Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Changes in the community composition of marine microbial eukaryotes across multiple temporal scales of measurement
(USC Thesis Other)
Changes in the community composition of marine microbial eukaryotes across multiple temporal scales of measurement
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
CHANGES IN THE COMMUNITY COMPOSITION OF MARINE MICROBIAL
EUKARYOTES ACROSS MULTIPLE TEMPORAL SCALES OF MEASUREMENT
by
Diane Young Kim
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)
May 2013
Copyright 2013 Diane Young Kim
i
Dedication
I dedicate my dissertation to my grandparents, Chang Hwan and Eun Sook Park, and my parents,
Kyung Sool and Kyung Hui Kim, for their sacrifices and unconditional love and support.
ii
Acknowledgments
I have many people to thank for the successful completion of this dissertation. I would first like
to thank my adviser, David Caron, who provided intellectual guidance and amazing research
opportunities to explore the diversity of microbial organisms in Antarctica and deep-sea
hydrothermal vents. Thank you for all of your encouragement and faith in my abilities
throughout graduate school. I am grateful for the support and guidance from my committee
members. Karla Heidelberg, John Heidelberg, Jed Fuhrman, and Steve Goodman, thank you for
many fruitful discussions and thoughtful comments. I also could not have asked for a better
group of mentors and colleagues throughout graduate school. Peter Countway, Astrid Schnetzer,
Adriane Jones, Beth Stauffer, Josh Steele, Anand Patel, Ivona Cetinic, Victoria Bertics, Lisa
Collins, Victoria Campbell, Alyssa Gellene, Alle Lie, my cohort and friends in the Marine
Environmental Biology section have been integral to my success in graduate school. Thank you
for making this so fun and unforgettable. The Wrigley Institute for Environmental Studies and
everyone involved in the Fall 2005 Catalina Semester changed the trajectory of my career path,
and I owe so much of who I am today to that time in my life.
I’m grateful to have been surrounded by such an amazing family throughout my life, who have
always supported and believed in me, no matter what. To Kim Tran, who has been and
continues to by my rock. Thank you for your love and patience. I could not have done this
without you.
The work in this dissertation was funded by National Science Foundation grants MCB-0703159,
MCB-0084231, OCE-9818953, OCE-0550829, OCE-1136818, the Wrigley Institute summer
fellowships, and a grant from the Gordon and Betty Moore Foundation.
iii
Table of Contents
Dedication i
Acknowledgments ii
Abstract v
Chapter 1: Rapid shifts in the structure and composition of a protistan assemblage
during bottle incubations affect estimates of total species richness
1
Abstract
Introduction
Methods
Results
Discussion
References
Tables
Figure Legends
Figures
1
2
5
11
17
26
34
36
38
Chapter 2: A combined sequence-based and fragment-based characterization of
microbial eukaryote assemblages provides taxonomic context for the Terminal
Restriction Fragment Length Polymorphism (T-RFLP) method
39
Abstract
Introduction
Methods
Results
Discussion
References
Tables
Figure Legends
Figures
39
40
43
49
54
59
67
70
73
iv
Chapter 3: Monthly, seasonal and interannual variability of microbial eukaryote
assemblages within and below the euphotic zone in the eastern North Pacific
79
Abstract
Introduction
Methods
Results
Discussion
References
Tables
Figure Legends
Figures
Supplemental Information
79
80
82
85
89
97
104
107
111
119
Chapter 4: Association networks reveal guilds of microbial eukaryotes with similar
temporal responses in the eastern North Pacific
125
Abstract
Introduction
Methods
Results
Discussion
References
Tables
Figure Legends
Figures
Supplemental Information
125
125
127
130
134
139
144
146
149
153
Alphabetized Bibliography 156
v
Dissertation Abstract
Microbial eukaryotes are critical components of marine ecosystems, contributing to vital
ecological and biogeochemical processes, but fundamental knowledge regarding patterns of
spatial and temporal variability of natural assemblages of these taxa is limited. Sequence-based
and fragment-based genetic approaches were used to characterize changes in community
composition and structure of microbial eukaryote assemblages across multiple timescales (days,
months, seasons, years). Short-term temporal changes in community composition and structure
were characterized for a 3-day bottle incubation experiment that consisted of 3 treatments (a
control, inorganic nutrient enrichment, and organic nutrient enrichment). Inorganic and organic
enrichments resulted in dramatic changes in community structure and substantially influenced
richness estimates, but community composition and structure also responded rapidly and
significantly even without nutrient additions. The relative abundance of some initially rare taxa
increased dramatically, implying that some taxa comprising the ‘rare biosphere’ responded to
take on ecologically important roles under changing environmental conditions. Long-term
temporal patterns of variability (monthly, seasonal and interannual) in the composition of natural
assemblages of microbial eukaryotes were examined at the San Pedro Ocean Time-series (SPOT)
station for 237 samples collected from four depths on cruises between September 2000 and
December 2010. The spatiotemporal variability of microbial eukaryote assemblages indicated
the presence of distinct communities within and below the euphotic zone at the SPOT station.
Month-to-month community similarity values (~51-61%) were relatively high for assemblages at
all depths, but assemblages at 5 m were temporally more dynamic compared to deeper
assemblages. Seasonality was apparent for microbial eukaryote assemblages within and below
the euphotic zone at 5 m and 150 m, but not at the deep chlorophyll maximum (DCM) which
vi
varied in depth seasonally, or at 500 m. Microbial eukaryote assemblages exhibited cyclical
patterns in nearly half of all the years and depths examined in this study, with an annual resetting
of communities during winter. Interannual variability was apparent at all depths and was a major
factor influencing community composition at our study site. Network analysis based on global
Spearman correlations identified highly correlated temporal patterns between microbial
eukaryote taxa for samples collected between September 2000 and December 2003, indicating
the presence of unique guilds of microbial eukaryote taxa at each depth with coordinated
responses over the 3 years. Our results provide new insight into the temporal changes within
natural assemblages of microbial eukaryotes on multiple timescales of variability, an essential
step for linking changes in microbial eukaryote communities brought about by environmental
fluctuations to overall ecosystem function.
Kim DY, Countway PD, Gast RJ, Caron DA (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.
1
Chapter 1
Rapid shifts in protistan community structure and composition during bottle incubations
affect estimates of total protistan species richness
Abstract
Changes in protistan community structure and composition were characterized through the
analysis of small-subunit ribosomal RNA gene (18S) sequences for a 3-day bottle incubation
experiment using western North Atlantic seawater. Cloning and sequencing was used to
investigate changes in perceived species diversity as a consequence of environmental
perturbation. The treatments included a control (unamended seawater), inorganic nutrient
enrichment, and enrichment with a complex organic mixture. Five clone libraries were
constructed and analyzed at the initial time of collection (t-0hr) and after 24 (t-24hr) and 72 (t-
72hr) hours for the control, and at t-72hr for the inorganic and organic enrichments, resulting in
an analysis of 1,629 partial 18S rDNA sequences that clustered into 241 Operational Taxonomic
Units (OTUs). Analysis of the clone libraries revealed that protistan assemblages were highly
dynamic and changed substantially at both the OTU-level and higher taxonomic classifications
during time frames consistent with many oceanographic methods used for measuring biological
rates. Changes were most dramatic in enrichments, which yielded community compositions that
were strongly dominated by one or a few taxa. Changes in community structure during
incubation dramatically influenced estimates of species richness, which were substantially lower
with longer incubation and especially with amendment, even though all samples originated from
the same parcel of seawater. Containment and enrichment of the seawater sample led to the
Kim DY, Countway PD, Gast RJ, Caron DA (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.
2
detection of otherwise undetected protistan taxa, suggesting that characterization of protistan
diversity in a sample only at the initial time of collection could lead to an underrepresentation of
unique taxa. Additionally, the rapid increase in relative abundances of some members of the
‘rare biosphere’ in our results implies an ecological importance of at least some of the taxa
comprising the ‘rare biosphere.’
Introduction
Protists are ubiquitous in marine environments and play critical functional roles at and near the
base of marine food webs, contributing to food web dynamics and global biogeochemical cycles
(Sherr and Sherr 1994; Barbeau et al. 1996). A major research effort within the field of
microbial ecology is to characterize the biological processes conducted by this diverse
assemblage of microorganisms, including primary production, herbivory and bacterivory.
Methodologies for accomplishing this goal often require incubations in containers for periods of
one to three days (Caron et al. 2000). For example, measurements of growth and grazing rates of
phytoplankton using the dilution technique require the addition of inorganic nutrients to samples
that are then incubated for periods of one to several days (Landry et al. 1995; Caron 2000; Sherr
and Sherr 2002). Traditional measurements of primary production also involve the incubation of
water samples for hours to days (Eppley 1968; Watt 1971). In addition, bottle incubations are
routinely used to investigate response to specific growth limiting factors (Hutchins and Bruland
1998) and changes in environmental parameters such as pH and pCO
2
concentrations on
microbial community structure and processes (Doney 2010). Major assumptions in the
performance of these experimental manipulations are that microbial eukaryotic assemblages do
Kim DY, Countway PD, Gast RJ, Caron DA (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.
3
not change substantially during these relatively short incubation periods and that processes
taking place in the bottles reflect those of natural assemblages (Dolan 2004; Agis et al. 2007).
Protistan assemblages have been shown to respond to perturbation and incubation on time
scales of hours to days using traditional bulk measurements such as chlorophyll a concentration
and cell counts (Agis et al. 2007), but the characterization of whole protistan assemblages using
traditional morphology-based approaches has rarely been performed because of the multitude of
methods required for the diverse array of protistan taxa present in most aquatic samples (Caron
2009a). Countway et al. (2005) used cloning and sequencing to demonstrate that protistan
community structure can change dramatically on time scales of hours to days in response to
containment. The authors analyzed a total of 970 partial 18S rDNA sequences to investigate
changes in community structure and composition over a 3-day incubation period. Results
revealed that more than 65% of the 165 OTUs were observed only at a single time-point,
implying rapid shifts in the relative abundances of OTUs in response to containment. The
authors also reported higher estimates of species richness (e.g. Chao-1 and ACE-1) when
sequences from all three time-points were combined, suggesting that shifts in taxonomic
composition induced by containment improved the detection of taxa that were present in the
water sample.
Molecular biological approaches for assessing microbial species richness and diversity
have been transforming the field of protistan ecology since the turn of the century. Genetic
methods are not limited by the diverse morphology-based taxonomies and thus offer the potential
for a single approach for characterizing the immense array of taxa present in most ecosystems.
These gene-based approaches, traditionally targeting the 18S rDNA for protists, have been
Kim DY, Countway PD, Gast RJ, Caron DA (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.
4
increasingly used to complement bulk measurements and traditional taxonomic approaches.
Environmental surveys of 18S rRNA genes have revealed numerous sequences of previously
uncultured and unknown protists, indicating unanticipated levels of protistan diversity in many
marine environments (Moon-van der Staay et al. 2001; Dawson and Pace 2002; Edgcomb et al.
2002; Massana et al. 2002; Stoeck and Epstein 2003; Lovejoy et al. 2006; Epstein and Lopez-
Garcia 2008). Sequence datasets have also revealed the global distribution of some OTUs, and
the application of culture-independent methods to laboratory manipulations have been able to
provide information regarding protistan community response to environmental change during
containment (Countway et al. 2005).
Despite their great potential, all extant methods for characterizing diversity have
detection limits. Therefore, various parametric and nonparametric methods are typically applied
to estimate total species richness in natural samples from molecular datasets. Major goals of this
study were 1) to determine whether genetic analysis of a parcel of seawater sampled initially and
following incubation and perturbation would result in similar estimates of species richness, and
2) to investigate changes in taxonomic composition that might take place during conditions
normally employed in oceanographic studies. This study is an extension and combined analysis
of sequences obtained by Countway et al. (2005). Results revealed that protistan community
structure changed dramatically on time scales used to measure rates of primary production,
herbivory and bacterivory. Results also showed that changes in OTU distribution or evenness
within the same parcel of seawater as a result of perturbation led to substantial differences in
diversity estimates.
Kim DY, Countway PD, Gast RJ, Caron DA (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.
5
Methods
Sample collection and experimental setup
Collection of seawater for the incubation experiment has been detailed in Countway et al.
(Countway et al. 2005) and briefly described here. Seawater was collected on board the R/V
Endeavor (University of Rhode Island) in August 2000 in the western North Atlantic (36°21’N,
75°14’W) from a depth of 7m using acid-rinsed (5%-HCl) Niskin bottles (General Oceanics Inc.,
Miami, FL). The samples were pre-screened (to reduce metazoan contribution to clone libraries)
using gravity filtration with in-line filters equipped with 200-µm mesh screening (Sefar America,
Monterey Park, CA). The resulting filtrate was used to rinse and then fill acid-cleaned 4-L
polycarbonate bottles. Two bottles were sacrificed at the time of experimental set-up as the t-0hr
samples, and the other 18 bottles were divided into the following 3 treatments (6 bottles per
treatment): control (no amendment), an inorganic nutrient enrichment (final concentrations of
10µM of nitrogen as NH
4
+
, 1µM of phosphorous as PO
4
3-
, 1nM of iron as Fe
2+
, and 0.1 nM of
Mn
2+
), and an enrichment with a complex organic mixture (0.0001% yeast extract, which is
comprised of a suite of organic and inorganic nutrients).
All bottles were incubated in a flow-through incubator on the deck of the ship to maintain
ambient water temperature, and were shaded with grey window screening to provide a 30%
incident solar irradiance exposure, similar to conditions at the depth of collection. Three bottles
from each treatment were sacrificed after 24hr of incubation, and the remaining three bottles
from each treatment were sampled after 72hr.
Chlorophyll a concentrations
Kim DY, Countway PD, Gast RJ, Caron DA (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.
6
Seawater samples (500ml) were vacuum filtered (~5 Hg) onto 25mm GF/F filters (Whatman), in
triplicate, for each bottle. Filters were folded in half and inserted into 5ml, 13mm glass test
tubes. Chlorophyll was extracted using 7ml of 90% acetone in the dark at -20°C. Samples were
warmed to room temperature in the dark following an extraction period of 18-24 hr, and
pigments were analyzed using a Turner fluorometer before and after acidification with 5% HCl.
DNA Extraction and Isolation for Cloning and Sequencing
All sequences were generated from the same parcel of seawater collected from the western North
Atlantic before and after incubation in bottles with or without the addition of inorganic nutrients
or a complex organic mixture. The protocol used to extract and isolate DNA for all samples has
been detailed in Countway et al. (Countway et al. 2005). Briefly, 2L from each t-0hr bottle, 1.5L
from each t-24hr bottle, and 1L from each t-72hr bottle were vacuum filtered (~5 Hg) onto 47-
mm GF/F filters (Whatman). Each filter was loosely rolled, placed into a 4-ml cryo-vial, and
frozen until extraction. DNA was extracted on board the ship using 1ml of hot (~70°C) lysis
buffer (100mM Tris (pH 8), 40mM EDTA (pH 8), 100mM NaCl, 1% SDS) and ~200µl of 0.5-
mm zircon beads. Each filter was put through three rounds of bead-beating (by vortexing at the
highest speed for 30 s) and heating (in a hot water bath set to ~70°C for 5 min). Lysates were
transferred into 2-ml micro-centrifuge tubes, adjusted to a final concentration of 0.7M NaCl and
1% CTAB (Hexadecyl-trimethyl-Ammonium-Bromide), and incubated for 10 min at 70°C.
After extraction using chloroform, aqueous layers were transferred to fresh tubes for
precipitation of nucleic acids with isopropanol. Tubes were centrifuged at >20,000 x g for
15min. The liquid was decanted and the pellets were air-dried, resuspended in sterile water, and
frozen at -20°C.
Kim DY, Countway PD, Gast RJ, Caron DA (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.
7
Processing of DNA extracts collected from the initial time point and control treatment
bottles has been detailed in Countway et al. (Countway et al. 2005). Additional t-0hr DNA
extract, as well as the DNA extracts from the t-72hr inorganic nutrients and the t-72hr organic
enrichment treatment bottles were processed for this study using the same methods, as briefly
noted below.
PCR for Cloning and Sequencing
Universal eukaryotic primers Euk-A (5’-AACCTGGTTGATCCTGCCAGT-3’) and Euk-B
(5’GATCCTTCTGCAGGTTCACCTAC-3’) were used to amplify full-length 18S rRNA genes.
PCRs were performed for the t-0hr sample (to augment t-0hr sequences obtained by Countway et
al. (2005)), as well as for the DNA extracts from the t-72hr inorganic nutrients and t-72hr
organic enrichment bottles. PCRs were comprised of 0.5 µM of each primer, 1X GoTaq Flexi
Green Buffer (Promega), 2.5 mM MgCl
2
, 250 µM dNTPs, 300 ng/µl BSA (Sigma), 2.5U of
GoTaq (Promega) in a total reaction volume of 50 µl. PCR amplifications were conducted using
a MoBio MyCycler and the following thermal protocol: 1x (95°C for 2 min), 35x (95°C for 30
sec, 55°C for 30 sec, 72°C for 2 min), 1x (72°C for 7 min), ∞ (4C°). Products from replicate
PCRs were pooled for each time-point and treatment, cleaned and concentrated using a Zymo
Clean and Concentrator-5 kit to account for within bottle amplification variability as well as to
ensure sufficient yield of DNA for subsequent steps.
Ligation and Transformation
PCR products were separated on a 1.2% SeaKem LE agarose gel for 10 min at 110V and another
50 min at 130V. The gel was stained for 30 min using SYBR Gold and PCR products were
visualized using a blue light transilluminator. PCR products ~1800bp in size were excised using
Kim DY, Countway PD, Gast RJ, Caron DA (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.
8
a sterile razor blade and the DNA products were isolated from the agarose gel, cleaned, and
concentrated using a MoBio gel extraction kit (MoBio). Recovered DNA was quantified using
Pico Green on a BioRad Versafluor fluorometer and ~25ng of DNA was used for ligation
reactions. Ligations were performed using Invitrogen TOPO-TA cloning kits. Ligation
reactions were cleaned and concentrated using a Zymo Clean and Concentrator-5 kit, repeated
with 2 additional wash steps with 80% ethanol to ensure a salt-free DNA solution, and then used
to transform TOP10 electrocompetent E. coli cells. Individual and discrete colonies were picked
with sterile toothpicks into 96-well culture blocks containing 1.25 ml TB + Kanamycin
(50µg/ml), sealed with an Airpore tape and incubated overnight at 37°C while shaking at 250
RPM to propagate the clones.
Sequencing
Sequencing was performed by the DNA Sequencing Core facility at the University of Florida
(http://langsat.biotech.ufl.edu), using a rolling-circle template amplification by Φ29 polymerase
with aliquots of the overnight clones followed by BigDye® Terminator (ABI) sequencing
reactions using the sequencing primer Euk-570F (Weekers et al. 1994). Only sequences with
Phred scores >20 were reported. Trimming was also based on Phred scores. Each sequence
chromatogram was also visually inspected using Chromas (Technelysium) for select conserved
regions of the DNA sequence as another quality check. This resulted in 305 and 298 high-
quality sequences for the t-72hr inorganic nutrient and the t-72hr organic enrichment treatments,
respectively. The sequencing effort also produced 56 additional sequences to augment the t-0hr
library examined by Countway et al. (Countway et al. 2005), resulting in a total of 549
sequences from t-0hr for the current study. Sequences from Countway et al. (Countway et al.
Kim DY, Countway PD, Gast RJ, Caron DA (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.
9
2005) from the t-24hr control (265 sequences) and t-72hr control (212 sequences) samples were
also included in this study. Average sequence length after quality controls and trimming for all
1,629 sequences presented in this study was 649 bp.
Sequence Analysis
All sequences were checked for chimeras using the RDP Chimera_Check v. 2.7 tool (Cole et al.
2003), and a local implementation of the pintail algorithm (Ashelford et al. 2005). Pairwise
comparisons of all sequences using ClustalW (Thompson et al. 1994) established a global
sequence similarity matrix that was used in the Microbial Eukaryotic Species Assignment
(MESA) program (Caron 2009b) to establish OTUs based on a 95% similarity threshold. The
95% sequence similarity level was established from an analysis of full-length SSU 18S rRNA
gene sequences of morphologically defined protists from NCBI’s GenBank to establish
approximate species-level distinctions for protistan taxa (Caron 2009b). Putative taxonomic
identifications of OTUs were assigned using BLAST (Altschul et al. 1997) against the ARB
SILVA database (v100). Assignment of putative taxonomic identifications of OTUs was
determined by e-values and a majority basis. Putative taxonomic identifications were
summarized as stacked rings depicting three phylogenetic levels—the supergroup (based on the
classification scheme discussed in Tekle et al (Tekle et al. 2009)), kingdom, and class/order
levels.
Nucleotide Sequence Accession Numbers
Partial length 18S rDNA sequences from the original t-0hr and control libraries were deposited
in GenBank by Countway et al. (Countway et al. 2005) (Accessions AY937465-AY938434).
Kim DY, Countway PD, Gast RJ, Caron DA (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.
10
Additional sequences generated for this study were deposited separately and have Accession
numbers JF304989-JF305644.
Multivariate Statistical Analysis
PRIMER (v6) & PERMANOVA+β18 (PRIMER-E Ltd) was used for similarity analyses of the
clone libraries. OTU abundance values were square-root transformed and a (rank) similarity
matrix was generated based on Bray-Curtis similarity coefficients (Bray and Curtis 1957). The
matrix was used for cluster analysis (along with the SIMPROF significance test) and non-metric
Multidimensional Scaling (MDS) analysis (Kruskal 1967). The settings for the MDS analysis
were as follows: number of restarts was set to 10000; the minimum stress value was set to 0.01;
and set to Kruskal fit scheme number 1. A stress value was calculated, which is based on the
variance between Bray-Curtis coefficient values and actual distances on the MDS plot, with
lower values depicting a better representation of similarity between samples along a 2-
dimensional space. PRIMER (v6) & PERMANOVA+β18 was also used to calculate the
following univariate diversity indices for the combined sequences as well as for each sample as
another measure of comparison: the Shannon-Weiner diversity index, the Simpson diversity
index (the reciprocal is presented for easier comparison), Pielou’s evenness index, and
Margalef’s richness index.
Species richness estimates
OTUs were ranked by abundance for each library and for all sequences combined for non-
parametric and parametric estimates of the total protistan species richness. SPADE (Chao 2010)
was used to generate the non-parametric estimates, Chao-1 and ACE-1. A parametric best-
model fitting approach was also used to estimate total protistan species richness using CatchAll
Kim DY, Countway PD, Gast RJ, Caron DA (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.
11
(Bunge and Barger 2008). 95% Confidence Intervals (C.I.) were included in all protistan
richness estimates.
Rank abundance curves and Venn diagram
Rank abundance curves were generated for each library in two formats: one in which OTU
numbers were the same in all libraries (using the t-0hr sample to normalize rank abundance) and
one in which rank OTUs were reorganized based on rank abundance within each individual
library. Shared and unique OTUs were also compared between the t-72hr clone libraries and the
results presented as a Venn diagram.
Results
Protistan diversity estimates and indices
A total of 1,629 partial 18S rDNA sequences were analyzed for this study (Table 1 and Figure
1). The sequences clustered into 241 OTU groups using a 95% sequence similarity threshold in
MESA (Microbial Eukaryotic Species Assignment), an automated OTU-calling program for
microbial eukaryotes that uses CLUSTAL-W aligned sequence files (program and threshold
setting mentioned in Materials and Methods and discussed in detail in Caron et al. (2009b)). A
rank abundance curve of the combined sequences revealed a typical logarithmic shape with most
of the OTUs (>92%) represented by less than 1% of the total number of sequences, while the
most abundant OTU represented ~15% of the total number of sequences (Figure 1).
Most of the sequences obtained at t-0hr and all of the t-24hr control and t-72hr control
sequences were presented in Countway et al. (Countway et al. 2005), which accounted for 970 of
the 1,629 sequences analyzed in this study. Countway et al. (Countway et al. 2005) reported that
the 970 sequences clustered into 165 OTUs, using the same MESA program and similarity
Kim DY, Countway PD, Gast RJ, Caron DA (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.
12
threshold for establishing OTUs. Based on the analysis of the 970 sequences, Countway et al.
(Countway et al. 2005) reported a Chao-1 estimate of total (observed plus unobserved) protistan
species richness of 282 (229-381, 95% C.I.) for data from all time-points combined.
The Chao-1 estimate for the combined sequences was 639 (486-888, 95% C.I.),
significantly higher than the estimate for the subset presented in Countway et al. (Countway et
al. 2005) (Table 1). Interestingly, the Chao-1 estimate for the t-0hr clone library alone after the
addition of 56 new sequences was not significantly different from the estimate for the combined
1,629 sequences (Table 1). ACE-1 and CatchAll estimates agreed more closely with each other
and yielded higher estimates than values obtained using the Chao-1 index (similar to results
reported by Zuendorf et al. (Zuendorf et al. 2006)) , but showed similar trends among the
samples as the Chao-1 estimator. Diversity estimates using the parametric and non-parametric
methods were highest for the t-0hr clone library and combined sequences (i.e. analysis of all
1,629 sequences), and generally lower for samples analyzed following incubation and
amendment (Table 1). Estimates obtained using the ACE-1 estimator did not detect significant
differences between the t-24hr and the three t-72hr samples, but they were all statistically less
than the ACE-1 estimate for the t-0hr sample. The CatchAll estimator (or the parametric best-
fitting model approach) yielded the lowest diversity estimate for the t-72hr sample enriched with
the organic mixture. The CatchAll approach also generated a significantly lower estimate for the
t-72hr sample enriched with inorganic nutrients relative to the t-24hr control, although the t-72hr
and t-24hr controls were not significantly different from each other. The Chao-1 estimates for
the three t-72hr libraries were not significantly different from each another, but were all
significantly less than the t-24hr control.
Kim DY, Countway PD, Gast RJ, Caron DA (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.
13
The individual clone libraries and all sequences combined resulted in different Shannon-
Weiner, Margalef, Simpson and Pielou values (Table 1). The t-0hr clone library, the t-24hr and
t-72hr control libraries, as well as the combined sequence dataset had the highest values for all
indices while the t-72hr samples that were enriched with inorganic nutrients or a complex
organic mixture had the lowest values.
Similarity of protistan assemblages
The protistan assemblages revealed by the five clone libraries (i.e. the individual treatments and
time-points) were approximately 29% similar to one another when analyzed by MDS, with
particular assemblages having substantially higher similarities (Figure 2). The enriched samples
at t-72hr were not significantly different from each other (p > 0.05) and clustered together at
approximately 62% similarity. However, they were significantly different (p < 0.05) from the t-
0hr sample and control treatments at 24 and 72 hours. The t-24hr control and t-72hr control
clustered at 52% similarity and were significantly different from the t-0hr sample. Results from
the cluster analysis (including the SIMPROF significance test, figure not shown) corroborated
results from the MDS analysis. The stress value for the MDS analysis was calculated based on
the variance between the expected distances from Bray-Curtis similarity values and the actual
distances on the 2-dimensional MDS plot. Stress values below 0.05 indicate accurate
representations of similarities between samples within a 2-dimensional space. The stress value
for this analysis was 0, implying highly accurate placements of the data points on the MDS plot.
Changes in protistan assemblages at the OTU-level
Rank abundance curves generated for each sample were used to visualize OTUs that were unique
for each sample, as well as changes in relative abundances of shared OTUs (left panel of
Kim DY, Countway PD, Gast RJ, Caron DA (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.
14
Figure 3). Rank abundance curves of all samples were first compared according to relative
abundances of OTUs in the t-0hr sample (left panel in Figure 3). OTUs in all enrichments and
incubations were then reorganized from most abundant to least abundant within each library
(right panel in Figure 3).
There were substantial changes in the relative importance of OTUs between the t-0hr
sample and the samples analyzed after incubation and/or perturbation (left panel in Figure 3).
Many initially abundant OTUs decreased in relative abundance after incubation and/or
amendment, and many were undetected in samples after 72hr of incubation. Conversely, several
OTUs that initially constituted <1% of total sequences were a significant portion of the
assemblage after incubation with and without enrichment. One OTU (shown as rank #26 in the
rank abundance curves on the left panel) comprised ~0.4% of the library in the t-0hr sample, but
increased in relative contribution to 3% and 6.6% in the t-24hr and t-72hr controls, respectively,
and to 24.9% and 52% in the samples taken after 72 hours of incubation with inorganic nutrients
and complex organics, respectively. OTU #26 was the most abundant OTU for the latter two
libraries (OTU rank #1 in the last two rank abundance curves on the right panel of Figure 3).
The best BLAST match for the sequences comprising this OTU using the ARB/SILVA (v100)
database resulted in a >98% similarity (average similarity of all sequences in this OTU ≈ 97%,
with an average length of 785 bp and average alignment length of 679 bp) to an Oligotrich ciliate
of the genus Strombidium.
Incubation and enrichment also resulted in the detection of many OTUs that were not
detected in the t-0hr sample (Figure 3, OTUs ranked 166 – 244 in the left panel). A total of 79
OTUs in the t-24hr and the three t-72hr treatments were unique from the OTUs detected in the t-
Kim DY, Countway PD, Gast RJ, Caron DA (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.
15
0hr sample. Additionally, 20 OTUs were only present in one or both of the t-72hr enriched
samples (i.e. not present in the t-0hr or t-24hr and t-72hr controls). Most of the unique OTUs
present in the t-72hr enriched samples were ‘rare’ (or represented by less than 1% of total
sequences in each respective library), with closest BLAST matches to Ciliophora (14),
Dinophyceae (1), Stramenopile (1), Fungi (1), and unknown (3) groups. Comparison between
the three t-72hr samples revealed that 84 OTUs were present in at least one or more of the t-72hr
clone libraries. Only 13 of these OTUs were shared across all three t-72hr libraries, and 56
OTUs (>66% of total OTUs in one or more of the t-72hr samples) were unique to a single t-72hr
library (Figure 4). Overall, only 4% of OTUs were present in all 5 libraries (although they
represented 39% of total sequences), while > 69% of the OTUs were unique to a single library.
All rank abundance curves were strongly dominated by a few taxa when the OTUs were
reorganized within each library (right panel of Figure 3). However, the most abundant OTUs
(i.e. OTUs comprised of >1% of total sequences) were substantially different after incubation
with and without amendment (Figure 3, both panels).
Taxonomic composition of clone libraries
Putative taxonomic identifications based on best BLAST matches using the ARB/SILVA
database (v100) were used to provide taxonomic information on the response of the protistan
community to the different treatments (Figure 5). This study was not intended to be an in-depth
characterization of phylogenetic relationships, and these taxonomic distinctions are not strictly
descriptive of functional role, but they can provide some insight into the potential ecological
roles of the OTUs. Supergroup classifications discussed in Tekle et al.(Tekle et al. 2009) were
employed for assigning putative taxonomic identities in this study.
Kim DY, Countway PD, Gast RJ, Caron DA (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.
16
The distribution of taxa among the individual clone libraries showed relatively minor
differences between the initial time of collection and after 24hr and 72hr of incubation without
enrichment at the supergroup, kingdom and class/order levels (Figure 5). The relative
contributions of Stramenopile, Dinophyceae, and Ciliophora groups increased in the control
bottles after 24hr of incubation and decreased following an additional 48hr of incubation,
although the contribution of these groups in the t-72hr control clone library still exceeded
relative abundances in the t-0hr clone library. In contrast, the taxonomic composition changed
dramatically in the bottles enriched with inorganic nutrients or organic compounds.
Chromalveolates, and in particular Ciliophora, increased in bottles treated with either inorganic
nutrients or a complex organic mixture after 72hr of incubation. Few Stramenopile sequences
were observed in the t-72hr enriched libraries.
