gobler and Sunda, 2012; Sunda et al., 2006). Numerous
studies have focused on how macro-organismal communities are altered or restructured in
response to these disturbance events (Estes et al., 2011), but less information is available on how
communities of microbial eukaryotes and bacteria respond to these forcing factors (Jones et al.,
2012; Jones et al., 2008; Michaloudi et al., 2008; Shade et al., 2010; Vigil et al., 2009) .
58

Lake Texoma is a brackish temperate lake located in the mid-southwestern United States.
The lake routinely experiences localized blooms of the toxic haptophyte alga Prymnesium
parvum and periodic salinity fluctuations related to meteorological events (Hambright et al.,
2010). We previously quantified cyclical patterns within the eukaryotic and bacterial community
structure (species richness and evenness) that reflected responses to gradual seasonal forcing and
distinct reactions to disturbance (Jones et al., 2012). In this paper we examined the high-level
taxonomic groupings of the microbial community (analogous to phylum-to-supergroup level) to
characterize the monthly taxonomic compositional changes in the metazoan (18s rDNA v9
region), microbial eukaryotic (18s rDNA v9 region) and bacterial (16s rDNA v6 region)
assemblages at two locations significantly structured by two disturbance events; a localized and
prolonged (4-month) bloom of the haptophyte alga P. parvum and a large (17cm) rain event,
during the course of one year in Lake Texoma.
P. parvum forms Ecosystem Disruptive Algal Blooms (EDABs) that are well documented
throughout the world (Evardsen and Imai, 2006). The alga produces a suite of toxins (Henrikson
et al., 2010; Igarashi et al., 1999) that can affect several trophic levels ranging from the death of
gill breathing organisms (Evardsen and Imai, 2006), altered life histories and fecundity of
crustaceous zooplankton (Cole, 1982; Nejstgaard and Solberg, 1996), and decreased mobility
and/or growth rates of ciliates, flagellated algal cells, diatoms, and filamentous cyanobacteria
(Fistarol et al., 2003). In addition, P. parvum is mixotrophic, combining phagotrophic and
phototrophic nutrition. The alga appears to be an obligate phototroph but it also readily engulfs a
variety of prey ranging from bacteria, to cryptophytes, small diatoms, ciliates, and heterotrophic
dinoflagellates (Martin-Cereceda et al., 2003; Skovgaard and Hansen, 2003; Tillmann, 1998;
Tillmann, 2003). While specific interactions between P. parvum and other taxa are well
59

documented, less is known about the degree to which blooms upset the trophic structure of the
total microbial community (Michaloudi et al., 2008).
Our study captured one of the largest P. parvum blooms documented in Lebanon Pool to
date (up to 180,000 cells/ml). The bloom lasted from January to April 2009 at this location,
while another location in the lake (Wilson Creek) experienced only low abundances of the alga
during the same period (<5000 cells/ml). The contrasting conditions at the two locations in this
study created a natural experiment in which to explore the direct effects of the P. parvum bloom
on the microbial community separated from changes due to seasonality. Additionally, both
locations experienced a large anomalous rain event at the beginning of May (17 cm in a single
day). This physical disturbance acted to reset the structure of the microbial communities at the
two sites, resulting in highly similar communities that remained similar for the remainder of the
one-year study. Further, the episodic nature of the rain event was reflected in rapid increases or
decreases in the taxonomic composition of the lake. We used sequence data to describe
ecological trends and captured altered trophic structures of the microbial communities uniquely
related to disturbance events.

Methods
Site description and sample collection
Lake Texoma is a temperate, low salinity reservoir (PSU ranges from 0.5-3) (Hambright et al.,
2010) on the border of Texas and Oklahoma in the southwestern United States. Two locations
were sampled monthly for one year from November 2008 to October 2009: Lebanon Pool (L.P.)
and Wilson Creek (W.C.). Wilson Creek was sampled twice during December to provide a true
biological replicate for the sampling and processing procedures. Near-shore water samples of
60

100-300 ml were filtered onto 47mm GF/F filters (Whatman) and stored at -80 for later DNA
extraction, PCR amplification and sequencing of the eukaryotic (18s) and bacterial (16s) rRNA
genes.
The lake experienced two major disturbance events during the study: 1) a localized and
prolonged bloom of the toxic alga Prymnesium parvum which occurred from January through
April in Lebanon Pool only, and 2) a shared episodic physical disturbance in the form of a
massive spring rain event and subsequent dramatic drop in salinity (from 2.5 to <0.25 PSU)
during May. These two disturbance events (the P. parvum bloom and the massive rain event) had
significant effects on both the eukaryotic and bacterial community structure that were clearly
overlaid onto the more gradual changes driven by cyclical seasonal changes in water temperature
and day length (Jones et al., 2012).

DNA sequence generation and processing
The v9 region of the 18s rRNA gene (Amaral-Zettler et al., 2009) and the v6 region of the 16s
rRNA gene (Sogin et al., 2006) were PCR amplified and sequenced via 454 titanium
pyrosequencing technology to assess diversity of the eukaryotic and bacterial communities. Fifty
DNA sequence libraries were generated (two genes, two locations, 12 months and one duplicate).
The open source software package MOTHUR v1.21.1 (http://mothur.org) was used for sequence
processing, alignments, and calling Operation Taxonomic Units (OTUs) (Schloss et al., 2009).
All raw reads were edited for quality (Q-score of ≥25, zero ambiguous bases, <8 homopolymers,
exact matches to the primer sequences (Huse et al., 2007)), and a good fit to SILVA ribosomal
small subunit (SSU) reference alignment. The 16s sequences were compared to the SILVA SSU
reference database (v 108) using stand alone BLAST+. Sequences identified as eukaryotic
61

plastids (~10% of the dataset) were removed from the 16s dataset and not considered in this
analysis. For details regarding the entire protocol from sequence generation to final processing
see Jones et al. (2012).
This paper aims to characterize the composition of the microbial community in the lake
using the best possible high-level taxonomic identifications. Our previous work was focused on
describing species richness and diversity trends, and therefore employed a different OTU calling
procedure (Jones et al., 2012). High quality 18s sequences (298,587) or 16s sequences (210,736)
from all datasets were clustered into OTUs using the single-linkage pre-clustering method (Huse
et al., 2010) and an average neighbor clustering method with a final threshold of "unique". Thus,
the sequences populating an OTU were identical save for single base-pair differences spawned
from highly abundant sequences that were detected and condensed during the pre-clustering step.
This approach yielded 27,140 eukaryotic OTUs and 13,129 bacterial 16s OTUs. Each OTU
represented a unique sequence variant. The number of sequences per sample (month/locations)
after screening for quality and clustering into OTUs varied from 7,293 to 23,148 in the 18s
rDNA dataset and from 2,948 to 15,098 in the 16s rDNA dataset. The February (19,013), March
(20,583) and April (16,982) samples in Lebanon Pool contained three of the largest numbers of
sequence per month. The number of sequences per OTU were converted into relative
abundances (percent values) within each sample in order to standardize our analyses.

Comparison to a previous study
The same raw sequence data set utilized in this analysis was used in a previous paper (Jones et
al. 2012), however the aims and data treatments in the two studies were different. The goal of
the previous study was to document and compare diversity patterns and environmental drivers of
62

species-level microbial (eukaryotic and bacterial) richness and community similarity throughout
the year. Each sequence library in that study was subsampled to normalize all libraries prior to
calling OTUs, in order to facilitate comparison of diversity indices across months and locations.
Moreover, global patterns in community structure in that study were assessed with OTUs
established at 97% sequence similarity, which has been shown to approximate species-level
OTUs for many microbial eukaryotes (Caron et al., 2009). Our approach in the present analysis
was to maximize the depth of sequencing in each sample, thus we did not subsample before
calling OTUs and instead standardized the data by converting the number of sequences per OTU
into relative abundances. The OTUs were comprised of identical sequences (100% similarity) to
avoid potentially condensing different species with 97% sequence similarity, yet different
taxonomic affiliations. The two datasets generated with the two approaches yielded greatly
different numbers of OTUs (9,770 vs. 27,140 in this study for 18s; 8,392 vs. 16,419 in this study
for 16s-with plastids removed), yet showed similar patterns in community similarity (see
results).

Taxonomic assignment of the 18s and 16s sequences
A representative sequence (one with the smallest total distance to all the other sequences within
an OTU) from each OTU was compared to the SILVA (SSU v108) and NCBI nt databases using
stand alone BLAST+, and run through the GAST pipeline (Huse et al., 2008). Each OTU was
assigned an identification based on a best match in SILVA or GAST which provided a nested
taxonomy in its results (Domain>Kingdom>Phylum>…>genus>…). A best match in SILVA or
NCBI was defined as ≥95% sequence similarity over 95% of the query sequence. A best match
in the GAST database was defined as a distance of 0.05 or less. More than 90% of the 16s
63

sequences, and over 66% of the all 18s sequences were assigned a taxonomy based on these
criteria, with two high level eukaryotic exceptions: fungi and cercozoa. There are fewer
sequenced representatives of the fungi and cercozoa in the SILVA and GAST databases, thus the
criteria for those two taxonomic affinities were relaxed to greater than 90% sequence similarity
over 95% of the query sequence. In some cases, a taxonomy within SILVA or GAST could not
be defined and the best match in NCBI nt database was used to assign an identification.
The overall taxonomic compositions and individual monthly distributions of higher
taxonomic groups within the eukaryotic and bacterial datasets at each location (Lebanon Pool or
Wilson Creek) were constructed by aggregating the sequences into broad groups such as the
"phyla" classifications provided by SILVA. Metazoan sequences (11,475) and metazoan OTUs
(830) comprised 4% and 3% of the 18s rDNA dataset, and were collected into a metazoan only
sub-dataset and analyzed separately.

Statistical analyses of community structure and similarity
The multivariate software package PRIMER v6.1.7 was used to compute Bray-Curtis similarity
values (Bray and Curtis, 1957) for the bacterial and eukaryotic communities between each month
and location, compute spearman correlations, perform permutation based statistical tests, build
cluster and non-metric multidimensional scaling (MDS) plots, and for the SIMPER analysis
(Clarke, 1993; Clarke and Warwick, 2001).
Bray-Curtis similarity values were computed from shared OTUs and relative abundances
of sequences within each OTU between individual pairs of samples. Bray-Curtis similarities
between the two locations and across the 12 months (25 samples in total) were calculated for the
following four datasets: 1) 18s-all sequences (protists, fungi, viridiplantae and metazoa), 2) 18s-
64

metazoa, 3) 18s-microbial eukaryotes (protists, fungi and viridiplantae), and 4) 16s-bacteria
(with plastids removed). Cluster diagrams based on average Bray-Curtis similarity scores for
each group were constructed from the Bray-Curtis similarity matrices for the 18s-metazoa, 18s-
microbial eukaryotic, and 16s-bacterial datasets. The SIMPROF permutation test was used to
identify statistically significant clusters. The same Bray-Curtis similarity matrices were
converted to ranks and then used to construct two-dimensional Multidimensional Scaling
diagrams (2D MDS) with a Kruskal fit of 1 and 25 restarts. Clusters of high average Bray-Curtis
similarity (25% or 35%), taken from the dendrograms were overlaid onto the MDS plots.
We used the clusters of high average Bray-Curtis similarity (see results) to organize the
individual samples within the 18s-metazoan, 18s-microbial eukaryotic and 16s-bacterial into
discrete clusters in which to explore differences related to disturbance and seasonality with the
SIMPER routine. Samples that were not statistically different on the basis of the SIMPROF test
were not split into different groups. Based on this approach, the 18s-metazoan dataset was
divided into four groups: 1) W.C. February through April, 2) L.P. February through April, 3)
L.P. May + W.C. May, and 4) L.P. June through October + W.C. June through October. The 18s-
microbial eukaryotic dataset was divided into six groups: 1) W.C. January + February 2) W.C.
March + April, 3) L.P. January through April, 4) L.P. May + W.C. May, 5) L.P. June + W.C.
June, and 6) L.P. July through October + W.C. July through October. The 16s-bacterial dataset
was divided into five groups: 1) W.C. February through April, 2) L.P. February through April, 3)
L.P. May + W.C. May, 4) L.P. June + W.C. June, and 5) L.P. July through October + W.C. July
through October. The SIMPER routine within Primer computes a simple average Bray-Curtis
similarity value for each user defined cluster and computes a simple average dissimilarity score
between the clusters. Further, because all shared OTUs do not contribute equally to the final
65

average Bray-Curtis similarity value for a given cluster of samples, the SIMPER routine was
used to explore the taxonomic identities behind each score.

Results
Taxonomic identification of the OTUs and sequences
A majority of the bacterial OTUs and sequences had (see methods) matches in both the SILVA
(Fig. 1B) and GAST (Fig. S1B) databases, and 62% of the OTUs and 88% of the sequences were
assigned a taxonomic identification (Fig. 1B and 1D). The GAST pipeline, and BLAST+ against
SILVA (SSU v108) overwhelming returned the same taxonomic identifications. More than 90%
of the bacterial sequences had a match to the NCBI nt database (Fig. S1D). The eukaryotic
taxonomic assignments were less well resolved than those for the bacterial sequences; 40% of
the eukaryotic OTUs and 53% of the sequences had matches in SILVA (Fig. 1A) and 31% of the
OTUs and 45% of the sequences had matches in the GAST databases (Fig. S1A). More than
70% of the eukaryotic sequences had a match to the NCBI nt database (Fig. S1C). There was an
uneven distribution of good matches within the different eukaryotic groups (Fig. 1C). A majority
of the 18s sequences with taxonomic affinities to the ciliates, diatoms, chrysophytes,
haptophytes, chlorophytes, and metazoa were assigned a taxonomy based on the ≥95% or ≥90%
(cercozoa only) sequence similarity criteria defined in the methods. The fungi were particularly
poorly identified using the SILVA database (Fig, 1C); 90% of the 18s sequences with fungal
affinities had matches of only ≤90% sequence similarity. However, most (80%) had good
matches (≥95% sequence similarity) to the NCBI nt database and were thus assigned as fungi.
Calling OTUs at 100% similarity generated a high degree of microdiversity. For example,
907 OTUs were assigned to the genus Prymnesium which comprised 28,672 sequences, although
66

a single Prymnesium OTU contained 89% of these sequences. When OTUs were clustered at
97% similarity (Jones et al., 2012), Prymnesium identity was assigned to 67 OTUs (out of a total
of 12,859 OTUs), comprised of 29,351 sequences. A single Prymnesium OTU (called at 97%
similarity) comprised 99% of the latter sequences. In this example, we avoided potentially
misclassifying 679 sequences as Prymnesium, by grouping our dataset into OTUs with a unique
level of similarity as opposed to 97% similarity.

High level taxonomic composition of lake Texoma
The high-level taxonomic aggregations of the 18s and 16s rDNA sequences over the year
revealed both compositional similarities and differences between Lebanon Pool and Wilson
creek (Fig. 2). Greater than 55% of the 18s rDNA sequences (Fig. 2A) were identified as one of
nine high-level classifications: ciliates, dinoflagellates, diatoms, chrysophytes, haptophytes,
cryptophytes, fungi, cercozoa (rhizaria), and metazoa, while 33 to 38% of the sequences could
not be identified with certainty (see Methods). Overall, the proportions of the diatoms (~10%),
chrysophytes (~5%), cryptophytes (~5%), and metazoa (~5%) were similar at both locations. The
proportion of haptophytes (14%, 5%) and fungi (9%, 2%) were markedly greater in Lebanon
Pool, while the proportions of ciliates (8%, 11%), chlorophytes (6%, 9%), and cercozoa (4%,
6%) were greater in Wilson Creek. A majority of the remaining 20% of the sequences that could
be identified were ichthyosporea, Pirsonia, euglenids, rhodophytes, and choanoflagellates.
Greater than 85% of the 16s rDNA sequences (Fig 2B) were identified as one of nine
high-level classifications of bacteria: alphaproteobacteria, betaproteobacteria,
gammaproteobacteria and deltaproteobacteria (all proteobacteria), sphingobacteria and
flavobacteria (both bacteroidetes), cyanobacteria, verrucomicrobia and actinobacteria, while 10%
67

of the sequences could not be identified with certainty. The relative proportions of each group
(~20%, to 3%) were very similar at both locations. The majority of the remaining 5% of the 16s
sequences were identified as planctomycetes, chloroflexi, and acidobacteria.

Divergent and parallel patterns of community structure within the metazoan, microbial
eukaryotic, and bacterial assemblages.
The metazoan, microbial eukaryotic, and bacterial community structures at the two sampling
sites responded uniquely to the site-specific biological disturbance event (a P. parvum bloom that
occurred from January to April in Lebanon Pool only), but experienced similar responses to the
episodic physical disturbance event (heavy rain/salinity decrease during May at both locations),
and eventually returned to similar community structures at the two locations (June/July through
October) that were similar to community structures observed at the start of the study (Fig. 3).
Spearman correlations between the Bray-Curtis similarity matrices previously reported (97%
level OTUs and the sequences standardized by subsampling (Jones et al. 2012)), and the
similarity matrices in this study ("unique" level OTUs with sequence counts converted to relative
abundances) were high: 0.988 between the two 18s datasets, and 0.998 between the 16s datasets
(without plastids) and p-values <0.05.
The 2D MDS representations of the metazoan (Fig. 3A), microbial eukaryotic (Fig. 3C),
and bacterial community (Fig. 3E) similarities each had low stress values (0.18, 0.1, and 0.08)
and revealed similar pictures of seasonal cyclicity and response to disturbance. Spearman
correlations between the three Bray-Curtis similarity matrices were good: metazoa and microbial
eukaryotes (0.65), microbial eukaryotes and bacteria (0.77), and metazoa and bacteria (0.51), all
p-values <0.001.
68

The metazoan communities formed four statistically discrete (SIMPROF) clusters of
high average similarity (Fig. 3A and Fig. S2A) that broadly separated the metazoan community
by season and disturbance. The clusters were: 1) November through April in Wilson Creek and
December in Lebanon Pool formed a cluster of 25% similarity, and the individual samples within
this group were not statistically different from one another, 2) February through April samples in
Lebanon Pool clustered at 15% similarity and the individual months were not statistically
different, 3) May samples at both locations clustered discretely at 30% similarity and were not
statistically different, 4) June through October at both locations formed a large cluster at 20%
similarity. The latter cluster had one significant break; however within each subcluster the
samples were not statistically different. July and September at both sites, October in Lebanon
Pool and July, September and August in Wilson Creek were not statistically different, and June
plus October in Lebanon Pool and June plus August in Wilson Creek were not statistically
different. The two December samples were 80% similar and not statistically different from one
another. The four clusters occupied three discrete regions on the MDS plot (Fig. 3A). The May
samples were placed adjacent to the Wilson Creek winter cluster, as the two were part of a larger
cluster at 15% similarity.
The microbial eukaryotic communities formed six discrete, significant (SIMPROF)
clusters of at least 25% group average similarity (Fig. 3B and Fig. S2B). Each distinct cluster
occupied discrete physical space on the MDS plot, presumably reflecting the microbial
eukaryotic community response to seasonality, disturbance, and recovery from disturbance. The
clusters were: 1) January plus February in Wilson Creek were 40% similar and not statistically
different, 2) March plus April in Wilson Creek were 28% similar and not statistically different,
3) January through April in Lebanon Pool were 40% similar; January and February were not
69

statistically different from each other, nor were March and April, 4) May samples at both
locations clustered discretely at 45% similarity and were not statistically different, 5) June
samples at both locations clustered at 30% similarity and were not statistically different, 6)
samples from July through October in both Lebanon Pool and Wilson Creek clustered with 30%
similarity. The two December samples were 80% similar and not statistically different from one
another.
The bacterial communities formed five statistically significant (SIMPROF) clusters with
at least 35% group average similarity (Fig. 3C and Fig. S2C). The arrangement of the clusters
reflects the bacterial communities’ response to the disturbance, recovery, and seasonality of the
system. The clusters were: 1) December plus January in Lebanon Pool and November through
April in Wilson Creek clustered at 35% similarity and the following pairs of samples within the
cluster were not statistically different: January at both locations and February plus March in
Wilson Creek, 2) February through April in Lebanon Pool clustered at 35% similarity and March
plus April were not statistically different, 3) May samples at both locations clustered discretely at
60% similarity, 4) June samples at both locations clustered discretely at 55% similarity and were
not statistically different, 5) July through October in both Lebanon Pool plus Wilson Creek
clustered with 50% similarity. The two December samples were 70% similar and not statistically
different from one another.

Contribution of taxonomic groups to the average Bray-Curtis similarity values for clusters of
samples defined and divided by seasonality and disturbance
The Bray-Curtis similarity analyses for the metazoan, microbial eukaryotic and bacterial datasets
(Fig. 3 and Fig S2) revealed patterns in community structure that reflected the natural history of
70

the lake, and suggested natural clusters of samples (see Methods) based on high average Bray-
Curtis similarity values. The SIMPER routine quantified the differences between the defined
clusters and deconstructed the individual contributions of each OTU to the overall average Bray-
Curtis similarity score. We aggregated the high-level taxonomic identifications of the OTUs in
these samples to describe the contributions of these groups to samples with high average Bray-
Curtis similarity scores.
The metazoan community showed marked differences related to disturbance events or
seasonality, and a majority (90%) of the similarity scores were described by OTUs classified as
one of four high level groups: crustacea, hexapods, rotifers, and nematodes (Fig. 4A). The algal
bloom disturbance event in Lebanon Pool from January to April divided the winter samples into
two distinct clusters separated by location (Fig. 3A; 4A). The communities in Wilson Creek from
February through April were composed of rotifers or crustacea, while the Lebanon Pool metazoa
during the same time period were comprised of nematodes and rotifers (Fig. 4A). The rain event
in early May resulted in high similarity of the metazoan community at both locations (Fig. 3A),
and crustacean OTUs largely contributed to that similarity (Fig. 4A). The metazoan community
at both locations from June through October were strongly dominated by rotifers.
The microbial eukaryotic community contained a greater diversity of high-level
taxonomic groups compared to the metazoa, yet still exhibited clear compositional distinctions
related to disturbance and seasonality (Fig. 4B). The microbial eukaryotic dataset contained a
high number of unidentified sequences (Fig. 1A and 2A) thus only 60 to 80% of the similarity
scores were described by one of the nine identified taxonomic groups. Most notably, the P.
parvum bloom in Lebanon Pool distinguished the winter assemblage (January through April) at
that location (Fig. 3B), and OTUs identified as haptophytes or fungi dominated the unique
71

community composition at that location (Fig. 4B). A diversity of OTUs (ciliates, diatoms,
haptophytes, chlorophytes, and cercozoa) comprised the microbial eukaryotic community in
Wilson Creek at that time (Fig. 4B).
The spring rain event resulted in high similarity for the May samples at both locations
(Fig. 3B) with an average Bray-Curtis similarity score of 45% defined largely by ciliate and
cercozoan OTUs (Fig. 4B). A diversity of taxa: ciliates, diatoms, cryptophytes, and chlorophytes
contributed to the average similarity scores (28% and 32%) found for summer/fall clusters (Fig.
4B).
The bacterial community contained a similar number of taxonomic groups to the
microbial eukaryotes, yet compositional changes and distinctions related to the disturbance
events were more subtle for bacteria than for the metazoan or the microbial eukaryotic
communities (Fig. 4C). More bacterial sequences were identified (Fig. 1B and 2B) and thus we
were able to capture 75 to 90% of the similarity scores with 7 major bacterial groups:
alphaproteobacteria, betaproteobacteria, flavobacteria, sphingobacteria, cyanobacteria,
verrucomicrobia, and actinobacteria (Fig. 4C). The algal bloom disturbance events structured
the bacterial communities (Fig. 3C), but to a lesser degree than observed for the eukaryotic
communities (Fig. 3A and B). The February to April samples in Lebanon Pool had one major
compositional difference (Fig. 4C), a lack of the actinobacteria in Lebanon Pool.
The rain event resulted in a single cluster for the May bacterial communities at both
sampling sites (Fig. 3C) with high similarity (60%). Two taxa, alphaproteobacteria and
betaproteobacteria, contributed to half this similarity score (Fig. 4C).
The two summer/fall bacterial clusters, June at both locations (55% average Bray-Curtis
similarity) and the July through October at both locations (48% average similarity), both
72

contained a diversity of taxa (Fig. 4C) with minor changes in the percent contributions. For
example actinobacteria contributed to more of the similarity in the June cluster, while
cyanobacteria contributed to more of the similarity in the July through October cluster.

Yearly taxonomic trends reflected the natural history of lake Texoma
Changes in abundances of the specific, high-level taxonomic groups in Lebanon Pool and Wilson
Creek reflected the environmental variability of the lake over the year (Fig. 5 and 6) including
the bloom of P. parvum from January to April in Lebanon Pool, a large spring rain (and resulting
episodic drop in salinity; 2.5 to <0.25) during May, and seasonal changes in chemical and
physical parameters (Jones et.al. 2012).
The algal bloom was captured in the high proportion of haptophyte sequences (Fig. 5A)
in Lebanon Pool (30-50%) from January to April. The family Prymnesiales comprised a
majority of the sequences in Lebanon Pool (Fig. 7C), while the family Pavlovales comprised a
one-month increase during April at Wilson Creek (Fig. 7D). Several high-level taxonomic
classifications exhibited differences between the two locations coincident with the timing and
duration of the bloom event. The proportion of fungi (Fig 5B) in Lebanon Pool increased (2 to
10%) from December to January, reached a maximum abundance (41%) in March, and declined
(<2%) through May. In contrast, the relative abundances of fungi in Wilson Creek were low
(<2%) from December to April. The proportion of chrysophytes in Lebanon Pool increased (2 to
10%) from December to January reached maximum abundances (15%) in April and declined
(<2%) in May (Fig. 5C). Chrysophytes in Wilson Creek exhibited a one-month episodic increase
in February (19%). Ciliates (Fig 5E), cercozoa (Fig. 5E), metazoa (Fig. 5F and Fig. 7A and B),
chlorophytes (Fig. 5G) and actinobacteria (Fig. 6G) were considerably less abundant in Lebanon
73

Pool compared to Wilson Creek over the four-month duration of the bloom. The metazoan
community throughout the year at both locations was primarily composed of rotifers and
crustacea (Fig. 7A and B), but neither group was detected during the P. parvum bloom in
Lebanon Pool. Dinoflagellates and diatoms showed notable differences between the two
locations during the course of the bloom. Diatoms (Fig 5I) were less abundant (<0.5%) in
Lebanon Pool compared to Wilson Creek (5 and 10%) during February and March.
Dinoflagellates (Fig 5H) were less abundant (<0.5%) in Lebanon Pool compared to Wilson
Creek (5%) in March and April.
Haptophytes, fungi, ciliates, cercozoa, chlorophytes, cyanobacteria, and
betaproteobacteria exhibited abrupt changes in abundance coincident with the rain event in May.
Haptophytes (both locations; Fig. 5A and Fig. 7A and B), fungi and chrysophytes in Lebanon
Pool (Fig. 5B and C) and cyanobacteria at both locations (Fig. 6H, Fig. 7E and F) sharply
decreased (from 18-30% to <2%) from April to May. Haptophytes, fungi, and chrysophytes
remained at low abundances at both locations for the remainder of the study (Fig. 5A-C), while
the cyanobacteria increased through July and sustained high abundances (20-25%) through the
fall (Fig. 6H and Fig. 7C and D). Ciliates (Fig. 5D), cercozoa (Fig. 5E), and betaproteobacteria
(Fig. 6B) increased sharply at both locations from April to May. These high abundances declined
to approximately half by June.
The microbial community was significantly affected and structured by the P. parvum
bloom and the rain event, while taxonomic responses related to seasonality were more subtle.
Haptophytes (Fig. 5A), chrysophytes (Fig. 5C), alphaproteobacteria (Fig.6A), and flavobacteria
(Fig. 6E) were more abundant during the winter-spring months of December through April or
May. Cryptophytes (Fig. 5J), diatoms (Fig. 5I), and cyanobacteria (Fig. 6H, Fig. 7E and F) were
74

more abundant in the summer-fall months of November, and June to October. The
gammaproteobacteria (Fig. 6C), sphingobacteria (Fig. 6D) and verrucomicrobia (Fig. 6F) did not
clearly exhibit trends related to seasonality or disturbance.
Cyanobacteria (Fig. 6H) were a major component (20% or greater) of the 16s assemblage
in November, April and from July through October at both locations. Cyanobacteria at both
locations decreased from the beginning of the study in November (≥20%) to approximately half
this value in December and remained low and variable through March. Following a precipitous
decrease that coincided with the May rain event, cyanobacteria at both locations increased
through the summer and remained at approximately 18-20% through October. Sequences
identified as Synechococcus (Fig. 7 E and F) were significant contributors to total cyanobacteria
at both locations (Fig. 7E and F).