Chlorophyll a measurements
The inorganic and organic enrichment treatments had higher concentrations of chlorophyll a than
the control treatment after 24hr of incubation, increasing by an order of magnitude from t-0hr to
approximately 1 µg/L (Figure 6). Chlorophyll a concentrations subsequently doubled in the
amended treatments after an additional 48hr of incubation (i.e. t-72hr). While there was no
difference in chlorophyll a concentrations between the inorganic nutrients and organic treatments
after 24hr of incubation, enrichment with a complex mixture of organics yielded higher
chlorophyll a concentrations compared to the sample enriched with inorganic nutrients as well as
the unamended sample after 72hr of incubation. Chlorophyll a concentrations in the control
treatment remained <0.5 µg/L over the course of the experiment (Figure 6).
Kim DY, Countway PD, Gast RJ, Caron DA (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.
17
Discussion
Rapid shifts in protistan community structure in response to containment and enrichment
Containment and enrichment can have dramatic effects on the taxonomic composition of
microbial communities. It is important to understand these effects because bottle incubations are
often employed for studying and measuring microbial processes that take place in nature.
Enclosure in a bottle may induce changes in microbial community structure that do not reflect
the community at the time of collection and thus call into question the rates inferred from
experimental incubations. Compositional changes of enclosed assemblages have been
demonstrated using traditional bulk measurements (Agis et al. 2007) and, more recently, a
molecular taxonomic approach using cloning and sequencing (Countway et al. 2005).
Changes in the taxonomic composition of bacterial assemblages have long been known
(Zobell 1943; Massana et al. 2001), but there have been few studies to characterize protistan
community dynamics during bottle incubations (Dolan 2004). One reason for this disparity is
the laborious nature and multitude of methods required for characterizing diverse protistan
assemblages using traditional morphology-based taxonomies. This situation has been improved
by recent advances in the application of molecular biological approaches and methods for
assaying microbial eukaryotic communities (Caron et al. 2004).
Countway et al. (Countway et al. 2005) used a combined genetic approach (sequencing
of small subunit ribosomal RNA genes and T-RFLP) to demonstrate substantial shifts in
protistan community composition and structure in response to containment on time-scales of 1-3
days. Experimental treatments conducted as a part of the same experiment examined the effects
of enrichment with inorganic nutrients and a complex mixture of organic and inorganic material
Kim DY, Countway PD, Gast RJ, Caron DA (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.
18
(yeast extract). Sequence analysis of these latter treatments, when combined with previous
results (Figure 1), revealed that changes in taxonomic composition were more pronounced in the
enrichments relative to the control (containment with no amendment) at the OTU-level (Figure
3) as well as at higher taxonomic classifications (the supergroup, kingdom, and class/order
levels; Figure 5). Community similarity analysis (MDS) indicated that enriched assemblages
were more similar to each other following 72hr of incubation than either were to assemblages
that were not enriched (Figure 2). These changes in the eukaryotic assemblage occurred more
rapidly than anticipated and are consistent with in situ observations of short-term temporal and
spatial changes in the dominant protistan taxa of assemblages in natural samples (as assessed by
a molecular fingerprinting approach) (Vigil et al. 2009). They also imply that relative
abundances of protistan taxa within microbial assemblages can be dynamic on time scales that
are often employed in oceanographic methods used to measure rates of primary production,
herbivory and bacterivory.
The eukaryotic taxonomic composition at t-0hr reflected a diverse assemblage of protists
(Figure 5) that did not change appreciably at the supergroup, kingdom and class/order levels
after 24hr and 72hr of incubation in the control treatment. In contrast, there were dramatic
changes at the higher order classifications in both enrichments, where ciliates comprised
approximately 92% of the taxonomic composition after 72hr of incubation following both types
of enrichment. The dominant OTUs also changed substantially in both amended treatments,
yielding rank abundance curves with fewer OTUs (Figure 3) and the dominance of the
oligotrichous ciliate phylotype, Strombidium sp. (OTU #26; bottom two curves on right panel of
Figure 3).
Kim DY, Countway PD, Gast RJ, Caron DA (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.
19
The dominance of ciliate phylotypes in the clone libraries of enriched samples after 72hr
of incubation was in apparent contradiction to the chlorophyll a data, which suggested high
phytoplankton biomass in the amended treatments (Figures 5 and 6). Phytoplankton biomass
(indicated by chlorophyll a concentration) increased rapidly in response to the addition of both
inorganic nutrients and a complex mixture of organic and inorganic compounds (yeast extract).
The relatively low abundance of sequences that might belong to potentially-photosynthetic
protists in the t-72hr enriched samples appears to indicate a trophic cascade in which herbivorous
ciliates rapidly converted phytoplankton biomass into ciliate biomass. The latter may have
dominated the overall biomass, even though the absolute amount of algal biomass also increased.
Growth rates of marine planktonic ciliates vary depending on the size of the ciliate and the
experimental temperature employed, with typical rates of <1 to ~4 d
-1
(Rose and Caron 2007). A
growth rate of 3.05 d
-1
has been reported for a Strombidium species at 20°C (Ohman and Snyder
1991), somewhat less than the experimental temperature used in the present study (≈ 25˚C). We
speculate that the rapid shift to dominance by Strombidium in the enriched treatments of our
study was promoted by the high potential maximal growth rate of Strombidium, the removal of
metazoan predators of ciliates by prescreening the water through 200 µm screening (see
Methods), and the enhancement of phytoplankton biomass (ciliate prey) by nutrient amendment
as reflected in increases in chlorophyll a concentrations (Figure 6).
High copy numbers of the 18S rRNA genes in ciliates also may have resulted in over-
representation of ciliates in the clone libraries relative to phytoplankton (Potvin and Lovejoy
2009). Nevertheless, changes in the relative abundance of the Strombidium phylotype in the t-
72hr enriched treatments relative to the control (unamended) treatment, and the high
Kim DY, Countway PD, Gast RJ, Caron DA (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.
20
reproducibility of clone libraries (Caron 2009b), imply that ciliates increased dramatically in
abundances in the enriched treatments. Additionally, oligotrichous ciliates are well known for
kleptoplasty, the ability to ingest photosynthetic prey and retain the prey chloroplasts in a
functional state (Stoecker et al. 2009). This phenomenon might, in part, account for the high
chlorophyll a concentrations in conjunction with the dominance of ciliate 18S genes observed at
t-72hr in the amended treatments.
Our results provide an account of rapid reassembly of protistan assemblages that may be
important when interpreting results from bottle incubation experiments. However, the results of
our study do not necessarily discount or discredit experimental results obtained from incubation
experiments because it is conceivable that protistan phylotypes that become dominant during
incubation carry out the same ecological processes at roughly equivalent rates as the protistan
assemblage present at t-0hr. This concept of ecological redundancy has been proposed as an
ecologically significant role for the protistan ‘rare biosphere’ (Caron and Countway 2009). A
much clearer understanding of how microbial communities (including protists) and their
attendant biogeochemical activities will respond to intentional or unintentional, large-scale
manipulations of environmental conditions (e.g. Fe fertilization and ocean acidification) will be
necessary in order to acquire the ability to predict the outcomes of those manipulations (Hutchins
et al. 2009; Strong et al. 2009). Continued advancements in experimental design and methods
for characterizing whole assemblages (including traditional and new methods) will provide a
better understanding of how microbial communities are organized in the environment, how
structure and composition change in response to changes in environmental conditions, the time-
scale at which changes occur, and how those changes might affect ecosystem function.
Kim DY, Countway PD, Gast RJ, Caron DA (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.
21
Unveiling more protistan diversity with time and perturbation: a focus on the ‘rare biosphere’
Changes in community structure induced by containment and enrichment of a single sample in
the present study led to the detection of substantially more OTUs compared to the analysis of the
sample at t-0hr (Table 1, Figures 3 and 4). Initially undetected OTUs were detected after
incubation and/or perturbation. Some of these OTUs were detected at relatively high
abundances, while most of them represented less than 1% of total sequences (i.e. rare) (Figure 3).
This implies that additional sequencing effort of the t-0hr sample may have revealed more
unique OTU groups, but it is also probable that many taxa would have remained undetected
without the imposed perturbation to community structure by containment and/or enrichment.
Regardless, our study demonstrated the immense diversity of a natural protistan assemblage, in
only tens of liters of seawater, which may still be an underrepresentation of protistan diversity
given that some molecular surveys have reported 183-191 unique taxa for ciliates alone
(Dopheide et al. 2008).
The long tail of rare OTUs observed in this study (Figure 1) appears to be a common
characteristic of microbial communities when assayed through the sequencing of SSU rRNA
genes from the environment (Sogin et al. 2006). Approximately 92% of the OTUs in the
analysis of our combined dataset were comprised of rare taxa. The ‘rare biosphere’ is a common
and interesting feature of microbial assemblages whose function and importance are not yet well
understood. Its seemingly ubiquitous presence, however, implies that it might be an ecologically
significant feature of microbial communities. A number of hypotheses are now emerging on the
importance of the microbial rare biosphere (Pedros-Alio 2007; Caron and Countway 2009).
Potential roles of the ‘rare biosphere’ for microbial communities include providing stability and
Kim DY, Countway PD, Gast RJ, Caron DA (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.
22
resilience to changing environmental conditions (including environmental stress) and/or playing
roles similar to keystone-species. The rapid changes in relative contribution of some members of
the protistan rare biosphere during the incubations conducted in our study suggest that at least
some constituents within the rare biosphere were alive and responsive to environmental
perturbation (Figure 3), suggesting that constituents of the ‘rare biosphere’ may play important
ecological roles under changing environmental conditions.
The identity and ecological function of many of the rare taxa remains an area of active
research. The ability to relate specific DNA sequences unequivocally to protistan morphotypes
is presently limited due to the relatively low percentage of morphologically well-described
protistan taxa that have been sequenced and appear in environmental 18S clone libraries. Thus,
it is difficult to provide names for many of the sequences observed at very low frequency in
clone libraries. New methods such as single-cell amplification using multiple displacement
amplification reactions (Stepanauskas and Sieracki 2007) show great promise for providing a
means of relating sequence information to individual cells. These approaches will allow
exploration of the composition of the rare biosphere and the ecological roles of organisms that
comprise it.
Effect of rapid shifts in community structure on diversity estimates
Characterizing marine microbial diversity is a fundamental requirement for predicting the
trajectory of marine ecosystems in the face of environmental change because of the inherent
links between biodiversity and ecosystem function (McGradySteed et al. 1997; Naeem and Li
1997). Genetic approaches that target the sequencing of SSU 18S rRNA genes and other genes
have become common, culture-independent approaches for characterizing natural microbial
Kim DY, Countway PD, Gast RJ, Caron DA (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.
23
assemblages (Pace et al. 1986). Early environmental surveys of protistan diversity using 18S
rDNA clone libraries revealed high levels of protistan diversity from a diversity of marine
environments (Diez et al. 2001; Moon-van der Staay et al. 2001; Dawson and Pace 2002;
Edgcomb et al. 2002; Massana et al. 2002; Moreira and Lopez-Garcia 2002; Stoeck and Epstein
2003; Countway et al. 2005; Worden 2006; Zuendorf et al. 2006; Not et al. 2007; Not et al.
2009; Rodriguez-Martinez et al. 2009; Shi et al. 2009; Aguilera et al. 2010; Collado-Mercado et
al. 2010). Significant numbers of unknown and uncultivated organisms also have been revealed
through 18S rDNA inventories, including several lineages of minute and morphologically non-
descript protists (e.g. MAST, picobiliphytes, etc.), potentially parasitic taxa (e.g. Alveolate
groups I and II), and taxa that thrive in extreme environments such as hydrothermal vents
(Lopez-Garcia et al. 2001; Edgcomb et al. 2002; Moreira and Lopez-Garcia 2003; Kolodziej and
Stoeck 2007; Takishita et al. 2007; Rodriguez-Martinez et al. 2009; Aguilera et al. 2010). These
approaches are revealing a large diversity of protistan taxa in individual water samples, although
they are not yet capable of completely characterizing the diversity present in most environments.
Thus, estimates of total species richness are still dependent on extrapolation from observed
species richness using various richness estimators.
Community structure and sequencing depth-of-coverage significantly affected estimates
of species richness using parametric and non-parametric approaches (Table 1). Estimates of
diversity using the Shannon-Weiner, Simpson, Margalef, and Pielou indices were also affected.
The seawater used to produce the clone libraries in this study came from a single sample
collected from the western N. Atlantic and, in theory, should have produced similar estimates of
protistan richness and diversity. However, estimates obtained for the samples following
Kim DY, Countway PD, Gast RJ, Caron DA (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.
24
incubation were substantially lower than the estimates for the t-0hr subsample, especially
following the addition of inorganic nutrients or complex organics (Table 1). Diversity estimates
have been shown to be dependent on the size of clone libraries (Youssef and Elshahed 2008) and
difference in the number and distribution of rare taxa in assemblage rank abundance curves
(Bunge and Barger 2008). Our results imply that estimates of species richness and diversity
were affected by changes in community structure, which was remarkably dissimilar for the two
enriched t-72hr clone libraries relative to the t-0hr, t-24hr and t-72hr control samples (Figures 2
and 3).
Interestingly, the t-0hr clone library yielded estimates of species richness that were not
statistically different from the combined sequence dataset using the Chao-1, ACE-1 and
CatchAll estimators despite the fact that the combined sequence dataset yielded nearly 50% more
unique OTUs than the t-0hr sample (241 vs. 165; Table 1). The similar estimates of species
richness presumably can be explained by the very large number of rare OTUs observed in the t-
0hr clone library (~92% of OTUs comprised less than 1% of the total sequences), their effect on
the shape of the rank abundance curve, and the manner in which the shape of the curve affected
the estimation of diversity by the various richness estimators. We anticipated that experimental
manipulation and incubation would reveal more protistan phylotypes, but the similar richness
estimates obtained for the combined dataset relative to the t-0hr sample was unanticipated.
In contrast to these findings, Countway et al. (Countway et al. 2005) obtained a Chao-1
species richness estimate of 282 for 970 of the 1,629 sequences from three time points (t-0hr, t-
24hr and t-72hr) for the samples that were not enriched. The Chao-1 estimated for the totaled
dataset in that study exceeded richness estimates determined for each of the three sampling times
Kim DY, Countway PD, Gast RJ, Caron DA (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.
25
analyzed separately. Nonetheless, the Chao-1 value obtained for that study (282) was
substantially lower than the value obtained for the larger t-0hr sample (1,010) or the totaled
dataset (639) in the present study. These findings indicate the sensitivity of these species
richness estimators to sampling depth and the shape of the rank abundance curve.
Conclusions
In summary, experimental manipulation and reanalysis of a single parcel of seawater revealed
rapid changes in protistan community structure, especially in response to modest amounts of
enrichment with inorganic nutrients or a complex mixture of inorganic and organic material
(yeast extract). Assemblages present at the end of traditional oceanographic bottle incubations
were substantially different than assemblages present at the initial time of collection. While
shifts in relative abundances of protistan OTUs revealed more protistan taxa in the combined
sequence analysis, changes in community structure as a result of enrichment resulted in lower
estimates of species diversity and richness in the individual t-72hr samples that were enriched.
Finally, the rapid increase in relative abundances of some members of the ‘rare biosphere’
observed in this study implies an ecological importance of at least some of the taxa comprising
the rare biosphere.
Kim DY, Countway PD, Gast RJ, Caron DA (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.
26
References
Agis M, Granda A, Dolan JR (2007) A cautionary note: Examples of possible microbial
community dynamics in dilution grazing experiments. J Exp Mar Biol Ecol 341, 176-183.
Aguilera A, Souza-Egipsy V, Gonzalez-Toril E, Rendueles O, Amils R (2010) Eukaryotic
microbial diversity of phototrophic microbial mats in two Icelandic geothermal hot
springs. Int Microbiol 13, 21-32.
Altschul SF, Madden TL, Schaffer AA, Zhang JH, Zhang Z, Miller W, Lipman DJ (1997)
Gapped BLAST and PSI-BLAST: a new generation of protein database search programs.
Nucleic Acids Res 25, 3389-3402.
Ashelford KE, Chuzhanova NA, Fry JC, Jones AJ, Weightman AJ (2005) At least 1 in 20 16S
rRNA sequence records currently held in public repositories is estimated to contain
substantial anomalies. Appl Environ Microbiol 71, 7724-7736.
Barbeau K, Moffett JW, Caron DA, Croot PL, Erdner DL (1996) Role of protozoan grazing in
relieving iron limitation of phytoplankton. Nature 380, 61-64.
Bray JR, Curtis JT (1957) An ordination of the upland forest communities of southern
Wisconsin. Ecol Monogr 27, 326-349.
Bunge J, Barger K (2008) Parametric models for estimating the number of classes. Biometrical J
50, 971-982.
Caron DA (2000) Protistan herbivory and bacterivory. Method Microbiol, Vol 30 30, 289-315.
Caron DA (2009a) New accomplishments and approaches for assessing protistan diversity and
ecology in natural ecosystems. BioSci 59, 287-299.
Kim DY, Countway PD, Gast RJ, Caron DA (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.
27
Caron DA, Countway PD (2009) Hypotheses on the role of the protistan rare biosphere in a
changing world. Aquat Microb Ecol 57, 227-238.
Caron DA, Countway PD, Brown MV (2004) The growing contributions of molecular biology
and immunology to protistan ecology: Molecular signatures as ecological tools. J
Eukaryot Microbiol 51, 38-48.
Caron DA, Countway, P.D., Savai, P., Gast, R.J., Schnetzer, A., Moorthi, S.D., Dennett, M.R.,
Jones, A.C. (2009b) Defining DNA-based operational taxonomic units for microbial-
eukaryote ecology. Appl Environ Microbiol 75, 5797-5808.
Caron DA, Dennett MR, Lonsdale DJ, Moran DM, Shalapyonok L (2000) Microzooplankton
herbivory in the Ross Sea, Antarctica. Deep Sea Res (II Top Stud Oceanogr) 47, 3249-
3272.
Chao A, and T.-J. Shen (2010) Program SPADE (Species Prediction And Diversity Estimation).
Cole JR, Chai B, Marsh TL, Farris RJ, Wang Q, Kulam SA, Chandra S, McGarrell DM, Schmidt
TM, Garrity GM, Tiedje JM (2003) The Ribosomal Database Project (RDP-II):
previewing a new autoaligner that allows regular updates and the new prokaryotic
taxonomy. Nucleic Acids Res 31, 442-443.
Collado-Mercado E, Radway JC, Collier JL (2010) Novel uncultivated labyrinthulomycetes
revealed by 18S rDNA sequences from seawater and sediment samples. Aquat Microb
Ecol 58, 215-228.
Countway PD, Gast RJ, Savai P, Caron DA (2005) Protistan diversity estimates based on 18S
rDNA from seawater incubations in the western North Atlantic. J Eukaryot Microbiol 52,
95-106.
Kim DY, Countway PD, Gast RJ, Caron DA (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.
28
Dawson SC, Pace NR (2002) Novel kingdom-level eukaryotic diversity in anoxic environments.
Proc Natl Acad Sci USA 99, 8324-8329.
Diez B, Pedros-Alio C, Massana R (2001) Study of genetic diversity of eukaryotic picoplankton
in different oceanic regions by small-subunit rRNA gene cloning and sequencing. Appl
Environ Microbiol 67, 2932-2941.
Dolan JRaM, K. (2004) The reliability of grazing rate estimates from dilution experiments: Have
we over-estimated rates of organic carbon consumption by microzooplankton? Ocean Sci
1, 1-7.
Doney SC (2010) The growing human footprint on coastal and open-ocean biogeochemistry.
Science 328, 1512-1516.
Dopheide A, Lear G, Stott R, Lewis G (2008) Relative eiversity and community structure of
ciliates in stream biofilms according to molecular and microscopy methods. Appl Environ
Microbiol 75, 5261-5272.
Edgcomb VP, Kysela DT, Teske A, Gomez AD, Sogin ML (2002) Benthic eukaryotic diversity
in the Guaymas Basin hydrothermal vent environment. Proc Natl Acad Sci USA 99,
7658-7662.
Eppley RW (1968) An Incubation Method for Estimating Carbon Content of Phytoplankton in
Natural Samples. Limnol Oceanogr 13, 574.
Epstein S, Lopez-Garcia P (2008) "Missing" protists: a molecular prospective. Biodivers
Conserv 17, 261-276.
Hutchins DA, Bruland KW (1998) Iron-limited diatom growth and Si : N uptake ratios in a
coastal upwelling regime. Nature 393, 561-564.
Kim DY, Countway PD, Gast RJ, Caron DA (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.
29
Hutchins DA, Mulholland MR, Fu FX (2009) Nutrient cycles and marine microbes in a CO2-
enriched ocean. Oceanography 22, 128-145.
Kolodziej K, Stoeck T (2007) Cellular identification of a novel uncultured marine stramenopile
(MAST-12 clade) small-subunit rRNA gene sequence from a Norwegian estuary by use
of fluorescence in situ hybridization-scanning electron microscopy. Appl Environ
Microbiol 73, 2718-2726.
Kruskal W (1967) Statistics Moliere and Adams,H. Am Sci 55, 416.
Landry MR, Kirshtein J, Constantinou J (1995) A refined dilution technique for measuring the
community grazing impact of microzooplankton, with experimental tests in the central
equatorial Pacific. Mar Ecol Prog Ser 120, 53-63.
Lopez-Garcia P, Rodriguez-Valera F, Pedros-Alio C, Moreira D (2001) Unexpected diversity of
small eukaryotes in deep-sea Antarctic plankton. Nature 409, 603-607.
Lovejoy C, Massana R, Pedros-Alio C (2006) Diversity and distribution of marine microbial
eukaryotes in the Arctic Ocean and adjacent seas. Appl Environ Microbiol 72, 3085-
3095.
Massana R, Guillou L, Diez B, Pedros-Alio C (2002) Unveiling the organisms behind novel
eukaryotic ribosomal DNA sequences from the ocean. Appl Environ Microbiol 68, 4554-
4558.
Massana R, Pedros-Alio C, Casamayor EO, Gasol JM (2001) Changes in marine
bacterioplankton phylogenetic composition during incubations designed to measure
biogeochemically significant parameters. Limnol Oceanogr 46, 1181-1188.
Kim DY, Countway PD, Gast RJ, Caron DA (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.
30
McGradySteed J, Harris PM, Morin PJ (1997) Biodiversity regulates ecosystem predictability.
Nature 390, 162-165.
Moon-van der Staay SY, De Wachter R, Vaulot D (2001) Oceanic 18S rDNA sequences from
picoplankton reveal unsuspected eukaryotic diversity. Nature 409, 607-610.
Moreira D, Lopez-Garcia P (2002) The molecular ecology of microbial eukaryotes unveils a
hidden world. Trends Microbiol 10, 31-38.
Moreira D, Lopez-Garcia P (2003) Are hydrothermal vents oases for parasitic protists? Trends
Parasitol 19, 556-558.
Naeem S, Li SB (1997) Biodiversity enhances ecosystem reliability. Nature 390, 507-509.
Not F, del Campo J, Balague V, de Vargas C, Massana R (2009) New Insights into the Diversity
of Marine Picoeukaryotes. PLoS One 4, e7143.
Not F, Valentin K, Romari K, Lovejoy C, Massana R, Tobe K, Vaulot D, Medlin LK (2007)
Picobiliphytes: A marine picoplanktonic algal group with unknown affinities to other
eukaryotes. Science 315, 253-255.
Ohman MD, Snyder RA (1991) Growth-kinetics of the omnivorous oligotrich ciliate
Strombidium spp. Limnol Oceanogr 36, 922-935.
Pace NR, Stahl DA, Lane DJ, Olsen GJ (1986) The analysis of natural microbial-populations by
ribosomal-RNA sequences. Adv Microb Ecol 9, 1-55.
Pedros-Alio C (2007) Dipping into the rare biosphere. Science 315, 192-193.
Potvin M, Lovejoy C (2009) PCR-based diversity estimates of artificial and environmental 18S
rRNA gene libraries. J Eukaryot Microbiol 56, 174-181.
Kim DY, Countway PD, Gast RJ, Caron DA (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.
31
Rodriguez-Martinez R, Labrenz M, del Campo J, Forn I, Jurgens K, Massana R (2009)
Distribution of the uncultured protist MAST-4 in the Indian Ocean, Drake Passage and
Mediterranean Sea assessed by real-time quantitative PCR. Environ Microbiol 11, 397-
408.
Rose JM, Caron DA (2007) Does low temperature constrain the growth rates of heterotrophic
protists? Evidence and implications for algal blooms in cold waters. Limnol Oceanogr 52,
886-895.
Sherr EB, Sherr BF (1994) Bacterivory and herbivory - key roles of phagotrophic protists in
pelagic food webs. Microb Ecol 28, 223-235.
Sherr EB, Sherr BF (2002) Significance of predation by protists in aquatic microbial food webs.
Antonie Van Leeuwenhoek International Journal of General and Molecular Microbiology
81, 293-308.
Shi XL, Marie D, Jardillier L, Scanlan DJ, Vaulot D (2009) Groups without cultured
representatives dominate eukaryotic picophytoplankton in the oligotrophic south east
Pacific Ocean. PLoS One 4, -.
Sogin ML, Morrison HG, Huber JA, Welch DM, Huse SM, Neal PR, Arrieta JM, Herndl GJ
(2006) Microbial diversity in the deep sea and the underexplored "rare biosphere". Proc
Natl Acad Sci USA 103, 12115-12120.
Stepanauskas R, Sieracki ME (2007) Matching phylogeny and metabolism in the uncultured
marine bacteria, one cell at a time. Proc Natl Acad Sci USA 104, 9052-9057.
Kim DY, Countway PD, Gast RJ, Caron DA (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.
32
Stoeck T, Epstein S (2003) Novel eukaryotic lineages inferred from small-subunit rRNA
analyses of oxygen-depleted marine environments. Appl Environ Microbiol 69, 2657-
2663.
Stoecker DK, Johnson MD, de Vargas C, Not F (2009) Acquired phototrophy in aquatic protists.
Aquat Microb Ecol 57, 279-310.
Strong A, Chisholm S, Miller C, Cullen J (2009) Ocean fertilization: time to move on. Nature
461, 347-348.
Takishita K, Yubuki N, Kakizoe N, Inagaki Y, Maruyama T (2007) Diversity of microbial
eukaryotes in sediment at a deep-sea methane cold seep: surveys of ribosomal DNA
libraries from raw sediment samples and two enrichment cultures. Extremophiles 11,
563-576.
Tekle YI, Parfrey LW, Katz LA (2009) Molecular data are transforming hypotheses on the origin
and diversification of eukaryotes. BioSci 59, 471-481.
Thompson JD, Higgins DG, Gibson TJ (1994) Clustal-W - Improving the sensitivity of
progressive multiple sequence alignment through sequence weighting, position-specific
gap penalties and weight matrix choice. Nucleic Acids Res 22, 4673-4680.
Vigil P, Countway PD, Rose J, Lonsdale DJ, Gobler CJ, Caron DA (2009) Rapid shifts in
dominant taxa among microbial eukaryotes in estuarine ecosystems. Aquat Microb Ecol
54, 83-100.
Watt WD (1971) Measuring primary production rates of individual phytoplankton species in
natural mixed populations. Deep-Sea Res 18, 329-&.
Kim DY, Countway PD, Gast RJ, Caron DA (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.
33
Weekers PHH, Gast RJ, Fuerst PA, Byers TJ (1994) Sequence variations in small-subunit
ribosomal-RNAs of Hartmannella vermiformis and their phylogenetic implications. Mol
Biol Evol 11, 684-690.
Worden AZ (2006) Picoeukaryote diversity in coastal waters of the Pacific Ocean. Aquat Microb
Ecol 43, 165-175.
Youssef NH, Elshahed MS (2008) Species richness in soil bacterial communities: A proposed
approach to overcome sample size bias. J Microbiol Meth 75, 86-91.
Zobell CE (1943) The effect of solid surfaces upon bacterial activity. J Bacteriol 46, 39-56.
Zuendorf A, Bunge J, Behnke A, Barger KJA, Stoeck T (2006) Diversity estimates of
microeukaryotes below the chemocline of the anoxic Mariager Fjord, Denmark. FEMS
Microbiol Ecol 58, 476-491.
Kim DY, Countway PD, Gast RJ, Caron DA (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.
34
Table 1 Number of sequences, OTU groups, and species richness and diversity information for a
single sample initially and following incubation with and without the addition of inorganic
nutrients or complex organic material. Information from individicual clone libraries and total
data (combined dataset). The reciprocal of the Simpson’s index, the Shannon-Wiener index, the
Margalef Richness index and the Pielou Evenness index values are listed as univariate measures
of diversity (richness and/or evenness). Richness estimates of total (seen and unseen) protistan
diversity using non-parametric approahes (Chao1 and ACE-1) and a parametric best-model
fitting approach (CatchAll) are also listed, including 95% Confidence Intervals (CI).
Kim DY, Countway PD, Gast RJ, Caron DA (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.
35
a
Generated with the automated OTU-calling program for microbial eukaryotes, MESA, using a 95% similarity threshold
b
,cBoth diversity indices take into account species richness and evenness. Larger values reflect higher species richness and more even distributions of species
d
Margalef's index (d) is a measure of richness and is calculated by d = (S-1)/logN, where S = # of OTUs and N = # of sequences
e
Pielou's Index (J') is a measure of equitability or evenness and uses the maximium possible value of Shannon diversity (H'max or log/S) and is calculated as J' =
H'/logS
a,b,c,d,e
Indices were calculated using PRIMER6 & PERMANOVA+β18 (PRIMER-E Ltd)
h
Parametric diversity estimator using John Bunge's CatchAll Estimator
f,g,
Non-parametric diveristy estimators using SPADE
Kim DY, Countway PD, Gast RJ, Caron DA (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.
36
Figure Legends
Figure 1 Rank abundance curve of the 1,626 partial-length 18S rDNA sequences that clustered
into 238 OTUs using MESA and a 95% sequence similarity threshold (A). Enlargement
showing details of the most abundant taxa from panel A (B).
Figure 2 A 2-dimensional non-metric multidimensional scaling (MDS) plot of five clone
libraries, based on calculated Bray-Curtis similarity coefficients after square-root transformations
of the relative contributions of each OTU in each library. Similarity levels of 28%, 39%, 50%
and 62% are depicted.
Figure 3 Individual rank abundance curves for five clone libraries with relative abundance (%)
on the y-axes and OTU ranks on the x-axes. The left panels display rank abundance curves of
OTUs that have been ranked according to abundances at t-0hr, from highest abundance to lowest
from left to right on the t-0hr curve. The right panels display rank abundance curves with each
library reorganized by rank in each respective library. The numbers of sequences are posted on
the top left corner of the individual rank abundance curves, and numbers of OTUs are also listed
for each sample, indicated by arrows. All y-axes are the same, and breaks were included in the
curves for the t-72hr inorganic nutrients and t-72hr complex organic enrichment samples to
facilitate comparison.
Kim DY, Countway PD, Gast RJ, Caron DA (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.
37
Figure 4 Venn diagram showing the number of shared and unique OTUs between the three t-
72hr clone libraries. A total of 81 OTUs were present among the three t-72hr clone libraries.
Figure 5 Taxonomic rings depicting three taxonomic levels of putative identifications of the 238
OTUs, based on best ARB BLAST hits. The inside ring is divided into supergroup levels
(colorless), the middle ring divides the supergroups into kingdom level distinctions, and the
outside ring divides the kingdom level groups further into the class/order level for the Ciliophora
and Dinophyceae kingdom groups only.
Figure 6 Chlorophyll a concentrations for the three different treatments (control, inorganic
nutrients and complex organics) at t-0hr, after 24hr, and after 72hr of incubation with standard
error bars.
Kim DY, Countway PD, Gast RJ, Caron DA (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.
38
Figure 1
Kim DY, Countway PD, Gast RJ, Caron DA (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.
39
Figure 2
Kim DY, Countway PD, Gast RJ, Caron DA (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.
40
Figure 3
Kim DY, Countway PD, Gast RJ, Caron DA (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.
41
Figure 4
Kim DY, Countway PD, Gast RJ, Caron DA (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.
42
Figure 5
Kim DY, Countway PD, Gast RJ, Caron DA (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.
43
Figure 6
Kim DY, Countway PD, Yamashito W, Caron DA (2012) A combined sequence-based and fragment-based characterization of microbial
eukaryote assemblages provides taxonomic context for the Terminal Restriction Fragment Length Polymorphism (T-RFLP) method. Journal of
Microbioogical Methods (in press).
39
Chapter 2
A combined sequence-based and fragment-based characterization of microbial eukaryote
assemblages provides taxonomic context for the Terminal Restriction Fragment Length
Polymorphism (T-RFLP) method
Abstract
Microbial eukaryotes in seawater samples collected from two depths (5 m and 500 m) at the USC
Microbial Observatory off the coast of southern California, USA, were characterized by cloning
and sequencing of 18S rRNA genes, as well as DNA fragment analysis of these genes. The
sequenced genes were assigned to operational taxonomic units (OTUs), and taxonomic
information for the sequence-based OTUs was obtained by comparison to public sequence
databases. The sequences were then subjected to in silico digestion to predict fragment sizes,
and that information was compared to the results of the T-RFLP method applied to the same
samples in order to provide taxonomic context for the environmental T-RFLP fragments. A total
of 663 and 678 sequences were analyzed for the 5 m and 500 m samples, respectively, which
clustered into 157 OTUs and 183 OTUs. The sequences yielded substantially fewer taxonomic
units as in silico fragment lengths (i.e., following in silico digestion), and the environmental T-
RFLP resulted in the fewest unique OTUs (unique fragments). Bray-Curtis similarity analysis of
protistan assemblages was greater using the T-RFLP dataset compared to the sequence-based
OTU dataset, presumably due to the inability of the fragment method to differentiate some taxa
and an inability to detect many rare taxa relative to the sequence-based approach. Nonetheless,
fragments in our analysis generally represented the dominant sequence-based OTUs and putative
Kim DY, Countway PD, Yamashito W, Caron DA (2012) A combined sequence-based and fragment-based characterization of microbial
eukaryote assemblages provides taxonomic context for the Terminal Restriction Fragment Length Polymorphism (T-RFLP) method. Journal of
Microbioogical Methods (in press).
40
identifications could be assigned to a majority of the fragments in the environmental T-RFLP
results. Our empirical examination of the T-RFLP method identified limitations relative to
sequence-based community analysis, but the relative ease and low cost of fragment analysis
make this method a useful approach for characterizing the dominant taxa within complex
assemblages of microbial eukaryotes in large datasets.
Introduction
Protists are ubiquitous in marine environments and are responsible for a range of ecological roles
including primary production, nutrient regeneration, symbioses and the transfer of energy and
carbon to higher trophic levels (Sherr et al. 2007; Caron et al. 2012). The diversity and
functional roles of protists have been topics of interest since microbial eukaryotes were first
observed by Antonie van Leeuwenhoek in the 17
th
century. Morphological features have
constituted the ‘gold standard’ for the taxonomy of these species since that time but DNA
sequence information increasingly has been employed to augment morphology-based taxonomic
schemes (Adl et al. 2005), to elucidate phylogenetic relationships among protistan lineages
(Tekle et al. 2009), and most recently to complement morphology-based approaches for the
study of protistan diversity and ecology (Caron et al. 2009; Marande et al. 2009).
The development and application of a DNA-based taxonomy as a tool to study protistan
ecology is alluring because of the potential to characterize natural protistan assemblages using a
single method. Protistan communities in nature are comprised of species that span several orders
of magnitude in size, and many can be morphologically nondescript (especially minute forms)
and/or present at abundances that are below the detection limits of traditional methods.
Characterization of the entire protistan assemblage in a sample by traditional methods is
Kim DY, Countway PD, Yamashito W, Caron DA (2012) A combined sequence-based and fragment-based characterization of microbial
eukaryote assemblages provides taxonomic context for the Terminal Restriction Fragment Length Polymorphism (T-RFLP) method. Journal of
Microbioogical Methods (in press).
41
therefore exceedingly difficult, involving multiple methods for collection, preservation and
observation, thus limiting the number of samples that can be analyzed and the completeness of
the assessment (Caron 2009). However, ecological studies to understand the diversity and
spatiotemporal variability of protistan assemblages require the collection and examination of
large numbers of samples. Sequence-based approaches offer the possibility of a practical and
potentially widely applicable alternative to traditional approaches for characterizing the diversity
and taxonomic composition of protistan assemblages. The application of these methodologies
within the last decade has revealed an unexpectedly large diversity of protists in numerous
marine environments (Diez et al. 2001b; Lopez-Garcia et al. 2001; Amaral-Zettler et al. 2002;
Edgcomb et al. 2002; Moreira and Lopez-Garcia 2002; Lopez-Garcia et al. 2003; Stoeck and
Epstein 2003; Gast et al. 2004; Massana et al. 2004; Romari and Vaulot 2004; Countway et al.
2005; Worden 2006).
A variety of genetic methods have recently been developed for assessing the diversity of
natural microbial assemblages. These methods can be grouped broadly into approaches that rely
on the analysis of sequences from genes such as small subunit ribosomal RNA or cytochrome
oxidase, and the analysis of DNA fragment sizes from genes or other genomic regions. Analyses
of community structure and diversity using DNA sequence information rely on grouping
sequences obtained from environmental samples into Operational Taxonomic Units (OTUs)
based on sequence similarity or phylogenetic relatedness. For microbial eukaryotes, some
studies have attempted to delineate OTUs in a manner that approximates morphological species
distinctions (Caron et al. 2009; Nebel et al. 2011). One recent analysis employed well-curated
18S rRNA gene sequences from GenBank for approximating protistan species using an
Kim DY, Countway PD, Yamashito W, Caron DA (2012) A combined sequence-based and fragment-based characterization of microbial
eukaryote assemblages provides taxonomic context for the Terminal Restriction Fragment Length Polymorphism (T-RFLP) method. Journal of
Microbioogical Methods (in press).
42
automated OTU calling program, Microbial Eukaryote Species Assignment (MESA) (Caron et
al. 2009).
Two popular DNA fragment approaches that have been employed are Terminal
Restriction Fragment Length Polymorphism (T-RFLP) and Automated Ribosomal Intergenic
Spacer Analysis (Liu et al. 1997; Fisher and Triplett 1999). T-RFLP has been commonly used to
characterize microbial eukaryote assemblages (Countway et al. 2005; Dopheide et al. 2008;
Euringer and Lueders 2008; Vigil et al. 2009; Joo et al. 2010; Steele et al. 2011; Balzano et al.
2012). OTU calling for the T-RFLP method is predicated on the assumption that different
species will yield DNA fragments of unique length. Limitations of the method relate to the small
number of taxonomic units (unique fragment sizes) that can be detected in T-RFLP patterns, and
the lack of taxonomic identity of the species giving rise to particular fragments (Dunbar et al.
2000; Osborn et al. 2000). These caveats reduce the usefulness of the method (and DNA
fragment-based approaches in general) for assessing the taxonomic composition of microbial
assemblages. However, the relatively low cost and rapid analysis relative to sequencing make T-
RFLP a useful method for analyzing large ecological datasets.
DNA fragment-based and sequence-based analyses of microbial eukaryote assemblages
were carried out in order to provide taxonomic context for the T-RFLP method. Samples were
collected from 5 m and 500 m as a part of the USC Microbial Observatory at the San Pedro
Ocean Time-series (SPOT) station located in the eastern North Pacific, where 237 samples from
multiple water column depths have been analyzed using the T-RFLP method (Kim et al. 2012).
Environmental clone libraries of 18S rRNA genes were constructed and sequenced, and OTUs
were assembled and taxonomies assigned. The sequences were then used to create in silico
Kim DY, Countway PD, Yamashito W, Caron DA (2012) A combined sequence-based and fragment-based characterization of microbial
eukaryote assemblages provides taxonomic context for the Terminal Restriction Fragment Length Polymorphism (T-RFLP) method. Journal of
Microbioogical Methods (in press).
43
restriction digests in order to assign taxonomic identity to fragments observed in the T-RFLP
patterns obtained from the same samples. Environmental T-RFLP fragments in our analysis
generally represented the dominant sequence-based OTUs, and fragment lengths obtained from
the in silico digests of sequence-based OTUs allowed confident assignments of taxonomic
identities to many of the fragments from the T-RFLP analysis.
Methods
Sample collection
Seawater samples were collected from 2 depths (5 m and 500 m) at the San Pedro Ocean Time-
series (SPOT) station in the eastern North Pacific (118° 24’ W, 33° 33’ N) on October 29, 2001
using 10 L Niskin bottles attached to a CTD sampler rosette. Seawater samples were pre-
screened through 200 µm and 80 µm mesh screening by gravity-filtration to reduce the
contribution of metazoa to subsequent DNA analyses. The <80 µm filtrate was collected in acid-
washed 20 L polycarbonate carboys and 2 L of the filtrate from each depth were vacuum filtered
(~5 Hg) onto 47 mm GF/F filters (Whatman) on board the ship. The filters were immediately
transferred to 15 ml BD Falcon tubes (BD Biosciences, San Jose, CA) containing 2 ml of 2x
Lysis Buffer (100 mM Tris (pH 8), 40 mM EDTA (pH 8), 100 mM NaCl, 1% SDS) and flash
frozen in liquid nitrogen.
DNA Extraction and Purification
DNA was extracted from samples after their return to the lab. A combination of chemical and
mechanical methods described in Countway et al. (2005) was used to extract, isolate and purify
nucleic acids. DNA pellets were resuspended in TE buffer (10 mM Tris, 1 mM EDTA, pH 7.5)
and frozen at -20°C until further processing.
Kim DY, Countway PD, Yamashito W, Caron DA (2012) A combined sequence-based and fragment-based characterization of microbial
eukaryote assemblages provides taxonomic context for the Terminal Restriction Fragment Length Polymorphism (T-RFLP) method. Journal of
Microbioogical Methods (in press).
44
Cloning and Sequencing of full-length 18S rRNA genes
Cloning for the two environmental samples was carried out according to Countway et al. (2005)
using Euk-A and Euk-B universal primers (Medlin 1988). Sequencing was carried out by the
Joint Genome Institute (JGI, Walnut Creek, CA) using two vector primers (M13F and M13R)
and one internal primer (Euk-570F). The resulting contigs were assembled at the JGI, and
resulted in a total of 1,094 and 823 full-length (or near full-length) 18S rRNA gene sequences for
the 5 m and 500 m samples, respectively. The sequences were generated as a part of a larger
global survey of protistan diversity and used in the current study to provide taxonomic context
for T-RFLP analysis performed on the same samples. The raw assembled reads were screened to
trim vector sequences and also checked for correct orientation using Geneious (Drummond et al.
2011) and MOTHUR (Schloss et al. 2009). The sequences were further screened for the Euk-A
and the Euk-570R primer sites (allowing up to 3 mismatches for each), to ensure that only
sequences that had the T-RFLP primer regions were used to construct a fragment database for
assigning identifications to specific fragment lengths, and for comparison to T-RFLP-based
characterizations of the microbial eukaryote assemblages. The resulting sequences were checked
for chimeras using a local implementation of Pintail (Ashelford et al. 2005) and the SIVLA
(v102) database for eukaryotes (Pruesse et al. 2007). These quality control measures resulted in
663 sequences for the 5 m sample and 678 sequences for the 500 m sample for further analyses
(total of 1,341 sequences). The sequences (from Euk-A to Euk-570R) can be found in the
GenBank database (http://www.ncbi.nlm.nih.gov/genbank/; Accessions JX841335-JX842675).
Kim DY, Countway PD, Yamashito W, Caron DA (2012) A combined sequence-based and fragment-based characterization of microbial
eukaryote assemblages provides taxonomic context for the Terminal Restriction Fragment Length Polymorphism (T-RFLP) method. Journal of
Microbioogical Methods (in press).
45
OTU calling and taxonomy assignment of 18S rRNA gene sequences
The sequenced regions from Euk-A to Euk-570R were extracted from the 1,341 18S rDNA
sequences passing quality control, and clustered into Operational Taxonomic Units (OTUs) using
the Microbial Eukaryote Species Assignment program (MESA) at a 95% sequence similarity
(Caron et al. 2009). The selection of the 95% sequence similarity threshold in MESA is
discussed in detail by Caron et al. (2009), and is designed to provide approximately species-
level, automated OTU calling for protistan sequences. Each sequence was assigned a taxonomic
identification based on the best BLAST+ match to the SILVA (v108) database (Pruesse et al.
2007). BLAST+ results against GenBank (Benson et al. 2010) largely corroborated the results
from the SILVA (v108) database and also provided identities to novel or unknown groups such
as Group I and II Alveolates that were not found in SILVA (v108). All sequences within an
OTU represented the same high-level phylogenetic group. Genus and species-level
identifications were assigned to each sequence-based OTU based on a majority basis.
Terminal Restriction Fragment Length Polymorphism (T-RFLP) Analysis
DNA extracts were quantified using Pico Green (Invitrogen) on a BioRad fluorometer, and
approximately 10 ng of DNA template was used for each PCR reaction for T-RFLP analysis.
PCR reactions consisted of final concentrations of 0.5 µM of each primer (Euk-A labeled with
D4 fluorochrome, 5’-D4-AAC CTG GTT GAT CCT GCC AGT-3’ (Medlin et al, 1988) and
unlabeled Euk-570R, 5’GCT ATT GGA GCT GGA ATT AC-3’ (Elwood et al., 1985)), 1x
GoTaq Flexi Colorless Buffer (Promega), 2.5 mM MgCl
2
, 250 µM dNTPs, 300 ng/µl BSA
(Sigma), 2.5U of GoTaq (Promega) in a total reaction volume of 50 µl. The following thermal
protocol was used for each PCR: 1x (95°C for 2 min), 35x (95°C for 30 sec, 55°C for 30 sec,
Kim DY, Countway PD, Yamashito W, Caron DA (2012) A combined sequence-based and fragment-based characterization of microbial
eukaryote assemblages provides taxonomic context for the Terminal Restriction Fragment Length Polymorphism (T-RFLP) method. Journal of
Microbioogical Methods (in press).
46
72°C for 2 min), 1x (72°C for 7 min). Three PCRs were performed for each sample, PCR
products were visualized on a 1.2% SeaKem LE agarose gel and the triplicate PCRs were pooled
before being purified and concentrated using a QIAquick PCR Purification kit (Qiagen).
Mung bean nuclease (10U) and mung bean nuclease buffer (New England Biolabs) were
added to the cleaned and concentrated PCR products and incubated for 60 min at 30°C to remove
single-stranded PCR artifacts (Egert and Friedrich 2003). Products digested with mung bean
nuclease were purified and concentrated using the QIAquick PCR purification kit and quantified
using Pico Green on a VersaFluor fluorometer (BioRad). Approximately 300 ng of DNA was
used for restriction digests. The HaeIII and MnlI enzymes were selected based on their ability to
produce a large number of unique fragments from in silico digestions of 18S rRNA gene
sequences obtained from publicly available databases (Countway et al. 2005). The restriction
digestion reactions comprised: 10 U each of either HaeIII or MnlI enzyme (New England
Biolabs), Buffer 2 (New England Biolabs), BSA (MnlI only) and ~300 ng of DNA in a total
reaction volume of 20 µl. All restriction enzymatic reactions were incubated at 37°C for ~15 h
and the enzymes were inactivated by heating at 80°C and 65°C for 20 min for the HaeIII and
MnlI enzymes, respectively.
The digested DNA products were precipitated using 20 µg of glycogen (Roche), 2 µl of 3
M sodium acetate (Sigma-Aldrich) and 50 µl of 95% ice-cold ethanol and centrifugation at
14,000 RPM for 15 min. The supernatant was decanted and DNA pellets were rinsed twice with
100 µl of 70% ice-cold ethanol and centrifugation at 14,000 RPM for 5 min. The supernatant
was decanted after each round and finally the DNA pellets were air-dried and then resuspended
in 40 µl of Sample Loading Solution (SLS; a deionized formamide solution, Beckman-Coulter).
Kim DY, Countway PD, Yamashito W, Caron DA (2012) A combined sequence-based and fragment-based characterization of microbial
eukaryote assemblages provides taxonomic context for the Terminal Restriction Fragment Length Polymorphism (T-RFLP) method. Journal of
Microbioogical Methods (in press).
47
Approximately 5-10 µl of the resuspended samples were combined with 0.5 µl of the DNA-Size-
Standard-Kit-600 (Beckman Coulter) and SLS up to a volume of 40 µl. The resulting terminally
labeled fragments (T-RFs) were detected on a Beckman CEQ8000 capillary gel-electrophoresis
system (with a detection range between 60 and 640 bp with 1 bp resolution). The CEQ8000
protocol consisted of: 2.9 kV at 60°C for 70 sec and injection times ranged between 9 and 15
seconds. The volume for each sample and injection times were adjusted so that maximum
relative fluorescent units (rfu) for each sample ranged between 80,000 and 125,000 rfu.
T-RFs from cultures of Ostreococcus sp. and Phaeocystis globosa were used as positive
controls for verifying fragment sizes in the T-RFLP analyses. The 18S rRNA genes from these
two species were cloned, and purified plasmids containing the genes were amplified using T-
RFLP primers, digested with HaeIII and MnlI enzymes, and run as standards in parallel with the
environmental samples. Fragment sizes of these two species on the CEQ8000 were compared to
fragment sizes predicted from in silico digestion of the complete, sequenced genes to ensure the
accuracy of fragment sizes determined by the CEQ8000.
Raw fragment data were analyzed using the Fragment Analysis module of the CEQ8000
software package (Beckman Coulter) using 1 bp-wide bins and a threshold of 0.5% peak area.
Fragment peaks were normalized to the total peak area for each sample to allow comparisons of
relative abundances of fragments between samples (Kaplan and Kitts 2004).
Taxonomy assignment to DNA fragments
A perl script was written to perform an in silico digestion of the 1,341 sequences obtained by
cloning and sequencing, using the HaeIII (GG’CC) and MnlI (CCTC(N)
7
’ or ‘(N)
6
GAGG)
restriction sites. The fragments resulting from the in silico digestions were grouped according to
Kim DY, Countway PD, Yamashito W, Caron DA (2012) A combined sequence-based and fragment-based characterization of microbial
eukaryote assemblages provides taxonomic context for the Terminal Restriction Fragment Length Polymorphism (T-RFLP) method. Journal of
Microbioogical Methods (in press).
48
fragment size (i.e. into fragment-based OTUs), and taxonomic identities of these in silico
generated fragment OTUs were assigned based on the identities assigned to the sequences as
described above. In cases where more than a single identity was assigned to a fragment size (the
case for the majority of fragment sizes), identity was based on the predominant taxonomic
identity of the sequences that generated that fragment size. In addition, each of the ten most
populated sequence-based OTUs from each sample was examined to determine if one or more
fragment sizes were generated for each sequence-based OTU. Most sequence-based OTUs
yielded a predominant fragment size, but many of the OTUs also yielded one or more minor
fragments.
Taxonomic identities were assigned to fragments generated by the environmental T-
RFLP analysis of the two samples using the taxonomies generated from the sequence-based
identifications assigned to fragments obtained from the in silico digestions of the 1,341
sequences. A T-RF database (including the taxonomic assignments for each fragment) generated
from the sequences from both samples was constructed using Microsoft Access. The fragment
databases generated from in silico digestion of the sequences from 5 and 500 m, and the
assignment of putative taxonomies to the T-RFs generated by the environmental T-RFLP
analysis, were handled separately for the two samples.
Multivariate analysis
Fragment and sequence data were square-root transformed to down-weight the influence of the
most dominant taxa in each sample before Bray-Curtis similarities were calculated in PRIMER
(v6) & PERMANOVA+β18 (PRIMER-E Ltd).
Kim DY, Countway PD, Yamashito W, Caron DA (2012) A combined sequence-based and fragment-based characterization of microbial
eukaryote assemblages provides taxonomic context for the Terminal Restriction Fragment Length Polymorphism (T-RFLP) method. Journal of
Microbioogical Methods (in press).
49
Results
Sequence-based OTU calling
A total of 663 and 678 full-length (or near full-length) 18S rRNA gene sequences were analyzed
from the 5 m and 500 m samples in this study, respectively. The Euk-A and Euk-570R region of
these 1,341 sequences was extracted, pooled and used for OTU-calling using MESA in order to
correspond with the environmental T-RFLP analysis. A total of 307 OTUs were obtained for the
combined dataset (Table 1), and the relative abundances of sequences within the dataset yielded
a rank abundance curve with a characteristic shape comprised of relatively few abundant taxa
and a large number of rare taxa (Fig. 1A). The sequences in the 5 m sample clustered into 157
OTUs while the 500 m sequences clustered into 183 OTUs. The individual rank abundance
curves for both samples were similar in shape to the curve including all data, although only 33
OTUs (~18% of the 5m OTUs and ~21% of the 500m OTUs) were shared between the two
samples (Figs. 1B and 1C). Community similarity analysis between the two depths based on
Bray-Curtis similarity of the sequence-based OTUs was only 19%.
The ten most abundant sequence-based OTUs in the 5 m sample (which comprised 261 or
39% of the 5 m sequences) consisted of five dinoflagellate taxa (Gyrodinium spp., Heterocapsa
spp., two unknown syndiniales and a dinoflagellate most closely related to a species described
from the Ross Sea, Antarctica (Gast et al. 2006)), two ciliates (Strombidium cf. basimorphum
and an unknown ciliate), a chlorophyte (Bathycoccus prasinos), a chlorarachniophyte
(Partenskyella glossopodia), and a copepod (Table 2). The ten most abundant sequence-based
OTUs in the 500 m sample (which constituted 287 or 42% of the 500 m sequences) consisted of
2 dinoflagellates (Gyrodinium spp. and Karlodinium spp.), 4 Group II alveolates, 2 polycistines
Kim DY, Countway PD, Yamashito W, Caron DA (2012) A combined sequence-based and fragment-based characterization of microbial
eukaryote assemblages provides taxonomic context for the Terminal Restriction Fragment Length Polymorphism (T-RFLP) method. Journal of
Microbioogical Methods (in press).
50
(Arachnosphaera myriacantha and Larcopyle butschlii), a ciliate (Varistrombidium kielum) and
a hydrozoan (Apolemia sp.) (Table 2).
Predicting fragment sizes from in silico digestion of 18S rDNA sequences
The 1,341 18S rRNA gene sequences were subjected to HaeIII digestion in silico in order to
predict and compare the DNA fragment sizes observed in the T-RFLP analysis of the same
samples. HaeIII in silico digestion resulted in the detection of 34% fewer OTUs (i.e. unique
fragment sizes) in the 5 m sample (104 in silico HaeIII fragments) and 46% fewer OTUs in the
500 m sample (98 in silico HaeIII fragments) relative to sequence-based OTUs (Table 1, Figs.
2A and 2B). In silico digestions of the 1,341 sequences were also conducted using the MnlI
restriction site, and yielded 94 fragments for the 5 m sequences and 91 fragments for the 500 m
sequences (data not shown). Only results from the analyses conducted using HaeIII are reported
due to the larger number of unique fragment sizes produced by that enzyme. The distribution of
fragment sizes produced by HaeIII in silico digestion of the 18S sequences for the two samples
resembled a typical rank abundance curve (Fig. 2A). The numbers of in silico fragments that
represented ≥0.5% of the total number of fragments in the two samples were 34 and 44, similar
to the number of fragments generally detected in environmental T-RFLP (see below).
Community similarity analysis between the two depths based on Bray-Curtis similarity analysis
of the relative abundances of in silico HaeIII fragments in the two datasets was 53%.
In silico digestion of some of the 18S rRNA gene sequences resulted in either a lack of
fragmentation or fragment sizes that were below the detection range for the environmental T-
RFLP method used in this study (60 bp). There were 74 and 34 sequences from the 5 m and 500
Kim DY, Countway PD, Yamashito W, Caron DA (2012) A combined sequence-based and fragment-based characterization of microbial
eukaryote assemblages provides taxonomic context for the Terminal Restriction Fragment Length Polymorphism (T-RFLP) method. Journal of
Microbioogical Methods (in press).
51
m samples, respectively, that did not have a HaeIII restriction site (in which cases the entire
length of the Euk-A and Euk-570R region was used in the in silico T-RFLP database).
Comparison of in silico fragments and sequence-based OTUs
In silico digestion of the sequences in each of the ten most abundant sequence-based OTUs
generally resulted in one dominant fragment size (Fig. 3). Additional fragment sizes were
present in all but one OTU (OTU-500m-1 resulted in a single fragment size), but the relative
abundances of the additional fragment lengths were minor in most cases. There were four cases
in which in silico digestion of the sequences in a sequence-based OTU resulted in two fragment
sizes of relatively equal proportions (OTU-500m-3, 4, 6 and 7), in which case both fragment
sizes were reported (Table 2; also, significant contributions of secondary fragments were
indicated in Fig. 2 by a ‘+’).
The dominant fragment sizes that corresponded with the 10 most abundant sequence-
based OTUs were also some of the most abundant fragment sizes following the in silico
digestion of the 5 m or 500 m sequences (Figs. 2A and 2B; Fragment OTUs marked with ‘*’ and
‘+’). For example, the most abundant fragment length resulting from in silico digestion of the
sequences from the 5 m sample (which constituted ~8% of the total in silico fragments in that
dataset, see Fig. 2A) was generated largely by sequences from the most abundant sequence-
based OTU (~93%). However, some fragments with high relative abundances were derived from
in silico digestions of several sequence-based OTUs. For example, the third most abundant in
silico fragment in the 5 m sample (Fig. 2A) was generated by the digestion of sequences that
clustered into 18 separate sequence-based OTUs.
Kim DY, Countway PD, Yamashito W, Caron DA (2012) A combined sequence-based and fragment-based characterization of microbial
eukaryote assemblages provides taxonomic context for the Terminal Restriction Fragment Length Polymorphism (T-RFLP) method. Journal of
Microbioogical Methods (in press).
52
Putative taxonomic identifications and phylogenetic groupings were assigned to the in
silico fragment sizes (i.e. fragment-based OTUs, right side of Table 2) associated with the most
abundant sequence-based OTUs (left side of Table 2) by determining the best BLAST+ matches
of the 5 m or 500 m sequences giving rise to that particular fragment size (see ‘Fragment-based
Identification’, right side of Table 2). In general, there was good correspondence between the
taxonomy derived for the dominant fragment size and the taxonomy of the sequence-based OTU
(left vs. right side of Table 2). However, some fragment sizes had minor contributions from
sequences that yielded different phylogenetic affiliations (note parentheses, right side of Table
2). For example, the fragment size of 239 bp was generated from sequences that had highest
similarities to dinoflagellates except 2 sequences that had high affinity to chlorophyte sequences.
This situation was not the case for the sequence-based OTUs, which yielded the same higher-
level phylogenetic affiliations for all sequences within an OTU.
Comparison of environmental T-RFLP fragments and in silico fragments
T-RFLP analysis of the 5 m and 500 m environmental samples resulted in 38 and 37 uniquely
sized fragments using HaeIII restriction digests, respectively (Table 1, Figs. 4A and 4B). These
results are consistent with an analysis of 237 T-RFLP profiles for samples collected from
multiple depths over the course of ~10 years at the USC Microbial Observatory, which resulted
in an average of ~35 fragments per sample (Kim et al. 2012). A total of 61 unique and 14 shared
HaeIII fragment sizes were obtained between the 5 m and 500 m samples (Figs. 4A and 4B).
The combined MnlI T-RFLP analysis resulted in a total of 58 unique fragment sizes, 10 of which
were shared between the two depths (data not shown). Community similarity between the two
depths based on Bray-Curtis similarity analysis of the relative abundances of fragments in the T-
Kim DY, Countway PD, Yamashito W, Caron DA (2012) A combined sequence-based and fragment-based characterization of microbial
eukaryote assemblages provides taxonomic context for the Terminal Restriction Fragment Length Polymorphism (T-RFLP) method. Journal of
Microbioogical Methods (in press).
53
RFLP datasets was ~40% with HaeIII digestion, compared to ~53% obtained using the in silico
HaeIII fragments and 19% obtained using the sequence-based OTUs.
Approximately 79% (30/38) of the fragment sizes from the environmental T-RFLP
analysis of the 5 m sample were found in the in silico HaeIII fragment database of the 5 m
sequences (Figs. 5A and 5B). Many of the common in silico fragments were observed at high
relative abundance in the T-RFLP analysis of the environmental sample, including 5 of the
fragment sizes associated with the most abundant sequence-based OTUs (marked with ‘*’ in Fig.
5B). The two most abundant 5 m in silico HaeIII fragments were not detected in the
environmental T-RFLP results, however. There were also discrepancies between the relative
abundances of some of the in silico fragments and T-RFs from the T-RFLP analysis.
T-RFLP analysis and quantitative PCR for characterizing the dynamics of Ostreococcus
The temporal dynamics of the prasinophyte Ostreococcus sp. in surface waters at the USC
Microbial Observatory site was examined and compared using fragment analysis and
quantitative PCR (qPCR). The relative abundances in T-RFLP patterns of a 259 bp fragment
whose sequences matched Ostreococcus were compared to results from a previous study that
developed and applied a qPCR approach for Ostreococcus (Countway and Caron 2006). The
relative abundances of the prasinophyte reflected by fragment analysis in samples collected from
a depth of 5 m at approximately monthly intervals between Sept 2000 and Sept 2002 at the USC
Microbial Observatory correlated well with abundances determined by qPCR (Fig. 6). Three
major peaks in abundances (May 2001, December 2001, June 2002) were documented using
both methods. Extending the T-RFLP results to the end of September 2003 revealed an
Kim DY, Countway PD, Yamashito W, Caron DA (2012) A combined sequence-based and fragment-based characterization of microbial
eukaryote assemblages provides taxonomic context for the Terminal Restriction Fragment Length Polymorphism (T-RFLP) method. Journal of
Microbioogical Methods (in press).
54
additional peak in relative abundance in May of 2003 for the fragment size identified as
Ostreococcus.
Discussion
T-RFLP analysis of environmental samples
Genetic approaches have yielded significant new insight into the immense diversity of microbial
eukaryotes from a variety of geographic locations, depths and times (Massana and Pedros-Alio
2008). There are few studies, however, that have characterized extensive spatial or temporal
scales of variation within microbial eukaryote assemblages. This is, in part, due to the logistical
challenges of collecting large numbers of samples over space or time, as well as the cost and
time required for generating DNA sequence information and analyzing large datasets. Thus,
despite significant advances in microbial and molecular eukaryote ecology, significant gaps in
our understanding of spatial and temporal dynamics within assemblages of microbial eukaryotes
still remain.
T-RFLP provides a relatively rapid and inexpensive method for the characterization of
the dominant taxa within microbial eukaryote assemblages, as demonstrated by the comparison
of results from fragment-based and sequence-based analyses of the same samples in the current
study (Figs. 2A and 2B; 5A and 5B). Fragment-based approaches have been used to characterize
natural assemblages of microbial eukaryotes including pico-eukaryotes from different oceanic
regimes (Diez et al. 2001a), identify distinct sea-ice and water assemblages of protistan taxa
from the Ross Sea in Antarctica (Gast et al. 2004), establish rapid changes in a protistan
assemblage during a bottle incubation experiment (Countway et al. 2005), characterize ciliate
communities in stream biofilms (Dopheide et al. 2008), document rapid shifts in dominant taxa
Kim DY, Countway PD, Yamashito W, Caron DA (2012) A combined sequence-based and fragment-based characterization of microbial
eukaryote assemblages provides taxonomic context for the Terminal Restriction Fragment Length Polymorphism (T-RFLP) method. Journal of
Microbioogical Methods (in press).
55
in an estuary system (Vigil et al. 2009), and monitor the dynamics of potentially harmful algal
bloom causing species (Joo et al. 2010). A primary advantage of the T-RFLP method in these
studies has been the ability to analyze the large numbers of samples that are often required for
ecological studies.