Discussion
Disturbance events are fundamental forces that affect biological communities, and there is a
growing recognition of the strong and even predictable role of disturbance in patterning
microbial communities. Rain and salinity decreases have been documented to reset or re organize
microbial community structure (Buskey et al., 2001; Jones et al., 2008) and recent studies have
documented changes within the microbial communities specifically in response to harmful algal
blooms (Fistarol G et al., 2004; Michaloudi et al., 2008; Teeling et al., 2012; Vigil et al., 2009).
We used clone library independent pyrosequencing of the 18s (v9) and 16s (v6) rDNA genes to
characterize the taxonomic composition at two locations: Lebanon Pool and Wilson Creek within
Lake Texoma over the course of one year. Changes in the relative abundances of high-level
75

taxonomic groups were examined, indicating ecological and biological trends distinctly related to
both seasonality and episodic disturbance.
We defined OTUs in this study as unique sequences in order to obtain the most precise
taxonomic identifications for the sequences, from which we constructed our high-level
taxonomic groupings. The approach resulted in a high degree of microdiversity in our samples:
27,140 eukaryotic OTUs from the non-subsampled dataset, compared to 12,859 OTUs called at
97% similarity from a nonsubsampled dataset, and 9,770 OTUs called at 97% similarity from a
subsampled dataset (~7000 sequences/sample; Jones et al. 2012). Nevertheless, the patterns of
eukaryotic and bacterial community similarity (Bray-Curtis index) observed throughout the year
at both locations were remarkably similar to those previously reported (OTUs formed at 97%
similarity from subsampled datasets) (Jones et al 2012). The total eukaryotic OTUs were further
divided in this study into metazoa and microbial eukaryotes (protists, fungi, and green plant), and
parallel patterns of community similarity within the metazoan, microbial eukaryotic and bacterial
assemblages were observed (Fig. 3A-C). The community similarity patterns detected from the
responses of high-level taxonomic groups in this study implied significant alterations of the
trophic structure of the lake distinctly related to the disturbance events (Fig.4A-C).

Assigning taxonomies to DNA sequences
A primary goal of this study was to extend the general descriptions of bacterial and microbial
eukaryotic species richness presented in Jones et al. (2012) to describe the taxonomic
composition of these assemblages, and to examine trophic structure within these communities at
a relatively high taxonomic level of characterization in the face of disturbance events and
seasonal forcing. Taxonomies were established at the highest resolution (i.e. unique sequences)
76

with subsequent compilation into high-level taxonomic groups to provide the most precise
identities possible. The practice of assigning a taxonomies to ribosomal sequence data is proving
to be a useful tool for discovering ecological trends (Andersson et al., 2009; Peura et al., 2012),
despite the known caveats behind extending PCR based sequence datasets into organismal
patterns (Medinger et al., 2010). Most (~80%) of the 16s sequences had excellent (>99%
identity) or good (>95% identity) matches to representative sequences in the SILVA SSU v108
reference database (Fig. 1B), and were confidently assigned to high-level taxonomic groups.
Fewer 18s sequences, compared to the 16s dataset, had good matches in SILVA (Fig.1A).
However, we were able to confidently assign high-level identities to 60% of the sequences. Two
eukaryotic groups, the fungi and cercozoan (Fig 1C) were particularly poorly resolved in our
dataset, perhaps because both groups are highly diverse, with deeply branching phylogenies
(Bass et al., 2009; James et al., 2006; Jones et al., 2011), and fewer sequenced representatives
were present in the reference database compared to the other eukaryotic groups. Their
significant contribution to the eukaryotic assemblages in our study suggests that these two
groups have ecological importance in lake ecosystems, particularly during times of disturbance.
Approximately 70% of the eukaryotic sequences had good matches in NCBI (Fig. S1C).
In some cases these matches provided taxonomic information, in particular for many of the fungi
and cercozoan sequences (Lefevre et al., 2007; Lefèvre et al., 2008), while in other cases the best
matches in NCBI were to sequences generated from environmental surveys. The matches with
environmental surveys provided reassurance that the sequences generated in this study were
likely not errors. Many of our sequences had good matches to an environmental survey
associated with a fish kill in a lake (Oikonomou et al., 2012). Overall the relatively high
percentage of unidentified eukaryotic sequences found in this study highlights the benefits of and
77

need for combining molecular and morphological analyses to fully describe eukaryotic microbial
diversity.

Changes in the community composition reflected underlying seasonality
Seasonal patterns for a few high-level taxonomic groups were detected within the year-long
dataset despite the fact that the microbial communities in this system were strongly shaped by
two disturbance events (the P. parvum bloom and the rain event). Seasonal patterns of metazoa,
microbial eukaryotes and bacteria in aquatic ecosystems are well documented (Andersson et al.,
2010; Berninger et al., 1993; Kan et al., 2006; Newton et al., 2011; Pinhassi and Hagström,
2000). Three high-level groups primarily composed of phototrophs (cryptophytes, Fig. 6J;
diatoms, Fig. 6I; cyanobacteria, Fig. 6H, particularly filamentous forms, Fig. 7F) were more
abundant during the summer/early fall and exhibited abundance patterns that reflected the
gradual warming of the surface water, and increasing photoperiod. Two eukaryotic lineages
(haptophytes, Fig. 6A, particularly prymnesiales, Fig. 7A; chrysophytes, Fig. 6C, which included
the heterotrophic Paraphysomonas) had higher abundances during the winter/early spring
months at both locations. The occurrence of the phototrophic P. parvum during the winter is not
surprising given that this species is mixotrophic and may supplement its nutrition in winter
months via heterotrophic processes. This alga routinely forms blooms during the winter months
throughout lake Texoma (Hambright et al., 2010; Roelke et al., 2010). Species of
Paraphysomonas have been shown to dominate enrichment cultures of small heterotrophic
protists from the winter/spring as opposed to the summer and fall (Lim et al., 1999).

78

A harmful algal bloom created contrasting trophic structures at the two locations
P. parvum produces a suite of toxins (Henrikson et al., 2010; Igarashi et al., 1999), has
mixotrophic capabilities (Skovgaard and Hansen, 2003; Tillmann, 1998), and is considered an
ecosystem disruptive algal bloom species (EDAB) (Sunda et al., 2006). P. parvum blooms have
well-documented negative effects on a range of species including some fish, zooplankton,
heterotrophic protists and phytoplankton (Cole, 1982; Fistarol et al., 2003; Martin-Cereceda et
al., 2003; Michaloudi et al., 2008; Nejstgaard and Solberg, 1996; Tillmann, 2003). Documented
negative associations between P. parvum and specific microbial taxa include ciliates (Fistarol et
al., 2003), chlorophytes (Michaloudi et al., 2008), cryptophytes (Uronen et al., 2007), diatoms
(Martin-Cereceda et al., 2003), dinoflagellates (Tillmann, 2003), crustacea (Remmel et al.,
2011), rotifers (Brooks et al., 2010), and cyanobacteria (Fistarol et al., 2003; Michaloudi et al.,
2008).
Lebanon Pool experienced a bloom of P. parvum from January to April of 2009, while
Wilson Creek did not (Jones et al., 2012). This site-specific difference allowed us to examine
effects related to the bloom separated from seasonality experienced at both locations. We have
previously documented the disruptive nature of a P. parvum bloom on microbial species richness
and diversity (Jones et al., 2012). In the present study we further demonstrated its effect on
metazoan, microbial eukaryotic, and bacterial community composition (Fig. 3) and revealed
altered trophic structures within the microbial communities at the two locations (Fig. 4). While
we did not measure toxins as part of this study, there was a documented fish kill related to a toxic
P. parvum bloom at a nearby location in the lake during February (Henrikson et al., 2010).
Haptophytes contributed extensively to the high similarity values for the eukaryotic
community from January through April in Lebanon Pool, while a diversity of taxa contributed to
79

these values in Wilson Creek (Fig. 4B). In addition, several other microbial taxa exhibited
contrasting trends at the two locations coincident with the timing of the bloom. Ciliates (Fig.
5D), cercozoa (Fig. 5E), metazoa (Fig. 5F; specifically rotifers and crustaceans, Fig. 7A and B,
respectively), chlorophytes (Fig. 5G), dinoflagellates (Fig. 5H), and actinobacteria (Fig. 6G)
occurred at low abundances or were undetectable in Lebanon Pool during the bloom.
Several of these trends were expected based on the published effects of P. parvum (see above),
but the negative relationships with cercozoa and actinobacteria have not been reported
previously. Cercozoa were consistently present at both locations during much of the year, but
their abundances decreased dramatically during the bloom only in Lebanon Pool (Fig. 4B and
5E). Most cercomonads are small bacterivorous flagellates (Bass et al., 2009), but they are a
highly diverse group (Bass and Cavalier-Smith, 2004). The sequences in our study were not well
represented in the SILVA database, and were defined as cercozoa based on good matches to the
cercozoan lineages from environmental surveys (Lefevre et al., 2007; Lefèvre et al., 2008). The
negative correlation between these taxa and P. parvum expands the range of taxa negatively
impacted by this alga.
Actinobacteria are typically very small (<0.1 µm
3
) and often abundant constituents of
lake ecosystems (Newton et al., 2011). These species were consistently present in the lake
during our study with the exception of Lebanon Pool during the P. parvum bloom (Fig. 4C and
Fig. 6G). We speculate that their conspicuous absence may indicate a negative effect from the
exposure of excreted toxins, or ecological vulnerability as preferential prey source (the suitability
of actinobacteria as prey is reviewed in (Newton et al., 2011)).
Fungi (Fig. 5B) were highly abundant and co-dominated with P. parvum during the
bloom in Lebanon Pool, but were not major components of the microbial eukaryotic community
80

in any other samples (Fig.7C). This result implies tolerance to the production of toxin by the alga
and/or avoidance of predation. These abilities apparently provided an ecological advantage for
the fungi, perhaps enhanced by relaxed competition from other species that were negatively
impacted by P. parvum. Fungi are diverse and common components of aquatic ecosystems
where they play a variety of ecological roles including the decomposition of biomass
(ascomycota), and parasitic infections of phytoplankton (chytridomycota) (Jobard et al., 2010).
However, our ability to classify the fungi was limited because fungal sequences were not well
represented in the SILVA reference database (Fig. 1C).
Chrysophytes (including chloroplast-bearing as well as heterotrophic forms) were also common
components of the eukaryotic community in Lebanon Pool over the course of the bloom (Fig.
5C), suggesting tolerance to, if not benefit from, the presence of P. parvum. However,
chrysophytes were also present at similar abundances in Wilson Creek (where no P. parvum
occurred) during the winter months, perhaps indicating a preference of chrysophytes for winter
conditions and a tolerance of these species for the ecological effects of the haptophyte. Many of
the chrysophyte sequences detected in this study could be identified to species in the
heterotrophic genus Paraphysomonas.

The rain event coincided with unique, short-lived communities that reflected the episodic nature
of the disturbance.
Storms are known to be major physical disturbances that can perturb (Jones et al., 2008; Shade et
al., 2010) and restructure communities (Buskey et al., 2001).The massive rain event in the
beginning of May during our study had two results: 1) it introduced or caused the formation of
microbial communities with distinctive taxonomic compositions relative to communities prior to
81

the event, and 2) it returned the divergent microbial communities at the two locations to highly
similar communities. The metazoan, microbial eukaryotic, and bacterial communities at both
locations for the month of May had high Bray-Curtis similarity values (Fig. 3) with unique
taxonomic profiles (Fig. 4).
The rain event both removed and added high-level taxonomic groups at the study sites.
We previously noted a dramatic decrease in chlorophyll coinciding with the salinity decrease in
May (Jones et al., 2012). Our analysis in the present study expanded our understanding of that
finding. Pronounced reductions in the contributions of photosynthetic taxa including
cyanobacteria (Fig. 6H), haptophytes (Fig. 5A) and chlorophytes (Fig. 5G) occurred with the rain
event. A marked decrease in the fungi also occurred in Lebanon Pool at the time of the event
(Fig. 5B). Conversely, the rain resulted in temporarily high abundances of ciliates (Fig. 5D),
cercozoa (Fig. 5E) and betaproteobacteria. It is unclear whether these taxa resulted from rapid
growth of endemic populations following changed ecological conditions after the rain, or if they
were advected to the sites from water entering the lake during and after the rain. After the event,
the all three components of the microbial communities at the two locations remained highly
similar (Fig. 3 and Fig. 4).

Conclusions
Pyrosequencing of 16s and 18s genes was conducted on monthly samples from two sites in Lake
Texoma in order to describe annual patterns of abundances of high-level taxonomic groups of
bacteria, protists and metazoa, and to examine the response of these microbial communities to
disturbance events. A subtle seasonal cycle was apparent for some microbial taxa, but a bloom of
the harmful alga, Prymnesium parvum, at one of the sites significantly disrupted community
82

structure and resulted in major trophic differences between the communities at the two sites. The
eukaryotic community at the time of the algal bloom in Lebanon Pool was dominated by the
toxic haptophyte, and fungi were the only other taxonomic group that appeared to benefit from
the presence of the harmful alga. Micrometazoa were undetected during the bloom and
abundances of actinobacteria were considerably decreased. A massive rain event in May
produced a strong physical disturbance at both lake sites resulting in dramatic changes in the
taxonomic composition of the microbial communities, and reestablishing similar microbial
communities at the two locations for the remainder of the year.

References
Allison, S.D., and Martiny, J.B.H. (2008) Resistance, resilience, and redundancy in microbial
communities. Proceedings of the National Academy of Sciences 105: 11512-11519.
Amaral-Zettler, L.A., McCliment, E.A., Ducklow, H.W., and Huse, S.M. (2009) A method for
studying protistan diversity using massively parallel sequencing of V9 hypervariable
regions of small-subunit ribosomal RNA genes. PLoS One 4: e6372.
Andersson, A.F., Riemann, L., and Bertilsson, S. (2009) Pyrosequencing reveals contrasting
seasonal dynamics of taxa within Baltic Sea bacterioplankton communities. ISME J 4:
171-181.
Andersson, A.F., Riemann, L., and Bertilsson, S. (2010) Pyrosequencing reveals contrasting
seasonal dynamics of taxa within Baltic Sea bacterioplankton communities. ISME J 4:
171-181.
83

Bass, D., and Cavalier-Smith, T. (2004) Phylum-specific environmental DNA analysis reveals
remarkably high global biodiversity of Cercozoa (Protozoa). Int J Syst Evol Microbiol 54:
2393-2404.
Bass, D., Howe, A.T., Mylnikov, A.P., Vickerman, K., Chao, E.E., Edwards Smallbone, J.,
Snell, J., Cabral Jr, C., and Cavalier-Smith, T. (2009) Phylogeny and Classification of
Cercomonadida: Cercomonas, Eocercomonas, Paracercomonas, and Cavernomonas gen.
n. Protist 160: 483-521.
Berninger, U.G., Wickham, S.A., and Finlay, B.J. (1993) Trophic coupling within the microbial
food web: a study with fine temporal resolution in a eutrophic freshwater ecosystem.
Freshwater Biology 30: 419-432.
Bray, J., and Curtis, J. (1957) An ordination of the upland forest communities of southern
Wisconsin. Ecol Monogr 27: 325-349.
Brooks, B.W., James, S.V., Valenti, J.T.W., Urena-Boeck, F., Serrano, C., Berninger, J.P.,
Schwierzke, L., Mydlarz, L.D., Grover, J.P., and Roelke, D.L. (2010) Comparative
toxicity of Prymnesium parvum in inland Waters Journal of the American Water
Resources Association 46: 45-62.
Buskey, E.J., Liu, H., Collumb, C., Guilherme, J., and Bersano, J.G.F. (2001) The decline and
recovery of a persistent Texas brown tide algal bloom in the Laguna Madre (Texas,
USA). Estuaries 24: 337-346.
Caron, D.A., and Countway, P.D. (2009) Hypotheses on the role of the protistan rare biosphere
in a changing world. Aquat Microb Ecol 57: 227-238.
84

Caron, D.A., Countway, P.D., Savai, P., Gast, R.J., Schnetzer, A., Moorthi, S.D., Dennett, M.R.,
Moran, D.M., and Jones, A.C. (2009) Defining DNA-based operational taxonomic units
for microbial-eukaryote ecology. Appl Environ Microbiol 75: 5797-5808.
Clarke, K.R. (1993) Non-parametric multivariate analyses of changes in community structure.
Australian Journal of Ecology 18: 117–143.
Clarke, K.R., and , and Warwick, R.M. (2001). Change in Marine Communities: An Approach to
Statistical Analysis and Interpretation, 2nd edn. edn (Plymouth, UK).
Cole, J.J. (1982) Interactions between bacteria and algae in aquatic ecosystems. Annual Review
of Ecology and Systematics 13: 291-314.
Connell, J.H. (1978) Diversity in tropical rain forests and coral reefs. Science 199: 1302-1310.
Estes, J.A., Terborgh, J., Brashares, J.S., Power, M.E., Berger, J., Bond, W.J., Carpenter, S.R.,
Essington, T.E., Holt, R.D., Jackson, J.B.C., et al. (2011) Trophic downgrading of planet
Earth. Science 333: 301-306.
Evardsen, B., and Imai, I. (2006). The ecology of harmful flagellates within Prymnesiophyceae
and Raphidophyceae. In Ecology of Harmful Algae, E. Granéli, and J. Turner, eds.
(Netherlands: Springer-Verlag), pp. 67-79.
Fistarol G, Legrand C, Selander E, Hummert C, Stolte, W.a., and E., G. (2004) Allelopathy in
Alexandrium spp.: effect on a natural plankton community and on algal monocultures.
Aquat Microb Ecol 35: 45–56.
Fistarol, G.C., Legrand C., and E, G. (2003) Allelopathic effect of Prymnesium parvum on a
natural plankton community. Mar Ecol Prog Ser 255: 115–125.
Floder, S., and Sommer, U. (1999) Diversity in planktonic communities: an experimental test of
the intermediate disturbance hypothesis. Limnol Oceanogr 44: 1114-1119.
85

Fuhrman, J.A. (2009) Microbial community structure and its functional implications. Nature
459: 193-199.
Gobler, C.J., and Sunda, W.G. (2012) Ecosystem disruptive algal blooms of the brown tide
species, Aureococcus anophagefferens and Aureoumbra lagunensis. Harmful Algae 14:
36-45.
Hambright, K.D., Zamor, R.M., Easton, J.D., Glenn, K.L., Remmel, E.J., and Easton, A.C.
(2010) Temporal and spatial variability of an invasive toxigenic protist in a North
American subtropical reservoir. Harmful Algae 9: 568-577.
Henrikson, J.C., Gharfeh, M.S., Easton, A.C., Easton, J.D., Glenn, K.L., Shadfan, M., Mooberry,
S.L., Hambright, K.D., and Cichewicz, R.H. (2010) Reassessing the ichthyotoxin profile
of cultured Prymnesium parvum (golden algae) and comparing it to samples collected
from recent freshwater bloom and fish kill events in North America. Toxicon : official
journal of the International Society on Toxinology 55: 1396-1404.
Huse, S.M., Dethlefsen, L., Huber, J.A., Welch, D.M., Relman, D.A., and Sogin, M.L. (2008)
Exploring microbial diversity and taxonomy using SSU rRNA hypervariable tag
sequencing. PLoS Genetics 4: e1000255.
Huse, S.M., Huber, J.A., Morrison, H.G., Sogin, M.L., and Welch, D.M. (2007) Accuracy and
quality of massively parallel DNA pyrosequencing. Genome Biol 8: R143.
Huse, S.M., Welch, D.M., Morrison, H.G., and Sogin, M.L. (2010) Ironing out the wrinkles in
the rare biosphere through improved OTU clustering. Environ Microbiol 12: 1889-1898.
Igarashi, T., Satake, M., and Yasumoto, T. (1999) Structures and partial stereochemical
assignments for Prymnesin-1 and Prymnesin-2: potent hemolytic and ichthyotoxic
86

glycosides isolated from the red ride alga Prymnesium parvum. Journal of the American
Chemical Society 121: 8499-8511.
James, T.Y., Letcher, P.M., Longcore, J.E., Mozley-Standridge, S.E., Porter, D., Powell, M.J.,
Griffith, G.W., and Vilgalys, R. (2006) A molecular phylogeny of the flagellated fungi
(Chytridiomycota) and description of a new phylum (Blastocladiomycota). Mycologia 98:
860-871.
Jobard, M., Rasconi, S., and Sime-Ngando, T. (2010) Diversity and functions of microscopic
fungi: a missing component in pelagic food webs. Aquatic Sciences - Research Across
Boundaries 72: 255-268.
Jones, A.J., Liao, V.T.S., Najar, F.Z., Roe, B.A., Hambright, D.K., and Caron, D.A. (2012)
Seasonality and disturbance: annual pattern and response of the bacterial and microbial
eukaryotic assemblages in a freshwater ecosystem. subb.
Jones, M.D., Forn, I., Gadelha, C., Egan, M.J., Bass, D., Massana, R., and Richards, T.A. (2011)
Discovery of novel intermediate forms redefines the fungal tree of life. Nature 474: 200-
203.
Jones, S.E., C., C., Kratz, T.K., Wu, J.T., Shadeand, A., and and McMahon, K.D. (2008)
Typhoons initiate predictable change in aquatic bacterial communities. Limnol Oceanogr
53: 1319-1326.
Kan, J., Byron, C.C., Wang, K., and Chen, F. (2006) Bacterioplankton community in
Chesapeake Bay: predictable or random assemblages. Limnol Oceanogr 51: 2157-2169.
Lefevre, E., Bardot, C., Noel, C., Carrias, J.F., Viscogliosi, E., Amblard, C., and Sime-Ngando,
T. (2007) Unveiling fungal zooflagellates as members of freshwater picoeukaryotes:
87

evidence from a molecular diversity study in a deep meromictic lake. Environ Microbiol
9: 61-71.
Lefèvre, E., Roussel, B., Amblard, C., and Sime-Ngando, T. (2008) The molecular diversity of
freshwater picoeukaryotes reveals high occurrence of putative parasitoids in the plankton.
PLoS One 3: e2324.
Lim, E.L., Dennett, M.R., and Caron, D.A. (1999) The ecology of Paraphysomonas imperforata
based on studies employing oligonucleotide probe identification in coastal water samples
and enrichment cultures. Limnol Oceanogr 44: 37-51.
Martin-Cereceda, M., G, N., and Young, J.R. (2003) Grazing by Prymnesium parvum on small
planktonic diatoms. Aquat Microb Ecol 33: 191-199.
Medinger, R., Nolte, V., Pandey, R.V., Jost, S., Ottenwalder, B., Schlotterer, C., and Boenigk, J.
(2010) Diversity in a hidden world: potential and limitation of next-generation
sequencing for surveys of molecular diversity of eukaryotic microorganisms. Molecular
ecology 19 Suppl 1: 32-40.
Michaloudi, E., Moustaka-Gouni, M., Gkelis, S., and Pantelidakis, K. (2008) Plankton
community structure during an ecosystem disruptive algal bloom of Prymnesium parvum.
J Plankton Res 31: 301-309.
Nejstgaard, J.C., and Solberg, P.T. (1996) Repression of copepod feeding and fecundity by the
toxic haptophyte Prymnesium patelliferum. Sarsia 81: 339-344.
Newton, R.J., Jones, S.E., Eiler, A., McMahon, K.D., and Bertilsson, S. (2011) A guide to the
natural history of freshwater lake bacteria. Microbiol Mol Biol Rev 75: 14-49.
88

Oikonomou, A., Katsiapi, M., Karayanni, H., Moustaka-Gouni, M., and Kormas, K.A. (2012)
Plankton microorganisms coinciding with two consecutive mass fish kills in a newly
reconstructed lake. TheScientificWorldJournal 2012: 504135.
Peura, S., Eiler, A., Hiltunen, M., Nykanen, H., Tiirola, M., and Jones, R.I. (2012) Bacterial and
phytoplankton responses to nutrient amendments in a boreal lake differ according to
season and to taxonomic resolution. PLoS One 7: 38552.
Pinhassi, J.a., and Hagström, Å. (2000) Seasonal succession in marine bacterioplankton. Aquat
Microb Ecol 21: 245–256.
Remmel, E.J., Kohmescher, N., Larson, J.H., and Hambright, K.D. (2011) An experimental
analysis of harmful algae-zooplankton interactions and the ultimate defense. Limnol
Oceanogr 56: 461-470.
Reynolds, C.S., Padisak, J.a., and Sommer, U. (1993) Intermediate disturbance in the ecology of
phytoplankton and the maintenance of species diversity: a synthesis. Hydrobiologia 249:
183-188.
Roelke, D.L., Grover, J.P., Brooks, B.W., Glass, J., Buzan, D., Southard, G.M., Fries, L., Gable,
G.M., Schwierzke-Wade, L., Byrd, M., et al. (2010) A decade of fish-killing Prymnesium
parvum blooms in Texas: roles of inflow and salinity. J Plankton Res 33: 243-253.
Schloss, P.D., Westcott, S.L., Ryabin, T., Hall, J.R., Hartmann, M., Hollister, E.B., Lesniewski,
R.A., Oakley, B.B., Parks, D.H., Robinson, C.J., et al. (2009) Introducing mothur: open-
source, platform-independent, community-supported software for describing and
comparing microbial communities. Appl Environ Microbiol 75: 7537-7541.
Shade, A., Chiu, C.-Y., and McMahon, K. (2010) Seasonal and Episodic Lake Mixing Stimulate
Differential Planktonic Bacterial Dynamics. Environ Microbiol 59: 546-554.
89

Shade, A., Read, J.S., Youngblut, N.D., Fierer, N., Knight, R., Kratz, T.K., Lottig, N.R., Roden,
E.E., Stanley, E.H., Stombaugh, J., et al. (2012) Lake microbial communities are resilient
after a whole-ecosystem disturbance. ISME J.
Skovgaard, A., and Hansen, P.J. (2003) Food uptake in the harmful alga Prymnesium parvum
mediated by excreted toxins. Limnol Oceanogr 48: 1161-1166.
Sogin, M.L., Morrison, H.G., Huber, J.A., Welch, D.M., Huse, S.M., Neal, P.R., Arrieta, J.M.,
and Herndl, G.J. (2006) Microbial diversity in the deep sea and the underexplored "rare
biosphere". PNAS USA 103: 12115-12120.
Sunda, W.G., Graneli, E., and Gobler, C.J. (2006) Positive feedback and the development and
persistence of ecosystem disruptive algal blooms. J Phycol 42: 963-974.
Teeling, H., Fuchs, B.M., Becher, D., Klockow, C., Gardebrecht, A., Bennke, C.M., Kassabgy,
M., Huang, S., Mann, A.J., Waldmann, J., et al. (2012) Substrate-controlled succession of
marine bacterioplankton populations induced by a phytoplankton bloom. Science 336:
608-611.
Tillmann, U. (1998) Phagotrophy by a plastidic haptophyte, Prymnesium patelliferum. Aquat
Microb Ecol 14: 155-160.
Tillmann, U. (2003) Kill and eat your predator: a winning strategy of the planktonic flagellate
Prymnesium parvum. Aquat Microb Ecol 32: 73-84.
Uronen, P., Kuuppo, P., Legrand, C.a., and T., T. (2007) Allelopathic effects of toxic haptophyte
Prymnesium parvum lead to release of dissolved organic carbon and increase in bacterial
biomass. Environ Microbiol 54: 183-193.
90

Vigil, P., Countway, P.D., Rose, J., Lonsdale, D.J., Gobler, C.J., and Caron, D.A. (2009) Rapid
shifts in dominant taxa among microbial eukaryotes in estuarine ecosystems. Aquat
Microb Ecol 54: 83-100.

Figure Legends
Figure 1: Graphical representations depicting the distributions of the best matches between the
18s (A and C) and 16s (B and D) sequences or OTUs against the SILVA v108 reference
database. Histograms of the percent similarity scores for the entire 18s (A) or 16s (B) OTU
(black bars) or sequence (gray bars) datasets. The Y axis represents the percentage of the total
number of sequences or OTUs. The X axis represents the percent similarity score against the
SILVA reference database from 75% to 100%. N/A stands for no returned similarity value.
Histograms of the percent similarity scores for 18s (C) and 16s (D) sequences within selected
high level taxonomic groups. The colors of the bars represent the binned percent similarity
scores of the best identifications: ≥ 95% (black), 90-95 (light gray), and <90 (dark gray). The Y
axis represents the percentage of the total number of sequences. The X axis represents the most
common high level taxonomic groups within the 18s or 16s datasets.

Figure 2: Relative abundances of sequences (as percents) of high-level taxonomies of 18s rDNA
(A) or 16s rDNA (B) datasets summed over the year (12 months) for Lebanon Pool (black) and
Wilson Creek (gray).

Figure 3: Annual patterns and inter-relationships of (A) metazoan, (B) microbial eukaryotic and
(C) bacterial community structures in Lebanon Pool (black circles) and Wilson Creek (gray
91

triangles) and Wilson Creek December duplicate (gray x) based on two-dimensional non-metric
multidimensional scaling representations of Bray-Curtis similarity values. OTUs were called at
the unique level of similarity and sequences were converted to relative abundances. Clusters of
samples with high average similarity are encircled in dotted lines: metazoa >25%, microbial
eukaryote >25% and bacteria >25% and >35%.

Figure 4: Contributions of high-level taxonomies to the overall simple average Bray-Curtis
percent similarity (black squares) of clusters of samples with high similarity values for A)
metazoan, B) microbial eukaryotic, and C) bacterial assemblages.

Figure 5: Annual distributions of the percentages of 18s rDNA sequences per month aggregated
into high level taxonomic classifications for Lebanon Pool (black circles and lines), Wilson
Creek (gray circles and lines) and Wilson Creek December duplicate (gray triangle). A)
haptophytes, B) fungi, C) stramenopile:chrysophytes, D) alveolate:ciliates, E) cercozoa, F)
metazoa, G) chlorophytes, H) alveolate:dinoflagellates, I) stramenopile:diatoms, and J)
cryptophytes.

Figure 6: Annual distributions of the percentages of 16s rDNA sequences per month aggregated
into high-level taxonomic classifications for Lebanon Pool (black circles and line), Wilson Creek
(gray triangles and lines), and Wilson Creek December duplicate (dark gray triangle). A)
alphaproteobacteria, B) betaproteobacteria, C) gammaproteobacteria, D) bacteroidetes:
sphingobacteria, E) bacteroidetes:flavobacteria, F) verrucomicrobia, G) actinobacteria, and H)
cyanobacteria.
92

Figure 7: Monthly percent contributions of individual groups within selected high-level
taxonomic classifications over the year at Lebanon Pool (A, C, E) and Wilson Creek (B, D and
F). The cumulative percents of the individual groups are presented as colored stacked areas
beneath the totals of their representative high-level classification presented as a simple line plot:
metazoa (A and B), haptophytes (C and D), and cyanobacteria (E and F).

93

94

95

96

98

99

100

101

Supplemental Information
Supplemental Figure 1: Graphical representations depicting the distributions of the best matches
between the 18s (A and C) and 16s (B and D) sequences or OTUs against the GAST (A and B)
or NCBI n/t (C and D) reference databases. Histograms of the 1-distance values for the entire 18s
(A) or 16s (B) OTU (black bars) and sequence (gray bars) datasets against the GAST database.
Histograms of the percent similarity scores for the entire 18s (C) or 16s (D) OTUs (black bars)
and sequence (gray bars) datasets against the NCBI n/t reference database. The Y axis represents
the percentage of the total number of OTUs or sequences. The X axis represents the 1-GAST
distance (A and B) from 0.75 to 1 or the percent similarity score in SILVA (C and D) from 75%
to 100%. N/A stands for no returned value.

Supplemental Figure 2: Hierarchical cluster diagrams of (A) metazoan, (B) microbial eukaryotic
and (C) bacterial community structures in Lebanon Pool (black circles) and Wilson Creek
December duplicate (gray x) based on group average clustering of Bray-Curtis similarity values.
OTUs were called at the unique level of similarity and sequences were converted to relative
abundances. Dotted lines represent clusters of samples that were not significantly different from
one another on the basis of the SIMPROF permutation test. Vertical dotted lines represent the
following group average similarity scores: metazoa >25%, microbial eukaryotes >25% and
bacteria >35%.
102

103

104

Chapter 3
Ecological patterns and relationships among bacteria and microbial eukaryotes derived
from network analyses over an annual cycle in a low salinity lake.

Abstract
Microbial communities are comprised of complex assemblages of highly interactive taxa. To
gain a better understand of how microbial communities are structured, it is important to describe
not only the diversity and composition of microbial communities but also to document patterns
across their connections and interactions, particularly in response to environmental change. We
employed network analyses to quantify and describe microbial interactions and co-occurrence
patterns between microbial eukaryotes (protists, fungi, green plants, and micrometazoa), and
bacteria at two locations within a low salinity (0.5-3.5) lake over an annual cycle. We previously
documented that the microbial community diversity and composition within Lake Texoma,
southwest United States, were significantly structured by both seasonal forces and a localized
biological harmful algal bloom disturbance event: one location experienced a bloom of the toxic
alga Prymnesium parvum that lasted from January to April. The networks revealed highly
interconnected consortia of taxa at both locations within the lake. Patterns of connectivity at both
locations reflected the seasonality of the lake, and a comparison of the winter communities at the
bloom and non-bloom locations revealed a localized response to the algal bloom disturbance
event. We also found conserved interactions at both locations within the lake, and a network
built from shared variables and connections suggest that a core microbial community exists
within the lake.

105

Introduction
Microbial communities form the foundations of aquatic food webs performing fundamental
ecosystem services, such as primary production, and nutrient recycling. (Azam et al., 1983;
Caron and Countway, 2009; Fuhrman, 2009). The microbial interactions between taxa and
higher trophic levels can be positive (symbiotic or mutualistic) (Cole, 1982) or negative such
predation (Sherr and Sherr, 2002) or parasitic (Ibelings et al., 2004). A goal of microbial ecology
is to quantify and describe patterns of interactions within complex assemblages of micro-
organisms, and to measure and document changes in response to environmental forcing features
such as seasonality (Gilbert et al., 2009; Gilbert et al., 2012), natural disturbance events (Jones et
al., 2008; Vigil et al., 2009) and experimental perturbations (Kim et al., 2011; Shade et al.,
2012).
Network analyses facilitate the exploration of many interconnected correlations at once,
and thus are useful tools for quantifying and characterizing complex biological systems (Ings et
al., 2009; Poulin, 2010; Proulx et al., 2005). Network analyses have been used to quantify spatial
and temporal relationships between microbes in natural communities (Barberan et al., 2012;
Eiler et al., 2012; Faust et al., 2012; Fuhrman, 2009; Steele et al., 2011). In this study we used
network analyses to visualize and quantify individual correlations between microbial metazoa,
protists, fungi, chlorophytes, bacteria and environmental parameters and to obtain a measure of
the connectivity and structure of the whole microbial community in a lake known to be affected
by seasonality and disturbance. Lake Texoma is a brackish (PSU 0.5-3.5), temperate lake in the
southwest United States that routinely experiences localized harmful algal blooms of
Prymnesium parvum and temporary salinity fluctuations (Hambright et al., 2010).
Harmful algal blooms are natural biological disturbance events that disrupt microbial
106

communities in aquatic ecosystems (Gobler and Sunda, 2012; Jones et al., 2008). P. parvum
forms Ecosystem Disruptive Algal Blooms (EDABs) that are well documented throughout the
world (Evardsen and Imai, 2006). P. parvum produces a suite of toxins (Henrikson et al., 2010;
Igarashi et al., 1999), has mixotrophic physiology (Tillmann, 1998) and thus can have far
reaching negative effects on taxa from many trophic levels including, gill breathing organisms,
(Evardsen and Imai, 2006), crustaceous zooplankton (Cole, 1982; Nejstgaard and Solberg, 1996),
ciliates and flagellated heterotrophic protists, small diatoms, and filamentous cyanobacteria
(Fistarol et al., 2003; Martin-Cereceda et al., 2003; Skovgaard and Hansen, 2003; Tillmann,
2003).
In a previous study we characterized patterns in microbial eukaryotic and bacterial
diversity measures and community similarity at two locations, Lebanon Pool and Wilson Creek
within Lake Texoma over the course of one year and documented changes related to seasonal
forcing and disturbance: a localized and prolonged biological disturbance event, a bloom of P.
parvum (up to 180,000 cells/ml) that lasted from January to April in Lebanon Pool but not
Wilson Creek. Here, network analyses characterized patterns of taxon-to-taxon interactions, and
revealed distinct features within the community structure related to seasonality and the harmful
algal bloom disturbance of the system.

Methods
Site description and sample collection
Lake Texoma is a temperate, brackish reservoir on the border of Texas and Oklahoma. Two
locations were sampled monthly for one year from November 2008 to October 2009: Lebanon
Pool (L.P.) and Wilson Creek (W.C.) (Jones et al., 2012). Near-shore water samples were
107

collected for measurements of temperature, salinity, dissolved oxygen, pH, extracted chlorophyll
a, Prymnesium parvum cell counts, and DNA sequencing of the eukaryotic (18s) and bacterial
(16s) rRNA genes. The lake experienced two disturbance events during the study: 1) a localized
and prolonged bloom of the toxic alga P. parvum that occurred from January through April in
Lebanon Pool only (Fig. S5A-C), and a shared, episodic, physical disturbance in the form of a
massive spring rain event and subsequent dramatic drop in salinity (from 2.5 to <0.25 PSU)
during May.

Calling and classifying microbial operational taxonomic units (OTUs)
A detailed description of the entire protocol for sample processing, sequence generation,
processing, and OTU calling can be found in Jones et al.(2012) and is briefly outlined here. The
v9 region of the 18s rRNA gene (Amaral-Zettler et al., 2009) and the v6 region of the 16s rRNA
gene (Sogin et al., 2006) were PCR amplified and sequenced via 454 titanium technology
pyrosequencing to assess the diversity of the eukaryotic and bacterial communities. Forty-nine
DNA sequence libraries were generated (two genes, two locations, 12 months). The open source
software mothur v.1.21.1 (http://mothur.org) was used for sequence processing, alignments, and
calling Operation Taxonomic Units (OTUs) (Schloss et al., 2009). The raw reads were edited for
quality (Q-score of ≥25, zero ambiguous bases, <8 homopolymers, exact matches to the primer
sequences (Huse et al., 2007), and a good fit to the SILVA ribosomal small subunit (SSU) 18s or
16s reference alignments. High quality sequences in the 18s-protists, fungi, green plants, and
metazoa (total =277,151) and 16s-bacteria and plastid (total =246,085) datasets were clustered
into OTUs using the single-linkage pre-clustering method (Huse et al., 2010) and an average
neighbor clustering method with a final threshold of 97% similarity. The number of sequences
108

per sample (month/locations) varied from 7,293 to 21,436 in the 18s rDNA dataset and from
3,398 to 20,107 in the 16s rDNA dataset after screening for quality and clustering into OTUs.
The number of sequences per OTU was converted into relative abundances for each sample in
order to standardize our analyses.
A representative sequence, one with the smallest total distance to all other sequences
within an OTU, was extracted from each OTU (12,859 18s rDNA and 16,274 16s rDNA). The
representative sequence was then compared using stand alone BLAST+ (Altschul et al., 1997) to
the SILVA small subunit ribosomal database (SSUv108, http://www.arb-silva.de/) (Pruesse et
al., 2007) and NCBI nt (www.ncbi.nlm.nih.gov) database. SILVA provides a nested taxonomy in
its results (Domain>Kingdom>Phylum>…>genus>…). Each representative sequence was
assigned an identification based on a best match: >95% sequence similarity over >95% of the
query sequence in SILVA. More than 90% of the bacterial sequences and >70% of the
eukaryotic sequences were assigned an ID with two high-level taxonomic exceptions, fungi and
cercozoa. The criteria for fungi and cercozoa were relaxed to >90% sequence similarity because
there are fewer sequenced representatives. A >95% sequence similarity over 99% of the query
match to NCBI nt was used to assign an identification in those cases where a taxonomy in
SILVA could not be defined.

Computing Bray-Curtis similarity values of the microbial community
Bray-Curtis similarity values based on the relative abundances of OTUs for the complete
eukaryotic (including metazoa) and bacterial (without plastids) datasets were computed in order
to measure the eukaryotic and bacterial community level similarity patterns over the year.
Individual Bray-Curtis similarity values for pairs of samples in sequential months at each
109

location, and for pairs of samples in the same month at both locations, were plotted along an
annual trajectory for microbial eukaryotic OTUs and the bacterial OTUs.

Computing correlations and visualizing the connections in network diagrams
Our network analyses were built from Spearman correlations computed over 12 data points (12
months) with each location treated separately. The variables included six environmental
parameters: temperature (Temp), salinity (PSU), pH, dissolved oxygen (DO), extracted
chlorophyll a (Chl), and P. parvum cell counts in cells/ml (Ppar) and 18s rDNA and 16s rDNA
(bacteria plus plastids) OTUs. An OTU was detected in at least four of the 12 months at a given
location and comprise at least 0.5% of the number of sequences during one or more months in
order to be included in the analysis. The relative abundances of the OTUs or absolute quantities
of the environmental parameters were transformed as described in (Ruan et al., 2006) and
pairwise Spearman correlations (using the average tie breaking method), with associated
permuted p-values and q-values (false discovery rates) (Storey, 2002), were computed using a
python script available at http://meta.usc.edu/softs/lsa/ (Xia et al., 2011).
The variables and correlations were imported into the open source data viewing platform
Cytoscape v.2.8.3 for network analyses and visual representations (network diagrams) of all the
significant correlations (edges) between the bacterial OTUs, eukaryotic OTUs and environmental
variables (nodes) (Cline et al., 2007; Shannon et al., 2003; Smoot et al., 2011). Correlations were
represented in the network diagrams if the p-value was significant (≤ 0.01, with associated q-
values of <0.032 for Lebanon Pool or <0.067 for Wilson Creek) (Steele et al., 2011), and when
the absolute value of the Spearman correlation was greater than 0.72. The relative abundances of
the OTU variables were put into eight size bins for visual representations in the network
110

diagrams based on the following criteria: 1) <0.1%, 2) 0.1-0.25%, 3) 0.25-0.5%, 4) 0.5-0.75%, 5)
0.75-1%, 6) 1-1.5%, 7) 1.5-2 %, 8) >2%. The environmental parameters were given the middle
size bin of 4.
The following three overarching networks were constructed: A) Lebanon Pool (222
variables and 4,787 correlations), B) Wilson Creek (247 variables and 3,788 correlations), and
C) a shared dataset, that contained the variables and correlations (of the same quality [positive or
negative], yet not necessarily the same quantity) found at both locations (134 variables and 510
correlations). The shared dataset was constructed using the advanced network merge Cytoscape
plugin.
Networks were visualized using the following four layouts in Cytoscape (Shannon et al.,
2003): 1) "degree-sorted circle layout" where nodes are arranged in a circle in order of those
nodes with the most correlations to other nodes (beginning at 6 o'clock) counter clockwise to the
fewest number of correlations, 2) "attribute circle layout "(nodes are organized in a circle based
on a user-defined attribute, in our case their taxonomic classification), 3) "edge-weighted, spring-
embedded layout" (nodes are strongly repelled or attracted as a function of their Spearman
correlation value, [spring strength = 50, spring rest length = 100, strength of disconnected spring
= 0.05, rest length of disconnected spring = 500, and strength to avoid collisions = 500]), 4) "un-
weighted force-directed layout" (the placement of the nodes is optimized based on the number of
correlations between the nodes and the value of the correlation is not taken into account [spring
coefficient = 1
-5
, spring length = 50, node mass = 3]).
The networks were undirected, meaning correlations between variables did not have
directionality. Overarching statistical properties (network density, average clustering coefficient,
and average shortest path length) of the networks were computed using the Network Analysis
111

Cytoscape plug-in (Assenov et al., 2008). The average network density is a normalized value (0
to 1) reflecting the average number of correlations per variable, and can be used to compare the
degree interconnectivity of different networks of varying sizes. The clustering coefficient (C.I)
for a variable is a ratio (0 to1) relating the number of connections between a variable's neighbors
(a neighbor is defined as a directly connected variable) and the total number of connections
possible between all neighbors of that variable. Values closer to1 indicate a highly
interconnected network, one in which every variable is correlated and connected to every other
variable. The shortest path length (L) for a variable represents the average number of
connections required for that variable to be connected to all other variables via the smallest
number of connections. Values closer to 1 indicate a highly interconnected network, one in
which every variable is directly connected to every other variable. A value of 2 would indicate
that each variable is at least indirectly connected to every other variable through, on average,
only one other variable. Each of our networks were statistically compared to a self-randomized
version, and to an identically sized Erdös-Réyni random model (Erdös and Réyni, 1960;
Shannon et al., 2003) using the Random Networks Cytoscape plug-in
(http://web.ecs.syr.edu/~pjmcswee), in order to establish that the connections we detected were
not generated at random. The log response ratios of the CI and L (observed values: value from
random network) were computed in order to normalize our values and compare the properties of
our networks to those generated in other studies. Log distributions of number of correlations per
variable (also known as degree distributions) of the networks and Erdös-Réyni random models
were plotted and trend lines fitted using SigmaPlot v11.0. The AllegroMCODE Cytoscape plug-
in, which implements the MCODE algorithm (Bader and Hogue, 2003) was used to find and
isolate highly intercorrelated and interconnected clusters of variables within the larger networks.
112

We used the following MCODE parameters: degree cutoff =2, node score cutoff =0.03, K-
score=2, and max depth 100.

Results
Patterns of eukaryotic and bacterial community composition and similarity
The 277,151 18s sequences and 246,085 16s sequences condensed into 12,859 18s OTUs and
16,274 16s OTUs, respectively at 97% similarity. Metazoa comprised 20% of the 18s OTUs and
14% of the 18s sequences. Eukaryotic chloroplasts comprised 5% of the 16s OTUs and 11% of
the 16s sequences.
The annual trajectories for the pairwise Bray-Curtis similarity values between adjacent
months at each location (Fig.1A and B) and the monthly between-site values (Fig.1C and D)
revealed periods of community stability and divergence related to disturbance events within the
lake for the eukaryotic and bacterial (no plastids) assemblages. The plots of the similarity values
between adjacent months in each location for the eukaryotic (Fig.1A), and bacterial (Fig.1B)
datasets revealed periods of high month-to-month similarity during the winter, followed by sharp
decreases between April and May coincident with the timing of the large rain event (from 60%
or 25% to <10% for the eukaryotes and from ~40% to ~15% for the bacteria). Month-to-month
similarities in the eukaryotic and bacterial communities at both locations increased through
August or September suggesting recovery from the physical disturbance. Both the eukaryotic and
bacterial communities at each location reached a second period of high month-to-month (~40%)
similarity in late summer from July to October. The plots of the monthly similarity values
between the locations demonstrated that the eukaryotic (Fig.1C), and bacterial (Fig.1D) datasets
at the two locations were each least similar (<10% for the 18s and from 20-30% for the 16s
113

datasets) from January to April or February to April, coincident with the timing of the P. parvum
bloom in Lebanon Pool. The eukaryotic and bacterial communities at the two locations were
highly similar (~50%) in May coincident with the rain event, and remained similar (30 to 50%)
through October.

Correlation and network analyses revealed highly interconnected microbial systems
OTUs detected in at least four of the 12 months at a given location and comprising at least 0.5%
of the sequences during one or more months were included for the Spearman correlation
analyses. Based on this criteria, the dataset for Lebanon Pool contained 106 18s OTUs including
8 metazoan OTUs, and 128 16s OTUs including 11 plastid OTUs, while the dataset for Wilson
Creek contained 137 18s OTUs including 6 metazoan OTUs, and 130 16s OTUs including 11
plastid OTUs. Both datasets included 6 environmental parameters (see Methods). An OTUs
average abundance did not directly correlate with its frequency of occurrence (Fig. S1A and B)
at either location. We previously demonstrated strong seasonality with the eukaryotic and
bacterial community similarity patterns (Jones et al., 2012). Thus it was not surprising given the
nature of our study site that our variable selection criteria resulted in biological variables that
exhibited seasonality in their annual abundance distributions. A majority of the OTUs selected
for the correlation analyses from both locations (~ 80%) contained more than 75% of their
sequence abundances in either the winter/spring (November to April) or summer/fall (May to
October) periods. See tables S1 and S2 for a complete list of OTUs and environmental variables
and their attributes.
A total of 28,680 (Lebanon Pool) and 37,128 (Wilson creek) pairwise Spearman
correlations were generated. The histograms of the associated p-values (Fig. S2) lacked a
114

uniform distribution, as would be expected under a null model, and instead were skewed towards
zero implying meaningful associations (Ruan et al., 2006). Spearman correlations with p-values
of ≤ 0.01 (associated q-values of <0.032-Lebanon Pool or <0.067-Wilson Creek) were included
in the network analyses. This threshold retained 21% of the correlations (4,787 between 222
variables) in Lebanon Pool and 14% of the correlations (3,788 between 247 variables) in Wilson
Creek. Overall, more positive (56% in Lebanon Pool and 63% in Wilson Creek) than negative
correlations were significant.
The number of significant correlations per variable (Fig. 2 and S3) ranged from 1 to 111
(Lebanon Pool) or from 1 to 88 (Wilson Creek) and did not correlate with a variable's average
relative abundance (Fig. 2), frequency of occurrence (Fig. S3), or taxonomic group (Table S1
and S2). The average number of correlations per variable in Lebanon Pool was 43 (Std. Dev=31)
and the average in Wilson Creek was 31 (Std. Dev=24). There were OTUs with high average
relative abundances and a large number of correlations (Fig. 2A and B open gray boxes) for
example: in Lebanon Pool a Prymnesium plastid (OTU1 6.7% average abundance and 10
occurrences), a diatom plastid (OTU10 3% average abundance and 12 occurrences), and a
SAR11(OTU4 1.8% average abundance and 12 occurrences), each had over 87 correlations. In
addition, we observed variables that fell outside the 95% confidence intervals for number of
correlations and average relative abundances (Fig. 2A and B shaded boxes). Two fungal OTUs,
one in Lebanon Pool and both in Wilson Creek were highly abundant yet had fewer than 5
correlations each (Fig. 2 upper left shaded boxes). In contrast, in Wilson Creek (Fig. 2B lower
right shaded box), two chlorophytes, a cercozoan, and a different fungal OTU each had low
relative abundances but a high (> 70) number of significant correlations.
115

The overwhelming majority (100% and 98%) of the variables and correlations within
Lebanon Pool (Fig. 3A) or Wilson Creek (Fig. 3B) resolved into one large highly interconnected
network. A network constructed from the nodes and correlations shared at both locations
contained 134 variables and 510 correlations (Fig. 3C). The 510 correlations were shared
between 102 of the variables, and represented correlations of the same quality (positive or
negative) independently found in Lebanon Pool and Wilson Creek, although a majority 70% of
the correlations were positive (Fig. 3C).
The number of correlations per variable within each dataset ranged from 1 to 111
(network density of 0.20) in Lebanon Pool (Fig. S4A), from 0 to 85 in Wilson Creek (network
density of 0.13) (Fig. S4B) and from 0 to 31 in the shared network (Fig. S4C) (network density
of 0.06). The distributions representing the frequency of the number of correlations per variable
for the LP, WC, and shared networks (Fig. S4 A-C, closed symbols) were each statistically
different from the Erdös-Réyni model distributions (Fig. S4 A-C, open symbols). The model
distributions each fit Poisson curves as expected with r
2
values of 0.80, 0.87, and 0.87 (Fig. S4A-
C, upper insets) (Barabasi and Oltvai, 2004). The node degree distributions for the LP and WC
networks did not fit power curves as has been found in some (Proulx et al., 2005) but not all
(Lima-Mendez and van Helden, 2009) biological networks. The distribution for the shared
network (Fig. S4C lower inset) had a moderate fit to power curve r
2
= 0.6.
Network analyses provide a picture of all the direct within a biological system. Various
topological properties, which include the average clustering coefficient, and mean shortest path
length (see Methods) describe the patterns of connections found within a network, can be
statistically compared against those from random networks, and can be used as points of
comparison with other experimental networks. The average clustering coefficients (CI = 0.601,
116

0.541, and 0.453) and mean shortest path lengths (L = 2.35, 2.62, and 2.75) of the Lebanon Pool,
Wilson Creek, and shared networks were each significantly different from the values computed
from the Erdös-Réyni random models built from the same number of variables and correlations,
and to self-randomized networks (Table 1).

The microbial networks revealed unique communities along broad seasonal divides
Force-directed, spring-embedded layouts within Cytoscape were used to organize the networks
and position the variables based on the strength and quantity of their correlations. Variables with
negative correlations were visually pulled apart, while those with positive correlations were
placed closely together. A majority (>90%) of the negative correlations within the microbial
association networks in both Lebanon Pool (Fig. 4A) and Wilson Creek (Fig. 4E) separated the
variables into two distinct subclusters. A majority (95%) of the positive correlations were
contained within the two subclusters at both Lebanon Pool (Fig. 4B) and Wilson Creek (Fig. 4F).
The known seasonality patterns (see Methods) of the variables were overlaid on the networks
(Fig. 4 C-D and G-H yellow highlights) and indicated that the two subclusters composed
primarily of variables with opposite seasonal abundance patterns: 90% of the variables with a
fall/winter (November to April) seasonality comprised one cluster (Fig. 4C and G, yellow
highlighted symbols) and 90% of the variables with a spring/fall (May/October) seasonality
comprised the other (Fig. 4D and H, yellow highlighted symbols). Variables with sequence
abundances more evenly distributed throughout the year were placed in either cluster (data not
shown). The winter/spring subnetwork in Lebanon Pool was smaller (33% of the variables)
compared to the summer/fall subnetwork (Fig. 4 A-D). The number of variables in the two
seasonal subnetworks in Wilson Creek were more equal (46/54% ) (Fig. 4 E-H).
117

Correlations found between algae and their plastids indicated the reliability of the network
approach
Positive correlations between plastid OTUs and photosynthetic alga OTUs highlighted likely
host/organelle pairs in the network (Fig. 5C and D). Each location contained 11 plastid OTUs
and all but one of them could be identified to at least phylum level. The number of total
correlations per plastid OTU ranged from 1 to 101, and they were directly and positively
correlated to eukaryotes, bacteria, metazoa, and environmental parameters (Fig. 5A and B).
The two haptophyte plastids, found in Lebanon Pool (Fig. 5C) were correlated with each
other (0.97), to P. parvum cell counts (0.81), and with two haptophyte OTUs (0.75 to 0.82). The
haptophyte and plastid in Wilson Creek (Fig. 5D) also correlated with each other (0.77 to 0.9),
and the haptophyte plastid correlated with P. parvum cell counts. Among diatom OTUs and
diatom plastids unique relationships emerged. For example, plastid OTU-10, and 18s OTU-14
had singular strong correlations (0.93, and 0.85) in both Lebanon Pool (Fig. 5C) and Wilson
Creek (Fig. 5D). An unidentified plastid OTU-69 had strong correlations with chlorophyte
plastids in both Lebanon Pool (0.90) and Wilson Creek (0.92), as well as strong correlations with
chlorophyte OTUs in both locations (0.87 in Lebanon Pool and 0.95 in Wilson Creek (Fig. 5C
and D).

Positive correlations revealed highly interconnected and distinct microbial assemblages within
seasonal subclusters
The negative correlations between the variables overwhelmingly separated the microbial
communities at each location into two seasonal (winter/spring and summer/fall) sub-assemblages
118

(Fig. 4), therefore we removed the negative correlations from the networks to investigate patterns
of positive associations among taxa within each seasonal subnetwork. The force-directed, spring-
embedded layout (Fig. 6 A and B) optimized the placement of the variables based on the number
and strength of the Spearman correlations values, thus variables with the strongest correlation
values were placed more closely together. The resulting visual representations of the positive
correlations also separated the variables based on their seasonal abundance patterns: in both Fig.
6A and B the cluster on the left was mostly comprised of variables with a winter/spring
seasonality and had network density scores of 0.3 (LP) and 0.2 (WC), while the cluster on the
right was predominantly comprised of variables with a summer/fall seasonal abundance pattern
and each had network density scores of 0.2. The MCODE algorithm identified the densest most
interconnected regions of the networks. The top two scoring networks within Lebanon Pool (CI
of 0.83 and 0.91) and Wilson Creek (CI of 0.89 and 0.87) occurred in each of the seasonal
subclusters (Fig. 6C and D yellow highlights).
The clusters identified with the MCODE algorithm exhibited qualitatively different
compositions of variables that reflected the community level responses to disturbance (a P.
parvum bloom from January to April in Lebanon Pool only) and seasonality (Fig. 7A-F). The
most interconnected region in the Lebanon Pool winter/spring community (Fig. 7A) was
comprised of 41 variables and 618 correlations, and had a clustering coefficient (CI) of 0.8. The
most interconnected region in the Lebanon Pool summer/fall (Fig. 7B) was comprised of 33
variables and 475 correlations and had a clustering coefficient (CI) of 0.9. The most
interconnected region in the Wilson Creek winter/spring community (Fig. 7C) was comprised of
42 variables and 572 correlations, and had a clustering coefficient (CI) of 0.8. The most
119

interconnected region in the Wilson Creek summer/fall community (Fig. 7D) was comprised of
47 variables and 855 correlations, and had a clustering coefficient (CI) of 0.9.
The highly interconnected winter/spring sub-cluster in Lebanon Pool (Fig. 7A and E)
contained the variables representing P. parvum cells counts, Prymnesium (OTU1) and two
Prymnesium related plastids (OTU-1 and OTU-12). These variables had similar annual relative
abundance patterns (Fig. S5A and C), and were highly interconnected with other 8 other
eukaryotic variables including: two fungal OTUs, two chrysophytes OTUs, one chlorophyte
OTUs and two unclassified OTUs (UC-32 and 118). The majority (71%) of this cluster was
composed of bacterial OTUs (32 in total), including two cyanobacteria (a Synechococcus [Syn-2]
plus a filamentous form [Cy-40]), and 16 (50%) alphaproteobacteria. Refer to Table S1 for a
complete list of the OTU identifiers and their taxonomic identifications.
The highly interconnected winter/spring cluster in Wilson Creek (Fig. 7C and F) was
equally composed of eukaryotic and bacterial OTUs (17 each). The eukaryotic community
included P. parvum cell counts, and one Prymnesium plastid OTUs. The abundances of P.
parvum cell counts and Prymnesium related OTUs were much lower in Wilson Creek compared
to Lebanon Pool (Fig. S5B and C). The eukaryotic community in Wilson Creek contained three
haptophyte, four chlorophyte, one chrysophyte, four cercozoan, two ciliate and two metazoan
OTUs. The bacterial community included a diversity of phylum-level groups, but no
cyanobacterial OTUs. Refer to Table S2 for a complete list of the OTU identifiers and their
taxonomic identifications.
The highly interconnected spring/summer communities in Lebanon Pool (Fig. 7B and G)
and Wilson Creek each were comprised of approximately the same number of eukaryotic and
bacterial OTUs (Fig. 7D and H). The eukaryotic community in Lebanon Pool included three
120

identified cryptophytes, two chlorophytes, one diatom plus a diatom plastid, four ciliates, and
two metazoan OTUs. The bacterial community in Lebanon Pool contained a diversity of taxa,
including two cyanobacteria (Fig. 7G). The eukaryotic community in Wilson Creek included
three diatoms, one cryptophyte, six chlorophytes, three ciliates, three cercozoan OTUs. The
bacterial community in Wilson Creek was dominated by cyanobacteria OTUs (50%) (Fig. 7H).