Analysis of DNA fragment data is also relatively straight-forward and is predicated on
the assumption that different species will generate unique DNA fragment sizes. Although there
are caveats associated with this assumption, the formation of OTUs from fragment data is less
subjective than clustering sequence data into OTUs by employing one of a range of similarity
values presently in use. Indeed, a number of similarity thresholds have been applied for
grouping SSU rRNA gene sequences from environmental surveys of microbial eukaryotes into
OTUs (Caron et al. 2009; Nebel et al. 2011). The difficulty in establishing an ecologically
meaningful threshold for forming OTUs includes fundamental complexities associated with the
species concepts that have been applied to protists.
We employed the conservative threshold of 95% derived by Caron et al. (2009) in this
study. The availability of morphologically well-defined protistan species and associated
sequences in GenBank in that study facilitated the analysis of 211 complete and partial 18S
rRNA gene sequences to determine a sequence similarity threshold for delineating approximately
species-level operational taxonomic units (OTUs) for microbial eukaryotes (Caron et al. 2009).
The comparison of sequence similarities between strains of the same species as well as different
species of the same genus resulted in an average similarity of 95% for delineating approximately
species-level OTUs. The authors reported that although this value is generally lower than those
employed by other studies (which have ranged between 97% and 99%) and most likely
Kim DY, Countway PD, Yamashito W, Caron DA (2012) A combined sequence-based and fragment-based characterization of microbial
eukaryote assemblages provides taxonomic context for the Terminal Restriction Fragment Length Polymorphism (T-RFLP) method. Journal of
Microbioogical Methods (in press).
56
overlooks physiological variability in some OTUs, it provides a conservative estimate of species
(i.e. OTU) richness for analyzing 18S rRNA gene sequence data.
Increasing the sequence similarity (%) used to cluster sequences into taxonomic units can
substantially inflate the number of OTUs detected in a sample (Caron et al. 2009), which can
arguably lead to a gross overestimation of species richness or diversity. Conversely, fragment-
based approaches provide a limited perspective of species richness (only characterizing the
dominant taxa), but may provide an assessment of diversity that is intermediate between
sequencing and morphological methods (Dopheide et al. 2008).
Limitations of T-RFLP
Since the introduction of the T-RFLP method in microbial ecology for the assessment of 16S
rRNA genes (Liu et al. 1997), limitations of the method as well methodological improvements
and assessments have been reported (Lukow et al. 2000; Egert and Friedrich 2003; Lueders and
Friedrich 2003; Abdo et al. 2006; Pandey et al. 2007; Schutte et al. 2008; Zhang et al. 2008;
Orcutt et al. 2009). One limitation of the T-RFLP method is an inability to characterize
members of the ‘rare biosphere.’ This seemingly ubiquitous feature of microbial eukaryote
assemblages (Figs. 1A-C) has been hypothesized to play an ecologically important role in
community response and reassembly (Pedros-Alio 2007; Caron and Countway 2009). The
functional roles and specific relationships of rare taxa within protistan assemblages are presently
unknown, but it is possible that at least some rare taxa play critical functional roles (e.g.,
keystone species). Such taxa that remain perpetually rare in a natural assemblage, but might still
be ecologically significant, would be undetected by T-RFLP analysis. For this reason, fragment-
based results are not appropriate for estimating total species richness (Dunbar et al. 2000).
Kim DY, Countway PD, Yamashito W, Caron DA (2012) A combined sequence-based and fragment-based characterization of microbial
eukaryote assemblages provides taxonomic context for the Terminal Restriction Fragment Length Polymorphism (T-RFLP) method. Journal of
Microbioogical Methods (in press).
57
Fragment-based approaches can also lead to higher estimates of community similarity than
sequenced-based approaches presumably due to the inability to differentiate some taxa and detect
rare taxa (Table 1; also, compare Figs. 1B and 1C to 2A and 2B as well as 4A and 4B). On the
positive side, changes in the dominant taxa of protistan assemblages can occur within hours to
days in response to changes in environmental conditions (Countway et al. 2005; Kim et al.
2011), and these exchanges can be captured by fragment analysis.
Improving T-RFLP analysis by providing taxonomic context using sequence-based analyses
DNA fragment analyses are most informative for characterizing changes within microbial
eukaryote assemblages when taxonomic context is provided for the fragments. The inability to
differentiate taxa, however, can complicate the assignment of taxonomic identifications to
fragment sizes. Fragment-based studies that have utilized publicly available sequences from
GenBank or SILVA to generate in silico T-RFLP databases or have selectively sequenced clones
with unique RFLP signatures for identifying fragments (Diez et al. 2001a; Fernandez-Guerra et
al. 2010) have noted this caveat. Here, we directly compared results from sequence-based and
fragment-based analyses of the same samples to provide taxonomic context for the T-RFLP
results.
Our combined sequence-based and fragment-based analysis of the same samples
provided confident assignment of taxonomic identities to many of the fragments resulting from
T-RFLP analysis (Figs 5A and 5B). Furthermore, our highly comparable results of the qPCR
results and T-RFLP results for characterizing the dynamics of Ostreococcus (Fig. 6) indicated
that at least some of the fragments in the current study were identified accurately using the
combined analysis approach, and that fragment data could be useful for investigating the
Kim DY, Countway PD, Yamashito W, Caron DA (2012) A combined sequence-based and fragment-based characterization of microbial
eukaryote assemblages provides taxonomic context for the Terminal Restriction Fragment Length Polymorphism (T-RFLP) method. Journal of
Microbioogical Methods (in press).
58
spatiotemporal distributions of some protists. We observed a strong correlation between
absolute 18S rRNA gene abundance (as measured by qPCR) of Ostreococcus and the relative
abundances of the fragment identified as Ostreococcus by T-RFLP. We suspect that this tight
relationship may have been in part due to the generally oligotrophic conditions of the USC
Microbial Observatory station, where monthly changes in bacterial or protistan assemblages are
generally not substantial (Fuhrman et al. 2006; Kim et al. 2012). A comparison of results from
sequence-based and fragment-based analyses also affirmed that fragments observed by T-RFLP
generally represented the most abundant sequence-based OTUs (Figs. 3A and 3B; 5A and 5B).
Conclusions
T-RFLP can provide a relatively rapid and inexpensive snapshot of the dominant taxa within
complex assemblages of microbial eukaryotes. The analysis of fragment data is relatively
straight-forward compared to the computational power required to analyze sequences. These
characteristics provide advantages for analyzing large numbers of samples that are often required
for ecological studies of microbial eukaryote assemblages. The combined sequence-based and
fragment-based analyses of the same samples in this study resulted in the confident assignment
of taxonomic identification to many of the fragments observed by T-RFLP and affirmed that
fragments generally represented the dominant sequence-based OTUs. Limitations of T-RFLP
exist and were demonstrated as a part of the current study, most significantly the inability of T-
RFLP to differentiate some taxa and detect rare taxa. The combined analysis of sequence-based
and fragment-based analyses of the same samples in the current study provided taxonomic
context for the analysis of T-RFLP data from approximately monthly samples collected from
multiple depths at the USC Microbial Observatory between 2000 and 2010 (Kim et al. 2012).
Kim DY, Countway PD, Yamashito W, Caron DA (2012) A combined sequence-based and fragment-based characterization of microbial
eukaryote assemblages provides taxonomic context for the Terminal Restriction Fragment Length Polymorphism (T-RFLP) method. Journal of
Microbioogical Methods (in press).
59
References
Abdo Z, Schuette UME, Bent SJ, Williams CJ, Forney LJ, Joyce P (2006) Statistical methods for
characterizing diversity of microbial communities by analysis of terminal restriction
fragment length polymorphisms of 16S rRNA genes. Environ Microbiol 8, 929-938.
Adl SM, Simpson AGB, Farmer MA, Andersen RA, Anderson OR, Barta JR, Bowser SS,
Brugerolle G, Fensome RA, Fredericq S, James TY, Karpov S, Kugrens P, Krug J, Lane
CE, Lewis LA, Lodge J, Lynn DH, Mann DG, McCourt RM, Mendoza L, Moestrup O,
Mozley-Standridge SE, Nerad TA, Shearer CA, Smirnov AV, Spiegel FW, Taylor MFJR
(2005) The new higher level classification of eukaryotes with emphasis on the taxonomy
of protists. J Eukaryot Microbiol 52, 399-451.
Amaral-Zettler L, Gomez F, Zettler E, Keenan BG, Amils R, Sogin M (2002) Eukaryotic
diversity in Spain's River of Fire. Nature 417, 137.
Ashelford KE, Chuzhanova NA, Fry JC, Jones AJ, Weightman AJ (2005) At least 1 in 20 16S
rRNA sequence records currently held in public repositories is estimated to contain
substantial anomalies. Appl Environ Microbiol 71, 7724-7736.
Balzano S, Marie D, Gourvil P, Vaulot D (2012) Composition of the summer photosynthetic
pico and nanoplankton communities in the Beaufort Sea assessed by T-RFLP and
sequences of the 18S rRNA gene from flow cytometry sorted samples. ISME J 6, 1480-
1498.
Benson DA, Karsch-Mizrachi I, Lipman DJ, Ostell J, Sayers EW (2010) GenBank. Nucleic Acids
Res 38, D46-D51.
Kim DY, Countway PD, Yamashito W, Caron DA (2012) A combined sequence-based and fragment-based characterization of microbial
eukaryote assemblages provides taxonomic context for the Terminal Restriction Fragment Length Polymorphism (T-RFLP) method. Journal of
Microbioogical Methods (in press).
60
Caron DA (2009) New accomplishments and approaches for assessing protistan diversity and
ecology in natural ecosystems. BioSci 59, 287-299.
Caron DA, Countway PD (2009) Hypotheses on the role of the protistan rare biosphere in a
changing world. Aquat Microb Ecol 57, 227-238.
Caron DA, Countway PD, Jones AC, Kim DY, Schnetzer A (2012) Marine protistan diversity.
Annu Rev Mar Sci 4, 467-493.
Caron DA, Countway PD, Savai P, Gast RJ, Schnetzer A, Moorthi SD, Dennett MR, Moran DM,
Jones AC (2009) Defining DNA-based Operational Taxonomic Units for microbial-
eukaryote ecology. Appl Environ Microbiol 75, 5797-5808.
Countway PD, Caron DA (2006) Abundance and distribution of Ostreococcus sp in the San
Pedro Channel, California, as revealed by quantitative PCR. Appl Environ Microbiol 72,
2496-2506.
Countway PD, Gast RJ, Savai P, Caron DA (2005) Protistan diversity estimates based on 18S
rDNA from seawater incubations in the western North Atlantic. J Eukaryot Microbiol 52,
95-106.
Diez B, Pedros-Alio C, Marsh TL, Massana R (2001a) Application of denaturing gradient gel
electrophoresis (DGGE) to study the diversity of marine picoeukaryotic assemblages and
comparison of DGGE with other molecular techniques. Appl Environ Microbiol 67,
2942-2951.
Diez B, Pedros-Alio C, Massana R (2001b) Study of genetic diversity of eukaryotic picoplankton
in different oceanic regions by small-subunit rRNA gene cloning and sequencing. Appl
Environ Microbiol 67, 2932-2941.
Kim DY, Countway PD, Yamashito W, Caron DA (2012) A combined sequence-based and fragment-based characterization of microbial
eukaryote assemblages provides taxonomic context for the Terminal Restriction Fragment Length Polymorphism (T-RFLP) method. Journal of
Microbioogical Methods (in press).
61
Dopheide A, Lear G, Stott R, Lewis G (2008) Relative eiversity and community structure of
ciliates in stream biofilms according to molecular and microscopy methods. Appl Environ
Microbiol 75, 5261-5272.
Drummond AJ, Ashton B, Buxton S, Cheung M, Cooper A, Duran C, Field M, Heled J, Kearse
M, Markowitz S, Moir R, Stones-Havas S, Sturrock S, Thierer T, Wilson A (2011)
Geneious v5.4, available from http://www.geneious.com/.
Dunbar J, Ticknor LO, Kuske CR (2000) Assessment of microbial diversity in four southwestern
United States soils by 16S rRNA gene terminal restriction fragment analysis. Appl
Environ Microbiol 66, 2943-2950.
Edgcomb VP, Kysela DT, Teske A, Gomez AD, Sogin ML (2002) Benthic eukaryotic diversity
in the Guaymas Basin hydrothermal vent environment. Proc Natl Acad Sci USA 99,
7658-7662.
Egert M, Friedrich MW (2003) Formation of pseudo-terminal restriction fragments, a PCR-
related bias affecting terminal restriction fragment length polymorphism analysis of
microbial community structure. Appl Environ Microbiol 69, 2555-2562.
Euringer K, Lueders T (2008) An optimised PCR/T-RFLP fingerprinting approach for the
investigation of protistan communities in groundwater environments. J Microbiol Meth
75, 262-268.
Fernandez-Guerra A, Buchan A, Mou XZ, Casamayor EO, Gonzalez JM (2010) T-RFPred: a
nucleotide sequence size prediction tool for microbial community description based on
terminal-restriction fragment length polymorphism chromatograms. Bmc Microbiol 10.
Kim DY, Countway PD, Yamashito W, Caron DA (2012) A combined sequence-based and fragment-based characterization of microbial
eukaryote assemblages provides taxonomic context for the Terminal Restriction Fragment Length Polymorphism (T-RFLP) method. Journal of
Microbioogical Methods (in press).
62
Fisher MM, Triplett EW (1999) Automated approach for ribosomal intergenic spacer analysis of
microbial diversity and its application to freshwater bacterial communities. Appl Environ
Microbiol 65, 4630-4636.
Fuhrman JA, Hewson I, Schwalbach MS, Steele JA, Brown MV, Naeem S (2006) Annually
reoccurring bacterial communities are predictable from ocean conditions. Proc Natl Acad
Sci USA 103, 13104-13109.
Gast RJ, Dennett MR, Caron DA (2004) Characterization of protistan assemblages in the Ross
Sea, Antarctica, by denaturing gradient gel electrophoresis. Appl Environ Microbiol 70,
2028-2037.
Joo S, Lee SR, Park S (2010) Monitoring of phytoplankton community structure using terminal
restriction fragment length polymorphism (T-RFLP). J Microbiol Meth 81, 61-68.
Kaplan CW, Kitts CL (2004) Bacterial succession in a petroleum land treatment unit. Appl
Environ Microbiol 70, 1777-1786.
Kim DY, Countway PD, Gast RJ, Caron DA (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.
Kim DY, Countway PD, Yamashito W, Tung C, Caron DA (2012) Monthly, seasonal and
interannual variability of microbial eukaryote assemblages within and below the euphotic
zone in the eastern North Pacific. (submitted).
Liu WT, Marsh TL, Cheng H, Forney LJ (1997) Characterization of microbial diversity by
determining terminal restriction fragment length polymorphisms of genes encoding 16S
rRNA. Appl Environ Microbiol 63, 4516-4522.
Kim DY, Countway PD, Yamashito W, Caron DA (2012) A combined sequence-based and fragment-based characterization of microbial
eukaryote assemblages provides taxonomic context for the Terminal Restriction Fragment Length Polymorphism (T-RFLP) method. Journal of
Microbioogical Methods (in press).
63
Lopez-Garcia P, Philippe H, Gail F, Moreira D (2003) Autochthonous eukaryotic diversity in
hydrothermal sediment and experimental microcolonizers at the Mid-Atlantic Ridge.
Proc Natl Acad Sci USA 100, 697-702.
Lopez-Garcia P, Rodriguez-Valera F, Pedros-Alio C, Moreira D (2001) Unexpected diversity of
small eukaryotes in deep-sea Antarctic plankton. Nature 409, 603-607.
Lueders T, Friedrich MW (2003) Evaluation of PCR amplification bias by terminal restriction
fragment length polymorphism analysis of small-subunit rRNA and mcrA genes by using
defined template mixtures of methanogenic pure cultures and soil DNA extracts. Appl
Environ Microbiol 69, 320-326.
Lukow T, Dunfield PF, Liesack W (2000) Use of the T-RFLP technique to assess spatial and
temporal changes in the bacterial community structure within an agricultural soil planted
with transgenic and non-transgenic potato plants. FEMS Microbiol Ecol 32, 241-247.
Marande W, Lopez-Garcia P, Moreira D (2009) Eukaryotic diversity and phylogeny using small-
and large-subunit ribosomal RNA genes from environmental samples. Environ Microbiol
11, 3179-3188.
Massana R, Castresana J, Balague V, Guillou L, Romari K, Groisillier A, Valentin K, Pedros-
Alio C (2004) Phylogenetic and ecological analysis of novel marine stramenopiles. Appl
Environ Microbiol 70, 3528-3534.
Massana R, Pedros-Alio C (2008) Unveiling new microbial eukaryotes in the surface ocean.
Curr Opin Microbiol 11, 213-218.
Medlin LK, Elwood, H. J., Stickel, S., Sogin, M. L. (1988) The characterization of enzymatically
amplified eukaryotic 16S-like rRNA-coding regions. Genetica 71, 491-499.
Kim DY, Countway PD, Yamashito W, Caron DA (2012) A combined sequence-based and fragment-based characterization of microbial
eukaryote assemblages provides taxonomic context for the Terminal Restriction Fragment Length Polymorphism (T-RFLP) method. Journal of
Microbioogical Methods (in press).
64
Moreira D, Lopez-Garcia P (2002) The molecular ecology of microbial eukaryotes unveils a
hidden world. Trends Microbiol 10, 31-38.
Nebel M, Pfabel C, Stock A, Dunthorn M, Stoeck T (2011) Delimiting operational taxonomic
units for assessing ciliate environmental diversity using small-subunit rRNA gene
sequences. Env Microbiol Rep 3, 154-158.
Orcutt B, Bailey B, Staudigel H, Tebo BM, Edwards KJ (2009) An interlaboratory comparison
of 16S rRNA gene-based terminal restriction fragment length polymorphism and
sequencing methods for assessing microbial diversity of seafloor basalts. Environ
Microbiol 11, 1728-1735.
Osborn AM, Moore ERB, Timmis KN (2000) An evaluation of terminal-restriction fragment
length polymorphism (T-RFLP) analysis for the study of microbial community structure
and dynamics. Environ Microbiol 2, 39-50.
Pandey J, Ganesan K, Jain RK (2007) Variations in T-RFLP profiles with differing chemistries
of fluorescent dyes used for labeling the PCR primers. J Microbiol Meth 68, 633-638.
Pedros-Alio C (2007) Dipping into the rare biosphere. Science 315, 192-193.
Pruesse E, Quast C, Knittel K, Fuchs BM, Ludwig WG, Peplies J, Glockner FO (2007) SILVA: a
comprehensive online resource for quality checked and aligned ribosomal RNA sequence
data compatible with ARB. Nucleic Acids Res 35, 7188-7196.
Romari K, Vaulot D (2004) Composition and temporal variability of picoeukaryote communities
at a coastal site of the English Channel from 18S rDNA sequences. Limnol Oceanogr 49,
784-798.
Kim DY, Countway PD, Yamashito W, Caron DA (2012) A combined sequence-based and fragment-based characterization of microbial
eukaryote assemblages provides taxonomic context for the Terminal Restriction Fragment Length Polymorphism (T-RFLP) method. Journal of
Microbioogical Methods (in press).
65
Schloss PD, Westcott SL, Ryabin T, Hall JR, Hartmann M, Hollister EB, Lesniewski RA,
Oakley BB, Parks DH, Robinson CJ, Sahl JW, Stres B, Thallinger GG, Van Horn DJ,
Weber CF (2009) Introducing mothur: open-Source, platform-independent, community-
supported software for describing and comparing microbial communities. Appl Environ
Microbiol 75, 7537-7541.
Schutte UME, Abdo Z, Bent SJ, Shyu C, Williams CJ, Pierson JD, Forney LJ (2008) Advances
in the use of terminal restriction fragment length polymorphism (T-RFLP) analysis of
16S rRNA genes to characterize microbial communities. Appl Microbiol Biot 80, 365-
380.
Sherr BF, Sherr EB, Caron DA, Vaulot D, Worden AZ (2007) Oceanic protists. Oceanography
20, 130-134.
Steele JA, Countway PD, Xia L, Vigil PD, Beman JM, Kim DY, Chow CET, Sachdeva R, Jones
AC, Schwalbach MS, Rose JM, Hewson I, Patel A, Sun FZ, Caron DA, Fuhrman JA
(2011) Marine bacterial, archaeal and protistan association networks reveal ecological
linkages. Isme J 5, 1414-1425.
Stoeck T, Epstein S (2003) Novel eukaryotic lineages inferred from small-subunit rRNA
analyses of oxygen-depleted marine environments. Appl Environ Microbiol 69, 2657-
2663.
Tekle YI, Parfrey LW, Katz LA (2009) Molecular data are transforming hypotheses on the origin
and diversification of eukaryotes. BioSci 59, 471-481.
Kim DY, Countway PD, Yamashito W, Caron DA (2012) A combined sequence-based and fragment-based characterization of microbial
eukaryote assemblages provides taxonomic context for the Terminal Restriction Fragment Length Polymorphism (T-RFLP) method. Journal of
Microbioogical Methods (in press).
66
Vigil P, Countway PD, Rose J, Lonsdale DJ, Gobler CJ, Caron DA (2009) Rapid shifts in
dominant taxa among microbial eukaryotes in estuarine ecosystems. Aquat Microb Ecol
54, 83-100.
Worden AZ (2006) Picoeukaryote diversity in coastal waters of the Pacific Ocean. Aquat Microb
Ecol 43, 165-175.
Zhang R, Thiyagarajan V, Qian PY (2008) Evaluation of terminal-restriction fragment length
polymorphism analysis in contrasting marine environments. FEMS Microbiol Ecol 65,
169-178.
Kim DY, Countway PD, Yamashito W, Caron DA (2012) A combined sequence-based and fragment-based characterization of microbial
eukaryote assemblages provides taxonomic context for the Terminal Restriction Fragment Length Polymorphism (T-RFLP) method. Journal of
Microbioogical Methods (in press).
67
Table 1 Summary of the data used to compare sequence-based and fragment-based OTUs: the
number of 18S rRNA gene sequences; the number of operational taxonomic units (OTUs, called
at 95% sequence similarity in MESA); the number of in silico fragments that resulted from in
silico digestion of the 5 m and 500 m sequences; the number of fragments arising from
environmental T-RFLP analysis of the samples, including matches to the in silico database.
Summaries are shown for each depth separately and combined. Values in brackets under the
combined analysis column are the numbers of OTUs or fragments that were found in both
samples.
Method used to derive OTUs
Number of
OTUs from
5 m
Number of
OTUs from
500 m
Number of OTUs from combined
analysis of 5 m and 500 m
(shared OTUs in brackets)
Sequence-based 157 183 307 (33)
In silico fragment-based 104 98 146 (56)
Environmental T-RFLP-based 38 37 61 (14)
Environmental T-RFLP-based OTUs
also found in the in silico fragment-
based OTUs
31 22 na
Kim DY, Countway PD, Yamashito W, Caron DA (2012) A combined sequence-based and fragment-based characterization of microbial
eukaryote assemblages provides taxonomic context for the Terminal Restriction Fragment Length Polymorphism (T-RFLP) method. Journal of
Microbioogical Methods (in press).
68
Table 2 Summary of the ten most abundant OTUs based on partial 18S sequences obtained for
the 5 m and 500 m samples, the terminally labeled fragment lengths predicted by in silico
digestion for those sequences, and a comparison of taxonomic identities assigned to the
sequence-based OTUs and the dominant fragment size for each OTU (see Figs. 3A and 3B). The
number of sequences in these sequence-based OTUs ranged between 14 and 87. Taxonomic
identifications and phylogenetic groups were determined from the sequences in each OTU as
described in the methods. All sequences within an OTU were then subjected to in silico HaeIII
restriction digestion. Not all sequences within a sequence-based OTU yielded the same fragment
size, in which case the dominant fragment size predicted for each OTU was reported (‘Predicted
in silico fragment size’; also reported in Fig. 3). Taxonomies were assigned to the dominant
fragment from each OTU based on the taxonomic identity of the sequences yielding that
fragment length, as well as the major phylogenetic group for each fragment (‘Fragment-based
Identification’). The phylogenetic group determined for a dominant fragment for each OTU was
generally but not always the same (see parentheses in last column).
Kim DY, Countway PD, Yamashito W, Caron DA (2012) A combined sequence-based and fragment-based characterization of microbial eukaryote assemblages provides taxonomic context for the
Terminal Restriction Fragment Length Polymorphism (T-RFLP) method. Journal of Microbioogical Methods (in press).
69
Depth
(m)
Seq-based
OTU (rank
abundance)
Number
of
sequences
Sequence-based Identification
Predicted
in silico
fragment
size (bp)
Fragment-based Identification
Taxonomy
Phylogenetic
Group
Taxonomy Phylogenetic Group
1 67 Gyrodinium spp. Dinoflagellate 239 Gyrodinium spp. Dinoflagellate (49/51)
2 33 Strombidium cf. basimorphum Ciliate 273 Strombidium spp. Ciliate (18/19)
3 29 Bathycoccus prasinos Chlorophyte 269 Bathycoccus prasinos Chlorophyte (27/27)
5
4 24 Paracalanus parvus Copepod 499 Paracalanus parvus Copepod (14/14)
5 22 unknown ciliate Ciliate 587 Strombidium sp. Ciliate (20/21)
6 21 uncultured syndiniales Dinoflagellate 279 uncultured syndiniales Dinoflagellate (24/26)
7 17 Dinophyceae sp. RS-24 Dinoflagellate 339 Pentapharsodinium tyrrhenicum Dinoflagellate (42/47)
8 17 Heterocapsa spp. Dinoflagellate 337 Heterocapsa spp. Dinoflagellate (28/32)
9 16 uncultured syndiniales Dinoflagellate 279 uncultured syndiniales Dinoflagellate (24/26)
10 15 Partenskyella glossopodia Cercozoa 282 Partenskyella glossopida Cercozoa (10/13)
1 87 Group II Alveolate Dinoflagellate 335 Group II Alveolate Dinoflagellate (139/139)
2 55 Arachnosphaera myriacantha Polycystine 333 Arachnosphaera myriacantha Polycistine (47/50)
3 29 Varistrombidium kielum Ciliate 273 Strombidium basimorphum Ciliate (12/12)
274 Novistrombidium sp. Ciliate (5/6)
500
4 18 Group II Alveolate Dinoflagellate 281 Group II Alveolate Dinoflagellate (8/9)
337 Group II Alveolate Dinoflagellate (27/28)
5 18 Apolemia sp. Cnidaria 279 Apolemia sp. Cnidaria (16/17)
6 18 Larchopyle butschlii Polycystine 181 Larchopyle butschlii Rhizaria (8/12)
270 Arachnosphaera myriacantha Rhizaria (9/10)
7 17 Group II Alveolate Dinoflagellate 337 Group II Alveolate Dinoflagellate (27/28)
338 Group II Alveolate Dinoflagellate (22/22)
8 16 Group II Alveolate Dinoflagellate 340 Group II Alveolate Dinoflagellate (35/35)
9 15 Karlodinium spp. Dinoflagellate 335 Group II Alveolate Dinoflagellate (139/139)
10 14 Gyrodinium spp. Dinoflagellate 239 Gyrodinium spp. Dinoflagellate (13/13)
Kim DY, Countway PD, Yamashito W, Caron DA (2012) A combined sequence-based and fragment-based characterization of microbial
eukaryote assemblages provides taxonomic context for the Terminal Restriction Fragment Length Polymorphism (T-RFLP) method. Journal of
Microbioogical Methods (in press).
70
Figure Legends
Figure 1 Rank abundance curves for microbial eukaryote assemblages collected from 5 m and
500 m at the USC Microbial Observatory site, based on partial sequences of 18S rRNA genes.
Sequences from the total dataset (1,341 18S rDNA sequences) were combined for OTU-calling
(A), and then separated by sampling depth (B, 5 m; C, 500 m). OTUs in the total dataset were
ranked from highest to lowest relative abundance (A), and the x-axes are the same for the three
rank abundance curves.
Figure 2 Rank abundance curves depicting the relative abundances of unique fragment sizes (i.e.
fragment OTUs) resulting from HaeIII in silico digestion of 18S rDNA sequences from two
samples collected at the USC Microbial Observatory site. Rank abundance curves of fragment
OTUs for the 5 m (A) and 500 m (B) samples were generated using the same sequences from
Fig. 1. The x-axes are arranged from highest to lowest abundance of the 5 m fragments.
Asterisks (*) indicate the fragments resulting from the in silico digestion of the ten most
abundant sequence-based OTUs in each sample (also see Fig. 3). Three of the most abundant
sequence-based OTUs yielded fragments of the same size (stacked asterisks). The in silico
digestion of four sequence-based OTUs from the 500 m sample yielded two fragment lengths of
relatively equal proportions (secondary fragments are indicated by ‘+’). There is a break in the
y-axis for the 500 m rank abundance curve to facilitate comparison between the samples.
Kim DY, Countway PD, Yamashito W, Caron DA (2012) A combined sequence-based and fragment-based characterization of microbial
eukaryote assemblages provides taxonomic context for the Terminal Restriction Fragment Length Polymorphism (T-RFLP) method. Journal of
Microbioogical Methods (in press).
71
Figure 3 Distributions of the fragment sizes resulting from HaeIII in silico digestions of the ten
most abundant sequence-based OTUs in the 5 m (A) and 500 m (B) samples. The ten most
abundant sequence-based OTUs comprised 39% and 42% of the total number of sequences in the
5 m and 500 m samples, respectively, and are depicted from most abundant on the left (OTU-
5m-1) to least abundant on the right (OTU-5m-10). The dominant fragment lengths associated
with each sequence-based OTU are labeled. Four of the sequence-based OTUs obtained from
the 500 m sample yielded two fragments of approximately similar relative abundances.
Figure 4 Rank abundance curves of HaeIII T-RFLP results obtained from environmental
samples collected at 5 m (A) and 500 m (B). The x-axes have been normalized to the relative
abundances of fragments in the 5 m sample to highlight unique and shared fragment sizes
between the samples. There is a break in the y-axis of the 500 m rank abundance curve to
facilitate comparison between the samples.
Figure 5 Comparison of rank abundance curves of (A) a fragment dataset obtained from HaeIII
in silico digestion of 663 partial 18S sequences from a 5 m sample (same data from Fig. 2A), and
(B) environmental T-RFLP fragments generated by HaeIII digestion of DNA from the same
sample (same data from Fig. 4A, rearranged for comparison to the in silico digestion in A). Both
x-axes are ordered by relative abundances of in silico fragments. Asterisks mark the fragment
sizes generated by the ten most abundant sequence-based OTUs (from Fig. 3A).
Kim DY, Countway PD, Yamashito W, Caron DA (2012) A combined sequence-based and fragment-based characterization of microbial
eukaryote assemblages provides taxonomic context for the Terminal Restriction Fragment Length Polymorphism (T-RFLP) method. Journal of
Microbioogical Methods (in press).
72
Figure 6 Time-series of changes in the absolute abundances of Ostreococcus sp. (filled triangles;
solid lines) as detected by quantitative PCR (Countway and Caron 2006) in samples collected at
approximately monthly intervals between September 2000 and September 2002 from the USC
Microbial Observatory site compared to relative abundances of a 259 bp T-RFLP fragment
(identified as Ostreococcus sp. from the database in this study) in the same samples (open
circles; dotted line). The time-series for the fragment dataset was extended to September 2003.
Kim DY, Countway PD, Yamashito W, Caron DA (2012) A combined sequence-based and fragment-based characterization of microbial
eukaryote assemblages provides taxonomic context for the Terminal Restriction Fragment Length Polymorphism (T-RFLP) method. Journal of
Microbioogical Methods (in press).
73
Figure 1
Kim DY, Countway PD, Yamashito W, Caron DA (2012) A combined sequence-based and fragment-based characterization of microbial
eukaryote assemblages provides taxonomic context for the Terminal Restriction Fragment Length Polymorphism (T-RFLP) method. Journal of
Microbioogical Methods (in press).