Conserved correlations in the shared network revealed a core microbial community in the lake
A total of 134 microbial OTUs, six environmental variables, and 510 interactions (352 positive)
were shared between the datasets from Lebanon Pool and Wilson Creek (Fig. 1C and Fig.8). The
510 interactions were contained across 102 variables (Fig. 8A). The edge-weighted, spring-
embedded layout was applied to the shared network using the signs of the correlations as the
repulsion/ attraction weight to position the variables (Fig. 8A). This approach resulted in two
subclusters of 32 and 63 variables separated by 95% of the negative correlations. The two
clusters contained variables with opposite patterns of seasonal abundance. The smaller cluster on
the left was comprised predominantly (90%) of nodes with higher abundances from November to
April, and the larger cluster on the right was comprised predominantly (81%) of nodes with
higher abundances from May to October.
The attribute circle layouts built from only the positive correlations within the two
seasonal clusters revealed relationships between taxonomic groups (Fig. 8 B and C). The smaller,
winter/spring subnetwork (Fig. 8B) was 70% composed of bacterial OTUs including 10
alphaproteobacteria, and four betaproteobacteria, and contained only six shared eukaryotic OTUs
which included a Prymnesium and plastid OTU. The alphaproteobacteria were the most highly
connected winter/spring group and constituted 66% of the correlations (Fig.8D). The larger
121

summer/fall subnetwork (Fig. 8C) contained an equal proportion of eukaryotic and bacterial
OTUs, including four diatoms, six ciliates, seven cryptophytes, and four plastids. This cluster
also contained 11 cyanobacteria OTUs including both Synechococcus and filamentous forms.
The cyanobacteria were the most highly connected spring/summer group and contained 59% of
the total connections (Fig. 8E).

Discussion
Network analyses are powerful applications for characterizing complex multi-member biological
systems (Proulx et al., 2005). Microbial assemblages are often comprised of a high diversity of
taxa with seemingly overlapping niches, that interact in complex ways thus network approaches
are a natural tool for quantifying and visualizing microbial occurrence patterns (Barberan et al.,
2012) and relationships (Eiler et al., 2012; Steele et al., 2011). The microbial eukaryotic and
bacterial assemblages in this study, exhibited seasonal patterns in community composition and
similarity (Fig. 1 A and B) over the year, and a prolonged (4-month) site-specific difference (Fig.
1 C and D) in community similarity directly related to a P. parvum algal bloom. (Jones et al.
2012). The same dataset was employed in this study to compare and contrast network properties
and identify specific microbe-to-microbe or microbe-to-environmental interactions related to
both seasonality and a biological disturbance event.

Ecological connections between the microbes and to their environment
Our network representations of the microbial assemblages of Lebanon Pool and Wilson Creek
represented non-random associations (Table 1) and describe microbial communities composed of
organisms with strong interactions and correlations and also demonstrate coordinated responses
122

to environmental forcing. Anticipated correlations between plastid OTUs and their eukaryotic
hosts which demonstrated the utility and ability of the networks to identify biological and
ecological correlations (Fig. 5).
The average clustering coefficient (CI) and average shortest path lengths (L) are two
examples of network statistics (Table 1) that describe the overarching features in a network
which can be used as points of comparison to other networks. The clustering coefficient is a local
property that represents the interconnectivity of the neighbors of a particular variable. Our
networks in Lebanon Pool, Wilson Creek, and the shared dataset, had high (0.6, 0.54, and 0.45)
average clustering coefficients, indicating that on average more than half of the neighbors of a
variable were also connected. Networks can also have regions or particular variables with very
high clustering coefficients. The MCODE algorithm identified highly interconnected subsets of
variables (CI = 0.83 to 0.91) within our networks (Fig. 6C and D yellow highlight and Fig. 7A-
D). The path length for a given variable is a global property, and it measures the average number
of connections to link a given variable with all other variables in the network. A path length of
one reflects a direct connection between two variables, a path length of two reflects an indirect
connections, through one variable. Our networks in Lebanon Pool, Wilson Creek, and the shared
dataset, had low (2.3, 2.6, and 2.7) average path lengths indicating that, on average, each variable
in our networks was connected to every other variable in the network through correlations
involving less than two other variables. The log response ratios of the average clustering
coefficients (1.1 to 1.5) and the mean shortest path length (0.21 to 0.32) are comparable to those
reported for the multi-domain microbial ecological network from an open ocean environment
(Steele et al., 2011). The low average path lengths and high clustering coefficients of the
networks together suggest that our networks have "small world properties" (Watts and Strogatz,
123

1998), meaning that the variables represent highly interconnected consortia of microbes, both
eukaryotes and bacteria with strong associations and interactions.

Networks reflected the natural history of the lake
The natural history (strong seasonality, and disturbance events) of the two locations in this study
provided an interesting framework for our network analyses. Both sampling locations exhibited
strong seasonal trends in community composition that reflected forces from both gradual
seasonal environmental changes and a physical disruption in the form of a massive spring rain
event during May (Jones et al., 2012) at both locations. The eukaryotic and bacterial
assemblages in both locations had two periods of community stability (winter/spring)
(summer/fall) defined by stretches of relatively high month-to-month similarity (Fig. 1A and B).
The seasonality of the system at each location was "organically" re-enforced by force-
directed spring-embedded visualizations of the networks (Fig. 4). The negative correlations in
our networks overwhelmingly reflected the opposite seasonal abundance patterns of the
variables, and two distinct (winter/spring and summer/fall) subnetworks of positively correlated
variables emerged from all the correlations in the lake over the year. Our monthly sampling
resolution was sufficient to resolve the seasonal dichotomy through our network analyses.
However, negative relationships could also indicate predator/prey or parasitic infection type
relationships, although such oscillations would be expected to happen over a shorter time scale.
The microbial assemblages in Lake Texoma were also strongly impacted by a prolonged
biological disturbance event: a bloom of the toxic alga P. parvum from January through April in
Lebanon Pool but not Wilson Creek (Jones et al., 2012). P. parvum is a classified as an
Ecosystem Disruptive Algal Bloom (EDAB) (Gobler and Sunda, 2012) and has demonstrated
124

direct negative effects on many members of the plankton (Martin-Cereceda et al., 2003;
Skovgaard and Hansen, 2003; Tillmann, 2003). The eukaryotic and bacterial communities had
low similarity values between the two locations during the time-frame of the bloom and this
translated into different community compositions and patterns of interaction during the
winter/spring season at the two locations (Fig. 1C and D and Fig. 7A, C, E and F).
The magnitude of the P. parvum bloom structured the community in Lebanon Pool and
appeared to disrupt the overall diversity of taxa and connections when compared to the
composition of the non-bloom community in Wilson Creek (Fig. 7C vs F). P. parvum related
variables (OTU-1, two plastid OTUs, and P. parvum cell counts) accounted for many of
correlations within the most highly interconnected subset of variables during the winter/spring.
This highly interconnected consortium in Lebanon Pool contained only 8 other eukaryotic OTUs,
including two fungi and three chrysophytes. A majority of the bacterial OTUs were
alphaproteobacteria (Fig.7E). The composition of and connections within in Lebanon Pool were
in stark contrast to the makeup of the most highly interconnected winter/spring subcluster within
Wilson Creek. P. parvum cells and OTUs were present over the winter in Wilson creek (Fig.
S7B and C), yet abundances remained at non-bloom conditions. The most highly connected
microbial community within Wilson Creek (Fig. 7F) during the winter/springs contained 17
eukaryotes including cercozoa, ciliates, chlorophytes, and a crustacean metazoan, as well as a
diversity of bacterial phyla.
In addition to the compositional differences between the two locations the ecosystem
disruptive nature of P. parvum may have been apparent within the properties of the networks
themselves. The biological disruption in Lebanon Pool may also have been the reason for the
smaller winter/spring subnetwork (75 variables) (Fig. 4A-D and 6A) compared to the
125

summer/fall (145 variables) subnetwork especially given that the seasonal subnetworks in
Wilson Creek were equal in size (121 and 125) (Fig. 4E-H, and 6B). The log response ratio of
the clustering coefficient in Lebanon Pool (Table. 1) was much lower (1.12) than the ratio in
Wilson Creek (1.47) or the joint network (1.51), indicating that on average, for the entire
network fewer of the neighboring nodes were also connected to each other. However, on a local
scale P. parvum was part of a highly interconnected region within the Lebanon Pool network
(Fig. 7A and E). The presence of P. parvum may select for a small number of unique taxa, that as
a suite are not interconnected to the taxa that occur under non-bloom conditions in Lebanon
Pool.

The shared network revealed a core set of shared taxa and connections in the Lake
Variables representing identical OTUs or environmental parameters and correlations of the same
quality (positive or negative) were represented in the shared network (Fig. 3C and Fig. 8). The
shared negative correlations separated the community along seasonal abundance patterns (Fig.
8A). The differences in community compositions (Fig. 1) between the two locations in the winter
were reflected in a smaller shared winter/spring network (Fig. 8B) compared to summer/fall (Fig.
8C). Overall the alphaproteobacteria formed a highly interconnected consortium common to both
locations in the winter (Fig. 8D) with very few eukaryotic connections common to both
locations. Several ciliates, cryptophytes, chlorophytes and cyanobacteria each had shared
interactions found in the summer/fall at both locations (Fig. 8E).
The shared relationships suggest a core microbial community within the lake, that
responds in parallel to seasonal environmental forcing and persists through the biological P.
parvum algal bloom disturbance. P. parvum radically alters the lake ecosystem leading to
126

contrasting community compositions and relationships. The shared community was comprised of
a small proportion of the organisms investigated in this study and an even smaller proportion of
the possible correlations suggesting that local differences (including disturbance events) may
govern many of the detectable interactions and patterns between the microbes.

References
Altschul, S.F., Madden, T.L., Schäffer, A.A., Zhang, J., Zhang, Z., Miller, W., and Lipman, D.J.
(1997) Gapped BLAST and PSI-BLAST: a new generation of protein database search
programs. Nucleic Acids Res 25: 3389-3402.
Amaral-Zettler, L.A., McCliment, E.A., Ducklow, H.W., and Huse, S.M. (2009) A method for
studying protistan diversity using massively parallel sequencing of V9 hypervariable
regions of small-subunit ribosomal RNA genes. PLoS One 4: e6372.
Assenov, Y., Ramírez, F., Schelhorn, S.-E., Lengauer, T., and Albrecht, M. (2008) Computing
topological parameters of biological networks. Bioinformatics 24: 282-284.
Azam, F., Fenchel, T., Field, J.G., Gray, S.J., L.A., M.-R., and F., T. (1983) The ecological role
of water-column microbes in the sea. Mar Ecol Prog Ser 10: 257-263.
Bader, G., and Hogue, C. (2003) An automated method for finding molecular complexes in large
protein interaction networks. BMC bioinformatics 4: 2.
Barabasi, A.-L., and Oltvai, Z.N. (2004) Network biology: understanding the cell's functional
organization. Nat Rev Genet 5: 101-113.
Barberan, A., Bates, S.T., Casamayor, E.O., and Fierer, N. (2012) Using network analysis to
explore co-occurrence patterns in soil microbial communities. ISME J 6: 343-351.
127

Caron, D.A., and Countway, P.D. (2009) Hypotheses on the role of the protistan rare biosphere
in a changing world. Aquat Microb Ecol 57: 227-238.
Cline, M.S., Smoot, M., Cerami, E., Kuchinsky, A., Landys, N., Workman, C., Christmas, R.,
Avila-Campilo, I., Creech, M., Gross, B., et al. (2007) Integration of biological networks
and gene expression data using Cytoscape. Nat Protoc 2: 2366-2382.
Cole, J.J. (1982) Interactions between bacteria and algae in aquatic ecosystems. Annual Review
of Ecology and Systematics 13: 291-314.
Eiler, A., Heinrich, F., and Bertilsson, S. (2012) Coherent dynamics and association networks
among lake bacterioplankton taxa. ISME J 6: 330-342.
Erdös, P.a., and Réyni, A. (1960) On the evolution of random graphs. Publ Math Inst Hung Acad
5: 17-61.
Evardsen, B., and Imai, I. (2006). The ecology of harmful flagellates within Prymnesiophyceae
and Raphidophyceae. In Ecology of Harmful Algae, E. Granéli, and J. Turner, eds.
(Netherlands: Springer-Verlag), pp. 67-79.
Faust, K., Sathirapongsasuti, J.F., Izard, J., Segata, N., Gevers, D., Raes, J., and Huttenhower, C.
(2012) Microbial co-occurrence relationships in the human microbiome. PLoS Computer
Biology 8: e1002606.
Fistarol, G.C., Legrand C., and E, G. (2003) Allelopathic effect of Prymnesium parvum on a
natural plankton community. Mar Ecol Prog Ser 255: 115–125.
Fuhrman, J.A. (2009) Microbial community structure and its functional implications. Nature
459: 193-199.
128

Gilbert, J.A., Field, D., Swift, P., Newbold, L., Oliver, A., Smyth, T., Somerfield, P.J., Huse, S.,
and Joint, I. (2009) The seasonal structure of microbial communities in the Western
English Channel. Environ Microbiol 11: 3132-3139.
Gilbert, J.A., Steele, J.A., Caporaso, J.G., Steinbruck, L., Reeder, J., Temperton, B., Huse, S.,
McHardy, A.C., Knight, R., Joint, I., et al. (2012) Defining seasonal marine microbial
community dynamics. ISME J 6: 298-308.
Gobler, C.J., and Sunda, W.G. (2012) Ecosystem disruptive algal blooms of the brown tide
species, Aureococcus anophagefferens and Aureoumbra lagunensis. Harmful Algae 14:
36-45.
Hambright, K.D., Zamor, R.M., Easton, J.D., Glenn, K.L., Remmel, E.J., and Easton, A.C.
(2010) Temporal and spatial variability of an invasive toxigenic protist in a North
American subtropical reservoir. Harmful Algae 9: 568-577.
Henrikson, J.C., Gharfeh, M.S., Easton, A.C., Easton, J.D., Glenn, K.L., Shadfan, M., Mooberry,
S.L., Hambright, K.D., and Cichewicz, R.H. (2010) Reassessing the ichthyotoxin profile
of cultured Prymnesium parvum (golden algae) and comparing it to samples collected
from recent freshwater bloom and fish kill events in North America. Toxicon : official
journal of the International Society on Toxinology 55: 1396-1404.
Huse, S.M., Huber, J.A., Morrison, H.G., Sogin, M.L., and Welch, D.M. (2007) Accuracy and
quality of massively parallel DNA pyrosequencing. Genome Biol 8: R143.
Huse, S.M., Welch, D.M., Morrison, H.G., and Sogin, M.L. (2010) Ironing out the wrinkles in
the rare biosphere through improved OTU clustering. Environ Microbiol 12: 1889-1898.
129

Ibelings, B.W., De Bruin, A., Kagami, M., Rijkeboer, M., Brehm, M., and Donk, E.V. (2004)
Host parasite interactions between freshwater phytoplankton and chytrid fungi
(Chytridiomycota). J Phycol 40: 437-453.
Igarashi, T., Satake, M., and Yasumoto, T. (1999) Structures and partial stereochemical
assignments for Prymnesin-1 and Prymnesin-2: potent hemolytic and ichthyotoxic
glycosides isolated from the red ride alga Prymnesium parvum. Journal of the American
Chemical Society 121: 8499-8511.
Ings, T.C., Montoya, J.M., Bascompte, J., Blüthgen, N., Brown, L., Dormann, C.F., Edwards, F.,
Figueroa, D., Jacob, U., Jones, J.I., et al. (2009) Review: Ecological networks – beyond
food webs. Journal of Animal Ecology 78: 253-269.
Jones, A.J., Liao, V.T.S., Najar, F.Z., Roe, B.A., Hambright, D.K., and Caron, D.A. (2012)
Seasonality and disturbance: annual pattern and response of the bacterial and microbial
eukaryotic assemblages in a freshwater ecosystem. subb.
Jones, S.E., C., C., Kratz, T.K., Wu, J.T., Shadeand, A., and and McMahon, K.D. (2008)
Typhoons initiate predictable change in aquatic bacterial communities. Limnol Oceanogr
53: 1319-1326.
Kim, D.Y., Countway, P.D., Gast, R.J., and Caron, D.A. (2011) Rapid shifts in the structure and
composition of a protistan assemblage during bottle incubations affect estimates of total
protistan species richness. Microb Ecol 62: 383-398.
Lima-Mendez, G., and van Helden, J. (2009) The powerful law of the power law and other
myths in network biology. Molecular BioSystems 5: 1482-1493.
Martin-Cereceda, M., G, N., and Young, J.R. (2003) Grazing by Prymnesium parvum on small
planktonic diatoms. Aquat Microb Ecol 33: 191-199.
130

Nejstgaard, J.C., and Solberg, P.T. (1996) Repression of copepod feeding and fecundity by the
toxic haptophyte Prymnesium patelliferum. Sarsia 81: 339-344.
Poulin, R. (2010) Network analysis shining light on parasite ecology and diversity. Trends in
Parasitology 26: 492-498.
Proulx, S.R., Promislow, D.E., and Phillips, P.C. (2005) Network thinking in ecology and
evolution. Trends Ecol Evol 20: 345-353.
Pruesse, E., Quast, C., Knittel, K., Fuchs, B.M., Ludwig, W., Peplies, J., and Glöckner, F.O.
(2007) SILVA: a comprehensive online resource for quality checked and aligned
ribosomal RNA sequence data compatible with ARB. Nucleic Acids Res 35: 7188-7196.
Ruan, Q., Dutta, D., Schwalbach, M.S., Steele, J.A., Fuhrman, J.A., and Sun, F. (2006) Local
similarity analysis reveals unique associations among marine bacterioplankton species
and environmental factors. Bioinformatics 22: 2532-2538.
Schloss, P.D., Westcott, S.L., Ryabin, T., Hall, J.R., Hartmann, M., Hollister, E.B., Lesniewski,
R.A., Oakley, B.B., Parks, D.H., Robinson, C.J., et al. (2009) Introducing mothur: open-
source, platform-independent, community-supported software for describing and
comparing microbial communities. Appl Environ Microbiol 75: 7537-7541.
Shade, A., Read, J.S., Youngblut, N.D., Fierer, N., Knight, R., Kratz, T.K., Lottig, N.R., Roden,
E.E., Stanley, E.H., Stombaugh, J., et al. (2012) Lake microbial communities are resilient
after a whole-ecosystem disturbance. ISME J.
Shannon, P., Markiel, A., Ozier, O., Baliga, N.S., Wang, J.T., Ramage, D., Amin, N.,
Schwikowski, B., and Ideker, T. (2003) Cytoscape: a software environment for integrated
models of biomolecular interaction networks. Genome Res 13: 2498-2504.
131

Sherr, E., and Sherr, B. (2002) Significance of predation by protists in aquatic microbial food
webs. Antonie van Leeuwenhoek 81: 293-308.
Skovgaard, A., and Hansen, P.J. (2003) Food uptake in the harmful alga Prymnesium parvum
mediated by excreted toxins. Limnol Oceanogr 48: 1161-1166.
Smoot, M.E., Ono, K., Ruscheinski, J., Wang, P.-L., and Ideker, T. (2011) Cytoscape 2.8: new
features for data integration and network visualization. Bioinformatics 27: 431-432.
Sogin, M.L., Morrison, H.G., Huber, J.A., Welch, D.M., Huse, S.M., Neal, P.R., Arrieta, J.M.,
and Herndl, G.J. (2006) Microbial diversity in the deep sea and the underexplored "rare
biosphere". PNAS USA 103: 12115-12120.
Steele, J.A., Countway, P.D., Xia, L., Vigil, P.D., Beman, J.M., Kim, D.Y., Chow, C.E.,
Sachdeva, R., Jones, A.C., Schwalbach, M.S., et al. (2011) Marine bacterial, archaeal and
protistan association networks reveal ecological linkages. ISME J 5: 1414-1425.
Storey, J.D. (2002) A direct approach to false discovery rates. Journal of the Royal Statistical
Society: Series B (Statistical Methodology) 64: 479-498.
Tillmann, U. (1998) Phagotrophy by a plastidic haptophyte, Prymnesium patelliferum. Aquat
Microb Ecol 14: 155-160.
Tillmann, U. (2003) Kill and eat your predator: a winning strategy of the planktonic flagellate
Prymnesium parvum. Aquat Microb Ecol 32: 73-84.
Vigil, P., Countway, P.D., Rose, J., Lonsdale, D.J., Gobler, C.J., and Caron, D.A. (2009) Rapid
shifts in dominant taxa among microbial eukaryotes in estuarine ecosystems. Aquat
Microb Ecol 54: 83-100.
Watts, D.J., and Strogatz, S.H. (1998) Collective dynamics of "small-world" networks. Nature
393: 440-442.
132

Xia, L., Steele, J., Cram, J., Cardon, Z., Simmons, S., Vallino, J., Fuhrman, J., and Sun, F. (2011)
Extended local similarity analysis (eLSA) of microbial community and other time series
data with replicates. BMC Systems Biology 5: S15.
133

134

Figure Legends
Figure 1: Annual distributions of pairwise Bray-Curtis percent similarities of microbial
communities calculated from OTUs called at 97% similarity and sequences converted into
relative abundances. Annual cycles of Bray-Curtis similarities between adjacent months in
Lebanon Pool (black circles and lines) and Wilson Creek (gray triangles and lines), for A)
eukaryotic, and B) bacterial assemblages. Annual cycles of the pairwise Bray-Curtis similarities
between the two locations, Lebanon Pool and Wilson Creek for each month for C) eukaryotic,
and D) bacterial assemblages.

Figure 2: The frequency on a log scale of a microbial OTUs relative abundance plotted against
its number of significant Spearman correlations with other OTUs or environmental variables, for
the data from Lebanon Pool (A) and Wilson Creek (B). Lines indicate the 95% confidence
intervals for the number of correlations per OTU, and its relative abundances. The boxes
highlight OTUs of interest: 1) the open boxes outlines in gray, in the upper right contain OTUs
with large (outside the 95% CI) relative abundances and numbers of significant correlations, 2)
the shaded boxes on the upper left contain OTUs with large (outside the 95% CI) relative
abundances and small (outside the 95% CI) numbers of significant correlations, and 3) the
shaded box on the lower right contains OTUs with small (outside the 95% CI) relative
abundances and large (outside the 95% CI) numbers of significant correlations.

Figure 3: Network diagrams depicting the microbial OTUs and environmental variables and
spearman correlations between the OTUs and environmental variables for the complete datasets
from A) Lebanon Pool (222 variables and 4,787 correlations), B) Wilson Creek (247 variables
135

and 3,788 correlations), and C) the shared dataset which contains the variables and correlations
common to both locations (134 variables and 510 correlations). The networks were visualized
using the following layouts in Cytoscape (A and B) the degree-sorted circle layout (the variables
were arranged in a circle beginning with the variable with the largest number correlations to
other variables (6 o'clock) counter clockwise to variable with the fewest number of correlations,
and (C) the attribute circle layout (variables were organized in a circle based on their taxonomic
classification). Connections drawn from positive Spearman correlations (>0.72 and p-values ≤
0.01) are black solid lines, and those from negative correlations (<-0.72 and p-values ≤ 0.01) are
gray dotted lines. Bacteria are red circles, eukaryotes are blue or purple (metazoa) diamonds,
environmental parameters are orange squares, and chloroplasts are green circles. The size of the
symbols in panels A and B reflects the average relative abundance. The shared network does not
contain information about the relative abundances of the variables.

Figure 4: Spearman correlations between the OTU variables within the network diagrams
revealed contrasting seasonal abundance patterns among the microbial taxa from Lebanon Pool
(A-D) and Wilson Creek (E-H). The networks were visualized with the edge-weighted, spring-
embedded layout (variables in the network were positioned based on the strength and sign of
their Spearman correlation values). OTU variables with 75% or more of their relative
abundances contained in the six month period of November through April (C and G) or May
through October (D and H) are highlighted in yellow. Connections drawn from positive
Spearman correlations are black solid lines, and those from negative correlations are gray dotted
lines. All correlations (4,787 and 3,788 [>0.72 or <-0.72 and p-values ≤ 0.01]) are displayed in
panels A, C, D, E, G and H. Only positive correlations (2,669 and 2,403) are displayed in panels
136

B and F. Bacteria are red circles, eukaryotes are blue or purple (metazoa) diamonds,
environmental parameters are orange squares, and chloroplasts are green circles.

Figure 5: Network representations of the plastid OTUs and their positive spearman correlations
to other OTUs and environmental variables from Lebanon Pool (A and C) and Wilson Creek (B
and D). Network diagrams were visualized with the force-directed layout (variables in the
network were positioned based on the strength of their Spearman correlation values). All direct
positive correlations between the plastids and all other variables (bacteria and eukaryotic OTUs
plus environmental variables) are shown for the data from Lebanon Pool (A) and Wilson Creek
(B). Selected individual positive correlations between a plastid OTUs and its likely
photosynthetic eukaryotic variable (18s OTUs or P. parvum cell count variable) are shown for
the data from Lebanon Pool (C) and Wilson Creek (D). Connections represent positive Spearman
correlations (>0.72 and p-values ≤ 0.01) and the exact values are written on the lines in C and D.
Bacteria are red circles, eukaryotes are blue or purple (metazoa) diamonds, environmental
parameters are orange squares, and chloroplasts are green circles. The size of the symbol reflects
the average relative sequence abundance. The number on the symbols refers to the OTU
identifier numbers. The following identification codes were used for the OTUs with good
taxonomic resolution: Hap (haptophyte), Chr (chrysophyte), Chl (chlorophyte), Cry
(cryptophyte), Dia (diatom), and Eug (euglenid), Dict (dictyochophyte), UC (unclassified), Ppar
(P. parvum cell counts).

Figure 6: Networks representing the positive spearman correlations between the microbial OTUs
and environmental variables from Lebanon Pool (A and C) and Wilson Creek (B and D). The
137

networks were visualized with the edge-weighted, spring-embedded layout (variables in the
network were positioned based on the strength their positive Spearman correlation values).
The visual spread in each network divided the variables by broad seasonal distribution patterns,
the left-hand cluster is primarily composed of variables with high abundances in winter/spring,
and the right-hand cluster contains primarily variables with higher abundances in the summer/fall
season. The MCODE algorithm was used to identify regions of highly interconnected clusters of
variables. The top two MCODE clusters in Lebanon Pool (C) and Wilson Creek (D) are
highlighted in yellow. Connections drawn from positive Spearman correlations are black solid
lines. Bacteria are red circles, eukaryotes are blue or purple (metazoa) diamonds, environmental
parameters are orange squares, and chloroplasts are green circles. The size of the nodes reflects
the average relative sequence abundance.

Figure 7: Highly interconnected clusters of variables extracted with the MCODE algorithm from
the networks representing all of the positive spearman correlations from Lebanon Pool (Fig. 6 C)
and Wilson Creek (Fig. 6D). The MCODE algorithm identified two highly interconnected
clusters from each location, and each cluster contained variables predominantly observed in one
season (November through April) or the other (May through October): A and E) winter/spring in
Lebanon Pool, B and G) summer/fall in Lebanon Pool, C and F) winter/spring in Wilson Creek
and D and H) summer/fall in Wilson Creek. The networks were visualized with the edge-
weighted, spring-embedded layout (variables in the network were positioned based on the
strength their positive Spearman correlation values). The size of the nodes reflects the average
relative sequence abundance. Bacteria are red circles, eukaryotes are blue or purple (metazoa)
diamonds, environmental parameters are orange squares, and chloroplasts are green circles.
138

Connections drawn from positive Spearman correlations are black solid lines (A-D). In panels
(E-H) the connections were erased for easier reading of the variable identifiers. The number on
the symbols refers to the OTU identifier numbers. The following identification codes were used
for the OTUs with good taxonomic resolution: Alp (alphaproteobacteria), Bet
(betaproteobacteria), Gam (gammaproteobacteria), Fla (flavobacteria), Sph (sphingobacteria),
Cy(cyanobacteria), Fun (fungi), Hap(haptophyte), Chr (chrysophyte), Chl (chlorophyte), Cer
(cercozoa), Cil (ciliate), Cry (cryptophyte), Dia (diatom), Rot (rotifer), Cru (crustacean), and Pol
(polychaete).

Figure 8: Diagrams representing the microbial OTUs, environmental variables and Spearman
correlations between the variables found in both Lebanon Pool and Wilson Creek. The shared
network in (A) was visualized with the edge-weighted, spring-embedded layout (variables in the
network were positioned based on the sign of their Spearman correlation values) and includes all
shared correlations (510) across 102 variables. Connections drawn from positive Spearman
correlations are connected by black solid lines, and negative correlations are represented by gray
dashed lines. The visual spread from the negative correlations in (A) divided the variables by
broad seasonal abundance patterns. The variables in the left-hand cluster were predominantly
observed from November through April, and the variables in the right-hand cluster were
predominantly observed from May through October. In panels B and C the positive correlations
across each seasonal subnetwork were visualized with the attribute circle layout (variables were
organized in a circle based on their taxonomic classification). In panels D and E the positive
correlations across each seasonal subnetwork were visualized with the un-weighted force-
directed layout (the placement of the variables was optimized based on the number of
139

correlations between the variables and the value of the correlation was not taken into account).
Panels B and D represent the 117 positive correlations across 38 variables from the winter/spring
cluster. Panels C and E represent the 236 positive correlations across 64 variables from the
summer/fall cluster. Bacteria are red circles, eukaryotes are blue or purple (metazoa) diamonds,
environmental parameters are orange squares, and chloroplasts are green circles. Connections
drawn from positive Spearman correlations are connected by black solid lines. The number on
the symbols (B-E) refer to the OTU identifier numbers, and the following three letter
identification codes were used: Hap(haptophyte), Chr (chrysophyte), Chl (chlorophyte), Cry
(cryptophyte), Dia (diatom), and Eug (euglena).
140

141

142

143

144

145

146

147

Table S1. A summary of the variables (18s and 16s rDNA OTUs plus environmental variables), in Lebanon Pool and their number of correlations,
best percent hit to SILVA (v108) reference database, nested taxonomic identifications, average relative abundance, binned abundance,

seasonal abundance: A (November to April), B (May to October), number of occurrences, and location of detection in the Lake Texoma

ID
#
Correlations Gene
Similarity
to SILVA Domain Phylum Class Other
Av
Abund
Binned
Abund
Seasonal
Abund
Abund
Rank Occur Location
LP_16_Otu1 91 Bact 94 Chloroplast Chloroplast Chloroplast Chloroplast 6.74 8 A 2 10 LP and WC
LP_16_Otu10 101 Bact 100 Chloroplast Chloroplast Chloroplast Odontella sinensis 3.14 8 B 5 12 LP and WC
LP_16_Otu100 77 Bact 100 Chloroplast Chloroplast Chloroplast Chloroplast 0.45 3 B 69 8 LP and WC
LP_16_Otu103 40 Bact 97 Bacteria Proteobacteria Deltaproteobacteria Bdellovibrionales 0.2 2 A 148 8 LP
LP_16_Otu104 70 Bact 98 Bacteria Proteobacteria Alphaproteobacteria Rhodospirillales 0.16 2 A 169 7 LP
LP_16_Otu107 49 Bact 98 Bacteria Cyanobacteria SubsectionI Synechococcus sp. EW15 0.21 2 C 143 8 LP and WC
LP_16_Otu108 5 Bact 100 Bacteria Actinobacteria Actinobacteria Frankiales 0.09 1 B 228 6 LP and WC
LP_16_Otu113 14 Bact 98 Bacteria Proteobacteria Alphaproteobacteria Rhodobacterales 0.15 2 A 181 8 LP
LP_16_Otu115 58 Bact 100 Bacteria Actinobacteria Acidimicrobiia Acidimicrobiales 0.25 2 B 124 8 LP and WC
LP_16_Otu118 4 Bact 100 Bacteria Proteobacteria Betaproteobacteria Burkholderiales 0.16 2 B 168 8 LP
LP_16_Otu119 2 Bact 100 Bacteria Actinobacteria Actinobacteria Frankiales 0.12 2 B 206 7 LP
LP_16_Otu12 90 Bact 97 Chloroplast Chloroplast Chloroplast Chloroplast 1.03 6 A 20 8 LP
LP_16_Otu120 52 Bact 97 Bacteria Cyanobacteria SubsectionIV SubgroupI 0.19 2 B 154 4 LP and WC
LP_16_Otu123 3 Bact 100 Bacteria Bacteroidetes Flavobacteria Flavobacteriales 0.16 2 B 170 9 LP
LP_16_Otu124 3 Bact 100 Bacteria Bacteroidetes Flavobacteria Flavobacteriales 0.13 2 A 193 7 LP
LP_16_Otu126 1 Bact 98 Bacteria Bacteroidetes Sphingobacteria Sphingobacteriales 0.16 2 C 176 9 LP
LP_16_Otu127 24 Bact 98 Bacteria Cyanobacteria SubsectionIV SubgroupI 0.18 2 A 159 4 LP
LP_16_Otu129 1 Bact 100 Bacteria Bacteroidetes Flavobacteria Flavobacteriales 0.31 3 C 105 7 LP and WC
LP_16_Otu13 38 Bact 100 Bacteria Verrucomicrobia Spartobacteria Chthoniobacterales 0.86 5 A 27 12 LP and WC
LP_16_Otu132 81 Bact 0 Bacteria Bacteria Bacteria Bacteria 0.14 2 B 191 7 LP
LP_16_Otu135 10 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Rhodospirillales 0.15 2 A 184 9 LP
LP_16_Otu136 78 Bact 100 Bacteria Chloroflexi SL56 marine group SL56 marine group 0.15 2 B 179 10 LP and WC
LP_16_Otu139 2 Bact 100 Bacteria Proteobacteria Betaproteobacteria Burkholderiales 0.19 2 C 155 12 LP and WC
LP_16_Otu14 69 Bact 97 Bacteria Proteobacteria Alphaproteobacteria Rhodobacterales 0.43 3 A 74 8 LP and WC
LP_16_Otu148 28 Bact 89 Bacteria Cyanobacteria SubsectionIII Pseudanabaena 0.2 2 B 150 5 LP
LP_16_Otu149 4 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Rhodobacterales 0.12 2 B 208 7 LP and WC
LP_16_Otu15 90 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Sphingomonadales 0.88 5 A 24 10 LP and WC
LP_16_Otu151 66 Bact 92 Bacteria Proteobacteria Betaproteobacteria Rhodocyclales 0.2 2 B 149 5 LP
LP_16_Otu152 19 Bact 100 Chloroplast Chloroplast Chloroplast Hemiselmis virescens 0.18 2 B 160 5 LP and WC
LP_16_Otu16 91 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Rhodobacterales 0.74 4 A 38 7 LP and WC
LP_16_Otu160 66 Bact 98 Bacteria Proteobacteria Alphaproteobacteria Rhizobiales 0.11 2 A 214 5 LP
LP_16_Otu162 28 Bact 100 Chloroplast Chloroplast Chloroplast Chloroplast 0.28 3 B 110 5 LP

148

LP_16_Otu163 3 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Sphingomonadales 0.12 2 B 209 6 LP and WC
LP_16_Otu165 40 Bact 0 Bacteria Bacteria Bacteria Bacteria 0.13 2 C 197 7 LP and WC
LP_16_Otu166 43 Bact 100 Bacteria Cyanobacteria SubsectionI SubsectionI 0.14 2 C 192 9 LP and WC
LP_16_Otu169 75 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Rhodobacterales 0.1 2 A 221 7 LP
LP_16_Otu17 84 Bact 100 Bacteria Proteobacteria Betaproteobacteria Burkholderiales 0.56 4 A 50 8 LP and WC
LP_16_Otu170 39 Bact 95 Chloroplast Chloroplast Chloroplast Pyramimonas olivacea 0.12 2 B 203 6 LP and WC
LP_16_Otu171 47 Bact 97 Chloroplast Chloroplast Chloroplast Chloroplast 0.33 3 C 97 9 LP and WC
LP_16_Otu175 11 Bact 98 Bacteria Cyanobacteria SubsectionIII uncultured 0.15 2 A 178 6 LP
LP_16_Otu18 74 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Sphingomonadales 0.31 3 A 104 8 LP and WC
LP_16_Otu182 76 Bact 0 Bacteria Bacteria Bacteria Bacteria 0.16 2 B 171 6 LP and WC
LP_16_Otu19 52 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Rhodobacterales 0.47 3 A 63 11 LP and WC
LP_16_Otu2 76 Bact 100 Bacteria Cyanobacteria SubsectionI Synechococcus 2.66 8 A 6 12 LP and WC
LP_16_Otu204 1 Bact 100 Bacteria Verrucomicrobia Opitutae vadinHA64 0.22 2 B 137 8 LP and WC
LP_16_Otu207 30 Bact 100 Bacteria Bacteroidetes Sphingobacteria Sphingobacteriales 0.12 2 B 211 8 LP
LP_16_Otu208 26 Bact 0 Bacteria Bacteria Bacteria Bacteria 0.08 1 A 231 4 LP
LP_16_Otu21 17 Bact 100 Bacteria Proteobacteria Betaproteobacteria Burkholderiales 0.76 5 C 35 10 LP and WC
LP_16_Otu212 76 Bact 100 Bacteria Bacteroidetes Sphingobacteria Sphingobacteriales 0.09 1 B 227 5 LP and WC
LP_16_Otu219 8 Bact 0 Bacteria Bacteria Bacteria Bacteria 0.16 2 B 174 4 LP
LP_16_Otu22 71 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Caulobacterales 0.4 3 A 80 6 LP and WC
LP_16_Otu23 28 Bact 100 Bacteria Bacteroidetes Flavobacteria Flavobacteriales 0.24 2 C 126 11 LP and WC
LP_16_Otu24 15 Bact 97 Chloroplast Chloroplast Chloroplast Chloroplast 0.8 5 A 31 7 LP
LP_16_Otu248 4 Bact 100 Bacteria Proteobacteria Betaproteobacteria Burkholderiales 0.08 1 B 234 4 LP
LP_16_Otu25 2 Bact 100 Bacteria Proteobacteria Gammaproteobacteria Xanthomonadales 0.66 4 A 43 4 LP
LP_16_Otu26 92 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Caulobacterales 0.5 3 A 57 6 LP and WC
LP_16_Otu261 18 Bact 91 Bacteria Cyanobacteria SubsectionIV SubgroupI 0.17 2 B 165 5 LP
LP_16_Otu268 83 Bact 100 Bacteria Proteobacteria Betaproteobacteria Nitrosomonadales 0.11 2 B 218 6 LP
LP_16_Otu269 52 Bact 97 Bacteria Proteobacteria Alphaproteobacteria Rhodospirillales 0.1 2 B 220 6 LP
LP_16_Otu27 19 Bact 100 Bacteria Proteobacteria Betaproteobacteria Burkholderiales 0.6 4 C 46 12 LP and WC
LP_16_Otu279 52 Bact 100 Bacteria Proteobacteria Alphaproteobacteria SAR11 clade 0.1 2 B 219 6 LP
LP_16_Otu28 73 Bact 100 Bacteria Bacteroidetes Flavobacteria Flavobacteriales 0.49 3 A 59 10 LP and WC
LP_16_Otu29 14 Bact 100 Bacteria Proteobacteria Betaproteobacteria Burkholderiales 0.41 3 C 79 12 LP and WC
LP_16_Otu3 17 Bact 97 Bacteria Verrucomicrobia Spartobacteria Chthoniobacterales 1.9 7 C 9 12 LP and WC
LP_16_Otu30 13 Bact 100 Bacteria Bacteroidetes Sphingobacteria Sphingobacteriales 0.13 2 A 194 6 LP and WC
LP_16_Otu33 66 Bact 100 Bacteria Bacteroidetes Sphingobacteria Sphingobacteriales 0.26 3 B 120 11 LP and WC
LP_16_Otu34 4 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Rhizobiales 0.45 3 C 68 12 LP and WC
LP_16_Otu347 8 Bact 0 Bacteria Bacteria Bacteria Bacteria 0.13 2 B 200 8 LP
LP_16_Otu35 11 Bact 100 Bacteria Bacteroidetes Flavobacteria Flavobacteriales 0.28 3 C 112 7 LP and WC
LP_16_Otu36 80 Bact 100 Bacteria Proteobacteria Gammaproteobacteria Xanthomonadales 0.52 4 C 55 12 LP and WC

149

LP_16_Otu37 6 Bact 95 Bacteria Actinobacteria Actinobacteria Frankiales 0.35 3 B 90 9 LP and WC
LP_16_Otu38 41 Bact 100 Bacteria Bacteroidetes Cytophagia Cytophagales 0.36 3 B 86 6 LP and WC
LP_16_Otu387 43 Bact 95 Bacteria Cyanobacteria SubsectionIV SubgroupI 0.11 2 B 216 5 LP
LP_16_Otu39 98 Bact 100 Bacteria Cyanobacteria SubsectionIII Planktothrix 0.44 3 A 72 8 LP and WC
LP_16_Otu4 90 Bact 97 Bacteria Proteobacteria Alphaproteobacteria SAR11 clade 1.82 7 B 10 12 LP and WC
LP_16_Otu40 82 Bact 100 Bacteria Cyanobacteria SubsectionIV SubgroupI 0.21 2 A 140 6 LP and WC
LP_16_Otu41 53 Bact 100 Bacteria Actinobacteria Acidimicrobiia Acidimicrobiales 0.35 3 C 91 9 LP and WC
LP_16_Otu42 80 Bact 100 Bacteria Bacteroidetes Sphingobacteria Sphingobacteriales 0.48 3 B 60 9 LP and WC
LP_16_Otu43 41 Bact 100 Bacteria Cyanobacteria SubsectionI Merismopedia 0.84 5 C 30 12 LP and WC
LP_16_Otu44 111 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Caulobacterales 0.34 3 A 92 7 LP
LP_16_Otu45 3 Bact 100 Bacteria Proteobacteria Gammaproteobacteria Xanthomonadales 0.25 2 C 123 11 LP and WC
LP_16_Otu46 62 Bact 100 Bacteria Proteobacteria Betaproteobacteria Burkholderiales 0.28 3 A 115 9 LP and WC
LP_16_Otu47 61 Bact 100 Bacteria Proteobacteria Betaproteobacteria Methylophilales 0.34 3 B 94 11 LP and WC
LP_16_Otu48 63 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Rhodobacterales 0.34 3 B 95 10 LP and WC
LP_16_Otu49 33 Bact 100 Bacteria Bacteroidetes Sphingobacteria Sphingobacteriales 0.35 3 B 89 7 LP and WC
LP_16_Otu5 88 Bact 97 Bacteria Bacteroidetes Sphingobacteria Sphingobacteriales 1.8 7 A 12 12 LP and WC
LP_16_Otu50 36 Bact 97 Bacteria Cyanobacteria SubsectionIII Leptolyngbya 0.41 3 C 77 6 LP and WC
LP_16_Otu51 69 Bact 100 Bacteria Cyanobacteria SubsectionI Synechococcus 0.6 4 B 47 10 LP and WC
LP_16_Otu53 74 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Sphingomonadales 0.12 2 A 202 6 LP and WC
LP_16_Otu54 4 Bact 100 Bacteria Verrucomicrobia Verrucomicrobiae Verrucomicrobiales 0.3 3 A 108 10 LP
LP_16_Otu55 6 Bact 100 Bacteria Cyanobacteria SubsectionI SubsectionI 0.37 3 C 84 12 LP and WC
LP_16_Otu56 95 Bact 100 Bacteria Actinobacteria Acidimicrobiia Acidimicrobiales 0.47 3 B 62 7 LP and WC
LP_16_Otu57 95 Bact 100 Bacteria Proteobacteria Betaproteobacteria Burkholderiales 0.26 3 A 122 10 LP and WC
LP_16_Otu58 91 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Rhodobacterales 0.38 3 A 81 7 LP
LP_16_Otu59 75 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Caulobacterales 0.28 3 A 109 11 LP
LP_16_Otu6 101 Bact 100 Bacteria Bacteroidetes Flavobacteria Flavobacteriales 1.04 6 A 19 9 LP and WC
LP_16_Otu62 100 Bact 100 Bacteria Cyanobacteria SubsectionIII Pseudanabaena sp. dqh15 0.36 3 A 85 8 LP
LP_16_Otu63 64 Bact 0 Bacteria Bacteria Bacteria Bacteria 0.26 3 A 121 7 LP
LP_16_Otu65 56 Bact 98 Bacteria Cyanobacteria SubsectionI Synechococcus 0.45 3 C 70 11 LP and WC
LP_16_Otu67 41 Bact 100 Bacteria Proteobacteria Betaproteobacteria Burkholderiales 0.24 2 B 125 11 LP
LP_16_Otu68 44 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Rhodobacterales 0.28 3 C 114 9 LP and WC
LP_16_Otu69 65 Bact 93 Chloroplast Chloroplast Chloroplast Chloroplast 0.89 5 B 23 6 LP and WC
LP_16_Otu7 84 Bact 95 Bacteria Lentisphaerae Lentisphaeria WCHB1-41 2.09 8 B 8 12 LP and WC
LP_16_Otu70 12 Bact 100 Bacteria Bacteroidetes Flavobacteria Flavobacteriales 0.46 3 B 66 8 LP and WC
LP_16_Otu73 71 Bact 100 Bacteria Bacteroidetes Sphingobacteria Sphingobacteriales 0.43 3 B 76 5 LP and WC
LP_16_Otu74 9 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Caulobacterales 0.22 2 A 136 10 LP and WC
LP_16_Otu75 29 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Rhizobiales 0.31 3 C 102 9 LP and WC
LP_16_Otu76 72 Bact 100 Bacteria Proteobacteria Betaproteobacteria Burkholderiales 0.2 2 A 144 6 LP

150

LP_16_Otu77 39 Bact 100 Bacteria Cyanobacteria SubsectionIII Prochlorothrix 0.53 4 B 53 7 LP and WC
LP_16_Otu78 34 Bact 98 Bacteria Cyanobacteria SubsectionIV SubgroupI 0.27 3 C 118 10 LP and WC
LP_16_Otu79 42 Bact 100 Bacteria Cyanobacteria SubsectionIV SubgroupI 0.36 3 B 87 7 LP and WC
LP_16_Otu8 66 Bact 100 Bacteria Proteobacteria Alphaproteobacteria SAR11 clade 0.5 3 A 58 8 LP and WC
LP_16_Otu81 27 Bact 98 Bacteria Cyanobacteria SubsectionI Synechococcus 0.28 3 B 111 9 LP and WC
LP_16_Otu82 64 Bact 98 Bacteria Bacteroidetes Flavobacteria Flavobacteriales 0.21 2 A 139 5 LP
LP_16_Otu84 40 Bact 0 Bacteria Bacteria Bacteria Bacteria 0.17 2 B 161 5 LP and WC
LP_16_Otu86 20 Bact 100 Bacteria Actinobacteria Acidimicrobiia Acidimicrobiales 0.23 2 C 132 11 LP and WC
LP_16_Otu87 80 Bact 100 Bacteria Proteobacteria Betaproteobacteria Burkholderiales 0.34 3 B 96 7 LP
LP_16_Otu88 4 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Sphingomonadales 0.15 2 C 186 11 LP and WC
LP_16_Otu89 8 Bact 100 Bacteria Bacteroidetes Flavobacteria Flavobacteriales 0.21 2 A 141 9 LP
LP_16_Otu9 67 Bact 100 Bacteria Bacteroidetes Flavobacteria Flavobacteriales 1.75 7 A 13 6 LP and WC
LP_16_Otu92 54 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Rhodobacterales 0.19 2 A 153 6 LP and WC
LP_16_Otu98 30 Bact 100 Chloroplast Chloroplast Chloroplast Chloroplast 0.33 3 C 98 8 LP
LP_18_Otu00001 59 Euk 100 Eukaryota Haptophyceae Prymnesiales Prymnesiaceae 14.3 8 A 1 11 LP and WC
LP_18_Otu00002 74 Euk 86 Eukaryota Fungi Dikarya Ascomycota 4.62 8 A 3 9 LP
LP_18_Otu00003 3 Euk 97 Eukaryota Katablepharidophyta Katablepharidaceae Katablepharis 0.56 4 C 51 10 LP and WC
LP_18_Otu00004 62 Euk 99 Eukaryota stramenopiles Bacillariophyta Coscinodiscophyceae 4.28 8 B 4 12 LP and WC
LP_18_Otu00005 2 Euk 88 Eukaryota Fungi Chytridiomycota Monoblepharidomycetes 1.82 7 C 11 12 LP and WC
LP_18_Otu00008 85 Euk 93 Eukaryota Viridiplantae Chlorophyta Prasinophyceae 0.27 3 B 116 8 LP and WC
LP_18_Otu00010 1 Euk 95 Eukaryota stramenopiles Chrysophyceae Chromulinales 2.35 8 A 7 6 LP and WC
LP_18_Otu00011 38 Euk 100 Eukaryota stramenopiles Chrysophyceae Chromulinales 1.24 6 A 16 6 LP and WC
LP_18_Otu00012 21 Euk 99 Eukaryota Metazoa Rotifera Monogononta 0.48 3 A 61 9 LP and WC
LP_18_Otu00013 41 Euk 93 Eukaryota Rhizaria Cercozoa Silicofilosea 0.75 4 B 36 7 LP and WC
LP_18_Otu00014 104 Euk 98 Eukaryota stramenopiles Bacillariophyta Coscinodiscophyceae 0.7 4 B 40 9 LP and WC
LP_18_Otu00015 38 Euk 86 Eukaryota Fungi Cryptomycota Rozella 1.4 6 A 15 8 LP
LP_18_Otu00016 38 Euk 95 Eukaryota Alveolata Ciliophora Intramacronucleata 0.87 5 B 26 7 LP and WC
LP_18_Otu00017 2 Euk 94 Eukaryota Alveolata Ciliophora Intramacronucleata 0.78 5 B 34 6 LP and WC
LP_18_Otu00019 37 Euk 91 Eukaryota Rhizaria Cercozoa Cercomonadida 0.45 3 A 71 7 LP and WC
LP_18_Otu00021 33 Euk 96 Eukaryota Alveolata Ciliophora Intramacronucleata 0.8 5 B 32 7 LP and WC
LP_18_Otu00022 31 Euk 96 Eukaryota Viridiplantae Chlorophyta Prasinophyceae 0.47 3 B 64 7 LP and WC
LP_18_Otu00023 20 Euk 80 Eukaryota Rhizaria Cercozoa Cercozoa sp. 0.85 5 A 28 4 LP
LP_18_Otu00024 6 Euk 100 Eukaryota Haptophyceae Pavlovales Pavlovaceae 0.31 3 A 103 10 LP and WC
LP_18_Otu00025 3 Euk 100 Eukaryota Metazoa Arthropoda Crustacea 0.13 2 B 195 6 LP and WC
LP_18_Otu00026 5 Euk 95 Eukaryota Alveolata Ciliophora Postciliodesmatophora 0.56 4 B 49 8 LP and WC
LP_18_Otu00027 37 Euk 98 Eukaryota Alveolata Ciliophora Intramacronucleata 0.15 2 B 187 7 LP and WC
LP_18_Otu00030 12 Euk 90 Eukaryota Alveolata Ciliophora Intramacronucleata 1.06 6 B 18 4 LP
LP_18_Otu00031 86 Euk 100 Eukaryota Cryptophyta Cryptomonadales Cryptomonadaceae 0.59 4 B 48 8 LP and WC

151

LP_18_Otu00032 59 Euk 82 Eukaryota Eukaryota Eukaryota Eukaryota 0.97 5 A 21 12 LP
LP_18_Otu00033 57 Euk 99 Eukaryota Viridiplantae Chlorophyta Mamiellophyceae 0.37 3 A 83 6 LP and WC
LP_18_Otu00035 80 Euk 99 Eukaryota Metazoa Rotifera Monogononta 0.91 5 B 22 6 LP and WC
LP_18_Otu00036 4 Euk 91 Eukaryota stramenopiles Chrysophyceae Chromulinales 1.11 6 B 17 4 LP
LP_18_Otu00039 84 Euk 91 Eukaryota Cryptophyta Cryptomonadales Cryptomonadaceae 0.65 4 B 45 6 LP and WC
LP_18_Otu00042 3 Euk 92 Eukaryota stramenopiles Pirsonia Pirsonia 0.3 3 A 107 4 LP and WC
LP_18_Otu00044 69 Euk 99 Eukaryota stramenopiles Bacillariophyta Coscinodiscophyceae 0.85 5 B 29 5 LP
LP_18_Otu00045 78 Euk 99 Eukaryota Cryptophyta Cryptophyta Cryptophyta sp 0.68 4 B 42 7 LP and WC
LP_18_Otu00047 90 Euk 100 Eukaryota Cryptophyta Cryptomonadales Cryptomonadaceae 0.43 3 B 75 7 LP and WC
LP_18_Otu00048 8 Euk 97 Eukaryota stramenopiles Bacillariophyta Coscinodiscophyceae 0.79 5 A 33 9 LP and WC
LP_18_Otu00051 67 Euk 93 Eukaryota Ichthyosporea Ichthyophonida Anurofeca 0.31 3 B 101 5 LP and WC
LP_18_Otu00053 8 Euk 93 Eukaryota Euglenozoa Euglenida Euglenales 0.19 2 A 152 8 LP and WC
LP_18_Otu00055 87 Euk 87 Eukaryota Fungi Fungi Fungi 0.75 4 A 37 5 LP and WC
LP_18_Otu00056 57 Euk 100 Eukaryota Haptophyceae Pavlovales Pavlovaceae 0.27 3 A 117 6 LP and WC
LP_18_Otu00058 44 Euk 100 Eukaryota Viridiplantae Chlorophyta Prasinophyceae 0.52 4 C 56 6 LP and WC
LP_18_Otu00059 22 Euk 100 Eukaryota stramenopiles Bacillariophyta Coscinodiscophyceae 0.87 5 B 25 12 LP
LP_18_Otu00060 52 Euk 87 Eukaryota Eukaryota Eukaryota Eukaryota 0.2 2 B 145 4 LP and WC
LP_18_Otu00062 78 Euk 95 Eukaryota Alveolata Ciliophora Intramacronucleata 0.15 2 B 185 6 LP and WC
LP_18_Otu00067 33 Euk 86 Eukaryota Fungi Dikarya Ascomycota 0.35 3 A 88 6 LP
LP_18_Otu00068 13 Euk 100 Eukaryota Metazoa Rotifera Monogononta 0.47 3 B 65 6 LP and WC
LP_18_Otu00070 20 Euk 98 Eukaryota Rhodophyta Bangiophyceae Bangiales 0.17 2 B 166 5 LP and WC
LP_18_Otu00072 71 Euk 88 Eukaryota Cryptophyta Cryptomonadales Cryptomonadaceae 0.23 2 B 130 5 LP and WC
LP_18_Otu00073 46 Euk 94 Eukaryota stramenopiles Bacillariophyta Bacillariophyceae 0.18 2 B 156 9 LP and WC
LP_18_Otu00076 34 Euk 98 Eukaryota Cryptophyta Cryptomonadales Hemiselmidaceae 0.24 2 B 129 5 LP and WC
LP_18_Otu00077 70 Euk 99 Eukaryota Alveolata Ciliophora Intramacronucleata 0.53 4 B 54 5 LP
LP_18_Otu00078 78 Euk 91 Eukaryota Alveolata Ciliophora Intramacronucleata 0.28 3 B 113 6 LP and WC
LP_18_Otu00080 60 Euk 92 Eukaryota stramenopiles Bacillariophyta Mediophyceae 0.18 2 B 158 8 LP and WC
LP_18_Otu00081 81 Euk 90 Eukaryota Cryptophyta Cryptomonadales Cryptomonadaceae 0.55 4 B 52 5 LP
LP_18_Otu00083 82 Euk 90 Eukaryota Cryptophyta Cryptomonadales Cryptomonadaceae 0.32 3 B 99 6 LP
LP_18_Otu00084 27 Euk 93 Eukaryota Cryptophyta