74
Figure 2
Kim DY, Countway PD, Yamashito W, Caron DA (2012) A combined sequence-based and fragment-based characterization of microbial
eukaryote assemblages provides taxonomic context for the Terminal Restriction Fragment Length Polymorphism (T-RFLP) method. Journal of
Microbioogical Methods (in press).
75
Figure 3
Kim DY, Countway PD, Yamashito W, Caron DA (2012) A combined sequence-based and fragment-based characterization of microbial
eukaryote assemblages provides taxonomic context for the Terminal Restriction Fragment Length Polymorphism (T-RFLP) method. Journal of
Microbioogical Methods (in press).
76
Figure 4
Kim DY, Countway PD, Yamashito W, Caron DA (2012) A combined sequence-based and fragment-based characterization of microbial
eukaryote assemblages provides taxonomic context for the Terminal Restriction Fragment Length Polymorphism (T-RFLP) method. Journal of
Microbioogical Methods (in press).
77
Figure 5
Kim DY, Countway PD, Yamashito W, Caron DA (2012) A combined sequence-based and fragment-based characterization of microbial
eukaryote assemblages provides taxonomic context for the Terminal Restriction Fragment Length Polymorphism (T-RFLP) method. Journal of
Microbioogical Methods (in press).
78
Figure 6
79
Chapter 3
Monthly, seasonal and interannual variability of microbial eukaryote assemblages within
and below the euphotic zone in the eastern North Pacific
Abstract
The monthly, seasonal and interannual variability of microbial eukaryote assemblages at four
depths (5 m, the deep chlorophyll maximum, 150 m and 500 m) were examined at the San Pedro
Ocean Time-series (SPOT) station in the eastern North Pacific. The depths spanned transitions
in temperature, light, nutrients and oxygen, and included a persistently hypoxic environment at
500 m. Terminal Restriction Fragment Length Polymorphism was used for high throughput
analysis of 237 samples that were collected between September 2000 and December 2010. The
patterns of spatiotemporal variability of microbial eukaryote assemblages indicated the presence
of distinct communities within and below the euphotic zone at the SPOT station, possibly
reflecting taxa that were specifically adapted for the conditions in the shallow and deep
environments at the sampling site. Month-to-month similarity values were relatively high for
assemblages at all depths (~51-61%), but the assemblage at 5 m was temporally more dynamic
than deeper assemblages. Seasonality was apparent for microbial eukaryote assemblages within
and below the euphotic zone at 5 m and 150 m, respectively, but not at the DCM or 500 m.
Microbial eukaryote assemblages exhibited cyclical patterns in nearly half of all the years and
depths examined in this study, implying an annual resetting of communities at specific depths.
Substantial interannual variability was detected for assemblages at all depths, but especially at 5
m and the DCM, and was more pronounced than monthly or seasonal changes in community
composition.
80
Introduction
Protistan abundances and community composition are governed by a multitude of environmental
processes including prey availability, top-down grazing pressure (e.g., viruses, other microbial
eukaryotes and metazoa), and abiotic factors (e.g., temperature, light, salinity and nutrients)
(Sherr et al. 2007). In turn, protistan assemblages have substantial impacts on ecosystem
processes and function. Improving our understanding of community assembly and reassembly,
and factors that control them, are essential to predicting ecosystem response to a changing
environment (McGradySteed et al. 1997; Naeem and Li 1997; Chapin et al. 2000).
Molecular snapshots of micro-eukaryote assemblages from varying geographic locations,
depths and times have provided a better understanding of the overall diversity, community
structure and distribution of microbial eukaryotes in marine ecosystems (Massana and Pedros-
Alio 2008; Vaulot et al. 2008; Caron et al. 2012). Molecular approaches have also been applied
to characterize short-term temporal dynamics (days to weeks) of protistan assemblages in natural
ecosystems (Vigil et al. 2009) and bottle incubations (Countway et al. 2005; Kim et al. 2011;
Weber et al. 2012). These studies have shown rapid shifts in community composition and
structure in response to changes in environmental conditions. Several studies have also
compared seasonal differences of microbial eukaryote assemblages (Massana et al. 2004; Romari
and Vaulot 2004; Lepere et al. 2006; Medlin et al. 2006; Aguilera et al. 2007; Countway et al.
2010; Nolte et al. 2010; Piwosz and Pernthaler 2010; Edgcomb et al. 2011; Orsi et al. 2011),
although most of these reports spanned relatively short time-scales (≤ 1-2 years). Multi-year
comparisons that capture interannual changes of microbial eukaryotes have focused largely on
bloom-forming conspicuous phytoplankton taxa such as diatoms and dinoflagellates (Venrick
1998; Kim et al. 2009; Hinder et al. 2012). Thus, significant gaps exist in our basic knowledge
81
regarding the natural variability of microbial eukaryote assemblages in the euphotic zone and
especially below the euphotic zone.
Terminal Restriction Fragment Length Polymorphism (T-RFLP) is a common molecular
fingerprinting approach that offers a faster and cheaper alternative to sequencing. Since its
introduction for the study of microbial ecology (Liu et al. 1997), the method has proven to be
highly reproducible and reliable for the characterization of dominant taxa within diverse
assemblages of microbes (Schutte et al. 2008; Rossi et al. 2009; Balzano et al. 2012; Kim et al.
2012).
Here, we characterized the vertical structure as well as the temporal (monthly, seasonal
and interannual) variability of microbial eukaryote assemblages at four depths (5 m, the deep
chlorophyll maximum, 150 m and 500 m) that spanned transitions in temperature, light, nutrients
and oxygen at the San Pedro Ocean Time-series (SPOT) station. Microbial eukaryote
assemblages were characterized using 18S rDNA-based T-RFLP analysis on 237 samples
collected at approximately monthly intervals on cruises between September 2000 and December
2010 from 5 m and the DCM (with the exception of a 3-year period between October 2006 and
February 2008), and between September 2000 and December 2003 from 150 m and 500 m. To
our knowledge, this represents one of the largest molecular-based studies of microbial eukaryote
assemblages in a marine environment. Patterns of spatiotemporal variability revealed distinct
assemblages of microbial eukaryotes in the shallow and deep environments at the SPOT station.
Seasonality was apparent for assemblages within the euphotic zone at 5 m and in the aphotic
zone at 150 m. Microbial eukaryote assemblages followed a cyclical pattern in some years,
indicating an annual resetting of communities at specific depths. Monthly changes in community
composition were modest compared to seasonal changes and especially to interannual
82
differences, which were the largest source of variability for microbial eukaryote assemblages in
this temperate coastal ecosystem.
Methods
Sample collection
Seawater samples were collected on board the R/V Seawatch or R/V Yellowfin from four depths
(5 m, the deep chlorophyll maximum, 150 m and 500 m) at approximately monthly intervals
between September 2000 and December 2010 at the San Pedro Ocean Time-series station
located in the eastern North Pacific (33°33’N, 118°24’W). The depth of the deep chlorophyll
maximum (DCM) was determined by in situ real-time fluorescence during each sample
collection.
Environmental data
In situ measurements of conductivity, temperature and depth (Sea-bird Electronics or SBE
911plus CTD), chlorophyll fluorescence (Wet Labs WETStar fluorometer), and dissolved
oxygen (SBE 13 sensor) were measured during the collection of each sample. Mixed layer
depths were estimated as the depths at which sigma-theta differed from surface (10 m) sigma-
theta by 0.125 kg m
-3
(Levitus 1982).
Chlorophyll a concentrations were measured for samples collected from 5 m and the
DCM, dissolved oxygen and nitrate concentrations were measured for samples collected every
10 m between the surface and 60 m, and every 100 m between 100 m and 500 m (Grasshoff et al.
2007). Depth-integrated (0-40 m) nitrate concentrations were calculated to reflect concentrations
within the mixed layer, and depth-integrated (60-200 m) oxygen concentrations calculated to
reflect oxygen levels near the oxycline.
83
Daily averages of sea surface temperature (SST) from NASA’s GOES Imager (Wu et al.
1999) and near-daily averages of SeaWiFS chlorophyll a concentrations from the GeoEye
Orbview-2 satellite (Hooker and McClain 2000) were obtained from the National Oceanic and
Atmospheric Administration (NOAA) CoastWatch Browser for the West Coast Regional Node
(http://coastwatch.pfeg.noaa.gov/coastwatch/).
Molecular characterization of microbial eukaryote assemblages
A total of 237 samples were analyzed for this study (Table 1), representing one of the largest
molecular-based characterizations of marine microbial eukaryotes. Terminal Restriction
Fragment Length Polymorphism (T-RFLP) using Euk-A (Medlin 1988) and Euk-570R (Weekers
et al. 1994) 18S rDNA primers was used, according to Countway et al. (2005). T-RFLP
fragments were assigned to high-level taxonomic groups based on putative identifications of
fragment lengths obtained from in silico digests of 1,341 partial 18S rRNA gene sequences from
a 5 m and 500 m sample at the SPOT station (Kim et al. 2012).
Multivariate analyses
T-RFLP results were normalized to peak area (Kaplan and Kitts 2004) and square-root
transformed to down-weight the contribution of highly dominant T-RFLP fragments in each
sample. PRIMER (v6) & PERMANOVA+β18 (PRIMER-E Ltd) was used to calculate Bray-
Curtis similarity values, which were used for subsequent non-metric multi-dimensional scaling
(MDS) and CLUSTER (including SIMPROF) analyses (Clarke 1993). The parameters used for
the SIMPROF test were: group average mode; permutations for mean profile: (1000); simulation
permutations: (999); significance level: (0.05). The parameters used for MDS analyses were:
Kruskal stress formula: (1); minimum stress: (0.01). Stress values were calculated for MDS
plots and reflect the level of distortion that results from representing similarity rankings between
84
many samples in 2-dimensional space. Stress values of less than 0.2 generally indicate an
accurate representation of similarity rankings. The RELATE test was used to calculate
Spearman Correlations between ranked similarities among monthly assemblages in each year
with other years from the same depth or against a cyclicity model (Clarke and Gorley 2006).
Analysis of similarity (ANOSIM) is the multivariate analog of ANOVA and was used to test the
null hypothesis of no significant compositional differences between depths. The ANOSIM test
statistic (R) ranges between 0 and 1, where 1 indicates completely different assemblages.
Shannon and Simpson diversity indices were calculated using normalized but not transformed T-
RFLP data. Since the two indices strongly correlated (Pearson correlations ranged between 0.88
and 0.94), only Shannon indices were presented.
Averages of pairwise Bray-Curtis similarity values
Averages of pairwise Bray-Curtis similarity values were used to compare the overall
(dis)similarity of microbial eukaryote assemblages at and between each depth. Samples from
each depth were also grouped by months apart (1-24 months lag), and averages of pairwise Bray-
Curtis similarity values were calculated for each group to examine seasonality. Assemblages
were determined to be seasonal if community similarity values were lowest between assemblages
that were 6- or 18-months apart (approximately opposite seasons) and highest between
assemblages that were 12- and 24-months apart (similar seasons), and resulted in a non-linear
regression of a sinusoidal pattern characteristic of seasonal temperature plots (see Supplementary
Figure S1A). Groups of samples that were more than 24 months apart reduced the power of the
analysis because of the increasingly smaller number of samples with increased lag, and were
excluded from the analyses.
85
Results
Environmental variables
Annual sea surface temperatures at the SPOT station ranged between ~13.5° C and ~21.5° C
(Figures 1A, S1A and S1D). Warmest temperatures occurred in August and September, and
coolest temperatures occurred between December and April. Coolest temperatures at the DCM
were observed between April and July, and warmest temperatures occurred between September
and January (Figure 1B). Temperature ranges were more constrained at 150 m and especially at
500 m, but the temporal variability of temperatures in these deep environments resembled the
patterns observed at the DCM (Figures 1B, 1C and 1D).
Mixed layer depths (MLDs) remained relatively shallow throughout the year, ranging
between approximately 15 and 50 m (Figure 1E). MLDs correlated with sea surface
temperatures (r
2
= 0.65; Supplementary Figures S1D and S1F) and were deepest between
December and February of each year before shoaling in March and April.
A persistent oxycline was present beginning near 50 m, and hypoxic conditions (< 1 mL
O
2
L
-1
) were observed below ~350 m throughout each year. The monthly averages of depth-
integrated (60-200 m) oxygen concentrations correlated with monthly averages of temperatures
at 150 m (r
2
= 0.66; Figures 1C and 1G). The depth of the DCM was generally found above the
MLD during winter months and below the mixed layer between March and November (Figure
1E), and correlated with the nitracline (r
2
= 0.4; Supplementary Figure S1H). Nutrient
concentrations at 5 m were low throughout each year (e.g., <0.3 µM nitrate at 5 m). Depth-
integrated (0-40 m) nitrate concentrations were highest between April and July (Figure 1F and
Supplementary Figure S1G). Highest chlorophyll a concentrations at 5 m and the DCM were
observed during April, although chlorophyll a concentrations were generally low throughout
86
each year (< ~2 µg L
-1
; Figures 1H, S1B and S1C). Interannual variability was observed for all
environmental parameters, reflected by the standard error bars in Figure 1 and detailed in
Supplementary Figure S1 (e.g., anomalously high chlorophyll a concentrations in September
2003, Supplementary Figure S1B).
Vertical distribution of microbial eukaryote assemblages
MDS and CLUSTER analyses of assemblages at four depths between September 2000 and
December 2003 resolved two significantly different communities of microbial eukaryotes—a
shallow community at 5 m and the DCM; and a deep community at 150 m and 500 m (R=0.78,
p=0.001) (Figure 2). Exceptions included assemblages collected on October 27, 2000 from all
four depths and on April 2, 2001 from 5 m (circled samples in Figure 2).
Assemblages of microbial eukaryotes at 5 m could not be differentiated from
assemblages at the DCM (R = 0.08). However, pairwise comparisons of community similarity
values between the two near-surface assemblages varied and were highest between November
and March, with the exception of January (~60-64%, Supplementary Figure S2). Community
similarity values between assemblages at 5 m and the DCM were lowest in September and
October (~49%).
The taxonomic composition of assemblages at 5 m and the DCM were distinct from
assemblages at 150 m and 500 m (Figures 3A-D and S3A-D). Shallow communities were
dominated by fragment lengths representative of arthropod, chlorophyte, ciliate, dinoflagellate,
haptophyte, stramenopile, telonemid, and uncultured alveolate (Group I and Group II) taxa.
Deep communities were largely comprised of cnidarian, euglenozoan, polycystine, and
uncultured alveolate (mostly Group II) taxonomic groups, as well as a smaller fraction of ciliate
and dinoflagellate taxa. OTUs that could be assigned a high-level taxonomic identification
87
accounted for an average of ~75-80% of total abundance in each sample (right panels of Figures
3A-D). Modest compositional differences were apparent in the monthly averages of relative
abundances of taxonomic groups at each depth (Supplementary Figures S3A-D).
Shannon Indices
There were no significant differences in the average number of OTUs detected between depths
(the analysis of 237 samples resulted in an average of 35 OTUs per sample), but a significant
difference (p<0.05) was detected in the averages of Shannon index values between assemblages
at 150 m and the DCM (Figure 4). Shallow assemblages resulted in relatively high Shannon
indices (averages of ~2.9 and ~3.0 for assemblages at 5 m and the DCM, respectively), and
assemblages at 150 m resulted in the lowest average Shannon index (~2.6). Assemblages at 150
m also resulted in the most variable Shannon indices between 2000 and 2003 (Figures 4 and
S4C). Assemblages at the DCM, on the other hand, maintained consistently high Shannon index
values (Figures 4 and S4B). The average Shannon index for assemblages at 500 m was ~2.8
(Figure 4).
Overall similarity of microbial eukaryote assemblages at specific depths
Surface assemblages (5 m and the DCM) had a larger number of highly disparate samples
compared to other depths between 2000 and 2003 (i.e., more samples that were > 1 or 2 standard
deviations away from the mean; Figures 5A-D). The lowest and highest average similarity
values (~43.6% and ~52.3%) were observed for assemblages at 5 m and 500 m, respectively
(Figure 5E). Assemblages at 150 m and the DCM resulted in intermediate averages that were
nearly identical. Additionally, there were a larger number of highly disparate samples collected
from 5 m as well as the DCM between 2000 and 2003 compared to samples collected between
2008 and 2010 (Figures 5A and 5B).
88
Monthly, seasonal and interannual variability of microbial eukaryote assemblages
Community similarity values between assemblages from adjacent months (i.e., 1-month apart or
lagged) were relatively high at all depths, ranging between ~51% and ~61% (Figures 6A-F).
Seasonality was not detected for assemblages at 5 m or the DCM when the first three years
(2000-2003) of samples were analyzed (Figures 6A and 6B), but a pattern of seasonality was
apparent for assemblages below the euphotic zone at 150 m (r
2
=0.61, Figure 6C). Lowest
community similarity values at 150 m were observed between assemblages that were 6 and 18
months apart (approximately opposite seasons), and high community similarity values returned
annually (at 12- and 24-month intervals). Non-linear regression resulted in a sinusoidal curve
that resembled seasonal sea surface temperatures in Supplementary Figure S1A. Seasonality was
not detected for assemblages at 500 m (Figure 6D).
The large number of highly disparate samples at 5 m and the DCM between 2000 and
2003 (Figures 5A and 5B) prompted analyses to test for seasonality in assemblages that were
collected between 2008 and 2010 at 5 m and the DCM, as well as in the entire dataset (2000-
2010) for both depths. Results revealed seasonality for assemblages at 5 m when the samples
that were collected between 2008 and 2010 were analyzed (r
2
=0.56; Figure 6E). A weak pattern
of seasonality was detected in the analysis of the entire dataset of the assemblages at 5 m (2000-
2010), but that analysis did not result in a good fit of a sinusoidal curve (Figure 6F). Seasonality
was not detected for assemblages at the DCM in the analysis of the last three years of the dataset,
nor in the entire dataset (data not shown).
Microbial eukaryote assemblages exhibited cyclical patterns of variability in
approximately half of the depths and years that were examined at the study site (Table 2).
Assemblages at 5 m displayed cyclical patterns in five out of the nine years that were examined,
89
and assemblages at the DCM in three years. The cyclical pattern observed for assemblages at
both 5 m and the DCM in 2003 were illustrated in MDS plots (Figures 7A and 7B, respectively).
Assemblages at 150 m and 500 m displayed cyclicity in one year (2003 and 2001, respectively)
(Figures 7C and 7D).
The ranked community similarity values between assemblages from different months in
some years significantly correlated (p<0.05) with the ranked community similarity values of
monthly assemblages from other years for assemblages at 5 m and the DCM, but not for
assemblages at 150 m or 500 m (Table 3). The strongest relationship (0.806) was observed for
assemblages at 5 m between 2005 and 2006, and MDS analyses showed very similar patterns
between months for the assemblages in these years (Figures 7E and 7F).
Assemblages at 5 m and the DCM between 2000 and 2003 largely clustered by year
(Figures 8A and 8B), with the exception of assemblages in December and January at 5 m (Figure
8A, December and January samples were marked with bars to the right of the CLUSTER
diagrams). Communities at 150 m and 500 m (Figures 8C and 8D) also showed interannual
variability, but the assemblages from each year did not cluster as tightly as observed for near-
surface assemblages. Assemblages at 150 m in December and January between 2000 and 2003
also shared high community similarity values (Figure 8C).
Discussion
The study site
The San Pedro Ocean Time-series (SPOT) station provides a coastal environment for comparison
to long-term open-ocean monitoring sites such as the Hawaiian Ocean Time-Series (Dore et al.
2008) and Bermuda Atlantic Time-Series (Steinberg et al. 2001) stations. The San Pedro basin
has a maximum depth of ~890 m and is bordered by Santa Catalina Island to the west and
90
subsurface sills to the north and south that rise to ~500 m. The station has an oxycline beginning
near 50 m and a persistent hypoxic environment below ~350 m as a result of terrigenous inputs
and restricted circulation in deep waters. Seasonal fluctuations in temperature during this study
were relatively restricted, especially at depth (Figures 1A-D), but seasonal patterns in
temperature and dissolved oxygen concentration were observed below the euphotic zone at 150
m (Figure 1C and 1G). Dissolved oxygen was not seasonally variable at 500 m (data not
shown), but temperatures at that depth displayed modest seasonal fluctuations (~0.5° C; Figure
1D). The water column was stratified throughout each year, punctuated by winter mixing events
to ~50 m (Figure 1E), which entrained nutrients into surface waters that stimulated
phytoplankton in the spring (Figures 1F and 1H). A deep chlorophyll maximum (DCM) was
also present throughout each year but varied in depth, with summer months characterized by a
deeper DCM (Figure 1E).
Distinct assemblages of microbial eukaryotes within and below the euphotic zone
The spatiotemporal variability of microbial eukaryote assemblages in this study showed that the
euphotic zone and deep environments at the SPOT station were comprised of distinct
communities, possibly reflecting taxa that are specifically adapted for conditions in those
environments (Figure 2). The persistent, shallow mixed layer (Figure 1E) maintained a physical
separation between unique shallow (<50 m) and deep taxa throughout the time-series study.
Previous studies have also reported that vertical gradients of environmental factors (e.g., light,
temperature, nutrients, and oxygen) at the SPOT station play an important role in shaping distinct
shallow and deep microbial communities (Countway et al. 2010; Schnetzer et al. 2011).
Additionally, diversity surveys have reported higher community similarity values between
microbial eukaryote assemblages within the euphotic zone at different locations than between
91
surface and deep-sea assemblages at the same location (Countway et al. 2007; Not et al. 2007),
suggesting that depth is more influential in shaping microbial communities than time or
geographic location.
Assemblages in the euphotic zone at the SPOT site were generally dominated by
arthropods, ciliates, chlorophytes, dinoflagellates, stramenopiles, and telonemids, common
taxonomic groups that have been previously characterized using microscopy and molecular
approaches (Moorthi et al. 2006; Countway et al. 2010; Fitzpatrick et al. 2010; Schnetzer et al.
2011; Steele et al. 2011; Kim et al. 2012) (Figures 3A, 3B, S3A and S3B). Deeper assemblages
were largely comprised of polycystines and group II alveolates, also consistent with previous
studies (Figures 3C, 3D, S3C and S3D). Only modest compositional differences were observed
between months over several years (Supplementary Figures S3A-D), although larger differences
would be expected if the communities were examined with finer taxonomic resolution (e.g.,
species-level). Restriction of phototrophic taxa to surface waters would be anticipated, but the
presence of unique taxa in the deep samples throughout this time-series study also indicated the
presence of taxa that were depth-restricted and possibly adapted to conditions found at our
deeper sampling depths.
The seasonal deepening of the mixed layer (Figure 1E) contributed to highest community
similarity values between assemblages at 5 m and the DCM in the months of November,
December, February and March between 2000 and 2003 (Supplementary Figure S2). Following
stratification and stabilization of the water column due to seasonal sea surface warming,
assemblages at 5 m and the DCM became less similar and were most dissimilar when the DCM
was observed below the MLD (Figures 1A, 1E and S2). Countway et al. (2010) also reported
92
higher similarities between assemblages at 5 m and the DCM in winter and spring months
compared to summer or fall assemblages at the SPOT station.
Surface assemblages (5 m and the DCM) displayed relatively high Shannon index values
(Figure 4), reflecting even relative abundances of taxa in surface waters at the SPOT station.
Assemblages at the DCM yielded consistently high Shannon values throughout each year
(Supplementary Figure S4B), presumably as a result of sampling the same biological feature
each month (i.e., the deep chlorophyll maximum) regardless of its depth. On average, the
environmental conditions just below the euphotic zone at 150 m selected for high dominance of
relatively few taxa. Assemblages at 150 m also displayed the most variable Shannon index
values (Figures 4 and S4C). Countway et al. (2010) and Schnetzer et al. (2011) reported some of
the highest diversity and richness estimates for assemblages at 150 m during summer and fall.
Significant differences in Shannon indices between seasons at 150 m were not detected in this
study because of considerable interannual variability at that depth (Supplementary Figure S4C).
It is not clear what environmental parameters influenced the variable Shannon indices at 150 m.
The average Shannon Index for assemblages at 500 m was comparatively high, an unexpected
finding because we anticipated that this chronically hypoxic environment might be highly
selective.
Community composition at 5 m was more variable over time than assemblages at other
depths (Figures 5A-E), presumably due to larger exposure to climatic fluctuations, including the
inherent patchiness of surface waters and the influence of mesoscale turbulence in the region
(Hickey et al. 2003; Dong et al. 2009). Assemblages at 500 m resulted in the highest average
similarity value, possibly due to temporally stable environmental conditions at that depth.
Horizontal advection is restricted by the sill of the basin (located ~500 m below the surface),
93
creating a relatively stable physical and chemical environment (Berelson 1991; Hickey 1991).
Sampling from a consistent biological feature (the DCM; Figure 1E) and at 150 m where
seasonal fluctuations in environmental conditions were dampened (Figures 1C and 1G) most
likely contributed to the intermediate averages obtained for assemblages at the DCM and 150 m
(Figure 5E).
Month-to-month variability and seasonality of microbial eukaryote assemblages within and
below the euphotic zone
Month-to-month changes in the composition of microbial eukaryote assemblages at the SPOT
station were modest compared to other studies that have characterized substantial shifts in
community composition within time-frames of hours to weeks (1-month time lag in Figures 6A-
F) (Countway et al. 2005; Vigil et al. 2009; Kim et al. 2011; Weber et al. 2012). The relatively
high similarity values between the microbial eukaryote assemblages from adjacent months in this
study were not surprising, considering the small seasonal changes in environmental parameters
observed at the station, especially at depth (Figures 1A-H).
Seasonality was detected for assemblages at 5 m in samples collected between 2008 and
2010, but not between 2000 and 2003 (Figures 6A and 6E). These results indicate that
interannual variability can obscure seasonal patterns at this study site. Average similarity values
between assemblages from opposite seasons ranged between ~44.6% and ~48.3% (6 and 18
month interval data points in Figures 6C and 6E), compared to month-to-month similarity values
of ~51-61% (1-month lag in Figures 6A-F). Seasonal successions among groups of conspicuous
phytoplankton in the southern California Bight have been previously characterized, including
diatoms that are abundant in the spring and dinoflagellates that dominate in the summer (Venrick
1998). More recently, molecular evidence revealed distinct communities of microbial
94
eukaryotes at 5 m and the DCM during different seasons in 2001 (Countway et al. 2010).
Seasonality was not detected for assemblages at the DCM in this study (Figure 6B), however,
which was possibly a consequence of sampling from a persistent biological feature in the water
column that varied in depth throughout each year (Figure 1E).
Assemblages of microbial eukaryotes below the euphotic zone at 150 m were also
seasonal (Figure 6C). To our knowledge, this is the first study to demonstrate seasonality for
microbial eukaryote assemblages below the euphotic zone and within an oxycline. The seasonal
pattern in community compositional changes is not surprising, considering the seasonal
variability in temperature and oxygen concentration observed at that depth (Figures 1C and 1G).
Additionally, 150 m is located near the base of the euphotic zone and presumably experiences
seasonal fluxes of dissolved and particulate materials from surface waters. Assemblages at 500
m did not exhibit seasonality (Figure 6D), which may be a result of the physically and
chemically stable environment at that depth.
Assemblages of microbial eukaryotes were cyclical in nearly half of the years and depths
examined in this study (Table 2), implying an annual resetting of communities at specific depths.
An annual resetting of community composition at 5 m and the DCM is mostly likely attributed to
winter mixing and light-limiting conditions, but it is not clear what environmental conditions
contributed to the annual resetting of assemblages in the chronically dark and cold environments
at 150 m and 500 m.
Interannual variability of microbial eukaryote assemblages
There were considerable interannual differences in community composition of microbial
eukaryotes at each depth, especially in surface waters (5 m and the DCM) at the study site
(Figures 8A-D). Assemblages at each depth exhibited an annual resetting of communities in
95
some but not all years (Table 2), and correlations between ranked similarities of monthly
assemblages were found in only a subset of our time-series data (Table 3). Interannual
variability seemed to dominate and even obscure monthly and seasonal changes in community
composition at our study site. Sources of interannual variability in the region include influences
by the colder and more nutrient-rich waters of the California Current as well as the warmer and
relatively nutrient-poor countercurrent when trade winds are weakened in the summer (Hickey et
al. 2003; Dong et al. 2009). The station is also surrounded by complex seafloor topography and
the Channel Islands that influence the circulation patterns in and around the station (Dong and
McWilliams 2007). Local and remote climatic events and wind stress also affect circulation
patterns as well as mixing of the water column in this coastal ecosystem (Hayward 2000). The
SPOT station is also located approximately 10 miles away from one of the largest urban centers
and busiest sea-ports in the world, and can experience substantial terrestrial input following large
storm events (Thunell et al. 1994).
Conclusion
This comprehensive study documented monthly, seasonal and interannual variability of the
community composition within microbial eukaryote assemblages at 5 m, the DCM, 150 m and
500 m in the eastern North Pacific at the SPOT station. Strong selective forces maintained
distinct assemblages of microbial eukaryotes between shallow (<50 m) and deep environments.
In general, month to month changes were small at all depths, but assemblages at 5 m displayed
the most temporal variability overall. Seasonality was apparent for microbial eukaryote
assemblages at 5 m and 150 m. Assemblages at each depth showed cyclical patterns in nearly
half of the years and depths examined in this study, suggesting an annual ‘resetting’ of the
communities at specific depths. Interannual differences were substantial for microbial eukaryote
96
assemblages at all depths, but especially at 5 m and the DCM, and represented a larger source of
variability than monthly or seasonal changes in community composition in this temperate
ecosystem.
97
References
Aguilera A, Zettler E, Gomez F, Amaral-Zettler L, Rodriguez N, Amils R (2007) Distribution
and seasonal variability in the benthic eukaryotic community of Rio Tinto (SW, Spain),
an acidic, high metal extreme environment. Syst Appl Microbiol 30, 531-546.
Balzano S, Marie D, Gourvil P, Vaulot D (2012) Composition of the summer photosynthetic
pico and nanoplankton communities in the Beaufort Sea assessed by T-RFLP and
sequences of the 18S rRNA gene from flow cytometry sorted samples. ISME J 6, 1480-
1498.
Berelson WM (1991) The flushing of 2 deep-sea basins, Southern California borderland. Limnol
Oceanogr 36, 1150-1166.
Caron DA, Countway PD, Jones AC, Kim DY, Schnetzer A (2012) Marine protistan diversity.
Annu Rev Mar Sci 4, 467-493.
Chapin FS, III, Zavaleta ES, Eviner VT, Naylor RL, Vitousek PM, Reynolds HL, Hooper DU,
Lavorel S, Sala OE, Hobbie SE, Mack MC, Diaz S (2000) Consequences of changing
biodiversity. Nature 405, 232-242.
Clarke KR (1993) Nonparametric multivariate analyses of changes in community structure. Aust
J Ecol 18, 117-143.
Clarke KR, Gorley RN (2006) PRIMER v6: User manual/Tutorial. In: PRIMER-E, Plymouth.
Countway PD, Gast RJ, Dennett MR, Savai P, Rose JM, Caron DA (2007) Distinct protistan
assemblages characerize the euphotic zone and deep sea (2500 m) of the western North
Atlantic (Sargasso Sea and Gulf Stream). Environ Microbiol 9, 1219-1232.
98
Countway PD, Gast RJ, Savai P, Caron DA (2005) Protistan diversity estimates based on 18S
rDNA from seawater incubations in the western North Atlantic. J Eukaryot Microbiol 52,
95-106.
Countway PD, Vigil PD, Schnetzer A, Moorthi SD, Caron DA (2010) Seasonal analysis of
protistan community structure and diversity at the USC Microbial Observatory (San
Pedro Channel, North Pacific Ocean). Limnol Oceanogr 55, 2381-2396.
Dong CM, Idica EY, McWilliams JC (2009) Circulation and multiple-scale variability in the
Southern California Bight. Prog Oceanogr 82, 168-190.
Dong CM, McWilliams JC (2007) A numerical study of island wakes in the Southern California
Bight. Cont Shelf Res 27, 1233-1248.
Dore JE, Letelier RM, Church MJ, Lukas R, Karl DM (2008) Summer phytoplankton blooms in
the oligotrophic North Pacific Subtropical Gyre: Historical perspective and recent
observations. Prog Oceanogr 76, 2-38.
Edgcomb V, Orsi W, Bunge J, Jeon S, Christen R, Leslin C, Holder M, Taylor GT, Suarez P,
Varela R, Epstein S (2011) Protistan microbial observatory in the Cariaco Basin,
Caribbean. I. Pyrosequencing vs Sanger insights into species richness. ISME J 5, 1344-
1356.
Fitzpatrick E, Caron DA, Schnetzer A (2010) Development and environmental application of a
genus-specific quantitative PCR approach for Pseudo-nitzschia species. Marine Biology
157, 1161-1169.
Grasshoff K, Kremling K, Ehrhardt M (2007) Methods of Seawater Anlaysis, Third edn. Wiley-
VCH GmbH, Weinhein, Germany.
99
Hayward TL (2000) El Nino 1997-98 in the coastal waters of Southern California: A timeline of
events. Cal Coop Ocean Fish 41, 98-116.
Hickey BM (1991) Variability in 2 deep coastal basins (Santa Monica and San Pedro) off
Southern California. J Geophys Res (Oceans) 96, 16689-16708.
Hickey BM, Dobbins EL, Allen SE (2003) Local and remote forcing of currents and temperature
in the central Southern California Bight. J Geophys Res (Oceans) 108.
Hinder SL, Hays GC, Edwards M, Roberts EC, Walne AW, Gravenor MB (2012) Changes in
marine dinoflagellate and diatom abundance under climate change. Nat Clim Change 2,
271-275.
Hooker SB, McClain CR (2000) The calibration and validation of SeaWiFS data. Prog
Oceanogr 45, 427-465.
Kaplan CW, Kitts CL (2004) Bacterial succession in a petroleum land treatment unit. Appl
Environ Microbiol 70, 1777-1786.
Kim DY, Countway PD, Gast RJ, Caron DA (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.
Kim DY, Countway PD, Yamashito W, Caron DA (2012) A combined sequence-based and
fragment-based characterization of microbial eukaryote assemblages provides taxonomic
context for the Terminal Restriction Fragment Length Polymorphism (T-RFLP) method.
J Microbiol Meth (in press).
Kim HJ, Miller AJ, McGowan J, Carter ML (2009) Coastal phytoplankton blooms in the
Southern California Bight. Prog Oceanogr 82, 137-147.
100
Lepere C, Boucher D, Jardillier L, Domaizon I, Debroas D (2006) Succession and regulation
factors of small eukaryote community. composition in a lacustrine ecosystem (Lake
pavin). Appl Environ Microbiol 72, 2971-2981.
Levitus S (1982) Climatological Atlas of the World Ocean, p. 191. NOAA.
Liu WT, Marsh TL, Cheng H, Forney LJ (1997) Characterization of microbial diversity by
determining terminal restriction fragment length polymorphisms of genes encoding 16S
rRNA. Appl Environ Microbiol 63, 4516-4522.
Massana R, Balague V, Guillou L, Pedros-Alio C (2004) Picoeukaryotic diversity in an
oligotrophic coastal site studied by molecular and culturing approaches. FEMS Microbiol
Ecol 50, 231-243.
Massana R, Pedros-Alio C (2008) Unveiling new microbial eukaryotes in the surface ocean.
Curr Opin Microbiol 11, 213-218.
McGradySteed J, Harris PM, Morin PJ (1997) Biodiversity regulates ecosystem predictability.
Nature 390, 162-165.
Medlin LK, Elwood, H. J., Stickel, S., Sogin, M. L. (1988) The characterization of enzymatically
amplified eukaryotic 16S-like rRNA-coding regions. Genetica 71, 491-499.
Medlin LK, Metfies K, Mehl H, Wiltshire K, Valentin K (2006) Picoeukaryotic plankton
diversity at the Helgoland time series site as assessed by three molecular methods.
Microb Ecol 52, 53-71.
Moorthi SD, Countway PD, Stauffer BA, Caron DA (2006) Use of quantitative real-time PCR to
investigate the dynamics of the red tide dinoflagellate Lingulodinium polyedrum. Microb
Ecol 52, 136-150.
Naeem S, Li SB (1997) Biodiversity enhances ecosystem reliability. Nature 390, 507-509.
101
Nolte V, Pandey RV, Jost S, Medinger R, Ottenwalder B, Boenigk J, Schlotterer C (2010)
Contrasting seasonal niche separation between rare and abundant taxa conceals the extent
of protist diversity. Mol Ecol 19, 2908-2915.
Not F, Gausling R, Azam F, Heidelberg JF, Worden AZ (2007) Vertical distribution of
picoeukaryotic diversity in the Sargasso Sea. Environ Microbiol 9, 1233-1252.
Orsi W, Edgcomb V, Jeon S, Leslin C, Bunge J, Taylor GT, Varela R, Epstein S (2011) Protistan
microbial observatory in the Cariaco Basin, Caribbean. II. Habitat specialization. ISME J
5, 1357-1373.
Piwosz K, Pernthaler J (2010) Seasonal population dynamics and trophic role of planktonic
nanoflagellates in coastal surface waters of the Southern Baltic Sea. Environ Microbiol
12, 364-377.
Romari K, Vaulot D (2004) Composition and temporal variability of picoeukaryote communities
at a coastal site of the English Channel from 18S rDNA sequences. Limnol Oceanogr 49,
784-798.
Rossi P, Gillet F, Rohrbach E, Diaby N, Holliger C (2009) Statistical assessment of variability of
Terminal Restriction Fragment Length Polymorphism analysis applied to complex
microbial communities. Appl Environ Microbiol 75, 7268-7270.
Schnetzer A, Moorthi SD, Countway PD, Gast RJ, Gilg IC, Caron DA (2011) Depth matters:
Microbial eukaryote diversity and community structure in the eastern North Pacific
revealed through environmental gene libraries. Deep Sea Res (I Oceanogr Res Pap) 58,
16-26.
Schutte UME, Abdo Z, Bent SJ, Shyu C, Williams CJ, Pierson JD, Forney LJ (2008) Advances
in the use of terminal restriction fragment length polymorphism (T-RFLP) analysis of
102
16S rRNA genes to characterize microbial communities. Appl Microbiol Biot 80, 365-
380.
Sherr BF, Sherr EB, Caron DA, Vaulot D, Worden AZ (2007) Oceanic protists. Oceanography
20, 130-134.
Steele JA, Countway PD, Xia L, Vigil PD, Beman JM, Kim DY, Chow CET, Sachdeva R, Jones
AC, Schwalbach MS, Rose JM, Hewson I, Patel A, Sun FZ, Caron DA, Fuhrman JA
(2011) Marine bacterial, archaeal and protistan association networks reveal ecological
linkages. ISME J 5, 1414-1425.
Steinberg DK, Carlson CA, Bates NR, Johnson RJ, Michaels AF, Knap AH (2001) Overview of
the US JGOFS Bermuda Atlantic Time-series Study (BATS): a decade-scale look at
ocean biology and biogeochemistry. Deep Sea Res (II Top Stud Oceanogr) 48, 1405-
1447.
Thunell RC, Pilskaln CH, Tappa E, Sautter LR (1994) Temporal variability in sediment fluxes in
the San-Pedro Basin, Southern California Bight. Cont Shelf Res 14, 333-352.
Vaulot D, Eikrem W, Viprey M, Moreau H (2008) The diversity of small eukaryotic
phytoplankton (<= 3 mu m) in marine ecosystems. FEMS Microbiol Rev 32, 795-820.
Venrick EL (1998) Spring in the California current: the distribution of phytoplankton species,
April 1993 and April 1995. Mar Ecol Prog Ser 167, 73-88.
Vigil P, Countway PD, Rose J, Lonsdale DJ, Gobler CJ, Caron DA (2009) Rapid shifts in
dominant taxa among microbial eukaryotes in estuarine ecosystems. Aquat Microb Ecol
54, 83-100.
103
Weber F, Del Campo J, Wylezich C, Massana R, Jurgens K (2012) Unveiling trophic functions
of uncultured protist taxa by incubation experiments in the brackish Baltic Sea. PLoS One
7, e41970.
Weekers PHH, Gast RJ, Fuerst PA, Byers TJ (1994) Sequence variations in small-subunit
ribosomal-RNAs of Hartmannella vermiformis and their phylogenetic implications. Mol
Biol Evol 11, 684-690.
Wu XQ, Menzel WP, Wade GS (1999) Estimation of sea surface temperatures using GOES-8/9
radiance measurements. B Am Meteorol Soc 80, 1127-1138.
104
Table 1 The total number of samples collected in each month from 4 depths (5 m, the depth of
the deep chlorophyll maximum (DCM), 150 m and 500 m) at the USC Microbial Observatory.
Samples were collected from 5 m and the DCM between September 2000 and December 2010,
with the exception of a 3-year period between October 2006 and February 2008. Samples were
collected from 150 m and 500 m between September 2000 and December 2003.
Depth Month Total
Jan Feb Mar Apr May Jun Jul Aug Sept Oct Nov Dec
5 m 6 6 8 7 8 8 9 9 9 8 5 7 90
DCM 6 6 8 7 8 7 9 9 7 8 5 6 86
150 m 3 2 3 2 4 2 3 3 4 4 1 3 34
500 m 3 2 2 2 3 1 3 2 3 3 1 2 27
All depths
combined
18 16 21 18 23 18 24 23 23 23 12 18 237
105
Table 2 Significant Spearman Correlations between the ranked similarities of monthly
assemblages in each year and a cyclicity model. The cyclicity model assumes a relatively
equidistant distribution of assemblages around a center. Spearman Correlations are reported
with p-values in brackets. Dashes represent results that were not significant, and the grey shaded
area represents no data. Years and depths that were depicted in MDS plots in Figures 7A-D are
marked with asterisks.
Year Depth
5 m DCM 150 m 500 m
2001 0.232 (0.001) - - 0.207 (0.018)*
2002 0.3718 (0.008) - - -
2003 0.223 (0.017)* 0.189 (0.025)* 0.356 (0.012)*
2004 - 0.232 (0.024)
2005 - -
2006 - 0.443 (0.017)
2008 - -
2009 0.219 (0.035) -
2010 0.320 (0.001) -
\
106
Table 3 Significant Spearman Correlations between ranked similarity values of monthly assemblages
between different years at 5 m, the DCM, 150 m and 500 m (there were no significant correlations for
assemblages at 150 m and 500 m). The table includes depth, years compared, Spearman Correlations and
p-values in brackets. Assemblages at 5 m in 2005 and 2006 (marked with asterisk) were depicted in MDS
plots in Figures 7E and 7F.
Depth
Years
compared
RELATE Statistic (p-vlaue)
5 m 2002, 2006 0.368 (.046)
5 m 2003, 2006 0.456 (.020)
5 m 2005, 2006* 0.806 (.025)
5 m 2002, 2009 0.349 (.047)
DCM 2003, 2004 0.598 (.014)
DCM 2003, 2010 0.461 (.020)
107
Figure Legends
Figure 1 Monthly averages of physical and chemical environmental parameters measured from
multiple depths at the USC Microbial Observatory between September 2000 and December
2003. In situ temperature at 5 m (A), the deep chlorophyll maximum (DCM; B), 150 m (C) and
500 m (D); the depths of the DCM and mixed layer (E); depth-integrated (0-40 m) nitrate
concentrations (F); depth-integrated (60-200 m) oxygen concentrations (G); and chlorophyll a
concentrations at 5 m and the DCM (H). The error bars represent the standard error of the
monthly means. Details of environmental variables are available in Supplementary Figure S1.
Figure 2 Multi-dimensional scaling (MDS) plot of T-RFLP results from seawater samples
collected approximately every month between September 2000 and December 2003 from 4
depths (5 m, DCM, 150 m, and 500 m) at the USC Microbial Observatory. The MDS plot is
based on Bray-Curtis similarity values of square-root transformed T-RFLP data. The stress
value is reported on the top right corner of the plot. Highly disparate samples that were observed
on 10/27/2000 at all four depths and on 4/2/2001 at 5 m are circled.
Figure 3 Relative abundances of high-level taxonomic groups based on T-RFLP profiles for
approximately monthly samples collected from 5 m (A) and the DCM (B) between 2000 and
2010 (with the exception of months between October 2006 and February 2008); and from 150 m
(C) and 500 m (D) between 2000 and 2003 at the USC Microbial Observatory. Gaps represent
missing data. The averages of relative abundances for each high-level taxonomic group are
presented on the right panel for each depth. Taxonomic identifications for the fragments were
108
assigned using a database of 1,341 18S rRNA gene sequences with assigned taxonomies that
were subjected to in silico restriction digestion to predict the length of terminally labeled
fragments at the study site (obtained from Kim et al., in prep.). Monthly averages of the relative
abundances of high-level taxonomic group are summarized for each depth in Supplementary
Figure S3.
Figure 4 Average Shannon diversity indices for assemblages collected between September 2000
and December 2003 at the USC Microbial Observatory from 5 m, the DCM, 150 m and 500 m.
The bars represent the standard errors of the means. Only dates that had samples from all 4
depths were included in the analysis (n=26). Significant differences are marked by asterisks
(ANOVA). Figure S4 details the Shannon index for each sample from each depth.
Figure 5 The averages of pairwise Bray-Curtis similarity values (plotted with s.e. bars) for
samples collected from 5 m (A), the DCM (B), 150 m (C) and 500 m (D) at the USC Microbial
Observatory. Samples collected between 2000 and 2010 from 5 m and the DCM, and samples
collected between 2000 and 2003 from 150 m and 500 m, were included in the analysis. Sample
sizes (n) are noted for each dataset. The solid grey lines represent the means of all pairwise
averages for each depth, and the dashed grey lines represent 1 and 2 standard deviations. The y-
axes reflect Bray-Curtis similarity values (%) and are standardized to facilitate comparison. The
averages of all pairwise comparisons for samples collected between September 2000 and
December 2003 from each depth are shown in panel E (and only dates that included samples
from each depth were included, n=26). Significant differences in the averages of pairwise
109
comparisons were detected between all depths (p<0.05, ANOVA) except the comparison
between the DCM and 150 m.
Figure 6 The averages of pairwise Bray-Curtis similarity values for all pairs of assemblages that
occurred 1 to 24 months apart. Results are shown for samples collected between September
2000 and December 2003 from 5 m (A), the DCM (B), 150 m (C) and 500 m (D); for samples
collected between January 2008 and December 2010 from 5 m (E); and for samples collected
between September 2000 and December 2010 from 5 m (F). Sample sizes (n) are noted for each
panel. The x-axes represent the number of months lagged and the dotted lines mark
approximately ‘opposite’ (6 and 18 months) or ‘similar’ (12 and 24 months) seasons. The y-axes
represent average pairwise Bray-Curtis similarity values (%) and display the same range to
facilitate comparison. Non-linear regression resulted in a good fit of a sinusoidal pattern only for
assemblages at 150 m between 2000 and 2003 (r
2
= 0.61) (C) and for assemblages at 5 m
between 2008 and 2010 (r
2
= 0.56) (E).
Figure 7 Two-dimensional MDS plots based on Bray-Curtis similarity values for assemblages of
microbial eukaryotes that were collected approximately every month in 2003 from 5 m (A), the
DCM (B), 150 m (C); in 2001 from 500 m (D); and in 2005 (E) and 2006 (F) from 5 m.
Trajectories of relative changes in community composition from month to month are plotted on
each MDS plot, and stress values are reported in the top right corner of each plot. Dashed arrows
represent a theoretical resetting of assemblages for years and depths that tested significantly
positive for cyclicity by the RELATE test (A-D, also see Table 2). Ranked similarities between
110
monthly assemblages in 2005 and 2006 at 5 m (E and F) were significantly correlated (also see
Table 3).
Figure 8 CLUSTER results for assemblages collected approximately every month between
September 2000 and December 2003 from 5 m (A), the DCM (B), 150 m (C), and 500 m (D),
based on Bray-Curtis similarity values on square-root transformed T-RFLP data. Bray-Curtis
similarity values (%) are on the x-axes and normalized. Solid lines denote differences that were
not significant (p > 0.05), and dotted lines denote significant differences (determined by
SIMPROF). December and January samples are marked by bars to the right of each CLUSTER
diagram.
111
Figure 1
112
Figure 2
113
Figure 3
114
Figure 4
115
Figure 5
116
Figure 6
117
Figure 7
118
Figure 8
119
Supplemental Information
Supplemental Figure S1 Physical and chemical parameters measured between September 2000
and December 2003 at or near the USC Microbial Observatory time-series station to supplement
the monthly averages in Figure 1. The solid vertical lines across all plots indicate January of
each year. Daily averages of sea surface temperature (A) and near-daily averages of SeaWiFS
chlorophyll a (B) data were obtained from NOAA’s CoastWatch database
(http://coastwatch.pfeg.noaa.gov/coastwatch/). Chlorophyll a concentrations were measured
from discrete water samples collected from 5 m (C). In situ temperatures at 5 m and 150 m (D)
and mixed layer depths (F) were determined by continuous CTD measurements. Depth-
integrated dissolved oxygen concentrations were determined by the Winkler titration method (E).
Depth-integrated (0-40 m) nitrate concentrations were calculated from nitrate concentration
measurements for samples collected from 10, 20, 30 and 40 m (G). The depth of the deep
chlorophyll maximum (DCM) was determined by in situ fluorescence and are plotted with
approximate nitracline depths (H). Gaps represent missing data.
Supplemental Figure S2 The averages of pairwise Bray-Curtis similarity values between
assemblages at 5 m and the DCM for months between 2000 and 2010 (plotted with standard
error bars).
Supplemental Figure S3 Monthly averages of relative abundances of high-level taxonomic
groups assigned to T-RFLP fragments obtained for samples collected between 2000 and 2010
from 5 m (A) and the DCM (B), and between 2000 and 2003 from 150 m (C) and 500 m (D) at
the San Pedro Ocean Time-series station. High-level taxonomic identifications were assigned
120
using the taxonomies assigned to 1,341 18S rRNA gene sequences that were digested in silico to
predict fragment size (detailed in Kim et al., in press).
Supplemental Figure S4 Shannon diversity indices for near monthly samples collected between
September 2000 and December 2003 from 5 m (A), the DCM (B), 150 m (C), and 500 m (D).
Solid grey lines denote the means of Shannon indices obtained for assemblages at each depth
(also shown in Figure 3), and dashed grey lines represent 1 and 2 standard deviations. The y-
axes are normalized to facilitate comparison.
121
Supplemental Figure S1
122
Supplemental Figure S2
123
Supplemental Figure S3
124
Supplemental Figure S4
125
Chapter 4
Association networks reveal guilds of microbial eukaryotes with similar temporal responses
in the eastern North Pacific
Abstract
Interactions between microorganisms shape microbial community composition in marine pelagic
environments, but our understanding of relationships between taxa within natural assemblages of
microbial eukaryotes remains limited. We characterized the temporal dynamics of microbial
eukaryote taxa using 18S rDNA based Terminal Restriction Fragment Length Polymorphism (T-
RFLP) for samples collected at approximately monthly intervals between September 2000 and
December 2003 from four depths (5 m, the deep chlorophyll maximum, 150 m and 500 m) at the
San Pedro Ocean Time-series station. Network analysis based on highly significant Spearman
correlations identified many groups of taxa that increased and decreased in relative abundance in
unison at specific depths over the ~3 years examined in this study. The number and strength of
correlations between taxa varied by depth, and were greatest for taxa at 150 m and lowest for
taxa at the DCM relative to other depths. Our results also showed conserved and divergent
associations for taxa found at multiple depths. Overall, our results revealed guilds of microbial
eukaryotes that displayed similar temporal responses to changing environmental conditions at
specific depths at our study site.
Introduction
Microbial eukaryotes play critical functional roles in marine ecosystems (Sherr et al. 2007;
Caron et al. 2012), and exist in complex consortia with many other microbial species in the
environment (Massana and Pedros-Alio 2008; Vaulot et al. 2008). Molecular-based
126
characterizations of microbial eukaryote communities have indicated that assemblages can be
highly responsive to even minor changes in environmental conditions, undergoing substantial
compositional changes on time scales of hours to days (Countway et al. 2005; Weber et al.
2012). On longer time-scales, studies have also demonstrated seasonality as well as non-random
and repeating patterns of temporal variability within natural assemblages of microbial eukaryotes
(Medlin et al. 2006; Aguilera et al. 2007; Countway et al. 2010; Piwosz and Pernthaler 2010;
Orsi et al. 2011; Kim et al. 2012b).
Hydrographic (abiotic physical and chemical) factors are key attributes shaping microbial
assemblages in marine environments (Fuhrman et al. 2006; Orsi et al. 2011), but interactions
between species also play a fundamental role in influencing community composition (Bertness
and Callaway 1994). Many types of relationships have been characterized between protistan
taxa, including predator-prey relationships, competition, and symbiosis (mutualism,
commensalism, parasitism). Characterizing relationships between microbial eukaryote species
largely relies on culture-based laboratory manipulations and are often limited to the examination
of a few to several species at a time. The role of interactions between species or functional
groups has long been studied for macro-organisms (Schoener 1983; Sih et al. 1985), but few
studies have attempted to characterize relationships between microbial species or trophic levels
within natural assemblages due to the challenges associated with observing and quantifying the
diversity of these taxa (Weber et al. 2012). One recent study at the SPOT station investigated
possible ecological connections between microbial organisms from all three domains at the depth
of the deep chlorophyll maximum (DCM) using network-based Local Similarity analysis (Steele
et al. 2011). In that study, it was shown that correlations between microbes were more common
than correlations between microbes and abiotic environmental parameters. That study also
127
indicated that microbial eukaryotes were more connected to other microbial eukaryotes than to
bacteria or archaea.
Network analysis is becoming more recognized as an effective tool to study complex
interactions between many variables and has been used to study metabolic networks within E.
coli (Wagner and Fell 2001), gene expression within a community of soil bacteria (Zhou et al.
2010), food web connections (Williams and Martinez 2000), possible ecological associations
between bacterial taxa (Deng et al. 2012), and even connections in the world wide web (for a
review of analyses of complex networks, see Albert and Barabasi, 2002). In this study, we used
network analysis based on Spearman correlations to identify highly correlated temporal patterns
of microbial eukaryote taxa at four depths (5 m, the deep chlorophyll maximum, 150 m and 500
m) in the eastern North Pacific for samples collected between September 2000 and December
2003. The strength and number of correlations between taxa varied by depth and some taxa were
associated with disparate guilds of microbial eukaryotes at different depths. In total, there were
246 significant positive correlations (p<0.001, q<0.05) between taxa at specific depths, showing
guilds of microbial eukaryotes that increased or decreased in relative abundance as a single unit
during the ~3 years of this study.
Methods
Sample collection
Seawater samples were collected on board the R/V Seawatch at approximately monthly intervals
between September 2000 and December 2003 from four depths (5 m, the deep chlorophyll
maximum or the DCM, 150 m, and 500 m). Samples were collected at the San Pedro Ocean
Time-series (SPOT) station (33°33’N, 118°24’W), as a part of the USC Microbial Observatory.
The depth of the DCM was determined by measurements of in situ fluorescence during the
128
collection of each sample and ranged between 15 and 50 m. Seawater samples were collected in
10 L Niskin bottles attached to a CTD sampling rosette and were pre-screened using gravity
filtration through 80 µm mesh net into 20 L acid-washed polycarbonate carboys. The filtrate
was filtered on to GF/F filters on board the cruise, added to 15 mL falcon tubes containing 2 mL
of 2x Lysis buffer (100 mM Tris (pH 8), 40 mM EDTA (pH 8), 100 mM NaCl, 1% SDS) and
flash frozen.
Molecular characterization of microbial eukaryote taxa
Nucleic acids were extracted and purified using a combination of mechanical and chemical
protocols, according to Countway et al. (2005). Terminal Restriction Fragment Length
Polymorphism (T-RFLP) was used to characterize the dominant taxa within natural assemblages
of microbial eukaryotes using Euk-A (labeled with a D4 fluorochrome) (Medlin et al, 1988) and
Euk-570R (unlabeled) (Elwood et al., 1985) PCR primers followed by HaeIII digestion, also
according to Countway et al. (2005). T-RFLP fragments were identified on a Beckman
CEQ8000, a capillary gel electrophoresis platform. Raw T-RFLP data were analyzed using a
peak area threshold of 0.5% and 1 bp-wide bins, and fragment peaks were normalized to total
peak area for each sample (Kaplan and Kitts 2004). Taxonomic identifications were assigned to
fragments based on putative identifications of fragment lengths obtained from in silico digests of
1,341 partial 18S rRNA gene sequences from a 5 m and 500 m sample collected at the SPOT
station (Kim et al. 2012a).
The T-RFLP results in this study were presented as part of a previous study that
characterized the monthly, seasonal and interannual variability patterns of microbial eukaryote
assemblages at the SPOT station (Kim et al. 2012b). This study analyzed the temporal
variability of individual taxa that contributed to the overarching temporal variability of microbial
129
eukaryote assemblages described in that study, and used network analyses based on global
Spearman correlations to identify possible guilds of temporally associated microbial eukaryote
taxa (see next section).
Statistics and network visualization of temporally correlated microbial eukaryote taxa
Global Spearman correlation analysis was used to detect associations between temporal
dynamics of microbial eukaryote taxa at and between specific depths at the study site. Only taxa
that were detected at least 3 times during the ~3 years were included in the anlaysis, which
included 97 OTUs from 5 m, 88 OTUs from the DCM, 84 OTUs from 150 m, and 71 fragments
from 500 m (approximately 52-57% of the total taxa that were detected at each depth during the
study) (Table 1). Global Spearman correlations and associated p-values and q-values (or false
discovery rates) (Storey 2002) were computed using a python script that can be found at
http://meta.usc.edu/softs/lsa/ (Xia et al. 2011). A maximum false discovery rate of 0.05 and a
corresponding maximum p-value of 0.001 were used to determine highly significant correlations
within the dataset. Only significant positive correlations were considered for this study in order
to identify groups of microbial eukaryote taxa that increased and decreased in relative abundance
as a single unit over the full ~3 years.
Significant positive correlations were visualized using Cystoscape v.2.8.2 (Shannon et al.
2003). Microbial eukaryote taxa were depicted as squares and positive correlations were
reflected by black lines. The sizes of the squares were manipulated to reflect average relative
abundances of taxa. Microbial eukaryote taxa were separated by depth using the Group
Attributes layout function in Cytoscape. Network Analyzer (Assenov et al. 2008) in Cytoscape
was used to calculate statistical properties of the networks for each depth and all depths
combined. Statistics included the clustering coefficient, which quantified the inherent tendency
130
for the networks to cluster into small groups (and ranged between 0 and 1, where 1 indicated a
highly interconnected network) (Watts and Strogatz 1998) and shortest average path length,
which quantified the average number of paths between any two taxa (e.g., a value of 2
represented an average of 2 links of separation between taxa) (Albert and Barabasi 2002). The
Random Networks plugin in Cytoscape was used to compare the clustering coefficients and
shortest average path lengths obtained for the networks from our dataset with the structure of
randomly generated networks using the Erdos-Renyi Model and an undirected approach (Albert
and Barabasi 2002). The log response ratios of clustering coefficients and shortest path lengths
were calculated for networks of taxa from specific depths and all depths combined to facilitate
the comparison of our results to other network-based studies (Gurevitch and Hedges 1999).
Results
Temporally correlated microbial eukaryote taxa at and between specific depths
A total of 400 significant (p<0.001, q<0.05) positive and negative global Spearman correlations
were detected between the temporal dynamics of microbial eukaryote taxa at and between the
four depths examined in this study for samples collected between September 2000 and December
2003 (Figure S1A). In total, 57,559 pairwise correlations were computed (Figure S1B). The
distribution of corresponding p-values and q-values for all pairwise comparisons were skewed to
zero and one, respectively, indicating that our results reflected meaningful associations (Figures
S1C and S1D) (Ruan et al. 2006). The 400 significant Spearman correlation values (p<0.001,
q<0.05) ranged between -0.68 and 0.90 and were all greater than |0.54|. Only positive
correlations (which comprised ~95% or 381 of the total significant correlations detected in this
study) were included for network analyses in order to identify microbial eukaryote taxa that
increased or decreased in relative abundance as a single unit over the ~3 years. The networks of
131
positive correlations included 75, 63, 71 and 53 taxa from 5 m, the DCM, 150 m and 500 m,
respectively (Table 1 and Figure 1).
Of the 381 positive correlations (p<0.001; q<0.05) detected in this study, 246 were
detected between taxa contained within the same depth (Table 1, bolded). Network results for
taxa at each depth were examined separately and showed that all depth-specific networks
resulted in similar topologies. Each depth-specific network consisted of multiple sub-clusters or
guilds of temporally correlated taxa and most of the sub-clusters were not connected (Figure 2,
150 m only). The network of taxa at 150 m was used as a representative example because this
network consisted of the most number of sub-clusters compared to the networks of taxa from
other depths. There were 84 significant positive correlations (p<0.001, q<0.05) detected
between the temporal variability of taxa at 150 m (Figure 2, Table 1). Sub-clusters were
comprised of as few as 2 to over 10 taxa and consisted of linear associations (one link per taxon)
as well as more complex and interconnected sub-clusters. Although most sub-clusters were
disconnected, a small fraction of guilds were joined together by one or two taxa. Finally, sub-
clusters included many taxa from the same high-level taxonomic group (Figure 2).
The number and strength of positive correlations between microbial eukaryote taxa
contained within the same depth were greatest for taxa at 150 m and lowest for taxa at the DCM
(Tables 1 and 2). Approximately 60% of the correlations detected for taxa at 150 m were with
other taxa from the same depth, while only ~32% of the correlations detected for taxa at the
DCM were with other DCM taxa (Table 1). Nearly half of the correlations detected for taxa at 5
m and 500 m were with taxa from the same respective depth (46% and 49%, respectively).
The network analyses of taxa at 5 m and the DCM resulted in the lowest clustering
coefficients (both 0.15), and the clustering coefficient for taxa at 150 m resulted in the highest
132
value (0.35) (Table 2). The clustering coefficient for assemblages at 500 m resulted in an
intermediate value of 0.21. The log response ratios for the networks of taxa at each depth as well
as the network of OTUs between all depths ranged between 0.92 and 1.76. The average shortest
path length ranged between 2.17 and 3.78 for each depth, but was considerably higher for the
network of taxa from all depths combined (7.34). The clustering coefficients and shortest path
lengths that were computed for taxa at each depth and all depths combined were significantly
different from the values obtained for the randomly generated networks.
In addition to temporally correlated taxa contained within a single depth, there were 135
correlations between temporal patterns of taxa from different water column depths (Table 1 and
Figure 1). Nearly half of the 76 positive correlations that were detected between temporal
dynamics of taxa at 5 m with taxa from other depths were with taxa at the DCM (Table 1).
Similarly, approximately 44% of the positive correlations between taxa at the DCM with taxa
from other depths were with taxa from 5 m. Nearly half of the 55 positive correlations detected
between the temporal patterns of taxa at 500 m with taxa from depths other than 500 m were
with taxa at the DCM.
Approximately 29.2% of fragments that were detected at least 3 times during the ~3 years
of this study were depth-specific (Figure S2), but the majority of taxa (i.e., fragments) were
detected at more than one depth. Approximately 21.8% of the fragments were detected at all
four depths, ~15.6% were detected at 5 m and the DCM only, and ~11.6% were detected at 150
m and 500 m only (Figure S2). Approximately 9.6% of taxa were detected in only the top three
or bottom three depths that were examined in this study, and the remaining ~12.2% of total
OTUs were detected at various combinations of depths. Despite the detection of some taxa (i.e.,
fragment lengths) at more than one depth, only 3 of the 135 significant positive correlations
133
detected between temporal patterns of taxa from different depths consisted of pairs of the same
taxa or fragment length.