0.43 3 B 73 6 LP
LP_18_Otu00085 65 Euk 100 Eukaryota Alveolata Ciliophora Intramacronucleata 0.2 2 B 146 6 LP and WC
LP_18_Otu00088 22 Euk 89 Eukaryota Fungi Dikarya Ascomycota 0.24 2 A 127 5 LP
LP_18_Otu00090 20 Euk 98 Eukaryota Metazoa Arthropoda Hexapoda 0.21 2 A 142 4 LP
LP_18_Otu00098 53 Euk 89 Eukaryota stramenopiles Synurophyceae Synurales 0.16 2 B 172 5 LP and WC
LP_18_Otu00101 25 Euk 94 Eukaryota Rhizaria Cercozoa Cercozoa sp. CC-2009a 0.31 3 B 100 5 LP
LP_18_Otu00105 70 Euk 100 Eukaryota Viridiplantae Chlorophyta Prasinophyceae 0.18 2 B 157 6 LP and WC
LP_18_Otu00108 63 Euk 97 Eukaryota stramenopiles Chrysophyceae Chromulinales 0.41 3 A 78 5 LP
LP_18_Otu00112 23 Euk 94 Eukaryota Choanoflagellida Acanthoecidae Savillea 0.37 3 A 82 6 LP

152

LP_18_Otu00114 70 Euk 98 Eukaryota stramenopiles Chrysophyceae Chromulinales 0.14 2 A 189 6 LP
LP_18_Otu00117 52 Euk 92 Eukaryota Cryptophyta Cryptomonadales Cryptomonadaceae 0.12 2 B 204 4 LP and WC
LP_18_Otu00118 93 Euk 92 Eukaryota stramenopiles Dictyochophyceae Pedinellales 0.15 2 A 182 6 LP
LP_18_Otu00119 47 Euk 95 Eukaryota Viridiplantae Chlorophyta Chlorophyceae 0.17 2 B 162 5 LP and WC
LP_18_Otu00120 19 Euk 96 Eukaryota Alveolata Ciliophora Intramacronucleata 0.23 2 B 134 8 LP and WC
LP_18_Otu00122 47 Euk 99 Eukaryota Metazoa Rotifera Monogononta 0.23 2 B 135 6 LP
LP_18_Otu00124 96 Euk 99 Eukaryota Viridiplantae Chlorophyta Chlorophyceae 0.23 2 B 131 6 LP
LP_18_Otu00127 60 Euk 94 Eukaryota Cryptophyta Cryptomonadales Cryptomonadaceae 0.19 2 B 151 7 LP
LP_18_Otu00137 1 Euk 96 Eukaryota stramenopiles Pirsonia Pirsonia formosa 0.15 2 C 183 8 LP
LP_18_Otu00140 13 Euk 100 Eukaryota Alveolata Dinophyceae unclassified Dinophyceae 0.15 2 B 188 8 LP and WC
LP_18_Otu00141 10 Euk 87 Eukaryota Rhizaria Cercozoa Vampyrellidae 0.12 2 B 205 5 LP
LP_18_Otu00151 25 Euk 93 Eukaryota Choanoflagellida Acanthoecidae Savillea 0.3 3 A 106 7 LP
LP_18_Otu00153 30 Euk 94 Eukaryota Viridiplantae Chlorophyta Chlorophyceae 0.34 3 A 93 7 LP
LP_18_Otu00154 52 Euk 88 Eukaryota Alveolata Dinophyceae Dinophyceae sp. 0.12 2 B 210 4 LP
LP_18_Otu00156 16 Euk 100 Eukaryota Viridiplantae Chlorophyta Chlorophyceae 0.23 2 A 133 12 LP
LP_18_Otu00163 34 Euk 95 Eukaryota stramenopiles Chrysophyceae Chrysophyceae sp. 0.16 2 A 173 7 LP
LP_18_Otu00164 2 Euk 97 Eukaryota Alveolata Ciliophora Intramacronucleata 0.13 2 B 198 5 LP
LP_18_Otu00169 2 Euk 92 Eukaryota Rhizaria Cercozoa Silicofilosea 0.08 1 B 230 4 LP
LP_18_Otu00175 39 Euk 96 Eukaryota Euglenozoa Euglenida Euglenales 0.2 2 B 147 7 LP
LP_18_Otu00183 5 Euk 100 Eukaryota Alveolata Ciliophora Intramacronucleata 0.13 2 A 196 7 LP
LP_18_Otu00191 26 Euk 96 Eukaryota Viridiplantae Chlorophyta Chlorophyceae 0.14 2 B 190 9 LP
LP_18_Otu00193 76 Euk 98 Eukaryota Cryptophyta Pyrenomonadales Chroomonadaceae 0.16 2 B 175 6 LP
LP_18_Otu00200 14 Euk 89 Eukaryota Alveolata Apicomplexa Colpodellidae 0.11 2 B 217 4 LP
LP_18_Otu00202 18 Euk 99 Eukaryota stramenopiles Bacillariophyta Coscinodiscophyceae 0.17 2 A 164 7 LP
LP_18_Otu00203 29 Euk 82 Eukaryota Alveolata Ciliophora Intramacronucleata 0.1 2 B 223 4 LP
LP_18_Otu00211 4 Euk 87 Eukaryota

0.15 2 B 180 5 LP
LP_18_Otu00237 3 Euk 94 Eukaryota Choanoflagellida Acanthoecidae Savillea 0.11 2 A 212 5 LP
LP_18_Otu00249 11 Euk 92 Eukaryota Alveolata Ciliophora Intramacronucleata 0.17 2 A 163 6 LP
LP_18_Otu00261 4 Euk 99 Eukaryota Metazoa Rotifera Monogononta 0.12 2 B 207 4 LP
LP_18_Otu00265 3 Euk 97 Eukaryota stramenopiles Chrysophyceae Chromulinales 0.16 2 A 167 4 LP
LP_18_Otu00281 70 Euk 99 Eukaryota Metazoa Rotifera Monogononta 0.11 2 B 215 5 LP
LP_18_Otu00320 8 Euk 88 Eukaryota Alveolata Apicomplexa Coccidia 0.1 2 B 224 4 LP
LP_Chla 1 Env 0 Chla Chla Chla Chla 1 4 C null 12 LP and WC
LP_DO 29 Env 0 DO DO DO DO 1 4 C null 12 LP and WC
LP_pH 19 Env 0 pH pH pH pH 1 4 C null 12 LP and WC
LP_Ppar 54 Env 0 Ppar Ppar Ppar Ppar 1 4 C null 6 LP and WC
LP_PSU 28 Env 0 PSU PSU PSU PSU 1 4 C null 12 LP and WC
LP_Temperature 45 Env 0 Temp Temperature Temperature Temperature 1 4 C null 12 LP and WC

153

Table S2. A summary of the variables (18s and 16s rDNA OTUs plus environmental variables), in Wilson Creek and their number of correlations,
best percent hit to SILVA (v108) reference database, nested taxonomic identifications, average relative abundance, binned abundance,
seasonal abundance: A (November to April), B (May to October), number of occurrences, and location of detection in the Lake Texoma

ID
#
Correlatio
ns
Gen
e
Similarit
y to
SILVA Domain Phylum Class Other
Av
Abun
d
Binne
d
Abund
Seasona
l Abund
Abun
d
Rank
Occu
r Location
WC_16_Otu1 42 Bact 94 Chloroplast Chloroplast Chloroplast Chloroplast 0.71 4 A 40 9 LP and WC
WC_16_Otu10 5 Bact 100 Chloroplast Chloroplast Chloroplast Odontella sinensis 1.74 7 B 13 12 LP and WC
WC_16_Otu100 3 Bact 100 Chloroplast Chloroplast Chloroplast Chloroplast 0.12 2 B 234 8 LP and WC
WC_16_Otu101 1 Bact 98 Bacteria Bacteroidetes Flavobacteria Flavobacteriales 0.2 2 C 162 8 WC
WC_16_Otu102 81 Bact 100 Bacteria Proteobacteria Betaproteobacteria Burkholderiales 0.23 2 A 148 9 WC
WC_16_Otu105 9 Bact 100 Bacteria Proteobacteria Betaproteobacteria Burkholderiales 0.2 2 C 163 10 WC
WC_16_Otu106 65 Bact 98 Bacteria Planctomycetes Phycisphaerae Phycisphaerales 0.16 2 B 196 8 WC
WC_16_Otu107 39 Bact 98 Bacteria Cyanobacteria SubsectionI Synechococcus 0.34 3 B 98 7 LP and WC
WC_16_Otu108 50 Bact 100 Bacteria Actinobacteria Actinobacteria Frankiales 0.19 2 A 172 9 LP and WC
WC_16_Otu109 13 Bact 0 Bacteria Bacteria Bacteria Bacteria 0.21 2 C 157 8 WC
WC_16_Otu11 7 Bact 100 Bacteria Bacteroidetes Cytophagia Cytophagales 1.33 6 A 20 12 WC
WC_16_Otu112 7 Bact 97 Bacteria Proteobacteria Betaproteobacteria Burkholderiales 0.23 2 C 145 10 WC
WC_16_Otu115 5 Bact 100 Bacteria Actinobacteria Acidimicrobiia Acidimicrobiales 0.12 2 B 231 6 LP and WC
WC_16_Otu117 1 Bact 100 Bacteria Bacteroidetes Flavobacteria Flavobacteriales 0.16 2 A 194 5 WC
WC_16_Otu120 44 Bact 97 Bacteria Cyanobacteria SubsectionIV SubgroupI 0.13 2 B 225 5 LP and WC
WC_16_Otu121 41 Bact 100 Bacteria Proteobacteria Betaproteobacteria Methylophilales 0.18 2 C 182 8 WC
WC_16_Otu122 2 Bact 100 Bacteria Chlorobi Chlorobia Chlorobiales 0.18 2 C 180 7 WC
WC_16_Otu129 7 Bact 100 Bacteria Bacteroidetes Flavobacteria Flavobacteriales 0.15 2 C 203 10 LP and WC
WC_16_Otu13 24 Bact 100 Bacteria
Verrucomicrobi
a Spartobacteria Chthoniobacterales 0.58 4 A 53 11 LP and WC
WC_16_Otu133 6 Bact 95 Chloroplast Chloroplast Chloroplast Chloroplast 0.13 2 A 227 5 WC
WC_16_Otu134 51 Bact 100 Bacteria
Verrucomicrobi
a
Candidatus
Methylacidiphilum
Candidatus
Methylacidiphilum 0.15 2 B 206 9 WC
WC_16_Otu136 6 Bact 100 Bacteria Chloroflexi SL56 marine group SL56 0.11 2 B 239 8 LP and WC
WC_16_Otu139 8 Bact 100 Bacteria Proteobacteria Betaproteobacteria Burkholderiales 0.21 2 B 154 12 LP and WC
WC_16_Otu14 51 Bact 97 Bacteria Proteobacteria Alphaproteobacteria Rhodobacterales 0.86 5 A 35 8 LP and WC
WC_16_Otu140 8 Bact 96 Bacteria Proteobacteria Betaproteobacteria Burkholderiales 0.13 2 A 215 4 WC
WC_16_Otu144 19 Bact 100 Bacteria Proteobacteria Betaproteobacteria Burkholderiales 0.13 2 B 216 8 WC
WC_16_Otu146 48 Bact 100 Bacteria Cyanobacteria SubsectionIV SubgroupI 0.2 2 B 161 5 WC
WC_16_Otu149 8 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Rhodobacterales 0.13 2 B 214 5 LP and WC
WC_16_Otu15 44 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Sphingomonadales 0.37 3 A 93 12 LP and WC
WC_16_Otu150 40 Bact 100 Bacteria Actinobacteria Acidimicrobiia Acidimicrobiales 0.19 2 A 177 5 WC
WC_16_Otu152 24 Bact 100 Chloroplast Chloroplast Chloroplast Hemiselmis 0.14 2 B 213 4 LP and WC
WC_16_Otu154 30 Bact 100 Chloroplast Chloroplast Chloroplast Mantoniela 0.09 1 A 255 4 WC

154

WC_16_Otu159 5 Bact 97 Bacteria Bacteroidetes Sphingobacteria Sphingobacteriales 0.14 2 A 207 5 WC
WC_16_Otu16 88 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Rhodobacterales 0.39 3 A 82 7 LP and WC
WC_16_Otu163 10 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Sphingomonadales 0.16 2 B 197 9 LP and WC
WC_16_Otu165 38 Bact 0 Bacteria Bacteria Bacteria Bacteria 0.14 2 B 212 6 LP and WC
WC_16_Otu166 24 Bact 100 Bacteria Cyanobacteria SubsectionI SubsectionI 0.25 2 B 136 9 LP and WC
WC_16_Otu167 59 Bact 97 Bacteria Proteobacteria Alphaproteobacteria Rhodobacterales 0.2 2 B 170 7 WC
WC_16_Otu17 45 Bact 100 Bacteria Proteobacteria Betaproteobacteria Burkholderiales 0.42 3 A 76 6 LP and WC
WC_16_Otu170 10 Bact 95 Chloroplast Chloroplast Chloroplast Pyramimonas olivacea 0.17 2 A 190 10 LP and WC
WC_16_Otu171 1 Bact 97 Chloroplast Chloroplast Chloroplast Chloroplast 0.18 2 C 185 11 LP and WC
WC_16_Otu173 4 Bact 100 Bacteria Cyanobacteria Cyanobacteria Cyanobacteria 0.18 2 A 179 9 WC
WC_16_Otu18 78 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Sphingomonadales 0.74 4 A 38 9 LP and WC
WC_16_Otu182 11 Bact 0 Bacteria Bacteria Bacteria Bacteria 0.14 2 B 208 6 LP and WC
WC_16_Otu19 88 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Rhodobacterales 0.49 3 A 63 11 LP and WC
WC_16_Otu193 51 Bact 90 Bacteria Cyanobacteria SubsectionIII Leptolyngbya 0.13 2 B 219 6 WC
WC_16_Otu198 62 Bact 93 Bacteria Cyanobacteria SubsectionIII Phormidium 0.09 1 B 252 4 WC
WC_16_Otu2 19 Bact 100 Bacteria Cyanobacteria SubsectionI Synechococcus 1.81 7 A 10 11 LP and WC
WC_16_Otu20 51 Bact 100 Bacteria Actinobacteria Actinobacteria Frankiales 0.88 5 A 32 8 WC
WC_16_Otu204 7 Bact 100 Bacteria
Verrucomicrobi
a Opitutae Opitutae 0.17 2 C 187 10 LP and WC
WC_16_Otu21 52 Bact 100 Bacteria Proteobacteria Betaproteobacteria Burkholderiales 0.97 5 A 28 10 LP and WC
WC_16_Otu212 6 Bact 100 Bacteria Bacteroidetes Sphingobacteria Sphingobacteriales 0.1 1 B 246 6 LP and WC
WC_16_Otu22 44 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Caulobacterales 0.6 4 A 46 7 LP and WC
WC_16_Otu228 36 Bact 100 Bacteria Bacteroidetes Sphingobacteria Sphingobacteriales 0.13 2 B 228 5 WC
WC_16_Otu23 27 Bact 100 Bacteria Bacteroidetes Flavobacteria Flavobacteriales 0.81 5 A 36 12 LP and WC
WC_16_Otu238 23 Bact 0 Bacteria Bacteria Bacteria Bacteria 0.09 1 A 256 6 WC
WC_16_Otu247 1 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Sphingomonadales 0.09 1 B 249 4 WC
WC_16_Otu253 35 Bact 100 Bacteria Proteobacteria Gammaproteobacter Enterobacteriales 0.08 1 B 260 7 WC
WC_16_Otu259 73 Bact 0 Bacteria Bacteria Bacteria Bacteria 0.13 2 B 226 4 WC
WC_16_Otu26 53 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Caulobacterales 0.45 3 A 70 7 LP and WC
WC_16_Otu264 37 Bact 0 Bacteria Bacteria Bacteria Bacteria 0.1 1 A 243 5 WC
WC_16_Otu27 34 Bact 100 Bacteria Proteobacteria Betaproteobacteria Burkholderiales 0.48 3 C 66 11 LP and WC
WC_16_Otu278 1 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Rhizobiales 0.1 1 B 248 7 WC
WC_16_Otu28 65 Bact 100 Bacteria Bacteroidetes Flavobacteria Flavobacteriales 0.26 3 A 133 8 LP and WC
WC_16_Otu286 1 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Caulobacterales 0.08 1 B 261 4 WC
WC_16_Otu29 15 Bact 100 Bacteria Proteobacteria Betaproteobacteria Burkholderiales 0.58 4 C 51 12 LP and WC
WC_16_Otu3 8 Bact 97 Bacteria
Verrucomicrobi
a Spartobacteria Chthoniobacterales 3.32 8 C 4 12 LP and WC
WC_16_Otu30 78 Bact 100 Bacteria Bacteroidetes Sphingobacteria Sphingobacteriales 0.44 3 A 73 6 LP and WC
WC_16_Otu31 54 Bact 100 Bacteria Actinobacteria Actinobacteria Micrococcales 0.5 3 A 62 11 WC
WC_16_Otu32 43 Bact 100 Chloroplast Chloroplast Chloroplast Chloroplast 0.72 4 A 39 6 WC
WC_16_Otu33 8 Bact 100 Bacteria Bacteroidetes Sphingobacteria Sphingobacteriales 0.58 4 C 52 12 LP and WC
WC_16_Otu34 16 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Rhizobiales 0.57 4 C 57 11 LP and WC

155

WC_16_Otu35 5 Bact 100 Bacteria Bacteroidetes Flavobacteria Flavobacteriales 0.39 3 A 85 12 LP and WC
WC_16_Otu36 16 Bact 100 Bacteria Proteobacteria Gammaproteobacter Xanthomonadales 0.37 3 C 92 11 LP and WC
WC_16_Otu37 8 Bact 95 Bacteria Actinobacteria Actinobacteria Frankiales 0.65 4 C 43 12 LP and WC
WC_16_Otu38 20 Bact 100 Bacteria Bacteroidetes Cytophagia Cytophagales 0.53 4 C 59 10 LP and WC
WC_16_Otu39 42 Bact 100 Bacteria Cyanobacteria SubsectionIII Planktothrix 0.21 2 A 153 8 LP and WC
WC_16_Otu4 10 Bact 97 Bacteria Proteobacteria Alphaproteobacteria SAR11 clade 3.9 8 C 1 12 LP and WC
WC_16_Otu40 9 Bact 100 Bacteria Cyanobacteria SubsectionIV SubgroupI 0.59 4 A 48 5 LP and WC
WC_16_Otu41 8 Bact 100 Bacteria Actinobacteria Acidimicrobiia Acidimicrobiales 0.54 4 C 58 12 LP and WC
WC_16_Otu42 37 Bact 100 Bacteria Bacteroidetes Sphingobacteria Sphingobacteriales 0.49 3 C 64 6 LP and WC
WC_16_Otu43 21 Bact 100 Bacteria Cyanobacteria SubsectionI Merismopedia 1.08 6 C 24 12 LP and WC
WC_16_Otu45 40 Bact 100 Bacteria Proteobacteria Gammaproteobacter Xanthomonadales 0.43 3 C 75 12 LP and WC
WC_16_Otu46 51 Bact 100 Bacteria Proteobacteria Betaproteobacteria Burkholderiales 0.29 3 A 121 9 LP and WC
WC_16_Otu47 2 Bact 100 Bacteria Proteobacteria Betaproteobacteria Methylophilales 0.44 3 C 74 12 LP and WC
WC_16_Otu48 66 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Rhodobacterales 0.59 4 B 49 10 LP and WC
WC_16_Otu49 35 Bact 100 Bacteria Bacteroidetes Sphingobacteria Sphingobacteriales 0.45 3 B 72 8 LP and WC
WC_16_Otu5 1 Bact 97 Bacteria Bacteroidetes Sphingobacteria Sphingobacteriales 0.99 5 C 27 12 LP and WC
WC_16_Otu50 45 Bact 97 Bacteria Cyanobacteria SubsectionIII Leptolyngbya 0.47 3 B 68 6 LP and WC
WC_16_Otu51 32 Bact 100 Bacteria Cyanobacteria SubsectionI Synechococcus 0.42 3 B 77 11 LP and WC
WC_16_Otu52 33 Bact 100 Bacteria
Verrucomicrobi
a Verrucomicrobiae Verrucomicrobiales 0.22 2 A 152 9 WC
WC_16_Otu53 63 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Sphingomonadales 0.37 3 A 91 8 LP and WC
WC_16_Otu55 23 Bact 100 Bacteria Cyanobacteria SubsectionI SubsectionI 0.3 3 C 111 10 LP and WC
WC_16_Otu56 14 Bact 100 Bacteria Actinobacteria Acidimicrobiia Acidimicrobiales 0.6 4 C 45 12 LP and WC
WC_16_Otu57 74 Bact 100 Bacteria Proteobacteria Betaproteobacteria Burkholderiales 0.14 2 A 210 10 LP and WC
WC_16_Otu6 41 Bact 100 Bacteria Bacteroidetes Flavobacteria Flavobacteriales 1.79 7 A 11 7 LP and WC
WC_16_Otu60 29 Bact 100 Bacteria Bacteroidetes Cytophagia Cytophagales 0.38 3 A 89 5 WC
WC_16_Otu61 26 Bact 100 Bacteria Actinobacteria Actinobacteria PeM15 0.58 4 C 50 12 WC
WC_16_Otu65 41 Bact 98 Bacteria Cyanobacteria SubsectionI Synechococcus 0.68 4 B 41 11 LP and WC
WC_16_Otu66 31 Bact 100 Chloroplast Chloroplast Chloroplast Isochrysis 0.37 3 A 94 4 WC
WC_16_Otu68 20 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Rhodobacterales 0.57 4 C 56 12 LP and WC
WC_16_Otu69 68 Bact 93 Chloroplast Chloroplast Chloroplast Chloroplast 0.24 2 B 141 5 LP and WC
WC_16_Otu7 11 Bact 95 Bacteria Lentisphaerae Lentisphaeria WCHB1-41 2.18 8 C 6 12 LP and WC
WC_16_Otu70 3 Bact 100 Bacteria Bacteroidetes Flavobacteria Flavobacteriales 0.18 2 B 184 8 LP and WC
WC_16_Otu71 19 Bact 100 Bacteria Bacteroidetes Sphingobacteria Sphingobacteriales 0.21 2 A 158 7 WC
WC_16_Otu73 43 Bact 100 Bacteria Bacteroidetes Sphingobacteria Sphingobacteriales 0.19 2 B 176 5 LP and WC
WC_16_Otu74 1 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Caulobacterales 0.28 3 B 125 7 LP and WC
WC_16_Otu75 53 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Rhizobiales 0.29 3 B 120 8 LP and WC
WC_16_Otu77 68 Bact 100 Bacteria Cyanobacteria SubsectionIII Prochlorothrix 0.57 4 B 54 4 LP and WC
WC_16_Otu78 50 Bact 98 Bacteria Cyanobacteria SubsectionIV SubgroupI 0.32 3 B 103 6 LP and WC
WC_16_Otu79 74 Bact 100 Bacteria Cyanobacteria SubsectionIV SubgroupI 0.45 3 B 71 5 LP and WC
WC_16_Otu8 56 Bact 100 Bacteria Proteobacteria Alphaproteobacteria SAR11 clade 1.77 7 A 12 7 LP and WC

156

WC_16_Otu81 51 Bact 98 Bacteria Cyanobacteria SubsectionI Synechococcus 0.35 3 B 97 9 LP and WC
WC_16_Otu84 52 Bact 0 Bacteria Bacteria Bacteria Bacteria 0.36 3 B 95 6 LP and WC
WC_16_Otu85 4 Bact 97 Bacteria Proteobacteria Deltaproteobacteria Bdellovibrionales 0.33 3 A 99 5 WC
WC_16_Otu86 7 Bact 100 Bacteria Actinobacteria Acidimicrobiia Acidimicrobiales 0.3 3 C 112 12 LP and WC
WC_16_Otu88 26 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Sphingomonadales 0.23 2 C 149 11 LP and WC
WC_16_Otu9 48 Bact 100 Bacteria Bacteroidetes Flavobacteria Flavobacteriales 0.39 3 A 84 7 LP and WC
WC_16_Otu90 20 Bact 100 Bacteria
Verrucomicrobi
a Verrucomicrobiae Verrucomicrobiales 0.17 2 A 186 8 WC
WC_16_Otu91 67 Bact 100 Bacteria Actinobacteria Actinobacteria Micrococcales 0.31 3 A 109 9 WC
WC_16_Otu92 47 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Rhodobacterales 0.3 3 A 115 6 LP and WC
WC_16_Otu94 1 Bact 0 Bacteria Bacteria Bacteria Bacteria 0.19 2 A 175 8 WC
WC_16_Otu95 33 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Rhodobacterales 0.2 2 A 164 6 WC
WC_16_Otu97 34 Bact 100 Bacteria Bacteroidetes Sphingobacteria Sphingobacteriales 0.26 3 C 129 10 WC
WC_16_Otu99 8 Bact 0 Bacteria Bacteria Bacteria Bacteria 0.24 2 C 138 8 WC
WC_18_Otu00001 18 Euk 100 Eukaryota Haptophyceae Prymnesiales Prymnesiaceae 0.92 5 A 30 10 LP and WC
WC_18_Otu00003 13 Euk 97 Eukaryota Katablepharid Katablepharidaceae Katablepharis 3.42 8 A 3 12 LP and WC
WC_18_Otu00004 60 Euk 99 Eukaryota stramenopiles Bacillariophyta Coscinodiscophyce 3.76 8 B 2 12 LP and WC
WC_18_Otu00005 1 Euk 88 Eukaryota Fungi Chytridiomycota Monoblepharidomy 0.86 5 C 34 12 LP and WC
WC_18_Otu00006 1 Euk 82 Eukaryota Rhizaria Cercozoa Cercomonadida 1.34 6 C 19 5 WC
WC_18_Otu00007 56 Euk 98 Eukaryota Alveolata Ciliophora Intramacronucleata 2.41 8 A 5 4 WC
WC_18_Otu00008 38 Euk 93 Eukaryota Viridiplantae Chlorophyta Prasinophyceae 2.02 8 B 8 6 LP and WC
WC_18_Otu00009 31 Euk 96 Eukaryota Viridiplantae Chlorophyta Chlorophyceae 1.45 6 A 15 8 WC
WC_18_Otu00010 47 Euk 95 Eukaryota stramenopiles Chrysophyceae Chromulinales 0.17 2 A 193 6 LP and WC
WC_18_Otu00011 20 Euk 100 Eukaryota stramenopiles Chrysophyceae Chromulinales 1.38 6 A 17 7 LP and WC
WC_18_Otu00012 7 Euk 99 Eukaryota Metazoa Rotifera Monogononta 1.97 7 A 9 6 LP and WC
WC_18_Otu00013 40 Euk 93 Eukaryota Rhizaria Cercozoa Silicofilosea 0.68 4 C 42 8 LP and WC
WC_18_Otu00014 53 Euk 98 Eukaryota stramenopiles Bacillariophyta Coscinodiscophyceae 1.39 6 B 16 11 LP and WC
WC_18_Otu00016 13 Euk 95 Eukaryota Alveolata Ciliophora Intramacronucleata 0.33 3 B 101 5 LP and WC
WC_18_Otu00017 20 Euk 94 Eukaryota Alveolata Ciliophora Intramacronucleata 0.42 3 B 79 6 LP and WC
WC_18_Otu00018 66 Euk 85 Eukaryota Eukaryota Eukaryota Eukaryota 0.76 5 A 37 8 WC
WC_18_Otu00019 82 Euk 91 Eukaryota Rhizaria Cercozoa Cercomonadida 2.15 8 A 7 8 LP and WC
WC_18_Otu00020 56 Euk 92 Eukaryota Rhizaria Cercozoa Thecofilosea 1.21 6 A 21 4 WC
WC_18_Otu00021 7 Euk 96 Eukaryota Alveolata Ciliophora Intramacronucleata 0.28 3 B 123 6 LP and WC
WC_18_Otu00022 4 Euk 96 Eukaryota Viridiplantae Chlorophyta Prasinophyceae 0.88 5 A 33 9 LP and WC
WC_18_Otu00024 5 Euk 100 Eukaryota Haptophyceae Pavlovales Pavlovaceae 1.49 6 A 14 10 LP and WC
WC_18_Otu00025 43 Euk 100 Eukaryota Metazoa Arthropoda Crustacea 1.37 6 A 18 8 LP and WC
WC_18_Otu00026 12 Euk 95 Eukaryota Alveolata Ciliophora