Conserved and varying guilds of microbial eukaryote taxa at different depths
Taxa that were detected at multiple water column depths at our study site revealed conserved
guilds of microbial eukaryotes at different depths. For example, the following 3 taxa were
detected at both 150 m and 500 m: Polycyst_333, Polycyst_269 and Euk_189, and were
associated in similar sub-clusters at each depth (Figure 3A and 3B). The global Spearman
correlation values between the temporal patterns of those taxa at the two depths ranged between
0.60 and 0.79, and time-series plots of these taxa over the ~3 years showed similar increases and
decreases in relative abundance over time between the connected taxa at each depth. Similar
examples of conserved guilds were observed between taxa that were found at both 5 m and the
DCM (data not shown).
There were also microbial eukaryote taxa that were detected at more than one depth, but
associated with different guilds of microbial eukaryotes at different depths (Figure 4). For
example, the changes in relative abundance of the fragment identified as Ostreococcus at 5 m
during the ~3 years strongly correlated with the temporal dynamics of a dinoflagellate
(Spearman correlation of 0.65), but Ostreococcus detected at the DCM temporally correlated
with a different taxon (an unknown eukaryote) (Spearman correlation of 0.62) during the ~3
years of our study (Figure 4A and B). The numerical responses of the fragment identified as
Ostreococcus at 5 m not significantly correlated with the temporal dynamics of Ostreococcus at
the DCM (Figure 4B), nor were either temporally correlated with other taxa within the same
respective depths. Ostreococcus at the DCM, however, was temporally correlated with a
dinoflagellate (Dino_280) at 5 m (not shown in Figure 4A). Similar examples of divergent
134
guilds for taxa that were detected at both 150 m and 500 m were also apparent (Figure 4C).
Figure 4C shows disparate sub-clusters or guilds of temporally correlated microbial eukaryote
taxa at 150 m and 500 m that included one taxon in common—Polycyst_228 (circled in Figure
4C).
Discussion
Network analysis for studying complex interactions
Network-based analyses are powerful for the characterization of interactions between many
variables within complex ecological or biological frameworks. Here, we used network analyses
based on global positive Spearman correlations to identify temporally correlated microbial
eukaryotes at our study site (Figure 1). A subset of our data (DCM only) was first presented in
Steele et al. (2011), but that analysis was based on localized temporal correlations with and
without time-lags and included bacterial and archaeal taxa as well. In both cases, network
analysis provided the ability to quantify and visualize complex associations within natural
assemblages of microbial organisms at the SPOT station.
Properties of the networks resulting from the analysis of temporal dynamics of microbial
eukaryotes at specific depths in this study were similar to properties observed in other ecological
studies, including Steele et al. (2011)’s analysis of bacterial, archaeal and protistan associations
and two network-based analyses of well characterized food webs in the Ythan estuary and
Silwood Park (Sole and Montoya 2001; Camacho et al. 2002; Montoya et al. 2006). Ecological
networks tend to exhibit similar patterns of organization because they are constrained by similar
processes of species interactions such as predation, competition and mutualism (Williams and
Martinez 2000; Strogatz 2001). Statistics such as the clustering coefficient and average shortest
path length allow the comparison of results between different network-based studies (Table 2),
135
and log response ratios of test statistics are used for comparative purposes in order to account for
differences in sample size and variance between studies. Clustering coefficients quantify the
inherent tendency for variables within a network to cluster together. Many ecological studies
have been described as having ‘small world’ properties (i.e., a high tendency to cluster together
than by random chance) (Albert and Barabasi 2002). The log response ratio of the cluster
coefficient reported in Steele et al. (2011)’s analysis of bacteria, archaea and protists at the DCM
using local similarity analysis was 1.71 (Steele et al. 2011). Network analyses of food webs in
the Ythan estuary and Silwood Park resulted in clustering coefficient log response ratios of 1.3
and 1.61, respectively (Sole and Montoya 2001; Camacho et al. 2002; Montoya et al. 2006).
Microbial eukaryote taxa at 5 m and the DCM resulted in slightly lower clustering coefficients
relative to these studies, but microbial eukaryotes at 150 m and 500 m resulted in intermediate to
high values compared to them (Table 2). Not surprisingly, the clustering coefficients from these
ecological studies were lower than the clustering coefficient obtained for in a study of the
metabolic network in E. coli (Wagner and Fell 2001). Average shortest path lengths quantify the
shortest average of links between any two taxa. In theory, no species within an ecosystem is
completely independent from other species (Cohen et al. 1990), and most species in food webs
are theorized to be two to three links from each other (Williams et al. 2002). Average shortest
path lengths for each depth in this study and the other ecological studies mentioned ranged
between 2 and 4.
Guilds of microbial eukaryotes
Our results revealed guilds of tightly coupled microbial eukaryotes that increased and decreased
in relative abundance as a single unit over ~3 years at 5 m, the DCM, 150 m and 500 m at our
study site (Figures 1-4). Considering the substantial interannual variability within assemblages
136
of microbial eukaryotes at each depth at the sampling site between 2000 and 2003 (Kim et al.
2012b) and the very conservative threshold that was used to determine significance in this study
(Figure S1), our results reflect robust correlations between the temporal dynamics of microbial
eukaryotes at the SPOT station. The simultaneous successes and failures of many taxa at each
depth and the similar taxonomic identities of taxa within guilds (Figures 2 and 3) support the
possibility that taxa within a guild shared similar environmental preferences. The guilds may
have also consisted of microbial eukaryotes that partitioned environmental niches. Although
taxonomic identifications could be assigned to many fragments detected in this study, it is not
possible to determine whether guilds were comprised of taxa with similar ecological activity or
complementary roles based on taxonomic identity alone. However, shared or complementary
roles between co-occurring and non-randomly associated microbial eukaryotes are possible. The
potential for direct interactions between temporally correlated taxa are also feasible. Possible
relationships between taxa include predation, symbioses (mutualism, commensalism, parasitism),
and/or competition. The tight coupling of many taxa could mean that groups of microbial
eukaryotes at our study site are co-evolving (Thompson 2009). The fragmentation of guilds at
each depth (Figure 2) was probably due to only dominant taxa being detected by the T-RFLP
method (Kim et al. 2012a). The T-RFLP approach as well as the monthly sampling resolution
also most likely contributed to the very small number of negative correlations detected between
temporal patterns of microbial eukaryotes at our study site (Figure S1A).
Differences in network-based correlations of microbial eukaryote taxa at different depths
Our results showed that the strength and number of correlations between taxa within the aphotic
zone were greater compared to the taxa at other depths (Tables 1 and 2), suggesting that specific
environmental conditions within the aphotic zone at our study site supported a more highly
137
connected community of microbial eukaryotes. The cold, dark and relatively nutrient-rich
environment at 150 m and 500 m may foster more interactions between taxa than would be
expected within the euphotic zone. For example, predator-prey relationships may be more
tightly coupled below the euphotic zone, where heterotrophy is likely the dominant mode of
nutrition for microbial eukaryotes. Interactions between taxa at 150 m were especially
prominent (Tables 1 and 2), but it is not clear why taxa were more tightly clustered at that depth
compared to 500 m. The very low number of connections between taxa at the DCM may have
been a result of sampling at varying depths.
Conserved and varying guilds of microbial eukaryotes found at different depths
Observations of conserved guilds of microbial eukaryotes at more than one depth (Figure 3)
suggest that there were strict or highly preferred relationships between at least some microbial
eukaryotes at our study site during the ~ 3 year study. It is possible that these connections
included more specialists than generalists. Some microbial eukaryotes, however, were
associated with different guilds of microbial eukaryotes at different depths of our study site
(Figure 4), which highlighted the diverse ecological relationships or interactions that can exist
within natural communities of functionally diverse microbial eukaryotes. For example, many
protists are capable of mixotrophic nutrition, and therefore can potentially exert influence on
multiple trophic levels (Sanders 2011). It is also possible that the varying guilds at the different
depths consisted of taxa that were functionally redundant. It is unclear how common functional
redundancy is within microbial assemblages at our study site. A recent study on bacterial
assemblages at 5 m at the SPOT station concluded that functional redundancy was not likely to
play a large role because of the observed predictable and repeating patterns of temporal
variability of the bacterial assemblage at that depth (Fuhrman et al. 2006). If we follow that
138
same rationale, it is not likely that functional redundancy played a large role within microbial
eukaryote assemblages, considering the seasonal and repeating patterns of temporal variability
that were apparent for microbial eukaryote assemblages within and below the euphotic zone at
the SPOT station (Kim et al. 2012b). However, functionally redundant organisms need not be
identical in physiology or specific environmental preferences, and it is likely that competition
between functionally redundant taxa lead to different ‘winners’ under minor differences in
environmental conditions. Indeed, there is mounting evidence to support the hypothesis that
functional redundancy plays a large role within natural assemblages of microbial eukaryotes
(Caron and Countway 2009).
Synchronicity throughout the water column
In addition to guilds of temporally correlated microbial eukaryotes at specific depths, there were
also 135 correlations between temporal dynamics of taxa from different water column depths
(Table 1 and Figure 1), showing a remarkable coordination between taxa throughout the water
column at our study site over the monthly, seasonal and interannual timescales that were
examined in this study. The synchronized changes in relative abundance between so many taxa
throughout the water column at our study site are an interesting finding, but are not well
understood. The correlations detected between depths were probably not a result or reflection of
ecological linkages. Since only 3 correlations consisted of the same taxa at different depths, it is
also not likely that correlations observed between depths were a result of sinking particulates.
139
References
Aguilera A, Zettler E, Gomez F, Amaral-Zettler L, Rodriguez N, Amils R (2007) Distribution
and seasonal variability in the benthic eukaryotic community of Rio Tinto (SW, Spain),
an acidic, high metal extreme environment. Syst Appl Microbiol 30, 531-546.
Albert R, Barabasi AL (2002) Statistical mechanics of complex networks. Rev Mod Phys 74, 47-
97.
Assenov Y, Ramirez F, Schelhorn SE, Lengauer T, Albrecht M (2008) Computing topological
parameters of biological networks. Bioinformatics 24, 282-284.
Bertness MD, Callaway R (1994) Positive Interactions in Communities. Trends Ecol Evol 9,
191-193.
Camacho J, Guimera R, Amaral LAN (2002) Robust patterns in food web structure. Phys Rev
Lett 88.
Caron DA, Countway PD (2009) Hypotheses on the role of the protistan rare biosphere in a
changing world. Aquat Microb Ecol 57, 227-238.
Caron DA, Countway PD, Jones AC, Kim DY, Schnetzer A (2012) Marine protistan diversity.
Annu Rev Mar Sci 4, 467-493.
Cohen JE, Briand F, Newman CM (1990) Community food webs: data and theory, illustrated
edn. Springer-Verlag, 1990.
Countway PD, Gast RJ, Savai P, Caron DA (2005) Protistan diversity estimates based on 18S
rDNA from seawater incubations in the western North Atlantic. J Eukaryot Microbiol 52,
95-106.
140
Countway PD, Vigil PD, Schnetzer A, Moorthi SD, Caron DA (2010) Seasonal analysis of
protistan community structure and diversity at the USC Microbial Observatory (San
Pedro Channel, North Pacific Ocean). Limnol Oceanogr 55, 2381-2396.
Deng Y, Jiang YH, Yang YF, He ZL, Luo F, Zhou JZ (2012) Molecular ecological network
analyses. BMC Bioinformatics 13.
Fuhrman JA, Hewson I, Schwalbach MS, Steele JA, Brown MV, Naeem S (2006) Annually
reoccurring bacterial communities are predictable from ocean conditions. Proc Natl Acad
Sci USA 103, 13104-13109.
Gurevitch J, Hedges LV (1999) Statistical issues in ecological meta-analyses. Ecology 80, 1142-
1149.
Kaplan CW, Kitts CL (2004) Bacterial succession in a petroleum land treatment unit. Appl
Environ Microbiol 70, 1777-1786.
Kim DY, Countway PD, Yamashito W, Caron DA (2012a) A combined sequence-based and
fragment-based characterization of microbial eukaryote assemblages provides taxonomic
context for the Terminal Restriction Fragment Length Polymorphism (T-RFLP) method.
J Microbiol Meth (in press).
Kim DY, Countway PD, Yamashito W, Tung C, Caron DA (2012b) Monthly, seasonal and
interannual variability of microbial eukaryote assemblages within and below the euphotic
zone in the eastern North Pacific. (submitted).
Massana R, Pedros-Alio C (2008) Unveiling new microbial eukaryotes in the surface ocean.
Curr Opin Microbiol 11, 213-218.
141
Medlin LK, Metfies K, Mehl H, Wiltshire K, Valentin K (2006) Picoeukaryotic plankton
diversity at the Helgoland time series site as assessed by three molecular methods.
Microb Ecol 52, 53-71.
Montoya JM, Pimm SL, Sole RV (2006) Ecological networks and their fragility. Nature 442,
259-264.
Orsi W, Edgcomb V, Jeon S, Leslin C, Bunge J, Taylor GT, Varela R, Epstein S (2011) Protistan
microbial observatory in the Cariaco Basin, Caribbean. II. Habitat specialization. ISME J
5, 1357-1373.
Piwosz K, Pernthaler J (2010) Seasonal population dynamics and trophic role of planktonic
nanoflagellates in coastal surface waters of the Southern Baltic Sea. Environ Microbiol
12, 364-377.
Ruan QS, Dutta D, Schwalbach MS, Steele JA, Fuhrman JA, Sun FZ (2006) Local similarity
analysis reveals unique associations among marine bacterioplankton species and
environmental factors. Bioinformatics 22, 2532-2538.
Sanders RW (2011) Alternative Nutritional Strategies in Protists: Symposium Introduction and a
Review of Freshwater Protists that Combine Photosynthesis and Heterotrophy. J
Eukaryot Microbiol 58, 181-184.
Schoener TW (1983) Field Experiments on Interspecific Competition. American Naturalist 122,
240-285.
Shannon P, Markiel A, Ozier O, Baliga NS, Wang JT, Ramage D, Amin N, Schwikowski B,
Ideker T (2003) Cytoscape: A software environment for integrated models of
biomolecular interaction networks. Genome Res 13, 2498-2504.
142
Sherr BF, Sherr EB, Caron DA, Vaulot D, Worden AZ (2007) Oceanic protists. Oceanography
20, 130-134.
Sih A, Crowley P, Mcpeek M, Petranka J, Strohmeier K (1985) Predation, Competition, and Prey
Communities - a Review of Field Experiments. Annu Rev Ecol Syst 16, 269-311.
Sole RV, Montoya JM (2001) Complexity and fragility in ecological networks. Proceedings of
the Royal Society B-Biological Sciences 268, 2039-2045.
Steele JA, Countway PD, Xia L, Vigil PD, Beman JM, Kim DY, Chow CET, Sachdeva R, Jones
AC, Schwalbach MS, Rose JM, Hewson I, Patel A, Sun FZ, Caron DA, Fuhrman JA
(2011) Marine bacterial, archaeal and protistan association networks reveal ecological
linkages. ISME J 5, 1414-1425.
Storey JD (2002) A direct approach to false discovery rates. J Roy Stat Soc B 64, 479-498.
Strogatz SH (2001) Exploring complex networks. Nature 410, 268-276.
Thompson JN (2009) The Coevolving Web of Life. American Naturalist 173, 125-140.
Vaulot D, Eikrem W, Viprey M, Moreau H (2008) The diversity of small eukaryotic
phytoplankton (<= 3 mu m) in marine ecosystems. FEMS Microbiol Rev 32, 795-820.
Wagner A, Fell DA (2001) The small world inside large metabolic networks. Proceedings of the
Royal Society B-Biological Sciences 268, 1803-1810.
Watts DJ, Strogatz SH (1998) Collective dynamics of 'small-world' networks. Nature 393, 440-
442.
Weber F, Del Campo J, Wylezich C, Massana R, Jurgens K (2012) Unveiling trophic functions
of uncultured protist taxa by incubation experiments in the brackish Baltic Sea. PLoS One
7, e41970.
143
Williams RJ, Berlow EL, Dunne JA, Barabasi AL, Martinez ND (2002) Two degrees of
separation in complex food webs. Proc Natl Acad Sci USA 99, 12913-12916.
Williams RJ, Martinez ND (2000) Simple rules yield complex food webs. Nature 404, 180-183.
Xia LC, Steele JA, Cram JA, Cardon ZG, Simmons SL, Vallino JJ, Fuhrman JA, Sun FZ (2011)
Extended local similarity analysis (eLSA) of microbial community and other time series
data with replicates. Bmc Syst Biol 5.
Zhou JZ, Deng Y, Luo F, He ZL, Tu QC, Zhi XY (2010) Functional Molecular Ecological
Networks. Mbio 1.
144
Table 1. Summary table of the total number of OTUs detected at 5 m, the deep chlorophyll
maximum, 150 m and 500 m between September 2000 and December 2003 at the San Pedro
Ocean Time-series station, the number of OTUs that occurred at least three times during the
study period, the number of OTUs that had at least one significant (p<0.001, q<0.05) positive
Spearman correlation with another OTU from any depth, and the number of significant positive
correlations between taxa contained within the same depth as well as between taxa from different
depths. The number of positive correlations between taxa within the same depth is bolded (and
percentages are reported in brackets). The total number of positive correlations between taxa
from different depths as well as the total number of positive correlations overall are also
presented in the table.
Dept
h
Tota
l
OT
Us
OTUs
that
occurr
ed at
least 3
times
OTUs
with
significa
nt
positive
correlati
ons
Positive
correlati
ons with
5 m
OTUs
Positive
correlati
ons with
DCM
OTUs
Positive
correlati
ons with
150 m
OTUs
Positive
correlati
ons with
500 m
OTUs
Total
positive
correlati
ons with
other
depths
Total
positive
correlati
ons
5 m 169 97 75
66
(46.5%)
36 24 16 76 142
DC
M
168 88 63 36
40
(32%)
20 26 82 124
150
m
147 84 71 24 20
84
(60%)
13 57 141
500
m
129 71 53 16 26 13
56
(49%)
55 114
145
Table 2. Clustering coefficients and average shortest path lengths for networks of significant
(p<0.001, q<0.05) positive Spearman correlations between microbial eukaryotes at 5 m, the deep
chlorophyll maximum, 150 m, and 500 m in samples collected approximately every month
between September 2000 and December 2003 at the San Pedro Ocean Time-series station.
Clustering coefficients and average shortest path lengths obtained for randomly generated
networks of the same size are also reported. The log response ratio of clustering coefficients and
average shortest path lengths are presented for each depth and the total dataset in order to
facilitate the comparison of our result to other network-based studies.
Dataset Cl
a
Cl
random
b
Cl/Cl
random
L
c
L
random
d
lr
Cl
e
lr
L
e
5 m 0.15
0.06
(± 0.01)
2.50 3.02
2.74
(± 0.02)
0.92 0.10
DCM 0.15
0.06
(± 0.01)
2.50 2.78
2.74
(± 0.02)
0.92 0.01
150 m 0.35
0.06
(± 0.01)
5.83 3.78
2.74
(± 0.02)
1.76 0.32
500 m 0.21
0.06
(± 0.01)
3.50 2.17
2.74
(± 0.02)
1.25 -0.23
All depths 0.22
0.06
(± 0.01)
3.67 7.25
2.74
(± 0.02)
1.30 0.97
a
Cl is the clustering coefficient
b
Cl
random
is the clustering coefficient for the randomly generated networks
c
L is the average shortest path length
d
L
random
is the average shortest path length for the randomly generated networks
e
lr are the log response ratios
146
Figure Legends
Figure 1 Network diagram of significant positive Spearman correlations (p<0.001, q<0.05) that
were detected between microbial eukaryote taxa within and between four depths (5 m, the deep
chlorophyll maximum, 150 m and 500 m) at the San Pedro Ocean Time-series station for
samples collected between September 2000 and December 2003. Taxa (squares) from each
depth are clustered together in a circle, and included 75, 63, 71 and 53 OTUs from 5 m, the
DCM, 150 m and 500 m, respectively. The positive Spearman correlation values ranged
between 0.53 and 0.90, and are depicted as black lines. The total number of significant positive
correlations detected between taxa within and between depths is summarized in Table 1.
Figure 2 Network diagram of significant positive Spearman correlations (p<0.001, q<0.05) that
were detected between the temporal dynamics of taxa at 150 m for samples collected between
September 2000 and December 2003 at the San Pedro Ocean Time-series station. There are 71
taxa (depicted as squares) and 84 correlations (depicted as black lines), which ranged between
0.53 and 0.82. Taxonomic identifications were assigned to fragments using a database of 1,341
18S rRNA gene sequences with assigned taxonomies that were subjected to in silico restriction
digestion to predict the length of terminally labeled fragments in a 5 m and 500 m sample
collected at the study site (obtained from Kim et al., in press.). Taxonomic identifications were
abbreviated as follows: alveolate=Alv; alveolate group I=AlvGrpI; alveolate group II=AlvGrpII;
acantharea=Acanth; ciliate=Cil; cnidarian=Cnid; dinoflagellate=Dino; euglena=Eug; no
match=Euk; polycystine=Polycyst; rhizaria=Rhiz). Fragment lengths of taxa are indicated
147
following the taxonomic abbreviations. There were 10 taxa that were temporally correlated only
with taxa from other depths.
Figure 3 Network diagrams of sub-clusters or guilds of microbial eukaryote taxa that were
temporally correlated at 150 m (A) and 500 m (B) between September 2000 and December 2003
at the San Pedro Ocean Time-series station. Black lines denote significant positive correlations
and are labeled with Spearman correlation values. Taxonomic identifications for the OTUs or
fragments were assigned using a database of 1,341 18S rRNA gene sequences with assigned
taxonomies that were subjected to in silico restriction digestion to predict the length of terminally
labeled fragments at the study site (obtained from Kim et al., in press.). Taxonomic
identifications were abbreviated as follows: alveolate=Alv; no match=Euk;
polycystine=Polycyst) and fragment lengths are indicated following each abbreviation. The sub-
clusters at each depth included three of the same taxa. Time-series plots of the changes in
relative abundances for all taxa over the ~3 years examined in this study are detailed on the right
panels for each depth. Both time-series plots consist of two y-axes and share one legend.
Figure 4 Network diagram of significant positive Spearman correlations (p<0.001, q<0.05) that
were detected between the temporal dynamics of Ostreococcus (Ostreo_259) and a dinoflagellate
(Dino_594) at 5 m and Ostreococcus (Ostreo_259) and an unknown eukaryote (Euk_322) at the
DCM in samples that were collected at the San Pedro Ocean Time-series station between
September 2000 and December 2003 (A). Black lines denote significant positive correlations
and are labeled with Spearman correlation values. Taxonomic identifications were assigned to
fragments using a database of 1,341 18S rRNA gene sequences with assigned taxonomies that
148
were subjected to in silico restriction digestion to predict the length of terminally labeled
fragments at the study site (obtained from Kim et al., in press.). Taxonomic identifications were
abbreviated as follows: alveolate=Alv; alveolate group I=AlvGrpI; ciliate=Cil;
dinoflagellate=Dino; euglena=Eugl; no match=Euk; polycystine=Polycyst). Fragment lengths
follow each abbreviation. Time-series plots of changes in relative abundances for all taxa in
panel A are shown in panel B. Network diagrams of guilds of microbial eukaryote taxa that
increased and decreased in relative abundance together as a unit at 150 m and 500 m are also
shown (C). The two guilds in panel C have one taxon in common, Polycyst_228 (circled).
Spearman correlation values for the networks in panel C ranged between 0.53 and 0.90.
149
Figure 1
150
Figure 2
151
Figure 3
152
Figure 4
153
Supplemental Information
Supplemental Figure S1 The distribution of significant Spearman correlation values (q<0.05,
p<0.001) (A), all Spearman correlation values (B), p-values (C), and q-values (D) for pairwise
comparisons of microbial eukaryote taxa at four depths in the eastern North Pacific between
September 2000 and December 2003.
Supplemental Figure S2 The percentage of unique and shared fragment lengths between 5 m,
the deep chlorophyll maximum, 150 m and 500 m in samples that were collected at
approximately monthly intervals between September 2000 and December 2003 at the San Pedro
Ocean Time-series station. Only fragments that were detected at least 3 times during the ~3
years (the same subset that was used for Spearman correlation analysis) were considered. There
were 147 unique OTUs detected when all depths were combined for the analysis. Depths are on
the y-axis, and percentages are listed on the x-axis (but not scaled) and categorized as follows:
all depths, shallow or deep, not depth-specific. Bars span the depths included for each
percentage listed on the x-axis.
154
Supplemental Figure S1
155
Supplemental Figure S2
156
Alphabetized Bibliography
Abdo Z, Schuette UME, Bent SJ, Williams CJ, Forney LJ, Joyce P (2006) Statistical methods for
characterizing diversity of microbial communities by analysis of terminal restriction
fragment length polymorphisms of 16S rRNA genes. Environ Microbiol 8, 929-938.
Adl SM, Simpson AGB, Farmer MA, Andersen RA, Anderson OR, Barta JR, Bowser SS,
Brugerolle G, Fensome RA, Fredericq S, James TY, Karpov S, Kugrens P, Krug J, Lane
CE, Lewis LA, Lodge J, Lynn DH, Mann DG, McCourt RM, Mendoza L, Moestrup O,
Mozley-Standridge SE, Nerad TA, Shearer CA, Smirnov AV, Spiegel FW, Taylor MFJR
(2005) The new higher level classification of eukaryotes with emphasis on the taxonomy
of protists. J Eukaryot Microbiol 52, 399-451.
Agis M, Granda A, Dolan JR (2007) A cautionary note: Examples of possible microbial
community dynamics in dilution grazing experiments. J Exp Mar Biol Ecol 341, 176-183.
Aguilera A, Souza-Egipsy V, Gonzalez-Toril E, Rendueles O, Amils R (2010) Eukaryotic
microbial diversity of phototrophic microbial mats in two Icelandic geothermal hot
springs. Int Microbiol 13, 21-32.
Aguilera A, Zettler E, Gomez F, Amaral-Zettler L, Rodriguez N, Amils R (2007) Distribution
and seasonal variability in the benthic eukaryotic community of Rio Tinto (SW, Spain),
an acidic, high metal extreme environment. Syst Appl Microbiol 30, 531-546.
Albert R, Barabasi AL (2002) Statistical mechanics of complex networks. Rev Mod Phys 74, 47-
97.
Altschul SF, Madden TL, Schaffer AA, Zhang JH, Zhang Z, Miller W, Lipman DJ (1997)
Gapped BLAST and PSI-BLAST: a new generation of protein database search programs.
Nucleic Acids Res 25, 3389-3402.
157
Amaral-Zettler L, Gomez F, Zettler E, Keenan BG, Amils R, Sogin M (2002) Eukaryotic
diversity in Spain's River of Fire. Nature 417, 137.
Ashelford KE, Chuzhanova NA, Fry JC, Jones AJ, Weightman AJ (2005) At least 1 in 20 16S
rRNA sequence records currently held in public repositories is estimated to contain
substantial anomalies. Appl Environ Microbiol 71, 7724-7736.
Assenov Y, Ramirez F, Schelhorn SE, Lengauer T, Albrecht M (2008) Computing topological
parameters of biological networks. Bioinformatics 24, 282-284.
Balzano S, Marie D, Gourvil P, Vaulot D (2012) Composition of the summer photosynthetic
pico and nanoplankton communities in the Beaufort Sea assessed by T-RFLP and
sequences of the 18S rRNA gene from flow cytometry sorted samples. ISME J 6, 1480-
1498.
Barbeau K, Moffett JW, Caron DA, Croot PL, Erdner DL (1996) Role of protozoan grazing in
relieving iron limitation of phytoplankton. Nature 380, 61-64.
Benson DA, Karsch-Mizrachi I, Lipman DJ, Ostell J, Sayers EW (2010) GenBank. Nucleic Acids
Res 38, D46-D51.
Berelson WM (1991) The flushing of 2 deep-sea basins, Southern California borderland. Limnol
Oceanogr 36, 1150-1166.
Bertness MD, Callaway R (1994) Positive Interactions in Communities. Trends Ecol Evol 9,
191-193.
Bray JR, Curtis JT (1957) An ordination of the upland forest communities of southern
Wisconsin. Ecol Monogr 27, 326-349.
Bunge J, Barger K (2008) Parametric models for estimating the number of classes. Biometrical J
50, 971-982.
158
Camacho J, Guimera R, Amaral LAN (2002) Robust patterns in food web structure. Phys Rev
Lett 88.
Caron DA (2000) Protistan herbivory and bacterivory. Method Microbiol, Vol 30 30, 289-315.
Caron DA (2009a) New accomplishments and approaches for assessing protistan diversity and
ecology in natural ecosystems. BioSci 59, 287-299.
Caron DA, Countway PD (2009) Hypotheses on the role of the protistan rare biosphere in a
changing world. Aquat Microb Ecol 57, 227-238.
Caron DA, Countway PD, Brown MV (2004) The growing contributions of molecular biology
and immunology to protistan ecology: Molecular signatures as ecological tools. J
Eukaryot Microbiol 51, 38-48.
Caron DA, Countway PD, Jones AC, Kim DY, Schnetzer A (2012) Marine protistan diversity.
Annu Rev Mar Sci 4, 467-493.
Caron DA, Countway PD, Savai P, Gast RJ, Schnetzer A, Moorthi SD, Dennett MR, Moran DM,
Jones AC (2009) Defining DNA-based Operational Taxonomic Units for microbial-
eukaryote ecology. Appl Environ Microbiol 75, 5797-5808.
Caron DA, Countway, P.D., Savai, P., Gast, R.J., Schnetzer, A., Moorthi, S.D., Dennett, M.R.,
Jones, A.C. (2009b) Defining DNA-based operational taxonomic units for microbial-
eukaryote ecology. Appl Environ Microbiol 75, 5797-5808.
Caron DA, Dennett MR, Lonsdale DJ, Moran DM, Shalapyonok L (2000) Microzooplankton
herbivory in the Ross Sea, Antarctica. Deep Sea Res (II Top Stud Oceanogr) 47, 3249-
3272.
Chao A, and T.-J. Shen (2010) Program SPADE (Species Prediction And Diversity Estimation).
159
Chapin FS, III, Zavaleta ES, Eviner VT, Naylor RL, Vitousek PM, Reynolds HL, Hooper DU,
Lavorel S, Sala OE, Hobbie SE, Mack MC, Diaz S (2000) Consequences of changing
biodiversity. Nature 405, 232-242.
Clarke KR (1993) Nonparametric multivariate analyses of changes in community structure. Aust
J Ecol 18, 117-143.
Clarke KR, Gorley RN (2006) PRIMER v6: User manual/Tutorial. In: PRIMER-E, Plymouth.
Cohen JE, Briand F, Newman CM (1990) Community food webs: data and theory, illustrated
edn. Springer-Verlag, 1990.
Cole JR, Chai B, Marsh TL, Farris RJ, Wang Q, Kulam SA, Chandra S, McGarrell DM, Schmidt
TM, Garrity GM, Tiedje JM (2003) The Ribosomal Database Project (RDP-II):
previewing a new autoaligner that allows regular updates and the new prokaryotic
taxonomy. Nucleic Acids Res 31, 442-443.
Collado-Mercado E, Radway JC, Collier JL (2010) Novel uncultivated labyrinthulomycetes
revealed by 18S rDNA sequences from seawater and sediment samples. Aquat Microb
Ecol 58, 215-228.
Countway PD, Caron DA (2006) Abundance and distribution of Ostreococcus sp in the San
Pedro Channel, California, as revealed by quantitative PCR. Appl Environ Microbiol 72,
2496-2506.