0.39 3 B 83 5 LP and WC
WC_18_Otu00027 4 Euk 98 Eukaryota Alveolata Ciliophora Intramacronucleata 1 5 B 26 9 LP and WC
WC_18_Otu00031 6 Euk 100 Eukaryota Cryptophyta Cryptomonadales Cryptomonadaceae 0.51 4 B 61 10 LP and WC
WC_18_Otu00033 27 Euk 99 Eukaryota Viridiplantae Chlorophyta Mamiellophyceae 0.95 5 A 29 4 LP and WC
WC_18_Otu00034 6 Euk 98 Eukaryota stramenopiles Chrysophyceae Spumella-like 1.13 6 A 23 11 WC

157

WC_18_Otu00035 4 Euk 99 Eukaryota Metazoa Rotifera Monogononta 0.38 3 B 86 6 LP and WC
WC_18_Otu00039 47 Euk 91 Eukaryota Cryptophyta Cryptomonadales Cryptomonadaceae 0.31 3 B 107 6 LP and WC
WC_18_Otu00040 3 Euk 84 Eukaryota Eukaryota Eukaryota Eukaryota 1.01 6 A 25 6 WC
WC_18_Otu00041 5 Euk 99 Eukaryota Alveolata Ciliophora Intramacronucleata 0.88 5 A 31 7 WC
WC_18_Otu00042 1 Euk 92 Eukaryota stramenopiles Pirsonia Pirsonia 0.26 3 A 130 4 LP and WC
WC_18_Otu00045 37 Euk 99 Eukaryota Cryptophyta Cryptophyta Cryptophyta sp 0.17 2 B 189 8 LP and WC
WC_18_Otu00047 33 Euk 100 Eukaryota Cryptophyta Cryptomonadales Cryptomonadaceae 0.31 3 B 110 8 LP and WC
WC_18_Otu00048 6 Euk 97 Eukaryota stramenopiles Bacillariophyta Coscinodisco.. 0.3 3 B 113 8 LP and WC
WC_18_Otu00051 16 Euk 93 Eukaryota Ichthyosporea Ichthyophonida Anurofeca 0.42 3 C 78 8 LP and WC
WC_18_Otu00052 10 Euk 89 Eukaryota Alveolata

0.57 4 B 55 4 WC
WC_18_Otu00053 5 Euk 93 Eukaryota Euglenozoa Euglenida Euglenales 0.3 3 A 116 10 LP and WC
WC_18_Otu00055 29 Euk 87 Eukaryota Fungi Fungi Fungi 0.13 2 A 224 4 LP and WC
WC_18_Otu00056 39 Euk 100 Eukaryota Haptophyceae Pavlovales Pavlovaceae 0.27 3 A 127 7 LP and WC
WC_18_Otu00058 26 Euk 100 Eukaryota Viridiplantae Chlorophyta Prasinophyceae 0.13 2 B 218 4 LP and WC
WC_18_Otu00060 73 Euk 87 Eukaryota Eukaryota Eukaryota

0.32 3 B 105 4 LP and WC
WC_18_Otu00061 44 Euk 89 Eukaryota Fungi Chytridiomycota

0.39 3 B 81 8 WC
WC_18_Otu00062 52 Euk 95 Eukaryota Alveolata Ciliophora Intramacronucleata 0.41 3 B 80 6 LP and WC
WC_18_Otu00064 44 Euk 99 Eukaryota Haptophyceae Prymnesiales Prymnesiaceae 0.64 4 A 44 6 WC
WC_18_Otu00066 13 Euk 100 Eukaryota Viridiplantae Chlorophyta Chlorophyceae 0.37 3 C 90 11 WC
WC_18_Otu00068 1 Euk 100 Eukaryota Metazoa Rotifera Monogononta 0.29 3 B 118 4 LP and WC
WC_18_Otu00069 4 Euk 99 Eukaryota Haptophyceae Pavlovales Pavlovaceae 0.59 4 A 47 4 WC
WC_18_Otu00070 70 Euk 98 Eukaryota Rhodophyta Bangiophyceae Bangiales 0.32 3 B 104 4 LP and WC
WC_18_Otu00072 46 Euk 88 Eukaryota Cryptophyta Cryptomonadales Cryptomonadaceae 0.26 3 B 134 5 LP and WC
WC_18_Otu00073 32 Euk 94 Eukaryota stramenopiles Bacillariophyta Bacillariophyceae 0.38 3 B 87 10 LP and WC
WC_18_Otu00074 4 Euk 95 Eukaryota stramenopiles Pirsonia Pirsonia 0.26 3 C 131 9 WC
WC_18_Otu00075 13 Euk 99 Eukaryota Viridiplantae Chlorophyta Prasinophyceae 0.23 2 C 147 6 WC
WC_18_Otu00076 24 Euk 98 Eukaryota Cryptophyta Cryptomonadales Hemiselmidaceae 0.33 3 B 102 4 LP and WC
WC_18_Otu00078 60 Euk 91 Eukaryota Alveolata Ciliophora Intramacronucleata 0.15 2 B 205 4 LP and WC
WC_18_Otu00079 6 Euk 99 Eukaryota stramenopiles Bacillariophyta

0.52 4 C 60 11 WC
WC_18_Otu00080 65 Euk 92 Eukaryota stramenopiles Bacillariophyta Mediophyceae 0.3 3 B 114 6 LP and WC
WC_18_Otu00085 7 Euk 100 Eukaryota Alveolata Ciliophora Intramacronucleata 0.33 3 B 100 9 LP and WC
WC_18_Otu00086 77 Euk 95 Eukaryota Alveolata Ciliophora Intramacronucleata 0.26 3 A 135 7 WC
WC_18_Otu00091 6 Euk 88 Eukaryota Alveolata Ciliophora Intramacronucleata 0.21 2 B 156 4 WC
WC_18_Otu00093 57 Euk 100 Eukaryota Metazoa Annelida Polychaeta 0.2 2 A 169 4 WC
WC_18_Otu00094 8 Euk 87 Eukaryota Alveolata Ciliophora

0.17 2 B 192 8 WC
WC_18_Otu00095 70 Euk 90 Eukaryota Viridiplantae Chlorophyta Chlorophyceae 0.28 3 B 124 4 WC
WC_18_Otu00097 6 Euk 91 Eukaryota Alveolata Ciliophora Intramacronucleata 0.16 2 A 198 5 WC
WC_18_Otu00098 70 Euk 89 Eukaryota stramenopiles Synurophyceae Synurales 0.21 2 B 160 4 LP and WC
WC_18_Otu00100 35 Euk 94 Eukaryota Alveolata

0.46 3 A 69 7 WC
WC_18_Otu00103 66 Euk 92 Eukaryota stramenopiles

0.48 3 A 65 5 WC
WC_18_Otu00105 71 Euk 100 Eukaryota Viridiplantae Chlorophyta Prasinophyceae 0.15 2 B 204 4 LP and WC

158

WC_18_Otu00106 4 Euk 100 Eukaryota Viridiplantae Chlorophyta Chlorophyceae 0.17 2 A 191 6 WC
WC_18_Otu00107 71 Euk 94 Eukaryota Haptophyceae Pavlovales Pavlovaceae 0.18 2 B 181 4 WC
WC_18_Otu00110 49 Euk 91 Eukaryota Rhizaria Cercozoa Thecofilosea 0.31 3 B 108 6 WC
WC_18_Otu00113 45 Euk 87 Eukaryota Rhizaria Cercozoa Cercomonadida 0.24 2 A 140 7 WC
WC_18_Otu00115 73 Euk 99 Eukaryota Euglenozoa Kinetoplastida Bodonidae 0.27 3 B 128 4 WC
WC_18_Otu00117 71 Euk 92 Eukaryota Cryptophyta Cryptomonadales Cryptomonadaceae 0.19 2 B 178 4 LP and WC
WC_18_Otu00119 73 Euk 95 Eukaryota Viridiplantae Chlorophyta Chlorophyceae 0.19 2 B 173 5 LP and WC
WC_18_Otu00120 7 Euk 96 Eukaryota Alveolata Ciliophora Intramacronucleata 0.24 2 B 139 6 LP and WC
WC_18_Otu00121 45 Euk 96 Eukaryota Viridiplantae Chlorophyta Prasinophyceae 0.22 2 A 150 8 WC
WC_18_Otu00123 11 Euk 98 Eukaryota Alveolata Ciliophora Intramacronucleata 0.16 2 B 199 5 WC
WC_18_Otu00125 2 Euk 91 Eukaryota Alveolata Ciliophora Intramacronucleata 0.14 2 A 209 6 WC
WC_18_Otu00126 10 Euk 91 Eukaryota Viridiplantae Chlorophyta Prasinophyceae 0.17 2 C 188 5 WC
WC_18_Otu00129 7 Euk 92 Eukaryota Rhizaria Cercozoa Thecofilosea 0.23 2 C 144 9 WC
WC_18_Otu00134 49 Euk 98 Eukaryota stramenopiles Bacillariophyta Bacillariophyceae 0.28 3 A 126 6 WC
WC_18_Otu00135 21 Euk 90 Eukaryota Rhizaria Cercozoa Cercomonadida 0.23 2 A 143 8 WC
WC_18_Otu00136 15 Euk 91 Eukaryota Rhizaria Cercozoa Thecofilosea 0.2 2 A 165 4 WC
WC_18_Otu00139 11 Euk 100 Eukaryota stramenopiles Eustigmatophyceae Eustigmatales 0.25 2 A 137 5 WC
WC_18_Otu00140 8 Euk 100 Eukaryota Alveolata Dinophyceae Stoeckeria 0.38 3 C 88 9 LP and WC
WC_18_Otu00146 14 Euk 100 Eukaryota Metazoa Rotifera Monogononta 0.3 3 A 117 5 WC
WC_18_Otu00157 4 Euk 86 Eukaryota Alveolata Dinophyceae Gymnodiniales 0.11 2 B 236 4 WC
WC_18_Otu00158 39 Euk 100 Eukaryota Haptophyceae Isochrysidales Isochrysidaceae 0.31 3 A 106 5 WC
WC_18_Otu00160 14 Euk 92 Eukaryota stramenopiles Chrysophyceae Chromulinales 0.19 2 A 171 6 WC
WC_18_Otu00161 48 Euk 87 Eukaryota Alveolata Ciliophora

0.11 2 B 240 5 WC
WC_18_Otu00165 6 Euk 98 Eukaryota Euglenozoa Kinetoplastida Bodonidae 0.18 2 B 183 6 WC
WC_18_Otu00168 7 Euk 86 Eukaryota Fungi Dikarya Basidiomycota 0.16 2 A 195 4 WC
WC_18_Otu00171 72 Euk 95 Eukaryota Viridiplantae Chlorophyta Chlorophyceae 0.12 2 B 235 5 WC
WC_18_Otu00176 49 Euk 95 Eukaryota Alveolata Ciliophora Intramacronucleata 0.13 2 B 222 5 WC
WC_18_Otu00178 21 Euk 100 Eukaryota Viridiplantae Chlorophyta Trebouxiophyceae 0.23 2 C 142 9 WC
WC_18_Otu00181 2 Euk 98 Eukaryota Alveolata Ciliophora Intramacronucleata 0.23 2 B 146 4 WC
WC_18_Otu00182 4 Euk 100 Eukaryota Cryptophyta Cryptomonadales Cryptomonadaceae 0.09 1 B 251 6 WC
WC_18_Otu00190 76 Euk 96 Eukaryota Viridiplantae Chlorophyta Chlorophyceae 0.1 1 B 245 5 WC
WC_18_Otu00208 2 Euk 93 Eukaryota Centroheliozoa Acanthocystidae Pterocystis 0.13 2 B 221 6 WC
WC_18_Otu00212 71 Euk 86 Eukaryota Rhizaria Cercozoa Cercozoa 0.11 2 B 237 4 WC
WC_18_Otu00215 6 Euk 99 Eukaryota Alveolata Dinophyceae Peridiniales 0.19 2 B 174 4 WC
WC_18_Otu00222 2 Euk 100 Eukaryota Viridiplantae Chlorophyta Trebouxiophyceae 0.2 2 C 168 9 WC
WC_18_Otu00223 71 Euk 89 Eukaryota Alveolata Ciliophora Intramacronucleata 0.09 1 B 250 5 WC
WC_18_Otu00225 1 Euk 95 Eukaryota Alveolata Ciliophora Intramacronucleata 0.26 3 C 132 9 WC
WC_18_Otu00229 5 Euk 81 Eukaryota Cryptophyta Cryptophyta Cryptophyceae 0.13 2 A 220 7 WC
WC_18_Otu00234 70 Euk 95 Eukaryota Viridiplantae Chlorophyta Chlorophyceae 0.12 2 B 229 4 WC
WC_18_Otu00243 68 Euk 89 Eukaryota stramenopiles Synurophyceae Synurales 0.15 2 B 202 4 WC
WC_18_Otu00247 68 Euk 91 Eukaryota Rhizaria Cercozoa Thecofilosea 0.1 1 B 244 5 WC

159

WC_18_Otu00252 30 Euk 98 Eukaryota Alveolata Apicomplexa Colpodellidae 0.11 2 A 241 6 WC
WC_18_Otu00259 33 Euk 99 Eukaryota stramenopiles Chrysophyceae Spumella-like 0.16 2 B 200 8 WC
WC_18_Otu00267 7 Euk 98 Eukaryota
Choanoflagellid
a Codonosigidae Desmarella 0.08 1 B 262 4 WC
WC_18_Otu00288 71 Euk 86 Eukaryota Fungi Blastocladiomycota

0.08 1 B 264 4 WC
WC_18_Otu00295 12 Euk 99 Eukaryota stramenopiles Bacillariophyta Bacillariophyceae 0.2 2 A 166 11 WC
WC_18_Otu00377 30 Euk 99 Eukaryota Viridiplantae Chlorophyta Chlorophyceae 0.11 2 C 242 5 WC
WC_18_Otu00394 38 Euk 92 Eukaryota Viridiplantae Chlorophyta Prasinophyceae 0.08 1 B 257 4 WC
WC_Chla 27 Env null Chla Chla Chla Chla 1 4 C null 12 LP and WC
WC_DO 45 Env null DO DO DO DO 1 4 C null 12 LP and WC
WC_pH 4 Env null pH pH pH pH 1 4 C null 12 LP and WC
WC_Ppar 63 Env null Ppar Ppar Ppar Ppar 1 4 A null 12 LP and WC
WC_PSU 29 Env null PSU PSU PSU PSU 1 4 C null 12 LP and WC
WC_Temperature 37 Env null
Temperatur
e Temperature Temperature Temperature 1 4 C null 12 LP and WC

160

Table S3. A summary of the variables (18s and 16s rDNA OTUs plus environmental variables) and correlations found in Lebanon Pool and Wilson Creek,
their best percent hit to SILVA (v108) reference database, nested taxonomic identifications

seasonal abundance pattern: A (November to April), B (May to October) in Lebanon Pool (LP) and Wilson Creek (WC)

ID # Correlations Gene
Similarity to
SILVA Domain Phylum Class Other
LP Seasonal
Abund
WC Seasonal
Abund
Chla null Env 0 Chla Chla Chla Chla C C

DO 9 Env 0 DO DO DO DO C C

Otu00001 3 Euk 100 Eukaryota Haptophyceae Prymnesiales Prymnesiaceae A A

Otu00003 null Euk 97 Eukaryota Katablepharidophyta Katablepharidaceae Katablepharis C A

Otu00004 14 Euk 99 Eukaryota stramenopiles Diatom Coscinodiscophyceae B B

Otu00005 null Euk 88 Eukaryota Fungi Chytridiomycota Monoblepharidomycetes C C

Otu00008 17 Euk 93 Eukaryota Viridiplantae Chlorophyta Prasinophyceae B B

Otu00010 null Euk 95 Eukaryota stramenopiles Chrysophyceae Chromulinales A A

Otu00011 4 Euk 100 Eukaryota stramenopiles Chrysophyceae Chromulinales A A

Otu00012 null Euk 99 Eukaryota Metazoa Rotifera Monogononta A A

Otu00013 3 Euk 93 Eukaryota Rhizaria Cercozoa Silicofilosea B C

Otu00014 21 Euk 98 Eukaryota stramenopiles Diatom Coscinodiscophyceae B B

Otu00016 6 Euk 95 Eukaryota Alveolata Ciliate Intramacronucleata B B

Otu00017 null Euk 94 Eukaryota Alveolata Ciliophora Intramacronucleata B B

Otu00019 12 Euk 91 Eukaryota Rhizaria Cercozoa Cercomonadida A A

Otu00021 3 Euk 96 Eukaryota Alveolata Ciliate Intramacronucleata B B

Otu00022 1 Euk 96 Eukaryota Viridiplantae Chlorophyta Prasinophyceae B A

Otu00024 null Euk 100 Eukaryota Haptophyceae Pavlovales Pavlovaceae A A

Otu00025 null Euk 100 Eukaryota Metazoa Arthropoda Crustacea B A

Otu00026 1 Euk 95 Eukaryota Alveolata Ciliate Postciliodesmatophora B B

Otu00027 null Euk 98 Eukaryota Alveolata Ciliophora Intramacronucleata B B

Otu00031 5 Euk 100 Eukaryota Cryptophyta Cryptophyte Cryptomonadaceae B B

Otu00033 8 Euk 99 Eukaryota Viridiplantae Chlorophyte Mamiellophyceae A A

Otu00035 null Euk 99 Eukaryota Metazoa Rotifera Monogononta B B

Otu00039 21 Euk 91 Eukaryota Cryptophyta Cryptophyte Cryptomonadaceae B B

Otu00042 null Euk 92 Eukaryota stramenopiles Pirsonia Pirsonia A A

Otu00045 12 Euk 99 Eukaryota Cryptophyta Cryptophyte Cryptophyta sp. CR-MAL11 B B

Otu00047 13 Euk 100 Eukaryota Cryptophyta Cryptophyte Cryptomonadaceae B B

Otu00048 null Euk 97 Eukaryota stramenopiles Bacillariophyta Coscinodiscophyceae A B

161

Otu00051 2 Euk 93 Eukaryota Ichthyosporea Ichthyophonida Anurofeca B C

Otu00053 null Euk 93 Eukaryota Euglenozoa Euglenida Euglenales A A

Otu00055 5 Euk 87 Eukaryota Fungi Fungi Fungi A A

Otu00056 9 Euk 100 Eukaryota Haptophyceae Pavlovales Pavlovaceae A A

Otu00058 6 Euk 100 Eukaryota Viridiplantae Chlorophyte Prasinophyceae C B

Otu00060 18 Euk 87 Eukaryota Eukaryota Eukaryota Eukaryota B B

Otu00062 10 Euk 95 Eukaryota Alveolata Ciliate Intramacronucleata B B

Otu00068 1 Euk 100 Eukaryota Metazoa Rotifera Monogononta B B

Otu00070 9 Euk 98 Eukaryota Rhodophyta Bangiophyceae Bangiales B B

Otu00072 17 Euk 88 Eukaryota Cryptophyta Cryptophyte Cryptomonadaceae B B

Otu00073 6 Euk 94 Eukaryota stramenopiles Diatom Bacillariophyceae B B

Otu00076 5 Euk 98 Eukaryota Cryptophyta Cryptophyte Hemiselmidaceae B B

Otu00078 19 Euk 91 Eukaryota Alveolata Ciliate Intramacronucleata B B

Otu00080 19 Euk 92 Eukaryota stramenopiles Diatom Mediophyceae B B

Otu00085 2 Euk 100 Eukaryota Alveolata Ciliate Intramacronucleata B B

Otu00098 22 Euk 89 Eukaryota stramenopiles Synurophyceae Synurales B B

Otu00105 29 Euk 100 Eukaryota Viridiplantae Chlorophyte Prasinophyceae B B

Otu00117 18 Euk 92 Eukaryota Cryptophyta Cryptophyte Cryptomonadaceae B B

Otu00119 21 Euk 95 Eukaryota Viridiplantae Chlorophyte Chlorophyceae B B

Otu00120 null Euk 96 Eukaryota Alveolata Ciliophora Intramacronucleata B B

Otu00140 null Euk 100 Eukaryota Alveolata Dinophyceae Stoeckeria B C

Otu1 14 Bact 94 Chloroplast Chloroplast Chloroplast Chloroplast A A

Otu10 2 Bact 100 Chloroplast Chloroplast Chloroplast Odontella sinensis B B

Otu100 3 Bact 100 Chloroplast Chloroplast Chloroplast Chloroplast B B

Otu107 11 Bact 98 Bacteria Cyanobacteria Synechococcus Synechococcus sp. EW15 C B

Otu108 null Bact 100 Bacteria Actinobacteria Actinobacteria Frankiales B A

Otu115 null Bact 100 Bacteria Actinobacteria Acidimicrobiia Acidimicrobiales B B

Otu120 14 Bact 97 Bacteria Cyanobacteria Cyanobacteria SubgroupI B B

Otu129 null Bact 100 Bacteria Bacteroidetes Flavobacteria Flavobacteriales C C

Otu13 2 Bact 100 Bacteria Verrucomicrobia Spartobacteria Chthoniobacterales A A

Otu136 4 Bact 100 Bacteria Chloroflexi SL56 marine group SL56 marine group B B

Otu139 1 Bact 100 Bacteria Proteobacteria Betaproteobacteria Burkholderiales C B

Otu14 15 Bact 97 Bacteria Proteobacteria Alphaproteobacteria Rhodobacterales A A

Otu149 null Bact 100 Bacteria Proteobacteria Alphaproteobacteria Rhodobacterales B B

Otu15 13 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Sphingomonadales A A

162

Otu152 2 Bact 100 Chloroplast Chloroplast Chloroplast Hemiselmis virescens B B

Otu16 25 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Rhodobacterales A A

Otu163 null Bact 100 Bacteria Proteobacteria Alphaproteobacteria Sphingomonadales B B

Otu165 8 Bact 0 Bacteria Bacteria Bacteria Bacteria C B

Otu166 10 Bact 100 Bacteria Cyanobacteria Cyanobacteria SubsectionI C B

Otu17 15 Bact 100 Bacteria Proteobacteria Betaproteobacteria Burkholderiales A A

Otu170 1 Bact 95 Chloroplast Chloroplast Chloroplast Pyramimonas olivacea B A

Otu171 null Bact 97 Chloroplast Chloroplast Chloroplast Chloroplast C C

Otu18 28 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Sphingomonadales A A

Otu19 18 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Rhodobacterales A A

Otu2 7 Bact 100 Bacteria Cyanobacteria Synechococcus Synechococcus A A

Otu204 null Bact 100 Bacteria Verrucomicrobia Opitutae Opitutae B C

Otu21 5 Bact 100 Bacteria Proteobacteria Betaproteobacteria Burkholderiales C A

Otu212 1 Bact 100 Bacteria Bacteroidetes Sphingobacteria Sphingobacteriales B B

Otu22 14 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Caulobacterales A A

Otu23 null Bact 100 Bacteria Bacteroidetes Flavobacteria Flavobacteriales C A

Otu26 15 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Caulobacterales A A

Otu27 5 Bact 100 Bacteria Proteobacteria Betaproteobacteria Burkholderiales C C

Otu28 23 Bact 100 Bacteria Bacteroidetes Flavobacteria Flavobacteriales A A

Otu29 1 Bact 100 Bacteria Proteobacteria Betaproteobacteria Burkholderiales C C

Otu3 1 Bact 97 Bacteria Verrucomicrobia Spartobacteria Chthoniobacterales C C

Otu30 2 Bact 100 Bacteria Bacteroidetes Sphingobacteria Sphingobacteriales A A

Otu33 null Bact 100 Bacteria Bacteroidetes Sphingobacteria Sphingobacteriales B C

Otu34 2 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Rhizobiales C C

Otu35 null Bact 100 Bacteria Bacteroidetes Flavobacteria Flavobacteriales C A

Otu36 5 Bact 100 Bacteria Proteobacteria Gammaproteobacteria Xanthomonadales C C

Otu37 1 Bact 95 Bacteria Actinobacteria Actinobacteria Frankiales B C

Otu38 null Bact 100 Bacteria Bacteroidetes Cytophagia Cytophagales B C

Otu39 16 Bact 100 Bacteria Cyanobacteria Cyanobacteria Planktothrix A A

Otu4 7 Bact 97 Bacteria Proteobacteria Alphaproteobacteria SAR11 clade B C

Otu40 2 Bact 100 Bacteria Cyanobacteria SubsectionIV SubgroupI A A

Otu41 2 Bact 100 Bacteria Actinobacteria Acidimicrobiia Acidimicrobiales C C

Otu42 12 Bact 100 Bacteria Bacteroidetes Sphingobacteria Sphingobacteriales B C

Otu43 3 Bact 100 Bacteria Cyanobacteria Cyanobacteria Merismopedia C C

Otu45 null Bact 100 Bacteria Proteobacteria Gammaproteobacteria Xanthomonadales C C

163

Otu46 13 Bact 100 Bacteria Proteobacteria Betaproteobacteria Burkholderiales A A

Otu47 null Bact 100 Bacteria Proteobacteria Betaproteobacteria Methylophilales B C

Otu48 20 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Rhodobacterales B B

Otu49 9 Bact 100 Bacteria Bacteroidetes Sphingobacteria Sphingobacteriales B B

Otu5 null Bact 97 Bacteria Bacteroidetes Sphingobacteria Sphingobacteriales A C

Otu50 15 Bact 97 Bacteria Cyanobacteria Cyanobacteria Leptolyngbya C B

Otu51 8 Bact 100 Bacteria Cyanobacteria Synechococcus Synechococcus B B

Otu53 18 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Sphingomonadales A A

Otu55 1 Bact 100 Bacteria Cyanobacteria Cyanobacteria SubsectionI C C

Otu56 3 Bact 100 Bacteria Actinobacteria Acidimicrobiia Acidimicrobiales B C

Otu57 26 Bact 100 Bacteria Proteobacteria Betaproteobacteria Burkholderiales A A

Otu6 12 Bact 100 Bacteria Bacteroidetes Flavobacteria Flavobacteriales A A

Otu65 10 Bact 98 Bacteria Cyanobacteria Synechococcus Synechococcus C B

Otu68 1 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Rhodobacterales C C

Otu69 26 Bact 93 Chloroplast Chloroplast Chloroplast Chloroplast B B

Otu7 7 Bact 95 Bacteria Lentisphaerae Lentisphaeria WCHB1-41 B C

Otu70 1 Bact 100 Bacteria Bacteroidetes Flavobacteria Flavobacteriales B B

Otu73 15 Bact 100 Bacteria Bacteroidetes Sphingobacteria Sphingobacteriales B B

Otu74 null Bact 100 Bacteria Proteobacteria Alphaproteobacteria Caulobacterales A B

Otu75 16 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Rhizobiales C B

Otu77 18 Bact 100 Bacteria Cyanobacteria Cyanobacteria Prochlorothrix B B

Otu78 9 Bact 98 Bacteria Cyanobacteria Cyanobacteria SubgroupI C B

Otu79 22 Bact 100 Bacteria Cyanobacteria Cyanobacteria SubgroupI B B

Otu8 14 Bact 100 Bacteria Proteobacteria Alphaproteobacteria SAR11 clade A A

Otu81 13 Bact 98 Bacteria Cyanobacteria Synechococcus Synechococcus B B

Otu84 11 Bact 0 Bacteria Bacteria Bacteria Bacteria B B

Otu86 1 Bact 100 Bacteria Actinobacteria Acidimicrobiia Acidimicrobiales C C

Otu88 null Bact 100 Bacteria Proteobacteria Alphaproteobacteria Sphingomonadales C C

Otu9 19 Bact 100 Bacteria Bacteroidetes Flavobacteria Flavobacteriales A A

Otu92 7 Bact 100 Bacteria Proteobacteria Alphaproteobacteria Rhodobacterales A A

pH null Env 0 pH pH pH pH C C

Ppar 9 Env 0 Ppar Ppar Ppar Ppar C A

PSU 2 Env 0 PSU PSU PSU PSU C C

Temperature 10 Env 0 Temperature Temperature Temperature Temperature C C

164

Supplemental Information
Supplemental Figure 1: The average relative percent abundance of a microbial OTU plotted
against its frequency of occurrence (4 to 12 months) in A) Lebanon Pool and B) Wilson Creek.

Supplemental Figure 2: Histograms of all the permuted p-values associated with all Spearman
correlations from A) Lebanon Pool and B) Wilson Creek.

Supplemental Figure 3: The frequency of a microbial OTUs occurrence plotted against its
number of significant Spearman correlations with other OTUs or environmental variables, for the
data from Lebanon Pool (A) and Wilson Creek (B). Lines indicate the 95% confidence intervals
for the number of correlations per OTU.