Countway PD, Gast RJ, Dennett MR, Savai P, Rose JM, Caron DA (2007) Distinct protistan
assemblages characerize the euphotic zone and deep sea (2500 m) of the western North
Atlantic (Sargasso Sea and Gulf Stream). Environ Microbiol 9, 1219-1232.
160
Countway PD, Gast RJ, Savai P, Caron DA (2005) Protistan diversity estimates based on 18S
rDNA from seawater incubations in the western North Atlantic. J Eukaryot Microbiol 52,
95-106.
Countway PD, Vigil PD, Schnetzer A, Moorthi SD, Caron DA (2010) Seasonal analysis of
protistan community structure and diversity at the USC Microbial Observatory (San
Pedro Channel, North Pacific Ocean). Limnol Oceanogr 55, 2381-2396.
Dawson SC, Pace NR (2002) Novel kingdom-level eukaryotic diversity in anoxic environments.
Proc Natl Acad Sci USA 99, 8324-8329.
Deng Y, Jiang YH, Yang YF, He ZL, Luo F, Zhou JZ (2012) Molecular ecological network
analyses. BMC Bioinformatics 13.
Diez B, Pedros-Alio C, Marsh TL, Massana R (2001a) Application of denaturing gradient gel
electrophoresis (DGGE) to study the diversity of marine picoeukaryotic assemblages and
comparison of DGGE with other molecular techniques. Appl Environ Microbiol 67,
2942-2951.
Diez B, Pedros-Alio C, Massana R (2001b) Study of genetic diversity of eukaryotic picoplankton
in different oceanic regions by small-subunit rRNA gene cloning and sequencing. Appl
Environ Microbiol 67, 2932-2941.
Dolan JRaM, K. (2004) The reliability of grazing rate estimates from dilution experiments: Have
we over-estimated rates of organic carbon consumption by microzooplankton? Ocean Sci
1, 1-7.
Doney SC (2010) The growing human footprint on coastal and open-ocean biogeochemistry.
Science 328, 1512-1516.
161
Dong CM, Idica EY, McWilliams JC (2009) Circulation and multiple-scale variability in the
Southern California Bight. Prog Oceanogr 82, 168-190.
Dong CM, McWilliams JC (2007) A numerical study of island wakes in the Southern California
Bight. Cont Shelf Res 27, 1233-1248.
Dopheide A, Lear G, Stott R, Lewis G (2008) Relative eiversity and community structure of
ciliates in stream biofilms according to molecular and microscopy methods. Appl Environ
Microbiol 75, 5261-5272.
Dore JE, Letelier RM, Church MJ, Lukas R, Karl DM (2008) Summer phytoplankton blooms in
the oligotrophic North Pacific Subtropical Gyre: Historical perspective and recent
observations. Prog Oceanogr 76, 2-38.
Drummond AJ, Ashton B, Buxton S, Cheung M, Cooper A, Duran C, Field M, Heled J, Kearse
M, Markowitz S, Moir R, Stones-Havas S, Sturrock S, Thierer T, Wilson A (2011)
Geneious v5.4, available from http://www.geneious.com/.
Dunbar J, Ticknor LO, Kuske CR (2000) Assessment of microbial diversity in four southwestern
United States soils by 16S rRNA gene terminal restriction fragment analysis. Appl
Environ Microbiol 66, 2943-2950.
Edgcomb V, Orsi W, Bunge J, Jeon S, Christen R, Leslin C, Holder M, Taylor GT, Suarez P,
Varela R, Epstein S (2011) Protistan microbial observatory in the Cariaco Basin,
Caribbean. I. Pyrosequencing vs Sanger insights into species richness. ISME J 5, 1344-
1356.
Edgcomb VP, Kysela DT, Teske A, Gomez AD, Sogin ML (2002) Benthic eukaryotic diversity
in the Guaymas Basin hydrothermal vent environment. Proc Natl Acad Sci USA 99,
7658-7662.
162
Egert M, Friedrich MW (2003) Formation of pseudo-terminal restriction fragments, a PCR-
related bias affecting terminal restriction fragment length polymorphism analysis of
microbial community structure. Appl Environ Microbiol 69, 2555-2562.
Eppley RW (1968) An Incubation Method for Estimating Carbon Content of Phytoplankton in
Natural Samples. Limnol Oceanogr 13, 574.
Epstein S, Lopez-Garcia P (2008) "Missing" protists: a molecular prospective. Biodivers
Conserv 17, 261-276.
Euringer K, Lueders T (2008) An optimised PCR/T-RFLP fingerprinting approach for the
investigation of protistan communities in groundwater environments. J Microbiol Meth
75, 262-268.
Fernandez-Guerra A, Buchan A, Mou XZ, Casamayor EO, Gonzalez JM (2010) T-RFPred: a
nucleotide sequence size prediction tool for microbial community description based on
terminal-restriction fragment length polymorphism chromatograms. Bmc Microbiol 10.
Fisher MM, Triplett EW (1999) Automated approach for ribosomal intergenic spacer analysis of
microbial diversity and its application to freshwater bacterial communities. Appl Environ
Microbiol 65, 4630-4636.
Fitzpatrick E, Caron DA, Schnetzer A (2010) Development and environmental application of a
genus-specific quantitative PCR approach for Pseudo-nitzschia species. Marine Biology
157, 1161-1169.
Fuhrman JA, Hewson I, Schwalbach MS, Steele JA, Brown MV, Naeem S (2006) Annually
reoccurring bacterial communities are predictable from ocean conditions. Proc Natl Acad
Sci USA 103, 13104-13109.
163
Gast RJ, Dennett MR, Caron DA (2004) Characterization of protistan assemblages in the Ross
Sea, Antarctica, by denaturing gradient gel electrophoresis. Appl Environ Microbiol 70,
2028-2037.
Grasshoff K, Kremling K, Ehrhardt M (2007) Methods of Seawater Anlaysis, Third edn. Wiley-
VCH GmbH, Weinhein, Germany.
Gurevitch J, Hedges LV (1999) Statistical issues in ecological meta-analyses. Ecology 80, 1142-
1149.
Hayward TL (2000) El Nino 1997-98 in the coastal waters of Southern California: A timeline of
events. Cal Coop Ocean Fish 41, 98-116.
Hickey BM (1991) Variability in 2 deep coastal basins (Santa Monica and San Pedro) off
Southern California. J Geophys Res (Oceans) 96, 16689-16708.
Hickey BM, Dobbins EL, Allen SE (2003) Local and remote forcing of currents and temperature
in the central Southern California Bight. J Geophys Res (Oceans) 108.
Hinder SL, Hays GC, Edwards M, Roberts EC, Walne AW, Gravenor MB (2012) Changes in
marine dinoflagellate and diatom abundance under climate change. Nat Clim Change 2,
271-275.
Hooker SB, McClain CR (2000) The calibration and validation of SeaWiFS data. Prog
Oceanogr 45, 427-465.
Hutchins DA, Bruland KW (1998) Iron-limited diatom growth and Si : N uptake ratios in a
coastal upwelling regime. Nature 393, 561-564.
Hutchins DA, Mulholland MR, Fu FX (2009) Nutrient cycles and marine microbes in a CO2-
enriched ocean. Oceanography 22, 128-145.
164
Joo S, Lee SR, Park S (2010) Monitoring of phytoplankton community structure using terminal
restriction fragment length polymorphism (T-RFLP). J Microbiol Meth 81, 61-68.
Kaplan CW, Kitts CL (2004) Bacterial succession in a petroleum land treatment unit. Appl
Environ Microbiol 70, 1777-1786.
Kim DY, Countway PD, Gast RJ, Caron DA (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.
Kim DY, Countway PD, Yamashito W, Caron DA (2012a) A combined sequence-based and
fragment-based characterization of microbial eukaryote assemblages provides taxonomic
context for the Terminal Restriction Fragment Length Polymorphism (T-RFLP) method.
J Microbiol Meth (in press).
Kim DY, Countway PD, Yamashito W, Tung C, Caron DA (2012b) Monthly, seasonal and
interannual variability of microbial eukaryote assemblages within and below the euphotic
zone in the eastern North Pacific. (submitted).
Kim HJ, Miller AJ, McGowan J, Carter ML (2009) Coastal phytoplankton blooms in the
Southern California Bight. Prog Oceanogr 82, 137-147.
Kolodziej K, Stoeck T (2007) Cellular identification of a novel uncultured marine stramenopile
(MAST-12 clade) small-subunit rRNA gene sequence from a Norwegian estuary by use
of fluorescence in situ hybridization-scanning electron microscopy. Appl Environ
Microbiol 73, 2718-2726.
Kruskal W (1967) Statistics Moliere and Adams,H. Am Sci 55, 416.
165
Landry MR, Kirshtein J, Constantinou J (1995) A refined dilution technique for measuring the
community grazing impact of microzooplankton, with experimental tests in the central
equatorial Pacific. Mar Ecol Prog Ser 120, 53-63.
Lepere C, Boucher D, Jardillier L, Domaizon I, Debroas D (2006) Succession and regulation
factors of small eukaryote community. composition in a lacustrine ecosystem (Lake
pavin). Appl Environ Microbiol 72, 2971-2981.
Levitus S (1982) Climatological Atlas of the World Ocean, p. 191. NOAA.
Liu WT, Marsh TL, Cheng H, Forney LJ (1997) Characterization of microbial diversity by
determining terminal restriction fragment length polymorphisms of genes encoding 16S
rRNA. Appl Environ Microbiol 63, 4516-4522.
Lopez-Garcia P, Philippe H, Gail F, Moreira D (2003) Autochthonous eukaryotic diversity in
hydrothermal sediment and experimental microcolonizers at the Mid-Atlantic Ridge.
Proc Natl Acad Sci USA 100, 697-702.
Lopez-Garcia P, Rodriguez-Valera F, Pedros-Alio C, Moreira D (2001) Unexpected diversity of
small eukaryotes in deep-sea Antarctic plankton. Nature 409, 603-607.
Lovejoy C, Massana R, Pedros-Alio C (2006) Diversity and distribution of marine microbial
eukaryotes in the Arctic Ocean and adjacent seas. Appl Environ Microbiol 72, 3085-
3095.
Lueders T, Friedrich MW (2003) Evaluation of PCR amplification bias by terminal restriction
fragment length polymorphism analysis of small-subunit rRNA and mcrA genes by using
defined template mixtures of methanogenic pure cultures and soil DNA extracts. Appl
Environ Microbiol 69, 320-326.
166
Lukow T, Dunfield PF, Liesack W (2000) Use of the T-RFLP technique to assess spatial and
temporal changes in the bacterial community structure within an agricultural soil planted
with transgenic and non-transgenic potato plants. FEMS Microbiol Ecol 32, 241-247.
Marande W, Lopez-Garcia P, Moreira D (2009) Eukaryotic diversity and phylogeny using small-
and large-subunit ribosomal RNA genes from environmental samples. Environ Microbiol
11, 3179-3188.
Massana R, Balague V, Guillou L, Pedros-Alio C (2004a) Picoeukaryotic diversity in an
oligotrophic coastal site studied by molecular and culturing approaches. FEMS Microbiol
Ecol 50, 231-243.
Massana R, Castresana J, Balague V, Guillou L, Romari K, Groisillier A, Valentin K, Pedros-
Alio C (2004b) Phylogenetic and ecological analysis of novel marine stramenopiles. Appl
Environ Microbiol 70, 3528-3534.
Massana R, Guillou L, Diez B, Pedros-Alio C (2002) Unveiling the organisms behind novel
eukaryotic ribosomal DNA sequences from the ocean. Appl Environ Microbiol 68, 4554-
4558.
Massana R, Pedros-Alio C (2008) Unveiling new microbial eukaryotes in the surface ocean.
Curr Opin Microbiol 11, 213-218.
Massana R, Pedros-Alio C, Casamayor EO, Gasol JM (2001) Changes in marine
bacterioplankton phylogenetic composition during incubations designed to measure
biogeochemically significant parameters. Limnol Oceanogr 46, 1181-1188.
McGradySteed J, Harris PM, Morin PJ (1997) Biodiversity regulates ecosystem predictability.
Nature 390, 162-165.
167
Medlin LK, Elwood, H. J., Stickel, S., Sogin, M. L. (1988) The characterization of enzymatically
amplified eukaryotic 16S-like rRNA-coding regions. Genetica 71, 491-499.
Medlin LK, Metfies K, Mehl H, Wiltshire K, Valentin K (2006) Picoeukaryotic plankton
diversity at the Helgoland time series site as assessed by three molecular methods.
Microb Ecol 52, 53-71.
Montoya JM, Pimm SL, Sole RV (2006) Ecological networks and their fragility. Nature 442,
259-264.
Moon-van der Staay SY, De Wachter R, Vaulot D (2001) Oceanic 18S rDNA sequences from
picoplankton reveal unsuspected eukaryotic diversity. Nature 409, 607-610.
Moorthi SD, Countway PD, Stauffer BA, Caron DA (2006) Use of quantitative real-time PCR to
investigate the dynamics of the red tide dinoflagellate Lingulodinium polyedrum. Microb
Ecol 52, 136-150.
Moreira D, Lopez-Garcia P (2002) The molecular ecology of microbial eukaryotes unveils a
hidden world. Trends Microbiol 10, 31-38.
Moreira D, Lopez-Garcia P (2003) Are hydrothermal vents oases for parasitic protists? Trends
Parasitol 19, 556-558.
Naeem S, Li SB (1997) Biodiversity enhances ecosystem reliability. Nature 390, 507-509.
Nebel M, Pfabel C, Stock A, Dunthorn M, Stoeck T (2011) Delimiting operational taxonomic
units for assessing ciliate environmental diversity using small-subunit rRNA gene
sequences. Env Microbiol Rep 3, 154-158.
Nolte V, Pandey RV, Jost S, Medinger R, Ottenwalder B, Boenigk J, Schlotterer C (2010)
Contrasting seasonal niche separation between rare and abundant taxa conceals the extent
of protist diversity. Mol Ecol 19, 2908-2915.
168
Not F, del Campo J, Balague V, de Vargas C, Massana R (2009) New Insights into the Diversity
of Marine Picoeukaryotes. PLoS One 4, e7143.
Not F, Gausling R, Azam F, Heidelberg JF, Worden AZ (2007a) Vertical distribution of
picoeukaryotic diversity in the Sargasso Sea. Environ Microbiol 9, 1233-1252.
Not F, Valentin K, Romari K, Lovejoy C, Massana R, Tobe K, Vaulot D, Medlin LK (2007b)
Picobiliphytes: A marine picoplanktonic algal group with unknown affinities to other
eukaryotes. Science 315, 253-255.
Ohman MD, Snyder RA (1991) Growth-kinetics of the omnivorous oligotrich ciliate
Strombidium spp. Limnol Oceanogr 36, 922-935.
Orcutt B, Bailey B, Staudigel H, Tebo BM, Edwards KJ (2009) An interlaboratory comparison
of 16S rRNA gene-based terminal restriction fragment length polymorphism and
sequencing methods for assessing microbial diversity of seafloor basalts. Environ
Microbiol 11, 1728-1735.
Orsi W, Edgcomb V, Jeon S, Leslin C, Bunge J, Taylor GT, Varela R, Epstein S (2011) Protistan
microbial observatory in the Cariaco Basin, Caribbean. II. Habitat specialization. ISME J
5, 1357-1373.
Osborn AM, Moore ERB, Timmis KN (2000) An evaluation of terminal-restriction fragment
length polymorphism (T-RFLP) analysis for the study of microbial community structure
and dynamics. Environ Microbiol 2, 39-50.
Pace NR, Stahl DA, Lane DJ, Olsen GJ (1986) The analysis of natural microbial-populations by
ribosomal-RNA sequences. Adv Microb Ecol 9, 1-55.
Pandey J, Ganesan K, Jain RK (2007) Variations in T-RFLP profiles with differing chemistries
of fluorescent dyes used for labeling the PCR primers. J Microbiol Meth 68, 633-638.
169
Pedros-Alio C (2007) Dipping into the rare biosphere. Science 315, 192-193.
Piwosz K, Pernthaler J (2010) Seasonal population dynamics and trophic role of planktonic
nanoflagellates in coastal surface waters of the Southern Baltic Sea. Environ Microbiol
12, 364-377.
Potvin M, Lovejoy C (2009) PCR-based diversity estimates of artificial and environmental 18S
rRNA gene libraries. J Eukaryot Microbiol 56, 174-181.
Pruesse E, Quast C, Knittel K, Fuchs BM, Ludwig WG, Peplies J, Glockner FO (2007) SILVA: a
comprehensive online resource for quality checked and aligned ribosomal RNA sequence
data compatible with ARB. Nucleic Acids Res 35, 7188-7196.
Rodriguez-Martinez R, Labrenz M, del Campo J, Forn I, Jurgens K, Massana R (2009)
Distribution of the uncultured protist MAST-4 in the Indian Ocean, Drake Passage and
Mediterranean Sea assessed by real-time quantitative PCR. Environ Microbiol 11, 397-
408.
Romari K, Vaulot D (2004) Composition and temporal variability of picoeukaryote communities
at a coastal site of the English Channel from 18S rDNA sequences. Limnol Oceanogr 49,
784-798.
Rose JM, Caron DA (2007) Does low temperature constrain the growth rates of heterotrophic
protists? Evidence and implications for algal blooms in cold waters. Limnol Oceanogr 52,
886-895.
Rossi P, Gillet F, Rohrbach E, Diaby N, Holliger C (2009) Statistical assessment of variability of
Terminal Restriction Fragment Length Polymorphism analysis applied to complex
microbial communities. Appl Environ Microbiol 75, 7268-7270.
170
Ruan QS, Dutta D, Schwalbach MS, Steele JA, Fuhrman JA, Sun FZ (2006) Local similarity
analysis reveals unique associations among marine bacterioplankton species and
environmental factors. Bioinformatics 22, 2532-2538.
Sanders RW (2011) Alternative Nutritional Strategies in Protists: Symposium Introduction and a
Review of Freshwater Protists that Combine Photosynthesis and Heterotrophy. J
Eukaryot Microbiol 58, 181-184.
Schloss PD, Westcott SL, Ryabin T, Hall JR, Hartmann M, Hollister EB, Lesniewski RA,
Oakley BB, Parks DH, Robinson CJ, Sahl JW, Stres B, Thallinger GG, Van Horn DJ,
Weber CF (2009) Introducing mothur: open-Source, platform-independent, community-
supported software for describing and comparing microbial communities. Appl Environ
Microbiol 75, 7537-7541.
Schnetzer A, Moorthi SD, Countway PD, Gast RJ, Gilg IC, Caron DA (2011) Depth matters:
Microbial eukaryote diversity and community structure in the eastern North Pacific
revealed through environmental gene libraries. Deep Sea Res (I Oceanogr Res Pap) 58,
16-26.
Schoener TW (1983) Field Experiments on Interspecific Competition. American Naturalist 122,
240-285.
Schutte UME, Abdo Z, Bent SJ, Shyu C, Williams CJ, Pierson JD, Forney LJ (2008) Advances
in the use of terminal restriction fragment length polymorphism (T-RFLP) analysis of
16S rRNA genes to characterize microbial communities. Appl Microbiol Biot 80, 365-
380.
171
Shannon P, Markiel A, Ozier O, Baliga NS, Wang JT, Ramage D, Amin N, Schwikowski B,
Ideker T (2003) Cytoscape: A software environment for integrated models of
biomolecular interaction networks. Genome Res 13, 2498-2504.
Sherr BF, Sherr EB, Caron DA, Vaulot D, Worden AZ (2007) Oceanic protists. Oceanography
20, 130-134.
Sherr EB, Sherr BF (1994) Bacterivory and herbivory - key roles of phagotrophic protists in
pelagic food webs. Microb Ecol 28, 223-235.
Sherr EB, Sherr BF (2002) Significance of predation by protists in aquatic microbial food webs.
Antonie Van Leeuwenhoek International Journal of General and Molecular Microbiology
81, 293-308.
Shi XL, Marie D, Jardillier L, Scanlan DJ, Vaulot D (2009) Groups without cultured
representatives dominate eukaryotic picophytoplankton in the oligotrophic south east
Pacific Ocean. PLoS One 4, -.
Sih A, Crowley P, Mcpeek M, Petranka J, Strohmeier K (1985) Predation, Competition, and Prey
Communities - a Review of Field Experiments. Annu Rev Ecol Syst 16, 269-311.
Sogin ML, Morrison HG, Huber JA, Welch DM, Huse SM, Neal PR, Arrieta JM, Herndl GJ
(2006) Microbial diversity in the deep sea and the underexplored "rare biosphere". Proc
Natl Acad Sci USA 103, 12115-12120.
Sole RV, Montoya JM (2001) Complexity and fragility in ecological networks. Proceedings of
the Royal Society B-Biological Sciences 268, 2039-2045.
Steele JA, Countway PD, Xia L, Vigil PD, Beman JM, Kim DY, Chow CET, Sachdeva R, Jones
AC, Schwalbach MS, Rose JM, Hewson I, Patel A, Sun FZ, Caron DA, Fuhrman JA
172
(2011a) Marine bacterial, archaeal and protistan association networks reveal ecological
linkages. ISME J 5, 1414-1425.
Steele JA, Countway PD, Xia L, Vigil PD, Beman JM, Kim DY, Chow CET, Sachdeva R, Jones
AC, Schwalbach MS, Rose JM, Hewson I, Patel A, Sun FZ, Caron DA, Fuhrman JA
(2011b) Marine bacterial, archaeal and protistan association networks reveal ecological
linkages. Isme J 5, 1414-1425.
Steinberg DK, Carlson CA, Bates NR, Johnson RJ, Michaels AF, Knap AH (2001) Overview of
the US JGOFS Bermuda Atlantic Time-series Study (BATS): a decade-scale look at
ocean biology and biogeochemistry. Deep Sea Res (II Top Stud Oceanogr) 48, 1405-
1447.
Stepanauskas R, Sieracki ME (2007) Matching phylogeny and metabolism in the uncultured
marine bacteria, one cell at a time. Proc Natl Acad Sci USA 104, 9052-9057.
Stoeck T, Epstein S (2003) Novel eukaryotic lineages inferred from small-subunit rRNA
analyses of oxygen-depleted marine environments. Appl Environ Microbiol 69, 2657-
2663.
Stoecker DK, Johnson MD, de Vargas C, Not F (2009) Acquired phototrophy in aquatic protists.
Aquat Microb Ecol 57, 279-310.
Storey JD (2002) A direct approach to false discovery rates. J Roy Stat Soc B 64, 479-498.
Strogatz SH (2001) Exploring complex networks. Nature 410, 268-276.
Strong A, Chisholm S, Miller C, Cullen J (2009) Ocean fertilization: time to move on. Nature
461, 347-348.
Takishita K, Yubuki N, Kakizoe N, Inagaki Y, Maruyama T (2007) Diversity of microbial
eukaryotes in sediment at a deep-sea methane cold seep: surveys of ribosomal DNA
173
libraries from raw sediment samples and two enrichment cultures. Extremophiles 11,
563-576.
Tekle YI, Parfrey LW, Katz LA (2009) Molecular data are transforming hypotheses on the origin
and diversification of eukaryotes. BioSci 59, 471-481.
Thompson JD, Higgins DG, Gibson TJ (1994) Clustal-W - Improving the sensitivity of
progressive multiple sequence alignment through sequence weighting, position-specific
gap penalties and weight matrix choice. Nucleic Acids Res 22, 4673-4680.
Thompson JN (2009) The Coevolving Web of Life. American Naturalist 173, 125-140.
Thunell RC, Pilskaln CH, Tappa E, Sautter LR (1994) Temporal variability in sediment fluxes in
the San-Pedro Basin, Southern California Bight. Cont Shelf Res 14, 333-352.
Vaulot D, Eikrem W, Viprey M, Moreau H (2008) The diversity of small eukaryotic
phytoplankton (<= 3 mu m) in marine ecosystems. FEMS Microbiol Rev 32, 795-820.
Venrick EL (1998) Spring in the California current: the distribution of phytoplankton species,
April 1993 and April 1995. Mar Ecol Prog Ser 167, 73-88.
Vigil P, Countway PD, Rose J, Lonsdale DJ, Gobler CJ, Caron DA (2009) Rapid shifts in
dominant taxa among microbial eukaryotes in estuarine ecosystems. Aquat Microb Ecol
54, 83-100.
Wagner A, Fell DA (2001) The small world inside large metabolic networks. Proceedings of the
Royal Society B-Biological Sciences 268, 1803-1810.
Watt WD (1971) Measuring primary production rates of individual phytoplankton species in
natural mixed populations. Deep-Sea Res 18, 329-&.
Watts DJ, Strogatz SH (1998) Collective dynamics of 'small-world' networks. Nature 393, 440-
442.
174
Weber F, Del Campo J, Wylezich C, Massana R, Jurgens K (2012) Unveiling trophic functions
of uncultured protist taxa by incubation experiments in the brackish Baltic Sea. PLoS One
7, e41970.
Weekers PHH, Gast RJ, Fuerst PA, Byers TJ (1994) Sequence variations in small-subunit
ribosomal-RNAs of Hartmannella vermiformis and their phylogenetic implications. Mol
Biol Evol 11, 684-690.
Williams RJ, Berlow EL, Dunne JA, Barabasi AL, Martinez ND (2002) Two degrees of
separation in complex food webs. Proc Natl Acad Sci USA 99, 12913-12916.
Williams RJ, Martinez ND (2000) Simple rules yield complex food webs. Nature 404, 180-183.
Worden AZ (2006) Picoeukaryote diversity in coastal waters of the Pacific Ocean. Aquat Microb
Ecol 43, 165-175.
Wu XQ, Menzel WP, Wade GS (1999) Estimation of sea surface temperatures using GOES-8/9
radiance measurements. B Am Meteorol Soc 80, 1127-1138.
Xia LC, Steele JA, Cram JA, Cardon ZG, Simmons SL, Vallino JJ, Fuhrman JA, Sun FZ (2011)
Extended local similarity analysis (eLSA) of microbial community and other time series
data with replicates. Bmc Syst Biol 5.
Youssef NH, Elshahed MS (2008) Species richness in soil bacterial communities: A proposed
approach to overcome sample size bias. J Microbiol Meth 75, 86-91.
Zhang R, Thiyagarajan V, Qian PY (2008) Evaluation of terminal-restriction fragment length
polymorphism analysis in contrasting marine environments. FEMS Microbiol Ecol 65,
169-178.
Zhou JZ, Deng Y, Luo F, He ZL, Tu QC, Zhi XY (2010) Functional Molecular Ecological
Networks. Mbio 1.
175
Zobell CE (1943) The effect of solid surfaces upon bacterial activity. J Bacteriol 46, 39-56.
Zuendorf A, Bunge J, Behnke A, Barger KJA, Stoeck T (2006) Diversity estimates of
microeukaryotes below the chemocline of the anoxic Mariager Fjord, Denmark. FEMS
Microbiol Ecol 58, 476-491.
Abstract (if available)
Abstract
Microbial eukaryotes are critical components of marine ecosystems, contributing to vital ecological and biogeochemical processes, but fundamental knowledge regarding patterns of spatial and temporal variability of natural assemblages of these taxa is limited. Sequence-based and fragment-based genetic approaches were used to characterize changes in community composition and structure of microbial eukaryote assemblages across multiple timescales (days, months, seasons, years). Short-term temporal changes in community composition and structure were characterized for a 3-day bottle incubation experiment that consisted of 3 treatments (a control, inorganic nutrient enrichment, and organic nutrient enrichment). Inorganic and organic enrichments resulted in dramatic changes in community structure and substantially influenced richness estimates, but community composition and structure also responded rapidly and significantly even without nutrient additions. The relative abundance of some initially rare taxa increased dramatically, implying that some taxa comprising the ‘rare biosphere’ responded to take on ecologically important roles under changing environmental conditions. Long-term temporal patterns of variability (monthly, seasonal and interannual) in the composition of natural assemblages of microbial eukaryotes were examined at the San Pedro Ocean Time-series (SPOT) station for 237 samples collected from four depths on cruises between September 2000 and December 2010. The spatiotemporal variability of microbial eukaryote assemblages indicated the presence of distinct communities within and below the euphotic zone at the SPOT station. Month-to-month community similarity values (~51-61%) were relatively high for assemblages at all depths, but assemblages at 5 m were temporally more dynamic compared to deeper assemblages. Seasonality was apparent for microbial eukaryote assemblages within and below the euphotic zone at 5 m and 150 m, but not at the deep chlorophyll maximum (DCM) which varied in depth seasonally, or at 500 m. Microbial eukaryote assemblages exhibited cyclical patterns in nearly half of all the years and depths examined in this study, with an annual resetting of communities during winter. Interannual variability was apparent at all depths and was a major factor influencing community composition at our study site. Network analysis based on global Spearman correlations identified highly correlated temporal patterns between microbial eukaryote taxa for samples collected between September 2000 and December 2003, indicating the presence of unique guilds of microbial eukaryote taxa at each depth with coordinated responses over the 3 years. Our results provide new insight into the temporal changes within natural assemblages of microbial eukaryotes on multiple timescales of variability, an essential step for linking changes in microbial eukaryote communities brought about by environmental fluctuations to overall ecosystem function.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Annual pattern and response of the bacterial and microbial eukaryotic communities in an aquatic ecosystem restructured by disturbance
PDF
Genetic characterization of microbial eukaryotic diversity and metabolic potential
PDF
Temporal variability of marine archaea across the water column at SPOT
PDF
Patterns of molecular microbial activity across time and biomes
PDF
Marine bacterioplankton biogeography over short to medium spatio-temporal scales
PDF
Characterizing protistan diversity and quantifying protistan grazing in the North Pacific Subtropical Gyre
PDF
Microbe to microbe: monthly microbial community dynamics and interactions at the San Pedro Ocean Time-series
PDF
Dynamics of marine bacterial communities from surface to bottom and the factors controling them
PDF
Spatial and temporal dynamics of marine microbial communities and their diazotrophs in the Southern California Bight
PDF
Ecological implications of daily-to-weekly dynamics of marine bacteria, archaea, viruses, and phytoplankton
PDF
Spatial and temporal investigations of protistan grazing impact on microbial communities in marine ecosystems
PDF
Plant-microbial interactions in mangrove sediments under different nutrient conditions
PDF
Marine protistan diversity, spatiotemporal dynamics, and physiological responses to environmental cues
PDF
The development of novel measures of landscape diversity in assessing the biotic integrity of lotic communities
PDF
Ecological patterns of free-living and particle-associated prokaryotes, protists, and viruses at the San Pedro Ocean Time-series between 2005 and 2018
PDF
Thermal diversity within marine phytoplankton communities
PDF
Disentangling the ecology of bacterial communities in cnidarian holobionts
PDF
Future impacts of warming and other global change variables on phytoplankton communities of coastal Antarctica and California
Asset Metadata
Creator
Kim, Diane Young
(author)
Core Title
Changes in the community composition of marine microbial eukaryotes across multiple temporal scales of measurement
School
College of Letters, Arts and Sciences
Degree
Doctor of Philosophy
Degree Program
Marine and Environmental Biology
Publication Date
01/03/2013
Defense Date
11/08/2012
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
18S rDNA,association network,biological oceanography,bottle incubation,interannual variability,microbial ecology,microbial eukaryote,molecular ecology,OAI-PMH Harvest,protist,seasonal variability,time-series
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Caron, David A. (
committee chair
), Fuhrman, Jed A. (
committee member
), Heidelberg, John F. (
committee member
), Heidelberg, Karla B. (
committee member
)
Creator Email
diane.y.kim@gmail.com,dianekim@usc.edu
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c3-126808
Unique identifier
UC11290471
Identifier
usctheses-c3-126808 (legacy record id)
Legacy Identifier
etd-KimDianeYo-1386.pdf
Dmrecord
126808
Document Type
Dissertation
Rights
Kim, Diane Young
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
Tags
18S rDNA
association network
biological oceanography
bottle incubation
interannual variability
microbial ecology
microbial eukaryote
molecular ecology
protist
seasonal variability
time-series