Supplemental Figure 4: Log distributions representing the number of significant spearman
correlations per microbial OTU or environmental variable within the networks from A) Lebanon
Pool, B) Wilson Creek, and C) Shared (both locations) . Closed symbols represent the
distribution from the experimental microbial association networks, and open symbols represent
distributions constructed from Erdös-Réyni model networks of the same size as the experimental
networks. The upper inset graphs in each panel show Poisson distributions fit to the Erdös-Réyni
model data with r
2
s of: A) Lebanon Pool = 0.8, B) Wilson Creek = 0.87, and C) Shared = 0.87.
The distribution for the shared microbial association network (panel C lower inset) had a
moderate fit r
2
of 0.6 to a power curve.

165

Supplemental Figure 5: Graphs depicting the annual trajectories of Prymnesium affiliated OTUs
(A and B) and P. parvum cell counts (C) in Lebanon Pool (A and C) and Wilson Creek (B and
C). Panels A and B show the relative abundances of 18s OTU #1 (solid symbols) plus 16s
OTU#1 and OTU#12 (open symbols) in Lebanon Pool (A) and Wilson Creek (B). Panel C shows
the abundance of P. parvum cell counts in Lebanon Pool (black symbols and lines) and Wilson
Creek (grey symbols and lines).

166

167

168

169

170

171

Bibliography

Allison SD, Martiny JBH (2008) Resistance, resilience, and redundancy in microbial communities.
Proceedings of the National Academy of Sciences, 105, 11512-11519.
Altschul SF, Madden TL, Schäffer AA, et al. (1997) Gapped BLAST and PSI-BLAST: a new generation of
protein database search programs. Nucleic Acids Research, 25, 3389-3402.
Amaral-Zettler LA, Gomez F, Zettler E, et al. (2002) Microbiology: eukaryotic diversity in Spain's River of
Fire. Nature, 417, 137-137.
Amaral-Zettler LA, McCliment EA, Ducklow HW, et al. (2009) A method for studying protistan diversity
using massively parallel sequencing of V9 hypervariable regions of small-subunit ribosomal RNA
genes. PLoS One, 4, e6372.
Andersson AF, Riemann L, Bertilsson S (2009) Pyrosequencing reveals contrasting seasonal dynamics of
taxa within Baltic Sea bacterioplankton communities. The ISME journal, 4, 171-181.
Andersson AF, Riemann L, Bertilsson S (2010) Pyrosequencing reveals contrasting seasonal dynamics of
taxa within Baltic Sea bacterioplankton communities. The ISME journal, 4, 171-181.
Assenov Y, Ramírez F, Schelhorn S-E, et al. (2008) Computing topological parameters of biological
networks. Bioinformatics, 24, 282-284.
Azam F, Fenchel T, Field JG, et al. (1983) The ecological role of water-column microbes in the sea.
Marine Ecology Progress Series, 10, 257-263.
Bader G, Hogue C (2003) An automated method for finding molecular complexes in large protein
interaction networks. BMC Bioinformatics, 4, 2.
Barabasi A-L, Oltvai ZN (2004) Network biology: understanding the cell's functional organization. Nature
Reviews Genetics, 5, 101-113.
172

Barberan A, Bates ST, Casamayor EO, et al. (2012) Using network analysis to explore co-occurrence
patterns in soil microbial communities. The ISME journal, 6, 343-351.
Bass D, Cavalier-Smith T (2004) Phylum-specific environmental DNA analysis reveals remarkably high
global biodiversity of Cercozoa (Protozoa). International Journal of Systematic and Evolutionary
Microbiology, 54, 2393-2404.
Bass D, Howe AT, Mylnikov AP, et al. (2009) Phylogeny and Classification of Cercomonadida:
Cercomonas, Eocercomonas, Paracercomonas, and Cavernomonas gen. n. Protist, 160, 483-521.
Behnke A, Engel M, Christen R, et al. (2011) Depicting more accurate pictures of protistan community
complexity using pyrosequencing of hypervariable SSU rRNA gene regions. Environmental
Microbiology, 13, 340-349.
Berninger UG, Wickham SA, Finlay BJ (1993) Trophic coupling within the microbial food web: a study
with fine temporal resolution in a eutrophic freshwater ecosystem. Freshwater Biology, 30, 419-
432.
Bray J, Curtis J (1957) An ordination of the upland forest communities of southern Wisconsin. Ecological
Monographs, 27, 325-349.
Brooks BW, James SV, Valenti JTW, et al. (2010) Comparative toxicity of Prymnesium parvum in inland
Waters Journal of the American Water Resources Association, 46, 45-62.
Bunge J (2011). Estimating the number of species with CatchAll. Paper presented at: PROCEEDINGS OF
PACIFIC SYMPOSIUM ON BIOCOMPUTING (Kohala Coast, Hawaii).
Buskey EJ, Liu H, Collumb C, et al. (2001) The decline and recovery of a persistent Texas brown tide algal
bloom in the Laguna Madre (Texas, USA). Estuaries, 24, 337-346.
Caraco N, Cole JJ, Raymond PA, et al. (1997) Zebra mussel invasion in a large, turbid river: phytoplankton
response to increased grazing ecology. Ecology, 78, 588-602.
173

Caron DA, Countway PD (2009) Hypotheses on the role of the protistan rare biosphere in a changing
world. Aquatic Microbial Ecology, 57, 227-238.
Caron DA, Countway PD, Savai P, et al. (2009) Defining DNA-based operational taxonomic units for
microbial-eukaryote ecology. Applied and Environmental Microbiology, 75, 5797-5808.
Chao A (2005). Species estimation and applications, Vol 12 (New York: Wiley).
Chao A, Lee SM (1992) Estimating the number of classes via sample coverage. Journal of the American
Statistical Association, 87, 210-217.
Clarke KR (1990) Comparisons of dominance curves. Journal of Experimental Marine Biology and
Ecology, 138, 143-157.
Clarke KR (1993) Non-parametric multivariate analyses of changes in community structure. Australian
Journal of Ecology, 18, 117–143.
Clarke KR, and , Warwick RM (2001). Change in Marine Communities: An Approach to Statistical Analysis
and Interpretation, 2nd edn. edn (Plymouth, UK).
Clarke KR, Somerfield PJ, Chapman MG (2006) On resemblance measures for ecological studies,
including taxonomic dissimilarities and a zero-adjusted Bray–Curtis coefficient for denuded
assemblages. Journal of Experimental Marine Biology and Ecology, 330, 55-80.
Clarke KR, Somerfield PJ, Gorley RN (2008) Testing of null hypotheses in exploratory community
analyses: similarity profiles and biota-environment linkage. Journal of Experimental Marine
Biology and Ecology, 366, 56-69.
Cline MS, Smoot M, Cerami E, et al. (2007) Integration of biological networks and gene expression data
using Cytoscape. Nature Protocols, 2, 2366-2382.
Cole JJ (1982) Interactions between bacteria and algae in aquatic ecosystems. Annual Review of Ecology
and Systematics, 13, 291-314.
Connell JH (1978) Diversity in tropical rain forests and coral reefs. Science, 199, 1302-1310.
174

Countway PD, Gast RJ, Dennett MR, et al. (2007) Distinct protistan assemblages characterize the
euphotic zone and deep sea (2500 m) of the western N. Atlantic (Sargasso Sea and Gulf Stream).
Environmental Microbiology, 9, 1219-1232.
Edgcomb V, Orsi W, Bunge J, et al. (2011) Protistan microbial observatory in the Cariaco Basin,
Caribbean. I. Pyrosequencing vs Sanger insights into species richness. The ISME journal, 5, 1344-
1356.
Edgcomb VP, Kysela DT, Teske A, et al. (2002) Benthic eukaryotic diversity in the Guaymas Basin
hydrothermal vent environment. Proceedings of the National Academy of Sciences of the United
States of America, 99, 7658-7662.
Eiler A, Heinrich F, Bertilsson S (2012) Coherent dynamics and association networks among lake
bacterioplankton taxa. The ISME journal, 6, 330-342.
Erdös Pa, Réyni A (1960) On the evolution of random graphs. Institute of Mathematics Hungarian
Academy of Sciences, 5, 17-61.
Estes JA, Terborgh J, Brashares JS, et al. (2011) Trophic downgrading of planet Earth. Science, 333, 301-
306.
Evardsen B, Imai I (2006). The ecology of harmful flagellates within Prymnesiophyceae and
Raphidophyceae. In Ecology of Harmful Algae, E. Granéli, and J. Turner, eds. (Netherlands:
Springer-Verlag), pp. 67-79.
Faust K, Sathirapongsasuti JF, Izard J, et al. (2012) Microbial co-occurrence relationships in the human
microbiome. PLoS Computer Biology, 8, e1002606.
Fistarol G, Legrand C, Selander E, et al. (2004) Allelopathy in Alexandrium spp.: effect on a natural
plankton community and on algal monocultures. Aquatic Microbial Ecology, 35, 45–56.
Fistarol GC, Legrand C., E G (2003) Allelopathic effect of Prymnesium parvum on a natural plankton
community. Marine Ecology Progress Series, 255, 115–125.
175

Floder S, Sommer U (1999) Diversity in planktonic communities: an experimental test of the
intermediate disturbance hypothesis. Limnology and Oceanography, 44, 1114-1119.
Fuhrman JA (2009) Microbial community structure and its functional implications. Nature, 459, 193-199.
Gihring TM, Green SJ, Schadt CW (2012) Massively parallel rRNA gene sequencing exacerbates the
potential for biased community diversity comparisons due to variable library sizes.
Environmental Microbiology, 14, 285-290.
Gilbert JA, Field D, Swift P, et al. (2009) The seasonal structure of microbial communities in the Western
English Channel. Environmental Microbiology, 11, 3132-3139.
Gilbert JA, Steele JA, Caporaso JG, et al. (2012) Defining seasonal marine microbial community dynamics.
The ISME journal, 6, 298-308.
Gobler CJ, Lonsdale DJ, GL B (2005) A review of the causes, effects, and potential management of
harmful brown tide blooms caused by Aureococcus anophagefferens (Hargraves et Sieburth).
Estuaries 28, 726-749.
Gobler CJ, Sunda WG (2012) Ecosystem disruptive algal blooms of the brown tide species, Aureococcus
anophagefferens and Aureoumbra lagunensis. Harmful Algae, 14, 36-45.
Hambright KD, Zamor RM, Easton JD, et al. (2010) Temporal and spatial variability of an invasive
toxigenic protist in a North American subtropical reservoir. Harmful Algae, 9, 568-577.
Heino J, Soininen J (2010) Are common species sufficient in describing turnover in aquatic
metacommunities along environmental and spatial gradients? Limnology and Oceanography, 55,
2397-2402.
Henrikson JC, Gharfeh MS, Easton AC, et al. (2010) Reassessing the ichthyotoxin profile of cultured
Prymnesium parvum (golden algae) and comparing it to samples collected from recent
freshwater bloom and fish kill events in North America. Toxicon, 55, 1396-1404.
176

Huse SM, Dethlefsen L, Huber JA, et al. (2008) Exploring microbial diversity and taxonomy using SSU
rRNA hypervariable tag sequencing. PLoS Genetics, 4, e1000255.
Huse SM, Huber JA, Morrison HG, et al. (2007) Accuracy and quality of massively parallel DNA
pyrosequencing. Genome Biology, 8, R143.
Huse SM, Welch DM, Morrison HG, et al. (2010) Ironing out the wrinkles in the rare biosphere through
improved OTU clustering. Environmental Microbiology, 12, 1889-1898.
Ibelings BW, De Bruin A, Kagami M, et al. (2004) Host parasite interactions between freshwater
phytoplankton and chytrid fungi (Chytridiomycota). Journal of Phycology, 40, 437-453.
Igarashi T, Satake M, Yasumoto T (1999) Structures and partial stereochemical assignments for
Prymnesin-1 and Prymnesin-2: potent hemolytic and ichthyotoxic glycosides isolated from the
red ride alga Prymnesium parvum. Journal of the American Chemical Society, 121, 8499-8511.
Ings TC, Montoya JM, Bascompte J, et al. (2009) Review: Ecological networks – beyond food webs.
Journal of Animal Ecology, 78, 253-269.
Jaccard P (1908) Nouvelles recherches sur la distribution florale. Bulletin de la Société Vaudoise des
Sciences Naturelles, 44, 223-270.
James TY, Letcher PM, Longcore JE, et al. (2006) A molecular phylogeny of the flagellated fungi
(Chytridiomycota) and description of a new phylum (Blastocladiomycota). Mycologia, 98, 860-
871.
Jobard M, Rasconi S, Sime-Ngando T (2010) Diversity and functions of microscopic fungi: a missing
component in pelagic food webs. Aquatic Sciences - Research Across Boundaries, 72, 255-268.
Jones AJ, Liao VTS, Najar FZ, et al. (2012) Seasonality and disturbance: annual pattern and response of
the bacterial and microbial eukaryotic assemblages in a freshwater ecosystem. Submitted.
Jones MD, Forn I, Gadelha C, et al. (2011) Discovery of novel intermediate forms redefines the fungal
tree of life. Nature, 474, 200-203.
177

Jones SE, C. C, Kratz TK, et al. (2008) Typhoons initiate predictable change in aquatic bacterial
communities. Limnology and Oceanography, 53, 1319-1326.
Kan J, Byron CC, Wang K, et al. (2006) Bacterioplankton community in Chesapeake Bay: predictable or
random assemblages. Limnology and Oceanography, 51, 2157-2169.
Kim DY, Countway PD, Gast RJ, et al. (2011) Rapid shifts in the structure and composition of a protistan
assemblage during bottle incubations affect estimates of total protistan species richness.
Microbial Ecology, 62, 383-398.
Kunin V, Engelbrektson A, Ochman H, et al. (2010) Wrinkles in the rare biosphere: pyrosequencing errors
can lead to artificial inflation of diversity estimates. Environmental Microbiology, 12, 118-123.
Lambshead PJD, Platt HM, Shaw KM (1983) The detection of differences among assemblages of marine
benthic species based on an assessment of dominance and diversity. Journal of Natural History,
17, 859-874.
Lefevre E, Bardot C, Noel C, et al. (2007) Unveiling fungal zooflagellates as members of freshwater
picoeukaryotes: evidence from a molecular diversity study in a deep meromictic lake.
Environmental Microbiology, 9, 61-71.
Lefèvre E, Roussel B, Amblard C, et al. (2008) The molecular diversity of freshwater picoeukaryotes
reveals high occurrence of putative parasitoids in the plankton. PLoS One, 3, e2324.
Lim EL, Dennett MR, Caron DA (1999) The ecology of Paraphysomonas imperforata based on studies
employing oligonucleotide probe identification in coastal water samples and enrichment
cultures. Limnology and Oceanography, 44, 37-51.
Lima-Mendez G, van Helden J (2009) The powerful law of the power law and other myths in network
biology. Molecular BioSystems, 5, 1482-1493.
178

Lopez-Garcia P, Philippe H, Gail F, et al. (2003) Autochthonous eukaryotic diversity in hydrothermal
sediment and experimental microcolonizers at the Mid-Atlantic Ridge. Proceedings of the
National Academy of Sciences of the United States of America, 100, 697-702.
Margulies M, Egholm M, Altman WE, et al. (2005) Genome sequencing in microfabricated high-density
picolitre reactors. Nature, 437, 376-380.
Martin-Cereceda M, G N, Young JR (2003) Grazing by Prymnesium parvum on small planktonic diatoms.
Aquatic Microbial Ecology, 33, 191-199.
McPherson RA, Fiebrich CA, Crawford KC, et al. (2007) Statewide monitoring of the mesoscale
environment: A technical update on the Oklahoma Mesonet. Journal of Atmospheric and
Oceanic Technology, 24, 301-321.
Medinger R, Nolte V, Pandey RV, et al. (2010) Diversity in a hidden world: potential and limitation of
next-generation sequencing for surveys of molecular diversity of eukaryotic microorganisms.
Molecular Ecology, 19 Suppl 1, 32-40.
Michaloudi E, Moustaka-Gouni M, Gkelis S, et al. (2008) Plankton community structure during an
ecosystem disruptive algal bloom of Prymnesium parvum. Journal of Plankton Research, 31, 301-
309.
Nejstgaard JC, Solberg PT (1996) Repression of copepod feeding and fecundity by the toxic haptophyte
Prymnesium patelliferum. Sarsia, 81, 339-344.
Newton RJ, Jones SE, Eiler A, et al. (2011) A guide to the natural history of freshwater lake bacteria.
Microbiology and Molecular Biology Reviews, 75, 14-49.
Oikonomou A, Katsiapi M, Karayanni H, et al. (2012) Plankton microorganisms coinciding with two
consecutive mass fish kills in a newly reconstructed lake. Scientific World Journal, 2012, 504135.
Parsons TR, Maita Y, Lalli CM (1984). A manual of chemical and biological methods for seawater analysis
(Oxford: Pergamon Press).
179

Pedros-Alio C (2006) Marine microbial diversity: can it be determined? Trends Microbiol, 14, 257-263.
Peura S, Eiler A, Hiltunen M, et al. (2012) Bacterial and phytoplankton responses to nutrient
amendments in a boreal lake differ according to season and to taxonomic resolution. PLoS One,
7, 38552.
Pinhassi Ja, Hagström Å (2000) Seasonal succession in marine bacterioplankton. Aquatic Microbial
Ecology, 21, 245–256.
Poulin R (2010) Network analysis shining light on parasite ecology and diversity. Trends in Parasitology,
26, 492-498.
Proulx SR, Promislow DE, Phillips PC (2005) Network thinking in ecology and evolution. Trends Ecol Evol,
20, 345-353.
Pruesse E, Quast C, Knittel K, et al. (2007) SILVA: a comprehensive online resource for quality checked
and aligned ribosomal RNA sequence data compatible with ARB. Nucleic Acids Research, 35,
7188-7196.
Redfield AC (1958) The biological control of chemical factors in the environment. American Scientist, 46,
205-211.
Remmel EJ, Kohmescher N, Larson JH, et al. (2011) An experimental analysis of harmful algae-
zooplankton interactions and the ultimate defense. Limnology and Oceanography, 56, 461-470.
Reynolds CS, Padisak Ja, Sommer U (1993) Intermediate disturbance in the ecology of phytoplankton
and the maintenance of species diversity: a synthesis. Hydrobiologia, 249, 183-188.
Roelke DL, Grover JP, Brooks BW, et al. (2010) A decade of fish-killing Prymnesium parvum blooms in
Texas: roles of inflow and salinity. Journal of Plankton Research, 33, 243-253.
Rothschild BJ, Ault JS, Goulletquer P, et al. (1994) Decline of the Chesapeake Bay oyster population: a
century of habitat destruction and overfishing Marine Ecology Progress Series, 111.
180

Ruan Q, Dutta D, Schwalbach MS, et al. (2006) Local similarity analysis reveals unique associations
among marine bacterioplankton species and environmental factors. Bioinformatics, 22, 2532-
2538.
Schloss PD, Westcott SL, Ryabin T, et al. (2009) Introducing mothur: open-source, platform-independent,
community-supported software for describing and comparing microbial communities. Applied
and Environmental Microbiology, 75, 7537-7541.
Scholin C, Gulland F, Doucette GJ, et al. (2000) Mortality of sea lions along the central California coast
linked to a toxic diatom bloom. Nature, 403, 80-84.
Shade A, Chiu C-Y, McMahon K (2010) Seasonal and Episodic Lake Mixing Stimulate Differential
Planktonic Bacterial Dynamics. Environmental Microbiology, 59, 546-554.
Shade A, Read JS, Youngblut ND, et al. (2012) Lake microbial communities are resilient after a whole-
ecosystem disturbance. The ISME journal.
Shannon P, Markiel A, Ozier O, et al. (2003) Cytoscape: a software environment for integrated models of
biomolecular interaction networks. Genome Research, 13, 2498-2504.
Sherr E, Sherr B (2002) Significance of predation by protists in aquatic microbial food webs. Antonie Van
Leeuwenhoek, 81, 293-308.
Simpson EH (1949) Measurement of Diversity. Nature, 163, 688.
Skovgaard A, Hansen PJ (2003) Food uptake in the harmful alga Prymnesium parvum mediated by
excreted toxins. Limnology and Oceanography, 48, 1161-1166.
Smoot ME, Ono K, Ruscheinski J, et al. (2011) Cytoscape 2.8: new features for data integration and
network visualization. Bioinformatics, 27, 431-432.
Sogin ML, Morrison HG, Huber JA, et al. (2006) Microbial diversity in the deep sea and the
underexplored "rare biosphere". Proceedings of the National Academy of Sciences of the United
States of America, 103, 12115-12120.
181

Sommer U, Gliwicz ZM, Lampert W, et al. (1986) The PEG-model of seasonal succession of planktonic
event in fresh waters. Arch Hydrobiol, 106, 433-471.
Steele JA, Countway PD, Xia L, et al. (2011) Marine bacterial, archaeal and protistan association
networks reveal ecological linkages. The ISME journal, 5, 1414-1425.
Storey JD (2002) A direct approach to false discovery rates. Journal of the Royal Statistical Society: Series
B (Statistical Methodology), 64, 479-498.
Sunda WG, Graneli E, Gobler CJ (2006) Positive feedback and the development and persistence of
ecosystem disruptive algal blooms. Journal of Phycology, 42, 963-974.
Teeling H, Fuchs BM, Becher D, et al. (2012) Substrate-controlled succession of marine bacterioplankton
populations induced by a phytoplankton bloom. Science, 336, 608-611.
Tillmann U (1998) Phagotrophy by a plastidic haptophyte, Prymnesium patelliferum. Aquatic Microbial
Ecology, 14, 155-160.
Tillmann U (2003) Kill and eat your predator: a winning strategy of the planktonic flagellate Prymnesium
parvum. Aquatic Microbial Ecology, 32, 73-84.
Uronen P, Kuuppo P, Legrand Ca, et al. (2007) Allelopathic effects of toxic haptophyte Prymnesium
parvum lead to release of dissolved organic carbon and increase in bacterial biomass.
Environmental Microbiology, 54, 183-193.
Vigil P, Countway PD, Rose J, et al. (2009) Rapid shifts in dominant taxa among microbial eukaryotes in
estuarine ecosystems. Aquatic Microbial Ecology, 54, 83-100.
Warwick RMCKR (1991) A Comparison of some methods for analysing changes in benthic community
structure. Journal of Marine Biology, 71, 225-244.
Watts DJ, Strogatz SH (1998) Collective dynamics of "small-world" networks. Nature, 393, 440-442.
Wetzel RG (2001). Limnology: Lake and River Ecosystems, Vol 3rd (Philadelphia: Academic Press).
182

Xia L, Steele J, Cram J, et al. (2011) Extended local similarity analysis (eLSA) of microbial community and
other time series data with replicates. BMC Systems Biology, 5, S15.
Zamor RM, Glenn KL, Hambright KD (2012) Incorporating molecular tools into routine HAB monitoring
programs: Using qPCR to track invasive Prymnesium. Harmful Algae, 15, 1-7.
Zheng Zhang SS, Lukas Wagner, and Webb Miller (2000) A greedy algorithm for aligning DNA sequences.
Journal of Computational Biology, 7, 203-214.

Abstract (if available)

Abstract Disturbance events are structuring forces that affect the species richness and composition of biological communities. Depending on their scale and duration, these events can have tremendous and long lasting effects on food web structure and ecosystem function. Consequences and examples of disturbance are well-documented for macrobial communities, but the degree to which disturbance affects microbial community structure is poorly understood. Pyrosequencing of 18s (v9 region) and 16s (v6 region) SSU rDNA genes was used to obtain monthly snapshots of eukaryotic and bacterial diversity, community structure, and community composition of the microbial assemblages in Lake Texoma, southwest United States. Microbial eukaryotic (metazoan, protist, fungi, and chlorophytes) and bacterial assemblages were characterized monthly at two locations (Lebanon Pool and Wilson Creek) for one year within the lake, significantly affected by two disturbance events: 1) a localized and prolonged (4-month) bloom of the toxic alga Prymnesium parvum and 2) a large (17 cm) rain event overlaid onto gradual seasonal environmental change. Interactions and co-occurrence patterns of the microbial taxa were examined using Spearman correlation analysis and visualized and quantified using network analyses. Eukaryotic species richness as well as both eukaryotic and bacterial community structure exhibited cyclical seasonal patterns, including distinct responses to the rain event. Patterns of connectivity within the microbial association networks at both locations revealed highly interconnected consortia of taxa and also negative correlations reflected the seasonality of the lake. ❧ The P. parvum bloom in Lebanon Pool but not Wilson Creek created a natural experiment in which to directly explore and compare the effects of an Ecosystem Disruptive Algal Bloom (EDAB) on the microbial community. Microbial species richness was unaffected by the bloom but the eukaryotic community structure (evenness) and the patterns of both eukaryotic and bacterial community similarity at bloom and non bloom sites were statistically distinct during the 4 months of the bloom. The two locations had contrasting taxonomic compositions of the microbial assemblages over the course of the bloom. The haptophyte (P. parvum) strongly dominated the eukaryotic community when it bloomed, although high abundances of fungi and chrysomonads were also observed. In contrast, the community at the non-bloom site (Wilson Creek) contained a wide diversity of common taxa: haptophytes, ciliates, cercozoa, chlorophytes, crustaceans and rotifers. Differences in the bacteria were more subtle, although the actinobacteria were largely absent during the P. parvum bloom. A comparison of the microbial association networks from the two locations during the P. parvum bloom disturbance revealed compositional differences, as well as differences within the interconnectivity patterns of the taxa. A large rain event resulted in massive restructuring of the microbial communities at both locations and rapid decreases or increases of particular microbial groups (cercozoa, ciliates, and betaproteobacteria). Our results indicate that the eukaryotic and bacterial assemblages were significantly structured by disturbance, overlaid on more gradual and subtle seasonal changes. Blooms of P. parvum are examples of particularly disruptive events with far-reaching consequences on microbial eukaryotic and bacterial community structure, trophic composition and interconnectivity within the microbial web as a whole.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Changes in the community composition of marine microbial eukaryotes across multiple temporal scales of measurement

PDF

Genetic characterization of microbial eukaryotic diversity and metabolic potential

PDF

Ecological implications of daily-to-weekly dynamics of marine bacteria, archaea, viruses, and phytoplankton

PDF

Microbe to microbe: monthly microbial community dynamics and interactions at the San Pedro Ocean Time-series

PDF

Using molecular techniques to explore the diversity, ecology and physiology of important protistan species, with an emphasis on the Prymnesiophyceae

PDF

Spatial and temporal investigations of protistan grazing impact on microbial communities in marine ecosystems

PDF

Dynamics of marine bacterial communities from surface to bottom and the factors controling them

PDF

The microbiome of gorgonian octocorals, Muricea, with a description of a novel, photosynthetic protistan symbiont

PDF

Characterizing protistan diversity and quantifying protistan grazing in the North Pacific Subtropical Gyre

PDF

Temporal variability of marine archaea across the water column at SPOT

PDF

Using sequencing techniques to explore the microbial communities associated with ferromanganese nodules and sediment from the South Pacific gyre

PDF

Paleoecology of Upper Triassic reef ecosystems and their demise at the Triassic-Jurassic extinction, a potential ocean acidification event

PDF

Phytoplankton bloom initiation in the Southern California Bight: a multi-year local and regional analysis

PDF

Spatial and temporal dynamics of marine microbial communities and their diazotrophs in the Southern California Bight

PDF

Disentangling the ecology of bacterial communities in cnidarian holobionts

PDF

Marine protistan diversity, spatiotemporal dynamics, and physiological responses to environmental cues

PDF

Distribution and impact of algal blooms leading to domoic acid events in southern California

PDF

B-vitamins and trace metals in the Pacific Ocean: ambient distribution and biological impacts

PDF

The early Triassic recovery period: exploring ecology and evolution in marine benthic communities following the Permian-Triassic mass extinction

PDF

Ecological recovery dynamics of the benthic and pelagic fauna in response to extreme temperature events and low oxygen environments developed during the early Triassic

Asset Metadata

Creator Jones, Adriane Clark (author)

Core Title Annual pattern and response of the bacterial and microbial eukaryotic communities in an aquatic ecosystem restructured by disturbance

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Marine and Environmental Biology

Publication Date 11/21/2013

Defense Date 10/11/2012

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag aquatic,bacterial,disturbance,Lake,microbial communities,microbial ecology,OAI-PMH Harvest,protists

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Caron, David A. (committee chair), Corsetti, Frank A. (committee member), Heidelberg, John F. (committee member), Heidelberg, Karla B. (committee member)

Creator Email adrianej@usc.edu,adrianejones@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-117954

Unique identifier UC11291427

Identifier usctheses-c3-117954 (legacy record id)

Legacy Identifier etd-JonesAdria-1334.pdf

Dmrecord 117954

Document Type Dissertation

Rights Jones, Adriane Clark

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA